After graduating from a U.S. or international medical school, a physician seeking to practice in the United States must complete a period of graduate medical education (GME) to become licensed to practice in a state or other U.S. jurisdiction. Most residency positions are filled through the National Resident Matching Program (NRMP), which assigns physicians to residency positions according to the preferences of the physician and the program. Each year, the NRMP publishes statistics on matched applicants by specialty, and those tables are often used to identify trends in physician manpower. Not all new residents find their positions in the NRMP, however; a substantial number find positions in other ways. In this report, I combined data from the NRMP with data from GMETrack, a database sponsored by the Association of American Medical Colleges (AAMC) and the American Medical Association (AMA) that contains the results of a national census of GME. Because it is possible to link individual residents with records of the NRMP, those who were unsuccessful in the NRMP can be distinguished from those who did not register with the NRMP and those who registered but failed to enter a rank order list of programs.
In view of the rapidly increasing number of graduates from both the MD-granting and DO-granting medical schools in the United States in an environment where the number of funded GME positions is limited, new U.S. graduates and international medical graduates (IMGs) are finding it increasingly difficult to obtain an initial residency position.1,2 A more nuanced understanding of the market for residency positions is timely and important. Planned changes by the NRMP will make it more difficult for programs and applicants to arrange placements outside the Match. The data presented here provide a measure of the magnitude of the changes that may be expected as a result of the NRMP policy change.
A recent article by Wetz et al3 used published aggregate data to estimate the number of residents in seven specialties who found positions outside the Match. The authors of that work assumed that all outside-the-Match surgery offers were for preliminary positions and that no outside-the-Match internal medicine offers were for preliminary positions—assumptions that were disputed in an NRMP rebuttal4 published alongside the article by Wetz et al.
I carried out the research reported here to determine the numbers and types of residents in residency programs who found their positions within and outside the NRMP Match.
The AAMC maintains a data warehouse containing records of medical school applicants, students, and residents. The resident data are derived from GMETrack, a database sponsored by the AAMC and the AMA that contains the results of a national census of GME. Data from the NRMP Match also are included in the AAMC collection, with permission from the NRMP. The AAMC assigns a unique personal identifier for each person in the tables of data, allowing those tables to be linked together for longitudinal studies. In this work, I combined data on 36,941 participants in the 2009 NRMP Main Residency Match data with GMETrack data on 25,432 residents who began a first-year residency in 2009–2010 to paint a comprehensive picture of the market for first-time residents.
This report differs from that of Wetz et al3 by using individual-level data instead of aggregates and by including all specialties open to new residents without prior GME. Because applicant-specific data are available, simplifying assumptions such as those used in the prior work are not required.
The analysis is limited to first-year specialties, programs that accepted new residents and were in a specialty identified by the Accreditation Council on Graduate Medical Education (ACGME) as accepting first-year residents.5 In addition, where the ACGME indicated that some but not all programs in the specialty accepted first-year residents, the study included only residency programs in those specialties that reported at least one 2009 graduate without prior GME and who was a graduate of either an MD- or DO-granting medical school. The specialties in which some programs are reported by the ACGME to accept first-year residents are anesthesiology, dermatology, emergency medicine, medical genetics, neurology, orthopedic surgery, preventive medicine, diagnostic radiology, and radiation oncology.5 A few urology programs were found to have accepted recent graduates as well.
This study included 2,915 residency programs. Of these, 2,736 (94%) reported one or more new residents in a first-year program, whereas the remaining 179 (6%) participated in the NRMP but did not report new residents in a first-year program to GMETrack. There were 149 first-year programs reporting new residents that did not participate in the NRMP. Residents were included in the analysis if they began a new program in a first-year specialty in 2009, but they were excluded if, in the prior year, they were in a different program in the same specialty or in a preliminary internal medicine or preliminary surgery program. Data on residents placed in the military match were provided to the NRMP by the Armed Services.
Residents were classified in two ways. First, they were classified into nine categories based on their participation in the NRMP Main Residency Match and their success in matching to a position. The nine categories are
* Military (n = 332)
* Withdrawn from the NRMP, matched in the osteopathic match (n = 118)
* Withdrawn from the NRMP for lack of certification by the Educational Commission for Foreign Medical Graduates (n = 35)
* Withdrawn from the NRMP; other reasons (n = 14)
* Withdrawn by applicant (n = 1,432)
* Matched (n = 20,287)
* Participated in the NRMP, but unmatched (n = 1,325)
* Inactive (registered with the NRMP, but did not submit a rank order list of programs) (n = 247)
* Not registered with the NRMP (n = 1,642)
The NRMP cooperates with the GME programs of the Army, Navy, and Air Force by withdrawing from the Match any applicants reported by the military to have been selected for a military residency. Military residents are similar to successful Match participants in that they participated successfully in a residency placement system. Applicants also are withdrawn by the NRMP if they are reported by the American Osteopathic Association to have been placed by the osteopathic match, if they are IMGs not yet certified by the ECFMG, or if they are ineligible for any other reason. Those labeled as unmatched certified a rank order list indicating their preferences in the Match, but were unsuccessful. New residents who voluntarily withdrew did so before the Match deadline, presumably because they already had found a position independently. Finally, a substantial number of residents who began a new first-year program never registered with the NRMP.
Second, new residents were classified according to their medical school experience. The categories are
* Current-year graduates of U.S. MD-granting medical schools (graduated between July 1, 2008 and June 30, 2009) (n = 15,131)
* Prior-year graduates of U.S. MD-granting medical schools (graduated before July 1, 2008) (n = 1,026)
* Late graduates of U.S. MD-granting medical schools (graduated after June 30, 2009) (n = 15)
* Students/graduates of U.S. DO-granting medical schools (n = 2,085)
* Students/graduates of Canadian medical schools (n = 28)
* Fifth Pathway applicants (applicants who have substituted a supervised clinical year in a U.S. medical school for the year of social service required to attain a medical degree from certain international medical schools) (n = 36)
* U.S. citizen students/graduates of international medical schools (n = 2,099)
* Students/graduates of international medical schools who are not U.S. citizens (n = 5,012)
This second classification is the same as that used by the NRMP, except that the NRMP does not recognize late graduates. Late graduates were expected to graduate on time when they registered with the NRMP, but their graduations were delayed because of failure to meet one or more requirements. If a student will be unable to enter residency training on July 1 in the year of the Match, that student is not eligible to participate in the NRMP, but occasionally obstacles to graduation arise after the Match deadline.
Data on available first-year residency positions are derived from three sources:
* Residency programs respond to the National GME Census with the number of first-year positions they expect in the coming year. Data from the 2008–2009 survey are published in the 2009 education issue of the Journal of the American Medical Association.6
* Residency programs participating in the NRMP register a quota of available first-year residency positions to be filled in the Match. Those data are published each year after the Match by the NRMP, along with the number of matches for each program.7
* By counting the number of new residents reported to GMETrack, the number of filled positions for each specialty can be estimated. Bearing in mind that a few programs do not participate in GMETrack and that some programs report residents only a year or more after they have begun training, this third number will be an underestimate. The ACGME also reports the number of residents entering GME, but their data do not clearly distinguish between those who are entering GME for the first time and those with prior GME experience.8
The study protocol was approved by the institutional review board of the American Institutes for Research.
Table 1 shows the number of first-year positions in each specialty from the three sources described above. The number of positions projected a year in advance for 2009–2010 is 26,429, whereas the number of actual new residents reported to GMETrack is 24,564. There were 22,420 positions listed with the NRMP, of which 21,337 were filled in the Match.
Table 2 reports Match success and success in finding a residency position for all candidates who registered with the NRMP. Ninety-six percent of matched applicants (20,553/21,337) were reported as residents in GMETrack, whereas 24% of those who were not matched (3,790/15,634) were nevertheless reported as residents.
Table 3 displays the status in the NRMP Main Residency Match by applicant type for all new residents in specialties available to new graduates (first-year specialties). There were 24,564 first-time residents reported in a first-year program in 2009, 368 with prior GME beginning a first-year program in a new specialty in 2009, and 500 restarting GME in a first-year program in 2009 after a year in which no GME was reported. All 25,432 of these residents could have used the NRMP to seek a first-year residency. Applicants who matched in the NRMP were 79.8% of new residents (20,287/25,432). Among residents who were seniors at U.S. MD-granting schools, 93.0% (14,076/15,131) found their positions in the Match.
Of those new residents who certified a rank order list of preferred residency programs in the Match, 1,325 were unsuccessful. This unmatched group, 5.2% of new residents (1,325/25,432), consists of those who ranked at least one program in a first-year specialty; that is, they tried unsuccessfully to use the Match but found a position after the Match on their own.
Applicants who found a position in the military match, 1.3% of new residents (332/25,432), are shown separately in Table 3. Those who had registered with the NRMP were automatically withdrawn. These 332 military residents account for less than half of the 692 resident matches reported to the NRMP in 2009 by the military; presumably, the others were in programs that do not participate in GMETrack.
New residents registered with the NRMP who found a position in the osteopathic match (n = 118) were withdrawn from the NRMP, and 49 others were withdrawn by the NRMP for various reasons. There were 1,432 new residents who withdrew themselves from the Match. An additional 247 new residents registered for the Match but did not certify a rank order list. There were 1,640 new residents who never registered for the Match.
Table 4 shows matched and unmatched applicants by status in the NRMP by specialty. The percentage of new residents who found their positions outside the Match (and not through the military) varies from as little as 4% in emergency medicine to 35% in diagnostic radiology. (Note that the number of residents in each specialty in Table 4 is based on the specialty of the program reporting to GMETrack. For integrated programs such as five-year programs in diagnostic radiology, first-year residents are often reported by a cooperating transitional, preliminary medicine, or preliminary surgery program.) In the primary care fields of family medicine, internal medicine, and pediatrics, 20.6% (11,476/14,450) of new residents obtained their positions outside the Match, close to the 19.6% found by Wetz et al.3
The number of positions projected a year in advance for 2009–2010 is greater than the sum of quotas for the NRMP (see Table 1). The quotas reported here are slightly smaller than the figures in the NRMP Results and Data book because the NRMP statistic includes a handful of positions intended to fill other than first-year vacancies. There are three likely reasons for the NRMP quotas to be smaller than the projected numbers. First of all, the projections are made a year in advance, and circumstances of the sponsoring institutions may have changed. Secondly, programs participating in the NRMP are permitted to accept in advance of the Match residents who are not seniors at U.S. MD-granting schools. Some programs take advantage of that option to recruit attractive IMGs, osteopathic graduates, or prior-year graduates, and those positions are excluded from the Match. And, finally, a few programs do not participate in the NRMP.
The NRMP database includes 21,337 candidates matched to a first-year position in 2009,7 leaving only 1,083 positions unfilled.1 Most vacant positions are filled during the so-called Match week scramble, when unsuccessful Match participants try to find a position among the remaining programs with openings.
Although the total number of new first-year residents reported to GMETrack is less than projected, the actual number of new residents in 2009 is probably close to the projected number in Table 1. On the basis of past experience, an additional 3.5% of residents who began GME in 2009–2010 will be reported to GMETrack in subsequent-year surveys. Add to that an estimated 3% of first-year positions in 2009–2010 in programs that will not report to GMETrack, and the number of filled first-year positions could be as high as 26,200. (I estimated the completeness of the GMETrack database by comparing the list of accredited programs and approved positions on the ACGME Web site with programs reporting to GMETrack.)
Using the same GMETrack database but with somewhat different analytical approaches, the AMA reports 25,070 new residents in 2009 without prior GME.6
For this work, I concentrated on the 36,971 participants in the NRMP in 2009 and the 25,432 residents beginning a first-year residency for the first time and reported in 2009 to GMETrack.
The data on Match participation and Match success in Table 2 are essentially the same as reported by the NRMP,7 but they are supplemented with information from the residency tracking system. Ninety-six percent of matched applicants were reported as residents in GMETrack (20,553/21,337). A few of those not reported may have failed to honor their NRMP contracts, but most are likely enrolled in the relatively small number of programs that do not participate in the National GME Census.
It is significant that 24.2% of NRMP participants who were not matched nevertheless found residency positions (3,790/15,634). Some of those individuals failed to submit rank order lists. Others submitted rank order lists but failed to match. Still others withdrew from the NRMP after finding positions on their own. It is important to recognize that workforce studies based on NRMP data alone9 present an incomplete picture of incoming residents.
Of the 25,432 new residents in 2009 shown in Table 3, 86.3% participated in the NRMP Match or the military match (21,944/25,432), and 81.1% matched to a position (20,619/25,432). The 5.2% in the group labeled unmatched (1,325/25,432) found positions after the Match, either during the Match week scramble or subsequently.
There were 13.7% of new residents who found their positions without participating in the Match (3,488/25,432); they either withdrew, were inactive, or never registered with the Match. As can be seen from Table 3, this group includes only 0.5% of U.S. seniors (80/15,131); most are osteopathic students/graduates, international medical students/graduates, or prior-year graduates.
Of particular interest are the 1,432 new residents who withdrew themselves from the Match. Because these withdrawals occurred before the Match deadline, and because these individuals were subsequently reported as residents, one may assume that they withdrew because they were offered and accepted positions prior to the Match. The majority of this group are international medical students/graduates and osteopathic students/graduates. (Applicants who are matched in the separate osteopathic match are automatically withdrawn from the NRMP, but they would not be included in Table 3 unless the program to which they were matched is jointly accredited by the ACGME and the American Osteopathic Association.)
An additional 247 new residents registered for the Match but did not certify a rank order list. Those residents may have found a position outside the Match before the Match deadline, or they may not have found a residency program with an interest in their candidacy before the Match deadline.
The 1,642 new residents who never registered for the Match may have found positions prior to the Match or after the Match. It seems likely that many if not most of them found positions prior to the Match because there are more opportunities for them before positions are filled in the Match. If these 1,642 are combined with the 1,599 who withdrew or were withdrawn, the 1,325 who were unsuccessful in the Match, and the 247 who were inactive, a total of 4,813, or 18.9% of new residents (4,813/25,432), found their positions outside of the NRMP and military placement systems.
There are two principal limitations to this analysis. Because less than 100% of residents are reported to GMETrack, it is not possible to account for all NRMP participants and nonparticipants. I estimate that records are missing for about 6.5% of new residents. Secondly, the study is focused only on first-year residency positions, positions available to new residents without prior GME. The NRMP also includes in the Main Residency Match second-year residency positions that require a preliminary GME year in internal medicine, general surgery, or transitional year. Those second-year positions are a possible subject for future analysis.
The findings stated above make clear that workforce studies that rely on NRMP statistics to establish national trends are seriously deficient. It is important to note that the NRMP will make an important change soon. For the 2013 Match, the NRMP will require programs that participate in the NRMP to use the Main Residency Match to list all positions in NRMP-participating specialties. As a consequence, the number of new residents who are able to find positions prior to the Match may be significantly reduced. Another potential effect of the policy change is that U.S. seniors will have more choices available to them because the pool of available positions will not be reduced by pre-Match placements.
Acknowledgments: The author wishes to thank the National Resident Matching Program for making available data on the Main Residency Match. Mona M. Signer, executive director of the NRMP, provided numerous suggestions and critically reviewed the manuscript. Robert V. Wetz, MD, associate chairman of medicine and program director for the Department of Medicine at Staten Island University Hospital, also read and commented on a draft of the paper.
Other disclosures: None.
Ethical approval: The study protocol was approved by the institutional review board of the American Institutes for Research.