An extension of Bayesian expected power and its application in decision making

Whether and how to conduct a clinical study is one of the critical decisions to be made every day in the pharmaceutical industry. Robust measurements on the probability of success (POS) of a study are crucial in the decision-making process. Among many factors that can affect the POS, establishment of the statistical alternative hypothesis (H(1)) is arguably the most important consideration. The classical power is universally used for assessing the probability of H(1) being accepted, given a certain value of the parameter of interest. If H(1) is composite, power is often provided for a range of plausible values of the parameter to reflect our uncertainty about the parameter. This could become inconvenient for decision makers and potentially introduce biases into the decision-making process. This paper proposes an extension of Bayesian expected power (eBEP) as a single metric for assessing POS defined as the probability that H(1) is accepted in a study as a tool for decision-making. eBEP is convenient, driven by scientific evidence, and systematically integrates the uncertainty, thus facilitates robust decision-making. The computation procedure of eBEP via Monte Carlo methods is provided. The application of eBEP is illustrated using a real-life bioequivalence study.