All Times EDT
Keywords: Historical data, real-world data, external control, matching, Bayesian extrapolation
In recent years, there has been a hyper-growth in the amount of utilizing historical clinical trial data and real-world data (RWD) in clinical trials. These data, obtained from sources other than randomized controlled trials (RCTs), offer additional opportunities to provide evidence of drug effectiveness and safety to efficiently accelerate the development of medical products. However, there are many challenges in using historical data or RWD in many aspects of clinical trials. One major aspect is related to the selection of the optimal statistical model for ensuring study compatibility and avoiding confounding effect. For example, numerous matching techniques have been considered for trials with external controlled arm (ECA) to utilize historical data or RWD, e.g., one-to-one matching, many-to-one matching, propensity score matching, etc. A comprehensive comparison of the pros and cons of these methods under different scenarios in clinical practice is needed. Furthermore, methods utilizing historical data or RWD often require prespecifying important parameters, such as the borrowing parameter and the prior distribution used in Bayesian extrapolation. How to successfully select these parameters needs to be studied in greater depth. Finally, it is still not clear how to plan the study size for a single arm study with or without concurrent control when a Bayesian borrowing is needed. These issues need to be carefully thought through before the single arm study can be launched.
Our proposed roundtable discussion is intended for audiences who are interested in discussing challenges associated with historical data or RWD for clinical trials. Speakers from regulatory agencies, pharmaceutical companies, and academia will share their latest research, practical trial examples, and potential solutions.