ObjectiveThe use of corticosteroids in patients with sepsis or septic shock has been controversial for many decades. Clinical studies have reported beneficial, as well as negative results. We conducted a meta-analysis to assess the clinical evidence and to evaluate treatment effects in specific subgroups of patients.DesignMeta-analysis.Data SourcesA comprehensive search of the literature revealed 49 publications investigating the effect of corticosteroids in patients with sepsis and septic shock.Study SelectionOnly ten of the 49 publications were prospective, randomized, controlled trials with an exact description of dosage and regimen.Date ExtractionTreatment effects on mortality were calculated as rate differences in each study (negative values favor steroids), and were combined with respect to the variability in each study.Data SynthesisOnly one study showed a significantly positive effect of steroid treatment. Overall, no positive effect was observed: -0.2% (95% confidence interval: -9.2, 8.8). There were no differences observed when comparing low- vs. high-dose or type of corticosteroid used. Comparing patients with proven Gram-positive or Gram-negative infection showed a slight but not significant difference. The Gram-negative group demonstrated better outcome (-5.6% vs. 1.8%). A quality rating of each trial showed a remarkable increase in quality over time. Adverse events (gastroin-testinal bleeding, secondary infections, hyperglycemia) were not more frequent in patients treated with steroids compared with controls.ConclusionsNo overall beneficial effect of corticosteroids in patients with septic shock was observed; however, there is some evidence for a positive effect in patients with Gramnegative septicemia.(Crit Care Med 1995; 23:1294-1303)