#AllHuesMatter: Addressing Health Disparities Perpetuated Through Clinical Decision Support Tools (Spotlight Poster)

Learning objective/purpose: The evolution of clinical decision support tools (CDSTs) has guided the proliferation of these tools within the healthcare landscape. These multiple tools are critical for telephone triage, as they provide guidance on the appropriate level of care, potentially reducing healthcare costs. Despite these advances to improve the quality of care, research suggests these tools may exacerbate health disparities among already marginalized populations. This evidence-based project delves into and discusses algorithmic bias within CDSTs. It addresses biases related to the skin assessment during the telephone triage process when utilizing CDSTs. Through this educational activity, participants will gain an understanding of how biases embedded in CDSTs can adversely impact disparities for darkly pigmented patients and learn about interventions implemented to mitigate these biases.

Description: Health disparities related to skin assessment biases woven into CDSTs are a complex and sensitive topic that can impact patients with darkly pigmented skin. The biases embedded into CDSTs primarily stem from the development of algorithms utilizing data that focuses on patients with lightly pigmented skin, and a lack of diverse interdisciplinary teams to help shape and develop these tools. Research affirms that normative descriptive words such as red, blue, pink, blanch, or pale during the skin assessment of darkly pigmented patients results in inappropriate diagnoses leading to poorer health outcomes. To address biases in CDSTs that contribute to health disparities, best practices support interventions incorporating comprehensive and inclusive language into clinician education and CDSTs related to the skin assessment of darkly pigmented patients.

Evaluation/outcome: A preliminary evaluation of nursing staff was completed to assess their knowledge on the differences of skin color changes in various patient populations and their comfortability related to this sensitive topic. Of 35 RN respondents, results revealed approximately 20% of nurses felt neutral to extremely uncomfortable asking additional questions pertaining to skin color.
The telephone triage CDST was evaluated for algorithm bias in skin assessment, which uncovered notable variations for the recommended dispositions based on positive and negative responses to questions related to skin color changes. Differences with the recommended level of care ranged from “911 NOW” for the positive response to skin color changes, referring to “blue color around lips,” compared to “within 24 hours” for a negative response.
Staff underwent training in inclusive skin assessment including images to aid in knowledge. In the post-education survey, 100% of respondents expressed intent to apply the knowledge gained and expressed confidence that it would improve their ability to assess patients more accurately. Furthermore, an interdisciplinary team is assessing modifications to the CDST, aiming to incorporate inclusive language and resources into the tool.