Steroids in Septic Shock: Which Patients? What Dose?

Abstract & Commentary

Synopsis: This meta-analysis of clinical trials of steroids in septic shock demonstrates that whereas short-term, high-dose administration actually worsens survival, giving lower doses for a longer period hastens shock resolution and improves survival.

Source: Minneci PC, et al. Ann Intern Med. 2004;141:47-56.

Minneci and colleagues at the national institutes of Health performed a MEDLINE search of randomized controlled trials of the effects of administration of steroids to patients with sepsis. They separated studies published before 1989 (a period in which high-dose steroids were used in the trials) from studies published after that date (when lower-doses were used). Criteria for inclusion in a meta-analysis of the combined findings were: randomized, controlled trial design; adult patients with sepsis or septic shock, with these conditions clearly defined; similar treatment of the steroid and placebo patients except for the study drug; and a specific end point (survival and/or discontinuation of vasopressor therapy). Two intensivist investigators reviewed all candidate studies and detemined their suitability for inclusion in the meta-analysis, which was then carried out by the larger group of authors.

Minneci et al examined a total of 1324 published studies, the majority of which (1162) did not meet criteria for clinical trials. Of the 162 screened clinical trials, only 11 published after 1988 met the predetermined inclusion criteria, and 6 of these were excluded because of inclusion of patients from previously reported trials (4), not having the outcome of interest (1), or having a design that precluded data interpretation (1). This left 5 trials published after 1988, on which the formal meta-analysis was performed. The results were compared to those of a previous meta-analysis of 9 similarly selected high-dose trials published before 1989.

The early studies administered short-course (1 day), high-dose steroids, starting as soon as possible (median < 2 hr) after diagnosis. The more recent trials administered lower doses, started steroids a median of 23 hours after enrollment, and continued them for a mean of 6 days (differences significant in each case). Eight trials published before 1989 demonstrated an adverse overall effect of steroid administration on survival (relative benefit, 0.89; 95% confidence interval [CI] 0.82-0.97; P = 0.008). However, aggregate data from the 5 recent trials showed a beneficial effect of steroids on survival (relative benefit, 1.23; 95% CI, 1.01-1.50; P = 0.036). The relationship between steroid dose and survival was linear, demonstrating benefit at low doses and increasing harm at higher doses (P = 0.02). In studies published before 1989, the effects of high-dose steroids on shock severity, as measured by vasopressor administration, were inconsistent; in the trials published after 1997 there was a clear (though statistically nonsignificant) trend toward more rapid shock reversal with steroids, although not all the studies provided data from which this could be determined.

Minneci et al conclude that short-course, high-dose steroids worsen survival in patients with sepsis, but that "physiologic" doses of hydrocortisone administered for 5-7 days improve survival and facilitate the more rapid resolution of shock.

Comment by David J. Pierson, MD

The use of steroids in septic shock has come full circle. The idea that large, pharmacologic doses of corticosteroids might blunt the inflammatory response and improve survival in life-threatening sepsis led to widespread administration of methylprednisolone at doses of a gram per day or more in the 1980s. Then came several studies showing not only a lack of benefit but also more frequent and more serious complicating infections in patients given steroids, with the suggestion of worsened survival, and most clinicians stopped using them during the 1990s. However, several clinical trials published in the last 6 years have suggested that steroids may be beneficial in sepsis if given at lower, physiologic doses, perhaps not from blunting the inflammatory response so much as from counteracting relative adrenal insufficiency. Whether adrenal insufficiency is what we are treating here is unclear, but the evidence is pretty good that many patients with septic shock have better outcomes if treated with low-dose hydrocortisone.

The meta-analysis of Minneci et al is not perfect, and its conclusions are heavily influenced by the findings of one study1 with substantially more patients than the others included in the analysis. However, results of the current report agree with those of a similar meta-analysis published earlier this year by Annane and colleagues,2 and together these papers represent the best available synthesis of this complicated but clinically important issue. As pointed out in the accompanying editorial by Luce,3 things will likely become much clearer when the results of a large ongoing multicenter trial4 become available. That trial, called CORTICUS, is studying the effects of hydrocortisone (50 mg IV every 6 hours for 5 days, tapered over the next 6 days) vs placebo in patients with septic shock who demonstrate a blunted response to corticotropin stimulation. After a thoughtful analysis of the current evidence for both benefit and harm, Luce3 concludes that all patients with septic shock should initially be started on low-dose "replacement" corticosteroids, and that this therapy should be continued in patients with findings on corticotropin stimulation that are consistent with either absolute or relative adrenal insufficiency. I agree. n


1. Annane D, et al. Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock. JAMA. 2002;288:862-871.

2. Annane D, et al. Corticosteroids for treating severe sepsis and septic shock. Cochrane Database Syst Rev. 2004;1:CD002243.

3. Luce JM. Physicians should administer low-dose corticosteroids selectively to septic patients until an ongoing trial is completed. Ann Intern Med. 2004;141:70-72.

4. Annanne D, et al. Corticosteroid insufficiency in acutely ill patients (letter). N Engl J Med. 2003;348:2157-2159.

David J. Pierson, MD Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington, and Editor of Critical Care Alert.