Virtual interviews reshape geographic diversity in OB-GYN residency matching

When the COVID-19 pandemic forced the elimination of in-person residency interviews in 2020, medical educators hoped the shift to virtual formats would democratize the residency match, allowing students from every corner of the country to compete on equal footing for training positions regardless of geography. The subsequent introduction of program signaling—an application tool designed to help applicants communicate genuine interest to a limited number of programs—was similarly expected to tame the runaway costs of applying broadly and, in doing so, perhaps reshape long-standing patterns of where doctors train. A new longitudinal study published in BMC Medical Education now offers the most comprehensive picture yet of how these reforms have actually played out, and its central finding is striking: for all the upheaval in the application process, the geographic DNA of the obstetrics and gynecology residency match has barely changed.

The study, led by Cecilia Mastrogiacomo and Mikaela R. Koch of the Renaissance School of Medicine at Stony Brook University and the David Geffen School of Medicine, together with colleagues Heidi Preis and Leah Kaufman, examined six consecutive match cycles spanning 2020 through 2025. The research team focused on a metric they call concordance: the percentage of each obstetrics and gynecology residency class that matched from the program’s own affiliated medical school, and the percentage that matched from a medical school located in the same state as the residency program. High concordance suggests that a program draws heavily from its institutional backyard; low concordance suggests a geographically diverse incoming class.

To construct this picture, the researchers turned to the Accreditation Council for Graduate Medical Education’s Fellowship and Residency Electronic Interactive Database Access, or FREIDA, a publicly available resource that catalogues residency programs across the United States. Every obstetrics and gynecology residency program with a corresponding medical school was included in the analysis, yielding a final dataset of 126 programs. For each program and each match year from 2020 to 2025, the team identified the medical schools of the residents who matched and calculated both institutional concordance—the share of new residents who attended the program’s home medical school—and state concordance, the share who attended any medical school within the program’s state. Because the same programs were tracked across multiple years, the data form a panel structure that permits rigorous longitudinal comparison.

The statistical architecture of the study reflected this repeated-measures design. Paired sample t-tests were used to evaluate whether concordance rates changed significantly between consecutive match cycles, effectively asking whether the transition into the signaling era—program signaling was formally adopted by the obstetrics and gynecology match in the 2023 cycle—coincided with measurable shifts in geographic draw. Complementing these pairwise comparisons, repeated measures analysis assessed overall trends across the full six-year window. Finally, the researchers employed cluster analysis, an unsupervised statistical technique that groups observations according to similarity, to determine whether programs naturally fall into distinct subgroups based on their concordance patterns rather than sitting on a continuous spectrum.

The results were largely reassuring for those who worried the reforms might entrench parochialism—or, conversely, disappointing for those who hoped they would unlock geographic mobility. Across all 126 programs, concordance rates remained relatively stable over the six match cycles. Programs that tended to hire their own graduates before the pandemic continued to do so afterward; programs with geographically diverse classes maintained that diversity. The most notable exception was a statistically significant dip and subsequent rebound in concordance between the 2022–2023 and 2023–2024 cycles, with paired comparisons yielding p-values of 0.03 and 0.003 respectively. This transient fluctuation, the authors note, coincides precisely with the rollout of program signaling, suggesting the intervention may have produced a brief period of adjustment before match behavior settled back into its established equilibrium.

The cluster analysis added a layer of nuance that the aggregate trend obscured. Rather than a uniform landscape, the researchers identified two distinct subgroups of programs: one cluster characterized by high concordance, meaning these programs consistently fill a substantial fraction of their classes from their home institution or home state, and another cluster with low concordance, drawing residents from a broader national pool. The existence of these two clusters, stable enough to be detected across multiple cycles, implies that geographic draw is not random but reflects durable institutional characteristics.

Those characteristics were not hard to find. Regional identity emerged as a significant correlate: programs located in the southern United States were more likely to belong to the high institutional concordance group, suggesting a stronger cultural or structural pull to retain local graduates in that region. Program scale and complexity also mattered. Residency programs offering more fellowships—subspecialty training tracks in areas such as maternal-fetal medicine, gynecologic oncology, or minimally invasive gynecologic surgery—tended to have lower state concordance rates. The authors interpret this as evidence that academically larger programs, with national reputations and subspecialty infrastructure, attract applicants from across the country, while smaller community-oriented programs rely more heavily on their regional pipelines.

The implications of these findings extend well beyond obstetrics and gynecology. The residency match has long been criticized for its escalating costs and geographic inequities. Students at medical schools without affiliated residency programs in their specialty of interest historically felt compelled to apply to dozens or even hundreds of programs, incurring thousands of dollars in application fees and travel expenses, simply to remain competitive. Virtual interviews, adopted universally during the pandemic and retained by most specialties afterward, were promoted partly as a cost-saving measure and partly as a geographic equalizer. Program signaling, introduced to counteract the information asymmetry created when applicants scatter their interest across vast numbers of programs, was intended to concentrate genuine enthusiasm and improve match efficiency.

Yet the new study suggests that neither intervention has materially altered the geographic composition of entering residency classes in obstetrics and gynecology. If concordance rates—the direct measure of how much programs recruit locally—remained stable through six cycles that encompassed both the virtual interview transition and the signaling rollout, then the deeper drivers of geographic matching likely lie elsewhere: in applicant preferences for proximity to family and support networks, in regional reputational networks between medical schools and residencies, in couples’ matching considerations, and in the structural gravitational pull that established training pathways exert on each new cohort.

For the field of graduate medical education, the study delivers a methodological contribution as much as a substantive one. By leveraging FREIDA data at the program level rather than relying on single-institution surveys or applicant self-reports, the researchers produced an objective, nationwide, multi-year measurement of geographic concordance. The panel design, paired t-tests, and cluster analysis together demonstrate a replicable framework that other specialties—many of which have adopted signaling in recent years—could readily apply to evaluate whether their own reforms are moving the needle on diversity outcomes.

The study also carries caveats worth noting. The analysis was limited to programs with an affiliated medical school, which excludes freestanding residency programs and may slightly skew the sample toward academic centers. FREIDA data depend on accurate program-level reporting, and the concordance metric captures where matched residents attended medical school but not the full geography of their personal origins. Moreover, the transient dip observed around the signaling transition, while statistically significant, resolved within a single cycle, underscoring how quickly the match system recalibrated. The research received no external funding and was conducted under IRB approval at Stony Brook University Hospital, with all data retrospective, unidentifiable, and publicly available.

What the study ultimately offers is a measured verdict on one of the most closely watched experiments in medical education. The pandemic-era reforms were conceived amid genuine optimism that technology and smarter signaling could flatten the residency landscape. Six match cycles later, the data indicate that the landscape has proven remarkably resilient. Geographic diversity in the obstetrics and gynecology match—measured by the proportion of residents training at their home institutions or within their home states—has persisted essentially unchanged. Whether that stability reflects a system that works well, or simply one that resists change, remains the question that educators, policymakers, and the students navigating each year’s match will now have to confront.