Title: The Geography of Health Care Utilization and Outcomes
1The Geography of Health Care Utilization and
Outcomes
- Jonathan Skinner
- Department of Economics, Dartmouth College
- The Dartmouth Institute for Health Policy and
Clinical Practice, Dartmouth Medical School - jon.skinner_at_dartmouth.edu
- November 9, 2010
- Institute of Medicine, Washington DC
2The Focus of This Talk Costs and Quality in the
Medicare Population
Congressional Budget Office, June 2010 (revised
Aug 2010), Extended baseline .
3Two Policy Questions
- What are the causes of geographic variation in
health care? - What are the consequences of geographic variation
in health care? - Todays objective Understanding how patient
cohorts can help us to answer each question
4Unexplained regional variation in Medicare
expenditures
Source Zuckerman, Berenson, Hadley, 2010
5Where do we go from here?
Source Zuckerman, Berenson, Hadley, 2010
6Choices, choices
- Complete risk adjustment using Medicare claims
data - Further risk adjustment using ecological (state
or county) data - Better risk adjustment approaches (biomarkers,
etc.) - Considering cohorts of patients with similar
diseases
7Diagnosis creep in high-intensity regions
- High-diagnosis
- 19 lower risk-adjusted costs
- 15 better risk-adjusted outcomes
8Choices, choices
- Complete risk adjustment using Medicare claims
data - Further risk adjustment using ecological (state
or county) data
9Potential Pitfalls of Using Ecological
Risk-Adjusters
Cutler and Sheiner, AER May 1999
10But Market-Level Explanations for Utilization
are Valid First, Chicago.
- Chicago-area hospitals sacrifice revenue as they
prepare for health care reform - By Mike Colias, Crains Chicago Business, July
12, 2010 - "It's not about building new facilities or
mergers and acquisitionsThe hospitals that will
fare best are the ones that are dealing most
seriously with getting ready for these
reimbursement changes and partnering with the
right doctors.
Michael Nugent, Navigant Consulting
11Then New York Fighting for Market Share
- Soaring cancer-care costs strain budgets -
Hospitals pour millions into new devices, talent
as margins thin - By Judith Messina, Crains New York Business,
August 22, 2010 - All told, the city's major hospitals have spent
more than 2 billion on cancer research and
treatment over the past five years, in a race to
carve out a bigger piece of a fast-growing pie. -
12Choices, choices
- Complete risk adjustment using Medicare claims
data - Further risk adjustment using ecological (state
or county) data - Better risk adjustment approaches (biomarkers,
etc.) - Considering cohorts of patients with similar
diseases
13Possible Low Variation Cohorts
- Hip fracture
- AMI
- Stroke
- Colon/Lung Cancer
- Risk-adjusted end-of-life cohorts (by condition
and presence of multiple conditions)
14Example AMI Cohorts
- One-year after AMI
- Risk adjustment for type of AMI (location of the
infarct), comorbidities (e.g., diabetes, COPD,
cancer, dementia), zip code income, poverty in
income, price adjustment, age-sex-race. - Currently Part A through 2005 could be Parts
A,B, and D through 2008 or 2009.
15Risk Price-Adjusted AMI and End-of-Life
Expenditures 2000-05 by Hospital
One-Year AMI Spending
N 1985, At least 400 AMIs
16Two Policy Questions
- What are the causes of geographic variation in
health care? - What are the consequences of geographic variation
in health care?
17Provider-Specific Measures of Quality Spending
Typically Look Like This
Survival/Quality
Spending
18This is waste
Survival/Quality
Spending
Low Cost
19But this waste is even more important!
Survival/Quality
Best practice
Spending
20A Simple Graph Spending vs. Survival/Quality of
Life
Survival/Quality
You are here
A
Spending
21A Tale of Two Hospitals Every New Adoption is
Medically Effective
Survival/Quality
Hospital X
Hospital Y
A
Spending
But hospital X gets better outcomes at lower
costs!
22Causes a Negative Correlation Between Spending
and Outcomes
Survival/Quality
Hospital X
Hospital Y
A
Spending
23Differences in Efficiency A Concrete Example
Some Firms Get Double the Output at the Same
Cost!
Deviation from Mean in Total Factor Productivity
Syverson, C., Market Structure and Productivity
A Concrete Example, JPE 2004.
24New Approaches Percent Men Age 80 Receiving PSA
Screening by HRR
Source Bynum, J., JAGS 58(4) April 2010
25One measure of a bad outcome, 2006-07
Feeding Tube Use ()
Brigham-Womens (MA) 4.7
Lawnwood Regional (FL) 37.5
UCLA 0.0
Cedars Sinai 14.6
US 6.3
Cohort Nursing home patients with advanced
dementia admitted to hospital
26Yet more ways to measure outcomes
- Changes over time in measures of care
- Deaths in hospitals or hospice, 2003-07 (D.
Goodman et al.) - Low or high quality prescription (Part D)
- Y. Zhang et al. (2010), N. Morden et al. (2010)
- Burdensome transitions for nursing home patients
(J. Teno, et al, 2010) - Multiple hospitalizations for UTI, pneumonia in
last 120 days - Transitions in the last 72 hours of life
- Lack of continuity among nursing homes
27Conclusions
- Challenges in measuring utilization and outcomes
but necessary to reward efficiency - Serious limits to current risk-adjustment
measures - Defining meaningful cohorts of patients
reasonable step forward - Looking ahead new approaches to measuring
outcomes