The contested meanings of race and ethnicity in medical research: A case study of the DynaMed Point of Care tool.

Abstract

Although the use of race and ethnicity for diagnostic purposes remains a controversial practice given the socially contingent meaning of the terms (Bowker and Star, 1999), health researchers continue to report possible relationships between health outcomes and race/ethnicity in the literature. As summaries of these types of studies are incorporated into commercial databases designed to provide medical practitioners with actionable information, there is a risk that the algorithms that drive the databases may unintentionally incorporate racist biases (O'Neil, 2016) in search reports that use race and ethnicity as query terms to identify findings to help in the diagnosis and treatment of particular patients. As a first step to unpacking this risk, we conducted a content analysis of the records and related citation trails in DynaMed's Point of Care (PoC) tool that refer to racial and ethnic research findings. Our analysis demonstrates that DynaMed does not control for how meanings of race and ethnicity are constructed in its entries, does not always accurately represent the nuanced and contingent nature of the findings about race/ethnicity that it cites, and relies on sources that are not always consistent with the 'evidence-based' criterion that the company self-promotes as a feature of its PoC tool. We conclude that, by failing to acknowledge the complex and contradictory ways that race and ethnicity may, or may not, correlate with the risk of a medical ailment, algorithmically-driven tools that use these concepts to establish group risks for medical ailments may unintentionally work to 'resuscitat[e] biological theories of race by modernizing old racial typologies that were based on observations of physical differences with cutting-edge genomic research' (Roberts, 2011: 567).