This document provides guidelines from an international panel of experts for the management of severe sepsis and septic shock. Some key recommendations include:
1) Early goal-directed resuscitation of septic patients within the first 6 hours, targeting specific goals for central venous pressure, mean arterial pressure, urine output, and central venous oxygen saturation.
2) Obtaining appropriate diagnostic studies like blood cultures before starting antibiotics to determine the causative organisms.
3) Early administration of broad-spectrum antibiotics, with reassessment to narrow coverage based on clinical and microbiological data.
4) Other recommendations address source control, fluid resuscitation, vasopressor use, glucose control, ventilation strategies, prophylaxis,
2. T he mortality rate of severe sepsis and septic shock. This process repre- METHODS
sepsis (infection-induced or- sented phase II of the Surviving Sepsis
The recommendations are graded based on
gan dysfunction or hypoper- Campaign, an international effort to in-
a modified Delphi methodology with categori-
fusion abnormalities) and crease awareness and improve outcome in
zation as previously described (Table 1,
septic shock (hypotension not reversed severe sepsis. Meeting expenses as well as adapted from Ref. 3). The methods for this
with fluid resuscitation and associated staff support for guidelines creation were document build on a 2001 publication spon-
with organ dysfunction or hypoperfu- provided by unrestricted industry educa- sored by the International Sepsis Forum and
sion abnormalities) in most centers re- tional grants as listed. There were no in- use the same method of grading recommen-
mains unacceptably high (1, 2). Similar dustry members of the committee. There dations (4). The supplement submission will
to an acute myocardial ischemic attack was no industry input into guidelines de- include background material, questions posed
and an acute brain attack, the speed and velopment and no industry presence at any that led to the recommendation, and expanded
appropriateness of therapy administered of the meetings. Industry awareness or rationale. This executive summary is targeted
in the initial hours after the syndrome de- comment on the recommendations was to be concise and user friendly for the bedside
velops likely influence outcome. A group of not allowed. The sponsors of the educa- clinician. The 2001 publication that was used
international critical care and infectious tional grants did not see the recommenda- as a starting point for the current process
disease experts in the diagnosis and man- tions until the manuscript was peer re- included a MEDLINE search for clinical trials
agement of infection and sepsis, represent- viewed and accepted for publication in final in the preceding 10 yrs, supplemented by a
ing 11 organizations, came together to de- form. Phase I of the Surviving Sepsis manual search of other relevant journals. Sub-
velop guidelines that the bedside clinician topics for each recommendation were cross-
Campaign was initiated in October 2002
could use to improve outcome in severe referenced to sepsis, severe sepsis, septic
with the Barcelona Declaration to improve
shock, sepsis syndrome, and infection. The
survival in severe sepsis, and phase III will
Surviving Sepsis Campaign guidelines consid-
be dedicated to the use of the management ered the evidence in the 2001 publication
Surviving Sepsis Campaign Management Guide-
lines Committee. Chairs: R. Phillip Dellinger, MD*; guidelines to evaluate the impact on clini- (through 1999) and repeated the process for
Henry Masur, MD; Jean M. Carlet, MD; Herwig Gerlach, cal outcome. A comprehensive document 2000 through 2003. The committee process
MD, PhD**. Committee Members: Richard J. Beale, created from the deliberations of the com- began in June 2003 with a meeting featuring
MD**; Marc Bonten, MD; Christian Brun-Buisson, MD; mittee will be submitted for publication as
Thierry Calandra, MD; Joseph A. Carcillo, MD; the first presentations of data and recommen-
Jonathan Cohen, MD**; Catherine Cordonnier, MD; E. a supplement. This document represents dations. Recommendations were discussed
Patchen Dellinger, MD; Jean-Francois Dhainaut, MD, an executive summary of the consensus and critiqued. Each clinical trial used to sup-
PhD; Roger G. Finch, MD; Simon Finfer, MD; Francois process with presentation of key recom- port recommendations was graded based on
A. Fourrier, MD; Juan Gea-Banacloche, MD; Maurene mendations. These recommendations are the methodology in Table 1 and included pres-
A. Harvey, RN, MPH**; Jan A. Hazelzet, MD; Steven M. ence or absence of important elements such as
Hollenberg, MD; James H. Jorgensen, PhD; Didier Keh, intended to provide guidance for the clini-
MD; Mitchell M. Levy, MD*; Ronald V. Maier, MD; cian caring for a patient with severe sepsis concealed randomization, blinded outcome
Dennis G. Maki, MD; John J. Marini, MD; John C. or septic shock, but they are not applicable adjudication, intention to treat analysis, and
Marshall, MD; Steven M. Opal, MD; Tiffany M. Osborn, for all patients. Recommendations from explicit definition of primary outcome. All ar-
MD; Margaret M. Parker, MD**; Joseph E. Parrillo, MD;
these guidelines cannot replace the clini- ticles were initially reviewed based on sub-
Graham Ramsay, MD*; Andrew Rhodes, MD; Jonathan group assignments and typically by two or
E. Sevransky, MD; Charles L. Sprung, MD, JD**; Antoni cian’s decision-making capability when he
three participants. Survival (28 –30 days) was
Torres, MD; Jeffery S. Vender, MD; Jean-Louis Vin- or she is provided with a patient’s unique
cent, MD, PhD**; Janice L. Zimmerman, MD. Associate the standard outcome measure used to assess
set of clinical variables. Although these rec- outcome benefit, and when an alternative was
Members: E. David Bennett, MD; Pierre-Yves Bochud,
ommendations are written primarily for used this is stated in the rationale. Where
MD; Alain Cariou, MD; Glenn S. Murphy, MD; Martin
Nitsun, MD; Joseph W. Szokol, MD; Stephen Trzeciak, the patient in the intensive care unit (ICU) strong trial evidence existed for outcome ben-
MD; Christophe Vinsonneau, MD. *Executive Commit- setting, many recommendations are appro- efit in critically ill populations known to con-
tee, Surviving Sepsis Campaign. **Steering Commit- priate targets for the pre-ICU setting. It tain a larger number of sepsis patients, these
tee, Surviving Sepsis Campaign. The Surviving Sepsis should also be noted that resource limita-
Campaign is administered jointly by the European So- trials were considered in determination of rec-
ciety of Intensive Care Medicine, International Sepsis tions may prevent physicians from accom- ommendation grading. A strict evidence-based
Forum, and the Society of Critical Care Medicine and is plishing a recommendation. methodology with a scoring system was not
supported in part by unrestricted educational grants
from Baxter Bioscience, Edwards Lifesciences, and Eli
Lilly and Company (majority sponsor).
Use of trade names or names of commercial Table 1. Grading system
sources is for information only and does not imply
endorsement by the Society of Critical Care Medicine. Grading of recommendations
The authors and the publisher have exercised A. Supported by at least two level I investigations
great care to ensure that drug dosages, formulas, and B. Supported by one level I investigation
other information presented in this publication are C. Supported by level II investigations only
accurate and in accord with the professional standards D. Supported by at least one level III investigation
in effect at the time of publication. Readers are, how- E. Supported by level IV or V evidence
ever, advised to always check the manufacturer’s Grading of evidence
product information sheet that is packaged with the I. Large, randomized trials with clear-cut results; low risk of false-positive (alpha) error of
respective products to be fully informed of changes in false-negative (beta) error
recommended dosages, contraindications, and the like II. Small, randomized trials with uncertain results; moderate-to-high risk of false-positive
before prescribing or administering any drug. (alpha) and/or false-negative (beta) error
Also published in Intensive Care Medicine (May). III. Nonrandomized, contemporaneous controls
Address requests for reprints to: R. Phillip Del- IV. Nonrandomized, historical controls and expert opinion
linger, MD, Cooper Health Systems, One Cooper Plaza, V. Case series, uncontrolled studies, and expert opinion
393 Dorrance, Camden, NJ 08103.
Crit Care Med 2004 Vol. 32, No. 3 859
3. used. The goal was total consensus, which was Central venous pressure: 8 –12 mm Hg apy is initiated as the clinical situation
reached in all recommendations except two. In Mean arterial pressure 65 mm Hg dictates.
those circumstances (recommendations C3 and Urine output 0.5 mL·kg 1·hr 1
H1), the solution was achieved with the inclu- Grade D
Central venous (superior vena cava) or mixed
sion of subrecommendations that expressed Rationale. Two or more blood cultures are
venous oxygen saturation 70%
some difference in expert opinion. When there recommended (6). Ideally, at least one blood
was difference of opinion about grading of a Grade B culture should be drawn through each lumen
clinical trial, an outside epidemiologist was con- of each vascular access device. Obtaining
Rationale. Early goal-directed therapy has
sulted. This occurred in one circumstance with been shown to improve survival for emergency blood cultures peripherally and through a vas-
resolution of differences. Each participant com- department patients presenting with septic cular access device is an important strategy. If
pleted a conflict of interest form, and individuals shock in a randomized, controlled, single-- the same organism is recovered from both
were not assigned to a subgroup topic if they had center study (5). Resuscitation directed to- cultures, the likelihood that the organism is
a potential conflict of interest. A full listing of all ward the previously mentioned goals for the causing the severe sepsis is enhanced. In ad-
potential conflicts of interest is included with initial 6-hr period of the resuscitation was able dition, if the culture drawn through the vas-
this article. Following that meeting, the process to reduce 28-day mortality rate. The consen- cular access device is positive much earlier
continued with further refinement of recom- sus panel judged central venous and mixed than the peripheral blood culture (i.e., 2 hrs
mendations through electronic communication venous oxygen saturation to be equivalent. earlier), it may offer support that the vascular
among committee members. A second meeting Either intermittent or continuous measure- access device is the source of the infection (7).
of core members of the committee occurred in ments of oxygen saturation are judged to be Volume of blood may also be important (8).
early October 2003. The document was finalized acceptable. Although lactate measurement
2. Diagnostic studies should be performed
and approved by the consensus committee and may be useful, it lacks precision as a measure
promptly to determine the source of the
by sponsoring organizations in December 2003. of tissue metabolic status. In mechanically
infection and the causative organism.
Evidence-based approaches are more ventilated patients, a higher target central ve-
Imaging studies and sampling of likely
readily applied to data from therapeutic trials. nous pressure of 12–15 mm Hg is recom-
sources of infection should be per-
Evaluation of diagnostic techniques is less mended to account for the increased intratho-
formed; however, some patients may be
racic pressure. Similar consideration may be
well suited to this approach. Readers will note too unstable to warrant certain invasive
warranted in circumstances of increased ab-
that the majority of the recommendations are procedures or transport outside of the
dominal pressure. Although the cause of
not supported by high-level evidence. Most are ICU. Bedside studies, such as ultrasound,
tachycardia in septic patients may be multi-
supported by expert opinion only. In order for may be useful in these circumstances.
factorial, a decrease in elevated pulse with
a general recommendation to carry a higher fluid resuscitation is often a useful marker of Grade E
level of evidence (grades A, B, C, or D), a improving intravascular filling.
supporting study or studies must have shown a Rationale. Diagnostic studies may identify a
clinical outcome difference. Studies showing 2. During the first 6 hrs of resuscitation of source of infection that must be drained to max-
physiologic changes that could be potential sur- severe sepsis or septic shock, if central ve- imize the likelihood of a satisfactory response to
rogates of clinical outcome benefit were not nous oxygen saturation or mixed venous therapy. However, even in the most organized
used by themselves as pivotal studies but were oxygen saturation of 70% is not achieved and well-staffed healthcare facilities, transport of
used to support the validity of studies showing with fluid resuscitation to a central venous patients can be dangerous, as can placing pa-
pressure of 8–12 mm Hg, then transfuse tients in outside-unit imaging devices that are
an outcome in a clinically important variable
packed red blood cells to achieve a hemat- difficult to access and monitor.
such as survival or length of ICU stay. A grade of
ocrit of 30% and/or administer a dobut-
A, B, or C required randomized trials. Recom-
amine infusion (up to a maximum of 20
mendations are graded and followed with ratio- g·kg 1·min 1) to achieve this goal. C. Antibiotic Therapy
nale. References are provided to support grades
A–D. In the committee’s deliberations, the grad- Grade B 1. Intravenous antibiotic therapy should
ing of a recommendation did not establish the be started within the first hour of rec-
Rationale. The protocol used in the study ognition of severe sepsis, after appropri-
level of priority or importance of a specific in- cited previously targeted an increase in mixed
tervention, only the degree of literature support. ate cultures have been obtained.
venous oxygen saturation to 70%. This was
Pediatric considerations are provided at the end achieved by sequential institution of initial fluid Grade E
of the document for aspects of management that resuscitation, then packed red blood cells, and
differ from adults. Recommendations are then dobutamine. This protocol was associated Rationale. Establishing vascular access and
grouped by category and not by hierarchy. with an improvement in survival (5). initiating aggressive fluid resuscitation is the
first priority when managing patients with se-
vere sepsis or septic shock. However, prompt
A. Initial Resuscitation B. Diagnosis infusion of antimicrobial agents is also a logical
1. The resuscitation of a patient in severe 1. Appropriate cultures should always be ob- strategy and may require additional vascular ac-
sepsis or sepsis-induced tissue hypoperfu- tained before antimicrobial therapy is ini- cess ports. Establishing a supply of premixed
sion (hypotension or lactic acidosis) tiated. To optimize identification of caus- antibiotics in an emergency department or crit-
should begin as soon as the syndrome is ative organisms, at least two blood ical care unit for such urgent situations is an
recognized and should not be delayed appropriate strategy for enhancing the likeli-
cultures should be obtained with at least
pending ICU admission. An elevated se- hood that antimicrobial agents will be infused
one drawn percutaneously and one drawn
rum lactate concentration identifies tis- promptly. Staff should be cognizant that some
through each vascular access device, un-
sue hypoperfusion in patients at risk who agents require more lengthy infusion time
are not hypotensive. During the first 6 hrs less the device was recently ( 48 hrs) whereas others can be rapidly infused or even
of resuscitation, the goals of initial resus- inserted. Cultures of other sites such as administered as a bolus.
citation of sepsis-induced hypoperfusion urine, cerebrospinal fluid, wounds, respi-
should include all of the following as one ratory secretions, or other body fluids 2. Initial empirical anti-infective therapy
part of a treatment protocol: should be obtained before antibiotic ther- should include one or more drugs that
860 Crit Care Med 2004 Vol. 32, No. 3
4. have activity against the likely pathogens broad-spectrum therapy usually must be con- Grade E
(bacterial or fungal) and that penetrate tinued for the duration of the neutropenia.
into the presumed source of sepsis. The 3. When a focus of infection amenable to
choice of drugs should be guided by the Grade E source control measures such as an
susceptibility patterns of microorganisms intra-abdominal abscess, a gastrointes-
Rationale. Use of antimicrobial agents with tinal perforation, cholangitis, or intes-
in the community and in the hospital. a more narrow spectrum and reducing the tinal ischemia has been identified as
duration of therapy will reduce the likelihood the cause of severe sepsis or septic
Grade D
that the patient will develop superinfection shock, source control measures should
Rationale. The choice of empirical antibi- with pathogenic or resistant organisms such be instituted as soon as possible follow-
otics depends on complex issues related to the as Candida species, Clostridium difficile, or ing initial resuscitation.
patient’s history (including drug intolerance), vancomycin-resistant Enterococcus faecium.
underlying disease, the clinical syndrome, and However, the desire to minimize superinfec- Grade E
susceptibility patterns in the patient’s com- tions and other complications should not take
Rationale. Case series and expert opinion
munity and in the healthcare facility. precedence over the need to give the patient
support the principle that rapid correction of a
The initial selection of an empirical anti- an adequate course of potent antimicrobials.
source of microbial contamination is essential
microbial regimen should be broad enough, to maximize survival of the severely septic
according to these criteria, covering all likely 4. If the presenting clinical syndrome is
determined to be due to a noninfec- patient with acute physiologic deterioration.
pathogens since there is little margin for error Intervention should only be undertaken fol-
in critically ill patients. There is ample evi- tious cause, antimicrobial therapy
should be stopped promptly to mini- lowing adequate resuscitation. Timely and
dence that failure to initiate appropriate ther- emergent intervention is particularly impor-
apy promptly (i.e., therapy that is active mize the development of resistant
pathogens and superinfection with tant for patients with necrotizing soft tissue
against the causative pathogen) has adverse infection or intestinal ischemia (19).
consequences on outcome (9 –12). other pathogenic organisms.
Although restricting the use of antibiotics, Grade E 4. If intravascular access devices are poten-
and particularly broad-spectrum antibiotics, is tially the source of severe sepsis or septic
important for limiting superinfection and for Rationale. Clinicians should be cognizant shock, they should be promptly removed
decreasing the development of antibiotic- that blood cultures will be negative in the after establishing other vascular access.
resistant pathogens, patients with severe sep- majority of cases of sepsis or septic shock.
sis or septic shock warrant broad-spectrum Thus, the decision to continue, narrow, or Grade E
therapy until the causative organism and its stop antimicrobial therapy must be made on
antibiotic susceptibilities are defined. At that the basis of clinician judgment and other cul- Rationale. Intravascular access devices are
point, restriction of the number of antibiotics ture results. thought to be the source of the majority of
and narrowing the spectrum of antimicrobial nosocomial bloodstream infections. When pa-
therapy is an important and responsible strat- tients develop sepsis of unknown source, it
D. Source Control may be reasonable to leave vascular access
egy for minimizing the development of resis-
tant pathogens and for containing costs. devices in place until the source of infection
1. Every patient presenting with severe
All patients should receive a full loading can be determined. However, when patients
sepsis should be evaluated for the pres-
dose of each antimicrobial. However, patients ence of a focus on infection amenable to have severe sepsis or septic shock of unknown
with sepsis or septic shock often have abnor- source control measures, specifically source, clinicians should consider removal
mal renal or hepatic function and may have the drainage of an abscess or local focus and replacement of vascular access devices to
abnormal volumes of distribution due to ag- on infection, the debridement of in- be a priority, even if the device is tunneled or
gressive fluid resuscitation. The ICU pharma- fected necrotic tissue, the removal of a surgically implanted (20, 21).
cist should be consulted to ensure that serum potentially infected device, or the defin-
concentrations are attained that maximize ef- itive control of a source of ongoing mi- E. Fluid Therapy
ficacy and minimize toxicity, (13–16). crobial contamination (17). (See Appen-
dix A for examples of potential sites See initial resuscitation recommendations
3. The antimicrobial regimen should al- needing source control.)
(A1–2) for timing of resuscitation.
ways be reassessed after 48 –72 hrs on
Grade E
the basis of microbiological and clinical 1. Fluid resuscitation may consist of natu-
data with the aim of using a narrow- Rationale. Healthcare professionals should ral or artificial colloids or crystalloids.
spectrum antibiotic to prevent the devel- engage specialists in other disciplines such as There is no evidence-based support for
opment of resistance, to reduce toxicity, radiology, surgery, pulmonary medicine, and one type of fluid over another.
and to reduce costs. Once a causative gastroenterology to obtain diagnostic samples
pathogen is identified, there is no evi- and to drain, debride, or remove the infection Grade C
dence that combination therapy is more source as appropriate.
effective than monotherapy. The dura- Rationale. Although prospective studies of
tion of therapy should typically be 7–10 2. The selection of optimal source control choice of fluid resuscitation in patients with
days and guided by clinical response. methods must weigh benefits and risks septic shock only are lacking, meta-analysis of
of the specific intervention. Source con- clinical studies comparing crystalloid and col-
Grade E trol interventions may cause further loid resuscitation in general and surgical pa-
complications such as bleeding, fistulas, tient populations indicate no clinical outcome
a. Some experts prefer combination therapy or inadvertent organ injury; in general, difference between colloids and crystalloids
for patients with Pseudomonas infections. the intervention that accomplishes the and would appear to be generalizable to sepsis
Grade E source control objective with the least populations (22–24). As the volume of distri-
physiologic upset should be employed, bution is much larger for crystalloids than for
b. Most experts would use combination ther- for example, consideration of percutane- colloids, resuscitation with crystalloids re-
apy for neutropenic patients with severe sepsis ous rather than surgical drainage of an quires more fluid to achieve the same end
or septic shock. For neutropenic patients, abscess (18). points and results in more edema.
Crit Care Med 2004 Vol. 32, No. 3 861
5. 2. Fluid challenge in patients with sus- Grade D 5. Vasopressin use may be considered in
pected hypovolemia (suspected inade- patients with refractory shock despite
quate arterial circulation) may be given Rationale. Although there is no high- adequate fluid resuscitation and high-
at a rate of 500 –1000 mL of crystalloids quality primary evidence to recommend one dose conventional vasopressors. Pending
or 300 –500 mL of colloids over 30 mins catecholamine over another, human and an- the outcome of ongoing trials, it is not
and repeated based on response (increase imal studies suggest some advantages of recommended as a replacement for nor-
in blood pressure and urine output) and norepinephrine and dopamine over epineph- epinephrine or dopamine as a first-line
tolerance (evidence of intravascular vol- rine (potential tachycardia, possibly disad- agent. If used in adults, it should be
ume overload). vantageous effects on splanchnic circula- administered at infusion rates of 0.01–
tion) and phenylephrine (decrease in stroke 0.04 units/min. It may decrease stroke
Grade E volume). Phenylephrine is the adrenergic volume.
agent least likely to produce tachycardia.
Rationale. Fluid challenge must be clearly Dopamine increases mean arterial pressure Grade E
separated from an increase in maintenance and cardiac output, primarily due to an in-
fluid administration. Fluid challenge is a term crease in stroke volume and heart rate. Nor- Rationale. Low doses of vasopressin may
used to describe the initial volume expansion epinephrine increases mean arterial pres- be effective in raising blood pressure in pa-
period in which the response of the patient to sure due to its vasoconstrictive effects, with tients refractory to other vasopressors, al-
fluid administration is carefully evaluated. little change in heart rate and less increase though no outcome data are available. Un-
During this process, large amounts of fluids in stroke volume compared with dopamine. like dopamine and epinephrine, vasopressin
may be administered over a short period of Either may be used as a first-line agent to is a direct vasoconstrictor without inotropic
time under close monitoring to evaluate the correct hypotension in sepsis. Norepineph- or chronotropic effects and may result in
patient’s response and avoid the development rine is more potent than dopamine and may decreased cardiac output and hepato-
of pulmonary edema. The degree of intravas- be more effective at reversing hypotension splanchnic flow. Most published reports ex-
cular volume deficit in patients with severe in patients with septic shock. Dopamine may clude patients from treatment with vaso-
sepsis varies. With venodilation and ongoing be particularly useful in patients with com- pressin if the cardiac index is 2 or 2.5
capillary leak, most patients require continu- promised systolic function but causes more L·min 1·m 2, and it should be used with
ing aggressive fluid resuscitation during the tachycardia and may be more arrhythmo- caution in patients with cardiac dysfunction.
first 24 hrs of management. Input is typically genic (25, 27–30). Studies show that vasopressin concentra-
much greater than output, and input/output tions are elevated in early septic shock, but
ratio is of no utility to judge fluid resuscitation 3. Low-dose dopamine should not be used with continued shock, concentrations de-
needs during this time period. for renal protection as part of the treat- crease to normal range in the majority of
ment of severe sepsis. patients between 24 and 48 hrs (33). This
has been called “relative vasopressin defi-
F. Vasopressors Grade B ciency” since in the presence of hypoten-
sion, vasopressin would be expected to be
1. When an appropriate fluid challenge Rationale. A large randomized trial and a elevated. The significance of this finding is
fails to restore adequate blood pres- meta-analysis comparing low-dose dopamine unknown. Doses of vasopressin 0.04 units/
sure and organ perfusion, therapy to placebo in critically ill patients found no min have been associated with myocardial
with vasopressor agents should be difference in either primary outcomes (peak ischemia, significant decreases in cardiac
started. Vasopressor therapy may also serum creatinine, need for renal replacement
be required transiently to sustain life output, and cardiac arrest (34 –36).
therapy, urine output, time to recovery of nor-
and maintain perfusion in the face of
mal renal function) or secondary outcomes
life-threatening hypotension, even G. Inotropic Therapy
when a fluid challenge is in progress (survival to either ICU or hospital discharge,
and hypovolemia has not yet been cor- ICU stay, hospital stay, arrhythmias). Thus,
the available data do not support administra- 1. In patients with low cardiac output de-
rected. spite adequate fluid resuscitation, dobu-
tion of low doses of dopamine to maintain or
tamine may be used to increase cardiac
Grade E improve renal function (31, 32).
output. If used in the presence of low
Rationale. Below a certain mean arterial blood pressure, it should be combined
4. All patients requiring vasopressors with vasopressor therapy.
pressure, autoregulation in various vascular should have an arterial catheter placed as
beds can be lost, and perfusion can become lin- soon as practical if resources are available. Grade E
early dependent on pressure. Thus, some pa-
tients may require vasopressor therapy to Grade E Rationale. Dobutamine is the first-choice
achieve a minimal perfusion pressure and main- inotrope for patients with measured or sus-
tain adequate flow. It is important to supple- Rationale. In shock states, measurement pected low cardiac output in the presence of
ment goals such as blood pressure with assess- of blood pressure using a cuff is commonly adequate left ventricular filling pressure (or clin-
ment of global perfusion such as blood lactate inaccurate, whereas use of an arterial cath- ical assessment of adequate fluid resuscitation)
concentrations. Adequate fluid resuscitation is a eter provides a more accurate and reproduc- and adequate mean arterial pressure. In the ab-
fundamental aspect of the hemodynamic man- ible measurement of arterial pressure. Mon- sence of measurements of cardiac output, hypo-
agement of patients with septic shock and itoring with these catheters also allows beat- tensive patients with severe sepsis may have low,
should ideally be achieved before vasopressors to-beat analysis so that decisions regarding normal, or increased cardiac outputs. Therefore,
are used, but it is frequently necessary to employ therapy can be based on immediate blood treatment with a combined inotrope/vasopressor
vasopressors early as an emergency measure in pressure information (25). Placement of an such as norepinephrine or dopamine is recom-
patients with severe shock (25, 26). arterial catheter in the emergency depart- mended. When the capability exists for monitor-
ment is typically not possible or practical. It ing cardiac output in addition to blood pressure,
2. Either norepinephrine or dopamine is important to appreciate the complications a vasopressor such as norepinephrine and an
(through a central catheter as soon as avail- of arterial catheter placement, which in- inotrope such as dobutamine may be used sep-
able) is the first-choice vasopressor agent to clude hemorrhage and damage to arterial arately to target specific levels of mean arterial
correct hypotension in septic shock. vessels. pressure and cardiac output.
862 Crit Care Med 2004 Vol. 32, No. 3
6. 2. A strategy of increasing cardiac index to b. Some experts would decrease dosage of syndrome [ARDS]) and with no absolute
achieve an arbitrarily predefined elevated steroids after resolution of septic shock. contraindication related to bleeding risk or
level is not recommended. relative contraindication that outweighs
Grade E the potential benefit of rhAPC (see Appen-
Grade A dix B for absolute contraindications and
Rationale. There has been no comparative
prescription information for warnings).
Rationale. Two large prospective clinical study between a fixed duration and clinically
trials that included critically ill ICU patients guided regimen. Two RCTs used a fixed dura- Grade B
who had severe sepsis failed to demonstrate tion protocol for treatment (39, 41), and inone
benefit from increasing oxygen delivery to su- RCT, therapy was decreased after shock reso- Rationale. The inflammatory response in
pranormal levels by use of dobutamine (37, lution and discontinued after 6 days (40). severe sepsis is integrally linked to procoagu-
38). The goal of resuscitation should instead lant activity and endothelial activation. The
be to achieve adequate levels of oxygen deliv- c. Some experts would consider tapering the inflammatory response in sepsis is procoagu-
ery or avoid flow-dependent tissue hypoxia. dose of corticosteroids at the end of therapy. lant in the early stages. rhAPC, an endogenous
anticoagulant with anti-inflammatory proper-
Grade E ties, has been shown, in a large, multiple-
H. Steroids center, randomized, controlled trial (50), to
Rationale. One study showed hemody-
1. Intravenous corticosteroids (hydrocortisone namic and immunologic rebound effects after improve survival in patients with sepsis-
200 –300 mg/day, for 7 days in three or abrupt cessation of corticosteroids (46). induced organ dysfunction.
four divided doses or by continuous infu- At present, risk assessment is best deter-
sion) are recommended in patients with d. Some experts would add fludrocortisone (50 mined by bedside clinical evaluation and judg-
septic shock who, despite adequate fluid g orally four times per day) to this regimen. ment. Given the uncertainty of risk assess-
replacement, require vasopressor therapy ment and the potential for rapid deterioration
Grade E
to maintain adequate blood pressure. of patients with severe sepsis and septic shock,
Rationale. One study added 50 g of once a patient has been identified as at high
Grade C risk of death, treatment should begin as soon
fludrocortisone orally (39). Since hydrocorti-
Rationale. One multiple-center, random- sone has intrinsic mineralocorticoid activity, as possible.
ized, controlled trial (RCT) with patients in se- there is controversy as to whether fludrocor-
vere septic shock showed a significant shock tisone should be added. J. Blood Product Administration
reversal and reduction of mortality rate in pa-
tients with relative adrenal insufficiency (defined 2. Doses of corticosteroids 300 mg hydro-
1. Once tissue hypoperfusion has resolved
as post-adrenocorticotropic hormone [ACTH] cortisone daily should not be used in and in the absence of extenuating circum-
cortisol increase 9 g/dL) (39). Two additional severe sepsis or septic shock for the pur- stances, such as significant coronary artery
smaller RCTs showed significant effects on pose of treating septic shock. disease, acute hemorrhage, or lactic acido-
shock reversal (40, 41). In the first study, pa- sis (see recommendations for initial resus-
Grade A
tients had more severe septic shock (systolic citation), red blood cell transfusion should
blood pressure 90 mm Hg despite vasopres- Rationale. Two randomized prospective occur only when hemoglobin decreases to
sors) than in the latter two studies (systolic clinical trials and two meta-analyses concluded 7.0 g/dL ( 70 g/L) to target a hemoglo-
blood 90 mm Hg with vasopressors). that for therapy of severe sepsis or septic shock, bin of 7.0 –9.0 g/dL.
high-dose corticosteroid therapy is ineffective or
a. Some experts would use a 250- g ACTH Grade B
harmful (47–50). There may be reasons to main-
stimulation test to identify responders ( 9 tain higher doses of corticosteroid for medical Rationale. Although the optimum hemo-
g/dL increase in cortisol 30 – 60 mins post- conditions other than septic shock. globin for patients with severe sepsis has not
ACTH administration) and discontinue been specifically investigated, the Transfusion
therapy in these patients. Clinicians should 3. In the absence of shock, corticosteroids Requirements in Critical Care trial suggested
not wait for ACTH stimulation results to should not be administered for the treat- that a hemoglobin of 7–9 g/dL (70 –90 g/L) is
administer corticosteroids. ment of sepsis. There is, however, no adequate for most critically ill patients. A
contraindication to continuing mainte- transfusion threshold of 7.0 g/dL (70 g/L) was
Grade E nance steroid therapy or to using stress not associated with increased mortality rate.
Rationale. One study demonstrated that an dose steroids if the patient’s history of Red blood cell transfusion in septic patients
incremental increase of 9 g/dL after 250- g corticosteroid administration or the pa- increases oxygen delivery but does not usually
ACTH stimulation test (responders) identifies tient’s endocrine history warrants. increase oxygen consumption (51–53). This
survivors of septic shock (42). A subsequent trial Grade E transfusion threshold contrasts with the tar-
demonstrated that stress dose steroids improved get of a hematocrit of 30% in patients with low
survival in those patients who failed to produce Rationale. There are no studies document- central venous oxygen saturation during the
this increase in cortisol with ACTH (nonre- ing that stress doses of steroids improve the first 6 hrs of resuscitation of septic shock.
sponders). Treatment with corticosteroids was outcome of sepsis in the absence of shock
ineffective in responders (39). Recommenda- unless the patient requires stress dose replace- 2. Erythropoietin is not recommended as a
tions for the identification of relative adrenal ment due to a prior history of steroid therapy specific treatment of anemia associated
insufficiency vary based on different cutoff levels or adrenal dysfunction. with severe sepsis but may be used when
of random cortisol, peak cortisol after stimula- septic patients have other accepted rea-
tion, incremental cortisol increase after stimu- sons for administration of erythropoietin
I. Recombinant Human such as renal failure induced compro-
lation, and combinations of these criteria (43–
Activated Protein C (rhAPC) mise of red blood cell production.
45). In patients with septic shock, clinicians
should consider administering a dose of dexa- 1. rhAPC is recommended in patients at high Grade B
methasone until such time that an ACTH stim- risk of death (Acute Physiology and Chronic
ulation test can be administered because dexa- Health Evaluation II 25, sepsis-induced Rationale. No specific information regard-
methasone, unlike hydrocortisone, does not multiple organ failure, septic shock, or ing erythropoietin use in septic patients is avail-
interfere with the cortisol assay. sepsis-induced acute respiratory distress able, but clinical trials in critically ill patients
Crit Care Med 2004 Vol. 32, No. 3 863
7. show some decrease in red cell transfusion re- with the goal of maintaining end- crease PaO2 when positive end-expiratory pres-
quirement with no effect on clinical outcome inspiratory plateau pressures 30 cm sure is applied through either an endotracheal
(54, 55). Patients with severe sepsis and septic H2O. (See Appendix C for a formula to tube or a face mask.
shock may have coexisting conditions that do calculate predicted body weight.)
warrant use of erythropoietin. 4. In facilities with experience, prone posi-
Grade B tioning should be considered in ARDS
3. Routine use of fresh frozen plasma to cor-
Rationale. Over the past 10 yrs, several patients requiring potentially injurious
rect laboratory clotting abnormalities in
multiple-center randomized trials have been levels of FIO2 or plateau pressure who are
the absence of bleeding or planned invasive
performed to evaluate the effects of limiting not at high risk for adverse conse-
procedures is not recommended.
inspiratory pressure through modulations in quences of positional changes.
Grade E tidal volume (60 – 63). These studies showed
differing results that may have been caused by Grade E
Rationale. Although clinical studies have
differences between airway pressures in the
not assessed the impact of transfusion of fresh Rationale. Several smaller studies and one
treatment and control groups (64, 65). The
frozen plasma on outcomes in critically ill larger study have shown that a majority of
largest trial of a volume- and pressure-limited
patients, professional organizations have rec- patients with ALI/ARDS respond to the prone
strategy showed a 9% decrease of all-cause
ommended fresh frozen plasma for coagulopa- position with improved oxygenation (72–76).
mortality in patients ventilated with tidal vol-
thy when there is a documented deficiency of The large multiple-center trial of prone posi-
umes of 6 mL/kg of predicted body weight
coagulation factors (increased prothrombin tioning for 7 hrs/day did not show improve-
(as opposed to 12 mL/kg) while aiming for a
time, international normalized ratio, or partial ment in mortality rates in patients with ALI/
plateau pressure 30 cm H2O (66).
thromboplastin time) and the presence of ac- ARDS; however, a post hoc analysis suggested
tive bleeding or before surgical or invasive improvement in those patients with the most
procedures (56 –58). 2. Hypercapnia (allowing PaCO2 to increase severe hypoxemia by PaO2/FIO2 ratio (75).
above normal, so-called permissive hy- Prone positioning may be associated with po-
4. Antithrombin administration is not rec- percapnia) can be tolerated in patients tentially life-threatening complications, in-
ommended for the treatment of severe with ALI/ARDS if required to minimize cluding accidental dislodgment of the endo-
sepsis and septic shock. plateau pressures and tidal volumes. tracheal tube and central venous catheters,
Grade B but these complications can usually be
Grade C avoided with proper precautions.
Rationale. A phase III clinical trial of high- Rationale. An acutely elevated PaCO2 may
dose antithrombin did not demonstrate any have physiologic consequences that include 5. Unless contraindicated, mechanically
beneficial effect on 28-day all-cause mortality vasodilation as well as an increased heart rate, ventilated patients should be maintained
in adults with severe sepsis and septic shock. blood pressure, and cardiac output. Allowing semirecumbent, with the head of the bed
High-dose antithrombin was associated with modest hypercapnia in conjunction with lim- raised to 45° to prevent the development
an increased risk of bleeding when adminis- iting tidal volume and minute ventilation has of ventilator-associated pneumonia.
tered with heparin (59). been demonstrated to be safe in small nonran-
5. In patients with severe sepsis, platelets domized series (67, 68). Patients treated in Grade C
should be administered when counts are larger trials that have the goal of limiting tidal
volumes and airway pressures have demon- Rationale. The semirecumbent position
5000/mm3 (5 109/L) regardless of has been demonstrated to decrease the inci-
apparent bleeding. Platelet transfusion strated improved outcomes, but permissive
hypercapnia was not a primary treatment goal dence of ventilator-required pneumonia (77).
may be considered when counts are Patients are laid flat for procedures, hemody-
5000 –30,000/mm3 (5–30 109/L) and in these studies (66). The use of hypercarbia is
limited in patients with preexisting metabolic namic measurements, and during episodes of
there is a significant risk of bleeding. hypotension. Consistent return to semirecum-
Higher platelet counts ( 50,000/mm3 acidosis and is contraindicated in patients
with increased intracranial pressure. Sodium bent position should be viewed as a quality
[50 109/L]) are typically required for indicator in patients receiving mechanical
surgery or invasive procedures. bicarbonate infusion may be considered in se-
lect patients to facilitate use of permissive ventilation.
Grade E hypercarbia.
6. A weaning protocol should be in place
Rationale. Guidelines for transfusion of and mechanically ventilated patients
3. A minimum amount of positive end-
platelets are derived from consensus opinion should undergo a spontaneous breathing
expiratory pressure should be set to pre-
and experience in patients undergoing chemo- trial to evaluate the ability to discon-
vent lung collapse at end-expiration. Set-
therapy. Recommendations take into account tinue mechanical ventilation when they
ting positive end-expiratory pressure
the etiology of thrombocytopenia, platelet dys- satisfy the following criteria: a) arous-
based on severity of oxygenation deficit
function, risk of bleeding, and presence of able; b) hemodynamically stable (with-
and guided by the FIO2 required to main-
concomitant disorders (56, 58). out vasopressor agents); c) no new po-
tain adequate oxygenation is one accept-
able approach. (See Appendix C.) Some tentially serious conditions; d) low
K. Mechanical Ventilation of experts titrate positive end-expiratory ventilatory and end-expiratory pressure
Sepsis-Induced Acute Lung pressure according to bedside measure- requirements; and e) requiring levels of
ments of thoracopulmonary compliance FIO2 that could be safely delivered with a
Injury (ALI)/ARDS face mask or nasal cannula. If the spon-
(to obtain the highest compliance, re-
1. High tidal volumes that are coupled with flecting lung recruitment). taneous breathing trial is successful,
high plateau pressures should be avoided consideration should be given for extu-
in ALI/ARDS. Clinicians should use as a Grade E bation (see Appendix D). Spontaneous
starting point a reduction in tidal vol- breathing trial options include a low
umes over 1–2 hrs to a “low” tidal vol- Rationale. Raising end-expiratory pressure level of pressure support with continu-
ume (6 mL per kilogram of predicted in ALI/ARDS keeps lung units open to partic- ous positive airway pressure 5 cm H2O or
body weight) as a goal in conjunction ipate in gas exchange (69 –71). This will in- a T-piece.
864 Crit Care Med 2004 Vol. 32, No. 3
8. Grade A duced if an intermittent assessment of the ier management of fluid balance in
depth of neuromuscular blockade is per- hemodynamically unstable septic patients.
Rationale. Recent studies demonstrate formed (92, 93).
that daily spontaneous breathing trials re- Grade B
duce the duration of mechanical ventilation
M. Glucose Control Rationale. Studies support the equivalence
(78 – 80). Although these studies had limited
of continuous and intermittent renal replace-
numbers of patients with documented ALI/
1. Following initial stabilization of patients ment therapies for the treatment of acute re-
ARDS, there is no reason to believe that
with severe sepsis, maintain blood glu- nal failure in critically ill patients (98, 99).In-
ALI/ARDS patients would have different out-
cose 150 mg/dL (8.3 mmol/L). Studies termittent hemodialysis may be poorly
comes from other critically ill patients. Suc-
supporting the role of glycemic control tolerated in hemodynamically unstable pa-
cessful completion of spontaneous breath-
have used continuous infusion of insulin tients. There is no current evidence to support
ing trials leads to a high likelihood of and glucose. With this protocol, glucose the use of continuous venovenous hemofiltra-
successful discontinuation of mechanical should be monitored frequently after ini- tion for the treatment of sepsis independent of
ventilation. tiation of the protocol (every 30 – 60 renal replacement needs.
mins) and on a regular basis (every 4 hrs)
L. Sedation, Analgesia, and once the blood glucose concentration
has stabilized. O. Bicarbonate Therapy
Neuromuscular Blockade in
Sepsis Grade D 1. Bicarbonate therapy for the purpose of
improving hemodynamics or reducing
1. Protocols should be used when sedation Rationale. A large single-center trial of vasopressor requirements is not recom-
of critically ill mechanically ventilated postoperative surgical patients showed signif- mended for treatment of hypoperfusion-
patients is required. The protocol should icant improvement in survival when continu- induced lactic acidemia with pH 7.15.
include the use of a sedation goal, mea- ous infusion insulin was used to maintain The effect of bicarbonate administration
sured by a standardized subjective seda- glucose between 80 and 110 mg/dL (4.4 – 6.1 on hemodynamics and vasopressor re-
tion scale. mmol/L) (94). Exogenous glucose was begun quirement at lower pH as well as the
simultaneously with insulin with frequent effect on clinical outcome at any pH has
Grade B monitoring of glucose (every 1 hr) and inten- not been studied.
2. Either intermittent bolus sedation or sity of monitoring greatest at the time of ini- Grade C
continuous infusion sedation to prede- tiation of insulin. Hypoglycemia may occur.
There is no reason to think that these data are Rationale. There is no evidence to support
termined end points (e.g., sedation
not generalizable to all severely septic pa- the use of bicarbonate therapy in the treat-
scales) with daily interruption/lighten-
ing of continuous infusion sedation with tients. Post hoc data analysis of the trial data ment of hypoperfusion-induced acidemia as-
awakening and retitration, if necessary, revealed that although best results were ob- sociated with sepsis. Two studies comparing
are recommended methods for sedation tained when glucose was maintained between saline and bicarbonate in patients with pH
administration. 80 and 110 mg/dL (4.4 and 6.1 mmol/L), 7.13–7.15 failed to reveal any difference in
achieving a goal of 150 mg/dL (8.3 mmol/L) hemodynamic variables or vasopressor re-
Grade B also improved outcome when compared with quirements between equimolar concentra-
higher concentrations. This goal will likely tions of bicarbonate and normal saline with
Rationale (L1 and L2). Mechanically ven-
reduce the risk of hypoglycemia. The control either therapy (100, 101).
tilated patients receiving continuous sedation
may have a significantly longer duration of of the blood glucose concentration appears to
mechanical ventilation as well as ICU and hos- be more important than the amount of insulin P. Deep Vein Thrombosis
pital length of stay (81). A daily interruption or infused (95, 96). The frequency of blood glu- Prophylaxis
lightening of a “continuous” sedative infusion cose determinations may require the use of
until the patient is awake may decrease the central or arterial catheters for blood sam- 1. Severe sepsis patients should receive
duration of mechanical ventilation and ICU pling. deep vein thrombosis (DVT) prophylaxis
stay (82). The use of sedation protocols in with either low-dose unfractionated hep-
mechanically ventilated patients has shown a 2. In patients with severe sepsis, a strategy arin or low-molecular weight heparin.
of glycemic control should include a nu- For septic patients who have a contrain-
reduced duration of mechanical ventilation,
trition protocol with the preferential use dication for heparin use (i.e., thrombo-
length of stay, and tracheostomy rates (83).
of the enteral route. cytopenia, severe coagulopathy, active
3. Neuromuscular blockers should be bleeding, recent intracerebral hemor-
avoided if at all possible in the septic Grade E rhage), the use of a mechanical prophy-
patient due to the risk of prolonged neu- lactic device (graduated compression
Rationale. When a glycemic control strat- stockings or intermittent compression
romuscular blockade following discon- egy is initiated, hypoglycemia is minimized by device) is recommended (unless contra-
tinuation. If neuromuscular blockers providing a continuous supply of glucose sub- indicated by the presence of peripheral
must be used for longer than the first strate. Initially, unless the patient is already vascular disease). In very high-risk pa-
hours of mechanical ventilation, either profoundly hyperglycemia, this is accom- tients such as those who have severe
intermittent bolus as required or contin- plished with 5% or 10% dextrose infusion and sepsis and history of DVT, a combination
uous infusion with monitoring of depth followed by initiation of feeding, preferably by of pharmacologic and mechanical ther-
of block with train of four monitoring the enteral route, if tolerated (97). apy is recommended.
should be used.
Grade A
Grade E N. Renal Replacement
Rationale. Although no study has been
Rationale. Prolonged skeletal muscle 1. In acute renal failure, and in the absence performed specifically in patients with severe
weakness has been reported in critically ill of hemodynamic instability, continuous sepsis, large trials confirming the benefit of
patients following the use of intermediate- and venovenous hemofiltration and intermit- DVT prophylaxis in general ICU populations
long-acting neuromuscular blockers (84 –91). tent hemodialysis are considered equiva- have included significant numbers of septic
The risk of prolonged paralysis may be re- lent. Continuous hemofiltration offers eas- patients (102–104). This benefit should be ap-
Crit Care Med 2004 Vol. 32, No. 3 865
9. plicable to patients with severe sepsis and sep- ment of intravascular support (110). On the tension of the newborn and sepsis in a
tic shock. basis of a number of studies, it is accepted that randomized controlled trial (116). When pedi-
aggressive fluid resuscitation with crystalloids atric patients remain in a normotensive low
Q. Stress Ulcer Prophylaxis or colloids is of fundamental importance to cardiac output and high vascular resistance
survival of septic shock in children (111, 112). state, despite epinephrine and nitrosovasodi-
1. Stress ulcer prophylaxis should be given There is only one randomized, controlled trial lator therapy, then the use of a phosphodies-
to all patients with severe sepsis. H2 re- comparing the use of colloid to crystalloid terase inhibitor should be strongly considered
ceptor inhibitors are more efficacious resuscitation (dextran, gelatin, lactated Ring- (117–119). Pentoxifylline (not available in the
than sucralfate and are the preferred er’s solution, or saline) in children with den- United States) improved outcome in prema-
agents. Proton pump inhibitors have not gue shock (111). All these children survived ture neonates with sepsis when given for 6
been assessed in a direct comparison regardless of the fluid used, but the longest hrs/day for 5 days in a randomized, controlled
with H2 receptor antagonists and, there- time to recovery from shock occurred in chil- trial (120).
fore, their relative efficacy is unknown. dren who received lactated Ringer’s solution. 4. Therapeutic End Points. Therapeutic
They do demonstrate equivalency in abil- Among patients with the narrowest pulse pres- end points are capillary refill of 2 secs, nor-
ity to increase gastric pH. sure, there was a suggestion that colloids were mal pulses with no differential between pe-
Recommendation: Grade A more effective than crystalloids in restoring ripheral and central pulses, warm extremities,
normal pulse pressure. Fluid infusion is best urine output 1 mL·kg 1·hr 1, normal men-
Rationale. Although no study has been initiated with boluses of 20 mL/kg over 5–10 tal status, decreased lactate and increased base
performed specifically in patients with severe mins, titrated to clinical monitors of cardiac deficit, and superior vena cava or mixed ve-
sepsis, large trials confirming the benefit of output, including heart rate, urine output, nous oxygen saturation 70%. When employ-
stress ulcer prophylaxis in general ICU popu- capillary refill, and level of consciousness. ing measurements to assist in identifying ac-
lations have included significant numbers of Children normally have a lower blood pressure ceptable cardiac output in children with
septic patients (105–108). This benefit should than adults and can prevent reduction in systemic arterial hypoxemia such as cyanotic
be applicable to patients with severe sepsis and blood pressure by vasoconstriction and in- congenital heart disease or severe pulmonary
septic shock. In addition, the conditions creasing heart rate. Therefore, blood pressure disease, arterial-venous oxygen content differ-
shown to benefit from stress ulcer prophylaxis by itself is not a reliable end point for assess- ence is a better marker than mixed venous
(coagulopathy, mechanical ventilation, hypo- ing the adequacy of resuscitation. However, hemoglobin saturation with oxygen. Optimiz-
tension) are frequently present in patients once hypotension occurs, cardiovascular col- ing preload optimizes cardiac index. As noted
with severe sepsis and septic shock. lapse may soon follow. Hepatomegaly occurs previously, blood pressure by itself is not a
in children who are fluid overloaded and can reliable end point for resuscitation. If a pul-
R. Consideration for Limitation be a helpful sign of the adequacy of fluid monary artery catheter is used, therapeutic
of Support resuscitation. Large fluid deficits typically ex- end points are cardiac index 3.3 and 6.0
ist, and initial volume resuscitation usually L·min 1·m 2 with normal perfusion pressure
1. Advance care planning, including the requires 40 – 60 mL/kg but can be much (mean arterial pressure/central venous pres-
communication of likely outcomes and higher (112–114). sure) for age.
realistic goals of treatment, should be 3. Vasopressors/Inotropes (Should Only Be 5. Approach to Pediatric Septic Shock. Fig-
discussed with patients and families. De- Used After Appropriate Volume Resuscita- ure 1 shows a flow diagram summarizing an
cisions for less aggressive support or tion). Children with severe sepsis can present approach to pediatric septic shock (121).
withdrawal of support may be in the pa- with low cardiac output and high systemic 6. Steroids. Hydrocortisone therapy should
tient’s best interest. vascular resistance, high cardiac output and be reserved for use in children with catechol-
low systemic vascular resistance, or low car- amine resistance and suspected or proven ad-
Grade E
diac output and low systemic vascular resis- renal insufficiency. Patients at risk include
Rationale. It is too frequent that inade- tance shock. Depending on which situation children with severe septic shock and purpura
quate physician/family communication char- exists, inotropic support should be started in (122, 123), children who have previously re-
acterizes end-of-life care in the ICU. The level the case of fluid refractory shock or a combi- ceived steroid therapies for chronic illness,
of life support given to ICU patients may not nation of an inotrope together with a vaso- and children with pituitary or adrenal abnor-
be consistent with their wishes. Early and fre- pressor or a vasodilator. Dopamine is the first malities. There are no strict definitions, but
quent caregiver discussions with patients who choice of support for the pediatric patient with adrenal insufficiency in the case of cate-
face death in the ICU and their loved ones may hypotension refractory to fluid resuscitation. cholamine-resistant septic shock is assumed
facilitate appropriate application and with- The choice of vasoactive agent is determined at a random total cortisol concentration 18
drawal of life-sustaining therapies. by the clinical examination. Dopamine- g/dL (496 nmol/L). There is no clear consen-
refractory shock may reverse with epinephrine sus for the role of steroids or best dose of
S. Pediatric Considerations or norepinephrine infusion (114). Pediatric steroids in children with septic shock. A post
patients with low cardiac output states may 30- or 60-min ACTH stimulation test increase
1. Mechanical Ventilation. Due to low benefit from use of dobutamine. The use of in cortisol of 9 g/dL (248 nmol/L) also makes
functional residual capacity, young infants vasodilators can reverse shock in pediatric pa- that diagnosis. Two randomized controlled trials
and neonates with severe sepsis may require tients who remain hemodynamically unstable used “shock dose” hydrocortisone (25 times
early intubation (109). The principles of lung- with a high systemic vascular resistance state, higher than the stress dose) in children, both in
protective strategies are applied to children as despite fluid resuscitation and implementa- dengue fever. The results were conflicting (124,
they are to adults. In premature infants, addi- tion of inotropic support (114, 115). Nitroso- 125). Dose recommendations vary from 1–2
tional attention is paid to avoiding hyperox- vasodilators with a very short half-life (nitro- mg/kg for stress coverage (based on clinical di-
emia to prevent retinopathy. prusside or nitroglycerin) are used as first-line agnosis of adrenal insufficiency) to 50 mg/kg for
2. Fluid Resuscitation. Intravenous access therapy for children with epinephrine-resis- empirical therapy of shock followed by the same
for fluid resuscitation and inotrope/vasopres- tant low cardiac output and elevated systemic dose as a 24-hr infusion.
sor infusion is more difficult to attain in chil- vascular-resistance shock. Inhaled nitric oxide 7. Protein C and Activated Protein C. Pro-
dren than in adults. The American Heart As- reduced extracorporeal membrane oxygen- tein C concentrations in children reach adult
sociation has developed pediatric advanced life ation (ECMO) use when given to term neo- values at the age of 3 yrs. This might indicate
support guidelines for emergency establish- nates with persistent pulmonary artery hyper- that the importance of protein C supplemen-
866 Crit Care Med 2004 Vol. 32, No. 3
10. tation either as protein C concentrate or as
rhAPC is even greater in young children than
in adults. There has been one dose finding,
placebo-controlled study performed using pro-
tein C concentrate. This study was not pow-
ered to show an effect on mortality rate but
did show a positive effect on sepsis-induced
coagulation disturbances (126, 127). No ran-
domized studies using rhAPC have been per-
formed.
8. Granulocyte Macrophage Colony Stim-
ulating Factor. Growth factors or white blood
cell transfusions are given to patients with
neutropenic sepsis secondary to chemother-
apy or white blood cell primary immune defi-
ciency. A randomized, controlled trial showed
improved outcomes in neonates with sepsis
and an absolute neutrophil count 1500/ L
(1.5 109/L) treated with a 7-day course of
granulocyte macrophage colony stimulating
factor (128, 129).
9. DVT Prophylaxis. Most DVTs in young
children are associated with central venous
catheters. Femoral venous catheters are com-
monly used in children, and central venous
catheter-associated DVT occurs in approximately
25% of children with a femoral central venous
catheter. There are no data on use of heparin
prophylaxis to prevent DVT in children.
10. Stress Ulcer Prophylaxis. No studies
have been performed in children analyzing the
effect of stress ulcer prophylaxis. Studies have
shown that the rate of clinically important
gastrointestinal bleeding in children occurs at
rates similar to adults (130, 131). As in adults,
coagulopathy and mechanical ventilation are
risk factors for clinically important gastroin-
testinal bleeding. Stress ulcer prophylaxis
strategy is commonly used in mechanically
ventilated children, usually with H2 blockers.
Its effect is not known.
11. Renal Replacement Therapy. Continu-
ous venovenous hemofiltration may be clini-
cally useful in children with anuria/severe ol-
iguria and fluid overload, but no large RCTs
have been performed.
12. Glycemic Control. In general, infants
are at risk for developing hypoglycemia Figure 1. Resuscitation of pediatric septic shock. Adapted from Ref. 121. *Normalization of blood
when they depend on intravenous fluids. pressure and tissue perfusion; **hypotension, abnormal capillary refill, or extremity coolness.
This means that a glucose intake of 4 – 6
mg·kg 1·min 1 or maintenance fluid intake 15. Intravenous Immunoglobulin. Poly- sis on ECMO; eight of the 12 patients survived,
with glucose 10% in NaCl 0.45% is advised. clonal intravenous immunoglobulin has with six leading functionally normal lives at a
There are no studies in pediatric patients been reported to reduce mortality rate and is median of 1 yr (range, 4 months to 4 yrs) of
analyzing the effect of rigid glycemic con- a promising adjuvant in the treatment of follow-up. Children with sepsis on ECMO do
trol using insulin. This should only be done sepsis and septic shock. In children, how- not perform worse than children without sep-
with frequent glucose monitoring in view of ever, all the trials have been small, and the sis at long-term follow-up (133–135).
the risks for hypoglycemia. totality of the evidence is insufficient to
13. Sedation/Analgesia. Appropriate seda- support a robust conclusion of benefit. Ad-
tion and analgesia for children who are me- SUMMARY AND FUTURE
junctive therapy with monoclonal intrave-
chanically ventilated are the standard of care, nous immunoglobulins remains experimen- DIRECTIONS
although there are no data supporting any tal (132).
particular drugs or drug regimens. Although evidence-based recommenda-
16. ECMO. ECMO has been used in septic
14. Blood Products. In the absence of data, shock in children, but its impact is not clear. tions have been frequently published in the
it is reasonable to maintain hemoglobin con- Survival from refractory shock or respiratory medical literature, documentation of im-
centration within the normal range for age in failure associated with sepsis is 80% in neo- pact on patient outcome is limited. The
children with severe sepsis and septic shock nates and 50% in children. There is one study next phase of the Surviving Sepsis Cam-
( 10 g/dL [100 g/L]). analyzing 12 patients with meningococcal sep- paign is targeted to implement a core set of
Crit Care Med 2004 Vol. 32, No. 3 867