Report Cards and Disparities in Post-Acute Care R. Tamara Konetzka, PhD

advertisement
Report Cards and Disparities in Post-Acute
Care
R. Tamara Konetzka, PhD
University of Chicago
Co-author: Rachel M. Werner, MD, PhD
Philadelphia VA and University of Pennsylvania
June 2009
Funding: AHRQ (R01 HS016478-01, PI Werner)
Two National Health Care Priorities
Improving the quality of health care
2. Reducing racial and ethnic disparities
 Can public reporting achieve both?
 Our goal is to answer this question
empirically using evidence from postacute care
1.
IOM: Differences, Disparities, and
Discrimination: Populations with Equal Access
to Health Care
Clinical Appropriateness
and Need
Patient Preferences
Minority
Non-Minority
Difference
Populations with Equal Access to Health Care
The Operation of Healthcare
Systems and the Legal and
Regulatory Climate
Disparity
Discrimination: Biases and
Prejudice, Stereotyping, and
Uncertainty
Potential Effects of Public Reporting
on Disparities

Report cards may reduce disparities
– Information may act as an “equalizer” among
consumers
– Low-quality providers have the greatest
incentive to improve

Report cards may increase disparities
– Consumers have differential access to the
information and ability to understand and use it
– Supply
– Low-quality providers may not have the
resources to invest in quality improvement
Background:
Disparities in Nursing Home Care


Nursing homes more segregated than other health care
sectors, mirroring residential segregation
Nursing homes in poor neighborhoods are lower quality
and tend to… (Mor, 2004)
–
–
–
–
–

Be high Medicaid
Be high minority
Have more deficiencies
Have lower staffing
Be more likely to be terminated from Medicare/Medicaid
Blacks and poorly educated patients more likely to enter
low-quality facilities for post-acute care (Angelelli,
Grabowski, Mor 2006)
Specific Research Questions
Do whites, blacks, and Hispanics benefit
differentially from improvements in postacute care quality under NHC?
 To what extent is the differential
improvement due to within-facility
changes vs. changes in the site of care?
 Does education mediate these
differences?

Data

Minimum Data Set (1999-2005)
– Mandatory for all residents of Medicare- and Medicaid-certified
nursing homes
– Detailed assessment data
– Source to calculate quality measures for Nursing Home Compare
– Used to calculate quality measures over study period

MedPAR
– Claims data on all non-managed-care Medicare beneficiaries
– Used to calculate rehospitalizations and several risk-adjustment
variables



OSCAR
Race=black, white, hispanic
6.3 million admissions from 8,139 SNFs
Reported quality measures
Technical definitions of measures from
CMS
 Follow CMS conventions

– 14-day assessment
– Resident exclusions from each measure
– Facilities with more than 20 cases during
target period
Dependent Variables:
Quality measures
Measure
Definition
Mean (SD)
No pain
% of residents who did not have
moderate or severe pain
76 (15)
No delirium
% of residents without delirium
96 (5)
Improved walking
% of residents whose walking
improved
Rehospitalization
% of residents who had a
rehospitalization within 30 days
since last discharge
8 (7)
22
Race
White (not hispanic)
 Black (not hispanic)
 Hispanic
 MDS instructions are to use self-report
 We omit Asian/Pacific Islander and Native
American due to small sample sizes.

Control variables

For all regressions:
– Age, gender, cognitive status, functional
status, comorbidities, past admissions, BMI,
end-stage disease, pain/delirium/walking
upon admission

For delirium:
– Prior residential history (as specified by CMS
for this measure)

Education: less than high school; high
school graduate; college graduate
Empirical approach
Qualityijt= β1NHCjt + ξ(NHCjtxRacei)+ θXit + ηj + εijt
– Qualityijt = quality for individual i in SNF j in year t
– NHCjt = indicator of Nursing Home Compare
▪ pre-post (1999-2002 vs. 2003-2005)
▪ set of year dummy variables
– NHCjtxRacei = interactions between race and NHC
– Xit = set of control variables
– ηj = SNF fixed effects
Percent of Residents Without Moderate/Severe Pain (adjusted)
0.77
0.765
0.76
0.755
black
0.75
hispanic
w hite
0.745
0.74
0.735
0.73
Pre-NHC
Post-NHC
Percent of Residents Without Delirium -- adjusted
0.968
0.966
0.964
black
0.962
hispanic
0.96
w hite
0.958
0.956
0.954
Pre-NHC
Post-NHC
Without Moderate/Severe Pain upon Admission
0.84
0.82
0.8
0.78
0.76
black
0.74
0.72
0.7
0.68
0.66
0.64
hispanic
white
Pre-NHC
Post-NHC
Percent of Residents with a 14-Day Assessment
0.675
0.67
0.665
black
0.66
hispanic
w hite
0.655
0.65
0.645
Pre-NHC
Post-NHC
Rehospitalizations, within-SNF, adjusted
0.245
0.24
0.235
black
0.23
hispanic
0.225
white
0.22
0.215
0.21
Pre-NHC
Post-NHC
Education
Education does not mediate the effects
 Stratification by education produces
similar results
 Consistent with most quality differences
by race being due to site of care

Conclusions
No clear conclusions as to effects of NHC
on disparities in post-acute care
 Use of 14-day assessments for quality
measurement remains problematic
 Broader-based measures such as
rehospitalization may be preferable

Download