Liberal or restrictive antimicrobial prophylaxis for surgical site infection: systematic review and meta-analysis of randomised trials

Background: Antimicrobial prophylaxis is widely used to prevent surgical site infection. Amid growing concern about antimicrobial resistance, we determined the effectiveness of antimicrobial prophylaxis.

Methods: We searched MEDLINE, EMBASE, CENTRAL, and WHO-ICTRP between January 1, 1990 and January 1, 2020 for trials randomising adults undergoing surgery to liberal (more doses) or restrictive (fewer or no doses) perioperative antimicrobial prophylaxis. Pairs of researchers reviewed articles and extracted data, and a senior author resolved discrepancies. The primary outcome measure was surgical site infection or bacteriuria for urological procedures. We calculated average risk difference (RD) with 95% confidence intervals and prediction intervals (PI) using random effects models, and present risk ratios (RR). We assessed evidence certainty using GRADE methodology, and risk of bias using the Cochrane Risk of Bias tool (PROSPERO: CRD42018116946).

Results: From 6593 records, we identified 294 trials including 86 146 patients. Surgical site infection occurred in 2237/44 113 (5.1%) patients receiving liberal prophylaxis vs 2889/42 033 (6.9%) receiving restrictive prophylaxis (RD -0.01 [-0.02 to -0.01]; relative risk 0.72 [0.67-0.77]; I²=52%, PI -0.05-0.02). There was a small benefit of prophylaxis in 161 trials comparing no prophylaxis with ≥1 dose (RD -0.02 [-0.03 to -0.02]; RR 0.58 [0.52-0.65]; I²=62%, PI -0.06-0.02). Treatment effect varied from a strong effect in urology to no benefit in 7/19 specialities. Tests for publication bias suggest 62 unreported trials and evidence certainty was very low. Treatment harms were reported in 43/294 trials.

Conclusions: A systematic review and meta-analysis of randomised trials revealed that more liberal antimicrobial prophylaxis is associated with a small reduction in the risk of surgical site infection, although antimicrobial harms are poorly reported. Further evidence about the risks of antimicrobial prophylaxis to inform current widespread use is urgently needed.