:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/69560634/race_norming_board_3.0.jpg)
Last month, the NFL pledged to discontinue “race norming.” Since the early 2000s, the league had been using the controversial medical practice — which included the assumption that Black players have lower cognitive function than white players — as part of a dementia test to determine payouts in a brain injury settlement.
Race norming in professional football first received broad public attention in August 2020, when former players Kevin Henry and Najeh Davenport sued the league, alleging that the practice prevented them from getting an award in the $1 billion-plus settlement over the NFL failing to protect players from the chronic risks associated with head injuries. Both players said they would have qualified for awards had they been white and had the dementia exam not relied on an algorithm that presumed they started out with lower cognitive function. Because of this assumption, Black players have to show a steeper cognitive decline, the lawsuit claimed.
While a judge dismissed Henry and Davenport’s lawsuit in March, the NFL announced an end to the practice in June amid the pressure of 50,000 petitions, media scrutiny, and the country’s increased attention to racial inequity.
“We are committed to eliminating race-based norms in the program and more broadly in the neuropsychological community,” an NFL spokesperson said in a statement. “Everyone agrees race-based norms should be replaced, but no off-the-shelf alternative exists, and that’s why these experts are working to solve this decades-old issue.”
As the NFL notes, race norming isn’t just employed in the league’s cognitive exam — it’s quietly used across medical specialties, from pulmonary medicine to cardiology. Over the past five years, with the growth of the Black Lives Matter movement, some medical students and practitioners have condemned race norming as harmful and discriminatory. In seeking to eliminate the practice, they note that its premise stretches back to slavery.
“Is it really likely that the average person of African ancestry is cognitively impaired when compared to the average white person?” David S. Jones, a Harvard historian and medical ethicist, told Vox. “I can’t think of how that could actually be true. And the assumption that it is true just sounds like white supremacist racism to me. We need to subject any claims like this to really strict scrutiny.”
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/22707328/GettyImages_87877899t.jpg)
Discussion about race-based practices in medicine is fraught. While medical communities largely no longer believe that race is rooted in biological differences — rather, race is a social construct — remnants of this outdated belief remain in institutionalized clinical practices. Legitimizing racial difference through medical tools and policies is still dangerous, according to a growing community of doctors and scholars, and sustains the bias and unequal care that Black patients and other patients of color have received for centuries.
How race norming is used in medicine
Race norming — also called “race correction,” “ethnic adjustment,” and ”race adjustment” — has been integrated into medical tools, from kidney stone risk calculators to oncology risk assessment tools, since at least the 1990s, Jones and his co-authors told Vox. Race is usually just one factor used to determine a person’s risk of illness or disease in these tools — factors like sex assigned at birth, severity of pains, and age are often also part of the algorithms.
These calculators — and the implementation of race correction — are intended to individualize risk so that doctors don’t apply a blanket decision to all patients and to remove physician bias when making difficult decisions. The calculators are meant to help guide medical treatment for high-risk patients who score above or below a certain threshold.
While these calculators have certainly been helpful in detecting some patients’ risk for disease and medical complications, some doctors say the race correction element can be problematic: Working on the racist assumption that Black and white bodies are different has led to algorithms that can harm Black patients and other patients of color.
Last year, Jones and residents Darshali Vyas and Leo G. Eisenstein identified a partial list of 13 tools across a variety of medical specialties (cardiology, nephrology, obstetrics, urology, oncology, endocrinology, and pulmonology) that use race correction. They concluded that race correction perpetuates race-based inequity, as it relies on stereotypes about people of color, often Black people, and can prevent these patients from receiving adequate care.
“By embedding race into the basic data and decisions of health care, these algorithms propagate race-based medicine ... in ways that may direct more attention or resources to white patients than to members of racial and ethnic minorities,” they wrote in the study.
For example, the STONE score helps doctors determine the likelihood of kidney stones in a patient who shows up to the emergency room with flank pain. Doctors enter five variables into the digital calculator (sex, duration of pain, whether the patient is experiencing nausea or vomiting, whether there’s blood in their urine, and race/origin) to predict whether a patient has kidney stones. If a doctor identifies a patient as non-Black, the calculator will add 3 points out of a possible 13 points total. Patients who receive a “high” score (10-13) on the scale have a 90 percent probability of kidney stones, according to the calculator.
After testing the calculator with 491 randomly selected patients, its creator concluded in a 2014 study that the score “reliably predicts” the presence of kidney stones. But in his research, he did not explain why race is incorporated as an element of the calculator. And assigning a lower score to Black patients could influence practitioners to avoid a “thorough evaluation for kidney stones in Black patients,” Vyas and her team wrote.
Other examples abound. For measuring kidney function, doctors estimate a patient’s glomerular filtration rate, or eGFR, to determine how much blood passes through filters in the kidneys. The equation is based on research from a 1999 paper that suggested that Black people have better kidney function than white people because Black people are supposedly more muscular, a notion that isn’t evidence-based and was used to justify slavery.
And the eGFR has had an impact on the treatment of Black kidney patients. In a recent article for Slate, emergency medicine physician Jennifer Tsai described the case of Jordan Crowley, a biracial patient with one Black grandparent and three white ones, who would be in line for a kidney if his doctors identified him as white in the eGFR equation. Because his doctor identifies him as Black, Crowley is waiting for his kidney function to decline even more so that he can be placed on the transplant list.
Tsai explained how a nonracially adjusted kidney function estimate could help Black patients get registered on the kidney transplant waiting list, reversing the grave statistic that nine of the 15 people who die every day waiting are Black. If race correction weren’t used in the eGFR, an October 2020 study found, close to a third of Black patients would have been reclassified as having a more severe kidney condition, prompting advanced care.
Then there is a tool called the Vaginal Birth After Cesarean (VBAC) calculator. For 14 years, many in the medical field used the VBAC calculator to predict the amount of risk a person would have with a vaginal birth after having a C-section in their previous pregnancy. The algorithm predicts higher risks for anyone who identifies as Black or Hispanic, yet the study used to develop the VBAC calculator also found that marital status and insurance type were correlates for VBAC success but weren’t included in the algorithm. The calculator, Vyas noted, could unnecessarily and disproportionately steer Black and Hispanic women away from a vaginal birth, which is usually less risky than a C-section.
“This dynamic is particularly troubling because Black people already have higher rates of maternal mortality,” Vyas and her co-authors wrote. After advocacy from Vyas and others in the field, race correction was removed from the calculator last month.
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/22707557/GettyImages_615301334t.jpg)
While there is no comprehensive study on the reach and impact of race correction — the medical field is just beginning to reckon with these calculations — scholars who spoke to Vox say the algorithms can affect everything from the type of medication patients receive to the type of access they have to specialists to the type of medical insurance coverage and clinical trials they qualify for. Some forms of race correction may be unconscious: Black patients are systemically undertreated for pain, and research has linked this trend to provider bias about pain perception.
“The motivation behind these calculators is not bad,” Vyas said, “but we get into a harmful situation when our default is to use race as a predictive tool instead of viewing it as an element that speaks to existing disparities.”
The role race has historically played in medicine
The key question — and tension — underlying race correction is whether humans of all races are fundamentally similar or different, Jones told Vox.
Jones, like many scholars and scientists, argues that humans are more similar than they are different — that all walking Homo sapiens came from a pool of ancestors who lived in East Africa and that humans are 99.9 percent identical in their genetic makeup.
“The argument I would make is that differences certainly exist but are relevant in very rare and specific circumstances,” Jones said. “We’re much better off most of the time assuming similarity.”
Race norming reflects the idea that race translates into specific biological differences. But race is a social construct that has been used to rank and sort people since as early as the 15th and 16th centuries. It should not be conflated with geography, economic stability, and other social factors that influence health across different communities — which may themselves reflect racism in society.
For example, medical students were taught for decades that sickle cell anemia was a Black genetic disease. Sickle cell anemia is genetic in the sense that it is a mutation that spread across certain populations as a selective advantage where malaria was common — but it’s less a marker of race than of geography. That’s why the trait is also carried by people of all races in Yemen, India, Greece, Italy, and other countries affected by malaria.
“Just because the risk of being a carrier is higher in people of recent African ancestry, it’s not zero for people who self-identify as white,” said Jones, who noted everyone should be screened for the disease.
This is one example of why racial categories are not good proxies for genetic or biological differences. “A patient in front of me who would go into the tool as Black ... could be someone equally from Haiti, from sub-Saharan Africa, or born and raised in the United States with mixed-race parents who present as Black,” Vyas said. “Using race as a stand-in for genes can create a lot of problems because the categories do not map with each other. There’s actually more variation at the genetic level within a race than between two given races.”
In fact, much of the racial differences that medical algorithm creators have found are likely a result of the stress and physiological consequences of racism, not of race itself, Vyas points out. “When we look at a data set and see that people of color have these different outcomes by race, the assumption there shouldn’t be that there is something intrinsically different there due to their race,” Vyas said. “Instead, we should really do that extra step of understanding why those disparities exist and figuring out how to combat the effects of racism on our patients. The experience of racism and the experience of being Black in America has consequences on people’s health.”
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/22707628/GettyImages_657566488t.jpg)
For example, poor living conditions or lack of access to clean drinking water or adequate health care can cause “weathering” or a high allostatic load, a measure for how stress contributes to disease, that leads to shorter life spans and the earlier onset of illness. This is a matter of racism impacting Black Americans’ health, not race. Such factors are sometimes called social determinants of health, and in a society that reflects structural racism, they will harm some communities more than others.
While even proponents of the algorithms acknowledge that race norming is problematic because race is a social construct, they warn that removing the practice without a better replacement can lead to the overdiagnosis of illnesses in Black patients in some instances. In the case of the eGFR kidney function estimate, for example, some researchers have argued that removing the race adjustment could lead to unintended consequences — like fewer Black people being able to donate their kidneys (since their kidneys might be deemed unfit for a transplant) and fewer Black people receiving certain drugs (if their kidneys are deemed unhealthy, Black patients might not be able to get particular antibiotics, for example).
Vyas says that more research may be necessary to fix these holes, but the longer it takes, the more harm is done. “A lot of times people are asking for tons of evidence to show that it’s okay to take out race correction. But in a lot of cases, that level of evidence was not required to put it in to begin with,” Vyas said.
While race norming is relatively modern, the United States has long been obsessed with assuming difference between races. Ideas about differences between Indigenous people and Europeans shaped the medical field; Europeans believed that Native Americans were racially inferior because they lacked immunity to measles and smallpox, diseases brought over by Europeans. Meanwhile, ideas that were propagated to sustain the enslavement of Africans — that they were mentally inferior, had thicker skin, and didn’t feel pain — inform race correction in medicine today, Jones said.
In a 2019 article that scrutinized the VBAC calculator, Vyas argued that the inclusion of race as a variable in the equation is based on ideas that connect to a “long history of racialized science.” An older study examining racial disparities in VBAC success had cited “ethnic variation in pelvic architecture” as a reason white women fare better with vaginal births. But as Vyas and her co-authors explain, these claims have “historically racist antecedents” and can be traced back to 1933 when scientists classified four different types of pelvic shapes and decided that the “gynecoid” pelvis, mostly found in white women, was ideally suited for childbirth. Conversely, the “anthropoid” pelvis, more common in nonwhite women, was first described in 1886 as a “degraded or animalized arrangement seen in lower races.”
Ideas about racial differences in lungs, meanwhile, can be traced to the work of Thomas Jefferson and the 19th-century physician Samuel Cartwright, according to Lundy Braun, a professor and author of the book Breathing Race Into the Machine. In 1787, Jefferson wrote in Notes on the State of Virginia that there was “a different structure in the pulmonary apparatus” between Black people and white people, laying the foundation for lung function studies and development of the spirometer, a tool that aids in the diagnosis of medical disease. Braun says Cartwright then took these ideas and created an empirical framework around them that doctors implemented at the time.
As for the NFL, the study at the heart of the Heaton norms, which the league used to conduct its dementia exams, is more recent — developed in 1991 and updated in 2004 — but not without problems. To determine how socioeconomic factors affect a person’s health, Dr. Robert Heaton used a sampling of non-Black people from across the country. But almost all the data on Black people was collected in San Diego, where the Black population is small and not representative of the diversity of Black people across the country, according to a document from an appeals advisory panel reviewer in the lawsuit.
“It is not commonly accepted practice to use norms that have not been validated and rigorously evaluated to be appropriate in the population in which they will be used,” Jennifer Manly, a professor of neuropsychology at Columbia University, wrote in a brief for plaintiffs suing the NFL. “It is not commonly accepted practice to assume that normative data is adequate for the individual being tested, especially if the norms are 25 years outdated [and] are regionally specific.”
These foundational medical ideas around race and the race norming that they birthed centuries later do not all stem from pseudoscience, Braun told Vox, but are what we accept as “normal” science. “It’s about a framework that was built from white supremacy and rarely got questioned,” she said.
The movement against race norming is growing
For the past five years, there has been a quiet movement to eradicate race correction. It began in part, according to Jones, with medical students in the fall of 2015 arriving at universities, energized about the Black Lives Matter movement after Freddie Gray died in Baltimore police custody.
“Medical students nationwide showed up and were disappointed by how medical schools teach about race and that medical school curriculums change slowly,” Jones said. “They were surprised that medical school had not kept up with the cutting edge of anti-racism, which wasn’t even the term used at the time.”
In those years, students called out the use of race in pulmonary function tests and kidney tests and challenged particular hospitals on their use of various calculators. In 2017, Beth Israel Deaconess Medical Center in Boston dropped the race variable in the eGFR kidney function calculator after a petition from medical students. In the years after, other institutions — including Zuckerberg San Francisco General, Mass General, University of Washington Medical Center, Brigham and Women’s Hospital, and Vanderbilt University Medical Center— followed.
The movement against race correction remains a key focus of organizations like White Coats for Black Lives, which has hundreds of chapters across the country.
“It’s taken six years, but now we have these gratifying examples of advocacy started by students coming to fruition,” Jones said. “My prediction is that five years from now, race correction will probably be rare. And if it does occur anywhere, it’ll be in very precise situations with a better evidence base than there is now.”
First and foremost, I want to again acknowledge that it is the many black students past and present at UW who pushed tirelessly for change in our curriculum and medical practices at great personal sacrifice that made this possible.
— Naomi Tweyo Nkinsi (@NNkinsi) May 26, 2020
So how did we make this change? We made it by questioning lecturers when the MDRD was taught. We did it by not letting the issue go and continuing to push discussions in class about the use of this equation.
— Naomi Tweyo Nkinsi (@NNkinsi) May 26, 2020
Some in Congress have taken notice. In September, the House Ways and Means Committee urged medical societies to rethink race correction, as organizations like the National Kidney Foundation and the American Society of Nephrology formed a task force to examine the use of race in diagnosing kidney disease. Sens. Elizabeth Warren (D-MA), Ron Wyden (D-OR), and Cory Booker (D-NJ) and Rep. Barbara Lee (D-CA) sent a letter to the Agency for Healthcare Research and Quality to prompt a review of race-based clinical algorithms.
According to Vyas, it will be easier to end certain instances of race correction than others. The recommendation to remove race correction in the eGFR, for example, will still need to be implemented on a hospital-by-hospital basis and requires sustained advocacy to actually implement since it is lab-based. But online calculators like the VBAC can be altered more readily.
“We don’t necessarily need to wait for tons of studies to prove that it’s safe to take out risk reduction. If there are questions to be answered clinically, we need to answer those questions. I am hopeful that this is a precedent that we can use as we weed out other examples,” Vyas said.
For Braun, eliminating race norming is just one step to addressing the role race plays in medicine and America’s institutions; it’s just as vital to understand why the practice was created.
“After the murder of George Floyd, every institution in the country said they’re going to embrace anti-racism, but I don’t even know what they think that is,” Braun said. “There will be tweaks when it comes to race norming, but to change life for masses of people who are oppressed is going to take more than tweaks, and that’s what we have to work for.”
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/22707707/GettyImages_1218030454t.jpg)
