Determination of hospitalization type by investigator case report form or adjudication committee in a large heart failure clinical trial (β-Blocker Evaluation of Survival Trial [BEST])

Background: End point committees are routinely used to adjudicate efficacy and safety end points in clinical trials. The 2,708-patient β-Blocker Evaluation of Survival Trial (BEST) originally determined hospitalization type via investigator case report forms (CRFs), which captured whether a hospitalization was due to worsening heart failure (HF). Recently, the BEST End Points Committee (EPC) completed a blinded adjudication of all hospitalizations, allowing a comparison of the CRF method to the EPC method of determining hospitalization type. We sought to compare the investigator-determined mode of hospitalizations with the adjudicated events, to quantify the degree of agreement, and to compare the clinical trial results by method of event classification. Methods: The BEST EPC reviewed all 5,086 hospitalizations that occurred in BEST. Events were identified using investigator-reported hospitalizations, as well as those documented by FDA Form 3500 (MedWatch) reports. Results: The investigators identified more HF hospitalization events than adjudication (2,466 vs 1,729, P < .0001, paired analysis). Eight hundred thirty-four (34%) HF hospitalizations identified in CRFs were not confirmed by adjudication. Ninety-seven (6%) adjudicated events were not identified by the investigator reported method. One thousand six hundred thirty-two events were similarly identified by both methods. Conclusions: The EPC adjudication identified fewer HF hospitalizations than did the investigator reported method with no change in the hazard ratio for this end point. Our findings suggest that independent end point committees may improve reliability through reduced variance, thus providing similar outcome results with fewer events and no increase in CIs.