Special Articles

2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions

Conference
Mitchell M. Levy, MD, FCCP; Mitchell P. Fink, MD, FCCP; John C. Marshall, MD; Edward Abraham, MD;
Derek Angus, MD, MPH, FCCP; Deborah Cook, MD, FCCP; Jonathan Cohen, MD; Steven M. Opal, MD;
Jean-Louis Vincent, MD, FCCP, PhD; Graham Ramsay, MD; For the International Sepsis Definitions Conference

Objective: In 1991, the American College of Chest Physicians and symptoms of sepsis, cell markers, cytokines, microbiologic
(ACCP) and the Society of Critical Care Medicine (SCCM) convened a data, and coagulation parameters. The subgroups corresponded
“Consensus Conference,” the goals of which were “to provide a electronically before the conference and met in person during the
conceptual and a practical framework to define the systemic inflam- conference. A spokesperson for each group presented the delib-
matory response to infection, which is a progressive injurious pro- eration of each group to all conference participants during a
cess that falls under the generalized term ‘sepsis’ and includes plenary session. A writing committee was formed at the confer-
sepsis-associated organ dysfunction as well.” The general defini- ence and developed the current article based on executive sum-
tions introduced as a result of that conference have been widely used mary documents generated by each group and the plenary group
in practice and have served as the foundation for inclusion criteria for presentations. The present article serves as the final report of the
numerous clinical trials of therapeutic interventions. Nevertheless, 2001 International Sepsis Definitions Conference.
there has been an impetus from experts in the field to modify these Conclusion: This document reflects a process whereby a group
definitions to reflect our current understanding of the pathophysiol- of experts and opinion leaders revisited the 1992 sepsis guide-
ogy of these syndromes. lines and found that apart from expanding the list of signs and
Design: Several North American and European intensive care symptoms of sepsis to reflect clinical bedside experience, no
societies agreed to revisit the definitions for sepsis and related evidence exists to support a change to the definitions. This lack
conditions. This conference was sponsored by the SCCM, The of evidence serves to underscore the challenge still present in
European Society of Intensive Care Medicine (ESICM), The Amer- diagnosing sepsis in 2003 for clinicians and researchers and also
ican College of Chest Physicians (ACCP), the American Thoracic provides the basis for introducing PIRO as a hypothesis-generat-
Society (ATS), and the Surgical Infection Society (SIS). ing model for future research. (Crit Care Med 2003; 31:1250 –1256)
Methods: The conference was attended by 29 participants KEY WORDS: sepsis; severe sepsis; septic shock; systemic in-
from Europe and North America. In advance of the conference, flammatory response syndrome; PIRO
five subgroups were formed to evaluate the following areas: signs

I
n 1991, the American College of sepsis-associated organ dysfunction as The 1992 statement from the ACCP/
Chest Physicians (ACCP) and the well” (1). Conference participants, under SCCM Consensus Conference introduced
Society of Critical Care Medicine the chairmanship of Roger C. Bone, MD, into common parlance the Systemic In-
(SCCM) convened a “Consensus sought to provide a broad series of defi- flammatory Response Syndrome (SIRS).
Conference” in an attempt “to provide a nitions that might ultimately improve The acronym provided a reference for the
conceptual and a practical framework to our collective ability to diagnose, moni- complex findings that result from a sys-
define the systemic inflammatory re- tor, and treat sepsis. Bone et al. also temic activation of the innate immune re-
sponse to infection, which is a progres- addressed the need for a formal sepsis sponse, regardless of cause. The statement
sive injurious process that falls under the research agenda to include the “standard- hypothesized that SIRS is triggered by lo-
generalized term ‘sepsis’ and includes ization of research protocols.” calized or generalized infection, trauma,

From Brown Medical School/Rhode Island Hospital, ment of Intensive Care, Erasme University Hospital, Brus- Malangoni, MD; John C. Marshall, MD; George Matus-
Medical Intensive Care Unit, Providence, RI (MML); Uni- sels, Belgium (J-LV); Department of Surgery, University chak, MD, FCCP; Steven M. Opal, MD; Joseph E. Parrillo,
versity of Pittsburgh Medical Center, Division of Critical Hospital, Maastricht, The Netherlands (GR). MD, FCCM, FCCP; Konrad Reinhart, MD; William J. Sib-
Care Medicine, Pittsburgh, PA (MPF, DA); Toronto General Conference participants: Mitchell M. Levy, MD, bald, MD, FCCM, FCCP; Charles L. Sprung, MD, JD,
Hospital, Division of Cellular and Molecular Biology, To- FCCM, FCCP (Co-Chair); Graham Ramsay, MD (Co-Chair); FCCM, FCCP; Jean-Louis Vincent, MD, PhD, FCCM, FCCP;
ronto, Ontario, Canada (JCM); University of Colorado, Edward Abraham, MD, FCCM; Derek Angus, MD, FCCP; Max H. Weil, MD, PhD, FCCM, FCCP.
Health Science Center, Denver, CO (EA); McMaster Uni- Robert Balk, MD, FCCP; Gordon Bernard, MD, FCCP; Address requests for reprints to: Mitchell M. Levy,
versity/St. Joseph’s Hospital, Division of Critical Care Julian Bion, MD; Joseph Carcillo, MD; Jean M. Carlet, MD, Rhode Island Hospital, Brown University School of
Medicine, Ontario, Canada (DC); Division of Investigative MD; Jonathan Cohen, MD; Deborah Cook, MD, FCCP; Medicine, 593 Eddy Street, Providence, RI 02903.
Sciences, Imperial College of Medicine, London, UK (JC); Jean-François Dhainaut, MD; Tim Evans, MD, FRCP, Copyright © 2003 by Lippincott Williams & Wilkins
Brown Medical School/Memorial Hospital of Rhode Island, FCCP; Mitchell P. Fink, MD, FCCM, FCCP; Donald E. Fry, DOI: 10.1097/01.CCM.0000050454.01978.3B
Infectious Disease Division, Providence, RI (SMO); Depart- MD; Herwig Gerlach, MD, PhD; Steve Lowry, MD; Mark A.

1250 Crit Care Med 2003 Vol. 31, No. 4

thermal injury, or sterile inflammatory the strengths and weaknesses of the cur- sult, the diagnostic criteria for AMI have
processes, i.e., acute pancreatitis. SIRS is rent definitions of sepsis and related con- been revised (9).
considered to be present when patients ditions. The second goal focused on the Unfortunately, a clinically useful set of
have more than one of the following clini- identification of ways to improve the cur- criteria for diagnosing sepsis and related
cal findings: body temperature, ⬎38°C or rent definitions. The final goal sought to conditions will necessarily be somewhat
⬍36°C; heart rate, ⬎90 min⫺1; hyperven- identify methodologies for increasing the arbitrary. There is no “gold standard”
tilation evidenced by a respiratory rate of accuracy, reliability, and/or clinical util- (such as the infarcted myocardium)
⬎20 min⫺1 or a PaCO2 of ⬍32 mm Hg; and ity of the diagnosis of sepsis. against which the diagnostic criteria can
a white blood cell count of ⬎12,000 cells The conference was held in Washing- be calibrated. Diagnostic criteria will be
␮L⫺1 or ⬍4,000 ␮L⫺1. ton, DC, in December 2001 and included judged successful if clinicians regard
The SIRS concept has been globally 29 participants from Europe, North them as an aid for decision-making at the
adopted by clinicians and investigators. A America, and the United Kingdom. Before bedside. The diagnostic scheme requires
MEDLINE search dated January 1992–May convening, five subgroups were formed sufficient sensitivity and specificity to be
2002 yielded almost 800 publications that to evaluate the signs and symptoms of a clinical aid.
mention SIRS in the title or the abstract. sepsis, cell markers, cytokines, microbio-
We did not record all MEDLINE terms used logic data, and coagulation parameters. SIRS
to identify all citations relevant to SIRS. Subgroup participants corresponded
Our goal was not to conduct a systematic electronically before meeting in person at The SIRS concept is valid to the extent
review of the literature, which does dictate the conference. A subgroup spokesperson that a systemic inflammatory response
the need to record the search strategy. presented individual deliberations to all can be triggered by a variety of infectious
Bone et al. defined sepsis as SIRS plus conference participants during plenary and noninfectious conditions. Signs of
infection, “severe sepsis” as sepsis associ- sessions. A writing committee, formed at systemic inflammation can and do occur
ated with organ dysfunction, hypoperfu- the conference, developed this article in the absence of infection among pa-
sion, or hypotension, and “septic shock” based on subgroup executive summary tients with burns, pancreatitis, and other
as sepsis with arterial hypotension, de- documents and the plenary sessions. Ad- disease states. However, the specific cri-
spite adequate fluid resuscitation. These ditional information was introduced for teria proposed in the 1992 consensus def-
general definitions are now widely used participant review after the conference initions are widely considered to be too
in practice and serve as the basis for nu- during telephone, E-mail, and live discus- nonspecific to be of utility in diagnosing a
merous clinical trial inclusion criteria. sions. This article serves as the final re- cause for the syndrome or in identifying a
Recent trial data relating to a number of port of the 2001 International Sepsis Def- distinct pattern of host response (2, 3).
new interventions have created a need to initions Conference. Although the clinical manifestations
revisit and modify the 1992 definitions to of systemic inflammation are protean,
better reflect our understanding of the the biochemical features may be more
pathophysiology of these syndromes (2, Definitions consistent. Investigators have detected
3). In addition, many clinicians believe elevated circulating levels of interleukin
that the 1992 consensus definition does Establishing working definitions for a (IL)-6 (10), adrenomedullin (11), soluble
not provide a clear definition of sepsis. A syndrome is inherently an imperfect pro- (s)CD14, sELAM-1, MIP-1␣ (12), extracel-
recent European Society of Intensive cess and one that requires periodic up- lular phospholipase A2 (13), and C-reac-
Care Medicine (ESICM)/SCCM physician dating on the basis of new insights into tive protein (14) in patients meeting the
attitudinal survey revealed that 71% of pathophysiology or the availability of new 1992 SIRS criteria. In the future, if sup-
respondees cited no common definition diagnostic tests. We point to the example ported by further epidemiologic data, it
of sepsis (4), despite the ACCP/SCCM of acute myocardial infarction (AMI) as a may be possible to use purely biochemi-
consensus conference criteria for sepsis, disease paradigm to illustrate this point. cal and/or immunologic, rather than
severe sepsis, and septic shock (1). Generally accepted diagnostic criteria for clinical, criteria to identify the inflamma-
This gap in clinician understanding AMI were formulated by the Joint Inter- tory response. It may be that inflamma-
and a concurrent increase in clinical trial national Society and Federation of Cardi- tion is present when the circulating con-
data provided the support needed for a ology/World Health Organization task centration of IL-6, procalcitonin (15–17),
review of the 1992 definitions of sepsis force in 1979 (5). Although AMI is easily or C-reactive protein are increased. No
and related conditions. The 2001 Interna- diagnosed when Q-waves are present on large prospective studies currently sup-
tional Sepsis Definitions Conference was the electrocardiogram, non-Q-wave AMI port such a conclusion.
sponsored by the SCCM, ESICM, ACCP, can be distinguished from unstable an-
American Thoracic Society, and the Sur- gina pectoris only by using biochemical Sepsis
gical Infection Society. Each of the spon- markers. Reflecting the contemporary
sors provided official representation at state of knowledge, the World Health Or- In contrast to SIRS, it is very impor-
the conference and during the prepara- ganization biochemical criterion for es- tant that clinicians and researchers have
tion of this article. tablishing the diagnosis of AMI was a the tools needed to recognize and diag-
total creatine kinase concentration nose sepsis promptly; effective therapies
Goals and Methods of greater than twice the upper limit of nor- for infection are widely and readily avail-
Conference mal (5). Subsequently, several more sen- able. As in 1992, we define sepsis to be the
sitive and specific biochemical markers of clinical syndrome defined by the presence
The overall goals of the conference myocardial cell death were introduced of both infection and a systemic inflam-
were threefold and began with a view of into clinical practice (6 – 8), and as a re- matory response. In considering whether

Crit Care Med 2003 Vol. 31, No. 4 1251

the diagnostic criteria for infection or Table 1. Diagnostic criteria for sepsis
systemic inflammation should be revised,
Infection,a documented or suspected, and some of the following:b
we adhere to several principles. The cri-
General variables
teria should be broadly useful both to Fever (core temperature ⬎38.3°C)
clinicians caring for patients at the bed- Hypothermia (core temperature ⬍36°C)
side and to researchers designing obser- Heart rate ⬎90 min⫺1 or ⬎2 SD above the normal value for age
vational studies and clinical trials to im- Tachypnea
Altered mental status
prove the understanding of sepsis and its Significant edema or positive fluid balance (⬎20 mL/kg over 24 hrs)
optimal treatment. The criteria should be Hyperglycemia (plasma glucose ⬎120 mg/dL or 7.7 mmol/L) in the absence of diabetes
sensitive enough to identify most patients Inflammatory variables
with the syndrome, while minimally sac- Leukocytosis (WBC count ⬎12,000 ␮L⫺1)
Leukopenia (WBC count ⬍4000 ␮L⫺1)
rificing inevitable specificity. The criteria
Normal WBC count with ⬎10% immature forms
should not be so cumbersome that clini- Plasma C-reactive protein ⬎2 SD above the normal value
cians will resist a commitment to mem- Plasma procalcitonin ⬎2 SD above the normal value
ory or application. Any laboratory- Hemodynamic variables
dependent criteria should use assays that Arterial hypotensionb (SBP ⬍90 mm Hg, MAP ⬍70, or an SBP decrease ⬎40 mm Hg in adults
or ⬍2 SD below normal for age)
either are widely available now or are
Sv៮ O2 ⬎70%b
likely to be generally available in the near Cardiac index ⬎3.5 L䡠min⫺1䡠M⫺23
future. The criteria should be applicable Organ dysfunction variables
to adult, pediatric, and neonatal patients. Arterial hypoxemia (PaO2/FIO2 ⬍300)
Infection. We defined infection as a Acute oliguria (urine output ⬍0.5 mL䡠kg⫺1䡠hr⫺1 or 45 mmol/L for at least 2 hrs)
Creatinine increase ⬎0.5 mg/dL
pathologic process caused by the invasion Coagulation abnormalities (INR ⬎1.5 or aPTT ⬎60 secs)
of normally sterile tissue or fluid or body Ileus (absent bowel sounds)
cavity by pathogenic or potentially patho- Thrombocytopenia (platelet count ⬍100,000 ␮L⫺1)
genic microorganisms. This definition, Hyperbilirubinemia (plasma total bilirubin ⬎4 mg/dL or 70 mmol/L)
Tissue perfusion variables
essentially the same one used in the 1992
Hyperlactatemia (⬎1 mmol/L)
document, is not perfect. For example, Decreased capillary refill or mottling
colitis caused by Clostridium difficile, re-
sults from overgrowth of this organism in WBC, white blood cell; SBP, systolic blood pressure; MAP, mean arterial blood pressure; Sv៮ O2,
the colon, which is certainly not sterile. mixed venous oxygen saturation; INR, international normalized ratio; aPTT, activated partial throm-
Furthermore, the clinical manifestations boplastin time.
of C. difficile colitis are not caused by the
a
Infection defined as a pathologic process induced by a microorganism; bSvO2 sat ⬎70% is normal
in children (normally, 75– 80%), and CI 3.5–5.5 is normal in children; therefore, NEITHER should be
bacteria invading normally sterile tissues
used as signs of sepsis in newborns or children; cdiagnostic criteria for sepsis in the pediatric
but, rather, by the cytopathic effects of an
population are signs and symptoms of inflammation plus infection with hyper- or hypothermia (rectal
exotoxin secreted by the organism. It is temperature ⬎38.5 or ⬍35°C), tachycardia (may be absent in hypothermic patients), and at least one
also important to point out that, fre- of the following indications of altered organ function: altered mental status, hypoxemia, increased
quently, infection is strongly suspected serum lactate level, or bounding pulses.
without being microbiologically con-
firmed. Accordingly, sepsis (i.e., infection
and the systemic response to it) may only sepsis. A high cardiac output is com- few, if any, patients in the early stages of
be strongly suspected, without being mi- monly observed following major surgical the inflammatory response to infection
crobiologically confirmed. procedures or multiple trauma. Arterial are diagnosed with sepsis via four arbi-
Systemic Inflammation in Response hypotension can be caused by many con- trary criteria. Instead, the clinician goes
to Infection. Because of the limitations of ditions other than sepsis, such as acute to the bedside, identifies myriad symp-
SIRS discussed above, we included a list left ventricular failure secondary to AMI toms, and regardless of an evident infec-
of possible signs of systemic inflamma- or hemorrhage. Coagulopathy can be tion, declares the patient to “look septic.”
tion in response to infection (Table 1). drug-induced and is associated with If no obvious source of infection exists,
Ultimately, this scheme seeks to codify many different diseases, in addition to the clinician then initiates a search for an
the physical and laboratory findings that sepsis. It is important that as a practitio- infectious origin of the signs and symp-
prompt an experienced clinician to con- ner “checks off the boxes” to establish the toms associated with sepsis. The use of
clude that an infected patient “looks sep- diagnosis of sepsis, only findings that the word “some” reflects the clinical re-
tic.” Findings indicative of early organ cannot be easily explained by other ality at the bedside, rather than an arbi-
dysfunction may be the first symptoms causes be included. The thresholds cho- trary list invented for the purpose of clin-
noted by clinicians when making this as- sen in Table 1 merits discussion. We have ical trial entry criteria. Should the
sessment. It is for this reason that we not chosen thresholds for each of the definition of sepsis reflect reality as seen
included findings such as hemodynamic criteria that are consistently abnormal in at the bedside, thereby facilitating a clin-
instability, arterial hypoxemia, oliguria, degree. The proposition is whether ical diagnosis, or should the definition
coagulopathy and altered liver function thresholds similar in degree of abnormal- enable investigators to develop clear and
tests among the list of criteria that can be ity confer similar prediction in sepsis. simple entry criteria for clinical trials? It
used to establish the diagnosis of sepsis. As a result, the group turned toward was the opinion of the group that facili-
It is important to emphasize that none the day-to-day “reality” for bedside clini- tating a bedside diagnosis should have
of the findings in Table 1 is specific for cians. Group consensus concluded that primacy over research entry criteria.

1252 Crit Care Med 2003 Vol. 31, No. 4

Severe Sepsis (Sepsis with Children and neonates maintain to regional lymph nodes (N), and distant
Organ Dysfunction) higher vascular tone than adults. There- metastases (M). Each domain is graded to
fore, the shock state occurs long before denote the extent of pathologic involve-
The definition of severe sepsis remains hypotension in children. Septic shock in ment. For any given tumor type, survival
unchanged and refers to sepsis compli- pediatric patients is defined as a tachycar- tends to correlate with certain TNM sub-
cated by organ dysfunction. Severe sepsis dia (may be absent in the hypothermic groups.
is now considered to be the most com- patient) with signs of decreased perfusion Using a variation of the TNM ap-
mon cause of death in noncoronary crit- including decreased peripheral pulses proach, we developed of a classification
ical care units. Approximately 150,000 compared with central pulses, altered scheme for sepsis— called PIRO—that
people die annually in Europe and alertness, flash capillary refill or capillary will stratify patients on the basis of their
⬎200,000 die annually in the United refill ⬎2 secs, mottled or cool extremi- Predisposing conditions, the nature and
States (18). Organ dysfunction can be de- ties, or decreased urine output (25). Hy- extent of the insult (in the case of sepsis,
fined using the definitions developed by potension is a sign of late and decompen- Infection), the nature and magnitude of
Marshall et al. (19) or the definitions used sated shock in children. the host Response, and the degree of con-
for the Sequential Organ Failure Assess-
comitant Organ dysfunction (Table 2). It
ment (SOFA) score (20). Organ dysfunc-
Developing a Staging System is important to emphasize that the PIRO
tion in severe sepsis in the pediatric pop-
concept is rudimentary; extensive testing
ulation can be defined using definitions for Sepsis
and further refinement will be needed
developed by Wilkinson et al. (21), Proulx
Despite the definitions for sepsis, se- before it can be considered ready for rou-
et al. (22), and Doughty et al. (23) or the
vere sepsis, and septic shock outlined tine application in clinical practice.
definitions used for the PEMOD and
above, these terms do not allow for pre- Predisposition. Premorbid factors
PELOD score (24).
cise characterization and staging of pa- have a substantial impact on outcome in
tients with this condition. A clinically sepsis, modifying both the disease pro-
Septic Shock cess and the approach taken to therapy.
useful staging system stratifies patients
Septic shock in adults refers to a state with a disease by both their baseline risk This point is emphasized by recent data
of acute circulatory failure characterized of an adverse outcome and their potential showing that genetic factors play a
by persistent arterial hypotension unex- to respond to therapy. Such systems, greater role in determining the risk of
plained by other causes. Hypotension is both formal and informal, are widely used premature mortality due to sepsis than
defined by a systolic arterial pressure be- in clinical medicine. Perhaps the best- they do in influencing the risk of prema-
low 90 mm Hg (or, in children, ⬍2 SD developed and most explicit approach to ture death from other common condi-
below normal for their age), a MAP ⬍60, disease stratification has evolved in on- tions, such as cancer or cardiovascular
or a reduction in systolic blood pressure cology. The TNM system, developed by diseases (27). Beyond genetic variability,
of ⬎40 mm Hg from baseline, despite Pierre Denoix in 1946 (26), classifies ma- however, the management of patients
adequate volume resuscitation, in the ab- lignant tumors based on descriptors for with sepsis, and hence the outcome of the
sence of other causes for hypotension. the primary tumor itself (T), metastases disease, is clearly influenced by factors

Table 2. The PIRO system for staging sepsis

Domain Present Future Rationale

Predisposition Premorbid illness with reduced Genetic polymorphisms in components In the present, premorbid factors impact on
probability of short term of inflammatory response (e.g., TIR, the potential attributable morbidity and
survival. Cultural or TNF, IL-1, CD14); enhanced mortality of an acute insult; deleterious
religious beliefs, age, sex. understanding of specific interactions consequences of insult heavily dependent
between pathogens and host diseases. on genetic predisposition (future).
Insult infection Culture and sensitivity of Assay of microbial products (LPS, Specific therapies directed against inciting
infecting pathogens; mannan, bacterial DNA); gene insult require demonstration and
detection of disease transcript profiles. characterization of that insult.
amenable to source control.
Response SIRS, other signs of sepsis, Nonspecific markers of activated Both mortality risk and potential to respond
shock, CRP. inflammation (e.g., PCT or IL-6) or to therapy vary with nonspecific measures
impaired host responsiveness (e.g., of disease severity (e.g., shock); specific
HLA-DR); specific detection of target mediator-targeted therapy is predicated on
of therapy (e.g., protein C, TNF, PAF). presence and activity of mediator.
Organ dysfunction Organ dysfunction as number Dynamic measures of cellular response Response to preemptive therapy (e.g.,
of failing organs or to insult—apoptosis, cytopathic targeting microorganism or early mediator)
composite score (e.g., hypoxia, cell stress. not possible if damage already present;
MODS, SOFA, LODS, therapies targeting the injurious cellular
PEMOD, PELOD). process require that it be present.

TLR, Toll-like receptor; TNF, tumor necrosis factor; IL, interleukin; LPS, lipopolysaccharide; SIRS, systemic inflammatory response syndrome; CRP,
C-reactive protein; PCT, procalcitonin; HLA-DR, human leukocyte antigen-DR; PAF, platelet-activating factor; MODS, multiple organ dysfunction
syndrome; SOFA, sepsis-related organ failure assessment; LODS, logistic organ dysfunction system; PEMOD, pediatric multiple organ dysfunction; PELOD,
pediatric logistic organ dysfunction.

Crit Care Med 2003 Vol. 31, No. 4 1253

such as the premorbid health status of host response has proven to be difficult to

T
the patient, the reversibility of concomi- characterize. Putative biologic markers of
tant diseases, and a host of religious and response severity include circulating lev- his document re-
cultural forces that shape the approach els of procalcitonin (16, 33), IL-6 (34, 35),
flects a process
toward therapy. It is also important to and many others. When a new mediator
appreciate that these multiple predispos- is identified, epidemiologic studies will be whereby a group
ing factors could influence both the inci- required to determine whether measure-
dence and the outcome in similar or con- ments of the compound can be useful for of experts and opinion lead-
flicting ways. They could also pose staging patients. Furthermore, the opti-
separate or different risks for each of the mal set of biologic markers for staging ers revisited the 1992 sepsis
different stages of infection, response, sepsis may depend on the nature of the guidelines and found that
and organ dysfunction. For example, im- therapeutic decision to be made. For ex-
munosuppression may increase a per- ample, an indicator of dysregulation of apart from expanding the
son’s risk of infection, decrease the mag- the coagulation system might be more
nitude of that person’s inflammatory valuable for making a decision about list of signs and symptoms of
response, and have no direct influence on whether to institute therapy with dro-
organ dysfunction. Similarly, a genetic trecogin alfa (activated) (36), whereas a
sepsis to reflect clinical bed-
polymorphism such as the TNF2 allele marker of adrenal dysfunction might be side experience, no evidence
may result in a more aggressive inflam- more useful for determining whether to
matory response to an invading organ- institute therapy with hydrocortisone exists to support a change to
ism. This might decrease a person’s risk (37).
of infection but increase that person’s Organ Dysfunction. By analogy with the definitions.
risk of an overly exuberant, and poten- the TNM system, the presence of organ
tially harmful, inflammatory response dysfunction in sepsis is similar to the
should that patient become infected. We presence of metastatic disease in cancer.
encourage researchers to explore further Certainly, the severity of organ dysfunc- tory of sepsis to define those variables
the complex interaction of the multiple tion is an important determinant of prog- that predict not only an adverse outcome
factors that predispose to the onset, nosis in sepsis (19, 38). Whether the se- but also the potential to respond to ther-
stages of progression, and outcome of verity of organ dysfunction can aid in apy. The parameters selected may well
sepsis. therapeutic stratification is less clear. vary depending on the aspect of sepsis
Infection. The site, type, and extent of Nevertheless, there is some evidence that being studied, being different, for exam-
the infection have a significant impact on neutralization of TNF, an early mediator ple, if the focus is the antibiotic treat-
prognosis. A bilateral bronchopneumonia in the inflammatory cascade, is more ef- ment of pneumonia, the evaluation of a
is a more extensive process than a local- fective in patients without significant or- novel inhibitor of tyrosine kinases, or the
ized pneumonia, and a generalized fecal gan dysfunction (39), whereas drotreco- optimizing of microcirculatory flow in
peritonitis is a more extensive process gin alpha (activated) may provide more sepsis. The methodologic challenge is at
than an appendicitis. By studying mortal- benefit to patients with greater as com- least as great as that faced by oncologists,
ity rates among patients randomized to pared with lesser disease burden (40). and the TNM system continues to evolve
receive placebo in recent randomized The modern organ failure scores can be more than half a century after its intro-
clinical trials of new agents for the adju- used to quantitatively describe the degree duction.
vant treatment of sepsis, it is apparent of organ dysfunction developing over the
that pneumonia and intra-abdominal in- course of critical illness (41). CONCLUSIONS
fections are associated with a higher risk The potential utility of the proposed
of mortality than are urinary tract infec- PIRO model lies in being able to discrim- The 2001 conference participants con-
tions. Patients with secondary nosoco- inate morbidity arising from infection vened with the belief that the body of
mial bacteremia experience a higher and morbidity arising from the response bench work since the 1991 sepsis defini-
mortality than those with catheter- to infection. Interventions that modulate tions conference may lead to a major
related or primary bacteremia (28). Sim- the response may impact adversely on the change in the definition of sepsis based
ilarly, there is evidence that the endoge- ability to contain an infection; con- on biomarkers. After a process of evi-
nous host response to Gram-positive versely, interventions that target the in- denced-based review and considerable de-
organisms differs from that evoked by fection are unlikely to be beneficial if the bate, the participants determined that the
Gram-negative organisms (29). Early morbidity impact is being driven by the use of biomarkers for diagnosing sepsis is
studies with antibodies directed against host response. Premorbid conditions es- premature. Given the length and focus of
endotoxin, for example, suggested that tablish a baseline risk, independent of the this article, we did not expand on how the
benefit was greatest in patients with infectious process, while acquired organ problem of defining sepsis has hampered
Gram-negative infection (30) or endotox- dysfunction is an outcome to be pre- progress. We realize that this issue has
emia (31) but that treatment might be vented. long been debated in the medical com-
harmful to patients with Gram-positive The PIRO system is proposed as a tem- munity, and we choose not to elaborate
infection (32). plate for future investigation and is a here.
Response. In general, current thera- work in progress, rather than a model to The primary issue debated was the im-
pies for sepsis target the host response, be adopted. Its elaboration will require portance of an accurate diagnosis of sep-
rather than the infecting organism. The extensive evaluation of the natural his- sis at the bedside when weighed against

1254 Crit Care Med 2003 Vol. 31, No. 4

Table 3. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference sis by rapid bedside assay for cardiac tropo-
nin T. Am Heart J 2002;133:596 –598
Primary consensus points 8. Puleo PR, Meyer D, Wathen C, et al: Use of a
● The current concepts of sepsis, severe sepsis, and septic shock seem to be robust definitions rapid assay of subforms of creatine kinase MB
and should remain as described 10 yrs ago. to diagnose or rule out acute myocardial
● Current definitions do not allow for precise staging of the host response to infection. infarction. N Engl J Med 2002; 331:561–566
● Signs and symptoms of sepsis are more varied than the initial criteria established in 1991.
● A list of these signs and symptoms, for the diagnosis of sepsis is presented. 9. Myocardial infarction redefined—A consen-
● The future lies in developing a staging system that will characterize progression of sepsis. A sus document of The Joint European Society
new system, PIRO, is proposed for characterizing and staging the host response to infection. of Cardiology/American College of Cardiol-
ogy Committee for the Redefinition of Myo-
SCCM, Society of Critical Care Medicine; ESICM, European Society of Intensive Care Medicine; cardial Infarction. J Am Coll Cardiol 2000;
ACCP, American College of Chest Physicians; ATS, American Thoracic Society; SIS, Surgical Infection 36:959 –969
Society; PIRO, stratifies patients on the basis of their predisposing conditions, the nature and extent 10. Taniguchi T, Koido Y, Aiboshi J, et al:
of the insult—in the case of sepsis, infection, the nature and magnitude of the host response, and the Change in the ratio of interleukin-6 to inter-
degree of concomitant organ dysfunction. leukin-10 predicts a poor outcome in pa-
tients with systemic inflammatory response
syndrome. Crit Care Med 1999; 27:
The fact that no new definitions for 1262–1264
the development of clear and simple en- 11. Ueda S, Nishio K, Minamino N, et al: In-
try criteria for clinical trials. Participants sepsis are introduced in this conference
creased plasma levels of adrenomedullin in
believe that the facilitation of bedside di- report is noteworthy. This document re- patients with systemic inflammatory re-
agnosis should have priority over stan- flects a process whereby a group of ex- sponse syndrome. Am J Respir Crit Care Med
dardized sepsis entry criteria for clinical perts revisited the 1992 sepsis consensus 1999; 160:132–136
trials. A standardized set of signs and definitions and found that apart from ex- 12. Stoiser B, Knapp S, Thalhammer F, et al:
symptoms that may aid enrollment into panding the list of signs and symptoms of Time course of immunological markers in
randomized trials remains to be devel- sepsis to reflect clinical bedside experi- patients with the systemic inflammatory re-
oped. Our conclusions can be summa- ence, no evidence exists to support any sponse syndrome: Evaluation of sCD14, sV-
rized as follows (Table 3). change in the definitions. This lack of CAM-1, sELAM-1, MIP-1 alpha and TGF-beta
evidence serves to underscore the chal- 2. Eur J Clin Invest 1998; 28:672– 678
lenge for clinicians and researchers still 13. Hietaranta A, Kemppainen E, Puolakkainen
1. Current concepts of sepsis, severe P, et al: Extracellular phospholipases A2 in
sepsis, and septic shock remain useful present in diagnosing sepsis in 2003 and
relation to systemic inflammatory response
to clinicians and researchers. Until fur- also provides the basis for introducing
syndrome (SIRS) and systemic complications
ther evidence arises that justifies alter- PIRO as a hypothesis-generating model
in severe acute pancreatitis. Pancreas 2002;
for future research.
ing these categories that describe the 18:385–391
host response to infection, they should 14. Takala A, Jousela I, Olkkola KT, et al: Sys-
remain as described 10 yrs ago. REFERENCES temic inflammatory response syndrome
without systemic inflammation in acutely ill
2. These definitions do not allow for 1. Members of the American College of Chest
patients admitted to hospital in a medical
precise staging or prognostication of Physicians/Society of Crit Care Med Consen-
emergency. Clin Sci (Lond) 1999; 96:
the host response to infection. sus Conference Committee: American Col-
287–295
lege of Chest Physicians/Society of Crit Care
3. While SIRS remains a useful con- 15. Sablotzki A, Borgermann J, Baulig W, et al:
Med Consensus Conference: Definitions for
cept, the diagnostic criteria for SIRS Lipopolysaccharide-binding protein (LBP)
sepsis and organ failure and guidelines for
published in 1992 are overly sensitive and markers of acute-phase response in pa-
the use of innovative therapies in sepsis. Crit
tients with multiple organ dysfunction syn-
and nonspecific. Care Med 1992; 20:864 – 874
2. Marshall JC: SIRS and MODS: What is their drome (MODS) following open heart surgery.
4. An expanded list of signs and symp- Thorac Cardiovasc Surg 2001; 49:273–278
relevance to the science and practice of in-
toms of sepsis may better reflect the tensive care? Shock 2000; 14:586 –589 16. Harbarth S, Holeckova K, Froidevaux C, et al:
clinical response to infection. 3. Vincent J-L: Dear SIRS, I’m sorry to say that Diagnostic value of procalcitonin, interleu-
5. The operational definitions of sepsis I don’t like you. Crit Care Med 1997; 25: kin-6, and interleukin-8 in critically ill pa-
372–374 tients admitted with suspected sepsis. Am J
may be refined and tested in the future
4. Ramsay G, Gerlach H, Levy MM, et al: An Respir Crit Care Med 2001; 164:396 –340
as we increase our understanding of 17. Duflo F, Debon R, Monneret G, et al: Alveolar
international sepsis survey: A Study of doc-
the immunologic and biochemical and serum procalcitonin: Diagnostic and
tors’ knowledge and perception about sepsis.
characteristics of these conditions. prognostic value in ventilator-associated
Crit Care Med 2003; 31:
6. We hypothesize that improvements 5. Nomenclature and Criteria for Diagnosis of pneumonia. Anesthesiology 2002; 96:74 –79
in the management of critically ill pa- Ischemic Heart Disease: Report of the Joint 18. Angus DC, Linde-Zwirble WT, Lidicer J, et al:
tients with serious infections may fol- International Society and Federation of Car- Epidemiology of severe sepsis in the United
low the development of a staging sys- diology/World Health Organization Task States: Analysis of incidence, outcome, and
Force on Standardization of Clinical Nomen- associated costs of care. Crit Care Med 2001;
tem for sepsis that can better
clature. Circulation 1979; 59:607– 609 29:1303–1310
characterize the syndrome on the basis 19. Marshall JC, Cook DJ, Christou NV, et al:
6. Falahati A, Sharkey SW, Christensen D, et al:
of predisposing factors and premorbid Implementation of serum cardiac troponin I Multiple organ dysfunction score: A reliable
conditions, the nature of the underly- as marker for detection of acute myocardial descriptor of a complex clinical outcome.
ing infection, the characteristics of the infarction. Am Heart J 1999; 137:332–337 Crit Care Med 1995; 23:1638 –1652
host response, and the extent of the 7. Antman EM, Grudzien C, Mitchell RN, et al: 20. Ferreira FL, Bota DP, Bross A, et al: Serial
resultant organ dysfunction. Detection of unsuspected myocardial necro- evaluation of the SOFA score to predict out-

Crit Care Med 2003 Vol. 31, No. 4 1255

 come in critically ill patients. JAMA 2002; eter-related bacteremia: A cohort and case- IL-6 as a marker of severity. Ann Surg 1992;
286:1754 –1758 control study in critically ill patients. Am J 215:356 –362
21. Wilkinson JD, Pollack MM, Ruttiman, et al: Respir Crit Care Med 2001; 163:1584 –1590 35. Panacek EA, Kaul M: IL-6 as a marker of
Outcome of pediatric patient with multiple 29. Opal SM, Cohen J: Clinical gram-positive excessive TNF-alpha activity in sepsis. Sepsis
organ system failure. Crit Care Med 1986; sepsis: Does it fundamentally differ from 1999; 3:65–73
14:271–274 gram-negative bacterial sepsis? Crit Care 36. Bernard GR, Vincent J-L, Laterre PF, et al:
22. Proulx F, Fagan M, Farrell CA, et al: Epide- Med 1999; 27:1608 –1616 Efficacy and safety of recombinant human
miology of sepsis and multiple organ dys- 30. Ziegler EJ, Fisher CJ Jr, Sprung CL, et al: activated protein C for severe sepsis. N Engl
function syndrome in children. Chest 1996; Treatment of Gram-negative bacteremia and J Med 2001; 344:699 –709
109:1033–1037 septic shock with HA-1A human monoclonal 37. Annane D, Sébille V, Charpentier C, et al:
23. Doughty LA, Carcillo JA, Kaplan SS, et al: antibody against endotoxin: A randomized, Effect of treatment with low doses of hydro-
Plasma nitrite and nitrate concentration and double-blind, placebo-controlled trial. cortisone and Fludrocortisone on mortality
multiple organ failure in pediatric sepsis. N Engl J Med 1991; 324:429 – 436 in patients with septic shock. JAMA 2002;
Crit Care Med 1996; 109:1033–1037 31. Wortel CH, von der Mohlen MAM, van De- 288:862– 871
24. Leteutre S, LaCroix J, LeClerc F, et al: Pedi- venter SJH, et al: Effectiveness of a human 38. Vincent J-L, Moreno R, Takala J, et al: The
atric Logistic Dysfunction Score. Med Decis monoclonal anti-endotoxin antibody (HA-1A) SOFA (Sepsis-related Organ Failure Assess-
Making 1997 in gram-negative sepsis: Relationship to en- ment) score to describe organ dysfunction/
25. Carcillo JA, Task Force Members, Fields AI: dotoxin and cytokine levels. J Infect Dis failure. Intensive Care Med 1996; 22:707–710
Clinical Practice Parameters for Hemody- 1992; 166:1367–1374 39. Marshall JC, Panacek EA, Teoh L, et al: Mod-
namic Support of Pediatric and Neonatal Pa- 32. McCloskey RV, Straube RC, Sanders C, et al: eling organ dysfunction as a risk factor, out-
tients in Septic Shock. Crit Care Med 2002; Treatment of septic shock with human come, and measure of biologic effect in sep-
30;1–13 monoclonal antibody HA-1A: A randomized, sis. Crit Care Med 2001; 28:A46
26. Denoix PX: Enquete permanent dans les cen- double-blind, placebo-controlled trial. Ann 40. Eli Lilly and Company: Briefing Document
tres anticancereaux. Bull Inst Natl Hyg 1946; Intern Med 1994; 121:1–5 for XIGRIS for the Treatment of Severe Sep-
1:70 –75 33. Hausfater P, Garric S, Ayed SB, et al: Useful- sis. http:www.fda.gov/ohrms/dockets/ac/01/
27. Gospodarowicz M, Benedet L, Hutter RV, et ness of procalcitonin as a marker of systemic briefing/3797b1_01_Sponsor.htm, August 6,
al: History and international developments infection in emergency department patients: 2001
in cancer staging. Cancer Prev Cont 1998; A prospective study. Clin Infect Dis 2002; 41. Cook R, Cook DJ, Tilley J, et al: Multiple
2:262–268 34:895–901 organ dysfunction: Baseline and serial com-
28. Renaud B, Brun-Buisson C, ICU-Bacteremia 34. Damas P, Ledoux D, Nys M, et al: Cytokine ponent scores. Crit Care Med 2001; 29:
Study Group: Outcomes of primary and cath- serum level during severe sepsis in human 2046 –2050

1256 Crit Care Med 2003 Vol. 31, No. 4