Predicting survival in cost-effectiveness analyses based on clinical trials

Abstract

This study deals with the question of how to model health effects after the cessation of a randomized controlled trial (RCT). By using clinical trial data on severe congestive heart failure patients, we illustrate how survival beyond the cessation of an RCT can be predicted based on parametric survival models. In the analysis, we compare predicted survival and the resulting incremental cost-effectiveness ratio (ICER) of different survival models with actual survival/ICER. Our main finding is that the results are sensitive to the choice of survival model and that an extensive sensitivity analysis in the CE analysis is required.