Finding the right words: A skills-based curriculum using verbal procedures to address implicit bias in clinical encounters

The educational intervention is grounded in the patient-informed [ 9 ], evidence-based framework of Implicit Bias Recognition and Management (IBRM) [ 46 , 47 ]. Briefly, the IBRM framework outlines a two-step process in which clinicians (students, trainees, and practicing clinicians) recognize when implicit bias is influencing an encounter and implement skills to manage the effect of that implicit bias to deliver optimal care. IBRM guides interventions that address both pathways by which implicit bias impacts communication in a clinical encounter (Figure) : Pathway 1 (patient lived experience leading to perceived bias in routine procedures) and Pathway 2 (clinician makes an assumption about the patient) [ 9 ]. Although this intervention is set in the context of implicit bias in the United States, implicit bias and discrimination affect patient-provider communication globally [ 48 – 58 ]. The intervention was delivered virtually or in-person, depending on the institution’s usual format or scheduling constraints. All sessions were conducted by the lead author (CMG) except for the in-person session with medical students in which trained faculty co-facilitated small groups following the same facilitator guide. To ensure consistency, CMG trained all facilitators who led the sessions. Both CMG and facilitators had significant prior experience teaching about implicit bias. We outlined the following learning objectives for the intervention: 1) Identify how perceived and actual implicit bias influence the outcomes of clinical encounters; 2) Develop strategies for establishing rapport with and effectively gathering data from patients despite perceived or actual implicit bias; and 3) Practice skills to restore rapport with patients once perceived or actual implicit bias has impacted the encounter. We conducted a preliminary program evaluation testing the efficacy of the intervention on the final learning objective (the skills practice). To ensure the relevance of the active learning exercise to patient care and communication skills, investigators delivered a short didactic (20–25 minutes long on average). This “mini-lecture” focused on the two pathways by which implicit bias can erode patient-clinician communication ( Figure 1 ), patient testimonials on the effect of Pathway 1 and Pathway 2, and research with patients describing the inflection point where clinician action is most effective once implicit bias has impacted an encounter [ 9 ]. That inflection point for clinician action requires communication skills to restore rapport with patients once evidence surfaces that implicit bias has impacted the encounter [ 9 ]. To start developing these communication skills, we had learners develop verbal procedures—personalized, verbatim statements—that can be used to restore rapport with patients after bias has negatively impacted the encounter [ 59 ]. For example, the physician may say, “I’m sorry, I see where you’re coming from, I should have explained why I was asking [ 59 ],” after the physician notices their patient’s verbal/nonverbal communication has changed (the patient may look down, become curt in their answers, lean back, etc.) We chose the term “procedure” to highlight the importance of practicing these verbatim statements to build “muscle memory,” much like learners practice manual procedures (e.g. drawing blood) in simulated settings prior to performing them on patients. To prompt the verbal procedures and ground them in a clinical context, learners were presented with two scenarios. The first scenario demonstrated how implicit bias impacts a clinical encounter through Pathway 1 (patient lived experience). It involved a young Black man who presents after injuring his arm. Due to his lived experience with bias and discrimination, he perceives bias when asked a routine question, “Do you smoke cigarettes?” He perceived bias when asked if he smokes cigarettes because he is seeking medical care for an arm injury and the question seems unrelated to his presenting symptom. He becomes quiet, reticent to answer questions, and visibly annoyed. The second scenario demonstrated how implicit bias impacts an encounter through Pathway 2 (actual clinician implicit bias). This scenario involved a young woman presenting as a new patient visit to establish care in a new city. She is requesting to continue her oral contraceptive, prescribed for endometriosis. The new physician suggests an alternative with better efficacy at preventing pregnancy, but the patient declines stating she only has sex with women. The physician then says, “Well, you never know.” After the physician states this assumption, the patient becomes visibly frustrated. The only modification to the cases was made for the virtual psychiatry grand rounds; the cases were adapted to include a mental health diagnosis to enhance the relevance of the activity for that audience. Participants developed their verbal procedure either through an audience response system (ARS—virtual cohorts) or a role-play (in-person cohorts). For learners participating virtually, after the didactic portion, they were presented with the two scenarios described above and displayed in the Text Box . Each scenario was delivered line by line with pauses in between to foster internal reflection. At the end of each scenario, learners were prompted to type a verbatim statement to restore rapport with the patient in the scenario after the patients’ verbal/nonverbal communication changes to suggest that implicit bias impacted the encounter. Then, through the ARS, participants submitted their verbal procedures. Participants in these virtual cohorts did not receive any feedback on their verbal procedures. In-person learners developed their verbal procedures through paired role-plays using worksheets (see Table 1 for a brief outline; full worksheets in the Appendix) and submitted them online using a QR code. Each worksheet was tailored to the role of the perspective of the “doctor” (Participant A) or “patient” (Participant B). Participants did not know each other’s perspectives going into the role-play. The worksheets begin by explaining the first scenario (described in 2.3.2) from the perspective of the “doctor” or the “patient.” To develop and practice a verbal procedure to restore rapport when implicit bias has impacted a clinical encounter through Pathway 1 ( Figure ), the “doctor’s” worksheet instructs the participant to formulate a verbal procedure and write it down. The “patient’s” worksheet asks the participant to reflect on their feelings (in character) and then instructs them to become quieter, withdrawn, and annoyed after being asked if they smoke cigarettes. The “doctor” and “patient” act out the scenario following the step-by-step instructions. The ”doctor” delivers a verbal procedure and receives feedback from “the patient” on how effective the verbal procedure was at restoring rapport. The “doctor” can then revise the original verbal procedure and deliver the revised verbal procedure in a repeat role-play, receiving feedback on this second attempt from the “patient.” The repeat role-play was the same scenario as the original role-play, thereby serving as a “do-over.” This repeat approach was adapted with permission from the Stanford Faculty Development Center for Medical Teachers [ 60 ]; it has been described by learners as “a gift” in prior IBRM interventions [ 18 ]. Worksheets then guide participants to switch roles, repeat the same series of steps to act out the second scenario (described in 2.3.2) and the “doctor” (Participant B) develops a verbal procedure to restore rapport with the “patient” (Participant A) when implicit bias has impacted the clinical encounter through Pathway 2 ( Figure ). An optional large-group debrief (up to five minutes) providing opportunities for questions, expression of challenging and surprising aspects of the exercise, and facilitators attempted to highlight successful approaches pending time constraints for both the ARS and role-play groups. To promote ongoing development and education for our learners, the submitted verbal procedures were shared back with learners post-session. Given the lack of practical skills-based components in most educational interventions on implicit bias [ 38 ], we sought to elucidate the effect of adding verbal procedures as a practical skills-based element of implicit bias teaching. To achieve this, we analyzed the final verbal procedures generated by the learners as a proxy for the development of new skills to apply in real-world situations when bias impacts a clinical encounter. Analysis of verbal procedures was approved as exempt research by both the Albert Einstein College of Medicine and the New York University Grossman School of Medicine Institutional Review Boards. Two investigators, blinded to whether the verbal procedures were developed through ARS or role-play, independently reviewed the verbal procedures, applied preliminary descriptors to elements within each verbal procedure, and then convened to review verbal procedures in duplicate, agree on codes and their definitions, and assign a point value. The two investigators iteratively developed a rubric to assign a point value to each code that was applied to each element within each verbal procedure; each code was assigned a point value based on being unhelpful (−1) or helpful (+1). This rubric was informed by our earlier work with racially and ethnically minoritized patients and their suggestions for restoring rapport once bias had impacted the encounter [ 9 ]. Patients in that study emphasized the importance of acknowledgment of the impact of implicit bias and the benefits of an apology from the clinician [ 9 ]. The positive or negative valence of the code was informed by those patient perspectives [ 9 ], literature on effective communication in apologizing and restoring rapport with another [ 61 – 64 ], and consensus between investigators when no prior evidence was available. Review of published literature across fields, including social psychology, supports the claim that an effective apology should be grounded in empathy [ 61 , 62 ]. Apologies grounded in empathy have been shown to prevent retaliation from the victim, as it demonstrates the offender seeks to remediate the relationship [ 61 ]. Effective apologies include those that: 1) Acknowledge an error has occurred; 2) Take accountability for the mistake; and 3) Validate the offendee’s experience and emotions [ 61 – 64 ]. Investigators arrived at a score for each verbal procedure by summing the point values of each code applied to each element within the verbal procedure. A verbal procedure was deemed unhelpful overall if its total score was 0 or –1; an overall helpful verbal procedure had a total score of at least 1. There was no minimum or maximum possible score.

From 2022–2024 we delivered the intervention across four settings to N=440 learners either virtually or in person. The virtual cohorts included 166 first-year medical students in Bronx, New York, USA and 153 attendees of a department of psychiatry grand rounds in Durham, North Carolina, USA, and lasted 60 minutes. The in-person cohorts included 103 first-year medical students in New York, New York, USA and 18 internal medicine residents in Brooklyn, New York, USA, and lasted 90 minutes. Attendance was expected for the three New York-based cohorts as the intervention was integrated into regularly scheduled curriculum for the learners; attendance was optional for the North Carolina-based cohort. Virtual delivery for two of the cohorts resulted from scheduling constraints precluding in-person facilitation of the intervention. For the cohorts based in New York, at least one of the investigators had direct knowledge of participants prior exposure to implicit bias content (i.e., several of the authors facilitated an introductory implicit bias session for these cohorts). For the cohort based in North Carolina the Department’s Vice Chair for Education confirmed that attendees would have participated in at least one introductory implicit bias educational session prior to delivery of the intervention we describe. Therefore, all participants had prior exposure to foundational content in implicit bias as a concept and its contributions to health disparities, including defining and increasing awareness about implicit bias and its impact on patients but without practical solutions to mitigate its impact. Additionally, role-play took learners 30 minutes to complete both scenarios, whereas ARS was shorter-learners completed both scenarios in under 10 minutes. The role-play cohorts did not have time for large-group debrief but some individuals stayed afterward to speak with the in-person facilitators. Given the differences between ARS without feedback and role-play with feedback as active learning educational strategies, we report the scores of the two virtual cohorts who used ARS separately from the two in-person cohorts who conducted the role-plays. We received 47 verbal procedures out of a total of 319 participants through ARS and 91 out of 121 through role-play, yielding a response rate of 15% and 75% respectively. Scores for submitted verbal procedures ranged from −1 to 6. See Table 2 for scoring criteria and exemplary helpful and unhelpful verbal procedures. Both the ARS and role-play cohorts submitted a higher percentage of overall helpful verbal procedures than unhelpful ones. For both cohorts combined, 91.3% (n=126) of verbal procedures were helpful, and only 8.6% (n=12) of verbal procedures were unhelpful. Separating scores by cohorts, for ARS, 83% (n=39) of verbal procedures were helpful, and 17% (n=8) of verbal procedures were unhelpful. For role-play, 95.6% (n=87) of verbal procedures were helpful and 4.4% (n=4) of verbal procedures were unhelpful. Verbal procedures developed through the role-plays included more helpful elements and therefore yielded higher scores than through ARS (patient lived experience, Pathway 1, Scenario 1: 2.21 vs 1.29, p=0.003; and actual clinician implicit bias, Pathway 2, Scenario 2: 2.46 vs 1.87 p=0.009). An exemplar high-scoring (score=4) verbal procedure was, “I’m sorry is everything ok? I noticed something changed about the way you’re speaking, and I realize the question I asked may have upset you. These are questions we ask every patient...” This helpful verbal procedure was submitted via role-play. A low scoring (score=0) verbal procedure that was submitted via ARS was “I notice that I might have upset you. Can you tell me about that?”.

In the future we expect to improve this intervention by analyzing verbal procedures before and after partner feedback to assess the effect of that feedback and repeated practice. We will also capitalize on lessons learned and ensure that all future sessions allot for at least thirty minutes to complete both role-plays (15 minutes per scenario) and time for large-group debrief at the end. Finally, we will collaborate with colleagues to advance the intervention from role-plays to simulations with standardized patients [ 8 , 68 ]. Performance in simulations predicts future clinical practice behaviors and is an important step in testing interventions seeking to improve patient care, while maintaining patient safety [ 69 – 72 ]. We anticipate that paired, guided role-plays are an efficient approach to foster skill development and practice in a myriad of other constructs within clinical encounters and the clinical learning environment. Although not as effective as role-play, ARS allows for development and articulation of verbal procedures and has the added benefit of reaching larger audiences virtually when pairing may not be feasible. We expect to test strategies to enhance engagement of audience members in developing verbal procedures through the ARS modality, especially in light of continued delivery of instruction through a virtual format [ 73 ]. Finally, we hope to be able to observe these behaviors in learners in the future capitalizing on the methodology of unannounced standardized patients to capture authentic clinician behaviors in actual clinical settings [ 74 ]. Our study has several limitations. Our sample size is too small to stratify findings by participant stage of training (e.g., student, trainee, faculty) or by geographic region. Also, the low response rate in the ARS group may indicate difficulty engaging some learners through virtual delivery of the intervention. We were unable to investigate differences between responders and non-responders in this group and if any differences might affect the outcomes of our intervention. Additionally, we have no baseline assessment of our learners’ ability to develop verbal procedures. We were unable to determine if learners would be able to identify when implicit bias has impacted a clinical encounter based on changes in patient verbal/nonverbal communication behaviors. Our first learning objective was to identify how perceived and actual implicit bias influence the outcomes of clinical encounters. In other words, for them to identify how implicit bias impacts an encounter via Pathways 1 and 2 (Figure) . Functionally, within a clinical encounter, the clinician must first be able to identify that bias has, in fact, impacted the clinical encounter before delivering a verbal procedure to restore rapport with a patient. Educationally, however, that is a separate skill that requires dedicated instruction if a clinician is not able to intuit a change in their patient’s demeanor. Teaching learners to identify when implicit bias has impacted their clinical encounter first, without teaching them skills to restore rapport with patients once they realize that implicit bias has impacted the encounter risks unintended consequences. Unintended consequences of implicit bias have already been reported, such as avoidant behaviors when someone becomes aware of their implicit bias, but does not have the skills to address its impact (e.g. avoiding Black people in an effort to do no harm) [ 75 ] and discomfort when students learn about implicit bias in the classroom, and then see it playing out in the clinical learning environment but do not know how to advocate for their patients or avoid providing biased care themselves [ 39 , 40 ]. Therefore, we chose to first provide instruction on restoring rapport, with plans to develop instruction on identifying when implicit bias impacted the clinical encounter. Finally, we do not have demographic data on participant race or sexual orientation. Patients from minoritized racial or sexual orientation backgrounds may have had different perspectives on if the verbal procedures were actually successful in restoring rapport after implicit bias has impacted an encounter, highlighting yet another opportunity for future research. We also do not have any data on participant satisfaction with the session. Our study findings may not be generalizable outside of the United States context. Nonetheless, we expect the core educational strategies of our intervention could inform development and study of future interventions in other global contexts. Data demonstrate that implicit bias—either perceived on routine statements an procedures due to patient lived experience or actual clinician implicit bias leading to assumptions about the patient—impacts patient-clinician communication. While many clinicians are aware of implicit bias and its impact, they struggle with addressing implicit bias when it has impacted their clinical encounters. Verbal procedures are a practical and effective way to develop and practice skills to restore rapport with patients in an effort to repair the patient-clinician relationship in situations where implicit bias has impacted the clinical encounter. Education in verbal procedures could be relevant to clinicians across the spectrum of training and practice and across disciplines.

Evaluation of our single-session intervention suggested achievement of the third learning objective. Some participants were able to generate verbal procedures with helpful elements using ARS or role-play. Recognizing the difference between ARS and role-play, we compared the verbal procedures developed through each of these active learning strategies. Verbal procedures developed through role-play had significantly more helpful elements than those developed using ARS. This is likely due to two factors: 1) the opportunity to revise the verbal procedure based on feedback from the role-play partner playing the “patient;” and 2) the in-person setting with the facilitator present likely led to increased engagement than the virtual learners watching the facilitator on a device. In the absence of an ability to conduct role-plays, however, ARS still offered a preliminary approach to generating verbal procedures to restore rapport with patients once implicit bias has impacted the encounter. As we collectively seek to eliminate healthcare disparities, clinician communication behaviors are an important intervention point [ 7 ]. Our intervention focuses on clinician communication behaviors that have been deemed relevant by patients to addressing the impact of implicit bias on clinical encounters [ 9 ]. We address two persistent challenges to developing interventions that address the impact of implicit bias by: 1) intentionally incorporating the experiences of minoritized patients; and 2) grounding our work in the extant literature on implicit bias [ 7 ]. By focusing specifically on the clinical encounter and placing participants in both the roles of the “patient” and “doctor” in role-plays to restore rapport with patients once bias has been perceived, our work further expands the approaches and outcomes of prior interventions in addressing the impact of implicit bias in the clinical learning environment. Prior interventions have incorporated role-plays focused on implicit bias and microaggressions in the learning environment for health professions students [ 18 , 19 , 24 , 65 , 66 ], and for addressing bias from patients toward members of the clinical care team [ 67 ]. By tasking participants to write verbatim verbal procedures, our current work also expands on our prior work using ARS to identify strategies in IBRM [ 20 ].

Effective communication between clinicians and patients is essential to excellent patient outcomes Hashim [ 1 ]. Implicit bias can negatively impact this communication when clinicians are communicating with racially minoritized and other marginalized patient populations [ 2 – 8 ]. Patients do not expect bias-free encounters, but they do expect when implicit bias has impacted rapport within the encounter, for this impact to be addressed by the clinician [ 9 ]. Communication skills to restore rapport with patients once bias has impacted an encounter are therefore a promising target for educators. Implicit bias is defined as unintentional and unconscious assumptions that individuals make about one another [ 10 ]. Implicit bias can negatively impact communication within clinical encounters. From the patient perspective this has resulted in less perceived patient centeredness [ 2 – 4 ] and difficulty remembering conversations following an encounter [ 4 ]. In our prior in-depth qualitative investigations with Black and Hispanic patients we found implicit bias can erode communication within clinical encounters through two pathways [ 9 ]. Pathway 1 stems from patients’ experiences with bias and discrimination in society as well as the healthcare system (patient lived experience). Patients may perceive bias via statements and/or procedures considered routine by clinicians. Although the clinician in this pathway did not do anything “wrong,” the effect of this perceived bias can be profound for patients [ 9 ]. Pathway 2 occurs through clinician assumptions (actual clinician implicit bias). These are often statements or actions that result as a coincidence of the clinician’s lived experience in a system that favors some and marginalizes others [ 9 , 11 ]. Regardless of the pathway by which implicit bias impacts an encounter, clinicians need to act when evidence that implicit bias has impacted the encounter surfaces; this evidence can often be recognized by changes in patient’s verbal/nonverbal communication [ 9 ]. When evidence that implicit bias has impacted the encounter surfaces, that is the inflection point where clinician action, or inaction, can affect the outcome of the clinical encounter. At that inflection point, evidence demonstrates that ignoring that impact leads to mistrust, avoidance of future healthcare, and other negative consequences [ 9 ]. However, verbal acknowledgement of that impact could lead to positive outcomes similar to if implicit bias had never impacted the encounter in the first place [ 9 ]. This last point is critical for learners to embrace. The expectation from patients is not that clinicians will have completely bias-free encounters, but rather that if evidence of the impact of implicit bias on the encounter surfaces, that clinicians can implement skills to address the impact of implicit bias and restore rapport with patients to optimize the outcome of that clinical encounter [ 9 ]. The importance of addressing the impact of implicit bias on clinical encounters is also supported by educators. Educators across four specialties agree implicit bias is an important knowledge topic to address [ 12 ]. Health professions educators deliver curricula on addressing implicit bias across the spectrum of clinical training and practice [ 13 – 37 ]. These curricula focus on awareness, acceptance, reflection, and some include strategy identification. Almost universally, however, these curricula fall short of skill development and practice in addressing implicit bias [ 38 ]. In our prior work, after two sessions focused on the impact of implicit bias on the persistent problem of health disparities, most students were aware of implicit bias and accepting of the potential for their biases to impact their future clinical care [ 39 ]. Unfortunately, these seven cohorts of students (N=1162) were unable to identify strategies to mitigate the impact of their implicit biases on their future clinical care [ 39 ]. Providing instruction about implicit bias without providing opportunities for skill development and practice in addressing implicit bias risks two unintended outcomes: 1) increased internal discomfort in learners and clinicians aware and accepting of their implicit biases without any skills to address their impact in their clinical encounters [ 39 , 40 ]; and 2) missed opportunities to improve the downstream effects of poor clinician communication with patients from populations experiencing healthcare disparities [ 7 , 41 – 45 ]. To address the gap in skills-based implicit bias curricula, the objective of our study is to describe an educational intervention targeting communication skills to restore rapport with patients once implicit bias has impacted a clinical encounter and provide preliminary efficacy data.