Review offers strategies for mitigating racial and ethnic disparities in health care algorithms
Abstract: https://www.acpjournals.org/doi/10.7326/M23-2960
URL goes live when the embargo lifts
A review of 63 studies related to health care algorithms found that use of these algorithms can mitigate, exacerbate, or not impact racial and ethnic disparities at all. The authors offer several strategies for health care systems to implement to mitigate these effects. The review is published in Annals of Internal Medicine.
Health care algorithms are frequently used to guide clinical decision making, resource allocation, and health care management. Although algorithms are developed to optimize specific processes of care, they may introduce or perpetuate racial and ethnic biases, leading to unequal treatment and contributing to or exacerbating unequal health outcomes.
Researchers from the University of Pennsylvania conducted a systematic review of 51 modeling, 4 retrospective, 2 prospective, 5 pre-post studies, and 1 randomized controlled trial. The authors found varying results, with some research indicating that health care algorithms mitigate racial and ethnic disparities, and other research indicating that these algorithms exacerbate these disparities or have no effect at all. After review, the authors identified seven strategies for potentially mitigating disparities in health care algorithms: removing an input variable, replacing a variable, adding race, adding a non–race-based variable, changing the racial and ethnic composition of the population used in model development, creating separate thresholds for subpopulations, and modifying algorithmic analytic techniques. According to the authors, these results highlight the need for more high-quality research, transparency, and monitoring of algorithms to detect and address biases in their application that may develop over time.
The author of an accompanying editorial from Priscilla Chan and Mark Zuckerberg San Francisco General Hospital and the University of California San Francisco expressed disappointment in the quality of available evidence included in the systematic review. The author suggests that clinical algorithms should be based on good research and need to consider important factors like socioeconomic factors and discrimination instead of just race. Factors that influence equity care go beyond race and include how society is structured, access to health care for some racial and ethnic groups, environmental factors affecting health, and the way some people of color may feel they are treated in the physician-patient relationship, especially considering the lack of representation in medicine.
Media contacts: For an embargoed PDF, please contact Angela Collom at acollom@acponline.org. To speak with the corresponding author Shazia Mehmood Siddique, MD, MSHP, please contact Shazia.Siddique@pennmedicine.upenn.edu.
