A Systematic Review of the Methodologies and Modelling Approaches Used to Generate International EQ-5D-5L Value Sets

The impact of demographic change on value set validity and obsolescence

PURPOSE

To investigate the contribution of demographic trends in countries' age and gender composition to value set validity and obsolescence.

METHODS

Time-trade off (TTO) valuation data from 3 EQ-5D-3L value sets of 20 years or older from the United Kingdom, Japan, and the United States were re-analyzed using Bayesian heteroskedastic Tobit models with sex and age group-specific scale parameters. Original value sets were obtained by weighting the original preference structures with the countries' original demographic composition at the time of the data collection. Updated value sets were created using the original preference structure weighted using the countries' most recent demographic composition. The differences between the original and updated value sets were monitored and compared based on 95% credible intervals.

RESULTS

The gender and age composition of the investigated countries changed in all 3 countries over time. The modelled health state preferences also depended on the respondents' gender and age. However, the overall impact of this demographic change on the investigated value sets was negligeable in all 3 countries and this finding was robust to accounting for the impact of ethnicity trends in the United States.

CONCLUSION

Value sets may become redundant and obsolete for various reasons, but demographic change was not identified as a contributing factor.

Exploring the relationship between condition severity and health-related quality of life in people with haemophilia A across Europe: a multivariable analysis of data from the CHESS II study

BACKGROUND

Haemophilia A (HA; Factor VIII deficiency) is a congenital X-linked bleeding disorder characterized by trauma-related or spontaneous bleeding events, most notably arising within the intraarticular space and resulting in chronic inflammation and degeneration of affected joints. Endogenous clotting factor activity relative to normal levels determines the severity of HA symptoms, as mild (> 5-40%), moderate (1-5%), or severe (< 1%). Within the current environment of rapid evolution in HA management, we seek to understand the interplay of condition severity and health-related quality of life (HRQoL) to characterise and differentiate unmet needs among people with HA (PwHA).

METHODS

A generalised linear regression model (GLM) was developed to explore the relationship between HA severity and EQ-5D-5 L index score from adult HA patients sampled in the "Cost of Haemophilia across Europe - a Socioeconomic Survey II" (CHESS II) cross-sectional, retrospective burden of illness study among adults with hereditary haemophilia A or B from eight European countries. HA patients of any severity with no active inhibitors during the 12 months prior to data capture and a completeEQ-5D-5 L response were included. A base GLM model was specified with covariates for demographic and clinical characteristics (age, body mass index, country, employment, HA severity, annual bleeding rate, problem joints, and chronic pain).

RESULTS

Of 381 evaluable patients, 221 (58.0%) had severe HA, 96 (25.2%) had moderate HA, and 64 (16.8%) had mild HA. Among the covariates included in the GLM model and after controlling for haemophilia-related outcomes, a significant association was observed between mild HA and higher EQ-5D-5 L index score (average marginal effects, 0.084; p = 0.016) relative to severe HA. Patient country of residence and magnitude of HA-related chronic pain were also associated with significant differences in index scores, with the latter showing a negative relationship with HRQoL outcomes.

CONCLUSIONS

Condition severity and chronic pain are significant predictors of HRQoL in PwHA. Durable bleeding protection and effective management of chronic pain have the potential to address unmet treatment needs in this population.

Population Norms for the EQ-5D-5L, PROPr and SF-6D in Hungary

OBJECTIVES

This study aimed to develop population norms for three preference-accompanied measures [EQ-5D-5L, Patient-Reported Outcomes Measurement Information System (PROMIS)-preference scoring system (PROPr) and Short-Form Six-Dimension (SF-6D)] in Hungary.

METHODS

In November 2020, an online cross-sectional survey was conducted among a representative sample of the Hungarian adult general population (n = 1631). Respondents completed the Hungarian versions of the EQ-5D-5L, PROMIS-29+2 version 2.1 and 36-item Short Form Survey version 1 (SF-36v1). The association of utilities with sociodemographic and health-related characteristics of respondents was analysed using multivariate regressions.

RESULTS

The proportion of respondents reporting problems ranged from 8 to 44% (self-care to pain/discomfort) on the EQ-5D-5L, 39-94% (physical function to sleep) on PROPr and 38-87% (role limitations to vitality) on the SF-6D. Problems related to physical function, self-care, usual activities/role limitations and pain increased with age, while mental health problems decreased in all three measures. In almost all corresponding domains, respondents indicated the fewest problems on the EQ-5D-5L and the most problems on the SF-6D. The mean EQ-5D-5L, PROPr and SF-6D utilities were 0.900, 0.535 and 0.755, respectively. Female gender (PROPr, SF-6D), a lower level of education (EQ-5D-5L, PROPr), being unemployed or a disability pensioner (EQ-5D-5L), being underweight or obese (SF-6D), lack of physical exercise (all) and polypharmacy (all) were associated with significantly lower utilities. PROPr yielded the lowest and EQ-5D-5L the highest mean utilities in 28 of 30 chronic health conditions.

CONCLUSIONS

This study presents the first set of Hungarian population norms for the EQ-5D-5L, PROPr and SF-6D. Our findings can serve as reference values in clinical trials and observational studies and contribute to the monitoring of population health and the assessment of disease burden in Hungary.

Content validity of the EQ-5D-5L with skin irritation and self-confidence bolt-ons in patients with atopic dermatitis: a qualitative think-aloud study

OBJECTIVES

Two bolt-on dimensions (skin irritation, self-confidence) have been developed for the EQ-5D-5L to improve its content validity and responsiveness in psoriasis. However, the two bolt-ons are not strictly psoriasis-specific and are potentially relevant in other skin conditions. This study aims to explore the content validity of the EQ-5D-5L with two bolt-ons in patients with atopic dermatitis (AD).

METHODS

In 2021-2022, qualitative, semi-structured interviews were conducted with 20 adult AD patients at a university dermatology clinic in Hungary. We aimed for a heterogeneous sample in terms of age, gender, education and disease severity. Patients completed the EQ-5D-5L with two bolt-ons using a think-aloud protocol. Probing questions were posed to investigate item relevance, potential conceptual overlaps, missing concepts and the appropriateness of the recall period. Interview transcripts were subjected to thematic analysis.

RESULTS

The EQ-5D-5L with the two bolt-ons covered the most important aspects of health-related quality of life in AD patients. Most patients found both the skin irritation and self-confidence bolt-ons relevant. Fifteen potential missing concepts were identified, but only two (social relationships, judgement by others) were identified by more than one patient. A smaller conceptual overlap was found between the skin irritation and pain/discomfort dimensions in 7 patients (35%). Half the patients expressed a preference for a recall period of 1 week rather than of 'today'.

CONCLUSIONS

The EQ-5D-5L with skin irritation and self-confidence bolt-ons showed good relevance, comprehensiveness and comprehensibility in patients with AD. However, in terms of comprehensiveness, social relationships and judgement by others (stigma) may be missing from the questionnaire.

Do people with different sociodemographic backgrounds value their health differently? Evaluating the role of positional objectivity

Objective

The fundamental disconnect between the actual and the perceived health of an individual raises considerable skepticism on the self-reported health data as it may be confounded by an individual's socio-economic status. In this light, the present study aims to assess if people with different sociodemographic backgrounds value their health differently.

Methods

The health-state valuation using time-trade off was performed in a cross-sectional survey among a representative sample of 2,311 adults from India. Individuals were selected using a multistage stratified random sampling from five Indian states to elicit their present health-state, and to perform the health-state valuation exercise using computer assisted personal interviewing. A single block of standardized health-states was valued by multiple individuals, each belonging to different socio-demographic group. The difference in the valuation of health was assessed using bivariate analysis. The impact of different sociodemographic factors on the health-state valuation was evaluated using Tobit regression model.

Results

Differences in the valuation of health were observed among different groups of age, religion, family type, state of residence, substance abuse, presence of ailments at the time of valuation, and number of dependent members in the household. Even after controlling for the severity of the administered health states, factors having a significant association with the valuation of health are age, religion, state of residence, substance abuse, family type, number of dependent members in the household, and presence of chronic or both acute and chronic ailments. Younger individuals place a higher value to their health as compared to their older counterparts. As compared to a healthy individual, a person with ailments rates the same health-state as worse.

Conclusion

Inequalities in self-reported ill-health cannot be attributed to positional objectivity; age, religion, state of residence, substance abuse, family type, dependents, and ailments impact individual health valuation.

EQ-5D-5L Value Set for Slovenia

OBJECTIVES

The aim of this valuation study was to produce a value set to support the use of EQ-5D-5L data in decision making in Slovenia.

METHODS

The study design followed the published EuroQol research protocol, and a quota sample was defined according to age, sex, and region. Overall, 1012 adult respondents completed 10 time trade-off and seven discrete choice experiment tasks in face-to-face interviews. The Tobit model was used to analyse the composite time trade-off (cTTO) data in order to generate values for the 3125 EQ-5D-5L health states.

RESULTS

The data showed logical consistency, with more severe states being given lower values. The greatest disutility was shown in the pain/discomfort and anxiety/depression dimensions. In the EQ-5D-5L value set, the values range from -1.09 to 1. With the exception of UA5 (unable to perform usual activities), all other levels on all health dimensions were statistically different from 0 and from each other. Compared with the existing EQ-5D-3L value set, there is a slightly lower share of 'worse than dead' states (32.1% compared with 33.7%) and the minimum value is lower.

CONCLUSIONS

Results have important implications for users of the EQ-5D-5L in Slovenia and regions. It is a robust and up-to-date value set and should be the preferred value set used in adults in Slovenia and in neighbouring countries without their own value set.

UK Valuation of EQ-5D-5L, a Generic Measure of Health-Related Quality of Life: A Study Protocol

OBJECTIVES

A high-quality and widely accepted UK EQ-5D-5L value set is urgently required to enable the latest version of EQ-5D scored using recent UK public preferences to inform policy including health technology assessments submitted to the National Institute for Health and Care Excellence. This article outlines the study protocol for the generation of a new EQ-5D-5L UK value set.

METHODS

Twelve hundred interviews will be undertaken using the composite time trade-off elicitation technique for 102 health states (86 from the international EQ-5D-5L valuation protocol, plus 16 with best predictive performance in an extended design used in the Native American EQ-5D-5L valuation). The sample will be UK adults (age ≥18 years) proportionately representative across England, Wales, Scotland, and Northern Ireland, representative for age, sex, ethnicity, and socioeconomic group, with inclusion of participants with/without health problems. Participants will choose to be interviewed via videoconference (by Zoom) or in-person in a central venue. Data quality will be rigorously assessed.

RESULTS

The value set will be generated using tobit random effects and heteroscedastic tobit models (with censoring at -1) using all data, excluding time trade-off values highlighted by participants as ones they would reconsider and data from interviewers failing protocol compliance. Quality and acceptance will be achieved by public involvement, regular Steering Group meetings, independent assessment of data quality at 4 time points, and final endorsement of data and analyses.

CONCLUSION

This study will produce a UK value set for the EQ-5D-5L for use in prospective and retrospective data sets containing EQ-5D-5L data.

Valuing the EQ Health and Wellbeing Short Using Time Trade-Off and a Discrete Choice Experiment: A Feasibility Study

OBJECTIVES

The EQ Health and Wellbeing Short (EQ-HWB-S) is a new generic measure that covers health and wellbeing developed for use in economic evaluation in health and social care. The aim was to test the feasibility of using composite time trade-off (cTTO) and a discrete choice experiment (DCE) based on an international protocol to derive utilities for the EQ-HWB-S and to generate a pilot value set.

METHODS

A representative UK general population was recruited. Online videoconference interviews were undertaken where cTTO and DCE tasks were administered using EuroQol Portable Valuation Technology. Quality control (QC) was used to assess interviewers' performance. Data were modeled using Tobit, probit, and hybrid models. Feasibility was assessed based on the distribution of data, participants, and reports of understanding from the interviewer, QC and modeling results.

RESULTS

cTTO and DCE data were available for 520 participants. Demographic characteristics were broadly representative of the UK general population. Interviewers met QC requirements. cTTO values ranged between -1 to 1 with increasing disutility associated with more severe states. Participants understood the tasks and the EQ-HWB-S states; and the interviewers reported high levels of understanding and engagement. The hybrid Tobit heteroscedastic model was selected for the pilot value set with values ranging from -0.384 to 1. Pain, mobility, daily activities, and sad/depressed had the largest disutilities, followed by loneliness, anxiety, exhaustion, control, and cognition in the selected model.

CONCLUSIONS

EQ-HWB-S can be valued using cTTO and DCE. Further methodological work is recommended to develop a valuation protocol specific to the EQ-HWB-S.

Interaction Effects in Health State Valuation Studies: An Optimal Scaling Approach

OBJECTIVES

This study aimed to introduce a parsimonious modeling approach that enables the estimation of interaction effects in health state valuation studies.

METHODS

Instead of supplementing a main-effects model with interactions between each and every level, a more parsimonious optimal scaling approach is proposed. 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 (ie, main effects) and the interactions between these domain-specific scales (ie, interaction effects). The number of interaction terms only depends on the number of health domains. Therefore, interactions between dimensions can be included with only a few additional parameters. The proposed models with and without interactions are fitted on 3 valuation data sets from 2 different countries, that is, a Dutch latent-scale discrete choice experiment (DCE) data set with 3699 respondents, an Australian time trade-off data set with 400 respondents, and a Dutch DCE with duration data set with 788 respondents.

RESULTS

Important interactions between health domains were found in all 3 applications. The results confirm that the accumulation of health problems within health states has a decreasing marginal effect on health state values. A similar effect is obtained when so-called N3 or N5 terms are included in the model specification, but the inclusion of 2-way interactions provides superior model fits.

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 biased value sets, particularly in stand-alone DCE with duration studies.

Urban/rural differences in preferences for EQ-5D-5L health states: a study of a multi-ethnic region in China

OBJECTIVES

This study aimed to compare health preferences for EQ-5D-5L health states between urban and rural populations in China.

METHODS

This study used pooled secondary data from two EQ-5D-5L valuation studies. Participants were recruited from Guizhou province and Chongqing municipality, China using quota sampling. Each participant was interviewed face-to-face to value a set of 15 or 16 out of 30 EQ-5D-5L health states using time trade-off (TTO) methods including composite TTO and other two variants. Regression analysis was used to compare health state preferences between urban and rural participants.

RESULTS

A total of 597 participants (urban: 55.44%; rural: 44.56%) completed the valuation interviews. Both univariate and multivariable linear regression analyses showed that rural participants tended to value health states lower than urban participants regardless of severity of health states. The unadjusted and adjusted overall mean differences between the two groups were - 0.041 (95% confidence interval [CI] - 0.077, - 0.004, p-value = 0.031) and - 0.040 (95% CI - 0.078, - 0.002, p-value = 0.038), respectively. Predictions for the 3125 health states based on rural participants' health preferences were lower than those based on urban participants' health preferences.

CONCLUSION

There were small, yet statistically significant, differences in EQ-5D-5L health states preferences between urban and rural populations in China. Future study aiming at establishing a national value set should pay more attention to the sample representativeness.