Moderator: Amanda Chase

PRESENTATION 1 - Biases Revealed! Lived Experience and Responses on the Implicit Association Test (IAT)    
Lynn Foster-Johnson    
Geisel School of Medicine at Dartmouth

PURPOSE Although female and minoritized physicians are often recipients of prejudice and microaggressions, little is known about the implicit and explicit biases and attitudes held by medical doctors identifying with one or more marginalizing characteristics. Understanding the attitudes and perspectives at the intersection of multiple marginalizing identities is crucial for effective diversity training.

METHODS We analyzed data from physicians completing the Implicit Association Test (IAT) between 2017 and 2020 through Project Implicit Bias (https://implicit.harvard.edu/implicit/) to investigate the differences in biases and attitudes by the number of marginalizing characteristics. Five publicly available datasets contained implicit and explicit bias measures examining attitudes towards race/ethnicity, skin tone, weight, disabilities, and transgender. We calculated a multiple marginalizing intersectionality score by summing affirmative responses to self-identifying as female, a minority race/ethnicity, or gender fluid/non-binary. We examined IAT responses by intersectionality levels (e.g., none, one, two or more) and associations between intersectionality, bias, and other attitudinal beliefs (e.g., political, poverty).

RESULTS Responses from 33,055 physician participants ranged across the five studies:  Transgender (n=610), Disability (n= 1,632), Skin Tone (n=4,259), Weight (n=5,327) and Race/Ethnicity (n=21,225). Overall, 29% reported as white males with no marginalizing characteristics, 50% reported one marginalizing characteristic, and 21% possessed two or more. Across all studies, physicians who self-described with several marginalizing characteristics held fewer biases against marginalized groups. In contrast, less-marginalized physicians held stronger biases for the dominant group (e.g., Whites, Light Skin, Thin or Abled-persons, Cisgender). Possessing several marginalizing characteristics was significantly associated with more liberal political views and greater preference for the study's target marginalized groups (e.g., Black, Dark Skin, Disabled, Transgender, Overweight).

CONCLUSION Physicians identifying as multiply marginalized have fewer biases against stigmatized groups than their less-marginalized counterparts, perhaps due to lived experiences as targets of overt discrimination or recipients of mundane humiliation and embarrassment in the clinical environment.

PRESENTATION 2 - An Innovative Pilot Program for Physician Assistant Assessment of Core Medical Knowledge and Continuous Learning    
Sheila Mauldin    
National Commission on Certification of Physician Assistants

PURPOSE The National Commission on Certification of Physician Assistants (NCCPA) certifies that physician assistants (PAs) meet clinical knowledge and reasoning standards. Utilizing adult learning theory, NCCPA developed an innovative longitudinal pilot program to assess core medical knowledge while fostering continuous learning and retention. We describe PA perspectives regarding the pilot and its impact on ongoing learning.

METHODS The web-based pilot program was implemented from 01/2019 to 12/2020 with over 18,000 PAs. We evaluated performance and surveyed participants quarterly, assessing their experiences. Key pilot components included spaced testing (25 quarterly questions), individualized feedback, and resources based on performance. PAs were asked about confidence level and question relevance to their practice enabling identification of content-area knowledge gaps and learning opportunities by relevance to practice. PAs immediately knew whether each answer was correct/incorrect, could view their overall percent correct, and compare performance to others. After each question, PAs were given rationales/references and had access to all feedback/resources for two years to enhance learning.

RESULTS Over 98% of PAs participating at the start remained in the pilot. PAs reported learning more from the pilot than other non-longitudinal NCCPA exams. Most (91%) said the rationales were useful; 93% read them for missed questions, and 71% indicated they would return to review them again. PAs also reported needing less preparation, decreased test anxiety, and less disruption to daily life while being more confident in performance, and that the process helped to refresh medical knowledge. However, most completed all 25 quarterly questions in less than two weeks not using available time to space-out questions. PAs demonstrated high performance on the pilot, and no differences were detected by age and specialty. All subgroups showed growth in performance.

CONCLUSIONS PAs view the pilot as fostering continuous learning. More research is needed on why PAs do not space questions throughout the entire allotted period.

PRESENTATION 3 - The Differential Diagnosis as a Snapshot of Clinical Reasoning    
Marconi Monteiro    
The University of Texas Medical Branch -Galveston

PURPOSE This study evaluated potential utilization of clinical skills assessment (CSA) data to obtain insight into students' clinical reasoning.

METHODS Following IRB approval, we accessed materials from our 2019 senior medical student CSA. Exam logistics: 15-minute standardized patient encounter followed by 10-minutes to complete a patient note (interview, physical exam, differential diagnosis (DDx), justification, and management). This work focused on the DDx (up to 3 listed) and justification; patient presentations studied:  back pain and cough (2 versions each). Scenarios targeted consideration of potential diagnoses. Two investigators reviewed each note; diagnoses scored as high priority (HP), feasible (F), not feasible (NF); justification as full, partial or absent. We analyzed frequency of diagnoses and quality of justification by case.

RESULTS Most students (n=212) listed three diagnoses. Students listing first diagnosis deemed high priority were: Back pain#1 (N=110): 89.1%; Back pain#2 (n=104) 73.1%; Cough#1 (n=106) 42.5%; Cough#2 (n=100) 76%. Students listing no high priority diagnosis ranged from 1% to 28.3%. While students offered many feasible diagnoses, some students also included one or more diagnoses not appropriate for the patient presentations. Diagnostic justifications varied widely, sometimes including data not documented in the interview or physical exam.

CONCLUSION Students generally identified one high priority diagnosis, with variation by case. Despite listing many feasible diagnoses, students did not consistently highlight the most important consideration for a given context.  Justification of diagnoses revealed limitations in integration and synthesis of patient data. Although accuracy of DDx and justification represent a window into student clinical reasoning, data from this work did not allow identification of student specific issues.