Risk of Ventricular Arrhythmia with Citalopram and Escitalopram

18 Pages·2016·0.51 MB·English

RESEARCHARTICLE
Risk of Ventricular Arrhythmia with
Citalopram and Escitalopram: A Population-
Based Study
ElenaQirjazi1,EricMcArthur2,DanielleM.Nash2,StephanieN.Dixon2,3,Matthew
A.Weir1,3,AkshyaVasudev4,5,RacquelJandoc2,LorneJ.Gula6,MatthewJ.Oliver7,
RonWald8,9,AmitX.Garg1,2,3*
1 DivisionofNephrology,DepartmentofMedicine,WesternUniversity,London,Ontario,Canada,2 Institute
forClinicalEvaluativeSciences,Toronto,Ontario,Canada,3 DepartmentofEpidemiology&Biostatistics,
a11111
WesternUniversity,London,Ontario,Canada,4 DivisionofGeriatricPsychiatry,DepartmentofPsychiatry,
WesternUniversity,London,Ontario,Canada,5 DivisionofClinicalPharmacology,DepartmentofMedicine,
WesternUniversity,London,Ontario,Canada,6 DivisionofCardiology,DepartmentofMedicine,Western
University,London,Ontario,Canada,7 DepartmentofMedicine,SunnybrookHealthSciencesCentre,
UniversityofToronto,Toronto,Ontario,Canada,8 KeenanResearchCentreintheLiKaShingKnowledge
InstituteofStMichael’sHospital,Toronto,Ontario,Canada,9 DivisionofNephrology,Departmentof
Medicine,UniversityofToronto,Toronto,Ontario,Canada
OPENACCESS *[email protected]
Citation:QirjaziE,McArthurE,NashDM,DixonSN,
WeirMA,VasudevA,etal.(2016)RiskofVentricular
Abstract
ArrhythmiawithCitalopramandEscitalopram:A
Population-BasedStudy.PLoSONE11(8):
e0160768.doi:10.1371/journal.pone.0160768
Background
Editor:HemachandraReddy,TexasTechnical
UniversityHealthSciencesCenter,UNITEDSTATES
Theriskofventriculararrhythmiawithcitalopramandescitalopramiscontroversial.Inthis
Received:April7,2016 studyweinvestigatedtheassociationbetweenthesetwodrugsandtheriskofventricular
Accepted:June14,2016 arrhythmia.
Published:August11,2016
Methods
Copyright:©2016Qirjazietal.Thisisanopen
accessarticledistributedunderthetermsofthe Weconductedapopulation-basedretrospectivecohortstudyofolderadults(meanage76
CreativeCommonsAttributionLicense,whichpermits years)from2002to2012inOntario,Canada,newlyprescribedcitalopram(n=137701)or
unrestricteduse,distribution,andreproductioninany
escitalopram(n=38436),comparedtothoseprescribedreferentantidepressantssertraline
medium,providedtheoriginalauthorandsourceare
orparoxetine(n=96620).Afterinverseprobabilityoftreatmentweightingusingapropen-
credited.
sityscore,thebaselinecharacteristicsofthecomparisongroupsweresimilar.Theprimary
DataAvailabilityStatement:Allrelevantdataare
outcomewasahospitalencounterwithventriculararrhythmiawithin90daysofanewpre-
withinthepaperanditsSupportingInformationfiles.
scription,assessedusinghospitaldiagnosticcodes.Thesecondaryoutcomewasall-cause
Funding:TheInstituteforClinicalEvaluative
mortalitywithin90days.
Sciences(ICES)isanon-profitresearchcorporation
fundedbyanannualgrantfromtheOntarioMinistry
ofHealthandLong-TermCare(MOHLTC).Partsof Results
thematerialinthecurrentreportarebasedondata
andinformationcompiledandprovidedbythe Citalopramwasassociatedwithahigherriskofahospitalencounterwithventricular
CanadianInstituteofHealthInformation(CIHI).The arrhythmiacomparedwithreferentantidepressants(0.06%vs.0.04%,relativerisk[RR]
researchwasconductedattheICESWesternfacility,
1.53,95%confidenceintervals[CI]1.03to2.29),andahigherriskofmortality(3.49%vs.
whichreceivesfinancialsupportfromtheAcademic
3.12%,RR1.12,95%CI1.06to1.18).Escitalopramwasnotassociatedwithahigherrisk
MedicalOrganizationofSouthwesternOntario,the
SchulichSchoolofMedicineandDentistryatWestern ofventriculararrhythmiacomparedwiththereferentantidepressants(0.03%vs.0.04%,RR
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 1/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
UniversityandtheLawsonHealthResearchInstitute. 0.84,95%CI0.42to1.68),butwasassociatedwithahigherriskofmortality(2.86%vs.
Theopinions,resultsandconclusionsreportedinthis 2.63%,RR1.09,95%CI1.01to1.18).
paperarethoseoftheauthorsandareindependent
fromthefundingsources.NoendorsementbyICES,
Conclusion
theOntarioMOHLTCorCIHIisintendedorshouldbe
inferred.Dr.AmitGargwassupportedbytheDr. Amongolderadults,initiationofcitalopramcomparedtotworeferentantidepressantswas
AdamLintonChairinKidneyHealthAnalytics.
associatedwithasmallbutstatisticallysignificantincreaseinthe90-dayriskofahospital
Researchpersonnelwhoworkedonthisprojectwere
supportedbytheLilibethCabertoKidneyClinical encounterforventriculararrhythmia.
ResearchUnit.
CompetingInterests:Dr.Garg’sinstitutionreceived
unrestrictedresearchfundingfromPfizer.Thisdoes
notaltertheauthors'adherencetoPLOSONE
policiesonsharingdataandmaterials.Theauthors Introduction
otherwisedeclarethattheyhavenorelevantfinancial
Selectiveserotoninre-uptakeinhibitors(SSRIs;e.g.,citalopram,escitalopram,paroxetineand
interests.
sertraline)arecommonlyprescribedantidepressants.[1–4]Citalopramandescitalopramhave
beenimplicatedinventriculararrhythmias,presumablybylengtheningtheQTintervalofthe
cardiaccycle.[5–18]TheFoodandDrugAdministration(FDA)andHealthCanadacaution
againsttheuseofcitalopramatdoses>20mg/dayinpatientsover65yearsofage).[19–22]
TheFDAwarningswerebasedonanunpublishedtrialof119patientsrandomizedtoplacebo
orcitalopram,demonstratinganincreaseinthecorrectedQTintervalwithcitalopram.[19]
Thesesafetywarningshavebeencontroversial,[23–25]withinconsistentfindingsinotherfol-
low-upstudies.[5,14,23,26–29]ManyofthesestudieswerelimitedbytheuseofQTprolonga-
tionratherthanventriculararrhythmiarisk,[14,26,28]ayoungpopulationcohort,[5,23,27,
28]lowstatisticalpower,[26]andnotaccountingforimportantconfoundingfactorsinthe
analysis.[29]Escitalopram(theSenantiomerofcitalopram)hasalsobeenassociatedwithQT
intervalprolongationandHealthCanadawarnsagainsttheuseof>10mg/dayofescitalopram
forpatients65yearsofageorolder.[5,7,14–15,28,30]Weconductedthislargepropensity
score-weightedpopulation-basedcohortstudyofolderadultstoinvestigatewhetherinitiating
citalopramorescitalopramintheoutpatientsettingisassociatedwithahigherriskofventricu-
lararrhythmia,comparedtoinitiatingsertralineorparoxetine(referentantidepressants).
Methods
DesignandSetting
Weconductedapopulation-basedretrospectivecohortstudyofolderadultsfromApril1,
2002toDecember31,2012inOntario,Canada,whohadreceivedanewoutpatientprescrip-
tionforcitalopram,escitalopram,sertralineorparoxetine(themostcommonlyprescribed
SSRIsinOntario).Ontariohasapproximately2millionresidents65yearsofageorolder,who
havefullcoverageforhospitalandphysicianservices,andprescriptiondrugs.[31]
Weuseddatasetsheldsecurelyinlinkable-fileswithoutanydirectpersonalidentifiers,and
analyzedattheInstituteforClinicalEvaluativeSciences(ICES).Patientinformationwasanon-
ymizedandde-identifiedpriortoanalysis.Thepre-specifiedprotocolwasapprovedbythe
ResearchEthicsBoardatSunnybrookHealthSciencesCentre(Toronto,Ontario,Canada).The
reportingofthisstudyfollowsguidelinesforobservationalstudies(seeS1Table).[32]
DataSources
Weascertainedpatientbaselinecharacteristics,druguseandoutcomedatausingrecords
fromeightlinkeddatabases.TheOntarioDrugBenefitdatabasecontainshighlyaccuraterec-
ordsforoutpatientprescriptionsdispensedtopatientsaged65yearsorolder(errorrateless
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 2/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
than1%).[33]TheOntarioRegisteredPersonsDatabaserecordsvitalstatistics,includingdate
ofdeath.TheCanadianInstituteforHealthInformation(CIHI)–DischargeAbstractData-
base,theCIHI—NationalAmbulatoryCareReportingSystemdatabase,andtheOntario
MentalHealthReportingSystemdatabasecontaindiagnosticandproceduralinformationon
allhospitalizations,emergencyroomandpsychiatricfacilityvisits.TheICESPhysicianData-
basereportsprescriberandspecialistreferraldata.TheOntarioHealthInsurancePlandata-
base(OHIP)includeshealthclaimsforphysicianservices,andtheCanadianOrgan
ReplacementRegisteridentifiespatientswithend-stagekidneydisease.Wehaveusedthese
databasespreviouslytoresearchadversedrugeventsandhealthoutcomes.[34–40]Theinfor-
mationobtainedwascomplete,exceptforneighbourhoodincomequintile(missingin0.3%of
patients)andprescriberspecialty(missingin12.7%ofpatients).
WeusedInternationalClassificationofDiseases9threvision(ICD9;pre-2002)and10threvi-
sion(ICD10;post-2002)codestoassessbaselineco-morbiditiesinthefiveyearspriortothe
receiptoftherelevantprescriptions(S2Table),inconcordancewithpriorstudies[34,36].We
assessedbaselinemedicationsandhealthcareuseinthe120daysand1yearpriortothedate
ofthenewSSRIprescription,respectively.
Patients
WeestablishedacohortofolderadultsinOntario,Canada,whoweredispensedanewoutpa-
tientprescriptionofatleast7daysforcitalopram,escitalopram,paroxetineorsertraline
betweenApril2002andDecember2012.Theprescriptiondatewasthecohortentrydate.
Patientswereseparatedintothreegroupsbasedontheirprescription:1)citalopram,2)escita-
lopramand3)referentantidepressants(paroxetineorsertraline).Paroxetineandsertraline
weregroupedtogetherastheyhavelowcardiactoxicity,andbothareprescribedforsimilar
indicationsascitalopramandescitalopram.[10,14–15,27]
Weexcludedfromanalyses:patientsintheirfirstyearofeligibilityforprescriptiondrug
coverage(age65)toavoidincompletemedicationrecords;thosewithantidepressantprescrip-
tionsinthe180dayspriortothecohortentrydatetoensurenewantidepressantuse;thosedis-
chargedfromhospitalinthetwodayspriortotheircohortentrydatetoensurenewoutpatient
prescriptions(patientscontinuingantidepressantsinitiatedinhospitalwouldhavetheiroutpa-
tientprescriptiondispensedonthesamedayorthedayafterhospitaldischarge);thosewitha
historyofventriculararrhythmia,cardiacarrestorimplantablecardiacdefibrillatortocapture
denovoarrhythmiceventsinfollowup;andthosewithnonstandarddailydrugdosestoensure
generalizabilitytousualcareandomitdataerrors.Patientswithmultipleeligiblestudydrug
prescriptionsenteredthecohortonceonthefirstprescription.
Outcomes
Weascertainedalloutcomeswithin90daysofthecohortentrydate,tomimicthedurationof
follow-upincorrespondingclinicaltrialsandtoavoidpotentialcrossoversbetweenthegroups
thatcouldoccurwithlongerfollowup.[5,14]Since,QTprolongationstartswithinhoursof
initiatingcitalopramorescitalopram,weexpecteddrug-relatedventriculararrhythmiasto
occursoonafterSSRIinitiation.[7,41–42]
Theprimaryoutcomewasatleastonehospitalencounter(emergencyroompresentationor
hospitaladmission)withventriculararrhythmia.Thesecondaryoutcomewasall-causemortal-
ity.DiagnosticcodesusedtoascertainoutcomesarelistedinS3Table(ICD10diagnosticcodes
wereusedtoassessventriculararrhythmiaandthiscodingsystemwasimplementedinCanada
in2002).Thesecodesareenteredintothedatabasesbytrainedpersonnelbasedonphysician-
recordeddiagnosesinpatients’medicalcharts.TheICD10codesforventriculararrhythmia
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 3/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
havenotbeenpreviouslyvalidated.However,theirsensitivityisexpectedtobelowasventricu-
lararrhythmiasfrequentlygoundetectedinroutinehealthcare(oftenoccuringoutsidehospital
settings,inunmonitoredpatients,orinthesettingofmulti-organmedicalillness).Previous
studiesassessingtheaccuracyofICD9andICD10codesforcardiacarrhythmia(ventricular
andsupraventricular)showapositivepredictivevalueexceeding80%.[43–46]Weperformed
anethics-approvedmanualreviewof202randomchartsinourregion,lookingathospital
encounters(emergencyvisitsoradmissions)withtheventriculararrhythmiacodesusedinthis
study,andconfirmedapositivepredictivevalueof92%(95%confidenceinterval[CI]87to
95%).All-causemortalitydataisaccuratelycodedinourdatasources,withasensitivityof
97.8%andspecificityof100%forthefindingofdeath.[47]
StatisticalAnalysis
Weusedinverseprobabilityoftreatmentweightsbasedonpropensityscorestoeliminatesys-
tematicdifferencesinthebaselinecharacteristicsofthecomparedgroupswhileretainingall
individualsintheanalysis.[48]Thepropensityscoresprovidedtheprobabilityofreceivinga
prescriptionfortheexposuredrug(citalopramorescitalopram)givenasetofmeasuredbase-
linecharacteristics.Scoreswerecalculatedusingmultivariablelogisticregressionmodelswith
48baselinecharacteristics(S4Table)–chosenbecauseoftheirpotentialinfluenceontheout-
comesorsegregationofpatientsbetweenthecomparedgroups.[49–51]Byapplyingpropensity
score-basedweightstothepatientsandoutcomes,wecreatedweightedgroupsthatwerewell-
balancedonallmeasuredbaselinecharacteristics.Wecomparedbaselinecharacteristics
betweenthegroupsusingstandardizeddifferences.Thismetricdescribesdifferencesbetween
thegroupmeansrelativetothepooledstandarddeviation—adifferencegreaterthan10%is
consideredmeaningful.[48,52]Wecalculatedrelativerisks(RR)with95%confidenceintervals
(CI)usinglog-binomialregressionmodelsaccountingfortheweights.
Wealsoevaluatedtheassociationforbothexposedgroups(citalopramandescitalopram)
withouroutcomesinpre-specifiedsubgroupsofpatients—definedbythepresenceorabsence
of:1)congestiveheartfailure,2)coronaryarterydisease,3)chronickidneydisease,and4)high
dose(S5Table).Wehypothesizedahigherriskinthepresenceoftheseconditions.Weidenti-
fiedchronickidneydiseaseusinganalgorithmofhospitaldiagnosticcodesvalidatedforolder
adultsinourregion.[53]Wedeterminedinteractionp-valuesbyincludinginteractiontermsin
theregressionmodels.Weinterpretedatwo-sidedp-valueoflessthan0.05asstatisticallysig-
nificant,andperformedallanalysisusingSASversion9.3(SASInstitute,Cary,North
Carolina).
Results
BaselineCharacteristics
Weidentified472001patientswithprescriptionsforthestudySSRIs.Afterapplyingourselec-
tioncriteria,wehad137701olderadultswithprescriptionsforcitalopram,38436forescitalo-
pramand96620forthereferentantidepressants(refertoS1Fig;lessthan10%ofpatients[40
015patients]wereexcludedfornon-standarddailydosesoftheSSRI).Themeanagewas76
yearsold(range66to105),and66%werewomen.Generalpractitionerswrote78%ofthepre-
scriptions.Thedistributionofbaselinecharacteristicsbeforeandafterpropensityscoreweight-
ingarepresentedinTable1forcitalopramandTable2forescitalopram.
Afterweighting,withexceptionofthedateofcohortentry,therewasnosignificantdiffer-
encebetweenthetwosetsofcomparisongroupsacrossallother77baselinecharacteristics
measuredinthisstudy(seeTables1and2).Theyearofcohortentrywasexpectedtobediffer-
entsinceescitalopramwasaddedtoOntario’sprovincialformularyin2008.Only6.5%ofthe
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 4/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
Table1. Baselinecharacteristicsforthecitalopramcohort(preandpostweighting).
Un-weighted Weighteda
Citalopram Paroxetineor Standardized Citalopram Paroxetineor Standardized
n=137701(%) Sertralinen=96 Differenceb(%) n=137701(%) Sertralinen=135746 Differenceb(%)
620(%) (%)
DEMOGRAPHICS
Age,years 76(7.4) 75(7.1) 19 76(7.4) 76(8.7) 2
Women 65.5 66.8 3 65.5 65.8 1
Ruralc 15.4 13.5 5 15.4 15.3 0
Longtermcare 6.9 3.4 16 6.9 6.3 2
Incomequintiled
One(lowest) 20.6 20.9 1 20.6 20.8 0
Two 21.2 21.9 2 21.2 21.8 1
Three(medium) 19.7 20.0 1 19.7 20.1 1
Four 19.0 18.6 1 19.0 18.5 1
Five(highest) 19.5 18.6 2 19.5 18.7 2
Yearofcohortentrye
2002–2005 41.6 60.7 39 41.6 58.1 33
2006–2009 39.8 25.1 32 39.8 26.7 28
2010–2012 18.6 14.2 12 18.6 15.2 9
COMORBIDITIESf
CharlsonComorbidityIndexg 0.74(1.1) 0.62(1.0) 12 0.74(1.1) 0.73(1.3) 1
Dementia 20.3 12.5 21 20.3 19.2 2
Schizophrenia/psychoticdisorders 4.0 3.4 3 4.0 4.4 2
Bipolardisorder 3.6 3.5 0 3.6 3.8 1
Unipolardepression/anxietydisorderh 21.6 20.6 2 21.6 21.0 1
Historyofself-harm 0.2 0.1 2 0.2 0.1 1
Majorhaemorrhage 5.7 4.7 4 5.7 5.6 0
HaemorrhagicStroke 0.6 0.4 3 0.6 0.5 1
IschemicStroke 4.7 3.2 7 4.7 4.5 1
TransientIschemicAttack 1.3 1.1 2 1.3 1.5 1
Chronicliverdisease 3.7 3.6 1 3.7 3.8 0
ChronicKidneyDisease 6.7 5.1 7 6.7 6.5 1
Congestiveheartfailure 15.8 13.7 6 15.8 15.7 0
Coronaryarterydiseasei 32.4 30.6 4 32.4 32.3 0
Angina 24.0 23.5 1 24.0 24.0 0
Acutemyocardialinfarction 4.5 3.9 3 4.5 4.5 0
Pacemaker 3.2 2.4 5 3.2 3.1 0
Atrialfibrillation/flutter 4.7 4.2 2 4.7 4.7 0
Peripheralvasculardisease 2.3 2.1 1 2.3 2.5 1
Chroniclungdisease 30.7 30.5 0 30.7 30.8 0
Cancerj 15.7 14.1 5 15.7 15.7 0
Alcoholism 2.4 2.4 0 2.4 2.5 0
Seizure 1.0 0.8 2 1.0 1.0 0
Acutekidneyinjury 2.4 1.6 6 2.4 2.3 1
Hospitalizationwithhyperkalemia 0.8 0.7 2 0.8 0.9 1
VenousThromboembolism 1.5 1.1 3 1.5 1.5 0
MEDICATIONSk
Anti-arrhythmics 2.2 2.1 1 2.2 2.3 1
Antipsychotics 7.1 5.0 9 7.1 6.9 1
Protonpumpinhibitors 31.2 27.1 9 31.2 31.0 0
Anti-emetic 2.2 1.7 3 2.2 2.0 1
(Continued)
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 5/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
Table1. (Continued)
Un-weighted Weighteda
Citalopram Paroxetineor Standardized Citalopram Paroxetineor Standardized
n=137701(%) Sertralinen=96 Differenceb(%) n=137701(%) Sertralinen=135746 Differenceb(%)
620(%) (%)
Lithium 0.3 0.3 1 0.3 0.3 0
Anti-lipemics 41.6 39.4 5 41.6 41.2 1
Antihypertensives 72.1 68.8 7 72.1 71.9 0
H2RAs 10.4 12.8 8 10.4 10.5 0
Pro-kinetics 4.6 4.1 3 4.6 4.5 1
Antidiabetics 16.3 15.0 3 16.3 16.0 1
Acetylsalicylicacid 9.6 11.5 6 9.6 9.7 0
Anticoagulants 10.4 7.7 9 10.4 10.2 1
Antiplatelet 6.5 4.7 8 6.5 6.3 1
Tri-cyclicantidepressants 1.4 1.6 1 1.4 1.5 1
Opioids 0.0 0.1 1 0.0 0.1 1
Anti-malarial 0.7 0.7 0 0.7 0.7 0
Anti-viral 0.0 0.0 0 0.0 0.0 1
Antibiotic 36.6 36.4 1 36.6 36.9 1
Antineoplastic 4.5 3.8 4 4.5 4.1 2
Benzodiazepine 39.6 42.8 5 39.6 40.3 1
NSAIDSl 21.5 24.0 6 21.5 21.7 1
Cholinesteraseinhibitors 0.0 0.0 0 0.0 0.0 0
Anticonvulsants 3.6 3.0 4 3.6 3.4 1
DOSEm
High 6.5 12.6 21 6.5 6.8 1
PRESCRIBER
GeneralPractitioner 76.6 78.5 5 76.6 76.9 1
Psychiatrist 2.6 2.4 1 2.6 2.6 0
Internist 0.8 0.7 1 0.8 0.8 0
Other 6.9 4.8 9 6.9 6.5 2
Missing 13.1 13.5 1 13.1 13.3 1
HEALTHCAREUSEn
NumberofHospitalizations
0 58.9 63.9 10 58.9 59.8 2
1to3 37.2 33.0 9 37.2 36.3 2
4to6 3.4 2.6 5 3.4 3.3 1
7to9 0.4 0.3 2 0.4 0.4 0
10to12 0.1 0.1 0 0.1 0.1 0
over12 0.0 0.0 0 0.0 0.1 1
NumberofEmergencyroomvisits
0 52.7 59.5 14 52.7 54.3 3
1to3 39.7 34.6 11 39.7 37.9 4
Over3 7.6 13.0 18 7.6 7.8 1
GeneralPractitionerVisits
0–4 13.9 15.9 6 13.9 14.5 2
5–9 22.4 24.2 4 22.4 22.2 0
10–14 19.5 20.2 2 19.5 19.2 1
15–19 12.9 13.0 0 12.9 12.9 0
20–24 8.6 8.2 1 8.6 8.5 0
25–29 5.7 5.3 2 5.7 5.7 0
(cid:1)30 17.1 13.2 11 17.1 16.9 1
(Continued)
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 6/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
Table1. (Continued)
Un-weighted Weighteda
Citalopram Paroxetineor Standardized Citalopram Paroxetineor Standardized
n=137701(%) Sertralinen=96 Differenceb(%) n=137701(%) Sertralinen=135746 Differenceb(%)
620(%) (%)
Athomephysicianservices 11.0 10.2 3 11.0 11.7 2
SpecialistConsultations
Psychiatristconsults 7.8 6.0 7 7.8 7.6 1
Nephrologistconsultso 6.0 5.1 4 6.0 6.0 0
Cardiologistvisits 42.1 37.9 9 42.1 41.7 1
Neurologistconsults 11.5 9.5 7 11.5 11.3 1
Diagnostictests/Interventions
Electrocardiogram 87.8 86.2 5 87.8 87.6 0
Stresstest 35.3 34.9 1 35.3 35.0 0
Echocardiography 41.7 38.1 7 41.7 41.4 1
CardiacCatheterization 6.8 6.3 2 6.8 6.5 1
HolterMonitor 21.1 19.3 5 21.1 21.0 0
Coronaryangiogram 7.5 6.8 3 7.5 7.0 2
ChestX-ray 78.4 76.1 5 78.4 78.2 0
Pulmonaryfunctiontest 24.8 25.0 0 24.8 25.1 1
Carotidultrasound 19.5 17.2 6 19.5 19.3 1
ComputedTomographyoftheHead 37.0 29.7 15 37.0 36.1 2
ComputedTomographyofotherarea 35.7 30.2 12 35.7 35.3 1
Mammogram 27.0 30.7 8 27.0 27.4 1
BoneMineralDensity 40.1 39.6 1 40.1 40.3 0
DatapresentedaspercentexceptforageandCharlsonComorbidityIndexwhicharepresentedasmean(standarddeviation).
Abbreviations:Non-steroidalanti-inflammatorydrug(NSAID)–excludesacetyl-salicylicacid,Acetylsalicylicacid(ASA),HistamineH2Receptorantagonist
(H2RA),Notapplicable(N/A)
aWeightedcohortbasedoninverseprobabilityoftreatmentweights,usingapropensityscorebasedon48baselinecharacteristics.
bStandardizeddifferencesarelesssensitivetosamplesizethantraditionalhypothesistests.Theyprovideameasureofthedifferencebetweengroups
dividedbythepooledstandarddeviation;avaluegreaterthan10%isinterpretedasameaningfuldifferencebetweenthegroups.
cDefinedasapopulation<10000people.
dIncomewascategorizedintofifthsofaverageneighbourhoodincomeonthecohortentrydate.
eTheyearofcohortentryisalsoreferredtoastheyearofcohortentrydate.
fComorbiditiesassessedbyadministrativedatabasecodesintheprevious5years.
gCharlsonComorbidityIndex[CharlsonME,PompeiP,AlexKL,MackenzieCR.Anewmethodforclassifyingprognosticcomorbidityinlongitudinalstudies:
developmentandvalidation.JChronDis1987;40(5):373–383.QuanH,SundararajanV,HalfonP,FongA,BurnandB,LuthiJC,etal.Codingalgorithmsfor
definingcomorbiditiesinICD-9-CMandICD-10administrativedata.MedCare2005;43(11):1130–1139.]wascalculatedusing5yearsofhospitalization
data.“Nohospitalizations”receivedascoreof0.
hTheprevalenceofdepressionislowsincedepressionisnotusuallyanin-patientdisorder,andthusoftennotcodedinthesourcedatabases.
iCoronaryarterydiseaseincludesreceiptofcoronaryarterybypassgraftsurgeryandpercutaneouscoronaryintervention.
jMajorcancersincludeesophagus,lung,bowel,liver,pancreas,breast,male/femalereproductiveorgans,aswellasleukemiasandlymphomas.
kBaselinemedicationuseassessedintheprevious120days.
lExcludesacetylsalicylicacid.
mRefertoS5Tablefordefinitionsofhighandlowdoses.
nHealthcareuseassessedintheoneyearpriortoSSRIprescription.
oBasedontheICESphysiciandatabase.
doi:10.1371/journal.pone.0160768.t001
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 7/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
Table2. Baselinecharacteristicsfortheescitalopramcohort(preandpostweighting).
Un-weighted Weighteda
Escitalopram Paroxetineor Standardized Escitalopram Paroxetineor Standardized
n=38436(%) Sertralinen=96 Differenceb(%) n=38436(%) Sertralinen=113 Differenceb(%)
620(%) 058(%)
DEMOGRAPHICS
Age,years 76(7.6) 75(7.1) 9 76(7.6) 75(4.5) 4
Women 63.0 66.8 8 63.0 63.2 0
Ruralc 12.8 13.5 2 12.8 12.8 0
Longtermcare 5.0 3.4 8 5.0 4.6 2
Incomequintiled
One(lowest) 18.9 20.9 5 18.9 20.0 3
Two 20.3 21.9 4 20.3 21.4 3
Three(medium) 19.6 20.0 1 19.6 20.2 2
Four 20.1 18.6 4 20.1 19.0 3
Five(highest) 21.1 18.6 6 21.1 19.3 4
Yearofcohortentrye
2002–2005 0.0 60.7 176 0.0 49.0 139
2006–2009 20.4 25.1 11 20.4 30.5 23
2010–2012 79.6 14.2 173 79.6 20.6 146
COMORBIDITIESf
CharlsonComorbidityIndexg 0.65(1.1) 0.62(1.0) 4 0.65(1.1) 0.64(0.7) 2
Dementia 19.8 12.5 21 19.8 18.6 4
Schizophrenia/psychoticdisorders 3.9 3.4 2 3.9 4.3 3
Bipolardisorder 4.0 3.5 2 4.0 4.3 2
Unipolardepression/anxietydisorderh 19.8 20.6 2 19.8 20.8 3
Historyofself-harm 0.2 0.1 3 0.2 0.1 2
Majorhaemorrhage 6.3 4.7 7 6.3 6.1 1
HaemorrhagicStroke 0.4 0.4 1 0.4 0.5 1
IschemicStroke 2.8 3.2 2 2.8 2.7 1
TransientIschemicAttack 0.8 1.1 3 0.8 1.3 6
Chronicliverdisease 3.7 3.6 0 3.7 4.0 2
ChronicKidneyDisease 7.2 5.1 9 7.2 6.9 1
Congestiveheartfailure 11.9 13.7 5 11.9 11.7 1
Coronaryarterydiseasei 28.3 30.6 5 28.3 27.9 1
Angina 18.3 23.5 13 18.3 18.3 0
Acutemyocardialinfarction 3.6 3.9 1 3.6 3.6 0
Pacemaker 3.2 2.4 5 3.2 3.1 1
Atrialfibrillation/flutter 2.9 4.2 7 2.9 3.5 4
Peripheralvasculardisease 1.6 2.1 4 1.6 2.1 5
Chroniclungdisease 28.7 30.5 4 28.7 28.6 0
Cancerj 15.3 14.1 3 15.3 15.3 0
Alcoholism 2.3 2.4 0 2.3 2.4 0
Seizure 0.7 0.8 2 0.7 0.8 3
Acutekidneyinjury 2.8 1.6 9 2.8 2.6 1
Hospitalizationwithhyperkalemia 0.7 0.7 0 0.7 0.8 2
VenousThromboembolism 1.3 1.1 1 1.3 1.3 1
MEDICATIONSk
Anti-arrhythmics 1.5 2.1 5 1.5 2.1 5
Antipsychotics 7.1 5.0 9 7.1 6.9 1
Protonpumpinhibitors 35.5 27.1 18 35.5 35.5 0
Anti-emetic 2.0 1.7 2 2.0 1.8 2
(Continued)
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 8/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
Table2. (Continued)
Un-weighted Weighteda
Escitalopram Paroxetineor Standardized Escitalopram Paroxetineor Standardized
n=38436(%) Sertralinen=96 Differenceb(%) n=38436(%) Sertralinen=113 Differenceb(%)
620(%) 058(%)
Lithium 0.3 0.3 0 0.3 0.4 2
Anti-lipemics 50.6 39.4 23 50.6 50.4 1
Antihypertensives 70.7 68.8 4 70.7 70.3 1
H2RAs 5.2 12.8 25 5.2 5.2 0
Pro-kinetics 4.5 4.1 2 4.5 4.5 0
Antidiabetics 17.7 15.0 7 17.7 17.5 1
Acetylsalicylicacid 4.4 11.5 24 4.4 4.4 0
Anticoagulants 9.4 7.7 6 9.4 9.2 1
Antiplatelet 7.5 4.7 12 7.5 7.4 0
Tri-cyclicantidepressants 1.1 1.6 4 1.1 1.5 4
Opioids 0.0 0.1 1 0.0 0.1 1
Anti-malarial 0.7 0.7 0 0.7 0.7 0
Anti-viral 0.0 0.0 0 0.0 0.0 1
Antibiotic 35.5 36.4 2 35.5 36.4 3
Antineoplastic 4.3 3.8 3 4.3 4.0 2
Benzodiazepine 35.5 42.8 13 35.5 36.0 1
NSAIDSl 17.4 24.0 16 17.4 17.5 0
Cholinesteraseinhibitors 0.0 0.0 0 0.0 0.0 0
Anticonvulsants 4.0 3.0 6 4.0 3.4 5
DOSEm
High 9.0 12.6 11 9.0 9.4 2
PRESCRIBER
GeneralPractitioner 81.0 78.5 6 81.0 81.2 1
Psychiatrist 4.5 2.4 12 4.5 4.5 0
Internist 0.5 0.7 3 0.5 0.5 0
Other 5.1 4.8 1 5.1 4.8 1
Missing 8.9 13.5 15 8.9 9.0 0
HEALTHCAREUSEn
NumberofHospitalizations
0 63.1 63.9 2 63.1 62.9 0
1to3 33.8 33.0 2 33.8 34.2 1
4to6 2.7 2.6 1 2.7 2.5 1
7to9 0.3 0.3 0 0.3 0.3 0
10to12 0.1 0.1 0 0.1 0.0 0
over12 0.0 0.0 0 0.0 0.0 0
NumberofEmergencyroomvisits
0 54.5 59.5 10 54.5 56.1 3
1to3 38.4 34.6 8 38.4 36.9 3
Over3 7.1 13.0 20 7.1 7.0 0
GeneralPractitionerVisits
0–4 14.6 15.9 4 14.6 14.6 0
5–9 25.5 24.2 3 25.5 23.8 4
10–14 21.5 20.2 3 21.5 20.2 3
15–19 12.7 13.0 1 12.7 13.2 1
20–24 7.8 8.2 1 7.8 8.5 3
25–29 4.7 5.3 3 4.7 5.5 4
(cid:1)30 13.2 13.2 0 13.2 14.2 3
(Continued)
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 9/18
RiskofVentricularArrhythmiawithCitalopramandEscitalopram
Table2. (Continued)
Un-weighted Weighteda
Escitalopram Paroxetineor Standardized Escitalopram Paroxetineor Standardized
n=38436(%) Sertralinen=96 Differenceb(%) n=38436(%) Sertralinen=113 Differenceb(%)
620(%) 058(%)
Athomephysicianservices 7.8 10.2 8 7.8 9.9 10
SpecialistConsultations
Psychiatristconsults 9.1 6.0 12 9.1 8.9 1
Nephrologistconsultso 7.1 5.1 8 7.1 7.1 0
Cardiologistvisits 44.9 37.9 14 44.9 44.8 0
Neurologistconsults 10.0 9.5 2 10.0 10.0 0
Diagnostictests/Interventions
Electrocardiogram 89.1 86.2 8 89.1 88.9 1
Stresstest 39.1 34.9 9 39.1 37.4 5
Echocardiography 47.7 38.1 20 47.7 47.4 1
CardiacCatheterization 7.1 6.3 3 7.1 5.9 7
HolterMonitor 24.2 19.3 12 24.2 24.0 1
Coronaryangiogram 8.0 6.8 5 8.0 6.4 8
ChestX-ray 77.7 76.1 4 77.7 77.4 1
Pulmonaryfunctiontest 26.7 25.0 4 26.7 26.6 0
Carotidultrasound 20.0 17.2 7 20.0 19.7 1
ComputedTomographyoftheHead 37.8 29.7 17 37.8 37.1 2
ComputedTomographyofotherarea 41.8 30.2 25 41.8 41.4 1
Mammogram 24.4 30.7 14 24.4 24.8 1
BoneMineralDensity 41.9 39.6 5 41.9 42.3 1
DatapresentedaspercentexceptforageandCharlsonComorbidityIndexwhicharepresentedasmean(standarddeviation).
Abbreviations:Non-steroidalanti-inflammatorydrug(NSAID)–excludesacetyl-salicylicacid,Acetylsalicylicacid(ASA),HistamineH2Receptorantagonist
(H2RA),Notapplicable(N/A)
aWeightedcohortbasedoninverseprobabilityoftreatmentweights,usingapropensityscorebasedon48baselinecharacteristics.
bStandardizeddifferencesarelesssensitivetosamplesizethantraditionalhypothesistests.Theyprovideameasureofthedifferencebetweengroups
dividedbythepooledstandarddeviation;avaluegreaterthan10%isinterpretedasameaningfuldifferencebetweenthegroups.
cDefinedasapopulation<10000people.
dIncomewascategorizedintofifthsofaverageneighbourhoodincomeonthecohortentrydate.
eTheyearofcohortentryisalsoreferredtoastheyearofcohortentrydate.
fComorbiditiesassessedbyadministrativedatabasecodesintheprevious5years.
gCharlsonComorbidityIndex[CharlsonME,PompeiP,AlexKL,MackenzieCR.Anewmethodforclassifyingprognosticcomorbidityinlongitudinalstudies:
developmentandvalidation.JChronDis1987;40(5):373–383.QuanH,SundararajanV,HalfonP,FongA,BurnandB,LuthiJC,etal.Codingalgorithmsfor
definingcomorbiditiesinICD-9-CMandICD-10administrativedata.MedCare2005;43(11):1130–1139.]wascalculatedusing5yearsofhospitalization
data.“Nohospitalizations”receivedascoreof0.
hTheprevalenceofdepressionislowsincedepressionisnotusuallyanin-patientdisorder,andthusoftennotcodedinthesourcedatabases.
iCoronaryarterydiseaseincludesreceiptofcoronaryarterybypassgraftsurgeryandpercutaneouscoronaryintervention.
jMajorcancersincludeesophagus,lung,bowel,liver,pancreas,breast,male/femalereproductiveorgans,aswellasleukemiasandlymphomas.
kBaselinemedicationuseassessedintheprevious120days.
lExcludesacetylsalicylicacid.
mRefertoS5Tablefordefinitionsofhighandlowdoses.
nHealthcareuseassessedintheoneyearpriortoSSRIprescription.
oBasedontheICESphysiciandatabase.
doi:10.1371/journal.pone.0160768.t002
PLOSONE|DOI:10.1371/journal.pone.0160768 August11,2016 10/18