Medicare and Medicaid Enrollment and Claims files - PowerPoint PPT Presentation

Loading...

PPT – Medicare and Medicaid Enrollment and Claims files PowerPoint presentation | free to download - id: 4e62d1-OWVlY



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Medicare and Medicaid Enrollment and Claims files

Description:

Medicare and Medicaid Enrollment and Claims files Objective To assess the disparities in colorectal cancer (CRC) screening between elderly dual Medicare-Medicaid ... – PowerPoint PPT presentation

Number of Views:237
Avg rating:3.0/5.0
Slides: 76
Provided by: Prefer898
Learn more at: http://msass.case.edu
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Medicare and Medicaid Enrollment and Claims files


1
Medicare and Medicaid Enrollment and Claims files
2
Note
  • Rather than a detailed description, these slides
    are intended to provide a broad overview of
    Medicare and Medicaid databases.
  • For detailed description of the contents of the
    file, please refer to the following website
  • Resdac.umn.edu

3
Medicare program description
  • Program started in 1965
  • Eligible population
  • Elderly
  • Disabled
  • End-stage renal disease (ESRD)
  • The dually (Medicare-Medicaid) eligible
  • Growing program and increasing number of
    enrollees, with the aging of the population (gt 40
    million people)
  • More women than men

4
Program description, contd
  • Part A inpatient hospital
  • Entitlement program
  • 98 of individuals 65 years of age or older are
    enrolled in Part A
  • Deductible applies for each spell of illness
  • Coinsurance applies
  • Part B outpatient care
  • must pay to be enrolled in Part B
  • Payment deducted from soc sec check
  • 96 of elderly and 90 of disabled are enrolled
    in Part B
  • Deductibles and coinsurance apply (with the
    exception of some services)

5
Program description, contd
  • Part D Drug benefit
  • Introduced in January 2006 to help address the
    needs of seniors with high out-of-pocket costs
    for medications
  • Potential coverage gap or so-called doughnut
    hole
  • Partial coverage for the first 2,250 of total
    drug costs in 2006 followed by a period of no
    coverage until patients reach cumulative
    out-of-pocket costs of 3,600 in 2006.
  • At the end of the gap period, catastrophic
    coverage begins (co-pay of 5 of drug costs or a
    pre-determined co-pay from that time onward)
  • 18.2 million beneficiaries were enrolled in
    Medicare Part D in 2006.
  • Schmittdiel et al. Am J Managed care, 2009
    15(3) 189-193.

6
Note on Dually Eligible population
  • Elderly people meeting certain eligibility
    criteria can enroll in state Medicaid programs,
    in which case Medicaid provides coverage for
    their premium, deductible and coinsurance (also
    referred to as state-buy-in)
  • Dually eligible people representing the poorest
    sickest and the frailest of the elderly
    population
  • Enrollment of the eligible elderly in Medicaid
    remains quite low
  • Implications in research

7
Program description, contd
  • Supplemental medical insurance (SMI) purchased
    individually or in group, intended to provide
    coverage
  • for services not covered under Medicare
  • Medicare deductible and coinsurance
  • Services NOT covered by Medicare
  • Eyecare
  • dentures

8
Program description, contd
  • Managed Care
  • Capitation based, based on the Adjusted Average
    Per Capita Cost (AAPCC 95 of fee-for-service
    (FFS) payment for the county, adjusted for age,
    gender, institutional status, and dual
    eligibility status)
  • Increased enrollment over time (15)
  • Managed care enrollment NOT uniform across
    geographic regions of the country
  • Absence of encounter level data ? Enrollees
    represented in the denominator files, but not in
    claims files
  • IMPLICATIONS IN ANALYSIS

9
Medicaid program description
Source the Kaiser Family Foundation (www.kff.org)
10
DATABASE STRUCTURE Enrollment files
  • Beneficiary identifier
  • Demographics (age, race, sex)
  • Enrollment spans
  • Eligibility categories (Medicaid)
  • Participation in managed care / state buy-in
    (Medicare)
  • Dual enrollment, spenddown (Medicaid)
  • County / address of residence
  • Vital Status (Medicare)
  • One record for each individual enrolled in the
    program during a given calendar/fiscal year

11
Enrollment file, contd
  • In Medicare, the information is finalized at the
    end of the quarter following a calendar year
  • NOT UPDATED BEYOND THAT POINT

12
Validity of demographic information
  • Denominator file is the recommended source of
    demographic variables for Medicare data analysis
  • Considered reliable for the most part
  • Age
  • Sex
  • Race (depending which category)
  • Place of residence (consider special category of
    elderly living half-time in Florida)
  • Vital Status -- date (not cause) of death

13
Note on HMO monthly indicators
  • Not possible to tell
  • whether disenrollment is by choice, or whether
    managed care program discontinued service in area
    of residence
  • Whether disenrollment is due to death (indicators
    set to zero after death ? check the date of
    death)
  • Switching between managed care plans

14
General use of the denominator file -- examples
  • Analyze shifts in demographics over time
  • Obtain denominators to calculate rates by
    region, demographics, eligibility category
  • Obtain months on FFS vs MC gt monitor managed
    care penetration rates over time and across
    regions (FFS months calculated as the difference
    between total months and MC months)
  • Obtain months on state buy-in (Medicaid
    eligibility also referred to as dual eligibles)

15
DATABASE STRUCTURE (non-pharmacy) Claims Files
  • Beneficiary ID
  • Dates of service
  • Diagnosis codes
  • Procedure codes
  • DRG code
  • Revenue center codes
  • Length of stay
  • Charges
  • Reimbursement amount

16
Pharmacy claims files
  • Beneficiary ID
  • DOB, Gender
  • Paid date
  • Date of service
  • Service Provider
  • Compound code
  • Quantity dispensed

17
DATABASE STRUCTURE CLAIMS FILES (organized by
state fiscal year, based on date of service)
I.
Institutional Claims Inpatient Hospital,
Outpatient Hospital, and Other Institutional
Header
Detail line item
II.
Non-Institutional Claims Physician, Clinic, Lab,
X-Ray, etc.
Header
Detail line item
Drug Claims
III.
18
MedPAR file
  • Hospital admissions for Medicare beneficiaries
  • One record per admission
  • Record carries
  • Beneficiary ID
  • Demographics (age, sex, race)
  • Date of birth date of death (if available)
  • Up to 10 ICD-9 diagnosis codes
  • Up to 10 ICD-9 procedure codes
  • Admission date discharge date
  • Source of admission
  • Discharge status
  • Provider ID ? possibility to link to American
    Hospital Assoc data to retrieve hospital
    characteristics.

19
MedPAR file, contd
  • Admission and discharge dates useful
    consistent diff1 agrees with LOS on record
  • Clinical information
  • DRGs (1 per stay)
  • Diagnoses (ICD-9 primary discharge dx 8
    secondary dx 1 injury)
  • Procedures up to 10 with corresponding dates
  • Admission diagnosis code
  • E codes to identify mechanism of injury coded
    inconsistently (depends of fiscal intermediaries)
  • Same applies to V codes (supplementary
    classification of factors influencing health
    status and contact with health services)

20
MedPAR file, contd
  • Pre-existing conditions and comorbidities can be
    identified using diagnosis codes
  • Difficulty to distinguish between pre-existing
    conditions and complications
  • Rule-out diagnosis (?)
  • Admission type common codes
  • emergent (patient admitted through the ER)
  • urgent (patient admitted to the first available
    and suitable accommodation)
  • Elective (patients condition permitted adequate
    time to schedule availability of suitable
    accommodations)

21
MedPAR file, contd
  • Discharge status (alive dead)
  • Discharge destination (e.g., home/self care SNF
    home health service care left AMA died)
  • Readmission vs. transfer
  • Detailed charge and reimbursement data

22
Part B files
  • Outpatient standard analytic file
  • Physician supplier / Carrier file
  • Important logistic issue
  • (ENORMOUS) SIZE OF FILES, EVEN WHEN LIMITED TO
    SPECIFIC PATIENT COHORTS
  • Files available in
  • 5 sample (nationwide)
  • 100 sample (nationwide), only for specific
    patient cohorts
  • Pre-defined cohorts
  • Records carrying specific diagnostic/procedure
    codes

23
Part D files
  • Medicare ID
  • DOB, Gender
  • Paid date
  • Service Provider
  • Compound code
  • Quantity dispensed
  • Days supply
  • Catastrophic coverage code
  • Out-of-pocket payment

24
Claims data analysis
25
Consider origin of claims data
  • Derived from reimbursement or PAYMENT OF BILLS
  • Data elements needed to pay the bill will be of
    higher quality
  • COMPLETENESS
  • ACCURACY (?)

26
What is recorded in claims data?
  • Conditions that are DIAGNOSED
  • Care RECEIVED (care needed but not received will
    not be recorded in claims data)
  • Services that are covered
  • Prescriptions filled, rather than written
  • Services that are billed for by the provider
  • ? Flu shots provided through grocery stores NOT
    recorded in claims data

27
What is recorded in claims data?, contd
  • Clinical information ? limited
  • No data on physiology (e.g., vital signs)
  • Test results NOT included
  • Exact timing (of events) not included

28
Data Quality
  • DOES A DATA ELEMENT IMPACT PAYMENT?
  • YES ?
  • Better quality (?)
  • Consistently recorded
  • Over-coding (?)
  • NO ?
  • Quality (?)
  • Consistently recorded (?)
  • Also consider
  • Source of the data (Provider, fiscal
    intermediary, CMS)
  • whether a field is required, and/or validated or
    edited by fiscal intermediaries

29
Working from population-based files
  • Define population who is your target
    population?
  • All persons in the numerator (events) must be
    eligible to be in the denominator
  • All persons in the denominator must be eligible
    to have the event
  • Consider newly eligible people those who have
    died those who move in/out of an area
  • Diagnostic criteria
  • Demographics
  • Specific coverage (Parts A, B, managed care)
  • Benefits program (elderly disabled ESRD)
  • Building cohort (Caution consistency of IDs
    across time/files)

30
Working from population-based files, contd
  • Define outcomes of interest (mortality,
    readmissions, visits, procedures)
  • Duration of time until outcome of interest
    (Survival)
  • Covariates
  • Demographics
  • Residence
  • comorbidities

31
Accuracy of claims data
  • Many studies to assess the accuracy of Medicare
    claims data
  • Comparisons between Medicare claims and
  • Medical records
  • SEER
  • State cancer registry
  • Sensitivity, positive predictive value differing
    by diagnosis/procedures
  • Higher sensitivity when including Part B data

32
Applications
  • Racial disparities (surgery for CR cancer
    cardiovasc. procedures )
  • disease prevalence (lung cancer)
  • Identification of beneficiaries with certain
    clinical conditions
  • Geographic variations in treatment/outcome
    patterns
  • Volume-outcome studies

33
Websites of interest
  • ResDAC Research Data Assistance Center
    http//www.resdac.umn.edu/
  • CMS -- Center for Medicare and Medicaid Services
    www.cms.gov

34
AN APPLICATION USING MEDICARE ENROLLMENT AND
CLAIMS FILES
35
Colorectal cancer screening in the Medicare
fee-for-service population does spillover from
managed care matter?
Koroukian SM, Litaker DL, Dor A, Cooper G.
Medical Care 2005 May43(5)445-52
36
Background
  • Health care access and service use may vary
    according to the level of managed care activity
    (MCA) in an area, affecting the care received by
    the uninsured and those with other forms of
    insurance coverage.
  • ? Medicare expenditures in fee-for-service (FFS)
    beneficiaries are lower in high MCA areas (Baker
    L) ? SPILLOVER EFFECT

37
Background, contd
  • The degree to which managed care influences
    preventive service delivery is unclear ?
    implications for the success of screening and
    early detection programs, especially among higher
    risk groups.

38
Background, contd
  • Managed care programs have favored receipt of
    preventive services. This is evidenced through
    the services covered in the benefits package.
  • the Medicare program began reimbursement in
    January 1998 for screening fecal occult blood
    testing (FOBT) and flexible sigmoidoscopy (FLEX)
    in average risk beneficiaries and screening
    colonoscopy (COL) in high-risk individuals.

39
Study Objective
  • To study colorectal cancer (CRC) screening among
    Medicare fee-for-service (FFS) beneficiaries in
    relation to the level of MCA in their county of
    residence.

40
Depiction of the spillover effect
FFS beneficiaries
Managed care enrollees
41
Conceptual Framework
  • Individual-level characteristics
  • Age
  • Race
  • Sex

COLORECTAL CANCER SCREENING
  • County-level socioeconomic attributes
  • Poverty
  • Education

?
REIMBURSEMENT POLICY
  • County-level Health Systems Characteristics
  • Availability of physician resources
  • Primary Care Physicians (PCP)
  • Specialists
  • Proportion of PCPs to total physician workforce
  • MANAGED CARE ACTIVITY

42
Methods
  • Cross-sectional study using
  • 1999 Medicare Denominator file
  • 1999 Outpatient Standard Analytic File (SAF)
  • 1999 Part B Physician/Supplier SAF
  • 1998 Area Resource File

43
Area Resource File (ARF)
  • County-level characteristics
  • Socioeconomic attributes
  • Poverty
  • Education
  • Physician resources
  • Primary Care physicians or PCPs (per 100,000
    residents)
  • Gastroenterologists (per 100,000 residents)
  • Percent PCPs to total physicians

44
Main Outcome Measures
  • Screening for colorectal cancer by one of three
    screening modalities
  • Fecal occult blood test (FOBT)
  • Flexible sigmoidoscopy (FLEX)
  • Colonoscopy (COL)
  • We also accounted for patients undergoing
    screening tests through more than one modality
  • Colonoscopy only (COL-ONLY)
  • Colonoscopy following FOBT and/or FLEX (COL-WFF)

45
Algorithm to identify colorectal cancer screening
procedures
A MULTI-STEP APPROACH
Identify colorectal procedures (FOBT, Flex, COL)
Surveillance test
YES
Procedure performed in the presence of symptoms?
Procedure performed in the presence of diagnostic
codes indicating the presence of personal or
family history of colon cancer?
NO
YES
Screening test
Diagnostic test
NO
For example, abdominal pain bleeding change
in bowel habits
46
Covariates
  • Main covariate Managed Care Activity (MCA) at
    the county-level, derived from total months of
    insurance under managed care (MC) and FFS as
    follows
  • MCA categorized as
  • Low MCA lt 10
  • Moderate MCA 10-30
  • High MCA gt 30

MCA MC months / (MC months FFS months)
47
Other Covariates
  • Individual-level
  • Age (5 age groups 65-69 70-74 75-79 80-84
    85)
  • Race (Caucasian Afr-American (AA) Other)
  • Sex
  • County-level
  • individuals age 65 with incomes at or below
    100 of the federal poverty level
  • adults with high school diploma
  • Availability of physician resources

48
Analysis
  • The analytic files were structured as follows
  • At the individual level one record per
    individual, including demographics, and
    dichotomous variables indicating whether the
    individual underwent screening procedure (by
    modality)
  • At the county level one record per county, with
    socio-demographic attributes (poverty,
    education) physician resource availability and
    MCA

49
Analysis, contd
  • Testing bivariate associations using chi-square
  • Multi-level logistic models to assess the
    likelihood of undergoing CRC screening procedure
    given the MCA level, after controlling for
    individual- and county-level characteristics
  • Multi-level models made it possible to account
    for clustering of screening activities within
    counties
  • Used the individual as the unit of analysis
  • Software packages SAS 8.12, HLM, ArcView

50
PERCENTILE DISTRIBUTION OF HMO PENETRATION RATE
(unit of analysis county)
51
FOBT Rates, by County (Adj by age, race, sex)
52
FLEX Rates, by County (Adj by age, race, sex)
53
COL Rates, by County (Adj by age, race, sex)
54
(No Transcript)
55
Distribution of Medicare beneficiaries by
individual characteristics and county-level MCA
( of total)
a 0.01 lt p 0.05 b 0.001 lt p 0.01 c p
0.001 all other p values gt 0.05 NOTE p-values
in the Moderate column refer to comparisons in
statistics between counties with low and moderate
MCA p-values in the High column refer to
comparisons in statistics between counties with
moderate and high MCA.
56
County characteristics by MCA level
a 0.01 lt p 0.05 b 0.001 lt p 0.01 c p
0.001 all other p values gt 0.05 NOTE p-values
in the Moderate column refer to comparisons in
statistics between counties with low and moderate
MCA p-values in the High column refer to
comparisons in statistics between counties with
moderate and high MCA.
57
Results from the multivariable analysis
predicting individuals use of CRC screening test
a 0.01 lt p 0.05 b 0.001 lt p 0.01 c p
0.001 all other statistics not significant at p
lt 0.05
58
Estimated Effect of an Increase in Managed Care
Activity (MCA) on Colorectal Cancer Screening for
10,568,067 Medicare Beneficiaries Residing in
Areas of Low MCA
a 0.01 lt p 0.05 b 0.001 lt p 0.01 c p
0.001 all other statistics not significant at p
lt 0.05
59
Principal findings
  • Positive association between county-level MCA and
    different forms of CRC screening among Medicare
    FFS beneficiaries
  • LEVELS ARE OF MODEST PUBLIC HEALTH IMPORTANCE
  • No consistent relationship between higher levels
    of MCA and increased likelihood of CRC screening

60
Strengths
  • Unique database combining Medicare claims
    denominator and claims files with the ARF
  • Use of both outpatient SAF and physician/supplier
    file to obtain a more complete account of
    screening services
  • High statistical power
  • Robust rate estimates
  • Methodological considerations
  • Unduplicated services across the outpatient SAF
    and physician/supplier file to avoid
    doublecounting

61
Limitations
  • Unable to account for differential in health
    status across regions with varying degrees of MCA
  • Managed care plans may be more likely to enroll
    healthier beneficiaries (cherry picking)
  • The extent to which these findings translate into
    actual improvements in CRC-related outcomes among
    FFS beneficiaries cannot be inferred from this
    study

62
Limitations, contd
  • Over-simplification of the managed care concept
    not accounting for the managedness of specific
    programs
  • Our algorithm to identify screening procedures
    (vs diagnostic or surveillance procedures) has
    not yet been tested for validity and reliability
    ? misclassification of procedures in/out of the
    screening category may have occurred.

63
Colorectal cancer screening in the elderly
population Disparities by dual Medicare-Medicaid
enrollment status
  • Koroukian SM, Xu F, Dor A, Cooper GS
  • Health Serv Res. 2006 Dec41(6)2136-54.

64
Background
  • Dually eligible Medicare-Medicaid beneficiaries
    represent the oldest, poorest, and frailest of
    the elderly population
  • 73 of duals have annual incomes of 10,000 or
    less, only 12 of non-duals may be identified
    with such income levels.
  • Duals are considerably more likely than
    non-duals
  • to rate their health as poor or fair (52 vs.
    26, respectively),
  • to present significant limitations in activities
    of daily living (over 30 vs. 11),
  • to reside in nursing homes (19 vs. 3)

65
Background, contd
  • On the other hand the above statistics imply
    that
  • nearly 50 rate their health as good, very good,
    or excellent
  • two thirds report no limitations in their
    activities of daily living
  • and 80 are community-dwelling
  • ? suggesting that a sizable proportion of duals
    may in fact benefit from preventive care.
  • Also, while most duals have low incomes,
    enrollment in Medicaid is likely to give them
    considerable leverage to access health services
    that they could not afford otherwise.

66
Objective
  • To assess the disparities in colorectal cancer
    (CRC) screening between elderly dual
    Medicare-Medicaid enrollees (or duals), the most
    vulnerable subgroup of the Medicare population,
    and non-duals.

67
Distribution of Duals and Non-Duals by
Demographic attributes
68
Proportion () of Duals and non-Duals undergoing
colorectal cancer screening procedures
Age- race- sex-adjusted rates of screening,
using direct risk adjustment method
69
Conclusions
  • Duals are significantly less likely than
    non-duals to undergo CRC screening, even after
    adjusting for individual- and county-level
    covariates. Future studies should evaluate the
    contribution of comorbidity and low socioeconomic
    status to these disparities.

70
Limitations
  • Simplistic way to account for dual status
  • Duals do not represent all low-income Medicare
    beneficiaries
  • A substantial proportion of non-duals may have
    low incomes
  • Inability to account for differential in
    comorbid/health status between duals and
    non-duals
  • Algorithms to account for screening/diagnostic/sur
    veillance procedures

71
Point of discussion
  • Comparing rates of colorectal cancer screening
    between survey data and claims data

72
Topics to follow in other sessions
  • More applications (especially with SEER-Medicare
    files)
  • Different approaches in identifying comorbid
    conditions from claims data
  • Logistics to request Medicare data

73
(No Transcript)
74
Deriving probability from odds ratios
  • Odds Exp coefficient p / (1-p)

Prob odds / (1 odds) Prob from log odds 1
/ (1 exp(-coeff))
exp(coeff) / (1
exp(coeff))
75
Additional details on how estimates in Table 3
were derived
  • Increasing from low to moderate MCA
    (n10,568,067)
  • Baseline screening in low MCA counties 10.25 ?
    odds of being screened 0.1025/(1-0.1025)0.1142
  • Odds ratio1.05 ? new odds in low MCA areas
    0.11421.050.1199
  • Prob0.1199/(10.1199)10.71
  • Additional of individuals in low MCA areas who
    would be screened
  • 10,568,067 (0.1071-0.1025) 49,000
About PowerShow.com