BACKGROUND AND OBJECTIVES: Failing Step 1 of the US Medical Licensing Examination (USMLE) or a delay in taking the exam can negatively affect a medical student's ability to match into a residency program. Unfortunately, identifying students at risk for failing Step 1 is challenging, but it is necessary to provide proactive educational support. The purpose of this study was to develop a strategy to identify students at risk for failing Step 1. METHODS: Using a retrospective study design, 256 students from the class of 2008 were eligible for the study. Independent variables included Medical College Admission Test (MCAT) scores and cumulative grades from years 1--2 of medical school. The dependent variable was their score on the USMLE Step 1. Variables with a significant univariate relationship were loaded into a series of binary logistic regression models. A receiver operating characteristic (ROC) curve examined the significant variables. RESULTS: Both year-2 standard score and the MCAT biological sciences score were significant as predictors of failure. The ROC curve provided a range of values for establishing a cutoff value for each significant variable. CONCLUSION: Using internal and external predictors, it is possible to identify students at risk for failing Step 1 of the USMLE.
BACKGROUND AND OBJECTIVES: Failing Step 1 of the US Medical Licensing Examination (USMLE) or a delay in taking the exam can negatively affect a medical student's ability to match into a residency program. Unfortunately, identifying students at risk for failing Step 1 is challenging, but it is necessary to provide proactive educational support. The purpose of this study was to develop a strategy to identify students at risk for failing Step 1. METHODS: Using a retrospective study design, 256 students from the class of 2008 were eligible for the study. Independent variables included Medical College Admission Test (MCAT) scores and cumulative grades from years 1--2 of medical school. The dependent variable was their score on the USMLE Step 1. Variables with a significant univariate relationship were loaded into a series of binary logistic regression models. A receiver operating characteristic (ROC) curve examined the significant variables. RESULTS: Both year-2 standard score and the MCAT biological sciences score were significant as predictors of failure. The ROC curve provided a range of values for establishing a cutoff value for each significant variable. CONCLUSION: Using internal and external predictors, it is possible to identify students at risk for failing Step 1 of the USMLE.