Corticosteroids in Sepsis: An Updated Systematic

Review and Meta-Analysis

Bram Rochwerg, MD, MSc1,2; Simon J. Oczkowski, MD, MSc, MHSc1; Reed A. C. Siemieniuk, MD2;
Thomas Agoritsas, MD, PhD2,3,4; Emilie Belley-Cote, MD1,2; Frédérick D’Aragon, MD, MSc5;
Downloaded from https://journals.lww.com/ccmjournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3/6NA8SYFqVTwJMhFNcntdHC7Mdsl4IJaSNlmodo/Vrk= on 08/16/2018

Erick Duan, MD, MSc1,2; Shane English, MD, MSc6,7; Kira Gossack-Keenan, BSc1;
Mashari Alghuroba, MSc1; Wojciech Szczeklik, MD, PhD1,8; Kusum Menon, MD, MSc9;
Waleed Alhazzani, MD, MSc1,2; Jonathan Sevransky, MD10; Per Olav Vandvik, MD, PhD11;
Djillali Annane, MD, PhD12; Gordon Guyatt, MD, MSc1,2

1
Department of Medicine, McMaster University, Hamilton, ON, Canada. Objective: This systematic review and meta-analysis addresses
2
Department of Health Research Methods, Evidence and Impact, McMas- the efficacy and safety of corticosteroids in critically ill patients
ter University, Hamilton, ON, Canada. with sepsis.
3
Division of General Internal Medicine, University Hospitals of Geneva, Data Sources: We updated a comprehensive search of MEDLINE,
Geneva, Switzerland.
EMBASE, CENTRAL, and LILACS, and unpublished sources for ran-
4
Division of Clinical Epidemiology, University Hospitals of Geneva,
Geneva, Switzerland. domized controlled trials that compared any corticosteroid to placebo
5
Centre de Recherche du Centre Hospitalier Universitaire de Sherbrooke or no corticosteroid in critically ill children and adults with sepsis.
et Faculté de médecine et des sciences de la santé, Université de Sher- Study Selection: Reviewers conducted duplicate screening of
brooke, Sherbrooke, QC, Canada.
citations, data abstraction, and, using a modified Cochrane risk of
6
Department of Medicine (Critical Care), University of Ottawa, Ottawa,
ON, Canada.
bias tool, individual study risk of bias assessment.
7
Clinical Epidemiology Program, Ottawa Hospital Research Institute,
Data Extraction: A parallel guideline committee provided input on
Ottawa, ON, Canada. the design and interpretation of the systematic review, including
8
Department of Intensive Care and Perioperative Medicine, Jagiellonian the selection of outcomes important to patients. We assessed
University Medical College, Krakow, Poland. overall certainty in evidence using Grading of Recommendations
9
Department of Pediatrics, University of Ottawa, Ottawa, ON, Canada. Assessment, Development and Evaluation methodology and per-
10
Division of Pulmonary, Allergy and Critical Care, Department of Medi- formed all analyses using random-effect models. For subgroup
cine, Emory University, Atlanta, GA.
analyses, we performed metaregression and considered p value
11
Department of Medicine, Innlandet Hospital Trust-Division, Gjøvik, Norway.
less than 0.05 as significant.
12
Hôpital Raymond Poincaré, Laboratory of Infection and Inflammation,
University of Versailles, Garches, France. Data Synthesis: Forty-two randomized controlled trials including
Supplemental digital content is available for this article. Direct URL cita- 10,194 patients proved eligible. Based on low certainty, cortico-
tions appear in the printed text and are provided in the HTML and PDF steroids may achieve a small reduction or no reduction in the rela-
versions of this article on the journal’s website (http://journals.lww.com/ tive risk of dying in the short-term (28–31 d) (relative risk, 0.93;
ccmjournal).
95% CI, 0.84–1.03; 1.8% absolute risk reduction; 95% CI, 4.1%
Drs. Rochwerg and Oczkowski are supported by McMaster University
Department of Medicine early career research awards. Dr. Siemien- reduction to 0.8% increase), and possibly achieve a small effect on
iuk disclosed off-label product use of corticosteroids (not licensed long-term mortality (60 d to 1 yr) based on moderate certainty (rel-
for treatment of sepsis). Dr. Sevransky’s institution received funding ative risk, 0.94; 95% CI, 0.89–1.00; 2.2% absolute risk reduction;
from the Marcus Foundation, and he received funding from the Society
of Critical Care Medicine (Associate Editor stipend). The remaining 95% CI, 4.1% reduction to no effect). Corticosteroids probably
authors have disclosed that they do not have any potential conflicts result in small reductions in length of stay in ICU (mean difference,
of interest.
–0.73 d; 95% CI, –1.78 to 0.31) and hospital (mean difference,
Address requests for reprints to: Bram Rochwerg, MD, MSc, Department
–0.73 d; 95% CI, –2.06 to 0.60) (moderate certainty). Corticoste-
of Medicine, Division of Critical Care, Juravinski Hospital, 711 Concession
St, Hamilton, ON L8V 1C1, Canada. E-mail: rochwerg@mcmaster.ca roids result in higher rates of shock reversal at day 7 (relative risk,
Copyright © 2018 by the Society of Critical Care Medicine and Wolters 1.26; 95% CI, 1.12–1.42) and lower Sequential Organ Failure
Kluwer Health, Inc. All Rights Reserved. Assessment scores at day 7 (mean difference, –1.39; 95% CI,
DOI: 10.1097/CCM.0000000000003262 –1.88 to –0.89) (high certainty). Corticosteroids likely increase the

Critical Care Medicine www.ccmjournal.org 1411

Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Rochwerg et al
risk of hypernatremia (relative risk, 1.64; 95% CI, 1.32–2.03) and
hyperglycemia (relative risk, 1.16; 95% CI, 1.08–1.24) (moderate
certainty), may increase the risk of neuromuscular weakness (rela-
tive risk, 1.21; 95% CI, 1.01–1.52) (low certainty), and appear to
have no other adverse effects (low or very low certainty). Subgroup
analysis did not demonstrate a credible subgroup effect on any of
the outcomes of interest (p > 0.05 for all).
Conclusions: In critically ill patients with sepsis, corticosteroids
possibly result in a small reduction in mortality while also possibly
increasing the risk of neuromuscular weakness. (Crit Care Med
2018; 46:1411–1420)
Key Words: corticosteroids; immune response; meta-analysis;
sepsis; systematic review
S
epsis—life-threatening organ dysfunction caused by
a dysregulated host immune response to infection
(1)—can result in systemic hypoperfusion and end-
organ dysfunction, and is associated with frequent morbidity
and mortality in the ICU (2). The mainstays of treatment are
early antibiotics and restoration of perfusion (IV fluids and
vasopressor therapy) (3); despite early and aggressive therapy,
mortality in septic shock remains as high as 30–40% (4).
Because the pathology of sepsis is characterized by a maladap-
tive host response to infection, experts have hypothesized that
immunomodulatory therapies may provide effective treatment
(5). Exogenous glucocorticoids are an attractive candidate given
their widespread availability, relatively low cost, and their demon-
strated ability to dampen the inflammatory cascade (5). In combi-
nation with mineralocorticoids, they may also address the relative
deficiency of cortisol that can occur in states of extreme stress (5).
Despite the physiologic rationale, clinical data examining the
effect of corticosteroids in sepsis have remained inconsistent,
leading to substantial variation in clinical practice worldwide. The
most recent Cochrane meta-analysis suggested that corticoste-
roids may reduce 28-day mortality in sepsis (relative risk [RR],
0.87; 95% CI, 0.76–1.00), although there was only low certainty
in the evidence, limited by imprecision, inconsistency, and the
potential for publication bias (6). Results from this review sug-
gested that patients with septic shock and those treated with a low
dose and a long course of corticosteroids had the highest likeli-
hood of benefit. Another review showed a similar point estimate,
but wider CI, and concluded no effect of corticosteroids in those
with sepsis (RR, 0.87; trial sequential analysis adjusted 95% CI,
0.74–1.08) (7). Since these reviews, a number of large randomized
controlled trials (RCTs) addressing the topic have been published
(8, 9), establishing the need for an updated systematic review.
This systematic review is part of the Rapid Recommendations
project, a collaborative effort from the Making GRADE the
Irresistible Choice (MAGIC) research and innovation pro-
gram (www.magicproject.org) (10). The aim of the project is
to respond to new potentially practice-changing evidence and
provide trustworthy practice guidelines in a timely manner. In
light of the two recently published large RCTs of 5,042 criti-
cally ill patients with sepsis (11, 12), we updated the previous
systematic review and meta-analysis (6) in order to inform a
1412 www.ccmjournal.org September 2018 • Volume 46 • Number 9
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
parallel clinical practice guideline (http://www.magicapp.org/
public/guideline/EZ1w8n).
METHODS
The protocol for this systematic review was registered on
PROSPERO (CRD42017058537) and published in full (13).
Substantive deviations from the published protocol are high-
lighted with an accompanying explanation.
Rapid Recommendation Guideline Panel
According to the Rapid Recommendations process, the guide-
line panel provided critical oversight and identified populations,
subgroups, and outcomes of interest for this review (14). The
panel included content experts, methodologists, and patients
or caregivers of patients with lived experience of sepsis. Patients
received personal training and support to optimize contributions
throughout the guideline development process. The patients on
the panel led the interpretation of the results based on what they
expected the typical patient values and preferences to be, as well
as the variation between patients. The four patients who were full
members of the guideline panel contributed to the selection and
prioritization of outcomes, values and preferences assessments,
and critical feedback to the protocol for the systematic review.
Data Sources and Searches
We performed a comprehensive search of MEDLINE, EMBASE,
CENTRAL, and LILACS for RCTs from December 1, 2014, to
January 10, 2018. As this was an updated systematic review, to
ensure adequate overlap, we searched from 3 months before the
last review (6) (supplemental electronic material 1, Supple-
mental Digital Content 1, http://links.lww.com/CCM/D757,
presents the MEDLINE search strategy). Keyword search terms
included corticosteroids, sepsis, and septic shock in a search that
did not apply any language restrictions. We searched conference
proceedings and abstracts from the last 3 years (2014–2017)
from the following societal meetings: The Society of Critical
Care Medicine Congress, The American Thoracic Society, and
The European Society of Intensive Care Medicine.
Study Selection
We performed all screening in duplicate with disagreements
resolved by discussion and third party adjudication as required.
After implementation of the search strategy, reviewers worked
in pairs to screen all potentially relevant citations and references.
Reviewers performed screening in two stages, initially assessing
titles and abstracts, and then full articles for those possibly eligible.
We captured reasons for exclusion at the full article review stage.
We included all RCTs comparing the use of corticosteroids
(including but not limited to hydrocortisone, methylpredniso-
lone, betamethasone, fludrocortisone, and dexamethasone)
with a corticosteroid-free comparator group in critically ill
patients with sepsis, excluding case reports, case series, and
observational studies. The population of interest included all
adults and children, excluding neonates, diagnosed with sep-
sis, severe sepsis, or septic shock, according to accepted criteria
(15–17). We included data from trials enrolling patients with

acute respiratory distress syndrome (ARDS) as long as the data
from patients with sepsis were reported separately.
We included the following outcomes: short-term mortality up to
31 days, long-term mortality (60 d to 1 yr), shock reversal at day 7
(defined as stable hemodynamic status for > 24 hr after withdrawal
of vasopressors), organ dysfunction at day 7 (using total Sequential
Organ Failure Assessment [SOFA] score [18]), ICU length of stay,
hospital length of stay, quality of life at 1 year using a validated index,
stroke, myocardial infarction, and adverse events including neu-
romuscular weakness, gastrointestinal bleeding, neuropsychiatric
effects (including delirium, confusion, mania, and others), hyperna-
tremia, superinfection, and hyperglycemia. Our published protocol
further separated short-term mortality into two separate endpoints,
28–31 days and 90 days. Given that all studies that reported 28- to
31-day mortality also reported 90-day mortality, we deviated slightly
from our protocol and included both endpoints from these studies
separately in our analysis. The short-term mortality from these stud-
ies was grouped with the other studies that also reported this 28- to
31-day endpoint, whereas the longer 90-day mortality data were
grouped with the 1-year mortality data. To assess for any difference
in long-term mortality between studies reporting at 60–180 days and
those that reported at 1 year, we performed subgroup analysis.
Data Extraction and Quality Assessment
Reviewers performed data extraction independently and in
duplicate using predefined data abstraction forms. A third
reviewer resolved disagreements. Abstracted data included study
title, first author, demographic data, details of the intervention
and control, primary and secondary outcome data, and risk of
bias (RoB) for each study. We performed our own data abstrac-
tion only for studies not included in the prior review (6), and for
outcomes or subgroups that were not previously reported.
RoB was assessed, independently and in duplicate, for each
outcome of individual studies using a modified Cochrane RoB
tool (19) that classifies RoB as “low,” “probably low,” “probably
high,” or “high” for each of the following domains: sequence
generation, allocation sequence concealment, blinding, selec-
tive outcome reporting, and other bias. We rated the overall
RoB as the highest risk attributed to any criterion. We assessed
the overall certainty of evidence for each outcome using the
Grading of Recommendations Assessment, Development and
Evaluation (GRADE) framework (20). Disagreements for RoB
and GRADE assessments were resolved by discussion. GRADE
assessments were also agreed upon with the linked British
Medical Journal Rapid Recommendation guideline panel.
Data Analysis
We used DerSimonian and Laird random-effects models in
Rev-Man 5.3 (The Cochrane Collaboration, Copenhagen, Den-
mark) to conduct the meta-analyses. Study weights were gen-
erated using the inverse of the variance. We present results as
RRs for dichotomous outcomes and as mean difference (MD)
for continuous outcomes, both with 95% CI. We examined the
potential for small study effects using a funnel plot if evidence
included more than 10 trials per outcome, using the Egger test
to detect funnel plot asymmetry for continuous outcomes (21)
Critical Care Medicine www.ccmjournal.org 1413
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Feature Articles
and the arcsine test for dichotomous outcomes. Funnel plots
and small study effect testing were performed using Stata 15
(StataCorp, College Station, TX).
We assessed for heterogeneity between studies using the chi-
square test for homogeneity, the I2 statistic, and visual inspection
of the forest plots. We considered the magnitude and direction of
heterogeneity when considering whether to rate down our cer-
tainty in the evidence for inconsistency. We explored prespecified
subgroups with random-effects metaregression using Stata 15 and
present interaction p values. Subgroups assessed included the fol-
lowing: high RoB versus low RoB hypothesizing that corticosteroids
would be more effective in trials with high RoB; patient subpopu-
lation (sepsis without shock vs septic shock, respiratory [ARDS/
community-acquired pneumonia (CAP)] vs no respiratory illness)
hypothesizing that corticosteroids would be more effective in those
with septic shock and pneumonia/ARDS; dose and duration of
corticosteroids (high dose/short course vs low dose/long course)
hypothesizing that corticosteroids would be more effective with
lower doses and when used for a longer duration; and corticoste-
roid molecule (hydrocortisone vs hydrocortisone plus fludrocor-
tisone vs methylprednisolone or prednisolone or prednisone vs
dexamethasone) hypothesizing that corticosteroids would be more
effective using agents with more mineralocorticoid effect. The pro-
tocol (13) included these subgroup analyses and a plan for assess-
ment for credible subgroup effect (22).
We performed two additional metaregression subgroup analyses
that were not predefined but were requested by the linked Rapid
Recommendations panel: impact of the mortality rate in the control
group and year of study publication (as a continuous variable) on
the effect of corticosteroids on mortality. The panel hypothesized
that corticosteroids would be more effective in trials with a higher
mortality rate in the control group and in older studies. Although
we planned for subgroup analysis based on age of patients (adults
vs children) and presence of critical illness–related corticosteroid
insufficiency, there was insufficient evidence examining these sub-
groups for metaregression. We also performed three additional
sensitivity analyses at the request of peer reviewers: we examined
the effect of corticosteroids on short-term mortality in those only
treated with long-course and low-dose corticosteroids, the effect of
short-term mortality only in the last two published RCTs (11, 12),
and the effect of corticosteroids in only studies at low or probably
low RoB. Absolute effects were calculated using the pooled baseline
prevalence from the control arm of included studies.
RESULTS
Description of Eligible Studies
The 2015 Cochrane review contributed 30 eligible studies
(n = 3,623) (23–52). Of the 7,202 citations identified in our
updated search, we reviewed the full text of 33, of which 12
new RCTs ultimately proved eligible (7–9, 11, 53–60) for a total
of 42 RCTs (Fig. 1).
Supplementary Table 1 (Supplemental Digital Content 2,
http://links.lww.com/CCM/D758) presents a detailed descrip-
tion of eligible trials, 20 of which were multicenter (7–9, 11,
23, 24, 26, 27, 31–33, 37, 38, 42, 44, 46, 49, 54, 57) and the
Rochwerg et al

of corticosteroid administered
varied, although most (n = 37;
9,283 patients) used what we
defined as low dose (< 400 mg/d
of hydrocortisone or equiva-
lent) and used them over a short
course of therapy (< 3 d) (7–9,
11, 24–26, 28–35, 37–42, 45, 46,
48–53, 55–60).
Supplemental material 2
(Supplemental Digital Content
1, http://links.lww.com/CCM/
D757) presents individual study
RoB. Approximately half of the
included studies were judged to
be at low (n = 10) (7–9, 11, 24,
38, 46, 49, 57, 59) or probably
low RoB (n = 11) (23, 25, 27, 28,
32, 35, 37, 45, 51, 52, 60).

Outcomes
Table 1 shows summary of
findings for all outcomes,
based on the meta-analysis of
identified trials, and includes
the certainty of the evidence
and the reasons for rating
down certainty. Interactive
tables summarizing the find-
ings are available at http://www.
magicapp.org/public/guideline/
EZ1w8n. Figures 2 and 3 and
supplemental material 3–18
(Supplemental Digital Content
1, http://links.lww.com/CCM/
D757) present forest plots,
metaregression results, and
funnel plots for all outcomes
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram demonstrating including sensitivity analysis.
study selection for inclusion in the quantitative meta-analysis. RCT = randomized controlled trial.
The pooled point estimates
suggest that corticosteroids
remainder single center. Twenty-four of the RCTs examined may result in a small reduction or no reduction in short-term
patients with septic shock (8, 9, 11, 24–26, 28–30, 32–34, 39, 43, mortality (RR, 0.93; 95% CI, 0.84–1.03; 1.8% absolute risk
46–48, 50, 53–58); five included patients with both CAP and reduction; 95% CI, 4.1% reduction to 0.8% increase, low cer-
sepsis (28, 35, 39, 42, 46) and four enrolled patients with the tainty) (Fig. 2 and Table 1) and possibly a small reduction in
ARDS and sepsis (32, 34, 37, 57). Three of the included studies long-term mortality (RR, 0.94; 95% CI, 0.89–1.00; 2.2% abso-
enrolled only children (41, 47, 55) and one enrolled both adult lute risk reduction; 95% CI, 4.1% reduction to no effect, mod-
and children, but reported the two groups separately (33). erate certainty) (Fig. 3 and Table 1).
The majority of studies used hydrocortisone (n = 27; 6,922 Pooled estimates suggest that corticosteroids probably result
patients) (7–9, 11, 25, 26, 28, 29, 31–35, 39, 41, 42, 47, 48, 50, 53– in small reductions in ICU length of stay (MD, –0.73 d; 95%
60); others used methylprednisolone (n = 6; 859 patients) (23, CI, –1.78 to 0.31; 12.40 vs 13.13 d, moderate certainty) and
27, 36, 37, 40, 49), prednisolone (n = 3; 308 patients) (45, 51, 52), hospital length of stay (MD, –0.73 d; 95% CI, –2.06 to 0.60;
or dexamethasone (n = 3; 333 patients) (30, 38, 44). Two stud- 31.30 vs 32.03 d, moderate certainty). Patients randomized to
ies used hydrocortisone in addition to fludrocortisone (1,541 receive corticosteroids had higher rates of shock reversal at day
patients) (11, 24), and two studies included both dexamethasone 7 (RR, 1.26; 95% CI, 1.12–1.42; 12.1% absolute risk increase;
and methylprednisolone groups (231 patients) (43, 47). The dose 95% CI, 5.6–19.5% increase, high certainty) and lower SOFA

1414 www.ccmjournal.org September 2018 • Volume 46 • Number 9

Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Summary of Findings Table Including Relative Effect Measure, Absolute Effect
TABLE 1.
Measure, Certainty of Evidence, and Narrative Plain Text Summary
Outcome Study Results
Timeframe and Measurements
Mortality Relative risk: 0.94
longer term (95% CI, 0.89–1.0)
(60 d to 1 Based on data from
yr) 6,438 patients in
nine studies
Mortality Relative risk: 0.93
short term (95% CI, 0.84–1.03)
(28–31 d) Based on data from
9,433 patients in 36
studies
Shock reversal Relative risk: 1.26
(1 wk) (95% CI, 1.12–1.42)
Based on data from
2,802 patients in 13
studies
Neuromuscular Relative risk: 1.21
weakness (95% CI, 1.01–1.45)
Based on data from
6,178 patients in
seven studies
Gastrointestinal Relative risk: 1.09
bleeding (95% CI, 0.86–1.38)
Based on data from 4,243
patients in 17 studies
Neuropsychiatric Relative risk: 0.58
events (95% CI, 0.33–1.03)
Based on data from
1,004 patients in five
studies
Hypernatremia Relative risk: 1.64
(95% CI, 1.32–2.03)
Based on data from 5,015
patients in six studies
Superinfection Relative risk: 1.02
(95% CI, 0.89–1.18)
Based on data from 4,519
patients in 21 studies
Stroke Relative risk: 2.07
(95% CI, 0.45–9.61)
Based on data from
1,105 patients in three
studies
Myocardial Relative risk: 0.91
infarction (95% CI, 0.45–1.82)
Based on data from 1,080
patients in three studies
Hyperglycemia Relative risk: 1.16
(95% CI, 1.08–1.24)
Based on data from 7,563
patients in 15 studies
Critical Care Medicine www.ccmjournal.org 1415
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Absolute Effect Estimates
No
Corticosteroids Corticosteroids
371/1,000 349/1,000
Difference: 22 fewer per 1,000
(95% CI, 41 fewer to 0 fewer)
254/1,000 236/1,000
Difference: 18 fewer per 1,000
(95% CI, 41 fewer to 8 more)
464/1,000 585/1,000
Difference: 121 more per 1,000
(95% CI, 56 more to 195 more)
250/1,000 303/1,000
Difference: 53 more per 1,000
(95% CI, 3 more to 130 more)
35/1,000 38/1,000
Difference: 3 more per 1,000
(95% CI, 5 fewer to 13 more)
59/1,000 34/1,000
Difference: 25 fewer per 1,000
(95% CI, 40 fewer to 2 more)
36/1,000 59/1,000
Difference: 23 more per 1,000
(95% CI, 12 more to 37 more)
161/1,000 164/1,000
Difference: 3 more per 1,000
(95% CI, 18 fewer to 29 more)
5/1,000 10/1,000
Difference: 5 more per 1,000
(95% CI, 3 fewer to 43 more)
30/1,000 27/1,000
Difference: 3 fewer per 1,000
(95% CI, 16 fewer to 25 more)
181/1,000 210/1,000
Difference: 29 more per 1,000
(95% CI, 14 more to 43 more)
Feature Articles
Certainty in Effect
Estimates (Quality Plain Text
of Evidence) Summary
Moderate Corticosteroids
possibly achieve a
Due to serious small reduction in
imprecisiona long-term mortality.
Low Corticosteroids may
Due to serious achieve a small
imprecision, border- or no reduction in
line inconsistency, short-term mortality.
and publication
biasb
High Corticosteroids
increase shock
reversal in the first
week.
Low Corticosteroids may
Due to serious impre- result in a small
cision and indirect- increase in neuro-
ness and borderline muscular weakness.
inconsistencyc
Low Corticosteroids may
Due to serious indi- have little or no dif-
rectness and ference on gastroin-
imprecisiond testinal bleeding
Low Corticosteroids may
Due to serious impre- achieve a small
cision and serious reduction in neu-
indirectnesse ropsychiatric events.
Moderate Corticosteroids proba-
Due to serious bly increase the risk
indirectnessf of hypernatremia.
Low Corticosteroids may
Due to serious impre- have little or no
cision and serious impact on superin-
indirectnessc fection.
Very low Whether or not corti-
Due to serious indi- costeroids impact
rectness and very the risk of stroke is
serious imprecisiong uncertain.
Very low Whether or not corticos-
Due to serious indi- teroids impact the risk
rectness and very of myocardial infarc-
serious imprecisiong tion is uncertain.
Moderate Corticosteroids prob-
Due to serious indi- ably increase the
rectnessh incidence of
hyperglycemia.
(Continued )
Rochwerg et al

(Continued). Summary of Findings Table Including Relative Effect

TABLE 1.
Measure, Certainty of Evidence, and Narrative Plain Text Summary
Absolute Effect Estimates
Certainty in Effect
Outcome Study Results No Estimates (Quality Plain Text
Timeframe and Measurements Corticosteroids Corticosteroids of Evidence) Summary

ICU length of Measured by: days 13.13 d 12.4 d Moderate Corticosteroids prob-
stay Based on data from 7,463 Due to serious impre- ably achieve a small
patients in 20 studies MD: 0.73 fewer (95% CI, 1.78 cisioni reduction in the
fewer to 0.31 more) duration of initial
ICU stay.
Hospital length Measured by: days 32.03 d 31.3 d Moderate Corticosteroids prob-
of stay Based on data from Due to serious impre- ably achieve a small
7,706 patients in MD: 0.73 fewer (95% CI, 2.06 cisionj reduction in the
18 studies fewer to 0.6 more) duration of hospitali-
zation.
Organ dysfunc- Measured by: Sepsis Organ 7.61 points 6.22 points High Corticosteroids
tion (1 wk) Failure Assessment decrease organ
score MD: 1.39 lower (95% CI, 1.88 dysfunction at 7 d.
Scale: 0–24 lower better lower to 0.89 lower)
Based on data from 1,986
patients in nine studies
Quality of life Not reported in any of the included studies
MD = mean difference.
a
Note that all studies that reported 90-d to 1-yr mortality also reported 28- to 30-day mortality: the 95% CI, around the pooled effect for short-term mortality
also excludes no effect: relative risk, 0.91 (95% CI, 0.84–0.98).
b
There is borderline inconsistency (I2 = 37%). There was also borderline publication bias (some visual asymmetry in the funnel plot), but statistical tests (Egger
test and the Trim and Fill method) did not detect statistically significant small study effects (p = 0.12).
c
Variable time of assessment. Variable assessment tool used. Varying magnitude. All decrease our certainty in this effect estimate. CI includes no difference.
d
Variable definition of gastrointestinal bleeding. Varying magnitude/clinical significance of gastrointestinal bleeding. All decrease our certainty in this effect
estimate. CI includes no difference.
e
Variable time of assessment. Variable assessment tool used. Varying magnitude. All decrease our certainty in this effect estimate.
Variable time of assessment. Variable assessment tool used. Varying magnitude. All decrease our certainty in this effect estimate. CI includes important harm.
f

g
Unclear how this outcome was abstracted or ascertained. Unclear clinical impact of these stroke or myocardial infarction events. Very few events and only two
studies.
h
Unclear how this outcome was abstracted or ascertained. Unclear clinical impact of these stroke or MI events. Very few events and only one study.
Variable time of assessment. Variable assessment tool used. Varying magnitude. All decrease our certainty in this effect estimate.
i

CI includes no difference.
j

scores at day 7 (MD, –1.39 points; 95% CI, –1.88 to –0.89; 6.22
vs 7.61 points, high certainty). None of the studies reported
quality of life outcomes.
Adverse Effects
Corticosteroids likely increase the rates of hypernatremia
(RR, 1.64; 95% CI, 1.32–2.03; 2.3% absolute risk increase; 95%
CI, 1.2–3.7% increase, moderate certainty) and hyperglyce-
mia (RR, 1.16; 95% CI, 1.08–1.24; 2.9% absolute risk increase;
95% CI, 1.4–4.3% increase, moderate certainty). Corticoste-
roids may increase the risk of neuromuscular weakness (RR,
1.21; 95% CI, 1.01–1.52; 5.3% absolute risk increase; 95% CI,
0.3–13.0% increase, low certainty). Results did not demon-
strate a convincing effect on other adverse effects including
gastrointestinal bleeding (RR, 1.09; 95% CI, 0.86–1.38; 0.3%
absolute risk increase; 95% CI, 0.5% decrease to 1.3% increase,
low certainty), superinfection (RR, 1.02; 95% CI, 0.89–1.18;
0.3% absolute risk increase; 95% CI, 1.8% decrease to 2.9%
increase, low certainty), and neuropsychiatric outcomes (RR,
0.58; 95% CI, 0.33–1.03; 2.5% absolute risk reduction; 95% CI,
4.0% fewer to 0.2% more, low certainty). We are very uncertain
about the effect of steroids on stroke (RR, 2.07; 95% CI, 0.45–
9.61; 0.5% absolute risk increase; 95% CI, 0.3–4.3% increase,
very low certainty) or myocardial infarction (RR, 0.91; 95%
CI, 0.45–1.82; 0.3% absolute risk reduction; 95% CI, 1.6%
decrease to 2.5% increase; very low certainty) (Table 1).
Subgroup Analyses
Subgroup analyses did not demonstrate a credible effect
for any of the predefined subgroups and on any of the out-
comes of interest (p > 0.05 for all outcomes) (Supplemen-
tary Material 19, Supplemental Digital Content 1, http://
links.lww.com/CCM/D757). Metaregression examining the
effect of control group mortality on short-term mortality
did not demonstrate a credible subgroup effect (p = 0.26).
There was also no effect based on year of study publication
(p = 0.89). Although we had planned for subgroup analysis
examining the effect of corticosteroids on adults when com-
pared with children, insufficient data precluded this analysis
for any of the outcomes of interest.

1416 www.ccmjournal.org September 2018 • Volume 46 • Number 9

Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
 Feature Articles

Figure 2. Forest plot for short-term mortality. df = degrees of freedom, M-H = Mantel-Haenszel.

DISCUSSION
Our systematic review demonstrates that the use of cortico-
steroids in sepsis may result in a small absolute reduction in
mortality of approximately 2%. In this most comprehensive
review to date, including several new large trials, the precision
of the pooled point effect estimates has increased substantially
(6, 7, 61). The CI around the pooled effect on mortality over
the short, but not the long term, includes a small risk of harm
(moderate quality for the latter because the upper boundary of
the CI is 1.0) (Table 1).
Subgroup analyses failed to identify credible effect modifi-
cation in patient population, study RoB, control group mor-
tality (severity of illness), corticosteroid dose, duration, or
molecule, or year of publication. Most of the evidence comes
from studies that used hydrocortisone with or without fludro-
cortisone over a long course (> 2 d) and at a low dose (under
400 mg/d of hydrocortisone or equivalent). Results showed
that it is likely that corticosteroids result in small reductions
in ICU and hospital length of stay (Table 1). Our results fail to
demonstrate an impact of corticosteroids on gastrointestinal
bleeding, superinfection, neuropsychiatric effects, stroke, or
myocardial infarction (Table 1). Based on low certainty evi-
dence, corticosteroids may result in a small increase in the risk
of neuromuscular weakness. Corticosteroids did increase the
risk of hyperglycemia and hypernatremia although the stud-
ies failed to provide information regarding whether these find-
ings required intervention or resulted in patient-important
sequelae. Ascertainment of these adverse outcomes in the
included studies was also variable, further contributing to the
lower certainty in evidence.
Strengths of this review include a comprehensive literature
search including unpublished sources, a published protocol
to which we adhered (13), application of GRADE methodol-
ogy to assess certainty in pooled estimates of effect, a priori
specification of possible effect modifiers including direction
of effect and subsequent metaregression to explore the effect

Critical Care Medicine www.ccmjournal.org 1417

Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Rochwerg et al
Figure 3. Forest plot for long-term mortality. df = degrees of freedom.
modification, and specification of both relative and absolute
effects. Content experts as well as methodologists and patients
or caregivers with personal experience with sepsis as part of the
Rapid Recommendation process provided critical input to our
protocol and assessment of the synthesized evidence.
Limitations include significant clinical heterogeneity over
studies conducted over a period of 60 years. Our exploration of
subgroup hypotheses, including the era in which the studies were
conducted, and the failure to identify any effect modification,
markedly diminishes this concern. All included studies enrolled
patients based on previous sepsis diagnostic criteria, although
we have no reason to believe that using the new Sepsis-3 criteria
would change the efficacy or harms of corticosteroids (16).
Although the demonstrated mortality effect may be impor-
tant, the certainty is limited by imprecision. Our review dem-
onstrates that the use of corticosteroids in septic shock is
associated with faster shock reversal and less organ dysfunc-
tion at day 7 when compared with those that do not receive
corticosteroids. Both outcomes are best viewed as surrogates
for mortality and, on the basis of evidence that persistent
organ dysfunction and shock may be associated with worse
long-term quality of life, for quality of life. Despite previous
suggestions that sicker patients might derive greater benefit
from corticosteroids in sepsis, we found no support for this
hypothesis in our subgroup analyses examining patients with
septic shock and study control group mortality rate. Harms
associated with the long-term use of corticosteroids (62) are
well documented; our review suggests, however, that adverse
effects may be minimal with the short-term use of corticoste-
roids in critically ill patients with sepsis.
The results of this meta-analysis reflect results of the recent
large RCTs. The Activated Protein C and Corticosteroids for
Human Septic Shock (APROCCHSS) trial enrolled 1,241
patients with septic shock from 34 centers in France and
1418 www.ccmjournal.org September 2018 • Volume 46 • Number 9
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
demonstrated a small reduction in 90-day mortality with
corticosteroids (RR, 0.88; 95% CI, 0.78–0.99) (11). The
Adjunctive Glucocorticoid Therapy in Patients with Septic
Shock (ADRENAL) trial enrolled 3,800 patients with septic
shock from 69 medical-surgical ICUs, and there was no effect
on 90-day mortality (odds ratio, 0.95; 95% CI, 0.82–1.10). The
results of both trials are consistent with our pooled estimates
of effect and the associated CIs (9).
Based on the results of this review, our best estimates suggest
a small absolute reduction in mortality with corticosteroids in
sepsis based on low-to-moderate certainty evidence. This abso-
lute reduction in death will be highest in the sickest patients
(e.g., those with septic shock); however, the relative effects of
corticosteroids appear consistent across all studied subgroups.
Previous clinical practice guidelines, before the publication of
ADRENAL and APROCCHSS, have made conditional recom-
mendations against corticosteroids in sepsis and conditional
recommendations for corticosteroids in the setting of septic
shock (3, 7). The results of this systematic review and meta-
analysis will inform a Rapid Recommendation (19) exam-
ining the role of corticosteroids in patients with sepsis and
septic shock.
CONCLUSIONS
The additional recent evidence from large trials has established
that if steroids do have on effect on mortality, that effect is
small. The most compelling evidence is of a 6% relative reduc-
tion in long-term mortality, translating to a 2% absolute risk
reduction, but the CI around the estimate includes no effect.
Point estimates suggest a reduction in length of stay in hospital
and ICU, but these effects are also small—less than a day—and
CIs include a possible increase in length of stay. Harms, how-
ever, are minimal, and informed individuals may as a result
conclude that net benefit is likely.

ACKNOWLEDGMENTS
We would like to express our gratitude to Jean Maragno and
Lois Cottrel for their guidance in designing and carrying out
our search strategy.
REFERENCES
1. Seymour CW, Liu VX, Iwashyna TJ, et al: Assessment of clinical cri-
teria for sepsis: For the Third International Consensus Definitions for
Sepsis and Septic Shock (Sepsis-3). JAMA 2016; 315:762–774
2. Rhee C, Dantes R, Epstein L, et al; CDC Prevention Epicenter
Program: Incidence and trends of sepsis in US hospitals using clini-
cal vs claims data, 2009-2014. JAMA 2017; 318:1241–1249
3. Rhodes A, Evans LE, Alhazzani W, et al: Surviving sepsis campaign:
International Guidelines for Management of Sepsis and Septic Shock:
2016. Intensive Care Med 2017; 43:304–377
4. Shankar-Hari M, Phillips GS, Levy ML, et al; Sepsis Definitions Task
Force: Developing a new definition and assessing new clinical criteria
for septic shock: For the Third International Consensus Definitions for
Sepsis and Septic Shock (Sepsis-3). JAMA 2016; 315:775–787
5. Annane D, Pastores SM, Arlt W, et al: Critical illness-related cortico-
steroid insufficiency (CIRCI): A narrative review from a Multispecialty
Task Force of the Society of Critical Care Medicine (SCCM) and the
European Society of Intensive Care Medicine (ESICM). Intensive
Care Med 2017; 43:1781–1792
6. Annane D, Bellissant E, Bollaert PE, et al: Corticosteroids for treating
sepsis. Cochrane Database Syst Rev 2015; (12):CD002243
7. Volbeda M, Wetterslev J, Gluud C, et al: Glucocorticosteroids for sep-
sis: Systematic review with meta-analysis and trial sequential analysis.
Intensive Care Med 2015; 41:1220–1234
8. Keh D, Trips E, Marx G, et al; SepNet–Critical Care Trials Group:
Effect of hydrocortisone on development of shock among patients
with severe sepsis: The HYPRESS randomized clinical trial. JAMA
2016; 316:1775–1785
9. Gordon AC, Mason AJ, Thirunavukkarasu N, et al; VANISH
Investigators: Effect of early vasopressin vs norepinephrine on kidney
failure in patients with septic shock: The VANISH randomized clinical
trial. JAMA 2016; 316:509–518
10. Siemieniuk RA, Agoritsas T, Macdonald H, et al: Introduction to BMJ
Rapid Recommendations. BMJ 2016; 354:i5191
11. Venkatesh B, Finfer S, Cohen J, et al: Adjunctive glucocorticoid
therapy in patients with septic shock. N Engl J Med 2018; 378:
797–808
12. Annane D, Renault A, Brun-Buisson C, et al; CRICS-TRIGGERSEP
Network: Hydrocortisone plus fludrocortisone for adults with septic
shock. N Engl J Med 2018; 378:809–818
13. Rochwerg B, Oczkowski S, Siemieniuk RA, et al: Corticosteroids in
sepsis: An updated systematic review and meta-analysis (protocol).
BMJ Open 2017; 7:e016847
14. Guyatt GH, Oxman AD, Kunz R, et al: GRADE guidelines: 2. Framing
the question and deciding on important outcomes. J Clin Epidemiol
2011; 64:395–400
15. Levy MM, Fink MP, Marshall JC, et al: 2001 SCCM/ESICM/ACCP/
ATS/SIS International Sepsis Definitions Conference. Critical Care
Med 2003; 31:1250–1256
16. Singer M, Deutschman CS, Seymour CW, et al: The Third International
Consensus Definitions for Sepsis and Septic Shock (Sepsis-3).
JAMA 2016; 315:801–810
17. Bone RC, Balk RA, Cerra FB, et al: Definitions for sepsis and organ
failure and guidelines for the use of innovative therapies in sepsis.
The ACCP/SCCM Consensus Conference Committee. American
College of Chest Physicians/Society of Critical Care Medicine. Chest
1992; 101:1644–1655
18. Minne L, Abu-Hanna A, de Jonge E: Evaluation of SOFA-based mod-
els for predicting mortality in the ICU: A systematic review. Crit Care
2008; 12:R161
19. Guyatt GH, Busse J: Evidence Partners. 2013. Available at: http://
distillercer.com/resources/methodological-resources/risk-of-bias-
commentary/. Accessed September 6, 2013
Critical Care Medicine www.ccmjournal.org 1419
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Feature Articles
20. Guyatt GH, Oxman AD, Vist GE, et al: GRADE: An emerging consen-
sus on rating quality of evidence and strength of recommendations.
BMJ 2008; 336:924–926
21. Higgins JP, Altman DG, Gotzsche PC, et al: The Cochrane
Collaboration’s tool for assessing risk of bias in randomised trials.
BMJ 2011; 343:d5928
22. Sun X, Briel M, Busse JW, et al: Credibility of claims of subgroup
effects in randomised controlled trials: Systematic review. BMJ 2012;
344:e1553
23. The Veterans Administration Systemic Sepsis Cooperative Study Group:
Effect of high-dose glucocorticoid therapy on mortality in patients with
clinical signs of systemic sepsis. N Eng J Med 1987; 317:659–665
24. Annane D, Sébille V, Charpentier C, et al: Effect of treatment with low
doses of hydrocortisone and fludrocortisone on mortality in patients
with septic shock. JAMA 2002; 288:862–871
25. Arabi YM, Aljumah A, Dabbagh O, et al: Low-dose hydrocortisone
in patients with cirrhosis and septic shock: A randomized controlled
trial. CMAJ 2010; 182:1971–1977
26. Bollaert PE, Charpentier C, Levy B, et al: Reversal of late septic shock
with supraphysiologic doses of hydrocortisone. Crit Care Med 1998;
26:645–650
27. Bone RC, Fisher CJ Jr, Clemmer TP, et al: A controlled clinical trial of
high-dose methylprednisolone in the treatment of severe sepsis and
septic shock. N Engl J Med 1987; 317:653–658
28. Briegel J, Forst H, Haller M, et al: Stress doses of hydrocortisone
reverse hyperdynamic septic shock: A prospective, randomized,
double-blind, single-center study. Crit Care Med 1999; 27:723–732
29. Chawla K, Kupfer Y, Goldman I, et al: Hydrocortisone reverses refrac-
tory septic shock. Crit Care Med 1999; 27:Supplement page 33A
30. Cicarelli DD, Vieira JE, Benseñor FE: Early dexamethasone treatment
for septic shock patients: A prospective randomized clinical trial. Sao
Paulo Med J 2007; 125:237–241
31. Confalonieri M, Urbino R, Potena A, et al: Hydrocortisone infusion
for severe community-acquired pneumonia: A preliminary randomized
study. Am J Respir Crit Care Med 2005; 171:242–248
32. Gordon AC, Mason AJ, Perkins GD, et al: The interaction of vasopres-
sin and corticosteroids in septic shock: A pilot randomized controlled
trial. Crit Care Med 2014; 42:1325–1333
33. Group CS: The effectiveness of hydrocortisone in the management of
severe infections. JAMA 1963; 183:462–465
34. Hu B, Li JG, Liang H, et al: The effect of low-dose hydrocortisone
on requirement of norepinephrine and lactate clearance in patients
with refractory septic shock. Zhongguo Wei Zhong Bing Ji Jiu Yi Xue
2009; 21:529–531
35. Liu L, Li J, Huang YZ, et al: The effect of stress dose glucocorticoid
on patients with acute respiratory distress syndrome combined with
critical illness-related corticosteroid insufficiency. Zhonghua Nei Ke
Za Zhi 2012; 51:599–603
36. Luce JM, Montgomery AB, Marks JD, et al: Ineffectiveness of high-
dose methylprednisolone in preventing parenchymal lung injury and
improving mortality in patients with septic shock. Am Rev Respir Dis
1988; 138:62–68
37. Meduri GU, Golden E, Freire AX, et al: Methylprednisolone infusion
in early severe ARDS: Results of a randomized controlled trial. Chest
2007; 131:954–963
38. Meijvis SC, Hardeman H, Remmelts HH, et al: Dexamethasone and
length of hospital stay in patients with community-acquired pneu-
monia: A randomised, double-blind, placebo-controlled trial. Lancet
2011; 377:2023–2030
39. Oppert M, Schindler R, Husung C, et al: Low-dose hydrocortisone
improves shock reversal and reduces cytokine levels in early hyperdy-
namic septic shock. Crit Care Med 2005; 33:2457–2464
40. Rezk N, Ibrahim A: Effects of methylprednisolone in early ARDS.
Egypt J Chest Dis Tuberc 2013; 62:167–172
41. Rinaldi S, Adembri C, Grechi S, et al: Low-dose hydrocortisone dur-
ing severe sepsis: Effects on microalbuminuria. Crit Care Med 2006;
34:2334–2339
42. Sabry N, Omar EED: Corticosteroids and ICU course of community
acquired pneumonia in Egyptian settings. Pharmacol Pharm 2011;
2:73–81
Rochwerg et al
43. Schumer W: Steroids in the treatment of clinical septic shock. Ann
Surg 1976; 184:333–341
44. Slusher T, Gbadero D, Howard C, et al: Randomized, placebo-con-
trolled, double blinded trial of dexamethasone in African children with
sepsis. Pediatr Infect Dis J 1996; 15:579–583
45. Snijders D, Daniels JM, de Graaff CS, et al: Efficacy of corticosteroids
in community-acquired pneumonia: A randomized double-blinded
clinical trial. Am J Respir Crit Care Med 2010; 181:975–982
46. Sprung CL, Annane D, Keh D, et al; CORTICUS Study Group:
Hydrocortisone therapy for patients with septic shock. N Engl J Med
2008; 358:111–124
47. Sprung CL, Caralis PV, Marcial EH, et al: The effects of high-dose
corticosteroids in patients with septic shock. A prospective, con-
trolled study. N Engl J Med 1984; 311:1137–1143
48. Tandan SM, Guleria R, Gupta N: Low dose steroids and adreno-
cortical insufficiency in septic shock: A double-blind randomized
controlled trial from India. In: Proceedings of the American Thoracic
Society Meeting, New York, NY, 2005:A24
49. Torres A, Sibila O, Ferrer M, et al: Effect of corticosteroids on treat-
ment failure among hospitalized patients with severe community-
acquired pneumonia and high inflammatory response: A randomized
clinical trial. JAMA 2015; 313:677–686
50. Valoor HT, Singhi S, Jayashree M: Low-dose hydrocortisone in pediat-
ric septic shock: An exploratory study in a third world setting. Pediatr
Crit Care Med 2009; 10:121–125
51. Yildiz O, Doganay M, Aygen B, et al: Physiological-dose steroid ther-
apy in sepsis [ISRCTN36253388]. Crit Care 2002; 6:251–259
52. Yildiz O, Tanriverdi F, Simsek S, et al: The effects of moderate-dose
steroid therapy in sepsis: A placebo-controlled, randomized study. J
Res Med Sci 2011; 16:1410–1421
1420 www.ccmjournal.org September 2018 • Volume 46 • Number 9
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
53. Branco RG, Amoretti CF, Garcia PCR, et al: Corticosteroid replace-
ment in sepsis induces lymphopenia: Results of a randomized con-
trolled trial. Pediatr Crit Care Med 2014; 15(4 Suppl):12
54. De Graaf H, Ramakrishnan KA, Pappachan J, et al: Evaluation of cortico-
steroid replacement therapy in children with severe septic shock - a ran-
domised intervention trial. Pediatr Crit Care Med 2014; 15(4 Suppl):141
55. El-Nawawy A, Khater D, Omar H, et al: Evaluation of early cortico-
steroid therapy in management of pediatric septic shock in pediatric
intensive care patients: A randomized clinical study. Pediatr Infect Dis
J 2017; 36:155–159
56. Lv QQ, Gu XH, Chen QH, et al: Early initiation of low-dose hydrocor-
tisone treatment for septic shock in adults: A randomized clinical trial.
Am J Emerg Med 2017; 35:1810–1814
57. Menon K, McNally D, O’Hearn K, et al; Canadian Critical Care Trials
Group: A randomized controlled trial of corticosteroids in pediatric septic
shock: A pilot feasibility study. Pediatr Crit Care Med 2017; 18:505–512
58. Mirea L, Ungureanu R, Pavelescu D, et al: Continuous administration
of corticosteroids in septic shock can reduce risk of hypernatremia.
Crit Care 2014; 18(Suppl 1):P239
59. Tongyoo S, Permpikul C, Mongkolpun W, et al: Hydrocortisone treat-
ment in early sepsis-associated acute respiratory distress syndrome:
Results of a randomized controlled trial. Crit Care 2016; 20:329
6 0. Meduri G, Golden E, Freire A, et al: Randomized clinical trial evalu-
ating the effects of low-dose prolonged hydrocortisone infusion on
resolution of MODS in severe sepsis. Chest 2009; 136(Suppl):154
61. Gibbison B, López-López JA, Higgins JP, et al: Corticosteroids in sep-
tic shock: A systematic review and network meta-analysis. Crit Care
2017; 21:78
62. Oray M, Abu Samra K, Ebrahimiadib N, et al: Long-term side effects
of glucocorticoids. Expert Opin Drug Saf 2016; 15:457–465