Abstract
AIMS: To investigate cardiovascular disease and mortality trends in control arm participants of diabetes cardiovascular outcome trials (CVOTs). METHODS: We electronically searched CVOTs published before October 2017. Data on all-cause mortality, cardiovascular mortality and events, and baseline characteristics were collected, along with study calendar years. Trends were estimated using negative binomial regressions and reported as rate ratio (RR) per 5-year intervals. RESULTS: 26 CVOTs, conducted from 1961 to 2015, included 86788 participants with 6543 all-cause deaths, 3265 cardiovascular deaths, and 7657 3-point major adverse cardiovascular events (3-P MACE; combined endpoint of cardiovascular death, nonfatal myocardial infarction, nonfatal stroke). In unadjusted analysis, there was an increasing trend for 3-P MACE rates over time (5-year RR 1.57; 95% CI 1.34, 1.84); a small increasing trend for cardiovascular disease mortality rates (1.13; 1.01, 1.26); and stable rates for all-cause death. Adjusting for age, sex, previous myocardial infarction, and diabetes duration, there was no evidence of trends for 3-P MACE or cardiovascular disease mortality rates, while reducing rates were observed for nonfatal myocardial infarction (5-year RR: 0.72; 0.54, 0.96), total stroke (0.76; 0.66, 0.88), and nonfatal stroke (0.60; 0.43, 0.82). CONCLUSIONS: In contrast to real-world data, there was no evidence of an improvement in all-cause and cardiovascular mortality in type 2 diabetes participants included in control arms of randomised clinical trials across 5 decades. Further studies should investigate whether and how dissimilarities in populations, procedures, and assessments of exposures and outcomes explain the differences between real-world setting and clinical trials.