A Pilot Discrete Choice Experiment to Explore Preferences for EQ-5D-5L Health States

Abstract

Background. The EQ-5D-5L has recently been developed to improve the sensitivity of the widely used three-level version. Valuation studies are required before the use of this new instrument can be adopted. The use of discrete choice experiments (DCEs) in this area is a promising area of research.Purpose. To test the plausibility and acceptability of estimating an Australian algorithm for the newly developed five-level version of the EQ-5D using a DCE.Methods. A choice experiment was designed, consisting of 200 choice sets blocked such that each respondent answered 10 choice sets. Each choice set presented two health state–duration combinations, and an immediate death option. The experiment was implemented in an online Australian-representative sample. A random-effects probit model was estimated. To explore the feasibility of the approach, an indicative algorithm was developed. The algorithm is transformed to a 0 to 1 scale suitable for use to estimate quality-adjusted life-year weights for use in economic evaluation.Results. A total of 973 respondents undertook the choice experiment. Respondents were slightly younger and better educated than the general Australian population. Of the 973 respondents, 932 (95.8 %) completed all ten choice sets, and a further 12 completed some of the choice sets. In choice sets in which one health state–duration combination dominated another, the dominant option was selected on 89.5 % of occasions. The mean and median completion times were 17.9 and 9.4 min, respectively, exhibiting a highly skewed distribution. The estimation results are broadly consistent with the monotonic nature of the EQ-5D-5L. Utility is increasing in life expectancy (i.e., respondents tend to prefer health profiles with longer life expectancy), and mainly decreases in higher levels in each dimension of the instrument. A high proportion of respondents found the task clear and relatively easy tocomplete.Conclusions. DCEs are a feasible approach to the estimation of utility weights for more complex multi-attribute utility instruments such as the EQ-5D-5L.