Voluntary resident attrition remains problematic despite recent changes in postgraduate general surgery training, including reduction of work hours. We conducted a prospective study of all postgraduate year (PGY)-1 and -2 trainees on the 2008 American Board of Surgery resident roster (ABS-RR) who completed the National Study of Expectations and Attitudes of Residents in Surgery (NEARS) survey after the American Board of Surgery In-Training Examination (ABSITE) in 2008 or 2009. Among 2,222 PGY-1 and -2 residents on the 2008 ABS-RR, 2,033 completed the NEARS survey in 2008 or 2009 (91.5%). The only demographic or programmatic variables associated with voluntary attrition on univariate analysis were PGY-1 status (9.4% risk vs 4.5% risk for PGY-2, p < 0.001) and program location (p = 0.03). Response differences (p < 0.01) were noted in 23 survey items. In multivariate modeling, PGY-2 status was protective against voluntary attrition (p < 0.001, hazard ratio [HR] 0.41), while programs located outside of the South (Northeast: p = 0.006, HR 2.39; Midwest: p = 0.01, HR 2.37; West: p = 0.10, HR 1.76) were associated with higher attrition. The attrition group more frequently reported that they had considered leaving training (p < 0.001, HR 2.59), that the personal cost of training was too great (p < 0.001, HR 2.89), that they were dissatisfied with their operative experience (p = 0.002, HR 1.89), and that they were not committed to completing their training (p < 0.001, HR 3.96). Using the estimated regression coefficient for each variable in the multivariate models, we calculated a risk score for individual residents; these scores were used to construct covariate-adjusted survivorship functions. Resident attitudes, PGY-1 status, and program location are most frequently associated with voluntary attrition. Our risk score calculation represents a novel potential tool for programs to quantify deficiencies in the training experience of residents, and develop targeted strategies to limit disaffection and improve resident retention.
Read full abstract