A systematic review of event rates in clinical trials in diabetes mellitus: the importance of quantifying baseline cardiovascular disease history and proteinuria and implications for clinical trial design.

BACKGROUND

Food and Drug Administration guidance now proposes that cardiovascular safety of new diabetes medicines be demonstrated. Consequently, trials should include a sufficient number of individuals with diabetes who are at high cardiovascular risk. We aimed to examine the impact of the presence of baseline cardiovascular disease and proteinuria, as binary criteria, on cardiovascular event rates in diabetes trials and to examine whether predicted primary end-point event rates are achieved.

METHODS

we searched Medline and EMBASE English language records (January 1998-June 2010) for randomized controlled trials of antiplatelet, lipid-lowering, antihypertensive, and glucose-lowering agents with >1,000 diabetic subjects reporting at least one of all-cause death, cardiovascular death, myocardial infarction, and stroke. Weighted mean event rates (events/1,000 patient-years) were calculated. Data from published power calculations were also compared with achieved event rates.

RESULTS

twenty-nine trials met inclusion criteria. Weighted mean event rates in diabetic subjects with and those without baseline cardiovascular disease were, respectively, 28.9 and 10.0 for all-cause death, 16.7 and 3.6 for cardiovascular death, 23.1 and 5.2 for myocardial infarction, and 12.1 and 5.4 for stroke. Event rates in diabetic subjects with and those without proteinuria were, respectively, 39.9 and 6.3 for all-cause death and 18.7 and 1.2 for cardiovascular death. Nine of 11 relevant trials achieved primary end-point event rates clearly lower than predicted.

CONCLUSIONS

trials including diabetic subjects without cardiovascular disease or proteinuria generate few events and require substantial participant numbers to achieve adequate power. However, the presence of coexisting cardiovascular disease or proteinuria increases event rates multiple-fold. Such data should aid investigators designing diabetes end-point trials.