Anda di halaman 1dari 4

13/07/2015

www.medscape.com/viewarticle/578785_print

www.medscape.com

AScoreIdentifyingSeriousBacterialInfectionsin
ChildrenWithFeverWithoutSource
AnnickGalettoLacour,MDSamuelA.Zamora,MDAlainGervaix,MD
PediatrInfectDisJ.200827(7):654656.

Theobjectiveofthestudywastodevelopasimpleclinicaltooltoidentifyseriousbacterialinfection
(SBI)inchildrenwithfeverwithoutasource.Foreachchild,aclinicalassessment,awhitebloodcell
count,aurineanalysis,adeterminationofCreactiveprotein,procalcitonin,andappropriatecultureswereperformed.Twohundred
twochildrenwerestudiedofwhom54(27%)hadSBI.Inthemultivariateanalysis,onlyprocalcitonin[oddsratio(OR):37.6],C
reactiveprotein(OR:7.8),andurinedipstick(OR:23.2)remainedsignificantlyassociatedwithSBI.Thesensitivityofthescorefor
theidentificationofSBIwas94%andthespecificity81%.Inthevalidationsetthesensitivityofthescorewas94%andthe
specificity78%.
Feverwithoutsource(FWS)inyoungchildrenremainsadifficultdiagnosticproblem,becauseclinicalsignsandsymptomsareoften
unreliablepredictorsofaseriousbacterialinfection(SBI).Manyclinicalstudieshaveaddressedthisproblem,andthecombination
ofaclinicalevaluationassociatedwithatotalanddifferentialleukocytecountarecommonlyusedscreeningmethods. [13]The
relativelypoorspecificityofthemarkersusedtoidentifySBI,takenindependently,urgesphysicianstogiveantibioticstothe
majorityofpatients.Inourstudy,weanalyzedthepredictivevaluesofdifferentmarkersinamultivariatelogisticregression
analysis.Ourgoalwastodevelopasimplescore,whichcouldbeeasilyperformedintheemergencyroomorintheofficetopredict
SBIinapediatricpopulationwithFWS.
Weperformedacombinedanalysisofdatacollectedfrom2prospectivelyandconsecutivelyenrolledcohortsofchildrenwithFWS
inasingleuniversitycenter. [4,5]Bothcohortstudieshadthesameinclusionandexclusioncriteriaandhadfollowedsimilar
methodology.ThestudyprotocolwasapprovedbytheEthicalCommitteeoftheChildandAdolescentDepartment,University
HospitalsofGeneva.Thestudyincludedallchildrenagedfrom7daysto36monthswhowereconsecutivelyadmittedtothe
EmergencyDepartmentoftheUniversityChildren'sHospitalofGenevawitharectaltemperatureabove38Candwithoutlocalizing
signsofinfectionintheirhistoryoratphysicalexamination.Criteriaofexclusionarenotifiedinthepreviousstudies. [4,5]Allchildren
hadaclinicalscorebasedontheInfantObservationScale(IOS), [6]aurineanalysiswithcultureandblooddrawnforwhitecell
count,determinationofCreactiveprotein(CRP),procalcitonin(PCT),andculture.Lumbarpuncturewasperformedwhen
meningitiswassuspected.Thepediatricresidentinchargeofthepatientdecidedwhichchildshouldreceiveantibiotics.Allchildren
hadaclinicalfollowupwithphysicalexaminationbyapediatricianinthefollowing48hoursorbyatelephonecontact.The
diagnosiswasregisteredattheendoftheclinicalfollowup.TechnicallaboratorydeterminationsanddefinitionofSBIs:bacteremia,
pyelonephritis,lobarpneumonia,bacterialmeningitis,andcriteriaofbenigninfectionaredescribedelsewhere. [4]
Thestudypopulationwasdividedbystratifiedrandomizationinaderivationset(2/3)andavalidationset(1/3).Thesensitivity,
specificity,negative,andpositivepredictivevaluesforthedetectionofaSBIweredeterminedinthederivationsetforthedifferent
laboratoryparametersusingthecutoffpointsderivedfromourpreviousstudies. [4,5]Univariatelogisticregressionwasperformed
consideringthedichotomizedpredictiveparametersasindependentvaluesandSBIastheoutcomevalue.Then,parameters
significantlyassociatedwithSBIwereenteredforwardstepwiseintoamultipleregressionmodelandonlythoseremaining
independentlysignificantly(P<0.05)associatedwithSBIwereretained.Foreaseofuseintheclinicalsetting,wethencreateda
LaboratoryscoreusingonlythepredictivevariablesindependentlyassociatedwithSBI.Thesensitivity,specificity,andpredictive
valuesoftheLaboratoryscoreweredeterminedinthederivationsetandinthevalidationset.
TwohundredtwentytwochildrenwereconsecutivelyincludedfromMarch1998toFebruary2002.Twentychildrenwereexcluded.
Thedataof202childrenwereanalyzed.Thefinaldiagnosiswas:SBIin54children(27%)(7bacteremia,40pyelonephritis,5lobar
pulmonarycondensation,1retropharyngealabscess,and1mastoiditis),benignfocalinfectionin26children(13%)(cystitis,acute
otitismedia,adenitis,Campylobactergastroenteritis),andprobableviralinfectionin122children(60%)(negativecultureandno
signsforfocalinfectionatclinicalfollowup).Onehundredthirtyfourof202(66%)ofthechildrenreceivedantibiotics.Thestudy
populationwasdividedinaderivationset(n=135)andavalidationset(n=67).The2setswerecomparableintermsofage,
fever,incidenceofSBI,clinicalobservationalscores,andlaboratoryparameters.Thesensitivity,specificity,andpredictivevalues
forthedifferentparametersassociatedwithSBIarelistedin.
PredictiveValue(%)ofDifferentVariablesBetweenChildrenWithandWithoutSevereBacterialInfections

http://www.medscape.com/viewarticle/578785_print

1/4

13/07/2015

www.medscape.com/viewarticle/578785_print

WefirstperformedunivariatelogisticregressionwithvariablespotentiallyassociatedwithSBI.PCT[oddsratio(OR):35.6]showed
thestrongestassociationfollowedbyCRP(OR:12.9),urinedipstick(OR:9),andleucocytosis(OR:3).LeftshiftandIOSscore
werenotstatisticallyassociatedwithSBI.
ThenPCT,CRP,urinedipstick,andleucocytosiswereenteredintoaforwardstepwisemultiplelogisticregressionmodeltoidentify
independentpredictorofSBI.ThePCTvalueremainedthemostsignificantpredictorofSBI(OR:37.695%CI:5.8243).The
othervariablesindependentlyassociatedwithSBIinthisanalysiswereCRP(OR:7.895%CI:230.4)andurinedipstick(OR:
23.295%CI:5.1104.8).LeucocytosiswasnotindependentlyassociatedwiththeoccurrenceofSBI(P=0.49).
Basedontheresultsofthelogisticregressionanalysis,wedevelopedariskindexscore,namedLaboratoryscore.Therelative
weightingofeachcomponentvariableoftheLaboratoryscorewasbasedonitsoddsratiointheunivariateanalysis.Twopoints
wereattributedtoPCTorCRPabovethecutoffvalues(0.5ng/mLand40mg/L,respectively)and4pointsforvaluesofPCTabove
2ng/mL,andforCRPabove100mg/L.Onepointwasattributedforapositiveurinedipstick().
LaboratoryScore

TheperformanceoftheLaboratoryscorewasthentestedbothonthederivationpopulationandthevalidationset().Inthe
derivationset,theLaboratoryscore(3)hadasensitivityof94%andaspecificityof81%.Whencomparedwiththeother
parameterscommonlyusedtopredictSBI,theLaboratoryscorehadthebestaccuracyassociatinggoodsensitivityandspecificity.
InthevalidationsettheLaboratoryscorehadsimilarperformanceswithasensitivityof94%(95%CI:7499)andaspecificityof
78%(95%CI:6487)().
PredictiveValue(%)ofDifferentVariablesBetweenChildrenWithandWithoutSevereBacterialInfections

http://www.medscape.com/viewarticle/578785_print

2/4

13/07/2015

www.medscape.com/viewarticle/578785_print

PredictiveValue(%)ofDifferentVariablesBetweenChildrenWithandWithoutSevereBacterialInfections

OurdatashowedthatPCT,CRP,andurinedipstickareindependentpredictorsofSBIinthispopulationofchildrenwithFWS.In
ourstudy,theIOSscoreandleftshiftwerenotstatisticallydifferentbetweenchildrenwithandwithoutSBI.Moreover,leucocytosis
wasnotanindependentpredictorofSBIwhenPCT,CRP,andurinedipstickhavebeentakenintoaccount.
Wehavedevelopedascoringsystem(Laboratoryscore)basedonthe3predictivevariablesindependentlyassociatedwithSBI:
PCT,CRP,andurinarydipstick.TheprincipaladvantageoftheLaboratoryscoreisitsgoodspecificity(81%)forthepredictionof
SBIassociatedwiththesecurityofahighsensitivity(94%).ThegoodspecificityoftheLaboratoryscoreshouldenablethereliable
selectionofchildrenwhoneedantibiotictreatment,withoutovertreatingchildrenwithviralinfection.Basedonthisstudy,if
antibioticshadsolelybeenadministratedforchildrenwithapositivescore,only40%ofthepopulationwouldhavereceived
antibiotics.Incomparison,basedontheclinician'sdecisions,morethan65%ofthestudiedpopulationreceivedantibiotics.Theuse
oftheLaboratoryscorecould,thus,substantiallyreduceantibioticuse.
Potentiallimitationsofourstudyshouldbeconsidered.Ourstudypopulationisrelativelysmallexplainingthewidenessofthe
confidenceintervalsaroundtheestimatesofsensitivityandspecificity.TheincidenceofSBI(27%)inourstudyseemshigherthan
reportedinotherstudies, [79]butsimilartotheincidenceofarecentstudyfromItaly(23%)thatanalyzedcomparablepopulationsof
childreninatertiaryhospital. [10]Thislikelyreflectsreferralbias,aspediatriciansreferillappearingchildrentoourhospitalforinitial
workup.BecausethisbiasaffectstheprevalenceofSBIinourpatientpopulation,thepredictivevaluesoftheLaboratoryscore
mustbeinterpretedwithcaution,andtheperformanceoftheLaboratoryscoremightvaryifappliedtoothercohortsofchildren.In
contrast,thesensitivityandspecificityofourscoringsystemarenotaffectedbythispotentialbias.Aninternalvalidationofthe
scorewasperformedonasubsetofthepopulation.However,thissampleissmallandthepotentialbiasassociatedwithourentire
populationremains.
Inconclusion,PCT,CRP,andurinedipstickareindependentpredictorsofSBIinthisstudy.Whitebloodcellcountisnotan
independentpredictor,whenthese3variablesaretakenintoaccount.ALaboratoryscoreincludingPCT,CRP,andurinedipstick
providesasecurityequivalenttothestandardworkup,iseasiertouse,andcouldconsiderablydiminishantibioticuseinchildren
withbenigninfection.However,childrenshouldbecarefullyfollowedup,toidentifythesmallproportionwithSBInotinitially
detectedbyapositivescore.Finally,theLaboratoryscoreshouldbeprospectivelyvalidatedandevaluatedindifferentclinical
settingsbeforeitsuseinclinicalguidelinesofchildrenwithFWS.
References

1. BaraffLJ,SchrigerDL,BassJW,etal.Practiceguidelineforthemanagementofinfantsandchildren0to36monthsofage
withfeverwithoutsource.Pediatrics.199392:112.

http://www.medscape.com/viewarticle/578785_print

3/4

13/07/2015

www.medscape.com/viewarticle/578785_print

2. BaraffLJ.Managementoffeverwithoutsourceininfantsandchildren.AnnEmergMed.200036:602614.
3. AmericanCollegeofEmergencyPhysiciansClinicalPoliciesCommitteeAmericanCollegeofEmergencyPhysiciansClinical
PoliciesSubcommitteeonPediatricFever.Clinicalpolicyforchildrenyoungerthanthreeyearspresentingtotheemergency
departmentwithfever.AnnEmergMed.200342:530545.
4. GalettoLacourA,ZamoraSA,GervaixA.BedsideprocalcitoninandCreactiveproteintestsinchildrenwithfeverwithout
localizingsignsofinfectionseeninareferralcenter.Pediatrics.2003112:10541060.
5. GalettoLacourA,GervaixA,ZamoraSA,etal.Procalcitonin,IL6,IL8,IL1receptorantagonistandCreactiveproteinas
identificatorsofseriousbacterialinfectionsinchildrenwithfeverwithoutlocalisingsigns.EurJPediatr.2001160:95100.
6. McCarthyP,SharpeM,SpieselS,etal.Observationscalestoidentifyseriousillnessinfebrilechildren.Pediatrics.
198270:802809.
7. ThayyilS,ShenoyM,HamalubaM,GuptaA,FraterJ,VerberIG.Isprocalcitoninusefulinearlydiagnosisofserious
bacterialinfectionsinchildren?ActaPaediatr.200594:155158.
8. PulliamPN,AttiaMW,CronanKM.Creactiveproteininfebrilechildren1to36monthsofagewithclinicallyundetectable
seriousbacterialinfection.Pediatrics.2001108:12751279.
9. BachurRG,HarperMB.Predictivemodelforseriousbacterialinfectionsamonginfantsyoungerthan3monthsofage.
Pediatrics.2001108:311316.
10. AndreolaB,BressanS,CallegaroS,LiveraniA,PlebaniM,DaDaltL.ProcalcitoninandCreactiveproteinasdiagnostic
markersofseverebacterialinfectionsinfebrileinfantsandchildrenintheemergencydepartment.PediatrInfectDisJ.
200726:672677.
ReprintAddress
AnnickGalettoLacour,MD,Dpartementdel'enfantetdel'adolescent,HUG,HpitaldesEnfants,rueWillyDonz6,1211
Genve14.Email:annick.galetto@hcuge.ch
PediatrInfectDisJ.200827(7):654656.2008LippincottWilliams&Wilkins

http://www.medscape.com/viewarticle/578785_print

4/4

Anda mungkin juga menyukai