|
Abstract:
|
As evident by PDUFA VII commitment letter and a potential update of the 21st Century Cures Act, among other documents, the regulatory environment for utilizing real-world evidence to support different stages of drug development has become progressively favorable. With a spectrum of degrees of reliance on real-world data during the trial conduct and analysis, it is imperative to devise methodologies that can rigorously evaluate and incorporate potentially heterogeneous evidence from disparate sources. One key aspect when conducting analysis is to assess both confirmatory and contradictory knowledge during prior construction to avoid cherry-picking advantageous information. To achieve this, propensity scores are employed to minimize selection bias by incorporating supplemental control subjects and accounting for their similarity in terms of pre-treatment characteristics to the subjects in the current trial. In this paper, we consider data borrowing in a single-arm setting. The simulation experiments show that the proposed method reduces prior and data conflict and improves the precision of the of the average treatment effect.
|
