Using decision theory, what is an appropriate sample size for a clinical tr
ial, with a binary endpoint? We present a program, suitable for actual plan
ning, which, with some extensions, implements Canner's solution to this que
stion. Examples with a discussion are given. Implications of a Bayesian app
roach are discussed. Bayesian and Neyman-Pearson approaches are compared. C
opyright (C) 2001 John Wiley & Sons, Ltd.