In many clinical trials, especially in cardiology, the primary endpoint is the time to first cardiovascular event, defined as the first of any constituent events such as acute myocardial infarction, cardiovascular death, or stroke. An Endpoint Adjudication Committee (EAC) adjudicates whether an investigator-reported event is a primary event constituent. The use of an EAC can delay interim analyses (IAs) of accumulating study data, and add uncertainty to predicting IA outcomes.
To mitigate delays imposed by adjudication processes, we consider two methods of supporting more “up-to-date” IAs. In the first, results of prior adjudications are used to predict classifications for unadjudicated potential events, resulting in an adjusted analysis where multiple events for each subject are weighted. In the second, Bayesian logistic regression is implemented, to predict classifications for unadjudicated potential events. We compare the performances of the Kaplan-Meier curves resulting from each of these methods to each other and to that of the “true” curve that would be obtained by waiting for all adjudications.