-
Special Article
Recommendations for the diagnosis and management of
corticosteroidinsufficiency in critically ill adult patients:
Consensus statements from aninternational task force by the
American College of Critical Care Medicine
Paul E. Marik, MD, FCCM; Stephen M. Pastores, MD, FCCM; Djillali
Annane, MD; G. Umberto Meduri, MD;Charles L. Sprung, MD, FCCM;
Wiebke Arlt, MD; Didier Keh, MD; Josef Briegel, MD;Albertus
Beishuizen, MD; Ioanna Dimopoulou, MD; Stylianos Tsagarakis, MD,
PhD; Mervyn Singer, MD;George P. Chrousos, MD; Gary Zaloga, MD,
FCCM; Faran Bokhari, MD, FACS; Michael Vogeser, MD
*See also p. 1987.From the Division of Pulmonary and Critical
Care Med-
icine, Thomas Jefferson University, Philadelphia, PA
(PEM);Critical Care Medicine Fellowship Program, Memorial
Sloan-Kettering Cancer Center, New York, NY (SMP); Critical
CareDepartment, Universite de Versailes Saint-Quentin en Yve-lines,
Hospital Raymond Poincare, Garches, France (DA);Division of
Pulmonary and Critical Care Medicine, Universityof Tennessee HSC,
Memphis, TN (GUM); Department ofAnesthesiology, Hadassah Hebrew
University Medical Cen-ter, Jerusalem, Israel (CLS); Division of
Medical Sciences,Institute of Biomedical Research, Endocrinology
& Metabo-lism, The Medical School, University of Birmingham,
Bir-mingham, UK (WA); Department of Anesthesiology and In-tensive
Care Medicine, Campus Virchow-Clinic, HumboldtUniversity, Berlin,
Germany (DK); Department of Anesthesi-ology, University of Munich,
Klinikum Grosshadern, Munich,Germany (JB); Department of Intensive
Care, VU UniversityMedical Center, Amsterdam, Netherlands (AB);
Departmentof Critical Care Medicine, Athens University, Medical
School,Athens, Greece (ID); Department of Endocrinology,
AthensPolyclinic, Athens, Greece (ST); Department of Medicine
andWolfson Institute of Biomedical Research, University
CollegeLondon, Jules Thorn Building, Middlesex Hospital, London,UK
(MS); First Department of Pediatrics, Athens UniversityMedical
School, Athens, Greece (GPC); Baxter Healthcare,Clintec Nutrition,
Deerfield, IL (GZ); Department of Trauma,
Stronger Hospital of Cook County, Chicago, IL (FB); Hospitalof
the University of Munich, Institute of Clinical Chemistry,Munich,
Germany (MV).
The American College of Critical Care Medicine (ACCM),which
honors individuals for their achievements and contri-butions to
multidisciplinary critical care medicine, is theconsultative body
of the Society of Critical Care Medi-cine (SCCM) that possesses
recognized expertise inthe practice of critical care. The College
has developedadministrative guidelines and clinical practice
parameters forthe critical care practitioner. New guidelines and
practiceparameters are continually developed, and current ones
aresystematically reviewed and revised.
Dr. Marik has received lecture fees from Eli Lilly andMerck. Dr.
Keh has received grant support from the GermanResearch Foundation
and German Ministry of Education andResearch (HYPRESS:
Hydrocortisone for Prevention of SepticShock). Dr. Sprung has been
a member of a data monitoringand safety committee for Artisan
Pharma, Novartis Corpora-tion, and Hutchinson Technology
Incorporated. He hasserved as a consultant for AstraZeneca, Eisai
Corporation, EliLilly, and GlaxoSmithKline. He has received grant
supportfrom the European Commission, Takeda, and Eisai
Corpora-tion. He has received lecture fees from Eli Lilly. Drs.
Sprung,Annane, Keh, Singer, and Briegel were investigators in
theCORTICUS study, which was supported by the EuropeanCommission,
the European Society of Intensive Care Medi-
cine, the European Critical Care Research Network,
theInternational Sepsis Forum, and the Gorham Foundation. Dr.Annane
has received grant support from the French Ministryof Health for
the prognostic value of a adrenocorticotrophichormone test in
septic shock; the French multicenter, ran-domized, controlled trial
on hydrocortisone plus fludrocorti-sone in septic shock; the
ongoing French multicenter 2 2factorial study that compares strict
glucose control vs. con-ventional treatment for steroid-treated
septic shock and hy-drocortisone alone vs. hydrocortisone and
fludrocortisone;and a French multicenter 2 2 factorial trial that
compareshydrocortisone plus fludrocortisone, activated protein C,
thecombination of the two drugs, and placebos for the treatmentof
septic shock. Dr. Pastores has received grant support formEisai
Medical Research (phase 3 trial of E5564 in severesepsis), and
Artisan Pharma (phase 2 sepsis with dissemi-nated intravascular
coagulation trial). The remaining authorshave not disclosed any
conflicts of interest with respect tothis article.
For information regarding this article,
E-mail:[email protected]
Copyright 2008 by the Society of Critical CareMedicine and
Lippincott Williams & Wilkins
DOI: 10.1097/CCM.0b013e31817603ba
Objective: To develop consensus statements for the diagnosis
andmanagement of corticosteroid insufficiency in critically ill
adult patients.
Participants: A multidisciplinary, multispecialty task force of
experts incritical care medicine was convened from the membership
of the Society ofCritical Care Medicine and the European Society of
Intensive Care Medicine.In addition, international experts in
endocrinology were invited to partici-pate.
Design/Methods: The task force members reviewed published
literatureand provided expert opinion from which the consensus was
derived. Theconsensus statements were developed using a modified
Delphi methodology.The strength of each recommendation was
quantified using the ModifiedGRADE system, which classifies
recommendations as strong (grade 1) or weak(grade 2) and the
quality of evidence as high (grade A), moderate (grade B), orlow
(grade C) based on factors that include the study design, the
consistencyof the results, and the directness of the evidence.
Results: The task force coined the term critical illnessrelated
corticosteroidinsufficiency to describe the dysfunction of the
hypothalamic-pituitary-adrenal axisthat occurs during critical
illness. Critical illnessrelated corticosteroid insufficiencyis
caused by adrenal insufficiency togetherwith tissue corticosteroid
resistance andis characterized by an exaggerated and protracted
proinflammatory response.Critical illnessrelated corticosteroid
insufficiency should be suspected in hypoten-sive patients who have
responded poorly to fluids and vasopressor agents, partic-ularly in
the setting of sepsis. At this time, the diagnosis of tissue
corticosteroidresistance remains problematic. Adrenal insufficiency
in critically ill patients is bestmade by a delta total serum
cortisol of
-
Severe illness and stress stronglyactivate the
hypothalamic-pitu-itary-adrenal (HPA) axis andstimulate the release
of adreno-corticotrophic hormone (ACTH) from thepituitary, which in
turn increases the re-lease of cortisol from the adrenal
cortex(13). This activation is an essential com-ponent of the
general adaptation to illnessand stress and contributes to the
mainte-nance of cellular and organ homeostasis.Adrenalectomized
animals succumb rap-idly to hemorrhagic and septic shock,
andsteroid replacement is protective againstthese challenges (4,
5).
Once considered a rare diagnosis inthe intensive care unit,
adrenal failureis being reported with increasing fre-quency in
critically ill patients with septicshock, severe community-acquired
pneu-monia, trauma, head injury, burns, liverfailure, HIV
infection, pancreatitis, aftercardiac surgery, after the use of
etomi-date, and in brain-dead organ donors (611). Adrenal failure
may be associatedwith structural damage to the adrenalgland,
pituitary gland, or hypothalamus;however, many critically ill
patients de-velop reversible failure of the HPA axis.
Although it is generally agreed thatadrenal failure may be
common in sub-groups of critically ill patients, the diag-nosis and
management of this disorderremains controversial, with poor
agree-ment among the experts. The objective ofthis task force was
therefore to developconsensus statements by experts in thefield
based on the best available scientificevidence (12).
METHODS
Experts were selected from the mem-bership lists of the Society
of CriticalCare Medicine (SCCM) and the EuropeanSociety of
Intensive Care Medicine (ES-ICM). Specific individuals were
selectedto represent geographic diversity and abroad range of
expertise on the basis oftheir published research. In addition,
en-docrinologists with expertise in this areawere invited to join
the task force.
The consensus statement was devel-oped using a modified Delphi
methodol-ogy (12). The Delphi method, originallydeveloped by the
RAND Corporation, is astructured process that uses a series
ofquestionnaires, each referred to as around, to both gather and
provide infor-mation (13, 14). With each round, theanswers are
modified based on the re-sponses of the previous round. The
rounds continue until group consensus/majority is reached. This
approach hasseveral distinct advantages. It allows theinclusion of
a large number of individualsacross diverse geographic locations
andwith a broad range of expertise. One of itskey advantages is
that unlike a face-to-face meeting of experts, it eliminates
thepossibility that a specific expert mightdominate the consensus
process. TheDelphi method helps to minimize the ef-fects of group
interactions and maximizesthe ability to elicit expert
knowledge.
The task force members individuallyand collectively undertook a
systematicsearch of published literature pertainingto the diagnosis
and treatment of adrenalfailure in critically ill adult patients
usingMedline, CINAHL, EMBASE, and the Co-chrane library. In
addition, the referencelists of relevant articles were reviewed
foradditional published works. Key wordsused in these searches
included pitu-itaryadrenal system, adrenal insuffi-ciency, adrenal
glands, pituitaryadrenalfunction tests, hydrocortisone,
glucocor-ticoids (GC), adrenal cortex hormones,glucocorticoid
receptor (GR), criticalcare, intensive care units, intensive
care,ARDS, shock septic, sepsis, and sepsissyndrome. A
comprehensive bibliogra-phy was developed, with the
referencesstored and cataloged using an electronicreference manager
(Reference Managerv11.1, Thompson ResearchSoft, Carlsbad,CA).
We used electronic mail to conductthe Delphi process. A list of
questions forreview was determined. Once a majorityagreement was
reached on each question,the strength of each recommendationwas
quantified using the Modified Gradesof Recommendation Assessment,
Devel-opment, and Evaluation (GRADE) systemdeveloped by the
American College ofChest Physicians (Appendix 1) (15). In all,there
were seven rounds until a majorityagreement was achieved on all the
ques-tions. In addition, the group met in Paris,France, in
September 2005 and again atthe Society of Critical Care Medicine
35thCritical Care Congress in San Francisco,CA, in January 2006 to
review theprogress of the Delphi process. The initialdraft of the
manuscript was written bythe Chair (P. E. Marik). The draft
manu-script was reviewed and iteratively editedby all members of
the task force.
A meta-analysis of randomized con-trolled trials that compared
the 28-daymortality and vasopressor dependency ofpatients with
septic shock and the 28-day
mortality of patients with acute respira-tory distress syndrome
(ARDS) who re-ceived either moderate-dose corticoste-roid or
placebo was performed. Four ofthe task force members (P. E. Marik,
D.Annane, S. M. Pastores, G. U. Meduri)reviewed the task force
bibliography forrelevant studies. Septic shock was definedby the
American College of Chest Physi-cians/Society of Critical Care
MedicineConsensus Conference and ARDS by theAmericanEuropean
Consensus Confer-ence (16, 17). Vasopressor dependencywas defined
as the requirement for a va-sopressor agent after 7 days of
treatmentwith a glucocorticoid (GC). The reviewersindependently
abstracted data from all el-igible studies. Data were abstracted
onstudy design, study size, corticosteroiddosage, vasopressor
dependency, and 28-day mortality. Study and data inclusionwas by
consensus. We used the randomeffects models using Review Manager
4.2(Cochrane Collaboration, Oxford, UK) forall analyses and
considered p .05 (two-sided) as significant. Summary effects
es-timates are presented as odds ratio with95% confidence
intervals. We assessedheterogeneity between studies using
theCochran Q statistic with p .10 indicat-ing significant
heterogeneity and the I2
with suggested thresholds for low (2549%), moderate (50 74%),
and high(75%) values (1821).
BACKGROUND
Exposure to hostile conditions resultsin a series of coordinated
responsesoften referred to as stress responsesorganized to enhance
survival; these in-clude a series of complex central andperipheral
adaptations. This stress re-sponse is mediated mainly by the
HPAaxis and the sympathoadrenal system,which includes the
sympathetic nervoussystem and the adrenal medulla (Fig. 1)(2224).
The HPA axis and the sympa-thoadrenal system are functionally
re-lated. Activation of the sympathoadrenalsystem results in the
secretion of epi-nephrine and norepinephrine from theadrenal
medulla and also leads to an in-creased production of inflammatory
cyto-kines, such as interleukin-6. Activation ofthe HPA axis
results in increased secre-tion from the paraventricular nucleus
ofthe hypothalamus of corticotropin-releasing hormone, a 41-amino
acid pep-tide, and arginine vasopressin. Cortico-tropin-releasing
hormone plays a pivotalintegrative role in the response to
stress.
1938 Crit Care Med 2008 Vol. 36, No. 6
-
Corticotropin-releasing hormone stimu-lates the production of
ACTH by the an-terior pituitary, causing the zona fascicu-lata of
the adrenal cortex to producemore GCs (cortisol in humans,
cortico-sterone in rats). Arginine vasopressin is aweak ACTH
secretagogue and vasoactivepeptide that acts synergistically with
cor-ticotropin-releasing hormone to increasesecretion of ACTH. The
increase in corti-sol production results in multiple
effects(metabolic, cardiovascular, and immune)aimed at maintaining
or restoring ho-meostasis during stress.
Cortisol Physiology, Synthesis,and Glucocorticoid Receptors
Cortisol is the major endogenous GCsecreted by the adrenal
cortex. More than90% of circulating cortisol is bound
tocorticosteroid-binding globulin, with10% in the free,
biologically activeform (25, 26). Corticosteroid-bindingglobulin is
the predominant binding pro-tein, with albumin binding a
lesseramount. During acute illness, particu-larly sepsis,
corticosteroid-binding glob-ulin levels fall by as much as 50%,
result-
ing in a significant increase in thepercentage of free cortisol
(27, 28). Thecirculating half-life of cortisol varies from70 to 120
mins. The adrenal gland doesnot store cortisol; increased
secretionarises due to increased synthesis underthe control of ACTH
(29). Cholesterol isthe principal precursor for steroid
bio-synthesis in steroidogenic tissue. In a se-ries of sequential
enzymatic steps, cho-lesterol is converted to pregnenolone andthen
to the end products of adrenal bio-synthesis, namely, aldosterone,
dehydro-epiandrostenedione, and cortisol (29).The first and
rate-limiting step in adrenalsteroidogenesis is the formation of
preg-nenolone from cholesterol. At rest andduring stress, about 80%
of circulatingcortisol is derived from plasma choles-terol, the
remaining 20% being synthe-sized in situ from acetate and other
pre-cursors (30). Experimental studiessuggest that high-density
lipoprotein isthe preferred cholesterol source of steroi-dogenic
substrate in the adrenal gland(31). Recently, mouse SR-B1
(scavengerreceptor, class B, type 1) and its humanhomolog (Cla-1)
have been identified asthe high-affinity high-density
lipoprotein
receptor mediating selective cholesteroluptake (3234). These
receptors are ex-pressed at high levels in the parenchymalcells of
the liver and the steroidogeniccells of the adrenal glands, ovary,
andtestis (35).
Cortisol exerts its effects after uptakefrom the circulation by
binding to intra-cellular glucocorticoid receptors (GRs)(3). These
receptors belong to a steroid-hormone-receptor superfamily of
tran-scription factors, which are made up of aC-terminal ligand
binding domain, a cen-tral DNA binding domain interactingwith
specific DNA sequences on targetgenes, and an N-terminal
hypervariableregion. The binding of cortisol to GR inthe cytoplasm
results in the activation ofthe steroid receptor complex via a
processinvolving the dissociation of heat shockproteins (heat shock
proteins 90 and 70)and FK-506 binding proteins (3638).
In-tracellularly, the cortisol-GR complexmoves to the nucleus,
where it binds as ahomodimer to DNA sequences called
glu-cocorticoid-responsive elements locatedin the promoter regions
of target genes,which then activate or repress transcrip-tion of
the associated genes. In addition,the cortisol-GR complex may
affect cellu-lar function indirectly by binding to andmodulating
the transcriptional activity ofother nuclear transcription factors,
suchas nuclear factor B (NF-B) and activa-tor protein-1. Overall,
GCs affect thetranscription of thousands of genes inevery cell of
the body. It has been esti-mated that GCs affect 20% of the
genomeof mononuclear blood cells (39).
GCs play a major role in regulatingthe activity of NF-B, which
plays a cru-cial and generalized role in inducing cy-tokine gene
transcription (40 42).NF-B is normally maintained in an in-active
form by sequestration in the cyto-plasm through interaction with
inhibi-tory proteins (IBs). On stimulation bylipopolysaccharide,
double-strandedDNA, physical and chemical stresses, andinflammatory
cytokines, the latent NF-B/IB complex is activated by
phosphor-ylation and proteolytic degradation ofIB, with exposure of
the NF-B nuclearlocalization sequence. The liberatedNF-B then
translocates to the nucleusand binds to promoter regions of
targetgenes to initiate the transcription of mul-tiple cytokines
(including tumor necrosisfactor-, interleukin-1, and
interleukin-6), cell adhesion molecules (e.g., intercel-lular
adhesion molecule-1, E-selectin),and other mediators of
inflammation.
Figure 1. Activation of the hypothalamic-pituitary-adrenal axis
by a stressor and the interaction withthe inflammatory response.
ACTH, adrenocorticotrophic hormone; CRH, corticotropin-releasing
hor-mone; IL-6, interleukin-6; IL-11, interleukin-11; LIF, leukemia
inhibitory factor; POMC, pro-opiomelanocortin; TGF-beta,
transforming growth factor-; TNF, tumor necrosis factor.
1939Crit Care Med 2008 Vol. 36, No. 6
-
GCs inhibit the activity of NF-B by in-creasing the
transcription of IBs and bydirectly binding to and inhibiting
NF-B(41, 42).
Cortisol has several important physio-logic actions on
metabolism, cardiovas-cular function, and the immune system(6, 43).
The metabolic effects of cortisolinclude an increase in blood
glucose con-centrations through the activation of keyenzymes
involved in hepatic gluconeo-genesis and inhibition of glucose
uptakein peripheral tissues such as the skeletalmuscles. In
addition, in adipose tissue,lipolysis is activated, resulting in
the re-lease of free fatty acids into the circula-tion. Cortisol
also has a permissive effecton other hormones, increasing
glucoselevels, including catecholamines and glu-cagon. Sustained
cortisol hypersecretionstimulates glucose production at the
ex-pense of protein and lipid catabolism andinsulin resistance.
Cortisol increases blood pressurethrough several mechanisms
involvingthe kidney and vasculature. In vascularsmooth muscle,
cortisol increases sensi-tivity to vasopressor agents such as
cat-echolamines and angiotensin II (44, 45).These effects are
mediated partly by theincreased transcription and expression ofthe
receptors for these hormones (44,45). Although the effect of GCs on
nitricoxide is complex, it seems to increaseendothelial nitric
oxide synthetase,thereby maintaining microvascular per-fusion
(4649). Cortisol has potent anti-inflammatory actions, including
the re-duction in the number and function ofvarious immune cells,
such as T and Blymphocytes, monocytes, neutrophils,and eosinophils,
at sites of inflammation.Cortisol decreases the production of
cy-tokines, chemokines, and eicosanoids andenhances the production
of macrophagemigration inhibitory factor (22, 50).
Dysfunction of the HPA AxisDuring Acute Illness
The acute stress response during crit-ical illness is
characterized by activationof the HPA and sympathoadrenal
systemaxis, with increased secretion of cortisol,an increase in the
percentage of free cor-tisol, and increased translocation of theGR
complex into the nucleus. Impor-tantly, there is increasing
evidence thatin many critically ill patients, this path-way may be
impaired (27, 51, 52). Thereported prevalence of adrenal
insuffi-ciency in critically ill patients varies
widely (077%), depending on the popu-lation of patients studied
and the diag-nostic criteria. However, the overall prev-alence of
adrenal insufficiency incritically ill medical patients
approxi-mates 1020%, with a rate as high as60% in patients with
septic shock. In astudy recently published by Annane et al.(53),
the prevalence of adrenal insuffi-ciency (as determined by
metyraponetesting) in patients with severe sepsis andseptic shock
was reported to be 60%. Themajor effect of adrenal insufficiency
inthe critically ill patient is manifestedthrough alterations in
the systemic in-flammatory response and cardiovascularfunction.
The mechanisms leading to dysfunc-tion of the HPA axis during
critical illnessare complex and poorly understood andlikely include
decreased production ofcorticotropin-releasing hormone, ACTH,and
cortisol and the dysfunction of theirreceptors. A subset of
patients may havestructural damage to the adrenal glandfrom either
hemorrhage or infarction,and this may result in long-term
adrenaldysfunction. Adrenal hemorrhage hasbeen described with blunt
abdominaltrauma, after major surgery, in dissemi-nated
intravascular coagulation associ-ated with sepsis, and in patients
withburns, heparin-induced thrombocytope-nia, the antiphospholipid
syndrome, HIVinfection, disseminated fungal infections,and
tuberculosis (3, 5459). In addition,patients who have been treated
long termwith adrenally suppressive doses of exog-enous GCs are
likely to develop secondaryadrenal insufficiency (3, 6). However,
itseems that most critically ill patients whodevelop adrenal
insufficiency develop re-versible dysfunction of the HPA axis
(6,60). Decreased production of cortisol orACTH is particularly
common in patientswith severe sepsis and septic shock (60).Annane
et al. (53) demonstrated an in-creased risk of adrenal
insufficiency inpatients with positive blood cultures andthose with
Gram-negative infections.
Clinical and experimental data indi-cate that the failure to
improve in sepsisand ARDS is frequently associated withfailure of
activated GRs to down-regulatethe transcription of inflammatory
cyto-kines, despite elevated levels of circulat-ing cortisol, a
condition defined as sys-temic inflammation-associated GCresistance
(61). Tissue corticosteroid re-sistance is a well-known
manifestation ofchronic inflammatory diseases, such aschronic
obstructive pulmonary disease,
severe asthma, systematic lupus ery-thematosus, ulcerative
colitis, and rheu-matoid arthritis (6265). It is thereforelikely
that acute inflammation, similar tochronic inflammation, may be
associatedwith tissue corticosteroid resistance (61).In
experimental models, endotoxin andproinflammatory cytokines have
beenshown to cause decreased GR nucleartranslocation (66 68). In an
ex vivomodel, Meduri et al. (69) demonstratedreduced nuclear
translocation of the GRcomplex in patients with fatal ARDS,
de-spite adequate cytoplasmic (and serum)levels of cortisol. It is
likely that multiplemechanisms cause systemic
inflamma-tion-associated GC resistance, includingdecreased GR
number, increased expres-sion of the beta isoform of the GR
(unableto bind ligand), altered ratio of chaperoneproteins (FK
binding proteins and heatshock protein 90), reduced affinity of
theGR for ligand, altered nuclear receptorcoactivators, reduced DNA
binding, de-creased histone acetylation, increased ac-tivity of the
P-glycoprotein membranetransport pump, and increased conver-sion of
cortisol to cortisone (61, 68, 7072). Furthermore, polymorphisms of
theGR and other pivotal genes may influencethe downstream effects
of the GCGR in-teraction (73, 74). Additional research inthis area,
particularly as it applies to crit-ically ill patients, is urgently
required.
Current evidence suggests that medi-ators released in patients
with critical ill-ness, and sepsis in particular, may
eitherstimulate or impair the synthesis and ac-tion of cortisol via
actions on the HPAaxis and the GR signaling system. The neteffect
of these opposing actions on theHPA axis and GR may be time
dependentand, in addition, depend on the severity ofillness and the
extent and pattern of me-diator production. Although the focus
onmost research has been in the area ofsepsis and ARDS, it is
likely that similarmechanisms operate in other
disorderscharacterized by significant systemic in-flammation,
including pancreatitis,burns, post-cardiopulmonary bypass, andliver
failure (7579).
RECOMMENDATIONS OF THETASK FORCE
Critical IllnessRelated CorticosteroidInsufficiency
Recommendation 1: Dysfunction of theHPA axis in critical illness
is best de-scribed by the term critical illness
1940 Crit Care Med 2008 Vol. 36, No. 6
-
related corticosteroid insufficiency(CIRCI).
Recommendation 2: The terms abso-lute or relative adrenal
insufficiencyare best avoided in the context of crit-ical
illness.
Dysfunction of the HPA axis in criticalillness is best described
by the term crit-ical illnessrelated corticosteroid insuffi-ciency
(CIRCI). CIRCI is defined as inad-equate cellular corticosteroid
activity forthe severity of the patients illness. CIRCImanifests
with insufficient GC-GRmediated down-regulation of proinflam-matory
transcription factors, leading topersistent elevation of
proinflammatorymediators over time. CIRCI occurs as aresult of a
decrease in adrenal steroidproduction (adrenal insufficiency) or
tis-sue resistance to GCs (with or withoutadrenal insufficiency).
Adrenal insuffi-ciency may arise due to dysfunction atany point in
the HPA axis. The termsabsolute or relative adrenal
insufficiencyare best avoided in the context of criticalillness
(80). CIRCI is a dynamic process(i.e., patients may not have CIRCI
at ad-mission to the hospital/intensive careunit but may develop
CIRCI during thecourse of their illness) (8183). CIRCI isusually a
reversible condition caused byproinflammatory mediators; however,
itmay also arise due to structural damageof the adrenal gland.
CIRCI may affect thebalance between proinflammatory
andanti-inflammatory pathways and therebyinfluence immune,
metabolic, vascular,and organ dysfunction.
Diagnosis of Adrenal Insufficiency
Recommendation 3: At this time, adre-nal insufficiency in
critical illness isbest diagnosed by a delta cortisol (after250 g
cosyntropin) of 9 g/dL or arandom total cortisol of 10 g/dL.
Strength of Recommendation: 2B
Recommendation 4: The use of freecortisol measurements cannot be
rec-ommended for routine use at this time.Although the free
cortisol assay hasadvantages over the total serum corti-sol, this
test is not readily available.Furthermore, the normal range of
thefree cortisol in critically ill patients iscurrently
unclear.
Strength of Recommendation: 2B
Recommendation 5: The ACTH stimu-lation test should not be used
to iden-tify those patients with septic shock orARDS who should
receive GCs.
Strength of Recommendation: 2B
The diagnosis of adrenal insufficiencyin critically ill patients
has been based onthe measurement of a random total se-rum cortisol
(stress cortisol level) orthe change in the serum cortisol in
re-sponse to 250 g of synthetic ACTH(ACTH stimulation test), the
so-calleddelta cortisol (6, 84). Both of these testshave
significant limitations in the criti-cally ill (85). Assays for
serum cortisolmeasure the total hormone concentra-tion (serum-free
cortisol plus the pro-tein-bound fraction). The consensus isthat
the free cortisol, rather than theprotein-bound fraction, is
responsible forthe physiologic function of the hormoneat the
cellular level (6, 50, 86). In mostcritically ill patients,
corticosteroid-binding globulin levels are decreased andthe
percentage of free cortisol is in-creased (27, 51, 52, 86, 87).
Furthermore,with acute stimulation of the adrenalgland, the
relative increase of free bioac-tive cortisol concentrations is
substan-tially more pronounced than the increaseof total cortisol
concentrations (27, 51,52, 8688). Consequently, in critically
illpatients, the total serum cortisol levelmay not accurately
reflect the free corti-sol level. This dissociation between
thetotal and free cortisol level is mostmarked in patients with a
serum albuminof 2.5 mg/dL (85, 87, 89).
Although measurement of the freecortisol level may arguably be
prefera-ble, this test is currently not widelyavailable. It is
likely, however, that withimprovement in laboratory techniquesand
clinical demand, this test will be-come commercially available
(90). Theinterpretation of the total serum cortisolconcentration is
further complicated bythe fact that the specificity,
sensitivity,and performance of the commerciallyavailable assays are
not uniform (91). It islikely that the variation in assay
charac-teristics might be even more significantin critically ill
patients, especially thosewith septic shock (91, 92).
Cross-reactiv-ity of the cortisol immunoassay with pre-cursors or
metabolites of cortisol thataccumulate in sepsis may account for
thisobservation.
Although a delta cortisol of 9 g/dLhas proven to be an important
prognosticmarker (9, 53, 93, 94), and a marker ofresponse to
treatment with corticoste-roids (95, 96), the ACTH stimulation
testhas a number of limitations. The deltacortisol is a measure of
the ability of the
adrenal gland to increase production ofcortisol in response to
ACTH; it does notassess the integrity of the HPA axis, theresponse
of the HPA axis to other stresses(i.e., hypotension, hypoglycemia),
or theadequacy of stress cortisol levels. In ad-dition, the ACTH
stimulation test may bepoorly reproducible, especially in
patientswith septic shock (97, 98). Despite theselimitations,
Annane et al. (53) have re-ported that a delta cortisol of 9
g/dLand a random total cortisol of 10 g/dLwere the best predictors
of adrenal insuf-ficiency (as determined by metyraponetesting) in
patients with severe sepsis/septic shock. Furthermore, although
the1-g ACTH stimulation test may be morephysiologic and have a
greater sensitivitythan the 250-g test, due to limited data,the 1-g
test dose is currently not rec-ommended (99). It should also be
appre-ciated that at present, we are unable tomeasure tissue GC
resistance or deter-mine the circulating cortisol level that
isrequired to overcome tissue resistance.
In those patients (severe sepsis, septicshock, and ARDS) most
likely to benefitfrom treatment with moderate-dose GCs,it is not
clear that treatment should bebased on the results of adrenal
functiontesting. To date, six randomized, placebo-controlled
studies have evaluated hydro-cortisone treatment (200300 mg/day)
inpatients with septic shock (95, 100103)(Figs. 2 and 3). In these
studies, morerapid shock reversal was noted in patientstreated with
hydrocortisone, and thisbenefit was noted in both ACTH respond-ers
(delta cortisol of 9 mg/dL) and non-responders (delta cortisol of 9
mg/dL)(Fig. 2). Furthermore, recent randomizedcontrolled studies in
patients with earlyARDS (treatment within 14 days) and se-vere
community-acquired pneumoniademonstrated improved outcome withGCs
(when compared with placebo), in-dependent of adrenal function
testing(see section below) (7, 104, 105). Thesedata suggest that in
patients with septicshock and early ARDS, the decision totreat with
moderate-dose corticosteroidsshould be based on clinical criteria
andnot on the results of adrenal functiontesting. The inability to
diagnose cortico-steroid tissue resistance may partly ex-plain
these observations.
Who to Treat with Glucocorticoids?
Recommendation 6: Hydrocortisoneshould be considered in the
manage-ment strategy of patients with septicshock, particularly
those patients who
1941Crit Care Med 2008 Vol. 36, No. 6
-
have responded poorly to fluid resusci-tation and vasopressor
agents.
Strength of Recommendations: 2B
The benefit of moderate-dose hydro-cortisone (200300 mg/day) in
patientswith septic shock has been evaluated insix randomized
controlled trials (95,100103, 106). A meta-analysis of thesesix
studies (including the recently com-pleted CORTICUS study)
demonstratesgreater shock reversal (at day 7) withhydrocortisone
but no benefit in terms ofmortality (Figs. 2 and 3). The
variability
in study size, inclusion criteria, and cor-ticosteroid dosing
limits the interpreta-tion of this meta-analysis. Nevertheless,the
French multicenter study and the re-cently completed European
multicenterstudy (CORTICUS) were better poweredto detect a survival
difference and deservefurther analysis. Annane et al. (95)
ran-domized 300 patients with refractory sep-tic shock (systolic
blood pressure of 90mm Hg for 1 hr, despite fluid resusci-tation
and the use of vasopressor agents)to treatment with hydrocortisone
(50 mgintravenously every 6 hrs) and oral
fludrocortisone (50 g daily) or matchingplacebo for 7 days. All
patients underwentan ACTH stimulation test. There was a30% decrease
in 28-day mortality in thehydrocortisonefludrocortisone
group(hazard ratio, 0.67; 95% confidence inter-val, 0.470.95; p
.02) (95). This benefitwas confined to the group of nonre-sponders
(delta cortisol of 9 g/dL).
The CORTICUS study was a double-blind, randomized,
placebo-controlledstudy performed in 52 centers through-out Europe
(106). A total of 500 patients(499 available to analyze) were
enrolled
Figure 2. Meta-analysis of treatment with moderate-dose
hydrocortisone on shock reversal at day 7 in patients with septic
shock grouped by response toadrenocorticotrophic hormone. RR,
relative risk; 95% CI, 95% confidence interval.
Figure 3. Meta-analysis of treatment with moderate-dose
hydrocortisone on 28-day survival in patients with septic shock.
RR, relative risk; 95% CI, 95%confidence interval.
1942 Crit Care Med 2008 Vol. 36, No. 6
-
between March 2002 and November2005. Inclusion criteria included
septicshock (systolic blood pressure of 90mm Hg, despite adequate
fluid resuscita-tion or need for vasopressors) and evi-dence of
organ dysfunction attributableto sepsis. Patients were randomized
tohydrocortisone (50 mg intravenously ev-ery 6 hrs for 5 days, then
50 mg intrave-nously every 12 hrs for 3 days, followedby 50 mg
intravenously daily for 3 days)or matching placebo. Patients did
not re-ceive fludrocortisone. Although the base-line
characteristics of the patients weresimilar, only 35% of the cohort
weremedical patients, with the abdomen be-ing the commonest source
of infection(48%). There was no difference in the28-day all-cause
mortality between thosepatients who received hydrocortisone
ascompared with placebo. Furthermore,there was no difference in
mortality be-tween the groups when stratified as re-sponders (delta
cortisol of 9 g/dL) ornonresponders (delta cortisol of 9 g/dL) to
the ACTH stimulation test. How-ever, the patients who received
hydrocor-tisone had more rapid resolution of shock(p .001 for
responders and p .06 fornonresponders). There were, however,more
episodes of new infection (not sta-tistically significant) and
septic shock (re-bound inflammation) in the hydrocorti-sone group.
The prevalence of otheradverse events, including critical
illnesspolyneuropathy, was similar betweengroups.
A number of factors may account forthe different results of the
French multi-center study and the CORTICUS study.The patients
enrolled in the French studywere sicker than those enrolled in
theCORTICUS study (28-day mortality in theplacebo arm of 61% vs.
31.5%). Further-more, the time window of enrollment was8 hrs in the
French study as comparedwith 72 hrs in the CORTICUS study. It
islikely that only patients at a high risk ofdeath will benefit
from corticosteroids,and this benefit may diminish with a de-
lay in instituting treatment. It is alsopossible that
improvements in the sup-portive care of critically ill patients
withseptic shock over the last decade haveincreased the survival of
patients withCIRCI who would otherwise have died.The demographics
and clinical character-istics of the patients enrolled in the
twostudies were quite different, with 40.1%of patients in the
French study beingsurgical patients as compared with 64.5%in the
CORTICUS study. Source controlmay be more important in
determiningthe outcome of sepsis in surgical patientsthan that of
adjunctive interventions.Furthermore, it is possible that
selectionbias affected the demographics and out-come of the
CORTICUS study. Althoughit has been suggested that clinical
equi-poise existed during enrollment into theCORTICUS study (107),
many intensivistscontinue to use corticosteroids in themanagement
of patients with septicshock (108, 109).
Given the different outcomes of theFrench and CORTICUS studies,
whatshould the clinician do? Considering thecentral role of
cortisol in modulating thestress response and recognizing the
po-tential suppressive effects of sepsis on theHPA axis and on GR
activity, the use ofmoderate-dose hydrocortisone seems ra-tional in
patients with septic shock poorlyresponsive to fluid and
vasopressor resus-citation. This is supported by recent datathat
demonstrate that up to 60% of pa-tients with severe sepsis and
septic shockhave adrenal insufficiency (53). The bestavailable
clinical evidence suggests thatmoderate-dose hydrocortisone results
insignificantly more rapid resolution ofshock (Fig. 2). The effects
of moderate-dose hydrocortisone on mortality seemless clear (Fig.
3). Nevertheless, based oncurrent data, hydrocortisone should
beconsidered in the management strategyof patients with septic
shock, particularlythose patients who have responded poorlyto fluid
resuscitation and vasopressoragents. As noted in Figure 2, more
rapid
resolution of shock was noted in bothresponders and
nonresponders. Thus, atthis time, it seems that the decision
totreat patients with septic shock shouldnot be based on the
results of a randomtotal cortisol level or the response toACTH. In
addition, it should be notedthat the administration of
hydrocortisoneduring septic shock has been demon-strated to reduce
the prevalence of post-traumatic stress disorder and improvethe
emotional well-being of survivors ofseptic shock (110).
Recommendation 7: Moderate-dose GCshould be considered in the
manage-ment strategy of patients with earlysevere ARDS (PaO2/FIO2
of 200) andbefore day 14 in patients with unre-solving ARDS. The
role of GC treat-ment in acute lung injury and less se-vere ARDS
(PaO2/FIO2 of 200) is lessclear.
Strength of Recommendations: 2B
Five randomized studies (n 518)have evaluated the role of GC
treatmentin patients with acute lung injury due
tocommunity-acquired pneumonia (7) andin patients with ARDS of
varied origins(104, 105, 111, 112). Varying doses (200750 mg of
hydrocortisone equivalents perday), dosing strategies
(infusion/bolus),and duration of therapy (732 days) wereused in
these studies. Due to the markeddifferences in study design and
patientcharacteristics, the cumulative summaryof these studies
should be interpretedwith some caution. Nevertheless, thesetrials
consistently reported that treat-ment was associated with
significant im-provement in PaO2/FIO2 (7, 104, 105, 111,112), a
significant reduction in markersof systemic inflammation (7, 104,
105,111, 112), duration of mechanical venti-lation (7, 104, 105,
111, 112), and inten-sive care unit length of stay (all with
pvalues of .05) (7, 104, 105, 111). Sub-group analysis (Fig. 4)
based on studiesthat investigated only treatment (methyl-
Figure 4. Effects of prolonged methylprednisolone treatment on
mechanical ventilationfree days at day 28. Reproduced with
permission from Meduri etal (114). WMD, weighted mean difference;
95% CI, 95% confidence interval.
1943Crit Care Med 2008 Vol. 36, No. 6
-
prednisolone) durations of 1 wk (n 295) (104, 105, 111) showed a
distinctincrease in the number of mechanicalventilationfree days
(weighted mean dif-ference, 5.59 days; 95% confidence inter-val,
3.497.68; p .001).
GC treatment in acute lung injuryARDS was not associated with
increasedrates of gastrointestinal bleeding or nos-ocomial
infections, and two of the studiesreported a reduction in the rate
of noso-comial infections, likely attributable tothe shorter
duration of mechanical ven-tilation (104, 105). In the two
random-ized trials (104, 111) that incorporatedinfection
surveillance, nosocomial infec-tions were frequently (56%)
identified inthe absence of fever. The combination ofGCs and
neuromuscular blocking agentssignificantly increases the risk for
pro-longed neuromuscular weakness (113).In the ARDS Network trial,
although bothgroups had similar exposure to paralyticagents (49%
vs. 42%; p .3), those ran-domized to methylprednisolone had ahigher
rate of serious events associatedwith myopathy or neuropathy (105).
Theother four trials did not report an in-creased rate of
neuromuscular complica-tions (7, 104, 111, 112).
A reduction in mortality was noted infour studies (7, 104, 111,
112). The ARDSNetwork trial reported increased 60-daymortality in
the subgroup randomized tomethylprednisolone after 14 days ofARDS
(105). This small subgroup (n 48), however, had large imbalances
inbaseline characteristics, and the mortal-ity difference lost
significance (p .57)when adjusting for these imbalances(114). The
two small clinical trials (n 68) (7, 111) showed marked reduction
inthe relative risk of death with GC therapy(2/39 [5%] vs. 11/31
[35%]; relative risk,0.15; 95% confidence interval, 0.04 0.59; p
.007). The three subsequentlypublished larger clinical trials (104,
105,112), when combined (n 400), achieved
a distinct reduction in the relative risk ofdeath (82/214 [38%]
vs. 98/186 [52.5%];relative risk, 0.78; 95% confidence inter-val,
0.640.96; p .02) (114). When an-alyzing the three trials
investigating cor-ticosteroids for durations of 1 wkinitiated
before day 14 of ARDS (n 245), mortality was equally
decreased(35/144 [24%] vs. 40/101 [40%]; relativerisk, 0.62; 95%
confidence interval, 0.430.90; p .01) (Fig. 5) (114).
The results of one randomized trial(111) indicate that 1
mgkg1day1
methylprednisolone, given as an infusionand tapered over the
course of 4 wks, isassociated with a favorable riskbenefitprofile
when secondary preventive mea-sures are implemented. These
measuresinclude 1) intensive infection surveil-lance, 2) avoidance
of paralytic agents,and 3) avoidance of rebound inflamma-tion with
premature discontinuation oftreatment that may lead to
physiologicdeterioration and reintubation. It shouldbe noted that
the premature and rapidtaper of corticosteroids in the ARDS
Net-work trial resulted in a deterioration ofthe PaO2/FIO2 and a
higher reintubationrate in the treatment group (105, 114).
Preliminary data suggest that GCsmay be of benefit in patients
with severecommunity-acquired pneumonia, liverfailure,
pancreatitis, patients undergoingcardiopulmonary bypass, and
duringweaning from mechanical ventilation (7,10, 11, 75, 96, 115).
The potential bene-fits of treatment with hydrocortisone inthese
patient subgroups and other criti-cally ill patients deserve
further investi-gation.
How to Treat
Recommendation 8: In patients withseptic shock, intravenous
hydrocorti-sone should be given in a dose of 200mg/day in four
divided doses or as abolus of 100 mg followed by a contin-uous
infusion at 10 mg/hr (240 mg/
day). The optimal initial dosing regi-men in patients with early
severe ARDSis 1 mgkg1day1 methylpred-nisolone as a continuous
infusion.
Strength of Recommendation: 1B
Recommendation 9: The optimal dura-tion of GC treatment in
patients withseptic shock and early ARDS is unclear.However, based
on published studiesand pathophysiological data, patientswith
septic shock should be treated for7 days before tapering,
assumingthat there is no recurrence of signs ofsepsis or shock.
Patients with earlyARDS should be treated for 14 daysbefore
tapering.
Strength of Recommendation: 2B
Recommendation 10: GC treatmentshould be tapered slowly and
notstopped abruptly.
Strength of Recommendation: 2B
Recommendation 11: Treatment withfludrocortisone (50 g orally
oncedaily) is considered optional.
Strength of Recommendation: 2B
Recommendation 12: Dexamethasoneis not recommended for the
treatmentof septic shock or ARDS.
Strength of Recommendation: 1B
Ideally, the dose of GC should be suf-ficient to down-regulate
the proinflam-matory response without causing im-mune-paresis and
interfering with woundhealing. Similarly, the duration of GCtherapy
should be guided by the durationof CIRCI and the associated
duration ofsystemic inflammation. The optimal doseand duration of
treatment with hydrocor-tisone/methylprednisolone remains to
bedetermined in well-controlled and well-powered studies. However,
the results ofpublished studies do allow us to make anumber of
recommendations. A number
Figure 5. Effects of prolonged glucocorticoid treatment
initiated before day 14 of acute lung injury-acute respiratory
distress syndrome on survival.Reproduced with permission from
Meduri et al (114). RR, relative risk; 95% CI, 95% confidence
interval.
1944 Crit Care Med 2008 Vol. 36, No. 6
-
of randomized controlled studies have in-vestigated the utility
of a high-dose,short-course treatment with corticoste-roids in
patients with ARDS and sepsis.Doses of methylprednisolone as high
as20 30 mg/kg body weight (10,000 to40,000 mg of hydrocortisone)
during thecourse of 24 hrs were investigated (116118). These
studies were unable to dem-onstrate an improved outcome, and
therewas a higher risk of complications in thepatients who received
high-dose cortico-steroids (116118). The literature there-fore does
not support the use of high-dose corticosteroids in critically
illpatients (except to prevent/treat rejectionin transplant
patients).
Myopathy and an increased risk of su-perinfections are more
common in pa-tients receiving 300 mg of hydrocorti-sone equivalents
per day (117, 118).Furthermore, while suppressing an exag-gerated
proinflammatory response, adose of 200300 mg of hydrocortisoneper
day does not seem to have immuno-suppressive effects (119, 120).
Based onthese data and the treatment protocolused in the French and
CORTICUS stud-ies, we recommend that patients withseptic shock be
treated with 50 mg ofhydrocortisone intravenously every 6 hrsor a
bolus of 100 mg, followed by a con-tinuous intravenous infusion at
10 mg/hr(340 mg the first day; 240 mg/day onsubsequent days). The
use of a continu-ous infusion of hydrocortisone has beenreported to
result in better glycemic con-trol, with less variability of blood
glucoseconcentration and a reduction in the staffworkload of
managing hyperglycemia(85, 121123). Treatment should con-tinue for7
days before tapering, assum-ing that there is no recurrence of
signs ofsepsis or shock. Hydrocortisone shouldbe tapered slowly and
not stoppedabruptly. The hydrocortisone dose shouldbe reduced every
23 days in small steps,unless there is clinical deterioration,which
would then require an increase inhydrocortisone dose. Abruptly
stoppinghydrocortisone will likely result in a re-bound of
proinflammatory mediators,with recurrence of the features of
shock(and tissue injury) (105, 119). In addi-tion, it should be
appreciated that GCtreatment itself results in down-regula-tion of
GR levels in most cells, potentiat-ing the rebound phenomenon with
theabrupt cessation of GC treatment (70).Currently, the optimal
dose and durationof therapy in patients with early severeARDS is 1
mgkg1day1 methylpred-
nisolone for 14 days, followed by a slowtaper while monitoring
indices of oxygen-ation.
Meduri et al. (124) demonstrated thatpersistent elevation of
inflammatory cy-tokines predicted a poor outcome in pa-tients with
ARDS. Recently, two longitu-dinal studies in patients with
severecommunity-acquired pneumonia foundhigh levels of circulating
inflammatorycytokines 3 wks after clinical resolutionof sepsis
(125, 126). The larger study,involving 1,886 patients, showed
hospitalmortality to be associated with highercirculating
inflammatory cytokine levelsand persistent elevation over time
(125).Furthermore, higher circulating inter-leukin-6 levels at
intensive care unit dis-charge were associated with increasedrisk
of death over 3 months (127). Thesedata support the concept of
immune dys-regulation in severe sepsis and ARDS (in-sufficient
corticosteroid activityCIRCI)and suggest that the duration of
treat-ment with GCs should be guided by theduration of elevation of
inflammatory cy-tokines (124). Further studies should ex-plore this
concept.
In the French study, patients in thetreatment group received
hydrocortisonetogether with fludrocortisone (50 gorally once
daily), whereas in the CORTI-CUS study patients received
hydrocorti-sone alone. It is unclear if the addition
offludrocortisone played a role in the favor-able outcome of the
French study. Thebenefit of the addition of fludrocortisonein
patients with septic shock is currentlybeing investigated in two
randomizedcontrolled trials comparing hydrocorti-sone alone vs.
hydrocortisone togetherwith fludrocortisone (www.ClinicalTrial.gov
NCT 00368381 and NCT00320099).Treatment with fludrocortisone is
consid-ered optional at this time.
Although treatment with dexametha-sone has been suggested in
patients withseptic shock until an ACTH stimulationtest is
performed, this approach can nolonger be endorsed. This
recommenda-tion is based on the fact that dexametha-sone leads to
immediate and prolongedsuppression of the HPA axis (limiting
thevalue of ACTH testing).
CONCLUSION
CIRCI is a complex and frequent dis-order of which our
understanding contin-ues to evolve. Although CIRCI may affecta
spectrum of critically ill patients, mostof the research has
focused on patients
with septic shock and ARDS. At this time,treatment with
moderate-dose corticoste-roids is recommended in patients
withseptic shock who have responded poorlyto volume resuscitation
and vasopressoragents. The consistent positive results re-ported in
patients with early severe ARDS(PaO2/FIO2 of 200) and
unresolvingARDS treated with GCs before day 14suggest that
treatment with moderate-dose GCs should be considered in
thesepatients. Tests of adrenal function are notroutinely required
in these patients. Therole of GCs in the management of pa-tients
with community-acquired pneu-monia, liver failure, pancreatitis,
thoseundergoing cardiac surgery, and othergroups of critically ill
patients requiresfurther investigation.
REFERENCES
1. Jurney TH, Cockrell JL Jr, Lindberg JS, etal: Spectrum of
serum cortisol response toACTH in ICU patients: Correlation with
de-gree of illness and mortality. Chest 1987;92:292295
2. Reincke M, Allolio B, Wurth G, et al:
Thehypothalamic-pituitary-adrenal axis in crit-ical illness:
Response to dexamethasone andcorticotropin-releasing hormone. J
ClinEndocrinol Metab 1993; 77:151156
3. Arlt W, Allolio B: Adrenal insufficiency.Lancet 2003;
361:18811893
4. Hinshaw LB, Beller BK, Chang AC, et
al:Corticosteroid/antibiotic treatment of adre-nalectomized dogs
challenged with lethal E.coli. Circ Shock 1985; 16:265277
5. Darlington DN, Chew G, Ha T, et al: Corti-costerone, but not
glucose, treatment en-ables fasted adrenalectomized rats to
sur-vive moderate hemorrhage. Endocrinology1990; 127:766772
6. Marik PE, Zaloga GP: Adrenal Insufficiencyin the critically
ill: A new look at an oldproblem. Chest 2002; 122:17841796
7. Confalonieri M, Urbino R, Potena A, et al:Hydrocortisone
infusion for severe commu-nity-acquired pneumonia: A
preliminaryrandomized study. Am J Respir Crit CareMed 2005;
171:242248
8. Dimopoulou I, Tsagarakis S: Hypothalamic-pituitary
dysfunction in critically ill pa-tients with traumatic and
nontraumaticbrain injury. Intensive Care Med 2005; 31:10201028
9. Tsai MH, Peng YS, Chen YC, et al: Adrenalinsufficiency in
patients with cirrhosis, se-vere sepsis and septic shock.
Hepatology2006; 43:673681
10. Eklund A, Leppaniemi A, Kemppainen E, etal: Vasodilatory
shock in severe acute pan-creatitis without sepsis: Is there any
placefor hydrocortisone treatment? Acta Anaes-thesiol Scand 2005;
49:379384
11. Fernandez J, Escorsell A, Zabalza M, et al:
1945Crit Care Med 2008 Vol. 36, No. 6
-
Adrenal insufficiency in patients with cir-rhosis and septic
shock: Effect of treatmentwith hydrocortisone on survival.
Hepatol-ogy 2006; 44:12881295
12. Definitions of ACCP evidence-based guide-lines, consensus
statements, and other re-views and projects. American College
ofChest Physicians, 2007. Available at:
http://www.chestnet.org/education/guidelines/definitions.php.
Accessed February 2, 2008
13. Fink A, Kosecoff J, Chassin M, et al: Con-sensus methods:
Characteristics and guide-lines for use. Am J Public Health
1984;74:979983
14. Williams PL, Webb C, Williams PL, et al:The Delphi
technique: A methodologicaldiscussion. J Adv Nurs 1994;
19:180186
15. Guyatt G, Gutterman D, Baumann MH, etal: Grading strength of
recommendationsand quality of evidence in clinical guide-lines:
Report from an American College ofChest Physicians Task Force.
Chest 2006;129:174181
16. Bone RC, Balk RA, Cerra FB, et al: Defini-tions for sepsis
and organ failure and guide-lines for the use of innovative
therapies insepsis: The ACCP/SCCM Consensus Confer-ence Committee.
American College of ChestPhysicians/Society of Critical Care
Medi-cine. Chest 1992; 101:16441655
17. Bernard GR, Artigas A, Brigham KL: TheAmerican-European
Consensus Conferenceon ARDS: Definitions, mechanisms, rele-vant
outcomes, and clinical trial coordina-tion. Am J Respir Crit Care
Med 1994; 149:818824
18. Cochran W: The combination of estimatesfrom different
experiments. Biometrics1954; 10:101129
19. Berlin JA, Laird NM, Sacks HS, et al: Acomparison of
statistical methods for com-bining event rates from clinical
trials. StatMed 1989; 8:141151
20. Higgins JP, Thompson SG: Quantifying het-erogeneity in a
meta-analysis. Stat Med2002; 21:15391558
21. Higgins JP, Thompson SG, Deeks JJ, et al:Measuring
inconsistency in meta-analyses.BMJ 2003; 327:557560
22. Chrousos GP: The hypothalamic-pituitary-adrenal axis and
immune-mediated inflam-mation. N Engl J Med 1995; 332:13511362
23. Carrasco GA, Van de Kar LD, Carrasco GA,et al:
Neuroendocrine pharmacology ofstress. Eur J Pharmacol 2003;
463:235272
24. Miller DB, OCallaghan JP, Miller DB, et al:Neuroendocrine
aspects of the response tostress. Metab Clin Exp 2002; 51:510
25. Dunn JF, Nisula BC, Rodbard D: Transportof steroid hormones:
Binding of 21 endog-enous steroids to both
testosterone-bindingglobulin and corticosteroid-binding globu-lin
in human plasma. J Clin EndocrinolMetab 1981; 53:5868
26. Mueller UW, Potter JM: Binding of cortisolto human albumin
and serum: The effect ofprotein concentration. Biochem
Pharmacol1981; 30:727733
27. Ho JT, Al-Musalhi H, Chapman MJ, et al:Septic shock and
sepsis: A comparison oftotal and free plasma cortisol levels. J
ClinEndocrinol Metab 2006; 91:105114
28. Dimopoulou I, Alevizopoulou P, Dafni U, etal:
Pituitary-adrenal response to humancorticotropin-releasing hormone
in criti-cally ill patients. Intensive Care Med 2007;33:454459
29. Arlt W, Stewart PM: Adrenal corticosteroidbiosynthesis,
metabolism, and action. En-docrinol Metab Clin North Am 2005;
34:293313
30. Borkowski AJ, Levin S, Delcroix C, et al:Blood cholesterol
and hydrocortisone pro-duction in man: Quantitative aspects of
theutilization of circulating cholesterol by theadrenals at rest
and under adrenocortico-tropin stimulation. J Clin Invest 1967;
46:797811
31. Yaguchi H, Tsutsumi K, Shimono K, et al:Involvement of high
density lipoprotein assubstrate cholesterol for steroidogenesis
bybovine adrenal fasciculo-reticularis cells.Life Sci 1998;
62:13871395
32. Acton S, Rigotti A, Landschultz KT, et al:Identification of
scavenger receptor SR-BIas a high density lipoprotein receptor.
Sci-ence 1996; 271:518520
33. Calco D, Gomez-Coronado D, LasuncionMA, et al: CLA-I is an
85-kD plasma mem-brane glycoprotein that acts as a high affin-ity
receptor for both native (HDL, LDL, andVLDL) and modified (OxLDL
and AcLDL)lipoproteins. Arterioscler Thromb Vasc Biol1997;
17:23412349
34. de la Llera-Moya M, Connelly MA, Drazul D,et al: Scavenger
receptor class B type 1affects cholesterol homeostasis by
magnify-ing cholesterol flux between cells and HDL.J Lipid Res
2001; 42:19691978
35. Liu J, Heikkila P, Meng QH, et al: Expres-sion of low and
high density lipoproteinreceptor genes in human adrenals. Eur
JEndocrinol 2000; 142:677682
36. Rhen T, Cidlowski JA: Antiinflammatory ac-tion of
glucocorticoidsnew mechanismsfor old drugs. N Engl J Med 2005;
353:17111723
37. Sinars CR, Cheung-Flynn J, Rimerman RA,et al: Structure of
the large FK506-bindingprotein FKBP51, an Hsp90-binding proteinand
a component of steroid receptor com-plexes. Proc Nat Acad Sci U S A
2003; 100:868873
38. Vermeer H, Hendriks-Stegeman BI, van derBB, et al:
Glucocorticoid-induced increasein lymphocytic FKBP51 messenger
ribonu-cleic acid expression: A potential marker forglucocorticoid
sensitivity, potency, and bio-availability. J Clin Endocrinol Metab
2003;88:277284
39. Galon J, Franchimont D, Hiroi N, et al:Gene profiling
reveals unknown enhancingand suppressive actions of
glucocorticoidson immune cells. FASEB J 2002; 16:6171
40. Auphan N, Didonato JA, Rosette C, et al:Immunosuppression by
glucocorticoids: In-
hibition of NF-kappa B activity through in-duction of I kappa B
synthesis. Science1995; 270:286290
41. Barnes PJ, Adcock I: Anti-inflammatory ac-tions of steroids:
Molecular mechanisms.Trends Pharmacol Sci 1993; 14:436441
42. Barnes PJ, Karin M: Nuclear factor-kB-Apivotal transcription
factor in chronic in-flammatory diseases. N Engl J Med
1997;336:10661071
43. Oelkers W: Adrenal insufficiency. N EnglJ Med 1996;
335:12061212
44. Collins S, Caron MG, Lefkowitz RJ: Beta-adrenergic receptors
in hamster smoothmuscle cells are transcriptionally regulatedby
glucocorticoids. J Biol Chem 1988; 263:90679070
45. Sakaue M, Hoffman BB: Glucocorticoids in-duce transcription
and expression of thealpha 1B adrenergic receptor gene in DTT1MF-2
smooth muscle cells. J Clin Invest1991; 88:385389
46. Limbourg FP, Huang Z, Plumier JC, et al:Rapid
nontranscriptional activation of en-dothelial nitric oxide synthase
mediates in-creased cerebral blood flow and stroke pro-tection by
corticosteroids. J Clin Invest2002; 110:17291738
47. Hafezi-Moghadam A, Simoncini T, Yang Z,et al: Acute
cardiovascular protective effectsof corticosteroids are mediated by
non-transcriptional activation of endothelial ni-tric oxide
synthase. Nat Med 2002;8:473479
48. Bobadilla NA, Tapia E, Jimenez F, et al:Dexamethasone
increases eNOS gene ex-pression and prevents renal
vasoconstric-tion induced by cyclosporin. Am J Physiol1999;
277:F464F471
49. Murata T, Hori M, Sakamoto K, et al: Dexa-methasone blocks
hypoxia-induced endo-thelial dysfunction in organ-cultured
pul-monary arteries. Am J Respir Crit Care Med2004; 170:647655
50. Cooper MD, Stewart PM: Corticosteroid in-sufficiency in
acutely ill patients. N EnglJ Med 2003; 348:727734
51. Widmer IE, Puder JJ, Konig C, et al: Corti-sol response in
relation to the severity ofstress and illness. J Clin Endocrinol
Metab2005; 90:45794586
52. Vogeser M, Groetzner J, Kupper C, et al:Free serum cortisol
during the postopera-tive acute phase response determined
byequilibrium dialysis liquid chromatogra-phy-tandem mass
spectrometry. Clin ChemLab Med 2003; 41:146151
53. Annane D, Maxime V, Ibrahim F, et al: Di-agnosis of adrenal
insufficiency in severesepsis and septic shock. Am J Respir
CritCare Med 2006; 174:13191326
54. Vella A, Nippoldt TB, Morris JC III: Adrenalhemorrhage: A
25-year experience at theMayo Clinic. Mayo Clin Proc 2001;
76:161168
55. Kovacs KA, Lam YM, Pater JL: Bilateralmassive adrenal
hemorrhage: Assessment of
1946 Crit Care Med 2008 Vol. 36, No. 6
-
putative risk factors by the case-controlmethod. Medicine 2001;
80:4553
56. Deeb SA, Rosenberg RB, Wilkerson RJ, et al:Adrenal
hemorrhage in a pediatric burn pa-tient. Burns 2001; 27:658661
57. Weyrich P, Balletshofer B, Hoeft S, et al:Acute
adrenocortical insufficiency due toheparin-induced thrombocytopenia
withsubsequent bilateral haemorrhagic infarc-tion of the adrenal
glands. Vasa 2001; 30:285288
58. Espinosa G, Santos E, Cervera R, et al: Ad-renal involvement
in the antiphospholipidsyndrome: Clinical and immunologic
char-acteristics of 86 patients. Medicine 2003;82:106118
59. Arnason JA, Graziano FM: Adrenal insuffi-ciency in the
antiphospholipid antibodysyndrome. Semin Arthritis Rheum
1995;25:109116
60. Briegel J, Scheelling G, Haller M, et al: Acomparison of the
adrenocortical responseduring septic shock and after complete
re-covery. Intensive Care Med 1996; 22:894899
61. Meduri GU, Yates CR: Systemic inflamma-tion-associated
glucocorticoid resistanceand outcome of ARDS. Ann N Y Acad Sci2004;
1024:2453
62. Ito K, Ito M, Elliott WM, et al: Decreasedhistone
deacetylase activity in chronic ob-structive pulmonary disease. N
Engl J Med2005; 352:19671976
63. Hew M, Bhavsar P, Torrego A, et al: Relativecorticosteroid
insensitivity of peripheralblood mononuclear cells in severe
asthma.Am J Respir Crit Care Med 2006; 174:134141
64. Chikanza IC, Kozaci DL: Corticosteroid re-sistance in
rheumatoid arthritis: Molecularand cellular perspectives.
Rheumatology2004; 43:13371345
65. Chikanza IC, Kozaci D, Chernajovsky Y: Themolecular and
cellular basis of corticoste-roid resistance. J Endocrinol 2003;
179:301310
66. Pariante CM, Pearce BD, Pisell TL, et al: Theproinflammatory
cytokine, interleukin-1alpha, reduces glucocorticoid
receptortranslocation and function. Endocrinology1999;
140:43594366
67. Liu LY, Sun B, Tian Y, et al: Changes ofpulmonary
glucocorticoid receptor andphospholipase A2 in sheep with acute
lunginjury after high dose endotoxin infusion.Am Rev Respir Dis
1993; 148:878881
68. Kino T, Ichijo T, Chrousos GP: FLASH in-teracts with p160
coactivator subtypes anddifferentially suppresses transcriptional
ac-tivity of steroid hormone receptors. J Ste-roid Biochem Mol Biol
2004; 92:357363
69. Meduri GU, Muthiah MP, Carratu P, et al:Nuclear
factor-kappaB- and glucocorticoidreceptor alphamediated mechanisms
inthe regulation of systemic and pulmonaryinflammation during
sepsis and acute respi-ratory distress syndrome: Evidence for
in-flammation-induced target tissue resistance
to glucocorticoids. Neuroimmunomodula-tion 2005; 12:321338
70. Schaaf MJ, Cidlowski JA: Molecular mecha-nisms of
glucocorticoid action and resis-tance. J Steroid Biochem Mol Biol
2002;83:3748
71. Creed TJ, Probert CS: Review article: Ste-roid resistance in
inflammatory bowel dis-ease. Mechanisms and therapeutic
strate-gies. Aliment Pharmacol Ther 2007; 25:111122
72. Lewis-Tuffin LJ, Cidlowski JA: The physiol-ogy of human
glucocorticoid receptor beta(hGRbeta) and glucocorticoid
resistance.Ann N Y Acad Sci 2006; 1069:19
73. Xu D, Buehner A, Xu J, et al: A polymorphicglucocorticoid
receptor in a mouse popula-tion may explain inherited altered
stressresponse and increased anxiety-type behav-iors. FASEB J 2006;
20:24142416
74. De Iudicibus S, Stocco G, Martelossi S, et al:Association of
BclI polymorphism of theglucocorticoid receptor gene locus with
re-sponse to glucocorticoids in inflammatorybowel disease. Gut
2007; 56:13191320
75. Marik PE, Gayowski T, Starzl TE, et al: Thehepatoadrenal
syndrome: A common yetunrecognized clinical condition. Crit CareMed
2005; 33:12541259
76. Murphy JF, Purdue GF, Hunt JL: Acute ad-renal insufficiency
in the patient withburns. J Burn Care Rehabil 1993; 14:155157
77. Winter W, Kamolz L, Donner A, et al: Hy-drocortisone
improved haemodynamics andfluid requirement in surviving but not
non-surviving of severely burned patients. Burns2003; 29:717720
78. De Waele JJ, Hoste EA, Baert D, et al: Rel-ative adrenal
insufficiency in patients withsevere acute pancreatitis. Intensive
CareMed 2007; 33:17541760
79. Gloor B, Uhl W, Tcholakov O, et al: Hydro-cortisone
treatment of early SIRS in acuteexperimental pancreatitis. Dig Dis
Sci 2001;46:21542161
80. Meyer NJ, Hall JB: Relative adrenal insuffi-ciency in the
ICU: Can we at least make thediagnosis? Am J Respir Crit Care Med
2006;174:12821283
81. Marik PE: The adrenal exhaustion syn-drome in patients with
liver disease. Inten-sive Care Med 2006; 32:275280
82. Van der Voort PHJ, Koopman M: Follow upcorticotropin
stimulating tests during ICUstay. Intensive Care Med 2004;
30(Suppl1):S61
83. Beishuizen A, Vermes I, Hylkema BS, et al:Relative
eosinophilia and functional adrenalinsufficiency in critically ill
patients. Letter.Lancet 1999; 353:16751676
84. Annane D, Sebille V, Troche G, et al: A3-level prognostic
classification in septicshock based on cortisol levels and
cortisolresponse to corticotropin. JAMA 2000; 283:10381045
85. Arafah BM: Hypothalamic-pituitary adrenalfunction during
critical illness: Limitations
of current assessment methods. J Clin En-docrinol Metab 2006;
91:37253745
86. le Roux CW, Chapman GA, Kong WM, et al:Free cortisol index
is better than serumtotal cortisol in determining
hypothalamic-pituitary-adrenal status in patients under-going
surgery. J Clin Endocrinol Metab2003; 88:20452048
87. Hamrahian AH, Oseni TS, Arafah BM: Mea-surement of serum
free cortisol in criticallyill patients. N Engl J Med 2004;
350:16291638
88. Vogeser M, Briegel J, Zachoval R: Dialyzablefree cortisol
after stimulation with Syn-acthen. Clin Biochem 2002; 35:539543
89. Salgado DR, Verdeal JC, Rocco JR: Adrenalfunction testing in
patients with septicshock. Crit Care 2006; 10:R149
90. Vogeser M, Mohnle P, Briegel J: Free serumcortisol:
Quantification applying equilib-rium dialysis or ultrafiltration
and an auto-mated immunoassay system. Clin ChemLab Med 2007;
45:521525
91. Vogeser M, Briegel J, Jacob K: Determina-tion of serum
cortisol by isotope-dilutionliquid-chromatography electrospray
ioniza-tion tandem mass spectrometry with onlineextraction. Clin
Chem Lab Med 2001; 39:944947
92. Cohen J, Ward G, Prins J, et al: Variability ofcortisol
assays can confound the diagnosisof adrenal insufficiency in the
critically illpopulation. Intensive Care Med 2006; 32:19011905
93. Rothwell PM, Udwadia ZF, Lawler PG: Cor-tisol response to
corticotropin and survivalin septic shock. Lancet 1991;
337:582583
94. Bollaert PE, Fieux F, Charpentier C, et al:Baseline cortisol
levels, cortisol response tocorticotropin, and prognosis in late
septicshock. Shock 2003; 19:1315
95. Annane D, Sebille V, Charpentier C, et al:Effect of
treatment with low doses of hydro-cortisone and fludrocortisone on
mortalityin patients with septic shock. JAMA 2002;288:862871
96. Huang CJ, Lin HC: Association between ad-renal insufficiency
and ventilator weaning.Am J Respir Crit Care Med 2006;
173:276280
97. Loisa P, Uusaro A, Ruokonen E: A singleadrenocorticotropic
hormone stimulationtest does not reveal adrenal insufficiency
inseptic shock. Anesth Analg 2005; 101:17921798
98. Bouachour G, Roy PM, Guiraud MP, et al:The repetitive short
corticotropin stimula-tion test in patients with septic shock.
AnnIntern Med 1995; 123:962963
99. Marik PE, Zaloga GP: Adrenal insufficiencyduring septic
shock. Crit Care Med 2003;31:141145
100. Chawla K, Kupfer Y, Tessler S: Hydrocorti-sone reverses
refractory septic shock. CritCare Med 1999; 27(Suppl):A33
101. Oppert M, Schindler R, Husuang C, et al:Low-dose
hydrocortisone improves shockreversal and reduces cytokine levels
in early
1947Crit Care Med 2008 Vol. 36, No. 6
-
hyperdynamic septic shock. Crit Care Med2005; 33:24572464
102. Bollaert PE, Charpentier C, Levy B, et al:Reversal of late
septic shock with supra-physiologic doses of hydrocortisone.
CritCare Med 1998; 26:645650
103. Briegel J, Forst H, Haller M, et al: Stressdoses of
hydrocortisone reverse hyperdy-namic septic shock: A prospective,
random-ized, double-blind, single-center study. CritCare Med 1999;
27:723732
104. Meduri GU, Golden E, Freire AX, et al:Methylprednisolone
infusion in patientswith early severe ARDS: Results of a
ran-domized trial. Chest 2007; 131:954963
105. The Acute Respiratory Distress SyndromeNetwork: Efficacy
and safety of corticoste-roids for persistent acute respiratory
dis-tress syndrome. N Engl J Med 2006; 354:16711684
106. Sprung CL, Annane D, Keh D, et al: Hydro-cortisone therapy
for patients with septicshock. N Engl J Med 2008; 358:111124
107. Annane D, Briegel J, Keh D, et al: Clinicalequipoise
remains for issues of adrenocor-ticotropic hormone administration,
cortisoltesting, and therapeutic use of hydrocorti-sone. Crit Care
Med 2003; 31:22502251
108. Annane D, Vignon P, Renault A, et al: Nor-epinephrine plus
dobutamine versus epi-nephrine alone for management of septicshock:
A randomised trial. Lancet 2007;370:676684
109. Russell JA, Walley KR, Singer J, et al: Vaso-pressin versus
norepinephrine infusion inpatients with septic shock. N Engl J
Med2008; 358:877887
110. Schelling G, Stoll C, Kapfhammer HP, et al:The effect of
stress doses of hydrocortisoneduring septic shock on posttraumatic
stressdisorder and health-related quality of life insurvivors. Crit
Care Med 1999; 27:26782683
111. Meduri GU, Headley S, Golden E, et al:Effect of prolonged
methylprednisolonetherapy in unresolving acute respiratorydistress
syndrome: A randomized controlledtrial. JAMA 1998; 280:159165
112. Annane D, Sebille V, Bellissant E: Effect oflow doses of
corticosteroids in septic shockpatients with or without early acute
respi-ratory distress syndrome. Crit Care Med2006; 34:2230
113. Leatherman JW, Fluegle WL, David WS, etal: Muscle weakness
in mechanically venti-lated patients with severe asthma. Am JRespir
Crit Care Med 1996; 153:16861690
114. Meduri GU, Marik PE, Chrousos GP, et al:Steroid treatment
in ARDS: A critical ap-praisal of the ARDS network trial and
therecent literature. Intensive Care Med 2008;34:6169
115. Halonen J, Halonen P, Jarvinen O, et al:Corticosteroids for
the prevention of atrialfibrillation after cardiac surgery: A
random-ized controlled trial. JAMA 2007; 14:15621567
116. Keh D, Sprung CL: Use of corticosteroidtherapy in patients
with sepsis and septicshock: An evidence-based review. Crit CareMed
2004; 32:S527S533
117. Annane D, Bellissant E, Bollaert PE, et al:Corticosteroids
for severe sepsis and septicshock: A systematic review and
meta-analysis. BMJ 2004; 329:480489
118. Minneci PC, Deans KJ, Banks SM, et al:Meta-analysis: The
effect of steroids on sur-vival and shock during sepsis depends
onthe dose. Ann Intern Med 2004; 141:4756
119. Keh D, Boehnke T, Weber-Cartens S, et al:Immunologic and
hemodynamic effects oflow-dose hydrocortisone in septic shock:A
double-blind, randomized, placebo-controlled, crossover study. Am J
RespirCrit Care Med 2003; 167:512520
120. Kaufmann I, Briegel J, Schliephake F, et al:
Stress doses of hydrocortisone in septicshock: Beneficial
effects on opsonization-dependent neutrophil functions.
IntensiveCare Med 2008; 34:344349
121. Loisa P, Parviainen I, Tenhunen J, et al:Effect of mode of
hydrocortisone adminis-tration on glycemic control in patients
withseptic shock: A prospective randomizedtrial. Crit Care 2007;
11:R21
122. Weber-Carstens S, Deja M, Bercker S, et al:Impact of bolus
application of low-dose hy-drocortisone on glycemic control in
septicshock patients. Intensive Care Med 2007;33:730733
123. Egi M, Bellomo R, Stachowski E, et al: Vari-ability of
blood glucose concentration andshort-term mortality in critically
ill pa-tients. Anesthesiology 2006; 105:244252
124. Meduri GU, Headley S, Kohler G, et al:Persistent elevation
of inflammatory cyto-kines predicts a poor outcome in ARDS:Plasma
IL-1 beta and IL-6 levels are consis-tent and efficient predictors
of outcomeover time. Chest 1995; 107:10621073
125. Kellum JA, Kong L, Fink MP, et al: Under-standing the
inflammatory cytokine re-sponse in pneumonia and sepsis: Results
ofthe Genetic and Inflammatory Markers ofSepsis (GenIMS) Study.
Arch Intern Med2007; 167:16551663
126. Lekkou A, Karakantza M, Mouzaki A, et al:Cytokine
production and monocyteHLA-DR expression as predictors of out-come
for patients with community-ac-quired severe infections. Clin Diagn
LabImmunol 2004; 11:161167
127. Yende S, Knong L, Weussfeld J: Inflamma-tory markers prior
to hospital dischargepredict subsequent mortality after commu-nity
acquired pneumonia: Proceedings ofthe American Thoracic Society. Am
J RespirCrit Care Med 2006; 3:A836
1948 Crit Care Med 2008 Vol. 36, No. 6
-
Appendix 1. Modified Grades of Recommendation Assessment,
Development, and Evaluation (GRADE) system for Grading the strength
of the evidence (15)
Grade of recommendation/description
Benefits vs. Risk and burdens Methodological quality
ofsupporting evidence
Implications
1A: Strong recommendation,high quality evidence
Benefits clearly outweigh riskand burdens or vise versa
RCTs without important limitationsor overwhelming evidence
fromobservational studies
Strong recommendation canapply to most patients in
mostcircumstances withoutreservation
1B: Strong recommendation,moderate quality evidence
Benefits clearly outweigh riskand burdens or vise versa
RCTs with important limitations orexceptionally strong
evidencefrom observational studies
Strong recommendation canapply to most patients in
mostcircumstances withoutreservation
1C: Strong recommendation,low quality or very low-quality
evidence
Benefits clearly outweigh riskand burdens or vise versa
Observational studies or case series Strong recommendation but
maychange when higher qualityevidence becomes available
2A: Weak recommendation,high quality evidence
Benefits closely balanced withrisk and burden
RCTs without important limitationsor overwhelming evidence
fromobservational studies
Weak recommendation, bestaction may differ depending
oncircumstances or patients orsocietal values
2B: Weak recommendation,moderate quality evidence
Benefits closely balanced withrisk and burden
RCTs with important limitations orexceptionally strong
evidencefrom observational studies
Weak recommendation, bestaction may differ depending
oncircumstances or patients orsocietal values
2C: Weak recommendation,low quality or very lowquality
evidence
Uncertainty in the estimates ofbenefits, risks, and
burdens;benefits risk and burdenmay be closely balanced
Observational studies or case series Very weak
recommendations;other alternatives may beequally reasonable
Reproduced with permission from Chest. RCT, randomized
controlled trial.
1949Crit Care Med 2008 Vol. 36, No. 6