Professor Elliott Fisher, Professor of Community and Family Medicine at Dartmouth Medical School and Director of the Center for Health Policy Research at the Dartmouth Institute for Health Policy and Clinical Practice spoke at the 5th HARC Forum in November 2009. The purpose of this forum was to consider how we can create and use new evidence about health system performance in order to inform policy and practice. Professor Fisher gave an overview of the internationally leading Dartmouth Atlas Project. This project has documented glaring variations in US healthcare delivery and has radically changed the way we think about effectiveness and efficiency of health care. HARC stands for the Hospital Alliance for Research Collaboration. HARC is a collaborative network of researchers, health managers, clinicians and policy makers based in NSW, Australia managed by the Sax Institute. HARC Forums bring members of the HARC network together to discuss the latest research and analysis about important issues facing our hospitals. For more information visit saxinstitute.org.au.

1. Welcome to the 5th HARC Forum
Monitoring variation in healthcare quality -
an evidence base to improve healthcare
HARC is a as a partnership between
the Sax Institute, Clinical Excellence Commission and the Greater Metropolitan Clinical Taskforce

2. Keynote Speaker
Professor Elliott Fisher
Professor of Community and Family Medicine at Dartmouth Medical School and Director of the Center
for Health Policy Research at the Dartmouth Institute for Health Policy and Clinical Practice
US Dartmouth Atlas of Health Care Project -
monitoring and explaining variation in healthcare to
improve the health system

3. Monitoring Variation in Health Care:
An approach to improving the
evidence base for practice and policy
Elliott S. Fisher, MD, MPH
Professor of Medicine and Community
and Family Medicine
Dartmouth Medical School
Director for Population Health and Policy
The Dartmouth Institute for Health Policy
and Clinical Practice

4. Houston, we’ve got a problem…
The usual suspects:
Rising costs
Uneven quality
Declining access to care
Some looming challenges
Loss of professional authority of physicians
Integrity and relevance of academic medicine

5. Houston, we’ve got a problem…
Traditional diagnoses: Prescriptions:
A shortage of money Spend more
A shortage of doctors Train more
A shortage of economists Focus on prices
Every system is perfectly designed to get the
results that it achieves
Paul Batalden
Insanity: doing the same thing day after day and
expecting different results
Einstein

6. Rethinking health care
Origins of the Dartmouth Institute – and Dartmouth Atlas Project
Science, December 14,
1973; Volume 182, pp
1102-08

7. Rethinking health care
A simple analytic framework; a shared vision
Application of epidemiologic methods to health care services
Define population at risk
Define events
Examine variations across relevant systems -- providers
Ask good questions
Organizational Development
Independent institute within Medical School
Interdisciplinary research group; all with departmental appointments
Governance through a shared vision:
Exploring the causes and consequences of unwarranted variations
Commitment to making a difference (locally, regionally, nationally)
Major long-term funding helps maintain focus

8. Rethinking health care
The Dartmouth Atlas of Health Care
Methods:
Population at risk – over 65
Compare Hospital Referral Regions
Events of interest -- many

9. Rethinking health care
The Dartmouth Atlas of Health Care
Categories of care
Safe and effective care
Preference sensitive care
Supply-sensitive care

10. Rethinking health care
Preference-sensitive care Transurethral Prostatectomy for
Benign Prostatic Hypertrophy
per 1000
15.0
13.0
TURP for BPH
11.0
9.0
7.0
5.0
3.0

11. Rethinking health care
Safe and Effective Care
30 Day Mortality Following CABG
30-Day Mortality Following CABG (%)
12.0
10.0
8.0
6.0
4.0
2.0

12. Rethinking health care
Spending – and supply sensitive care
Medicare Spending
per capita
15,000
Medicare spending per enrollee
13,000
11,000
9,000
7,000
5,000

13. Rethinking health care
Spending – and supply sensitive care
Hospital and Physician Spending
last 2 years of life at
Inpatient + Part B spending per decedent USN&WR top 10 hospitals
120,000
UCLA Medical Center 72,793
100,000 New York-Presbyterian 69,962
How can the best medical care in the world cost twice as much as
Johns Hopkins 60,653
the best medical care in the world?
80,000 UCSF Medical Center 56,859
Uwe Reinhardt
Univ. of Washington 50,716
60,000 Mass. General 47,880
Barnes-Jewish 44,463
Duke University Hosp. 37,765
40,000
Mayo Clinic (St. Mary's) 37,271
Cleveland Clinic 35,455
20,000

14. Rethinking health care
The Dartmouth Atlas of Health Care
1. What we know – 3 case studies
2. What I think we’ve learned
3. Translating evidence to policy
4. Is there reason for hope?

15. Preference Sensitive Care
Building the evidence: the Prostate Patient Outcome Research Team
Exploring the causes of variations in TURP for BPH
Interdisciplinary team; multiple methods
Focus groups of urologists to determine clinical theories
Preventive hypothesis: must operate early in a progressive disease
Quality of life hypothesis
Clinical research: decision-analysis, cohort studies
No survival benefit from surgery
Benefit of surgery depends upon patients’ values (symptoms vs sexual
dysfunction)
Patients’ values differed dramatically
Implications:
Broadly applicable – orthopedics, cardiology, oncology, etc
Need for accurate information on risks and benefits
Structured approach to supporting informed patient choice
Studies have now demonstrated effectiveness of decision aids

16. Preference Sensitive Care
Building the evidence: the Prostate PORT
Interdisciplinary team; multiple methods
Focus groups of urologists to determine clinical theories
Preventive hypothesis: must operate early in a progressive disease
Quality of life hypothesis
Clinical research: decision-analysis, cohort studies
No survival benefit from surgery
Benefit of surgery depends upon patients’ values (symptoms vs sexual
dysfunction)
Patients’ values differed dramatically
Implications:
Broadly applicable – orthopedics, cardiology, oncology, etc
Need for accurate information on risks and benefits
Structured approach to supporting informed patient choice
Studies have now demonstrated effectiveness of decision aids

17. Preference Sensitive Care
Translating evidence into policy
Underlying problems
Inadequate information on risks and benefits of biologically targeted treatments
Provider-dominated decision-making
Remedies
Outcomes research (comparative effectiveness)
Informed patient choice
Policy implications and progress
Major investment in comparative effectiveness research
National standards now include informed choice as core quality measure
Many integrated delivery systems are moving to adopt shared decision-making
States moving to require informed patient choice as legal standard

18. Safe and Effective care
Population: Patients undergoing Coronary Artery Bypass Graft
Providers: Cardiovascular surgery centers in New England
Northern New England
Cardiovascular Disease
Study Group
Origins: threatened public
report of unadjusted CABG
Eastern Maine mortality rates
Medical Center
Fletcher Allen
Health Care
New England rates varied two
Dartmouth Maine
Hitchcock Medical fold: 3.1% to 6.3%
Medical Center Center
Surgeons agreed to collect
Concord Hospital relevant clinical data

19. Safe and Effective care
Population: Patients undergoing Coronary Artery Bypass Graft
Providers: Cardiovascular surgery centers in New England
7
6
In-Hospital Mortality Rate
Adjusted mortality
no less variable: 5
2.3% to 6.3% 4
Near death experience 3
of study group
2
Now over 20 years of 1
experience; 100+
papers published; all 0
sites still participating 1 2 3 4 5
in 3 meetings per year Center
O’Connor et al, JAMA, 1991;266:803-809

20. Safe and Effective Care
Improvement achieved as research advanced
6
Initial intervention-data feedback,
site visits and CQI training
5
Mortality rate (%)
4
3
2 Mode of death study- low
output heart failure major
Process mapping
cause of in-hospital
1 and identification
mortality
of high leverage
areas
0
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year

21. Safe and Effective Care
Translating evidence into policy
Underlying problems
Inadequate data: on patient attributes, process of care and outcomes
Small numbers, lack of follow-up prevent learning
Flawed professional model: individual responsibility and autonomy
Remedy
Technical: registries to support ongoing study of variation in outcomes
Define relevant local care system / teams (to allow comparison with others)
Measure and compare processes and outcomes
Cultural: create teams and systems capable of learning
Engage clinicians in practice-based research and improvement
Enable reflective practice – timely, relevant feedback
Policy implications and progress
Potential benefits – to patients, physicians, managers and policy-makers -- of
practice-based research networks is substantial
Lack of government support remains challenge. Most efforts are voluntary, self-
funded.

22. Variations in spending
Building the evidence base
“How can the best medical
care in the world cost twice
as much as the best medical
care in the world?

23. Variations in spending
Building the evidence base
Health implications of variations in spending
Study population: Medicare patients with AMI, colon cancer, hip fracture
Comparison: across (1) regions; (2) academic medical centers – grouped
according to “intensity” – price and illness adjusted spending.
Measures: content, quality and outcomes of care
Robert Wood Johnson Foundation
National Institute of Aging
California Healthcare Foundation
Aetna Foundation
Wellpoint Foundation
United Healthcare Foundation
(1) Fisher et al. Ann Intern Med: 2003; 138: 273-298
(2) Baicker et al. Health Affairs web exclusives, October 7, 2004
(3) Fisher et al. Health Affairs, web exclusives, Nov 16, 2005
(4) Skinner et al. Health Affairs web exclusives, Feb 7, 2006
(5) Sirovich et al Ann Intern Med: 2006; 144: 641-649
(6) Fowler et al. JAMA: 299: 2406-2412

24. What do they get more of?
Effective Care: benefit clear for all If bar on this side
Reperfusion in 12 hours (Heart attack) higher spending
Aspirin at admission (Heart attack)
regions get more
Mammogram, Women 65-69
Pap Smear, Women 65+
Pneumococcal Immunization (ever)
Preference Sensitive: values matter
Total Hip Replacement
Total Knee Replacement
Back Surgery
CABG following heart attack
Supply sensitive: often avoidable care
Total Inpatient Days
Inpatient Days in ICU or CCU
Evaluation and Management (visits)
Imaging
Diagnostic Tests
0.5 1.00 1.5 2.0 2.5
Ratio of rate in high spending to low spending regions

25. Outcomes and Quality
High spending compared to low spending regions
Physician’s Patient-Perceived
Health Outcomes
Perceptions Quality
Worse Lower satisfaction
No gain in survival
communication with hospital care
Greater difficulty Worse access to
No better function
ensuring coordination primary care
Greater perception No sense that care
of scarcity is rationed
(1) Fisher et al. Ann Intern Med: 2003; 138: 273-298
(2) Baicker et al. Health Affairs web exclusives, October 7, 2004
(3) Fisher et al. Health Affairs, web exclusives, Nov 16, 2005
(4) Skinner et al. Health Affairs web exclusives, Feb 7, 2006
(5) Sirovich et al Ann Intern Med: 2006; 144: 641-649
(6) Fowler et al. JAMA: 299: 2406-2412

26. The paradox of plenty
Pop Quiz….
If we cut spending so that all U.S. regions were receiving the
same per-capita amount as the lowest spending regions,
which of the following would apply:
1. U.S. healthcare spending would fall by 20% to 30%
2. The Medicare Trust Fund might survive a few years past it’s
predicted collapse in 2017 (the year I become eligible).
3. We could send a third of the U.S. healthcare workforce to Africa and
improve the health of both continents.
4. All of the above.

27. Variations in spending
What’s going on? General attributes of U.S. health care
Assumption that more is better
Inadequate information on risks and benefits
Growing tension between science and professionalism --
and -- market approach (health care as a commodity)

28. Variations in spending
What’s going on? Exploring causes of regional variations
Patient
Malpractice
Demand
Less than 10% of
Little difference
difference

29. Variations in spending
What’s going on? Exploring causes of regional variations
Patient
Malpractice Supply & payment
Demand
Less than 10% of
Little difference Powerful influence
difference
Explains less than
50% of difference

30. Variations in spending
What’s going on? Exploring causes of regional variations
Patient
Malpractice Supply & payment
Demand
Less than 10% of
Little difference 32% 65% Powerful influence
difference
4.0 higher 50 higher
40 Explains less than
3.0
30 50% of difference
2.0
20
1.0
10
Regional
Spending Low High Low High
Hospital Beds Medical Specialists

31. Variations in spending
What’s going on? Exploring causes of regional variations
Patient
Malpractice Supply & payment
Demand
Less than 10% of
Little difference Powerful influence
difference
Explains less than
50% of difference

32. Variations in spending
What’s going on? Exploring causes of regional variations
Patient
Malpractice Supply & payment
Demand
Less than 10% of
Little difference Powerful influence
difference
Explains less than
50% of difference
New York Times
August 18, 2006

33. Variations in spending
What’s going on? Exploring causes of regional variations
Patient
Malpractice Supply & payment
Demand
Less than 10% of
Little difference Powerful influence
difference
Explains less than
50% of difference

34. Variations in spending
Exploring causes -- gray area decisions
Evidence-based decisions – drawn from guidelines
Doctors sometimes disagreed – but was unrelated to regional
differences in spending
Gray area decisions (more judgment required):
For a patient with well-controlled high blood pressure and no other
medical problems, when would you schedule the next visit?

35. Variations in spending
Exploring causes -- gray area decisions
Clinical evidence is an important -- but limited --
influence on clinical decision-making.
Policy Environment
Physicians practice within a local organizational
(e.g. payment system)
context that profoundly influences their decision-making.
Payment system ensures that existing capacity
is fully utilized. Physicians adapt to available resources:
more referrals, more admissions, more ICU stays
Local
Organizational Context
Consequence: reasonable individual clinical and local (e.g. capacity - culture)
decisions lead, in aggregate, to higher utilization rates,
greater costs -- and inadvertently -- worse outcomes
The more complicated care
becomes, the more likely
Clinical Physician - Patient
mistakes are to occur.
Evidence Encounter
Hospitals are dangerous places if Professionalism
you don’t need to be there.

36. Culture? Capacity? Both?
Differences in spending and practice across top academic centers
Medicare beneficiaries with chronic illness, 2001-2005

37. Variations in spending
Exploring causes: case studies beginning to shed light
2006 Spending 92-06 Growth
McAllen $14,946 8.3%
La Crosse $5,812 3.9%

38. Variations in spending
Exploring causes: case studies beginning to shed light
“Here … a medical community came to
treat patients the way subprime mortgage
lenders treated home buyers: as profit
centers.”
Atul Gawande
2006 Spending 92-06 Growth
McAllen $14,946 8.3%
La Crosse $5,812 3.9%

39. Variations in spending
Exploring causes: case studies beginning to shed light
“Here … a medical community came to
treat patients the way subprime mortgage
lenders treated home buyers: as profit
centers.”
Atul Gawande
2006 Spending 92-06 Growth
McAllen $14,946 8.3%
La Crosse $5,812 3.9%
“…a culture that focuses on
the wellbeing of the
community, not just the
financial health of our system.”
Jeff Thompson, MD
CEO Gunderson-Lutheran
La Crosse, WI

40. What I think we know
Putting the pieces together: the IOM system of effect
La Crosse McAllen
Aims
Micro-system Organization
How care is provided e.g. capacity, policies
to each patient practices, norms
Environment
e,g, payment, regulations
measures, culture
Institute of Medicine: Crossing the Quality Chasm

41. 41
What I think we know
Underlying problems – and principles to guide reform
Underlying problem Key principles
Confusion about aims – what Clarify aims: Better health, better care
we’re trying to produce lower costs
Bad data allow MDs to discount it, Better information that engages
and public to assume that medicine physicians, creates tension for change,
is science and that more is better. supports improvement; informs consumers
Flawed conceptual model. Health New model: It’s the system. Establish
is produced by individual actions of teams and organizations accountable for
“good” clinicians, working hard. aims and capable of improving practice
Wrong incentives reinforce model, Rethink our incentives: Realign
reward fragmentation, induce more incentives – both financial and
care and entrepreneurial behavior. professional – with aims.

42. Translating evidence to policy
Aims and Performance Measurement
Emerging alignment on aims: National Priorities Partners
Improving population health
Improving safety & reliability and coordination of care
Engaging patients in managing their care and making informed decisions
Eliminating overuse
Performance measurement: the critical lever
National Quality Forum “Episode measurement framework”
Key notions
Core question: how did the patient do over the relevant time-course?
Value: best judged from the patient’s perspective; is multidimensional
Requires organizational accountability – over time

43. Translating evidence to policy
Performance Measurement – across episodes
Post AMI Trajectory 1 (T1)
Relatively healthy adult
Assessment of
Preferences Focus on:
• Quality of Life
• Functional Status
Population at Risk
• 20 Prevention Strategies
10 Prevention • Rehabilitation
(no known CAD) • Advanced care planning
Post Acute/
20 Prevention Acute
Rehabilitation 20 Prevention
(CAD no prior AMI) Phase
Phase
20 Prevention
(CAD with prior AMI) PHASE 2 PHASE 3 PHASE 4 Post AMI Trajectory 2 (T2)
Advanced Care Planning Adult with multiple co-morbidities
PHASE 1 Focus on:
• Quality of Life
Living w/ Illness/Disability (T1) • Functional Status
Staying Healthy Getting Better
Coping w/ End of Life (T2) • 20 Prevention Strategies
• Advanced Care Planning
Traditional model Needed model • Advanced Directives
• Palliative Care/Symptom Control
Autonomy Accountability
Individual Responsibility
Episode begins – Shared Episode ends –
Responsibility
onset of symptoms 1 year post AMI
© NQF

44. Translating evidence to policy
Implications for Health Information Technology
Effective registries are thus critical for a learning health system
To learn -- we need to know:
Patient attributes and risks (including biologic markers)
Specific targeted biologic interventions performed
Attributes of system -- delivery methods -- where care provided
Health outcomes, patient experience and costs
Infrastructure would support
Comparative effectiveness research: compare biologic and delivery
system interventions, controlling for patient and local attributes.
Comparative performance assessment: compare providers and local
systems, controlling for patient attributes
Both are critical

45. Translating evidence to policy
Organizational Accountability and Payment Reforms
Accountable Care Organizations
Principles:
Establish provider organizations that can effectively manage the full
continuum of care as a real or virtually integrated local delivery system
Performance measurement – to support improvement and accountability
Payment reform: establish target spending levels; shared savings
Potential ACOs
Integrated delivery systems (Kaiser-Permanente, Group Health)
Physician Networks; Hospital that employ primary care physicians
Insight from research:
Most physicians already practice within coherent local networks
Performance measurement at group level feasible
Feasible to develop spending targets for most U.S. networks
Fisher et al. Creating Accountable Care Organizations, Health
Affairs 26(1) 2007:w44-w57.

46. Translating evidence to policy
Organizational Accountability and Payment Reforms
Early pilots promising; many organizations supportive
Physician Group Practice demonstration successful
Congressional Budget Office scored as cost-saving
Support from key stakeholders has solidified
ACOs accepted as component of current bills
Support for extensive pilots, rapid expansion in House bills
Senate Finance – voluntary program (not pilot) by 2012
Initiatives at state and local level
Brookings-Dartmouth supporting pilot development in multiple sites
Pilots to start January 2010 in two (or more) sites (VA, KY, TX)
Learning collaborative underway with 40+ health systems
Massachusetts, Vermont, others moving forward

47. Rethinking health care
The Dartmouth Atlas of Health Care
1. What we know – 3 stories
2. What I think I know
3. From insight to action
4. Is there reason for hope?

48. 48
Is there reason for hope?
Theories of change – can they help frame our thinking
Health care is a complex adaptive system
Autonomous actors continuously adapt their behavior
System held in place by “attractors”, self-reinforcing behaviors
Change in complex systems occurs through:
Exploring variation and paradox to create a tension for change;
Creating better alternatives (better policies, better models of care)
Supporting interaction and learning – so others can see new ways to go
New attractors (performance measurement, payment, positive deviants)
Implications for research – and policy
Public reporting: creates a tension for change, raises good questions
Policy relevant research: has undermined flawed assumptions;
suggested path toward reform
Registries, practice networks, have developed evidence, engaged
clinicians, engaged local systems and communities

49. Is there reason for hope?
Theories of change – have stimulated new conversations
“How do they do that?” Common themes
conference Shared aims, accountable to community
Everett, WA Portland, ME Strong foundation of primary care
Sacramento, CA Sayre, PA
La Crosse, WI Richmond, VA
Physician engagement as leaders
Cedar Rapids, IA Asheville, NC Organizational support important
Temple, TX Tallahassee, FL
Use of data to drive change
Lighter colors = lower spending

51. Is there reason for hope?
Theories of change – have stimulated new conversations
“How do they do that?” Common themes
conference Shared aims, accountable to community
Everett, WA Portland, ME Physician engagement as leaders
Sacramento, CA Sayre, PA
La Crosse, WI Richmond, VA
Strong foundation of primary care
Cedar Rapids, IA Asheville, NC Organizational support important
Temple, TX Tallahassee, FL
Use of data to drive change
Lighter colors = lower spending High self-efficacy; high morale

52. Have we made a difference?
Trends in important things
in U.S. healthcare
uninsured, spending
1973 2009
1983

53. Have we made a difference?
Trends in important things
in U.S. healthcare
insights, evidence, tests of change
1973 2009
1983

54. 54
Have we made a difference?
Exploring variation has helped to advance knowledge and policy
Underlying problem Key principles
Confusion about aims – what Clarify aims: Better health, better care
we’re trying to produce lower costs
Bad data allow MDs to discount it, Better information that engages
and public to assume that medicine physicians, creates tension for change,
is science and that more is better. supports improvement; informs consumers
Flawed conceptual model. Health New model: It’s the system. Establish
is produced by individual actions of teams and organizations accountable for
“good” clinicians, working hard. aims and capable of improving practice
Wrong incentives reinforce model, Rethink our incentives: Realign
reward fragmentation, induce more incentives – both financial and
care and entrepreneurial behavior. professional – with aims. Question pay
for performance.