Slide 1 Bending the cost curve Addressing the problem of “supply-sensitive” care Elliott S. Fisher, MD, MPH Professor of Medicine Center for the Evaluative

Slide 1

Bending the cost curveAddressing the problem of “supply-sensitive” care

Elliott S. Fisher, MD, MPHProfessor of Medicine

Center for the Evaluative Clinical SciencesDartmouth Medical School

Senior AssociateVA Outcomes Group

White River Junction, Vermont

CECSCenter for the EvaluativeClinical Sciences

Slide 2

Variations in practice and spendingacross U.S. Regions

Slide 3

The paradox of plenty: cross sectional evidenceWhat do higher spending regions -- and systems -- get?

Technical quality worseNo more major elective surgeryMore hospital stays, visits, specialist use, tests, procedures

Content / Quality of Care1,2

Slightly higher mortalityNo better function

Health Outcomes1,2

Worse communication among physiciansGreater difficulty ensuring continuity of careGreater difficulty providing high quality care

Physician-reported quality5

Patient-reported quality1,3 Lower satisfaction with hospital careWorse access to primary care

(1) Ann Intern Med: 2003; 138: 273-298 (2) Health Affairs web exclusives, October 7, 2004(3) Health Affairs, web exclusives, Nov 16, 2005(4) Health Affairs web exclusives, Feb 7, 2006(5) Ann Intern Med: 2006; 144: 641-649

More hospital beds per capita (32%)More medical specialists (65%) and internists (75%)

Resource levels1

Supply sensitive care

Slide 4

Trends in spending and qualityWhat do higher spending regions -- and systems -- get?

Regions with greatest spending growth had smallest gains in heart attack survival

Skinner, Health Affairs, February 2006

Slide 5

Differences in spendingWhat are the underlying causes?

Explains less than 10% of state differences in spendingLittle impact on growth in utilization across states

Malpractice environment3,4

Capacity strongly correlated, but explains less than 50%Payment system ensures all stay busy

Capacity / payment system5

No difference in decisions with strong evidenceMore likely to intervene in “gray” areas(when to see patient, when to refer, when to admit)

Clinical judgment6,7

(1) Pritchard et al. J Am Geriatric Society; 46:1242-1250, 199(2) Anthony et al, under review(3) Kessler et al. Quarterly Journal of Medicine 1996;111(2):353-90(4) Baicker, Chandra, NBER Working Paper W10709(5) Fisher et al. Ann Intern Med: 2003; 138: 273-298(6) Sirovich et al. Archives of Internal Medicine. 165(19):2252-6.(7) Sirovich et al, J Gen Intern Med. 2006;21(Suppl4):164.

Slight preference for specialist care in high spendingNo difference for tests (if MD says not needed)No difference in preferences for aggressive EOL care

Patient preferences?1,2

Slide 6

Likely diagnosisLocal capacity and culture drive practice and spending

Physician - PatientEncounter

Clinical EvidenceProfessionalism

Clinical evidence (e.g. RCTs, guidelines) and principles of professionalism are a critically important -- but limited -- influence on clinical decision-making.

Consequence: reasonable individual clinical and local decisions lead, in aggregate, to higher utilization rates,greater costs -- and inadvertently -- worse outcomes

LocalOrganizational Context(e.g. capacity - culture)

Policy Environment(e.g. payment system)

Physicians practice within a local organizationalcontext and policy environment that profoundlyinfluences their decision-making. Payment systemensures that existing (and new capacity) is fullyutilized -- and generously rewards growth.

Slide 7

Some examplesA payment system that rewards growth and higher intensity care…

Management of coronary artery disease -- the case of Elyria, Ohio

PercutaneousCoronary Interventions

Age-sex-race adjustedrate per 1000 enrollees in

2003

Slide 8



PercutaneousCoronary Interventions

Age-sex-race adjustedrate per 1000 enrollees in

2003

Slide 9



New York Times, August 18, 2006

Slide 10



Use of erythropoetin (under current payment system)

New York Times, May 9, 2007

Slide 11



Use of erythropoetin (under current payment system)

Differences in use of physician workforce across academic medical centers

Dartmouth Atlas of Health Care 2006

Mayo Duke UCSF UCLA Cedars

Hospital days (L6M)* 12.9 14.0 13.2 19.2 23.1

Physician visits (L6M)* 23.8 23.3 30.4 52.1 71.3

Medical specialist FTE (L2Y)** 8.4 8.8 9.0 22.9 29.9

Primary care FTE inputs (L2Y)** 7.0 6.4 10.8 9.3 12.8

Total Physician FTE (L2Y)** 20.3 21.1 24.5 40.6 52.2

* Measures are per person / per decedent** Measures are per 1000 decedents

Slide 12

Some thoughts on moving forwardWe need to consider underlying causes of rising costs, poor quality

Failure to recognize key role of local system (capacity, clinical culture) asdriver

Assumption that more is betterEquating less care with rationing

Payment system that rewards morecare, increased capacity, high margintreatments, entrepreneurial behavior

Foster development of local organizations(delivery systems) accountable for care (withincentives to limit future growth)

Balanced information on risks / benefitsComprehensive performance measures

Reform of payment system (long term)Shared savings as interim approach

Underlying cause General Approach

Slide 13

Organizational accountability and incentives to slow growth Per-beneficiary spending in EHMS (n = 4772) sorted into quintiles

by magnitude of per-beneficiary growth (1999-2003)

$4000

$3000

$2000

1999 2003

Averagespending*

on MD servicesper beneficiary

at EHMS

* Using standardized payments, using 2003 RVU** Percent increase calculated relative to average 1999 per-beneficiary spending

Absoluteincrease

per benef.

$936

$198

$431

$551

$675

Percentincrease 99-03**

46%

10%

21%

27%

33%

AverageAnnual Rate

9.9%

2.4%

4.8%

6.1%

7.3%

Slide 14

Organizational accountability and incentives to slow growth Per-beneficiary spending in EHMS by BETOS category (highest

and lowest quintiles of per-beneficiary growth (1999-2003)

0

500

1000

1500

2000

2500

3000

3500

19992003

Lowest Growth Quintile

Med

icar

e sp

end

ing

per

en

roll

ee

OtherMajor ProceduresMinor ProceduresTestsImagingE and M

6%

29%

18%

2%

0%

27%

10%

19992003

Highest Growth Quintile

27%

80%

65%

38%

19%

116%

46%

Percent increase in per-beneficiary spending

Each Quintile includes approximately 20% of the Medicare population

Differences in growth likely due to: • active recruitment of physicians • physician location decisions • expansion of facilities (imaging)

Control of spending will require alteringincentives for growth

Slide 15

Payment reform Challenges and opportunities

Barriers to comprehensive payment reform are substantial

Public opposition to capitation; provider concern about bearing risk

Development of other prospective payment approaches years away

Might “shared savings” approaches help in the interim?

Key notion: establish target growth rate; reward physician groups that achieve per-beneficiary spending growth below the target with portion of savings

Theory being tested in the Physician Group Practice demonstration

Has important advantages:• Preserves fee-for-service payment (a plus for patients and MDs)• Provides incentive to avoid increases in capacity (and to reduce capacity where feasible); and to improve care in domains previously ignored:

care coordination, end-of-life care• Can be done with existing claims data

Documents

Slide 1 Bending the cost curve Addressing the problem of “supply-sensitive” care Elliott S. Fisher, MD, MPH Professor of Medicine Center for the Evaluative