A System For Comparing Outpatient Use Between VA And

A System for Comparing OutpatientUse between VA and MedicarePaul Hebert, PhDChuan-Fen Liu, PhD MPHHSR&D Center of Innovation, VA Puget Sound Health Care SystemHERC Cyberseminar, October 21, 2015

Background: Dual use of VA and Medicare isextensive Medicare eligibleVeterans can use care inVA or Medicare (CMS) There is substantial dualuse of Medicare byVeterans Hynes et al (2007) 46% ofDual-eligible Veterans useboth VA and Medicareoutpatient servicesVETERANS HEALTH ADMINISTRATIONPERCENT OF VETERANS BYUSE OF OUTPATIENT CAREVA only18%Dual use46%Medicareonly36%Hynes DM et al, Medical Care 2007. 45:3; pp 214-2232

VA Reliance has a U-shaped distribution Liu et al (2011) tracked apanel of Veterans for fouryears Reliance is U-shaped More Veterans areheavily reliant on CMSfor specialty care thanprimary care VA reliance decreaseswith ageVETERANS HEALTH ADMINISTRATIONDistribution of patients by reliance onVA for specialty care, 2001-2004MedicarereliantVAreliantLiu CF et al, Medical Care 2011. 49:10; pp 911-173

VA reliance may be a marker for value Dually-eligible Veterans“voting with their feet” mayreflect the overall value ofVA care at VA a facility Wagner et al (2015)- use ofMedicare increasedfollowing large scaleadverse events (LSAE) at aVA facility.Adjusted odds of outpatientsurgery use following LSAE at Q032.521.510.50Q0 (ref)Q1Q2VAQ3Q4MedicareWagner TH, et al. BMJ Qual Saf 2015;24:295–302VETERANS HEALTH ADMINISTRATION4

Different data structures complicate VA-CMScomparisons VA and Medicare comparisons are complicated by differentdata systems at each institution Medicare: billing system VA: clinical care, resource use Burgess, Liu et al (2011): describe methods for comparing VAuse and Medicare use across three dimensions Primary care Specialty care Psychiatric careVETERANS HEALTH ADMINISTRATION5

Objective Describe a framework for a more comprehensivecomparisons of outpatient use at VA and Medicare (CMS) Build on the algorithms developed by Burgess/Liu tocompare outpatient use between VA and Medicare Expand to different types of visits Expand to include different dimensions of visits IIR-10-150-3 (Hebert) Report on VA reliance over time PACT National EvaluationVETERANS HEALTH ADMINISTRATION6

Framework for a morecomprehensive comparisons ofoutpatient use at VA andMedicare (CMS)Paul L. Hebert, PhD

Framework Extend previous methods by Burgess and Liu Add more categories within the type-of-visit Add more dimensions to the visitDimensionType of visitDescriptionWhy did the veteran receive care?Type of serviceType of providerLocation of visitCost of visitWhat did we do for him/her?Who provided the service?Where was the service provided?How much did it cost? (Work-in-Progress)VETERANS HEALTH ADMINISTRATION8

VA and CMS administrative data sourcesDimensionVA data elementsType of service CPT codesType of provider First-listed providerspecialtyLocation of visit Stop codesCost of visitType of visitCMS data elementLine-item CPT codesLine-item specialty codeLine-item place ofservice codeHERC/DSS costsLine-item reimbursed(WIP)amountService location providerWIP work in progressVETERANS HEALTH ADMINISTRATION9

Structure of Medicare Claims: HFCA 1500 CMS Outpatient claims derive from the HCFA 1500 formPlace of service provideslocation of serviceProcedure codes providetype of serviceRendering provider IDidentifies type of providerCharges provide cost ofvisitVETERANS HEALTH ADMINISTRATION10

Type of Service Type of service derived from Berenson-Eggers Type of Service(BETOS) Codes Assigns every HCPCS code to only one BETOS code HCPC codes include all CPT codes used by CMS, plus CMSspecific codes 17000 HCPC codes mapped to 98 BETOS codes Consists of readily understood clinical categories Developed for analyzing the growth in Medicare expenditures. Is stable overtime, and is relatively immune to minor changes intechnology or practice patterns. BETOS-HCPC map available for download from www.cms.govVETERANS HEALTH ADMINISTRATION11

Type of Service- BETOS codesMajor BETOS CategoryExamplesEvaluation and Management (E&M)- 13 subgroupsM1B: Office visit established patientM3: Emergency departmentM5B: Specialist-PsychiatryProcedures - 45 subgroupsP3C: Knee replacementP6A: Minor procedure-skinImaging – 18 subgroupsI2C: MRI BrainTests- 12 subgroupsT2C: EKGDurable medical equipment- 7subgroupsD1D: WheelchairOther- 7 SubgroupsO1B: Chiropractic careUnclassified- 4 subgroupsZ1: Local codesVETERANS HEALTH ADMINISTRATION12

Type of service17000 HCPC codes98 BETOS codes33 Type ofService CodesVETERANS HEALTH ADMINISTRATION13

Location of Visit 53 CMS Place of Service codes and VA Stop codes mapped to 12 locationcodesCode NamePlace of Service Classification1Clinic11 (Office), 22 (Outpatient Hospital), 50 (FQHCs), 2Home12 (Home)3Residential13 (Assisted Living Facility), 32 (Nursing Facility), 4567Retail ClinicUrgent careEmergency RoomInpatient Hospital17 (Walk-in Retail Clinic)20 (Urgent Care Facility)23 (Emergency Room – Hospital)21 (Inpatient Hospital), 51 (Inpatient Psychiatric Facility), 89Ambulatory SurgicalSkill Nursing24 (Ambulatory Surgical Center)31 (Skill Nursing Facility), 61 Inpatient Rehabilitation Facility)101198AmbulanceLabOther41 (Ambulance – Land), 42 (Ambulance – Air or Water)81 (Independent Laboratory)The rest of the codesVETERANS HEALTH ADMINISTRATION14

Type of Provider Derived from CMS specialty codes 127 CMS specialty codesderived from ProviderTaxonomy Codes Reduced to 32 specialty codes Mapped VA specialty codes tothese 32 categories Used the first listed provider forthe primary stop code as the onlyprovider for the visitVETERANS HEALTH ADMINISTRATIONTop provider type codesPrimary matologyPulmonary/Critical CarePain MedicineNeurology15

Classification of type of visit Classify each encounter between a Veteran and the healthcare system (VAor CMS) as one of 8 mutually exclusive categories:Emergency carePrimary careMental healthSpecialty careRehab careTele-careDiagnosticOther Do this in three steps Identify the location of the visit Identify the type of service Identify the type of provider Combine to make type of visitVETERANS HEALTH ADMINISTRATION16

Classification of Visit TypeVETERANS HEALTH ADMINISTRATION17

Results: VA reliance over timeChuan-Fen Liu, PhD

Objective To assess time trends in reliance on VA outpatient carefor various types of visits, and types of providers amongMedicare-eligible Veterans from 2003 - 2012VETERANS HEALTH ADMINISTRATION19

Methods Repeated cross-sectional time series analysis from 2003to 2012 Data sources: VA administrative data and Medicareclaims Study Sample All patients in Primary Care Management Module (PCMM) Enrolled in both Medicare parts A and B Excluded Medicare Advantage (MA) enrollees because of nomedical claims availableVETERANS HEALTH ADMINISTRATION20

Classification of Outpatient Visits Four types of face-to-face visits Primary careSpecialty mental health careMedical specialty careSurgical care Provider specialty Service typeVETERANS HEALTH ADMINISTRATION21

VA Reliance at System Level Proportion of visits occurred in VA for a specific visittype in a given year1. Sum up visits provided in VA and Medicare, respectively,across all patients2. VA reliance VA visits/(VA visits Medicare visits) Ranging from 0 to 1 Adjusted for age, gender, and race by standardizing thestudy population in a given year to the distribution ofstudy population in 2010VETERANS HEALTH ADMINISTRATION22

Number of VA primary care patients withMedicare FFS was stable over timeNumber of Primary Care PatientsMillions76Number of Patients54321200320042005Medicare FFSVETERANS HEALTH ADMINISTRATION200620072008Medicare MA2009201020112012VA ONLY23

Results of Classification of Visits in 2012Number of Visits (millions)Visit TypeVAMedicareEmergency DepartmentPrimary CareMental HealthSurgical CareMedical Specialty CareRehab .24.64.81.04.20.46.58.32.00.03.62.1VETERANS HEALTH ADMINISTRATION24

1005Millions1520Adjusted Total Visit Counts in VA andMedicare by Visit Type: 2003 and 20122003 20122003 20122003 20122003 2012SurgicalMental HealthPrimary CareSpecialtyVACMSVA visit count: Bottom section of the barMedicare visit count: Upper section of the barVETERANSHEALTHADMINISTRATIONAdjusted forage, gender,and race using the study population distribution in 201025

Time Trends of VA Reliance by Visit Type10085.682.3% VA 06200820102012YearPrimary CareSurgical CareMental HealthSpecialty CareAdjusted for age, gender, and race usingthe study population distribution in 2010VETERANS HEALTH ADMINISTRATION26

Increase in VA reliance for specialty care due toan increase in VA visits, not drop in MedicarePrimary CareMental 1020152000200520102015CYVACMSGraphs by Visit TypeVA and CMS visits indexed to 1.0 in 2003VETERANS HEALTH ADMINISTRATION27

Across Board Increase in Use of VA SpecialtyCareMedical ulmonary/Critical CarePain MedicineNeurologyOtolarygologyNephrologyGITotal CPT countsin 2012*(millions)5.55.13.03.02.31.91.81.01.00.9VA Reliancein 2012% Increase of VA Usefrom 2003 to 258%168%178%154%167%178%*VA CMSVETERANS HEALTH ADMINISTRATION28

Top 10 Service types in 2012: CMS, VA and VAreliance (millions)CMSVATotalVA RelianceLabs16.539.656.171%E&M General Office Visits11.510.121.647%OtherAmbulatory and Minorprocedures9.913.623.558%6.12.78.831%E&M Other Physical Therapy2.41.43.736%Other Drugs1.72.74.362%Other testing-Other1.62.44.059%VETERANS HEALTH ADMINISTRATION29

Use of imaging in the VA and CMS, 2003 /ultra- heartImaging-echo/ultra- 012CYVACMSGraphs by Service typeVETERANS HEALTH ADMINISTRATION30

VA reliance Summary Large increase in VA reliance on specialty careresulted from more VA use High reliance on VA mental health care and trendingupward Reliance on VA primary care stable over timeVETERANS HEALTH ADMINISTRATION31

Limitations Trends from repeated cross sectional analysis differfrom longitudinal trends for a given cohort Did not yet adjust for comorbidityVETERANS HEALTH ADMINISTRATION32

Challenges in Classification Visits AcrossSystems Complex process, but understand details of crosssystem use Billing records vs encounter data Some codes only in VA, eg. Tele-care Changes in CPT coding over time Policy changes Changes in reimbursement rules Map BETOS codes annuallyVETERANS HEALTH ADMINISTRATION33

Acknowledgement HSR&D funded study IIR-10-150 PACT national evaluation funded by Office ofPatient Care Services Data analysis team Elaine Hu, MS Adam Batten, BS Haili Sun, PhDVETERANS HEALTH ADMINISTRATION34

ype of service derived from Berenson-Eggers Type of Service T (BETOS) Codes Assigns every HCPCS code to only one BETOS code HCPC codes include all CPT codes used by CMS, plus CMS specific codes 17000 HCPC codes mapped to 98 BETOS codes Consists of readily understood clinical categories