A SAMPLE-SIZE-OPTIMAL BAYESIAN PROCEDURE FOR SEQUENTIAL PHARMACEUTICAL TRIALS

Consider a pharmaceutical trial where the consequences of different de cisions are expressed on a financial scale. The efficacy of the new dr ug under consideration has a prior distribution obtained from the unde rlying biological process, animal experiments, clinical experience, an d so forth. Berry and Ho (Biometrics 44, 219-227) show how these compo nents are used to establish an optimal (Bayes) sequential testing proc edure, assuming a known constant sample size at each decision point. W e show in this article how it is also possible to optimize further, wi th respect to the sample-size rule. This last component of the design, which is missing from most sequential procedures, has the potential t o yield considerably larger expected net gains (equivalently, consider ably smaller Bayes risks).