The impact of marital status on health care utilization amon...

A substantial literature supports a positive effect of marriage on health. Married individuals, especially men, have a longer life expectancy than the unmarried ... TheimpactofmaritalstatusonhealthcareutilizationamongMedicarebeneficiaries:Medicine March2019-Volume98-Issue12 Previous Article Next Article ArticleasEPUB ExportAllImagestoPowerPointFile AddtoMyFavorites Colleague'sE-mailisInvalid YourName: Colleague'sEmail: Separatemultiplee-mailswitha(;). Message: Thoughtyoumightappreciatethisitem(s)IsawinMedicine. Yourmessagehasbeensuccessfullysenttoyourcolleague. Someerrorhasoccurredwhileprocessingyourrequest.Pleasetryaftersometime. EndNoteProciteReferenceManager Savemyselection ResearchArticle:ObservationalStudyPandey,KiranRajMD,MSa,∗;Yang,FanPhDb;Cagney,KathleenA.PhDc;Smieliauskas,FabricePhDd;Meltzer,DavidO.MD,PhDa,e;Ruhnke,GregoryW.MD,MS,MPHeEditor(s):Sayed.,MazenEl AuthorInformation aTheCenterforHealthandtheSocialSciences,UniversityofChicago,IL bDepartmentofBiostatisticsandInformatics,UniversityofColoradoDenver,Aurora,CO cDepartmentofSociology,UniversityofChicago dDepartmentofPublicHealthSciences,UniversityofChicago eSectionofHospitalMedicine,DepartmentofMedicine,UniversityofChicago,Chicago,IL60637,USA. ∗Correspondence:KiranRajPandey,TheCenterforHealthandtheSocialSciences,UniversityofChicago,5841SouthMarylandAvenue,MC1000,Chicago,IL60637,USA(e-mail:[email protected]). Abbreviations:ADL=activitiesofdailyliving,AOR=adjustedoddsratio,CI=confidenceinterval,IADL=instrumentalactivitiesofdailyliving,MCBS=MedicareCurrentBeneficiarySurvey,OPD=outpatientdepartment,OR=oddsratio,SNF=skillednursingfacility. Noneoftheauthorshaveanyconflictofinteresttodeclare. SupplementalDigitalContentisavailableforthisarticle. Supplementaldigitalcontentisavailableforthisarticle.DirectURLcitationsappearintheprintedtextandareprovidedintheHTMLandPDFversionsofthisarticleonthejournal'sWebsite(www.md-journal.com). ThisisanopenaccessarticledistributedunderthetermsoftheCreativeCommonsAttribution-NonCommercialLicense4.0(CCBY-NC),whereitispermissibletodownload,share,remix,transform,andbuilduptheworkprovideditisproperlycited.Theworkcannotbeusedcommerciallywithoutpermissionfromthejournal.http://creativecommons.org/licenses/by-nc/4.0 Medicine: March2019-Volume98-Issue12-pe14871 doi:10.1097/MD.0000000000014871 Open SDC Metrics Abstract ToexplainpriorliteratureshowingthatmarriedMedicarebeneficiariesachievebetterhealthoutcomesathalftheperpersoncostofsinglebeneficiaries,weexamineddifferentpatternsofhealthcareutilizationasapotentialdriver. UsingtheMedicareCurrentBeneficiarySurvey(MCBS)data,wesoughttounderstandutilizationpatternsinmarriedversuscurrently-not-marriedMedicarebeneficiaries.Weanalyzedtherelationshipbetweenmaritalstatusandhealthcareutilization(classifiedbasedonsettingofcareutilizationintooutpatient,inpatient,andskillednursingfacility(SNF)use)usinglogisticregressionmodeling.WespecifiedmodelstocontrolforpossibleconfoundersbasedontheAndersenmodelofhealthcareutilization. Basedon13,942respondentsintheMCBSdataset,12,929hadcompletedata,thusformingtheanalyticsample,ofwhom6473(50.3%)weremarried.Ofthese,58%(vs.36%ofthosecurrently-not-married)weremale,45%(vs.47%)wereage>75,24%(vs.70%)hadahouseholdincomebelow$25,000,18%(vs.14%)hadexcellentself-reportedgeneralhealth,and56%(vs.36%)hadprivateinsurance.Comparedtounmarriedrespondents,marriedrespondentshadatrendtowardhigheroddsofhavingarecentoutpatientvisit(unadjustedoddsratio(OR)1.11,95%confidenceinterval(CI)1.04–1.19,adjustedoddsratio(AOR)1.10,(CI)0.99–1.22),andloweroddsintheyearpriortohavehadaninpatientstay(AOR0.84,CI0.72–0.99)oraSNFstay(AOR0.55,CI0.40–0.75). BasedonMCBSdata,oddsofself-reportedinpatientandSNFusewereloweramongmarriedrespondents,whileunadjustedoddsofoutpatientusewerehigher,comparedtocurrently-not-marriedbeneficiaries. 1Introduction Therearereasonstosuspectthathealthcareutilizationpatternsmaydifferaccordingtomaritalstatus.First,healthcarecostsamongmarriedMedicarebeneficiariesareapproximatelyhalfofthosenotmarried;[1]thisdifferenceincostsisobservedamongthenonelderly(age<65)aswell,butfortheelderly,thedifferenceissignificantlygreater.[2]Second,socio-demographicfactorshavebeendemonstratedtobeimportantdeterminantsofhealthcareutilization.[3–5]Third,unmarriedindividualshavereducedaccesstoresourcesthatmayaffectutilization(e.g.healthinsuranceanddisposableincome)thanthosewhoaremarriedandmayengageinriskierhealth-relatedbehaviors,[6,7]possiblyimpactingutilization.Fourth,previousresearchsuggeststhatbeingmarriedispredictiveofbetterhealthstatus,[8–10]perhapsattributabletomoreeffectivepatternsofutilization.Thisstudyanalyseshealthcareutilizationpatternsbasedonmaritalstatusandaimstoexploretheimportanceofandmechanismsbywhichmaritalstatusmayinfluencehealthcareutilization,costs,andoutcomes. Asubstantialliteraturesupportsapositiveeffectofmarriageonhealth.Marriedindividuals,especiallymen,havealongerlifeexpectancythantheunmarried,[10,11]bettermentalhealth,[12]andreportgreatersatisfactionwithoverallqualityoflife.[13]Theobservedrelationshipbetweenmarriageandhealthwasinitiallythoughttobedueto“marriageselection”(i.e.,marriedindividualsmaybeself-selectedbasedonhealth-relatedcharacteristics,attitudestowardhealth,orbehavioralfactors).[14]However,thereisagrowingliteratureregardingtheconceptof“marriageprotection.”Thisconceptimpliesaprotectiveroleofastrongsocialrelationshipthatmayresultinbetterhealthbecausespouses(especiallywomen)functionascaretakers,providingphysicalandemotionalsupport.[15–17]TheideathatsocialcircumstancesimpacttheproductionofhealthisalsosupportedbytheGrossmanmodelofhealthcapital.[18]AccordingtotheGrossmanmodel,individualsarenotonlymereconsumersbutalsoproducersofhealthcare,andthatsuchproductionofhealthcareisimpactedbyinvestmentsanindividualmakestoimprovetheirhealth,oneofwhichisleveragingpersonalsocialcapital,suchasmaritalstatus.[19] However,becausetheliteraturedoesnotcorroboratethesameextentofhealthbenefitsamongunmarriedcohabitingcouples,marriagemayinfluencehealthstatusinwaysotherthanthe“supportfunction”ofcohabitation.[6,13]Cohabitingcouplesarelesslikelytoshareresources(e.g.healthinsurancebenefitsorfinances)thanmarriedones.Inananalysisofyoungmarriedwomen(age24–34),thepositivehealtheffectsofmarriagewerefoundonlyamongtheunemployed,suggestingthattherelationshipbetweenmarriageandhealthcareutilizationmaybemodifiedbyadditionalsocialcircumstances.[20]Inaddition,risk-adjustedhealthcareexpendituresareloweramongmarriedcouplesdespitebetterhealthoutcomes(definedastheextentofchangeinorpreservationofhealthstatusfollowinganintervention).Thesefindingsraisethepossibilitythatmaritalstatusmayinfluencehealthoutcomesthroughefficient(improvedoutcomesforagivencostorreducedcostswithoutadecrementinhealthoutcomes)utilizationofhealth-relatedresources.Theaforementioneddisparateassertionsandconclusionsleadtotwoquestions:First,domarriedcouplesdifferintheirpatternsofhealthcareutilization,andsecond,ifsuchadifferencetrulyexists,isitresponsibleforthepreviouslyobserveddifferencesincostsandhealthoutcomes?Inthisanalysis,wefocusonansweringtheformerquestion. Toourknowledge,thepublishedliteraturedoesnotincludearigorousanalysisoftheassociationbetweenmaritalstatusandpatternsofutilizationacrosssettingsofhealthcareprovision.Mostresearchonhealthcareutilizationisfocusedonspecificdiseaseconditions,asinglesettingofcare,oramongpatientswithaparticularinsurancestatus,muchofwhichisnotgroundedinarobusttheoreticalframework.[21–25]Somestudieshavefocusedoncostsinsteadofserviceutilizationastheprimaryoutcomemeasure.[1] Understandinganydifferenceinhealthcareutilizationassociatedwithmaritalstatusisimportant,particularlyaspopulationdemographicsshifttowardunmarriedstatusandfamilystructureschange.[26–28]Ifthereareindeedsignificantdifferencesinpatternsofhealthcareutilizationaccordingtomaritalstatus,thisinsightmayproveusefulinthefuturetoimprovehealthoutcomesandmaximizetheefficiencyofhealthcareprovision.Inconductingthisstudy,wehypothesizedthatmarriedMedicarebeneficiarieswouldexhibitamoreefficientpatternofhealthcareutilizationacrosssettingsofcare. 2Methods 2.1Conceptualframework WeusedthewidelyrecognizedAndersenmodelofhealthcareutilizationastheconceptualmodeltoexploretherelationshipbetweenmaritalstatusandhealthcareutilization.TheAndersenmodelisconceptuallyrelevantforanalyzingthisrelationshipbecausecharacterizationsofsocialstructureareacentralcomponentofthismodelandthereforegermanetoourresearchquestion.[29]Themodelimpliesthathealthcareutilizationisdeterminedbypredisposingfactors,enablingfactorssuchascommunityresources,andneedordemand-relatedfactors.[3] TheoriginalAndersenmodelwasproposedin1968tostudyhealthcareutilizationbyfamilies,anditslatermodificationshavebeenwidelyacceptedandusedasaconceptualmodeltounderstandindividualhealthcareutilization.[29–31]Basedonthisconceptualmodel,weextractedvariablesfromourdatasetpotentiallyimportantinconfoundingtherelationshipbetweenmaritalstatusandhealthcareutilizationtominimizeanyinaccuracyoftheeffectestimate.Thepotentialconfoundingfactorsforwhichwecontrolledincludedemographicinformation(age,race,sex,education,andincome),healthstatus(self-ratedgeneralhealth,difficultywalking,andmeasuresoffunctionallimitations),insurancestatus(additionalprivateinsuranceordualMedicareandMedicaidcoverage),andtendencytousehealthcare,whichweascertainedusingreceiptofaninfluenzavaccinationasaproxymeasure.[32] 2.2Dataandvariables WeusedtheMedicareCurrentBeneficiarySurvey(MCBS),alongitudinalin-personquarterlysurveythatfollowsanationallyrepresentativesampleofMedicarebeneficiariesforaperiodof3years.[33,34]TheMCBSisacomplexrandomsocialsurveydesignedtorepresentallcurrentMedicareenrolleesintheUnitedStatesandPuertoRico.Weusedthe2013MCBSAccessToCarePublicUseFile,alimitedde-identifieddatasetof13,924surveyparticipantsthatincludes459variablesobtainedfromtheMCBSRound67aswellasdatafromadministrativesources.[35] Ouroutcomesofinterestwereoutpatientdepartment(OPD)visits,inpatienthospitalizationandskillednursingfacility(SNF)use.OPDusewasmeasuredovertherecentpastandwasobtainedthroughsurveyrespondentreport.Thesurveyquestionnaireascertainsinformationbasedoneachrespondent'smostproximateOPDvisitationpriortothecurrentsurveyadministration.Becausethesurveywasconductedevery4months,theOPDvariableislikelytoreflectOPDuseinthefewmonthspriortotheinterview.InpatientandSNFusewerederivedfromtheadministrativedatalinkage,measuringutilizationovera1-yearperiodpriortotheinterview.Giventheleft-skeweddistributionoftheoutcomevariables,wedichotomizedallthreeintozeroormorethanzerovisits/admissions. Inouranalysis,wecodedmaritalstatusasadichotomousvariablebetweenmarriedandcurrently-not-married,thelatterbeinganaggregatecategoryincludingrespondentswhowerewidowed,divorcedornevermarried.Ourdecisiontospecifythisaggregatecategorywasbasedonpriorliteraturesuggestingsimilarhealthrisks(e.g.,mortalityrates)acrosstypesofcurrentlyunmarriedindividuals.[7,36] Thedatasetincludedagebrackets(<65,65–75,>75)ascategoricalvariables,whichweusedassuchintheanalysis.Wecategorizedsex,race,andhouseholdsize.Weusedself-ratedhealthstatusasarepresentativepredisposingfactor.Additionally,weincludeddifficultywalkingandavariablerepresentingdegreeoffunctionallimitation(measuredbythenumberofActivitiesofDailyLiving(ADL)andInstrumentalADLlimitations)aspotentialconfoundingvariables,allofwhichwecategorized.WemeasuredaccesstocarebasedonthepresenceofadualMedicareandMedicaidinsurancebenefitorprivateinsurance.Weusedinfluenzavaccinationintheyearpriortothesurveyasaproxymeasureofhealth-relatedbehaviorandtendencytousehealthcare.Forallcategoricalvariables,weusedthefirstcategoryasthereferencecategory. 2.3Analyticapproach Wecreatedmultivariatelogisticregressionmodelsbasedonthe2013MCBSPublicUseFilecross-sectionaldatatoestimatetherelationshipbetweenmaritalstatusandeachofthethreeoutcomevariables(OPD,inpatient,andSNFuse).Giventheskewed(towardzero)distributionoftheoutcomevariables,weusedlogisticregressionmodelsratherthannegativebinomialorPoissonmodelsforstatisticalefficiencyandinterpretabilityadvantages.Wecontrolledforpotentialconfoundingfactorsaswasdetailedabove.Foreachoutcomevariable,thefirstmodelestimatedtheunadjustedrelationshipbetweenmaritalstatusandthatmeasureofutilization.Thefinalmultivariate-adjustedmodelincludedvariablesrepresentingdemographicfactors,predisposingfactors,insurancestatus,healthbehavior,andtendencytoutilizehealthcare.ModelspecificationwasbasedontheAndersenmodelinadditiontotheaforementionedstatisticalconsiderations. Wealsotestedadditionalmodelsforeachofthethreeoutcomevariablesbyincludingseveraladditionalvariablesmeasuringfactorsofpotentialimportancelikehouseholdstructure(acategoricalvariabledenotinghouseholdcomposition),andalimitedsetofself-reportedcomorbidconditions.TheP-valueswerederivedusingPearson'sChi-squaredtestoncontingencytablesofeachvariableversusdichotomizedmaritalstatus.StatisticalanalyseswereperformedusingRv.3.3.2(inadditiontothebasepackage,weusedtidyverse,plyr,lmtest,sjPlot,stargazer).WedidnotseekInstitutionalReviewBoardapprovalinconductingthisstudybecauseweusedapubliclyavailablede-identifieddatasetcreatedbasedontheMCBS. 2.4Subgroupanalysis Inordertoexploretheassociationofnonspousalcohabitation(e.g.,withsiblingsorchildren)andhealthcareutilization,wefurtheranalyzedthecurrently-not-marriedsubgroup.Withinthissubgroup,wecomparedutilizationamongthosewhowerelivingalonetothosewhowerelivingwithothernon-spousalhouseholdmembers.Wecreatedlogisticregressionmodelsspecifyinghouseholdsize(dichotomousvariablewith1ormorethan1membersascategories)asthemainpredictorvariables,theoutcomevariablesremainingaspreviouslydescribed.Wecontrolledforpotentialconfoundingfactorsidenticaltothewaydescribedabove. 3Results Ofourinitialstudysampleof13,924respondents,995wereexcludedbecauseofmissingdata,creatingthefinalanalyticsampleof12,929observations.Table1exhibitsthedistributionofhealthcareutilizationacrossthethreesettingsofcarepreviouslydescribed,aswellasthedistributionofrespondents’maritalstatus.Table1:Frequencydistributionofthestudysample,accordingtomaritalstatusandmeasuresofutilizationacrosssettingsofcare.Table2demonstratesthat,ofthe12,929respondentsintheanalyticsample,6473(50.3%)werecurrentlymarried,theremaining6456(49.7%)havingbeencurrently-not-marriedatthetimeofsurveyadministration.Marriedbeneficiariesweremorelikelytobemale,older,whiteandhavehigherincomeandeducationcomparedtothosecurrently-not-married.Theyalsohadbetterself-ratedhealthstatus,fewerfunctionallimitations,andweremorelikelytohaveprivateinsurance.Thosecurrently-not-marriedtendedtohavelowerincome,belongtoracialminorities,haveworseself-reportedhealthstatus,andweremorelikelytohavedualMedicareandMedicaidcoverage.Table2:MedicareCurrentBeneficiarySurvey(MCBS)2013respondentcharacteristics.Table3demonstratestheadjustedoddsratios(AORs)measuringtheassociationbetweenmaritalstatusandeachprespecifiedmeasureofutilization.Wesubdividethefollowingaccordingtoeachmeasureofutilization.Table3:HealthcareuseinMedicareBeneficiaries-LogisticRegressionModels.3.1Outpatientutilization Currentlymarriedrespondentshadhigherunadjustedoddsofhavinganoutpatientvisitintherecentpast,althoughthisrelationshipfailedtomaintainstatisticalsignificanceintheadjustedmodel(unadjustedOR1.11,95%confidenceintervalCI1.04–1.19,AOR1.10,CI0.99–1.22).IntheOPDmodel,age>75predictedreduceduseofOPD(OR0.82,[CI0.72–0.92].Similarly,HispanicswerelesslikelytohavehadanOPDvisitcomparedtowhites(AOR0.71,CI0.81–0.93).Eachdecrementinself-ratedhealthstatuswasassociatedwithanincreasinglygreaterlikelihoodofhavinghadanOPDvisit(e.g.poorhealthstatus,AOR1.91,CI1.60–2.29).Receiptofinfluenzavaccinationandhavingsupplementalprivateinsurancewerealsoassociatedwithhavinghadanoutpatientvisitintherecentpast. 3.2Inpatientutilization Marriedrespondentswerelesslikelytohaveaninpatientstayintheyearpriortotheinterview(unadjustedOR0.77,CI0.69–0.86,AOR0.84,[CI0.72–0.99]).Olderage,malesex,difficultywalking,poorself-ratedhealth,supplementalprivateinsurance(AOR2.03,CI1.78–2.33),dualMedicare/Medicaidcoverage(AOR1.88,CI1.59–2.23),wereassociatedwithagreateroddsofhospitalization. 3.3SNFutilization MarriedstatuswasassociatedwithamarkedlyreducedoddsofSNFuse(unadjustedOR0.46,CI0.36–0.58,AOR0.55,CI0.40–0.75).Largerhouseholdsizewasassociatedwithatrendsuggestinga29%reductioninSNFuse,albeitnotquitetostatisticalsignificance.Increasingage,lowself-ratedhealthstatus,anddifficultywalkingpredictedagreateragreateruseofSNF,asdidthereceiptofaninfluenzavaccination(OR1.87,CI1.38–2.59),privateinsurance(OR1.68,CI1.30–2.19),anddualMedicare/Medicaidcoverage(OR1.72,CI1.22–2.41). Weperformedadditionalanalysesthroughinclusionofvariablescodifyingcomorbiditiesandhouseholdstructure(todenotethenatureofrelationshipsinmulti-memberhouseholds).However,noneofthesecharacteristicsweresignificantpredictorsofutilizationanddidnotmodifytherelationshipbetweenmaritalstatusandtheutilizationmeasures(notreported). 3.4Additionalandsubgroupanalyses Inordertounderstandtherelativeeffectofspousalcohabitationversusnon-spousalcohabitationonutilization,weperformedasubgroupanalysisamongrespondentscurrently-not-married.Inthisanalysis,weusedadichotomizedhouseholdsizevariableastheprimarypredictorandthesameoutcomemeasures,controllingforthesamesetofpotentialconfoundersusedinthemainanalysis.Wedidnotfindaconsistentassociationbetweenhouseholdsizeandutilizationamongrespondentswhowerecurrentlynotmarried(Appendix1,https://links.lww.com/MD/C893). 4Discussion ThisanalysiswasaimedatelucidatingtherelationshipbetweenmaritalstatusandpatternsofhealthcareutilizationamongasampleofMedicarebeneficiaries.MaritalstatuswaspredictiveofreduceduseofinpatientandSNFservices,withatrendtowardgreateruseofoutpatientservices. Thereareseveralimportantinsightsfromourstudythatmayhaveimplicationsfromclinicalandpolicyperspectives.ReducedutilizationofmorecostlyinpatientandSNFcare,withgreaterutilizationofoutpatientservices,supportstheassertionthatthesocialcomponentsofmarriagemayresultinamoreefficientallocationofhealthcareresources.Priorliteraturehasidentifiedatrendtowardsubstitutionofinpatientutilizationwithadditionalcarereceivedintheoutpatientsetting.[37,38]Outpatientvisitsenhancecontinuityofcare,whichhasbeenassociatedwithreducedhospitalizationrates.[39,40]Inparticular,increaseduseofoutpatientservicesreduceshospitalizationratesamongpatientswithAmbulatoryCareSensitiveConditions.[41–43]Infact,asmanyas9%–22%ofhospitaladmissionsarethoughttobeavoidablewithadequateprovisionofoutpatientcare.[44]Therefore,agreaterrelianceonreadilyavailableoutpatientcaremaybeonemechanismbywhichmarriedpeopleachievebetterhealthoutcomes.Additionally,outpatientvisitsarealsomuchlesscostlythaninpatientandSNFstays,[2,45]creatingopportunitiesforefficientprovisionofcarewhilepreservingthecontinuitythatisthecornerstoneoflongitudinalphysician-patientrelationships.ThetendencytosubstitutecostlyinpatientandSNFserviceswithlessexpensiveoutpatientservicesislikelyanimportantreasonwhymarriedMedicarebeneficiarieshavesignificantlylowerhealthcarecoststhanthosecurrently-not-married.Inadditiontoourprimaryfindingsregardingtherelationshipbetweenmaritalstatusandhealthcareutilizationacrossmultiplesettingsofcare,ouranalysisrevealsthatolderMedicarebeneficiarieshaveloweroutpatientutilization,buthigherinpatientandSNFutilization.Asexpected,thosewithworseself-ratedhealthstatus,agreaternumberoffunctionallimitations,andadditionalinsurancebenefitshavegreaterutilizationacrossalldomainsofcare.Thesefindingsareconsistentwithprioranalyses.[46–48] Ourfindingssupporttheassertionthatmaritalstatusisanimportantpredictorofhealthcareutilization.Althoughtheobservationalnatureofthedatasourcelimitsourabilitytodrawtruecausalinferences,thisanalysisdoesexhibitastrongassociation,robusttomultivariatemodeling,betweenmaritalstatusandhealthcareutilizationinthethreedomainsanalyzed.This‘‘spousaleffect”isalsocorroboratedbyourfindingthatunmarriedcohabitatingindividualsdonothavethesamepatternofreducedinpatientandSNFutilizationwithatrendtowardgreateroutpatientutilization.Thesefindingsraisethepossibilitythatmarriagemayinfluencehealthstatusnotonlythroughthesupportandprotectionthatmarriageoffers,butalsothroughamoreefficientpatternofhealthcareutilization. Inordertotranslatethesefindingsintorelevantpolicy,itisimportanttounderstandthecausalmechanismsforthis‘‘spousaleffect”inhealthcareutilization.Ourstudydoesnotallowustounambiguouslydelineatethesemechanisms.However,severalmechanismscouldexplaintheobservedassociations.First,spousesmayfunctionasin-homecaretakers,therebyobviatingtheneedforformalhealthcare.[17]Forexample,spousesmayassistwithmedicationadherence,preparingandencouragingtheconsumptionofhealthymeals,orensuringattendanceatphysicianappointments.[49]Second,theremaybepsychologicalandphysicalhealtheffectsofspousalcohabitation,andthatmayconsequentlyinfluencethevolumeanddistributionofhealthcareutilization.[5,15]Third,marriagehasalsobeennotedtocreatewhathasbeenreferredtoas“sparecapacity”–theabilitytodedicateone'stime,effort,andavailablehealthcareresourcestoimprovehealthasaresultofdivisionoflaborandsharedresponsibilitieswithinthehome.[50]Similarly,marriagealsofacilitatesengagementinresourcesharingandmutualinvestment.Fourth,theremaybeaselfselectionintomarriage,oftenreferredtoas“marriageselection,”resultingfromunobservedcharacteristicsthatinfluencebothhealthcareutilizationandhealth.[14,51]Wehavetriedtocontrolforthispotentialselectionasrigorouslyasourdatawouldallowthroughregressioncontrols.Consequently,webelievethatourestimationreflectstheeffectofmarriagemediatedpredominantlybythefirstthreemechanisms. WebelievethetrendtowardgreaterOPDutilizationobservedamongmarriedrespondentsismediatedbyacombinationoffactors:(a)greateraccesstoresources,includinghighertotalincomeandmoreprivateinsuranceenablingaccesstouninterruptedcare;(b)greaterhealth-enhancinghomesupportandassistanceimprovingcompliancewithoutpatientappointments;(c)andhigherlevelsofeducationcontributingtointelligentandcost-effectivedecisionsregardingtheappropriatesettinginwhichtopursuecare.ThevalidityoftheseassertionsissupportedbythefactthattherelationshipbetweenmaritalstatusandOPDutilizationisweakenedwhenadjustedforthesefactors,suggestingthattheobservedrelationshipisinpartmediatedthroughthesemechanisms. Ourstudyhasimportantstrengths.TheMCBSisanationally-representativesurveyadministeredtoalargesampleofMedicarebeneficiariesacrosstheUnitedStates.Thedepthofthedatasetallowedustoadjustforawiderangeofpotentialconfoundingfactors.Ouroutcomemeasuresalsorepresentthebroaddomainsofhealthcareprovision,conferringamoreglobalperspectiveonthehypothesizedfindings.Themodelsweconstructedwerebasedonawell-validatedconceptualmodel,clinicalconsiderations,andpotentialpolicyrelevance.Asweindicatedabove,forcasesinwhichpriorliteratureexists,ourfindingsareconsistentwiththoseofotherinvestigatorsusingdifferentdatasources.Finally,ouranalysisproducedrobustestimateswithpotentiallyimportantimplications.Thepositiveassociationweobservedbetweenworseself-ratedhealthstatusandreceiptofaninfluenzavaccination(representativeofpropensitytoseekmedicalattention)withutilizationacrossallthreesettingsofcaresupportsthevalidityoftheoutcomemeasureswechose. Themajorlimitationofthisstudyisthatthisisacross-sectionalanalysisofsecondarydatathatassessedanassociationbutnotacausaleffect.Althoughweestablishedastrongassociationbetweenmaritalstatusandtheutilizationmeasures,itispossiblethatunobservedclinicalcharacteristicsorsocialdifferencesthatcorrelatewithmarriageactuallymediateourfindings.However,webelievethatthedepthofthedataallowingustoperformamultivariateanalysisamonganationallyrepresentativesamplerendersourfindingsvaluableforplausiblecausalinference.Anotherlimitationimportanttoacknowledgeisourinabilitytoderivearelationshipbetweenmaritalstatusandhealthoutcomes.WewerealsonotabletodelineatethediagnosesorclinicalcircumstancesassociatedwithinpatientandSNFuse.Ifthereducedutilizationamongthemarriedwasforambulatorycaresensitiveconditions,thiswouldbeofparticularimportanceforhealthsystemstructureandqualitymeasurement.Incorporatingsuchdetailedclinicalinformationwouldbeanimportantdirectionforfutureresearchinthisarea.Althoughwehavecontrolledformeasuresofperceivedneedforhealthcare,wewerenotabletorigorouslycontrolforcomorbidconditionsusingvalidatedmethodssuchastheCharlsonorElixhausermethodsforriskadjustment.[52,53]However,asAndersenpointsoutinthe1995reassessmentofhismodel,perceivedneeds(asmaybereflectedinself-ratedhealthstatus)maybeabettermeasureofcare-seekingbehaviorcomparedtoobjectivemeasuresofhealthcareneed(e.g.comorbidityburden).[29]Sincewedidnothaveaccesstorigorousmeasuresofrespondents’propensitytoseekmedicalattention,weusedthereceiptofaninfluenzavaccinationasaproxymeasure.Despiteitsfacevalidityassuch,wecannotstatewithcertaintythefidelityorrobustnesswithwhichitascertainsanindividual'stendencytopursuehealthcareservices. Futureresearchmightexaminewhetherincreasingoutpatientserviceconsumptionamongunmarriedindividualsreducestheirhospitalizationratesand/orSNFuse,andanyquantifiableeffectontheiroutcomesorthecostsincurred.Thiswouldlendcredencetothesubstitutioneffectdescribedabove.The‘‘spousaleffect”maybemorethansimplypsychological,thuscreatingdiscerniblepatternsofbehaviorinwhichspousesengageincaregivingandhealthcaresystemnavigation,yieldingmeasurablehealthbenefits.Rigorousknowledgeinthisregardcouldfacilitatethedesignoftargetedhealthcareinterventionsandhomehealthservicesforunmarriedpeople.Suchinterventionsmayrealizebenefitsbysimultaneouslyimprovingobjectivehealthoutcomesandpatient-centeredness,whilereducingcosts. Theimpactofsocialcircumstancesonhealthcareutilizationpatternsisnotwellunderstood,butwebelievethatourfindingsaidinelucidatingtherelationshipbetweenmaritalstatusandutilizationofhealthcare.ThisstudyprovidesimportantcorroborativeevidencetosupportsucharelationshipamongMedicarebeneficiaries.Therefore,weconcludethatpatternsofhealthcareusemayexplainatleastsomeofthevariationsinoutcomesandcostsobservedacrosssocialgroups.Longitudinal,prospectivelyspecifiedstudiesmightdeterminewithgreaterconfidencetheextenttowhichatruecausaleffectunderliesthisrelationship.Althoughwehaveexploredthemechanismsthatmayunderlieourfindings,futureinvestigationsmaybedirectedatmoredefinitivelyelucidatingthefactorsthatmediatetheobservedrelationshipsbetweenmaritalstatusandutilization.Agreaterunderstandingofthesefactorsmayhaveimportantimplicationsforclinicalcare,socialpolicy,andimprovementsinhealthsystemdesign,whichmayholdthepotentialtooptimizehealthandmitigatetheeconomicchallengesofhealthcareprovision. Authorcontributions Conceptualization:KiranRajPandey,GregoryW.Ruhnke. Datacuration:KiranRajPandey. Formalanalysis:KiranRajPandey. Methodology:KiranRajPandey,FanYang,KathleenA.Cagney,FabriceSmieliauskas,DavidO.Meltzer,GregoryW.Ruhnke. Software:KiranRajPandey. Supervision:FanYang,KathleenA.Cagney,FabriceSmieliauskas,DavidO.Meltzer,GregoryW.Ruhnke. Writing–originaldraft:KiranRajPandey. Writing–review&editing:KiranRajPandey,FanYang,KathleenA.Cagney,FabriceSmieliauskas,DavidO.Meltzer,GregoryW.Ruhnke. KiranRajPandeyorcid:0000-0002-8277-766X.References [1].BanerjeeS.DifferencesinOut-of-PocketHealthCareExpensesofOlderSingleandCoupleHouseholds.2016. CitedHere [2].BanerjeeS.Utilizationpatternsandout-of-pocketexpensesfordifferenthealthcareservicesamongamericanretirees.EBRIIssueBrief2015;411:1–20. CitedHere [3].AndersenR,NewmanJF.SocietalandindividualdeterminantsofmedicalcareutilizationintheUnitedStates.MilbankQuarterly2005;83:Online-only-Online-only.doi:10.1111/j.1468-0009.2005.00428.x. CitedHere [4].DunlopDD,ManheimLM,SongJ,etal.Genderandethnic/racialdisparitiesinhealthcareutilizationamongolderadults.JGerontolBPsycholSciSocSci2002;57:S221–33. CitedHere [5].deBoerAGE,WijkerW,deHaesHCJ.Predictorsofhealthcareutilizationinthechronicallyill:areviewoftheliterature.HealthPolicy1997;42:101–15. CitedHere [6].WaiteLJ.Doesmarriagematter?Demography1995;32:483–507. CitedHere [7].LillardLA,WaiteLJ.Tildeathdouspart:maritaldisruptionandmortality.AmJSociol1995;100:1131–56. CitedHere [8].VerbruggeLM.Maritalstatusandhealth.JMarriageFam1979;41:267doi:10.2307/351696. CitedHere [9].MendesdeLeonCF,AppelsAW,OttenFW,etal.Riskofmortalityandcoronaryheartdiseasebymaritalstatusinmiddle-agedmeninTheNetherlands.IntJEpidemiol1992;21:460–6. CitedHere [10].CoombsRH.Maritalstatusandpersonalwell-being:aliteraturereview.FamRelat1991;40:97doi:10.2307/585665. CitedHere [11].HuY,GoldmanN.Mortalitydifferentialsbymaritalstatus:aninternationalcomparison.Demography1990;27:233–50. CitedHere [12].SimonRW.Revisitingtherelationshipsamonggender,maritalstatus,andmentalhealth.AmJSociol2002;107:1065–96. CitedHere [13].StackS,EshlemanJR.Maritalstatusandhappiness:a17-NationStudy.JMarriageFam1998;60:527doi:10.2307/353867. CitedHere [14].GoldmanN.Marriageselectionandmortalitypatterns:inferencesandfallacies.Demography1993;30:189–208. CitedHere [15].CobbS.Socialsupportasamoderatoroflifestress.PsychosomMed1976;38:300–14. CitedHere [16].EspinosaJ,EvansWN.Heightenedmortalityafterthedeathofaspouse:marriageprotectionormarriageselection?JHealthEcon2008;27:1326–42. CitedHere [17].ArnoPS,LevineC,MemmottMM.Theeconomicvalueofinformalcaregiving.HealthAff(Millwood)1999;18:182–8. CitedHere [18].GrossmanM.Ontheconceptofhealthcapitalandthedemandforhealth.JPolitEcon1972;80:223–55. CitedHere [19].GrossmanM.TheHumanCapitalModeloftheDemandforHealth.Cambridge,MA:NationalBureauofEconomicResearch;1999. CitedHere [20].WaldronI,HughesME,BrooksTL.Marriageprotectionandmarriageselection--prospectiveevidenceforreciprocaleffectsofmaritalstatusandhealth.SocSciMed1996;43:113–23. CitedHere [21].RizzoJA,FriedkinR,WilliamsCS,etal.HealthcareutilizationandcostsinaMedicarepopulationbyfallstatus.MedCare1998;36:1174–88. CitedHere [22].LuberMP,MeyersBS,Williams-RussoPG,etal.Depressionandserviceutilizationinelderlyprimarycarepatients.AmJGeriatrPsychiatry2001;9:169–76. CitedHere [23].HentonFE,HaysBJ,WalkerSN,etal.DeterminantsofMedicarehomehealthcareserviceuseamongMedicarerecipients.NursRes2002;51:355–62. CitedHere [24].YangM,BarnerJC.Useoftheandersenhealthcareservicesutilizationbehavioralmodeltounderstandtherelationshipbetweenhealthinsurancecoverageandhealthcareservicesutilizationamongtheelderly.ValueHealth2001;4:172doi:10.1046/j.1524-4733.2001.40202-271.x. CitedHere [25].JoungIM,vanderMeerJB,MackenbachJP.Maritalstatusandhealthcareutilization.IntJEpidemiol1995;24:569–75. CitedHere [26].USCensusBureau.HistoricalMaritalStatusTables.2016.Availableat:https://www.census.gov/data/tables/time-series/demo/families/marital.html.[accesseddateAugust9,2017]. CitedHere [27].PewResearchCenter.RecordShareofAmericansHaveNeverMarried.PewResearchCenterSocialandDemographicTrends2014.Availableat:http://www.pewsocialtrends.org/2014/09/24/record-share-of-americans-have-never-married/[accesseddateMay8,2017]. CitedHere [28].USCensusBureau.Theriseoflivingalone.2016.Availableat:https://www.census.gov/data/tables/time-series/demo/families/households.html.[accesseddateAugust11,2017]. CitedHere [29].AndersenRM.Revisitingthebehavioralmodelandaccesstomedicalcare:doesitmatter?JHealthSocBehav1995;36:1–0. CitedHere [30].AndersenR,NewmanJF.SocietalandindividualdeterminantsofmedicalcareutilizationintheUnitedStates.MilbankMemFundQHealthSoc1973;51:95–124. CitedHere [31].AdayLA,AndersenR.Aframeworkforthestudyofaccesstomedicalcare.HealthServRes1974;9:208–20. CitedHere [32].ButzA,HuttonN,JoynerM,etal.HIV-infectedwomenandinfants:Socialandhealthfactorsimpedingutilizationofhealthcare.JNurseMidwifery1993;38:103–9. CitedHere [33].AdlerGS.AprofileoftheMedicareCurrentBeneficiarySurvey.HealthCareFinancRev1994;15:153–63. CitedHere [34].BriesacherBA,TjiaJ,DoubeniCA,etal.MethodologicalissuesinusingmultipleyearsoftheMedicarecurrentbeneficiarysurvey.MedicareMedicaidResRev2012;2:doi:10.5600/mmrr.002.01.a04. CitedHere [35].CMS.MedicareCurrentBeneficiarySurvey2013AccesstoCare,PublicUseFileUserGuide.2013. CitedHere [36].LundR,HolsteinBE,OslerM.Maritalhistoryfromage15to40yearsandsubsequent10-yearmortality:alongitudinalstudyofDanishmalesbornin1953.IntJEpidemiol2004;33:389–97. CitedHere [37].GoldsmithJC.Thehealthcaremarket:canhospitalssurvive?HarvBusRev1980;58:100–12. CitedHere [38].StoeckleJD.TheCitadelcannothold:technologiesgooutsidethehospital,patientsanddoctorstoo.MilbankQ1995;73:3–17. CitedHere [39].GillJM,MainousAG.Theroleofprovidercontinuityinpreventinghospitalizations.ArchFamMed1998;7:352–7. CitedHere [40].MainousAG,GillJM.Theimportanceofcontinuityofcareinthelikelihoodoffuturehospitalization:issiteofcareequivalenttoaprimaryclinician?AmJPublicHealth1998;88:1539–41. CitedHere [41].WeissmanJS.Ratesofavoidablehospitalizationbyinsurancestatusinmassachusettsandmaryland.JAMA1992;268:2388doi:10.1001/jama.1992.03490170060026. CitedHere [42].BindmanAB,GrumbachK,OsmondD,etal.Preventablehospitalizationsandaccesstohealthcare.JAMA1995;274:305–11. CitedHere [43].SilverMP,BabitzME,MagillMK.AmbulatorycaresensitivehospitalizationratesintheagedMedicarepopulationinUtah,1990∗to1994∗:arural-urbancomparison.JRuralHealth1997;13:285–94. CitedHere [44].TianY,DixonA,GaoH.Emergencyhospitaladmissionsforambulatorycare-sensitiveconditions:identifyingthepotentialforreductions.TheKing'sFundDataBriefing2014. CitedHere [45].GalarragaJE,MutterR,PinesJM.Costsassociatedwithambulatorycaresensitiveconditionsacrosshospital-basedsettings.AcadEmergMed2015;22:172–81. CitedHere [46].ShahMN,RathouzPJ,ChinMH.Emergencydepartmentutilizationbynoninstitutionalizedelders.AcadEmergMed2001;8:267–73. CitedHere [47].FriedTR,BradleyEH,WilliamsCS,etal.Functionaldisabilityandhealthcareexpendituresforolderpersons.ArchInternMed2001;161:2602–7. CitedHere [48].ChinMH,ZhangJX,MerrellK.DiabetesintheAfrican-AmericanMedicarepopulation.Morbidity,qualityofcare,andresourceutilization.DiabetesCare1998;21:1090–5. CitedHere [49].SimeonovaE.Marriage,bereavementandmortality:theroleofhealthcareutilization.JHealthEcon2013;32:33–50. CitedHere [50].CastevensP.Theimpactoffamilystructureonsparecapacityandhealthcareutilization.HealthWatch2014.31–8. CitedHere [51].GoldmanN,KorenmanS,WeinsteinR.Maritalstatusandhealthamongtheelderly.SocSciMed1995;40:1717–30. CitedHere [52].CharlsonM,SzatrowskiTP,PetersonJ,etal.Validationofacombinedcomorbidityindex.JClinEpidemiol1994;47:1245–51. CitedHere [53].ElixhauserA,SteinerC,HarrisDR,etal.Comorbiditymeasuresforusewithadministrativedata.MedCare1998;36:8–27. CitedHere Keywords:healthcareutilization;inpatient;marriage;medicare;outpatient;skillednursingfacility SupplementalDigitalContent MD_2019_03_07_PANDEY_MD-D-18-03821_SDC1.docx;[Word](95KB) Copyright©2019theAuthor(s).PublishedbyWoltersKluwerHealth,Inc.Viewfullarticletext Source TheimpactofmaritalstatusonhealthcareutilizationamongMedicarebeneficiaries Medicine98(12):e14871,March2019. Full-Size Email +Favorites Export ViewinGallery Colleague'sE-mailisInvalid YourName: Colleague'sEmail: Separatemultiplee-mailswitha(;). Message: Thoughtyoumightappreciatethisitem(s)IsawinMedicine. Yourmessagehasbeensuccessfullysenttoyourcolleague. Someerrorhasoccurredwhileprocessingyourrequest.Pleasetryaftersometime. FollowMedicine®onSocialMedia! ArticleasEPUB ExportAllImagestoPowerPointFile AddtoMyFavorites Colleague'sE-mailisInvalid YourName: Colleague'sEmail: Separatemultiplee-mailswitha(;). Message: Thoughtyoumightappreciatethisitem(s)IsawinMedicine. Yourmessagehasbeensuccessfullysenttoyourcolleague. Someerrorhasoccurredwhileprocessingyourrequest.Pleasetryaftersometime. EndNoteProciteReferenceManager Savemyselection KeywordHighlighting Highlightselectedkeywordsinthearticletext. healthcareutilization inpatient marriage medicare outpatient skillednursingfacility SearchforSimilarArticles Youmaysearchforsimilararticlesthatcontainthesesamekeywordsoryoumay modifythekeywordlisttoaugmentyoursearch. healthcareutilization\r,inpatient\r,marriage\r,medicare\r,outpatient\r,skillednursingfacility ArticlesinPubMedbyKiranRajPandey,MD,MS ArticlesinGoogleScholarbyKiranRajPandey,MD,MS OtherarticlesinthisjournalbyKiranRajPandey,MD,MS Dataistemporarilyunavailable.Pleasetryagainsoon. Work-relatedstress,burnout,andrelatedsociodemographicfactorsamongnurses:Implicationsforadministrators,research,andpolicy AstressmanagementinterventionforadultslivingwithHIVinNigeriancommunitysettings:Aneffectsstudy EffectofrationalemotivebehaviortherapyonstressmanagementandirrationalbeliefsofspecialeducationteachersinNigerianelementaryschools ThestudyonthecorepersonalitytraitwordsofChinesemedicaluniversitystudentsbasedonsocialnetworkanalysis Participationincancerscreeningamongfemalemigrantsandnon-migrantsinGermany:Across-sectionalstudyontheroleofdemographicandsocioeconomicfactors Thiswebsiteusescookies.Bycontinuingtousethiswebsiteyouaregivingconsenttocookiesbeingused.ForinformationoncookiesandhowyoucandisablethemvisitourPrivacyandCookiePolicy. Gotit,thanks!