All Times EDT
Keywords: External Trial Subjects, Hybrid Control, Randomized Clinical Trials, Type I Error
Leveraging external trial data in clinical trials has been a topic receiving increased attention. Of interest is the promising ethical benefit to reduce the number of subjects in control groups who may not receive treatment benefit and the promising operational benefit to design and conduct studies more efficiently. Recent research reveals issues with using hybrid controls by statistical matching in baseline characteristics, mainly because the contemporaneous and operational differences have not been considered between the external data and the trial data. Conditional borrowing approaches can substantially control both bias and type I error, but do have certain small type I error inflation in some circumstances. This presentation discusses several strategies that can improve the control of the type I error without great loss in power. Simulation results will be discussed and a case study on how best to apply the conditional hybrid approach will be discussed.