Eight RCTs (n=805) and 14 observational (n=4,485) studies were included. Four RCTs were low quality (Jadad score 1 or 2 out of 5) and four were high quality (Jadad score 4). Four RCTs reported blinded outcome assessment.
RCTs
Pre-operative aspirin was associated with a statistically significant increase in postoperative bleeding MD 104.9 mL (95% CI: 19.2, 190.6; p=0.016; seven studies) and reoperation, OR 2.52 (95% CI: 1.18, 5.38; p=0.017; six studies) compared to control. There was moderate heterogeneity for bleeding (p=0.093, I2 41.2%). This was no longer present when studies were grouped by aspirin dose. Aspirin >325 mg was associated with a statistically significant increase in postoperative bleeding, MD 229.6 mL (95% CI: 18.7, 440.5; p=0.033; three studies). There was no significant difference between aspirin < 325 mg and control, MD 65.3 mL (95% CI: -20.2, 150.8; p=0.134; four studies). There was significant heterogeneity between these subgroups (p=0.094). Heterogeneity in the main analyses remained after grouping studies by publication date and quality. There was no statistically significant difference between aspirin and control in transfusion requirements (four studies), perioperative MI (three studies) or death (five studies).
Observational studies
Pre-operative aspirin was associated with a statistically significant increase in postoperative bleeding (10 studies) and transfusion requirements (11 studies) compared to control; statistically significant heterogeneity was found for both analyses (p<0.01, I2 71.3% for postoperative bleeding and p =0.007, I2 54.7% for transfusion requirements). There was no statistically significant difference between aspirin and control in reoperation rates (eight studies) or perioperative MI (one study).
Funnel plots were asymmetrical for bleeding and mortality among RCTs suggesting publication bias.