This is a time of great uncertainty for graduate medical education (GME). There have been strong calls for the expansion of the number of residents trained, in consideration of the looming large increase in demand for health care and to meet the training needs of rapidly increasing numbers of medical school graduates.1,2 Despite this, the president’s most recent federal budget proposal calls for a reduction of the Medicare support of GME.3 In addition, there is much discussion concerning the appropriate specialty mix of GME with emphasis on expanding the training of practitioners of adult primary care.4
The system by which we train medical residents in the United States is a complex one. The majority of funding is provided by governmental agencies, principally Medicare, but a significant portion is provided by the training institutions themselves.5 Overall governance is provided by the professions through entities such as the Accreditation Council for Graduate Medical Education (ACGME) and the American Osteopathic Association (AOA). However, decisions concerning which specialties to provide training in are made by the training institutions themselves—at this point almost entirely teaching hospitals—which might have strong financial incentives when choosing specialties to support training.
It becomes important, therefore, to understand these institutions’ values regarding the desirability of training programs in the various specialties. To contribute to this understanding, we present here the results of a survey of New York State teaching hospitals that we conducted in 2009. At that time, there was considerable interest in the expansion of GME, so we asked these institutions whether they were currently contemplating GME expansion and how they would deploy additional federal funds at current rates to establish new residencies and enlarge current ones. We focused on the core residencies because it is these that would have to be expanded if the total resident output were to be increased. We chose to study New York State because it afforded a large but manageable sample, recognizing that it is not similar to other states with regard to some aspects of GME. However, we believe that the basic determinants of the desirability of residencies for sponsorship by teaching hospitals must be common to most teaching hospitals.
We believe it is important to consider these determinants now for two reasons. First, in view of the pending large increase in demand for health care in the not-too-distant future, GME will almost certainly be expanded in the long run, whatever the source of funds will be to do so. We note that as we write this report, bipartisan legislation has been introduced in Congress to increase the number of Medicare-supported training of resident physicians.6 Hospital preferences will play a key role in determining which specialties will be expanded, especially if new programs are supported in whole or in part by these institutions themselves. Second, if Medicare funding decreases in the short term, hospitals’ valuation of training in the various specialties will affect decisions regarding downsizing of programs.
From the ACGME’s public Web site,7 we identified all ACGME sponsoring institutions in New York State in 2008. (Sponsoring institutions are teaching hospitals and other entities with primary responsibility for residencies.) We obtained the following information on each: the designated institutional officials (DIOs), the residency programs offered, and the number of positions approved and filled for each residency. AOA sponsoring institutions were not surveyed because of difficulty in obtaining basic institutional data from the AOA. Our focus was on first-certification (“core”) residencies (listed in Table 1)—that is, primary, not subspecialty, residencies—because these are essential to expanding the number of residency positions available for medical school graduates.
Of the 58 sponsoring institutions in New York State, 10 were single-specialty institutions, most with subspecialty residency programs. We excluded these from the study. Thus, there were 48 sponsoring institutions eligible for the study. In an attempt to maximize the response rate, we introduced the survey by a priority mail letter followed by an e-mail. We made as many as three follow-up phone calls to nonrespondents. To make the survey user-friendly, a separate version was tailored to each institution so that respondents could easily view their active programs in a “drop down” file when that was necessary for a response.
The Web-based survey elicited information from the DIOs about their desire to expand their existing residency programs and add new residency programs (1) using currently available funds, and (2) with increased state and federal funding at current rates. For expansion of existing programs, we asked them to estimate the number of positions they would add. We also asked all DIOs, “Besides the current caps, do any of the following factors or concerns limit your hospital/medical center’s ability to add new residency programs?” The choices are listed, with responses, in Table 3. We developed the survey and piloted it by asking the DIOs of Stony Brook University teaching hospitals to take it. Their comments were used to develop the final version.
Because all sponsoring institutions in the survey were hospitals, hospital systems, or medical schools with an affiliated hospital, we also determined bed capacity for each using the 2009 American Hospital Association guide,8 which reports data collected in 2008 and, therefore, closely matches the year in which ACGME residency information was reported. For hospital systems, we aggregated the beds of all hospitals in each system.
The project was reviewed and approved by the Stony Brook University human subjects committee institutional review board (#104622).
Thirty-six of the 48 sponsoring institutions (75%) responded. There was no difference between responders and nonresponders with regard to either number of existing residency programs in place or bed capacity. We noted that half of the core residencies in the sponsoring institutions that responded reside in medium-sized hospitals (300–600 beds) with fewer than 10 residency programs.
In the absence of additional funding, 5 (14%) of the 36 responding sponsoring institutions were planning to add new residency programs, and 8 (22%) were planning to add new positions to current programs. If additional funding was available, 14 (39%) would add new programs and 17 (47%) would add new positions. Three (8%) were contemplating the downsizing or elimination of one or more residency programs. The distribution of interest in sponsoring new programs and adding positions to current programs by bed capacity and number of existing programs is shown in Figure 1. There was little interest in expansion among the largest institutions, which, we conclude, were essentially saturated with regard to training residents in first-certification specialties. However, interest existed in medium-sized sponsoring institutions with less than a full complement of first-certification residencies and, therefore, the capacity to add new programs and new positions to current programs.
Adding new programs
Table 1 lists the specialties of interest to sponsoring institutions with the capacity to add a program—that is, those without a residency program in that specialty. Because all specialties cited for development without funding were also cited for development with funding, we considered the “development with funding category” as representing the true total interest. The greatest interest was in emergency medicine, with seven (one-third) sponsoring institutions without an emergency medicine residency program desiring one. This was followed by plastic surgery (4; 13%). Beyond that, there was interest in adding a wide variety of programs, with only one or two institutions interested in each. Notably, despite considerable capacity, there was little interest in adding family medicine programs. Assuming new family medicine programs would each have six entry-level positions (an estimate based on review of recent ACGME-approved programs), the desired two new programs would graduate only 12 additional family physicians each year.
Adding positions to existing programs
As in Table 1, we assumed that “expansion with funding” represented the total interest. The greatest interest was in expansion of internal medicine (Table 2). Assuming that approximately 30% of internal medicine residents go on to practice primary care,9 this expansion would produce 23 additional primary care internists per year. (This assumption is based on findings of a 2005 study, but we believe the numbers are not significantly different today.) This was followed by several largely hospital-intensive programs including diagnostic radiology, obstetrics–gynecology, and general surgery. The desired expansion of family medicine programs would produce eight additional practitioners a year.
The programs identified for downsizing or elimination were dermatology, family medicine, ophthalmology, pathology, preventive medicine, and physical medicine and rehabilitation.
Barriers to expanding GME
The responses to the questions about the potential barriers to expansion of GME programs are listed in Table 3, sorted by whether the institutions were interested in adding new programs with additional funds or not. A substantial percentage in both groups thought that adequacy of space and number of potential full-time faculty were limiting issues. Not surprisingly, this concern was somewhat more common among those that would not expand their programs. It is also not surprising that those that were not interested in expansion were substantially more likely to list as concerns the adequacy of current reimbursement rates (41% versus 20%) and the ability to recruit residents of preferred quality (41% versus 10%). As expected, those not interested in establishing new programs or expansion of current ones were more likely to be concerned about adequacy of patient volume. “Other infrastructure issues” were commonly cited by both but not further elaborated on.
We focused on first-certification (“core”) medical residencies because they are the rate-limiting step in the education of future physicians in the United States. Although many states will license physicians without completion of a residency, as a practical matter, virtually all physicians must complete an accredited residency in the United States prior to practice in the United States, wherever their medical school training has been. Our findings indicate a modest interest in and/or a capacity for expanding GME (about 50%) among ACGME sponsoring institutions at the time of our study. However, this interest was largely focused on hospital-based and hospital-intensive programs.
If additional funds were available, by far the most sought-after program would have been emergency medicine. This was followed mostly by other hospital-intensive programs. Notably, there was little interest in establishing programs in family medicine among these institutions. However, a significant number of institutions said they would expand existing internal medicine residencies. Figure 1 shows that GME growth would be primarily in medium-sized hospitals with fewer than 10 residency programs. It is not surprising that the large centers had no capacity to add new first-certification programs. However, we do note that only two of these were interested in expanding them. Also of note is the rather modest interest in or capacity for training in other specialties considered to be in need of additional practitioners, such as psychiatry.
These findings suggest that resident “productivity” considerations were of major concern in determining which residencies were of interest for development or expansion. Emergency medicine residents substantially extend the capacity of physicians practicing emergency medicine at a fraction of the cost of additional fully trained specialists. Furthermore, this specialty has become much more important to hospitals, as there has been a marked increase in use of emergency departments as sources of hospital admissions in recent years.10 Similarly, internal medicine residents are, in a real sense, the “workhorses” of inpatient medicine and less expensive than their current substitutes, full-time hospitalists. Family medicine residents, on the other hand, spend substantially less time training in the hospital setting and more time in outpatient settings, which are generally considered to have a less positive fiscal impact on hospitals. In the same vein, only one of the programs listed as candidates for elimination or downsizing (pathology) is hospital intensive. In view of the changing nature of that specialty, with a great deal of outsourcing to large independent laboratories, the need for pathology residents may be diminishing.
The barriers to expansion that were cited were not inconsequential and were largely shared by both the institutions interested in expansion and also by those which were not. They included inadequate faculty in specialties of interest, inadequate space, and inadequate current reimbursement rates. Thus, the present data reflect desires for expansion and probably overestimate the actual expansion that would occur if additional funds were available at current rates.
We do not fully know to what extent current nonteaching hospitals could provide sites for new residency programs. Responses from a survey of nonteaching hospitals in New York State11 indicate that they almost all perceive the same barriers to establishing programs as do teaching hospitals for expansion of programs. In addition, they note a lack of meaningful association with current teaching institutions as a major impediment to establishing residency programs. Accordingly, relatively few seem interested in developing GME. These findings are quite similar to those of a study in Florida.12 We do note that nonhospital clinical providers are, at a modest funding level, targeted for development of new primary care residencies by the Affordable Care Act. It is not yet clear how successful this program has been or will be in generating new primary care residencies. However, it should be pointed out that at the time of our study, ACGME standards still resulted in about 50% of family medicine residents’ training time being hospital based.13 Thus, the capacity for development of new primary care residencies remains dependent on the interest of these institutions.
We acknowledge potential limitations to the study. There are inherent uncertainties in survey-based studies that are directed to a single person in an institution. For example, the directors of GME at teaching hospitals might have a greater interest in establishing residency programs than would directors of other types of hospitals. Additionally, we cannot assume that these findings from New York State can be simply generalized to the United States as a whole. There may well be different valuations of specialties such as family medicine across the country. However, we believe that the productivity-related incentives to establish hospital-based or hospital-intensive residency programs among current teaching hospitals in New York are likely be the same nationwide. Finally, even though our study findings are from a few years ago, we believe they are instructive and relevant and set the stage for future studies.
For example, in a recently completed national survey of residency sponsoring institutions, we asked respondents to indicate the degree to which they valued core residencies from both operational (and financial) as well as educational (and public service) points of view.14 Internal medicine, general surgery, and emergency medicine were rated highly in both categories, whereas dermatology, nuclear medicine, and preventive medicine were rated poorly in both categories. This substantiates the idea that hospital-intensive residencies are the most desired by current teaching hospitals.
Conclusions: Physician Workforce Implications
The primary finding of this study is that sponsoring institutions’ envisioned provision of additional support for GME at current rates would likely do relatively little to increase the output, from current teaching hospitals, of physicians practicing adult primary care medicine. Our data strongly suggest that hospitals have very little interest in developing family medicine programs despite considerable potential capacity to do so. Indeed, there was some interest in eliminating family medicine programs. There was interest in expanding ongoing internal medicine programs. However, a survey of internal medicine residents a few years before our study indicated that only about 30% of them would ultimately practice primary care medicine,9 a proportion that we believe has not changed significantly. Thus, our data suggest that if all responding institutions could mount the program development and expansion they were interested in, they would increase the yearly output of physicians practicing adult primary care medicine in New York State by approximately 53 (20 in family medicine and 23 in general internal medicine). By way of reference, the American Medical Association indicates that in 2009, the year of this study, there were 65,679 clinically active physicians in New York State; 2,901 self-identified as family physicians or general practitioners, and 18,570 self-identified as internists with no subspecialty practice.15
We believe that these observations lead to at least two inferences. First, when funding for GME is increased, as it eventually must be (whatever the source of funds) because of a rapidly increasing demand for health care, teaching hospitals will require greater incentives than currently exist to develop programs in family medicine and, perhaps, other specialties considered to be in need of expansion but of limited “productivity” such as psychiatry. One might ask whether there will be a demand by medical school graduates for more positions in these specialties. It might be pointed out in this regard that because of recent rapid expansion, it has been predicted that the output of U.S. medical schools will increase by 30% by the end of this decade and may exceed entry-level residency positions if there are no increases.16 Second, if, in the short term, funding for GME is decreased, the apparent value system of current teaching hospitals may put family medicine programs at greater risk than others for downsizing or elimination.
Acknowledgments: The authors would like to acknowledge that Maria Relevo, Stephen Colon, Aleef Rahman, MPH, MBA, Kadhambari Sridhar, MPH, and Erin Healy, MPH, RN, made substantial contributions to this research.