Automatic enrollment and state health reform

1
Automatic enrollment and state health reform

• Stan Dorn
• Senior Research Associate
• Urban Institute
• 202.261.5561
• sdorn_at_ui.urban.org

State Coverage Initiatives Program AcademyHealth D
enver Colorado August 3, 2007
2
Overview

1. Enrollment models
2. Data issues
3. Applying auto-enrollment to state coverage reforms

3
Preliminary topic Why enrollment matters
4
If you build it, will they come?
5
Why enrollment matters

• Necessary to accomplish the goal of coverage
  expansion
• Cost offsets with eligible but un-enrolled when
  they get sick, they will use services, and the
  state will pay
• Standard enrollment growth curve creates
  political vulnerability for example, see next
  slide

6
So one year in, we have a plan that, even if no
more concessions to liberal advocates are made,
falls 20 percent short of its stated goal.
At one year, Mass. healthcare plan falls short By
Sally C. Pipes    May 15, 2007

7
Part I Basic enrollment models
8
Traditional public benefits model

• Governments role
• Provide program information outreach
• Process applications
• Individual must
• Apply
• Provide individual information showing
  eligibility
• Complete the application process

9
Implications of traditional model

• Denies coverage to eligible people who
• Do not apply
• Do not complete the process
• It takes several years for a new program to reach
  many of its targeted beneficiaries
• High ongoing administrative costs for state
• BUT
• Familiarity means less risk, culture shock,
  uncertainty, mid-course adjustment after initial
  stumbles
• Permits covert caseload controls that lower cost
  with less risk of successful opposition
• Procedural barriers prevent waste, fraud and
  abuse
• Reduced outreach may never come to public
  attention

10
A different model Auto-enrollment

• Mechanisms
• Default enrollment
• Data-driven enrollment
• Proactively facilitated enrollment
• Promise lessening the historic tension between
  safeguarding program integrity and simplifying
  application procedures.
• More eligible people get covered
• A smaller percentage of ineligible people get
  covered
• Operational administrative costs drop (after
  infrastructure development)
• Happened with WIC and NSLP

11
Basic principle Newtons First Law of Motion

• An object at rest tends to stay at rest

12
Examples of auto-enrollment

• SCHIP vs. Medicare Part D
• Retirement savings
• Medicare Part B
• Community-based, proactive facilitation of child
  health enrollment
• Retention of health coverage in Louisiana
• Massachusetts CommCare

13
Example 1 SCHIP vs. Low-Income Subsidies (LIS)
for Medicare Part D
Effective 10/1/97
Food stamps, after 2 years 31 take-up
Source Selden, et al., 2004 (MEPS data).
14
Data-driven enrollment Medicare Part D,
low-income subsidies (LIS)

• Without application, automatically enrolled in
  drug plan, with LIS, if received Medicaid or SSI
  the prior year
• Can apply to SSA

15
Example 1, continued
Total enrollment 74
Source CMS enrollment data. Calculations by
Urban Institute.
16
Example 2 retirement savings
Sources Etheredge, 2003 EBRI, 2005 Laibson
(NBER), 2005.
17
Example 3 Medicare Part B
Sources NASI, 2006 Remler and Glied, 2003..
Note Medicare Savings Programs (MSP) help
Qualified Medicare Beneficiaries (QMB) with
income up to 100 FPL and Specified Low-Income
Beneficiaries (SLMB) with income between 101 and
120 FPL.
18
Example 4 Community-based facilitators of child
health enrollment
Source Flores, et al., Pediatrics, 12/05.
19
Example 5 Retention in Louisiana
Source Summer and Mann, Georgetown University
Health Policy Institute (prepared for
Commonwealth Fund), June 2006. Note other policy
changes included telephone contact, rather than
forms, to supplement data.
20
Example 6 Auto-enrollment in Massachusetts,
based on prior uncompensated care pool
Source Commonwealth Connector Authority, June
2007 (unpublished data).
21
Part II Data issues
22
Cross-cutting data issues

• Privacy
• Funding IT development

23
Privacy

• Practical strategy inform families, in advance,
  about information disclosure.
• Provide chance to opt out
• Builds trust
• State law changes may be needed to access data
• Safeguards of confidentiality, data security

24
Building IT infrastructure

• Enhanced FMAP via MMIS (90 for start-up, 75 for
  operations) is denied to eligibility systems,
  by federal regulation from
• MITA todays MMIS
• Add eligibility data to EHRs
• Offset with lower operating costs
• NSLP case study in MN 80 savings, net

The 1970s
25
Part III Applying Auto-Enrollment to State
Coverage Reforms
26
Where can auto-enrollment help? Three crucial
functions

• Identifying the uninsured
• Determining eligibility
• Enrollment into coverage

27
Function Number One Identifying the Uninsured

• Key life events
• Master list

28
Key life event strategy

• Its a key life event if it includes
• Many uninsured
• Existing mechanism on which to build
• Examples
• Health care visits (e.g., at hospitals, CHC/s)
• State EITC forms (if state has EITC)
• W-4 forms (wage withholding when starting work)
• Applications for unemployment insurance
• Child ages off Medicaid/SCHIP or parents
  insurance
• Annual start of school, child health forms

29
Critical piece of key life event strategy a
nearly effortless form

• Check one box to indicate
• Uninsured
• Want coverage
• Want state officials to examine otherwise
  confidential data to determine eligibility
• Permission to contact household to follow-up
• SSN (to facilitate data matching)
• Uninsured person seeking coverage (essential to
  FFP)
• Household adults (cant be required of
  non-applicants, but can request, to facilitate
  eligibility determination phrase carefully!)
• Maybe one or two facts unavailable from other
  data
• Citizenship?
• Resist temptation to add!!!

30
Identifying the uninsured through master list
comparison

• Simple idea compare list of insured with list of
  all group members
• People on one list but not the other are probably
  uninsured

31
Wheres the list of insured people?
Where?

• Medicaid/SCHIP
• Private coverage
• DRA Section 6035 (TPL)
• Each state must require insurers to provide
  information re enrollment of Medicaid
  beneficiaries
• Explicitly applies to group plans under ERISA
• CMS developing mechanism

32
Listing all group members

• Statewide lists are incomplete but useful
  starting points
• More targeted lists are promising. E.g.
• Compare public program records with
  Medicaid/SCHIP enrollment records to identify the
  potentially uninsured
• For Medicaid enrollment does not have to await
  info re private coverage, since the privately
  insured qualify for Medicaid

33
Example poor, uninsured parents
Source Dorn and Kenney. Notes (1) Poor parents
have the following characteristics their income
is at or below the FPL they are ages 18 to 64
and they live with a stepchild, biological child,
or adopted child under age 18. (2) Analysis
based on 2002 NSAF. (3) NSLP is the National
School Lunch Program.
34
Health Coverage Among Poor Parents Whose Families
Participated in Means-Tested Nutrition Programs
or Whose Children Received Medicaid, 2002
Source Dorn and Kenney.
High-impact, efficient intervention via SPA
35
Function Number Two Determining Eligibility
Dn it all, sir! Am I not eligible?

1. Define eligibility based on data
2. Express Lane Eligibility
3. Using data to target intensive application
   assistance

Illus. John McLenan, A Tale of Two Cities, 1859
36
Defining eligibility based on data Medicare Part
B means-testing

• Traditionally, Part B premiums received 75
  percent subsidy for all enrollees
• Under Medicare Modernization Act (MMA), Part B
  subsidy is means-tested, starting 1/07

37
For purposes of Medicare Part B, how is 2007
income determined?

• 2005 tax year income determines Part B income for
  all of 2007
• It does not matter if you won the lottery in 2006
  or 2007
• BUT - if you come forward and show your income is
  lower in 2007 than 2005 and you qualify for
  deeper subsidies, your 2007 income controls!

38
Applying this model to state coverage initiatives

• Pure disregard income above taxable income
  during the most recent available tax year
• 1902(r)(2)
• Adjusted disregard such income, as modified by
  more recent income information from state
  workforce agencies
• New hires and quarterly earnings data
• Either way
• Continuous eligibility, regardless of
  post-enrollment changes in household
  circumstances
• No asset test data not as good re assets
• Other eligibility pathways remain open

39
Is this reasonable?

• Fewer possibilities of error MEQC/PERM
• Key prior SPA approval
• But low-income people dont file tax forms!
• Income information still reported 1099, W-2
• If state provides EITC, most low-income people
  file
• 86 of eligible families with children, 45
  without children
• Income changes are more common with working
  families than with seniors
• With proposed Administration tax credits for
  low-income workers, prior-year tax data
  determined eligibility

40
Express lane eligibility

• Concept if another means-tested program has
  already found a family to have sufficiently low
  income to qualify for Medicaid or SCHIP, enroll
  the family in Medicaid or SCHIP!

But there are obstacles to overcome!
41
Most low-income, uninsured children live in
families that receive means-tested nutrition
assistance
Source Dorn and Kenney, Urban Institute
(prepared for Commonwealth Fund), June 2006.
Notes (1) Analysis based on 2002 NSAF. (2) NSLP
is the National School Lunch Program. (3)
Low- Income is at or below 200 of the FPL.
42
Obstacle methodologies

• Problem each program has its own methodology
• Generally, Medicaid will determine families to
  have lower income than will other programs
• But not always e.g.,food stamps, excess shelter
  cost deduction
• Upshot health program must recalculate
  eligibility, family may need to reapply

43
Overcoming methodology obstacle

• Pick non-health program with income threshold far
  below Medicaids. E.g, with children
• Medicaid to 150 FPL (after disregards)
• Free school lunch - 130 FPL (gross income)
• SSA 1902(r)(2) income disregard. E.g.
• Disregard all income above net family income
  found by food stamp program
• FS net income limit 100 FPL
• 1115 waiver to disregard methodological
  differences
• Budget neutrality unspent SCHIP allocations

44
Will federal government say yes?

• Uncharted terrain - but
• Bush Administration supported Express Lane in
  context of Frist-Bingaman bill in 109th Congress
  (S. 1049)
• CMS already provides more aggressive Express Lane
  eligibility into low-income subsidies (LIS) for
  Medicare Part D
• Auto-enrollment from MSP into LIS, even though
• 6 states waive asset test for MSP, and LIS has
  asset test
• 18 states disregard in-kind income for MSP
• 10 states define household to include resident
  grandchildren
• Statutory standard Substantially the same

45
Proposed legislation

• Express lane becomes state option or
  demonstration
• Children and adults
• Extra federal money for IT connections between
  health agencies and others
• More access to federal data
• Context
• SCHIP reauthorization

46
At a minimum, can use data to target intensive
application assistance

• With children, can provide presumptive
  eligibility, then follow-up to transition to
  ongoing coverage

47
More on data-based targeting of intensive
application assistance

• To target, use income data from multiple sources
• In gathering income data, notify re (a) possible
  use for health coverage and (b) how to opt out of
  such use
• Simplifying application process
• Phone calls, not forms (send cards, ask family to
  call at convenient time)
• Pre-populate forms with income estimates, ask for
  corrections
• Use MCOs to provide assistance?
• Leveraging someone elses dollars BUT
• Conflict of interest

48
Function Number Three enrollment into coverage

• Default enrollment
• Phone-activated insurance cards

49
Default enrollment

• Youre eligible! Well enroll you unless you say
  no.
• Example NYC enrolled 13,000 children based on
  Food Stamp data. Parents could decline, but only
  2 did.
• Probably best without premiums
• Risks
• Wrong address
• Capitated payments, no services
• Strategies
• To start capitated payments, MCO must confirm
  w/family
• Partial withhold of capitation until 1 service
  provided
• In large part, base default enrollment shares on
  preventive services to prior default enrollees
• Monitor real-time encounter data

50
Phone-activated insurance cards

• Idea from Ruth Kennedy, director of child health
  for LA Medicaid and SCHIP
• Send cards, with strip of tape saying, Call
  toll-free number to activate
• Voice prompts can allow choice of plan
• Can use with premiums

51
The Auto-Enrollment motto
Applications? We dont need no stinkin
applications!
52
Summary

• For new state initiatives to succeed, enrollment
  and retention methods must be effective
• The more you ask people to do, the fewer people
  will do it
• If you want new initiatives to cover as many
  eligible individuals as possible, consider
  automatic mechanisms to
• identify the uninsured
• determine eligibility and
• enroll people into coverage.