Integration of oral health-related quality of life instruments

Integration

of

oral

health-related

quality

of

life

instruments

Mike

T.

John

*,

Daniel

R.

Reissmann

,

Asja 

Celebic

,

Kazuyoshi

Baba

,

Dóra

Kende

,

Pernilla

Larsson

,

Ksenija

Rener-Sitar

a_{DepartmentofDiagnosticandBiologicalSciences,UniversityofMinnesota,7-536MoosTower515DelawareStreetSE,Minneapolis,MN55455,USA} b_{DepartmentofProstheticDentistry,CenterforDentalandOralMedicine,UniversityMedicalCenterHamburg-Eppendorf,Hamburg,Germany} c

DepartmentofProsthodontics,UniversityofZagreb,andSchoolofDentalMedicineandClinicalHospitalCentre,Zagreb,Croatia

d

DepartmentofProsthodontics,ShowaUniversity,Tokyo,Japan

e

DepartmentofProsthodontics,UniversityofPécs,Pécs,Hungary

f

CentreofOralRehabilitation,ProstheticDentistry,Norrköping,Sweden

g

DepartmentofProsthodontics,UniversityofLjubljana,Ljubljana,Slovenia

h_{DepartmentofProsthodontics,UniversityDentalClinics,UniversityMedicalCenterLjubljana,Ljubljana,Slovenia}

ARTICLE INFO

Articlehistory: Received18April2016

Receivedinrevisedform23June2016 Accepted24June2016

Keywords:

Oralhealth-relatedqualityoflife Measurement Questionnaire Cross-sectionalstudy Validity Dimensionality ABSTRACT

Objectives: To integrate items from two widely used oral health-related qualityof life (OHRQoL) questionnaires,theGeneral OralHealth Assessment Index(GOHAI) andtheOralImpacts onDaily Performances(OIDP),aswellasculturally-specificitemsoftheOralHealthImpactProfile(OHIP)intoa four-dimensionalOHRQoLmodelconsistingofOralFunction,OrofacialPain,OrofacialAppearance,and PsychosocialImpact.

Methods:SubjectscamefromanancillarystudyoftheDimensionsofOralHealth-RelatedQualityofLife Project(N=267patients,meanageSD:54.017.2years,58%women.)Patientsfilledintheoriginal49 itemsof OHIPand 22additional OHRQoL itemsina cross-sectionalstudy.Theseadditional items consistedof7culturallyspecificOHIPitemsand15GOHAIorOIDPitemswithuniquecontentnotcovered inOHIP-49.Beforedatacollection,threeexpertshypothesizedtowhichofthefourOHRQoLdimensions theseitemsbelong.Hypothesesweretestedincorrelationanalysesbetweenthe22itemsandthefour dimensionscoresthatwerederivedfromOHIP-49.

Results:Fiveofthe22itemsdidnotprovidesufficientinformationtowhichdimensiontheybelong.In16 oftheremaining17items,thepatternofcorrelationcoefficientsfittedexperts’apriorihypotheses. Acceptanceof16ofthe17hypotheseswasinterpretedasevidencethatadditional(notinOHIP-49 contained)OHRQoLitemscanbeassignedtoOralFunction,OrofacialPain,Orofacial Appearance,and PsychosocialImpact.

Conclusion:ItemsofthreeOHRQoLinstrumentscanbeintegratedintoadimensionalOHRQoLmodel consistingofOralFunction,OrofacialPain,OrofacialAppearance,andPsychosocialImpact.

Clinicalsignificance:OralFunction,OrofacialPain,OrofacialAppearance,andPsychosocialImpactcanserve asasimpleandclinicallyappealingsetoforalhealth-relatedqualityoflife(OHRQoL)dimensionsand thereforeprovideanopportunityforsimpler,butpsychometricallyimprovedOHRQoLmeasurementin thefuture.

ã2016ElsevierLtd.Allrightsreserved.

1.Introduction

Oralhealth-relatedqualityoflife(OHRQoL)isapopularconcept toassesstheimpactoforalconditionsandtheeffectsofdental interventions.ManyinstrumentsexisttomeasureOHRQoL.This situationatteststothedentalcommunity’sinterestinOHRQoL,but it also challenges comparabilityand communication of results.

Therefore, from a conceptual point of view, reduction of the numberofinstrumentsseemsnecessarytoimprove standardiza-tion for using OHRQoL assessment globally. In addition, to be confidentintheinstrumentscores,psychometricpropertiessuch asreliability,validity,responsiveness,andinterpretabilityneedto beknown.Fromapracticalpointofview,thiscanonlybeachieved foralimitednumberofquestionnaires.

However,insteadof abandoningsomeOHRQoL instruments, combining them with other instruments – a synthesis of questionnaires – seems a promising strategy that can achieve betterquality instrumentswhile reducing their quantity.Some

*Correspondingauthor.

E-mailaddress:mtjohn@umn.edu(M.T.John).

http://dx.doi.org/10.1016/j.jdent.2016.06.006

0300-5712/ã2016ElsevierLtd.Allrightsreserved.

ContentslistsavailableatScienceDirect

Journal

of

Dentistry

studieshavealreadyutilizedseveralOHRQoLinstrumentstogether

[1–4]andotherstudieshaveevensubmittedthemasacombined measure ofOHRQoL in analyses [5,6]. However,a moreformal assessmentwhetherinstrumentscanbecombinedhasnotbeen performed yet. The prerequisite for a synthesis would be that instruments measure the same underlying latent attribute. Substantial correlations among OHRQoL questionnaire scores attest to this proposition empirically [2,4,6,7]. However, these results apply to the OHRQoL global measure, the instrument summaryscore,butinstrumentcombinationneedstobedoneon the dimension level because these are the attributes that are actuallymeasuredandOHRQoLisjusttheglobalumbrellaterm. Conceptually,thehypothesizedconstructsthat aremeasuredby theGeneralOralHealthAssessmentIndex(GOHAI)andtheOral Impacts on Daily Performances (OIDP) two frequently used OHRQoLquestionnaires aresimilartothefourdimensionsof OHRQoL(OralFunction,OrofacialPain,Orofacial Appearance,and PsychosocialImpact)measuredbyOHIP[8,9],themostwidelyused OHRQoLinstrument.Therefore,OHIP’sdimensionsmayserveas theframeworktointegratetheotherOHRQoLinstruments’items. Ifthispropositionwouldhold,GOHAIandOIDPitemsaswellasthe existing culture-specific OHIP items [10–12], i.e., additional indicatorsoftheOHRQoLconstructmeasuredwithOHIP,should fall into the four OHIP-based dimensions. A relevant target populationwhere sucha newgroupingof OHRQoLitemscould be studied would be prosthodontic patients. They could be consideredtypicaldentalpatientsbecausetheirnumberislarge, theyusuallysufferfromtoothlossresultingfromthetwomajor oraldiseases, caries and periodontitis,and several otherdental treatmentsareoftenperformedinconjunctionwithprosthodontic therapy[13].

Our study aimed to integrate items from two widely used OHRQoLquestionnaires,theGeneralOralHealthAssessmentIndex (GOHAI)andtheOralImpacts onDailyPerformances (OIDP),as wellasculturallyspecificitemsoftheOralHealthImpactProfile (OHIP)intoafour-dimensionalOHRQoLmodelconsistingofOral Function, Orofacial Pain, Orofacial Appearance, and Psychosocial Impact.

2.Methods

2.1.Studysetting,studydesign,andsubjects

ThisisanancillarystudyinitiatedwithintheDimensionsofOral Health-RelatedQualityofLife(DOQ)Project[13].TheDOQProject aims to identify OHIP’s dimensional structure, i.e., how many importantcomponentsthemultidimensionaloralhealth percep-tionhas.Theprojectanalyzes49-itemOHIP[14]datafromgeneral populationsubjectsandprosthodonticpatientsfromsixcountries (Croatia, Germany,Hungary,Japan,Slovenia, and Sweden)with validated OHIP instruments [11,12,15_–18]. The international collaboratorsofthe DOQProjectcame fromtheDepartmentof Prosthodontics,UniversityofZagreb,Croatia;theDepartmentof ProstheticDentistry,UniversityMedicalCenter Hamburg-Eppen-dorf,Germany;theDepartmentof Prosthodontics,Universityof Pécs,Hungary;theDepartmentofProsthodontics,Showa Univer-sity,Japan;theDepartmentofProstheticDentistry,Universityof Ljubljana,SloveniaandtheCentreofOralRehabilitation,Prosthetic Dentistry,Norrköping,Sweden.In each participatingcenter,we targetedaconsecutivesampleofprosthodonticpatientstostudya new,one-weekrecall periodforOHIPinacross-sectional study

[19].Inthisstudy,patientswereassessedontwooccasionswhen theirOHRQoLwas assumedtobestable.Specifically,they were assessedeithertwicebeforethestartofprosthodontictreatment ortwiceaftertheendoftreatment.Onaverage,twoweekselapsed betweenassessments.Theorderinwhichpatientscompletedthe

two forms was determined by random assignment. The study design is similartotest-retest studiesthat wereperformed for testing the psychometric properties for language-specific OHIP versionsinthesesixcountries.However,insteadofreceivingtwo OHIPswiththesamerecallperiod,oneOHIPformhadanew,7-day recallperiodandtheotherOHIPformhadthecommonlyused one-monthrecallperiod.Forthepresentstudy,onlytheOHIPdatawith the one-month recall period were used. For details see our previous publication[19]. InstitutionalReviewBoardsreviewed andapprovedthestudiesinCroatia(EthicalBoardoftheSchoolof DentalMedicine,UniversityofZagreb,05-PA-26-55/06,PDS-103/ 09-10), Hungary (Ethics Committee of the University of Pecs FacultyofMedicine,2009–3562),Germany(InstitutionalReview BoardoftheMedicalAssociationinHamburg,Germany,PV3530), Japan (Ethics Committees of Showa University, #2007-29), Slovenia (National Medical Ethic Committee of theRepublic of Slovenia,124/02/13),andSweden(RegionalEthicsReviewBoardat LinköpingUniversityHospital,72/04,M208-07).

2.2.Dataanalysis

WestudiedthemostpopularOHRQoLinstruments,i.e.,OHIP, GOHAI, and OIDP. The OHIP-49 served to identify OHRQoL dimensions. Seven additional, culture-specific OHIP items, 15 GOHAIand12OIDPitemswereinvestigatedwhethertheyfitinto theOHRQoLdimensions.Beforethestudy,threeexperts(KRS,DRR, MTJ) independently assessed the GOHAI and the OIDP item content’soverlapwithOHIPtominimizepatients’burdenandto improve complianceby avoidingthe useof very similaritems. Then,inadiscussiontheexpertsreachedaconsensus.Theexperts also separated themes in GOHAI and OIDP items when they appearedtogetherinaparticularquestion.Forexample,aGOHAI questionasks‘Howoftendidyouhavetroublebitingorchewing anykindsoffood,suchasfirmmeatorapples?’BecauseOHIPasks ‘Haveyouhaddifficultychewinganyfoodsbecauseofproblems with your teeth, mouth, jaw or dentures?’, only ‘biting’ was selectedasaseparatetheme/problemtoberetainedinanewitem with unique content. Takingculture-specific and original OHIP items’ content into account,12 GOHAI-related items,and then threeOIDP-relateditemshaduniquecontent.Togetherwiththe seven culture-specificOHIP items, they constituted a setof 22 OHRQoLitemswithadditional,notinOHIP-49capturedOHRQoL content.

Toavoidtheinfluenceofmethodseffects,weadaptedGOHAI andOIDPitemstotheOHIPframeworkanditsrecallperiod.For example,thepreviousGOHAIitemwasadaptedto‘Haveyouhad difficultybitinganyfoods becauseofproblemswithyourteeth, mouth, jawor denturesin thelast month?’ Availablelanguage versionsoftheGOHAI[1,4,20,21]andtheOIDP[22–24]wereused toguidetheadaptationprocess.Then,thethreeexpertsassigned the22itemstothefourOHIP-baseddimensionsofOralFunction, Orofacial Pain, Orofacial Appearance, and Psychosocial Impact. Reproducibilityofthismethodwasinvestigatedbeforeandfound tobeadequate[25].First,theexpertsdidthisindividuallyandthen synthesizedfindingsbyconsensus.Formostitems,assignmentasa functional, painful,aestheticalorpsychosocialproblemwas not difficult.Itemstheexpertscouldnotclearlyclassifyintooneofthe fourdimensionswereassignedtoPsychosocialImpact,because,in their opinion, this dimension represented the broader, more general impactfromoral conditions.Anexamplewould bethe culture-specificOHIPitem‘bitingcheeks’wheretheexpertscould notdecidewhetheritwasafunctionalorapainfulproblem.This proceduregenerated22hypothesesaboutOHRQoLitems belong-ingtooneofthefourOHRQoLdimensions.

Basedontheresults fromtheDOQProject [8],asa marker for these dimensions, we calculated a dimension score

averagingtheitemsperdimensionsothatthedimensionscore wouldbeonthesame0–4scaleasthe22itemresponses.Then, we correlated dimension scores and item responses using a Spearman rank correlation coefficient, resulting in four coef-ficientsper item.

Weperformedtheitem-to-dimensionclassificationsimilarlyas inexploratoryfactoranalysis(EFA),exceptforthedifferencethatin EFAthelatentdimensionsarenotknown;inourstudy,wehadfour alreadyexistingdimensions.InEFA,afterextractionofacertain numberofunknownfactors,theitems’correlations(loadings)with thesefactorsareinspectedtoinformaboutthedimensions.Only salientitems,i.e.,itemsexceedingacertainthreshold,arefurther consideredbecauseonlytheyareconsideredinformative.Weused inapreviousEFAacriterionof0.45asathresholdforsaliency[9]. Therefore,inthepresentstudyonlyitemswithacorrelationof 0.45wereconsideredinformativeaboutacertaindimensionand includedinhypothesistesting.Theitem’shighestvalueamongthe correlationswitheachofthefourdimensionswasconsideredan indicationtowhich dimensionthisitem belonged. When data-basedandhypothesis-baseddimensionalassignmentagreed,our hypothesiswasprovedandaccepted.Overall,toacceptorreject ouroverarchinggoalthatthesetofitemsfromthetwoOHRQoL instrumentsandtheadditionalOHIPitemscanbeintegratedinto the four dimensions, we expected that 75% of our individual hypotheseswouldbeaccepted.

3.Results

3.1.Characterizationofprosthodonticpatientsinsixcountries

Sociodemographicandclinicalcharacteristicsofthe267adult patientsfromsixinternationalprosthodonticstreatmentcenters varied substantially (Table 1). Overall, patients’ mean age was 54.0years.Fifty-eightpercentofthepatientswerewomen.The majorityofpatientshad acombination offixedprosthodontics andtheirownnaturalteeth(57%),butnumbersvariedfrom18% inJapan, upto96%in Slovenia.Abouthalf ofthestudy’stotal patientswereassessedaftertreatment,withSloveniahavingthe lowest number (32%) and Japan having the highest (68%). OHRQoLimpairmentdifferedacrosscountries,fromaminimum of18.2OHIPpointsinSwedenupto49.8OHIPpointsinGermany (average of 37.2 OHIP points). These differences in OHRQoL impairmentacrosscenterswereclinicallyrelevantascompared toOHIP-49’s Minimal Important Difference of six points [26]. Whentranslatedintoanabsoluteproblemfrequencyaccordingto Reissmannetal.[27],betweennine(Sweden)and25(Germany) impactsperdayoverthelastmonthoccurredinthesixtreatment centers.

3.2.Hypothesestesting

Fiveitemsdidnotreachthethresholdforsalientcorrelations (Table2).‘Swallowcomfortably’hadverylowcorrelationsacross dimensions. ‘Avoid laughing’ had low correlations of similar magnitudetoOralFunction,OrofacialAppearance,andPsychosocial Impactand,asexpected,averylowcorrelationwithOrofacialPain. ‘Jointnoises’, ‘dry mouth’, and ‘biting cheeks’ correlatedrather equally,butlowwithallfourdimensions.

Forthe17remainingitems,onehypothesiswasrejected.The item ‘eating in front of people’was hypothesizedtobelong to PsychosocialImpact.Itcorrelatedsubstantiallywiththisdimension (r=0.60)butevenslightlyhigherwithOralFunction(r=0.62).For 16items,thehighestcorrelationwasinthepredicteddimension andconsequently,16hypotheseswereaccepted.Fortwoofthese 16correlations,equallyhighcorrelationvaluesoccurredinanother dimension.‘Eatingwithothers’correlatedwithPsychosocialImpact (predicted) but also with Oral Function. ‘Medication for pain’ correlatedwithPsychosocialImpact(predicted)butequallywith OrofacialPain.

Overall, because 16 out of 17 individual hypotheses were accepted,wealsoacceptedouroverarchinghypothesisthattheset of OHRQoL items, having a unique content when compared to OHIP-49,canbeassignedtofourOHRQoLdimensions.

Ten items belonging to Oral Function, Orofacial Pain, and OrofacialAppearancehadtheirsecondhighestcorrelationalways withPsychosocialImpact.Incontrast,whenthehighestcorrelation forsixitemswasobservedtobePsychosocialImpact,thesecond highestcorrelationwasevenlydistributedamongtheotherthree dimensions.

4.Discussion

ItemsfromtwowidelyusedOHRQoLquestionnaires,theGOHAI and the OIDP,as wellas culture-specific OHIP items could be integratedintoa4-dimensionalOHRQoLmodelthatisbasedon themostwidelyusedOHRQoLquestionnaire,theOHIP.Findings indicatethefourOHIPdimensionsofOralFunction,OrofacialPain, OrofacialAppearance,andPsychosocialImpactseemtobe general-izabletootherOHRQoLquestionnaires.

4.1.Indirecteffectsoforalconditions

Furthermore, a pattern emerged, that for indicators of Oral Function, Orofacial Pain, and Orofacial Appearance, the second highestcorrelationobservedwasalwaysforPsychosocialImpact. Thisfindingmightbeinterpretedasevidenceforindirecteffects from oral conditions that we previously hypothesized [28]. Whereasmost oralconditions have a primary effecton oneof

Table1

Sociodemographiccharacteristics,denturestatus,proportionoffollow-upassessments,andOHRQoLimpairmentinprosthodonticspatientsofsixcountries. All N=267 Croatia N=59 Germany N=37 Hungary N=49 Japan N=50 Slovenia N=50 Sweden N=22 Females[%] 58.4 59.3 54.1 63.3 56.0 58.0 59.1 Age(SD)[years] 54.0(17.2) 62.3(13.2) 54.5(13.8) 46.8(15.4) 68.6(8.7) 40.0(16.1) 46.0(15.4)

Removabledenturestatus[%(N)]: Noremovable 56.6(151) 47.5(28) 48.6(18) 67.3(33) 18.0(9) 96.0(48) 68.2(15) Oneremovable 35.2(94) 37.3(22) 40.5(15) 26.5(13) 80.0(40) 4.0(2) 9.1(2) Twocomplete 6.0(16) 15.3(9) 10.8(4) 4.1(2) 2.0(1) 0.0(0) 0.0(0) Unknown 2.2(6) 0.0(0) 0.0(0) 2.0(1) 0.0(0) 0.0(0) 22.7(5)

Post-treatmentassessments[%] 51.3 47.5 64.9 51.0 68.0 32.0 45.4

thosethreedimensionsOralFunction,OrofacialPain,andOrofacial Appearance(e.g.,TMDonOrofacialPain,toothlossonOralFunction and/orOrofacialAppearance),conditionsareoftenassociatedwith asubstantialpsychosocialinfluence,i.e.,theyhavesimultaneous or, morelikely, subsequent PsychosocialImpact. Several authors have already proposed a stepwise (dimension to dimension) hierarchicalframeworkfororalhealthimpacts.Lockerproposed thatoraldiseasescanleadtoimpairment(astructuralcondition), further leading to Functional Limitations, Pain/Discomfort and Handicap.Inhismodel,Disabilityisanintermediatestageinthe transitionfromlesstomoreimpairedperceivedoralhealth,with Handicapbeingthefinalstage[29].Cross-sectionaldatafromthe generalpopulationwerelaterinterpretedassupportforthismodel withFunctionalLimitation,aswellasPainandDiscomfortasafirst, Disabilityasasecond,andHandicapasathirdleveloforalhealth impact [30]. While our findings share similarities with the previousstudy,theyaremoreinlinewithaframeworkpresented for the OIDP [31]. These authors examined hypothesized Pain, Discomfort,FunctionalLimitation,andDissatisfactionwith Appear-anceas“Level2IntermediateImpacts”(Level1isImpairment)and Physical, Psychological and Social Performances as “Level 3 Intermediate Impacts”. We believe that this sequence – that primary functional, painful, and aesthetical problems have subsequentpsychosocialimpacts–representsatypicalsituation when theprimary influence is strong.While this is a frequent

scenario,anisolatedprimaryinfluenceonPsychosocialImpactcan alsooccur,asdemonstratedbefore[28].

4.2.OHRQoLdimensions

That OHIP-based dimensions are applicableto other instru-ments is not surprising. All OHRQoL instruments measure the sameconstruct.Consequently,instrumentscoresshouldcorrelate highly. Infact, pairwisecorrelations betweenOHIP,GOHAI, and OIDP’stotalscoresprovidedempiricalevidenceforanoverlapin content.Forexample,GOHAIandOHIP-14correlatedhighly,witha valueof 0.83 inSwedish dental patients[4], 0.76 inelderly Germans [1],and 0.73 in elderlyJapanese [2] (direction of the correlation differed depending on coding of scores). OIDP and OHIP-14alsocorrelatedhighly,withavalueof0.78inBritishdental hospitalpatients[7]andwith0.76inBrazilianpostpartumwomen

[3].WhenthethreeOHRQoLinstrumentsandafourth question-naire (Oral Health Quality of Life-UK, OHQoL-UKã [32]) were studiedtodeterminetheassociationbetweentoothloss oral health’s major physical health indicator and OHRQoL, an associationwaspresentandseemedtobeindependentfromthe specificOHRQoLinstrumentused[33].However,evidencethatthe measurescorrelatedwellwiththeirdimensionsisnotavailable. Methodological reasons may be responsible for this situation. WhilefactoranalyseshavebeenperformedforOHIP-14,GOHAI,

Table2

Hypothesistestingfor22OHRQoLitems(havinguniquecontentcomparedtoOHIP-49)belongingtofourOHRQoLdimensions(basedonOHIP-49)in267prosthodontics patientsfromsixcountries.

and OIDP,they haveconsistently missed GOHAI’s hypothesized dimensionsofPhysical Function,PsychosocialFunction,andPain/ Discomfort[34],aswellasOIDP’shypothesizeddimensionsPain, Discomfort,FunctionalLimitation, DissatisfactionwithAppearance, andPhysical,PsychologicalandSocialPerformances[31].However, authorsperformedfactoranalysesusingOHIPinconjunctionwith GOHAIorwithOIDP[5,6].Thisindicatedthatauthorsthoughtthat OHRQoL items/instruments are compatible enough to be com-bined. Resulting findings for OHIP-OIDP identified Functional Limitations,Pain-Discomfort,and Psychosocial Impacts as dimen-sions identical to the four-dimensional OHRQOL (based on OHIP),exceptforOrofacialAppearance.WebelievethatOrofacial Appearanceissoimportantforindividuals onparwithOrofacial PainandOral Function that thiscomponent ofperceivedoral healthdeservesaplaceinthetheoreticalstructureofOHRQoL.For theGOHAI-OHIPanalysis,acomparisonisnotsostraightforward becausesixfactorswerefoundbutnotnamedinthestudy[6].

Our findingscorroboratepreviousresultswell.Aspreviously mentioned,itemsdidnotonlycorrelatehighlywiththepredicted dimension but also notably with other dimensions. This was observedbefore[8]andisanindicationthatastronggeneralfactor underliesOHRQoL,andthatthis constructcanprobablybestbe described by a hierarchical model with four first-order factors (dimensions)andahigher-order,generalfactormodel[8].Findings arealso in linewithprevious findings [9] that notall existing OHRQoLindicators(items)arewellsuitedtobeincorporatedinto OHRQoLdimensions.Methodologicalreasonsarepartly responsi-ble.Forexample,theOHIPitem‘Havedentalproblemsmadeyou miserable’missedpreviouslywith0.42ourthresholdforasalient correlation (loading) and was not included in the Psychosocial Impactdimensionevenifthisitemseemstobeaveryplausible indicator. The same situation may apply to the item ‘avoid laughing’ in the current study. Based on the OHIP item ‘avoid smiling’beinga strongindicatorforOrofacialAppearance,‘avoid laughing’ hada substantialcorrelationwiththis dimensionbut alsowithPsychosocialImpact.Moreimportantthan methodologi-calinfluencesisprobablythefactthatitemssuchas‘jointnoises’ and‘drymouth’aresymptomsoforalconditions.Thepresenceof symptomsin OHIP was previously observedfor theOHIP item ‘stalebreath,’ which could not be assigned to one of thefour dimensions[9].Conceptually,symptomsoforaldiseasesshould notbeincorporatedinadimensionalOHRQoLmodel.Finally,we alsoencounteredproblemswithsomeitems.Theitem‘swallow comfortably’didnotshowaninterpretablepatternofcorrelations, whereasanitemwithsimilarcontent(‘difficultyswallowing’)did. ThisindicatesthatOHRQoLitemswillstillneedmore methodo-logicalworkregardinghowtheyperformacrosscultures.

Overall,whilethefindingsofastronggeneralOHRQoLfactor andindirecteffectsfromonedimensiontoanotherarerelevant fromaconceptualpointofviewtoelucidatemechanismsoforal conditions_’in_fluencesonindividuals,theyalsopointtotechnical challenges.Itemshavingaconsiderablecorrelationwithmorethan onedimensionwillmakedimensionsmoresimilar.Thissituation challenges the usefulness of dimension scores. Further item analysesare necessary, leadingtothedeletion of certainitems and the addition of more specific items, to refine OHRQoL dimensions.Alignmentoforalhealthwithhealth-relatedquality oflifemeasurementisalsorecommended[35].

4.3.Strengthsandlimitations

For studies using questionnaires, it should be verified that scoresarevalidand reliableinthetargetpopulation.Whilethe OHIP-49hasdocumentedsufficientpsychometricpropertiesinall sixstudycountries[11,12,15–18],specificGOHAI[1,4,20,21] and OIDP[22–24]languageversionsandtheextraOHIPquestionswere

onlyavailableforsomesettings.Wedidnotperformarigorous cross-cultural adaptation, which is a limitation. However, we believethat theintroducedmeasurement errorwouldprobably have biased findings against the observed pattern. It seems unlikely thattranslation errorandotherinfluenceswould have movedresultsclosertoaplausible,clinicallyinterpretablepattern offindings.

We performed a correlation analysis to study the research questionandwecombineddatawithouttakingintoaccountthat theycamefromdifferentcountries,i.e.,ourdatahadahierarchical structure.Moresophisticatedanalysesexist,e.g.,factoranalysis, but these statistical techniques require more subjects than available. On the other hand, researchers have used other techniques besides factor analysis before[25,30] toinvestigate dimensionality. Our robust correlation analysis, not relying on manystatisticalassumptions,providedasoundandinterpretable pattern offindings. Finally,westudiedGOHAIand OIDP.While these two instrumentsare the two mostwidely used OHRQoL instrumentsbesidestheOHIP,manyotherinstrumentsexist[36]. Alltheinstrumentssharemanysimilaritiesandtheitemsofthe twowestudiedshouldbeseenasasampleofOHRQoLindicatorsin general.

5.Conclusion

Thestudyadvancesourtheoryoforalhealth.WhileOHIP-based dimensionscouldbegeneralizedtoOHRQoLdimensionsbasedon conceptualreasonsbefore,wenowhavemoreevidencethatOral Function, Orofacial Pain, Orofacial Appearance, and Psychosocial Impact can serve as a simple and clinically meaningful set of OHRQoLdimensions becauseseveralinstruments’itemsfit into thesedimensions.

The study also points to future practical improvement of OHRQoLmeasurement.Whileshortinstrumentsthatarebasedon thefour-dimensionalOHRQoLmodelsuchasthe5-itemOHIP[37]

alreadyexistandwerevalidated[38],OHRQoLmeasurementcan beadvanced.ThemanyexistingOHRQoLinstrumentscanintegrate theiritemsintothedimensions.Onlyfourinstruments–onefor each OHRQoL dimension – are necessary to comprehensively measure the major patient-perceived influences from oral disordersanddentalinterventions.

Conflictsofinterest None.

Acknowledgements

Research reportedin this publicationwas supportedby the National Institute of Dental and Craniofacial Research of the NationalInstitutesofHealthunderAwardNumberR01DE022331.

AppendixA.Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound, in the online version, at http://dx.doi.org/10.1016/j.jdent.2016. 06.006.

References

[1]A.J.Hassel,C.Rolko,U.Koke,J.Leisen,P.Rammelsberg,AGermanversionofthe GOHAI,CommunityDent.OralEpidemiol.36(2008)34–42.

[2]K. Ikebe, T. Hazeyama, K. Enoki, S. Murai, T. Okada, R. Kagawa, et al., ComparisonofGOHAIandOHIP-14measuresinrelationtoobjectivevaluesof oralfunctioninelderlyJapanese,CommunityDent.OralEpidemiol.40(2012) 406–414.

[3]B.H.Oliveira,P.Nadanovsky,PsychometricpropertiesoftheBrazilianversion oftheoralhealthimpactprofile-shortform,CommunityDent.OralEpidemiol. 33(2005)307–314.

[4]C.Hagglin,U.Berggren,J.Lundgren,ASwedishversionoftheGOHAIindex: psychometricpropertiesandvalidation,SwedishDentalJournal29(2005) 113–124.

[5]J.Montero,M.Bravo,M.P.Vicente,M.P.Galindo,J.F.Lopez,A.Albaladejo, Dimensionalstructureoftheoralhealth-relatedqualityoflifeinhealthy Spanishworkers,HealthQual.LifeOutcomes8(2010)24.

[6]A.J.Hassel,B.Steuker,C.Rolko,L.Keller,P.Rammelsberg,I.Nitschke,Oral health-relatedqualityoflifeofelderlyGermans—comparisonofGOHAIand OHIP-14,CommunityDent.Health27(2010)242–247.

[7]P.G.Robinson,B.Gibson,F.A.Khan,W.Birnbaum,Validityoftwooral health-relatedqualityoflifemeasures,CommunityDent.OralEpidemiol.31(2003) 90–99.

[8]M.T.John,L.Feuerstahler,N.Waller,K.Baba,P.Larsson, A.Celebic,etal., Confirmatoryfactoranalysisoftheoralhealthimpactprofile,J.OralRehabil.41 (2014)644–652.

[9]M.T.John,D.R.Reissmann,L.Feuerstahler,N.Waller,K.Baba,P.Larsson,etal., Exploratoryfactoranalysisoftheoralhealthimpactprofile,J.OralRehabil.41 (2014)635–643.

[10]H.Murray,D.Locker,D.Mock,H.C.Tenenbaum,Painandthequalityoflifein patientsreferredtoacraniofacialpainunit,J.Orofac.Pain10(1996)316–323. [11]M.T.John,D.L.Patrick,G.D.Slade,Thegermanversionoftheoralhealthimpact profile—translationandpsychometricproperties,Eur.J.OralSci.110(2002) 425–433.

[12] M.Yamazaki,M.Inukai,K.Baba,M.T.John,Japaneseversionoftheoralhealth impactprofile(OHIP-J),J.OralRehabil.34(2007)159–168.

[13]M.T.John,D.R.Reissmann,L.Feuerstahler,N.Waller,K.Baba,P.Larsson,etal., Factoranalysesoftheoralhealthimpactprofile—overviewandstudied population,J.Prosthodont.Res.58(2014)26–34.

[14]G.D.Slade,A.J.Spencer,Developmentandevaluationoftheoralhealthimpact profile,CommunityDent.Health11(1994)3–11.

[15]A.Szentpetery,G.Szabo,G.Marada,I.Szanto,M.T.John,Thehungarianversion oftheoralhealthimpactprofile,Eur.J.OralSci.114(2006)197–203. [16]N.Petricevic,A.Celebic,M.Papic,K.Rener-Sitar,Thecroatianversionofthe

oralhealthimpactprofilequestionnaire,Coll.Antropol.33(2009)841–847. [17]K.Rener-Sitar,A.Celebic,N.Petricevic,M.Papic,D.Sapundzhiev,A.Kansky,

etal.,Theslovenianversionoftheoralhealthimpactprofilequestionnaire (OHIP-SVN):translationandpsychometricproperties,Coll.Antropol.33 (2009)1177–1183.

[18]P.Larsson,T.List,I.Lundstrom,A.Marcusson,R.Ohrbach,Reliabilityand validityofaSwedishversionoftheoralhealthimpactprofile(OHIP-S),Acta Odontol.Scand.62(2004)147–152.

[19]N.G.Waller,M.T.John,L.Feuerstahler,K.Baba,P.Larsson,S.Peršic,etal.,A 7-dayrecallperiodforaclinicalapplicationoftheoralhealthimpactprofile questionnaire,Clin.OralInvestig.20(2016)91–99.

[20]M.Naito,Y.Suzukamo,T.Nakayama,N.Hamajima,S.Fukuhara,Linguistic adaptationandvalidationofthegeneraloralhealthassessmentindex(GOHAI) inanelderlyJapanesepopulation,J.PublicHealthDent.66(2006)273–275.

[21]A.Lomšek,B.Artnik,R.Gašperši9c,Theinfluenceofsocio-demographicfactors onself-perceptionoforalhealthinperiodontalpatients,J.Clin.Periodontol.39 (Suppl.13)(2012)157–158.

[22]M.Naito,Y.Suzukamo,H.O.Ito,T.Nakayama,DevelopmentofaJapanese versionoftheoralimpactsondailyperformance(OIDP)scale:apilotstudy,J. OralSci.49(2007)259–264.

[23]A.L. Ostberg, P. Andersson, M. Hakeberg, Cross-cultural adaptation and validationoftheoralimpactsondailyperformances(OIDP)inSwedish,Swed. Dent.J.32(2008)187–195.

[24]V.Lajnert,R.Gržic,N.Radica,D.Šnjaric,S.Špalj,Translationandvalidationof theCroatianversionoftheoralimpactsondailyperformances(OIDP)scale, Vojnosanit.Pregl.73(2015)136.

[25]M.T.John,Exploringdimensionsoforalhealth-relatedqualityoflifeusing experts’opinions,Qual.LifeRes.16(2007)697–704.

[26]M.T.John,D.R.Reissmann,A.Szentpétery,J.G.Steele,Anapproachtodefine clinical significanceinprosthodonticpatients,J.Prosthodont.18(2009) 455–460. [27]D.R.Reissmann,I.Sierwald,G.Heydecke,M.T.John,Interpretingoneoral

healthimpactprofilepoint,HealthQual.LifeOutcomes11(2013)12. [28]M.T.John,K.Rener-Sitar,B.Baba,A. Celebic,P.Larsson,G.Szabo,etal.,Patterns

ofimpairedoralhealth-relatedqualityoflifedimensions,J.OralRehabil.43 (2016)519–527.

[29]D.Locker,Measuringoralhealth:aconceptualframework,CommunityDent. Health5(1988)3–18.

[30]N.Nuttall,G.D.Slade,A.E.Sanders,J.G.Steele,P.Allen,S.Lahti,Anempirically derivedpopulation-responsemodeloftheshortformoftheoralhealthimpact profile,CommunityDent.OralEpidemiol.34(2006)18–24.

[31]S.Adulyanon,A.Sheiham,Oralimpactsondailyperformances,in:G.D.Slade (Ed.),MeasuringOralHealthandQualityofLife,UniversityofNorthCarolina, DepartmentofDentalEcology,ChapelHill,1997,pp.151–160.

[32]C.McGrath,R.Bedi,Anevaluationofanewmeasureoforalhealthrelated qualityoflife–OHQoL-UK(W),CommunityDent.Health18(2001)138–143. [33]A.E.Gerritsen,P.F.Allen,D.J.Witter,E.M.Bronkhorst,N.H.Creugers,Toothloss

andoralhealth-relatedqualityoflife:asystematicreviewandmeta-analysis, HealthQual.LifeOutcomes8(2010)126.

[34]K.A.Atchison,Thegeneraloralhealthassessmentindex,in:G.D.Slade(Ed.), MeasuringOralHealthandQualityofLife,UniversityofNorthCarolina, DepartmentofDentalEcology,ChapelHill,1997,pp.71–80.

[35]D.R.Reissmann,Alignmentoforalhealth-relatedwithhealth-relatedquality oflifeassessment,J.Prosthodont.Res.60(2016)69–71.

[36]E.Skaret,A.NordrehaugÅstrøm,O.Haugejorden,OralHealth-relatedquality oflife(OHRQoL)—reviewofexistinginstrumentsandsuggestionsforusein oralhealthoutcomeresearchinEurope,in:D.M.Bourggeois,J.C.Llodra(Eds.), HealthSurveillanceinEurope–EuropeanGlobalOralHealthIndicators DevelopmentProject–2003ReportProceedings,QuintessenceInternational, Paris,2004,pp.99–110.

[37]M.T.John,D.L.Miglioretti,L.LeResche,T.D.Koepsell,P.Hujoel,W.Micheelis, Germanshortformsoftheoralhealthimpactprofile,CommunityDent.Oral Epidemiol.34(2006)277–288.

[38]A.Naik,M.T.John,N.Kohli,K.Self,P.Flynn,Validationoftheenglish-language versionof5-itemoralhealthimpactprofile,J.Prosthodont.Res.60(2016)85–91.