Bayesian Clinical Trial Design For Categorical Endpoint Models

In the conduct of sequential clinical trials, primary statistical issues include design , monitoring and reporting. Currently, approaches built upon frequentist inference methodology predominate. Focusing on the design aspect, our objective is the development of a very general Bayesian framework permitting multiple arms with multiple patient endpoints and multiple stopping criteria. It is important to model patient response as a multidimensional phenomenon. We do so by representing it as a categorical outcome described through cells in a multiway contingency table. Speciication of a sequential Bayesian design requires a likelihood, a prior and a stopping rule. We model the likelihood using a product multinomial density, the prior using previous data available on the standard or placebo arm and employ multiple stopping criteria to deene the stopping rule. After a detailed discussion of these components , we ooer some qualitative results on the behavior of such designs and then present an example to illustrate the entire development. We conclude with a brief discussion of Bayesian post-posterior analysis.