Adaptive Bayesian randomized trials: realizing their potential.

Abstract

Bayesian adaptive designs offer the flexibility to adapt the conduct of a trial in response to accumulating evidence. This ability of the trial to adapt is especially appealing when studying an intervention that evolves rapidly. Examples of design adaptations include dropping arms from a study, sample size re-estimation, or stopping a trial early. The choice of adaptive design must match the objectives of the trial, and this paper gives examples of designs suitable for exploratory trials, explanatory trials, and trials assessing cost-effectiveness.