Technological advances in real world data (RWD) captured from healthcare sources have enabled generation of an expanding body of real-world evidence (RWE) on the use of medical products. These novel sources of evidence can increase efficiencies of clinical trials by reducing sample size and/or shortening trials duration, but have yet to be fully utilized. One application of RWD that could significantly impact the conduct of clinical trials is the use of these data as external controls. Of special interest are hybrid randomized controlled trial designs, which supplement internal control arms with patients? level data from real-word data sources. D issimilarity between internal and external controls has the potential to negatively impact the trial (e.g., decrease power, inflate type I error rate) as compared to randomized control trials. Bayesian methods which adaptively adjust the influence of external controls on the analysis of the trial data can help to mitigate these issues and balance the risks and rewards associated with this type of complex trial designs. Through our collaboration with the Department of Biostatistics at the University of North Carolina (UNC) we are developing an adaptive borrowing approach with subject-specific discounting parameters specifically suited for time-to-event analyses. The proposed project would allow us to expand the UNC collaboration and develop a novel decision framework (simulation tools, including R-Packages and where computationally feasible SAS macros, and a set of study design considerations) allowing reliable application of our method when using hybrid clinical trials for regulatory decision making. We would focus on the following aims: (1) evaluation of the hybrid designs and their operating characteristics, when combined with sequential monitoring and possibly use of adaptive randomization, (2) assessment of possible extensions of the method beyond time-to-event settings when applied to diseases in different therapeutic areas, including rare diseases and (3) development of R- Packages supporting study design simulations and offering training workshops on the use of the packages to review staff at the FDA. Where computationally feasible, we will develop SAS macros as well and make these publicly available. To achieve our aims, we will utilize data from completed clinical trials, RWD sources and simulation studies. We plan to hold annual mini-conferences cross academia and industry to explore how operating characteristics of the proposed designs could be utilized for regulatory decision making and develop a recommended list of sensitivity analyses that would support regulatory submissions based on hybrid study designs. Our overarching objective is to make our developed decision framework publicly available.
Technological advances in real world data (RWD) captured from healthcare sources are continuously improving, making it possible to utilize these data for research purposes such as external controls to study arms in clinical trials. Of particular interests are hybrid randomized controlled trial designs combining internal control arms with patients? data from real-world data sources, and hence resulting in fewer patients required to be enrolled in a clinical trial. We will utilize novel statistical approaches and data from completed clinical trials, different RWD sources, and simulation studies to evaluate select hybrid designs, develop a novel decision framework in this setting and make policy changing recommendations.