Grant: Travel Awards - Institute Community Support

Objective: Advancements in the estimation of discrete choice experiment (DCE) data allow for the quantification of individual-specific utility or quality of life values. With a view to improve the validity of DCE-derived estimates of parents' willingness to pay (WTP) for a quality of life gain, we investigate incorporating a choice certainty question into the elicitation and estimation of DCE data. Approach: The DCE elicited parents' preferences surrounding drug treatment for juvenile idiopathic arthritis (JIA). Parents chose between two alternatives differing on 6 attributes: drug administration, child reported pain, participation in activities, side-effects, school days missed, and drug cost. Respondents rated their certainty for each choice on a Likert scale (very uncertain to very certain). A mixed logit model allowed estimation of individual-specific utility values. Respondents' choice certainty was incorporated into the model by re-weighting the likelihood function. The D-error metric was used to investigate improvement in precision. Monte Carlo simulation techniques were used to generate WTP estimates. Results: 105 parents of children with JIA at The Hospital for Sick Children, Toronto, completed 16 choice and choice-certainty questions. The D-error statistic demonstrated a 53% gain in efficiency with the choice certainty approach.In the standard approach, the mean WTP for a child who moved from responding inadequately to methotrexate and adequately to etanercept was $2,105 per month; 95% of the simulated individual-level WTP estimates fell between $360 and $12,743. When choice certainty was accounted for, mean WTP was $2,167 and 95% of the simulated individual-level estimates were between $405 and $11,866. Conclusion: Parents' of children with JIA are willing to pay $2,167 per month to improve children's well-being. The choice certainty approach improved the precision of the individual-level population WTP estimates.