COMPARING BEST WORST SCALING AND CONJOINT ANALYSIS TO MEASURE CAREGIVER PREFERENCES FOR THE BENEFITS AND RISKS OF EMERGING TREATMENTS FOR DUCHENE MUSCULAR DYSTROPHY

Purpose: To compare best worst scaling and conjoint analysis in measuring caregivers' preferences for the benefits and risks of emerging treatments for Duchenne muscular dystrophy (DMD).

Method: Within a single survey, two different preference elicitation methods were used to measure caregivers' treatment preference. Using a main-effects orthogonal array, 18 potential treatments were developed using six attributes (each defined across three levels) that were previously identified using a community-engaged approach. For each treatment profile, respondents were asked to identify what they viewed as the best and worst feature. They were also asked "If the treatment were real, would you use it for your child?" Good research practices were used in analyzing the data, and unadjusted results were compared graphically.

Result: The caregiver survey respondents (n=119) were more often married (90%), Caucasian (92%), biological mothers (67%). As seen in figure 1, there were qualitative differences between the two approaches, but most could be accounted for by differences in scale associated with the different nature of the dependent variable. While the benefits and risks were similarly evaluated, major differences were identified for nausea and post-market data knowledge about the drug, where monotonicity was observed for BWS, but not for the conjoint analysis.

Conclusion: BWS and conjoint analysis produced similar results for benefits and risks, but not for the other attributes. The lack of monotonicity for nausea and post-market data in the conjoint analysis could not be explained by stratifying by disease severity or via latent class analysis, leading us to assume that is was due to some unobserved framing effect. More research is needed to study differences between stated-preference methods.

Figure 1. Comparison of two methods