Preclinical predictors of surgery NBME exam performance

BACKGROUND: The passing score for the National Board of Medical Examiner's Surgery Subject Examination (NBME-SS) was raised at an urban academic medical school, resulting in a doubling of the number of students failing their surgical clerkship. We hypothesized that at-risk students could be accurately identified and offered an intense tutorial program to enhance success on their exam. METHODS: Preclinical academic records for the graduating class of 2006 were obtained and included gender, ethnicity, preclinical grades, and preclinical NBME's exams, MCAT, and USMLE scores. Simple and multiple linear regression analyses were used to identify correlates with the surgery shelf exam scores and data from the class of 2007 were used for validation. RESULTS: Data were collected on 194 students. On univariate analysis, MCAT, USMLE, and all individual preclinical NBME scores were positively correlated with the NBME-SS (P<0.05). Variables that remained predictive using multiple linear regression were scores on the pathology and physical diagnosis exams (model r(2)=0.48). The single strongest predictor of failure was performance on the pathology shelf exam (ROC area was 0.85 with P<0.0001), which also held true for the class of 2007. CONCLUSIONS: There is a strong correlation between poor performance on the third-year NBME-SS and the second-year pathology NBME scores. The reason for this correlation is unclear but may be related to the integrative thinking and clinical application skills requisite for surgical decision-making first required by preclinical students during their pathology course. Whether efforts to assist medical students with this skill set will be effective warrants further study.