The Project HOPE CENTER ADESCRIPTIVEANALYSISOF for MEDICAREHOSPITALEPISODES HEALTH AFFAIRS WITHCRITICALCAREBILUNGS: IMPUCAT10NSFORBUNDUNGSERVICES FORPRICING FinalReport July30.1993 ProjectHOPECenterforHealthAffairs 7500OldGeorgetownRoad,Suite600 Bethesda.Maryland20814 7500OldGeorgetownRoad,Suite600,Bethesda,Maryland 20814-6133 ADESCRIPTIVEANALYSISOF MEDICAREHOSPITALEPISODES WITHCRITICALCAREBILUNGS: IMPUCATIONSFORBUNDUNGSERVICES FORPRICING FinalReport July30,1993 ProjectHOPECenterforHealthAffairs 7500OldGeorgetownRoad,Suite600 Bethesda,Maryland20814 "^r TableofContents Section Page ListofRguresandTables iv Preface vi Abstract vii Summary viii Section 1. Introduction 1 Section2. Background 3 2.1 DefinitionofCriticalCareServices 3 2.2 PhysicianBillingsforCriticalCareServices 5 Section3. DataFileConstmctionandMethods 7 3.1 OverviewofApproach 8 3.2 ConstructionofCriticalCareEpisodes 9 Section4. Results: HospitalStayEpisodeswithCriticalCareBillingsand ICU/CCUUse 16 4.1 Demographics 16 4.2 CodingandFrequencyofCriticalCareUse 16 4.3 CostsandChargesAssociatedwithCriticalCareUse 19 4.4 AnalysisofVariations 22 4.5 UseofRelatedandLocalHCPCCodes 31 Section5. Discussion 35 5.1 SummaryofKeyRndings 35 5.2 VariationsinCodingandBilling 36 55..34 TImhpeliNcaetwionCrsitfiocralBuCnadrleinCgoadnedsAlternativePaymentSchemes .. 3407 Section6. Conclusion 43 References 45 u TableofContents(continued) Appendices: A. TKaeblyeMAe-a1s:uMreesanbsyaDnRdGCoefficientsofVariationsfor A-1 B. ListofHCPCCodesUsedMoreThan500timesin 53,091 CriticalCareHospitalStayEpisodes B-1 C. ListofCarrierLocalCodesforProcedures RelatedtoCriticalCare C-1 lU ListofFiguresandTables Figures Page Figure1: CriticalCarePhysicianBillingCodes 4 Figure2: InitialRieConstructionSteps 11 Tables Table1: UseofCriticalCarePhysicianServices, 1987 6 Table2: FrequencyDistributions: DemographicVariables 17 Table3: CodingofCriticalCareEpisodes: FrequencyofCodeUse ..18 Table4: CombinationsofCriticalCareCodesUsedin AtLeastOnePercentofCases 20 Table5: AllowedChargesforCriticalCareServices 21 Table6: FrequencyDistributions: TotalAllowedChargesfor AllBilledCriticalCareServices 23 Table7: ComparisonsofKeyVariables: MeansandCoefficients ofVariation 24 Table8: Diagnosis-RelatedGroupsUsed inMoreThanOnePercentofCases) 26 Table9: AComparisonofCriticalCareCodesBilled, TotalAllowedCharges,andICU/CCUDaysByDRG 27 Table10: AComparisonofCriticalCareAllowedCharges,Total HospitalCharges,andtheRatioofCriticalCareCharges toTotalHospitalChargebyDRG 28 Table11:SAerCvoimcepsarainsdonAllofOtBihlelringPsarftorBCrSietricvailceCsarbeyPDhRysGician 29 Table12: HCPCCodesUsedOver7,000TimesDuring CriticalCareEpisodes 32 IV ListofFiguresandTables(continued) TableA-1: MeansandCoefficientsofVariationforKeyMeasures byDRG A-1 Prefece Likemanydata-intensiveresearchprojectsusingverylargedatafiles,thisone beganwithoverlyambitiousaimsthatweregraduallymodifiedastheprojectevolved. Thedifficultiesinvolvedinmanipulatingtheselargefilesplayedalargeroleinthe eventualdevelopmentandexecutionoftheproject. LouGarrisonservedasthe projectleader. VeraKuriantzickandBethMcLaughlinweretheotherkeyanalystson theproject. YiminNganofSocialandScientificSystems,Inc.,didthebulkofthe analyticprogramming. PortiaDeFilippesofProjectHOPEwasresponsibleforthe initialcomputationsattheHCFADataCenter. SheldonRetchinofProjectHOPE servedasaconsultantreviewerandmademanyusefulsuggestionseariyoninthe project. StaffoftheHCFAOfficeofResearchandDemonstrationsweresupportiveand patientthroughouttheproject. Themembersoftheprojectteamremaingratefulfor tthheeedfaftoartpsroofbloeumrsi.nitiWaleHwCoFulAdparoljseoctlikoefftiocert,haRnokbeorutrGsruubbseerq,uteontaspsriosjtecutsoifnfiocveerrsc-oming MichaelBorowitzandWilliamBuczko-fortheirassistance. Thisfinalreportbenefitted fromthecommentsofanonymousHCFAreviewersonaneariierdraft. Thisreport describestheworkcompletedunderthisprojectandshouldbeofassistancetoORD andfutureresearchersinterestedintheprovisionofcriticalcareservices. VI Summaiy Thepurposeofthisprojectwas(1)toanalyzebillingandutilizationpatternsfor physiciancriticalcareservicesprovidedtoMedicarebeneficiairesand(2)toassess theimplicationsofthesefindingsforbundlingtheseservicesforpricing. Thestudy usedthe1987PartA/PartBMergedFile,representingafivepercentsampleofall beneficiaries. Extensivecomputationswererequiredtoisolateandconstructhospital stay"episodesofcare"involvingcriticalcareuse. Billingsrelatedtothesixprincipal criticalcarecodeswerelinkedtospecifichospitalstaysthatincludedreporteddaysin intensiveorcoronarycareunits. Theprincipalanalyticfileconsistedofsome53,000 hospitalstayepisodesinvolvingbothICU/CCUuse(aPartAbilling)andcriticalcare billings(PartB). Asubstantialvolumeofhospitalizations(over74,000discharges)involving criticalcare,asIndicatedbyreportedICU/CCUdays,didnothavereporteduseofany ofthesixcriticalcarevisitcodes. Itisimportanttonotethattheseepisodeswerenot studiedindetail. Thedetailedanalysesfocusedonthe53,000hospitalstayepisodes withbothICU/CCUuseandcriticalcarePartBbillings,andyieldedseveralkey findings: 1) Evenamongthosedischargesindicatingsomecriticalcarephysician use,therewasconsiderablevariabilityinbillingpattemsintermsofthe typesandfrequencyofcodesused. 2) Totalbillingsforfollow-upcriticalcarevisitcodesweremuchhigherthan thosefortheinitialvisitcodes: theyaccountedforabout73percentof totalcriticalcarephysicianspending. 3) Comparedtototalhospitalchargesandotherphysicianchargesper episode,averagecriticalcarephysiciancharges($270.80perepisode) weresmallonaverage-about2percentoftheformerand12percentof thelatter. 4) AmongdischargeshavingbothICU/CCUuseandcriticalcarebillings, perhospitalstayspendingoncriticalcarephysicianservices(mean coefficientofvariation(CV)=167percent)variedmoreonaveragethan viii thatforeitherhospitalcharges(meanCV=150percent)orother physiciancharges(meanCV=124percent). 5) Amongthesehospitalepisodeswithcriticalcarephysicianbillings,on average,fewerthanonecriticalcareservice(0.55onaverage)wasbilled perdayintheICU/CCU. 6) Forthisanalyticsample,episodesfallingintoonly25DRGsaccounted for70.3percentofallhospitalepisodeswithcriticalcarephysician servicebillings. 7) Amongthese25DRGs,averagewithin-DRGvariabilityforcriticalcare physicianbillings(meanCV=127percent)wasgreaterthanthatfortotal hospitalcharges(meanCV=103percent)orfortotalotherphysician charges(meanCV=91 percent). 8) Acrossthese25DRGs,thevariabilityintheaveragecriticalcarebilling (CVacrossthemeans=34percent)waslowerthanfortotalhospital charges(CVofmeans=71 percent)andforotherPartBbilledservices (CVofmeans=106percent). 9) CriticalcarebillingsperICU/CCUdayvariedlittle(CVofmeans=10 percent)acrossthese25DRGs;theaveragenumberofICU/CCUdays variedmore(CVofmeans=38percent). 10) Within-DRG,forthese25DRGs,theaveragevariabilityinICU/CCUdays wasmoderate(meanCV=110percent),comparedtothe127percentfor criticalcarechargesandthe103percentfortotalhospitalcharges. 11) Forthecriticalcareepisodesinthese25mostfrequentlyoccurring DRGs,boththenumberofotherphysicianbillings(21.2procedurecodes onaverage)andthetotalbillings($1774.84)farexceededthenumber billedcriticalcarecodes(2.83)andtherelatedbillings($262.36). Thesefindingsindicatethatcodingpracticesandbillingsforcriticalcare physicianservicesin 1987exhibitedsubstantialvariability. Onahospitalepisode basis,paymentsforphysiciancriticalcareserviceswerequitesmall(andmore variable),ascomparedtothetotalepisodecost. Thetotalphysicianbillfortheentire episodewasamorestableandmuchlargerproportionofthetotalcost. Asa potentialcase-mixadjuster,DRGsperformedbetterforallPartBphysicianservices andtotalhospitalchargesthanforcriticalcareservices. Thenumberofcriticalcare IX codesbilledwasconsistentlyandsignificantlylessthanthenumberofICU/CCUdays reportedforthe25DRGsinvolvingsubstantialcriticalcareuse. Allofthissuggests that: (1)someclarificationisneededregardingthecodingofcriticalcareservicesand (2)theremaybelimitedutilityinbundlingtheseservicesaloneonaDRG-basisorper diembasis(intheICU). Severalimportantlimitationsofthisanalysisshouldbehighlighted. First,it examinedindetailonlyaportionoftheMedicarepatientsreceivingcriticalcare. In particular,thesedischargesreflectedanarrowdefinitionofahospitalstayepisode Involvingcriticalcareuse. Second,noadjustmentsweremadeforgeographiccostof livingdifferences,billingpracticesofsurgeonsandresidents,orlocalcodes. Third, thevariationsincodingandbillingpracticesintroduceerrorintothesubsequent analysisofspendingpatterns. RecentchangesincriticalcarecodingunderthenewMedicarefeeschedule addressthiscodingissueaswellastheincentivetobillfollow-upcriticalcarecodes. However,theymakeitlessfeasibleandlesslogicaltodevelopaseparateperepisode paymentbundleforcriticalcare. Theyalsodonotmakeitanyeasiertoconduct researchonthistopic,since(a)someproceduresarebundledinthecriticalcare constantattendancecodesand(b)subsequenthospitalvisitcodesaretobeusedfor routinecaretocriticalcarepatients. Globalsurgicalbillingpolicyandbillingby residentsfurthercomplicatesthesituation. Thus,futureresearchonphysician servicesprovidedtocriticalcarepatientsislikelytorequiredetailedchartreview. Basicresearchonthestructureandoperationofcriticalstaffingwouldalsobean importantcontribution, Rnally,researcheffortstosupportpoliciestocontainphysician inpatientservicegrowthmightbetterfocusonanalysisofthepotentialforbundlingall inpatientphysicianservicesonanepisodebasis. Thereisalsoaneedforoutcomes researchoncriticalcaretoaddressitscosteffectiveness.