Twenty six RCTs were included for review (n=11,343). Randomisation was unclear in 21 studies. Allocation concealment was unclear in 22 studies. Eight studies had double or triple blinding. Seventeen studies had adequate baseline comparability. All but one study defined eligibility criteria. Sample size calculation was not reported or was unclear and intention-to-treat analysis was unclear or inadequate for 22 studies.
Prophylactic antibiotics significantly reduced the risk of wound infection compared to no antibiotics or placebo (RR 0.19, 95% CI 0.12 to 0.31, p<0.00001; seven studies, n=3,065). There was no evidence of significant statistical heterogeneity. There was no significant difference between systemic antibiotics and antibiotic impregnated cement (three studies, n=2,388), between cephalosporins and teicoplanin (five studies, n=2,625), between cephalosporins and penicillin derivatives (three studies, n=386) and between first and second generation cephalosporins (eight studies n=2,879) in the rate of wound infection. Statistical heterogeneity for these analyses was low or absent.