Does the 12-item General Health Questionnaire contain ...

The General Health Questionnaire (GHQ-12) consists of 12 items, each assessing the severity of a mental problem over the past few weeks using a ... Skiptomaincontent Advertisement SearchallBMCarticles Search Doesthe12-itemGeneralHealthQuestionnairecontainmultiplefactorsanddoweneedthem? DownloadPDF DownloadPDF Research OpenAccess Published:11November2004 Doesthe12-itemGeneralHealthQuestionnairecontainmultiplefactorsanddoweneedthem? FeiGao1,NanLuo2,JulianThumboo3,CalvinFones4,Shu-ChuenLi5&…Yin-BunCheung6 Showauthors HealthandQualityofLifeOutcomes volume 2,Article number: 63(2004) Citethisarticle 91kAccesses 95Citations 6Altmetric Metricsdetails AbstractBackgroundThe12-itemGeneralHealthQuestionnaire(GHQ-12)iswidelyusedasaunidimensionalinstrument,butfactoranalysestendedtosuggestthatitcontainstwoorthreefactors.NotmuchisknownabouttheusefulnessoftheGHQ-12factors,iftheyexist,inrevealingbetween-patientdifferencesinclinicalstatesandhealth-relatedqualityoflife.MethodsWeaddressedthisissueinacross-sectionalsurveyofout-patientswithpsychologicaldisordersinSingapore.Theparticipants(n=120)completedtheGHQ-12,theBeckAnxietyInventory,andtheShort-Form36HealthSurvey.ConfirmatoryfactoranalysiswasusedtocomparesixpreviouslyproposedfactorstructuresfortheGHQ-12.Factorscoresofthebest-fittingmodel,aswellastheoverallGHQ-12score,wereassessedinrelationtoclinicalandhealth-relatedqualityoflifevariables.ResultsThe3-factormodelproposedbyGraetzfittedthedatabetterthanaunidimensionalmodel,two2-factormodels,andtwoother3-factormodels.However,thethreefactorswerestronglycorrelated.Theirvaluesvariedinasimilarfashioninrelationtoclinicalandhealth-relatedqualityoflifevariables.ConclusionsThe12-itemGeneralHealthQuestionnairecontainsthreefactors,namelyAnxietyandDepression,SocialDysfunction,andLossofConfidence.Nevertheless,usingthemseparatelydoesnotoffermanypracticaladvantagesindifferentiatingclinicalgroupsoridentifyingassociationwithclinicalorhealth-relatedqualityoflifevariables. BackgroundRecentstudiesofdiseaseburdenhavedemonstratedtheimportanceofpsychologicaldisorders.Forinstance,depressionwasthefourthleadingcauseofdiseaseburden,accountingfor4.4%oftotaldisabilityadjustedlifeyearsintheworldin2000[1].The12-itemGeneralHealthQuestionnaire(GHQ-12)hasbeenwidelyusedinmanycountriesfordetectingpsychologicalmorbidity.SomemajornationalstudiessuchastheBritishHouseholdPanelSurvey(BHPS)alsoemploythisinstrument[2].Calibrationofthisinstrumentmaythereforecontributesignificantlytoalargecommunityofresearchers.WhilethelongerversionsoftheGHQarenormallyconsideredmultidimensional,theGHQ-12isoftenregardedasmeasuringonlyasingledimensionofpsychologicalhealth.Forexample,Corti[3]analyzedtheGHQ-12dataintheBHPSandmaintainedthatthehighCronbach'salphavalueindicatedtheunidimensionalityofthisinstrument.However,severalauthorssuggestedthattheGHQ-12containedtwoorthreeclinicallymeaningfulfactors.Usingprincipalcomponentanalysis,Politietal.[4]identifiedtwofactors:generaldysphoriaandsocialdysfunction.AndrichandvanSchoubroeck[5]suggestedthatthepositivelywordeditemsformedonefactorandthenegativelywordeditemsformedanother.Graetz[6],Martin[7]andWorselyandGribbin[8]proposedthreedifferent3-factormodels.Inamulti-centrestudy,althoughconsiderablebetween-centrevariationwasfound,thefinalsolutiontendedtohaveeithertwoorthreefactors[9].Usingconfirmatoryfactoranalysis(CFA)toanalyzetheBHPSdata,Cheung[10]comparedvariousmodelsandfoundthatthe3-factormodelproposedbyGraetz[6]gavethebestfit.Thefactorsareanxietyanddepression(4items),socialdysfunction(6items),andlossofconfidence(2items).InastudyofemployeesinNewZealand,Kalliathetal[11]alsoemployedCFAtocomparevariousmodels.TheyalsofoundthatGraetz's3-factormodelgavebettergoodness-of-fitthantheothers.However,theymaintainedthatnoneofthemodelstheyexaminedgaveasufficientlevelofgoodness-of-fit.Hencetheymodifiedtheinstrumenttoproposeashort(8-item)versionofGHQ.InastudyofcollegestudentsandyoungadolescentsinAustralia,FrenchandTait[12]foundthatGraetz'smodelnotonlyfittedthedatabetterthanothermodels,butalsosatisfactorilyachievedsomefitindicestargetssuchasComparativeFitIndex>0.95.InastudyofaruralpopulationinAustralia[13],themodelofWorselyandGribbinfittedbestandthatofGraetzwassecondbest.WhilethestructureoftheGHQ-12hasbeenstudiedusingfactoranalysismethods,theconstructvalidityandusefulnessofthoseresultingfactorsarenotoftentested.Thequestioniswhethertheadditionalinformationprovidedbythe2or3factors,iftheyexist,isclinicallyuseful.Inotherwords,willmultiplescoresbemoreusefulthanatotalsinglescoreinhelpingustounderstandrespondents'healthstatus?Thepurposeofthisstudywasthereforetwo-fold.First,weaimedtocomparethepreviouslyproposedmodelsoftheGHQ-12inanorientalpopulationandidentifythebest-fittingone.ItwasnotourobjectivetoassesstheirabsoluteleveloffitortoderivenewmodelorversionoftheGHQ.Second,weaimedtoassesswhetherthefactorsidentifiedrelatetoclinicalandhealth-relatedqualityoflifevariablesindifferentways.MethodsSubjectsandstudydesignAconsecutivesampleofoutpatientswithanxietydisordersand/ordepressivedisorderswasrecruitedfromapsychiatricclinicatatertiaryhospitalinSingapore.Inclusioncriteriawerethepresenceofanyanxietydisorderand/ormajordepressivedisorder,literacyinEnglishorChinese,andcompletionofaninformedconsentform.Patientswithorganicbrainsyndromeorpsychosiswereexcluded.Duringroutineconsultationvisits,diagnosesofrecruitedpatientswereascertainedbyapsychiatristusingDSM-IVcriteriaandtheseverityoftheirpsychiatricdisorderswasassessedusingaClinicalGlobalImpression(CGI)scale,whichrangesfrom1(verymild)to5(verysevere).PatientswerethengivenaquestionnairecontainingtheGeneralHealthQuestionnaire(GHQ-12)[14],theBeckAnxietyInventory(BAI)[15],andtheShortForm-36HealthSurvey(SF-36)[16]forself-completion.IdenticalEnglishandChinesequestionnaireswerepreparedforsubjectstoselectaccordingtotheirpreference.Aresearchassistantcheckedreturnedquestionnairesforcompleteness.InstrumentsTheGeneralHealthQuestionnaire(GHQ-12)consistsof12items,eachassessingtheseverityofamentalproblemoverthepastfewweeksusinga4-pointscale(from0to3).Thescorewasusedtogenerateatotalscorerangingfrom0to36,withhigherscoresindicatingworseconditions[14].TheChineseversionofGHQ-12usedinthisstudyhadbeenvalidated[17,18].Apreviousstudyofthe60-and30-itemversionsofEnglishandChineseGHQyieldedcomparablescalescores,suggestingequivalenceforthetwolanguageversions[19].TheBeckAnxietyInventory(BAI)isavalidandreliableself-reportchecklistforanxietysymptoms[15].Thisinstrumentconsistsof21items,eachdescribingananxietysymptomforarespondenttoassesshowmuchheorshehasbeenbotheredbythesymptomoverthepastweekona4-pointscale.Responsestoallitemsaresummeduptoatotalscorerangingfrom0to63,withhigherscoresindicatingmoresevereanxiety.AChineseBAIwasdevelopedbytheauthorsusingforward-andback-translationprocedures,andrefinedafterapilotstudyofsubjectswithanxietydisorders[20].TheShortForm36HealthSurvey(SF-36)[16]isa36-itemquestionnaireassessingfunctionalhealth-relatedqualityoflife(HRQoL)in8domains:physicalfunctioning,rolelimitationsduetophysicalproblems,bodilypain,generalhealth,vitality,socialfunctioning,rolelimitationsduetoemotionalproblems,andmentalhealth.Theinstrumentyieldseachdomainascorerangingfrom0to100,withhigherscoresindicatingbetterHRQoL.ThevalidityandreliabilityofSF-36havebeenextensivelydocumented[21].InSingapore,boththeUKEnglish[16]andChinese(HongKong)[22]versionsofSF-36havebeenvalidated[23,24]andthesetwolanguageversionsappeartobeequivalent[25].StatisticalanalysisVariousfactorstructuresoftheGHQ-12weretestedbyconfirmatoryfactoranalysis.ModelIwasunidimensional.ModelIIAcontained2factors:GeneralDysphoriaandSocialDysfunction[4].ModelIIBalsocontained2factors:positivelywordeditemsformingonefactorandnegativelywordeditemsforminganother[5].ModelIIIAcontained3factors:Cope,StressandDepress,identifiedbyMartin[7].ModelIIIBwasthe3-factormodelproposedbyGraetz[6]:AnxietyandDepression,Socialdysfunction,andLossofConfidence.ModelIIICwasalsoa3-factormodel:Anhedonia-Sleepdisturbance,SocialPerformanceandLossofConfidence[8].IntheconfirmatoryfactoranalysisthenumberoffactorsandtherelationshipbetweenfactorsandobservedGHQ-12itemswerepre-specifiedaccordingtothemodels.Theloadingofanitemonafactorwithinamodelwasestimatedusingthemaximumlikelihoodmethod.Methodologistshaveemphasizedthatitisdesirabletousedifferentindicatorstoexamineamodel'sgoodness-of-fit[26].Thefitofthesixmodelswasassessedbythreemeasures.TheAkaike'sInformationCriterion(AIC)penalizesthemaximumloglikelihoodofamodelaccordingtoitsnumberofparameters.AmodelwithalowerAICismoreplausiblethanonewithahigherAIC.Insteadofshowingrelativefitness,theComparativeFitIndex(CFI)assessesthefitofamodelitself.Thevaluesrangebetween0and1.ACFIlargerthan0.90indicatesanacceptablemodel.(HuandBentler[27]suggestedthataCFIvalueabove0.95indicatesanacceptablemodel.Inalatersectionwewilldiscussthemorestringentcutoff.)TheRootMeanSquareofApproximation(RMSEA)assessesamodel'samountoferror.AnRMSEAvaluelargerthan0.08indicatestoomucherror.Thebest-fittingmodelwasexaminedindetail.TheKruskal-WallistestwasusedtocomparetheGHQ-12overallandfactorscoresofpatientswithdifferentdiagnosis.Pearson'scorrelationcoefficient(r)wasusedtoassesstheassociationbetweenGHQ-12scoresandvariousvariables,namelyBeckAnxietyInventory,ClinicalGlobalImpressionandSF-36scores.TheFisher'sZtransformationwasusedtoproduce95%confidenceinterval.ResultsandDiscussionAtotalof120participants(63manand57women)wereincludedintheanalysis(Table1).Most(90%)respondentswereChinese;themean(SD)agewas43.1(12.7).SixtysixpercentoftheparticipantschosetoadministeranEnglishversionofthequestionnaire.ThemeanscoresofclinicalandHRQoLdatareportedbytherespondentsinbothgenderwereshowninTable1.Mentendedtohavelessanxiety,betterclinicalglobalimpression,andhigherSF-36scores.Table1Mean(SD)clinicalandSF-36health-relatedqualityoflifevaluesbygenderFullsizetable Table2showsgoodness-of-fitstatisticsforthe1-,2-and3-factormodels.The3-factormodel(IIIB)proposedbyGraetz(1991)wasthebestintermsofallthreefitstatistics.ItgavethelowestAICandRMSEAandhighestCFI.ItsCFIwas0.935.AllsixmodelsproducedRMSEA'swhichexceeded0.08.Theone-dimensionalmodel(ModelI)hadthehighestAIC,highestRMSEAandlowestCFI.Table2Goodness-of-fitofsixconfirmatoryfactoranalysismodels(N=120)(a),(b) Fullsizetable Figure1displaysthestandardizedfactorloadingsandbetween-factorcorrelationofmodelIIIB.Thefactorloadingsrangedbetween0.72and0.90.Thethreefactorswerestronglycorrelated.Thecorrelationbetweenfactor1(AnxietyandDepression)andfactor2(SocialDysfunction)was0.89.Thecorrelationbetweenfactor2andfactor3(LossofConfidence)was0.83.Thatbetweenfactor1and3was0.90.Thesestrongcorrelationssuggestthateveniftherewereinfactthreefactors,inpracticeitmaybeverydifficulttodiscernthem.Figure1Standardisedfactorloadingsandbetween-factorcorrelationsofGraetz'smodel[6].BoxesrepresentGHQ-12items;ellipsesrepresentfactors.One-wayandtwo-wayarrowsindicatefactorloadingsandbetween-factorcorrelations,respectively.Fullsizeimage HavingestablishedthatGraetz's3-factormodelfittedthedatabetterthantheothermodels,wecalculatedthefactorscoresasunweightedsumsoftheitemsconcerned.Fromfigure1wecouldseethattheloadingsoneachfactordidnotvarysubstantially.Hencewechosetouseunweightedsumsforsimplicity.Table3showsthemean(SD)factorscoresandtheoverallGHQ-12scorebyclinicaldiagnosis.Somepatientshadmultiplediagnoses;wecategorizedthemintooneofthreemajorclinicaldiagnoses.ThethreefactorscoresandtheoverallGHQ-12scoresbehavedinfairlysimilarways.Allfourscoresweresignificantlydifferentbetweenpatientswithandwithoutdepression;nonewassignificantlydifferentbetweenpatientswithandwithoutgeneralanxietydisorder.PatientswithpanicdisorderhadlowerscoresonthefactorLossofConfidence(difference=0.68;P=0.043).TheSDofthetwodiagnosisgroupspooledwasabout1.75;thebetweengroupdifferencewasthereforeapproximatelyabout0.4SD.Table3Comparisonofmean(SD)valuesofGHQ-12scoresbyclinicaldiagnosis.Fullsizetable Table4presentstheresultsofthecorrelationof3factorsofGraetz'smodelandBAI,ClinicalGlobalImpressionScore,andSF-36scales.The3factorswerecorrelatedwiththe10clinicalandHRQoLvariablestoverysimilardegree.Table4Pearson'scorrelationcoefficients(95%confidenceintervals)betweenGHQ-12scoresandclinicalandhealth-relatedqualityoflifevariablesFullsizetable Severalpreviousconfirmatoryfactoranalysesfoundthatthe3-factormodelofGraetzgavebetterfittosurveydatafromAustralia[12],Britain[10]andNewZealand[11].InthisstudyweexaminedtheissueinanAsianpopulationinSingapore,whosemembersaremainlyethnicChinese.Allthreegoodness-of-fitindicesemployed,namelyAIC,CFIandRMSEA,agreedthatthe3-factormodelofGraetzout-performedtheotherfivemodels.TheCFIvaluewas0.935.Conventionally,aCFIof0.90orlargeristakenasevidenceofsufficientfit.AmorestringentcriterionofCFIlargerthan0.95hasrecentlybeenproposedanddebated[27,28].TheRMSEAalsoindicatedthateventhebest-fittingmodeldidnotfitwell,usingthecut-offof0.08asacriterion.However,ouraimistocomparethemodelsratherthantomodifytheinstrument.Soforourpurposeitisthecomparisonofthegoodness-of-fitofthesixmodelsthatmatters,nottheabsolutevaluesofthefitindices.Weconsiderthe"correctness"and"usefulness"ofamodeltwofairlyseparateissues.Althoughthegoodness-of-fitofGraetz'smodelwaslimited,weproceededtoexaminethefactorscoresinrelationtoexternalcriteriainordertoreachaconclusionabouttheusefulnessofthemodel.Theone-dimensionalmodelwastheworstaccordingtoallthreegoodness-of-fitindices.ThethreefactorsinthemodelproposedbyGraetzwerefoundtobestronglycorrelatedwitheachother,withcorrelationcoefficientsintheneighborhoodof0.8to0.9.Suchstrongcorrelationssuggestthateveniftherewereindeedthreedifferentfactors,inpracticeitisquitedifficulttodifferentiatethem.ThestudyofFrenchandTait[12]alsoshowedstrongcorrelationbetweenthefactors,whichledtheauthorstorecommendthatitmaybeprudenttousetheoverallscoreratherthanoverinterpretthefactorswithintheGHQ-12.WeexaminedthethreefactorscoresandtheoverallGHQ-12scoreinrelationtoclinicaldiagnoses.Thefourscoresbehavedinfairlysimilarways.AlthoughtheLossofConfidencescalewassignificantlydifferentbetweenpatientswithandwithoutpanicdisorderwhiletheotherthreescalesdidnotshowsignificantdifferencesbetweenthetwogroupsofpatients,thedifferencewasonlyabout0.4SD.Thisissmallerthanarecommendedthreshold(0.5SD)correspondingtominimalclinicallyimportantdifferencesforhealthstatesquestionnaires[29].WealsoexaminedtheassociationbetweenthethreeGHQscoresandtheBeckAnxietyInventory,aclinicalimpressionscore,andthe8scalesoftheSF-36.ThethreefactorswereassociatedwiththeclinicalandHRQoLvariablestosimilardegrees.Twolimitationsofthestudyshouldbenoted.Firstly,thesamplesizewassomewhatsmallforconfirmatoryfactoranalysis.Secondly,theparticipantswereclinicalcases.ThishomogeneitymighthavemadeitmoredifficulttodetectvariationsinGHQ-12scores.Webelievethatthequestionabouttherelativeplausibilityofvariousfactormodelshavebeensufficientlyansweredbythisandseveralpreviousstudies[10–12].Nevertheless,futurestudiesofnon-clinicalparticipantsbasedonlargersamplesizeswillbehelpfultofurtherassessthepracticalusefulnessofthefactorsoftheGHQ-12.ConclusionsSeveralstudies,includingthepresentone,havefoundthatGraetz's3-factormodeloftheGHQ-12ismoreplausiblethanothermodels.However,thefactorswerestronglycorrelatedanddifficulttodiscern.OuranalysisofthethreeGHQscoresinrelationtoclinicalvariablesandaspectsofhealth-relatedqualityoflifedidnotappeartobemoreinformativethananalysisofasingleoverallGHQ-12score.Assuch,fromapragmaticpointofviewweconsideritacceptabletousethisinstrumentasaone-dimensionalmeasure.Unlessonehasspecificquestionsthatarebestansweredbyasubsetofthethreefactors,thereisnoneedtoconsiderthemulti-dimensionality. ReferencesUstunTB,Ayuso-MateosJL,ChatterjiS,MathersC,MurrayCJ:Globalburdenofdepressivedisordersintheyear2000.BrJPsychiatry2004,184:386–392.10.1192/bjp.184.5.386CAS  PubMed  Article  GoogleScholar  WigginsRD,SchofieldP,SackerA,HeadJ,BartleyM:SocialpositionandminorpsychiatricmorbidityovertimeintheBritishHouseholdPanelSurvey1991–1998.JEpidemiolCommunityHealth2004,58:779–787.10.1136/jech.2003.015958CAS  PubMedCentral  PubMed  Article  GoogleScholar  CortiL:Forbetterorworse?Annualchangeinsmoking,self-assessedhealthandsubjectivewell-being.InInChangingHouseholds:TheBritishHouseholdPanelSurvey1990–1992.Editedby:BuckN,GershunyJ,RoseD,ScottJ.Colchester:UniversityofEssex;1994:199–219. GoogleScholar  PolitiPL,PiccinelliM,WilkinsonG:Reliability,validityandfactorstructureofthe12-itemGeneralHealthQuestionnaireamongyoungmalesinItaly.ActaPsychiatrScand1994,90:432–437.CAS  PubMed  Article  GoogleScholar  AndrichD,vanSchoubroeckL:TheGeneralHealthQuestionnaire:apsychometricanalysisusinglatenttraittheory.PsycholMed1989,19:469–485.CAS  PubMed  Article  GoogleScholar  GraetzB:MultidimensionalpropertiesoftheGeneralHealthQuestionnaire.SocPsychiatryPsychiatrEpidemiol1991,26:132–138.CAS  PubMed  Article  GoogleScholar  MartinAJ:Assessingthemultidimensionalityofthe12-itemGeneralHealthQuestionnaire.PsycholRep1999,84(3Pt1):927–935.CAS  PubMed  Article  GoogleScholar  WorsleyA,GribbinCC:Afactoranalyticstudyofthetwelve-itemGeneralHealthQuestionnaire.AustNZJPsychiatry1977,11:260–272.CAS  PubMed  Article  GoogleScholar  WernekeU,GoldbergDP,YalcinI,UstunBT:ThestabilityofthefactorstructureoftheGeneralHealthQuestionnaire.PsyMed2000,30:823–829.10.1017/S0033291799002287CAS  Article  GoogleScholar  CheungYB:Aconfirmatoryfactoranalysisofthe12-itemGeneralHealthQuestionnaireamongolderpeople.IntJGeriatrPsychiatry2002,17:739–744.10.1002/gps.693PubMed  Article  GoogleScholar  KalliathTJ,O'DriscollMP,BroughP:AconfirmatoryfactoranalysisoftheGeneralHealthQuestionnaire-12.StressandHealth2004,20:11–20.10.1002/smi.993Article  GoogleScholar  FrenchDJ,TaitRJ:MeasurementinvarianceintheGeneralHealthQuestionnaire-12inyoungAustralianadolescents.EurChildAdolescPsychiatry2004,13:1–7.10.1007/s00787-004-0345-7PubMed  Article  GoogleScholar  CampbellA,WalkerJ,FarrellG:ConfirmatoryfactoranalysisoftheGHQ-12:canIseethatagain?AustNZJPsychiatry2003,37:475–483.10.1046/j.1440-1614.2003.01208.xPubMed  Article  GoogleScholar  GoldbergDP,WilliamsP:AUser'sGuidetotheGeneralHealthQuestionnaire.Windsor:NFER-Nelson;1988. GoogleScholar  BeckAT,EpsteinN,BrownG,SteerRA:Aninventoryformeasuringclinicalanxiety:psychometricproperties.JConsultClinPsychol1988,56:893–897.10.1037//0022-006X.56.6.893CAS  PubMed  Article  GoogleScholar  WareJE,SnowKK,KosinskiM,GandekB:SF-36HealthSurveyManualandInterpretationGuide.Boston:TheHealthInstitute,NewEnglandMedicalCenter;1993. GoogleScholar  ChanDW,ChanTS:Reliability,validityandthestructureoftheGeneralHealthQuestionnaireinaChinesecontext.PsycholMed1983,13:363–371.CAS  PubMed  Article  GoogleScholar  PanPC,GoldbergDP:AcomparisonofthevalidityofGHQ-12andCHQ-12inChineseprimarycarepatientsinManchester.PsycholMed1990,20:931–940.CAS  PubMed  Article  GoogleScholar  ChanDW:TheChineseversionoftheGeneralHealthQuestionnaire:doeslanguagemakeadifference?PsycholMed1985,15:147–155.CAS  PubMed  Article  GoogleScholar  LuoN,FonesCS,ThumbooJ,LiSC:Factorsinfluencinghealth-relatedqualityoflifeofAsianswithanxietydisordersinSingapore.QualLifeRes2004,13:557–565.10.1023/B:QURE.0000018484.89711.e2PubMed  Article  GoogleScholar  McHorneyCA,WareJEJr,LuJF,SherbourneCD:TheMOS36-itemShort-FormHealthSurvey(SF-36):III.Testsofdataquality,scalingassumptions,andreliabilityacrossdiversepatientgroups.MedCare1994,32:40–66.CAS  PubMed  Article  GoogleScholar  LamCLK,GandekB,RenXS,ChanMS:TestsofscalingassumptionsandconstructvalidityoftheChinese(HK)versionoftheSF-36HealthSurvey.JClinEpidemiol1998,51:1139–1147.10.1016/S0895-4356(98)00105-XCAS  PubMed  Article  GoogleScholar  ThumbooJ,FengPH,SohCH,BoeyML,ThioST,FongKY:ValidationoftheChineseSF-36forqualityoflifeassessmentinpatientswithsystemiclupuserythematosus.Lupus2000,9:708–712.10.1191/096120300673421268CAS  PubMed  Article  GoogleScholar  ThumbooJ,FongKY,MachinD,ChanSP,LeongKH,FengPH,ThioST,BoeML:AcommunitybasedstudyofscalingassumptionsandconstructvalidityoftheEnglish(UK)andChinese(HK)SF-36inSingapore.QualLifeRes2001,10:175–188.10.1023/A:1016701514299CAS  PubMed  Article  GoogleScholar  ThumbooJ,FongKY,ChanSP,MachinD,FengPH,ThioST,BoeyML:TheequivalenceofEnglishandChineseSF-36versionsinbilingualSingaporeChinese.QualLifeRes2002,11:495–503.10.1023/A:1015680029998PubMed  Article  GoogleScholar  MarshHW,BallaJR,HauKT:Anevaluationofincrementalfitindices:aclarificationofmathematicalandempiricalproperties.InInAdvancedStructuralEquationModeling:IssuesandTechniques.Editedby:MarcoulidesGA,SchumackerRE.Mahwah,NJ:LawrenceErlbaumAssociates;1996:315–354. GoogleScholar  HuL,BentlerP:Cutoffcriteriaforfitindexesincovariancestructureanalysis:conventionalcriteriaversusnewalternatives.StructuralEquationModeling1999,1:1–55.Article  GoogleScholar  MarshHW,HauKT,WenZ:Insearchofgoldenrules:commentonhypothesis-testingapproachestosettingcutoffvaluesforfitindexesanddangersinovergeneralizingHuandBentler's(1999)findings.StructuralEquationModeling2004,11:320–341.10.1207/s15328007sem1103_2Article  GoogleScholar  NormanGR,SloanJA,WyrwichKW:Interpretationofchangesinhealth-relatedqualityoflife:theremarkableuniversalityofhalfastandarddeviation.MedCare2003,41:582–592.10.1097/00005650-200305000-00004PubMed  GoogleScholar  DownloadreferencesAuthorinformationAuthorsandAffiliationsClinicalTrialsandEpidemiologicalSciences,NationalCancerCentre,11HospitalDrive,169610,SingaporeFeiGaoInstituteofHealthEconomics,#1200,10450JasperAvenue,Edmonton,Alberta,T5J3N4,CanadaNanLuoDepartmentofRheumatologyandImmunology,SingaporeGeneralHospital,OutramRoad,169608,SingaporeJulianThumbooDepartmentofPsychologicalMedicine,NationalUniversityofSingapore,5LowerKentRidgeRoad,119074,SingaporeCalvinFonesDepartmentofPharmacy,NationalUniversityofSingapore,18ScienceDrive4,117543,SingaporeShu-ChuenLiMRCTropicalEpidemiologyGroup,IDEU,ITD,LondonSchoolofHygieneandTropicalMedicine,London,WC1E7HT,UKYin-BunCheungAuthorsFeiGaoViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarNanLuoViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarJulianThumbooViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarCalvinFonesViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarShu-ChuenLiViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarYin-BunCheungViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarCorrespondingauthorCorrespondenceto Yin-BunCheung.AdditionalinformationAuthors'contributionFGcarriedouttheconfirmatoryfactoranalysis,interpretedthefindings,anddraftedpartofthemanuscript.NLdesignedthestudy,participatedinthedevelopmentofthestatisticalframeworkandinterpretedthefindings.JTparticipatedinthestudydesign,discussionofthestatisticalframework,andtheinterpretationoffindings.CFparticipatedinthestudydesignandcarriedoutthedatacollectionandclinicalassessments.SCLparticipatedinthestudydesignanddiscussionandinterpretationoffindings.YBCconceivedofthestudy,developedthestatisticalframework,carriedoutpartofthestatisticalanalysis,anddraftedpartofthemanuscript.Allauthorsreadandapprovedthefinalmanuscript.Authors’originalsubmittedfilesforimagesBelowarethelinkstotheauthors’originalsubmittedfilesforimages. Authors’originalfileforfigure1RightsandpermissionsReprintsandPermissionsAboutthisarticleCitethisarticleGao,F.,Luo,N.,Thumboo,J.etal.Doesthe12-itemGeneralHealthQuestionnairecontainmultiplefactorsanddoweneedthem?. HealthQualLifeOutcomes2,63(2004).https://doi.org/10.1186/1477-7525-2-63DownloadcitationReceived:26October2004Accepted:11November2004Published:11November2004DOI:https://doi.org/10.1186/1477-7525-2-63SharethisarticleAnyoneyousharethefollowinglinkwithwillbeabletoreadthiscontent:GetshareablelinkSorry,ashareablelinkisnotcurrentlyavailableforthisarticle.Copytoclipboard ProvidedbytheSpringerNatureSharedItcontent-sharinginitiative KeywordsGHQfactorstructurepsychologicalhealth DownloadPDF Advertisement HealthandQualityofLifeOutcomes ISSN:1477-7525 Contactus Submissionenquiries:[email protected] Generalenquiries:[email protected]