Picture of Housing and Health: Medicare and Medicaid Use Among Older Adults in HUD-Assisted Housing
Executive Summary
The Lewin Group
March 2014
Overview
For the U.S. Departments of Health and Human Services (HHS) and Housing and Urban Development (HUD), the Lewin Group and its sub-contractors, Leading Age and the Moran Company, explored the potential for publicly-subsidized senior housing to serve as a platform for efficiently managing the population health of low-income older adults with various levels of physical and mental health risk. We developed a demonstration design and piloted creating a new dataset linking HUD tenant and Medicare/Medicaid claims data in 12 jurisdictions across the country.
Study Objective
This study task explored the feasibility of matching HUD administrative data to the HHS Centers for Medicare and Medicaid Services (CMS) administrative data in order to determine the extent to which this resource could track health and housing outcomes, and whether this approach could reliably support future research and policy analysis.
Data
We focused on 2008 individual-level administrative data provided for the 12 geographic areas from both HUD and CMS. The HUD multifamily and public housing data came from the Tenant Rental Assistance Certification System and the Public and Indian Housing Information Center. Medicare administrative data came from the Medicare Beneficiary Summary File and includes Medicare Parts A, B, and D enrollment, payments, and utilization, as well as information about chronic conditions. The Medicaid Analytic eXtract Person Summary file, based on state submission of Medicaid administrative data, provided enrollment, payment and utilization for Medicaid-covered services.
Study Objectives and Results
We provide a detailed description of the study results in the Summary Report and in Appendix C. The chart below provides a brief overview of the study objectives and corresponding results.
| Objectives | Results |
|---|---|
| Match Rate between HUD and CMS data | Among HUD-assisted individuals age 65 or older in 2008, 93% matched to Medicare based on Social Security number (SSN) only; 85% matched to Medicare when requiring a match on SSN, gender, and date of birth. |
| Estimate HUD-assisted Medicare beneficiaries enrollment in Part D subsidy assistance programs or Medicaid | Health insurance assistance programs for Medicare beneficiaries reduce out-of-pocket health care expenses, which result in savings for HUD. Among HUD-assisted Medicare beneficiaries age 65 years or older:
While we would not expect the proportion to be 100%, as not all HUD-assisted Medicare beneficiaries meet the eligibility requirements for these programs because the income criteria for HUD-assisted housing can be higher than that for health insurance assistance, there may be room for improvement in the enrollment rates. |
| Compare Medicare and Medicaid payment and utilization among Medicare and Medicaid enrollees (MMEs) | Most of our analyses to compare HUD-assisted individuals to unassisted individuals focused on Medicare beneficiaries dually enrolled in Medicaid (MMEs), age 65+ with fee-for-service (FFS) coverage compared to unassisted MME beneficiaries in the community in order to examine comparable groups in terms of economic status because this factor drives social determinants of health, use of services, and resulting payment. Health conditions:
Medicare payments and utilization:
Medicaid payments and utilization:
|
Discussion
This study demonstrates the feasibility and utility of linking HUD tenant data to CMS administrative data. The descriptive results summarized above highlight key areas for future analysis to better understand the health and health care utilization of HUD-assisted elderly individuals enrolled in Medicare. This includes supplementing current data sources with additional CMS data, refining matching algorithms and study samples to better determine HUD-assisted elderly individuals' eligibility and enrollment in Medicare programs, providing distributional analyses, and conducting multivariate regressions to determine if the differences observed in descriptive comparisons remain after adjusting for confounders.
| The Full Report is also available from the DALTCP website (http://aspe.hhs.gov/office_specific/daltcp.cfm) or directly at http://aspe.hhs.gov/daltcp/reports/2014/HUDpic.shtml. |