Jazmin Evans was waiting on dialysis for four years before finally, on the Fourth of July last year, she received a kidney transplant. “Now I say the fireworks are for me,” said Evans, who was diagnosed with kidney disease when she was 17.
She would have been waiting even longer had it not been for a shift, in 2021, in the way that physicians calculate kidney function for Black patients like her.
In a panel Wednesday at the STAT Summit, Evans, who advocates for equity in health care and organ transplant systems, joined two experts in the use of race-based clinical algorithms. They discussed the work that led race to be removed from a commonly-used calculator to estimate patients’ kidney function, and the challenges facing similar efforts to address race-based algorithms in other areas of medicine. The topic was a subject of STAT’s investigative series Embedded Bias, coauthored by moderator Usha Lee McFarling and this reporter.
When Evans found out that the kidney function calculator, called eGFR, had made her kidney function appear better because she was Black, “I was enraged, to say the least,” she said. Because of the race-based estimates, thousands of Black patients were placed further down kidney transplant waitlists — a disparity now being rectified as a new, race-free tool is adopted across the country and patients like Evans are granted back the time they should have accrued on the waitlist.
But patients are rarely aware of the fact that race, a socially-constructed category, is factored into dozens of other calculators that inform clinical decisions. “It was completely ubiquitous,” said David Jones, a professor at Harvard Medical School who teaches the culture of medicine and has researched the role of race in clinical algorithms. When he and his students started looking at those tools five years ago, “it was very easy to look in medical textbooks or in MDCalc and find tool after tool after tool that would ask, is the patient Black or white, and then, like eGFR, would produce a different answer,” he said. “There’s no way to justify that practice.”
Pulmonology is another field that has grappled with race-based clinical tools. Race was initially incorporated into clinical equations to enable more personalized care after decades of clinical research had used white populations as a default. “I was taught, and really believed, that the way that we could do this is by having this greater precision of these racially-specific algorithms and equations,” said pulmonologist Aaron Baugh, assistant professor of medicine at the University of California San Francisco.
But over time, Baugh continued to see Black patients who, despite having the “same” lung function as white patients according to their lung function tests, were doing worse. Eventually, he and other researchers asked, “Are we really sure that they have the same lung function after all? Or is the way that we are interpreting these lung function tests misleading us and giving them a falsely higher suggestion of lung health?” Baugh said. That turned out to be the case — a finding seen widely that contributed to the American Thoracic Society’s recommendation, in 2023, to adopt race-neutral lung function tests.
As clinicians have attempted to strike out the harmful effects of race adjustment in other clinical algorithms, they have faced stiff resistance. Conversations are often stymied by the discomfort surrounding conversations about race and racism. “In order to really deal with issues of race, not just in our health care system, but in our country as a whole, we have to be comfortable with being uncomfortable,” said Evans.
Some clinicians also worry that efforts to remove race from clinical tools could harm the disadvantaged populations that race-based equations were designed to better represent. Collecting medical data from diverse patient populations is still critical, said Jones, both to understand health disparities and combat them. “No one I work with would recommend having a colorblind medical system or colorblind public health data,” he said.
The solution, said Baugh and Jones, is in measuring and acting on the underlying traits that race represents when they are incorporated into clinical algorithms, to deliver truly individualized care — not care that lumps people into buckets based on their race. “Racism, wealth, privilege — all these things are attached to this label of race,” said Baugh. “But why would you measure the label rather than saying this is a big umbrella of different qualities?”
Clinicians and researchers developing algorithms to predict a patient’s risk have a growing arsenal of variables to consider. “You can think about genetics. You can think about the cumulative exposures that we have over our lifetimes. You can think about experiences of racism,” said Jones. Those categories can help inform the development of new, more precise clinical calculators that avoid the potential to introduce disparities by stratifying patients into crude, race-based categories.
But even when new approaches are accepted, it takes time to change minds and the medical system. To implement race-neutral changes in lung function testing, Jones pointed out, each of the country’s 6,000 decentralized hospitals has to reprogram its electronic health record — an expensive and time-consuming bureaucratic endeavor.
“This is going to be hard work,” said Evans. “It’s not just another ‘Check! We’re done.’” As patients wait for change, she counsels them to ask the hard questions of their doctors. “They’re human, too,” she said.