Observations of Score Changes Between USMLE Step 1 and Step 2 Among Students of Different Demographic Groups in a...

The transition of the United States Medical Licensing Examination Step 1 to a pass/fail scoring system is reshaping its role in medical students' residency placements. This compels institutions to rethink Step 2 preparation strategies, raising concerns about a clerkship's impact on various student groups. Traditionally, medical schools followed the traditional block rotation model for clerkships, which limits longitudinal learning, and many schools are switching to longitudinal integrated clerkships and longitudinal interleaved clerkships (LInCs). The growth in longitudinal popularity sparks concern for the success of diverse medical student groups as there is minimal research regarding LInC students' USMLE performance. Our study aims to identify which student groups at Kirk Kerkorian School of Medicine at the University of Nevada, Las Vegas (KSOM) saw the greatest improvement in their USMLE Step scores after completing the LInC clerkship model.

Utilizing institutional data from KSOM, 145 students from 3 KSOM cohorts' Step 1 and Step 2 3-digit scores and their self-identified demographic information prior to the change in Step 1 grading were categorized by admissions and initial performance factors. Binary groups were created for each variable. Descriptive statistics and t-tests (including Levene's test) gauged score change significance (P < .05) within these groups. Changes were assessed by subtracting Step 1 from Step 2 scores, identifying groups showing significant score improvements after completing the LInC clerkship.

Analysis revealed significant score improvements between Step 1 and Step 2 for the following groups: females, students with low socioeconomic status, and students who originally received lower Step 1 scores.

This study underscores the significance of gender, socioeconomic status, and prior exam performance in clerkship model development given the changes to Step 1 scoring. Further research should discern whether the observed score changes are attributed to the LInC model or its associated testing model.