Sunteți pe pagina 1din 61

Corticosteroids for treating severe sepsis and septic shoc k (Review)

Annane D, Bellissant E, Bollaert PE, Briegel J, Keh D, Kupfer Y

D, Bellissant E, Bollaert PE, Briegel J, Keh D, Kupfer Y This is a reprint of

This is a reprint of a Cochrane review, prepared and maintaine d by The Cochrane Collaboration and published in The Cochrane Library 2010, Issue 12

Library 2010, Issue 12 http://www.thecochranelibrary.com Corticosteroids for treating severe sepsis and septic shock
Library 2010, Issue 12 http://www.thecochranelibrary.com Corticosteroids for treating severe sepsis and septic shock

T A B L E O F C O N T E N T S

HEADER

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

1

. PLAIN LANGUAGE SUMMARY

ABSTRACT

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

1

2

SUMMARY OF FINDINGS FOR THE MAIN COMPARISON

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

2

BACKGROUND

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

5

OBJECTIVES

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

5

METHODS

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

6

RESULTS

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

8

Figure 1.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

9

Figure 2.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

12

Figure 3.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

15

Figure 4.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

16

Figure 5.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

17

Figure 6.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

18

Figure 7.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

19

DISCUSSION

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

21

AUTHORS’ CONCLUSIONS

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

22

ACKNOWLEDGEMENTS

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

23

REFERENCES

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

23

CHARACTERISTICS OF STUDIES

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

27

DATA AND ANALYSES

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

56

WHAT’S NEW

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

57

. CONTRIBUTIONS OF AUTHORS

HISTORY

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

58

58

DECLARATIONS OF INTEREST

 

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

59

SOURCES OF SUPPORT

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

59

NOTES

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

59

INDEX TERMS

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

59

[Intervention Review]

Corticosteroids for treating severe sepsis and septic shoc k

Djillali Annane 1 , Eric Bellissant 2 , Pierre Edouard Bollaert 3 , Josef Briegel 4 , Didier Keh 5 , Yizhak Kupfer 6

1 Critical Care Department, Hôpital Raymond Poincaré, Assistance Publique - Hôpitaux de Paris, Garches, France. 2 Centre d’Investigation Clinique INSERM 0203, Hôpital Pontchaillou, Rennes, France. 3 Intensive Care Unit, Hôpital Central, Nancy, France. 4 Klinik für Anaesthesiologie, Klinikum der Universitaet, Munich, Germany. 5 University Clinic of Anesthesiology and Intensive Care Medicine CCM/CVK, Charité-Campus Virchow Clinic, Charité Universitätsmedizin Berlin, Berlin, Germany. 6 Division of Pulmonary and Critical Care Medicine, Maimonides Medical Center, New Yor k, USA

Contact address: Djillali Annane, Critical Care Department, Hôpital Raymond Poincaré, Assistance Publique - Hôpitaux de Paris, 104. Boulevard Raymond Poincaré, Garches, Ile de France, 92380, France. djillali.annane@rpc.ap-hop-paris.fr .

Editorial group: Cochrane Anaesthesia Group. Publication status and date: New search for studies and content updated (no change to conclusions), published in Issue 12, 2010. Review content assessed as up-to-date: 31 October 2010.

Citation: Annane D, Bellissant E, Bollaert PE, Briegel J, Keh D, Kupfer Y. Corticosteroids for treating severe sepsis and septic shock. Cochrane Database of Systematic Reviews 2004, Issue 1. Art. No.: CD002243. DOI: 10.1002/14651858.CD002243.pub2.

Copyright © 2010 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.

A B S T R A C T

Background

Sepsis may be complicated by impaired corticosteroid production. Giving corticosteroids may benefit patients.

Objectives

To examine the effects of corticosteroids on death at one month in sepsis.

Search strategy

We searched CENTRAL ( The Cochrane Library Issue 3, 2009), MEDLINE (October 2009), EMBASE (October 2009), LILACS (October 2009), reference lists of articles, and also contacted trial authors.

Selection criteria

We included randomized and quasi-randomized controlled trial s of corticosteroids versus placebo or supportive treatment in severe sepsis and septic shock.

Data collection and analysis

All review authors agreed the eligibility of trials. One review author extracted data, which was checked by the other review authors and the primary author of the paper whenever possible. We obtained some missing data from the trial authors. We assessed the methodological quality of the trials.

Main results

We identified 25 trials, of which 20 (17 randomized and three quasi-randomized trials) could be pooled in a meta-analysis.

Corticosteroids did not change 28-day mortality (20 trials, n = 2138, relative risk (RR) 0.87, 95% confidence interval (CI) 0.74 to 1.01; random-effects model). There was significant heterogeneity that was partly related to the dosing strategy. Treatment with a long course of low dose corticosteroids significantly reduced 28-day mortality (RR 0.84, 95% CI 0.72 to 0.97; P = 0.02), increased the proportion of shock reversal by day seven (six trials, n = 965, RR 1.35, 95% CI 1.16 to 1.57; random-effects model) and day 28 (six trials, n =

952, RR 1.12, 95% CI 1.02 to 1.23), reduced the sepsis-related or gan failure assessment (SOFA) score by day seven (five trials, n = 916, RR -1.47, (95% CI -2.01 to -0.92), and survivors’ length of stay in the intensive care unit (eight trials, n= 622, RR -4.49, 95% CI -7.04 to -1.94), without inducing gastroduodenal bleeding (13 trials, n = 1594, RR 11.12, 95% CI 0.81 to 1.53), superinfection (14 trials, n = 1917, RR 1.01, 95% CI 0.82 to 1.25), or neuromuscul ar weakness (three trials, n = 811, RR 0.63, 95% CI 0.12 to 3.35). Corticosteroid increased the risk of hyperglycaemia (nine trials, n = 1434, RR 1.16, 95% 1.07 to 1.25) and hypernatraemia (three trials, n= 805, RR 1.61, 95% CI 1.26 to 2.06).

Authors’ conclusions

Overall, corticosteroids did not change mortality in severe se psis and septic shock. A long course of low dose corticosteroids reduced 28-day mortality without inducing major complications; metabolic disorders were increased.

P L A I N L A N G U A G E S U M M A R Y

A long course (five days or more) of a low dose of corticosteroids may be considered as an adjunct therapy in patients with septic shock.

Septic shock is the most severe form of infection. It may also interfere with the production of corticosteroids, a key hormone f or host defence against infection. This review showed that corticoster oids did not impact on mortality overall. However, the trials conducted after the scientific community agreed upon a consensual definition of sepsis, in 1992, suggested that a low dose of corticosteroids for five days or more improved the survival of patients in septic shock without causing harm They did develop some metabolic disorders. Trials performed before 1992 showed no benefit from a short course of high dose corticosteroids.

S U M M A R Y O F F I N D I N
S U M M A R Y O F F I N D I N G S F O R T H E M A I N C O M P A R I S O N [Explanation]
long course low dose corticosteroids compared to placebo or standard therapy for severe sepsis and septic shock
Patient or population: patients with severe sepsis and septic shock
Settings: in patient
Intervention: long course low dose corticosteroids
Comparison: placebo or standard therapy
Outcomes
Illustrative comparative risks* (95% CI)
Relative effect
No of Participants
(studies)
Quality of the evidence
(GRADE)
Comments
(95% CI)
Assumed risk
Corresponding risk
placebo
or
therapy
standard long course low dose
corticosteroids
All-cause mortality
Follow-up: 28 days
Medium risk population
RR 0.84
(0.72 to 0.97) 1
1228
⊕⊕⊕
(12 studies)
moderate
2, 3
45
per 100
38 per 100
(32 to 44)
Meta-analysis of all
doses/duration of corti-
costeroids showed non-
significant decrease in
mortality, RR 0.87 (0.74,
1.01)
Intensive care unit mor-
tality
Medium risk population
RR 0.81
(0.63 to 1.04)
1082
⊕⊕⊕
(8 studies)
moderate
2
37
per 100
30 per 100
(23 to 38)
Number of patients with
shock reversal
Follow-up: 7 days
Medium risk population
RR 1.35
(1.16 to 1.57)
965
⊕⊕⊕
(6 studies)
moderate
4
50
per 100
68 per 100
(58 to 79)
for treating severe sepsis and septic shock (Review) 3Corticosteroids
Copyright © 2010 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.

for treating severe sepsis and septic shock (Review) 4Corticosteroids

Copyright © 2010 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.

SOFA score Scale from: 0 to 24 (most abnormal). Follow-up: 7 days The mean sofa
SOFA score
Scale from: 0 to 24 (most
abnormal).
Follow-up: 7 days
The mean sofa score The mean SOFA score
916
⊕⊕⊕
ranged across control
groups from
2 to 9.5
in the intervention groups
was
(5 studies)
moderate
2
1.47
lower
(2.01 to 0.92 lower)
Length of ICU stay
The mean length of icu
The mean Length of ICU
622
⊕⊕⊕
stay ranged across con- stay in the intervention
(8 studies)
moderate
2
trol groups from
groups was
11
to 38 days
4.49
lower
(7.04 to 1.94 lower)
Number
of patients with
Medium risk population
RR 1.01
(0.82 to 1.25)
1917
⊕⊕⊕
adverse events
(14 studies)
moderate
2, 4
superinfections
16
per 100
16 per 100
(13 to 20)
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the
assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Metaregression analysis also showed evidence for a dose and duration response: longer duration and low dose are associated with better treatment response

2 Different definitions of sepsis, concomitant therapies, and baseline risk of death; and some inconsistency with large trials showing non-significant results.

3 Funnel plot showed asymmetry which may be consistent with publication bias but not downgraded.

4 Studies from both short and long dose corticosteroids.

B A C K G R O U N D

Sepsis is present when a site of infection is apparent in a patient and there is evidence of body-wide, systemic inflammation. Systemic inflammation is usually defined by two or more criteria. These are fever or low body temperature, an increase or decrease in white blood cells, an increase in the heart rate, and rapid breathing ( ACCP/SCCM 1992 ; Bone 1991 ). Septic shock is when sepsis is combined with organ dysfunction or hypoperfusion and a fall in systemic blood pressure that does not improve even when heal th staff give intravenous fluids. The current incidence of severe sepsis in industrialised countries ranges from 50 to 100 cases per 100 ,000 population, with a short term mortality of 20% to 50% ( Annane 2003 ; Finfer 2004 ; Martin 2003 ; Padkin 2003 ; The EPISPESIS group 2004 ). People usually die from hypotension or progressive multiple organ failure (Annane 2005 ; Parrillo 1993 ).

Researchers have explored the biological mechanisms of shock for potential interventions. Corticosteroids have been a partic- ular focus because of their influence on the immune response. In sepsis, the hypothalamic-pituitary gland hormonal pathway to the adrenal glands stimulates corticosteroid production ( Chrousos 1995 ). These hormones affect inflammation through the white blood cells, cytokines (proteins that influence the immune re- sponse), and nitric oxide production. In septic shock, cytokine s may suppress the cortisol response to adrenocorticotropin hor - mone ( Hotta 1986 ; Jaattela 1991 ). This causes poor adrenal ac- tivity in almost half of patients (Annane 2000 ; Lipiner 2007 ; Rothwell 1991 ) and body tissues possibly become resistant to cor- ticosteroids (Meduri 1998a ) through fewer corticosteroid recep- tors or receptors with lower affinity (Barnes 1995 ; Huang 1987 ; Molijn 1995 ). Early studies demonstrated that a pharmacological dose of corticosteroids prolonged survival in septic animals (Fabian 1982 ). More recent studies in rodents demonstrated that lower dose of corticosteroids, for example 0.1 mg/kg of dexamethasone, also improved haemodynamic and organ function, modulated th e inflammatory response favourably, and prolonged survival (Heller 2003 ; Tsao 2004 ; Vachharajani 2006 ). LIkewise, in healthy vol- unteers challenged with endotoxin, a low dose of corticosteroids, for example 10 mg of prednisolone, blocked the release of pro-in- flammatory cytokines, prevented endothelial cell and neutrophil activation, and inhibited the acute phase response without al tering the coagulation and fibrinolysis balance ( de Kruif 2007 ). Studies in patients with septic shock showed that a short course of corti- costeroids may result in a rebound in the systemic inflammator y response (Briegel 1994 ; Keh 2003 ). In addition, it is now recog- nized that increased pro-inflammatory cytokine release can be sus- tained for more than a week in patients with severe sepsis (Kellum 2007 ). Finally, a recent randomized controlled study in 82 patients with septic shock compared a seven-day to a three-day course of 200 mg/day hydrocortisone (Huh 2007 ). This study found lower mortality in patients treated for seven days (32% versus 24%). For these reasons, we would anticipate that corticosteroid treatment is of benefit in human septic shock and that a different dose or

duration of steroid may differentially affect the patient re sponse to treatment.

Initial research used high doses of corticosteroids, usually given as a single bolus, in an attempt to block any potential burst in pr o- inflammatory cytokines. A systematic review and meta-analysis of trials of corticosteroids in sepsis or septic shock included 10 r an- domized and placebo controlled trials with a total of 1329 patients

( Lefering 1995 ). The result showed no significant difference in ab- solute mortality rate, but there was significant heterogeneity across the trials. The various subgroup analyses included whether or not the infection was related to Gram-negative bacteria, and compared low and normal doses of corticosteroids. Another systematic re- view and meta-analysis of trials of corticosteroids in sepsis or sep- tic shock included nine randomized and placebo controlled trial s accounting for 1232 patients (Cronin 1995 ). The authors did not include one unpublished study ( Rogers 1970 ) since they could not clarify whether the treatment was allocated in a random order or not. This systematic review showed no significant difference in relative risk of death, but there was significant heterogeneity across the trials. These systematic reviews did not show any significant increase in gastrointestinal bleeding or superinfection associated with corticosteroids.

As a result of these reviews, most clinicians will not recommend the use of high doses of corticosteroids in severe sepsis. Howe ver, the reviews covered a period from 1966 to 1993 and did not ex- clude a possible benefit of a lower dose ( 300 mg of hydrocor- tisone or equivalent per day) and longer duration ( 5 days) of treatment, as observed in more recent randomized, placebo con- trolled trials (Annane 2002 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Cicarelli 2007 ; Confalonieri 2005 ; Huh 2007 ; Keh 2003 ; Mikami 2007 ; Oppert 2005 ; Rinaldi 2006 ; Tandan 2005 ; Yildiz 2002 ). In these trials, the major expectations for corticosteroids were a reduction in shock duration and improvement in survival with no increase in adverse events. A recent multicentre trial con- firmed that corticosteroids reduced shock duration in septic shock patients but failed to demonstrate any survival benefit and sug- gested an increased risk of superinfections ( Sprung 2008 ). There- after, recent international guidelines suggest using corticosteroids only in septic shock patients who are poorly responsive to fluid replacement and vasopressors ( Dellinger 2008 ).

We therefore aim to systematically review the effects of corticos- teroids in patients with severe sepsis and septic shock.

O B J E C T I V E S

To examine the effects of corticosteroids on death at one month in patients with severe sepsis and septic shock, and to examine whether dose and duration of corticosteroids influence patient responses to this treatment.

M E T H O D S

Criteria for considering studies for this review

Types of studies

We included randomized or quasi-randomized controlled trials

with or without blinding.

Types of participants

We included children and adults with septic shock defined by the following criteria ( ACCP/SCCM 1992 ).

1. Documented infection defined as culture or Gram stain of

blood, sputum, urine, or normally sterile body fluid that is

positive for a pathogenic microorganism; or a focus of infection identified by visual inspection (e.g. ruptured bowel with the presence of free air or bowel contents in the abdomen found at the time of surgery; wound with purulent drainage).

2. At least two symptoms of a systemic inflammatory response

syndrome, such as fever (body temperature > 38 °C) or hypothermia (< 36 °C), tachycardia (> 90 beats per minute), tachypnoea (> 20 breaths per minute), or hyperventilation (arterial carbon dioxide tension (PaCO 2 ) < 32 mm Hg), and

abnormal white blood cell count (> 12,000 cells/ml or < 4000 cells/ml), or more than 10% immature band of neutrophils.

3. At least one sign of organ dysfunction, that is metabolic

acidosis, arterial hypoxaemia (arterial oxygen tension [PaO 2 ]:

fractional inspired oxygen [FiO 2 ] < 250 mm Hg), oliguria (< 30 ml/h for at least 3 h), coagulopathy, or encephalopathy.

4.

Hypotension (persisting systolic arterial pressure below 90

mm

Hg) that is refractory to fluid resuscitation and which needs

vasopressor support, that is more than 5 µg/kg of body weight per minute of dopamine or any dose of either epinephrine or norepinephrine). We included data from trials of sepsis, sepsis syndrome, or acute respiratory distress syndrome if separate data were available for pa-

tients with septic shock, or when contact with the authors resul ted in provision of the data.

Types of interventions

Intervention

Intravenous treatment with any type of corticosteroid prepar ation (for example cortisone, hydrocortisone, methylprednisolone , be- tamethasone, or dexamethasone).

Low dose corticosteroid was defined by a total dose per day of 300

mg or less of hydrocortisone (or equivalent); otherwise it woul d be considered to be a high dose of corticosteroid. A long course for the intervention was defined by a full dose treatment duration

of five days or more; otherwise it was considered as short course treatment.

Control

Standard therapy (which may have included antibiotics, fluid re- placement, inotropic or vasopressor therapy, mechanical ventila- tion, renal replacement therapy) or placebo.

Types of outcome measures

Primary outcomes

The 28-day all-cause mortality

Indeed, this was the primary outcome measure in most of the randomized controlled trials on sepsis that have been conducte d in the past 15 years (Annane 2009b ). Most of the studies performed before 1992 looked at 14-day or hospital mortality rates. We used these data to compute the pooled analysis on 28-day mortality, unless actual 28-day mortality rates could be obtained from studies authors.

Secondary outcomes

Intensive care unit mortality

Hospital mortality

Number of patients with shock reversal (as defined by stable

haemodynamic status for at least 24 hours after withdrawal of

vasopressor therapy) at day 7 and at day 28

Number of organ dysfunction-free days (as defined in individual studies)

Length of stay in the intensive care unit (for all patients and for survivors only)

Length of hospital stay (for all patients and for survivors

only)

Adverse events (i.e. gastrointestinal bleeding and

superinfection or any other adverse effects or complications of corticosteroid treatment)

Search methods for identification of studies

We attempted to identify all relevant studies regardless of language or publication status (published, unpublished, in press, and in progress).

Electronic searches

We originally searched the Cochrane Infectious Diseases Group Trials Register for relevant trials (to August 2003) using th e search terms: ’sepsis’; and ’septic shock’. Full details of the Cochr ane In- fectious Diseases Group’s methods and the journals they hand- search are published in The Cochrane Library in the section on Cochrane Review Groups.

In this updated version, we searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2009, Issue 3) using the search terms: ’sepsis’; ’septic shock’; ’steroids’; and ’corticosteroids’ (for detailed search strategy see Appendix 1). We also searched (up to Octobre 2009) MEDLINE, EMBASE and LILACS using the topic search terms in combination with the search strategy for identifying trials developed by The Cochr ane Collaboration ( Higgins 2005 ). (For detailed search strategy see Appendix 2 (MEDLINE), Appendix 3 (EMBASE) and Appendix 4 (LILACS))

Searching other resources

We checked the reference lists of all trials identified by the above methods, and contacted authors to identify any additional pub- lished or unpublished data. We also searched the proceedings of the annual meetings of major critical care medicine symposia, th at is Society of Critical Care Medicine, American Thoracic Society, the International Symposium on Intensive Care and Emergency Medicine, the American College of Chest Physicians, and Euro- pean Society of Intensive Care Medicine (1998 to 2009). Finally, we searched for ongoing randomized controlled trial s (October2009) in the metaRegister of Controlled Trials using the search terms: ’septic shock’; ’sepsis’; steroids’; ’corticosteroids’; ’adrenal cortex hormones’; and ’glucocorticoids’ ( www.controlled- trials.com/mrct/active ).

Data collection and analysis

Study selection

All review authors checked the titles and abstracts identified with the search strategy. All authors examined in full any trial th at po- tentially met the inclusion criteria. Whenever possible, one author was blinded to the journal in which the paper was published, th e authors, the institution, and the magnitude and direction of the re- sults. Five authors (Djillali Annane, Pierre Edouard Bollae rt, Josef Briegel, Didier Keh, and Yizhak Kupfer) evaluated all the tr ials. We decided which trials fitted the inclusion criteria and graded their methodological quality. We resolved any disagreement be- tween the five authors by discussion with the sixth author (Eric Bellissant) until a consensus was reached. One author (Djillal i An- nane) contacted study authors for clarification, where necessary.

Assessment of methodological quality

We graded generation of allocation sequence and allocation con- cealment as adequate, unclear, or inadequate according to Juni 2001 . We described the method for blinding as double blind (method described and placebo(s) or dummy technique meant nei- ther the participant nor the care provider or assessor knew which treatment was given); single blind (participant or the care provider

or assessor was aware of the treatment given); and open (all par ties were aware of the treatment). We described, whenever possible , who among patients, care-givers, data collectors, outcome asses- sors, and data analysts remained blinded ( Devereaux 2001 ). We considered loss to follow up as adequate (90% or more of the par- ticipants randomized were included in the analysis), unclear (not reported), or inadequate (< 90% participants randomized into the trial were included in the analysis). We resolved any disagree ments between the five authors by discussion with the sixth author (Er ic Bellissant) until a consensus was reached. We contacted study au- thors for clarification, where necessary.

Data extraction

DA drew up a standard data extraction form and four other au- thors (PEB, JB, DK, YK) amended and validated the design of the form prior to data abstraction. Four authors (DA, PEB, JB, DK) independently extracted data and DA systematically contacted the authors of the trials to provide missing data where possible . DA entered (DA secretary independently reentered all data to achieve a double entry) the data into the computer and five authors (EB, PEB, JB, DK, YK) checked the accuracy of data that were entered using the original articles.

Data analyses

1. We performed intention-to-treat analyses. We performed al l

statistical calculations using RevMan 5 or STATA/IC version

10.0 (Stata Corp, College Station, Texas) as appropriate.

2. We calculated a weighted treatment effect across trials. We

expressed the results as RRs with 95% confidence intervals (CI) for dichotomous outcomes, and mean differences (MD, 95% CI) for continuous outcomes. We considered methods based on the random-effects model for all analyses. Indeed, we suspecte d

that we would observe heterogeneity across the studies as the y were conducted over a wide period of time (almost half a century

between the first and the last trials) and the rationale on which the studies were designed varied greatly over time, with mar ked differences in treatment strategies between studies conducte d prior to and after the early 90s.

3. To identify potential sources of heterogeneity (when I 2

20%), we a priori sought to conduct a subgroup analysis based on ’dose and duration’, that is long course (at least five days) of low dose (< 300 mg/day of hydrocortisone or equivalent). This subgroup analysis allowed the evaluation of a strategy base d on new developments in the understanding of the role of corticosteroids in host response to sepsis, tested in trials

performed after 1992 (Annane 2002 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Cicarelli 2007 ; Confalonieri 2005 ; Huh 2007 ; Keh 2003 ; Meduri 1998b ; Mikami 2007 ; Oppert 2005 ; Rinaldi 2006 ; Sprung 2008 ; Tandan 2005 ; Yildiz 2002 ). Older trials used a short course (one to four bolus doses within 24 hours) of high dose corticosteroids (more than 300 mg of

hydrocortisone or equivalent) as an anti-inflammatory approach, while the most recent trials used low dose corticosteroids for a long period of time (at least five days) as a hormonal replacement strategy. To further explore the putative interaction betwe en steroid dose and duration and the magnitude of effect, we considered performing a meta-regression analysis using 28-day all-cause mortality as the dependent variable, and dosage and duration of corticosteroids as predictors. The meta-regression analyses were computed using STATA/IC version 10.0 (Stata Corp, College Station, Texas). We also a priori tested the interaction between baseline severity of illness and the

magnitude of effect in a meta-regression analysis using mortality rates in controls as the predictors.

4. We conducted sensitivity analyses for generation of

allocation sequence, concealment of allocation, and blinding.

5. We sought evidence of publication bias using the funnel

plot method. We computed, using STATA/IC version 10.0 (Stata Corp, College Station, Texas), a contour enhance funnel plot ( Peters 2008 ). This graphical analysis used the log of the RR and the standard error of the RR. Contours illustrating the statistical significance of the study effect estimates were plotted

from a two-tailed test.

6. We assessed the validity of the subgroup analysis (dose and

duration) on the basis of the following criteria: (1) subgroup comparisons within studies rather than between studies; (2) hypothesis preceded the analysis; (3) one of very few hypotheses; (4) large and consistent difference across studies; and (5) exter nal evidence to support the results ( Guyatt 2008b ). When subgroup analyses met these criteria and were found to be statistically significant, we applied Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria to evaluate the quality of evidence ( Guyatt 2008a ).

R E S U L T S

Description of studies

See: Characteristics of included studies ; Characteristics of excluded studies ; Characteristics of ongoing studies . Our search results are detailed in Figure 1 .

Figure 1. Search flow diagram

Figure 1. Search flow diagram Corticosteroids for treating severe sepsis and septic shock (Review) Copyright ©

The search strategy yielded 35 randomized controlled trials that evaluated corticosteroids in severe sepsis or septic shock. Of these, we excluded 10 (see Characteristics of excluded studies ). We in- cluded the remaining 25 trials and have described them below (se e Characteristics of included studies ). We identified nine additional randomized trials of prolonge d treatment with a low dose of corticosteroids from trials registries (but these nine studies were not included in the analysis, see Characteristics of ongoing studies ). One of these trials was to com- pare hydrocortisone alone to the combination of hydrocortisone and fludrocortisone but this trial did not start due to lack of fund- ing ( NCT00368381 2008 ). One of the trials has a 2 x 2 factorial design to also allow evaluation of the effects of the recombinant activated protein C ( NCT00625209 2008 ). The trials included only adults except one trial evaluating treatment with hydrocorti- sone in children with severe sepsis (NCT00732277 2008 ).

Source of information

In addition to the extracted data from the publications, we ob- tained unpublished information from 15 trials by contacting the primary authors ( Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Cicarelli 2007 ; Confalonieri 2005 ; Keh 2003 ; Meduri 2007 ; Oppert 2005 ; Rinaldi 2006 ; Sprung 1984 ; Sprung 2008 ; Tandan 2005 ; Yildiz 2002 ) (Appendix 5). In one case, contact with the authors did not allow provision of additional information (Luce 1988 ). For one trial we could use only published data.

Trial centres

Seven trials were multicentre trials (that is more than two centres) ( Annane 2002 ; Annane 2010 ; Bone 1987 ; CSG 1963 ; Confalonieri 2005 ; Sprung 2008 ; VASSCSG 1987 ).

Age of participants

One study enrolled both children and adults ( CSG 1963 ). Another trial included only children (Slusher 1996 ). All remaining trials included only adults.

Description of participants

Eight trials included both severe sepsis and septic shock patients ( Bone 1987 ; Klastersky 1971 ; Lucas 1984 ; Luce 1988 ; Slusher 1996 ; VASSCSG 1987 ; Wagner 1955 ; Yildiz 2002 ) and only one study provided separate data for septic shock (Bone 1987 ). Two trials included patients with severe sepsis ( Confalonieri 2005 ; Rinaldi 2006 ).The remaining trials focused only on septic shock patients treated by a vasopressor. Two trials included only septic shock patients with adrenal insufficiency as defined by a cortisol increment of less than 9 µg/dl after a corticotropin bolus ( Huh

2007 ; Tandan 2005 ). In five trials, a short corticotropin test was systematically performed at baseline ( Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Oppert 2005 ; Sprung 2008 ).

Control

Two studies did not use a placebo and corticosteroid therapy was compared to standard therapy, that is antibiotics, fluid resuscita- tion, and vasopressor when needed ( Lucas 1984 ; Rinaldi 2006 ). In one study a placebo was used in one centre only (Sprung 1984 ). In another trial ( Wagner 1955 ), a placebo was available only at the end of the study. Thereafter, in the first 85 patients, corticosteroid therapy was compared to standard therapy, that is antibiotics, fluid resuscitation, and vasopressor when needed, and to a placebo in the last 28 patients. In one trial that compared hydrocortisone to hydrocortisone plus fludrocortisone did not use a placebo of flu- drocortisone for technical reasons ( Annane 2010 ). In the remain- ing trials, the corticosteroid therapy was compared to a placebo.

Corticosteroid dose and treatment course

Twelve trials tested the effects of a long course (> 5 days) with low dose hydrocortisone (Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Confalonieri 2005 ; Huh 2007 ; Oppert 2005 ; Rinaldi 2006 ; Sprung 2008 ; Tandan 2005 ; Wagner 1955 ), one trial tested prednisolone ( Yildiz 2002 ), and one trial tested the effects of dexamethasone (Cicarelli 2007 ). In six trials ( Annane 2002 ; Bollaert 1998 ; Huh 2007 ; Oppert 2005 ; Sprung 2008 ; Tandan 2005 ) the effects of corticosteroids were analysed in patients with adrenal insufficiency. In one trial ( Huh 2007 ) the authors compared hydrocortisone 50 mg intravenously ever y six hours when given for three days versus seven days. Anothe r trial compared a seven-day treatment with hydrocortisone to a seven-day treatment with the combination of hydrocortisone pl us fludrocortisone ( Annane 2010 ). Seven trials tested the effects of a short course with a large dose of methylprednisolone ( Bone 1987 ; Luce 1988 ; Schumer 1976 ; Sprung 1984 ; VASSCSG 1987 ), of dexamethasone ( Lucas 1984 ; Schumer 1976 ; Sprung 1984 ), or of betamethasone ( Klastersky 1971 ).

Outcomes

Thirteen trials explicitly reported 28-day mortality rates (Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Cicarelli 2007 ; Confalonieri 2005 ; Huh 2007 ; Klastersky 1971 ; Oppert 2005 ; Sprung 2008 ; Tandan 2005 ; Yildiz 2002 ). For three trials contact with the primary author of the paper allowed record- ing of 28-day mortality rates ( Meduri 2007 ; Rinaldi 2006 ; Sprung

1984 ). Three trials reported only 14-day mortality rates ( Bone 1987 ; Lucas 1984 ; VASSCSG 1987 ). Five trials reported only hospital mortality rates (CSG 1963 ; Luce 1988 ; Schumer 1976 ; Slusher 1996 ; Wagner 1955 ;). One trial reported only two deaths among 113 patients during hospital stay ( Wagner 1955 ) and one trial did not report mortality rates ( Keh 2003 ). Seven trials explicitly reported the intensive care unit (ICU) mor- tality rate (Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Briegel 1999 ; Confalonieri 2005 ; Meduri 2007 ; Sprung 2008 ) and the primary author of two additional trials provided this outcome ( Chawla 1999 ; Rinaldi 2006 ). Hospital mortality rates were avail- able for 16 trials (Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Confalonieri 2005 ; Klastersky 1971 ; Lucas 1984 ; Luce 1988 ; Meduri 2007 ; Rinaldi 2006 ; Schumer 1976 ; Sprung 1984 ; Sprung 2008 ; Wagner 1955 ; Yildiz 2002 ). Eight trials reported the rate of shock reversal at day seven ( Annane 2002 ; Bone 1987 ; Bollaert 1998 ; Briegel 1999 ; Chawla

1999 ; Oppert 2005 ; Sprung 1984 ; Sprung 2008 ) and eight tri- als ( Annane 2002 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Huh 2007 ; Sprung 2008 ; Tandan 2005 ) the rate of shock rever- sal at day 28. Five trials reported the numbers of organ dysfunc- tion at seven days, that is SOFA score ( Annane 2002 ; Cicarelli 2007 ; Oppert 2005 ; Rinaldi 2006 ; Sprung 2008 ). The length of ICU stay was reported in 10 trials (Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Confalonieri 2005 ; Huh 2007 ; Meduri 2007 ; Rinaldi 2006 ; Sprung 2008 ) and the length of hospital stay in nine trials (Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Chawla 1999 ; Confalonieri 2005 ; Meduri 2007 ; Slusher 1996 ; Sprung 2008 ; Yildiz 2002 ).

Risk of bias in included studies

The detailed methodological quality of individual trials are re- ported in the risk of bias tables, Figure 2, and in Appendix 6.

Figure 2. Methodological quality summary: review authors’ judgements about each methodological quality item for each included study.

each methodological quality item for each included study. Corticosteroids for treating severe sepsis and septic shock

Randomization

In three trials, we considered that the randomization (method of generation of allocation sequence) was inappropriate to minimize selection bias, that is based on hospital numbers in two trial s ( Lucas 1984 ; Wagner 1955 ) and on a card system in one study ( Schumer 1976 ); the method for generation of allocation sequence was judged inadequate. In four trials the method was unclear ( CSG 1963 ; Huh 2007 ; Klastersky 1971 ; Slusher 1996 ) and was deemed adequate in the remaining trials. We judged the method for allocation concealment to be adequate in all but eight trial s. In two trials assignment of treatment was based on hospital numbers ( Lucas 1984 ; Wagner 1955 ) and in one trial on unsealed envelopes ( Schumer 1976 ). In one trial, in one of the two participating centres the investigators enrolling patients could have fore seen the upcoming assignment as the local ethical committee refused to accept blind allocation ( Sprung 1984 ). In four trials the method for allocation concealment was not reported ( CSG 1963 ; Huh 2007 ; Klastersky 1971 ; Slusher 1996 ).

Blinding

In five trials, blinding was inadequate ( Annane 2010 ; Lucas 1984 ; Rinaldi 2006 ; Sprung 1984 ; Wagner 1955 ). In one study ( Wagner 1955 ), blinding of treatment administration and of outcome as- sessment was used only at the end of the study (for the last 28 patients among 113 patients). Two trials used open labelled treat- ments (Annane 2010 ; Rinaldi 2006 ;). In another trial (Lucas 1984 ) the authors stated that “steroids were administered in a non blinded manner, because a previous unpublished double-blind study of steroid therapy for patients caused uniform deferve scence in the steroid-treated patients, thereby permitting an accurate pre- diction of steroid supplementation by the nursing personnel”. In the fifth trial (Sprung 1984 ), the local ethical committee of one of the two centres did not permit double-blind allocation and ad- ministration of treatment. Then, blinding was not possible for 40 out of the 59 patients included in the trial. The remaining trials were deemed as appropriately double blinded. In three additional trials the method for ensuring blinding was not reported (CSG 1963 ; Klastersky 1971 ; Schumer 1976 ).

Withdrawal

Eleven trials ( Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Briegel 1999 ; Keh 2003 ; Lucas 1984 ; Meduri 2007 ; Oppert 2005 ; Rinaldi 2006 ; Sprung 2008 ; VASSCSG 1987 ) explicitly provided the number of, and reasons for, withdrawals or losses to follow up. In one trial only 500 out of the 800 expected patients were recruited, mainly due to a low recruitment rate after loss of equipoise among investigators (Sprung 2008 ).

Intention-to-treat analysis and adherence to the protocol

Thirteen trials explicitly reported the use of intention-to-treat analysis (as the primary analysis) and the number of, and rea- sons for, non-adherence to the protocol ( Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Bone 1987 ; Briegel 1999 ; Chawla 1999 ; Confalonieri 2005 ; Keh 2003 ; Meduri 2007 ; Oppert 2005 ; Rinaldi 2006 ; Sprung 2008 ; VASSCSG 1987 ). One trial reported only the use of intention-to-treat analysis (Luce 1988 ). The re- maining trials provided no information about these criteria. How- ever, the number of analysed participants matched the number of randomized participants except for five of all the trials. In one trial, 191 participants were randomized in the placebo group and 190 were analysed for the mortality outcome (Bone 1987 ). In two trials ( Annane 2002 ; Sprung 2008 ), one patient withdrew his or her consent and 499/500 and 299/300 randomized patients were analysed, respectively. In two trials, contact with the primar y au- thor allowed us to get information for patients who were dropped out from the analysis ( Oppert 2005 ; Rinaldi 2006 ). In the first study, seven randomized patients (five corticosteroid group and two placebo group) were not analysed ( Oppert 2005 ). Four of these patients (two corticosteroid group and two placebo group) were discharged alive from the ICU and then were definitely lost to follow up. The three remaining patients (corticosteroid group) died, two before receiving hydrocortisone and the last at study day 17. In the second study, 12 out of 52 patients were dropped out of the study, six in the control group and six in the corticosteroid group (Rinaldi 2006 ). Nine patients (four in the control group) were excluded as they developed renal failure. Two of the contr ol patients died in the ICU at day five and day seven, respectively. Three of the corticosteroid-treated patients died, at day five, six, and 28 respectively. Three other patients (two control group) were excluded as they developed septic shock. They all died at days three, five, and six, respectively.

Explicit definition of septic shock

Twelve trials provided an explicit definition of severe sepsis or sep- tic shock (as defined in the method section of this review) ( Annane 2002 ; Annane 2010 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Cicarelli 2007 ; Huh 2007 ; Keh 2003 ; Oppert 2005 ; Sprung 1984 ; Sprung 2008 ; Tandan 2005 ). Nine trials provided a definition of severe sepsis and on septic shock without referring to the nee d for vasopressor agents ( Bone 1987 ; Klastersky 1971 ; Lucas 1984 ; Luce 1988 ; Rinaldi 2006 ; Schumer 1976 ; Slusher 1996 ; VASSCSG 1987 ; Yildiz 2002 ). The definition for severe sepsis or septic shock was not explicitly given in two studies (CSG 1963 ; Wagner 1955 ). Two trials explicitly defined severe sepsis due to community ac- quired pneumonia (Confalonieri 2005 ; Mikami 2007 ). In one

trial on early acute respiratory distress syndrome (ARDS), contact with the primary author confirmed that explicit definitions of se- vere sepsis and septic shock were used (Meduri 2007 ).

Effects of interventions

See: Summary of findings for the main comparison Long course low dose corticosteroids compared to placebo or standard therapy

for severe sepsis and septic shock We did not pool the data from two trials that included chil- dren ( CSG 1963 ; Slusher 1996 ), one crossover trial (Keh 2003 ), one trial that compared two durations of hydrocortisone treat- ment (Huh 2007 ), and one trial that compared hydrocortisone to the combination of hydrocortisone plus fludrocortisone ( Annane

2010 ).

28-day all-cause mortality

Data for 28-day mortality were available for 16 trials; among these two trials had no corticosteroid-free arm. In addition, we used data on 14-day mortality (n = 3 trials) or hospital mortality (n = 3 trials). Thus, we computed data from 20 trials (17 randomized and three quasi-randomized trials) that accounted for 2384 par - ticipants. There were 416/1220 participants in the treated gr oup that died by day 28 compared to 424/1164 participants in the placebo group. There was significant heterogeneity in the results (Chi 2 = 34.18, P = 0.02, I 2 = 44%). The RR of dying at 28 days was 0.87 (95% CI 0.74 to 1.01, P = 0.07; random-effects model) (Analysis 1.1). We analysed separately the 14 studies for which 28-day mortality was available and the three studies reporting only 14-day mor tal- ity (Analysis 1.2). There were 291/718 deaths at 28 days in the corticosteroid-treated group and 293/654 deaths in the control group (RR 0.88, 95% CI 0.77 to 1.01, P = 0.06; random-effects model) with moderate heterogeneity in the results (Chi 2 = 15.00, P = 0.31, I 2 = 13%). In the studies reporting only 14-day mortality rates, there were 93/326 deaths in the corticosteroid-treated group and 77/326 deaths in the control group (RR 1.21, 95% CI 0.94

to 1.58, P = 0.14; random-effects model), without heterogeneity in the results (Chi 2 = 1.36, P = 0.51, I 2 = 0%). Heterogeneity across trials may have been explained by diffe rent therapeutic regimens and different populations. Subgroup analysis on the 12 trials that tested a long course (5 days) of low dose corticosteroids ( 300 mg hydrocortisone or equivalent) showed less heterogeneity across the trials (Chi 2 = 12.89, P = 0.30, I 2 = 15%) and a RR of dying at 28 days of 0.84 (95% CI 0.72 to 0.97, P = 0.02) in favour of the corticosteroid group (Analysis 1.3). We also conducted a sensitivity analysis by excluding the trial on community acquired pneumonia (Confalonieri 2005 ). This

analysis still showed a RR of 0.87 (95% CI 0.77 to 0.98, P = 0.02) in favour of the corticosteroid group and almost no heterogene ity (Chi 2 = 10.09, P = 0.43; I² = 1%). Subgroup analyses on the trials that tested a short course of high dose corticosteroids showed significant heterogeneity across the trials (Chi 2 = 18.63; P = 0.005, I 2 = 68%) and a RR of dying at 28 days of 0.94 (95% CI 0.69 to 1.30; random-effects model) (Analysis 1.3). Differences in methodological quality across the trials may have also accounted for the observed heterogeneity in the results. Sub- group analyses based on the trials with an adequate method for generation of the allocation sequence showed a RR of dying at 28 days of 0.92 (95% CI 0.99 to 1.07) (Analysis 1.4). Similarly, subgroup analyses based on studies with adequate allocation con- cealment showed a RR of dying at 28 days of 0.90 (95% CI 0.76 to 1.06); and subgroup analyses on double-blind trials showed a RR of dying at 28 days of 0.91 (95% CI 0.79 to 1.05) (Analysis

1.4).

One trial of a large dose of corticosteroids was a statistical outlier and was excluded from the meta-regression analysis ( Schumer 1976 ). Meta-regression analysis confirmed the positive interaction between dose and duration of corticosteroid treatment and sur vival with: a lower RR of dying with prolonged duration treatment at a low dose (P = 0.01) (Figure 3), lower daily doses (P = 0.02)

( Figure 4), and lower cumulative doses (P = 0.02) ( Figure 5 ). Meta- regression showed less interaction of mortality rate in the control group with corticosteroid effects (P = 0.06).

Figure 3. Figure displays the relative risk of death plotted against the duration of glucocorticoid treatment at full dose, given in hours. Individual studies are depicted by a bubble, which size indicates the weight of that particular study in the meta-regression analysis. There were significant association between the relative risk of death and the time of treatment at full dose (P = 0.015).

of death and the time of treatment at full dose (P = 0.015). Corticosteroids for treating

Figure 4. Figure displays the relative risk of death plotted against the dose of glucocorticoid treatment at study day 1and expressed in hydrocortisone equivalent. Ind ividual studies are depicted by a bubble, which size indicates the weight of that particular study in the meta-regression analysis. There were significant association between the relative risk of death and the time of treatment a t full dose (P = 0.022).

death and the time of treatment a t full dose (P = 0.022). Corticosteroids for treating

Figure 5. Figure displays the relative risk of death plotted against the cumulated dose of glucocorticoid treatment , expressed in hydrocortisone equivalent. Individual studies are depicted by a bubble, which size indicates the weight of that particular study in the meta-regression analysis. There were significant association between the relative risk of death and the time of treatment a t full dose (P = 0.018).

death and the time of treatment a t full dose (P = 0.018). Subgroup analysis of

Subgroup analysis of patients with adrenal insufficiency showed no heterogeneity in the results. There were 135/288 deaths in the treated group and 145/275 in the placebo group. The RR of dying was 0.88 (95% CI 0.76 to 1.02) (Analysis 1.4). Funnel plot analysis, including all trials, suggested some asymme- try ( Figure 6 ). Contour enhanced funnel plot analysis including trials of a long course of low dose corticosteroids also suggested significant asymmetry (P = 0.01) ( Figure 7).

Figure 6. Funnel plot of comparison: 1 Steroids versus control, outcome: 1.1 28-day all-cause mortality.

versus control, outcome: 1.1 28-day all-cause mortality. Corticosteroids for treating severe sepsis and septic shock

Figure 7. Scatter plot of the intervention effect estimates (RR) from the 12 individual studies on prolonged treatment with low-dose corticosteroids against each stud y precision (standard error of log[RR]). Lines define areas according to significance levels, i.e. 1%, 5% and 10%). Because the precision of the intervention effect estimate is proportional to the study’s size, effect estima tes from small studies scatter more widely at the bottom of the graph, with the spread narrowing among larger studies.

the graph, with the spread narrowing among larger studies. In one trial comparing hydrocortisone alone to

In one trial comparing hydrocortisone alone to hydrocortisone

plus fludrocortisone, the hazard ratio of death was of 0.94 (95%

CI 0.73 to 1.21) ( Annane 2010 ).

Intensive care unit (ICU) mortality

Data were available in eight trials, accounting for 1082 partici- pants. All these trials investigated a long course of a low dose of corticosteroids. There were 226/558 participants in the treated group and 239/524 participants in the placebo group that died in the ICU. There was some heterogeneity in the results (Chi 2 = 12.86, P = 0.08, I 2 = 46%). The RR of dying in the ICU was 0.81 (95% CI 0.63 to 1.04; random-effects model) (Analysis 1.6).

Hospital mortality

We could extract data for hospital mortality from 15 trials that accounted for 1672 participants. There were 344/866 participants

in the treated group compared to 355/806 in the control group that died in hospital. There was significant heterogeneity in the results (Chi 2 = 27.95, P = 0.01, I 2 = 50%). The RR of dying in hospital was 0.83 (95% CI 0.68 to 1.00; random-effects model) (Analysis 1.7). Subgroup analysis on 10 trials that tested a long course ( 5 days) of low dose corticosteroids ( 300 mg hydrocortisone or equiva- lent) showed less heterogeneity in the results and a RR of dying in hospital of 0.85 (95% CI 0.72 to 1.00; random-effects model) in favour of the corticosteroids group. Subgroup analyses on the tri- als that tested a short course of high dose corticosteroids showed a significant heterogeneity across the trials (Chi 2 = 16.81, P = 0.002, I 2 = 76%) and a RR of dying at 28 days of 0.84 (95% CI 0.52 to 1.36, P = 0.47; random-effects model) (Analysis 1.7). Differences in methodological quality across the trials may al so have accounted for the observed heterogeneity in the results. Sub-

group analyses based on trials with an adequate method for ge n- erating the allocation sequence showed a RR of dying in hospital

dysfunction. Five additional studies reported the changes of an organ dysfunction score (sepsis-related organ failure assessment,

of

0.90 (95% CI 0.77 to 1.06), with less heterogeneity across the

SOFA) within the first week from randomization (Annane 2002 ;

studies (Analysis 1.7). Similarly, subgroup analyses based on stud-

Cicarelli 2007 ; Oppert 2005 ; Rinaldi 2006 ; Sprung 2008 ). The

ies

with adequate allocation concealment showed a RR of dying

weighted mean difference in the SOFA score at day seven was -1.47

in

hospital of 0.89 (95% CI 0.76 to 1.05), and subgroup analyses

(95% CI -2.01 to -0.92, P < 0.00001) in favour of corticosteroids.

on

blinded trials showed a RR of dying in hospital of 0.89 (95%

There was no heterogeneity across the studies (Chi² = 4.09, P =

CI

0.75 to 1.05) (Analysis 1.7).

0.39; I² = 2%) (Analysis 1.8).

In

one trial comparing hydrocortisone alone to hydrocortisone

plus fludrocortisone, the RR of death was 0.94 (95% CI 0.77 to

1.14) ( Annane 2010 ).

Shock reversal at day seven

We could extract data from eight trials that accounted for 1268 participants. There were 418/658 participants in the treated group and 315/610 in the placebo group that had shock reversed at day seven. There was significant heterogeneity in the results (Chi 2 = 21.48, P = 0.003, I 2 = 67%). The RR of having shock reversed

at day seven was 1.29 (95% CI 1.06 to 1.58, P = 0.01; random-

effects model) in favour of the corticosteroids group (Analysis 1.8). Heterogeneity in the results could be explained by difference s in treatment strategies used in the various trials. Two trials evalu- ated one or two boluses of high dose corticosteroids ( Bone 1987 ;

Sprung 1984 ) while the six remaining trials all studied a replace- ment therapy with low dose (200 to 300 mg) hydrocortisone for more than five days. When analysing the six trials with a simi-

lar therapeutic strategy, there was less heterogeneity in the results.

Then, there were 308/485 in the treated group and 226/480 pa- tients in the placebo group that had shock reversed at day seven.

The RR of having shock reversed was 1.35 (95% CI 1.16 to 1.57,

P

= 0.0001) in favour of the corticosteroids group (Analysis 1.8).

In

one crossover trial, hydrocortisone was given for three day s at

a

dose of 240 mg per day (Keh 2003 ). Although this trial could not provide information on shock reversal at day seven, it showed that at day three fewer hydrocortisone patients required nore- pinephrine treatment than placebo-treated patients (6/20 ver sus 14/20, P = 0.025).

Shock reversal at day 28

We could extract data from six trials accounting for 952 partici- pants. There were 322/481 participants in the treated group that had shock reversed at day 28 and 276/471 in the placebo group. There was little heterogeneity in the results. The RR of having

shock reversed was 1.12 (95% CI 1.02 to 1.23, P = 0.02) in favour

of the corticosteroids group (Analysis 1.8).

Number of organ dysfunction-free days

This outcome was not reported in the various trials. In one study ( Briegel 1999 ) the corticosteroids treatment was associated with

a non-significant (P = 0.18) trend to earlier resolution of organ

Length of stay in the intensive care unit (ICU)

In eight trials (n = 1083) (Annane 2002 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Confalonieri 2005 ; Meduri 2007 ; Rinaldi 2006 ; Sprung 2008 ), the weighted mean difference for the length of stay in the ICU was -3.11 (-5.79 to -0.43, P = 0.02) with some heterogeneity across the studies (Chi² = 9.72, P = 0.21, I² = 28%). We could extract data from these trials on 622 intensive care unit survivors. The mean difference for the length of stay in the ICU was -4.49 days (95% CI -7.04 to -1.94, P = 0.0006). There was no heterogeneity across the studies (Chi² = 2.78, P = 0.90; I² = 0%) (Analysis 1.10).

Length of hospital stay

In seven trials (n = 1031) (Annane 2002 ; Bollaert 1998 ; Chawla 1999 ; Confalonieri 2005 ; Meduri 2007 ; Sprung 2008 ; Yildiz 2002 ) we could extract data for all patients and for 552 hospital survivors. There was no evidence for a difference between the two groups (WMD -2.54, 95% CI -7.93 to 2.47) (Analysis 1.11).

Adverse events

Gastroduodenal bleeding

We could extract data from 13 trials. There were 65/827 partici- pants in the treated group and 56/767 in the placebo group that had an episode of gastroduodenal bleeding. There was no hete ro- geneity in the results. The RR of having gastroduodenal blee ding was 1.12 (95% CI 0.81 to 1.53) (Analysis 1.12).

Superinfection

We could extract data from 14 trials. There were 184/983 partici- pants in the treated group and 170/934 participants in the placebo group that had an episode of nosocomial infection. There was no heterogeneity in the results. The RR of having superinfection was 1.01 (95% CI 0.82 to 1.25) (Analysis 1.12).

Hyperglycaemia

The number of participants who presented with hyperglycaemia was reported for nine trials. There was no heterogeneity in the results. The RR of having hyperglycaemia was 1.16 (95% CI 1.07 to 1.25) (Analysis 1.12). One trial comparing tight glucose control to standard care did not find any benefit in normalizing blood glucose levels in corticos- teroid-treated septic shock patients ( Annane 2010 ).

Hypernatraemia

The number of participants who presented with hypernatraemia was reported for three trials. There was no heterogeneity in the results. The RR of having hypernatraemia was 1.61 (95% CI 1.26 to 2.06) (Analysis 1.12).

Neuromuscular weakness

The number of participants who presented with neuromuscular weakness was reported for three trials. The RR of having neu- romuscular weakness was 0.63 (95% CI 0.12 to 3.35) (Analysis

1.12).

The main results are summarized in the ’Summary of findings’ table 1.

D I S C U S S I O N

In this review, we performed a comprehensive search of the lite r- ature with no restriction on language so we can assume that there was a very limited risk of missing important trials. The asymmet- rical funnel plot for the primary outcome of this review suggests some publication bias. However, potential sources of an asym- metrical funnel plot include selection biases, poor methodological quality of smaller studies, true heterogeneity, artefacts, or chance ( Egger 1997 ). Visual inspection of the funnel plot suggests a small- study effect (that is, among small studies the positive ones are more likely to be published). Nevertheless, our thorough search strategy and the need to enrol studies in public clinical trial registries may have decreased the risk of missing any randomized control trial. As discussed in the review, all studies on low dose corticosteroids had acceptable methodological quality. True heterogeneity seems to be a more plausible explanation of the observed asymmetrical funnel plot. Indeed, the effect of low dose corticosteroids on mortality may be proportional to the basal risk of death, and CORTICUS included patients at lower risk of death ( Sprung 2008 ). In ad- dition, smaller intervention effects in the CORTICUS trial may have resulted from an improved standard of care during the decade that separated most of the smaller trials and CORTICUS. Finally, the asymmetrical funnel plot may have been due to chance.

According to the primary objective of this systematic review, we only included the trials that compared corticosteroids to standard

therapy alone or a placebo. One trial had a crossover design ( Keh 2003 ) and we could obtain none of the foreseen outcomes for this review. This trial concluded that prolonged treatment with a low dose of hydrocortisone improved haemodynamic and im- mune outcomes. Another trial compared three days versus seven days of hydrocortisone therapy and suggested better outcomes in patients treated for seven days ( Huh 2007 ). Two other trials have included children ( CSG 1963 ; Slusher 1996 ). We considered that pooling the results of the remaining 20 trials in a meta-analy sis was acceptable.

At the time this review was written, three of the 24 trials were published only as an abstract ( Chawla 1999 ; Huh 2007 ; Tandan 2005 ). Nevertheless, the primary investigators of two of these stud- ies ( Chawla 1999 ; Tandan 2005 ) provided sufficient unpublished data to compute the primary outcome and several secondary out- comes for this review, allowing us to include these trials in th e meta-analysis. Both published and unpublished data were available for 13 trials ( Annane 2002 ; Bollaert 1998 ; Briegel 1999 ; Chawla 1999 ; Cicarelli 2007 ; Confalonieri 2005 ; Keh 2003 ; Meduri 2007 ; Oppert 2005 ; Rinaldi 2006 ; Sprung 1984 ; Sprung 2008 ; Yildiz 2002 ) and the primary author of each trial validated the data ex- traction form. For one study, contact with the primary investigator did not result in additional data (Luce 1988 ).

We chose to convert the outcome measures that correspond to censored data into dichotomous variables, that is, the propor tion of participants with a particular event after one week and four weeks, or at ICU or hospital discharge.

Overall, this review showed no evidence of an effect of corticos- teroids on 28-day mortality, ICU mortality, or on hospital mortal- ity from severe sepsis or septic shock. However, for these outcomes the nominal P values were close to 0.05 and there was strong het- erogeneity in the results. Sensitivity analyses based on tr ials with adequate generation of allocation sequence, trials with adequate allocation concealment, or double-blind trials also failed to show any benefit from corticosteroids.

Interestingly, when sorting the trials by year, almost all trials con- ducted before 1992 yielded a relative risk of dying at 28 days that was above one, and almost all trials conducted after 1992 yielded a relative risk below one. The date coincided with harmonization of the definition of severe sepsis and septic shock ( ACCP/SCCM 1992 ) and with a new understanding of the role of the adrenal glands in survival following septic shock ( Rothwell 1991 ). Before 1992 there was no standard definition of sepsis and no gradation of patients into sepsis, severe sepsis, and septic shock. This is why trials conducted before 1992 were more likely to include patients with very different risks of death. Most trials designed after 1992 focused on septic shock (the group of patients with the highest risk of death) using the same definition, which requires the ne ed for vasopressor agents to maintain blood pressure and organ per- fusion. These trials also used the same strategy with a long course

( 5 days) of low dose ( 300 mg) hydrocortisone or equivalent. Sensitivity analyses on the trials of a long course ( 5 days) of low dose corticosteroids ( 300 mg of hydrocortisone or equivalent) showed that, in patients with septic shock, this treatment signifi- cantly reduced 28-day all-cause mortality.

Although this subgroup analysis is a between-study and not within-study hypothesis, we thought its validity was acceptable according to recently proposed criteria ( Guyatt 2008b ). First, the hypothesis for an interaction between dose and duration and cor- ticosteroid effects on mortality was defined a priori. Second, we conducted only three subgroup analyses (based on methodological quality of studies, dose and duration, and baseline risk of death). Third, treatment effect was large, about a 6.6% absolute diff er- ence in mortality, and rather consistent between 28-day and hos- pital mortality (RRs of 0.84 and 0.85, respectively). Meta-regres- sion analysis further confirmed the interaction of the dose and duration with corticosteroids effects on mortality. Fourth, there is strong external evidence supporting these results. Exper imental and human studies have shown that a dose of 300 mg or less of hydrocortisone or equivalent can reverse the systemic inflamma- tory response, endothelial activation, and coagulation disorders secondary to an infection (Annane 2005 ); thus arguing against the use of higher doses. Moreover, at these low doses corticosteroids have been shown to improve rather than to suppress innate im- munity in patients with septic shock ( Kaufman 2008 ). It is now established that severe sepsis results in a sustained pro-inflamma- tory state, arguing against a short course of treatment ( Kellum 2007 ). Similarly, one randomized controlled trial has compared a short course of treatment (three days) with a longer course (seve n days) ( Huh 2007 ). This study suggested both reduction in shock duration and mortality, in favour of the seven-day strategy.

However, we judged the quality of evidence as moderate rather than high because one of the two largest trials on a long course of low-dose corticosteroids did not find a survival benefit (Sprung 2008 ). In addition, there are still several differences between th e various trials conducted after 1992 that should be pointed out. First, two trials included both severe sepsis and vasopressor-depen- dent septic shock patients (Slusher 1996 ; Yildiz 2002 ), one trial focused on severe sepsis due to community acquired pneumonia ( Confalonieri 2005 ), two trials included only septic shock with documented adrenal insufficiency ( Huh 2007 ; Tandan 2005 ), and one trial included only patients with severe sepsis who were f ree of vasopressor therapy ( Rinaldi 2006 ). Second, there was no stan- dardization of concomitant therapy. For example, in some trials patients may have received anti-thrombin III supplementation or intravenous polyclonal immunoglobulins ( Briegel 1999 Sprung 2008 ) while these treatments were not given in the other trials. The duration of shock also may have varied from one trial to anothe r. Some trials included only early septic shock ( Annane 2002 ; Huh 2007 ; Tandan 2005 ) while other trials included late septic shock ( Bollaert 1998 ; Chawla 1999 ; Cicarelli 2007 ) or both early and

late septic shock ( Briegel 1999 ; Keh 2003 ; Sprung 2008 ). Finally, one trial used both hydrocortisone and fludrocortisone and found a survival benefit (Annane 2002 ). One trial of 509 septic shock patients found a non-significant 3% absolute reduction in in-hos- pital mortality when fludrocortisone was added to hydrocortisone ( Annane 2010 ).

The beneficial effects observed on mortality with a long course of low dose corticosteroids may be related to the favourable effe ct of the treatment on the duration of shock. Indeed, this review showed that treatment with a long course of low dose corticos- teroids resulted in a substantial reduction in shock duration with fewer patients remaining on vasopressor therapy by day seve n and day 28. Treatment with a long course of low dose corticosteroids may also attenuate the severity of inflammation ( Confalonieri 2005 ; Keh 2003 ; Mikami 2007 a; Oppert 2005 ; Rinaldi 2006 ) and the intensity and duration of organ system failure ( Briegel 1999 ; Confalonieri 2005 ; Keh 2003 ; Oppert 2005 ; Sprung 2008 ), as shown in this review by a marked decrease in the SOFA score at seven day. In addition, subsequent to the favourable effect on car- diovascular and other organ function, the corticosteroid therapy resulted in a substantial shortening of intensive care unit l ength of stay. Finally, this review also showed no evidence of effect of corticosteroids on the rates of gastroduodenal bleeding or super- infection, or on the proportion of patients with acquired neur o- muscular weakness. Corticosteroids were associated with an in- creased risk for developing hyperglycaemia and hypernatraemia. One randomized controlled trial suggested that continuous in- fusion of hydrocortisone resulted in fewer episodes of hyper gly- caemia than bolus administration (Loisa 2007 ). One trial on 509 corticosteroid-treated septic shock patients did not find any be n- efit in normalizing blood glucose levels ( Annane 2010 ).

The current review cannot provide recommendations to select sep- tic shock patients who may be the best candidates for a long course of low dose corticosteroids. The analysis of seven trials, incl uding patients with adrenal insufficiency, suggested a non-significant re- duction in the risk of death. However, studies did not use the same definition for adrenal insufficiency. Further studies are needed to determine the best diagnostic tool for adrenal insufficiency in se- vere sepsis or septic shock patients (Cooper 2003 ).

A U T H O R S ’ C O N C L U S I O N S Implications for practice

Overall, corticosteroids did not impact on 28-day all-cause mor- tality in severe sepsis and septic shock. Meta-analysis of a subgroup of 12 trials investigating prolonged courses of low dose corticos- teroids showed a favourable impact on all-cause mortality. The dose used in these studies was 200 to 300 mg of hydrocortisone (or equivalent), by bolus intravenous injection or continuous in- fusion, for roughly a week. Although the evidence is not partic- ularly robust, we suggest that treatment should be given at f ull

dose for at least 100 hours only in adults with vasopressor-de pen- dent septic shock. There is insufficient evidence from this review to support either an abrupt or gradual interruption of treatment. The evidence accumulated from eight trials uniformly does not support the use of a short course of high dose corticosteroids in severe sepsis or septic shock.

Implications for research

The criteria for adrenal insufficiency in septic shock remain to be defined.

Ongoing trials should clarify:

1. the role of a long course of low dose corticosteroids for

treating septic shock in children;

2. the role of a long course of low dose corticosteroids for

treating severe sepsis, particularly for patients with community

acquired pneumonia;

3. the additional role of mineralocorticoid replacement;

4. the potential interaction of corticosteroids and activated

protein C in patients with septic shock.

The role of a long course of low dose corticosteroids for treating septic shock needs to be evaluated in developing countries to e x- tend generalizability.

The optimal timing to start treatment, the optimal dose of hy - drocortisone (or equivalent), and the duration and modality of withdrawal of treatment require further trials.

A C K N O W L E D G E M E N T S

We would like to thank Dr R DeGaudio, Dr GU Meduri, Dr M Oppert, Dr C Sprung, and Dr S Tandan for providing us with unpublished data.

This review was initially developed within the Infectious Diseases Group, supported by a grant from the Department for Interna- tional Development, UK. The review was transferred to the Anaes- thesia Group in May 2005.

We would like to thank Prof Harald Herkner (content edi- tor), Marialena Trivella (statistical editor), Peter Minneci, Charles Natanson, Gordon Guyatt, Matthias Briel (peer reviewers) and Karen Hovhannisyan (Cochrane Anaesthesia Review Group Trial s search co-ordinator) for their help and editorial advice during the preparation of this updated review.

R E F E R E N C E S

References to studies included in this review

Annane 2002 {published and unpublished data} Annane D, Sebille V, Charpentier C, Bollaert PE, François B, Korach JM, et al.Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock. JAMA 2002; 288 (7):862–71. [PUBMED: 12186604]

Annane 2010 {published and unpublished data} COIITSS Study Investigators, Annane D, Cariou A, Maxime V, Azoulay E, D’honneur G, Timsit JF, et al.Corticosteroid treatment and intensive insulin therapy for septic shock in adults: a randomized controlled trial. JAMA 2010; 303 :341–8.

Bollaert 1998 {published and unpublished data}

Bollaert PE, Charpentier C, Levy B, Debouverie M, Audibert G, Larcan A. Reversal of late septic shock with supraphysiologic doses

of hydrocortisone. Critical Care Medicine 1998; 26 (4):645–50.

[PUBMED: 9559600]

Bone 1987 {published data only} Bone RG, Fisher CJ, Clemmer TP, Slotman GJ, Metz CA, Balk RA.

A controlled clinical trial of high-dose methylprednisolone in the

treatment of severe sepsis and septic shock. New England Journal of

Medicine 1987; 317 (11):653–8.

Briegel 1999 {published and unpublished data} Briegel J, Forst H, Haller M, Schelling G, Kilger E, Kuprat G, et al.Stress doses of hydrocortisone reverse hyperdynamic septic shock:

a prospective, randomized, double-blind, single-center study. Critical Care Medicine 1999; 27 (4):723–32. [PUBMED:

10321661]

Chawla 1999 {published and unpublished data} Chawla K, Kupfer Y, Tessler S. Hydrocortisone reverses refractory septic shock. Critical Care Medicine 1999; 27 (1):A33.

Cicarelli 2007 {published data only} Cicarelli DD, Vieira JE, Benseñor FE. Early dexamethasone treatment for septic shock patients: a prospective randomized clinical trial. Sao Paulo Medical Journal 2007; 125 :237–41. [PUBMED: 17992396 ]

Confalonieri 2005 {published data only} Confalonieri M, Urbino R, Potena A, Piatella M, Parigi P, Giacomo

P, et al.Hydrocortisone infusion for severe community acquired

pneumonia: a preliminary randomized study. American Journal of

Respiratory and Critical Care Medicine 2005; 171 :242–8.

CSG 1963 {published data only} Cooperative Study Group. The effectiveness of hydrocortisone in the management of severe infections. JAMA 1963; 183 :462–5.

Huh 2007 {published data only} Huh JW, Lim CM, Koh Y, Hong SB. Effect of low doses of hydrocortisone in patient with septic shock and relative adrenal insufficiency: 3 days versus 7 days treatment. Critical Care Medicine 2007; 34 Suppl :A101.

Keh 2003 {published and unpublished data} Keh D, Boehnke T, Weber-Cartens S, Schulz C, Ahlers O, Bercker S, et al.Immunologic and hemodynamic effects of “low-dose” hydrocortisone in septic shock: a double-blind, randomized, placebo-controlled, crossover study. American Journal of Respiratory and Critical Care Medicine 2003; 167 (4):512–20. [PUBMED:

12426230 ]

Klastersky 1971 {published data only} Klastersky J, Cappel R, Debusscher L. Effectiveness of betamethasone in management of severe infections. A double-blind study. New England Journal of Medicine 1971; 284 (22):1248–50. [PUBMED: 4929896]

Lucas 1984 {published data only} Lucas C, Ledgerwood A. The cardiopulmonary response to massive doses of steroids in patients with septic shock. Archives of Surgery 1984; 119 (5):537–41. [PUBMED: 6712466 ]

Luce 1988 {published and unpublished data} Luce JM, Montgomery AB, Marks JD, Turner J, Metz CA, Murray JF. Ineffectiveness of high-dose methylprednisolone in preventing parenchymal lung injury and improving mortality in patients with septic shock. American Review of Respiratory Diseases 1988; 138 (1):

62–8. [PUBMED: 3202402 ]

Meduri 2007 {published and unpublished data} Meduri GU, Golden E, Freire AX, Taylor E, Zaman M, Carson SJ, et al.Methylprednislone infusion in early ARDS: result of a randomised controlled trial. Chest 2007; 131 (4):954–63. [PUBMED: 17426195]

Oppert 2005 {published and unpublished data} Oppert M, Schindler R, Husung C, Offerman K, Graef KJ, Boenisch O, et al.Low-dose hydrocortisone improves shock reversal and reduces cytokine levels in early hyperdynamic septic shock. Critical Care Medicine 2005; 33 :2457–64. [PUBMED: 16276166 ]

Rinaldi 2006 {published data only} Rinaldi S, Adembri C, Grechi S, DE Gaudio R. Low-dose hydrocortisone during severe sepsis: effects on microalbumineria. Critical Care Medicine 2006; 34 :2334–9. [PUBMED: 16850006]

Schumer 1976 {published data only} Schumer W. Steroids in the treatment of clinical septic shock. Annals of Surgery 1976; 184 (3):333–41. [PUBMED: 786190 ]

Slusher 1996 {published data only} Slusher T, Gbadero D, Howard C, Lewinson L, Giroir B, Toro L, et al.Randomized, placebo-controlled, double blinded trial of dexamethasone in African children with sepsis. Pediatric Infectious Diseases Journal 1996; 15 (7):579–83. [PUBMED: 8823850]

Sprung 1984 {published and unpublished data} Sprung CL, Caralis PV, Marcial EH, Pierce M, Gelbard MA, Long WM, et al.The effects of high-dose corticosteroids in patients with septic shock. A prospective, controlled study. New England Journal of Medicine 1984; 311 (18):1137–43. [PUBMED: 6384785 ]

Sprung 2008 {published and unpublished data} Sprung C, Annane D, Keh D, Moreno R, Singer M, Freivogel K, et al.The CORTICUS randomized, double-blind, placebo- controlled study of hydrocortisone therapy in patients with sep tic shock. New England Journal of Medicine 2008; 358 :111–24. [PUBMED: 18184957]

Tandan 2005 {unpublished data only} Tandan SM, Guleria R, Gupta N. Low dose steroids and adrenocortical insufficiency in septic shock: a double-blind randomised controlled trial from India. Proceedings of the American Thoracic Society Meeting. 2005:A24.

VASSCSG 1987 {published data only} The Veterans Administration Systemic Sepsis Cooperative Stud y Group. Effect of high-dose glucocorticoid therapy on morta lity in patients with clinical signs of systemic sepsis. New England Journal of Medicine 1987; 317 (11):659–65. [PUBMED: 2888017]

Wagner 1955 {published data only} Wagner HN, Bennett IL, Lasagna L, Cluff LE, Rosenthal MB, Mirick GS. The effect of hydrocortisone upon the course of pneumococcal pneumonia treated with penicillin. Bulletin of Johns Hopkins Hospital 1955; 98 :197–215. [PUBMED: 13304518]

Yildiz 2002 {published data only} Yildiz O, Doganay M, Aygen B, Guven M, Keleutimur F, Tutuu A. Physiologic-dose steroid therapy in sepsis. Critical Care 2002; 6 (3):

251–9. [PUBMED: 12133187]

References to studies excluded from this review

Cicarelli 2006 {published data only} Cicarelli DD, Bensenor FEM, Vieira JE. Effects of single dose of dexamethasone on patients with systemic inflammatory response. Sao Paulo Medical Journal 2006; 124 :90–5. [PUBMED: 16878192

]

Hahn 1951 {published data only} Hahn EO, Houser HB, Rammelkamp CH, Denny FW, Wannamaker LW. Effect of cortisone on acute streptococcal infections and post-streptococcal complications. Journal of Clinical Investigation 1951; 30 :274–81.

Hughes 1984 {published data only} Hughes GS Jr. Naloxone and methylprednisolone sodium succinate enhance sympathomedullary discharge in patients with septic shock. Life Sciences 1984; 35 (23):2319–26. [PUBMED: 6390057]

Kaufman 2008 {published and unpublished data} Kaufmann I, Briegel J, Schliephake F, Hoelzl A, Chouker A, Hummel T, et al.Stress doses of hydrocortisone in septic shoc k:

beneficial effects on opsonization-dependent neutrophil functions . Intensive Care medicine 2008; 34 :344–9. [PUBMED: 17906853 ]

McKee 1983 {published data only} McKee JI, Finlay WE. Cortisol replacement in severely stressed patients. Lancet 1983; 1 (8322):484. [PUBMED: 6131207 ]

Meduri 1998b {published and unpublished data}

Meduri GU, Headley AS, Golden E, Carson SJ, Umberger RA, Kelso T, et al.Effect of prolonged methylprednisolone therapy i n unresolving acute respiratory distress syndrome: a randomized controlled trial. JAMA 1998; 28