Patient-Level Meta-Analysis of Low-Dose Hydrocortisone in Adults with Septic Shock.

BACKGROUND: Trials and study-level meta-analyses have failed to resolve the role of corticosteroids in the management of patients with septic shock. Patient-level meta-analyses may provide more precise estimates of treatment effects, particularly subgroup effects. METHODS: We pooled individual patient data from septic shock trials investigating the adjunctive use of intravenous hydrocortisone. The primary outcome was 90-day all-cause mortality, and it was also analyzed across predefined subgroups. Secondary outcomes included mortality at intensive care unit and hospital discharge, at 28 and 180 days, and vasopressor-, ventilator-, and organ failure­free days. Adverse events included superinfection, muscle weakness, hyperglycemia, hypernatremia, and gastroduodenal bleeding. RESULTS: Of 24 eligible trials (n=8528), 17 (n=7882) provided individual patient data, and 7 (n=5929) provided 90-day mortality. The marginal relative risk (RR) for 90-day mortality of hydrocortisone versus placebo was 0.93 (95% confidence interval [CI], 0.82 to 1.04; P=0.22; moderate certainty). It was 0.86 (95% CI, 0.79 to 0.92) for hydrocortisone with fludrocortisone and 0.96 (95% CI, 0.82 to 1.12) without fludrocortisone. There was no significant differential treatment effect across subgroups. Hydrocortisone was associated with little to no difference in any of the secondary outcomes except vasopressor-free days (mean difference, 1.24 days; 95% CI, 0.74 to 1.73; high certainty). Hydrocortisone may not be associated with an increase in the risk of superinfection (RR, 1.04; 95% CI, 0.95 to 1.15; low certainty), hyperglycemia (RR, 1.05; 95% CI, 0.98 to 1.12; low certainty), or gastroduodenal bleeding (RR, 1.11; 95% CI, 0.83 to 1.48; low certainty). Hydrocortisone may be associated with an increase in the risk of hypernatremia (RR, 2.01; 95% CI, 1.56 to 2.60; low certainty) and muscle weakness (n=2647; RR, 1.73; 95% CI, 1.49 to 1.99; low certainty). CONCLUSIONS: In this patient-level meta-analysis, hydrocortisone compared with placebo was not associated with reduced mortality for patients with septic shock. (Funded by "Programme d'Investissements d'Avenir," a research Professorship from the National Institute of Health and Care Research, Leadership Fellowships from the National Health and Medical Research Council of Australia, and Emerging Leaders Fellowship from the National Health and Medical Research Council of Australia; PROSPERO registration number, CRD42017062198.)