Racial/ethnic inequities in health are well described. Members of minority groups disproportionately suffer from conditions such as cardiovascular disease, diabetes, asthma, cancer, and other conditions.1,2 Inequities also exist within the health care system.3–6 The Institute of Medicine (IOM) report Unequal Treatment documented racial/ethnic disparities in the diagnosis and treatment of various conditions that persisted even after controlling for possible confounders such as comorbidities, stage of disease, age, site of care, and insurance status.7 The IOM detailed strong evidence of racial discrimination at all levels of health care and posited that health care provider bias and stereotyping may contribute to disparities. Prominent figures have recently advocated for institutional change to address these issues on a structural level.8
Race is not a biologically based, genetic, or innate difference but, instead, a social classification or construct that is imprecise, though with real consequences that are difficult to capture.9–11 Racial bias, prejudice, and discrimination are often subconscious or inadvertent. Studies using techniques such as the Implicit Association Test (IAT), an indirect measurement of social cognition, have found evidence of physician bias based on race and stereotypes regarding patient race and medical noncompliance.12–15 Studies have documented that this IAT-measured bias correlates in some instances with poor quality of patient interaction and treatment differences.12,15,16 Cultural competence in health care is one strategy to reduce or eliminate racial/ethnic disparities in health care.17,18 Another strategy is to examine how subconscious bias may be reinforced in medical education via inequity in representation of people of color within didactic materials.
Concomitant with racial/ethnic inequities in health care, there is general agreement on the poor representation of women’s health in preclinical course work.19 Curricular gender/sex bias can be defined as the omission of gender-related issues, stereotyping, imbalance in presentation of issues, application of study results to women even if they were not included in the research, and ignoring gender inequality in health and illness.20 The underrepresentation of women in various learning materials may hinder students’ education and actually lead to a false perception of the male body as the standard patient, thus diminishing familiarity with the female body.21,22
In addition to sex and racial bias in the formal medical curriculum, a widely recognized “hidden curriculum” may also contribute to bias in medical education. This unwritten curriculum is considered to be a subtler collection of knowledge, skills, attitudes, and behaviors that are neither consciously nor unconsciously intended by those who created the curriculum to be learned but are still perceived and/or absorbed by students.19,23 Part of this hidden curriculum consists of messages inadvertently communicated in lectures and may be a consequence of the larger structure or culture of the institution.24 In one study, clinical “case studies” were examined within a school’s medical school curriculum and found to be consistent with a “standard in medical education, in which white, male, and heterosexual are placed in a central, normative position.”19 Media studies have suggested that repeated exposure to images may alter our perceptions, expectations, and acceptance of others along with perpetuating or endorsing stereotypes.25 Although imagery presented to students in textbooks or PowerPoint slides is likely considered part of the formal curriculum, it may be better understood as a component of the “hidden curriculum.”
PowerPoint has become a large component of content delivery in course work, often taking the place of textbooks. Our thorough literature search revealed that to our knowledge, research using PowerPoint slides as the data source has not been performed regarding this visual aspect of the medical curriculum. In this study, we analyze the images presented to first- and second-year medical students at one school of medicine in the preclinical didactic course work. We describe the proportion of representations of individuals by gender and racial/ethnic group to better characterize the potential hidden curriculum presented to students. In this article, sex is defined as biological (genetic and anatomic) features or characteristics distinguishing women and men, whereas gender has a broader definition than sex and includes biological, social, and psychological qualities or properties of the spectrum of gender identity. As discussed above, race is a social category, and similarly, ethnicity can be defined as an element within a sociocultural system that defines and structures the relations among persons by essential group differences, though the classification system will vary cross-culturally. Racial/ethnic categories referenced in this study are reflections of the language used in the data or articles from which they were drawn.
Our intent was to determine whether the representation of humans in lecture material presented to preclinical medical students at our university is biased or inequitably distributed relative to the general population, in terms of the gender/race/ethnicity of persons used in lecture slide images. Specifically, what proportion of images depicted white persons or persons of color? And, what proportion of images presented male or female individuals?
Sample and procedures
Two of us (G.C.M., J.K.) examined the PowerPoint presentations (decks) available from course Web sites for 33 (out of a total of 36 courses, 3 were unavailable for analysis) courses in the preclinical medical school curriculum at the main campus of University of Washington School of Medicine (see Supplemental Digital Appendix 1 at http://links.lww.com/ACADMED/A330). We developed and agreed on the coding criteria initially and together established that we were coding for clear, commonly understood representations of humans, then noting when sex and race were clearly identifiable. Two of us (G.C.M., J.K.) independently analyzed half of the data set each after establishing reasonable interrater reliability (as described below). This study was deemed to be exempt from the institutional review board (IRB) approval process by the university’s IRB.
To ensure a reasonable interrater reliability, we randomly selected approximately 10% of the decks for a third coder (E.S.) to evaluate (Supplemental Digital Appendix 2, http://links.lww.com/ACADMED/A330). This coder underwent training to ensure that consistent criteria were used. We compared the totals for each of the five categories (human, female, male, white, person of color) for each set of two coders, and a two-sample t test assuming unequal variance was performed to test for a significant difference in the number of images rated in a particular category for each rater-pair. The coding differences were not significant except for between coders 2 and 3 regarding what was characterized as “human.” To determine which data to include for lectures that were coded in duplicate for interrater reliability purposes, we flipped a coin for each PowerPoint lecture deck.
We included images that were illustrations and photographs with a “human form,” defined as an outline or silhouette of a human body or external body part. Images excluded were single internal organs; internal anatomy cross-sections without a clear segment of skin, hair, or other surface markings; and depersonalized schematic diagrams. We initially tallied included images as “human,” and then, if possible, we further categorized them into either male or female. Any human images were also coded into further categories, white or person of color, if possible, regardless of ability to code for a sex category. Although we did not categorize images on the basis of the self-identification of the subjects, this approach reflects real-life experience—namely, that individuals frequently make race- and gender-based assessments of others based exclusively or predominantly on appearance, as explored in the introduction and discussion. Context of the images, such as text, lecturer, or subject, were not considered in this study.
To categorize sex, we looked at the human images for indicators such as the presence of genitals, reproductive organs, structures relating to sex-specific features (e.g., the testicular artery or presence of a placenta), secondary sex characteristics, “figure” (e.g., noticeable narrow or wide hips), or any indication in the accompanying text that the subject was male or female.26 Most coding decisions were ultimately made on the basis of more than one indicator. Images in which sex was indeterminate (either gender neutral, ambiguous, or insufficient information) were not assigned a category, though the image was still classified as “human.”
Coders categorized race in two categories, white or person of color (POC), based on physical characteristics such as skin color/complexion, facial features, hair type, or textual cues, and this led to a determination. The category POC included black/African American, Chicano/Latino, Native American, Middle Eastern, and Asian. The category white included Anglo, European, light-skinned individuals who could not otherwise be identified as POCs. Again, we ultimately made most coding decisions on the basis of more than one indicator. Images for which this was indeterminable (either ambiguous or insufficient information) were not assigned a category.
We performed a subanalysis of several courses for further detail and comparison with previous studies on the topic. Hormones/Nutrients and Reproduction were chosen to investigate the gender representation in courses with a general emphasis on the female reproductive endocrine function and organ/organ systems, respectively. Anatomy and Embryology was selected because of its status as a foundation in medical science as well as being the subject of other studies on this topic (e.g., anatomical textbooks).
Table 1 shows the total numbers of human images by sex and race/ethnicity. In total, we analyzed 34,219 slides. Of the 4,033 images that could be coded by sex, 60.5% (2,438) were male. Of the 5,230 images that could be coded by race/ethnicity, 78.4% (4,100) were white.
Table 2 shows that of the images coded by sex in the Hormones/Nutrients course, 59.5% (154) were male. However, of the images coded by race/ethnicity, 26.6% (89) were POCs, the highest of any of the subanalyzed courses.
Table 3 shows that the Reproduction course was the only course of the three to demonstrate a majority of female images, with 62.4% (226) female. The distribution of white persons and POCs was similar to the overall results, with 82.9% (377) white.
Table 4 shows a greater difference between female and male images, 36.8% (99) and 63.2% (170), respectively, in the Anatomy and Embryology course than the overall results. There was also a greater disparity between racial/ethnic-coded images, with white images making up 92.5% (369).
Finally, in Table 5, to evaluate the potential influence of images from reproductive course work, we subtracted the Reproduction course results from the overall results. Of the images coded by sex, the modified total was 62.7% (2,302) male. The modified total for images coded by race/ethnicity was 78.0% (3,723) white.
This study is the first known investigation of PowerPoint slides as preclinical didactic material examined for equal representation of images by sex and race. Our findings compare with previous studies of medical and general education materials.19–22,27 Our concern is that the inequity in representation of women and POCs will affect patient care delivery by physicians-in-training who are exposed to limited diversity in educational materials by failing to accurately portray the full diversity of patients. The findings demonstrate not only a majority of male images but also one of white images. Earlier studies excluded reproductive-focused materials out of concern that the emphasis on female subject material would skew the sample. In this study we included reproductive and other female-focused subject matter and nevertheless discovered a majority of male images. Our findings cannot be dismissed as sample bias or error because all available courses were examined.
The preponderance of white, male images in teaching slides might reflect the source of such images and the greater availability of these in anatomical textbooks, for example. With the advent of image search engines, and the expanded access to images they represent, it is worth discussing the optimal representation of women and persons of color in medical education.
As of 2011 in the United States, women made up 50.79% of the population and men 49.21%.28 Although women and men suffer many common illnesses at relatively similar rates, women’s reproductive health requires far more attention than male reproductive health, and the entire field of obstetrics–gynecology is a testament to this fact. It seems reasonable that there should be a larger proportion of female than male images in medical school didactics, beyond what was found in the reproductive course that we examined. The less controversial proposal is that the images should be representative of the general population.
Whether the proportion of images of white people and POCs should be representative of the general population or otherwise distributed should also be considered. According to the U.S. Census, white persons make up 78.1% of the U.S. population. Whereas the U.S. Census defines a minority as anyone who is not single-race white and not Hispanic (the proportion of white persons who are not Hispanic is 63.4%), this study included Hispanic and Middle Eastern individuals (for which there is currently no census category) as persons of color. Therefore, our findings may overrepresent those in the POC classification, which points to even greater inequity of representation.
The U.S. population has become increasingly diverse racially and ethnically. The non-Hispanic white population is growing at the slowest rate; Hispanic, Asian, and black population growth are the fastest. The U.S. Census Bureau recently released a set of estimates projecting that the United States will become a “majority-minority” nation by 2043, meaning that the largest single group will continue to be the non-Hispanic white population, but no group will constitute a majority.29,30 In preparing medical students for future practice, medical schools must consider patient demographics when designing educational materials.
There are several limitations to our study. The data collection was restricted to one school—a large, multicenter public university—and contained only PowerPoint slide images for those lectures presented at the main campus, and is thus not necessarily generalizable to other schools. Additionally, we are limited in that our racial/ethnic data collection of “persons of color” does not represent the true diversity of various general subcategories (such as African American, Latino, Southeast Asian, etc.). Another limitation involves the classification of sex as male or female in the typical binary approach, which did not capture the nuances of gender. Although our categorization of the images was subject to interpretation, the coding differences were not significant except for between coders 2 and 3 regarding what was characterized as “human.” This difference could be related to different interpretations of images that would be considered human (e.g., cartoons, stick figures). Because our intent was to use commonly understood ideas about representations of sex and race, and coders agreed to exclude ambiguous images, we believe this is a small possible limitation.
Between 2009 and 2011, matriculation at the University of Washington School of Medicine was majority women, at 54%. The students were 65.7% Caucasian, 10% unreported, and 24.3% persons of color. Data on faculty are incomplete. The impact of these images in the slides cannot be fully understood outside of the setting in which they were presented. However, this study highlights implications for further research, which should analyze the context and narrative in the images presented in the curriculum. A qualitative analysis could be performed on images in the medical school curriculum, perhaps focusing on common stereotypes. Although the results are for one medical school, our study provides a methodological road map for other medical schools to analyze their didactic materials. The impact of images used in medical school curricula could be further assessed by examining attitudes and knowledge of medical students regarding sex, race, and cultural competence as they progress in their careers.
Increasing awareness of image content, stereotyping, and the need for equitable representation among faculty, while also providing access to alternative images, may improve gender and racial/ethnic equity in imagery in medical education materials. One solution to inequity in didactic imagery would be to create an image repository or bank with a diverse selection of skin pigmentation, gender representation, and other characteristics. These images could be offered without copyright issues and could also be selected to minimize stereotypes or stigmatization of vulnerable or underserved populations (see Supplemental Digital Appendix 3 for examples, http://links.lww.com/ACADMED/A330). Finally, medical education materials could be improved by community-based research with underrepresented populations to determine how best to represent them in teaching. It would be ideal to train medical students with images in their course work that represent the patient population they will serve.
Acknowledgments: The authors wish to thank Roger Rosenblatt, Colleen Martin, Sherry Martin, Jim Martin, Rick Fisher, Robert Jones, Dale Terasaki, Hedy Lee, Kelly Treder, Brett Bell, and Wen Wei Loh.
1. Centers for Disease Control and Prevention. CDC health disparities and inequalities report. MMWR Morb Mortal Wkly Rep. 2013;62(suppl 3). http://www.cdc.gov/minorityhealth/CHDIReport.html
. Accessed December 3, 2015.
2. Williams DR, Mohammed SA. Discrimination and racial disparities in health: Evidence and needed research. J Behav Med. 2009;32:2047.
3. Braveman PA, Egerter SA, Mockenhaupt RE. Broadening the focus: The need to address the social determinants of health. Am J Prev Med. 2011;40(1 suppl 1):S4S18.
4. Braveman P. Health disparities and health equity: Concepts and measurement. Annu Rev Public Health. 2006;27:167194.
5. Solar O, Irwin A. A Conceptual Framework for Action on the Social Determinants Of Health. Social Determinants of Health Discussion Paper 2 (Policy and Practice). 2010.Geneva, Switzerland: World Health Organization.
6. Williams DR, Sternthal M. Understanding racial–ethnic disparities in health: Sociological contributions. J Health Soc Behav. 2010;51(suppl):S15S27.
7. Institute of Medicine. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. 2002.Washington, DC: National Academies Press.
8. Bassett MT. #BlackLivesMatter—a challenge to the medical and public health communities. N Engl J Med. 2015;372:10851087.
9. Witzig R. The right to identity: Implications of using subjectively-assigned race in US healthcare. Soc Med. 2014;8:4352.
10. Beagan B. Micro inequities and everyday inequalities: “Race,” gender, sexuality and class in medical school. Can J Sociol. 2001;26:583610.
11. Jones CP. Levels of racism: A theoretic framework and a gardener’s tale. Am J Public Health. 2000;90:12121215.
12. Green AR, Carney DR, Pallin DJ, et al. Implicit bias among physicians and its prediction of thrombolysis decisions for black and white patients. J Gen Intern Med. 2007;22:12311238.
13. Sabin JA, Rivara FP, Greenwald AG. Physician implicit attitudes and stereotypes about race and quality of medical care. Med Care. 2008;46:678685.
14. Johnson RL, Roter D, Powe NR, Cooper LA. Patient race/ethnicity and quality of patient–physician communication during medical visits. Am J Public Health. 2004;94:20842090.
15. Cooper LA, Roter DL, Carson KA, et al. The associations of clinicians’ implicit attitudes about race with medical visit communication and patient ratings of interpersonal care. Am J Public Health. 2012;102:979987.
16. Sabin JA, Marini M, Nosek BA. Implicit and explicit anti-fat bias among a large sample of medical doctors by BMI, race/ethnicity and gender. PLoS One. 2012;7:e48448.
17. Betancourt JR, Green AR, Carrillo JE, Ananeh-Firempong O 2nd. Defining cultural competence: A practical framework for addressing racial/ethnic disparities in health and health care. Public Health Rep. 2003;118:293302.
18. Expert Panel on Cultural Competence Education for Students in Medicine and Public Health. Cultural Competence Education for Students in Medicine and Public Health: Report of an Expert Panel. 2012.Washington, DC: Association of American Medical Colleges and Association of Schools of Public Health.
19. Turbes S, Krebs E, Axtell S. The hidden curriculum in multicultural medical education: The role of case examples. Acad Med. 2002;77:209216.
20. Dijkstra AF, Verdonk P, Lagro-Janssen AL. Gender bias in medical textbooks: Examples from coronary heart disease, depression, alcohol abuse and pharmacology. Med Educ. 2008;42:10211028.
21. Lawrence SC, Bendixen K. His and hers: Male and female anatomy in anatomy texts for U.S. medical students, 1890–1989. Soc Sci Med. 1992;35:925934.
22. Alexanderson K, Wingren G, Rosdahl I. Gender analyses of medical textbooks on dermatology, epidemiology, occupational medicine and public health. Educ Health. 1998;11:151163.
23. Murray-García JL, García JA. The institutional context of multicultural education: What is your institutional curriculum? Acad Med. 2008;83:646652.
24. Hafferty FW. Beyond curriculum reform: Confronting medicine’s hidden curriculum. Acad Med. 1998;73:403407.
25. Banks JA. Banks JA, Banks CAM. Knowledge construction and popular culture. Handbook of Research on Multicultural Education. 2003:San Francisco, Calif: Jossey-Bass; 211227.
26. Mendelsohn KD, Nieman LZ, Isaacs K, Lee S, Levison SP. Sex and gender bias in anatomy and physical diagnosis text illustrations. JAMA. 1994;272:12671270.
27. Giacomini M, Rozee-Koker P, Pepitone-Arreola-Rockwell F. Gender bias in human anatomy textbook illustrations. Psychol Women Q. 1986;10:413420.
28. U.S. Census Bureau. Current population survey, annual social and economic supplement, 2011. http://www.census.gov/population/age/data/2011comp.html
. Published November 2012. Accessed December 3, 2015.
29. U.S. Census Bureau. 2011 population and housing unit estimates. http://www.census.gov/popest/index.html
. Published 2011. Accessed December 3, 2015.
30. U.S. Census Bureau. 2012 national population projections. http://www.census.gov/population/projections/data/national/2012.html
. Published 2012. Accessed December 3, 2015.