Medicare Advantage Payment Extra Payments, Enrollment & Quality of Care Lauren Hersch Nicholas

advertisement
Medicare Advantage Payment
Extra Payments, Enrollment &
Quality of Care
Lauren Hersch Nicholas
Columbia University
AcademyHealth
June 3, 2007
Research funded by the John A. Hartford Foundation Hartford Doctoral Fellows Program and the
Commonwealth Fund
Outline
Relationship between payments to Medicare
Advantage plans and enrollment
 Quality of Care in Medicare Advantage vs. FFS
 Effects of Extra payments on quality of care in
Medicare Advantage

Methods: Data Sources




Medicare Enrollment file provides average
demographics at county-insurance status level
Area Resource File for county health system
characteristics
CMS Medicare Advantage Ratebooks
State Inpatient Dataset from Healthcare Cost and
Utilization Project
 Repeated
cross-sections 1999-2004
 inpatient discharge abstract for universe of
hospitalizations
 AZ, FL, NJ and NY data
Payment Model
Fixed Effects Regression
MAc,t = β1Pay c,t + β2Rate c,t + β3X c,t + β4C + β5Y + ε c,t
Where MA is Medicare Advantage Penetration
Pay is the extra payment amount (per enrollee per month)
Rate is a vector containing the payment rate and its square
X is a vector of county health systems characteristics including a constant (total
doctors, general practitioners, hospitals, hospital beds, ambulatory care
centers, skilled nursing facilities, HMO headquarters, per capita income)
C is a vector of county fixed effects
Y is a vector of year fixed effects
Counties weighted by number of Medicare enrollees
Results: Payment Rates and
Enrollment
Payment Rate
-0.0022
(.0008)**
Rate Squared
0.000003
(0.0000006)**
F test of Instruments
12.45
First-Stage F-test
51.85
Enrollment in Medicare Advantage is increasing with
payment rates up to $807 per enrollee per month (through
2004)
Measuring Quality: AHRQ
Hospitalization Classifications









Preventable: Could be managed/prevented by
effective primary care
Higher rates indicate inadequate quality of or
access to outpatient care
Asthma
Chronic heart disease
Congestive heart failure
Diabetes Complications
Hypertension
Kidney/Urinary Infections
Pneumonia
Source: United States Agency for Healthcare Research and Quality (2003). Data for Monitoring the Health Care Safety Net
Measuring Access: AHRQ
Hospitalization Classifications







Referral- Sensitive: Discretionary, often
elective, technology-intensive procedures,
require referring physician
Low rates of procedures may suggest
barriers to service use
Coronary angioplasty
Coronary Bypass
Hip Replacement
Organ Transplant
Pacemaker insertion
Source: United States Agency for Healthcare Research and Quality (2003). Data for Monitoring the Health Care Safety Net
Data: County-Insurance Status
Level Cells




ICD-9-CM diagnostic codes used to identify
preventable, referral-sensitive and reference
hospitalizations
Restrict sample to adults 65+ with FFS Medicare
or MMC as primary payer
Calculate rates of each type of hospitalization
per 1,000 enrollees
Weight cells by number of enrollees
Positive Selection or Better Care? Rates of Hospitalization for MA and FFS Enrollees
120
113.5
Rate per 1,000 Enrollees
100
77
80
70.3
60
43.6
40
20
15.2
10.4
0
Preventable
Referral-Sensitive
Type of Hospitalization
Source: State Inpatient Data, AZ, FL, NJ,
NY, 1999 - 2004
Medicare Advantage
FFS Medicare
Reference
Quality Models:
MA vs. FFS
Hc,i,t = β0 + β1MMCc,t + β2X c,i,t + β3M c,t + β4Yr + ε c,i,t (2)
where MMC status is estimated using payment rates
Effects of Extra Payments on MA Quality
Hc,i,t = β0 + β1Extrac,t + β2X c,i,t + β3M c,t + β4Yr + ε c,i,t (2)
limited to MA sample
Extra Payments = MA Rate - (FFSA /Avg RiskA
+ FFSB/Avg
RiskB)
Results: MMC vs. FFS
Hospitalization Rates
(MMC Coefficient)
Preventable
Mean
Rate
65
(13.2)
OLS
Year FE
-23.1
(3.3)**
4.10
(19)
Referral-Sensitive
106
(48.8)
-31.4
(6.8)**
-39.1
(35.8)
Reference
14
(3.2)
-3.1
(0.64)**
0.18
(2.8)
* Significant at 5%
IV
** Significant at 1% Clustered standard errors in parentheses
Results: Effect of Extra Payments
Hospitalization Rates
Preventable
Payment Effect at
Coefficient $121
.011
1.33
(.0089)
Referral-Sensitive
.051
(.031)
6.2
Reference
.001
(.002)
.12
* Significant at 5%
** Significant at 1% Clustered standard errors in parentheses
Summary - (1)
No significant differences in hospitalization
rates once we address selection bias
 IV point estimate for referral
hospitalizations relatively unchanged, may
indicate reduced access to elective
procedures under MMC
 MMC enrollment may not provide higher
quality preventative care relative to FFS

Summary - (2)
Payments to MMC plans in excess of
average FFS spending are associated with
more hospitalizations of all kinds
 Difference is not statistically nor
substantively significant
 Extra payments do not appear to improve
quality or access for MA enrollees

Implications
Little significant evidence of quality
differences between MMC and FFS
 Extra payments to Medicare Advantage
plans may not buy improved quality, but
little evidence that enrollees trade quality
for lower out-of-pocket spending either

Download