In this issue of Academic Medicine, Ray, Bishop, and Dow1 propose changing the National Resident Matching Program (NRMP) to address problems they perceive with the Match. They suggest that applicants and residency programs should directly negotiate with each other in a free market to fill residency positions on a rolling basis, whenever applicants are ready to begin graduate medical education (GME) rather than after four years of medical school. In response, we consider three questions: How valid are their reasons for changing the Match? What is the likelihood that this free-market approach would succeed? What alternatives for improving the Match are available?
Validity of Reasons for Changing the Match
Evidence both supports and challenges the reasons Ray and colleagues cite in their proposal for modernizing the Match: the rise in applicant and application numbers and the imbalanced ratio of applicants to residency positions; the ensuing consequences of the increased applications and uneven ratio; changes that undermine the algorithm that drives the Match; and the “July effect.”
Applicant and application numbers and their consequences
As Ray and colleagues note, there are more total applicants than available residency positions, contributing to an impressive rise in application numbers as applicants apply to multiple programs to secure a position and to an imbalance in the ratio of applicants to positions. Notably, an imbalanced ratio has existed in the Match since the 1980s.2 However, in 2016 the ratio of positions per active U.S. senior was 1.53, the second highest since 1976 after a high of 1.54 in 2014.2 Additionally, in 2016 the match rate among U.S. seniors of 93.8% was in the historical range of 92% to 95%, and 79.2% of U.S. seniors matched to one of their top three choices.2 Ray and colleagues also suggest the Match disadvantages couples and other applicants with complex choices. Yet, in 2016 the match rate to PGY-1 positions for couples was 95.7%, the highest on record.2 Admittedly, exceptions to these rates do occur in competitive specialties. As for residency programs, the 2016 fill rate was 96.2%.2 In our view, the 2016 Match results do not support the authors’ contention of a broken system but, rather, indicate it sufficiently accommodates the preferred choices of both applicants and programs, as the NRMP observes.2
We concur with Ray and colleagues that negative consequences result from the rise in applicants and applications. The literature also describes a number of these consequences.3 , 4 Applicants see the Match as highly competitive. Consequently, they submit more applications, interview with more programs, feel more pressure to audition for residency spots through away-electives, and spend more money on the process than previously. They divert their attention away from their fourth-year studies, and the quality of their education may decline. Similarly, residency programs, desirous of filling their positions, interview more applicants and perhaps spend limited time and resources on less-interested applicants. Inundated with applications, programs may make decisions about applicants based on limited information from standardized assessments (e.g., the United States Medical Licensing Examination) designed for another purpose, rather than on broad-based assessments. Applicants, too, may make choices without a full understanding of the critical information necessary to determine whether a program is a good fit for them.
Ray and colleagues view the imbalance of applicants and positions, along with the imperfect information that programs and applicants have about each other, as violating assumptions underlying the NRMP algorithm. But a more recent report5 than the authors’ source6 has shown that relative numbers of applicants and positions do not affect the production of a stable match, the marker of the algorithm’s success.5 Moreover, equal numbers of applicants and positions have rarely existed,2 not at the outset of the Match when it was hailed a success nor at the time research on the original algorithm and its revision (to better meet the needs of applicants, including couples) received the Nobel Prize for Economic Sciences.7 It also seems likely that programs and applicants have rarely had full information about each other to shape their decision making. We therefore disagree with the authors’ claim that the algorithm’s assumptions are not being met.
The “July effect”
Ray and colleagues acknowledge that the “July effect”—a perceived deterioration in patient outcomes that coincides with the July 1 start date of new residents—is controversial. Nevertheless, as part of their proposed free-market approach, they advocate rolling start dates for GME. Studies have not consistently found a “July effect,” and those that have done so have suffered a significant limitation8: failure to control for the level or type of supervision provided to residents. We believe that altering the Match to address an unconfirmed phenomenon is ill advised. Instead, we recommend that programs improve supervision of residents, and we encourage programs to implement Ray and colleagues’ suggestion of assigning clinical duties according to residents’ capabilities, not their year of training.
In short, we believe the reasons prompting Ray and colleagues’ proposal can, on the whole, be challenged.
Likelihood of Success
Despite the questionable validity of the authors’ stated reasons for adopting a free-market approach to matching applicants and programs, we applaud Ray and colleagues’ effort to improve problems with the Match. Accordingly, we explore their proposal’s likelihood of success.
Application numbers and associated negative effects
The merit of the proposed approach lies in its potential to reduce the numbers of applications and interviews students feel they must complete as well as the numbers of applications and interviews programs feel they must entertain. The anticipated reduction in student and program workload could alleviate some current issues, such as stress and inefficient use of time and financial resources. Achieving these advantages, however, would depend on the following circumstances. First, students would have to limit the number of programs to which they apply. Second, programs would need to find the initial applicant pool satisfactory and promptly respond to applicants with a rejection or an interview invitation. Third, programs would be required to make offers—and applicants to accept or decline them—within a limited time frame. Whether students and programs would behave accordingly is unknown.
Reliance on limited information
With the potential for fewer applications and interviews, the proposed free-market approach could permit applicants and programs to get to know each other better in several ways. As Ray and colleagues envisage, applicants who receive rejections early could modify their strategy to increase their chances of gaining a residency position. Similarly, programs unable to fill positions could alter their tactics midstream and, in an effort to choose applicants more effectively, answer the call for more holistic assessment of applicants.9
However, programs and applicants, we surmise, could still encounter constraints on obtaining better information. Ray and colleagues’ notion that programs could improve their benefit or salary offers midstream would be difficult to put into practice. Budgets are usually set annually without much leeway for exigencies, and variations in offers could quickly breed discontent among residents within the program. Additionally, implementation of reliable, broader assessment methods would require expertise, trained examiners, and financial resources. Administration of multiple mini-interviews or simulation-based exercises throughout the year would necessitate considerable administrative prowess and could diminish the validity of these measures. Further, students would still have a finite amount of time for audition electives and interviews, and programs would still have limited space and time for interviews. The projected change might exacerbate these existing problems or introduce new ones. For example, students could feel pressured to audition sooner, compromise their curriculum earlier, and miss out on the benefits of a rigorous final year.10 Programs could decide to choose students they already know—from their own institution or from audition electives—and thereby decrease diversity among their residents.
Applicants with complex choices
Couples and applicants applying to preliminary and advanced programs might find the proposed free-market process more straightforward. However, the approach might have undesirable consequences for military applicants and the programs interested in them because these applicants must postpone their decision making until they obtain waivers from the military. Also, students who apply to programs in their preferred specialty as well as to backup programs in a less desired specialty might prematurely settle for a guaranteed spot in the latter.
Ray and colleagues suggest their proposal could help lessen the predicted shortage of physicians in the United States, including shortages in primary care and psychiatry.11 Achieving this goal under their proposal would require a steady stream of students and residents moving through the educational continuum, at differing rates, in less time than a traditional curriculum stipulates. Ultimately, the rate-limiting step to accommodating the education of more residents within a program at any one time is the restriction placed by Medicare caps or other funding on the number of positions allocated to residency programs. Indeed, it is the insufficient number of residency positions that is at least partly responsible for a physician shortage and for several issues with the Match, described above.
In summary, we envision that the authors’ proposal could succeed to an extent but only under certain circumstances and not without creating its own problems.
Barriers to Success
We see several barriers that prompt us to consider Ray and colleagues’ proposed approach as premature.
The success of Ray and colleagues’ proposal fundamentally depends on the state of competency-based education. However, competency-based education is not yet sufficiently developed12 to support the proposed modification of the Match. Most pressing for the success of the authors’ proposal is the need for criteria to determine a learner’s readiness for residency or unsupervised practice other than time spent in a formal education program. To our knowledge, a comprehensive set of valid assessments that would certify the necessary competence, regardless of time spent, is not yet available, although work toward that end is progressing—for example, with specialty milestones in GME13 and entrustable professional activities in undergraduate medical education (UME).14 Substitution of demonstrable competence for time spent would also require approval by accrediting bodies, specialty boards, and state licensing agencies—all of which currently stipulate the length of GME and UME programs.
To create the steady stream of graduates competent to begin residency throughout the year required by the proposal, medical schools would need not only to develop competency-based curricula further but also to solve major logistics issues. For example, in the absence of knowing when and how many students will qualify for graduation during a year, planning for needed resources—such as faculty coverage, clinical site capacity, tuition available, and first-year class size—becomes unpredictable. Perhaps medical schools, too, would have to contemplate rolling admissions. Ray and colleagues characterize obstacles as “neither minor nor insurmountable.”1 We view these obstacles as far from minor, especially for faculties already overwhelmed by the complexities of competency-based education.12
Ray and colleagues are silent about how the process of finding and filling residency positions would be managed. We wonder how the NRMP would be involved and what agency, if any, would ensure that abuses of medical students, which once characterized the process before the Match, do not reemerge.
The literature contains several potential solutions to problems with the Match.3 , 4 Among the more promising, in our view, are the following:
- Partially restrict application numbers by implementing multiple Match rounds with limits placed on application numbers for the first few rounds and retaining the Supplemental Offer and Acceptance Program for the final round;
- Standardize dates for programs to invite applicants for interviews and for students to reply;
- Deemphasize match rates as a quality measure in UME and board pass rates in GME accreditation;
- Begin student advising early and offer more accurate advice about applicants’ choices in relation to their qualifications and abilities;
- Provide accurate and more transparent information about programs; and
- Assess applicants holistically.
Increasing the number of residency positions is a more fundamental solution worthy of concerted attention and action.
As we noted above, evidence challenges Ray and colleagues’ stated reasons for changing the Match. In response, we ask: Why change the Match when it is not as flawed as the authors imply? Although problems exist with the Match, the proposed changes might not eliminate them; rather, the changes could exacerbate some and introduce others. The critical barrier to the proposal’s success is the immaturity of competency-based education. We conclude that the time for this proposal has not yet arrived but encourage continuing discussion of it along with more immediate practical solutions and ways to increase the number of residency positions.
1. Ray C, Bishop SE, Dow AW. Rethinking the Match: A proposal for modern matchmaking. Acad Med. 2018;93:45–47.
3. Gliatto P, Karani R. Viewpoint from 2 undergraduate medical education deans: The residency application process: Working well, needs fixing, or broken beyond repair? J Grad Med Educ. 2016;8:307–310.
4. Berger JS, Cioletti A. Viewpoint from 2 graduate medical education deans: Application overload in the residency match process. J Grad Med Educ. 2016;8:317–321.
5. Peranson E, Randlett RR. The NRMP matching algorithm revisited: Theory versus practice. National Resident Matching Program. Acad Med. 1995;70:477–484.
6. Gusfield D, Irving RW. The Stable Marriage Problem: Structure and Algorithms. 1989.Cambridge, MA: MIT Press.
8. Young JQ, Ranji SR, Wachter RM, Lee CM, Niehaus B, Auerbach AD. “July effect”: Impact of the academic year-end changeover on patient outcomes: A systematic review. Ann Intern Med. 2011;155:309–315.
9. Prober CG, Kolars JC, First LR, Melnick DE. A plea to reassess the role of United States Medical Licensing Examination Step 1 scores in residency selection. Acad Med. 2016;91:12–15.
10. Richards CJ, Mukamal KJ, DeMelo N, Smith CC. Fourth-year medical school course load and success as a medical intern. J Grad Med Educ. 2017;9:58–63.
12. Sklar DP. Competencies, milestones, and entrustable professional activities: What they are, what they could be. Acad Med. 2015;90:395–397.
13. Li ST. The promise of milestones: Are they living up to our expectations?J Grad Med Educ. 2017;9:54–57.
14. Lomis K, Amiel JM, Ryan MS, et al. Implementing an entrustable professional activities framework in undergraduate medical education: Early lessons from the AAMC Core Entrustable Professional Activities for Entering Residency pilot. Acad Med. 2017;92:765–770.