Medicaid Analytic eXtract MAX

1
Medicaid Analytic eXtract (MAX)

• Presentation to the
• Academy Health Annual Research Meeting
• San Diego, California
• Dave Baugh, CMS/ORDI
• June 8, 2004

2
What is MAX?

• Person-based Medicaid data used for
• Research/evaluation
• Epidemiology/quality
• Statistics/forecasting
• Calendar Year (begins 1999, SMRF - prior years)
• Event Based
• Occurrence of eligibility
• Dates of service
• Final action events (hospital stays, visits,
  etc.)
• Derived from MSIS (7 calendar quarters)

3
Why Do We Need MAX?

• Eligibility
• Retroactive eligibility in proper chronology
• Eligibility codes verified and improved
• Eligibility data added to each claim
• Services (Claims)
• Final action events (interim claims combined)
• Organized by dates of service
• Type of service verified and regrouped
• Person Summary File
• Calendar year eligibility and summary of claims
• Not available from MSIS

4
MAX Data Sets

• Person Summary File
• Eligibility (annual and monthly)
• Managed care enrollment
• Utilization and Medicaid payment by type of
  service
• Service Files
• Inpatient hospital
• Long term care
• Prescription drug
• Other Services
• Service file records include
• Fee-for-service
• Prepaid plans - premium payments and encounters
  (incomplete)

5
Medicaid Data Enhancements Beginning
1999

• More detail Medicaid eligibility
• Dual (Medicare and Medicaid) status
• Medicaid case number
• Enrollment in prepaid plans
• Other eligibility (e.g. TANF, SCHIP)
• Services
• More diagnoses and procedures
• More data (waiver enrollment, hospital cost
  centers)
• Additional types of service (e.g. DME/supplies,
  adult day care)
• Maternal delivery indicator

6
MAX Data Linkages

• To Medicare Enrollment Data Base (EDB)
• Best way to identify dual eligibles
• Begin and end dates of Medicare eligibility
• Other Medicare data (e.g. Medicare HIC, date of
  death)
• To Medi-Span and First Data Bank
• Prescription drugs
• Link on National Drug Code (NDC)
• Therapeutic classes (clinical use)
• Other FDB data (e.g. generic, OTC or prescribed
  drug)
• Other linkages (Agreement with SSA)

7
MAX/SMRF DataAvailability

• Who has access?
• Privacy Act and HIPAA regulations apply
• Research protocols must be reviewed
• A Data Use Agreement (DUA) must be filed
• A CMS processing fee may apply
• Access to Medi-Span and First Data Bank data
  restricted
• What data are available?
• Years prior to CY 1999 (SMRF)
• 1992-1998 25-29 states full data
• 1987-1991 5 states, data quality?
• Years after CY 1998 (MAX) all States
• CY 1999 available now
• CY 2000 available beginning mid-2004

8
MAX/SMRF DataDocumentation

• Documentation on the Web via
• www.cms.gov/researchers/max
• Data Dictionaries
• Better descriptions of data elements
• Improved source information
• Addition of user notes
• Data Validation Reports
• Data Anomaly Reports
• Valid data, but unexpected results (e.g. broken
  time series, new covered service)
• Data inconsistencies (cant be fixed)

9
Medicaid Data LimitationsMSIS and MAX

• Data not reported
• Some desired beneficiary characteristics
• Some aggregate payments
• Provider characteristics
• Incomplete data
• Periods of ineligibility
• Third-party insurance coverage and payments
• Services for persons in prepaid plans
• Service detail for dual eligibles
• Drug payments are prior to rebates
• Program and operational variation

10
Estimates of Dual andFull Medicaid Benefit Dual
Eligibles(Using MAX)
11
Linkage to Medicare (EDB)Two Steps

• Not available Name and Address
• Not used (initially)
• Medicare Health Insurance Claim (HIC)
• Medicaid dual status
• Step 1 - Linking criteria
• For Aged - SSN and gender
• For Disabled, either
• SSN and date of birth (DOB), or
• SSN, gender and two of three elements in DOB
• Step 2 Linking criteria for step 1 non-links
• Medicaid SSN to EDB claim account number (CAN),
  plus
• Gender and DOB

12
Setting Dual EligibilityAfter the link

• For each linked eligibility record
• Monthly Medicaid eligibility is compared to
  spells of Medicare eligibility
• An dual indicator is set when dates overlap
• This indicator confirms dual status
• By month
• For the year (ever a dual in the year)

13
Estimating Dual EligiblesAdjusting for bias

• Estimates adjust for
• Undercounting
• Medicaid-reported duals not linked
• With no SSN
• With incorrect/non-matching SSNs
• Estimates do not adjust for
• Undercounting
• Medicare-reported duals not linked
• Overcounting
• Medicaid persons eligible in more than one state
• More than one Medicaid eligible per SSN

14
Estimating Dual EligiblesAlternative Estimates

• Best Estimate of Duals
• Confirmed duals (linked to EDB), plus
• Medicaid eligibles not linked to EDB, but
• Identified as dual eligibles by Medicaid, and
• Had at least one claim in the year where Medicare
• copayment and/or deductible was paid by
  Medicaid
• Upper Bound Estimate of Duals
• Same as above except for
• Medicaid eligibles not linked to EDB replace
  and with and/or, plus
• Estimate not gt total aged and disabled eligibles

15
Full Medicaid Benefit Dual EligiblesBackground
on Estimates

• State reporting of dual status in MSIS
• Incomplete at best starting in 1999
• For Calendar Year 1999
• 11 states reported gt 50 unknown type
• 21 states reported gt 20 unknown type
• 5 states reported no full duals
• One state did not report type of dual

16
Full Medicaid Benefit Dual EligiblesMethodology

• Lower Limit estimate
• By state, allocated unknown type proportionally
• For the six states identified above -
• Produced a combined percent of full benefit to
  total dual eligibles for all other states.
• Multiplied this percent by the total number of
  dual eligibles in the state.
• Best estimate
• Same as above, except
• Allocated all unknown type to full benefit

17
Dual and Full Medicaid Benefit Dual Estimates
(ever in 1999)

• National estimates (50 states and D.C.)
• Dual eligibles
• Confirmed (from EDB) 6.823 million
• Best estimate 6.881 million
• Upper bound estimate 7.288 million
• Full Medicaid Benefit Dual eligibles
• Lower bound estimate 5.916 million
• Best estimate 6.091 million
• Restricted Benefits estimate 6.015 million