Abstract
Importance Over-application and interviewing are believed to be widespread in residency recruitment. These may have increased during the 2021 virtual recruitment season. The increase does not correspond to an increase in available residency positions and likely results in more interviews with low probabilities of yielding matches. Prior work demonstrates that such marginal interviews are identifiable ‑ from key explanatory factors like same-state for interviewee and program ‑ in sufficient volume to allow programs to substantially decrease interviews. Objective To evaluate the importance of same-state relationships in primary care and to determine the extent of over-interviewing in the 2021 virtual recruitment season. Design The National Resident Matching Program and Thalamus merged match (outcomes) and interview (explanatory variables) data from primary care specialties (family medicine, internal medicine, pediatrics). Data were analyzed by logistic regression, trained on the 2017-2020 seasons, and projected on the 2021 season for testing. Setting The setting was the 2017-2021 main residency matches. Participants This comprised 4,442 interviewees applying to 167 residency programs in primary care. Intervention This included the transition to virtual recruitment from in-person recruitment in the 2021 residency recruitment season. Measurements A total of 20,415 interviews and 20,791 preferred programs with program and interviewee characteristics and match outcomes were included. Results Same-state geographic relations predicted match probability in primary care residency interviews better than medical school/residency affiliation, with 86.0% of interviewees matching consistently with their preferences for the same state. Same-state was more effective than medical school affiliations with programs in predicting matching. Eliminating interviews with less than a 5% probability of matching (upper 95% prediction limit) removed 31.5% of interviews. Conclusions and relevance The large number of low-match probability interviews demonstrates over-interviewing in primary care. We suggest that programs eliminate interview offers to applications falling below their chosen match probability threshold.