Study Objective Is Medicaid Coverage as Good as than Being Uninsured?

advertisement
Is Medicaid Coverage as Good as
Private Insurance or No Better
than Being Uninsured?
Study Objective
• Compare Medicaid beneficiaries to both
privately insured and uninsured people
– Medical care use and spending
– Short-term change in health
Jack Hadley, Ph.D.
• Draw inferences about relative efficiency of
Medicaid coverage
(George Mason University)
AcademyHealth Annual Research Meeting, June 4, 2007
Research supported by the Kaiser Family Foundation
1
Major Methodological Issue:
The Identification/Endogeneity
Problem
2
For Example
• Many people have Medicaid because of poor
health: explains higher use and worse outcomes
• Medicaid beneficiaries typically have the lowest
incomes, least education, and very weak
family/community supports—Are study results
due to Medicaid coverage or the characteristics of
Medicaid beneficiaries?
• People are not randomly assigned to
different insurance states
• Unobserved differences in characteristics
and/or health may explain both insurance
status and use/outcomes
3
4
Research Design
Data: MEPS-HC
• Use longitudinal data; measure change in
health
• Focus on exogenous adverse health events:
accidents and new chronic conditions
• Measure insurance status prior to adverse
health shock in order to establish
exogeneity
• Pooled data from 1999 – 2004
• Two-year rotating panels w/ 5 interviews
• New health conditions after baseline trigger
special module of questions, including date
of accident or when new condition first
noticed/diagnosed, if
– caused by an accident or
– condition on HHS “priority” list
5
6
1
Multivariate Regression:
Dependent Variables
Sample
Medical Care Use
• 12,787 accidents and 8,963 new chronic
conditions among non-elderly
– Received any care at all
– Categorical quantities of specific types of care
(inpatient, outpatient, office-based, ER, RX)
– Total charges and spending for care associated with
condition
– Medicaid about 10% of each sample
– Uninsured 13% (accidents) and 16% (new
chronic)
Short-Term Change in Health
• Constant insurance status 2 months before,
month of, and month after adverse health
event
– Change in health status from round before to round
after health shock (e.g., excellent to poor)
– Fully recovered from accident
7
8
Multivariate Regression:
Control Variables
Results: Service Use and
Spending
• Dummy variables for specific conditions, overall
severity (self-rated), and accident type and
location
• Baseline health and sociodemographic
characteristics (from interview before event)
• Dummy variables for census region, metro status,
year, month of event, survey round; time between
event and follow-up interview
• Service Use – Medicaid use greater than
uninsured and similar to privately insured
• Spending – Medicaid spending is greater
than uninsured and similar (new chronic) or
slightly lower (accidents) than privately
insured
9
Results: Short-Term
Change in Health
10
Conclusions: Clear-Cut
Accidents
– Medicaid beneficiaries do better than uninsured and
similar to privately insured in terms of full recovery
– But change in health status same as uninsured, and both
worse than privately insured
• Uninsured receive less care and have worse
outcomes than privately insured
New Chronic Conditions
– 1st post-event round: Medicaid change in health similar
to privately insured, and both better than uninsured
– 2nd post-event round: Medicaid change in health moves
closer to uninsured
11
• Care received by Medicaid recipients is not
more costly than care received by privately
insured
12
2
Sidebar: Medicaid Managed Care
vs. Medicaid FFS
Conclusions: Not So Clear
• If Medicaid health outcomes not
significantly better than uninsured’s
• Supplementary analysis divides Medicaid
beneficiaries into managed care (MMC) and FFS
into populations
• General Results
– May be indicative of lower quality care in
Medicaid compared to private insurance
– Methodology may still not be able to control
for effects of
– Service use and spending higher in FFS population
– Similar changes in short-term health measures
• unobserved differences between Medicaid
beneficiaries and privately insured
• unobserved Medicaid program differences
• Implies MMC more efficient than FFS: same
change in health for lower cost
13
14
Appendix: Service Use
Sidebar: Emergency Dept. Use
(odds ratios from ordered and binomial logistic regressions,
relative to privately insured)
Accidents
Accidents
– No difference in ED use by uninsured,
Medicaid, and privately insured
New Chronic Conditions
– Significantly higher use by uninsured
– Medicaid and privately insured have similar ED
use
New Chronic Condition
Medicaid
Uninsured
Medicaid
Received Any Care
1.16
0.45a
1.92a
0.52a
ED Visits
1.15
0.99
0.88
1.45a
Outpatient Visits
0.87
0.63a
1.18
0.68a
Office Visits
0.89
0.59a
1.04
0.77a
Prescription Drugs
1.13
0.65a
1.13
0.76a
Uninsured
a. Significantly different from privately insured, p<=.01
15
Appendix: Spending
Appendix: Short-Term Change in Health
(percentage difference from log-linear regressions,
relative to privately insured)
Accidents
Medicaid
16
(odds ratios from logistic regressions, relative to privately insured)
New Chronic Condition
Uninsured
Medicaid
Uninsured
All Cases
Medicaid
Uninsured
Accidents
Total Charges
0.04
-0.79a
0.23
-0.33b
Not Fully Recovered
1.03
1.70a
Total Payments
-0.11
-1.33a
0.07
-0.70a
Decline in Health Status
1.26a
1.22a
1.10
1.37a
Cases w/ Positive Charges
Total Charges
Total Payments
New Chronic Conditions
0.02
0.01
0.07
-0.23a
-0.18b
-0.93a
-0.13
-0.78a
a.
Significantly different from privately insured, p<=.01
b.
Significantly different from privately insured, 0.1<p<=.05
Decline in Health Status
a. Significantly different from privately insured, p<=.01
17
18
3
Appendix: Medicaid Managed Care vs
Medicaid Fee-for-Service
Accidents
New Chronic Condition
Medicaid
Managed
Care
Medicaid
FFS
Medicaid
Managed
Care
Medicaid
FFS
Percentage Difference in
Spending (relative to private
insurance)
-0.22c
-0.002
-0.18
0.32c
Short-Term Decrease in SelfReported Health Status (odds ratio
relative to private insurance)
1.29b
1.22c
1.04
1.17
19
4
Download