Relationship Between Residency Match Distance From Medical School and Virtual Application, School Characteristics, and Specialty Competitiveness

Introduction: The COVID-19 pandemic has disrupted the residency match process by eliminating away rotations and changing from in-person to virtual interviews. In this study, we explore the impact of the COVID-19 pandemic on the geographic match distance of United States (US) senior medical students across all specialties. Methods: We collected publicly available student match data between 2018 and 2021 from US allopathic medical schools and calculated match distance between medical school and residency training using a novel metric - the “match space.” Match space was codified by whether the student matched at their home institution, home state, adjacent state, same or adjacent US census division (non-adjacent state) or skipped at least one US census division. Adjusting for covariates, ordinal logistic regression correlated school and specialty characteristics with match distance pre- and post-pandemic for all specialties. We defined and ranked specialty competitiveness using predictive values from factor analysis. Results: A total of 34,672 students representing 66 medical schools from 28 states matched into 26 specialties in 50 states and Canada. Fifty-nine percent of students were from public institutions, and 27% of schools ranked in the top 40 for research. The mean percentage of in-state students by school was 60.3% (range 3-100%). Match space was lower after the pandemic (adjusted odds ratio (OR) 0.94, 95% CI 0.90-0.98; p=0.006), from schools with higher in-state percentages (OR 0.74, 95% CI 0.72-0.76), from top National Institutes of Health-funded institutions (OR 0.88, 95% CI 0.85-0.92), from the Northeast (OR 0.71, 95% CI 0.67-0.75; Midwest reference), and the West (OR 0.67, 95% 0.60-0.74). Match space was higher for students graduating from private schools (OR 1.11, 95% CI 1.05-1.19), from the South (OR 1.62, 95% CI 1.2-1.33), and matching into more competitive specialties (OR 1.08, 95% CI 1.02-1.14). The top five most competitive specialties were Plastic Surgery, Neurosurgery, Dermatology, Orthopedic Surgery, and Otolaryngology. Internal Medicine ranked eighth. Conclusions: After the COVID-19 pandemic, students graduating from US allopathic schools matched closer to their home institution. Students attending public schools, schools with more in-state matriculants, and schools with higher research rankings also matched closer to their home institutions. Specialty competitiveness and US census region also impacted match distance. Our study adds insight into how geographic match patterns were influenced by school, specialty choice, and the pandemic.


Introduction
In the United States (US), senior medical students and residency programs utilize the National Resident Matching Program (NRMP) to interview and fill residency positions each year [1]. The COVID-19 pandemic has disrupted this process by eliminating away rotations and changing from in-person to virtual interviews [2]. Despite efforts to optimize the virtual experience, it remains challenging for programs to gauge genuine applicant interest and for applicants to obtain a sense of a program's culture and environment [3][4][5]. The transition to a virtual residency match has yielded significant benefits for applicants through reduced travel costs and more schedule flexibility, yet the reduced cost and general process uncertainty have led applicants to apply and interview with more programs, resulting in record high-rank order list lengths for programs and applicants alike [6,7].
These changes have exacerbated the pre-existing trend of application inflation, which led the Electronic Residency Application Service (ERAS®) to introduce a supplemental application for a growing number of specialties since the 2022 match cycle. In addition to their most meaningful experiences, applicants provide geographic preferences and "preference signals" to identify a limited number of preferred programs [8]. These changes reflect the growing recognition of the importance of geography as a factor in the residency match process [9,10]. We have previously shown that school and specialty characteristics impact geography; however, little is known about how adaptations to the COVID-19 pandemic have influenced residency match geography across all specialties [11]. As we conducted this study, Cotner et. al. found supporting evidence that the virtual match process would result in students staying closer to their home institutions, but their analysis was limited to the same state versus different states [12]. We and others have found varying results for different specialties: no difference in matched distance for plastic surgery [13], neurology [14], or cardiology fellowship [15]; for otolaryngology [16], dermatology [17], and orthopedic surgery [18], students matched closer to their home institution.
Our study aimed to investigate pre-and post-pandemic trends in the geographic placement of senior allopathic medical students for post-graduate training and to determine which factors impacted where applicants matched across all specialties. To do so, we utilized a metric that we called "match space" to quantify the non-linear geographic relationship between medical schools and residency programs [11].
Abstracts of earlier versions of this study were presented at Association of American Medical Colleges (November 8-10, 2021, virtual meeting), the Southern Society of General Internal Medicine Regional Meeting (February 9-10, 2022, virtual meeting), the Society of General Internal Medicine, Orlando, FL (April 6-9, 2022), and the Alliance for Academic Internal Medicine (AAIM), Academic Internal Medicine Week 2022, Charlotte, NC (April 11, 2022).

Study design, setting, and participants
We include the post-COVID-19 matched data to our prior publication with pre-COVID-19 matched data [11]; see the conceptual framework in Figure 1.

Residency Match Placement.
We included students graduating from US allopathic medical schools who were matched into residency programs during in-person interviews (pre-COVID-19, 2018-2020) and during virtual interviews (post-COVID-19, 2021) for all specialties. We excluded medical schools without publicly available match lists or schools with fewer than 50 graduating students. We also excluded students who matched into combined residencies, interventional radiology, oral and maxillofacial surgery, child psychiatry, non-residency fellowships, or preliminary-only positions. We collected matched residency program data for each student.
We obtained medical school characteristics from publicly available datasets. Public/private medical school designation was obtained from the Medical School Admission Requirements™ (MSAR®) online database [19]. School reputation was determined using a surrogate of the top 40 National Institutes of Health (NIH) funding based on US News and World Report rankings [20]. Finally, the percentage of in-state matriculants was obtained from the Association of American Medical Colleges [21].
Specialty characteristics including US senior applicants per position, US senior fill rate, and number of post-graduate year (PGY) positions available were collected using 2021 NRMP data [7]. The salary was collected from the 2020 annual physician compensation reports [22,23].

Statistical analysis
To examine specialty competitiveness, we used exploratory factor analysis of the specialty characteristics, retained items with Eigenvalues ≥ 1.0, and rotated factor loading > 0.4. Then, we ranked specialties based on the predictive value of the factor with the greatest Eigenvalue. We obtained Spearman rho correlations of the characteristics and assessed internal consistency reliability with Cronbach's alpha (and considered alpha > 0.9 to be excellent and > 0.8 to be good).
We use the same statistical approach as in our prior study [11]. Match space was codified by whether the student matched at their home institution, home state, adjacent state, same or adjacent US census division (non-adjacent state), or skipped at least one US census division (5-point ordinal scale). We used ordinal logistic regression to explore the relationship between the match space and the following variables: pre-and post-COVID-19 match, school type, top 40 NIH ranking, percentage of in-state matriculants, US census region, and specialty competitiveness. Instead of using the full range for continuous variables (e.g. percentage of in-state matriculants) and for a more parsimonious interpretation of the statistical analyses, we collapsed data into quartiles [11]. A positive odds ratio meant that the match space was incrementally larger on the 5-point scale. All analyses were performed using StataCorp (2021, Stata Statistical Software: Release 17, College Station, TX: StataCorp LLC) with the logit module; we used a significance level of P<0.05.

Results
We included 35,065 students (26,102 pre-COVID-19 for the match year 2018-2020; 8,963 post-COVID-19 for the match year 2021) from 66 medical schools, representing 28 states across the four US census regions and nine divisions; Table 1 shows studied variables.

Characteristic Value
Year, n (%)    Based on the ranking of the predictive value for Factor 1, the top five most competitive specialties were Plastic Surgery, Neurosurgery, Dermatology, Orthopedic Surgery, and Otolaryngology. Internal Medicine ranked eighth. The least competitive specialties were Pediatrics, Neurology, Psychiatry, Pathology, Family Medicine, and Physical Medicine and Rehabilitation, see Table 3.   Graduating US medical school seniors' match space. Values indicate odds ratios (triangles) and 95% confidence intervals adjusted by the other variables using ordinal logistic regression of match space (1 = home institution; 2 = home state; 3 = adjacent state; 4 = the same or adjacent US Census division (and non-adjacent state); 5 = skipped at least one US Census division). Asterisks indicate a P < 0.001. Variables include pre-post COVID-19 pandemic, private vs. public medical school designation, school from top 40 NIH funding, percentage of in-state matriculants (collapsed into quartiles), geographic location (vs. Midwest), and specialty competitiveness (vs. low competitiveness). Specialty competitiveness was determined based on predicted value using Factor Analysis: very high (plastic surgery, neurosurgery, dermatology, orthopedic surgery, otolaryngology), high (urology, radiation oncology, internal medicine, cardiothoracic surgery, general surgery), medium (diagnostic radiology, vascular surgery, obstetrics/gynecology, emergency medicine, anesthesiology), and low (pediatrics, neurology, psychiatry, pathology, family medicine, physical medicine, and rehabilitation).

NIH = National Institute of Health
As compared with less competitive specialties (referent), the match space was higher for students applying to medium competitive specialties (OR 1.08, 95% CI, 1.03 to 1.13) or high competitive specialties (OR 1.10, 95% CI, 1.05 to 1.16), but not for very high competitive specialties (OR 0.99, 95% CI, 0.93 to 1.07), see Figure 2.

Discussion
In this analysis of 34,672 medical students, we explored geographic trends of residency placement for all specialties before and after the onset of the COVID-19 pandemic. Using the "match space" metric to quantify the geographic relationship between medical schools and residency programs, we found an overall decrease in match space during the COVID pandemic when compared to prior years; however, match space varied significantly depending on the characteristics of the medical school and specialty.

School characteristics
For the 2021 application cycle, graduates of private medical schools were more likely to match farther from their home institution than those from public schools, and graduates of schools with higher in-state percentages matched closer than those with lower percentages. This is consistent with pre-pandemic findings and with previous years and is reassuring for public schools, which often have a mission to develop physicians for their state [24,25].
We used NIH funding as a surrogate of school reputation when considering how reputation may impact match space. In the 2021 application cycle, graduates of the top 40 NIH-funded schools matched closer to their home institution, also consistent with pre-pandemic match trends. We suspect this is because top NIHfunded schools often have strong residency options that attract internal applicants.
Finally, we examined the US census region of each medical school to determine the match distance relative to other regions. Interestingly, applicants from the South had the highest match space, whereas the Northeast and West had lower match space relative to the Midwest. This finding suggests that applicants from the South match into programs significantly farther away from their home institution compared to applicants from other regions. This may reflect the density of training programs; however, further study is needed to determine why southern medical schools appear to have a lower retention rate of their graduates.

Specialty characteristics
Specialties were categorized by competitiveness from low to very high based on exploratory factor analysis including salary, US-senior fill rate, step 1 United States Medical Licensing Examination (USMLE) average, step 2 USMLE average, AOA percentage of matched applicants, and percentage of matched applicants from a top 40 NIH-funded school. In our analysis, we found medium and highly competitive specialties match farther compared to the least and most competitive specialties. We suspect that applicants applying to the most competitive specialties are more likely to match at their home program, which may mitigate some of the effects of fewer residency programs and positions that would otherwise increase match space.

Limitations
We recognize some limitations in our study. First, we included schools with publicly available match information, which could introduce selection bias and limit generalizability; however, we obtained a large sample size of individual matches from both public and private medical schools over four years. Second, match location depends on a mutual agreement between both the program and applicant, and thus we are unable to determine how much match distance is attributable to the student or residency program. This may be an area of further study by adding student and program rank lists in addition to match lists.

Implications
Other studies have investigated the geographic impact of the pandemic with mixed results but were either limited to a single specialty or dichotomized by in-state vs. out-of-state [12,26,27]. As programs and leaders in medical education grapple with decisions about away rotations and virtual or in-person interviews, they must consider safety, cost, and equity for all applicants against the backdrop of growing application inflation [28,29]. US graduates applied to an average of 73 programs in 2021 compared to 65 in 2019, and international medical graduates applied to an average of 147 programs compared to 137 in 2019 [30]. An additional, and perhaps less recognized, factor to consider is the geographic implication of these decisions on applicants. The piloting of secondary applications within certain specialties shows a growing recognition of the importance of match geography; however, additional research and understanding are needed.

Conclusions
After the COVID-19 pandemic, students graduating from US allopathic schools matched closer to their home institution. Students attending public schools, schools with more in-state matriculants, and schools with higher research rankings also matched closer to their home institutions. Specialty competitiveness also impacted match distance; as compared with less competitive specialties, the match space was higher for students applying to medium competitive specialties (Diagnostic Radiology, Vascular Surgery, Obstetrics/Gynecology, Emergency Medicine, Anesthesiology) or high competitive specialties (Urology, Radiation Oncology, Internal Medicine, Cardiothoracic Surgery, General Surgery). We included more specialties than any other published study and more nuance in the measurement of match distance through the calculation of match space.

Additional Information Disclosures
Human subjects: Consent was obtained or waived by all participants in this study. University of Alabama at Birmingham Institutional Review Board issued approval IRB-300006047. This study was deemed exempt. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue.

Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: Disclaimer: The opinions expressed in this article are those of the authors alone and do not reflect the views of the Department of Veterans Affairs.