Clinical Question: Does long-term treatment with low-dose corticosteroids decrease mortality caused by severe sepsis or septic shock?
Setting: Inpatient (intensive care unit only)
Study Design: Meta-analysis (other)
Synopsis: A previous meta-analysis has shown that high-dose, short-term corticosteroids do not offer a benefit in patients with severe sepsis or septic shock (Crit Care Med 1995;23:1294–303). In this meta-analysis, investigators reexamined whether high-dose, short-term corticosteroids offer any benefit in patients with severe sepsis or septic shock and also looked at the role of low-dose, long-term steroids. The authors searched several databases for potential studies, dividing into pairs to determine the eligibility of each study. Four reviewers independently abstracted the data, contacting authors for missing data when necessary. They identified 16 randomized or quasi-randomized studies evaluating a combined 2,063 patients. Most studies were double-blind and had adequate concealment of allocation. High-dose, short-term therapy, which was defined as more than 300 mg of hydrocortisone for less than five days, had no effect on mortality rates at 28 days. However, in the five trials with low-dose (i.e., 300 mg or less), long-term steroids, all-cause mortality was approximately 20 percent less in the treated patients (relative risk = 0.8; 95 percent confidence interval [CI], 0.67 to 0.95). One death would be prevented for every nine patients treated with long-term courses of low-dose steroids instead of no additional treatment (number needed to treat [NNT] = 8.8; 95 percent CI, 5.4 to 35.7). Mortality rates in the intensive care unit also were lower (NNT = 9.1; 95 percent CI, 5.5 to 55). Steroid treatment did not increase the risk of gastrointestinal bleeding, super infections, and hyperglycemia.
Bottom Line: Although high doses of corticosteroids are ineffective for the treatment of severe sepsis or septic shock, less than 300 mg of hydrocortisone given for at least five days decreases short-term and long-term mortality rates. (Level of Evidence: 1a–)