Access to health information online has resulted in increased internet use to !nd medical education resources.1,2 The United States (US) population is comprised of 60.1% Whites, 18.5% Hispanics or Latinos, 13.4% Blacks or African Americans, with trends toward increasing diversity. Our objectives were to analyze skin color distributions on the American Academy of Dermatology (AAD) public education website.
A cross-sectional study of “Diseases & Conditions†on the AAD public education website was conducted to identify skin colors, June 30 – July 2, 2020. Both authors categorized images by Fitzpatrick skin type. Images were labeled as indeterminate due to lighting, image framing, or disease extent. The analysis included those conditions with greater than seven images to ensure more representative sets.
On average, skin types I/II represented 55%, III/IV 24%, and V/VI 17% of images, with 4% indeterminate. Types V/VI were highly represented in vitiligo and keloid sections. Sarcoidosis and scar pages had roughly equal representations of all skin types (Figure 1). For stock images, 351/490 (72%) depicted Fitzpatrick I/II, 51/490 (10%) III/IV, 43/490 (9%) V/VI skin types, with 7/490 (1%) indeterminate, and 38/490 (8%) mixed. Mixed images depicted a group with various skin colors.
A cross-sectional study of “Diseases & Conditions†on the AAD public education website was conducted to identify skin colors, June 30 – July 2, 2020. Both authors categorized images by Fitzpatrick skin type. Images were labeled as indeterminate due to lighting, image framing, or disease extent. The analysis included those conditions with greater than seven images to ensure more representative sets.
On average, skin types I/II represented 55%, III/IV 24%, and V/VI 17% of images, with 4% indeterminate. Types V/VI were highly represented in vitiligo and keloid sections. Sarcoidosis and scar pages had roughly equal representations of all skin types (Figure 1). For stock images, 351/490 (72%) depicted Fitzpatrick I/II, 51/490 (10%) III/IV, 43/490 (9%) V/VI skin types, with 7/490 (1%) indeterminate, and 38/490 (8%) mixed. Mixed images depicted a group with various skin colors.
The AAD public education website represents diverse arrays of skin colors. Conditions with fewer Fitzpatrick III–VI images were supplemented with text about skin color. Examples are “People with skin of color can get rosaceaâ€, “10 tips for clearing acne in skin of colorâ€, “How can an African American woman treat scalp psoriasis?â€, “Can signs of psoriasis vary with skin toneâ€, “Can you get psoriasis if you have skin of color?â€, and ringworm presentations in skin of color.
Skin cancer images in darker skin types are sparse (Figure 2); these deficits are partially compensated by supplementary text. In the “Skin cancer in people of color†section, it is emphasized that skin cancer diagnosis in people of color is often detected in late stages, with recommendations for improved identification. In the cutaneous T-cell lymphoma section, increasing prevalence in African Americans and presentation of “light spots†in skin of color is cited. Eight commonly used dermatology textbooks and web-based resources also reported scarce skin cancer dark skin images (V/VI). These sources, on average, had lower proportions of dark skin images for melanoma (2.9%) and basal cell carcinoma (2.4%) compared to the AAD public education website." Diagnostic accuracy of skin neoplasms using google reverse images was significantly lower in skin of color (IV– VI).4 The 5-year survival of patients with malignant melanoma
Skin cancer images in darker skin types are sparse (Figure 2); these deficits are partially compensated by supplementary text. In the “Skin cancer in people of color†section, it is emphasized that skin cancer diagnosis in people of color is often detected in late stages, with recommendations for improved identification. In the cutaneous T-cell lymphoma section, increasing prevalence in African Americans and presentation of “light spots†in skin of color is cited. Eight commonly used dermatology textbooks and web-based resources also reported scarce skin cancer dark skin images (V/VI). These sources, on average, had lower proportions of dark skin images for melanoma (2.9%) and basal cell carcinoma (2.4%) compared to the AAD public education website." Diagnostic accuracy of skin neoplasms using google reverse images was significantly lower in skin of color (IV– VI).4 The 5-year survival of patients with malignant melanoma
