VP&S Guidelines for Promoting an Anti-Bias and Inclusive Curriculum
Excellence in patient care is a value of Columbia University Vagelos College of Physicians and Surgeons (VP&S). These guidelines for educators are intended to combat bias which has historically existed and continues to exist in medical education, and which can hinder our ability to provide excellent medical care. Through these guidelines, co-created by VP&S students, faculty, and administrators, we aim to foster dialogue and encourage community learning, particularly around issues of bias in medicine. To learn more about this initiative, please see our main page. We encourage students to offer feedback (examples of best practices and opportunities for improvement) on any learning experience on our feedback.(link is external and opens in a new window)
The Social Construction of Race
A key principle that informs our approach to an accurate, scientifically up-to-date curriculum that corrects historical bias is that race is a social construct (as opposed to a phenotypic trait such as skin pigmentation). Racial categories reflect historically created social labels rather than meaningful biological distinctions. Some of these labels are still in use today (e.g., US census data) as social categorizations. It is important for health care professionals to understand the nuanced distinctions between race, ancestry, and genetics and to avoid perpetuating the false notion that differences in health outcomes along racial lines are due to biological differences between races. For more information on the social construction of both race and gender see below for Addendum A.
With this context in mind, these guidelines identify 6 areas for consideration when developing curricular materials and teaching students:
- Be mindful of language, attitudes, and behaviors.
- Be comprehensive in representations of healthy/ “normal.”
- Be comprehensive in representations of pathology.
- Avoid stereotypes in representations of pathology.
- Explore structural reasons for differences in health outcomes.
- Acknowledge limitations of research.
1. Be mindful of language, attitudes, and behavior.
Example: Informal language is often a useful tool when teaching. However, educators risk alienating or distracting the learner from the teaching point, particularly if using informal language in reference to a particular identity group (see below for a partial list of domains of identity).
Solution: Educational research indicates that informal language, attitudes, and behavior in the educational setting powerfully affect student experiences. Educators should strive to use precise, accurate language. Examples:
- Eliminating the use of outdated, racist, and imprecise terms, e.g., “Oriental(link is external and opens in a new window)” to describe an Asian person or “Caucasian(link is external and opens in a new window)” to describe a white person.
-
- Census categories are recommended
- Disaggregation(link is external and opens in a new window) should be employed when possible
-
- Using person-first(link is external and opens in a new window) language, e.g., “a person with diabetes” instead of “a diabetic” or “a person with schizophrenia” instead of “a schizophrenic.”
- Using precise gender-related language, e.g., “people with uteruses” instead of “women” if the relevant point is about the presence of a uterus rather than the person’s expressed gender identity - see below for Addendum A.
- Avoiding stereotypes in examples (e.g. consistently referring to nurses as "she").
2. Be comprehensive in representations of healthy/“normal.”
Example: A textbook may describe healthy gums as being “coral pink” in color, when in fact healthy gums of persons of color may be pigmented; similarly, most examples of couples may depict heterosexual partners. Additionally, patients with disabilities are often only represented when their disability is discussed.
By broadening examples and descriptions of normal, educators can take advantage of opportunities to sharpen students' clinical acumen, thus preparing them to provide the highest quality care to the broadest possible patient population. Additionally, they may assist in the recognition of disability as existing within the spectrum of normal.
Solution: When describing human structure, function, or behavior, consider whether a representation applies universally or whether more varied descriptions are needed for scientific accuracy. For example, an educator may present images of gums of different normal pigmentations. Similarly, teachers can strive to craft cases highlighting patients with maximal variety that mirror those that students will encounter in clinical settings.
3. Be comprehensive in representations of pathology.
Example: Hyperbilirubinemia can present clinically as jaundiced skin. However, in highly melanated-skin, jaundice may be clinically difficult to appreciate.
Solution: An educator may discuss how disease presentations may vary across populations. In the example above, the educator can discuss challenges of identifying skin color changes in highly melanated-skin and recommend focusing on palms and sclera for clinical clues in these patients. Alternatively, in a case where the descriptors used for pathology are consistent across groups, such as peau d’orange to refer to the dimpled appearance of cutaneous lymphatic edema, the educator could be explicit about the universality of the clinical sign.
4. Avoid stereotypes in representations of pathology.
Example: A discussion on sexually transmitted infections (STIs) may use “typical” case examples to illustrate disease pathology and epidemiology as shorthand for highest risk (e.g. young people). This type of reinforcement of pattern recognition may cause students to inadvertently miss diagnoses in clinical practice.
Solution: Consider using varied case examples that illustrate both populations at highest risk (guided by evidence-based knowledge of population prevalence) while avoiding the impression that only those populations are at risk.
For instance, consider discussion of a geriatric person with an STI or sexual health of a woman who has sex with women. Another example of teaching that counteracts stereotypes and biases in clinical thinking might be to craft a case of a man who has sex with men who presents with a fever and turns out not to have HIV or an STI but rather has a common infectious process that any immunocompetent person might develop.
Further, consider discussing structural reasons behind high prevalence of certain pathologies in populations - such as discrimination.
5. Explore structural reasons for differences in health outcomes
Example: Medical literature now describes many examples of health disparities by race, socioeconomic status, and other variables. However, the mediators of those disparities are not always known and, if known, are not always discussed. This may leave some students with a misguided impression that genetic or biological differences drive such disparities.
Solution: Clarify potential reasons for health disparities. Conversations that focus on the structural reasons for health outcomes (e.g., discrimination, education, housing) may help students develop a more nuanced understanding of how structural racism, socioeconomic status, unconscious bias, and other factors impact health care. (Please see Addendum A for further discussion of this topic.)
6. Acknowledge limitations of research.
Example: It is common to encounter medical studies that enroll people in proportions that don’t represent the US population. The generalizability of those findings to underrepresented populations may be limited and lead to poor clinical outcomes.
Solution: If underrepresentation in study subjects by gender identity, sex, race, socioeconomic status, or another meaningful variable may limit the study’s generalizability, consider disclosing this as an educational point.
Acknowledging the limitations of research allows students to become aware of gaps in clinical literature and engage in the challenge of advancing medical knowledge. Scholarly dialogue may be encouraged even when study subjects appear similar because categories are often poorly defined (e.g., using a sample of a Yoruba population in Nigeria to compare “Africans” to “white Americans”). Even when racial minorities are included in research, the categorization of broad racial groups (e.g. African American, Asian) can often obscure precise outcomes, making it important for educators to consider when data disaggregation (link is external and opens in a new window)is appropriate.
Addendum A
The social construction of race (continued)
It is important to understand and convey race as a social construct to avoid perpetuating the false notion that differences in health outcomes along racial lines are due to biological differences between races and genders.
Because the concept of human races was created as a political tool, the meaning/definitions of racial classifications have changed internationally with time and political context. For example, the Nazi’s use of genocide to create a “master race.” Thus, racial categories reflect social conventions rather than biological distinctions. While these racial categorizations often facilitate individuals in forming communities around similarities in cultures or lived experiences (e.g., the deaf community,) they are not fixed categories that map onto scientific differences.
For instance, the relatively high US prostate cancer mortality among Black people are thought to be strongly related to structural factors, such as poorer access to high quality preventive and uro-oncologic care, as opposed to primarily a difference in prostatic histology or pathophysiology.
An educational framework that identifies race as a social construct and emphasizes structural antecedents of health and disease helps educators convey to their students an accurate understanding of a) structural inequities that drive many current health disparities and b) precision diagnoses and treatments related to culture, ethnicity, ancestry, and genetics.
The social dimensions of gender
Gender is rooted in socially constructed roles, behaviors, expressions, and identities of girls, women, boys, men, and gender non-binary people. A person’s gender is not necessarily indicative of an individual’s genetic makeup.
For example, of two people who identify as women, one may carry a Y chromosome while the other does not; one may have a uterus or ovaries or breasts or a penis while the other does not.
The portrayal of gender as distinct from biological phenomena such as genitalia or chromosomes offers students specific and accurate descriptions of medical findings and treatments.
Resources to learn more:
1. Bias
- NEJM review article(link is external and opens in a new window)that covers the topics in this curriculum (link is external and opens in a new window)(link is external and opens in a new window)
- Important review of bias in medicine. NEJM article Series: Recognizing Historical Injustices in Medicine and the (link is external and opens in a new window)Journal(link is external and opens in a new window)
The Journal and other medical institutions have historically advocated and justified the mistreatment of groups on the basis of their race, ethnicity, religion, gender, and physical or mental conditions. To grapple with this history, we have commissioned an independent group of historians to examine various aspects of these biases and injustices. The series is meant to start a conversation, to help us learn from our mistakes and equip us to prevent new ones from occurring.
- An excellent book by geneticists that detangles the concepts of race, ancestry and genetics. Graves, J. L., & Goodman, A. H. (2021). Racism, not race: Answers to frequently asked questions.Columbia University Press.
2. Language
3. Dermatologic variation: It can be difficult to find examples of pathology on a wide variety of skin pigmentations. Two useful resources are:
- The PCOM library has assembled a comprehensive list(link is external and opens in a new window)of databases, websites, and electronic books.
- VisualDx in UpToDate has images on a wide variety of skin tones – including their “skin of color” collection.
4. Research
- The NAS has assembled this comprehensive guide to increasing precision in medicine; a one-page summary is on the website. National Academies of Sciences, Engineering, and Medicine. (2024). Rethinking race and ethnicity in biomedical research(link is external and opens in a new window).