Claims Data to Ascertain Clinical Events: Lost in Translation.

Accurate identification and classification of clinical events is essential to conducting clinical trials.1 The event rates in a study are highly dependent on the methods of clinical event identification, the intensity of the surveillance efforts, the accuracy of data collection, the criteria for validating events, and the overall quality control methods.2 The established standards for clinical event identification, as well as detection of adverse events, in randomized clinical trials are based on study personnel contacting study participants and their family or caregivers in conjunction with reviewing medical record documentation. Study site investigators document and classify events. Most clinical trials also use clinical events classification committees to review all relevant documentation to identify and classify clinical end points. Because these processes are time consuming, are highly resource intensive, and substantially contribute to the expense of randomized clinical trials, there has been considerable and growing interest in alternative approaches to event identification and adjudication that are more efficient and provide sufficient accuracy.1,2 Some clinical trials have begun to use medical claims and/or electronic health record data, instead of dedicated study coordinators, to identify clinical end points and adverse events as well as coded diagnoses, instead of clinical event classification committees, to classify events. In this issue of JAMA Cardiology, Guimarães and colleagues,3 using data from the Treatment With Adenosine Diphosphate ReceptorInhibitors:LongitudinalAssessmentofTreatmentPatterns and Events After Acute Coronary Syndrome (TRANSLATE-ACS) prospective observational study, evaluate the incident rates and level of agreement between claims-based ascertainment of myocardial infarction, stroke, and bleeding events in the first year after myocardial infarction and physician-based review of medical records.3 The authors found in their secondary analysis that claims-based ascertainment resulted in lower incident rates, appearing to miss events, and had modest to poor agreement with physician adjudication, appearing to misclassify events. These findings should serve as a wake-up call. Despite the momentum toward the use of less expensive, pragmatic trials and the financial attractiveness of alternative events identification and classification methods, using claims data alone for study follow-up and event adjudication may not offer, at least currently, the accuracy necessary to replace traditional approaches. Caution is needed when interpreting any study that uses these alternative methods. This analysis by Guimarães and colleagues helps highlight that there remains much to learn about optimal strategies for making clinical trials more efficient while still producing high-quality evidence.