Methods for safety and endpoint ascertainment: identification of adverse events through scrutiny of negatively adjudicated events

Abstract Background The primary goal of phase 2 and 3 clinical trials is to evaluate the safety and effectiveness of therapeutic interventions, and efficient and reproducible ascertainment of important clinical events, either as clinical outcome events (COEs) or adverse events (AEs), is critical. Clinical outcomes require consistency and clinical judgment, so these events are often adjudicated centrally by clinical events classification (CEC) physician reviewers using standardized definitions. In contrast, AEs are reported by sites to the trial coordinating center based on common reporting criteria set by regulatory authorities and trial sponsors. These different requirements have led to the development of separate tracks for COE and AE review. Main body Potential COEs that fail to meet standardized definitions for CEC adjudication – i.e. negatively adjudicated events (NAE) – may meet criteria for AEs. Trial oversight practices require the sponsor to process AEs regardless of how the AEs are submitted; therefore, review of NAEs may be necessary to ensure that important AEs do not go unreported. The Duke Clinical Research Institute (DCRI) developed and implemented a process for scrutinizing NAEs to detect potential missed serious AEs. Initial experience with this process across two trials suggests that approximately 0.2% of NAEs are serious unexpected AEs that were not otherwise reported and another 1.5% are serious expected AEs. Conclusions Given their infrequent concealment of serious AEs in two large trials assessing cardiovascular outcomes, routine scrutiny of NAEs to identify AEs is not recommended at this time, though it may be useful in some trials and should be carefully considered by the trial team. Closer integration of data across safety surveillance and endpoint adjudication systems may enable scrutiny of NAEs when indicated while limiting complexity associated with this process.