PURPOSE: To assess the effect of systemic corticosteroids on the incidence of gastrointestinal bleeding in adult critically ill patients.

METHODS: We systematically reviewed randomised clinical trials comparing systemic corticosteroids administered for more than 24 h with placebo/no treatment in adult critically ill patients. Trial selection, data abstraction and risk of bias assessments were performed in duplicate. We used trial sequential analysis (TSA) to assess the risk of random errors and the grading of recommendations, assessment, development, and evaluations (GRADE) approach to assess the quality of evidence. The primary outcome was the incidence of clinically important gastrointestinal bleeding within 90 days. The secondary outcome was the incidence of gastrointestinal bleeding of any severity within 90 days.

RESULTS: Twenty-five trials (n = 14,615) reported data for the primary outcome and 55 trials (n = 21,792) for the secondary outcome. The pooled incidence of clinically important gastrointestinal bleeding was 2.3% in the corticosteroid group and 1.8% in the control group (RR, 1.26; 95% CI, 1.01-1.57; I2 = 0%, TSA-adjusted CI 0.51-3.14). We observed no difference in the risk of gastrointestinal bleeding of any severity (RR, 1.10; 95% CI, 0.92-1.32; I2 = 0%, TSA-adjusted CI 0.87-1.38). The GRADE quality of evidence was low (risk of bias and imprecision).

CONCLUSIONS: We observed an overall low incidence of clinically important gastrointestinal bleeding among adult critically ill patients. Corticosteroids may slightly increase the incidence of clinically important gastrointestinal bleeding, but not bleeding of any severity. Rarity of events, infrequent trial reporting and high risk of bias reduced the quality of evidence.