Effect of resident and assessor gender on entrustment-based observational assessment in an internal medicine residency program

Background: Implicit gender bias leads to differences in assessment. Studies examining gender differences in resident milestone assessment data demonstrate variable results. The purpose of this study was to determine if observational entrustment scores differ by resident and assessor gender in a program of assessment based on discrete, observable skills. Methods: We analyzed overall entrustment scores and entrustment scores by Accreditation Council for Graduate Medical Education (ACGME) core competency for 238 residents (49% female) from 396 assessors (38% female) in one internal medicine residency program from July 2012 to June 2019. We conducted analyses at 1-12 months, 1-36 months, 1-6 months, 7-12 months, and 31-36 months. We used linear mixed-effect models to assess the role of resident and assessor gender, with resident-specific and assessor-specific random effect to account for repeated measures. Results: Statistically significant interactions existed between resident and assessor gender for overall entrustment at 1-12 months (p < 0.001), 1-36 months (p< 0.001), 1-6 months (p<0.001), 7-12 months (p=0.04), and 31-36 months (p<0.001). However, group differences were not statistically significant. In several instances an interaction was significant between resident and assessor gender by ACGME core competency, but there were no statistically significant group differences for all competencies at any time point. When applicable, subsequent analysis of main effect of resident or assessor gender independently of one another revealed no statistically significant differences. Conclusions: No significant differences in entrustment scores were found based on resident or assessor gender in our large, robust entrustment-based program of assessment. Determining the reasons for our findings may help identify ways to mitigate gender bias in assessment.