Graduate Medical Education, 1997-1998

MEDICAL EDUCATORS have long observed that the growth in the number of graduate medical education (GME) trainees, now well beyond the number needed for the national physician workforce, parallels the amount and availability of government funding for GME. This growth is mainly attributable to Medicare funding and to a lesser extent state support, either directly or through Medicaid.^{1 - 6} In 1996, knowing that the Congress would address revisions in the Medicare legislation in early 1997, many groups became interested in formulating recommendations for adjusting the growth in GME positions. Six major professional medical organizations (the American Association of Colleges of Osteopathic Medicine, American Medical Association, American Osteopathic Association, Association of Academic Health Centers, Association of American Medical Colleges, and National Medical Association) met in December 1996 to find areas of common agreement and make recommendations, as a Consensus Statement on the Physician Workforce, to guide the administration and members of Congress in their policy deliberations regarding physician oversupply.⁷

The consensus statement proposed 7 recommendations. The first proposed reducing the number of GME positions funded by the federal government to a number closer to that of the graduates of US allopathic medical schools. The second proposed reestablishing the J-1 Exchange Visitor Program as a true exchange program. The third addressed the problems of physician maldistribution. The fourth called for a new all-payer system of GME support. The fifth proposed transitional funding for institutions that choose to downsize their GME programs. The sixth acknowledged the need for a stable source of support for teaching hospitals. The final recommendation called for a new public- and private-sector workforce advisory body to have oversight over this entire process.

The Consensus Statement on the Physician Workforce reflected a broad spectrum of opinions, held not only by the 6 organizations, but also by other medical educators and policymakers. Therefore, it is not surprising that most of these recommendations were included in the Balanced Budget Act (BBA) of 1997, either as direct provisions or as issues to be studied over the next 2 years by the Medicare Payment Advisory Commission (MedPAC) and the Bipartisan Commission on the Future of Medicare.⁸ In addition, several states have taken initiatives with similar intent. Of the critical GME issues, the BBA set a limit for the direct graduate medical education (DGME) payment based on the number of residents in the institution as of December 31, 1996. The BBA included provisions that provided transitional funding for institutions that voluntarily enter into major GME downsizing over a 5-year period. Additionally, the BBA provided special consideration in several areas for primary care. Among these are exemptions from some of the downsizing requirements if there is a shift of positions to primary care, and payment for combined residencies is enhanced if the combined programs are in primary care. Also, prior to the adoption of the BBA, 42 hospitals in New York had set a precedent for the downsizing experiment by gaining agreement that the Health Care Financing Administration (HCFA) would provide transitional funding.⁸

The BBA gave MedPAC and the Bipartisan Commission, among other requirements, the responsibility to study and make recommendations for policy regarding international medical graduates (IMGs), stable sources of funding, appropriate support for pediatric programs, particularly in free-standing children's hospitals, and the determination of how many physicians are needed by each major specialty.

Accurate tracking of changes in numbers of trainees is essential to the assessment of the effects of policy changes. Since the early 1900s, the American Medical Association has collected data on GME in the United States and published the results in the annual medical education issue of JAMA. The total number of resident physicians reported in JAMA for each year since 1980 is reported in Figure 1. With the variety of public policy initiatives being introduced and the definitive studies to be done, it is now more important than ever to collect and provide complete and accurate GME data. In addition, it is important to collaborate with other organizations in the GME data collection effort to reduce the reporting burden placed on program directors and teaching institutions, improve the quality of the information, and agree on the size and distribution of GME.

In 1996, we reported from our GME survey results that a few programs had already reduced in size, even though no major initiatives for reductions had been implemented. Furthermore, we found that other programs had increased their numbers. The result was very little net change in total aggregate numbers, thus masking the outcome and producing what we termed "the shifting sands of GME."⁹ In our GME report in 1997, we reported that before the federal and state initiatives had been implemented there was a trend among programs to reduce numbers of first-year positions. With increasing numbers of IMGs remaining in the subspecialties, the changes in total number of residents were minimal. Thus, one must be cautious in ascribing what is cause and what is effect, as some events seem to precede the correcting stimulus.

Now federal and several major state initiatives are being implemented, which will directly affect the GME numbers. In this year's report on GME, we have incorporated historical data along with current 1997-1998 GME data to provide a more complete reference basis for following and analyzing changes in GME in response to each of the major public policy initiatives. Extensive GME data for 1997-1998 are provided in this issue of THE JOURNAL in Appendix II.

The Annual Survey of Graduate Medical Education Programs was mailed to 7861 Accreditation Council for Graduate Medical Education–accredited and combined specialty programs in July 1997. The programs were given a deadline of September 30, 1997, to complete either the electronic or paper version of the survey. This year the survey data collection process was aided by the endorsement and conjoint support from the Association of American Medical Colleges, Texas Medical Association, American College of Surgeons, Center for Health Workforce Studies at SUNY, Albany, Virginia Center for Generalist Medicine, Office of Statewide Health Planning and Development in California, and University of California, San Francisco, Center for the Health Professions. Each organization communicated with its constituents to assist with response management and improve data quality. In addition, the American Medical Association conducted follow-up communications using broadcast faxes, reminder letters, and telephone calls. For programs unwilling to participate fully, abbreviated surveys were mailed December 15, 1997, to elicit, at the least, information on critical elements. The data collection ended on January 31, 1998.

The number of programs providing resident-level responses totaled 7459 (94.9%), including 525 abbreviated responses. We did have some type of response from 96.5% of the programs; however, some returns were program level (FREIDA) data only. As in previous years, for the 402 programs that did not provide updated resident census information, the resident physicians reported from the last received survey were moved into their next year in the program or graduated, and new residents were added from the 1997 National Resident Matching Program, when available.

Interpreting the data requires an understanding of the definitions for program year 1 and graduate year 1 (GY1) positions. We use the term GY1 to refer only to those entry-level residency positions available to individuals directly out of medical school (even though some individuals may have had prior GME training). Program year refers to the current year of training within a specific program and may or may not correspond to the graduate level. Thus, a resident could be in program year 1 of the current program (eg, cardiology) but in his or her fourth year of GME.

Over the past 5 years, the aggregate number of residents enrolled in all GME programs (including combined specialty programs), as well as the number who entered GY1 positions without prior GME training (first-time entrants), has remained essentially constant (Figure 2). In contrast, the number of residents in entry-level GY1 positions, regardless of their prior US training status, has been reduced by 5.8% over the past 5 years. During 1997, 11% of the resident physicians in GY1 positions had already received some prior US GME. The reduction in overall size of GY1 positions may be a consequence of HCFA's providing the full DGME payment only for the minimum number of years required to become eligible for first board certification.

The difference in the number of US allopathic medical school graduates and the total number of residents in GME programs is made up of IMGs, graduates of American Osteopathic Association–accredited medical schools, and graduates of Canadian medical schools. These numbers for 1993 through 1997 are given in Figure 3. The total numbers of osteopathic and Canadian physicians in GME programs has been stable over the past 5 years. However, the total number of US allopathic medical school graduates in GME has decreased 4.4% since 1993, while the total number of IMGs has increased 12.4% since 1993. Further, the decrease of 3107 US medical school graduates in GME since 1993 was about the same as the increase in IMGs (2810) during the same period.

As shown in Table 1, the number of IMGs entering the system as first-time GY1 residents with no prior GME experience has actually decreased and the number of US medical graduates has remained constant. The increased total number of IMGs in GME is explained by the fact that the IMGs tend to move into subspecialty training or to move from one discipline to another. Thus, of the IMGs in GY1 positions in 1997-1998, 5230 had no prior GME experience while another 1027 (16.4%) had 1 or more years of prior US GME. In contrast, only 8.5% of US medical school graduates were in GY1 positions for the second time. It appears that a higher percentage of US medical graduates leave GME training to enter practice after completion of sufficient years to provide eligibility for first board certification rather than enter subspecialty training or change disciplines.

This is especially true in the primary care disciplines of family practice, internal medicine, pediatrics, internal medicine/pediatrics, internal medicine/family practice, as illustrated in Figure 4. As the number of residents training in primary care increased over the last 5 years from 36777 to 41378 (12.5%), the number of IMGs training in these specialties peaked at 13025 in 1995 but dropped to 12185 in 1997 and is basically unchanged since 1993 (−1.3%). In addition, the number of residents training in the subspecialties of family practice, internal medicine, and pediatrics has declined by 8.0% in total; however, the number of IMGs training in these subspecialties has increased 22.9% since 1993. The majority of IMGs training in primary care specialties were in internal medicine. Figure 5 shows that the percentage of IMGs was 40.2% in 1997, down from the peak of 42.5% in 1995. In contrast, the percentage of IMGs in the internal medicine subspecialties has continued to rise every year for the past 5 years, reaching 47.6% in 1997. Similarly, the percentage of IMGs in general pediatrics (33.6%) (Figure 6) peaked in 1993 and has been dropping each year subsequently and is now 25.5%. As with internal medicine, the percentage of IMGs in the pediatric subspecialties, currently 45.4%, has consistently risen over the past 5 years.

International medical graduates continue to obtain GME opportunities in the United States and now make up 26% of the entire GME population. This is an increase of 2810 IMGs since 1993. Although the overall number continues to rise, the number of new-entrant IMGs has dropped over the last 5 years (Table 1). In 1997, there was a net increase, but it was attributable principally to an additional 195 US citizen IMGs over the previous year (Appendix II, Table 4). Physician workforce policymakers should keep in mind when dealing with IMG issues that IMGs do not represent a homogeneous population. For example, as shown in Appendix II, Table 6, only 34.8% of IMGs are on J-1 or J-2 exchange visitor visas. In contrast, 49% are either US citizens or permanent residents and another 9% are on H-1 or H-1B temporary worker visas.

If the United States either has or is approaching a physician excess, then market forces should reflect increasing difficulty for physicians to find suitable practice opportunities for some disciplines and in certain geographic areas. We have attempted to assess this phenomenon over the past 3 years. The results of our surveys indicate that there is a significant problem for some specialties and subspecialties in certain areas. The results of these studies have been detailed in 2 previous publications.^{10 - 11} The third report appears elsewhere in this issue of THE JOURNAL.¹²

An example of the value of collaboration in data collection and analysis can be seen in the early observations of the New York hospital downsizing demonstration project. As indicated earlier, that project, which negotiated with HCFA prior to the legislative authorization in the BBA to decrease the number of residents training in New York State, is responsible for training about 15% of all allopathic residents. Our 1997 GME data initially showed a small increase in the number of residents training in New York since 1996 (Appendix II, Table 2). With further investigation by the Center for Health Workforce Studies at SUNY, Albany, it was found that some underreporting of data occurred during 1996 and the 39 hospitals participating in the demonstration project reduced the number of residents by 5%. Most of the reductions were in first-year positions and will lead to additional reductions in future years. In addition, there was little or no change in the nondemonstration hospitals.

The perception that the United States is on the verge of physician excess has led a number of individuals and groups to seek the federal government's intervention to bring the production better in line with the nation's need. The government has been petitioned because it is widely believed that the excess of physicians is the result of unlimited federal financial support of GME programs, regardless of need.

As federal and state initiatives are put in place to influence different outcomes in GME, it is imperative to monitor the changes quickly and critically to be certain that the changes are, in fact, what has been intended and that they neither fall short nor go beyond the objective. In addition, it is important for organizations to collaborate on data collection and reporting efforts. Agreement on the size and distribution of the GME workforce is critical. The timeline in the establishment of GME programs is much too long to tolerate errors in recalibration in numbers of programs and positions. These data are intended to provide, together with Appendix II, baseline and recent year trend data from which GME changes can be critically and meaningfully monitored.