Keywords: Bayesian Clinical Trial Design, Historical Data, Information Borrowing, Adaptive Design
We develop a Bayesian adaptive design which facilitates information borrowing from a historical trial using subject-level control data while assuring a reasonable upper bound on the maximum type I error rate and lower bound on the minimum power. First, one negotiates how much information may be borrowed from the historical trial, then constructs a data-based prior to be used for design and analysis of the new trial. At an interim analysis, one examines the degree of prior-data conflict. If there is too much conflict between the new trial data and the prior, the prior is discarded and the study proceeds to the final analysis where a non-informative prior is used. Otherwise, the trial is stopped early and the informative prior is used for analysis. We demonstrate our method by designing a new cardiovascular outcomes trial (CVOT) that borrows from the SAVOR trial, one of the first completed CVOTs. Cardiovascular outcome trials are commonly used to evaluate cardiovascular risk for new therapeutic agents intended for the treatment of Type 2 Diabetes Mellitus per FDA guidance. These trials are substantial in size and can take years to complete using traditional strategies.