Editor’s Note: Linking Social Determinants, Research, and Racism
By now, most health care professionals, from social workers to health information management professionals, are familiar with the term social determinants of health, which encompasses the conditions in which people are born, grow, live, work, and age. To a degree, these factors can also be linked to racism.
We know there are health inequities. Multiple studies have documented instances where disparities in medical treatment were directly tied to ignorance, which has led to differences in disease prevalence and health outcomes. There’s clearly a problem with health inequities, but when it comes to better understanding how the body operates, can race be totally ignored?
“Racial Bias in Health Care Artificial Intelligence,” an infographic published by the National Institute for Health Care Management (NIHCM) Foundation, sheds lights on racial bias in clinical care outcomes. Algorithms are used to identify patients with complex health needs in order to provide more comprehensive care management. While race and ethnicity are often correlated with outcomes because minority patients routinely have different health outcomes from white patients, not accounting for race in the algorithms employed by artificial intelligence tools can have devastating consequences.
For example, Black patients who were considerably sicker than white patients were given the same risk score because the algorithm assigned those numbers based on past health care spending, a flawed approach based on the fact that Black patients spend less than their white counterparts for a given level of health.
Interestingly, in an NIHCM Foundation–sponsored webinar, David Jones, MD, PhD, pointed out that "race-based medicine often has a defensible logic. Researchers have documented race differences in disease prevalence and therapeutic outcomes. Clinicians, in response, have factored race into diagnostic tests, risk calculators, and treatment guidelines."
The problem comes when many of the race-adjusted tools favor white patients. Nevertheless, it would be foolish to disregard the potential of artificial intelligence. To that end, the NIHCM Foundation suggests that biases in algorithms can be addressed by the following:
• adding more diverse and complete data;
• incorporating equity into the design of algorithms; and
• addressing the lack of inclusion in artificial intelligence.
If we can somehow manage to align the technology with what is clearly happening in patients’ day-to-day lives, it would be a huge first step to creating a health care system that works for everyone.