Medscape sat down with Jenna Lester, MD, associate professor and director of the Skin of Color Program at the University of California, San Francisco (UCSF), to discuss the factors that contribute to the lack of diverse medical imagery in dermatology and what medical professionals can do to lessen such disparities and improve care for impacted patients.

Medscape: Can you briefly describe the most significant disparities in dermatologic conditions diagnosed in patients of color?

Jenna Lester, MD: First, I'd like to open our conversation by stating that race and ethnicity are social constructs that should not be mistakenly viewed as tied to any intrinsic biological or genetic difference.^[1,2] Rather, race and ethnicity have arisen through a complex interplay between several factors, including geographic, social, and cultural influences that impact the biology of a given population.^[1,2] This can, in turn, affect the types of diseases or health conditions an individual or group of individuals develop. While individuals with darker skin color can develop the same skin conditions as those with lighter skin color, racial/ethnic disparities do exist in dermatologic care.^[3]

Certain dermatologic conditions and disease processes may manifest differently in individuals with darker skin tones.^[3,4] For example, melasma is a hyperpigmentation disorder that commonly affects women with darker skin color (eg, Hispanic, Asian) and is triggered by a range of factors including sun exposure and estrogen, resulting in the formation of dark spots or patches that can be tan, brown, or bluish gray.^[5,6] Post-inflammatory pigmentary alteration (or post-inflammatory hyperpigmentation) is another condition where dark spots form following acne formation and other triggers, such as atopic dermatitis, allergies, injuries, and cosmetic procedures. While this can occur in individuals with any skin tone and its underlying causes are varied, it is often more severe and noticeable in those with darker skin due to the development of dark, flattened spots on the body that can be brown to black in coloration.^[7] On the other hand, certain skin disorders like central centrifugal cicatricial alopecia (CCCA) appear to have a predilection for specific groups of individuals (Black women over 30 years of age), but the underlying reasons for this remain unclear.^[8]

Other conditions and diseases that may look different on darker skin include psoriasis, atopic dermatitis, eczema, and many other skin conditions, primarily because of how inflammation appears in dark skin.^[9,10] Inflammation is usually depicted as bright pink or red in light skin but looks very different in dark skin. Therefore, conditions characterized by the prominent appearance of inflammation can have distinct manifestations in people with dark skin tones. For example, in patients with early-stage Lyme disease, a characteristic "bullseye" rash called erythema migrans, develops. This rash takes on a dark red, brown, or purple color in patients with darker skin tones, while it appears bright pink to red in color in patients with light skin.^[11-14] The issue is further exacerbated by the fact that the word erythema itself originates from the Greek word for red and thus is a descriptor for red color, grounded in the way inflammatory conditions appear in White skin.^[15] As a result, many dermatologists advocate using a different word that more directly describes inflammation than redness since those 2 things are different.

Medscape: What is the role of dermatologic imagery/illustrations in diagnosing skin conditions?

Dr Lester: In dermatology, we are trained to recognize patterns and colors. As a dermatology resident, you spend a significant amount of time analyzing many photos of different skin conditions and understanding how those visual features may be connected to a particular underlying disease pathology. This takes me back to a study I conducted with a group of UCSF medical students. We looked at photos (N = 5026) in 2 common textbooks and a teaching set and observed that most images depicted dermatologic conditions in patients with lighter skin (~22% to 32% of images depicted skin of color).^[16] However, we discovered an alarming exception. When images were analyzed on the individual chapter level, those highlighting sexually transmitted infections tended to use darker skin types (up to 58% of images) to model the condition.^[16] Similar reports have shown that images of darker skin tones make up a mere 4% to 19% of images found in medical texts, supporting our findings.^[1,17-20] These data suggest a systemic issue is rooted within our dermatology training curriculums, posing a critical problem that is especially detrimental to learners. This can have long-term implications, as relying on images that lack diversity can make it difficult for those same students, who might eventually become dermatologists, to identify and diagnose common conditions in patients with darker skin.^[21-23]

Medscape: Can you describe some key factors contributing to the underrepresentation of diverse skin tones in medical imagery?

Dr Lester: One potential contributing factor that may promote underrepresentation in medical imagery is that individuals of color, especially Black, Latinx, and Indigenous people, tend to access dermatologic specialty less frequently than White patients.^[24,25] Additionally, the ability to capture photos of a given skin condition may depend on several things, including the clinician's interest in using it as an example for educational purposes, as well as the patient's willingness to allow an image to be taken.^[26,27] The latter may be of particular concern for groups historically impacted by systemic racism and an untrustworthy medical system.^[16,19] Given this context, physicians should be sure to obtain explicit permission from patients to take photos and use them in particular scenarios. For example, if you are capturing a photo to monitor how a patient's disease changes over time, it has ethical and legal implications to repurpose those photos for use in research, education, or commercial uses. Therefore, it is important to ask patients when and how they would be most comfortable with their images being used.^[26-28]

Medscape: What are the consequences of skin tone misrepresentation in medical education resources used to train dermatologists and other healthcare professionals?

Dr Lester: Lack of diverse medical imagery, as I mentioned earlier, can impact one's ability to make an accurate diagnosis. When clinicians cannot accurately diagnose common skin conditions in individuals with darker skin tones, this can have several consequences. Patients may receive a delayed diagnosis, preventing them from identifying the disease in its early stages where it might be more amenable to therapy.^[16,19,20] Patients might also be misdiagnosed, potentially increasing unnecessary suffering and worse outcomes,^[29,30] or receive unnecessary invasive procedures such as biopsies for common skin conditions that do not present "classically" due to differences in presentation on darker skin.^[3,16] For example, melanoma does not typically present in sun-exposed areas in patients with darker skin. In fact, among Black patients, melanoma develops independent of ultraviolet (UV) exposure.^[31,32] Although the data around melanoma mortality in Black patients are complex and imperfect, the information we have now shows that melanoma is associated with poorer clinical outcomes (eg, diagnosis of late-stage melanoma, increased all-cause mortality) in patients with ethnic/racial backgrounds (eg, non-Hispanic Black patients), which may in part be due to delays in diagnosis.^[33-36]

Medscape: Are there strategies healthcare professionals and medical educators can implement to help reduce disparities promoted by the lack of diverse skin tone representation in medical texts?

Dr Lester: By 2060, it is expected that ~57% of the US population will have non-White skin.^[37,38] Therefore, it will become increasingly critical for dermatologists to address health inequities among racial/ethnic groups, which as we just discussed, may be driven by underrepresentation of skin tone in medical imagery. First, improving access is key. There are realities of our healthcare system that make it very difficult for some people to access dermatologic care. For example, many patients on a public insurance option (ie, Medicare or Medicaid), can find it challenging to obtain dermatology care.^[39,40] This access issue is further compounded by location. A recent study showed that there are fewer dermatologists in counties that have more racial diversity (ie, United States).^[3,29,41] It is also important that there be continued efforts to increase diversity in the dermatology workforce.^[3,29] When patients are treated by clinicians who look like themselves, they may be more receptive to seeking treatment.^[3] I'm a firm believer that all physicians are responsible for caring for all patients and these data on racially concordant visits represent the fact that, as a healthcare system, we have more work to do to become trustworthy and make patients feel comfortable regardless of who they are seeing. This is an attainable goal.

As dermatologists, we must also be cognizant that common disorders can present differently when assessing patients of color. This means that efforts are needed to increase the repertoire of available images demonstrating dermatologic disorders on darker skin tones.^[16] Accordingly, more experts in skin of color should become involved in authoring commonly used medical textbooks.^[16] Similarly, healthcare professionals should help expand our current databases of dermatologic photos by imaging common dermatoses in patients with darker skin tones.^[16] The Journal of the American Academy of Dermatology recently published a list of valuable resources to help clinicians access images and examples of skin conditions in diverse skin tones.^[42] There is also a growing body of print literature and atlases depicting dermatologic images of darker skin tones that clinicians can refer to, including the Atlas and Synopsis for Skin of Color, Dermatology Atlas for Skin of Color, and A Diverse and Inclusive Atlas just to name a few.^[9,32]

Medscape: Do you have any concluding remarks you would like to share with our learners?

Dr Lester: The role of artificial intelligence (AI) as a medical tool to aid clinicians in diagnosis and treatment decisions is beginning to emerge. However, before AI can be widely deployed in dermatology to make skin diagnoses, we must first address how the algorithms are trained. With each AI device, we should examine training data to make sure it is diverse and is not perpetuating the same biases we have as clinicians. Research shows that algorithms perform better on diverse skin tones if they have been trained on diverse skin tones. Another reasonable step would be to develop studies focused on developing modified scales for precisely estimating skin tone with reduced bias, which may one day lead to more objective AI technology that can be used in dermatologic care.^[43-48]

Watch Dr Lester's video vignette to learn more about the potential role of AI technology in dermatology care and education.^[43,46,49]

If image biases fail to be addressed, AI technology will have difficulty moving forward as a medical tool to potentially help diagnose and inform treatment decisions. Therefore, AI and machine learning technologies must be developed with diversity and inclusivity in mind.^[47,48] This is important since the underrepresentation of medical imagery can perpetuate biases that are detrimental in several ways, including worsening healthcare gaps and outcomes for patients of color.

AI is also being studied to assess medical teaching materials for biases in skin tone imagery. The recent STAR-ED study revealed that an automated machine learning algorithm can significantly detect imbalances in skin tone representation among 4 commonly used medical textbooks, suggesting AI may one day help medical educators, publishers, and clinicians assess skin tone diversity in the materials that are used to educate and train the future dermatology workforce.^[50]