skip to Main Content

>> Back to Research
EuroQol Research Projects – This table lists the ongoing and completed research projects that have been funded by the EuroQol Research Foundation. You can also see a downloadable version of the table here)

Project IdTitleAbstractProject PI / Applicant NameWorking GroupApproved Budget (EUR)StatusStart YearEnd Year
1588-RAAn exploratory study on the constructs of health-related quality of life and mental well-being: results from a Belgian population surveyA substantial number of studies estimated mapping functions between a preference-based measure, such as EQ-5D, and other non-preference-based measures, but few have looked into the concepts being measured by the present measures to check the legitimacy of mapping. Concepts such as health-related quality of life (HRQoL) and mental well-being are vaguely defined in health outcomes research and there is no agreement on their definitions and measurement. A crude analysis of the PubMed citations indicates that EQ-5D is associated with terms such as health status (29%), health-related quality of life (85%), quality of life (81%), well-being (6%), patient-reported outcome (9%), and/or satisfaction (10%), but what does EQ-5D measure? These concepts are not well clarified. In this study, we aim to use a series of statistical methods to explore the constructs of HRQoL and mental well-being, as defined by EQ-5D and GHQ-12, respectively. Specifically, we aim to understand to what extent the items of EQ-5D and GHQ-12 associate/overlap with each other. The second aim is to provide empirical evidence to distinguish the constructs of concepts including health, quality of life, mental well-being and satisfaction, which are all vaguely defined and often used interchangeably. We will use the data collected in the ‘Great Corona Study’ (GCS) in Belgium. The GCS study included EQ-5D-5L, GHQ-12, overall satisfaction score, the Brief Resilience Scale (BRS) and the 6-item Revised UCLA Loneliness Scale (ULS-6). We will perform a set of statistical analyses, including multidimensional scaling and exploratory factor analysis to attain the research aims.Zhuxin MaoDescriptive Systems16980Ongoing20232023
1585-RAMaking composite time trade-off method sensitive for worse-than-dead statesProblems were reported regarding the discriminatory validity of the composite time trade-off (cTTO) in eliciting utilities for health states worse than dead (WTD). The negative values do not correlate with state severity, which may be interpreted as the lead-time TTO part (LT-TTO) being insensitive. However, other explanations are possible related to some peculiarities in how the negative values are distributed: (1) the absence of near-zero values; (2) many observations censored at -1; (3) a peak at -0.5. We aim to see how selected modifications of cTTO can bring about the negative association between severity and negative utility. We plan to use small and well-targeted modifications, to retain the overall gist of cTTO and pinpoint the cause of the current problems. Based on the above three distribution characteristics, we plan to use three corresponding modifications. (1) Removing the comparison vs immediate death (used in cTTO to sort a state as WTD or better than dead), and instead using only the LT-TTO version of the sorting question (10 years in 11111 vs 10 years in 11111 followed by 10 years in a given state). (2) Eliciting <-1 utilities with additional questions after all time in 11111 was traded in LT-TTO. (3) Changing the iterative procedure in LT-TTO from bisection to titration with random starting point. Based on power calculations, we plan to conduct 440 online interviews with trained interviewers and interviewees recruited from an online platform. The primary endpoint will be a correlation between the level sum score and the negative utility.Michał JakubczykValuation89500Ongoing20232024
1591-RAA Systematic Scoping Review to Sythesise Evidence on Health-Related Quality of Life Measures in AfricaDespite widespread promotion and documentation of the use of health-related quality of life(HRQoL) measures in policy planning and resource allocation, evidence from low-income settings, particularly in Africa, is limited. The aim of this review is to summarise the available evidence on the use of HRQoL measures to date. Specifically, to i) synthesise the available evidence on the types of HRQoL measures used; ii)understand the rationale for selection and extent of use of HRQoL measures in clinical practise, clinical trials, industry, and health technology assessment; and iii) describe the key characteristics of different studies, including their recruitment, populations, data collection methods, and use of other variables/measures. A systematic search of literature will be conducted in five databases using pre-determined keywords (Medline, Embase, CINAHL, PsycINFO, Scopus). A grey literature search, and hand searching of reference lists from the included studies will be carried out. Data on study characteristics and HRQoL measures will be extracted using a customised data charting table. The general characteristics of studies and HRQoL data will be analysed using descriptive statistics and thematic analysis. The findings will be presented in the form of tables and narrative summary. The current study will provide evidence on the use of HRQoL measures in low-income settings, as well as highlight EQ-5D performance, utilisation, and application challenges. This will contribute to the instruments’ global applicability. Furthermore, by providing population and unique country characteristics, types of data and optimal modes of collection, EQ-5D application areas, and end-users, it will be used as baseline evidence to design large-scale population level studies across Africa.Begashaw Melaku GebresillassiePopulations and Health Systems, Education and Outreach22200Ongoing20232023
1600-RAPopulation norms and inequalities based on EQ-5D-5L general population surveys (POPS 2): Pilot on MethodologyThe co-investigators of the POPS 2 pilot will pursue the full POPS 2 project after the pilot demonstrated feasibility for obtaining appropriate datasets and the inclusion of sufficient socio-economic information within these datasets. In this second pilot project, we propose to conduct the necessary preliminary work to review methods for standardizing EQ-5D norms across countries and approaches to inequality assessment. The aims are to: (1) Identify standardized reporting of population norms for EQ-5D-5l datasets; (2) Reviewing and select inequality measures to be used in the main project; (3) Drafting tables and data presentation to use in the main POPs 2 study; and (4) Execute a pilot analysis on 1 to 2 country datasets. The results of this project will directly be used to create the planned EQ-5D-5L population norms booklet, which has been the overarching aim of the overall POPS 2 project. Furthermore, investigating the methodology for cross-country comparisons (aims (1) and (2)), especially pertaining to inequality assessment, is expected to generate other research topics for members of the Health Inequalities Special Interest Group (HISIG). Additionally, this pilot will cross-collaborate with other EuroQol funded population level projects which also expressed interest in assessing inequalities. As the limited pilot project seeks to use only 1 to 2 datasets, it can be conducted relatively easily. The finalized method from aims (1) to (3) (including any syntax that can be produced) will be tested on 1 to 2 easily obtainable datasets identified from the POPS 2 pilot.You-Shan Feng, PhDPopulations and Health Systems25000Ongoing20232024
1602-RAThe effect of perspective, duration and views on life after death on valuation of severe states of EQ-5D-Y-3LBackground: Some studies found that the range of the EQ-5D-Y-3L utility scale differs between perspectives, i.e. severe health states receive lower utility when adults take their own perspective compared to the perspective of a child (i.e. a proxy perspective). Such differences were not only observed when adults decide for a child (proxy 1 perspective) but also when they imagine what a child wants itself (proxy 2 perspective). These results suggest that states can be better-than-dead (BTD) in one perspective and worse-than-dead (WTD) in the other. Yet, it is unclear why this happens or why some studies found no differences between perspectives. Objectives: In this proposal, we aim to compare the proportion of states considered WTD between the adult perspective and both proxy perspectives. Two additional factors that may interact with perspective are explored: the duration of the health state and people’s views on life after death. Methods: We employ the BTD method (Stalmeier et al., 2007) in an online sample, which asks individuals to choose between life in a health state and immediate death. These discrete choices are offered with 3 different perspectives (adult-own, proxy 1 and proxy 2), 5 different health states (21111, 22222, 23333, 33323, and 33333) and 5 different durations (no duration, 3 months, 1 year, 5 years, and 10 years). In order to elicit sufficient variation in views on life after death, we split the sample into equal groups of religious and non-religious persons and include some questions on attitude to death and euthanasia.Arthur AttemaValuation, Youth38990Ongoing20232024
1572-RAAcceptability, validity and responsiveness of the EQ-5D-3L in people with multiple sclerosis: A psychometric analysis of data from the UK MS Register**Background**: Over the last two decades, concerns have been raised about the responsiveness of the EQ-5D to detect health changes and intervention effects. Such evidence is especially lacking for people living with Multiple Sclerosis (MS), likely due to the scarcity of suitable sources of longitudinal data. In addition, the content validity of the EQ-5D has been questioned in relation to MS, and indications regarding construct validity are mixed. **Aim of research**: To assess the psychometric properties of the EQ-5D-3L in people with MS in terms of acceptability, content, convergent and discriminative validity, and responsiveness. **Proposed methods**: Analysis will be undertaken using data from the UK MS Register, a prospective, longitudinal, cohort study with n=10,600 current participants. Since 2011, the UK MS Register has collected data biannually on a range of socio-demographic and clinical characteristics, and patient-reported outcome measures including the EQ-5D-3L, MS-specific measures of health-related quality of life and disability, the preference-based Multiple Sclerosis Impact Scale-Eight Dimensions (MSIS-8D), and further symptom-specific scales. The EQ-5D-3L will be evaluated in terms of acceptability (missing values), distribution properties (floor and ceiling effects), convergent (Spearman's correlation) and discriminative validity (one-way ANOVA, independent t-tests, standardised effect sizes and absolute mean values), and responsiveness (standardised response means and effect sizes and paired t-tests). All analyses will be conducted in comparison to the disease-specific and preference-based MSIS-8D. **Benefits to the group**: The study will provide evidence of the psychometric performance of the EQ-5D-3L in MS, a long-term condition that has major impacts on physical, psychological and cognitive components of health-related quality of life. This study will also reveal how the EQ-5D-3L performs compared to the disease-specific MSIS-8D. Finally, this study results will provide the necessary groundwork for future investigations of the EQ-5D-5L in this population.Elizabeth Tompkins (known as Elizabeth Goodwin)Descriptive Systems24941Ongoing20232024
1575-VSValuation of the EQ-5D-Y-3L in Poland**Background:** In the Polish health technology assessment guidelines (AOTMiT 2016), the EQ-5D questionnaires are preferred instruments to calculate quality-adjusted life years (QALYs). Poland has national value sets for both adult versions of the EQ-5D. However, no value set exists for the EQ-5D-Y, what limits its use in HTA. **Objectives:** To produce EQ-5D-Y-3L value set for Poland based on preferences of the general adult population. As a secondary objective, we aim to explore the impact of modifications of cTTO protocol (change of the sorting question) on the discriminatory power of LT-TTO and the face validity of the negative utilities. **Methods:** We will follow the international EQ-5D-Y-3L valuation protocol including discrete choice experiment (DCE) and composite time trade-off (cTTO) tasks. The EQ-VT software will be used to collect data by computer-assisted personal interviews (200 respondents – standard cTTO and 100 respondents – experimental cTTO) and an online panel survey (1000 respondents, DCE). For both surveys, a non-probability quota sampling will be used. For DCE survey quotas will be set for age, gender, geographical region, size of town, level fo education to reflect the composition of the Polish general population. **Expected results: ** We expect that directly measured EQ-5D-Y-3L value set will strengthen the position of EQ-5D as a leading and officially recommended utility measurement instrument in Poland.Dominik GolickiValuation, Youth56447Ongoing20232024
1583-RATesting the ordinal relationship between TTO utilities and the ranking of EQ-5D health states : examining the commensurability of preferences in the MVH datasetEstimation models based on 5L valuation study data combine data from DCE and TTO in a single "hybrid". This approach is not what was intended when the EQ-VT protocol was initially specified. As originally planned, DCE responses would enable the creation of a set of latent scale values for 5L states, with these being transformed to the conventional 0-1 format using TTO preferences. The combination of DCE and TTO preferences relies on a critical assumption - that they are commensurate and in effect, measure the same construct. Unfortunately, the EQ-VT protocol does not collect data that allows us to test this property. Luckily, the original MVH protocol enables us to precisely target the problem. Despite being based on the analysis of 3L preferences, the (definitive) UK 3L valuation study provides an ideal benchmark for the investigation of respondent-level ordinal consistency. The proposed study will examine the extent to which the rank order of EQ-5D health states as explicitly determined by each respondent, is consistent with the ordering of the SAME states as inferred from the TTO values subsequently assigned to them.Paul KindValuation20000Ongoing20232023
1574-RAAlternative approach to value set construction – accounting for interpersonal utility comparisons taboo based on axiomatic approachWhen constructing value sets, the utilities elicited from respondents are averaged and extrapolated to other states via econometric methods. Meanwhile, in economics, the interpersonal utility comparisons are seen as problematic (to say the least). Even if the utilities for each respondent are expressed on identical QALY scale, u(full health)=1, u(dead)=0), this does not imply that the difference between these two states is perceived equally by all respondents. Recently, Jakubczyk (2022, attached) showed how standard averaging of utilities may, first, distort preferences by giving more impact to some individuals and, second, violate some reasonable properties of preference aggregation. As a solution, it was proposed that prior to averaging the utilities elicited from an individual should be rescaled to a [0,1] range (i.e. also for states worse than dead). However, this approach does not fully solve all identified problems. Meanwhile, the axiomatic approach to aggregating utilities has been developed in the general economic literature for decades, for instance, DeMeyer and Plott (1971) or Kaneko and Nakamura (1979). There are three aims. (1) To attempt to use the existing economic literature to construct axioms relevant to health state utility aggregation and propose aggregation method satisfying the selected axioms. (2) The new aggregation method may use non-linear transformations (e.g. Kaneko and Nakamura use logarithmic function), which will require new analytical approaches (e.g., to avoid log(0) problems). Hence, I aim to design and implement such an approach. (3) To check the feasibility of the developed methods in actual datasets using Polish EQ-5D-5L valuation data.Michał JakubczykValuation19400Ongoing20232023
1596-RAGetting personal: scoping the potential of using OPUF to develop an EQ-5D-5L-based a decision aid for routine clinical practiceThe EQ-5D-5L is the most widely used generic measure of health-related quality of life; it is commonly applied in health economic evaluations and population health surveys to inform policy decision making. However, despite its widespread use, the EQ-5D-5L is generally not used in clinical practice on the individual patient level. A major barrier for use in routine care is the lack of an easy way to summarise and convey the complex information contained in EQ-5D-5L health states. It would be inappropriate to apply a social value set to an individual patient, since the preference weights would not reflect the patients’ view of what aspects of their health are most important to them . Other methods for summarising changes on the five dimensions are also not patient-centric: the level sum score is too crude, and psychometric methods would still rely on group-level statistical analyses. None of these are meaningful to any single patient reporting their EQ-5D health state profile. A newly developed health valuation method, called Online elicitation of Personal Utility Functions (OPUF) approach, has the potential to overcome this barrier. The OPUF approach can generate personal EQ-5D-5L value sets, for each respondent. In principle this would allow patients to both self-report their health on the EQ-5D-5L, and (using OPUF) to communicate which aspects of their health problems are most important in determining their overall health related quality of life, from their own perspective. This can provide important information of relevance to decisions to treat, and choices between treatments, allowing clinicians to better understand which health problems are most bothersome to patients, and how side effects of treatments can impact their HRQoL. It might allow summarising changes in health states over time in terms of (personal) utility, which may help patients to better understand the EQ-5D-5L and the trade-offs they face in making treatment decisions. This line of research is in its very early stages. We thus propose a small research project with limited scope, to map opportunities, barriers, stakeholders, potential use cases, etc. For this, we will conduct a targeted literature review, and semi-structured expert interviews. Findings will be summarised in a (future-state) patient journey map, and used to inform an internal discussion and a roadmap for further research for consideration by the Executive Committee.Paul SchneiderValuation, Populations and Health Systems21880Ongoing20232023
1561-RAEQ-5D as an add-on generic measure in psoriasis, when excellent disease-specific measures are present: its psychometrical and clinical value in a representative Swedish cohort# Background Psoriasis is a chronic, immune-mediated, inflammatory skin disorder that may largely impact patients’ health-related quality of life (HRQoL), particularly if other organ systems are also involved (e.g., joints, mental health). The Psoriasis Area and Severity Index (PASI) and Dermatology Life Quality Index (DLQI) are the most widely used psoriasis-specific self-report instruments to measure psoriasis severity. However, these may be insensitive to capture the impact of beyond the skin symptoms. The inclusion the EQ-5D, might overcome this limitation. The Swedish National Quality Register for Systemic Treatment of Psoriasis (PsoReg) is a patient-based register. PsoReg collects information on age, sex, BMI, pregnancy status, consumption of tobacco and alcohol, family history, comorbidities, clinical type/degree of psoriasis, historical and current treatment, medication details and its side effects, PROMs (DLQI, PASI, EQ-5D-3L), and patient-perceived discomfort associated with psoriasis. Unique is the option to link to other national registries, e.g., education, health care costs, and sick leave. The current study uses patient data admitted between April 2006 and December 2022 (n=78,691), without linkage. # Research questions • How is EQ-5D associated with the disease-specific DLQI and PASI, both in terms of coverage/overlap, and in 'target' (severity range)? • Assuming that the EQ-5D has unique information (non-overlap), (1) is this specific to particular EQ domains, and (2) is non-overlap specific to particular background characteristics or disease features of the patient. # Significance The primary aim is to determine the added value of EQ-5D to the default disease-specific PROMs in psoriasis (rather than the reverse question!), building on our own research on mental health and costs in psoriasis. This fast track study also prepares for a larger study with PsoReg including comprehensive linkage expanding on the 'added value' concept. Positive collaboration with the register holder can also be a showcase in Sweden supporting expansion of registry-based EQ research.Sun SunDescriptive Systems, Populations and Health Systems24975Ongoing20232024
1582-RADiscrepancies between EQ-5D-5L self- and proxy ratings in dementiaBackground: Health-Related Quality of Life (HRQoL) has become an important parameter in dementia research, emphasizing the importance of valuing the perspective of patients with dementia (PwD). PwDs cognitive decline affects the validity of self-ratings. Therefore, proxy-proxy ratings by informal caregivers are commonly used, which however often result in discrepancies between both ratings. To date, discrepancies between EQ-5D-5L self- and proxy-proxy ratings, associated factors, and their stability over time have not been analyzed to full extent. Inconclusive evidence to what extent differences between self- and proxy-proxy ratings affect the prediction of objective adverse events, like hospitalizations, is also missing. Aim of research: To examine discrepancies between EQ-5D-5L self- and proxy-proxy ratings in dementia and analyze associated factors explaining the impact on variations and objective adverse events. Proposed methods: The analysis will be based on data from n= 246 PwD and n=246 caregivers of a cluster-randomized, controlled intervention trial (InDePendent) initiated in 2020 in Germany. Self and proxy-proxy EQ-5D-5L and EQ-VAS versions were administered at baseline and six and twelve months after baseline, resulting in n=662 self- and n=662 proxy-proxy assessments. Descriptive analyses (distribution of responses, ceiling and floor effects) and different methods of agreement (Cohen's Kappa, Intra Class Correlation) will be used to describe discrepancies between self- and proxy-proxy ratings on the dimension level, index and VAS. We will use multiple linear and panel data regression to assess socio-demographic (age, sex, education, living situation, working status caregiver) and clinical factors (patients' mobility, activities of daily living, depression, unmet care needs and healthcare utilization, and caregivers' burden and time spent on informal care) associated with the discrepancy between self- and proxy-proxy ratings. Multiple logistic regression analyses will be used to assess the predictive validity of both ratings on subsequent adverse objective health events (institutionalization, hospitalization, care grade increase, and other healthcare utilization). Benefits to the group: The proposed study will extend the knowledge about the impact of the administration of the EQ-5D-5L self- and proxy-proxy version in dementia by adding new evidence about the discrepancies between both ratings. We will provide evidence on factors influencing these discrepancies over time, by identifying predictors of differences between both versions. Finally, results will demonstrate if the EQ-5D-5L self- or proxy-proxy ratings have a higher predictive validity on objective adverse health events and outcomes, which is important information for the administration of the EQ-5D-5L in future research in dementia.Maresa BuchholzDescriptive Systems24700Ongoing20232024
1504-RATesting EQ-5D-5L bolt-ons in patients with sleep & sleep breathing disorders: an exploratory study for making EQ-5D a clinically attractive patient-reported outcomes measure.Recently, the PIs had concluded a systematic review of qualitative evidence on lived experience among patients with sleep breathing disorder which provided rich qualitative data on how the condition impacted patients’ lives from their perspectives. Based on results of this review, we think that existing sleep bolt-on may not be optimal for use as a PROM in this patient population. At the same time, we had also identified several relevant dimensions which are absent in the standard EQ-5D descriptive system. However, there are existing bolt-ons available, such as relationship, energy/tiredness, and memory/concentration. Therefore, we propose testing in 2 phases all these potentially useful bolt-ons in a clinical population with sleep and sleep breathing disorder. In phase one, we will work with clinicians and patients to assess the content validity in terms of relevancy, comprehensibility, and comprehensiveness of EQ-5D together with the bolt-ons. In phase two, 200 patients with sleep and sleep breathing disorders will be interviewed with EQ-5D-5L and two sleep bolt-on variants plus other bolt-ons relevant for the condition, together with condition-specific PROMs. The bolt-ons will be compared with standard EQ-5D-5L and condition-specific PROMs for their ceiling effects, and sensitivity (using F-statistic and AUC) to severity of sleep and sleep breathing disorder. We hope this project will help improve existing bolt-ons particularly the sleep bolt-on, and inform the EuroQol Group of the potentials of the bolt-ons for use as PROMs in therapeutic areas, where the standard EQ-5D alone is considered inadequate.Nan LUODescriptive Systems55860Ongoing20232024
1517-RAEQ-5D for proxy assessment of nursing home residents: A systematic review of feasibility and measurement propertiesBackground: Elderly are the fastest growing age group in most developed countries and are significant users of long-term care facilities such as nursing homes. The health and well-being of nursing home residents is a vital goal of nursing home care, with health-related quality of life (HRQoL) being widely identified as a quality measure. However, as many nursing home residents suffer from dementia, their HRQoL may only be assessed by observers such as nursing home staff and family members. The EQ-5D is a commonly used generic HRQoL measure and its proxy version has been tested and used in the nursing home setting. Aims: This review aims to synthesize evidence on the feasibility and measurement properties of the proxy-reported EQ-5D for assessing nursing home residents worldwide. Methods: A comprehensive search of eight databases will be conducted from 1990 to November 2021. Two independent reviewers will perform the study selection, data extraction and risk of bias assessment. Any disagreements will be resolved via discussion with a third reviewer. Data analysis: The evidence from the included studies with measurement properties will be aggregated. The quality of EQ-5D will be determined and rated as 'sufficient (+)', 'inconsistent (±)', or 'insufficient (−)' if ≥75%, 25%–74% and <25% of the relevant studies had a 'positive' rating, respectively. The overall quality of the evidence for each population will be rated using modified GRADE criteria. Where possible, random-effects meta-analysis, subgroup analysis, and meta-regression was conducted to explore factors affecting the measurement properties and supplement the COSMIN-derived summaries, respectively.Nan LuoDescriptive Systems25000Ongoing20222023
1481-PHDIncorporating informal carers' quality of life in health economic evaluation using the EQ-5DWhile, traditionally, economic evaluations considered the costs and benefits of the person receiving care only, recent international guidelines on conducting economic evaluations recommend including costs and benefits of informal carers when adopting a societal perspective. In the UK, the ‘Health-Related Quality of Life Task & Finish Group’ (HRQOL T&F group) that was initiated by the National Institute for Health and Care Excellence (NICE) put together a set of ‘minimum evidence requirements’, recommending that carer quality of life (QoL) should be measured using the EQ-5D. Given these recent developments, there is an urgent need for further research, examining how carer QoL can be included in health economic evaluation using the EQ-5D. This requires the assessment of the appropriateness of the EQ-5D in capturing caregiving impacts when compared with other QoL measures. Additionally, incorporating carer outcomes, also referred to as ‘spillover effects’, into health economic evaluation requires further investigation into the extent of potential ‘double-counting’. It has been argued that spillover effects may already be implicitly included in patient’s utilities. Finally, there are challenges in combining ‘carer QALYs’ and ‘patient QALYs’ in an economic evaluation that requires further research. The aim of this PhD research program is to address current methodological issues concerning the inclusion of carer outcomes in health economic evaluation using the EQ-5D tools. This ambitious program of research will address the following research questions: (1) What is the psychometric performance of the EQ-5D tools in capturing carer outcomes for use in economic evaluation? (2) What is the extent of double-counting when including carer outcomes in economic evaluation using the EQ-5D? (3) Should ‘carer QALYs’ be weighted differently than ‘patient QALYs’ in economic evaluation? A systematic literature review will be undertaken, examining the performance of the EQ-5D tools in capturing carer outcomes. Additionally, a secondary data analysis will be conducted, assessing the psychometric properties of the EQ-5D-5L in informal carers in Australia. The extent to which members of the general public consider spillover effects when valuing EQ-5D health states will be examined in a time trade-off exercise, supplemented with some think-aloud interviews. Finally, an online survey, comprising a person trade-off exercise, will be administered to a representative sample in Australia, assessing whether the general public assigns different weights to ‘patient QALYs’ and ‘carer QALYs’ The PhD student will be based at Monash University within the School of Public Health and Preventive Medicine, supervised by EuroQol member Dr Lidia Engel. Additional supervisory support will be provided by Prof Cathy Mihalopoulos (Monash University) to meet Australian University guidelines around supervisory panels.Lidia EngelDescriptive Systems, Valuation101000Ongoing20232026
1534-EOOrganizing an EQ-5D-Y workshop in Asia at the ISPOR Asia Pacific Summit 2022 (a virtual event)Substantial research efforts have been devoted to validate the EQ-5D-Y and to establish its value sets in Asia. It is also crucial to inform the users and researchers about the progress and the availability of this instrument. ISPOR Asia-Pacific Summit provides a great platform to present this instrument and its use to users and researchers from this region. After discussion, this team decided to deliver an introductory workshop to the attendees of 2022 ISPOR Asia-Pacific Summit. In this workshop, we aimed to help the users and researchers in this Asia-Pacific region to know about EQ-5D-Y and the research progress within this region. We also aimed to provide researchers practical guidance and tips on obtaining and using this instrument in their studies. The workshop entitled ‘The EQ-5D-Y(Youth) in the Asia-Pacific: What is it, how to get it, and how to use it?’ was accepted and in total, we delivered a one-hour workshop covering 4 pre-recorded pitches, two presentations and a Q&A session. The pre-recorded pitches were prepared incorporating the EuroQol whiteboard animations and were subtitled with both Chinese and English. As a virtual event, the workshop was recorded and can be played by registered attendees. After the meeting, the 4 pitches were placed online at Zhihu.com (similar to Quora, which shares knowledge), which can be accessed by researchers, students, and clinicians in China and beyond. In total, we counted 20+ live attendees and 229 hits on Zhihu.com (from 20 Sep to 05 Nov 2022). The number of attendees viewed the workshop at ISPOR was not formally recorded.Zhihao YangEducation and Outreach7680Completed20222022
1487-RATerror Management Theory: a new observation window on TTO and VASWe address two issues. The first issue is the discussion about the use of TTO and VAS methods in health valuation measurement, and about the health state ‘death’ in valuation tasks. The second issue addresses the lack of interpretation of existing TTO determinants. We tackle these issues using a framework linking TTO values to Terror Management Theory (TMT), a theory describing sociopsychological defenses against death thoughts. Such thoughts are also relevant in the TTO. An intermediary value ‘prolonging life’ was proposed to link TTO to TMT defenses. Aims: In two studies, we address 1) the occurrence of death thoughts in TTO and VAS, the latter with ‘a state equivalent to dead’, 2) the relation between death thoughts on the one hand, and TTO and VAS on the other hand, 3) whether the intermediary value ‘prolonging life’ is tied to TMT defenses, and 4) the relation between the value ‘prolonging life’ in relation to death thoughts and the TTO. Methods: The validated Death Thought Accessibility (DTA) questionnaire is used, that quantifies the occurrence of unconscious death thoughts. In both studies, experimental designs from TMT are used to produce death thoughts, which are subsequently related to the TTO, VAS, or the value ‘prolonging life’. Relevance This proposal used a new observation window to investigate TTO and VAS. It promises a coherent way to think about existent TTO determinants, making ad hoc explanations unnecessary. The proposal underpins the framework linking TMT and the TTO.Peep FM StalmeierValuation65150Ongoing20222024
344-VSTesting the feasibility and acceptability of the EQ-5D-Y-3L valuation protocol in adolescents and adults in PakistanRecently, the Youth Working Group agreed upon a protocol for the valuation of the 3-level version of EQ-5D-Y. The proposed protocol suggested a minimum of 5 health states to be valued by C-TTO, combined with DCE. The sample of respondents is drawn from adult general population and are asked to think of preference for health for children at the age of 10. In our research, the suggested protocol serves as a standard while we take additional steps to test parts of the suggested methodology. In our case, we will test to determine whether preferences elicited from adolescents differ from those elicited from adults for child health states. Furthermore, the qualitative component will be used for in-depth analysis of the preference elicitation method with a subset of the adolescent as well as adult participants to draw out key themes from the experience for comparison among the two groups. All testing of the mixed minimum C-TTO/DCE design will be done in the context of testing and strengthening the valuation protocol in the context of EQ-5D-Y-3L for Pakistani population, validation of EQ-3D-Y-3Ll Urdu version and also arriving at a value set for the EQ-5D-Y in Pakistan, thus using a representative and sufficient large sample.madeeha malikValuation89507Ongoing20222024
1505-RAPhase-2 study of the Global HTA Agency Survey projectThe views of the HTA agencies on HTA methods can have a significant influence on the generation and use of EQ-5D data. In 2019 we proposed an EuroQol research project to assess the current practices, views and needs of HTA agencies around the world with regard to the use of utility data. The Exec approved the budget for the phase-1 study which aimed to develop and pilot a survey for future roll-out. Here we proposed the phase-2 study. The overall aim is to understand HTA practitioners’ practices, preferences and views regarding measurement and use of health-state values. We will conduct an online survey of personnel in HTA agencies (the survey we developed in phase-1). We will use a 2-stage sampling and recruitment procedure to survey 50 HTA agencies. Both statistical and content analysis will be performed. Findings will be reported to the EuroQol Group and globally at ISPOR meetings.Nan LuoEducation and Outreach57500Ongoing20222023
1530-RARe(re)visiting negative composite time trade-off utilities – can threshold hypothesis really save the day?Background: To fully understand health preferences, the utilities of states worse than dead (WTD) must be measured. It is disputed how credibly such negative utilities are elicited with the composite time tradeoff method (cTTO), as these utilities do not correlate with EQ-5D-5L state severity (the insensitivity hypothesis). In spite of this, a recent explanation has been put forward in terms of the variation in the propensity of respondents to consider a state WTD (the threshold explanation). Purpose: To demonstrate that (i) the threshold explanation fails to falsify the insensitivity hypothesis and that (ii) a negative correlation should indeed be obtained if the cTTO results are sensitive to severity. Methods: Utilising data from the Polish EQ-5D-5L valuation study, I replicate the analysis behind the insensitivity hypothesis and the threshold explanation. Following this, I modify the data in two opposite ways: (A) randomly reshuffling utilities (removing sensitivity), (B) imputing utilities based on regression models (assuring sensitivity) or simulations with sensitivity assumed a priori. This is followed by a determination of how the analyses respond to the changes. Results: Reshuffling does not affect the results underlying the threshold explanation; hence, this explanation is compatible with the insensitivity hypothesis. Imputation and simulation show that in reasonable situations a negative correlation between negative utility and severity prevails. Conclusion: cTTO seems largely insensitive to severity for WTD states.Michał JakubczykValuation11400Completed20222022
180-RAA fast-track proposal for supporting a fresh PhD graduate to do post-doc research work on EQ-5DThis short post-doc research project is proposed for Annushiah who recently graduated from the PhD program of Universiti Sains Malaysia. Annushiah’s PhD work surrounds the Malaysian EQ-5D-5L valuation study. She is one of the very few young EQ-5D researchers in Malaysia. She hopes to do post-doc work but cannot find opportunities partly because of COVID-19. In this project, Annushiah will be mentored by Nan Luo to complete several EQ-5D related projects, including drafting a research proposal for seeking government funding and publishing two manuscripts from her PhD work. This project will help Annushiah to continue her career as an academic researcher. Her work will support the use and research of EQ-5D instruments in Malaysia and beyond.Nan LuoOthers20000Ongoing20202021
1447-RAInvestigating the dimensionality of wellbeing instruments and their added value in explaining health and wellbeingBackground & aim Recently, there has been renewed interest in measuring health by adding items (bolt-ons) to traditional measures. This may be related to the shift from cure to care as a result of the increasing number of patients with chronic diseases, whereby (also) other aspects of health may need to be measured to evaluate interventions. Moreover, interventions are increasingly extending to sectors outside of health care, such as social care, which may call for outcome measures to broaden to well-being rather than just a broader definition of health to fully capture the effects of these interventions. The primary aims of this study are to explore the dimensionality of wellbeing measures, and how these dimensions and items loading on them relate to health and wellbeing. Furthermore, the implications of identifying these dimensions for assessing content validity of wellbeing measures will be discussed. Proposed methods Data will be obtained from an existing dataset consisting of 1002 respondents from the general Dutch population. The respondents completed a visual analogue scale (VAS) on health and happiness, together with the EQ-5D-5L and a diverse set of wellbeing outcome measures that are entirely build on input from patients and the general population. An exploratory analysis will be performed to assess the dimensionality of the wellbeing measures. Also the EQ-5D-5L will be included in the item pool. In this way latent factors, and items loading on them, will be identified that were not related to the EQ-5D-5L. Subsequently, the impact of the identified factors and individual items, using linear regressions and latent regressions, will be assessed. Regressions will be fitted to determine whether these factors and related items can explain the variation in health and wellbeing as measured by the health VAS and happiness VAS. The possibility of extending the impact assessment using path analysis and structured equation modelling, both based on the dimensionality assessment, will be investigated. Finally, the information from the regression analyses can be used for assessing content validity of wellbeing measures (e.g. by comparison of beta coefficients, leave out method etc.). These implications will be discussed. Relevance for EuroQol Group The expected relevance for the EuroQol group are a gain in understanding of how the EQ-5D-5L relates to wellbeing measures based on the perspective of the patient/general population and insight into the content validity of the EQ-HWB. This also could support work understanding the relationship between the EQ-5D-5L and EQ-HWB.M. Elske van den Akker-van MarleDescriptive Systems24540Ongoing20222023
466-RATesting the validity of EQ-5D-5L respiratory bolt-ons in a large Australian datasetObjectives: The EQ-5D has been used to assess health related quality of life (HRQoL) in respiratory conditions including chronic obstructive pulmonary disease (COPD) and asthma. However, the core descriptive system may not be sensitive to all the HRQoL impacts of respiratory conditions. To increase the sensitivity of the descriptive system, two bolt-on questions “Limitations in physical activities due to shortness of breath” and “Breathing problems,” have been developed (EQ-5D-5L+R1 and EQ-5D-5L+R2). Psychometric comparisons are required to understand the performance and sensitivity added by the bolt-ons in comparison to other validated instruments. This is important to inform the work of the EuroQol group in the further development of bolt-ons, and the research agenda around bolt-ons. Therefore, this study tested the psychometric characteristics of the EQ-5D-5L+R using a large dataset collected in Australia. Take-away results: o The results show that adding the respiratory bolt-on to the EQ-5D-5L improved the instrument's descriptive sensitivity in a sample of people with respiratory conditions. o The bolt-on ‘‘breathing problems’’ did not have a high correlation with other dimensions which shows that it is a more independent dimension, than the physical activity limitation bolt-on. The “Limitations in physical activities due to shortness of breath” domain might have an overlap with the other EQ-5D domain “usual activities”, which might cause limited sensitivity of the EQ-5D-5L+R1. o The bolt-on instruments are not suggested to be used as a stand-alone instrument however, they can be used alongside the EQ-5D-5L to estimate so-called ‘‘bolt-on’’ QALYs in addition to the calculation of the standard QALYs to show the potential change in treatment impact when a condition-specific domain is included in the economic evaluation. The choice of bolt-on may be driven by whether overall problems or limitations are being measured. Implications and next steps: o Results suggest a level of validity of the bolt-ons in an Australian population, and therefore to facilitate further use, a value set would be beneficial. The values set can be used alongside the EQ-5D in the economic evaluation of new interventions and treatments for respiratory conditions. However, before using the values to measure QALYs (Quality-adjusted life years), future studies should check the validity and responsiveness of the bolt-ons in different respiratory conditions. o Further research is required to investigate whether the EQ-5D-5L+Rs will be more responsive to changes in the health status of patients longitudinally.Mina BahrampourDescriptive Systems24070Ongoing20222022
1493-RAAn investigation into the psychometric performance of the EQ-PSO in patients with atopic dermatitis in the UK and GermanyThe widespread use of the EQ-5D measures stems in large part from the fact that they are generic and preference based, meaning they can be used to estimate utility across a wide range of conditions. To be amenable for valuation, a HRQoL measure must be short. This necessitated brevity, combined with the need to be generic, means that some aspects of health or disease symptoms will be missed, leading to a lack of validity or sensitivity in some conditions. Bolt-ons have been proposed to improve the coverage of the EQ-5D in such conditions. The EQ-PSO was developed to improve the performance of the EQ-5D in psoriasis. It consists of the core EQ-5D-5L dimensions, as well as two bolt-ons covering skin irritation and self-confidence. While the EQ-PSO was developed as psoriasis-specific, it is also likely to be relevant in other skin conditions which present with similar symptoms, for example atopic dermatitis (AD). Evidence on the psychometric performance of the EQ-PSO in AD patients would be of value to determine the broader relevance of these bolt-ons. This study therefore aims to investigate the psychometric performance of the EQ-PSO in patients with atopic dermatitis in the UK and Germany. Data will be collected via online survey. The convergent and known-group validity of the EQ-5D-5L, EQ-PSO and dermatology life quality index (DLQI) will be compared. The structural validity of the EQ-PSO will be explored using factor analysis. Dependency between itch and pain/discomfort will be explored through item correlations and regression analysis. The addition of a sleep bolt-on will be explored to examine whether this adds to the performance of the measure or whether this is already captured within the existing items.Hannah PentonDescriptive Systems49950Ongoing20222023
1483-TVGScientific international exchange project in the context of the EQ-sponsored PhD project (PHD-287) on inequality research with orthopedic registry data.This project is the product of the combined effort of dr. J. Poeran, dr. G.J. Bonsel (EuroQol member) and J. Bonsel. The hosting partner is the New York based Mount Sinai Center for Clinical and Outcomes Research, which is worldwide one of the largest orthopedic research units, with as one of its focus registry data analysis. The director dr. J. Poeran, with Dutch origins, will personally supervise the project. The purpose is to further develop analytical techniques (from Janssen MF, Haagsma J, Bonsel GJ) previously used with the Dutch quality registry data, in particular for inequity analysis. Furthermore, to investigate the potential incorporation of EQ-5D in the Mount Sinai data, and to learn from the organizational and other incentives applied in the research institute, to involve clinicians as end-user of registry data, PROMs in particular. Specific Topic: utilization and outcome inequalities of regional anesthesia in hip and knee surgery in the United States of America.Joshua BonselPopulations and Health Systems5750Ongoing20222022
1463-PHDReporting heterogeneity in health description and valuation: identification, correction, and sourcesEuroQol instruments are often used to compare health (gains) across groups defined by disease, demographic, socioeconomic, and other characteristics. Systematic differences in the way in which groups respond to the instruments – *reporting heterogeneity* (RH) – bias comparisons. For example, *socioeconomic health inequality* will be underestimated if lower socioeconomic groups report health more positively. Such differences in reporting styles may, in part, arise from differences in health knowledge and beliefs. RH is potentially present in EQ-5D descriptions of health levels within dimensions, in EQ VAS global assessments of health, and in EQ-5D value sets representing preferences between levels and across dimensions. To identify and correct RH in these EuroQol instruments, some external anchor is required. For example, a respondent’s evaluation of the health of a vignette description of functioning within a health domain identifies RH, which can then be purged from the respondent’s evaluation of their own health, using the same instrument. This method relies on the assumptions of *vignette equivalence* (VE) – all respondents interpret a vignette in the same way – and *response consistency* (RC) – a respondent uses the same style to report own health and that of the vignettes. There is limited and mixed evidence on the validity of these assumptions, particularly in the context of EuroQol instruments. This PhD project will test for RH in data obtained with EuroQol instruments and it aims to develop methods to purge these instruments of RH. In collaboration with the research team, the student will review vignettes that have been developed for use with the EQ-5D descriptive system, with a view to developing new vignettes for which the identification assumptions (VE and RC) are more plausible (WP1). Next, we will develop and test methods that use vignettes data to identify and correct RH in categorical ratings of health obtained with EQ-5D (WP2). The project will also evaluate the use of vignettes to purge RH from EQ VAS scores (WP3). In addition to identifying RH and correcting for it in data obtained with EuroQol instruments, the project will explore sources of RH. In particular, we aim to test the hypothesis that it arises, in part, from differences in health knowledge and beliefs (WP4, WP5). Evaluation of a health state may be contingent on ability to recognize its risks and on beliefs about the relative position of that state. Finally, we will apply the methods developed in WP1-WP5 to purge RH from estimates of socioeconomic health inequality obtained with EuroQol instruments (WP6).Teresa Bago d'UvaPopulations and Health Systems253328Ongoing20222026
206-RAEstimating an EQ-5D-Y-3L value set in the United KingdomThe international EQ-5D-Y-3L valuation protocol has now been successfully implemented in several studies. This experience was discussed in a three-day workshop organised by the EuroQol Group and summarised in a report. In general, there seems to be consensus that the use of discrete choice experiments (DCE) is appropriate to elicit preferences for health states that will be experienced by children. However, the most appropriate method to anchor latent scale DCE values onto the QALY scale is still uncertain. In addition, the report recommends a clearer engagement with relevant HTA agencies. This revised application has been prepared following these recommendations and present a research study to estimate a number of EQ-5D-Y value sets under different normative scenarios of perspectives and sample representativeness in the UK. The proposed methodology will be discussed before the study commences with an advisory group and relevant stakeholders and revised if necessary. We will conduct an online DCE with a representative sample of 1,000 adolescents (11-17 years old) and 1,000 adults (18 year and older). Participants will complete the DCE tasks using their own perspective or for a 10-year old child. An adult sample of 200 participants will complete a C-TTO task from the same two perspectives, which will be used to anchor DCE values onto the QALY scale. Finally, we will explore the impact of using value sets under different normative judgments of perspective and UK sample representativeness on QALY gains using patient-level data from a clinical trial with long-term follow-up.Oliver Rivero-AriasYouth360500Ongoing20222024
1494-RAAn investigation of differential item functioning related to age, gender and education in the EQ-5D-5L using ordinal logistic regressionA key assumption for the validity of patient reported outcome measure (PROM) data is measurement invariance, which requires that PROM items and response options are interpreted the same across respondents. If measurement invariance is violated, PROMs exhibit differential item functioning (DIF), whereby individuals from different groups with the same underlying health respond differently, potentially biasing scores. This study investigates DIF in the EQ-5D-5L using ordinal logistic regression (OLR) in the Multi-Instrument Comparison (MIC) dataset. DIF will be examined in relation to age, gender and education. OLR will be used to examine whether DIF results in meaningful differences in expected EQ-5D-5L level sum scores. Effect sizes will be examined to see whether DIF indicates meaningful score differences.Hannah PentonDescriptive Systems, Populations and Health Systems24750Ongoing20222023
1489-RAEstimating interactions between the health domains in stand-alone DCE valuation studiesOBJECTIVE: In a previous EQ sponsored study (i.e. EQ Project 415-RA), we proposed a parsimonious modelling approach that allows for a full set of two-way interactions between the EQ-5D health domains. In the absense of a discrete choice experiment (DCE) dataset that was suitably optimized to identify a full set of two-way interactions, the proposed modelling approach was applied to an existing EQ-5D-3L dataset without duration. In this project, the previously developed modelling approach has been applied to a stand-alone EQ-5D-5L DCE with duration and EQ-5D-EQ-5D-3L composite time trade-off (cTTO) dataset. These analyses are aimed at providing important input for a) the anchoring process in the upcoming stand-alone DCE duration protocol, and b) provide evidence that the valuation of the 5 health domains are potentially not independent, which has ramifications for a bolt-on valuation strategy that the VWG needs to develop in the near future. METHODS: Instead of supplementing a main effects model with interactions between each and every level, a more parsimonious optimal scaling approach was used. This approach is based on the mapping of health-state levels onto domain-specific continuous scales. The attractiveness of health states is then determined by the importance-weighted optimal scales (i.e. main effects) and the interactions between these domain-specific scales (i.e. interaction effects). The number of interaction terms only depends on the number of health domains. As a result, interactions between dimensions can be included with only a few additional parameters. EMPIRICAL APPLICATIONS: The proposed models with and without interactions are fitted on three valuation datasets from two different countries, i.e. an Australian EQ-5D-3L time-trade-off (TTO) dataset with N=400 respondents, and a Dutch EQ-5D-5L DCE with duration dataset with N=788 respondents - and a Dutch latent-scale discrete choice experiment (DCE) dataset with N=3,699 respondents that was already analyzed in the previous project. RESULTS: Important interactions between health domains were found in all three applications. The results confirm that the accumulation of health problems within health states has a decreasing marginal effect on health state values. Moreover, conform the prior hypothesis, particularly the DCE with duration value set depends strongly on the inclusion of interaction effects. CONCLUSIONS: The proposed interaction model is parsimonious, produces estimates that are straightforward to interpret, and accommodates the estimation of interaction effects in health state valuation studies with realistic sample size requirements. Not accounting for interactions is shown to result in profoundly biased QALY tariffs, particularly in stand-alone DCE with duration studies.Marcel Jonker, PhDValuation44500Completed20222023
1479-TVGResearch visit to University of Auckland to facilitate collaborative working and engage with local stakeholders to promote routine PROM data collection and analysis in the New Zealand healthcare system.Nils GutackerPopulations and Health Systems11490Ongoing20222022
1484-RAThe impact of traffic-light color coding in discrete choice health-state valuationsAim: The proposed study aims to replicate the results of a previously EQ sponsored study (i.e. EQ project 2015430) based on a sample of UK instead of Dutch respondents, and hopes to confirm the substantial benefit of including level overlap and color coding on respondents’ response efficiency in discrete choice experiments. In addition to seeing whether the results from one country (i.e. the Netherlands) can be replicated in another (i.e. the UK), the current proposal also aims to compare the relative efficiency and potential biases that are introduced when using traffic-light color coding instead of either no colors or shades-of-purple color coding. These findings are intended to inform the upcoming stand-alone DCE valuation protocol that the Valuation Working Group is (about to be) piloting in Trinidad and Tobago. Methods: A randomized controlled trial with 6 study arms will be used to systematically investigate the impact of level overlap and different types of color coding on the choice consistency and attribute attendance of respondents in health-state valuation discrete choice experiments. Each of the six study arms consists of 500 respondents and is based on a Bayesian D-efficient DCE designs with QALY balanced optimization. Based on the experimental study arms, four different hypotheses will be investigated. First, the impact of color coding and overlap on the drop-out rate will be determined. Second, the impact of color coding and overlap on attribute attendance will be investigated. Third, the impact of color coding and overlap on respondents’ choice consistency is estimated and compared between study arms, and fourth, their impact on the relative attribute and level importances is determined. All estimations will be based on Bayesian MIXL models and model specifications that are similar to those used in two preceeding publications (see attachments).Marcel Jonker, PhDValuation11200Ongoing20222022
1475-RAExploring the content validity of the EQ-PSO bolt-ons in chronic skin conditions other than psoriasisBackground: Two psoriasis-specific bolt-ons (skin irritation and self-confidence) have been proposed for the EQ-5D-5L. The two bolt-on dimensions may be relevant in other chronic dermatological conditions; however, evidence on their validity is only available in psoriasis patients. Objective: To conduct qualitative research to explore the content validity of the EQ-5D-5L and EQ-PSO bolt-ons in skin conditions other than psoriasis. The study will also explore the content validity in individuals of different ethnicities. Methods: A targeted literature review will be conducted to identify prevalent skin conditions where the domains may be relevant. The review will aim to identify two chronic skin conditions to focus on. A qualitative study will then be conducted to explore the content validity of the EQ-PSO bolt-ons in the selected conditions. Individual interviews will be conducted until data saturation is reached, which is expected to be when 15-25 interviews with patients with each condition have been conducted. Recruitment will aim to include individuals with the selected conditions, including patients of different socioeconomic background, severities and ethnicities, in the UK. The interviews will aim to assess the three elements of content validity: comprehensiveness, comprehensibility and relevance. Interviews will consist of concept elicitation questions to explore important areas of health-related quality of life in the selected skin conditions and cognitive interview questions about the relevance of and potential overlaps between dimensions, appropriateness of wording and recall period, and important health concepts not captured by the EQ-5D-5L and EQ-PSO bolt-ons. Data will be analysed using thematic and content analysis.Andrew LloydDescriptive Systems80760Ongoing20222023
1455-RAA comparison of the EQ Health and Wellbeing (EQ-HWB) and EQ-5D-5L instruments.The EQ Health and Wellbeing (EQ-HWB) is a newly developed generic measure developed to capture both quality of life and wellbeing. The measure has been developed to be used in economic evaluation across health, social care and public health. As part of a feasibility study recently conducted in the United Kingdom (UK), a UK value set has been produced. Further evidence is required to assess the psychometric performance and validity of the measure. The objective of this study is to investigate the construct validity and convergent validity of the EQ-HWB and to conduct comparative analysis of the performance of the EQ-HWB and EQ-5D-5L including an assessment of the degree of agreement between the measures. Two data sets collected by the Extending the QALY (E-QALY) project will be used which provide data from a mixed sample of the UK general public, patients and carers. Further analysis will be undertaken to examine whether there are differences between versions of the EQ-HWB measure used in the two separate datasets.Emily McDoolDescriptive Systems, Valuation20420Ongoing20222023
1404-RARevised title: Are there any challenges in valuing Y-5L arising from the descriptive system? A multi-country study Previous title: A multi-country pilot study of EQ-5D-Y-5L valuationAbstract (revised): Background: EQ-5D-Y-5L (Y-5L) is being considered for beta status. The YPWG has identified gaps in evidence specific to valuation of Y-5L. It is important that we check now that valuation of Y-5L does not reveal concerns about the descriptive system levels and labels. Aims: (i) Identify whether the descriptive system for the Y-5L (e.g., level labels) presents any problems when states from it are presented in valuation tasks (e.g., preference reversals/logical inconsistency arising from a lack of clarity about severity levels) (ii) Establish the feasibility of eliciting stated preferences for the Y-5L for both adult and adolescent members of the general public (iii) Provide initial evidence on the characteristics of Y-5L values, compared to the Y-3L, e.g. with respect to minimum values. Additional methodological questions are addressed, relating to the role of specified duration ranges in DCE+ duration tasks. Methods: Research questions specific to to valuation of Y-5L (i.e. over and above the wider research questions also relevant to Y-3L) have been identified. These are amenable to investigation using online DCE, complemented by evidence on ‘response scaling’ of level labels. We propose a multi-country study involving: Australia, Canada, China, Netherlands and Spain. The sample includes both adult and adolescent members of the general public. The research design is primarily based on latent scale DCE. The adult sample will additionally complete DCE+duration or DCE+dead tasks. The research design is driven by hypothesis testing; the data will not be used to produce Y-5L value sets. Impact: Research will be undertaken June-Nov 2022, with results available to support decisions about Y-5L status. The research also provides a first step toward the Y-5L value sets which will be required.Nancy DevlinValuation, Youth169563Ongoing20222023
1411-VSValuing health‐related quality of life: An EQ‐5D‐5L value set for MoroccoBackground: An EQ-5D-5L value set allows the computation of Quality Adjusted Life Years (QALYs), which may impact healthcare interventions and decision making. To date, a value set for the EQ-5D-5L based on the health state preferences of the general Moroccan population has never been established. Aim: The aim of this study is to generate a value set for the EQ-5D-5L version based on data from a representative sample of the Moroccan population using the EQ-VT protocol. Methods: For data collection, we will use the international research protocol developed by the EuroQol group to elicit the health preferences using: (i) composite time trade-off (cTTO) and (ii) discrete choice experiment (DCE). The cTTO design includes 86 EQ-5D-5L health states grouped into 10 blocks, each block with 10 health states. The DCE comprises 196 pairs EQ-5D-5L health states grouped into 28 blocks, each block with 7 choice pairs. Results: The utilities of the Moroccan population will allow us to understand in depth the impact of the pathology, socio-economic, socio-demographic and cultural aspect on the Moroccan people QoL’s. The obtained results would be critical for routine clinical practice, and can therefore provide more evidence for decision-makers in healthcare systems. Conclusion: This study support use of EQ-5D-5L data for healthcare decision-making in a Moroccan context. We will use the Moroccan EQ-5D-5L value set to implement the PROMs (Patient Reported Outcomes Measures) in the primary healthcare services allowing a better monitoring and management of patients with chronic diseases.Abdelghafour MARFAKValuation21000Ongoing20222023
1414-EOThe First EuroQol Latin American Academy MeetingThis meeting was held in Trinidad and Tobago at the Trinidad Hilton on November 9th and 10th, 2022 in collaboration between the EuroQol Research Group and the University of the West Indies. The organizing committee comprised Henry, Victor (cohost), Elly, Bas, Bram, and Mandy.Henry BaileyEducation and Outreach98113Completed20222022
1413-RAFitting mixed logit models with a garbage class instead of manually screening for respondents with low data qualityOBJECTIVES: The upcoming stand-alone discrete choice experiment (DCE) valuation protocol requires the Group to think about efficient and reliable approaches to assess DCE data quality. This manuscript introduces the garbage class mixed logit (MIXL) model as a convenient and performant alternative to manually screening for respondents with low data quality. METHODS: Garbage classes are typically used in latent class logit analyses to designate or identify group(s) of respondents with low data quality. Yet the same concept can be applied to achieve an automated selection of respondents in MIXL models as well. RESULTS: Based on a re-analysis of four DCEs, including an EQ-5D-5L dataset, it is shown that the garbage class MIXL model and root likelihood (RLH) tests have indistinguishable empirical accuracy. Previous research has shown that the latter has superior performance compared to internal validity tests (such as repeated and dominant choice tasks), which means that also garbage class MIXL models have excellent sensitivity and specificity. The advantage of garbage class MIXL models, however, is that they require no user effort and produce preference estimates that do not depend on statistical cut-off values. CONCLUSIONS: Including a garbage class in MIXL models removes the influence of respondents with a random choice pattern from the MIXL model estimates, provides an estimate of the number of low-quality respondents in the dataset, and avoids having to manually screen for respondents with low data quality based on internal and/or statistical validity tests. Although less versatile than the combination of standard MIXL estimates with separate assessments of data quality and sensitivity analyses, the proposed garbage class MIXL model provides a fully automated and reliable alternative that is applicable to both DCE with and without duration data but particularly relevant for the upcoming EQ-DCE-VT protocol.Marcel JonkerValuation25000Completed20222022
365-RAMeasuring Health-Related Quality of Life (HRQoL) in the Youngest PopulationThe Toddler and Infant (TANDI) HRQoL measure for use in children aged 0-3 years is designed for completion by proxy and was developed in English in South Africa. It was based on the EuroQol model of health status and aimed to be part of the EuroQol family of instruments to allow measurement and valuation of health across the lifespan. The TANDI was developed following a rigorous process which included the generation and refinement of the item pool through review of the literature (EQ Project 2014_200), cognitive interviews with caregivers of children, a round table meeting with paediatric experts (EQ Project 2016_180) and two rounds of an online Delphi panel with international experts in paediatric health and HRQoL (14). A preliminary set of 11 items (dimensions) were quantitatively tested in children with and without illness and reduced, according to a priori criteria to six dimensions and a Visual Analogue scale measuring general health (15). The final version of the TANDI includes six dimensions: movement, play, pain, relationships, communication and eating, with (currently) three levels of severity in each (Appendix B). The measure has retained the EQ-VAS (15). Content validity and psychometric properties of the TANDI have only been tested in South Africa with good results. Concurrent validity of dimensions showed moderate to strong correlations to instruments measuring similar constructs. Internal consistency reliability was good (α= 0.83). The TANDI was able to discriminate between known groups (children with acute illness, chronic illness and those from the general population). Test-retest results showed no variance for dimension scores of movement and play, and high agreement for pain (83%), relationships (87%), communication (83%) and eating (74%) and the VAS scores were highly correlated (ICC = 0.76; p < 0.001). Intellectual property (IP) of the TANDI has been transferred to the EuroQol Research Foundation and it is currently classified as an experimental version.Janine VerstraeteYouth426330Ongoing20222025
404-RABetter than dead? – but this or that? Testing how framing impacts viewing a health state as worse than dead.Objective: Current approaches to health state valuation rely on credible classification of states as either `better than dead' (BTD) or `worse than dead' (WTD). We investigate how the evaluation of health states is affected by the framing of the BTD/WTD distinction in pairwise comparison tasks. Methods: We conducted an online survey with 361 participants to compare the propensity to value a state as WTD under six frames: derived from regular time trade-off (TTO; frame A) or from lead-time TTO (LT-TTO; frame B), dismissing with immediacy of death (C) or with the process of dying (D), corresponding to CUA by measuring whether extending lifetime is desirable (E) or how health improvements from a given state are perceived (F). Each participant valued 9 EQ-5D-5L health states using three frames. The frames were compared in several approaches to confirm the robustness against indirect comparisons or respondent heterogeneity and inattentiveness. Results: The odds of a state being considered WTD, compared with frame A, increase 2.7-fold (1.5-fold) in frame B and E, respectively, and decrease >5-fold in frame F. Frames C and D do not differ significantly from frame A. Accepting euthanasia and being <40 years old independently increase the odds of considering a state WTD. Conclusions: Different framings of the question whether a state is WTD or BTD, even if theoretically equivalent, yield substantially different results. In particular, whether a state is considered WTD differs greatly between regular TTO and LT-TTO and between the methods used in valuation and in the CUA context.Michał JakubczykValuation39404Completed20222023
1543-RAVMC proposal for developmental work to support development of gender neutral language in the EuroQol suite of measures(Note: This project has been discussed by email between the VMC (Sarah and Jennifer) and Elly and Bernhard. The project is being entered in to the portal so their is a project number allocated to it ad a budget stated for the external researchers and advisors to invoice against). The VMC wish to explore the use of gender-neutral terms in PROMS generally and in the EuroQol instruments in particular and is thus seeking funding to undertake this work. At the 2021 Strategic Planning Meeting the VMC agreed in principle that one of the roles of the Committee is to ensure that the EuroQol instruments are as inclusive as possible. In recent years there has been an increasing realization that gendered language may alienate certain respondents and may exclude some groups from participating in surveys. However, before the VMC proceeds to alter existing wording or introduce new phrases, there needs to be a strong theoretical foundation upon which to make these changes. The need for this strong foundation has become evident with each of our VMC/RWS meetings where the complexity associated with gender neutrality is evident. This proposal is requesting support for external-to-VMC expertise to undertake a scoping review with VMC members. Specifically: 1) For a researcher (external to the VMC) to lead a scoping review of published literature to identify, describe and understand the strategies used by others in addressing gender inclusiveness/neutrality in questionnaires 2) Engage advisors to provide specialist advice. We would like at least one advisor to come from the LGBTQIA+ (Rainbow) community to help ensure findings are interpreted in a non-discriminatory manner and one advisor from a region where there may be cultural/policy challenges to the use of gender neutral/inclusive language. Apart from providing the VMC with a theoretical foundation on which to base any further action on this topic, the project will deliver: A paper prepared for submission to a peer-reviewed journal, a survey of EuroQol members, and a report to the VMC and Executive.Sarah DerrettOthers20900Ongoing20222023
427-RAConceptualising bolt-ons: identifying key questionsThe notion of adding ‘bolt-on’ items to the EQ-5D has been discussed since its conception. However, there remains a lack of consensus on how bolt-ons should be developed and in what circumstances (if any) they ought to be used. The purpose of this project is to develop a foundation on which to build future bolt-ons research. Specifically, we seek to identify the key conceptual questions that are outstanding in order to inform the future development of a conceptual framework for bolt-ons and, ultimately, protocols for their development and use. For instance, it is important to identify questions relating to bolt-on descriptive systems, the valuation of bolt-ons, and how and when to use bolt-ons. We will do this by i) conducting a review of the literature, ii) leading a workshop with relevant EuroQol members, and iii) inviting the wider EuroQol membership to complete a survey. We will work closely with members of EuroQol committees and working groups to establish a supportive infrastructure for bolt-ons research. This collaborative project will enable us to understand how a conceptual framework for bolt-ons may be identified and how future research may address outstanding questions.Chris SampsonDescriptive Systems, Valuation42840Ongoing20222022
442-RAEQ-HWB and EQ-HWB-S in Indonesia: content validity, interviewer administered version, and test-retestThe EuroQol Group developed an instrument intended to measure health and social care related quality of life, called EQ Health and Wellbeing instrument (EQ-HWB). It has long (25 items) and short (9 items) version intended to capture a broader range of symptoms and functioning aspects which may be relevant to general public members, patients, their families, social care users and carers. Further research is needed to validate EQ-HWB and compare it with existing instruments. This study tried to collect evidence about: (i) content and face validity, (ii) develop the Interviewer Assisted (IA) version and check its agreement with the self-completion version, (iii) test-retest reliability of the Indonesian EQ-HWB and EQ-HWB-S. The present study will be added into another EuroQol funded project aimed to validate the IA version of EQ-5D-5L. After translating the two EQ-HWB versions into Indonesian language, we will interview 45 respondents (15 literate, 15 illiterate-low literacy and 15 patients) with varying socio-demographics using a think-aloud concurrent protocol. The full sample of literate groups will complete the IA and SC versions together to obtain agreement of the versions. All respondents will complete the EQ_HWB and EQ-HWB-S two times with 2 weeks interval for test-retest. Interview data will be analyzed using a thematic analysis approach. Agreement between the IA and SC dimensions’ responses and test-retest reliability will be assessed by Weighted Kappa (Kw).Fredrick Dermawan PurbaEQ-HWB24825Ongoing20222023
462-RAEQ-5D 3L in the 2017 National Health Survey for ChileThe Chilean National Health Survey (ENS) is part of the Comprehensive Social Protection System that the Government is promoting. It is a powerful tool used by the Chilean Ministry of Health (MoH) to find out what diseases and what treatments men and women aged 15 years + living in Chile are receiving. The information provided by this survey is of vital importance to formulate prevention plans, care and health policies for people who need it. A team of more than 100 interviewers and nurses surveyed the field in the 15 regions of the country in urban and rural areas, applying questionnaires and medical examinations at the homes of the selected people (N~6000). On the 2017 ENS version, the MoH has made an enormous effort, incorporating the measurement of approximately 60 health problems along with the main risk factors, protectors and their determinants in people's health. EQ-5D 3L questionnaire was also included in this round, along side biometric measures and biomarkers obtained form blood and urinary samples. Researchers from the MoH (Health Technology Assessment Unit) have contacted me to ask for methodological support to estimate EQ-5D population norms for several subgroups. The MoH also need to know how often it would be convenient to include a questionnaire such as EQ-5D to estimate population health. The main aim of this research proposal is to estimate official EQ-5D 3L population norms according to the need of the MoH doing a preliminary analysis of the data and drafting a research agenda for future research projects.Victor ZaratePopulations and Health Systems14840Ongoing20212022
435-RAEmploying Episodic Future Thinking to reduce the distortion of time preference in TTOBackground: EQ-5D valuation with time trade-off (TTO) may be biased by time preferences, i.e., when respondents discount the life duration traded off, TTO valuation may inaccurately reflect disutility of health impairments. Different methods have been proposed to address time preferences, but these typically rely on strong assumptions or ex-post (and error-prone) measurement of time preference. Objectives: Our main motivation is to explore if the use of Episodic Future Thinking (EFT) may reduce the distortion of time preference in TTO valuation of EQ-5D-5L. Besides exploring the effect of EFT on TTO, we also explore the following questions: i) How does EFT affect discounting as elicited in a discounting task? ii) Does EFT yield the same TTO utilities as correcting for discounting? Methods: A sample of 150 students will be recruited for this experiment, which has a control condition and an EFT condition. All data will be collected in 1-on-1 interviewer guided sessions with trained interviewers. The EFT treatment group is introduced to the EFT task in which participants imagine their lives in the next 10-20 years, whereas the control group completes a filler task. Next, all respondents will value up to 10 EQ-5D-5L states with composite TTO and their time preferences will be measured in a discounting task.Arthur AttemaValuation37470Ongoing20222023
460-RAPopulation health impact of the COVID-19 pandemic (POPCORN): third waveThis research proposal concerns Wave 3 of the Population health impact of the COVID-19 pandemic (POPCORN) study; further indicated with POPCORN-W3. The POPCORN study is a longitudinal study which investigates the effects of the COVID-19 pandemic on health-related quality of life (HRQL) of the general population, and to study the role of socio-economic status and other determinants of HRQL. Wave 1 was April-June 2020 and Wave 2 was May-June 2021. POPCORN-W3 will collect HRQL data, by a.o. EQ-5D-5L in six of the nine countries of POPCORN-W1. It is timed two years after W1, the start of the COVID-19 pandemic. The study design permits cross-sectional analysis and individual partial repeat data analysis. The aims of this study are to (1) assess HRQL, measured by EQ-5D-5L, two years after the start of the COVID-19 pandemic of the persons from the general population of six countries and investigate country level (CF) and individual (IF) factors associated with lower HRQL; and (2) assess change in HRQL among 3 strata: (1) COVID-19 patients, (2) patients with chronic condition(s), and (3) healthy participants and investigate CF and IF factors favoring positive change. CF include socioeconomic impact, stringency of measures and vaccination strategy. IF include level of/change in social position, living situation, medical/chronic disease status including vaccination. Methods: a web-based survey among respondents of POPCORN-W1; we expect about 35% response from W1. Deliverables: Two scientific papers that will be submitted to peer-reviewed scientific journals and a presentation at the EuroQol Plenary/Academy meeting and at the HISIG.Juanita HaagsmaPopulations and Health Systems111160Ongoing20222024
449-RAAssessing the health of Ethiopian Adolescents using the EQ-5D-Y-3L: A cross-sectional studyIntroduction: Health-Related Quality of Life (HRQoL) is a multi-dimensional concept, which focuses on the effect of a health condition and its treatment on a person’s daily life. In similar to the adult population assessing HRQoL of the young population will have a great importance in public health research and the evaluation of treatments. Even though, the child friendly version EQ-5D-Y is getting more attention evidences on health status of the adolescent population especially in resource limited, low-income countries (e.g. Ethiopia) are limited. This research will have crucial implications on measuring HRQoL outcomes, resource allocation, planning of health care interventions, policy/decision makers for adolescent populations in Ethiopia. The Young Population Working Group (YPWG) has also expressed interest in receiving proposals presenting population reference data using EQ-5D-Y. Objectives: To assess the population health status based on the EQ-5D-Y-3L among adolescents (12-17 years) in Ethiopia. Methods: Participants will be asked to complete an Amharic version of EQ-5D-Y-3L paper based instrument containing background information. Descriptive statistics will be used to report health according to the five separate dimensions of EQ-5D-Y-3L and EQ-VAS. Chi-square test or the Fisher’s and Mann-Whitney U-test will be used to assess for statistically significance difference between groups of categorical and continuous variables respectively. Multiple logistic and linear regression analyses will also be used to assess for statistically significance of association between reported problems in EQ-5D-Y-3L and VAS score with respondents socio-demographic, and parent’s employment status respectively. Five thousand adolescents (12–17 years) will be recruited from 40 schools in Addis Ababa.Goitom Molalign TakeleYouth25000Ongoing20222022
426-RAHealth-Related Quality of Life in children dependent on technology for breathingMedical advancement has enabled children to survive congenital airway anomalies, rare diseases and critical illnesses with medical technology including tracheostomies and long-term ventilation to support breathing. The number of technology-dependent children is increasing globally with the increased use of healthcare services. There is a clear decrement in HRQoL across multiple instruments, including the EQ-5D, in adult patients with tracheostomies. Most studies in children have studied the impact of family and/or caregiver burden of a child dependent on technology for breathing. None of the evidence includes the EQ-5D-Y. Thus this study aims to assess the psychometric performance, including reliability and responsiveness to change, of the EQ-5D-Y-3L, EQ-5D-Y-5L (proxy and self-report) and TANDI in children dependent on technology in South Africa. Furthermore, the impact of caring for these children will be assessed with the EQ-5D-5L and CarerQol.Janine VerstraeteYouth38855Ongoing20212023
432-RAUsing the online personal utility functions (OPUF) tool to explore issues in the valuation of EQ-5D-YThe valuation of EQ-5D-Y is associated with a wide range of methodological considerations. Whilst a valuation protocol exists, discussions around the most appropriate approaches to valuation have continued and a considerable amount of methodological research is ongoing. Two key considerations are the choice of valuation methodology and the perspective to use in valuation exercises. The personal utility function approach was introduced as a more direct and reflective valuation method compared to existing alternatives and has recently been developed as an online tool (OPUF). The OPUF tool has several advantages over other methods: 1) it can produce individual-level utility functions; 2) relatively few participants are needed to derive a value set; and 3) it produces a value set using a single method, with values being anchored using an innovative trading technique that locates dead within the descriptive system for each respondent (the LOD task). This project aims to utilize the OPUF tool to explore issues relating to the valuation of EQ-5D-Y health states, which can provide several useful insights. Firstly, the LOD task can be used to anchor DCE data and may avoid some of the issues observed with TTO data, potentially providing a useful and less resource-intensive alternative. Secondly, the impact of different perspectives on EQ-5D-Y value sets can be examined further, and heterogeneity can be explored in greater detail than in past/ongoing studies. Alongside these key objectives, the study would also enable a methodological comparison between OPUF and DCE, further contributing to the development of this promising valuation method.David MottValuation, Youth72150Ongoing20222023
444-RAComparing the psychometric properties of the EQ-5D-3L and EQ-5D-5L in Chinese patients with pompe diseaseThe primary aim of this study is to evaluate the psychometric properties of the EQ-5D in a group of patients with pompe disease (PD) in China; The secondary aims are to compare the performance between EQ-5D-3L and EQ-5D-5L. Research team will collaborate with China pompe disease care center, the biggest PD patient organization in China, to recruit participants. A total of 100 PD patients, who met our inclusion criteria, will be invited to participate in an online cross-sectional survey. The data of EQ-5D-5L and participants’ social and health characteristics will be collected. The feasibility and accessibility, convergent and discriminant validity, and exploratory power of the EQ-5D-3L and EQ-5D-5L will be evaluated as well as the agreement between them. This study will provide empirical evidence on the performance of the EQ-5D in patients with a rare and severe disease.Ricahrd XuDescriptive Systems24160Ongoing20222023
452-RABehavioural Groups for composite TTO dataFor TTO, the common utility model used in health state valuation is 𝑈𝑖𝑗 = 𝑉𝑗 + 𝜀𝑖𝑗 , where i indicates respondent and j indicates health state. This model assumes that that all respondents have the same underlying values 𝑉𝑗 and that all variability in the observations is due to random error 𝜀𝑖𝑗. However, we know that preference heterogeneity is present in TTO data, for example, some respondents give worse than dead responses, while others don’t. The aim of the study is to investigate the preference heterogeneity present in EQ-VT composite TTO data, by using machine learning algorithms in order to identify groups of respondents with different data patterns. Specifically, cluster analysis will be used to analyse response patterns in the TTO data with respect to scale use, and with respect to domain preference. Estimating regression models separately for these groups will allow us to explore differences in preference structures and investigate if there are interaction terms that are important for particular groups of respondents, but that are hidden when all data is analysed jointly. Furthermore, we will investigate if the structures and interaction terms found in the TTO data are also present in the DCE data. Lastly, we aim to assess the robustness and generalisability of the method and results by applying them across EQ-VT studies in multiple countries.Mark OppeValuation45760Ongoing20222022
450-RADeveloping tools (Stata, R and Excel) for calculating utility values, analysing and reporting data from the EQ-5D family of instrumentsEuroQol Group has recently funded 2 books: “Methods for analysing and reporting EQ-5D data” and “EQ-5D-5L value sets”. These 2 books together with the previous EQ-5D value sets book will be the main source of guidance for EQ-5D users. The webpage of EuroQol Group now includes a section where some analysis tools can be found. However, those available tools have 3 main weakness, firstly they focus on calculating values, either direct calculation or via crosswalk methods, secondly, they are in separate macros files forcing users to modify the codes to fit their purpose and thirdly no tools for reporting EQ-5D dimensions/EQ-VAS are available. Drs Ramos-Goñi and Rivero-Arias have made some effort in the past to develop tools either for calculating values (eq5d and eq5d5l in Stata) , and for reporting EQ-5D dimensions/EQ-VAS (eq5dds in Stata). All these commands have been standard tools for many EQ-5D users, however, nowadays those tools become to be obsolete. Therefore, the aim of this proposal is to develop the following tools: 1) A single command (eq5d-suite in both Stata and R) which, through options, can either report on the descriptive system (either transversal or longitudinal) or calculate values from the most recent available values sets including EQ-5D-Y ones. 2) Replicate the proposed eq5d-suite command in a VBA environment as an add-on that users can download and install within their Excel package. By providing to users the suggested commands/add-ons EuroQol will improve EQ-5D users experience by facilitating EQ-5D data management, analysis, and reporting.Juan M. Ramos-GoñiDescriptive Systems, Populations and Health Systems, Youth, Education and Outreach78320Ongoing20222022
401-RATesting content validity of the EQ-5D-Y in ChinaEQ-5D-Y is a preference-based instrument designed for measuring and valuing health-related quality of life (HRQOL) of respondents aged 8-18 years old. It has demonstrated satisfactory measurement properties in children/adolescents with and without certain health conditions in China, and has gained more and more popularity in HRQOL researches. However, its content validity has not been assessed in China yet. Hence, the study aimed to assess the content validity of the EQ-5D-Y in general and diseased children/adolescents in China. Children and adolescents with various health conditions (e.g., anemia, dermopathy, asthma, congenital heart disease and leukemia) in a tertiary hospital in Shanghai, China would be recruited to self-complete the EQ-5D-Y questionnaire and then have a semi-structured interview. The first section of the interview consisted of a series of open-ended questions to elicit concepts and understanding on health, and then respondents would answer questions about the relevance and comprehensiveness of the descriptive system in the second sections. Thematic analysis employing open, focused and axial coding was used to identify the themes and subthemes from the interviews. Each question item and corresponding responses would be entered into Nvivo, and all the scripts would be analyzed by sections in accordance with the questions.pei wangYouth18800Ongoing20212023
416-RAMind the gap. Psychological distance in EQ-5D-Y valuation**Background.** The work leading up to and following the publication of the EQ-5D-Y-3L valuation protocol has shown that the use of a 10-year old child’s perspective may yield higher utilities for TTO, lower data quality, and lower variance for TTO. Most of the work in this area explored different reasons why the use of child perspectives could affect EQ-5D valuation, without testing a specific hypothesis or using a particular theoretical framework to explain these effects. **Objectives**. Our proposal aims to use construal level theory (CLT) to explain why the use of child or adult perspectives can affect EQ-5D valuation. CLT explains individuals thinking and reasoning as a function of psychological distance. The higher the psychological distance between the individuals and the concepts they are asked to consider, the more abstract their reasoning, which may affect EQ-5D valuation. Our objective is to answer the following research question: ‘How do the four types of psychological distance (social, temporal, spatial and hypothetical) affect: i) mean EQ-5D-Y utilities, ii) variance of EQ-5D-Y utilities and iii) internal validity of EQ-5D-Y utilities.’ **Methods**. A convenience sample of 150 students will be recruited to take part in in-person cTTO interviews led by an experienced interviewer. Each student will value (at least) 3 EQ-5D-Y-3L health states in four perspectives in random order. A combination of existing and innovative valuation perspectives is used to compare the influence of different types of psychological distance, as well as increases in psychological distance on EQ-5D valuation.Stefan LipmanValuation, Youth24890Ongoing20222023
396-RACorrespondence between directly-reported and recalled HRQoL collected at 1 week, 1month and 2 months post-diabetic ketoacidosis: from patients and proxies perspective (Resubmission)Introduction: Diabetic ketoacidosis results in adverse neurocognitive outcomes which lead to decreased HRQoL. Retrospective assessment of pre-event HRQoL is frequently used to measure change from pre- to post-event of HRQoL. However, retrospective measurement may be confounded by recall bias. Recall bias can be influenced by the measurement scale or the instrument that is used, the measurement schedule, and the presence of a substantial health event during the follow up period. Therefore, the present study will assess whether the EQ-5D-5L would be used to measure HRQoL in recall situations of acute disease conditions. Objectives: It will evaluate the correspondence between directly-reported EQ-5D summary and EQ-VAS scores collected at 1 week (T1) and 1 month (T2) post-DKA, and recalled scores of 1 week (T1) collected at 1 month (T2) and 2 months (T3) post-DKA, and recalled scores of 1 month (T2) collected at 2 months (T3) from patients and proxies. Methods: 200 patients with DKA and 200 proxies will be recruited from September to December 2021. The EQ-5D-5L tool, which is under development into Tigrinya language (under cognitive debriefing phase, TRF2225), will be used for data collection via a postal invitation and phone call to non-responders. We will compute paired t-test and Intraclass correlation coefficient to compare T1–T2, T1–T3, and T2–T3 correspondence of direct (i.e., the EQ-5D outcome at that moment) versus recalled outcomes for the EQ-5D summary, the dimensions, and the EQ-VAS scores separately for patients and their proxies. Overall P-values< 0.05 will be considered as statistically significant.Afewerki GebremeskelDescriptive Systems24620Ongoing20222023
420-RACORFU: a COVID-19 follow-up studyBackground: Many patients who have had a COVID-19 infection keep reporting complaints months after infection, even those with a relatively mild infection. These complaints in formerly COVID-19 patients can have a significant impact on quality of life. Long-term complaints are prevalent in former COVID-19 patients ranging from those who had a mild infection without hospital admission to those who have had mechanical ventilation at the ICU ward. Aims: The aim of this project is to focus on long term complaints up to 2 years after infection. Sub-aims of the study for which data of the POPCORN cohort will be used are: 1) describing the prevalence and nature of long-term complaints after COVID-19 infection up to two years after, and in relation to health-related quality of life; and 2) development of a prediction model to be able to estimate an individual’s probability of long-term complaints using readily-available predictors. Both aims will help to understand the usefulness of the EQ-5D-5L in assessing HRQoL in patients who continue to experience symptoms after COVID-19 and it application in prediction models of long COVID. Deliverables: Two scientific papers that will be submitted to peer-reviewed scientific journals.Juanita HaagsmaPopulations and Health Systems15660Ongoing20222023
409-RADeveloping a value set for the EQ-5D-Y-3L in the United StatesEQ-5D-Y-3L valuation studies are currently underway in different countries following the recently published international protocol. EQ-5D-Y valuation consists of a different, if not more complex, set of challenges compared with conducting adult EQ-5D valuations. Among other things, questions remain surrounding the framing of discrete choice experiment (DCE) and composite time trade off (cTTO) considering the views about a 10-year-old child, as is suggested by the published protocol. Further, in previous EQ-5D-Y-3L valuation studies, there has been limited engagement with key stakeholders, which could have significant implications for uptake and application. The US valuation team is proposing two main aims of this study. First, we aim to elicit views of representatives of key stakeholder groups/potential users (i.e., industry, payers, health systems, patient reps, academia) to understand user needs, concerns, and inform the design of the valuation study. Feedback from the focus group meeting will inform the second aim and main purpose of the study, which is to estimate a US value set for the EQ-5D-3L-Y.Ning Yan GuValuation, Youth174740Ongoing20212023
398-RATo what extent does EQ-5D reflect the health concepts of Chinese: a scoping review**Background: ** EQ-5D was found to have a high ceiling effect and be insensitive in detecting differences in health status in China. In recent years, the legitimacy of applying EQ-5D in the Chinese cultural setting has been questioned. Given health can be conceptualised differently in China, it can be argued that EQ-5D may miss some important questions for Chinese populations in describing health. However, no systematic investigations into this potential issue have been made and what important health dimensions to Chinese but not included in EQ-5D remains unknown. **Aims: ** This study aims to systematically review relevant studies on the Chinese concepts of health and health-related quality of life, in order to identify those important health dimensions in a Chinese cultural setting but have not been included in EQ-5D. **Methods: ** A scoping review will be undertaken to attain the research aim. First, a systematic literature search will be conducted in Chinese databases CNKI and Wanfang, as well as English databases EMBASE, MEDLINE and the EuroQol Asian Academy meeting proceedings. We will review articles on the following topics: 1) reporting HRQoL measures that were developed in a Chinese cultural setting; 2) constructing a conceptual framework of health or health-related quality of life specifically to Chinese culture; 3) conducting qualitative interviews to explore Chinese people’s understandings of health. Second, two bilingual reviewers will screen titles, abstracts and full-text articles independently. Third, data will be extracted and synthesised. Fourth, a list of potentially important health dimensions that are not covered by EQ-5D will be identified.Zhuxin MaoDescriptive Systems21240Ongoing20212022
234-VSValuing Health-Related Quality of Life: Developing an EQ-5D-5L Value Set for GhanaThe Ministry of Health and other stakeholders of health have initiated the process of requiring Health Technology Assessment (HTA) as a formal process for priority setting in Ghana. Progress made include an HTA pilot study, and an HTA technical working group. The development of a Ghanaian EQ-5D-5L value set will provide a more appropriate basis for conducting HTA for decision-making in the Ghanaian health system, and support the process of institutionalizing HTA as a formal requirement. This allow for the EQ-5D-5L to be used in HTA related projects in Ghana presently and most importantly in the future when HTA is fully institutionalised. This study aims to provide the first EQ-5D-5L value set for Ghana and collect qualitative data on the ‘worse than death’ hypothesis to explore the contextual 'value of life' in Ghana. We will collect data from a sample of the Ghanaian population by adopting the methodological approach used for the Australian study that valued EQ-5D-5L using the adapted version of the standard EQ-VT protocol (EQVT-Lite). This approach has been demonstrated to produced comparable values in line with other value sets using the standard approach. To estimate utility values, we will model the TTO and DCE data separately, and also use ‘hybrid’ models. We will conduct an in-depth qualitative interview with 10% of the sample to explore the contextual, religious and cultural influence of Ghanaians with regards to ‘worse than death’ health state, and how Ghanaians value life. These data will be analysed using an inductive thematic analytical approach.Brendan MulhernValuation55779Ongoing20212023
299-RAValidation of EQ-5D-5L in critical care (EuroQoL Working Groups Project Request for Proposal)Critical care research typically reports outcomes which are patient-important, specifically mortality and morbidity. However, healthcare providers, patients and families are not only interested in patient survival, but also health-related quality-of-life (HRQoL) before, during and after critical care. Prior critical care research has not routinely described patient-reported outcomes like HRQoL. Patient-reported health utility can be elicited using various indirect, generic preference-based value measures. The most common generic method is the EQ-5D-5L. It has advantages including accuracy at low utilities compared to other HRQoL tools, no licensing fee for non-commercial use, and a built-in visual analog scale (VAS) for self-rating a patient’s health status. It’s more user-friendly than other instruments (e.g. SF-36). EQ-5D-5L has proxy instruments in addition to patient-reported instruments. Measuring baseline HRQoL in critically ill patients can be used to predict potential downstream outcomes. Prior research has shown that low HRQoL prior to critical care admission is associated with a grim prognosis in terms of survival, and leads to deterioration in the HRQoL post-discharge (23–28). The EQ-5D-5L has not been previously validated in critical care. Patient-reported outcomes like HRQoL and quality-adjusted life-years (QALYs) are being increasing used and recognized as important endpoints to measure. With more patients surviving their critical illness, documenting ICU survivors HRQoL (patient-reported psycho-social and physical functional domains) becomes important in its own right. Therefore, the objectives of this proposal are to: (1) validate the EQ-5D-5L-5L in the critical care setting; and (2) compare proxy and self-complete versions of the EQ-5D-5L-5L in critically ill patients.Vincent LauDescriptive Systems, Valuation, Populations and Health Systems39680Ongoing20212023
370-RAA research programme to support and strengthen the use of EQ-5D instruments in China**Background** China is a strategically important country for EQ-5D instruments in terms of both market size, potential growth in use of HRQoL, and influence across the region. There has already been some use of EQ-5D instruments in China, but there are also some important impediments (e.g., high ceiling effects) and emerging threats (e.g., Chinese-developed HRQoL measures and SF-6Dv2 Chinese value sets) which may limit more widespread use of EQ-5D instruments. **Aim** The overall aim of this programme of work is to support the use of the EQ-5D instruments, to ensure their relevance and maximise their usefulness, and to engage their large-scale users in China. **Data** Our primary data source is Nanjing CDC adult survey which used 3L and SF-12v2 in 2020, and switched to 5L in 2021, covering 30,000 respondents each year. Working in collaboration with CDC, we will add methodological components to the survey in 2021. We will also use the nationally representative National Health Services Surveys (NHSS) data which uses 3L, and collect existing datasets containing both 3L and 5L from Chinese researchers for secondary analysis. **Studies** The programme of work consists five studies: (1) Exploring the effects of data collection mode and processes on the ceiling effects of 3L; (2) Comparing 3L and 5L in general population; (3) Validation of the international 3L-5L crosswalk algorithms; (4) Comparing EQ-5D and SF-6Dv2 in general population; (5) Developing a good practice guideline. We will use both qualitative and quantitative methods as described in detail in the proposal.Nancy DevlinDescriptive Systems, Populations and Health Systems143900Ongoing20212024
389-RAA head-to-head comparison of measurement properties of EQ-5D-Y-3L and EQ-5D-Y-5L in children and adolescents with heart diseasesCongenital heart disease (CHD) is one of the most common birth defects globally, with a long-term impact on health-related quality of life (HRQoL). The EQ-5D-Y was adapted from EQ-5D through a rigorous adaptation process with inputs from children and adolescences. This approach facilitates the comparison of HRQoL in adults and younger populations, using the same domains of HRQoL. While the first version of EQ-5D-Y uses a 3-level response (EQ-5D-Y-3L), recently, a 5-level EQ-5D-Y (EQ-5D-Y-5L) was developed to reduce the ceiling effects and to enhance sensitivity. But the evidence on comparing the measurement properties of these two instruments is limited with mixed findings. Furthermore, there are no reported studies for evaluating the measurement properties of either of these two instruments in pediatric patients with CHD. We aim head-to-head comparison of measurement properties of EQ-5D-Y-3L and EQ-5D-Y-5L in pediatric patients (age 8-18 years) with CHD. The study will recruit a total of 200 patients for psychometric validation from pediatric cardiology outpatient clinics of two major public hospitals in Singapore. The patients will self-administer EQ-5D-Y-3L, EQ-5D-Y-5L, along with non-preference-based generic (PedsQL Generic Core) and disease-specific (PedsQL Cardiac Module) HRQoL instruments at baseline, 2 weeks (for test-rest reliability) and 6 months (for responsiveness). Measurement properties (discriminative ability, test-retest reliability, convergent validity, ease of use, and responsiveness) will be evaluated and compared between EQ-5D-Y-3L and EQ-5D-Y-5L. The findings of our study will guide the selection of a preference-based HRQoL instrument for clinical research and reimbursement policy decision-making in children and adolescents with CHD.Mihir GandhiYouth25000Ongoing20212023
377-VSEQ-5D-5L Slovenia national value setThis study aims to develop EQ-5D-5L value set for Slovenia from a common (quota) sample. 1000 respondents, representative of the Slovenian general population (age, gender and regions) will complete composite time trade-off tasks and DCE tasks. The computer-assisted personal interviews (EQ-VT) will be conducted, the interviewers will be recruited among university students from various regions, each interviewer will conduct around 70 interviews, depending on the sample size in regions. It is assumed that 12 interviewers will need to be recruited and educated. The interviewer script with the provided instructions on standards and goals interviewers should achieve and text suggestions for what to say, will be prepared where missing. Regarding the selection of health states, the protocol for EQ-5D-5L valuation studies will be followed: 86 health states will be included in the design of the cTTO task, divided in to 10 cTTO blocks. The DCE design will include196 health states distributed over 28 blocks of seven pairs of states. Blocks as well as the order of health states within the blocks will be presented in a random order both for the cTTO and DCE. The latest QC protocol will be followed. The reports for regular meetings with EuroQol VT support group will be based on the QC reports used in EQ-5D-Y study conducted in Slovenia in 2020. The QC criteria defined in the protocol will be used. The analysis will result in the EQ-5D-5L value set, which will be compared to the other value sets in the CEE region and Europe.Valentina Prevolnik RupelValuation24620Ongoing20212022
304-PHDThe sequential relief of child health problems: a preference path elicited by a kaizen taskIn a paired comparison for child health valuation, respondents are commonly asked to choose between two children who are suffering (e.g., Child A versus Child B). To make this choice, respondents face a heartbreaking dilemma, causing some to feel guilty for abandoning one child, to question their role as accomplices, and to challenge the merits of the investigation. Instead of choosing between children, an alternative task may show a single profile and ask the respondent to improve the child’s health by sequentially relieving problems. Although such evidence is statistically identical to its multinomial counterpart, this kaizen task (i.e., the Japanese term for continuous improvement) is more realistic, similar to clinical decision-making that personalizes treatment to best meet a patient’s changing needs. In the proposed PhD project, we hypothesize that responses to paired comparisons and kaizen tasks agree but that the kaizen task will be preferred by respondents and will render more precise values through a shorter survey, regardless of the child’s age and problem duration. Our aims include: 1. To develop and implement a kaizen task for child health valuation and compare the preference evidence, response times, and respondent feedback between paired comparisons and kaizen tasks. Craig et al. (2015) conducted the first EQ-5D-Y valuation and published their value set in *Health Economics* (attached). Based on its paired comparison results and subsequent work, Drs. Craig, Rand, and Hartman developed a kaizen task for EQ-5D-5L valuation (under review). Currently, Drs. Craig and Rivero-Arias are collecting preference evidence from paired comparisons and kaizen tasks in their study of children’s surgery outcomes. This first aim extends this line of independent research through primary data collection using LimeSurvey in collaboration with the EuroQol Research Foundation. 2. To assess the effects of child age and problem duration on the preference evidence and their agreement. The 2015 study found that the value of EQ-5D-Y outcomes varies by child age (9 and 11) and problem duration (1 and 2 years) due to the potential implications for child development. Likewise, it is possible that response time, respondent feedback, and task agreement vary by child age and problem duration. For this second aim, a second wave of primary data collection will test for these hypothesized differential effects. 3. To examine individual heterogeneity in the preference evidence and their discordance by task. Respondents may differ in their preferences as well as the agreement between their paired comparison and kaizen task responses. If heterogeneity or discordance exists, it is important to understand their source, which may lead to further innovations. Using the data collected, we will conduct latent-class analyses to test for heterogeneity in the preference evidence and discordance by task. If funded, this three-year PhD project will begin with a systematic literature review (Paper #1) followed by two waves of primary data collection (Papers #2 and #3) and latent-class analyses of respondent heterogeneity (Paper #4). Each aim will be accomplished in collaboration with the EuroQol Office using LimeSurvey and R code, which will facilitate dissemination and integration for future research.Maksat JumamyradovValuation, Education and Outreach102150Ongoing20222025
316-EOThe EQ-5D-5L value set for Italy – Dissemination and OutreachFollowing the successful completion of the Italian EQ-5D-5L valuation study (Finch et al. Social Science & Medicine 2022), we successfully organized and performed a workshop targeting different stakeholder groups (clinical scientific associations, patient advocacy groups, reimbursement and procurement organizations, healthcare industry), aimed at promoting the new tariff for the EQ-5D-5L in Italy and disseminating the findings of the project. The dissemination of the Italian EQ-5D-5L Valuation study findings has been complemented by the preparation and release of a video tutorial to illustrate the use of the EQ-5D-5L tariffs estimated based on preferences of a representative sample of the Italian population.Oriana CianiValuation, Education and Outreach17440Completed20222022
309-RAIllustrating the empirical impact of applying different value sets: easy-to-read graphs and tables for stakeholdersAim: The objective is to perform by-country comparison of 3L and 5L value sets and cross-walks for all countries where -3L and -5L value sets are available. The output is a graphical and tabular representation of the performance and discriminatory ability of the different value sets across a wide range of severity, along with a standardized set of value set properties. Methods: First, a set of properties of each value set will be collated, including modality of kernel density plots, percentage of health states with values less than 0, mean level transitions, and range of scale. Second, a weighted resampling-based method applied to empirical data with parallel self-reports of -3L and -5L health states will be used to simulate the sensitiveness of each value set to small changes in health, along a wide spectrum of severity. Discriminative ability will be assessed using F-ratios. Graphical representation will be used to illustrate the impact of applying different descriptive systems and value sets. Significance: The EuroQol group advocates for the use of the EQ-5D-5L descriptive system, but recommendations by HTA agencies lag behind. There remains significant confusion among users and industry about the implications and advantages of moving to the -5L, as well as the implications of using a -3L versus -5L value set. This work is intended to expand the scope of work and methods used to examine US value sets to understand and illustrate the implications of choosing between EQ-5D measures and value sets in other countries around the world.Kim RandValuation31400Ongoing20212022
295-RAThe feasibility, acceptability and validity of the EQ-HWB for use in a hard-to-reach population of carers of children experiencing adversity.Background Most Health-related Quality-of-Life (HRQoL) measures focus on health rather than broader concepts of well-being, making it difficult to measure cost-effectiveness in the Social Care Sector (such as aged care and out-of-home care for children). The EuroQol Health and Wellbeing measure (EQ-HWB (25 items), EQ-HWB-S (nine items)) was designed to capture a broader range of important outcomes that include the effects of health conditions on carers, and may be appropriate to measure quality of life in this sector; however, the measure needs to be tested in the field. We have a unique opportunity to analyse data using the EQ-HWB-S in a hard to reach population of families where a child is experiencing adversity. Aims (Aim 1) To determine feasibility, acceptability, validity and sensitivity to change of the experimental version of the EQ-HWB-S using quantitative analysis. (Aim 2) To determine face and content validity and reliability of the instrument to measure important and relevant quality of life aspects in the context of adversity through qualitative interviews. Methods (Aim 1) EQ-HWB-S data from 100 Australian families over 2 collection cycles, 6 months apart, will be analysed to determine feasibility, acceptability, known groups validity and sensitivity to change. The EQ-HWB-S will be administered through the Childhood Adversity Centre for Research Excellence (CRE), thus making this study highly cost-effective. (Aim 2) Face and content validity will be established through qualitative semi-structured interviews with a subset of parents/carers of children experiencing adversity. Test-retest reliability will be conducted at the start and end of the interviews.Cate BaileyEQ-HWB61758Ongoing20212022
310-RACANDOUR study: Using EQ-5D-5L to assess the impact of global use of COVID-19 vaccines on health-related quality of life**Aims**: To understand how COVID-19 vaccination has impacted various aspects of health using the EQ-5D-5L instrument and related measures. **Methods**: We will undertake the third follow-up of our global CANDOUR survey study, which is a unique multi-country internet-based survey of around 20,000 individuals from 17 countries (~1100 per country). Respondents were chosen to be geographically diverse (i.e. from all continents) and represent roughly half the world’s population: Australia, Brazil, Canada, Chile, China, Colombia, France, India, Italy, Ghana, Kenya, Russia, Spain, Tanzania, Uganda, United Kingdom, United States. We have collected the EQ-5D-5L in two survey waves (Dec 2020 and May 2021) and a wide variety of other information including health comorbidities as well as wider information (e.g. household income).This study will collect an additional wave of the CANDOUR survey in September 2021, supplementing the two waves already collected. The study’s longitudinal design will provide a unique opportunity to obtain repeated measures on survey participants to better understand the impacts of the COVID-19 pandemic and vaccination programmes on self-assessed health across a wide range of countries. **Outcomes:** We intend to use longitudinal methods (e.g. fixed-effects models) to measure the impact of vaccination on EQ-5D-5L. We will also be able to stratify the effects by a large number of covariates to identify heterogeneous treatment effects. The aim will be to produce unique comprehensive global estimates of the overall impacts of both the pandemic and vaccination on health and to undertake methodological work to support use of the EQ-5D-5L to measure global health.Professor Philip ClarkePopulations and Health Systems99518Ongoing20212023
289-RACOVID-19 and EQ-5D-5L health state valuationBackground: We investigate whether and how the COVID-19 pandemic has influenced general population health state values. Changes would have important implications, as general population values are used in health resource allocation. Data: In Spring 2020, participants in a UK general population survey rated two EQ-5D-5L states, 11111 and 55555, as well as dead, using a visual analogue scale (VAS) from 100=best imaginable health to 0=worst imaginable health. Participants answered questions about their pandemic experiences, including COVID-19’s effect on their health and quality of life, and their subjective risk/worry about infection. Analysis: VAS ratings for 55555 were transformed to the full health=1, dead=0 scale. Tobit models were used to analyse VAS responses, as well as multinomial propensity score matching (MNPS) to create samples balanced according to participant characteristics. Results: Of 3,021 respondents, 2,599 were used for analysis. There were statistically significant, but complex associations between experiences of COVID-19 and VAS ratings. For example, in the MNPS analysis, greater subjective risk of infection implied higher VAS ratings for dead, yet worry about infection implied lower ratings. People whose health was affected by COVID-19 rated 55555 higher, whether the effect on health was positive or negative. Conclusion: The results complement previous findings that COVID-19 may have impacted EQ-5D-5L health state valuation, and different aspects of the pandemic have different effects.Edward WebbValuation24950Completed20212022
354-RAContent and face validity of the EQ-HWB and EQ-HWB-S in a sample of patients, members of the general public and social care users in ItalyThe EuroQol Group recently developed two new instruments intended to measure health and social care related quality of life. The instruments are referred to as EQ Health and Wellbeing instrument (EQ-HWB) and EQ-HWB-S. The EQ-HWB and EQ-HWB-S are intended to capture a broader range of symptoms and functioning aspects which may be relevant to general public members, patients, their families, social care users and carers. Further research is needed to test and validate the instruments. As part of the Italian valuation study, the EQ-HWB has been collected alongside the EQ-5D. As an Italian translation was not available, a process of forward and backward translation, with reconciliation of the translated versions, was employed. To decide for the optimal translation, 20 patients, members of the general public and social care users with varying socio-demographics were interviewed using a think-aloud concurrent protocol. The generated data can be used for assessing the content and face validity of the EQ-HWB and EQ-HWB-S. This will be done by analyzing the data using framework analysis.Camilla FalivenaDescriptive Systems, EQ-HWB24760Ongoing20212022
348-PHDThe social value of avoiding poor health states in childrenAims: This PhD will contribute to our understanding of the methods used in the valuation of child health states, furthering knowledge on the issues of anchoring child preference-based measures onto the QALY scale and weighting child and adult QALY gains. The work will explore three related questions: 1. What drives people’s preferences towards valuing health gains differently when they are experienced by different age groups? 2. Do members of the public have a different relative value for quality of life compared to length of life for children of different ages, and between children and adults. If so, what are the implications for the valuation of the EQ-5D-Y? 3. Using example case studies, what is the impact of applying age weights to QALY gains on incremental cost effectiveness ratios and on the uncertainty of findings? Methods: This PhD will use both qualitative and quantitative analysis. Study 1. ** Qualitative analysis of Person Trade Off discussions ** This study will apply framework analysis to focus group data previously collected within the QUOKKA project. During the focus groups respondents will deliberate and discuss their responses to trade off questions relating to the relative value of health gains to patients of different ages. This will complement the QUOKKA projects exploration of the social value of a child relative to an adult QALY. Study 2. ** Exploring the willingness to trade length versus quality of life for children ** This study explores whether a child QALY derived from EQ-5D-Y valuation protocol can be compared to an adult QALY derived from EQ-VT protocol. It will seek to value states described by the EQ-5D-Y which are considered equivalent when experienced by different age groups. These states will be derived through focus groups, and the equivalence of the quality of life decrement from the states tested empirically based on simple ranking and sorting tasks. These health states will then be valued by a representative sample of the public using an interview and the compositive Time Trade Off component of the EQ-5D-Y international valuation protocol with variability across the age of the hypothetical person experiencing ill-health. The study will use mixed methods and seek to understand any differences in willingness to trade and consequent utility values. Study 3. ** Exploring the impact of child QALY weights on cost-effectiveness models*** Understanding the magnitude of potential differences in QALY weights on decision making will help highlight important research gaps. Study 3 will develop a number of cases studies to explore the impact of applying age weights to QALY gains within the context of decision models including discrete event simulation models and Markov cohort models. The project will include implications for threshold values and explore the impact of social value QALY weights upon estimates of opportunity cost. This PhD will be a complement to the QUOKKA study led by Prof Devlin but produces completely new outputs.Tessa PeasgoodValuation, Youth117903Ongoing20222025
319-RAAn examination of the psychometric performance of the EQ-5D in haemophilia: A systematic literature reviewPsychometric methods may be employed to inform assessment of the appropriateness of an instrument for use within a particular population. The aim of this review is to examine the appropriateness of the EQ-5D for measuring health-related quality of life in people with haemophilia by examining all published evidence relating to the psychometric performance of the EQ-5D compared with alternative preference based measures and non-preference based measures. A systematic search will be undertaken to identify studies reporting the psychometric performance of the EQ-5D and alternative measures in people with haemophilia. Data will be extracted to assess known-group validity, convergent validity, responsiveness, reliability, acceptability, and feasibility. Data will be extracted separately for the dimensions and utility index where this is reported.Antony MartinDescriptive Systems25000Ongoing20212021
335-RAAn extensive pilot phase in the Egyptian EQ-5D-5L valuation study - Lessons learnedBackground: In EQ-5D-5Lvaluation studies, interviewers training before and during data collection is undertaken to increase the quality of the collected data and reduce interviewer effects. Each interviewer continues to conduct pilot interviews until acceptable protocol compliance is achieved and interviewer effects are minimized as indicated by the QC tool. In some studies, only the first 5 or 10 interviews are excluded then interviewers start actual interviews. Since the pilot phase is time consuming and may increase the cost of the valuation studies, it is not clear if an extensive pilot phase can standardize the performance of the interviewers and improve the quality of the collected data. Aim: To investigate the effect of an extensive pilot phase on improving the face validity, quality of the collected data, protocol compliance, reducing interviewer effect and clustering of responses. Methods: This study will further explore the data collected in the Egyptian EQ-5D-5L valuation study, where 1303 interviews were conducted during the period between July 2019 to March 2020 by twelve interviewers. Among the collected data, 298 interviews were pilot and 1005 interviews were actual, three interviewers were excluded from the study along with the interviews they had conducted (N=113), thus leaving 216 pilot and 974 actual interviews. The study will investigate the effect of the pilot phase on improving protocol compliance, reducing the interviewers’ effect and clustering of responses. In addition, it will make use and compare the QC report indicators during the pilot phase and the actual data collection.Samar FaridValuation14400Completed20212021
317-RAInvestigating the aspects of HRQoL covered by the descriptive system and the added value of the respiratory bolt-ons (EQ-5D-5L+R): breathing problem and limitations in physical activities due to shortness of breath among patients suffering from asthma in Ethiopia: A mixed method studyIntroduction: Asthma is one of the major non-communicable diseases, with its recurrence and severity of the symptoms results affecting the health-related quality of life(HRQoL) of the patient. In the EQ-5D, limitation of physical activity is somewhat captured by the usual activity and mobility item, while breathing problem is a very important aspect of physical discomfort, other aspects such as shortness of breath, coughing, and wheezing are not explicitly mentioned. This research aims to address the additional advantage of using the respiratory bolt-on(EQ-5D-5L+R) in people with asthma. Aim: To investigate how well the aspects of HRQoL are captured by the EQ-5D descriptive system and the additional advantage of using the respiratory bolt-on(EQ-5D-5L+R) in people with asthma in Ethiopia. Method: A mixed study method will be used. First, an in-depth qualitative interviews with 30 patients will be undertaken to investigate how well the health aspects important for people with asthma are captured by the EQ-5D descriptive system and the additional advantage of using the EQ-5D-5L+R questionnaire. Both EQ-5D-5L and EQ-5D-5L+R will be administered to participants. The qualitative interviews will be recorded using an audio recording device, transcribed, and translated. Findings from the interviews will be discussed in focus group discussions with participants who interviewed to make sure their concern is captured, to feed off each other’s ideas, to get useful information that individual interviews does not provide. Second, psychometric properties between EQ-5D-5L and EQ-5D-5L+R will be compared in terms of distributional effect, convergent validity, informativity, and explanatory power among 500 asthmatic patients.Goitom MolalignDescriptive Systems25720Ongoing20212022
277-VSValuing health-related quality of life: an EQ-5D-5L value set for AustriaIn recent years, several countries adopted mandatory health economic guidelines for public drug reimbursement decisions to facilitate efficient resource allocation. These decisions are supported by health technology assessment (HTA) bodies. In June 2020, the Austrian Ministry of Health published recommendations for further actions in the establishment of systematic HTA in the Austrian healthcare system. Despite the increasing interest in using PROs also in Austria, nationally valid tools and methods are lacking for the measurement and valuation of HRQoL. This project aims to develop the currently lacking national standards for HRQoL valuation for the EQ-5D-5L questionnaire which is the most widely used generic HRQoL instrument in the context of evaluative studies and HTAs. Following training by the EuroQol Group, 1100 online interviews will be conducted to derive preference-based valuation for the EQ-5D-5L for the Austrian general population. Study sample will be stratified according to population size, age, gender and educational level distributions in each federal state. The preference elicitation techniques of discrete choice experiment (DCE) and time-trade-off (TTO) will be used as required by the EuroQoL group. Each interview will have 10 TTO and 7 DCE tasks and last 1 hour. Deidentified data will be collected according to strictest data security regulations using the standard software package EQ-VT provided by the EuroQol Group. For the derivation of the final valuation tariff, data will be analysed using a hybrid regression model which combines responses obtained through the TTO and DCE tasks.Judit SimonValuation0Ongoing2022
355-RAFeasibility, reliability and validity and of the EQ-5D-Y (3L&5L) in children and adolescents with ADHDBackground and purpose: Facing the lack of validity studies of self-reported preference measures of health-related quality of life, especially in mental-health conditions and social behavior disorders, the purpose is to analyze the feasibility, reliability and validity of EQ-5D-Ys (-3L and -5L) in children and adolescents with attention deficit hyperactivity disorder (ADHD). Methods: We will perform a cross-sectional design. The Spanish online versions of EQ-5D-Ys, KiDSCREEN-10, and Strengths and Difficulties Questionnaire, including the subscale of hyperactivity, will be collected using a web-based questionnaire survey in 300 youngs between 8 and 17 years old recruited by National association of ADHD. A test-retest will be performed in a subsample (n=60). The feasibility will be tested by the proportions of inadequate responses (missing responses or inconsistent) and time to complete. The reproducibility, distribution, psychometrics and Intra-Class coefficient of the EQ-5D-Ys summary score, EQ-VAS and KS-10 index will be calculated. The construct validity will be tested by: 1) comparing known groups (children/adolescents, sex, severity of hyperactivity, severity of mental health problems, family composition, the time of physical activity and the time of use screens); 2) Convergent and discriminant validity will be mainly assessed by Spearman correlation matrixes and a principal component analysis with varimax rotation to identify the comparative relevance of factors underlying the constructs of pairwise instruments. Expected implications: EuroQol Group could lead the first validation of self-reported preference measure in ADHD, increase the evidence to support the use of EQ-5D-Ys in condition-specific populations, and compare the perform of -3L and -5L.Narcis GusiYouth24950Ongoing20212022
315-RAThe relation between the EQ-5D-5L and fatigue and cognition problems: does the EQ-5D-5L capture persistent symptoms of infectious disease?The EQ-5D-5L might lack sensitivity for disease specific health complaints. This cross-sectional study analyzed whether fatigue and cognition problems are captured by the EQ-5D-5L in a Q-fever patient population with persistent fatigue/cognitive problems. Out of 432 Q-fever patients, 373 (86%) reported severe fatigue and 387 (90%) reported cognition problems. EQ-5D-5L utility and EQ VAS scores of Q fever patients reporting severe fatigue/cognitive problems were significantly lower compared to patients reporting less severe fatigue/cognitive problems. Fatigue and cognition problems in Q-fever patients were partially captured by the EQ-5D-5L dimensions. The addition of fatigue to the EQ-5D-5L slightly improved explained variance for the EQ VAS. This potentially also accounts for patients who experience other infectious disease sequelae that are characterised by fatigue and cognitive problems, such as COVID-19.Juanita HaagsmaDescriptive Systems16080Completed20212022
358-RA{(RRM + RUM) + DCE} *EQ-5D-5L = PreferenceBackground: Traditionally, researchers extract latent coefficients for EQ-5D-5L using the conventional random utility maximization (RUM). Recently, a new model, the Random Regret Minimization (RRM or µRRM) postulates that people making choices by minimizing the anticipated regret. Specifically, the µRRM models the attribute-specific scale factor (µ), which determines whether each attribute shall be processed using RUM or RRM. To date, there are limited applications of µRRM in healthcare settings. Specifically, decision rules on the size of the scale factor (µ) have remained arbitrary. Study Aims: Our study aims are 1) to apply the RRM modelling approach using both the EQ-5D-5L in a discrete choice experiment (DCE) and, 2) to ascertain the decision rules for the scale factor (µ) in different survival/risk scenarios. Methods: We plan to conduct a DCE study on the basis of 3 hypothetical survival/risk scenarios. Health profiles are defined by the EQ-5D-5L health states. Amazon Mechanical Turk (MTurk) will be used for data collection. Included respondents will be randomized into three surveys, each with one of the following risk scenarios: 1) living with chronic conditions and a life expectancy of 10 years, 2) living with acute conditions with a life expectancy of 6 months, and 3) living with confirmed acute infectious disease such as COVID-19. Choices by each group will be analyzed using RUM and µRRM separately and, head-to-head comparisons will be made.Ning Yan GuValuation24820Ongoing20212023
343-RAThe Psychometric Properties of the EQ-5D-5L among Ethiopian Cervical Cancer Patients: A Longitudinal StudyIntroduction: The EuroQol five-dimensional five-level (EQ-5D-5L) instrument is widely used as a generic patient-reported outcome measure. The psychometric properties of the instrument have been proven in general population and multiple disease conditions. However, there is limited evidence about validity, feasibility, responsiveness, minimal clinically important difference (MCID) in cervical cancer patients, especially in low-income settings. Therefore, examining the psychometric properties of the EQ-5D-5L in patients with cervical cancer in Ethiopia is important to conduct future outcome studies in Ethiopia and other low-income settings. Objective: To examine the validity, feasibility, responsiveness, MCID, and usefulness of the EQ-5D-5L in patients with cervical cancer in Ethiopia. Methods: An interviewer-administered, face-to-face longitudinal study will be conducted between July 2021 and May 2022 among 422 patients with cervical cancer at the oncology center in Ethiopia. Participants who fulfil eligibility criteria will be interviewed using EQ-5D-5L and cervical cancer-specific Quality of Life module of the European Organization for Research and Treatment of Cancer (EORTC QLQ-CX24) before initiation and three months post-treatment. Discriminant validity will be evaluated based on specific groups that are anticipated to have different levels of utility due to their performance status. Criterion validity will be estimated using Spearman correlation coefficient. Test-retest reliability will be measured by intraclass correlation coefficients (ICC). Responsiveness will be analyzed by effect size, standardized response mean (SRM), and criterion responsiveness. The MCID will be estimated using distribution-based, anchor-based, and instrument-defined methods. The percentage of patients exceeding the MCID will be reported. All statistical analysis will be performed using R software.Girma Tekle GebremariamDescriptive Systems23387Ongoing20212022
302-EOIntroduction to Latent Classes in Health Valuation: A Workshop ProposalSupported by the EuroQol Research Foundation, all researchers were invited to participate in an introductory workshop on choice and latent-class analysis in health valuation. Taught by Benjamin M. Craig and Suzana Karim, this free online workshop had three group sessions, two tutoring sessions, and two hands-on exercises using the R programming language and paired comparison data. Any researcher could register for any session for free. Attendees who participated in all sessions and complete both exercises will receive a EuroQol Workshop Certificate shortly. Session 1 (10 June): Applied Choice Analysis in Health Valuation The first session covered the conditional logit and heteroskedastic logit models. Dr. Craig presented the slides and led the discussion (40 minutes followed by a 10-minute break), and Ms. Karim introduced the Dutch EQ-5D-Y paired comparison data and its R code (40 minutes). After the session, attendees were invited to complete an optional hands-on exercise in R and discuss their results during a follow-up tutoring session (30 minutes), allowing for personalized feedback on how to estimate an EQ-5D value set (pits scaled). Session 2 (17 June): Latent Class Analysis in Health Valuation The second session extended the heteroskedastic logit model to its latent class counterpart, introducing grade-of-memberships, model selection, and interpretation. After the session, attendees were invited to complete a second exercise in R and discuss their results during a follow-up tutoring session (30 minutes), allowing for personalized feedback on how to estimate an EQ-5D value set (pits scaled) for each latent class. Session 3 (24 June): Applications and Advanced Topics in Health Valuation In the third session, Dr. Craig covered applications and advanced topics, including time preferences, response behaviors, scale-adjusted latent class (SALC) analyses, and random parameters. After the break, Ms. Karim briefly introduced three examples from her dissertation (Dutch BWS, Peru EQ-5D-5L, and US EQ-5D-5L). They concluded this session with a discussion on econometric extensions and survey methods in health valuation. For Whom Anyone interested in health valuation, who understands the basics of the EQ-5D descriptive system and had estimated a conditional logit. Some knowledge of R and R studio was encouraged but not required. Stable internet access was required. Outcome This introductory workshop was designed to enhance the technical capacity of early career researchers (ECRs). Gaining the capacity to estimate EQ-5D value sets using choice data also prepares ECRs for advanced topics (e.g., time preferences).Benjamin CraigEducation and Outreach22540Completed20222022
349-RAPsychometric properties of the EQ-5D in rare ataxia diseases (EQ-5D-ATAX)**Background:** The recent NICE "method of health technology evaluation" consultation demonstrated that there is only limited or no evidence of validity and responsiveness of the EQ-5D in rare diseases. Spinocerebellar ataxia (SCA) and Friedreich ataxia (FA) are rare (2-4 cases per 100.000 people) autosomal, inherent, and neurodegenerative diseases, characterized by muscle weakness, sensory loss, imbalance, poor coordination, and limited cognitive and mental health. However, evidence about the psychometric performance of health-related quality of life measures, like the EQ-5D, is currently lacking. **Aim of research:** To assess the acceptability, validity and responsiveness of the EQ-5D-3L in patients with ataxia diseases (SCA and FA). **Proposed methods:** The analysis will be on data of different prospective, longitudinal observational ataxia cohort studies (European Friedreich Ataxia Registry (EFACTS), European Spinocerebellar Ataxia Type 3/Machado-Joseph Disease Initiative (ESMI); EUROSCA, European Spinocerebellar Ataxia Registry) carried out at up to 17 European study centers with n=799 and n=605 patients with SCA and FA, respectively. The EQ-5D-3L and further socio-demographic (age, sex) and clinical variables (ataxia severity, non-ataxia symptoms, functional deficits, activities of daily living, comorbidity) were assessed via interviews at baseline and two years later (follow-up). The EQ-5D-3L will be evaluated in terms of acceptability, ceiling effects, convergent and discriminative validity (construct validity), and responsiveness. **Benefits to the group:** The study will determine if the EQ-5D performed well in a rare disease that tremendously affects patients' physical and mental health, adding new evidence about the acceptability, validity and responsiveness to the current paucity of knowledge in rare ataxia diseases.Bernhard MichalowskyDescriptive Systems24240Ongoing20212023
321-VSA multi-country EQ-5D-Y valuation study in Asia (Resubmission of project 154-2020RA)There is a growing interest and need for economic evaluation of paediatrics products and services in Asia. However, there is currently a lack of health-related quality of life instruments for the valuation of paediatric health outcomes in the region. EQ-5D-Y has exhibited good psychometric properties in several paediatric groups in Asia, and a valuation protocol has been developed for establishing country-specific EQ-5D-Y value sets. The PIs of the EQ-5D-5L valuation study in Malaysia, Singapore, Taiwan, and Vietnam came together to propose a multi-country EQ-5D-Y valuation study. The primary aim is to concurrently establish the EQ-5D-Y value set for the four countries using the recently published EQ-5D-Y valuation protocol. We will also collect additional data to address some of the methodological issues in EQ-5D-Y valuation. Specifically, we aim to: a) use a pilot DCE study to examine whether the child's age used in the valuation tasks will have a significant effect on valuation; b) use a larger cTTO design to explore what is the best strategy for modeling cTTO and DCE data. This project will make available four country-specific EQ-5D-Y value sets for economic evaluation of paediatric products in South-East Asia. It may also contribute to the future development of the EQ-5D-Y valuation protocol by providing new evidence from a culturally different region. We are aware of the current COVID-19 pandemic situation. We plan to start data collection in April 2022, if this project is funded.Nan LuoValuation, Youth147810Ongoing20222023
357-RAA comparative investigation of inequality measures for EQ-5D outcomesBackground: In a recent study we explored consistency among a set of commonly used inequality measures by applying them to index values, EQ VAS scores and level sum scores based on a large multi-country EQ-5D-3L data set. Within-outcome correlation coefficients were very high for EQ VAS and level sum data but lower for index values. The Entropy Index was particularly inconsistent when compared to the other measures when index values were used as the basis. This inconsistency was not observed when level-sum and EQ VAS values were used as the bases. Aims: In this study we propose to carry out a more in-depth analysis and investigation of the different inequality measures and corresponding outcomes. We have identified 6 key issues on which this study can potentially improve the understanding and status of EQ-5D as a basis for evaluating inequality (health-and beyond-health). Methods: We plan to revisit the Gallup dataset with a more focused/detailed investigation into 6 key issues concerning EQ-5D inequality for which this dataset is particularly well-suited. Inequality results for each country will be re-examined using ICC analyses. EQ-5D dimensions/levels will be added to the analysis. Approaches to data visualization of EQ-5D inequality for a large data set will be evaluated to identify an optimal framework. Inequality analyses will be rerun using different value sets to test for value-set effects. Learnings from these investigations will be leveraged to produce a comprehensive analysis of EQ-5D inequality for this dataset which will extended to capture inequality along some unique Gallup Indices.Henry BaileyPopulations and Health Systems57200Ongoing20212022
367-RAPopulation Health Primary Data Collection ProjectThe intent of this proposal is to create research infrastructure for the EQ group, not a specific research project. As such, we have not specified concrete research questions at this time. Rather, our intent is to maximize the infrastructure that would support future research projects which have specific research objectives/questions. A roster of various projects/objectives/questions would be developed throughout the project, but particularly during year 1, by soliciting specific objectives and measures from the membership and EQ Working Groups. Developing a large research infrastructure such as this could (or perhaps, should) be guided by multiple research objectives, as the volume of data could support a variety of research interests. For the proposed project, the breadth of potential research topics would be characterized under two broad domains: 1. Population Health Status - Developing country-specific population norms - Disease burden in general population - Social determinants of health and population health inequities - Personal health behaviours/attitudes and health status - Health care system expectations/experience - Local (i.e., regional, within country) or International comparisons (for all of above) 2. Comparative performance/Validation of instruments - EQ-5D-5L - EQ-HWB - Selected ‘bolt-on’ dimensions Beyond these initial ideas, suggestions for specific research objectives/questions would be invited from EQ membership/Working Groups. These suggestions would then be discussed, debated and decided by the Project Team. Ideally suggested specific research questions would be supported by specific measures to be included in the surveys. Some allowance for country-specific data collection would be desirable.Jeff JohnsonPopulations and Health Systems1439446Ongoing20212024
361-RAMulti instrument comparison study extension: focus on EuroQol instruments, psychometric protocols, psychometric analysis and view to international replicationThis study proposes to leverage and extend the QUOKKA Multi-Instrument Comparison (MIC) study already funded by the Australian Government to compare a suite of paediatric generic and condition specific instruments in a sample of 4000 children and their carers with 1000 recruited through a tertiary Children’s Hospital. The QUOKKA-MIC will assess the acceptability, validity and responsiveness of measures but without a specific focus on EuroQol measures. This proposal aims to produce detailed analysis of EuroQol instruments to capitalise on the extensive data generated via QUOKKA-MIC, and to collect further data targeted at addressing questions of specific strategic and scientific interest to the EuroQol Group. Three areas of research activity are proposed: THEME A: The addition of new samples of children to the MIC to add more conditions, more serious conditions and conditions expected to rapidly change. Testing of proxy versus self-report across ages 6 to 10 years. Additional psychometric testing of EuroQol measures (TANDI, EQ-5D-Y for age 2-4 years, EQ-5D adult version for use in children >12 years, EQ-HWB for carers, EQ-5D 3L, and 5L) THEME B: Analysis of psychometric performance including the development and testing of an explorative crosswalk (or mapping function) between the EQ-5D-Y-3L and EQ-5D-Y-5L. To develop an international protocol for the psychometric testing of EuroQol instruments. THEME C: To develop a rationale, approach, protocol and study team for an international application of the QUOKKA-MIC.Kim DalzielYouth392247Ongoing20212023
352-RAExploring the potential of using EQ-5D-3L versus EQ-5D-5L to assess the value of a national large-scale health care improvement initiative in rheumatology in SwedenSeveral national quality improvement programs are under development in Sweden. In rheumatology, two person-centered care pathways have been developed. The Swedish Rheumatology Quality Register (SRQ) is closely linked to the care-pathways and data from the registry will be used to assess the consequences of the implementation. SRQ has been collecting EQ-5D-3L data since 2008 and EQ-5D has been discussed as a potential measure in the assessment of the pathways. Nevertheless, the responsiveness of 3L has been questioned and the registry is considering replacing it with EQ-5D-5L. The aims of the current study are: 1) to make recommendations regarding a replacement of EQ-5D-3L with the 5L version 2) to assess commensurability of 5L compared to 3L in this patient population and to develop a mapping function to enable comparability between newly collected 5L data and historical 3L data. Data will be collected in at least three rheumatology units in Sweden in patients with Rheumatoid Arthritis (RA) during the first year of disease (early RA) as well as in patients with established RA and Systemic Lupus Erythematosus (SLE) on one-two occasions. Cross-sectional 3L and 5L data will be compared in terms of distributional characteristics, known groups and convergent validity. Responsiveness of longitudinal 3L and 5L data will be assessed on descriptive data (inconsistencies, level moves, PCHC, index of superiority) and values (standardized response mean, standardized effect size, relative efficiency). A mapping function will be estimated, based on descriptive data and values. Findings will be validated in a separate cohort of SLE patients.Malin RegardtPopulations and Health Systems70870Ongoing20212025
342-RAUnderstanding the views of Canadians on valuing health for children and adolescentsMeasuring and valuing youth health for clinical and economic evaluations have been acknowledged as an important but challenging area. The publication of the international protocol for valuing the EQ-5D-Y-3L generates an interest in undertaking a valuation study for the youth version among a few countries including Canada. However, there are still a number of aspects in the current protocol that are open to debate. The issues include, but are not limited to, the target population for valuation study, and the perspective and framing of the valuation tasks. In Canada, limited research has been conducted on valuing youth health states and its implication to national health technology assessment and subsequent health care coverage policy making. This proposal is aimed to understand the views of Canadians on valuing health for children and adolescents.Feng XieValuation, Youth44925Ongoing20212022
285-PHDTo capitalize on the clinical value of EQ-5D - communicating and predicting patient outcomes on the individual patient levelTo support value-based health care, many hospitals are developing dashboards to collect patient reported outcomes (PROM), with future aspirations to use results to support decision making at the hospital and individual patient level. In the proposed PhD project, our objective is to demonstrate the value of the EQ-5D to support clinical decision making at the individual level, using data from smaller scale databases that can be realistically collected within a short time and within a health care organization. The proposed project is a combination of qualitative focus groups and interviews, survey methods and prediction modelling. Our aims are: 1. To develop validated displays for both the descriptive system as well as the VAS score that accurately and comprehensibly present EQ-5D-5L outcome data of a single patient, in that moment, over time, and present outcomes of different treatment options. In WP1, we will develop and test different display formats of the EQ-5D-5L descriptive system and VAS that will enable patients to compare their own health state at that moment with their own historical data and future health state after intervention(s). Subsequently, we will elicit patient, caregiver, health professional and population preferences for different display formats of the EQ-5D outcomes and test the impact of display format on comprehension of information that is presented in the displays. 2. To demonstrate responsiveness of the EQ-5D-5L to changes in health status and the ability of predictive models to offer reliable estimates of future EQ-5D health status on the individual level. In WP2, we will study the responsiveness of the EQ-5D to changes in functioning and health status as a result of intervention or disease progression (or improvement) on the individual level for four different case studies in osteoarthritis, juvenile idiopathic arthritis, chronic obstructive pulmonary disease and colorectal cancer. Then, we will determine the predictive power for both the descriptive system of the EQ-5D as well as the VAS using a risk based and an effect-based approach. Finally, we will determine how database size will impact reliability of predictions of future health status. 3. To demonstrate proof of concept by implementing optimal displays with the most promising predictive models for at least one case study. In WP3, we will implement findings in an existing web-based tool and test feasibility and usability in ten patients. If funded, this four year PhD project will result in one systematic literature review (Paper #1), two user centered design studies with different sample sizes and research methods (Paper #2 and #3) and a minimum of two papers on the responsiveness and predictive power of EQ-5D-5L in two different case studies (Paper #4 and #5). If the EQ-5D could prove value as an instrument for clinical decision support, this could increase the implementation potential of the EQ-5D in large scale databases as a result of patient effort directly benefitting the same patient.Janine van TilDescriptive Systems, Valuation, Populations and Health Systems185161Ongoing20212025
313-PHDMeasuring and valuing health for children and adolescents in EthiopiaEthiopia is the second most populated country and one of the fastest-growing economies in Africa. Population between the ages of 10 and 19 accounts for 42 percent of Ethiopia's total population. Its recent development of HTA system recommends the use of societal preferences for valuing health states and social health insurance. Therefore, there is a rising need for research on youth health measurement and valuation in this country. The goal of this PhD project is to improve health measurement and valuation for children and adolescents in Ethiopia. Specific objectives include to assess the equivalence in measurement properties between EQ-5D-Y-3L and -5L Amharic interview-assist (IA) and proxy modes of administration, to develop a value set for the EQ-5D-Y-3L using the international valuation protocol, and to investigate the difference in EQ-5D-Y values elicited from adults and adolescents. First, a systematic literature review will be undertaken on the topic of measurement equivalence across modes of administration. Then, the psychometric performance of IA and proxy of both Y-3L and Y-5L will be tested and compared using a cross-sectional study design among young population aged 5–15 years with a known health conditions (n = 600) and a control group of 'healthy' school children (n = 300). The instruments will be administered again 10 days after the first interview using the same mode of administration to a subgroup of participants from a general school sample and those with chronic disease conditions in order to evaluate test-retest reliability and responsiveness in those who have changed health conditions. Third, the valuation study will be conducted in accordance with the international EQ-5D-Y valuation protocol. The fourth part is a methodological add on study connected to the main valuation study: DCE and TTO values for EQ-5D-Y-3L health states will also be elicited from adolescents to enable comparison of adult proxy and adolescent self perspectives, which has not been done for TTO yet.Abraham GebregziabiherValuation, Youth115000Ongoing20212024
330-PHDEvaluation of the EQ-5D-Y as a child PROM in tertiary hospitals for high impact childhood conditionsIncreasingly clinicians, health service providers and patients see value in the use of patient reported outcome measures (PROMs) to inform clinical practice. As a short generic health measure, validated across a wide age range and with strong psychometric properties, the EQ-5D-Y is a promising tool for routine use as a clinical PROM for children. There is a paucity of research regarding the use of PROMs in children and the use of EQ-5D-Y as a clinical PROM. This project aims to assess the clinical utility of the EQ-5D-Y as a routine PROM in managing clinical care in a tertiary paediatric children’s hospital. A knowledge to action framework will inform a four stage, mixed methods approach: Project 1. **Systematic review** of the literature to understand how generic PROMs have been used to inform clinical decisions across the lifespan. Project 2. **Determine clinical utility of the EQ-5D-Y in a paediatric tertiary hospital** Using data from the QUOKKA- Multi Instrument Comparison Study for Children which involves 1500 families recruited via tertiary hospital setting with a range of conditions selected based on their high prevalence and high impact on quality of life of children. Clinical utility of the EQ-5D-Y will be assessed through analysis of selected outpatient clinic data to show feasibility and acceptability to patients and clinicians. Clinical trust in the data will be established through validity and responsiveness compared to established clinical endpoints. Project 3. **Understand patient and clinician perspectives** via qualitative semi-structured interviews and/or focus groups to understand their preferences on the use of PROMs in the clinical setting and to collaboratively design how PROMs could be feasibly incorporated and fedback into hospital outpatient clinical practice. The co-design will include visual formatting of the PROM information and ways of scoring of the EQ-5D-Y. Project 4. **Pilot cluster randomised control trial (RCT) trialling two alternative methods of presenting the EQ-5D-Y as a PROM to clinicians and families** across a number of outpatient clinics/ clinical groups for 3 months. The two methods of incorporating and feeding back the EQ-5D-Y as a PROM for clinicians will be informed by stage 1, 2 and 3 using a codesign process. A mixed methods evaluation will be conducted to include changes in clinical practice, confidence, knowledge and outcomes. Qualitative information on the usefulness, feasibility and sustainability of the trialled methods for incorporating the EQ-5D-Y as a PROM will also be obtained. This PhD will be a complement to the QUOKKA study led by Prof Devlin but produces new outputs based on use of EQ-5D-Y as a routine measure in a clinical hospital setting.Kim DalzielYouth124858Ongoing20222025
287-PHDMeasuring Health-Related Quality of Life in Orthopedic Clinical Practice**General** In this PhD-project the broad potential of EQ-5D in Orthopedics is demonstrated, thereby enhancing use of the Dutch arthroplasty registry; it also adds a registry (in scoliosis, a chronic condition). By working from within the setting and with a multi-facetted program we believe we will add to wider implementation in practice, beyond the mere data collection in registries. Two projects have already started, in part sponsored by EQ. The proposal has been reviewed by members of the LSA working group. **Collaborations formed** The proposed research program not only enables a PhD achievement of the candidate, but will also build up an expertise center at the department of Orthopedics at the Erasmus Medical Center (EMC) on the best use of the EQ-5D, given the use at stake, and other available outcome measures. Direct links to the department of Public Health (Haagsma, EMC), other clinical centers, the Dutch registry (LROI) are formed. Also, it internationally reaches out (APERSU; not restricted to this), within feasible limits. This program invests in long-term stakeholder engagement. **Working packages** The proposal includes 5 projects resulting in at least 6 peer-reviewed papers. Key element is level-participation of users, and a strong validation component using other questionnaires and clinical outcomes/decisions. The below mentioned projects function as seeding projects; they will develop new projects for junior researchers/students. 1) *Systematic review on use of PROMs specifically EQ-5D as quality improving tool, with a focus on Orthopedics* The study will provide lessons for the EQ-group to further widen use via e.g. teaching or research. 2) *Using Dutch arthroplasty registry data to show EQ-5D inequalities in Orthopedics* The first study to incorporate findings from the EQUIMETRICS project, providing evidence either supporting or refuting the suitability of the EQ instrument compared to other disease-specific measures in measuring Health Inequities. Methods will be shared at an appropriate time. 3) *Using Dutch arthroplasty registry data to study the impact of COVID-19 in Orthopedics* Besides the societal relevance of reporting an impact this study will show the relative performance of EQ-5D to assess such an impact in comparison to other widely spread measures (e.g. the Oxford set). 4) *A sub-study from the POPCORN-survey to illustrate the COVID-19 impact in the general population with joint disease* The EQ-5D is used as population metric to evaluate health-related quality of life during a pandemic; the evidence will strengthen the position of the EQ-5D/EQ-VAS in this regard. 5) *A registry founding project in scoliosis children in which also the performance of the EQ-5D-A is compared to the EQ-5D-Y* This project will show whether the Youth version translates longitudinally into the Adult version, enabling longitudinal follow-up. **PhD-candidate** The PhD candidate has a double qualification (medicine, health economics) and is currently working as physician at the a.o. COVID-19-ward, aiming to become an orthopedic surgeon in the future. We regard this broad background as one of the unique features. His COVID-19 expertise could benefit the current EQ work in that direction (he already supported the POPCORN wave 2 questionnaire on long-term COVID-19 follow-up).Joshua BonselPopulations and Health Systems, Youth112680Ongoing20212024
345-PHDStatistical methods for handling and analyzing EQ-5D-5L data in randomized clinical trialsThere is a growing interest in measuring patient-reported outcomes (PROs) in clinical research. The EQ-5D-5L has been often included to supplement disease-specific PRO instruments in randomized clinical trials (RCTs). The EQ-5D-5L data, if analyzed properly, can produce important evidence for two purposes: estimating the treatment effect between arms and deriving health utilities to support economic modeling. However, the analysis and reporting of the EQ-5D-5L in RCTs is rather limited, due at least in part to the lack of methodological guidance in analyzing utility data. Among those RCTs with published EQ-5D data, there is noticeable heterogeneity in choosing statistical models for data analyses. This PhD project aims at producing empirical evidence in comparing a wide range of statistical models in estimating treatment effect using the EQ-5D-5L and deriving health utilities for economic models. Furthermore, the output from this research program can be used to develop practical guidances on analyzing the EQ-5D-5L in the RCT setting. Such guidances can potentially improve the quality of analyzing and reporting the EQ-5D-5L in this context.Feng XieOthers90000Ongoing20202025
320-VSResubmission of pre-approved EQ Project 20190450: Re-estimating the EQ-5D-5L value set for ChinaChina was one the first few countries completing the EQ-5D-5L value set study. Similar to other studies done in the first wave, the China study suffered from limitations like insufficient interviewer training and monitoring. For example, the China 5L valuation study showed signs of interviewer effects: respondents from Beijing had significantly higher TTO values than respondents from other cities. A particular concern is that the China valuation study only included the urban population; the rural population was not sampled due to budget constraint. The rural population accounts for over 50% of China’s population, and it is a focus of the ongoing healthcare reform because it is long underserved. With the rapid development of HTA and Pharmacoeconomics, a growing number of economic evaluation studies in China will use instruments with a national value set such as EQ-5D. Recently, the SF-6D V2 value set of China using a nationally representative population sample (urban respondents: 1600; rural respondents: 1600) has been published. Based on the latest 2020 Pharmacoeconomics Guideline of China (forthcoming in English), SF-6D V2 will effectively be the most preferred utility instrument in China, if the EQ-5D-5L value set is not re-estimated by including rural population. Therefore, we aim to estimate a more representative EQ-5D-5L value set for China using a sample including both urban and rural respondents and the latest EQ-VT protocol. For this study, we will use a new DCE design with 240 pairs which used overlapping technique.Nan LuoValuation147810Ongoing20212023
351-RABehind the scenes: a mixed method investigation of the impact of quality control procedures on interviewers performanceBackground: The EuroQol Valuation Technology (EQ-VT) protocol is currently employed by valuation studies of the EQ family of instruments across the world. Protocol compliance is supported by the standardised interviewer training, instructions manual and the quality control (QC) process. All evidence in support of the QC originates from quantitative indicators, and so far, no studies have investigated the qualitative aspects of interviewers’ performance in the context of EQ-VT. Objectives: To investigate how the interaction between the interviewer and the respondent affects data quality in EQ-VT interviews and identify patterns that influence data quality. Methods: This study is based on EQ-VT interviews carried out as part of the Italian EQ-5D-5L valuation study, whereby approximately one-fifth of all interviews were video-recorded. We will use both quantitative (i.e. parameters measured during the QC) and qualitative data (i.e. video-recordings and transcribed texts of interviews and written account of the feedback provided to the interviewers). A three-step approach to data analysis will be employed: (1) Using conversational analysis, we will identify dimensions of interviewers’ conversational practices in the transcripts of a subset of interviews; (2) We will develop a scoring system of verbal and non-verbal skills by which a randomly selected 110 video-recorded interviews will be scored; (3) Evidence from the first two steps together with reports of QC feedback provided and the quantitative evidence will be triangulated to identify interaction cues related to data quality.Michela MeregagliaValuation38140Ongoing20212022
322-RAInvestigating the development of a multi layered “Deep Dive” measure of health-related quality of life based on the EQ-5D: A pilot studyThere have been discussions within the DSWG and elsewhere about the potential to develop an instrument based on the EQ-5D that combines the benefits of both preference based and profile measures. This can be defined as a multi layered ‘Deep Dive’ approach to measuring health and quality of life and includes a higher-level preference-based measure (i.e. EQ-5D descriptive system) (Layer 1), and a set of items associated with each dimension included in the preference measure (Layer 2). This would generate a system including both a preference-based values and profile scoring, enhancing EQ-5D measurement properties without affecting the descriptive system. The aim of this pilot study is to investigate the potential development of a multi layered ‘Deep Dive’ instrument based on the EQ-5D. The results will inform a larger research programme to develop a Deep Dive instrument led by the DSWG. The study includes two stages. The first stage will examine the conceptual basis for such an instrument, and will consider theoretical, conceptual and methodological issues, and the benefits and limitations. Stage one will result in documents for consultation, and to facilitate the early involvement of the membership in future initiatives. The second stage will test the feasibility of developing a Deep Dive. This will be done by applying psychometric methods to existing datasets to investigate the association between the EQ-5D dimensions and overlapping items from other measures. This tests the feasibility of increasing the amount of information provided by the dimensions included in the EQ-5D descriptive system.Brendan MulhernDescriptive Systems40280Ongoing20212022
337-RAMeasurement properties of the EQ-5D in diseases of the upper respiratory tract: a systematic review**Aims:** To perform a systematic review on EQ-5D (EQ-5D-3L or EQ-5D-5L) psychometric properties in diseases of the upper respiratory tract (including such anatomical areas as nose, pharynx, and larynx). **Methods:** The review protocol will be registered in the PROSPERO database, before the start of a systematic search. The review will follow the PRISMA guidance on systematic reviews and meta-analyses. The following electronic databases will be searched: MEDLINE (PubMed), EMBASE (Elsevier) and EuroQol Group publication database. Additional references will be obtained from reviewed articles. Two authors will independently screen the titles and abstracts of studies resulting from the searches and then – full texts of selected articles. The quality of included studies will be assessed according to COSMIN guidelines. Data from studies meeting the inclusion criteria will be extracted using a pre-determined extraction form. Measurement properties will be summarised based on the type of property assessed (feasibility, distributional properties, content validity, construct validity, test-retest reliability, responsiveness). We will try to explain the heterogeneity in results between studies. Results will be presented in a narrative and tabular form. In the case of a sufficient number of studies and their homogeneity, results will be pooled. Clinical indications with sufficient data on EQ-5D psychometric properties and those which should be studied in the future will be indicated.Dominik GolickiDescriptive Systems29865Ongoing20212022
339-RASystematic Review of Measurement Properties of the EQ-5D in Hematologic Cancers**Aims:** To perform a systematic review on EQ-5D (EQ-5D-3L or EQ-5D-5L) psychometric properties in hematologic cancer patients. **Methods:** The review protocol will be registered in the PROSPERO database, before the start of a systematic search. The review will follow the PRISMA guidance on systematic reviews and meta-analyses. The following electronic databases will be searched: MEDLINE (PubMed), EMBASE (Elsevier) and EuroQol Group publication database. Additional references will be obtained from reviewed articles. Two authors will independently screen the titles and abstracts of studies resulting from the searches and then – full texts of selected articles. The quality of included studies will be assessed according to COSMIN guidelines. Data from studies meeting the inclusion criteria will be extracted using a pre-determined extraction form. Measurement properties will be summarised based on the type of property assessed (feasibility, distributional properties, content validity, construct validity, test-retest reliability, responsiveness). We will try to explain the heterogeneity in results between studies. Results will be presented in a narrative and tabular form. In the case of a sufficient number of studies and their homogeneity, results will be pooled. Clinical indications with sufficient data on EQ-5D psychometric properties and those which should be studied in the future will be indicated.Dominik GolickiDescriptive Systems34390Ongoing20212022
329-RAEQ VAS: What does it measure? A structured analysis of the EQ VAS in national population surveys (the Health Survey of England)**Background**: Global self-assessed health status (SAH) is a powerful predictor of mortality and morbidity. The EQ VAS, a measure of SAH, has been included in the EQ-5D measure since its inception but has been underutilized and understudied. **Aims**: We propose a study that leverages the Health Survey of England (HSE), an annual population health survey, to understand which dimensions beyond the EQ-5D health profile contribute to variation in the EQ VAS. We also aim to develop a conceptual model for EQ VAS. **Methods**: HSE for the years 2017/2018, which includes EQ-5D-5L data, will be used in these analyses. Using the conceptual framework by Jylhä 20091 and Picard et al 20132, the relationships between objective (e.g. diagnosis, medications), subjective (contextual) health variables, as well as “non-health” (e.g. socioeconomic) variables and EQ VAS will be examined using regression analysis (including block and Lasso approaches). Secondly, a targeted literature review on conceptual models for SAH will be conducted. Lastly, we will hold a workshop with interested EuroQol members to clarify a conceptual model for the VAS and develop a study to test that model. **Significance for the EuroQol group**: Despite changes to the descriptive system EQ VAS has remained a largely unmodified element of the EQ-5D family of instruments. However, the EQ VAS remains underexplored and underexploited as a brief and comprehensive health measure. It is of strategic importance for the EuroQol group to understand ways in which the EQ VAS can be used and interpreted, especially in larger scale studies.You-Shan FengDescriptive Systems, Populations and Health Systems55230Ongoing20212022
326-RAA feasibility study of applying PAPRIKA to the EQ-HWBBackground: The EQ Health and Wellbeing Short (EQ-HWB-S) is a new measure developed for use in the evaluation of health and social care interventions. Composite time trade-off (cTTO) and discrete choice experiments (DCE) are currently being used to value the measure. However, these methods are cognitively challenging, and combined with the challenge of imagining EQ-HWB-S states, there are concerns about how well respondents engage with the tasks. One approach to minimise burden is an adaptive DCE using the Potentially All Pairwise RanKings of all possible Alternatives (PAPRIKA) method which constrains what respondents see to two dimensions with a separate binary search to identify the location of 'dead'. The aim of this study is to assess the feasibility and applicability of PAPRIKA to value the EQ-HWB-S. Method: A mixed method study with members of the general public will be undertaken inclduing:1) qualitative interviews (n=16) to assess views of PAPRIKA and provide information to develop the online survey; 2) an online PAPRIKA survey (n=300); 3) follow-up interviews (n=24) of those who complete the online survey. Qualitative data will be analysed using a framework approach to identify feasibility and applicability of this approach to EQ-HWB-S. Quantitative data will be analysed and compared to data from the on-going EQ-HWB-S cTTO and DCE study. Results from the study will inform on the feasibility of the approach which offers an alternative to valuing the EQ-HWB-S and potentially bolt-ons are a larger pool of items from the EQ-HWB which has 25 items.Clara MukuriaValuation, EQ-HWB52460Ongoing20212023
318-RAMeasurement properties of the EQ-5D-5L among non-small cell lung cancer patients on active treatments in ChinaBackground: Target treatments and immune checkpoint inhibitors have changed the therapeutic landscape for non-small cell lung cancer (NSCLC) treatment in recent years. These new treatments extend patients’ overall survival and with less adverse effect compared with chemotherapies. Interests in conducting cost-effectiveness between NSCLC treatments requires HRQL collected from this population. However, there is limited information on the performance of the EQ-5D-5L for NSCLC in China. Aim: To test the measurement properties of the EQ-5D-5L in measuring HRQL of NSCLC patients on active treatments in China. Methods: This study will piggy-back on a health survey that will interview 800 NSCLC patients on active treatments from 16 hospitals across 7 provinces/cities in China in 2021. Patients will be recruited using a quota sampling frame based on: 1) the distribution of NSCLC incidence rates (age and sex) in China, age and sex distribution of the general population, and the distribution of NSCLC treatments. Both EQ-5D-5L and EORTC QLQ-C30 will be administered. Demographic and disease- and treatment-related information will also be collected. To investigate the measurement performance of the EQ-5D-5L, we will examine: 1) response patterns using ceiling and floor effects and distribution across severity levels of each dimension; 2) convergent construct validity using Spearman’s correlation against the QLQ-C30; 3) known-group validity by type of treatment, cancer stage, and QLQ-C30 domain scores; and 4) informativity and discriminatory power using the Shannon and Shannon evenness indices.Xuejing JinDescriptive Systems, Populations and Health Systems39040Ongoing20212022
274-RATesting the psychometric properties of two respiratory bolt-onsA key goal of the EuroQol Group is to continuously improve the EQ-5D descriptive system. One of the ways is to utilize dimensional bolt-ons that may better capture the health-related quality of life (HRQoL) of patients with certain diseases. Recently, Hoogendoorn et al. had developed two respiratory bolt-on dimensions to improve the responsiveness of EQ-5D in assessing respiratory diseases, such as chronic obstructive pulmonary disease (COPD) and asthma. The importance and feasibility of the two bolt-ons have been demonstrated in a valuation study. However, their psychometric properties are unknown. Therefore, we propose to test the construct validity, test-retest reliability and responsiveness of the two respiratory bolt-ons in patients with COPD and asthma. This study will piggyback on an existing EuroQol research project to prospectively collect data from patients with COPD and asthma (n=200) using the two respiratory bolt-ons. The construct validity of the two bolt-ons will be assessed in the form of convergent validity and known-groups validity by examining their association with clinical variables and a disease-specific HRQoL measure. The test-retest reliability of the bolt-ons will be assessed using patients whose health status are known to be stable over time and responsiveness will be assessed using patients with improved or worsened health status. This project will provide the first evidence for the psychometric properties of two respiratory bolt-ons. Such evidence will be useful to inform the decision on the movement from the beta version to the approved version as well as future research for the two bolt-ons.Descriptive Systems24790Ongoing20212022
243-RAThe relationship between the EQ-5D-5L “anxiety/depression” dimension and anxiety and depression symptomsBackground: The EQ-5D-5L “anxiety/depression” (A/D) dimension is a so-called composite dimension. The forced single response relies on the respondent's ability to choose a level when the levels of the anxiety and depression problems, taken apart, may differ. This study investigates the properties and use of the A/D dimension. Study aims: The aims of this study are: (1) to establish empirically the descriptive aggregation rules of respondents to report health; (2) to investigate the relationship between the composite A/D dimension, separate anxiety (A) and depression (D) dimensions and anxiety and depression symptoms as measured with the Generalized Anxiety Disorder–7 (GAD-7) and Patient Health Questionnaire–9 (PHQ-9). Methods: This study is a secondary data-analysis of data from the POPCORN study. These data were collected using a web-based survey among general population samples from 9 countries (total n=24,689). The primary outcome measures of our study are the EQ-5D-5L A/D dimension level score, separate A and D dimension level scores (5L) and the PHQ-9 and GAD-7 score. We will use head-to-head comparisons to compare outcomes of the EQ-5D-5L A/D domain with the separate A and D domains. We will examine the frequency of reported problems on the composite AD dimension and separate A and D dimensions by constructing cross tables with AD and A and D. We will assess the Spearman rank correlation coefficient to examine the convergent validity of the AD dimension and the A and D dimension and the GAD-7 and the PHQ-9.Juanita HaagsmaDescriptive Systems19560Ongoing20212022
227-RAComparison of the Afaan-Oromo language version of the EQ-5D-Y-3L and the EQ-5D-Y-5L performance among children and adolescents in EthiopiaIntroduction: In the 10th Euro-Qol joint call for proposals, the Younger Populations Working Group (YPWG) welcomes proposals aiming to validate the five-level EQ-5D-Y in non-English speaking countries, as well as testing the psychometric properties of the extended version in different health conditions. With the already established HTA in Ethiopia and clinicians and researchers interest in measuring health-related quality of life (HRQoL), this research will have crucial implications on measuring HRQoL outcomes, resource allocation, and planning of health care interventions in child and youth populations in Ethiopia. Objectives: To culturally adapt the self-complete versions of the EQ-5D-Y-3L and EQ-5D-Y-5L into Afaan-Oromo (spoken by 35 million native speakers) and to assess and compare the instruments’ psychometric properties in healthy, and children and adolescents with a range of health conditions. • Methods: Both the Y-3L and Y-5L instruments will be translated into Afaan-Oromo following the procedures set by the Euro-Qol Group’s VMC guidelines. Data will be collected from main-stream schools in Addis Ababa and from Tikur Anbessa Specialized Hospital (TASH). Participants will be asked to complete the two versions of the EQ-5D-Y instrument. Descriptive statistics will be used to describe the characteristics for the five separate dimensions of EQ-5D-Y-3L/5L and EQ-VAS. The feasibility, test-retest reliability and known groups’ validity of the EQ-5D-Y-3L and EQ-5D-Y-5L self-complete versions will be tested in a cross-sectional study in paediatric patients aged 8–15 years with any of three health conditions, acute lymphocytic leukemia, congestive heart failure, and asthma (n≈100 for each condition) and healthy children’s (school) (n≈200).Goitom MolalignYouth22080Ongoing20212022
282-RARandomised equivalence study to compare online interviews versus face-to-face interviews to value the EQ-5D-5L using cTTO: Australian armIntroduction: Valuation studies using composite time trade-off (cTTO) interviews have historically been conducted face-to-face. The COVID-19 pandemic forced a number of valuation studies to conduct their interviews online via videoconference. These studies reported that online interviews appeared feasible and acceptable; however, participants were not randomly allocated and there was no reporting of cTTO value equivalence. This study builds on its sister study from the UK and aims to assess the acceptability and equivalence of in person face-to-face interviews with online videoconferencing interviews on cTTO valuation outcomes and explore the impact of interview mode on attendance and on data quality. Methods: Participants were recruited via an external market research company from Greater Melbourne and Regional Victoria. A multi-stage stratified sampling approach was used, with quotas based on Australian Bureau of Statistics (ABS) data for age, gender, geographical location, education level and income. Consenting participants were randomly allocated to complete a cTTO interview face-to-face or online with 1 of 4 trained interviewers. Participants who refused the interview after randomisation were contacted to offer reasons. Participants completed cTTO tasks for the same 10 EQ-5D-5L health states using EQ-VTv2 software. Mean and SD cTTO value overall and for each health state, participant understanding, data quality, demographic characteristics, participant preference, participant engagement and participant feedback were all compared across interview mode. Statistical equivalence for cTTO values for each state was tested using two one-sided t-tests by mode. Finally, regression analysis was completed to assess the impacts of interview mode on cTTO value while controlling for demographic characteristics of the participants. Results: Mean cTTO values were shown to be equivalent for mild health states and showed no significant difference for serious health states. Regressions analysis showed that mode of interview did not have any significant impact on mean cTTO values (p=.817). The proportion of individuals who did not accept their interview in the face-to face group (21.6%) was significantly larger than the proportion in the online group (1.8%) (p<0.0001). No between group difference in demographic characteristics was found. No significant difference was found between groups for participant engagement, understanding or feedback. No significant difference was found between groups for any indicators of data quality. Overall, 151 (37.4%) participants reported that they would have chosen to be interviewed online if given the choice, 51 (12.7%) would have preferred to be interviewed face-to-face and 201 (49.9%) did not mind. Discussion: The results from this study indicate that cTTO values obtained from face-to-face and online interviews are statistically significantly equivalent for mild health states. Regression analysis found no statistically significant association between health state value and mode when all the ten health states were combined. The data generated by both interview modes was of high quality and showed no significant difference. The majority of participants had no preference for interview mode however a larger proportion of participants preferred to be interviewed online rather than face-to-face, this preference was supported by the larger proportion of participants who did not accept their interview when randomised to face-to-face. Both online and face-to-face interviews appear to be equivalent and acceptable for conducting cTTO interviews.Tessa PeasgoodValuation182848Completed20212022
198-RAGoing beyond life expectancy– Examining health inequalities in quality adjusted life expectancy (QALE) in AustraliaWe undertook a study to examine the potential health inequality using quality-adjusted life expectancy (QALE) as an index for future public health policy formulation. Via surveying over 2000 participants, it was found that the gap in QALE is wider than that in life expectancy when comparing the major cities with the remote/very remote areas in Australia. There appeared some intersectional differences in utility values, life expectancy and QALE.Lan GaoPopulations and Health Systems24640Completed20212022
245-RAThe search for a task to measure time preference in EQ-5D valuation: systematic review, experiment and application to stand-alone DCEBackground: There is ample evidence that time preference, i.e. the importance individuals assign to health now and in the future, can influence EQ-5D valuation, for EQ-VT and the upcoming stand-alone DCE protocol. There is, however, no consensus on how time preference could and should be measured in valuation of EQ-5D, with different solutions suggested for EQ-VT and stand-alone DCE. Aim: This project, split into three parts, aims to answer the following research questions: 1) Which methods for measuring time preference exist and how do they differ? 2) Which methods appear promising for use in valuation of EQ-5D? 3) What is the internal consistency, level of difficulty, and time needed for completion for promising methods for measuring time preference? 4) Can a promising method for measuring time preferences at an individual level be included into stand-alone DCE valuation? Methods: Part 1 of this project involves a systematic review of methods for measuring time preference developed in various disciplines. Evidence synthesis will involve categorising methods by: e.g. i) the parametric assumptions made, ii) the amount of questions required, iii) difficulty for respondents, iv) evidence about internal consistency and v) applicability to EQ-5D. From this review, a selection of promising methods for measuring time preference will be obtained. In Part 2, the feasibility and reliability of these methods is determined in an experiment with a general public sample. After a go/no go decision based on the results of part 2, we apply one of these methods in stand-alone DCE valuation of EQ-5D.Stefan A. LipmanValuation134400Ongoing20212023
235-RADo EQ-5D valuations differ in palliative care settings? A discrete choice experimentEvidence suggests that people value health differently in different social care contexts. If this is true, EQ-5D values may be sensitive to the context in which they have been obtained. Therefore, it is important to understand how EQ-5D values might vary, given the context. A debate has recently focused on the health state values elicited from the general population when used to assess palliative and end-of-life interventions. We argue that the trade-off between different domains of EQ-5D may change fundamentally as people progressed through different palliative health care states. This study proposes to contribute to the growing literature on the prioritisation of health care by exploring the valuation of EQ-5D-5L (EQ-5D) in palliative situations using a discrete choice experiment (DCE). Palliative situations will be described and contextualised using a vignette model of presentation, developed by a 3-stage iterative process, including exploratory qualitative work, expert panel discussion, and pilot testing. A DCE will elicit preferences for EQ-5D health states in the context of the palliative care vignettes.Irina KinchinValuation47600Ongoing20212023
20210010Translating EQ-5D into Hausa and validating the new version in hip replacement.not availableAbdulhamid MusaOthers1898Ongoing20212022
236-RADoes priority setting when deciding between adults and children correspond to valuation of EQ-5D(-Y)?Background: The EQ-5D-Y instrument is increasingly used to value health states in children aged 8-15. However, there are still several methodological issues surrounding this instrument, which raised doubt about its validity. An important concern is that adults are willing to trade off substantially less time in the TTO task for children than for themselves, resulting in higher utilities for the same health states in children than in adults. It is not clear if this reflects that impaired health states are indeed giving more utility in children than in adults, or that it is an artefact of other factors. Aim: This project aims to test if priority setting decisions in healthcare allocation tasks between adults and children correspond to the TTO utilities generated by the EQ-5D and EQ-5D-Y, respectively, when accounting for equity weights. We also aim to test if and how this correspondence change when replying TTO utilities by VAS weights. Methods: We elicit VAS scores and TTO utilities for a wide spectrum of health states in a controlled, computerized, lab experiment. In a person trade-off task, the subjects have to choose between allocating a scarce healthcare budget to a group of children and a group of people of their own age. By changing the amount of people in each group, we can determine the relative weight given to these two groups. We then test the ability of the generated VAS and TTO utilities to predict these weights, while controlling for equity preferences by separate questions.Arthur AttemaValuation, Youth39360Completed20212021
232-RAMeasurement properties of EQ-5D-Y and other commonly used generic preference-based measures for children and adolescents: a systematic reviewBackground & study aim Preference-weighted measures (PWMs) of health status/health-related quality of life play an essential role in estimating Quality-adjusted life-years (QALY) for use in economic evaluations of health care products and interventions. However, as PWMs are first and foremost intended to accurately reflect respondent health status, ideally, they should demonstrate good psychometric properties for the population in question. This study aimed to systematically review published evidence on the measurement properties of commonly used PWMs for children and adolescents. Methods Three electronic databases (PubMed, Medline, and PsycINFO) were searched for articles assessing the psychometric properties (content validity, construct validity (including convergent validity and known-group validity), test-retest reliability, and responsiveness) of the PWMs of interest (AQoL-6d, CHU9D, HUI2, HUI3, and EQ-5DY). The COsensus-based Standards for the selection of health Measurement INstruments methodology (COSMIN) guidelines were used to assess a) the methodological quality of the studies included and b) the psychometric performance of the instruments covered. Data were analysed overall as well as by population (country and disease group) and perspective (self-report or proxy-report). This study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (CRD42021277296). Results In total, 42 articles were included in this systematic review. HUI was tested only in patient populations, CHU9D was tested mostly in general population samples, while EQ-5D-Y was tested in both populations. Overall, there was high-quality evidence supporting sufficient construct validity for all instruments except AQoL-6D. Evidence for supporting test-retest and responsiveness was of low quality. Evidence for content validity was minimal and therefore not extracted and synthesized for all PWMs. Conclusion This review provides updated evidence on the measurement properties of existing generic PWMs for children and adolescents. This review only included articles published in the English language and results are limited and might be biased. More studies are needed to further assess those instruments in a wider range of population types, countries, and disease groups.Nan LuoDescriptive Systems, Youth22800Completed20212021
217-EOPromoting and Supporting the use of EQ-5D instruments in ChinaSince 2017, China’s National Healthcare Security Administration requires economic evaluation (mainly cost-utility analysis) reports when they renew/extend the drug formulary for the national healthcare insurance plans, which cover more than 95% of the Chinese population. Pharma and consulting companies and researchers who conduct economic evaluations in China are therefore increasingly interested in preference-based measures. In addition, health care providers and researchers are interested in patient-centred care, especially in the oncology area. However, relevant Chinese stakeholders do not have a lot of opportunities to systematically learn the details of the EQ-5D instruments, such as how to obtain the instrument, how to use EQ-5D in their studies, and how to apply the instruments beyond economic evaluations. Furthermore, there is a lack of effective communication channels between the EuroQol Group and Chinese EQ-5D end-users. Therefore, this proposed project aims to promote and support the use of EQ-5D instruments in China through webinars and other knowledge translation channels; by increasing awareness about the EQ-5D instrument, EQ-5D researchers (EuroQol Group members), and EQ-5D-related studies; providing guidance on how to obtain the EQ-5D instruments for commercial and non-commercial uses, how to apply EQ-5D in economic evaluations and beyond economic evaluations; and showcasing successful examples of large-scale applications of EQ-5D to encourage relevant Chinese stakeholders to experiment in this aspect. We anticipate 1000 attendees for each webinar on average. We will also establish and pilot a platform for direct communication to the EuroQol Group for Chinese EQ-5D end-users.Xuejing JinEducation and Outreach22400Ongoing20212022
20200060Grant to develop the outcomes-research component of the intended Capacity-2 studyCapacity-2 aims to extend the current multicenter Capa­city-clinical registration with 6 month outcome data, including EQ-5D/EQ-VAS; these data would allow to document whether - as suggested by preliminary imaging and pathophysiological findings - whether specific cardiovascular complications arise after a significant covid infection requiring hospital admission. Following the received information, about 30-40 (Dutch) hospitals participate in Capacity at large. The currently projected follow-up study aims to follow up at least 500 patients, with the option that more hospitals join (incl. international). The developmental award covers IT-associated costs which relate tot the projected outreach and data collection 6 month after first discharge of qualifying patients (see Capacity-2 protocol). Part of the data collection includes clinical parameters (incl. history taking), imaging data, and biomarkers. The other part of the data collection consists of a dedicated health questionnaire, primarily containing PROMs and some data required to process or interprete the outcome data. This questionnaire will also contain a.o. the EQ-5D and the EQ-VAS.F AsselbergsPopulations and Health Systems29500Ongoing2020
225-RAAssessing and comparing psychometric properties of both 3L/5L of EQ-5D-Y and adult EQ-5D versions in adolescents with prevalent disease conditions in EthiopiaIntroduction: The EQ-5D-Y user guide suggests that either the EQ-5D-Y or adult EQ-5D versions can be used for respondents aged 12-15 years. The Younger Populations Working Group (YPWG) has expressed interest in receiving proposals to compare the psychometric performance of new, extended version of the EQ-5D-Y in non-English speaking countries in this age group. Given that the simpler version of EQ-5D-Y (the EQ-5D-3L) may be preferred in some contexts, we believe it is of interest to test both versions with a range of health impairments to gain information on the dimensions, items and the psychometric performance of the EQ-5D-Y-5L instrument. Objectives: To assess and compare the psychometric properties of the 3L and 5L versions of EQ-5D-Y and the adult EQ-5D in healthy adolescents aged 12 – 17 and those with a range of health conditions. Method: Both 3L/5L of EQ-5D-Y and adult EQ-5D versions (in Backward translation phase (TRF2225)), are under development into Tigrinya language following the procedures required by the EuroQol Group’s Version Management Committee. The feasibility, reliability, and known group validity of both 3L/5L of EQ-5D-Y and adult EQ-5D versions will be tested and compared in a cross-sectional study in adolescent patients aged 12–17 years with pulmonary TB and epilepsy health conditions and in a control group of ‘healthy’ adolescents (n≈ 480). Test-retest reliability and responsiveness will also be examined after 10 to 15 days of the first visit. Preference for the two versions will also be assessed.Abraham GebregziabiherDescriptive Systems, Youth24693Ongoing20212021
224-RAA review of the impact of a one day versus a seven-day recall period on domains from the EQ-5D and EQ-HWB instrumentsMethod: This targeted review integrates quantitative and qualitative literature across health, economics, and psychology to explore the effect of a one-day (or ‘24 hour’) versus seven-day (or ‘one week’) recall period. We extracted results relevant to six domains with generic health relevance: physical functioning, pain, cognition, psychosocial wellbeing, sleep-related symptoms and aggregated disease-specific signs and symptoms. Quantitative studies compared weekly recall scores with the mean or maximum score over the last seven days or with the same day recall score. Results: Overall, across the 24 quantitative studies identified, 158 unique results were identified. Symptoms tended to be reported as more severe and health-related quality of life (HRQoL) lower when assessed with a weekly recall than a one-day recall. A narrative synthesis of 33 qualitative studies integrated patient perspectives on the suitability of a one-day versus seven-day recall period for assessing health state or quality of life. Participants indicated a preference for one-day recall for reporting symptoms, except where conditions were characterised by high symptom variability, or where PROMs concepts required integration of infrequent experiences or functioning over time. Conclusion: This review identified a clear trend toward higher symptom scores and worse quality of life being reported for a seven-day compared to a one-day recall. A better understanding of the impact of using different recall periods within PROMs and HRQoL instruments will help contextualise future comparisons between instruments.PeasgoodDescriptive Systems24720Completed20212021
194-RAAssessing older people's health related quality of life in aged care settings: unravelling the EQ-5D self-report proxy conundrum.This project focused on a program of work to address the unique challenges in measuring quality of life with older people in aged care. There is ongoing debate surrounding the ability of frail older people with cognitive impairment and dementia to self-report their own health-related quality of life (HRQoL). Despite increasing calls for inclusivity and self-assessment of HRQoL where ever possible, proxy-assessment of older people’s HRQoL (by family members or health/aged care professionals) is often used as the default option. This project applied an innovative mixed methods approach, incorporating qualitative think aloud and eye tracking technology in older people who self-completed the EQ-5D-5L, along with proxy family member data collection of HRQoL using the EQ-5D-5L from two proxy perspectives; proxy-proxy and resident-proxy. The project has facilitated a more detailed understanding of the impact of cognitive impairment and dementia on self-assessed HRQoL for older people in residential aged care settings using the EQ-5D-5L. The project has also provided an assessment of the impact of proxy perspective on the inter-rater agreement of self- and proxy-reported quality of life.Julie RatcliffeDescriptive Systems, Populations and Health Systems97287Completed20212022
214-RAMeasurement properties of the EQ-5D-Y: a systematic review**Aims:** To perform a systematic review on the EQ-5D-Y (EQ-5D-Y-3L, EQ-5D-Y-5L) psychometric properties. **Methods:** The following electronic databases will be searched: MEDLINE, EMBASE and EuroQol Group publication database. Additional references will be obtained from reviewed articles. No study design restrictions will be implemented. The search strategy will be focused on the EQ-5D-Y instrument. Studies on EQ-5D-Y measurement properties will be selected during the abstract selection process and full-text screening. Two authors will independently screen the titles and abstracts of studies resulting from the searches, and then – full texts of selected articles. The quality of studies will be assessed according to the determined criteria. We will not exclude any relevant studies but will highlight any concerns about quality. Data from studies meeting the inclusion criteria will be extracted using a pre-determined extraction form. Measurement properties will be summarized based on the type of property assessed (validity, test-retest reliability, responsiveness, feasibility, distributional properties, informativity). We will try to explain the heterogeneity in results between studies. Results will be presented in a narrative and tabular form. Clinical areas with sufficient data on EQ-5D-Y psychometric properties and areas which should be studied in the future will be indicated.Dominik GolickiDescriptive Systems, Youth31140Completed20212021
196-RAValidity of the EQ-5D-3L and EQ-5D-5L in advanced MelanomaThere is limited evidence on the validity of the EuroQol-5Dimensions (EQ-5D) in advanced melanoma. This study aims to assess and compare the validity and responsiveness of the EQ-5D-3L and the EQ-5D-5L in patients with advanced melanoma. We will use data from the high-quality nation-wide Dutch Melanoma Treatment Registry (DMTR) containing data on patient and disease characteristics and, for about 1280 patients, data on Patient Reported Outcome Measures (PROMs) including the EQ-5D-3L, EQ-5D-5L, Visual Analogue Scale (VAS) and the Functional Assessment of Cancer Therapy for Melanoma (FACT-M). For assessing the validity of the EQ-5D, we will first compare feasibility, content and construct validity, and responsiveness of the three- and the five-level version. After that, we will assess validity by comparing observed and predicted outcomes of the three- and five-level version. To do so, we will first regress the EQ-5D-3L and EQ-5D-5L scores on the FACT-M scores to obtain a mapping algorithm for each version. After that, we will predict EQ-5D index scores for both patient groups (3L and 5L) using our developed mapping algorithm for its counterpart version (3L for 5L and vice versa). We can then head-to-head compare observed and predicted EQ-5D scores to assess validity of both EQ-5D versions. This study will add evidence to the current body of knowledge of the EuroQol group regarding the validity and responsiveness of the EQ-5D in advanced melanoma. It will also give valuable insights into the validity of using the EQ-5D-3L in comparison to the EQ-5D-5L.Margreet FrankenDescriptive Systems25000Ongoing20212022
215-2020RATranslating the ‘Methods for analysing and reporting EQ-5D data’ into ChineseThe interests of using EQ-5D is growing rapidly in China. Many Chinese studies were published on Chinese journals. Unlike the studies published in English journals, the studies published in Chinese received little attention from the EuroQol Group. Moreover, both the reviewers and editors of the Chinese journals often lack sufficient knowledge about EQ-5D. For such reasons, the use of EQ-5D including data collection, data analysis and report were often not well prepared or conducted in many studies published on Chinese journals. Although many Chinese EuroQol members have been trying to help the local researchers/users as best as they could, there are always some users could not get assistance when they need. Recently, Devlin et al published a book ‘*Methods for analysing and reporting EQ-5D data’*. We felt this could be extremely helpful for users. However, due to limited English fluency for some Chinese researchers, especially those who chose to publish on Chinese journals, the original English version may not be well understood easily. Hence, in this project, we aim to translate the book into Chinese together by a group of Chinese EuroQol members, and make it available for all interested users.Zhihao YangEducation and Outreach26530Ongoing20212023
216-RAExploring the validity of EQ-5D-5L in Indigenous people of CanadaBackground. Worldwide, there is a need to recognize the distinct health needs of Indigenous people, address inequalities of health experienced by Indigenous people, and improve the appropriateness of health care services and measures of chronic disease. We propose to investigate the application of traditional approaches to measuring and valuing health, based on a better understanding of preferences and values of Indigenous people of Canada. We aim to explore the validity of the EQ-5D and traditional choice-based health valuation tasks, and make recommendations for the use of EQ-5D-5L in Indigenous people and its application in evaluating effectiveness of healthcare. Methods. This project will use mixed methods, informed by an on-going systematic review to describe the application, development, and performance of preference-based measures in Indigenous people. Face and content validity of the EQ-5D-5L and interpretation of traditional stated choice and health valuation tasks will be explored using a qualitative and think-aloud approach. We will further investigate current approaches by examining stated preferences for attributes of health using multiple stated choice techniques (TTO and DCE) to explore differences between Indigenous and non-Indigenous people. Expected Results. This research will determine the extent to which the EQ-5D-5L captures health as conceptualized by Indigenous people, and explore whether current approaches to measuring health state preferences fully represents their values and preferences. Recommendations will be made regarding the application of EQ-5D-5L and valuation methods used in the EQ-VT protocol in these populations, and potential modifications or methodologic approaches that may benefit this group.Jeff JohnsonDescriptive Systems119441Ongoing20212023
221-RATesting two alternative TTO methods for valuation of EQ-5D-Y health states by trading life years in adulthoodIntroduction: One drawback of the current cTTO method for the Y valuation is that it works by asking respondents to consider shortening a 10-years-old child’s life for better health. Although it is hypothetical, the cTTO task could be upsetting and abhorrent to some respondents. Also, there is concern that adult respondents are unwilling to trade child life years, and consequently many studies reported cTTO values of Y valuation studies higher than the cTTO values of corresponding adult states. In order to overcome this potential issue, we conceptualized, and pilot tested two alternative TTO variants named the Parent TTO (PTTO) and lag-time TTO (LTTO). Both methods ask respondents to trade adult life years. We hypothesized that they have higher acceptability and would generate lower values compared to cTTO. Method: We collected PTTO and LTTO data for the 10 health states included in the EQ-5D-Y valuation protocol from a general population sample in China. The data collection was piggybacked on the China EQ-5D-Y valuation study, using three interviewers with experience using EQ-VT (two interviewers participated in the China Y study, one interviewer participated in two methodological study). For comparison, the cTTO data, including cTTO values, feedback questions and time etc. were drawn from the China EQ-5D-Y study. We compared the methods in terms of acceptability (using three feedback questions: easy to understand, easy to tell the difference, easy to make the decision), feasibility (time to complete the practice task, time to value the 10 states) and characteristics of TTO values (mean and data distributions). Results: In total, 304 participants were included (cTTO: 100; LTTO: 102; PTTO: 102) in this study. On a 5-point Likert scale, the mean score of the ‘easy to understand’ question was 1.18 (SD: 0.58), 1.45 (SD: 0.91) and 1.65 (SD: 1.02) for cTTO, LTTO and PTTO respectively. The mean score of the ‘easy to differentiate’ question was 1.45 (SD: 0.91), 1.94 (SD: 1.08) and 1.86 (SD: 1.24) and the mean score of the ‘difficult to decide’ question was 3.61 (SD: 1.29), 2.97 (SD: 1.33) and 3.02 (SD: 1.50) respectively. The mean (SD) time spent on the wheelchair example was 276.34 (147.51), 350.33 (140.28) and 454.44 (139.92) seconds for cTTO, LTTO and PTTO, respectively. The mean (SD) time spent on valuing each of the 10 states was 102.97 (29.48), 134.66 (49.69), 141.72 (47.07). The mean (SD) TTO values of all 10 states were 0.463 (0.494), 0.387 (0.555) and 0.123 (0.710). All tests were significantly different when using cTTO method as references, except that the mean value comparison between cTTO and LTTO. LTTO and PTTO showed clear clusterings at 0 and -1 respectively; PTTO had more values on the negative value range. Discussion: By designing and testing these two alternative TTO methods to trade-off life adult years, we found participants did not find the TTO tasks more acceptable and feasible, but these two methods do produce values that are more similar to the EQ-5D-5L values. We also found the TTO values may be affected by the parental status and age of the respondents suggesting that researchers to pay attention to the sample representativeness when conducting an EQ-5D-Y valuation study.Zhihao YangValuation, Youth49760Completed20212021
242-RAInterviewer Administered and Self-Complete versions of EQ-5D-5L: agreement and psychometric propertiesThe EuroQol Group has recently developed the Interviewer Administered (IA) version of EQ-5D. This IA version may be useful for respondents who are illiterate, low socio-economic status, low vision problem, patients who are unable to self-complete the questionnaire, or remote telephonically collected data. Usually the self-completed (SC) version was used and then the interviewer read out loud the items to their respondents (assisted SC). Countries such as Indonesia and India could utilize this IA version for collecting data from the aforementioned groups. It remains to be tested whether different modes of administration: IA, SC, and assisted SC, are equivalent. We will recruit 600 respondents in three groups: literate, illiterate, and patient groups. Characteristics for sampling are different residences (urban and rural areas), age group (below 30, 30-50, and above 50 years), and different types of job (no, casual, and permanent). Sociodemographic and health condition questionnaire, EQ-5D-5L SC and IA versions, General State of Health will be completed by the respondents. The IA and SC version will be completed by the literate groups, IA and assisted SC by the illiterate and patient groups, in randomly order three days apart. Individual interview will be done with illiterate groups to investigate the feasibility of completing EQ-VAS with the IA version, because no visual prompt is available for visualizing the vertical line of EQ-VAS. Agreement between dimensions’ responses, feasibility, ceiling effect, classification efficiency (or informativity) will be calculated and compared between the three versions. Interview data will be analysed using a thematic analysis approach.Fredrick Dermawan PurbaDescriptive Systems44370Ongoing20212022
223-RAContent validity and measurement properties of the EQ-5D-3L and EQ-5D-5L in epilepsy patients in a low-income settingAbstract Introduction: Although EQ-5D has been used in several studies in epilepsy patients, most have used the EQ-5D-3L. The few studies that have systematically assessed the instrument’s psychometric properties have produced inconsistent results and some research has indicated that the instrument may not perform well in this population. Notably, however, very few studies have examined the performance of the EQ-5D-5L in epilepsy and there has been no direct comparison of the 3L and 5L versions in epilepsy. Additionally, there has been no research assessing EQ-5D’s measurement properties in epilepsy in low-income countries, where the burden of disease is greatest, and no investigation anywhere of the intrument’s content validity for use in epilepsy. The aim of this study is therefore to perform comparative assessment of the reliability, validity, and usefulness of the 3L and 5L versions in epilepsy patients managed in a low-income setting. Method: Mixed-method study will be conducted in epilepsy patients attending Ayder Comprehensive Specialized Hospital, Ethiopia. Measurement properties of the EQ-5D-3L and EQ-5D-5L will be assessed in approximately 200 adult patients. Patients will complete EQ-5D-3L and EQ-5D-5L, EQ-VAS, and PHQ-9. Socio-demographic information will also be collected and medical records consulted to categorise patients on key clinical parameters for known groups’ assessment. The two EQ-5D versions will be compared in terms of feasibility, test-retest reliability, ceiling effects, discriminative power and validity. Patient preferences for the two versions will be elicited. Content validity will be investigated in FGDs of epilepsy patients, their caregivers and through interviews with clinical staff.Meles TekieDescriptive Systems24167Ongoing20212021
222-RAPilot of the UK EQ-5D-5L TTO valuation to assess equivalence and feasibility of online interviews and face-to-face interviews during the COVID-19 pandemicObjective: Recent evidence has established the feasibility of generating time trade-off (TTO) utility values using online videoconference (video) interviews, but has not assessed whether the results are equivalent to values elicited via face-to-face in-person interviews, that were widely used prior to the COVID-19 pandemic. This study examines the equivalence, feasibility and acceptability of video and in-person interviews in generating TTO values. The study further aims to inform the choice of mode of administration for the new UK EQ-5D-5L valuation of the EQ-5D-5L and future valuation studies. Methods: Sample participants in Sheffield and Oxford, England were recruited using a blended approach of different methods to attract different people to be interviewed, and were sampled based on age, gender, ethnicity, and index of multiple deprivation. Participants were allocated to be interviewed either via video or in-person by 6 trained interviewers. Participants completed TTO tasks for the same block of 10 EQ-5D-5L health states using the EQ-VTv2 software. Feasibility, acceptability and equivalence was assessed across mode using: sample representativeness; participant understanding, engagement and feedback; participant preferred mode of interview; data quality; mean utility and distribution of values for each health state; and regression analyses assessing the impact of mode whilst controlling for the sociodemographic characteristics of participants. Results: The video and in-person samples had statistically significant differences in terms of ethnicity and income levels but were broadly similar across all other characteristics. Video interviews generated marginally lower quality data across some criteria. Participant understanding and feedback was positive and similar across modes. TTO values were similar across modes (mean and distribution for each state); whilst mean TTO values were lower for the in-person interviews for the most severe states, most noticeably the worst state, in most regression analyses the mode dummy was insignificant. There was no clear preference of mode across all individuals, though the characteristics of participants preferring to be interviewed across each mode differs. Discussion: The results demonstrate that video and in-person TTO interviews are feasible and acceptable. Both generated good-quality data, though video interviews had lower quality data across some criteria. Whilst TTO values differed across the modes for the more severe states, mode does not appear to be the cause. The sample is highly educated across both modes, and it is possible that data quality and TTO values by mode could differ in a less educated sample. The study results suggest that whilst TTO data collection using either mode is feasible, acceptable and will not in itself affect TTO values, the characteristics of people willing to be interviewed differs for each mode, and this in turn may impact on sample representativeness for some characteristics and the TTO values.Donna RowenValuation171453Completed20202022
192-VSValuation of the EQ-5D-Y in HungaryBackground: The Hungarian health technology assessment guidelines recommend the use of the EuroQol instrument family in quality-adjusted life year calculations. However, no national value set exists for the EQ-5D-Y-3L or any other youth-specific instrument. Objective: This study aims to develop a national value set of the EQ-5D-Y-3L for Hungary based on preferences of the general adult population. Methods: This study followed the international valuation protocol for the EQ-5D-Y-3L. A representative sample of the Hungarian general adult population in terms of age and gender was recruited. Overall, 996 respondents completed online discrete choice experiment (DCE) tasks and 200 respondents completed composite time-trade-off (cTTO) tasks by computer-assisted personal interviews. Adults valued hypothetical EQ-5D-Y-3L health states considering the health of a 10-year-old child. DCE data were modeled using a mixed logit model with random-correlated coefficients. Latent DCE utility estimates were mapped onto observed mean cTTO utilities using ordinary least squares regression. Results: For each domain, the value set resulted in larger utility decrements with more severe response levels. The relative importance of domains by level 3 coefficients was as follows: having pain or discomfort>feeling worried, sad or unhappy>mobility>doing usual activities>looking after myself. Overall, 12.8% of all health states had negative utilities in the Hungarian value set, with the pits state (33333) having the lowest predicted utility of -0.503. Conclusion: This study developed a national value set of the EQ-5D-Y-3L for Hungary. The value set enables to evaluate the cost-utility of health technologies for children and adolescents based on societal preferences in Hungary.Fanni RenczValuation, Youth2800Completed20202022
128-RAA global survey of HTA agencies for their views on health utility instruments and data: protocol developmentOrganizations which are most interested in using the EuroQol Group’s instruments are health technology assessment (HTA) agencies. However, to our knowledge, only limited formal ‘market’ research has been conducted to understand these users preferences with regard to the ‘science and technology’ underlying our instruments and/or their unmet needs. The ultimate goal of this research is to assess the preferences and needs of HTA agencies around the world with regard to the collection and use of health-state utility data. In this first phase of the project, we developed a study protocol for conducting a survey that can be sent to HTA agencies across the world. We first conducted a scoping review to understand the issues and challenges in collecting, evaluating, synthesizing, and using health utility data for reimbursement decision making. We then interviewed members of HTA agencies from Singapore, Indonesia, Canada, England, Norway, Australia, New Zealand, and Argentina, to understand the issues and challenges in their work with regard to the use of health utility instruments and to get their feedback on a draft survey. Based on this, we developed a survey that can be sent to HTA agencies to elicit this feedback.Luo NanOthers50400Completed20202022
188-RAA comparison of proxy 1 and proxy 2 of EQ-5D-Y: validity, reliability and responsivenessWhen a proxy assesses a patient’s PRO, the proxy can take his or her own perspective (i.e. proxy-proxy perspective) or the patient’s perspective (i.e. proxy-patient perspective). EQ-5D-Y is available in both proxy versions – Proxy-1 and Proxy-2 which represent the proxy-proxy and proxy-patient perspectives, respectively. Currently, only South African studies compared the two proxy versions of EQ-5D-Y and findings were not consistent. As a result, no definitive recommendation can be made to users of EQ-5D-Y when proxy version is needed. The primary aim of this project is to evaluate the relative merit of the two proxy versions of EQ-5D-Y in paediatric patients. The key question we intend to answer is: Which version, Proxy-1 or Proxy-2, has better agreement with the self-complete version of EQ-5D-Y? The secondary aim is to investigate the factors affecting the agreement with the self-complete versions and to assess and compare the construct validity, test-retest reliability, and responsiveness of the two proxy versions. This study will be piggybacked onto a validation of the self-complete EQ-5D-Y in Asian children with asthma or eczema. A total of 200 patients and their parents/legal guardians will be asked to complete a survey form separately at two different hospital visits. The construct validity, reliability, and responsiveness of the two proxy versions will be investigated and compared, as well as agreement with the self-complete version using Gwet’s AC, intra-class correlation coefficient (ICC), and Bland-Altman plots as appropriate. This project will provide new information on the measurement properties of the proxy-1 and proxy-2 of EQ-5D-Y.Nan LuoYouth22120Ongoing20202024
20200050Startup & support cost for the UK 5L valuation studyBernhard SlaapValuation18312Completed20202021
133-RAAssessment and comparison of the feasibility and measurement properties of the EQ-5D-Y-3L and EQ-5D-Y-5L self-complete versions in the Tigrinya language and comparison with the CHU-9DThe EuroQol Group has recently produced a new version of the EQ-5D-Y (the EQ-5D-Y-5L), which has increased the number of levels of severity in each dimension to five from the original three. Before being made widely available, the feasibility, distributional and measurement properties of the EQ-5D-Y-5L need to be assessed and compared with the standard EQ-5D-Y-3L. It is also of interest to compare their performance with that of other generic multi-attribute utility instruments (MAUI) for use in pediatric populations. The objectives of this study are to assess and compare the feasibility, distributional characteristics, discriminatory power, and measurement properties of the Tigrinya versions of the EQ-5D-Y-3L and EQ-5D-Y-5L in healthy children and children with a range of health conditions, and to compare the results with those of the CHU-9D. The EQ-5D-Y-3L and EQ-5D-Y-5L will have been previously adapted into Tigrinya (a semitic language spoken by approximately 9 million native speakers, primarily in Ethiopia and Eritrea) following the EuroQol Version Management Committee guidelines. Their feasibility, distributional properties, discriminatory power, and known groups’ and convergent validity with other, attribute-specific scales will be tested in a cross-sectional study in paediatric patients aged 8–15 years with any of 8 health conditions (n≈520) and a control group of 'healthy' schoolchildren (n≈100). Test-retest reliability and responsiveness will be assessed in a sub-group of the initial sample who will participate in a second visit 1–3 months from baseline. The Tigrinya version of the CHU-9D will be included alongside the two versions of EQ-5D-Y and the results compared.Abraham Gebregziabiher WelieYouth67070Ongoing20222023
69-RAA mixed methods approach to testing alternative recall periods for EQ-5D (2nd revision)Improving EQ-5D descriptive systems is an important task of the Descriptive System Working Group (DSWG) of the EuroQol Group, and it has been explored through increasing levels and dimensions. This project explores whether changing the time frame, or recall period, of EQ-5D may improve its sensitivity for patients whose symptoms are episodic. The project will be carried out in two phases. In phase 1, we will work with clinicians and patients to identify two new recall periods that may enhance the ability of EQ-5D to capture the effect of episodic symptoms on health. In phase 2, patients with asthma or COPD will be interviewed with EQ-5D-5L and an EQ-5D-5L variant, and the St George Respiratory Questionnaire. The EQ-5D-5L variants will be compared with EQ-5D-5L for their ceiling effects and sensitivity (using F-statistic and AUC). This project will inform the EuroQol Group of alternative recall periods that may help further improve the current EQ-5D descriptive systems.Nan LuoDescriptive Systems65280Ongoing20202022
151-RADoes rapid fluctuation of health over time affect the actual time span that is used by trauma patients when thinking of ‘your health today’?Background: The standard reference period of the EQ-5D is ‘your health today’. However, the ‘true’ reference period that is used by the respondent when filling out the EQ-5D may be shorter or longer than ‘your health today’, or may focus on specific (health) events in that time period. Patients with health conditions with rapid fluctuations of health over time as measured by other instruments or through observation, such as injuries, may use another reference period than what is asked for ('today'). Repeated measurements of the EQ-5D and cognitive debriefing items in a sample of trauma patients can give more insight into the reference period that is used, and the relation between the fluctuation pattern and the actual reference period. We will investigate these heuristic rules and the impact - if any - of specific factors, e.g. recovery patterns, and non-specific factors like age, sex and level of education. Study aims: Determine the actual reference period that was used when thinking of 'your health today' and the heuristic rule that was used to combine the full experience of that actual reference period in one overall 'signifying' number or level. Methods: The sample consist of trauma patients who filled out the EQ-5D-3L and cognitive debriefing questions at repeated measurements. Uni- and multivariate regression analysis will be used to determine if recovery pattern and patient characteristics are associated with the actual reference period and the heuristic rule. Output: A scientific paper that describes the results of this study.Juanita HaagsmaDescriptive Systems13400Ongoing20202022
82-RAValuing health in children: an examination of age, perspective and methodological effects (REVISED)This research seeks generate evidence that can be used to inform the choice of methods for obtaining values for child/adolescent health states. Primary objectives: * To evaluate whether and how the age of a hypothetical child/adolescent described within a valuation task undertaken by adults affects their valuations * To examine differences between adult health state values and child/adolescent health state values, whilst controlling for potential differences driven by other changes in perspective (own health vs. health of another individual) * To examine whether TTO and VAS lead to different patterns of responses in the context of health state values elicited using different perspectives (adult/other/child/adolescent) Secondary objectives: * To discern whether respondents’ valuations of health states elicited using different perspectives (adult/other/child/adolescent) reflect their wider views about how health care resources should be prioritised between adults and children * To generate evidence on how respondents interpret valuation tasks requiring them to adopt an adult/other/child/adolescent perspective and on the motivations and reasoning behind their responses * To examine whether the parental status of respondents (and the ages of the children they parent) affects valuations Computer-assisted personal interviews will be carried out with 300 members of the UK general public (sample will be representative of the general population in terms of age group, gender, social grade and parental status). The interview will comprise composite TTO tasks, VAS tasks, priority-setting tasks, and follow-up debrief and probing questions. The TTO and VAS tasks will be presented using four different perspectives: adult – own health; adult – other person’s health, 4-year-old child, and 10-year-old child.Koonal ShahValuation, Youth126330Ongoing20202021
124-VS**Valuing the EQ-5D-Y-3L in Belgium using the new protocol**In Belgium, health economic valuations are a mandatory part of the Health technology appraisal (HTA) process, both for adult and for paediatric indications. The guidelines for this process were developed by the Belgian Health Care Knowledge Centre (KCE). To date, only a VAS-based 3-level valuation set is available for adults in Belgium; no value sets for any HRQoL instruments for children exist in Belgium. The KCE encourages a youth valuation study and will incorporate its results into the HTA guidelines. Recently, a new youth valuation protocol has been proposed for developing country-specific EQ-5D-Y value sets. The primary aim of this study is to conduct a EQ-5D-Y valuation study for Belgium using this new protocol, including an online panel to collect DCE data (n=1,000) and face-to-face TTO interviews (n=200) for anchoring the DCE data. This study also includes a methodological arm to assess whether the valuation differs between children and adolescents. For this second study aim, new respondents will be recruited to value 14 DCE pairs: 7 from a child and 7 from an adolescent perspective. A new, efficient design will be developed and we calculated that 500 respondents would be sufficient to estimate a main-effects model and to test for the impact of perspective on the value set. This double-pronged project will fill in knowledge gaps in youth valuation, firstly by estimating a EQ-5D-Y value set reflecting the health preferences of the Belgian general population. And secondly, by furthering our understanding of a possible age-related effect within this paediatric population.Dewilde SarahValuation, Youth138130Ongoing20202021
148-RACombining health and social outcomes using the EQ-5D-5L and the ASCOT – development of a pilot value setTo allocate scarce health care resources equitably across a population we have to be able to measure and value all the outcomes that matter to the population. Current measurement systems focus mostly on health, and rarely on both health and social outcomes together. This project provides a novel solution to combining the measurement of health and social outcomes, and to valuing them on the same scale. It will deliver a new and original approach to combine health and social dimensions and value them on the same scale. A key advantage of our approach is that it is based on combining the EQ-5D-5L with another well-established instrument – the ASCOT – to measure and value health-related quality of life and social care-related quality of life together. The advantage of such an approach is that it uses the EQ-5D-5L as the basis for combining the domains of health and social care. This builds on earlier EuroQol funded work by the project team examining preferences for the EQ-5D-5L and ASCOT dimensions, using a discrete choice experiment approach. In this study we will investigate the overlap between EQ-5D-5L using psychometric methods, develop a descriptive system based on both instruments, and then pilot a DCE approach to valuing that descriptive system. We will use the approach to develop a value set that is based on combining items from the ASCOT and EQ-5D-5L.Rosalie VineyDescriptive Systems, Valuation55540Ongoing20202021
83-VSAn Australian Value Set for the EQ-5D-YThe aim of the work is to develop an Australian value set for the 3-level EQ-5D-Y. Australia is a major market for EuroQol, with widespread use of generic instruments. While the EQ-5D-Y is used, other pediatric instruments (particularly the CHU-9D) are more widely employed, largely because they have existing value sets. This project will develop this value set for the EQ-5D-Y-3L using the recently accepted protocol, combining face-to-face TTO tasks and online DCE. Interviews will take place in three Australian centres/states (Perth (Western Australia), Melbourne (Victoria), and Sydney (New South Wales)), and in regional centres in each of the three states. Rigorous quality control methods will be used to minimise interviewer effects. We will extend the standard methodology by allowing exploration of different ways of anchoring the latent scale DCE data using the TTO data (or not). This will be done using the TTO data exclusively (either the worst health state 33333 or all the TTO data), or by including additional paired DCE tasks containing either duration or dead, both of which can be used as anchoring devices.Richard NormanValuation, Youth135918Ongoing20202023
111-RADeveloping and testing a version of EQ-5D-Y for use in children aged 2-5 years using a mixed methods approach**Background**: Few HRQoL measures are available for estimation of child QALYs with a nearly clear absence of measures for children under 5 years, despite younger children being greater consumers of health care services than older children. The EQ-5D-Y is recommended for children 4-7 years using parent proxy, but there has been relatively little research into its use for children under 4 years. There could be a key opportunity to extend the use of EQ-5D-Y for proxy completion for children aged 2-4 years. **Aims**: (a) To establish if the EQ-5D-Y can be used for children 2-4 years and to co-design a suitable adaptation for testing with parents/carers. (b) To validate the adapted EQ-5D-Y with proxy completers alongside CHU9D and PedsQL. **Methods**: Two focus groups with parents/carers of well and sick children recruited from the Royal Children’s Hospital Melbourne will be conducted, to explore EQ-5D-Y use in children 2-4 years and to inform the design of adaptations. A survey will then be administered to 400 proxy completers for children 2-4 years including 1) adapted version of EQ-5D-Y, 2) the PedsQL 2-4 years, 3) CHU9D for under 5 years, and 4) questions on presentation and comprehension. The survey will be re-sent to the same participants 2 weeks later. Assessment will be made of acceptability, version preference, convergent validity, known group validity, reliability and responsiveness, and instrument comparisons. **Deliverables**: a report with recommendations about whether EQ-5D-Y can be used for children 2-4 years and proposed adaptations.Kim DalzielYouth87200Ongoing20202022
92-2020RAInvestigating the effect of interaction terms in modelling EQ-5D value set and its impact on sample size requirements using three VAS saturated dataA recent report filed by the EEPRU raised several issues regarding the design, data quality and modelling of the UK EQ-5D-5L study. Among all the issues raised, the most prominent one is the directly valued TTO states only covered less than 3% of all possible 3,125 health states despite the fact that other instruments used even smaller proportions of states to develop their value sets. Historically, the valuation study of EQ-5D has favoured the design of a small set of health states with sufficient responses per health state, e.g. in the case of current EQ-VT design, 80 unique states with 100 responses per health state (excluding the 5 mildest states and state 55555). A main consideration of this design choice is that with 100 responses per health state, the observed means are robust. However, a limitation of this design is the small coverage limited the possibility of identifying higher order interaction term as questioned by the EEPRU report. In this study, we aim to explore the interaction effects in three existing EQ-5D VAS saturated dataset (one 5L and two 3L). We will 1) investigate models with interaction terms in the full data of 3 saturated EQ-5D respectively; 2) investigate the possible consequences for sample size if we want to use the interaction model, i.e., we will gradually lower the number responses/states to the point where the interaction model identified in the full data can be constructed while the number of responses/states are minimal.Zhihao YangValuation30580Ongoing20202023
100-RAValuation of the EQ-5D-3L-Y in the Netherlands and an investigation on different proxy perspectivesBackground: There is increasing interest in preference-accompanied measures of health for paediatric populations. The EQ-5D-Y-3L is one of such instruments, but the lack of a Dutch value set prevents its use in economic evaluations of health care interventions in the Netherlands. Furthermore, EQ-5D-Y-3L health states are commonly valued from the perspective of a 10-year old child. The use of this proxy perspective (henceforth called proxy-proxy) has been a source of discussion. Other perspectives could also be used, such as a substitute perspective (henceforth called proxy-substitute): i.e. adults considering what they think a 10-year old child would decide for itself. Objectives: Our main objective was to derive a value set for the EQ-5D-Y-3L for the Netherlands. Furthermore, we explore how the outcomes, dispersion and response patterns of composite time trade-off (cTTO) valuation differs between proxy-substitute and proxy-proxy perspectives. Methods: Composite Time Trade Off (cTTO) data were collected using videoconferencing interviews, with each respondent completing 10 cTTO tasks. A target sample of 400 respondents was collected, with half of the respondents being randomized to the proxy-proxy arm, and the other half to the proxy-substitute arm. Discrete Choice Experiment (DCE) data were collected using an online survey, with a target sample of 1000 respondents each completing 15 paired comparisons. DCE data were analysed using a 10-parameter mixed logit model and anchored to the QALY scale using the mean observed cTTO values. Results: The level 3 weight for pain/discomfort was the largest, followed by feeling worried, sad or unhappy, usual activities, mobility and self-care. Health state values ranged between 1 and -0.218. The use of proxy-proxy and proxy-substitute preferences yielded different EQ-5D-Y-3L valuation outcomes. For states in which children had a lot of pain and were very worried, sad or unhappy, respondents’ valuations were lower in proxy-proxy rather than proxy-substitute preferences. (by about 0.2). Within-subjects variation across health states was lower for proxy-substitute preferences than proxy-proxy preferences. Analyses of response patterns suggest that data for proxy-substitute preferences were less clustered. Conclusions: This study generated a Dutch value set for the EQ-5D-Y-3L, which can be used for the computation of quality-adjusted life years (QALY’s) for economic evaluations of healthcare interventions in paediatric populations. There are systematic differences between cTTO responses given by adults deciding for children and adults considering what children would want for themselves. Besides warranting further qualitative exploration, such differences contribute to the ongoing normative discussion surrounding the source and perspective used for valuation of child and adolescent health.Bram RoudijkValuation, Youth81427Completed20202021
85-RAComparing the psychometric properties of EQ-5D-Y-5L and EQ-5D-Y-3L in children with osteogenesis imperfecta in ChinaThe objective of this study was to evaluate and compare the psychometric properties of the EQ-5D-Y-3L, EQ-5D-Y-5L, and CHU-9D with the “standard” instrument, PedsQL, in a sample of children and adolescents with Osteogenesis Imperfecta (OI). A web-based cross-sectional survey of pediatric OI patients was conducted to collect data. The EQ-5D-Y-3L, EQ-5D-Y-5L, CHU-9D, and PedsQL were used to assess the health-related quality of life of the participants. Construct validity, including convergent and divergent validity, known-group validity, and test-retest reliability, were examined to confirm the psychometric properties of the instruments. A total of 157 children and adolescents with OI participated in the study. Few samples reported full health status. A strong ceiling effect was observed for all EQ-5D-Y and most CHU-9D dimensions. Both EQ-5D-Y and CHU-9D showed statistically significant correlations with the corresponding PedsQL subscales. A strong correlation was also identified between EQ-5D-Y-3L and EQ-5D-Y-5L. The test-retest reliability for the EQ-5D-Y-3L, EQ-5D-Y-5L, and CHU-9D was acceptable. The EQ-5D-Y instruments showed a better known-group validity than CHU-9D in differentiating patients in different risk groups. The results show that the EQ-5D-Y instruments and CHU-9D are reliable and valid and that the EQ-5D-Y-3L performed slightly better than EQ-5D-Y-5L regarding convergent validity and its ability to discriminate.Richard Huan XUYouth16295Completed20202022
137-RANavigating antithrombotic therapies with the EQ-5D: An analysis of the COMPASS TrialDespite the the inclusion of the EQ-5D in many RCTs, published evidence on measuring treatment effects using the EQ-5D in these clinical trials has been sparse. The Cardiovascular Outcomes for People Using Anticoagulation Strategies (COMPASS) trial is a multinational RCT that evaluated efficacy and safety of antithrombotic therapies with 27,395 cardiovascular disease (CVD) patients enrolled at 602 centres in 33 countries. In the trial, the intervention group with rivaroxaban combined with aspirin improved the composite clinical endpoint that consists of cardiovascular death, stroke, or myocardial infarction, but it also increased the risk of major bleeding. Net clinical benefit analyses demontrated a 22% reduction in a composite of vascular and bleeding outcomes, but this did not take into accout the patients preference. The COMPASS trial provides an ideal clinical context to demonstrate the value of the EQ-5D as a preference-based measure of patient-reported outcomes. The EQ-5D measures the overall impact of both treatment benefits (decrease in CV events) and harms (bleeding) on patient’s health-related quality of life. We propose to analyze the EQ-5D data collected at baseline and at the two-years follow-up during the COMPASS trial. We will estimate the treatment effects on EQ-derived health utilities as well as the utility decrements associated with major clinical events to support economic evaluations. This proposed analysis will generate imporant evidence using the EQ-5D to faciliate the interpretation of clinical findings.Feng XieOthers22740Ongoing20202020
79-RAThe psychometric properties, feasibility and usefulness of the EQ-5D-5L in Ethiopian stroke patients- A mixed-methods longitudinal studyStroke is the second leading cause of death globally and significantly impacts the health-related quality of life (HRQoL) of survivor. While there are some established longitudinal studies of stroke patients internationally, little is known about health-related quality of life (HRQoL) outcomes following the first-ever stroke in Ethiopia (classified as a low-income country), where the population exceeds 110 million people. Therefore, this study aimed: 1) to assess the feasibility of administering the EQ-5D-5L to patients and their proxy respondents during hospital admission and again 3 and 6 months following an acute onset illness (first-ever stroke); 2) to examine further the reliability and validity of the EQ-5D-5L applied to stroke patients and their proxy; 3) to examine patient-proxy agreement on the domains and summary scores of the EQ-5D-5L; and 4) to qualitatively investigate, within patient-proxy dyads, patient and proxy perspectives of the reasons for any differences in HRQoL reporting on the descriptive profile between patients and their proxies. At baseline, 200 patient-proxy dyads were recruited between November 2021 and November 2022, and in the second and third visits, 152 (76%) and 133 (66.5%) patient-proxy dyads were included. At baseline, hospital admitted patients with ischemic stroke (n=128; 64%), hemorrhagic stroke (n=66; 33%), and cardioembolic stroke (n=6; 3%) were recruited. The Amharic version of self-completed and interviewer-administered EQ-5D-5L were completed by the stroke patients and the proxy self-complete and interviewer administered versions were completed by their proxy respondents. At all three time points participants (both patients and proxies) preferred the interviewer-administered version of the tool. Demographic and clinical data were collected at baseline, and the degree of disability or dependence in the daily activities of stroke patients was measured using the Modified Rankin Scale (mRs) at baseline and follow up visits. A qualitative interview was done with 15 dyads to investigate potential explanations for the observed slight discrepancy in EQ-5D profiles between patients and proxies at three time points of an interview. The EQ-5D-5L response distribution demonstrated that higher proportion of patients and proxies reported problems on usual activities (UA) and mobility (MO) dimensions while fewer problems were reported on selfcare, anxiety/depression (A/D) and pain/discomfort (P/D) dimension regardless of the visit period. The proportion of reported no problems (level 1)/problems (level 2 to 5) for each dimension of EQ-5D-5L were compared across the interview period between patient and proxy using the chi-square test. No significant difference in reported no problems/problems proportion was observed across all dimensions. Weighted Cohen’s Kappa statistics were computed to compare the agreement between patient and proxy rating for each dimension of EQ-5D-5L at baseline and follow up visits. At baseline, a lower agreement was observed for P/D (0.32) and A/D (0.43) dimensions, while a higher agreement was observed for MO and selfcare (SC) dimensions. In all dimensions, an improved agreement was observed in visits two and three compared to the baseline report. Differences between patient and proxy utility scores (based on EQ-5D index and EQ VAS) were assessed with paired t-tests at baseline and follow up visits. No significant difference in mean EQ-5D index and EQ VAS was observed between patient and proxy at baseline and visit two; however, at the third visit a significant difference was observed in EQ-5D index or EQ-VAS score with t (131) = -0.88, p = .01 and t (126) = -2.68, p = .008 respectively. EQ-5D index reported by a patient at baseline shows a strong correlation (r=0.88) with the EQ-5D index reported by a caregiver at baseline; however, moderate correlation was observed with EQ VAS score (r=0.7 for patients and r=0.66 for caregiver). The EQ VAS and EQ-5D index showed a moderate to strong correlation with the mRs score. The mean utility based on the type of stroke was reported in which higher mean utility was observed in cardioembolic stroke than in ischemic and hemorrhagic stroke; however, the difference was not significant based on the type of stroke. A significant difference in mean utility was observed among clients with different levels of stroke disability based on the mRs score (good known group validity). Overall, there was no significant difference between the patient and proxy-reported EQ-5D-5L profiles or utility values at admission. There was strong agreement/correlation in reported values in admission and follow up visits.Yared Belete BelayDescriptive Systems24000Ongoing20202021
157-RATowards a patient-reported summary score for EQ-5D - revision (20190210)One of the greatest strengths of the EQ-5D lies in the available of country-specific value sets, facilitating the calculation of quality-adjusted life years (QALYs). EQ-5D value sets are specifically designed for that purpose, which is reflected both in the underlying methodology (stated preferences) and whose preferences are sought (by convention the general public). For purposes other than the estimation of QALYs, there is no clear rationale for using any value set to summarize profile data. However, it seems that for these (‘non-QALY’) purposes values are often used as a means of ‘merely’ summarizing scores for the five dimensions of EQ-5D. The aim of this study is to explore simple methods to develop summary scores for EQ-5D for non-economic (evaluation) purposes, such as comparing and monitoring population health, monitoring patients’ health and aiding shared decision making. We will calculate patient-reported summary scores using dimension-specific rating scales for each of the five EQ-5D dimensions (both 3L and 5L) based on two large international datasets. The rating scale methodology provides a tested psychometric approach, allowing for a more refined assessment of the underlying position of the level responses for each dimension. Several weighting approaches to aggregate the five rating scale scores into a single summary score will be tested, using the overall EQ-VAS and by applying psychometric scaling approaches based on the level scores. The resulting summary scores will be tested in a number of external datasets for various measurement properties, including distributional properties, convergent validity, and responsiveness.Bas JanssenPopulations and Health Systems68890Ongoing20202023
76-RAA PhD grant to investigate the valuation of worse-than-dead health statesHealth utility is measured on an interval scale anchored by 0 and 1, where 0 corresponds to being dead, 1 full health, and negative values worse-than-dead (WTD) health states. While health-state valuation methods have advanced significantly in the past decades, there are still many unresolved issues in estimating the utility of WTD health states. Some prominent ones are: 1) lack of a uniform method for eliciting better and worse than dead states; 2) lack of a theoretically sound method for bounding negative values; 3) maximum endurable time; and 4) lack of sensitive elicitation methods. We propose a PhD project to systematically investigate the issue of insensitive valuation methods for WTD health states. The overall aim of the project is: 1) to ascertain the reasons for the challenges in eliciting negative utility; and 2) to develop and test new utility elicitation methods for WTD health states. Given that the cTTO used in the EQ-VT for estimating EQ-5D-5L value sets represents the state-of-the-art health-state utility elicitation method, this project will focus on analyzing the measurement issues of cTTO and testing new designs that may improve its performance. Four studies will be conducted: 1) a scoping review of articles on the valuation of WTD health states; 2) an investigation into states valued at -1 in cTTO tasks using an additional question; 3) a qualitative study of the thought processes for valuation of WTD health states using cTTO; and 4) developing new TTO variants and testing their performance in valuing WTD health states.Nan LuoValuation150000Ongoing20212025
104-RAUsing TTO data to anchor DCE data and produce EQ-5D-Y value sets: comparing alternative approaches using existing and simulated dataThe valuation protocol for the EQ-5D-Y-3L recommends that both discrete choice experiment (DCE) data and composite time trade-off (cTTO) data are collected. The DCE provides the key information on the relative importance of the different levels and dimensions, and the cTTO is intended to provide data to anchor the DCE data onto the full health to dead scale. However, there are multiple ways in which to conduct this anchoring. These include rescaling based on the mean cTTO values ("worst state rescaling"), mapping DCE onto mean cTTO values ("linear mapping"), and hybrid modelling. To date, there has been little research into the relative merit of these approaches. This research aimed to fill these gaps by comparing the different anchoring approaches using previously collected EQ-5D-Y-3L valuation data. Overall, it was found that, whilst the value sets produced by each anchoring method are similar, there are some notable differences which are generalisable. For example: 1) Both worst state rescaling and linear mapping will maintain the relative importance scores from the DCE data as they are linear transformations of the DCE data. 2) Worst state rescaling will produce value sets with a larger scale than both linear mapping and hybrid modelling due to prediction error. 3) Hybrid modelling and linear mapping will produce value sets that better align with the cTTO data than worst state rescaling as they use more cTTO data. Based on the current EQ-5D-Y valuation protocol, the linear mapping approach is likely to be viewed as advantageous over the worst state rescaling approach (makes better use of the data collected) and the hybrid modelling approach (insufficient data). However, the final decision on which approach to use is likely to involve some normative considerations. To aid decisions, researchers should engage with relevant stakeholders and test the sensitivity of their results based on different anchoring approaches.David MottValuation, Youth24860Completed20202020
134-RAValuing well-being alongside health: What can and should be done? Project number 20190750 (Revised)Recent research has focussed on widening the classification system used to measure health to capture aspects beyond health to focus on dimensions such as quality of life. This presents challenges for the methods used to elicit utility values as well as around the scope of the QALY itself and its role in informing cost-effectiveness analyses. This PhD will contribute to methodological developments and understanding in the area of eliciting preferences for states that combine both health and wellbeing and that may contain a large number of both overlapping and independent dimensions. The feasibility and appropriateness of different preference elicitation methods for use to value a widened classification system have not been fully explored. The PhD will involve a literature review and primary research using a mixed methods approach involving both qualitative and quantitative research. In particular, the PhD will explore how discrete choice experiments (DCE) can be designed and undertaken to meet the challenges raised through a widened classification system including overlapping concepts and a large number of dimensions (potentially beyond the number that can be considered all at the same by a research participant). This will include both quantitative aspects of design alongside qualitative exploration of how participants undertake the DCE tasks, what influences their choices, and whether they understand the tasks. Towards the end of the project an online survey will be used to determine the feasibility of the use of DCE using a design building upon the findings from the qualitative research.John BrazierValuation132598Ongoing20202024
96-EOKnowledge translation for the use of EQ-5D as a PROM for routine outcome measurement in health systemsOur aims with this knowledge translation (KT) project were to synthesize and disseminate our key messages on the use of EQ-5D as a PROM for routine outcome measurement in health systems, and the use of routinely collected EQ-5D data to support decision-making at various levels within the system. Specific aims were to: • Increase awareness about the use of EQ-5D as a PROM in health systems, and highlight key implementation approaches and considerations. • Showcase applications of EQ-5D implementation within the health system, and highlight implementation methods, challenges and lessons learned. • Provide guidance on the use and reporting of routinely collected EQ-5D data to support decision-making at the micro, meso, and macro levels within the health system. • Highlight key methodological considerations and challenges in the use of routinely collected EQ-5D data. • Engage the community of EQ-5D end-users in guiding knowledge and research development around the use of EQ-5D as a PROM in large-scale applications.Fatima Al SayahPopulations and Health Systems, Education and Outreach24116Completed20202021
89-RAInvestigating the aspects of HRQoL covered by pain/discomfort and the added value of the psoriasis bolt-ons (EQ-PSO) among patients suffering from skin diseases (Revised).Introduction: Dermatologic conditions do not constitute a direct threat to life, but their chronic and incurable character has a negative impact on the health related quality of life (HRQoL). In the EQ-5D, physical discomfort is captured by the pain/discomfort item, while pain is clearly a very important aspect of physical discomfort, other aspects such as itching, skin irritation and the indirect impact of this on skin appearance, self-confidence, and relationship difficulties are not explicitly mentioned. The objective of this study is to investigate how well aspects of HRQoL are captured by the EQ-5D descriptive system, and more specifically by the pain-discomfort item, and the additional advantage of using the psoriasis bolt-ons (EQ-PSO questionare) in people with skin diseases. Method: First in-depth qualitative interview with 30 patients will be undertaken to investigate how well the health aspects important for people with skin diseases (psoriasis) are captured by the EQ-5D descriptive system, more specifically by the pain-discomfort item and the additional advantage of using EQ-PSO questionare. Both EQ-5D-5L and EQ-PSO questionnaire will be administered to participants. The qualitative interviews will be recorded using an encrypted audio recording device, transcribed and translated. For the analysis, statements will be grouped into themes. Subsequently, thematic content analysis will be used to investigate further issues. Findings from the interviews will be discussed in FGDs with participants who did the interview to make sure their concern is captured, to feed off each other’s ideas, to get useful information that individual interviews does not provide.Abraham GebregziabiherDescriptive Systems23960Ongoing20222023
91-RAEQ-5D-5L in productivity assessment according to the type of occupationEQ-5D-5L is a descriptive system, often used to describe health states and evaluate treatment effects. Such measurements converted into utilities can be used in cost-utility analyses. In many countries, not only the direct costs, but also the costs of reduced productivity are accounted for (aka indirect costs). Indirect costs are generated by absenteeism (a worker missing from work) or presenteeism (a worker present but less productive). Hence, for cost-utility analyses to be reliable, it is important to understand how health states affect productivity. Krol et al. (2014) showed that productivity loss (presenteeism and absenteeism) can be predicted by EQ-5D-3L description. Authors, however, indicate that the results may differ between the types of occupation (also raised by Lamers et al., 2005, and Brouwer et al., 2005). In the proposed research, we would like to assess how the productivity loss (absenteeism and presenteeism) changes with EQ-5D-5L, also accounting for the type of occupation. We plan to conduct a survey in which we will ask respondents about their type of occupation, health status within EQ-5D-5L, number of hours missed from work due to illness and degree in which illness affected their productivity while they were at work. In the main part of the survey, the respondents will evaluate hypothetical health states and their impact on ability to work and predicted productivity. The results will show whether EQ-5D-5L is more sensitive than EQ-5D-3L in measuring impact on productivity and will enable using EQ-5D-5L (also historically collected data) to measure indirect cost of illnesses.Beata KońDescriptive Systems24226Completed20202021
109-RAImproving predictive precision in valuation studies using non-parametric techniquesBackground: Health utilities derived from value sets for the EQ-5D-5L are commonly used in economic evaluations, however the precision of the value sets is of the same order of magnitude as reported minimum important differences (MIDs), which typically range from 0.05 to 0.1. We examined whether modelling spatial correlation among health states could improve the precision of the value sets. Methods: Using data from 7 EQ-5D-5L valuation studies (Canada, China, Germany, Indonesia, Japan, Korea and the Netherlands) we compared the predictive precision of the published linear model, a recently proposed 8-parameter level-scale model, and two Bayesian models with spatial correlation. Predictive precision was quantified through the root mean squared error (RMSE) for out-of-sample predictions of state-level mean utilities on omitting individual states, as well as omitting blocks of states. Results: In all seven countries, on omitting single health states, Bayesian models with spatial correlation improved upon the published linear model: the RMSEs for the originally published models were 0.060, 0.055, 0.060, 0.061, 0.039, 0.050 and 0.087 for Canada, China, Germany, Indonesia, Japan, Korea and the Netherlands respectively, and could be reduced to 0.044, 0.049, 0.051, 0.053, 0.037 0.037 and 0.086 respectively on using spatial correlation. On omitting blocks of health states, Bayesian models with spatial correlation led to smaller RMSEs in just one country, while the 8-parameter model led to smaller RMSEs in 5 of the 7 countries. Discussion: Bayesian models incorporating spatial correlation and the 8-parameter models offer promising approaches to improving the precision of value sets for the EQ-5D-5L. The differential performance of the Bayesian models on omitting single states compared to omitting blocks of states suggests that designing valuation studies to capture more health states may further improve precision. We suggest that Bayesian models with spatial correlation and 8-parameter models be considered as candidate models when creating value sets, and that alternative designs be explored; this is vital given that the prediction errors in value sets need to be smaller than the MID of the instrument.Eleanor PullenayegumValuation32000Completed20202021
138-RAPreference heterogeneity in health valuation: Peru EQ-VT dataBackground: Preference heterogeneity in health valuation has become a topic of greater discussion among health technology assessment agencies. To better understand heterogeneity within a national population, valuation studies may identify latent groups that place different absolute and relative importance (i.e., scale and taste parameters) on the attributes of health profiles. Objective: Using discrete choice responses from a Peruvian valuation study, we estimated EQ-5D-5L values on a quality-adjusted life-year (QALY) scale accounting for latent heterogeneity in scale and taste, as well as controlling heteroskedasticity at task level variation. Method: We conducted a series of latent class analyses, each including the 20 main effects of the EQ-5D-5L and a power function that relaxes the constant proportionality assumption (i.e., discounting). Taste class membership was conditional on respondent-specific characteristics as well as their experience with the time trade-off tasks. Scale class membership was conditional on behavioral characteristics such as survey duration and self-stated difficulty level in understanding tasks. Each analysis allowed the scale factor to vary by task type and time (i.e., heteroskedasticity). Results: The results indicate three taste classes: a quality-of-life oriented class (33.35%) that placed the highest value on levels of severity, a length-of-life oriented class (26.72%) that placed the highest value on life span, and a middle class (39.71%) with health attribute effects lower than the quality class and life span effect lower than the length-of-life oriented class. The EQ-5D-5L values ranged from -2.11 to 0.86 (quality-of-life oriented class), from -0.38 to 1.02 (middle class), and from 0.36 to 1.01 (length-of-life oriented class). The likelihood of being a member of the quality-of-life class was highly dependent on whether the respondent completed the time-tradeoff tasks (p-value <0.001). The results also show two scale classes as well as heteroskedasticity within each scale class. Conclusion: Accounting for taste and scale classes simultaneously improves understanding preference heterogeneity in health valuation. Future studies may confirm the differences in taste between classes in terms of the effect of quality of life and life span attributes. Furthermore, confirmatory evidence is needed on how behavioral variables captured within a study protocol may enhance analyses of preference heterogeneity.Suzana KarimValuation24480Completed20202022
119-RAA qualitative study on the content validity of the EQ-5D-5L and EQ-PSO bolt-on in patients with psoriasis in HungaryObjectives: A number of bolt-ons have been proposed for the EQ-5D, including two psoriasis specific bolt-ons, skin irritation and self-confidence. The study investigates and compares the relevance and comprehensiveness of these psoriasis specific bolts and the EQ-5D-5L and explores the potential conceptual overlaps between the existing five dimensions and the two bolt-ons. Methods: Psoriasis patients were purposively sampled according to age and gender. Semi-structured interviews, where participants were asked to complete the EQ-5D-5L, the EQ VAS and the bolt-ons while thinking aloud, were conducted. Probes were used to investigate the thought processes of patients regarding the dimensions, wording, recall period and relevant concepts not captured by the EQ-5D-5L and bolt-ons. Data were analysed thematically. A focus group was used to confirm the findings. Results: Overall, 16 patients completed the interviews. Sixteen and fifteen patients considered skin irritation and self-confidence relevant areas to describe psoriasis problems. Three patients considered itching a form of discomfort, and thus, pointed out a potential overlap between pain/discomfort and skin irritation. Twelve patients reported overall 10 general health- or psoriasis-related concepts that are not captured by the EQ-5D-5L, including itching, social relationships and sex life. Eleven patients reported that the recall period of the EQ-5D-5L might be subject to bias because of the daily or within-day fluctuations of their symptoms. Conclusions: The skin irritation and self-confidence bolt-ons are particularly pertinent and contribute to improve content validity of the EQ-5D-5L in patients with psoriasis. There is only a minor conceptual overlap between the pain/discomfort and skin irritation dimensions.Fanni RenczDescriptive Systems24990Completed20202021
127-RADeveloping Scoring Methods for the 25-item EQALY InstrumentThe main aim of this project was to develop a non-preference scoring system for the EQ-HWB and EQ-HWB-S using exploratory item response theory, confirmatory factor analysis and confirmatory item response theory. Datasets from the EQ-HWB psychometric studies conducted in United States, Australia, and United Kingdom were used. Developing a scoring system is especially important for the 25-item EQ-HWB which is not easily scored using standard preference elicitation techniques. We have been able to prove the validity of using a level sum score (LSS) for a single score as well as two subscales of the EQ-HWB-S. Although we were able to rule out certain models for the EQ-HWB, such as modelling the positive and negative items separately, , the original theoretical structure of the items and the 1- and 2-factor model for the EQ-HWB, it is less clear which of the other investigated models is the most suited for scoring. It was determined that a 3- and a 6-factor model were the most recommended. However, due to the limitations of the data used for these analyses, it was determined that we cannot firmly recommend a scoring structure for the EQ-HWB until the 3- and 6-factor models can be tested using data based on the current version of the EQ-HWB.You-Shan FengDescriptive Systems, EQ-HWB81430Completed20202021
84-RAAssessing the impact of COVID-19 on population health using the longitudinal panel surveys in the US, Sweden and Norway1. The EQ-5D-5L is sensitive to capture the HRQoL changes over time in measuring the impact of an unprecedented infectious disease pandemic with a prolonged duration (2 years to date and on-going) 2. The EQ-5D-5L was able to differentiate the HRQoL differences between groups, for example, young adults vs. older adults, thus rendering important public health policy recommendations on the needed attention for the younger population in the US. 3. EQ-5D-5L utility values enabled the estiamtion of over 1 million lives lost due to COVID-19's negative impact on the youth population in the US. Public health policy should consider the mental health impact of the policy on the youth popualtion 4. It is important to understand people's health behavior under uncertainty, including people's perceptions and acceptability of mandate interventions such as vaccinationNing Yan GuPopulations and Health Systems24480Completed20202020
2016030A PROMs based patient decision aid for patients considering total knee arthroplasty: development and a pilot randomized controlled trial.The proposition is that patients do not have realistic expectations about the outcomes of total knee arthroplasty (TKA)and might change their decision aboutsurgery if they were better informed prior to the surgeon consult. Unrealistic expectations can lead to some patients having surgery that they later regret. We propose using routinely-collected Patient Reported Outcome Measures (PROMs),including the EQ-5D,in patients who have previously considered TKA surgery. We believe this information can better set patients’ expectations, leading to improved quality decisionsand more appropriate surgeries. For example, since patientswith less severe pain systematically overestimate the amount they would benefit from surgery, we hypothesize the decision aids would lead to some patients delayingsurgery.We will test this hypothesis in two phases: 1) Finalizethe development of a PROMs decision aid for patients considering TKA (originally developed with funding from EuroQol)to ensure it is acceptable and understandable to patients and providersandcollect PROMs in patients choosing non surgical management;and 2) Conduct a pilot randomized trial to evaluate the feasbility and potential effectiveness(based on decision quality) of the decision aidvs usual carein 280 patients considering TKA. Ultimately, the pilot findings will inform a future multicentre implementation trial which would evaluate the impact of the decision aid being used province-wide in Alberta. This research could have importanthealth policy consequences, as health care systems continue to struggle with long waiting timesfor TKA.With the trend for more and younger patients with less severe pain and better function demanding surgery, it is critical to identify appropriate surgical candidates. If successful, our studywould demonstrate the role of the EQ-5D in helping guide patient and physician decisions.Nick BansbackOthers210714Completed20152017
20190510Design and Analysis Considerations when using the EQ-5D alongside clinical trials or observational studies for economic evaluation: PhD studentship and development of tools for analysts and researchersThe EQ-5D instrument is widely used alongside randomized controlled trials and as part of observational studies to inform the development of cost effectiveness models. Perhaps surprisingly very little guidance has emerged on design and analysis considerations for analysts who are interested in employing the EQ-5D instrument as part of their study. Trial sponsors commonly analyse EQ-5D by trial arm and report that there is no difference in HRQL (sometimes in contrast to findings for other PROs). But the structure of economic models is often described in terms of discrete disease states over a patients’ lifetime. Health economic modellers have long recognized the limitations of standard approaches to trial data analysis and have instead employed regression modellingto understand the impact of events and health states on utility. Prominent examples include the analysis of the UKPDS trial1; work in COPD 2; oncology3, and cardiovascular disease4. We believe a lot more work can be usefully conducted to provide guidance on a wide range of issues including modelling approaches, handling patient drop out and providing guidance regarding the optimal design of studies designed to measure the impact of events at random points in time. This application is seeking funding for a studentship at the London School of Hygiene & Tropical Medicine to explore this work using available datasets.Andrew LoydPopulations and Health Systems156995Ongoing20212023
20200020Establishment of a UK EuroQol Group and initial 1 day meetingWe propose to organise a one-daymeeting in the UK for researchers with an interest in the development and application of theEQ-5Dor other EuroQol measures. The meeting is designed to be researcher-led and aimed at UK and Ireland based researchers primarily. This is a day to discuss EQ-5D related research issues (e.g. DCE/TTO valuation, E-QALY development, non-economic applications of EQ-5D etc , as examples) rather than the use of the tool in clinical studies. The intention and purpose here is to 1.Support the development of early career researchers/ students who are interested in EQ-5D application & methods2.Provide a forum for non-EuroQol members to discuss and debate their EQ-5D related research (with a focus on methods rather than use of the tool)3.Provide a day for academic researchers, NICE/ SMC employees, industry and others to networkand so therefore facilitate more EQ-5D related research in the UKLouise LongworthEducation and Outreach18180Ongoing20202022
20200010Describing the Worse than Dead in Youth ValuationPurposeTrading life years for children is difficult, not only in the context of Time Trade Off valuation task but also in a real-life situation, where adults have to take medical decision for children which involve a trade-off between mortality and changes in quality of life. This difficult situation captured in the preliminary result of the Indonesian EQ-5D Youth valuation study using cTTO. People tend to be reluctant to give up life years for children, indicated by few worse than dead responses (only 10% from total observations), and wide differences of mean and SD for the most severe state compare to the other bad health states. Due to the limited number of severe states in Indonesian EQ-5D Youth valuation study, firm conclusion of adult’s reluctance to trade-off in trading life years for children could not be well-established. Understanding the reasoning behind adult’s decision in trading life years for children will lead to improvement in the youth valuation study design in the near future. Therefore, the aims of this study are: a) to reinforce the finding from Indonesian EQ-5D-Y valuation study by collecting additional data using cTTO arranged dominantly with severe health states and b) explore respondent’s thought process when trading life years for children (or being reluctant), particularly in the worse-than-dead values. MethodsFifty respondents with an age above 17years old in rural and urban area at Jakarta and Bandung (Indonesia) will be invited to participate in this study. Interview will consist of standard cTTO interview, then followed by semi-structured qualitative interview. To make it comparable with the previous study, the cTTO will consist of 12 health states whereas half of them will have misery index greater than 12. This design is chosen to provoke a higher number of worse-than-dead responses so that respondent’s thought process behind worse-than-dead responses could be betterunderstood. The qualitative interview will be focus in the background thinking of the worse-than-dead responses. Quantitative data from cTTO interview will be analyse using STATA and qualitative data will be analyse using content and thematic analysis.Titi FitrianiValuation, Youth13366Ongoing20202021
20191070A comparison of methods to evaluate DCE response qualityProvide a brief summary of your research proposal, including the aim and general methods proposed (max 250 words). Please note that the abstract of research will be posted on the Members’ Area of the EuroQol website for successful applications. Aims:Toprovide an overview of all methods used to evaluate DCE response qualityand toassess the relative performance and feasibility of these methods when applied to EQ-5D-5Ldiscrete choice data.Methods:A structured review of DCE research published in the past threeyears (2017-2019) will beconducted to summarize the various approaches that are used to establish DCE response quality. Because different methods will require different types of choice tasks, new databased on asingle combinedDCE design (that incorporates all of these requirements)will be collected. In addition, 20% simulated random respondents will be added to the data, which will serve as a benchmark sample of respondents that should at the minimum be identifiedas having inadequate response quality. Therelative performance of the identifiedapproaches will be evaluated using a congruence analyses of their respective predictions. In addition, simulated evidence on the statistical probability that each method correctly identifies respondents as having consistently violated the DCE assumptions will be calculated. Based on both sets of results, a recommendation will be formulated and presented to the EuroQol Group with respect to the methods that can be used to successfully evaluateDCE response quality.Marcel JonkerValuation42840Ongoing20202020
20190151extension of 20190150: Exploring non-iterative time trade-off methods for valuation of EQ-5D-5L health states: an on-line experimentThis proposal is an extension of analready approved study proposal EQ-Project 20190150R1: Exploring non-iterative time trade-off methods for valuation of EQ-5D-5L health states. In that proposal, we aimed to test two non-iterative time trade-off methods (non-stopping and open-ended time trade-off)using face-to-face interview. Encouraged by the discussions of the online cTTO study in the Brussels Plenary Meeting, we felt the non-iterative TTO methods may have the potential advantages for online administration.Therefore,in this proposal, we aim to develop and test the online version ofthese two non-iterative TTO methods, as anadditional component of the approved study.Zhihao YangValuation23350Ongoing20202023
20190900An investigation of the ‘shrinking factor’ model for predicting vision and cognition bolt-on values elicited from the general publicThis projectbuilds on the recent study of Yang et al, entitled“Modelling TTO values of vision bolt-on and self-care bolt-off health states: can bolt-on and bolt-off value sets be built upon EQ-5D value set?”. In that study, we demonstrated that modified main-effects models which contain a‘shrinking factor’ can provide fairly good predictions of vision bolt-on values. That study desgin, however, had several limitations, including using university student samples, involving only the vision bolt-on, and possibly being biased by order/learning effects in health-state valautiontask etc. In this project, we aim to test the ‘shrinking factor’ model with vision and cognition bolt-on values elicited from two general popualtion samples. A total of 600 members of the general public will be recruited and interviewed face-to-face using EQ-PVT. Participants will be randomised to value one of the three types of health states (EQ-5D, vision bolt-on EQ-5D, and cognition bolt-on EQ-5D). A 5-step procedure will be used to evalute the ‘shrinking factor’ model for its performance in predicting the vision and cognition bolt-on values. Both the additive and multiplicative (a constrained main-effects model) models will be assessed for use in the ‘shrinking factor’ model. The comparator will be the standard models used in valuation studies. The ‘shrink facor’ model is apromising method that could allow development of bolt-on value sets from existing EQ-5D value setswhile maintaining the relative importanceofthe core dimensions unaltered. This project will shed more light on the potential of this method by using cTTO data to be generated from a study design very similar to that of the EQ-5D-5L valuation studies.Nan LuoDescriptive Systems, Valuation56400Ongoing20202023
20180301Extension to 20180300: What aspects of quality of life are important to people with experience of cognitive or visual impairment? A qualitative investigationThere is a series of ongoing work on the development of descriptors for bolt-ons to the EQ-5D, driven by the need to capture domain-specific Qol not measured by the EQ-5D. One such project is currently underway exploring the development of bolt-ons for cognition and vision. As part of this project substantive qualitative data collection with individuals and carers experiencing these health conditions has taken place. The pre-planned analysis has revealed that a broad range of quality of life domains beyond those with a vision and cognition focus, and beyond the domains currently covered by the EQ-5D descriptive system, are important to these people. As such, we seek additional funding to explore these domains using a detailed qualitative thematic analysis, which is important to ensure content and construct validity of further quantitative work. The outcomes of this work will lead to a robust and comprehensive list of quality of life domains of importance for people with cognition and vision impairments. Moreover, it will provide a comparison of this list with existing measure of quality of life measures that cover the domains of cognition and vision (e.g. Adults Social Care Outcomes Toolkit (ASCOT)). Therefore, the project will deliver evidence about the suitability of the EQ-5D in these health conditions and inform its future development. This will yield important insights for future bolt-on development work as well as research seeking to develop broader quality of life descriptive systems, such as the E-QALY.Katie PageDescriptive Systems14800Ongoing20202022
20191000Testing the appropriateness of EQ-5D in a socioeconomically disadvantaged populationHealth inequality has been prominent in the policy agenda. Apart from using objective health indicators such as mortality and morbidity to measure health inequalities, with an increasing awareness with people’s feeling and satisfaction about their own health status,health-related quality of life (HRQoL)has become an essential part in health inequality measurement nowadays. Only with a HRQoL measure that is valid and reliable across populations, can the extent and nature of health inequality in terms of HRQoL be well captured. However, HRQoL measures, such as EQ-5D,are less likely to be validated with people who are socioeconomically disadvantaged, for example, those people living in underdeveloped rural areas or with lower education attainment.While, in fact, it has already been reported that thosesocioeconomically disadvantagedpeople in China had difficulty in understanding the questionnaire.With limited research investigated the understandings of HRQoL andtheacceptability of HRQoLmeasures among rural populations in underdeveloped areas, this study proposes to explore how they may understand health as described by EQ-5D. If potential barriers for them in using EQ-5D to reflect their actual health status are identified, it will indicate how we can promote an appropriate use of EQ-5D in this population.Zhuxin MaoDescriptive Systems12550Ongoing20202020
20191110Meeting Asia Policy Makers at HTAi 2020Efficient use of scarce health care resource is of paramount importance for Asia due to its large population and projected increases in chronic diseases. There is a growing interest in developing health technology assessment system to support evidence-based reimbursement and coverage policy making in this region. Over the past few years, the EuroQol Groupincreases its support of EQ-5D related research in Asia which certainly promote the EQ-5D instruments in the region. However, our engagement in Asia has been largely limited to academia. We propose a half-day meeting prior to the 2020 HTAi Annual Meeting inBeijing to engage Asian HTA agencies and health policy makers and communicate on the recent development in patient-reported outcome measures and the potential role the EQ-5D instruments can play.Feng XieEducation and Outreach19800Ongoing20202023
20191180Improved anchoring of stand-alone DCE duration value sets by incorporating immediate death and maximum endurable time; a mixed methods approach based on the EQ-5D-5L and E-QALY instrumentsAims:The primary aim of this research is to improve the anchoring of stand-alone DCE duration value sets for the EQ-5D-5L and E-QALY instruments on the (0-1) QALY scale, which is achievedusing an improved DCE format and the incorporation of Maximum Endurable Time (MET) in the analyses of health state preferences, resulting in more reliable and more reliably anchored tariffs and in a more accurate representation of respondents’ preferences for living in relativelypoor health states.Methods:A new DCE with duration format is proposed that combines a) matched pairwise choice tasks that includeperfect health with b) matched pairwise choice tasks that includeimmediate death. This newformat has the advantageof including both perfect health and immediate death in the DCE (i.e. the traditional QALY anchor points) while constraining the complexity of the choice tasksand ensuring that respondents use the correct multiplicative utility function for health state valuations. The performance of the DCE with duration format will be established using a combination of qualitative and quantitative research.The new formatis anticipated to significantly reduce the complexity of the choice tasks compared to standardDCE duration choice tasks whileimproving the identification of respondents’ time preferences.The latteris crucial for being able to correctly extrapolatehealth-state preferences towardsa duration of zero (i.e. extrapolated immediate death), which haspreviously been established as key to reducing differences between stand-alone DCE and time trade-off (TTO) based tariffs. The improved identification of respondents’ time preferenceswill also allow for an investigation of whether the concept of Maximum Endurable Time (MET) is required to correctly establish the QALY anchor points using a DCE duration format, and whether METcan explain the commonly observed disparity between extrapolated immediate death and immediate death when modeled as an alternative specific parameter in the DCE.Finally, the performance of the format will be evaluated with respect to obtaining stand-alone E-QALY tariffs.Marcel JonkerValuation87931Ongoing20202021
20191020Estimating an EQ-5D-Y value set for ChinaIntroduction: In 2020, the EuroQol Group published an international protocol to estimate EQ-5D-Y value set. In the protocol, DCE data is used as the primary preference data to model the relative importance of five health dimensions and cTTO data is used to anchor the DCE modelling results onto the QALY scale. This study aims to estimate an EQ-5D-Y value set for China following this protocol. To better understand the role of cTTO data in estimating EQ-5D-Y value set, we adopted a larger cTTO design. We also explored the possibility of estiamting a cTTO alone value set and assessed the feasibilty of completing EQ-5D-Y cTTO task in Chinese general public. Methods: Overall, 150 choice sets and 28 EQ-5D-Y health states were valued using DCE and cTTO methods with two independent samples, respectively. General public from 14 different regions were recruited using quota sampling method to achieve representativeness. We compared two modelling strategy: 1) fit the DCE data with mixed logit model with corrleated coefficients and a subsequent mapping procedure for anchoring; 2) fit the DCE and TTO data jointly in a hybrid model. Two criteria 1) coefficient significance and monotonicity; 2) prediction accuracy of the observed cTTO values were used to assess the models. For estiamting a cTTO only value set, we estimated the two models using 3 different estimators: 1) ordinary least square (OLS) as the base model; 2) a heteroscedastic model; 3) a random-effects intercept model. For assessing the feasibilty of using cTTO method valuing EQ-5D-Y states, we examined the distribution of the cTTO data using histogram and by plotting the mean values of individual health states against the level sum score (LSS), an indicator of overall severity of the health states (e.g., the LSS of 33333 is 15). We assessed the feasibility of the cTTO tasks in terms of survey completion time and participant-reported task difficulty. Results: In total, 1,476 individuals participated in the study, with 1,058 participated the DCE survey and 418 participated the cTTO survey. The highest mean TTO value is 0.924 for state 11112 and the lowest mean TTO value is -0.088 for state 33333. The hybrid model with an A3 term performed the best and was selected to estimate the value set. Data density is highest between the range of 0.5 to 1, with 21.89% being negative values (i.e., <0). Less than 2% of the values were at -1. The highest and lowest mean cTTO (SE) value is 0.924 (SD: 0.011) for state 11112 and -0.088 (SD: 0.025) for state 33333, respectively. State 33333 is the only state with a negative mean value. 74.16%, 59.33% and 11.48% participants agreed that the cTTO tasks easy to understand, easy to differentiate the health states and easy to decide their answer, respectively. When modelling the cTTO data without DCE data, all models achieved monotonicity, but 1 coefficient (level 2 usual activities, p-value=0.075) in the OLS model was not significant at the 0.05 level. The prediction accuracy of the models was similar, with the random effects model exhibiting the lowest cross validation MAE. Discussion: Following the international protocol and using a larger cTTO design, this study established the EQ-5D-Y value set using a hybrid model for China. Future EQ-5D-Y valuation study could consider using a larger cTTO design for estimating the value set. Moreover, it is possible to establish an EQ-5D-Y value set using cTTO data alone. Given it is still not clear how the preferences measured by cTTO and DCE methods differ and any method combining both preference data seems arbitrary. The use of DCE method to obtain relative importance and use cTTO value to anchor which on a QALY scale may not be necessary.Zhihao YangValuation, Youth45540Completed20202021
20180750Developing a value set for the child-friendly EQ-5D health-related quality of life instrument EQ-5D-Y in Hong KongThe EQ-5D-Y is currently useto assess health-related quality of life of health states experienced by young populations.However, no country-specific EQ-5D-Y value set is currently availableand recent research has shown that just applying adult value sets to EQ-5D-Y states is not appropriate.Developingvaluation studies for younger population presents challenges to study design and normative problems not present or important in adult valuations.In The proposed study aims to develop avalue set for theEQ-5D-Y in HKfollowing the recommended valuation proposal for EQ-5D-Y studies by the EuroQol Group. The study will include a discrete choice experiment (DCE) exercise to estimate the relative importance associated to each EQ-5D-Y level in each dimension, and a composite time-trade off (C-TTO) to anchor the latent scale DCE value set into the quality-adjusted life years (QALY) 0-1 scale.Both the DCE and the C-TTO will be administered face-to-face in computer-assisted personal interviews by trained interviewersusing the EQ-PVTplatform.DCE and c-TTO responses will be collected using adults from the general population and two different perspectives: adults responding the elicitation tasks from a child perspective and adult’s own perspective. Preferences for both perspectives and elicitation mode will be compared to identify differences and potential impact in real practice. The final value set will be estimated using responses from the child perspective.Eliza WongValuation, Youth0Ongoing20192021
20190580Severity and EQ-5D (SEVQ). How can EQ5D-utilitites capture notions of severity for priority setting in health care.The two-page proposal on which this is based is attached. The EuroQol Exec requested a full-length proposal with more details, and for the requested budget to be capped at 150k €The urgent need for fair, equitable, and publicly acceptable priority setting criteria is attested to by an array of government white papers and scholarly works. QALY-based cost-effectiveness (utilitarian) alone is often considered insufficient. In order to accommodate other ethical concerns, various forms of severity criteria, to be applied on top of or alongside QALY models, have been suggested. However, there is no unambiguous answer to what severity is, and the topic has recently been debated byeconomists and philosophers in several publications.SEVQ aimstoinvestigate linksbetween the descriptive system of EQ-5D and ‘severity’, and to investigate to what extent respondents prioritise the most severely ill, the worst off, or tends towards utility maximization.Mathias BarraValuation383641Ongoing20202023
20191010Psychometric properties, feasibility and usefulness of the extended EQ-5D-Y-5L in children with prevalent disease conditions in EthiopiaBackground: Type 1 diabetes mellitus and heart diseases are the most common medical problems among children in Ethiopia. Children's health-related quality of life (HRQoL) outcome measure are currently receiving attention. EQ-5D-Y-5L health survey in children has been used in different countries. Objective: To assess the psychometric properties of the Amharic EQ-5D-Y-5L Beta version and, thereby, to test whether the measure is feasible, reliable, and valid to use among healthy and children with prevalent disease conditions. Methods. Data were collected using a pen-and-paper survey that presents the self-complete Amharic EQ-5D-Y-5L Beta version and EQ-VAS. A cluster sampling strategy was used to select 200 study participants from ten sub-cities in Addis Ababa city, Ethiopia. A supervisor selected one class from the elementary school of each sub-city using simple random sampling techniques. Within each class (n=40-60), 20 students between the age of 8 and 12 were randomly selected and invited to complete a self-complete Amharic EQ-5D-Y-5L Beta version and EQ-VAS. For children with the prevalent disease conditions (diabetes mellitus, congestive heart failure, rheumatic heart disease) participants were identified from Tikur Anbessa Specialized Hospital (TASH) of diabetic and cardiac clinics. A sample of 55 study participants (23 congestive heart failure, 17 rheumatic heart disease, and 15 diabetes mellitus) who had unchanged health state based on the general health item received the questionnaire again 10 days after the first task for the test-retest procedures to investigate reliability of the instrument. Two-way mixed-effects models absolute intraclass correlation coefficient (ICC) was used to test reliability of the instrument while Kruskal-Wallis rank test with pairwise comparison was used to assess the known group validity of the instrument. Results. The mean (SD) age of the participants in this study was 10.35 (1.42) years of which 51.00% were female. A sample of 450 participants (200 school children, 65 diabetes mellitus, 70 Rheumatic heart disease, and 115 congestive heart failure) were recruited and 437 participants (200 school children, 62 diabetes mellitus, 66 Rheumatic heart disease, and 109 congestive heart failure) were included in the sample for analysis. All of the EQ-5D-Y-5L dimensions and the EQ-VAS had strong intraclass correlation coefficients (ICC), which fall within the range of good agreement for all respondents (ICC ranges from 0.52-0.77, p 0.001). The differences in self-complete Amharic EQ-5D-Y-5L Beta version and EQ-VAS scores between disease groups demonstrate the self-complete Amharic EQ-5D-Y-5L Beta version's significant known group validity. Conclusion: The Amharic EQ-5D-Y-5L Paper BETA Self-Complete appears to be a suitable measure for assessing HRQoL in various child population groups in Ethiopia, according to this study.Abraham GebregziabiherYouth18200Ongoing20192020
20191030Investigating response heterogeneity in the EQ-5DSystematicdifferences in the ways that people use and interpretresponse categories can introduce variationwhen using self-reports (like the EQ-5D) tocompare health or quality-of-life across heterogeneous patient orpopulation groups.Inter-group comparisons using these self-reported measureswhich experience such variation may bring into questionany perceived findings.Self-reportedhealthmeasures can also be affected by response style;a respondent’s tendency to systematically respond to questionnaire items ina given way regardless of item content. Given the growing demand for the routine collection of self-reported health measures like the EQ-5D in registries, cohort studiesand national health surveys there is obviousvalue in understanding whether and to what extent the EQ-5D suffers from response heterogeneity. This study proposes to undertake a series of analyses using a range of different datasets to understand Version 10SEPT2018Page 2if response heterogeneity introducesvariation intothe EQ-5D, and if so the likely extent of this. We will explore availablesecondary datasets that include the EQ-5D(e.g.UK PROMs, GP Patient Survey, Multi-Instrument Comparison (MIC)study,and Cancer 2015 cohort study) to identify which dataset is most appropriate for answering our study question. These datasetsoffer an opportunity to test for the existence of two types of response heterogeneity: reporting heterogeneity (using Rasch analysis and/orlogistic regression) and different response styles (using variations on the orderedprobit and latent class approaches). We will also explore if response heterogeneity changesover time usinglongitudinal datasetsPaula LorgellyPopulations and Health Systems52400Ongoing20202021
20190780Valuation of the EQ-5D-5L in the Kingdom of Saudi Arabia (KSA)Objectives:Saudi MOH is one of the first entities across the Middle East region which started the adoption of economic assessmentsin reimbursement decision making. The cost-utility analysis will be soon adopted as a tool for economic evaluation and for assessing the cost-effectiveness of new and existing health technologies. Since no country specific Utility values exist in KSA, theprimary objective of this study (the Saudi Arabia EQ-5D-5L Valuation Study) is to develop a value set for this instrument in KSA which can be used to support the health utility estimation in future economic evaluations for reimbursement decision making in KSA.Methodology:Asingle country, population based cross-sectional survey study based on the EuroQol Valuation Technology (EQ-VT) study protocol generated by the EuroQol group and administered using computer-assisted personal interviews. A minimum of 1,000 respondents will be targetedwhere a Quota sampling will be adopted to obtain a representative sample of the Saudi population taking into consideration the following factorsRespondents’ location; 13 mainregions of KSA Different age groupsGenderEducationEmploymentEligible population should be (1) Saudis 18+ years old; (2) able to understand the tasks, contents and techniques of the interview process (as judged by the interviewer) ;(3) able to give informed consent, (4) do not have current acute illness or cognitive impairments that would interfere with the interview conduct.Direct advertisements & flyers in addition to digital recruitment will beused for recruiting the study population. Third-party vendor will be contracted by the Saudi MOH for the following activities:Study documents developmentIRB submissionsRespondents’ recruitment, field Interviews, data collection &quality checksData analysis & reportingAhmed AljedaiValuation0Ongoing20192021
20180710Valuation of EQ-5D-5L in Uganda and exploration of a ‘lite’ protocolThe primary objective of this project is to develop the EQ-5D-5L value set for Uganda. A total of 500 participants will be recruitedfrom fourregions across Uganda (Central, Western, Eastern, and Northern) to form a representative sample of the general Ugandan population in terms of age, sex, ethnicity, andeducation attainment. Each will be asked to complete 20 composite time trade-off (cTTO) tasksby trained interviewers. Thisdesign will enable the inclusion of more EQ-5D-5L health states(compared to the 86 health states used in the standard valuation study), allowing for greater precision in the parameterestimation.The EuroQol Valuation Technology (EQ-VT) softwarewill beused in data collection. TheTTO based model will be used to develop the value setfor Uganda. A secondary objective is to evaluate the performance of a ‘lite’ valuation protocol. Half of the data will be used to estimate the value set and then the results will be compared to those using all the data. This would provide guidance on the design of a ‘lite’ protocoland benefit future valuation studies, especially in low and middle-income countries.Anadditional sample of 160 participants will be recruited from the four regions to conduct a health inequality survey. This includes a series of trade-off questions presented usingweb-based software. The responses will be used to estimate aninequality aversion parameter for Ugandawhich can be used as part ofdistributional cost-effectivenessanalysesof interventions and policies.Mark SculpherValuation47579Completed20202021
20191050Using EQ-5D to inform real-world decision making : a cross-sectoral perspectiveMultiple agencies responsible for planning and delivery of services in separate (and ostensibly unrelated) public sectors are being encouraged to share/exchange data. LARIA’s annual conference this year includes several speakers from PHE with workshops specifically targeting the central issue of improving information flows and data exchange. The perspective of researchers within separate sectors is largely limited by the scope defined by the services for which they are responsible. The historic emphasis on using EQ-5D for “health” applications has naturally reinforced the view within the health sector this is the specific role for which it is best suited. Alternative uses of EQ-5D – outside the health sector – are to be found and this suggests a possible role for such data in providing a linking mechanism between data sets on a cross-sectoral basis. This proposal is designed to test that challenge by bringing together relevant stakeholders, research academics and policy/planning advisors from Local Authorities, health and criminal justice sectors. LARIA has expressed a general willingness to support the proposal across its regional and national networks should funding support be made available.Paul KindEducation and Outreach11390Ongoing20192020
20190600Does EQ-5D cover the most undesirable health problems in different cultures? A study of seven countries using a mixed methodsWhen using patient reported outcomes (PROs) measures such as EQ-5D internationally, it is important to assess the extent to which their content is valid and relevant across different countries and cultures. If it is shown that content is appropriate for use in different regions, it can strengthen the case for using the instrument in a wide range of cultural settings. A recent study in Asia revealed that some health concepts which are important to Asianpopulations are not included in EQ-5D. Nevertheless, that study was limited to 4 countries and the methodology used was only intended to explore perceptions of health and health problems inNan LuoDescriptive Systems172632Ongoing20202023
20190920Understanding adult preferences in the valuation of child and adolescent health states measured with the EQ-5D-Y: A qualitative approach.BackgroundThe EQ-5D-Y was developed to measurehealth-related quality of life(HRQOL)in children and adolescents aged 8 to 15. The valuation of the measure has confronted the EuroQol Group with several challenges. For example, evidence shows that adults value HRQOL for a 10-year old child higher than for themselves, resulting in a relatively narrower value range for the EQ-5D-Y. This may be due to a difference in the value of life years, maximum endurable time, and extrinsic goals between children and adults; however, the reasons behind this difference are currently unclear. Evidence further shows that adults value health differently for children of different ages. Therefore, the current reference of a 10-year old child may insufficiently reflect preferences regarding, for example, a 15-year old adolescent. To date, no qualitative studies have been published that provide insight into the reasoning of adults while valuing EQ-5D-Y health states. Aim: To examine (i) the difference in value adults attribute to their own HRQOL and that of a (10-year old) child and (15-year old) adolescent, (ii) how considerations—that are not directly related to the described health states—influence adults’ valuation, and (iii) whether (and why) adults attribute different value to similar health states described for a child and an adolescent and the influence of time/duration in this context.Methods:Adult members of the general population (n=25) will perform composite c-TTO and DCEtasks associated with EQ-5D-Y health states using a think-aloud protocol and, subsequently, participate in a semi-structured interview with retrospective verbal probes.Vivian Reckers-DroogYouth38000Completed20202021
20190440Proposal of the 2nd EuroQol Asia Academy MeetingEQ-5D is the most widely used quality of life measure in health economics and outcomes research around the world including the Asian region. The 1stEuroQol Asia Academy Meeting was successfully held in Guangzhou, China on 11-12 June 2019, bringing 38 researchers from Asian countries and 5 senior EuroQol members together for sharing, learning, discussing, and collaborating their research and expertise. There were 40 abstracts submitted of which 26 were discussed in the meeting: 14 as paper and 12 as poster presentation. A range of topics were covered, including (i) the similarities and differences of health-related quality of life concepts between Asia and the western counties, (ii) additional EQ-5D dimensions that specific to Asian population (e.g., sleep and eat), (iii) problems (and solutions offered) of completing the EQ-5D questionnaires and valuation study in rural and illiterate respondents, (iv) the application of EQ-5D youth version in Asian countries that also possible to be used by adult respondents who have difficulties to complete the adult version (v) valuation studies, (vi) application of the questionnaires in different populations. Several interesting findings were resulted from studies employed qualitative method, rarely discussed in the general plenary meeting. Ideas of collaboration were discussed, among them are: investigate the (bolt-ons) dimensions thatspecific to Asian population in more countries, (ii) reviewpsychometric evidences of EQ-5D in local (non-English) journals, (iii) investigate the best way to obtain EQ-5D data from difficult-to-reach groups of participants, such as rural and illiterates. Allattendees felt that this meeting was beneficial for their on-going studies or future study plans.More specifically, the senior EuroQol members who were present all found that the level of the papers presentedwas good and the discussion useful. The EuroQol associates who attended the meeting (beyond the organizing committee; Michael Herdman, Paul Kind and Aureliano Finch) all support a next meeting, as they considered such meeting a timelyinvestment, given the quality of the papers and the dynamics of the area. This second meeting will thus continue the momentum created by the first meeting: to foster sharing of research projects and results, expertise and experience. Collaboration projects being held could also be reported. The format of meeting will be similar to the usual general meeting: started with presentations about the EuroQol Group and the EQ-5D questionnaires, followed by paper discussion and guided poster presentations. Discussions about future plan and collaborations will be held at the end of the meeting. We can expect the prominent experts and researchers that attended the first meeting and the ones who were not able last time will participate, especially when the announcement is made at the earliest time.Fredrick PurbaEducation and Outreach67500Completed20222022
20190370Compare the TTO and DCE modelling results on individual levelBackground So far, little attention has been given to the question what level of agreement is required to support the hybrid model, and what level of agreement can realistically be expected from DCE and TTO data. In this study we investigate the relationship of DCE coefficients and TTO coefficients on individual level. Methods A total of 210 university students were recruited from Guizhou Medical University, China to complete both TTO and DCE tasks. All participants valued 31 states using TTO and completed 55 choice tasks in the DCE. The set of states in the TTO tasks included a 25-state orthogonal design + 5 mildest states + 55555. A Bayesian efficient design algorithm was used to construct Bayesian D-efficient design with 60 pairs (55 unique pairs + 1 dominated pair + 4 repeated pairs) for the other respondents. The priors of the DCE design were from a sample of 36 students who completed a design of 40 pairs of DCE design constructed using priors from previous study. Students completed the TTO and DCE tasks in two separate face-to-face session, with the assistance of a trained interviewer. The interval between two sessions was <7 days. All responses were modelled by student using OLS (for TTO) and conditional logit (for DCE) using a 5-parameter (5P) model. 20-parameter (20P), 10-parameter (10P) main effects model were used to compare two sets of coefficients on aggregate level. Results 210 university students completed both sessions. For the TTO session. On average, each student spent 59.37 (SD: 18.93) minutes and used 8.97 (SD: 2.05) moves in the TTO tasks. 174 students completed the design of 60 DCE tasks (including 1 dominant pair test and 4 repeated pair tests) and 36 students completed the design of 40 DCE tasks. Among the 174 students of completing 60 DCE tasks, 166 students passed the dominant pair tests. 49, 67, 39 and 18 students passed 4, 3, 2, 1 repeated-pair tests, respectively. For the coefficients, the average Pearson correlation was 0.435 (SD: 0.462, median: 0.567, range: -0.859 to 0.999). Respondents whose DCE data was of better quality according to R2 showed better consistency in their TTO and DCE modelling results. On aggregate level, two sets of coefficients had a Pearson correlation coefficient of 0.885 for the 5P model, 0.906 for the 10P model and 0.871 for the 20P model. Discussion Our results showed that on individual level, there are some evidence that the two sets of coefficients are associated, but not for all respondents. A main limitation of this study is the limited power of modelling coefficients on individual level. On aggregate level , it is clear that the linear relationship between two sets of coefficients exist, but varied with the severity levels of the coefficients.Zhihao YangValuation33720Completed20192020
20190200Validation of the Chichewa versions of the EQ-5D-Y-3L and the EQ-5D-Y-5L in MalawiObjectives The EuroQol Group has developed an extended version of the EQ-5D-Y with five response levels for each of its five dimensions (EQ-5D-Y-5L). The psychometric performance of the three response level version (EQ-5D-Y-3L) has been reported in several studies involving both healthy children as well as clinical populations, but that of the EQ-5D-Y-5L has only been reported in a few studies. This study aimed to translate and psychometrically evaluate the Chichewa (Malawi) versions of both the EQ-5D-Y-3L and EQ-5D-Y-5L. Methods The EQ-5D-Y-3L, EQ-5D-Y-5L and PedsQL™ 4.0 self-report Chichewa versions were administered to children and adolescents aged 8-17 years in Blantyre, Malawi. Both the EQ-5D-Y-3L and the EQ-5D-Y-5L were evaluated using gold standard psychometric methods, including missing data, floor/ceiling effects, reliability (internal consistency) and validity (convergent, discriminant, known-group and empirical). Results A total of 289 participants (95 healthy, and 194 chronic and acute) self-completed the questionnaires. There was little problem with missing data (0.7). Convergent validity tested with PedsQL™ 4.0 self-report was found to be satisfactory (correlation >0.4). However, correlation between some of the dimensions for both versions of the EQ-5D-Y and the PedsQL™ 4.0 sub-scales was mixed. While there was evidence of discriminant validity with respect to gender and age (no association with EQ-5D-Y-3L and EQ-5D-Y-5L sum and utility scores), it was not evident for school grade (p<0.05). As regards empirical validity, the EQ-5D-Y-5L was 31%-91% less efficient than the EQ-5D-Y-3L at detecting differences in health status using external measures. Conclusions The EQ-5D-Y-3L and EQ-5D-Y-5L were both found to be reliable and valid for use among children and adolescents although with some limitations. Further, psychometric testing is required for test re-test reliability and responsiveness that could not be carried out in this study due to COVID-19 restrictions.Janine VerstraeteDescriptive Systems14200Completed20192020
20170491Valuing Health-Related Quality of Life: An EQ-5D-5L Value Set for Belgium: Request for budget extentionThe aim of this study is to generate a value set for Belgium by means of 1000 computer-assisted face-to-face interviews in a representative sample of the general Belgian population. To both ensure the representativeness of the sample and avoid a too wide and scattered spread of individuals to be interviewed, a multistage, stratified, cluster samplingwith unequal probability design will be applied, with provinces and municipalities used as clusters and age and gender used as strata.Health state values will be elicited using both the composite time trade-off (cTTO) and the discrete choice experiment (DCE) tasks. Different models will be tested. For the DCE task alone, a conditional logit regression model will be used. For the cTTO task alone, a modelling strategy similar to the one used in the reassessment of the English value set (Feng et al, 2016){Feng, 2016 #11}will be followed, taking into account censoring-like characteristics of the TTO scale, coefficient order of magnitude consistency, semi-continuous nature of the TTO scale, and the heterogeneity of the respondents ‘personal’ TTO scale. A Tobitmodel similar to the one used in the Dutch value set (Versteegh et al, 2016),{Versteegh, 2016 #10}taking into account censoring-like characteristics of the TTO scale, will also be considered. Finally, the hybrid cTTO and DCE model described in Feng et al, 2016{Feng, 2016 #11}and Ramos-Goni et al, 2017{Ramos-Goni, 2017 #7}will be used. The performance of the models will be assessed on the basis of internal consistency, goodness of fit and parsimony.Gerkens SophieValuation17630Completed20172019
20190630EUROQOL Satellite Symposium - Applying Quality of life Measurements for Clinical and Economic Researchn the Brazilian Public Health System, recommendations for the coverage and reimbursement of new medicines, devices, or equipment are centralized in a committee (National Commission for Technology Incorporation in the Unified Health System). To be approved, the claimant must provide health technology assessments (HTAs). These include measures of safety, efficacy, effectiveness, and cost-effectiveness to substantiate the decision-making process. There is no restriction on the type of economic model that can be submitted for review. The recommendations are presented as legal documents and economic analysis guidelines.2 This document discusses focus on the possibilities of the utility measures, without any reference to sources or methods or the selection of the best evidence for utility parameters. Health insurance plans in the private sector have a similar process through an independent agency (National Regulatory Agency for Private Health Insurance and Plans) that is responsible for defining a mandatory list of procedures and medicines to be included in all plans. Brazilian preference research was sparse in both the public and private sectors. In this context, utility is a concept adopted from economics that refers to preferences for a specific health state or outcome. This preference (or weight) is usually based on a large group of people representing the general population. Utility is a proxy of quality of life and reflects the preferences of individuals or society for any particular set of health outcomes. Utility expresses the health-related quality of life (HRQOL) in a single value scored on a scale anchored on 1 = “full health” and 0 = “death,” usually derived from “off-the-shelf” preference-based measures such as the EQ- 5D questionnaire. Some health states may be considered worse than death and given negative utility estimates.Utilities are used for informing cost–utility models and can sometimes be obtained from different sources; it is important to create a hierarchy when multiple estimates are available. In a cost–utility analysis, competing health technologies are compared in terms of their cost per “year in full health.” The quality-adjusted life-year (QALY) is one such widely used measure that combines a person’s life expectancy and the value of their HRQOL in a single estimate. The HRQOL can be expressed in utilities for economic analysis. It is important to distinguish between the utility weights and the profiles or health states. The measurement process using multiattribute instruments starts by asking for a description of individual health states, called profiles. The selected profiles are converted to utilities compared with a table called the value set, which contains weights for each profile. These weights are usually collected from the general population (valuation process) and represent preferences for each possible health state. In one Brazilian state, in 2011, value set was developed for the Short-Form Six-Dimension (SF-6D) instrument.9 For 3-level version of EQ-5D (EQ-5D-3L), the value set was developed nationally in 2013.10 Nonetheless, in the Brazilian HTA ecosystem, utility estimates based on Brazilian samples are limited. Considerations of the source and type of utility values are especially important in a modeling context, in which the lack of transparency, including the lack of a hierarchy for utility data sources, is a major issue for any estimation and could compromise model reliability. The absence of clear guidelines permits flawed modeling practices, given that an ad hoc evidence selection can result in cherry picking. An extreme example of this practice would be feeding a model with data to uphold the owner’s perspective, creating a false favorable impression of a particular technology. In recent years, the process of decision-making using cost–utility data in Brazil has improved. Because of the lack of confidence in some economic models, the cost–utility models have a mixed impact on real-world judgments. Transparently built models can hopefully support better decisions for the rational use of health resources. Few international guidelines discuss how to select the best utility data from different perspectives in the Brazilian context. This document aims to present the first version of the Brazilian guidelines for a utility measurement supporting an economic analysis. A glossary of technical terms was created. As an initial step, a rapid review of the literature was conducted on July 6, 2020, based on an adapted search strategy from MEDLINE,EMBASE, and LILACS databases and the websites of The University of Sheffield, the Decision Support Unit of the National Institute for Health and Care Excellence (NICE), and the EuroQol Group. Additional individual search strategies were adapted for each recommendation to gather the main recommendations and debates previously published on utility measurement issues for economic analyses. Approximately 110 documents were reviewed. The experts were selected based on their previous experience with utility measurement instruments or economic models. The the external reviewer was selected for his experience with both topics. Another important criterion was to be a manager or member of the 2 incorporated sectors in Brazil: public health (Brazilian Public Health System) and insurance plans (National Regulatory Agency for Private Health Insurance and Plans). The review results were synthesized in a brief report. Then, in October 2020, these findings were discussed during a 5-hour online workshop, in which 5 lectures were presented by international speakers followed by debates. The complete program is available in the attached material. The expert group included a wide range of stakeholders—including government representatives, industry, academic groups, international guests, and a patient representative—and regulatory agencies, including the private health sector An adapted Delphi panel technique with 4 iterations was adopted for the construction of the final report. The draft report was sent to representatives, who used it to write the final report and who are listed here as coauthors. The recommendations are based on the proposal with the most approvals, but controversial topics are included in the recommendation or discussion. The text includes the rationale for the final decisionMARISA SANTOSEducation and Outreach17976Completed20222022
20190690Understanding the routine collection and use of EQ-5D data in large-scale applications within the healthcare system (Apersu supports)There is an increasing use of EQ-5D as a patient-reported outcome measure in large-scale applications within healthcare systems around the world. However, we still lack guidance on the proper use of EQ-5D in these settings, and further, on how best to use EQ-5D data at various levels within the system. The overall aimsof this projectare to examine the use of EQ-5D in routineoutcome measurement within the healthcare system, and explore whether/how EQ-5D data can be used to inform decision-making at the micro (e.g., screening and monitoring of health outcomes, clinical management), meso (e.g., assessing and evaluating outcomeswithin a clinical unit), and macro levels(e.g., evaluatingthe performance of the healthcare system). Three projects in three different clinical areas are proposed: 1) the use of EQ-5D at the micro and macro levels in community outpatient and speciality rehabilitation; 2) the use of EQ-5D at the micro and meso levels in epilepsy patients; and 3) the use of EQ-5D at the meso and macro levels in primary care. Each of the projects will involve the use of mixed methods including a quantitative study (involvingsecondary data analysis) and a qualitative study (involving primary data collectionbased on realist evaluation methodology). Findings from these projects will provide a framework for the use of EQ-5D data in large-scale applications, particularly those involving routine outcome measurement within the healthcare system, and provide guidance on how to enhance the use of this data to support decision-making at various levels within health care systems.Jeffrey JohnsonPopulations and Health Systems241000Ongoing20202023
20190280Elicitation of a Norwegian EQ-5D-5L value setThe valuation study will be undertaken using EQ-VT in a representative sample of the Norwegian general population recruited through multistage random sampling and quota sampling. Six geographical areas of acute hospital catchment will be randomly sampled with proportional allocation. Within the six areas, locations for data collection will be sampled including clubs and societies, healthcare providers, housing associations and workplaces. Possible locations will be stratified into relevant groups to increase homogeneity of respondents across locations within each stratum and randomly selected for data collection. Stratification of locations will be designed to enhance recruitment of groups who have greater demands on their time and hence less likely to participate in such studies, including those in full-time employment with children. At the respondent level, quota sampling will be based on the composition of the general population including age, gender and education level for each area. Data collection will be undertaken in two three month periods starting in August 2019 and March 2020 with the aim of six interviewers completing four interviews each per day. The interviews will follow the EQ-VT software with ten cTTO questions and seven DCE questions. An additional, eleventh cTTO task will be included for respondents with health states below full health "11111" who will be asked to value their own health state. This data will contribute to a further study objective comparing the valuation of experience­ based and hypothetical health states. Statistical modelling of the survey data in R, STATA or equivalent will be used to derive the value set and scoring algorithm. Approaches to modelling will reflect the characteristics of the data. The scoring algorithm will be available to Norwegian and international users through the Institute's website.Andrew GarrettValuation0Ongoing20192022
20190470Bringing Patients Back to the PROM. Development and delivery of a workshop to promote the use of the EQ-5D to inform patientsA workshop was developed to help attendees learn how routine PROM data collection strategies could be improved by engaging the very patients who are completing the PROMs. The workshop started by explaining routine PROMs collection – how it differs from research collection – and the problems in terms of response rates and how the data is used. It then went on to describe shared decision-making (SDM) using the 3 talk model by Elwyn et al. Attendees were asked to consider how SDM worked in their clinical context. We then went through each of the 3 talks, providing examples of how PROMs could be used in each. Videos from patients and clinicians who have used examples were provided. Graphic demos were also given. Attendees where then put into groups to consider what might work in their context – given the question “why should a patient report their health outcome” The workshop finished by giving further resources, and a discussion of how new initiatives can engage patients. 3 workshops were given and a website was created.Nick BansbackPopulations and Health Systems36450Completed20202020
20190430Efficient designs for valuation studies that use DCEs with mapping to TTO health statesDiscrete choice experiments (DCEs) have been of interest to the EuroQol Group for some time, are included in the EQ-VT, and have been used to create value sets for the EQ-5D-5L. Moreover, both the e-QALY and the EQ-5D-Y may use DCEs for the purposes of valuation. Given the resource-intensive nature of the time trade-off (TTO), valuation using DCEs isappealing. However, current approaches to anchoring the latent utilities derived from a DCE to the cardinal scale may be suboptimal. When the DCE is implemented without duration, anchoring is typically achieved by including some TTO tasks. In this project we will consider the optimal number of health states to be valued using TTO, whether these health states should be clustered at the severe end of the utility scale or spread evenly, and whether, given a fixed number of respondents and number of tasks per respondent, it is better to have more health states valued by fewer respondents or fewer health states valued by more respondents. Furthermore, we will consideranalytic strategies to improve predictive precision by (a) extending the hybrid model to incorporate non-linear associations between latent and TTO utilities, and (b) incorporating spatial correlation among health states in the hybrid model. Making more efficient use of the available data can allow researchers to use a smaller sample size while maintaining the same predictive precision. We will develop mathematical models, implement them in R, and test them using simulation.Functions to fit the proposed models will be released through Github.Eleanor PullenayegumValuation39917Ongoing20202023
20190290Malaysian EQ‐5D‐5L Value Set Symposium and WorkshopA half-day symposium followed by a half-day workshop was organized to inform healthcare decision makers and researchers on the availability of a Malaysian EQ-5D-5L value set and provide guidance on its applications, specifically in economic evaluations and outcomes research.Asrul ShafieEducation and Outreach20097Completed20192019
20190350A comparison of DCEs with choice sets of size 2 and DCEs with various choice set sizes for the valuation of the EQ-5DRecent work in the construction of DCEs has demonstrated that allowing the DCE to have choice sets of unequal sizes can be at least as efficient as DCEs with constant choice set sizes. It was conjectured that varying choice set sizes may be preferred by the respondents which might lead to more engaged respondents. We conducted two field experiments to check if this conjecture is true or not. In the first field experiment we compared six designs with choice sets of varying sizes (2, 3 or 4) and with options described by the EQ-5D-5L. In the second field experiment we compared three designs with choice sets of varying sizes (2, 4 or 6) and where options were pasta sauces. In the second field experiment designs with varying choice set sizes were perceived to get easier as the survey progressed but that was not the case in the first field experiment. Whether this was because of the nature of the options (pasta sauce choice is more familiar) or the larger variation in choice set sizes is not clear.Deborah StreetValuation14675Completed20192020
20190130The performance of EQ-5D-5L in various disease groups with different durationsThe recall period of ‘today’ used in EQ-5D is different with other health-related quality of life questionnaires, for instance using‘pasttwo weeks’ or ‘during the past 4 weeks.’ Different types of illness: chronic, acute and periodic attackimpacted HRQOL of its patient in different ways. Yet no previous investigation comparing different recall periods with the one being used in EQ-5D and how they will affect the way patients/respondents respond to EQ-5D.A sample of 675 patientsfrom 9groups: 50 patients with tuberculosis, 75 HIV/AIDS, 100 T2DM, 75 dengue, 75 typhoid, 75 traumatic injury, 75 asthma, 75 Allergic rhinitis, and75 psoriasiswill be invited to participate in this study. For the chronic ones (Tb, HIV/AIDS, T2DM), patients will complete severalrecall period versions(i.e.today, past week, past month) in the first month, followed by monthly data completionfor 5 months. For the acute ones (dengue, typhoid, traumatic injury), two consecutive data collection will be done: the first day of hospitalization and the day of hospital discharge. Recall period to be asked are: ‘today’and ‘since the first day you are hospitalized’. For the periodic attackones (asthma, allergic rhinitis, psoriasis), modified recall period will be ‘last time you experienced the illness’in two times data collection. A subset of each group will be interviewed in-depth to elaborate what do they think and how do they decided when they completed the EQ-5D-5L using the different recall periods. Feasibility, test-retest, responsiveness, and correlation with relevant disease specific will be analyzed between different recall periods and severity level. This investigation will provide insight of how EQ-5D performs in these various diseases with regard to difference of duration and fluctuations of severity.Fredrick PurbaDescriptive Systems35600Ongoing20192021
20190670Cross-cultural Validity and Reliability Testing of the Toddler and Infant (TANDI) Health Related Quality of Life Measure, an experimental version of the EQ-5D-Y ProxyBackground The EuroQol Group is exploring the development of a Health-Related Quality of Life measure for Toddlers and Infants (EQ-TANDI) aged 0-36 months. This study aims to report on the cross-cultural adaptation and validity of the South African Afrikaans EQ-TANDI. Methods The development of the Afrikaans EQ-TANDI followed the EuroQol guidelines including forward-backward translation and cognitive interviews with 10 caregivers of children aged 0-36 months. Thereafter 162 caregivers of children 0-36 months of age were recruited from a paediatric hospital in-patient (inpt) and outpatient (outpt) facility. The EQ-TANDI, Ages and Stages Questionnaire, FLACC and dietary information were completed by all caregivers. The distribution of dimension scores, Spearman’s correlation, analysis of variance and regression analysis were used to explore the validity of the EQ-TANDI. Results The descriptive system of the EQ-TANDI was generally well understood and accepted by caregivers. The correlation coefficients for concurrent validity were significant and moderate for pain and weak and significant for the other dimensions hypothesised to correlate. Known groups were compared and inpts had a significantly higher report of pain (X^2=7.47, p=0.024), more problems reported across all EQ-TANDI dimensions (recorded on the level sum score) (KW-H=3.809, p=0.05) and reported significantly worse health on the Visual Analogue Sale (VAS) (KW-H=15.387, p<0.001). There were no age-related differences except for a lower report of problems with movement in the 0-12 month group (X^2=10.57, p=0.032). Conclusion: The Afrikaans version of the EQ-TANDI is well understood and accepted by caregivers and valid for use with children 0-36 months in South Africa.Janine VerstraeteDescriptive Systems, Youth8851Completed20192020
20180340R1Trickling down to explain the valuation of worse than dead states: towards more valid values.Background: Recent studies concluded that for health states considered worse than dead (WTD), as measured with the Time Trade-Off (cTTO) method, negative mean values were insensitive to health state severity, which represents a validity problem for the cTTO. However, the aforementioned studies analysed negative values in isolation, which causes selection bias as the value distribution is truncated. Aim: To investigate the validity of aforementioned studies and of negative values in general. Methods: The ‘threshold explanation’ was formulated: beyond a certain severity threshold, preferences change from better than dead (BTD) to WTD. This threshold differs between respondents. Thus, negative values across severity are obtained from different respondents, and responses added for higher severity contribute negative values close to zero , explaining the aforementioned insensitivity. This explanation was tested using data from the Dutch EQ-5D-5L valuation study. Respondents valued 10 health states. Based on respondents’ number of WTD preferences, segments were constructed, containing respondents with similar severity thresholds. Using regression models for each individual respondent, we examined the relation between values and severity, and compared respondents between segments. Results: Negative values, when analysed in isolation, were insensitive to severity. However, for individual respondents and within most segments, cTTO values and severity were negatively related. For individual respondents, negative slopes were steeper for segments with more WTD preferences, as predicted by the threshold explanation. Discussion: Analysing negative values in isolation leads to biased estimates. Analyses of cTTO values for individual respondents refute the insensitivity of negative cTTO values.Bram RoudijkValuation14600Completed20192019
20190310Extending the QALY Valuation Study in EnglandObjective: The first aim was to test the feasibility of using time trade-off (cTTO) and discrete choice experiment (DCE) administered using the EuroQol Valuation Technology research protocol to derive utilities for the EQ-HWB-S. The second aim was to review the acceptability of the new measure for decision-making with informed members fo the public. Methods: EQ-HWB-S utilities were elicited using cTTO and DCE tasks with adaptations to fit the new measure. Participants (target n=600) from the UK general population were sampled based on age, sex and ethnicity. Interviews were undertaken using video-conferencing. Quality control (QC) steps were used to assess interviewers’ performance throughout the study. Data were modelled using linear, Tobit, probit and hybrid models. Feasibility was assessed based on the evaluation of the distribution of cTTO data, QC assessment and regression modelling results. Regression results were assessed based on theoretical considerations, monotonicity and statistical significance. For the second aim, members of the NICE Public Involvement Programme Expert Panel were invited to participate, with volunteers selected to represent varying age, gender, health and caring responsibilities. To familiarise the group with the measure and the source of the weights, each person completed a valuation interview (time trade-off (TTO) and discrete choice experiment (DCE)). This was followed by a cognitive debrief and information giving group session, where the weights from the feasibility valuation study were presented. Two subsequent separate focus groups obtained views regarding the measure, the utility weights, the sample (including exclusions) and the methods used. All sessions took place online. Focus groups were recorded, transcribed, and analysed using a framework approach. Results: There were 521 participants who provided cTTO and DCE data. The demographic characteristics were broadly representative of the UK general population although participants were more educated and there were slightly more females. Interviewers met QC requirements. cTTO values ranged between -1 to 1 with increasing disutility associated with more severe states. The hybrid Tobit heteroscedastic model had values ranging from -0.384 to 1. Pain, mobility, daily activities, sad/depressed had the largest disutilities followed by loneliness, anxiety, exhaustion, control and cognition in this model. Twelve people (50% female, aged 28-74) completed the interviews and nine attended the focus groups to assess feasibility. EQ-HWB-S was viewed positively due to the inclusion of dimensions such as exhaustion and loneliness. Some missing dimensions were identified (e.g. coping, sleep), but existing dimensions were considered to cover some of these (e.g. sleep covered by exhaustion). There was surprise at the small utility decrements for anxiety, control and exhaustion relative to other dimensions. Weights were seen as reflecting societal norms, respondent experience or knowledge, the composition of the sample and the interpretation of items. There were concerns that the valuation survey sample was not diverse or large enough to adequately represent the values of those who would be impacted by decisions based on EQ-HWB-S. Participants only supported data exclusions where it could be evidenced by multiple sources that the respondent did not understand or fully engage in the exercise. Other exclusions were considered problematic either because the data could reflect true preferences or for ethical reasons. DCE was preferred to TTO, but participants suggested TTO could be improved by providing more background information, different practice states, particularly the wheelchair state, and offering post-survey debriefs.Clara MukuriaValuation201450Completed20202020
20190020QALY MICI (IBD QALY)Objectives: QALY-MICI is a propective transversal survey, the primary objective of which is to collect health states using EQ-SD-SL in the French context for patients who suffer from Inflammatory Bowel Disesase {IBD: Crohn's disease and Ulcerative Colitis) and to provide utility values based on the French Value set. EQ-SD data will be associated to data on history of disease, past treatments, current treatments, activity scores. Socio-demographic data include type of activity and occupation, as well as size of residential area. EQ-SD data will be compared to a specific Qol questionnaire for IBD, the Short Inflammatory Bowel Disease Questionnaire (SIBDQ), for external validation. Methods: Three sources will be used. An online questionnaire will be sent to members of the MICI Observatory managed by AFA, the French IBD Patient Association. A mail questionnaire shall be sent to patients of the EPIMAD registry, covering four departments of the north and north-west of France. Direct interviews shall be conducted by the department of gastro­ enterology of Hopital Saint-Antoine, Paris, during follow-up visits and day care. Expected number of respondents is 3,300. Expected results: Health States and Utility values for IBD patients; multivariate analysis of utility values according to disease phenotype and treatments; comparison of EQ-SD and SIBDQ. The final data base will be made available for academic research under request after approval by the Scientific Committee of the research. This research has received an approval from official ethics committees and is registered on the site of ANSM, the French Medicines Agency.Gerard De PouvourvilleDescriptive Systems15000Ongoing20192019
20190260Budget transfer from international Psychometric surveys to TUOSData collection budget to test the psychometric performance of candidate items for the EQALY measure. Linked to projects: 20180600, 20180580; 20180520; 20180460 and 2018 0450Tessa PeasgoodDescriptive Systems, EQ-HWB76216Completed2018
20180450Extending the QALY – Psychometric testing of the items in Germany to support the item selection for the new measureThe aim of thisproject is to support the item selection for a new instrument development (Extending the QALY (E-QALY) project) that aims tocapture the changes in quality of life from interventions in health and social care that would be fit for the purpose of inclusion in economic evaluations. The core domains for the instrument have been identified and a pool of potential items has been proposed. Face validity has been undertaken in 5 countries (Argentina, Australia, China, Germany and USA) as well as the UK. Psychometric work based on a large survey is being undertaken in the UK from September through to December 2018 in order to identify items for the new measure. This proposed study will collect and analyse additional Germandata to support the psychometric work and future item selection.Data on the German E-QALY test items, plus EQ-5D-5L and other existing instruments, will be collected from a sample of 500 members of the public, of whom 400 will report a health condition using a commercial panel. This German data will be analysed in line with the analysisof the UK survey data using basic psychometrics and Item Response Theory (IRT). These findings will be interpreted in the light of evidence from the face validity workin Germany, and country specific recommendations will be madefor the item selection process. These will be discussed in conjunction with other international evidence to support item selection for the new instrument.Wolfgang GreinerDescriptive Systems, EQ-HWB16900Ongoing20192019
20180310Non-parametric approach to valuing the EQ-5D-5LThe EuroQol Group uses an international protocol to standarizevaluation study design and preference elicitationacross countries. Followingdata collection, modeling the relationship between health states and elicited utilities is a key step in value set development. In theory, there are a large number of model specifications due to numerous possible interactions between dimensions. Obviously it is not feasible to investigate all of them. A common, practical approach is that the research team, based on prior evidence, or even arbitrarily, select and compare a few candidate models and identify the best one as the preferred model for the value set. Unfortunately, the true model specification is unknown. As a result, the final model used to produce the value set of the EQ-5D-5L differ across countries, despite the fact that they all follow the same protocol. We proposed a non-parametric approachto developing the value setwhich circumvents the need to specify the model. In our previous investigation using Canadian valuation study data, the non-parametric approach demonstrated substantial gains in in-sample and out-of-sample performance compared with commonly used parametric models.We are proposing to further assess the performance of the non-parametric approach in other valuation studies.If demonstrated to be effective for developing value sets, this approach wouldeliminate the need for specifying models, a known source of variations betweenvalue setsthat we should try to minimize. By using the non-parametric approach, we could standardize the modeling process and greatly facilitate comparisonsbetween country-specific value sets.Feng XieValuation28600Ongoing20182019
20180670EQ-5D-3L as an outcome indicator: analysis of its performance in a longitudinal study of patients receiving palliative careIt is well recognised that quality of life as measured by patient-reported outcomes (PROs) deteriorates near the end of life. This deterioration can be captured using the EQ-5D-3L, however, the extent to which the EQ-5D-3L dimensions and EQ-VAS are prognostic in advanced incurable cancer is unknown. This project encompassed three elements: - Using a simple prognostic model we demonstrate that the EQ-5D-3L self-care dimension alone can replace clinician-assessed performance status in predicting survival in advanced incurable cancer with bone metastases. Further, we illustrate that whilst the prognostic value of the self-care dimension is greatest for those with short survival, this diminishes in those with longer prognosis. Conversely the EQ-VAS offers greater prognostic separation at these later time points. The optimal measures for inclusion in future models will need to optimise model performance in line with the time points relevant to the clinical decision problem. - A one day inter-disciplinary workshop was convened to build a consensus on the use and challenges to including PROs in prognostic and predictive modelling in cancer care. Our consensus highlights the crucial need, particularly relevant in PRO models, to identify the decision problem and consider challenges of implementation at the earliest possible stage in model development. Beyond this we highlight key challenges to PRO capture in routine care and consider the impact of these challenges on prognostic and predictive models. - Finally, using longitudinal EQ-5D-3L data from patients undergoing palliative radiotherapy for bone metastases, we identify how the relationship between the EQ-5D-3L dimensions and the EQ-VAS changes with proximity to death. This analysis informs the wider discussions about the need to reconsider the evaluative space when appraising care outcomes for patients near the end of life.Katie SpencerPopulations and Health Systems17532Ongoing20192019
20180630'Equimetrics' of the EQ-5D. Measuring inequalities in health in the UK, Netherlands, and Italy to assess the potential of the EQ-5D-3L and 5L as outcome measures and determinants of income inequalityStudy aims:Explorationin a large 3-country dataset(UK, IT, NL):(part 1) the potential of the EQ-5Das an outcomemeasure in health inequality analysis, relative to an Overall Health Satisfaction (OHS) measure(as used in population surveys) and EQ-VAS;(part 2)the potential of the EQ-5Das determinantin income inequalityanalysis,as opposed to Health Satisfaction and EQ-VAS.Throughout the study 3L -5L comparisons are included, and an investigation of possible differences between countries and translations. Partof the study process is a consultation of the Health Inequalities SIG.Standard techniques are usedto describe and analyze inequalities(Gini; Concentration Index, Slope Index of Equality (SII), Relative Index of Equality (RII); decomposition analysis).Results are set against available literature both in terms ofcountry inequalities level and psychometrics with regard to 3L vs 5L.This study on focusses on the performanceof EQ-5Din this area, noton comparing methodologicalapproachesto inequality analysis.Methods:Three representative population datasets are available(each n>3500),for which country-specific value setsareavailable (except for 5L in onecountry).All analysis will be performed with and withoutstandardizationaccording to theEuropeanStandard Population. Part 1. A. Comparison of country-specific differences in health status, measured with the EQ-5D (Level SumScore, utility), by country (UK vs. Netherlands vs. Italy); in terms of descriptive statistics,and in terms of inequalities.B. Stratification of respondents according to health complaints; re-assessment of average health leveland inequalities. A. and B. will result in the reference data for C.C.Repeat the preceding analyses with OHS and EQ-VAS to assess congruence with EQ-5D results. D. Deductionfrom the results whichEQ-5D outcome choice performs best (across countries), and whether 3L or 5L shows better results.From country-specific patterns of inequalities (3L-5L) and additional analysis on distributionsit will be assessedwhether differential response functioning and/or translation bias exists,and if it influences previous conclusionson inequalities.Differencesbetween results according to the Level Sum Score (LSS)and the country-specific utilities are tested. Part 2.A. Determination of EQ-5D impact (in terms of dimension, categorized LSS, deciles of utility-score; the continuous scores of LSSor utility; 3L vs.5L) on household income as suchand on income distribution(men/women separately). B. Determination of best performing use of the EQ-5D (ranked data). We will check the EQ-VAS as determinantsimilarly. Part 3(during Part 1/2). Presentation of preliminary results and choices during face to face health inequalities special interest group (HISIG)meetingto further discussion on best use of EQ-5D. Results:Performance of EQ-5D in equality analysis (relative to other measures) as outcome and as determinant; within this: performance of options within EQ-5D, e.g. 3L vs. 5L. Additional: information on potentialbias through language and other effects. Further: discussion support of the HISIG.Juanita HaagsmaPopulations and Health Systems69700Ongoing20192022
20180350Extending the QALY Valuation Study in the UK: A feasibility study of applying different valuation methods to a health and wellbeing classification systemOBJECTIVES: The aim was to assess whether existing valuation methods were suitable for the nine item EQ Health and Wellbeing Short (EQ‐HWB‐S). METHODS: EuroQol Portable Valuation Technology (EQ‐PVT) which uses composite time trade‐off (cTTO) and discrete choice experiments (DCE) was modified for the EQ‐HWB‐S. Volunteer non‐academic University of Sheffield staff were recruited. A mixed methods approach involving qualitative interviews and assessment of quantitative data was used to assess the applicability and feasibility of EQ‐PVT to EQ‐HWB‐S. Participants valued six states using cTTO (three EQ‐HWB‐S and three EQ‐5D‐5L) and four EQ‐HWB‐S states using DCE. RESULTS: Nineteen participants with mean (SD) age 48.2 (13.0) were interviewed. Mean TTO values were ordered as expected with higher mean values for the mild EQ‐HWB state compared to the moderate and severe states. Most participants found it fairly or very easy to understand cTTO questions for both EQ‐HWB‐S 94.7% (18/19) and EQ‐5D 89.5% (17/19). Pain, activities and depression were considered key drivers for respondents’ choices. Additional information in the EQ‐HWB‐S was useful in helping to imagine what life would be like but it could also be overwhelming and make the tasks difficult. ‘Coping’ was a problematic item as it was either used as an overall assessment of the state or ignored in favour of participants’ perceived ability to cope with the state. ‘Coping’ was replaced with ‘control’ which did not have the same problems. Participants generally preferred DCE to TTO. DCE presentations with overlap but with simple formatting were preferred. CONCLUSIONS: A modified standardised valuation has been successfully applied to health and wellbeing states defined by the EQ‐HWB‐S. A full feasibility study is now required.Tessa PeasgoodValuation48775Completed20192020
20180770An EQ-5D-5L value set for the Swedish populationThe aim of this proposal is to develop a Swedish value set for EQSD-SL. We will do it by using standard data collection methods developed by the EuroQol group. Participants will be recruited in four regions in north, south and central of Sweden. Rural and urban participants will be included. 20% will be over 65 years and 80% in working age. Age, gender and rural/urban will reflect the Swedish population. The 800 informants will answer cTTO- and DCE-questions in face-to-face interviews. PI will conduct first 50 interviews before the training of interviers. Training of 12 interviewers will be conducted according to EuroQol procedures. The analysis will have the possibility to use both TTO and DCE data (hybrid modell).Klas Goran SahlenValuation80000Ongoing20192021
20180260Testing the robustness of the German EQ-5D-5L value set for people with health impairmentsObjectives: The national EQ-5D-5L value set for Germany is based on the average preferences of the general population. However, in Germany, there is an ongoing debate about the appropriateness of using general population preferences and whether patient preferences should be used instead. Therefore, the research study used the data from the German EQ-5D-5L valuation study and checked the published German value set for robustness against health impairments. Methods: Subgroups were built on the self-reported health measured by the EQ-5D-5L. To identify groups which significantly influence the value set model, different regression models were tested while controlling for preference heterogeneity. Backward selection based on the akaike information criterion (AIC) lead to significant subgroups analyzed in more detail. For each significant subgroup the value set model was separately estimated and comparison between models was done. Socio-demographics of the subgroups have been considered. Results: Three significant dummies were identified: health state 11111, severity level 5 to 7 and reported problems in dimension pain/discomfort. The six resulting value sets of the subgroups were compared to the national German EQ-5D-5L value set. It turned out, that there are only marginal deviations. The mean absolute deviation had a range from 0.004 to 0.013. No different densities were identified for the decrements from the different value sets. Control for socio-demographics did not change the results. Conclusions: People with self-reported health impairments do not have different EQ-5D-5L health state preferences than the German general population in general. Further research is needed if the ‘chronification’ of a health impairment to a chronic disease, leads to valuing health states differently by patients and the general population.Wolfgang GreinerValuation14400Completed20182019
20180490Test of the minimal number of C-TTO states in the valuation protocol of the EQ-5D-3L-YRecently, the Youth Working Groupagreed upon a protocol for the valuation of the 3-level version of EQ-5D-Y. The proposed protocol suggested a minimum of 5 health states to be valued by C- TTO, combined with DCE. We want to test whether such minimum of 5 TTO valuation is justified, by sampling an additional 20 health state comprise of 18 orthogonal health state and 2 health state with a higher level of the so called ‘misery index’. The primary comparison is the comparison of the DCE/5 C- TTO model with the hybrid DCE/25 C- TTO model, as the hybrid model seems, at least at this moment in time, to be the most used model when establishing EQ-5D value sets. The secondary analysis is the comparison between the DCE/5 C-TTO model with the 25 C- TTO model, thus based on C- TTO data only. To convert the DCE/C-TTO onto the full health-dead QALY scale we will use 4 methods: 1) anchoring DCE values using C-TTO value for worst state, 2) mapping DCE values onto the 5 TTO 3) mapping DCE values onto the 25 C-TTO and 4) combining DCE and the C- TTO values in a hybrid model. In total there will be 8 analyses: primary x 4 scaling methods + secondary x 4. All testing will be done in the context of arriving at a value set for the EQ-5D-Y in Indonesia, thus using a representative and sufficient large sample. Therefore this study will arrive at insights about how well the different of anchoring method perform in practice, using the minimum suggested health states and possible extensions of that minimum set. Secondly, the study will arrive an Indonesian value set for the EQ-5D-Y. An amount of 1,200 respondent will be drawn from a representative sample of Indonesian adult general population. There will be 2 groups of respondents: - The DCE group, consist of 1,000 participants living in 7 cities and its rural surrounding in Indonesia- The C-TTO group: consist of 200 participants living in 2 cities and its rural surrounding in Indonesia; Jakarta and Bandung, IndonesiaParticipants will be interviewed face-to-face and asked to complete the valuation task by using a preferred health state for the perspective of a 10 years old child.Titi FitrianiValuation, Youth29609Ongoing20192020
20180620Validation and Comparison of the Psychometric Properties of the EQ-5D-3L-Y and EQ-5D-5L-Y in the United StatesAim:The psychometric properties of the EQ-5D-3L-Y and EQ-5D-5L-Y in the United States are unknown.The aims of this study are to 1)test the psychometric properties (including feasibility, reliability, and validity) of the EQ-5D-5L-Y in children and adolescents with arthritis and leukemia and 2)compare the psychometric properties (including feasibility, redistribution, discrimination, and validity) of the EQ-5D-3L-Y and EQ-5D-5L-Y in children and adolescents with arthritis and leukemia.Methods:This study will recruit a total of 400 children and adolescents, including 100 patients with arthritis, 100 patients with leukemia, and 200 healthy individuals.Children and adolescents with arthritis and leukemia will be recruited from hospitals. Healthy individuals will be recruited from schools and matched with patients byage, gender, and race.Study participants will be asked to complete the survey in the order of 1) EQ-5D-5L-Y, 2) PedsQL, 3) self-rated health, 4) demographic questions, and 5) EQ-5D-3L-Y.They will be invited to complete the second survey seven to ten days after the baseline survey.The psychometric properties of the EQ-5D-5L-Ywill be tested through feasibility(using missing values), reliability(using Cohen's kappa statisticandintra-class correlation coefficient), convergent validity(using Spearman's rho), and known-groups validity(using a priori hypotheses). The psychometric properties of the EQ-5D-3L-Y and EQ-5D-5L-Ywill be compared through feasibility(using missing values and ceilingeffects), redistribution(using consistent response pairs), discrimination(using Shannon index and Shannon Evenness index), convergent validity (using Spearman's rho), and known-groups validity (using a priori hypothesesand relative efficiency statistic).Minghui LiYouth29200Ongoing20192023
20180730Validation of the UK English version of EQ-5D-Y-5L in South AfricaBackground:The standard EQ-5D-Y-3L has been used since it’s development in 2010 to measure health in the general population and in clinical samples in South Africa[1]–[4]. There have been concerns regarding the ceiling effects and whether the three-level response options are sensitive enough to changes in health. The EuroQol Foundation’s Younger Populations Working Group has recently developed a new version of the EQ-5D-Y with fivelevels of severity in each EQ-5D-Y dimension. The increased level descriptive system, the EQ-5D-Y-5L, has not yet been fully tested for validity or reliability in children between eight and 15 years. Aim:To assess reliability and validity of the UK English version of EQ-5D-Y-5L, in children and adolescents aged eight to 15 years in South Africa. Methods:Participants willinclude children and adolescents who areAcutely Ill (AI), Chronically Ill (CI) (with disease specific groups) and children of the General Population (GP) aged 8-15 years. In a large sample of childrenand adolescents, the construct validity of the EQ-5D-Y-5Lwill be established by comparing the results obtained to those from the PedsQLgeneric module.The known group validity will be established by comparison of results between Ill (AI andCI)and GP children. The discriminant validity will be determinedby comparing the EQ-5D-Y-5L scores within AI, CI and GP children who according to their classification on the self-rated health scale as well as within a sub-groups of chronically-ill children(with Cerebral Palsy)who are classified as mild, moderate or severe. Test-retest reliability will be established in the responses of the EQ-5D-Y proxy administered one day apart in group of childrenand adolescents from the GP and who have a stablechronicillness. Responsiveness will be explored through test/retest of AI childrenand adolescents at baseline and after24 and72 hours.Janine VerstraeteYouth45680Completed20192020
20180290Generation of an EQ-5D-5L value set for the Mexican populationAim: To develop an EQ-SD-SL value set for the Mexican general population. Background: There is strong interest in using EQ-SD-SL to support decision making in Mexico, for two reasons. First, HTA processes are used by the General Health Council of Mexico (CSG) to determine which new technologies are recommended for funding. To date, HTA has relied mainly on cost per life year gained and this is to be extended to include quality of life and QALYs. Secondly, the Mexican Secretariat of Health has as one of its main responsibilities monitoring of health care quality provided by public and private organizations. The General Directorate of Quality and Health Education, (DGCES) is in charge of such monitoring. Since 2013, the DGCES has been actively participating in the OECD Health Care Quality Indicators project. The DGCES has a well-established monitoring system based upon indicators that measure several aspects of health care quality from the provider perspective (DGCES, 2018) and wishes to extend that to include patient reported outcomes. An EQ-5D-SL value set for Mexico is required to support both uses of the EQ-SD-SL. Methods: The study will use the EQ-VT protocol and data quality monitoring processes that have emerged as best practice from previous value set studies. Value sets will be modelled from TTO data, DCE data, and both data combined via hybrid modelling. Deliverables: The principal deliverable will be a manuscript reporting the value set. Timelines: The project is planned to commence in September 2018 and to last for 18 months.Cristina GuttierezValuation0Ongoing20182020
20180300Development and psychometric testing of EQ-5D-5L bolt-on descriptors for vision and cognition: A study in the UK and AustraliaThe use of ‘bolt-on’ descriptors has been proposed to improve the sensitivity and responsiveness of the EQ-5D in certain contexts. The Descriptive System Working Group (DSWG) hascalled for proposalsto progress the bolt-on agenda by focussing on two areas, namely cognition and vision. However, to date, there has been limited research to develop appropriate descriptors using qualitative research or comprehensivepsychometric testing. The aim of this project is to identify vision (5L-Vis) and cognition (5L-Cog) bolt-on descriptors for the selfand proxy-completed adult versions of the EQ-5D-5L, using qualitative and quantitative methods. All stages of the researchwill include people with different types of experience of visual or cognitive impairment, including patients and carers. Phase 1 of the study will involve a literature review and the conduct of a series of focus groups. The purpose of Phase 1 is to identify relevant concepts and terminology in the definition of health-related quality of life with respect to visual impairment and cognitive impairment. In Phase 2, candidate descriptors will be administered in qualitative interviews. Cognitive debriefing willbe used to assess the acceptability and face validity of the bolt-ons. Phase 3 will entail quantitative analyses of data from a large sample to assess the psychometric properties of the selected 5L-Vis and 5L-Cog bolt-ons, using classical psychometric anditem response theory methods. The research will identify vision and cognition bolt-ons that can beappended to the EQ-5D-5L descriptive system andaresuitable for valuation and further testing.Brendan MulhernDescriptive Systems118785Ongoing20192020
20170520Development of health-related quality of life (EQ-5D-5L) value set for IndiaThe present study aims to develop EuroQol five-dimensional (EQ-5D-5L) health states value set for Indian population. A cross-sectional survey using the EuroQol Group’s Valuation Technology (EQ-VT) software will be undertaken in representative sample of 2700 respondents.The respondents will be selected from 12 districts in 6different states of India using a multistage stratified random sampling technique.The participants will be interviewed in a face to face setting using CAPI (computer assisted personal interviewing) technique. Time trade off (TTO) valuation will be done using 10 composite (cTTO) tasks and 7 discrete choice experiment (DCE) tasks.The demographic data will be analyzed by descriptive statistics. TTO values will be modeled using main effects model that will includeconstant and 20 main effectsderived from the EQ-5D-5L descriptive system, using ordinary least squares (OLS)and tobit models. The DCE data will be modeled under random utility using the conditional logit model. Hybrid modeling approach using both c-TTO and DC data to estimate the potential value setwill be applied. This method will combine the utility values elicited in the c-TTO for the 86 health states with utility values elicited in the DC experiment for 196 pairs of states. Sensitivity analysis will be conducted to explore the impactof severely inconsistent responders.Value set for EQ-5D-5L health state will be estimated for the Indiangeneral population. This will be helpfulin clinical practice/research for better monitoring of health-relatedquality of life. The scores can be used as an important input that better reflect Indian population’s preference for health technology assessment research. In addition, the results can be used for international comparison in order to understand similarities and differencesof health preference across populations.Shankar PrinjaValuation0Ongoing20172019
20180170R1Evaluation of routinely Measured PATtient reported outcomes in HemodialYsis care (EMPATHY) Trial: A Cluster Randomized Controlled TrialThe data collected for this project was used for 2 analyses: 1) to examine how the routine use of PROMs (including the EQ-5D-5L) influences patient-clinician communication in in-centre hemodialysis units in Northern Alberta and; 2) to describe the burden of depressive and anxiety symptoms reported by adults on in-centre hemodialysis in Northern Alberta, using PROMs, and understand patients’ and nurses’ perceptions of managing such symptoms. Both analyses employed a concurrent, longitudinal, mixed-methods research design. The quantitative data came from a multi-centre cluster randomized controlled trial (a separately funded study) of 17 hemodialysis units in northern Alberta that introduced a PROMs intervention. Patient-clinician communication was assessed using a modified Communication Assessment Tool (CAT). Depressive and anxiety symptoms were assessed using the Patient Health Questionnaire - 2 item (PHQ-2) and Generalized Anxiety Disorder - 2 item (GAD-2), respectively. Using purposeful sampling, patients and nurses were invited for interviews. Field notes were documented from dialysis unit observations. Patients’ responses to open-ended survey questions and nurses’ electronic chart notes related to mental health were compiled. Thematic and content analyses were used. For analysis 1, PROM use did not substantively improve patient-clinician communication. There was a small positive change in mean total CAT scores (range 1-5) from baseline to 12-months in PROM use units (0.25) but little difference from control group units that did not use PROMs (0.21). The qualitative findings provide in-depth insights into why PROM use did not improve patient-clinician communication; the purpose of PROM use was not always understood by patients and clinicians; PROMs were not implemented as originally intended in the trial, despite clinician training; PROM completion was seen to challenge communication; and PROM use was perceived to have limited value. For analysis 2, 29% screened positive for depressive symptoms, 21% for anxiety symptoms, and 16% for both. From patient (n = 10) and nurse (n = 8) interviews, unit observations (n=6), patient survey responses (n = 779) and nurses’ chart notes (n = 84), we discerned that PROMs (ESAS-r: Renal/EQ-5D-5L) had the potential to identify and prompt management of mental health concerns. However, opinions differed about whether mental health was within kidney care scope. Nonetheless, participants agreed there was a lack of mental health resources.Jeffrey JohnsonPopulations and Health Systems46983Completed20182020
20180070R1Use EQ‐PVT to develop a cancer patient preferences based EQ‐5D‐5L value setObjectives: To assess the feasibility of estimating an EQ-5D-5L value set using a small study design in cancer patients and to cross-compare the EQ-5D-5L health state preferences of cancer patients with those of the general public in China. Methods: Patients with clinically diagnosed cancers were recruited from two hospitals in Shanghai, China. In face-to-face interviews using the EQ-PVT survey, 31 health states (divided into three blocks) and 80 pairs of health states (divided into eight blocks) were valued by the cancer patients using the cTTO and DCE methods, respectively. cTTO data was modelled alone or jointly with DCE data. Forty-eight models using different model specifications (cross-attribute level effect [CALE] and additive models), model assumptions for effects (random/fixed), data heteroscedasticity (yes/no), and censoring (yes/no) were estimated. The best-performed model was identified in terms of monotonicity of estimated model coefficients and out-of-sample prediction accuracy. Results: Data of all the 221 cancer patients who participated in the study were included in this analysis. The hybrid CALE model using both the TTO and DCE data performed better in terms of prediction accuracy (Lin’s concordance coefficient=0.989; root mean squared error=0.058). Compared to the Chinese general population, the EQ-5D-5L value set based on cancer patients’ preferences were higher except for health states characterized by severe or extreme pain/discomfort. The rank of dimensional importance based on the general population was mobility (0.303), pain/discomfort (0.266), anxiety/depression (0.227), self-care (0.222) and usual activities (0.205), while in cancer patients, the order was pain/discomfort (0.421), anxiety/depression (0.267), mobility (0.262), usual activities (0.212) and self-care (0.211). Conclusion: This study demonstrated the feasibility of using a small design for developing EQ-5D-5L value sets based on cancer patients’ health preferences. Since there were signs of differences between the preferences of cancer patients and general public, it may be valuable to develop patient-specific value sets for use in clinical decision-making and economic evaluation.Zhihao YangValuation10250Completed20182019
20180140Validity, Responsiveness and Test-Retest of EQ-5D-3L-Y and EQ-5D-5L-Y and their proxy versions in Pediatric Patients in IndonesiaThe EQ-5D-5L-Y predicted to be more sensitive than EQ-5D-3L-Y, butthere is little evidence about the differencereported.For this reason, the Youth Working Group has prioritized research: comparing the 5L to the 3L in young patients. Theobjectives of this study areto compare the 5L-Y to3L-Yand assess the validity, reliability,and responsiveness of the 5L-Y across 5chronic diseaseswith different severities from the perspective of patients and parentin Indonesia.A sample of 250pediatric patients between 8-16 years old of 5 patient groups; chronic malnutrition, acute lymphoblastic leukemia, persistent asthma, major thalassemiaand hemophilia, will be invited to participate in this study.Data collection will target children withsevere health status among these patients group, while moderate and mild condition will alsoinvolve as part of severitiesvariation. Estimated population will be 100 chronic malnutrition, 30 acute lymphoblastic leukaemia, 70 persistent asthma, 25 major thalassemiaand 25hemophilia patients.Participantwill be interviewed and asked to complete the EQ-5D-3L-Y, EQ-5D-5L-Y, PedsQLGeneric Core Scales –Short Form 15, PedsQL Disease-Specific Modulesand General State of Health question. Their mothers will be asked to complete the proxy versions of the EQ-5D-3L-Y, EQ-5D-5L-Y, and of the PedsQL. For the patient groups there will be three measurements in time: 1) baseline, 2) the test-retestadministration after 3 days and 3)afollow up to measure responsiveness.The 3L-Y and 5L-Y will be comparedin terms of feasibility (missing values), redistribution properties, ceiling effect, discriminatory power, convergent validity and known-groups validity. Test-retestreliability will be assessed by comparing the results of the questionnairesfrom baseline and at 3 days after the first intake. Responsiveness will be the difference between baseline and the interview after the last intensive treatment. By administering the two versions of EQ-5D-Y in groupsof patients, we can determine the performance of the two in discriminating subjects between groups and in groups with different degrees of severity.Frederik PurbaYouth35263Ongoing20182022
20180160Testing 4 cognition bolt-on items in a community dwelling elderly groupIn responseto the 5thjoint call for proposals, we would like to submit this proposal to test 4bolt-on items for cognition. Several studies have identified cognition and vision as areas where the EQ-5D may be strengthened with bolt-on items in order to achieve wider usage. Some research has been conducted for cognition bolt-on candidates, but the measurement properties of cognition bolt-on items are hardly tested. In this study, we aim to test 4 candidate cognition bolt-on items for their psychometric properties and added value to the EQ-5D-5L in a community dwelling elderly group (N=500 to 600). We focus on discriminatory power of the cognition items and the new index score that may be generated from a cognition bolt-on EQ-5D-5L.Nan LuoDescriptive Systems16000Ongoing20182019
20180040R1NABackground: The island of Bermuda has embarked on a programme of reform for its health sector-including reorganization of the ways in which healthcare is delivered, evaluating changes to the services that are offered,and to the ways inwhich servicesare delivered. The analyses that are to be undertaken would be well served withdata concerningpreferences in health and population health status. Aims:The objectives of this project are to create an EQ-5D-3L value set,and to conduct a population normssurvey using EQ-VAS for Bermuda.Methods: EQ-PVTwill be used for the valuation study. 4 Interviewers from the Bermuda Health Council will be trained by experienced EQ-5D valuation study practitioners. The interviewers will collect data from a representative sample of 350 respondents. TheEQ-VAS study will be web-based with a target of 1,000 respondents.Henry BaileyValuation8800Ongoing20182018
20180150Furthering the DCE research agenda: Comparing anchoring and design methods for the valuation of EQ-5DRecently there has been debate within the EuroQoL Executive Committee and Valuation Working Group (VWG) about whether it is appropriate to use DCE as a stand-alone valuation method, and a form of DCE with durationwould potentially be a strong alternative. Two approaches that have been recently with promising resultsinclude presentingtriplets including pairs of EQ-5D-5L health states with duration, and a third option of either ‘immediate death’or ‘full health’. Questions about the optimal way to design the experiment are also important, and the two triplet methods have used theoretical (generator based) and algorithm-basedapproaches to selectingan efficient design.Further workabout these methods is requiredto support the executive committee in their decision making around the use of DCE. To inform this, the EQ-VT studiescurrently being carried out in Denmark and Peru have included an additionalset of questionsat the end of the face-to-face EQ-VT interview to collect data onboth approaches developed using both methods of generating the designed experiment. This results in four designsfor testing:1.Tripletswith immediate death developed usingagenerator design(usual method)2.Tripletswith immediate death using an efficient algorithmicdesign3.Tripletswith full health developed using agenerator design 4.Tripletswith full health developed using an efficient algorithmicdesign(usual method)The aim of this proposal isto requestfunding support for further analysis and dissemination of thedata from thePeruvian and Danish triplet add on studiesthat arecurrently underway.Brendan MulhernValuation14800Ongoing20182019
20180130The EQ-5D-5L valuation study in EgyptBackground: Egypt, the most populous country in the Middle East, has a population of 96.7million citizens, residing in 27 governorates. It is classified as a low- middle income country. No value set exists for either EQ-5D versions in Egypt or other Arabic speaking countries. Using other countries utility values carries some risks in not representing the views and preference of the Egyptian population. Aim: To develop a value set for the EQ-5D-5L based on societal preferences in Egypt to be used in economic evaluation studies and to support resources allocation decisions. Methodology: The valuation protocol will be administered using the EuroQol Group valuation technology (EQ-VT-2.0). The adult Egyptian participants will be recruited from different Egyptian governorates representing all geographical areas of the country as per the population distribution. Participants will be recruited through personal contact and from public places like shopping malls, university campuses, governmental authorities, parks and sports facilities using multi stratified quota sampling to select a representative sample in terms of age, gender, education and residence (urban/rural). Written informed consent will be obtained from all participants. The interview will take place at the interviewer’s office or the participant’s work place or home depending on the participant’s preferences.Samar FaridValuation0Completed20182020
20170310R2Interval TTO valuation approach (2nd revision)Background:Great improvements have been made with the refinement of the C-TTO part of the EQ-VT. However well explained, the C-TTO task stillrequires a great level of engagement. One of the most challenging aspects of the task for the respondent is to give a precise point of indifference, indeed, in any new data the proportion of responses given after reversing the direction on the iterative procedure keeps being low, indicating “satisficing”effects. In this proposal, we hypotheses that changing the C-TTO termination rule by: “a specific number of moves”and/or“a specific interval width”of values where the true preference is, whichwilllead tomore accurateinformation compared tothe current C-TTO task, as by definition we will avoid the satisficingeffect.Aims: The aims of this proposal are twofold. The first is to showthat changing the termination rule in C-TTO will lead to a reduction in response inaccuracy. The second aim is to calibrate the proposed interval C-TTO approach in terms of required number of moves and width of the interval responses. Method: We will collect 100 interviews.50%of respondent will value 10health states using currentTTO while the other 50% will use theintervalTTO. We will use the EQ-PVTin a one-one face to faceenvironment. We suggest using “15 moves”or “a width of one year”on the interval response as termination rules.In other words, if a respondent startscirclingin aninterval of one year the task will stopor,in other case,the respondent will be forced to make 15 adaptive choices in order to finish the task. This will allow a posterior calibration of the task, i.e., we will be able to apply different termination rules using same data, e.g. 2 years’interval width and 10 moves.This can be done by examiningthe path responses on the collected data.Juan M. Ramos-GoñiValuation22375Ongoing20172018
20170640Testing the potential of multiplicative models for efficient EQ-5D bolton/off valuation study designPrediction of bolt-on/offEQ-5D health state valuescan be based on new value sets independent of original EQ-5D value sets. Nevertheless, it would be more efficient to link bolt-ons to existing EQ-5D value sets. For this purpose, the effect of bolt-on/off items on valuation of EQ-5D dimensions should be elucidated. Previous studies found the effect of bolt-on items on valuation of EQ-5D health states was complex and failed to inform efficient valuation designs for bolt-on/off EQ-5D. Recently, multiplicative models were found to be superior to addictive models (which was used in almost all previous bolt-on valuation studies) and have been used to estimate 5L value sets. This new type of models provides a new opportunity to exploring the relationship between bolt-on/off and valuation of EQ-5D health states. The purpose of this project is to test the multiplicative models for their potential in exploring the effect of bolt-on/off items on valuation of EQ-5D health states. The specific research questions are: 1) are multiplicative models superior to the additive main-effects models for modelingvisionbolt-on (5L+VI) and self-care bolt-off (5L-SC) valuation data? 2: Are common parameters of the multiplicative models for 5L, 5L-SC and 5L+VI or ratios of the parameters constant? We propose to elicit the utility values of 30 5L, 29 5L-SC, and 31 5L+VI health states from 600 members of the general public using the c-TTO method. Each respondent will be asked to value arandom block of 5L, 5L-SC, or 5L+VI health states (n=14 to 16). Data will be modeled with both additive and multiplicative models and the model results will be examined to answer the research questions raised.Nan LuoDescriptive Systems, Valuation42300Completed20182019
20170530A Chinese value set for the EQ-5D-YThe first attempts to make national value set for the EQ-5D-Y are on their way as the Exec has agreed upon proposals for Spain, the UK, the Netherlands and Germany. This valuation task arrives at a latent scale though DCE in an online sample. These value sets studies also included a methodological research question about the age dependency of EQ-5D-Y values. Age dependency was found in Spain, the UK, and Germany, but not so clear in the Netherlands. The Dutch study also assessed another methodological question of whether the relative value of life years over quality of life is related to whom the tradeoffs apply. The study found the presence of a latent variable affecting differently on the tradeoff ratios for adults and children. In the present study we developed the first comparable value set for China, and further explored the two methodological questions in China. In order to make the results compatible with the studies done in Spain, the UK, the Netherlands and Germany, we used the similar protocol, notably the Dutch and work in close cooperation with the original researchers. The Dutch study was administered using an online survey. It includes three tasks: latent scale DCE, DCE duration questions, and questions assessing preference towards different kinds of QALY compositions (e.g. 2 years in full health, or 4 years in utility of 0.5).All respondents will receive the same tasks but will be randomized into three arms for three hypothetical persons in different ages(10-years, 15-years or 40-years). Our study translated the survey into Chinese language version, and add an additonal arm (70-years) as well as localize questions (e.g. the demographic questions). The survey provided data to addressed three aims, including: 1. Derive the preliminary value set of the EQ-5D-Y for China. 2. Assess age dependency of EQ-5D-Y values 3. Explore the presence of a latent variable affecting differently on time trade-off ratios for adults and children in ChinaJan BusschbachValuation14450Completed20182018
20170470Recent experiences using the EuroQol EQ-5D instrument In Latin America: The 3L & 5L; Public Health And Economic Evaluations; Newer Time Trade Off Variants And Discrete ChoiceExperimentsEQ-5D research has been presented in all six previous meeting of the ISPOR Latin America Consortium, starting on 2007 in Cartagena (Colombia) with two workshops: 1) HEALTH VALUATION RESEARCH IN LATIN AMERICA: AN AGENDA FOR CHANGE (Kind&Zarate), 2) EQ-5D: PERSPECTIVES AND POSSIBILITIES FOR LATIN AMERICA(Zarate&Badia&Kind&Augustovski), followed by several poster and oral presentations performedthrough the years during the followregionalmeetings.Inthe last ISPOR LA meeting, which took place inSantiago (Chile)in 2015, two EuroQol members received best oral presentations awards (Augustovsky&Zarate) for EQ-5D research performed in Uruguay and Chilei.Given thestrong influence of the EuroQol group in the Region, there have been many EQ-5D valuation studies performed in Argentina, Chile, Brazil, Uruguay, Colombia and recently in Ecuador. EQ-5D 3L has been included in national population surveys in Argentina (2005) and Chile (2016),bothinitiatives were organized by their local Ministriesof Health in each nation.Despite the steady increase of EQ-5D research, that have seen almost 100 publications related with Latin Americain Pubmedii, there are stillmany countries in Latin America where there should bea highinterest to measure and value self-perceived health, but alow level of knowledge about how to do it properly based on the best scientific evidence. For the coming 2017 ISPOR LA Regional meeting, a group of EQ-5D members and non-member hassubmitted oneworkshop and fourpresentations (one of them already selected for an award). All activitiesaim to provide crucial information to Latin American decision makers regarding the different uses of the EQ-5D instrument in Public Health and Economic Evaluation, as well as give examples of such activities performed in different countries in Central and South America.The Conference expectsclose to 1.000 attendees and we aim to cover at least 10%-20% of them by direct contactthough our fivepresentations already confirmedby the ISPOR LA ConferenceCommittee.victor zarateOthers12602Ongoing20172017
20170230Estimating the EQ-5D-5L Value Set for the PhilippinesThe Philippine National Formulary Executive Committee, a body tasked to selecting which medicines can be bought by the Philippine government, has been conducting health technology assessments. In the past, the group uses DALYs or QALYs weighted using Thai value sets. The group would like to standardize assessments and utilize QALYs weighted using the Philippine value set. The goal of this research is to generate the Philippine EQ-5D-5Lvalue set. The value set will be estimated using the standard methodology using the EQ-VT software translated to appropriate local languages. A purposive sample of 1,720 Filipino adults will be recruited to valuate 86 health states using TTO and 196 health state pairs using DCE. Previous modelling approaches as well as the hybrid approach will be explored to generate the valuation model. An exploratory study to assess equivalence of the translated software will be also done through a survey of bilingual speakers who will valuate the same set using two different languages.Hilton LamValuation25080Ongoing20172018
20170330Psychometric validation of the Chinese version of EQ‐5D‐Y for China in three medical conditionsAim: The Chinese version of EQ‐5D‐Y for china was recently developed accroding to the EuroQol Group’s translation guidelines. However, the measurement properties of this new EQ‐5D‐Y langauge version is unknown.This study aims to pscychometrically validate the Chinese version of EQ‐5D‐Y for China in three patient populations, including children and adolescents with juvenile idiopathic arthritis, congenital heart disease, and childhood leukemia. Methods: The Chinese EQ‐5D‐Y questionnaire will be administered together with Pediatric Quality of Life Inventory (PedsQL, a valdiated HRQOL instrument in China) to children and adolescents with juvenile idiopathic arthritis, congenital heart disease,and childhood leukemia (100 cases for each condition) seen in two tertiary hosptials in Shanghai, China.Clinical data of the study samples will be retrieved from medical records and the young patients who are in stable health status will be asked to complete the EQ‐5D‐Y questionnaire a second time within a week. A group of 200 healthy school children and adolecents will be asked to complete the EQ‐5D‐Y anf PedsQL questionnaires. Validity will be assessed by testing a priori hypotheses relating EQ‐5D‐Y to PedsQL and clinical measures. Test‐retest reliability will be evaluated using Cohen’s kappa statistic for the 5 EQ‐5D‐Y dimensions and the intra‐class correlation coefficient for the index score (using the EQ‐5D‐3L value set for China since no EQ‐5D‐Y value sets are avaialble) and EQ‐VASNan LuoYouth31900Ongoing20172018
20170360PROMs and PREMs, their interaction: bias or added value? On the dependency between EQ5D5L (stand alone PROM), and validated PREMs in a large sample of recently delivering women, ranging from healthy to severely affectedBackground EQ-5D5L rapidly disseminates as PROM, beyond the EQ-supported Canadian APERSU and the Swedish program. EQ-5D5L is often part of PROMs/PREMs datasets, with many emerging questions. Within EQ Request for Proposals, the PROM (EQ5D)-PREM relation is assigned priority: independent value of EQ5D5L is vital to its broader use, one wonders whether some PREM questions (EQ-style) could be added to EQ-5D5L, allowing at some stage combined preference measurement. The study dataset used here anticipates the compulsory Dutch national framework for perinatal outcome performance measurement, starting Jan 1, 2018 where EQ-5D5L is part of. We anticipate 1. questions on yes/no mutal casemix correction of PROMs and PREMs, and 2. the need for overall judgement scores (combining PROM and PREM), for which this study could create building blocks. Aims and research questions - General: to study in a large dataset with the mutual relationship between PROMs and PREMs, with multiple measures (=stated EQ research priority). - Specific: 1. using the health VAS scale (EQ-VAS) as overall health outcome, we will establish the role - if any - for PREM-outcomes as a added health-unrelated factor/confounder, beyond the known role of the EQ-5D5L domains and sociodemographic factors for EQ-VAS. - Specific: 2. using a 10-point quality of care/PREM scale (Picker) as overall client PREM, we establish the role - if any - for PROM-outcomes as confounder of PREM outcome, that is: a role beyond the predictive role of 8 independent, descriptive PREM domains (ReproQ-8D), detailed procedural and setting information, and sociodemographic factors on the Picker PREM scale. - Results of 1. and 2. are important for casemix decisions in using EQ5D and PREMs together. - Specific: 3. to explore systematically, per domain of EQ-5D5L (PROM) the influence (çarry-over effect) of PREM domains where it is assumed that this effect should be minimal. - Specific: 4. to test whether the interactions of 1.-3. (if any) are essentially unrelated to personal factors (deprivation) which supports straightforward use by clients of PROM and PREM data. - Methodological side aim: 5. Profiting from the unique dataset, we study - in the context of the preceding analysis - the different relation of predictors to VAS outcome in the upper vs the lower part of a VAS scale in this context in general. In pilot work, an accidental finding pointed to a general mechanism we can address. - All analyses will use EQ-MIDs as proposed by the Canadian APERSU group, and published MIDs of the PREMs used to calibrate findings in terms of 'relevant'. Data A large (n>8000) national dataset of recently delivering women, with data on PROMs (EQ5D5L, EQ-VAS, condition specific PROM mother, condition specific PROM child -via mother report), PREMs (8-dimensional WHO questionnaire called ReproQ, plus a Picker PREM score, and domain-preference. It covers about 70% of the pregnancy units in the Netherlands, and includes (after cleaning) 3800 antenatal, and 4800 postnatal responses, more than 3000 of these are paired (longitudinal). Units are available as anonymous, nominal variable. Special data feature: outcome data-triads The critical feature is that both PROM and PREM data are available in 3 operationalizations: - as a multidimensional profile (EQ5D: 5 separate domains, ReproQ-8D: 8 separate domains) - as a weighted sumscore of both profiles (preference and non preference-based) - as a VAS (EQ-VAS; Picker-10 point scale for overall process quality) Both in the PROM and PREM case, the associate VAS/scale can (but not necessarily will) cover more than the contents covered by the multidimensional descriptive profile. This feature underlies many of the unique analytical options of our study. Analysis The dataset has been prepared for 90%, as part of an ongoing fast track proposal. After datacleaning, we will apply essentially the same analytical strategy using EQ-VAS & EQ5D5L as PROM data, and Picker-scale & ReproQ 8D as PREM data, where we try to strive for a uniform protocol as this informs on perhaps mechanisms of confounding common to PROMs and PREMs. Aim 1. First we will predict EQ-VAS from EQ5D5L descriptive domain data, adding variables in blocks, as personal background, health factors (=concept-related), ending with PREM (concept-unrelated) data as candidate predictors. Main effects, interaction; multi-level analysis. This approach shows the independent minimal impact of PREMs - if any - (determinant) on PROMs (outcome). Aim 2. With reversed roles, we do the same with PREM as primary outcome: EQ-VAS for the PROM analysis is replaced by the Picker-PREM scale, while EQ-domains are replaced by their ReproQ- domains. The other covariates are the same. We start here 8 domains of ReproQ, then backrgound, then procedural information, ending with in this context unrelated health information of mother (EQ as PROM) and newborn (a specific PROM for the baby, validated). Aims 3 (and 4). With stratified analysis we will carefully compare the relations domain-wise; in some cases the PROM-PREM relation could be in part causal, while in others this is impossible. This analysis takes on board the within-PROM and within-PREM dependencies. We will explore whether casemix correction of a PROM with PREM data, and the reverse, should be considered. Aim 5 (methods): the test of scale dependency will done by stratifying the dataset into 3 (perhaps more) groups/levels according to health level / quality of care level, testing whether VAS-relations are scale-dependent with different approaches. Answers provided- Whether the EQ-5D5L instrument is suitable as stand alone maternal PROM in pregnancy care performance measurement. What bolt-on could be considered, if any. - To what extent EQ-5D5L (as a PROM) and validated PREMs provide sufficiently independent performance measures in a practical real life situation - What PREM domains can be considered to be part of an independent PREM in EQ-style, if such an option should be deemed desirable. - Whether casemix adjustment of EQ5D (as PROM) with PREMs results should be considered, and the reverse, as both measures often coincide in current performance frameworks - Methods: whether we should change the current interpretation, and analytical practice of EQ-VAS data (including experience-based studies), which assumes uniform effects of predictors across the scale.Gouke BonselPopulations and Health Systems33350Ongoing20172018
20170290Testing and comparing the Spanish version of EQ-5D-3L-Y and EQ-5D-5L-Y in general and cancer young populationBackground. There is a need to compare the performance of EQ-5D-3L-Y and EQ-5D-5L-Y, and there is no validation of none of them in young patients with cancer, one of the most relevant and prevalent market in children. The aim is to assess and compare the psychometric properties of the Spanish version of EQ-5D-Y of to classification system (5 and 3 levels) and its validation in general and young children with cancer.Participants 1400 children and adolescents (8-18 years old) from primary and secondary school of Extremadura, Castilla-León and Andalucía (Region of Spain) and 100 children with cancer from Region Associations of Cancer respectively will take part of the study. Methods. A core set of questionnaire will apply to assess the HRQoL composed by EQ-5D-(3L&5L)-Y, Kidscreen-27, SDQ questionnaire, Cantril Ladder and VAS. Statistics descriptive, missing values and reliability method will perform to analyses the aim of this research. To compare the performance of 3L-Y AND 5L-Y will apply the Shannon index and the Shannon Evenness indexto assess discriminatory power in the VAS for each level of dimension in both 3L and 5L questionnaires.Narcis GusiYouth14900Ongoing20172019
20170280Revisiting EQ-5D-3L tariffs – An international collaboration between Slovenia end PortugalIntroduction: The two primary objectives were (a) to develop first logically consistent TTO based EQ-5D-3L value sets for Slovenia and (b) to revisit earlier developed VAS-based EQ-5D-3L value sets. Methods: Between September 2005 and April 2006, face-to-face interviews with 225 individuals in Slovenia were conducted. Protocols from the Measurement and Value of Health study were followed closely. Each respondent valued 15 health states out of a total of 23. Model selection was informed by the criteria monotonicity/logical consistency. Predictive accuracy was assessed in terms of the mean square difference between out-of-sample predictions and corresponding observed means, as well as Lin’s Concordance Correlation Coefficient. Results: Modelling was based on 2,717 VAS and 2,831 TTO values elicited from 225 respondents. A 6-parameter a constrained regression model with a supplementary power term was selected for VAS and TTO value sets, as it produces monotonic values, and proved superior in terms of out-of-sample predictive accuracy over the tested alternatives. Conclusion: This is the first EQ-5D-3L TTO-based value set in Slovenia and the second in Central and Eastern Europe (besides Poland). It is also the first monotonic and logically consistent VAS value set in Central and Eastern Europe. Comparisons with Polish and UK TTO values show considerable differences, mostly due to mobility with having a substantially greater weight in Slovenia. The UK value set generally produces lower values and the Polish value set higher values for mild states.Valentina RupelValuation15000Completed20172017
20170180An EQ-VT study of heart disease patientsOriginally the study was aimed to investigate the impact of heart disease patients’ demographic, socioeconomic, and disease characteristics on the valuation of EQ-5D-5L health states. That is, investigate to what extent these characteristics affect differences in valuation between heart disease patients and the general population. As an EQ-5D-5L valuation study in the Singapore general population was ongoing at that time, we planned to recruit only heart disease patients and use the data from the Singaporean general population value set study to achieve the planned objectives. However, the official Singapore general population value set for EQ-5D-5L is still not available, so we decided to publish the valuation data from heart disease patients only. We still aim to compare the valuation data from heart disease patients and the general population once the official value set will be available. Below is the abstract of the publication based on heart disease patient data only. Objectives: Several studies have shown that patients with heart disease value hypothetical health states differently from the general population. We aimed to investigate the health preferences of patients with heart disease and develop a value set for the 5-level EQ-5D (EQ-5D-5L) based on these patient preferences. Methods: Patients with confirmed heart disease were recruited from 2 hospitals in Singapore. A total of 86 EQ-5D-5L health states (10 per patient) were valued using a composite time trade-off method according to the international valuation protocol for EQ-5D-5L; 20-parameter linear models and 8-parameter cross-attribute level effects models with and without an N45 term (indicating whether any health state dimension at level 4 or 5 existed) were estimated. Each model included patient-specific random intercepts. Model performance was evaluated for out-of-sample and in-sample predictive accuracy in terms of root mean square error. The discriminative ability of the utility values was assessed using heart disease-related functional classes. Results: A total of 576 patients were included in the analysis. The preferred model, with the lowest out-of-sample root mean square error, was a 20-parameter linear model including N45. Predicted utility values ranged from 20.727 for the worst state to 1 for full health; the value for the second-best state was 0.981. Utility values demonstrated good discriminative ability in differentiating among patients of varied functional classes. Conclusions: An EQ-5D-5L value set representing the preferences of patients with heart disease was developed. The value set could be used for patient-centric economic evaluation and health-related quality of life assessment for patients with heart disease.Nan LuoValuation0Ongoing20172018
20170020Comparison of different model specifications for the frequentist estimation of random effect hybridThe overall aim of the proposed project is the development of a theoretical framework for hybrid models capturing intra‐person correlation and the corresponding estimation of these models within a frequentist framework using different approaches to estimate the DCE‐component of the hybrid as well as the Random Effect, respectively. The first part of the project provides an in‐depth discussion of current hybrids for the EQ‐5D and based on this existing literature develops a theoretical model specification of a hybrid model, which captures the different scales of TTO‐ and DCE‐values and accounts for intra‐person correlation. The second part of the project consists of a simulation study allowing the assessment of different methods to estimate such a Random Effect hybrid‐model and Logit‐ as well as Probit‐parts for the DCE data. The statistical and computational performances of the newly developed estimators will be compared among themselves and against existing estimators proposed in the literature and the corresponding estimation‐functions in R will be documented and made available. Regardless of the methods used for estimating the Random Effect, our implementation will be rooted in the frequentist framework which has the appeal that potential users have all familiar tests available and can interpret the model and inferential results the ‘textbook way’.Wolfgang GreinerValuation22200Ongoing20172017
20170010Guidance on methods for analysing data from EQ-5D instrumentsAt the heart of the EuroQol Group’s vision and mission statements is an ambition “to support individuals and organizations across the world seeking to use those instruments” and “to actively promote the transfer of knowledge and evidence regarding the use, analysis, and interpretation of measures developed by the EuroQol Group” https://www.euroqol.org/euroqol/mission.html. Yet, despite nearly 30 years of the use of EQ-5D and related instruments developed by the EuroQol Group across the world, we have never provided comprehensive guidance to users on how to analyse the EQ-5D data they have collected. The User Guides focus on the collection of data. Brief guidance is available to users of EQ-5D value sets (Devlin and Parkin 2006 ). Krabbe and Weijnen (2003) touch on some ways of analysing EQ-5D profile data, but it is not a current reflection of available methods. Despite widespread use of EQ-5D, it has been noted that the data “are often under-reported, and inadequately analysed. The bottom line is – if you collect these data from your patients, you should be committed to making sure you learn as much as possible from what they tell you” (Devlin 2016 ). A particularly common weakness is to restrict analysis to the EQ Index, without analysing the profile data or EQ-VAS. The aim of this project is to produce “Guidance on analysing EQ-5D data”, drawing together the various ways in which EQ-5D data (EQ-5D profiles, EQ-VAS and EQ Index) can be analysed and reported, and providing clear guidance to users. The Guidance will be accompanied by open-source code in R, STATA, SPSS and SAS, to ensure it is as accessible as possible to all potential users, and the production of a slide deck which can be used in training workshops and webinars. Our goal is to ensure that, where EQ-5D data are collected, whether from patients or the general public, users are encouraged to make full use of those data to gain insights that ultimately help to improve patient care and peoples’ health.Nancy DevlinEducation and Outreach36225Ongoing20172017
2016640Building values sets based on TTO results by averaging model predictions and actually observed meansWhen constructingvalue setsbased on TTO, usually a subset of states is used in a study, and the results are extrapolated by econometric modelling. Also in saturated studies, the value set is typically built entirely from model predictions. Then, for each given state, the utilities actually observed in the sample are used only insofar as they impact the model parameters, and the most relevant information, whose validity does not rely on any model specification, is only indirectly used (a loss of information, especially if frequent states were used in valuations). We aim to propose a method of constructing value sets by averaging the model predictions and the directly observed values (for states included in the study), with weights based on comparing the model fit (hence, the variances of the predictions) and the variance of direct observation. This methodological study will mostly use theoretical considerations and working on simulated dataset, in order to discover how capable various approaches are of discovering the true (known only for simulated datasets) mean values. We also plan the feasibility study, based on Polish 3L study. We plan to build on three streams of literature: 1) linear mixed effects models (designed to account for deviations from the mean across several dimensions, e.g. a state, not only a respondent), 2) Gaussian random field models (nonparametric models designed to directly take into account local information about observations mean and variance), and 3) Bayesian averaging methods (which allow effectively weighting evidence from different prediction models).Michał JakubczykValuation23400Ongoing20172017
2016750EQ-5D-5L in pregnancy. Antenatal and postnatal HRQOL, the impact of poor outThe general aim of this study is to describe, in terms of the EQ-5D5L: (A) the health in pregnancy and the maternity period, using a representative large nationwide dataset (n>8000), which - apart from the EQ data - have been used for developing an international instrument to study care delivery in pregnancy/maternity units. (B) describe health inequities along several criteria (income class, education, living area, ethnic background) using conventional direct methods, in general and after subdivision into clinically relevant subgroups. (C) to predict in women without antenatal clinical problems (>80%), postnatal health from 1. antenatal health, 2. personal characteristics, 3. delivery course, 4. the reported health of the baby, and, 5. facility/pregnancy unit. We will apply some known group comparisons to establish sensitivity of EQ-domains for common clinical conditions (episiotomy and ceasarean section: pain/discomfort; ceasarean section: mobility; poor outcome child: anxiety&depression).Gouke BonselPopulations and Health Systems14950Ongoing20162017
2016650A qualitative approach to understanding what aspects of health are important to people Ð Australian extensionThe March 2016 call for proposals states the following: “given that the EuroQol Group operatesglobally, generalizability of the importance of dimensions of health is not only relevant between different types of populations but also across cultures/countries”.Recently a project with the same team as this proposal (PI Koonal Shah)titled “A qualitative approach to understanding what aspects of health are important to people” was funded. Key aims of that project includedevelopingand piloting an approach to identifywhat aspects of healthare important to people and to obtain the views of patients and members of the public about what aspects of health are important to them. The project includes four stages:Stage 1: Review of English language definitions of key conceptsStage 2: Development of a survey designed to understand what aspects of health are important to people, with piloting in a focus groupStage 3: Administration of the revised survey to patient and general public samples (n=200), with the aim of generating a list of potential domains for a generic classification system, Stage 4: Checking / triangulation of the stage 3 results using a second focus groupGiven the need to understand what aspects of health are important to people internationally, the aim of this research project is to extend the Shah et al proposal to repeat Stage 3 with an Australian population (including general population and patient groups) using the same survey.Both the DSWG and exec reviewer noted that a limitation of the Shah et al project was that it was UK only. This extension is designed to address those concerns by covering a different culture and region.The study is a low cost extension to collect data from a second English speaking country using the same methodology(in line with DSWG recommendations), and will inform the ongoing research agenda around the measurement of health and health-related quality of life beyond the existing EQ-5D descriptive system.Brendan MulhernDescriptive Systems19750Ongoing20172017
2016710Going beyond health related quality of life – towards a broader QALY measure for use across sectorsAbstract Please provide a short summary of your completed research. This information will be included in the EuroQol database on funded research and may also be presented on the research section of the EuroQol website Quality adjusted life years (QALYs) are widely used around the world to inform health care decisions, such as the reimbursement of pharmaceutical products. Existing measures for estimating QALYs are mostly limited to health-related quality of life and focused on physical health. This will miss aspects of quality of life shown to be important to many patients, particularly those with long terms conditions and those reviewing social (i.e. non-medical) care. Furthermore, these measures are not designed to assess the impact on informal carers. We have developed a new generic measure of the impact of health and social care interventions on the lives of services users and their carers. The EQ Health and Wellbeing (EQ-HWB) and a short version called the EQ-HWB-s (based on a sample of items from the longer version) were developed using qualitative evidence from service users (health and care services) and carers, along with psychometric evidence, collected in six countries.John BrazierDescriptive Systems94367Completed20172018
2016610A Randomised Controlled Trial of the effect of Short-Stretch Inelastic Compression bandages on Knee Function following total knee arthroplasty: Comparison of EQ-5D-3L and EQ-5D-5LObjectives: The aim of this research is to undertake a comparison of differentEQ-5D-3L and EQ-5D-5L value sets (UK, SP, NL) in a population of 2600patients withknee arthroplasty for osteoarthritis in the UKover a 12 month period. Methods:The comparison will take place as part of arandomised controlledtrial which aims to assess the effectiveness and cost-effectiveness of a two layer compression bandage versusstandard bandage post-operatively on patient reported outcomes in total knee arthroplasty patients. The trial aims to recruit 2600 participants and as part of the trial EQ-5D-3L data will be collected at baseline (pre-operatively)and 6 months (via theUKNHS Patient Reported Outcome Measures(PROMs)programme) and at 12 months(via a questionnaire). There is the opportunity toalso collect data on the 5L version in this population of patients, in order to undertake a comparison of the value sets, i.e. the 3L value sets versus the 5L value sets. The proposed research will assess the sensitivity to change of the 5L value setsby administering boththe3L and 5L versionsat baseline and at 12 monthsvia paper-based questionnaires. The projectwill be the firstto compare the EQ-5D-3L and EQ-5D-5Lvalue sets using panel data forthis population of patients, to our knowledge. The comparison will assess whether the new value sets for the 5L version aremore sensitive to the changes compared with the previous 3L value sets.Juan M. Ramos-GoñiDescriptive Systems14850Ongoing20162019
2016540Health utilities used in economic evaluations of cancer treatmentsCost-utility analysis has been widely used in economic evaluations for cancer treatments, which is required to inform coverage and reimbursement decisions in many countries. Health utilities are a key set of measures to capture health benefits to which many economic evaluations are often found sensitive. We will use cost utility analyses included in the Tufts CEA Registry to identify the source studies on health utilities for cancer patients. By reviewing and extracting all these source studies, we aim to 1) understand to what extent the EQ-5D has been used to measure health utility in oncology cost utility analysis; and 2) identify whether or not the use of health utilities in oncology economic evaluations has been appropriate. Proper use of health utilities for cancer economic evaluation is extremely important given their implication in resource allocation decision-making.Feng XiePopulations and Health Systems14958Ongoing20162017
2016580The feasibility of using the EQ-VT program to conduct the EQ-5D-5L valuation study in rural ChinaBackgroundThe EQ-5D-3L valuation study has been conducted in bothurban and rural areas in China using the Paris protocol. Urban and rural residents were able to understand and finish valuation tasks using time boards and cards in face-to-face interviews. The EQ-5D-5L valuation study, however, has been conducted in urban Chinaonly.Urban residents did not have problemsin using the EQ-VT programin computer-assisted personal interviews.The large urban-rural disparity in computer and Internet access poses a significant threat to the successfulness of using of the EQ-VT program to conduct the EQ-5D-5L valuation study in rural China. No studies have examined the feasibility of using the EQ-VT program among rural residents in China. AimsThe aimof the study isto explore the feasibility of using the EQ-VT program in computer-assisted personal interviews to conduct the EQ-5D-5L valuation study in rural China.MethodsThe proposed study is a primary research. A convenience sampling method will be used to recruit 120 participants lived in rural China. Quotas will be set based on age, gender, and education level.Computer-assisted personal interviewswill beconducted using the EQ-VT program. In the end of the interview, each participant will be asked to complete a supplemental questionnaire examining the difficulty in using the EQ-VT program. If participants have comments or suggestions on the EQ-VT program, they will be invited to joina focus group discussion. The study will analyze quantitative data from the supplemental questionnaire and qualitative data from the focus group to identify, examine, and interpret reasons ofhavingdifficulties in using the EQ-VT program. Based on study findings, strategies to help participants better understanding and completing the EQ-5D-5L valuation tasks using the EQ-VT program in rural China will be proposed.Minghui LiValuation15000Ongoing20162023
2016280Valuing EQ-5D-5L in Australia: A comparison of the EQ-VT protocol and DCE with durationThe EQ-5D is widely used to inform Australian health care reimbursement decisions. An Australian study to value EQ-5D-5L using DCE with duration (DCETTO) funded by the National Health and Medical Research Council (NHMRC) is currently underway, led by the proposed investigators. However no data based on the preferences of the Australian population collected using cTTO and DCE without duration as recommended by the EQ-VT protocol are available. It is important for both methodological development and decision science to compare values generated using different methods. This study aims to: 1). Collect preference data in Australia using the EQ-VT protocol. 2). Compare valuation data and the values produced by c-TTO and DCE data (from EQ-VT) and online DCETTO valuation data (from the NHMRC study) to inform ongoing research around the development of TTO and DCE. 3). To compare the application of the two methods in terms of practical issues around the administration of the tasks, respondent and data collection burden. We will collect preference data from an Australian general population sample using the EQ-VT protocol. The sample for the EQ-VT will be drawn from the sample undertaking the online DCETTO, enabling within person comparison of methods. To estimate utility values, we will model the TTO and DCE data separately, and also use -5D-5L value sets produced using the EQ-VT and DCETTO protocols will be compared statistically to disentangle the extent to which different valuations (using similar functional forms) are driven by elicitation method, and the underlying model of preferences.Rosalie VineyValuation80000Ongoing20162018
2016230The use and research of EQ-5D instruments in East and South-East Asia: a systematic reviewThe use and research of EQ-5D in Asia are both increasing. However,it is not entirely clear how strong is the increase or what populations have be involved. Furthermore, for the use of the instrument in the region, what are the primary and secondary purposes of usage is unknown; for the research of the instrument, how EQ-5D has been assessed and to what extent it has been found to be appropriate for use in Asians are still elusive. The aims of this project are 1) to describe the use of EQ-5D, and 2) to summarize the measurement properties of EQ-5D in east and south-east Asian populations where the vast majority of the researchand use of EQ-5D in Asia is from. To achieve the study aims, we propose asystematic review approach to identify and analyse the relevant publications in the literature.A systematic literature review(SLR), following the Cochrane guideline, willbe conducted. The search will target at relevant publications using data fromthe following Asian-Pacificcountries: China(includingHong Kong), Indonesia, Japan, Malaysia, the Philippines, Singapore, South Korea, Taiwan, Thailand, and Vietnam. Atwo-stage SLR has been designed: a review covering major electronic databases will be conducted first as stage one; following by stage two where non-English papers published in local peer-reviewed journals which are not indexed in the major electronic databaseswill be hand-searched. The search will cover the period from 1995 to 2015, with no language restriction.Nan LuoEducation and Outreach74000Ongoing20162017
2016410The impact of color coding and the optimal degree of overlap in discrete choice experimentsBackground: in EQ-project 201543, we established that implementing attribute level overlap allows increasing the number of choices if the DCE-part of the EQVT gets an update, as it promotes choice consistence. Questions remain about the optimal amount of overlap, and about the interpretation of the results on color-coding to help respondents identify overlap. Aim: to collect additional data using the exact same setup as a study that was previously conducted in the Netherlands, yet now in an English speaking country (i.e. the UK). These data may validate the results of the previous Dutch study, investigate optimal amount of attribute overlap, and test the hypotheses that color coding will have a different effect on the English than the Dutch version. Methods: this study varies choice task complexity by implementing attribute-level overlap and color coding as strategies to reduce task complexity. The individual and combined impact of these strategies on choice behaviour is evaluated using Bayesian mixed logit models that allow for the evaluation of choice task variability and identification of the separate and combined effects of color coding and overlap relative to a base-case scenario without these strategies. Study arms: A randomized controlled trial with 8 study arms will be used to investigate the impact of color coding and various degrees of attribute-level overlap on the standard EQVT DCE format (i.e. without death or duration). All DCE designs will be Bayesian efficient and QALY balanced optimized.Elly StolkValuation19375Ongoing20162016
2016250The relationship between the EQ-5D and surgical outcomes in a large Australian registry of percutaneous intervention patientsQuality of life following percutaneous coronary intervention (PCI) in patients with coronary artery bypass graft surgery (CABG) has been reported as lower than non-CABG patients, however previous reports pre-date modern developments in PCI and cardiac surgery. This study aimed to examine the 30-day QoL after PCI between patients with and without prior CABG using a contemporary dataset. A retrospective analysis of the Victorian Cardiac Outcomes Registry was undertaken. This study included 36,799 patients who completed the EQ-5D questionnaire that was used to assess the 30-day QoL and was compared between groups with and without prior CABG at baseline. Most of the participants were older than 65 years, more than half were male and had PCI due to acute coronary symptoms (ACS) and nearly 90% of patients received drug eluting stents. Compared to the ‘no prior CABG’ group, the ‘CABG’ group had a significantly higher rate of reporting a health problem (OR 1.30, 95% CI 1.10–1.53), presence of a problem in mobility (OR 1.42, 95% CI 1.15–1.75), personal care (OR 1.49, 95%CI 1.13–1.97) and usual activities (OR 1.39, 95%CI 1.15–1.68), pain/discomfort (OR 1.31, 95%CI 1.11–1.54), and anxiety/depression (OR 1.20, 95%CI 1.02–1.42). Despite modern developments in both PCI and CABG, our study showed a consistent negative association between prior CABG status and 30-day QoL following PCI. There is a need for better targeted cardiac rehabilitation in patients with prior CABG to address their greater relative risk of experiencing poor health.Richard NormanPopulations and Health Systems15000Ongoing20162022
2016260Comparing DCE designs that can be used to value EQ-5D-5LThe use of Discrete Choice Experiments (DCE) to estimate values for EQ-5D has grown in recent years. To estimate the parameters with confidence, a key feature of these studies is the method used to select the choice sets to be valued. However studies comparing designsdeveloped for the purpose of valuing EQ-5D that usedifferent construction methods and different software have not been carried out. Also the impact of sample size across designshas also not been fully investigated. The aim of this project is to use simulation methods and primary data to compare three prominent approaches to the design of DCE studies usingEQ-5D-5L.The approaches are:Approach 1: An optimal generator developed design Approach 2: An Ngene developed D-Optimal design with zero priors using a modified Federov swapping algorithmApproach 3: An Ngene developed D-Optimal design with non-zero priors using a modified Federov swapping algorithmWe willtest the approaches using a DCE presenting pairs ofEQ-5D-5L profiles. Three designs for each approach including 100 choice setswill be developed.Firstly all designswill be explored using simulationto assessthe ability of each design to return an unbiased estimate of each of a number ofan assumed set of coefficients. Secondly, data will be collected online for thedesign from each of the three approaches that the simulation testing suggests is the most efficient. This will allow for primary data comparisons of the models across the approaches and in comparison to the results from the simulation stage.Brendan MulhernValuation34000Ongoing20162017
20160502-day entry level course describing development and current status of EQ-5D "technologies" for Russianot availablePaul KindEducation and Outreach4350Ongoing20162016
2015090Revisiting the MVH study: new methods for modelling UK valuations for the EQ-5D-3LBackgroundThe EQ-5D-5L value set for England project, usingthe EQ-VT protocol, was completed in 2016. A notable aspect of that study is the innovative econometric approach usedto modelthe EQ-5D-5L value set. Compared to the methods used in the UK MVH study, this approach differs in at least five ways:(1) the probability distribution of the errors at the top and the bottom scale of modelling the TTO data is handled differently;(2) in addition to treatingTTO data as continuous variable, further exploration is also conducted to treat TTO data as intervals;(3) account is taken of the heteroscedasticity in the errors of modelling TTO data;(4) heterogeneity in the respondent’s values is explicitly modelled;and (5) different types of preference data obtained from respondents are modelled simultaneously within a ‘hybrid’ approach. Thesemethod can beused in other data sets and can also be applied to earlier valuation datae.g.obtained for the EQ-5D-3L. Revisiting the UK MVH study data and applying those same methods will (a) facilitate more direct comparison between the characteristics of the values in the UK MVH EQ-5D-3L and England EQ-5D-5L value sets, and (b) allow us to determine what difference these modelling approaches would make to the UK value set estimated for the EQ-5D-3L. Insights from this paper willbe relevant to the wider issue of how the 3L and 5L compare; in particular, how differences between the 3L and 5L value sets affect the distribution of 3L Index and 5L Index data. AimsTo apply the econometric approaches that were developed in the EQ-5D-5L value set study for England to the EQ-5D-3L data from the UK MVH study, in order toexplorethe differences in values that result. Data MVH data are available from n=3,395 participants; a representative sample of the non-institutionalized adult population of England, Scotland, and Wales. To facilitate comparisonswith the results from the MVH study, the modelling analysis in our modelling workwill use the same 2,997 respondents as in the MVH study. Three types of dataavailable from the MVH study, i.e. TTO, VAS and ranking data, will be used in the modelling work. For each respondent, the dataconsists of 12 TTO tasks, 12 VAS tasks, and 15ranking tasks. MethodsWe will estimate models both for TTO data alone and hybrid models of TTO with other preference data (VAS valuation and ranking data) applying the same ideas underlying the 5LEnglish valuation study with respect to the error distribution, heterogeneity, heteroscedasticity and censoring. Special characteristics of the MVH data/modelling will be addressed, including the rescaling of values < 0; N3 term, and the availability of both VAS data and ranking data. Resultsand conclusionsAnalysis is now underwayandresults will be available for the plenary paper deadline. Our paper will discuss the implications of our results for understanding differences between the characteristics of 3L and 5L value sets and their use when applied to profile data e.g.in cost-effectiveness analysis.Ben Van HoutValuation36000Ongoing20152016
2015250Intercept investigation: Does the value drop from full health to any EQ-5D problems reflect preferences, or is it an artefact of the valuation method?Background: In all published EQ-5D tariffs, both for the -3L and the -5L versions, the drop in HRQoL associated with the movement from state 11111 to the second best state is greater by two orders of magnitude than any other single incremental impairment. One hypothesis that has been proposed is that this drop is related to the wording of “full” or “perfect health”, as opposed to the less perfect state 11111. However, when tested, this hypothesis has not found support in empirical data. That leaves two primary candidate explanations for the drop: it might reflect “true” preferences, or it may be an artefact stemming from a tendency among respondents in valuation studies to strengthen contrast in the valuation tasks.Kim RandValuation34750Ongoing20152017
2015420A City-wide survey of HrQoL in children using EQ-5D-YPaul KindYouth15274Ongoing20152015
2013300The impact of framing effects on EQ-5D-5L valuationsDiscrete Choice Experiments (DCE) for health state valuation, involving duration as an attribute (DCEduration) or death as an alternative (DCEdeath), are increasingly promoted. Especially DCEdurationis regardedas a promising technique. Nevertheless, DCEdurationfor health state valuation is still in its infancy and results that are obtained warrant further investigation. A consequential result of the DCEdurationvaluation technique compared to Time Trade Off (TTO) is that DCEdurationcurrently classifies much more states as worse than dead (WTD) and produces lower average health state values. Given the important consequences for cost-effectiveness analyses, and the broad acceptance of health state values derived using TTO, the question inevitably arises: what is the explanation for this difference? It may be clear that remaining problems in WTD estimation using TTO contribute to the problems. However, they are probably not the sole explanation. We hypothesise that framing effects also contribute to the differences, because ‘losses and gains’ and ‘immediate death’ are defined differently in these methods. Against this background we propose to 1.Examine how alternative editing or framing of the questions affects outcomes in TTO, DCEdurationand DCEdeath, and in turn2.Explore to what extent framing differences provide an explanation for across-method differencesWe hypothesize that the differences between TTO and DCE may at least in part be reconciled by examining and manipulating the framings. If this hypothesis is correct, this study will contribute significantly to the acceptance of DCE as method for health state valuation in future protocols.Elly StolkValuation172300Ongoing20142015
2013220Feasibility of the use of EQ-5D in quantitative benefit-risk assessmentAim: Market authorization for pharmaceuticals will only be granted to products that demonstrate sufficient levels of quality, safety, and efficacy for the indicated patient population. Usually, limited information regarding the safety and efficacy of a new pharmaceutical is available when the decision whether to grant the product a market authorization license needs to be made by regulatory authorities. A main issue in quantitative benefit-risk assessment is the weighing of benefits and risks on a single scale.1 Data regarding benefits consist of clinical trial endpoints and are usually disease-specific whereas ‘risks’ consist of the adverse drug reactions (ADRs) that are reported during a clinical study. In the assessment of a pharmaceutical’s benefit-risk profile, a decision maker will have to weigh a positive outcome (such as an increase in the progression-free survival) against the occurrence of a range of ADRs in a proportion of patients. This project intends to determine whether EQ-5D data is capable of capturing the negative effect of ADRs on a patient’s quality of life. If EQ-5D indeed is able to measure such risks, this could be an important step towards the use of EQ-5D data in quantitative benefit-risk assessment. General Methods: Before a benefit-risk assessment using EQ-5D data can be performed, the feasibility and validity of this approach needs to be assessed. This study will examine the sensitivity of the EQ-5D in the measurement of the impact of experienced ADRs on health-related quality of life. If the results of our study would confirm the sensitivity of the EQ-5D in measuring health effects of ADRs, this could make the EQ-5D applicable for quantitative benefit-risk assessment. Performing a full benefit-risk assessment is beyond the scope of this study.Mark OppeOthers33000Ongoing20132014
20190250Psychometric testing of proxy EQ-5D-5L-Y in caregivers of paediatric patients with idiopathic scoliosisProxyEQ-5D-3Lquestionnaire for youth (EQ-5D-3L-Y), which has three severity levels,is a health utility measure for caregivers to rate the child/adolescent’s health-related quality of life (HRQOL). An extended version of the questionnaire, namely EQ-5D-5L-Y, which has 5 severity levels, could be more sensitive and discriminative in adult version. Both questionnaires rated by patients’ proxy are alternative measurement for child’s and adolescent’s HRQOL. However, no comparison betweenthe proxy version of EQ-5D-3L-Y and EQ-5D-5L-Y has been reported so far.This studyaims to assess the feasibility of the proxy EQ-5D-5L-Y from caregivers and compare it with proxy EQ-5D-3L-Y in Hong Kong children and adolescent. Redistribution properties, relative efficiency and informativity of the three questionnaires will also be compared. One hundred consecutive patients will be recruited from the paediatrics and spine clinics in a public Hospital in Hong Kong. The inclusion criteria for the patients are: 1) Having a diagnosis of juvenile (JIS) or adolescent idiopathic scoliosis (AIS), 2) aged 8-17 years, and 3) having the ability to read and converse in Chinese. They will be interviewed and asked to complete the EQ-5D-3L-Y and EQ-5D-5L-Y and provide their socio-demographics and clinical characteristics, while their caregivers will be interviewed with proxy version of EQ-5D-3L-Y and EQ-5D-5L-Y. Responsiveness will be assessed by asking patients and their caregivers to complete the questionnaires at 3- 9 months again after baseline survey.Carlos WongYouth7300Completed2019
2016130SOME NEW STRATEGIES FOR ELICITING AND MODELING UTILITY VALUES OF MULTI-ATTRIBUTE HEALTH STATESThis project (i.e. workshop) has no direct scientific implications for the EuroQol Group. However,several future research topics emerged during the preparation of the workshop. Future research on the TTO part of the EQ-VT program was discussed, although no specific topics were identified. Basically, we think that more research is needed in this direction since there were many contradicting findings from the MAT project. TTO as the main valuation technique of the EuroQol Group deserves further scientific investigation to improve its validity and friendliness to both interviewers and interviewees. Another research topic is interval regression for modelling TTO data collected by EQ-VT. The advantage of the new modelling strategy is attractive but has not been fully confirmed. The team has discussed future steps to exploring those research topics. There are several strategic implications. Frist, the workshop disseminated the state of the art research conducted within the EuroQol Group to the greater research community. Specifically, we demonstrated how the EQ-VT program works and explained how the EQ-5D-5L value sets can be developed. Second, we had useful interactions with EQ-5D users and researchers on other multi-attribute preference-based instruments during and after the workshop. The interactions were efficient in clarifying questions regarding the design of the EQ-VT and explaining the challenges in generating EQ-5D value sets. Has the workshop reached its goals? Yes, we think so. Our goals were to collect feedback on our research work conducted within the EuroQol Group, to disseminate our latest research findings to the greater research community, and to enhance the leadership status of the EuroQol Group in health valuation research. While it is difficult to assess how successful we were in achieving the last goal, we achieved the first two goals through two presentations and a discussion session with workshop attendees.Nan LuoValuation10115Completed2016
2015100The distribution of the EQ-5D-5L Index in patient populationsBackground:EQ-5D data are often summarised by anEQ-5D index, whosedistribution foritsoriginal version, the EQ-5D-3L,often shows in patient populations two distinct groups, arising fromboth the distribution of ill health and how the index is constructed (Parkin et al., 2016). Todate, there is little evidence about the distribution of the EQ-5D-5L index.Aims:The aimsof this study are:to explore whether or not the EQ-5D-5L index distribution also demonstrates clustering; to test the extent to which clustering of EQ-5D-5L profile data drivesany observed clustering of the EQ-5D-5L index,and the extent to which clusters result fromthe value sets used to create the index; and to discussthe implications of our results for statistical analysis of EQ-5D-5L index data.Data:Data from Cambridgeshire Community Services NHS’s electronic patient records data warehousewere analysed.EQ-5D-5L profiles before treatmentwere obtained for30,284 patients across three patient groups: community rehabilitation services (N=6,919); musculoskeletal therapy services (N=19,999); and nursing services (N=3,366).Methods:The EQ-5D-5L indexis calculated using both a ‘mapped’ (crosswalk) value set (MVS) and the English value set (EVS).We examinedthe distribution of 1,730 of the 3,125 profiles described by the EQ-5D-5L to check forclustering of the EQ-5D-5L index. The k-meanscluster method and the Calinski–Harabaszpseudo-F indexstopping rule were used to search for the clusters in the index. We examined the impact on the results of using different initial values in the clustering analysis. Results:Clustering within the EQ-5D-5L index distribution is suggested by both clustering methods, forthe three patient groupsand all patients together.For the all patients’data, we foundtwo robust clusters for the MVS-based index, compared to three robust clusters for the EVS-based index. The EQ-5D-5L profile dataalonedonotobviously drive the index clusters.Conclusion:The results highlight the importance of undertakingcareful exploratory data analysisfor health related quality of life measures such as the EQ-5D,to ensure that statistical testingtakesaccount ofclusteringand other features of the data distribution.Key words:EQ-5D; EQ-5D-5L Index; EQ-5D-5L profile; EQ-5D-5L value sets; Clustering Analysis; Patient Reported Outcomes (PROs); NHS PROMs.Nancy DevlinValuation9750Completed2015
2014200Development of a proxy English Health-Related Quality of Life (HRQoL) instrument for children under six years of age, derived from the EQ-5D-Y: Part 1Background and Aim: Although the EQ-5D-Y was developed for self-report by children over the age of seven years, there is no instrument in the EuroQoL stable developed and validated specifically for younger children and infants. In order to investigate the need for the development of such an instrument, a systematic review of Health Related Quality of Life (HRQoL) measures for children under seven years of age was undertaken. The COSMIN (1)checklist for reviewing outcome measures was used as a framework.Study Design:The search strategy was based on published literature and expert knowledge. A data abstraction form was developed, following COSMIN (2). Information on bibliographic details, instrument development, completion by self/proxy, domains, number of items, response options, frame of reference, recall period, scoring and psychometric results pertaining to validation, reliability and practicality was recorded. Results:Fifteen generic HRQoL measures were identified. The development procedure for most of the instruments was based on the existing literature and expert opinion and all authors claimed to have taken age-appropriateness into account. Many of the measures were extended for proxy use to five years of age but few were developed specifically for children under the age of seven years. Domains were similar across measures and in accord with WHO’s definition of health and components of the HRQoL namely: physical; emotional; social and cognitive. Content validity was judged from the developmental process of the measure. Face validity was reported based on results of the pilot or field testing during the development process. Despite therequirements of FDA guidelines for proxy measures, observable characteristics of the domains were poorly defined and no measure was based purely on observable behaviour. The majority of measures rely on either domain scoring and/or a single summary score.The scoring algorithms for the measures are generally poorly defined. The criterion validity of most of the instruments was established through comparison to a previously developed and validated HRQoL measure. The divergent validity was generally examinedby confirming differences between known-groups. The reliability as measured by internal consistency was reported α≥0.70 for only eight of the measures. Conclusion:There is a need to develop a comprehensive generic, preference-based HRQoL measure for children under the age of seven years as none of the measures reviewed satisfy all of the criteria. The specific aspects that were deficient included: a) The development procedure did not include focus groups of parents of children in the age group to establish the age appropriate behaviour which would represent the HRQoL construct and b) observable domains were not explicitly included and observable behaviour that would inform proxy report was not defined; c) the scoring systems were not necessarily psychometrically sound and the scoring algorithms, where developed, where opaque. There is a lack of preference based measures and several of the instruments were lengthy and valuation sets would be difficult to develop. There may be a need to develop more thanone instrument to measure HRQoL in different age categories for children under the age of seven years.Jennifer JelsmaYouth15000Completed2014
20180740What contributes to variations in self-reported health for the general population and for ten condition-specific patient groups in England? - An empirical analysis using repeated cross-sectional general practice data with 2.9 million patient records.Background Significant variations in the UK’s population health have been widely observed. Little is known about individual and health service factors associated with these variations in health or about how much of the unexplained variation is at patient and health service levels. Aim This report explores which factors explain the variation in patients’ self-reported health, measured by the EQ-5D-5L instrument, for the English general population and for ten patient groups with self-reported chronic conditions. We quantify how much of the variation that observed factors could not explain can be attributed to patients, general practice, and Clinical Commissioning Groups (CCGs) characteristics and whether they differ across the general population and ten patient groups. Methods We linked data from three data sources - the NHS’s General Practitioner (GP) Patient Survey, Quality and Outcomes Framework, and GP workforce databases, over five years between 2012/13 to 2016/17. The resulting dataset covers almost all GP practices in England, with a nationally representative sample for the general population (N=2,913,294). Additionally, we generated ten subsamples for each chronic condition. We use ordinary least squares, mixed-effects, practice fixed- and random-effects models to identify factors associated with EQ-5D-5L index variation for the general practice population and for groups of patients with one of ten specific conditions. We also calculate ICCs using variances estimated from mixed-effects models to decompose the unexplained variation in the EQ-5D-5L index to patient, GP, and CCG levels. We explored the impact of missing values on mixed-effects models using multiple imputation. Results Some explanatories (such as multimorbidity, gender, deprivation, employment status, smoking status, satisfaction with access to the practice and good overall experience with the practice) had similar and plausibly signed effects for all 11 sets of individuals. But there were differences for those with different conditions. For example, those reporting their ethnicity as Black had lower EQ-5D-5L index if they had Alzheimer/dementia or angina but higher EQ-5D-5L index if they have asthma, diabetes, hypertension, or mental health problems. Better clinical quality for specific conditions was associated with better patient reported health only for three conditions (asthma, diabetes, epilepsy). For the general practice population, the unexplained variation in the EQ-5D-5L index was almost entirely at patient level (99.57%) with very small general practice (0.29%) and CCG (0.14%) contributions. Ten condition-specific patient groups shared similar findings for the decomposition of unexplained variation as that from the general population. Conclusion The findings from our study identifies factors that explain the variation in health for the general population and patient groups with chronic conditions in England. Policymakers could develop interventions to reduce or eliminate the effects for some of those factors in our list, such as smoking and deprivation status of individuals, or patient satisfaction with access to their general practice, to improve health and to reduce variation in health.Yan FengPopulations and Health Systems46800Completed20192020
2016180Exploring the possibilities for developing a EuroQol instrument for use in very young children: a workshop on feasibility, relevant issues, and potential methodologyLiterature is clear that self-report health in young children is challenging• Therefore, proxy (cross-informant) responses are neededEvidence about the agreement between children and informant information vary• But observational measures might be more accurate that proxy (subjective) measures in young childrenWe should think about developing a conceptual framework to use when measuring HRQoL in young children• Should reflect changes over time (e.g. 0-2, 2-3, 4-5 years)Jennifer JelsmaYouth14821Completed2016
2015300QALY-balanced DCE designs for health state evaluationsBackground:Discrete choice experiments(DCE) are increasingly used for health state valuation. However, the valuesderived from initial DCE studies varied widely. We hypothesize that these findings indicate presence of unknown sources of bias that must be recognized and minimized. Against this background, we studied whether values derived from a DCE are sensitive to how well the design spansthe QALY range. Methods:We constructed an experiment involving three variants of DCE tasks for health state valuation: standard DCE, DCE-death, and DCE-duration. For each type of DCE, an experimental design was generated under two different conditions, enabling a comparison of health state values derived from standard Bayesian efficient DCE designs with values derived from ‘QALY balanced’ designs that control for coverage of the severity range in health state selection. About 3,000 respondents participated in the study and were randomly assigned toone of the 6 study arms. Results:Imposing the QALY-balance restriction hada large effect on health states sampled for the DCE-durationapproach. The standard efficient design returned a skewed distribution of selected health statesand this introduced bias. The choice probability of bad health states was underestimated, and time tradeoffs to avoid bad states were overestimated, resulting in too low values.Imposing the same restriction had limited effect in the DCE-death approach and standard DCE. Conclusion:Variation in DCE derived values can be partially explained by differences in how well selected health statesspanned the severity range. Imposing ‘QALY balance’ on DCE duration designs is a validity requirement.Mark OppeValuation49250Completed2015
2015180Exploring non-iterative TTO (ENITTO)Background – Time Trade-Off (TTO) usually relies on “iteration”, which is susceptible to bias. Discrete Choice Experiment with duration (or DCETTO) is free of such bias, but respondents find this cognitively more challenging. This paper explores non-iterative TTO (NITTO). In NITTO respondents see a series of independent pairwise choices without iteration (similar to DCETTO), but one of the two scenarios always involves full health for a shorter duration (similar to TTO). Methods – We examine the performance of three different types of NITTO relative to DCETTO (i.e. four “Types” of choice tasks). Each Type is presented in two ways: (a) verbally tabulated (as in a DCE); and (b) with visual aids (as in a TTO). Thus, there are eight survey variants, each with 12 experimental choice tasks and a 13th task with a logically determined answer. Data on the 12 experimental choices obtained from an online survey of 6,618 respondents are modelled by variant using conditional logistic regressions. Results – The overall results suggest that NITTO seems feasible. Fewer respondents found the tasks difficult compared to DCETTO; the great majority of respondents “pass” the logical consistency test; and data can be modelled to produce interpretable coefficients. However, some relatively mild states appear to have almost implausibly low values, and the range of predicted values between state 22222 and 55555 seems narrow. A high proportion of respondents are non-trading, always choosing the shorter survival in full health over the longer survival in worse health.Aki TsuchiyaValuation87870Completed2015
20180510Valuing the EQ-5D-Y-3L in Germany, Spain and SloveniaThere is an increasing interest in understanding thevariations in health state preferences derived fromadolescents or from adults with the perspective of a child.This project will(1) explore those differences in health preferences using the EQ-5D-Y-3Land (2) develop a value set for the EQ-5D-Y-3L for Germany, Spain and Slovenia.Health state preferences from adolescentsvaluing health states for themselvesand from adultsvaluingchildren’shealth stateswill be assessed to compare thoseand to fill the gap of the non-existence of value sets for the EQ-5D-Y-3L.700 adolescents and 1000adults will be includedin Slovenia, Spain and Germany, each. In thefirst sub-study, an online survey will be used to obtain health state preferencesby discrete choice experiments (DCE)The proposal Version 26JAN2017Page 2follows the EuroQol Group’s valuation protocol for the EQ-5D-Y-3L. To anchor the DCE latent scale values derived for the EQ-5D-Y-3Lhealthstates, the data will be combined with those of a time trade-off (TTO)survey.TheTTO values will be used to rescale the latent scale valuesfrom the DCE study. Therefore, a second sub-study will be conducted using computer assisted personal interviews (CAPI) to ask adult respondents to value health states using TTO tasks. This decision of using the TTO method to anchor the DCE values is in accordance with the agreed valuation protocol for the EQ-5D-Y-3L. Therefore, a sample of 200 respondents will be recruitedin each countryWolfgang GreinerValuation, Youth95840Completed2019
2013110The feasibility and usefulness of using the EQ-5D-Y as a routine measure of outcome in a facility for children with chronic illness1.Background:The development of the EQ-5D-Y, an outcome measure of HRQoL in children, was undertakenby an international task team and published in 2010 (1). Following the feasibility, reliability and validity studies,undertaken in the main with typically developing children, a recommendation was made to apply the measure in a longitudinal, clinical study to determine its responsiveness to change over a period of time and to monitor effects of treatment (2). .2Aim:The overall aim of the study was thereforeto investigate thepsychometric properties and feasibilityand usefulness of the EQ-5D-Ywhen used to monitor the HRQoL of children within different institutional contexts. This included monitoring the changes in HRQoL over time,assessingthe feasibility of the process of data collection as well as examining how useful the collected data were to the health providers.3Methods:Ethical approval was gained from the Human Research Ethics Committee of the University of Cape Town. A longitudinal, analytical descriptive study design wasused within four different contexts. Children were recruited from a mainstream school(MS); aconvalescent homefor children with chronic diseases(Chronic Centre, CC); special school (SS) for childrenwith physical disabilities and an acute care (AC) hospital. The EQ-5D-Y was the primary outcome measure. The PedsQL(1)was used as a parallel measure to assess HRQoL andthe WeeFIM(2)to assess function.Appropriatelanguage versions of each instrument were used (i.e. English, Afrikaans and isiXhosa, translated under the direction of the researcher if no version available). Repeated measures were taken at three month intervals, or, in the case of acutely ill children, at admission and at discharge. A self-designed questionnaire was used during interviews with the therapists who were treating the childrento assess feasibility and usefulness.4Results:4.1ParticipantsA total of 224 children were recruited, approximately 130 from the Institutions and 100 from the MS (See Figure 1). The mean age of the children at the various institutions was 10.5 years (Standard Deviation (SD) =1.45), with a minimum age of 7.0 years and a maximum of 13.8 yearsand therewas no significant difference in the mean ages between the institutions (p=.379).Health conditions ranged from cerebral palsy(18), to spina bifida (10)spina bifida, HIV(8) and diiabetes, muscle diseases, neoplasms, appendicitis and joint injuries.Jennifer JelsmaYouth25932Completed2013
2016330Understanding the relationship between clinical quality of primary care and patient self-reported health on the EQ-5D in EnglandAbstract Background:There is very limitedempirical evidence in the English National Health Service (NHS)–or indeed from any country –on how the clinical quality ofprimary care impactson patients’ self-reported health. This study sets out to address this gap in the literature, using EQ-5D data. Aims:The primary aim of this paper is to explore what policy-relevant general practice factors, in particular the clinical quality of primary care, affect patients’ self-reported healthin England as measured by the EQ-5D-5L instrument. Data:We usedata from the GP Patient Survey (GPPS) and the national Quality and Outcomes Framework (QOF) datasets in England between financial years 2012/13 and 2015/16. The two datasets are linked by general practice and financial year. The linked dataset covers most general practices in England with around 7,500 to 8,000 practices included in each year.In the GPPS dataset, EQ-5D-5L data are recorded as the number(and %) of patients,at general practice level,ineach level and dimension. From this, we derive two measures of health for each general practice: (1) the practice-level EQ-5D-5L level sum scoreand (2) the practice level EQ-5D-5L indices.To measure the clinical quality of general practices based on the QOF dataset, we calculate two types ofmeasures: the practice level population achievement rate and the proportion of clinical QOF points achieved out of the total achievable clinical QOF points. Methods:We estimate linear panel data multiple regression models with controls for patient and practice characteristics by Ordinary Least Squares (OLS)method and with random and fixed practice effects. In sensitivity analyses, we explore the effects of non-response in the GPPS, lagged effects of clinical quality, and of including non-clinical QOF indicators inthemeasure of clinicalquality. Results:Clinical quality and patients’self-reported health exhibit considerable variation between general practices but not within practices over time. Theabsolute unconditional cross-section correlations between the clinical quality and EQ-5D measures in 2015/16 are between 0.0626and 0.0748. TheOLS and practice random effect multiple regression models suggest that clinical quality has a statisticallysignificant positive effect on improving patients’self-reported healthmeasured by the EQ-5D-5L instrument. The results are consistent whenapplying two proxies for clinical quality measurements and two proxies for patients’self-reported healthmeasurements. However, once we allow for unobserved time invariant practice characteristics with fixed effects regressions the effects are not significant. Discussion:This study is one of the first attempts to use national GPPS and QOF datasets in England to explore the relationship between clinical quality of primary care and patients’ self-reported health as measured by the EQ-5D. The findings suggest that improvements in general practice clinical quality as measured by QOF indicators are not associated with improvements in patient self-reported health as measured by EQ-5D.This may be because there is a lag in the effect of clinical quality on health or because the QOF indicators are a mix of process measures and intermediate outcome measures.Yan FengPopulations and Health Systems35246Completed2016
117-RAConceptual challenges in the valuation of health in children and adolescentsConceptually, the valuation of health states in children raises several questions, including: 1. Source: Whose preferences should be elicited? (adults, children, both?) 2. Perspective: Whose health should be valued? (self today, self as a child, own child, hypothetical child, etc.) If a child health perspective is taken, how should the age of the child be described, if at all? 3. Methods: How should be preferences be elicited and anchored? (TTO, DCE, etc.) Are there conceptual reasons why health state selection methods in a child health valuation study should differ from those for an adult health valuation study? This proposed project aims to study key conceptual issues in the valuation of health in children. This will be done via a combination of desk research, meetings to exchange/discuss ideas, and an expert workshop. It will contribute to the current debate and support the EuroQol Group in interpreting (and potentially expanding) its research agenda in this area. While several EuroQol-funded empirical studies in this area are likely to show how the choice of approach influences the outcomes of health state valuation, they are unlikely to provide a clear indication about which approach to prefer. The question of which approach to choose would then shift from an empirical one to a normative one, and our proposed research will provide insights into how such a question can be answered. The project is intended to be a collaboration between early career researchers who are actively working on EQ-5D-Y and conceptual issues in health state valuation research.Koonal ShahValuation, Youth44980Completed2020
2013240Indonesian valuation and validation study of the EQ-5D-5L and its application in families living in unhealthy circumstances and in breast cancer patientsIndonesia is with more than 238 million people the world’s fourth’s most populous country. Given the large gap between its under-resourced health care system and its strong and persistent economic growth (6% per year), a rise in new pharmaceutical products,and illness prevention projects with corresponding cost-effectiveness analyses are to be expected in the coming years. The introduction and use of the validated EQ-5D-5L can perform a pivotal role here. Therefore, the aimsof this study are:1. To translate the EQ-5D-5L in Bahasa Indonesia (BI), to obtain age and gender norms, and to derive a preference valuation set for the translated EQ-5D-5L health states from the Indonesiangeneral public. 2. To validate the EQ-5D-5L by its application inan impoverished population group in Jakarta. Methods: Ad 1. Translationof the 5 levels versionof the EQ-5D, the EQ-VT and the guide book will be carried out according to guidelines of the EuroQol Group Executive office. Next, arandom sample (N=1,000) representative forthe Indonesianpopulationwill be asked to value a subset of EQ-5D-5L health states usingthe EuroQol Valuation Technology (EQ-VT). Regression techniques will be used to estimate anIndonesianvalue set of all possible 3,125 EQ-5D-5L health states from the empirical validated subsample of health states. Test-retest reliability of the valuationswill be determined in a subgroup (n=200)and there we will examine the relation between the quality of the interviews (adherence to the protocol) and the valuation task (EQ-VT).Age and gender (distribution) norms of the EQ-5D-5L descriptive system and EQ-VAS for the general population will be obtained as well. Ad 2: Validation: The construct validity of the EQ-5D-5L will be determined in boththe general population and in 250 families, living in unhealthy circumstances in the urban area of Jakarta and/or Bandung. The construct validity will be based on the comparison of the EQ-5D-5L with a number of additional quality of life measures (among which the WHOQOL-Bref) and a comparison between the healthy population and the impoverished sample.In addition, the EQ-5D-5L will be completed by about 200 Indonesian patients with breast cancer in Bandungwho participate in a Dutch Cancer Foundation funded project that ischaired by two of the principleinvestigators of the presentEUROQOLproposal(SS and JP). The patients’scores will be stratified according to cancer stage (I-IV) and compared with the scores of the other samples. Reports and presentations will be provided for the Indonesian stakeholders (government officials, physicians and water sanitation companies)and for the international scientific community.Jan BusschbachValuation71100Completed2014
2014110Assessing the health of the general population in England: how do the EQ-5D 3L and 5L versions compare?not available Valuation8250Completed2014
2013140A Japanese valuation study for the EQ-5D-5LThis survey was conducted as a s upplemental analy sis of "A Japanese valuati on study for the EQ-5D-5L (project number 2013140)", to assess the impact of implementing汁anking"process in to EQ-VT software . Then, in dependent study proposal for this research was not developed. For information purpose, study proposal of project number 2013140 was attached as appendix 1, "Appendix I original Research proposal.pdf'.Shunya IkedaValuation73500Completed2013
2014010Valuing EQ-5D-5L health states using EQ-VT: Does the Life A health description and/or the ordering of dimensions matter?Objectives: Studies to produce utility values for the EQ-5D-5L instrument are ongoing internationally. These include the valuation of 10 EQ-5D-5L health states using thecompositetime trade-off (TTO) method. In some of the studies carried out to date, relatively low mean TTO values formild health states have been observed. It is hypothesised that this is because the health states under evaluation are being compared to “full health”, whereas in previous studies they were compared to 11111 (the “best” health state in the descriptive system). The objective is to assess differences in TTO valuations using two different comparators (full health and 11111).Methods: Preferences for EQ-5D-5L health states were elicited from a broadly representative sample of the UK general public. TTO data were collected using computer-assisted personal interviews, carried out in respondents’ homes. Respondents were randomly allocated to one of two arms: in arm 1 the comparator health state was full health; in arm 2 the comparator health state was 11111. After completing 10 TTO valuations, respondents were asked follow-up questions which sought to examine their interpretations of the term “full health”.Results: Interviews with 443 respondents were completed between May and October 2014. We do not observe a statistically significant difference in mean values for the mild health states across arms. The two arms produced data of similar quality. We observe evidence of interviewer effects.Health state 11111 was given a value of 1 by 98.2% of the respondents who valued it; yet the majority of respondents who self-reported as being in 11111 did not consider themselves to be in “best imaginable health”. Vision and energywere mentioned by respondents as examples of important aspects of health not covered by the existing dimensions.Conclusions:EQ-5D-5Lvalues elicited using the composite TTO approach are not greatly affected by whether full health or 11111 is used as the comparator health state.Brendan MulhernValuation89950Completed2014
139-RAPreference heterogeneity in health valuation: Dutch BWS dataBackground Respondents in a health valuation study may have different sources of error (i.e., heteroskedasticity), tastes (differences in the relative effects of each attribute level), and scales (differences in the absolute effects of all attributes). Although prior studies have compared values by preference-elicitation tasks (e.g., paired comparison [PC] and best-worst scaling case 2 [BWS]), no study has yet controlled for heteroskedasticity and heterogeneity (taste and scale) simultaneously in health valuation. Methods Preferences on EQ-5D-5L profiles were elicited from a random sample of 380 adults from the general population of the Netherlands, using 24 PC and 25 BWS case 2 tasks. To control for heteroskedasticity and heterogeneity (taste and scale) simultaneously, we estimated Dutch EQ-5D-5L values using conditional, heteroskedastic, and scale-adjusted latent class (SALC) logit models by maximum likelihood. Results After controlling for heteroskedasticity, the PC and BWS values were highly correlated (Pearson's correlation: 0.9167, CI: 0.9109-0.9222) and largely agreed (Lin's concordance: 0.7658, CI: 0.7542-0.7769) on a pits scale. In terms of preference heterogeneity, some respondents (mostly young men) failed to account for any of the EQ-5D-5L attributes (i.e., garbage class), and others had a lower scale (59%; p-value: 0.123). Overall, the SALC model produced a consistent Dutch EQ-5D-5L value set on a pits scale, like the original study (Pearson's correlation:0.7295; Lin's concordance: 0.6904). Conclusions This paper shows the merits of simultaneously controlling for heteroskedasticity and heterogeneity in health valuation. In this case, the SALC model dispensed with a garbage class automatically and adjusted the scale for those who failed the PC dominant task. Future analysis may include more behavioral variables to better control heteroskedasticity and heterogeneity in health valuation.Suzana KarimValuation20100Completed2020
20190070Making both ends neat: exploring the effects of modifying the TTO on non trading and all in tradingThe corner answers in time trade-off, TTO, non-trading, NT; and assigning utility of -1 —all-in-trading, AIT) occur frequently and impact the resulting modelsand value sets. NT resultsin poor discriminatory power for mild states. AIT means the average utility estimates are biased upwards(censoring may help but requires parametric assumptions).We aim to test if modifying the TTO protocol lowersthe frequency of NT/AIT (if yes, it would shed light on the reasons behind). Specifically, we hypothesise that NT may stem froma coarse-grained list of answers in TTO(half-a-year intervals) or psychological biases(e.g., reluctance to give up years because of the loss aversion as in Prospect Theory). AIT may result fromutilityfor severe statesbeing discontinuous at 0 (i.e.,the sole fact of having to endure 55555, for whatever short 𝑇, reduces the utility),analogously to present bias or certainty effectobserved in behavioural economics.We will use four arms in online TTO data collection: a control arm (resemblingcTTO) and three experimental arms (ca. 200 respondents in each arm). In the experimental arms,we will ask respondents if finer granularity would get them to trade or add years in full health to a diseased state or use diseased state in both compared profiles in the worse-than-dead part. We will compare the proportion of NT/AIT answers in groups of mild/severe states between the arms and also see if modifications improve the association(counterintuitive in the current protocol)between state severity and utilityin worse-than-dead part.Michał JakubczykValuation83200Completed2019
2014150Discrete choice modeling from a different angleIntroduction: Preference-based measurement methods are frequently used to obtain values expressing the quality of health-state descriptions. A crucial assumption in these methods is that respondents pay equal attention to all information components presented in the response task. So far, there is no solid evidence that respondents are fulfilling this condition or are using shortcuts. The aim of our study is to explore the attendance to various information cues presented in the discrete choice response tasks. Methods: Eye tracking was used to study the eye movements and fixations on specific information areas. This was done for seven discrete choice response tasks comprising health-state descriptions. A sample of 10 respondents participated in the study. Videos of their eye movements were recorded and are presented here graphically. Statistics were computed for length of fixation and number of fixations, so differences in attendance were demonstrated for particular attributes in the tasks. Results: All respondents completed the survey. Respondents were fixating on the left-sided health-state descriptions slightly longer than on the right-sided health-state descriptions. Fatigue was not observed, as the time spent did not decrease in the final response tasks. We noted that the time spent on the tasksdepended on the difficulty of the task and the amount of information presented. Conclusion: Eye tracking proved to be a feasible method to study the process of paying attention and fixating on health-state descriptions in the discrete choice response tasks. Eye tracking facilitates the investigation of whether respondents fully understand the information in health descriptions or whether they ignore particular elements.Paul KrabbeValuation250000Completed2014
2015290The Effect of Timing, Duration, and Lifespan on Choicenot availableBenjamin CraigValuation68439Completed2015
115-RAFacing death to understand the construct validity of the (c)TTO method: a conceptual approachThe construct validity of health utilities, e.g. the (c)TTO, is an under researched area. Existing determinants of (c)TTO are ill-understood. The search for (c)TTO-determinants and their interpretation has not been guided by theory. This proposal focusses on the role of death. In the search for concepts or determinants related to the (c)TTO, the word "dead" has been hopelessly missing. This is an omission as preferences for death as well as trading life are central to the (c)TTO, and this will induce death thoughts. Our proposal focusses on the thoughts of dead induced by the (c)TTO. Terror Management Theory (TMT) is used to understand psychological responses to thoughts of dead. TMT starts from death related anxiety, that occurs after death thoughts. Defense mechanisms, such as bolstering cultural values, are called into action to reduce this anxiety. As both TMT and the (c)TTO involve thoughts of death, we aim to study whether TMT can be used to improve our understanding of (c)TTO determinants. Relevance 1) This proposal is about the construct validity of the (c)TTO. 2) TMT enriches the construct validity of the (c)TTO by embedding it in the concepts of TMT. 3) Connecting (c)TTO and TMT guides the search for new (c)TTO-determinants, as well as interpreting previously found (c)TTO-determinants.Peep StalmeierValuation, Youth21120Completed2020
2016470US valuation study of the EQ-5D-5LObjective: To derive a value set for the EQ-5D-5L from the preferences of the United States (US)adult general population based on the standardized protocol developed by the EuroQol Group. Methods: Respondents from the US adult population were quota sampled based on age, gender, ethnicity, and race. Trained interviewers guided participants in completing composite time trade-off (cTTO) and discrete choice experiment (DCE) tasks using the EuroQol Valuation Technology software and routine quality control measures. Data were modeled using random effects linear regression and Tobit for cTTO data, conditional logit for DCE data, and a hybrid model that combined cTTO and DCE data. Model performance was compared based on statistical significance and logical ordering of coefficients, goodness-of-fit, and theoretical considerations.Results: Of 1,134 respondents, 1,102 provided valid responses on the basis of the quality control criteria and interviewer judgement. The distribution of age, sex, race, ethnicity, education, chronic conditions, and general health status were similar to the 2015 US Census. The hybrid model was selected as the preferred model to generate the value set, which provided logically ordered, statistically significant coefficients for each of the dimensional levels. Values ranged from -0.699 (55555) to 1 (11111), with 23.5% of all predicted health states scoring < 0 (i.e., worse-than-dead). Conclusions: We applied the official, consensus-based international EuroQol valuation protocol in a sample representative of the US population to derive a TTO/DCE-based value set for the EQ-5D-5L that can be used for economic evaluations and decision-making in US health systems. As we utilized an internationally established protocol, the US-based value set can also facilitate cross-country comparisons.Simon PickardValuation283000Completed2017
2014070EQVT iteration experimentTTO data from recent EQVT valuation studies displays troubling patterns, including spikes at 1, .5, 0, -.5, and -1; a lack of non-integer year responses; high levels of logically inconsistent responses, and possible censoring of values below .5 or 0 for a number of respondents. The (ping-pong) routing, has been suggested as part of the reason for the observed spikes. This proposal is submitted under the umbrella of the ongoing effort to improve the EQ-VT system, and aims to determine if two approaches to altering the routing procedure may improve the data quality, particularly in terms of reduced spikes and reduced level of inconsistent responses The alternative procedures are based on the observation that the spikes of elicited values correspond to respondents stating indifference very early in the current TTO routing. At least two distinct effects may contribute to this: satisficing (i.e. respondents stating indifference early in order to shorten the task), anda form of sequential anchoring (inadequate adjustment from the values presented in the TTO task). The two modifications we suggest testing are1) Randomized starting point and routing:Reducing the danger of routing-specific bias through randomizing respondents to a range of different routing procedures. If anchoring and satisficing is at play in all conceivable routing procedures, introducing a variety of routes will distribute the overall bias, so that the common variance across different routing procedures should be a less biased measure of net preferences.As a benefit, varied routing may enable analyses to determine to what extent values are censored at 0.2) Non-stopping TTO:Satisficing is a problem in TTO because the procedure ends as soon as respondents state indifference, creating an incentive for stating indifference at values corresponding to early choices. The second modification attempts to remove this incentive through forcing respondents to perform the same amount of work for all possible elicited values. This is achieved by not stoppingthe preference elicitation once a point of indifference is identified. Deliverables: (1) report to the Exec/VMWG with description findings and suggestions for how to handle routing in the EQVT, (2) Scientific paper on routing bias and satisficing effects in TTO valuation, (3) paper on a non-stopping TTO, and (4) paper on censoring.Kim RandValuation159450Completed2014
20180030R1Measuring Quality of Life in the general population in Romania: an EQ5D-5L value set and population norms for Romania (QoLROClear, transparent and consistent criteria for decision-makingand healthcare resource allocation have recently become the focus of many health system strategic plans across the globe. In the last few years, in Romania, several factors have converged to create a more favourable environment for the development of health technology assessment (HTA). However, some challenges still remain. This research aims to simultaneously elicit value setsand develop population norms for both EQ-5D-3L and 5L in the same sample of Romanian respondents. Amulti-stage stratified sampling procedure will be used to recruit 2,016people representative of the Romanian general population in terms of historical region, degree of urbanisation, age and sex. Data will be collected through face-to-face interviews conducted in respondents’ homes. Preferences for EQ5D3L and 5L health states will be elicited from the same respondents using composite time trade-off(cTTO)techniques. Additionally,discrete choice experiments (DCE) will be used to elicit preferences for the 5L version. Linear regression models will be used to estimate the 3L value set based on the cTTO data. A hybrid model using the data from both the cTTO and DCE will be used to estimate the 5L value set. This project will enable local expertise development, thus addressing one of the challenges highlighted as potentially hindering theeffective implementation of HTA in Romania. Additionally, it will benefit the EuroQoL Foundation by creating a new demand for EQ-5D as the instrument will be the only utility measure in Romania for which a local value setand population normswill be available.Elena OlariuValuation4996Completed2018
2015200Feasibility and application of the EQ-5D in elderlyDue to the aging population and changes in the morbidity structure, health-related quality of life (HRQoL) measurement in the elderly gets more relevant. However, special characteristics of the elderly population, such as impaired cognitive functioning or visual impairments, seem to make an accurate measurement of HRQoL challenging. The EQ-5D is a widely used generic preference-based instrument, which is psychometrically sound with the general population, but it may not be suitable for use in elderly. Therefore, we need to examine whether the EQ-5D is a feasible measure applicable for the use in the elderly population. Hence, this study aims to systematically identify and analyse the available literature which examined feasibility properties of the EQ-5D specifically for the elderly. We conducted a systematic search in the PubMed, PsycInfo and the EuroQol database to identify articles published in English or German without any restrictions to the publication date. Studies were included, if (1) the minimum age of the sample was reported and included participants were at least 65 years old and (2) HRQoL as a primary or secondary outcome was measured by the EQ-5D. Two independent reviewers conducted the literature sifting process. We extracted the sample and study characteristics, including outcomes referring to the feasibility of the EQ-5D as a measure of HRQoL in the elderly in a qualitative synthesis. We identified 15 studies reporting information on feasibility based on four criteria: missing values, completion rates, time required to complete the instrument and -more broadly –qualitative statements referring to the completion of the measure. Only two of the included studies examined the EQ-5D-5L, the remaining 13 reported on the 3L version. Our findings suggest that missing values or comprehension problems (especially with the VAS) are slightly more prevalent in older age groups. However, these aspects are well within an acceptable range and still considerably lower than in other measures such as the SF-36 or ICECAP-O, which are frequently used in the elderly. Furthermore, older respondents seem to have a higher propensity of requiring some degree of assistance or even an interviewer-based approach, but still on a quite low level.Overall, evidence from the reviewed literature indicates that the EQ-5D is a comprehensible and short measure, which is also highly applicable in older respondents. The EQ-5D is a well-established instrument used to measure and evaluate health-related quality of life. Both versions the EQ-5D-3L and 5L demonstrated good measurement properties across many conditions and populations, strengthening the EQ-5D’s role as a basis for calculating quality-adjusted life years for use in clinical and economic evaluations. In the light of an aging population, the EQ-5D also showed adequate measurement properties in the elderly population. When evaluating healthcare services for elderly people, the measure should also be feasible and applicable to the use in an elderly population. The ease of applying the EQ-5D in an elderly sample may have an impact on data quality given that feasibility issues increase the risk of measurement error, i.e. missing responses ormulti-coded answers, or even complete study withdrawal. Yet, little is known about these properties of the EQ-5D in this population. This study aims to examine the feasibility of the EQ-5D-3L and 5L in the elderly general population by exploring the impact of age on the completion of the EQ-5D instruments further controlling for gender and the presence of long-term conditions. A prerequisite for this analysis is the availability of uncleansed, self-report EQ-5D health data from the general population. We pool across five waves of secondary datasets from the GP Patient Survey from years 2012, 2016 and 2017, where respondents self-report their health in either EQ-5D-3L or 5L depending on the wave of fieldwork. First, descriptive analysis was undertaken to analyse the distribution and proportion of missing values and completion rates stratified by age and EQ-5D version. Secondly, logistic regression models were specified to quantify the impact of age, gender and potential long-term conditions on the completion of each of the EQ-5D instruments. The total sample comprises around 4.36 million observations, of which 2.88 million respondents report their health in 5L and 1.47 million in 3L, respectively. Respondents being 65 years and above have marginally more missing values in each dimension compared to younger respondents. The highest share was observed for the oldest age group in the dimension anxiety/depression (3L 9.1% vs. 5L 7.6%), otherwise being below 5%. Consequently, completion rates (observed and predicted) decreased with age and at a higher rate after the age of 64. Evidence from our study suggests that both the EQ-5D-3L and 5L have good feasibility properties. The proportion of missing values is acceptably low across all age groups but compares more favourably for the 5LGiven the ongoing ageing trend in European countries, we expect a major increase in the proportion of elderly people with a corresponding increase in long-term care needs. While in Europe the majority of long-term care is provided by informal caregivers in a home-based setting, little is known about the impact of different care settings on the (health-related) quality of life (HRQoL) of care-dependents. Hence, this study aims to assess differences in (HR)QoL between elderly respondents receiving informal care and elderly respondents receiving formal care based on EQ-5D-3L, 5L and the age-specific WHOQOL-old. In addition to that, we aim to add to the scarce evidence of the EQ-5D’s feasibility and test-retest properties in the elderly population. Our findings suggest that differences in (HR)QoL between different care settings seem indeed to be prevalent, pointing towards higher levels of reported (HR)QoL in recipients of formal care and those in a nursing homes. Generally, the EQ-5D shows good properties in terms of response distribution, feasibility and test-retest statistics and this is irrespective of the version. Moreover, we conclude that using the WHOQOL-old alongside EQ-5D in this sample added further information ondifferent aspects of QoL from care-dependents.The EQ-5D is one of the most prominent instruments for the measurement and valuation of health-related quality of life. Due to its generic descriptive system it can be used to measure health across different disease areas and age groups. Given the growing demand of HRQoL data for clinical and economic appraisal targeted at the elderly age group the EQ-5D was extensively used in this population, too. Reference values are a helpful tool to facilitate comparisons against other groups. For example, to assess disease burden of patients, for researchers as baseline or control measurement or to determine the effects of treatments. In order to generate reference values for the German elderly population (≥65 years), we analysed a sub-set of data from the German valuation study to present benchmark values for mean utilities and VAS values as well as with regard to the response distribution stratified by age and gender. Instead of using only two age groups to characterise the elderly population, we used four age groups (65-69, 70-74, 75-79 and >79 years) to facilitate a more detailed examination of age-related HRQoL. Generally, our findings correspond well to earlier findings from a large German population norm study. Nonetheless, a more detailed provision of reference values for the elderly population seems helpful given that HRQoL in the oldest old is more volatile and differs from the young elderly, while the sought age categorisation of conventional population norms studies seems to mask these differences. Our findings may help researchers, clinicians and health economists to compare their results with those of the German general elderly population.Wolfgang GreinerPopulations and Health Systems154900Completed2015
2016010Writing up a manuscript titled 'Estimating a time trade-off EQ-5D-5L value set for China' for publicationBackground: The availability of the 5-level EQ-5D descriptive system (EQ-5D-5L) provides the opportunity of developing a new preference-based instrument for assessing health technology assessment.Objectives: To estimate an EQ-5D-5L value set for China using the health preferencesof residents living in the urban areas of the country. Methods: The values of a subset of EQ-5D-5L defined health states (n=86) were elicited using the time trade-off (TTO) technique from a sample of urban residents (n=1,271) recruited from 5 Chinese cities. In computer-assisted personal interviews, participants each completed 10 TTO tasks. Two additive and two multiplicative regression models were evaluated for their performance in describing the relationship between TTO values and health-state characteristics using a cross-validation approach. Final values were generated using the best performed model and a rescaling method.Results: The 8- and 9-parameter multiplicative models unanimously outperformed the 20-parameter additive model using a random or fixed intercept in predicting values for out-of-sample health states in the cross-validation analysis and their coefficients were estimated with lower standard errors. The prediction accuracy of the two multiplicative models measured the by mean absolute error and intraclass correlation coefficientwas very similar, thus favoring the more parsimonious model.Conclusions: The 8-parameter multiplicative model performed best in the study and therefore was used to generate the EQ-5D-5L value set for China. We recommend using rescaled values whereby 1.0 represents the value of full health in economic evaluation of health technologies in China whenever EQ-5D-5L data is available.\Nan LuoValuation4500Completed2016
2016590Extension of Variability in DCE Results Ð October 2016Background: To date, most models of health preferences are based on logits or probits. Alternatively, ratio-or angle-based models may perform better, such as Zermelo-Bradley-Terry (ZBT). Aims: to comparelogit and ZBT models for the analysis of health preferences. Methods: Using paired comparison responses from seven EQ-5D-5L valuationstudies (Netherlands, China, Singapore, Spain, Canada, Uruguay and Korea), we estimated each model separately (i.e., 20 level and one ancillary parameter) and predicted 3125 EQ-5D-5L values. Specifically, fourteen models (7 countries ×2 models) were estimated by maximum likelihood with respondent-level clusters and evaluated by their predictive validity as well as the sign and significance of their parametersand values. Results: The ZBT produced fewer disordered levels (0 vs. 8) and fewer insignificant parameters (p-value<0.01; 5 vs. 32) compared to the logit. Apart from content and statistical considerations, the logit and ZBT estimates were significantly different for 92 (65.7%) level parameters and 6816 (31.2%) EQ-5D-5L values (p-value<0.01). Across the seven countries, ZBT consistently produced a higher pseudo-likelihood and found moredifferences between countries (109 vs. 86parameters, 27696vs. 2090027696X values). Conclusions: Compared to the logit, ZBT fit the choice databetter and producedfewer inconsistencies between levels.The EQ-5D-5L values were significantly different between the models.With improved fit and content validity, differences between countries became more evident. The results here confirm previous findings and hint at the potential benefits of geometric analysis in other forms of health preference research.Mark OppeValuation9500Completed20162017
2015020Valuation of the EQ-5D-5L+RThe responsiveness of the EQ-5D-3L to patient-level changes over time in respiratory diseases likeasthma and COPD appears limited.Therefore we exploredthe potential of adding a respiratory dimension to the current EQ-5D descriptive system, i.e. a bolt-on dimension.Aim of the study was to develop a respiratory bolt-on for the EQ-5D-5L and assess the impact of the developed respiratory bolt-on on EQ-5D-5L health state valuations and the explanatory power of the EQ-5D-5L. The first part of the study, described in this report, aimedto explore the impact of different respiratory items on the explanation of the EQ-5D VAS scores and to develop a respiratory bolt-on for the EQ-5D-5L. To identify which of the respiratory complaints of asthma and COPD patients was the most promising candidate for a bolt-on three different routes were followed. Firstly, a content review of the EQ-5D and six disease-specific health-related quality of life measures for asthma and COPD was performed. Disease-specific questionnaires included in this content review were the asthma quality of life questionnaire (AQLQ), the asthma control questionnaire (ACQ), the St. Georges Respiratory Questionnaire (SGRQ), the COPD Assessment Test (CAT), the Clinical COPD Questionnaire (CCQ), and the mMRC dyspnea scale. Secondly, a post-hoc analysis of data from three clinical studies in which both EQ-5D and disease-specific quality of life were measured was done. Multivariate regression analyses were performed to investigate how much the proportion of variance in EQ-5D VAS scores that is explained (i.e. the R2) increased when we added respiratory items of the different disease-specific questionnaires tothe EQ-5D dimension scores. Thirdly, a principal component analysis was performed for the disease-specific health-related quality of life questionnaires in combination with the EQ-5D to investigate which items from the questionnaires form different constructs apart from the EQ 5D domains.Results for the content review showed that all disease-specific questionnaires include questions about the impact of asthma/breathlessness on physical activities. The EQ-5D includes a question about the impact on physicalactivities in terms of the impact on mobility. Five of the disease-specific questionnaires address asthma or COPD-specific symptoms, while the EQ-5D addresses symptoms more broadly in terms of pain/discomfort. Impact on functioning was included in four disease-specific questionnaires as well as the EQ-5D. Domains that were included in several disease-specific questionnaires, but not in the EQ-5D, were impact on sleep and use of medication. The post-hoc regression analyses showed that in four out of five questionnaires, the ACQ, AQLQ, CCQ and MRC, questions about the impact of asthma/shortness of breath on physical activities were found to explain the highest additional R2. For the SGRQ items about the problems the chest condition causes had the highest additional R2. Results of the PCA showed that for multiple questionnaires separate constructs were found on which none of the EQ-5D items loaded. These constructs either related to symptoms or impact of breathlessness on activities. Based on the combined results of the post-hoc regression analyses and the PCA we concluded that symptoms and the impact of shortness of breath on physical activities were the two most appropriate items/domains for inclusion in the respiratory bolt-on.Several possible formulations of the new respiratory bolt-onwere tested for clarity, comprehensiveness and relevance in five asthma patients, six COPD patientsand five non-patients. Based on these interviews we concluded that patients and non-patients considered a question on the impact of shortness of breath on their ability to do physical activities highly relevant and important. However, they think that the added value to the existing questions of the EQ-5D is limited. A question about symptoms is regarded of less importance, but is considered to have no overlap with the existing questions of the EQ-5D. Based on these findings we decided to include both candidate respiratory bolt-ons in the valuation study to see which of the two has the most impact on the valuation.Maureen Rutten-van MolkenValuation0Completed2015
20190550Public perspectives on patient preferences: an extension to project 2016560Aim:While differences between values derived from hypothetical and experienced health state valuation exercises are established, there is a more limited understanding of why these differences exist, what members of the public think about the differences, and how they may be practically and acceptably resolved. This research will extenda previousEuroQol project(2016560),byexploring these three domains within a general population samplein three new focus groups. This will allow us to assess the saturation of themes and probe deeper the perceived implications of differences between hypothetical and experienced valuations, and the practicalities of proposed solutions in reconciling such differences. Methods:Members of the general public (n=15) will be recruited to three focus groups, purposively sampled to achieve approximately equal numbers of men and women and a spread of ages. In the focus groups, participants will complete an introductory task that involves indicating their expectations of five consequences (enjoyment, relationships, independence, dignity, and activities) for an EQ-5D-5L moderate mental health state (11333) and a moderate physical health state (33311). Second, participants will be shown a short visually-enhanced presentation of recent data collected for the two abovementioned health states, from experience (patients) or expectation (general public), differences and similarities will be highlighted. Finally, a semi-structured discussion will be held, where participants will discuss theirviews on the differences and on potential, acceptable solutions for preference elicitation that may minimise such differences. Focus groups will be recorded, transcribed verbatim, and analysed using Framework analysis to identify common themes and sub-themes.Milad KarimiValuation22315Completed2019
2013170Further Exploration into using DCE for EQ5D-5L Valuations (FEDEV)The EuroQol Group funded the PRET-AS project, which explored Discrete Choice Experiments with duration, or DCETTO, to value EQ-5D-5L states, using an on-line survey with 1800 respondents. Scenarios were based on EQ-5D-5L states combined with one of three levels of duration. This produced a logically consistent set of predicted values for the 3,125 EQ-5D-5L states. The proposed study, Further Exploration into using DCETTO for EQ5D-5L Valuations (FEDEV) is the natural next stage of DCETTO research in the EuroQol Group. FEDEV has three aims. First: to improve the duration attribute of the DCETTO design (objectives 1-2). Second: to examine alternative approaches to designing and modelling DCETTO data (objectives 3-4). Third: to use DCETTO as a vehicle to examine more general question related to EQ-5D-5L (objectives 5-8). Objective 1: Increase the proportion of choice sets where duration varies across the pairs. Objective 2: Increase the number of levels in the duration attribute. Objective 3: Explore strategies to incorporate non-zero priors in the design. Objective 4: Explore an alternative approach to the design based on states and matched duration.Objective 5: Examine the effect of varying the order of presenting EQ-5D dimensions. Objective 6: Examine the effect of splitting up the two composite EQ-5D dimensions. Objective 7: Examine the effect of bolting off one EQ-5D dimension at a time. Objective 8: Examine the incidence of choosing shorter over longer duration of a severe state FEVED will develop five DCETTO designs, conduct an on-line survey (n=4000), and analyse the data using econometric techniques.Aki TsuchiyaValuation80145Completed2013
2015140Valuation of the EQ-5D in countries with limited research resources: investigating the potential of shrinkage analysisIntroduction Resource-constrained countries have difficulty conducting large EQ-5D valuation studies, which limits their ability to conduct cost-utility analyses using a value set specific to their own population. When estimates of similar but related parameters are available, shrinkage estimators reduce uncertainty and yield estimators with smaller mean square error (MSE). We hypothesize that health utilities based on shrinkage estimators can reduce MSE and mean absolute error (MAE) when compared to country-specific health utilities. MethodsWe conducted a simulation study (1000 iterations) based on the observed means and standard deviations (or standard errors) of the EQ-5D-3L valuation studies from 14 counties. In each iteration, the simulated data was fitted with the model based on the country-specific functional form of the scoring algorithm to create country-specific health utilities (“naïve” estimators). Shrinkage estimators were calculated based on the empirical Bayes estimation methods. The performance of shrinkage estimators was compared with those of the naïve estimators over a range of different sample sizes based on MSE, MAE, mean bias, standard errors and the width of confidence intervals.ResultsThe MSE of the shrinkage estimators was smaller than the MSE of the naïve estimators on average as theoretically predicted. Importantly, the MAE of the shrinkage estimators was also smaller than the MAE of the naïve estimators on average. In addition, the reduction in MSE with the use of shrinkage estimators did not substantially increase bias. The degree of reduction in uncertainty by shrinkage estimators is most apparent in valuation studies with small sample size.Conclusion Health utilities derived from shrinkage estimation allow valuation studies with small sample size to “borrow strength” from other valuation studies to reduce uncertainty.Eleanor PullenayegumValuation21300Completed2015
20190460Valuation of the EQ-5D-5L in ItalyBackgroundItalyis the fourth largest economy of Europe, and it has the fourth largest population. The healthcare system is decentralized between national and regional levels. The national level ensures that a nationally statuary benefit package is available to every resident in every region. Regions are allowed to reimburse additional technologies not included in the national benefit package. Economic evaluations are used for pricing and reimbursement decisions at both the national and regional levels. The EQ-5D is generally used as the main outcome for economic evaluations, as demonstrated by more than 40 patent requests from Italy only in 2018. Although an EQ-5D-3L value set is available and commonly used, an EQ-5D-5L valuation study has not yet been conducted. The current research proposal aims at developing a value set for the EQ-5D-5L in Italy.Proposed methodsPreferences will be collected using the EuroQol standardized valuation protocol available on lime survey(EQ-VT V2).This will involve using the composite time trade-off (cTTO) alongside adiscrete choice experiment (DCE). The minimum sample size will be of 1,000 participants. Participants will be recruited in public places, such as universities, libraries and hospitals, using 10recruiters i.e. interviewers in 5 cities. Interviews will be conducted by 10 interviewers selected among PhD students, researchers and other academic affiliates. Standard QCcontrols will be employed. To inform the modelling choices, GLS, Tobit, logit, probit, multiplicative and hybrid models will be fitted to the data. Models will be assessed in terms of monotonicity, errors and predictive accuracy.Aureliano Paolo FinchValuation83950Completed2020
140-RAAssessing regional variations in the impact of COVID-19 on quality of life as a function of regional social isolation controls in the U.S.Background: Many public health officials and organizations have advocated for population-wide social isolation controls (SICs) to help slow the spread of novel Coronavirus Disease 2019 (COVID-19). However, SICs may also cause significant reductions in quality-of-life as individuals are restricted from going about their normal daily routines or usual activities. There is evidence suggesting quarantines and extended periods of social isolation have a detrimental effect on mental and physical health. From intrapersonal feelings of anxiety and depression to interpersonal violence and domestic abuse, social isolation controls (SICs) are likely to have unintended consequences on the health and quality of life as the population remains in lockdown. Study Aims: This study aims to characterize the regional variation in COVID-19’s impact on quality-of-life in the U.S., with focus on the effects of region-specific SIC policies. Methods: Quality-of-life data will be taken from EQ Project 84-2020RA, a longitudinal panel survey funded by the EuroQoL group in April 2020. Information on the specific social isolation policies enacted by U.S. states will be collected from the Kaiser Family Foundation’s online repository. Precise social distancing data will be used to capture the true effects of region-specific policies. We will fit a partial proportional odds model with fixed effects to estimate the change in EQ-5D-5L scores within regions as a function of the change in SIC policy and social distancing, and employ a difference-in-difference approach, compare changes between regions.Nadine ZawadzkliPopulations and Health Systems30740Completed2020
202-RAUsing EQ-5D-Y to capture children’s HRQoL in economic evaluation: conceptual and empirical challenges for cost effectiveness modelling and implications for modellers and decision makers.The specific consideration of health-related quality of life in children is extremely important in Health Technology Assessment, but also raises a number of challenging questions in terms of interpretation and use of data in policy making. Treating QALYs in child and adult populations as equivalent is a strong assumption due to differences in descriptive systems, in valuation, and in framing. Importantly, the range of scores in youth instruments are typically smaller than observed for adults, which can lead to higher ICERs for quality of life improving interventions in children, a consequence that may be unreasonable. Considering children differently can also lead to perverse effects in economic evaluation, such as significant artefactual discontinuities in health-related quality of life in models of cost effectiveness that span childhood into adulthood. This proposal brings together experts from within and outside the EuroQol Group to build a conceptual framework for consideration of such issues, and will lay out a roadmap to help researchers and policy makers. This will identify the range of interpretations of data that we observe in EQ-5D-Y value sets, and identify a range of innovative, practical solutions based on each interpretation. We will explore the perspectives on these matters that already exist, and build in the views of our world-leading methodologists and practitioners in our Expert Advisory Panel.Nancy DevlinValuation, Youth24930Completed2020
2015210Reducing biases in adaptive Time Trade-Off using non-transparent methodsThis paper offers a new explanation of the disparity between utilities elicited using standard (or Classic) and Time Trade-Off (TTO) and utilities elicited using Chaining. we refer to “Classic” as the utility of a health state obtained from a Time Trade-off question where the end points are Full Health and Death. We refer to “Chained” as the utilities that are obtained from a TTO question where at least one of the end points is neither Full Health or Death. Previous literature has observed that the two sets of utilities are different, and they have been explained by anchoring effects. We present and alternative explanation based on the different Evaluation Mode that Classic and Chained TTO use, namely, Separate Evaluation Mode in the case of TTO and Join Evaluation Mode in the case of Chained TTO. In this paper, we present the result of a study conducted among the Spanish general population (n=346) to separate the role of anchoring from the role in the Evaluation Mode as potential explanations of the disparity between Classic and Chained utilities. All subjects had to estimate the utility of three health states (say A, B and C) using the standard TTO. They also had to answer three direct TTO questions (A vs B, A vs C and B vs C) to estimate Chained utilities. The Chained procedure always started with a comparison between two health states with the same duration (10 years). Our study also observed differences between Classic and Chained utilities. The disparity was explained by the large amount of preference reversals observed. People preferred one health state in standard TTO and another one in the choice between (A, 10 years) vs (B, 10 years). The discrepancy between Classic and Chained utilities vanished for those subjects who did not do preference reversals. Those results suggest that the difference between Classic and Chained utilities is mainly produced by the different Evaluation Mode and not so much because of anchoring effects. We present the implications of our results for preference elicitation methods for health states.Jose Luis Pinto-PradesValuation110500Completed2015
2015130Evaluating consistency between DCE and TTO valuations using multivariate mixed models and multivariate latent class analysisBackground: In health economics there has been interest in using discrete choice experiments (DCEs) to derive preferences for health states in lieu of previously established approaches like time trade-off (TTO). We examined whether preferences elicited through DCEs correlate and agree with preferences elicited through TTO tasks. Methods: We used data from 1073 respondents to the Canadian EQ-5D-5L valuation study. Multivariate mixed effects models specified a common likelihood for the TTO and discrete choice data, with s eparate but correlated random effects for the TTO and DCE data, for each of the five EQ-5D-5L dimensions. Multivariate latent class models allowed separate but correlated latent classes for the DCE and TTO data. Results: Correlation between the random effects for the two tasks ranged from 0.17 for usual activities to 0.57 for anxiety and depression, reaching statistical significance for all dimensions except usual activities. Latent classes for the TTO and DCE data both featured one latent class capturing participants attaching large disutilities to pain/discomfort, another capturing participants attaching large disutility to anxiety/depression, and the third class capturing the remainder. Agreement in class membership was poor (kappa coefficient: 0.081; 95% Credible Interval 0.033 to 0.13). Fewer respondents expressed strong disutilities for problems with anxiety/depression or pain/discomfort in the TTO than the DCE data (17% vs. 55%, respectively).Conclusions: Stated preferences using TTO and DCEs show reasonably strong association across dimensions, but poor agreement at the level of individual health states within respondents. J oint models that assume agreement between DCE and TTO have been used to develop national value sets for the EQ-5D-5L. This work indicates that when combining data from both techniques, methods requiring association but not agreement are needed.Eleanor PullenayegumValuation21650Completed2015
20180380The added value of bolt on dimensions: a systematic review of studies that analyzed the performance of proposed bolt-ons for the EQ-5DAbstract of researchBackground: The question whether dimensions should be added to the EQ-5D from a generic perspective has been debated and researched since its launch in the beginning of the 1990s. Several authors suggested valuable add-ons and subsequently the performance of severalof these bolt-ons were tested in varying (patient) populations. However, currently, an overview of bolt-on studies, performanceand the metrics that have been used to test performance is lacking. Study aim: The aim of this study is to systematically review all published studies on the (psychometric) performance of EQ-5D bolt-ons. This includes both descriptive and valuation studies. Methods: First, a systematic literature search will be conducted in EMBASE, MEDLINE and other databases. This search will be appended with a search of key papers and the EuroQol Plenary Meeting Proceedings from 1991. Second, two reviewers will screen, titles, abstracts and full-text articles independently. Third, information on the bolt on (description of the items, response categories (3L, 5L, other), language, source/origin of the bolt on, study population, cognitive debriefing of the bolt-on and (psychometric) performance metrics)will be extracted. We willcategorizethe psychometric methods that were used to test performance intothe following groups: descriptive data/preference weighted data, cross-sectional vs longitudinal-intervention-panel data, and redundance vs. predictive value vs. Shannon’s indices.Results:The study will provide an overview of bolt-on studies, format and origin of bolt-onsthat were tested,setting in which these bolt ons were studied, andan overview ofpsychometric methods and performance of these bolt-ons.Juanita HaagsmaDescriptive Systems21054Completed2018
77-RAPublic health impact of the COVID-19 pandemia: inequity of its effects and the role of health policies.The Population health impact of the COVID-19 pandemic (POPCORN) study, sponsored by the EuroQol, explored the impact of the COVID-19 pandemic on health-related quality of life (HRQL) and mental well-being of the general adult population of nine countries and investigate the role of individual and country level factors. Data were collected in the first phase of the pandemic (Spring 2020). Results showed that stringency of government response (at country level) had a small positive relation with HRQL and a strong negative relation with mental well-being, although the magnitude of effects differed for healthy persons and diseased persons. Moreover, the youngest age group (18-29 years old) had lowest mental well-being. Results among a sub-group of respondents from New York State showed that the negative impact of the pandemic was aggravated in diseased persons, in part through limited access to health care.Juanita HaagsmaPopulations and Health Systems173280Completed2020
20190990Regional differences in health-related quality of life in England: EQ-5D in national surveys of the general populationThis research investigates regional differences in health-related quality of life (HRQoL) as described by EQ-5D in England over a 20 year period.Data will be extracted from the Health Survey forEngland (HSE), an annual survey of nationally representative residents in England that between 1996 and 2013 periodically included EQ-5D,supplemented by data from other national surveys, including ONS Omnibus and the General Household Survey (GHS) which also included EQ-5D. All surveys used essentially similar sampling methods and in total provide EQ-5D data on some 150,000 respondents. Regions will be classified in terms of the degree of urbanisation (village / town / urban), Government Office (North East / North West / Yorkshire / East Midlands / West Midlands / East/ London / South West / South East) and IMD quintile score (from least deprived to most deprived). Descriptive analyses by region will be conducted. Multiple regression models will be constructed to predict HRQoL (percentages of respondents reporting problems in each dimension of EQ-5D, VAS scores and utility scores) in separate regions and in different years.Zhuxin MaoPopulations and Health Systems8800Completed2020
2014025Supplementary funding 5L value set study England: SheffieldBen Van HoutValuation24000Completed2014
2016310A head‐to‐head comparison of EQ‐5D‐3L and EQ‐5D‐5L index scores: more levels is better responsiveness?The EQ‐5D‐5L questionnaire (5L) has demonstrated better measurement properties than the EQ‐5D‐3L questionnaire (3L) in many cross‐sectional studies. In the only known head‐to‐head comparison of 3L and 5L using longitudinal data, the 5L index was not more responsive than the 3L index to health improvement experienced by stroke survivors. The 5L was found no more sensitive than SF‐6D. However, all these findings may be false because those were based on the interim ‘crosswalk’ algorithm for 5L. It is also unclear whether bolt‐on items can improve the responsiveness of EQ‐5D. This project is aimed to 1) compare the responsiveness of 3L and 5L index scores to the treatment benefit of cataract surgery (primary objective) ;2) compare the responsiveness of the two EQ‐5D index scores, the SF‐6D score, and the HUI3 score to cataract removal benefit (sedondary objective); and 3) evaluate the effect of a vision ‘bolt‐on’ item on the responsiveness of the 3L index score (secondary objective). We propose to interview 210 cataract patients face‐to‐face using a battery of questionnaires including the 3L with a vision ‘bolt‐on’ item, 5L, SF‐6D, and HUI3 before and after their cataract removal surgeries. Responsiveness of 5L, 3L, vision bolt‐on, SF‐6D, and HUI3 will be assessed in terms of Cohen’s effect size (d), standardized response mean, and Fstatistic (squared t statistic) based on the preoperative and postoperative scores generated from these instruments. Relevant country‐specific value sets of these intruments will be applied to the specific comaprisons.Nan LuoDescriptive Systems68450Completed2016
20180520Psychometric assessment of the E-QALY item pool in the United StatesThis proposal is a subproject related to the psychometric validation phase of The ‘Extending the QALY’ collaboration between the EuroQol group and UK based researchers, centrally coordinated by Sheffield University, that aims to develop a broad measure of quality of life for use in economic evaluations across health and social care. The project consists of six stages, and this proposal concerns the fourth stage. The overall aim of this study is to test the psychometric validity of the E-QALY item pool using US data. We will collect online data from a sample of 500 US cancer patientsusing a range of instruments relevant to health and social care. This includes the E-QALY item pool measures of health-related quality of life (EQ-5D-5LandPROMIS-29),and measures of wellbeing (WEMWBS). We will conduct psychometric analysis similar to that being conducted by the UK E-QALY team to better understand the generalizability of the findings of the UK to other predominant ly English speaking countries to support the item reduction and selection process. As health care policy and reimbursement in the USA can have global ramifications, it is useful to include the USA in the measure development process which may enhance the acceptability of the instrument in the future.Simon PickardEQ-HWB18400Completed2019
20180530Agreement between proxy EQ‐5D‐Y and self‐reported EQ‐5D‐Y in a paediatric patient groupStudy Design.Prospective cohort study.Objective.To compare feasibility of self-reported and proxy-reported youth version of EuroQoL Five-Dimension Three-LevelQuestionnaire (EQ-5D-3L-Y), to estimate the agreement of healthoutcome between patients with adolescent idiopathic scoliosis(AIS) and their proxies, and to examine factors that may affectpatient-proxy agreement.Summary of Background Data.The EQ-5D-3L-Y question-naire has both self-reported and proxy-reported versions. Despiteprevious studies have indicated that proxies tended to respondwith higher or lower levels of severity in specific dimensionsthan patients report, the level of agreement between childrenwith AIS and their proxies remained unknown.Methods.A consecutive sample of patients with AIS and theircaregivers were recruited. Feasibility was tested according to theproportion of missing responses. Agreements between self-reportand proxy EQ-5D-3L-Y were evaluated using percentage agree-ment, Gwet agreement coefficients and the intraclass correlationcoefficients. Linear regressions and logistic regressions were con-ducted to assess the factors associated with the agreement in healthoutcome between self-reported and proxy-reported EQ-5D-3L-Y.Results.A total of 130 patient-proxy pairs were involved in thestudy. Agreement of EQ-5D-3L-Y responses between the self-report and proxy version was good for ‘‘Feeling worried/sad/unhappy’’ dimension, and very good for other dimensions. Pooragreement in visual analog scale score was observed betweenpatient and proxy versions. Proxy’s education level, patient’scurvature type, and treatment modality were the significantdeterminants of the agreement in ‘‘Mobility,’’ ‘‘usual activities,’’and ‘‘pain/discomfort’’ dimension, respectively.Conclusion.Proxy-reported EQ-5D-3L-Y demonstrates goodfeasibility and satisfactoryagreement with patient version.Proxy’s education appears to have positive influence in agree-ment between patient-proxy dyads.Key words:adolescent idiopathic scoliosis, agreement, EQ-5D-3L-Y, feasibility, proxy.Level of Evidence:2Spine 2020;45:E799–E807Carlos WongDescriptive Systems, Youth5000Completed2019
20190120Prediction of injury recovery patterns: can multiple measurements of EQ-5D data be used in dynamic prediction models?Background:The understanding of recovery patterns of acute onset conditions supports timely identification of specific patient groups with poor prognosis.Speed of recovery may depend on many factors, including complications that may develop after onset.To predict who faces a poor prognosis, one can use dynamic prediction models. These models use time-fixed data that is available at baseline as well as time-dependent data that is measured longitudinally. Multiple measurements of EQ-5D data have not yet been used in dynamic prediction models of prognosis. Aims:The aims of our study are 1) to assess performance of dynamic prediction models that use the EQ-5D data (summary index, EQ-VAS score, dimensions, cognition bolt on) as time-dependent covariate to predict injury prognosis; and 2) to compare performance of dynamic prediction models that use EQ-5D data versus health utilities index (HUI)data. Methods:We will use data from a large cohort of trauma patients who filled out the EQ-5D and HUI at 1 week, 1, 3, 12 and 24 months post-injury to develop dynamic prediction models that provide individual estimates of good recovery probabilities given patient’s current health status using various time-fixed predictors (e.g. gender) and time-dependent predictors. We will use the landmark approach to develop the dynamic prediction models. Output:Testing the use of multiple EQ-5D measurements in dynamic prediction models and assessing the performance of these models in a patient group with varying recovery patterns may pave the way for further application of longitudinal assessments of EQ-5D data (e.g. PROMs) in prediction modeling.Juanita HaagsmaDescriptive Systems52000Completed2019
2014080Comparison of the EQ-5D-5L to measures of well-being and capability in an older populationFormal and informal social care support is an important input for individuals with long term conditions (LTCs) as they can help maintain well-being even where health is not improving. Health related quality of life measures such as EQ-5D may capture some of the impact of social care but may miss out on aspects such as independence and control and new measures such as the ICECAP-O, ASCOT or well-being have been recommended as alternatives. The overall aim of this study is to compare the new 5 level version of EQ-5D to these new ‘beyond health measures’ in a longitudinal study in order to examine the extent to which EQ-5D already captures the ‘beyond health’ domains covered in these measures. LTCs are more prevalent in older populations therefore participants aged over 65 with and without LTCs will be recruited from the South Yorkshire Cohort, an on-going study, and followed up for a year. The EQ-5D as well as the measures that go beyond health will be administered in a survey alongside health and social service use questions. The psychometric properties of the EQ-5D-5L viz. reliability, validity and responsiveness will be tested against several indicators. The ability of the EQ-5D-5L to predict well-being compared to the other measures will also be tested. Results will provide information on the performance of the EQ-5D-5L in a large elderly population as well as whether it captures aspects that go beyond health such as social care use.John BrazierOthers75616Completed2014
2013160Use of EQ-5D in elderly populations (Call for proposal 1)INTRODUCTION:At present the world is undergoing a demographic revolution, with a sizeable and ongoing increase in the absolute and relative numbers of older people in both developed and developing countries. According to the fact that chronic illnesses and disabilities has a high prevalence between the elderly, it further emphasises the importance of appropriate assessment of health related QoL (HRQoL) within the elderly. The aim of the review was to identify to what extent the QoL/wellbeing of these patients can be measured with the EQ-5D generic instrument. This paper reviews the use of EQ5D in elderly populations (not specific disease groups in old age) with special attention given to the distinction between frail elderly and non-frail elderly and between users and non-users of social/long-termcare. METHODS: A systematic literature search was conducted using the PubMed, EBSCO, JSTOR, Springerlink, Science direct and Wiley online library databases to identify articles that examined the EQ-5D questionnaire in connection to with the elderly. RESULTS: 23 articles, which address the EQ-5D questions in connection to the elderly, have been found. The presentation of the results is prepared from different angles. Firstly, the results are presented according to the countryof research, then according to the type of the study and then according to the target population of the study(i.e. frail elderly in comparison to non-frail elderly and users vs. non-users of social/long term care). The latter division is described in details with special attention given to EQ-5D instrument. CONCLUSION:It can be summarized that EQ-5D is a good instrument to assess HRQoL and also a useful predictor of mortality and first hospitalization. Therefore it can be confirmed that EQ-5D can help to find differences between frail and non-frail elderly and users and non-users of social/long term care.Valentina Prevolnik RupelOthers18000Completed2013
20190220Exploring the relationship between EQ-5D-5L and PROMIS-29The PROMIS set of measures represent an important body of work and there is increasing use of them both in the US and further afield, including in the UK and Australia. They are accompanied by utilities via PROPr. Both EQ-5D and PROMIS-29 are potential candidates for use in evidence for HTA. However, our literature review (paper 1 from this project) revealed there is very little evidence about how they compare in terms of measurement and valuation. In paper 2, we compared all theoretical values generated by PROPr based on PROMIS-29 instrument and compared these with the EQ-5D-5L value sets for US, Australia and England. There are very substantial differences in the properties of these values, and PROPr utilities have some unusual characteristics. In Paper 3 we examine and compare the psychometric performance of PROMIS-29 and EQ-5D-5L in a cross-section sample of patients with diverse health problems. The measurement and valuation properties of each instrument interact in complex ways. We conclude that which instrument is chosen is likely to have an important bearing on QALY estimates but how they compare depends on the nature and severity of the condition of interest.Nancy DevlinDescriptive Systems, Valuation, Populations and Health Systems54239Completed2019
2014190Understanding relationships between the EQ-5D and Personal Well-beingnot availableSarah DerrettOthers9997Completed2014
20190890Individual vs. proxy, child vs. adult – A systematic study of different perspectives applied in TTO valuation for EQ-5D-Y•EQ-5D-Y health states are valued by adults considering that the states affect a 10-year-old child. Earlier work has shown that this perspective may yield different valuations, compared to the perspective used in valuation of adult EQ-5D instruments. However, it is unclear exactly why these differences occur, as the perspective introduced for EQ-5D-Y involves two main changes: i) child health states are valued instead adults’ health states, and: ii) others’ health states are valued instead of valuing health states for oneself. Disentangling the influence of these two changes could be of empirical or normative importance. •In our study, we disentangled these effects by using four different perspectives in EQ-5D-Y valuation with visual analogue scale (VAS) and time trade-off (TTO) methods. Overall, differences between perspectives were consistent, but the direction of this impact depends on the health state and respondent. For VAS, the effect of perspective on outcomes of valuation depends on severity and variance was higher when deciding for children. EQ-5D-Y states valued with TTO on behalf of others (i.e. children or adults) receive higher valuations - but lower variances. •Our study suggests that the use of different perspectives may have a small effect on EQ-5D-Y valuation, but will lead to systematic heterogeneity. Seeing as it is unclear why differences between respondents occur, the search for the empirical and normative implications of perspective used in EQ-5D-Y is far from over.Stefan LipmanYouth37800Completed2020
2015390Can social care needs and well-being be explained by EQ-5D? Analysis of the Health Survey for England dataset.Introduction: It is becoming increasingly important for decision makers to be able to measure and compare diverse outcomes arising from health and socialcare interventions. Use of different outcome measures for assessment in the two sectors creates difficulties for comparisons and for capturing the multiple outcomes in a single metric. EQ-5D is the preferred measure of health-related quality of life (HRQoL) by key decision-makers in the UK. However, there is currently little clarity about how it is associated with other outcomes used in social care. Method: We empirically compared responses to EQ-5D with responses to measures of social care needs and subjective wellbeing using a large cross-sectional dataset of the UK general population, the Health Survey for England.Bivariate and multivariate regression analyses wereused to examine whethersocial care needs,measured by the Barthel Index,in people aged 65 years and over could beexplained by EQ-5Dresponsesand wellbeing measures; and how much ofthe variation in mental health and subjective well-being in the general population, as captured by GHQ-12 and WEMWBS, are explained by the EQ-5D. Results: Our study found that responses indicating poor health and wellbeing in EQ-VAS, EQ-5D Index, GHQ-12 summary score and WEMWBS pointspredicted a need for social care. Investigation of dimensions used to construct the measuresfound that two of the EQ-5D dimensions -self-care and pain/discomfort -were significantly associated withneed for social care. In addition, individuals with extreme anxiety/depression(compared to notanxious/depressed) and unable to perform usual activities (compared to no problem in performing usual activity) were more likely to report need for social care. None of the GHQ-12 dimensions were significantly associated with the Barthel Index. Two dimensions from WEMWBS,‘been feeling useful’ and ‘had energy to spare’ were significantly associated with the Barthel index. Amongst background characteristics, age and limiting longstanding illness were significant predictors of social care need across all the models. Conclusion: The need for social care,which largely depends on one’s ability to perform personal day-to-day activities,appears to be significantly associated with those EQ-5D dimensions that capture physical health and mental health.This offers the EuroQol group a strong ground to pursue how best EQ-5D instrument could be made available in capturing outcomes of social care, increasing comparability in economic evaluations between the health and social care sectors. This may includereviewingdescriptive system and/or any future development of additional or amended EQ-related questionnaire.Jeshika SinghValuation11800Completed20152016
2016450The role of EQ-5D value sets based on patient preferences in the context of hospital choice in the national PROM programme in EnglandBackground: The English NHS has been collecting EQ-5D data for all publicly-fundedpatients undergoing four types of elective surgery since April 2009. One of the intended aims ofthese data is to inform patients’ choice of hospital. To this end NHS Digital publishes hospitalperformance indicators reflecting the change in EQ-5D utility scores. These are calculated usingthe UK TTO general population value set.While the approach of using general population value sets is justified in many applications,for example when evaluating new technologies for adoption into collectively funded healthcaresystems, it appears indefensible when HRQoL data are used to inform individual patients’decisions about where to receive care. Instead the weights should reflect patients’ individualpreferences, or if this is infeasible a close approximation thereof based on samples of patientswith similar characteristics.Aims: This study explores whether using weights derived from previous patients instead ofgeneral population weights leads to a) different rankings of hospital providers and b) changes inproviders deemed positive or negative performance outliers; both of which might arguably affectprospective patients’ choice of hospital.Methods: We use EQ-5D health state data and EQ-VAS data from over 190,000 patientswho underwent hip and knee replacement surgery in England between April 2012 and March2015 to derive procedure-specific patient population tariffs. We test for i) internal consistency,ii) model fit and iii) predictive performance (root mean squared error) in an external sample ofpatients (April 2015 to March 2016). We apply these tariffs to the EQ-5D health state databefore and after surgery to calculate case-mix adjusted hospital performance estimates. Changesin hospital ranking are calculated for each patient for their five closest hospitals. Changes inoutlier status are calculated for all providers in England. We also test for patient heterogeneityin valuations using latent class modelling.Results: Patient population VAS tariffs assign lower decrements to pain & discomfort andhigher decrements to anxiety & depression than the UK general population TTO tariff. The preferred patient tariff structure differs across conditions but includes additional level 2 and 3interactions compared to the basic MVH-A1 specification. We find coefficient estimates to berelatively stable over time.Hospital rankings vary depending on the weights used and a smaller number of providersare identified as being positive outliers when patient weights rather than general populationvalues are used. Moreover, a significant proportion of patients has no well performing hospital(as identified by change in EQ-5D index score) in their local vicinity.Latent class analysis identifies up to four latent types of patients with significantly differentvalue functions. Group membership is associated with observable characteristics, such as age,gender, relative deprivation, and symptom duration.Conclusion: Patients may be better served with hospital performance estimates that reflecttheir own values and are therefore more relevant to their decisions. This finding has importantimplications for the national PROM programme in England.Nils GutackerPopulations and Health Systems25250Completed2016
2016290A qualitative study to explore the interpretation and relevance of the EQ‐5D questionnaire in 4 Asian countriesA large body of literature has demonstrated the psychometric properties of the EQ-5D questionnaire in many different cultures. In contrast, qualitative evidence on the content validity of the instrument (i.e. relevance and adequacy of the health dimensions included in its descriptive system) is very limited especially in Asia. This project aims to explore the appropriateness of the EQ‐5D questionnaire for use in a number of Asian populations including Chinese, Japanese, Koreans, Malays, and Indians. First, we aim to ascertain the relevance and adequacy, or content validity,of the EQ‐5D descriptive system in the target populations. Second, we aim to understand the thought processes these Asian populations use to rate their own health with the EQ‐VAS. We propose to collect qualitative data from nationals of China, Japan, SouthKorea,and Singapore on their views on health and EQ‐5D.In each participating country,personal and/or focus‐group interviews will be conducted to elicit what the important health dimensions are and how important the EQ‐5D dimensions are to local people, and how they interpret the EQ‐VAS and make decisions in the process of evaluating their health with the scale.All interviews will be recorded, transcribed verbatim, and translated into English for thematic analysis. the analysis will focus on identifying participants’ opinions about theEQ‐5D as a measure of health status, with a particular focus on the content and the relevance and appropriateness of the dimensions,how the measure could be improved, what is missing, the cognitive processes participants use when completing the instrument.Nan LuoDescriptive Systems86519Completed2016
2015040Using routine collection of the EQ-5D to enhance shared decision making: a proof of concept studyThe propositionfor this project is that patients are not sufficiently informed of the outcomes of total knee arthroplasty (TKA)in order to make an appropriate decision on whether they desire surgery. We propose that by using routinely-collected Patient Reported Outcome Measures (PROMs) from patients who have considered undergoingTKA surgery(including those whohave and have not), we can better inform patients about what they can expect from surgery, leading to improved decisions. Since patients systematically overestimatebenefits, and underestimate harms, we hypothesize this will lead to patients with more moderate pain choosing to delay having surgeryNick BansbackPopulations and Health Systems9400Completed2015
141-RATest-retest reliability of the EQ-5D-5L in the general population of the UK, Italy and the NetherlandsBackground: EQ-5D-5L data of the general population is collected for multiple purposes, e.g. to monitor the health status of the general population or establish population norms. Essential for each of these purposes is that the EQ-5D-5L is reliable. Previously several studies have investigated the test-retest reliability of the EQ-5D-5L in the general population administered via an in-person interview. Whether the test-retest reliability of the EQ-5D-5L administered via a web-based survey is similar compared to the test-retest reliability of the EQ-5D-5L administered via an in-person interview remains to be investigated. Aims: The aim of this study is to assess the test-retest reliability of EQ-5D-5L administered via a web-based survey for the general population of the UK, Italy and the Netherlands. Methods: The sample consists of 1864 members of the general population (aged 18-75 years) of the UK, the Netherlands and Italy who completed the EQ-5D-5L on two consecutive occasions. We will use Gwet’s agreement coefficient (GAC) test to determine the test-retest reliability of the EQ-5D-5L dimensions and the intraclass correlation coefficient (ICC) to determine the test-retest reliability of the EQ-VAS, EQ-5D sum level score and the EQ-5D-5L index value. GAC and ICC will be assessed for the respondents of the UK, Italy and Netherlands separately and for specific subgroups within a country. Output: A paper submitted to a peer-reviewed international scientific journal that describes the test-retest reliability of the EQ-5D-5L administered via a web-based survey in multiple countries and within subgroups of a country.Juanita HaagsmaDescriptive Systems15240Completed2020
20190030Use of the VAS in the EQ-5D-YBackground:The VAS is incorporated into all versions of the EQ-5D and in the adult versions, there is a box in which the respondent records the value he/she assigned on the VAS. There is no corresponding box in the EQ-5D-Y. Aim:The VMCwould like to pilot the inclusion of the box in addition to the VAS in the EQ-5D-Y versions. We also wish to examine the comprehension of the younger children with regard tothe VAS.Methods:Bothquantitative and post-coded open question data will be collected. Participants will include children of ages eight to ten years of age attending main stream schools, 60 ineach of these age categories. In addition, five ofthe children in each age group who participated in the first part of the study will beselected using systematic sampling to take part in a structured interview on the use of the VAS.Analysis: The data entry assistants will be blinded as to both scores on the VAS as the responses will be duplicatedand one of the VAS scores will be blocked out. The inter-classcorrelation co-efficient (ICC) will be calculated for absolute agreement at the individual level for each age group (8-8.9; 9-9.9; 10-10.9) and the limits of agreement will be plotted for each age category using Bland Altman plots. The time taken to enterthe data for each form of VAS will be compared.We anticipate that the two responses will be equivalent in the older children but that there might be a discrepancy in the younger children. The interview script will be transcribed and post-coded based onthematic analysis.Jennifer JelsmaDescriptive Systems, Youth9239Completed2019
2012050Extension EQ-5D-5L study England for other UK countriesEarlier this year, our research team was awarded funding from the National Institute of Health Research (NIHR) Policy Research Programme (PRP), supported by the Department of Health, to undertake an EQ-5D-5L value set for Englandstudy. The research is being undertaken as a collaboration between the Office of Health Economics and Sheffield University. Data collection is currently underway, with sampling and face to face interviews being undertaken throughout England by the social survey company Ipsos MORI. As the research funding is provided under the auspices of England’s Department of Health, we were advised by the Department of Health that the relevant definition of the ‘general public’, from whom to collect data, should be restricted to ‘the general public of England’ only –not the wider UK –and that funding could not be provided for the collection of any data outside England. This issue of whether this study should be focused just on England, or on the UK, had for some time been the subject of extensive discussion with the Department of Health –in the two years prior to submitting our proposal, this issue had been discussed in a number of key meetings between the EuroQol Group and the Department of Health. We had also consulted with colleagues in NICE (Carole Longson and Bhash Naidoo) and the Scottish Medicines Consortium (SMC) (Andrew Walker) on this issue.The 5L value set for England which we will produce for England as an outcome of our current work will be used by the Department of Health in England-specific applications -for example, in the analysis of data from the English NHS PROMs programme, and fromthe Health Survey for England. However, other decision making bodies that are very important users of the EQ-5D are responsible for slightly different populations. For example, NICE’s remit covers both England and Wales. The SMC makes decisions regarding health care technologies in Scotland. Both NICE and the SMC currently use the Dolan (1997) value set for the UK. Importantly, the system of value based pricing for new medicines, to be introduced in 2014, will be UK-wide -not just for England. As estimates of QALY gains (and therefore EQ-5D-5L data) will be central to VBP, a UK value set for EQ-5D-5L would ensure health states are valued in a manner fitting with the populations affected by VBP decisions.We are concerned that by restricting the value set to England, and failing to capitalise on the current opportunity to extend data collection efforts to Wales, Scotland and Northern Ireland, that this could potentially:(a)compromise the acceptability and perceived relevance of the new value set to key users such as NICE and in future decisions regarding VBP; and(b)adversely affect the use of the EQ-5D-5L in the UK.Because of these concerns, we sought advice from the England Department of Health on how to approach the Departments of Health in each of the ‘devolved governments’ of Scotland, Northern Ireland and Wales. Since funding for the EQ-5D-5L value set for Englandstudy was confirmed in May this year, we have been vigorously pursuing contacts in each of those countries, via email and teleconferences, to explain the study and to explore their willingness to provide funding to support data collection in each country.Initially, there was strong interest in principle from all three countries. The Office of the Chief Scientist in Scotland was subsequently confirmed the availability of funds to support the collection of additional data from the Scottish general public. Unfortunately, and despite the efforts of the people we were in contact with in Northern Ireland and Wales (which included the Chief Medical Officers and Chief Pharmacists in each case), they were unable to secure the (relatively modest) budgets required in each country to facilitate data collection.Having now exhausted all other avenues for seeking external funding for data collection in Wales and NI, we therefore ask the EuroQol Group to consider providing funding for that. Details of the aims and budget request are provided below.Nancy DevlinValuation51226Completed20192020
2014180EQ-5D-5L Electronic Measurement Equivalence ProjectThe overall aim of this study was to provide empirical evidence regarding the measurement equivalence of data collected with various data collection modes (paper, handheld, tablet,interactive voice response [IVR], and web) for the EQ-5D-5L. The objectives of this research were to:1.Collect and analyze test-retest data of the paper version of the EQ-5D-5L; and2.Quantitatively assess (via crossover studies) the measurement equivalence of various modes of data collection (i.e., paper, handheld, tablet, IVR, and web) of the EQ-5D-5L.The testing associated with these objectives involved a series of quantitative assessments of the various modes of the EQ-5D-5L. Several study samples were recruited from the UK general population to meet the objectives and conduct the quantitative assessments. The first group of participants consisted of 240 UK subjects divided across four independent samples (i.e., 60 subjects per sample). These four samples were used to assess the test-retest reliability of the paper EQ-5D-5L, as well as to test the agreement in the scores produced by the paper measureversus the three screen-based implementations of the EQ-5D-5L (i.e., handheld, tablet, and web). The second group of participants consisted of 61 UK subjects recruited to test the agreement in the scores produced by the paper, web, and IVR versions. Test-retest analysis of the paper EQ-5D-5L and analyses of the agreement of the scores of the original paper mode and the electronic modes (i.e., handheld, tablet, web, IVR) were conducted. Specifically, a two-period, repeated measures design was used for assessing the test-retest reliability of the paper form in one sample of 60 subjects. A three-period repeated measures design, with subjects equally allocated to order of completion, was used for testing the agreement among the various modes (i.e., handheld, tablet, web, IVR) in the separate samples of subjects. Statistical analysis consisted of testing mean differences based on a dependent sample t-test for the test-retest analysis, and repeated measures analysis of variance (ANOVA) for the three period contrasts. Estimation of agreement was based on the intraclass correlation coefficient (ICC). Analysis of the test-retest data resulted in mean differences near zero and ICC values above 0.90 for both the index and the global health rating scale (i.e., EQ VAS) scores. These data were also used to set the threshold values for the three period crossover assessments. For the mean differences, one-half of the standard deviations of 0.080 for the index and 8.0 for the EQ VAS was computed. Hence the equivalence intervals for mean differences were set as -0.040 to 0.040 for the index and -4.0 to 4.0 for the EQ VAS. The thresholds for denoting equivalence on the ICC for the three period comparisons were ICC lower 95% confidence interval (CI) ≥0.911 for the index and ≥0.940 for the EQ VAS.For the three period crossover designs, all of the mean differences for the index and EQ VAS were wholly contained in the equivalence intervals. Further, the ICC 95% lower CIs for the index and EQ VAS scores showed values above the thresholds for denoting equivalence across all comparisons in each sample. No significant mode-by-order interactions were present in any ANOVA model. These data, taken collectively, provide strong evidence that the data produced by the paper and electronic versions of the EQ-5D-5L are highly correlated and produce negligible differences in mean scores at the group level.Stephen CoonsOthers96430Completed2014
20191120Population norms and inequalities based on EQ-5D-5L general population surveys (POPS 2 Project)Populationnorms and inequalities based on EQ-5D-5L general public surveys (POPS 2 Project)One factor associated with the success ofthe EQ-5D instrument has been the ease of constructingcomparable international datasets, partlydue to the existence of user guides and standards of analysis. A book summarizing results from international EQ-5D-3L general population surveys on self-reportedhealth (“Self-Reported Population Health: An International Perspective based on EQ-5D”) continues to be in very high demand and has been downloaded over 50,000 timessince publication in 2014.A new edition of this book is necessary for a number of reasons. First, new population datasets have become available based on the EQ-5D-5L,including countries hitherto not being represented, including countries outside of Europe. These data were usually collected under higher quality control. Second, the selection of indicators (both economic and epidemiological) was elementary, where some amendments weremade due torecent guiding sources. The aim of the current proposal is to produce a new edition of population health data, including norms and inequality indicators based on EQ-5D-5L surveys conducted in representative samples of general populations. Specifically, the following are the key objectives:1)Identify EQ-5D-5L datasets from general population surveys. Create a database archive using standardized structure2)Generate EQ-5D-5L population norms3)Generate inequality indicators, both fromepidemiological and economic origin4)Make cross-country comparison of EQ-5D-5L population dataOutputs of the project will support future research activities of the EuroQol Group’s Health Inequalities Special Interest Group, such as future in-depth analyses of health inequalities using the dataset archive.The project would also have implications for other working Groups, such as the Valuation and the Education & Outreach Working Groups.Agota SzendePopulations and Health Systems15000Completed2020
2016080Variation in health state preferences across local and international populations: East doesn't meet WestThe 2016 CADTH Symposium is the main scientific venue for researchers who conduct health technology assessment/health economics in Canada and government agencies that are involved in decision making in health care. It provided an excellent opportunity to: (1) promote awareness of The EQ-5D-5L Canadian Valuation study and the availability of the value set; (2) present research on geographic variation of preferences, and differences in the relationship between elicitation methods across regions and discuss its implications; (3)highlight the relationships between different preference-based elicitation techniques based on international valuation studies using a standardized protocol (EQ-VT) that demonstrates the state of the science supported by the EuroQol group to inform HTA around the world. Both are areas of scientific and policy-based interest to the EuroQol Group. These topics were of specific interest to policy makers and researchers in Canada, given that provinces are responsible for the delivery of health care services.Feng XieEducation and Outreach10871Completed2016
20190810travel scholarship for Olivia ErnstssonJeffrey JohnsonEducation and Outreach7819Completed2019
20190830travel scholarship for Xueyun ZengNan LuoEducation and Outreach6190Completed2019
20190660Funding application: Early Career Researcher Meetingnot availableKoonal ShahEducation and Outreach48275Completed2019
2016560Investigating the difference between hypothetical and experienced valuations: the case of mistaken expectationsLittle is known about why experienced and hypothetical health state preferences differ. Patients and the public may have inherently different preferences, but it is also possible that the public does not have accurate expectations about what life is like in ill-health. Hypothetical health state valuations may thus differ from experienced based valuations, for example, if the public is mistaken in how ill-health affects their enjoyment of life. This study investigates the extent to which: (i) the public’s expectations about life in a health state are associated with their valuation of that health state, and (ii) the public’s expectations are similar to the experience of patients.We collect expectations of the public on five consequences (enjoyment, relationships, independence, avoiding being a burden, and activities) using an online survey of 1300 UK residents (completed August 2017). Participants were asked to imagine living in five EQ-5D-5L health state, to value them using the Visual AnalogueScale (VAS), and to report their expectations on the six consequences for them. Data on the experience of patients was available from the Multi Instrument Comparison study, where individuals self-reported the EQ-5D-5L and five consequences.To assess theextent to which the public’s expectations about life in a health state are associated with their valuations of that health state, we compare the explanatory power of two models of the VAS valuations, one using the EQ-5D-5L classification system as independent variable, and one using the consequences as independent variables. To assess the extent to which the public’s expectations are similar to the actual experience of patients, data on public expectations and patient experiences were jointly modelled. Separate regression models were estimated for each of the consequences (enjoyment, relationships, independence, avoiding being a burden, and activities) using the EQ-5D-5L classification system as independent variables with interaction effects to account for the differences between expectations and experience. For the first aim, preliminary analysis suggests that the expected consequences better explain VAS values than the health state classification system itself, indicating that health is at least partly valued based on the expected consequences.For the second aim, preliminary analysis suggests that the public generally overestimates the effect of ill health on the consequences (e.g., they expect ill health to affect their relationships more than patients report it actually does). The difference between the public’s expectations and patients’ experience differs for the health dimensions: (a) the general public overestimate the effectsof mobility and self-care problems more than they overestimate usual-activities and pain/discomfortand (b) the general public underestimate the effect of anxiety/depression problems.The results of the preliminary analysis study suggest that differences between patient and public preferences may be driven by differences between expectations and experiences of what life is like in ill-health, and that these differences differ systematically per health dimension. In addition, the EQ-5D-5L tariff that elicits preferences using hypothetical valuations of the general public may overestimate the importance of mobility compared to anxiety/depression, potentially suggesting that existing tariffs could be a sub-optimal reflection of the publics’ informed preferences.Milad KarimiValuation69541Completed20162017
20190140Valuing health benefits for children and adolescents: Qualitative research examining the impact of perspective and respondents’ priorities around adult and child healthAim: The valuation of child and adolescent health-related quality of life raises methodological considerations, such as whose preferences, what perspective, and which elicitation technique. Prior quantitative work has shown that these choices impact on elicited preferences, but has not explored why. This qualitative project aimed to produce a better understanding of the impact of: a) perspective and elicitation technique; and b) views around, and prioritisation of, child/adult health, on the valuation of child and adolescent health states by adults.Methods: Six focus groups with 30 members of the UK adult general public were conducted and analysed using Framework analysis. The focus groups had two stages. First, participants individually valued two EQ-5D-Y (youth version of EQ-5D) health states usingTTO and DCE pairwise comparison tasks (without duration), iteratively assuming four perspectives: themselves; another adult; a 10-year old child; themselves as a 10-year old child. Second, a semi-structured discussion explored: task understanding and whatinformed responses; potential differences by perspective and task; views on the prioritisation of child/adult health; and which techniques should be used for informing policy. Results: Participants views were often heterogeneous, but key findings emerged. Typically, participants had a different willingness to trade-off life years for children versus themselves, and this differed depending upon the health profile and who was imagined by participants. The same health states were often viewed as having a different impact on utility for children than adults, but in an inconsistent way across participants. While no strong preferences for prioritising child health over working-age adult health were exhibited, there was consensus over a ‘fair innings’ argument and prioritising younger people over the elderly. Adopting other perspectives was generally more problematic than an own perspective for participants, with them finding the latter preferable.Conclusions: If an adult sample is used to value child and adolescent specific health states, we recommend that an own perspective is used, since this is likely to minimise methodological problems associated with the use of different perspectives, including emotion and bias. If a child perspective is used we recommend that sampling is representative for parental status of children aged below 18, since this can impact on preferences.Donna RowenValuation, Youth43419Completed2019
2015170Monitoring neurotrauma patient outcomes in Bandung, Indonesia: A feasibility studyIntroduction: Almost a quarter of Indonesia's 250 million citizens live in Western Java. The Dr. Hasan Sadikin Hospital (RSHS), in Bandung City, is a major referral hospital for the region. RSHS Neurosurgery Unit admits patients following major trauma or other neurological conditions. Little is currently known about neurosurgery patient outcomes in Indonesia, particularly regarding factors preceding admission and following discharge. Prior to our study, patients were often discharged with limited follow-up. Our study sought to describe the characteristics of patients admitted to, and discharged from, the Neurosurgery Unit and to assess the feasibility of undertaking follow-up telephone interviews post-discharge. Of 184 consenting patients, 46% were admitted because of trauma/injury, 37% brain tumour, 9% sudden health event (mostly cerebrovascular accident), and 8% other conditions. The majority had a mild (77%) or moderate (20%) GCS on admission; mean age was 41 years. Trauma patients spent an average of 8 days in hospital; non-trauma patients 21 days. Most (97%) were discharged back to their pre-admission residence. Despite the majority (88%) living more than one hour away from RSHS, 83% were able to attend a clinical follow-up appointment. Clinicians reported that the process of undertaking this study, and obtaining patients' contact details, enabled a much higher follow-up appointment rate than had previously occurred. By the time of the one-month interview: five patients had not reached the scheduled follow-up date, six had died, six were uncontactable; 154 completed an interview. Of these, a high proportion reported extreme problems pre-discharge for the dimensions of mobility (21%), self-care (23%), and usual activities (38%). Proportions had reduced by one-month, but were still high (12%, 12% and 18% respectively). Most reported no or slight problems with pain/discomfort and anxiety/depression at both time points (85% - 95% and 96% - 97% respectively). Conclusion: This study has shown it is feasible to follow neurosurgery patients by telephone one month after hospital discharge. A high proportion of patients still reported difficulties with mobility, self-care and usual activities. Data collected at the remaining two and three-month interviews will continue to provide valuable infonnation about both longer-tenn outcomes and feasibility of follow-up. Methods: Eligible patients were aged .!:18 years, admitted to RSHS Neurosurgery Unit (19/10/2015 - 17/02/2016), and subsequently discharged. Baseline data was collected on admission from clinical notes by registered nurses. An in-person interview (conducted in the Indonesian language) was held 1-2 days pre­ discharge; follow-up telephone interviews were scheduled for one, two and three months post-discharge. Socio-demographic and clinical characteristics, health pathways, Glasgow Coma Scale (GCS), and EQ-5D- 5L data were collected. Data were recorded on a patient case report form and entered into an electronic database. Statistical analyses included proportions for categorical data, means for normally distributed data, and a student t-test for comparison of means. Results: There were 217 patients admitted to the Neurosurgery Unit during the recruitment period; 33 died before discharge. All surviving patients (n=184) consented to participate but four were discharged prior to an interview being arranged and nine were still in hospital, therefore 171 (93%) completed a pre-discharge interview.Sarah DerrettPopulations and Health Systems54157Completed2015
2016520Measurement Properties of the EQ-5D-5L: Comparative Performance with other Measures and Determining Minimal Important Differences in Canadian Populationsnot availableJeffrey JohnsonPopulations and Health Systems120000Completed20162018
2012030EQ-5D-5L valuation study in Hong KongA report of the study is available here: Wong ELY, Ramos-Goñi JM, Cheung AWL, Wong AYK, Rivero-Arias O. Assessing the Use of a Feedback Module to Model EQ-5D-5L Health States Values in Hong Kong. Patient. 2018 Apr;11(2):235-247. doi: 10.1007/s40271-017-0278-0. PMID: 29019161; PMCID: PMC5845074.Carlos WongValuation0Completed2012
2016730Workshop to discuss the legitimacy, estimation, and uses of the Minimal Important Difference (MID) with EQ-5DThere is clearly muddled language used in the literature and wide variations in methodologies. Despite this, there appears to be a tendency for researchers to quote one or two key papers when justifying the use of a single “off the shelf” MID despite the variation in MIDs across patient groups (e.g.the 0.074 averagefrom Walters & Brazier, 2005). In addition, there is no evidence of any exploration of how or whether a MID could be used with the descriptive systemrather than an index; the Pareto classification of change might provide a starting point for thisfor example(Devlin, Parkin and Browne, 2010).There appears to be some evidence that MIDs are different depending on whether they are used for an improvement or a deterioration in health, although this difference does not appear to be large. There is also a clear focus on the index (and EQ-VAS to a lesser extent) but this is not necessarily appropriate given that indices are typically based on general population values. It isalsocurrentlyunclear as to how the state transition approach might fit in alongside or compare with the more common anchor-based and distribution-based methods.Michael HerdmanEducation and Outreach17550Completed2017
75-EOPresenting E-QALY study results in 2020 ISOQOLThe extending the QALY project (E-QALY), funded by the UK MRC and EuroQoL Research Foundation aimed to develop a broad generic measure of quality of life for use in economic evaluation across health care, social care and public health. The study had 5 development stages: 1) establishing the domains; 2) generating a long list of items; 3) testing the face validity of the items; 4) psychometric testing of the items; 5) valuation of selected items. Work was undertaken in six countries: England, Argentina, Australia, China, Germany and USA. Extensive consultation with stakeholders, experts and the public was used to refine the domain structure and identify high performing items to measure each selected domain. We aim to present the development process of the E-QALY project at the 2020 ISOQOL Annual Conference, Prague, Czech.

The symposium will discuss the opportunities and challenges of developing a new international preference-based measure. John will provide the rationale for developing the measure and an overview of the project. Clara will present findings from Stages 1 and 2 where the domains and items for the measure were identified. Zhihao will focus on the face validation results with a particular focus on China (Stage 3). Ole will discuss the psychometric testing with particular focus on Germany (Stage 4). Donna will present the findings from the valuation stage. Finally, Madeleine King will start the discussion by providing her view on the approach taken and the resulting measure.
Zhihao YangDescriptive Systems19400Completed
103-RAUsing EQ-5D to measure health status in Chinese populations during the COVID-19 pandemicSince the new coronavirus (COVID-19) first emerged in Wuhan China at the end of 2019, the virus has spread to more than 200 countries and territories around the world. China was the first country to report COVID-19 cases and to impose strict lockdown, quarantine and outdoor restrictions. This research aims to investigate the health status of Chinese people during this pandemic. It can also show the usefulness of EQ-5D in describing health status in this global health crisis. We carried out an online questionnaire survey to collect health status information from the Chinese general population. EQ-5D-5L was used as the major component of our survey. Respondents’ sociodemographic information was also collected. The survey was distributed via the WeChat platform to Chinese mainland citizens living in Hubei (the lockdown province) and outside Hubei between February and March 2020, and to Chinese citizens living abroad between mid-March and April 2020. Data of the current survey will be compared with data collected in previous Chinese National Health Services Surveys, in terms of the likelihood of having any problems in all EQ-5D dimensions and mean VAS scores. Among respondents living in their home country, people living in Hubei province will be compared with those living outside Hubei. Health status of overseas respondents will be compared with that of respondents at home.Zhuxin MaoPopulations and Health Systems5280Completed2020
2015230Funding proposal for an scholarship for international travel and cooperationTo startup and to develop a detailed protocol for methodological valuation research in China, using the opportunity of an already funded talented PhD student (Zhihao Yang), we want the research group (PhD - (co)supervisors and PhD candidate) to meet in Rotterdam. To facilitate such meeting, we applied for EuroQol Funding. The PhD student Zhihao Yang was already in Rotterdam, just like his supervisor Jan Busschbach and the co-supervisor Elly Stolk. The second co-supervisor Nan Luo is located in Singapore, and therefore we ask for funding the time, travel and housing of Nan Luo for two weeks, from 21st Sep to 2nd Oct, 2015. We also asked for funding for the time Jan Busschbach and Elly Stolk during the stay of Nan Luo, to facilitate the cooperation. At that time Elly was still employed by the university, and therefore we asked funding for her university time for this cooperation.Jan BusschbachEducation and Outreach20500Completed2015
20200030travel scholarship Lucky NgwiraBackground & aims: Preference-based measurements (instruments) of childhood health-related quality of life (HRQoL) in sub-Saharan Africa (sSA) are limited. This work set out: i) to explore how the existing instruments are adapted for use in sSA; ii) to establish if the adaptation process aligns with local health perceptions in Malawi; and iii) psychometrically validate the adapted instruments. Methods: Step 1: Cross-cultural adaptation of childhood preference-based HRQoL instruments The adaptation process included the following two stages. i) Translation Two independent translators forward and backward translated selected instruments (EQ-5D-Y, EQ-5D-Y-5L, PedsQL™ 4.0 child self-report, and PedsQL™ 4.0 teen self-report) into Chichewa language for Malawi. The translated versions were piloted (cognitive interviews) in a sample of participants (n=38), aged 8-17years from the main referral hospital and primary and secondary schools within Blantyre city, the commercial capital of Malawi. ii) Cognitive interviews process a) Self completion of the questionnaire For the EQ-5D-Y-5L only, a card exercise was employed to assess the correct hierarchical ranking of response options. All the Chichewa translated instruments were administered to different sets of participants for self-completion. If necessary, children who struggled to understand instruction for self-completion of the questionnaires were given assistance but not on how to respond to them. b) Cognitive interviews Each child participating in the interviews then took part in a one-to-one interview about their understanding of the questionnaire. The interviews were open ended, and participants were asked to comment on how they understood each question and response options. Following piloting and making necessary changes, pre-final Chichewa versions were sent to developers for approval. Step 2: Psychometric validation The approved Chichewa versions were administered to a larger sample (n=298, age 8-17 years) consisting of healthy (n=95) and sick (chronic and acute, n=194) children to assess psychometric performance. The EQ-5D-Y and EQ-5D-Y-5L have five dimensions each with three and five response options respectively. The PedsQL™ 4.0 has 23 items and five response options each. The EQ-5D-Y and EQ-5D-Y-5L scores were generated in two different ways. HRQoL sum scores were generated by adding numeric values of response levels and utility scores were generated using USA utility values to evaluate psychometric performance and utility performance (empirical validity) respectively. The PedsQL™ 4.0 scores were generated following an algorithm provided by the instrument developers. The instruments performance were then assessed using standardized psychometric criteria: item performance (missing data >5%); internal consistency reliability (Cronbach's alpha =0.7); and validity (convergent of EQ-5D-Y sum scores and PedsQL™ 4.0 scores >0.4; discriminant: no association between HRQoL scores with gender & grade; known groups between healthy and sick groups: effect size >0.5). Additionally, since the EQ-5D-Y and EQ-5D-Y-5L are utility instruments, they were assessed for empirical validity (relative efficiency of the EQ-5D-Y-5L relative to the EQ-5D-Y, where >1 means the former performing better than the latter). Content validity was later evaluated by firstly developing a conceptual framework for health from four focus group discussions (FGDs) composed of 6-12 participants, aged 8-17 years. The framework was conceptualized using axial and selective coding FGDs to generate main concepts which were considered important by children and adolescents in a Malawian setting. The concepts from the model were compared against those of the existing EQ-5D-Y dimensions to assess if the EQ-5D-Y contains all the relevant dimensions that children and adolescent consider important in this setting. Results: Ten preference-based HRQoL instruments were identified and none of these had been developed in low- and/or middle- income countries. Of these, only two (HUI3 and EQ-5D-Y had been used in a sSA setting. : i) EQ-5D-Y (preference-based), and ii) PedsQL™ 4.0 self-report (non preference-based) were selected for cross-cultural adaptation into Chichewa and subsequently psychometrically validated in a Malawian setting. i) Translation There were several conceptual and linguistic problems identified from the translation process. Most of these were identified and resolved by translators during the consensus process. However, translation issues such as those referring to ‘life’, ‘health’, and ‘healthy’ were problematic. Some translation issues including those pertaining to qualifiers for responses and dimension equivalence were only resolved during cognitive interviews. ii) Cognitive interviews process a) Self-completion of the questionnaires Children, 0.7) for all instruments. The correlation of EQ-5D-Y (EQ-5D-Y-5L) sum scores and PedsQL™ 4.0 overall scores was >0.4, providing evidence of convergent validity; but the correlation between some of the EQ-5D-Y dimensions and PedsQL™ 4.0 self-report sub-scales was mixed. There was no association between HRQoL scores with gender and age but the association with school grade (p<0.05) showed lack of discriminant validity. For empirical validity, the EQ-5D-Y-5L was 31%-91% less efficient than the EQ-5D-Y at detecting differences in health status using external measures. Discussion and conclusion: Health, life and healthy were complex terms to translate, and even with help from cognitive interviews these were found to be problematic. The cognitive interviews were instrumental in isolating translation problems and appropriate hierarchical ordering. The psychometric findings demonstrate that the Chichewa (Malawi) PedsQL™ 4.0 child self-report and PedsQL™ 4.0 teen self-report established reliability, but evidence of their validity was mixed. While the PedsQL™ 4.0 child self-report met most psychometric criteria, the missing data and lack of known-groups validity means this should be used with caution in younger children and in those with different health conditions. The study did not find much difference between the EQ-5D-Y and EQ-5D-Y-5L in terms of missing data, reliability and known-group validity. The EQ-5D-Y had problems with known-groups validity. The EQ-5D-Y-5L demonstrated reliability and validity except discriminant validity with grade. Whereas the EQ-5D-Y-5L performed better than the EQ-5D-Y in terms of discriminant validity, the latter outperformed the EQ-5D-Y-5L in the utilization of response options as well as in terms of empirical validity. The EQ-5D-Y and EQ-5D-Y-5L were found to be reliable for use among children and adolescents in relation to economic evaluation. However, psychometric testing is required for test re-test reliability and responsiveness that could not be carried out in this study due to COVID-19 restrictions. In conclusion, this research produced Chichewa EQ-5D youth versions which have potential for use in this setting among children and adolescents in Malawi but with caveats.Lucky NgwiraOthers14700Ongoing2020
20170210Testing the impact of potential bolt-ons on preferences using pairwise choices: A pilot study.Background:It is accepted that the EQ-5D may miss dimensions relevant for some conditions. Whenthis happens, a possible solution is adding bolt-ons to expand its descriptive system. A recent study has identified eight candidate bolt-ons for the EQ-5D-5L: relationships, hearing, life satisfaction, speech, cognition, vision, energy/vitality and sleep (Finch et al., 2016). Developing and incorporating bolt-ons into EQ-5D values sets is complex and costly, so it is important to establish which ones are likely to impact on preferences. This study teststhe use of pairwise choices for the selection of bolt-on dimensions by examininghow bolt-ons alter public preferences for the EQ-5D-5L health states. Methods:Preferences of the general population were collected using an online survey of 1040 UK residents. The survey presented participants with pairwise choices of EQ-5D-5L health states without and with bolt-ons of hearing, sleeping, cognition, energy and relationships. Three health state pairs were chosen among those where responders’ preferences were in approximately equal proportions in the discrete choice experiment of the EQ-5D-5L value set for England study (Devlin et al., 2016). Participants were asked to imagine living in those health states for 10 years and then die, and to choose their preferred state. To assess whether bolt-ons changed preferencesstatistical testing of differences in proportions of responses between the pairwise choices without bolt-ons and the corresponding pairwise choices with the bolt-ons wasperformed using Z tests. Todiscriminate and help inform the selection between bolt-ons, differences in the proportions of preferences between bolt-on at the same level for the same health state pairs were tested using Z tests.Results:Each of the individual bolt-ons had a significant impact on preferences for the EQ-5D-5L. The extent of this impact varied according to the bolt-ons and their severity level, as well as the health states to which they were added. Additions of bolt-ons at level 1 generally resulted in differences of ± 10 percentage pointscompared to the same pairwise choice without bolt-ons. These were not statistically significant. Additions of bolt-ons at level 3 generally produced a statistically significant reduction in the percentage ofindividuals choosing the health state to which the moderate level was added. Addition of bolt-ons at level 5 produced a further reduction. At level 5 hearing had the largest impact, followed by energy, cognition and relationships with similar impacts, andsleeping with the smallest impact. By contrast, at severity level 3 cognition producedthe largest switch in preferences, followed by hearing, relationships and sleep, with energy registering the smallest switch. Conclusions:Use ofsimple pairwise comparisons of health states is a feasible technique toassess the potential importance of bolt-ons. However, the relative weight responders place on different health problems is not constant across levels of severity between bolt-ons.A FinchDescriptive Systems, Valuation9313Completed2017
2016490Statistical analysis and preparation of a manuscript describing the responsiveness, and impact on outcomes of administering the EQ‐5D‐Y via internet to children/youths with Type I Diabetes Mellitus in clinical practice.OBJECTIVEThe objective of the study was to assess the potential for using an internet-administered version of EQ-5D-Y as a routine outcome measure in children/youths with T1DMin clinical practice. Specifically, to assess its capacity to detect changes in health comparing with other measures of Health-Related Quality of Life (HRQOL). METHODSThis is a secondary data analysis of a longitudinal study designed totest whether the systematic monitoring of HRQOLin Spanish pediatric patients with T1DM helps improve their daily life measured with KIDSCREEN-27. Of 205 potential candidates, 136 were consecutively recruited between July and December 2014 from a list of outpatient with T1DM attending to paediatric endocrinology departments at 5 hospitals in the Barcelona province (Spain). Paediatricians were randomly allocated to either the intervention or the control group. The intervention consisted of discussing HRQOL scores between the doctor and the patient at each routine visit. The EQ-5D-Y, the KIDSCREEN-27, and the Strengths and Difficulties Questionnaire (SDQ) were completed, and clinical variables were collected, during the first and last visit of the study in both groups. Changes between evaluations on dimensions of the EQ-5D-Y, EQ-VAS, EQ-5D-YSummaryscore, KIDSCREEN-27 physical and psychological well-being dimensions, KIDSCREEN-10 index, and SDQ were described and compared between groups. Sensitivity, Specificity, and Positive and Negative Predictive Values (PPV/NPV) of the EQ-5D-Y dimensions were estimated. Multivariate models of repeated measures were constructed to assess the HRQL impact of monitoring health over time. RESULTSAt the end of follow-up response rate was 82.6%. Mean of age was 14 years old, and mean time from diagnosiswas 5.56 years.The comparisonbetween 1stand 4thfollow-up visits showed statistically significant improvement in the intervention group for: ‘Worried/Sad/Unhappy dimension’(28.6% and 15.7%, p=0.035), the SummaryEQ-5D-Y score (9.29 and 5.29, p=0.013), the EQ-VAS (78.5 and 83.2, p=0.009), the KIDSCREEN-10 index,and the two KIDSCREEN-27 dimensions. No differences between both visits were found in the control group, except for the SDQ. The GEE showed no differenceson EQ-5D-Y change between groups.EQ-5D-Y dimensions and Summaryscore showed good sensitivity to detect stability (38.5%-95.5%) in health status and high specificity to detect worsening (89.7%-100%) and improvement (77.4%-100%) when using changes in theKIDSCREEN as the criterion measure.EQ-5D-Y dimensions and Index also showed a very high PPV (70.4%-100%) for detecting improvement and a high NPV to discard worsening (71.2%-82.7%) and stability (75%-96.6%).CONCLUSIONThe electronic EQ-5D-Y is wellaccepted by children/youths with T1DMto be used as a routine outcome measure in clinical practice.Results confirm ourhypothesis that mental health improved in the intervention group,as measured by Worry/Sad/Unhappy of EQ-5D-Y and Psychological Well-Being of KIDSCREEN-27. Also, as hypothesized,the EQ-5D-Y shows a high capacity for the detection ofHRQL improvement in these patients, but our results did not confirm the hypothesis about its capacity to detect worsening.Since this latest issue is related to the low proportion of patients experiencing worsening in our sample, further studies are needed to test this capacity.Luis RajmilYouth15000Completed20162017
2016270The EQ-5D-Y in a clinical sample of children and adolescents with asthma, diabetes and rheumatoid arthritis: convergent validity, agreement between self-and parent-reports as well as sensitivity to changePurpose: The EQ-5D-Y assesses child and adolescent health-related quality of life(HRQoL) by means of five items on mobility, self-care, usual activities, pain/discomfortand anxiety/depression as well as a Visual Analogue Scale (VAS) on the self-ratedcurrent health state. The present study investigates predictors of self-reported HRQoLaccording to the EQ-5D-Y in a mixed clinical sample of chronically ill children andadolescents based on longitudinal data.Methods: Data from the German Kids-CAT study on children and adolescents withasthma, diabetes and juvenile arthritis gathered over a period of six months wereanalyzed (n=310; aged 7 to 17 years; 48 % female). Socio-demographic as well asdisease- and health-specific predictors were investigated. Generalized linear mixedmodels and linear mixed models served to examine both the items and the VAS of the EQ-5D-Y.Results: Ceiling effects for the EQ-5D-Y were detected in the analyzed sampleindicating low burden of disease. Results of longitudinal regression-based analysesrevealed effects of health complaints on all investigated domains of HRQoL. Further,besides age- and gender-specific effects, associations of HRQoL with disease control,the duration of the disease as well as mental health problems were found.Conclusions: Our findings support the content validity of the EQ-5D-Y and underlinethe necessity to address subjective health complaints and mental health problems inthe care of children and adolescents with asthma, diabetes and juvenile arthritis.Ulrike Ravens-Sieberer Ravens-SiebererYouth15000Completed2016
2013260Project Save TTO valuationsnot availableAlexander AronsValuation10000Completed2013
2016340Uses and applications of EQ-5D in Latin America & the Caribbean - Systematic Review and bibliometric studynot availableLucila Rey AresEducation and Outreach14992Completed2016
2016160Conventional and perceived change in health-related quality of life of trauma patients: what role does recall bias play?Background:Retrospective assessment of pre-injury health-related quality of life (HRQL) is frequently used to measurechange from pre- to post-injury HRQL. However, retrospective measurement may be confounded by recall bias. It isassumed that presence of recall bias is influenced by several factors, such as the measurement scale or the instrumentthat is used, the measurement schedule, and the presence of a substantial health event during the follow up period.This study empirically tests these assumptions by comparing pre-injury EQ-5D summary scores, EQ-5D profiles andvisual analogue scale (EQ-VAS) scores of trauma patients, as recorded 1 week and 12 months post-injury, respectively.Methods:A sample of 5371 adult trauma patients who attended the Emergency Department (ED) followed by hospitaladmission, received postal questionnaires 1 week (T1) and 12 months (T2) post-injury. The questionnaires contained itemson pre-injury health, in terms of EQ-5D3L and EQ-VAS.Results:Onethousandonehundredsixty-sixcompleteddatapairs with T1 and T2 pre-injury data were available. Meanpre-injury EQ-5D summary scores were 0.906 (T1) and 0.905 (T2), respectively, with moderate intertemporal agreement(intraclass correlation coefficient (ICC) T1T2 = 0.595). In absolute terms, 442 (37.9%) respondents reported adifferent pre-injury EQ-5D profile at T2 compared to T1. The least stable EQ-5D dimension was pain/discomfort (20.2% reported a change). Mean T2 pre-injury EQ-VAS score was significantly higher than meanT1 pre-injury EQ-VAS score (T2 84.6 versus T1 83.3). Multivariable logistic regression analysis indicated thatlower educational level, comorbiddisease and having PTSD symptoms were independent predictors ofchange of pre-injury EQ-5D profile.Conclusions:Despite one third of respondents reported a different pre-injury health level, if asked for on twointerview occasions separated by 1 year, on the group level this difference was nil (EQ-5D summary score) to small (EQ-VAS). The consistent symmetrical pattern of change suggests random error to play the largest role. Intertemporal reliabilitywas the same in EQ-5D profiles vs. EQ-VAS scores, ruling out scale effects. Particularly certain trauma subgroupsshowed highest distortion. While group comparisons may be trusted, in pre-post analysis and repeated measure analysisthe individual injury impact and recovery pattern may be wrongly estimated.Keywords:Health-related quality of life, EQ-5D, Recall bias, Trauma populationGouke BonselPopulations and Health Systems64281Completed2016
2016530Application study of the EQ-5D-5L in oncology: linking self-reported quality of life of patients with metastatic colorectal cancer to clinical data of a German tumor registryOBJECTIVES Approximately one out of eight cancers in Germany affects the bowel and more than 6% of the population are diagnosed with colorectal carcinoma during their lifetime. The EuroQoL EQ-5D ques-tionnaire is widely used in oncology to generate quality of life (QoL) weights and corresponding health states. Aim of this study is to generate EQ-5D utilities by linking clinical data of a German colorectal cancer registry to self-reported QoL measures from the EQ-5D-5L to explore the relation-ship between disease-specific health states and health-related QoL (hrQoL). METHODSThe study sample included metastatic colorectal cancer patients currently recruited in the German Tumor Registry Colorectal Cancer. The EQ-5D-5L was administered once as paper version to pa-tients who were at the start or at later stages of treatment. Potentially relevant comorbidities, disease-specific health states (e.g. metastases) and symptoms (e.g. nausea), and treatment status weredefined by literature review and medical experts. Data was drawn from the clinical registry and the EORTC QLQ-C30 questionnaire. Multivariate regression models will be calculated to explore the health state specific and comorbidity dependent decrements on QoL. RESULTSIn total, n=758 questionnaires were sent to patients, n=535 were returned, and n=503 were finally included in the data analysis. Mean age was 66.76 years and 62.23% were male. 70.68% of patientshad at least one comorbidity and the most frequent comorbidity was hypertension (42.37%). The overall mean hrQoL based on EQ-5D-5L for patients with metastatic colorectal cancer was 62.12with the Visual Analog Scale (VAS).CONCLUSIONSThis pilot study linking clinical registry data to hrQoL data shows a new opportunity for a cross-sectional study design. The implementation of EQ-5D-5L in metastatic colorectal cancer patients showed reduced hrQoL compared to the general population (mean VAS 85.15). Results from the regression analyses will be presented.Wolfgang GreinerPopulations and Health Systems51575Completed2016
20180230A fast-track proposal: write and publish a paper comparing DCE data from 11 Asian EQ-5D-5L valuation studiesSo far, 11Asian regions(China, Indonesia, Japan, South Korea, Malaysia, Singapore, Thailand, Philippine, Hong Kong, Vietnam, and Taiwan)have conducted their EQ-5D-5L valuation studies following EQ-VT protocol. Currently, there is no research comparing health preference focusing on Asian population. Notably, publishedresearchesaimedto compare the difference in the value set. A downside of directly comparing value set is that the differences identified may stem from 1) preference heterogeneity and 2) methodological differencesand it is difficult to detangle such differences .Despite the fact that all EQ-5D-5L valuation studies followed the EQ-VT protocol, different versions of EQ-VT protocol were developed to improve the quality of data collected using TTO method. Nevertheless,the design of DCE stayed intactfor all these valuation studies. Additionally, the DCE task was performed after the TTO task, which ensured that respondent already familiar with the EQ-5D-5L health description. Therefore, the highly standardized DCE data collection protocol offered a unique opportunity to study the health preference between countries.In this study, we utilize the DCE dataset of eleven Asian regions:first, each region’s data set wasmodelledusing a conditional logit main effect model. For comparison, we createddummy variable to represent different countries andmodelledtwo countries data together with the country variable as an interactionterm to test the differenceof the main effect variables. Next, Wecompared the relative importance of dimensions and relative utility decrements between each of the 5 levels for each country.Zhihao YangValuation6720Completed2018
2015460The Relative Value of Social Outcomes in Health Technology AssessmentA key outcomein the evaluationof health technologies is the quality adjusted life year (QALY)which is oftenestimatedusing health measuressuch as EQ-5D. However, the impact of many interventions extends beyond a narrow definition of health to include non-health impactssuch associal care. This means that there are circumstances where the QALY doesnot capture the full value of an intervention.In response to this,instruments with a broaderperspective such as ASCOT, which measures social care related quality of life, havebeen developed. Given the growthin available measures, it is important that decision-makers have tools to assess value for money consistently. However, eliciting preferences for different aspects of QoL,to allow for the combined benefits to be captured within the same framework,has not been tested. We investigate the relationship between health and social care aspects of QoL when assessed jointly by combining theEQ-5D-5L and ASCOTin an online discrete choice experiment (DCE). Each respondent completed 15DCEtasks from an underlying design of 300 choice sets. Analysis used conditional logit regression to estimate coefficient decrements for each attribute and examine theirrelative importance. Latent class and mixed logit modelling were used to understand heterogeneity in preferences.Overall 975 people completed the survey. The results suggest evidence oftrading acrosshealth and social careaspects, as indicated by differences in the magnitude of the coefficients across the different aspectsincluded. There is evidence of preference heterogeneityacross the dimensions, particularly at the more severe levels.We have used an established valuation methodology to demonstrate that it is possible to value concepts measuring different aspects of quality of life on the same underlying scale. The results inform further valuation work eliciting preferences for instrumentscombining outcomes from multiple perspectives.Brendan MulhernOthers14680Completed2015
20190910Case-mix adjustment of EQ-5D health profiles for the purpose of hospital performance assessmentThere is increasing interest in using routinely collected EQ-5D data to compare the performance of healthcare providers in terms of their patients’improvementsin health-related quality of life. An essential condition forsuch comparisons to be valid is that they shouldadjust for differences between providers in thecharacteristics of thepatients that they treat; a process known as case-mix adjustment. The prevalentapproach is to estimate a case-mix adjustment model that relates patient and provider characteristics directly toEQ-5D valuesand to use thisto generate provider-specific outcomesin terms of changes in adjusted EQ-5D values. However, this approach haslimitations. For example,it does not make full use of the information provided by the underlying EQ-5D profile data,and it generates results that are specific to the value set used. We propose an alternative approach that will generatefor providers the mix of EQ-5D profilesadjusted for patient and provider characteristics, with associated probabilities. These can then be used to generate EQ-5D valuesor analysed in profile formfor use in performance comparisons. This approach relies on predicting dimension-specific case-mix adjusted responses using a multivariate regression framework. In this project we will use a simulation study to investigate the statisticalproperties of both methods, how well they can identify performance variation, and which context-specific factors should guide the choice of methodology in applied research. We will follow this up with real-world case studies to establish feasibility of the proposed method in commonly-encountered performance assessment settings.Nils GutackerPopulations and Health Systems34660Completed2020
2013190Overview of psychometric properties of EQ-5D in a range of conditionsAims: EQ-5D has been widely used in many different health conditions and the amount of evidence of its psychometric properties has increased over recent years. The primary aim of this study was to review papers reporting systematic reviews of the validity, reliability and/or responsiveness of EQ-5D. A secondary aim was to review other evidence about the extent to which EQ-5D covers the dimensions of health affectedby various health conditions.Methods: Medline and Embase were searched for systematic reviews of the performance of EQ-5D. Supplementarysearches were carried out in Cochrane Library, Web of Science, reference lists of included studies, the EuroQol databaseand hand searching of EuroQol Scientific Plenary Proceedings. In addition the website of the Oxford Patient Reported Outcome Measures (PROMs) Group was searched for reports. Papers providing an overview of evidence relating to the validity, reliability and/or responsiveness of EQ-5D based on a systematic review were included. Data were extracted using a template designed specifically for the study.Results: 25reviewswere identified in this studyand a further 18reviewswere identified from the Oxford PROMgroupwebsite. The papers covered a range of conditions. Overall there was evidence of good to fair performance of EQ-5D in depression, diabetes (type 2), rheumatoid arthritis, skin conditions, cancer, cardiovascular disease, asthma, personality disorder and urinary incontinence. The evidence was mixed in COPD, dementia, schizophrenia and vision disorders, and poor for hearing disorders. The was little or no evidence for liver transplantation, venous leg ulcers, haemophilia, skin cancer, systemic lupus erythematous, bipolar disorderand low back pain; although the limited evidence showed positive results for liver transplantation, haemophilia and leg ulcers.Conclusions:This study has identified several reviews of the performance of EQ-5D. Most of the evidence suggests good psychometric properties of EQ-5D; however there are particular concerns about its ability to capture the impact of dementia, schizophrenia, visual impairment and hearing disorders. Further research is encouraged to: (i) review the literature for areas where reviews have not yet been published; (ii) conduct more primary research into the psychometric properties of EQ-5D where data are lacking; and (iii) further explore the use of bolt-ons where performance of EQ-5D is poor.Louise LongworthValuation5800Completed2013
2014210Health-related quality of life and perceived burden of informal caregivers of patients with rare diseases in EuropePurposeToevaluate whether the EQ-5D discriminates HRQOL between caregivers’ burden levels, and to examine which caregivers and patients characteristics significantly relate to caregivers’ HRQOL.MethodsIn this study, participants were long-term informal caregivers ofpersonssuffering rare diseases (RD) in several Europeancountriesincluded in the BURQOL-RD study. Besides descriptive statistics, correlation analysis to examinethe relationshipsbetween caregiver HRQOL and self-perceived burden of caring measured by the Zarit scale. Multinomial logistic regression models were used to explore the role of explanatory variables on each domain of caregivers HRQOL measured by EQ-5D.ResultsWe obtained dataon self-perceived burden of caring and HRQOLfrom 825 caregiversincluded in the BURQOL-RD database. CaregiversHRQOL is inversely correlated with burdenof caring(Pearson correlation coefficient=-0.180; P-value < 0.0001). Mobility dimensionwas significantly associated with patientsage, time devoted to care by secondarycaregivers, patient genderand patient utility score.Duration of caring of0 to 5 years are 2.7 times more likely of reporting‘slight problems’ than ‘no problems’in self-care dimension. Patients ́age, burden scoresand patient utilityscoresignificantlypredict the capacityof caregivers to performactivities of daily living(usual activities dimension).Employedcaregiversare lesslikely of reporting ‘slight problems’in pain/discomfort dimensionthan unemployed caregivers. Furthermore, patient age and patient utility score explain differences in thisdimension. ConclusionsThe EQ-5D instrument is sensitive to measure differences in HRQOL betweencaregivers with different levels of burden of care. Patient utility score is strongly associated with all dimensions of EQ-5D instrument exceptanxiety/depression in which perceived care burden plays an important role.Juan M. Ramos-GoñiValuation7875Completed2014
122-RAAn exploratory study on the use of the recall period and the impact of different formats in two fluctuating conditionsThe EQ-5D uses the recall period “today”. Other HRQoL measures use different recall periods e.g. 7 days, presented in different formats e.g. bold lettering and in different locations e.g. in the item labels. EQ research has focused on assessing the impact of different recall periods in different conditions. However, qualitative studies suggest respondents might often fail to account for the recall period when self-reporting their health. It is also not clear whether using different formats improves respondents use of recall periods. This study will examine these two questions in an online sample of arthritis and migraine patients. Participants will be randomized to one of 3 blocks. Participants in Block 1 will receive EQ-5D-5L versions with recall period “today” and “the last 2 weeks”, presented in the standard format. Participants in Block 2 will receive EQ-5D-5L versions with recall period “today” and “the last 2 weeks” presented in bigger, bold lettering. Participants in Block 3 will receive EQ-5D-5L versions with recall period “today” and “the last 2 weeks”, with the recall period located in the item labels. A VAS exercise will be used to investigate the degree of fluctuation of the condition in the last 2 weeks and cognitive debriefing questions will examine participants’ thought processes when answering the EQ-5D-5L with different recall periods. Paired t-tests will be used to analyse differences between recall period versions overall and Wilcoxon signed ranked test to compare at the dimensions level. Results will be interpreted based on responders health fluctuations and cognitive debriefing questions.Aureliano Paolo FinchDescriptive Systems26766Completed2020
2016421The follow up meeting with Indonesian Ministry of Health and HTA Committee: EQ-5D inHTA and non-HTA research and the EQ-5D-5L value setThis year in the value set of the Indonesia 5 level version of the EQ-5D will be published in scientific journal. We will also publish the population norm scores. Thearrival of the Indonesia 5 level value set and norm scores comes timely, as Indonesia has started in 2014 a societalhealth insurance system, and (cost-)effectiveness is determined as one of the leading principles for reimbursement. In September 2016, we organized a symposium to introduce this value set to the stakeholders. Policy makers from Ministry of Health(MoH), National Health Insurance, and Indonesian Food and Drug Administration were present together with researchers from various universities in Indonesia. In collaboration with The Center of Health Financing and Security (Pusat Pembiayaan dan Jaminan Kesehatan/PPJK) MoH, we manage to secure a follow up meeting with the Indonesian HTA committee and higher level officials of MoH in Jakartain March 2017. The aimsof thisfollow up meeting areto promote EQ-5D-5L as recommended tool in the Indonesian HTA guideline and to introduce EQ-5D usefulness in theHTA and non-HTA research.Jan BusschbachOthers0Completed2017
20170560Framing effects when valuing EQ-5D-Y health states in a latent scale DCEnot availableRichard NormanValuation, Youth37576Completed2018
20170630Understanding Self-perceived Health in Latin AmericaBackground:Some studies depict the relationships between self-reported health socio-demographic factors and clinical conditions in the general population. The relationship between self-reported health with broader life domains, like the perception of law and order, corruption, civic engagement, and others has been less studied. In Latin America and the Caribbean, notably, there have been very few studies looking into these issues. The Gallup World Poll is a survey conducted periodically in more than 160 countries around the world. In the 2007 and 2008 version, the poll included 130 countries (20 from LAC) and self-reported health using EQ-5D-3L instrument. Clustering is a term used to refer to a setoftechniques that find subgroups in a data set, such that observations within each subgroup are as similar to each other as possible and as different as possible to observations in other subgroups. “Similarity” between observation is determined by the selection of a similarity/dissimilarity measure. The choice of the appropriate measure is made based on the type of data and knowledge of the data being analyzed. After choosing the dissimilarity measure, the clustering algorithm must be selected. For our study, we chose partitioning algorithms which include the K-means clustering algorithm and the Partitioning Around Medoids (PAM) algorithm. In PAM clustering, each cluster is associated with a medoid, the object with the smallest dissimilarity to all other objects in the cluster, and is named after it. Objectives:a) Sociodemographic segmentation: Find subgroups of respondents determined by patterns of sociodemographic variables and assess if belonging to such subgroups predicts a difference in the EQ-VAS score.b) EQ-5D segmentation: Find subgroups determined by response to all items of EuroQol combined with age, education and income.Methods:The Gallup World Poll is a Survey conducted in over 150 countries, with randomly selected, nationally representative adult samples using a set of core questions. Most items havea dichotomous response set and are organized into22indexes.Survey samples are nationally representative of adults over 15 years old. Survey data is weighted to account for sampling, design effects, and margin of error.We eliminated cases not corresponding to one of the following Latin-Americancountries. 1. Argentina2. Belize3. Bolivia4. Brazil 5. Chile6. Colombia7. Costa Rica8. Dominican Republic9. Ecuador10. El Salvador11. Guatemala12. Guyana13. Honduras14. Mexico15. Nicaragua16. Panama17. Paraguay18. Peru19. Trinidad and Tobago20. Uruguay21. VenezuelaWe chose to work with the R statistical package and to limit the amount of computing power necessary for clustering algorithms, we chose to work with data for the year 2008. We created two datasets, one for each previously specified objective. For sociodemographic segmentation, we used the22indexes that summarize information from different domains, and we created a dissimilarity matrix using Euclidean distance, given that all variables were numeric. For EQ-5D segmentation we used the following variables:• Household Income per capita (International Dollars) • Age• Own Health Today –Visual analog scale coded as a continuous variable with a range from 0 to 10. • Mobility• Self-Care• Usual activities• Pain/Discomfort• Anxiety/Depression• EducationSince the data included numeric and categorical variables, we generated a dissimilarity matrix using Gower distance.We later implemented a PAM algorithm, accounting for weighted observations for both objectives. We used weightedaverage Silhouette to determine the best way to cluster the data. This process was applied to both distance matrices. In both cases, a range of 2 to 10 clusters was explored. Once every observation was assigned to a cluster through the described methods,we assessed whether belonging to a specific cluster predicted a change in EQ-5D-VAS score through linear regression. We included the EQ-5D-VAS as an outcome variable, cluster as a categorical variable and age and education as covariables.Results:After selecting the variables to be analyzed and restricting the observations to those recorded in 2008, both datasets had 18,609 observations from 19 (Belize and Guyana were not included in 2008).For sociodemographic segmentation, average Silhouette was highestfor two clusters. Clustering produced clusters 7365and 18131. Cluster 18131concentrated 32,07% of observations and had better(statistically significant) values in all indicators included in the analysis. These clusters were significantly associated with a difference in the value of self-rated health on EuroQol. Belonging to cluster 18131was associated with an average increaseof 0.8994points in the EQ-VAS 0-10 scale, holdingsex,education and age constant, compared to belonging to cluster 7365.Using the criteria of the highest average Silhouette, we selected 8clusters as the best way to group the data for the EQ-5D-Based segmentation. Four clusters were predominantly male and four were predominantly female respondents. Clusters 15706 and 2016 gathered most of the respondents with no problems in any of the five domains and had the higher VAS score in males and females, respectively.We found a group of predominantly males with high EQ-VAS score that reported moderate anxiety depression.Discussion:In our study we explored different segmentation tools in a large Data set that includes data of self-reported health and other aspects from 19 LAC countries. These tools are increasingly used in healthcare to tailor clinical and public health interventions to heterogeneous populations In our first analysis, we showed thataclustering algorithm was able to identify a subgroup of observations defined by favorable values in the Index variables, andthatthese observations also correlated with higher values in the EQ-VAS. In our second analysis, we identifiedinteresting patterns defined by the five domains and VAS score of the EQ-5Daccording to sex. Our findings show that clustering analyses arenot only useful but are also versatile techniquesthat can be applied to the study of health-related quality of life and help to untangle the differences that lie within different cultural and social subgroups.victor zaratePopulations and Health Systems89890Completed2018
239-RAA review of the methodologies implemented in the valuation of generic and condition-specific measures of child and adolescent health statesThis research seeks to provide an up-to-date review of the methods currently being used to value child/adolescent health states. Primary objective: • To perform a comprehensive review of the methodologies implemented in the valuation of preference-based, generic and condition-specific measures of child/adolescent health Secondary objective: • To review the extent to which debrief and follow-up questions are used to understand respondents’ valuations of child/adolescent health states Strategic scoping searches will be run in the Embase and Medline electronic databases to identify valuation studies of generic and disease-specific measures of child and adolescent health. General hand-searching will also be performed to identify further studies of interest to ensure the completeness of the review. Relevant studies will be identified and selected for full-text review and data extraction. In particular, we are interested in extracting information regarding who valued the child/adolescent health states, the perspectives through which they were valued, how their preferences were elicited, and the age(s) of the children/adolescents whose health was valued. Measures will be taken to ensure the rigour of the review at each stage of the study. This research will provide the EuroQol Group with an up-to-date picture of the methodologies that have been implemented in the valuation of generic and disease-specific measures of child and adolescent health states, the results of which will inform future versions of the EQ-5D-Y valuation protocol.Andrew LennyValuation, Youth14840Completed2021
102-RAThe impact of COVID-19 on health status, based on the EQ-5D-5L, of adults visiting emergency departments and primary care clinics in Alberta, Canada**Background**: COVID-19 is placing a significant burden on individuals’ physical, mental, and social functioning. As such, it is imperative to explore the impact of this pandemic on the health-related quality of life (HRQL) of populations to inform public health policies and strategies. **Aims**: This proposed study aims to - Examine the impact of COVID-19 on health-related quality of life (HRQL), including physical and mental aspects of health, as measured by the EQ-5D-5L, of adults visiting emergency departments and primary care clinics in the province of Alberta. - Explore whether the impact of COVID-19 is disproportionate in certain groups of the population based on socio-demographic data and deprivation, and as such, identify the most adversely affected groups in this population. **Methods**: We will use data from two on-going large-scale surveys from health service users that have been conducted by the Health Quality Council of Alberta for several years; as such, these surveys will provide data “before”, “during”, and “after” COVID-19. EQ-5D-5L dimension level data and EQ-index and VAS scores before, during and after COVID-19 will be examined in the overall samples, and by population sub-groups to identify the most affected groups of the patient population. **Implications**: This study will provide evidence on the impact of COVID-19 on HRQL trends in adults with various health conditions. Quantifying this impact and identifying the most affected groups in the population will inform public health policies and strategies to alleviate the detrimental consequences of COVID-19 on population health status.Fatima Al SayahPopulations and Health Systems31955Completed2020
20190860Think of the children: rationale for and implications of the perspective in EQ-5D-Y-3L health state valuation•The EQ-5D-Y facilitates the description and measurement of child and adolescent health-related quality of life (HRQOL) in economic evaluations of health technologies. The recently published standardized valuation protocol prescribes the use of a child perspective, rather than the individual perspective used for valuation of adult EQ-5D health states. An EQ-5D-Y value set is not yet available; however, the limited evidence so far indicates that a child perspective yields different utilities for similar EQ-5D health states. •The choice for and implications of using a child perspective for EQ-5D-Y health state valuation has received little attention. As it is important to understand the reasons for this change and what influence a different perspective has, the objective of this paper is to address this gap in the literature and discuss the rationale for using a child perspective, in comparison to using the conventional individual perspective, and the normative and practical implications of a child perspective for EQ-5D-Y value sets and estimation of HRQOL gains in children and adolescents in economic evaluations. •Although the use of an individual perspective for valuation of EQ-5D-Y health states is possible, it may not be suitable or desirable. However, the discussion of several implications of using a child perspective makes it clear that using this perspective may impact (the development of) EQ-5D-Y value sets and ultimately also reimbursement decisions in healthcare. Further research is necessary for gaining insight into the extent to which this impact is (not) empirically and normatively justified.Stefan LipmanYouth19700Completed2020
20190850Validation of the EQ-5D plus respiratory bolt-on in the birmingham copd cohort studyObjectives: Arespiratory bolt-on dimension for the EQ-5D-5Lhas recently been developed and valued by the general public. Thiscurrent study aimed to validate the EQ-5D-5L plusrespiratory dimension (EQ-5D-5L+R) in a large group of COPD patients. Methods: Validation was undertaken with data from the Birmingham COPD cohort study (BCCS), alongitudinal UK study of COPDprimary care patients. Data on the EQ-5D-5L+R were collected from 1,008 responding participants during a follow-up questionnairein 2017 and combined with (previously collected)data on patient and disease characteristics. Descriptive and correlationanalyses were performed on the EQ-5D-5L+Rdimensions and utilities,inrelation toCOPD characteristicsand compared to the EQ-5D-5L without respiratory dimension. Multivariate regression models wereestimated to test whether regression coefficients ofclinical characteristics differed between the EQ-5D-5L+Rutilityandthe EQ-5D-5L utility. Results:Correlation coefficients for the EQ-5D-5L+Rutility with COPD parameters were slightly highercompared to the EQ-5D-5Lutility. Both instruments displayed discriminant validitybut analyses in clinical subgroups of patients showed larger absolute differences in utilitiesfor the EQ-5D-5L+R. In the multivariate analyses,only the coefficientfor the CAT score was higher for the model using the EQ-5D-5L+R utilityas outcome. Conclusions:This study showed that the addition of a respiratory domain to the EQ-5D-5L led to small improvementsin the instrument’s performance. Comparability of the EQ-5D across diseases, currently considered one of its strengths, would have to be traded-off against a modest improvement in utilitydifference when adding the respiratory dimension.Martine HoogendoornDescriptive Systems14700Completed2019
2016020EQ DCE: Crowdsourcing innovation in valuation specificationABSTRACT Background: Health is not bought and sold openly. In order to understand the valueofits attributes, health preference researchers conductsurveys using paired comparisons (“Which do you prefer?”). In the absence of external data with which to validate their findings, resulting models, often based on conventional practices and assumptions,may be suboptimal in terms of predictive validity.Aims: To engage a large group of scientists in order to identify specifications and approaches to model selection that better predict health preferences. Methods: Announced in March 2016, the competition attracted 18 teams from around the world. In April,an exploratory survey was fielded:4074 US respondents completed 20 out of 1560 paired comparisons by choosing between two health descriptions (e.g., longer lifespan vs better health). The paired comparison data were distributed to all teams. By July, eight teams had submitted their predictions for 1600 additional pairs and described their analytical approach. After these predictions had been posted online, a confirmatory study was fielded (4148additional respondents).Results: In September, the victorious team, “Discreetly Charming Econometricians” led by Michał Jakubczyk, was announced at the EuroQol Plenary. This team achieved the smallest chi square 4391.54(a predefined criterion)usingtwo sub-models: one comparinga health descriptionvs immediate death and the other comparing twohealth descriptions.Conclusions: We demonstratedthe diversity of analytical approaches and highlighted the importance of predictive validity in health preference research. Although the generalizability to specific applicationsis unknown, the models that are more predictive, even if difficult to be immediately applied in health technology assessment (e.g. may not yield utilities directly assigned to specific health states), may inspireand informdebate among health preference researchers.Benjamin CraigValuation123632Completed2016
20190840Aversion to inequalities in health by EQ-5D domainBACKGROUND: On average, poorer people in the UK live shorter, and less healthy, lives than richer people. Evidence suggests the UK-public don’t like this inequality, and that they are willing to prioritise gains in life-expectancy to poorer people over richer people. Nobody has tested whether they are equally willing to prioritise poorer people for interventions that improve quality-of-life. If we want to include “inequality aversion”in cost-effectiveness analyses, it would be useful to know whether or not this is the case. We hypothesise that the UK-public are more willing to prioritise poorer people for life-expectancy gains than quality-of-life gains. METHODS: We fielded an online person-trade-off (PTO) study in 1,502 members of the UK-public. Participants were briefed on inequalities in health between socioeconomic groups, and then randomised to PTO exercises about one of ten health-gain types (nine forms of quality-of-life and life-expectancy). In each PTO, participants made choices between interventions that would benefit different numbers of people from either the poorest or richest fifth of society. We logically varied the number of people in each group in response to each participant’s response, in order to determine the relative priority they placed on improving the health of each group. Every participant completed two PTO exercises: one for a 3 QALY gain and one for a 0.5 QALY gain. We tested our hypothesis by comparing participant responses in the life-expectancy arm and each quality-of-life arms. This was done separately for the the two gain-magnitudes. RESULTS: The UK-public are more willing to prioritise poorer people for interventions that improve lifeAki TsuchiyaOthers6678Completed2019
2016170A systematic review of the measurement properties of EQ-5D-5LBackgroundSince the introduction of the five-level ver-sion of the EQ-5D (5L), many studies have comparativelyinvestigated the measurement properties of the originalthree-level version (3L) with the 5L version.ObjectiveThe aim of this study was to consolidate theavailable evidence on the performance of both instruments.MethodsA systematic literature search of studies in theEnglish and German languages was conducted (2007–January 2018) using the PubMed, EMBASE, and Psy-cINFO (EBSCO) databases, as well as the EuroQolResearch Foundation website. Data were extracted andassessed on missing values, distributional properties,informativity indices (Shannon’s H0and J0), inconsisten-cies, responsiveness, and test–retest reliability.ResultsTwenty-four studies were included in the review.Missing values and floor effects (percentage reporting theworst health state) were found to be negligible for both 3Land 5L (\5%). From 18 studies, inconsistencies rangedfrom 0 to 10.6%, although they were generally well below5%, with 9 studies reporting the most inconsistencies forUsual Activities (mean percentage 4.1%). Shannon’sindices were always higher for 5L than for 3L, and all butthree studies reported lower ceiling effects (‘11111’) for 5Lthan for 3L. There is mixed and insufficient evidence onresponsiveness and test–retest reliability, although resultson index values showed better performance for 5L on test–retest reliability.ConclusionOverall, studies showed similar or bettermeasurement properties of the 5L compared with the 3L,and evidence indicated moderately better distributionalparameters and substantial improvement in informativityfor the 5L compared with the 3L. Insufficient evidence onresponsiveness and test–retest reliability implies furtherresearch is needed.Ines BuchholzDescriptive Systems21356Completed2016
20170430Comparison of 3L and 5L health profiles and valuation in patients with ischemic heart diseasesBackground EQ‐5D‐5L is developed from EQ‐5D‐3L to mitigate ceiling effects and improve discriminativeability. Value sets for EQ‐5D‐5L are also established using a new composite time trade‐off method (cTTO) with or without discrete choice experiment. It is not entirely clear whether or how the EQ‐5D‐5L index is more sensitive than the EQ‐5D‐3L index. Regardless of their comparative sensitivity,how the descriptive system, 3L or 5L, impacts on the sensitivity of the index score cannot be assessed with existing value sets because those we redeveloped using different valuation methods, interviewers, and respondents. Study aims (i) To compare EQ‐5D‐5L and EQ‐5D‐3L items in heart disease patients; (ii) to compare EQ‐5D‐5L and EQ‐5D‐3L in dicesusing existing value sets; (iii) to explore the effect of descriptive system on the sensitivity of EQ‐5D indices. Methods The study will piggy back on a study of 500 heart disease patients. Patients will report their health using both EQ‐5D‐5L and EQ‐5D‐3L questionnaires. In addition, patients with be asked to valueten EQ‐5D‐5L and six EQ‐5D‐3L health states, in a random order, using the cTTO tasks administered by EQ‐VT. Differences in self‐reported EQ‐5D‐5L and EQ‐5D‐3L profiles and indices will be assessed by examining their association with each other and clinical variables using appropriate statistical models. Existing 3L and 5L value sets including those derived from the general UK population will be used. Experimental 5L and 3L value sets based on the preferences of the heart disease patients in this study will be developed in order to study the effect of the 3L and 5L descriptive systems on the sensitivity of the utility‐based indices.Nan LuoDescriptive Systems10000Completed2017
2016300Methods for development of a generic descriptive systemBackground: Hundreds of standardised instruments have been developed to measure health-related quality of life and health status, with a range of approaches used in the development. The EuroQol Group are currently exploring the potential for developing a new descriptive system. A key part of the process of this development will require the selection of items for inclusion in the instrument. Various methods for item selection are available and the Group will need to decide which to adopt in the development of any future EQ instrument. The aim of this study was to provide a contemporary overview of methods recommended in guidelines for item selection in instrument development and methods used by instrument developers. Method:A focused review of key standards and guidelines was used to identify recommended methods and psychometric evaluations. A systematic review of the literature was conducted to identify the methods actually used in empirical studies reporting development processes of descriptive systems and any additional methods not included in the guidelines. The search was conducted using COSMIN’s precise search filter for measurement properties, restricted to include health related quality of life studies and focusing on development of descriptive systems, and a total of 553 abstracts were identified. Supplementary searches were conducted by reviewing bibliographies and citations of the included studies to identify complete set of articles on development of the descriptive system where possible.Results: A total ofeleven guidance documents were consulted and a data extraction template was created. A total of 61 full text articles were included. Item generation involved three approaches: developing items de novo, identifying items from several existing measures or using a single measure to derive a short form measure. Item selection was determined by the instrument developers in few cases, but most reported use of empirical work. The guidelines describe two development phases after item generation: pretesting and field-testing of questionnaire but this was not always followed. Some studies determined items based on relevance and importance, assessed from interviews and rating studies. The most common approach was to apply statistical analyses to datasets containing completed questionnaires. Three general types of statistical models were applied: regression analysis, factor analytic techniques and item response theory models. These measurement models were used to establish construct validity, however in most cases further hypothesis testing was carried out to assess known group, convergent and divergent validity. The majority used a combination of statistical methods; with exploratory factor analysis and item response theory being the most common. Psychometric criteriaincluded descriptive statistics of the responses such as missing values, range and distribution of responses and internal consistency.Discussion: The methods used to identify items were similar across studies but differed considerably for item selection.Construct validity and internal consistency was assessed in most of the studies, but few instrument developers relied completely on content validity. Few of the statistical methods identified are interchangeable in terms of purpose, but may generate very different instruments. The developers seldom explained why one statistical model was chosen over anotherLouise LongworthDescriptive Systems19500Completed2016
2015070Comparing the predictive accuracy of different main-effects regression models on left-out EQ-5D-5L health states: 20-parameter additive model vs. 8, 9, and 11 parameter multiplicative models.BackgroundThe conventional method for modeling of EQ-5D-5L health state values in national valuation studies is an additive20-parameter main-effect regression model. Statistical models with many parameters are at increased risk of overfitting; fitting to noise and measurement error, rather than the underlying relationship. ObjectiveTo compare the 20-parameter main-effect model to simplified,non-linear,multiplicative regression models in terms of how accurately they predict mean values of out-of-sample health states.MethodsWe used data from the Spanish, Singaporean, and Chinese EQ-5D-5L valuation studies. Four models were compared: an 8-parameter model with single parameters per dimension, multiplied by cross-dimensional parameters for levels 2, 3, and 4; 9 and 11-parameter extensions with handling of differences in the wording of level 5;and the“standard” additive 20-parameter model. Fixed and random intercept variants of all models were tested using two cross-validation methods: leave-one-out at the level of valued health states, andof health state blocks used in EQ-5D-5L valuation studies. Mean absolute error, Lin’s Concordance Correlation Coefficient, and Pearson’s R between observed health state means and out-of-sample predictionswere compared.ResultsPredictive accuracy was generally bestusing random intercepts.The 8, 9, and 11-parameter modelsoutperformed the 20-parameter model in predictingout-of-sample health states.Discussion and conclusionSimplified non-linear regression models look promising, and should be investigated further using other EQ-5D-5L datasets.To avoid overfitting, cross-validation is recommended in model selection in future EQ-5D valuation studies.Kim RandValuation8500Completed2015
2014120LSHA WorkshopPaul KindPopulations and Health Systems10000Completed2014
2015160Relationship between behavioural risk factors for poor health and the EQ-5D: Prospective analyses in a New Zealand cohortIntroduction: Physical inactivity, tobacco smoking, high body mass index and hazardous alcohol consumption are risk factors for poor health. However, their relationship with the EQ-5D, a commonly used measure of general health status, is poorly understood. This study examines the relationship between these factors, both individually and in combination, with the EQ-5D-3L. Methods: Data from the Prospective Outcomes of Injury Study (n=2856) was analysed using multi-variable logistic regression modelling to examine pre-injury relationships between physical activity, tobacco smoking, body mass index and alcohol consumption and each of the five EQ-5D dimensions. Logistic regression modelling was also used to calculate the predicted probability of problems within each of the EQ-5D dimensions for each possible combination of the four behavioural risk factors (i.e. 16 possible combinations were examined for each of the five dimensions). Graphs were produced to show the 16 combinations of risk factors with each of the dimensions. This appears to be a novel approach in terms of presenting EQ-5D findings. Results: Physical inactivity was associated with being more likely to have problems with mobility. Tobacco smoking was associated with being less likely to have problems with pain or discomfort whereas high alcohol consumption was associated with being more likely to have problems with pain or discomfort. Body mass index was not associated with any of the EQ-5D dimensions. However, the combination of obesity and physical inactivity (with or without the other behavioural risk factors) corresponded to a larger probability of problems in almost all dimensions. Conclusions: Our results support earlier findings that large scale population studies should consider collecting information about key behavioural factors alongside EQ-5D health outcomes. It also appears useful to consider specific combinations of behavioural risk factors when assessing general health status using the EQ-5D.Helen HarcombePopulations and Health Systems14905Completed2015
152-RAHead-to-head comparison of the EQ-5D-3L and EQ-5D-5L: Psychometric properties in Dementia**Background:** Several studies demonstrated the advantages of the EQ-5D-5L over its 3-level version as a preference-based summary measure of health-related quality of life (HRQoL). However, assessing HRQoL in cognitively impaired patients poses several challenges due to the missing intellectual capacity, which could be much more challenging using five instead of three levels. **Aim of research:** To assess the psychometric properties of both self reported and proxy-reported EQ-5D-5L in comparison to the EQ-5D-3L in cognitively impaired patients living with dementia. **Proposed methods:** The analysis will be based on an already ongoing interventional study conducted in German primary care. EQ-5D-3L, EQ-5D-5L, and Quality of Life in Alzheimer´s Diseases are assessed via interviews as self- and proxy-ratings by caregivers. The EQ-5D-3L and 5L will be evaluated in terms of acceptability, agreement, ceiling effects, redistribution properties, inconsistency, informativity (Shannon and Shannon Evenness indices), convergent and discriminative validity, and responsiveness. Measurement probabilities will be assessed in consideration of the mode of administration (self vs. proxy-ratings) and level of cognitive impairment. **Benefits to the group:** The study will determine if the EQ-5D-5L is preferable to the three-level version as a self- and/ or proxy-rating for cognitively impaired patients. It provides important empirical evidence about the comparative performance of 5L vs 3L.Bernhard MichalowskyDescriptive Systems20840Completed2020
2015190No kidding: the validity of discrete choice experiments in children - a pilot studyObjective: To assess the feasibility and validity of using adiscrete choice experiment format to elicit health preferences in children by comparing illogical choices and choice blindness rates between adults and children; and toexplore the relationship between personality traitsand health-state choices.Methods: A convenience sample of adults and children (8-17 years old) were recruited aroundChicago, USA. Apersonality inventorywas administered,followed by pairwise comparisons of6health-state scenarios. Health statedescriptionswere based ona simplified 3 dimension version of EQ-5D (mobility, pain, depression, each with 3 levels). For 2 scenarios, theinterviewer switchedthe respondent’s preferred choiceto identify choice blindness. Odds of choice blindnessand health-state choice among pairswas assessed using logistic regression.Results: One-hundred and onerespondents were recruited (44% adults). Comparing children to adults, there was no significant difference in the rate of illogical preferences, i.e. selecting a dominated state, (9% vs. 12%) or in preferring dead to the worst health-state (54% vs. 64%) (p>0.05). Choice blindness rates were significantly higherin children (35%) thanadults (9%) (p<0.01). Adjusting for gender and personality traits, the odds of choice blindness in children was 6.6(CI 95%: 1.8-23.8, p=0.004). Conscientiousness was significantly associated with health-state choice in 3 of the 6 models predicting health state choice (using p<0.1as a threshold).Conclusions: The results of this exploratory study suggest it is feasible to conduct choice experimentsin children; however, children are significantly more likely to demonstrate choice blindness.Psychological traits may be noteworthy predictors of health-state choices, withConscientiousness independently associated with severalhealth-state choices.Simon PickardYouth13500Completed2015
20180640Insight into the higher health state valuation for children compared to adults: effect of 3 valuation methodsINTRODUCTION: Available evidence comparing adult versus child stated preferences consistently report higher utility values for children, using the VAS, composite-TTO, DCE-with-death, DCE-with-duration, lag-time-TTO, and location-of-dead-approach, than for adult healthstates (HS). Explanations for these findings have been proposed but not verified.AIM: The aim of this research is to confirm the higher valuation for child compared to adult HS; and to investigate why respondents value child HS differently.METHODS: Eightygeneral public respondents from the UK, Belgium and The Netherlands were invited for a 1.5-hour face-to-face interview. Respondents valued four EQ-5D HS from two perspectives (8-year-old child, 40-year-old adult) using VAS and TTO. Thirty-two of the respondents participated in think-aloud interviewing. Quantitative analyses were carried out in SAS; audio-recorded interviews were transcribed,and a thematic analysis was conducted in NVIVO. Interviews were coded by two researchers using presumption-focused coding. Statements, nodes and themes were reviewed cyclically until consensus was reached within the research team. The aim was to obtain a framework with a limited number of sub-themes(nodes)and uniform statements within each node. A conceptual frameworkwasdevelopeddepicting the relationships between nodes, themes and the life years (LY) trade-off. Qualitative statements made by respondents were contrasted with their quantitative responses. RESULTS: Quantitative results: Statistically significantlyhigher utilitieswere found forchild versusadult TTO values: 0.026, 0.112, 0.377 and 0.294 higher utilities for mild, moderate, severe and worst HS (all p<0.001), whereas significantly higher VAS ratings were only observed for severe HS.Qualitative results:1,221 pages of transcripts were reviewed, resulting in 274coded statements. Fragments were categorised in 5 themes that were present both in child and adult valuation, though with a different interpretation. Two themes encompassed General principles on the value of life: Staying alive is important(Life is worth living even with impaired HRQoL, for children and adults)and Inter-generational responsibility and dependency(All lives are precious: childhood is the foundation yearsand a time for havingnew experiences; adulthood is an important time to take care of the family). Three themes were identified as ConversionFactors:Awareness of poor HRQoL and ability to make decisions(children have difficulties comprehending poor HRQoL and their parents are making choices for them, whereas adults are able to assess their HRQoL and decide for themselves); Adapting and coping(children being flexible and resilient; adults having experience with dealing with difficulties); and the Practical organisation of care(children being unconditionally cared for by their parents; adults being able to organise and pay for care).The five themes were combined into a conceptual framework, labelled the “Circle of life”.Mixed methods: Comparing statements on the value of childhood as a time for new experiences and adults’ roles in life to prepare their children for adulthood, with the same respondents’ TTO values, confirmed the concordance between the qualitative and quantitative results.CONCLUSION: Adult respondents revealed having a lower willingness-to-trade LY for achieving higher HRQoL in children. Child-specific value sets will pose a challenge for pharmacoeconomic evaluation and resource allocation as this might have implications for access to health care for children.Sarah DewildeYouth50225Completed2019
20180580EQALY International Psychometric Analysis: Argentina studyThe aim of thisproject is to support item selection for a new instrument development (Extending the QALY (E-QALY project) that aims tocapture the changes in quality of life from interventions in health and social care that would be fit for the purpose of inclusion in economic evaluations. The core domains for the instrument have been identified and a pool of potential items have been proposed. Face validity has been undertaken in 5 countries (Argentina, Australia, China, Germany and USA) as well as the UK. Psychometric work based on a large survey is being undertaken in the UK from September through to December 2018 in order to identify items for the new measure. This proposed study will collect and analyse additional Argentina data to support the psychometric work and future item selection. It is the only Spanish speaking country in this initial multi-country team, Spanishbeingthe second most common language in the world. We will also explore the relationship of HRQoL and finnancial coping, patient and family coststhe whole sample andin different disease groups.Data on the E-QALY test items, plus EQ-5D-5L and other existing instruments, will be collected from a sample of 500 members of the public, of whom 400willreport a health condition using a commercial panel. This data will be analysed in line with the analysisof the UK survey data using basic psychometrics and Item Response Theory (IRT). These findingswill be interpreted in the light of evidence from the face validity work, and country specific recommendations will be made. These will be discussed in conjunction with other international evidence to support item selection for the new instrument.Federico AugustovskiDescriptive Systems, EQ-HWB20425Completed2019
2016200Clik 2016Paul KindPopulations and Health Systems16535Completed2016
20190820travel scholarship for Qingqing ChaiNan LuoEducation and Outreach8240Completed2019
20180460Psychometric assessment of the eQALY item pool in AustraliaThe ‘Extending the QALY’ project aims to develop a broad measure of quality of life for use in economic evaluation across health and social care. While the project consists of six stages, the proposed research plan is concerned with the fourth stage (psychometric validation).This follows assessment of the face-validity of the E-QALY items in Australia as part of stage 3 of the wider project. The overall aim of this study is to test the psychometric validity of the E-QALY item pool using data collected from Australian respondents. We will collect online data from a sample of 500 people in Australia with a self-reported health condition (covering a range of physical and mental conditions)and a healthy sample, using a range of instruments with different measurement frameworks. This includes the E-QALY item pool, measures of health-related quality of life (EQ-5D-5L and AQoL-8D), and a measure of wellbeing (WEMWBS). It is expected that this study will complement the psychometric analysis conducted by the UK E-QALY team for various reasons: (i) the E-QALY item pool will comprise additional items that were important to Australians in the face validity work, but were excluded by the UK E-QALY team; (ii) the item pool will include new items that were identified during the face-validity study in Australia; (iii) the inclusion of the AQoL-8D will extend the information available to psychometrically test the E-QALY item pool as it includes a number of domains that overlap with those developed by the E-QALY team; (iv) the Australian site will recruit additional patient groups that are not well coveredin the UK psychometric study. Overall, findings from the Australian psychometric study will provide additional data to support the item reduction and selection process.Brendan MulhernDescriptive Systems, EQ-HWB19200Completed2019
20190960An exploratory analysis of the Pain/Discomfort dimension of the EQ-5D-5L in people living with physical and mental health conditionsResearch into the Pain/Discomfortdimensionof EQ-5D instruments has identified problems with interpretation because two different aspects of health are combined in a single question. Studies have also shown that interpretation difficulties may arise because of the wording of response options, particularly concerning the differences between ‘slight’ and ‘moderate’ (levels 2 and 3), and ‘severe’ and ‘extreme’ (levels 4 and 5). This study will explore what is behind the Pain/Discomfort dimension of the EQ-5D-5L, drawing comparisons to other pain and discomfort items collected in the Extending the Quality-Adjusted Life Year (E-QALY) psychometric study in Australia. Analyses will include: descriptive statistics of response data; correlation analyses 2exploring the associations between EQ-5D-5L Pain/Discomfort and other E-QALY pain and discomfort questions; regression analyses assessing the extentto whichthe E-QALY pain and discomfort items are captured by EQ-5D-5L Pain/Discomfort; and discriminatory power analyses of the EQ-5D-5L with and without other E-QALY pain and discomfort questions. Furthermore, we will explore variation in reporting patterns for pain and discomfort questions across individuals living with mental health and physical health conditions, respectively. This sub-analysis arose from findings of the face validity study of the E-QALY project in Australia, which indicated that people with a mental health condition associated the Pain/Discomfort questions with emotionalpain or discomfort rather thanphysicalsensations. Overall, findings from this study will provide a better understanding of the Pain/Discomfortdimensionof the EQ-5D-5L, including implications for the complementary use of the EQ-5D-5L and E-QALY, and the future bolt-on agenda.Lidia EngelDescriptive Systems15000Completed2020
20170540Validity and Reliability testing of the EQ-5D-Y Proxy version 1 in young children.One of the key foci of the World Health Organisation (WHO) and its member states is the improvement of child health globally (1)(2). As a result there has been an increase in measures to monitor child health, including Health Related Quality of Life (HRQoL) in children and adolescents has increased overthe last two decades (3).HRQoL measures aim to capture the subjective multi-dimensional constructs of HRQoL namely physical, social and psychological functioning which are relevant to health (4,5). By definition, an individual’s HRQoL is subjective and should be elicited by self-report whenever possible, even from children (6). This is not always possibly as there are those who are either too young or cognitively unaware to self-report and so it is necessary to rely onproxy report (6–12). Olderchildren can provide reliable and valid responses to measures like the EQ-5D-Y (13), Paediatric Quality of Life Measure (PedsQL) (14–16), Kidscreen (17–19), KINDL and Kiddy-KINDL (20). However, for younger children below the age of 7-8 years it is necessary to rely on proxy report. The EQ-5D-Y was developed and validated for children and adolescents aged 8-18 years by an international team under the EuroQol group (13,21–24). The youth version of the instrument was derived from the EQ-5D-3L, an adult HRQoL instrument. The primary use of the EQ-5D is to support the estimation of quality adjusted life years (QALYs) for economic evaluation. In recent years a large number of treatments have been developed for paediatric conditions which has increased the need for a child specific version of the EQ-5D, with an associated economic tariff. The EQ-5D-Y, like the adult version, includesfive dimensions but hasage appropriate descriptions.The EQ-5D-Y Proxy is a direct adaptation of the EQ-5D-Y for proxy completionand is currently recommended by the EuroQoL Foundation for use in children aged 4-8 years (25).Proxy version 1requires the respondent to rate the child’s health as the proxy perceives it to be. Proxy version 1 hasbeen validated in a Spanish study in children over six years of age (26)and the performance hasbeen examined in a few studies with children older than six years (27,28). Scalone et al (2011) tested the EQ-5D-Y proxy for suitability of the domains (from the child’s perspective) on ill children and healthy childrenattending kindergarten or school from age 4-11years (29). The authors concluded thatthe domains on the current EQ-5D-Y proxy self-careand usual activities are not considered suitable for younger children between the ages of 4-5 (29). Verstraete andJelsma (2017) report that the EQ-5D-Y Proxy version was acceptable tocaregivers of ill and typically developing children and performs well in children aged four years and older,but was not appropriate for very young children(30).Jennifer JelsmaYouth24510Completed2018
20180600Initial psychometric testing of E-QALY and 3L-5L comparison in ChinaInterim report with flagged problematic items to contribute to a collective international judgement on item selection. Final report delivered to EuroQol group. One paper about the 3L and 5L comparison submitted to an EuroQol Plenary meeting and an international peer-reviewed journal for publicationNan LuoDescriptive Systems, EQ-HWB26400Completed2018
2014160Comparing valuation of the EQ-5D-Y and the EQ-5D-3L: The impact of wording and perspectiveIntroduction: Since the launch of the EQ-5D-Youth (EQ-5D-Y) in 2010, the instrument has been included in an increasing number of studiesthat consider health states for children and adolescents,with an accompanying demand for value sets. Given the similarities between the EQ-5D-Y and the EQ-5D-3L and that more than 20 value sets are currently available for the EQ-5D-3L, a natural starting point is to evaluate whether state valuations differ whenparticipants complete valuation tasks with different combinations of instruments and perspectivesadopted. Methods: The studywasconducted in Germany, Netherlands, Spain, and the UK. The valuation interview startedwith background questions, followed bya rankingtask, composite TTO (cTTO)tasks (9 states) and discrete choice experiments with dead (DCE+dead)tasks (9 pairs of health states). Debriefing questions were used to elicit respondents’opinions on the tasks and the perspectives they were asked to take. Seventeenhealth states were includedand divided in two blocks.Half of the respondentsreceivedthehealth states from theEQ-5D-Ydescriptive system, the other half valued health states based on the EQ-5D-3L. These two arms were further randomised into two groups that were asked to value health states either as if they were experienced by“themselves” (adult perspective) or by“a 10 year old child” (child perspective),obtaining four study groups.Descriptive analysis was used to examinesample background characteristics andcTTO and DCEresponses. 2-way MANOVAanalysis was used to compare cTTO responses and post-hoc comparisons wereadjusted using Bonferronicorrection. OLS regression was performed using N3 model. All results werepresented by the four arms of the study design. Results:The total sample consisted of 805 members of the adult general population, spread equally across the participating countries. Background characteristicswere broadly similaramong the four study arms. 2-way MANOVA analysis found a significant effect in the interaction between the instrument and theperspective for the observed cTTO values. Post-hoc comparisons after Bonferronicorrection confirmed that the wording affect values only on the EQ-5D-3L version, the wording affect values only for adults perspective and when crossing EQ-5D-3L for adults and EQ-5D-Y for child, the values are also affected. Estimated utilities for the pits state(33333)were different with the lowest estimated in the EQ-5D-3L and adult perspective (-0.309), followed by the EQ-5D-3L and child perspective (-0.227), then by EQ-5D-Y and adult perspective (-0.171) and finally for the EQ-5D-Y and child perspective (-0.151).Conclusion:Preliminary results from this study identified an interaction between the wording of the descriptive system valued (EQ-5D-3L versus EQ-5D-Y) and perspective (adult versus child) influencing values given to health states. Such outcome suggests that perspective and wording cannot be considered in isolation. The extent of the impact of this interaction on the values still needs to bedetermined in further analyses.Wolfgang GreinerYouth167070Completed2014
142-RAPerformance of the EQ-5D-Y Interviewer Administered Version in young children aged 5-8 yearsBackground: The standard EQ-5D-Y-3L has been used since it’s development in 2010 to measure health in the general population and in clinical samples in South Africa [1]–[4]. As Health Related Quality of Life (HRQoL) is subjective it is recommended that self-report is collected as far as possible [5]. Currently, the EQ-5D-Y self-complete is recommended for children aged 7-8 years. It is however suggested that children as young as five years may be able to reliably report on their HRQoL [5]. The aim of this study is to investigate whether children aged 5-8 years can self-report on their HRQoL using the newly developed EQ-5D-Y-3L interviewer administered (IA) version. Aim: The aim of the research study is to determine the validity, reliability and feasibility of the EQ-5D-Y-3L IA in children between 5-8 years compared to children aged 8-10 years. The secondary aim is to establish the performance of the IA version compared to the self-complete version in children aged 8-10 years. Methods: Participants will include children aged 5-10 years with a functional disability, receiving Orthopaedic intervention, chronic respiratory illness and General Population. Children will be asked to complete the EQ-5D-Y IA, Mood and Feelings Questionnaire and the Faces Pain Scale. Their physical functioning will be scored on the WeeFIM. Test-retest reliability will be established with a second completion of the EQ-5D-Y two days later. Feasibility will be established through, time taken to complete the measure and perceived difficulty in answering the EQ-5D-Y.Janine VerstraeteYouth38392Completed2020
20190010QALY estimation for HTA: The EuroQol approach. 2019 HTAsiaLink Annual ConferenceHTAsiaLink is the regional network of HTA agencies in Asia (www.htasialink.org). The annual meetings of HTAsiaLink are attended by more than 100 analysts and administrators of HTA agencies in Asia as well as academics. It is a platform for HTA agencies to learn from each other’s experiences and share information. The annual meetings provide a good platform for the EuroQol Group to reach out and educate Asian HTA practitioners with the EuroQol development and research work done in recent years, such as the Asian 5L valuation studies. HTA practitioners may not know the EQ-5D instruments or the value sets well. Also importantly, the HTAsiaLink meeting is a good opportunity to spread the news among Asian HTA practitioners and researchers that EQ-5D is now free for any non-commercial use. Recently, some leaders in HTAsiaLink expressed their serious concerns about the license fees for using EQ-5D instruments. They may not know the new fee policy recently implemented by the EuroQol Group and be conservative or even resistant to the EuroQol instruments. I am invited by the organizing chair of 2019 HTAsiaLink annual meeting to give a 30-mins presentation on the topic of QALY estimation for HTA: The EuroQol approach. This will be the first time to introduce the EQ-5D instruments in the annual meeting of this most influential Asian HTA community. Because of the misunderstanding of HTAsiaLink leaders about the EuroQol fee policies, I will emphasize the new policy in my presentation as well as in the social activities of this 3-day conference. The synopsis of my presentation is as below. Cost-utility analysis (CUA) is the most popular form of economic evaluation in health technology assessment. The EQ-5D instruments developed by the EuroQol Research Group have been used worldwide to collect health-related quality of life data for estimating qualityadjusted life years (QALYs) in CUA. Being newcomers in health technology assessment, Asian countries are following the steps of Western countries to establish their own evaluation systems and methods. For estimating QALYs, the first EuroQol instrument (i.e. EQ-5D-3L) was introduced to Asia in 1990s and a lot of research and development have been done in the region especially in recent years. However, it is not entirely clear how the EuroQol instruments have been used in Asian countries and, more importantly, how appropriate it is to use such instruments given the very different cultures between the East and the West. This preconference workshop is designed for participants to evaluate the latest evidence on the use and research of EuroQol instruments in Asia, identify and prioritise existing issues and knowledge gaps, and brainstorm ideas for future work in this area. To inform and facilitate the discussion among participants, recent research and development on EuroQol instruments in Asia will be presented in the workshop. Those include EQ-5D-5L value set studies, a 4-country Version 26JAN2017 3 qualitative study of the EuroQol health concepts, a systematic review of the psychometric properties of preference-based instruments, and a systematic review of the use of preferencebased instruments in CUA.Nan LuoEducation and Outreach2880Completed20192019
20170260Exploring potential innovations in hybrid modeling (REVISED)The aim of this project was to incorporate the power function into the estimation of cTTO-based weights and to test the effect of the power function on the 20-parameter model between countries (the Netherlands, Spain), response interpretation (point, interval), heteroskedasticity (constant, hetero) and censoring (none, -1). Examination of the formulae results revealed two limitations with cTTO responses: power cannot be estimated using cTTO responses alone and there is an overlap/or gap between BTD and WTD errors, which prevents interval estimation using the power function and cTTO responses. The benefits of relaxing the constant proportionality assumption is largely a shift in the values up, smaller confidence intervals, and improved face validity (e.g., fewer negative effects-coded coefficients). An exploratory analysis was conducted to attempt to address the two limitations by widening the intervals and incorporating DCE dead data. These pragmatic techniques appear successful and suggest specific changes to the EQ-VT protocol and hybrid analyses.Benjamin CraigValuation10000Completed20172018
20180250The EQ-5D-5L INSTRUMENT: PAST, PRESENT AND FUTUREA 1-hour workshop was delivered on Sunday, 9 September 2018 at 3:45pm in Keio Plaza Hotel (Tokyo, Japan). It was in the program of the ISPOR 8th Asia-Pacific Conference. The title of the workshop was “The EQ-5D-5L INSTRUMENT: PAST, PRESENT AND FUTURE”.Nan LuoOthers9660Completed2018
201802702nd EQ-5D training workshop for ChinaA half-day EQ-5D training workshop on 26 October 2018 in Fuzhou, China as the pre-conference short course of the 6th South China Pharmacoeconomics Forum.Nan LuoEducation and Outreach9400Completed20182018
20170410Comparing the EQ-5D-3L and EQ-5D-5L in a cohort of cancer patientsImprovements in HRQoLas measured by the QLU-C10D (derived from the condition specific EORTC QLQ-C30 instrument) appear to be associated with smaller changes in utility quantified by thematched5L compared to the 3L-For small reductions in HRQoL there is little to no difference between the 3L and the 5L group. For the large falls in HRQoL the 5L tariffsgenerated changes of smaller magnitude than the 3L tariffs-The crosswalk (a) loses the increased sensitivity of the 5L (if it detects more change) but (b) it stretches out utility values across a larger range (the 3L range), and hence gains or losses are larger and more in line with the 3L tariffs We conclude that this research demonstrates the potential effects of switching between instruments on policy recommendations resulting from cost-utility analysis in an oncology setting-3L data generally suggest larger health gains from effective interventions than 5L data, which will tend towards more attractive incremental cost-effectiveness ratios (ICERs) for interventions that improve quality of life. Attached are the presentations to the EuroQol Academy meeting 2018 (slides in the taskforce session and poster),and thepaper and presentation for the EuroQol Plenary meeting 2018.Richard NormanDescriptive Systems40000Completed20172017
2014020Supplementary funding 5L value set study England: LSENancy DevlinValuation57794Completed2014
2013230Establishing a tool for endorsing EQ-5D valuation studiesnot availableFeng XieValuation38500Completed2013
20180010Extending the QALY - Testing face and content validity with patients, social-care users and carers in GermanyIn total, we interviewed 30 respondents, the detailed demographic information was provided below in Table 1. We used convenient sampling method and did not set quota for the sample in terms of the gender, age, education and ethnicity. In overall, the sample was young and with high level of education. There were 5 respondents from interviewer 1 that did not fill in the self-reported questionnaire, which caused some missing information. The only variable we set quota on the sample was the category of the respondents (carers, physical patients, mental patients). Please note that in the self-completed questionnaire, we used ‘whether you have long-standing conditions’ to find out whether the respondent have physical or mental problems, but some respondents did not think their problems are ‘long-standing’ and therefore did not indicate their condition in the questionnaire. In this case, the interviewers probed that condition that the respondent was having and put down in the interview transcript. We recruited 2 postgraduate students and 1 recently graduated master student as interviewers from School of Public Health, Fudan University, Shanghai. The two interviewers had qualitative interview experience and all three interviewers were trained and briefed by the investigators before formal data collection. Interviews were conducted between 26th July and 17th August 2018. All respondents were recruited from two hospitals in Shanghai, No.10 Hospital of Shanghai and Zhongshan Hospital of Fudan University. Most respondents were recruited from the outpatient services, some were recruited from inpatient services. Interviewers approached the respondents and explained the purpose of this study and the takes they will be expected to do. Once the respondents consented, the interviewers took the respondents to a quiet place (e.g. meeting room, doctor’s office) and conducted the interview. All interviews were conducted following the International interview protocol of the E-QALY study. We did adjust the question of asking education attainment in the self-completed questionnaire according to the Chinese education system.Wolfgang GreinerDescriptive Systems, EQ-HWB9213Completed20182018
2013080An investigation into the meaning of, and the relationship between, Happiness, Health and Health status measurementHappiness, health and health status measurement was the title of the triple H project.Happiness was an inspiration. However it turned out tobeonly a segment of a set offeelings. And the set of feelings was only a segment of the set of subjective well-beingmeasures. In turn the set of subjective well-being measures then was only a segment of well-being measures. So it may be disappointing,but happinessis not the golden key to open upaPandora’s Boxfor the EuroQol Group. It might even be a fata morgana.Health perception and health status measurement provided most of the inspiration for thesuggestions below. Feelings are includedand that might be an area of research where theexperience of the EuroQol Group can be helpful. And there seems to be ample scope formore sophisticated measurementsof those feelings. Over time the perception of health haschanged. In 1987 measuring health status with a very few marker dimensions wastrendsetting, but theICF and questionnairesbased on the ICFhave arrived on thesceneasoffshoots. Thesemight be game changers in the realm of health status measurement.Without doubt the creativity and experience of the Group will ensure thatthis groupofresearchers will stay ahead of the game. The suggestions belowhopefullyprovide inspirationfor that endeavourFrank De CharroOthers30000Completed20132014
2013270Understanding participant's responses to the EQ-VT tasks; A qualitative studyPurpose: The EuroQol-Valuation-Technology (EQVT) uses traditional time-trade-off (tTTO) for health states better than dead and lead-time-TTO (LT-TTO) for states worse than dead to elicit a value (-1.0 to +1.0) for each health state. In the Canadian EQ-5D-5L Valuation studywhich used the EQVT platform, we observed an unexpected peak in frequency of “0” values and few negative values, particularly in the range of 0 to -0.5. To better understand this finding, we soughtto explore respondents’thought processeswhile valuing a health state, andtheir understanding of the tTTO andLT-TTOexercises.Methods: Qualitative semi-structured interviewswere conducted with EQVT task respondents. Questions focused on valuations of health states as:(a) Same as dead in tTTO, (b) Worse than dead in tTTO but changed to same as dead in the LT-TTO, (c) Worse than dead in LT-TTO, and (d) Worse than dead in LT-TTO with trading off all 10 years. Data were analyzed using content and thematic analysis. Results: Mean age ofparticipants (N=70) was 40±18.1years, 60%female, and 76% Caucasian. Participantsprovided similar reasons for valuing a health state same as or worse than dead. Many participants expressed confusion about worse than dead valuations, distinction between same as andworse than dead, and the transitionfrom tTTO to LT-TTO. A few indicated that theaddition of 10 years of full health in the LT-TTO influenced their valuations.Conclusions: The transition from tTTO to LT-TTO in the EQVT wasconfusing to participants, wherebysome health state valuations around this transition appeared to be arbitrary.Feng XieValuation24940Completed2013
2013100Preference Inversion in the EQ-5D-5LBackground: Evidence on preference inversion(i.e., where respondent preferences contradict the adjectival scalein health-related quality of life [HRQoL])has become increasingly apparent for the EQ-5D-5L, specifically in the last level of each item. This perception has anecdotally deteredthe adoption of the 5Lin the United States. Aims: 1.To directly test preference inversion between levels 4 and 5 of each itemby asking single intradomain pairs:Pr(L4>L5) L5) > Pr(L3>L4) Methods: To test for these inversionsempirically, I propose to create a survey containing a series of paired comparisons(listed below)using a nationally representative sample of the United States respondents. For each of the 10 intradomain and 200 interdomain pairs, 50 responses will be collected such that 2,635 respondents each complete 4 pairs (i.e., 50x(10+200)= 2635x4).Implications: Regardless of outcome, a better understanding of preference inversion has substantial implications for the EQ-5D development, particularly for its valuation and dissemination.Benjamin CraigValuation9182Completed20132013
20180500Reliability and validity of using EQ-5D-5L among healthy and adolescents with major mental disorders in EthiopiaBackground: TheEQ-5D, a generic measure of health-related quality of life (HRQoL) has been validated and used in healthy and with different disease conditionsof children and adolescents. The objective of this study was to test whetherEQ-5D-5L is feasible, reliable, and valid to use among healthy and adolescents with major mental disorders.Methods: A cluster sampling strategy was used to select 2,00 study participants from all of the ten sub-districts comprising the Butajira Rural HealthProgramme(BRHP). At each of the subdistrict, the supervisor selected one class from the school, then after identification of the class 20 students from the class between the age of 12 and 17 were interviewed using an Amharic EQ-5D-5L and EQ-VAS by trained and supervisedinterviewer. Participant with major mental disorders including; schizophrenia, bipolar and depressive disorders were recruited from Butajira major mental disorder center. For each case 100 participants were interviewed consecutively. Two way mixed-effects model absolute intraclass correlation coefficient, Kruskal-Wallis rank test and chi-square(X2)-tests were used to determine the reliability, and validityof the instrumentrespectively. Results: Fourinterviewers completed 500 interviews. From participants who completed the interview, a total of 497 (201 generaladolescents and 296 of major mental disorders) participants formed the sample for analysis.Intraclass correlation coefficient (ICC)was high for all EQ-5D-5L dimensionsand EQ-VAS, which lieswithin the goodagreement rangeacross all respondents (ICC > 0.7, p< 0.001). Theresultsshow that the Amharic EQ-5D-5L has a significant known group validity as shown by the difference in scores among various disease group, age category, educational status, presence of chronic illness history and religion of participants.Conclusions: The Amharic EQ-5D-5L appears to be an appropriate measure for measuring health related quality of life in various adolescent population groups in Ethiopia.Abraham GebregziabiherDescriptive Systems13450Completed20192020
2013200Extension of the labels within the EQ-5D-YThe objectiveof this study is to extent the labels within the EQ-5D-Y and thereforeto develop a four-ora five-level version of the EQ-5D-Y. The study aims to identify appropriate level labels for a revised EQ-5D-Y version withfour or five levels within each dimensionand to test the alternative versions concerning their comprehensibility and feasibility to get the finally resulting new EQ-5D-Y.UK, Germany, Spain, Italy, Sweden and China will take part in the study. During the first phase of the study, in each country a sample of 60 children and adolescents aged 8-15 from the general population of school children is needed. Within the second phaseof the study, a sample of30 children and adolescents aged 8-15(15 healthy childrenand15 children under treatment for a health condition)has to be recruited in each country.Firstly,a pool of possible labels has to be developedby a literature review. Theidentified labels will be tested and rated by conducting face-to-face interviews. By using response scaling tasks,respondentswill be asked how theywould rate the severity of a label on a visual analog scale (VAS).In the second phase, the developed new versions of the EQ-5D-Y will be tested by cognitive interviews after self-completing the questionnaires.By analysing the results of the cognitive interviews it would be possible to decide about the final version of the new EQ-5D-Y which is comprehensible, acceptable and easy in its use.Wolfgang GreinerYouth135400Completed2013
2015240EQ-5D-5L VALUATION IN POLANDObjective Cost-utility analysis gain importance also in Central and Eastern Europe (CEE). We aimed to develop Polish utility tariff for EQ-5D-5L health states.Methods Face-to-face, computer-assisted interviews were collected in the representative sample. Each respondent followed a standardizedprotocol to collect 10composite time trade off (cTTO) and 7discrete choice experiment (DCE) observations. In the Bayesian approach, several model specifications were compared based on model fit, the usabilityof the final value set, and how they reflect the elicitation procedure (e.g. censoring). A hybrid approach (using cTTO and DCE data) was employedin the final set, which wascomparedwith the existing ones: EQ-5D-3L and EQ-5D-5L cross-walk.ResultsData from 1252 respondents (11,480 cTTO valuations and 8764 DCE pairs) were collectedin June-October, 2016. The final model accounted for random parameters, error scaling with fattails, censoring at -1, unwillingness to trade in TTO by the religious, and Cauchy distribution in DCE. Pain/discomfort impacts the utility most: disutility equal to 0.575 when at level 5. In the value set, 4.4% of EQ-5D-5L states are worse than dead. Thenew value set has a comparable range (minimum at -0.590 compared to -0.523) and same ordering of the first three dimensions (pain/discomfort, mobility, self-care) as EQ-5D-3L and EQ-5D-5L cross-walk value sets. Moreover,it is more sensitive to mild health worsening.Conclusions The new value set guarantees consistency with past decisions in cost-utility studies,while offering betterassessment of even mild health gains. It can be an option for CEE countries lacking own value sets.Dominik GolickiValuation80000Completed2015
2015340EQ-5D for monitoring population health: a comparison of general population survey data in Argentina, Brazil, Chile and UruguayObjectives: To describeself-reportedhealth related quality of life (HRQL) fromgeneral population of Argentina, Brazil, Chile and Uruguay and to report population norms based on EQ-5D.Methods:We included data from the 2009 national health risk survey in Argentina (n=41,392) and from the participants of theEQ-5Dvaluation studies Brazil(n=3,362, Viegas Andrade 2013),Chile (n=2000, Zárate 2011) and Uruguay (n=792). We estimated mean VAS values and EQ-5D index by age, gender and the presence of limitations in any of the EQ-5D domains. We also explored the most frequent health states (those which include at least 90% of the population). Results:Self-rated mean VAS was 76.5 (IC95% 76.2-76.8), 83.8 (83.3-84.3), 75.7 (74.8-76.6) and 80 (78.6-80.9) in Argentina, Brazil, Chile and Uruguay respectively. The percentage of individuals without limitations was 60.9%, 44.3%, 49.8%, and 44.6% in Argentina, Brazil, Chile and Uruguay respectively. The EQ-5D TTO-based index valuesshowed slightly higher values in Argentina and Uruguay (0.910; CI95% 0.907-0.913 and 0.957; CI95% 0.951-0.962) than in Brazil and Chile (0.892, CI95% 0.888-0.897 and 0.80, CI95% 0.79-0.81). In all countries, both VAS and EQ-5D index values decreased with increasing age, andwere lowerin females. Out of the 243possible EQ-5D-3L health states only eight were present in at least 90% of the populationin Argentina, nine in Brazil, and 19 in Chile. In Uruguay, only 33 out of the possible 3125EQ-5D-5L health states were referred at least by 90% of the participants. In general, the prevalent states were the ones with mild limitations in pain-discomfort, anxiety-depression ormobility. Conclusions: This is the first collaborative work between Latin-American EQ-5D researchers from four countries. This study summarizespopulation norms for the region that can be used in decision-makingas well as in researchLucila Rey AresPopulations and Health Systems15000Completed20152017
2016720REDCap Evaluation: Instrument ConstructionOn 5 October 2016, Benjamin M. Craigand Juan Manuel Ramos Goñiconducted a brief evaluation of REDCap (1-week trial license) for discrete choice experiments (DCEs).Based on its findings, Arnd Jan Prause led a team to install REDCap on a virtual Unix server licensed by the EuroQol Research Foundation (https://mq-rc.codedose.com/). Installation was completed on 28 November 2016.The aim of thisfast-track proposal is to extend the evaluation by constructing two survey instruments for EQ-5D-5L valuation using DCEs (United States and Germany). This entails three steps: (1) the construction of the data dictionaries and randomization lists in English; (2) the translation of these files into German; and (3) the installation these files into REDCap. Each instrument will have three components: screener (consent, demographics), health (EQ-5D-5L and EQ VAS), and DCE (28 paired comparisons; two sets of 14 pairs).If this project is successful, we will further extend this evaluation and propose that (1) the instruments be fielded using online samples of 350 U.S. and 350 German respondents to collect 50 responses for each of the 196EQ-VT pairs (50 × 196/28 = 350); (2) that hyperlinks to the instruments be placed on the EuroQol website to aid in dissemination and feedback; (3) that an abstract on REDCap valuation be submitted to the Plenary meeting; and (4) that we visit Vanderbilt Universityto demonstrate the instrumentsand to solicit their counsel on the design. The longer term objective is to provide an alternative platform for future valuation research.Benjamin CraigValuation11680Completed20162017
20170100Measuring and valuing patient preferences in randomized clinical trialsRandomized clinical trial (RCT) is the ‘gold standard’ design to produce high quality evidence on safety and efficacy which is required for regulatory approval of new drugs. In Canada, cost effectiveness evidence is also formally considered to assess the overall value of new drugs before reimbursement decisions are made. Canadian Institutes of Health Research who requests cost effectiveness analysis be part of its RCT proposals catalyzes knowledge exchange between clinical researchers and health economists. As a result, more RCTs are incorporating the cost effectiveness design in which measuring and valuing patient preferences is a key component. Basic knowledge on how to measure and value patient preferences alongside a RCT is useful and perhaps necessary for clinical researchers, especially those who don’t have access to health economists. This first part of the workshop will introduce how to measure and value patient preferences using direct approaches through commonly used elicitation techniques and indirect approaches using pre-developed preference-based instruments. Hands-on exercises with the EQ-5D-5L used as an example to allow participants to practice these methods. The EQ5D-5L is a widely used instrument with its Canadian specific value set was recently developed. The strengths and limitations of applying these methods in the setting of RCT will be reviewed. Relevant recommendations from recently released Washington Cost Effectiveness Panel and CADTH Economic Evaluation Guidelines will also be highlighted. The second part of this workshop will use a hypothetical RCT to incorporate the EQ-5D-5L in the trial design, data collection and analysis, and interpretation, as well as its use in the cost effectiveness analysis.Feng XieValuation3690Completed20172017
2012040Bolt-on proposalOver the last few years there has been discussion within the EuroQol Group about the development of condition specific "bolt-ons" to the standard EQ-5D. Bolt-ons are extensions of the standard five dimension instrument, implying that a dimension or multiple dimensions will be added to the EQ-5D as to broaden the coverage of the health measurement continuum. Bolt-ons can be of substantial scientific and strategicinterest to the EuroQol group. There are several reasons for this interest. The most obvious and arguablymost important reason is user demand. Disease specific utility measures have begun to be developed and published in the literature. Also, agencies outside the EuroQol group are interested in funding research in this area. From a scientific point of view there are further grounds for exploring the development of condition specific bolt-ons to the EQ-5D. There is evidence that the EQ-5D does not capture all possible variations in disability caused by various conditions, in areas such as cognition (Krabbe et al, 1999), vision or hearing loss, (Espallargues et al, 2005; Brazier et al 2004), but also sexual function, incontinence or changes in someone’s appearance (Kaarlola et al, 2004). Two recent studies by Pickard and Lin (2010) and Solem et al (2011) show a number of condition areas where developingbolt-ons to the EQ-5D could be justified.Moreover, more and more so-called ‘mapping’algorithms to the EQ-5D are being used that are based on a link function between the EQ-5D and a condition specific non-preference based measure, often by means of a simple regression on the index values. Two recent review studies showed that many of these mappings are of very poor quality (Brazier et al, 2010, Mortimer et al, 2008), but the increasing use of these mappingsindicates that there is an interest for index values for extensions to the EQ-5D.There are two known studies that can be seen as initial bolt-on endeavours. The first is the addition of a cognitive dimension (Krabbe et al, 1999) and the second a sleep dimension (Yang et al, 2008). Both were not condition specific. In the cognition study valuations were performed and showed systematically different values compared to the standard EQ-5D, whereas content validity improved. It was concluded that the addition of the cognition dimension was justified. The secondbolt-on valuation study showed that the sleep dimension had no significant incremental disutility and therefore adding a sleep dimension to the EQ-5D was discouraged. Two recent studies aimed at the development of bolt-ons to the EQ-5Din the areas of vision and psoriasis (Yang et al, 2012; Swinburn et al, 2012). Vision had a significant impact on the EQ-5D-3L although differences between the EQ-5D-3L states and the corresponding states including the Vision dimension were not significant. Two psoriasis related dimensions (Skin irritation and Self-confidence)had a significant impact on the EQ-5D-5L.Bas JanssenDescriptive Systems40950Completed20122013
20170400Deriving EQ-5D-5L preference weights for DenmarkDeriving EQ-SD-SL preference weights for Denmark The project has two main aims: First, to derive a set of Danish preference-based weights for the EQ-SD-SL. The current weights were published in 2009 for the EQ-SD-3L (Wittrup-Jensen et al. 2009). We will follow the EuroQol Group's standardised valuation (EQ-VT) protocol to derive local preferences for the EQ-SD-SLquestionnaire, which includes two different tasks: composite time trade-off (TTO) and force-choice paired comparisons discrete choice experiment (DCE) (Oppe et al. 2014). Second, to further explore the DC methodology by adding a new task after the standard EQ-VT protocol. This is likely to be a head-to-head comparison of two DC approaches using a triplet format, one of which has been developed by the Dutch team, and the other by the Australian team. Specifically, we are interested in assessing the effect of duration in association with the choice task, and the use of dead. We will be discussing the details of these investigations with the EuroQol Group's Valuation Working Group (VWG).Claire GudexValuation55000Completed2017
20170160Can we use a 25 health states to estimate a EQ-5D-5L value set instead of “EQ-VT standard” 86?To construct the value set of EQ-5D-5L, the EuroQol Group designed a standardized valuation protocol (the EQ-VT). The experimental design of the EQ-VT includes a set of 86 health states needs to be valued using composite time trade-off (cTTO) method. These 86 health states were selected using a Monte Carlo procedure which was applied to a set of simulated data. It is currently unknow Version 26JAN2017Page 2how the 86 state cTTO design of the EQ-VT performs compared to other possible designs.We have already investigate this issueby using a so-called ‘saturation dataset’, which contains the observed VAS valuesfor all 3,125 health states in EQ-5D-5L. Using this dataset, we tested different design choices(i.e. different subset of health states). We included the subset of current EQ-VT design to see how well they can predict the observed values from the saturation data by computing the Mean absolute error (MAE) for all 3,125 health states. The result showedthe current EQ-VT selection of 86 health states performed well in estimating VAS valuesfor EQ-5D-5L health states. In addition, we found that a small orthogonal design with 25 health states performed also well. This indicates a reduction in design size (i.e. the number of states used for direct valuation/included in the design) is possible, which would increase the feasibility/reduce the burden of doing an EQ-VT study.The results from our work using the saturation data have a limitation in that the results werebased on VAS instead of cTTO.Therefore we plan a 3-arms study to confirm and further investigate our findings incTTO. In this study, we plan to collect cTTO data on two most prominent design choices as found in the VAS study: the orthogonal design with 25 states and the Bayesian design also with 25 states. Further, we will collect cTTO data for the current 86 EQ-VT as one control arm. To make this investigation feasible and less costly, we will collect the data using 500 Chinese students as respondents. The logistics for this study will benefit of the experience of similar logistics employed during the VAS EQ-5D-5L saturation in China. This will assure that the data is collected swiftly and validlJan BusschbachValuation37250Completed20172018
2013060Testing feasibility, reliability and validity of the health-related quality of life instrument EQ-5D-Y in children and adolescents with asthma- a cross-sectional pilot studyAnn Charlotte EgmarYouth2880Completed2013
2016670Measuring health-related quality of life in trauma patients: what is the added value of extending the EQ-5D3L and the EQ-5D5L with a cognitive domain?IntroductionThe EQ-5D is frequently used to understand the development of health-related quality of life (HRQL) following injury. However, the lack of a cognition dimension is generally felt as disadvantageous as many injuries involve cognitive effects. We aimed to assess the added value of a cognitive dimension in a cohort of injury patients.MethodsWe analyzed EQ-5D-3L extended with cognition (EQ-5D + C) dimension responses of 5346 adult injury patients. We studied dimension dependency, assessed the additional effect of the cognitive dimension on the EQ-VAS, and, using the EQ-VAS as a dependent variable, determined the impact of EQ-5D and EQ-5D + C attributes in multivariate regression analyses.ResultsExtreme cognitive problems combined with no problems on other dimensions are uncommon, whereas severe prob-lems on other dimensions frequently occur without cognitive problems. The EQ-VAS significantly decreased when cognitive problems emerged. Univariate regression analyses indicated that all EQ-5D + C dimensions were significantly associated with the EQ-VAS. Exploratory analyses showed that using any set of five of the six EQ-5D + C dimensions resulted in almost identical explained variance, and adding the remaining 6th dimension resulted in a similar additional impact.ConclusionsThe addition of the cognition dimension increased the explanatory power of the EQ-5D-3L. Although the increase in explanatory power was relatively small after the cognition dimension was added, the decrease of HRQoL (meas-ured with the EQ-VAS) resulting from cognitive problems was comparable to the decreases resulting from other EQ-5D dimensions.KeywordsGouke BonselDescriptive Systems79750Completed20172019
2013130The Relationship between Time, Sequencing, and Precision: Considerations for Choice ExperimentsObjective: This study estimatesthe effect of sequence on response precision and behaviorin health valuation studies.Methods: Time trade-off(TTO)and paired comparison responses from 6 health valuation studies—4 US, 1 Spanish, and 1 Dutch—were examined(22,225 respondents)to test whether task sequence influencesresponse precision(e.g., rounding),responsechanges and median response times.Each study used a computer-based instrumentthat randomized task sequence among a national sampleof adults, age 18 or older,from the general population.Results: For both TTO and paired comparisons, median response times decreasedwith sequence (i.e., learning), but tendedto flatten after the first 3 tasks. Although the paired comparison evidence demonstratedthat sequence hadno effect on response precision, the frequency of rounded TTO responses (to either 1-yearor 5-year units) increasedwith sequence. Conclusion: Based on these results, reducingthe numberor randomizing paired comparison tasksdoes not appear to influenceresponseprecision; however, generalizability, practicality,and precautionary considerations remain. Overall, participantslearnedto respondefficiently within the first 3tasksand didnot resort to satisficing, but may have roundedtheir TTO responses.Benjamin CraigValuation24899Completed2013
2014130Understanding the relationship between health behaviors, attitudes and perceptions of HRQL using the EQ-5DBackground: Strategies to improve public health may benefit from targeting specificlifestyles associated with poor health behaviors and outcomes. The aim of this studywas to characterize and examine the relationship between health and lifestyle-relatedattitudes (HLAs) and self-rated health and life-satisfaction.Methods: Secondary analyses were conducted on data from a 2012 communitywellness survey in Kirklees, UK. Using a validated HLA tool, respondents (n=9,130)were categorized into 5 segments: health conscious realists (33%), balancedcompensators (14%), live-for-todays (18%), hedonistic immortals (10%), andunconfident fatalists (25%). Multivariate regression was used to examine whetherHLAs could explain self-rated health (personal score) using the EQ-5D visual analogscale (EQ-VAS), the gap between personal and EQ-5D societal weighted score(societal score), and life-satisfaction. Health conscious realists served as the referencegroup.Results: Adjusted analyses found significant differences in the self-rating of health byHLA, with lowest mean EQ-VAS scores among unconfident fatalists (-9.0, 95% CI -9.9,-8.2) and live-for-todays (-2.0, 95%CI -2.8, -1.1). Unconfident fatalists had a largernegative difference between personal and societal scores (-0.027, 95% CI -0.017, -0.038), and were most likely to have low life-satisfaction (OR: 3.51, 95% CI 2.92, 4.23).Significance: Segmentation by HLA explains differences in self-rated health and life-satisfaction, with unconfident fatalists being a distinct segment with significantly worsehealth perceptions and life-satisfaction. In addition to informing conceptual frameworksof health, targeted health promotion efforts may benefit from considering the HLAsegment that predominates a patient group, especially unconfident fatalists, an areawhere further research is needed.Simon PickardValuation21000Completed2014
2015320A fuzzy approach to time trade-off experiment in EQ-5D-3L valuationAbstractUtilities of health states are often estimated to support public de-cisions in healthcare. People’s preferences may be imprecise, for lack of ac-tual trade-off experience. We show how to elicit the utilities accounting forimprecision (represented as fuzzy sets), discover the main drivers of impreci-sion, and compare several approaches to modelling health state utility datain the fuzzy setting. We extended the time trade-o↵(TTO) questionnaire, toelicit utilities of states defined in EQ-5D-3L descriptive system (health de-scribed by five dimensions) in 184 respondents. Our study demonstrates thatrespondents are capable of assessing own imprecision and a rigorous mathe-matical modelling is possible. The imprecision is larger than as inferred fromthe standard TTO method and is larger than estimation error, even in oursmallish sample. The analysis shows that non-trading behaviour in TTO oftenresults from imprecision, rather than lexicographic preferences for longevityover quality. People are especially imprecise in assessing the impact of usualactivities (one of dimensions) on utility; also, the internal inconsistency of ahealth state increases the imprecision. Fuzzy least squares seems best suitedto assign disutilities to individual dimensions, while separately modelling the location of preferences and amount of imprecision seems best to extrapolateimprecision for out-of-sample health states (to produce value sets). Accountingfor imprecision changes the results little, if crisp parameters are estimated.Michał JakubczykValuation27500Completed2016
2013150Non-additive impact of dimensions on the index values of health states: an analysis of interaction effects to avoid misspecification of the value function with unsaturated valuation datasetsSince it is common to include only a subset of health states (or health profiles) in any valuation study, statistical modelling is commonly applied to establish the effect of the health state on the elicited values. The resulting regression formula enables a value prediction for all health profiles, in terms of a predicted group mean, commonly called the value set or 'tariff'. The validity of this formula, essentially an extrapolation, depends on both features of the collected valuation data, and the statistical model used. The standard approach for obtaining VAS or TTO observations for this purpose has been the definition of a deliberate selection of health states. This selection has often been on average rather mild, that is rather healthy (e.g. the MVH set), and the subset rather small, with between 17 and 43 health states of 243 possible states being valued. The standard approach for modeling has been linear regression (not multilevel) with as main effects the descriptive variables of the health states. In case of the EQ-5D-3L these descriptive variables are represented by 10 dummy variables to cover the 5 trichotomized dimensions. Interaction models have also been used for several reasons, the most common interaction model defined by the so called N3 term. This standard approach is vulnerable for misspecification of the value function; past decisions on design and analysis were mainly a combination of theoretical reasoning and partial studies, with many issues undecided. These issues matter in view of extension of the system to EQ-5D-5L which requires the tariff to be re-estimated. To arrive at empirical evidence we subjected (1) two existing saturated EQ-5D-3L datasets and (2) simulated datasets (based on these observed data) to a systematic analysis. Our conclusions can be split into 3 parts: part 1 refers to the selection of health states, and their number, part 2 refers to the required number of responses per health state, an issue connected to inter- and intra-individual reliability, and part 3 refers to choices in the analysis of such incomplete datasets in particular if interaction terms are considered. A provisional conclusion on TTO (vs. VAS) is made, and the relevance for EQ-5D-5L is mentioned. Deliberate selection of health states, taking features into account of the assumed position of a health state on the 0-1 scale, or the composition of attributes such as the avoidance of complex or unlikely (‘stressed’) states, influences the probability of misspecification. This selection effect is strong, if the so called sampling ratio (the number of health states empirically valued divided by the number of all health states defined by the classification system) is equal or below 1:10. In general misspecification is difficult to catch in 1 quantitative measure: the performance in the upper part of the scale (the six states with at least four level 1s) dominates most measures. Here our judgments are, in model terms, non-linear and non-additive. Misspecification on average appears the least when severe and/or complex (‘stressed’) states dominate the selection of health states. When mild states dominate, then obviously the estimation benefits for the very best states, but these models perform, however, worse along the rest of the scale, which becomes clear with reasonably low sampling ratios (i.e. small subsets). Excluding the most upper part, misspecification is the least if so called stressed health states are overrepresented. In visual terms the fit is even much better. The MVH-selection represents a selection prone to more misspecification, since it is not only mild but also avoids stressed states. The sampling ratio of the health state is a critical feature for misspecification. In a case of >100 respondents per health state (hence little uncertainty on the mean), a sampling ratio of 1:10 or better (higher, say 1:8) is perhaps good enough (health states subject to the above selection criteria). Below 1:10 health state sampling, the bias of the health state selection criteria becomes stronger, like effects of low reliability. Lower sampling ratios than 1:10 are therefore hazardous. Any further conclusions on the sampling ratio depend on the reliability of the mean. So far we are not informed on inter-individual and intra-individual variation, which - with the number of respondents/observations per state - determines the reliability of the mean. If reliability is high, a limited number of observations per health state (say 10) suffices to obtain stable VAS means, but in practice 10 observations is not enough. This is even more so in TTO were reliability seems lower, and were lower values are a mix of values (and transformations) obtained by 2 different methods for better than dead and worse than dead. Returning to the sampling ratio in case of VAS values: if the number of respondents per health state inevitably is small, the number of health states cannot be small. And also, if the sampling ratio is low, interaction analysis is not advocated as it increases misspecification of the means of the health state values that were not part of the subset empirically valued, regardless of the source of the error. The interconnection between sampling ratio en reliability makes a simple conclusion on the minimal number of respondents hard to make. In TTO values and models, biases seem to have a stronger impact because models were quite distorted due to the lower end with negative values. Here the observations per health state should be even higher than in VAS models, and medians instead of means might be considered. Our modelling approach was simple in this regard; the effects of statistical or explanatory treatment of individual variation remains to be investigated (by multilevel analysis of modelling individual effects), as this might substantially decrease the confidence interval of the mean values as has been shown in similar datasets. In general terms interaction models (beyond main effect models) are likely to introduce misspecification, but one or two interaction terms may be conditionally profitable. The 1s and 3s interaction model was not particularly beneficial to improve model results above main effects models. The reason for the instable results are the irregularities in the interplay between domains along the 0-1 scale. The N3 model approach appears fairly accurate in large samples. Rather than that it changes lower state estimates, it generally rises the estimations of the better and best values. The better fit of the upper states compensates fit losses at the middle and lower end. In particular small sample models gained with the N3 interaction term, but on 1 occasion predictions with a small subset and N3 were wrong. In practice we do not known the truth, and in view of the other results the probability of such 'of the road' models which can not be recognized should not be underestimated. The driving force behind misspecification are the irregularities at the top of the scale (the jump of 11111 to the states with just one domain level changed say 12111). A lower sampling ratio (small subset) rapidly increases misspecification to unacceptable levels, in particular if means themselves are uncertain (e.g. with few observations per health state such as in the Australian design). This in turn increases interaction analysis risks. Therefore we regard main effects models superior (less risky) if sampling ratios are below 1: 10. Interaction strategies - if any - should be modest like N3. It seems useful to investigate in analytical strategies at the individual level to cope with the observed irregularities and to increase reliability of the (adjusted) means per health state. As said, TTO model results performed much less well compared to VAS, due to unreliable means and the effect of (large) negative values. Still most systematic observations were the same as in VAS. Also at the upper part and, to a lesser degree at the lower part of the scale misspecification follows the same pattern as in VAS. Considerable difference however existed in the lowest part of the scale. Our study showed that preference reversals (of parameter coefficients) and inconsistencies in terms of coefficients should be reinterpreted. These reversals surprisingly were design and analysis artefacts rather than 'wrong data'. As our data showed, the effect of an 'unlucky' choice of health states, low respondent numbers and/or of 'unlucky' data analysis (interaction term choices) may be provide utterly wrong models without internal options to check. Our pilot study provides warnings, and a methodological consistent strategy to find an optimal mix of design and analytical choices. Our study, however, did not provide a best strategy, not for EQ-5D-3L, nor for EQ-5D-5L. The following guidance for future studies may nevertheless be given. First of all the observed means must be stable. Whether is the case with 20, 30 or more observations per health state is to be established. We guess that below 30 respondents per health state the reliability of the observed mean easily becomes too low. Second, health state selections can be chosen selectively (mainly to improve the upper part of the scale), but a modelling strategy (to be developed) which would cover the upper jump/gap better with say 1 parameter might be favourable. For the remainder of the scale, overrepresentation of stressed states is better or, as we assume, a factorial design. Our study showed that there is no reason to avoid difficult, complex states. The removal of inconsistent respondents is of little effect. Finally additive main effects models are a robust and appropriate choice but modelling the individual effects seems an attractive complement. This pilot study is subject to limitations such as the datasets used, the valuation methods studied, the limited number of options in terms of health states choices, respondent numbers, and many regression features. We could not decide on one best parameter of misspecification, where new ones are proposed to cover the applicational requirements better. A final limitation is that at this stage little can be said on the bolt-on strategy. At this stage the most important lesson is that a sufficient number of health states and respondents is required, and interaction analysis should not be a first strategy.Gouke BonselValuation26136Completed2013
2015010Revisiting TTOnot availableAnna LugnerValuation26250Completed2015
2016040Assessing the validity of the 86 health state selection for EQ-5D-5L VT studies: are we using the most efficient and valid set of health states to predict health states that exist in practice?Introduction When constructing an EQ-5D value set, a small number of health states will be directly valued by general public, and subsequently values for all health states are predicted using a regression model. A criterion that has been used for selecting health states for direct valuation, is whether the health states are ‘implausible’ for respondents. Health states being ‘implausible’ refers to those health states that respondent might find unrealistic. Inclusion of implausible states has been assumed to compromise respondents’ engagement with the task and to increase uncertainty around obtained health states’ values. Nevertheless, ‘implausible health states’ can be found in almost all valuation studies. In this study, we first, rated the ‘implausibility’ of all 3,125EQ-5D-5L health states using a student panel and explored the impact of ‘implausibility’ on data modelling; second, we explored the thought process of the respondents used to value ‘implausible’ health states. Methods1,600 students from Guizhou Medical University China, were recruited and each student valued around 200 EQ-5D-5L health states using EQ-VAS. Of those, 890 students provided feedback on the implausibility of health states they valued using a binary scale (implausible/plausible). In addition, 9 students were invited to share their views on ‘implausible health states’ in one on one semi-structured interviews and 12 students were invited in two focus group discussion (6 students each). An ‘implausibility’ score (ranging from 0 to 1, indicating the proportion of respondents who rated a health state as ‘implausible’; a higher score meaning more implausible) was calculated for each health state. Four OLS main effect model were fitted using implausible/plausible/all observations with/without additional implausible variable. Goodness of fit and misprediction effect were compared between models. Thematic analysis was performed for qualitative data by two independent researchers. ResultsApproximately 30% of the 3,125 EQ-5D-5L health states were rated as‘implausible’ by at least 50% of respondents. In the 20 dummy variables main effect model for estimating value set, adding the ‘implausibility’ information in the model as an addictive variable improved the model fit and lowered the MSE with a beta=-3.231(P=0.000) with 95% CI: -4.430, -2.032 (on EQ-VAS). Besides, excluding the implausible observations from the model will result in higher prediction error (MAE from 3.088 to 3.444). The qualitative work showed that some respondents encountered problems whenvaluing ‘implausible health states’. The most frequently reported categories of ‘implausible health states’ was: ‘health states with contradicting levels between dimensions’. The major reason for difficulty in valuing health states reported was ‘lack of experience/knowledge with severe health problems’. Respondents tended to give low value to ‘implausible health state’. Strategies respondents used to value ‘implausible’ state include ‘rewrite the health state to rationalize the contradictions’, ‘focus on certain dimensions and value’ and ‘refer to previous valued health states’.Discussions Our findings suggested that within the EQ-5D-5L set of 3,125 health states ‘implausible health states’ were not rare and respondents tended give low value to ‘implausible health states’. The understanding of ‘implausible health states’ differed over respondents. Nevertheless, inclusion of those implausible health states improves the model to estimated values sets. In cTTO task, the effect of the ‘implausible health state’ may not be problematic as respondent reactions can be controlled by the interviewer. In on-line DCE studies however, without further assistance from interviewers, the inclusion of implausible health state may irritate, confuse respondent and leads to more randomness in the data.ConclusionImplausible health states do exists, are more difficult to value and tent to have lower values, but excluding such state from modelling will result in higher prediction error in overall. Hence, it is not recommended to remove such states from a valuation study.Jan BusschbachValuation62200Completed20162017
20190050panel presentation titled "Putting Patients at the Centre of Health Care: The use of patient-reported outcome measures (PROMs) in the Healthcare System” for CAHSPR annual conferenceThe presentation has been heldFatima Al SayahPopulations and Health Systems, Education and Outreach4211Completed2019
20180280EQ-5D-Y valuation study in JapanAs The Japanese Ministry of Health, Labour and Welfare (MHLW) plansto introducefull-scale cost-effectiveness analysis for the pricing of drugs and medical devices inFY2019 (April 2019),the National Institute of Public Health (NIPH), to which the principal investigator belongs, was asked to develop preference-based measures andscoring formula. Accordingly,this study aimsto derive ascoring weight to convert Japanese EQ-5D-Y responses to EQ-5D-Y index values in timeforimplementation(i.e., by the end of March 2019). Valuation methods areprincipally similar tothose used in other countriesand in accordance withthe methodological consensus of the EuroQOL EQ-5D-Y valuation group (May29,2018).We ask 1000 laypeople (aged 20 yearsor older) to respond to both the DCE and TTO in aface-to-face survey;200 respondents arerecruitedfrom five cities (Tokyo, Osaka, Fukuoka,Niigata, and Okayama), as in the EQ-5D-5L valuation survey.To collect respondentdata, EQ-VPT developed by the EuroQOL group is used. Before starting the valuation survey, EQ-VPTneeds to be translatedintoJapanese. DCE responsesare analyzed by a conditional logit model including dummy variables indicating each level of five domains. Latent DCE scores are converted to utility scale by mapping to TTO scores. Finally,we can obtain a scoring formula forEQ-5D-Y thatreflectsJapanese preferences. It may bemeaningful to compareitwiththeJapanese EQ-5D-5L preference weight.Takeru ShiroiwaYouth0Completed20182019
2015220The effect of chronic conditions on valuation of EQ?5D?5L health statesObjectives: To compare the utility values of the 5-level EuroQoL-5Dimension (EQ-5D-5L) health states elicited from heart disease patients, cancer patients, and members of the general public.Methods: In a face-to-face interview, each participant was asked to value 10 EQ-5D-5L health states using acomposite Time Trade-Off method. Utility values between the two patient groups and the general public group were compared using ordinary least-square regression models. Results: 157 heart disease patients, 169 cancer patients, and 169 members of general public were included in the analysis. Pooling utility values for all health states, heart disease patients and cancer patients had mean utility values lower by 0.11 points (P-value=0.014) and 0.06 points (P-value=0.148), respectively, compared to the general population. After taking into account demographic and socioeconomic characteristics, differences in health state utility values between the patient populations and the general population were not statistically significant, except that heart disease patients gave higher utility values (mean difference=0.08; P-value=0.007) to mild health states than the general population. Gain in utility values defined as utility value of a better health state minus that of a poorer health state was higher according to utility values derived from heart disease patients compared to the general population, with and without adjustment of demographic and socioeconomic characteristics.Conclusions: No simple patterns exist in health state utility values elicited from patients and the general population.Using utility values derived from the general population may under-estimate the comparative effectiveness and cost-effectiveness of treatments or interventions for certain type of disease.Nan LuoValuation22950Completed2015
71-RADoes the scoring matter? The impact of using different EQ-5D scoring methods on cost-utility results1. What is already known about the topic? Currently, a 3-level (3L) and a 5-level (5L) version of the EQ-5D is available. Both EQ-5D-3L and EQ-5D-5L health states can be converted to utility scores using either a value set or an interim (reverse) crosswalk. A recent study suggested that country-specific EQ-5D-5L crosswalks and value sets produce similar cost-utility outcomes. This study, however, only used empirical data of two trial-based economic evaluations and cost-utility outcomes derived using the EQ-5D-3L were not compared with those derived using the reverse crosswalk. Therefore, it remains unclear whether (reverse) crosswalks and value sets can also be used interchangeably in other patient populations. 2. What does the paper add to existing knowledge? This study provides novel information on the impact of using EQ-5D-3L and EQ-5D-5L value sets or (reverse) crosswalks on cost-utility outcomes for different health conditions (i.e., depression, low back pain, osteoarthritis, and cancer), different disease severity levels (i.e., mild, moderate, and severe), different treatment effect sizes (i.e., small, medium, and large), and different countries (i.e., the Netherlands, the United States, and Japan). The results indicate that the choice of utility scoring method can have a relevant impact on the probability of cost-effectiveness, especially when treatment effect sizes are small. 3. What insights does the paper provide for informing health care-related decision making? Crosswalks may be used when value sets are missing for a specific country or jurisdiction. However, our findings indicate that reimbursement decisions may change depending on the use of either value sets or (reverse) crosswalks. Therefore, when EQ-5D value sets are not available, rresearchers and decision-makers should be aware that the use of (reverse) crosswalk can impact the decision, especially in situations with small effect sizes.Judith BosmansValuation24440Completed2020
20190040monetary support for ongoing Tunisian valuation studyThe results of the study were published here: Chemli J, Drira C, Felfel H, Roudijk B, Al Sayah F, Kouki M, Kooli A, Razgallah Khrouf M. Valuing health-related quality of life using a hybrid approach: Tunisian value set for the EQ-5D-3L. Qual Life Res. 2021 May;30(5):1445-1455. doi: 10.1007/s11136-020-02730-z. Epub 2021 Jan 14. PMID: 33447958; PMCID: PMC8068700.Hajer FalfelValuation5050Completed2019
2014090Prevention of misspecification of the EuroQol tariff through optimal choices on design and analysis. An empirical and simulation based analysis using existing datasets.Introduction. The EQ-5D valuation approach classifies health into profiles, which subsequently receive a unique value from an existing value function. Such a value function requires: (1) a sample of respondents valuing a subset of profiles, covering a small part of all possible profiles, <5% for EQ-5D-5L; (2) modelling of the data, allowing for interpolation of values of the remaining profiles. This is inevitable as no respondent can directly value all states (to create a 'saturated' dataset). Multiple choices on design and analysis in valuation studies are therefore present, where optimality (lack of misspecification) is determined by sample size, health state design, and the role of individual effects on the values and the use of the scale. We developed a simulation-analysis tool (SAT), which allows for finding an optimal set of choices. Methods. VAS data from the multinational EQ-5D-5L pilot valuation studies were used to derive a main effects model, including a random intercept for all individuals, as a prior for all simulations. The resulting coefficients (20 dummies, 1 intercept, variance contribution of the random individual effect (here 20%)) were input parameters for SAT. SAT consists of: I) a generator of a defined sample of saturated data records of hypothetical respondents; II) a tool to use any sub-selection of states to mimic a particular design; this can be repeated n times; III) a regression tool which estimates the models on the datasets resulting from I and II; IV) an analytical tool that compares the values of these regression models to the original simulated data (using a predefined part of the 3125 profiles not used in III); V) a presentation tool showing the misspecification pattern of the particular design. Design choices involved were: the numbers of respondents (200-2000), the proportion of states valued (40-800); state selection criteria (random, mild/top, severe/bottom, flat, stressed).Misspecification was expressed as mean absolute error (MAE). Scale-dependence of misspecification (top, bottom of VAS scale) was explored.Results. With 1000 respondents and 120 states, MAE was around 0.47, independent from selection (mild, severe, etc.). Using 40 states increased MAE (0.60), more in case of mild state selection (0.65). Increasing beyond 120 states did not significantly improve results. Over the range of 200-2000 respondents, the mean MAE decreased almost according to the inverse square root of the sample (MAE with n=2000 about 0.33). The MAE consistently differs along the VAS scale, being highest on top; this scale dependent misspecification is to a small degree dependent on set selection, and disappears if the number of states or number of respondents is high.Conclusions. If the underlying value function is adequately represented by a main effects model, then misspecification primarily depends on the number of respondents, as long as the number of states is around 80 - 120. The impact of set type is small, although mild selections performed worse. To avoid scale dependent heterogeneity of MAE, 500 seems the minimal number of respondents. We hypothesize that the 20% individual variance is the hitherto underestimated factor in misspecification, where designs could be more efficient (lower n) if the model contains a specified individual component rather than a random intercept. SAT is an extremely powerful and valuable tool to guide design choices in valuation studies.Gouke BonselValuation75000Completed2014
2016620A head-to-head comparison of eight country-specific EQ-5D-3L and EQ-5D-5L value sets in eight patient groups and a student cohort - Why discriminatory power varies with versions, value sets and subgroups?The current study described the first head-to-head comparison of seven country-specific EQ-5D-3L and EQ-5D-5L value sets comparisonin a comprehensive pooled dataset. Our study had two research questions:Do 5L utility values show better performancethan 3L utilityscores in terms of discriminatory poweras a direct result of improved descriptiveprecision?and: What are the factors affectingthis performance? A large multinational dataset, including 3L and 5L data for eight patient groups and a student cohort, was used to evaluate the value sets of Canada, China, England/UK, Japan, the Netherlands, Spain and South Korea.We used distributional analyses and two types of methods exploring discriminatory power: relative efficiency as assessed by the F-statistic, and an area under the curve for receiver-operating characteristics approach (AUROC). Four factors were identified as underlying to differences in discriminatory power: descriptive performance, distributional effects, valuation effects, and country-specific effects. An in-depth investigation led to many interesting insights and robust conclusions.Bas JanssenDescriptive Systems15000Completed20162017
20180430EuroQol Annual meeting of Asian region: sharing and networking between EQ-5D researchers of Asian countries.To summarize,this EuroQol Asian Academy Meetingfunded by EuroQol Groupcould be seen a promotion of the EuroQol Group as keyplayer inthe development of HRQOL and HTA studies in the Asian region. We planned the secondmeetingin 2020 that help the EQ-5D researchersin Asia to share their results and foster collaborations.Fredrick PurbaEducation and Outreach48825Completed20192019
2013250Examining interviewer performance in the Dutch EQVT studiesBackground:TheDutch EQ-5D-5L valuation studywas one of the first applications of the new protocol for EQ-5D-5L valuation studies. To contribute information on performance of the protocol, a subsample of the face-to-face interviews with respondents was recorded to evaluate interviewers’ protocol compliance and identify possible issues with the protocol. Methods:Video recordings of 157out of the 1003 participants in the Dutch EQ-5D-5L valuation study were available for analyses containing screen captures, mouse movements, and sound recordings of interviewer and respondent conversations.These interviews had been completed by 6 different interviewers, three of whom had substantial prior experience with conducting TTO interviews. We developeda protocol adherence checklist formultiple interviewer and respondent performance indicators. These indicators were identified based on a subset (n=20) of the 157interviews as well as direct input from a subset of the interviewers. Subsequently all videos were scoredusing the checklist. Results:There were significant differences between interviewers on several of the performance indicators in both the introduction to the composite time trade-off task(cTTO)as well as during the valuation of health states.The differences were not associated with particularresponse patterns. Interviewers were consistent in describing the 'wheelchair example' and explaining the cTTO set-up; they were inconsistent in switching to the worse-than-dead task, and how they handled respondent mistakes.Amongst the novice interviewers, one had persistent problems with the transition to the worse than dead part of the cTTO interview.Generally the experienced interviewers followed the protocol more closely.Discussion:We obtained insight into the performance of interviewers during elicitation of health state preferences using the composite time trade-offtask.The differences between interviewers did not appear to directly affect responses in a systematic way.Alexander AronsValuation39375Completed2014
20180420The impact of partial profile designs on the valuation of health statesPartial profile presentation of discrete choice experiment (DCE) choice tasks can be used to simplify health state valuations and reduce the burden on participating respondents. However, research has shown that presenting choice tasks with partial profilescan produce nonequivalent coefficient estimates to the case of presenting as full profiles. This study aimed to examine whether respondents’ tendency to infer missing attribute information can bring the discrepancy of parameter estimates between DCEs withfull versus partial profile presentation. At first, DCE data with two study arms (one with the full profile presentation and the other with the partial profile presentation) were used to examine the differences in EQ-5D-5L parameter estimates depending onthe presentation format. Then, three different models capturing the effect of respondents’ inference for absent attributes in choice tasks with the partial profile presentation were investigated based on three different data sets with varying methods to simulate respondents’ inference behavior. The results showed that DCEs with the partial profiles produced significantly different estimates from the case of the full profile presentation. However, we did not find significant evidence on the effect of inferring absent attributes in DCEs with the partial profile presentation. Nonetheless, we recommend being cautious about using choice tasks with partial profiles as respondents are likely to have structurally different health state values depending on the presentation format.Sesil LimValuation24100Completed2019
2015280EQ-5D-3L in Hematologic Malignant Neoplasms: a Systematic Review of Health State Utility ValuesBACKGROUND:A rapid development of new drugs and new indications for registered drugs in the area of hematologic malignancies is observed, which may increase the need for cost-utility analyses and health state utilities estimates.AIM:To systematically review data on EQ-5D use in patients with hematologic malignancies (ICD-10: C81-C96and D46) and sum up health state utilities, descriptive system results,and psychometric properties in this area.METHODS:In February 2016,we ran a systematic search in MEDLINE (PubMed), EMBASE, Cochrane Library, EuroQol website, CEA Registry, LILACS (VHL), ProQuest, CRD,and Web of Science. Additional references were obtained from reviewed articles. The search strategy combined MESH, EMTREE,and free-text terms respecting the population and the EQ-5D instruments (EQ-5D-3L, EQ-5D-5L or EQ-5D-Y). Two authors independently screened titles and abstracts of studies found duringthe searches, and then at least two authors screenedfull texts of selected articles. Data from studies meeting the inclusion criteria were extracted using a pre-determined extraction form. RESULTS:Three hundred eighty eight abstracts were initially identified and screened,and 101 records were screened in full text. Finally, 54 full text papers (49 referring to EQ-5D-3L, 4 to EQ-5D-5L,and 1 to EQ-5D-Y) were included in the review. The most popular clinical diagnoses included: multiple myeloma (20 papers), chronic myeloid leukemia (7), acute myeloid leukemia (5), myelodysplastic syndrome (4), Hodgkin lymphoma (3), follicular lymphoma (3),andnon-Hodgkinlymphomas (ICD10: C82-C88; 6 papers). In total, 332 health state utility values (including 301 EQ-5D-3L values), ranging from -0.28 to 0.98 were reported in 46papers. The highest number of health state utility values (HSUVs)was reported for multiple myeloma (n=133; range: -0.025 to 0.923) and chronic myeloid leukemia (n=49; range: -0.28 to 0.91). EQ descriptive system and EQ VAS results were presented in 12 and 23papers, respectively. In terms of EQ-5D psychometric properties, validity, reliability and responsiveness weredescribed in 11, 1,and 3 papers, respectively. CONCLUSIONS:A largenumber of EQ-5D-derived health state utilities for hematologic malignant neoplasms was identified, although huge inequalities in HSUVs’ availability between diseases were also noticed. Details of review results will be presented in a systematic way.Dominik GolickiDescriptive Systems14800Completed2015
20170401Deriving EQ-5D-5L preference weights for Denmark - request for budget extensionDeriving EQ-5D-5L preference weights for DenmarkThe project has two main aims:i)To derive a set of EQ-5D-5L preference-based weights for Denmark by using the EQ -VT protocol that includes composite time trade-off (TTO) and DCE as paired comparisonsii)Tofurther explore DCE methodology by adding a new task after the standard EQ-VT protocol, which is triplet DCE with duration (same as that used in the Peru 5L valuation study)Current status of project:-Approx. 810 interviews completed to date-Good quality data for cTTO, and standard DCE-Interviewers are performing well -All participants are also asked to completetriplet DCE-Poster on interviewer experiences will be presented at ISPOR 2019 Reason for request for extra funding from EuroQol:-A lower than expected acceptance rate for participating in interviews led to too slow data collection (7 months to complete approx. 400 interviews), loss of 4 of the 7 initial interviewers, and higher study costs. -In July 2019 we made a pilot with the market research company Epinion for them to contact potential respondents and recruit them to the study. This has worked well, and we are continuing our collaboration with Epinion-We request extra funding to cover costs related to participant recruitment by Epinionandto employment and training of four new interviewers.Project ExtensionWe also request formal acceptance that the project completion date is extended to December 2020.Lars EhlersValuation40946Completed20182020
2015370EQ-5D: the ABC approach to measuring and valuing health in Latin America (W15 workshop: 5th ISPOR Latin America meeting/Chile 2015EQ-5D is currently being used to measure outcomes in a wide range of applications –in the measurement of population health in national/regional surveys, in calculating QALYs for economic appraisals, in the assessment of patient reported outcomes and as a performance indicator in benchmarking health providers.The EQ-5D-3L descriptive system comprises 5 generic dimensions of health: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each dimension has 3 levels: no problems, some problems, extreme problems. A respondent is asked to indicate her health state by ticking in the box against the most appropriate statement in each of the 5 dimensions. This classification defines a total of 243 unique health profiles, which can be later transformed into an index by attaching social preferences of a particular relevant population, which enables statistical comparisons between different intervention groups. Given the lack of social values for EQ-5D in many LA countries, a series of methods have been proposed to indirectly derive such valuations. Those alternatives represent only short-term solutions and should be avoided when local valuation data is available. Scientific evidence supports the fact that health valuations tend to vary from country to country and also within a society whenits level of cultural heterogeneity is high. Decision makers in LA should always try to collect direct valuation data when economic evaluation methods are being considered as part of the planning and regulation of health care.victor zarateEducation and Outreach22900Completed20152015
2015310The impact of duration on EQ‐5D‐5L value sets derived from a Discrete Choice Experimentnot availableRichard NormanValuation31200Completed2015
20170450Publishing a paper titled “Cost-utility analysis using EQ-5D: does how the utility values are derived matter?”The aim of this project is to publish a paper in a peer reviewed journal. The paper mainly examines the effect of using different EQ-5D value sets (including the UK 3L and 5L value sets) on cost-effectiveness analysis. The paper has been drafted but needs more work to be publishable. The draft abstract of the paper is as below: Objectives: This study aimed to evaluate how country-specific EQ-5D-5L value sets would affect cost-effectiveness results. The impact of crosswalk- and mapping-derived EQ-5D-3L values was also examined. Methods: The evaluations were based on a cost-utility analysis of hemodialysis (HD) and peritoneal dialysis (PD) for patients with end-stage renal disease (ESRD) in which primary EQ-5D-5L and Short Form-12 (SF-12) data was collected for quality-adjusted life year estimation. Individual-level utility values were generated using EQ-5D-5L value setsfrom seven countries(5L), EQ-5D-3L Version 18 NOV 2015Page 2 value sets from seven countries via crosswalk (c3L), and four algorithms mapping from SF-12 to EQ-5D-3L (m3L). Cost-effectiveness was assessed for non-diabetic and diabetic patients separately, by comparing incremental cost-effectiveness ratios (ICERs). Results: ICERs based on 5L values ranged from S$55379 to S$62222 per QALY for non-diabetic patients and S$69008 to S$77406 per QALY for diabetic patients. ICERs based on c3L values ranged from S$56210 to S$61947 per QALY for non-diabetic patients and S$71843 to S$82459 per QALY for diabetic patients. 5L and c3L value sets of the same country tended to produce similar ICERs for non-diabetic patients, but 5L generated lower ICERs for diabetic patients. ICERs based on m3L values (range: S$82588 to S$92506 per QALY) were much higher than others. Conclusions: Our study shows that currently available approaches to generating EQ-5D utility values would lead to varied ICER values and sometimes different decisions (using S$60000 per QALY as the threshold) in cost-utility analysis. It seems that developing and designating an EQ-5D value set of own country, such as a country-specific 5L value set, is imperative for achieving valid and consistent economic evaluation for health technology assessmenNan LuoDescriptive Systems5000Completed20172018
2013090Course Title: Discrete choice for health state valuationCourse offered to EQ members in 2013. Faculty: Estehr de Bekker-Grob, Marcel Jonker, Elly StolkElly StolkEducation and Outreach9450Completed20132013
2016360Health related quality of life measurement -uses in economic evaluation and population health.not availableFederico AugustovskiEducation and Outreach15000Completed20162016
2015080Valuing EQ-5D-3L health states in a representative sample of the Colombian population : a proposed retest extensionPaul KindValuation9502Completed2015
2015150Values for severe states: a methodological and cultural viewIntroduction: Scaling severe states can be a difficult task. First, the method of measurement affects whether a health state is considered better or worse than dead. Second, in discrete choice experiments, different models to anchor health states on 0 (dead) and 1 (perfect health) produce varying amounts of health states worse than dead. Research question: Within the context of the QALY model, this paper provides insight into the value assigned to dead and its consequences for decision making. Our research questions are: 1) What are the arguments set forth to assign dead the number 0 on the health-utility scale? 2) What are the effects of the position of dead on the health-utility scale on decision making? Methods: A literature review was conducted to explore the arguments set forth to assign dead a value of 0 in the QALY model. In addition, scale properties and transformations were considered. Results:The review uncovered several practical and theoretical considerations for setting dead at 0. In the QALY model, indifference between two health episodes is not preserved under changes of the origin of the duration scale. Ratio scale properties are needed for the duration scale to preserve indifferences. In combination with preferences and zero conditions for duration and health, it follows that dead should have a value of 0. Conclusions: The health-utility and the duration scale have ratio scale properties and that dead should be assigned the number zero. Furthermore, the position of dead should be carefully established, as it determines how life saving and life improving values are weighed in cost-utility analysis.Peep StalmeierValuation251305Completed2015
2016370A qualitative approach to understanding what aspects of health are important to peopleThe aims of this research project are:To understand what concepts and definitions should form the basis for a EuroQol generic classification systemTo develop and pilot an approach to identifying what aspects of health1are important to people To obtain the views of patients and members of the public about what aspects of health are important to themTo produce a list of candidate items for potential inclusion in a generic classification systemThe project will comprise four stages:Stage 1: Review of English language definitions of the following concepts: health, quality of life, health-related quality of life, and wellbeingStage 2: Development of a survey designed to understand what aspects of health are important to people, with piloting of the draft survey in a focus groupStage 3: Administration of the revised survey to patient and general public samples (n=200), with the aim of generating a list of potential domains for a generic classification system, ranked by importanceStage 4: Checking / triangulation of the stage 3 results using a second focus groupThe study will inform the ongoing research agenda around the measurement of health and health-related quality of life beyond the existing EQ-5D descriptivesystem.Koonal ShahDescriptive Systems86500Completed20172018
2016060HTAi Workshop: Engaging patients and general public in health technology assessment: Measuring and valuing health preferencesFeng XieEducation and Outreach23220Completed20162016
2015450A head-to-head comparison of nine country-specific EQ-5D-3L and EQ-5D-5L value sets in eight patient groups and a student cohortnot availabeBas JanssenDescriptive Systems14250Completed20162016
20170130Scoring Methods for the EQ-5D instrument: theoretical background and empirical analyses (revised application)Introduction: The EQ-5D is a preference-based measure developed for economic analysis of health interventions. Yet, most of its applications are in other settings, where preference-based approaches have no more merit than other scoring approaches. In this study, we explore reflective and formative approaches to summarize the EQ-5D-5L for non-economic applications. Methods: We examined: 1) reflective approach modelling EQ-5D-5L items on one latent factor using confirmatory factor analyses (CFA) and item response theory (IRT); 2) formative approach conceptualizing the items as causal and health scales as reflective using Multiple Indicators, Multiple Causes (“external” MIMIC) models; and 3) hybrid approach which modeled some EQ-5D-5L items as formative and others as reflective indicators of one latent factor (“internal” MIMIC). Nine datasets [population (4), patient (3), mix (2)] containing the EQ-5D-5L were analyzed to examine the robustness of the results. Results: CFA showed all items loaded well (0.7 to <0.95) except for AD (loadings 0.192 to 0.619, excluding one outlier). The best fitting external MIMIC modelled the HUI3 and SF-36 subscales on two factors (RMSEA=0.147). The best fitting internal MIMIC model defined MO, AD and, PD as causal, and SC and UA reflective indicators, confirming the findings of Gamst-Klaussen et al. 2017. Conclusion: Neither the reflective or formative approach was best – the AD and SC dimensions were problematic, respectively. A hybrid model to scoring the EQ-5D-5L is suggested for future research. Most datasets analyzed were from Western, developed countries; despite similar findings across datasets, the results may not be generalizable to all countries.thomas kohlmannValuation37500Completed20172018
20190150R1Exploring non-iterative time trade-off methods for valuation of EQ-5D-5L health statesIn current EQ-VT protocol,compositetime trade-off (cTTO) method is used as the primaryvaluationmethod.The cTTO method usesa fixed starting point and iterative route, which are suspectable to cause bias.Discrete choice experiment (DCE) is another method adopted in the EQ-VT protocol to value EQ-5D health states. However, the DCE task does not rely on the iterations. If we could come up with a new TTO variant without the iterative procedures, the TTO task maybe less cognitive demanding for the respondent and theresultantdata will be free from the iterative related bias.In some EQ-VT studies, ‘spikes’ were found at the values of 1 and 0.5 in the histogramsof the aggregated c-TTO data, which is likely to be caused by anchoring effects or satisficing.Another benefit of not using the iterative process is that the valuationtasks canbe done with minimal supervision by interviewers and couldbe done even without an interviewer. This way, we could also minimize the potential interviewer effect.In summary,good quality c-TTO data can only be collected in tightly controlled interviews such as in national valuation studiesaiming at estimate national value set. The required quality control measures are expensive. This creates a demand for simplified protocols (e.g. online/on paper/with lesswell-trained interviewersetc.)In this study, we aim to test two non-iterative TTO tasks, one known as non-stopping TTO and one known as open-ended TTO. The n-TTO was compared with c-TTO in a pilot study in Norway and Singapore. It was found that n-TTO did not outformed c-TTO but n-TTO was less time-consuming and one variant of n-TTO performed similarly well as c-TTO. Given that was the very first pilot of n-TTO, we think these results are encouraging and that n-TTO should be further tested with better designs. A paper reporting these findings has been accepted for discussion in the coming EuroQol Plenary meeting in Brussels.The o-TTO method has beenused regularly by thegroupof Lisa Properin Boston as a way of obtaining TTO scores from postal surveys, but has not been studied head to head against the c-TTO method. We will further refine these two methods to mimic the EQ-VT framework and pilot test these two methods in terms of data distribution, logical consistency, and acceptability. Based on the results of the pilot study, we will fully design the data collection protocol for these two methods (e.g. interview script, warm-up exercises etc.)and compare theirperformancesagainst the cTTO method in valuing EQ-5D-5L health states.Additionally, the current Chinese EQ-5D-5L value set isbased on the preferences of urban respondents only. This hampered the increasing use of EQ-5D-5L in Chinaas the rural population accounted for over 50% of thetotalpopulation and the preferences of urban and rural population may differ.As a second aim of this study, we would like to comparethe preferences of the urban population and the rural population in China.Zhihao YangValuation68600Ongoing2019
2015380Alberta EQ-5D end-user conference and APERSU Scientific Advisory Committee meeting fall 2015The planned event is intended to start with the end-user conference that would last for one and half days. After that conference, there would be a SAC meeting that would last the afternoon of the second day. That arrangement would allow SAC members to participate in the meeting, and discuss with the end-users prior to making the decisions about the research directions of APERSU. This arrangement would also allow key EuroQol representatives to participate in the end-user meeting, serve as experts in the various discussions, and also look forward for potential collaboration projects with Alberta research teams. The initial agenda for the meeting is attached (Appendix A). The end-user meeting is intended to be for about 50 people, including APERSU staff and leadership, HQCA and AHS representatives in the initiative and people they invite internally, Alberta Health (ministry) representatives, primary health care and strategic clinical networks representatives, and key provincial PROMs researchers at the University of Alberta and University of Calgary. Briefing of the end-user meeting would be made available at the APERSU website. The aim of the SAC meeting is to discuss the research agenda of the APERSU and thus the first version of the APERSU research agenda for the coming years will be the second outcome of the meeting.Arto OhinmaaEducation and Outreach15000Completed2015
2016690DCE Learning curvesABSTRACT: Introduction: The use of discrete choice experiments (DCEs) to value health outcomes captured by the EQ-5D has gained popularity in recent years. Research on DCE’s suggests they may be used as a standalone technique to produce values on a quality-adjusted life year (QALY) scale, and online DCE’s may one day supplant more conventional interviewer-based techniques due to their lower costs and ease of use. However, DCE’s are not without their own issues. The complexity of the choice task may be cognitively burdensome, forcing respondents to adopt heuristics. This may induce biases in the results.Common heuristics in DCE may include magnification (non-attenuation; e.g., only looking at lifespans), horizontal blindness (always choose left option), or favoring the status quo (e.g., choosing the option that most resembles current health). The sequence order may also play an important role in heuristics: arespondent may begin the survey with undeveloped preferences and base responses on previous questions, not true preferences. Some individuals may experience burn out and only resort to heuristics measures after a certain number of questions. Methods: Using two recently completed DCE surveys of the US general population (N=15,292), I tested for common heuristic approaches to an EQ-5D paired comparison task. An example paired comparison task may ask for an individual if they prefer (A) 10 years in health state 33333 or (B) 6 years in health state 11111. Each survey asked between 20-30 paired comparisons. Results: The analysis showed significant evidence of heuristic patterns and sequence effects in each study. As the survey progresses, individuals increasingly prefer the option with the longer lifespan (p-value<-0.01). Time trade-off (TTO) pairs suffer from a significant right bias where there is nearly a 5% difference (59% vs 54%) in choice probability depending on whether or not the option was presented on the right. The pairs an individual has previously seen also has an influence on future choices (p-value<0.01). Individuals exposed to easier pair choices in a moderate state (one option was preferred 80% of the time) at the beginning of the survey were more likely to exhibit lexicographic preferences for the entire survey than those who were given difficult pair choices with severe health states (these had a 50/50 split) to start (19% vs 15%). Conclusion: It is important for researchers to consider these issues when designing a choice survey, particularly the interpretation of sequence effects. Many of these biases can be reduced or even prevented given a proper design. Recent econometric advancements in choice modelling (i.e. power function, heterogeneity, lexicographic clusters) can also help reduce the effects of heuristics and sequence on values for health outcomes.John HartmanValuation14000Completed20162017
2013020A Korean valuation study for the EQ-5D-5LBackground The EuroQol Group recently developed a new questionnaire-the EQ-5D-5L-which has 5 levels per dimension. However, the social values for EQ-5D-5L health states have depended on each country. This study aimed to estimate Korean preference weights for EQ-5D-5L based on values elicited from Korean population applying the EuroQoL Valuation Technology (EQ-VT) program and the standard protocol by the EuroQol group. Methods Of 1,085 general populations were recruited using multi-stage quota sampling method in Korea. Each respondent valued 10 health states using the composite time trade-off (cTTO) and 7 health states using discrete choice experiment (DCE). The EQ-VT program was developed by the EuroQol group and translated into Korean with the Korean research team. Computer-assisted, face-to-face interviews were conducted. A range of predictive models were explored using cTTO and DCE data, respectively. The most appropriate model was determined after assessing goodness of fit, logical consistency, and parsimony. Results Model with dummy variables for each level of severity and dimension and the term that picked up whether any dimension in the state was at 4th or 5th level was selected as the best predictive model. Only with cTTO data, the final model has all coefficients that were statistically significant and logically consistent. In addition, it was parsimonious. This model had a mean absolute error of 0.027 and none out of 86 exceeded 0.1 of absolute error. Conclusions The final model in this paper appeared to predict the utilities of the states, which was valuated directly. This model could be used interpolate quality weights of all EQ-5D-5L health states.Min-Joo WooValuation34500Completed20132016
2016740A Dutch tariff for the Euroqol-5D-YouthBackground: It is unknown how framing of age impacts the valuation of health states within the context of the EQ-5D-Youth. A complicating factor in studying age dependency is that values are usually derived via trade-offs between Qol and life years. Those trade-offs may be calibrated differently across groups, rendering obtained values incomparableAim: To compare health state values across different age groups while accounting for time preferences. Methods: Participants completed a DCE that offered choices between two EQ-5D-Y states with a lifespan attribute attached. The choice model captured the value of a year in full health, disutility determined by EQ-5D-Y, and a discount rate. In addition, a 'QALY composition’ task captured strength of preference for different ways to produce a number QALYs, e.g. 2 years in full health or 4 years in 50% QoL. Participants were randomly assigned to fill out the survey for a hypothetical person of 10, 15 or 40 year. Results: 1938 people administered the survey for a child (35%) adolescent (36%) or adult (29%).Controlling for time preferences, we found strong agreement of health state values across age groups, except for pain which was valued lower for children (-0.43 v -0.52 for extreme pain). We found heterogeneous preferences for QALY composition in all arms and no clear pattern of differences. The probability distribution over response options varied most when levels for lifespan orseverity were at the extremes of the spectrum. People were less certain of their responses for children.Discussion: It seems not possible to compare DCE duration results across age groups, without accounting for differences in time preferences. Likely this result generalizes to TTO. The framing of age on health state values is limited but that might reflect local context in the Netherlands and may not generalize to other countries.Brigitte EssersYouth39750Completed20172018
2016600Pre-conference EQ-5D Short Course for South China Pharm-economics Forum 2016One half-day short course on EQ-5D on December 9, 2016 for attendees of the 4thPharmacoeconomic Forum of South China in Dongguan, ChinaNan LuoEducation and Outreach13790Completed20162016
2016110EQ-5D-5L valuation in Poland: a methodological extensionObjective Cost-utility analysis gain importance also in Central and Eastern Europe (CEE). We aimed to develop Polish utility tariff for EQ-5D-5L health states.Methods Face-to-face, computer-assisted interviews were collected in the representative sample. Each respondent followed a standardizedprotocol to collect 10composite time trade off (cTTO) and 7discrete choice experiment (DCE) observations. In the Bayesian approach, several model specifications were compared based on model fit, the usabilityof the final value set, and how they reflect the elicitation procedure (e.g. censoring). A hybrid approach (using cTTO and DCE data) was employedin the final set, which wascomparedwith the existing ones: EQ-5D-3L and EQ-5D-5L cross-walk.ResultsData from 1252 respondents (11,480 cTTO valuations and 8764 DCE pairs) were collectedin June-October, 2016. The final model accounted for random parameters, error scaling with fattails, censoring at -1, unwillingness to trade in TTO by the religious, and Cauchy distribution in DCE. Pain/discomfort impacts the utility most: disutility equal to 0.575 when at level 5. In the value set, 4.4% of EQ-5D-5L states are worse than dead. Thenew value set has a comparable range (minimum at -0.590 compared to -0.523) and same ordering of the first three dimensions (pain/discomfort, mobility, self-care) as EQ-5D-3L and EQ-5D-5L cross-walk value sets. Moreover,it is more sensitive to mild health worsening.Conclusions The new value set guarantees consistency with past decisions in cost-utility studies,while offering betterassessment of even mild health gains. It can be an option for CEE countries lacking own value sets.Dominik GolickiValuation29850Completed2015
2016700Selecting health attributes; the patients perspectiveObjectives To describe the concept of health, traditionally, the WHO definition is used. During the time this definition was first set out, morbidity mainly featured infectious diseases, while today chronic diseases are much more prevalent. Medicine has changed from ‘illness-centered ’to ‘patient centered’ by integrating the patient perspective. Additionally, with increasing life expectancy, more people receive ‘care’, instead of ‘cure’. In line with the shift from viewing disease as a state, to thinking of it as a process, the life-span perspective takes into account how a persons perceived health condition depends on how they grow, develop, and decline. There is also a concern that existing health instruments, which predominantly rely on consensus and expert opinion of health care professionals, have no adequate content coverage of the concept of health from the perspective of the individual patient, especially in regard to particular patient groups. These changes and concerns indicate that there is a demand for measures including other health aspects to evaluate treatment effectiveness. Therefore, the aim of the present study was to identify studies that reported on the development of instruments from the patients’ perspective and to describe which health aspects are relevant from this perspective. Methods Systematic review to identify studies that reported on the development of instruments from the patients’ perspective. Attributes were extracted from the papers, and categorized into domains and duplicate or equal terms were merged. The final list of attributes is presented, and compared to existing preference-based instruments (e.g., EQ-5D, HUI, SF-6D). ResultsAn electronic database search initially generated 855 records. After screening the titles and abstracts, 680 papers were excluded. The main reasons were that the paper reported on a ‘hard’ clinical outcome. Seventy-six papers remained for full text screening. During this process another 32 papers were excluded for various reasons. Mainly because of the concept (i.e. quality of care). From the remaining 44 studies, 801 attributes that are important from the patients’ perspective were extracted and categorized in four domains (mental, physical, social and other). Physical symptoms represented the largest group. Conclusions Existing generic preference-based instruments are relatively older instruments compared to the generic instruments developed with patient input, and therefore might not be representative for the current health care practice (patient-centered care). The generic instruments developed with patient input do not cover the entire life span. Youth and elderly are not represented in these measures. Therefore, the content coverage of the existing instruments might be inadequate.KM VermeulenDescriptive Systems44393Completed20172018
169-RAValue sets for EQ-5D-5L: A compendium, comparative review & user guideBackground: EQ-5D value sets are usually published in peer reviewed scientific journals, from which the scoring algorithms and other relevant information may be extracted. Szende et al. (2007) bundled information of all EQ-5D-3L value sets available at the time into a single book, which was well received as an accessible and useful resource for users. Since then, many EQ-5D-5L value sets have been published, yet no centralised resource such as the Szende et al. (2007) book exists for EQ-5D-5L value sets to date. Aim: The aim of this project is to develop a value sets book for the EQ-5D-5L. This book will include information on all EQ-5D-5L valuation studies and value sets so far, and inform our users on how this data has been collected and how the value sets differ from each other. Methods: Several researchers will be invited to contribute to one or more chapters of this book. These researchers are experts in the field and have contributed significantly to the EuroQol valuation protocols. Furthermore, information from (published) manuscripts of valuation studies will be collected, summarised and presented in a structured format, as in the previous book by Szende et al. (2007). In cases where information is missing from the manuscripts, this information will be requested from the PIs of those studies.Bram RoudijkValuation89072Completed2020
2016500Second Annual End-user and Scientific Advisory Committee MeetingOur proposalwas to host a meeting, notundertakea study. Through our meeting we educated and informed attendees on theuse ofEQ-5D and PROMs,including HQCA and AHS partners, Alberta Health Ministry Representatives and researchers from universities and organizations across Alberta and Canada.This has paved the way for more end-users of the EQ instrument which will lead to increased applications and more data being generatedand therefore, moreinformation to potentially call upon for future research.We also gained insight from end-users, particularly in the primary care setting, as to important implementation research questions to address. This input helps to inform our research agenda, about how EQ-5D is applied in PROMs programs.For our 3rd Annual APERSU End-User meeting, we will put the call out earlier to obtain a keynote speaker. We were waiting for a very long time to for a response from one of the people we contacted, in hindsight, we should have looked for alternates sooner. We have had comments from some potential attendees that the dates of our meeting conflicted with other meetingsbut with the diverse group of potential participants, it is difficult to get a time that would suit all.Jeffrey JohnsonPopulations and Health Systems15000Completed20162016
2016571(Proposal extension) Anchoring discrete choice experiment values at 0=dead for the EQ-5D-Y: additional data collection to control for instrumentObjectives:To date there have been no value sets tosupport the use of the EQ-5D-Y in cost-utility analysis. DCE can be used to obtain values on a latent scale, but thesevaluesrequire anchoringat 0 = dead to meet the conventions of QALY estimation.The primary aim of this study is to compare four preference elicitationmethods for anchoring EQ-5D-Y values. Methods:Four methods were tested: VAS, DCE (with a duration attribute), lag-time TTO and therecently developed‘location-of-dead’(LOD)element of the PUF approach. Adult respondents were asked to value both EQ-5D-3Lhealth states from an adultperspective(considering their own health)and EQ-5D-Y health states from a child perspective (considering the health of a 10yearold child). All respondents completed valuation tasks using all four methods,under both perspectives. For a subset of respondents the instrument was controlled for, i.e. EQ-5D-Y healthstates were valued under both perspectives. Results:Three-hundred and forty-nineinterviews were conducted. Overall, respondents gave lower values under the adult perspective compared to child perspective, withsomevariation across methods. The meanTTO value for health state33333was about equal to dead in the child perspective and worse than deadin the adult perspective. The meanVAS rescaled value for33333 was also higher in the child perspective thaninthe adult perspective. The DCE with duration results followed the samepattern, i.e. positive child perspective valuesandnegative adult perspective values, though themodels were notconsistent. The LODmedian rescaled value for 33333 was negative under both perspectives, and higher in the child perspective. When asked directly about their prioritisation preferences, 65%of respondentsindicatedthat treating adults andtreatingchildrenshould have same priority. Discussion:There was broad agreement across all methods. Values for 33333 tendedto be negativeforthe adult perspectiveand closer to 0for the child perspective. Potentialcriteria for selectinga preferred anchoring methodare presented. We conclude by discussingthe decisionmaking circumstances under whichutilities and QALY estimates for children and adults need to be commensurate in order to achieve allocative efficiency.Nancy DevlinValuation5600Completed20172018
2016680Screen size & data qualityIntroductionThe EQ-5D is frequently used to understand the development of health-related quality of life (HRQL) following injury. However, the lack of a cognition dimension is generally felt as disadvantageous as many injuries involve cognitive effects. We aimed to assess the added value of a cognitive dimension in a cohort of injury patients.MethodsWe analyzed EQ-5D-3L extended with cognition (EQ-5D + C) dimension responses of 5346 adult injury patients. We studied dimension dependency, assessed the additional effect of the cognitive dimension on the EQ-VAS, and, using the EQ-VAS as a dependent variable, determined the impact of EQ-5D and EQ-5D + C attributes in multivariate regression analyses.ResultsExtreme cognitive problems combined with no problems on other dimensions are uncommon, whereas severe prob-lems on other dimensions frequently occur without cognitive problems. The EQ-VAS significantly decreased when cognitive problems emerged. Univariate regression analyses indicated that all EQ-5D + C dimensions were significantly associated with the EQ-VAS. Exploratory analyses showed that using any set of five of the six EQ-5D + C dimensions resulted in almost identical explained variance, and adding the remaining 6th dimension resulted in a similar additional impact.ConclusionsThe addition of the cognition dimension increased the explanatory power of the EQ-5D-3L. Although the increase in explanatory power was relatively small after the cognition dimension was added, the decrease of HRQoL (meas-ured with the EQ-VAS) resulting from cognitive problems was comparable to the decreases resulting from other EQ-5D dimensions.John HartmanValuation14000Completed20162017
2013040An investigation of EQ-5D-5L values in the United Arab Emirates: a feasibility studyBackground:Nofive-level EuroQolfive-dimensional questionnaire(EQ-5D-5L) value sets are currently available in the Middle East toinform decision making in the region’s health care systems.Objec-tives:To test the feasibility of eliciting EQ-5D-5L values from ageneral public sample in the United Arab Emirates (UAE) using theEuroQol Group’s standardized valuation protocol.Methods:Valueswere elicited in face-to-face computer-assisted personal interviews.Adult Emiratis were recruited in public places. Respondents com-pleted 10 time trade-off tasks and 7 discrete choice experiment tasks,followed by debriefing questions about their experience of completingthe valuation tasks. Descriptive analyses were used to assess the facevalidity of the data.Results:Two hundred respondents were inter-viewed in December 2013. The face validity of the data appears to bereasonably high. Mean time trade-off values ranged from 0.81 for themildest health state (21111) to 0.19 for the worst health state in theEQ-5D-5L descriptive system (55555). Health states were rarely valuedas being worse than dead (6.2% of all observations; 10% of allvaluations of 55555). In a rationality check discrete choice experimenttask whereby a health state (55554) was compared with another thatlogically dominated it (55211), 99.5% of the respondents chose thedominant option. Most of the respondents stated that their religiousbeliefs influenced their responses to the valuation tasks.Conclusions:Our results suggest that it is feasible to generate meaningful health-state values in the UAE, though some adaptation of the methods maybe required to improve their acceptability in the UAE (and othercountries with predominantly Arab and/or Muslim populations).Keywords:EQ-5D-5L, quality of life, religion, stated preference, TTO,United Arab Emirates, utilities.Copyright&2015, International Society for Pharmacoeconomics andOutcomes Research (ISPOR). Published by Elsevier Inc.Nancy DevlinValuation20000Completed20132013
2016190National EQ-5D Symposium and Valuation WorkshopA total of 18 potential interviewers were selected to undergo the three‐day valuation workshop. Two esteemed EuroQol Board Members, together with the Investigator Team moderated the workshop. The entire workshop was held at Universiti Sains Malaysia, Penang. On the first day of the workshop, held at the Main Meeting Room of the School of Pharmaceutical Sciences, the participants were exposed to the various concepts and interview techniques pertaining to the Valuation Study. A series of comprehensive lectures were delivered by experts from the Investigator Team and by one of the invited representative from EuroQoL. The participants were also familiarized to the interview script by are corded mock session of the interview (Appendix1), besides practising the script among colleagues. These cond day of the workshop included mock interview sessions organized at the lobby are alocated beside the university’s Main Library. Each participant interviewed a minimum of five individuals using the official script of the Valuation Study. The Investigator Team evaluated each participant’s performance to ensure only well‐prepared interviewers are selected to participate in the Valuation Study. The final day of the workshop included a round‐table discussion held again at the Main Meeting Room of the School of Pharmaceutical Sciences. The participants had a chance of sharing their experience of conducting the mock interviews and raise any queries pertaining to the interviews. Generally, the participants shared that they enjoyed their new experience of interviewing and encountering interviewees from different walks of life. The invited guests from EuroQoL were generous in sharing their expert advice and experience in conducting such interviews. Additionally, the performance of the participants’ mock interviews was also formally reviewed using official quality control methods adopted by EuroQoL. Each of the 18 participants passed the performance review and were selected to represent the Valuation Study Team of Interviewers. Overall, the moderators of the workshop were very satisfied with the participation, enthusiasm, and energy of the participants of the workshop and hoped that that would follow through the entire Valuation Study.Asrul ShafieValuation14519Completed2016
2015410Writing a manuscript titled 'valuation of health outcomes using the time trade-off technique: the EuroQol protocols' for publication in the journal of PharmacoeconomicsThe time trade‐off (TTO) valuation technique is widely used to determine utility values of health outcomes to inform calculations of quality adjusted life years (QALYs) for use in economice valuation. Protocols for implementing TTO vary in aspects including the trade‐off framework, iteration procedure and its administration model and method, training of respondents and interviewers, and quality control of data collection. The most widely studied and utilized TTO valuation protocols are the Measurement and Valuation of Health (MVH) protocol, the Paris protocol, and the EuroQol Valuation Technology (EQ‐VT) protocol, all developed by members of the EuroQol Group. The MVH protocol and its successor, the Paris protocol, are developed for valuation of EQ‐5D‐3L health states. Both protocols are designed for a trained interviewer to elicit preferences from a respondent using the conventional TTO framework with a fixed time horizon of 10years and an iteration procedure combining bisection and titration. Developed for valuation of EQ‐5D‐5L health states, the EQ‐VT protocol adopts a composite TTO framework and makes use of computer technology to facilitate data collection. Training and monitoring of interviewers and respondents is a pivotal component of the EQ‐VT protocol. Research is under way aiming to further improve the EuroQol protocols which form an important basis for the current practice of health technology assessment in many countries.Nan LuoValuation13912Completed2015
20180320User-friendly tool to optimize (D)efficient DCE duration designs for the EQ-5D-Y / EQ-5D-5L.Objectives:This article explains how to optimize Bayesian D-efficient discrete choice experiment (DCE) designs for theestimation of quality-adjusted life year (QALY) tariffs that are unconfounded by respondents’time preferences.Methods:The calculation of Bayesian D-errors is explained for DCE designs that allow for the disentanglement of respondents’time and health-state preferences. Time preferences are modelled via an exponential, hyperbolic, or power discount functionand the performance of the proposed DCE designs is compared with that of several conventional DCE designs that do not takenonlinear time preferences into account.Results:Based on the achieved D-error, asymptotic standard error, and estimated sample size to obtain statistically significantestimates of the discount rate parameters, the proposed designs outperform the conventional DCE designs.Conclusions:We recommend that applied researchers use appropriately optimized DCE designs for the estimation of QALYtariffs that are corrected for time preferences. The TPC-QALY software package that accompanies this article makes therecommended designs easily accessible for health-state valuation researchers.Keywords:discrete choice experiment, health state valuation, quality-adjusted life year, time preferences.VALUE HEALTH. 2019Marcel JonkerValuation15750Completed2019
20190410US population norms for the EQ-5D-5LAim: The EQ-5D-5L is usedin many health-related fields, including patientclinical assessmentand cost-utility analyses. Inthese applications, a set of norms would be useful to allow for comparisons betweenthe population of interestand the general population. The aim of this study is to estimate population normsfor the EQ-5D-5L visual analogue and index-based scores.Methods: Data from the US EQ-5D-5L face-to-face (andpotentially the online)valuation studieswhich used age, gender and race/ethnicity-based quota sampling forthe US adult general populationwill serve as the data source. Conceptual and statistical considerations will be evaluated for appropriateness of using the face-to-face and/or the online study. Descriptive statistics (mean, standard deviation, interquartile range and 95% confidence intervals) will be reportedby gender/age bands for the VAS and EQ-5D-5L index-based summary scores using the value set based on the international protocol (Pickard et al, 2019).Significance: The availability of US 5L norms will help facilitate useand interpretationof the EQ-5D-5L in the United States for both economic and non-economic purposes, thereby encouraging use of the measure.Simon PickardValuation17600Completed2019
2013280An Irish valuation study for the EQ-5D-5LThe objective of this study is to derive a preference valuation set for the EQ-5D-5L health states from the Irish general public, following the standard protocol for such studies developed by the EuroQol Groupand examine variations across identifiable population groups in respect of preferences. A random sample (N=1,000) from the Irish population will be asked to value a standard subset of EQ-5D-5L health states,usingcompositetime-trade-off and discrete choices, in face to face interviews. Interviews will be conducted by a small team of postgraduate researchers with background knowledge of valuation methods, and who will be trained in the interviewing process and EQ-VT software.Regression analyses will be applied to the valuations of those health states. By means of the estimated regression coefficients, which together constitute the so-called Irish value set, valuations for all possible 3,125 EQ-5D-5L health states willbe derived. This set of social health status preference valuations ensures the application of the EQ-5D-5L in cost-utility analysis for health care policy and clinical assessment in Ireland.Variations in preferences related to private health insurance status will be examined to shed light on potential differences in preferences for health. The study will provide an opportunity to investigate emergent methodological issues relating to EQ-5D-5Land Iand wewould welcome the opportunity for this study to facilitate this.Ciaran O'NeillValuation14000Completed2014
2014040Japanese participation in the MAT-endorsed ranking tasknot availableShunya IkedaValuation40097Completed2014
2013070Deriving social values using the EQ-5D-5L in the general population of Uruguay.The results of this study have been published here: Augustovski F, Rey-Ares L, Irazola V, Garay OU, Gianneo O, Fernández G, Morales M, Gibbons L, Ramos-Goñi JM. An EQ-5D-5L value set based on Uruguayan population preferences. Qual Life Res. 2016 Feb;25(2):323-333. doi: 10.1007/s11136-015-1086-4. Epub 2015 Aug 5. Erratum in: Qual Life Res. 2016 Feb;25(2):335. PMID: 26242249.Federico AugustovskiValuation50796Completed2013
20170440Creating a laboratory for testing differences between the 3L and 5L Index in patient populations: simulating profile and valuation dataWe created the ‘simulation laboratory’ as a suite of modules with associated data sets by iterative development of software. In doing this, the specific form and structure of the laboratory evolved from the prototype plan but resulted in the same functionality. We presented our initial plan at the mid-year meeting in Budapest in 2018 and demonstrated the results of the final structure at the plenary meeting in Lisbon later that year. Since then, we have concentrated on making the software more user friendly, written a user guide and added to the associated data sets.David ParkinDescriptive Systems42875Completed20172018
2013030Development of Thai population-based preference scores for EQ-5D-5LSUMMARY This research aimed to develop the population-based preference scores for the EQ-5D-5L (the 5L) Thai version. To elicit population-based preference score, face-to-face interviews using the EQ-VT protocol was undertaken in 12 provinces across Thailand. A 1,207 representative sample was recruited using a stratified stage sampling and quota sampling by age and sex. Six interviewers were recruited and well-trained for 1 month prior to work. For TTO valuation, 86 health states were grouped into 10 blocks. Each block contained 10 health states. For the Discrete Choice Experiment (DCE) valuation, 196 health states grouped into 28 blocks of 7 pairs of health states were used. For each participant, the block used for TTO valuation and DCE valuation were randomly selected by the EQ-VT software. The Thai algorithm were estimated using the lead-time TTO method and random effect model with only main effect. It had no constant and interaction terms as shown in below table. No inconsistency among 3,125 possible health states was found. Since no constant in the algorithm, the full health (11111) can be directly calculated from this algorithm without assume to be equal to 1.00. The second best score was 0.968 for state 11112 and the worst score is -0.283 for worst state (55555). The number of health state which the preference score is lower than zero (worse than dead) is 126 health states (4.03%). Mobility had the greatest impact on preference score. The DCE valuation generated inconsistent coefficient in the regression model, indicating the need to further examine the cause of inconsistency and how to deal with these problems. The 5L could be recommended as a preferred health-related quality of life measure in Thailand.Juntana PattanaphesajValuation38244Completed20132014
2016660How to capture fluctuating health impairments: Testing intensive longitudinal assessment of the EQ-5D-5L in multiple sclerosisFor each topic, a summary of findings, implications for the ILAincluding the respective rationale, and sample quotes from participants are listed in Appendix FJohn BrazierDescriptive Systems49437Completed20172018
20190940Health state utility rescaling and interpersonal comparisonsFunding applicationfor a one-week research visitThis proposal aims to bring together three ongoing projects:Jakubczyk, Sampson, and van Hout/Schneider independently conceived researchrelating to utility rescaling. While the analyticalapproaches and rationales differ, all three share the notion that EQ-5D health state utilities are not directly comparable across individuals. Most notably, comparisons between those individuals who do, and those who do not consider some health states to be worse than dead are problematic. Therefore,utilityvalues need to be rescaled (e.g. min-max normalisation), before they can be aggregated into a social value set. A collaboration between the three research projectswill be established toexchange knowledge and conduct comparative analyses. Funding is sought for a one-week period of focused collaborative work. During this period,the rescaling methods and their conceptual foundationswill be discussed, compared and contrasted. This will allow further refinement and contribute to a deeper, more systematicunderstanding of interpersonal utility comparisons in the context of EQ-5D health state utility values.The collaboration will result in a working paper and also create‘rescaled’ EQ-5D 5L value sets, based on preference data from England and/or Poland.Ben Van HoutValuation12200Completed2020
2013290Investigating the validity of values worse than dead estimated using DCE with durationBackground: DCE with duration (DCEtto) is an alternative protocol for deriving population value sets for instruments such as the EQ-5D-5L. Valuation studies of the EQ-5D-5L so far have used the EQ-VT which only included DCE without duration questions. In Canada, we sought to estimate the DCEtto by combining the EQ-VT with additional DCEtto questions. Our objectives were twofold: 1) to determine the feasibility of generating the DCEtto from this combination of EQ-VT and DCEtto questions; and 2) to further investigate the validity of the DCEtto. The latter involved comparing individual values from the conventional TTO and the DCEtto.Methods: Data from 1107respondents in the Canadian EQ-5D-5L valuation survey are used. Each respondent completed the EQ-VT (composite TTO, and DCE without duration) followed by an additional survey which included 2 DCEtto tasks. These DCEtto tasks where selected using experimental design theory whereby the DCE without duration tasks were assumed to have equal duration, and combined with the DCEtto data to create a single dataset amenable for the analysis. For our first objective we useda conditional logit modelto estimate coefficients from the combined data. For our second objective ran mixed logit models and used hierarchical bayes to estimate individual values from the DCEtto. We then compared these to the TTO values.Results: The DCEtto provided coefficients that were statistically significant and logically ordered. Values ranged from 1 to -0.71(for health state 55555). Thiscontrasted to a range of 1 to 0for the TTO, suggesting clear differences in cardinal values derived from the two different protocols. The DCEtto values were better distributed than the TTO, which for severe health states suffered from clusters of values at 1, 0.5, 0 and -1. However, there was limited consistency between severe health states between the DCEtto and TTO. Conclusions: Values for the EQ-5D-5L were successfully generated using the DCEtto by simply adding two questions to the EQ-VT. The model generated plausible values, and the model coefficients were similar to those seen in previous DCEtto studies.The DCEtto values for severe health states were lower than those generated from the conventional TTO.Nick BansbackValuation18750Completed2014
2016510ISPOR Workshop: DETERMINING UTILITY OF MULTI-ATTRIBUTE HEALTH STATES: NEW MEASUREMENT AND ANALYTIC APPROACHESWe drafted the workshop proposals together, based on the workshop previously held at the ISPOR 7thAsia-Pacific Conferencein Singapore. Following the conference in Singapore, the slides and content was refined and discussed by email prior to the trip, and face-to-face in Vienna. The workshop was conducted Wednesday November 2nd, in the very last slot of the conference. The contents were, very briefly, background and introduction (Nan), quality control issues (Mark), DCE-modeling and constrained models (Kim), and the hyreg command (Juanma). We extensively explained how to set-up the data file to run the commands, but also we focusedthe attention on interpretation of the output, specifically the most complex, i.e., the heteroscedastic models.Juan M. Ramos-GoñiValuation9885Completed20162016
2016350Symposium Sponsorship & Panel Session: 38th Annual North American Meeting of the Society for Medical Decision Makingnot availableDavid WhitehurstPopulations and Health Systems15000Completed20162016
2016440Taiwan valuation study for the EQ-5D-5LObjectives:This study was aimed to develop thefive-level EuroQol five-dimensional (EQ-5D-5L)value set for utility calculation in the Taiwan population. Methods: The 1000 representative sample aged 20 years and over was recruited from six geographic regions in Taiwan. In the computer-assisted interview using EQ-VT 2.0, participants completed 10 composite time trade-off (C-TTO) tasks and 7 pair discrete choice experiment (DCE)tasks. The value setsweregenerated and compared in C-TTO, DCE and hybrid modeling using twenty dummy variables as main effectsapproach.Results: A representative sample of the Taiwan general population completed the valuation study. The hybriddatawithleft-censored at -1was considered as themost preferred model for EQ-5D-5Lvalue set in Taiwan.The possible values for the predicted utility ranged from -1.0259to 1. Conclusions:A representative EQ-5D-5L value set derived from hybrid modeling is now available for Taiwan.The EQ-5D-5L tariffcan facilitatethe health economic evaluations and health decision making on resource allocation under national health insurance program in Taiwanafterward.Hsiang-Wen LinValuation24000Completed20162017
2015260Hybrid model in RThis was not a study in the normal sense, but the process has been helpful in identifying several issuesrelated to hybrid modelling that call for further investigation. The R hyreg command will make hybrid regression available to a larger audience, and unlike STATA, is completely free. The xreg function will hopefully simplify future modelling efforts both related to EQ-5D, and in general.Kim RandValuation14150Completed2015
120-RAThe role of time and lexicographic preferences in valuation of EQ-5D-Y for health states better and worse than dead.Background: There is evidence that adapted wording influences the utility assigned to health states derived from EQ-5D-Y, compared to health states derived from EQ-5D-3L. As a result, EuroQol has proposed a valuation protocol for EQ-5D-Y (Ramos-Goni et al., 2020). In this protocol, instead of an individual perspective as is usual in EQ-5D-5L valuation, adults value health states that are described to affect a 10-year-old child (i.e. a proxy perspective). Aim: Our study aims to explore the degree to which time and lexicographic avoidance of immediate death can explain why EQ-5D-Y utilities elicited with a proxy perspective are generally higher than those from an individual perspective. We answer the following research questions: 1) What is the influence of time preferences on EQ-5D-Y utilities and can this influence be corrected for? 2) What is the difference between EQ-5D-Y utilities elicited by composite Time Trade-off (cTTO) and EQ-5D-Y utilities elicited by lead-time TTO exclusively? 3) If a difference exists, is this related to time preferences, lexicographic preference for immediate death (PFID), or both? Methods: We measure time preferences for each individual (i.e. discounting of a 10-year-old child’s life years) by means of the direct method. The role of lexicographic PFID will be assessed by comparing TTO utilities for 2 severe and 2 mild health states elicited in two experimental arms: cTTO is and lead-time TTO only. The role of time and lexicographic preferences can be studied by comparing these 2 arms before and after correcting for time preferences.Stefan A. LipmanValuation, Youth25050Completed2020
20170510Valuation of the EQ-5D-5L in the French populationAccording to the World Health Organization (WHO), quality of life is defined as “an individual's perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards and concerns. It is a broad ranging concept affected in a complex way by the person's physical health, psychological state, personal beliefs, social relationships and their relationship to salient features of their environment.Quality of life measures are obtained from responses provided by subjects to standardized questionnaires. Today there are multiple scales of quality of life related to health: they are considered to be specific if they explore quality of life related to a particular disease, orgeneric if they are common toany condition. Among the existing generic tools, theEuroQolEQ-5Dquestionnaire istoday the most widely used descriptive system of health statesto capture health related quality of life in the world(1,3). Each health state generated by the questionnaire is associated to a so-called “tariff” (or utility value), which reflects its relative value ranging from -1 to 1. These values are then used to assess the outcomes of health products, servicesor programs in cost-effectiveness analyses. Associated to a time spent in a given health state, they yield a number of Quality AdjustedLife Years (QALYs), which are literally the number of life years provided by a given intervention weighted by the quality of life experienced by the beneficiary of the treatment. The EQ-5D questionnaireis adescriptive system of health statesbased on five dimensions, which are considered as being essential: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each dimensionis reported with three levels of severity (EQ-5D-3L) and more recently with five (EQ-5D-5L). In the 3level questionnaire, respondents can score each dimension on three levels, thus leading to 243health states(plus death, scored 0 by convention). For each health sate a value from -1 (theoretically considered asthe worst possible health state)to 1 (perfect health) is associated, the estimation of which is based on surveys of stated preferences in a representative sample of the population of each country. In France, the valuation of EQ-5D-3L health states was undertaken by the Chair of Health Systems at ESSEC Business School and the French value set is recommendedby the French High Health Authority(HAS)for the submission of cost-effectiveness studies when required by regulation for access to coverage and pricing. (4).The 3 level version of EQ-5D has been considered by many researchers as limited in capturing small differences in health levels for certain conditions. This lack of sensitivity has been the main motivation for the EUROQOL consortium to develop the 5 level version of the EuroQol questionnaire, the EQ-5D-5L. Throughout the development of the 5L, the five dimension of the original EQ-5D-3L are maintained but the number of levels for each dimension was increased to 5 (5). This new version of the questionnaire leads to a total of 3125 health states (55) compared with only 243 from the 3-level previous version. Because of its better accuracy to describe the health values related to a given health state, EQ-5D-5L will now become the reference in the coming years. Indeed, several publications have demonstrated an increased sensitivity of this new version to capture variations in health related quality of life of patients (6-15).Gerard De PouvourvilleValuation25000Completed2017
20190360Comparison of EQ-5D-3L and 5L value sets/scoring for US usersAim: Directly-elicited and mapped value sets for both the EQ-5D-3L and the EQ-5D-5L are now available in the US. As stakeholders contemplate transitioning from one instrument to the other, an important consideration for decision makers will be the comparability (and differences between)value sets/scoring functions for the 3L and 5L. The objective of this study is to examine the comparability of scoring functions for the 3L and 5Lin the US.Methods: The properties of the theoretical value set over the entire spectrum of health states described by the value sets will be evaluated. These characteristics include: modality of kernel density plots, percentage of health states with values less than 0, mean level transitions, and range of scale. The value sets will then be applied to respondents from the Crosswalk, US valuation, and if available, the simulation lab (Parkin et al)to compare empirical properties. Bland-Altman plots will be generated to compare agreement between value sets. Discriminative ability will be assessed by ability of value sets to distinguish between known groups using F-ratios. Responsiveness will be analyzed using effect size and standardized response means. Ceiling and floor effects of value sets will be assessed using proportion of utility values greater than 0.95 and less than 0, respectively.Significance: As there has been significant discussion ex-US regarding the transition between descriptive systems and thereby value sets,we propose this work in anticipation of similar interest and issues among users of the EQ-5D in the US. The results of these analyses will inform applications of the EQ-5D in the US.Simon PickardValuation25400Completed2019
20190170Which utility function do respondents (actually) use when completing DCE-duration choice tasks?not availableMarcel JonkerValuation29000Completed2019
2013320Reintroduction of the ranking task in EQ-5D valuation. Improved data quality and reduced level of inconsistencies?Background:The EuroQol group has observed several problematicissues related to the quality of valuation data collected using the EQ-VT system. The objective of this study was to test if reintroducing a ranking task prior to TTO valuation could help to reduce inconsistencies and improve the quality of thedata in future valuation studies.Method:The study was performed in Spain (Canary Island). Respondents were assigned to three different study arms. Thecontrol group (Arm 1) wasthe current EQ-VT version, which did not include a ranking task. The second arm (Arm 2)wasthe EQ-VT preceded by a ranking task similar to the one used in the MVH protocol, in which the respondents wereasked to rank the target health states using physical cards. The last arm (Arm 3)wassimilar to the second, with ranking prior to the EQ-VT. Unlike the second branch, the ranked states remainedvisible through the Composite TTO (C-TTO) tasks, and the C-TTO states order weredetermined by the rank order. We usedjust one specific block of 10 C-TTO states. A total sample size of 600(~200 per experimental arm) respondentswererandomly selected with representatives in term of age and gender ofCanary Islands population. All interviews wereperformed face to face. The interviewers performed similar number of interviews in the different experimental branches. We comparedthe number of C-TTO-based inconsistencies.Results:The final sample size was 600 respondents (196, 205 and 199 for arms 1, 2 and 3 respectively). The percentages of times when the state 55555 was not valued as the worst state on C-TTO responses by study arm were 13.27%, 10.73% and 13.57% for arms 1, 2, 3 respectively. However, none of the observed differences in levels of inconsistent responses were statistically significant.Conclusions:Under the light of theseresults it does not make sense to increment the current burden of EQ-VT by adding the ranking task as it does not improve the data quality neither the spikes on C-TTO observed values distributions.Juan M. Ramos-GoñiValuation64000Completed2014
2015050Important aspects of (full) health not captured by EQ-5DBackground:Studies to produce value sets for preference-based measures of health require a full health upper anchor to be defined if the values are to be used to calculate quality-adjusted life years.A recentempirical study assesseddifferences in time trade-offvaluations using two comparator health states: 11111 and full health. The primary results of thatstudy suggest that EQ-5D-5L health state values are not greatly affected by whether 11111 or full health is used as the comparator state.The study also generated data on how survey respondents interpreted those descriptions, and on what aspects of health they consider to be missing from the EQ-5D descriptive system.Aims:To report the methods and findings of an analysis of the qualitative data collected relating to the following questions: Whatmakes 11111 and full health different? What important aspects of health are not captured by the five EQ-5D dimensions?Methods:Data were collected froma broadly representative sample of the UK general publicvia face-to-face interviews. After completinga valuation questionnaire using EQ-VT, respondents completeda short pen-and-paper follow-up questionnaire. Thefollow-up tasks were developed by the study team to elicit additional information about the comparability between 11111 and full health and respondents’ interpretations of those concepts; and to inform the ongoing research agenda around the measurement of health beyond the existing EQ-5D descriptive system.A combination of descriptive and thematic techniques were used to analyse the data. Results:Interviews were conducted in 2014. Data are available for 436 respondents. A sizeable minority (30%) of respondents did not agree when asked whether they considered 11111 and full health to be the same as each other. When asked what makes them different, many respondents suggested that the five EQ-5D dimensions are not exhaustive of all conditions and health problems. Sensory deprivation (particularly vision and hearing) and mental health were the aspects of health most commonly suggested by respondentsas being important but not captured by EQ-5D.Discussion and Conclusions:To some extent, the findings of this study support the choice of areas in which developmental bolt-on work has been conducted to date. Further research is required to establish whether and what people would be willing to trade for improvements in the aspects of healthnot currently included in the EQ-5D.We hope that this study can provide a basis for more detailed qualitative and quantitative research to inform further review of theEQ-5D descriptive system.Koonal ShahValuation7500Completed2015
2014140Quantification of EQ-5D health-state values by scaling similarity data (studies 1 and 2)Objective: It is assumed that the 'Q' in Quality-adjusted life-years (QALYs) is unidimensional. However, this assumption has not been sufficiently tested which is what the current study aims to do. It is important to note that the response task might influence the dimensionality of the data. The current study should be considered a pilot study using a novel methodology to test the dimensionality of rank-ordered health states.Methods:The rank data of 2997 respondents from the Measurement and Valuation of Health (MVH) study was exploded into paired comparison data. Pairwise choice probabilities were transformed into proximity data by |,5.0ijijPwhere ijis the dissimilarity between health states iand jand ijPisthe probability of health state ibeing preferred to health state j. The proximity data were entered into a similarity matrix. Subsequently, this matrix scaled with metric and non-metric multidimensional PROXSCAL algorithms in SPSS in 1, 2, and 3 dimensions. Dimensional fit was assessed usinginterpretability,stress measures,and scree plots.Results:Based on interpretability the unidimensional solution had the best fit. State '33333' was scaled at the end of the continuum while all the states with a single attribute level at level 2 were scaled at the other end of the continuum. In terms of stress and the scree plot a two-dimensional solution would indicate a meaningful improvement over the unidimensional solution. However, the interpretability was lower compared to the unidimensional solutions. The three-dimensional solutions did not decrease stress measures meaningfully, nor increase the interpretability. Discussion: Based on the current findings we can infer that EQ-5D-3L rank-based data is scaled best unidimensionally, even though statistical fit leaves room for improvement. Other types of data could be put through the protocol of the current study to test the assumption of unidimensionality as well. The time trade-off data would seem like a prima candidate for this type of follow-up research.Alexander AronsValuation56250Completed2014
2016090Developments in preference-based measures of health: scoring approaches and guidanceSummary upon completion of workshop: This workshop helped to promote the research and agenda of the EQ group, particularly its work in valuation. ISOQOL organizers were happy with the workshop as there was a lot of interest, with 18 registrants. Of them, 12 completed the workshop evaluation survey designed by ISOQOL. It is attached. To summarize, the ISOQOL workshop was positively received by the attendees. Several were conducting valuation studies and they were very interested in the content. Thus, although it was designed for the relative novice, the content was current and relevant to scientists with different levels of experience. There was a slight issue with the materials, as the email containing the pre-reading papers was not received by all the attendees. This was a logistic issue that was mediated by the ISOQOL office and was beyond our control. Otherwise, the ratings were generally high, and many indicated interest in a full-day workshop.Simon PickardEducation and Outreach13050Completed20162016
2016120A comparison of the validity of the two proxy versions of the EQ-5D-Y instrument in acutely ill and chronically ill children in South Africa. A cross sectional analytical descriptive studynot availableJennifer JelsmaYouth5000Completed20162016
2016210Measuring concordance between patient and proxy raters of EQ-5D using fuzzy set theoryBACKGROUND: The reliability of raters is important to evaluate in quality of life research. For instance, proxy raters may be sought when patients have difficulty to respond on their own behalf. It is, thus, vital to understand how multiple assessments interrelate and develop informative measures of concordance (MCs). However, commonly used MCs have limitations. Pearson’s (Spearman’s ) correlation coefficient, 𝜌, focuses on the quality of fit, not reflecting systematic differences in scores. Linear regression coefficients, 𝛼, are diluted (drawn towards 0) and heteroscedasticity can beproblematic. Intraclass correlation coefficients (ICC) may face conceptual difficulties as the observations are not exchangeable (patient vs proxy) and the judges vary between pairs. Additionally, all of the above (but regression coefficients) are symmetric. Our aim was to propose the use of MCs derived from fuzzy set theory and illustrate their usefulness. METHODS: Data from a longitudinal natural history of stroke study was used, which included patient & proxy assessments of the EQ-5D-3L (n=124 dyads). We focused on the EQ VAS assessment in four waves (months: 0, 1, 3, and 6). VAS assessments were rescaled into [0,1] intervals, and the score was interpreted as a membership function from fuzzy set theory measuring how much a given patient’s health state resembles perfect health. We first defined several asymmetric MCs measuring the degree to which a patient considering their own state as perfect (say, 𝑥 ∈[0, 1]) implies the proxy considering this state perfect (𝑦∈[0, 1]), or vice versa. For that purpose we used various OR operators (Zadeh, bounded sum, product) along with 𝑥 ⇒𝑦⇔¬𝑥∨𝑦 equivalence. We then defined the both-way counterparts as conjunction of implications plus a ratio of a (scalar) cardinality of intersection and of union. All the measures yield values in [0,1]. The results were averaged for the whole dataset. We measured the asymmetry of implications as the difference between 𝑥⇒𝑦 and 𝑦⇒𝑥. Bootstrapping was used to calculate 95% confidence intervals (95%CI).RESULTS: For the first wave we obtained standard MCs: 𝜌= 0. 359, 𝛼= 0. 301 or 𝛼= 0. 427(explaining patient with proxy, or vice versa, notably both <1), 𝐼𝐼𝐼(2, 1)= 0. 31. We got the assessment of patient ⇒ proxy with fuzzy measures in the range 0.583–0.858 and of proxy ⇒patient: 0.678–0.953. The implication was stronger in the latter direction (statistically significant, i.e. bootstrap 95%CI above 0 for all the measures). The differences for other waves did not show this significance. Hence, considering a given state as perfect by proxy implies agreeing by patient (considering health worsened by patient implies agreeing by proxy) more strongly than vice versa, but the asymmetry disappears in time. The symmetric MCs for the first wave ranged from 0.536–0.811 (0.711 for ratio of cardinalities).DISCUSSION: Concordance between various utility assessments can be defined using fuzzy set approach, also allowing to define a directed (asymmetric) version, and reflect possibly non-trivial interrelations. Statistical inference is still possible, e.g. via bootstrapping. To build on this basic conceptual framework, future research and testing on different datasets will help to verify properties and enhance interpretation of the results.Simon PickardValuation7500Completed20162016
20170170Deriving Social Values for the EQ-5D 5L in Peru implementing a “Lite” protocolThe main deliverablesplanned were: a draft of preliminary results and a draft of report/paper to be shared with the EuroQol team, that was accepted and going to be presented at the Brussels plenary; technical report published with the peruvianauthorities with the study details, analysis, results, and the peruvian eq-5d-3l value sets.We aimed to deliver at least 2 papers to be published in an international peer reviewed journal: a) one about the valuation set using the lite protocol and hybridmodelling (to be undertaken in the near future)b) about the comparison of the DCEvariants. A collaboration was done with Australian and Danish team and another paper sent to the plenary lead by Alice Yu about the DCE triplet comparison. After the revision of the project, and resubmission, the teaching-distance learning material was not planned nor produced.Federico AugustovskiValuation39700Completed20172018
2014060Separation of the BTD and WTD task in TTOBACKGROUND: The first series of EQ-5D-5L valuation studies reported in their TTO data many inconsistent responses involving state 55555 (20%) or the mild state (10%). Mistakes at the sorting question identifying whether a state was considered better than dead (BTD) or worse than dead (WTD), strategic behavior, learning effects, interviewer effects, and ordering effects may have contributed to the inconsistency rate. AIMS: To explore if composite TTO data consistency is promoted by 1.A feedback module offering respondents the opportunity to review their responses and take the wrong ones out, if any. 2.Separation of the BTD and WTD task, moving all WTD questions to the end of the TTO task METHODS: The effect of the modifications was tested in the Netherlands and Hong Kong using a 2-arm study with a crossover design. In each country, six interviewers conducted interviews. Three started using the current EQVT, switching to the split version after 25 (the Netherlands) or 30 (Hong Kong) interviews, and back after another 25 until reaching the target 400 samples. The others started in the split version and also switched between the versions in a same manner as the previous group. We compared the study arms on consistency, interview duration, and cognitive debriefing, thereby also considering the impact of interviewer (learning) effects. The feedback module was offered to all respondents. The same effects were evaluated, but ́within subject ́. RESULTS In the Netherlands and Hong Kong respectively 404 and 403 people participated in the experiment. In the Netherlands, 17.8% of respondents provided one or more inconsistent responses. The FB module lowered this rate to 10.6% (p=0.003). In Hong Kong, 31.8% of respondents provided one or more inconsistent responses. The FB module lowered this rate to 22.3% (p=0.003). Completion of the FB module took around 2 min. About one-third of the respondents used the FB module to remove one or more responses. In total about 5% of responses was removed. In the Netherlands, the inconsistency rates were higher in the current version than in the split version, both before the FB module (19.5% vs. 16.1%, p=0.37) and after (13.2% vs. 8.0%, p=0.095). In Hong Kong, the opposite was found, both before the FB module (28.0% vs. 35.5%, p=0.107) and after (16.0% v. 28.6%, p=0.002). DISCUSSION Both the FB module and the separation of the BTD and WTD tasks improved consistency in the Netherlands, but in Hong Kong only the positive impact of the former was confirmed.Elly StolkValuation61750Completed2014
2016140Two small DCE projectsBackground:Although much progress has been made on identification efficiency in the design of discrete choice experiments (DCE; e.g., D-efficiency), less is known about allocation efficiency, such as number of pairs and number of responses per pair. The aim of this project was to investigate the effectof number of pairs and pair-specific sample sample size on the variability in DCE results(i.e., standard deviation and error). Methods: We examined data from three DCEvaluation studies for the EQ-5D-5Lfrom the Netherlands: two of the studies used face-to-face interviews and onewas an online self-completed survey.This analysis was conducted in three steps:First, we conducted a series of analyses to create a “base case” specificationto act as the primary comparator forsimulationson variability. In these analyses, we applied2 cumulutative density function (CDF) specifications(sigmoidal[logit]and ratio-based[Bradley-Terry]) and 3 estimation techniques (Weighted least squares (WLS) using the observed choice probabilities, WLS using the predicted choice probabilities and Maximum Likelihood). Theresulting base case included 186 pairs with a mean of 183 observations per pair, weighted according to mode of administration(30% face-to-face; 70% online)and specifieda 20-parameter main effects logitCDF estimated by maximum likelihood. Second, we simulated 30, 40, 50, 60, 70, 80, 90 and 100 responsesper pairfor 186 pairs and estimated the base-case model. For each of the 1000 bootstrap iterations, we compared the results to the base caseusing three criteria: median coefficient standard error, the number of rejected predictions, and chi square of predictions. Third, weran aBayesian efficient design algorithm that takeslevel balance into account to generate design sizes of 25, 30, 40, 50, 75, 100 and 150 pairs drawn from the 186 pair design of the base case.We simulated 30, 50 and 70 responsesper pair for each design sizes:25, 50, 100 and 150 pairs(similar to the second step). Results:Among the findingsfrom this brief study, three results areparticularly noteworthy for the EuroQol Group: (1) future attempts tomanuallyenterpairs should take scale into account(i.e., adding the 10 mild pairs to the EQ-VT DCE may have created a bias in logit estimates); (2)the concept of the “pragmatic p” may serve as the basis for determining number of responses per pair and range of pair probabilities in future studies; and (3) Infuture EQ-VT studies, the number of responses per pair should increase (>50) and the number of pairs should decrease (~150).Mark OppeValuation10500Completed2016
2015270HRQoL among patients seeking treatment for abuse of illicit substancesThe project has been severely delayed due to a number of factors. First, the regulations around patient data security were changed, resulting in a long process of getting everything approved. When this was in place, co-investigator Espen Arnevik had changed his formal position, and Rand was considerably more busy than anticipated. The result was an intermittent process of writing by relay, eventually resulting in submission of a manuscript in March of 2020. Fortunately for us (unfortunately for the field of research), there has been very little in the way of publication on HRQoL in SUD patients, and the manuscript is still relevant. Main methods 178 SUD patients were administered EQ-5D-3L at treatment start. SUD patients and a general population sample in the same age range were compared in terms of reported EQ-5D-3L health states, problems by dimension, UK index values, and EQ VAS scores. We investigated specific drug dependence, mental health disorders, sex, age, and education as predictors of EQ-5D-3L values and EQ VAS scores. Anxiety/depression dimension scores were compared to Hopkins symptom Checklist (HSCL-25) scores. Main results 91.6% of the patient sample reported problems on the EQ-5D-3L, with 29.8% reporting extreme problem. Corresponding proportions from the general population sample were 39.8% and 3.0%. The most frequently reported problems were for anxiety/depression, pain/discomfort, and usual activities. Mean UK index values were .59 (SUD) and .90 (general population). Mean VAS scores were 59.9 (SUD) and 84.1/general population). Regression analyses identified phobic anxiety diagnosis and cocaine dependence as statistically significant predictors of higher EQ-5D-3L index scores. Conclusions EQ-5D may be a useful and practical instrument for monitoring HRQoL in SUD patients, though there is need to demonstrate that the (ideally -5L version) instrument is sensitive to changes, not just to overall symptom burden.Kim RandOthers12000Completed2015
20190270Modelling dependence in EQ-VT DCE data: impact on value setsBackground.In health economics, there has been interest in using discrete-choice experiments (DCEs) to derive pre-ferences for health states in lieu of previously established approaches like time tradeoff (TTO). We examined whetherpreferences elicited through DCEs are associated and agree with preferences elicited through TTO tasks.Methods.We used data from 1073 respondents to the Canadian EQ-5D-5L valuation study. Multivariate mixed-effects modelsspecified a common likelihood for the TTO and discrete-choice data, with separate but correlated random effects forthe TTO and DCE data, for each of the 5 EQ-5D-5L dimensions. Multivariate latent class models allowed separatebut associated latent classes for the DCE and TTO data.Results.Correlation between the random effects for the 2tasks ranged from20.12 to 0.75, with only pain/discomfort and anxiety/depression having at least a 50% posteriorprobability of strong (.0.6) correlation. Latent classes for the TTO and DCE data both featured 1 latent class cap-turing participants attaching large disutilities to pain/discomfort, another capturing participants attaching large disu-tility to anxiety/depression, and the third class capturing the remainder. Agreement in class membership was poor(kcoefficient: 0.081; 95% credible interval, 0.033–0.13). Fewer respondents expressed strong disutilities for problemswith anxiety/depression or pain/discomfort in the TTO than the DCE data (17% v. 55%, respectively).Conclusions.Stated preferences using TTO and DCEs show association across dimensions but poor agreement at the level of indi-vidual health states within respondents. Joint models that assume agreement between DCE and TTO have been usedto develop national value sets for the EQ-5D-5L. This work indicates that when combining data from both tech-niques, methods requiring association but not agreement are needed.Eleanor PullenayegumValuation31800Completed2019
2013120Rescaling relative health-state values from discrete choice experiments unbiased onto the QALY metricBackground:More attention is devoted nowadays to simple discrete choice(DC)tasksbetween descriptions of health states to obtain ordinal data that can subsequently usedin statisticalmodels to derivehealth-state values. Such values are often used for the computationof quality-adjusted life years(QALYs). A requirement of the conventional QALYs is that there should be a fixed localization of dead atzero. One strategy in DC experiments is to include the state‘dead’ as a choice optionthat may be preferredover states considered as worse than dead. However, this may become cumbersome if subjects have difficulty with the appraisal of ‘dead’ and consider all health states at least better than dead.Objective: To introduce a novel method that calibrates health states and the localization of ‘dead’together in a less biased way.Methods: Data from the Measurement and Valuation of Health study were analyzed. Three different probit modelswere estimated. Thefirst model(classical Thurstonianmodel)used response data fromassessmentsofholistic(no consideration of the levels of the attributes)health states andthe state ‘dead’. Subsequently, the values for the best and worst state of this modelwere used as anchors for another model without‘dead’. This second model used dummies for thelevels of thehealth-state attributesand was calibrated on the range of the first model. The result of thesecombinedmodelswerecompared to a ‘traditional’model. Here,attribute-level dummy variables and a ‘dead’-dummy variablewere analyzed.Results: All three modelsproduced regression coefficientsthat werelogically orderedand estimated health-state values wererathersimilar.Sensitivity analysis of the novel calibration method showed that the proportion of people who considered certainstates as worsethandead does not affecttherelative health-state values.Conclusion: The novel calibration method proved to be feasible in practice. Several theoretical advantages and limitations of this method were identified and discussed.Alexander AronsValuation30000Completed20132016
20180610An investigation of constructing EQ-5D-5L value sets by censoring time-trade off data at 0One study submitted to the EuroQol Plenary meeting for discussion.2.One manuscript submitted to an International peer-reviewed journal for publication.Zhihao YangValuation14200Completed2018
2013310Test of reference dependency in EQ-5D-5L health state valuationsBackground: Health state valuations provided by patients (i.e. experienced-based preferences) and non­ patients are generally not the same. Ultimately, the decision of which health-state values to use is a normative one, and research into the similarities and differences of values obtained from both perspectives is generated as basis for such decision. Thus far, the discussion has largely overlooked the fact that general population samples are very heterogeneous and comprise many respondents in impaired health states. Similar as patients, these respondents have a unique vantage (or reference) point that can influence their preferences. This potential source of bias is usually not recognized and even when it is acknowledged, the empirical implications are not investigated. This study seeks to address this issue by investigating the existence and empirical relevance of health state reference dependency in health state valuations. Methods: The reference dependency hypothesis is tested in a discrete choice experiment (DCE). A Bayesian DCE design consisting of 8 sets of 24 (matched pairwise) choice tasks was used, each set providing full identification of the EQ5D parameters for individual respondents. Mixed logit models were used to estimate the utility decrements associated with the severity levels in each dimension of the EQ5D. Reference dependency was subsequently modeled relative to respondents' own (self-assessed) EQ-5O health states using 5 additional parameters. These reference dependency parameters capture, separately for each EQ5D dimension, the multiplicative increase or decrease in health state utility as a result of valuing health state levels that are better than or equal to respondents' own health. Results: The DCE was administered to a Dutch nationally representative sample of 799 respondents. The majority of respondents in our sample was in good health, reflected by the share reporting problems in any of the dimensions: 24% in MO, 4% in SC, 27% in VA, 48% in PD and 22% in AD. Modeling the respondents' choices without taking reference dependency into account produced statistically significant and logically consistent parameter estimates. When reference dependency was taken into account, statistically significant and consistent parameters estimates were obtained for all severity levels and reference dependency parameters, except for self-care (where 96% of the sample reported not to have any health problems). Our results indicate that respondents in impaired health lower than or equal to the health state levels under evaluation have, on average, approximately 30% smaller health-state decrements. Despite these differences, a correction for reference dependency did not result in a social tariff with stronger health-state decrements; in fact, the social tariffs obtained from the models with and without reference dependency are close to identical. Conclusion: The results confirm our hypothesis that reference dependency can be observed in a random sample from the general population. On the one hand, reference dependency is found to have a major impact on respondents' individual preferences. On the other hand, the limited number of respondents with health impairments does not appear to bias the social tariff as obtained from a random sample of the general population, which is reassuring evidence.Matthijs VersteeghValuation28200Completed2013
20170390Request for funding to cover expenses relating to a proposed visit of Henry Bailey to spend 1 month at the University of Leeds with Paul KindThisis a preliminary report on ourstudy that investigates the links between EQ-5D and happiness. Firstwe review what has been published about EQ-5D and happiness. We then go on to a preliminary investigation of the relationships between happiness ratings and EQ-5D VAS and utility values in 5 data sets being: RLMS (Russia), GEM (4 Caribbean Basin countries), CliK (West Yorkshire), HSE and a Gallup International dataset covering Latin America. We then compare the findings of our analyses with those of previous studies that have investigated happiness and EQ-5D. Graham et al (2011) conducted the first analysis thatattempt to identify the impact of the 5 dimensions of EQ-5D on happiness. In a Gallup poll conducted in 2007, 14,000 respondents from 18 Latin American countries completed a survey that included the EQ-5D-3L instrument, a 0-10 VAS and Cantril’s ‘Ladder of Life Scale’, a 0 to 10 scale anchored at ‘best possible life’ (10) and ‘worst possible life’(0). This was the first data set that allowed EQ-5D dimensions and VAS scores to be mapped on to life satisfaction or happiness data. EQ-5D score was found to be strongly and significantly correlated with subjective health status (VAS) but much less so with life satisfaction (Cantril’s). Material goods (assets, telephones etc) were found to be more important to life satisfaction than to health satisfactionbyvirtue of having higher regression coefficients in the former. Levels 2 and 3 in EQ-5D dimensions were found to have less impact as explanatory variables in regression models with life satisfaction as the dependent variable than those with subjective health satisfaction as the dependent variable.Dolan and Metcalfe (2012) analysed the impact of the EQ-5D domains on Subjective Well Being (SWB) using regression models and life satisfaction questions:Overall, how satisfied are you with your life? (Using a 0-6 scale)Overall how did you feel yesterday? oRespondents had to express the strength of feeling on a 0-6 scale friendly, lethargic, stressed, happy, sad, calm, angry, tired, depressed and worried. oThe difference between average of positive and negative affect was used to create a new variable: ‘Day affect’Dolan and Metcalfe regressed Life Satisfaction (LS) and Day Affect (DA) separately on the EQ-5D domains using a utility model taken from Shaw et al (2006, i.e. including the D1, I2, and I3 interaction terms). This allowed a comparison of the impact of the EQ-5D dimensions on the LS and DA models with health preference (i.e. by comparing coefficients in the SWB models with the Shaw et al (2006) model). AD2 had a similar impact on both utility and SWB models. The remaining coefficients in the SWB models were much smaller than in the utility models, several of them were counter-intuitive (e.g. taking positive values), and none of the MO, SC or UA2 coefficients were significant at the 5% level in the two SWBmodels. Mukuria and Brazier (2013) used a SWB measure taken from the SF-6D instrument on EQ-5D states in a sample of 15,824 patients from a hospital in Wales. The SWB measure used in this study was the SF-36 question: Have you been happy?Respondents had five response options: All of the time; most of the time; some of the time; a little of the time; and none of the time. Ordered logit analysis produced a model with logically consistent (level 3 greater impact than level 2) and significant odds ratios at the 5% level (except MO3, SC3, UA3 and PD30). Anxiety/Depression was found to have the biggest impact.Karimi and Brazier (2016) investigated the meanings of and interrelationships between health, health-related quality of life (HRQoL) and Quality of Life (QoL). They pointed out that health is one (small) element of QoL, and that QoL is influenced by many ‘non-health’ factors. Hence, we should expect that EQ-5D dimensions would have different effects on HRQoL and QoL. Karimi and Braizer (2016) concluded that the HRQoL concept overlaps both health and QoL factors.Even within the domain of ‘health’ Shah et al (2016) reported that over 40% of their sample of 436 respondents identified factors that they considered to be important to health that were not included in EQ-5D. Sun et al (2016) analysed data from a survey undertaken in China (n=8,000) which included EQ-5D, EQ-VAS and a happiness question taken from the World Values Survey. Mean SWB was observed to increase with self reported health and with VAS values, and to decrease with increasing levels of problems on EQ-5D dimensions. OLS and ordered logit models showed logically consistent and significant (to the 5% level) coefficients for MO and AD with the latter having the highest absolute values.Henry BaileyEducation and Outreach6300Completed20172017
2016390Analyzing self-perceive health status using the EQ-5D in Latin America: 1st approach to the Gallup 2007 World Survey (19 countries)The Gallup World Poll (GWP) has been conducted globally since 2005, and it represents an important tool for local decision-making. The 2017 GWP version reaches out to 160 countries covering 99% of adult population in the world. It takes representative samples for each country in order to extract information about main global issues such as Law and Order, Food and Shelter, Job Creation, Migration, Financial Life, Personal Health, Civic Engagement and Evaluative Well-Being. Also, some specific topics are included in each version according to regional scenario and needs. Summary measures of population health are known to be of great importance due to the potential use in monitoring the health of communities, informing policymakers and being an input for cost-effective analysis as well as other health outcomes research. The 2007-2008 GWP version considered perceptions of Health-related quality of life (HRQoL) in 22 Latin-American countries representative of people aged 15 and over. This report presents the first regional overview of the health perceptions of Latin-American people by analyzing the data from the GWP.victor zaratePopulations and Health Systems15000Completed20162017
20191190HRQoL and Health Literacy among informal caregivers to persons with dementiaAs part of a larger project financed by the Norwegian Research Council, PATHWAYS, on health and ageing, PhD candidate Kristin Häikiö has conducted a survey of ~200 informal carers for elderly people with dementia. Aside from demographics and other background variables regarding both careperson and the one receiving care, the survey was trained upon health literacy, time spent on various caring tasks, perceived carer burden(Relative Stress Scale), and quality of life(EQ-5D-5L). Preliminary analyses (regression models predicting utility scores using health literacy, including relevant background variables) indicate that health literacy is a statistically significant predictor of HRQoL as measured by the EQ-5D-5L among carepersons. This material has gone unused as of yet, the thesis focused elsewhere. As the PhD project funding is out in early2020, and Häikiöis set to deliver her thesis, there is risk that it will not be published upon. The proposal is for funding of three weeks of extra time for the PhD student to write up a paper on the findings regarding health literacy and utility scores, and two days of supervision on the interpretation and use of EQ-5D data.Kim RandOthers6800Completed2020
2013210HTA and quality of life measurement in patientsValentina Prevolnik RupelOthers0Completed2013
2016070EuroQol workshops back-to-back with ISPOR 7th Asia-Pacific Conference in Singapore on September 3, 2016One full-day regional meeting on September 3, 2016 (including a morning session for EQ-5D users in Asia and an afternoon by-invitation session for policy makers and HTA practitioners in Asia)Nan LuoEducation and Outreach70926Completed20162016
2015430The impact of overlap and color coding on response efficiency in discrete choice experimentsObjective To test the hypothesis that level overlap and color coding can mitigate or even preclude the occurrence of attribute non-attendance in discrete choice experiments. Methods A randomized controlled experiment with five experimental study arms was designed to investigate the independent and combined impact of level overlap and color coding on respondents’ attribute non-attendance. The systematic differences between the study arms allowed for a direct comparison of observed drop-out rates and estimates of the average number of attributes attended to by respondents, which were obtained using augmented mixed logit models that explicitly incorporated attribute non-attendance. Results In the base-case study arm without level overlap or color coding, the observed drop-out rate was 14% and respondents attended on average only 2 out of 5 attributes. The independent introduction of level overlap and color coding both reduced the drop-out rate to 10% and increased attribute attendance to 3attributes. The combination of level overlap and color coding, however, was most effective: it reducedthe drop-out rate to 8% and improved attribute attendance to 4 out of 5 attributes. The latter essentially removed the need to explicitly accommodate for attribute non-attendance when analyzing the choice data. Conclusion Based on the presented results, the use of level overlap and color coding are recommendable strategiesto reduce the drop-out rate and improve attribute attendance in discrete choice experiments.Elly StolkValuation59725Completed2015
20180560Non-Linear Time Preferences in Discrete Choice ExperimentsIntroduction One issue currently preventing the wider adoption of discrete choice experiments (DCEs) involves the differences in scale found between these studies and those using a time trade-off (TTO) valuation technique. Specifically, values for severe health problems are often much lower using a DCE approach. However, research has also shown that DCE models specifying non-linear time preferences produce utility estimates on a similar scale to the TTO. While multiple discount factors have been proposed, no study has directly compared different discount methods using EQ-5D data. Methods We compare three different discount factors (exponential, hyperbolic, power) to constant proportionality using two recently completed DCE studies from the US (N=2,260) and Australian (N=944) general population. Each of the four specifications is judged by face validity (utility values, number of disordered levels), model fit (AIC, BIC), and predictive validity (Χ2 and Lin’s 𝜌, using an 80/20 split). All tests are performed using both the conditional logit and Zermelo-Bradley-Terry (ZBT) models to examine the consistency across valuation methods. Results Across both studies and regression models, we find that specifications incorporating non-linear time preferences have higher utility values for severe health problems than those using constant proportionality. Models specifying non-linear time preferences also had better fit (lower AIC and BIC) and greater predictive validity (higher Χ2 and Lin’s 𝜌). The hyperbolic and power discount functions produce nearly identical results (<0.001 difference in utility values for both models in the Australian study). Discussion We find additional evidence that DCE studies should account for non-linear time preferences in the valuation model. However, we do not find evidence of a dominant discount function. The exponential discount function produced fewer disordered levels than the hyperbolic or power functions, but it had poorer model fit and predictive validity across three of the four study/model combinations.John HartmanValuation14800Completed20192019
2014100Two EQ workshops at ISPOR AsiaNan LuoEducation and Outreach8720Completed2014
20170250Cross-sectional validity and feasibility of the self-report EQ-5D-Y as a generic Health Related Quality of Life outcome measure in children and adolescents with Juvenile Rheumatoid Arthritis (JRA) in Western Cape, South AfricaJuvenile IdiopathicArthritis (JIA) is an autoimmune, chronic rheumatic disorder with episodic flare ups and remissions. Itoccurs in children 16 years and younger, causing short orlong-term disability (5). Impairments resulting from the inflammatory processes and leading to disability include on-going pain, limited range of motion, joint stiffness and decreased physical fitness. Other manifestations may include behavioural problems, social isolation, depression and anger(6). Skeletal growth may be delayed or distorted in severe cases of JIA, as growth hormone (GH) levels are low and premature fusion of the epiphyses at the hipmay occur (7). There is little consensus regarding the precise prevalence of JIAworldwide due to a lack of standard diagnostic criteria(8). The prevalence rate in developed countries varies in different population studies from seven to 400 per 100 000 children(9), (10), (11), (12). There are no accurate prevalence studies for South Africa (8), it would howeverappearthat there is a higherprevalence in North America and Europe compared to Asia and Africa(11).Limited researchon African or South African populations with JIAis thought to be due to limited resources being directed at the high burden of infectious diseases such as Human Immunodeficiency Virus (HIV)and Tuberculosis (TB)in these regions, rather than to JIA(8). The cause of the disease is thought to beasthe result of an interaction of genes.In 2004, a study was performed in South Africa which investigated the association of the HLA-DRB1 allele with rheumatoid arthritis. The study showed a strong relationship between erosive rheumatoid arthritis and the presence of such alleles, withconsistencyacross race groups –similar to a studyin Europe, North America and Asia(13). The pathophysiology of JIAincludes the presence of autoantibodies or autoreactive T cells (14), which attack the body, causing the joint synoviumto swell and produce excess fluid, resulting in the breakdown of cartilage and friction at the joint site with movement-leading to stiffness, pain and swelling. No cure exists, however first line treatment includes NSAID’s, as well as corticosteroid injections, physical therapy and disease modifying anti-rheumatic drugs (DMARD’s). DMARD’s are a group of disease modifying drugs which aim to slow down the progression of the disease(15).The psychometric properties of the diseasespecificPedsQLRheumatology Modulehavebeen compared with other disease specific outcome measures, such asthe Juvenile Arthritis Quality of LifeQuestionnaire (JAQQ), the Paediatric Rheumatology Quality of Life Scale (PRQL), and the Childhood Arthritis Health Profile (CAHP), but not with a generic outcome measure(16). A study conducted bySawyer et al. 2003in Southern Australia investigated the relationship between HRQoL, physical functioningand pain in children with JIA, using the PedsQL Rheumatology Module Arthritis module. The study indicated that the children experienced themost problems with physical functioningand theleast problems in social functioning. The children with higher levels of pain, reported lower levels of functioning in the physicaland emotional dimensions(17).The Juvenile Arthritis Multidimensional Assessment Report(JAMAR) has beenused in clinical practice to improve the care of children with JIA(18)(19)(1). The performance of the JAMAR was favourably compared to the PedsQL Rheumatology Module (1). The EQ-5D-Y has been used to assess HRQoL in children with a variety of health conditions, including JIA(20)(21)(22)(23)but its performance has not been compared to disease specific outcomes. There are at least three outpatient clinics in Cape Town which are dedicated to the management of children with JIA, one of which is at alarge Paediatric Hospital and the other two are based in General Hospitals. In total, there were 219childrenfrom 8-15 years,registeredatthese clinics in 2017, however not all these children attend the outpatient clinics regularly. Many of these children default clinics during a period of remission and only reappear if they relapse. Children typically attend these outpatient facilities with their parentsmonthly if they have been started on DMARDs and every 4 months thereafter if they are on DMARDs and biannually or annually if they are stable off DMARDs. The JAMARis the only outcome measure that is currently used to monitor functional and HRQoL change in the children. However, it is a lengthy instrument and not routinely applied. As the joint damage and symptoms affect HRQoL and can persistinto adulthood in 50% of children (24), it is important to trackHRQoL. The generic EQ-5D-Ymight be appropriate for this use as it can used until the age of 18 years and, while not interchangeable with the adult EQ-5D, it at least articulates with this version which could be used in adulthood.It was hypothesised that the EQ-5D-Y, whereas not yielding the details of the child’s condition to the extent of the JAMAR, would show concurrent validity with the JAMAR. It could thus be used not only to track changes in the child’s HRQoL overtime but also as a screening for increased inflammatory activity. The JAMAR might then be applied in the cases who reported disease flare or a decrease in their HRQoL.Jennifer JelsmaYouth4850Completed20172017
2016320New methods for analysing the distribution of EQ-5D observations in data setsBackground: EQ-5D profile data are often under-analysed, but can yield important insights into levels of and changes in patient and population health. One characteristic is the extent to which they cluster together in a small number of profiles or are dispersed evenly over many profiles. This can have implications for interpreting statistical analysis of the corresponding EQ values (Index)data, and for clinical management of patients.Aims:This paper aims to develop methods for describing observed distributionsof EQ-5D profiles and to explore the properties of the new methods compared with existing ones (e.g. Shannon’s Index). We investigate the methods using both real, from the EnglishNHS, and simulated EQ-5D data, and show how they can be used to generate new insights into, for example, the differences between the three-and five-levelversions of the EQ-5D with respect tohow profile data are clustered. Methods: We report three methods we have developed to characterise and summarise the distribution of health states in patient reported outcome (PRO) data within a sample or population of patients: the Health State Density Index(HSDI), Health State Density Curve(HSDC)and estimated Power Law functions (PLFs).We compare these andexisting methods from information theory (e.g. Shannon’s Index), in examining the distribution of EQ-5D health profilesin three data sets:across three groups of patients in Cambridgeshire Community Services NHS’s electronic patient records for the EQ-5D-5L; the Health Survey for England 2014 for the EQ-5D-3L; andthe NHS PROMs programme for theEQ-5D-3L. The properties of the various methods are further examined using simulated data sets. Results:Each method has different properties and will give different insights into patients’ data. For example, the Shannon index (absolute and relative) is not sensitive to random variations but decreases slowly with “rare health states”. The HSDI decreases slowly with random variations and is strongly affected by “rare” health states with largedecreases towards zero (total inequality).Conclusions:These methods can be used by researchers to better understand the characteristics of EQ-5D profile data. They can also be used by clinicians to understand the degree to which their patients’ needs are homogeneous or characterised by distinct sub-groups, with implications for treatment planning. Finally, the methods can also be used as a way of comparing differences between instruments, such as the 3L and 5L, in measuring health. Acknowledgements and disclaimersThis project was funded by a research grant from the EuroQol Research Foundation. All views expressed are those of the authors, and not necessarily those of the EuroQol Research Foundation.The authors are grateful to Nils Gutacker, Thomas Kohlmann,and participants at the Health Economists’Study Group (HESG) Winter Meeting (Birmingham 2017)and the 34thEuroQol Plenary (Barcelona 2017), and to Nigel Rice, for helpful comments onearlier versionsof this paper.Nancy DevlinDescriptive Systems, Populations and Health Systems34500Completed20162017
20190500Funding a two-day Academy Workshop in Cape Town in January or February 2020.Aims: Tobuild capacity in the field of measurement of HRQoL, using the EQ-5Dand to stimulate interest in the continued and expanded use of the EQ-5D in Africa.Outcomes: Greater utilisation of the EQ-5D as a patient-reported outcome (PRO) measure across the different countries and contexts; improved methodology, analysis and reporting of studies using the EQ-5D; establishment of researcher networks within and between countries; possible interest in doing national or language valuation studies; greater attention to methodological research about EQ-5D; and an increased interest in HTA and CUA by health economists and policy makers. Participants: These will include threeinternational experts to act as facilitatorsand 46participants. Researchers from Africa whohave published several papers using the EQ-5D as an outcome measure, particularly methodological papers and who have used the measure for HTAwill be invited.Participants will be identified by trawling through the literature.Researcherswho are already affiliated and/or supported by the Foundationand other stakeholdersidentified by snowballing will also be invited.Travel bursaries will be awarded to 20colleagues from the African region and two from South Africa.Programme: Each session will be devoted to a specific theme, based on theparticipants’ experience, needs and expectations identified through a pre-meeting survey. These themes are likely to include e.g. validation of the different versions, use of the EQ-5D in clinical and population-basedstudies and valuation studies. Introductory sessions will be offered on the use forthe EQ-5D in HTA, validation and valuation methodology if required.(250words)Jennifer JelsmaEducation and Outreach74030Completed20202020
20170460Comparison between EQ-5D-3L-Y and EQ-5D-5L-Y in a patient population in Hong KongThis study was design to compare five-level (EQ-5D-5L-Y) and three-level EQ-5D-Y questionnaires in a pediatric patient population in juvenile (JIS) or adolescent idiopathic scoliosis (AIS) in Hong Kong. A consecutive sample of idiopathic scoliosis patients aged 8–17 years were recruited from a referral outpatient scoliosis center at Hong Kong, China in October 2017 and completed the two versions of EQ-5D-Y. After obtaining informed consent, eligible patients were asked to complete the 5LY and 3LY questionnaires, EQ-VAS and answer the questions about their socio-demographic and clinical characteristics. All respondents completed 5LY first followed by 3LY. Test-retest reliability and responsiveness were assessed by asking patients to complete the questionnaires at 1-2 weeks and 3-9 months again through telephone interview after baseline survey. The feasibility of 3LY and 5LY for use in Hong Kong children and adolescents has been assessed, and their redistribution properties, informativity and psychometric properties have been compared.Nan LuoDescriptive Systems, Youth1850Completed20172018
20180080Extending the QALY project in the United States – testing the face and content validity of a preliminary list of itemsSemi-structured cognitive interviews were conducted with respondents with arange of chronic conditions in Chicago ,Illinois .The mesandissues related toapool of candidate items identify by the E-QALYresearchgroup were discussed with respondents until saturation was reached(n=19). Overall, while respondents were able to understand the items and respons escales ,there was a general preference for positively stated items and for a response scale based on frequency of experience. Respondent preferences for different items was generally guided by the consideration of the context in which the instrument is intended to be used. The assessment of self-reported burden wa sconsidered to be relevant in the American context.Simon PickardDescriptive Systems, EQ-HWB15000Completed20182018
2016100Valuation of EQ-5D-5L for the Malaysian populationObjectives‐ Primarily the aim is to develop an EQ‐5D‐5L value set reflecting the health preferences of the Malaysian adult population and secondarily to psychometrically validate the Malay and English versions of the descriptive system in Malaysians. Methods‐ 1137 respondents of the Malaysian general public were sampled with quotas for urbanicity, gender, age, and ethnicity in a nationwide study in 8 locations. The study was conducted using a standardized protocol involving the EuroQol Valuation Technology (EQ‐VT) computer‐assisted interview system. Respondents were administered 10 composite time‐trade off (C‐TTO) tasks and 7 discrete choice experiment (DCE) tasks. Both linear main‐effects and constrained non‐linear regression models of C‐TTO only data and hybrid models combining C‐TTO and DCE data were explored to determine an efficient and informative model for value set prediction. Additionally, respondents completed the EQ‐5D‐3L questionnaire on paper and selected respondents were given paper form of EQ‐5D‐5L for completion within two weeks for test‐retest reliability. Ceiling effects, response redistribution, informativity, and convergent validity were compared between EQ‐5D‐5L and ED‐5D‐3L separately by Malay and English language versions. Results‐ The hybrid 8‐parameter model was chosen to represent the Malaysian value set, as it displayed greater out‐of‐sample predictive accuracy other models with the estimated value set ranging from ‐0.442 to 1. In general, ceiling effects reduced in the EQ‐5D‐5l version with higher relative informativity and convergent validity exhibited for both Malay and English versions. Test‐retest reliability testing revealed moderate to poor agreements on all EQ‐5D‐5L dimensions. Conclusions‐ The presence of a validated Malaysian EQ‐5D‐5L value set will facilitate its application in research and health technology assessment activities.Asrul ShafieValuation60193Completed2015
2015110Estimation of the EQ-5D-5L value set for PortugalAims:Any decision-making, mainly in a context of limited resources, has to have arationaleand be transparent. The EQ-5D is a generic preference-based quality of life measure considereduseful to support clinical and policy decisions, by providing utility values that can be easily converted into QALYsto be integrated in cost-utility economic evaluations.However, its first 3L version showssignificant ceiling effect. Therefore, a 5L version was designed with five levels ofproblems intensity for each dimension, allowing utility values for all 3,125 health statesmeasured.The purpose of this study was to estimate the value system for Portuguese population in what concernsthe different health states described by the EQ-5D-5L description system.Methods: Health states values were elicited according to the EuroQoL Valuation Technology (EQ-VT) protocol using composite time trade-off (cTTO) and a discrete choice experiment (DCE). Data were collected between October’2015 and July’2016 using computer-assisted personal interviews in arepresentative sample of the Portuguese population aged above 18 years andstratified by age and gender.cTTO data weremodelledby fixed and random effects models, such as OLS, variable effects GLS, censored GLS, censored Tobit, interval GLS censored at 1, and interval GLS censored at 1 and 0. Model performance was assessed according to specific criteria: internal consistency, statistical significance of the estimated coefficients, and goodness of fit evaluation.The ability for the model to tackle censored data was also evaluated. DCE data were not modelled.Results: A sample of 1,451 respondents revealed aVAS mean score of78.9 in a 0-100 scale from the worst to the besthealth. Themodels showed no inconsistenciesand the models with best performance was the interval GLS censored at 1.Conclusions: This research provided the Portuguese value set for the EQ-5D-5L based on cTTO and these findings are to be usedin economic evaluationsto be conducted in Portugal.Pedro FerreiraValuation47000Completed2015
2012020German tariff for EQ-5D-5LThe initial preparation of the study was undertaken between November 2012 and September 2013. Due to the results of the first wave of EQ-5D-5L valuation studies in 2012/2013and the following moratorium, it was decided to postpone the data collection of the main study in Germany and to conduct an initial experimental pre-studyto test the protocol updates in Germany.The final report was already signed-off and the results were published (Ludwig et al. Valuation of the EQ-5D-5L with composite time trade-off for the German population -an exploratory study. Health Qual Life Outcomes 2017;15(1):39).The results of the pre-study contributed to the development of EQ-VT version 2.0 (inclusion of feedback module). After consulting the Valuation Methodology Working Group in September 2014, it was decided to apply the improved valuation protocol2.0 in the German EQ-5D-5L valuation study for the first time worldwide. Interviewer training and pilot interviews were carried out in November2014. Data were collected from December 2014 to March 2015by 19 professionalinterviewers in six cities and surroundslocated in different parts of Germany. A detailed analysison the complete data set was carried out from May2015to March 2017 and (preliminary)results were stepwise presented on several conferences. A manuscript for journal submission was prepared and submitted in August 2017.Wolfgang GreinerValuation62505Completed2012
2012010The EQ-5D-5L Valuation Study in ChinaBackground: In order to estimate a country-specific EQ-5D-5L value set, time trade-off (TTO) values of a subset of EQ-5D-5L health states were recently elicited from a general population sample (N=1250) in China using the EuroQol Group’s EQ-VT system. The purpose of this study was to evaluate the ability of various linear regression models to predict disutility (or utility) values of EQ-5D-5L health states.Methods: Characteristics of EQ-5D-5L health states were modelled to predict disutility values. Model specifications tested included main effects of different levels and types of EQ-5D-5L health problems, main effects with selected first-order interaction terms, and extended N3 and D1 models. All models were also tested with addition of one of two alternative terms, M1 (1, if slight problems in only 1 dimension and no problems in the remaining dimensions; 0, otherwise) and M2 (1, if slight problems in 1 or 2 dimensions and no problems in the remaining dimensions; 0, otherwise). Each model was estimated with 70% of the study sample and assessed for goodness of fit and prediction bias with the remaining 30% of the sample. Model prediction bias was visually examined by plotting the predicted and actual values for observed health states. Results: For all model specifications tested, the random-effects model was more efficient than the fixed-effects model (p>0.05 for all, Hausman test). The M1 and M2 terms were statistically significant in almost all models. However, the estimated effect of ‘slightly anxious/depressed’ turned negative in most models when the M1 term was added. All criteria considered, the model contained all main effects, the M2 term, and a fixed constant (defined as ‘M2 model’) or a varying constant as specified by the D1 term (defined as ‘D1-M2 model’) was most suitable for estimating the EQ-5D-5L TTO values in China. The two variants of the M2 model exhibited identical fit of the 30% of the sample which was not used for model estimation, with the mean absolute error being 0.0573 and the number of absolute errors >0.05 and >0.10 being 41 and 15, respectively. Conclusions: This study provides the algorithm for estimation of the EQ-5D-5L value set in China. The significance of the M terms suggests that Chinese rate the mildest EQ-5D-5L health states differently in the TTO exercise. Future studies should investigate whether this is the case in other populations and explore the reasons for this valuation behaviour.Gordon LiuValuation61875Completed2012
2013180Transforming latent utilities to health utilities: Can one function fit other countries?Discrete choice experiments(DCEs) area promising alternative to more resource intensive preference elicitation methodssuch as the time trade-off(TTO),as thesimplicity ofpairwise comparisons lends itself to online completion, which can save time and money. However, modelingDCE dataproduceslatent utilities which are on an unknown scale. Therefore, latent utilities need to be transformed before they can be usedin quality-adjusted life year calculations. Weaimed to explore transformationfunctionsfrom DCE-derived latent utilitiestoTTO-derived health utilities. We usedEQ-5D-5L valuation data from 8different countries that collectedboth DCE and TTO datausing astandardized protocol. Results foundless variation in the function that transformedlatent utilities to health utilitiesin the western countries than in the eastern countries. While a global transformation function is not recommended,results suggestregional transformationfunctionscould be potentially used to derivehealth utilities fromDCEdata.Feng XieValuation9950Completed2013
2013010States worse than Dead: Exposing the measurement properties of the Better than Dead preference methodBackground: Traditionally, the valuation of health states worse than dead suffers from 2 problems: [1] the use of different elicitation methods for positive and negative values necessitating arbitrary transformations to map negative to positive values and [2] the inability to quantify that value is time dependent. The Better than Dead (BTD)-method is a health state valuation method where states with a certain duration are compared to dead. It has the potential to overcome these problems. Objectives: To test the feasibility of the BTD-method to estimate values for the EQ-5D system. Methods: A representative sample of 291 Dutch respondents (age 18-45 years) was collected. In a web-based questionnaire, preferences were elicited for a selection of 50 different health states with 6 durations between 1 and 40 years. Random effect models were used to estimate effects of socio-demographic and experimental variables, and estimate values for the EQ-5D. Test-retest reliability was assessed in 41 respondents. Results: Important determinants for better than dead were a religious life stance (OR 4.09 [2.00 - 8.36]) and educational level. The fastest respondents more often preferred scenarios to dead and had lower test-retest reliability, 0.45 vs. 0.77 and 0.84, for fast, medium and slow response times. Value estimates showed a small number of so-called maximal endurable time states. Conclusion: Valuating health states using the BTD-method is feasible and reliable. Further research should explore how the experimental setting modifies how values depend on time.Peep StalmeierValuation15100Completed20132013
2016430Issue panel and Workshop at ISPOR Singapore 2016A 1-hour workshop plus a 1-hour Issue panel, both conducted by Nan Luo, Kim Rand-Hendriksen, Mark Oppe and Juan Manuel Ramos-Goñito ISPOR 7thAsia-Pacific Conference, Sep3-6, 2016, Suntec Singapore Convention & Exhibition Center. SingaporeJuan M. Ramos-GoñiEducation and Outreach13490Completed20162016
20190800Measuring and Valuing Patient-Reported Outcomes in Economic Evaluations of Health Care WorkshopThe format of the presentation is outlined below, and the aims of the workshop. There are several strategic advantages to the EuroQol Group support of this workshop. This is the second annual conference for the China National HTA in Beijing. Approximately 800 attendees went last year, it is expected more will attend this year. There is interest in understanding approaches to valuing health in health technology assessment. Drs. Xie and Pickard are established academicians who have extensive experience presenting on this topic, and have previously co-presented workshops at ISOQOL, HTAi and ISPOR. Dr. Xie is an invited plenary speaker at the conference, and has established ties with the Chinese Ministry of Health that is involved in organizing the conference. This workshop is an opportunity to further build upon developing a relationship with the ministry and HTA organizations in China. The EuroQol Group has strategic interest in promoting understanding of approaches to valuing health. Although we are not explicitly “promoting” the use of the EQ-5D, a more nuanced approach is intentionally used here, where the EQ-5D will be used for exercises and featured as the foremost utility measure used globally. China has great strategic value for the group, as it is rapidly developing its own pharmaceutical industry, and this educational outreach initiative broaden efforts by the EuroQol group for goodwill relationships in Asian countries, particularly China.Simon PickardEducation and Outreach7400Completed2019
2013050Assessing test-retest reliability of the EuroQol Group Valuation Technology in valuing the EQ-5D-5Lnot available yetFeng XieValuation69332Completed20132013
2015440Valuing health states 'in context'Introduction: In EQ-5D valuation studies, health states have by convention beenpresented as a series of five sentences, each describing a health dimension and severity‘level’. Since the introduction of the EQ-5D-5L, the valuationof the descriptive system may have become more challenging. Preference inversions, for example between levels 4 and 5,have been identified in various EQ-5D-5L valuation studies. A solution could be to mimic the way in which patients self-report their own health using the EQ-5D-5L descriptive system, where the ordinal structure of levels 1 to 5 is clear. The aim of this study is to develop and test the feasibility of presenting health states in the ‘context’ of the whole descriptive system, and to understand how this affects valuation data.Methods: A two-arm discrete choice experiment (DCE) study was conducted online (n=993). Respondents were randomly allocated to either the control (standard EQ-VT display), or the ‘context’ arm (health states described in the context of the descriptive system).Each respondent wasrandomly allocated to two of 28 blocks (from the EQ-VT experimental design),and completed14 DCE tasks plus two fixed pairs, chosen to address specific research questions. The survey also includedbackground and feedback questions. Differences across arms wereassessed using regression analyses, and the numbersof correctly ordered coefficients and logically inconsistent responseswere examined. Differences across arms weretested on the pooled dataset.Results and discussion:Presenting health states ‘in context’ is feasible, and we demonstrate that this can significantly reduce the number of respondents selecting logically dominated health states, particularly for the labels ‘severe’ and ‘extreme’(χ²=46.02, p<0.001). Comparing conditional logit modelling results between arms, we find that coefficients are ordered as expected for both arms (apart from the ordering of MO2 and MO3 for the control arm, but this is not statistically significant), but the magnitude of decrements between levels islarger for the context arm, particularly for levels 4 and 5. The likelihood ratio statistic of the restricted pooled model is 42.00, which leads us to reject the null hypothesis of preference homogeneity between the two arms, suggesting that preferences were significantly different between arms. Our results could have implications for valuing EQ-5D states using DCE andother methods such as time trade-off (TTO). It may also haveimplications for the requirement to label each level.Koonal ShahValuation33500Completed2015
20170480Valuing Health-State: An EQ-5D-5L Value Set for EthiopiansObjectives:Thereis a growing interest inHealth Technology Assessment and economic evaluationsin developing countries like Ethiopiausing measures such as EQ-5D-5L to undertake cost utility analysis. However, the main challengeislack of local population value setswhich are recommended in most countries as they reflect local preferences.Hence, this study aimed to obtain social preference and establish an EQ-5D-5L value set from Ethiopian general population.Methods:Anationally representative sample (n= 1,050) wasrecruited using stratified multi-stage quota sampling technique. Face-to-face, computer-assisted interviews usingthe EuroQol Portable Valuation Technology (EQ-PVT) software of composite time trade-off(c-TTO) and discrete choice experiments (DCE)were undertakento elicit preference score. EQ-PVT protocol feasibility was pilottested in a sample of the population (n=110).A hybrid regression model combining c-TTO and DCE data was used to estimate the final value set.Results: In the pilot study,acceptability of the tasks was good and there were no special concerns with undertaking TTOtask. The predicted value for the EQ-5D-5L ranged from-0.719to 1. The coefficients generated from a hybrid model were logically consistent.Anxiety/depression showed the most impact to utility decrementof 0.4578, and self-care influenced the least 0.2224. The maximum predicted value beyond full health was 0.9741 for the ‘11112’ health state. Conclusions:This study established Ethiopian value set for EQ-5D-5L on the basis of c-TTO and DCE. It is expected to facilitate health economic evaluations, health-related quality of life research and to inform decision making in Ethiopia.Keywords:Discrete choice experiment, EQ-5D-5L, Ethiopia, Health state valuation, Quality of life,Time trade-off, Utility.Abraham GebregziabiherValuation28833Completed2017
2014050Hong Kong participation BTD-WTD split experimentCarlos WongValuation9443Completed2014
2014170Directly eliciting personal utility functions: a feasibility study of an innovative approach to valuing HRQoLIntroduction: In 2015, we introduced the concept of Personal Utility Functions (PUFs) at the Scientific Plenary meeting (Devlin et al 2015), and shared some promising early results from pre-piloting.In contrast to existing valuation methods –all of whichrely on presenting health states and inferring from respondents’responses what their preferences are with respect to the descriptive system, the PUF approach asks respondentsto consider the dimensions and levels of the EQ-5D, and takes them through a series of tasks to help them attach weights to reflect the importance of these relative to each other.We also ask respondentsto identify their anchor for dead = 0.An important emphasis of the approach is on helping respondentsreflect and deliberate on their answers as they construct their PUF. Subsequent to the Plenary, we further improved our methods for eliciting PUFs, and conducted pilot interviews with n=76 members of the general public in England and a convenience sample of n=30health outcomes professionals working for Pharmerit (various locations). The aimsof this paper areto (a) report the results from the pilot and (b) demonstrate the way in which the PUFs of individuals can be modelled to obtain a value set. M