A Novel Communication Rating Scale to Mitigate the Effect of Implicit Bias

Implicit bias recognition and management (IBRM) training is a promising strategy for improving clinician communication but is underevaluated.

To describe the development and psychometric properties of the Respect, Empathize, Listen, Ask, Talk, and Engage (RELATE) rating scale, based on a theoretical framework of communication skills, to mitigate negative influences of implicit bias by using strategies for interrupting unconscious decision-making in patient-centered communication.

In this cross-sectional study, between September 2019 and April 2022, standardized patients used the RELATE rating scale to evaluate communication skills of participating clinical trainees at an academic medical center who completed an IBRM training and engaged in simulated clinical encounters with standardized patients representing those with low socioeconomic status and African American, Latino or Hispanic, and African immigrant patients. Statistical analyses were ongoing from June 2021 to January 2024.

Gwet agreement coefficient (AC) was calculated using data from the 2019 to 2020 cohort year to assess standardized patient interrater reliability (IRR). To assess construct validity and internal consistency reliability, factor analysis techniques were used, and Cronbach α was calculated using data from 3 cohorts (2019-2020, 2020-2021, and 2021-2022).

Twenty-seven standardized patients generated 226 independent RELATE ratings for 123 consenting clinical trainees (14 family medicine residents [11.4%], 48 internal medicine residents [39.0%], and 61 doctor of nursing practice students [49.6%]; mean [SD] age, 30.4 [4.1] years; 82 [66.7%] female), who each had 2 RELATE scores. Of the 3-level item responses of whether a behavior was observed ("present," "partial," or "absent"), for all 19 items, at least 1 trainee (0.4%) was noted who did not complete the evaluated behavior, and all items had some responses of "partial" (range, 13 [5.8%] to 70 [31.0%]). Twelve of the 19 RELATE items (63.2%) showed either substantial or almost perfect IRR (Gwet AC, >0.60), and 6 items (31.6%) had moderate IRR (Gwet AC, 0.41-0.60). Factor analyses resulted in a final 4-factor solution with excellent model fit indices and strong factor loadings: respect (4 items), empathy (5 items), listening and talking (6 items), and engaging in partnership with patients (4 items). Each factor showed sound internal consistency (Cronbach α range, 0.64-0.77).

This cross-sectional study of the RELATE rating scale demonstrated high construct validity and good internal consistency in simulated clinician-patient encounters. The findings suggest that the RELATE rating scale is an efficient, theoretically consistent tool to evaluate implicit bias-management communication behavior among clinicians.