Moving Medicaid Data Forward: Medicaid Enrollment Overview and Data Sources A Mathematica Policy Research Forum Washington, DC February 7, 2017 Craig Thornton Maggie Colby Robin Rudowitz Thomas DeLeire
Welcome Craig Thornton Senior Fellow 2
Moving Medicaid Data Forward Medicaid as a whole is the largest health insurance program in the United States 74 million current beneficiaries More than half of Americans will be covered by Medicaid at some point in their lives Total Medicaid expenditures exceeded $532 billion in 2015 Medicaid accounts for the largest single item in many state budgets Substantial policy interest in pursuing new options 3
Moving Medicaid Data Forward 1. Understanding T-MSIS (October 2016) History Current functionality How should CMS make T-MSIS data available to the research community? 2. Medicaid Enrollment 4
Today s Panelists Presenters Robin Rudowitz Kaiser Family Foundation Maggie Colby, Mathematica Discussant Thomas DeLeire, Georgetown University 5
Moving Medicaid Data Forward, Forum 2: Medicaid Enrollment -- Overview and Data Sources Robin Rudowitz, Associate Director for Program for Medicaid and the Uninsured Kaiser Family Foundation February 7, 2017
Figure 7 Before we talk about enrollment, it is important to understand what the minimum eligibility standards are. ACA established minimum eligibility standards for adults, but the Supreme Court ruling effectively made these levels optional for states. 138% 138% 138% 138% 74% 25% Children Pregnant Women Parents Other Adults Seniors and People with Disabilities (tied to SSI Coverage) NOTE: Parent minimums vary across states; median minimum shown. 138% FPL is $16,394 for an individual and $27,821 for a family of three in 2016. 0%
Figure 8 All states have taken up options to expand eligibility beyond federal minimum standards. Number of States: 51 49 44 44 32 33 21 Cover Children at or above 138% FPL Cover Pregnant Women at or above 138% FPL Cover Adults at or above 138% FPL Seniors and People with Disabilities >75-100% FPL Working People Medically Needy with Disabilities Coverage Buy-in Allow People in Need of Long- Term Care to Qualify up to 300% SSI SOURCE: Based on results from a national survey conducted by the Kaiser Commission on Medicaid and the Uninsured and the Georgetown University Center for Children and Families, 2017 and the Kaiser Commission on Medicaid and the Uninsured Medicaid Financial Eligibility Survey for Seniors and People with Disabilities, 2015.
Figure 9 Due to state flexibility and the effective option of the ACA Medicaid expansion, median Medicaid eligibility levels for adults varies across expansion and non-expansion states. 138% 138% 2013 2017 90% 48% 44% 0% 0% 0% Parents Other Adults Parents Other Adults States that Have Implemented the Medicaid Expansion as of Jan. 2017 (32 states, including DC) States that Have Not Implemented the Medicaid Expansion as of Jan. 2016 (19 states) NOTE: 2017 levels are based on state-reported eligibility levels as of Jan. 2017. Levels are based on 2016 federal poverty levels for parents in a family of three, which is $20,160, and individual childless adults, which is $11,880. SOURCE: Based on results of a national survey conducted by the KCMU and the Georgetown Center for Children and Families, 2013 and 2017.
Figure 10 The ACA modernizes the Medicaid application and enrollment experience in all states. PAST Apply in person Multiple options to apply ACA Vision No Wrong Door to Coverage Provide paper documentation Electronic verification $ Data Hub # Medicaid CHIP Marketplace Wait for eligibility determination Real-time determination Dear, You are eligible for
Figure 11 How many people are covered by Medicaid????? Seems like a simple question, but hard to answer! Need to know how someone is trying to use the information: Trends over time? Effects of expansion? Medicaid as a share of overall coverage? Are we talking about point in time or ever enrolled in the year? Full or partial benefit enrollees? What question you are trying to answer determines which number / data source is best.
Figure 12 A few sources contain Medicaid enrollment data, but each has benefits and challenges. Data Source Latest Available Key Benefits MSIS 2011 Data by eligibility group Can link enrollment and spending HMA / KFF Enrollment Snapshots CMS Performance Metric Data December. June 2013 November 2016 MBES Data 2016, Quarter 1 Survey Data (CPS, NHIS, ACS) 2015 (CPS & ACS), 2016 (NHIS, some quarters) More recent than MSIS Data by eligibility group Trend data over time Updated monthly Breaks out data for kids Shows increases from pre- ACA baseline (summer 2013) Data for Group VIII (expansion group) Linked to claims data Puts Medicaid in context with other coverage Challenges Latest available data is from 2011 Not all data publicly available No longer collected Cannot separate Medicaid / CHIP No detail by eligibility group No trend data before baseline Inability to trend prior to the ACA No characteristics Medicaid undercount Need analytic capacity to use data
Figure 13 The double bars show how Medicaid enrollment compares to spending. Disabled 15% Elderly 9% Disabled 42% Adults 27% Elderly 21% Children 48% Adults 15% Children 21% Enrollees Total = 68.0 Million Expenditures Total = $397.6 Billion SOURCE: KCMU/Urban Institute estimates based on data from FY 2011 MSIS and CMS-64. MSIS FY 2010 data were used for FL, KS, ME, MD, MT, NM, NJ, OK, TX, and UT, but adjusted to 2011 CMS-64.
Figure 14 Trend data is useful to understand the implications of economic downturns and policy changes. Annual Percentage Changes, FY 1998 FY 2017 Total Medicaid Spending 12.7% Medicaid Enrollment 13.2% 4.7% 10.4% 8.7% 6.8% 7.5% 3.2% 0.4% 9.3% 8.5% 5.6% 7.7% 6.4% 4.3% 3.2% 1.3% 3.8% 5.8% 7.6% 6.6% 3.1% 7.8% 7.2% 9.7% 4.8% 2.3% 3.2% 1.5% 6.8% 10.5% 5.3% 5.9% 4.5% 3.9% 3.3% 0.2% -0.5% -1.9% -4.0% 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Proj. NOTE: For FY 1998-2013, enrollment percentage changes are from June to June of each year. FY 2014-2016 reflects growth in average monthly enrollment. Spending growth percentages refer to state fiscal year. FY 2017 data are projections based on enacted budgets. SOURCE: Enrollment growth rates for FY 1998-2013 are as reported in Medicaid Enrollment June 2013 Data Snapshot, KCMU, January 2014. FY 2014-2016 are based on KCMU analysis of CMS, Medicaid & CHIP Monthly Applications, Eligibility Determinations, and Enrollment Reports, accessed October 2016. Historic Medicaid spending growth rates are derived from KCMU Analysis of CMS Form 64 Data. FY 2016-2017 data are derived from the KCMU survey of Medicaid officials in 50 states and DC conducted by Health Management Associates, October 2016.
Figure 15 The performance metric data is useful in understanding how enrollment has changed since the implementation of the Affordable Care Act (ACA). Monthly Medicaid/CHIP enrollment (in millions) 56.4 74.4 Pre-ACA (2013) November 2016 NOTE: Data not available for CT and ME. Summer 2013 baseline enrollment data based on monthly average for Jul Sept. 2013. Expansion status groupings based on states with expansions in effect. SOURCE: CMS, Medicaid & CHIP: Monthly Application and Eligibility Reports, October 2013 November 2016.
Figure 16 Most of the growth in Medicaid enrollment was in expansion states and the West from Summer 2013 to November 2016. Among 49 states reporting data for both periods: Expansion Status Region All Non- Expansion States 18% Other Expansion States 40% CA 26% NY 4% WA 4% OH KY 4% 4% Medicaid Expansion in Effect 82% Midwest 14% Northeast 13% South 29% West 44% Total Net Change in Medicaid/CHIP Enrollment= 17 Million NOTE: Data not available for CT and ME. Summer 2013 baseline enrollment data based on monthly average for Jul Sept. 2013. Expansion status groupings based on states with expansions in effect. SOURCE: CMS, Medicaid & CHIP: Monthly Application and Eligibility Reports, October 2013 November 2016.
Figure 17 The MBES data shows the distribution of Medicaid enrollees across traditional and expansion groups. Medicaid Enrollment, January-March 2016 Traditional Medicaid 80% Expansion Group - Newly Eligible 15% Expansion Group - Not Newly Eligible 5% Total Enrollment = 73.4 Million Enrollees NOTES: Data for January through March 2016 for 30 states that implemented the Medicaid expansion as of January 2016 (Louisiana expanded Medicaid in on 7/1/16 and has no data reported. There is no data reported for North Dakota. Enrollment data reflect the highest enrollment for each state during the quarter. SOURCE: KCMU analysis of data from Medicaid Budget and Expenditure System (MBES).
Figure 18 Survey data can show how Medicaid fits into the overall coverage picture. Other Public 2% Uninsured 9% Medicare 14% Employer 48% Medicaid/ CHIP 20% Non-Group 7% SOURCE: Health insurance coverage: KCMU analysis of 2015 data from the 2016 ASEC Supplement to the CPS.
Figure 19 Looking ahead: Congress will be debating major policy changes that will affect Medicaid and data can help inform the debate: ACA Repeal / Replacement Financing Changes (Block Grant or Per Capita Cap) Estimating the impact of these changes and then tracking the impact going forward depends on data CMS has been collecting new data through T-MSIS, but it is not yet publicly available Continued collection of Medicaid data is a collaboration between states and the federal government
Data Sources for Medicaid Enrollment Research: What Is Available, Needed, and Hoped for? Moving Medicaid Data Forward: Medicaid Enrollment Overview and Data Sources February 7, 2017 Maggie Colby Associate Director of Health Research and Senior Researcher
Sources for Medicaid Enrollment Estimates National survey data American Community Survey (ACS) National Health Interview Survey (NHIS) and Medical Expenditure Panel Survey (MEPS) Aggregate-level administrative data Medicaid and Children s Health Insurance Program (CHIP) Performance Indicators Medicaid Budget and Expenditure System (MBES) Individual-level administrative data Medicaid Statistical Information System (MSIS)/Medicaid Analytic extract (MAX) Transformed MSIS 21
National Survey Data 22
National Surveys Advantages Can be used to estimate eligible populations and take-up rates Breadth of survey topics allows researchers to consider intersection of Medicaid enrollment with other demographic, household, health status, and health expenditure patterns Limitations Medicaid participation (as in most other social programs) is significantly undercounted Undercount varies by survey generally 20% to 40% (see summary in Boudreaux et al. 2013) Undercount research predates 2014 Medicaid expansions have people gotten more or less accurate in answering these questions? Timeliness release is generally delayed at least a year 23
American Community Survey Fielded by the Census Bureau Now available through 2015 Minnesota Population Center at the University of Minnesota releases Integrated Public Use Microdata Series, which facilitates research that requires understanding family relationships Health insurance question added in 2008 Current health insurance status Can produce statistically reliable single-year health insurance estimates at the state level and at some substate levels Twelve independent monthly samples, so estimates are roughly an average for the full year Researchers found 76% correct reporting of Medicaid enrollment for 2008 linked MSIS/ACS data (66% to 88% by state) Uninsured and employer-sponsored insurance about as likely as incorrect responses 24
National Health Interview Survey (NHIS) Fielded by the Centers for Disease Control and Prevention Currently available through 2015 Assesses coverage status at time of interview and length of time without coverage for noncovered household members Current questions introduced in 1997 Questionnaire revisions planned for 2018 Cannot provide annual state-level data, but some state-level estimates can be obtained by combining years Can also combine states to produce annual estimates (e.g., expansion vs. non-expansion; see Cohen and Martinez 2014) 25
Medical Expenditure Panel Survey (MEPS) Fielded by the Agency for Healthcare Research and Quality Household component collects data from subsample of participants in the prior year s NHIS Smallest geographic region available is census region (Northeast, Midwest, South, West) Panel design with several rounds of interviewing covering two full calendar years Unique ability to track what happens as panel members exit Medicaid Contains detailed information on duration of insurance coverage and uninsurance status Can be linked with detailed medical expenditure data 26
Aggregate-Level Administrative Data 27
Performance Indicators and MBES Advantages Timely and easy to use for broad monitoring of trends Enrollment data from Medicaid and CHIP Performance Indicator available monthly at the state level through November 2016 Enrollment data from the Medicaid Budget and Expenditure System (MBES) available quarterly at the state level through Q12016 Only MBES provides data on the number of Group VIII Newly Eligible enrollments (i.e., expansion adults) Limitations No ability to unpack the counts Cannot fully reconcile MBES and performance indicator data given different methodologies Treatment of limited-benefit enrollees Ever-enrolled vs. point-in-time 28
Individual-Level Administrative Data 29
MSIS and MAX Advantages Both contain universe of individuals eligible for Medicaid for at least one day Researchers have flexibility to examine different subgroups of beneficiaries and track them over time MAX is a research-ready version of MSIS Final-action claims, reconciled to the calendar year the services took place Limitations Demographic variables relatively limited Standard eligibility categories not designed to track Medicaid expansion population or other key policy-relevant groups (e.g., pregnant women) No insight into what happens to beneficiaries during periods when they are not Medicaid eligible Timeliness states submitted MSIS quarterly; MAX required seven quarters of MSIS data As of July 2016, MAX 2013 data available for 20 states As of January 2017, MAX 2014 data available for11 states 30
T-MSIS: New Possibilities Monthly submissions improved timeliness seems possible New variables available on beneficiaries Demographic Veteran status, marital status, language preferences, family income, household size Eligibility Pregnancy, citizenship and immigration, disability and chronic conditions, SSI status, level-of-care status, reasons for eligibility change New files collected from states Managed care, provider, third-party liability Opportunity to understand how different enrollees get connected to plans and providers 31
With Some Caveats Work remains to understand data quality, expected ranges, and fluctuations over time for new data elements in T-MSIS Usability is likely to vary by state at first Researchers should proceed cautiously with new data elements New T-MSIS data formats may also affect old data elements available in MSIS 32
So We ve Talked About What s Available What s Needed and Hoped for? 33
My Researcher Wish List (1) Retain eligibility hierarchy in administrative data What are all the ways someone could be eligible, besides their primary basis of eligibility? Would help model the impact of policy alternatives Promptly update aggregate reporting requirements as major new policy initiatives are implemented Avoid 2014 challenge of estimating new eligibility vs. welcome-mat effects following Medicaid expansion 34
My Researcher Wish List (2) Update survey validation research to understand: Magnitude of the Medicaid undercount post-expansion Potential sources of cross-state variation in the undercount Is there more confusion in states that: Use managed care organizations to deliver care? Enroll Medicaid beneficiaries in qualified health plans offered through the Marketplaces? Support Medicaid beneficiaries enrollment in employer-sponsored insurance? Brand their Medicaid programs using language that does not include the word Medicaid? 35
For Discussion: What s on Your Wish List? 36
Discussant Thomas DeLeire Georgetown University 37
Questions? For online and in-person attendees Please state your name, affiliation, and whether your question is directed at a specific panelist Webinar audience can submit questions for our speakers now using the Q&A widget Please fill out your feedback card before you leave 38
Mark your calendars! The Moving Medicaid Data Forward Forum Series Forum 3: June 8, 2017 (topic: TBD) Forum 4: October 12, 2017 (topic: TBD) 39
Networking Reception Starts Now Mathematica Lobby, 12 th Floor 4:30 5:30 p.m.