MEPS: A National Information Resource to

advertisement
MEPS: A National Information Resource to
Support
pp
Health Research and Inform Policy
y
and Practice
Steven B. Cohen PhD
Joel W. Cohen PhD & Jessica C. Banthin PhD
Presentation
 AHRQ mission and emphasis on





information and research efforts that
translate into policy and practice
MEPS overview
Program outreach and impact
Research Update
Modeling and Simulation Studies
MEPS
S Data
a a Products
oduc s and
a d Dissemination
sse
a o
Medical Expenditure Panel
Survey (MEPS)
Annual Survey of 14,000 households:
provides national estimates of health care
use, expenditures, insurance coverage,
sources off payment,
t access to
t care and
d
health care quality
Permits studies of:
 Distribution of expenditures and sources of
payment
 Role of demographics, family structure,
insurance
 Expenditures for specific conditions
 Trends over time
MEPS Components
p
 Household Component (HC)
 Medical Provider Component (MPC)
 Insurance Component (IC)
MEPS Household Component
Sample Design
Oversampling of policy relevant domains
1996
Minorities (Blacks & Hispanics)
1997
Minorities
Low income
Children with activity limitations
Adults with functional limitations
Predicted high expenditure cases
Elderly
1998--2001
1998
Mi iti
Minorities
2002--2008
2002
Minorities, Asians, Low Income
2009+
Minorities Asians
Minorities,
14,000 households; ~32,000 persons
HC - Purpose
p
 Estimates annual health care use and




expenditures
Provides distributional estimates
Supports person and family level
analysis
Tracks changes in insurance coverage
and
d employment
l
t
Longitudinal design; linkage to NHIS
Key Features of MEPS
MEPS--HC
 Survey of U.S. civilian noninstitutionalized population
 Sub
Sub--sample of respondents to the National Health
Interview Survey (NHIS)
–
Linkage to NHIS
 Oversample of minorities and other target groups
 Panel Survey – new panel introduced each year
– Continuous data collection over 2 ½ year period
– 5 inin-p
person interviews ((CAPI))
– Data from 1st year of new panel combined with
data from 2nd year of previous panel
MEPS Overlapping Panels
((Panels 13 and 14))
MEPS Household
Component
MEPS Panel 13 20082009
1/1/2008
NHIS
2007
Round 1
1/1/2009
Round 2
Round 3
NHIS
2008
Round 4
Round 1
Round 5
Round 2
Round 3 Round 4
MEPS Panel 14
2009-2010
Round 5
MEPS - Integrated Survey
g Features
Design
 National Health Interview Survey serves




as sample
l fframe ffor Household
H
h ld
Component
Census Bureau Business Register serves
as Insurance Component sample frame
Secondary data on health care measures
supplement surveys
Linked surveyy of medical providers
p
Distinct data sources linked for
longitudinal
g
analyses
y
Evaluation of Accuracy of MEPS after
Adjustments for Survey Nonresponse
 MEPS has overlapping panel design: 1st year of new




panel combined with data from 2nd year of previous
year’s panel to yield annual data
Multiplicative response rates: product of NHIS RR and
MEPS RR (multiplicative function of round specific RR:
3 rounds for new panel/5 rounds for old panel)
Detailed adjustments for survey nonresponse:
NHIS to MEPS round 1/MEPS round 1 to round 3: to
derive annual estimates for year t
MEPS round 3 to round 5: annual estimates for year
t+1.
Testing for Panel Effect
Capacity of MEPS to Produce
Comparable NHIS Estimates of Health
I
Insurance
Coverage
C
Medical Provider
Component
Purpose
 Compensate for household item
nonresponse
 Gold standard for expenditure estimates
 Greater accuracy and detail
 Imputation source
 Supports methodological studies
Medical Provider
Component
Targeted Sample
 All associated hospitals and associated
physicians
 Sample of associated officeoffice-based physicians
 All associated home health agencies
 All associated pharmacies
Data Collected
 Dates of visit
 Diagnosis and procedure codes
 Charges (except Rx) and payments
MPC: Correction Source for
Item Nonresponse
p
Source for event level
expenditures
Household
Reported
Nonresponse
reported
nonresponse
1Recalibrated
Provider
reported
reported
nonresponse
nonresponse
MEPS value - Yij
Yij = Provider $ij
Yij = Provider $ij
Y = Household $ij
1 ij
Yij = Imputed $ij
as necessary based on analyses of
concordance
b t
between
sources
Determination of Factors for
Expenditure Imputation
Hot Deck Imputation:
Classification Variables for Donors and
Recipients
Factors
associated with
predicting
medical
expenditures
Factors
associated with
item
nonresponse
Collection of Rx Data in
MEPS
 ~8,000
8,000 pharmacies sampled annually
– data on prescribed medicines purchased by
households
 Data obtained:
– Medication Name
– National Drug Code (NDC)
– Quantity Dispensed
– Strength
St
th and
dF
Form
– Sources of Payment
– Amount
A
t Paid
P id b
by E
Each
hS
Source
MEPS Insurance
p
Component
Annual survey of 40
40,000
000 establishments
National and State Level estimates of employer
sponsored coverage:
 Availability of health insurance
 Access to health insurance
 Cost of health insurance
 Benefit and payment
y
provisions of private
health insurance
Published Estimates from the
MEPS--Insurance Component
MEPS
 Each year the MEPSMEPS-IC produces 280 tables of
State
St
Statet -level
l
l estimates
ti t ffor privateprivate
i t -sector
t employers:
l
– Premiums,
– Contributions,
C t ib ti
– Enrollments,
– Take
Take--up rates, and
– Other (i.e., percent of employees with a choice
off plans)
l
)
 Survey began in 1996 with estimates for 40 States
 Since
Si
2003,
2003 estimates
i
are available
il bl ffor allll S
States
Current Capacity
p
y
AHRQ’ss MEPS data and research findings
AHRQ
provide national and state specific estimates
of:
 the uninsured population – by length of time,
availability of offers, income level
 the characteristics of employer sponsored
coverage – availability, employee take up,
premium costs (employer/employee)
 health care utilization, expenditures, source of
payment,
p
y
, and health status p
profiles by
y
insurance coverage status
Trends in medical care costs,
coverage and use
Impact of economic and behavioral factors, payment and
individual demand on health care service utilization
and expenditures
 Distribution of expenditures, concentration and
persistence of high levels
 Expenditures for chronic conditions: focus on patients
with multiple chronic conditions
 Trends
T d in
i prescription
i ti medications
di ti
b
by d
drug class
l
Research on Health Insurance
 Tracks overall health insurance status of the
U.S. population
– Estimates of uninsured by population
characteristics
– Duration of spells of uninsurance
– Trends in estimates of the uninsured
 More focused research examines
– Factors associated with insurance take up
– Financial consequences of being uninsured
– Relationship between uninsurance and health
status
MEPS
Definition and estimation of uninsured
 Types
T
off estimates
ti t off uninsured
i
d – calendar
l d
year focus:
1 First half of calendar year
1.
2. Annual profiles
3 Two consecutive years
3.
4. Point in time
5. Long
5
Longo g-term
e u
uninsured:
su ed 4 co
consecutive
secu e yea
years
s
 As a longitudinal
g
survey
y MEPS can examine
health insurance dynamics, changes in
coverage, and spells without insurance
Economic Research
Infrastructure
 Data infrastructure to support intramural, extramural
work on cost and financing,
financing efficiency and quality,
quality
access, disparities.
– Significant intramural expertise and activity
– Large and growing use by extramural researchers
 Data Center onon-site for work with MEPS
 The
Th link
li k between
b t
research
h and
d data
d t
– Substantive expertise reflected in design of AHRQAHRQ-
sponsored
p
data resources and tools
– Maintain and increase quality, integrity, and relevance
through researcher
researcher--informed data improvements,
substantive and technical assistance
Assistance to Congress on Coverage
Trends and Cost
 Provision of AHRQ research findings to inform
health policy
– national estimates of the long term uninsured
– estimates of the number of uninsured children
eligible for CHIP
– state estimates of the availability and cost of
employer sponsored coverage
– concentration of health care expenditures
 Fast
Fast--track responses to requests from CBO, CRS,
Senate and House Committees and
Congressional
g
staff
Recent Collaborations
CDC, Medicaid Chronic Disease Directors and National
Pharmaceutical Council

Development of tool to calculate prevalenceprevalence-based state
state-specific Medicaid and total cost estimates for :
heart diseases, stroke, hypertension, congestive heart failure, diabetes,
and cancer

Cost estimation tool based on Medical Expenditure Panel
Survey (MEPS) data

consideration of enhancements that identifies the impact
p
of specific
p
i t
interventions
ti
on costs
t and
d health
h lth outcomes
t
National Academy of Sciences: IOM Study on Health Insurance Status
and Its Consequences
 Assessment of health status and financial burdens faced by the
uninsured,
i
d focus
f
on impact
i
t off coverage trends
t
d over prior
i decade
d
d
 Visible use of AHRQ’s MEPS as a sentinel data base
Recent Collaborations
Kaiser Permanente Care Management Institute:
Concentration of Health Care Expenditures
 Effort focused on the 1% of population accounting
f 25% off medical
for
di l expenditures
dit
 coordination in development of predictive models
 study impact of enhanced care management on
cost and health
NAS Panel: Research Program on the Design of
National Health Accounts
 Focus on measuring changes in the population’s
health within an health accounts framework
 Examination
E
i ti off costt and
d value
l off h
health
lth care
 Visible use of AHRQ’s MEPS as a sentinel data
base
MEPS Informs Consumers’ Checkbook
Guide to Health Plans




Annual
publication
Rates every
plan available
to federal
employees
and retirees
Compares
likely cost of
various plan
options
p
to
employee
Example:
Estimated
2007 cost to
average
family of 4
with head of
household
under 55
years of age
Approximate Yearly Cost to You ($)
Plan
Cod
e
Plan Name
Yearly
Premiu
m ($)
If Your
Health Care
Usage were
Low
If Your
Health Care
Usage were
Average
If Your
Health
Care
Usage
were High
Yearly Limit
on Cost to
You
Excluding
Dental ($)
Local Plans
E35
Kaiser-St
1210
1420
2670
4800
8880
E32
Kaiser-Hi
2480
2590
3340
4680
7230
JP2
M.D. IPA
2190
2340
3300
5170
7990
JN5
Aetna Open
AccessBasic
1420
1630
3090
5900
8880
JN2
Aetna Open
Access-Hi
Access
Hi
3080
3260
4570
7100
10540
222
Aetna
HealthFund
CDHP
1310
1310
3770
7700
13260
2G2
CareFirst
Bl Ch i
BlueChoice
2250
2480
3680
6030
10510
Modeling
g and Simulation Efforts
In prior decade,
decade MEPS predecessor survey
(NMES) used to model costs and impacts
of various proposed reforms
– Costs of reform to households
– Costs to nation
– Changes in coverage
– Tax impacts
Modeling
g and Simulation Efforts
Using MEPS data
data, these areas remain our
strengths, with addition of:
– Medicaid/ SCHIP eligibility simulation model
– Expenditures by service (including
prescription drug expenditures)
– State level estimates of coverage and
expenditures ((largest
g
states by
y population))
– Improved tax simulation model
– Employer
p y survey
y data by
y state
Research Uses of the
M di l E
Medical
Expenditure
dit
P
Panell S
Survey
Advancing
Excellence
in Health
Care
Research Objectives
 Provide analytic oversight of survey
 Guide construction of analytic files
 Conduct policy relevant research
 Provide technical assistance
Research Areas
 Health insurance
 Use and expenditures
 Access, quality and satisfaction
 Health status and health behaviors
 Health care reform
Health insurance status of civilian
noninstitutionalized p
population
p
under age
g
65,first half 1996-2007
Private
Public only
y
Uninsured
80
70
60
68 7
68.7
69 2
69.2
70 4
70.4
19.2
18.9
17.8
70 4
70.4
69 9
69.9
69 1
69.1
67 9
67.9
67.1
65.8
64.9
65.0
63.1
18.2
18.8
18.5
18.8
19.0
19.5
19.4
20.6
15.6
15.6
2005
2006
Perc
cent
50
40
30
17.9
20
16 3
16.3
10
12.1
11.9
11.8
11.7
11.9
12.1
13.5
14.2
15.2
1996
1997
1998
1999
2000
2001
2002
2003
2004
0
2007
Source: Center for Financing, Access, and Cost Trends, AHRQ, Medical Expenditure Panel Survey, Household Component
Summary Tables 1996–2006
Number of uninsured under age 65
MEPS, 1996
1996–2007
2007
Any time in year
Number in millions
62 0
62.0
62.2
59 1
59.1
60
58.5
61 7
61.7
First half of year
61.9
62.9
63.9
65.8
45.9
47.0
48.1
49.8
34.4
35.8
2004
2005
61.7
44.5
44.2
31.6
32.1
31.0
28.7
31.5
31.3
32.0
33.7
1996
1997
1998
1999
2000
2001
2002
2003
42.0
42.6
43.8
45.7
Full year
68.0
50.1
53.5
40
20
37.1
0
2006
2007
Source: Center for Financing, Access, and Cost Trends, AHRQ, Household Component of the Medical Expenditure Panel
Survey, 1996–2006 Full-Year Files and 1996–2007 Point-in-Time Files
Number of children under age 18 by allyear insurance status
MEPS, 1996–2006
Private
Number in
n millions
40
39.8
39.7
41 4
41.4
42 0
42.0
Public only
40 4
40.4
39.4
Uninsured
38.9
38.6
37.4
37.3
37.0
17 7
17.7
18.4
19.2
30
20
14.1
16.1
16.5
10.9
11.8
11.6
11.3
12.4
70
7.0
71
7.1
65
6.5
5.3
63
6.3
56
5.6
5.1
4.6
4.7
4.1
4.1
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
10
0
Source: Center for Financing, Access, and Cost Trends, AHRQ, Household Component of the Medical Expenditure Panel
Survey, 1996–2006 Full-Year Files
Health Insurance Status of Adults Ages 21-64:
December 2006
Any iinsurance
A
Individual Market
Employment-related
E
l
t l t d
Public Coverage
90
78 3
78.3
80
Perce
ent
70
60
69.5
70.6
67.5
65.6
85.2
82.5
53.6
50
40
30
20
10
4.2
9.9
9.3
3.2
10.2
3.9
15 3
15.3
6.1
0
21-29
21
29
30-44
30
44
45-54
45
54
55-64
55
64
Source: Center for Financing, Access, and Cost Trends, AHRQ, Household Component of the
Medical Expenditure Panel Survey, 2006.
Retiree Coverage for Adults, Ages 55-64:
December 2006
Any retiree
A
ti
coverage
Dependent
Policyholder
P
li h ld
Dep., Pholder Age 65+
20
16.8
15.6
Perce
ent
14
9.4
10
7.9
32
3.2
2.2
0.1
0
Men
Women
Source: Center for Financing, Access, and Cost Trends, AHRQ, Household Component of the
Medical Expenditure Panel Survey, 2006.
Health Insurance Premiums Employee/Employer Contributions for
Single Coverage 1996 - 2006
Advancing
Excellence
in Health
Care
2006
$788
$
2005
$723
2004
$671
2003
$606
2002
$565
2001 $498
2000 $450
1999 $420
1998 $383
$3,330
$
,
$3,268
$3,034
$2,875
$2,624
$2 391
$2,391
$2,205
$1,905
$1,791
1997 $320
$1,731
1996 $342
$1 650
$1,650
Employee
p y
Contribution
Premiums
increased 3.2% &
employee
contributions
t ib ti
increased 9.0%
over 2005,
continuing the
trend from
previous years
years.
Employer
Contribution
AHRQ MEPS Insurance Component Tables 19961996-2006
Health Insurance Premiums Employee/Employer Contributions for
Family Coverage 1996 - 2006
Advancing
Excellence
in Health
Care
2006
$2,890
$
,
2005
$2,585
2004
$2,438
2003
$2,283
2002
$8,491
$
,
$8,143
$7,568
$6,966
$1,987
2001
$1 741
$1,741
2000
$1,614
1999 $1,438
1998 $1,382
1997 $1,305
$1 2
1996 $1,275
$6,482
$5 768
$5,768
$5,158
$4,620
$4,208
$4,027
$3 6 9
$3,679
Employee
Contribution
Employer
Contribution
Premiums
increased 6.1%
& employee
contributions
increased 11.8%
over 2005,
continuing the
trend from
previous years.
AHRQ MEPS Insurance Component Tables 19961996-2006
Advancing
Excellence
in Health
Care
Average Annual Health Insurance Premium per
Enrolled Employee at PrivatePrivate-Sector Establishments
Offering
g Health Insurance: US and Ten Largest
g
States, 2006
State
UNITED STATES
California
T
Texas
New York
Florida
Illinois
Pennsylvania
Ohio
Michigan
New Jersey
Georgia
Single
Coverage
$4,118
$4,036
$4,133
$4 133
$4,605
$3,936
$4,245
$4,277
$4,054
$4,446
$4,471
$3,873
$3 873
Employee-Plus
EmployeePlus-One Coverage
$7,988
$7,989
$8,081
$8 081
$8,779
$7,735
$7,984
$8,764
$7,884
$8,654
$8,791
$7,609
$7 609
Family
Coverage
$11,381
$11,493
$11,690
$11 690
$12,075
$11,046
$11,781
$11,794
$10,967
$11,452
$12,233
$10 793
$10,793
AHRQ MEPS Insurance Component Tables - 2006
Premiums for employer-sponsored
health insurance, family coverage, private
sector by firm size,
size 1996
1996–2006
2006
$12,000
$11 438
$11,438
$11,381
$11,000
$11,095
$$10,000
,
$9,000
All firms
Small firms
L
Large
firms
fi
$8,000
$7,000
$6,000
$4,957
$5,000
$4,954
$4,938
$4,000
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
Employer costs for employer-sponsored
health insurance, family coverage, private
sector
t b
by firm
fi
size,
i
1996–2006
1996 2006
$9,000
$8,590
$8,491
$8,000
$7,994
$7,000
All firms
Small firms
$6,000
Large firms
$5,000
$4,000
$3,000
$3,702
$3,679
$3,571
$2 000
$2,000
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
Source: Center for Financing, Access, and Cost Trends, AHRQ, Medical Expenditure Panel Survey – Insurance Component,
2003, Tables II.C.2, II.C.3, II.D.2, II.D.3, II.E.2 and II.E.3
Employee contributions for
employer-sponsored health insurance, family
coverage, private
i t sector
t by
b firm
fi
size,
i
1996–2006
1996 2006
$3,500
$3,101
$3,000
$2,890
$2,848
$2,500
All firms
Small firms
Large firms
$2,000
$1,367
$1,500
$1,275
$1,255
$1 000
$1,000
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
Source: Center for Financing, Access, and Cost Trends, AHRQ, Medical Expenditure Panel Survey – Insurance Component,
2003, Tables II.C.2, II.C.3, II.D.2, II.D.3, II.E.2 and II.E.3
Private Sector Employees With
FSAs 2001
FSAs,
2001--2005
Advancing
Excellence
in Health
Care
Percent of employees with FSAs
100%
80%
60%
40%
20%
0%
2001
2002
2003
< 10 employees
10-24 employees
100-999
100
999 employees
1000+ employees
2004
2005
25-99 employees
Source: 2001-2005 Medical Expenditure Panel Survey – Insurance Component, private sector establishments.
Average Deductibles With and
Without FSAs
Advancing
Excellence
in Health
Care
Average Deductibles
With and Without FSAs
900
0.70
800
0.60
0.50
600
500
0.40
400
0.30
300
0.20
200
0.10
100
0
0.00
2001
2002
2003
2004
2005
Year
Percent of employees with FSAs
Average deductible with FSA
Average deductible without FSAs
Source: 2001-2005 Medical Expenditure Panel Survey – Insurance Component, private sector establishments.
Proportion
Averag
ge deductible
700
Percentt
Advancing
Excellence
in Health
Care
Percentage of single workers who are
offered health insurance and take up offered coverage
according to health insurance preferences
100
90
80
70
60
50
40
30
20
10
0
Weak preferences
Uncertain preferences
Strong preferences
Offered
Taking
up
Healthy enough
Offered
Taking
up
Offered
Not worth cost
Taking
up
Take risks
Preference measure
Monheit and Vistnes, "Health Insurance Enrollment Decisions: Preferences for Coverage, Worker Sorting,
and Insurance Take-up" Inquiry (45)1 Summer 2008.
Percent of adults ages 18 – 64
uninsured all year by health insurance
preferences,
f
2001
Advancing
Excellence
in Health
Care
Perrcent unins
sured
35
30
25
20
15
Weak preferences
Uncertain preferences
Strong preferences
10
5
0
Healthy
y
enough
Not worth
cost
Take risks
Can
overcome
illness
Preference measure
Monheit and Vistnes, "Health Insurance Enrollment Decisions: Preferences for Coverage, Worker Sorting,
and Insurance Take-up" Inquiry (45)1 Summer 2008.
Distribution of Workers by Plan Choice
A il bilit 2006
Availability,
Not offered insurance
Offered a choice of plans
Offered 1 plan
70
58.4
Perce
ent
60
50
42.2
40
30
35.23
25.59
22.56
20
16.01
10
0
From own jjob
Source: MEPS-HC, 2006
Through
g an HIEU
Take-up Rates for Wage-Earners age 21-64
Wh H
Who
Have Off
Offers, 2006
Offered 1 plan
Offered a choice of plans
100
90
Perce
ent
80
76.31
80.49
73.39
72.06
70
60
50
40
30
20
10
0
From own job
Source: MEPS-HC, 2006
Through an HIEU
Percentage distribution of health care
spending, by type of service, U.S. civilian
noninstitutionalized population,
population 2006
Total expenses = $1
$1.03
03 trillion
Hospital inpatient
2.0
3.3
3
3
3.6
Office-based
29.7
Prescription medicines
Hospital outpatient
7.4
Dental
8.7
Emergency room
23 7
23.7
21.6
Home health care
Other medical services
and equipment
Note: Percentages may not add to exactly 100.0 due to rounding.
Source: Center for Financing, Access, and Cost Trends, AHRQ, Household Component of the Medical Expenditure Panel Survey, 2006
Source of payment distribution for health care
spending,
di
by
b age, 2006
All ages
41.7
Private Insurance
7.1
Medicare
Out of Pocket
Medicaid
Other
8
8.7
23.5
19.0
<18
years
y
18–64 years
50.7
55.5
65+ years
15.2
2.4
7.3
14.1
0.5
6.6
4.6
20.5
7.4
23.7
9.7
20.8
60.9
Note: Percentages may not add to exactly 100.0 due to rounding.
Source: Center for Financing, Access, and Cost Trends, AHRQ, Household Component of the Medical Expenditure Panel Survey, 2006
Trends in Concentration
Percen
ntage off expend
ditures
1977
100
90
80
70
60
50
40
30
20
10
0
1987
1996
2006
97 97 97 97
70 70 69
63
55 56 56
48
38 39 38
31
27 28 28
21
Top 1% Top 2% Top 5% Top 10% Top 50%
Population ranked by expenditures
Source: National Medical Care Expenditure Survey, 1977; National Medical Expenditure Survey, 1987; Medical
Expenditure Panel Survey, 1996 and 2006.
Percentage of population w
P
with same
percenttile rank in 200
06
Persistence in the level of
health care expenditures, U.S. civilian
noninstitutionalized population
population, 2005 to 2006
100
75.3
74 2
74.2
Top 50%
Lower 50%
80
62.9
60
53.5
40.3
40
20
31.3
18.1
0
Top 1%
Top 5%
Top 10%
Top 20%
Top 30%
Percentile rank by health care expenditures, 2005
Source: Center for Financing, Access, and Cost Trends, AHRQ, Household Component of the Medical Expenditure
Panel Survey, HC-106 (Panel 10, 20052006)
Advancing
Excellence
in Health
Care
15 Highest Cost Conditions, 2006
($ in billions))
 Heart Disease ($78)
 Trauma ($68)
 Cancer ($58)
 Mental
M t l Disorders
Di d
($57)
 Pulmonary Conditions
($51))
($
 Hypertension ($49)
 Diabetes ($48)
 Osteoarthritis ($38)
 Back Problems ($35)
 Hyperlipidemia
H perlipidemia ($26)
 Circulatory Conditions




($26)
Kidney Disease ($26)
Central Nervous
System
y
Disorders ($26)
($ )
Upper GI Disorders
($21)
Other Endocrine,
Endocrine
Nutritional and Immune
Disorders ($20)
Source: Center for Financing, Access and Cost Trends, Agency for Healthcare Research and Quality:
Medical Expenditure Panel Survey, 2006
Expenditures on Chronic Conditions as a
Percent of Total Expenses for adults age 18 &
over,
over by age,
age 2005
Percent
100
80
60
51.8
40
56.6
58.9
55-64
65 and older
49.6
29.2
20
0
18 & over
18-34
35-54
Note: Estimates do not include expenses for dental care and other medical equipment and services.
Source: Center for Financing, Access, and Cost Trends, AHRQ, Household Component of the Medical Expenditure Panel
Survey, 2005
Cost of Obesity
From Wang, Beydoun, and Liang, “Will All Americans Become Overweight or Obese? Estimating the
Progression and Cost of the US Obesity Epidemic,” Obesity October 2008.
Prescription Drugs
From Donohue, Huskamp, and Zuvekas, “Dual Eligibles With Mental Disorders And Medicare Part D:
How Are They Faring?” Health Affairs May/June 2009.
Regression Model Goodness of Fit:
Hosmer--Lemeshow Tests for Elderly RX
Hosmer
Expenditures > 0
Advancing
Excellence
in Health
Care
Residuals as % of M
R
Mean
10
5
0
-5
1
2
3
4
5
6
7
8
9
10
9
10
-10
-15
-20
Deciles of Predicted Expenditure Distribution
EEE
Residuals
s as % of Mean
10
5
0
-5
1
2
3
4
5
6
7
8
-10
-15
-20
Deciles of Predicted Expenditure Distribution
GGM
Gamma
Poisson
Linear OLS
Steven C. Hill and G.
Edward Miller.
“Health Expenditure
Estimation and
Functional Form:
Applications of the
Generalized Gamma
and Extended
Estimating
Equations Models.”
Forthcoming in
Health Economics.
Economics
Advancing
Excellence
in Health
Care
Comparisons of cumulative distributions
of Medicare expenditures in the MEPS to
other measures
Zuvekas and Olin, “Accuracy of Medicare Expenditures in the Medical Expenditure Panel Survey,”
Inquiry 46:92-108 (Spring 2009)
Advancing
Excellence
in Health
Care
Journal Articles Using
MEPS
AHRQ Modeling
and
d Si
Simulation
l ti Eff
Efforts
t
Overview
Division of Modeling and Simulation activities to
support health reform:
 Reconciliation of MEPS expenditure estimates
to National Health Expenditure Accounts
(NHEA)
 Additional data products
 Basic Research
 Microsimulation models
MEPS--HC: Data Products
MEPS
 MEPS is one
one--stop data source for many key
components of health policy microsimulation
models
 Virtually all major health models use MEPS
data in some way – most often they use the
individual level medical expenditure data
 The Modeling Division produces several data
products and tools to enhance MEPS for policy
simulation (available in the Data Center)
MEPS--HC: Augmented
MEPS
g
Data
 Federal and state income tax simulations (from
NBER TAXSIM)
 2002 data aligned
g
to NHEA and p
projected
j
forward to 2016
 Imputed employer contributions (regression(regressionbased IC models)
 Other enhancements:
– Immigration, citizenship status through 2005
– Fully imputed jobs variables
Importance of Reconciling
MEPS to NHEA
 Benchmarked, projected data are critical to all
models and questions
 NHEA and MEPS provide the two most
comprehensive
h
i estimates
ti t off health
h lth care
spending in the U.S.
 Reconciling estimates from both sources serves
as an important quality assurance exercise for
both.
 Augmented MEPS files include expenditures
adjusted for survey underreporting and more
Simulated Taxes
 MEPS collects detailed income and asset
data that support simulation of federal,
state, payroll, and property taxes
 Simulations produce estimates of: tax
payments marginal tax rates
payments,
 Send data files to NBER’s TAXSIM
 Make
M k further
f th refinements
fi
t and
d
calculations inin-house
MEPS-HC: Basic Research
MEPSy Simulations
to Inform Policy
Elasticities are key parameters in most
microsimulation models:
 Premium elasticity of taketake-up (Blumberg,
Ni h l B
Nichols,
Banthin)
thi )
 Tax
Tax--price elasticity of group coverage
(Selden&Bernard)
 Tax
Tax--price elasticity of self
self--employed (Selden)
 Tax subsidies,, winnerswinners-losers,, and withinwithin-firm
incidence of employer contributions
(Bernard&Selden)
 Burden of health care (Banthin&Bernard)
– Within
Within--year burdens (Selden)
Recent research on Role of Assets
in Insurance Enrollment
Questions
 What is the difference in wealth between
insured and uninsured families?
 How much better can we predict demand for
insurance using asset and wealth data?
Wealth, Income and the Affordability of Health Insurance
D.
D Bernard
Bernard, JJ. Banthin
Banthin, and W.
W Encinosa
Differences in asset holdings
byy insurance status
 Median net worth of privately insured families
was 23.2 times that of the uninsured
 Among families w/ access to employer
coverage, median net worth of privately
insured families was 15.4 times that of the
uninsured
 Among families in the individual market,
median
di nett worth
th off privately
i t l iinsured
d ffamilies
ili
was 34.6 times that of the uninsured
The role of wealth in private insurance
enrollment: simulation results
 The standard income model performs
relatively well for the employer coverage
market.
 In
I the
th individual
i di id l market,
k t th
the wealth
lth model
d l
performs significantly better
 The standard model overestimates enrollment
for those with low wealth and underestimates
enrollment for those with high wealth
 Standard
St d d model
d l estimates
ti t are misleading
i l di ffor
two subpopulations: low income and high
wealth, high income and low wealth
Simulation Models at
AHRQ
Modeling Division develops and maintains
two key simulation models:
KIDSIM
PUBSIM
New model in progress:
Employer
p y _SIM
KIDSIM
 Detailed statestate-specific Medicaid and CHIP
eligibility simulations for children and parents
 Yields most accurate estimates of eligible
uninsured
i
d children
hild
(CBO lletter,
tt JJuly
l 2007)
 Model used to estimate
Track progress over time
take--up rates
take
crowd--out rates
crowd
Simulated take up of coverage under possible
expansion
– Net costs of public coverage for children
–
–
–
–
 Currently updating model to 2007
Children Eligible for Public
g , 19971997-2005 KIDSIM
Coverage,
E lig ib le C h ild re n
(m illio n s )
50
40
All Eligible
Medicaid Eligible
CHIP Eligible
30
20
10
0
1997 1999 2001 2003 2005
P e r c e n t E li g ib le w h o a r e
U n inn s u r e d
Percent Eligible but Uninsured
Children,, 20002000-2005 KIDSIM
25.0%
All Eligible
Eli ibl
Medicaid Eligible
CHIP Eligible
20 0%
20.0%
15 0%
15.0%
10.0%
0 0%
2000 2001 2002 2003 2004 2005
Percent of Eligible Families with
Uninsured Child,, 2005 KIDSIM
50%
40%
30%
20%
10%
0%
43%
21%
16%
All CHIP
All
26%
Medicaid Medicaid-CHIP
Eligible Families with 2+ Children
EligibleI li ibl
Ineligible
P e rc e n t M e d ic a id C H I P F a m i li e s w i t h
U n i n s u r e d C h ild
Medicaid-CHIP Families with
MedicaidUninsured Child by CHIP Policy, 2005
KIDSIM
27 4%
27.4%
30.0%
20.4%
20.0%
10.0%
0.0%
p
CHIP
Medicaid Expansion
Anyy Separate
p
CHIP
CHIP Program Structure
PUBSIM
 Builds on KIDSIM for all nonnon-elderly
adults (esp. childless adults)
 Detailed statestate-specific Medicaid
Medicaid, CHIP
and state funded programs - eligibility
simulations
 Simulated disability status based on
health and employment status
MEPS--IC: Data
MEPS
 Large sample of establishments
(n=42,700 with response rate of 81%)
– Leading
g data source employer
p y offers,, taketake–
–
–
–
up, employer/employee premiums
State level estimates
D
Data
released
l
d iin tabular
b l fform
Limited public access to data files at
Census Data Centers
Most models use MEPSMEPS-IC estimates to
benchmark premiums in simulation models
based on other data (e.g.,
(
Kaiser/HRET)
/
)
StatisticallyStatistically
y-Linked IC
IC--HC Data
 “Holy Grail”: information on all establishment
employees and their families (e.g.,
(e g administer
HC to all employees at IC establishments)
– Do workers have spouses with offers? Children
with public eligibility?
– Health status of workers and their families
– Family income (all sources) and marginal tax
rates
 Next best: statistical match of HC data to IC
establishments,
t bli h
t aligning
li i th
the synthetic
th ti
workforces to IC workforce characteristics
– Used for tax subsidy estimates in Selden and
Gray (Health Affairs
Affairs, 2006)
– Updated model under construction
MEPS--IC: Employer
MEPS
p y _SIM
 Selden & Gray (HA
(HA,, 06) “populated”
populated establishments
with HC workers using statistical matching and raking
post--stratification
post
–
Enabled estimates of tax subsidy by estab characteristics
 Under new initiative at Census, we gain access to
MEPSMEPS
S-IC
C data to recreate this data resource
–
–
–
–
Tax subsidy estimates
Estimates of p
premiums facing
g workers who do not take up
p
offered coverage
Microsimulation of reforms
Responses to capped subsidies
Average Subsidy per Plan,
by
y Type
yp ((2006))
2006 $
4000
3500
3000
2500
2000
1500
1000
500
0
l
l
A
e
l
ng
i
S
il y
m
a
F
MEPS IC and HC: Selden & Gray, Health Affairs 2006
10
00
>
99
-9
10
0
-9
9
25
10
0
-2
4
2000
1800
1600
1400
1200
1000
800
600
400
200
0
-9
2006
6$
Average Subsidy per Worker
y Firm Size ((2006))
by
Firm Size
MEPS IC and HC: Selden & Gray, Health Affairs 2006
Average Subsidy per Worker
by Establishment Wage Mix (2006)
3000
2500
2006 $
2000
1500
1000
500
0
Low
d le
Mid
h
Hig
Oth
er
Predominant Wage Level
MEPS IC and HC: Selden & Gray, Health Affairs 2006
Other Applications
pp
 Updated tax subsidy estimates
 Best data source for certain policy-
relevant estimates
– e.g., average employee premium
contribution among workers turning down
offered coverage
 Database for policy reform simulations
 Behavioral
B h i l research
h iinto
t d
determinants
t
i
t off
firm offers, “crowd-out”, and more
Conclusion
 MEPS data are critical for most microsimulation
models of health reform
 AHRQ supports policy analyses based on MEPS
through
g various activities:
–
–
–
–
Public use files,
Enhanced databases, linkages and other tools
Basic research and reporting
reporting,
Micro--simulation models
Micro
 Future work includes
–
–
–
–
Statistical linkage between HC & IC
Enhanced data files
Basic behavior research on parameters used in modeling
Mi
Microsimulation
i l ti model
d ld
development
l
t
Medical Expenditure Panel Survey
DISSEMINATION OF
INFORMATION AND DATA
PRODUCTS
O UC S
MEPS Dissemination
 An onon-line interactive statistical
computer system (MEPS
(MEPS--NET)
 Data center for use of nonnon-public data
 Website
 Workshops
MEPS Website
www meps ahrq gov
www.meps.ahrq.gov











Overview of MEPS and Frequently Asked Questions
(FAQs)
( Qs)
Staff Reports using MEPS
 Findings/Statistical Briefs/Chart books
Data Tables of Estimates
Public Use Files (microdata)
MEPSnet Interactive Query Tool
Survey Methodology Reports
Survey Questionnaires and Other Collection Materials
Data product availability and ordering information
MEPS data workshop information and schedule
Mailing list
list, List server and e
e--mail for technical
assistance
Data Center Information
Greater Access to AHRQ
Restricted Data
Use of Census Bureau Research Data
Centers (RDC
(RDC)) to improve accessibility of
non--public AHRQ data
non
Examples of off
off--site approved projects in RDC’s:
Columbia University - Department of Health Policy and Management Sherry Glied – “The Tax Treatment of Health Insurance Revisited”
University of Michigan - Economic Research Initiative on the
Uninsured, – Matthew Rutledge – “Estimation of Adverse Selection
& Moral Hazard in Health Insurance”
Insurance
Growing utility of AHRQ Data Center
MEPS HC Data Also Accessible at
Census Bureau RDC’s
RDC s
 Research Data Files will be accessible at
the 9 regional Census Bureau RDC’s
((NY,NC,
, , MI,, IL,, MD,, CA,, MA))
 AHRQ will approve projects
 Will require Census Bureau Special Sworn
Status
 Census user fees will apply
Presentation Summary
y
 MEPS overview and analytic capacity
 Program outreach and impact
 Research Update
 Modeling efforts to inform health initiatives
 MEPS Data Products and Dissemination
Download