United States Medical Licensing Examination Step 1 and 2 Scores Predict Neuroradiology Fellowship Success

Purpose: Many neuroradiology programs use United States Medical Licensing Examination (USMLE) scores to assess fellowship candidates. The authors hypothesized that because they are taken several years before fellowship, USMLE scores would correlate poorly with success in fellowship training as measured by faculty evaluations. Methods: USMLE scores from 10 years of neuroradiology fellows (n = 73) were compared with their cumulative mean E*Value scores from their fellowship years and their best-to-worst rankings within their fellowship years. If available, subspecialty certification scores were also factored as an outcome. Linear correlation and regression analyses were performed adjusting for gender, medical school site, and practice setting after fellowship. Results: USMLE Step 1, 2, and 3 scores were available for 69, 64, and 56 fellows, respectively. Fellowship E*Value scores and rankings showed statistically significant (P < .05) correlations with all USMLE scores, but most strongly for Step 1 scores and E*Value grade (r = 0.443) and rank (r = 0.370). The mean USMLE Step 1 score of the top-ranked fellows (234.5) was significantly higher than that of the bottom-ranked fellows (217.7). The correlations of E*Value score and rank with USMLE Step 1 and 2 scores remained after adjusting for gender and American versus foreign medical school, but the medical school site attended also was an independent predictor of fellowship evaluations. Subspecialty certification scores did not show correlations but were underpowered (n = 28). Conclusions: USMLE Step 1 and 2 scores correlated significantly with success in neuroradiology fellowship, measured by faculty assessments of the six core competencies. Using the scores as a means of assessing candidates for positions is justified.