Current issues of ACP Journal Club are published in Annals of Internal Medicine


Corticosteroids do not reduce mortality in sepsis and septic shock

ACP J Club. 1996 Mar-April;124:47. doi:10.7326/ACPJC-1996-124-2-047

Source Citation

Lefering R, Neugebauer EA. Steroid controversy in sepsis and septic shock: a meta-analysis. Crit Care Med. 1995 Jul;23:1294-303.



To determine the effect of corticosteroid therapy on mortality from sepsis, septic shock, or severe infections.

Data Sources

Studies were identified using MEDLINE (1966 to 1992), bibliographies of relevant articles, personal files, and contact with experts.

Study Selection

Randomized controlled trials were selected if they studied corticosteroid therapy given to patients with sepsis, septic shock, or severe infections; if the type of drug, exact dosage, and route of administration were stated; and if mortality was an end point. Studies were excluded if patients had typhoid fever, the toxic shock syndrome, or bacterial meningitis or if patients with sepsis were only a subgroup of the study population.

Data Extraction

Data obtained from each study included sample size and patient characteristics; medication, dose, and duration of therapy; mortality rate; and adverse effects. Results were combined using a model that considers the number of patients in each sample and the variability among all studies. Treatment effect was calculated as the difference in mortality rates between the treatment and control groups; a negative value represents a beneficial effect of corticosteroids.

Main Results

10 studies, with a total of 1329 patients, met the selection criteria. The pooled mortality rate difference was -0.2% (95% CI -9.2% to 8.8%). No difference in treatment effect was detected for high- compared with low-dose regimens. From the 4 studies reporting these data, the pooled mortality rate difference was -5.6% (CI -21.4% to 10.1%) in gram-negative and 1.8% (CI -15.1% to 18.6%) in gram-positive infections. 2 studies contributed significantly to the heterogeneity of the observed mortality rates; the same analyses done after these studies were excluded also failed to show a treatment effect. Corticosteroid administration had no effect on the occurrence of gastrointestinal bleeding (2.3%, CI -0.7% to 5.4%); secondary infection (0.4%, CI -4.4% to 5.2%); or hyperglycemia (0.2%, CI -4.0% to 4.4%).


Corticosteroids do not reduce mortality in patients with sepsis, septic shock, or severe infections.

Source of funding: Not stated.

For article reprint: Mr. Rolf Lefering, Biochemical and Experimental Division, Department of Surgery, University of Cologne, Cologne, Germany. FAX 49-221-893864.


Speculation has persisted for 4 decades about the potential benefit of corticosteroids in sepsis and septic shock. Scores of investigations have addressed this issue, but consensus has been elusive. Using meta-analysis, Lefering and Neugebauer made a useful framework for assessing the published data. Using different analytic strategies, Cronin and colleagues (1) published another well-executed meta-analysis 1 month later in which they pooled the results from 9 of 10 studies selected by Lefering and Neugebauer and reached the same conclusions.

The debate has been fueled by the publication of a study by Hinshaw and colleagues (2) that showed a substantial benefit of corticosteroids in animals. Survival was 100% in baboons challenged intravenously with an otherwise lethal dose of Escherichia coli when corticosteroids were given concomitantly, but survival decreased steadily to 65% when steroid administration was postponed by 4 hours. In striking contrast to clinical practice, the animals in this study were healthy, and the investigators knew precisely which pathogen was present, the site and extent of infection, and the time of treatment relative to the course of the disease.

Neither the meta-analysis by Lefering and Neugebauer nor the one by Cronin and colleagues found corticosteroid therapy to be harmful, but Cronin and colleagues saw a trend toward higher mortality associated with secondary infections (relative risk 1.7, CI 0.7% to 4.1%). Further, results from some studies suggest that corticosteroids or other modulators of inflammation may be detrimental in sepsis.

Therefore, corticosteroids are not indicated for treatment of confirmed or suspected sepsis. Additional clinical trials might be considered if the factors associated with benefit in animals can be better defined and if clinical methods that permit an earlier diagnosis of sepsis and its cause are developed.

Dante L. Landucci, MD
Robert L. Danner, MDNational Institutes of HealthBethesda, Maryland, USA


1. Cronin L, Cook DJ, Carlet J, et al. Corticosteroid treatment for sepsis: a critical appraisal and meta-analysis of the literature. Crit Care Med. 1995;23:1430-9.

2. Hinshaw LB, Beller-Todd BK, Archer LT. Current management of the septic shock patient: experimental basis for treatment. Circ Shock. 1982;9:543-53.