ABSTRACT

This chapter considers the problem of making inferences about the efficacy of tamoxifen and cyclophosphamide, and their possible interaction, in the treatment of patients with early stage breast cancer. Synthesis of beliefs based on previous trial evidence and on the results of a specific factorial designed trial is used to assess the current weight of evidence for the use of tamoxifen and cyclophosphamide. This is achieved by using a Bayesian analysis of a fully parametric proportional hazards regression model in which the baseline hazard is assumed to be constant over time. This example illustrates how the Bayesian philosophy is able to formalize the assessment of the current clinical evidence.