IMPORTANCEDefinitionsofsepsisandseptic

shockwerelastrevisedin2001.Considerable
advanceshavesincebeenmadeintothe
pathobiology(changesinorganfunction,
morphology,cellbiology,biochemistry,
immunology,andcirculation),management,
andepidemiologyofsepsis,suggestingthe
needforreexamination.
KEYFINDINGSFROMEVIDENCESYNTHESIS

Limitationsofpreviousdefinitionsincluded
anexcessivefocusoninflammation,the
misleadingmodelthatsepsisfollowsa
continuumthroughseveresepsistoshock,
andinadequatespecificityandsensitivityof
thesystemicinflammatoryresponse
syndrome(SIRS)criteria.Multiple
definitionsandterminologiesarecurrentlyin
useforsepsis,septicshock,andorgan
dysfunction,leadingtodiscrepanciesin
reportedincidenceandobservedmortality.
Thetaskforceconcludedthetermsevere
sepsiswasredundant.
RECOMMENDATIONSSepsisshouldbe

definedaslifethreateningorgandysfunction
causedbyadysregulatedhostresponseto
infection.Forclinicaloperationalization,
organdysfunctioncanberepresentedbyan
increaseintheSequential[Sepsisrelated]
OrganFailureAssessment(SOFA)scoreof2
pointsormore,whichisassociatedwithan
inhospitalmortalitygreaterthan10%.Septic
shockshouldbedefinedasasubsetofsepsis
inwhichparticularlyprofoundcirculatory,
cellular,andmetabolicabnormalitiesare
associatedwithagreaterriskofmortality
thanwithsepsisalone.Patientswithseptic
shockcanbeclinicallyidentifiedbya
vasopressorrequirementtomaintainamean
arterialpressureof65mmHgorgreaterand
serumlactatelevelgreaterthan2mmol/L
(>18mg/dL)intheabsenceofhypovolemia.
Thiscombinationisassociatedwithhospital
mortalityratesgreaterthan40%.Inoutof
hospital,emergencydepartment,orgeneral
hospitalwardsettings,adultpatientswith
suspectedinfectioncanberapidlyidentified
asbeingmorelikelytohavepooroutcomes

typicalofsepsisiftheyhaveatleast2ofthe
followingclinicalcriteriathattogether
constituteanewbedsideclinicalscoretermed
quickSOFA(qSOFA):respiratoryrateof
22/minorgreater,alteredmentation,or
systolicbloodpressureof100mmHgorless.
CONCLUSIONSANDRELEVANCEThese

updateddefinitionsandclinicalcriteriahould
replacepreviousdefinitions,offergreater
consistencyforepidemiologicstudiesand
clinicaltrials,andfacilitateearlier
recognitionandmoretimelymanagementof
patientswithsepsisoratriskofdeveloping
sepsis.
Sepsis,asyndromeofphysiologic,
pathologic,andbiochemicalabnormalities
inducedbyinfection,isamajorpublichealth
concern,accountingformorethan$20billion
(5.2%)oftotalUShospitalcostsin2011.1
Thereportedincidenceofsepsisis
increasing,2,3likelyreflectingaging
populationswithmorecomorbidities,greater
recognition,4and,insomecountries,
reimbursementfavorablecoding.5Although
thetrueincidenceisunknown,conservative
estimatesindicatethatsepsisisaleading
causeofmortalityandcriticalillness
worldwide.6,7Furthermore,thereisincreasing
awarenessthatpatientswhosurvivesepsis
oftenhavelongtermphysical,psychological,
andcognitivedisabilitieswithsignificant
healthcareandsocialimplications.8
A1991consensusconference9developed
initialdefinitionsthatfocusedonthethen
prevailingviewthatsepsisresultedfroma
hostssystemicinflammatoryresponse
syndrome(SIRS)toinfection(Box1).Sepsis
complicatedbyorgandysfunctionwas
termedseveresepsis,whichcouldprogressto
septicshock,definedassepsisinduced
hypotensionpersistingdespiteadequatefluid
resuscitation.A2001taskforce,recognizing
limitationswiththesedefinitions,expanded
thelistofdiagnosticcriteriabutdidnotoffer
alternativesbecauseofthelackofsupporting

evidence.10Ineffect,thedefinitionsofsepsis,
septicshock,andorgandysfunctionhave
remainedlargelyunchangedformorethan2
decades.

TheProcessofDevelopingNew
Definitions
Recognizingtheneedtoreexaminethe
currentdefinitions,11theEuropeanSocietyof
IntensiveCareMedicineandtheSocietyof
CriticalCareMedicineconvenedataskforce
of19criticalcare,infectiousdisease,
surgical,andpulmonaryspecialistsinJanuary
2014.Unrestrictedfundingsupportwas
providedbythesocieties,andthetaskforce
retainedcompleteautonomy.Thesocieties
eachnominatedcochairs(DrsDeutschman
andSinger),whoselectedmembersaccording
totheirscientificexpertiseinsepsis
epidemiology,clinicaltrials,andbasicor
translationalresearch.
Thegroupengagediniterativediscussions
via4facetofacemeetingsbetweenJanuary
2014andJanuary2015,email
correspondence,andvoting.Existing
definitionswererevisitedinlightofan
enhancedappreciationofthepathobiology
andtheavailabilityoflargeelectronichealth
recorddatabasesandpatientcohorts.
Anexpertconsensusprocess,basedona
currentunderstandingofsepsisinduced
changesinorganfunction,morphology,cell
biology,biochemistry,immunology,and
circulation(collectivelyreferredtoas
pathobiology),forgedagreementonupdated
definition(s)andthecriteriatobetestedin
theclinicalarena(contentvalidity).The
distinctionbetweendefinitionsandclinical
criteriaisdiscussedbelow.Theagreement
betweenpotentialclinicalcriteria(construct
validity)andtheabilityofthecriteriato
predictoutcomestypicalofsepsis,suchas
needforintensivecareunit(ICU)admission
ordeath(predictivevalidity,aformof
criterionvalidity),werethentested.These

explorationswereperformedinmultiplelarge
electronichealthrecorddatabasesthatalso
addressedtheabsence(missingness)of
individualelementsofdifferentorgan
dysfunctionscoresandthequestionof
generalizability(ecologicvalidity).12A
systematicliterature
Box1.SIRS(SystemicInflammatoryResponse
Syndrome)
Twoormoreof:Temperature>38Cor<36C
Heartrate>90/minRespiratoryrate>20/minorPaCO2
<32mmHg(4.3kPa)

Whitebloodcellcount>12000/mm or<4000/mm or
>10%immaturebands
FromBoneetal.9

reviewandDelphiconsensusmethodswere
alsousedforthedefinitionandclinical
criteriadescribingsepticshock.13
Whencompiled,thetaskforce
recommendationswithsupportingevidence,
includingoriginalresearch,werecirculatedto
majorinternationalsocietiesandother
relevantbodiesforpeerreviewand
endorsement(31endorsingsocietiesare
listedattheendofthisarticle).

IssuesAddressedbytheTask
Force
Thetaskforcesoughttodifferentiatesepsis
fromuncomplicatedinfectionandtoupdate
definitionsofsepsisandsepticshocktobe
consistentwithimprovedunderstandingof
thepathobiology.Adefinitionisthe
descriptionofanillnessconcept;thus,a
definitionofsepsisshoulddescribewhat
sepsisis.Thischosenapproachallowed
discussionofbiologicalconceptsthatare
currentlyincompletelyunderstood,suchas
geneticinfluencesandcellularabnormalities.
Thesepsisillnessconceptispredicatedon
infectionasitstrigger,acknowledgingthe

currentchallengesinthemicrobiological
identificationofinfection.Itwasnot,
however,withinthetaskforcebriefto
examinedefinitionsofinfection.
Thetaskforcerecognizedthatsepsisisa
syndromewithout,atpresent,avalidated
criterionstandarddiagnostictest.Thereis
currentlynoprocesstooperationalizethe
definitionsofsepsisandsepticshock,akey
deficitthathasledtomajorvariationsin
reportedincidenceandmortalityrates(see
laterdiscussion).Thetaskforcedetermined
thattherewasanimportantneedforfeatures
thatcanbeidentifiedandmeasuredin
individualpatientsandsoughttoprovidesuch
criteriatoofferuniformity.Ideally,these
clinicalcriteriashouldidentifyallthe
elementsofsepsis(infection,hostresponse,
andorgandysfunction),besimpletoobtain,
andbeavailablepromptlyandatareasonable
costorburden.Furthermore,itshouldbe
possibletotestthevalidityofthesecriteria
withavailablelargeclinicaldatasetsand,
ultimately,prospectively.Inaddition,clinical
criteriashouldbeavailabletoprovide
practitionersinoutofhospital,emergency
department,andhospitalwardsettingswith
thecapacitytobetteridentifypatientswith
suspectedinfectionlikelytoprogresstoa
lifethreateningstate.Suchearlyrecognition
isparticularlyimportantbecauseprompt
managementofsepticpatientsmayimprove
outcomes.4
Inaddition,toprovideamoreconsistentand
reproduciblepictureofsepsisincidenceand
outcomes,thetaskforcesoughttointegrate
thebiologyandclinicalidentificationof
sepsiswithitsepidemiologyandcoding.

IdentifiedChallengesand
Opportunities
AssessingtheValidityofDefinitions
WhenThereIsNoGoldStandard
Sepsisisnotaspecificillnessbutrathera

syndromeencompassingastilluncertain
pathobiology.Atpresent,itcanbeidentified
byaconstellationofclinicalsignsand
symptomsinapatientwithsuspected
infection.Becausenogoldstandard
diagnostictestexists,thetaskforcesought
definitionsandsupportingclinicalcriteria
thatwereclearandfulfilledmultipledomains
ofusefulnessandvalidity.

ImprovedUnderstandingofSepsis
Pathobiology
Sepsisisamultifacetedhostresponsetoan
infectingpathogenthatmaybesignificantly
amplifiedbyendogenousfactors.14,15The
originalconceptualizationofsepsisas
infectionwithatleast2ofthe4SIRScriteria
focusedsolelyoninflammatoryexcess.
However,thevalidityofSIRSasadescriptor
ofsepsispathobiologyhasbeenchallenged.
Sepsisisnowrecognizedtoinvolveearly
activationofbothproandantiinflammatory
responses,16alongwithmajormodifications
innonimmunologicpathwayssuchas
cardiovascular,neuronal,autonomic,
hormonal,bioenergetic,metabolic,and
coagulation,14,17,18allofwhichhave
prognosticsignificance.Organdysfunction,
evenwhensevere,isnotassociatedwith
substantialcelldeath.19
Thebroaderperspectivealsoemphasizesthe
significantbiologicalandclinical
heterogeneityinaffectedindividuals,20with
age,underlyingcomorbidities,concurrent
injuries(includingsurgery)andmedications,
andsourceofinfectionaddingfurther
complexity.21Thisdiversitycannotbe
appropriatelyrecapitulatedineitheranimal
modelsorcomputersimulations.14With
furthervalidation,multichannelmolecular
signatures(eg,transcriptomic,metabolomic,
proteomic)willlikelyleadtobetter
characterizationofspecificpopulation

subsets.22,23Suchsignaturesmayalsohelpto
differentiatesepsisfromnoninfectiousinsults
suchastraumaorpancreatitis,inwhicha
similarbiologicalandclinicalhostresponse
maybetriggeredbyendogenousfactors.24
Keyconceptsofsepsisdescribingitsprotean
naturearehighlightedinBox2.

VariableDefinitions
Abetterunderstandingoftheunderlying
pathobiologyhasbeenaccompaniedbythe
recognitionthatmanyexistingterms(eg,
sepsis,severesepsis)areused
interchangeably,whereasothersare
redundant(eg,sepsissyndrome)oroverly
narrow(eg,septicemia).Inconsistent
strategiesinselectingInternational
ClassificationofDiseases,NinthRevision
(ICD9),andICD10codeshavecom
poundedtheproblem.
Sepsis
Thecurrentuseof2ormoreSIRScriteria
(Box1)toidentifysepsiswasunanimously
consideredbythetaskforcetobeunhelpful.
Changesinwhitebloodcellcount,
temperature,andheartratereflect
inflammation,thehostresponsetodanger
intheformofinfectionorotherinsults.The
SIRScriteriadonotnecessarilyindicatea
dysregulated,lifethreateningresponse.SIRS
criteriaarepresentinmanyhospitalized
patients,includingthosewhoneverdevelop
infectionandneverincuradverseoutcomes
(poordiscriminantvalidity).25Inaddition,1
in8patientsadmittedtocriticalcareunitsin
AustraliaandNewZealandwithinfection
andneworganfailuredidnothavethe
requisiteminimumof2SIRScriteriatofulfill
thedefinitionofsepsis(poorconcurrent
validity)yethadprotractedcourseswith
significantmorbidityandmortality.26
Discriminantvalidityandconvergentvalidity
constitutethe2domainsofconstructvalidity;
theSIRScriteriathusperformpoorlyonboth
counts.

Box2.KeyConceptsofSepsis
Sepsisistheprimarycauseofdeathfrominfection,
especiallyifnotrecognizedandtreatedpromptly.Its
recognitionmandatesurgentattention.
Sepsisisasyndromeshapedbypathogenfactorsand
hostfactors(eg,sex,raceandothergenetic
determinants,age,comorbidities,environment)with
characteristicsthatevolveovertime.What
differentiatessepsisfrominfectionisanaberrantor
dysregulatedhostresponseandthepresenceoforgan
dysfunction.
Sepsisinducedorgandysfunctionmaybeoccult;
therefore,itspresenceshouldbeconsideredinany
patientpresentingwithinfection.Conversely,
unrecognizedinfectionmaybethecauseofnewonset
organdysfunction.Anyunexplainedorgandysfunction
shouldthusraisethepossibilityofunderlyinginfection.
Theclinicalandbiologicalphenotypeofsepsiscanbe
modifiedbypreexistingacuteillness,longstanding
comorbidities,medication,andinterventions.
Specificinfectionsmayresultinlocalorgan
dysfunctionwithoutgeneratingadysregulatedsystemic
hostresponse.

OrganDysfunctionorFailure
Severityoforgandysfunctionhasbeen
assessedwithvariousscoringsystemsthat
quantifyabnormalitiesaccordingtoclinical
findings,laboratorydata,ortherapeutic
interventions.Differencesinthesescoring
systemshavealsoledtoinconsistencyin
reporting.Thepredominantscoreincurrent
useistheSequentialOrganFailure
Assessment(SOFA)(originallytheSepsis
relatedOrganFailureAssessment27)(Table
1).28AhigherSOFAscoreisassociatedwith
anincreasedprobabilityofmortality.28The
scoregradesabnormalitybyorgansystem
andaccountsforclinicalinterventions.
However,laboratoryvariables,namely,PaO2,
plateletcount,creatininelevel,andbilirubin
level,areneededforfullcomputation.
Furthermore,selectionofvariablesandcutoff
valuesweredevelopedbyconsensus,and
SOFAisnotwellknownoutsidethecritical
carecommunity.Otherorganfailurescoring
systemsexist,includingsystemsbuiltfrom

statisticalmodels,butnoneareincommonuse.

SepticShock
Multipledefinitionsforsepticshockare
currentlyinuse.Furtherdetailsareprovided
inanaccompanyingarticlebyShankarHari
etal.13Asystematicreviewofthe
operationalizationofcurrentdefinitions
highlightssignificantheterogeneityin
reportedmortality.Thisheterogeneity
resultedfromdifferencesintheclinical
variableschosen(varyingcutoffsforsystolic
ormeanbloodpressurediverselevelsof
hyperlactatemiavasopressoruse
concurrentneworgandysfunction
definedfluidresuscitationvolume/targets),
thedatasourceandcodingmethods,and
enrollmentdates.

ANeedforSepsisDefinitionsfor
thePublicandforHealthCare
Practitioners
Despiteitsworldwideimportance,6,7public
awarenessofsepsisispoor.29Furthermore,
thevariousmanifestationsofsepsismake
diagnosisdifficult,evenforexperienced
clinicians.Thus,thepublicneedsan
understandabledefinitionofsepsis,whereas
healthcarepractitionersrequireimproved
clinicalpromptsanddiagnosticapproachesto
facilitateearlieridentificationandanaccurate
quantificationoftheburdenofsepsis.

Results/Recommendations
DefinitionofSepsis
Sepsisisdefinedaslifethreateningorgan
dysfunctioncausedbyadysregulatedhost
responsetoinfection(Box3).Thisnew
definitionemphasizestheprimacyofthe
nonhomeostatichostresponsetoinfection,
thepotentiallethalitythatisconsiderablyin
excessofastraightforwardinfection,andthe
needforurgentrecognition.Asdescribed

later,evenamodestdegreeoforgan
dysfunctionwheninfectionisfirstsuspected
isassociatedwithaninhospitalmortalityin
excessof10%.Recognitionofthiscondition
thusmeritsapromptandappropriate
response.
NonspecificSIRScriteriasuchaspyrexiaor
neutrophiliawillcontinuetoaidinthe
generaldiagnosisofinfection.Thesefindings
complementfeaturesofspecificinfections
(eg,rash,lungconsolidation,dysuria,
peritonitis)thatfocusattentiontowardthe
likelyanatomicalsourceandinfecting
organism.However,SIRSmaysimplyreflect
anappropriatehostresponsethatisfrequently
adaptive.Sepsisinvolvesorgandysfunction,
indicatingapathobiologymorecomplexthan
infectionplusanaccompanyinginflammatory
responsealone.Thetaskforceemphasison
lifethreateningorgandysfuncionis
consistentwiththeviewthatcellulardefects
underliephysiologicandbiochemical
abnormalitieswithinspecificorgansystems.
Underthisterminology,severesepsis
becomessuperfluous.Sepsisshouldgenerally
warrantgreaterlevelsofmonitoringand
intervention,includingpossibleadmissionto
criticalcareorhighdependencyfacilities.

ClinicalCriteriatoIdentifyPatientsWith
Sepsis
Thetaskforcerecognizedthatnocurrent
clinicalmeasuresreflecttheconceptofa
dysregulatedhostresponse.However,as
notedbythe2001taskforce,manybedside
examinationfindingsandroutinelaboratory
testresultsareindicativeofinflammationor
organdysfunction.10Thetaskforcetherefore
evaluatedwhichclinicalcriteriabest
identifiedinfectedpatientsmostlikelyto
havesepsis.Thisobjectivewasachievedby
interrogatinglargedatasetsofhospitalized
patientswithpresumedinfection,assessing
agreementamongexistingscoresof
inflammation(SIRS)9ororgandysfunction
(eg,SOFA,27,28LogisticOrganDysfunction

System30)(constructvalidity),and
delineatingtheircorrelationwithsubsequent
outcomes(predictivevalidity).Inaddition,
multivariableregressionwasusedtoexplore
theperformanceof21bedsideandlaboratory
criteriaproposedbythe2001taskforce.10
Fulldetailsarefoundintheaccompanying
articlebySeymouretal.12Inbrief,electronic
healthrecorddataof1.3millionencountersat
12communityandacademichospitalswithin
theUniversityofPittsburghMedicalCenter
healthsysteminsouthwesternPennsylvania
werestudied.Therewere148907patients
withsuspectedinfection,identifiedasthose
whohadbodyfluidssampledforcultureand
receivedantibiotics.Twooutcomes
hospitalmortalityandmortality,ICUstayof
3daysorlonger,orbothwereusedto
assesspredictivevaliditybothoveralland
acrossdecilesofbaselineriskasdetermined
byage,sex,andcomorbidity.

Forinfectedpatientsbothinsideandoutside
oftheICU,predictivevaliditywas
determinedwith2metricsforeachcriterion:
theareaunderthereceiveroperating
characteristiccurve(AUROC)andthe
changeinoutcomescomparingpatientswith
ascoreofeither2pointsormoreorfewer
than2pointsinthedifferentscoring
systems9,27,30acrossdecilesofbaselinerisk.
Thesecriteriawerealsoanalyzedin4
externalUSandnonUSdatasetscontaining
datafrommorethan700000patients(cared
forinbothcommunityandtertiarycare
facilities)withbothcommunityandhospital
acquiredinfection.
InICUpatientswithsuspectedinfectionin
theUniversityofPittsburghMedicalCenter
dataset,discriminationforhospitalmortality
withSOFA(AUROC=0.74;95%CI,0.73
0.76)andtheLogisticOrganDysfunction
System(AUROC=0.75;95%CI,0.720.76)
wassuperiortothatwithSIRS(AUROC=

0.64;95%CI,0.620.66).Thepredictive
validityofachangeinSOFAscoreof2or
greaterwassimilar(AUROC=0.72;95%CI,
0.700.73).ForpatientsoutsidetheICUand
withsuspectedinfection,discriminationof
hospitalmortalitywithSOFA(AUROC=
0.79;95%CI,0.780.80)orchangeinSOFA
score(AUROC=0.79;95%CI,0.780.79)
wassimilartothatwithSIRS(AUROC=
0.76;95%CI,0.750.77).
BecauseSOFAisbetterknownandsimpler
thantheLogisticOrganDysfunctionSystem,
thetaskforcerecommendsusingachangein
baselineofthetotalSOFAscoreof2points
ormoretorepresentorgandysfunction(Box
3).ThebaselineSOFAscoreshouldbe
assumedtobezerounlessthepatientis
knowntohavepreexisting(acuteorchronic)
organdysfunctionbeforetheonsetof
infection.PatientswithaSOFAscoreof2or
morehadanoverallmortalityriskof
approximately10%inageneralhospital
populationwithpresumedinfection.12Thisis
greaterthantheoverallmortalityrateof8.1%
forSTsegmentelevationmyocardial
infarction,31aconditionwidelyheldtobelife
threateningbythecommunityandby
clinicians.Dependingonapatientsbaseline
levelofrisk,aSOFAscoreof2orgreater
identifieda2to25foldincreasedriskof
dyingcomparedwithpatientswithaSOFA
scorelessthan2.12
Asdiscussedlater,theSOFAscoreisnot
intendedtobeusedasatoolforpatient
managementbutasameanstoclinically
characterizeasepticpatient.Componentsof
SOFA(suchascreatinineorbilirubinlevel)
requirelaboratorytestingandthusmaynot
promptlycapturedysfunctioninindividual
organsystems.Otherelements,suchasthe
cardiovascularscore,canbeaffectedbyiat
rogenicinterventions.However,SOFAhas
widespreadfamiliaritywithinthecriticalcare
communityandawellvalidatedrelationship
tomortalityrisk.Itcanbescored
retrospectively,eithermanuallyorby

automatedsystems,fromclinicaland
laboratorymeasuresoftenperformed
routinelyaspartofacutepatient
management.Thetaskforcenotedthatthere
areanumberofnovelbiomarkersthatcan
identifyrenalandhepaticdysfunctionor
coagulopathyearlierthantheelementsused
inSOFA,buttheserequirebroadervalidation
beforetheycanbeincorporatedintothe
clinicalcriteriadescribingsepsis.Future
iterationsofthesepsisdefinitionsshould
includeanupdatedSOFAscorewithmore
optimalvariableselection,cutoffvalues,and
weighting,orasuperiorscoringsystem.

ScreeningforPatientsLikelytoHave
Sepsis
Aparsimoniousclinicalmodeldeveloped
withmultivariablelogisticregression
identifiedthatany2of3clinicalvariables
GlasgowComaScalescoreof13orless,
systolicbloodpressureof100mmHgorless,
andrespiratoryrate22/minorgreater
offeredpredictivevalidity(AUROC=0.81;
95%CI,0.800.82)similartothatofthefull
SOFAscoreoutsidetheICU.12Thismodel
wasrobusttomultiplesensitivityanalyses
includingamoresimpleassessmentofaltered
mentation(GlasgowComaScalescore<15)
andintheoutofhospital,emergency
department,andwardsettingswithinthe
externalUSandnonUSdatasets.
Forpatientswithsuspectedinfectionwithin
theICU,theSOFAscorehadpredictive
validity(AUROC=0.74;95%CI,0.730.76)
superiortothatofthismodel(AUROC=
0.66;95%CI,0.640.68),likelyreflectingthe
modifyingeffectsofinterventions(eg,
vasopressors,sedativeagents,mechanical
ventilation).Additionoflactatemeasurement
didnotmeaningfullyimprovepredictive
validitybutmayhelpidentifypatientsat
intermediaterisk.
Thisnewmeasure,termedqSOFA(forquick
SOFA)andincorporatingalteredmentation,
systolicbloodpressureof100mmHgorless,

andrespiratoryrateof22/minorgreater,
providessimplebedsidecriteriatoidentify
adultpatientswithsuspectedinfectionwho
arelikelytohavepooroutcomes(Box4).
Becausepredictivevaliditywasunchanged
(P=.55),thetaskforcechosetoemphasize
alteredmentationbecauseitrepresentsany
GlasgowComa
Scalescorelessthan15andwillreducethe
measurementburden.AlthoughqSOFAis
lessrobustthanaSOFAscoreof2orgreater
intheICU,itdoesnotrequirelaboratorytests
andcanbeassessedquicklyandrepeatedly.
ThetaskforcesuggeststhatqSOFAcriteria
beusedtopromptclinicianstofurther
investigatefororgandysfunction,toinitiate
orescalatetherapyasappropriate,andto
considerreferraltocriticalcareorincrease
thefrequencyofmonitoring,ifsuchactions
havenotalreadybeenundertaken.Thetask
forceconsideredthatpositiveqSOFAcriteria
shouldalsopromptconsiderationofpossible
infectioninpatientsnotpreviously
recognizedasinfected.

DefinitionofSepticShock
Septicshockisdefinedasasubsetofsepsis
inwhichunderlyingcirculatoryandcellular
metabolismabnormalitiesareprofound
enoughtosubstantiallyincreasemortality
(Box3).The2001taskforcedefinitions
describedsepticshockasastateofacute
circulatoryfailure.10Thetaskforcefavored
abroaderviewtodifferentiatesepticshock
fromcardiovasculardysfunctionaloneandto
recognizetheimportanceofcellular
abnormalities(Box3).Therewasunanimous
agreementthatsepticshockshouldreflecta
moresevereillnesswithamuchhigher
likelihoodofdeaththansepsisalone.

ClinicalCriteriatoIdentifySepticShock
Furtherdetailsareprovidedinthe
accompanyingarticlebyShankarHarietal. 13
First,asystematicreviewassessedhow
currentdefinitionswereoperationalized.This

informedaDelphiprocessconductedamong
thetaskforcememberstodeterminethe
updatedsepticshockdefinitionandclinical
criteria.Thisprocesswasiterativeand
informedbyinterrogationofdatabases,as
summarizedbelow.
TheDelphiprocessassessedagreementson
descriptionsoftermssuchashypotension,
needforvasopressortherapy,raised
lactate,andadequatefluidresuscitation
forinclusionwithinthenewclinicalcriteria.
Themajority(n=14/17;82.4%)oftaskforce
membersvotingonthisagreedthat
hypotensionshouldbedenotedasamean
arterialpressurelessthan65mmHg
accordingtothepragmaticdecisionthatthis
wasmostoftenrecordedindatasetsderived
frompatientswithsepsis.Systolicblood
pressurewasusedasaqSOFAcriterion
becauseitwasmostwidelyrecordedinthe
electronichealthrecorddatasets.
Amajority(11/17;64.7%)ofthetaskforce
agreed,whereas2(11.8%)disagreed,thatan
elevatedlactatelevelisreflectiveofcellular
dysfunctioninsepsis,albeitrecognizingthat
multiplefactors,suchasinsufficienttissue
oxygendelivery,impairedaerobic
respiration,acceleratedaerobicglycolysis,
andreducedhepaticclearance,also
contribute.32Hyperlactatemiais,however,a
reasonablemarkerofillnessseverity,with
higherlevelspredictiveofhighermortality. 33
Criteriaforadequatefluidresuscitationor
needforvasopressortherapycouldnotbe
explicitlyspecifiedbecausethesearehighly
userdependent,relyingonvariable
monitoringmodalitiesandhemodynamic
targetsfortreatment.34Otheraspectsof
management,suchassedationandvolume
statusassessment,arealsopotential
confoundersinthehypotensionvasopressor
relationship.
ByDelphiconsensusprocess,3variables
wereidentified(hypotension,elevatedlactate
level,andasustainedneedforvasopressor

therapy)totestincohortstudies,exploring
alternativecombinationsanddifferentlactate
thresholds.Thefirstdatabaseinterrogated
wastheSurvivingSepsisCampaigns
internationalmulticenterregistryof28150
infectedpatientswithatleast2SIRScriteria
andatleast1organdysfunctioncriterion.
Hypotensionwasdefinedasameanarterial
pressurelessthan65mmHg,theonly
availablecutoff.Atotalof18840patients
withvasopressortherapy,hypotension,or
hyperlactatemia(>2mmol/L[18mg/dL])
aftervolumeresuscitationwereidentified.
Patientswithfluidresistanthypotension
requiringvasopressorsandwith
hyperlactatemiawereusedasthereferent
groupforcomparingbetweengroup
differencesintheriskadjustedoddsratiofor
mortality.Riskadjustmentwasperformed
withageneralizedestimatingequation
populationaveragedlogisticregression
modelwithexchangeablecorrelation
structure.
Riskadjustedhospitalmortalitywas
significantlyhigher(P<.001comparedwith
thereferentgroup)inpatientswithfluid
resistanthypotensionrequiringvasopressors
andhyperlactatemia(42.3%and49.7%at
thresholdsforserumlactatelevelof>2
mmol/L[18mg/dL]or>4mmol/L[36
mg/dL],respectively)comparedwitheither
hyperlactatemiaalone(25.7%and29.9%
mortalityforthosewithserumlactatelevelof
>2mmol/L[18mg/dL]and>4mmol/L[36
mg/dL],respectively)orwithfluidresistant
hypotensionrequiringvasopressorsbutwith
lactatelevelof2mmol/L(18mg/dL)orless
(30.1%).
Withthesame3variablesandsimilar
categorization,theunadjustedmortalityin
infectedpatientswithin2unrelatedlarge
electronichealthrecorddatasets(University
ofPittsburghMedicalCenter[12hospitals;
20102012;n=5984]andKaiserPermanente
NorthernCalifornia[20hospitals;20092013;
n=54135])showedreproducibleresults.
Thecombinationofhypotension,vasopressor

use,andlactatelevelgreaterthan2mmol/L
(18mg/dL)identifiedpatientswithmortality
ratesof54%atUniversityofPittsburgh
MedicalCenter(n=315)and35%atKaiser
PermanenteNorthernCalifornia(n=8051).
Theserateswerehigherthanthemortality
ratesof25.2%(n=147)and18.8%(n=
3094)inpatientswithhypotensionalone,
17.9%(n=1978)and6.8%(n=30209)in
patientswithlactatelevelgreaterthan2
mmol/L(18mg/dL)alone,and20%(n=
5984)and8%(n=54135)inpatientswith
sepsisatUniversityofPittsburghMedical
CenterandKaiserPermanenteNorthern
California,respectively.
Thetaskforcerecognizedthatserumlactate
measurementsarecommonly,butnot
universally,available,especiallyin
developingcountries.Nonetheless,clinical
criteriaforsepticshockweredevelopedwith
hypotensionandhyperlactatemiaratherthan
eitheralonebecausethecombination
encompassesbothcellulardysfunctionand
cardiovascularcompromiseandisassociated
withasignificantlyhigherriskadjusted
mortality.Thisproposalwasapprovedbya
majority(13/18;72.2%)ofvoting
members13butwarrantsrevisiting.The
ControversiesandLimitationssectionbelow
providesfurtherdiscussionaboutthe
inclusionofbothparametersandoptionsfor
whenlactatelevelcannotbemeasured.

RecommendationsforICD
CodingandforLayDefinitions
Inaccordancewiththeimportanceof
accuratelyapplyingdiagnosticcodes,Table2
detailshowthenewsepsisandsepticshock
clinicalcriteriacorrelatewithICD9CMand
ICD10codes.Thetaskforcealsoendorsed
therecentlypublishedlaydefinitionthat
sepsisisalifethreateningconditionthat

ariseswhenthebodysresponsetoinfection
injuresitsowntissues,whichisconsistent
withthenewlyproposeddefinitions
describedabove.35Totransmitthe
importanceofsepsistothepublicatlarge,the
taskforceemphasizesthatsepsismayportend
death,especiallyifnotrecognizedearlyand
treatedpromptly.Indeed,despiteadvances
thatincludevaccines,antibiotics,andacute
care,sepsisremainstheprimarycauseof
deathfrominfection.Widespreadeducational
campaignsarerecommendedtobetterinform
thepublicaboutthislethalcondition.

ControversiesandLimitations
Thereareinherentchallengesindefining
sepsisandsepticshock.Firstandforemost,
sepsisisabroadtermappliedtoan
incompletelyunderstoodprocess.Thereare,
asyet,nosimpleandunambiguousclinical
criteriaorbiological,imaging,orlaboratory
featuresthatuniquelyidentifyaseptic
patient.Thetaskforcerecognizedthe
impossibilityoftryingtoachievetotal
consensusonallpoints.Pragmatic
compromiseswerenecessary,soemphasis
wasplacedongeneralizabilityandtheuseof
readilymeasurableidentifiersthatcouldbest
capturethecurrentconceptualizationof
underlyingmechanisms.Thedetailed,
dataguideddeliberationsofthetaskforce
duringan18monthperiodandthepeer
reviewprovidedbybodiesapproachedfor
endorsementhighlightedmultipleareasfor
discussion.Itisusefultoidentifytheseissues
andprovidejustificationsforthefinal
positionsadopted.
Thenewdefinitionofsepsisreflectsanupto
dateviewofpathobiology,particularlyin
regardtowhatdistinguishessepsisfrom
uncomplicatedinfection.Thetaskforcealso
offerseasilymeasurableclinicalcriteriathat
capturetheessenceofsepsisyetcanbe
translatedandrecordedobjectively(Figure).
Althoughthesecriteriacannotbeall
encompassing,theyaresimpletouseand

offerconsistencyofterminologytoclinical
practitioners,researchers,administrators,and
funders.Thephysiologicandbiochemical
testsrequiredtoscoreSOFAareoften
includedinroutinepatientcare,andscoring
canbeperformedretrospectively.
Theinitial,retrospectiveanalysisindicated
thatqSOFAcouldbeausefulclinicaltool,
especiallytophysiciansandother
practitionersworkingoutsidetheICU(and
perhapsevenoutsidethehospital,giventhat
qSOFAreliesonlyonclinicalexamination
findings),topromptlyidentifyinfected
patientslikelytofarepoorly.However,
becausemostofthedatawereextractedfrom
extractedUSdatabases,thetaskforce
stronglyencouragesprospectivevalidationin
multipleUSandnonUShealthcaresettings
toconfirmitsrobustnessandpotentialfor
incorporationintofutureiterationsofthe
definitions.Thissimplebedsidescoremaybe
particularlyrelevantinresourcepoorsettings
inwhichlaboratorydataarenotreadily
available,andwhentheliteratureaboutsepsis
epidemiologyissparse.
NeitherqSOFAnorSOFAisintendedtobea
standalonedefinitionofsepsis.Itiscrucial,
however,thatfailuretomeet2ormore
qSOFAorSOFAcriteriashouldnotleadtoa
deferralofinvestigationortreatmentof
infectionortoadelayinanyotheraspectof
caredeemednecessarybythepractitioners.
qSOFAcanberapidlyscoredatthebedside
withouttheneedforbloodtests,anditis
hopedthatitwillfacilitateprompt
identificationofaninfectionthatposesa
greaterthreattolife.Ifappropriatelaboratory
testshavenotalreadybeenundertaken,this
mayprompttestingtoidentifybiochemical
organdysfunction.Thesedatawillprimarily
aidpatientmanagementbutwillalsoenable
subsequentSOFAscoring.Thetaskforce
wishestostressthatSIRScriteriamaystill
remainusefulfortheidentificationof
infection.
Somehavearguedthatlactatemeasurement

shouldbemandatedasanimportant
biochemicalidentifierofsepsisinaninfected
patient.Becauselactatemeasurementoffered
nomeaningfulchangeinthepredictive
validitybeyond2ormoreqSOFAcriteriain
theidentificationofpatientslikelytobe
septic,thetaskforcecouldnotjustifythe
addedcomplexityandcostoflactate
measurementalongsidethesesimplebedside
criteria.Thetaskforcerecommendations
shouldnot,however,constrainthe
monitoringoflactateasaguidetotherapeutic
responseorasanindicatorofillnessseverity.
Ourapproachtohyperlactatemiawithinthe
clinicalcriteriaforsepticshockalso
generatedconflictingviews.Sometaskforce
memberssuggestedthatelevatedlactate
levelsrepresentanimportantmarkerof
crypticshockintheabsenceof
hypotension.Othersvoicedconcernaboutits
specificityandthatthenonavailabilityof
lactatemeasurementinresourcepoorsettings
wouldprecludeadiagnosisofsepticshock.
Nosolutioncansatisfyallconcerns.Lactate
levelisasensitive,albeitnonspecific,stand
aloneindicatorofcellularormetabolicstress
ratherthanshock.32However,the
combinationofhyperlactatemiawithfluid
resistanthypotensionidentifiesagroupwith
particularlyhighmortalityandthusoffersa
morerobustidentifierofthephysiologicand
epidemiologicconceptofsepticshockthan
eithercriterionalone.Identificationofseptic
shockasadistinctentityisofepidemiologic
ratherthanclinicalimportance.Although
hyperlactatemiaandhypotensionare
clinicallyconcerningasseparateentities,and
althoughtheproposedcriteriadifferfrom
thoseofotherrecentconsensusstatements,34
clinicalmanagementshouldnotbeaffected.
Thegreaterprecisionofferedbydatadriven
analysiswillimprovereportingofboththe
incidenceofsepticshockandtheassociated
mortality,inwhichcurrentfiguresvary4
fold.3Thecriteriamayalsoenhanceinsight
intothepathobiologyofsepsisandseptic
shock.Insettingsinwhichlactate

measurementisnotavailable,theuseofa
workingdiagnosisofsepticshockusing
hypotensionandothercriteriaconsistentwith
tissuehypoperfusion(eg,delayedcapillary
refill36)maybenecessary.
Thetaskforcefocusedonadultpatientsyet
recognizestheneedtodevelopsimilar
updateddefinitionsforpediatricpopulations
andtheuseofclinicalcriteriathattakeinto
accounttheiragedependentvariationin
normalphysiologicrangesandin
pathophysiologicresponses.
Abbreviations:ICD,InternationalClassificationof
Diseases;MAP,meanarterialpressure;SOFA,
Sequential[Sepsisrelated]OrganFailure
Assessment.

27

Includedtrainingcodes.

Suspectedinfectioncouldbedefinedasthe
concomitantadministrationoforalorparenteral
antibioticsandsamplingofbodyfluidcultures(blood,
urine,cerebrospinalfluid,peritoneal,etc).Forexample,
ifthecultureisobtained,theantibioticisrequiredtobe
administeredwithin72hours,whereasiftheantibiotic
isfirst,thecultureisrequiredwithin24hours.

12

Considersaperiodasgreatas48hoursbeforeandup
to24hoursafteronsetofinfection,althoughsensitivity
analyseshavetestedwindowsasshortas3hoursbefore
and3hoursafteronsetofinfection.

12

Withthespecifiedperiodaroundsuspectedinfection,
assessforshockcriteria,usinganyvasopressor
initiation(eg,dopamine,norepinephrine,epinephrine,
vasopressin,phenylephrine),anylactatelevel>2
mmol/L(18mg/dL),andmeanarterialpressure<65
mmHg.Thesecriteriarequireadequatefluid
resuscitationasdefinedbytheSurvivingSepsis
Campaignguidelines.

Implications
Thetaskforcehasgeneratednewdefinitions
thatincorporateanuptodateunderstanding
ofsepsisbiology,includingorgandys
function(Box3).However,thelackofa
criterionstandard,similartoitsabsencein
manyothersyndromicconditions,precludes

unambiguousvalidationandinsteadrequires
approximateestimationsofperformance
acrossavarietyofvaliditydomains,asout
linedabove.Toassistthebedsideclinician,
andperhapspromptanescalationofcareif
notalreadyinstituted,simpleclinicalcriteria
(qSOFA)thatidentifypatientswithsuspected
infectionwhoarelikelytohavepoor
outcomes,thatis,aprolongedICUcourse
anddeath,havebeendevelopedand
validated.
Thisapproachhasimportantepidemiologic
andinvestigativeimplications.Theproposed
criteriashouldaiddiagnosticcategorization
onceinitialassessmentandimmediate
managementarecompleted.qSOFAorSOFA
mayatsomepointbeusedasentrycriteria
forclinicaltrials.Thereispotentialconflict
withcurrentorgandysfunctionscoring
systems,earlywarningscores,ongoing
researchstudies,andpathwaydevelopments.
Manyofthesescoresandpathwayshave
beendevelopedbyconsensus,whereasan
importantaspectofthecurrentworkisthe
interrogationofdata,albeitretrospectively,
fromlargepatientpopulations.Thetaskforce
maintainsthatstandardizationofdefinitions
andclinicalcriteriaiscrucialinensuring
clearcommunicationandamoreaccurate
appreciationofthescaleoftheproblemof
sepsis.Anaddedchallengeisthatinfectionis
seldomconfirmedmicrobiologicallywhen
treatmentisstarted;evenwhen
microbiologicaltestsarecompleted,culture
positivesepsisisobservedinonly30%to
40%ofcases.Thus,whensepsis
epidemiologyisassessedandreported,
operationalizationwillnecessarilyinvolve
proxiessuchasantibioticcommencementor
aclinicallydeterminedprobabilityof
infection.Futureepidemiologystudiesshould
considerreportingtheproportionof
microbiologypositivesepsis.
Greaterclarityandconsistencywillalso
facilitateresearchandmoreaccuratecoding.
ChangestoICDcodingmaytakeseveral
yearstoenact,sotherecommendations

providedinTable2demonstratehowthenew
definitionscanbeappliedintheinterim
withinthecurrentICDsystem.
Thedebateanddiscussionthatthisworkwill
inevitablygenerateareencouraged.Aspects
ofthenewdefinitionsdoindeedrelyon
expertopinion;furtherunderstandingofthe
biologyofsepsis,theavailabilityofnew
diagnosticapproaches,andenhanced
collectionofdatawillfueltheircontinued
reevaluationandrevision.

Conclusions

Theseupdateddefinitionsandclinicalcriteria
shouldclarifylonguseddescriptorsand
facilitateearlierrecognitionandmoretimely
managementofpatientswithsepsisoratrisk
ofdevelopingit.Thisprocess,however,
remainsaworkinprogress.Asisdonewith
softwareandothercodingupdates,thetask
forcerecommendsthatthenewdefinitionbe
designatedSepsis3,withthe1991and2001
iterationsbeingrecognizedasSepsis1and
Sepsis2,respectively,toemphasizetheneed
forfutureiterations.