The hospital anxiety and depression rating scale: A cross ...

General practitioners' (GPs) diagnostic skills lead to underidentification of generalized anxiety disorders (GAD) and major depressive ... Skiptomaincontent Advertisement SearchallBMCarticles Search Thehospitalanxietyanddepressionratingscale:Across-sectionalstudyofpsychometricsandcasefindingabilitiesingeneralpractice DownloadPDF DownloadPDF Researcharticle OpenAccess Published:14December2005 Thehospitalanxietyanddepressionratingscale:Across-sectionalstudyofpsychometricsandcasefindingabilitiesingeneralpractice IngridOlssøn1,ArnsteinMykletun2,3&AlvADahl4  BMCPsychiatry volume 5,Article number: 46(2005) Citethisarticle 51kAccesses 385Citations Metricsdetails AbstractBackgroundGeneralpractitioners'(GPs)diagnosticskillsleadtounderidentificationofgeneralizedanxietydisorders(GAD)andmajordepressiveepisodes(MDE).SupplementofbriefquestionnairescouldimprovethediagnosticaccuracyofGPsforthesecommonmentaldisorders.TheaimsofthisstudyweretoexaminetheusefulnessofTheHospitalAnxietyandDepressionRatingScale(HADS)forGPsby:1)ExaminingitspsychometricsintheGPs'setting;2)Testingitscase-findingpropertiescomparedtopatient-ratedGADandMDE(DSM-IV);and3)ComparingitscasefindingabilitiestothatoftheGPsusingClinicalGlobalImpression-Severity(CGI-S)rating.MethodsInacross-sectionalsurveystudy1,781patientsinthreeconsecutivedaysinSeptember2001attended141GPsgeographicallyspreadinNorway.Sensitivity,specificity,optimalcutoffscore,andAreaunderthecurve(AUC)fortheHADSandtheCGI-SwerecalculatedwithGeneralizedAnxietyQuestionnaire(GAS-Q)asreferencestandardforGAD,andDepressionScreeningQuestionnaire(DSQ)forMDE.ResultsTheHADS-Ahadoptimalcutoff≥8(sensitivity0.89,specificity0.75),AUC0.88and76%ofpatientswerecorrectlyclassifiedinrelationtoGAD.TheHADS-Dhadbyoptimalcutoff≥8(sensitivity0.80andspecificity0.88)AUC0.93and87%ofthepatientswerecorrectlyclassifiedinrelationtoMDE.ProportionsofthetotalcorrectlyclassifiedattheCGI-Soptimalcut-off≥3were83%ofpatientsforGADand81%forMDE.ConclusionTheresultsindicatethatadditionofthepatients'HADSscorestoGPs'informationcouldimprovetheirdiagnosticaccuracyofGADandMDE. PeerReviewreports BackgroundAnimportanttaskforgeneralpractitioners(GPs)istodiagnoseandtreatdepressionsandanxietydisorders,whichareamongthemostcommonandamenablementaldisordersintheirpractice[1].TheliteratureshowsthattheGPs'diagnosticskillsconcerningthesecommondisordersaremoderatelygood[1–8],andsomewhatbetterformajordepressiveepisodes(MDE)thanforgeneralizedanxietydisorder(GAD)[8].Aprospectivecohortstudyofdepressioninprimarycare,foundthattheWHO-5wellbeingindex(WHO-5,5items)hadsignificantlyhighersensitivitythantheGPs'clinicaldiagnosiswhentheCompositeInternationalDiagnosticInterview(CIDI)wasusedasgoldstandard[9].Thedepressionmoduleofthebriefpatienthealthquestionnaire(B-PHQ,9items)hadsignificantlyhigherspecificitythanGPs'clinicaldiagnoses,andGPs'diagnoseshadsignificantlyhigherspecificitythantheWHO-5.ThesensitivityandspecificityofscreeninginstrumentsforGADingeneralpracticehashardlybeeninvestigated[10,11].Reviews[12,13]showthattheHospitalAnxietyandDepressionRatingScale(HADS)[14]iswidelyusedasabriefself-ratinginstrumentforbothdimensionalandcategoricalaspectsofanxietyanddepressioninbothepidemiologyandspecialistcare.InthesesettingsthepsychometricpropertiesoftheHADSareexcellent[15,16].Untilnowthefactorstructure,theinternalconsistency,andtheinter-correlationandhomogeneityoftheHADSsub-scaleshavenotbeendescribedinthecontextofgeneralpractice.Thecase-findingabilitiesoftheHADSinrelationtoDSM-III/DSM-IVandICD-10definedanxietydisordersanddepressionsbytheuseofascore≥8ascut-offareconsideredgoodwithfewfalsenegatives,butadefiniteproportionoffalsepositives.Inclinicalpracticeapositivescreeningtypicallyresultsinfurtherevaluation.ConsideringthebrevityandfeasibilityoftheHADS,itshouldbeusefulforscreeningofpatientsingeneralpractice,butstudiesoftheHADSfromthatpartofthehealthservicesarefewandinconsistentastocut-offscoresforcaseness[17–20].Thesepointsindicatetheneedformoredataonthecase-findingabilitiesoftheHADSsubscalesingeneralpractice.AimsofthestudyThisstudyfromNorwegiangeneralpracticehasthefollowingaims:1)ToexaminethepsychometricfeaturesoftheHADSratedbypatientsintheprimarycaresetting;2)Totestthecase-findingpropertiesoftheHADSinrelationtothediagnosesofGADandMDEbasedonpatient-ratingoftheirdiagnosticcriteriaaccordingtoDSM-IVasreferencestandards;and3)TocomparethecasefindingabilitiesoftheHADSratedbypatientstothatofGPsusingtheClinicalGlobalImpression-Severity(CGI-S).MethodsProcedureThestudyisbasedonacross-sectionalstudyofGP'sandtheirpatientscarriedoutinGermany,Scandinavia,andFinland[8,21].Aflow-chartoverthestudydesignisshowninFigure1.Essentialfeaturesofthedesignwere:1)SamplingofGPsgeographicallyspreadinNorway;2)DuringthreeconsecutivedaysinSeptember2001alltheGPs'patientswereinvitedtotakepartinastudyratingthemselvesonthethreequestionnairesconcerninganxietyanddepression;.3)Blindtothepatients'ratings,theGPsfilledintheCGI-SinordertoratetheclinicalseverityofeventualGADandMDEintheirpatients.Exclusioncriteriaforpatientswere:age<16years,languagedifficulties,patientswhorequiredhelptocompletethequestionnaires,andpatientswhocameforprescriptiononly,orforanaccident/emergency.Figure1Flowchartstudydesign.Fullsizeimage SamplingofGPsTheGPsinvariouspartsofNorwaywererecruitedasaconveniencesampleamongthoseregisteredinthedatabaseofWyethNorwayLtd.TheproceduralinformationtotheGPswasgiveninwriting,andnospecialtrainingofthemforthestudywasundertaken.Among141participatingGPs,136wereeligibleand133gavedemographicdata.NinetyGPs(68%)weremenand43(32%)women.Theyhadbeenworkinginprimarycareforameanof15(SD7)and11(SD7)years,respectively,and118(89%)ofthemworkedingrouppractice.TheGPsconsultedwithameanof21.1(SD5.1)patientsonanaverageday.TherewerenosignificantdifferencesbetweengendersofGPswithregardtonumberofconsultations.SamplingofpatientsAmongthe1,781patientsincludedinthestudy,1,385(78%)hadvalidratingsontheDSQandtheHADS-D,and1,238(70%)ontheGAS-QandtheHADS-A.Theproportionofwomeninthetwosampleswas64%and63%,respectively,withameanageof45(SD17)forwomenand50(SD17)yearsformen.FurtherdemographiccharacteristicsofpatientsareshowninTable1.Thenon-complyingpatientsdidnotdiffersignificantlyfromthecompliersastoageandgender,whichweretheonlydataatdisposalforattritionanalyses.Table1Samplecharacteristics.Fullsizetable DiagnosticcriteriaandinstrumentsPsychiatricclassificationsystemslikeDSM-IVandICD-10arebasedonthepresenceorabsenceofvariousoperationalizeddiagnosticcriteria.Whenstructuredinterviewsareused,thepatientsareaskedforthepresenceofthediagnosticcriteriabyaninterviewer.Incontrast,inthisstudythepatientsratethemselvesthediagnosticcriteriaforGAD(DSM-IV)ontheGeneralizedAnxietyQuestionnaire(GAS-Q)andforMDE(DSM-IV)ontheDepressionScreeningQuestionnaire(DSQ),andthesepatientratingsareusedasdiagnosticreferencestandardinthisstudy.TheGAS-QisamodificationoftheAnxietyScreeningQuestionnaire[22],andisaself-ratingquestionnairedevelopedtodiagnoseGADaccordingtoDSM-IVandICD-10.TheGAS-Qconsistsof20itemscoveringthediagnosticcriteriaforGADintheDSM-IV.Test-retestreliabilityoftheGAS-Qoveratwo-dayretestperiodshowedakappavalueof0.74forthediagnosisofGAD.CongruentvaliditycomparingGAS-QdiagnosiswiththeDSMIValgorithmforGADoftheCompositeInternationalDiagnosticInterviewshowedakappaof0.72[23].TheDSQwasmadeforpatient-ratingofMDEaccordingtoDSM-IVandICD-10[24]andwaschosenasourreferencestandard.TheDSQisan11itemquestionnaireinwhichdiagnosticcriteriaareratedonathreepointscale,supplementedbythreequestionstoassesstheageatfirstandcurrentepisode,andthenumberofepisodesaccordingtothecriterionAofMDEinDSM-IV.ConsistentwiththeDSM-IVcriteria,adiagnosisofMDEwasassignedwhenatleastfiveoftheitemswereratedaspositivebythepatient.IntheGermanpartoftheEuropeanstudy,theinternalconsistencyoftheDSQshowedaCronbach'scoefficientalphaof0.83[25].Test-retestreliabilityoveratwo-dayperiodfoundakappavalueof0.82forMDE[8].TestsoftheDSQdiagnosisversusdiagnosisofMDEbasedonstructuredinterviewshowedakappa0.89[26].TheHADSconsistsofsevenitemsforanxiety(HADS-A)andsevenfordepression(HADS-D).Theitemsarescoredonafour-pointscalefromzero(notpresent)tothree(considerable).Theitemscoresareadded,givingsub-scalescoresontheHADS-AandtheHADS-Dfromzeroto21.InthisstudyvalidHADSsubscalescoresweredefinedashavingansweredatleastfiveofsevenitemsonboththeHADS-AandtheHADS-D.Inordertobevalidinpatientswithsomaticproblems,theHADSitemswerebasedonthepsychologicalaspectsofanxietyanddepression.Theanxietyitemswereconcentratedongeneralanxiety,andfiveoftheitemswereclosetothediagnosticcriteriaofGAD.Thedepressionitemswerebasedonanhedonia,whichisconsideredtobeoneoftheessentialcriteriaofdepression[27].TheconcurrentvalidityoftheHADScomparedtootherquestionnairesforanxietyanddepressionisdescribedbetween0.60and0.80forbothsub-scales[13].TheCGI-Sisastandardizedassessmenttoolthatiswidelyusedasanoutcomemeasureinresearch[28].TheCGI-Shadthefollowingwording:"InyourclinicaljudgementhowseverelydoesthispatientsufferfromMDE/GAD?"TheratingsofCGI-Swere:1=notillatall,2=aborderlinecase,3=onlymildlyill,4=moderatelyill,5=seriouslyilland6=extremelyseriouslyill.TheCGI-Sscalewasdichotomisedinto1–2=notill,3–6=ill,butwealsoexploredthefrequencyofcasesbyaCGI-Sscoreof≥2(=borderlinecase).StatisticalmethodsThestatisticalanalyseswerecarriedoutwiththeSPSSforWindows,version11.0.PrincipalComponentAnalysis(PCA)withobliquerotationwasperformedtoexplorethefactorstructureoftheHADS.InternalconsistencyoftheHADS-AandtheHADS-DwastestedusingCronbach'scoefficientalpha.Pearson'scorrelationcoefficientwasusedforestimationoftheoverlapbetweenthesubscales.Sensitivityandspecificitywerecalculatedfordifferentcut-offvaluesfortheHADS-A,theHADS-D,andtheCGI-SinrelationtotheprevalencerateofGADidentifiedwithGAS-QandtherateofMDEidentifiedwithDSQ.Sensitivitiesandspecificitiesbyoptimalcut-offwereusedtocalculatetheratesoftrueandfalsepositiveandnegativecases.TheReceiverOperatingCharacteristics(ROC-curve)weredepictedgraphically,andtheAreaUndertheCurve(AUC)werecalculatedfortheHADS-A,theHADS-DandtheCGI-SagainsttheGAS-QandtheDSQasreferencestandards.Theassociationsofageandgendertocasenessontheinstrumentswereexaminedbylogisticregressionanalyses.Allsignificancetestsweretwo-tailed,andp-values<.05werereportedassignificant.ethicsthecommitteeformedicalethicsofhealthregioneastofnorwayapprovedthi sstudy.theparticipantsdeliveredinformedconsentafterwritteninformationaboutthestudy.wyethnorwayltdpai dthegpsafixedsumofeur15perpatientinadditiontotheirnormalsalary.noemployeesofwyethltd.werepresentinan yofthegeneralpracticesduringthedayofinclusion.thenationalstudyleadercoordinatedthestudy figure2receiveroperatingcurvesforhads-aandcgi-sfordetectinggad.fullsizeimage figure3receiveroperatingcurvesforhads-dandcgi-sfordetectingmde.fullsizeimage comparisonofgp-ratedandpatient-ratedcaseidentificationusingthegps formdenosignificantdifferencewasobservedbetweenratesoftruepositivedisorder references googlescholar goldbergd article pubmed jenckssf:recognitionofmentaldistressanddiagnosisofmentaldisordersinprimarycare.jama.1985 feinsteinre kesslerrc:theepidemiologyofpureandcomorbidgeneralizedanxietydisorder:areviewandevaluationofrecentres earch.actapsychiatrscandsuppl.2000 wittchenhu henkelv pubmedcentral hoyervonj beesdok herrmannc:internationalexperienceswiththehospitalanxietyanddepressionscale bjellandi zigmondas mooreys mykletuna elrufaieoe lamcl wilkinsonmj l allgulanderc krausep winters h watsond guyw:clinicalglobalimpressionsscale.ecdeuassessmentmanualforpsychopharmacology.usdepthealth statisticsandresearchonphysiciansinnorway. crawfordjr spinhovenph pre-publicationhistorythepre-publicationhistoryforthispapercanbeaccessedhere:http: downloadreferencesacknowledgementshospitalinnlandettrust ingridolss authors openaccess thisarticleispublishedunderlicensetobiomedcentralltd.thisisanopenaccessarticleisdistributedunderthet ermsofthecreativecommonsattributionlicense https: reprintsandpermissionsaboutthisarticlecitethisarticleolss bmcpsychiatry5 providedbythespringernatureshareditcontent-sharinginitiative keywordsgeneralizeanxietydisordermajordepressiveepisodecompositeinternationaldiagnosticinterviewsoma ticproblemgeneralizeanxietydisorder downloadpdf advertisement bmcpsychiatry issn:1471-244x contactus submissionenquiries:bmcpsychiatry generalenquiries:orsupport>