Maximum Difference Scaling To Enhance Insight In Qualitative Payer Research

Abstract

Multiple-Criteria Decision Analysis (MCDA) considers multiple criteria in complex decision-making environments, helping to understand needs and preferences in healthcare. Here we assess the benefits of MCDA vs Likert preference scales to understand payer preferences. Multiple MCDA methods can be applied. Qualitative payer research often has restricted sample sizes and interview duration does not permit using lengthy assessments. Maximum Difference scaling has been validated for small sample sizes. From an online platform, respondents select and weight attributes and criteria relevant to their decision-making context. Payers from national, regional and local levels from France, Germany, Spain, Sweden the UK (n=5 per country) and US (n=15) underwent in-depth interviews to understand their opinions on the attributes of a novel product profile. Likert (7-point scale) and Maximum Difference exercises were completed. Median Likert scores were calculated, and hierarchical Bayesian analyses were performed on the Maximum Difference data. Likert scale results show that respondents tended to avoid extreme scores, known as ‘central tendency bias’, resulting in a restricted score range of 3-6. Thus, although interview findings provided more granularity, isolated scores were not sufficiently spread to confidently state any one attribute was preferred over another. With Maximum Difference, total score differentiation was more pronounced, with a 12 point spread between maximum and minimum values. Payer research is a key pre-market step to understand opportunities for pricing, reimbursement and market access. Unlike market research, the aim for market access research is robust insight. While Likert scales are frequently used, easy to construct and implement, validity and reproducibility are among their weaknesses. In our comparison of approaches to capturing payer preferences for product attributes, we have demonstrated that a short, well designed Maximum Difference exercise can produce clearer and less biased preference data than a Likert scale, even with a small sample size.