Capturing growth curves of medical students' clinical skills performance.

Introduction: A benefit of a milestone or Entrustable Professional Activity (EPA) assessment framework is the ability to capture longitudinal performance with growth curves using multi-level modelling (MLM). Growth curves can inform curriculum design and individualised learning. Residency programmes have found growth curves to vary by resident and by milestone. Only one study has analysed medical students' growth curves for EPAs. Analysis of EPA growth curves is critical because no change in performance raises concerns for EPAs as an assessment framework.

Methods: Spencer Fox Eccles School of Medicine-University of Utah students' workplace-based assessment ratings for 7 EPAs were captured at 3 time-points in years 3-4 of AY2017-2018 to AY2020-2021. MLM was used to capture EPA growth curves and determine if variation in growth curves was explained by internal medicine (IM) clerkship order.

Findings: A curvilinear slope significantly captured 256 students' average ratings overtime for EPA1a-history-taking, EPA2-clinical reasoning, EPA3-diagnostics, EPA5-documentation and EPA6-presentation, and a linear slope significantly captured EPA9-teamwork ratings, p ≤ 0.001. Growth curves were steepest for EPA2-clinical reasoning and EPA3-diagnostics. Growth curves varied by students, p < 0.05 for all EPA ratings, but IM clerkship rotation order did not significantly explain the variance, p > 0.05.

Discussion: The increase in ratings from Year 3 to Year 4 provides validity evidence for use of EPAs in an assessment framework. Students may benefit from more curriculum/skills practice for EPA2-clinical reasoning and EPA3-diagnostics prior to year 3. Variation in student's growth curves is important for coaching and skill development; a one size fits all approach may not suffice.