The 3 nonprocedural primary care specialties (internal medicine, family medicine, and pediatrics) accounted for 10091 (46.2%) of the NRMP-offered PGY-1 residency positions (table 1). The 7 disciplines (primary care plus procedural/lifestyle) analyzed in this study represented 15205 (69.6%) of the PGY-1 positions offered through the match (table 1). Of these, 64% were filled by USMDs and 31.7% matched with non-USMDs (4.3% went unmatched). An additional 1262 Pre S positions (not included in either the primary care or the procedural/lifestyle groups) were offered in the match, and only 38.1% matched with USMDs (table 1).

The total number of PGY-1 positions available, in the 3 primary care and 4 procedural and/or lifestyle-oriented specialties studied, was 18030, with 2825 (15.7%) of these positions being offered outside the match (table 1). Practically all USMDs are required to take residency offers through the match (in fact in 2007, 91% of the available 15667 USMDs were recruited via the match); thus, virtually all outside-the-match offers are made to non-USMDs. Variation in percentages for outside-the-match offers was found by specialty and ranged from 7% in obstetrics-gynecology to 23% in internal medicine. Four specialties—family medicine, internal medicine, transitional year, and Pre S—appear to be prematching more than 20% of their total available positions (table 1). The percentage of NRMP-offered positions taken by USMDs ranged from 87% in transitional year to 42% in family medicine (38% in Pre S). The percentage of outside-the-match offers was highly inversely correlated with the specialty's “success” in matching USMDs (SROT = −0.87) (figure 2).

A total of 8304 (46.1%) of all PGY-1 positions (in the specialties studied) in 2007 were offered to non-USMDs in and out of the match (table 1). Most of these offers were made in the 3 nonprocedural primary care specialties (internal medicine, family medicine, and pediatrics) and they represent 54.7% of the total 12558 primary care positions offered (table 1). Fully 19.6% (2467) of these positions were offered outside the match and they represent 35.9% of all primary care offers to non-USMDs (table 1). The procedure-oriented/lifestyle-oriented specialties (obstetrics-gynecology, categorical surgery, preliminary internal medicine positions, and transitional year) matched with an average of 78.9% USMDs and offered only 358 of their total positions outside the match (table 1). The percentage of the total PGY-1 positions in these specialties accepted by non-USMDs was 26.2% (with 6.5% offered outside the match) (table 1).

In internal medicine, the percentage of outside-the-match offers was significantly and inversely associated with the number of fellowship programs (P < .0001) offered by these internal medicine programs (table 2). After excluding data from states with few internal medicine programs, the association remained significant (table 2).

About 1 in 5 positions in nonprocedural, primary care specialties are offered outside the match. Also, 54.7% of all primary care PGY-1 positions were offered to non-USMDs, of whom 35.9% took positions outside the match, whereas 3977 non-USMDs who registered for the match withdrew before the match. While around 1200 offers (Mona Signer, MPH, E-mail communication, December 15, 2009) were withdrawn for candidates' failure to meet Educational Commission for Foreign Medical Graduates testing requirements, the remaining candidates (2777) are presumed to have been offered and to have accepted outside-the-match offers, closely approximating the 2825 outside-the-match offers made in the specialties studied. This observation suggests that most (70%) of the withdrawals were to take offers outside and before the match, though proof of this is lacking. In procedure-oriented/lifestyle-oriented specialties, only 26.2% of all positions are filled by non-USMDs and only 6.5% of all such positions are estimated to be offered outside the match.

The possible effect of the current unregulated system of offering outside-the-match residency positions has been to create a recruitment method in graduate medical education (predominantly in primary care specialties) characterized by program directors “giving” positions primarily to non-USMDs who are eager to secure a position rather than take their chances in the match. This situation is similar to that which existed for USMDs before the match (before 1952) and needs to be addressed. Our estimates suggest that outside-the-match offers occur to an extent previously unrecognized, especially in 4 specialties (family medicine, internal medicine, Pre S, and transitional year). Furthermore, the amount of prematching in internal medicine may be linked to the lack of availability of fellowship positions.

It may be useful to consider why program directors choose to offer and candidates choose to accept positions outside the match. Program directors may do so to address 2 recruitment issues: (1) to prevent losing interested, non-USMD candidates to other programs offering outside-the-match contracts and (2) to avoid the “scramble.” Offering positions outside the match reduces a program's match quota, thereby limiting the number of unmatched positions. The “scramble” has been likened to college basketball's “March Madness,”⁸ and candidates taken in the postmatch period seem to perform less favorably when compared to those accepted in the match.⁹

The potential situation for candidates is similar. Non-USMDs, correctly perceiving their generally less competitive status relative to USMDs, may, if they wish to compete in the match, choose to apply to more programs and risk needing to “scramble” if they fail to match. Reginald Fitz, at Boston University, summarized a student's perspective in 1939: “… there are very few men who have the conceit to pass up a very good appointment in 1 locality offered early simply on the gamble of competing for a somewhat more desirable appointment made later in another locality.”¹⁰ Non-USMDs also may be disadvantaged in the “scramble,” as they may not receive the dean's office assistance with phone calls and paperwork; additionally, they may experience difficulty getting through to program directors who hope to fill their unfilled position(s) with USMDs. Also, the timeline for obtaining a visa is shorter the longer one waits for a position, making a prior-to-the-match offer additionally enticing for foreign citizens.

The literature^11–,14 indicates perceived match violations, including promises to applicants before the match, thus putting in question the ethics and morals of program directors and candidates alike. The NRMP has considered proposals to limit outside-the-match offers¹⁵ and to conduct a second match for unmatched candidates.¹⁶ Neither has been adopted, in part because of concerns expressed by program directors “dependent” on international medical graduates, who might have difficulty getting visas in time for their July start dates. Our study represents a preliminary attempt to elucidate the amount of prematching occurring in the United States. Since prematching data are not officially reported anywhere, our calculations may be overestimating or even underestimating the extent of out-of-match offers. If, however, it is confirmed that primary care programs (particularly internal medicine and family medicine) need to consistently, year after year, recruit 1 of 5 applicants outside of the match, then this negative finding should be further examined. These proposals should be reconsidered in the context of our study findings. We advise the community against drawing strong conclusions from these initial results, which analyze only a fraction of 1 academic year's match. Further research, by independent groups and spanning multiple academic years, is needed to confirm or disprove our findings and to analyze prematching trends in all specialty areas.

One limitation of this study is that the AAMC Data Book⁵ provides only the programs' reporting of both the total number of residents working in each specialty (without a breakdown by year of training) and the number of positions offered in total and in the PGY-1 year. When the total number of positions offered (Table F7 of AAMC Data Book⁵) is compared to the total number of residents reported (Table F8 of AAMC Data Book⁵), there is a 3% difference in nonprocedural primary care specialties and a 1% difference in procedure-oriented specialties. This difference between positions offered and filled should have little effect on the study, as it probably is evenly distributed over all years of residency training. Since preliminary internal medicine positions and Pre S positions are a part of—but much different than (with regards to a career)—the overall internal medicine and surgery program of a given institution, respectively, we analyzed them separately from their categorical counterparts. In doing so, we had to assume that no prematching occurs in preliminary internal medicine or categorical surgery positions. The actual number is likely a small percentage, and the assumption does not significantly affect the results of this study. Also, we assumed that all available postmatch positions were given to non-USMDs. Since 9% of all graduating allopathic students are theoretically available in the postmatch period, it is reasonable to assume that some postmatch positions are being filled by USMDs. For a given specialty, this would probably represent a very small percentage of their positions; this assumption also does not alter the prematch findings of this study. Finally, the data in the AAMC Data Book⁵ are entered online by the training programs through GME Track Census and may overstate or understate the number of positions offered and/or filled. It is the official source of the data used by the AMA and the AAMC, and it is unlikely that programs would willingly provide inaccurate information.