Using Data to Inform Policy: Principles and Some Examples from Health Care

1
Using Data to Inform PolicyPrinciples and Some
Examples from Health Care

• Scott Leitz
• Director, Office of Health Policy and Research
• Minnesota Department of Health
• August 20, 2006

2
Overview of Presentation

• The Importance of Data to the Policy Process
• Some general observations and thoughts on using
  data to inform policy
• Four (not mutually exclusive) areas of influence
• Framing the Issue
• Informing the Policymaker (and the public) and
  the debate
• Making the Case
• Developing the Solution
• Summary

3
The Importance of Data to the Policy Process

• An old saw
• The plural of anecdote is not data
• Legislators and policymakers are there to
  legislate and make policy
• Do so in the presence or absence of data to
  inform their decisions
• Will use data to inform their decision but in
  absence of data, still need to make decision
• Data and information availability doesnt always
  guarantee theyll be used to inform the
  decisionbut lack of data guarantees that they
  wont

4
1. Know Your Audience

• Know your audience
• A legislative audience is different than an
  academic audience is different from the general
  public
• What story do you want to tell?
• Educational?
• Promotional?
• How best is that story told?

5
2. Communicate to that Audience

• No single or clear rule of thumb on how best to
  communicate data. The mode depends on the
  audience
• One pagers
• Chart packs
• Talking points
• But in all cases, a limit number of takeaway
  messages (and clarity as to what those messages
  are) is important

6
3. National data is great State data is better
local data is the best

• The Tip ONeil doctrine all politics is local
• Legislators and policymakers
• Will use information when its available
• Also have a belief that their city/county/state
  is different (or at least want to see whether
  they are)
• Any data helps, but the more localized the
  information is, the more relevant it becomes to
  the policymaker
• Example uninsured low income kids
• Nationwide 7.1
• VT 1.4 CT, MA 2.4
• TX 14.4 MT 10.6

7
4. Be a Healthy Skeptic about the Source of Data

• The explosion of internet sites, think tanks, and
  advocacy organizations has led increased
  availability of information on nearly any topic
• On one hand great, more information. On the
  other not as great, as it can make sorting a
  more difficult task.
• Because some organizations can bring idiological
  perspectives use a healthy skepticism in
  choosing the studies/data sources to highlight
• Refereed journals are safest fact-based
  descriptive statistical information from
  non-partisan think tanks are also generally
  safe

8
5. Get to know your local university and state
agency resources (and make sure they get to know
you)

• Academic researchers are often looking for a
  policy outlet for their research
• State analysts oftentimes know and understand the
  different sources of data available
• Opportunities to leverage the interest of
  academics in seeing their work used to bring data
  to bear on the topic of interest to you
• Example Rhode Island Medicaid program
• Excellent in-house analysis but also a great
  interactive partnership with Brown University

9
6. And get to know your state agencies,
legislative staff, advocates, and the press/media

• Staff at state agencies and legislative staffers
  can be valuable resources in reaching
  policymakers
• Their understanding of your story is important
  they can sometimes help you to refine your
  message
• The media can help you tell a story, especially
  if they can have data and a compelling personal
  story

10
7. Money Talks (or at least makes people
interested in listening)

• Programmatic outcomes are important
• did we improve the health outcomes of our
  population
• But oftentimes not sufficient to sell the
  program
• Outcomes tied to cost savings or
  cost-effectiveness are important additional
  component
• See Rhode Island RIteCare, Minnesota
  uncompensated care analysis

11
8. Be mindful of the language thats used in
presenting data findings (and be honest with it)

• How things are phrased matter
• And affects credibility and believability

12
Between 1999 and 2004, Florida experienced a
dramatic increase in the number of uninsured
individuals under age 65
Source 1999 and 2004 Florida Health Insurance
Studies, referenced by Allison Hall at the State
health Research and Policy Interest Group
meeting, Feburary 7, 2006.
13
Between 1999 and 2004, Florida experienced a
slight increase in the number of uninsured
individuals under age 65
Source 1999 and 2004 Florida Health Insurance
Studies, referenced by Allison Hall at the State
health Research and Policy Interest Group
meeting, Feburary 7, 2006.
14
Framing The Issue
15
Issue Framing

• Why does the issue matter?
• What does the data tell us about the issue?
• 2 Examples
• Obesity
• Insurance coverage trends
• Especially related to low income kids and adults

16
Obesity

• Obesity has been shown to be associated with
  increased risk of many chronic diseases,
  including
• Type 2 diabetes, cardiovascular disease, several
  types of cancer, musculoskeletal disorders, sleep
  apnea and respiratory problems, stroke, and
  gallbladder disease
• Between 1987 and 2001, obesity prevalence
  increased 10.3 percentage points, while normal
  weight prevalence declined 13 percentage points
  (Thorpe, Health Affairs, 2004).

17
Obesity Trends Among U.S. AdultsBRFSS, 1985
(BMI 30, or 30 lbs overweight for 5 4
person)
18
Obesity Trends Among U.S. AdultsBRFSS, 1986
(BMI 30, or 30 lbs overweight for 5 4
person)
19
Obesity Trends Among U.S. AdultsBRFSS, 1987
(BMI 30, or 30 lbs overweight for 5 4
person)
20
Obesity Trends Among U.S. AdultsBRFSS, 1988
(BMI 30, or 30 lbs overweight for 5 4
person)
21
Obesity Trends Among U.S. AdultsBRFSS, 1989
(BMI 30, or 30 lbs overweight for 5 4
person)
22
Obesity Trends Among U.S. AdultsBRFSS, 1990
(BMI 30, or 30 lbs overweight for 5 4
person)
23
Obesity Trends Among U.S. AdultsBRFSS, 1991
(BMI 30, or 30 lbs overweight for 5 4
person)
24
Obesity Trends Among U.S. AdultsBRFSS, 1992
(BMI 30, or 30 lbs overweight for 5 4
person)
25
Obesity Trends Among U.S. AdultsBRFSS, 1993
(BMI 30, or 30 lbs overweight for 5 4
person)
26
Obesity Trends Among U.S. AdultsBRFSS, 1994
(BMI 30, or 30 lbs overweight for 5 4
person)
27
Obesity Trends Among U.S. AdultsBRFSS, 1995
(BMI 30, or 30 lbs overweight for 5 4
person)
28
Obesity Trends Among U.S. AdultsBRFSS, 1996
(BMI 30, or 30 lbs overweight for 5 4
person)
29
Obesity Trends Among U.S. AdultsBRFSS, 1997
(BMI 30, or 30 lbs overweight for 5 4
person)
30
Obesity Trends Among U.S. AdultsBRFSS, 1998
(BMI 30, or 30 lbs overweight for 5 4
person)
31
Obesity Trends Among U.S. AdultsBRFSS, 1999
(BMI 30, or 30 lbs overweight for 5 4
person)
32
Obesity Trends Among U.S. AdultsBRFSS, 2000
(BMI 30, or 30 lbs overweight for 5 4
person)
33
Obesity Trends Among U.S. AdultsBRFSS, 2001
(BMI 30, or 30 lbs overweight for 5 4
person)
No Data lt10 1014
1519 2024 25
34
Obesity Trends Among U.S. AdultsBRFSS, 2002
(BMI 30, or 30 lbs overweight for 5 4
person)
(BMI ?30, or 30 lbs overweight for 54 person)
No Data lt10 1014
1519 2024 25
35
Obesity Trends Among U.S. AdultsBRFSS, 2003
(BMI 30, or 30 lbs overweight for 5 4
person)
No Data lt10 1014
1519 2024 25
36
Obesity Trends Among U.S. AdultsBRFSS, 2004
(BMI 30, or 30 lbs overweight for 5 4
person)
No Data lt10 1014
1519 2024 25
37
Impact of Obesity on Rising Medical Spending

• Obesity-related health spending accounted for 27
  of inflation adjusted per capita health spending
  increases
• 41 of heart disease spending
• 38 of diabetes-related spending
• Thorpe, October 2004, Health Affairs
• Medicare will spend 34 more on an obese person
  than on someone of normal weight
• Lakdawalla, et al., Health Affairs, September 2005

38
Insurance Coverage Changes

• Proportion of people lacking health insurance
  increased to 15.7 in 2004, up from 14.0 in
  2000.
• 45.8 million people in the U.S. lack health
  insurance.

39
Change in Number of Uninsured, 2000 to 2004
All nonelderly
Adults
Children
6.3 million
6.0 million
-0.3 million
SourceJohn Holahan and Allison Cook, Changes in
Economic Conditions and Health Insurance
Coverage, 2000-2004, Health Affairs web
exclusive, November 1, 2005.
40
Change in Number of Uninsured Adults, by Income
2000 to 2004
All Adults
6.3 million
lt200 FPG
200-400
400
4.2 million
1.5 million
0.6 million
SourceJohn Holahan and Allison Cook, Changes in
Economic Conditions and Health Insurance
Coverage, 2000-2004, Health Affairs web
exclusive, November 1, 2005.
41
Percentage-point Change in Sources of Coverage of
Low Income Adults, 2000 to 2004
SourceJohn Holahan and Allison Cook, Changes in
Economic Conditions and Health Insurance
Coverage, 2000-2004, Health Affairs web
exclusive, November 1, 2005.
42
Percentage of Firms Offering Health Benefits, by
Firm Size
Source Kaiser/HRET Survey of Employer-Sponsored
Health Benefits, 2002 to 2005.
43
Family Health Insurance as a Percentage of Median
Family Income
Sources Len Nichols, calculations based on data
from the Kaiser Family Foundation, AHRQ and CPS.
44
Percentage-point Change in Sources of Coverage of
Low Income Children, 2000 to 2004
SourceJohn Holahan and Allison Cook, Changes in
Economic Conditions and Health Insurance
Coverage, 2000-2004, Health Affairs web
exclusive, November 1, 2005.
45
Informing the Policymakers

• Or the Generally Interested Audience

46
Informing Policymakers and Informing the Debate

• A key role that data plays is establishing a set
  of facts that inform policymakers of the
  landscape
• Serves a critical educational role
• Necessary precondition to making your case
• Information can then be used by advocates on
  either side of the issue, with the general shared
  base of knowledge
• From the perspective of how to inform policy with
  data
• Sets up a common level of education and
  understanding is important for the more complex
  discussions on specific issues
• When thinking about framing of issues, can be
  useful to start with these common sets of
  agreement

47
A Few Examples
48
Health Spending is Highly Concentrated among a
Relatively Few People
Source Berk and Monheit, The Concentration of
Health Care Expenditures, Revisited, Health
Affairs, March/April 2001. Expenditure estimates
for civilian non-institutionalized population.
49
Health Care Spending as a Portion of the Gross
Domestic Product
Source Centers for Medicare and Medicaid
Services.
50
Increases in Health Insurance Premiums Compared
to Other Indicators, 1988-2004
13.9
2.3
2.2
Estimate is statistically different from the
previous year shown at plt0.05. Estimate is
statistically different from the previous year
shown at plt0.1. Note Data on premium increases
reflect the cost of health insurance premiums for
a family of four. Source KFF/HRET Survey of
Employer-Sponsored Health Benefits 1999-2004
KPMG Survey of Employer-Sponsored Health
Benefits1993, 1996 The Health Insurance
Association of America (HIAA) 1988, 1989, 1990
Bureau of Labor Statistics, Consumer Price Index
(U.S. City Average of Annual Inflation (April to
April), 1988-2004 Bureau of Labor Statistics,
Seasonally Adjusted Data from the Current
Employment Statistics Survey (April to April),
1988-2004.

51
Medicaid Births as a Percentage of Total Births
in the US, 1993 to 2000.
Source MCH Update 2002 State Health Coverage of
Low Income Pregnant Women, Children, and Parents.
52
Informing the Debate

• Example Minnesota Governors proposal in 2003
  to
• Eliminate General Assistance Medical Care
• Reduce eligibility for subsidized health
  insurance coverage for single adults from 175
  FPG to 75 FPG, and for parents from 275 FPG to
  175 FPG
• Hospital and provider concerns raised
• Request from Governors office What will be the
  resulting uncompensated care?

53
Methodology

• Used a variety of state of Minnesota data sources
  including
• State health expenditure accounts
• Hospital and provider reporting of uncompensated
  care
• Enrollment and expenditure figures from our state
  Medicaid agency
• Household insurance survey data
• Published academic literature

54
Results Estimated Impact on Number of Uninsured

• Minnesota has relatively low uninsured rate
  estimated at 5.7 at time of analysis
• Number of uninsured increase by the following
  relative to current levels
• Baseline estimate, 2003 272,000 total
• 2004 32,206
• 2005 50,577
• 2006 57,476
• 2007 63,108

55
How Do These Estimated Increases in Uncompensated
Care at Hospitals Compare to Current Levels?
88
80
63
34
28

56
Making the Case
57
Making the Case

• Just as data can be useful in providing a common
  set of facts and also to frame an issue for
  policymakers, it can also be used to make a
  specific case about the effectiveness of programs
• Following Example
• Borrowed in its entirety from
• Presentation by Deborah Florio, RI Department of
  Human Services
• June 25, 2005
• Presentation to Academy Health-Annual Research
  Meeting

58
How are Medicaid SCHIP Weathering the Fiscal
Storm in Rhode Island?
Presentation to Academy Health-Annual Research
Meeting By Deborah Florio RI Department of Human
Services June 25, 2005
59
Protecting RIte Care From Going Under Through
Value Based Purchasing

• Ten year effort of using Research and Data
  Analysis to highlight successes and define
  improvement needs
• Document What We Do and Why It Matters
• Use of Descriptive Materials to Tell the Story

60
Rite Care Approach

• Set Goals Early
• Identify Measurable Indicators
• Establish a Baseline
• Implement Program Intervention
• Monitor Trends
• Evaluate Impact
• Make Midcourse Corrections
• Provide Concise Information to Target Audience

61
RIte Care

• Created in 1994 under Medicaid RD waiver with
  the following goals
• Reduce uninsurance for low-income children and
  families
• Improve access, service quality and health status
  for the covered population
• Control the rate of growth in Medicaid
  expenditures for the eligible population

62

• Goal 1
• Reduce uninsurance for
• low-income children and families

63
Percent Uninsured Rhode Island Children lt 18
Years Old - 1994-2003
2nd
1st
2nd
3rd
1st
3rd
8th
25th
2nd
2nd
Data Source Medicaid Research and Evaluation
Project, RI Access Project US Bureau of the
Census, Current Population Surveys 1994-2002
(September estimates)
64
Goal 2Improve access, service quality and
health status for the covered population
65
Methods

• Oversight and monitoring of Health Plan contracts
  
• site visits
• encounter data analysis
• Health Plan Performance incentives
• Trend access, quality and health outcome
  indicators for all enrollees
• RIte Care Member Satisfaction Survey

66
Monitoring Trends
67
RIte Care Lead Screening Rates Improve Percent of
Two Year Olds with Timely Recommended Screening


GAO report NHANES estimates Patrick
Vivier, MD, Phd, 1997
68
Infant Mortality Rate Declines in Rhode Island
Infant Mortality by Insurance Status 1990-1999
Data Source Medicaid Research Evaluation
Project Center for Child Family Health,
Department of Human Services Linked Birth Death
File 1990-99, Division of Family Health,
Department of Health (n905) Deaths per 1000
births to Infants 0-364 days 3 year moving
average
69
Goal 3Control the rate of growth in Medicaid
expenditures for the eligible population
70
RIte Care Cost-Efficient A few states have
revamped their organizational and management
systems to ensure better access to medical care
while keeping costs under control. Rhode Island
stands out in this respect. Governing Magazine,
Feb 2004
71
Lessons Learned

• Start Early to Establish a Baseline
• Ensure Access to Data
• Use the Data in a Variety of Ways (Speak to
  Legislature, Governor, Advocates, Grantors)
• Integrate Research into the Medicaid Program
• Use Interdisciplinary Team
• Supplement Research and Evaluation with Outside
  Funding
• Monitor Long Term Goals and Improvements

72
Developing the Solution and Assessing its effects

• Finally, data can also be used to identify
  potential solutions to issues that have been
  raised and then to assess and monitor the effects
  of the solution
• Example Arkansas Childhood Obesity Initiative

73
(No Transcript)
74
2005 Statewide Results (Arkansas)BMI
Classifications for All Students
At Risk for Overweight 17
Healthy Weight 60
Overweight 21
Underweight 2
75
Act 1220 Activities and Impact

• Elimination of all vending machines in public
  elementary schools statewide
• Requires professional education of all cafeteria
  workers
• Public disclosure of pouring contracts
• Assessment and development of physical activity
  and health education standards
• Parent advisory committees for all schools
• Child Health Advisory Committee
• Annual BMI Assessment on every child
• AND
• Evaluation component through U of AR School of
  Public Health

76
In Summary

• Remember again that
• Legislators and policymakers are there to
  legislate and make policy
• And will do so in the presence or absence of data
• So, while the plural of anecdotes is not data
  sometimes the plural of anecdotes can be
  legislation
• Data effectively used can help to inform the
  process

77
Contact Information
Scott Leitz, Director Office of Health Policy and
Research Minnesota Department of Health Phone
651-201-3565 Email scott.leitz_at_state.mn.us