Similarities and Differences of Bayesian Designs and Adaptive Designs for Medical Devices: A Regulatory View

Abstract

This article provides a regulatory view of the design of clinical trials using Bayesian statistics and adaptive methods. The foci are the similarities and differences between Bayesian and adaptive designs for clinical studies for medical devices and the contrast with trials of pharmaceutical drugs. The critical role of the U.S. Food and Drug Administration in such designs is highlighted. The previous device experience with Bayesian trials has provided some important insight into the planning of adaptive trials. This article discusses both frequentist and Bayesian adaptive designs and presents several real examples of designs for medical device studies. The crucial role that simulations play in assessing the operating characteristics of the designs is emphasized. The unique advantage of predictive modeling in the Bayesian adaptive design is explored. For adaptive designs there are challenging problems including operational bias and logistical issues. The reporting of both Bayesian and adaptive trials is discussed.