Six RCTs (974 patients) were included.
Less than 10% of the enrolled participants were excluded after randomisation. The main reasons for removal were that the patient did not receive the intervention as stipulated, or the patient converted to a higher risk surgical procedure. The quality scores ranged between 1 and 4; the authors reported that no interaction between study quality and the results was found.
Overall infection.
ABP did not significantly reduce the rate of overall infection (2.8% compared with 4.4%). The OR was 0.69 (95% CI: 0.34, 1.43). There was no evidence of statistical heterogeneity (Q=2.14, d.f.=5, P=0.83). The use of RDs gave a difference of 1.2% (95% CI: -3.1, 0.7) in favour of antibiotic therapy, and the NNT to prevent one infection was 63 (95% CI: 24, infinity) patients.
Surgical site infections.
ABP did not significantly reduce the rate of surgical site infections (2.1% compared with 2.9%). The OR was 0.82 (95% CI: 0.36, 1.86). There was no evidence of statistical heterogeneity (Q=1.76, d.f.=5, P=0.88). The use of RDs gave a difference of -0.1% (95% CI: -2.4, 0.5) in favour of antibiotic therapy, and the NNT to prevent one infection was 121 (95% CI: 32, infinity) patients.
Other site infections.
ABP did not significantly reduce the amount of other site infections (0.7% compared with 1.5%). The OR was 0.82 (95% CI: 0.18, 1.90). There was no evidence of statistical heterogeneity (Q=1.78, d.f.=5, P=0.88). The use of RDs gave a difference of -0.3% (95% CI: -1.6, 0.9) in favour of antibiotic therapy, and the NNT to prevent one infection was 131 (95% CI: 40, infinity).