The benefit of corticosteroids in severe sepsis and septic shock remains controversial. We examined the benefits and risks of corticosteroid treatment in severe sepsis and septic shock and the influence of dose and duration. We searched the CENTRAL, MEDLINE, EMBASE, and LILACS (through March 2009) databases as well as reference lists of articles and proceedings of major meetings, and we contacted trial authors. Randomized and quasi-randomized trials of corticosteroids vs placebo or supportive treatment in adult patients with severe sepsis/septic shock per the American College of Chest Physicians/Society of Critical Care Medicine consensus definition were included. All reviewers agreed on trial eligibility. One reviewer extracted data, which were checked by the other reviewers and by the trials' authors whenever possible. Some unpublished data were obtained from the trials' authors. The primary outcome for this review was 28-day mortality. We identified 17 randomized trials (n = 2138) and 3 quasi-randomized trials (n = 246) that had acceptable methodological quality to pool in a meta-analysis. Twenty-eight-day mortality for treated vs control patients was 388/1099 (35.3%) vs 400/1039 (38.5%) in randomized trials (risk ratio [RR], 0.84; 95% confidence interval [CI], 0.71-1.00; P = .05; I(2) = 53% by random-effects model) and 28/121 (23.1%) vs 24/125 (19.2%) in quasi-randomized trials (RR, 1.05, 95% CI, 0.69-1.58; P = .83). In 12 trials investigating prolonged low-dose corticosteroid treatment, 28-day mortality for treated vs control patients was 236/629 (37.5%) vs 264/599 (44%) (RR, 0.84; 95% CI, 0.72-0.97; P = .02). This treatment increased 28-day shock reversal (6 trials; 322/481 [66.9%] vs 276/471 [58.6%]; RR, 1.12; 95% CI, 1.02-1.23; P = .02; I(2) = 4%) and reduced intensive care unit length of stay by 4.49 days (8 trials; 95% CI, -7.04 to -1.94; P < .001; I(2) = 0%) without increasing the risk of gastroduodenal bleeding (13 trials; 65/800 [8.1%] vs 56/764 [7.3%]; P = .50; I(2) = 0%), superinfection (14 trials; 184/998 [18.4%] vs 170/950 [17.9%]; P = .92; I(2) = 8%), or neuromuscular weakness (3 trials; 4/407 [1%] vs 7/404 [1.7%]; P = .58; I(2) = 30%). Corticosteroids increased the risk of hyperglycemia (9 trials; 363/703 [51.6%] vs 308/670 [46%]; P < .001; I(2) = 0%) and hypernatremia (3 trials; 127/404 [31.4%] vs 77/401 [19.2%]; P < .001; I(2) = 0%). Corticosteroid therapy has been used in varied doses for sepsis and related syndromes for more than 50 years, with no clear benefit on mortality. Since 1998, studies have consistently used prolonged low-dose corticosteroid therapy, and analysis of this subgroup suggests a beneficial drug effect on short-term mortality.
