The Good, the Bad, and the Fitting: A Hierarchical Bayes Model for Patient Preferences Elicited Through Discrete Choice Experiments

• Activity Number: 180
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Marketing
• Author(s): Anna Liza Malazarte Antonio*+ and Catherine Crespi and Robert E. Weiss and Christopher Saigal
• Companies: University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles
• Keywords: Bayesian model ; multiattribute options ; best-worst choice experiments ; conjoint analysis ; marginal utilities ; random effects

Abstract

Discrete choice experiments are used in health care settings to improve patient outcomes. In our study, patients needing prostate biopsies participated in a discrete choice experiment designed to elicit their preferences among health states that might occur subsequent to prostate cancer treatment. Perfect health being ideal but improbable, the health states implicitly force patients to make tradeoffs between different treatment side effects (e.g., longer or shorter expected lifespan, better or worse sexual or urinary functioning) and other treatment attributes (surgical or non-surgical, family and/or doctor support). In our particular study design, patients were presented with sets of four health states and asked to choose their most and least preferred, reducing respondent burden but leaving some preferences indeterminate. We propose a hierarchical Bayes random-effects rank-ordered multinomial logit model to estimate subject-specific regression coefficients that rank-order the various treatment side effects and options for each individual. The model accounts for missing rank information by marginalizing over all possible permutations of unranked profiles.