Back to 2017 Program
Who Makes it to the End? A Novel Predictive Model for Identifying Surgical Residents at Risk of Dropping Out
, Jonathan Abelson1
, Jialin Mao1
, Frank Lewis2
, Fabrizio Michelassi1
, Richard H Bell3
, Art Sedrakyan1
, Julie Sosa4
1New York Presbyterian Hospital - Weill Cornell Medicine, New York, NY;2American Board of Surgery, Philadelphia, PA;3Lewis Katz School of Medicine, Temple University, Philadelphia, PA;4Duke University Medical Center, Durham, NC
OBJECTIVE(S): Attrition in graduate surgical education is 15-35% despite ACGME work-hour reforms. No prospective nationwide study has evaluated factors contributing to trainee loss.
METHODS: This is a nationwide 8-year prospective cohort study of general surgery interns from the Class of 2007. Initial survey results were linked with ABS data (ABSITE dates/scores, residency completion, board status, and program characteristics). Non-parametric classification and regression tree (CART) analysis identified risk factors at the resident level for training non-completion using successive binary divergences of covariates.
RESULTS: There were 1048 interns in 2007. Matched data were available for 80%, representing 83% of residencies. 788(94%) residents had ≥1 ABSITE, and 672(80%) completed training. Gender was the most important predictor of attrition; drop-out for men was 17% vs. 24% for women. For men, the next most important predictor was program size; larger programs had higher drop-out (23% vs. 17% smaller programs). Lowest drop-out was among non-Hispanic married white men at smaller non-academic programs outside the northeast (<6%). Among women, the most important factor was race, with 30% of non-white women leaving vs. 20% for whites. For non-white women, attrition was highest at academic programs (35% vs. 30% non-academic). White women at large academic programs experienced higher drop-out (25% vs 11% smaller programs). Lowest drop-out was among white women at small community programs with a relative in medicine (5%).
CONCLUSIONS: We evaluated attrition in the first longitudinal national cohort study using an individualized predictive model that allows identification of residents at risk and creates a framework for interventions.
Back to 2017 Program