Use of Historical Data in Design

The goal of clinical research of new disease treatments is to evaluate the potential benefits/risks of a new treatment, which generally requires comparisons to a control group. The control group is selected to characterize what would have happened to a patient if they had not received the new therapy. Although a randomized controlled trial provides the most robust clinical evidence of treatment effects, there may be situations where such a trial is not feasible or ethical, and the use of external or historical controls can provide the needed clinical evidence for granting conditional or accelerated regulatory approvals for novel drugs. Data from previous clinical trials and real-world clinical studies can provide evidence of the outcomes for patients with the disease of interest. However, many methodologic considerations must be considered such as the appropriateness of data sources, the specific data that needs to be collected, application of appropriate inclusion/exclusion criteria, accounting for bias, and statistical adjustment for confounding. This chapter reviews the settings in which the use of historical controls or comparators may be appropriate, as well as study designs, limitations to using real-world comparators, and analytic methods to compare real-world data to clinical trial data. When these considerations are appropriately handled, use of historical controls can have immense value by providing further evidence of the efficacy, effectiveness, and safety in the development of novel therapies and increase efficiency of the conduct of clinical trials and regulatory approvals.