Cortisol levels and adrenal response in severe community-acquired pneumonia: a systematic review of the literature.
ABSTRACT Our aim was to review the literature on the prevalence and impact of critical-illness related corticosteroid insufficiency (CIRCI) on the outcomes of patients with severe community-acquired pneumonia (CAP).
We reviewed Cochrane, Medline, and CINAHL databases (through July 2008) to identify studies evaluating the adrenal function in severe CAP. Main data collected were prevalence of CIRCI and its mortality.
We screened 152 articles and identified 7 valid studies. Evaluation of adrenal function varied, and most studies used baseline total cortisol levels. The prevalence of CIRCI in severe CAP ranged from 0% to 48%. Among 533 patients, 56 (10.7%) had cortisol levels of 10 μg/dL or less and 121 patients (21.2%) had cortisol levels of 15 μg/dL or less. In a raw analysis, there was no significant difference in mortality when patients with cortisol levels less than 10 μg/dL (8.6 vs 15.5%; P = .55) or less than 15 μg/dL (12.4 vs 16%; P = .38) were compared with those with cortisol above these levels. In the meta-analysis, relative risk for mortality were 0.81 (confidence interval, 0.39-1.7; P = .59; χ(2) = 1.04) for cortisol levels less than 10 μg/dL and relative risk was 0.67 (confidence interval, 0.4-1.14; P = .84; χ(2) = 1.4) for cortisol levels less than 15 μg/dL.
A significant proportion of patients with severe CAP fulfilled criteria for CIRCI. However, CIRCI does not seem to affect the outcomes. Noteworthy, the presence of elevated cortisol levels is associated with increased mortality and may be useful as a prognostic marker in patients with severe CAP.
- [show abstract] [hide abstract]
ABSTRACT: Community-acquired pneumonia (CAP) is a common illness, with the majority of patients treated out of the hospital, yet the greatest burden of the cost of care comes from inpatient management. In the past several years, the management of these patients has advanced, with new information about the natural history and prognosis of illness, the utility of serum markers to guide management, the use of appropriate clinical tools to guide the site-of-care decision, and the finding that guidelines can be developed in a way that improves patient outcome. The challenges to patient management include the emergence of new pathogens and the progression of antibiotic resistance in some of the common pathogens such as Streptococcus pneumoniae. Few new antimicrobial treatment options are available, and the utility of some new therapies has been limited by drug-related toxicity. Ancillary care for severe pneumonia with activated protein C and corticosteroids is being studied, but recently, inpatient care has been most affected by the development of evidence-based "core measures" for management that have been promoted by the Centers for Medicare and Medicaid Services, which form the basis for the public reporting of hospital performance in CAP care.Chest 04/2007; 131(4):1205-15. · 5.85 Impact Factor
- Pneumologia (Bucharest, Romania). 57(4):239-45.
- [show abstract] [hide abstract]
ABSTRACT: Serum cortisol levels rise in response to the stress of critical illness but the optimal range of serum cortisol in such settings is not clearly defined. The objectives of this study were to determine the range of serum cortisol levels in a group of medical intensive care unit patients with severe sepsis/septic shock using uniform criteria, and to correlate serum cortisol levels to mortality. In a prospective observational fashion, 100 medical intensive care unit patients at Northwestern Memorial Hospital in Chicago were enrolled within 48 h of developing severe sepsis/septic shock as defined by the American College of Chest Physicians/Society of Critical Care Medicine. A serum cortisol level was measured during the morning hours in the first 48 h of developing severe sepsis/septic shock. The severity of critical illness was measured by the Acute Physiology and Chronic Health Evaluation II (APACHE II) score. The average patient age was 63 +/- 17 years, 54 patients were men. The average APACHE II score for all patients was 23 +/- 7. In-hospital and 90-day mortality were 51% and 60%, respectively. Four patient groups were defined a priori based on morning serum cortisol levels and their in-hospital mortalities were as follows: group 1 (cortisol < or = 345 nmol/l), n = 11, mortality 54%; group 2 (cortisol 345-552 nmol/l), n = 19, mortality 53%; group 3 (cortisol 552-1242 nmol/l), n = 54, mortality 41%; and group 4 (cortisol > or = 1242 nmol/l), n = 16, mortality 81% (P < 0.01). Cortisol levels were elevated in most patients with septic shock. Cortisol levels less than 552 nmol/l occurred in 30% of patients with septic shock but the mortality in these patients was not significantly increased. Serum cortisol levels > or = 1242 nmol/l were associated with significantly higher mortality.Clinical Endocrinology 01/2004; 60(1):29-35. · 3.40 Impact Factor
Cortisol levels and adrenal response in severe
community-acquired pneumonia: A systematic review of
Jorge I.F. Salluh MD, PhDa,b,⁎, Cássia Righy Shinotsuka MD, MSca,c,
Márcio Soares MD, PhDa, Fernando A. Bozza MD, PhDc,d,
José Roberto Lapa e Silva MD, PhDe, Bernardo Rangel Tura MD, PhDd,
Patrícia T. Bozza MD, PhDb, Carolina Garcia Vidal MDf
aIntensive Care Unit and Postgraduate Program, Instituto Nacional de Câncer, Rio de Janeiro, Brazil, 20230-130
bImmunopharmacology laboratory, Fundação Oswaldo Cruz, Rio de Janeiro, Brazil, 21040-900
cD'Or Institute for research and Education (IDOR), 22281-100
dICU, Evandro Chagas Institute, Fundação Oswaldo Cruz, Rio de Janeiro, Brazil, 21040-900
ePulmonary Diseases Department, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, 21941-913
fService of Infectious Diseases, Hospital Universitari de Bellvitge, Feixa Llarga s/n, 08907 L’Hospitalet de Llobregat
Objectives: Our aim was to review the literature on the prevalence and impact of critical-illness related
corticosteroid insufficiency (CIRCI) on the outcomes of patients with severe community-acquired
Methods: We reviewed Cochrane, Medline, and CINAHL databases (through July 2008) to identify
studies evaluating the adrenal function in severe CAP. Main data collected were prevalence of CIRCI
and its mortality.
Results: We screened 152 articles and identified 7 valid studies. Evaluation of adrenal function varied,
and most studies used baseline total cortisol levels. The prevalence of CIRCI in severe CAP ranged from
0% to 48%. Among 533 patients, 56 (10.7%) had cortisol levels of 10 μg/dL or less and 121 patients
(21.2%) had cortisol levels of 15 μg/dL or less. In a raw analysis, there was no significant difference in
mortality when patients with cortisol levels less than 10 μg/dL (8.6 vs 15.5%; P = .55) or less than
15 μg/dL (12.4 vs 16%; P = .38) were compared with those with cortisol above these levels. In the
meta-analysis, relative risk for mortality were 0.81 (confidence interval, 0.39-1.7; P = .59; χ2= 1.04)
☆This study is original and was not previously submitted to another primary scientific journal.
☆☆Financial support: institutional departmental funds.
★Conflicts of interest: none.
★★Authors' contributions: JIFS, CRS and CGV contributed to the study conception and design, carried out and participated in data analysis and drafted the
manuscript. MS, JRLS, FAB, PTB conceived the study, and participated in its design and coordination, supervised the data analysis and helped to draft the
manuscript. BRT contributed in data analysis and helped to draft the manuscript. All authors read and approved the final manuscript.
⁎Corresponding author. Instituto Nacional de Câncer-INCA, Centro de Tratamento Intensivo-10° Andar, Rio de Janeiro–RJ, CEP: 20230-130, Brazil.
Tel.: +55 21 2506 6120; fax: +55 21 2506 6205.
E-mail addresses: firstname.lastname@example.org, email@example.com (J.I.F. Salluh).
0883-9441/$ – see front matter © 2010 Published by Elsevier Inc.
Journal of Critical Care (2010) xx, xxx–xxx
ARTICLE IN PRESS
for cortisol levels less than 10 μg/dL and relative risk was 0.67 (confidence interval, 0.4-1.14; P = .84;
χ2= 1.4) for cortisol levels less than 15 μg/dL.
Conclusions: A significant proportion of patients with severe CAP fulfilled criteria for CIRCI. However,
CIRCI does not seem to affect the outcomes. Noteworthy, the presence of elevated cortisol levels is
associated withincreasedmortality andmaybeusefulasaprognosticmarkerinpatientswith severeCAP.
© 2010 Published by Elsevier Inc.
Community-acquired pneumonia (CAP) is associated
with significant morbidity and mortality and is the most
common cause of death from infectious diseases in critically
ill patients . Patients with severe CAP often require
intensive care unit (ICU) admission, and despite major
advances in supportive care, an exceedingly high mortality
rate is observed . A recent study evaluating factors
associated with early death in patients with CAP reinforces
the classical concept that some deaths are not only dependent
on antibiotic efficacy but also on other factors, especially
inadequate host response . The hypothalamic-pituitary-
adrenal axis plays a major role in the regulation of the host's
response to infection , and a strong association between
elevated cortisol levels and severity of illness and the risk of
death have been demonstrated [5-7]. Moreover, the presence
of an inadequate adrenal response or adrenal dysfunction or,
as more recently defined, critical illness-related corticoste-
roid insufficiency (CIRCI) may also be helpful in identifying
patients with severe infections at high risk of death [5,8-10].
Complex changes in the endocrine system have been
described in critical illness . Severe infections and the
immune host response to microorganisms are frequently
implicated in the pathogenesis of adrenal response present in
critically ill patients. Clinical and experimental data have
demonstrated that pro and antiinflammatory mediators lead to
decreased production and delivery of cortisol, overcome local
tissue regulation of cortisone/cortisol ratio, and induce down-
regulation of glucocorticoids receptors . Thus, it can be
easily noticed that the adrenal response is a complex
phenomenon in critical illness and its diagnosis can be
misleading. Moreover, its epidemiology and impact on the
In the present article, we reviewed the medical literature,
identified, and analyzed studies that evaluated the adrenal
function in patients with severe CAP. We describe the
frequency of CIRCI and whether it plays a significant role on
the outcomes of patients with severe CAP.
2.1. Search strategy, study selection,
data collection, and analysis
We performed a systematic search of Medline, Cochrane
database, and CINAHL (from 1966 to July 2008) to identify
full-text English language publications that evaluated the
adrenal function in adult hospitalized with severe CAP.
Inclusion criteria were established a priori. Major MESH
search terms included community-acquired infections,
pneumonia, adrenal insufficiency, adrenal failure, cortisol,
corticosteroids, and glucocorticoids. Additional published
reports were identified through a manual search of citations
from retrieved articles. Only original peer-reviewed studies
evaluating the adrenal function in adult patients with CAP
were selected and analyzed. The abstracts of all articles were
used to confirm our target population, and the corresponding
full-text articles were reviewed for the presence of data
evaluating the adrenal function of adult nonimmunocom-
promised patients with CAP. Two investigators (JIFS and
CRS) independently identified the eligible literature. Pre-
defined variables were collected, including year of publica-
tion; study design (prospective/retrospective, cohort/clinical
trial); number of patients included; and hospital mortality
and length of stay, oxygenation, frequency of septic shock,
mechanical ventilation, and pneumonia severity stratifica-
tion. Additional unpublished data were obtained by elec-
tronic mail from most authors. Any inconsistencies between
the 2 investigators (JIFS and CRS) in interpretation of data
were resolved by consensus. Standard descriptive statistics
were applied to describe and compare the populations.
For evaluated homogeneity of studies, using Q Cochran
test and I2, the measure of effect was relative risk calculated
using Mantel-Haenszel approach. All meta-analytic proce-
dures were performed using R software version 2.10.1 and
the package r meta version 2.16. Statistical analyses were
carried out with the open source statistical language and
environment R 2.9.0 [R Foundation for Statistical Comput-
The initial literature search yielded 152 articles, and 145
studies were excluded based on their titles and abstracts. The
reasons for exclusion are shown in Fig. 1. Eventually, we
found and analyzed 7 studies that evaluated the adrenal
function of patients with CAP.
3.1. Description of studies and
Different design and patient selection were observed in
most studies. Overall, 533 patients were enrolled in 7
2J.I.F. Salluh et al.
ARTICLE IN PRESS
studies that evaluated the adrenal function in patients with
CAP (Table 1).
Overall, the studies evaluated a heterogeneous population
of patients with CAP, ranging from mild CAP to those
presentingwithsepticshockandrespiratory failure (Table1).
Only 4 studies evaluated exclusively patients with severe
CAP requiring ICU admission [8,13,17,20]. Feldman et al
 included only critically ill patients with CAP, median
Simplified Acute Physiology Score was 11.5 and ICU
mortality was 33%. In the retrospective study performed by
Salluh et al , 65% of the patients (n = 26) met the criteria
for severe pneumonia according to the British Thoracic
Society guidelines. These patients were severely ill as
indicated by high Acute Physiology And Chronic Health
Evaluation (APACHE) II scores (median, 16; 12-19;
interquartile range, 25%-75%). In addition, a significant
proportion of patients (70%; n = 28) received mechanical
The ICU and hospital mortality rates were 22.5% and 32.5%,
respectively. Brivet et al  evaluated 38 patients, and 71%
(n = 27) were diagnosed as severe CAP according to the
American Thoracic Society criteria. Hospital mortality was
31.5%, and 27 patients (71%) needed mechanical ventilation.
In the prospective study of Salluh et al , 72 CAP patients
admitted to the ICU were evaluated. Patients were stratified
with the CURB-65 [Confusion, Urea, Respiration, Blood
pressure and Age 65 or more], median APACHE II score was
14 (11-17; interquartile range, 25%-75%), 27.8% of the
patients received invasive mechanical ventilation, and 32%
patientspresented with septicshock.TheICU and in-hospital
mortality were 13.8% and 16.7%, respectively. Among the
studies that evaluated critically ill patients with CAP, median
APACHE II ranged from 11 to 14, there was a high
prevalence of mechanical ventilation (27.8%-71% of
patients) and also a high prevalence of septic shock (32%-
47.5% of patients). The ICU mortality varied from 13.8% to
22%, and hospital mortality ranged from 16.7% to 32.5%.
Christ-Crain et al  included 278 consecutive patients
with suspected CAP admitted to the hospital, and 60% of
patients (n = 167) were classified as severe CAP (pneumonia
V had in-hospital mortality rates of 16% and 21%,
respectively. Only 4 patients were hypotensive at presenta-
tion, and there is no available information about the use of
mechanical ventilation or vasopressors or the need for ICU
admission. Mikami et al  evaluated all patients admitted
and who needed mechanical ventilation or ICU admission.
Seventeen patients (54.8%) were diagnosed as severe CAP
(PSI classes IV and V), and only one patient died (3.2%). In
the study conducted by Gotoh et al , all CAP patients
admitted to the hospital were evaluated. Most patients (69%;
Flowdiagram ofstudies selectedandreasonsfor exclusion.
Clinical studies evaluating the adrenal function in patients admitted with severe CAP
ReferenceNo. of patients Patient category Study designEnd points
Feldman et al
Fine et al 
18 Severe CAPProspective single center cohortFrequency of endocrine changes
40 Severe CAP Retrospective single center
Evaluate cortisol levels
Mikami et al
23 Moderate to severe CAP Prospective single center cohort
within an open-label prospective
randomized controlled trial
CAP at emergency
Severe CAPRetrospective single center
Moderate to severe CAP Prospective single center cohort
Hospital length of stay,
antimicrobial therapy duration, and
time to stabilize vital signs
Correlation of adrenal function with survival Christ-Crain
et al 
Brivet et al
Gotoh et al
278Prospective cohort study
38Correlation of cortisol levels with
Correlation of ACTH, cortisol and
cortisol after cosyntropin-stimulation
test with survival, and length of
Correlation of baseline cortisol levels
and cortisol after cosyntropin
stimulation with survival
Salluh et al
72 Severe CAPProspective single center cohort
3 Cortisol levels and adrenal response in severe CAP
ARTICLE IN PRESS
n = 44) had severe CAP (PSI classes IV and V). There is no
available information on septic shock, mechanical ventila-
tion, or ICU admission, and 7 patients (10.9%) died during
hospitalization. Among the studies that evaluated a non-ICU
population of patients admitted with CAP, hospital mortality
ranged from 3.2% to 21% of patients and was significantly
lower than the critically ill population, as expected.
3.2. Diagnosis and prevalence of CIRCI
Diagnostic criteria of CIRCI have only recently been
defined as a random total cortisol of 10 mg/dL or less or a δ
serum cortisol of 9 μg/dL or less after adrenocorticotropic
hormone (ACTH) administration of 250 μg . As a result,
several different criteria to address the adrenal function were
used in each the selected studies. Among all, only total
random cortisol levels were available for all patients. From
533 patients, 121 patients (21.2%) had baseline cortisol
levels of 15 μg/dL or less and 56 (10.7%) had cortisol levels
of 10 μg/dL or less. Christ-Crain et al  evaluated total
and free cortisol levels in patients with CAP. In the whole
study cohort, 54 patients (19.4%) had random total cortisol
levels of 15 μg/dL or less and 30 patients (10.8%) had total
cortisol levels of 10 μg/dL or less. Assessing only patients
with PSI class IV and V (n = 147), 22 patients (14.9%) had
total cortisol levels of 15 μg/dL or less, and 9 patients (6%)
had total cortisol levels of 10 μg/dL or less .
Corticotropin stimulation test was not performed. Feldman
et al  in an earlier study could not observe any case of
CIRCI in patients with lobar pneumonia requiring ICU
admission. Only baseline cortisol and ACTH levels were
evaluated. The ACTH levels were nonsignificantly lower in
nonsurvivors than in survivors, but values were not reported.
Salluh et al  evaluated 40 patients with severe CAP.
Random plasma cortisol levels were obtained, 5 patients
(12.5%) had levels of 10 μg/dL or less and 19 patients (48%)
had levels of 15 μg/dL or less. The ACTH levels or a
corticotropin stimulation test were not obtained. Mikami et al
 evaluated the adrenal function of 23 patients with CAP.
One patient (4.3%) had baseline cortisol of 10 μg/dL or less,
and 7 patients (40.3%) had levels of 15 μg/dL or less. A
corticotropin (250 μg) stimulation test was performed, and
the diagnostic criteria were fulfilled by 10 patients (43%).
Critical illness-related corticosteroid insufficiency was not a
predictive factor for either hospital length of stay or duration
of intravenous antibiotic administration. No data on disease
severity of this subgroup is available; only, there was no
difference in disease severity or other clinical background
between patients with or without CIRCI . Gotoh et al
 evaluated 64 patients hospitalized due to severe CAP
and found that 2 patients (3%) had cortisol levels of 10 μg/
dL or less and 12 patients (19%) had cortisol levels of 15 μg/
dL or less. When corticotropin test was used as a diagnostic
criterion of CIRCI, 13 patients (20%) fulfilled the diagnostic
criterion . Brivet et al  evaluated 38 severe CAP
patients, 1 patient (2.7%) had cortisol levels of 10 μg/dL or
less and 9 patients (25%) had cortisol levels of 15 μg/dL or
less. A corticotropin test was not performed. Finally, Salluh
et al  enrolled 72 patients with CAP admitted to the ICU.
Seventeen (23.6%) had baseline cortisol levels of 10 μg/dL
or less, and 20 patients (27.7%) had cortisol levels of 15
μg/dL or less. Corticotropin stimulation test was performed
in all patients, and 13 (18%) were diagnosed as having
CIRCI based on this criterion. Overall, the prevalence of
CIRCI varied from 2.7% to 48% of patients, ranging from
2.7% to 23.6% when cortisol level of less than 10 μg/dL was
used as CIRCI criteria and from 14.9% to 48% when cutoff
was cortisol level of less than 15 μg/dL. Only 3 studies
performed corticotropin stimulation test, and the prevalence
of CIRCI according to these criteria were 18% and 43%
[8,15,18] (Table 2).
3.3. Adrenal response and mortality
A total of 81 patients (15.2%) died during hospital stay.
In a crude analysis, there was no significant difference in
mortality between patients with CIRCI when compared to
the non-CIRCI group (7/56 [8.6%] vs 74/477 [15.5%]; P =
Prevalence of CIRCI and mortality in the clinical studies according to different CIRCI criteria
No. of patients Cortisol b 10 μg/dLCortisol ≥ 10 μg/dLCortisol b 15 μg/dL Cortisol ≥ 15 μg/dL
0 18 (33.3%)
In the study by Feldman et al , no patients presented low cortisol levels. Numbers in parenthesis represent mortality in the groups of patients with cortisol
level of less than 10 and 10 μg/dL or greater and less than 15 and 15 μg/dL or greater.
⁎P = .55 (comparing mortality between cortisol b 10 μg/dL and ≥ 10 μg/dL).
⁎⁎P = .38 (comparing mortality between cortisol b 15 μg/dL and ≥ 15μg/dL).
4 J.I.F. Salluh et al.
ARTICLE IN PRESS
.55). When a baseline cortisol cutoff level of 15 μg/dL to
define CIRCI was applied, again there was no difference in
mortality (15/121 [12.4%] vs 66/412 [16.0%]; P = .38)
(Table 3). When ICU vs non-ICU patients were compared,
no significant difference in mortality was found in CIRCI
patients when a cortisol cutoff level of less than 10 μg/dL
was applied (5/23 [21.8%] vs 2/33 [6%]; P = .11).
However, when a cortisol cutoff level of less than 15 μg/
dL was used to define CIRCI, there was a significant
difference in mortality between ICU vs non-ICU patients
with adrenal dysfunction (11/48 [22.9%] vs 4/69 [5.8%];
P = .009) (Table 4). Only 3 studies have used corticotropin
test to define CIRCI [8,15,18]. According to this criteria,
when CIRCI vs non-CIRCI patients were compared, there
was no significant difference in mortality (5/30 [16.6%] vs
15/121 [12.3%]; P = .55).
In the meta-analysis, when a cuttoff of basal cortisol level
of less than 10 μg/dL was applied, we computed data for
only 3 studies [8,16,20], due to the small number of CIRCI
patients in the other studies. Relative risk for mortality was
0.81 (IC, 0.39-1.7; P = .59; χ2= 1.04) (Fig. 2). When
cortisol of less than 15 μg/dL was used as criteria, 5 studies
were included [8,16-18,20]. Relative risk was 0.67 (IC, 0.4-
1.14; P = .84; χ2= 1.4) (Fig. 3).
The current systematic review and meta-analysis com-
prehensively evaluates the role of cortisol levels and the
diagnosis of CIRCI on mortality in patients with CAP.
Analyzing the 7 selected studies, we could conclude that a
diagnosis of CIRCI has no significant effect on mortality
even when different cutoffs (baseline cortisol levels b 10 μg/
dL or b 15 μg/dL) are considered. Our meta-analysis also has
demonstrated no significant difference between CIRCI vs
non-CIRCI patients. However, it suggests a possible
association between high cortisol levels and mortality,
which could make cortisol a useful biomarker for assessing
prognosis in patients with severe CAP.
Regarding the impact of adrenal response on the
outcomes, 2 of the evaluated studies thoroughly investigated
and found an association between plasma cortisol and
mortality [16,19]. These results are in accordance with those
obtained from patients with severe sepsis [5-7]. Christ-Crain
et al  observed that cortisol levels were directly
associated with disease severity (as measured by the PSI
score) and hospital mortality and concluded that cortisol
levels are good predictors of severity and outcome in CAP.
In this study, the prognostic accuracy of free cortisol for
patients with CAP was not better than total cortisol. A total
cortisol cutoff value of 34.8 μg/dL was superior to that of
leukocyte count, C-reactive protein, and procalcitonin to
predict death and improve the prognostic accuracy compared
with the PSI alone . Salluh et al  reported in a
prospective study that there was no difference in ICU and
hospital mortality between patients diagnosed with CIRCI
and those who were not. Nonetheless, in this ICU population
of patients with severe CAP, baseline total cortisol levels
were significantly higher in nonsurvivors than in survivors.
Also, baseline cortisol was the best predictor of death when
compared with other laboratorial parameters (D-dimer and C-
reactive protein) and scores (APACHE II, CURB-65, and
SOFA). In this study, δ cortisol or postcorticotropin cortisol
were not able to distinguish survivors from nonsurvivors.
These data support the notion that although the presence of
CIRCI is not associated with worse outcomes, elevated
cortisol levels are associated with disease severity and in-
Finally, it should be acknowledged that, despite the
finding that low cortisol levels are not associated with
worse outcomes in severe CAP, it does not mean that
patients with severe CAP will not benefit from corticoste-
roids. Despite recent literature that challenges the role of
adrenal function status in relation to the response to
corticosteroids , there is also evidence of benefit of
corticosteroids in patients with septic shock  and in
selected patients with severe CAP . Therefore, this issue
is still a source of intense debate that should be evaluated in
future clinical trials.
However, significant heterogeneity in study design and
patient selection is observed among the studies and could
CAP according to different criteria of adrenal dysfunction
Pooled analysis of mortality in patients with severe
Cortisol b 10 μg/dL
Cortisol N 10 μg/dL
Cortisol b 15 μg/dL
Cortisol N 15 μg/dL
aFor comparisons between survivors and nonsurvivors using
cortisol level of less than 10 μg/mL as CIRCI criteria.
bFor comparisons between survivors and nonsurvivors using
cortisol level of less than 15 μg/mL as CIRCI criteria.
patients with severe CAP diagnosed with CIRCI according
Pooled analysis of mortality in ICU and non-ICU
Cortisol b 10 μg/dL
Cortisol b 15 μg/dL
aFor comparisons between survivors and nonsurvivors using
cortisol level of less than 10 μg/dL as CIRCI criteria.
bFor comparisons between survivors and nonsurvivors using
cortisol level of less than 15 μg/dL as CIRCI criteria.
5Cortisol levels and adrenal response in severe CAP
ARTICLE IN PRESS