Discussion
Early Years before the Ophthalmology Match (1916–1979)
Until 1979, the method used by ophthalmology departments to select new resident trainees
had remained unchanged since ophthalmology graduate medical education began. Applicants
contacted programs at any point during their medical school years. Programs would
interview students and offer positions that would commence two, three, or even more
years later. These positions would be contingent upon the applicant first graduating
from medical school and completing an internship, after which ophthalmic training
as a postgraduate year (PGY) 2 physician would begin. For some, the lengthy span between
offer and start date related to military obligations to be fulfilled after medical
school. But over time, it became apparent that junior and even sophomore medical students
were being offered postgraduate positions. The offers frequently included expiration
dates, colloquially known as “exploding offers,” whereby the applicant had a limited
amount of time to accept or decline. It was not unheard of for an offer to be valid
for only a few days, or even less.
Applicants were therefore faced with the option of accepting an offer, or declining
and hoping that a better offer might be extended later. Retina specialist James C.
Folk, MD, recalls securing a residency offer from Pittsburgh Eye and Ear (now University
of Pittsburgh Medical Center, Eye and Ear Institute) in the prematch era when he was
a fourth year medical student: “I applied in the fall of 1976 and got an immediate
interview because they had an opening. They offered a spot at the end of the interview”
(email, February 8, 2021). Folk accepted the offer without applying to, or interviewing
with, any other residency programs (J.C. Folk, MD, email, February 8, 2021).
Establishment of NRMP in 1953 Becomes Model for Ophthalmology Match
The process for obtaining ophthalmology residency positions prior to 1979 mirrored
the process for internship positions before 1953. In that year, the Association of
American Medical Colleges (AAMC) instituted a matching program to pair applicants
with internships, with a process that offered structure and fairness to applicants
and programs alike.
The compelling origin story of the NRMP has been well-documented.[1]
[2]
[3] Briefly, the AAMC proposed a matching algorithm that was tested for one application
cycle; but when the algorithm was introduced, medical students rebelled.[3]
[4] Led by W. Hardy Hendren III of Harvard University, the students asserted that the
algorithm favored the programs substantially.[4] A nationwide committee of medical students developed a better matching algorithm,
which the AAMC accepted and used thereafter.[4]
The NRMP has been a model for the ophthalmology match. Since the NRMP's inception
in 1953, many developments have improved graduate medical education. For example,
the Medicare Act of 1965 markedly increased graduate medical education spending.[5]
[6] Additionally, the formation of the American Board of Ophthalmology (ABO) in 1916,
with its work establishing standards and certifications for trainees, has promoted
excellence in the field of ophthalmology.[7] Most important was the ABO requirement that, to become Board certified, a doctor
must devote a specified period of time solely to ophthalmic graduate medical education.[8]
The 1962 Gale–Shapley Matching Algorithm
In 1962, economists David Gale, PhD, and Lloyd Shapley, PhD, described an algorithm
for matching entities from two groups equitably as an attempt to solve the “stable
matching problem.”[9] They did this unaware that the NRMP had, for many years, already been performing
matches according to the work they theorized.[10] When the economists later learned of this, they agreed that the NRMP had already
utilized this type of matching.[11] In 2012, Shapley and colleague Alvin Roth of Harvard University received the Nobel
Prize in economics for work on matching algorithm development, Gale having died in
2008.[10]
Ophthalmology Match Founded by Bruce Spivey, MD, and August Colenbrander, MD, PhD
(1979)
Ophthalmologist Bruce Spivey, MD ([Fig. 1]), was instrumental in founding the ophthalmology match. Born in 1934 in Cedar Rapids,
Iowa, Spivey's training included medical school and residency at the University of
Iowa, a pediatric ophthalmology fellowship and a Master's in Education degree from
the University of Illinois. Spivey went on to become chairman of the ophthalmology
department at Pacific Medical Center (now California Pacific Medical Center) in San
Francisco in 1971. In his capacity as ophthalmology department chairman, Spivey was
able to attend meetings of the AUPO, which at the time limited membership to include
only department heads.
Fig. 1 Bruce E. Spivey, MD, in the early 1970s (photo with permission of Bruce Spivey, MD).
Spivey realized, in discussions with other department leaders, that the resident selection
system had drawbacks limiting both programs and medical students. Applicants were
pressured to commit to a career in ophthalmology early in their medical school years,
to forgo other options, and to take the “bird in the hand” offer. Programs were pressured
to extend offers early, to attract high-quality students. This timetable often made
it difficult to evaluate the professional potential of inexperienced students who
were barely entering their clinical years. At that time, the AUPO concentrated on
improving the resident selection process, yet never considered a matching program.
Awareness of these limitations motivated Spivey to work, over the course of several
years, to persuade a majority of department leaders that a matching program would
be prudent. The structure would include interviewing rising senior medical students
early in autumn. Students would generate a rank order list of programs that they wished
to attend, and programs would create a list of students they would be willing to have
as residents. The two lists would then be processed through a matching algorithm.
The goal was to match each student with the program they ranked highest that wished
to have them as residents. This would be accomplished well before the NRMP match so
that applicants could make suitable plans for their PGY-1 year, or alternate career
plans, if they failed to match into an ophthalmology program.
The proposed ophthalmology match was initially met with pushback from some ophthalmology
residency programs. Programs considered more prestigious thought that a match system
would diminish their prominence by leveling the differences in status among programs.
Programs that were smaller or considered less prestigious also feared being disadvantaged,
concerned that students would only apply to larger programs. Spivey and others worked
hard to convince all programs of the benefits. These included being able to assess
candidates' abilities more accurately as rising seniors, further along in their studies
at the time of selection than under the traditional system. Spivey also noted that
there had always been more candidates than positions nationwide, which would allow
smaller and lesser-known programs to draw from a larger pool of candidates. He told
the larger, better-known programs that if they were truly strong departments, then
they would have no problems filling their residency slots with excellent applicants:
“You should be able to recruit against the smaller programs” Spivey recalled telling
them (oral communication, February 16, 2021). Frederick T. Fraunfelder, MD, who served
as AUPO president during this time, recalled that persuading uncertain department
heads to support the ophthalmology match was best done at the AUPO annual meetings,
usually in smaller, private conversations during social events (oral communication,
March 4, 2021).
In the fall of 1977, the AUPO voted to sponsor a matching program to begin the following
year, for students graduating in the spring of 1979.[12] Spivey recalls that the vote was far from unanimous, “not even 90%” (oral communication,
February 16, 2021), but that all programs elected to participate once the majority
vote determined that the ophthalmology match would be established. The fact that there
were more small programs and that these supported the match helped get the idea approved.
Spivey oversaw the ophthalmology match while ophthalmologist August “Gus” Colenbrander,
MD, PhD ([Fig. 2]), created the matching algorithm and served as program coordinator.[12]
Fig. 2 August Colenbrander, MD, PhD, in 2016 (photo with permission of the Department of
Ophthalmology, California Pacific Medical Center).
Colenbrander, an ophthalmology colleague of Spivey at Pacific Medical Center, had
a background in mathematics and computing, ideal for the development of a matching
algorithm. Colenbrander was born in 1931 and raised in Leiden, the Netherlands, where
his father was Professor and chair of the ophthalmology department. Colenbrander completed
medical school at Leiden University, the Netherlands, in 1959, later completing his
residency in ophthalmology and receiving his PhD at Utrecht University in 1964.[13] When asked by his department chief to work on a new disease coding system, Colenbrander
put the information on punch cards and learned about computers, cutting-edge technology
for the 1960s.
Colenbrander gave a talk about this work at a European ophthalmology meeting where
he met Spivey, who invited him to tour U.S. ophthalmology departments including his
own at the time, at the University of Iowa. This led to Colenbrander becoming a visiting
Professor at the University of Iowa beginning in 1969, and then relocating permanently
to the United States with his wife and two children. In 1971, when Spivey became chair
at Pacific Medical Center, Colenbrander moved to San Francisco as well, working at
Pacific Medical Center for more than 25 years, specializing in low vision and becoming
director of the Vision Rehabilitation service.[13]
Colenbrander, a gifted mathematician, was the ideal candidate to serve as program
coordinator for the ophthalmology match and develop its matching algorithm. He developed
the algorithm on his own, saying that he had never read about or in any way studied
either the NRMP match or Shapley's game theory. Colenbrander said that it was simply
an “iterative process”; he had never before thought of residency selection or matching
strategies (oral communication, February 10, 2021). The original code was a “spaghetti
code” but was later switched to a more stable Structured Query Language (SQL) code
(A. Colenbrander, MD, PhD, oral communication, February 10, 2021).
The first ophthalmology matches were performed manually by Colenbrander, who enlisted
the help of several Pacific Medical Center departmental secretaries. This was tedious
work: as the assistants could only tolerate the work for a few hours a day, it took
several weeks to complete. While neither Spivey nor Pacific Medical Center received
any money from the AUPO for this work, eventually “Gus got some money” (B.E. Spivey,
MD, oral communication, February 16, 2021). The money, Colenbrander recalled, came
from Pacific Medical Center (oral communication, February 10, 2021).
Early Years of the Ophthalmology Match
The first years of the ophthalmology match went quite smoothly. All accredited programs
participated, with few complaints and no ill-will. Spivey felt that it “helped that
ophthalmology was and is a collegial field” (oral communication, February 16, 2021).
Colenbrander recalled that a few programs attempted a work-around whereby they submitted
rank order lists with the same number of applicant names as positions available, and
that those applicants submitted rank order lists naming only that one particular program.
This seemed like an obvious attempt by the program to circumvent the match by promising
those applicants a position, clearly beyond the spirit of the match. Spivey and other
AUPO members held private discussions with those department leaders to persuade them
to “play by the rules” (B.E. Spivey, MD, oral communication, February 10, 2021).
It helped that from the very beginning, there were more applicants than positions.
While the AUPO thought that the ratio would be approximately 2 to 3:1, Colenbrander
recalled it actually being closer to 1.5:1 (oral communication, February 10, 2021).
In these early years, Colenbrander would call programs with some unfilled positions
and ask them to add more names to their rank order list. Colenbrander also made sure
to present data each year at the annual AUPO meeting. He recalled that “people liked
that I came to the AUPO and talked to them about how the match worked. I wasn't the
chairman so I wasn't a threat to anyone. Bruce (Spivey) did the politics, I did the
match” (oral communication, February 10, 2021).
AAMC Effort in the 1980s to have NRMP Take over Ophthalmology Match
The ophthalmology match has encountered several obstacles since inception. Robert
E. Kalina, MD, who served as AUPO president in the early years of the ophthalmology
match, recalled that the AAMC routinely asked for the AUPO to relinquish authority
over the ophthalmology match to the NRMP (oral communication, March 18, 2021). Spivey
attended AAMC meetings and recalled “getting a lot of sniping from them” about the
ophthalmology match (oral communication, February 16, 2021). For instance, at their
September 1983 meeting, the AAMC board discussed in detail their desire to usurp control
over the ophthalmology match.[14]
1980s' Ophthalmology Match Algorithm Challenge Prompts Change
A second early obstacle came in the form of objections to the actual match algorithm.
In 1981, Kevin Jon Williams, MD, then an internal medicine resident at the University
of Chicago, realized that both the NRMP and the ophthalmology match algorithms slightly
favored the program over the applicant.[15] This was a subtle mathematical issue. Williams spent many years attempting to publish
his observation so that it would gain recognition by the NRMP. A series of articles,
first in the New England Journal of Medicine in 1981 and 15 years later in Academic Medicine. It was only after the latter publication, in 1996, that Colenbrander learned of
the issue and reviewed his own algorithm. He realized that it did indeed slightly
favor the programs.[16]
[17] Colenbrander immediately changed the algorithm for the ophthalmology match to slightly
favor the applicant, something that the NRMP did not do for some time.[18]
Colenbrander recalled that he reevaluated the prior 5 years of the match with the
new algorithm to see if the old and new algorithms produced different results; his
reevaluation demonstrated that only a few subtle changes would have occurred (A. Colenbrander,
MD, PhD, oral communication, February 10, 2021). Colenbrander thought that the NRMP
was slow to change because “they didn't have people who fully understood programming”
(A. Colenbrander, MD, PhD, oral communication, February 10, 2021).
Ophthalmology Match Survives 2002 Antitrust Challenge
In 2002, a threat to the ophthalmology match loomed when a group of Family Medicine
residents filed a lawsuit against the AAMC, claiming that the NRMP violated antitrust
laws.[19]
[20] While the ophthalmology match was not a party to the suit, it was clear that a judicial
ruling dissolving the NRMP would set a precedent applicable to the ophthalmology match,
which would be halted.
At the heart of this class action lawsuit, Jung v Association of American Medical Colleges,[19] were allegations that the NRMP imposed anticompetitive restraints on medical residents
in violation of the federal antitrust laws embodied in the Sherman Act.[19]
[20]
[21]
[22]
[23]
[24] The plaintiffs alleged that the AAMC required medical students to use the NRMP to
obtain graduate medical education and, concomitantly, accept offers, wages, and working
conditions without appropriate ability to negotiate terms[22]; limited overall U.S. training positions; limited residents' abilities to transfer
employment to a different program; and allowed programs to share information and enter
into agreements that allowed them to keep residents' salaries low and in some instances,
suppress the right to negotiate.[19]
[20]
[21]
[22]
[23]
[24]
Drs. Kalina and Spivey monitored the lawsuit carefully due to their relationships
with the AAMC. At that time, Spivey was president of the Council of Medical Specialty
Societies which was “very afraid of the suit” (B.E. Spivey, MD, oral communication,
February 16, 2021). Many legal experts expected the plaintiffs to prevail in court
on the theory that the NRMP violated antitrust principles of the federal Sherman Act
by entering into collusive agreements that suppressed competition regarding residents'
salaries and hours.[22] Other legal analysts envisaged that the context of medical education and professional
tradition might impact the case, and argued for a legislative solution; indeed, the
antitrust laws themselves contemplate exemptions.[20]
[22]
A fortuitous, serendipitous event led to just such a legislative solution. W. Hardy
Hendren, III, MD, who had led the initial battle to establish a more fair internship
match as a medical student in 1953, was now a renowned pediatric surgeon at Harvard
University. Hendren and a colleague happened to meet their state's senior U.S. Senator
Ted Kennedy one weekend on the Boston docks after they had all been sailing. The two
physicians explained the dilemma to Senator Kennedy and extolled the virtues of the
NRMP.[4]
Shortly thereafter, Kennedy and Senator Judd Gregg of New Hampshire undertook a federal
legislative initiative to protect the NRMP.[19]
[23]
[25] Specifically, the Senators added a rider to an unrelated federal spending bill,
the Pension Funding Equity Act of 2004, expressly stating that the NRMP was exempt
from the antitrust law, and that, moreover, such exemption was to be retroactive.[23]
[25] Some objected to the rider on the basis that it had not been adequately debated.[23]
The bill, with the attached rider, was signed into law by President George W. Bush
in April of 2004.[25] The lawsuit against the NRMP was dismissed shortly thereafter, in August of 2004,
with the court ruling that the recently enacted federal legislation exempted the NRMP,
and all graduate medical education programs, from the antitrust law claims.[19] As such, the validity of the NRMP was assured by federal legislation[25] and confirmed by the judicial ruling.[19]
[24]
The San Francisco Residency and Fellowship Matching Services (SF Match)
In the early 1990s the matching process became more formalized as the San Francisco
Residency and Fellowship Matching Services organization (T.R. Losch, CMP, email, August
2, 2021). This organization, and the ophthalmology residency and fellowship match
program it administers, are each known as “SF Match.”[26] The program is now considered an affiliate of the American Academy of Ophthalmology
(AAO) and is run out of the AAO office. SF Match has an executive committee composed
of ophthalmologists from the AUPO and SF Match director Timothy R. Losch, CMP (T.R.
Losch, CMP, oral communication, April 1, 2021). The SF Match sees the AUPO as a “client”
and these two organizations work together closely, continuously striving to improve
the matching process (T.R. Losch, CMP, oral communication, April 1, 2021). For example,
in 1985, ophthalmology fellowships were added to the ophthalmology residency matching
program. Later in the 1980s, the introduction of a common application eliminated the
need for applicants to complete unique forms for each program. In the 1990s, the match
began to embrace digital technology that streamlined the ophthalmology match application
process and was less reliant on paper-based methods.[12] SF Match engaged Xerox to develop a computer program to register documents submitted
by applicants as they arrived. By 2011, the entire process was completed online. The
application, the submission of the rank order list, and the results were now performed
electronically. For the first year of the online process, the plan to simultaneously
email all applicants their match results on the posted date ended with a failure of
the SF Match email system (T.R. Losch, CMP, oral communication, April 1, 2021). Applicants,
such as Christy Benson, MD, anxiously awaited emails that never came (Christy Benson,
MD, oral communication, March 5, 2021). Later that day, the match results were sent
via the traditional system: a FAX to each medical school's Dean's office from which
the results could be relayed to the applicants (T.R. Losch, CMP, oral communication,
April 1, 2021).
Contemporary Trends and Future Directions
In recent years, the matching process has met challenges and opportunities related
to life in the Internet era. Social media now allows students everywhere to communicate
with each other via online forums or communities such as those found on Reddit, Twitter,
Facebook, or other digital platforms. Now, applicants may easily share information
and impressions about residency programs in a manner not traditionally possible.
For many years, SF Match has made statistical information publicly available shortly
after each year's results were known. This includes information such as number of
applicants; number of positions; average number of applications per position; average
number of interviews per applicant; average U.S. Medical Licensing Examination (USMLE)
step one score of successful and unsuccessfully matched candidates; and the percentage
of those matched who received their first choice, second choice, third choice (and
so on) program. This information created a situation whereby applicants knew that
not all would find a position, and that increasing the number of programs one applied
to might increase the odds of success.
As a result of this information and online discussions about the increasingly intense
competition for limited positions, the number of programs applied to per applicant
soared. For example, in 1994, 639 applicants applied to 442 available positions, with
the average applicant applying to 41 residency programs.[27] In 2021, 677 applicants applied to 499 available positions, with the average applicant
applying to 79 residency programs. The average USMLE step one score of matched applicants,
often used by programs as an initial device to screen applicants, had risen from 228
in 1994 to 245 in 2021.[26] By creating a common application, the SF Match had simplified the task of the applicant
who could now easily apply to as many programs as desired simply by paying a modest
fee to do so.
Some observers note that, at least in the more general NRMP, simply increasing the
number of programs applied to did not materially increase the odds of matching into
a desired program.[28] Moreover, the increase in the number of programs applied to per applicant, a phenomenon
known as “overapplication,” increases the workloads of departments, which may need
to screen hundreds of applications to select a manageable number for interviews. Nevertheless,
many applicants feel compelled to apply to many programs and accept as many interviews
as possible, a time-consuming and costly endeavor, particularly given the constraints
of typical medical students.
In recent years, the AUPO and SF Match have continued to work closely together. The
AUPO has a “Match Oversight Committee” that interacts extensively with SF Match. While
no residents or medical students sit on the committee, the AUPO conducts detailed
surveys of selected applicants to gain data on their experiences with the match and
entire application process, according to Misha Syed, MD, AUPO Program Directors Council
member (oral communication, March 24, 2021).
The COVID-19 pandemic forced changes to the application system for the 2021 ophthalmology
match. In-person interviews were discontinued and replaced with virtual video interviews.
To prevent applicants from monopolizing the limited interview spots available, SF
Match, in conjunction with the AUPO, set a limit of 20 interviews per applicant, with
SF Match coordinating interview scheduling. As the results of this new system seemed
agreeable to most applicants and programs, it was continued for an additional year.
Whether or not this system will stay in place is unknown, although the rising proportion
of digital natives as physicians and increasing role of technology suggest that as
the match continues to evolve, it will embrace digital trends.[29]
SF Match also administers a residency match for plastic surgery and numerous fellowship
matches in ophthalmology, otolaryngology, anesthesiology, and other fields.[26] Future directions for the ophthalmology match may include a couples match, analogous
to that of the NRMP, which would allow married couples or other partners to seek residency
programs in the same geographic area.[30] Others have suggested using a slightly different algorithm, claiming that doing
so would improve match results for both applicants and programs.[31]
The success of the ophthalmology match is indeed laudable. Several specialties tried
to establish matching programs in the 1960s but did not succeed, most notably psychiatry[32] and pediatrics. These failed soon after initiation, mainly due to a lack of full
cooperation by the departments.[33] As such, the success of the ophthalmology match as the first successful specialty
medicine match stands as a notable achievement that has benefitted programs and applicants
alike. Today, the ophthalmology match is thriving, offering broad access to applicants
nationwide, promoting geographic and demographic diversity among medical residents.
The ophthalmology match continues to have a significant impact on the field of ophthalmology
by promoting excellence in graduate medical education.
