Pacific C
Stochastic Analysis of the Cost-Effectiveness Frontier (306432)
*Daniel F. Heitjan, Southern Methodist UniversityKeywords: Bayesian analysis, cost-effectiveness, efficient frontier, simulation model
Suppose you wish to conduct a cost-effectiveness (CE) analysis for a set of potential treatments, for which you have estimates of costs and effectiveness from some combination of clinical trials, observational studies, or empirically-informed simulation exercises. The first step in the analysis is to identify all those treatments that lie on the CE frontier; the next is to select the most effective treatment whose incremental CE ratio (compared to the next less effective one on the frontier) is less than the current willingness-to-pay. The literature on incorporating uncertainty in a comparison of two treatments, as from a clinical trial or observational study, is voluminous, but the topic of comparing several treatments when there is uncertainty has received little study. We address the latter problem from a Bayesian perspective, which renders both the synthesis and analysis of the data conceptually straightforward. We propose to present posterior probabilities of two kinds of events: i) specific configurations of CE frontiers, and ii) indicators of whether specific treatments are on the frontier. One can readily extract these probabilities from any Monte Carlo posterior com