How we describe skin — and why we never ask your race

Sources

Retrieved and read for this guide; reviewed 2026-09-13. A source listed here says what it establishes and what it does not — a citation is not an endorsement.

1. [1]Integrating skin color assessments into clinical practice and research: A review of current approaches · Harvey, Alexis, Okeke et al. J Am Acad Dermatol (Europe PMC, PMID 38342247), 2024 · read at abstract levelEstablishes: Identified 17 skin classification systems. Fitzpatrick is the most widely used but has significant limitations, and current instruments have notable deficiencies.Does not establish: An endorsement of any single tool for consumer use.
2. [2]Rethinking the use of population descriptors in dermatology trials and beyond: disentangling race and ethnicity from skin color · Harvey, Lester, Jaleel, Takeshita, McMichael et al. Archives of Dermatological Research (Europe PMC, PMID 40252110), 2025 · read at abstract levelEstablishes: Consensus statement from a Skin of Color Society summit. Race and ethnicity categories do not accurately reflect biological differences. Recommends specific descriptors such as skin color, used alongside social factors, and objective color tools.Does not establish: A consumer self-report questionnaire — it specifies none.
3. [3]Recommended practices in skin tone ML research · Google, 2026Establishes: Lists 'DON'T: Equate skin tone to race'. Explains that Monk intended to decouple tone and race, since racial and ethnic groups span many tones, and notes that lighting affects perceived tone.Does not establish: Guidance written for health tools — it addresses machine-learning research.
4. [4]Skin cancers in skin types IV–VI: Does the Fitzpatrick scale give a false sense of security? · Goon, Banfield, Bello, Levell. Skin Health and Disease (PMC), 2021Establishes: Fitzpatrick devised the scale in 1975 to choose UVA doses for PUVA treatment of psoriasis, originally in lighter-skinned patients; it describes burn and tan response, not skin colour. The authors argue its 'never/rarely burns' labels for types V–VI are inaccurate and can falsely reassure, and that skin cancers in darker skin are often found late.Does not establish: New incidence data or a validated replacement scale — it offers neither.
5. [5]Limitations of Fitzpatrick Skin Type as a Proxy for Skin Color and Race · Garg, McMichael. Cutis (Europe PMC, PMID 42594399), 2026Establishes: An editorial exists arguing that Fitzpatrick skin type should not stand in for skin color or race (the title's own direction).Does not establish: Any specific argument or figure — only the record's metadata was read. Corroboration only.
6. [6]International Expert Consensus on Defining Skin of Color and Delivering Equitable Dermatologic Care · Lim HW et al. Int J Dermatol (PMC lists 2025; publisher lists 2026). PMID 41195697. Wiley / PMC, 2025Establishes: Literature review plus international expert panel. 'Skin of color' is acceptable but imperfect. Alternatives such as 'melanin-rich skin' were proposed. 'Ethnic skin' is rejected as imprecise, and national descriptors are preferred in parts of Asia. Fitzpatrick is practical but designed for fair skin. Do not use race alone as a proxy; combine skin tone with self-identification.Does not establish: A universally adopted standard — it acknowledges ongoing debate.
7. [7]Fitzpatrick Skin Type Self Reporting Versus Provider Reporting: A Single-center, Survey-based Study · J Clin Aesthet Dermatol (JCAD), 2024Establishes: Patient and provider Fitzpatrick ratings often disagreed, and providers tended to underestimate patients' ability to burn. Asking about 'becoming darker' instead of 'tanning' matched patients' self-assessment closely.Does not establish: Anything beyond a single centre, or a validated new instrument.
8. [8]Sun Advice for Skin of Colour · British Association of Dermatologists (Patient Hub), 2018Establishes: Darker skin can burn, though rarely in UK conditions. Advises checking palms, soles, nails, genitals and mucous membranes, and shade, clothing and SPF 30+ with UVA protection; mineral or iron-oxide sunscreens may help with hyperpigmentation. Advises a 10 microgram vitamin D supplement year-round and short summer sun exposure.Does not establish: Individualized advice, or conditions outside UK latitudes.
9. [9]Improving skin tone representation across Google · Google (The Keyword), 2022Establishes: Announces the 10-shade Monk Skin Tone Scale, developed with Harvard sociologist Dr. Ellis Monk and openly released for research and product use.Does not establish: Licence wording — it contains none.
10. [10]Skin Tone Research @ Google: Get started (Monk Skin Tone Scale) · Google, 2026Establishes: Official page for the Monk Skin Tone Scale. States the licence as Creative Commons Attribution 4.0 International, with the usage citation: Monk, Ellis. "Monk Skin Tone Scale," 2019. https://skintone.google. Warns that on-screen colors may look off and points to the original Orbs/Swatches files.Does not establish: Any health or UV-response meaning for the tones — it gives none. The page is client-rendered; its text was read from the site's own JavaScript bundle.
11. [11]Consensus and subjectivity of skin tone annotation for ML fairness · Schumann, Olanubi. Google Research blog, with preprint arXiv:2305.09073, 2023Establishes: Monk ratings of the same person varied by annotator region (US raters chose MST-5–7 where raters in India chose MST-3–5). Pooling diverse raters converged near expert values. MST-E example images are licensed only for annotator training, not for training ML models.Does not establish: That consumers can self-select a Monk tone matching measured skin color — it does not test this.
12. [12]Limitations of ITA for Skin Type Estimation Under Uncontrolled Imaging Conditions · Alipour, Burke, Courtney. Skin Research and Technology (PMC), 2026Establishes: ITA (Individual Typology Angle) is computed from L* and b* colorimetry values. From uncontrolled photos, fixed-threshold ITA mapped to Fitzpatrick groups reached 22.9% accuracy, below the 51.3% majority-class baseline; it failed entirely for type VI and was sensitive to lighting, especially in darker skin.Does not establish: Anything about ITA measured with a calibrated colorimeter in a clinic.
13. [13]How to fade dark spots in darker skin tones · American Academy of Dermatology, 2025Establishes: Advises tinted broad-spectrum SPF30+ sunscreen with iron oxide and treating the underlying cause. Lists azelaic, glycolic and kojic acids, retinoids and vitamin C as ingredients. Says to avoid irritating products, liquid bleach, and imported products with unlisted steroids or mercury. Fading takes 6–12 months for lighter spots and can take years for deeper ones.Does not establish: Any ranking of the ingredients, or trial evidence for them. Not individual advice.
14. [14]Postinflammatory hyperpigmentation · DermNet (NZ), 2015Establishes: PIH is more intense and longer-lasting in darker skin. Inflammation drives melanin production, with an epidermal form and a dermal (macrophage) form. Physical treatments can worsen PIH by injuring the epidermis.Does not establish: Any trial data or comparative efficacy — it gives none. The page was last updated in December 2015.
15. [15]Keloids: Overview · American Academy of Dermatology, 2022Establishes: People of all races get keloids. Black people are reported to have the greatest risk, and people of Asian, Latin American or Mediterranean descent are also at raised risk. Family history is present in about a third to a half of cases. Keloids do not fade by themselves.Does not establish: Individual risk, or piercing-specific prevention (that is on the treatment page).
16. [16]Keloid scars · NHS (UK), 2023Establishes: Higher likelihood is listed for South Asian, Chinese, African Caribbean or Black African descent, age 10–30, pregnancy and prior keloids. If at higher risk, avoid tattoos and piercings. Early acne treatment reduces scarring. Keloids cannot be prevented outright.Does not establish: Numeric risk.
17. [17]Razor bump remedies for men with darker skin tones · American Academy of Dermatology, 2022Establishes: Razor bumps come from coarse, curly hair curving back into the skin. Persistent bumps can leave deep grooves and raised scars. Gives shaving technique tips and advises seeing a dermatologist if bumps persist.Does not establish: Other body sites or other genders — it is written for men.
18. [18]Representations of race and skin tone in medical textbook imagery · Louie, Wilkes. Social Science & Medicine (Europe PMC, PMID 29501717), 2018 · read at abstract levelEstablishes: Content analysis of 4,146 textbook images: racial proportions roughly matched the US population, but skin tones shown were 74.5% light, 21% medium and 4.5% dark.Does not establish: Clinical consequences.
19. [19]Disparities in dermatology educational resources · Ebede, Papier. J Am Acad Dermatol (Europe PMC, PMID 17010750), 2006 · read at abstract levelEstablishes: Found insufficient representation of dark skin in dermatology educational images.Does not establish: Figures — the summary read gave none.
20. [20]Diversity in Dermatology Clinical Trials: A Systematic Review · Charrow, Xia, Joyce, Mostaghimi. JAMA Dermatology (Europe PMC, PMID 28055072), 2017 · read at abstract levelEstablishes: Across 626 randomized trials from 2010–2015, about 60% of US trials reported race; among those that did, 74.4% of participants were White.Does not establish: Effects on treatment outcomes.
21. [21]Assessment of Changes in Diversity in Dermatology Clinical Trials Between 2010-2015 and 2015-2020 · Chen et al. JAMA Dermatology (Europe PMC, PMID 35080592), 2022 · read at abstract levelEstablishes: Race reporting rose from 59.8% to 71.9% of trials. Only 38.1% of trials had at least 20% non-White participants in both periods.Does not establish: Trials outside the US specifically.
22. [22]Basal Cell Carcinoma, Squamous Cell Carcinoma, and Cutaneous Melanoma in Skin of Color Patients · Hogue, Harvey. Dermatologic Clinics (Europe PMC, PMID 31466591), 2019 · read at abstract levelEstablishes: Documents diagnostic and outcome disparities in skin cancer among patients with skin of color, and knowledge gaps caused by under-representation in research.Does not establish: Any figures — the abstract gives none.
23. [23]Diagnostic Disparities in Erythema Visibility: A Call to Redefine Inflammatory Assessment in Diverse Skin Tones · Forsyth A et al. Cureus 2025;17(10):e94930. Cureus / PMC, 2025Establishes: Melanin masks redness, so inflammation may look violaceous, grey or brown. Redness-based scoring and trial criteria can underestimate or exclude darker-skinned patients. Recommends using palpation (warmth, swelling), texture and patient-reported symptoms such as itch and pain.Does not establish: Validated questionnaire wording. Cureus is a lower-barrier journal; treat as supporting commentary.
24. [24]Common Dermatologic Conditions in Skin of Color · Frazier WT, Proddutur S, Swope K. Am Fam Physician 2023;107(1):26-34. American Academy of Family Physicians, 2023Establishes: Clinical review with SORT ratings. Reported prevalence: DPN about 33% in skin of color, PFB 45–83% in affected groups, and keloids higher in types III–VI than I–II. Recommendations with SORT grades: stopping shaving for at least 8 weeks for PFB (SORT C); avoiding close shaves and tight collars or helmets for AKN; laser for AKN (SORT B). Notes erythema looks different on darker skin.Does not establish: The primary prevalence studies, which were not retrieved. One PIH percentage in the fetched page looked garbled and is not used.
25. [25]Disparities in dermatology AI performance on a diverse, curated clinical image set · Daneshjou et al. Science Advances (DOI 10.1126/sciadv.abq6147); preprint arXiv:2203.08807, 2022Establishes: On the pathology-confirmed Diverse Dermatology Images set, leading dermatology AI models lost 27–36% ROC-AUC, performing worse on dark skin tones and uncommon diseases. Dermatologists also did worse on these images. Fine-tuning on diverse data narrowed the gap.Does not establish: Consumer apps in general.
26. [26]Skin conditions that lasers can treat · American Academy of Dermatology, 2021Establishes: Darker skin is more prone to burns and dark marks after laser treatment. Advises choosing a board-certified dermatologist experienced with lasers on darker skin, avoiding tanning beforehand, and disclosing all medications and supplements.Does not establish: Test spots or specific settings — it mentions neither.
27. [27]Find skin cancer: How to perform a skin self-exam · American Academy of Dermatology, 2023Establishes: A self-exam covers the whole body using mirrors, including fingernails, palms, soles, between the toes, scalp and buttocks. See a dermatologist for any spot that is different from the others or changing.Does not establish: That self-exams lower deaths from skin cancer — it does not show this.
28. [28]Finding skin cancer in darker skin tones · American Academy of Dermatology, 2025Establishes: In darker skin, skin cancer often appears on the palms, soles, fingers, toes, nails, mouth, buttocks, anus or genitals, and is often advanced at diagnosis. Warning signs include changing spots, dark streaks in nails, and sores that don't heal.Does not establish: Statistics — it gives none.