Associations between outpatient and inpatient serv

RESEARCHARTICLEAssociationsbetweenOutpatientandInpatientServiceUseamongPersonswithHIVInfection:APositiveorNegativeRelationship?JohnA.Fleishman,RichardD.Moore,RichardConviser,PerrinB.Lawrence,P.ToddKorthuis,andKellyA.GeboObjective.Toexaminetheprospectiveassociationbetweenfrequencyofoutpatientvisitsandsubsequentinpatientadmissions.DataSources.Medicalrecorddataon13,942patientswithHIVinfectionseenin10HIVspecialitycaresitesacrosstheUnitedStates.StudyDesign.ThisobservationalstudyfollowedacohortofHIV-infectedpatientswhowereincareinthefirsthalfof2001.Numbersofinpatientadmissionsandout-patientvisitswerecalculatedforeachpatientforeach3-monthperiod,from2001through2004.Analysis.Negativebinomialandlogisticregressionanalysesusingrandom-effectsmodelsexaminedtheeffectsofinpatientadmissionsandoutpatientvisitsinthepreviousperiodoninpatientandoutpatientserviceutilization,controllingforbackgroundchar-acteristicsandHIVdiseasestage.Results.For3-monthperiods,between5and9percentofpatientshadaninpatientadmission.Thelinearassociationbetweennumberofoutpatientvisitsandanyinpatientadmissioninthesubsequentperiodwaspositive(adjustedoddsratio51.05;95percentconfidenceinterval[CI]51.04,1.06).However,patientswithzeroprioroutpatientvisitshadsignificantlygreateradmissionratesthanthosewithonepriorvisit.HospitalizationrateswerealsohigheramongthosewithapriorhospitalizationandthosewithmoreadvancedHIVdisease.Conclusions.TheseresultssuggestaJ-shapedrelationshipbetweenoutpatientuseandinpatientuseamongpersonswithHIVdisease.Thoseinworsehealthhavegreaterutilizationofbothinpatientandoutpatientcare.However,havingnooutpatientvisitsmayalsoincreasethelikelihoodofsubsequenthospitalization.Althoughoutpatientcarecannotbejustifiedasacost-savingmechanism,maintainingregularclinicalmonitoringofpatientsisimportant.KeyWords.HIVinfection,inpatientserviceuse,outpatientserviceuseNoclaimtooriginalU.S.governmentworks.rHealthResearchandEducationalTrustDOI:10.1111/j.1475-6773.2007.00750.x76InthefirstdecadeoftheHIVepidemic,effortsweremadetodevelopcom-munity-basedsystemsofcare,inparttoreducethefrequencyandcostofinpatienttreatment(Moretal.1994).Theseeffortsculminatedin1990withthepassageoftheRyanWhiteCAREAct,whichprovidesfundstosupportandenhancecommunity-basedcareforHIVdisease(Bowenetal.1992).Onepremiseofsuchprogramsisthatappropriateoutpatientcarecouldreducetheincidenceofcostlyinpatientepisodes.Closeroutpatientdiseasemanagementcouldpotentiallyavoidhospitalizationbydiscoveringandtreatingincipientconditions.Ifthisiscorrect,onewouldexpecttofindanegativerelationshipbetweenoutpatientandinpatientutilization.Alternatively,sickerpatientsmightconsumehigherlevelsofbothinpatientandoutpatientcare,leadingtoapositiverelationshipbetweeninpatientandoutpatientutilization.RelativelyfewstudiesofpersonswithHIVhaveinvestigatedtheasso-ciationbetweeninpatientandoutpatientserviceuse.Mostresearchexaminesinpatientandoutpatientuseinseparateanalyses(e.g.,Shapiroetal.1999;Fleishmanetal.2005).Cunninghametal.(1996)reportedapositiveassoci-ationbetweennumberofambulatoryvisitsandtheprobabilityofhavinghadapriorhospitalization.However,thesamplewasrecruitedfrompatientshos-pitalizedforHIV-relatedconditions,raisingthepotentialforselectionbias.PezzinandFleishman(2003)developedacomplexeconometricmodeltoexaminetheeffectofnumberofambulatoryvisitsoninpatientandemergencydepartmentuse.Overall,theassociationbetweennumberofambulatoryvisitsandnumberofinpatientepisodeswasnotsignificant.Thesestudieswerebasedondatacollectedbeforetheintroductionofhighlyactiveantiretroviraltherapy(HAART)in1996.Itisnotclearifsimilarrelationshipswouldbeobtainedinthecurrenttreatmentenvironment.TheintroductionofHAARTwasassociatedwithsubstantialdeclinesininpatientcensusesofpatientswithHIV(TorresandBarr1997;FleishmanandHellinger2001;Gebo,Diener-West,andMoore2001;Pauletal.2002).Despitethereductions,HIV-relatedinpatientutilizationisstillfarfromrare;among15,211HIV-infectedpatientsfromseveralcaresitesin2001,20.4percenthadoneorAddresscorrespondencetoJohnA.Fleishman,Ph.D.,CenterforCostandFinancingStudies,AgencyforHealthcareResearchandQuality,540GaitherRoad,Rockville,MD20850.RichardD.Moore,M.D.,M.H.S.,PerrinB.Lawrence,M.P.H.,andKellyA.Gebo,M.D.,M.P.H.,arewiththeDepartmentofMedicine,JohnsHopkinsUniversitySchoolofMedicine,Baltimore,MD.RichardConviser,Ph.D.,wasformerlywiththeHIV/AIDSBureau,HealthResourcesandSer-vicesAdministration,Rockville,MD.Hiscurrentaddressis:2780LorraineAve.,Missoula,MT.P.ToddKorthuis,M.D.,iswiththeDepartmentofMedicine,OregonHealthandScienceUni-versity,Portland,OR.OutpatientandInpatientServiceUseamongPersonswithHIVInfection77moreinpatientepisodes(Fleishmanetal.2005).InpatientcareremainsamajorcomponentofoverallexpensesfortreatingHIVinfection(Geboetal.2006).Thisstudyexaminestheassociationbetweenambulatoryvisitsandin-patientadmissions,usingdatafrom2001to2004inalargesampleofHIVpatientsincare.Weexaminewhetherambulatorycarehelpstopreventsub-sequentinpatientcare(i.e.,anegativerelationship)orwhetheroutpatientandinpatientcarearecomplementary(i.e.,apositiver