Current Controversies in Sepsis Management

Stephanie R. Moss, MD; Hallie C. Prescott, MD, MSc


Semin Respir Crit Care Med. 2019;40(5):594-603. 

In This Article

Steroids in Septic Shock—Sometimes, Always, or Never?

Owing to their myriad anti-inflammatory properties, glucocorticoids have been considered for treatment of septic shock for decades. Short courses of high-dose (6,000–42,000 mg hydrocortisone equivalent over 24 hours)[23] steroids became common practice in the mid-1970s after a study by Schumer demonstrated a marked mortality benefit.[24] However, subsequent studies not only failed to replicate the results,[25–27] but also found increased mortality due to secondary infection among steroid-treated patients.[26] Two subsequent meta-analyses found no mortality benefit,[23,28] and a third meta-analysis excluding the Schumer study found decreased survival among patients treated with high-dose steroids.[29]

However, the debate did not end there (Table 1). In the 1990s, the concept of "relative adrenal insufficiency" in sepsis gained traction.[30–32] Some patients with septic shock were found to have abnormal responses to adrenocorticotropic hormone stimulation testing, and these patients seemed to have worse outcomes than those with a normal adrenal response.[33,34] There was renewed interest in using lower "stress-dose steroids" (200–300 mg of hydrocortisone daily) for longer courses to manage relative adrenal insufficiency in vasopressor-dependent septic shock but avoid immunosuppression and secondary infections.[31,35] Several small randomized controlled trials (RCTs) found decreased time to shock reversal with stress-dose steroids,[36–39] and some demonstrated decreased mortality.[36,38] Meta-analyses pooling studies of low-dose and longer course (5–7 days) corticosteroid therapy showed improvement in survival and shock reversal.[29,40] Based on these results, the 2004 SSC guidelines recommended empiric stress-dose hydrocortisone (200–300 mg daily) in patients with septic shock who require vasopressors despite adequate fluid resuscitation.[41] The largest study included in these meta-analyses, Annane et al, demonstrated both a survival benefit and improved time to shock reversal in patients who were nonresponders to an adrenocorticotropin stimulation test.[38]

Following the 2004 SSC guidelines, a larger trial evaluating steroid therapy in septic shock was published, the CORTICUS trial (n = 499).[42] CORTICUS found no difference in 28-day mortality (34.3 vs. 31.5%, p = 0.51), regardless of corticotropin response. Time to shock reversal was faster in the steroid-treated group, but the rate of superinfection was higher.

Updated meta-analyses, including data from CORTICUS, came to differing conclusions.[43–45] Based on the available evidence, the 2016 SSC guidelines recommend intravenous (IV) hydrocortisone when adequate fluid resuscitation and vasopressors do not achieve hemodynamic stability (weak recommendation, low quality of evidence).[8]

Given the differing results of trials evaluating glucocorticoids in septic shock, variation in practice is to be expected. The PROGRESS registry is an international database including 276 contributing ICUs in 37 countries documenting use of vasopressors and steroids in patients with severe sepsis (n = 8,968).[46] It confirmed regional variability in steroid usage, with the highest rates in Europe (51.1%) and the lowest rates in Asia (21.6%). It also found that 14% of all patients received steroids despite not being on vasopressors.[46] The PROGRESS registry demonstrates that despite conflicting evidence, steroid use is widespread, including in situations (lack of vasopressor requirement) where it is not recommended.

In 2018, two large, multicenter RCTs of adjunctive glucocorticoids were published—ADRENAL (n = 3,658 patients) and APROCCHSS (n = 1,241 patients).[47,48] Whereas prior studies examined 28-day mortality, the primary outcome in both these trials was 90-day mortality. ADRENAL randomized patients to a continuous infusion of hydrocortisone (200 mg/kg) for 7 days, or until discharge from the ICU. Meanwhile, APROCCHSS randomized patients to hydrocortisone 50 mg IV every 6 hours plus fludrocortisone 50 μg daily for 7 days. (In the earlier years of the APROCCHSS trial, patients were also randomized in a factorial fashion to activated protein C, prior to its removal from the market in 2011.)

ADRENAL found no difference in 90-day mortality.[47] However, patients randomized to steroids fared better on several secondary endpoints, including faster time to shock reversal (3 vs. 4 days, p < 0.001), faster time to extubation (6 vs. 7 days, p < 0.001), shorter ICU length of stay (10 vs. 12 days, p < 0.001), and lower rate of blood transfusion (37.0 vs. 41.7%, p = 0.004). There was no difference in new bacteremia or fungemia at day 14.[47] By contrast, APROCCHSS had lower 90-day mortality in the steroid arm (43.0 vs. 49.1%, p = 0.03). Secondary end-points were likewise better in the steroid arm: greater vasopressor-free days (17 vs. 15 days, p < 0.001) and organ-failure-free days (14 vs. 12 days, p = 0.003).[48]

The conflicting mortality findings from these two studies have generated much discussion. There are several differences that could potentially explain the discrepant results. Control group mortality was substantially higher in APROCCHSS (49 vs. 29%). However, even in the sicker subset of 903 ADRENAL patients who would have met enrollment criteria for APROCCHSS, the difference in mortality was still nonsignificant, and the relative risk reduction associated with steroids was less than in APROCCHSS. Thus, differences in illness severity alone do not fully explain the discrepant results. Other potential explanations include the use of fludrocortisone, the trial location (France for APROCCHSS vs. Australia, United Kingdom, New Zealand, Saudi Arabia, and Denmark for ADRENAL), or the trial enrollment dates (2008–2015 for APROCCHSS vs. 2013–2017 for ADRENAL). Regardless of the differences in mortality, however, both studies suggest overall benefit from steroids in secondary outcomes, with no signal for increased harm.[47,48]

Since the publication of the ADRENAL and APROCCHSS trials, two updated meta-analyses have been published. The first evaluated 42 RCTs enrolling 10,194 patients and found that steroid treatment had a relative risk (RR) of 0.93 (95% confidence interval [CI]: 0.84–1.03) for short-term (28-day) mortality and a RR of 0.94 (95% CI: 0.89–1.00) for long-term (60-day to 1 year) mortality.[49] They concluded that there may be a possible small versus no reduction in mortality with steroids.[49] The second meta-analysis evaluated 37 RCTs of 9,564 patients and found decreased 28-day mortality with steroids (RR: 0.90; 95% CI: 0.82–0.98) but no difference for 90-day mortality (RR: 0.94; 95% CI: 0.85–1.03).[50] Both meta-analyses found earlier shock reversal and lower sequential organ failure assessment (SOFA) scores at day 7 with steroid treatment and neither showed a difference in rates of superinfection.[49,50] These meta-analyses both conclude that although the absolute benefit is small, there is low risk of adverse events, and therefore there is likely net benefit to low-dose glucocorticoids, particularly in the sickest patients.[49,50]

In summary, there remains controversy regarding the role of glucocorticoids in septic shock. RCTs do not show a consistent mortality benefit. However, they do show a consistent decrease in time to shock reversal with steroid treatment. Our practice is to use glucocorticoids on a case-by-case basis for patients with high or persistent (i.e., cannot down-titrate within 8–12 hours) vasopressor requirements.