EuHEA Seminar - Edward Webb (University of Leeds)

Edward Webb

Abstract

Objectives 

National EQ-5D value sets are used to guide healthcare resource allocation, with the justification being that they represent the general public’s preferences. Value sets can remain in use for a long time, particularly in the UK where health technology assessment still uses values from 1993. There are concerns that over time value sets may no longer reflect the public’s preferences, but there is little understanding of why and how health preferences evolve over time. Many disparate factors can have an influence, including demographic changes, evolving societal attitudes and advances in preference measurement technology. We present a method which disentangles the influence on EQ-5D value sets of demographic shifts, e.g. an ageing population, from other potential sources such as changing attitudes or different measurement techniques.

Methods

In 2018 an EQ-5D-5L valuation exercise was carried out using a discrete choice experiment with values anchored to the full health=1, dead=0 scale using a visual analogue scale (VAS) exercise. The influence of two sets of demographic char- acteristics was estimated: (1) age and gender; (2) a range of characteristics including age, gender, occupation, number of children and long-term health conditions. For (1), DCE responses were analysed using multinomial logit, for (2) DCE responses were anal- ysed using the least absolute shrinkage and selection operator (LASSO). In both cases VAS responses were analysed using linear regression. Data from the Office for National Statistics was used to re-weight values to match the UK population’s characteristics in terms of the first set of characteristics, i.e. age and gender, in each year from 1990-2040. The Health Survey for England was used to re-weight values to match the population in terms of the second set of characteristics in each year from 1993-2018. 

Results 

All demographic characteristics included in the analysis influenced survey re- spondents’ valuation of at least one EQ-5D-5L level. However, values remained largely stable over time. The greatest changes between 1990 and 2040 were seen for anxi- ety/depression level 4, at -0.016 and self-care level 4, at -0.010. Including a wider range of characteristics sometimes influenced whether the utility decrement associated with a level increased or decreased over time. 

Conclusion 

No evidence was found that demographic shifts affect EQ-5D-5L values, even over a 50 year period. The method is generalisable, and similar exercises could inform future EQ-5D life-cycle research.