Skip to main content
Dear Editor,
We write as two long-standing family physician researchers working in the public primary-care network of Spain’s National Health System. The brief report by Ding and colleagues on offering four bandage shades, recently published in The Annals of Family Medicine (1), is modest in scale yet methodologically sound and analytically transparent. Its results—44 consecutive vaccine-centre attendees, 63 % of whom selected a shade matching their skin and nearly half of whom reported a more positive view of the clinic—add quantitative weight to the intuition, shared by many clinicians, that apparently trivial material cues can affect how welcome patients feel. We found both the design (an anonymous, validated survey adapted from established satisfaction instruments) and the framing (quality-improvement, not hypothesis-testing) appropriate, and the authors’ restraint in interpretation exemplary. Their work invites reflection on how primary care in other publicly funded systems might adopt, extend, and critically evaluate similar inclusion-centred micro-interventions.
The study also resonates with the empirical literature linking small cues of respect to trust. In Spanish primary care, rapport is diagnostically valuable because many presentations are non-specific or evolve over time. When relational empathy is high, as measured by the Consultation and Relational Empathy (CARE) instrument (2), patients are more willing to re-attend, disclose sensitive information, and adhere to advice. The low-cost gesture of letting a patient choose a bandage whose colour does not advertise the wound can therefore be framed as a structural support for empathic consultation rather than mere cosmetics. Importantly, CARE studies show no consistent association between empathy scores and consultation length; what matters are attentiveness and tailoring, exactly the qualities implicitly signalled by colour-matched dressings.
Historical parallels strengthen the case. Lorna Roth’s analysis of the Shirley Card, the light-skinned photographic standard that calibrated film chemistry for decades, illustrated how technological defaults become cognitive norms (3). Because early emulsions were balanced on Caucasian skin, darker subjects were systematically under-exposed; users internalised the message that deviation from “normal” skin required correction. The persistence of that bias into digital sensors and colour spaces is well documented. Louie and Wilkes showed that 74 % of textbook images in major medical schools depict light skin and that darker tones are not only fewer but often stereotyped (4). Such visual under-representation shapes diagnostic mental models long before practitioners meet diverse patients.
These technological legacies now intersect with machine learning. Adelekun and colleagues demonstrated that core dermatology texts still under-represent Fitzpatrick IV–VI skin, with only 15 % of photographs depicting these tones (5). Guo et al. mapped the same problem in convolutional networks trained for skin-cancer triage: most U.S. datasets draw overwhelmingly from a few states and lighter phenotypes, a limitation that carries forward into published algorithms (6). Hardware is not exempt. The 2020 correspondence by Sjoding et al. in NEJM confirmed that pulse-oximetry readings mis-estimate hypoxaemia three times more often in Black than White patients (7). In other words, if clinicians ignore the skin-tone dimension of device design, they risk embedding inequity at point-of-care.
Proposed remedies include richer annotation schemes for complexion. The open-source Monk Skin Tone (MST) scale, developed with Google (8), and follow-up psychometric work comparing MST with Fitzpatrick and cosmetics-industry palettes (9), provide practical tools for auditing datasets and vendor catalogues. López-Pérez et al. recently showed that a variational auto-encoder trained on the Fitzpatrick17k archive reconstructs lighter phenotypes more faithfully than darker ones, even when uncertainty estimates are added (10). Such work suggests that inclusive physical artefacts (e.g. multi-shade dressings) and inclusive digital artefacts (e.g. balanced image corpora) are mutually reinforcing.
From a Spanish vantage-point, several pragmatic lines of action follow. First, we could replicate Ding et al.’s survey across urban and rural centres with higher migrant density, embedding it within the established Primary-Care Information System (SIA-AP). A stepped-wedge rollout would let each practice serve as its own control while giving procurement teams lead time to secure diversified supplies. Second, medical faculties could pair the intervention with an audit of teaching materials, benchmarking slides against MST quantiles and updating image banks accordingly. Third, regional health services negotiating electronic-health-record upgrades should require that AI decision-support modules disclose the complexion distribution of their training data, echoing the European Health Data Space draft clauses on algorithmic transparency.
The letter by its nature is prospective, and our enthusiasm must be tempered by attention to confounding factors. Ding et al. report only 44 volunteers; the confidence intervals around their effect estimates are therefore wide, and the setting—a COVID-19 vaccination walk-through—differs from routine consulting where the doctor rather than the nurse applies dressings. Moreover, patient–clinician concordance in ethnicity or language was not measured, and both may moderate the perceived value of inclusive consumables. The intervention itself risks tokenism if not accompanied by broader competence, such as the ability to recognise urticaria on dark skin or explain why cyanosis looks grey rather than blue.
Nonetheless, size and scope should not preclude adoption. The bandage intervention exemplifies the incremental, test-and-refine ethic of quality improvement that fits public primary care: low capital outlay, negligible training burden, clear patient-reported outcomes. We advocate framing such gestures within national equity strategies rather than isolated “diversity days”. Spain’s evolving primary-care roadmap already contains language on person-centredness and cultural adaptation; what remains is to translate that language into purchasing specifications and feedback loops.
Finally, we note the symbolic power of the intervention. A choice of bandage shade is not a panacea for deep structural inequities, yet symbolism matters precisely because it attaches values to everyday objects. When a patient sees medical consumables designed with them in mind, they receive a non-verbal assurance that the system expects their presence and plans for their comfort. Conversely, an environment where every wall poster shows pale skin and every wound dressing contrasts starkly with brown arms signals that darker bodies are after-thoughts. Our duty as family physicians, committed to continuity and community orientation, is to recognise and correct such signals.
In our opinon, Ding et al. remind us that inclusivity can be operationalised at the scale of adhesive strips. By integrating evidence from visual-culture studies, device-performance research, and algorithmic fairness, we can craft primary-care environments in which diversity is reflected not only in rhetoric but in the artefacts that touch patients’ skin. We urge Spanish colleagues to pilot, measure, and iteratively improve these simple innovations, documenting both intended and unintended effects. Doing so will enrich the evidence base on which future procurement, teaching, and technological standards are set, and will bring us one step closer to a practice where every patient, whatever their complexion, sees themselves reflected in the tools of care.
REFERENCES
1. Ding H, Schiller EA, Cascio JL, McLean I, Islam S, Arabadjian M, et al. Cultivating Inclusivity in Medicine: The Impact of Offering Bandage Options of Various Skin Shades on Patient Experience. Ann Fam Med. 2024 Mar 1;22(2):172–172.
2. Mercer SW, Maxwell M, Heaney D, Watt GC. The consultation and relational empathy (CARE) measure: development and preliminary validation and reliability of an empathy-based consultation process measure. Fam Pract. 2004 Dec 1;21(6):699–705.
3. Roth L. Looking at Shirley, the Ultimate Norm: Colour Balance, Image Technologies, and Cognitive Equity. Can J Commun. 2009 Mar 30;34(1):111–36.
4. Louie P, Wilkes R. Representations of race and skin tone in medical textbook imagery. Soc Sci Med. 2018 Apr 1;202:38–42.
5. Adelekun A, Onyekaba G, Lipoff JB. Skin color in dermatology textbooks: An updated evaluation and analysis. J Am Acad Dermatol. 2021 Jan 1;84(1):194–6.
6. Guo LN, Lee MS, Kassamali B, Mita C, Nambudiri VE. Bias in, bias out: Underreporting and underrepresentation of diverse skin types in machine learning research for skin cancer detection—A scoping review. J Am Acad Dermatol. 2022 Jul 1;87(1):157–9.
7. Sjoding MW, Dickson RP, Iwashyna TJ, Gay SE, Valley TS. Racial Bias in Pulse Oximetry Measurement. N Engl J Med. 2020 Dec 16;383(25):2477–8.
8. Monk E. The Monk Skin Tone Scale [Internet]. OSF; 2023 [cited 2025 Apr 29]. Available from: https://osf.io/pdf4c_v1
9. Heldreth CM, Monk EP, Clark AT, Schumann C, Eyee X, Ricco S. Which Skin Tone Measures Are the Most Inclusive? An Investigation of Skin Tone Measures for Artificial Intelligence. ACM J Responsib Comput. 2024 Mar 20;1(1):7:1-7:21.
10. López-Pérez M, Hauberg S, Feragen A. Are generative models fair? A study of racial bias in dermatological image generation [Internet]. arXiv; 2025 [cited 2025 Apr 29]. Available from: http://arxiv.org/abs/2501.11752