OBJECTIVES/BACKGROUND: As greater use is made of patient preference data, increasing attention is given to which preference elicitation methods are appropriate. The default method is the discrete choice experiment. However, this might not be appropriate, for instance, where sample sizes are small and/or the attribute list is long. This was the case with a quantitative benefit-risk assessment (qBRA) of treatments for aortic stenosis (AS) - surgical aortic valve replacement (SAVR) and transcatheter aortic valve replacement (TAVR). Instead, an adaptive swing weighting (ASW) exercise was employed. This paper describes ASW, reports on the rationale for its adoption, and lessons learned in the process of implementing it to elicit AS patients’ preferences.

METHODS: Two alternative benefit-risk frameworks with 12 attributes were identified through a literature review. SAVR and TAVR performance against these attributes was extracted from existing publications. 93 patients with AS were recruited from advocacy organization databases. An online preference survey included: attribute definitions; practice elicitation questions, the ASW exercise, and questions on numeracy and literacy, and clinical status. Responses were converted into attribute weights, which were incorporated into a qBRA.

RESULTS: In all qBRA scenarios, patients preferred TAVR to SAVR due to the value they by placed on lower short-term mortality, reduced procedural invasiveness, and quicker time to return to normal quality of life more than offsetting SAVR’s longer period of proven durability. Despite the ASW identifying significant preference heterogeneity and some potential non-attribute attendance, the validity of responses was supported by: heterogeneity being partly explained by participant age, only a small proportion of respondents straight-lining, and some measures of response consistency.

CONCLUSIONS: A BRA based on ASW was able to generate robust findings in the context of small sample size and large number of treatment attributes. Further work should test participants’ attendance to ASW tasks.