Title: NAHDOCDC COOPERATIVE AGREEMENT
1NAHDO-CDC Cooperative Agreement Webinar -
Medicaid Data An Essential Public Health Data Set
September 22, 2004
2Overview
- Significance of Medicaid Data to Public Health
- Denise Love, NAHDO
- Medicaid Data vs. Public Health Data Sets
- Jeffrey Geppert, Stanford University
- Applied Lessons
- Nevada, Christopher Thompson
- Utah, Wu Xu
3Significance Medicaid Data and Public Health
- Medicaid provides coverage to over 44 million
Americans - Jointly funded by federal and state governments
- Second largest and fastest growing component of
expenditures (general funds and all spending
sources) (NASBO 2003) - The largest source of health care financing for
people with disabilities (NHIS)
4Significance Medicaid Data and Public Health
- Provide primary care and other services to
Medicaid beneficiaries - Change in safety net role in respond to increase
enrollment in Medicaid managed care (e.g., Early
and Periodic Screening, Diagnosis, and Treatment
Program (EPSDT) services) - Emphasis on disease prevention and health
promotion - Developing partnerships with Medicaid
5Medicaid Data Sharing and Integration
- Medicaid provides data about an important
population - Data requests must fit within HIPAA exemptions to
circumvent requirements for individual
authorization - Medicaid and public health need to coordinate and
evaluate their joint and respective interests and
requirements to make data sharing work for them
and the public interest
6Permission Encouragement for Data Partnerships
- Interagency agreement between HRSA, CDC and then
HCFA, issued October 1998. - To facilitate collaborations for intrastate data
sharing between State Medicaid and State Health
Agencies. - Includes model agreement that covers all Federal
Medicaid requirements.
7Permission Encouragement for Data Partnerships
- Change especially for Medicaid from use of
Medicaid data only for direct administrative
purposes of Medicaid and only by Medicaid staff.
8Potential Barriers to Data Partnerships
- Confidentiality, Privacy, Security concerns
regarding linkage of existing data internally
between and within programs of same states
public health and health services, and sometimes
social services. - Institutional Inertia Most but not all recognize
change from prior Medicaid specific policy,
regarding administrative use only.
9Potential Barriers to Data Partnerships
- May have state laws, regulations or policies -
dating from original Federal guidance - that
still slow or limit ability to link datasets,
even though Federal guidance has changed. - Confusion around HIPAA
-
10Standards for Privacy of Individually
Identifiable Health Information (45 CFR Parts 160
and 164 Sec. 164.512)
- PHI may be disclosed by a covered entity without
consent or authorization - To a public health authority authorized by
law..for purposes of preventing, controlling
disease, injury, disability, vital events ...and
conduct of public health surveillance and
investigations
11Technical Capacity Issues
- Knowledge of structure and content of databases
to be linkedMedicaid is often many separate
files. - Knowledge of content data clinical coding,
programmatic variables. - The availability of analytic software and ability
to use it properly to make matches. - Data Warehouse approach is successful in some
states (alternative to dataset-by-dataset,
case-by-case linkage)
12Medicaid program questions
- Utilization of health care services
- Pre- and post-natal care
- Ambulatory care sensitive conditions
- Preventive services
- Among beneficiaries with chronic and acute
conditions - FFS and managed care
13Medicaid program questions
- Health promotion and disease prevention
- Prevalence of diabetes, asthma
- Diagnosis and treatment of STDs, TB and HIV
- Disease surveillance and environmental health
- Diagnosis and treatment of mental health and
alcohol/drug abuse
14Medicaid program questions
- Program Evaluation
- Impact of public health nurses on utilization
among foster care children - Home visitation and education programs
- Utilization in areas with community and
school-based public health clinics
15Medicaid data differ from public health data sets
- Advantages
- Wide range of conditions
- Treatment in addition to diagnosis
- Variations in intensity and cost
- Access to care
- Outcomes of care (morbidity and mortality)
- Long-term outcomes (5-year, 10-year)
16Medicaid data differ from public health data sets
- Advantages (continued)
- Samples versus the universe
- Self-report versus objective
- Longitudinal versus panel
- Trends over time
- Impact of change in policy on individuals rather
than populations - Exogenous sources of variation
17Medicaid data differ from public health data sets
- Claims data vs. event data
- Requests for payment from providers for
inpatient, physician, pharmacy, etc. - Separate files and formats
- Data linkage
- Claims and individual line items
- ICD-9-CM diagnosis and procedure, CPT, NDC coding
18Medicaid data differ from public health data sets
- Rules for constructing an event
- Claims and visits
- Screening and conditions
- Inclusions and exclusions
- Episodes of care
- Prevalence versus incidence
- Censoring
- Health care home
19Medicaid data differ from public health data sets
- Managed care and carve out services result in
fragmented data - Eligibility versus enrollment
- Enrollment and disenrollment
- Managed care enrollment
- Carve-out for mental health, pharmacy, dental,
etc. - Changing eligibility
20Medicaid data differ from public health data sets
- Historical data may not reside in the data
warehouse - Enrollment
- Eligibility category
- County of residence
- Managed care or fee-for-service
- Health care home
21Case Study Prenatal Outcomes in Nevada
- Center for Health Information Analysis,
University of Nevada Las Vegas (CHIA) - Data sources Medicaid paid claims, Medicaid
eligibility, managed care encounter and case
management/risk assessment data for pregnant
women
22Case Study Prenatal Outcomes in Nevada
- Project aims
- Cost of care in 1Y by mother's length of
enrollment in Medicaid and amount of prenatal
care received - Effectiveness of the initial prenatal screening
assessment in identifying high-risk pregnancies - Compare pregnancy outcomes (e.g. newborn costs
and LOS) for Medicaid births to those of the
general population - Effectiveness of FFS and MCO for high-risk
pregnancies
23Case Study Prenatal Outcomes in Nevada
- Preliminary Findings
- Lower newborn costs associated with managed care
enrollment, length of Medicaid enrollment and
prenatal risk assessment - Newborns born to mothers enrolled less than one
month prior to giving birth were nearly 3 times
as likely to be high cost - Nearly linear relationship between the length of
enrollment (i.e. available for prenatal care) and
newborn cost - When controlling for length of enrollment,
newborn costs were slightly higher for mothers in
managed care. The differences were generally
small and may be due to the inclusion of
capitated payments rather than costs based on
encounters
24Case Study Prenatal Outcomes in Nevada
- Challenges
- Linking mother records to newborn records (i.e.
family versus individual identifiers) - 3 of the newborns account for 43 of the costs
(1 of the newborns account for 25 of the costs) - Lack of managed care encounter data (under 25k)
for outcomes and costs - Separately identifying the impact of Medicaid
enrollment versus risk assessment
25Utah Example 1 Using Medicaid Data for MCH
Reporting
- Questions Medicaid low birth weight rate, etc.
- Process
- Data sharing agreement (1997)
- Linking Medicaid eligibility to birth
certificates (1998-1999) - MCH defined its business needs in Medicaid Data
Warehouse (2000) - MCH analysts access Medicaid DW (2001)
26Utah Example 2 Using Population Data to Support
Medicaid Program Evaluation
- Questions Has the PCN 1115 waiver improved its
enrollees health status? Impact on statewide
uncompensated care? - Process
- Designed the evaluation with Medicaid input to
address their administrative needs - Used general population data as reference
- Provided timely feedback to the PCN program. PCN
timely adjusted their procedures and strategies.
27Summary
- Medicaid goals may be different than the public
health entity goals - More program-oriented
- Data access may be tied to benefit to Medicaid
program - Outcome may not be a public report, but instead
an internal discussion
28Summary
- Medicaid data differ from public health data
sets - Claims data vs. event data
- Rules for constructing an event
- Managed care and carve out services result in
fragmented data - Historical data may not reside in the data
warehouse
29Summary
- Medicaid and ALL of its stakeholders must be
involved at all stages - Policy/outcome question development
- Research design
- Interpretation of results
- Implementation of interventions
- Public health programs are stakeholders (chronic
disease, injury, MCH, epidemiology, health
planning)
30Summary
- Building trust between Medicaid and other state
agencies essential for continued data access - State agencies should employ data sharing
agreements to guide permissible data sharing and
dissemination of Medicaid data (CMS-HRSA-CDC
model)
31Benefits to sharing Medicaid data (HRSA News
Brief 1999)
- Encourage development of integrated state
information systems - Assure appropriate, accessible, cost-effective
care for underserved/vulnerable populations - Improve the technical capacity of states to
analyze data from multiple sources for policy
support/program evaluation - Promote measurement using common performance
measures across programs - Better use of Medicaid encounter data to assist
in public health surveillance and improve care
for Medicaid enrollees
32Resources
- Federal Data Sharing Agreement
- http//www.hcfa.gov/medicaid/smd10228.htm
- Medicaid HIPAA Page
- http//www.cms.hhs.gov/medicaid/hipaa/adminsim/
- Medicaid Primer, Public Health Foundation
- http//www.phf.org/Medicaid_primer.pdf
- Medicaid Managed Care Statistics (CMS)
- http//www.cms.hhs.gov/medicaid/mcaidsad.asp