Purpose
The main purpose of the present study was to analyze student growth curves of entrustable professional activities (EPAs) for an Emergency Medicine (EM) clerkship. While empirical work on EPA data has been done for several undergraduate medical education core clerkships, little such work has been done for emergency medicine.

Method
A total of 481 (271 women - 56%; 210 men - 44%) third or fourth-year medical students at the University of Minnesota Medical School, with mean age = 24.3 (SD=2.6) at matriculation, participated. Students were enrolled in a required four-week EM clerkship during 3 academic years, 2021-22, 2022-23, and 2023-24 across 13 core EPAs. Regression analyses were employed to examine if the growth curve of each of the EPAs follow a curvilinear structure as hypothesized in learning theory. To assess reliability or dependability of the EPA assessments we employed a two-facet generalizability analysis (Ep2 coefficients).

Results
There were 7,707 EPA-based assessments (mean = 16 [SD=5.0] assessments per student), provided by 793 assessors; 494 (62.3%) residents, and 299 (37.7%) faculty members. The mean number of ratings per assessor was 9.7 (SD=12.93; range: 1 to 129). The growth curves for most EPAs ratings followed the predicted negative exponential curve (EPAs 1,2,3,5,6,8,10,12) but not for others (4,7,9,11,13). The slope of the growth curves showed variation by EPA. Ep2 coefficients 0.70 were achieved with 6 raters assessing on 6 assessment occasions.

Conclusions
The results of our analyses are consistent with the negative exponential learning theory. Acceptable Ep2 coefficient with 6 assessors. Our overall results, indicate that the EPA framework is a practical approach to assessing real-world competency within the Emergency Medicine clerkship.