Corticosteroids probably lower short-term mortality in sepsis; longer-term effects remain uncertain
Across randomized trials in people with sepsis, corticosteroids probably reduced early mortality compared with placebo or usual care. Longer-term mortality and some harms remained uncertain.
Based on the published abstract. The full paper may contain additional methods, results and limitations.
The 30-second takeaway
For people with sepsis, corticosteroids probably reduced early mortality compared with placebo or usual care, based on moderate-certainty evidence. The pooled longer-term mortality result was less certain and allowed both lower and higher mortality. The review also found uncertainty about muscle weakness and about whether continuous infusion differs from intermittent dosing.
Differences between trial results lowered confidence in the early mortality estimate. An updated search had not yet been incorporated, and evidence for several harms and dosing comparisons remained limited or uncertain.
What researchers found
All-cause mortality
Versus placebo or usual care: risk ratio 0.89 (95% CI 0.84–0.95), with moderate-certainty evidence. This is a relative comparison, not an absolute death-rate difference.
Harms had different certainty
The review found that corticosteroids may make little or no difference to superinfection, meaning an additional infection. Evidence about muscle weakness was much less certain. An uncertain harm estimate cannot establish that the treatment is harmless.
Administration remained an open comparison
Trials comparing continuous infusion with intermittent doses provided very low-certainty evidence. The review could not establish which approach produced better mortality, hospital-stay, or harm outcomes. This is a separate question from corticosteroids versus usual care.
The original publication
Corticosteroids for treating sepsis in children and adults.
Annane D, Briegel J, Granton D et al.
Cochrane Database Syst Rev · 2025
- PubMed ID
- 40470636
- Record checked
Linked publication notices
AI-assisted research and writing. This explains one selected publication; it is not a complete review of everything known. Our approach.
One more question, understood.
Keep track of the research you’ve explored.