Corticosteroid treatment for sepsis: A critical appraisal and meta-analysis of the literature

Abstract

OBJECTIVE: To determine the effect of corticosteroid therapy on morbidity and mortality in patients with sepsis. DATA SOURCES: We searched for published and unpublished research using MEDLINE, EMBASE, and the Science Citation Index, manual searching of Index Medicus, citation review of relevant primary and review articles, personal files, and contact with primary investigators. STUDY SELECTION: From a pool of 124 potentially relevant articles, duplicate independent review identified nine relevant, randomized, controlled trials of corticosteroid therapy in sepsis and septic shock among critically ill adults. DATA EXTRACTION: In duplicate, independently, we abstracted key data on population, intervention, outcome, and methodologic quality of the randomized controlled trials. DATA SYNTHESIS: Corticosteroids appear to increase mortality in patients with overwhelming infection (relative risk 1.13, 95% confidence interval 0.99 to 1.29), and have no beneficial effect in the subgroup of patients with septic shock (relative risk 1.07, 95% confidence interval 0.91 to 1.26). Studies with the highest methodologic quality scores also suggest a trend toward increased mortality overall (relative risk 1.10, 95% confidence interval 0.94 to 1.29). A similar trend was observed for patients with septic shock (relative risk 1.12, 95% confidence interval 0.95 to 1.32). No difference in secondary infection rates was demonstrated in corticosteroid-treated patients with sepsis or septic shock. However, there was a trend toward increased mortality from secondary infections in patients receiving corticosteroids (relative risk 1.70, 95% confidence interval 0.70 to 4.12). The occurrence rate of gastrointestinal bleeding was increased slightly in the treatment group (relative risk 1.17, 95% confidence interval 0.79 to 1.73). CONCLUSIONS: Current evidence provides no support for the use of corticosteroids in patients with sepsis or septic shock, and suggests that their use may be harmful. These trials underscore the need for future methodologically rigorous trials evaluating new immune-modulating therapies in well-defined critically ill patients with overwhelming infection.