-
1644 Definitions for Sepsis and Organ Faikire (Bone et a!)
accplsccm consensus conferenceDefinitions for Sepsis and Organ
Failure andGuidelines for the Use of Innovative Therapies
inSepsis
THE ACCP/SCCM CONSENSUS CONFERENCE COMMITTEE:RogerC. Bone, M.D.,
FG.C.P, ChairmanRobertA. Balk, M.D., F.C.C.PFrank B. Cerra, M.D.R.
Philip Dellinger, M. D. , F. C. C.P
An American College ofChest Physicians/Society of CriticalCare
Medicine Consensus Conference was held in North-brook in August
1991 with the goal of agreeing on a set ofdefinitions that could be
applied to patients with sepsis andits sequelae. New definitions
were offered for some terms,while others were discarded. Broad
definitions of sepsisand the systemic inflammatory response
syndrome wereproposed, along with detailed physiologic parameters
bywhich a patient may be categorized. Definitions for severesepsis,
septic shock, hypotension, and multiple organ dys-function syndrome
were also offered. The use of severity
Alan M. Fein, M.D., F.C.C.PWilliam A. Knaus; M.D.Roland M. H.
Schein, M.D.William j Sibbald, M.D. , F. C.C.P
scoring methods when dealing with septic patients wasrecommended
as an adjunctive tool to assess mortality.Appropriate methods and
applications for the use andtesting of new therapies were
recommended. The use ofthese terms and techniques should assist
clinicians andresearchers who deal with sepsis and its
sequelae.
(Chest 1992; 101:1644-55)
MODS = multiple organ dysfunction syndrome; SIRSsystemic
inflammatory response syndrome
n American College of Chest Physicians/Society ofCritical Care
Medicine Consensus Conference
was held in Chicago in August 1991 with the goal ofagreeing on a
set of definitions that could be appliedto patients with sepsis and
its sequelae. It was thegoal of this conference to provide both a
conceptualand a practical framework to define the
systemicinflammatory response to infection, a progressive,injurious
process that falls under the generalized termsepsis and includes
sepsis-associated organ dysfunc-tions as well. We expect that the
broad definitionsproposed in this report will improve our ability
tomake early bedside detection of the disease possible,and thus
allow early therapeutic intervention. In
For editorial comment see page 1481
addition, the standardization ofresearch protocols willbe
possible, as will improved dissemination and appli-cation
ofinformation derived from clinical studies. Wehope that the
continuing research on the inflammatoryresponse to infection will
allow us to understand thecellular and immunologic mechanisms that
cause sep-sis and related organ dysfunctions and, sometimes,death.
We recognize the limitations of the definitionswe have proposed and
urge further studies to validatethese clinical concepts, critical
phases, and measures
Reprint requests: Dr. Bone, Rush-Presbyterian St. b.ske
MedicalCenter, 1753 West Congress Parkway, Chicago 60612
of inflammation by using more sophisticated riskstratification
and other tools of evaluation.
Two other issues are also addressed in this articleand serve to
round out the discussion of the causesand treatment of sepsis: (1)
the utilization of severity-of-illness scoring systems that allow
the consistentevaluation, description, and risk prognostication
ofpatients with sepsis; and (2) guidelines for the use ofinnovative
therapies in severe sepsis.
SEPSIS
The systemic response to infection has been termedsepsis. 1.2
Sepsis is an increasingly common cause ofmorbidity and mortality,
particularly in elderly, im-munocompromised, and critically ill
patients. Sep-sis has been reported to be the most common causeof
death in the noncoronary intensive care unit.46 Itsrising
incidence, new etiologies, and appearance innew populations of
patients have been related tochanging demographics and the
increased use of morepotent and broader-spectrum antibiotics,
immunosup-pressive agents, and invasive technology in the
treat-ment of inflammatory, infectious, and neoplastic dis-eases.34
Recent clinical trials have been undertakento evaluate both
conventional and innovative therapiesin the treatment
ofsepsis.7#{176} However, interpretationsof these results have been
obscured by the use ofvarying definitions for the following terms:
infection,bactenmia, sepsis, septicemia, septic syndrome, and
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
BLOOD BORNE INFECTION
CHEST I 101 I 6 I JUNE, 1992 1645
septic shock.72 An additional source of confusion hasbeen the
application of the terms sepsis and septicsyndrome to noninfectious
inflammatory states. 13.14Several editorials and position papers
have recentlyattempted to provide a framework for the
standardi-zation and simplification of this termino1ogy37 Toadvance
these processes, this consensus conferencewill offer
recommendations for the standardization ofterminology.
The standardization of terminology is necessary toeliminate
confusion in communication for both clini-cians and researchers
concerning sepsis and its se-quelae. By standardizing terms, such
as sepsis, theability to compare protocols and evaluate
therapeuticinterventions is significantly improved. The
followingdefinitions should be used as general guidelines in
thedesign of future investigations into potential newdiagnostic and
treatment modalities.
Recommendation 1
The term sepsi in popular usage, implies a clinicalresponse
arising from infection. It is apparent that asimilar, or even
identical, response can arise in theabsence ofinfection. We
therefore propose the phrasesystemic inflammatory response syndrome
(SIRS) todescribe this inflammatory process, independent ofits
cause (Fig 1).
This systemic inflammatory response can be seenfollowing a wide
variety of insults and includes, but is
not limited to, more than one of the following
clinicalmanifestations: (1) a body temperature greater
than38#{176}Cor less than 36#{176}C;(2) a heart rate greater than
90beats per minute; (3) tachypnea, manifested by arespiratory rate
greater than 20 breaths per minute,or hyperventilation, as
indicated by a PaCO2 of lessthan 32 mm Hg; and (4) an alteration in
the whiteblood cell count, such as a count greater than 12,000/cu
mm, a count less than 4,000/cu mm, or the presenceof more than 10
percent immature neutrophils(bands). These physiologic changes
should representan acute alteration from baseline in the absence
ofother known causes for such abnormalities, such aschemotherapy,
induced neutropenia, and leukopenia.
Rationale: The systemic inflammatory response isseen in
association with a large number of clinicalconditions. Besides the
infectious insults that mayproduce SIRS, noninfectious pathologic
causes mayinclude pancreatitis, ischemia, multiple trauma andtissue
injury, hemorrhagic shock, immune-mediatedorgan injury, and the
exogenous administration of suchputative mediators of the
inflammatory process astumor necrosis factor and other
cytokines.
A frequent complication ofSIRS is the developmentof organ system
dysfunction, including such well-defined clinical conditions as
acute lung injury, shock,renal failure, and multiple organ
dysfunction syn-drome (MODS). The term MODS also stems from
thisconsensus conference, and its definition will be dis-
FIGURE 1. The interrelationship between systemic inflammatory
response syndrome (SIRS), sepsis, andinfection.
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
1646 Defindions for Sepsis and Organ Failure (Bone eta!)
cussed later in this report.It is likely that similar
pathogenesis and pathophys-
iology underlie the various clinical entities that con-stitute
SIRS . Future definitions may take into accountthe pathogenetic
mechanism in descriptions of theresponse. Further work is needed to
characterize theclinical and prognostic significance of SIRS and
itsassociated sequelae.
Recommendation 2
When SIRS is the result of a confirmed infectiousprocess, it is
termed sepsis. In this clinical circum-stance, the term sepsLs
represents the systemic inflam-matory response to the presence of
infection.
Rationale: Sepsis has been well recognized as asystemic
inflammatory response to an active infectiousprocess in the host.
The use of a broad-based clinicaldefinition of the septic process
may facilitate studiesof the pathogenetic mechanisms involved in
the pro-duction of the systemic inflammatory response toinfection,
as well as the noninfectious causes of SIRS.An improved
understanding of these mechanisms willlead to improved therapeutic
management.
Recommendation 3
In an attempt to improve the written and verbalcommunication
concerning infection as it relates toSIRS, we recommend the
adoption of the followingnomenclature and definitions for several
commonlyused terms (Table 1):
infection is a microbial phenomenon characterizedby an
inflammatory response to the presence ofmicroorganisms or the
invasion of normally sterilehost tissue by those organisms.
Bacteremia is the presence of viable bacteria in theblood. The
presence of viruses, fungi, parasites, andother pathogens in the
blood should be described in asimilar manner (ie, viremia,
fungemia, parasitemia,etc).
Septicemia has been defined in the past as thepresence of
microorganisms or their toxins in theblood. However, this term has
been used clinicallyand in the medical literature in a variety of
ways,which has added to confusion and difficulties in
datainterpretation. Septicemia also does not adequatelydescribe the
entire spectrum of pathogenic organismsthat may infect the blood.
We therefore suggest thatthis term be eliminated from current
usage.
Sepsis is the systemic inflammatory response toinfection. In
association with infection, manifestationsof sepsis are the same as
those previously defined forSIRS, and include, but are not limited
to, more thanone of the following: (1) a temperature greater
than38#{176}Cor less than 36#{176}C;(2) an elevated heart
rategreater than 90 beats per minute; (3) tachypnea,manifested by a
respiratory rate greater than 20
breaths per minute or hyperventilation, as indicatedby a PaCO2
ofless than 32 mm Hg; and (4) an alterationin the white blood cell
count, such as a count greaterthan 12,000/cu mm, a count less than
4,000/cu mm;or the presence of more than 10 percent
immatureneutrophils. To help identify these manifestations
assepsis, it should be determined whether they are apart of the
direct systemic response to the presenceofan infectious process.
Also, the physiologic changesmeasured should represent an acute
alteration frombaseline in the absence ofother known causes for
suchabnormalities.
Recommendation 4
Sepsis and its sequelae represent a continuum ofclinical and
pathophysiologic severity. The degree ofseverity may independently
affect prognosis. Someclinically recognizable stages along this
continuumthat may adversely affect prognosis include the
follow-ing:
Severe sepsis is defined as sepsis associated withorgan
dysfunction, hypoperfusion abnormality, or sep-sis-induced
hypotension. Hypoperfusion abnormali-ties include lactic acidosis,
oliguria, and acute altera-tion of mental status.
Sepsis-induced hypotension is defined by the pres-
Table 1 -Definitions
Infection = microbial phenomenon characterized by an
inflamma-tory response to the presence of microorganisms or the
invasionof normally sterile host tissue by those organisms.
BaCt#{128}I-emia = the presence ofviable bacteria in the
blood.Systemic inflammatory response syndrome (SIRS)= the
systemic
inflammatory response to a variety of severe clinical insults.
Theresponse is manifested by two or more ofthe following
conditions:(1) temperature >38#{176}Cor 90 beats perminute; (3)
respiratory rate >20 breaths per minute or PaCO,12,000/cu mm,10%
immature (band) forms
Sepsi.s = the systemic response to infection, manifested by two
ormore of the following conditions as a result of infection:
(1)temperature >38#{176}C or 90 beats perminute; (3) respiratory
rate >20 breaths per minute or PaCO,12,00Wcu mm,10% immature
(band) forms.
Severe sepsis= sepsis associated with organ dysfunction,
hypoper-fusion, or hypotension. Hypoperfusion and perfusion
abnormali-ties may include, but are not limited to lactic acidosis,
oliguria,or an acute alteration in mental status.
Septic shock= sepsis-induced with hypotension despite
adequatefluid resuscitation along with the presence of perfusion
abnor-malities that may include, but are not limited to, lactic
acidosis,oliguria, or an acute alteration in mental status.
Patients who arereceiving inotropic or vasopressor agents may not
be hypotensiveat the time that perfusion abnormalities are
measured.
Sepsis-induced hypotension a systolic blood pressure
-
CHEST I 101 I 6 I JUNE. 1992 1647
ence of a systolic blood pressure of less than 90 mmHg or its
reduction by 40 mm Hg or more frombaseline in the absence ofother
causes for hypotension(eg, cardiogenic shock).
Septic shock is a subset of severe sepsis and isdefined as
sepsis-induced hypotension, persisting de-spite adequate fluid
resuscitation, along with thepresence of hypoperfusion
abnormalities or organdysfunction. Patients receiving inotropic or
vasopres-sor agents may no longer be hypotensive by the timethey
manifest hypoperfusion abnormalities or organdysfunction, yet they
would still be considered to haveseptic shock.
Rationale: Recent, large-scale, multicenter, pro-spective
studies of sepsis have suggested that there isa continuum of
severity encompassing both infectiousand inflammatory components.
The condition beginswith infection and potentially leads to sepsis
withorgan system dysfunction and septic shock.7#{176} Bacte-remia
and hypotension may occur as a part of thisprocess. While
recognizing that the disease processforms a continuum of severity,
an analysis of severalclinical trials has indicated that definable
phases existon that continuum which characterize populations
atincreased risk of morbidity and mortahty.72 Onesuch phase should
be termed severe sepsis or sepsiswith organ system dysfunction.
Some have previouslyused the term septic syndmme to describe this
phaseof the septic process. However, the term septicsyndrome has
been applied to a variety of inflamma-tory states and it now
appears to be both confusingand 134 We, therefore, recommend
thatthe term septic syndrome no longer be used.
These critical stages in the septic process are likelyto have
independent prognostic implications;18.19 how-ever, this hypothesis
has not been tested in large-scale, prospective, multicenter
trials. Risk assessmentmay be a more appropriate approach to
identifyingpatients likely to develop morbidity and mortality.The
development and refinement of such evaluationtools is encouraged.
Previous studies have shown thatseptic shock, as defined above, is
associated withincreased mortality. 11.19.22
Conclusion
We have provided both a conceptual and a practicalframework for
the definition of the systemic inflam-matory response to infection
(sepsis). The applicationof these broad definitions will improve
early bedsidedetection and permit early intervention in sepsis.
Inaddition, the standardization ofresearch protocols willbe
enhanced, as will application ofinformation derivedfrom clinical
studies. We believe that the early iden-tification of the
inflammatory response to infectionwill enhance our understanding of
the cellular andimmunologic mechanisms that can cause sepsis
and
organ dysfunction and, in the most severe cases, death.Because
of the limitations that are inherent in thesedefinitions, we urge
further studies that utilize moresophisticated risk stratification
and other tools ofevaluation to validate the clinical concepts,
criticalphases, and measures of inflammation that are impor-tant
for the clinical treatment of sepsis.
MULTIPLE ORGAN DYSFUNCTION SYNDROME
The increasing incidence ofmorbidity and mortalitycaused by
multiple organ failure has paralleled im-provements in the
life-support technologies availableto patients admitted to an
intensive care unit (ICU).As newer technologies for the monitoring
and supportof patients sustaining life-threatening critical
illnessbecame established, retrospective clinical studiesfound that
a major threat to survival was not theunderlying illness, or even a
single complicationthereof, but rather a process ofprogressive
physiologicfailure of several interdependent organ systems.The
terms progressive or sequential organ failure,multiple organ
frilure,m and multiple systeims organf ailure were thereby
introduced to describe an evolv-ing clinical syndrome that was
characterized by thedevelopment of otherwise unexplained
abnormalitiesoforgan function in critically ill patients. The
phenom-enon that these terms describe is clearly increasing
inprevalence, as a result not only of improvements inlife-support
technology (both medications and devices)but also of the
application of these technologies to anincreasingly high-risk
patient population.
Conventional terminology is considered inadequateto accurately
characterize this syndrome. Thus, cmi-cal descriptions ofthe organ
failure syndrome emergedin an arbitrary and retrospective fashion.
Criteria fordefining abnormalities of specific organ function
havealso been widely dissimilar from one study to anotherand, for
the most part, have been predicated on theconcept of organ failure,
a dichotomous event that iseither present or absent, rather than
organ dysfunc-tion, a continuum of physiologic derangements.
Thestatic criteria used in current epidemiologic descrip-tions
preclude the possibility ofdynamically changingorgan function that
characterizes the syndrome as itis encountered clinically. This
issue, if it is not soonaddressed, could potentially hinder future
advancesin the treatment of this syndrome.
Early clinical studies of multiple organ failureidentified
occult infection as the most importantclinical correlate of the
syndrome.m However,recent work has shown that organ system
dysfunctioncan evolve in the absence of an untreated focus
ofinvasive infection and can be reproduced expen-mentally by the
infusion of a diverse spectrum ofendogenously derived mediators of
inflammation.#{176}Futhermore, recent work has demonstrated a
complex
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
condary MOD$
Recovery
1648 Definitions for Sepsis and Organ Failure (Bone et a!)
interrelationship among individual organs, such thatfailure of
one may establish an amplification processthat hastens injury to
another.
Our understanding of the pathophysiology of organdysfunction and
failure in critically ill patients isimproving. In contrast,
descriptions of the epidemi-ology of this syndrome remain meager.
Availablereports focus primarily on disease that is severe,perhaps
at a point in the disease course where inter-ventions may no longer
be anticipated to have potentialfor success.
The purpose of this statement is to propose aconceptual
framework for future studies ofthe clinicalphenomenon of organ
system dysfunction in criticalillness, and to lay the foundations
for common termi-nology and criteria to describe the syndrome.
Recommendation 1
The detection of altered organ function in theacutely ill
patient constitutes a syndrome that shouldbe termed multiple organ
dysfunction syndrome. Theterminology dyifunction identifies this
process as aphenomenon in which organ function is not capableof
maintaining homeostasis. This process, which maybe absolute or
relative, can be more readily identifiedas a continuum of change
over time. An example ofrelative organ dysfunction is found in the
patient witha normal cardiac output and systemic oxygen deliverywho
exhibits evidence of inadequate tissue oxygena-tion (eg, lactic
acidosis).
The proposed acronym also identifies multiple organdysfunction
as a syndrome. In this context, MODSis proposed to describe a
pattern of multiple andprogressive symptoms and signs that are
thought tobe pathogenetically related.
Rationale: In contrast to the static descriptions thathave been
used previously, the proposed change interminology emphasizes the
dynamic nature of theprocess under discussion. Thus, the following
points
Death
FIGURE 2. The different causes and results ofprimary and
secondarymultiple organ dysfunction syndrome (MODS).
must be recognized:1. MODS describes a continuum oforgan
dysfunc-
tion, although specific descriptions ofthis contin-uous process
are not currently available.
2. The recognition of early organ abnormalitiesmust be improved
so that treatment can beinitiated at earlier stages in the
evolution of thissyndrome.
3. Changes in organ function over time can beviewed as an
important element in prognostica-tion. When applied to MODS,
existing measuresofillness severity provide only a snapshot in
timeof this dynamic process, and are generally with-out reference
to the natural course ofthe disease.
4. MODS is subject to modulation by numerousfactors at varying
time periods, both interven-tional and host-related.
Recommendation 2
MODS may be described as being either primaryor secondary.
Rationale: MODS develops by two relatively dis-tinct, but not
mutually exclusive, pathways. PrimaryMODS is the direct result of a
well-defined insult inwhich organ dysfunction occurs early and can
bedirectly attributable to the insult itself. An example ofprimary
MODS is organ dysfunction as the immediateresult oftrauma (eg,
pulmonary contusion, renal failuredue to rhabdomyolysis, or the
coagulopathy due tomultiple transfusions). In primary MODS, the
partic-ipation ofan abnormal and excessive host
inflammatoryresponse in both the onset and progression of
thesyndrome is not as evident as it is in secondary MODS(Fig
2).
Secondary MODS develops, not in direct responseto the insult
itself, but as the consequence of a hostresponse, and is identified
within the context of SIRS.SIRS is also a continuous process, and
describes anabnormal host response that is characterized by
ageneralized activation of the inflammatory reaction inorgans
remote from the initial insult. When the processis due to
infection, the terms sepsis and SIRS aresynonymous. Given that
SIRS/sepsis is a continuousprocess, MODS may be understood to
represent themore severe end of the spectrum of severity of
illnessthat characterizes S I RS/sepsis. Thus, secondaryMODS
usually evolves after a latent period followingthe inciting injury
or event, and is most commonlyseen to complicate severe
infection.
Recommendation 3
Since criteria that are universally applicable inquantifying the
individual organ dysfunctions com-prised by MODS cannot be proposed
at this time, acomprehensive and continuously updated data base
toclinically test and validate optimal criteria for describ-
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
RISK DISTRIBUTIONS OF 519 ICU ADMISSIONSFOR SEPSIS ACCORDING TO
CATEGORICAL
DEFINITION OF SEPTIC SYNDROME *N OF CASES
0 10 20 30 40 50 60 70 80 90
DAY 1 HOSPITAL MORTALITY RISK
- NO DEFINITION(N211) DEFINITION (N.308)*
SEPTIC SYNDROME WITH OR WITHOUT SEPTICSHOCK (Bone,Crit Care Med
1989;17:389)
FIGURE 3. Risk distribution of 519 sepsis patients who either
met (n 308) or did not meet (n = 211) thecriteria for sepsis
syndrome (see reference 40 for further details).
CHEST I 101 I 6 I JUNE, 1992 1649
ing this syndrome must be established.Rationale: Data are
insufficient to justify a recom-
mendation of universally applicable criteria that couldserve as
a validated operational template. The assign-ment of criteria for
measuring organ dysfunctionshould not occur a priori, but should
result from anempiric process in which specific variables are
testedagainst outcome. By doing so, the predicted accuracyof
individual variables, groups of variables, and levelsofabnormality
can be defined in a manner that reflectscurrent clinical
practice.
SEVERITY-OF-ILLNESS SCORING SYSTEM
A common theme in the two previous sections ofthis consensus
statement is that we are treating moreseverely ill patients at
later stages of illness. It is alsoapparent that many of these
patients who have morecomplex illnesses may be suffering from a
combinationof chronic and acute disease. The nature of
diseasepresentation is changing, and patient or host responseis
exemplified by the proposed introduction of newsyndromes, such as
SIRS and MODS. The recognitionand treatment of these syndromes
would not havebeen possible without our advanced diagnostic
andlife-support capabilities. It is also emphasized, how-ever, that
patients with both of these syndromespresent somewhere along a
continuum of illness Se-verity, and that an accurate description of
that contin-uum is essential to appropriate usage of these
terms.
The rationale for using scoring systems, therefore,
is to ensure that the increased complexity of diseasein patients
currently being treated is consistentlyrepresented in evaluations
and descriptions. A specificgoal of severity scoring systems is to
use these impor-tant patient variables to describe the relative
risks ofpatients and, thereby, to identify where, along
thecontinuum of severity, the patient resides. This willreduce the
variation due to patient factors so that theincremental impact of
new or existing therapy can bemore precisely determined. It is also
hoped thatmore precise measurements of patient risk will leadto new
insights into disease processes and serve as atool with which
clinicians can more accurately monitorpatients and guide the use of
new therapies, such asmonoclonal antibodies.
In this regard, it is increasingly being recognizedthat the end
point of severity scoring can be morethanjust a score representing
the degree of physiologicdisturbance. Severity scoring can be used,
in conjunc-tion with other risk factors (eg, disease etiology
orpatient selection criteria), to anticipate and evaluateoutcomes,
such as hospital mortality.37 These proba-bility estimates can be
calculated at the time a patientpresents for care or for entry into
a clinical trial; thus,they can serve as a pretreatment control.
They canalso be updated during the course of therapy,
therebydescribing the course of illness and providing analternative
for the evaluation of response. The meth-ods for calculating these
dynamic probability estimatesare not as developed, however, as are
those for initial
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
8070
60
50
40
30
20
10
0
1650 Definitions for Sepsis and Organ Faikire (Bone eta!)
RISK DISTRIBUTIONS OF 519 ICU ADMISSIONSFOR SEPSIS ACCORDING TO
THE DEFINITION
OF SEPTIC INFLAMMATORY RESPONSE SYNDROMENo. of Patient8
- -- -----.--..- -- ----.--
1O 10- 20- 30.- 40- 50- 60- 70- 80- 90-
Icu Day 1 Mortality Risk
- SIRS (N503) No SIRS (N16)
FIGURE 4. Risk distribution of the same 519 sepsis patients as
in Figure 3, according to whether they metcriteria for SIRS.
or presentation risk assessment.To illustrate the value of
combining initial severity
scoring or probability risk estimation with the newlyproposed
definition for SIRS, the study group re-viewed the application of
hospital mortality risk esti-mation to a group of519 adult patients
with a primarydiagnosis of sepsis upon admission to medical
andsurgical ICU.#{176} These patients frequently lacked aninitial
microbial source of infection, such as bacterialpneumonia, but were
still identified and treated assuffering primarily from infection.
As such, theyrepresent a subgroup ofpatients with sepsis for
whomthe new definitions, such as SIRS, would be appropri-ate (Fig
1).
Figure 3 illustrates the distribution of hospitalmortality risk
calculated on the initial day of ICUtreatment for these 519
patients, according to whetherthe patient met criteria for the
definition of sepsissyndrome, as defined by Bone et al. It can be
seenthat the risk distribution of the 308 (59 percent)patients
meeting the criteria for this syndrome is notsubstantially
different from that for the 211 (41 per-cent) patients who did not
fulfill the criteria.
In contrast, the result depicted in Figure 4 illus-trates that
when the SIRS definition was applied tothe same 519 patients with a
primary clinical diagnosisof sepsis, it identified 96.9 percent
(503/519). Whatthis initial application of the new definition
hasachieved, therefore, is an increase in the number ofpatients
classffied by the definition. This is important,
since the previous definition (sepsis syndrome) ex-cluded many
of these patients, although their esti-mated risks were equivalent
to those of patients whowere included.#{176}
Finally, Figure 5 demonstrates that, for the 503patients meeting
the criteria for SIRS, the estimatedmortality risks calculated on
the initial day of ICUtreatment accurately predicted the subsequent
actualhospital mortality rates. This demonstrates the
currentcapability ofseverity scoring and risk estimation whenused
in combination with a broad, encompassingclinical definition like
SIRS. Further details on thesemethods are available#{176} and will
be the subject of asubsequent report from this consensus
conference.These findings led to the following recommendations.
Recommendation 1
Because of the increasing complexity of patientpresentation, the
use of new terminology, such asSIRS, with its various etiologies
(Fig 1) should becombined with risk stratification or probability
riskestimation techniques in order to measure the positionof an
individual patient along the continuum of sever-ity.
Rationale: The major change in patient presentationhas been an
increase in the complexity ofillness. Also,more severely ill
patients are being treated at laterstages of their illness.
Accurate identification of pre-treatment risk can improve the
precision of the eval-uation of new therapies. Such risk estimation
can also
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
10 10- 20- 30- 40- 50- 60- 70- 80- 90-ICU Day 1 Mortality
Risk
CHEST I 101 I 6 I JUNE, 1992 1651
be useful in monitoring the utilization ofnew therapiesand in
refining the indications for specific treatmentsby identifying risk
levels when certain therapiesappear to be efficacious. The use of
this approach isparticularly important when the patient is a
candidatefor, or a participant in, a clinical trial.
Recommendation 2
When patients are identified as having SIRS orMODS, sequential
(daily or more frequently) riskstratification or probability
estimation techniquesshould be applied to describe the course of
thesesyndromes.
Rationale: At our current level of understanding(and measurement
capabilities), we determine thecourse of SIRS by relying primarily
on sequentialmeasurements of physiologic changes. These
physio-logic changes correlate with subsequent outcome. Inthe
future, it may be possible to extend this measure-ment to include
metabolic changes. Such advancesin measurement capabilities may be
especially impor-taut in characterizing the course of MODS. As
em-phasized previously, the exact criteria for, and descrip-tion
of, MODS have yet to be determined.
Recommendation 3
Priority should be given to building on severityscoring and
other predictive and descriptive efforts.This will allow the
development of a comprehensive
model of disease progression that will have implica-tions for
the investigation of new syndromes such asSIRS and MODS.
RatiOflal4: The development of a comprehensivemodel for a
syndrome, such as SIRS, is a complexprocess. Such a model must
describe the pre-ICUtreatment interval with respect to time and
therapyand the changes in physiologic and metabolic param-eters
over time, while remaining unaffected by varia-tions in practice
styles. Ideally, the variables involvedwould be independent, would
distinguish control ordisease-initiating effects from response
effects, andwould be path-independent for an individual
patient.
Conclusion
While examples of such model systems exist, anumber of
significant problems remain. We are cur-rently unable to determine
which physiologic, clinical,or metabolic variables cause, and which
are causedby, the inflammatory response. We have difficultylabeling
and then reliably identifying diseases. We alsoknow that the
measurement of a number of variablescurrently in use, such as the
level ofoxygen consump-tion or the cardiac output, may vary with
practicestyle. Very large data bases are also necessary toestablish
whether any reduced or streamlined data setretains its validity.
Despite these and other substantialobstacles, it is becoming
apparent that a small numberof variables can accurately capture
most of the mean-
RISK DISTRIBUTIONS OF503 ICU Admissions With SIRS.
80
70
60
50
40
30
20
10
0
No. of Patients Actual Hospital Mortality (%)100
908070
60H 50
4030
20
10
- Hospital Mortality No. of Patients
FIGURE 5. Risk distribution of 503 septic patients who met the
criteria for SIRS. This demonstrates therelationship between risk
of hospital mortality, calculated on the first day of the ICU stay,
and actualhospital mortality rates.
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
1652 Definitions for Sepsis and Organ Faikire (Bone et a!)
ingful data defining physiologic response currentlypresent in
large data sets. This provides encourage-ment that, while the ideal
model for describing theresponse of patients with complex
illnesses, such asSIRS, does not currently exist, it is a goal
worthattempting to achieve.
GUIDELINES FOR THE USE OF INNOVATIVE
THERAPIES IN SEVERE SEPSIS
Innovative therapy in severe sepsis usually involvesan attempt
to alter the systemic inflammatory responseof a patient. Such forms
of therapy are quite differentfrom supportive therapy or therapies
that are directedat the causative organism (eg, antibiotics or
surgicalprocedures). Despite the use of these current thera-pies,
the morbidity and mortality rates in severe sepsisremain high. Over
the last 10 to 15 years, newantibiotics and increasingly
sophisticated critical carehave had little impact on the mortality
rate of thisdisease.
A variety of innovative therapies aimed at themediators of
inflammation have recently undergoneclinical trials, and this line
of investigation is likely tocontinue. To recruit appropriate
numbers of patientswho meet the entry criteria for the various
studiesand to gain significant statistical power, multicentertrials
are usually necessary. The results of these trialsshould be used as
a guide to the rational use of newtherapies that are developed.
Multicenter clinical trials and product developmentare very
expensive. This will impose significantlyincreased costs on the
product being investigated if itbecomes clinically available. The
majority of theseinnovative therapies will, therefore, entail
significantexpenses for the health care consumer, although
theirimpact on the total cost of health care remains to
bedetermined.
There are well-established guidelines for conduct-ing clinical
tests, including adherence to goodclinical practice and the
protection ofhuman subjects.These guidelines are particularly
important in clinicaltrials that investigate the causes of sepsis.
The issue oflanguage was a central focus of this conference, andwe
recommend the use of the terminology discussedearlier in this
report in conjunction with that used inpublished peer-reviewed
clinical trials..#{176}The use ofestablished terminology may make
it possible to morecompetently compare the results of efficacy
trials forinnovative agents in the treatment of sepsis.
Thecomparison of trials would also be facilitated bystandardized
trial design, data collection, and report-ing of data. Further, it
may be anticipated that prob-lems with terminology will only get
worse as additionalagents are tested, either alone or in
combination.
The choice of patients for entry into trials shouldbe as
selective as possible, particularly when the trial
targets a subgroup ofthe septic population (eg, patientswith
Gram-negative sepsis). Patients whose underlyingdisease is not
likely to permit them to survive thestudy should be excluded, as
should those who arenot candidates for aggressive medical therapy.
As ourability to define specific subgroups improves, the
entrycriteria to future trials should reflect these changes.
The design of trials in sepsis research should
entailwell-defined end points, including the reporting ofdeaths
from any cause, through a minimum of28 daysafter study enrollment.
Overall hospital mortality, aswell as the resolution oforgan
dysfunction, should alsobe reported. We encourage researchers to
report datarelevant to the cost oftherapy and quality oflife.
Theanalysis of adverse outcomes in the overall group, aswell as in
the treatment group or any other group ofinterest, should be
presented. The reporting of resultsshould include a detailed
analysis that demonstratesthe comparability ofnoninvestigational
treatments andpatient characteristics among groups. It is
importantto address potential predictors of clinical outcome,such
as underlying disease and the referral source ofthe patients, and
to ensure that they are treatedadequately in the statistical
analysis of the data.Severity-of-illness scoring systems should be
used inthe stratification of patient risk to the extent that
theindividual scoring system has been independentlydemonstrated to
predict outcome in septic patients.The interval between fulfillment
of entry criteria andadministration of experimental intervention,
as wellas other indicators of possible lead time bias, shouldbe
noted and analyzed.
In the approved use of new therapies for treatingsepsis, the
selection of suitable patients to receivethese innovative
treatments is an important consider-ation for the clinician. The
expected impact of thetreatment on the disease process of the
patient shouldbe considered; the prospective identification of
pa-tients for whom the treatment would be most effica-cious is
important. The morbidity and mortality ratesthat would have
occurred in the absence of theinnovative therapy and the safety
profile ofthe productshould also be appraised. A patient who is to
be treatedwith an innovative therapy should have a
clinicalpresentation that matches the entrance criteria usedin the
clinical trial for that therapy. However, for someentrance criteria
(is, temperature, heart rate, andrespiratory rate), rigid limits
were set for the purposeofconducting a clinical trial. When
objective data haveallowed a definitive diagnosis of the target
populationfor which the innovative therapy is intended, excep-tions
to the listed entrance criteria can be made . Forexample, in
considering a therapy that utilizes anti-bodies raised against
endotoxin for a patient who hasa positive blood culture for
Gram-negative bacteria,tachypnea, tachycardia, and hypotension, the
failure
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
CHEST I 101 I 6 I JUNE, 1992 1653
to meet temperature criteria from a previous clinicaltrial
should not be used as a reason to withholdtreatment. Also, since
the Food and Drug Administra-lion has access to a larger data base
on any givenagent, FDA labeling indicators may be different
fromclinical trial entrance criteria, and may thus preemptthem.
Patients for whom exclusion criteria might be over-looked
include those who are younger than 18 yearsold, those who are
pregnant, and those with uncon-trolled hemorrhage. Such patients
are frequentlyexcluded from clinical trials as a matter of
protocol,not because of any anticipated risk to efficacy. Forthese
excluded populations, treatment may offer asignificant benefit. The
individual physician mustassess the risks or benefits of therapy
for each patient.Patients with burns, neutropenia, and
transplantedorgans may also be excluded from clinical trials dueto
the potentially disparate effects of therapy. Ifinnovative therapy
is found to be beneficial in nonex-cluded groups, then additional
clinical trials may beappropriate for certain excluded groups, such
as thethree groups mentioned above. In the interim, poten-tial
risks and unproven benefits should be consideredprior to any
decision to institute therapy. Finally,patients who are receiving
less than full support arefrequently not included in clinical
trials. A decision totreat such patients must be made with ethical
andcost-containment considerations in mind; it must beremembered
that the efficacy of treatment in thisgroup is unknown.
For those therapies that are directed at bacterialinfections or
the products of bacteria, such as endo-toxin, it is important to
obtain information about thespecific etiologic agent so that the
appropriate treat-ment method can be used. The utilization of
previousculture results and current Gram stains of specimensfrom
suspected sites of infection are imperative forgood decision
making. For example, a patient withGram stain evidence
ofStaphylococcus as the probableinfecting agent is not likely to
benefit from theutilization of antiendotoxin antibodies.
These innovative therapies are typically character-ized as
having a potentially important influence onpatient outcome, a
substantial impact on health carecosts, and a restrictive set
ofindications for use. Thesecharacteristics necessitate an
increased responsibilityon the part of the corporations developing
and mar-keting the therapy to provide scientifically
appropriateassistance and education to the clinicians or
institu-tions who must determine how and when to use it.This
information should be helpful in the selection ofindividuals who
could potentially benefit from thetherapy. This responsibility may
also necessitate ad-ditional studies or documentation in the future
tomeet changing requirements for the formal approval
of the therapy.In the absence of published data supporting
altera-
tions in the frequency or amount of an agent thatshould
beadministered in innovative therapy, physi-cians should dose
precisely as was done in the meth-odology ofthe clinical trial that
showed efficacy. Unlesssupported by published literature, the
effect of inno-vative therapy in the treatment of recurrent
sepsiscannot be predicted. However, in those patientssuspected of
having accelerated drug clearance, as incases of plasmapheresis and
massive bleeding, redos-ing may be considered, although there are
no availabledata on what effect, if any, these conditions have
onthe bioavailability of the agent or on its therapeuticeffect.
Most innovative therapies will be expensive andintended only for
specific populations of patients.These populations will be
identified through theresults ofclinical trials. Overutilization is
an importantissue, particularly if the identification of the
targetpopulation is difficult and there is no anticipatedtoxicity.
Equally important is the assurance that therewill not be
underutilization of the innovative therapyin patient groups that
would be likely to benefit.Methods of ensuring proper patient
selection willvary, based on the character of the particular
medicalinstitution. Each hospital must consider its own situ-ation
and devise appropriate methods to ensure thatthe proper innovative
therapy for sepsis is employed.
Potential mechanisms for accomplishing this goalinclude the
placement of physicians with expertise inthe diagnoses of the
particular disorders to be treatedin a position to guide the
utilization of innovativetherapies. An approval process involving
these physi-cians may be warranted. These physicians should
bereadily available, so that treatment with innovativetherapy is
not delayed; on-site expertise is preferableto off-site expertise.
When a specially trained physi-cian is not available, the use of
patient selectioncriteria checklists to assist the prescribing
physicianmay be helpful. The checklists may also be useful forother
situations in which the initial contact physicianis not an expert
in sepsis and innovative therapy.
With the input of physicians and pharmacists, sys-tems should be
designed and implemented prospec-tively as a quality assurance
mechanism to evaluatethe utilization of innovative therapy. Patient
selectionchecklists may help prevent overutilization; monitor-ing
for underutilization is more difficult. It is especiallyimportant
for individual physicians to exercise cautionin the use of
potentially deleterious therapies.THE ACCP/SCCM CONSENSUS
CONFERENCE COMMIT-TEE:
Conference Chairman:Roger C. Bone M.D., F.C.C.P, Dean, Rush
Medical College; Vice-President for Medical Affairs,
Rush-Presbyterian-St. Lukes MedicalCenter, Chicago
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
1654 Definitions for Sepsis and Organ Faikan (Bone eta!)
Session Chairmen:Robert A. Balk, M.D. , F.C.C.P, Associate
Professor of Medicine,Rush-Presbyterian-St. Lukes Medical Center,
ChicagoFrank B. Cerra, M.D., Professor ofSurgery and Director of
CriticalCare, Department of Surgery, University of Minnesota
Hospital,MinneapolisR. Philip Dellinger, M.D., F.C.C.P, Associate
Professor and Direc-tor of Critical Care, Department of Medicine,
Baylor College ofMedicine, DallasAlan M. Fein, M.D., FC.C.P,
Director, Pulmonary and CriticalCare Medicine Division, and
Associate Professor of Medicine,SUNY Health Science Center,
Winthrop-University Hospital, Mm-cola, NYWilliam A Knaus, M.D.,
Director, PCU Research Unit, GeorgeWashington University,
Washington, DCRoland M. H. Schein, M.D., Director of Critical Care
Medicineand Associate Professor ofMedicine, Department ofVeterans
AffairsMedical Center, University of Miami School of Medicine,
MiamiWdhiamj Sibbald, M.D., FC.C.P, Chief, Program in Critical
Care,and Professor of Medicine, University ofWestern Ontario,
London,Ontario, Canada
Faculty:
Jerome H. Abram.s, M.D. ,Assistant Professor ofSurgery,
Universityof Minnesota, MinneapolisGordon R. Bernard, M.D.,
F.C.C.P, Associate Professor of Mcdi-cine, Vanderbilt University
Medical School, NashvilleJames W Biondi, M.D., Pulmonary and
Critical Care Section, YaleUniversity School of Medicine, New
HavenJames E. Calvin, Jr., M.D., Associate Professor ofMedicmne,
Rush-Presbyterian-St. Lukes Medical Center, ChicagoRobert Demling,
M.D., Professor of Surgery, Harvard MedicalSchool, BostonPatrick j
Fahey, M.D., FC.C.P, Chief, Pulmonary and CriticalCare, Loyola
University Medical Center, Maywood, IllCharles I. Firher Jr., M.D.,
Associate Professor and Director,Center for Critical Care Research,
Case Western Reserve Univer-sity, ClevelandCory Franklin, M.D.,
Associate Professor of Medicine, Universityof Health
Sciences/Chicago Medical School, North Chicago, IllKennethj
Gorelick, M.D., Vice President, Medical and RegulatoryAffairs,
Genelabs Technologies, mc, Redwood City, CalifMark A. Kelley, M.D.,
FC.C.P, Vice Dean, Clinical Affairs,University of Pennsylvania
Medical Center, PhiladelphiaDennts C. Maki, M.D., F.C.C.P,
Professor of Medicine and Head,Section of Infectious Diseases;
Attending Physician, Center forTrauma and Life Support, University
ofWisconsin Medical School,Madison
John C. Marshall, M.D. ,Assistant Professor of Surgery,
UniversityofToronto, and Toronto Hospital, TorontoWilliam W
Merrill, M.D., Chief, Pulmonary Section, West HavenVA Medical
Center, West Haven, ConnJohn P Pribble, Pharin.D., Assistant
Director ofClinical Research,Synergen, mc, Boulder, CobEric C.
Rackou, M.D., FC.C.P, Professor and Chairman, Depart-ment of
Medicine, St. Vincents Hospital and Medical Center ofNew York, New
York Medical College, New YorkTimothy C. Rodell, M.D. ,Vice
President, Operations and ProductDevelopment, Cortech, Inc, Denver;
Clinical Assistant Professorof Medicine, University of Colorado
Health Sciences Center,DenverJohn N. Sheagren, M.D., Chairman,
Department of InternalMedicine, Illinois Masonic Medical Center;
Professor of Medicine,University of Illinois College of Medicine,
ChicagoMichael Slices; M.D., F.C.C.P, Assistant Professor and
Director,Respiratory Care Center, Rush-Presbyterian-St. Lukes
MedicalCenter, ChicagoCharles L. Sprung, M.D. ,F.C.C.P, Director,
Intensive Care Unit,Hadassah Hebrew University Medical Center,
Jerusalem, IsraelRichard C. Straube, M.D., Director, Infectious
Diseases, Centocor,Inc, Malvern, PaMartin I. Tobin, M.D., F.C.C.P,
Professor of Medicine, LoyolaUniversity ofChicago Stntch School of
Medicine, Maywood, IllGordon M. Trenholme, M.D. ,Professor
ofMedicine, Rush MedicalCollege, ChicagoDouglas Wagner, M.D.,
Senior Staff Scientist, ICU Research,George Washington University
Medical Center, Washington, DCC. Douglas Webb, Ph.D., Director,
Anti-Infectives, RoerigfPfizerPharmaceuticals, New YorkJanice C.
Wherry, M.D., Ph.D., Associate Director, Clinical Re-search, Cutter
Biological, Miles, Inc, Berkeley, CalifHerbert? Wzedemann, M.D.,
Chairman, Department of Pulmonaryand Critical Care Medicine,
Cleveland Clinic Foundation, Cleveland
Cornelius H. Wortel, M.D., Adjunct Clinical Assistant Professor
ofMedicine, Critical Care/Emergency Medicine Section, Depart-ment
of Medicine, Ben Taub General Hospital, Houston
EDITORS Nom: The editorial on page 1481 and this report
appearsimultaneously in the June, 1992 issue of Critical Care
Medicine(20: 724-26; 20: 864-74).
REFERENCES
1 Balk BA, Bone RC. The septic syndrome: definition and
clinicalimplications. Crit Care Clin 1989; 5:1-8
2 Ayres SM. SCCMS New Horizons Conference on Sepsis andSeptic
Shock. Crit Care Med 1985; 13:864-66
3 Increase in national hospital discharge survey rates for
septice-mia: United States, 1979-1987. MMWR 1990; 39:31-4
4 Parrillo JE, Parker MM, Natanson C, et al. Septic
shock:advances in the understanding of pathogenesis,
cardiovasculardysfunction, and therapy. Ann Intern Med 1990;
113:227-42
5 Manship L, McMillin RD. Brown JJ. The influence of sepsisand
multisystem organ failure on mortality in the surgicalintensive
care unit. Am Surg 1984; 50:94-101
6 Niederman MS. Fein AM. Sepsis syndrome, the adult respira-tory
distress syndrome, and nosocomial pneumonia: a commonclinical
sequence. Cliii Chest Med 1990; 11:663-65
7 Bone RC, Fisher CJ Jr. Clemmer iT, et al. A controlled
clinicaltrial ofhigh-dose methylprednisobone in the treatment of
severesepsis and septic shock. N EngI J Med 1987; 317:653-58
8 Veterans Administration Systemic Sepsis Cooperative
StudyGroup. Effect of high-dose glucocorticoid therapy on
mortalityin patients with clinical signs of systemic sepsis. N Engl
J Med1987; 317:659-65
9 Ziegler EJ, Fisher CJ Jr. Sprung CL, et al. Treatment of
gram-negative bacteremia and septic shock with HA-lA
humanmonocbonal antibody against endotoxin. N Engl J Med
1991;324:429-36
10 Greenman RI, Schein RMH, Martin MA, et al A
controlledclinical trial ofE5 murine monocbonal 1gM antibody to
endotoxinin the treatment ofgram-negative sepsis. JAMA 1991;
266:1097-1102
11 Bone RC, Fisher CJ Jr, Clemmer iT, et al. Sepsis syndrome:
avalid clinical entity. Crit Care Med 1989; 17:389-93
12 Bihari DJ. Sepflcemia the clinical diagnosis. J
AntimicrobChemother 1990; 25:C1-7
13 Sibbald WJ, Marshall J, Christou N, et al. Sepsis: clarity
ofexisting terminology . . . or more confusion? Crit Care Med1991;
19:996-98
14 Marshall J, Sweeney D. Microbial infection and the
septicresponse in critical surgical illness. Arch Surg 1990;
126:17-23
15 Bone RC. Sepsis, the sepsis syndrome, multi-organ failure:
aplea for comparable definitions. Ann Intern Med 1991;
114:332-33
16 Bone RC. Lets agree on terminology: definitions ofsepsis.
CritCare Med 1991; 19:973-76
17 Sprung CL. Definitions ofsepsis: have we reached a
consensus?Crit Care Med 1991; 19:849-51
18 Sprung CL, Peduzzi PN, Shatney CH, et al. Impact of
enceph-alopathy on mortality in the sepsis syndrome. Cnt Care
Med1990; 18:801-06
19 Kreger BE, Craven DE, McCabe WR. Gram-negative bacte-remia:
IV Re-evaluation ofclinical features and treatment in 612patients.
Am J Med 1980; 68:344-50
20 Pepe PE, Potkin lIT, Reus DH, Hudson LD, Carrico CJ.
Clinicalpredictors ofthe adult respiratory distress syndrome. Am J
Surg1982; 144:124-30
21 Pepe FE, Hudson LD, Cameo CJ. Early application of
positiveend-expiratory pressure in patients at risk for the adult
respira-tory distress syndrome. N Engl J Med 1984; 311;281-86
22 Weil MH, Shubin H, Biddle M. Shock caused by
gram-negativemicroorganisms. Ann Intern Med 1964; 60:384-400
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014
-
CHEST I 101 I 6 I JUNE, 1992 1655
23 Skillman JJ, Bushnell LS, Goldman H, Silen W.
Respiratoryfailure, hypotension, sepsis, and jaundice: a clinical
syndromeassociated with lethal hemorrhage from acute stress
ulcerationofthe stomach. Am J Surg 1969; 117:523-30
24 fllney NL, Batley GL, Morgan A? Sequential system
failureafter rupture of abdominal aortic aneurysms: an
unsolvedproblem in postoperative care. Ann Surg 1973;
178:117-22
25 Eiseman B, Beart R, Norton L. Multiple organ failure.
SurgGynecol Obstet 1977; 144:323-26
26 Fry DE, Pearlstein L, Fulton RL, Polk HC. Multiple
systemorgan failure: the role ofuncontrolled infection. Arch Surg
1980;115:136-40
27 Bell RC, Coalson JJ, Smith JD, Johanson WG. Multiple
organsystem failure and infection in adult respiratory distress
syn-drome. Ann Intern Med 1983; 99:293-98
28 Baue AE. Multiple, progressive, or sequential systems
failure:a syndrome ofthe 1970s. Arch Surg 1975; 110:779-81
29 Polk HC, Shields CL. Remote organ failure: a valid sign
ofoccult intra-abdominal infection. Surgery 1977; 81:310-13
30 Goris RJA, Boekhorst TAP, Nuytinck JKS, Gimbrere JSF.Multiple
organ failure: generalized autodestructive inflamma-tion? Arch Surg
1985; 120:1109-15
31 Norton LW. Does drainage of intraabdominal pus
reversemultiple organ failure? Am J Surg 1985; 149:347-50
32 Marshall JC, Christou NV, Horn H, Meakins JL. The
microbi-obogy of multiple organ failure: the proximal GI tract as
anoccult reservoir ofpathogens. Arch Surg 1988; 123:309-15
33 Tracey KJ, Fong Y, Hesse DG, Manogue KR, Lee At, et al.
Anti-cachectinfFNF monoclonal antibodies prevent septic shockduring
lethal bacteremia. Nature 1987; 330:662-64
34 Okusawa 5, Gelfand JA, Ikejima T, Connolly RJ, Dianarelbo
CA.Interleukin 1 induces a shock-like state in rabbits:
synergismwith tumor necrosis factor and the effect of
cyclooxygenaseinhibition. J Clin Invest 1988; 81:1162-72
35 Wallace JL, Steel G, Whittle BJR, Lagente V, Vargaftig
B.Evidence for platelet-activating factor as a mediator of
endo-toxin-induced gastrointestinal damage in the rat: effects of
threeplatelet-activating factor agonists. Gastroenterology 1987;
93:765-73
36 Sculier JP, Bron D, Verboven N, Klastersky J. Multiple
organfailure during interleukin-2 and LAK cell infusion.
IntensiveCare Med 1988; 14:666-67
37 Knaus WA, Wagner DP, LynnJ. Short-term mortality
predictionsfor critically ill hospitalized adults: science and
ethics. Science1991; 254:389
38 Knaus WA, Wagner DP, Draper EA. The value of
measuringseverity of disease in clinical research on acutely ill
patients.J Chronic Dis 1984; 37:455-63
39 Knaus WA, Wagner DP, Draper EA, Zimmerman JE, et al.
TheAPACHE III Prognostic System: risk prediction of
hospitalmortality for critically ill hospitalized adults. Chest
1991;100:1619-38
40 Knaus WA, Sun X, Nystrom P0, Wagner DR Evaluation
ofdefinition ofsepsis. Chest 1992; 101:1656-62
41 Cerra FB, Hypermetabolism, organ failure, and
metabolicsupport. Surgery 1987; 191:1-27
42 Coleman W1, Siegel JH, Giovanni I, DeGaetano A, et
al.Probability and patient state space. Int J Chin Monit
Comput1990; 7:201-15
43 Abrams JH, Barke BA, Cerra FB. Quantitative evaluation
ofclinical course in surgical ICU patients: the data conform
tocatastrophe theory. J Trauma 1984; 24:1028-37
44 Chalmers TC, Smith H, Blackburn B, Silverman B, SchroederB,
Reitman D, et al. A method for assessing the quality of arandomized
control trial. Controlled Chin Trials 1981; 2:31-49
45 Meinert CL. Clinical trials: design, conduct and analysis.
NewYork: Oxford University Press, 1986
Downloaded From: http://journal.publications.chestnet.org/ on
05/29/2014