A new approach to comparing EQ-5D-5L value sets: What really makes a difference to people?
Background
EQ-5D-5L value sets play a vital role worldwide in health technology assessment and measuring population health. However, it is difficult to compare the impact of using different value sets, for example if an older value set is replaced by a new one, or if two or more value sets are available.
Objectives
1) Reformulating the standard EQ-5D-5L health-state utility model in terms of meaningful, interpretable parameters: range of utility from best to worst state, relative importance of each dimension, and decrements associated with levels 2-4 as a fraction of the level 5 decrement;
2) Providing formulae for the marginal effects of the reformulated model’s parameters on utilities;
3) Empirically estimating marginal effects for several value sets and patient groups.
Data and Methods
The 2017/8 Health Surveys for England collected EQ-5D-5L responses from around 13,000 people. Participants with a long-term condition were divided into patient groups according to how the condition affected them (vision/mobility/mental health/etc.). Participants with long-lasting pain were divided into groups of varying severity using the Chronic Pain Grade Scale (CPGS). The marginal effects of parameters from 48 EQ-5D-5L value sets were calculated for each patient group. In countries with multiple value sets available, differences in parameters were compared to differences in mean utility.
Results
Parameters representing value set range had consistently large marginal effects across value sets and patient groups. Increased importance of self-care relative to other dimensions always increased utility, though with variable effect sizes. The direction and magnitude of the marginal effects of other dimensions’ relative importance varied across value sets and patient groups. The magnitude of all parameters’ marginal effects increased with condition severity.
In within-country comparisons the magnitudes of parameter differences were related to utility differences, with the smallest and largest differences for both seen when comparing the Japanese and Peruvian value sets respectively. Utility differences were increasing in disease severity: the utility of people with the worst chronic pain level differed by 0.425 depending on which Peruvian value set was used.
Conclusion
This study presents innovative methods for comparing EQ-5D-5L value sets and assessing what aspects have the biggest impact on patient utilities. Value set range was a major driver of utility differences, and people with more severe conditions were most likely to be impacted by using alternative value sets.
Speaker:
Ed Webb is an associate professor at the Academic Unit of Health Economics, University of Leeds. He studies preferences and decision-making in a wide variety of contexts and using a range of methods. His other research interests include the measurement and valuation of health, as well as interactions between health, work and wellbeing.