Are Final Residency Milestones Correlated With Early Fellowship Performance in Pediatrics?

Abstract

Purpose Milestones have been used to assess trainees across graduate medical education programs and reflect a developmental continuum from novice to expert. This study examined whether residency milestones are correlated with initial fellowship milestone performance in pediatrics. Method This retrospective cohort study used descriptive statistics to assess milestone scores from pediatric fellows who began fellowship training between July 2017 and July 2020. Milestone scores were obtained at the end of residency (R), middle of the first fellowship year (F1), and end of the first fellowship year (F2). Results Data represent 3,592 unique trainees. High composite R scores, much lower F1 scores, and slightly higher F2 scores were found over time for all pediatric subspecialities. R scores were positively correlated with F1 scores (Spearman ρ = 0.12, P < .001) and F2 scores (Spearman ρ = 0.15, P < .001). Although scores are negligibly different when trainees graduate from residency, there were differences in F1 and F2 scores among fellows in different specialties. Those who trained at the same institution for residency and fellowship had higher composite milestone F1 and F2 scores compared with those who trained at different institutions (P < .001). The strongest associations were between R and F2 scores for the professionalism and communication milestones, although associations were still relatively weak overall (rs = 0.13–0.20). Conclusions This study found high R scores and low F1 and F2 scores across all shared milestones, with weak association of scores within competencies, indicating that milestones are context dependent. Although professionalism and communication milestones had a higher correlation compared with the other competencies, the association was still weak. Residency milestones may be useful for individualized education in early fellowship, but fellowship programs should be cautious about overreliance on R scores due to the weak correlation with F1 and F2 scores.