DoeslongtermcareprovisionreducehospitalbedblockingEvidencefromapoli - PDF document

Doeslong-termcareprovisionreducehospitalbed-blocking?EvidencefromapolicyreforminPortugal(Preliminaryandincomplete,pleasedonotcirculateorcitewithoutpermission)AnaMouraAbstractExcessivelengthsofhospitalstayareamongtheleadingsourcesofineciencyinhealthcare.Onereasonforexcessivelengthsofhospitalstaycanbebed-blocking.Istudywhetherlong-termcare(LTC)provisionreduceshospitalbed-blocking.Usingindividualdata

onemergencyinpatientadmissionsatPortuguesehospitalsduring2000-2015,Iimplementatriple-dierencesdesign.ThisdesignexploitsvariationinthetimingofentryofLTCprovidersacrossregionsoriginatingfromthestaggeredintroductionofthepublicLTCNetwork.Italsoexploitsvariationinlengthsofstaybetweenregularpatientsandpatientsexhibitingsocialfactorsthatputthematriskofbed-blocking,suchaslivingalone,havingnofamilytocare,andha

vinginadequatehousing.Indthattheentryofthersthome-careteaminaregionreducesthelengthofstayofindividualslivingaloneandthosewithinadequatehousingby4daysrelativetoregularpatients.Thesereductionsinlengthofstaydonotaectthetreatmentreceivedwhileatthehospital.Reductionsinlengthofstayupontheentryoftherstnursinghomeoccuronlyforpatientswithhighcareneeds.Thebedsfreedupbybed-blockersareusedtoadmitmo

reelectivepatients.Keywords:Long-termcare;hospitalbed-blocking;delayeddischarges.JELcodes:H51;I10;I18;J14. TilburgUniversity,DepartmentofEconometricsandOperationsResearch.Address:TilburgUniversity,P.O.Box90153,5000LETilburg,TheNetherlands.E-mail:a.c.moura@tilburguniversity.edu.IamthankfultoJanBooneandMartinSalmforextensiveadviceandsupport.Accesstotheindividualhospitaldataispossibleunderthedata-sharinga

greementbetweenAdministrac~aoCentraldoSistemadeSaude,I.P.(ACSS)andtheHealthEconomics&ManagementKnowledgeCenteratNovaSchoolofBusinessandEconomics,withwhichIamaliated.IthankHugoLopesforprovidingmethedataontheroll-outofthepubliclong-termcarenetwork.1 1IntroductionWhilethereisageneralconsensusamongeconomistsandpolicy-makersthathealthcaresystemsarewastefulandinecient(GarberandSkinner,2008),p

ointingoutspecicsourcesofineciencycanbechallenging(Einavetal.,2019).TheWorldHealthOrganizationconsidersexcessivelengthsofhospitalstayasoneoftheleadingsourcesofineciencyinhealthcare(WHO,2010).Onereasonforexcessivelengthsofhospitalstaycanbebed-blocking.Bed-blockingoccurswhenapatientisclinicallyttobedischargedbutrequiressomeformofsupportoutsidethehospital,suchasashortstayatanursinghomefaci

lityorhome-help,whichisnotreadilyavailable.Ifnosafedischargearrangementscanbemade,thepatientremainsinthehospitaluntilasafetransitiontothenextstageofcareprovisionispossible,resultinginlongerlengthsofhospitalstay.Thesearenotinconsequential.Theyimplyhigherhospitalcosts,havepotentiallydetrimentalimpactsonpatients'healthoriginatingfromincreasedrisksofmobilityloss,nosocomialinfections,andloneliness,andcancreated

elaysforpatientsawaitingelectivecare(Mur-VeemanandGovers,2011).Bed-blockingisagrowingpolicyconcernindevelopedcountries.Thisismotivatedbyseveraltrends.Duringthelastdecades,thereweresignicantincreasesinlifeexpectancyandconsequentlyrisingshareoftheelderlyinthepopulation.Moreover,chronicdiseasesbecametheleadingcauseofillness,disability,anddeath.Whilelargelymanageableoutsidethehospital,chronicdiseaseslimit

patients'abilitytoliveindependently.Thesedemographictrendsputpressureonexistinginstitutionalarrangementswithinthehealthsystemandcallforareorganizationofcaredelivery(Harper,2014).Socialtrends,suchastheriseinfemalelaborforceparticipationandthedeclineofinter-generationalhouseholds,inturn,threatenexistinginformallong-termcarearrangements(LakdawallaandPhilipson,2002).Inthisstudy,Iexaminewhethertheprovisionoflon

g-termcare(LTC)reduceshospitalbed-blocking.IfocusonthePortuguesecase.InPortugal,onarandomdayin2019,4.7%ofbedsinpublichospitalswereoccupiedwithpatientswhowereclinicallyreadytobedischargedbutlackedlong-termcaresupport,amountingtoover80,000delayedbed-days.1,2Iexploitplausiblyexogenousvariationintheavailabilityoflong-termcareoriginatingfromtheintroductionofthepublicLTCNetworkbythePortuguesegovernment.Before200

6,LTCserviceswerenotwithinthescopeofthePortugueseNationalHealthSystemandindividuals 1Resultsfromasnapshot-censuscarriedoutbythePortugueseAssociationofHospitalManagers(APAH).Seehttps://apah.pt/portfolio/barometro-de-internamentos-sociais/.2Figuresconveyingthemagnitudeofbed-blockingindierentcountriesarenoteasilyavailable.InSweden,theshareofbed-blockerswasabout7%in1992(StyrbornandThorslund,1993).In2006,6

.1%ofallhospital-daysintheNetherlandswerebed-blockingdays(Mur-VeemanandGovers,2011).InEngland,duringDecember2019alone,thenumberofdelayedbed-daysreached148,000,a15%increasecomparingtothesamemonthofthepreviousyear(https://www.theguardian.com/society/2020/feb/23/bed-blocking-highest-since-2017-hospitals-nhs).2 reliedheavilyoninformalcareprovidedbyfamilymembers.ThepublicLTCNetworkwasintroducedin2006,withtheaim

ofllinginthisgapinservicecoverage.Itcompriseshighlysubsidizednursinghomefacilities(NH)andteamsprovidinghome-care(HC),whichoperateincoordinationwithhospitals,aimingateasingthetransitionofpatientsacrossdierentsettingsofcareprovision.TheLTCNetworkwasintroducedinastaggeredfashion,meaningthatdierentregionsexperiencedtheentryofLTCprovidersatdierentpointsintime.Usingindividualdataontheuniverse

ofemergencyinpatientadmissionsatpublichospitalsinPortugalbetweentheyears2000and2015,Irstshowthatthereisagroupofpatientsexhibitingacomplexcombinationofhealthandsocialneeds.Thisgroupincludesindividualswholivealone,havenofamilytocareforthem,andhaveinadequatehousingconditionsorotherunfortunateeconomiccircumstancesthatmighthinderatimelydischarge.Thedierencesinlengthofstaybetweenthesepatientsandpatient

swhodonotexhibitsocialneedsaresizable.Forexample,individualswithnofamilytocarehavelengthsofstaythatare,onaverage,over20dayslongerthanregularpatientsaftercontrollingfordemographics,comorbidities,andmedicaldiagnoses.Throughoutthepaper,Irefertopatientswhoexhibitbothhealthandsocialneedsasbed-blockers,asopposedtoregularpatients,whoexhibitnosocialneeds.Intheempiricalanalysis,Iuseatriple-dierencesdesigntocom

parethelengthofstayofbed-blockersandthelengthofstayofregularpatientsbeforeandaftertheentryoftherstNHandtherstHCteaminaregion.ThisidenticationstrategyexploitsvariationinLTCavailabilityacrossregionsandtimeoriginatingfromthestaggeredintroductionoftheLTCNetwork.Italsoexploitsvariationinlengthsofstaybetweenregularpatientsandbed-blockers.Indevidenceofreductionsinthelengthofstayofbed-blockersr

elativetoregularpatientsfollowingtheentryofthersthome-careteaminaregion.Thesereductionsamountto4days,onaverage,forindividualslivingaloneandthosewithinadequatehousingandothereconomicissues.Indnosignicantreductionsinthelengthofstayofbed-blockersrelativetoregularpatientsfollowingtheentryoftherstnursinghomeinaregion.Sinceadmissiontoanursinghomerequireshigherlevelsofdisabilityanddependency,t

heaveragebed-blockermightnotbeatriskofbeingadmittedtoanursinghome.Consistentwiththishypothesis,Indreductionsinthelengthofstayofbed-blockersrelativetoregularpatientsfollowingtheentryoftherstnursinghomeinaregiononlywhenrestrictingthesampletopatientswithhighcareneeds,suchasthosewithastrokediagnosis.Finally,theentryoftherstnursinghomeandhome-careproviderinaregionhasaprecisezeroimpactonthelengtho

fstayofregularpatients,meaningthatreductionsinthelengthsofstayofbed-blockersrelativetoregularpatientsoriginateonlyfromreductionsinthelengthofstayofbed-blockers.Theseresultsarerobusttoalternativemodelspecications,outcomevariables,relevant3 regions,andtreatmentdenitions.Anevent-studydesignconveysthattheresultsarenottypicallydrivenbydierencesinunderlyingpre-treatmenttrends.Ialsoshowthattheentry

ofLTCprovidersinaregionisnotassociatedwithchangesinthecodingfrequencyofthesocialfactorsusedtoidentifybed-blockers.Consistentwiththelongerlengthofstayofbed-blockersbeingunnecessary,Indnoreductionsintheintensityofcarereceivedbybed-blockersduringtheirhospitalstayaftertheentryofLTCprovidersinaregion.Finally,thebedsfreedupbybed-blockersdonotremainunoccupied:Indevidenceofincreasesinprogrammedactivityfo

llowingtheentryofthersthome-careteaminaregion.WhiletheseresultsshowthatLTCprovisioncanreducehospitalbed-blocking,theresultingreductionsinlengthofstaytakesometimetomaterializeanddonotfullyclosethegapbetweenbed-blockersandregularpatients.Iinvestigatetheroleoftheaccumulationofrelationship-specicexperiencebetweenahospital-regionpairinexplainingtheseresults.Thendingsfromthisanalysisconveythatalar

genumberofinteractionsbetweenahospital-regionpairisneededtogeneratemeaningfulreductionsinthelengthofstayofbed-blockers.RelatedLiterature.Thispaperrelatestoseveralstrandsoftheliterature.3ItspeaksmostdirectlytoasmallbutgrowingliteraturestudyingtheimpactsofLTCavailabilityonhospitalbed-blocking(Forder,2009;Gaughanetal.,2015,2017a,b).Unlikeexistingstudies,IexploitvariationintheavailabilityofLTCinducedbyapolicyr

eformandassessitsimpactsonbed-blocking.Duetothenatureofthereform,IamabletoseparatetheeectsofdistincttypesofLTCproviders,namelynursinghomesandhome-careteams.Additionally,Iproposetheuseofinformationonsocialneedstoidentifypatientsatincreasedriskofbed-blockingandcircumventthefactthattheexactlengthofthedelayistypicallyunobserved(withHolmasetal.(2013)anotableexception).Thesesocialneedsincludetheavailab

ilityoffamilysupportandhousingconditions,whichhavebeenshowntoaectbothLTCuse(Diepstratenetal.,2020;Lopesetal.,2018)andhospitalbed-blocking(Costaetal.,2012;Bryanetal.,2006;McDonaghetal.,2000).ArelatedliteraturefocusesonthesubstitutabilityofLTCandacutehospitalcare.ThisliteraturendslittletonosubstitutionbetweenLTCandhospitallengthofstay(McKnight,2006;Forder,2009;Gaughanetal.,2015,2017a;Bakxetal.,2020

;Costa-Fontetal.,2018;Kumpel,2019).IdeviatefromexistingstudiesinthatIassessthesubstitutabilityofLTCandacutecareusingindividualdataonpatientswhoarehospitalized.Givenmyresearchdesign,Ianalyzesubstitutionofacutecarewithrespecttobothhome-careandnursinghome 3Thereisalsoanextensiveliteratureonbed-blockingoutsideeconomics.Scholarsinmedicineandhealthpolicyhaveelaboratedonthecausesofbed-blocking,characterizedthea&

#11;ectedpopulation,andquantiedtheassociatedmonetarylosses(see,forexample,Bryanetal.,2006;Hendyetal.,2012;Costaetal.,2012).Withinoperationsresearchandhealthcaremanagement,theoptimizationofpatient owshasbeenwellstudied(McCleanandP.,2006;El-Darzietal.,1998;Katsaliakietal.,2005;OsorioandBierlaire,2007andmanyothers).4 care,separatelyforregularpatientsandbed-blockers.Indnoevidenceofsubstitutionbetween

LTCandacutecareforregularpatients.Forbed-blockers,Iestimatetheelasticityofdelayedlengthofstaywithrespecttoboththesupplyofnursinghomebedsandofplacesinhome-careteamsbetween-0.2and-0.3.Myndingthatreductionsinbed-blockingleadtoincreasesinprogrammedactivityrelatestoadiscussionontheinternalallocationofresourceswithinahospital,whichgoesbacktoHarris(1977).Thespeciccaseofinteractionsbetweenemergencyandele

ctiveactivitieswasrecentlystudiedbyJoharetal.(2013).Thispaperalsoprovidesnewinsightsonthefactorscontributingtoperpetuatebed-blocking.FernandezandForder(2008)studytheimportanceofnancialresourcesallocatedtotheLTCsector.Holmasetal.(2010)showthatmonetaryincentivestoreducebed-blockingcancrowd-outagents'intrinsicmotivationandendupbeingcounterproductive.BecausetheacuteandLTCsettingsareorganizedandfunded

separatelyinmanycountries(Siciliani,2014),coordinationdicultiesacrossthetwosettingsarelikelytooccur(Cebuletal.,2008).Theroleofcoordinationdicultiesinperpetuatingbed-blockinghasbeenstudiedbyFernandezetal.(2018).Drawingon(Kellogg,2011),Iproposeanalternative,albeitcomplementarymechanismbasedontheaccumulationofrelationship-specicexperiencebetweenpairsofhospitalsandteamsresponsibleforndingva

canciesintheLTCNetwork,whicharebasedinthepatient'sregionofresidence.Myndingssuggestthatonlythepairswiththelargestnumberofinteractionsareabletoaccumulatealevelofexperiencethatallowsformeaningfulreductionsinthelengthofstayofbed-blockers.Thiscanexplainwhypreviousliteraturehasfoundthatlargerhospitals,withahighnumberofadmissions,seemtomanagedischargesmoreecientlyandthushavelessdelayeddischarges(DeVold

eretal.,2020).Finally,andmorebroadly,thispaperrelatestorecentworkzoominginonspecicaspectsofthehealthcaresectortoidentifysourcesofwasteandineciency.AlargepartofthisliteraturehasfocusedoninteractionsbetweentheacutecareandLTCsettings(DoyleJretal.,2017;Einavetal.,2018;Eliasonetal.,2018;Jinetal.,2018;Einavetal.,2019;Kumpel,2019,andmanyothers).Byandlarge,thisliteraturepointstotheLTCandpost-acutesector

sasasourceofineciencyinthehealthcaresystem.Mypapertakesadiametricallyopposedstand,suggestingthatLTCprovisionmightbeawaytoreduceinecienciesassociatedwithbed-blockingintheacute-caresetting.Inparticular,mybaselineestimatessuggestthatLTCprovisiongeneratesa30%reductioninannualbed-blockingcostsincurredbyhospitals.Theremainderofthispaperisorganizedasfollows.Section2providesinstitutionalbackgroundonthePo

rtuguesehospitalandLTCsettings.Section3describesthedatausedintheanalysis.Section4describesmyempiricalapproach.Section5presentstheresultsandSection6elaboratesonpotentialmechanisms.Finally,section7concludes.5 2InstitutionalSetting2.1InpatientcareInPortugal,mostinpatientcareisprovidedbypublichospitalsbelongingtotheNationalHealthSystem(SNS).TheSNSispredominantlynancedthroughgeneraltaxationandcharacterized

byuniversalcoverageandaccesstocarethatismostlyfreeatthepointofuse(Sim~oesetal.,2017).InpatientcareprovidedbyhospitalsbelongingtotheSNSispaidviaDiagnosis-RelatedGroups(DRGs).ADRGgroupspatientswhohavesimilarconsumptionofresourcesbasedontheirmedicaldiagnosis,treatmentreceived,anddemographicprole.Thereareover600distinctgroupsinthecurrentDRGsystemandeachhasanassociatedprice.DRGsareusedtosetanannualprospect

iveglobalbudgetforinpatientcareprovidedbyeachhospital,whichamountsto75-85%oftotalinpatientrevenuesofSNShospitals(Mateus,2011).Overall,hospitalshavenonancialincentivestokeeppatientsforlongerthannecessary.SincehospitalsarepaidaccordingtothenumberandtheDRGofpatientstheyreceive,DRG-basedfundingprovidesincentivesforhospitalstotreatmorepatientsandtocutcosts,possiblybyreducinglengthsofstay.Toaccountforcompli

catedpatientswhoselengthofstaymightbeextraordinarilylong,hospitalsgetanadditionaldailypaymentforeachdayinexcessofanuppertrim-pointdenedbylawforthepatient'sDRGuntildischarge.Whilethetrim-pointisDRG-specic,thedailyamountforthedaysinexcessofthetrimpointisnot.Trim-pointsanddailyamountswereupdatedbythegovernmentatseveralpointsduringmystudy-period.Formostofthestudy-period,thedailyamountwasbetweene80and

e90.2.2Long-termcareLong-termcareiscareneededbyindividualswithsomedegreeofdependency.Itincludeshealthcare(i.e.rehabilitationafteranacutecareepisode),personalcare(i.e.personalhygiene),helpwithactivitiesofdailyliving(i.e.housework,meals),andaccommodationforindividualswhocannotliveindependently(Siciliani,2014;Norton,2000).Itcanbeprovidedeitherformallyorinformally.Formallong-termcareisprovidedbytrainedpersonne

latthepatient'shomeoratinstitutionssuchasnursinghomes.Informalcareisprovidedbyrelatives,friends,orneighbors.Before2006,formalLTCserviceswerenotwithinthescopeoftheSNSandindividualsreliedheavilyoninformalcare.WhileLTCservicescouldbepurchasedfromprivateproviders,namelynon-protreligiousinstitutions(Misericordias)(Sim~oesetal.,2017),theircostshadtobepaidforoutofpocket,whichtookanancialtollonmanyu

sers(Santana,2010).6 Table1:OverviewoftheorganizationbytheLTCNetwork Nursinghome(NH)Home-care(HC) Startofroll-out20062008ProvidersPrivatePublicFundingPublicPublicSet-upGovernmentcontractswithex-istingprovidersTeamscreatedinprimarycarecentersPriceHighlysubsidized(means-tested)co-paymentsFreeServices24-hourmedicalcare,rehabil-itation,food,hygiene,accom-modation,etc.Preventivecare,helpwithADLs,food,hygiene,me

dica-tion,etc. In2006thePortuguesegovernmentestablishedtheNationalNetworkforLong-TermIntegratedCare(RNCCI,henceforththeNetwork),asajointeortoftheMinistryofHealthandtheMinistryofLaborandSocialSecurity(Decree-Law101/2006).Thiswasmotivatedbyconcernsregardingdemographic,social,andepidemiologicaltrendsthatputpressureonexistingcarearrangements,includingincreasinglifeexpectancyandconsequentrisingshareofelder

lyinthepopulation,thedeclineofinter-generationalhouseholds,andtheincreasingshareofindividualslivingwithmultiplecomorbidities.Importantly,theNetworkwasnotaimedofreducingbed-blocking,whichisarecenttopicinthepublicdebate.TheNetworkcomprisestwodistinctsettingsofcareprovision:homeandcommunity-basedservices(HC)andnursinghomes(NH).Table1providesanoverviewofthesetwosettings.Theyareorganizedverydierently.TheNH

settingoperatesinamodelofpublicfundingandprivateprovisioninwhichthegovernmentcontractswithexistingproviders,namelytheMisericordias,whowerealreadyactiveincareprovisionforseveraldecades.4Theservicescontractedincludearound-the-clockmedicalcare,rehabilitation,accommodation,catering,personalhygiene,etc.TherearedierenttypesofNHfacilitiesthatcatertopatientswithdierentcareneeds.Sometargetindividuals

whonolongerneedacutehospitalcarebutstillrequireintensivemedical,nursing,andrehabilitationcareforarelativelyshortperiodoftime|thesecanbethoughtofasprovidingpost-acutecare.OtherNHfacilitieshavelessintensivemedical,nursing,andrehabilitationcomponents,mainlycateringtoindividuals 4Morerecentlythegovernmentstartedcontractswithprivate,for-protprovidersandalsoestablishedsomepublic-ownedfacilities.Theseamounte

dto16%and2%ofNHproviderscontractedasof2015,theendofmystudy-period.7 withchronicillnessesandhighfunctionaldependency,whomightneedlongerstays.UndertheNHcontractsthegovernmentpaysprovidersanadministrativelysetdailypriceforthecareprovidedtoindividualswhoareintheNetwork.In2015,dailypricesweresetbetweene60ande105,dependingontheintensityofcareprovidedbytheNHfacility.Thedailypriceiseitherfullypaidorhighly-subsidiz

edbythegovernment.TheHCsettingoperatesinamodelofpublicprovisionandpublicfunding.Thegovernmentestablishedspecializedteamsinprimarycarecentersthatvisitpatientsintheirhomes.HCteamsprovideservicessuchaspreventivecare,helpwithactivitiesofdailyliving,medication,personalhygiene,etc.Theycatertoindividualswithdependencywhoneedalowerfrequencyofmedicalandrehabilitationcareandarestillabletoliveinthecommunity.BecauseHC

teamsbelongtoprimarycarecenters,theyfallundertheSNSandarefreeofchargetousers.ThecontractingofNHteamsstartedin2006andtherstHCteamsstartedoperatingonlyin2008.Figure1showstheyearofentryoftherstnursinghomefacility(ontheleftpanel)andthersthome-careteam(ontherightpanel)acrossACESregions.ACESisthePortugueseacronymforPrimaryCareCenterGroupsandtheseareasarerelevantfororganizingprimarycaredelivery.5Th

emajorityofACESregionsexperiencedtheentryoftherstNHin2006and2007andtheentryoftherstHCteaminbetween2008and2010. Figure1:EntryyearoftherstNHunitandtherstHCteamacrossACESregions 5Thereare55ACESregionsinPortugal.ThedensemunicipalitiesofLisbon,Porto,andVilaNovadeGaiahavemorethanoneACES.Becausepatientlocationsarerecordedatthemunicipalitylevelintheinpatientdata,IcollapsetheseACESatthemunicipal