US Graduate Medical Education, 1999-2000

Prior reports on graduate medical education (GME) in the United States have demonstrated several consistent trends.^{1 - 3} The number of US graduates from allopathic medical schools (USMDs) entering GME programs has remained stable during the past 7 years, while the number of graduates from osteopathic medical schools (USDOs) entering Accreditation Council for Graduate Medical Education (ACGME)–accredited and combined specialty GME programs has continued to increase. The number of international medical school graduates (IMGs) who are US citizens or permanent residents and are first entering GME in the United States has also continued to increase. The same is true for the number and percentage of first-year positions filled by individuals with prior US GME. Although the total number of programs is still increasing, the number of specialty programs continues to decrease, while the number of subspecialty programs is increasing.

The Medicare Payment Advisory Commission is developing a plan for a Teaching Hospital Adjustment that would combine historic payments for direct and indirect GME, but it has not addressed several areas required by congressional mandate, including (1) federal policies regarding IMGs; (2) the dependence of medical schools on service-generated income; (3) projections for changes in physician supply and specialty needs; and (4) methods for establishing a suitable number, distribution, and mix of health care professionals.⁴ Efforts to determine the appropriate number and distribution of health care professionals also must address the needs of teaching institutions, as well as those of the rest of the health care delivery system.

The absence of a systematic strategy to guide GME is not due to a lack of effort on the part of concerned members of the medical education and health services communities. Since its authorization by US Congress in 1986, the Council on Graduate Medical Education (COGME) has been studying this problem and has published 14 reports and multiple resource articles addressing physician workforce issues.^{5 - 7} A consensus statement developed by 6 major professional organizations in 1996 recommended that the number of entry-level positions in GME be aligned more closely with the number of graduates of accredited US medical schools.⁸ The Council on Medical Education of the American Medical Association (AMA) made a more specific recommendation to gradually reduce the number of entry-level positions to 120% of the number of 1997 graduates of US MD- and DO-granting medical schools, and the AMA adopted this recommendation as policy in June 1999. In this year's report, we focus on these concerns by examining (1) entry level positions in GME; (2) characteristics of IMGs; (3) trends in the number of specialty and subspecialty programs; (4) the racial/ethnic diversity of USMDs; and (5) sites of primary care vs non–primary care training.

The 1999-2000 Annual Survey of Graduate Medical Education Programs was mailed in July 1999 to all programs accredited by the ACGME and to combined specialty programs (a total of 7946 active programs). Combined specialty programs are recognized by individual specialty boards as fulfilling the requirements for eligibility for board certification but are not accredited as programs by the ACGME. However, the participating specialty programs are accredited individually by the ACGME. Osteopathic residency programs are accredited by the American Osteopathic Association, not by the ACGME, and, therefore, were not included. In addition, some subspecialty programs are not recognized for accreditation by the ACGME.

The format of the survey was electronic for most programs, although several hundred program directors completed paper versions. Complete and early responses (by September 30, 1999) were rewarded with an early listing of program information in FREIDA Online (Fellowship and Residency Electronic Interactive Database Access; http://www.ama-assn.org/freida), an interactive database containing information on all accredited and combined specialty programs. FREIDA Online is used extensively by medical students researching specialties and programs in preparation for the National Resident Matching Program. Numerous attempts were made to encourage program directors to respond, including broadcast faxes, reminder letters, letters from state or specialty societies encouraging participation, and telephone calls. Program directors who had not returned a survey by mid-December 1999 were faxed an abbreviated survey, which asked for basic program- and resident-level information. Data collection was closed on January 31, 2000.

The total number of programs providing information on residents was 7560 (95.1% response rate), a slight numerical and proportional decrease from the previous year. Only 388 programs submitted abbreviated rather than full surveys. We also received program-only information from 17 programs, and another 56 provided limited information by telephone. For these 73 programs, as well as the other 313 programs from which we did not receive resident-level data, resident training year was advanced into the next year of training or the residents were "graduated," based on information from last year's survey. New residents were added from the 1999 National Resident Matching Program when possible.

There were 97,989 residents enrolled in ACGME-accredited and combined specialty GME programs in academic year 1999-2000 (Table 1; see additional data in Appendix II). Figure 1 compares the total number of residents, the number of residents in graduate year 1 (GY1) positions (entry-level GME positions requiring no previous GME, although some residents may have had prior training), and the number of residents in GY1 positions without prior US GME during the past 7 years. Compared with last year, the number of total residents this year increased slightly, with similar slight increases in the number of GY1 residents with and without prior training.

Because many proposed physician workforce policies have suggested limiting or monitoring the number of GY1 positions available as the entry point to the US physician workforce, we examined the characteristics of resident physicians in those positions and the historical record of their specialty/subspecialty career choices.^{5 - 6} Table 2 presents characteristics of physicians in GY1 positions, with or without prior US GME, during the past 3 academic years. Residents are categorized as USMDs, USDOs, or IMGs, with IMGs further divided by citizenship or visa status. Graduates of Canadian medical schools are typically categorized as US allopathic or osteopathic medical school graduates (USMGs) because of the similarity of medical education in the 2 countries; for the analyses reported here, they have been categorized by immigration/citizenship status.

Residents in GY1 positions with previous US GME are statistically significantly more likely to be IMGs (P<.001), particularly IMGs with permanent residency status. The proportion of IMGs in GY1 positions with prior US GME who are citizens (native or naturalized) or have permanent residency status has increased during the last 3 years to comprise a quarter of these residents. The number of IMGs with visas, with or without prior GME, has decreased proportionately. Graduate year 1 residents who are USDOs, both with and without prior GME, have also increased. One quarter of USDO GY1 resident physicians have had prior GME, a proportion that has held steady. Most residents in GY1 positions with prior US GME have had only 1 year of previous training. As a group, IMGs are more likely to have 2 or more years of previous training (Table 3). In this instance, IMGs who are citizens or permanent residents are similar to those with visas.

Overall, the number and percentage of residents in GY1 positions with prior US GME have continued to increase for USMDs, USDOs, and IMG citizens or permanent residents. Each resident in this group has already been included in previous years as a new entrant into the workforce. Only 22,320 (87.5%) of the GY1 positions in 1999-2000 reflect new physicians entering the GME pipeline (Figure 1).

The citizenship status of physicians who have completed training in a program and are pursuing additional GME is presented in Table 4. International medical graduates who are native-born US citizens, naturalized citizens, or permanent residents tend to have specialty/subspecialty career intentions similar to USMGs.⁹ But IMGs with temporary immigration status, namely those with H or J visas, may make training and career choices reflective of their more transitory positions and are more likely to train in a subspecialty. In 1999-2000, 44% of IMGs with temporary citizenship status entered another program after completing a first residency, compared with 27% of all graduating residents.

The difference between the proportions of USMDs and IMGs pursuing additional training is also apparent within individual specialties. Table 5 presents the number of residents completing core programs, along with numbers and percentages of USMDs and IMGs who were known to be pursuing additional training for 1999-2000. Considering primary care programs only, USMDs and IMGs are equally likely to complete a family practice residency and to pursue further training, while IMGs are more likely than USMDs to pursue additional training after completing an internal medicine program (49.6% vs 42.8%; P<.001). International medical graduates finishing pediatrics programs are twice as likely to obtain additional training than USMDs (41.2% vs 19.4%; P<.001).

Although the total number of ACGME-accredited and combined specialty programs increased negligibly during the past year (Table 1), there are now 5.8% more programs than in 1994-1995. This growth is principally the result of a 12.7% increase in the number of subspecialty programs since 1994-1995, averaging 83 additional subspecialty programs per year (Table 6). The number of residents enrolled in subspecialty programs has varied during the past several years, with an increase of 6.6% between 1998-1999 and 1999-2000. The average subspecialty program has between 3 and 4 residents, a number that has held steady for the past few years. The proportion of subspecialty programs without any residents has also varied slightly: 17% in 1997-1998, 22.3% in 1998-1999, and 19.0% in 1999-2000.

The total number of residents training in primary care specialties and primary care combined specialties has remained relatively constant during the past 3 years (Figure 2). The decline observed last year did not continue this year. The rate of rise of the number of USMDs in primary care has decreased but the absolute number continues to increase. During the past 6 years, the number of USMDs in primary care training programs has increased by 23.9%. During the same period, the number of IMGs in such programs has declined by 12.3%.

Efforts to increase numbers of underrepresented minority students in US medical schools have been hampered by recent state legislative and federal judicial outcomes.¹⁰ Although decreases in applications and enrollment have been observed,¹¹ it is too soon for these changes to appear in the racial/ethnic characteristics of residents, even for first-year entrants. Figure 3 presents these characteristics for GY1 USMDs for the past 4 years. For the second year in a row, the number of white GY1 USMDs has declined, decreasing 2.0% (n = 196). During this same period, the number of Hispanic GY1 USMDs has increased by 67 (10.5%). The number of Asian USMDs has increased by 204 (11.0%), while the number of blacks has declined by 76 (7.1%) from 1998-1999. Information on race/ethnicity was provided for slightly more USMDs this year compared with last year, with 76 fewer residents with unknown or unavailable race/ethnicity data.

Table 7 shows the distribution of USMDs in ACGME-accredited primary care programs for 1999-2000. Excluding Asians, who are more likely to be enrolled in internal medicine programs, the distribution of white, black, and Hispanic residents is similar, with 40.3% to 43.0% in internal medicine, 34.6% to 35.7% in family practice, and 22.5% to 24.0% in pediatrics (Table 7). Overall, 38.5% of all USMD residents are training in these 3 specialties. Significantly higher proportions of black and Hispanic residents (45.0% and 43.5%, respectively) are training in these primary care specialties compared with white and Asian residents (P<.001).

Since patient care increasingly occurs in the ambulatory setting, we reviewed the proportion of time residents spend beyond the first year of training within a program in various ambulatory settings in individual primary care specialties and non–primary care specialties for the past 3 academic years (Table 8). In the primary care specialties, residents in family practice spend more time than others in community ambulatory settings (37.7%), while those in pediatrics spend more time in hospital outpatient settings (46.6%). Time spent in managed care settings has decreased during the past 3 years for all primary care specialties, which may be a result of the decline in the number of staff model health maintenance organizations. Clearly, there is a difference in the amount of training time spent in outpatient settings between primary care and non–primary care programs, with the largest differences observed in ambulatory community-based settings that are not attached to a hospital. These differences have increased over time, but only because of a decrease in time spent in these settings by residents in non–primary care programs, not because of an increase on the part of primary care residents.

Some trends identified in recent years have not persisted. The number of residents leaving a GME program without completing it in 1999-2000 (n = 3674) increased slightly from 1996-1997; the growth is due to the number of residents transferring programs. The total number of IMGs and the number of IMGs with visas has rebounded to levels near those of 1997-1998. The total number of residents, which declined last year, is again at the level recorded 3 years ago.

In the absence of any oversight mechanism to regulate the numbers of physicians in the workforce, GME in the United States has increased in the number of both programs and residents. We find no evidence that the Balanced Budget Act of 1997 has had its intended effect of slowing the growth of GME. Although the number of residents with federal reimbursement has been restricted and federal support for the indirect costs of GME has decreased, the number of physicians in training continues to increase. As noted herein, the number of entry-level positions has continued to increase since 1997, from 24,516 to 25,498, and is now approaching the level reached in 1994. The number of GY1 residents with prior US GME has increased faster than the number of GY1 residents overall. The characteristics of residents in GY1 positions are increasingly heterogeneous. The number of subspecialty programs has continued to increase since 1993 and the number of residents in those programs exceeds the level recorded in 1994. Despite projections of negative financial balances for 59% to 68% of US hospitals over the next several years,¹² there has been no systematic effort to control the number of residents in training.

Workforce projections and recommendations from COGME,^{5 ,7} while thorough and thoughtfully developed, have had little effect. Similarly, a consensus statement developed by 6 major professional organizations in 1996⁸ to address the projected physician oversupply has not noticeably affected the size or balance of US GME. Since workforce planning involves so many unknown variables, it is essential that any system to control the number and specialties of physicians to be flexible and responsive to a changing environment and health care system.