Medicare Managed Care and Primary Care
Quality: Examining Racial/Ethnic Effects
across States
Jayasree Basu, Ph.D.
AcademyHealth, 2009
Session title: Disparities
Background
 Medicare Modernization Act of 2003 sparked renewed
interest in Medicare managed care (MMC)
 Medicare spends about $10 billion more each year on
beneficiaries enrolled in MMC plans —calls for
evaluation
 Understudied topic: program’s effectiveness in
reducing racial and ethnic disparities in quality of
health care delivery and access
Study Objective
 To assess the role of MMC plans in providing quality
primary care compared to FFS Medicare in three
states (NY, CA, FL) across three racial ethnic groups
(White, African American, and Hispanic)
 The performance will be measured in terms of
providing better quality primary care, defined as
lowering the risk of preventable hospital admissions
Hypotheses
 Managed care plans will reduce preventable
hospitalizations (PH) across racial groups through care
coordination and provision of preventive care
 Relative to FFS, improved care coordination in HMO
plans will reduce PH for minorities more than whites
 H1: PHMA < PH
FFS
 H2: PHMA(Minorities|Whites) < PH
FFS(Minorities|Whites)
Patient Selection
 States : NY, CA, FL, Year: 2004
 Hospitalized Medicare FFS and Medicare advantage (MA) plan
enrollees (Age 65 and over)
 Patient level data on MA versus FFS enrollment as recorded in
the confidential files of discharge database of the three states
All three states had higher penetration than US average in 1994
increasing further by 2000. CA reached 54% penetration rate
By 2000.
MMC penetration by State and
US
2000
CA
NY
FL
US
1998
1996
1994
0%
State source: InterStudy
20%
40%
60%
Data
NY
CA
FL
 Hospital discharge data (HCUP-SID, AHRQ) for elderly
Medicare (age 65+), 2004
 Medicare managed care plans available in 2004 were
predominately HMO types (96-99%)
 Inpatient discharge data linked to area resource files, US
Census, AHA, Interstudy, HRSA’s spatial data warehouse
 Multivariate cross sectional framework with patient-level
data for each State
Variables

Individual patient characteristics:
–
–

Three Racial ethnic groups

Whites

African Americans (AA)

Hispanics
Type of insurer, age groups, gender, severity of illness, indirect
severity indicators, severity*HMO, race*HMO
Contextual data: socio-demographic conditions and provider
characteristics in each Primary care service area (PCSA) where
patients live
–
PCSA : smallest geographic area considered as discrete service area for
primary care--validated in previous research
Design
 PH admissions compared with admissions for “marker
conditions” for each State in each Racial group
PH
Marker
• Sensitive to primary
care
• Urgent, insensitive to
primary care
• Ex: Severe ENT
infections, UTI,
COPD, Tuberculosis,
Hypertension etc.,
• Appendicitis with
appendectomy, acute
MI, gastrointestinal
obstruction, fracture of
hip/femur
Analysis
 Unit of analysis = patients
 Multivariate Logistic regression models with odds of PH
admission compared to marker admission
– by each Racial group and Pooled models
– multilevel data, adjusting for area-level clustering, by state
RESULTS
Odds ratios < 1 in all racial groups in each
state, and lower for minorities than whites
Odds Ratios
CA
FL
PH versus
marker
0.85
0.93
0.89
0.82
0.75
(PH/marker) MA = (PH/marker) FFS
---------------------------------------------------------0.82
0.70
0.71
1.30
1.20
1.10
1.00
0.90
0.80
0.70
0.60
0.50
0.40
NY
White
AA
Hispanic
% Difference in Odds of PH Admissions:
MA versus FFS Enrollees
White
AA
Hispanic
CA
-18
-30
-29
NY
-7
N.S.
-15
FL
-11
-18
-25
CA and FL had greater reductions in PH among MA enrollees by
racial groups, minorities in particular
Odds Ratios of PH Admissions versus Marker
Admissions: Race*HMO Interactions
AA / White
CA
NY
FL
0.83 (p=.000)*
N.S.
0.90 (p=.153)
Hispanic / White
0.87 (p=.012)*
N.S.
0.81 (p=.000)*
*MA plans significantly reduced racial gaps in PH/marker in CA and FL
Summary

MA plans reduced PH rates relative to marker rates

In all racial groups, minority MA enrollees had lower risks
of PH admissions (versus marker admissions) relative to
their FFS counterparts

Minority MA enrollees had greater reductions in PH
admissions relative to white MA enrollees

CA and FL: Interaction effect in pooled model shows
statistically significant difference between minorities and
white MA enrollees in PH rates
Conclusion

MA plans had beneficial impacts in terms of improving
quality primary care by reducing preventable
hospitalizations in all three states

The benefit also spilled over to different racial and ethnic
subgroups

In CA and FL, MA plans resulted in significant reductions
in racial and ethnic differences in preventable
hospitalization rates
Implications

MA plans added value to the quality of primary care to the
elderly by racial groups

Greater reduction of PH rates among minority subgroups
indicates favorable role of MA plans in achieving
racial/ethnic equalities

Care management provided in Medicare HMOs may have
implications for future strategies to reduce racial ethnic
gaps and improve quality of primary care

Future research should evaluate the MMC programs by
plan types using more recent data
APPENDIX
RESULTS
Odds Ratios of PH Admissions (relative to
Marker): MA VS. FFS enrollees
CA
NY
FL
White
AA
Hispanic
0.82
0.70
0.71
0.93
N.S.
0.85
0.89
0.82
0.75
PH Admissions
severe ENT infections
chronic obstructive
pulmonary disease
diabetes
convulsions
hypoglycemia
kidney infection
asthma
angina
congestive heart failure
bacterial pneumonia
tuberculosis
hypertension
cellulitis
gastroenteritis requiring
hospitalization
urinary tract infection
dehydration
pelvic inflammatory disease
nutritional deficiencies
certain dental conditions
Marker Admissions:
The Comparison Group
 Diagnoses for which provision of timely and effective
outpatient care is likely to have little impact on the need
for hospital admission
 Agreement among practitioners on clinical criteria for
admission:
appendicitis with appendectomy
acute myocardial infarction
gastrointestinal obstruction
fracture of hip/femur