Primum non nocere: An updated meta-analysis on aspirin use in primary prevention of cardiovascular disease in patients with diabetes

Abstract

Individuals with diabetes have a twoto four-fold increased risk of vascular events. A pivotal role in this setting is played by an exaggerated platelet activation. This has led to increasing interest over recent decades in developing interventions aimed at reducing cardiovascular risk in this population. Although the effectiveness of aspirin in secondary prevention of cardiovascular disease (CVD) in diabetes is established, its use in primary prevention is still controversial and current European Society of Cardiology guidelines on CVD prevention do not provide a clear recommendation. A number of randomized trials have reported on the role of aspirin in primary prevention of CVD in people with diabetes, but the majority of these studies were poorly powered with regard to the number of people with diabetes and reported results from subgroups. The recently published ASCEND trial provided new evidences specifically assessing this issue in a large population of diabetic patients. We performed an updated study-level meta-analysis to evaluate the efficacy and safety of aspirin in primary prevention of CVD in diabetic patients. Electronic databases were searched for clinical trials that randomized diabetic patients without previous cardiovascular events to aspirin or placebo. The process was performed according to the PRISMA statement. Two authors extracted data on patient characteristics and outcomes. Efficacy outcomes were major adverse cardiovascular event (MACE), myocardial infarction (MI), stroke, and cardiovascular and all-cause mortality. Safety endpoints were major bleedings and malignancies. Net adverse clinical events (NACEs) were computed as the unweighted sum of haemorrhagic stroke, major extracranial bleeding, ischaemic stroke and MI and as weighted sum of fatal equivalent of the above events using weights and methods previously reported. Random-effects risk ratios (RRs) were estimated using a Mantel–Haenszel method with a person-year approach. Heterogeneity was calculated using the I test and publication bias was visually assessed with funnel plots. Univariate meta-regression for unadjusted log-RR was performed to explore the potential moderator effect of mean age, aspirin dose, smoking status, hypertension, gender, type of diabetes, concomitant statin therapy and compliance to study treatment on MACE. Statistical analyses were conducted using Review Manager version 5.3 and OpenMeta-Analyst. Five trials met our inclusion criteria and were included in the analyses (N1⁄4 24,037; mean follow-up 5.43 1.56 years; Table 1). Aspirin reduced the risk of MACE compared with placebo (1.5% vs. 1.7% person-year; RR: 0.90; 95% confidence interval (CI): 0.83–0.97; p1⁄4 0.004; number needed to treat (NNT) over the mean follow-up1⁄4 88). As shown in Figure 1 the risk of MI, stroke, and cardiovascular and all-cause mortality did not differ between the two groups. Aspirin significantly increased the risk of major bleedings compared with placebo (0.53% vs. 0.43% person-year; RR: 1.24; 95% CI: 1.06–1.45; p1⁄4 0.03; number needed to