Although only a surrogate for clinical ability or performance, the results of highly reliable standardized examinations are among the few readily available measures of these qualities. They are therefore widely used and will continue to be used in this way until other tools are developed for the direct assessment of clinical competence [4].

The two largest groups of examinees, approximately equal in size, are residents who have graduated from medical schools in the United States and foreign-born physicians who have graduated from schools outside of the United States and Canada (international medical graduates [IMGs]). The large numbers of persons in these groups who take examinations have permitted statistically valid comparisons. (Other groups-graduates of schools of osteopathic medicine, graduates of Canadian medical schools, and native-born U.S. citizens who have graduated from medical schools abroad-are smaller and have not been included in published analyses.)

During the past 3 years, a trend in the performance of U.S. graduates and IMGs on the ITE has become increasingly apparent. The scores of IMGs at all three levels of residency are higher than the scores of U.S. graduates [5], and the gap is widening [6] (Table 1). The differences are highly statistically significant, although their clinical importance can be questioned. Most meaningful, because of the high proportion of examinees, are the data on the performance of PGY-2 residents.

Why are these data noteworthy? Awareness of an imbalance in the United States between physician supply and the nation's needs is developing [7]. Some see the core problem as an excess in the total physician supply; others see it as an imbalance in the distribution of physicians across specialties or regions [8]. Some persons with the former viewpoint have gone so far as to propose the closing of some U.S. medical schools [9]. Others believe that a reduction in the number of entry-level residency positions, with or without a redistribution of trainees among specialties, is the way to solve the "workforce problem." A specific proposal that has generated widespread support is that the total number of PGY-1 positions in the United States be limited to 110% of the number of graduates of U.S. schools of allopathic and osteopathic medicine (USMGs) [10]. A likely consequence of the implementation of this proposal would be substantially fewer opportunities for IMGs to enter PGY-1 residency programs because there would probably be considerable pressure to ensure a first-year residency position in a teaching hospital for almost every graduate of a U.S. medical school.

At present, the system does not limit the number of PGY-1 positions offered, as long as the accreditation requirements of the various residency review committees are met. As a consequence, the graduate medical education enterprise in the United States has expanded to include more than 100 000 positions, about 25 000 of which are PGY-1 positions. This number is about 1.5 times the number of graduating USMGs [11]. Considering the enormous amount of care provided by residents and the fiscal benefit to teaching hospitals (from Medicare and other sources) for each resident, this unchecked growth is not surprising. The net effect of the 110% restriction or a similar formulaic constraint would be the denial of a residency in the United States to more than 5000 IMGs each year. Yet the ITE results suggest that the large pool of IMGs who enter internal medicine residencies in the United States perform at least as well in the cognitive domain as the graduates of U.S. schools. The explanation for this phenomenon is probably multifactorial and includes the often substantial period of residency training for IMGs in their countries of origin before they enter U.S. residencies. Would a change in the graduate medical education system result in an overall change in quality because of the exclusion of this large number of IMGs?

There is little doubt that a bias against IMGs has existed within graduate medical training in the United States, despite the undeniably outstanding accomplishments of many individual persons from that group [12]. Some evidence also indicates that IMGs have been more likely than U.S. graduates to seek subspecialty fellowship training after residency [13]. And although many IMGs enter residencies with visas that require that they return to their home countries after training, most IMGs actually end up practicing medicine in the United States because of waivers and other options [13]. On the other hand, IMGs have been more likely than U.S. graduates to care for the vulnerable populations in the inner-city areas of the United States [14]. Certainly, as trainees, IMGs provide considerable care for millions of persons who would have no other immediate options if the highly IMG-dependent graduate medical education system were to contract in many of the largest U.S. cities, especially those in the Northeast and Midwest [8].

Many believe, also, that the United States should continue to provide educational opportunities for IMGs who truly do intend to return to countries whose citizens need the skills that these physicians acquire in U.S. residency and fellowship programs. On balance, considering the needs filled by IMGs and their ability as reflected by ITE scores, a simplistic curtailment of the opportunity for thousands of IMGs to enter U.S. residency programs is unlikely to be in the best interest of the United States.

What, then, are the elements of a logical and appropriate solution to the physician workforce problems in the United States?

1. If, as seems very likely, more physicians are graduating from U.S. residency programs than are necessary to provide for the health of U.S. citizens, public funding should subsidize the education and training of only the number and specialty mix of physicians needed for patient care and biomedical research.

2. If it conforms to the values of the United States, as a long-standing and honorable tradition suggests, the provision of graduate medical education for physicians whose new skills will benefit the health of the citizens of their countries should be done with funding specifically identified for the purpose and with the certainty that the physicians so trained will fulfill their original intentions.

3. If we wish to correct an imbalance among the specialties, let us do so either by regulation or through the operation of market forces, not by denying residency training to whole categories of applicants, such as IMGs.

4. If there is a continuing need to link graduate medical education with the care of vulnerable populations because no other realistic options for such care exist-and this is a link that has a distinguished history in public hospitals and many private teaching hospitals-then let us plan to meet this need explicitly.

Considerable study and work would be necessary to meet these goals. However, we cannot continue to ignore the growing physician workforce problem and the malfunctions of the educational system that contribute to it. On the other hand, we should not yield to the temptation of assuming that all of our problems can be solved by denying access to graduate medical education in the United States to a group of persons who seem, at least in internal medicine, to be among the more able residents.