|
Abstract:
|
Drug development involves a long costly journey from discovery to registration during which continuous learning about the efficacy and safety of candidate drugs is critical for decision making and resource allocation. In the clinical stage, the efficacy of a drug is traditionally established with a properly powered Phase 2 proof-of-concept study. However, for a variety of reasons, small efficacy studies are used for an early readout of the potential efficacy of the drug in many development programs. Limited efficacy data may be collected from Phase 1 studies, or sponsors conduct very small efficacy studies with the intent of de-risking prior to larger resource spend. Such studies are usually severely under-powered, and thus not reliable enough to appropriately inform go/no-go decisions solely based these data. We examine risks associated with under-powered studies and propose a framework for interpreting and contextualizing results from such studies using Bayesian methodology. Hypothetical examples based on real patient data are presented for illustration.
|
