Four RCTs were included (n=989). All four studies reported randomisation methods, three reported allocation concealment, three were double-blinded and in the fourth the outcome assessor for surgical site infection was blinded to the treatment group. Two studies used intention-to-treat analysis. Three studies reported losses to follow-up.
Supplemental oxygenation was associated with a statistically significant reduction in surgical site infection using a fixed-effect model (RR 0.70, 95% CI 0.52 to 0.94). There was no significant difference between treatments when a random effects model was used (RR 0.74, 95% CI 0.39 to 1.43). The authors stated in the results section that moderate to heterogeneity was found (I2=77%).
Two studies reported a statistically significant reduction in surgical site infection in the supplemental oxygenation group, one reported no significant difference between treatment groups and one reported a significantly higher surgical site infection rate in the supplemental oxygenation group; this last study accounted for all of the heterogeneity. After removal of the study that accounted for the heterogeneity, the relative risk of surgical site infection associated with supplemental oxygenation was 0.55 (95% CI 0.38 to 0.80). It was unclear whether a fixed-effect or random-effects model was used for this analysis.
There were no significant differences between treatment in duration of hospital stay (three studies) or mortality (two studies).
There was no evidence of publication bias.