Early career burnout among Dutch nurses: A process captured in a Rasch model

(1)

ContentslistsavailableatScienceDirect

Burnout

Research

j ou rn a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / b u r n

Early

career

burnout

among

Dutch

nurses:

A

process

captured

in

a

Rasch

model

Jan

Alexander

de

Vos

a,b,∗

,

André

Brouwers

c

_,

_Tineke

_Schoot

d

_,

_Ron

_Pat-El

c

_,

_Peter

_Verboon

c

_,

Gérard

Näring

e

a_Human_Concern_Foundation,_Baden_Powellweg₃₀₅_M,_Amsterdam,_The_Netherlands

b_Twente_University,_Department_of_Psychology,_Health_and_Technology,_Enschede,_The_Netherlands c_Faculty_of_Psychology_and_Educational_Sciences,_Open_University,_Heerlen,_The_Netherlands d_Faculty_of_Health,_Zuyd_University_of_Applied_Sciences,_Maastricht,_The_Netherlands e_Behavioural_Science_Institute,_Radboud_Universiteit_Nijmegen,_The_Netherlands

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received28October2015

Receivedinrevisedform25April2016 Accepted7June2016 Keywords: Burnout Nurses Earlycareer Dimensionality Itemresponse Raschmeasurement

a

b

s

t

r

a

c

t

ThisstudyinvestigatesburnoutamongDutchnursinggraduatesasaprocessbytestinga sequential-developmentalmodel.Asampleof237respondentswasrecruitedfromapopulationofDutchearlycareer nurses.BurnoutwasmeasuredwiththeDutchversionoftheMaslachBurnoutInventoryHumanServices Survey(MBI-HSS,Maslach&Jackson,1981;Schaufeli&vanDierendonck,2000).First,thedimensionality wastestedwithconﬁrmatoryfactoranalysis.Aresultingone-dimensionalmodelwithexhaustionand depersonalisationwasthenanalysedusingaRaschratingscalemeasurementmodel.Raschanalysis showedthatthesedataadheredwelltoasequential-developmentalmodel.Burnoutamongearlycareer nursesmaybeoperationalizedasaprocessthatstartswithfatigueasaresultofstrainandendswith severeexhaustionanddepersonalisationtowardspatients.Personalaccomplishmentdevelopsrelatively independently.AseparateRaschanalysisonthepersonalaccomplishmentitemsrevealedascalewith almostsimilaritemlocations,resultinginredundantinformation.

1. Introduction

For nursing graduates the experience of transition from school-basedexperiencetoprofessionalpracticecanbestressful (Duchscher&Myrick,2008).Scientistshavenotedalreadydecades ago that younger nurses show a greater vulnerability towards burnoutthanoldernurses(Bartz&Maloney,1986),andthiswas alsonotedinmorerecentstudies(Lee&Wang,2002;Winwood, Wineﬁeld,&Lushington,2006).Onlyinthelastdecadethisissue istheexplicitfocusofstudiesamongnewgraduatenurses.Such studiesshowseriouslevelsofmentalhealth problemsand high turnover(Laschinger,Finegan&Wilk,2009).AstudybyBeecroft andcolleagues(2008)showedthat30%ofnewgraduateshadhigh turnoverintentions.BowlesandCandela(2005)foundanactual turnoverrateof30%intheﬁrstyearand57%aftertwoyears.In anotherstudy,66% ofnew graduateswere experiencingsevere burnout(Cho,LaschingerandWong,2006).

∗ Correspondingauthorat:HumanConcernFoundation.BadenPowellweg305 M,Amsterdam,TheNetherlands.

E-mailaddress:s.devos@humanconcern.nl(J.A.deVos).

Newprofessionalsmayinitiallyfeelinadequatelypreparedfor theiroccupationalrole,whichledDuchscher(2009)todescribethis transitionasthemoststressful,frustrating,discouragingand dis-illusioningstageintheprocessofprofessionalroleadaptationfor nursinggraduates.Cherniss(1980)hasexplainedthisprocessasan identitycrisisamongnoviceprofessionals.Thismayleadto unsuc-cessfuloccupationalsocializationorearlycareerburnout(Cherniss, 1980;Gustavsson,Hallsten,&Rudman,2010).

Burnoutisgenerallyexplainedasapsychologicalsyndrome con-sisting of emotional exhaustion, depersonalisation and reduced personalaccomplishmentthatcanoccuramongindividualswho workwithpeople(Maslach&Jackson,1981;Schaufeli,Maslach, & Marek, 1993). Emotional exhaustion refers to feelings of being overextended emotionally and physically, and depleted of resources. Depersonalisation refers to a cynical and distant attitude towards one’s work and the people one works with. Reduced personal accomplishment is the tendency to evaluate one’sachievementsatworknegatively(Maslach&Leiter,1997). Emotionalexhaustionisusuallyconsideredtobethecentral qual-ityofburnoutandthemostobviousmanifestationofthesyndrome (Taris,LeBlanc,Schaufeli,&Schreurs,2005).

http://dx.doi.org/10.1016/j.burn.2016.06.001

(2)

Scientistshaveproposedfourmajorprocessmodelstodescribe therelationshipsbetweentheburnoutdimensions.Cherniss(1980) described a process of burnout consisting of three stages. The ﬁrststageinvolvesanimbalancebetweenresourcesanddemand, i.e.stress.The secondstageis theimmediate, short-term emo-tional response to this imbalance, characterized by feelings of anxiety,tension,fatigueandexhaustion.Thethirdstageconsists ofanumberofchangesinattitudeandbehaviourleadingto deper-sonalisation.

LeiterandMaslach (1988)described a processmodel where burnout starts with emotional exhaustion, which then triggers depersonalisation,which leadstoreduced personal accomplish-ment.Feelingsofbeingoverextendedariseﬁrstasaconsequence of a demanding work environment. To cope with exhaustion and stress, the overextended individual then distances himself fromothersandbecomescynicalinhisresponsestowardothers. Oncedepersonalisationoccurs,theindividualmaybegintofeel ineffectiveonthejob,andtheself-evaluationofhisorher accom-plishmentsislikelytobecomenegative.

Whereas in the previous models elevated levels of emo-tional exhaustion signify the onset of burnout, Golembiewski, MunzenriderandStevenson(1986)consideremotionalexhaustion tobetheﬁnalstageofburnout.Golembiewskietal.describea pro-cessofburnoutthatstartswithdepersonalisation,whichtriggers reducedpersonalaccomplishmentandendsinemotional exhaus-tion.Theystatethat depersonalisationariseswhenprofessional detachmentin dealing withothers is nolonger functional and therebyreducestheabilitytoformnecessaryrelationshipswith othersandtoperformwellonthejob.Asaconsequence,the self-evaluationofjobaccomplishmentbecomesnegative,theabilityto copewithjobdemandsdeclines,whichinturninducesfeelings ofbeingoverextended.Lee andAshforth(1993)describea pro-cessmodelofburnoutinwhichemotionalexhaustionispositively relatedtodepersonalisation,andwherepersonalaccomplishment developsindependentlyfromdepersonalisation. Theystatethat elevatedlevelsofemotionalexhaustionhaveadirecteffecton feel-ingsofpersonal accomplishmentratherthan indirectlythrough depersonalisation(Lee&Ashforth,1993;Tarisetal.,2005).

Thesefourmajorprocessmodelshaveincommonthatthey consideredburnouttobeaone-dimensionalsyndromeconsisting ofseveralstages.However,theunidimensionalityofburnouthas beendebated,aswellastheexactcausalrelationshipsbetweenthe burnoutdimensions(Schaufeli&Buunk,2003).

Tarisetal.(2005)investigatedwhetherthereweresequential relationshipsbetweentheburnoutdimensionsbyreviewing ear-lierresearchandanalysingtwolongitudinalsets(ateachersample andasampleofoncologycareproviders).TheChernissmodelwas notincludedinthereviewbecausethestudyonlyconsidered mod-elswiththreedimensionsofburnout(Tarisetal.,2005).Thereview providednoconvincingsupportforanyparticularprocessmodel. ThelongitudinalstudytestedthemodelsofLeiterand Maslach (1988),Golembiewskietal.(1986),LeeandAshforth(1993)and anintegrationofthelattertwo.Thedatasupportedthemodelsof Lee&Ashforth(1993)andofLeiter&Maslach(1988),whichboth proposethatburnoutstartswithemotionalexhaustionandends withreducedpersonalaccomplishment.Noevidencewasfound forthemodelofGolembiewskiandassociates(1986)thattakes depersonalisationasastartingpoint.

In the integrationof the model of Lee and Ashforth (1993) andLeiterandMaslach(1988)thatTarisetal.construed,burnout startswithemotionalexhaustionthattriggersdepersonalisation. Thismodelassumesthat persistentdepersonalisation maylead toreducedpersonalaccomplishment,butitalsocontainsa sep-aratedirecteffectofemotionalexhaustion onreduced personal accomplishment. Furthermore they tested a feedback effect of depersonalisationovertimeontheotherburnoutdimensions.This

integratedmodelofTarisetal.(2005)showedthebestfitofthe dataincomparisontotheothermodels.Higherlevelsof exhaus-tionindeedtriggeredhigherlevelsofdepersonalisation,andhigher levelsofdepersonalisationwereassociated withlowerlevelsof personalaccomplishment.Furthermore,afeedbackeffectof deper-sonalisationonexhaustionover time wasreported(Tarisetal., 2005).Thelasttwoeffectswereonlyfoundintheteachersample, butnotamongoncologycareproviders.ParkerandSalmela-Aro (2011)alsocomparedandcontrastedseveralmajormodelsforthe developmentofschoolburnoutinafour-wavelongitudinal sam-pleofhighschoolstudents.Theirresultssuggestthatthemodelof Tarisetal.(2005)providedasignificantlybetterfittothedatathan othermodels.

Gustavssonetal.(2010)appliedaninnovativeapproachtothe Cherniss(1980)modelinastudyofearlycareerburnoutamong nurses. Personal accomplishment is not a part of the Cherniss model.Theburnoutquestionnairethattheyusedintheirstudy,the OldenburgBurnoutInventory(OLBI;Demerouti,&Bakker,2008) alsoonlymeasuresexhaustionanddisengagement.Theyassumed thatearlycareerburnoutis asequentialdevelopmentalprocess thatcanbedescribedinanitemresponsetheory.Theirfindings indicateaprocessconsistingofthreelinkedstageswherefirst cri-sisandstraintriggerexhaustion,thenhigherlevelsofexhaustion areassociatedwithhigherlevelsofdefensivecoping,whichfinally leadstoevenhigherlevelsofexhaustionanddisengagement.

Gustavssonetal.appliedaprobabilisticitemresponseapproach. Theone-parameteritemresponsemeasurement(Raschmodel)has gainedwidespreadrecognitionforitsmeasurementpropertiesand usageinconstructingandevaluatingmeasuresof developmental-sequentialorhierarchicalattributes(Gustavssonetal.,2010;Smith &Smith,2004).Thismethodassumesthatitemscoresare deter-minedbythelocationoftherespondentandtheitem,andprovides informationaboutitemsthatdonotﬁtthemodel.Thismethod assumesanorderofbothparticipantsandsymptomsalonga sin-gle latentcontinuumreﬂecting an individualgrowth sequence. Thisaspectofitem responsetheorymakesitpossibletomodel ahypothesizeddevelopmentalprocess.Ascorederivedfromthis modelthenrepresentstheseverityoftheburnout.If,forexample, apersonhasdevelopeddepersonalisationafteraperiodofsevere emotionalexhaustion,themodelwillassumeandtestthat emo-tionalexhaustionwillnothavedisappeared,butstillispresent. ApplyingthistotheTarismodelwouldresultinthepredictionthat areducedsenseofpersonalaccomplishmentcanonlydevelopafter depersonalisationhasbeenreported.

InmostcountriestheOLBIisnotusedtoassessburnout,but rathertheMaslachBurnoutInventory(MBI).Theaimofthe cur-rentstudyistoinvestigatewhethertheitemsoftheMBIcanalso beorderedasasequential-developmentalprocessbyusinganitem responseapproach. InlinewiththeCherniss(1980)model and inlinewiththeMaslach(1988)modelwehypothesizethatthe emotionalexhaustionanddepersonalisationaspectsofthe MBI-HSScanbemodeledasasequential-developmentalprocesswhere bothitemsandindividualscanbelocatedinspecificallydefined uniquephasesasanindicatorofaburnoutprocess.Drawingonthe complexfindingsregardingthedevelopmentofpersonal accom-plishmentfromtheTarisetal.(2005)model,wewilltestwhether personalaccomplishmentalsofitsinaone-dimensionalmodelor hastoberegardedasaratherindependentprocess.

Maslach and Leiter(1997)proposedthat engagement isthe oppositeofburnout.SchaufeliandBakker(2003)statethatburnout andengagementaretwodistinctdimensions,althoughnegatively correlated.SchaufeliandBakker(2003)defineworkengagement asapositive,fulfilling,work-relatedstateofmindthatis charac-terizedbyvigor,dedicationandabsorption.Inordertovalidateour findings,wewillcalculatecorrelationsoftheRaschburnoutscores

(3)

ofparticipantswithscoresonthethreeengagementdimensions. Wehypothesizethattheywillallbenegative.

2. Method 2.1. Participants

In 2009, an estimated 149.000 nurses were active in the Netherlands (Van der Windt, Smeets, & Arnold, 2009). Divided intobranchestherewere71.500nursesactiveinhospitals,24.900 nursesinthementalhealthcaresector,15.600nursesinthe dis-abledcaresector,22.400nursesintheelderlycaresector,8.700 nursesinthehomecaresectorand4.900nursesinothersettings. TheintakeofnewnursingstudentsintheNetherlandsin2008was 3.583nursesforProfessionalEducationLevel(Bachelor,HBO),and 6.542forIntermediateLevel(MBO,VanderWindtetal.,2009a).In 2009–2010,2.600newbachelorstudentsgraduated,whileabout 4.000newnursesonintermediatelevelgraduatedin2007–2008 (VanderWindt,vanderVelde,&vanderKwartel,2009).This num-berisfairlystableinthelastyears,butashortageofnursesonthe labourmarketispredicted(VanEssen,Kramer,vanderVelde&van derWindt,2016).Anincreasedneedforcareforelderlypeopleand forpeoplewithchronicillnessbecauseofalongerlifeexpectancy (VanEssenetal.,2016)andagingofthenursingpopulationmight causethiseffect(Hellenthal,2012).

Forpracticalreasonsaconveniencesamplewasused.The tar-getsamplepopulationwasreachedwiththeassistanceofhigher professionalschoolsofnursingstudies.

Participantswerenursingstudentsandgraduates fromZuyd UniversityofAppliedSciencessituatedinLimburg,andtheHAN UniversityofAppliedSciences,situatedinGelderland.Limburgand Gelderlandare bothprovincesin theNetherlands. Toachieve a higherresponserate,acalltofilloutthequestionnairewasmade ina magazine fornurses(ZorgannoNU),and throughTwitter. Thefollowinginclusioncriteriawereused.First,participantshad towork(ataninternship)atleastsixmonthspriortothisstudy. Theyhadtoworkinanursingsettingthroughouttheperiodof thisstudy.Second,participantshadtohaveamaximumworking experienceofthreeyearssincegraduation.Althoughacriterionof threeyearsforearlycareerburnoutisratherarbitrary,weused thecriterioninaccordancewithGustavssonandassociates(2010). FromthedatabaseofnursingstudentsandgraduatesfromZuyd UniversityofAppliedSciences,323nurses(260studentsand63 graduates)metthecriteriaandwereaskedtofillouta question-naire. Theresponse was39%(127 participants).Due toprivacy requirements,theHANUniversityofAppliedSciencescontacted studentsandalumnithemselvesfromtheirowndatabase.Asthey didnotprovideinformationaboutthepopulationsize,itwasnot possibletocalculatetheresponserate.Thedisseminationofthe questionnaireatHANUniversityofAppliedSciencesresultedinan additional110participantswhofilledinthequestionnaire.There wasnoadditionalresponsethroughTwitterandtheappealviathe magazine.

Thisresultedinatotalof237participants(127participantsfrom ZuydUniversityand110participantsfromHANUniversity).The averageagewas24years,(SD=7.7)witharangeof17–58years. Inoursample209participantswerefemale(88%)and28 partici-pantsweremale(12%).Theaveragenumberofworkinghoursper weekwas31h(SD=6.9),witharangeof10–50h.74participants (31%)wereworkingandhadgraduatedfromschool,79participants (33%)wereworkinginaninternshipwhilestillattendingschool, 67participants(28%)hadjustﬁnishedaninternshipand17 partic-ipants(7%)hadadifferentjob/schoolcharacteristic.Ofthetotal sample,120respondents(51%) wereworkingina general hos-pital,48respondents(20%)inmentalhealthcare,7respondents

(3%)inthehandicappedcare,24respondents(10%)inanursing home,26(11%)oftherespondentsinthehomecareand12(5%) inanothercaresetting.Theeducationalleveloftherespondents wasHighereducationallevel(Bachelor,HBO,level5,201 respon-dents,85%),educationallevel(MBO,level3&4,21respondents,9%) andacademicorelse(15respondents,6%).Thescoresonthe MBI-HSS wereM=1.39,SD=0.92 foremotional exhaustion,M=0.80, SD=0.61fordepersonalisation,andM=4.11,SD=0.73forpersonal accomplishment.Thepercentagesofaveragetolowscores com-paredtoa Dutchnormgroupofhealthcare workerswere81.1% foremotionalexhaustion,90.8%fordepersonalisationand79.8% forpersonalaccomplishment,whileconsecutively,18.9%,9.2%and 20.2% of thesample scoredhigh tovery high(Schaufeli& van Dierendonck,2000).

Permissiontocarryoutthestudywasgrantedbythe Commis-sionofNursingStudiesfromZuydUniversityofAppliedSciences. TheHANUniversityofAppliedSciencestooknoticeofthisdecision andsubsequentlyalsogaveapproval.Participantswereinformed aboutthepurposeof theresearchandthepossibility to termi-natetheirparticipationatanytime.Thiswasstatedinaninformed consentformwhichparticipantshadtosignbeforeﬁllingoutthe questionnaire.

2.2. Design

Thedesignofthis studyis across-sectionalsurvey.Byusing anitemresponseapproachitispossibletomodelatheorydriven underlyingprocessofburnoutbasedontheseverityofsymptoms, representedinitems,andthelevelofseverityoftheburnoutina person.

Minimal samplesizecriterion is 100respondents.For Rasch analysiswithprincipalcomponentanalysis,100personsisauseful criteriontoobtainreliableﬁndings(Arrindell&vanderEnde,1985). InadditionitisadvisedforRaschmeasurementtohaveatleast 10observationsforeachcategorywithpolytomousdata(Linacre, 2002).TheMBI-HSShas7categoriesperitem,meaningthat,atleast 70respondentsarerequiredforRaschanalysis.

Allrespondentswereaskedtoﬁllinananonymousonline ques-tionnairebye-mail.Thequestionnairewasmadeavailablethrough Thesistools,asoftwareprogramforconductingonlinestudies. 2.3. Instruments

Self-report instruments were used, measuring background characteristics,dimensionsofburnout andengagement.For the measurementofburnoutthereareseveralquestionnaires. Gustavs-sonandassociates(2010)usedtheOldenburgBurnoutInventory, which is a questionnairemeasuringthe burnout dimensionsof exhaustionanddisengagementandtheoppositepolesofburnout, energyandengagement(Demerouti&Bakker,2008;Kristensen, Borritz,Villadsen,&Christensen,2005).Thequestionnairehasgood psychometriccharacteristics.However,itisnotappropriateforthis researchbecausethedimensionpersonalaccomplishmentis miss-ing.Themostwidely-usedquestionnaireistheMaslachBurnout Inventory(Maslach&Jackson,1981).Itissaidthatover90%ofthe journalarticlesanddissertationsusetheMBI-HSSforthe measure-mentofburnout(Schaufeli,Bakker,Hoogduin,Schaap,&Kladler, 2001;Schaufeli&Enzmann,1998).Becauseofthewidelyaccepted useandgoodpsychometriccharacteristics,theDutchversionofthe MaslachBurnoutInventory-HumanServicesSurvey(MBI-HSS)was used(Maslach&Jackson,1981;Schaufeli&Dierendonck,2000).The MBI-HSSconsistsof20items,whichmeasureemotionalexhaustion (8items),depersonalisation(5items)andpersonal accomplish-ment(7items).Theitemsaremeasuredona7-pointLikertscale, rangingfrom1(never)to7(always).Highscoresonthescales emotionalexhaustion and depersonalisation,andlow scoreson

(4)

personalaccomplishmentindicateburnout.Examplesofitemsare: “Ifeelfrustratedbymyjob”(emotionalexhaustion),“Iworrythat this job is hardeningme emotionally” (depersonalisation), and “InmyworkIdealwithemotionalproblemsverycalmly” (per-sonalaccomplishment).Reportedinternalconsistencyisgenerally high, withCronbach’s alphasof 0.89 for emotional exhaustion, 0.69fordepersonalisation,and0.84forpersonalaccomplishment (Maslach&Jackson,1981;Schaufeli&vanDierendonck,2000).In ourstudyCronbach’salphasare 0.88for emotional exhaustion, 0.57fordepersonalisationand0.95forpersonalaccomplishment. Thealphaofdepersonalisationcannotdirectlybecomparedtothe alpha’sfromtheotherscalesasithasfeweritemsthanpersonal accomplishment or emotional exhaustion. Correcting thealpha withtheSpearman-Brownpredictionformularesultedinanalpha of0.65fordepersonalisation.

Workengagement wasmeasuredby theDutchengagement questionnaire(UBES,Schaufeli&Bakker,2003).Thequestionnaire consistsof17itemsmeasuringvigor(6items),dedication(5items) andabsorption(6items)andusesthesame7-pointLikertscale,as theburnoutquestionnaire,rangingfrom1(never)to7(always). Theinternalconsistencyinthisstudywasadequate,with Cron-bach’salphasof0.83 forvigor,0.88 fordedication,and0.71 for absorption.

2.4. Analysis

A Rasch measurement model was used. This measurement modelisamathematicalmodeldescribinghowthedifferent indi-vidual’s particular responses to specific items in a multi-item instrument shouldbe ordered. The Rasch model is partof the itemresponsetheory.In comparisontoclassictesttheory,item responsetheory(moderntesttheory)isdifferentinits assump-tionsofmeasurement.Forinstance,inclassictesttheory,ameasure ofreliabilityalwaystakesintoaccountameasureofdistribution (dispersion),mostlikelythevariance,whichisbasedupona distri-butionorpopulation(Verhelst,1993).Reliabilityisforthisreason alwayspopulationdependent.Second,inclassictesttheoryatrue scoreisalwaystest-specific,andforthisreasonitisnotalways clearifascoreisindicativefortheabstractconceptthetestaims tomeasure.In itemresponsetheory,ascoreisnot test-specific butdependentoftheindividualresponsesandtheitems.Third, withclassictesttheoryitisnotpossibletoempiricallyexamine thetestlevel,i.e. whetherthevariables areof ordinalor inter-vallevel. By using item response theory, thetest level follows fromthetheory.AnimportantassumptioninRaschmodelisthe unidimensionalityofresiduals.Toassesstheunidimensionalityof theburnout scalesConfirmatory factoranalysis(CFA) wasused. Thefollowingmodelsweretested:(1)aone-factormodelwithall itemsfromthethreeMBI-HSSsubscales;(2)atwo-factormodel withpersonalaccomplishmentasafactorandemotional exhaus-tionanddepersonalisationasaone-dimensionalfactor;(3)athree factor modelwitha correlation betweendepersonalisation and emotionalexhaustion, but nocorrelationwithpersonal accom-plishment;(4)and orthogonalthree factormodel basedonthe threetheoreticalconstructs, emotionalexhaustion, depersonali-sation,and personalaccomplishments;and(5)anobliquethree factormodelwiththesamefactorstructureasmodel4.To inter-pretamodel’sfit,thefollowingindicatorswereused:RootMean SquaredErrorofApproximation(RMSEA)below0.05and Compara-tiveFitIndex(CFI)andTucker-LewisIndex(TLI)above0.95indicate goodfit(Browne&Cudeck,1992)andRMSEAbelow0.08andCFI andTLIabove0.90 indicateacceptablefit(Hu&Bentler, 1999). Theresultsofa CFAwereusedasastartingpointfortheRasch Analysis.Afterthat,aPartialCreditModel(PCM)wasestimated inordertoallowfortheidentificationofitemswhere response orderingisnotcumulative.ThedatawasanalysedwithR

statisti-Table1

Conﬁrmatoryfactoranalysis:Fitstatisticsofthetestedburnoutmodels.

model AIC ␹2_(df) _CFI _TLI _RMSEA

1 1-factor 12033.180 1600.17(152) 0.378 0.300 0.214 2 2-factor 10722.70 287.70(151) 0.941 0.933 0.066 3 3-factor,nocorrelationPA 10697.10 262.10(151) 0.952 0.946 0.059 4 3-factororthodogonal 10801.58 368.58(152) 0.907 0.895 0.083 5 3-factorcorrelated 10699.15 260.15(149) 0.952 0.945 0.060 Notes:AIC=Akaike’sinformationcriterion,␹2₌_chi-square,_df₌_degrees_of_freedom,

CFI=Comparativeﬁtindex,TLI=Tucker-LewisIndex,RMSEA=Rootmeansquare errorofapproximation.

calsoftware(RCoreTeam,2013).CFA’swereconductedwiththe

lavaan-package(Rosseel,2012)andRaschmodelsweretestedwith theeRm-package(Mair&Hatzinger,2007).ThescoresontheMBI PersonalAccomplishmentscalewerereversed beforetheywere usedintheCFAandintheRaschanalysissothathigherscoresin theseanalyseswerealwaysassociatedwithmoreseverelevelsof burnout.

3. Results

3.1. Conﬁrmatoryfactoranalysis

Thedimensionalityoftheburnoutconstructmeasuredwiththe (MBI-HSS)wastestedwithconﬁrmatoryfactoranalysis(CFA).For thisanalysisthescoresofthepersonalaccomplishmentscalewere reversedﬁrst.TheresultsaresummarizedinTable1.

ItemEE8(“end ofmyrope”)consistentlycameupasa Hey-woodcase(negativeestimatedvariance)andwasremovedfrom furtheranalysis. The one factor model showed a very poor fit (model1;CFI=0.38,TLI=0.30,RMSEA=0.21),indicatingthatthe MBI-HSSmeasuredamultidimensionalconstruct,whichviolated theRaschassumptionofunidimensionality.Thefitofthe alterna-tivemodelsrangedfromacceptabletogood.Thebestmodelwas thethree-factormodelwithoutcorrelationsforfactorPA(model 3;CFI=0.95,TLI=0.95,RMSEA=0.06).Thedifferencebetweenthe fullycorrelatedthree-factormodelandmodel3wasnot signifi-cant(␹2(2)=1.94,p=0.622).However,theinterfactorcorrelation betweendepersonalisationandemotionalexhaustionwasr=0.77. Duetotheveryhighoverlapbetweenthesetwofactors,model 2(depersonalisationandemotionalexhaustionasone unidimen-sionalfactor,versuspersonalaccomplishment)waschosenasthe preferredmodel,whichhasanacceptablefit(CFI=0.94,TLI=0.93, RMSEA=0.07). The interfactor correlation between PA and the combined scale (EEand DP) wasr=0.08, t(207)=1.11, p=0.28. Based on theCFA, emotional exhaustion and depersonalisation could be considered unidimensional, whereas personal accom-plishmentneededtobeconsideredasadistinctseparatefactor. ThiswasconfirmedbytheAndersonLikelihoodRatio(LR)testfor testingtheunidimensionalityofresiduals.Thecombinedscale(EE andDP)wasunidimensional,Anderson-LR(103)=63.11,p=0.99, andthePAscalewasunidimensional,Anderson-LR(431)=209.53, P=0.99.Forthisreason,Raschanalysiswasperformedontheitems fromthedepersonalisationandemotionalexhaustionscalesand separatelyonthepersonalaccomplishmentscale.

3.2. Veriﬁcationoftheorderofresponsecategories

Despite the good ﬁt of the MBI-HSS in the factor analysis, anevaluationoftheinitialsymptomseverityandresponse-scale thresholdsrevealed that all 11 items demonstrated disordered responsecategories.Themostseveredisorderingoccurredinitems pertainingtodepersonalisationandwasmostlikelycausedbythe limitednumberofresponseshigher>2.Tocounterthis,the orig-inalsevenresponsesinthedisordereditemswerecollapseduntil

(5)

Table2

RecodedMaslachBurnoutInventoryresponsescale.

Originalscale 0 1 2 3 4 5 6 Recodedscale EE1drained 0 1 2 3 3 3 4 EE2usedup 0 1 2 3 3 4 5 EE3fatigued 0 1 2 3 3 3 4 EE4strain 0 1 1 1 1 2 2

EE5burnedout 0 1 2 2 2 2 3

EE6frustrated 0 1 2 2 3 3 3

EE7workingtoohard 0 1 2 3 3 4 4

DP1objects 0 1 2 2 2 3 3

DP2callous 0 1 1 1 1 1 2

DP3hardening 0 1 1 1 2 2 2

DP4don’treallycare 0 1 2 3 3 3 3 Note:EEitemsareemotionalexhaustionitems.DPitemsaredepersonalisation items.

Table3

Symptomseverity,anditemﬁtstatisticsinaRaschmodelfortheMBIExhaustion andDepersonalisationitems.

MSQ Location Chi2₍₂₀₂₎ _p-value _Outﬁt _Inﬁt

EE3fatigued −0.31 173.35 0.93 0.85 0.86 EE7workingtoohard 0.15 226.67 0.11 1.12 1.02 EE1drained 0.18 143.32 1.00 0.71 0.73 EE2usedup 0.87 162.43 0.98 0.80 0.81 EE6frustrated 1.15 134.65 1.00 0.66 0.69 DP1objects 1.15 209.47 0.34 1.03 1.06 EE5burnedout 1.17 144.42 1.00 0.71 0.79 DP4don’treallycare 1.99 299.61 0.00 1.48 1.46 DP3hardening 2.66 146.65 1.00 0.72 0.78 EE4strain 3.81 65.76 1.00 0.32 0.87 DP2callous 4.27 745.41 0.00 3.67 1.02 Note:EEitemsareemotionalexhaustionitems.DPitemsaredepersonalisation items.

Table4

ItemfitstatisticsinaRaschmodelfortheMBIPersonalaccomplishmentscale. MSQ Location Chi2₍₂₀₂₎ _p-value _Outfit _Infit

PA6achievement 1.54 163.53 0.99 0.79 0.81 PA3positiveinﬂuence 1.50 213.31 0.39 1.02 1.04 PA1empathy 1.45 157.83 1.00 0.76 0.71 PA4relaxed 1.43 133.09 1.00 0.64 0.62 PA2problemsolving 1.40 162.27 0.99 0.78 0.76 PA5cheerup 1.40 157.67 1.00 0.76 0.75 PA7emotionalproblems 1.05 285.96 <0.001 1.37 1.32 Note:PA=personalaccomplishment.

allitemshadmonotonouslyincreasingthresholds.Theemotional

exhaustionitemsingeneralneededtheleastcollapsing;the

deper-sonalisationitemneededthemostcollapsing.Theﬁnalrecoding

resultedintothreecategories(0–2)forthreeitems,fourcategories

(0–3)forfouritems,ﬁvecategories(0–4)forthreeitems,andsix

categories(0–5)foroneitem;allinorderofsymptomseverity.The

resultingrecodedMBI-HSSscoringisshowninTable2.

3.3. Itemseverityhierarchy

InTable3theitemlocationsaresorted,fromtoptobottom,in ascendingorderofsymptomseverity.Howtheselocationsrelateto thethresholdsonthelatentvariablesissummarizedinthe person-itemMapinFig.1(emotionalexhaustion,depersonalisation)and Fig.2(personalaccomplishment).Atthesethresholds,itisequally likelyapersonwillbeclassiﬁedintoadjacentcategories,and there-foretoobtainoneoftwosuccessivescoresontheitem.Themeanof

thethresholdlocationswithinanitemisrepresentedbythe loca-tionparameter.Thelocationsrangefrom−0.31to4.27logits,where ‘callous(DP2)’wastheitemrepresentingthemostseveresymptom and‘fatigued(EE3)’wastheleastseveresymptom.Theitem-fitto theRaschmodelisalsosummarized inTable3.BoththeOutfit andInfitprovideinformationregardingresidualsincomparisonto thepredictedmodel.IfOutfitandInfitareequalto1,theobserved varianceisequaltothepredictedvariance.AnInfitof1.02 ‘work-ingtoohard’(EE7)meansthattheitemhas2%morevariancethan thepredictedmodel(Lai,Cella,Chang,Bode,&Heinemann,2003). Alltheitems,except‘effectively(DP2)’and‘don’treallycare(DP4)’ fitwithintheRaschmodelandhaveInfitandOutfitMeanSquared Errorsthatareproductiveformeasurement.Theitems‘effectively (DP2)’and‘don’treallycare(DP4)’sufferfromalowoutfitMSQ, indicatingthatrespondentswithlowburnout,scoreloweronthose itemsthanexpected,whereasrespondentswithhighburnoutscore higherontheseitemsthanexpectedbasedontheRaschmodel.This ismostlikelyduetothereductionofthescalerangeandthefew respondentswithscores>2ontheseitems.

3.4. Constructvalidityoftheraschburnoutscale

AnindividualscoreofapersonontheRaschscaleforburnout isexpressedinathetascore.Correlationsofthethetascoreswith theclassicalMBIscoreswerehigh,r=0.93,p<0.001foremotional exhaustion,andmoderatelystrongr=0.59,p<0.001for deperson-alisation.In order tovalidate theRasch burnout modelagainst otherconstructs,wealsoincludedameasureofworkengagement (UBES,Schaufeli&Bakker,2003).Thecorrelationsofthetascores forburnoutwithvigor,r=−0.45,p<0.001,dedication,r=−0.42, p<0.001,andabsorption,r=−0.21,p<0.001,areallnegativeand signiﬁcant,whichisindicativeofgoodconstructvalidity.

3.5. Personalaccomplishmentscale

Raschanalysisonthepersonalaccomplishmentscaleshowed that noreordering ofscores onitemswas necessary.Allitems showedmonotonouslyincreasingscores.Allitemlocationswere centredononeandahalfstandarddeviationhigherthanthe aver-agelatentPA.Therangeinlevelofpersonalaccomplishmentin theseitemsislow,withalocationof1.05forPA7andalevelof 1.45forPA6.TheresultsaresummarizedinTable4.The correla-tionoftheRaschthetascoreforburnoutwiththeRaschthetascore forpersonalaccomplishmentwasr=0.08,n.s.TheburnoutRasch scaleandthepersonalaccomplishmentscalearenotrelatedtoone another.

4. Discussion

Thisstudyshowsahypothesizedmodelofearlycareerburnout amongnursesasasequential-developmentalprocessconsistingof severalphases,wherepersonalaccomplishmentseemstodevelop independently. The hypothesis about the unidimensionality of emotionalexhaustionanddepersonalisationcouldbeaccepted,but onlyafterpersonalaccomplishmentwasidentifiedasaseparate dimensionandanalysedseparatelyinasecondRaschmodel.Early careerburnoutstartswithfatigueandendswithsevereexhaustion anddepersonalisationtowardspatients.Theitemsmeasuring per-sonalaccomplishmentallhaveanalmostsimilardifficultyleveland similarthresholds.Seenfromameasurementperspective,theygive thesameinformation,whichmakesthemnotverysuitableto mea-surechangesinpersonalaccomplishmentthatmightbeassociated withchangesinlevelsofburnout.Forthispurpose,newitemsfor measuringpersonalaccomplishmentwouldhavetobedeveloped. Our study is a conceptual replication of the findings that Gustavsson etal.(2010) reportedintheirworkonearlycareer

(6)

BDP4 BDP1 BEE6 BEE5 BEE2 BEE7 BEE1 BEE3 −3 −2 −1 0 1 2 3 4 Latent Dimension 2 1 1 2 1 2 1 2 4 3 2 1 3 2 1 2 1 3 2 1 3

Person−Item Map

Person Parameter Distribution

Fig.1.Person-ItemMapshowinglocationparametersoftherecodeditemsacrossthelatentburnoutvariableforsubscalesemotionalexhaustionanddepersonalisation.The topoftheﬁgureshowsthesampledistributionoverthelatentvariable.

burnoutamong Swedishnurses.Burnoutclearlydevelopsalong thelinesofemotionalexhaustion anddepersonalisationas out-linedbyCherniss(1980)andbyLeiterandMaslach(1988).But partiallyinlinewiththeTarisetal.(2005)model,personal accom-plishmentseemstodeveloponanindependenttrack.Acomparison oftheemotionalexhaustionanddepersonalisationﬁndingsofour studywiththeGustavssonetal.(2010)studyresultsinstriking similarities.

ThesymptomseverityhierarchyinTable3showsagoodfitfor theone-dimensionalsequentialdevelopmentalmodelofburnout fromCherniss(1980).Apersonwhoendorsesanitemwillmost likelyalsoendorseitemswithequallyseveresymptoms. There-fore,theitemswithsimilarseveritycanbeclusteredtogetherin threestages.Theleastseverecomplaintsarecapturedinemotional exhaustionitems(‘fatigued’(EE3),‘drained’(EE1), and‘working toohard’(EE7)representingthefirststageofcrisisandstrain.In thesecondstage,emotionalexhaustionpersists(‘usedup’(EE2), ‘frustrated’(EE6),‘burnedout’(EE5)),withastartofdefensive cop-ing(‘objects’(DP1),‘don’treallycare’(DP4)).Inthelastphasethe mostsevereexhaustionitems(‘strain“(EE4))arelocated,together withtwodepersonalisationitems(“hardening”(DP3)and“callous” (DP2)).Gustavssonetal.(2010)labelledtheirthreestagesasa “suf-feringfrompressure”stage,followedbya“frustration”stage,and finallya“withdrawalanddetachment”stage.Thesamelabelscan easilybeappliedtothethreestagesinourfindings.We concep-tuallyreplicatedaRaschmodelforburnoutfortheMBIthatwas earlierfoundwiththeOLBI.IfourfindingswiththeMBIcanbe repli-cated,amucheasierone-dimensionalscreeningtoolforburnout mightaresult.

Although many scientists will state that reduced personal accomplishmentisadimensionofburnout(Lee&Ashforth,1993; Maslach & Jackson, 1981; Taris et al., 2005), other researchers questionthis (Friedman, 1993; Kristensenet al.,2005; Schutte, Toppinen,Kalimo, &Schaufeli,2000).Our ﬁndingssuggest that personalaccomplishmentdevelopslargelyindependentfromthe othertwoburnoutdimensions.EarlierworkbyFriedman(1993)on burnoutinteachersalsoshowedthatdepersonalisationand emo-tionalexhaustiondevelopstoa climaxofburnout,independent frompersonalaccomplishment.Theseresultsalsosupportearlier workbyWalkeyandGreen(1992),whoshowedthatscoresonthe MBIquestionnairecanbeexplainedbytwomainfactors.Theﬁrst isexplainedasthe‘coreofburnout’andconsistsofthescales emo-tionalexhaustionanddepersonalisation;andthesecondfactoris personalaccomplishment,whichdevelopsindependentlyfromthe otherscales.

Schutteandassociates(2000)statethatthedimensionpersonal accomplishmentmaynotbepartoftheburnoutconcept.Houkes, Winants,Twellaar,andVerdonk(2011)investigatedtheetiologyof burnoutandfoundnoevidenceofpersonalaccomplishmentasa dimensionofburnoutamongmen.Forwomen,theydidﬁnd evi-dencefor personal accomplishmentasa dimensionof burnout, butreducedpersonal accomplishmentwasonlytriggeredwhen higherlevelsofdepersonalisationarose.Onemightconcludethat therelationofpersonalaccomplishmentwiththecoredimension ofburnout iscomplicatedand thatgenderhastobetakeninto account.Oursamplemainlyconsistedofwomanandthisprecludes analysesthatmightshedmorelightonthegenderissueinburnout.

(7)

BPA6 BPA3 BPA1 BPA4 BPA5 BPA2 BPA7 −5 −4 −3 −2 −1 0 1 2 3 4 5 6 7

Latent Dimension

6 5 4 3 2 1 6 5 4 3 2 1 6 5 4 3 2 1 6 5 4 3 2 1 6 5 4 3 2 1 6 5 4 3 2 1 6 5 4 3 2 1

Person−Item Map

Person Parameter Distribution

Fig.2.Person-ItemMapshowinglocationparametersoftheitemsacrossthelatentburnoutvariableforthesubscalepersonalaccomplishment.Thetopoftheﬁgureshows thesampledistributionoverthelatentvariable.

Incomparisonwithnormtables,the“classical”meanscoreon personalaccomplishmentofthenursesinoursampleisaverage (Schaufeli&vanDierendonck,2000).Duchscher(2009)statesthat lackofpracticeexperienceandconﬁdence,unrealisticperformance expectationsfromtheirsurroundings(institution,colleagues)and themselves,lossofcontrolovertheirprofessionalrole,and inse-curities in communicating and relating to new colleagues, are commonamongnursesduringtheinitialmonthsofwork.

Thereareseverallimitationsinthisstudy.Theﬁrstoneconcerns thedesignofthestudy,across-sectionalsurveyconsistingofonly onemeasurement.Totestahypothesisaboutthedevelopmentofa syndromeovertimeisdifﬁcult.Alongitudinaldesignwithatleast twomomentsofmeasurementwouldbemoreappropriate, more-overadesignwithmuchmoremeasurementssuchasinexperience samplingmethodologymightberequired.

The second shortcoming concerns the sample, which cov-eredanormalworkingpopulation,i.e.relativelyhealthyworkers, whereasthosewhoareill,orhavelefttheirworkofstress-related problemsarenotconsidered.Thereforeourresearchmightsuffer fromwhatiscalledthe“healthyworkereffect”(Schaufelietal., 2001).Becauseweusedaconveniencesample,onemightargueon theotherhandthatmorenurseswhosufferedfromburnout com-plaintschosetoparticipateinthisstudy.Asthemodelshouldbe applicabletoexactlythispopulation,thisshouldnotposeathreat tothevalidityofthemodel.

Thisstudyhastheoreticaland practicalimplications. Under-standingtheprocessof earlycareerburnout canfacilitateearly

recognitionofburnout among nurses.A clearscoretomeasure thedegreeof burnoutortheburnout phasesomeoneis experi-encinggivespracticalbenefitssuchasearlytreatment,directions fortreatmentandprevention.Thesefindingsshouldbereplicated forotherprofessions,andamongworkerswhoalreadysufferfrom workstress-relatedhealthproblems.Theuseoflongitudinal stud-iestoreplicatethesefindingsisrecommendedtoconfirmearly careerburnoutasaprocessstartingwithexhaustion.

References

Arrindell,W.A.,&vanderEnde,J.(1985).Anempiricaltestoftheutilityofthe observations-to-variablesratioinfactorandcomponentsanalysis.Applied PsychologicalMeasurement,9(2),165–178.http://dx.doi.org/10.1177/ 014662168500900205

Bartz,C.,&Maloney,J.P.(1986).Burnoutamongintensivecarenurses.Researchin Nursing&Health,9(2),147–153.http://dx.doi.org/10.1002/nur.4770090210 Beecroft,P.C.,Dorey,F.,&Wentem,M.(2008).Turnoverintentioninnewgraduate

nurses:amultivariateanalysis?TheJournalofAdvancedNursing,62(1),41–52. Bowles,C.E.,&Candela,L.E.(2005).FirstjobexperiencesofrecentRNgraduates: improvingtheworkenvironment.TheJournalofNursingAdministration,35(3), 130–137.

Browne,M.W.,&Cudeck,R.(1992).Alternativewaysofassessingmodelﬁt. SociologicalMethods&Research,21(2),230–258.http://dx.doi.org/10.1177/ 0049124192021002005

Cherniss,C.(1980).Staffburnout:jobstressinthehumanservices.BeverlyHills,CA: SagePublications.

Cho,J.,Laschinger,H.K.S.,&Wong,C.(2006).Workplaceempowerment,work engagementandorganizationalcommitmentofnewgraduatenurses. CanadianJournalofNursingLeadership,19(3),43–60.

(8)

Demerouti,E.,&Bakker,A.B.(2008).TheOldenburgBurnoutInventory:agood alternativetomeasureburnoutandengagement.InJ.R.B.Halbesleben(Ed.), Handbookofstressandburnoutinhealthcare.Hauppauge,NY:NovaScience. Duchscher,J.E.B.(2009).Transitionshock:theinitialstageofroleadaptationfor

newlygraduatedRegisteredNurses.JournalofAdvancedNursing,65(5), 1103–1113.http://dx.doi.org/10.1111/j.1365–2648.2008.04898.x Duchscher,J.B.,&Myrick,F.(2008).Theprevailingwindsofoppression:

understandingthenewgraduateexperienceinacutecare.NursingForum, 43(4),191–206.http://dx.doi.org/10.1111/j.1744-6198.2008.00113.x Friedman,I.A.(1993).Burnoutinteachers:theconceptanditsuniquecore

meaning.EducationalandPsychologicalMeasurement,53(4),1035–1044.http:// dx.doi.org/10.1177/0013164493053004016

Golembiewski,R.T.,Munzenrider,R.F.,&Stevenson,J.G.(1986).Phasesofburnout: developmentsinconceptsandapplications.NewYork:Preager.

Gustavsson,J.P.,Hallsten,L.,&Rudman,A.(2010).Earlycareerburnoutamong nurses:modellingahypothesizedprocessusinganitemresponseapproach. InternationalJournalofNursingStudies,47(7),864–875.http://dx.doi.org/10. 1016/j.ijnurstu.2009.12.007

Hellenthal,A.(2012).Verpleegkundigenaanhetwerk[Nursesatwork]..Retrieved February162016from http://www.cbs.nl/NR/rdonlyres/571EAABA-EBC5-4B48-A66B-9656CF2F6ED0/0/2012k1v4p55art.pdf

Houkes,I.,Winants,Y.,Twellaar,M.,&Verdonk,P.(2011).Developmentofburnout overtimeandthecausalorderofthethreedimensionsofburnoutamongmale andfemaleGPs.Athree-wavepanelstudy.BMCPublicHealth,11(1),240. http://dx.doi.org/10.1186/1471-2458-11-240

Hu,L.T.,&Bentler,P.M.(1999).Cutoffcriteriaforﬁtindexesincovariance structureanalysis:conventionalcriteriaversusnewalternatives.Structural EquationModeling:AMultidisciplinaryJournal,6(1),1–55.http://dx.doi.org/10. 1080/10705519909540118

Kristensen,T.S.,Borritz,M.,Villadsen,E.,&Christensen,K.B.(2005).The CopenhagenBurnoutInventory:anewtoolfortheassessmentofburnout. Work&Stress,19(3),192–207.http://dx.doi.org/10.1080/02678370500297720 Lai,J.,Cella,D.,Chang,C.H.,Bode,R.K.,&Heinemann,A.W.(2003).Itembanking

toimprove,shorten:andcomputerizeself-reportedfatigue:anillustrationof stepstocreateacoreitembankfromtheFACIT-Fatiguescale.QualityofLife Research,12,485–501.

Laschinger,H.K.S.,Finegan,J.,&Wilk,P.(2009).NewGraduateBurnout:the impactofprofessionalpracticeenvironment,workplacecivility,and empowerment.NursingEconomics,27(6),377–383.

Lee,R.T.,&Ashforth,B.E.(1993).Alongitudinalstudyofburnoutamong supervisorsandmanagers:comparisonsbetweentheLeiterandMaslach (1988)andGolembiewskietal.(1986)models.OrganizationalBehaviorand HumanDecisionProcesses,54(3),369–398.http://dx.doi.org/10.1006/obhd. 1993.1016

Lee,I.,&Wang,H.H.(2002).Perceivedoccupationalstressandrelatedfactorsin publichealthnurses.JournalofNursingResearch,10(4),253–260.http://dx.doi. org/10.1097/01.jnr.0000347606.91295.76

Leiter,M.P.,&Maslach,C.(1988).Theimpactofinterpersonalenvironmenton burnoutandorganizationalcommitment.JournalofOrganizationalBehavior, 9(4),297–308.http://dx.doi.org/10.1002/job.4030090402

Linacre,J.M.(2002).Optimizingratingscalecategoryeffectiveness.Journalof AppliedMeasurement,3(1),85–106.

Maslach,C.,&Jackson,S.E.(1981).Themeasurementofexperiencedburnout. JournalofOrganizationalBehavior,2(2),99–113.http://dx.doi.org/10.1002/job. 4030020205

Maslach,C.,&Leiter,M.P.(1997).Thetruthaboutburnout:howorganizationscause personalstressandwhattodoaboutit.NewYork:Jossey-Bass.

Parker,P.D.,&Salmela-Aro,K.(2011).Developmentalprocessesinschoolburnout: acomparisonofmajordevelopmentalmodels.LearningandIndividual Differences,21(2),244–248.http://dx.doi.org/10.1016/j.lindif.2011.01.005 RCoreTeam.(2013).R:Alanguageandenvironmentforstatisticalcomputing.

Vienna,Austria:RFoundationforStatisticalComputing[URL]. http://www.R-project.org/

Rosseel,Y.(2012).Lavaan:anRpackageforstructuralequationmodeling.Journal ofStatisticalSoftware,48(2),1–36.

Schaufeli,W.B.,&Bakker,A.B.(2003).Deutrechtsebevlogenheidschaal:handleiding [Theutrechtengagementscale:manual].Utrecht:UniversiteitUtrecht. Schaufeli,W.B.,Bakker,A.B.,Hoogduin,K.,Schaap,C.,&Kladler,A.(2001).Onthe

clinicalvalidityofthemaslachburnoutinventoryandtheburnoutmeasure. Psychology&Health,16(5),565–582.http://dx.doi.org/10.1080/

08870440108405527

Schaufeli,W.B.,&Buunk,B.P.(2003).Burnout:anoverviewof25yearsofresearch andtheorizing.InM.J.Schabracq,J.A.M.Winnubst,C.L.Cooper,&J.Wiley (Eds.),Thehandbookofworkandhealthpsychology(pp.383–425).Chichester UK:Wiley.

Schaufeli,W.B.,&Enzmann,D.(1998).Theburnoutcompaniontostudyand practice:acriticalanalysis.London:TaylorFrancis.

Schaufeli,W.B.,&vanDierendonck,D.(2000).UBOSutrechtseburnoutschaal: handleiding[DutchMBI-NLmanual].Lisse:SwetsTestPublishers. Schaufeli,W.B.,Maslach,C.,&Marek,T.(1993).Professionalburnout:recent

developmentsintheoryandresearch.Washington:Taylor&Francis.

Schutte,N.,Toppinen,S.,Kalimo,R.,&Schaufeli,W.(2000).Thefactorialvalidityof theMaslachBurnoutInventory-GeneralSurvey(MBI-GS)acrossoccupational groupsandnations.JournalofOccupationalandOrganizationalPsychology, 73(1),53–66.http://dx.doi.org/10.1348/096317900166877

Smith,E.V.,&Smith,R.M.(2004).IntroductiontoRaschmeasurement:theory, modelsandapplications.Minnesota:MapleGrove.

Taris,T.W.,LeBlanc,P.M.,Schaufeli,W.B.,&Schreurs,P.J.G.(2005).Arethere causalrelationshipsbetweenthedimensionsoftheMaslachBurnout Inventory?Areviewandtwolongitudinaltests.Work&Stress,19(3),238–255. http://dx.doi.org/10.1080/02678370500270453

VanderWindt,W.,Smeets,R.C.K.H.,&Arnold,E.J.E.(2009).Regiomarge2009:De arbeidsmarktvanverpleegkundigen,verzorgendenensociaalagogen2009–2013 [Region2009:Thejobmarketfornurses,caregiversandsocialworkers 2009–2013].Utrecht:Prismant.

VanderWindt,W.,vanderVelde,F.,vanderKwartel,A.,&Prismant.(2009).Arbeid inzorgenwelzijn2009[Labourincareandwell-being2009].[Retrievedfrom]. ww.azwinfo.nl/handlers/ballroom.ashx?function=download&id=16 VanEssen,G.,Kramer,S.,vanderVelde,F.,&vanderWindt,W.(2016).Arbeidin

zorgenwelzijn,jeugdzorgenkinderopvang2015[Labourincareandwell-being youthandchildcare].[Retrievedfrom].http://www.caop.nl/ﬁleadmin/ bestanden/documenten/nieuws-agenda/2016/AZWrapport.pdf

Verhelst,N.D.(1993).Itemresponstheorie[Itemresponsetheory].InP.F.Sanders (Ed.),Psychometrieindepraktijk[PracticeofPsychometrics].Arnhem,The Netherlands:CitoInstituteforTestDevelopment.

Walkey,F.H.,&Green,D.E.(1992).Anexhaustiveexaminationofthereplicable factorstructureofthemaslachburnoutinventory.Educationaland PsychologicalMeasurement,52(2),309–323.http://dx.doi.org/10.1177/ 0013164492052002006

Winwood,P.C.,Wineﬁeld,A.H.,&Lushington,K.(2006).Work-relatedfatigueand recovery:thecontributionofage,domesticresponsibilitiesandshiftwork. JournalofAdvancedNursing,56(4),438–449.http://dx.doi.org/10.1111/j. 1365-2648.2006.04011.x