Selecting and Integrating Data Sources in Benefit–Risk Assessment: Considerations and Future Directions

A key challenge in benefit–risk (B-R) assessment of a medication is the multitude of data sources and the changing quality and relevance of these sources during the medication lifecycle. At the time of regulatory approval, B-R assessment is largely based on data from controlled clinical trials and preclinical studies. Following approval, data informing B-R accumulate in a broader, larger sample of subjects in clinical practice and post-approval studies, and it can be challenging to know how to appropriately aggregate and compare these datasets. In this article, we critically evaluate different data sources that may be used in B-R assessment, including controlled trials, observational data, and spontaneous reports. We demonstrate how these sources may be used to create an effects table (summary of evidence of key efficacy and safety data), using dabigatran, a nonvitamin K antagonist oral anticoagulant (NOAC), as a case example. We discuss how to compare quantitatively across studies when populations are disparate (baseline characteristics, susceptibility to adverse effects) or when there are data from multiple sources. We also discuss future directions of B-R assessment using emerging data sources and methods and the potential to use mechanistic insights from preclinical and clinical studies to support B-R assessment.