Eliciting and Anchoring Health State Preferences Using Discrete Choice Experiments Among Adults, Adolescents, and Children

Objective We aimed to compare EQ-5D-Y-5L health state preferences among children, adolescents, and adults in Canada using a discrete choice experiment (DCE), and to explore the feasibility of a rescaling latent DCE using anchoring tasks collected from adolescents. Methods An online survey was conducted to elicit preferences for EQ-5D-Y-5L health states from children (aged 12–15 years), adolescents (aged 16–17 years), and adults (aged ≥ 18 years). All respondents completed 12 latent DCE tasks. Adults and adolescents were randomly assigned to three additional anchoring tasks using a DCE with duration or with dead. The tasks were framed from the perspective of a 10-year-old child for adults and their own perspective for children and adolescents. Respondents provided feedback on the difficulty of latent DCE tasks. Mixed logit models were used to analyze latent DCE data. Anchored DCE models using duration/dead tasks were estimated and compared between adults and adolescents. Results Overall, 546 children, 508 adolescents, and 908 adults were included in the analyses. A higher proportion of children indicated it easy to complete DCE tasks compared with adolescents and adults. Monotonicity of coefficients were observed in latent DCE models among adults but not among children and adolescents. Anchored DCE modeling performed better in adults than in adolescents regarding monotonicity and statistical significance of coefficients, and the DCE with duration performed slightly better than the DCE with dead. Conclusions There were differences in health state preferences elicited using DCEs between children/adolescents and adults. Anchoring tasks appeared feasible for adolescents, with a DCE with duration performing slightly better than a DCE with dead.