Eleven studies (n=at least 2,976) were included in the review: nine were retrospective (n=at least 2,778) and two were prospective (n=198). It was not possible to be certain of patient numbers as two studies reported the number of operations rather than number of patients. Newcastle-Ottawa scores ranged from 5 to 8 out of a possible 9. Follow-up was for 30 days postoperatively or for duration of hospital stay.
The total risk of postoperative complications was statistically significantly higher in patients taking corticosteroids (OR 1.41, 95% CI 1.07 to 1.87; seven studies), as was the risk of infectious postoperative complications (OR 1.68, 95% CI 1.24 to 2.28). The results were not significantly different when a random-effects model was used. An adjusted analysis that took into account the effect of confounding variables was performed for infectious complications. Adjustment by a factor of 0.9 gave an adjusted odds ratio of 1.79 (95% CI 1.32 to 2.43). There was no evidence of statistically significant heterogeneity for either analysis. Sensitivity analyses that included only studies that scored at least 7 points on the Newcastle-Ottawa scale did not materially alter the results.
There was no statistically significant difference between patients who took doses higher than 20mg/day and those who took lower doses for total complications (OR 1.35, 95% CI 0.93 to 1.97; three studies) or in infectious complications (OR 1.17, 95% CI 0.73 to 1.87; two studies). There was a statistically significantly higher rate of total complications in patients who took more than 40mg/day compared with those who took lower doses (OR 2.04, 95% CI 1.28 to 3.26; two studies); this difference was also apparent for infectious complications (OR 9.16, 95% CI 1.51 to 55.42; one study).
There was no evidence of publication bias.