Technology Diffusion, Hospital Variation, and Racial Disparities Among Elderly Medicare Beneficiarie

1
Technology Diffusion, Hospital Variation, and
Racial Disparities Among Elderly Medicare
Beneficiaries 1989-2000

• Peter W. Groeneveld, MD, MS
• Sara B. Laufer, MA
• Alan M. Garber, MD, PhD

2
Healthcare Disparities and Geographic Variation

• Racial disparities in medical procedure use may
  partially be explained by small area geographic
  differences in procedure availability.
• Explained 95 of the difference in knee
  replacement rates between white and latina
  women.
• Within localities, there are large differences in
  technology utilization rates among hospitals.
• Skinner J, et al. Racial, ethnic, and
  geographic disparities in rates of knee
  arthroplasty among Medicare patients.
• N Engl J Med 20033491350-9.
• Selby JV, et al. Variation among hospitals in
  coronary-angiography practices and outcomes after
  myocardial infarction in a large health
  maintenance organization. N Engl J Med.
  19963351888-1896.

3
Research Questions

• Do differences in major medical procedure rates
  among hospitals help explain racial disparity in
  healthcare?
• Do hospitals with larger black inpatient
  populations provide more/less equal care?
• As medical technologies diffuse through the
  marketplace, do racial disparities decrease?

4
Setting

• 20 random selection of elderly Medicare
  beneficiaries hospitalized between 1989-2000.
• Medicare Provider Analysis and Review (MEDPAR)
  administrative records.

5
Selection Criteria for Procedures

• Performed in sufficient volume among the elderly
  throughout 1989-2000.
• Substantial growth in volume and in number of
  hospitals offering procedure during the 1990s.
• Performed in inpatient setting.
• Influenced DRG assignment.

6
Emerging Procedures and Their Indicator Diagnoses
7
Cohort Formation / Outcomes

• Hospitalization with indicator diagnosis,
  1989-2000.
• Linked to subsequent hospitalizations within 90
  day period.
• Outcomes
• Procedure within 90 days of admission or
• Death prior to 90 days without procedure or
• Survive 90 days without procedure.

8
Multinomial Logit Model

• Logit (outcome) ß1race ß2t yeart ß3t
  race yeart ß4 black9_20
  ß5t black9_20 yeart ß6 blackgt20
  ß7t blackgt20 yeart ?k covariates e
• Covariates sex, age, zip code-level
  income/education, Charlson comorbidity, academic
  hospital
• Standard errors adjusted for data clustering by
  hospital and ZIP code.

9
Sub-cohorts
10
Which Patients are Admitted to Hospitals with
gt20 Black Inpatient Populations?
of whites/blacks admitted to hospitals with
gt20 black inpatient population
11
Racial Disparity for Five Emerging Technologies
1.4
1.2
1
Odds ratio for blacks receiving procedure
0.8
0.6
0.4
0.2
0
Aortic Valve
IMA CABG
Dual-Chamb
Vena Cava
L/LS Spine
Replcmt
Pacer
Interrpt
Fusion
12
Procedure use at hospitals with gt20 black
inpatients
13
Comparison of Disparities at Hospitals with gt20
or lt9 Black Inpatients
14
Conclusions

• Hospitals with larger black inpatient populations
  had generally lower procedure rates for their
  patients.
• These hospitals also had greater levels of racial
  disparity.
• Substantial racial disparities persisted in the
  use of several emerging medical technologies
  during the 1990s.

15
Limitations

• Administrative data were insufficiently detailed
  to determine who met definitive procedural
  criteria.
• Possible that systematic differences exist
  between the accuracy and detail of MEDPAR records
  for whites and blacks.

16
Implications

• The quality and innovativeness of care provided
  by hospitals with gt20 black inpatient
  populations is critical to the provision of more
  equal healthcare.
• Policy initiatives to improve racial disparities
  in healthcare should concentrate on the mediating
  role of these hospitals.

17
END
18
Procedure Rate Growth 1989-1999
Procedures per 10,000 elderly Medicare
Beneficiaries
19
Covariates

• Race (black or white).
• Sex, age, ZIP-code-level race-specific
  income/education, urban location.
• Charlson comorbidity index.
• Black inpatient population () of center in which
  patient hospitalized.