financing graduate medical education in a changing health care environment · 2018. 12. 15. · 3....

Financing GraduateMedical Education in aChanging Health Care

Environment

COUNCIL ON GRADUATE MEDICAL EDUCATION

Fifteenth Report

D E C E M B E R 2 0 0 0

COUNCIL ON GRADUATE MEDICAL EDUCATION

Fifteenth Report

Financing GraduateMedical Education in aChanging Health CareEnvironment

December 2000

U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICESHealth Resources and Services Administration

iiFIFTEENTH REPORT OF COGME

The views expressed in this document are solely those of theCouncil on Graduate Medical Education and do not

necessarily represent the views of theHealth Resources and Services Administration

nor the U.S. Government.

iiiFIFTEENTH REPORT OF COGME

List of Tables and Figures ..................................................................................................... v

The Council on Graduate Medical Education .................................................................. vii

Acknowledgements ............................................................................................................... xi

Preface ................................................................................................................................. xiii

Executive Summary ............................................................................................................... 1Overview ......................................................................................................................................................1The Changing Health Care Environment ....................................................................................................2

Current Financing of Graduate Medical Education ....................................................................................3

Residency Training in Community-Based Settings ....................................................................................5Alternative Models for GME Financing Reform ........................................................................................6

Recommendations for GME Financing Reform..........................................................................................8

Introduction .......................................................................................................................... 17Background ............................................................................................................................................... 17

Purpose ..................................................................................................................................................... 17

Methodology ............................................................................................................................................. 17Organization of Report .............................................................................................................................. 18

The Changing Health Care Environment .......................................................................... 19

Current Financing of Graduate Medical Education ........................................................ 23Overview ................................................................................................................................................... 23

Medicare Payments ................................................................................................................................... 23

Medicaid Support for GME ...................................................................................................................... 34Health Resources and Services Administration Programs ....................................................................... 36

Other Direct Funding Sources .................................................................................................................. 38

Residency Training in Community Settings ...................................................................... 39Current Training in Ambulatory/Community Settings ............................................................................. 39

Accreditation Requirements ..................................................................................................................... 40

Potential Financial Barriers to Training in Ambulatory Settings ............................................................. 41Site Visit Findings ..................................................................................................................................... 42

GME Reform Proposals ...................................................................................................... 49All-Payer Proposals ................................................................................................................................... 49Commission on the Future of Medicare ................................................................................................... 51

MEDPAC Proposal ..................................................................................................................................... 51

Alternative Models for GME Funding ............................................................................... 53Policy Goals for GME Funding ................................................................................................................ 53

Policy Considerations for Community-Based Training ........................................................................... 53

Health Care Provider Model ..................................................................................................................... 54Education Model ....................................................................................................................................... 55

Planning Model ......................................................................................................................................... 56

Performance Model ................................................................................................................................... 57

Table of Contents

ivFIFTEENTH REPORT OF COGME

Recommendations for GME Financing Reform ............................................................... 58

Bibliography and Footnotes ................................................................................................ 67

APPENDICES

A. Explicit Medicaid GME Payments, FY1998 ............................................................... 73

B. Program Requirements for Residency Educationin Ambulatory/Community-Based Settings ................................................................. 75

C. Site Visits and Interviews .............................................................................................. 77

vFIFTEENTH REPORT OF COGME

LIST OF TABLES AND FIGURES

Table 1: Medicare GME Payments to Hospitals .................................................................................. 23

Table 2: Hospital Per Resident Teaching Physician Costs, FY1997 ................................................... 26

Table 3: Distribution of Hospitals by Teaching Intensity and Low-Income Patients ......................... 29

Table 4: Ambulatory/Community-Based Rotations for Program Year 1 Residents ............................ 39

Figure 1: Hospital Total Margin by Teaching Status, 1994-1998 ......................................................... 20

Figure 2: Hospital Medicare Inpatient Margin (ExcludingGraduate Medical Education) by Hospital Group, 1994-1998 ............................................. 20

Figure 3: Comparison of FY 1997 Per Resident Amounts and GME Costs ......................................... 24

Figure 4: Components of FY 1997 GME Cost Per Resident ................................................................ 25

Figure 5: Distribution of FY1997 Medicare Per Resident Payments ................................................... 27

Figure 6: GME Costs and Per Resident Amounts by Proportion of Primary CareResidents, 1997 ...................................................................................................................... 28

Figure 7: Proportion of Clinical Experience in Ambulatory Settingsfor Categorical Internal Medicine Residency Programs ....................................................... 40

Figure 8: Internal Medicine Ambulatory Care Sites Used for Continuity Experiences ....................... 40

viFIFTEENTH REPORT OF COGME

viiFIFTEENTH REPORT OF COGME

The Council on Graduate Medical Education(COGME) was authorized by Congress in1986 to provide an ongoing assessment of

physician workforce trends, training issues andfinancing policies, and to recommend appropriateFederal and private sector efforts to address identi-fied needs. The legislation calls for COGME toadvise and make recommendations to the Secre-tary of the Department of Health and Human Serv-ices (DHHS), the Senate Committee on Health, Edu-cation, Labor and Pensions, and the House ofRepresentatives Committee on Commerce. The HealthProfessions Education Partnerships Act of 1998 reau-thorized the Council through September 30, 2002.

The legislation specifies 17 members for theCouncil. Appointed individuals are to include rep-resentatives of practicing primary care physicians,national and specialty physician organizations, in-ternational medical graduates, medical student andhouse staff associations, schools of medicine andosteopathy, public and private teaching hospitals,health insurers, business, and labor. Federal repre-sentation includes the Assistant Secretary forHealth, DHHS; the Administrator of the Health CareFinancing Administration, DHHS; and the ChiefMedical Director of the Veterans Administration.

Charge to the CouncilThe charge to COGME is broader than the name

would imply. Title VII of the Public Health ServiceAct, as amended, requires COGME to provide ad-vice and recommendations to the Secretary andCongress on the following issues:

1. The supply and distribution of physicians inthe United States.

2. Current and future shortages or excesses ofphysicians in medical and surgical specialtiesand subspecialties.

3. Issues relating to international medical schoolgraduates.

4. Appropriate Federal policies with respect tothe matters specified in items 1-3, includingpolicies concerning changes in the financingof undergraduate and graduate medical edu-cation (GME) programs and changes in thetypes of medical education training in GMEprograms.

5. Appropriate efforts to be carried out by hospi-tals, schools of medicine, schools of osteopa-

The Council on Graduate Medical Education

thy, and accrediting bodies with respect to thematters specified in items 1-3, including ef-forts for changes in undergraduate and GMEprograms.

6. Deficiencies and needs for improvements indata bases concerning the supply and distri-bution of, and postgraduate training programsfor, physicians in the United States and stepsthat should be taken to eliminate those defi-ciencies.

In addition, the Council is to encourage enti-ties providing graduate medical education to con-duct activities to voluntarily achieve the recommen-dations of the Council specified in item 5.

COGME ReportsSince its establishment, COGME has submit-

ted the following reports to the DHHS Secretaryand Congress:

• First Report of the Council (1988)

• Second Report: The Financial Status of Teach-ing Hospitals and the Underrepresentation ofMinorities in Medicine (1990)

• Scholar in Residence Report: Reform in Medi-cal Education and Medical Education in theAmbulatory Setting (1991)

• Third Report: Improving Access to HealthCare Through Physician Workforce Reform:Directions for the 21st Century (1992)

• Fourth Report: Recommendations to ImproveAccess to Health Care Through PhysicianWorkforce Reform (1994)

• Fifth Report: Women and Medicine (1995)

• Sixth Report: Managed Health Care: Implica-tions for the Physician Workforce and Medi-cal Education (1995)

• Seventh Report: Physician Workforce Fund-ing Recommendations for Department ofHealth and Human Services’ Programs (1995)

• Eighth Report: Patient Care Physician Supplyand Requirements: Testing COGME Recom-mendations (1996)

• Ninth Report: Graduate Medical EducationConsortia: Changing the Governance ofGraduate Medical Education to Achieve Phy-sician Workforce Objectives (1997)

viiiFIFTEENTH REPORT OF COGME

• Tenth Report: Physician Distribution andHealth Care Challenges in Rural andInner-City Areas (1998)

• Eleventh Report: International Medical Gradu-ates, The Physician Workforce and GME Pay-ment Reform (1998)

• Twelfth Report: Minorities in Medicine (1998)

• Thirteenth Report: Physician Education for aChanging Health Care Environment (1999)

• Fourteenth Report: COGME PhysicianWorkforce Policies: Recent Developments andRemaining Challenges in Meeting NationalGoals (1999)

COGME Resource Papers• Process by which International Medical

Graduates are Licensed to Practice in theUnited States (1995)

• Preparing Learners for Practice in a ManagedCare Environment (1997)

• International Medical Graduates: ImmigrationLaw and Policy and the U.S. PhysicianWorkforce (1998)

• The Effects of the Balanced Budget Act of1997 on Graduate Medical Education (2000)

• Update on the Physician Workforce (2000)

• Evaluation of Specialty Physician WorkforceMethodologies (2000)

Other COGME Publications• Council on Graduate Medical Education: What

is it? What has it done? Where is it going?(2000)

ixFIFTEENTH REPORT OF COGME

MembersChairDavid N. Sundwall, M.D.PresidentAmerican Clinical Laboratory AssociationWashington, D.C.

Macaran A. Baird, M.D.Medical DirectorMayo Management Services, Inc.Rochester, Minnesota

Regina M. Benjamin, M.D., M.B.A.Family PracticeSpanish Fort, Alabama

F. Marian Bishop, Ph.D., M.S.P.H.Professor and Chairman EmeritaDepartment of Family and Preventive MedicineUniversity of Utah School of MedicineSalt Lake City, Utah

Jo Ivey Boufford, M.D.Dean, Robert F. Wagner Graduate School of

Public Service, New York UniversityNew York, New York

William ChingMedical StudentNew York University School of MedicineNew York, New York

Ezra C. Davidson, Jr., M.D.Associate Dean, Primary Care and ProfessorDepartment of Obstetrics and GynecologyKing/Drew Medical CenterLos Angeles, California

Vice ChairCarl J. Getto, M.D.Dean and Provost, SouthernIllinois University School of MedicineSpringfield, Illinois

Kylanne GreenVice PresidentInova Health SystemsFairfax, Virginia

Ann KempskiAssistant Director Health PolicyAmerican Federation of State, County and

Municipal EmployeesWashington, D.C.

Members, Council on Graduate Medical Education

Lucy Montalvo-Hicks, M.D, M.P.H.State of California Department of Social ServicesSan Diego, California

Susan Schooley, M.D.Chair, Department of Family PracticeHenry Ford Health SystemDetroit, Michigan

Donald C. Thomas III, M.D.ConsultantNorthridge, California

Douglas L. Wood, D.O., Ph.D.President, American Association of Colleges of

Osteopathic MedicineChevy Chase, Maryland

Designee of the Assistant Secretary for HealthNicole Lurie, M.D., M.S.P.H.Principal Deputy Assistant Secretary for HealthU.S. Department of Health and Human ServicesWashington, D.C.

Designee of the Health Care FinancingAdministration

Tzvi M. HefterDirector, Division of Acute CareCenter for Health Plans and ProvidersHealth Care Financing AdministrationU.S. Department of Health and Human ServicesBaltimore, Maryland

Designee of the Department of Veterans AffairsGloria Holland, Ph.D.U.S. Department of Veterans AffairsWashington, D.C.

Statutory MembersDavid Satcher, M.D.Assistant Secretary for Health and Surgeon

GeneralU.S. Department of Health and Human ServicesWashington, D.C.

Nancy-Ann Min DeParleAdministrator, Health Care Financing

AdministrationU.S. Department of Health and Human ServicesBaltimore, Maryland

xFIFTEENTH REPORT OF COGME

Thomas L. Garthwaite, M.D.Undersecretary for HealthVeterans Health AdministrationU.S. Department of Veterans AffairsWashington, D.C.

COGME Staff

Carol Bazell, M.D., M.P.H.DirectorDivision of Medicine and Dentistry

Stanford M. Bastacky, D.M.D., M.H.S.A.Acting Executive SecretaryChief, Policy and Special Projects BranchDivision of Medicine and Dentistry

Jerilyn K. Glass, M.D., M.P.H.Acting Deputy Executive Secretary

P. Hannah Davis, M.S.Staff Liaison, Graduate Medical Education

Financing and Policy Workgroup

John Rodak, Jr., M.S. (Hyg.), M.S.(H.S.A.)Staff Liaison, Physician Workforce Workgroup

C. Howard Davis, Ph.D.Staff Liaison

Richard Diamond, M.D., M.P.A.Staff Liaison

Jerald M. KatzoffStaff Liaison

Helen K. Lotsikas, M.A.Staff Liaison

Anne PattersonSecretary

Eva M. StoneProgram Analyst and Committee Management

Specialist

Velma ProctorSecretary

COGME Workgroup on AmbulatoryPrograms and Financing

COGME MEMBERS

Carl J. Getto, M.D., Chair

Macaran A. Baird, M.D.

Kylanne Green

Tzvi M. Hefter

Ann Kempski

Nicole Lurie, M.D., M.S.P.H.

Lucy Montalvo-Hicks, M.D., M.P.H.

Donald C. Thomas III, M.D.

Douglas L. Wood, D.O., Ph.D.

COGME STAFF

P. Hannah Davis, M.S., Staff Lead

Jerald M. Katzoff

xiFIFTEENTH REPORT OF COGME

The Council wishes to acknowledge the ef-forts of the many people who contributed tothe preparation of this report. Barbara Wynn,

of RAND Health Corporation assiduously studiedand analyzed the complex issues of GME financ-ing and wrote the basis of the report under con-tract, with assistance from Kimberly Gavin andJennifer Kawata. Representatives at sites studiedfor alternative GME financing programs generouslycontributed their time and openly shared helpfulinformation. Members of an expert panel contrib-uted many helpful insights during an all-day meet-ing in Washington D.C. and reviewed an early draftof the report. The expert panel members are JohnCoombs, M.D., Associate Dean, University ofWashington School of Medicine; Ronald D. Franks,M.D., Dean of Medicine and Vice President forHealth Affairs, East Tennessee University Collegeof Medicine; Larry Greene, M.D., Director, Centerfor Policy Studies in Family Practice and PrimaryCare, American Academy of Family Physicians;Tim Henderson, M.S.P.H., Director, Primary CareResource Center, National Conference of State Leg-

islatures; Denise Holmes, Director, Plan Adminis-tration Bureau, Michigan Department of Commu-nity Health; Robert J. Mossad, M.D., ProfessorEmeritus, Montefiore; Gordon T. Moore, M.D.,Associate Professor, Harvard Medical School/Harvard Pilgrim Health Care; and Michael Opipari,D.O., Executive Vice President and Chief MedicalOfficer, Horizon Health System.

The Council wishes to acknowledge the guid-ance and leadership of P. Hannah Davis, M.S., whowas Project Officer and chief staff liaison for thisstudy, and Jerald M. Katzoff who assisted her inthis endeavor. The Council expresses its gratitudeto F. Lawrence Clare, M.D., senior advisor, Divi-sion of Medicine and Dentistry for his expertise inGME financing. The Council also appreciates thecommitment from Carol Bazell, M.D., M.P.H, Di-rector, Division of Medicine and Dentistry, andStanford M. Bastacky, D.M.D., M.H.S.A., COGMEActing Executive Secretary in bringing this report for-ward and to fruition.

Acknowledgements

xiiFIFTEENTH REPORT OF COGME

xiiiFIFTEENTH REPORT OF COGME

The purpose of this report of the Council onGraduate Medical Education (COGME) isto examine current funding mechanisms for

graduate medical education and to evaluate alter-native Federal financing policies within the con-text of the Council’s recommendation for a stablefinancing mechanism that would provide adequatesupport for ambulatory training. COGME’s 13thReport, Physician Education for a Changing HealthCare Environment, explored the educationalchanges needed to train physicians who are well-equipped to provide high-quality, effective and ef-ficient care and emphasized the need for an appro-priate balance of traditional and community-basedtraining experiences. The 14th Report, COGMEPhysician Workforce Policies: Recent Developmentsand Remaining Challenges in Meeting NationalGoals, called for a stable all-payer financing mecha-nism for graduate medical education that wouldprovide adequate support for ambulatory training.The current report builds on the earlier COGMEreports and does not revisit the earlier analyses lead-ing to the recommendations for expanded commu-nity-based training opportunities and all-payer fi-nancing. Rather, the report analyzes alternativefinancing policies within the context of those rec-ommendations. Readers are referred to the earlier

Preface

reports for findings and recommendations dealingwith medical education curriculum and quality andphysician workforce issues.

This report focuses on graduate medical edu-cation financing issues. There is a close connec-tion between GME and issues related to under-graduate medical education and nursing and alliedhealth education. While the latter issues are ex-tremely important, they are not within the scope ofthis report.

COGME considers this report to be a valuablecontribution to the discourse on GME financing.While the report’s recommendations may be con-troversial, they provide the foundation for furtherdiscussion of these complex issues. Developmentof these concepts is an iterative process. COGMEbelieves this report is beneficial in stating past po-sitions and new ideas all together in one place, witha thorough explanation of their basis. This is alsothe only report that documents the results of se-lected site visits of successful residency programswith community-based training. COGME intendsto use this report as a basis for further discussion ofGME financing in a national forum of stakehold-ers that it plans to hold in 2001.

xivFIFTEENTH REPORT OF COGME

1FIFTEENTH REPORT OF COGME

Executive Summary

This report, Financing Graduate Medical Edu-cation in a Changing Health Care Environ-ment, reviews current funding mechanisms

for graduate medical education (GME) and recom-mends actions that should be taken to respond tothe changing health care environment. Care is in-creasingly provided within the context of managedsystems of care that require clinical learning expe-riences across the continuum of care. The currentsystem of financing GME has inherent limitationsand disincentives for expanding training in com-munity-based sites. Stable all-payer financing isneeded that will provide adequate support for train-ing in community settings.

OVERVIEWAs used in this report, GME is clinical training

in an approved residency program following gradu-ation from schools of medicine, osteopathy, den-tistry and podiatry. The training is required for cer-tification in a specialty and is approved by anon-governmental accrediting organization for thespecialty. The residency program varies in lengthdepending on the specialty. Most residency pro-grams are sponsored by a hospital, medical school,or educational consortium. There are about 100,000residents in 8,000 different residency programs. Theresidents, who are serving a form of apprenticeship,provide patient care under the supervision of ateaching physician. Teaching hospitals serve as theprimary training sites for most residency programs.Training occurs in both the inpatient setting and inthe ambulatory-clinics of the teaching hospital. Inaddition, community hospitals and other commu-nity-based sites provide training opportunities.

Residency training should be relevant to cur-rent daily physician practice and address the careof the individual patient in their cultural and socialcontext. With the growth of managed systems ofcare and the movement of services to outpatientsettings, residency programs have expanded train-ing opportunities in community settings. As usedin this report, the term “community settings” de-scribes settings that are representative of the envi-ronment in which most residents will eventuallypractice. Under this definition, the processes of careand educational outcomes are the determining fac-tors in identifying a community setting rather thanits location per se. A comprehensive range of ex-periences is necessary in order to provide opportu-

nities to follow the patient across each componentof an integrated delivery system. Community-basedsettings such as health centers and clinics, physi-cian offices, schools and workplaces, nursinghomes, hospices and home care, community hos-pitals, and managed care organizations can offeressential experiences to complement those at aca-demic health centers. For some specialties, com-munity training will occur in hospital-based ambu-latory sites since this is where the specialtycommonly practices.

The costs of GME are difficult to determinebecause teaching occurs in tandem with patient careand research. There are direct GME costs, whichinclude the resident’s stipend, payments to teach-ing physicians, program administration costs andother costs directly attributable to educational ac-tivities. In addition, there are indirect GME costs.These are higher patient care costs associated withteaching hospitals, such as treating sicker patients,using more diagnostic tests, and longer patient vis-its or hospital stays. Direct GME costs for a singleresidency program are typically incurred by mul-tiple entities: the program sponsor, the faculty prac-tice plan affiliated with the sponsoring institution,and the hospitals and ambulatory sites that providetraining. Each site’s direct costs for GME dependon its negotiations and arrangements with otherentities involved in the training program over is-sues such as which party will assume the costs ofthe resident’s salary and teaching physician com-pensation.

Patient care revenues provide most of the sup-port for GME. However, Medicare and, in someStates, Medicaid, make explicit payments to teach-ing hospitals for their GME costs. These paymentsrecognize that equipping future physicians with thecompetencies to provide high quality care is in thepublic interest. Private payers have also tradition-ally paid higher amounts to teaching hospitals tosupport the costs of training residents, and to someextent, the charity care provided by teaching hos-pitals. In 1998, uncompensated care representedrevenue losses of 7.8 percent and 5.4 percent of thetotal costs of academic health centers and othermajor teaching hospitals, respectively (MedPAC,2000b).

Faculty practice plan revenues are anothersource of support for clinical faculty time spent inacademic activities. Faculty practice plan revenues


may also provide direct support to medical schoolor department funds that are used to support gradu-ate as well as undergraduate medical education.State-support for GME typically occurs throughappropriations to State-operated medical schoolsor residency training grants (about $185 million).In addition, Federal appropriations under the Pub-lic Health Service Act support primary care resi-dency programs and other health professional edu-cation ($300 million) and children’s teachinghospitals ($40 million). Other sources of fundinginclude research grants, endowments, and founda-tion grants. The Department of Veterans Affairs(DVA) and the Department of Defense (DoD) sup-port about 15 percent of residency positions. Thus,the flow of funds among the participants in GMEactivities is complex and frequently involves cross-subsidies between medical schools, teaching hos-pitals, and other training sites.

Competitive pressures associated with the moveto managed care (capitated financing arrangements)have eroded the private payer subsidies for teach-ing and charity care. Medicare and Medicaid pay-ment reductions in the Balanced Budget Act of 1997(BBA) also added to the financial pressures onteaching institutions. The Council on GraduateMedical Education’s (COGME) 13th Report, Phy-sician Education for a Changing Health Care En-vironment (1999a), concluded that the current sys-tem of funding GME through teaching hospitals hasinherent limitations and disincentives for develop-ing ambulatory clinical training experiences andcommunity-based educational programs. The finan-cial and service needs of teaching hospitals com-pete with the educational need to expand trainingopportunities in community settings. The uncertain-ties and financial pressures inherent in the chang-ing health care environment suggest changes areneeded in the way GME is financed. In its 14thReport, COGME Physician Workforce Policies:Recent Developments and Remaining Challengesin Meeting National Goals (1999b), COGME calledfor a stable, all-payer financing mechanism forGME that would provide adequate funding for train-ing in ambulatory settings.

The current report builds on COGME’s earlierreports by examining current funding mechanismsfor GME and assessing their implications for de-veloping community-based educational programs.It evaluates alternative Federal financing policieswithin the context of the Council’s recommenda-tion for a stable financing mechanism that wouldprovide adequate support for ambulatory training.(Readers are referred to the earlier reports for find-ings and recommendations dealing with medicaleducation curriculum and physician workforce is-sues.)

This report recommends the creation of a GMEfund that would supplement current Federal fund-ing for GME with funds from third-party payers.Total GME funding would be sufficient to supporthigh-quality and efficient training of an appropri-ately sized and balanced physician workforce. Therecommended fund allocation policies are designedto encourage an appropriate balance between tra-ditional and community training in all hospital-sponsored specialty programs. Payments for directGME costs would be made to program sponsorswho would be held accountable for educational andworkforce objectives. Separate payments would bemade to clinical training sites to support their higherpatient care costs attributable to teaching activities.The report recommends a separate GME accountfor funding special projects and programs directedat building high-quality community-based trainingcapacity and achieving specific workforce priori-ties. In addition to the recommendations on GMEfunding , the report recommends increased supportfor “safety net “ hospitals and community provid-ers that serve a disproportionate share of low-in-come patients and have high uncompensated carecosts. It is important that refinements in the GMEallocation methodologies not adversely affect hos-pitals that provide significant charity care. Any re-ductions in GME payments to these hospitalsshould be offset by higher support for uncompen-sated care until specific funding for such servicesis provided.

THE CHANGING HEALTH CAREENVIRONMENT

Recent trends in the health care delivery sys-tem have major implications for how GME pro-grams are operated and how they are financed. Themost significant financing changes are associatedwith the growth of managed systems of care andincreased competition within health care markets.Teaching hospitals tend to have higher costs thatput them at a competitive disadvantage with com-munity hospitals in competing for managed carecontracts. In the past, private payers have subsidizedthe educational and other missions of teachinghospitals through higher payments. Competitivepressures have eroded these subsidies for publicmissions. The competitive pressures are evidencedin a decline in the private payer payment-to-costratio for teaching hospitals from 1.25 in 1989(ProPAC, 1992) to 1.15 and 1.05, respectively, foracademic health centers and other major teachinghospitals by 1998 (MedPAC, 2000b).

Recent changes in Medicare and Medicaidfunding for GME have added to the financial pres-


sures on teaching institutions. Medicaid managedcare growth has reduced Medicaid revenues andpayments for serving a disproportionate share oflow income patients that public and other safetynet teaching hospitals rely on to support their char-ity care missions. For major teaching hospitals thatserve low-income patients, the reduction in Med-icaid revenues has been accompanied by an increasein uncompensated care losses (IOM, 2000). In ad-dition, the Balanced Budget Act of 1997 (BBA)reduced Medicaid payments for hospitals servinga disproportionate share of low-income patients$10.4 billion over five years (CBO, 1997). The BBAprovisions affecting the Medicare indirect teach-ing adjustment and the disproportionate share ad-justment were estimated to reduce payments toteaching hospitals $5.6 billion and hospitals serv-ing low income patients by $0.6 billion over fiveyears (CBO, 1997). Other Medicare provisions,such as reductions in the annual updates for infla-tion and the implementation of prospective paymentsystems for hospital outpatient services and post-acute care providers will also affect teaching hos-pital revenues. The Balanced Budget RefinementAct of 1999 (BBRA) restored an estimated $700million of the Medicare cuts and added $100 mil-lion in direct GME payments (CBO, 1999), a rela-tively modest amount compared to the total BBAcuts.

The issue of whether Medicare’s support shouldcontinue at its current levels was debated duringconsideration of the BBA. The debate centered onconcerns over the solvency of the Medicare Part Atrust fund and the impact of continued Medicarespending cuts in the face of a competitive healthcare market. The uncertainties of continued reli-ance on Medicare, Medicaid and private pay rev-enues to fund GME have reinforced the conclusionsheld by COGME and others that major changes areneeded in the way GME is financed. The Council’s14th Report (1999b) reiterated COGME’s longstanding recommendation for an all-payer financ-ing system that would spread the costs of prepar-ing a well-qualified physician workforce equitablyacross all payers. COGME believes there contin-ues to be value in exploring alternative financingpolicies that would enhance support for training anappropriate number and balance of physicians whoare well-equipped to provide high quality, effec-tive and efficient care.

CURRENT FINANCING OFGRADUATE MEDICAL EDUCATION

As noted above, GME is currently fundedthrough a variety of mechanisms. Medicare and, inmost States, Medicaid make explicit payments to

teaching hospitals for the costs of GME. An under-standing of the GME payment policies used by theseprograms is fundamental to any discussion regardinghow all-payer GME funds should be allocated.

MEDICARE PAYMENTS

In FY2000, Medicare will pay an estimated $2.7billion (including $200 million for managed careenrollees) in direct GME payments and $5.1 bil-lion (including about $700 million for managed careenrollees) in indirect payments (HCFA, 2000). Pay-ments are linked to services provided to Medicarebeneficiaries. There are three overarching concernsthat suggest broader based funding is preferable torelying heavily on Medicare payments for patientcare services as the only Federal financing mecha-nism. These issues are:

• Restricting payments to teaching hospitals foreducational costs impedes the development ofresidency programs in non-hospital ambula-tory and managed care settings.

• Linking educational payments to services fur-nished to Medicare patients concentrates Fed-eral support on providers with high Medicareutilization and offers little support to provid-ers with low Medicare utilization such aschildren’s hospitals and Federally qualifiedhealth centers (FQHCs). It provides little sup-port for preventive medicine and other resi-dency programs that do not involve direct pa-tient care services.

• Paying for educational costs through patientcare payments alone is not an effective mecha-nism for achieving specific workforce priori-ties such as improving the specialty and geo-graphic distribution of the physicianworkforce.

Recent changes in national policy have begunto address these concerns, but COGME believesfurther changes are in order.

DIRECT GME PAYMENTS

Medicare’s payments for direct GME costs arebased on the number of residents at the hospital(and ambulatory settings if the hospital assumessubstantially all of the training costs), a hospital-specific per resident amount based on 1984 costsupdated for inflation, and Medicare’s share of hos-pital inpatient days. Medicare’s average per resi-dent payment was $22,350 in FY1997 (HCFA/HCRIS, 1997). Until the BBA, the Medicare lawauthorized direct GME payments only to hospitals.Beginning January 1, 1998, Medicare may make


direct GME payments to other provider entities. Thehospital-specific per resident amounts vary widelybased on historical accounting practices and finan-cial arrangements between the program sponsor andthe teaching sites. The differences do not appear tobe related to factors such as cost of living or thequality of the residency programs (Anderson, 1996).The Balanced Budget Refinement Act of 1999 madea modest change to reduce disparities in the Medi-care per resident amounts. The provision raises theminimum payment to 70 percent of a national wage-adjusted per resident amount. The annual inflationupdates for per resident amounts that are above 140percent of the wage-adjusted national average arereduced for FY2001-FY2005.

IME P AYMENTS

Medicare pays acute care hospitals for inpatientservices based on a prospectively determined ratethat takes into account average resources requiredto treat Medicare beneficiaries in the same diagno-sis-related group. Medicare makes an indirect teach-ing adjustment to the standard rate to pay for addi-tional patient care costs attributable to teachingactivity that are not captured as direct GME costs.Although based on the hospital’s ratio of residents-to-beds, the adjustment also compensates the hos-pital for higher patient care costs typically associ-ated with other activities provided in conjunctionwith GME, i.e., clinical research, specialized carefor complex patients, and charity care. However,the level of involvement in the other public mis-sions varies across teaching hospitals and teachingintensity is not a good measure to use to supportthese other public missions.

IME payments are about $1.5 billion higherthan the costs attributable to teaching intensityalone (MedPAC, 2000b). The difference betweencurrent IME payment levels and the analyticallyjustified levels represents amounts that could beredirected to support educational activities to in-crease community-based training capacity or toincrease support for uncompensated care. It is im-portant that refinements in the IME payment meth-odology not adversely affect hospitals that providecharity care. Major teaching hospitals with highuncompensated care costs rely on IME paymentsto support their charity care. Any refinements inthe IME allocation methodology that reduces pay-ments to these hospitals should be offset by highersupport for uncompensated care until specific fund-ing for such services is provided.

L IMITATIONS ON THE NUMBER OF RESIDENTS

RECOGNIZED BY MEDICARE

The Balanced Budget Act of 1997 limits thenumber of residents that are counted for Medicare

payment purposes to the number working at thehospital during its cost reporting period ending in1996. The limits are in line with COGME’s goal toreduce the number of residents (COGME, 1999b).However, hospital-specific limits impede the abil-ity of program directors to shift residents for edu-cational reasons. The limit is also problematic inpopulation growth areas where expanding residencyprograms would improve the geographic distribu-tion of physicians. The Balanced Budget Refine-ment Act of 1999 allows a 30 percent expansion inrural residency programs and recognizes that ur-ban hospitals may operate a rural residency track.Although it is too early to assess its overall impact,the 30 percent tolerance allows for only negligibleexpansions in relatively small residency programs.

PAYMENTS FOR RESIDENCY TRAINING IN NON-HOSPITAL BASED SETTINGS

The Balanced Budget Act of 1997 made sev-eral significant changes in Medicare payment forresidency training in ambulatory settings. First, thelegislation authorized direct payments to be madeto entities other than hospitals. Medicare-partici-pating Federally qualified community health cen-ters, rural health clinics, Medicare+Choice organi-zations, and other entities designated by Medicaremay be paid for direct GME if the provider incursall or substantially all of the costs of training at thesite. It is too early to determine the effect these pro-visions may have on support for community-basedtraining.

Second, the law allows the hospital to includein their IME resident count (as well as direct GMEresident count) the time residents spend in patientcare activities at non-hospital settings if the hospi-tal incurs all or substantially all of the training costsat the ambulatory site. Since the time residents spentin non-hospital settings could not be included inthe hospital’s resident count prior to the BBA, theresidents who were training at ambulatory sites in1996 are not represented in the hospital’s residentlimit for IME payments. Hospitals that had alreadydeveloped community-based training sites by 1996receive no benefit from the change unless there arereductions in other residency programs at the hos-pital. By contrast, hospitals that developed com-munity-based training sites after the BBA was en-acted, could include those residents in their IMEcount after moving them into community-basedsettings.

TEACHING PHYSICIAN SERVICES

The direct GME payment is intended to coverthe hospital’s compensation to teaching physiciansfor time spent on GME program administration and


general teaching and supervision of residents. Inaddition, payment may be made under the Medi-care physician fee schedule for professional servicesfurnished by the physician or by a resident underthe medical direction of the teaching physician. Toavoid paying for the same service twice, Medicarerules were revised in 1996 to require that the teach-ing physician be present during the key portion ofa billable service. For evaluation and managementservices and minor procedures, the teachingphysician’s presence must be the same as it wouldbe had the teaching physician personally performedthe entire visit or procedure. Some teaching insti-tutions believe the policy is inconsistent with theeducational goal of encouraging progressively in-dependent decision-making. Additional documen-tation requirements and heightened attention tocompliance has resulted in more time spent ondocumentation and less time on teaching.

MEDICAID PAYMENTS

Under the Medicaid law, States have consider-able flexibility to determine what services will becovered and the payment methodologies that willbe used to pay for covered services. Federal match-ing funds apply to the State’s expenditures. Statesspent about $2.3 billion through Medicaid patientfunds in 1998, or approximately 7 percent of Med-icaid inpatient hospital expenditures, to supportGME (Henderson, 1999). Most Medicaid programssupport GME in their payments for inpatient hos-pital services following Medicare-like methodolo-gies. However, some States have approval fromHCFA to link some or all of Medicaid GME pay-ments to specific State workforce objectives. Theseprograms provide funding models that should beconsidered within the context of GME financingreform. For example:

• Michigan and New York have established sepa-rate incentive pools to achieve specificworkforce objectives.

• Tennessee and Oklahoma make GME pay-ments to medical schools.

• In Minnesota, advanced nursing, pharmacy,dental and physician assistant training pro-grams as well as medical and dental schoolsare eligible for medical education payments.

• Utah is seeking a State plan amendment forMedicaid GME funds to flow to a consortiumconsisting of the State, the single universitymedical school in the State, hospitals and com-munity-based providers, and private payers. Awaiver request is also under development forMedicare funds to flow to the consortium.

RESIDENCY TRAINING INCOMMUNITY-BASED SETTINGS

The growth of managed care has been accom-panied by reductions in hospital utilization and ashift of services from inpatient to ambulatory set-tings (U.S. Congress, 1999). The shift to ambula-tory settings makes it increasingly important forresidency programs to provide training opportuni-ties in community settings that are representativeof the environment in which residents will eventu-ally practice (COGME, 1999a). To practice effec-tively in the changing health care environment, resi-dents in all specialties need a comprehensive rangeof experiences that include opportunities to followthe patient across each component of an integrateddelivery system. Community-based settings suchas health centers and clinics, physician offices,schools and workplaces, nursing homes, hospicesand home care, community hospitals, and man-aged care organizations can offer essential expe-riences to complement those at academic healthcenters.

There is general agreement that training expe-riences in the community are important and shouldbe expanded. However, as evidenced by the lack ofspecific requirements for community training bythe accrediting organizations for most specialties,there is no consensus on what constitutes the ap-propriate balance between traditional and commu-nity-based experiences. For most specialties, train-ing opportunities will be expanded as needed tomaintain an adequate patient base for teaching.However, without the impetus from the accreditingorganizations to move training into the communitysettings, financial and other considerations at thehospital as well as the community site mayoutweigh the educational goal of providing com-munity-based training experiences.

FINANCIAL BARRIERS TO COMMUNITY-BASED TRAINING

There are several financial barriers to expand-ing community-based training. Ambulatory train-ing tends to be less efficient and more faculty-in-tensive than inpatient training. In the inpatientsetting, faculty can teach students at different edu-cational levels during patient rounds. In the ambu-latory setting, patients are available generally onlyfor a short period of time. Space constraints andthe focus on clinical productivity limit opportuni-ties for teaching multiple students at the same time(Lave, 1989; Boex, 1997; Philibert, 1999). In addi-tion, physician revenues for ambulatory servicestend to be lower than revenues for services furnishedin the inpatient setting. Thus, patient care revenues


generated when residents are involved in ambula-tory care are generally less than when the care is ininpatient settings. Further, there is a distinctionbetween revenues generated by a hospital-basedclinic, which include Medicare (and in many StatesMedicaid) GME payments, and revenues generatedby community clinics. The latter do not receiveMedicare payments directly (unless they qualifyunder the BBA provision) and, in the case of com-munity health centers, have few privately insuredpatients and high uncompensated costs.

Some studies have tried to measure the effectof teaching on the ambulatory site’s productivityand net revenues. These studies are directed at thequestion of whether there are GME costs in ambu-latory settings that are not recovered through pa-tient care revenues. The findings suggest net costsare site-dependent and reflect factors such as theamount of time spent on teaching relative to pa-tient care, patient flow and the efficiency of the prac-tice, payer mix, the physicians’ incomes and resi-dents’ salaries. The studies support a widespreadbelief that residency programs “break even.” Firstyear residents generate productivity losses. Secondyear residents generally have no effect on produc-tivity. Net increases in productivity and revenuesby the third year of residency training in ambula-tory sites offset the first-year losses (Diamond,1993; Flannagan, 1995; Lave, 1989).

Another potential barrier is the impact residentrotations to community-based settings have on theprimary training site and its teaching faculty. Manyresidents have significant teaching responsibilitiesfor medical students that may make community-based rotations more difficult. It is possible this is-sue may diminish as medical students increasinglyreceive training in community settings. Residentsalso provide coverage for teaching physicians andcontribute to the services billed by teaching physi-cians. Resident replacement costs can be a signifi-cant factor in decisions to rotate residents to non-hospital settings. The cost of replacing residentswill depend on the relative cost of the staff that arereplacing them, their productivity, and their abilityto generate revenue.

SITE VISIT FINDINGS

As part of this study, site visits or telephoneinterviews were conducted with individuals in-volved with selected residency programs with com-munity-based training sites. The interviews in-cluded both programs with long-standingcommunity-based training sites and relatively newprograms that are developing collaborative arrange-ments with managed care organizations or commu-

nity providers in medically underserved areas. Theinterviews focused on the funding sources and ar-rangements used to support ambulatory training.The findings reflect the circumstances of the pro-grams where the interviews were conducted. How-ever, similar themes emerged from the interviewswith different sites that are also consistent withother information on GME financing issues.

Key findings from the site visits are that financ-ing arrangements for community-based trainingvary, reflect local circumstances, and frequentlyinvolve other interrelationships as well. An indi-vidual program may have arrangements for teach-ing with hospital-based clinics, hospital-operatedand hospital-affiliated physician practices, commu-nity health centers, and individual clinician-physi-cians in private practice. The financing arrange-ments differ for each site depending on a numberof factors, including payer mix and the intensity ofthe teaching effort. The financing issues for hospi-tal-based clinics are quite different than those forcommunity clinics and physician practices. Thevariety of arrangements suggests that decisions onhow GME funds should be allocated among thevarious participants in a given program are bestmade at the local level. A single national policythat allocates funds between hospital and commu-nity-based sites using a pre-determined formuladoes not acknowledge the myriad of existing ar-rangements for community-based training thatcould be disrupted.

Most teaching in community-based sites occursthrough volunteer faculty. During the interviews,some concern was expressed that clinician-educa-tors in the community are becoming less willing toteach without compensation. Commonly cited rea-sons are competitive pressures for clinical produc-tivity and the Medicare teaching physician rules.Specific policies may be needed to assure commu-nity physicians receive appropriate compensationfor their teaching activities.

ALTERNATIVE MODELS FOR GMEFINANCING REFORM

The changing health care environment and theuncertainties of continued reliance on Medicare,Medicaid and private pay revenues to fund GMEhave led COGME and others to conclude signifi-cant changes are needed in the way GME is fi-nanced. In addition to COGME, advocates of anall-payer fund include the Pew Health ProfessionsCommission, the Commonwealth Fund’s Taskforceon Academic Health Centers, and the 1997 Con-sensus Statement on the Physician Workforce by


associations representing physicians and teachinginstitutions. Two bills (H.R. 1224 and S.210) wereintroduced in the 106th Congress that would pro-vide all-payer funding for GME. Underlying theall-payer proposals is an assumption that GME is apublic good that merits broad support. An alterna-tive assumption made by the Medicare PaymentAdvisory Commission is that GME results in en-hanced patient care that should be recognized inhealth care service-related payments to teachinghospitals.

POLICY CONSIDERATIONS FOR GMEFUNDING

GME funding policies should meet the follow-ing objectives:

• Provide a stable funding mechanism that isresponsive to the community yet consistentwith the national workforce objectives;

• Enable health care institutions to compete onprice and quality by subsidizing higher costsattributable to educational activities and un-compensated care without supporting ineffi-ciencies;

• Create adequate support and appropriate in-centives for developing community-based edu-cational programs;

• Encourage effective and efficient educationalmodels that promote improved ways to meethealth care needs;

• Foster mechanisms that will stabilize the totalnumber of physicians while improving thespecialty and geographic distribution of thefuture physician workforce; and,

• Hold recipients of Federal and State funds ac-countable for producing needed public goods.

Educating physicians in the environment inwhich they will eventually practice requires expan-sion of residency training in community settingsfor most specialties. The BBA eliminated some dis-incentives for a hospital to rotate residents to non-hospital settings by paying IME for the resident’stime in the non-hospital setting. However, the BBAlimits on FTE resident counts do not benefit thosehospitals that were already training residents incommunity-based settings. A modification in thelaw to count the residents who were in non-hospi-tal settings in the 1996 base year would remedythis problem. The remaining disincentives to ro-tate residents to community-based settings are moredifficult to resolve:

• As the recipient of the funds, the hospital is inthe stronger bargaining position regarding who

should bear the direct costs of training in theambulatory setting.

• Hospital service demands compete with edu-cational needs to rotate residents to commu-nity-based settings.

• The hospital’s patient care needs and finan-cial interests rather than physician workforceneeds may determine the numbers and spe-cialty mix of residents and residency programs.

• Accountability is difficult to establish becausethe program sponsor, rather than the hospitalhas the ultimate responsibility for the conductof the educational program. In hospital-spon-sored programs, this problem arises with re-spect to resident rotations to other hospitals.

Fundamental issues to consider in evaluatingalternative models for distributing all-payer GMEfunds are:

• Who should receive payments directly fromthe fund;

• How the funds should be allocated among thereceiving entities; and,

• How to establish accountability for the funds.

There are four basic models that potentiallycould be used to distribute GME funds. The differ-ences between three models are based on whichentities would receive payments directly from thefund: health care providers, educational institutions,or GME planning bodies. The fourth model wouldlink payment to specific performance measures. Themodels are not mutually exclusive and a combina-tion is needed to achieve the policy objectives forthe GME fund.

HEALTH CARE PROVIDER MODEL

The health care provider model links paymentsfor clinical training to patient care activities. Thisis the approach Medicare uses. It treats clinicaltraining costs as patient care costs as opposed toeducational costs. It is the most appropriate modelfor the indirect costs of clinical training since thesecosts reflect the impact of the teaching activity onthe patient care costs of the site where the trainingtakes place. The health care provider model pro-vides no support for training that does not occur inpatient care settings. Residency programs in pre-ventive medicine would typically receive little fund-ing through this model.

There are several options regarding which en-tity should receive direct GME payments for train-ing in community-based settings:


• Consistent with Medicare’s current rules, pay-ment could flow to the hospital or the com-munity training site depending on which en-tity bears substantially all the training costsfor the resident’s time at the site, i.e. pays theresident’s salary and reasonable compensationfor teaching.

• An alternative would be for funding to followthe resident to the community site regardlessof which entity bears the cost.

• Another option would be to pay pro rataamounts to the hospitals and the community-based ambulatory sites participating in the resi-dency program based on their relative sharesof GME program costs. One approach wouldrequire the affiliated hospitals and community-based training sites to agree on their relativeshares of program costs. A variation would beto designate a fixed percentage of the paymentto hospitals for training in community settings.

EDUCATION MODEL

The education model treats direct GME costsas an educational cost as opposed to a patient carecost. It should give educational needs more weightin deciding what and where residency training oc-curs. The model is appropriate for direct GME costsonly. Payment would flow to a single entity thatwould be accountable for how the funds were ex-pended. This could be to a medical school or pro-gram sponsor. There is potential to introduce otherelements into the allocation formula, such as pro-gram quality or meeting specific educational ob-jectives or workforce priorities. Making direct GMEpayments to medical schools would emphasize moststrongly that Federal GME funds are to support edu-cation rather than patient care. However, fundingthrough the medical schools for residency programsthat are sponsored by community-based hospitalswould require a major shift in program account-ability and funding. Another alternative, givingvouchers to residents, adds value only if there is aregulatory apparatus to determine the number ofpositions to be funded and which residents shouldreceive funding.

PLANNING MODEL

Under the planning model, the GME fundswould flow through a GME planning and coordi-nating body. This is an entity whose primary func-tions would be to assess the health care needs ofthe community and allocate funds based on localworkforce considerations. Funding allocation de-cisions could support local health care needs more

readily than distributions directly to health careproviders or program sponsors based on nationalallocation formulae.

The funds could be distributed to GME con-sortia or to the States. Payment to consortia couldbe either a natural outgrowth of paying the programsponsor or could result from deliberate policies toencourage consortia. Experience to date suggeststhat providing payment only to consortia would bepremature. Many States have established commis-sions or task forces related to GME and physicianworkforce. Some are permanent groups with thestructure and expertise that could become physi-cian workforce planning bodies in their State(COGME, 1999b).

PERFORMANCE MODEL

The performance model would link payment toachieving specific performance measures or objec-tives. Payment could be formula-driven based onmeeting specific educational or workforce objec-tives. These objectives could be in the area of cur-riculum content and quality, training opportunities,or workforce objectives. Alternatively, there couldbe support for specific projects or demonstrationsneeded to support infrastructure development orworkforce goals. Examples of the types of programsthat could be funded include the targeted PublicHealth Service grants for primary care residencyprograms, support for developing academic-com-munity partnerships to serve medically underservedpopulations, consortia development, and faculty de-velopment programs for clinician-educators in thecommunity.

The performance model is more suitable as amechanism for making supplemental payments thanas a primary payment mechanism. Educationalquality measures and workforce priorities are notsufficiently defined to be used to determine all fundallocations. Also, if all funding were predicated onmeeting specific performance measures, significantyear-to-year fluctuations in funding could occur thatwould be inconsistent with the need for stable GMEfunding.

RECOMMENDATIONS FOR GMEFINANCING REFORM

COGME’s 14th Report called for a stable, all-payer financing mechanism for graduate medicaleducation (GME) that would provide adequatefunding for training in ambulatory settings. Thisreport builds on the 14th Report by recommendinghow the all-payer GME funds should be distrib-


uted. The recommendations are based on COGME’spolicy goals for GME funding.

RECOMMENDATION 1

CREATE A GME FUND THAT COMBINESFEDERAL FUNDING TO SUPPORT GRADUATEMEDICAL EDUCATION WITH ALL-PAYERFUNDS.

To assure financing policies are consistentacross Federal programs and reflect nationalworkforce priorities, the various Federal fundingstreams for GME that is provided by non-Federalinstitutions (i.e., excluding DoD and DVA) shouldbe combined into a single fund and supplementedwith all-payer funds obtained through a modest sur-charge of private insurance premiums. The GMEfund would include amounts that would otherwisebe paid under current formulae for Medicare fordirect GME and indirect payments to teaching hos-pitals, the Federal portion of Medicaid paymentsthat are implicitly GME payments, and theChildren’s Hospital GME fund. In addition, theHRSA Title VII grants for GME, e.g., primary careresidency training grants, would be included in aset-aside fund for specific workforce goals.

Within the general GME fund, five separateaccounts should be established for:

• Medicare direct GME payments;

• non-Medicare direct GME payments;

• Medicare IME payments;

• non-Medicare IME payments; and,

• targeted payments to support specificworkforce and educational objectives.

The separate Medicare and non-Medicare ac-counts are needed as a transitional measure. Theywould assure full funding on behalf of Medicarepatients if contributions from other payers are notsufficient. Also, since the Medicare funds are cur-rently being paid, changes in the allocation of theMedicare funds should be phased-in or offset byadditional funding from the non-Medicare ac-counts. A transition may not be needed for the non-Medicare funds. To the extent they represent newfunding streams, funds in the non-Medicare ac-counts should be allocated consistent with preferredpolicies from the outset.

While the GME fund would not include fund-ing for residency training in DoD and DVA spon-sored programs, the Federal budget for those pro-grams should be consistent with the policyobjectives for the GME fund. Residency trainingin these programs has significant impact on the size

and specialty composition of the physicianworkforce.

a. GME should be broadly supported by all-payers.

Explicit funding for GME should be spreadmore broadly across all sectors of society. Apermanent and stable funding source, such aspremium contributions from all health insur-ance plans, should supplement current Fed-eral funding for GME. In the long run, Medi-care and Medicaid’s contribution to the GMEfund should be proportionate to the percent-age of insured population represented by theirenrollees.

b. Funding from all sources should be sufficientto support high-quality, efficient training ofan appropriately sized physician workforce.

Total aggregate funding should be sufficientto support the efficient training of an appro-priately sized physician workforce. Togetherwith payments from other sources (primarilypatient care revenues and State funds), GMEfunding should be adequate to train the num-ber of physicians required to meet current andfuture national health care needs. Additionalfunding would not be in the public interestsince it could contribute to a continuing sur-plus of physicians.

In the past, COGME has recommended thatthe total number of physicians entering firstyear residency should not exceed the numberof U.S. medical school graduates in 1993 plus10 percent (COGME, 1994). The Council’s 14thReport found that a reduction of 3,386 first yearpositions in 1997-1998 was needed to meet the110 goal. In view of recent changes in the healthcare delivery system since its initial recommen-dations were issued, COGME plans to reviewthe 110 goal and its target of a 50/50 mix of pri-mary care and other specialists.

Most of the increase in the total number ofresidents in recent years is attributable to anincrease in the number of graduates from med-ical schools outside the United States. Sup-port should be discontinued for new exchangevisitors (J-1 visa) residents. As COGME haspreviously recommended (1997), exchangevisitor residents should be funded by alterna-tive sources, such as home country financingor foreign aid.

A conceptual framework should be used to es-tablish an appropriate level of Federal supportfor funded residency positions. Because GMEis a joint product with patient care services,patient care revenues cover some direct GME


costs. Consideration also needs to be given toissues such as maintenance of effort for cur-rent State funding through the Medicaid pro-gram and grant programs and whether all resi-dent activities required for accreditation in anapproved program should be funded. UnderMedicare, only resident time spent in patientcare activities is supported.

RECOMMENDATION 2

IME ACCOUNTS SHOULD PAY HOSPITALSAND OTHER CLINICAL TRAINING SITES ASAPPROPRIATE FOR THE INDIRECT COSTS OFEDUCATIONAL ACTIVITIES.

IME accounts should be created to subsidizehigher patient care costs associated with residencytraining. The funds should be allocated to hospi-tals and, to the extent it is empirically supported,to other clinical training sites that incur indirectteaching costs, including hospital outpatient clin-ics and community-based settings. Initially thereshould be separate accounts for Medicare and non-Medicare patients in order to assure the indirectcosts for Medicare patients are fully funded. In thelong run, a single account would be appropriate.

a. IME payments should be set at no more thanthe analytically justified level for teachingactivities.

Paying more than the analytically justifiedamount would subsidize inefficient providersand give teaching institutions a competitiveedge over non-teaching institutions. For Medi-care inpatient services, MedPAC’s (2000b)current estimate is a 3.1 percent adjustmentfor each 0.1 increment in the resident-to-bedratio after other refinements are made to theMedicare prospective payment system. Basedon this estimate, Medicare IME payments wouldbe $1.5 billion lower than the 5.5 percent adjust-ment provided by the BBA. The difference couldbe targeted toward achieving specific workforceand educational goals (see Recommendations4 and 6) or toward supporting uncompensatedcare (see Recommendation 8). A transitionwould be needed to the extent reductions inMedicare IME payments are not offset by in-creases in non-Medicare IME funding.

b. Research is needed to determine the appro-priate IME payment formulae.

Research is needed to refine the Medicare IMEadjustment and to determine the appropriateIME teaching adjustment for non-Medicarehospital inpatients. Medicare’s adjustmentshould be based on the higher costs attribut-

able to teaching activities. For non- Medicarepatients, the adjustment should be directed at“leveling the playing field” between teachingand non-teaching hospitals. It does not needto cover the full indirect teaching costs ifteaching hospitals are able to command a pre-mium for quality or specialized services.

Ideally, the IME payment formula should notreflect higher costs indirectly attributable toother teaching hospital missions, e.g. special-ized services, uncompensated care, and re-search. Subsidies for those public goodsshould be directed toward the hospitals pro-ducing them through separate funding streams.Reducing the adjustment to an analyticallyjustified level for teaching would reduce in-centives to train more physicians than neces-sary. It would also eliminate confusion be-tween funding for the teaching mission andfunding to support charity care. Higher costsattributable to serving low-income patientsand uncompensated care costs should be rec-ognized through a separate funding mecha-nism (which would also distribute paymentsto non-teaching institutions serving low-in-come patients. See Recommendation 8). Re-finements in the IME payment methodologyshould not reduce the total level of supportfor hospitals with significant uncompensatedcare until specific funding for such services isprovided. Reductions in the IME payment for-mula should be accompanied by refinementsin the prospective payment system to incorpo-rate better case-mix and severity measurements.An additional adjustment for research-inten-sive hospitals may also be appropriate.

Additional research is also needed to deter-mine the extent to which there is an indirectteaching effect on costs when resident train-ing takes place in hospital outpatient and non-hospital settings. If empirical research findsthere is an indirect teaching effect on the costsof services provided in ambulatory/commu-nity settings, the IME account should pay forthese services as well as inpatient hospitalservices.

RECOMMENDATION 3

DIRECT GME ACCOUNTS SHOULD PAYPROGRAM SPONSORS OR THEIR DESIGNEESFOR THE DIRECT COSTS OF GRADUATEMEDICAL EDUCATION.

Direct GME costs are educational costs thatshould be supported through payments to the spon-soring institution ultimately responsible for the


graduate medical education program. Payment al-location decisions should be made at the local levelbecause the tremendous variety of existing arrange-ments cannot be accommodated at the nationallevel. By making payments to either the sponsor-ing institution or its designees, the sponsor can de-termine the most appropriate recipient of the fundsbased on local circumstances for a particular pro-gram. For example, a sponsoring institution maydecide to retain maximum control over the fundsand receive them directly, elect to continue histori-cal arrangements having the funds flow through theteaching hospital, or may choose to have a consor-tium distribute the funds. The same election wouldnot need to apply to each program sponsored bythe institution. Regardless of which entity receivedthe funding, the sponsoring institution would be ac-countable for the funds being expended to supporta high quality training program with the appropri-ate balance of hospital and community-based train-ing experiences.

a. There should be written agreements betweenthe program sponsor and training sites indi-cating the sponsor is assuming substantiallyall of the training costs and describing howGME payments will be allocated.

The program sponsor or its designee must as-sume all or substantially all of the direct costsof operating the residency program as a con-dition of receiving direct GME payments.Written agreements should be required be-tween the sponsoring institution and clinicaltraining sites to formalize the negotiation pro-cess and to increase accountability for thefunds. The agreements should detail how thedirect GME funds will be allocated betweenthe sponsor and the training site, identifywhich entity will pay resident salaries andfringe benefits, and specify teaching physiciancompensation arrangements for supervisingresidents. The goal is to strengthen the nego-tiating position of community–based siteswithout jeopardizing long-standing relation-ships between academic institutions and com-munity training sites. A sponsoring hospitalmay have a disincentive to rotate residents tocommunity-based training sites if all directGME funds automatically follow the residentto a community-based training site throughdirect payments from the GME fund or avoucher system.

b. Model agreements and information on directGME costs should be made available to fa-cilitate equitable agreements between thesponsor and the sites.

Local circumstances should determine how di-rect GME payments are allocated to teaching

sites. However, benchmarking informationshould be provided to facilitate the negotia-tion process, including:

• breakdown of GME payments into threecomponents based on average direct GMEcosts: resident salaries and related costs,teaching physician compensation, and anadministrative and overhead cost compo-nent.

• benchmarks for teaching physician compen-sation and the added time per teaching ses-sion when residents are present in commu-nity-based practices on short-term rotationsand on an on-going basis; and,

• model agreements between institutionalsponsors and community-based sites.

c. Require separate reporting of resident timespent in inpatient hospital, hospital outpa-tient and community settings.

At present, there is no formal accounting forthe time residents spend in each type of train-ing site. Standard definitions should be devel-oped to distinguish hospital outpatient settingsfrom community settings. Community settingsshould be broadly defined to include both hos-pital-operated and community-based sites thatare representative of the environment in whichresidents will eventually practice. The deter-mining characteristics are the processes of carerather than proximity to the hospital or pro-vider ownership. Community settings addressthe care of the individual patient in the con-text of the population of which the patient is amember. They teach residents to deliverculturally effective care to an ethnically andracially diverse population.

RECOMMENDATION 4

ESTABLISH A NATIONAL AVERAGE PERRESIDENT PAYMENT FOR DIRECT GMECOSTS.

The base payment for direct GME costs shouldvary only for differences in the cost of living acrossgeographic areas. For Medicare payments, thereshould be a transition from the hospital-specific perresident amounts to the national per resident pay-ment. The length of the transition will depend onadditional payments for non-Medicare patients.These can help compensate for any reductions inMedicare payments. At the end of the transition,separate Medicare and non-Medicare accountswould no longer be necessary. Higher paymentsmay be appropriate for training in community-basedsettings. In addition, there should be an incentive


payment for programs that meet specific workforceor educational objectives.

a. Base total direct GME payments on the netcosts of supporting an appropriately sizedworkforce.

Ultimately, total direct GME funding shouldbe based on the net costs of educating an ap-propriately sized physician workforce. Estab-lishing a fixed payment per resident shouldprovide incentives for efficiency in the edu-cational process. However, the costs of effi-ciently delivering high quality GME and theextent to which these costs are offset by pa-tient care revenues has not been determined.As an interim policy, either the average perresident amounts or average GME costs perresident could be assumed to represent thetotal costs of an efficient program. Total costsbased on the FY1997 average per residentamount updated for inflation and the 110 per-cent target are estimated at $6 billion forFY2000.1* A lower funding amount would beappropriate since the per resident amounts donot take into account patient care revenues at-tributable to GME.

b. Provide higher payments for training in com-munity settings.

When training occurs in a community setting,the sponsoring institution continues to incursome supervisory physician and overheadcosts. The community setting incurs some di-rect GME costs as well (for example, to com-pensate the community physician for teach-ing) even if the sponsoring institutioncontinues to pay resident salaries and fringebenefits. As a result, total GME costs may behigher when residents rotate to community-based settings than when they remain in hos-pital-based settings. A higher payment fortraining in community-based settings wouldbe appropriate if the net total costs (after tak-ing any additional patient care revenue intoaccount) are higher in the community-basedsettings. Research is needed to determinewhether this is the case. To counter any disin-centive that might currently exist for commu-nity-based rotations, a temporary policy mightbe to increase the component of the per resi-dent amount attributable to teaching physiciancompensation by a fixed percentage, e.g. 25percent. The higher payment could apply inall community settings even though the ratio-nale for the payment is primarily applicable

to settings off the hospital premises. Thiswould provide an incentive for hospitals toturn training in ambulatory clinics into expe-riences that are more representative of com-munity physician practices.

c. Provide incentive payment for meeting spe-cific workforce and educational objectives.

In addition to the base per resident payment,there should be an incentive payment for meet-ing specific workforce and educational objec-tives. Programs that meet one or more of theobjectives would be eligible for a bonus onthe national average base payment. The incen-tive payment should be established as a fixedpayment rather than a pool so that the ben-efits do not erode as additional programs meetthe objectives and qualify for payment. Thebonus payment could be awarded based on:

• Participation in a broad-based consortia ofthe sponsoring institution(s) for residencyprograms in an area, hospitals and commu-nity providers participating in GME activities,and community representatives. The consor-tia would have to be designated by the spon-soring institutions to receive all direct GMEfunds. Bonus payments would be made if theconsortia has a formal process to identify thehealth care needs of the community, engagein workforce planning, and promote commu-nity-based training opportunities;

• Number of graduates that provide signifi-cant amounts of care to medicallyunderserved populations;

• Percentage of time residents spend provid-ing care to medically underserved popula-tions; and,

• Quality of the residency program.

d. Research is needed to understand variationin direct GME costs by specialty and setting.

Most research regarding residency trainingcosts was conducted at a limited number ofsites and before the growth of managed sys-tems of care. A better understanding of differ-ences in the net costs of training across resi-dency programs and training sites is neededto refine the payment allocation methodology.A generic financial model should be used toexamine systematically issues such as:

• Whether there are significant differences inthe amount of teaching physician involve-ment between primary care and non-pri-mary care residency programs and betweeninitial residencies and fellowships;

* Numbered footnotes are listed on pages 71-72 in the “Bibli-ography and Footnotes” section.


• Whether there are significant differences inimpact on clinical productivity and net phy-sician practice revenue when teaching oc-curs in ambulatory settings relative to inpa-tient settings and between initial residenciesand fellowships;

• How direct GME costs are affected by thepresence of students in other health profes-sions at the training site and by residentsteaching medical students; and,

• Factors that affect the efficiency and qual-ity of the educational process.

RECOMMENDATION 5

CONTINUE THE BALANCED BUDGET ACTOF 1997 LIMITS ON THE NUMBER OF RESI-DENTS WITH MODIFICATIONS.

In concept, the Balanced Budget Act of 1997limits on the residents that will be recognized byMedicare are consistent with the goal of reducingthe future physician workforce and should be car-ried over to eligibility for payments under the GMEfund. However, hospital-specific limits are not anappropriate long-term way to deal with physiciansupply issues. The limits hamper a programdirector’s ability to move residents among hospitalprograms for educational reasons. In geographicareas with physician shortages, the limits precludeexpansions in needed residency programs.

a. Modify the caps to apply to sponsoring insti-tutions rather than hospitals.

The sponsoring institution is to be held ac-countable for educational outcomes andworkforce objectives. Therefore, the limitsshould apply to the number of residents in theprograms sponsored by the institution ratherthan the hospitals serving as training sites forthe program. This will provide the flexibilityto move residents between hospitals and othersettings. Consortia that meet certain workforceplanning objectives should be able to workunder an aggregate limit for multiple sponsor-ing institutions.

Applying the limit to sponsoring institutionsis consistent with making direct GME pay-ments to sponsoring institutions. However, itcomplicates IME payments to individual hos-pitals. One option would be to apply no limitfor IME as long as the sponsoring institutionswhose residents are training at the hospital areunder their caps. An adjustment would beneeded only if the total number of residentsin programs at the sponsoring institution(s) ex-

ceeded an aggregate cap. The three-year roll-ing average and the one-year cap on the resi-dent-to-bed ratio should continue to apply. Therolling average provides a form of transitionpayments to hospitals that reduce the numberof residents and slows the recognition of newresidents in the IME count.

b. Include residents in non-hospital settingsregardless of who paid the resident’s salaryin the 1996 base year count.

Residents who were working in non-hospitalsettings were not included in a hospital’s 1996base year count for Medicare IME payments.They were included in the direct GME countonly if the hospital incurred substantially allof the training costs. The limit applicable to asponsoring institution should be adjusted to in-clude all resident time in non-hospital settingsregardless of who paid the resident’s salary.

c. Allow adjustments in the Balanced BudgetAct of 1997 limits to improve the distributionof physician workforce.

Further research is needed to understand theimpact of the Balanced Budget Act of 1997limits and to develop policies that will resultin a better geographic balance in the physi-cian workforce while encouraging an overallreduction in the number of physicians. Appro-priate indicators of adequate distribution byspecialty are needed. As an interim measure,the limits should not apply to rural residencytraining tracks if their graduates practice pre-dominately in rural areas. In addition, the lim-its should not apply to primary care residencyprograms whose graduates practice predomi-nately in those States with low physician-to-population ratios.

RECOMMENDATION 6

ESTABLISH AN ACCOUNT FOR FUNDINGSPECIAL PROJECTS AND PROGRAMS DI-RECTED AT BUILDING HIGH-QUALITY COM-MUNITY-BASED TRAINING CAPACITY ORACHIEVING SPECIFIC WORKFORCE GOALS.

At least 10 percent of the GME fund should beset aside to support specific projects and programsdirected at building high-quality community-basedtraining or achieving specific workforce priorities.The types of projects and programs that should befunded include:

• Primary care residency program grants;

• Faculty development grants to support train-ing of community clinician teachers;


• Information technology infrastructure devel-opment to link patient care records at teach-ing hospitals and community sites within or-ganized systems of care;

• Incentive programs to reward residents that fo-cus their practice on medically underservedpopulations;

• Transition funds to cover residency replace-ment costs in hospitals with high uncompen-sated care patient loads; and,

• Demonstration projects involving develop-ment of broad-based consortia.

RECOMMENDATION 7

MODIFY THE MEDICARE RULES RELATEDTO TEACHING PHYSICIANS TO EMPHASIZETHE TEACHING PHYSICIAN’S OVERALL RE-SPONSIBILITY FOR THE MANAGEMENT OFTHE PATIENT’S CARE AND TO REDUCE THEIMPORTANCE OF DOCUMENTATION.

The Medicare rules pose two challenges tograduate medical education: 1) the supervision rulesmake it more difficult for residents to become pro-gressively independent, and 2) the documentationrequirements detract from the amount of time avail-able for teaching and resident supervision. Thereis some evidence that the rules may adversely af-fect the willingness of community physicians toparticipate in teaching programs. The rules shouldbe revised to address these concerns.

a. Establish different rules for residents in fel-lowship programs.

HCFA’s rationale for its teaching physicianrules rests on: 1) making payment only whenthere is an identifiable physician service to anindividual patient and 2) avoiding duplicatepayment for the physician’s supervisory time.Duplicate payment should not be an issue withresidents who are beyond their initial resi-dency period. These residents count as only.5 FTE and resident salaries and fringe ben-efits comprise only 43 percent of the per resi-dent amount.

The rules on teaching physician supervisionshould be revised for residents who are be-yond their initial residency program to permitMedicare billing if:

• the teaching physician is immediately avail-able;

• reviews with each resident during or imme-diately after the visit the patient’s medicalhistory and care; and,

• documents his or her participation in thereview and direction of services.

Accreditation standards for the residency pro-gram should be relied upon to determine issuesregarding supervision requirements for spe-cific services and resident-to-preceptor ratios.

An alternative would be to allow residents be-yond their initial residency period to bill forservices furnished in the hospital if an elec-tion is made to forego a direct GME paymentfor the resident’s time.

b. Evaluate the impact of the teaching physi-cian rules.

The issue of when physician billing is appro-priate for care provided by residents is not lim-ited to the Medicare program. It will remainrelevant under an all-payer fund. There is aneed to evaluate formally the administrativeand teaching burden associated with the cur-rent Medicare rules and their impact on thequality of clinical training and patient care.Particular attention should be paid to the ef-fect in hospital ambulatory clinics and com-munity-based settings where there is a lowresident-to-preceptor ratio.

c. Develop clear and reasonable documentationrequirements.

There is need for additional guidance and com-mon understanding of what constitutes ad-equate documentation of a teachingphysician’s participation in the care of patientsinvolving residents. HCFA should work withthe academic physician community and theOffice of Inspector General to develop reason-able standards that do not compromise highquality clinical education. The standardsshould provide a reasonable means fordocumenting the teaching physician’s involve-ment in the care of the patient and assuringappropriate payments without imposing un-due administrative burden. They should betested in a variety of teaching settings and spe-cialty programs before implementation.

RECOMMENDATION 8

PROVIDE ADDITIONAL SUPPORT FORHOSPITALS AND COMMUNITY-BASED TRAIN-ING SITES THAT SERVE A DISPROPORTION-ATE SHARE OF LOW INCOME PATIENTS.

In the absence of national health insurance,“safety net” providers should be provided with ad-ditional funding to cover uncompensated care costs.


Major teaching hospitals provide substantial un-compensated care. Faculty practice plans also fur-nish charity care. Uncompensated care is not aneducational cost. However, it affects the trainingsite’s ability to provide high quality educationalexperiences. More importantly, teaching institutionsthat furnish high amounts of uncompensated carerely on current GME funding to support their char-ity care. As changes are made in the IME paymentmethodology, the current level and distribution of

DSH payments should be examined to assure thefunds are well targeted to subsidize uncompensatedcare. The subsidies should apply to hospitals forboth inpatient and outpatient services and to com-munity-based providers. Without additional sup-port, GME is not sustainable in community-basedtraining sites with a high volume of uncompensatedcare. These sites cannot generate the patient care rev-enues needed to support their educational activities.


BACKGROUNDThe 13th Report of the Council on Graduate

Medical Education (COGME), Physician Educa-tion for a Changing Health Care Environment, em-phasized that the comprehensive preparation of allmodern physicians requires training experience inboth traditional and community settings. Notingthat care is increasingly provided in the context ofintegrated delivery systems, the Council concludedthat effective medical practice requires an under-standing of the characteristics of the populationserved and the attributes of the delivery system inwhich care is provided. COGME recommended thateducational programs address the care of the indi-vidual in the context of the patient’s communityand include clinical learning experiences across thecontinuum of care. Training should include oppor-tunities to participate in a multi-disciplinary teamapproach to patient care. The Council advocatedthe development of high quality, community-basedclinical teaching opportunities with a faculty in-corporating community clinician-teachers.

COGME found that the current system of fund-ing graduate medical education (GME) throughteaching hospitals has inherent limitations and dis-incentives for developing ambulatory clinical train-ing experiences and community-based educationalprograms. In its 14th Report, COGME PhysicianWorkforce Policies: Recent Developments and Re-maining Challenges in Meeting National Goals,COGME called for a stable, all-payer financingmechanism for GME that would provide adequatefunding for training in ambulatory settings. TheCouncil noted recent developments to support am-bulatory training, including actions by residencyreview committees, medical schools, foundations,States, and changes in Medicare financing policies.However, the Council also expressed concern overwhether the changes are adequate to support a suf-ficient shift to ambulatory settings and suggestedthat further revisions in financing policies may benecessary.

PURPOSEThe purpose of this report is to create a policy

framework and set of recommendations for changesin GME financing policies. The recommendedchanges are intended to provide appropriate incen-tives for high quality residency training in both tra-ditional and community settings. As used in this

Introduction

report, the term “community settings” describessettings that are representative of the environmentin which residents will eventually practice. Underthis definition, the processes of care and educationaloutcomes are the determining factors in identify-ing a community setting rather than its location perse. Both hospital-based clinics and community-based ambulatory care providers may serve as com-munity training sites.

The report builds on COGME’s 13th and 14thReports by examining current funding mechanismsfor GME and assessing their implications for ex-panding training in community settings. It evalu-ates alternative Federal financing policies withinthe context of the Council’s recommendation for astable financing mechanism that would provideadequate support for ambulatory training. Consid-eration is also given to the implications of the al-ternative policies on other COGME goals. Thesegoals include:

• Reduce the rate of growth in the supply of U.S.physicians;

• Increase the number of generalists;

• Increase the diversity of the physicianworkforce;

• Promote a rational system of physicianworkforce planning; and,

• Preserve safety net providers.

METHODOLOGYThe report’s findings and recommendations

were developed through analysis and synthesis ofinformation derived from a variety of sources: re-view of the literature and recent GME reform pro-posals, primary and secondary data analyses, con-sultation with an expert panel and interviews. Theliterature review provided background informationon current GME financing policies and issues. Par-ticular attention was given to issues related to train-ing in community-based settings. The review ofrecent GME proposals provided an inventory of fi-nancing policies that have been advanced by oth-ers and provided the foundation for developing al-ternative payment models.

The data analyses were directed at several ques-tions, including the amount and type of ambula-tory training that is occurring, the financial statusof teaching hospitals, and variations in the level of


GME costs and Medicare support across teachinghospitals. The analyses drew from secondary datasources as well as Medicare cost reports and otherdata files. The results have relevance to recommen-dations regarding the appropriate level of publicsupport for GME and allocation of the funds.

On-site and telephone interviews and input froman expert panel were used to determine the basicmodels used for clinical training experiences inambulatory settings. The interviews focused on theflow of funds to support GME and the financial ar-rangements between program sponsors, hospitals andcommunity-based training sites. Attention was givento identifying barriers to community-based train-ing and the characteristics of programs that havesubstantial community-based training. The inter-views were with individuals involved in both programswith long-standing community-based training sitesand relatively new programs that are developing col-laborative arrangements with managed care organi-zations or community providers in medicallyunderserved areas. Where possible, the residencyprogram director and individuals at the sponsoringhospital and the community sites were interviewed.The report’s findings reflect the circumstances ofthe programs where the interviews were conducted.However, similar themes emerged from the inter-views with different sites that are also consistentwith other information on GME financing issues.The analyses of alternate financing mechanisms area synthesis of the information obtained during theinterviews, from the data analyses and literaturereview, and consultation with the expert panel.

ORGANIZATION OF REPORT“The Changing Health Care Environment” sec-

tion, beginning on page 19, provides an overviewof changes in the health care environment that havemajor implications for how GME programs areoperated and financed. Concerns over the solvencyof the Medicare trust fund, the impact of continuedMedicare spending cuts in the face of a competi-tive health care market, and the weak linkage be-tween Medicare GME funding and physicianworkforce goals have led COGME and others tosupport alternative funding mechanisms for GME.

The “Current Financing of Graduate MedicalEducation” section, beginning on page 23, reviewscurrent GME funding sources. It begins with a de-tailed analysis of Medicare payments and policieswith particular attention to policies that may affectresidency training in community-settings. It thensummarizes other major sources of GME funding,including Medicaid and other State programs andprograms funded under the Public Health ServiceAct.

The “Residency Training in Community Set-tings” section, beginning on page 39, contains back-ground information on residency training in com-munity-based settings. Drawing on informationgathered from the site visits and the literature re-view, it begins with information on the amount andtype of ambulatory training that is currently takingplace and the requirements of the Residency Re-view Committees that pertain to ambulatory train-ing. A literature review on the barriers to trainingin community-based training sites follows. The sec-tion concludes with the findings from the site vis-its. The visits highlighted the wide variety of cur-rent financial arrangements between teachinghospitals and community-based training sites andthe impact of Medicare payment provisions on resi-dency training in ambulatory settings.

The “GME Reform Proposals” section, begin-ning on page 49, summarizes recent proposals forGME reform, including proposals for an all-payerfund.

The “Alternative Models for GME Funding”section, beginning on page 53, follows with ananalysis of alternative models for allocating GMEfunds. The discussion addresses the fundamentalissues: who should receive payments directly fromthe fund; how should funds be allocated amongthe receiving entities; and, how to establish ac-countability among the recipients for the funds.

The “Recommendations for GME FinancingReform” section, beginning on page 59, containsrecommendations for GME financing reform withinthe context of a stable, broad-based financingmechanism. The recommendations address the leveland sources of funding and issues related to alloca-tion of the funds.


Recent trends in the health care delivery sys-tem have major implications for how GME programs are operated and how they are fi-

nanced. These trends affect the settings in whichpatient care is delivered and in which residencytraining should occur. They also have import forthe continued reliance on patient care revenues asthe main source of funding for GME. This sectionprovides an overview of the uncertainties and fi-nancial pressures inherent in the changing healthcare environment that suggest stable broad-basedfunding through an all-payer fund is needed to sup-port GME.

The most significant changes in the health caredelivery system are associated with the growth ofmanaged systems of care and increased competi-tion within health care markets. Managed caregrowth has been rapid. In 1990, 71 percent of indi-viduals covered under employer group health planswere enrolled in a conventional fee-for-service plan.By 1998, only 14 percent were enrolled in conven-tional fee-for-service and 86 percent were in someform of managed care: traditional health mainte-nance organizations (30 percent), HMO point ofservice plans (22 percent), and preferred providerorganizations (34 percent) (U.S. Congress, 1999).Over the same period, Medicare enrollment in man-aged care plans grew from 3.3 to 15.4 percent (U.S.Congress, 1999) while 54 percent of the Medicaidpopulation was enrolled in managed care by 1998(HCFA, 1998). Since 1998, Medicare managed careenrollment has declined (HCFA, 2000).

Managed care activities such as rate negotia-tion, selective contracting, and utilization manage-ment affect patient care revenues generated by aca-demic health centers and other providers involvedin GME. The impact on specific institutions is af-fected by the managed care penetration rate in thehealth care market, the presence of other teachinghospitals, and the nature of the contractual arrange-ments with the health plan. Currently, health planscommonly offer multiple products involving diversecontracting arrangements with a broad providernetwork. HMO contracts are commonly with hos-pital systems and physician networks that manageutilization and may assume financial risk. HMOpoint-of-service products expand consumer choiceby allowing coverage of non-network services. Pre-ferred provider organization products typically in-volve contracts with individual physicians on a dis-counted fee-for-service basis with limited

utilization management (Lake, 2000; Robinson,1999).

Faculty practice plan revenues that tradition-ally support GME have been affected by the Medi-care physician fee schedule as well as competitivepressures associated with the move to managed care.The 1998 Medical Group Management Associationand Association of American Medical Colleges’Faculty Salary Survey found overall inflation-ad-justed growth in clinical faculty salaries over thepreceding five years was essentially flat. Consis-tent with the impact of the Medicare fee schedule,there was greater downward pressure on facultysalaries in the procedurally oriented departmentsrelative to faculty in departments with a high con-centration of generalists (AAMC, 1999a). At thesame time, medical school departments have be-come increasingly dependent on faculty practiceplans. The survey indicated patient care revenuesprovided 60 percent of all medical school depart-mental revenue in 1997 (AAMC, 1999b).

Reductions in hospital inpatient utilization anda shift of services from inpatient to ambulatory set-tings have accompanied the growth of managedsystems of care. Between 1990 and 1998, hospitalinpatient days declined 15 percent while outpatientvisits increased 50 percent (U.S. Congress, 1999).Efforts to increase operating efficiencies and im-prove negotiating power with managed care planshas led to hospital consolidations into regional andnational systems (Levit et al., 2000) and expansionof hospital-owned or affiliated physician practices(Retchin, 2000). The competitive pressures are evi-denced in a decline in the average hospital payment-to-cost ratio for services covered by private payersfrom 1.27 in 1990 to 1.14 in 1998 (MedPAC,2000b). In many areas, health care markets arecontinuing to evolve change as unsuccessful hos-pital mergers are dissolved and unprofitable hospi-tal-owned physician practices are sold, frequentlyat a loss.

Teaching hospitals tend to have higher costs thatput them at a competitive disadvantage with com-munity hospitals in competing for managed carecontracts for non-specialized care.2 Private payerssubsidize the educational and other missions ofteaching hospitals through higher payments. How-ever, competitive pressures are eroding the subsi-dies for public missions. The private payer payment-to-cost ratio for teaching hospitals was 1.25 in 1989(ProPAC, 1992). By 1998, it had fallen to 1.15 and

The Changing Health Care Environment


❖❖❖❖❖❖❖❖❖❖

FIGURE 2Hospital Medicare Inpatient Margin (Excluding Graduate

Medical Education) by Hospital Group, 1994-1998

35%30%25%20%15%10%

5%0%

Perc

ent M

argi

n

❖❖❖❖❖❖❖❖❖❖

❖❖❖❖❖

Source: MEDPAC, 2000a.

Year1994 1995 1996 1997 1998

All Hospitals

Academic MedicalCenter

Major Teaching(non-AMC)

Other Teaching

Nonteaching❖❖❖❖❖

percent, respectively, relative to an average 4.6 per-cent margin for non-teaching hospitals. However,Medicare margins are higher for teaching hospi-tals than non-teaching hospitals (see Figure 2). Thehigher Medicare margins are largely attributable toMedicare’s support for the public missions of teach-ing hospitals through GME and DSH payments. In1998, academic medical centers and other majorteaching hospitals had Medicare inpatients marginsof 24.6 and 26.2 percent, respectively, relative toaverage Medicare inpatient margins of 13.8 percentfor other teaching hospitals and 9.3 percent for non-teaching hospitals. The issue of whether Medicare’ssupport should continue at its current levels wasdebated during consideration of the BBA. The de-bate centered on concerns over the solvency of theMedicare Part A trust fund and the impact of con-tinued Medicare spending cuts in the face of a com-petitive health care market.

Medicaid and Medicare payment re-ductions in the BBA (Public Law 105-33)have added to the financial pressures onhealth care providers. Hospitals serving adisproportionate share of low-income(DSH) patients are most affected by theMedicaid reductions. Medicaid paymentsto these hospitals were reduced $10.4 bil-lion over five years. Medicare provisionsin the BBA that affect all hospitals wereestimated to reduce payments by $33.6 bil-lion over five years. In addition, Medicarepayments to hospitals for indirect medicaleducation costs and for serving a dispro-portionate share of low-income patientswere reduced over five years by $5.6 bil-lion and $0.5 billion, respectively. The leg-islation also carved out an estimated $4billion in implicit GME payments to Medi-care + Choice organizations that will bepaid directly to teaching hospitals for GME(CBO, 1997). Other BBA provisions, suchas implementation of prospective paymentsystems for hospital outpatient services,home health services, and skilled nursingfacility services, will also affect hospitalrevenues.4 The Balanced Budget Refine-ment Act of 1999 (BBRA) restored an esti-mated $700 million of the teaching and dis-proportionate share cuts and added $100million in direct GME payments (CBO,1999), a relatively modest amount whencompared to the total BBA cuts.

The BBA contained provisions in-tended to align Medicare payments forGME more closely with workforce policy.These provisions included:

1.05, respectively, for academic health centers andother major teaching hospitals (MedPAC, 2000b).In addition, Medicaid managed care growth hasreduced the Medicaid revenues and payments forserving a disproportionate share of low incomepatients (DSH) that public and other safety netteaching hospitals rely on to support their charitycare missions. For safety net hospitals, Medicaidrevenue losses have been accompanied by an in-crease in uncompensated care (IOM, 2000). In1998, uncompensated care represented losses of 7.8percent and 5.4 percent of the total costs of aca-demic health centers and other major teaching hos-pitals, respectively (MedPAC, 2000b).

Hospital margins are a common measure of fi-nancial performance.3 Teaching hospitals tend tohave lower total margins than non-teaching hospi-tals (see Figure 1). In 1998, major and minor teach-ing hospitals had average margins of 2.3 and 4.1

❖❖❖❖❖❖❖❖❖❖

FIGURE 1Hospital Total Margin by Teaching Status

1994-1998

8%7%6%5%4%3%2%1%0%

Perc

ent M

argi

n

❖❖❖❖❖❖❖❖❖❖

❖❖❖❖❖

Source: MEDPAC, 2000a.

Year1994 1995 1996 1997 1998

All Hospitals

Major Teaching

Other Teaching

Nonteaching❖❖❖❖❖


• A hospital-specific cap on the number of resi-dents that will be recognized for Medicare pay-ment and use of a rolling average to determineresident counts;

• Modification to the indirect medical educationadjustment paid to hospitals when the hospi-tal bears substantially all training costs to rec-ognize resident time in non-hospital settings;

• Authority to pay certain non-hospital entities(Federally qualified health centers, rural healthclinics, and Medicare + Choice organizations)for direct costs of GME if they incur substan-tially all training costs;

• Transition payments to hospitals electing toparticipate in a voluntary resident reductionplan; and,

• Authority for GME consortia demonstration.

In addition, the BBRA included a provision toreduce the variation in the per resident amounts used

to determine Medicare payments for direct GMEcosts. Even with these changes, however, the link-age between Medicare payments for patient careservices and workforce priorities remains weak.

The uncertainties of continued reliance onMedicare, Medicaid and private pay revenues tofund GME have led COGME and others to con-clude major changes are needed in the way GME isfinanced. The Council’s 14th Report (1999b) reit-erated its long standing recommendation for the de-velopment of an all-payer financing system thatwould spread the costs of preparing a well-quali-fied physician workforce equitably across all pay-ers. An all-payer fund would also afford an oppor-tunity to strengthen the relationship between GMEfunding and physician workforce priorities. Ideally,it would provide stable, broad-based funding tosupport the training of an appropriate number andspecialty mix of physicians who are well-equippedto provide high quality, effective and efficient care.


OVERVIEWThere are both direct and indirect costs associ-

ated with operating residency training programs.Direct costs include resident salaries and fringebenefits, teaching physician compensation, programadministration costs, and allocated institutionaloverhead costs. These costs are typically incurredby program sponsors and/or participating trainingsites. Indirect costs are the higher patient care costsincurred by clinical teaching sites that are attribut-able to their educational activities. Total costs forGME are difficult to estimate because of the jointproduction of patient care, education and research.A Lewin Group study estimated the higher costs ofeducation-related activities at teaching hospitals tobe $18.1 billion in 1997 (Commonwealth, 1997).The estimate does not include costs assumed bymedical schools, faculty practice plans, and com-munity-based sites or the imputed value of teach-ing by volunteer faculty.

Most GME financing occurs through patientcare revenues. Medicare and, in most States, Med-icaid make explicit payments to teaching hospitalsfor the costs of GME. Other payers implicitly sup-port GME through higher payments for patient care.In addition, faculty practice plan revenues are an-other source of support for clinical faculty timespent in academic activities. Plan revenues may also

Current Financing of Graduate Medical Education

provide direct support to medical school or depart-ment funds that are used to support graduate as wellas undergraduate medical education. State-supportfor GME typically occurs through appropriationsto State-operated medical schools or residencytraining grants. In addition, Federal appropriationsunder Title VII and VIII of the Public Health Serv-ice Act support primary care residency programsand other health professional education. Accord-ing to AAMC estimates, the major funding sourcesin FY1995 for direct GME costs incurred by non-Federal hospitals were Medicare (34%), Medicaidand private payers (57%), and other sources suchas State and local government appropriations, phi-lanthropy, and revenues from faculty practice plans(9%) (Fishman, 1996). The flow of funds amongthe participants in GME activities is complex andfrequently involves cross-subsidies between medi-cal schools, teaching hospitals, and other trainingsites.

MEDICARE PAYMENTSThe Medicare program recognizes the higher

costs associated with teaching activities throughexplicit payments for GME. In FY2000, Medicarewill pay an estimated $2.7 billion (including $200million for managed care enrollees) in direct GMEpayments and $5.1 billion (including about $700million for managed care enrollees) for IME(HCFA/OACT, 2000). Medicare’s payments forearlier years’ GME costs are shown in Table 1. To-tal payments increased 48 percent between 1990and 1998 (U.S. Congress, 1999).

Medicare has supported GME as a public goodsince its inception in 1966. Although the paymentmethodologies have changed over the years,Medicare’s support has always been linked to theprovision of patient care services to Medicare ben-eficiaries. The committee report accompanying theoriginal Medicare law indicated that the costs ofmedical education should be borne by the commu-nity. However, until the community assumed thecosts, the report specified Medicare would share inthe costs because educational activities contributeto quality of care within the institution. The reportfurther noted that the intent was to support activi-ties traditionally carried on by providers and thatMedicare would not share in increased costs result-ing from a redistribution of costs from educationalinstitutions to providers (U.S. Congress, 1965).

TABLE 1Medicare GME Payments

to Hospitals($ BILLIONS)

IME PAYMENTS DIRECT GMEPAYMENTS

1990 .................................. 2.91 1.761991 .................................. 3.21 1.891992 .................................. 3.67 2.361993 .................................. 4.09 2.551994 .................................. 4.50 2.611995 .................................. 5.10 2.741996 .................................. 5.55 2.861997 .................................. 5.16 2.431998 .................................. 4.99 2.10

Source: U.S. Congress (1999).Payments include GME amounts implicit

in managed care capitation rates.


FIGURE 3Comparison of FY1997 Per

Resident Amounts and GME Costs

$140,000

$120,000

$100,000

$80,000

$60,000

$40,000

$20,000

$0Low

Source: FY1997 HCRIS cost reports.

Hospital Quartiles by Per Resident Amount

36,738

65,622

Med-Low Med-High High Average

54,236

82,445

70,839

99,466104,687

123,857

71,468

97,066

PRA GME Costs

The subsections that follow provide a generaloverview of current Medicare payments associatedwith GME. Specific issues are noted for each fund-ing stream that have implications for reformingFederal GME financing policies. There are alsothree overarching concerns that suggest broader-based funding is preferable to relying heavily onMedicare payments for patient care services as theonly Federal funding mechanism. They will be dis-cussed more fully in the context of an evaluation ofpotential changes in GME financing mechanisms.These issues are:

• Restricting payments to teaching hospitals foreducational costs impedes the development ofresidency programs in non-hospital ambula-tory and managed care settings.

• Linking educational payments to services fur-nished to Medicare patients concentrates Fed-eral support on providers with high Medicareutilization. It offers little support to providerswith low Medicare utilization such aschildren’s hospitals and Federally qualifiedhealth centers (FQHCs).

• Paying for educational costs through patientcare payments alone is not an effective mecha-nism for achieving specific workforce priori-ties such as improving the specialty and geo-graphic distribution of the physicianworkforce.

Recent changes in national policy have begunto address these concerns, but COGME believesfurther changes are in order.

DIRECT GME PAYMENTS TO HOSPITALS

Medicare’s direct GME payments are intendedto cover the program’s share of expenses that aredirectly attributable to GME: the resident salaries,teaching physician compensation for supervisingresidents, and associated overhead costs. Paymentis based on the number of residents at the hospital(and ambulatory settings if the hospital assumessubstantially all of the training costs), a hospital-specific per resident amount based on 1984 costsupdated for inflation, and Medicare’s share of hos-pital inpatient days. Until the BBA, the Medicarelaw authorized direct GME payments only to hos-pitals. Beginning January 1, 1998, Medicare maymake direct GME payments to other provider enti-ties. Policy issues concerning Medicare’s directGME payments to hospitals are discussed below.Medicare GME financing issues related to commu-nity-based training are treated later in this sectionas a separate discussion.

PAYMENTS REFLECT HISTORIC COSTS

Medicare’s total payments for direct GME costsfor 1990-1998 are shown in Table 1. Payments area function of the hospital’s per resident amount, thenumber of residents, and the ratio of Medicare in-patient days to total inpatient days.5 A hospital’sper resident amount is based on its 1984 GME costs.It has been adjusted for inflation annually based onthe rate of increase in the Consumer Price Indexfor Urban Consumers (CPI-U).6 On average, Medi-care pays about 30 percent of the per resident

amount. Actual increases inMedicare direct GME pay-ments exceeded the increasein the CPI-U until 1997. Pay-ments started to decline in1997 with reductions in resi-dent counts and increases inMedicare managed care en-rollments.

Direct GME costs perresident have increased morerapidly than the per residentamounts. Figure 3 comparesthe per resident amounts ap-plicable to cost reportingperiods beginning in Federalfiscal year (FY) 1997 to di-rect GME costs per residentincurred by hospitals for thesame period. Hospitals havebeen divided into quartilesbased on their per residentamounts. Direct GME costs


FIGURE 4Components of FY1997 GME Cost Per Resident


GME Cost Quartiles

$180,000

$160,000

$140,000

$120,000

$100,000

$80,000

$60,000

$40,000

$20,000

$0Low

11,786

14,012

26,065

Med Low

21,764

21,521

39,348

Med High

28,876

32,083

46,830

High

47,921

55,566

59,580

Average

26,087

28,942

41,724

Overhead

Other Costs

Res. Salaries

per resident were 36 percent higher than the aver-age per resident amount. The rate of increase forhospitals with low per resident amounts has beenhigher relative to hospitals with high per residentamounts. As a result, hospitals with low per resi-dent amounts received 56 percent of Medicare’sshare of their FY1997 GME costs while hospitalswith a high per resident amount received 85 per-cent of Medicare’s share.

THERE IS WIDE VARIATION IN PER RESIDENT

AMOUNTS

The hospital-specific per resident amounts varywidely based on historical accounting practices andfinancial arrangements between program sponsorsand teaching sites in 1984. The resident-weightedmean of the high per resident amount group is 2.8times the resident-weighted mean of the low perresident amount group. Even within the same geo-graphic area, the variation can be substantial. Ear-lier studies have found that the per resident amountsdo not vary systematically by factors such as costof living, mix of program specialties, or quality(Anderson, 1996).

Analysis of cost reports beginning in FY1997indicates the differences in resident salary levelsaccount for a relatively small amount of the varia-tion in direct GME costs. Figure 4 compares thecomponents of direct GME costs by per residentcost groups.7 The analysis includes 787 hospitalsthat reported having ten or more residents. Most

cost variation is attributable to differences in theother direct program costs (including teaching phy-sician compensation) and in allocated overheadcosts. The components for other direct costs andallocated overhead costs per resident in the highcost group are about four times those costs in thelow cost group.

HCFA recently began to collect compensationdata on teaching physicians in connection with re-finements to the hospital wage index. These dataafford a more direct measure of differences in teach-ing physician compensation. The reported amountsshould reflect wage-related payments to teachingphysicians who are either employed or under con-tract to provide teaching or administrative servicesrelated to the hospital’s GME programs. The analy-sis used data for teaching hospitals with ten or moreresidents and at least $10,000 in GME costs. Notall teaching hospitals reported their teaching sala-ries. After removing statistical outliers, 352 hospi-tals reporting teaching compensation remained inthe analysis. Table 2 compares teaching physiciancompensation per resident to the average GME costper resident for those hospitals. The average teach-ing physician compensation per resident is $24,695.The average for hospitals in the low per residentGME cost quartile is $11,459 while the average forhospitals in the high cost quartile is $42,983. Thedifference is largely attributable to the number ofcompensated teaching physician hours per resident:178 in the low cost group compared to 614 in thehigh cost quartile. Standardizing for differences in

the cost of living across geographicareas does not significantly changethe results.

One might expect teaching phy-sician compensation per resident tobe higher in hospitals that sponsorrelatively few residents since admin-istrative costs associated with oper-ating residency programs are spreadover fewer residents. Table 2 com-pares teaching physician compensa-tion amounts across hospitalsgrouped by total number of residents.The per resident average teachingphysician compensation and hoursdecline as the size of the residencyprogram increases. However, they arefairly comparable across hospitalswith fewer than 200 residents.8

The question of whether primarycare residency programs have higherteaching physician costs is closelyrelated to the issue of whether thereshould be specialty-differentials in


Total .............. 352 34,133 $24,695 364 $68

Low ................. 58 7,425 $11,459 178 $64

Med Low ......... 92 10,814 $19,524 284 $69

Med High ...... 102 9,093 $27,972 423 $66

High .............. 100 6,801 $42,983 614 $70

<=50 ........... 173 4,464 $31,013 488 $64

<=100 ............. 65 4,582 $30,210 435 $69

<=200 ............. 62 8,593 $28,390 409 $69

<=300 ............. 31 7,589 $23,057 339 $68

<=400 ............. 11 3,640 $18,373 288 $64

>400 ............. 10 5,265 $15,238 210 $73

100% .......... 73 1,942 $31,832 534 $60

76-99% .......... 76 3,547 $32,108 434 $74

51-75% ........ 102 10,564 $29,243 416 $70

26-50% .......... 77 16,900 $19,124 292 $65

0-25% .......... 24 1,180 $29,729 425 $70

Source: FY2001 Hospital Wage Index Survey (HCFA website). FY 1997 HCRIS cost reports usedto establish groupings. Based on 352 hospitals, with at least 10 residents and $10,000 in GME

aggregate approved amounts, reporting teaching physician compensation.

TABLE 2Hospital Per Resident Teaching Physician Costs, FY1997

Number ofHospitals

TotalNumber ofResidents

AverageTeachingPhysician

Compensation

AverageTeachingPhysician

Hours

AverageHourly Rate

HOSPITAL GROUPING

GME COSTQUARTILE

NUMBER OFRESIDENTS PER

HOSPITAL

PROPORTION OFRESIDENTS IN

PRIMARY CARE

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

○

the per resident amounts. Primary care specialtiesare more dependent on hospital support becausetheir revenues for physician services typically pro-vide less faculty support relative to those gener-ated by non-primary care specialties (IOM, 1989).Also, primary care residency programs are oftenthe only programs offered by minor teaching hos-pitals. As a result, overhead is spread over fewerresidents. This suggests that teaching compensa-tion per resident might be higher in primary careresidency programs relative to other programs.Table 2 compares the teaching physician compen-sation costs per resident in hospitals with only pri-mary care/OB-GYN residency programs to thosecosts in hospitals that train residents in other resi-dency programs. Hospitals offering other programsare grouped based on the proportion of residentsthat are in residency programs other than primarycare/OB-GYN. There is no consistent pattern acrosshospitals.

The BBRA made a modest change in the Medi-care payment methodology to reduce disparities in

per resident amounts. The provision raises the mini-mum payment to 70 percent of a national wage-adjusted average per resident amount. The per resi-dent amounts for hospitals that are above 140percent of the wage-adjusted national average arefrozen for FY2001-FY2002 and limited to infla-tion increases of CPI-U minus 2 percentage pointsfor FY2003 through FY2005. HCFA estimates 27.3percent of teaching hospitals will benefit from the70 percent floor and receive $33.2 million in addi-tional payments during FY2001. The agency esti-mates 14.6 percent will be affected by the ceilingand will receive $16 million less in GME payments(DHHS, 2000b).9

PAYMENTS DEPEND ON MEDICARE PATIENT

LOAD

Medicare’s share of a hospital’s aggregate pay-ment amount (the product of the per residentamounts and number of FTE residents) is based onits ratio of Medicare inpatient days to total days.On average, Medicare’s share is approximately 30

percent of the aggregatepayment amount. There areno substantial differences inthe average Medicare shareacross the payment groups.However, individual hospi-tals can have significantlydifferent Medicare shares. Inparticular, hospitals withlow Medicare utilizationlevels, such as children’shospitals and certain hospi-tals with high charity carecaseloads receive littleMedicare support. The dis-tribution of Medicare perresident payments (Medi-care’s share of the per resi-dent amount) is shown inFigure 5. The average resi-dent-weighted Medicarepayment was $22,350 inFY1997.

PAYMENTS DO NOT

REFLECT SPECIALTY

DIFFERENCES INRESIDENCY PROGRAM

COSTS

There are two hospital-specific per residentamounts under Medicare.One amount applies to resi-dency programs in primary


on the proportion of their residents that are in pri-mary care and OB-GYN residency programs. Hos-pitals with a higher percentage of their residents inprimary care and OB-GYN training programs ap-pear to have higher per resident amounts and coststhan hospitals with a relatively smaller proportionof primary care and OB-GYN residents. There is amarked difference between hospitals with more than50 percent primary care and OB-GYN residents andthose with less than 50 percent primary care andOB-GYN residents.11

In determining appropriate levels of Federalsupport, specialty differences in direct GME costsare less important than the impact of teaching onthe training sites net costs (total costs minus pa-tient care revenues). Faculty supervising residentsin subspecialty programs tend to have higher clini-cal revenues although the differences are lesseningas a result of the Medicare physician fee schedule.The Medicare rules count residents who are beyondtheir initial residency period as 0.5 FTE for pur-poses of determining direct GME payments. Theinitial residency period is defined as the minimumnumber of years needed for initial board certifica-tion. It is based on the first residency program aresident enters and is not modified or extended if aresident switches to another residency program.Approximately 12 percent of first-year residencyprograms are filled by residents with prior U.S. resi-dency training (COGME, 1999b). Further researchis needed to understand whether the 0.5 FTEweighting reasonably approximates specialty

care (family medicine, general internal medicine,general pediatrics, osteopathic general practice),preventive medicine, geriatric medicine and obstet-rics-gynecology. The other per resident amountapplies to all other residency programs. The aver-age per resident amount for primary care and OB/GYN programs is 5.6 percent higher than the perresident amounts for other programs at a particularteaching hospital. The difference stems from a two-year freeze imposed by the Omnibus Budget Rec-onciliation Act of 1993 on the inflation updates inper resident amounts for all programs other thanprimary care and OB-GYN. The difference does notreflect actual variations in direct GME costs be-tween primary care and other residency programs.10

In teaching hospitals with multiple residencyprograms, primary care residency programs aresometimes subsidized by other residency programsthat generate relatively higher clinical revenues.However, based on the FY1997 cost reports, 393hospitals with 6,124 residents only participate inprimary care residency training and have no cross-subsidization potential. This could become an is-sue if hospital-specific per resident amounts are re-placed by a national average per resident amount.If GME costs are higher in programs with substan-tial community-based training needs and little op-portunity for cross-subsidization, adequate supportfor these programs may require per resident pay-ment amounts by specialty or groups of special-ties. Figure 6 compares the per resident amountsand the FY1997 direct GME costs in hospitals based

FIGURE 5Distribution of FY1997 Medicare Per Resident Payments


Num

ber o

f Hos

pita

ls

40

35

30

25

20

15

10

5

0

Medicare Payment Per Resident ($000)

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81+


differences in net program costs. The BBRA re-quires the Medicare Payment Advisory Commis-sion to make recommendations in its May 2001report regarding the appropriateness of using theinitial residency period distinction in determiningdirect GME payment.

INDIRECT MEDICAL EDUCATIONPAYMENTS TO HOSPITALS

Medicare pays acute care hospitals for inpatientservices based on a prospectively determined ratethat takes into account average resources requiredto treat Medicare beneficiaries in the same diagno-sis-related group. An indirect teaching adjustmentis added to the standard rate to pay for additionalpatient care costs attributable to teaching activitythat are not captured as direct GME costs. Althoughbased on the hospital’s ratio of residents-to-beds,the adjustment also compensates the hospital forhigher patient care costs typically associated withother activities provided in conjunction with GME,i.e., clinical research, specialized care for complexpatients, and charity care. Payments exceed theempirically justified level based on teaching inten-sity alone.

Medicare costs per case have grown moreslowly than increases in IME payments. Changesin cost per case reflect changes in practice patternssuch as a decline in the average length of stay inaddition to price inflation and case mix changes(which also affect IME payments per case). Thedifferential growth rates are largely accountable for

the high average Medicaremargins for teaching hos-pitals shown in Figure 2.

The remainder of thissubsection explores sev-eral policy issues involv-ing IME payments thathave import for GME fi-nancing reform.

INTENDED PURPOSE OF

THE IME A DJUSTMENT

The IME adjustmentwas established at the out-set of the Medicare pro-spective payment systemto account for the higherinpatient costs per dis-charge incurred by teach-ing hospitals. The commit-tee reports accompanyingthe legislation noted that

the adjustment was being provided “as a proxy toaccount for a number of factors which may legiti-mately increase costs in teaching hospitals.” Theconferees expressed concern about the ability ofthe prospective payment system to account fullyfor factors such as severity of illness and additionalcosts associated with teaching residents (U.S. Con-gress, 1983). Fearing that teaching hospitals mightbe adversely affected, the initial IME adjustmentwas double its empirically derived amount.

Medicare IME payments were intended to com-pensate teaching hospitals for their higher costs.While the formula is based on the ratio of residents-to-beds, the payments are an important source ofsupport for other public missions such as charitycare and research. It is on these grounds that reduc-tions in the IME adjustment to the empirically jus-tified amount are generally opposed. However, thelevel of involvement in the other public missionsvaries across teaching hospitals and teaching in-tensity is not a good measure to use to support otherpublic missions.

Teaching hospitals as a group furnish a dispro-portionate share of charity care. However, not allteaching hospitals furnish substantial care to lowincome patients nor are all hospitals that serve low-income patients teaching hospitals. Table 3 showsthe distribution of hospitals and residents by teach-ing intensity and percentage of low-income pa-tients.12 Among hospitals with a low-income pa-tient percentage more than 35 percent, 229 areteaching hospitals and 661 are non-teaching. Thisrepresents 26 percent of all teaching hospitals and

FIGURE 6GME Costs and Per Resident Amounts by

Proportion of Primary Care Residents, 1997

$140,000

$120,000

$100,000

$80,000

$60,000

$40,000

$20,000

$0


Percentage of Residents in Primary Care and OB-GYN Specialties

PRA GME Costs

100%

83,075

116,576

75-99%

81,010

112,215

51-75%

79,424

109,909

26-50%

67,686

84,516

0-25%

58,380

79,048

Average

73,486

97,235


0 - 15 16 - 25 26 - 35 > 35 Total

Number of Hospitals

All Teaching .................................... 382 292 160 299 1133

<.10 ............................................... 232 158 74 94 558

.11 - .25 ........................................ 103 80 38 56 277

.26 - .45 ........................................ 38 31 18 65 152

>.45 ............................................... 9 23 30 84 146

Non Teaching .................................. 1715 936 571 661 3883

ALL .................................................. 2097 1228 731 960 5016

Source: PPS Proposed FY2001 Impact File. Low-income patients defined aspercentage of Medicare patients who are entitled to SSI plus percentage of all

patients who are non-Medicare patients entitled to Medicaid.

TABLE 3Distribution of Hospitals by Teaching Intensity

and Low-Income Patients

Low Income Patient PercentRATIO OF RESIDENTS TO BEDS

17 percent of non-teaching hospitals. However, one-third of all teaching hospitals, including 337 majorteaching institutions, have a low-income patientpercentage of less than 15 percent. This suggeststhat support for hospitals serving low-income pa-tients should be addressed through targeted pay-ments for serving a disproportionate share of low-income patients (DSH) and other subsidies ratherthan the IME adjustment. At the same time, it isimportant that refinements in the IME paymentmethodology not adversely affect hospitals thatprovide significant charity care. Reductions in IMEpayments to these hospitals should be offset byhigher support for uncompensated care until spe-cific funding for such services is provided.

Similarly, not all teaching hospitals are engagedin substantial research. Even among the 125 aca-demic health centers, research funding is concen-trated on a few institutions. In 1996, 50 percent ofall NIH funds were awarded to 19 academic healthcenters (Commonwealth, 1999). Further analysisis needed to understand whether research-intensiveteaching hospitals have higher costs relative to thenon-research intensive teaching hospitals after con-trolling for teaching intensity. This analysis hasimport for GME financing reform. If there is a re-search effect on costs, the issues will be whetherthe effect should be recognized in GME fundingand whether differential payments should flow toresearch-intensive institutions. In this regard,Medicare’s policy has been that research costsabove usual patient care costs should not be reim-bursed.

EMPIRICALLY DERIVED LEVEL ISLOWER

The IME adjustment is a functionof the hospital’s teaching intensity,Medicare discharges, case mix indexand wage index. Teaching intensity ismeasured by the ratio of the hospital’sresidents-to-beds in the payment for-mula for operating costs. In FY2000,the adjustment will increase a teach-ing hospital’s payments about 6.5 per-cent for each .10 increment in its ratioof residents-to-beds. A hospital with aresident-to-bed ratio of .25 would re-ceive a 14 percent add-on to its pro-spective payment rate for operatingcosts. The per case dollar value of theadjustment will be higher for hospitalslocated in relatively high wage areasand for hospitals which treat a rela-tively high proportion of resource-in-tensive patients. The estimated averageIME adjustment per resident in

FY2000 is $48,300. The BBRA reduces the adjust-ment to 6.25 percent for each .10 increment in thehospital’s resident-to-bed ratio in FY2001, and to5.5 percent in FY2002 and thereafter.13 AdditionalIME payments are made for capital-related costsand for Medicare managed care enrollees.

IME payments in excess of the analytically sup-ported level have been counter-productive to phy-sician workforce goals. The excess represents“profit” that provides an incentive to hospitals toincrease the number of residents and, until the BBA,rotate residents primarily to hospital-based clini-cal training sites. As discussed below, the BBAmoderated these incentives by establishing a capon the number of residents and allowing IME pay-ments under certain conditions when residents ro-tate to non-hospital settings.

Estimates of the teaching effect on Medicareinpatient operating costs vary and are influencedby the way the regression formula is specified.14

The Prospective Payment Assessment Commissionestimated the teaching adjustment at 4.1 percent in1997 (ProPAC, 1997). The Congressional BudgetOffice (2000) estimates the Medicare programwould save $5.8 billion over five years if the ad-justment were reduced to this level. More recentMedPAC (2000b) analyses estimate the adjustment at3.1 percent after other refinements are made to thepayment system. The 3.1 percent adjustment wouldresult in a $1.5 billion reduction in annual IME pay-ments compared to payments based on the 5.5 per-cent adjustment provided in the BBRA for FY2002.


The difference between current IME payment levelsand the analytically justified levels represents amountsthat could be redirected. One potential use would beto support educational activities that would increasecommunity-based training capacity. Another wouldbe to improve charity care support by making tar-geted payments to those institutions that actuallyprovide substantial amounts of charity care.

AMBULATORY RESIDENTS ARE INCLUDED INTHE IME C OUNT

The initial regulations implementing the pro-spective payment system included only residentsassigned to the inpatient portion of the hospital inthe IME payment formula. The formula was subse-quently revised to include all residents working ineither the inpatient or outpatient portions of theacute care hospital. More recently, the BBA per-mitted residents working outside the hospital to beincluded in the count provided the hospital incurssubstantially all costs of the residency training pro-gram. (Implementation of this provision is dis-cussed later in this subsection in the context ofMedicare policies for non-hospital training sites).The more inclusive resident count protects againstdisincentives to rotate residents outside the inpa-tient area of the hospital. However, it is not as gooda technical measure of the effect of teaching on in-patient costs per case. An alternative would be torestrict the count to residents in the hospital inpa-tient area and to provide IME or incentive paymentsto train residents in outpatient hospital and com-munity-based settings.15

DIRECT PAYMENTS TO HOSPITALS FORMANAGED CARE ENROLLEES

Until the BBA, teaching hospitals lost Medi-care support for GME when Medicare beneficia-ries enrolled in a managed care plan. The capita-tion rates paid to managed care plans for Medicareenrollees were based on fee-for-service paymentsand included an implicit amount for GME; how-ever, managed care plans were not required to passthe funds on to teaching hospitals. Nationally, about3.2 percent of the average capitated paymentamount represented implicit GME payments (U.S.Congress, 1999). In areas with high managed carepenetration, the revenue loss could be significant.16

The BBA recognizes teaching hospitals are ata competitive disadvantage relative to non-teach-ing hospitals for managed care contracts and maynot be able to command the premium necessary tosupport their educational missions. The legislationprovides for a carve-out of the GME amounts im-plicit in payments to Medicare + Choice plans and

direct payments to teaching hospitals for GME costsattributable to Medicare + Choice enrollees. Theprovision is being phased-in over five years and isestimated to result in $4.0 billion in additional pay-ments to teaching hospitals over that period (CBO,1997). Payment is based on the amount of directGME and IME payments that would have been paidto the hospital if the beneficiary had been enrolledin traditional Medicare fee-for-service.17 Thus, pay-ment is linked to Medicare managed care dischargesat the teaching hospital. If the teaching hospital’sshare of Medicare managed care enrollees is lowrelative to its share of fee-for-service enrollees, to-tal Medicare support will still be less than what thehospital would have received if the beneficiarieshad remained in fee-for-service. A higher propor-tion of inpatient care is provided to managed careenrollees at non-teaching hospitals and on an out-patient basis. One implication is that more fundsmay be carved out of the capitated payments thanare paid to teaching hospitals on behalf of Medi-care + Choice enrollees.

COGME’s 14th Report (1999b) recommendedthat a portion of the GME carve-out should be dis-tributed based on performance consistent with na-tional physician workforce goals. The amount couldbe based on the difference between the carve-outfrom the Medicare + Choice rates and the amountspaid to teaching hospitals.18 The encounter databeing collected from managed care plans providesan opportunity to evaluate the relative proportionsof managed care and fee-for-service beneficiariesreceiving inpatient care at teaching hospitals.

The BBA did not provide a similar pass-thoughfor payments made to hospitals for serving a dis-proportionate share of low-income patients. About2.1 percent of the capitated payment to Medicare +Choice organizations in FY1995 represented im-plicit DSH payments (U.S. Congress, 1999).

LIMITATIONS ON THE NUMBER OFRESIDENTS RECOGNIZED BY MEDICARE

The BBA made several changes in how resi-dents are counted for Medicare payment purposes.The count is now based on a three-year rolling av-erage. The number of allopathic and osteopathicresidents that can be counted is limited to the num-ber reported in the hospital’s cost reporting periodending in 1996. (The limit does not apply to dentalor podiatric residents). The changes were intendedto temper growth in the number of residents by lim-iting the number Medicare will pay for. For pro-grams that are reducing their size, the rolling-aver-age provides transition payments to supportresidency replacement costs.


Medicare’s limit on resident counts is consis-tent with COGME’s goal to reduce the number ofresidents. However, hospital-specific limits impedethe ability of program directors to shift residentsfor educational reasons. The receiving hospital maynot be able to absorb the residents under its count.The rules allow the limit to be applied on an aggre-gate basis to affiliated hospitals. While this providessome flexibility, it relies on cooperation among theparties and adds considerable administrative com-plexity to the rules. The limit is particularly prob-lematic in population growth areas where expand-ing residency programs would improve thegeographic distribution of physicians. The BBAallows an exception for rural hospitals that estab-lish new residency training programs. However, itdoes not allow expansion of existing rural programsor an exception for rural tracks sponsored by urbanhospitals. The BBRA made some revisions in thelaw to allow a 30 percent expansion in rural resi-dency programs and to recognize that urban hospi-tals may operate a rural residency track. The provi-sions are an improvement over the BBA limits.However, if the rural hospital’s resident count isrelatively small, the 30 percent tolerance allows foronly minimal expansions in residency programs.For example, a rural hospital with an establishedfamily medicine program with nine residents (threeper year) would only be allowed to add one resi-dent position for each year. Rural hospitals that areaffiliated with urban teaching hospitals often haveeven smaller base year counts that would precludeany significant increase in the absolute number ofresidents.

PAYMENTS FOR RESIDENCY TRAINING INNON-HOSPITAL BASED SETTINGS

Prior to the Balanced Budget Act of 1997, thetime a resident spent in hospital outpatient depart-ments counted in the IME adjustment for inpatienthospital services. The time spent in non-hospitalsettings did not count. For direct GME payments,the time a resident spent in patient care activitiesin non-hospital settings counted if the hospital hada written agreement with the site and incurred allor substantially all of the costs of the training. Regu-latory policies assumed the hospital incurred sub-stantially all training costs if it paid the resident’ssalary and fringe benefits. The teaching physicianrules apply if the resident in the non-hospital set-ting is counted by the hospital for direct GME pay-ments. However, if the hospital does not includethe resident in its direct GME count, a resident whois fully licensed to practice medicine may bill forservices under Part B in a non-hospital setting. Inthis case, the teaching physician may not bill un-

der Part B for services furnished by the residentunder the physician’s supervision.

Medicare pays certain ambulatory care provid-ers on a reasonable cost basis. Subject to reimburse-ment limits, Medicare shares in any indirect teach-ing costs incurred by these providers. Sincepayment is based on each provider’s costs ratherthan a fixed rate, an explicit payment for indirectteaching costs is generally not needed. However,the reimbursement limits are not adjusted for teach-ing activities. As a result, full sharing in GME costsdoes not occur in situations where a provider’s costsexceed its reimbursement limits (e.g. an urban Fed-erally qualified health center with costs in excessof its year 2000 per visit limit of $96.02). Prior tothe BBA, providers paid on a reasonable cost basisincluded hospitals for most outpatient services,skilled nursing facilities, home health agencies,Federally qualified community health centers andrural health centers. The BBA established prospec-tive payment systems for hospital outpatient ser-vices, home health services, and services providedby skilled nursing facilities. These are settingswhere residency training should take place in or-der to follow patients across the full continuum ofcare. As the payment methodologies change, caremust be taken to ensure against erosion in existingsupport for residency training in these settings.Additional study is needed to determine whetherthere is an indirect teaching cost that should be rec-ognized through an adjustment to the reimburse-ment limit or prospective payment for ambulatorycare providers. The evidence to date does not sub-stantiate the need for an IME adjustment. For ex-ample, in implementing the hospital outpatient PPS,HCFA found teaching did not have a significanteffect on hospital outpatient costs (DHHS, 2000a).Other studies of residency training programs in am-bulatory settings support a widely held belief thatover the course of a three-year residency program,residents have little impact on net costs (Diamond,1993; Lave, 1989).

The BBA made several significant changes inMedicare payment for residency training in ambu-latory settings. First, the legislation authorized di-rect payments to entities other than hospitals. Medi-care-participating Federally qualified communityhealth centers, rural health clinics, Medicare +Choice organizations, and other entities designatedby the Secretary may be paid for direct GME if theprovider incurs all or substantially all of the costsof training at the site. (Implementing regulationsdid not expand the list of eligible entities.) Second,the law allows hospitals to include in their IMEresident count (as well as direct GME residentcount) the time residents spend in patient care


activities at non-hospital settings if the hospital in-curs all or substantially all of the training costs atthe ambulatory site. In determining which entityincurs substantially all of the training costs, imple-menting regulations look to which entity pays theresident’s salary and fringe benefits and teachingphysician compensation while the resident is at thenon-hospital site. The rules provide that the follow-ing criteria must be met before the hospital mayreceive payment:

• There must be a written agreement betweenthe hospital and the ambulatory site.

• The parties must agree that the hospital bearsthe costs for the resident’s salary and fringebenefits and that the teaching physician com-pensation is reasonable.

• The resident must be engaged in patient careactivities as part of an approved residency pro-gram.

If the hospital bears substantially all of the costs,any GME costs incurred by the ambulatory site arenot allowable costs for Medicare payment purposes.On the other hand, if the ambulatory site incurssubstantially all of the costs, Medicare pays its shareof the GME costs incurred by the site. Direct GMEcosts are not subject to the productivity standardsand per visit limits applicable to Federally quali-fied health centers and rural health clinics. Paymentto Medicare + Choice organizations is based on thepercentage of plan enrollees who are Medicare ben-eficiaries and is limited to the resident’s salary andreasonable teaching physician compensation.

Payment will be made to either the hospital orthe ambulatory site, but not both. In most cases, itis advantageous for payment to flow to the hospi-tal:

• The hospital usually has higher Medicare uti-lization than the ambulatory site.

• The hospital would receive both direct GMEand IME if it bears substantially all of the costs.The ambulatory site would receive direct GMEin addition to payment for direct patient carecosts or the capitated payment rate. Generally,the hospital’s payments would be substantiallymore than the ambulatory provider’s payments.

However, there is a complicated interactionbetween the BBA payments to ambulatory sites andthe limitations on the number of residents that canbe included in a hospital’s resident count. If thehospital exceeds its 1996 resident limits, it may beadvantageous for an ambulatory site entitled to di-rect GME payments (i.e., FQHC, RHC or Medi-care + Choice program) to assume the residency

training costs and be paid based on Medicare’s shareof the costs. If the hospital does not include theresidents in its count and the residents are fullylicensed, another option would be for the resi-dents to bill for their patient care services underPart B.

For IME payments, the time residents spent innon-hospital settings could not be included in ahospital’s resident count prior to the BBA. The resi-dents who were training in non-hospital settings in1996 are not represented in the hospital’s residentlimit for IME payments. With the BBA limits, theseresidents cannot be included in the hospital’s IMEcount unless there are offsetting reductions in otherresidency programs at the hospital. In essence, theBBA removes the disincentive for new shifts in resi-dents from hospital to community-based trainingsites. However, hospitals that had already developedcommunity-based training sites by 1996 are disad-vantaged. They receive no benefit from the changeunless there are reductions in other residency pro-grams at the hospital. The provisions disproportion-ately impact family medicine residency programs.An estimated 250-350 residents training in non-hospital settings in 1996 were not included in hos-pitals’ IME count. Forty percent of family medi-cine programs are the sole residency program inthe hospital and cannot benefit from downsizing inother programs at the hospital (Davis, 2000).

TEACHING PHYSICIAN SERVICES

The direct GME payment is intended to coverthe hospital’s compensation to teaching physiciansfor time spent on GME program administration andteaching and supervision of residents. Specifically,the per resident amounts include an average pay-ment for teaching physician supervision of residentsbased on 1984 GME costs. In addition, paymentmay be made under the Medicare physician feeschedule for professional services furnished by thephysician or by a resident under the medical direc-tion of the teaching physician. Medicare rules gov-ern when a teaching physician can appropriatelybill Medicare when a resident is involved in the careof his or her patient. The rules are intended to avoidpaying for the same service twice. They limit PartB payments to situations involving a teachingphysician’s direct, identifiable service to an indi-vidual patient. The teaching physician rules havebeen clarified and revised since the per residentamounts were initially determined. As a result, theremay be inconsistencies between teaching physiciancompensation amounts included in the per residentamounts and what would be considered teachingphysician activities payable through the per resi-dent amounts under current rules.


PRESENCE DURING THE KEY PORTION OF THE

VISIT

For many years, HCFA’s payment criteria forprofessional billings were based on an attendingphysician concept. The attending physician crite-ria included: reviewing the patient’s history andphysical examination and personally examining thepatient within a reasonable period of admission;confirming or revising the diagnosis; determiningthe course of treatment to be followed; ensuringthat any supervision needed by the interns and resi-dents is furnished; and making frequent reviews ofthe patient’s progress throughout the period of care(BHI, 1969).

The insurance carriers that process Medicareclaims inconsistently applied standards regardingwhether the attending physician needed to be physi-cally present when a billable service was providedand the documentation required to substantiate thebilling requirements were met (Rabb, 1997). Ef-fective July 1996, the attending physician conceptwas replaced by a requirement that a teaching phy-sician be present to perform or observe the residentperform the “key portion” of each billable service.For evaluation and management services and mi-nor procedures, the teaching physician must bepresent for all portions as if the teaching physicianhad personally performed the entire visit or proce-dure. In the case of surgery or a complex or dan-gerous procedure, the teaching physician must bepresent during all critical portions of the procedureand must be immediately available to furnish serv-ices during the entire procedure or service. For psy-chiatric services, the physical presence requirementmay be met by observation of the service by use ofa one-way mirror, video equipment, or similar de-vice (HCFA, 1997).

The rules provide a limited exception to thephysician presence rule in hospital outpatient de-partments or other ambulatory settings that receivedirect GME payments and provide coordinatedcomprehensive care not limited to a body system.Generally, the exception applies to primary careteaching clinics that provide coordinated compre-hensive care. The exception applies only to resi-dents with more than six months training in an ap-proved residency program that furnish evaluationand management services of low to mid-level com-plexity. The teaching physician must be immedi-ately available and:

• Direct the care of no more than four residentsat a time;

• Have no other responsibilities (including su-pervision of other personnel);

• Review with each resident during or immedi-ately after each visit the patient’s medical his-tory and care; and,

• Document his or her participation in the re-view and direction of the services.

The purpose of the teaching physician rules isto prevent duplicate payments by the Medicare pro-gram and to assure Part B payments are made onlywhen a teaching physician provides a personal iden-tifiable service to an individual patient. Residentsact more independently as they progress throughthe training program. The need for supervision ismore extensive early in the residency program andis less toward the end. The Medicare rules for PartB billing require the teaching physician to bepresent during the portion of the evaluation andmanagement visit that determines the level of theservice billed regardless of the program year (PGY).This creates incentives for an inefficient and inap-propriate teaching model. It does not recognize thatthe teaching physician’s needed level of involve-ment decreases as the resident’s proficiency in-creases.

DOCUMENTATION REQUIREMENTS

For evaluation and management services, ateaching physician must personally document his/her presence and participation in the key compo-nents that demonstrate the appropriate level of serv-ice (e.g., participation in history, examination, andmedical decision-making). Medicare instructionsnote that if the resident previously obtained anddocumented key elements of the service, the teach-ing physician need not repeat the documentationin detail. Instead, the teaching physician’s docu-mentation may be brief summary comments thatconfirm or revise the key elements of the service.The combined teaching physician-resident entriesmust be adequate to substantiate the level of serv-ice required by the patient and billed to Medicare.

In 1996, the Department of Health and HumanServices’ Office of the Inspector General (OIG) ini-tiated the Physicians at Teaching Hospitals nation-wide review for compliance with rules governingteaching physicians and the level of the servicesbeing billed. The review standards and method ofconducting the audits have been controversial. AGeneral Accounting Office (GAO) review of theOIG audits concluded that the OIG’s review stand-ards and audit methodology were reasonable; how-ever, the GAO suggested that a risk-based approachfocusing on the most problem-prone institutionswould be a more efficient use of resources (GAO,1998). Heightened attention to compliance withbilling requirements and concerns over potential


OIG audits under the False Claims Act has led manyteaching institutions to adopt more extensivedocumentation requirements than those required bythe Health Care Financing Administration, therebydecreasing faculty time for teaching.

MEDICAID SUPPORT FOR GMEUnder the Medicaid law, States have consider-

able flexibility to determine what services will becovered in addition to basic required services andthe payment methodologies that will be used to payfor covered services. Federal matching funds ap-ply to the State’s expenditures. Nearly all Statesand the District of Columbia support GME underthe Medicaid program. Using the results of anAAMC-sponsored survey, the National Conferenceof State Legislatures estimates States spent about$2.3 billion through Medicaid patient funds in 1998,or approximately 7 percent of Medicaid inpatienthospital expenditures, to support GME (Henderson,1999). Appendix A summarizes the State-level sur-vey findings. Explicit GME payments are identi-fied separately from amounts for GME that areimplicit in payments to managed care plans. Themanaged care plans have discretion regardingwhether to pass these amounts on to teaching hos-pitals through higher payments. Specific findingswere:

• Forty-three States and the District of Colum-bia make GME payments under their fee-for-service programs. Of these, 23 make both di-rect GME and IME payments.

• Thirty-three States and the District of Colum-bia include GME funds in their Medicaid man-aged care capitated payments. Of these, 16carve-out the GME payments and make ex-plicit payments to teaching hospitals or otherteaching programs.

• Among those that do not support GME throughMedicaid, only Illinois and Puerto Rico havea large number of residents. One of the remain-ing States (South Dakota) has a medicalschool; three States (Alaska, Idaho, and Mon-tana) that do not have medical schools haveagreements with the University of Washing-ton for undergraduate education and one ormore affiliated family medicine graduate train-ing programs.

Most Medicaid programs support GME in theirpayments for inpatient hospital services followingMedicare-like methodologies. However, someStates have or are in the process of obtaining ap-proval from HCFA to link some or all of Medicaid

GME payments to specific State workforce objec-tives. For example:

• Michigan and New York have established sepa-rate incentive pools to achieve specificworkforce objectives.

• Tennessee and Oklahoma make GME pay-ments to medical schools.

• In Minnesota, advanced nursing, pharmacy,dental and physician assistant training pro-grams as well as medical and dental schoolsare eligible for medical education paymentsfrom the State. The State is seeking approvalto include the Medicaid managed care carve-out amounts in the education fund.

• Utah is developing a consortium consisting ofthe State, the single university medical schoolin the State, hospitals and community-basedproviders, and private payers. The State isseeking a waiver from HCFA to have Medi-care and Medicaid GME funds flow to the con-sortia instead of the teaching hospitals.

Tennessee’s funding for GME was approved aspart of the TennCare waiver process. Other Stateswith innovative GME funding have used the Stateplan amendment process to obtain HCFA approval.State plans must be consistent with the Medicaidlaws and regulations. In contrast to the Medicarelaw, the Medicaid GME provisions are not verydetailed. As a result, States have considerably moreflexibility than the Medicare program to institutechanges in GME funding. In addition, many Stateshave established commissions or task forces relatedto GME and physician workforce. Features of se-lected State programs that are directed at meetingspecific workforce or GME objectives are summa-rized below.

MICHIGAN

Since 1997, Medicaid GME-related payments,including managed care carve out amounts, haveformed three pools. For the State’s FY2000, thepools are funded as follows: 1) a $164.5 millionhistoric cost pool that pays teaching hospitals at1995 levels; 2) a $20 million primary care pool thatpays hospitals based on the number of primary careresidents and its indigent patient load; and, 3) a $3.5million innovations pool that that awards competi-tive grants to GME consortia involving a hospital,a university, and a managed care organization. Hos-pitals that receive funding must account for howfunds are being used to support the policy goalsand priorities of the Medicaid program. Initial pro-posed performance measures are the percentageof graduates practicing in the State and the


percentage participating in Medicaid (Salsberg,1997; Holmes, 2000).

MINNESOTA

Minnesota established the Medical Educationand Research Cost (MERC) Fund in 1997 to sup-port clinical training costs. The fund is supportedwith general-fund tax revenue and the State’s to-bacco settlement. The 1999 fund was comprised of$10 million in State appropriated funds and $10.6million in Federal matching funds. A State planamendment to allow a Medicaid managed careGME carve-out amount ($12 million) is pendingHCFA approval. The fund is intended to “level theplaying field” for teaching hospitals and other pro-viders. Support is for the unrecovered costs of clini-cal training for medical students and residents, den-tal students and residents, pharmacists,chiropractors, advanced practice nurses and physi-cian assistants. Funds are allocated based on thenumber of FTE trainees and the average cost foreach eligible program. Institutional sponsors applyfor funds on behalf of their programs. The fundsflow through the sponsors to protect the integrityof the teaching programs and to obtain completeinformation about the programs. Sponsors are re-quired to self-report program costs and to pass fundsthrough to the sites where the clinical training oc-curs. Funding is a uniform percentage of eachinstitution’s program costs (7 percent in 1999 com-pared to estimated unrecovered costs of 15 percent).Medical residency programs received 86 percentof the funds. The payment was $10,661 per resi-dent compared to an estimated cost of $156,631per resident Accountability is currently limited toprogram costs and numbers of trainees. Consider-ation is being given to other measures, such as per-centage of graduates who practice in-State (MDH,2000; Blewett and Weslowski, 2000).

NEW YORK STATE

New York’s GME Reform Incentive Pool is fo-cused on reducing the number of physician train-ees in the State, increasing the number of primarycare physicians, and increasing residency trainingin ambulatory settings. In 1999, the $54 millionpool was distributed to 72 hospitals. Hospitals withnon-primary care residency programs had to meetminimum qualifying criteria of a 2 percent reduc-tion in specialty residency positions before com-peting for the funds. Distribution is based on per-formance in meeting State workforce goals. Theseinclude: downsizing residency programs, increas-ing the percentage of underrepresented minorities,increasing the percentage of residents training inunderserved areas, maintaining quality training with

continuity of care experiences, and increasing thenumber of ambulatory training sites. The priorityof increasing the percentage of primary care resi-dents was dropped as duplicative of the upweightingformula used to distribute the majority of GMEfunds. Funding is set at $31 million in 2000(NYSDOH, 2000; Calman, 2000; Cunningham,2000).19

TENNESSEE

When TennCare (the State’s Medicaid managedcare program) was initially implemented in 1995,all GME funding was eliminated. Funding was re-stored in 1996 with an agreement to flow fundsthrough the four medical schools after a transitionperiod. The agreement was to base funding on thenumber of filled residency positions and to providefinancial incentives to encourage primary care train-ing and residency placements in medicallyunderserved areas. Over a 5-year transition, $48million in GME funding to teaching hospitals wasreplaced by funding to medical schools, with 100percent of the funds flowing to medical schoolsbeginning in 2000. In addition, funds are allocatedto support primary care stipends for residents whoagree to practice in an underserved area for at leastone year. By 2000, 50 percent of the aggregate resi-dency positions sponsored by the medical schoolsare to be in primary care specialties (Franks, 1999).The primary care target spawned a cooperative ef-fort among the State’s academic health centers tomeet the 50 percent target jointly instead of indi-vidually (Meyer and Blumenthal, 1999). Eachmedical school has its own target and will receivereduced funding if it fails to meet its target.(Salsberg, 1997; Henderson, 2000) There is no capon the number of residency positions but the fixedfunding amounts serve to discourage resident ex-pansions (Meyer and Blumenthal, 1999). Annualreporting is required concerning the proportion ofresidency positions that are in primary care and theproportion of patients treated by residents that arein TennCare or medically uninsured.

To date, TennCare’s GME funding policies havemade the State’s academic health centers more ac-countable and responsive to local community healthneeds and priorities (Meyer and Blumenthal, 1999).It is too early to assess whether paying the medi-cal schools is having a significant effect onworkforce goals. The greatest impact may be feltby East Tennessee State University which, as acommunity-based medical school, did not previ-ously have access to Medicaid GME funds. Theincreased funding allowed the medical school topay community physicians for teaching (Franks,1999).


UTAH

Utah passed legislation in 1997 establishing anall-payer GME financing system that would includeprivate payers/purchasers and, pending HCFA ap-proval, Medicare and Medicaid GME funds. ACouncil on Medical Education is charged with de-termining the number and type of training positionsthat will be supported and how the funds will beallocated. The nine-member Council has represen-tatives from the public, health insurance industryand institutional sponsors of medical education.Fund distributions are to be balanced geographi-cally and by training settings, encouragemultidisciplinary training, and provide stable fund-ing for accredited programs. Direct GME funds willflow to the sponsoring institutions based on theCouncil’s assessment of heath care needs (Utah,2000; Salsberg, 1997).

HEALTH RESOURCES ANDSERVICES ADMINISTRATIONPROGRAMS

NATIONAL HEALTH SERVICE CORPS

Title III of the Public Health Service Act au-thorizes the National Health Service Corps throughthe end of FY2000. The NHSC mission is to in-crease access to primary health care services inhealth professional shortage areas (HPSAs) throughsite development and the preparation, recruitment,and retention of community-responsive, culturallycompetent primary care physicians and other healthprofessionals. The FY1999 appropriation was for$75 million in scholarship/loan repayment and $37million in field support. There are four basic pro-grams to assist with the cost of health professionstraining:

• Participants in the NHSC scholarship programhave a minimum two-year commitment andmust provide one year of service to the NHSCfor each year of participation in the scholar-ship program.

• Participants in the NHSC loan repayment pro-gram may repay the qualified educationalloans if they choose to provide primary careservices in a priority-designated HPSA for aminimum of two years.

• Thirty-five States participate in the Federal-State Loan Repayment partnership to obtainservices of health professionals by matchingtheir repayment of educational loans.

• The Federal-State-Community ScholarshipProgram supports students in primary care

health professions who are committed to re-turning to their underserved communities. TheFederal share is 40 percent.

The Student/Resident Experiences and Rota-tions in Community Health (SEARCH) program isa NHSC partnership with States to support com-munity-based training experiences. Other than theSEARCH program, the NHSC has had little directinvolvement in developing academic health center-community linkages that would expand commu-nity-based training sites for residency programs.One limiting factor is the authorizing language forthe NHSC, which limits the creditable time ofNHSC scholars and repayers to time spent in theprovision of primary care services. Time spent inother activities, such as clinical preceptorships, doesnot count under current law for repayment purposes.Periodic reauthorization for the NHSC provides anopportunity to assess whether there are additionalways the NHSC program could facilitate residencytraining in underserved areas.

TITLE VIITitles VII and VIII of the Public Health Service

Act are designed to expand the supply of primaryhealth care providers, improve the geographic dis-tribution of health professionals, and increase ac-cess to primary health care services in both urbanand rural underserved areas. Title VII programs in-clude support for educating physicians and otherhealth professionals while Title VIII programs sup-port nursing education.

Several Title VII programs have been signifi-cant sources of support to medical schools andteaching hospitals in developing community-basedtraining capacity. Primary care training grants pro-vide support for planning, developing and operat-ing residency programs in general internal medi-cine and general pediatrics, family medicine, andpreventive medicine programs. Title VII funds alsosupport geriatric education centers and geriatrictraining fellowships for physicians and other healthprofessionals.

In addition, Title VII grants support the creationof area health education centers (AHECs) andhealth education and training centers. The missionof the AHEC program is to improve the supply anddistribution of health care professionals throughcommunity/academic educational partnershipsthat eventually will become self- or State-sup-ported. Strategies to accomplish the mission includeforming linkages between educational resources andhealth care delivery systems in underserved commu-nities and creating collaborative community-based


education for primary care residents, health pro-fessionals, and students.

Funding for Title VII residency training pro-grams was relatively flat during the period FY1990-1998 and did not keep pace with inflation. Therange for family medicine grants over the periodwas $38-49 million; the average annual appropria-tion was $44 million compared to an annual appro-priation of $40 million during the 1980s. Fundingfor general internal medicine/pediatric training hasbeen constant at about $17 million annually since1980 (HRSA, 2000).

The Health Professions Education PartnershipsAct of 1998 (P.L. 105-392) made significantchanges in the administration of Title VII funds thatare consistent with COGME goals. The reauthori-zation clustered the separate training grant pro-grams for general internal medicine and pediatrics,family medicine, and physician assistants into asingle grant program that also includes general andpediatric dentistry. The aggregate funding level wasset at $80 million for FY1999. Floors were estab-lished for the level of funding for each component.Priority funding is to be given to residency pro-grams 1) with a high percentage of graduates thatenter and remain in primary care, or 2) train indi-viduals from disadvantaged backgrounds (includ-ing racial and ethnic minorities underrepresentedin primary care practice). In addition, special con-sideration is to be given to programs that prepareresidents to care for underserved populations andother high risk groups such as the elderly, individu-als with HIV/AIDS, substance abuse, homeless, andvictims of domestic violence. Grantees must estab-lish and meet specific performance goals forworkforce objectives. As in past years, the Presi-dent’s budget request for FY2000 does not includea Title VII appropriation.

The legislation also established a separate pro-gram to foster interdisciplinary, community-basedlinkages through support to develop and operatetraining programs involving two or more disci-plines, model demonstration programs, faculty de-velopment, trainee stipends and technical assis-tance. The AHEC authorization now requires eachcenter receiving funds specifically to designate aremote geographic area or medically underservedpopulation to be served by the center. The centermust:

• Assist in the planning and development oftraining programs to meet the health person-nel of the area;

• Arrange and support rotations for primary careresidents in a center-operated or affiliated pro-gram with at least four first-year positions;

• Conduct and participate in interdisciplinarytraining; and,

• Have an advisory board with at least 75 per-cent of the members being health service pro-viders and consumers from the area served bythe center.

The Title VII funding structure provides aframework for targeting Federal outlays consistentwith physician workforce and education goals.However, the level of support is less than one per-cent of total Federal funding for workforce.

CHILDREN’S TEACHING HOSPITAL FUND

Public Law 106-129 amended the Public HealthService Act to establish a new program to supportGME activities of freestanding children’s teachinghospitals. The fund was established through a bi-partisan effort in recognition that children’s teach-ing hospitals have low Medicare utilization ratesand do not receive the same level of Federal sup-port for GME as other teaching hospitals. The pro-vision authorizes payments in FY2000 and FY2001for both direct GME and IME associated with ap-proved GME programs. The payment methodolo-gies follow the general framework of the Medicaremethodologies. Direct GME payments are basedon a wage-adjusted national average per residentamount and the number of residents at each insti-tution using Medicare rules for counting residents.The IME adjustment is to take into account casemix and the number of residents. Unlike MedicareGME payments, however, funding is not an entitle-ment but requires annual appropriations. TheFY2000 appropriation is $40 million, which trans-lates into slightly less than $12,000 per resident.By contrast, Medicare’s GME payments per resi-dent is about $73,200.20

The payment methodologies for the children’steaching fund are established in the authorizing leg-islation. Payment is not explicitly linked with phy-sician workforce objectives. However, the Govern-ment Performance Results Act requires thatperformance measures be established for the pro-gram. This will provide an opportunity to assesshow each facility performs on specific workforce-related indicators. The proposed measures are:maintaining the number of residents, increasing theproportion of training in rural and underserved ar-eas, and monitoring the hospitals’ financial statusand the proportion of Medicaid and uncompensatedcare patients. Data collected for this program couldserve as a template for information that could be gath-ered from all residency programs. Such informa-tion would increase accountability of program spon-sors to meet educational and workforce objectives.


OTHER DIRECT FUNDINGSOURCES

DEPARTMENT OF VETERANS AFFAIRS

Eleven Veterans Affairs Medical Centers(VAMCs) sponsor or are the primary clinical sitefor residency programs and 97 others are partici-pating institutions in GME programs (JAMA,1999). The DVA funds 8,900 residency positions,or about 10 percent of all residency positions. About32,000 residents annually receive some training atthe DVA. The DVA has made substantial efforts overthe past decade to shift care to ambulatory settingsand to focus on primary care access supported byspecialized care for chronically ill and seriously illpatients. Residency training has been aligned withthe new health care delivery system by moving moretraining into ambulatory settings that useinterprofessional teams and away from traditionalhospital settings (DVA, 1997).

DEPARTMENT OF DEFENSE

Thirty-two Department of Defense facilitiessponsor or are the primary clinical training sitesfor residency programs and five others are partici-pating institutions (JAMA, 1999). DoD funds about3,000 residency positions annually.

STATE AND LOCAL APPROPRIATIONS

In addition to Medicaid funding, States supportGME through funding to State-operated medicalschools, training grants for family medicine resi-dencies, and workforce policy and planning activi-ties that frequently are coordinated with HRSAfunding. For example, Maryland provides about $1million to support two AHECs, has a loan re-pay-ment program for primary care physicians whoagree to serve in medically underserved areas forat least two years, and provides incentives to edu-cational institutions training health professionals

in short supply, including family physicians(Salsberg, 1997). States spend an estimated $185million to support family medicine residency pro-grams (Henderson, 2000).

FOUNDATION SUPPORT FOR COMMUNITY-BASED GRADUATE MEDICAL EDUCATION

PARTNERSHIPS FOR QUALITY EDUCATION

(PQE)

The Partnerships for Quality Education (PQE)supports the work of primary care residency pro-grams working in partnership with managed careorganizations. During Phase I, (1996-1999) PewCharitable Trusts awarded $8.3 million in grants to66 programs.. Six partnerships received sizablegrants to develop new models of educational col-laboration between academic programs and man-aged care organizations. The remaining granteesreceived $10,000 awards.

The Robert Wood Johnson Foundation is fund-ing the second phase of the PQE projects. $8.9 mil-lion in grants have been given to 28 residency pro-grams and 25 nurse practitioner programs todevelop new curricula, teaching strategies and sites(PQE, 2000).

COMMUNITY PARTNERSHIPS IN GRADUATE

MEDICAL AND NURSING EDUCATION

Ending in 2000, W.K. Kellogg Foundation’sCommunity Partnerships in Graduate Medical andNursing Education initiative emphasized commu-nity-based multi-disciplinary health care that meetsthe needs of a vulnerable population. Critical de-sign features of the six projects funded through thisinitiative were: the governing board is a partner-ship of health professionals and community repre-sentatives, funds flow through the community part-ner, and there is at least 1:1 matching fromuniversity or State sources (Kellogg, 1999).


The growth of managed systems of care andthe shift of service delivery to ambulatorysettings make it increasingly important for

residency programs to provide training opportuni-ties in community settings. As used in this report,

the term “community settings” describes settingsthat are representative of the environment in whichresidents will eventually practice. Under this defi-nition, the processes of care and educational out-comes are the determining factors in identifying acommunity setting rather than its location per se.Training should be relevant to current daily physi-cian practice and address the care of the individualpatient in the context of the population of whichthe patient is a member.

This section focuses on the financial arrange-ments and issues associated with residency train-ing in community settings. It begins with overviewsof how much training is currently being done inambulatory/community-based settings and the rel-evant accreditation requirements. A discussion ofpotential financial barriers to expanding trainingopportunities in the community follows. The dis-cussion is based on a literature review. The sectionconcludes with a summary from the site visits andinterviews conducted with individuals involvedwith selected residency programs.

CURRENT TRAINING INAMBULATORY/COMMUNITYSETTINGS

COGME’s 13th Report (COGME, 1999a) iden-tified the following core curricular domains thatphysicians need to practice effectively in the chang-ing health care environment:

• Health systems financing and delivery

• Evidence-based medicine

• Ethics and the management of dual responsi-bilities and conflicts of interest

• Patient-physician communication

• Leadership, teamwork and organizationalchange

• Quality measurement and improvement

• Systems-based care

• Medical informatics

• Teaching managed care

Competency in these domains requires learn-ing experiences that are not typically available inthe hospital outpatient departments of academic healthcenters. A comprehensive range of experiences is

Residency Training in Community Settings

TABLE 4Ambulatory/Community-Based Rotations

for Program Year 1 Residents

Average Percent of Trainingby Setting

Non-Hospital

AmbulatoryCare

HospitalOutpatient

Clinic

ManagedCare

Setting

SPECIALTIES

Family Practice ........................................ 20.5 15.7 12.5

Geriatric Medicine (FP) ........................... 34.7 13.3 23.1

Internal Medicine ..................................... 22.6 11.0 10.6

Internal Medicine Subspecialties:

CardioVascular (IM) ............................ 18.4 5.6 11.5

Gastroenterology (IM) ......................... 28.2 10.9 15.4

Geriatric Medicine (IM) ....................... 30.2 15.3 5.5

Hematology (IM) ................................. 37.3 7.5 3.5

Nephrology (IM) .................................. 28.3 8.8 11.4

Oncology ................................................. 40.7 6.0 8.8

Rheumatology ......................................... 48.3 4.1 5.2

Pediatrics ................................................ 39.9 11.7 15.4

Dermatology ............................................ 65.6 11.2 16.4

Emergency Medicine ............................... 16.8 2.8 7.7

Neurology ................................................ 26.9 2.8 7.3

Obstetrics and Gynecology ..................... 33.1 11.8 16.7

Ophthalmology ........................................ 74.7 8.7 8.9

Otolaryngology ........................................ 48.3 4.1 5.2

Psychiatry ............................................... 12.4 3.9 9.2

Physical Medicine and Rehabilitation ...... 23.8 4.7 6.2

General Surgery ...................................... 20.0 6.5 7.8

Urology ................................................... 34.2 7.4 12.0

Source: AMA FREIDA, 2000 (categories are not mutually exclusive).


FIGURE 7Proportion of Clinical Experience in Ambulatory Settings for

Categorical Internal Medicine Residency Programs

Source: ABIM, 1998a.

50%

40%

30%

20%

10%

0%PGY-1

123456789011234567890112345678901

12345678901231234567890123

123456789011234567890112345678901

123456789012123456789012123456789012

PGY-2 PGY-3

27.3%

36.1%

43.9%2.3%2.4%3.4%

5.3%

10.0%

18.0%

123456789012123456789012123456789012123456789012

123456789012312345678901231234567890123

Other

VA Clinic

Hospital-SponsoredSatellite Clinic

Emergency Departmentor Trauma Center

Office-Based Practice

Hospital-BasedAmbulatory Clinic

123412341234

123451234512345

Source: ABIM, 1998b.

FIGURE 8Internal Medicine Ambulatory Care Sites

Used for Continuity Experiences

100%

80%

60%

40%

20%

0%

27.3%

1234567890123456712345678901234567

123456789012345678

123456789123456789123456789123456789012345671234567890123456712345678901234567123456789012345671234567890123456712345678901234567

1234123412341234

Urgent Care (2%)

Other (3%)

Closed Panel HMO (4%)

Gov’t Sponsored Clinic (6%)

VA Clinic (11%)

Private Physician Office (14%)

Community-Based FacultyPractice Site (15%)

Hospital-Based AmbulatoryClinic (44%)

123412341234

123451234512345

1212

needed that includes opportunities to follow thepatient across each component of an integrated de-livery system. Community-based settings such ashealth centers and clinics, physician offices, schoolsand workplaces, nursing homes, hospices and homecare, community hospitals, and managed care or-ganizations can offer essential experiences tocomplement those at academic health centers. Forsome specialties, community training will occur in

hospital-based ambulatory sites since this is wherethe specialty commonly practices.

Table 4 is based on aggregated data from theannual survey of GME program directors as reportedon the American Medical Association’s on-line Fel-lowship and Residency Electronic Interactive Data-base (FREIDA) system. The survey asked the pro-gram directors to estimate the amount of time spentin the first program year in hospital outpatient clin-

ics, non-hospital ambulatory sites,and managed care settings. The cat-egories are not mutually exclusive.For example, time spent in an outpa-tient clinic could also be time spentin a managed care setting. As expected,residents in primary care residency pro-grams generally spend more time innon-hospital clinics than do residentsin other programs. In addition, residentsin obstetrics and gynecology and indermatology spent on average morethan 10 percent of their first year innon-hospital ambulatory care.

More detailed information on in-ternal medicine residency programsis available from the Internal Medi-cine Residency Training Survey,1997-1998. Figure 7 indicates thepercentage of time spent in ambulatorycare increases throughout the residencyprogram. Overall, the survey showed

that 37 percent of clinical training experiences overthe three years are spent in ambulatory settings. Thisrepresents a significant increase over the past de-cade. A 1986-1987 survey of internal medicine de-partments of academic health centers found that 14percent of residency training occurred in ambula-tory settings (Levinsky, 1998). However, the ma-jority of time is spent in hospital outpatient or emer-gency departments and may not provide theessential learning opportunities for effective medi-cal practice. Sites for continuity experiences in theinternal medicine residency programs are shown inFigure 8. The most common continuity sites werethe hospital-based ambulatory clinic (44 percent).The next two most common sites, community-basedfaculty practices (15 percent) and private physicianoffices (14 percent) were chosen far less often.

AAAAACCREDITCCREDITCCREDITCCREDITCCREDITAAAAATION REQUIREMENTTION REQUIREMENTTION REQUIREMENTTION REQUIREMENTTION REQUIREMENTSSSSSThe Accreditation Council for Graduate Medi-

cal Education (ACGME) is a non-governmentalbody that develops general accreditation standardsfor all medical residency programs. Within theACGME, there is a Residency Review Committee


(RRC) for each specialty that consists of represen-tatives appointed by the American Medical Asso-ciation, the appropriate specialty board, and in mostcases, the professional college or other professionalassociation within the specialty. Subject to theACGME’s approval, each RRC establishes specificaccreditation standards for its specialty. The RRCsalso review the residency training programs to as-sess whether each substantially complies with ac-creditation requirements. Generally, physicians areeligible for board certification by a member of theAmerican Board of Medical Specialties (ABMS)only if they have been trained in an ACGME-ac-credited program. Also, Medicare support for allo-pathic medical residency training programs is lim-ited to programs accredited by the ACGME orcertified by the ABMS.

Since accreditation is critical for maintainingresidency programs, RRCs have tremendous poten-tial to influence the development of ambulatory andcommunity-based training opportunities. However,with the exception of family medicine, the RRCshave established at most relatively general require-ments for community training. The primary carespecialties, the internal medicine sub-specialties,obstetrics and gynecology, neurology, and psychia-try have minimum training requirements for anambulatory continuity experience. Other special-ties do not require an ambulatory continuity expe-rience. Except for family medicine, the require-ments do not distinguish between hospital-basedor community-based continuity clinic experiences.It is not clear the extent to which the hospital-basedcontinuity clinics provide a community setting com-parable to the environment in which the residents willeventually practice. Both hospital-based and commu-nity-based continuity clinics provide the opportu-nity to follow a panel of patients longitudinally overthe course of the residency. However, the typicalhalf day per week schedule means that the residentis frequently not at the clinic when a patient has anacute condition that requires immediate treatment.Appendix B summarizes current accreditation re-quirements for the primary care specialties and se-lected other specialties with a significant propor-tion of private practice outside the hospital setting.

Generally, specialties that do not require conti-nuity experience do at least have provisions forambulatory training. However, among these special-ties, only physical and rehabilitation medicine es-tablishes minimum requirements. For some, suchas surgical specialties, a specific outpatient require-ment may be unnecessary in that training will au-tomatically follow the shift of services to ambula-tory settings in order to maintain the requisitepatient workload.

Beginning July 1999, all internships and resi-dency programs accredited by the American Os-teopathic Association must occur in OsteopathicPostdoctoral Training Institutions (OPTIs). OPTIsare training consortia comprised of one or morecolleges of osteopathic medicine and one or moreosteopathic hospitals. The OPTI structure is in-tended to enhance the quality and efficiency of theeducational programs and to promote partnershipsand collaboration among academic medicine, hos-pitals, and community based health care facilitiesin order to provide broader clinical training oppor-tunities (AOA, 2000).

There is general agreement that training expe-riences in the community are important and shouldbe expanded. However, as evidenced by the lack ofspecific RRC requirements for community train-ing, there is no consensus on what constitutes theappropriate balance between traditional and com-munity-based experiences for most specialties. Formost, training opportunities will be expanded asneeded to maintain an adequate patient base forteaching. However, without the RRC impetus tomove training into the community settings, finan-cial and other considerations at the hospital as wellas the community site may outweigh the educa-tional goal of providing training experiences in set-tings that are representative of the environment inwhich residents will eventually practice.

POTENTIAL FINANCIALBARRIERS TO TRAINING INAMBULATORY SETTINGS

Quality experiences in ambulatory settings requirea strong and interested faculty, a patient base that isclinically diverse, and a site that integrates trainingand trainees into the daily operation of the practice(Boex, 1997; COGME, 1999a). A major barrier toexpanding community-based training is the lack ofan academic reward structure for community teach-ing physicians (COGME, 1999a). Other financial bar-riers to training in community-settings that are com-monly cited in GME literature are discussed below.

IMPACT ON NET EDUCATIONAL COSTS

Ambulatory training tends to be less efficientand more faculty-intensive than inpatient training.In the inpatient setting, faculty can teach studentsat different educational levels during patient rounds.In the ambulatory setting, patients are availablegenerally only for a short period of time. Space con-straints and the focus on clinical productivity limitopportunities for teaching multiple students at the sametime (Lave, 1989; Boex, 1997; Philibert, 1999).


In addition, physician revenues for ambulatoryservices tend to be lower than revenues for servicesfurnished in the inpatient setting. Thus, patient carerevenues generated when residents are involved inambulatory care are generally less than when thecare is in inpatient settings. Further, there is a dis-tinction between revenues generated by a hospital-based clinic, which include Medicare (and in manyStates Medicaid) GME payments, and revenuesgenerated by community clinics. The latter do notreceive Medicare payments directly (unless theyqualify under the BBA provision) and, in the caseof community health centers, have few privatelyinsured patients and high uncompensated costs.

IMPACT ON AMBULATORY CAREPROVIDERS

A critical issue in expanding ambulatory train-ing is the impact residency training programs haveon the costs and clinical productivity of the ambu-latory provider. Studies in this area are fairly datedand have had mixed results. One reason is the costsare different across the variety of settings in whichambulatory training takes place (Lave, 1989) andoften involve medical and other health professionsstudents as well as residents. Another reason is thedifferent approaches studies have taken to measur-ing the effects the presence of residents has on costs.

• Some studies take an incremental accountingcost approach or time and motion approach todetermine the additional costs when residents(or residents and medical students) are presentcompared to non-teaching sites or non-teach-ing physicians at teaching sites (Jones, 1997;Ricer, 1998; Kosecoff, 1987).

• Others measure the opportunity costs of hav-ing residents. That is, the forgone clinical rev-enue for the time preceptors spend in teachingthat could have been spent generating patientrevenues (Gamble and Lee, 1991; Ricer, 1997;Flanagan, 1995).

• The “replacement cost” approach measurescosts that the provider would incur to replacethe services of the resident (Delbanco andCalkins, 1988; Gavett and Mushlin, 1986).

Some studies have tried to measure the effectof teaching on the ambulatory site’s productivityand net revenues (Diamond, 1993; Lave, 1989).These studies are most closely directed at the ques-tion of whether there are unrecovered GME costsin ambulatory settings. The findings suggest netcosts are site-dependent and reflect factors such asthe amount of time spent on teaching relative topatient care, patient flow and the efficiency of the

practice, payer mix, the physicians’ incomes andresidents’ salaries (Lave, 1989). The studies sup-port a widespread belief that residency programs“break even.” First year residents generate produc-tivity losses. Second year residents generally haveno effect on productivity. Net increases in produc-tivity and revenues by the third year of residencytraining in ambulatory sites offset the first-yearlosses (Diamond, 1993; Flanagan, 1995; Lave,1989). When residents are in the inpatient setting,they assist the teaching physicians and reduce thetime spent by faculty members in caring for pa-tients. This time saving does not appear to carryover to ambulatory settings where spending timewith the resident going over the case allows theteaching physician less time to spend on patient care(Boex, 1997).

In addition to potential cost impacts, teachingcan affect the ability of the ambulatory site to at-tract patients in a competitive market. Patient sched-uling in continuity clinics can be problematic be-cause residents are not always at the site. Waitingtimes are frequently longer and visits take longeras well, sometimes leading to patient preferencesthat residents not be involved in their care.

IMPACT ON THE SPONSORING HOSPITALAND FACULTY

Medical schools and faculty benefit from thepresence of residents. Many residents have signifi-cant teaching responsibilities for medical studentsthat may make community-based rotations moredifficult. It is possible this issue may diminish asmedical students increasingly receive training incommunity settings. Residents also provide cover-age for teaching physicians and contribute to theservices billed by teaching physicians. Residentreplacement costs can be a significant factor in deci-sions to rotate residents to non-hospital settings. Thecost of replacing residents will depend on the relativecost of the staff who are replacing them, their pro-ductivity, and their ability to generate revenue.

Residents in community-based settings can pro-vide subspecialty referrals and referrals to the teach-ing hospital. Most studies do not provide credit forthese referrals when evaluating the cost of rotatingresidents outside the hospital (Stern, 1977; Lave,1989).

SITE VISIT FINDINGSAs part of this study, site visits or telephone

interviews were conducted with individuals in-volved with selected residency programs with com-munity-based training sites. Both programs with


long-standing community-based training sites andrelatively new programs that are developing col-laborative arrangements with managed care orga-nizations or community providers in medicallyunderserved areas were included. Geographic di-versity was a selection criteria coupled with someemphasis on selecting sites located in States withinnovative GME funding. The sites that were in-volved in the interviews are listed in Appendix C.

The interviews focused on the funding sourcesand arrangements used to support ambulatory train-ing. Of necessity, most of the information gatheredon community-based training concerned primarycare residency programs. With a few notable ex-ceptions, ambulatory training in other specialtiesis typically hospital-based and organized by depart-ment. Where possible, the views of individuals witheducational and financial responsibilities were ob-tained from the perspective of the sponsoring insti-tution as well as the community provider. The goalwas to identify both features that have facilitatedcommunity-based training and barriers to expand-ing community-based training sites that should beconsidered in evaluating alternative GME financ-ing policies. The report’s findings reflect the cir-cumstances of the programs where the interviewswere conducted. However, similar themes emergedfrom the interviews with different sites that are alsoconsistent with other information on GME financ-ing issues.

The key findings from the site visits are dis-cussed below.

SITE VISIT FINDING 1 – THERE IS A WIDE RANGE

OF COMMUNITY SETTINGS FOR RESIDENCY

TRAINING OPPORTUNITIES.

An individual program may have arrangementsfor teaching with hospital-based continuity clinics,hospital-owned and hospital-affiliated physicianpractices, community health centers, and individualclinician-educators in private practice. The charac-teristics of the most common sites are describedbelow.

HOSPITAL-BASED CLINICS

The most prevalent model for community-based training is the teaching hospital clinicthat provides continuing and comprehensivecare to a panel of patients. Residents are as-signed a panel of patients that they followthroughout their residency. The continuityclinic may be based at a hospital that is part ofan academic health center, a local public hos-pital, or a community hospital. Populationsserved by hospital-based continuity clinics tend

to be sicker and poorer than patients treated incommunity-based private practices. Payer mixranges from largely private paying to largelyindigent with Medicaid or no insurance cover-age. Payer mix has significant impact onwhether the faculty practice plan income is ableto provide most or all physician salary supportfor the clinic. Teaching faculty typically receivebase compensation with incentive paymentsbased on clinical productivity. Teaching is re-quired as part of the contractual arrangementbut is not necessarily taken into account in es-tablishing productivity levels for incentivecompensation. There may also be physiciansunder contract that are engaged in patient careonly and do not teach. Resident salaries are paidby the hospital.

Community hospital clinics also serve as resi-dency training sites for either continuity clin-ics or block rotations. There is typically a smallcore faculty that is salaried by the communityhospital. The majority of the teaching is bycommunity attending physicians who receivelittle or no compensation from the hospital. Thecommunity hospital assumes the costs for resi-dent salaries and fringe benefits and may payan overhead amount to the sponsoring hospi-tal as well.

COMMUNITY HEALTH CENTERS

Until the growth of integrated delivery systems,most non-hospital residency training occurredin freestanding community health centers orother clinics serving indigent patients. The sitesare used to fulfill continuity clinic requirementsas well as for block rotations. Physicians atthese sites are salaried by the center. Typically,there has been no explicit compensation eitherfrom the site or the hospital to the physicianfor any additional time required to teach resi-dents. Arrangements vary regarding whetherthe hospital or the site pays the resident’s salary.

HOSPITAL-OWNED PHYSICIAN PRACTICES

While some teaching institutions have had along tradition of operating clinics in the com-munity, most hospital-owned community clin-ics opened in response to market pressuresduring the 1990’s. Teaching institutions pur-chased or established community-based prac-tices to maintain their market position in thecommunity. The driving force was the desireto establish a physician network for referralsand to compete more effectively for managedcare contracts. However, these hospital-ownedcommunity practices also afford expanded


opportunities for quality community-basedtraining when the physicians practicing in thosesites are committed to being clinician teach-ers. When the site is used for teaching, a typi-cal arrangement is for the preceptor to hold aclinical faculty appointment only. Teachingmay or may not be required as part of the con-tractual arrangement. Generally, it is not takeninto account in establishing productivity lev-els for incentive compensation. Resident sala-ries are paid by the teaching institution.

HOSPITAL-AFFILIATED PHYSICIAN PRACTICES

As part of a strategy to develop a network tocompete for managed care contracts, teachinghospitals also affiliated with physician privatepractices. Those that serve as teaching sites largelyrely on volunteer faculty. These are communityphysicians with clinical faculty appointments whodo not receive monetary compensation for theirteaching. They may be rewarded through “inkind” benefits such as access to university fa-cilities, faculty development activities, and rec-ognition functions. Arrangements vary regard-ing whether the teaching institution (hospitalor medical school) or the physician practicepays the resident’s salary. These sites are morecommonly used for block rotations than forcontinuity experiences. Frequently, the residentworks one-on-one with a preceptor.

MANAGED CARE ORGANIZATIONS

Some residency programs have had long-stand-ing relationships with group or staff modelHMOs. Typically, the teaching physicians aresalaried by the HMO and hold clinical facultyappointments at the medical school in university-affiliated programs. Arrangements vary regard-ing whether the site or the medical school/teaching hospital pays the resident’s salary.

Residency training also occurs in physicianpractices that are members of independentpractice associations and preferred providerorganizations. The amount of training in thesesettings (which could include hospital-basedas well as community-based sites) is more dif-ficult to establish. The agreements are with thephysicians rather than the managed care orga-nization. With increased managed care penetra-tion, it is likely significant amounts of com-munity-based training takes place in settingsthat have one or more managed care contracts.

OTHER SETTINGS

Indian Health Service clinics and Departmentof Veterans Affairs (DVA) clinics also serve as

residency training sites for both continuity andblock rotations. The DVA, with its emphasison ambulatory care, is a significant sponsor ofambulatory training. There is also a wide-vari-ety of other settings that typically serve as atraining site for short-term rotations. These in-clude other health care providers such as nurs-ing homes and hospices, school and employer-based programs, public health agencies,administrative offices of managed care orga-nizations, etc.

SITE VISIT FINDING 2 – THE FINANCING AR-RANGEMENTS FOR AMBULATORY TRAINING SITES

VARY, REFLECT LOCAL CIRCUMSTANCES, AND

FREQUENTLY INVOLVE OTHER INTERRELATION-SHIPS AS WELL.

The financing arrangements differ across sitesdepending on a number of factors, including:

• Payer mix and sources of support for theteaching site – The sponsoring institution ismore likely to provide support to the trainingsite if it has high uncompensated care costs.Sites with adequate third-party support aremore likely to absorb the teaching costs.

• Intensity of the teaching effort – The spon-soring institution is more likely to provide sup-port if the community site is engaged in teach-ing on an on-going basis for several residentsthan if the site intermittently teaches residentson short-term rotations (e.g. one month a year).

• Opportunity costs and disruption – Thesponsoring institution is more likely to com-pensate the community provider if the teach-ing activity poses significant scheduling chal-lenges for the community provider than if theresidents can be integrated into the practicefairly easily. Resident-pairing (matching resi-dents with complementary schedules) andother strategies to maintain a constant num-ber of residents at the site increases produc-tivity and reduces the need for support.

• Other financial arrangements with the site –Other arrangements with the site may affectarrangements for GME. For example, the sitemay be compensated as a clinical training sitefor undergraduate medical students. The sitemay have other relationships with the hospi-tal as part of its physician referral network.

The variety of arrangements suggests that de-cisions on how GME funds should be allocated arebest made at the local level. A single national policythat allocates funds between hospital and commu-nity-based sites using a pre-determined formula


does not acknowledge the myriad of existing ar-rangements for community-based training thatcould be disrupted. After examining this issue ex-tensively, the New York State Council on GraduateMedical Education (NYSCOGME) reached a simi-lar conclusion. That Council decided that strength-ening the HCFA requirements for a written agree-ment between the hospital and the training site wasmore appropriate than paying the site directly forGME (Calman, 2000; Cunningham, 2000).

SITE VISIT FINDING 3 – THERE IS SOME EVI-DENCE THAT HCFA’ S REVISED DEFINITION OF

“ ALL OR SUBSTANTIALLY ALL OF THE COSTS” OF

NON-HOSPITAL TRAINING IS AFFECTING FINAN -CIAL ARRANGEMENTS WITH COMMUNITY TRAIN-ING SITES.

At several sites, financial support was beingprovided for the first time to a community-basedsite or was being increased. These involved long-standing academic-community partnerships with afew community health centers and a managed careorganization. HCFA’s revised rules governing thewritten agreement between the hospital and thecommunity site as well as general fairness werecited as reasons for the change.21 No situationswhere arrangements with physicians in private prac-tice have been affected were identified.

It is too soon to assess whether HCFA’s ruleshave had a significant impact on the financial ar-rangements between hospitals and community sites.It is not clear whether community sites have suffi-cient understanding of the rules and the potentialfor negotiating appropriate financial arrangementswith the hospital. Additional education and provi-sion of benchmark data on physician compensa-tion could facilitate the process.

SITE VISIT FINDING 4 – RESIDENTS ARE NOR-MALLY PAID BY THE SPONSOR INSTITUTION . THE

SALARY AND FRINGE BENEFIT COSTS ARE TYPI-CALLY REIMBURSED BY THE HOSPITAL THAT RE-CEIVES GME FUNDING.

Typically, the sponsoring institution pays theresident’s salary and fringe benefits costs. It is re-imbursed by a hospital to which the resident ro-tates for either inpatient or ambulatory training. Thehospital claims the resident for Medicare GMEpayments based on the time the resident spends atthe hospital. At two sites involving residents inuniversity-based programs rotating to rural hospi-tals, the rural hospital does not claim the residentsfor direct GME payments. In one, the hospital stillpays the resident’s salary. In the other, the medical

school picks up 50 percent of the resident’s salaryand the remainder is paid by a physician group af-filiated with the rural hospital.

Arrangements for rotations to non-hospitalcommunity sites are more varied. When Medicarefunding is involved, the hospital generally assumesthe resident’s salary costs. When no Medicare GMEfunding is involved, the hospital and community-site may share the costs or the site may pay thehospital for the resident. A single program may havemultiple arrangements that are site-specific.

SITE VISIT FINDING 5 – MONETARY COMPENSA-TION FOR TEACHING PHYSICIANS IN COMMUNITY

SETTINGS IS UNCOMMON. THERE IS SOME EVI-DENCE THAT ARRANGEMENTS ARE CHANGING INTEACHING-INTENSIVE SITES.

Within hospital-based ambulatory clinics, aca-demic faculty are expected to serve as clinical pre-ceptors. The amount of teaching, however, is oftennot specified in the contract and may differ amongthe faculty in the department. A few departmentshave developed ways to take additional clinicalteaching responsibilities into account in setting pro-ductivity levels for incentive compensation. How-ever, most appear to rely on faculty to assume ad-ditional teaching sessions without supplementalcompensation. Several sites mentioned that theywere exploring how to give appropriate recogni-tion to faculty with a disproportionate share of clini-cal teaching responsibilities.

Monetary compensation explicitly for teachingin community-based sites rarely occurs, evenamong hospital-owned sites with employed physi-cians. When employed physicians are involved, theresidency program directors believe they have morecontrol over teaching quality even if there are nospecific financial rewards for teaching. Several pro-gram directors expressed concern that they have lesscontrol over the quality of the teaching experiencewith volunteer community physicians. They be-lieved it was easier to establish accountability whenmonetary rewards, even token amounts, are paid tocommunity physicians.

SITE VISIT FINDING 6 – THERE IS CONCERN THAT

CLINICIAN -EDUCATORS IN THE COMMUNITY ARE

BECOMING LESS WILLING TO TEACH WITHOUT

COMPENSATION.

There is already a scarcity of highly qualifiedclinician-educators who are willing to teach with-out compensation, particularly in communities withmultiple medical schools competing for clinicaltraining sites for both medical students and


residents. A common strategy has been to incorpo-rate teaching into community-based practices thatare under the direct control of the sponsoring insti-tution. This arrangement provides more control overthe selection of clinician-educators, accountabilityfor the quality of the training experience, and sta-bility among the faculty.

Several program directors expressed concernover current difficulties in placing residents in com-munity-based sites with volunteer faculty. Eventhose who are currently experiencing few problemsquestioned whether they would be able to findhighly qualified clinicians who will be willing toteach without compensation in the future. Fre-quently cited reasons for concern were the com-petitive pressures for clinical productivity and theMedicare teaching physician rules. At the sametime, the program sponsors did not believe they hadthe financial resources to pay community clinician-educators.

SITE VISIT FINDING 7 – GRANT FUNDS HAVE

BEEN A KEY FACTOR IN DEVELOPING TRAINING

OPPORTUNITIES IN COMMUNITY-BASED SET-TINGS.

HRSA primary care residency training grants,State funds such as California’s Song Brown Fund,and foundation grants were consistently reportedas important “seed money” to start the dialogue withcommunity-based sites and to provide the funds forcurriculum and faculty development. For sites in-volving academic partnerships with communityproviders serving indigent populations, on-goinggrant support has been particularly important. Un-compensated care makes it difficult for these sitesto become self-sustainable.

SITE VISIT FINDING 8 – MEDICARE’S TEACHING

PHYSICIAN RULES HAVE HAD MIXED IMPACT.THERE IS CONSIDERABLE CONCERN THAT THEY

ARE INTERFERING WITH HIGH QUALITY GRADU-ATE MEDICAL EDUCATION.

Medicare’s teaching physician rules have hada mixed impact on residency training sites in com-munity settings. Some sites have been unaffectedby the rules. These include managed care sites andnon-hospital sites where the resident’s time is notclaimed for direct GME payments and the residentbills under Part B. There has also been minimalimpact in clinics with low Medicare utilization, pro-vided the Medicaid program has not adopted theMedicare rules.

The impact in other community settings is in-fluenced by a number of factors. These include: the

ratio of residents-to-preceptors; whether the ruleswere adopted by the State Medicaid agency;whether the site applies the rules to all patients; thesite’s teaching physician supervision levels beforethe rules were implemented; whether the primarycare exception applies; and, how the documenta-tion requirements are interpreted. Several sites ex-pressed little concern over the rules. A few indi-cated that they thought the rules have improved thequality of teaching supervision and foster bettercontinuity of care if the patient needs care laterwhen the resident is not available. Other sites hadserious concerns about adverse effects on theirteaching programs. Key points that were made in-clude the following:

• The requirements for Part B billing for evalu-ation and management services are counter tothe educational goal of progressively indepen-dent action by the resident.

• The requirements also endanger teaching innon-hospital settings. These frequently involveone-on-one resident to preceptor relationships.If the preceptor has to be present for the keyportion of all evaluation and management vis-its, the preceptor cannot see patients at thesame time. Productivity is greatly reduced andmay preclude teaching at the site.

• The supervision requirements for the primarycare exception exceed the requirements of theaccrediting organizations. The rules have re-quired an increase in the number of teachingphysicians per session and have reduced pro-ductivity by precluding the supervisory phy-sician from seeing other patients.

• Lack of clarity on the documentation require-ments and desire to avoid any errors that mightlead to problems under the False Claims Acthave resulted in unnecessarily conservativebilling practices, overly burdensome documen-tation procedures, and reluctance of commu-nity physicians to teach residents.

SITE VISIT FINDING 9 – THE BALANCED BUD-GET ACT OF 1997 LIMITS ON FTE RESIDENT

COUNTS ARE AFFECTING FLEXIBILITY TO MOVE

RESIDENTS AMONG HOSPITALS AND TO EXPAND

NEEDED PROGRAMS.

The Balanced Budget Act of 1997 limits onresident counts has had unintended consequencesfor several GME programs. One issue revolvesaround residents who were training in ambulatorysites in 1996 and are not included in the hospital’scount. Several sites raised the equity issue of not


rewarding programs that had already taken actionto expand ambulatory training capacity.

In several situations, hospitals are still work-ing to resolve issues of who “owned” residents whowere training in ambulatory settings in 1996. Hos-pitals at their BBA limit do not want to assume thecosts for residents that cannot generate IME pay-ments. The hospitals are reluctant to serve as a train-ing site unless they receive direct GME and IMEpayments for the time the resident spends at thehospital. This makes it difficult to move residentsfor educational purposes.

Finally, the issue of the long-term appropriate-ness of an across-the-board limit on growth in resi-dency programs in light of geographic imbalancesin physician supply was raised.

SITE VISIT FINDING 10 – TRAINING IS BEING EX-PANDED IN AMBULATORY SETTINGS WITHOUT IM -PACTING NEGATIVELY ON RESIDENT COVERAGE

ON INPATIENT SERVICES.

Several factors seem to have combined to fa-cilitate the expanding ambulatory training capac-ity without adversely impacting on resident cover-age on inpatient services. Following is a summaryof what was reported during the interviews:

• The RRC requirements have been influentialis creating a common understanding that resi-dency training in ambulatory settings must in-

crease. They have shifted the dialogue fromwhether the shift in training should occur tohow it can occur.

• Most sites have experienced reductions in in-patient census that reduces the amount of cov-erage needed on inpatient services.

• Hospital mergers and consolidations have fa-cilitated the shift to outpatient training.

• Several programs have found hospitalists canbe a cost-effective replacement for residents.Fee-for-service billings and productivity sav-ings from efficient care management byhospitalists can offset their higher salary costsand shorter hours.

This issue was framed within the context ofresident replacement costs for shifting training fromhospital inpatient services to ambulatory services.Most shifts have been from inpatient to hospital-based clinics. It appears a shift to community-basedtraining sites where the hospital assumes the sal-ary costs but does not receive the benefit of theresident’s services remains an issue. The issue ofreplacing the resident’s services includes both theresident’s patient care services and the resident’stime spent teaching and supervising medical stu-dents. Also, the issue may be different for publicteaching hospitals and other hospitals with highuncompensated care caseloads where hospitalistscannot generate sufficient revenue to offset theircompensation.


The changing health care environment and theuncertainties of continued reliance on Medi-care, Medicaid and private pay revenues to

fund GME have led COGME and others to con-clude major changes are needed in the way GME isfinanced. Major features of recent proposals foralternative GME financing mechanisms are high-lighted below. Underlying the all-payer proposalsis an assumption that GME is a public good thatmerits broad support. An alternative assumptionmade by the Medicare Payment Advisory Commis-sion is that GME results in enhanced patient carethat should be recognized in health care service-related payments.

ALL-PAYER PROPOSALSConcerned that the competitive marketplace

will not provide adequate support for GME,COGME’s 14th Report (1999b) recommends thedevelopment of an all-payer financing system thatwould spread the costs of preparing a well-quali-fied physician workforce across all payers. In mak-ing its recommendation, COGME noted that itwould be important to build in mechanisms forpublic accountability to assure the all-payer fundssupported workforce needs and were adequate forambulatory training. In addition to COGME, otheradvocates of an all-payer fund include the PEWHealth Professions Commission, the Common-wealth Fund’s Taskforce on Academic Health Cen-ters, and the 1997 Consensus Statement on the Phy-sician Workforce by associations representingphysicians and teaching institutions. The specific-ity and details of the recommendations made bythese advocates of all-payer funding vary. Two bills(H.R. 1224 and S.210) were introduced in the 106thCongress that would provide all-payer funding forGME. Key elements of the all-payer proposals aresummarized below.

PEW HPEW HPEW HPEW HPEW HEALEALEALEALEALTHTHTHTHTH P P P P PROFESSIONSROFESSIONSROFESSIONSROFESSIONSROFESSIONSCCCCCOMMISSIONOMMISSIONOMMISSIONOMMISSIONOMMISSION

The PEW Commission (1998) recommends anall-payer financing mechanism exclusively dedi-cated to supporting entities involved in the clinicaleducation of physicians, advanced practice nurses,and physician assistants. (Funding for other publicgoods, i.e. research, highly specialized and charitycare would come from other funding mechanisms.)

GME Reform Proposals

All entities providing clinical education, includingconsortia and children’s hospitals, would be eligiblefor payment.

Total funded positions would not exceed 110percent of the number of U.S. medical school gradu-ates in 1997 and the BBA limits on FTE residentcounts would continue to apply at the institutionallevel.

• There would be a uniform per resident amountfor direct GME costs that would vary only forexternal reasons such as geographic differ-ences in the cost of living. To ensure an ad-equate supply of generalist physicians, pay-ment would be made only for residents in theirinitial residency period and only to institutionsthat continue to offer at least the number ofgeneralist positions they offered in 1997.

• Eligibility for IME payments would be con-sistent with those for direct GME payments.IME payments would be distributed amongteaching hospitals, affiliated academic insti-tutions, and non-hospital training sites. To re-duce disincentives for decreasing the numberof residents, a significant portion of IME pay-ments to teaching hospitals would be basedon historical IME revenues (rather than thecurrent resident-to-bed ratio).

To replace residents furnishing care to under-served populations, the Commission would expandthe National Health Service Corps’ loan repaymentprogram and allow specialists to participate whereneeded.

COMMONWEALTH FUND

The Commonwealth Fund’s Taskforce on Aca-demic Health Centers (1997) recommends level-ing the competitive playing field for academichealth centers and other teaching hospitals by es-tablishing explicit mission-related all-payer financ-ing for their social missions. The taskforce recom-mends that educational support be loweredcommensurate with decreasing the number of newphysician trainees to 110 percent of the graduatesof U.S. medical schools.

• Payments for direct GME costs would be madeon a per resident basis.

• Payments for IME and other indirect costs ofsocial missions would be linked to patient care.


The Taskforce recommends that payment besite-neutral so that education and other academichealth center missions occur in the most appropri-ate location. Demonstration projects are recom-mended as a way to evaluate new institutional ar-rangements that may accelerate the transition ofprograms into ambulatory and community-basedsettings.

CONSENSUS STATEMENT ON THEPHYSICIAN WORKFORCE

Associations representing major stakeholdersin GME issued a consensus statement in 1997 onGME financing and workforce issues (AACOM etal, 1997). The recommendations on physician sup-ply issues include:

• Limit Federal funding of residency positionsto the number sufficient to allow graduates ofU.S. medical schools to enroll in an approvedresidency program.

• Provide Federal funding to expand trainingopportunities in medically underserved com-munities. Provide incentives to encourage stu-dents to become generalist physicians andpractice in underserved areas.

• Provide transition funding for hospitals thatlose residents.

The statement supported an all-payer fund fordirect GME costs only. Funds would flow to enti-ties that incur the costs of GME, whether they behospital-based or not, or to consortia that have beendesignated to receive funds on behalf of the enti-ties that incur the costs. The statement also calledfor a national workforce advisory body to monitorand periodically assess the size and specialty com-position of the physician workforce.

ALL-PAYER GRADUATE MEDICALEDUCATION ACT

The All-Payer Graduate Medical Education Act(H.R. 1224, 106th Congress, 1st Session)22 wouldsupplement current Medicare payments to teach-ing hospitals with additional payments for patientscovered by non-governmental payers.

• Hospitals would be required to provide an as-surance that at least 20 percent of the fundswould be used to compensate teaching physi-cians.

• Payment for direct GME costs would be basedon a national wage-adjusted per residentamount (resident salaries and fringe benefitsonly) and the percentage of hospital revenue

(inpatient and outpatient) attributable to patientswith non-governmental third party coverage.

• IME payments would be based on the amountsthat would have been payable had the patientsbeen entitled to Medicare.

• If necessary, pro-rata reductions would bemade in payments so that total payments donot exceed the amounts in the all-payer fund.

Conforming changes would be made in Medi-care to require the assurance for teaching physi-cian compensation and to base Medicare’s share ofGME payments on the percentage of hospital rev-enues (rather than percentage of inpatient days)attributable to Medicare beneficiaries. There are nospecial provisions pertaining to training in com-munity-based settings. The Secretary is required tostudy the appropriate level of documentation thatshould be required as a condition of payment forPart B teaching physician services. The BBA lim-its on FTE resident counts would apply and theSecretary would be required to develop a plan toreduce the number of residents beginning July 2005to 110 percent of the graduates from U.S. medicalschools. The Secretary would be required to moni-tor the distribution of specialties to assure there isan adequate number of primary care physicians.Hospitals receiving IME payments would be re-quired to report annually on how they contributedto education, improvements in clinical services andresearch infrastructure, and the provision of com-munity services.

MEDICAL EDUCATION TRUST FUND ACTOF 1999

The Medical Education Trust Fund Act of 1999(S.210, 106th Congress, 1st Session)23 would es-tablish separate Medicare and non-Medicare fundsfor direct GME and IME payments to teaching hos-pitals. There is also a medical school account. To-tal non-Medicare funding is estimated at $8 bil-lion.24 The funds would be generated through a 1.5percent tax on health insurance premiums andhealth-related administrative services and 5 per-cent of Federal spending on Medicaid acute careservices.

The payment formulae for the non-Medicareaccounts mirror current Medicare payment meth-odologies. There are no special provisions forcommunity-based training. The legislation providesfor an Advisory Commission to study a number ofissues, including policies designed to expand eli-gibility for GME payments to institutions other thanteaching hospitals. Funding is authorized for dem-onstration projects.


COMMISSION ON THE FUTUREOF MEDICARE

The Balanced Budget Act of 1997 created theNational Bipartisan Commission on the Future ofMedicare to examine issues related to the futuresolvency of the Medicare program. One of itscharges was to explore broad-based GME financ-ing alternatives. A GME Study Group recom-mended carving direct GME payments out of theMedicare program (Breaux-Thomas, 1999). DirectGME funding would continue either through amandatory entitlement or multi-year discretionaryappropriation program separate from Medicare. Theproposal also recommended exploring funding IMEand disproportionate share payments outside theMedicare program. The full commission failed toreach consensus on this or other recommendationsand did not issue a final report (National Biparti-san Commission, 1999).

MEDPAC PROPOSALThe Balanced Budget Act of 1997 required the

Medicare Payment Advisory Commission(MedPAC) to examine Federal policies that affectGME, Medicare’s payments to teaching hospitals,and workforce issues. In its August 1999 report,the Commission concluded that the Medicare dis-tinction between direct and indirect costs is an ac-counting artifact and that both represent costs ofproviding patient care. Instead of making separatepayments for direct GME costs as an educationalcost, the Commission recommended that these costsbe combined with other patient care costs and thatMedicare’s payments be adjusted to reflect thehigher costs of providing enhanced patient care inteaching hospitals. In addition, the Commission

recommended that a teaching adjustment be devel-oped for other settings where residents or otherhealth professionals train if the added value of pa-tient care justifies their higher costs. The MedPACreport concluded that Medicare’s primary purposeis to provide beneficiaries with access to care andthat specific targeted programs may be more ap-propriate vehicles for achieving physicianworkforce goals.

In its June 2000 report, MedPAC refined therecommendations in its earlier report (2000b). TheCommission recommends that a revised Medicareteaching hospital adjustment be set at a level thatwould provide an aggregate teaching hospital sub-sidy comparable to that provided under current law(about $1.5 billion). The report defines the subsidyas the difference between payment amounts andhigher costs per case attributable to teaching ac-tivities. The Commission is concerned that reduc-ing the subsidy below the levels established by theBBA could place undue financial strain on teach-ing hospitals.

The June 2000 report recommends that thehigher costs attributable to teaching activity be de-termined after folding direct GME costs for inpa-tient care into the Medicare inpatient cost base. Inmeasuring teaching intensity, the full resident countrather than inpatient resident count would be usedin the resident-to-bed ratio. The full resident countshould avoid creating an incentive to shift residentsfrom outpatient to inpatient settings. Direct GMEcosts for outpatient and other settings would con-tinue based on a hospital-specific per resident pay-ment amount determined from outpatient GMEcosts. The goal would be to eliminate any financialincentives hospitals might have to shift residentsamong settings.


This section begins with a summary ofCOGME’s overall goals for funding GME.It is followed by a discussion of specific

policy considerations relevant to residency train-ing in community-based settings. These subsectionsprovide a policy framework for evaluating alterna-tive models for distributing GME funds. Fundamen-tal issues to consider in evaluating the models are:

• Who should receive payments directly fromthe fund;

• How to allocate the funds among the receiv-ing entities; and,

• How to establish accountability for the funds.

The remainder of this section analyzes fourpotential models for distributing GME funds. Thedifferences between three models are based onwhich entities would receive payments directlyfrom the fund: health care providers, educationalinstitutions, or GME planning bodies. The fourthmodel would link payment to specific performancemeasures. The models are not mutually exclusiveand a combination is needed to achieve the policyobjectives for the GME fund. Recommendationsfor GME funding follow in the next section,“Recommedations for GME Financing Reform.”

POLICY GOALS FOR GMEFUNDING

GME funding policies should meet the follow-ing objectives:

• Provide a stable funding mechanism that isresponsive to the community yet consistentwith national workforce objectives;

• Enable health care institutions to compete onprice and quality by subsidizing higher costsattributable to educational activities and un-compensated care without supporting ineffi-ciencies;

• Create adequate support and appropriate in-centives for developing community-based edu-cational programs;

• Encourage effective and efficient educationalmodels that promote improved ways to meethealth care needs;

• Foster mechanisms that will stabilize the totalnumber of physicians while improving the

specialty and geographic distribution of thefuture physician workforce; and,

• Hold recipients of Federal and State funds ac-countable for producing needed public goods.

POLICY CONSIDERATIONS FORCOMMUNITY-BASED TRAINING

Educating physicians in the environment inwhich they will eventually practice requires expan-sion of residency training in ambulatory settingsfor all specialties for most training in community-based settings. The BBA eliminated some disincen-tives for a hospital to rotate residents to non-hospi-tal settings by paying IME for the resident’s timein the non-hospital setting. However, the BBA lim-its on FTE resident counts operate against thosehospitals that were already training residents incommunity-based settings. A modification in thelaw to count the residents who were in non-hospi-tal settings in the base year would remedy this prob-lem. The remaining disincentives to rotate residentsto community-based settings are more difficult toresolve:

• As the recipient of the funds, the hospital is inthe stronger bargaining position regarding whoshould bear the direct costs of training in theambulatory setting.

• Hospital service demands compete with edu-cational needs to rotate residents to commu-nity-based settings.

• The hospital’s patient care needs and finan-cial interests rather than physician workforceneeds may determine the numbers and spe-cialty mix of residents and residency programs.

• Accountability is difficult to establish becausethe program sponsor, rather than the hospital,has the ultimate responsibility for the conductof the educational program.

To some extent, fund allocation policies couldreduce these disincentives.

• Clear guidance might be provided regardingreasonable financial agreements between thehospital and community-based sites. In addi-tion, public accounting of how the funds weredispersed could be required.

• Empirical research is needed to determine ifan IME adjustment is also justified for other

Alternative Models for GME Funding


settings (e.g. hospital outpatient and commu-nity health centers). If the research finds thatteaching increases patient care costs in thosesettings, support for indirect teaching costscould flow directly to the ambulatory sites.

• Elimination of Medicare utilization as a pay-ment factor will result in more comparablepayments across teaching sites.

• The funding formulae could be modified toreinforce workforce objectives. For example,relatively higher payments could be made forresidents in primary care programs or fortraining in community-based settings. Re-search is needed if the adjustments are to beestablished on an empirical rather than solelypolicy basis.

• Payments could be based on a historical resi-dent count rather than current resident count.This has the advantage of providing neutralincentives for an individual hospital regard-ing the number of residents it trains and wherethe training takes place. However, it also re-duces flexibility to move residents among hos-pitals and settings consistent with educationalneeds.

The remainder of this section discusses poten-tial payment models that could be used to distrib-ute the GME funds. The models focus on the gen-eral issues regarding who should receive directpayments from the GME fund and how to establishaccountability for the payments. Allocation poli-cies designed to expand community-based train-ing opportunities can be built into each of thesemodels.

HEALTH CARE PROVIDER MODELThe health care provider model links payments

for clinical training to patient care activities. Thisis the approach Medicare uses. It treats clinicaltraining costs as patient care costs as opposed toeducational costs. It is the most appropriate modelfor the indirect costs of clinical training. These costsreflect the impact of the teaching activity on thecosts of the health care provider where the trainingtakes place. The health care provider model doesnot provide support for training that does not occurin patient care settings. Residency programs in pre-ventive medicine would typically receive little sup-port under this model.

Under Medicare, the health care provider modelis also used to support direct GME costs. DirectGME costs for a single residency program are typi-cally incurred by multiple entities: the program

sponsor, the faculty practice plan affiliated with thesponsoring institution, the hospital that is the pri-mary training site for the residency program, andthe community hospitals and ambulatory provid-ers that serve as additional training sites. Eachprovider’s direct costs for GME depend on its ne-gotiations and arrangements with other entities in-volved in the training program over issues such aswhich party will assume the costs of the resident’ssalary or the opportunity costs of teaching. Typi-cally, when training occurs in a non-hospital set-ting, the non-hospital site incurs some but not allof the training costs. At many community-basedtraining sites, the hospital continues to incur theresident’s salary costs and program administrativecosts. The community-based site incurs site-spe-cific direct costs and may incur indirect costs. Thesite’s costs are funded to some extent by patientcare revenues. The community-based site shouldbe paid directly for indirect teaching costs that aresupported by empirical research. The issue is howfunding might be allocated for direct GME coststhat are not recovered through patient revenues.

PAY PROVIDERS DIRECTLY

Medicare pays a hospital for the time residentsspend in training at that hospital regardless ofwhether the hospital or another hospital or educa-tional institution bears the costs of the residencytraining program. Different rules apply to non-hos-pital settings. Direct GME payments for trainingin non-hospital settings depends on which entitybears substantially all the training costs for theresident’s time at the site, i.e. pays the resident’ssalary and reasonable compensation for teaching.Only the site or the hospital can receive paymentfor the time the resident spends in the non-hospitalsetting; there is no provision for splitting the pay-ment based on the proportion of costs incurred byeach entity. If GME fund payments for direct GMEcosts were made directly to health care providers,there are three basic options for distributing thefunds when community-based training is involved:

• One option would be for the funding to followthe resident. That is, payments for direct GMEcosts would flow to the community-based siteregardless of which entity bears the cost. Thiswould require that virtually all agreements berenegotiated between hospitals and the com-munity-based sites. The policy has appeal as ameans to encourage non-hospital providers tobecome training sites; however, it could becounter-productive to expanding community-based training opportunities. Hospital spon-sors may prefer to retain control over GME


funding and avoid negotiating with commu-nity-based sites by rotating residents to theirown ambulatory clinics. It reduces the com-munity site’s accountability to meet the expec-tations of the program sponsor. It does not pro-vide flexibility to have different remunerationand reward systems for preceptors based onextent of teaching involvement. This optionwould also require an administratively burden-some mechanism to identify and pay eachcommunity-based clinician involved in resi-dency training programs.

• A second option would be to pay pro-rataamounts to the hospitals and the community-based ambulatory sites participating in the resi-dency program based on their relative sharesof direct GME program costs. This optionwould not be as disruptive of historical ar-rangements as the first option; however, itwould require the affiliated hospitals and com-munity-based training sites to agree on theirrelative shares of program costs. The adminis-trative burden to track each agreement and de-termine appropriate payment amounts wouldbe great. A less burdensome variation wouldbe to establish a fixed allocation betweenhospitals and community sites based on nationaldata on their relative shares of direct GME costs.

• A third option would be to pay the entity thatbears substantially all the costs for the com-munity-based rotations. As under currentMedicare policy, it would require negotiationand agreement between the hospital and thecommunity-based sites regarding which entitybears substantially all the costs. It would beless burdensome than the second option be-cause no cost determinations would be re-quired and only one party would be paid forthe resident’s time at the community site. It isthe least disruptive to existing arrangementsbetween the entities participating in residencytraining. For the same reason, it is least likelyto increase community-based training withoutsupplemental policies.

DESIGNATE FUNDS FOR TRAINING INCOMMUNITY SETTINGS

A payment or voucher earmarked for trainingin community settings could reduce the adminis-trative burden of tracking resident time spent incommunity-based sites. This could be accomplishedby requiring the hospital that assumes substantiallyall of the training costs to provide a minimumamount, e.g. 20 percent of the direct GME fundpayment to community training sites. Alternatively,

a voucher could be used to pay for resident timespent in community-based settings.25 The vouchercould be given to the hospital that bears substan-tially all of the training costs. Only a community-based site could redeem the voucher. This portionof the GME fund payment would be forfeited if itwere not used for community-based training. Ei-ther approach would assure funding reaches com-munity-based sites with less administrative burdenthan alternatives to pay the sites directly. Therewould be less local flexibility to determine the ex-tent of training in community settings and an ap-propriate financial arrangement to support the di-rect GME costs of the training sites. However,hospitals would have the flexibility to pay varyingamounts to different community sites within theoverall requirement.

EDUCATION MODELThe education model treats GME costs as an

educational cost as opposed to a patient care cost.It should give educational needs more weight indeciding what and where residency training occursand affords more opportunity to link payment withworkforce objectives. The model is appropriate fordirect GME costs only. Payment would flow to asingle entity that would be accountable for how thefunds were expended. This could be to a medicalschool or the program sponsor. Some have sug-gested the residency program director as the po-tential recipient of the GME funds. However, pay-ing the program director could reduce theopportunities within the sponsoring institution forcross-subsidization of GME programs and for col-lective decision making on workforces issues. Inaddition, the program director does not have theultimate responsibility for the residency program.

INSTITUTIONAL SPONSOR

This alternative pays the GME funds to the in-stitutional sponsor that has ultimate responsibilityfor the GME program. Using the ACGME defini-tion, the sponsor could be a university or medicalschool, hospital, health department or public healthagency, an organized health care delivery system,a consortium or other organization whose primarypurpose is to provide education and/or health careservices (AMA, 1999). If sponsors of the programwere paid, funding could be contingent on meetingspecific educational or workforce objectives. Thereis potential to introduce other elements into the al-location formula, such as program quality. This ismore difficult to do when funds flow directly toproviders.


For hospital-sponsored programs, this optionrepresents the status quo. For programs that aresponsored by educational institutions, the relation-ship between the hospital and the educational in-stitution will determine if paying the program spon-sor would represent a significant change in flow ofdirect GME funds. It would be unnecessarily bur-densome to interrupt existing arrangements whenthe sponsoring institution is satisfied with thosearrangements. These situations could be addressedby allowing the sponsor to designate another en-tity to receive the payments. Payment would bemade only to the sponsoring institution or its des-ignated entity. The program sponsor would be ac-countable for allocating the funds to clinical train-ing sites and for holding the sites accountable forhigh quality training experiences.

MEDICAL SCHOOLS

Paying medical schools would emphasize moststrongly that the GME funds are to support educa-tion rather than patient care. Currently, the univer-sity or the medical center/affiliated hospital spon-sors most residency programs at academic healthcenters. Medical schools sponsor relatively fewprograms. As a result, this option would create amismatch between responsibility for the programand fund accountability at many academic healthcenters. Alignment could require disruptive changesin program sponsorship.

Some residency programs are sponsored bycommunity-based hospitals. Many of these are fam-ily medicine residency programs. Funding throughthe medical schools for these programs would re-quire a major shift in program accountability andfunding.

RESIDENT VOUCHER SYSTEM

The resident voucher system under the educa-tion model differs from the designated funds orvoucher discussed above under the health care pro-vider payment model.

• The resident controls the funds rather than theprogram or hospital that incurs the costs forthe residency program.

• The voucher is not limited to funding for com-munity-based training experiences. Rather, itis for all training experiences.

Resident vouchers are intended to create com-petition among residency programs. However, it isnot clear that the voucher system is significantlydifferent from other payment models. Programsalready compete for residents and GME payments

are based on where the resident trains. The resident’sability to hold the program accountable is limitedonce the resident has selected a program. Thevoucher adds value only if there is a regulatory ap-paratus to determine the number of positions to befunded and which residents should receive fund-ing.

PLANNING MODELUnder the planning model, the GME funds

would flow through a GME planning and coordi-nating body. This is an entity whose primary func-tions would be to assess the health care needs ofthe community and allocate funds based on localworkforce considerations. In contrast to the educa-tion model, the planning model could afford States,payers and consumers a stronger role in how thefunds are allocated to support workforce objectives.The GME funds could be distributed to a broad-based GME consortium or to the States. Under ei-ther option, funding allocation decisions could sup-port local health care needs more readily thandistributions directly to health care providers orprogram sponsors based on national allocation for-mulae.

CONSORTIA

Payment to consortia under the GME fundcould be either a natural outgrowth of paying theprogram sponsor or could result from deliberatepolicies to encourage consortia. Several definitionsof consortium are in use.

• The ACGME defines a consortium as “two ormore organizations or institutions that havecome have come together to pursue commonobjectives.” A consortium may be a sponsor-ing institution if “it is formally established asan on-going entity with a documented com-mitment to GME” (AMA, 1999).

• The Association of American Medical Col-leges and the Center for the Health Professionsat University of California at San Franciscobuilt on the ACGME definition in their 1996survey of 30 consortia. They defined GMEconsortia as “formal partnerships involvingtwo or more separate organizations involvedin graduate medical education, formed to re-organize or strengthen medical education andcharacterized by shared or joint decision mak-ing” (Cox, 1996).

• The BBA defined a consortium as a teachinghospital and one or more of the following: amedical school, another teaching hospital, a


Federally qualified health center, a medicalgroup practice, a managed care entity, or anentity furnishing outpatient services.

The AAMC/UCSF study surveyed 30 multi-disciplinary consortia meeting its definition (anestimated 80 percent of all such consortia) (Cox andDower, 1996). Key findings from the survey included:

• Most consortia included a medical school andteaching hospitals; other health professionsschools, health care providers, or public agen-cies were rarely involved.

• Fifty percent controlled GME expenditures,while 10 percent controlled expenditures forsome members. GME expenditures remainedunder the exclusive control of the individualmembers in 40 percent of the consortia.

• Few had dealt with medical education in acomprehensive way. Sixty percent had consid-ered clinical resources for community-basedtraining sites. However, only 43 percent hadaddressed issues related to developing clini-cian-educators in the community.

• Physician workforce achievements were mixedand largely mirrored the remainder of the aca-demic community.

COGME’s 9th Report, Graduate Medical Edu-cation Consortia: Changing the Governance ofGraduate Medical Education to Achieve WorkforceObjectives (1997) suggested that the consortiumconcept provides the organizational flexibility todraw upon the expertise of the broad group of stake-holders needed to restructure medical education andachieve substantive physician workforce reform.The Council envisioned consortia that would in-clude medical schools, teaching hospitals and com-munity training sites and would promote an inter-disciplinary approach to health care delivery.COGME found that consortia as a group have notinstituted changes that would influence physiciansupply and distribution. COGME concluded thatcurrent consortia do not have the authority and re-sources to achieve workforce objectives. To be ef-fective, consortia need local sponsorship authorityand access to financial resources. Performance todate for consortia suggests that providing paymentonly to consortia would be premature.

The Medicare consortia demonstration projectsauthorized by the BBA are intended to fosterworkforce objectives. Payments would flow to theconsortia. However, the financial rewards are mini-mal. Under the statute’s budget neutrality provi-sions, HCFA cannot pay more to the consortia thatit would otherwise pay to the participating entitiesfor direct GME costs. The demonstration does not

involve IME payments. Additional incentives areneeded to encourage effective models that are re-sponsive to the health care needs of the commu-nity. Consortia of medical schools and teachinghospitals may not be sufficiently broad-based to beresponsive to the health care needs of the commu-nity. Many are limited to a few specialty programs,and do not involve all resident programs in an area.Broad-based consortia of hospitals, community-based providers, educational institutions, purchas-ers and community representatives, such as the Utahconsortium, should be encouraged through devel-opment funds. Beyond developmental support, anypreferential funding for consortia should be tied toachieving specific workforce objectives rather thanto meeting the formal definition of a consortium.

STATES

Many States have established commissions ortask forces related to GME and physicianworkforce. Some are permanent groups with thestructure and expertise that could become physi-cian workforce planning bodies in their State. A1998 assessment by the Center for Health Work-force Studies concluded at least seven States hadpermanent organizations with some authority foradvising on GME policies that could become phy-sician workforce planning groups: New York, Illi-nois, North Carolina, Minnesota, Louisiana, Ari-zona, and Utah (COGME, 1999b).

PERFORMANCE MODELThe performance model would link payment to

achieving specific performance measures or objec-tives. Payment could be formula-driven based onmeeting specific educational or workforce objec-tives. Alternatively, there could be support for spe-cific projects or demonstrations needed to supportinfrastructure development or workforce goals. Theperformance model is more suitable as a mecha-nism for making supplemental payments than as aprimary payment mechanism. Educational qualitymeasures and workforce priorities are not suffi-ciently defined to be used to determine all fund al-locations. Also, if all funding were predicated onmeeting specific performance measures, significantyear-to-year fluctuations in funding could occur thatwould be inconsistent with the need for stable GMEfunding.

INCENTIVE PAYMENTS

The GME funding formulae could provide in-centive payments for meeting specific objectives.These objectives could be in the area of curriculum


content and quality, training opportunities, orworkforce objectives. One approach would be touse an incentive pool along the lines of those em-ployed by New York or Michigan (described in thesection “Current Financing of Graduate MedicalEducation,” in the subsection “Medicaid Supportfor GME”). Limiting the size of the pool, however,provides less stable funding since the amount offunding for individual participants declines as thepercentage of participants meeting the objectivesincreases. The alternative would be fixed incentivepayments that any participant would receive if itmet specific performance criteria. Incentive pay-ments could also be used to reward residents for

career decisions that advance workforce objectivesto improve the specialty and geographic distribu-tion of the physician workforce.

INFRASTRUCTURE SUPPORT

Some funding could be set aside for infrastruc-ture development. Examples of the types of pro-grams that could be funded include the targetedPublic Health Service primary care training grants,support for developing academic-communitypartnerships to serve medically underserved popula-tions, consortia development, and faculty developmentprograms for clinician-educators in the community.


Graduate medical education (GME) shouldmeet community needs and remain currentas new technology and evolving health sys-

tems affect the way care is delivered. The fundingpolicies should provide incentives to support highquality training in both traditional and communitysettings. It should encourage training innovationand improved ways to meet patient needs. Fundsshould be allocated to regional or local levels con-sistent with national workforce priorities. However,Federal support must be sufficiently flexible to al-low some funding allocation decisions to be madeat the local level. At the same time, recipients ofthose funds must be accountable for producing:

• an appropriate number and specialty mix ofphysicians distributed across geographic areasconsistent with current and future nationalhealth care needs; and,

• educating physicians who are well-equipped toprovide high quality, effective and efficient care.

The recommendations outlined below are basedon these principles. The recommendations are madewithin the context of an all-payer GME fund. How-ever, some recommendations could also have ap-plicability to current GME funding policies underthe Medicare program.

RECOMMENDATION 1

CREATE A GME FUND THAT COMBINESFEDERAL FUNDING TO SUPPORT GRADUATEMEDICAL EDUCATION WITH ALL-PAYER FUNDS.

To assure financing policies are consistentacross Federal programs and reflect nationalworkforce priorities, the various Federal fundingstreams for GME that is provided by non-Federalinstitutions (i.e., excluding DoD and DVA) shouldbe combined into a single fund and supplementedwith all-payer funds obtained through a modest sur-charge of private insurance premiums. The GMEfund would include amounts that would otherwisebe paid under current formulae for Medicare fordirect GME and indirect payments to teaching hos-pitals, the Federal portion of Medicaid paymentsthat are implicitly GME payments, and theChildren’s Hospital GME fund. In addition, theHRSA Title VII grants for GME, e.g., primary careresidency training grants, would be included in a set-aside fund for specific workforce goals.

Within the general GME fund, five separateaccounts should be established for:

• Medicare direct GME payments;

• non-Medicare direct GME payments;

• Medicare IME payments;

• non-Medicare IME payments; and,

• targeted payments to support specificworkforce and educational objectives.

The separate Medicare and non-Medicare ac-counts are needed as a transitional measure. Theywould assure full funding on behalf of Medicarepatients if contributions from other payers are notsufficient. Also, since the Medicare funds are cur-rently being paid, changes in the allocation of theMedicare funds should be phased-in or offset byadditional funding from the non-Medicare ac-counts. A transition may not be needed for the non-Medicare funds. To the extent they represent newfunding streams, funds in the non-Medicare ac-counts should be allocated consistent with preferredpolicies from the outset.

While the GME fund would not include fund-ing for residency training in DoD and DVA-spon-sored programs, the Federal budget for those pro-grams should be consistent with the policyobjectives for the GME fund. Residency trainingin these programs has significant impact on the sizeand specialty composition of the physicianworkforce.

a. GME should be broadly supported by all-payers.

Explicit funding for GME should be spreadmore broadly across all sectors of society. Apermanent and stable funding source, such aspremium contributions from all health insur-ance plans, should supplement current Fed-eral funding for GME. In the long run, Medi-care and Medicaid’s contribution to the GMEfund should be proportionate to the percent-age of insured population represented by theirenrollees.

b. Funding from all sources should be sufficientto support high-quality, efficient training ofan appropriately sized physician workforce.

Total aggregate funding should be sufficientto support the efficient training of an appro-priately sized physician workforce. Togetherwith payments from other sources (primarilypatient care revenues and State funds), GMEfunding should be adequate to train the number

Recommendations for GME Financing Reform


of physicians required to meet current andfuture national health care needs. Additionalfunding would not be in the public interestsince it could contribute to a continuing sur-plus of physicians.

In the past, COGME has recommended thatthe total number of physicians entering firstyear residency should not exceed the numberof U.S. medical school graduates in 1993 plus10 percent (COGME, 1994). The Council’s14th Report found that a reduction of 3,386first year positions in 1997-1998 was neededto meet the 110 goal. In view of recent changesin the health care delivery system since its ini-tial recommendations were issued, COGMEplans to review the 110 goal and its target of a50/50 mix of primary care and other specialists.

Most of the increase in the total number ofresidents in recent years is attributable to anincrease in the number of graduates frommedical schools outside the United States.Support should be discontinued for new ex-change visitors (J-1 visa) residents. AsCOGME has previously recommended (1997),exchange visitor residents should be fundedby alternative sources, such as home countryfinancing or foreign aid.

A conceptual framework should be used to es-tablish an appropriate level of Federal supportfor funded residency positions. Because GMEis a joint product with patient care services,patient care revenues cover some direct GMEcosts. Consideration also needs to be given toissues such as maintenance of effort for cur-rent State funding through the Medicaid pro-gram and grant programs and whether all resi-dent activities required for accreditation in anapproved program should be funded. UnderMedicare, only resident time spent in patientcare activities is supported.

RECOMMENDATION 2

IME ACCOUNTS SHOULD PAY HOSPITALSAND OTHER CLINICAL TRAINING SITES ASAPPROPRIATE FOR THE INDIRECT COSTS OFEDUCATIONAL ACTIVITIES.

IME accounts should be created to subsidizehigher patient care costs associated with residencytraining. The funds should be allocated to hospi-tals and, to the extent it is empirically supported,to other clinical training sites that incur indirectteaching costs, including hospital outpatient clin-ics and community-based settings. Initially thereshould be separate accounts for Medicare and non-

Medicare patients in order to assure the indirectcosts for Medicare patients are fully funded. In thelong run, a single account would be appropriate.

a. IME payments should be set at no more thanthe analytically justified level for teachingactivities

Paying more than the analytically justifiedamount would subsidize inefficient providersand give teaching institutions a competitiveedge over non-teaching institutions. For Medi-care inpatient services, MedPAC’s (2000b)current estimate is a 3.1 percent adjustmentfor each 0.1 increment in the resident-to-bedratio after other refinements are made to theMedicare prospective payment system. Basedon this estimate, Medicare IME paymentswould be $1.5 billion lower than the 5.5 per-cent adjustment provided by the BBA. Thedifference could be targeted toward achievingspecific workforce and educational goals (seeRecommendations 4 and 6) or toward support-ing uncompensated care (see Recommenda-tion 8). A transition would be needed to theextent reductions in Medicare IME paymentsare not offset by increases in non-MedicareIME funding.

b. Research is needed to determine the appro-priate IME payment formulae.

Research is needed to refine the Medicare IMEadjustment and to determine the appropriateIME teaching adjustment for non-Medicarehospital inpatients. Medicare’s adjustmentshould be based on the higher costs attribut-able to teaching activities. For non- Medicarepatients, the adjustment should be directed at“leveling the playing field” between teachingand non-teaching hospitals. It does not needto cover the full indirect teaching costs ifteaching hospitals are able to command a pre-mium for quality or specialized services.

Ideally, the IME payment formula should notreflect higher costs indirectly attributable toother teaching hospital missions, e.g. special-ized services, uncompensated care, and re-search. Subsidies for those public goodsshould be directed toward the hospitals pro-ducing them through separate funding streams.Reducing the adjustment to an analyticallyjustified level for teaching would reduce incen-tives to train more physicians than necessary.It would also eliminate confusion betweenfunding for the teaching mission and fundingto support charity care. Higher costs attribut-able to serving low-income patients anduncompensated care costs should be recognized


through a separate funding mechanism (whichwould also distribute payments to non-teach-ing institutions serving low-income patients.See Recommendation 8). Refinements in theIME payment methodology should not reducethe total level of support for hospitals with sig-nificant uncompensated care until specificfunding for such services is provided. Reduc-tions in the IME payment formula should beaccompanied by refinements in the prospec-tive payment system to incorporate better case-mix and severity measurements. An additionaladjustment for research-intensive hospitalsmay also be appropriate.

Additional research is also needed to deter-mine the extent to which there is an indirectteaching effect on costs when resident train-ing takes place in hospital outpatient and non-hospital settings. If empirical research findsthere is an indirect teaching effect on the costsof services provided in ambulatory/communitysettings, the IME account should pay for theseservices as well as inpatient hospital services.

RECOMMENDATION 3

DIRECT GME ACCOUNTS SHOULD PAYPROGRAM SPONSORS OR THEIR DESIGNEESFOR THE DIRECT COSTS OF GRADUATEMEDICAL EDUCATION.

Direct GME costs are educational costs thatshould be supported through payments to the spon-soring institution ultimately responsible for thegraduate medical education program. Payment al-location decisions should be made at the local levelbecause the tremendous variety of existing arrange-ments cannot be accommodated at the nationallevel. By making payments to either the sponsor-ing institution or its designees, the sponsor can de-termine the most appropriate recipient of the fundsbased on local circumstances for a particular pro-gram. For example, a sponsoring institution maydecide to retain maximum control over the fundsand receive them directly, elect to continue histori-cal arrangements having the funds flow through theteaching hospital, or may choose to have a consor-tium distribute the funds. The same election wouldnot need to apply to each program sponsored bythe institution. Regardless of which entity receivedthe funding, the sponsoring institution would be ac-countable for the funds being expended to supporta high quality training program with the appropri-ate balance of hospital and community-based train-ing experiences.

a. There should be written agreements betweenthe program sponsor and training sites in-dicating the sponsor is assuming substan-

tially all of the training costs and describinghow GME payments will be allocated

The program sponsor or its designee must as-sume all or substantially all of the direct costsof operating the residency program as a con-dition of receiving direct GME payments.Written agreements should be required be-tween the sponsoring institution and clinicaltraining sites to formalize the negotiation proc-ess and to increase accountability for thefunds. The agreements should detail how thedirect GME funds will be allocated betweenthe sponsor and the training site, identifywhich entity will pay resident salaries andfringe benefits, and specify teaching physiciancompensation arrangements for supervisingresidents. The goal is to strengthen the nego-tiating position of community–based siteswithout jeopardizing long-standing relation-ships between academic institutions and com-munity training sites. A sponsoring hospitalmay have a disincentive to rotate residents tocommunity-based training sites if all directGME funds automatically follow the residentto a community-based training site throughdirect payments from the GME fund or avoucher system.

b. Model agreements and information on directGME costs should be made available to fa-cilitate equitable agreements between thesponsor and the sites

Local circumstances should determine how di-rect GME payments are allocated to teachingsites. However, benchmarking informationshould be provided to facilitate the negotia-tion process, including:

• breakdown of GME payments into threecomponents based on average direct GMEcosts: resident salaries and related costs,teaching physician compensation, and anadministrative and overhead cost compo-nent.

• benchmarks for teaching physician compen-sation and the added time per teaching ses-sion when residents are present in commu-nity-based practices on short-term rotationsand on an on-going basis; and,

• model agreements between institutionalsponsors and community-based sites.

c. Require separate reporting of resident timespent in inpatient hospital, hospital outpa-tient and community settings

At present, there is no formal accounting forthe time residents spend in each type of train-ing site. Standard definitions should be


developed to distinguish hospital outpatientsettings from community settings. Communitysettings should be broadly defined to includeboth hospital-operated and community-basedsites that are representative of the environmentin which residents will eventually practice.The determining characteristics are the pro-cesses of care rather than proximity to thehospital or provider ownership. Communitysettings address the care of the individual pa-tient in the context of the population of whichthe patient is a member. They teach residentsto deliver culturally effective care to an ethni-cally and racially diverse population.

RECOMMENDATION 4

ESTABLISH A NATIONAL AVERAGE PERRESIDENT PAYMENT FOR DIRECT GMECOSTS.

The base payment for direct GME costs shouldvary only for differences in the cost of living acrossgeographic areas. For Medicare payments, thereshould be a transition from the hospital-specific perresident amounts to the national per resident pay-ment. The length of the transition will depend onadditional payments for non-Medicare patients.These can help compensate for any reductions inMedicare payments. At the end of the transition,separate Medicare and non-Medicare accountswould no longer be necessary. Higher paymentsmay be appropriate for training in community-basedsettings. In addition, there should be an incentivepayment for programs that meet specific workforceor educational objectives.

a. Base total direct GME payments on the net costsof supporting an appropriately sized workforce.

Ultimately, total direct GME funding shouldbe based on the net costs of educating an ap-propriately sized physician workforce. Estab-lishing a fixed payment per resident shouldprovide incentives for efficiency in the edu-cational process. However, the costs of effi-ciently delivering high quality GME and theextent to which these costs are offset by pa-tient care revenues has not been determined.As an interim policy, either the average perresident amounts or average GME costs perresident could be assumed to represent thetotal costs of an efficient program. Total costsbased on the FY1997 average per residentamount updated for inflation and the 110 per-cent target are estimated at $6 billion forFY2000.26 A lower funding amount would beappropriate since the per resident amounts donot take into account patient care revenues at-tributable to GME.

b. Provide higher payments for training in com-munity settings

When training occurs in a community setting,the sponsoring institution continues to incursome supervisory physician and overheadcosts. The community setting incurs some di-rect GME costs as well (for example, to com-pensate the community physician for teach-ing) even if the sponsoring institutioncontinues to pay resident salaries and fringebenefits. As a result, total GME costs may behigher when residents rotate to community-based settings than when they remain in hos-pital-based settings. A higher payment fortraining in community-based settings wouldbe appropriate if the net total costs (after tak-ing any additional patient care revenue intoaccount) are higher in the community-basedsettings. Research is needed to determinewhether this is the case. To counter any disin-centive that might currently exist for commu-nity-based rotations, a temporary policy mightbe to increase the component of the per residentamount attributable to teaching physician com-pensation by a fixed percentage, e.g. 25 percent.The higher payment could apply in all commu-nity settings even though the rationale for thepayment is primarily applicable to settings offthe hospital premises. This would provide anincentive for hospitals to turn training in ambu-latory clinics into experiences that are more rep-resentative of community physician practices.

c. Provide incentive payment for meeting spe-cific workforce and educational objectives.

In addition to the base per resident payment,there should be an incentive payment for meet-ing specific workforce and educational objec-tives. Programs that meet one or more of theobjectives would be eligible for a bonus onthe national average base payment. The incen-tive payment should be established as a fixedpayment rather than a pool so that the ben-efits do not erode as additional programs meetthe objectives and qualify for payment. Thebonus payment could be awarded based on:

• Participation in a broad-based consortia ofthe sponsoring institution(s) for residencyprograms in an area, hospitals and commu-nity providers participating in GME activi-ties, and community representatives. The con-sortia would have to be designated by thesponsoring institutions to receive all directGME funds. Bonus payments would be madeif the consortia has a formal process to iden-tify the health care needs of the community,engage in workforce planning, and promotecommunity-based training opportunities;


• Number of graduates that provide signifi-cant amounts of care to medically under-served populations;

• Percentage of time residents spend provid-ing care to medically underserved popula-tions; and,

• Quality of the residency program.

d. Research is needed to understand variationin direct GME costs by specialty and setting.

Most research regarding residency trainingcosts was conducted at a limited number ofsites and before the growth of managed sys-tems of care. A better understanding of differ-ences in the net costs of training across resi-dency programs and training sites is neededto refine the payment allocation methodology.A generic financial model should be used toexamine systematically issues such as:

• Whether there are significant differences inthe amount of teaching physician involve-ment between primary care and non-pri-mary care residency programs and betweeninitial residencies and fellowships;

• Whether there are significant differences inimpact on clinical productivity and net phy-sician practice revenue when teaching oc-curs in ambulatory settings relative to inpa-tient settings and between initial residenciesand fellowships;

• How direct GME costs are affected by thepresence of students in other health profes-sions at the training site and by residentsteaching medical students; and,

• Factors that affect the efficiency and qual-ity of the educational process.

RECOMMENDATION 5

CONTINUE THE BALANCED BUDGET ACTOF 1997 LIMITS ON THE NUMBER OF RESI-DENTS WITH MODIFICATIONS.

In concept, the Balanced Budget Act of 1997limits on the residents that will be recognized byMedicare are consistent with the goal of reducingthe future physician workforce and should be car-ried over to eligibility for payments under the GMEfund. However, hospital-specific limits are not anappropriate long-term way to deal with physiciansupply issues. The limits hamper a programdirector’s ability to move residents among hospitalprograms for educational reasons. In geographicareas with physician shortages, the limits precludeexpansions in needed residency programs.

a. Modify the caps to apply to sponsoring insti-tutions rather than hospitals.

The sponsoring institution is to be held ac-countable for educational outcomes andworkforce objectives. Therefore, the limitsshould apply to the number of residents in theprograms sponsored by the institution rather thanthe hospitals serving as training sites for the pro-gram. This will provide the flexibility to moveresidents between hospitals and other settings.Consortia that meet certain workforce planningobjectives should be able to work under an ag-gregate limit for multiple sponsoring institutions.

Applying the limit to sponsoring institutionsis consistent with making direct GME pay-ments to sponsoring institutions. However, itcomplicates IME payments to individual hos-pitals. One option would be to apply no limitfor IME as long as the sponsoring institutionswhose residents are training at the hospital areunder their caps. An adjustment would beneeded only if the total number of residentsin programs at the sponsoring institution(s) ex-ceeded an aggregate cap. The three-year roll-ing average and the one-year cap on the resi-dent-to-bed ratio should continue to apply. Therolling average provides a form of transitionpayments to hospitals that reduce the numberof residents and slows the recognition of newresidents in the IME count.

b. Include residents in non-hospital settingsregardless of who paid the resident’s salaryin the 1996 base year count.

Residents who were working in non-hospitalsettings were not included in a hospital’s 1996base year count for Medicare IME payments.They were included in the direct GME countonly if the hospital incurred substantially allof the training costs. The limit applicable to asponsoring institution should be adjusted toinclude all resident time in non-hospital settingsregardless of who paid the resident’s salary.

e. Allow adjustments in the Balanced BudgetAct of 1997 limits to improve the distributionof physician workforce.

Further research is needed to understand theimpact of the Balanced Budget Act of 1997limits and to develop policies that will resultin a better geographic balance in the physi-cian workforce while encouraging an overallreduction in the number of physicians. Appro-priate indicators of adequate distribution byspecialty are needed. As an interim measure,the limits should not apply to rural residencytraining tracks if their graduates practice


predominately in rural areas. In addition, thelimits should not apply to primary care resi-dency programs whose graduates practice pre-dominately in those States with low physician-to-population ratios.

RECOMMENDATION 6

ESTABLISH AN ACCOUNT FOR FUNDINGSPECIAL PROJECTS AND PROGRAMS DI-RECTED AT BUILDING HIGH-QUALITY COM-MUNITY-BASED TRAINING CAPACITY ORACHIEVING SPECIFIC WORKFORCE GOALS.

At least 10 percent of the GME fund should beset aside to support specific projects and programsdirected at building high-quality community-basedtraining or achieving specific workforce priorities.The types of projects and programs that should befunded include:

• Primary care residency program grants;

• Faculty development grants to support train-ing of community clinician teachers;

• Information technology infrastructure devel-opment to link patient care records at teach-ing hospitals and community sites within or-ganized systems of care;

• Incentive programs to reward residents that fo-cus their practice on medically underservedpopulations;

• Transition funds to cover residency replace-ment costs in hospitals with high uncompen-sated care patient loads; and,

• Demonstration projects involving develop-ment of broad-based consortia.

RECOMMENDATION 7

MODIFY THE MEDICARE RULES RELATEDTO TEACHING PHYSICIANS TO EMPHASIZETHE TEACHING PHYSICIAN’S OVERALL RE-SPONSIBILITY FOR THE MANAGEMENT OFTHE PATIENT’S CARE AND TO REDUCE THEIMPORTANCE OF DOCUMENTATION.

The Medicare rules pose two challenges tograduate medical education: 1) the supervision rulesmake it more difficult for residents to become pro-gressively independent, and 2) the documentationrequirements detract from the amount of time avail-able for teaching and resident supervision. There issome evidence that the rules may adversely affect thewillingness of community physicians to participatein teaching programs. The rules should be revised toaddress these concerns.

a. Establish different rules for residents in fel-lowship programs

HCFA’s rationale for its teaching physicianrules rests on: 1) making payment only whenthere is an identifiable physician service to anindividual patient and 2) avoiding duplicatepayment for the physician’s supervisory time.Duplicate payment should not be an issue withresidents who are beyond their initial residencyperiod. These residents count as only .5 FTE andresident salaries and fringe benefits compriseonly 43 percent of the per resident amount.

The rules on teaching physician supervisionshould be revised for residents who are be-yond their initial residency program to permitMedicare billing if:

• the teaching physician is immediately avail-able;

• reviews with each resident during or imme-diately after the visit the patient’s medicalhistory and care; and,

• documents his or her participation in thereview and direction of services.

Accreditation standards for the residency pro-gram should be relied upon to determine is-sues regarding supervision requirements forspecific services and resident-to-preceptor ratios.

An alternative would be to allow residents be-yond their initial residency period to bill forservices furnished in the hospital if an elec-tion is made to forego a direct GME paymentfor the resident’s time.

b. Evaluate the impact of the teaching physi-cian rules

The issue of when physician billing is appro-priate for care provided by residents is not lim-ited to the Medicare program. It will remainrelevant under an all-payer fund. There is aneed to evaluate formally the administrativeand teaching burden associated with the cur-rent Medicare rules and their impact on thequality of clinical training and patient care.Particular attention should be paid to the ef-fect in hospital ambulatory clinics and com-munity-based settings where there is a lowresident-to-preceptor ratio.

c. Develop clear and reasonable documentationrequirements

There is need for additional guidance and com-mon understanding of what constitutes ad-equate documentation of a teaching physi-cian’s participation in the care of patientsinvolving residents. HCFA should work with


the academic physician community and the Of-fice of Inspector General to develop reasonablestandards that do not compromise high qual-ity clinical education. The standards shouldprovide a reasonable means for documentingthe teaching physician’s involvement in thecare of the patient and assuring appropriatepayments without imposing undue adminis-trative burden. They should be tested in a va-riety of teaching settings and specialty pro-grams before implementation.

RECOMMENDATION 8

PROVIDE ADDITIONAL SUPPORT FORHOSPITALS AND COMMUNITY-BASED TRAIN-ING SITES THAT SERVE A DISPROPORTION-ATE SHARE OF LOW INCOME PATIENTS.

In the absence of national health insurance,“safety net” providers should be provided with ad-

ditional funding to cover uncompensated care costs.Major teaching hospitals provide substantial un-compensated care. Faculty practice plans also fur-nish charity care. Uncompensated care is not aneducational cost. However, it affects the trainingsite’s ability to provide high quality educationalexperiences. More importantly, teaching institutionsthat furnish high amounts of uncompensated carerely on current GME funding to support their char-ity care. As changes are made in the IME paymentmethodology, the current level and distribution ofDSH payments should be examined to assure thefunds are well targeted to subsidize uncompensatedcare. The subsidies should apply to hospitals forboth inpatient and outpatient services and tocommunity-based providers. Without additionalsupport, GME is not sustainable in community-based training sites with a high volume of uncom-pensated care. These sites cannot generate the pa-tient care revenues needed to support theireducational activities.


BIBLIOGRAPHY

American Association of Colleges of OsteopathicMedicine, American Medical Association,American Osteopathic Association, Associationof Academic Health Centers, Association ofAmerican Medical Colleges, and National Medi-cal Association. Consensus Statement on thePhysician Workforce, 1997.

American Board of Internal Medicine Foundation.Resident Survey 1997-98. National Study ofGraduate Education in Internal Medicine, 1998a.http://www.abim.org, last accessed June 5, 2000.

American Board of Internal Medicine Foundation(1998b). “Summary of Key Findings from theInternal Medicine Residency Training Survey1997-1998,” National Study of Graduate Educa-tion in Internal Medicine. http://www.abim.org/nasgim/key9798.htm, last accessed June 5, 2000.

American Medical Association. “AMA GraduateMedical Education Directory 1998-1999.” Wash-ington, D.C.: American Medical Association,1999.

American Medical Association. “AMA FREIDAOnline Reporting System.” Washington, D.C.:American Medical Association, 2000. http://www.ama.org, last accessed May 2000.

American Osteopathic Association. “OsteopathicGraduate Medical Education FAQ.” http://www.am-osteo-assn.org, last accessed March 10,2000.

Anderson G.F. “What Does Not Explain the Varia-tion in the Direct Costs of Graduate MedicalEducation,” Academic Medicine, Volume 71,Number 2, pp. 156-165, 1996.

Anderson G.F. Greenberg G. and Lisk C. “AcademicHealth Centers: Exploring a Financial Paradox,”Health Affairs, Volume 18, Number 2, pp. 156-165, March/April 1999.

Association of American Medical Colleges. “FactSheet: Trends in Medical School Faculty Sala-ries,” Washington, D.C.: Association of Ameri-can Medical Colleges, December 1999a. http://www.aamc.org, last accessed May 2000.

Association of American Medical Colleges.“AAMC Press Release.” Washington, D.C.: As-sociation of American Medical Colleges, May14, 1999b. http://www.aamc.org, last accessedMay 2000.

Association of American Medical Colleges. “FactSheet: Impact of 1999 Balanced Budget Act of1997 Refinement Act on DGME PaymentChanges for Teaching Hospitals.” Washington,D.C.: Association of American Medical Colleges,January 2000. http://www.aamc.org, last ac-cessed May 2000.

Blewett L.A. and Weslowski V. “New Roles for Statesin Financing Graduate Medical Education:Minnesota’s Trust Fund,” Health Affairs, Volume19, Number 1, pp. 248-252, January/February 2000.

Boex J.R., Boll A., Hogan A. Meservey P., RubinR., and Seifer S. “Assessing Ambulatory PrimaryCare Training: Costs, Methods, and Quality.”Bureau of Health Professions, Health Resourcesand Services Administration, with additionalsupport from the W.K. Kellog Foundation andthe New York State Department of Health, No-vember 1997.

Boex J.R., Blacklow R., Boll A., Fishman L.,Gamliel S., Garg M., Gilchrist V., Hogan A.,Meservey P., Pearson S., Politzer R., Veloski J.J.“Understanding the Costs of Ambulatory Careand Training,” Academic Medicine, Volume 73,Number 9, September 1998.

Breaux-Thomas Proposal. “Building a Better Medi-care for Today and Tomorrow.”National Biparti-san Commission on the Future of Medicare.March 16, 1999 version. http:rs9.loc.gov/medi-care/bbmtt, last accessed October 10, 2000.

Bureau of Health Insurance, Intermediary Letter379, 1969.

Calman N., personal communication, February 2000.

Commonwealth Fund Task Force on AcademicHealth Centers. “Leveling the Playing Field: Fi-nancing the Mission of Academic Health Cen-ters.” The Commonwealth Fund, 1997.

Commonwealth Fund Task Force on AcademicHealth Centers. “From Bench to Bedside: Pre-serving the Research Mission of AcademicHealth Centers.” The Commonwealth Fund, April1999.

Congressional Budget Office. “Budgetary Implica-tions of the Balanced Budget Act of 1997,” De-cember 1997. http://www.cbo.gov/docs.shtml,last accessed May 2000.

Bibliography and Footnotes


Congressional Budget Office. “Conference Reporton H.R. 3194 Consolidated Appropriations Act,2000,” December 8, 1999. http://www.cbo.gov/docs.shtml, last accessed May 2000.

Congressional Budget Office. “Budget Options,Annual Report to the House and Senate Com-mittees on the Budget.” March 2000. http://www.cbo.gov/docs.shtml, last accessed May2000.

Council on Graduate Medical Education. “NinthReport: Graduate Medical Education Consortia:Changing the Governance of Graduate MedicalEducation to Achieve Physician Workforce Ob-jectives.” Rockville, Maryland: Department ofHealth and Human Services, June 1997.

Council on Graduate Medical Education. “EleventhReport: International Medical Graduates, thePhysician Workforce, and GME Payment Re-form.” Rockville, Maryland: Department ofHealth and Human Services, March 1998.

Council on Graduate Medical Education. “Thir-teenth Report: Physician Education for a Chang-ing Health Care Environment.” Rockville, Mary-land: Department of Health and Human Services,March 1999a.

Council on Graduate Medical Education. “Four-teenth Report: COGME Physician WorkforcePolicies: Recent Developments and RemainingChallenges in Meeting National Goals.”Rockville, Maryland: Department of Health andHuman Services, March 1999b.

Cox M. and Dower C. “Graduate Medical Educa-tion Consortia: A National Survey.” AmericanAssociation of Medical Colleges and Center forthe Health Professions. University of California:San Francisco, 1996.

Cunningham M., personal communication, March2000.

Davis P. H. “The Effects of the Balanced BudgetAct of 1997 on Graduate Medical Education: ACOGME Review.” Council on Graduate Medi-cal Education Resource Paper. U.S. Departmentof Health and Human Services, March 2000.

Delbanco T.L and Calkins D.R. “Costs and Financ-ing of Ambulatory Medical Education,” Journalof General Internal Medicine, Volume 3, Sep-tember 1988.

Department of Health and Human Services, “Of-fice of Inspector General; Medicare Program;Prospective Payment for Hospital OutpatientServices,” Federal Register, Volume 65, Number68, 18434-18820, April 7, 2000a.

Department of Health and Human Services, “Medi-care Program: Changes to the Hospital InpatientProspective Payment Systems and Fiscal Year2001 Rates,” Federal Register, Volume 65, Num-ber 148, 47504-47211, August 1, 2000b.

Department of Veterans Affairs. “Graduate Medi-cal Education Realignment, Strategies for Meet-ing VHA’s Goals for Graduate Medical Educa-tion Realignment,” Office of AcademicAffiliations. http://www.va.gov/OAA/docs/G M E D o c s / G e n e r a l / A r t i c l e _ G M E _PCSTRAT3.asp, last accessed May 29, 2000.

Department of Veterans Affairs, Office of AcademicAffilitations. “Strategies for Meeting VHA’sGoals for Graduate Medical Education Align-ment.” September 22, 1997. www.va.gov/OAA/Docs/General/Article, last accessed May 29,2000.

Desmarais H.R. and Hash, M.M. “Financing Gradu-ate Medical Education: The Search for NewSources of Support,” Health Affairs, Volume 16,Number 4, pp. 48-63, July/August 1997.

Diamond H.S., Fitzgerald L.L., Day R. “An Analy-sis of the Cost and Revenue of an ExpandedMedical Residency,” Journal of General Inter-nal Medicine, Volume 8, November 1993.

Fishman L. E. Medicare Payments with an Educa-tion Label: Fundamentals and the Future. Asso-ciation of American Medical Colleges: Washing-ton, D.C., 1996.

Flanagan T., Mitchner B., Weyl-Feyling D., andLaros R.K. “The Cost of Teaching Residents Out-patient Obstetrics and Gynecology in a Univer-sity Medical Center,” Obstetrics and Gynecol-ogy, Volume 86, Number 6, December 1995.

Franks, R.D., personal communication, November1999.

Gamble J.G., and Lee R. “Investigating WhetherEducation of Residents in a Group Practice In-creases the Length of the Outpatient Visit,” Aca-demic Medicine, Volume 66, Number 8, August1991.

Gavett J.W., and Mushlin A.I. “Calculating theCosts of Training in Primary Care, Medical Care,Volume 24, pp. 301-312, 1986.

General Accounting Office. MEDICARE: Concernswith Physicians at Teaching Hospitals (PATH)Audits. Report to the Chairman, Subcommitteeon Health, Committee on Ways and Means,House of Representatives. GAO/HEHS-98-174,July 23, 1998. http://www.gao.gov, last accessedMarch 2000.


Hanft R., and White C. “Financing Ambulatory CareTraining of Primary Care Physicians: Sites, Costsand Options for the Future.” June 1997.

Health Care Financing Administration. HospitalCost Report Information System. 1997.

Health Care Financing Administration. “MedicareHealth Insurance Manual.” 14, Section 15016ff,Revision 1566 (5/97). http://www.hcfa.gov/pubforms, last accessed March 1, 2000.

Health Care Financing Administration. “ManagedCare Penetration Rates by States as of Decem-ber 31, 1998.” http://www.hcfa.gov/medicaid/omcpr98.htm, last accessed June 2000.

Health Care Financing Administration. “Medicare:A Profile.” July 2000. http://www.hcfa.gov, lastaccessed October 10, 2000.

Health Care Financing Administration. Office of theActuary, personal communication, January 2000.

Health Resources and Services Administration,personal communication from American Acad-emy of Family Practice (AAFP), January 31,2000.

Henderson T.M. “Medicaid Support for GraduateMedical Education: Should There Be a NationalPolicy?” A Policy Brief for The Center for theHealth Professions, University of California, SanFrancisco, August 1998.

Henderson T.M. “Funding of Graduate MedicalEducation by State Medicaid Programs.” Asso-ciation of American Medical Colleges, April1999.

Henderson T.M. “Medicaid’s Role in FinancingGraduate Medical Education,” Health Affairs,Volume 19, Number 1, pp. 221-229, January/February 2000.

Holmes D., Director, Plan Administration and Cus-tomer Service, Department of CommunityHealth, State of Michigan, personal communi-cation, 2000.

Institute of Medicine. “On Implementing a NationalGraduate Medical Education Trust Fund.” Com-mittee on Implementing a National GraduateMedical Education Trust Fund, Division ofHealth Care Services. Washington, D.C.: Na-tional Academy Press, 1997.

Institute of Medicine. “Primary Care Physicians:Financing their GME in Ambulatory Settings.”A Report of a Study by a Committee of the Insti-tute of Medicine, Division of Health Care Serv-ices. Washington, D.C.: National Academy Press,1989.

Institute of Medicine, Committee on the ChangingMarket, Managed Care, and the Future Viabilityof Safety Net Providers. “America’s Health CareSafety Net. Intact but Endangered,” Ed. byMarion Ein Lewin and Stuart Altman. Washing-ton, D.C.: National Academy Press, 2000.

Jones T.F., Culpepper L., and Shea C. “Analysis ofthe Cost of Training Residents in a CommunityHealth Center,” Academic Medicine, Volume 70,Number 6, June 1995.

Jones T.F. “The Cost of Outpatient Training of Resi-dents in a Community Health Center,” FamilyMedicine, Volume 29, Number 5: 347-52, May1997.

Journal of the American Medical Association.“Graduate Medical Education.” Appendix II. Vol-ume 282, No. 9, September 1, 1999.

Karp S., Langschwager L., Gamliel S., andFernandez, S. “The Need for ComprehensiveData on Educational Affiliations between Aca-demic Health Centers and Managed Care Orga-nizations,” Academic Medicine, Volume 72,Number 5, May 1997.

Kellogg Foundation, “Community Partnerships: ASummary of National Public Policy Objectives.”# 489, 8/99. See also “Community Partnershipsin Graduate Medical and Nursing Education,”Item #487.

Kosekoff J., Brook R., Fink A., Kamberg C., RothP., Goldberg G., Linn L., Clark V., Newhouse J.,Delbanco T., et al. “Providing Primary GeneralMedical Care in University Hospitals: Efficiencyand Cost, Annals of Internal Medicine, Volume107, Number 3, pp.399-405, September 1987.

Lake T., Gold M., Hurley R., Sinclair M., andWaltman S. “Health Plans’ Selection and Pay-ment of Health Care Providers, 1999.” Final re-port by Mathematica Policy Research, Inc. to theMedicare Payment Advisory Commission. Wash-ington, D.C.: May 16, 2000.

Lave J. R., “The Cost of Graduate Medical Educa-tion in Outpatient Settings, a Report of a Studyby a Committee of the Institute of Medicine,Division of Health Care Services.” Paper com-missioned by the Committee to Study Strategiesfor Financing Graduate Medical Education forPrimary Care Physicians in Ambulatory Settings.Washington, D.C.: National Academy Press, pp.144-172, 1989.

Levinsky N.G., “A Survey of Changes in the Pro-portions of Ambulatory Training in InternalMedicine Clerkships from 1986-1987 to 1996-1997,” Academic Medicine, Volume 73:1114-1115, 1998.


Levit K., Cowan C., Lazenby H., Sensenig A.,McDonnell P., Stiller J., Martin A., and the HealthAccounts Team, “TRENDS: Health Spending In1998: Signals Of Change,” Health Affairs, Vol-ume 19, Number 1, pp. 124-132, January/Febru-ary 2000.

Medicare Intermediary Letter 372 (1969).

Medicare Payment Advisory Commission. “Re-thinking Medicare’s Payment Policies for Gradu-ate Medical Education and Teaching Hospitals.”Washington, D.C., August 1999.

Medicare Payment Advisory Commission. “Reportto the Congress: Medicare Payment Policy.”Washington, D.C., March 2000a.

Medicare Payment Advisory Commission. “Reportto the Congress: Selected Medicare Issues.”Washington, D.C., June 2000b.

MEDSTAT Group, ”Comparison of Admissions toAcademic and Non-Academic Medical Centersfor Capitated and Fee-for-Service Plans,” May29, 1997. http://www.aahp.org, last accessed June2000.

Meyer G.S. and Blumenthal D. “TennCare andAcademic Health Centers: The Effect of PublicSector Health Care Reform in Tennessee.” Un-published manuscript, March 1999.

Minnesota Department of Health (MDH) website.http://health.state.mn.us/divs/hpsc/hep/merc, lastaccessed May 2000.

National Bipartisan Commission on the Future ofMedicare. Minutes from the March 16, 1999meeting. . http:rs9.loc.gov/medicare, last ac-cessed October 10, 2000.

New York State Department of Health (NYSDOH),Graduate Medical Education (COGME) website.http://www.health.state.ny.us/nysdoh/gme/main.htm, last accessed June 2000.

Partnerships for Quality Education website. http:/www.pqe.org, last accessed in April 2000.

Peinado S. and Eisenberg J.M. “Financing Gradu-ate Medical Education in Primary Care: Optionsfor Change,” The Journal of Family Practice,Volume 31, Number 6: 637-644, 1990.

PEW Health Professions Commission. “Beyond theBalanced Budget Act of 1997: StrengtheningFederal GME Policy.” University of California, SanFrancisco, Center for Health Professions, 1998.

Philibert I. “Medical Education In Ambulatory CareSettings An Annotated Bibliography of SelectedWorks,” Second Edition. Washington, D.C.: As-sociation of American Medical Colleges, March1999.

Prospective Payment Assessment Commission.“Medicare and the American Health Care Sys-tem: Report to the Congress,” June 1992

Prospective Payment Assessment Commission.“Report and Recommendations to the Congress,”March 1, 1997.

Public Law 106-113, “Balanced Budget RefinementAct of 1999.”

Public Law 105-33, “Balanced Budget Act of 1997:Conference Report to Accompany H.R. 2015.”House of Representatives, 105th Congress, 1stSession, Report 105-217.

Rabb H. Letter from the HHS General Counsel tothe American Medical Association and the As-sociation of American Medical Colleges, July 11,1997.

Retchin S. M. “Three Strategies Used by AcademicHealth Centers to Expand Primary Care Capac-ity,” Academic Medicine, Volume 75, Number 1,pp.15-22, January 2000.

Ricer R.E., Filak A.T., David A.K. “Determiningthe Costs of a Required Third-Year Family Medi-cine Clerkship in an Ambulatory Setting,” Aca-demic Medicine, Volume 73, Number 7, pp. 809-811, July 1998.

Robinson J. C. “The Future of Managed Care,”Health Affairs, Volume 18, Number 2, pp. 7-24,March/April 1999.

Salsberg E. “State Strategies for Financing Gradu-ate Medical Education,” Center for HealthWorkforce Studies, School of Public Health,SUNY at Albany, August 1997.

Stern R.S., Jennings M., Delbanco T.L., Dorsey J.L.,Stoeckle J.D., and Lawrence R.S. “GraduateEducation in Primary Care: An Economic Analy-sis,” The New England Journal of Medicine, Vol-ume 297, Number 12, September 22, 1977.

U.S. Congress. House of Representatives, HouseReport No. 213, 89th Congress, 1st Session 32,1965.


U.S. Congress. “Ways and Means Committee Re-port.” House of Representatives, House ReportNumber 98-25, March 4, 1983.

U.S. Congress. “Medicare and Health CareChartbook.” House of Representatives, Commit-tee on Ways and Means, 106th Congress, 1stSession, WMCP:106-4, March 17, 1999.

University of California. “Sixth Report: ChangingDirections in Medical Education. 1999 Updateon Systemwide Efforts to Increase Training ofGeneralists.” University of California, July 1999.

Utah Health Policy Commission website. http://www.hlunix.hl.state.ut.us/hpc/meded, last ac-cessed May 2000.

FOOTNOTES

1 Based on 83,422 residents and $71,468 averageper resident cost incurred by hospitals in FY1997,full funding would require $6 billion.

2 In 1994, academic health center’s market shareof HMO patients was 80 percent of their marketshare of other privately insured patients. (Com-monwealth, 1997). A study of 1994 private sec-tor health data found that severity and geographicadjusted hospital payments per admission in highmanaged care penetration areas were signifi-cantly lower for managed care plans than fee-for-service plans. For teaching hospitals, managedcare payments were about 30 percent lower thanthe fee-for-service payments (MEDSTAT, 1997).

3 Margins are defined as revenues minus expensesdivided by revenues.

4 The $33.6 billion includes provisions affectingPPS updates, outlier policy, transfer policy, baddebt and outpatient services. The extent to whichthe Medicare cuts are offset by the carve-outamounts will depend on whether managed careplans take account of the carve-out amounts innegotiating payments with teaching hospitals.

5 For example, if a hospital’s per resident amountwas $100,000 and the hospital had 10 residentsand 35 percent Medicare utilization, thehospital’s Medicare GME payment would be$350,000 ($100,000 x 10 residents x .35).

6 The Omnibus Reconciliation Act of 1993 pro-vided for a two-year freeze on per residentamounts for programs other than those in pri-mary care and OB-GYN. In addition, the Bal-anced Budget Refinement Act of 1999 affectedthe updates for per resident amounts that exceed140 percent of the national average.

7 The analysis excludes 428 hospitals that reportedfewer than 10 residents. These residency pro-grams are more likely not to incur certain costs(e.g. teaching physician compensation for rotat-ing residents) and could distort a comparison ofthe components of GME costs.

8 Pearson’s correlation coefficient was .51; prob >lRl .0021.

9 HCFA calculated the FY1997 average per resi-dent amount (standardized for wage differentials)to be $68,684. The FY 1997 70 percent floor is$47,925 and the 140 percent ceiling is $95,850.

10 Information on direct GME costs by specialtyprograms is not readily available. For example,only aggregate direct GME costs are reported onthe Medicare cost report. However, the residentcount is separately provided for residents in pri-mary care and OB-GYN programs and in otherprograms.

11 Pearson’s correlation coefficient for the per resi-dent amount =.17; prob > lRl = .00004. The co-efficient for GME costs = .14; prob > lRl =.0037.

12 The low-income patient percentage defined isconsistent with the formula used to distributeMedicare DSH payments. It is the sum of thehospital’s percentage of inpatients who are Medi-care patients who are entitled to SSI plus the per-centage of inpatients who are non-Medicare pa-tients who are entitled to Medicaid.

13 The Balanced Budget Act of 1997 reduced thepayment from 7.7 percent in FY1998 to 6.0 per-cent in FY2000, and 5.5 percent in FY2001 andsubsequent years. The Balanced Budget Refine-ment Act of 1999’s modification in the phasedreduction increases payments to teaching hospi-tals by $600 million during FY2000-FY2004relative to the Balanced Budget Act of 1997.

14 Medicare’s indirect teaching adjustment for capi-tal costs was established empirically, based onthe effect of teaching on total Medicare inpatientoperating costs per discharge. The formula isbased on the ratio of residents to average dailycensus and is therefore not directly comparableto the adjustment for operating costs. The adjust-ment increases payment by [e raised to the powerof (.2822 times the ratio of residents to averagedaily census)] minus 1.


15 MedPAC’s June 2000 report recommends thattime spent in both inpatient and outpatient set-tings be counted in determining the refined Medi-care teaching hospital adjustment.

16 Medicare managed care enrollment varies sig-nificantly across geographic areas. For example,24 percent of beneficiaries were enrolled in man-aged care plans in California in 1998 comparedto fewer than 7 percent in New York. Managedcare penetration tends to be concentrated in ur-ban centers where teaching hospitals are also con-centrated. In 1998, 22.5 percent of Medicare ben-eficiaries residing in urban centers (generally thecore cities of metropolitan areas) were enrolledin risk-based managed care plans compared to11.7 percent in outlying urban areas (U.S. Con-gress, 1999).

17 The $4 billion in additional payments assumesmanaged care plans do not adjust for the carve-out by reducing their payments.

18 Section 541 of the Balanced Budget RefinementAct of 1999 provides for up to $60 million innursing and allied health education direct pay-ments for Medicare managed care enrollees anda proportionate reduction in direct GME pay-ments under the GME carve out. The provisionis effective January 1, 2000.

19 The NYS formula for distributing IME paymentsprovides a higher payment to primary care, pre-ventive medicine and emergency medicine pro-grams meeting State criteria than programs inother specialties.

20 The estimated Medicare per resident direct GMEpayment in FY 2000 is $24,900. The estimatedper resident IME payment in FY 2000 is $48,300.

21 To qualify for GME payments for residents in anon-hospital setting, the hospital must assumesubstantially all of the training costs in the set-ting. As discussed more fully in the section en-titled “Current Financing of Graduate MedicalEducation” (beginning on page 23), there mustbe a written agreement between the hospital andthe community site that indicates that the hospi-tal is paying the resident’s salary and fringe ben-efits and reasonable compensation for teachingphysicians.

22 Introduced by Rep. Cardin on March 23, 1999.

23 Introduced by Senator Moynihan on January 19,1999. The same legislation was also introducedas H.R. 2771 by Rep. Lowey on August 5, 1999.

24 Estimate assumes $5 billion in tax on privatehealth insurance and $3 billion from Federal pay-ments for Medicaid acute care services.

25 In this report, hospital-based sites that providetraining that is representative of the environmentin which the residents will ultimately practiceare considered community settings.

26 Based on 83,422 residents and $71,468 averageper resident cost incurred by hospitals in FY1997,full funding would require $6 billion.


APPENDIX A: Explicit Medicaid GME Payments, FY1998

Alabama .................................. $0 $10 Montana .................................. * *

Alaska ...................................... * * Nebraska .............................. $5.0 $0

Arizona ...............................$17.8 N/R Nevada ................................. $8.4 $0

Arkansas .............................. $5.7 $0.00 New Hampshire ................... $2.1 $0

California ..........................$129.1 $0 New Jersey ........................ $20.0 $23.4

Colorado ............................... $8.0 $0 New Mexico ......................... $4.4 $0

Connecticut .......................... $6.0 $1.5 New York ............................. $812 $0

Delaware .............................$1.07 $0.2 North Carolina .................. $102.5 $0

District of Columbia ...........$15.2 $0 North Dakota ........................ $.93 $0

Florida ................................... N/R N/R Ohio ................................. $115.7 $28.9

Georgia ...............................$70.0 $0 Oklahoma ........................... $15.7 $0

Hawaii .................................... N/R N/R Oregon ................................. $8.6 $0

Idaho ....................................... * * Pennsylvania ...................... $66.6 $0

Illinois ...................................... * * Puerto Rico ............................. * *

Indiana ............................... $12.0 $3.0 Rhode Island ......................... N/R N/R

Iowa ................................... $43.8 N/R South Carolina ................... $57.8 $0

Kansas .................................. $7.7 $1.9 South Dakota .......................... * *

Kentucky ............................$49.7 $12.4 Tennessee .......................... $46.3 $0

Louisiana ............................ $50.0 $0 Texas .................................. $40.0 $0

Maine ................................... $2.4 $0 Utah ..................................... $4.0 $1.0

Maryland ............................$54.8 $0 Vermont ............................... $.63 $0

Massachusetts ...................$25.0 $0 Virginia ............................... $16.1 $0

Michigan ............................... N/R N/R Washington ........................ $63.5 $0

Minnesota ..........................$39.0 $19.0 West Virginia ......................... N/R N/R

Mississippi .........................$12.5 $3.1 Wisconsin .......................... $25.0 $12.0

Missouri .............................$26.7 $0 Wyoming ............................. $.06 $0

Italicized amounts are estimated.

N/R = Not Reported

*The Medicaid Agency reported that it does not pay for GME.

Source: Henderson, 2000.

STATE

TOTAL GME PAYMENTS(Millions of Dollars)

ExplicitPayments

ImplicitPayments

STATE

TOTAL GME PAYMENTS(Millions of Dollars)

ExplicitPayments

ImplicitPayments


SPECIALTY RESIDENCY REVIEW COMMITTEE REQUIREMENT

FAMILY PRACTICE ...................................... The primary training site is the model office or family practiceclinic where each resident must provide continuing, compre-hensive care to a panel of patient families. The clinic must befor the exclusive use of the residency. Residents may not beaway for remote assignments more that 1month in the first yearand two months in each of the last years.

INTERNAL MEDICINE ................................... The clinical settings must include a minimum of one-third ofthe time in ambulatory sites and a minimum of one-third of thetime in inpatient sites. Over the 36 months of training, at least½ day each week must be spent managing a panel of generalinternal medicine patients in continuity. Total required emergencymedicine experience must not exceed 3 months in 3 years oftraining for a resident.

INTERNAL MEDICINE SUBSPECIALTIES .......... Minimum requirement for experience in ambulatory settings thatincludes consultative as well as continuing care at least ½ dayper week for at least 18 months.

Continuing care experience must occur at least ½ day per weekfor at least 24 months.

The ambulatory care program must compromise a minimum of33% of the resident’s time. In a long-term care institutional set-ting, each resident must have 12 months of continuing longitu-dinal clinical experience. Experience with home visits and hos-pice care must also be included. Residency program mustprovide residents with experiences in an ambulatory care set-ting at least ½ day each week over 24 months of training.

The residency program must provide experiences in ambula-tory settings at least ½ day per week throughout 2 years.

Each resident must have equivalent of 1 full day per week ofcontinuity care experience the first year and at least ½ day perweek in second year.

PEDIATRICS ............................................... A minimum of 50% of clinical training should be devoted toambulatory experiences. This may include all assignments inthe continuity practice, emergency and acute care, and com-munity-based experiences, as well as the ambulatory portion ofnormal/term newborn, subspecialty, behavior/development, andadolescent experiences.

Residents must devote at least one-half day per week to theircontinuity experience throughout the three years, and an addi-tional one-half day session per week is suggested.

APPENDIX B: Program Requirements for ResidencyEducation in Ambulatory/Community-Based Settings

– CARDIOVASCULAR DISEASE

– GERIATRIC MEDICINE

– HEMATOLOGY

– ONCOLOGY

– RHEUMATOLOGY


SPECIALTY RESIDENCY REVIEW COMMITTEE REQUIREMENT

DERMATOLOGY .......................................... Adequate exposure to both outpatients and inpatients is neces-sary. It is essential that an active outpatient service furnish suf-ficient clinical material representing the broad array of diseasesseen by the dermatologist.

EMERGENCY MEDICINE ............................... No specific requirement for ambulatory/community-based train-ing outside the emergency department. There must be a struc-tured resident experience involving pre-hospital care (EMS).

NEUROLOGY .............................................. Program must include 6 months outpatient experience and mustinclude a resident longitudinal/continuity clinic with attendanceby each resident 1/2 day per week.

OBSTETRICS AND GYNECOLOGY ................... Primary and preventative care experiences should occupy theequivalent of at least 6 months of the 4 years of residency. Ex-periences should be strongly oriented toward ambulatory careand must include at a minimum a 4-month rotation in generalmedicine or family practice medicine and a 1-month rotation inemergency medicine. Residents should have a continuity clinicexperience at least ½ day per week for 3 years.

OPHTHALMOLOGY ...................................... Residents should be responsible for the care of an adequatenumber of outpatients representing a broad range of ophthalmicdisease. Each resident should participate in a minimum of 3,000outpatient visits.

OTOLARYNGOLOGY ..................................... Residents must have an opportunity to see patients, establishprovisional diagnoses, and initiate preliminary treatment planson an outpatient service that operates in relation to an inpatientservice used in the program. Experience should be provided inoffice practice procedures and management.

PSYCHIATRY .................................. Inpatient clinical experiences must be no more that 18 months.Outpatient experiences must include at least one year of conti-nuity experience. Clinical experience should include continuity-based mental health activities. Psychiatric consultation in medi-cal, surgical, and community settings.

PHYSICAL MEDICINE AND ............................ Residents must spend at least 1/3 of their time in care of outpa-REHABILITATION tients. Residents should gain fundamental understanding of

types of patients and care provided in the continuum of reha-bilitation care in community rehabilitation facilities.

UROLOGY .................................................. Residents must participate in continuity of care through pre-and post-operative clinics.

Source: AMA, 1999


As part of this study, site visits or telephone interviews were conducted with individuals involvedwith selected residency programs with community-based training sites. The sites that were in-volved in the interviews are listed below.

Beth Israel/Institute for Urban Family Health Family Medicine Residency Program. As theonly family medicine residency program operating in Manhattan, the program’s focus isquality training of physicians that will meet the needs of urban underserved. Beth Israeland The Institute jointly own a Medicaid managed care plan and the Institute’s familypractice centers have contracts with other HMOs. One family practice center serves as atraining site.

Boston University/Boston Medical Center Internal Medicine Residency Program. The Bos-ton University/Boston Medical Center residency programs have a long-history of in-volvement with Boston’s network of community health centers. Of particular interest isthe role of the Center for Community Health, Education, Service and Research, a con-sortium that places medical residents and advanced practice nursing students in interdis-ciplinary teams in the community health centers. The consortium has a Kellogg Commu-nity Partnerships grant.

East Tennessee State University. Founded in 1974 to alleviate a primary care shortage, thiscommunity-based medical school has 14 training sites outside its primary service area.The medical school used a Kellogg grant to develop multi-disciplinary training sites intwo rural communities. The medical school is the direct recipient of Medicaid GMEfunds under Tenncare.

Georgetown University Medical Center. For several years, Georgetown University operated ajoint medicine/pediatrics residency program with Kaiser Permanente where the residents’continuity experience was in a Kaiser community site. This program has now been movedout of the Kaiser site into a university-owned suburban physician clinic.

George Washington University. The university owns an HMO. The medical school is nowcommunity-based, having recently sold its hospital. As a partner in the Washington Re-gional Academic and Community Consortium (WRACC), George Washington Univer-sity has four residency programs participating in a Kellogg-funded Community Partner-ships demonstration involving an interdisciplinary faculty practice at six communityclinics.

Harvard Pilgrim Health Care/Brigham and Women’s Primary Care Residency Program.In 1992, the Department of Ambulatory Care and Prevention was started by HarvardMedical School and Harvard Pilgrim Healthcare as the first medical school departmentto be based in a freestanding HMO. It is a model of an academic medical center-man-aged care organization partnership. Participants spend 70 percent of their time practicingambulatory medicine at Harvard Pilgrim Health Centers in one-on-one relationships withpreceptors and 30 percent of their time at Brigham and Women’s on inpatient services.

Henry Ford Health System. In affiliation with Case Western Reserve, the Henry Ford HealthSystem operates residency programs with sites in the Detroit area, including communityhealth centers, and 31 Henry Ford Medical Centers located throughout the State of Michi-gan. Most community-based training takes place within the Henry Ford System.

APPENDIX C: Site Visits and Interviews


Michigan State University. The university houses community-based schools of allopathic andosteopathic medicine. The osteopathic consortia is a Statewide network of 17 hospitalswith the osteopathic college and over 1,000 interns that have pooled funds for centraladministrative expenses, as well as formal academic programs and workshops. This State-wide campus system is overseen by a governance board and osteopathic GME commit-tee representative of all members. Among the allopathic residency programs, the Depart-ment of Family Practice sponsors a network of affiliated family medicine residencyprograms throughout the State. The internal medicine program is affiliated with twoLansing hospitals and the MSU Medical Clinic. The program uses a “firm” approachthat teams small groups of residents with a full-time academic general internist for thecare of their continuity primary care patients on both an outpatient and inpatient basis.The Institute for Managed Care is an interdisciplinary managed health care initiative foreducation and research created through funding by the Blue Care Network of Mid Michi-gan to advance education in the managed health care system. The Institute has a PQEgrant to develop a multidisciplinary curriculum for community-based training.

Montefiore/Albert Einstein School of Medicine. Montefiore and the Albert Einstein School ofMedicine have unified clinical departments that are responsible for academic and clini-cal programs. Montefiore has a long-standing history of training in community clinics inthe Bronx. The hospital has expanded its network of hospital-owned clinics in order tocompete more effectively for managed care contracts. As these clinics develop strongclinical practices, they are used as residency training sites.

Ohio University. The College of Osteopathic Medicine is part of Centers for Osteopathic Re-gional Education (CORE). CORE, the first Osteopathic Post-Doctoral Training Institu-tion accredited by the AOA, combines the College, 14 osteopathic teaching hospitals inOhio and three other osteopathic medical schools located in Iowa, Maine, and Pennsyl-vania. CORE trains 472 interns/residents, 376 of which are in primary care specialties.

University of California, Los Angeles. Over a 5-year period, UCLA changed from 47 percentprimary care residency programs to 54 percent. The increase in primary care residencypositions has led to continuing efforts to increase community-based training sites. Ex-pansions have occurred through a network of university-owned primary care offices aswell as increased use of county health centers. Building on a successful program atHarbor/UCLA, the family medicine residency program is moving out of the hospital andinto a county health center.

University of Washington. The WWAMI network involves 45-50 residency training sites in fiveStates (Washington, Wyoming, Alaska, Montana, Idaho). About 30 sites provide oppor-tunities for 2-month block residency rotations to rural community sites. Training sites forresidency programs within the Seattle area include a hospital-based clinic, a university-owned network of physician clinics, community health centers, private practice officesincluding the not-for-profit University of Washington Physicians Network (UWPN) or-ganized under the auspices of UWAMC. The Affiliated Family Practice Residency Net-work consists of sixteen residency programs in the five-State region. Sponsors of pro-grams in the network include Group Health Cooperative of Puget Sound and four programswith rural training tracks.

financing graduate medical education in a changing health care environment · 2018. 12. 15. · 3....

Documents