SDOH May be Better Predictor of Health Outcomes than Race

June 15, 2022

Race has long been used as a predictor of health outcomes, but a recent panel of medical bias experts at the Philadelphia Alliance for Capital and Technologies argued that doing so may be hurting members of marginalized communities. Instead, all of the social determinants of health that intersect with race should be considered holistically. The panel, including Seun Ross of Independence Blue Cross, noted that race is a generalizing social construct and that historical medical data is riddled with race-related bias. As such, datasets that distill all the complex factors surrounding race into a single variable actually train systemic bias into algorithms.

According to Katie Adams “When this biased data is used to train healthcare algorithms, the healthcare industry’s structural racism is perpetuated. To illustrate this point, Ross brought up an example of racist bias that has been historically present in the U.S. healthcare system: when Black and White patients present the same behavioral health symptoms, Black patients are more likely to be diagnosed with schizophrenia and White patients are more likely to be diagnosed with mood disorders like depression and anxiety. If engineers use historical health records of Black patients being treated for schizophrenia in the U.S. to train a machine learning program, it would arguably use racist predictors for its calculations and continue to over-diagnose schizophrenia among Black patients, Ross said.”

To read more, click here.