It was created with the best intentions: a government policy asking researchers to collect racial data to help address health disparities. But it may have had an unintended opposite effect, paving the way for the problematic use of race in an array of medical decision-making tools.

Starting in the 1990s, the National Institutes of Health began requiring the collection and reporting of racial data in its funded research. It was a major pivot point, revealing in stark, undeniable numbers racial divides in health outcomes. But to a new generation of researchers, that quantification enabled the development of algorithms and other medical decision-making tools that misused race as a health risk factor.

Poorly understood correlations between race and outcomes were embraced in medical decision-making tools as a way to make disease-risk calculations more precise, though the race data were actually quite imprecise.

It was a time when many in medicine, new to handling race data, used it in a sloppier way than they would today. Many categorized research subjects as simply being Black, Hispanic, or Asian without thinking about the complex ancestry within those groups and how that could impact medical decision-making tools such as algorithms.

Many also still considered race a biological explanation for differences, and not, as scientists agree today, a socially created category — with a weak relationship to genetic differences — that may be more connected to characteristics like income or neighborhood. In some instances, researchers devising new algorithms and other medical decision-making tools uncritically accepted faulty ideas about racial differences that date back to America’s slavery era.

“All we have is old research that was accepted under a lower standard of rigor,” said Lou Hart, medical director of health equity at Yale New Haven Health System. The federal government said, “‘You have to diversify your clinical trials. You have to report out this type of information and publish literature.’ And so people did.”

Eliseo Pérez-Stable, director of the National Institute on Minority Health and Health Disparities, said the collection of racial data “categorically” did not lead to the creation of race-based algorithms and other medical decision-making tools. “The reason we collect race and ethnicity, and we should collect other things like socioeconomic status that we don’t, is because they influence health outcomes in ways that we don’t fully understand,” he said.

Read the full article about medical decision-making tools by Usha Lee McFarling and Katie Palmer at STAT News.