Thirty-two RCTs (4,260 patients) were included.
Less than 10% of the patients were removed from the analysis after treatment allocation. The most common reasons for removal were pre-operative infected urine and non-satisfactory pre-operative and/or post-operative urine culture. Randomisation was adequately concealed in 47% of the trials, and double-blinding with placebo use was present in 30% of the studies.
Bacteriuria.
Antibiotic prophylaxis significantly decreased the rate of post- operative bacteriuria from 26 to 9.1%. The RRR was 65% (95% CI: -55, -72). There was no statistical evidence of heterogeneity (Q=34, d.f.=31, p=0.31).
Septicaemia (8 RCTs, 1,979 patients).
Antibiotic prophylaxis significantly decreased the rate of post- operative septicaemia from 4.4 to 0.7%. The RRR was 77% (95% CI: -55, -88). There was no statistical evidence of heterogeneity (Q=2.7, d.f.=7, p=0.91).
Bacteriuria by antibiotic class.
Aminoglycosides (4 RCTs) significantly decreased the risk of bacteriuria. The RRR was 55% (95% CI: 0, -80, p=0.051).
Co-trimoxazole (3 RCTs) significantly decreased the risk of bacteriuria. The RRR was 64% (95% CI: -4, -87, p=0.041).
First-generation cephalosporins (3 RCTs) significantly decreased the risk of bacteriuria. The RRR was 66% (95% CI: -36, -82, p<0.01).
Second-generation cephalosporins (4 RCTs) significantly decreased the risk of bacteriuria. The RRR was 63% (95% CI: -28, -81, p<0.01).
Third-generation cephalosporins (13 RCTs) significantly decreased the risk of bacteriuria. The RRR was 67% (95% CI: -55, -76, p<0.01).
Quinolones (4 RCTs) significantly decreased the risk of bacteriuria. The RRR was 92% (95% CI: -75, -98, p<0.01).
Nitrofurantoin (1 RCT), penicillin (1 RCT) and beta-penicillin (1 RCT) did not significantly decrease the risk of bacteriuria.
Statistical heterogeneity within subgroups was not detected but there was significantly heterogeneity among subgroups (chi-squared test = 14, d.f.=8, p=0.09).
All cephalosporins combined (20 RCTs, 2,874 patients) significantly decreased the risk of bacteriuria. The RRR was 66% (95% CI: -57, -73, p<0.01). Statistical heterogeneity was not detected (Q=5, d.f.=7, p=0.64).
Bacteriuria by treatment duration. Bacteriuria was significantly decreased by all treatment duration protocols and there was no statistical heterogeneity among subgroups (chi-squared test = 2.3, d.f.=2, p=0.31).
The RRR was 72% (95% CI: -42, -87) for extended-course protocols (4 RCTs), 68% (95% CI: -56, -77) for short-course protocols (17 RCTs), and 57% (95% CI: -41, -68) for single-dose protocols (12 RCTs).
Cephalosporins significantly decreased rates of bacteriuria in all treatment duration protocols, and statistical heterogeneity between subgroups was detected (chi-squared test = 9.4, d.f.=2, p<0.01). The RRR was 82% (95% CI: -37, -95, p<0.01) for extended-course protocols (1 RCT), 71% (95% CI: -63, -77, p<0.01) for short-course protocols (12 RCTs), and 52% (95% CI: -38, -62, p<0.01) for single-dose protocols (7 RCTs).
Bacteriuria rates by methodological characteristics.
No significant differences in bacteriuria rates were found between studies in which allocation concealment was adequate and those with unclear or inadequate concealment. No differences were found between double-blind and partially-blind placebo-controlled trials.
Adverse effects.
Overall, nineteen adverse effects of antibiotics were reported. Two severe responses required antibiotic withdrawal. The adverse effects included rashes, gastrointestinal upsets, nausea and confusion.