Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • NEWS

Is a racially-biased algorithm delaying health care for one million Black people?

A woman is treated by two nurses.

A woman receives dialysis, a common treatment for people with kidney disease. Black people in the United States are almost four times as likely to experience kidney failure as are white people.Credit: AB Forces News Collection/Alamy

One million Black adults in the United States might be treated earlier for kidney disease if doctors were to remove a controversial ‘race-based correction factor’ from an algorithm they use to diagnose people and decide whether to administer medication, a comprehensive analysis finds.

Critics of the factor question its medical validity and say it potentially perpetuates racial bias — and that the latest study, published on 2 December in JAMA1, strengthens growing calls to discontinue its use.

“A population that is marginalized and much less likely to have necessary resources and support is the last group we want to put in a situation where they’re going to have delays in diagnosis and treatment,” says nephrologist Keith Norris at the University of California, Los Angeles, who argues for retiring the correction until there’s clear evidence that it’s necessary.

On the flip side, others say that the correction is based on scientific data that can’t be ignored, although they, too, agree that its basis on race is a problem.

A correction arises

Researchers introduced the correction factor2 in the late 1990s to take into account results showing that, on average, Black people in the United States tend to have higher blood levels of a molecule called creatinine than do white people — despite having similar kidney function. Creatinine levels are a marker of how well a person’s kidneys filter waste from the body. Doctors feed the measurement, along with other information, into algorithms that calculate a person’s estimated glomerular filtration rate (eGFR) to evaluate kidney function. High creatinine levels lead to a low eGFR, which is a sign of kidney disease; the correction inserts a multiplier of about 1.2 when calculating the eGFR of Black people, potentially making their kidneys seem healthier than they actually are.

Variations of race-corrected eGFR algorithms are now used in more than 90% of pathology labs across the United States, according to the College of American Pathologists in Northfield, Illinois.

In the past few years, institutions including Beth Israel Deaconess Medical Center in Boston, Massachusetts, have dropped the correction factor. None of them has yet released data on the action’s impact. In August, the American Society of Nephrology in Washington DC and the National Kidney Foundation in New York City convened a task force to evaluate whether the entire medical community should stop using it. The group’s initial recommendations are expected by the end of the month, with a final decision on whether to continue using the correction due in the first half of next year.

Those who want to abolish the correction say it perpetuates the problematic idea that people of different ethnicities have different biology. Also, the eGFR algorithms are just an estimate of kidney function that describe a collection of “noisy data”, says nephrologist Rajnish Mehrotra of the University of Washington in Seattle, one of the institutions that has dropped the correction. The question, he says, is whether you’re willing to perpetuate the false idea that race reflects biological differences for a “small gain in precision” that you might get from using it to evaluate kidney function.

Others worry that simply removing the correction could cause harm. Mathematical analyses such as the one in JAMA, they say, do not represent real-world health outcomes: it’s unclear whether removing the race multiplier would actually help or hurt the health of those one million Black adults. Neil Powe, an internal-medicine specialist at the University of California, San Francisco, and a co-author of the study, points out that removing the factor could lead to over-diagnosis of kidney disease in Black people, causing burdens such as extra medical bills and denying them access to medicines, such as diabetes drugs considered too risky for those with unhealthy kidneys. The correction arose because of creatinine data, he says — and as long as creatinine alone is used as a biomarker to gauge kidney function, researchers can’t just ignore those data, which have been replicated3 for US study participants multiple times.

Assessing a correction

In the United States, kidney disease disproportionately affects Black people. They are currently almost four times as likely as white people to experience kidney failure.

It’s tough to tell whether — or how much — the race-corrected algorithm has worsened this crisis, because the rate of disease is affected by other factors influenced by systemic racism, including socio-economic inequalities and a lack of health insurance, scientists say. “Race corrections to eGFRs are most likely contributing, but it’s not reasonable to expect that removing the race correction will automatically solve all inequities for Black Americans with chronic kidney disease,” says Nwamaka Eneanya, a nephrologist at the University of Pennsylvania in Philadelphia who advocates dropping the correction.

In the latest analysis, researchers including Powe aimed to assess what would happen if they removed the race-based correction factor for a representative group of people. The team examined the medical records of 9,522 Black people included in the National Health and Nutrition Examination Survey, a programme run by the US Centers for Disease Control and Prevention that maintains a national database of health statistics.

Although it was unsurprising that dropping the correction would increase the number of Black people diagnosed with kidney disease, “the size of the effect surprised us”, says Arjun Manrai, a computational health researcher at Harvard Medical School in Boston, who led the study. Removing it would lead to a change in diagnosis for 3.5% of Black adults from ‘disease free’ to having early-stage kidney disease (extended to the US population, this would be one million Black adults). Removing it would also shift the status of 29% of Black patients from having early-stage to advanced disease. Getting rid of the correction, says Manrai, could drastically alter these people’s access to medications for common conditions such as hypertension or diabetes, because the drugs can have side effects on the kidneys. Overall, it would also increase the number of Black people with kidney disease who are eligible to receive a transplant.

Correcting a correction

Andrew Levey, a nephrologist at Tufts University in Boston, is one of the researchers who originally established the correction factor. Although he has come to question whether the multiplier should be used, he isn’t certain that dropping it is a perfect solution. The latest study does show that more Black people could be diagnosed earlier if the correction were removed, he says, but that doesn’t mean they will all benefit. “Some of them will probably be helped by getting into treatment earlier because they’re on their way to getting kidney disease, and some of them won’t be helped,” he adds. “It’s a typical tension we face between diagnosing too early versus too late.”

A better solution, says Levey, might be to develop an algorithm that relies on biomarkers beyond creatinine. On 7 December, he and his colleagues published an eGFR algorithm that has no race-based correction factor and instead uses multiple biomarkers, in addition to creatinine4.

Until such algorithms are vetted for wider clinical use, Levey and others suggest talking to patients about how their race might be used in clinical decisions. He adds: “I don’t think that we have been transparent in speaking with our patients about how we do this.”

Nature 588, 546-547 (2020)



  1. 1.

    Diao, J. A. et al. J. Am. Med. Assoc. (2020).

    Article  Google Scholar 

  2. 2.

    Levey, A. S. et al. Ann. Int. Med. 130, 461–470 (1999).

    PubMed  Article  Google Scholar 

  3. 3.

    Peralta, C. A. et al. Am. J. Nephrol. 31, 202–208 (2010).

    PubMed  Article  Google Scholar 

  4. 4.

    Inker, L. A. et al. Am. J. Kidney Dis. (2020).

    Article  Google Scholar 

Download references


Nature Careers


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing


Quick links