P3.153: Attrition Among Pediatric Residents and Fellows by Race/Ethnicity, Gender, and Citizenship: 2001-2021

Background: Declining interest in pediatrics residency training among medical doctorate students within the US over recent years has stimulated attention to innovative recruitment efforts. Although the number of pediatric residency positions has increased over time, the fill rate during the main residency match markedly decreased over the last two years. Similarly, many pediatric subspecialty fellowships remain unfilled year after year. While recent literature has focused on recruitment efforts and workforce trends, little research exists on attrition, or lack of completion of training of a residency or fellowship program, among pediatric trainees. The objective of this study was to evaluate the associations between pediatric trainee attrition and race/ethnicity, gender, and citizenship.

Methods: This cross-sectional study of pediatric resident and fellow attrition utilized 2001-2021 de-identified data extracted by the Association of American Medical Colleges (AAMC) from the Graduate Medical Education (GME) Track Survey, which receives a high response rate (~95%). Dataset measures provided included identification#, year entered, specialty, race/ethnicity (American Indian/Alaska Native [AI/AN], Asian, Black, Hispanic, Native Hawaiian/Pacific Islander [NH/PI], Other, White), gender (M/F), citizenship (Y/N), attrition (Y/N), and attrition type (dismissed, transferred, withdrawn). An underrepresented in medicine (URiM) category was created based on the AAMC definition, which included identification as AI/AN, Black, Hispanic, and NH/PI. Logistic regression models controlling for year entered, URiM status, gender, and citizenship were fit to assess relationships with attrition outcomes. If the overall test was significant (defined as p< 0.05), pairwise comparisons within independent variables (URiM vs non-URiM, M vs F, etc.) were analyzed to detect significant effects. Odds ratios and 95% confidence intervals were calculated for independent measures. SAS (version 9.4, Cary, NC, USA) was used for all analyses.

Results: 85,958 records were included; 31 duplicate entries and 2 with missing attrition data were excluded, yielding a final sample of 85,927, 60,867 (70.8%) of which were residents. Table 1 highlights sample characteristics. In a model evaluating effects of year, race/ethnicity, gender, and citizenship of residents, all were significantly associated with any attrition (p < 0.0001, p=0.0014, p < 0.0001, p=0.0003, respectively), with URiMs, males, and non-citizens having significantly higher odds of dismissal (Tables 2a-d). In the same model of fellows, year and citizenship were significantly associated with any attrition (p < 0.0001 and p=0.0001), with URiMs and males having significantly higher odds of dismissal and non-citizens higher odds of withdrawal (Tables 2e-g).

Conclusion: This study is one of the first to highlight the impact of resident and fellow attrition on the pediatric workforce. From 2001-2021, year, race/ethnicity, gender, and citizenship were significantly associated with any resident attrition, while year and citizenship were significantly associated with any fellow attrition. Future research should examine the higher attrition rates among certain groups that our study found.