Non-Purulent Skin and Soft Tissue Infections in the ... · Non-purulent skin and soft tissue infections (SSTIs) describe infections of the superficial epidermis and dermis (erysipelas)

Non-PurulentSkinandSoftTissueInfectionsintheEmergencyDepartment

KrishanYadav

ThesissubmittedtotheFacultyofGraduateandPostdoctoralStudiesinpartial

fulfillmentofthedegreerequirementsfortheM.Sc.inEpidemiology

SchoolofEpidemiology

FacultyofMedicine

UniversityofOttawa

©KrishanYadav,Ottawa,Canada,2018

ii

AbstractNon-purulentskinandsofttissueinfections(SSTIs)involvetheepidermisand

dermisandarecommonlymanagedintheemergencydepartment(ED).Current

guidelineslackevidencetoguidecliniciansonoptimalmanagement.Theaimofthis

thesiswastodescribetheepidemiologyofadultswithnon-purulentSSTIs

presentingtotheED.Secondarygoalsweretoidentifyfactorsassociatedwithoral

antibiotictreatmentfailure;describeanoutpatientparenteralantibiotictherapy

(OPAT)clinic-to-EDprogram;anddeterminephysicianrationaleforselecting

intravenoustherapy.Weconductedahealthrecordsreviewandprospective

observationalcohortstudy.

Therewassignificantphysicianpracticevariationandanunexpectedlyhigh

hospitalizationrate.Weidentifiedfourfactorsassociatedwithoralantibiotic

treatmentfailure(tachypnea,chroniculcers,historyofmethicillinresistant

Staphylococcusaureuscolonizationorinfection,andcellulitisinthepast12months).

AnED-to-OPATclinicprogramwasfoundtobesafe,withlowtreatmentfailure

ratesandhighpatientsatisfaction.

iii

ExecutiveSummaryNon-purulentskinandsofttissueinfections(SSTIs)describeinfectionsofthe

superficialepidermisanddermis(erysipelas)orinvolvementofthedeepdermis

andsubcutaneoustissue(cellulitis).Theseinfectionsarecommonlydiagnosedand

managedintheemergencydepartment(ED)setting.Therearenostudiesthathave

describedtheepidemiologyofadultswithnon-purulentSSTIswhoaremanagedin

theED.Furthermore;currentguidelinesarebasedonexpertopinionandlack

evidencetoguidephysiciansontheoptimalrouteofantibiotictherapy(oralversus

intravenous).

Theaimofthisthesisprojectwastodescribetheepidemiologyofadultswithnon-

purulentSSTIswhopresenttotheED.Secondarygoalswereto:(1)identify

predictorsassociatedwithoralantibiotictreatmentfailure;(2)describeemergency

physicianrationaleforselectingtheintravenousroute;and(3)describetherateof

treatmentfailure,adverseeventsandoverallpatientsatisfactionforadultstreated

atanoutpatientparenteralantimicrobialtherapy(OPAT)clinicafterinitial

diagnosisandmanagementintheED.Tomeetthesegoals,weconductedahealth

recordsreviewandprospectiveobservationalcohortstudy.

Thefindingsfromthisthesisprojectareimportantfirststepstobetterunderstand

whatfactorspredisposetooralantibiotictreatmentfailure,whyemergency

iv

physiciansselecttheintravenousroute,andpatientoutcomesinanED-to-OPAT

clinicmodelformanagementofnon-purulentSSTIs.

ContributionsoftheAuthorsDr.KrishanYadavisthefirstauthorofbothmanuscriptsandwasprimarily

responsibleforstudydesign,datacollection,statisticalanalysisandwritingofthe

manuscripts.Bothmanuscriptswereco-authoredbyDr.Yadav’sthesissupervisors

Dr.IanGStiellandDr.GeorgeWells,andthesisadvisorycommitteememberDr.

KathrynSuh.Twomedicalstudents(Mr.JohnMacisaacandMr.DarmynRitchie)

aidedinchartreviewanddatacollectionforthehealthrecordsreview.Mr.Jordan

Bernickprovidedhelpfulfeedbackregardingstatisticalanalysis.Drs.Eaglesand

Thiruganasambandamoorthyprovidedhelpfulsuggestionsandfeedbackforboth

manuscripts.Drs.Stiell,WellsandSuhprovidedvaluablefeedbackthroughoutthe

entireprocess.

v

AcknowledgmentsIwouldliketothankmyprimarythesissupervisorDr.IanGStiellforhisimmense

support,mentorshipandguidanceoverthepastthreeyears.Iwillforeverbe

gratefulfortheopportunitytolearnfromDr.Stiell;Ihopetomakehimproudwith

myplannedfutureresearchendeavors.Itwasanequalhonourandprivilegetohave

theguidanceandexpertiseofmyco-supervisor,Dr.GeorgeWells.Ilearnedagreat

dealandalwaysenjoyedourmeetingstodiscussstatistics,researchandlifein

general.IoweahugedebtofgratitudetoDr.KathrynSuh.Herexpertise,

encouragementandmentorshipweregreatlyappreciated.Ilookforwardtofuture

researchcollaborationswithDr.Suhandherteam.

AsincerethankstoAngelaMarcantonio,CatherineClementandtheDepartmentof

EmergencyMedicineresearchteam.Ifeelveryfortunatetohavehadthe

opportunitytoworkwithandlearnfromsuchaterrificgroup.

Thankyoutomyfamilyforallofyoursupportandlove.MywifeJesshasbeen

integraltothisentireprojectandincrediblysupportiveeverystepoftheway–I

cannotthankyouenough.Iameternallygratefultomyparentsforencouragingme

topursuemydreamsandtofurthermyeducationeverystepoftheway.Finally,I

dedicatethisthesistotwoveryspeciallittlepeople:AnandYadavandPriyanka

(Bean)Yadav.Idohopethatthisworkonedayinspiresyoutopursueandachieve

yourowngoalsanddreams.

vi

TableofContentsAbstract......................................................................................................................................ii

ExecutiveSummary...............................................................................................................iiiContributionsoftheAuthors.............................................................................................iv

Acknowledgments....................................................................................................................v

TableofContents....................................................................................................................viListofTables............................................................................................................................ix

ListofFigures..........................................................................................................................xi

AbbreviationsUsedintheText........................................................................................xiiChapterOne:Introduction...................................................................................................11.Introduction...................................................................................................................................12.Rationale.........................................................................................................................................23.ThesisGoalsandObjectives......................................................................................................2ChapterOne........................................................................................................................................3ChapterTwo.......................................................................................................................................3ChapterThree....................................................................................................................................3ChapterFour.......................................................................................................................................4ChapterFive........................................................................................................................................4

ChapterTwo:Background...................................................................................................61.Introduction...................................................................................................................................62.Background....................................................................................................................................62.1Classification..............................................................................................................................................62.2Incidence&BurdenofDisease..........................................................................................................72.3Etiology&RiskFactors.........................................................................................................................82.4ClinicalFeatures&Complications....................................................................................................92.5Diagnosis..................................................................................................................................................102.6CurrentGuidelines...............................................................................................................................102.7OralversusIntravenousTherapy..................................................................................................122.8EDTreatment.........................................................................................................................................142.9QuestionsSurroundingOptimalEDManagementofSSTIs................................................212.10TreatmentFailure..............................................................................................................................212.11PredictorsofTreatmentFailureWithOralAntibiotics.....................................................252.12OralVersusTheIntravenousRoute:EmergencyPhysicianRationaleforSelectingIntravenousTherapy..................................................................................................................................26

3.Rationale......................................................................................................................................27References........................................................................................................................................29

ChapterThree:PredictorsofOralAntibioticFailureforNon-PurulentSkinandSoftTissueInfectionsintheEmergencyDepartment...............................................35ChapterOverview..........................................................................................................................35Introduction.....................................................................................................................................39Methods.............................................................................................................................................40

vii

StudyDesignandSetting..........................................................................................................................40Population.......................................................................................................................................................40StudyProtocolandDataAbstraction..................................................................................................41OutcomeMeasures......................................................................................................................................42StatisticalAnalysis.......................................................................................................................................43SampleSizeandFeasibility......................................................................................................................44

Results...............................................................................................................................................45Discussion.........................................................................................................................................48InterpretationofResults...........................................................................................................................48PreviousStudies...........................................................................................................................................49Strengths..........................................................................................................................................................49Limitations......................................................................................................................................................49ClinicalImplications...................................................................................................................................51ResearchImplications................................................................................................................................52

Conclusions......................................................................................................................................52References........................................................................................................................................53Figures...............................................................................................................................................56Tables.................................................................................................................................................57SupplementaryAppendix...........................................................................................................65

ChapterFour:OutpatientParenteralAntibioticTherapyFollowingEmergencyDepartmentTreatmentofNon-PurulentSkinandSoftTissueInfections.........77ChapterOverview..........................................................................................................................77Abstract.............................................................................................................................................80Methods.............................................................................................................................................83StudyDesignandSetting..........................................................................................................................83StudyPopulation..........................................................................................................................................84IntravenousAntibioticTreatment........................................................................................................84DataCollection..............................................................................................................................................84OutcomeMeasures......................................................................................................................................85DataAnalysis..................................................................................................................................................86SampleSize.....................................................................................................................................................87

Results...............................................................................................................................................87Discussion.........................................................................................................................................89InterpretationofResults...........................................................................................................................89PreviousStudies...........................................................................................................................................90Strengths..........................................................................................................................................................90Limitations......................................................................................................................................................91ClinicalImplications...................................................................................................................................92ResearchImplications................................................................................................................................93

Conclusion........................................................................................................................................93References........................................................................................................................................94Figures...............................................................................................................................................97Tables.................................................................................................................................................98SupplementaryAppendix.........................................................................................................105

ChapterFive:Discussion..................................................................................................114Introduction...................................................................................................................................114InterpretationofResults...........................................................................................................114PreviousStudies...........................................................................................................................116Strengths.........................................................................................................................................118

viii

Limitations.....................................................................................................................................118ClinicalImplications...................................................................................................................120ResearchImplications................................................................................................................122Conclusions....................................................................................................................................124References......................................................................................................................................126

AppendixA:OttawaHealthScienceNetworkResearchEthicsBoardApprovalLetter......................................................................................................................................129

AppendixB:MethodsforDevelopingaMultivariableLogisticRegressionModelforPredictorsofOralAntibioticTreatmentFailure.................................130Introduction...................................................................................................................................130ExploratoryAnalysis...................................................................................................................130Outliers...........................................................................................................................................................131MissingData.................................................................................................................................................131ExcludingVariables...................................................................................................................................131Associationsbetweencategoricalvariables...................................................................................133Correlationbetweencontinuousvariables.....................................................................................133

LogisticRegressionAnalysis....................................................................................................134UnivariateAnalysis...................................................................................................................................134AssessingforInteraction........................................................................................................................138MultivariableLogisticRegression:PreliminaryModel.............................................................139SecondaryAnalysis(ComparingOralvs.IntravenousTreatmentGroups).....................141

Discussion.......................................................................................................................................143Conclusion......................................................................................................................................144References......................................................................................................................................146

ix

ListofTablesChapterThreeTable1.BaselineCharacteristicsofAdultswithNonpurulentSkinandSoftTissueInfections

seenintheEmergencyDepartment(N=500)..................................................................................57Table2.PresentingPatientandInfectionCharacteristics(N=500).................................................58Table3.EmergencyDepartmentTreatment(N=500)...........................................................................59Table4.AntibioticTreatmentfor352PatientsDischargedfromtheED......................................60Table5.AdverseEventsfor352PatientsDischargedfromtheED..................................................61Table6.OutpatientParenteralAntibioticTherapy(OPAT)ClinicData(N=85).........................62Table7.TreatmentFailurewithOralAntibiotics(N=85of288PatientsTreatedwitha

Minimumof48HoursofOralTherapy).............................................................................................63Table8.PredictorsAssociatedwithOralAntibioticTreatmentFailureUsingMultivariable

LogisticRegression(N=288)...................................................................................................................64TableS1.VariableDefinitions...........................................................................................................................72TableS2.UnivariateAssociationwithOralAntibioticTreatmentFailurefor288EDPatients

TreatedwithaMinimumof48HoursofOralTherapy...............................................................73TableS3.TreatmentFailurewithIVAntibiotics(N=12of212PatientsTreatedwitha

Minimumof48HoursofIVTherapy).................................................................................................74TableS4.SecondaryAnalysisComparingOralversusIntravenousAntibioticGroupsfor

CategoricalVariablesforall500Patients.........................................................................................75TableS5.SecondaryAnalysisComparingOralversusIntravenousAntibioticGroupsfor

ContinuousVariablesforall500Patients.........................................................................................76ChapterFourTable1.BaselineCharacteristicsofAdultswihtNon-PurulentSkinandSoftTissue

Infections(SSTIs)seenintheED(N=153).....................................................................................98Table2.PresentingPatientandInfectionCharacteristics(N=153).................................................99Table3.IntravenousAntibioticTreatmentAdministeredWhileintheED(N=153).............100Table4.IntravenousAntibioticPrescriptionsforPatientsDischargedfromtheED(N=153)

............................................................................................................................................................................101Table5.EmergencyPhysicianRationaleforIVAntibioticsforall153Patients.......................102Table6.OutpatientParenteralAntibioticTherapy(OPAT)ClinicData(N=137)....................103Table7.Outcomesat14DaysfromIndexEDVisitfor137PatientsNotLosttoFollow-up104TableS1.ReasonsforHospitalizationwithin14DaysofEDVisitfor137Patientswho

attendedtheirOPATClinicAppointment........................................................................................113AppendixBTableB1.UnivariateAssociationofCharacteristicswithTreatmentFailureoftheStudy

Participants(N=288)................................................................................................................................132TableB2.AssessmentofSimpleCollinearityBetweenCategoricalVariablesUsingthePhi

Coefficient......................................................................................................................................................133

x

TableB3.CorrelationBetweenContinuousVariablesUsingPearsonCorrelationCoefficient(r)......................................................................................................................................................................134

TableB4.UnivariateLogisticRegressionforAgeasaPredictorofOralAntibioticTreatmentFailure.............................................................................................................................................................135

TableB5.UnivariateLogisticRegressionforSystolicBloodPressureasaPredictorofOralAntibioticTreatmentFailure................................................................................................................135

TableB6.UnivariateLogisticRegressionforHeartRateasaPredictorofOralAntibioticTreatmentFailure......................................................................................................................................136

TableB7.UnivariateLogisticRegressionforTemperatureasaPredictorofOralAntibioticTreatmentFailure......................................................................................................................................136

TableB8.UnivariateLogisticRegressionforRespiratoryRateasaPredictorofOralAntibioticTreatmentFailure................................................................................................................136

TableB9.SimpleUnivariateLogisticRegressionforCategoricalVariables...............................137TableB10.AssessmentofPossibleInteractionTerms........................................................................138TableB11.MultivariableLogisticRegressionModelofPredictorsAssociatedwithOral

AntibioticTreatmentFailure(N=288).............................................................................................139TableB12.MultivariableLogisticRegressionModelUsingBackwardsSelectionof

PredictorsofOralAntibioticTreatmentFailure(N=288).......................................................140TableB13.FinalMultivariableLogisticRegressionModelofPredictorsAssociatedwithOral

AntibioticTreatmentFailure(N=288).............................................................................................140TableB14.SecondaryAnalysisComparingOralversusIntravenousAntibioticGroupsfor

CategoricalVariablesforall500Patients.......................................................................................142TableB15.SecondaryAnalysisComparingOralversusIntravenousAntibioticGroupsfor

ContinuousVariablesforall500Patients.......................................................................................143

xi

ListofFiguresChapterTwoFigure1.Classificationofskinandsofttissueinfections(SSTIs).........................................................8Figure2.EDTreatmentPathwaysforPatientswithSSTIs..................................................................15ChapterThreeFigure1.FlowDiagramofPatientEligibilityandOutcomes...............................................................56FigureS1.StandardizedCaseRecordForm................................................................................................65ChapterFourFigure1.FlowDiagram........................................................................................................................................97FigureS1.OPATClinicReferralForm..........................................................................................................105FigureS2.StandardizedCaseRecordForm..............................................................................................106

xii

AbbreviationsUsedintheTextSSTI–skinandsofttissueinfections

ED–emergencydepartment

OPAT–outpatientparenteralantibiotictherapy

CA-MRSA–communityacquiredmethicillinresistantstaphylococcusaureus

MRSA–methicillinresistantstaphylococcusaureus

ABSSI–acutebacterialskinandskinstructureinfections

IDSA–InfectiousDiseaseSocietyofAmerica

CREST–clinicalresourceefficiencysupportteam

PICC–peripherallyinsertedcentralcatheter

CCAC–communitycareaccesscentre

LHIN–localhealthintegrationnetwork

CDER–centerfordrugevaluationandresearch

CI–confidenceinterval

OR–oddsratio

ICD-10-CA–InternationalClassificationofDiseases,10threvision,Canada

IQR–interquartilerange

1

ChapterOne:Introduction

1.IntroductionSkinandsofttissueinfections(SSTIs)refertoaspectrumofdiseaseprocessesthat

encompassbacterialinfectionsoftheepidermisanddeeperdermallayers.Severity

rangesfromsuperficialprocesses(erysipelasandcellulitis)tolife-orlimb-

threateninginfectionsofdeeperstructures(necrotizingfasciitis).Non-purulent

SSTIsarethesubsetofinfectionsthatdonotcontainpus(acollectionofwhiteblood

cells,nonviabletissuesanddegradedcellularcontents).Non-purulentSSTIsare

commonandaccountforapproximately14.5millioncasesannuallyintheUnited

Statesalone.1However,cliniciansarefacedwithmanychallengesinbothdiagnosis

andmanagementofthisseeminglysimplediseaseentity.

SSTIsremainaclinicaldiagnosis.Therearemanydiagnosticmimics,includingbut

notlimitedto:deepvenousthrombosis;stasisdermatitis;hematoma;gout;and

contactdermatitis.Attimesitmaybedifficulttodifferentiatenon-purulentSSTIs

fromabscesses.Thisisimportant,asthelatteristreatedwithsurgicalincisionand

drainageasopposedtoantibioticsalone.Therearenospecificlaboratoryteststo

confirmthediagnosis.The2014InfectiousDiseaseSocietyofAmericaguidelinesin

factrecommendagainstroutinelaboratorytestingfornon-purulentSSTIs–

includingbloodtests,biopsyculturesorswabs.2Oncethediagnosisismade,

cliniciansareexpectedtoselectappropriateantimicrobialtherapy.Decidingon

optimaltreatmentiscomplicatedbytheneedtoselectnotonlythecorrect

antimicrobialagent,butalsothecorrectrouteofdelivery(oralversusintravenous).

2

Duetothelackofevidence,currentguidelinesonmanagementofSSTIsarebased

onexpertopinion.2,3

2.RationaleAccuratediagnosisandappropriatetreatmentofSSTIsarecriticalforpatientsafety,

optimizingclinicaloutcomeanddecreasinghealthcarecostsandoverallburden.The

epidemiologyofSSTIshasbeenwelldocumentedintwodistinctpopulations:

hospitalizedpatientsandthosewhoseekcareatphysicians’offices.4-6However,

therearenostudiesthathavedescribedtheepidemiologyofadultswithSSTIswho

seekcareintheEmergencyDepartment(ED).Thisconstitutesasignificantevidence

gapwhenoneconsidersthatSSTIsaccountforasmuchas3%ofallEDvisits.7A

thoroughunderstandingoftheepidemiologyofnon-purulentSSTIsintheED

patientpopulationwouldprovideanimportantbasistowarddevelopingmore

appropriateevidence-basedrecommendationsformanagementofthiscommon

condition.

3.ThesisGoalsandObjectivesTheprincipalgoalofthisthesisistoobtainathoroughunderstandingofthe

epidemiologyofnon-purulentSSTIsinadultswhoaremanagedinaCanadianED

setting.Asecondarygoalwillbetoidentifyanyvariablesindependentlyassociated

withtheoutcomeoftreatmentfailurewithoralandintravenousantibioticsat14

daysfromtheindexEDvisit.Thesegoalswillbeaddressedthroughahealthrecords

review(StudyA)ofallpatientspresentingtotheEDwithnon-purulentSSTIs.A

3

thirdgoalwillbetodescribethefunctionandoutcomesofanOutpatientParenteral

AntibioticTherapy(OPAT)CliniconadultEDpatientswithSSTIsthatarefeltto

requireintravenoustherapy.Thisstudywillspecificallydescribetheoutcomeof

OPATtreatmentfailureandemergencyphysicianrationaleforselectingintravenous

antibioticsfornon-purulentSSTIs.AprospectiveobservationalcohortstudyofED

patientstreatedattheOPATClinicwillbeconducted(StudyB)toaddressthisfinal

goal.

Theobjectivesofeachchapterinthisthesisareasfollows:

ChapterOneAbriefintroductionthatprovides:(a)therationaleandimportanceofresearching

EDpatientswithnon-purulentSSTIs;(b)thethesisgoalsandobjectives;and(c)the

objectivesofeachchapter.

ChapterTwoThischapterprovidesthebackgrounddiscussionconcerningthepathophysiology,

diagnosisandoutpatientmanagementofnon-purulentSSTIsintheEDsetting.

ChapterThreeAmanuscriptofthehealthrecordsreview(StudyA).Thegoalsofthemanuscript

areto:1)describetheepidemiologyofnon-purulentSSTIsinadultswithSSTIswho

presenttotheED;and2)identifyvariablesindependentlyassociatedwiththe

4

clinicaloutcomeoftreatmentfailurewithoralantibioticsat14daysfromtheindex

EDvisit.

ChapterFourAmanuscriptoftheprospectiveobservationalcohortstudy(StudyB).Thegoalsof

themanuscriptareto:1)describeoutcomesoftheOPATClinicinEDpatients

includingOPATtreatmentfailureandpatientadverseevents;2)assesspatient

satisfactionwithOPAT;and3)identifyemergencyphysicianrationaleforselecting

intravenousantibiotics.

ChapterFiveAdiscussionbasedonthecombinedresultsofStudyAandStudyB.Inadditionto

summarizingtheresultsanddrawingconclusions,implicationsforfutureresearch

willbedetermined.

5

References

1. RaffAB,KroshinskyD.Cellulitis:AReview.JAMA.2016;316(3):325-337.doi:310.1001/jama.2016.8825.

2. StevensDL,BisnoAL,ChambersHF,etal.Practiceguidelinesforthediagnosisandmanagementofskinandsofttissueinfections:2014updatebytheInfectiousDiseasesSocietyofAmerica.ClinInfectDis.2014;59(2):e10-52.doi:10.1093/cid/ciu1444.

3. ClinicalResourceEfficiencySupportTeam(2005)Guidelinesonthemanagementofcellulitisinadults.CREST,Belfast..

4. Perello-AlzamoraMR,Santos-DuranJC,Sanchez-BarbaM,CanuetoJ,MarcosM,UnamunoP.Clinicalandepidemiologicalcharacteristicsofadultpatientshospitalizedforerysipelasandcellulitis.EurJClinMicrobiolInfectDis.2012;31(9):2147-2152.

5. ZervosMJ,FreemanK,VoL,etal.Epidemiologyandoutcomesofcomplicatedskinandsofttissueinfectionsinhospitalizedpatients.JClinMicrobiol.2012;50(2):238-245.

6. PallinDJ,EspinolaJA,LeungDY,HooperDC,CamargoCA,Jr.EpidemiologyofdermatitisandskininfectionsinUnitedStatesphysicians'offices,1993-2005.ClinInfectDis.2009;49(6):901-907.doi:910.1086/605434.

7. PallinDJ,EganDJ,PelletierAJ,EspinolaJA,HooperDC,CamargoCA,Jr.IncreasedUSemergencydepartmentvisitsforskinandsofttissueinfections,andchangesinantibioticchoices,duringtheemergenceofcommunity-associatedmethicillin-resistantStaphylococcusaureus.AnnEmergMed.2008;51(3):291-298.doi:210.1016/j.annemergmed.2007.1012.1004.Epub2008Jan1028.

6

ChapterTwo:Background

1.IntroductionUncomplicated,non-purulentskinandsofttissueinfections(SSTIs)describe

infectionsofthesuperficialepidermisanddermis(erysipelas)oradditionaldeeper

involvementincludingthedeepdermisandsubcutaneoustissue(cellulitis).1

Patientswithnon-purulentSSTIspresentwithredness,painandswellingofthe

involvedskin.Itisimportanttonotethatthereisanabsenceofanypurulent(i.e.

pus-containing)materialorabscess.TheEmergencyDepartment(ED)physician

mustfirstestablishthediagnosis,andthendecideontheoptimalagent,dose,

frequency,routeandsettingforantimicrobialtherapy.Thesedecisionshavebeen

complicatedfurtherbytheincreasingprevalenceofcommunityacquired

methicillin-resistantStaphylococcusaureus(CA-MRSA).2,3Appropriatediagnosis

andmanagementofSSTIsiscrucialinordertopreventcomplicationssuchas

bacteremiaorthedevelopmentofnecrotizingfasciitis.

2.Background

2.1ClassificationSSTIsmaybesubdividedintopurulentornon-purulentcategories(Figure1).

PurulentSSTIsareinfectionsthatcontainpus,whichisacollectionofwhiteblood

cells,nonviabletissueanddegradedcellularcontents.PurulentSSTIsinclude

furuncles(purulentmaterialfromthehairfollicle–alsocalledaboil);carbuncles

(collectionoffuruncleswithcommunicatingtracts);andabscesses(deeperwalled-

offcollectionofpurulentmaterial).Whenfeasiblethissubgroupistreatedwith

7

surgicalincisionanddrainage.Antibioticsarerecommendedonlyiftherearesigns

ofsystemicillness.4Non-purulentSSTIsarenotamenabletosurgicaldrainage.

Instead,thissubgroupistreatedwithantibiotics.Non-purulentSSTIsmaybe

furthersubdividedintouncomplicatedversuscomplicatedcategories.

Uncomplicatednon-purulentSSTIsincludecellulitisanderysipelas.Complicated

non-purulentSSTIsconsistoflife-threateningnecrotizinginfectionsofdeeper

tissues,includingthefasciaormuscle.Necrotizinginfectionsarerapidlyprogressive

andcreatesignificantdestructionoftissue.Patientswithnecrotizinginfections

classicallypresentwithseverepainoutofproportiontothefindingsonclinical

exam,andareatriskoflimblossorevendeath.Thefocusofthisthesisisonthe

mostcommonsubset:uncomplicated,non-purulentSSTIs.

2.2Incidence&BurdenofDiseaseSSTIsareacommonconditiondiagnosedandmanagedinEDsandcarrysignificant

financialburdenonhealthcaresystemsglobally.From1997to2005,thenumberof

AmericansseekingmedicalcareforSSTIsincreasedby50%,with14.2millionvisits

in2005alone.5PatientswithSSTIsaccountforupto3%ofallEDvisitsintheUnited

States,translatingto3.4millionvisits.6,7AlthoughCanadiandataarelacking,a

singleVancouverEDdiagnosed2234patientswithaSSTIbetweenJanuary2003

andSeptember2004,representing2%ofallEDvisits.8A2011reportbythe

CanadianInstituteforHealthInformationfoundthatcellulitiswasthe5thand7th

mostcommonreasonforanEDvisitinthe45to64andgreaterthan65yearsage

groups,respectively.9

8

Figure1.Classificationofskinandsofttissueinfections(SSTIs)

SSTIsareresponsibleforasignificanthealthcaresystemburdenduetohospital

admissionandsubsequentcosts.SSTIsmanagedintheEDresultina13.9–15.2%

hospitaladmissionrate,withthemostcommoncitedreasonbeingtheneedfor

intravenousantibiotics.10,11HospitalizationforcomplicatedSSTIs,termedacute

bacterialskinandskinstructureinfections12,resultinanaveragecostof$8023with

ameanhospitallengthofstayof4.9days.13Onestudyfoundthatfrom2005to

2011,therateofadmissionforABSSIincreasedby17%,accountingfor2%ofall

hospitaladmissionsintheUnitedStates.14

2.3Etiology&RiskFactorsGroupAstreptococcus(Streptococcuspyogenes),β-hemolyticgroupB,CandG

streptococci,andStaphylococcusaureusarethemostcommonoffendingbacteria.

Typically,aportalofentrythatdisruptstheprotectivecutaneousbarrier

9

predisposesapatienttodevelopinganSSTI.Examplesincludeskintrauma,surgical

incisions,injectiondruguse,andulcerformation.Additionalriskfactorsfornon-

purulentSSTIsinclude:lymphedema,venousinsufficiency,obesity,priorhistoryof

cellulitisandtineapedis.Ofnote,case-controlstudiesfailedtofindanyassociation

observedwithdiabetesmellitus,alcoholorsmoking.15,16Inaminorityofcases,the

sourceofinfectionmaybehematogenousseedingoridiopathic.

2.4ClinicalFeatures&ComplicationsPatientstypicallypresentwithpain,redness,swellingandindurationoftheaffected

skin.Patientswithcellulitishaveirregular,patchybordersofaffectedreddened

skin.Conversely,thebordersarewelldemarcated,raisedandpalpableinpatients

witherysipelas.Insomecases,theremaybelymphangitis(erythematousstreaking

alongtheaffectedextremityalongthedistributionofthebloodvessels)or

lymphadenopathy(painfulswellingoflymphnodes).Aminorityofpatientsmay

alsoexhibitsystemicsignssuchasfeverortachycardia.Iftreatedwithappropriate

antimicrobialtherapy,uncomplicatednon-purulentSSTIsshouldresolvewithinfive

tosevendays.Patientsmayexperiencemildpainandrednessforseveraldays

beyondthistimeframeastheresidualinflammationsubsides.

Somepatientsmaysuffercomplicationssuchassepsisorbloodstreaminfections

(bacteremia)thatwarranthospitaladmission.Onoccasion,apparentnon-purulent

infectionsdevelopintowalled-offabscessesthatrequiresurgicalincisionand

drainage.Ifnotpromptlytreated,cellulitisinvolvingapreexistingulcermay

10

progresstoinfectunderlyingbone(osteomyelitis),whichoftenrequiresseveral

weeksofantimicrobialtherapy.Seriousadverseeventssuchaslimbamputationor

deatharerare.

2.5DiagnosisWoundculturesarenotpossiblefornon-purulentSSTIs.Asystematicreviewfound

therateofbacteremiasecondarytocellulitisorerysipelasrangedfromlessthan1%

to7.9%.17Bloodculturesaretypicallylowyieldandrarelychangemanagement.

Thus,adjuncttestsareunhelpfulandthediagnosisofSSTIsremainsaclinicalone.

However,diagnosisofuncomplicatednon-purulentSSTIsremainschallenging.A

recentlargecross-sectionalstudyfounda30.5%misdiagnosisrateoflower

extremitycellulitis(termed‘pseudocellulitis’).Thestudyauthorsestimatedthis

misdiagnosisratewouldresultin50,000to130,000unnecessaryhospitalizations

and$195to$515millionUSdollarsinexcesshealthcarespending.18Thedifficulty

intheaccuratediagnosisofSSTIsisduetoseveralmimics,suchas:stasisdermatitis;

lymphedema;gout;deepveinthrombosis;andcutaneousdrugeruption.

2.6CurrentGuidelinesOwingtoalackofhighqualityevidence,empirictreatmentguidelinesbasedon

expertopinionhavebeenpublishedtoaidclinicians.4,19-21Therearecurrentlyno

CanadianguidelinesforthemanagementofSSTIs.Despitetheincreasingburdenof

thiscommonpresentation,currentevidenceislackingregardingtheoptimal

11

managementofSSTIs.ArecentCochranereviewconcludedthattheoptimal

antimicrobialtherapyforSSTIsremainsunclear,asnotworandomizedcontrolled

trials,amongthe25identifiedstudies,comparedthesametwoantibiotic

regimens.21Recentstudieshaveshownpromisefornovelonce-weeklyparenteral

lipoglycopeptideantibioticsforthetreatmentofSSTIs,whichmaybeespecially

usefulincommunitiesthatlacktheresourcestoimplementoutpatientparenteral

antibiotictherapy(OPAT).Thesehavethepotentialtodecreasethehealthcare

systemburden,lowercomplicationratesfromrepeatintravenousdoses,and

provideaddedconvenienceforpatients.22,23

PracticeguidelinespublishedbytheInfectiousDiseaseSocietyofAmerica(IDSA)

arebasedonexpertconsensus,againhighlightingthelackofpublisheddatato

determineanoptimalmanagementstrategy.4IntheUnitedKingdom,theEron

classificationsystemwasdevelopedbyanexpertpanelandhasbeenincorporated

intotheClinicalResourceEfficiencySupportTeam(CREST)guidelines.19,24An

alternativeDundeeclassificationwasdevelopedbasedonretrospectivedata,and

hasbeenshowntoreducethenumberofpatientstreatedbyparenteralantibiotics

by70%incomparisontotheEronclassification.25,26However,theDundee

classificationschemehasnotbeenincorporatedintoanyconsensusguidelines.

Ithasbeenpostulatedthatonlyasparseinoculumofbacterialpathogencauses

SSTIs,andthattheassociatedsignificantinflammatoryresponseisresponsiblefor

thephysicalfindingsofpain,rednessandindurationthatpatientsexperience.27

12

Basedonthisrationale,Hepburnandcolleaguesreportedthatashorter5-day

courseofantibioticswasequallyefficaciousasalonger10-daycourse.28Current

guidelinesrecommendthatuncomplicatednon-purulentSSTIsshouldbetreated

withanti-streptococcalantibioticsforadurationof5days.4

2.7OralversusIntravenousTherapySelectingtheoptimalrouteofantimicrobialtherapyisakeydecisionpointinthe

managementofSSTIsforemergencyphysicians.Oralantibiotictherapyholds

severaladvantagesovertheparenteralroute,includinglowerriskofcomplications,

decreasedcost,andincreasedpatientconvenienceandcomfort.29,30Oraltherapyis

generallypreferred,inparticularforinfectionsinotherwisewellappearing

immunocompetenthostswithoutsignsofsystemicillness,suchasfeverand

vomiting.

Patientswhoreceiveintravenouslinesareatriskofbothlocalandsystemic

intravenouscatheter-relatedinfections,suchasbacteremia.Thereisalsoariskof

thrombophlebitis–apainfulinflammationoftheveinduetoathrombus.

Intravenousformulationsaremorecostlythantheiroralcounterparts.Thereisalso

theadditionalcostofmedicalsupplies(intravenouslines,tubing,needles,etc.)and

trainedhealthcarepersonneltoadministerthemedication.InareaswhereanOPAT

serviceisnotavailable,thereisalsoasignificantcostincurredduetohospital

admission.Intravenoustherapycarriestheaddedinconvenienceofalackof

13

mobilityforpatientsandariskofhavingtoreturntotheEDduetocomplications

suchasablockedordislodgedperipheralintravenousline.

Theonlyabsoluteindicationsforintravenoustherapyare:1)aninabilitytoswallow

pillsorabsorbthemedicationfromthegastrointestinaltract;and2)theoral

medicationachievespoorconcentrationsinthecirculation(poorbioavailability).

TheformerindicationisuncommonintheEDforpatientswithSSTIs.Furthermore,

mostoralantibioticsusedtotreatSSTIshavegoodtoexcellentbioavailability.30,31

Therearecurrentlynostudiesthathaveexaminedwhyemergencyphysiciansselect

intravenousantibiotictherapy.

Onlytwostudieshavecomparedoralversusintravenoustherapywithinthesame

antibioticclassforthetreatmentofSSTIs.Jorupandcolleaguesconductedasmall

quasi-randomizedtrialcomparingoralversusintravenouspenicillinforerysipelas,

andfoundnobenefitwithintravenoustherapy.32Theoutcomesassessedinthis

studywerefeverduration,hospitallengthofstay,andsickleave.Amorerecent

smallrandomizedtrialbyAboltinsandcolleaguesfoundoralcephalexinwasnon-

inferiortointravenouscefazolinfortheprimaryoutcomeofdaysuntilno

advancementintheareaofcellulitis.33Unfortunately,neithertrialwaspoweredto

detectadifferenceintreatmentfailure,whichisanoutcomethatismoreclinically

importantandimpactful.Todate,therearenopublishedstudiesthathave

demonstratedabenefitofintravenousantibioticsoveroraltherapyforthe

managementofnon-purulentSSTIs.ItstandstoreasonthatifSSTIscanbe

14

adequatelytreatedwithoraltherapy,thenthisisclearlymorepreferable,giventhe

increasedcostandrisksassociatedwithintravenoustherapy.

2.8EDTreatmentSeveraltreatmentpathwaysexistforEDpatientsdiagnosedwithSSTIs(Figure2).

Theemergencyphysicianmustdecideonthemostappropriaterouteofantibiotic

therapy.Themajorityofpatientsaretreatedasoutpatientswithoralantibiotics.If

patientsareeithersystemicallyillorarefelttohavefailedoraltherapy,thenthe

intravenousrouteisselected.Forpatientsinwhichintravenoustherapyischosen,

patientsareeitheradmittedtohospitalormayreceivetheirtreatmentinthe

community,whichisalsoreferredtoasoutpatientparenteralantibiotictherapy

(OPAT).

2.8.1OralAntibioticTherapyThemajorityofpatientsdiagnosedwithnon-purulentSSTIsintheEDaretreatedas

outpatientswithoralantibiotics.Currentguidelinesrecommenda5-daycourseof

oralantibiotictherapy.4Despitethis,surveyedCanadianemergencyphysicians

indicatedthattheymorecommonlyprescribea7or10-daycourseoforal

antibiotics.34Patientsaretypicallyaskedtofollowupwiththeirfamilyphysicianin

48to72hoursforareassessment,butareinstructedtoreturntotheEDif

symptomsworsen(spreadingerythemaorsignsofsystemicillness).Some

emergencyphysiciansmarkthebordersoferythemawithapen,tohelppatients

determineiftheerythemaisspreadingwelloutsideofitsinitialborders.

15

Figure2.EDTreatmentPathwaysforPatientswithSSTIs

2.8.2Parenteral(Intravenous)AntibioticTherapyIntravenousantibioticsarechosenifpatientsaresystemicallyill,haveasevere

clinicalpresentationbasedonthephysician’simpression,orareunabletotolerate

oraltherapy.Thesepatientsareeithertreatedinthehospitalordischargedhometo

receiveOPAT(Figure2).

16

2.8.2.1InpatientParenteralAntibioticTherapyInpatientcaremaybewarrantedforseveralreasons.First,theremaybeconcerns

aboutapatient’sabilitytobecompliantwithoutpatienttherapy.Thismayinclude

socialissuessuchashomelessness,aninabilitytotravel,psychiatricillness,and

injectiondruguse.Second,theclinicianmaydecidethattheinfectionissevere

enoughtorequirecloserobservationandmanagementasaninpatient.Third,the

patientmayalreadyhavecomplicationssuchassepsisand/orunstablevitalsigns

(e.g.hypotension,tachycardia).Fourth,aminorityofpatientsmaybeworsening

clinically,despitealreadyreceivingintravenousantibioticsinthecommunity,and

aredeemedtohavefailedOPAT.

2.8.2.2OutpatientParenteralAntibioticTherapy(OPAT)TherearethreemethodsofintravenousantibioticdeliveryusingtheOPATmodel:

1)intheED;2)viaatrainednurseorself-administrationinthepatient’shome;or

3)viaatrainednurseatanambulatoryclinic.Firstreportedin1974,OPATis

generallydefinedastheadministrationofatleasttwodosesofparenteral

antimicrobialsondifferentdayswithoutinterimhospitalization.35OPATismost

commonlyusedtotreatSSTIs,butmayalsobeutilizedforavarietyofother

infections,including:urinarytractinfections,osteomyelitis,centralnervoussystem

infectionsandendocarditis.36,37ThefollowingdiscussionaboutOPATisinreference

toSSTIsonly.

AppropriatepatientselectioniscriticaltothesuccessofanOPATprogram.

Followingdiagnosis,theemergencyphysicianmustdecidethatthepatientrequires

17

parenteralratherthanoralantibiotictherapy.Therearecurrentlynopublisheddata

regardinghowemergencyphysiciansdecidethatparenteraltherapyisrequiredfor

patientswithSSTIs.Moreover,therearenoevidence-basedguidelinestohelp

cliniciansdecideonwhentheparenteralrouteisappropriate.Inadditiontothe

infectionitself,severalotherfactorsmayplayaroleinphysiciandecision-making,

suchas:comorbidities,triagevitalsignsandsignsofsystemicillness.Socialfactors

mustalsobeconsidered.Forexample,homelesspatientsorthosewithoutameans

oftravelmaybepoorcandidatesforOPAT.ManyOPATprogramsexclude

intravenousdrugusersduetosafetyconcerns,andonlyonesmallobservational

studysuggestsOPATinthispatientpopulationmaybesafeandeffective.38Finally,

useofOPATmustbeacceptabletothepatientandcaregivers.

Followingpatientselection,theemergencyphysicianmustselecttheoptimal

antimicrobialagent.Antibioticswithlonghalf-lives,suchasceftriaxone,maybe

attractivechoicesduetolessfrequentdosingthatwouldbemoreconvenientfor

patients.However,unnecessarilybroad-spectrumantibioticscarrytheriskof

selectingforantimicrobialresistanceduetoalterationsinthegutflorawith

subsequentovergrowthofresistantbacteria.Whensurveyed,76.5%ofCanadian

emergencyphysicianspreferredcefazolinasthefirstchoiceparenteralantibioticfor

SSTIs.34VancomycinorotherantimicrobialswithactivityagainstCA-MRSAshould

beconsideredinpatientswithapurulentSSTIandriskfactors,suchashepatitisCor

substanceabuse.39

18

Patientsareusuallydischargedwithaperipheralintravenouscatheterinplacefor

furtherantimicrobialdoses.Aperipherallyinsertedcentralcatheter(PICC),whichis

amorepermanentintravenouscatheterthatcangenerallyremaininplaceforthe

durationoftreatment,maybeplacedatalaterdateifitisdeterminedthat

prolongedtherapyisrequired.Thefinalstepistodeterminetheappropriatefollow-

upforpatientsreceivingOPAT.Dependingonlocallyavailableresources,patients

canreceivefollow-upwithemergencyphysicianintheED,withafamily

practitioner,orinahospitalclinicsettingwithaninfectiousdiseasespecialist.

2.8.2.2.1ReturntotheEDIncommunitiesthatdonothaveresourcestoadministerintravenousantibiotics

outsideofthehospital,patientsareaskedtoreturntotheEDforsubsequentdoses.

Anemergencyphysicianusuallyperformsaclinicalreassessmentwithin24to72

hourstodetermineiftheinfectionisrespondingtotherapy.Patientswithagood

responsearesteppeddowntooraltherapy,whereasthosewithlittleresponsebut

nosignsofsystemicillnessandnoworseningmaybecontinuedwithintravenous

therapy.ThesepatientswouldbescheduledforanotherEDreassessmentatalater

date.Patientswhoseinfectionsareclinicallyworseningdespiteparenteraltherapy

areadmittedtohospital.

Thisapproachisclearlycumbersomeforpatients,particularlyforthosewhodonot

haveanappropriatemeanstotraveltoandfromaclinicorhospital.Furthermore,

reassessmentatalaterdateisrarelywiththesameemergencyphysician,which

19

makesitdifficulttoprovideanobjectiveclinicalassessmentregardingresponseto

therapy.AnotherapproachinvolvestheuseofEDobservationunitstomonitor

patientswithSSTIsintheEDforupto24hoursbeforemakingafinaldecisionto

admittohospitalormanageasanoutpatient,resultinginhighadmissionrates

rangingfrom29.2to38%.40,41ItisgenerallyacceptedthatSSTIstakeatleast48to

72hoursbeforeimprovementmaybenoted.Itisthereforeunlikelythatashortstay

inanobservationunitwouldimpactthedecisiontoadmitordischargeapatientand

wouldinsteadaddtoEDcrowding.

2.8.2.2.2CommunityCareAccessCentre(CCAC)&FamilyPhysicianFollow-UpInOntario,Canada,theprovisionofhealthcareinthecommunityorathomeis

managedbyLocalHealthIntegrationNetworks(LHINs).TheLHINofaspecific

regionoverseestheCommunityCareAccessCentre(CCAC),whichprovidesnursing

servicesforwoundcareandadministrationofparenteralantibiotics.Apharmacist

isalsoinvolvedinmedicationreview.InOntario,CCACservicesareanintegralpart

oftheOPATmodel.Mostpatientsvisitaclinictoreceiveintravenousdosesbya

trainednurse.Asmallersubsetofpatients(usuallyolderpatientswhocannot

travel)receivesintravenousdoseswithintheirownhome.Incommunitieswith

CCACservicesbutwithouttimelyaccesstoinfectiousdiseasespecialists,patients

areadvisedtoseekaclinicalreassessmentfromtheirfamilyphysician.

20

2.8.2.2.3CCAC&ReturntoaHospitalClinicManyurbanhospitalshavesetupclinicsrunbyinfectiousdiseasespecialiststo

followandmanagepatientsbeingtreatedinthecommunitywithintravenous

antibiotics.Forexample,patientsatTheOttawaHospitalwhoaretreatedwithOPAT

receivefollow-upandreassessmentwithaninfectiousdiseasespecialistatthe

OPATClinic.Thismodelprovidesseveraltheoreticaladvantages:1)decreased

hospitaladmissions;2)increasedpatientconvenience;and3)decreasedEDvisits

forantimicrobialtherapyandreassessments.Whilethisintuitivelyappearstobethe

idealapproach,nopublisheddatacurrentlyexistregardingtheefficacyofOPATin

thecommunitysetting.42

2.8.2.2.4OPATAdverseOutcomesAdministrationofOPATforSSTIsisnotwithoutrisk.Thereisthepotentialof

progressionofinfectiondespiteOPAT,line-relatedcomplicationsoradversedrug

reactionsthatmaywarrantsubsequenthospitaladmission.Petraketal.43

recommendedthatstudiesabouttheefficacyofOPATshouldincorporaterobust

definitionsfortreatmentfailure.

ThereportedhospitaladmissionratefollowingOPATtreatmentfailurerangesfrom

2.6%to8%.43-45However,thesestudiesexaminedavarietyofinfectiousdiseases,

makingitdifficulttodeterminetheclinicalfailurerateforSSTIsinparticular.A

retrospectivestudyreportedareadmissionrateof5.5%forcellulitis,althoughthe

reasonsforhospitalizationwerenotdescribed.46Thereisalackofstudies

concerningnon-purulentSSTIsusingrobustdefinitionsforOPATtreatmentfailure.

21

2.9QuestionsSurroundingOptimalEDManagementofSSTIsBeforedeterminingtheoptimaloutpatientmanagementstrategyforEDpatients

withSSTIs,therearethreekeyquestionsthatarise:

1) Whatisthedefinitionof(i)oralantibiotictreatmentfailureand(ii)OPAT

treatmentfailure?

2) Aretherepredictorsoftreatmentfailurewithoralantibiotics?Inother

words,whichpatientsrequireparenteraltherapy?

3) Whenconsideringtheoptimalrouteoftherapy,whatistheEDphysician’s

rationalewhenselectingparenteraltherapy?

2.10TreatmentFailure

2.10.1TreatmentFailureasanOutcomeintheEDSettingIn2013,theUSCenterforDrugEvaluationandResearch(CDER)releasedguidelines

tohelpresearchersstandardizefutureclinicaltrialsforSSTIs.TheCDERdefines

clinicalresponseasareductioninlesionsizegreaterorequalto20%comparedto

baseline,evaluated48to72hoursaftertherapyisinitiated.12Basedonthis

definitionoftreatment‘improvement’,onecaninferthattreatmentfailurewould

thenbedefinedasareductioninlesionsizelessthan20%at48to72hours.

ThisdefinitionoftreatmentfailureisproblematicintheEDsetting.First,the

majorityofEDphysicianswillonlyseepatientswithSSTIsduringasingle

encounter.Second,manypatientsareprescribedoralantibioticsforSSTIsinthe

22

clinicsettingandthenpresenttotheED.Theemergencyphysicianisleftwiththe

dilemmaofwhethertheircurrentpresentationtrulyreflectsatreatmentfailure.The

emergencyphysicianalsooftendoesnothaveanyobjectivedata(photographsor

medicalrecords)toreasonablyestimatethelesionsizeandseveritywhenoral

antibioticswereinitiated.

AnumberofstudiesintheEDsettinghavereportedtreatmentfailureasan

outcome.Murrayandcolleaguesdefinedtreatmentfailureasanyofthefollowing:

specialistconsultation,hospitaladmission,surgicalprocedure(e.g.incisionand

drainage),an‘upgrade’fromoraltoparenteraltherapy,orachangeinclassof

intravenousantibioticsduetolackofclinicalresponse.47Thisdefinitionwasbased

onexpertopinionandtheauthors’ownobservationoftreatmentpatternsintheir

EmergencyDepartment.A2011EDstudyofpediatricpatientsthatdefined

treatmentfailureusedasimilardefinition,onceagainbasedonexpertopinion.48

Petersonandcoworkersdefinedtreatmentfailureassubsequenthospitalization,a

changeinclassofantibiotic,oraswitchfromoraltoparenteraltherapy.49Patients

requiringsubsequentincisionanddrainagewereexcludedfromthisdefinition.

Ofnote,theauthorsreportingtreatmentfailureasanoutcomeinthe

aforementionedstudiesdidnotspecifyatimeframeatwhichclinicalassessmentfor

treatmentfailureshouldbeundertaken.Thisisclinicallyimportantasassessinga

patienttoosoonaftertheinitiationofantibiotictherapymayresultinan

inappropriatediagnosisoftreatmentfailure.Additionally,itisnotablethattheseED

23

studiesdidnotincorporateachangeinlesionsizeintotheirdefinitionsoftreatment

failure,asthisisimpracticalintheEDsetting.Arecentsystematicreviewidentified

19randomizedcontrolledtrialswithreportedtreatmentfailureratesforcellulitis

rangingfrom6to37%.50Theauthorsspeculatedthatthishighdegreeofvariability

mightbeduetodifficultywithdiagnosisandconfusionwithcellulitismimics.In

addition,variabilityintreatmentfailureratesislikelyalsorelatedtothelackofa

uniformdefinitionfortreatmentfailure.

2.10.2TreatmentFailureDefinitionThereiscurrentlynovalidateddefinitionoftreatmentfailurefollowingantibiotic

therapyforSSTIs.Duetotheuniquenatureofthepatientpopulationandcare

deliveredintheED,itisimportanttouseadefinitiontailoredtothissetting.Usinga

specificpercentageinreductionoflesionsizeisimpracticalforthefollowing

reasons:1)physiciansmaynotdocumentdimensionsoftheinfection;2)manyED

(electronicmedical)recordsdonotallowforphotographstobeuploaded;and3)

patientsmaybealreadyonoralantibiotictherapyinitiatedbyafamilyphysician

anddonotpresentwithdocumentationoflesionsizeatonsetoftherapy.Instead,it

ismorepragmatictouseclinicaljudgmentwhendeterminingtreatmentfailure.It

wouldbeappropriatetoconcludeatreatmentfailurehasoccurredifthepatient

reportssignificantspreadofinfection,therearesystemicsignsofillness(e.g.fever,

tachycardia)oriftheclinicianfeelstheinfectionisseveredespiteanappropriate

durationoftherapy.

24

TheUnitedStatesFoodandDrugAdministrationrecommendsthatclinicalresponse

totreatmentshouldbeassessedat48to72hoursfrominitiatingtherapy.51When

surveyed,amajorityofCanadianemergencyphysiciansselected48hoursasthe

optimaltimeframefordeterminingiftreatmentfailurehadoccurredfollowing

initiationofantibiotictherapy.34Afterreviewoftheliterature40,47-49,52-54and

discussionwithlocalexpertsinemergencymedicineandinfectiousdisease,the

followingtwodefinitionsoftreatmentfailurefornon-purulentSSTIswere

developed:

A. OralAntibioticTreatmentFailure

Oralantibiotictreatmentfailureisdefinedasanyofthefollowingoutcomesthat

occurswithin14daysoftheinitialEDvisitandafteraminimumof48hoursoforal

antibiotictherapy:(i)subsequenthospitaladmissionforanSSTI;(ii)achangein

classoforalantibioticowingtoprogressionofinfectionandnotduetointolerance

orallergy;or(iii)achangeinantibioticroutefromoraltointravenoustherapy.

B. OPATTreatmentFailure

OPATtreatmentfailureisdefinedassubsequenthospitaladmissionaftera

minimumof48hoursofOPATforanyofthefollowing:(i)infectionprogression;(ii)

line-relatedcomplications(e.g.bacteremia,thrombophlebitis,venous

thromboembolism);or(iii)drug-relatedcomplications(e.g.Clostridiumdifficile

colitis).

Adverseeventssuchasoperativedebridement,amputationordeathwerefelttobe

unhelpfulcomponentsoftheaforementioneddefinitionsbecausetheyarerare.

25

2.11PredictorsofTreatmentFailureWithOralAntibioticsIdentifyingpredictorsoffailurewithoralantibioticsisimportantfortworeasons.

First,havingknowledgeofpredictorswithoraltherapywouldhelpemergency

physiciansidentifythesubsetofpatientsthatshouldbestartedonintravenous

therapyattheinitialvisit,limitingtheoveruseofparenteraltherapyand

subsequentlyreducingtheassociatedadverseeventsandcosts.Second,theriskof

sepsisandsubsequenthospitaladmissionsmaybereducedinpatients

inappropriatelystartedonoraltherapy.TherearecurrentlynoEDstudiesreporting

riskfactorsforfailurewithoralantibiotics.Anunderstandingoftheseriskfactors

wouldbecriticaltowarddevelopingevidence-basedcriteriaforpatientsthat

requireintravenoustherapy.

AsystematicreviewofSSTIsmanagedintheEDobservationunitreportedthat

fever,leukocytosisandknownMRSAexposurewerethemostcommonlyreported

riskfactorsfortreatmentfailure.55Petersonandcoworkersidentifiedfiverisk

factorsindependentlyassociatedwithtreatmentfailureforcellulitis:fever,chronic

legulcers,chronicedemaorlymphedema,priorcellulitisinthesamearea,and

cellulitisatawoundsite.49However,thesestudiesdidnotdiscriminatebetween

failureswithoralversusintravenoustherapy.

26

2.12OralVersusTheIntravenousRoute:EmergencyPhysicianRationaleforSelectingIntravenousTherapyTheidealantimicrobialagentinthetreatmentofnon-purulentSSTIswouldpossess

thefollowingcharacteristics:(i)optimalpharmacokineticpropertiesagainstthe

causativeorganism;(ii)minimaladverseeffects;(iii)inexpensive;and(iv)

acceptabletopatientsandhealthcareproviders.Oralantibioticsarepreferredto

intravenousantibioticsinthattheyposefewerrisksofadverseevents,arecheaper,

andarelessinvasivetopatients.Themainpointofcontentionliesintheabilityof

theantimicrobialagenttoachieveadequateconcentrationsinbloodandtissue,for

anadequatedurationtoarrestbacterialgrowth.Therefore,anoralagentwouldbe

clearlypreferableifitcanachievehightissueandserumconcentrationsthatare

comparabletotheirparenteralcounterparts.Forexample,oralcephalexinachieves

excellent(90to100%)bioavailability31andisanexcellentchoicetotreatnon-

purulentSSTIs.

YetED-basedstudieshaveshownthatintravenousantibioticsaremorecommonly

administered47,49andthatantibioticoveruseiscommon.56Mayandcoworkers

reportedthatwhile87%ofsurveyedattendingemergencyphysiciansfeltthat

antibioticswereoverusedintheED,only10%believedtheythemselves

overprescribedantibiotics.57Oneplausibleexplanationisthatpatientspresentingto

theEDmayhavemoresevereinfections.Otherpossibleexplanationsincludea

perceptionthatparenteraltherapyissuperior,aconsiderationofsignificant

comorbidities,orconcernsaboutcompliancewithoralmedications.

27

Surprisingly,thereisscantliteratureaddressingemergencyphysiciandecision-

makingregardingantibioticprescribing57,andtherearenopublishedstudies

describingphysicianrationaleforselectingintravenousantibioticsforinfections.

Identifyingthefactorsphysiciansconsiderwhenoptingforparenteraltherapy

wouldbehighlyusefulforcarryingoutclinicaltrialstotesttheseperceived

indications.

3.RationalePatientscommonlyseekEDcarefornon-purulentSSTIs,whichrequireextensive

healthcareresourcesforbothinpatient(hospitalresources)andoutpatient(CCAC;

physicianfollowup)management.Identifyingpredictorsoftreatmentfailurewith

oralantibioticswouldallowemergencyphysicianstomoreappropriatelyselect

patientsthatrequireintravenoustherapy.Thismayalsohelpreducethenumberof

patientsinappropriatelytreatedwithmorecostlyparenteraltherapythatcarries

theaddedriskofadverseevents.Furthermore,determiningthereasonsbehind

selectingparenteraltherapyisacriticalfirststepinunderstandingemergency

physiciandecision-makingforthemanagementofSSTIs.Finally,despitebeingthe

mostcommonindicationforOPAT,treatmentfailureintheOPATsettinghasnot

beendescribedforSSTIsinparticular.Thus,itisimportanttodescribetheoverall

performanceandsafetyofOPATtherapyforSSTIsintheCanadiansetting.

AccuratediagnosisandappropriatetreatmentofSSTIsarecriticalforpatientsafety

anddecreasinghealthcarecostsandoverallburden.TheepidemiologyofSSTIshas

28

beenwelldocumentedintwodistinctpopulations:hospitalizedpatientsandthose

whovisitphysicians’offices.58-60However,therearenostudiesthathavedescribed

theepidemiologyofadultswithSSTIswhoareseenandmanagedintheEDsetting.

AsamajorityofEDadultpatientsaretreatedasoutpatients,thisconstitutesa

significantevidencegap.Athoroughunderstandingoftheepidemiologyofnon-

purulentSSTIsseenintheEDisanimportantbasisfordevelopingmoreappropriate

evidence-basedrecommendationsforoptimalmanagementofthiscommon

condition.

29

References1. StevensDL,EronLL.Cellulitisandsoft-tissueinfections.AnnInternMed.

2009;150(1):ITC11.

2. WallinTR,HernHG,FrazeeBW.Community-associatedmethicillin-resistantStaphylococcusaureus.EmergMedClinNorthAm.2008;26(2):431-455,ix.

3. QuallsML,MooneyMM,CamargoCA,Jr.,ZucconiT,HooperDC,PallinDJ.Emergencydepartmentvisitratesforabscessversusotherskininfectionsduringtheemergenceofcommunity-associatedmethicillin-resistantStaphylococcusaureus,1997-2007.ClinInfectDis.2012;55(1):103-105.doi:110.1093/cid/cis1342.Epub2012Mar1028.


5. HershAL,ChambersHF,MaselliJH,GonzalesR.Nationaltrendsinambulatoryvisitsandantibioticprescribingforskinandsoft-tissueinfections.ArchInternMed.2008;168(14):1585-1591.

6. PallinDJ,EganDJ,PelletierAJ,EspinolaJA,HooperDC,CamargoCA,Jr.IncreasedUSemergencydepartmentvisitsforskinandsofttissueinfections,andchangesinantibioticchoices,duringtheemergenceofcommunity-associatedmethicillin-resistantStaphylococcusaureus.AnnEmergMed.2008;51(3):291-298.doi:210.1016/j.annemergmed.2007.1012.1004.Epub2008Jan1028.

7. PallinDJ,CamargoCA,Jr.,SchuurJD.Skininfectionsandantibioticstewardship:analysisofemergencydepartmentprescribingpractices,2007-2010.WestJEmergMed.2014;15(3):282-289.doi:210.5811/westjem.2013.5818.18040.Epub12014Jan18046.

8. StenstromR,GrafsteinE,RomneyM,etal.Prevalenceofandriskfactorsformethicillin-resistantStaphylococcusaureusskinandsofttissueinfectioninaCanadianemergencydepartment.[ErratumappearsinCJEM.2009Nov;11(6):570].CJEM,Can.2009;11(5):430-438.

9. InformationCIfH.ASnapshotofHealthCareinCanadaasDemonstratedbyTop10Lists.2011;https://secure.cihi.ca/free_products/Top10ReportEN-Web.pdf.AccessedSeptember25,2017.

30

10. VenkateshAK,DaiY,RossJS,SchuurJD,CappR,KrumholzHM.VariationinUShospitalemergencydepartmentadmissionratesbyclinicalcondition.MedCare.2015;53(3):237-244.doi:210.1097/MLR.0000000000000261.

11. TalanDA,SalhiBA,MoranGJ,etal.Factorsassociatedwithdecisiontohospitalizeemergencydepartmentpatientswithskinandsofttissueinfection.WestJEmergMed.2015;16(1):89-97.

12. FDAGuidelineABSSSI2013.2013.

13. PollackCV,Jr.,AminA,FordWT,Jr.,etal.Acutebacterialskinandskinstructureinfections(ABSSSI):practiceguidelinesformanagementandcaretransitionsintheemergencydepartmentandhospital.JEmergMed.2015;48(4):508-519.

14. KayeKS,PatelDA,StephensJM,KhachatryanA,PatelA,JohnsonK.RisingUnitedStatesHospitalAdmissionsforAcuteBacterialSkinandSkinStructureInfections:RecentTrendsandEconomicImpact.PLoSOne.2015;10(11):e0143276.doi:0143210.0141371/journal.pone.0143276.eCollection0142015.

15. BjornsdottirS,GottfredssonM,ThorisdottirAS,etal.Riskfactorsforacutecellulitisofthelowerlimb:aprospectivecase-controlstudy.ClinInfectDis.2005;41(10):1416-1422.Epub2005Oct1413.

16. DupuyA,BenchikhiH,RoujeauJC,etal.Riskfactorsforerysipelasoftheleg(cellulitis):case-controlstudy.BMJ.1999;318(7198):1591-1594.

17. GundersonCG,MartinelloRA.Asystematicreviewofbacteremiasincellulitisanderysipelas.JInfect.2012;64(2):148-155.doi:110.1016/j.jinf.2011.1011.1004.Epub2011Nov1011.

18. WengQY,RaffAB,CohenJM,etal.CostsandConsequencesAssociatedWithMisdiagnosedLowerExtremityCellulitis.JAMADermatol.2016.


20. ConsensusDocumentontheManagementofCellulitisinLymphoedema.BritishLymphologySociety.2016.

21. KwakYG,ChoiSH,KimT,etal.ClinicalGuidelinesfortheAntibioticTreatmentforCommunity-AcquiredSkinandSoftTissueInfection.InfectChemother.2017;49(4):301-325.doi:310.3947/ic.2017.3949.3944.3301.

31

22. KilburnSA,FeatherstoneP,HigginsB,BrindleR.Interventionsforcellulitisanderysipelas.CochraneDatabaseSystRev.2010(6):CD004299.doi:004210.001002/14651858.CD14004299.pub14651852.

23. CoreyGR,KablerH,MehraP,etal.Single-doseoritavancininthetreatmentofacutebacterialskininfections.NEnglJMed.2014;370(23):2180-2190.doi:2110.1056/NEJMoa1310422.

24. BoucherHW,WilcoxM,TalbotGH,PuttaguntaS,DasAF,DunneMW.Once-weeklydalbavancinversusdailyconventionaltherapyforskininfection.NEnglJMed.2014;370(23):2169-2179.doi:2110.1056/NEJMoa1310480.

25. EronLJ,LipskyBA,LowDE,NathwaniD,TiceAD,VolturoGA.Managingskinandsofttissueinfections:expertpanelrecommendationsonkeydecisionpoints.JAntimicrobChemother.2003;52(Suppl1):i3-17.

26. MarwickC,BroomhallJ,McCowanC,etal.Severityassessmentofskinandsofttissueinfections:cohortstudyofmanagementandoutcomesforhospitalizedpatients.JAntimicrobChemother.2011;66(2):387-397.doi:310.1093/jac/dkq1362.Epub2010Oct1095.

27. PhoenixG,DasS,JoshiM.Diagnosisandmanagementofcellulitis.BMJ.2012;345:e4955.(doi):10.1136/bmj.e4955.

28. SachsMK.Cutaneouscellulitis.ArchDermatol.1991;127(4):493-496.

29. HepburnMJ,DooleyDP,SkidmorePJ,EllisMW,StarnesWF,HasewinkleWC.Comparisonofshort-course(5days)andstandard(10days)treatmentforuncomplicatedcellulitis.ArchInternMed.2004;164(15):1669-1674.

30. LiHK,AgweyuA,EnglishM,BejonP.Anunsupportedpreferenceforintravenousantibiotics.PLoSMed.2015;12(5):e1001825.doi:1001810.1001371/journal.pmed.1001825.eCollection1002015May.

31. CyriacJM,JamesE.Switchoverfromintravenoustooraltherapy:Aconciseoverview.JPharmacolPharmacother.2014;5(2):83-87.doi:10.4103/0976-4500X.130042.

32. MacGregorRR,GrazianiAL.Oraladministrationofantibiotics:arationalalternativetotheparenteralroute.ClinInfectDis.1997;24(3):457-467.

33. Jorup-RonstromC,BrittonS,GavlevikA,GunnarssonK,RedmanAC.Thecourse,costsandcomplicationsoforalversusintravenouspenicillintherapyoferysipelas.Infection.1984;12(6):390-394.

34. AboltinsCA,HutchinsonAF,SinnappuRN,etal.Oralversusparenteralantimicrobialsforthetreatmentofcellulitis:arandomizednon-inferiority

32

trial.JAntimicrobChemother.2015;70(2):581-586.doi:510.1093/jac/dku1397.Epub2014Oct1021.

35. YadavK,GatienM,Corrales-MedinaV,StiellI.Antimicrobialtreatmentdecisionfornon-purulentskinandsofttissueinfectionsintheemergencydepartment.CJEM.2017;19(3):175-180.doi:110.1017/cem.2016.1347.Epub2016Aug1017.

36. TiceAD,RehmSJ,DalovisioJR,etal.Practiceguidelinesforoutpatientparenteralantimicrobialtherapy.IDSAguidelines.ClinInfectDis.2004;38(12):1651-1672.Epub2004May1626.

37. SeatonRA,BarrDA.Outpatientparenteralantibiotictherapy:principlesandpractice.EurJInternMed.2013;24(7):617-623.doi:610.1016/j.ejim.2013.1003.1014.Epub2013Apr1018.

38. GardiolC,VoumardR,CochetC,deValliereS.Settingupanoutpatientparenteralantimicrobialtherapy(OPAT)unitinSwitzerland:reviewofthefirst18monthsofactivity.EurJClinMicrobiolInfectDis.2016;35(5):839-845.doi:810.1007/s10096-10016-12606-z.Epub12016Feb10017.

39. HoJ,ArchuletaS,SulaimanZ,FisherD.Safeandsuccessfultreatmentofintravenousdruguserswithaperipherallyinsertedcentralcatheterinanoutpatientparenteralantibiotictreatmentservice.JAntimicrobChemother.2010;65(12):2641-2644.doi:2610.1093/jac/dkq2355.Epub2010Sep2623.

40. VayalumkalJV,SuhKN,ToyeB,RamotarK,SaginurR,RothVR.Skinandsofttissueinfectionscausedbymethicillin-resistantStaphylococcusaureus(MRSA):anafflictionoftheunderclass.CJEM.2012;14(6):335-343.

41. VolzKA,CanhamL,KaplanE,SanchezLD,ShapiroNI,GrossmanSA.IdentifyingpatientswithcellulitiswhoarelikelytorequireinpatientadmissionafterastayinanEDobservationunit.AmJEmergMed.2013;31(2):360-364.

42. MayL,MullinsP,PinesJ.DemographicandtreatmentpatternsforinfectionsinambulatorysettingsintheUnitedStates,2006-2010.AcadEmergMed.2014;21(1):17-24.

43. ChapmanAL.Outpatientparenteralantimicrobialtherapy.BMJ.2013;346:f1585.(doi):10.1136/bmj.f1585.

44. PetrakRM,SkorodinNC,FliegelmanRM,HinesDW,ChundiVV,HartingBP.ValueandClinicalImpactofanInfectiousDisease-SupervisedOutpatientParenteralAntibioticTherapyProgram.OpenForumInfectDis.2016;3(4):ofw193.eCollection2016Oct.

33

45. ChapmanAL,DixonS,AndrewsD,LilliePJ,BazazR,PatchettJD.Clinicalefficacyandcost-effectivenessofoutpatientparenteralantibiotictherapy(OPAT):aUKperspective.JAntimicrobChemother.2009;64(6):1316-1324.doi:1310.1093/jac/dkp1343.Epub2009Sep1319.

46. Hoffman-TerryML,FraimowHS,FoxTR,SwiftBG,WolfJE.Adverseeffectsofoutpatientparenteralantibiotictherapy.AmJMed.1999;106(1):44-49.

47. ZhangJ,MooreE,BousfieldR.OPATforcellulitis:itsbenefitsandthefactorsthatpredisposetolongertreatment.EurJClinMicrobiolInfectDis.2016;35(6):1013-1015.doi:1010.1007/s10096-10016-12631-y.Epub12016Apr10015.

48. MurrayH,StiellI,WellsG.Treatmentfailureinemergencydepartmentpatientswithcellulitis.CJEM.2005;7(4):228-234.

49. MistryRD,ScottHF,ZaoutisTE,AlpernER.Emergencydepartmenttreatmentfailuresforskininfectionsintheeraofcommunity-acquiredmethicillin-resistantStaphylococcusaureus.PediatrEmergCare.2011;27(1):21-26.doi:10.1097/PEC.1090b1013e318203ca318201c.

50. PetersonD,McLeodS,WoolfreyK,McRaeA.Predictorsoffailureofempiricoutpatientantibiotictherapyinemergencydepartmentpatientswithuncomplicatedcellulitis.AcadEmergMed.2014;21(5):526-531.

51. ObaitanI,DwyerR,LipworthAD,etal.Failureofantibioticsincellulitistrials:asystematicreviewandmeta-analysis.AmJEmergMed.2016;34(8):1645-1652.doi:1610.1016/j.ajem.2016.1605.1064.Epub2016May1626.

52. AdminstrationFaD.GuidanceforIndustry:AcuteBacterialSkinandSkinStructureInfections:DevelopingDrugsforTreatment.

53. AminAN,CerceoEA,DeitelzweigSB,PileJC,RosenbergDJ,ShermanBM.Hospitalistperspectiveonthetreatmentofskinandsofttissueinfections.MayoClinProc.2014;89(10):1436-1451.doi:1410.1016/j.mayocp.2014.1404.1018.Epub2014Jun1425.

54. JenkinsTC,KnepperBC,McCollisterBD,etal.Failureofoutpatientantibioticsamongpatientshospitalizedforacutebacterialskininfections:Whatistheclinicalrelevance?AmJEmergMed.2016;34(6):957-962.doi:910.1016/j.ajem.2016.1002.1013.Epub2016Feb1012.

55. PallinDJ,BinderWD,AllenMB,etal.Clinicaltrial:comparativeeffectivenessofcephalexinplustrimethoprim-sulfamethoxazoleversuscephalexinalonefortreatmentofuncomplicatedcellulitis:arandomizedcontrolledtrial.ClinInfectDis.2013;56(12):1754-1762.doi:1710.1093/cid/cit1122.Epub2013Mar1751.

34

56. AbetzJW,AdamsNG,MitraB.Skinandsofttissueinfectionmanagementfailureintheemergencydepartmentobservationunit:asystematicreview.EmergMedJ.2016:2016-205950.

57. MayL,HarterK,YadavK,etal.Practicepatternsandmanagementstrategiesforpurulentskinandsoft-tissueinfectionsinanurbanacademicED.AmJEmergMed.2012;30(2):302-310.doi:310.1016/j.ajem.2010.1011.1033.Epub2011Jan1028.

58. MayL,GudgerG,ArmstrongP,etal.Multisiteexplorationofclinicaldecisionmakingforantibioticusebyemergencymedicineprovidersusingquantitativeandqualitativemethods.InfectControlHospEpidemiol.2014;35(9):1114-1125.doi:1110.1086/677637.Epub672014Jul677623.

59. Perello-AlzamoraMR,Santos-DuranJC,Sanchez-BarbaM,CanuetoJ,MarcosM,UnamunoP.Clinicalandepidemiologicalcharacteristicsofadultpatientshospitalizedforerysipelasandcellulitis.EurJClinMicrobiolInfectDis.2012;31(9):2147-2152.

60. ZervosMJ,FreemanK,VoL,etal.Epidemiologyandoutcomesofcomplicatedskinandsofttissueinfectionsinhospitalizedpatients.JClinMicrobiol.2012;50(2):238-245.

61. PallinDJ,EspinolaJA,LeungDY,HooperDC,CamargoCA,Jr.EpidemiologyofdermatitisandskininfectionsinUnitedStatesphysicians'offices,1993-2005.ClinInfectDis.2009;49(6):901-907.doi:910.1086/605434.

35

ChapterThree:PredictorsofOralAntibioticFailureforNon-PurulentSkinandSoftTissueInfectionsintheEmergencyDepartment

ChapterOverviewThefollowingisamanuscriptpreparedforpublicationbasedonahealthrecords

review.Theobjectivesofthishealthrecordsreviewwere:1)todescribethe

epidemiologyofadultpatientswithnon-purulentskinandsofttissueinfectionswho

presenttotheemergencydepartment;and2)toidentifyriskfactorsassociatedwith

oralantibiotictreatmentfailure.

InAppendixAacopyoftheapprovalletterfromtheOttawaHealthScience

NetworkResearchEthicsBoardisprovided.

InAppendixBadetaileddescriptionisprovidedofthemethodsusedtodevelopa

multivariablelogisticregressionmodelforpredictorsoforalantibiotictreatment

failure.

Dr.KrishanYadavisthefirstauthorofthismanuscriptandwasresponsibleforthe

studydevelopment,datacollection,monitoringofdataabstraction,statistical

analysisandwritingofthemanuscript.Thismanuscriptwasco-authoredbyDr.Ian

Stiell,Dr.KathrynSuhandDr.GeorgeWells.JordanBernickprovidedvaluableinput

regardingdataanalysis.Dr.DebraEaglesandDr.Venkatesh

Thiruganasambandamoorthyprovidedvaluablefeedbackthroughouttheprocess.

Mr.JohnMacisaacandMr.DarmynRitchiewereinvolvedwithdataabstraction.

36

PredictorsofOralAntibioticFailureforNon-PurulentSkinandSoftTissue

InfectionsintheEmergencyDepartment

KrishanYadav1,KathrynSuh2,DebraEagles3,JohnMacIsaac4,DarmynRitchie4,

JordanBernick5,VenkateshThiruganasambandamoorthy3,GeorgeWells5,6,IanG

Stiell3

1DepartmentofEmergencyMedicine,UniversityofOttawa

2DepartmentofMedicine,DivisionofInfectiousDiseases,UniversityofOttawa

3DepartmentofEmergencyMedicine,TheOttawaHospitalResearchInstitute,

UniversityofOttawa

4DepartmentofUndergraduateMedicine,UniversityofOttawa

5CardiovascularResearchMethodsCentre,UniversityofOttawaHeartInstitute

6DepartmentofEpidemiologyandCommunityMedicine,UniversityofOttawa

Correspondenceto:KrishanYadav

Email:[email protected]

Date:January30,2018

WordCount:3248

Acknowledgments:Theauthorswouldliketothankthefollowingindividualsfor

theirassistanceinthisstudy:My-LinhTran,SherylDomingo,AngelaMarcantonio

andCatherineClement.ThisstudywasfundedbyagrantfromtheDepartmentof

EmergencyMedicine,UniversityofOttawa,Ontario,Canada.

37

Abstract

Introduction

Currentguidelinerecommendationsforoptimalmanagementofnon-purulentskin

andsofttissueinfections(SSTIs)arebasedonexpertconsensus,ratherthanon

evidence.Thereiscurrentlyalackofevidencetoguideemergencyphysicianson

whentoselectoralversusintravenousantibiotictherapy.

Methods

Weperformedahealthrecordsreviewofadults(age≥18years)withnon-purulent

SSTIstreatedattwotertiarycareemergencydepartments(EDs).Patientswere

excludediftheyhadapurulentinfectionorinfectedulcerswithoutsurrounding

cellulitis.Multivariablelogisticregressionwasusedtoidentifypredictors

independentlyassociatedwithoralantibiotictreatmentfailureafteraminimumof

48hoursoforaltherapy.

Results

Weenrolled500patients(meanage64years,279male(55.8%)and126(25.2%)

withdiabetes).Thehospitaladmissionratewas29.6%.Of288patientswhohad

receivedaminimumof48hoursoforalantibiotics,therewere85oralantibiotic

treatmentfailures(29.5%).Tachypneaattriage(oddsratio[OR]=6.31;95%CI

1.80,22.08),chroniculcers(OR=4.90;95%CI1.68,14.27),historyofMRSA

colonizationorinfection(OR=4.83,95%;CI1.51,15.44),andcellulitisinthepast

38

12months(OR=2.23,95%CI=1.01,4.96)wereindependentlyassociatedwithoral

antibiotictreatmentfailure.

Conclusion

Thisisthefirststudytoevaluatepotentialpredictorsoforalantibiotictreatment

failurefornon-purulentSSTIsintheED.Weobservedahighhospitaladmissionrate

andpracticevariabilityregardingantimicrobialagentandroute.Tachypneaat

triage,chroniculcers,historyofMRSAcolonizationorinfectionandcellulitiswithin

thepastyearwereindependentlyassociatedwithoralantibiotictreatmentfailure.

Emergencyphysiciansshouldconsidertheseriskfactorswhendecidingonoral

versusintravenousantimicrobialtherapyfornon-purulentSSTIsbeingmanagedas

outpatients.

39

IntroductionUncomplicated,non-purulentskinandsofttissueinfections(SSTIs)describe

bacterialinfectionsofthesuperficialepidermisanddermis(erysipelas)ordeeper

dermisandsubcutaneoustissue(cellulitis)inwhichpatientsexperienceredness,

painandindurationoftheinvolvedskin.Non-purulentandpurulentSSTIsarea

commonclinicalproblem,accountingforupto3%ofallemergencydepartment

(ED)visitsintheUnitedStates,translatingto3.4millionvisits.1,2AlthoughCanadian

dataarelacking,asingleVancouverEDdiagnosed2234patientswithaSSTI,

representing2%ofallEDvisits.3Oncethediagnosisofanon-purulentSSTIismade,

theemergencyphysicianmustselecttheappropriateantibioticagent,dose,

durationandroute(oralorintravenous).

Duetoalackofhighqualityevidence,empirictreatmentguidelinesarebasedon

expertopinion.4-6Selectingtheappropriateantibioticrouteforoutpatient

managementisakeydecisionpoint.Oraltherapyholdsseveraladvantagesoverthe

parenteralroute,including:lowerriskofcomplications,decreasedcost,increased

patientconvenienceandcomfort.7-9Intravenoustherapymaybeselectedifapatient

hasfailedoraltherapy,issystemicallyunwellorhasasevereinfectionbasedonthe

clinician’simpression.Themainadvantageoftheintravenousrouteisoptimizing

bioavailability,whichisespeciallyusefulinpatientswithswallowingdifficultyora

gastrointestinalmalabsorptionsyndrome.Theintravenousrouteiscostlier,less

convenientforpatientsandhasanaddedriskofadverseevents.Therearecurrently

nostudiesthathaveaimedtoidentifypredictorsassociatedwithtreatmentfailure

40

withoralantibiotictherapy.Identificationofsuchpredictorswouldallow

emergencyphysicianstobetterselectpatientsthatrequireintravenousantibiotics.

Thismayalsohelpreducethenumberofpatientsinappropriatelytreatedwith

parenteraltherapy.

Theprimaryobjectiveofthisstudywastoidentifyriskfactorsassociatedwithoral

antibiotictreatmentfailurefornon-purulentSSTIs.Asecondaryobjectivewasto

describetheepidemiologyofadultswithnon-purulentSSTIspresentingtotheED.

Methods

StudyDesignandSettingWeperformedahealthrecordsreviewofconsecutiveadultpatientspresentingto

theEDwithdiagnosisandmanagementofanon-purulentSSTI.Thestudy

populationwasenrolledfromtheOttawaHospitalCivicandGeneralEDs,bothbeing

tertiarycareadultEDswithacombined170,000patientvisitsannually.TheOttawa

HealthScienceNetworkResearchEthicsBoardapprovedtheprotocolwithoutthe

needforinformedconsent.

PopulationWeenrolledaconsecutivesampleofpatientsmeetingeligibilitycriteriathat

presentedtotheEDoveraseven-monthperiod(January1toJuly31,2016).Eligible

patientswereadults(age≥18years)presentingtotheEDanddiagnosedwitha

non-purulentSSTIthatwastreatedwitheitheroralorintravenousantibiotics.

Patientswereconsideredeligibleiftheywerealreadytakingantibioticsattheindex

EDvisit.Weexcludedpatientsforthefollowingreasons:(i)patientspresentingfora

41

follow-upvisit(i.e.nottheindexEDvisitforthisclinicalproblem);(ii)age<18

years;(iii)adiagnosisofapurulentskinabscesswhereanincisionanddrainage

procedurewasperformed;(iv)infectedulcerswithoutsurroundingcellulitisor

erysipelas;or(v)necrotizinginfections.

StudyProtocolandDataAbstractionInordertominimizebias,wetookspecificstepswithrespecttocaseselection,

abstractortraining,definitionofvariables,useofastandardizedcaserecordform,

regularmeetingsandoversightofabstractorsinaccordancewithaccepted

methodologyforchartreviews.10-13WeidentifiedeligiblecasesbyInternational

ClassificationofDiseases,10threvision,Canada(ICD-10-CA)diagnosiscodesofL03*

(cellulitis,unspecified)andA46(erysipelas).Relevantpatientdatawereobtained

fromtheelectronichealthrecord(physicianandnursingnotes,outpatient

parenteralantibiotictherapy(OPAT)clinicrecords).

Theprincipalinvestigator(KY)trainedtwomedicalstudents(JM,DR)ontheuseof

theelectronichealthrecordssystem.Allvariablesandtheprimaryoutcomeof

interestwereexplicitlydefined(seesupplementaryappendix)apriori.Weuseda

standardizedcaserecordform(seesupplementaryappendix)toabstractdata.The

caserecordformwaspilotedtoremoveambiguousitemsandensurethedata

collectioninstrumentwasrobust.Thedataabstractorsheldregularmonthly

meetingstoresolveanydisagreementsbyconsensus.Theprincipalinvestigator

monitoredtheperformanceofthedataabstractorsbyreviewing25%ofthesample.

42

Cohen’skappastatisticwasusedtoassessinterobserveragreementforincluded

subjectsandtheprimaryoutcome.

OutcomeMeasuresTheprimaryoutcomewastreatmentfailurewithoralantibiotics.Thereiscurrently

novalidateddefinitionoftreatmentfailureintheliterature.TheUnitedStatesFood

andDrugAdministrationrecommendsthatclinicalresponsetotreatmentshouldbe

assessedat48to72hoursfrominitiatingtherapy.14Whensurveyed,amajorityof

Canadianemergencyphysiciansselected48hoursastheoptimaltimeframefor

determiningiftreatmentfailurehadoccurredfollowinginitiationofantibiotic

therapy.15Afterreviewoftheliterature16-22anddiscussionwithlocalexpertsin

emergencymedicineandinfectiousdisease,wedevisedatreatmentfailure

definition.Treatmentfailurewithoralantibioticswasdefinedasanyofthe

followingoutcomesoccurringafteraminimumof48hoursoforalantibioticsandat

nolaterthan14daysfromtheindexEDvisit:(i)subsequenthospitaladmissionfor

aSSTI;(ii)achangeinclassoforalantibioticowingtoprogressionofinfectionand

notduetointoleranceorallergy;or(iii)achangeinantibioticroutefromoralto

intravenoustherapyowingtoprogressionofinfectionandnotduetointoleranceor

allergy.

Asecondaryoutcomeofinterestwastreatmentfailurewithintravenousantibiotics.

Thiswasdefinedasanyofthefollowingoutcomesafteraminimumof48hoursof

intravenousantibioticsandatnolaterthan14daysfromtheindexEDvisit:(i)

43

subsequenthospitaladmissionforaSSTI;or(ii)achangeinclassofintravenous

antibioticowingtoprogressionofinfectionandnotduetointoleranceorallergy.

Thefollowingbaselinedemographicsandclinicaldatawereabstracted:patientage

andgender;comorbidities;EDtriagevitalsigns;andinfectioncharacteristics.We

anticipatedthataccurateinfectiondimensions(lengthandwidth)mightnotbe

consistentlyrecordedonpatientcharts.Inordertoestimatethepercentbody

surfaceareaofaffectedskin,themodifiedLund-Browderchartwasused.23,24ED

treatmentvariableswereabstractedasfollows:antibiotictreatmentapproach;and

settingforsubsequentintravenousantibiotics(ifchosen).Adverseoutcomes

includedantibioticeventsandintravenouscatheter-relatedevents.

StatisticalAnalysisTheprevalenceofnon-purulentSSTIsintheEDpopulation,theproportionof

patientswhoreceivedoralversusintravenoustherapy,andthepatientswhohada

treatmentfailurewerecalculated.Continuousdataarepresentedasmeanswith

standarddeviationsormedianswithaninterquartilerange(IQR,Q1–Q3)for

normallyandnon-normallydistributeddata,respectively.Categoricaldataare

presentedasproportionswith95%confidenceintervals.

Weemployedunivariateanalysestoexamineallclinicalvariableshypothesizedto

beriskfactorsfortreatmentfailurewithoralantibiotictherapy(seesupplementary

appendix).Variableswithp-valuesof0.10orlesswereconsideredformultivariable

analysis.Abackwardsselectionprocedurewasusedtoobtainamultivariable

44

logisticregressionmodeltodetermineclinicalpredictorsindependentlyassociated

withtheprimaryoutcomeoftreatmentfailurewithoralantibiotics.TheHosmer-

Lemeshowstatisticwasusedtoassessmodelfit.SAS(version9.4,SASInstitute,

CaryNC)wasusedfordescriptivestatistics,univariateandmultivariablelogistic

regressionanalysis.

SampleSizeandFeasibilityPreviousstudieshavesuggestedthataminimumof10eventspervariableis

requiredtoavoidbiasedestimateswhendevelopingmultivariableprediction

models.25-27Weestimatedthatnomorethanfivepredictorvariableswouldbe

includedinamodeltopredicttreatmentfailurewithoralantibiotics.Basedonthe

10eventspervariableworkingrule,aminimumof50treatmentfailureswouldbe

requiredtodeveloparobustmodel.Treatmentfailureratesreportedinthe

literaturerangefrom6to37%.28PreviouslypublishedstudiesinCanadianEDshave

indicatedthattheapproximatetreatmentfailurerateofSSTIswithantibiotics

rangesfrom18.7to20.5%.16,17Assumingaconservativeestimateofan18%

treatmentfailurerate,270patientswouldberequiredtoobtainaminimumof50

oralantibiotictreatmentfailures.Weestimatedthatupto40%ofpatientsmightbe

treatedwithintravenoustherapy.Therefore,wedeterminedanoverallsamplesize

of500patientswouldensurethatwesurpassedtheminimumrequirednumberof

patientstreatedwithoralantibiotics.FromJanuary1,2014untilDecember31,

2014,therewere2286uniquevisitstoeithertheCivicorGeneralcampusofThe

OttawaHospitalwithanICD-10diagnosisofcellulitisorerysipelas.Evenafter

45

accountingforpurulentSSTIs,ahealthrecordsreviewofpatientswithnon-purulent

SSTIsoversevenmonthswouldyieldwellovertheminimumrequiredsamplesize.

ResultsOvertheseven-monthstudyperiod,666caseswerescreenedforeligibilityand500

patientsmettheinclusioncriteria(Figure1).Thekappastatisticforincludedcases

betweentheprimaryinvestigator(KY)andeachabstractor(JMandDR)was0.96

(95%CI0.93,0.99)and0.91(95%CI0.86,0.97),respectively.Thekappastatistic

fortheprimaryoutcomeoforalantibiotictreatmentfailurewas0.94(95%CI0.90,

0.98).Ofthe500enrolledpatients,126(25.2%)haddiabetesand87(17.4%)hada

historyofcellulitisintheprior12months(Table1).Themostcommonlocationof

infectionwastheleg(54.2%)andmostinfections(80.2%)wereestimatedtobe

<5%totalbodysurfacearea(Table2).

Ofthetotalcohort,354patients(70.8%)receivedanintravenousantibioticinthe

ED,with148patients(29.6%)admittedtohospitalforfurtherparenteraltherapy.

Themostcommonoralagentusedwascephalexinandthemostcommonparenteral

agentwascefazolin(Table3).Ofpatientsreceivingintravenousantibiotics,20.6%

receivedtwoormoreantibiotics.

Ofthe352patientsthatweremanagedasoutpatients,themajority(61.4%)

receivedsolelyoralantibioticprescriptions(Table4).Asignificantproportionof

46

outpatients(19.9%)receivedanintravenousantibioticdoseintheEDbutwere

dischargedwithanoralantibioticprescription.Ofthe222patientsreceivingoral

antibioticprescriptions,cephalexinwasmostcommonlyselected(77.4%).Ofthe

136patientswhowereprescribedoutpatientintravenousantibiotics,99patients

(72.8%)werereferredtotheOPATclinicforassessmentandfollowupwithan

infectiousdiseasespecialist.Aminoritywasaskedtofollowupwiththeirprimary

careproviderorreturntotheEDforfollowup.Cefazolin(68.4%)wasthemost

commonlyprescribedintravenousantibiotic.

Asignificantnumberofoutpatients(40.6%)returnedtotheEDwithin14days

(Table5).Themajorityofpatientsreturnedforscheduledrepeatintravenous

antibiotics,becauseofadelayinarrangingintravenousantibioticsintheoutpatient

setting.Asmallproportionofunscheduledvisits(5.4%)wasforaworsening

infectionthatrequiredhospitaladmission.Therewerefewadverseeventsfor

outpatients:2.8%withadislodgedorblockedperipheralintravenousline;1.7%of

withgastrointestinalsymptomsand0.6%ofwitharashattributedtotheprescribed

antibiotic.

Forthe99patientsreferredtotheOPATclinic,85.8%ofpatientsattendedtheir

appointment(Table6).Emergencyphysiciansdiagnosedcellulitiswithahigh

degreeofaccuracy(96.5%),withonlythreepatientshavinganalternatediagnosis

assignedbytheinfectiousdiseasephysicians.Themediantimetofollowupforthe

47

firstOPATclinicvisitwas4daysandtherewasamedianof2clinicvisits.Patients

receivedamediandurationof7daysofintravenousantibiotics.

Of288patientswhoweretreatedwithatleast48hoursoforalantibiotics,85

patients(29.5%)sufferedanoralantibiotictreatmentfailure(Table7).Treatment

failuresweremanagedasfollows:51patients(60.0%)wereswitchedtooutpatient

intravenousantibiotics;30patients(35.3%)werehospitalizedforintravenous

therapy;and4patients(4.7%)wereswitchedtoadifferentclassoforalantibiotic.

Of212patientstreatedwithatleast48hoursofintravenousantibiotics,12patients

(5.7%)sufferedanintravenousantibiotictreatmentfailure(seethesupplementary

appendix).

Predictorsassociatedwithoralantibiotictreatmentfailureusingmultivariable

logisticregressionareshowninTable8.Tachypneaattriage(oddsratio[OR]=

6.31;95%CI1.80,22.08),chroniculcers(OR=4.90;95%CI1.68,14.27),historyof

MRSAcolonizationorinfection(OR=4.83;95%CI1.51,15.44),andcellulitisinthe

past12months(OR=2.23;95%CI1.01,4.96)werefoundtobeindependently

associatedwithoralantibiotictreatmentfailure.TheHosmer-Lemeshowchi-square

testyieldedap-valueof0.604(χ2=1.853,degreesoffreedom=3)andtheC-

statisticwas0.709.Thisindicatesthatourmodelhasgoodfit.

48

Discussion

InterpretationofResultsThisstudydescribesadultpatientspresentingtotheEDfornon-purulentSSTIs.A

substantialproportionofpatientswasadmittedtohospitalforfurthermanagement.

Weobservedsignificantpracticevariationwithrespecttoselectionofantimicrobial

routeandagent.AnumberofpatientsreceivedasingleintravenousdoseintheED

followedbyoutpatientoraltherapy,despitealackofevidencetosupportthis

approach.Thevariabilityintreatmentapproachreinforcesthelackofagreement

amongstemergencyphysiciansontheoptimalapproachtotherapyforthiscommon

clinicalcondition.

Wefoundanoralantibiotictreatmentfailurerateof29.5%,whichwashigherthan

expected.Murrayetal.16reportedanoralantibiotictreatmentfailurerateof6.8%,

butthiswasasmallsamplesize(2of29patients).Petersonetal.17reportedanoral

antibiotictreatmentfailurerateof21.0%.Asneitherstudyusedastricttimecutoff

intheirdefinitionoftreatmentfailure,somepatientsmayhavebeenclassifiedasan

oralantibiotictreatmentfailureprematurely.Weidentifiedpotentialriskfactorsfor

failurewithoralantibiotics.Tachypneaattriage,chroniculcers,historyofMRSA

colonizationorinfectionandcellulitiswithinthepastyearwereindependently

associatedwithoralantibiotictreatmentfailure.Theseriskfactorsmaybe

consideredaspotentialindicationsforintravenoustherapy.Thehightreatment

failureandhospitaladmissionratesareofconcern.Thesefindingsmayinpart

reflectthelackofevidencetoguideemergencyphysiciansontheoptimalrouteof

antimicrobialtherapy.

49

PreviousStudiesThemostrecentInfectiousDiseaseSocietyofAmericaguidelinessuggest

intravenousantibioticsfor‘moderate’(signsofsystemicillness)or‘severe’(failed

oraltherapy,signsofsystemicillness,clinicalsignsofdeeperinfection,or

immunocompromised)infections.4TheBritishClinicalResourceEfficiencySupport

Team(CREST)guidelinesrecommendoraltherapyin‘ClassI’patients,definedas

havingnosignsofsystemictoxicityandno‘uncontrolled’comorbidities,whichwas

notexplicitlydefined.5Duetoalackofevidence,theseguidelinesarebasedon

expertopinion.AstudybyPetersonetal.identifiedpredictorsoffailurewith

outpatientantibioticsforcellulitis,butdidnotdistinguishbetweenoralversus

intravenousroutes.17ArecentsurveyofCanadianemergencyphysiciansrevealed

that94.4%ofrespondentswouldconsideraclinicaldecisionruletopredictoral

antibiotictreatmentfailure.15Todate,evidenceregardingtheoptimalrouteof

antimicrobialtherapyfornon-purulentSSTIsislacking.

StrengthsThisisthefirststudytoidentifypotentialpredictorsassociatedwithoralantibiotic

treatmentfailurefornon-purulentSSTIs.Therewasexcellentagreementbetween

dataabstractorsforbothinclusionofstudiesandtheprimaryoutcome.Thestudy

findingsmaybetterguideemergencyphysicianstodeterminewhenoralantibiotic

treatmentfailureislikely–andwhentoselectintravenoustherapyattheonsetof

treatment.

LimitationsThishealthrecordsreviewhasseveralpotentiallimitations.First,potentially

clinicallyimportantvariables(infectionsizeandobesity)mayhavebeeninaccurate

50

ornotdocumented.Obtainingaccuratemeasuresofinfectionsizewasnotpossible

asitwasseldomdocumentedinthemedicalrecord.Weinsteadattemptedto

estimateinfectionsizeusingaLund-Browderburnchartasasurrogatefortotal

bodysurfaceareaofaffectedskin.Obesitywasnotconsistentlydocumentedinthe

medicalchartandwasnotvalidated(e.g.bycalculatingthebodymassindex).We

attemptedtomitigatethisbyreviewingallelectronichealthrecordsinthe6months

priortoandaftertheindexvisittoidentifyifthiscomorbiditywasdocumented.

Second,thedataabstractorswerenotblindedtothestudyoutcome.Thisisunlikely

tohaveresultedinsignificantbiasastheprimaryoutcomewasstrictlydefined

usinga48-hourcutoffforconsiderationoftreatmentfailure.Inaddition,therewas

excellentinter-observeragreementfortheprimaryoutcome.Weattemptedto

minimizebiasbytrainingabstractors,holdingregularmeetings,validating25%of

charts,definingvariablesapriori,andusingastandardizedcaserecordformin

accordancewithacceptedmethodologyforchartreviews.10-13

Third,thereisnovalidateddefinitionoforalantibiotictreatmentfailure.Following

areviewoftheliterature14-22,29wedevelopedacompositeendpointdefinitionafter

discussionandconsensusamonglocalexpertsinemergencymedicineand

infectiousdisease.Disadvantagesofusingacompositeendpointincludemisleading

resultsifonecomponentoftheoutcomeheavilydrivestheresult,especiallyifitis

theleastpatient-importantoutcome.However,95.3%ofthetreatmentfailureswere

51

drivenbythetwomostpatientcentredoutcomes:hostpitaladmissionfor

intravenoustherapyorswitchfromoraltointravenousoutpatienttherapy.

Fourth,therewasasmallamountofmissingdata(<2.5%),whichwasassumedtobe

missingcompletelyatrandom.Wehandledthemissingdatausingacompletecase

analysis.Lastly,wewereunabletomeasureadherencetotreatmentduetothe

natureofthestudydesign.

ClinicalImplicationsOurfindingsrevealimportantclinicalimplications,havingdemonstratedsignificant

practicevariabilitywithrespecttoselectionofantimicrobialagentandroute.

PatientswhoreceivedOPAThadamediandurationofantibiotictherapythatwas

longerthantheguideline-recommendedfivedays.4Patientsrequiringintravenous

therapyhavemorecomplicatedinfections,whichmayexplainwhytheyrequireda

longerdurationoftherapy.Thisvariationintreatmentapproachcoupledwitha

highhospitaladmissionrateislikelyduetoalackofevidence-based

recommendationsforoptimaltherapy.Severalriskfactorsassociatedwithoral

antibiotictreatmentfailurewereidentified.Webelievethatsuchfactorsshouldbe

consideredwhendecidingontheoptimalrouteoftherapy.Ultimately,ourfindings

highlightthatfurtherstudiesarecriticaltoimprovetreatmentofthiscommon

clinicalcondition.

52

ResearchImplicationsFutureresearchshouldinvolveaprospectivestudytofurtherassessthesepotential

riskfactorsfortreatmentfailureidentifiedinourstudyandideallyderiveaclinical

decisionruletoguideemergencyphysiciansontheoptimalrouteofantimicrobial

therapy.Furthermore,studiesexaminingrationaleforselectingintravenoustherapy

wouldprovidebetterinsightregardingphysiciandecision-making.

ConclusionsThisisthefirststudytoevaluatepotentialpredictorsoforalantibiotictreatment

failurefornon-purulentSSTIsintheED.Weobservedahighhospitaladmissionrate

andpracticevariabilityregardingantimicrobialagentandroute.Tachypneaat

triage,chroniculcers,historyofMRSAcolonizationorinfectionandcellulitiswithin

thepastyearwereindependentlyassociatedwithoralantibiotictreatmentfailure.

Emergencyphysiciansshouldconsidertheseriskfactorswhendecidingonoral

versusintravenousantimicrobialtherapyforpatientswithnon-purulentSSTIs.

53

References1. PallinDJ,EganDJ,PelletierAJ,EspinolaJA,HooperDC,CamargoCA,Jr.

IncreasedUSemergencydepartmentvisitsforskinandsofttissueinfections,andchangesinantibioticchoices,duringtheemergenceofcommunity-associatedmethicillin-resistantStaphylococcusaureus.AnnEmergMed.2008;51(3):291-298.doi:210.1016/j.annemergmed.2007.1012.1004.Epub2008Jan1028.

2. PallinDJ,CamargoCA,Jr.,SchuurJD.Skininfectionsandantibioticstewardship:analysisofemergencydepartmentprescribingpractices,2007-2010.WestJEmergMed.2014;15(3):282-289.doi:210.5811/westjem.2013.5818.18040.Epub12014Jan18046.

3. StenstromR,GrafsteinE,RomneyM,etal.Prevalenceofandriskfactorsformethicillin-resistantStaphylococcusaureusskinandsofttissueinfectioninaCanadianemergencydepartment.[ErratumappearsinCJEM.2009Nov;11(6):570].CJEM,Can.2009;11(5):430-438.



6. ConsensusDocumentontheManagementofCellulitisinLymphoedema.BritishLymphologySociety.2016.

7. LiHK,AgweyuA,EnglishM,BejonP.Anunsupportedpreferenceforintravenousantibiotics.PLoSMed.2015;12(5):e1001825.doi:1001810.1001371/journal.pmed.1001825.eCollection1002015May.

8. MacGregorRR,GrazianiAL.Oraladministrationofantibiotics:arationalalternativetotheparenteralroute.ClinInfectDis.1997;24(3):457-467.

9. CyriacJM,JamesE.Switchoverfromintravenoustooraltherapy:Aconciseoverview.JPharmacolPharmacother.2014;5(2):83-87.doi:10.4103/0976-4500X.130042.

10. GilbertEH,LowensteinSR,Koziol-McLainJ,BartaDC,SteinerJ.Chartreviewsinemergencymedicineresearch:Wherearethemethods?AnnEmergMed.1996;27(3):305-308.

54

11. BadcockD,KellyAM,KerrD,ReadeT.Thequalityofmedicalrecordreviewstudiesintheinternationalemergencymedicineliterature.AnnEmergMed.2005;45(4):444-447.

12. LowensteinSR.Medicalrecordreviewsinemergencymedicine:theblessingandthecurse.AnnEmergMed.2005;45(4):452-455.

13. KajiAH,SchrigerD,GreenS.Lookingthroughtheretrospectoscope:reducingbiasinemergencymedicinechartreviewstudies.AnnEmergMed.2014;64(3):292-298.doi:210.1016/j.annemergmed.2014.1003.1025.Epub2014Apr1018.









55


23. HettiaratchyS,DziewulskiP.ABCofburns:pathophysiologyandtypesofburns.BMJ.2004;328(7453):1427-1429.

24. WachtelTL,BerryCC,WachtelEE,FrankHA.Theinter-raterreliabilityofestimatingthesizeofburnsfromvariousburnareachartdrawings.Burns.2000;26(2):156-170.

25. ConcatoJ,PeduzziP,HolfordTR,FeinsteinAR.Importanceofeventsperindependentvariableinproportionalhazardsanalysis.I.Background,goals,andgeneralstrategy.JClinEpidemiol.1995;48(12):1495-1501.

26. PeduzziP,ConcatoJ,FeinsteinAR,HolfordTR.Importanceofeventsperindependentvariableinproportionalhazardsregressionanalysis.II.Accuracyandprecisionofregressionestimates.JClinEpidemiol.1995;48(12):1503-1510.

27. PeduzziP,ConcatoJ,KemperE,HolfordTR,FeinsteinAR.Asimulationstudyofthenumberofeventspervariableinlogisticregressionanalysis.JClinEpidemiol.1996;49(12):1373-1379.

28. ObaitanI,DwyerR,LipworthAD,etal.Failureofantibioticsincellulitistrials:asystematicreviewandmeta-analysis.AmJEmergMed.2016;34(8):1645-1652.doi:1610.1016/j.ajem.2016.1605.1064.Epub2016May1626.

29. FDAGuidelineABSSSI2013.2013.

56

FiguresFigure1.FlowDiagramofPatientEligibilityandOutcomes

PatientsScreened

(N=666)

ExcludedPatients,total(N=166)Didnotmeetinclusioncriteria(N=62)

Didnotreceiveantibiotic(N=25)NotdiagnosedwithSSTI(N=37)

Metexclusioncriteria(N=104)

I&Dofabscess(N=40)Follow-upvisit(N=58)Ulcerwithoutcellulitis(N=4)DeclinedConsent(N=2)

EligiblePatients

(N=500)

Home(N=352)(70.4%)

HospitalAdmission(N=148)(29.6%)

57

Tables

Table1.BaselineCharacteristicsofAdultswithNonpurulentSkinandSoftTissueInfectionsseenintheEmergencyDepartment(N=500)

Variable N=500Age(years),mean± SDRange

Male(%)HospitalSite(%)

TOHCivicCampusTOHGeneralCampus

ComorbiditiesDiabetesmellitusCellulitisinpast12monthsCoronaryarterydiseaseCongestiveheartfailureHistoryofMRSAinfectionorcolonizationPeripheralvasculardiseaseLiverdiseaseChronickidneydiseaseActivecancerLymphedemaObesityInjectiondruguseOrgantransplantrecipient

TakingantibioticsatthetimeofEDpresentationOralIV

64±1918–98

279(55.8)

278(55.6)222(44.4)

126(25.2)87(17.4)58(11.6)48(9.6)43(8.6)40(8.0)37(7.4)35(7.0)34(6.8)33(6.6)27(5.4)14(2.8)4(0.8)

85(17.0)13(2.6)

SD=standarddeviation;TOH=TheOttawaHospital;MRSA=methicillinresistantStaphylococcusaureus;ED=emergencydepartment;IV=intravenous

58

Table2.PresentingPatientandInfectionCharacteristics(N=500)

Variable N=500TriageVitalSignsTemperature,°C(mean±SD)HeartRate,beats/min(mean±SD)BloodPressure,mmHg(mean±SD)RespiratoryRate,breaths/min(median,IQR)OxygenSaturation,%(median,IQR)

InfectionLocation(%)LegFootArmHandFaceTorsoGroin

InfectionCharacteristics(%)ChroniclegulcersSurgicalsiteinfectionBiteSizeTBSA<5%TBSA5–10%TBSA>10%

LaboratoryTestsWhitebloodcellcountordered(%)Whitebloodcellcount,×109/L(median,IQR)

36.6±0.987±19136±2418,16–1897,96–98

271(54.2)85(17.0)51(10.2)37(7.4)29(5.8)22(4.4)5(1.0)

56(11.2)30(6.0)12(2.4)

401(80.2)97(19.4)2(0.4)

378(75.6)9.2,7–13

SD=standarddeviation;IQR=interquartilerange;TBSA=totalbodysurfacearea

59

Table3.EmergencyDepartmentTreatment(N=500)

AntibioticSelection Numberofpatients,N=500N(%)

IVantibioticsonlyOralantibioticsonlyOralandIVantibioticsNumberofIVAntibiotics(N=354)OneTwoThree

OralAntibioticsCephalexinClindamycinAmoxicillin-ClavulanateCiprofloxacinTrimethoprim-SulfamethoxazoleDoxycyclineAmoxicillin

IVAntibioticsCefazolinCeftriaxoneVancomycinPiperacillin-TazobactamClindamycinCiprofloxacinMeropenem

339(67.8)146(29.2)15(3.0)

281(79.4)64(18.1)9(2.5)

121(24.2)11(2.2)9(1.8)7(1.4)6(1.2)4(0.8)3(0.6)

202(40.4)80(16.0)59(11.8)46(9.2)40(8.0)3(0.6)3(0.6)

IV=intravenous

60

Table4.AntibioticTreatmentfor352PatientsDischargedfromtheED

OutpatientManagement Numberofpatients,N=352N(%)

PrescribedoralantibioticsPrescribedIVantibioticsPrescribedoralandIVantibioticsOralantibioticinEDandsenthomeonoralantibioticIVantibioticinEDandsenthomeonIVantibiotic*IVantibioticinEDandsenthomeonoralantibioticIntendedlocationforoutpatientIVantibioticsCCACandOPATclinicCCAConlyED

OralantibioticsprescribedCephalexinClindamycinAmoxicillin-ClavulanateTrimethroprim-sulfamethoxazoleCiprofloxacinDoxycyclineAmoxicillin

IVantibioticsprescribedCefazolinCeftriaxoneClindamycinVancomycinMeropenemMultipleIVAntibiotics

216(61.4)130(36.9)6(1.7)

146(41.5)136(38.6)70(19.9)

99(28.1)26(5.6)11(2.2)

222(63.1)172(48.9)19(5.4)13(3.7)7(2.0)5(1.4)5(1.4)1(0.3)

136(38.6)93(26.4)31(8.8)4(1.1)3(0.8)1(0.3)4(1.1)

IV=intravenous,CCAC=communitycareaccesscentre;OPAT=outpatientparenteralantibiotictherapy;ED=emergencydepartment*6patientsweredischargedwithbothintravenousandoralantibiotics

61

Table5.AdverseEventsfor352PatientsDischargedfromtheED

AdverseEvents Numberofpatients,N=352N(%)

ReturntotheEDwithin14DaysReasonforreturnEDvisitRepeatantibioticsForSSTIandnoadmissionUnrelatedmedicalproblemForSSTIandhospitaladmissionDiagnosedwithabscessrequiringI&D

AdversedeviceeventsDislodged/blockedperipheralIVlineOther*

AdverseantibioticeventsNauseaand/orvomitingRashDiarrhea

143(40.6)

60(17.0)39(11.1)21(6.0)19(5.4)4(1.1)

10(2.8)0(0)

4(1.1)2(0.6)2(0.6)

*Other=thrombophlebitis,lineinfectionorbacteremiaED=emergencydepartment;IV=intravenous;SSTI=skinandsofttissueinfection;I&D=incisionanddrainage

62

Table6.OutpatientParenteralAntibioticTherapy(OPAT)ClinicData(N=85)

OPATClinicMetric Numberofpatients,N=85N(%)

Cellulitiswasthecorrectdiagnosis**Alternatediagnoses,totalGoutLymphedemaVenousStasis

Timetofollow-up(days),median,IQRNumberofvisits,median,IQRDurationofIVantibiotics(days),median,IQRDurationofIVandoralantibiotics(days),median,IQR

82(96.5)111

4,3–6

2,1–3

7,7–14

19,14–28

*Atotalof99patientswerereferredtotheOPATclinicfromtheemergencydepartmentOPAT=outpatientparenteralantibiotictherapy;IV=intravenous;IQR=interquartilerange

63

Table7.TreatmentFailurewithOralAntibiotics(N=85of288PatientsTreatedwithaMinimumof48HoursofOralTherapy)

OralAntibioticTreatmentFailures Numberofpatients,N=85N(%)

PatientoutcomesSwitchedtooutpatientIVantibioticsHospitalizedforIVantibioticsSwitchedtooutpatientoralantibioticsofdifferentclass

TreatmentfailureoninitialEDvisit*TreatmentfailureonreturnEDvisitwithin14days

51(60.0)30(35.3)4(4.7)

68(80.0)17(20.0)

*Patientwasalreadyon≥48hoursoforalantibiotictherapyattimeofindexEDvisitIV=intravenous;ED=emergencydepartment

64

Table8.PredictorsAssociatedwithOralAntibioticTreatmentFailureUsingMultivariableLogisticRegression(N=288)

PredictorVariable AdjustedOR 95%CI PValueTachypneaattriage(RR>20)ChroniculcersHistoryofMRSAcolonizationorinfectionCellulitisinthepast12monthsChronickidneydiseaseDiabetesmellitus

6.314.904.832.232.601.70

1.80,22.081.68,14.271.51,15.441.01,4.960.82,8.220.87,3.32

0.0040.0040.0080.050.100.12

TheHosmer-Lemeshowchi-squaretestyieldedap-valueof0.604(χ2=1.853,degreesoffreedom=3).C-statistic=0.709.Thisindicatesnoevidenceofpoorfit.RR=respiratoryrate;MRSA=methicillinresistantStaphylococcusaureus;OR=oddsratio;CI=confidenceinterval

65

SupplementaryAppendixFigureS1.StandardizedCaseRecordForm

CASE#:____________________ AbstractorLastName:____________________DoesthiscasemeetINCLUSIONcriteria?oYes oNo(if‘No’,stophere)

- InclusionCriteria:o Age≥18yearso DiagnosedwithSSTI(cellulitisorerysipelas)o Prescribed/receivedanantibioticforSSTI

DoesthiscasemeetEXCLUSIONcriteria?oYes oNo(if‘Yes’,stophere)- ExclusionCriteria:

o Incisionanddrainageofanabscesso Follow-upvisit(i.e.nottheindexvisitforthisSSTI)o Infectedulcerwithnocellulitis

DemographicsHospital: qTOHCivic qTOHGeneralGender: qMale qFemaleDateofbirth:__________/_____/_____(Y/M/D)DateofindexEDvisit:__________/_____/_____(Y/M/D)PastMedicalHistory(qNoneofbelow)qActiveCancer(onchemotherapyorpalliative)qChronickidneydiseaseqChronicvenousinsufficiencyqCongestiveheartfailureqCoronaryarterydiseaseqCorticosteroids(systemic)qDiabetesqHepaticdiseaseqHIVqIntravenousdruguse(IVDU)qLymphedema/VenousStasisqMRSApositiveqObesityqOrgantransplantrecipientqPeripheralvasculardiseaseqPriorcellulitis

66

InfectionCharacteristicsLocationofcellulitis: qArm qLeg qHand qFoot qTorso qGroin qAxilla qFaceChroniculcers: qYes qNoSurgicalsiteinfection: qYes qNoBite(humanoranimal) qYes qNoMaximumlengthoferythema(cm):__________qNotrecordedMaximumwidthoferythema(cm):__________qNotrecordedAreaoferythema(cm2):__________qNotrecorded Areadeterminedby:qDiagram qPhotograph

qText(dimensions) qText(non-specific)PercentTBSAinvolved: q<5% q5–10% q10–20% q>20%TriageVitalSignsTemperature:__________°CHeartrate:__________bpmRespiratoryrate:__________breaths/minuteSystolicBP:__________mmHgSaO2:__________% qRAqSupplementalO2WBCCount:__________(x109/L)CurrentTreatment(PriortoEDVisit)IspatientcurrentlytakingantibioticsforSSTI? oYes oNo

Ifyes,completethefollowing(checkallthatapply):Antibioticpatientiscurrentlytaking:qPOAmoxicillin-Clavulanate(Clavulin) qPOAmoxicillin qPOTMP-SMX(Septra) qPOCephalexin(Keflex)qPOClindamycin qPOCiprofloxacinqPOPenicillinqIVCeftriaxone qIVCefazolin(Ancef)qIVClindamycin qIVPipercillin-TazobactamqIVVancomycin qIVPenicillin

qIVCiprofloxacin qIVMeropenem qOther:_________________________ Duration(fromstarttoEDvisit;days):____________________

67

EDTreatment(whatwasgivenintheED)Route:

qOralAntibioticsonly? qIVAntibioticsonly? qOralandIVAntibiotics?

NumberofOralAntibiotics: q0 q1 q2NumberofIVAntibiotics: q0 q1 q2OralCephalexin(Keflex)? qYes qNo

IfYes,Dose(mg): q250 q500 q1000OralAmoxicillin-Clavulanate? qYes qNo

IfYes,Dose(mg): q500 q875 OralAmoxicillin? qYes qNo

IfYes,Dose(mg): q250 q500 q1000OralClindamycin? qYes qNo

IfYes,Dose(mg): q300 q450 q600OralTMP-SMX(Septra)? qYes qNo

IfYes,Dose(mg): q800(1DStab) q1600(2DStabs)OralCiprofloxacin? qYes qNo

IfYes,Dose(mg): q250 q500 q1000IVCefazolin(Ancef)? qYes qNo

IfYes,Dose(mg): q500 q1000 q1500q2000IVCeftriaxone? qYes qNo

IfYes,Dose(mg): q1000 q1500 q2000IVClindamycin? qYes qNo

IfYes,Dose(mg): q300 q450 q600q900IVMeropenem? qYes qNo

IfYes,Dose(mg): q500 q1000 q1500q2000IVVancomycin? qYes qNo

IfYes,Dose(mg): q500 q1000 q1500q2000IVPiperacillin-Tazobactam? qYes qNo

IfYes,Dose(mg): q3375 q4500 IVCiprofloxacin? qYes qNo

IfYes,Dose(mg): q200 q400

68

WhatwasEDpatientdisposition?qHome qHospitaladmission(Ifadmitted,lengthofstay(days):__________)Ifsenthome:WherewillpatientgetsubsequentIVdoses? qOPAT/CCACqCCAConlyqReturntoEDqNoIVABXIfsenthome:EDPrescription:OralCephalexin(Keflex)? qYes qNo

IfYes,Dose(mg): q250 q500 q1000Frequency: qBID qTID qQID Duration(days):____________________

OralAmoxicillin-Clavulanate? qYes qNo

IfYes,Dose(mg): q500 q875 Frequency: qBID qTID qQID Duration(days):____________________

OralAmoxicillin? qYes qNo


OralClindamycin? qYes qNo


OralTMP-SMX(Septra)? qYes qNo

IfYes,Dose(mg): q800(1DStab) q1600(2DStabs)Frequency: qOD qBID qTID Duration(days):____________________

OralCiprofloxacin? qYes qNo

IfYes,Dose(mg): q250 q500 q1000Frequency: qOD qBID qTID Duration(days):____________________

69

IVCefazolin(Ancef)? qYes qNoIfYes,Dose(mg): q500 q1000 q1500q2000

Frequency: qBID qTID qQID Duration(days):____________________

IVCeftriaxone? qYes qNo

IfYes,Dose(mg): q1000 q1500 q2000Frequency: qOD qBID Duration(days):___________________

IVClindamycin? qYes qNo

IfYes,Dose(mg): q300 q450 q600q900Frequency: qBID qTID qQID Duration(days):___________________

IVMeropenem? qYes qNo


IVVancomycin? qYes qNo

IfYes,Dose(mg): q500 q1000 q1500q2000Frequency: qOD qBID qTID Duration(days):___________________

IVPiperacillin-Tazobactam? qYes qNo

IfYes,Dose(mg): q3375 q4500 Frequency: qBID qTID qQID Duration(days):___________________

IVCiprofloxacin? qYes qNo

IfYes,Dose(mg): q200 q400Frequency: qBID qTID qQID Duration(days):___________________

70

Ifsenthome:ReturntoEDwithin14days: oYes oNo Ifyes(chooseONEonly):

oReturntoEDforSSTIandhospitaladmission oReturntoEDforSSTIbutnoadmission oReturntoEDwithabscess(Ifyes,I&Dperformed? oYes oNo) oReturntoEDforrepeatantibiotics:numberofsubsequentvisits_____oReturntoEDforunrelatedmedicalproblem

Ifsenthome:ChangestotreatmentinEDatrepeatvisit: oYes oNo qN/A

Ifyes(chooseONEonly):oChangefromPOtoIVantibiotics(nameofnewantibiotic:_______________)oChangetodifferentPOantibiotic(nameofnewantibiotic:_______________) oChangetodifferentIVantibiotic(nameofnewantibiotic:_______________)oChangetodifferentantibioticdose(newdose:________________mg)Iftreatmentwasaltered,reasonforchangeintreatmentplan:

oWorseninginfection oNoimprovementoVomiting oDiarrhea oRash/hivesoOther Indicate:_____________________

POTreatmentFailure? qYesqNo Ifyes,reasonfortreatmentfailure(checkallthatapply): qStep-upfromPOtoIVantibiotic qChangeinclassofantibiotic(duetoworseninginfection) qInitialoutpatientmanagementwithsubsequenthospitaladmissionIVTreatmentFailure? qYesqNo Ifyes,reasonfortreatmentfailure(checkallthatapply): qChangeinclassofantibiotic(duetoworseninginfection) qInitialoutpatientmanagementwithsubsequenthospitaladmission

71

Ifsenthome,AdverseEvents: Adversedrugreaction: qYesqNo Ifyes: qRash qNauseaonly(novomiting) qVomiting qDiarrhea qOther:____________________

Adversedeviceevent: qYesqNo Ifyes: qBlockedperipheralIV qDislodgedperipheralIV qLinethrombosis qLineinfection

Subsequenthospitaladmissionat14daysfromindexEDvisit?qYesqNo Ifyes: qForworseningSSTI

qForunrelatedmedicalproblem(specify:_______________)IfreferredtoOPATClinic(qYes qNo[STOPhere]):DidpatientattendOPATappointment? qYesqNo–losttofollow-upCellulitiswascorrectdiagnosis: qYesqNo IfNO–specifyalternatediagnosis:____________________NumberofdaysfrominitialEDvisittofirstOPATappointment:__________TotalnumberofvisitstoOPATclinic:__________TotalnumberofdaysonIVantibiotics:__________TotalnumberofdaysonIVandPOantibiotics:__________(qN/A)

72

TableS1.VariableDefinitions

Variable DefinitionActivecancer Cancerbeingtreatedwithchemotherapyor

palliative;documentedonEDROTorrecentconsult

Chronickidneydisease Impairedrenalfunction≥3months.CKDshouldbechartedoneitherEDrecordoftreatment(ROT)orrecentconsult

Chronicvenousinsufficiency Duplexultrasoundshowingvenousreflux(reversedflowfor>0.5seconds)orchartedontheEDROTorrecentconsult

Congestiveheartfailure DocumentedonROTorrecentconsult.Coronaryarterydisease Priormyocardialinfarction,percutaneous

coronaryinterventionorasotherwisestatedoncardiologyconsult,EDROTorrecentconsult

Corticosteroids(systemic) PatienttakingoralorintravenouscorticosteroidsatthetimeofEDvisit

Diabetesmellitus Takinginsulinororalmedications(e.g.metformin,gliclazide,etc.)atthetimeofEDpresentation,ordocumentedonrecentconsultorEDROT

Hepaticdisease DocumentationonEDROTorrecentconsultofanyofthefollowing:hepatitisB,hepatitisCorlivercirrhosisfromanycause

Injectiondruguse DocumentationonEDROTorrecentconsultofeitherintravenousorsubcutaneousinjectionofmedicationsorrecreationaldrugs

Lymphedema DocumentedonEDROTorrecentconsultMRSApositive Documentedonelectronichealthrecordtohave

knownMRSAcolonizationorpriorMRSAinfectioninthepast12months

Obesity DocumentedonEDROTorrecentconsultOrgantransplantrecipient DocumentedonEDROT,operativereportsora

recentconsult.Includesfollowingtransplants:heart,liver,kidney,andbonemarrow

Peripheralvasculardisease Operativereportofbypassgraftingforperipheralvasculardisease,ordocumentedonEDROTorrecentconsult

Priorcellulitis DocumentedEDvisitfornon-purulentcellulitisorerysipelasinthepast12months

Recentconsult Consultonelectronichealthrecordwithin3monthsofthepatient’sindexEDvisit

73

TableS2.UnivariateAssociationwithOralAntibioticTreatmentFailurefor288EDPatientsTreatedwithaMinimumof48HoursofOralTherapyVariable

OralAntibioticTreatmentFailure(N=85)

NoOralAntibioticTreatmentFailure(N=203)

PValue*

Age,years(mean± SD)Malegender(%)Comorbidities(%)ChronickidneydiseaseCongestiveheartfailureCoronaryarterydiseaseDiabetesmellitusHepaticdiseaseIntravenousdruguseLymphedemaObesityPeripheralvasculardiseaseHistoryofMRSAinfectionPriorcellulitisinpast12months

VitalSignsTemperature,°C(mean±SD)HeartRate,beats/min(mean±SD)BloodPressure,mmHg(mean±SD)RespiratoryRate,breaths/min(median,IQR)OxygenSaturation,%(median,IQR)

InfectionCharacteristics(%)ChronicUlcersBiteSurgicalSiteInfectionSize:TBSA≥5%

67±19

48(56.5)

8(9.4)8(9.4)12(14.1)23(27.1)5(5.9)3(3.5)5(5.9)3(3.5)5(5.9)11(12.9)19(22.4)

36.6±0.884±17140±2216,16–18

97,96–98

13(15.3)1(1.2)7(8.2)14(16.5)

63±20

110(54.2)

7(3.4)10(4.9)17(8.4)32(15.8)12(5.9)1(0.5)9(4.4)5(2.5)12(5.9)5(2.5)19(9.4)

36.3±0.783±16138±2416,16–18

97,96–98

7(3.4)7(3.4)13(6.4)25(12.3)

0.15

0.72

0.040.150.140.030.990.090.560.690.990.0010.003

0.010.680.690.85

0.11

0.00030.440.580.35

*UsingChiSquaredorFisher’sExactTestforcategoricalvariables;t-testsfornormallydistributedcontinuousvariables;Wilcoxontestsfornon-normallydistributedcontinuousvariablesSD=standarddeviation;IQR=interquartilerange;MRSA=methicillinresistantStaphylococcusaureus;TBSA=totalbodysurfacearea

74

TableS3.TreatmentFailurewithIVAntibiotics(N=12of212PatientsTreatedwithaMinimumof48HoursofIVTherapy)

IVAntibioticTreatmentFailures Numberofpatients,N=12N(%)

PatientOutcomesHospitalizedforIVantibioticsSwitchedtooutpatientIVantibiotics

TreatmentfailureoninitialEDvisit*TreatmentfailureonreturnEDvisitwithin14days

11(91.7)1(8.3)

5(41.7)7(58.3)

*Patientwasalreadyon≥48hoursofIVantibiotictherapyattimeofindexEDvisitIV=intravenous;ED=emergencydepartment

75

TableS4.SecondaryAnalysisComparingOralversusIntravenousAntibioticGroupsforCategoricalVariablesforall500Patients

Variable(n)

OralAntibioticGroupFrequency,n(%)

(n=288)

IVAntibioticGroupFrequency,n(%)

(n=212)

PValue*

Male(279) 158(54.9) 121(57.1) 0.62

ChronicUlcers(56) 20(6.9) 36(17.0) 0.0004SurgicalSiteInfection(30)

20(6.9) 10(4.7) 0.30

Bite(12) 8(2.8) 4(1.9) 0.52

>5%TBSA(99) 39(13.5) 60(28.3) <0.0001

Coronaryarterydisease(58)

29(10.1) 29(13.7) 0.21

Congestiveheartfailure(48)

18(6.2) 30(14.1) 0.003

Chronickidneydisease(35)

15(5.2) 20(9.4) 0.07

Chronicvenousinsufficiency(10)

8(2.8) 2(0.9) 0.20

Diabetesmellitus(126)

55(19.1) 71(33.5) 0.0002

HepaticDisease(37) 17(5.9) 20(9.4) 0.14HIV(8) 2(0.7) 6(2.8) 0.08IVDU(14) 4(1.4) 10(4.7) 0.02Lymphedema(33) 14(4.9) 19(9.0) 0.07MRSAHistory(43) 16(5.6) 27(12.7) 0.005Obesity(27) 8(2.8) 19(9.0) 0.002Priorcellulitis(87) 38(13.2) 49(23.1) 0.004PVD(40) 17(5.9) 23(10.8) 0.04

OnSupplementalO2(36)

12(4.2) 24(11.7) 0.002

*UsingChi-SquaredorFisher’sExactTestIV=intravenous;TBSA=totalbodysurfacearea;HIV=humanimmunodeficiencyvirus;IVDU=intravenousdruguse;MRSA=methicillinresistantStaphylococcusaureus;PVD=peripheralvasculardisease

76

TableS5.SecondaryAnalysisComparingOralversusIntravenousAntibioticGroupsforContinuousVariablesforall500Patients

Variable(n)

OralAntibioticGroup

Frequency,n(%)(N=288)

IVAntibioticGroup


PValue*

Age(mean±SD) 64.4±20.0 64.3±16.9 0.95SystolicBP(mean±SD) 139±23 131±25 0.0007Heartrate(mean±SD) 84±16 90±21 <0.0001Temperature(mean±SD) 36.4±0.8 36.8±1.2 <0.0001RespiratoryRate(median,IQR)

16(16–18) 18(16–20) 0.02

SaO2(median,IQR) 97(96–98) 97(96–98) 0.88WBCcount(median,IQR) 8.2(6.6–11.2) 10.4(7.7–14.7) <0.0001*Usingt-testsfornormallydistributedvariables,Wilcoxontestsfornon-normallydistributedvariablesSD=standarddeviation;IQR=interquartilerange;IV=intravenous;BP=bloodpressure;HR=heartrate;SaO2=oxygensaturation;WBC=whitebloodcell

77

ChapterFour:OutpatientParenteralAntibioticTherapyFollowingEmergencyDepartmentTreatmentofNon-PurulentSkinandSoftTissue

Infections

ChapterOverviewInthepreviouschapter,weidentifiedpotentialpredictorsassociatedwithoral

antibiotictreatmentfailurefornon-purulentskinandsofttissueinfections.Most

patientsinthisstudycohortreceivedatleastonedoseofintravenousantibiotic,and

morethanonethirdofdischargedpatientswereprescribedoutpatientparenteral

antibiotictherapy.Duetothestudydesign,itwasnotpossibletoidentifyemergency

physicianrationaleforselectingintravenoustherapy,whichisanimportantfirst

stepinthedecisiontorefertoanoutpatientparenteralantibiotictherapyprogram.

Furthermore,itwasnotpossibletodescribeoutpatientparenteralantibiotic

therapyclinicprocessesortodeterminepatientsatisfactionwiththistreatment

approach.

Thefollowingisamanuscriptpreparedforpublicationbasedonaprospective

observationalcohortstudy.Theobjectivesofthemanuscriptwere:1)describethe

performanceofanemergencydepartment-to-outpatientparenteralantibiotic

therapyclinicprograminthetreatmentofadultswithnon-purulentskinandsoft

tissueinfections;2)identifyemergencyphysicianrationaleforselecting

intravenoustherapy;and3)determinepatientsatisfactionwiththistreatment

approach.

78

InAppendixAacopyoftheapprovalletterfromtheOttawaHealthScience

NetworkResearchEthicsBoardisprovided.

Dr.KrishanYadavisthefirstauthorofthismanuscriptandwasresponsibleforthe

studydevelopment,datacollection,dataanalysisandwritingofthemanuscript.

Thismanuscriptwasco-authoredbyDr.IanStiell,Dr.KathrynSuhandDr.George

Wells.Dr.DebraEaglesandDr.VenkateshThiruganasambandamoorthyprovided

valuablefeedbackthroughouttheprocess.

79

OutpatientParenteralAntibioticTherapyFollowingEmergencyDepartment

TreatmentofNon-PurulentSkinandSoftTissueInfections

KrishanYadav1,KathrynSuh2,DebraEagles3,Venkatesh

Thiruganasambandamoorthy3,GeorgeWells4,5,IanGStiell3

1DepartmentofEmergencyMedicine,UniversityofOttawa

2DepartmentofMedicine,DivisionofInfectiousDiseases,UniversityofOttawa

3DepartmentofEmergencyMedicine,TheOttawaHospitalResearchInstitute,

UniversityofOttawa

4CardiovascularResearchMethodsCentre,UniversityofOttawaHeartInstitute

5DepartmentofEpidemiologyandCommunityMedicine,UniversityofOttawa

Correspondenceto:KrishanYadav

Email:[email protected]

Date:January30,2018

WordCount:2978

Acknowledgments:Theauthorswouldliketothankthefollowingindividualsfor

theirassistanceinthisstudy:My-LinhTran,SherylDomingo,AngelaMarcantonio

andCatherineClement.ThisstudywasfundedbyagrantfromtheDepartmentof

EmergencyMedicine,UniversityofOttawa,Ontario,Canada.Thisstudywasfunded

byagrantfromtheDepartmentofEmergencyMedicine,UniversityofOttawa.

80

AbstractIntroduction

Emergencydepartment(ED)patientswithnon-purulentskinandsofttissue

infections(SSTIs)requiringintravenousantibioticsmaybemanagedviaoutpatient

parenteralantibiotictherapy(OPAT).Todate,therearenoprospectivestudies

describingtheperformanceofanED-to-OPATclinicprogram.Furthermore,there

arenostudiesthathaveexaminedphysicianrationaleforintravenoustherapy,

despitethisbeingacriticalfirststepinthedecisiontorefertoanOPATprogram.

TheprimaryobjectivewastodeterminetheOPATtreatmentfailurerateforadults

withnon-purulentSSTIswhoareinitiallymanagedintheED.

Methods

Weconductedaprospectiveobservationalcohortstudyofadults(age≥18years)

withnon-purulentSSTIsreceivingparenteraltherapyattwotertiarycareEDs.

Patientswereexcludediftheyhadpurulentinfectionsorcouldnotprovideconsent.

OPATtreatmentfailurewasdefinedaprioriashospitalizationafteraminimumof

48hoursofOPATfor:(i)infectionprogression;(ii)line-relatedcomplications(e.g.

bacteremia,thrombophlebitis,venousthromboembolism);or(iii)drug-related

complications(e.g.Clostridiumdifficilecolitis).Secondaryoutcomeswereto

describeOPATclinicprocessesandadverseevents,assesspatientsatisfaction,and

identifyEDphysicianrationaleforselectingintravenousantibiotics.Theemergency

physiciancompletedaformdocumentingrationaleforintravenoustherapy,

81

infectionsizeandchoiceofantimicrobialagent,doseandduration.Patient

satisfactionwasassessedata14-daytelephonefollowup.

Results

Weenrolledaconsecutivesampleof153patients(meanage60years,82male

(53.6%)and38(24.8%)withdiabetes).Atotalof137patients(89.5%)attended

theirOPATclinicappointment.Ofthe101patientsprescribedcefazolin,50.5%

received1000mgand48.5%received2000mg.TherewerelowratesofOPAT

treatmentfailure(4.4%).Noneoftheadverseperipheralintravenouslineevents

(10.9%)andadverseantibioticevents(8.0%)requiredhospitalization.Patients

reportedahighdegreeofsatisfactionwithtimelinessofclinicreferral(median

score9outof10)andoverallcarereceived(medianscoreof10outof10).Thetop

5reasonsgivenbyphysiciansforselectingintravenoustherapywere:clinical

impressionofseverity(52.9%);failedoralantibiotictherapy(41.8%);diabetes

(17.6%);severepain(7.8%);andperipheralvasculardisease(7.8%).

Conclusion

Thisisthefirststudytoidentifyphysicianrationalefortheuseofintravenous

antibioticsforSSTIs.Therewassignificantvariabilityinantibioticprescribing

practicesbyEDphysicians.ThisprospectivestudydemonstratesthatanED-to-

OPATclinicprogramfornon-purulentSSTIsissafe,hasalowrateoftreatment

failuresandresultsinhighpatientsatisfaction.

82

Introduction

Firstdescribedin1974,outpatientparenteralantibiotictherapy(OPAT)isdefined

astheadministrationofatleasttwodosesofparenteralantimicrobialsondifferent

dayswithoutinterimobservation.1OPATisanattractiveoptionforadultswithnon-

purulentskinandsofttissueinfections(SSTIs)whorequireintravenousantibiotics.

Patientsdischargedfromtheemergencydepartment(ED)canreceiveOPATfollow

upinanumberofsettings:returntotheED,afamilyphysicianclinic,ahomecare

clinicrunbynurses,oradedicatedOPATclinic.Duetodifficultieswithprimarycare

access2andassociationsbetweenEDovercrowdingandincreasedadverseevents3,

anED-to-OPATclinicprogrammaybeapreferredoptionduetoimportant

advantages:1)decreasedhospitaladmissions;2)increasedpatientconvenience;

and3)decreasedEDvisits.

ED-basedstudieshaveshownthatintravenousantibioticsarefrequently

administered4,5andthatantibioticoveruseiscommon.6Surprisingly,thereisscant

literatureregardingEDantibioticprescribingforSSTIs.7-9Furthermore,thereareno

publishedstudiesdescribingEDphysicianrationaleforselectingintravenousover

oralantibiotics,despitethisbeingacriticalfirststepinthedevelopmentofanOPAT

program.1,8AdministrationofOPATisnotwithoutrisk.Potentialadverseeventsfor

anyinfectiontreatedwithOPATincludeinfectionprogression,peripheralline-

relatedcomplicationsoradverseantibioticeventsthatmaywarrantsubsequent

hospitalization.9,10PriorstudiesreportthehospitaladmissionratefollowingOPAT

rangingfrom2.6to8%.9,11,12However,thesestudiesincludedanumberof

83

infections,makingitdifficulttodeterminethetrueclinicalfailurerateforSSTIsin

particular.Arecentretrospectivestudyreportedahospitaladmissionrateof5.5%

foracellulitisOPATcohort,althoughthereasonsforhospitalizationwerenot

described.13

TheprimaryobjectiveofthisstudywastodeterminetheOPATtreatmentfailure

rateforadultswithnon-purulentSSTIswhoareinitiallymanagedintheED.

SecondaryobjectivesweretodescribeOPATclinicprocesses(timetofirst

appointment,totalnumberofclinicvisitsanddurationoftherapy),adverseevents,

assesspatientsatisfaction,andidentifyemergencyphysicianrationaleforselecting

intravenousantibiotics,

Methods

StudyDesignandSettingWeconductedaprospectiveobservationalcohortstudyofadultpatientswithnon-

purulentSSTIsinitiallymanagedintheEDwithintravenousantibioticsandreferred

totheOPATclinic.ThestudypopulationwasenrolledfromtheOttawaHospital

CivicandGeneralEDs(twotertiarycareadultEDswithacombined170,000patient

visitsannually).AllenrolledpatientswerereferredtotheOPATclinic,locatedatthe

OttawaHospitalCivicCampus.TheOttawaHospitalOPATclinicwasestablishedin

2014andoperatesonthreehalfdaysperweekbyappointmentonly.TheOttawa

HealthScienceNetworkResearchEthicsBoardapprovedthisstudy.

84

StudyPopulationWeenrolledaconsecutivesampleofpatientsmeetingeligibilitycriteriafrom

January15toJune20,2017.Eligiblepatientswereadults(age≥18years)

presentingtotheEDanddiagnosedwithanon-purulentSSTIthatwasfeltbythe

emergencyphysiciantorequireintravenousantibioticsandfollowupintheOPAT

clinic.Patientswereexcludedforthefollowingreasons:(i)age<18years;(ii)

diagnosisofapurulentSSTIwhereanincisionanddrainageprocedurewas

performed;(iii)necrotizinginfections;(iv)significantcognitiveorverbal

impairmentsuchthatinformedconsentwasnotfeasible;or(v)thosewhowerenot

localresidentsorwhodidnothaveatelephone.

IntravenousAntibioticTreatmentTheemergencyphysicianselectedtheagent,dose,frequencyanddurationof

parenteralantibioticattheirowndiscretion.Allpatientsreceivedintravenous

antibioticsinthecommunityviathelocalhomecareprogram.Patientsreturnedto

theEDforsubsequentintravenousdosesiftherewasadelayinestablishing

homecareasanoutpatient.Patientsthenfollowedupwithaninfectiousdisease

specialistattheOPATclinic,atthenextavailableclinicdate(rangingfromtwoto

tendays).Theinfectiousdiseasespecialistwasresponsiblefordeterminingif

furtherintravenoustherapywaswarrantedorifthepatientcouldbesteppeddown

tooraltherapy.

DataCollectionAllpatientswereassessedbyemergencyphysiciansorresidentssupervisedby

attendingemergencyphysicians.Studydetailsweredistributedbyelectronicmailto

familiarizephysicianswiththestudy.Theemergencyphysicianwasresponsiblefor

85

completingtheOPATreferralform(seethesupplementaryappendix),whichalso

requiredthephysiciantoindicaterationaleforselectingintravenoustherapy.

Physicianswereaskedtousearulerontheright-marginofthereferralformto

obtainaccurateinfectiondimensions.Patientswereprovidedwithaninformation

sheetoutliningthestudydetailsandverbalconsentwasobtainedfora14-day

telephonefollow-upcall.Theprincipalinvestigator(KY)screenedallreferralstothe

OPATclinictoensurethatnocasesweremissed.

Allpatientsreceiveda14-daytelephonefollowuptoassesspatientsatisfaction.

Participantswereconsideredlosttotelephonefollowupiftheycouldnotbe

reachedafteramaximumofthreephonecalls.Theprincipalinvestigatorabstracted

allrelevantclinicaldatafromtheelectronichealthrecord(EDphysicianandnursing

notes,OPATclinicrecords)ontoastandardizedcaserecordform(seethe

supplementaryappendix).Eligiblepatientswereenrolledregardlessofwhetherthe

OPATreferralformwasfullycompleted.

OutcomeMeasuresTheprimaryoutcomewasOPATtreatmentfailure.Thereisnoestablished

definitionofOPATtreatmentfailureforSSTIsintheliterature.TheUnitedStates

FoodandDrugAdministrationrecommendsthatclinicalresponsetotreatment

shouldbeassessedat48to72hoursfrominitiatingtherapy.14Whensurveyed,a

majorityofCanadianemergencyphysiciansselected48hoursastheoptimal

timeframefordeterminingiftreatmentfailurehadoccurredfollowinginitiationof

antibiotictherapy.15Afterreviewoftheliterature4,5,16-20anddiscussionwithlocal

86

expertsinemergencymedicineandinfectiousdisease,wedevisedadefinitionof

OPATtreatmentfailure.Patientswereconsideredtohaveatreatmentfailureifa

patientwassubsequentlyadmittedtohospitalafteraminimumof48hoursofOPAT

foranyofthefollowingreasons:(i)worseninginfection;(ii)line-related

complications(e.g.bacteremia,venousthrombosis);or(iii)drug-related

complications(e.g.Clostridiumdifficilecolitis).

SecondaryoutcomesincludedEDphysicianrationaleforselectingtheintravenous

route,OPATclinicdataandpatientsatisfaction.Thetreatingphysicianwasallowed

toselectmorethanonereasonforselectingintravenoustherapy.Forclinicdata,we

recordedtimetofirstvisit,totalnumberofvisits,andthenumberofpatientslostto

follow-up.Ata14-daytelephonefollowup,weaskedpatientstogiveanumerical

ratingfromone(leastsatisfied)toten(mostsatisfied)withrespecttotimelinessof

referralfromtheEDandoverallpatientsatisfaction.Adverseoutcomesincluded

antibioticeventsanddeviceevents.

DataAnalysisWeuseddescriptivestatisticstodescribetheproportionofpatientswhohadan

OPATtreatmentfailure,adverseantibioticordeviceevents,OPATclinicdataand

patientsatisfaction.Continuousdataarepresentedasmeanswithstandard

deviationsormedianswithaninterquartilerange(IQR,Q1–Q3)fornormallyand

non-normallydistributeddata,respectively.Categoricaldataarepresentedas

proportionswith95%confidenceintervals.

87

SampleSizeTheOPATclinicattheOttawaHospitaltreatsapproximately30newpatientswith

SSTIspermonth.Afteradjustingfor10%dropout,weestimatedaconsecutive

enrolmentofapproximately135patientsovera5-monthperiod.Thissamplesize

wasbasedonfeasibilityduetofundingandtimeconstraints.

ResultsWescreened214casesreferredtotheOPATclinicoverthefive-monthstudyperiod

andidentified153eligiblecases(Figure1).Atotalof137patients(89.5%)attended

theirclinicappointment.Oftheremaining16patients(10.5%),fivepatientswere

admittedtohospitalpriortotheirclinicappointmentand11werelosttofollowup.

Forpatientswhoattendedtheirappointment,wewereabletocontact118patients

(86.0%)fora14-daytelephonefollowup.

Tables1and2highlightthebaselinepatientandinfectioncharacteristics,

respectively.Themeanageofenrolledpatientswas60years,andalmostaquarter

ofpatientshaddiabetes(24.8%)orlymphedema(23.5%).Almosthalfofpatients

(48.4%)weretakingoralantibioticsatthetimeoftheirpresentationtotheED.The

mostcommonlocationofinfectionwastheleg(45.1%)andthemedianareaof

erythemawas150cm2(IQR40–300).Emergencyphysiciansorderedbloodtestsin

themajorityofthesepatients(68.0%)andthemedianwhitebloodcellcountwas

9.0×109/L(IQR6.8–11.6).Thetreatingphysicianorderedbloodculturesfor22

patients(14.4%);therewereclinicallysignificantpositivebloodcultures(one

positivebloodculturewasconsideredtobeacontaminant).

88

Table3showsthevariationinantimicrobialtherapyadministeredintheED.Most

patientsreceivedasingleintravenousantibiotic(93.5%).Cefazolinwasthemost

commonlyadministeredantibioticandwasadministeredin99patients(64.7%).Of

thosethatreceivedcefazolin,33patients(33.3%)and66patients(66.7%)received

1000mgand2000mg,respectively.

InTable4,thevariationinantibioticprescribingpracticesforpatientssenthome

fromtheEDispresented.Cefazolinwasthemostcommonlyprescribedantibiotic

(66.0%).Thechosendoseofcefazolinfor101patientswasasfollows:1000mg(n=

51,50.5%);1500mg(n=1,1.0%);and2000mg(n=49,48.5%).Aftercefazolin,

themostcommonlyprescribedantibioticswereceftriaxone(19.6%)and

clindamycin(10.5%).

Emergencyphysicianrationaleforselectingintravenousantibioticsisshownin

Table5.Treatingphysiciansprovidedatotalof22differentrationales.Thetopfive

reasonsforselectingintravenoustherapywere:clinicalimpression(52.9%);failed

oralantibiotictherapy(41.8%);diabetes(17.6%);severepain(7.8%);and

peripheralvasculardisease(7.8%).

Ofthetotal153patientsreferredtotheOPATclinic,137patients(89.5%)attended

theirscheduledappointment.TheOPATclinicdataareshowninTable6.The

emergencyphysicianandinfectiousdiseasespecialistdiagnosisofanon-purulent

89

SSTIwereconcordantin93.4%ofcases.Themediantimetothefirstclinicvisitwas

5days(IQR4–7).Themediandurationofintravenoustherapywas9days(IQR7–

14).Patientsreportedahighdegreeofsatisfactionwithtimelinessofclinicreferral

(medianscore9outof10)andoverallcarereceived(medianscoreof10outof10).

Ofthe118patientswhocouldbereachedfortelephonefollowup,110patients

(93.2%)indicatedtheywouldpreferfollow-upwiththeOPATcliniciftheyrequired

intravenousantibioticsinthefuture.

Table7shows14-dayoutcomesforthe137patientswhoattendedtheirinitial

OPATclinicappointment.Themajorityofpatients(63.5%)returnedtotheED

within14days.Themostcommonreasonwasareturnforscheduledintravenous

dosesifthehomecareprogramcouldnotbeinitiatedintimeforthenextrequired

dose.OnlysixpatientssufferedanOPATtreatmentfailure(4.4%);allweredueto

worseninginfection.Fifteenpatients(10.9%)hadanadversedeviceevent(blocked

ordislodgedperipheralintravenousline)andtherewerenocasesoflineinfection

orbacteremia.Elevenpatients(8.0%)hadanadverseantibioticeventwiththe

majorityexperiencingdiarrhea.Noneoftheadversedeviceorantibioticevents

resultedinhospitalization.

Discussion

InterpretationofResultsThisprospectiveobservationalcohortstudydescribesED-to-OPATclinic

performanceinthemanagementofadultswithnon-purulentSSTIs.Therewas

90

significantpracticevariationamongstemergencyphysiciansregardingantibiotic

agent,doseandduration.Cefazolinwasthemostcommonlyprescribedparenteral

antibioticandemergencyphysiciansweresplitregardingthedose.Thisisthefirst

studytoprospectivelyevaluateemergencyphysicianrationaleforintravenous

antibiotics.Thetopfivereasonsforselectingintravenoustherapywere:clinical

impression;failedoralantibiotictherapy;diabetes;severepain;andperipheral

vasculardisease.TherewasaverylowrateofbothOPATtreatmentfailuresand

adverseantibioticorperipherallineevents.Patientsreportedaveryhighdegreeof

satisfactionwiththeircare.OurfindingsstrengthentheargumentthatanED-to-

OPATclinicmodeliseffective,safe,andresultsinahighdegreeofpatient

satisfaction.

PreviousStudiesOurstudyfoundalowOPATtreatmentfailurerate(4.4%)thatissimilarto

previouslypublishedstudies.9,11,12However,thisisthefirstprospectivestudyto

assessOPATfornon-purulentSSTIsinparticular.Appropriatepatientselectionfor

outpatientintravenoustherapyisacriticalfirststeptothesuccessofanOPAT

program.1,8However,therearenopublishedstudiestodatethathaveexamined

physicianrationaleforselectingparenteraltherapy.Currentguidelines1,21,22list

patientsatisfactionasakeyelementtoanOPATprogram.Wefoundahighdegreeof

patientsatisfactionthatissimilartopriorstudies.12,23

StrengthsAconsecutivesampleofpatientswasenrolledandwescreenedallOPATclinic

referralssoasnotmissanypotentialcasesoverthestudyperiod.Thedatawere

91

collectedprospectivelyandwewerethusabletoaccuratelyidentifyphysician

rationaleforselectingparenteraltherapy.Thisisthefirstprospectivestudythat

providesimportantinsightintophysiciandecision-makingregardingpatient

selectionforintravenoustherapy.

LimitationsTherearestudylimitationsthatwarrantmention.First,thisstudywasconductedat

twotertiarycareEDsbuthadasmallsamplesizeduetofeasibility.Somepatients

didnotattendtheirclinicappointmentandwerelosttofollow-up(7.2%).Itis

possiblethatsomeofthesepatientslosttofollowupmayhaveexperienceda

treatmentfailure.However,wefeelitisunlikelythatpatientswhoreceived

treatmentatourcentrewouldpresenttoanotherhospitalforfurthercare.Another

limitationwasthat14.0%ofpatientswhoattendedtheclinicvisitcouldnotbe

reachedbytelephoneordeclinedconsent.Itispossiblethattheirpatient

satisfactionscoresmayhavedifferedsignificantlyfromthegroupthathada

completefollowup.Third,theconfidenceintheresultsmaybeaffectedbythesmall

samplesize.

Afurtherlimitationisthatwecouldnotrecordpatientweightorcategorizethe

degreeofobesityduetofeasibilityinobtainingthisdata.Itispossiblethatatleast

someofthevariationobservedinantimicrobialdosing,durationandfrequency

mightberelatedtopatientbodyhabitus.However,wefeelthisisunlikelybecause

thesmallproportionofpatientswithobesity(12.4%)wouldnotfullyexplainthe

highdegreeofobserveddosingvariability.

92

Lastly,thereisnovalidateddefinitionofOPATtreatmentfailure.Petraketal.

recommendedthatfuturestudiesexaminingOPATefficacyshouldincluderobust

definitionsfortreatmentfailure.11Weattemptedtodeveloparobustdefinitionby

reviewingtheliterature4,5,16-20andreachingconsensusamonglocalexpertsin

emergencymedicineandinfectiousdisease.Thisdefinitionwasacomposite

endpoint.Disadvantagesofusingacompositeendpointincludemisleadingresultsif

onecomponentoftheoutcomeheavilydrivestheresult,especiallyifitistheleast

patient-importantoutcome.However,thecomponentsofourdefinitionallresulted

inthesamefinaloutcomeofhospitaladmission.

ClinicalImplicationsThereareimportantclinicalimplicationsthatdeservemention.Ourstudy

demonstratedahighdegreeofvariationinantimicrobialprescribingpractices.

Whilethismayinpartbeduetovariationinpatientpresentation(e.g.animalbite,

antibioticallergy),theresultsstillreflectalackofconsensusamongstemergency

physiciansregardingoptimalmanagementofthiscommonclinicalcondition.For

example,thedoseforthemostcommonlyprescribedantibiotic(cefazolin)was

nearlyevenlysplit–whichcannotbeexplainedbyheterogeneityinpatient

presentationalone.TheratesofadverseeventsandOPATtreatmentfailurewere

verylow,suggestingthattheED-to-OPATclinicmodelissafe,effectiveandresultsin

highpatientsatisfaction.Thismodelcanbeintroducedinothercommunitiesto

potentiallydecreasehospitaladmissionsandassociatedhealthcarecosts.

93

ResearchImplicationsOurstudyidentifiedreasonsemergencyphysiciansinstituteintravenousantibiotic

therapyfornon-purulentSSTIs.Futurestudiesshouldseektoassesswhether

patientswiththeseidentifiedrationalestrulyrequireintravenoustherapy,or

whethertheycanbetreatedwithlessinvasiveandcheaperoralantibiotics.Wealso

identifiedalargedegreeofvariationwithantibioticprescriptionpractices.

Randomizedclinicaltrialscomparingvariousdosesanddurationsofintravenous

therapywillaidinmakingmorerobustguidelinestoaidemergencyphysicians

whenselectingtheappropriateantibioticroute.

ConclusionThisisthefirststudytoidentifyphysicianrationalefortheuseofintravenous

antibioticsforSSTIs.Therewassignificantvariabilityinantibioticprescribing

practicesbyEDphysicians.ThisprospectivestudydemonstratesthatanED-to-

OPATclinicprogramfornon-purulentSSTIsissafe,hasalowrateoftreatment

failuresandresultsinhighpatientsatisfaction.

94

References1. TiceAD,RehmSJ,DalovisioJR,etal.Practiceguidelinesforoutpatient

parenteralantimicrobialtherapy.IDSAguidelines.ClinInfectDis.2004;38(12):1651-1672.Epub2004May1626.

2. WilsonK,RosenbergMW.AccessibilityandtheCanadianhealthcaresystem:squaringperceptionsandrealities.HealthPolicy.2004;67(2):137-148.

3. McCuskerJ,VadeboncoeurA,LevesqueJF,CiampiA,BelzileE.Increasesinemergencydepartmentoccupancyareassociatedwithadverse30-dayoutcomes.AcadEmergMed.2014;21(10):1092-1100.doi:1010.1111/acem.12480.



6. MayL,HarterK,YadavK,etal.Practicepatternsandmanagementstrategiesforpurulentskinandsoft-tissueinfectionsinanurbanacademicED.AmJEmergMed.2012;30(2):302-310.doi:310.1016/j.ajem.2010.1011.1033.Epub2011Jan1028.

7. MayL,GudgerG,ArmstrongP,etal.Multisiteexplorationofclinicaldecisionmakingforantibioticusebyemergencymedicineprovidersusingquantitativeandqualitativemethods.InfectControlHospEpidemiol.2014;35(9):1114-1125.doi:1110.1086/677637.Epub672014Jul677623.

8. TimbrookTT,CaffreyAR,OvalleA,etal.AssessmentsofOpportunitiestoImproveAntibioticPrescribinginanEmergencyDepartment:APeriodPrevalenceSurvey.InfectDisTher.2017;6(4):497-505.doi:410.1007/s40121-40017-40175-40129.Epub42017Oct40119.

9. KamathRS,SudhakarD,GardnerJG,HemmigeV,SafarH,MusherDM.GuidelinesvsActualManagementofSkinandSoftTissueInfectionsintheEmergencyDepartment.OpenForumInfectDis.2018;5(1):ofx188.doi:110.1093/ofid/ofx1188.eCollection2018Jan.

10. GilchristM,SeatonRA.Outpatientparenteralantimicrobialtherapyandantimicrobialstewardship:challengesandchecklists.JAntimicrobChemother.2015;70(4):965-970.doi:910.1093/jac/dku1517.Epub2014Dec1023.

95


12. HodgsonKA,HuynhJ,IbrahimLF,etal.Theuse,appropriatenessandoutcomesofoutpatientparenteralantimicrobialtherapy.ArchDisChild.2016;101(10):886-893.doi:810.1136/archdischild-2015-309731.Epub302016May309710.



15. ZhangJ,MooreE,BousfieldR.OPATforcellulitis:itsbenefitsandthefactorsthatpredisposetolongertreatment.EurJClinMicrobiolInfectDis.2016;35(6):1013-1015.doi:1010.1007/s10096-10016-12631-y.Epub12016Apr10015.






96



23. ChapmanAL,SeatonRA,CooperMA,etal.Goodpracticerecommendationsforoutpatientparenteralantimicrobialtherapy(OPAT)inadultsintheUK:aconsensusstatement.JAntimicrobChemother.2012;67(5):1053-1062.doi:1010.1093/jac/dks1003.Epub2012Jan1031.


25. CorwinP,ToopL,McGeochG,etal.Randomisedcontrolledtrialofintravenousantibiotictreatmentforcellulitisathomecomparedwithhospital.BMJ.2005;330(7483):129.Epub2004Dec2016.

97

FiguresFigure1.FlowDiagram

PatientsScreened(N=214)

ExcludedPatients,total(N=61)Abscess(N=17)Osteomyelitis(N=16)Infectedulcerwithoutcellulitis(N=15)NotdiagnosedwithSSTI(N=10)Givenoralantibiotics(N=1)Previouslyenrolled(N=2)

EligiblePatients(N=153)

AttendedOPATclinic(N=137)(89.5%)

Admittedtohospital(N=5)(3.3%)Losttofollowup(N=11)(7.2%)

14-Daytelephonefollowup(N=118)(86.0%)

Losttofollowup(N=12)(8.8%)Declinedconsent(N=7)(5.2%)

98

TablesTable1.BaselineCharacteristicsofAdultswithNon-PurulentSkinandSoftTissueInfections(SSTIs)seenintheED(N=153)

Variable N=153Age(years),mean± SDRange

Male(%)HospitalSite(%)

TOHCivicCampusTOHGeneralCampus

ComorbiditiesDiabetesmellitusLymphedemaPriorcellulitisinpast12monthsObesityPeripheralvasculardiseaseCoronaryarterydiseaseActivecancerChronickidneydiseaseCongestiveheartfailureLiverdiseaseInjectiondruguseHistoryofMRSAcolonizationorinfectionOrgantransplantrecipient

MedicationsCurrentlytakingoralantibiotics

60±1921–100

82(53.6)

80(52.3)73(47.7)

38(24.8)36(23.5)26(17.0)19(12.4)12(7.8)10(6.5)8(5.2)8(5.2)7(4.6)7(4.6)7(4.6)5(3.3)1(0.6)

74(48.4)

SD=standarddeviation;TOH=TheOttawaHospital;MRSA=methicillinresistantStaphylococcusaureus;IV=intravenous

99

Table2.PresentingPatientandInfectionCharacteristics(N=153)

Variable N=153TriageVitalSignsTemperature,°C(mean±SD)HeartRate,beats/min(mean±SD)BloodPressure,mmHg(mean±SD)RespiratoryRate,breaths/min(mean±SD)OxygenSaturation,%(median,IQR)

InfectionLocation(%)LegFootArmHandFaceTorso

InfectionCharacteristics(%)ChronicLegUlcersBiteSurgicalSiteInfectionInfectiondimensionsrecorded(%)Areaoferythema,cm2(median,IQR)

LaboratoryTestsWhitebloodcellcountordered(%)Whitebloodcellcount,×109/L(median,IQR)

BloodCultureSentNegativeBloodCultureContaminantPositiveBloodCulture

36.5±0.786±17138±2117±2

97,96–98

69(45.1)21(13.7)23(15.0)19(12.4)16(10.5)5(3.3)

20(13.1)11(7.2)4(2.6)

124(81.0)150,40–300

104(68.0)9.0,6.8–11.6

22(14.4)21(13.7)1(0.6)0(0)

SD=standarddeviation;IQR=interquartilerange;TBSA=totalbodysurfacearea

100

Table3.IntravenousAntibioticTreatmentAdministeredWhileintheED(N=153)

AntibioticSelection Numberofpatients,N=153N(%)

IVantibioticsonlyOralandIVantibioticsNumberofIVantibioticsOneTwo

IVantibioticagentandsingledoseCefazolinDose=1000mgDose=2000mg

CeftriaxoneDose=1000mgDose=2000mg

ClindamycinDose=300mgDose=600mgDose=900mg

VancomycinDose=1000mgDose=1500mgDose=2000mg

Piperacillin-Tazobactam(3375mg)Meropenem(1000mg)Ciprofloxacin(400mg)

Oralantibiotics*ClindamycinCiprofloxacinTrimethoprim-Sulfamethoxazole

148(96.7)5(3.3)

143(93.5)10(6.5)

99(64.7)33(21.6)66(43.1)37(24.2)31(20.3)6(3.9)

17(11.1)1(0.6)12(7.8)4(2.6)5(3.3)3(2.0)1(0.6)1(0.6)2(1.3)2(1.3)1(0.6)

2(1.3)2(1.3)1(0.6)

IV=intravenous*Giveninadditiontointravenousantibiotics

101

Table4.IntravenousAntibioticPrescriptionsforPatientsDischargedfromtheED(N=153)

EDIVAntibioticPrescription Numberofpatients,N=153N(%)

CefazolinSingleDose1000mg1500mg2000mg

FrequencyTwicedailyThreetimesdailyFourtimesdaily

Duration,days(median,IQR)Range

CeftriaxoneSingleDose1000mg2000mg

FrequencyOncedailyTwicedaily


ClindamycinSingleDose300mg450mg600mg900mg

FrequencyTwicedailyThreetimesdaily


VancomycinPiperacillin-TazobactamMeropenemCiprofloxacin

101(66.0)

51(33.3)1(0.6)49(32.0)

5(3.3)94(61.4)2(1.3)

7,7–73–14

30(19.6)

25(16.3)5(3.3)

1(0.6)29(19.0)7,7–93–10

16(10.5)

1(0.6)1(0.6)11(7.2)3(2.0)

1(0.6)15(9.8)7,7–75–14

3(2.0)2(1.3)2(1.3)1(0.6)

IQR=interquartilerange(Q1–Q3);Range=min–max;IV=intravenous,ED=emergencydepartment

102

Table5.EmergencyPhysicianRationaleforIVAntibioticsforall153Patients

RationaleforIVAntibiotics* Numberofpatients,N(%)N=153

ClinicalImpressionFailedoralantibiotictherapyDiabetesSeverepain(>8/10)PeripheralvasculardiseaseBitePriorSSTIthatrequiredIVantibioticsSocial/complianceissuesAbnormalskinatinfectionsite**PriororalantibioticfailurePriorSSTIinsameareaHypotensionorfeverandtachycardiaRapidlyspreadingerythemaorlymphangitisIndwellingIVcatheterBlunttraumaEarinvolvementInjectiondruguseImmunocompromised

81(52.9)64(41.8)27(17.6)12(7.8)12(7.8)7(4.6)5(3.3)5(3.3)3(2.0)2(1.3)2(1.3)2(1.3)2(1.3))1(0.6)1(0.6)1(0.6)1(0.6)1(0.6)

*Emergencyphysiciansindicated>1rationaleforsomepatients**Surgicalsiteinfection,underlyingburn,underlyingmelanomaIV=intravenous;SSTI=skinandsofttissueinfection

103

Table6.OutpatientParenteralAntibioticTherapy(OPAT)ClinicData(N=137)

OPATClinicMetric Numberofpatients,N(%)

N=137Cellulitisconfirmedasdiagnosis*AlternatediagnosesAbscessOsteomyelitisDrugrashStasisdermatitis

Timetofollow-up(days),median,IQRRange

Totalclinicvisits,median,IQRTotalDurationofIVantibiotics(days),median,IQRTotalDurationofIVandoralantibiotics(days),median,IQRPatientfollowupat14daysPatientsatisfaction(scaleof1to10**)TimelinessofreferraltoOPATclinicOverallsatisfactionwithcarereceived

PatientpreferenceforfollowupinfutureOPATclinicFamilydoctorEDUnknown(losttofollowup)

128(93.4)

3(2.2)3(2.2)2(1.4)1(0.7)

5,4–71–18

2,1–3

9,7–14

17,12–28

118(86.1)

9,8–1010,9–10

110(80.3)5(3.6)3(2.2)19(13.9)

*Theinfectiousdiseasespecialistagreedwiththeemergencyphysiciandiagnosisofanon-purulentSSTI**1isleastsatisfiedand10ismostsatisfiedOPAT=outpatientparenteralantibiotictherapy;IV=intravenous;IQR=interquartilerange,Q1–Q3

104

Table7.Outcomesat14DaysfromIndexEDVisitfor137PatientsNotLosttoFollow-up

AdverseEvents Numberofpatients,N=137,N(%)

ReasonforReturnEDVisitWithin14DaysRepeatIVantibioticdosesNumberofrepeatvisits,median,IQRRange

ForSSTIanddischargedhomeForSSTIandhospitaladmission(OPATTreatmentFailure)DiagnosedwithabscessrequiringI&DUnrelatedmedicalproblem

AdverseDeviceEventsBlockedperipheralIVlineDislodgedperipheralIVlineThrombophlebitis,lineinfectionorbacteremia

AdverseAntibioticEventsDiarrheaRashOralthrushNauseaand/orvomiting

55(40.1)1,1–21–5

18(13.1)6(4.4)

4(2.9)4(2.9)

15(10.9)9(6.6)6(4.4)0(0)

11(8.0)8(5.8)2(1.4)1(0.7)0(0)

IQR=interquartilerange(Q1–Q3);Range=min–maxED=emergencydepartment;IV=intravenous;SSTI=skinandsofttissueinfection;I&D=incisionanddrainage

105

SupplementaryAppendixFigureS1.OPATClinicReferralForm

106

FigureS2.StandardizedCaseRecordForm

CASE#:____________________ AbstractorLastName:____________________DemographicsHospital: qTOHCivic qTOHGeneralGender: qMale qFemaleDateofbirth:__________/_____/_____(Y/M/D)DateofindexEDvisit:__________/_____/_____(Y/M/D)PastMedicalHistory(qNoneofbelow)qActiveCancer(onchemotherapyorpalliative)qChronickidneydiseaseqChronicvenousinsufficiencyqCongestiveheartfailureqCoronaryarterydiseaseqCorticosteroids(systemic)qDiabetesqHepaticdiseaseqHIVqIntravenousdruguse(IVDU)qLymphedema/VenousStasisqMRSApositiveqObesityqOrgantransplantrecipientqPeripheralvasculardiseaseqPriorcellulitisInfectionCharacteristicsLocationofcellulitis: qArm qLeg qHand qFoot qTorso qGroin qAxilla qFaceChroniculcers: qYes qNoSurgicalsiteinfection: qYes qNoBite(humanoranimal) qYes qNoMaximumlengthoferythema(cm):__________qNotrecordedMaximumwidthoferythema(cm):__________qNotrecordedAreaoferythema(cm2):__________qNotrecorded Areadeterminedby:qDiagram qPhotograph

qText(dimensions) qText(non-specific)PercentTBSAinvolved: q<5% q5–10% q10–20% q>20%

107

TriageVitalSignsTemperature:__________°CHeartrate:__________bpmRespiratoryrate:__________breaths/minuteSystolicBP:__________mmHgSaO2:__________% qRAqSupplementalO2WBCCount:__________(x109/L)BloodCulturessent? oYes oNo IfYes: oPositive oNegative oContaminated IfPositive: oMSSA oMRSA oGroupAStrep oPseudomonas oProteus oStrepPneumoniae oSerratia oEikenella oVibrio oH.influenzae oContaminatedCurrentTreatment(PriortoEDVisit)IspatientcurrentlytakingantibioticsforSSTI? oYes oNo

Ifyes,completethefollowing(checkallthatapply):Antibioticpatientiscurrentlytaking:qPOAmoxicillin-Clavulanate(Clavulin) qPOAmoxicillin qPOTMP-SMX(Septra) qPOCephalexin(Keflex)qPOClindamycin qPOCiprofloxacinqPOPenicillin qPODoxycyclineqIVCeftriaxone qIVCefazolin(Ancef)qIVClindamycin qIVPipercillin-TazobactamqIVVancomycin qIVPenicillin

qIVCiprofloxacin qIVMeropenem qOther:_________________________ Duration(fromstarttoEDvisit;days):____________________

108

EDTreatment(whatwasgivenintheED)Route:

qOralAntibioticsonly? qIVAntibioticsonly? qOralandIVAntibiotics?

NumberofOralAntibiotics: q0 q1 q2 q3NumberofIVAntibiotics: q0 q1 q2 q3OralCephalexin(Keflex)? qYes qNo

IfYes,Dose(mg): q250 q500 q1000OralAmoxicillin-Clavulanate? qYes qNo

IfYes,Dose(mg): q500 q875 OralAmoxicillin? qYes qNo

IfYes,Dose(mg): q250 q500 q1000OralClindamycin? qYes qNo

IfYes,Dose(mg): q300 q450 q600OralTMP-SMX(Septra)? qYes qNo

IfYes,Dose(mg): q800(1DStab) q1600(2DStabs)OralCiprofloxacin? qYes qNo

IfYes,Dose(mg): q250 q500 q1000IVCefazolin(Ancef)? qYes qNo

IfYes,Dose(mg): q500 q1000 q1500q2000IVCeftriaxone? qYes qNo

IfYes,Dose(mg): q1000 q1500 q2000IVClindamycin? qYes qNo

IfYes,Dose(mg): q300 q450 q600q900IVMeropenem? qYes qNo

IfYes,Dose(mg): q500 q1000 q1500q2000IVVancomycin? qYes qNo

IfYes,Dose(mg): q500 q1000 q1500q2000IVPiperacillin-Tazobactam? qYes qNo

IfYes,Dose(mg): q3375 q4500 IVCiprofloxacin? qYes qNo

IfYes,Dose(mg): q200 q400

109

EDPrescription:OralCephalexin(Keflex)? qYes qNo


OralAmoxicillin-Clavulanate? qYes qNo

IfYes,Dose(mg): q500 q875 Frequency: qBID qTID qQID Duration(days):____________________

OralAmoxicillin? qYes qNo


OralClindamycin? qYes qNo


OralTMP-SMX(Septra)? qYes qNo

IfYes,Dose(mg): q800(1DStab) q1600(2DStabs)Frequency: qOD qBID qTID Duration(days):____________________

OralCiprofloxacin? qYes qNo

IfYes,Dose(mg): q250 q500 q1000Frequency: qOD qBID qTID Duration(days):____________________

OralDoxycycline? qYes qNo

IfYes,Dose(mg): q100 q200 Frequency: qOD qBID qTID

Duration(days):____________________

110

IVCefazolin(Ancef)? qYes qNoIfYes,Dose(mg): q500 q1000 q1500q2000

Frequency: qBID qTID qQID Duration(days):____________________

IVCeftriaxone? qYes qNo

IfYes,Dose(mg): q1000 q1500 q2000Frequency: qOD qBID Duration(days):___________________

IVClindamycin? qYes qNo


IVMeropenem? qYes qNo


IVVancomycin? qYes qNo

IfYes,Dose(mg): q500 q1000 q1500q2000Frequency: qOD qBID qTID Duration(days):___________________

IVPiperacillin-Tazobactam? qYes qNo

IfYes,Dose(mg): q3375 q4500 Frequency: qBID qTID qQID Duration(days):___________________

IVCiprofloxacin? qYes qNo

IfYes,Dose(mg): q200 q400Frequency: qBID qTID qQID Duration(days):___________________

111

ReturntoED(RTED)within14days: oYes oNo Ifyes(chooseONEonly):

oReturntoEDforSSTIandhospitaladmission oReturntoEDforSSTIbutnoadmission oReturntoEDwithabscess(Ifyes,I&Dperformed? oYes oNo) oReturntoEDforrepeatantibiotics:numberofsubsequentvisits_____oReturntoEDforunrelatedmedicalproblem

IfRTED:ChangestotreatmentinEDatrepeatvisit: oYes oNoIfyes(chooseONEonly):oChangefromPOtoIVantibiotics(nameofnewantibiotic:_______________)oChangetodifferentPOantibiotic(nameofnewantibiotic:_______________) oChangetodifferentIVantibiotic(nameofnewantibiotic:_______________)oChangetodifferentantibioticdose(newdose:________________mg)Iftreatmentwasaltered,reasonforchangeintreatmentplan:

oWorseninginfection oNoimprovementoVomiting oDiarrhea oRash/hivesoOther Indicate:_____________________

OralAntibioticTreatmentFailure? qYesqNo Ifyes,reasonfortreatmentfailure(checkallthatapply): qStep-upfromPOtoIVantibiotic qChangeinclassofPOantibiotic(duetoworseninginfection) qInitialoutpatientmanagementwithsubsequenthospitaladmissionOPATTreatmentFailure? qYesqNo

Ifyes(i.e.hospitalizedfrom48hours–2weeks),reasonfortreatmentfailure(checkallthatapply):

qWorseninginfection qDrug-relatedcomplication qLine-relatedcomplicationAdverseEvents: Adversedrugreaction: qYesqNo Ifyes: qRash qNauseaonly(novomiting) qVomiting qDiarrhea qOther:____________________ Adversedeviceevent: qYesqNo Ifyes: qBlockedperipheralIV qDislodgedperipheralIV qLinethrombosis qLineinfection

Subsequenthospitaladmissionat14daysfromindexEDvisit?qYesqNo Ifyes: qForworseningSSTI

qForunrelatedmedicalproblem(specify:_______________)

112

OPATClinic:DidpatientattendOPATappointment? qYesqNo(STOPHERE)Cellulitiswascorrectdiagnosis: qYesqNo IfNO–specifyalternatediagnosis: qDVT qLymphedema qStasisdermatitis qAbscess qHematoma qDrugrash NumberofdaysfrominitialEDvisittofirstOPATappointment:__________TotalnumberofvisitstoOPATclinic:__________TotalnumberofdaysonIVantibiotics:__________TotalnumberofdaysonIVandPOantibiotics:__________IVAntibioticsfor>7d? qYesqNoIVAntibioticsfor>10d? qYesqNoTimetostep-down(days)fromIVtoPOantibiotics:__________

RationaleforIVantibiotics(checkallthatapply):

qFailedoralantibiotictherapy

qHypotension

qSeverepain(>8/10)

qClinicalimpression(severeinfection)

qDiabetes

qPainoutofproportion

qPeripheralvasculardisease

qIndwellingIVcatheter

qBite

qSocial/complianceissues

qOther:______________________________________________________________________Patientagreedtotelephonecall? oYes oNo oLosttofollow-upPatientexpectedtobereferredtoanoutpatientclinicformanagementofinfection?oYes oNoSettingpatientwouldpreferforreceivingfollow-upinthefuture?oOPATclinic oED oFamilydoctorSatisfactionwiththetimelinessofthereferraltotheOPATclinic._____/10OverallsatisfactionwithcareintheOPATclinic:_____/10

113

TableS1.ReasonsforHospitalizationwithin14DaysofEDVisitfor137PatientswhoattendedtheirOPATClinicAppointment

ReasonforHospitalization Numberofpatients,N(%)N=137

WorseninginfectionOthermedicalproblemAcutecoronarysyndromeRenalfailureGoutHipfracture

6(4.4)

4(2.9)1111

114

ChapterFive:Discussion

IntroductionThegoalofthisthesiswastodescribetheepidemiologyofadultswithnon-purulent

SSTIswhopresenttotheED.Additionalgoalswereto:(a)identifypredictors

associatedwithoralantibiotictreatmentfailure;(b)determineemergency

physicianrationaleforselectingintravenoustherapy;and(c)describethe

performanceofanED-to-OPATclinicprogram.Thischapterwillsummarizethe

resultsofthehealthrecordsreview(ChapterThree)andtheprospective

observationalcohortstudy(ChapterFour)inordertohighlightclinical

recommendationsandareasforfutureresearch.

InterpretationofResultsThehealthrecordsreview(ChapterThree)describedtheepidemiologyofadults

withnon-purulentSSTIsseekingcareintheED.Therewasasurprisinglyhigh

hospitaladmissionrate,whichinpartreflectstheneedformorerobustevidence-

basedguidelinestoimprovemanagementofthiscommonclinicalcondition.Many

patientshaddiabetesandcellulitiswithinthepreceding12months.Wedidobserve

alargevariationinselectionoftheantimicrobialagentandroute.Themajorityof

patientsreceivedatleastonedoseofintravenousantibioticintheED.Despitealack

ofevidencetosupportthispractice,manypatientsreceivedasingleintravenous

doseandweresenthomewithanoralantibioticprescription.Theobserved

variabilityintreatmentapproachhighlightsaclearlackofagreementamongst

emergencyphysiciansregardingoptimalantimicrobialtherapyforthiscommon

115

clinicalcondition.Duetothestudydesign,thereasonsbehindselectingintravenous

therapyremainedunclear.

Wefoundanoralantibiotictreatmentfailureratethatwashigherthanexpectedin

thehealthrecordsreviewwhencomparedtopriorstudies.1,2Thesepriorstudies

didnotuseastricttimecutofftodefinetreatmentfailure,whichmayhaveledto

somepatientsbeingmisclassifiedastreatmentfailureorsuccess.Weused

multivariablelogisticregressiontoidentifyfactorsassociatedwithoralantibiotic

treatmentfailure.Adetaileddescriptionofthemethodsusedtodevelopa

multivariablelogisticregressionmodelcanbefoundinAppendixB.Wefirstused

clinicalreasoningtodeterminewhichvariablescouldplausiblypredictoral

antibiotictreatmentfailure.Ultimately,sevenvariableswerechosenforinclusion

intoamultivariablemodel.Fourvariableswerefoundtobeassociatedwiththe

primaryoutcome:tachypneaattriage;chroniculcers;ahistoryofMRSA

colonizationorinfection;andpriorcellulitisinthelast12months.Thisisthefirst

studytoidentifypotentialpredictorsassociatedwithoralantibiotictreatment

failure.

Toaddresstheissuesweidentifiedsurroundingintravenousantibioticuse,we

conductedaprospectiveobservationalcohortstudy(ChapterFour)to:(a)describe

theperformanceofanED-to-OPATclinicmodel;and(b)identifyphysicianrationale

forselectingintravenoustherapy.WefoundlowratesofOPATtreatmentfailure.

Noneoftheadverseperipheralintravenouslineeventsoradverseantibioticevents

116

wasseriousenoughtorequirehospitalization.Thediagnosisofnon-purulentSSTIs

wasconcordantbetweenemergencyphysiciansandinfectiousdiseasespecialistsin

themajorityofcases.Patientsreportedahighdegreeofsatisfactionwithtimeliness

ofreferralandoverallcarereceived.OurfindingssuggestthatanED-to-OPATclinic

modelissafeandacceptabletopatients.

Therewassignificantemergencyphysicianpracticevariationintheprospective

observationalcohortstudy.Themostcommonlyprescribedintravenousantibiotic

wascefazolin;physicianswerealmostevenlysplitregardingthechosendose.The

topfivereasonsforselectingtheintravenousroutewere:clinicalimpression;failed

oralantibiotictherapy;diabetes;severepain;andperipheralvasculardisease.Oral

antibioticfailurewasthesecondmostcommonreasonforselectingtheparenteral

route,whichisinterestinggiventhelackofavalidateddefinitionoftreatment

failure.Thepresenceofcomorbidities(diabetesandperipheralvasculardisease)as

rationaleforintravenoustherapywasnoteworthygiventhatneitherwasfoundto

beassociatedwithoralantibioticfailureinthehealthrecordsreview.

PreviousStudiesThe2014InfectiousDiseaseSocietyofAmericaguidelinesarelargelybasedon

expertopinion,andsuggestintravenousantibioticsfor‘moderate’(signsofsystemic

illness)or‘severe’(failedoraltherapy,signsofsystemicillness,clinicalsignsof

deeperinfection,orimmunocompromised)infections.3TheBritishClinicalResource

EfficiencySupportTeam(CREST)guidelinesrecommendoralantibioticsin‘ClassI’

117

patients,whichisdefinedasalackofsystemicsignsofillnessandtheabsenceof

‘uncontrolledcomorbidities’.4Thelatterwasnotexplicitlydefined.Thereare

currentlynoCanadianguidelinesforthisimportanttopic.Petersonandcoworkers

identifiedpredictorsoftreatmentfailurewithoutpatientantibiotictherapy,butdid

notdistinguishbetweenoralversusintravenousroutes.2Despitethelackof

evidence,arecentnationalsurveyofCanadianemergencyphysiciansfoundthat

94.4%ofrespondentswouldconsideraclinicaldecisionruletopredictoral

antibiotictreatmentfailure.5

TheobservedlowOPATtreatmentfailurerateintheprospectiveobservational

cohortstudywassimilartopreviouslypublishedstudies.6-8Currentguidelineslist

patientsatisfactionasacriticalelementofanOPATprogram.9-11Wefoundahigh

degreeofpatientsatisfactionthatissimilartopreviousstudies.Anadditionalaimof

ourstudywastoanswertheclinicalquestion:whydoemergencyphysiciansselect

theintravenousrouteforantimicrobialtherapy?Wefeltthatthiswasanimportant

clinicalquestion,asappropriatepatientselectionforoutpatientintravenous

therapyisacriticalfirststeptothesuccessofanOPATprogram.9,12Thereareno

previouslypublishedprospectivestudiesthathaveexaminedphysicianrationalefor

selectingtheintravenousroute.

118

StrengthsThehealthrecordsreviewisthefirststudytoidentifypotentialpredictors

associatedwithoralantibiotictreatmentfailurefornon-purulentSSTIs.Weused

strictinclusionandexclusioncriteria,predefinedvariables,astandardizedcase

recordform,andheldregularmeetingsbetweendataabstractorstoresolve

outstandingissuesandtoperformappropriateoversight.Thesespecificstepswere

takeninordertominimizebias,inaccordancewithacceptedmethodologyforchart

reviews.13-16Therewasexcellentagreementamongdataabstractorsforbothstudy

inclusionandtheprimaryoutcome.

Fortheprospectiveobservationalcohortstudy,weenrolledaconsecutivesampleof

patientsandscreenedallOPATclinicreferralstoensurenoeligiblecaseswere

missed.Thedatawerecollectedprospectively,whichallowedustoaccurately

identifyemergencyphysicianrationaleforselectingtheintravenousroute.Thisis

thefirststudytoprovideimportantinsightintophysiciandecision-making

regardingselectionofintravenousantibiotictherapy.

LimitationsThehealthrecordsreviewhasseveralpotentiallimitations.First,wefeltthatthe

sizeofaninfection(areaoferythema)couldbeaclinicallyimportantmarkerof

infectionseverity.However,wesuspectedthatmostemergencyphysiciansmight

notaccuratelydocumentinfectionsizeoromitthisinformation.Inordertomitigate

thisproblemforthehealthrecordsreview,wedevelopedanapriorimethodto

estimateinfectionsizeusingtotalbodysurfaceareaextrapolatedfromaLund-

119

Browderburndiagram.17,18Giventhelimitationsofthestudydesign,thisprovided

thebestestimateofinfectionsize.Second,obesityasacomorbiditymayhavenot

beenconsistentlydocumentedinthechart.Toaddressthisissue,wereviewed

electronichealthrecordsinthesixmonthspriortoandaftertheindexEDvisitto

determineifobesitywasdocumentedelsewhere.

Third,thedataabstractorswerenotblindedtotheprimaryoutcomeoforal

antibiotictreatmentfailure.However,weusedadefinitionoforalantibiotic

treatmentfailurethatincorporatedaclearminimumof48hoursoforaltherapy

beforeapatientcouldbeconsideredapossibletreatmentfailure.Therewasalso

excellentinterobserveragreementfortheprimaryoutcome.Therefore,wefeelthe

lackofblindingofthestudyabstractorswouldnothaveintroducedsignificantbias.

Afourthpotentiallimitationofthehealthrecordsreviewwasoverfittingofdata

duetothenumberofvariablesconsideredaspossiblepredictorsoforalantibiotic

treatmentfailure.Weinitiallyinvestigated18variablesusingunivariateanalysis

andidentifiedsevenvariablesthatwereassociatedwiththeprimaryoutcome.

Thesesevenvariableswerethenenteredintoamultivariablemodel.However,this

wasahealthrecordsreviewintendedtohelpidentifypotentialpredictors.Basedon

theresultsofthehealthrecordsreview,weintendtoassessasmallernumberof

variablesinafutureprospectivecohortstudy.

120

Alimitationofbothstudies(healthrecordsreviewandprospectiveobservational

cohortstudy)isthelackofavalidateddefinitionfororalorintravenousantibiotic

treatmentfailure.Wefirstreviewedtheliterature1,2,5,19-24andthendeveloped

treatmentfailuredefinitionsafterdiscussionamonglocalexpertsinemergency

medicineandinfectiousdisease.

Fortheprospectiveobservationalcohortstudy,thereareadditionallimitationsthat

deservemention.First,thesamplesizewaslimitedduetofeasibility.Second,some

patientsdidnotattendtheirclinicappointment.Itispossiblethatsomeofthese

patientslosttofollow-upexperiencedatreatmentfailureandsoughtcareat

anotherhospital.However,wefeelitwouldbeunlikelyforpatientstovisita

hospitaldifferentfromthecentrethatinitiatedtheircareforaninfection.Third,it

wasnotfeasibletoaccuratelyweighpatientstodeterminebodymassindexand

degreeofobesity.Whileitispossiblethattheobservedvariationinantibioticdosing

maybeinpartduetopatientbodyhabitus,thiswouldbeaminorissueasonlya

smallpercentageofpatientswererecordedasobeseinthisstudycohort.

ClinicalImplicationsTheresultsfromthehealthrecordsreviewhighlightimportantclinicalimplications

giventhepracticevariationwithrespecttoantibioticagent,dose,frequencyand

route.Inaddition,wefoundanunexpectedlyhighhospitaladmissionrateandoral

antibiotictreatmentfailurerate.Thesefindingsarelikelyduetoalackofevidence-

basedrecommendationstoguidecliniciansonoptimaltreatmentofthiscommon

121

clinicalcondition.Weidentifiedriskfactorsassociatedwithoralantibiotic

treatmentfailure.Thesefactorsshouldbeconsideredbyemergencyphysicians

whendecidingontheoralversusintravenousrouteofantibiotictherapy;thismay

inturndecreaseoveruseofmoreinvasiveandcostlyintravenoustreatment.

Theprospectiveobservationalcohortstudyalsoidentifiedalackofconsensus

regardingantibioticprescribingpracticesamongemergencyphysicians.Itisthe

firststudytoidentifyemergencyphysicianrationaleforselectingintravenous

therapyfornon-purulentSSTIs.Thisisanimportantfirststeptobetterunderstand

physiciandecision-makingwhenselectingtheintravenousroute.TherateofOPAT

treatmentfailurewaslow,andnoneoftheadverseantibioticeventsoradverse

peripheralintravenouslineeventsrequiredhospitalization.AnED-to-OPATclinic

modelwasfoundtobeeffective,safeandresultsinahighdegreeofpatient

satisfaction.Thismodelmaybeintroducedinothercommunitiestopotentially

reducehospitaladmissionsandoverallhealthcarecosts.

Tosummarize,basedonthetwostudiesconducted,wecanmakethefollowing

clinicalrecommendations:

1. Emergencyphysiciansshouldconsidercertainfactors(tachypnea,chronic

legulcers,ahistoryofMRSAcolonizationorinfection,andpriorcellulitisin

thepast12months)whendecidingonintravenous(overoral)antibiotic

therapyfornon-purulentSSTIs.

122

2. AnED-to-OPATclinicmodelissafeandresultsinahighdegreeofpatient

satisfaction.Thismodelmaybeimplementedinothercommunitiestohelp

reducehospitaladmissionsandoverallhealthcarecosts.

ResearchImplicationsThefindingsfromthisthesisprojecthavehighlightedimportantissuesthatweaim

toaddressinfuturestudies:

1. Developandvalidatearobustdefinitionfororalandintravenousantibiotic

treatmentfailureusingtheDelphimethod.Thesedefinitionsmaybeusedin

futurestudiesonthistopic.

2. Conductaprospectiveobservationalcohortstudytoderiveamodelto

identifyfactorsassociatedwithoralantibiotictreatmentfailure.Wewill

selectafewernumberofpotentialpredictorvariablesaprioribasedonthe

resultsofthisthesisproject.Thismodelwillthenbevalidatedinaseparate

EDpopulation.

3. Designandimplementapilotrandomizedcontrolledtrialcomparingoral

versusintravenousantibioticsinthetreatmentofadultpatientswithnon-

purulentSSTIs.

PriorED-basedstudieshaveuseddifferenttreatmentfailuredefinitionsasthe

primaryoutcome.1,2,22Futurestudieswouldbemoreimpactfulbyincorporatinga

uniformandrobusttreatmentfailuredefinition.WeplantousetheDelphimethod

todevelopandvalidatedefinitionsoforalandintravenousantibiotictreatment

123

failure.TheDelphimethodisasystematicapproachthathasbeenusedtodevelop

clinicaldefinitions25andguidelines26basedonexpertopinion.Consensuswillbe

definedapriorias≥75%,whichisthemostcommonthresholdusedinpriorDelphi

studies.27Relevantstakeholderswillincludeemergencyphysicians,infectious

specialists,pharmacistsandpatients.TheDelphiprocesswillbeconductedin

sequentialroundsuntilconsensushasbeenachieved.Structureddefinitionsfororal

andintravenousantibiotictreatmentfailureforSSTIsmaybeutilizedinfuture

studies.

ChapterThreeidentifiedpossiblepredictorsassociatedwithoralantibiotic

treatmentfailure.Weplantoconductaprospectiveobservationalcohortstudyto

identifypredictorsassociatedwithoralantibiotictreatmentfailure.Theresearch

teamwillrecordinfectionsizeandbodymassindex(measureofobesity)foreligible

patients,andbothvariableswillbeassessedaspotentialpredictorsoftreatment

failure.InChapterFour,severalphysicianrationalesforselectingtheintravenous

routewereidentified,includingdiabetesandperipheralvasculardisease.Forour

plannedprospectivecohortstudy,afewernumberofpredictorvariableswillbe

selectedbasedontheresultsofthehealthrecordsreview(ChapterThree),

prospectiveobservationalcohortstudy(ChapterFour)andexpertopinionfrom

emergencyphysicianandinfectiousdiseasespecialists.Fortheprimaryoutcome,

wewillusethedefinitionoforalantibiotictreatmentfailuredevelopedfromour

plannedDelphistudy.Thisstudywillmoreaccuratelyidentifywhichfactorspredict

oralantibiotictreatmentfailure,andmayhelpemergencyphysiciansdecidewhich

124

patientsrequireintravenousversusoralantimicrobialtherapy.Thismayhelplimit

overuseofintravenousantibioticsandlowerratesofadverseeventsandoverall

healthcarecosts.

Ultimately,weaimtoconductafuturerandomizedcontrolledtrialcomparingoral

versusintravenousantibiotictherapyfornon-purulentSSTIs.Beforeconducting

suchalarge-scalestudy,weintendtodesignandimplementapilotrandomized

controltrialtoassessfeasibility.EligibleadultEDpatientswithnon-purulentSSTIs

wouldberandomizedtoreceiveeitheroralcephalexinorintravenouscefazolin,and

theprimaryoutcomewouldbeoralorintravenousantibiotictreatmentfailure.

ConclusionsThisthesisprojecthasaccomplishedspecificgoalsregardingnon-purulentSSTIs

thataremanagedintheED.Ourfirststudy(healthrecordsreview)describesthe

managementofallpatientswithnon-purulentSSTIspresentingtotheED.Thestudy

findingshighlighttheheterogeneityinantimicrobialtreatmentrouteandagent.

Furthermore,weobservedbothanunexpectedlyhighhospitaladmissionrateand

oralantibiotictreatmentfailurerate.Ultimately,wewereabletodevelopa

multivariablelogisticregressionmodeloffactorsassociatedwithoralantibiotic

treatmentfailure.Oursecondstudy(prospectiveobservationalcohortstudy)

showedthatanED-to-OPATclinicmodelisbothsafeandresultsinhighpatient

satisfaction.Wealsoidentifiedphysicianrationalesforselectingtheintravenous

route.

125

Theresultshavehighlightedkeyareasinwhichfuturestudiesarerequiredtobetter

informemergencyphysiciansregarding:(a)optimaltherapyofthiscommonclinical

condition;and(b)appropriatepatientselectionforoutpatientintravenoustherapy.

Boththeidentifiedfactorsassociatedwithoralantibioticfailureandemergency

physicianrationalesforintravenoustherapydeservefuturestudy.Morerobust

evidenceregardingwhichfactorstrulypredictoralantibioticfailurewouldhelp

emergencyphysiciansmoreappropriatelyselectpatientsforintravenoustherapy

andreducetheincidenceoftreatmentfailure.Weareoptimisticthatthiswillinturn

resultinbetterpatientcare,animportantimprovementintheappropriateuseof

intravenousantibiotictherapy,andareductioninhealthcarecosts.

126

References1. MurrayH,StiellI,WellsG.Treatmentfailureinemergencydepartment

patientswithcellulitis.CJEM.2005;7(4):228-234.








9. TiceAD,RehmSJ,DalovisioJR,etal.Practiceguidelinesforoutpatientparenteralantimicrobialtherapy.IDSAguidelines.ClinInfectDis.2004;38(12):1651-1672.Epub2004May1626.

10. ChapmanAL,SeatonRA,CooperMA,etal.Goodpracticerecommendationsforoutpatientparenteralantimicrobialtherapy(OPAT)inadultsintheUK:aconsensusstatement.JAntimicrobChemother.2012;67(5):1053-1062.doi:1010.1093/jac/dks1003.Epub2012Jan1031.


127

12. GilchristM,SeatonRA.Outpatientparenteralantimicrobialtherapyandantimicrobialstewardship:challengesandchecklists.JAntimicrobChemother.2015;70(4):965-970.doi:910.1093/jac/dku1517.Epub2014Dec1023.

13. GilbertEH,LowensteinSR,Koziol-McLainJ,BartaDC,SteinerJ.Chartreviewsinemergencymedicineresearch:Wherearethemethods?AnnEmergMed.1996;27(3):305-308.

14. BadcockD,KellyAM,KerrD,ReadeT.Thequalityofmedicalrecordreviewstudiesintheinternationalemergencymedicineliterature.AnnEmergMed.2005;45(4):444-447.

15. LowensteinSR.Medicalrecordreviewsinemergencymedicine:theblessingandthecurse.AnnEmergMed.2005;45(4):452-455.

16. KajiAH,SchrigerD,GreenS.Lookingthroughtheretrospectoscope:reducingbiasinemergencymedicinechartreviewstudies.AnnEmergMed.2014;64(3):292-298.doi:210.1016/j.annemergmed.2014.1003.1025.Epub2014Apr1018.

17. HettiaratchyS,DziewulskiP.ABCofburns:pathophysiologyandtypesofburns.BMJ.2004;328(7453):1427-1429.






128



25. Shankar-HariM,PhillipsGS,LevyML,etal.DevelopingaNewDefinitionandAssessingNewClinicalCriteriaforSepticShock:FortheThirdInternationalConsensusDefinitionsforSepsisandSepticShock(Sepsis-3).JAMA.2016;315(8):775-787.doi:710.1001/jama.2016.0289.

26. SunBC,ThiruganasambandamoorthyV,CruzJD.Standardizedreportingguidelinesforemergencydepartmentsyncoperisk-stratificationresearch.AcadEmergMed.2012;19(6):694-702.doi:610.1111/j.1553-2712.2012.01375.x.

27. DiamondIR,GrantRC,FeldmanBM,etal.Definingconsensus:asystematicreviewrecommendsmethodologiccriteriaforreportingofDelphistudies.JClinEpidemiol.2014;67(4):401-409.doi:410.1016/j.jclinepi.2013.1012.1002.

129

AppendixA:OttawaHealthScienceNetworkResearchEthicsBoardApprovalLetter

130

AppendixB:MethodsforDevelopingaMultivariableLogisticRegressionModelforPredictorsofOralAntibioticTreatment

Failure

IntroductionWeconductedahealthrecordsreviewof500adultemergencydepartment(ED)

patientswithnon-purulentskinandsofttissueinfections(SSTIs).Ouraimwasto

identifypredictorsassociatedwiththeprimaryoutcomeoforalantibiotictreatment

failure.Ofthetotalcohortof500patients,288patientsweretreatedwithoral

antibioticsand85patientsexperiencedanoralantibiotictreatmentfailure.

ExploratoryAnalysisTableB1summarizesthecharacteristicsofthe288patientstreatedwithoral

antibioticsbasedonwhetherornottheyexperiencedatreatmentfailure.For

continuousvariables,weplottedhistogramstoassessthedistributionofthe

variables.Age,systolicbloodpressure,heartrateandtemperaturewere

approximatelynormallydistributed.Respiratoryrate,oxygensaturation,white

bloodcellcountandareaoferythemawerenon-normallydistributed.Comparing

thepatientswhodidanddidnotexperienceanoralantibiotictreatmentfailure,

categoricalvariableswereassessedusingthechi-squareorFisher’sexacttest,and

continuousvariableswereassessedusingt-testsfornormallydistributedvariables

andWilcoxontestsfornon-normallydistributedvariables.

131

OutliersTherewasasingleoutlierforeachofthefollowingvariables:whitebloodcellcount

(75×109/L),respiratoryrate(40breaths/minute)andoxygensaturation(82%).

Thepatientwithahighwhitebloodcellcounthadleukemia.Noneoftheseoutliers

arephysiologicallyimpossible,andsothesewerenotremoved.

MissingDataThevariableareaoferythemawasmissingin88.2%ofcases.Becausewefeltthat

infectionsizemaybeapotentiallyimportantpredictorfortheprimaryoutcome,we

estimatedthepercenttotalbodysurfaceareaforallcasesusingaLund-Browder

burndiagram.1,2Whitebloodcellcountwasmissingin24.4%ofcases.Allremaining

continuousvariableshadlessthan4%missingdata.

ExcludingVariablesThevariableareaoferythemawasexcludedbecauseofasignificantproportionof

missingdata.Thevariableswhitebloodcellcountandoxygensaturationwere

excludedbecausewedetermineditdidnotmakeclinicalsensethateithervariable

wouldbeassociatedwithtreatmentfailure.Fourvariableshadasparsedistribution

(i.e.lowfrequency(<1%)cellsoncrosstabsofthevariablesbyoralantibiotic

treatmentfailure):bite,HIV,injectiondruguseandorgantransplantrecipient;we

removedthesevariablesforconsiderationaspotentialpredictorsinourmodel.

Thuswewereleftwith18variables(5continuous,13categorical)thatwefeltcould

plausiblybeassociatedwithoralantibiotictreatmentfailure.

132

TableB1.UnivariateAssociationofCharacteristicswithTreatmentFailureoftheStudyParticipants(N=288)

Variable(N)

OralTreatmentFailure


NoOralTreatmentFailureFrequency,n(%)

(N=203)

PValue*

Male(158) 48(56.5) 110(54.2) 0.72Chroniculcers(20) 13(15.3) 7(3.4) 0.0003Surgicalsiteinfection(20) 7(8.2) 13(6.4) 0.58Bite(8) 1(1.2) 7(3.4) 0.44>5%TBSA(39) 14(16.5) 25(12.3) 0.35Coronaryarterydisease(29) 12(14.1) 17(8.4) 0.14Congestiveheartfailure(18) 8(9.4) 10(4.9) 0.15Chronickidneydisease(15) 8(9.4) 7(3.4) 0.04Chronicvenousinsufficiency(8)

1(1.2) 7(3.4) 0.44

Diabetesmellitus(55) 23(27.1) 32(15.8) 0.03Liverdisease(17) 5(5.9) 5(5.9) 0.99HIV(2) 2(2.4) 0(0) 0.09Injectiondruguse(IVDU)(4) 3(3.5) 1(0.5) 0.08Lymphedema(14) 5(5.9) 9(4.4) 0.56HistoryofMRSAcolonizationorinfection(16)

11(12.9) 5(2.5) 0.001

Obesity(8) 3(3.5) 5(2.5) 0.69Priorcellulitisinpast12months(38)

19(22.4) 19(9.4) 0.003

Peripheralvasculardisease(17)

5(5.9) 12(5.9) 0.99

Age(mean±SD) 67.1±19.0 63.3±20.3 0.15Systolicbloodpressure(mean±SD)

140±22 138±24 0.69

Heartrate(mean±SD) 84±17 83±16 0.68Temperature(mean±SD) 36.6±0.8 36.3±0.7 0.01Respiratoryrate(median,IQR)

18(16–18) 16(16–18) 0.85

SaO2(median,IQR) 97(96–98) 97(96–98) 0.11WBCcount(median,IQR) 8.6(7.0–12.1) 8.2(6.4–11.1) 0.21TBSA=totalbodysurfacearea;HIV=humanimmunodeficiencyvirus;SaO2=oxygensaturation;WBC=whitebloodcell;SD=standarddeviation;IQR=interquartilerange(Q1–Q3)

133

AssociationsbetweencategoricalvariablesWeconsideredseveralcategoricalvariableswhereitwasclinicallyplausiblethat

theymaybecollinear.TableB2showsthedegreeofcollinearitybetweenthe

categoricalvariablesusingthePhicoefficient.3

TableB2.AssessmentofSimpleCollinearityBetweenCategoricalVariablesUsingthePhiCoefficient

Gender ChronicUlcers

%TBSA

DM Lymphedema MRSA Obesity PriorSSTI

PVD

Gender 0.05 -0.02 0.07 -0.13 0.0001

0.03 0.005 -0.005

ChronicUlcers

0.05 0.05 0.13 0.08 0.12 0.08 0.17 0.24

%TBSA -0.02 0.05 DM -0.07 0.13 0.16 0.05 0.12 0.15 0.12 0.22Lymphedema -0.13 0.08 0.17 0.05 0.005 0.15 0.11 0.01MRSA 0.0001 0.12 0.04 0.12 0.005 0.05 0.16 -0.01Obesity 0.03 0.08 -0.008 0.15 0.15 0.05 0.15 -0.005PriorSSTI 0.005 0.17 0.12 0.12 0.11 0.16 0.15 -0.02PVD -0.005 0.24 0.11 0.22 0.01 -0.01 -0.005 -0.02 DM=diabetesmellitus;TBSA=totalbodysurfacearea;MRSA=priormethicillinresistantStaphylococcusaureus(MRSA)colonizationorinfection;PriorSSTI=cellulitiswithinthepast12months;PVD=peripheralvasculardisease

InterpretationofthePhicoefficientwasasfollows:strongpositiveassociation(0.7

to1.0);weakpositiveassociation(0.3to0.7);littleornoassociation(-0.3to0.3);

weaknegativeassociation(-0.7to-0.3);andstrongnegativeassociation(-1.0to-

0.7).Therewerenostrongpositiveornegativeassociations.Becausetherewasno

highdegreeofcollinearity,wedidnotneedtoconsiderexclusionofanyofthe

categoricalvariablesinTableB2.

CorrelationbetweencontinuousvariablesInTableB3thecorrelationbetweencontinuousvariablesusingthePearson

correlationcoefficient(r)areprovided.ThesamerangesforthePhicoefficientwere

134

usedtointerpretthePearsoncorrelationcoefficients.Therewerenostrongpositive

ornegativelinearassociationsbetweenthecontinuousvariables,asnoneofthe

Pearsoncorrelationcoefficientsapproached+1or-1.

TableB3.CorrelationBetweenContinuousVariablesUsingPearsonCorrelationCoefficient(r)

Pearsonrcoefficient

Age SBP HR Temp RR SaO2

Age 1 0.12 -0.23 -0.001 0.07 -0.24SBP 0.12 1 -0.07 -0.05 -0.04 -0.01HR -0.23 -0.07 1 0.25 0.20 0.03Temp -0.008 -0.05 0.25 1 0.28 -0.08RR 0.07 -0.04 0.19 0.28 1 -0.08SaO2 -0.24 -0.01 0.03 -0.08 -0.09 1SBP=systolicbloodpressure;HR=heartrate;Temp=temperature;RR=respiratoryrate;SaO2=oxygensaturation

LogisticRegressionAnalysis

UnivariateAnalysisFollowingtheexploratoryanalysis,therewere18variables(5continuousand13

categorical)thatwefeltmeritedconsiderationaspotentialpredictorsbasedon

clinicalsensibility.

Forthe5continuousvariables,weconsideredclinicallymeaningfulcutoffsand

performedunivariatelogisticregression.Inaddition,wedidexplorearangeof

cutoffstodetermineiftherewereanytrends.Wedeterminedage≥65yearstobe

themostclinicallyrelevant,whichisthedefinitionemployedbythemostrecent

geriatricEDguidelines.4Forthetriagevitalsigns,thefollowingcutoffswere

selected:systolicbloodpressure<90mmHg(clinicallyrelevanthypotension);heart

135

rate≥100beatsperminute(clinicaldefinitionoftachycardia);respiratoryrate>20

breathsperminute(clinicaldefinitionoftachypnea);andtemperature≥38.0°C

(clinicaldefinitionoffever).Theunivariatelogisticregressionanalysisofthe

continuousvariablesatprespecifiedcutoffsaresummarizedinTablesB4–B8.

Thesecutoffswereselectedbasedonstandarddefinitionsandclinicalreasoning

andareshownwithvariousothercutoffstoshowanypotentialtrends.

TableB4.UnivariateLogisticRegressionforAgeasaPredictorofOralAntibioticTreatmentFailure

PredictorVariable

Estimate PValue OR 95%CI

Age≥50 0.40 0.21 1.49 0.80–2.75Age≥55 0.38 0.20 1.46 0.82–2.62Age≥60 0.35 0.21 1.42 0.82–2.43Age≥65* 0.21 0.42 1.24 0.74–2.06Age≥70 0.02 0.95 1.01 0.61–1.69Age≥75 0.15 0.57 1.16 0.69–1.95Age≥80 0.22 0.44 1.25 0.71–2.18Age≥85 0.25 0.44 1.29 0.67–2.46

*SelectedasthemostclinicallyrelevantcutoffOR=oddsratio;CI=confidenceintervalTableB5.UnivariateLogisticRegressionforSystolicBloodPressureasaPredictorofOralAntibioticTreatmentFailure

PredictorVariable


SBP<85 0.90 0.53 2.45 0.15–39.64SBP<90* 0.90 0.52 2.45 0.15–39.64SBP<100 0.04 0.95 1.04 0.26–4.14SBP<110 -0.36 0.49 0.70 0.25–1.96

*SelectedasthemostclinicallyrelevantcutoffSBP=systolicbloodpressure;OR=oddsratio;CI=confidenceinterval

136

TableB6.UnivariateLogisticRegressionforHeartRateasaPredictorofOralAntibioticTreatmentFailure

PredictorVariable


HR≥90 -0.14 0.61 0.87 0.50–1.50HR≥100* 0.33 0.35 1.39 0.70–2.77HR≥110 0.59 0.22 1.80 0.70–4.66HR≥120 0.90 0.16 2.46 0.69–8.74

*SelectedasthemostclinicallyrelevantcutoffHR=heartrate;OR=oddsratio;CI=confidenceintervalTableB7.UnivariateLogisticRegressionforTemperatureasaPredictorofOralAntibioticTreatmentFailure

PredictorVariable


T≥38.0* 1.14 0.09 3.13 0.82–11.98T≥38.5 2.01 0.08 7.44 0.76–72.6

*SelectedasthemostclinicallyrelevantcutoffT=temperature;OR=oddsratio;CI=confidenceintervalTableB8.UnivariateLogisticRegressionforRespiratoryRateasaPredictorofOralAntibioticTreatmentFailure

PredictorVariable


RR≥20 0.14 0.70 1.15 0.57–2.30RR>20* 1.78 0.004 5.95 1.78–19.91RR≥22 1.78 0.004 5.95 1.78–19.91RR≥24 3.05 0.004 21.19 2.60–172.3

*SelectedasthemostclinicallyrelevantcutoffRR=respiratoryrate;OR=oddsratio;CI=confidenceintervalAtourchosencutoffof65years,agewasnotassociatedwithoralantibiotic

treatmentfailure.ThesamewastrueforavarietyofothercutoffsasshowninTable

4.Hypotensionandtachycardiawerealsonotassociatedwiththeprimaryoutcome

(TablesB5andB6).However,feverandtachypneawereassociatedwiththe

137

primaryoutcome(TablesB7andB8),andthesevariableswereselectedforthe

multivariablemodel.

TableB9showstheunivariatelogisticregressionfor13categoricalvariables.We

foundfivevariablesthatwereassociatedwiththeprimaryoutcome(p-value≤0.10):

chroniculcers;chronickidneydisease;diabetesmellitus;ahistoryofMRSA

colonizationorinfection;andpriorcellulitisinthepast12months.

TableB9.SimpleUnivariateLogisticRegressionforCategoricalVariables

PredictorVariable Estimate PValue OR 95%CIMale 0.09 0.72 1.10 0.66–1.83

ChronicUlcers 1.62 0.0009 5.06 1.94–13.17Surgicalsiteinfection 0.27 0.58 1.31 0.50–3.41

TBSA>5% 0.34 0.35 1.40 0.69–2.86Coronaryarterydisease 0.59 0.14 1.80 0.82–3.95Congestiveheartfailure 0.70 0.16 2.01 0.76–5.27Chronickidneydisease 1.07 0.04 2.91 1.02–8.30Diabetesmellitus 0.68 0.03 1.98 1.08–3.65Lymphedema 0.30 0.60 1.35 0.44–4.14

HistoryofMRSAcolonizationorinfection

1.77 0.001 5.89 1.98–17.51

Obesity 0.37 0.62 1.45 0.34–6.20Priorcellulitisinlast

12months1.02 0.004 2.79 1.39–5.59

Peripheralvasculardisease -0.005 0.99 0.99 0.34–2.92TBSA=totalbodysurfacearea;MRSA=methicillinresistantStaphylococcusaureus;OR=oddsratio;CI=confidenceinterval

Insummary,followinganexploratoryanalysisandunivariatelogisticregression

analysis,weidentifiedsevenvariableswithap-value≤0.10thatwechosetoinclude

inamultivariablelogisticregressionmodel:

138

1. Fever(temperature≥38.0°C)

2. Tachypnea(respiratoryrate>20breaths/minute)

3. Chroniculcers

4. Chronickidneydisease

5. Diabetesmellitus

6. PastMRSAcolonizationorinfection

7. Priorcellulitisinthepast12months

Giventhattherewere85identifiedcasesoforalantibiotictreatmentfailure,testing

7variableswouldsatisfytheruleof10.5-7

AssessingforInteractionWhereitseemedclinicallyplausible,weassessedforpossibleinteractionbetween

thesevenpotentialpredictorvariables.Investigationforpotentialinteractionterms

issummarizedinTableB10.Noneoftheinteractiontermswassignificant(defined

asp≤0.05),andsonointeractiontermswerebroughtintoamultivariablemodel.

TableB10.AssessmentofPossibleInteractionTerms

PredictorVariable Estimate PValueDiabetes*CKD -0.26 0.82

ChronicUlcers*Diabetes 13.60 0.99ChronicUlcers*PVD 1.32 0.37

ChronicUlcers*Lymphedema 0.06 0.96Priorcellulitis*MRSAhistory 0.36 0.79Priorcellulitis*Lymphedema -1.86 0.17

Priorcellulitis*PVD 15.13 0.99Priorcellulitis*Diabetes -1.57 0.08

Fever*Tachypnea 13.66 0.99CKD=chronickidneydisease;PVD=peripheralvasculardisease;MRSA=methicillinresistantStaphylococcusaureus;PVD=peripheralvasculardisease

139

MultivariableLogisticRegression:PreliminaryModelWeperformedmultivariablelogisticregressionincludingsevenpotentialpredictor

variableswithp≤0.10intheunivariateanalysis.TableB11showsthepreliminary

model.

TableB11.MultivariableLogisticRegressionModelofPredictorsAssociatedwithOralAntibioticTreatmentFailure(N=288)

PredictorVariable Estimate PValue AdjustedOR 95%CITachypnea(RR>20) 1.64 0.01 5.15 1.41–18.86Chroniculcers 1.57 0.004 4.80 1.64–14.04HistoryofMRSA

colonizationorinfection1.56 0.008 4.78 1.49–15.31

Priorcellulitisinlast12months

0.76 0.06 2.14 0.96–4.74

Chronickidneydisease 1.08 0.08 2.94 0.89–9.70Diabetesmellitus 0.53 0.13 1.70 0.86–3.63Fever(T≥38.0) 0.82 0.34 2.26 0.42–12.24

TheHosmer-Lemeshowchi-squaretestyieldedap-valueof0.938(chi-squared=0.409,degreesoffreedom=3).C-statistic=0.710.

RR=respiratoryrate;MRSA=methicillinresistantStaphylococcusaureus;T=temperature;OR=oddsratio;CI=confidenceinterval

TableB12showsthebackwardsselectionlogisticregressionmodelincludingthe

samesevenvariableswithaninclusioncriteriasetasap-value≤0.15.The

backwardsselectionmodeldroppedonevariable(fever).

140

TableB12.MultivariableLogisticRegressionModelUsingBackwardsSelectionofPredictorsofOralAntibioticTreatmentFailure(N=288)



Chronickidneydisease 1.05 0.08 2.87 0.87–9.46Priorcellulitisinpast

12months0.78 0.06 2.17 0.98–4.82

Diabetesmellitus 0.55 0.11 1.74 0.88–3.43TheHosmer-Lemeshowchi-squaretestyieldedap-valueof0.611(chi-squared=

1.819,degreesoffreedom=3).C-statistic=0.710.RR=respiratoryrate;MRSA=methicillinresistantStaphylococcusaureus;T=temperature;OR=oddsratio;CI=confidenceinterval

Asafinalstep,theremaining6variablesfromthebackwardsselectionmodelwere

placedintoamultivariablelogisticregressionmodel.Thefinalmodelisshownin

TableB13.

TableB13.FinalMultivariableLogisticRegressionModelofPredictorsAssociatedwithOralAntibioticTreatmentFailure(N=288)



Priorcellulitisinpast12months

0.80 0.05 2.23 1.01–4.96

Chronickidneydisease 1.59 0.10 2.60 0.82–8.22Diabetesmellitus 0.53 0.12 1.70 0.87–3.32

TheHosmer-Lemeshowchi-squaretestyieldedap-valueof0.604(chi-squared=1.853,degreesoffreedom=3).C-statistic=0.709.

TheHosmer-Lemeshowgoodnessoffit(p=0.604)indicatesnoevidenceofpoorfit

andtheC-statistic(0.709)suggestthemodelhasgoodfit.Thevariablesthat

141

remainedsignificant(p≤0.05)inthemultivariablemodelwere:tachypnea,chronic

ulcers,historyofMRSAcolonizationorinfection,andpriorcellulitisinthepast12

months.Sincediabeticsarepronetodevelopinglowerextremityulcers8-10,asafinal

step,wecheckedforpossibleinteractionbetweendiabetesmellitusandchronic

ulcers.Thisyieldedap-value=0.99,indicatingnosignificantinteractionbetween

thesevariables.

SecondaryAnalysis(ComparingOralvs.IntravenousTreatmentGroups)Weconductedasecondaryanalysiscomparingtheoralantibiotictreatmentgroup

(N=288)andintravenousantibiotictreatmentgroup(N=212).Theresultsare

summarizedinTablesB14andB15.

142

TableB14.SecondaryAnalysisComparingOralversusIntravenousAntibioticGroupsforCategoricalVariablesforall500Patients

Variable(n)

OralAntibioticGroupFrequency,n(%)

(n=288)

IVAntibioticGroupFrequency,n(%)

(n=212)

PValue*

Male(279) 158(54.9) 121(57.1) 0.62

ChronicUlcers(56) 20(6.9) 36(17.0) 0.0004SurgicalSiteInfection(30)

20(6.9) 10(4.7) 0.30

Bite(12) 8(2.8) 4(1.9) 0.52

>5%TBSA(99) 39(13.5) 60(28.3) <0.0001

Coronaryarterydisease(58)

29(10.1) 29(13.7) 0.21

Congestiveheartfailure(48)

18(6.2) 30(14.1) 0.003

Chronickidneydisease(35)

15(5.2) 20(9.4) 0.07

Chronicvenousinsufficiency(10)

8(2.8) 2(0.9) 0.20

Diabetesmellitus(126)

55(19.1) 71(33.5) 0.0002

HepaticDisease(37) 17(5.9) 20(9.4) 0.14HIV(8) 2(0.7) 6(2.8) 0.08IVDU(14) 4(1.4) 10(4.7) 0.02Lymphedema(33) 14(4.9) 19(9.0) 0.07MRSAHistory(43) 16(5.6) 27(12.7) 0.005Obesity(27) 8(2.8) 19(9.0) 0.002Priorcellulitis(87) 38(13.2) 49(23.1) 0.004PVD(40) 17(5.9) 23(10.8) 0.04*UsingChi-SquaredorFisher’sExactTestIV=intravenous;TBSA=totalbodysurfacearea;HIV=humanimmunodeficiencyvirus;IVDU=intravenousdruguse;MRSA=methicillinresistantStaphylococcusaureus;PVD=peripheralvasculardisease

143

TableB15.SecondaryAnalysisComparingOralversusIntravenousAntibioticGroupsforContinuousVariablesforall500Patients

Variable(n)

OralAntibioticGroup


IVAntibioticGroup


PValue*

Age(mean±SD) 64.4±20.0 64.3±16.9 0.95SystolicBP(mean±SD) 139±23 131±25 0.0007Heartrate(mean±SD) 84±16 90±21 <0.0001Temperature(mean±SD) 36.4±0.8 36.8±1.2 <0.0001RespiratoryRate(median,IQR)

16(16–18) 18(16–20) 0.02

SaO2(median,IQR) 97(96–98) 97(96–98) 0.88WBCcount(median,IQR) 8.2(6.6–11.2) 10.4(7.7–14.7) <0.0001*Usingt-testsfornormallydistributedvariables,Wilcoxontestsfornon-normallydistributedvariablesSD=standarddeviation;IQR=interquartilerange;IV=intravenous;BP=bloodpressure;HR=heartrate;SaO2=oxygensaturation;WBC=whitebloodcell

Patientsweremorelikelytoreceiveintravenousantibioticsiftheyhadchronic

ulcers,largeareaofinvolvement(totalbodysurfacearea>5%),congestiveheart

failure,diabetes,injectiondruguse,ahistoryofMRSAcolonizationorinfection,

priorcellulitisinthepast12monthsandperipheralvasculardisease.Patients

treatedwithintravenousantibioticsalsohadahighermedianrespiratoryrateand

highermeansystolicbloodpressure,heartrateandtemperature.Theseresults

suggestthatemergencyphysiciansappeartobemorelikelytoadminister

intravenousantibioticstopatientswithtriagevitalsignsorcomorbiditiesthatwere

associatedwithtreatmentfailureinthefinalmodel(TableB13).

DiscussionOurstudyidentifiedpotentialriskfactorsassociatedwithoralantibiotictreatment

failureforadultspresentingtotheEDwithnon-purulentSSTIs.Thenumberof

144

variablesunderconsiderationwasreducedbyeliminatingvariableswith>20%

missingdataorwithsparsedistributions.Wechosetoonlyincludevariableswhere

itwasclinicallyplausiblethattheymaybeassociatedwiththeprimaryoutcome.We

didnotidentifyahighdegreeofcollinearitybetweenvariables.Followingthe

exploratoryanalysis,univariatelogisticregressionidentifiedsevenvariables

associatedwiththeprimaryoutcome.

Wethendevelopedafinalmultivariablelogisticregressionmodelthatincludedsix

variables.Fourvariableswereindependentlyassociatedwithoralantibiotic

treatmentfailure:tachypnea;chroniculcers;ahistoryofMRSAcolonizationor

infection;andpriorcellulitisinthepast12months.TheHosmer-Lemeshow

goodness-of-fitsuggestsnoevidenceofpoorfitandtheC-statisticindicatesthatthe

modelhasgoodfit.Wealsolookedatpossibleinteractionbetweenindependent

variablesbutdidnotidentifyanysignificantinteractionterms.

Oneimportantlimitationtoourapproachwasthatthedatamightbeoverfitted.

However,asthiswasahealthrecordsreview,thepurposewastotryandidentify

potentialassociationswithoralantibiotictreatmentfailure.Weplantoconduct

futureprospectivestudiesincorporatingfewervariablestodevelopamodel.

ConclusionToourknowledge,thisisthefirststudytoidentifypredictorvariables

independentlyassociatedwithoralantibiotictreatmentfailure.Emergency

145

physiciansshouldconsidertheseriskfactorswhenselectingtheappropriateroute

ofantimicrobialtherapyforapatientwithanon-purulentSSTI.Futureprospective

studieswillbeconductedthatassessfewervariablesbasedonthefindingsofthis

study.

146

References1. HettiaratchyS,DziewulskiP.ABCofburns:pathophysiologyandtypesof

burns.BMJ.2004;328(7453):1427-1429.


3. BreaughJ.Effectsizeestimation:factorstoconsiderandmistakestoavoidJournalofManagement.2003;29(1):79-97.

4. Geriatricemergencydepartmentguidelines.AnnEmergMed.2014;63(5):e7-25.doi:10.1016/j.annemergmed.2014.1002.1008.

5. ConcatoJ,PeduzziP,HolfordTR,FeinsteinAR.Importanceofeventsperindependentvariableinproportionalhazardsanalysis.I.Background,goals,andgeneralstrategy.JClinEpidemiol.1995;48(12):1495-1501.

6. PeduzziP,ConcatoJ,FeinsteinAR,HolfordTR.Importanceofeventsperindependentvariableinproportionalhazardsregressionanalysis.II.Accuracyandprecisionofregressionestimates.JClinEpidemiol.1995;48(12):1503-1510.

7. PeduzziP,ConcatoJ,KemperE,HolfordTR,FeinsteinAR.Asimulationstudyofthenumberofeventspervariableinlogisticregressionanalysis.JClinEpidemiol.1996;49(12):1373-1379.

8. SinghN,ArmstrongDG,LipskyBA.Preventingfootulcersinpatientswithdiabetes.JAMA.2005;293(2):217-228.doi:210.1001/jama.1293.1002.1217.

9. LevyL,ZeichnerJA.Dermatologicmanifestationofdiabetes.JDiabetes.2012;4(1):68-76.doi:10.1111/j.1753-0407.2011.00151.x.

10. ForbesJM,CooperME.Mechanismsofdiabeticcomplications.PhysiolRev.2013;93(1):137-188.doi:110.1152/physrev.00045.02011.

Documents

Non-Purulent Skin and Soft Tissue Infections in the ... · Non-purulent skin and soft tissue infections (SSTIs) describe infections of the superficial epidermis and dermis (erysipelas)