Critical Data Literacy: Addressing race as a variable in a preclinical medical education session: Poster

Hello, my name is Stacy Brody. I am a Reference and Instruction Librarian at the Himmelfarb Health Sciences Library at The George Washington University. My colleague Susan Koenig and I created the poster “Critical Data Literacy: Addressing race as a variable in a preclinical medical education session”. We describe the changes made to a librarian-led informatics class to incorporate discussions of race in clinical research and algorithms. 

In the past year, discussions and critiques of the teaching of race in medicine and the inclusion of race as a variable in biomedical research crescendoed. Amidst continuing acts of racial violence and disparities in health and health care, medical students and practitioners have challenged race correction factors in clinical decision-making tools. These conversations have been carried on in the pages of journals like the New England Journal of Medicine and in an increasing number of medical school class – or Zoom – rooms, as well as in professional conversations among medical librarians. 

As I personally tried to engage in anti-racist action, I was also reviewing the content of librarian-led informatics session for the preclinical medical curriculum. 

I was aware of university guidelines describing the appropriate use of race in case vignettes. When I saw the case vignette for this session on chest pain and the Atherosclerotic Cardiovascular Disease Risk calculator, I noticed that the races of the patient examples were included. I started to adjust the case, removing race, until I noticed the calculator required race as an input. 

The options are white, African American, and other*. 

So, I had to leave the example patient cases as is. I had to keep the designation of race in the session. 

I felt uneasy.

I felt especially uneasy since patients in the case examples were not all described as white or African American. One was Hispanic. Another Middle Eastern. 

Is this patient white? Do they identify as white? Who am I to say? If I was having these questions, would medical students have similar questions? I felt the librarians needed to be prepared to address these.

This led be down a rabbit hole of research into who was included in the cohorts used to derive the algorithm and the cohorts on whom the algorithm was validated. I found the studies of external generalizability demonstrated variation, at times over or underestimating risk, depending on the population. 

I revisited the Office of Management and Budget categories of race and ethnicity and saw how they have been implemented in biomedical research and clinical care. 

I heard more about others raising similar issues with clinical algorithms, particularly related to kidney function. 

The more I read and listened in this space, the more I recognized the need to update the session materials. I added the session objective: Discuss factors to consider in the derivation and application of clinical risk calculators. I added slides describing the general development process for clinical algorithms and prompting students to think critically about the derivation and validation of clinical algorithms. I added content addressing the “who” of the Framingham Study and the later pooled cohorts, from which calculators were derived. I highlighted potential for over- and under-estimation of risk and risk-enhancing factors to encourage more critical application of the tool in the context of a larger conversation with a patient. 

I led a discussion of these changes with the librarians prior to delivering the session. These are difficult topics to address. We agreed to focus the conversation via the lens of critical data literacy – knowing how a dataset is developed can empower you to apply the resultant tool in clinical practice and to improve data collection in future research endeavors. 

The delivery of the session was challenging via WebEx, as this class was delivered 100% virtually. There was not enough time, particularly after navigating breakout sessions and technical issues, to dive into questions for discussion or engage students in ways we would have liked and which we anticipate will be more possible in person. 

After delivering this session once, we recognize a few areas for improvement and a few areas in which more support materials are needed.

We were also able to distill down the key messages that resonated with students:
• First, be aware of the composition of the dataset so that you know of possible limitations to application and areas for improvement
• Second, use the evidence to treat the individual patient

In the future, we hope to continue to encourage critical data literacy through this and other librarian-led sessions. We hope to encourage students to raise critical questions and to raise awareness of the “how” of evidence creation and the “who” of the evidence, in order to apply the evidence to clinical practice.