Combining Hypothesis Testing with Bayesian Analyses for Dose Response and Dose Selection
View Presentation View Presentation
*David Gruben, Pfizer  Neal Thomas, Pfizer 

Keywords: Bayesian, Emax, Dose response, Dose selection, Phase 2, Phase 3, Clinical Trial

Data from an early study of a binary endpoint was used to inform the design and analysis of a second study. In particular, a Bayesian analysis with priors from the early study will give support to a fitting a 4-parameter Emax model, allowing model convergence when a frequentist approach may only allow for a 3-parameter model. Ease of implementation of the analysis, including discussion of software, will be covered. The actual data from the second study will not yet be available, so simulations may be presented. Definitions of dose and dose-selection criteria will be given and the results of a Bayesian analysis will be contrasted with that of a frequentist analysis.