Keywords: Historical borrowing, Bayesian, Clinical Trials
In an effort to increase the efficiency of drug development, innovative clinical trial designs have been proposed where historical control data are used in the design and analysis of a new trial. Including historical controls can potentially reduce the number of subjects needed in the new trial and increase the precision of estimating treatment effects. However, there is potential bias that may lead to false decision making. In this talk, we will first discuss the opportunities and challenges in using historical controls in modern clinical trials. Then we will propose a practical approach to leveraging historical controls in the design and analysis of a new trial. In particular, we will discuss several design considerations including compatibility of historical controls, methodology for analysis, evaluation of potential bias and impact on power and sample size. Finally, we will illustrate this proposed approach with an example in an immunology study.