What Physician Shortage?An Evidenced-Based Perspective

David C. Goodman, MD MS

Professor of Pediatrics and ofHealth Policy

The Center for Health Policy ResearchDartmouth Medical School

Hanover, NH

May 2009

Workforce Research at The Center for Health Policy Resarch

• John Wennberg, MD MPH • Elliott Fisher, MD MPH• Sam Finlayson, MD MS• Chiang-hua Chang, MS• George Little, MD• Therese Stukel, PhD• Jonathan Skinner, PhD• Julie Bynum, MD• Scott Shipman, MD MPH• Douglas Staiger, PhD• James Weinstein, MD MS• Dongmei Wang, MS• Sally Sharp, SM• Stephanie Raymond• Phyllis Wright-Slaughter, MHA• Daniel Gottlieb, MS• Kristen Bronner, MA• Megan McAndrews, MBA, MS• David Bott, PhD• Stephen Mick, PhD (VCU) • Jia Lan, MS• Nancy Marth, MS• Jon Lurie, MD MS• Ken Schoendorf, MD MPH (CDC/NCHS)

• The Robert Wood Johnson Foundation

• Mithoefer Center for Rural Surgery

• National Institute on Aging

• Health Resources and Services Administration

• WellPoint Foundation

• Aetna Foundation

• United Health Foundation

• California HealthCare Foundation

Collaborators Support

The Workforce Crisis

• Why do many believe that there is a workforce crisis?

• Would patients benefit from higher physician training rates?

• Should we “interfere” with market forces?

• How should we build our workforce and training programs?

U.S. Workforce Policy: From Surplus to Shortage

• 1997: Surplus of physicians.

• 2005: Council on Graduate Medical Education 16th report declares an impending physician shortage.

• 2006: AAMC recommends 30% increase in medical school enrollment and lifting of the Medicare GME funding cap.

Physician Training - 2000

US Medical Grads~16,000 per yr

Graduate Med Educationentry = ~22,000 per yr

Clinical Practice

International Medical Grads~6,000 per year

IncreaseGraduate Medical

Education

Medicare GME: ~$8 billionplus Medicaid $$

IncreaseUS Medical School

Enrollment

Total Revenue $~60 billionless care/research $~19 b

What is the evidence for an impending shortage?

• Growing population, particularly of the elderly.

• Increases in age-specific utilization rates.

• Economic expansion: “GDP is destiny”.

• In other words, “demand” is increasingly rapidly; failing to anticipate “demand” with more physicians will lead to a shortage.

AAMC Projected National Supply & Shortfall of Physicians with GME Expansion

Source: Salsberg. International Medical Workforce Meeting. 2008.

Baseline Supply

Additional Supply from

Robust GME Expansion

Shortfall

How large is the shortfall?

AAMC Projected National Supply & Shortfall of Physicians with GME Expansion

Source: Salsberg. International Medical Workforce Meeting. 2008.

Baseline Supply

Additional Supply from Robust GME Expansion

Shortfall

How large is the shortfall?

The 2020 “Shortfall” in Physicians

Council on Graduate Medical Education. Sixteenth Report. 2005.

0

200,000

400,000

600,000

800,000

1,000,000

1,200,000

1,400,000

Supply Demand Need

1,076,000

972,000

1,240,000

1,027,00

1,173,000

1,086,000

Physician Supply, Demand, and Need in the U.S. 2020

“Shortfall” = ~90,000 or ~10%

An alternative approach:What are the desirable outcomes of investing

in the medical workforce?

• Access: to care when it is wanted and needed.

• Quality: Care that is technically excellent and personally compassionate.

• Outcomes: Care that improves the health and well being of patients and populations.

• Costs: Care that is affordable to the patient and to society.

If we agree on the desirable outcomes...

Then the question is:

What are the most effective and efficient ways to achieve these ends?

Is there evidence that access, quality, and outcomes are sensitive

to physician supply, per se?

www.dartmouthatlas.org

John Wennberg Lead Author

Co-authors:

Elliott Fisher, MD MPH

David Goodman, MD MS

Jonathan Skinner, PhD

The Per Capita Supply of Physicians Varies ~200% Across Regions

Post-GME clinicians per 100K population age sex adjusted - 2005

Dartmouth Atlas Hospital Referral Regions

50

75

100

125

150

175

200

225Specialists Generalists

10%

200%

40

50

60

70

80

90

100

110

120

Clinically Active Physicians per 100,000 ResidentsClinically Active Physicians per 100,000 Residentsby Hospital Referral Region (2005), age-sex adjustedby Hospital Referral Region (2005), age-sex adjusted

215215 to to 316316 (57)(57)200200 to < to < 215215 (54)(54)185185 to < to < 200200 (63)(63)170170 to < to < 185185 (67)(67)118118 to < to < 170170 (65)(65)Not PopulatedNot Populated

Regional variation in physician supply is not explained by:

• Patient health status or health riskChan R, et al. Pediatrics 1997.Goodman D, et al. Pediatrics 2001.Wennberg J. Ed. Dartmouth Atlas of Health Care. Various editions. 1996 - 2006.Fisher E, et al. Ann Int Med 2003.

Neonatologists

**

**

*

* * *

*

**

*

*

**

**

*

* *

*

*

* *

*

*

***

*

*

*

*

*

***

*

*

*

*

*

*

**

*

**

*

*

*

*

*

*

*

*

* *

*

*

*

*

*

**

* *

*

*

*

*

* *

**

*

* *

*

*

* *

*

*

**

***

*

*

*

**

*** * *

*

*

*

*

*

**

*

**

* *

**

*

**

*

*

*

***

**

*

*

*

*

*

**

*

**

*

*

*

*

*

* *

*

*

*

*

*

*

*

**

*

*

*

*

* *

*

*

**

*

**

*

*

*

*

*

**

**

**

** *

*

*

*

*

*

**

*

*

*

*

*

*

*

**

**

*

**

*

* *

*

*

*

*

*

*

*

*

*

*

*

*

**

*

**

*

***

*

*

*

*

*

*

*

*

**

*

* ***

*

***

*

*

**

* *

0

5

10

15

20

25

30

4 5 6 7 8 9 10 11 12 13

Percent Low Birth Weight

Neo

nato

logi

sts

per

10,

000

birt

hs R2=0.04

Goodman, et al. Pediatrics, 2001.

Are neonatologists located where newborn needs are greater?

(246 Neonatal Intensive Care Regions)

There is virtually no relationship between regional physician supply and health

needs.

2.0

4.0

6.0

8.0

10.0

12.0

3.0 6.0 9.0 12.0 15.0 18.0

Acute Myocardial InfarctionRate per 1,000 Medicare Enrollees

Car

diol

ogis

ts p

er 1

00K

Source: Wennberg, et al. Dartmouth Cardiovascular Atlas

There is virtually no relationship between regional physician supply and health

needs.

Are cardiologists located where cardiac needs are greater?(306 Hospital Referral Regions, Dartmouth Atlas)

Regional variation in physician supply is not explained by:

• Patient health status or health risk• Patients preference for care

Fisher E, et al. Ann Int Med 2003.NIA-CMS beneficiary survey, forthcoming.

No difference in preferences for aggressive care (dying in hospital, mechanical ventilation, or drugs that would lengthen their life, but make them feel worse)

No differences in concerns about getting too little (or too much) treatment

So what?

Despite the idiosyncratic location of physicians...

maybe more physicians leads to better health outcome.

Do areas with higher physician supply have better health outcomes?

Source: Goodman, Fisher, et al. New Engl J Med, 2002.

• Logistic models 1995 USbirth cohort

• N = 3.8 million live births

• Dependent variable:28 day mortality

Very Low Low Medium High Very High0.8

0.9

1

1.1

Mortality Adj.Odds Ratio

Quintile of Physician Capacity in Neonatal Intensive Care Regions

Neonatologists

Better Outcomes Inefficient Care

Beyond a very low supply, outcomes are

insensitive to physician supply.

With Similar Outcomes, Many Health Care Systems Deliver Care with Far Fewer Physicians

Standardized Physician Labor Input During Last 6 Months of Life Among Medicare Cohorts

(Full Time Equivalents per 1,000 beneficiaries)

Mean Age

Total FTEs Primary Care

Medical Specialists

NYU Medical Center 82 28.3 8.8 15.0

RWJ University Hospital (NJ) 80 19.8 4.3 12.2

Montefiore Med Center (NY) 83 16.5 6.5 7.1

MA General Hospital 80 15.3 6.3 5.5

Johns Hopkins Hospital 77 12.2 5.0 3.9

Yale-New Haven 82 10.6 3.4 4.4

UC, San Francisco 81 9.4 4.7 3.2

Mayo, Rochester MN 81 8.9 3.0 3.9

Strong Memor., Rochester,NY 81 8.1 3.8 2.4

Source: Goodman, Wennberg, Chang, Health Affairs,March/April 2006.

FTE Primary Care Physician Labor Inputs per 1,000 Decedents During the Last Two Years of Life

3.03.0

7.07.0

11.011.0

15.015.0

19.019.0

23.023.0

FT

E p

rimar

y ca

re la

bor

inpu

ts p

er 1

,000

FT

E p

rimar

y ca

re la

bor

inpu

ts p

er 1

,000

Cedars-Sinai Med Ctr 14.6NYU Medical Center 13.2Mass General 11.5Elliot Hospital 9.8Fletcher Allen 8.1Catholic Med Center 7.7Maine Medical Center 7.0Mayo Clinic (St. Mary's) 6.8Dartmouth-Hitchcock 6.5

FTE Medical Specialist Labor Inputs per 1,000 Decedents During FTE Medical Specialist Labor Inputs per 1,000 Decedents During the Last Two Years of Lifethe Last Two Years of Life

4.04.0

8.08.0

12.012.0

16.016.0

20.020.0

24.024.0

28.028.0

32.032.0

FT

E m

edic

al s

peci

alis

t la

bor

inpu

ts p

er 1

,000

FT

E m

edic

al s

peci

alis

t la

bor

inpu

ts p

er 1

,000

Cedars-Sinai Med Ctr 31.6NYU Medical Center 30.1Mass General 11.7Maine Medical Center 10.0Mayo Clinic (St. Mary's) 8.9Fletcher Allen 8.8Elliot Hospital 7.7Catholic Med Center 6.9Dartmouth-Hitchcock 6.9

Are Technical Quality and Patient Satisfaction Better with More Physicians?

Physicians Per Capita

Lowest Quintile

Highest Quintile

Ratio highest to

lowest

Total physicians per capita by Hospital Referral Regions (2005)

169.4 271.8 1.60

CMS Compare Composite Scores (2005)

Acute myocardial infarction 91.0 93.1 1.02

Congestive heart failure 84.1 88.6 1.05

Pneumonia 79.5 79.2 1.00

Goodman DC, Fisher ES. New England J Med, 2008.

Are Technical Quality and Patient Satisfaction Better with More Physicians?

Physicians Per Capita

Lowest Quintile

Highest Quintile

Ratio highest to

lowest

Total physicians per capita by Hospital Referral Regions (2005)

169.4 271.8 1.60

CMS Compare Composite Scores (2005)

Acute myocardial infarction 91.0 93.1 1.02

Congestive heart failure 84.1 88.6 1.05

Pneumonia 79.5 79.2 1.00

Medicare access and satisfaction (2005)

Ever had a problem and didn't see a doctor? (% No) 91.7 93.2 1.02

Do you have a particular place for medical care? (% Yes) 95.0 95.5 1.01

Satisfied with ease of getting to the doctor? (% Yes) 94.9 94.7 1.00

Satisfied with doctor's concern for overall health? (% Yes) 95.5 95.7 1.00

Satisfied with quality of medical care? (% Yes) 96.7 97.0 1.00

Goodman DC, Fisher ES. New England J Med, 2008.

Why is there such a weak association between workforce supply and outcomes?

Examples of Medical Decision Uncertainty that Lead to Different Labor Demand

• 84 y.o with mild CHF, diabetes, and new onset back pain that is poorly controlled with oral opiates.

– Admit to the hospital?

• 69 y.o with COPD (Nighttime O2) and two recent episodes of bronchitis with ER visits.

– Consultation with a pulmonologist? Revisit every 2, 4, 6 months?

• 65 y.o. with new lumbar disc herniation.

Inpatient back surgery per 1,000 Medicare enrollees (2005)

1.0

3.0

5.0

7.0

9.0

11.0

Bac

k su

rger

y pe

r 1,

000

enro

llees Minneapolis 5.0

Binghamton 4.4Rochester 3.8Buffalo 3.3Syracuse 3.2White Plains 2.7Elmira 2.6Albany 2.6Miami 2.4Manhattan 1.9East Long Island 1.9Bronx 1.8

So what?

Yes, physician are located idiosyncratically.

And maybe outcomes aren’t sensitive to physician supply.

Still, would an increase in physician training rates cause any harm?

High Physician Supply/Cost Regions:

• Less likely to provide primary care.

• Lower perceived access by patients.

• No better patient satisfaction.

• Worse technical quality.

• No better, and sometimes worse outcomes

• Physicians perceive care to be less available, less able to provide quality care.

Sirovich B, et al. Ann Int Med 2006. Sirovich B, et al. Arch Int Med 2005.Wennberg J. Ed. Dartmouth Atlas of Health Care. Various editions. 1996 - 2006.Fisher E, et al. Ann Int Med 2003; Fisher E, at al. Health Affairs 2004; Fisher E, et al. Health Affairs 2005.Goodman D, et al. Health Affairs 2006.

Number of Atlas Regions byPhysicians per 100,000 population

Where do more physicians go?

Source: Goodman. Health Affairs, 2004.

For every physician that settled in a low supply region, 4

physicians settled in a high supply region.

These are the regions associated with lower quality

and higher costs.1999

Number of Atlas Regions byPhysicians per 100,000 population

Num

ber

of R

egio

ns

1979

What about the costs of expanding medical schools and removing the Medicare GME

funding cap?

No published estimates...

probably an additional $5-10 billion per annum in training costs.

(NIH ~ $28 billion; CDC ~ $8 billion)

Medicare Costs and Non-Interest Income by Source as a Percent of GDP

2019 Part A trust fund goes brokePart B and D premiums soar

% GDP

Where would you invest $5-10 billion per annum of public money in the health care system?

• Implementation of the U.S. Preventive Services Task Force recommendations.

• Greater implementation of Cochrane Collaboration recommendations.

• Increasing NIH funding.• Rewarding health care systems for improved outcomes.• Expanding insurance coverage to children (S-CHIP).

• Increasing physician training rates?

Since when did we start trusting market forces to deliver good health care?

Does “Demand” Equal Consumer “Wants?”

• Consumers can judge quality.(e.g. Consumers Report)

• Lot’s of sellers.

• Consumers are the sole decider.

• Consumers pay the full price (no subsidization).

• Demand = what consumers want.

• Markets work well.

• Evidence-base is imperfect.

• Patients do not have full information.

• There are fewer “sellers.”

• Patients look to physicians to make recommendations.

• Insurers pay the price at the time of the “purchase” decision.

• Demand = utilization

• Market failure.

Autos Medical Care

Market forces are like gravity...Each help you get where you want to go,

but you wouldn’t want to throw away the steering wheel and brakes.

Restoring Accountability to Health Workforce Planning

• Decisions about numbers and specialty mix of physician training are left to each training hospital.

• Council on Graduate Medical Education has a narrow policy brief (i.e. physician training only, no dedicated staff) and consists entirely of physicians, primarily from teaching hospitals.

• Public dollars pays for most medical training.

• Permanent Health Workforce Commission

– Public interests and workforce goals should be clearly stated.

– Broad membership (nurses, public health expts., patients, docs)

– Should advice on health workforce, not just physician workforce.

– Dedicated staff support

– Increasingly regulatory responsibility to insulate the deliberations from training program and provider self-interests.

Source: Goodman DC. JAMA, September 10, 2008.

Beyond the workforce “crisis”

• Physician supply varies 2 - 3 fold, generally without differences in outcomes (health status, quality, access, satisfaction).

• Health care systems are adaptable to varying levels of physician supply.

• Expansion of physician training will be costly, and could exacerbate many of our current health care ills.

• Workforce planning in the U.S. lacks coordination and depends on the individual decisions of hundreds of teaching hospitals.

• Physician training resources should be redirected towards health systems delivering efficient care, and preference-based care.

• A robust primary care workforce is necessary but not sufficient for improved systems of care.

• The medical home can only succeed with payment reform and redesign of health care systems to integrated delivery systems.