Estimating non-VA Health Care Costs

1
Estimating non-VA Health Care Costs

• Todd H. Wagner
• Ciaran Phibbs
• Mark W. Smith

2
Who we are
Mark
Todd
Ciaran
3
Learning Objectives

• After this talk, you will
• Understand whether you need non-VA data
• Know the strengths and weaknesses for different
  sources of non-VA data

4
Do you need non-VA data?

• Many veterans have a choice in where they get
  care
• Many veterans who get care from VA facilities
  also get care from non-VA providers (e.g.,
  Medicare, Medicaid)
• Perspective and objectives these should dictate
  your data needs

5
Example

• Any examples of studies that require non-VA data?

6
Sources of non-VA data

• Medicare data
• Fee Basis
• Bills from providers
• Self-report
• All sources have strengths and weaknesses

7
Medicare Data
8
Medicare Data for Veterans

• Medicare is health insurance for people over age
  65 or those with a disability
• VIReC maintains
• Medicare Data for all VA enrollees from 1999
  through 2003
• Note the delay this may be critical for clinical
  trials.

9
Medicare Institutional Claims

• AKA Part A (except outpatient)
• Inpatient (short/long)
• Outpatient (Part B)
• Home Health (Part A B)
• Hospice
• Skilled Nursing Facilities
• One file for each type of claim

10
Medicare Non-Institutional Claims

• AKA Part B
• Physician/supplier file
• Physician, NPs, and other professionals
• Clinical Laboratories
• Ambulance services
• Ambulatory Surgery Centers
• Durable Medical Equipment (DME) file

11
Medicare File Types

• Research Identifiable Files (RIFs)
• Beneficiary Encrypted Files (BEFs)
• Limited Data Set (LDS)
• Downloadable files (PUFs)

12
Charges in Medicare Data

• Charges reflect billed amount.
• Charges gt Costs.
• Adjust charges using cost-to-charge ratio (CCR).
• Cost to charge ratio is calculated from Medicare
  Hospital Cost Report

13
Medicare Payments

• Payments reflect amount paid by Medicare. This
  reflects
• Co-payments, deductibles, coinsurance
• Benefit limitations
• Wages, disproportionate share, IME
• Direct medical education
• Outlier payment
• Reimbursement Amount DRG Price Outlier
  Payment Individual Payment Other Insurance
  Payment

14
Fee Basis
15
Overview of Fee Basis Program

• Pays for care at non-VA facilities when
• it is the only source available, or
• VA could save money
• Full range of services covered
• Mostly pre-arranged limited emergent care

16
Fee Basis files

• Subset of all VA contract care
• Non-VA PTF has detail on hospital stays some
  overlap with Fee Basis files
• Substantial utilization unaccounted for
• SAS format at Austin

17
Highlights of Financial Data

• Amount claimed
• Amount paid
• often much less than amount claimed
• Many variables relating to FMS record-keeping
  invoice date, processing date, check number,
  check date, cancel code, etc.

18
User Notes

• Each paid invoice has a separate record.
• Invoices may be sent LONG after services are
  rendered
• Search for records in many years

19
Using Fee Basis Files Cautions

• Beware of missing decimal places
• ICD diagnosis codes
• Payment amounts
• Care in community nursing homes, state veterans
  homes, and some non-VA hospitals may also be
  recorded in other files
• e.g., contract nursing home care appears in DSS
  outpatient files

20
Bills from Providers
21
Collecting Billing Data

• With consent, you can attempt to collect hospital
  bills
• We are doing this for a few clinical trials
• Mixed success typically only done for inpatient
  costs

22
Method

1. Use self-report to identify utilization
2. Ask patient for name of hospital and approximate
   date
3. Have patient sign HIPAA release
4. Contact hospital for UB92
5. Cost adjust the charges reported on the bill

23
Self-Report
24
Limits with Administrative Data

• Obtaining and analyzing claims data can be costly
  and time consuming
• Administrative data can be incomplete or
  inaccurate
• Limited benefits
• Out-of-plan or out-of-pocket utilization
• Capitated health plans

25
What is Self Report?

• Cognitive process of recalling information
• Ample opportunity for distortion and error
  (Khilstrom et. al 2000)
• Self-report not valid when people lack the
  cognitive capabilities

26
Modifiable Components

1. Recall timeframe
2. Type of utilization
3. Utilization frequency
4. Questionnaire design
5. Mode of data collection

27
Questionnaire Design

• No standards exist
• Pretest Dillman (2000)
• Use counts for responses (not categories)

28
Self-Reported Costs

• Self-reported costs are assumed poor
• Imputing costs from self-reported utilization can
  introduce biases

29
Additional Readings

• Fee Basis Technical Report 18
• Bhandari and Wagner. Self-Reported Utilization of
  Health Care Services Improving Measurement and
  Accuracy (2006, MCRR). Available upon request.

30
Additional Links

• VIReC Manages the VA Medicare Data
• www.virec.research.med.va.gov
• ResDAC (Research Data Assistance Center) for
  Medicare data
• www.resdac.umn.edu
• Medicare and Medicaid
• www.cms.hhs.gov

31
Additional Viewings

• 2005 HERC Courses
• Talk on Medicare Data (Yu)
• Talk on the Fee Basis (Smith)
• http//www.herc.research.med.va.gov/resources/trai
  ning_course_archives.asp

32
Next Health Economics Course

• May 18, 2006
• Effectiveness, Patient Preferences and Utilities
• Patsi Sinnott, PT, PhD, MPH