Implementing Core Entrustable Professional Activities in Undergraduate Medical Education: A Psychometric Study

This study examines the feasibility and psychometric results of an assessment of entrustable professional activities (EPAs) as a core component of the clinical program of assessment in undergraduate medical education, assesses the learning curves for each EPA, explores the time to entrustment, and investigates the dependability of the EPA data based on generalizability theory (G theory) analysis.

Third-year medical students from the University of Minnesota Medical School in 7 required clerkships from May 2022 through April 2023 were assessed. Students were required to obtain at least 4 EPA assessments per week on average from clinical faculty, residents supervising the students, or assessment and coaching experts. Student ratings were depicted as curves describing their performance over time; regression models were used to fit the curves.

The complete class of 240 (138 women [58.0%] and 102 men [42.0%]) third-year medical students at the University of Minnesota Medical School (mean [SD] age at matriculation, 24.2 [2.7] years) participated. There were 32,614 EPA-based assessments (mean [SD], 136 [29.6] assessments per student). Reliability analysis using G theory found that an overall score dependability of 0.75 (range, 0-1) was achieved with 4 assessors on 4 occasions. The desired level of entrustment by academic year end was met by all 240 students (100%) for EPAs 1, 6, and 7, 237 (98.8%), 236 (98.3%), and 218 (90.8%) students for EPAs 2, 5, and 9, respectively, 197 students (82.1%) for EPA 3, 178 students (74.2%) for EPA 4, and 145 students (60.4%) for EPA 12. The most rapid growth was for EPA 2 (β 0 = .286), followed by EPA 1 (β 0 = .240), EPA 4 (β 0 = .236), and EPA 10 (β 0 = .230).

The study findings suggest that EPA ratings provide reliable and dependable data to make entrustment decisions about students' performance.