The factor structure of the 12-item general health ...

The 12-item General Health Questionnaire (GHQ-12) is a commonly used screening instrument for measuring mental disorders. Skiptomaincontent Advertisement SearchallBMCarticles Search Thefactorstructureofthe12-itemgeneralhealthquestionnaire(GHQ-12)inyoungChinesecivilservants DownloadPDF DownloadePub DownloadPDF DownloadePub Research OpenAccess Published:26September2016 Thefactorstructureofthe12-itemgeneralhealthquestionnaire(GHQ-12)inyoungChinesecivilservants YingLiang1,LeiWang2&XicanYin1  HealthandQualityofLifeOutcomes volume 14,Article number: 136(2016) Citethisarticle 13kAccesses 57Citations 1Altmetric Metricsdetails AbstractBackgroundThe12-itemGeneralHealthQuestionnaire(GHQ-12)isacommonlyusedscreeninginstrumentformeasuringmentaldisorders.However,fewstudieshavemeasuredthementalhealthofChineseprofessionalsorexploredthefactorstructureoftheGHQ-12throughinvestigationsofyoungChinesecivilservants.MethodThisstudyanalysesthefactorstructureoftheGHQ-12onyoungChinesecivilservants.Respondentsinclude1051participantsfromsixcitiesineasternChina.ExploratoryFactorAnalysis(EFA)isusedtoidentifythepotentialfactorstructureoftheGHQ-12.ConfirmatoryFactorAnalysis(CFA)modelsofpreviousstudiesarereferredtoformodelfitting.ResultsTheresultsindicatetheGHQ-12hasverygoodreliabilityandvalidity.AlltenCFAmodelsarewellfittedwiththeactualdata.ConclusionAllthetenmodelsarefeasibleandfitthedataequallywell.TheChineseversionoftheGHQ-12issuitableforprofessionalgroupsandcanserveasascreeningtooltodetectanxietyandpsychiatricdisorders. BackgroundThe12-itemGeneralHealthQuestionnaire(GHQ-12)originatedfromageneralhealthquestionnairedesignedbyGoldberg,whichreflectsthementalhealthconditionofrespondentsby12-itemself-assessmentresults.Ithasbeenwidelyappliedtoclinicalpatient’psychologicalassessmentandself-assessmentofthegeneralpopulation[1].LittleresearchhasbeeninitiatedtosurveyChineseoccupationalgroupsbytheGHQ-12;meanwhile,noresearchershavedetectedthefactorstructureoftheGHQ-12viainvestigationsofthehighanxiety-riskoccupationalgroups,includingthoseyoungcivilservantsworkingatgrassrootsgovernmentagencies.Inthispaper,theGHQ-12isusedinthemeasurementofyoungChinesecivilservants’mentalhealthtoanalysethefactorstructureofGHQ-12inthisoccupationalgroup.ApplicationofGHQ-12GeneralHealthQuestionnaire(GHQ),developedbyBritishscholarGoldbergin1972,isoneofthemostpopularandwidelyusedscreeninginstrumentsforrecognitionandmeasurementofmentalhealth[2].TheoriginalGHQconsistsof60items,andnowtherearemultipleversions,includingGHQ-1,GHQ-12,GHQ-20,GHQ-28andGHQ-30.Amongthese,theGHQ-12isthemostpopularduetoitssimplicity[2,3].TheGHQ-12consistsof12items,eachofwhichisevaluatedbyfourindexes.Thetwomostcommonlyusedscoringtypesarethebi-modal(0-0-1-1)andLikertscoringmethods(0-1-2-3)[4].TheGHQ-12hassatisfactoryreliability[5–7]andgoodsensitivityandspecificity[8,9].Ithasbeenappliedtodifferentpopulationsindifferentcountriestostudyitsreliabilityandvalidity,andtoexplorethementaldisorderscharacteristicsofthepopulationsinpreviousstudies.Theserespondentshaveincludedbothadolescents[4,7,10]andadultcommunityresidents[1,11].Numerousstudieshavefocusedonoccupationalgroups,includingnurses[12],academicstaff[13],cohort[11]andcivilservants[14]etc.Simultaneously,differentlanguageversionsoftheGHQ-12haveprovenacceptable,(e.g.,Arabic[9],AustralianEnglish[13],German[2],Spanish[10]andSwedish[15]).AlthoughfewstudieshavefocusedontheapplicabilityoftheChineseversionoftheGHQ-12,ithasbeenproventhattheChineseversionisreliableandvalid[16,17].ScholarshaveappliedtheChineseversionoftheGHQ-12touniversitystudents[16],andtheresultshaveshownthattheinstrumentisacceptable.Ithasalsobeenusedtoevaluatethementaldisordersofsecondaryschoolstudents,reachingthesameconclusion[17].Thestudiesmentionedabovewerefocusedonstudents,butthereisscarceliteratureontheapplicabilityoftheGHQ-12tospecificChineseoccupationalgroups.AlthoughpreviousstudieshaveusedtheGHQ-12tomeasurethementalhealthofcivilservantsandnurses,noresearchhasbeendoneonyoungcivilservantsingrassrootsgovernmentagencies,agroupthathasspecificoccupationalandphysicalcharacteristics.MentaldisordersofcivilservantsVariouspopulationsinChinasufferedfrommentaldisorders[18–21].Amongthese,civilservantsareprofessionalswithahighriskofmentaldisorders[22].Civilservantsareresponsibleformanagingthestate,andtheylinkaveragecitizensandthegovernment[23].Thenatureoftheirworkandtheorganisationalenvironmentdeterminesthespecificityoftheirwork,andtheiroccupationalhealthhasspecificcharacteristics.Infact,previousstudieshaveshownthatthehealthconditionofthisoccupationalgroupisnotoptimistic,andevencanbeworrisome [24].Civilservantsareunderhighwork-relatedpressure,resultinginpoormentalandphysicalhealthandlackofsupport[25,26].Inaddition,theirhigh-intensityworkmakesthemmorevulnerabletodisease.Mostyoungcivilservantswhohaverecentlyenteredtheexecutiveauthoritiesareinlower-levelpositions.Marmot[27]conductedastudyfindingthatmortalityrateduetocoronaryheartdiseaseofmalecivilservantsinlow-levelpositionswasthreetosixtimeshigherthanthatofmalecivilservantsinhigherpositionsinLondon.Civilservantsinlow-levelpositionshavebeenemployedforashorttime [28],havesignificantemploymentpressure [29],tendnottoplaysports andaremoresusceptibletohighbloodpressureandbloodsugar [30].Inaddition,negativehealthbehavioursandhabitsamongcivilservantsarequitecommon,includingaddictedtotobaccoandalcohol,lackofexerciseandatendencytowardchronicdiseaseslikeinsomnia[31–34].Thus,concernsaboutthehealth,especiallythementaldisorders,ofspecialoccupationalgroupsareofgreatimportanceandurgency.ExistingfactorstructuresoftheGHQ-12Atpresent,theexistingresearchesonthefactorstructureoftheGHQ-12havebeenfullofcontroversy,astheresultshavebeeninconclusive.Themajorityofpreviousstudiesextractedonetothreefactorsfrom12items.Initially,theGHQ-12wasdesignedasaunidimensionalscalethatallprojectsincludedinamainfactor,butonlyafewscholarssupportedthisoriginalone-factormodelinsubsequentempiricalstudies.Banks[5]andhiscolleaguesconductedthreeseparateinvestigations,totestthevalidityforthreetypesofsamples:employeesofanengineeringfirm,recentleaversandunemployedmen.TheresultsshowedthatthescalingpropertiesoftheGHQ-12aresufficientlygoodtojustifytheuseofasingle-scalescore.However,theoriginalunidimensionalmodelmaybeunabletoprovideresearcherssufficientinformationbecauseofitssimplicity[5].Sincethen,anumberofstudieshaveexploreditspossiblefactorstructuresandrevealeddifferentdimensionsinthestructureoftheGHQ-12,mainlywithtwoorthreefactors.AsurveyofteachersinWesternAustraliafoundthattheGHQ-12containstwodimensions,andtheyweredescribedseparatelyinastatementbypositiveandnegativewords[35].Thisresultalsorepeatedlyshowedtheadequatefittothedata[15,36,37].Inanotherstudy,usingalargesampleofyoungAustralians(n = 8998),Graetz[10]extractedthreefactorsfromtheGHQ-12:SocialDysfunction(includingitems1,3,4,7,8and12),AnxietyandDepression(includingitems2,5,6and9),LossofConfidence(includingitems10and11).Withtheapplicationoftheconfirmatoryfactoranalysis,manystudieshavesuggestedthatthisthree-factormodel,comparedtoothermodels,hasabettergoodnessoffit[7,38,39].Simultaneously,otherscholarshavealsofoundthattheGHQ-12hasthreedimensions[9,40],butnamedthefactorsdifferentlyfromGraetz[10].However,thevalidityandusefulnessofthesemulti-dimensionalsolutionshavefrequentlybeenquestionedbecauseofthehighdegreeofcorrelationbetweenthefactors[7,41].Otherstudieshavefocusedontheconnectionsbetweenwordingeffectandnegativevocabulary.Hankins[12]foundthroughstatisticsfromtheHealthSurveyforEnglandthattheunidimensionalmodel,withcorrelatederrorsonthenegativelywordeditems,wassuperiortothetwo-factor(positivelyandnegativelywordeditems)orthree-factormodelsproposed[42].ThestudyresultsofLi[17]andAguadoareconsistentwiththis.Bothcontendthataslongasthewordingeffectiscontrolled,theGHQ-12isaunidimensionalmodel[17,43].Ingeneral,althoughtherehavebeenmanystudiesonthefactorstructureoftheGHQ-12,theyhavenotreachedaunanimousconclusion.ThusexplorationandverificationofthefactorstructureoftheGHQ-12arestillnecessary.ResearchobjectivesAsitisaconstantlyimprovedandwidelyusedmeasurementtoolformentalhealth,theGHQ-12hasbeenappliedtodifferentpopulationsinmanycountries.Theresearchsubjectshavemostlybeenmembersofthegeneralpopulationandcertainoccupationalgroups.Nopaperthathasmadeitssampleyoungcivilservantsworkingatgrassrootsgovernmentagencies.Whatismore,theChineseversionoftheGHQ-12hasnotbeenresearchedextensively;fewstudieshaveexaminedthefactorstructureandpsychologicalcharacteristicsofthisscaleinChineseoccupationalgroups.Inthispaper,weusetheGHQ-12tomeasurethementalhealthofyoungChinesecivilservantsandtoanalysetheGHQ-12factorstructureinthisoccupationalpopulation.Thisstudyhasthefollowingthreeaims:(1)CreativelyapplyingtheGHQ-12toyoungcivilservantsworkingingrassrootsgovernmentagenciestobroadenthescopeofuseandacceptanceoftheGHQ-12;(2)ResearchingtheGHQ-12factorstructureofoccupationalgroupsinChinatoprovidemoreextensiveclinicalandempiricaldatafortheimprovementanddevelopmentoftheGHQ-12;(3)ThroughstudyingthereliabilityandvalidityoftheGHQ-12topromoteitsuseinChineseclinicalmeasurementsandindividualself-assessment.MethodsSamplesThesubjectsofthisstudyareyoungcivilservantsworkingat24grassrootsgovernmentagencies.Wedesignedandconductedacross-sectionalstudyandselectedsixcitieswithcomparativelydevelopedeconomyineasternChina(Nanjing,Shanghai,Suzhou,Hangzhou,Yangzhou,Wenzhou).Thesecitiesareunderbetterurbanconstruction,andtheyhavemorecompleteanddiversegovernmentagencies,sotheymakerepresentativesampleregions.Werandomlyselectedcivilservantsunder45 yearsoldworkinginadministrativedepartmentsbelowthecountylevel.Intheprocessofdistributingthequestionnaire,weaskedwhetherrespondentswerewillingtoaccepttheinvestigationandstateprinciples,suchastheconfidentialityandauthenticity,inadvance.Respondentswereaskedtocompletethequestionnaireindependently.Iftheydidnotunderstandanyitem,theycouldconsulttheinvestigatoronthespot.Finally,1200questionnairesweredistributed,(i.e.,200foreachcity).Atotalof1051questionnaireswerereturned,makingtheresponserate87.58 %.InstrumentsThisstudyusedtheGHQ-12proposedbyGoldberg[44]tomeasurethementalhealthofyoungChinesecivilservants.Thisquestionnaireincludes12items(sixpositivelywordeditems(e.g.,Haveyoubeenabletoconcentrateonwhateveryouaredoing?)andnegativelywordeditems(e.g.,Haveyoulostmuchsleepoverworry?)).Weadoptedthefour-pointLikertscale,witheachitemrangingfrom0to3.Fornegativelywordeditems,‘0’indicatedNotatall,‘1’indicatedSeldom,‘2’indicatedUsualand‘3’indicatedMorethanusual,whilepositivelywordeditemswerereverselyscored.Allitemswereaddedtoobtainthetotalscore,makingthescorerange0-36(withahigherscoreindicatingworsementalhealth).Scoresoverthecut-offpointof12couldbeclassifiedascases[8].StatisticalanalysisFirst,thescoredistributionsforeachitemandthewholescalewerecalculatedtounderstandthementalhealthofyoungChinesecivilservants.Then,weconductedareliabilityanalysisoftheGHQ-12scale.Finally,weusedfactoranalysistoexploreandvalidatetheGHQ-12factorstructureoftherespondents.Inthefactoranalysisprocess,thetotalsample(n = 1051)wasrandomlysplitintoanExploratoryFactorAnalysis(EFA)sample(n = 525)andaConfirmatoryFactorAnalysis(CFA)sample(n = 526)bySPSS19.0.Then,EFAwasconductedtoexploretheunderlyingfactorstructureoftheGHQ-12intheEFAsampleusingtheprincipalaxisextractionmethodwithvarimaxrotations.ParallelanalysiswasthemethodusedtodecidethenumberoffactorsextractedfromEFA[45].CFAwithmaximumlikelihoodestimation(MLR)wasreferredtoformodelfit.TheselectionofthemodelsforCFAwasbasedonaliteraturereview.Wereferredtoalargenumberofpreviousstudiesandselectedthemodelwiththebestfitfromeachvalidationstudy.Asmentionedintheintroduction,mostofthestudiesextractedonetothreefactorsfrom12items.ThemodelsmoststronglysupportedtheunidimensionalmodelwithwordingeffectbyHankins[41],AndrichandVanSchaubroecks’two-dimensionalmodel[35]andthethree-dimensionalmodelbyGraetz[10],whichhavebeensupportedandvalidatedbymanyotherscholars.Furthermore,anadditionalfivetwo-dimensionalandthree-dimensionalmodels[6,9,40,46,47]werealsoincluded,whichhavebeenproventofitwellbypreviousstudies.Finally,couplingtheunidimensionalmodelfromtheoriginalGHQ-12designedbyGoldbergandourmodelresultsofEFA,CFAwasthenperformedbytestingtencompetingfactormodelswithone-,two-andthree-dimensionalsolutionsoftheCFAsample.Thetentypesofmeasurementmodelswereasfollows:(1)unidimensionalmodel(original),(2)unidimensionalmodelwithcorrelatederrorsonthenegativelywordeditemsbyHankins[41],(3)two-dimensionalmodelproposedbyAndrichandVanSchaubroeck[35](positivelywordeditems(includingitems1,3,4,7,8and12)andnegativelywordeditems(includingitems2,5,6,9,10and11)),(4)two-dimensionalmodelbySchmitzetal.[46](Anxiety/Depression(includingitems1,2,6,7,10and11),SocialPerformance(includingitems4,5,8,9and12)),(5)two-dimensionalmodelbyPolitietal.[6](Dysphoria(includingitems2,5,6,9,10and11)andSocialDysfunction(includingitems1,3,4,7and8),(6)three-dimensionalmodelbyGraetz[10](SocialDysfunction(includingitems1,3,4,7,8and12),AnxietyandDepression(includingitems2,5,6and9),LossofConfidence(includingitems10and11),(7)three-dimensionalmodelbyFarrell[47](Anxiety(includingitems2,5,10,11and12),Depression(includingitems1,6,7,8and9)andSocialDysfunction(includingitems3and4),(8)three-dimensionalmodelbyDaradkehetal.[9](GeneralDysphoria(includingitems5,6,9,10and11),LackofEnjoyment(includingitems1,2,7,8and12)andSocialDysfunction(includingitems3and4),(9)three-dimensionalmodelbyMartin[40](Cope(includingitems1,3,4,8and11),Stress(includingitem2,5and7)andDepression(item6,9and12),(10)unidimensionalmodelobtainedfromEFA.ThereasonsweconductedEFAbeforeCFAareasfollows:First,previousstudiesonthefactorstructureoftheGHQ-12havebeenfullofcontroversy,andtheyhavebeeninconclusive.Thenamesofeachfactorinthesamedimensionsmodelalsovaryintheexistingfactordimensionalmodel.Second,sofar,nostudieshaveconductedEFAonChineseoccupationalgroups,althoughYe[16]andLi[17]haveappliedEFAtoChinesestudents.Last,tothebestofourknowledge,thisisthefirststudyoftheGHQ-12factorstructureonyoungChinesecivilservants.Youngcivilservantsareaspecificoccupationalgroupatahighriskofmentaldisorders.Theybareagreaterworkpressureandhigherpropensityformentaldisorders[25,26].Theresultsofthegeneralpopulationorotheroccupationalgroupsmaynotbeapplicabletoyoungcivilservants.WeusedEPIDATA3.1doubleentrytotestthesamplequality.Thesignificancelevelforalltestswas0.05.WeperformedtheCFAmodelwithAMOS21.0andEFAmodelsandotherstatisticaltestswithSPSS19.0.Thus,weadoptedthisshortanalysistoexplorethereliabilityandvalidityoftheGHQ-12onyoungChinesecivilservants.ResultsDemographicsThedemographicandsocioeconomiccharacteristicsofrespondentsmainlyincludedgender,age,education,employmentlevelandmaritalstatus.Ofthe1051respondents,61.5 %weremale,and38.5 %werefemale.Intermsofage,41.5 %ofrespondentswereaged20–29years,48.4 %ofrespondentswereaged30–39and10.1 %wereaged40–45.Educationlevelsoftheseyoungcivilservantsweredividedintofourcategories:Juniorcollegeandbellow,Undergraduate,GraduateandDoctorate.MostrespondentsfellintotheJuniorcollegeandbelow(32.7 %)andUndergraduate(42.6 %)categories,andmostrespondentswerestaffandbelow(41.4 %),andnotmarried(46.8 %).Inaddition,theEFAsampleandtheCFAsamplehadsimilardistributionsofage,educationandemploymentgrade.TheCFAsamplehadahighratioofrespondentswhoweremaleormarried.Ingeneral,thedemographicsoftherespondentsarerepresentativeoftheyoungChinesecivilservants.Table 1providesdescriptivestatisticsofthesociodemographiccharacteristicsoftherespondentsinthisstudy.Table1SociodemographiccharacteristicsofparticipantsFullsizetable DescriptiveandhealthcharacteristicsTable 2showstheoverallandindividualitemscoresoftheGHQ-12.TheGHQ-12itemsweretotheleftonbothsidesofthescoredistributionexceptitem10,andthekurtosiscoefficientswerelessthan0.TheaveragescoreoftheGHQ-12was23.62(SD = 7.92),whichwasfarhigherthanthecut-offpointof12[8],andofalltherespondents,86.49 %scoredgreaterthanorequalto12andcouldbeusedascases.Thisindicatedthatthehealth,especiallythementalhealth,oftherespondentswasinverypoorcondition.Inparticular,thehighestaveragescoreswereforitems1,9and11,whichweremorethan2.10.Oftheseitems,theaveragescoreofitem9was2.14(SD = 1.08),asthehighest,indicatingthatthemajorityofrespondentsfeltunhappyanddepressed.Whiletheaveragescoreofitem12was2.06(SD = 1.06),themajorityofrespondents(74.3 %)scored2or3points,andonly13.8 %ofrespondentsscored0,showingthatingeneraltherespondentsdidnotfeelhappy.Table2DescriptivestatisticsforGHQ-12itemsandsummaryscores(N = 1051)Fullsizetable ReliabilityandcorrelationsanalysisofGHQ-12Cronbach’salphareliabilitywasobtainedusingtheSpearman-Brownformula.Table 3showsthereliabilityandcorrelationsbetweenitemsandfortheoverallGHQ-12scale.Cronbach’salphaoftheGHQ-12was0.844(>0.8),indicatingthatthescalehasagoodreliabilityandcorrelations.Item11hadthehighestcorrelationcoefficient,whichwas0.609.Correlationcoefficientsfortherestoftheitemsandthetotalscorerangedfrom0.324to0.606.TheCronbach’salphacoefficientswhenanitemwasdeletedwereover0.8,butbelowthetotalcoefficients,theyrangefrom0.824to0.837whenanitemwasdeleted,indicatingthateachitemwasnecessaryandofequalimportance.Table3CorrelationsbetweenitemsandforoverallGHQ-12scaleFullsizetable FactoranalysisPreviousstudiesonthefactorstructureoftheGHQ-12havebeenfullofcontroversy,andtheyhavebeeninconclusive.Inthefactoranalysisprocess,thetotalsample(n = 1051)wasrandomlysplitintotheEFAsample(n = 525)andtheCFAsample(n = 526)bySPSS19.0.UsingthedatafromtheEFAsample,EFAwasfirstlyusedtoidentifythepossiblelatentvariablesoftheGHQ-12.Table 4showstheresultsoftheEFAofthescaleusingtheprincipalaxisextractionmethodwithvarimaxrotation.Kaiser-Meyer-Olkin(KMO)measuredthesamplingadequacyoftheGHQ-12as0.908(>0.9),andtheBartletttestresultspassedthesignificancetest(approximatedchi-square = 1249.186,df = 66,P