The use of composite endpoints in cardiovascular outcome trials for diabetes: A review of 22 randomized clinical trials published since 2008

Abstract Aim To describe the use of composite endpoints (CEs) in cardiovascular outcome trials (CVOTs) of type 2 diabetes and to evaluate the significance of the individual outcomes included within these CEs from the perspectives of both patients and clinicians. Secondary objectives were to estimate the gradient of treatment effects and events across outcomes. Materials and Methods Eligible studies were randomized controlled trials assessing CV outcomes for patients with diabetes from 2008 and onwards. Trials were identified by searching the reports from the CVOT Summit of the Diabetes & CV Disease EASD (European Association for the Study of Diabetes) Study Group. The individual outcomes comprising the CE were compared for differences in importance for patients and clinicians, proportion of events, and effect size. Results We included 22 trials randomizing a mean of 8098 patients to an active intervention or a comparator group for an average of 33 months (standard deviation 16). All primary outcomes were CEs, and from a patient perspective there was no gradient of importance across outcomes in 22 of 22 (100%) CEs, while the gradient was small in 22 of 22 (100%) from a clinician perspective. The gradient of effect was moderate to large in 9 of 18 (50%) reporting studies, while assessment of events was available in 15 of 22 studies (68%), finding that three of 15 (20%) had a gradient of effect of more than 5% points between included outcomes. In 10 of 22 (45%) trial reports, the results were not clearly presented as based on a CE. Conclusions To avoid misinterpretation, clinicians and regulatory authorities should be careful when interpreting the results of trials, of which the main outcomes are CEs.