Seven RCTs (11,618 patient) were included in the review. Two trials scored a maximum of 5 points for quality. Allocation concealment, placebo or active comparator control, and blinding of participants and outcome-assessors was reported in five RCTs. Loss to follow-up was reported in six RCTs. Intention-to-treat analysis was reported in four RCTs. No trials reported on potential baseline differences.
There were suggestions of the beneficial effect of aspirin, but the results did not reach statistical significance for major adverse cardiovascular events (RR 0.91, 95% CI 0.82 to 1.00; six RCTs), myocardial infarction (RR 0.85, 95% CI 0.66 to 1.10; seven RCTs), stroke (RR 0.84, 95% CI 0.63 to 1.11; six RCTs), cardiovascular death (RR 0.95, 95% CI 0.71 to 1.27; five RCTs) or all-cause mortality (RR 0.95, 95% CI 0.85 to 1.06; five RCTs). There was evidence of heterogeneity for the analyses of myocardial infarction (Ι²=53.1%), stroke (Ι²=47.4%) and cardiovascular death (Ι²=41.1%). Subgroup analysis did not substantially change the results.
In the aspirin groups, there were higher rates of hemorrhagic complications (RR 2.50, 95% CI 0.77 to 8.10), gastrointestinal bleeding (RR 2.13, 95% CI 0.63 to 7.25), or gastrointestinal events not resulting in bleeding (RR 2.92, 95% CI 0.17 to 50.23) compared with control but these did not reach statistical significance.
Ninety-two people would need to be treated to prevent one major cardiovascular event. The estimate of the 'the likelihood of being helped versus harmed metric' was 6 (range 2 to 8).
There was no significant evidence of publication bias.