OBJECTIVES: This research explores the application of patient preferences, as estimated by a discrete choice experiment (DCE), to hypothetical clinical trial results in childhood autism spectrum disorder (ASD), as measured by (between-group) changes from baseline. In childhood ASD, the Childhood Autism Rating Scale–2nd edition (CARS2) instrument is used to assess severity and change.

METHODS: A DCE was conducted, estimating the value of improvements within the standard version of CARS2, i.e. CARS2-ST. Caregivers were asked to participate, expressing preferences from the perspective of the child. As such, “utility” values were obtained for CARS2-ST-based profiles. Subsequently, hypothetical changes in CARS2-ST were simulated, as if measured in a trial, and the corresponding results in terms of “utility” were analysed: (1) A 3-point improvement was applied to a single patient; (2) A 3-point average group improvement was simulated under 3 conditions, wherein the improvement was derived randomly across the CARS2-ST, derived from the 6 attributes with the greatest preferences, and derived from the 6 attributes with the lowest preferences.

RESULTS: For one patient, a 3-point improvement in the CARS2-ST, increased the CARS2-ST numeric score by 6.5%; however, the latent utility function increased by 21% when this concerned the attributes with highest preference weights; and 6.3% when this concerned the attributes with the lowest preference weights. Using simulated trial data, the three scenarios resulted in very different impacts on latent utility. Depending on the source of CARS2-ST improvement, there was substantial variation in the value to patients.

CONCLUSIONS: The assessment of clinical trial measures should, where possible, account for the preferences of relevant groups, e.g., patients, caregivers and clinicians. Conclusions made from non-preference-based scoring alone can hide meaningful information on how a treatment will affect a patient’s health-related quality of life.