Monte Minute‎ > ‎Monte Minute‎ > ‎

Steroids and Sepsis

posted Mar 5, 2015, 5:36 PM by Kevin Hauck

A wonderful Monte Minute brought to you by our rising Firm 1 chief resident, Rebecca Braunstein!

Steroids and Sepsis

By Rebecca Braunstein



So, pressors and shock. This is a topic we encounter often over residency, probably most in the MICU but not infrequently on the floors as well. The question is, when should we initiate steroids in sepsis?

So the first question is, why would we do that? The theoretical answer is to help an  altered or insufficient HPA axis during a time of extreme stress, leading to insufficient cortisol and mineralocorticoid. Annane et al (same as group who subsequently conducted a large RCT described below) demonstrated in an observational study JAMA 2000 that ACTH stim test had prognostic value—specifically, that higher initial cortisol level (>34) and an INADEQUATE response to ACTH stim test (change in cortisol of <9) were associated with increased mortality.

But does treating with steroids work? The earliest trials in the 1980s-1990s had varying results, ranging from no mortality benefit to as beneficial as more rapid reversal of shock and earlier ability to withdraw pressors. A concern in treating with steroids includes side effects, especially immunosuppression and superinfection.

Two larger RCT’s were subsequently done, which are the papers we most frequently reference.

 A word on types of steroids: hydrocortisone is a pharmacologic form of cortisol. It has both glucocorticoid and mineralocorticoid activity (about 50/50). Fludrocortisone (used in trial #1 below only) is a pure mineralocorticoid.

1) French trial JAMA 2002 – Annane et al.

- Design: 300 patients in shock on pressors, randomized/double blinded, randomized to receive: treatment with hydrocortisone 50 Q6hr and fludrocortisone 50mcg qday (mineralocorticoid activity only) for 7 days versus placebo. All patients underwent an ACTH stim test to assess adrenal response, and classified into responder (ie adequate adrenal response) and non-responder categories.


For primary end point of 28 day mortality:

Main finding: non-responders ie inadequate adrenal response group: decrease in 28 day mortality (hazard ratio 0.67, confidence interval 0.47-0.95)

No statistically significant difference in the following groups:

All patients: no change in 28 day mortality (hazard ratio 0.65 with confidence interval 0.39-1.07)

Adequate adrenal response group: no change in 28 day mortality (HR 0.97, CI 0.32-2.99)

 This sort of makes sense, right? The patients who were unable to mount a robust stress response benefited from the hydrocortisone/fludrocortisone replacement therapy. Of note, there was no increase in adverse effects secondary to steroids in the treatment group.

2) CORTICUS trial- NEJM 2008

- Design: 499 patients with septic shock (not necessarily on pressors), multicenter, randomized/double blinded, randomized to receive: hydrocortisone 50 Q6hr vs placebo for 5 days. ACTH stim test was done to determine adrenal response and divided into similar groups as above, responders and non-responders.


In summary, there was no mortality benefit in any group or in all patients. The treatment group had faster reversal of shock and increased incidence of superinfection, although not statistically significant.

Here are a few numbers to chew on—

For 28 day mortality, RR 1.09 (0.84-1.41) in all patients. For non-responders, RR 1.09 (0.77-1.52) and for responders, RR 1.00 (0.68-1.49).

Risk of superinfection in treatment arm vs placebo was 1.27 (0.96-1.68)— not quite significant. 


What should we make of these conflicting data?

Differences in studies:

1) Annane study had sicker patients based on clinical prediction scores, and the fact that they were all steroid dependent – this is reflected in total mortality in both placebo groups (61% in Annane vs 32% in CORTICUS).

2) Time to onset of enrollement into trial Annane et al trial within 8 hrs versus 72 hours in CORTICUS.



A number of meta-analyses have been performed. With all of this data together, the main conclusions have been:

1) Steroids are most likely to benefit patients with more severe or refractory septic shock (where fluids and pressors fail to restore hemodynamic stability).  Steroids may be harmful in septic shock of lesser severity.

2) Based on doses used in trials, “stress” steroids is recommended to be hydrocortisone 50mg Q6hr for 5-7 days with some degree of taper. Fludrocortisone is not included based on the idea that hydrocortisone has mineralocorticoid activity and a different trial (COIITSS) which did not show a difference in outcomes with addition of fludrocortisones to hydrocortisone.

3) ACTH stim test is not routinely recommended. has failed to reliably identify patients with septic shock that benefit from corticosteroid use.