Winter 2021

Field Correction

Race-based medicine, deeply embedded in clinical decision making, is being scrutinized and challenged

Racism in Medicine Issue

  • by Stephanie Dutchen
  • 25 minute read

A young Black man arrives in the emergency room, doubled over in pain from a sickle cell crisis. “It’s an act,” says the attending physician dismissively. “I think he just wants drugs.” The attending refuses to prescribe the opioids he might give to a white patient in similar straits.

Andrea Reid, MD ’88, associate dean for student and multicultural affairs for the Program in Medical Education and director of the Office of Recruitment and Multicultural Affairs at Harvard Medical School, witnessed too many such scenes as a trainee in Boston-area hospitals in the 1980s and ’90s.

“It was awful,” she says. “There was bias that reflected in the management of some patients, especially those who didn’t look like they were in pain.” After watching this scenario play out in the emergency department and on the wards, Reid quietly began to direct some of the sickle cell patients toward her outpatient clinic for continuity care.

Studies confirm what Reid and countless other people of color in the United States have known for decades: Black and brown patients are systematically undertreated for pain. When treating pain from broken bones to appendicitis, clinicians—often white clinicians—give darker-skinned patients, including children, lower doses of analgesics than they do white patients, less potent medicines, or nothing at all.

Assuming that a Black or Latino man in pain is a drug user represents race-based discrimination that might be rectified through anti-bias training. The disparity in pain management, however, is also driven by biases that are more insidious because they appear to be based in science.

Many clinicians have heard or been formally taught that Black people don’t feel pain as acutely as white people because they have different biology. Black bodies have fewer nerve endings than white bodies, they’ve been told. Black skin is thicker than white skin, they’ve learned. Digging deeper reveals that these notions, as old as transatlantic slavery, have no evidence behind them. Yet a 2016 survey in PNAS of white medical students and residents found that half of the respondents still believe and act on them.

Andrea Reid
Andrea Reid

“These are not remote, historical issues around race-based treatment recommendations,” says Reid. “These are contemporary issues, with disturbing articles coming out every year.”

Troubled waters

For centuries, race-based medicine in the United States has aimed to identify biological differences between racial groups that could then be used to tailor health care to members of those groups. This has been done at times in bad faith, particularly in past eras, and at other times with good intentions, reflecting earnest attempts to provide better care to historically underserved populations. A growing number of patients, researchers, and clinicians, however, argue that regardless of intent, race-based medicine is simply another form of bias masquerading as evidence-based practice. Scientific meaning can’t be derived from race when race is widely acknowledged as a social construct with definitions that vary around the world, they say, and the faulty endeavor too often results in substandard care for patients of color.

“Race medicine is bad medicine,” summarized Harvard Law School alumna Dorothy Roberts, a professor of law, sociology, and civil rights at the University of Pennsylvania, in a 2015 TEDMED Talk. “It’s poor science, and it’s a false interpretation of humanity.”

The practice ranges far beyond pain management. Race has become baked in to the way U.S. clinicians are supposed to assess disease risk, make diagnoses, plan treatments, and gauge outcomes. Tools that attempt to correct for race span at least eight clinical specialties, according to a 2020 roundup in the New England Journal of Medicine. On a scale doctors use to determine whether flank pain is likely the result of the presence of a kidney stone, being any race but Black counts as much as having blood in the urine does. Patients identified as American Indian/Alaska Native, Asian American, Black, Hispanic or Latinx, or Native Hawaiian/Pacific Islander are given higher risk values than white patients when assessing the safety of undergoing the most common heart surgeries, on top of concrete risk factors like age and obesity. Race-based “corrections” also seep into electronic health records and machine learning algorithms, where adjustments and the rationales behind them are not always apparent to care providers or patients.

The results of these corrections affect the full spectrum of care, from what medicines and dosages are selected to which patients are granted access to specialists, organ transplants, clinical trials, and insurance coverage.

“Any institution should be looking at race-based medicine as an important area of bias. Many medical schools are asking: How did it get into the curricu­lum, and how do we get it out?”

For the moment, it’s largely unclear which race-based medical practices alleviate health disparities and which perpetuate or worsen them. The reasoning can be circular: Black study participants have in the past displayed poorer lung function than white patients, therefore lung function must be naturally lower in Black people, therefore standards related to lung health should be lowered for Black patients and treatment recommendations made accordingly. Those familiar with the history of racial discrimination suspect, and indeed have begun to demonstrate, that patients of color end up being underserved—a double misfortune in cases where race-based medicine is meant to improve quality of care. The lack of clarity can put standards of care at odds with the professional and personal values of clinicians, trainees, and students, especially those of color.

Darshali Vyas, MD ’19, an HMS clinical fellow in medicine at Massachusetts General Hospital, has published several commentaries on the flaws of race-based decision calculators, including as first author of the 2020 NEJM roundup. The impetus to investigate arose from “a tension many of us feel between how we learned about race in medical school as a social construct and how it is used clinically,” she wrote on Twitter in June 2020.

Questions about the legitimacy of race-based medicine have cropped up for decades from both within and outside the medical profession. That skepticism has gained steam in the past few years as movements such as Black Lives Matter and Missing and Murdered Indigenous Women have drawn the nation’s attention to racial inequities in killings involving police and beyond. Demand is growing to see the data that justify divergent medical practices, old and new.

David Jones
David Jones

As sleuths scratch the surface, they’re discovering that evidence is often missing, mixed up with other variables, or reliant on racist premises and stereotypes—findings that deeply discomfit some and come as no surprise to others.

“As historians, we trace the footnotes back to see where these practices come from,” says David Jones, MD ’01, the A. Bernard Ackerman Professor of the Culture of Medicine at Harvard and HMS and co-author with Vyas of the NEJM paper. “Sometimes the footnote trail leads back to nineteenth-century scientific racism.”

“We need to figure out the ways in which our longstanding history of racism within medicine has obscured our modern-day understanding of race and the role it plays in health and health care,” says Troy Amen, a fifth-year MD-MBA student at HMS. “Unless you admit that you may have biases, there’s no way to move forward and fix these historical artifacts that pervade the way we teach our students and provide less adequate care to patients.”

The key, experts say, is to first acknowledge there’s a problem. Then more people—clinicians, researchers, social scientists, students, medical educators, institutional leaders, and funding bodies—can join efforts to solve it.

Thoroughly assessing race-based medicine entails reaching consensus on what is meant by race and elucidating the factors that race stands for as a rough proxy. It requires research that sorts biological from social contributors to health and determines how many observed differences in health and disease represent natural traits that track to some extent with what we call race and how many reflect the consequences of living as a particular perceived race in the United States. It calls for raising clinician awareness and encouraging inquiries into where and why race-based clinical decisions are made. And it necessitates changes in medical education.

“Any thoughtful institution at this time should be looking at race-based medicine as an important area of bias,” says Reid. “Many medical schools, including HMS, are asking: How did it get into the curriculum, and how do we get it out?”

False premises

Scholars have documented how modern conceptions of race and racial hierarchies arose alongside the Western enslavement of people from Africa. People invested in white supremacy at the time were known to conveniently discover or outright invent biological differences between newly defined races to justify inhumane treatment. Some of those theories remain embedded in medicine.

Spirometry is an oft-cited example. Diagnosing or monitoring the status of lung conditions such as asthma and chronic obstructive pulmonary disease commonly involves this test, in which a machine measures the force and volume of a patient’s exhalations, calculates the lung capacity, and determines whether it’s within normal range. The ranges considered normal are adjusted downward for shorter and older people and women, who’ve been shown to have lower lung capacity than taller and younger people and men of comparable health. Ranges also are lowered for Black, Hispanic, and Asian people.

That’s where things get dicey. Lundy Braun, a professor of pathology and laboratory medicine and of Africana studies at Brown University, traced the race correction in spirometry to eighteenth- and nineteenth-century claims that Black people had less lung capacity than white people and that this deficit would benefit from physical labor. The most influential “study,” by pro-slavery physician Samuel Cartwright, did not account for the effects of enslavement on Black people’s observed lung health. A contemporaneous study finding no difference in lung capacity between Black and white Union Army soldiers failed to gain traction in the medical field.

Recent studies have linked diminished lung capacity with poverty and exposure to pollution, both of which disproportionately occur among people of color because of structural racism. The combination of ingrained ways of thinking and the possible confusing of innate biological differences with disparities in environment, income, education, and access to good food and health care means that today, use of spirometry may be writing off poorer lung health in marginalized populations as normal and contributing to underdiagnosis and insufficient treatment.

“By using these lung function algorithms, are we blinding ourselves (and society) to the health harms of structural racism, effectively normalizing lung damage that Black Americans suffer from dirtier air, dirtier jobs, and substandard medical care?” asked Adam Gaffney, an HMS instructor in medicine at Cambridge Health Alliance, and colleagues in a September 2020 perspective in STAT.

Similarly, Vyas, Jones, and colleagues have written about how slavery-era stereotypes regarding the shape of Black and white women’s pelvises continue to appear in textbooks and to factor into clinical decisions such as whether to recommend attempting vaginal birth after a cesarean delivery (VBAC). Some textbooks and articles refer to “ethnic variation in pelvic architecture”; others note that white women’s pelvises have a so-called gynecoid shape while Black women’s pelvises have a nongynecoid shape associated with more complications during VBAC. Such language goes back decades to claims that gynecoid pelvises are better suited for childbirth than nongynecoid ones. Certain other factors statistically shown to correlate with successful vaginal birth, such as marital and insurance status, are excluded from consideration. Doubts strengthen in light of findings from Sweden and Canada that show that adding or docking points for race does not improve predictions.

“Are pelvises different in Canada?” asks Jones rhetorically.

Investigating and teaching the history of science is crucial, say concerned historians and medical students, Jones and Amen among them. Clinical decisions that include race must incorporate not only medical and scientific evidence but also social and historical context, urge researchers such as Arjun Manrai, an HMS assistant professor of biomedical informatics and of pediatrics at Boston Children’s Hospital. Those looking to investigate differences among races or apply race-based practices must be careful not to directly perpetuate racism or assist purveyors of it, they warn.

“As we move forward in doing research, we have to be extremely vigilant that we don’t hand ammunition to people who will distort factoids and exploit prejudice to do harm,” says Isaac Kohane, the Marion V. Nelson Professor of Biomedical Informatics and head of the Department of Biomedical Informatics in the Blavatnik Institute at HMS.

Until the knot of race, racism, and health gets unraveled, clinicians and professors must be frank with patients and medical students, says Reid.

“When we use race in medicine and medical education, there ought to be a reason, and we need to say what that reason is,” she says. “If you’re discussing race-based differences—whether for incidence, prevalence, risk, symptoms, diagnosis, treatment, or outcomes—talk about the causes of these racial differences. If we don’t know, say that it needs additional research.”

“As we move forward in doing research, we have to be extremely vigilant that we don’t hand ammunition to people who will distort factoids and exploit prejudice to do harm.”

Race-based medicine’s checkered history has made many professionals and patients of color cautious about new claims. In 2005, the FDA approved the first, and to date only, race-specific drug in the United States: BiDil, a combination of two established medicines to treat heart failure. After initially failing to convince regulators that the drug was effective, BiDil’s designers conducted a clinical trial in Black participants only and successfully reapplied with the race-specific data. The hope that BiDil could help close racial gaps in heart health clashed with complaints about the study methods—namely, that the team hadn’t provided an explanation for why a heart failure drug would work differently in Black people nor shown convincingly that BiDil was any more or less effective in other populations—and with Black communities’ well-founded distrust of the medical establishment, as Jones explained in a volume he co-edited about BiDil in 2008. Sales lagged, and the company was sold not long after the drug’s release.

Medicine by proxy

A common assertion holds that race and ethnicity serve as proxies for the true drivers of variations in health. Whether those drivers will prove to be inborn, a result of social determinants, or a mix of both remains hotly debated.

“Find me a person who thinks ‘Hispanic’ is a meaningful biological category,” says Jones.

“It’s easier to say ‘race,’” says Reid, “but let’s name it for what it is: racism and the impact on people who’ve been assigned an arbitrary designation.”

In a 2018 New York Times op-ed, David Reich, a professor of genetics at HMS, wrote, “I am worried that well-meaning people who deny the possibility of substantial biological differences among human populations are digging themselves into an indefensible position, one that will not survive the onslaught of science.”

Until scientists unearth the true drivers and reveal how closely or poorly they correspond with racial groupings, experts question the usefulness of using race as a placeholder.

“Race is a bad proxy,” said Roberts in her TEDMED Talk. “It’s just a distraction.”

Yet excluding race from consideration altogether risks not only failing to notice injustices but also contributing to them.

“Ignoring race is not the forward-looking solution,” says Manrai. “Seemingly race-blind approaches still harbor insidious bias.”

Researchers, including Manrai and Kohane, are working to identify underlying causes of disease so clinicians can ditch racial proxies in favor of providing more precise care, moving from generalized to individualized medicine.

“In many cases, we don’t have good substitutes for self-reported race and ethnicity,” says Kohane. “I’m confident that in the future, with the help of bioinformatics and machine learning, we will understand enough about genetics and physiology to replace these imperfect and often misused labels with objective measurements. They’re a way station as we travel toward something more scientifically robust.”

Troy Amen
Troy Amen

Some argue that what we call race really refers to shared ancestry or geographic origins. Sickle cell trait, for example, which also protects against malaria, is most prevalent among populations from swaths of Africa and the Mediterranean where malaria is endemic. But if race stands in for ancestries or environmental adaptations that leave marks in DNA, then labels like “Black” aren’t useful when Africa contains more genetic diversity than the rest of the world, researchers point out. They say medicine would be better served by asking patients which regions their ancestors hailed from or which ethnic groups they belong to. “It’s certainly possible that a recent immigrant from Ghana, another from Ethiopia, and someone else whose ancestors from West Africa became enslaved, while all considered ‘Black,’ could have medically significant differences between them,” says Jones, “and these would be missed by current practices.”

Better yet, says Manrai, practitioners could skip the guesswork and do the gene sequencing to find out whether a person has a disease-associated variant or mutation. That would require making clinical sequencing technology more accessible and a standard of care. Researchers also must continue to ferret out relevant genes while ensuring that genetic data sets and analyses represent enough people from a variety of backgrounds to reach valid conclusions. Manrai, Kohane, and colleagues reported in a 2016 NEJM study that a disproportionate number of Black people were being misdiagnosed as having hypertrophic cardiomyopathy, a thickening of the heart muscle that can trigger fatal arrhythmias, because gene variants believed to contribute to the disease had not been studied in enough Black participants.

It’s unclear how much value genetics adds when studies have revealed greater genetic diversity within racial categories than between races and when individual variants typically contribute small effects. When researchers uncover genes that do appear to differ in frequency or function from one racial group to another, they should make sure the differences are meaningful and get applied responsibly, experts emphasize.

In a 2006 article in the Harvard Review of Psychiatry, Jones and HMS professor of psychiatry Roy Perlis, MD ’97, pointed to an allele that makes some people metabolize a certain drug faster than others do. The clinical conclusion was that since the allele appears in 2 percent of white people but 8 percent of Black people, “we should treat them differently,” says Jones. “But the finding means 92 percent of Black Americans don’t have that allele.” It would be better to assume that most people aren’t rapid metabolizers until proven differently, regardless of race, they wrote.

Kohane and Manrai use biomedical informatics technologies to discover drivers, genetic or otherwise, that currently seem race-associated and to flag false signals.

“Large data sets allow you to tease apart what’s real and what’s rumor,” says Kohane.

An area of particular interest to Manrai is a common test known as estimated glomerular filtration rate, or eGFR. The test measures circulating creatinine, a metabolic waste product, and uses it to calculate kidney function. More creatinine suggests the kidneys aren’t filtering well—unless the patient is Black, male, or young. In those cases, slightly higher levels are considered normal. Why the race correction? It stems from a statistical analysis that found that Black individuals had greater measured kidney function at the same creatinine levels, age, and sex as non-Black people, explains Manrai. Some proposed that this was due to higher muscle mass—but that common hypothesis lacks evidence. Manrai and colleagues are using machine learning to investigate other ways to accurately estimate eGFR without using race. These approaches simultaneously consider environmental context, blood and urine biomarkers, and specific metabolites, he says.

“Declining to study racism risks sounding like it’s not a scholarly subject or something worth changing.”

In the meantime, facing the possibility that eGFR uses a racist algorithm, a handful of hospitals, including Mass General, Brigham and Women’s Hospital, and Beth Israel Deaconess Medical Center, have dropped race as a correction factor. Not everyone is convinced that’s the best solution. Manrai and colleagues, including third-year MD student James Diao, Herman Taylor, MD ’80, and Neil Powe, MD ’81, attempted to quantify the effects of race adjustment on Black patients in a December 2020 JAMA paper. They found that without race correction, more Black patients would be diagnosed with chronic kidney disease and at more advanced stages of disease, qualify for specialist care and insurance coverage of nutrition therapy and education, and become eligible for kidney transplants. Yet more also would be disqualified from donating kidneys, be prescribed lower doses of important medications, or be considered at higher risk for adverse events from those medicines. Interpreting whether dropping the correction factor serves Black patients better than current practice gets even trickier when, as Jones points out, it’s not clear whether the status quo is correct.

Wrestling with race as a proxy also involves issues of implementation. Instead of asking the patient’s race, clinicians sometimes make assumptions based on appearance. When they do get asked, patients don’t always go into the intricacies of their identities and ancestries. A woman might simply say she considers herself Black when she has a rich history of Black, white, and Indigenous forebears from three continents. Some can’t share the specifics of their ancestry because family histories were erased during the enslavement of Africans or the genocide of Indigenous people. Even if such nuances get articulated, studies haven’t been done to fully assess the meaning of multiracial identities in health, nor of distinctions within defined groups, such as Navajo versus Aleut for someone labeled “American Indian/Alaska Native.” Tools such as eGFR calculators and spirometers aren’t designed to process anything beyond single, broad racial categories, either. In fact, many of the tests that employ race correction simply sort patients into “Black” and “not Black.”

Manrai confronted that fact firsthand when his mother, who’d developed diabetes, underwent an eGFR test to monitor her failing kidneys. An immigrant from India, she didn’t consider herself Black, and she didn’t understand why everyone not Black was assumed to be alike. Research participants of South Asian ancestry had not been well represented in the studies that established the race correction factors. Equally unsure how to get the most accurate results, her nephrologist decided to average the Black and non-Black calculations.

Getting past racial proxies is complicated, but biomedical researchers and practitioners can handle the complexity, and the effort is worth it.

“People say it’s hard,” says Jones. “But we put a man on the moon. Surely we can figure out descriptive epidemiology.”

“Declining to study racism risks sounding like it’s not a scholarly subject or something worth changing,” says Jalen Benson, a second-year MD student at HMS. “I haven’t experienced endometriosis, but that doesn’t mean I can’t learn about it and become an OB/GYN. You put in the work to become an expert.”

Nip in the bud

As a child, Reid gathered from the way doctors treated her family members that it was considered acceptable for Black patients in the United States to have higher blood pressures and higher prevalence of kidney disease than white patients. She scratched her head when she read in her mother’s copies of Psychology Today that there were racial differences in certain psychiatric disorders but no explanations why. She itched to know what drove the gap in life expectancy between Black and white men.

Reid figured answers would come when she enrolled in HMS as a medical student in the 1980s. But lectures and clinical training proved disappointing on that front.

“It’s really hard to speak up when you’re at the bottom of the ladder. There’s fear about academic and professional reper­cussions and whether people will be afraid to work with you.”

Her pathophysiology professors commented, usually in passing, about how “Black patients have more this or that disease, but they rarely explained why this was true,” she recalls. Few mentioned the social determinants of health. Only when she started residency did conversations, mainly instigated by students and trainees, dig for the roots of racial disparities in conditions such as high blood pressure and heart disease.

Now, as co-chair of the Program in Medical Education’s Task Force to Address Racism, Reid is coordinating efforts to reduce racism in medical education at HMS and its affiliated hospitals, including places where racism intersects with race-based medicine. The task force will tackle individual instances of racism—the easier part, Reid says—as well as the culture in which those occurrences are embedded.

“An instance is when one professor makes one comment about a race-based clinical indicator without explaining what race is a proxy for,” she says. “A culture is where that is accepted as an appropriate way to teach.”

Part of the solution involves recruiting and retaining more teaching faculty from groups underrepresented in medicine. Amen, who participates in an HMS initiative to diversify faculty and founded a group called Hope Medical Scholars to widen the student pipeline, thinks such efforts could help solve a problem that struck him when he began his coursework: that the persistent and unexamined linking of race with “undesirable states” such as poverty and intravenous drug use can create harmful pattern recognition in some students and reinforce existing stereotypes in others. In turn, he says, this can perpetuate health disparities and the “othering” of patients of color.

“It’s really tough, especially when the people teaching you this material are older white men,” says Amen. “I wonder if this would play out differently if more professors were people of color. Maybe they’d no longer say, ‘Race is a risk factor,’ but instead, ‘This might be an association.’ Just having someone who is Black or Hispanic teach you that could make the pattern recognition very different.”

Task force members agree that the onus to challenge race-based medicine and to unearth the basis for racial disparities in health can’t continue to fall on faculty, staff, students, and trainees of color, who already must shoulder the burdens of individual and systemic racism on themselves, their colleagues, and their patients. At the same time, Reid and colleagues strive to elevate the concerns of students of color to HMS leadership and to empower students to speak up when they encounter racism in their medical education.

The power structures inherent in the U.S. medical profession can make the latter decision exceptionally fraught.

“It’s really hard to speak up when you’re at the bottom of the ladder,” says Amen. “There’s a lot of fear about academic and professional repercussions and whether people will be afraid to work with you. Oftentimes it’s easier to be silent and move on and hope things get better, either on their own or as you get more power.”

It took three years before Amen felt comfortable enough to publish an essay on the website Medium about a racist incident he experienced during a clinical rotation.

Benson, too, has struggled many times with the conflicting desires to question race-based medical decisions but avoid jeopardizing his career. Once, his cohort was taught in the clinic that they should administer different blood pressure drugs to Black patients than to white patients. Benson says that when he asked why, his attending replied, circularly, that it was the standard of care.

“I’m a second-year medical student,” Benson says. “I’m sure not going to tell my attending they’re wrong or argue with them about the controversy in the field.”

“At Harvard, we feel so ready for change.”

Arguing could result in being characterized as aggressive or combative in faculty evaluations, which would torpedo his chances of becoming a doctor as a Black man, Benson says. So until he climbs enough rungs to feel safe enacting change, he bites his tongue.

That includes memorizing race-based treatment recommendations that he believes can harm the patients he wants to serve in order to pass the U.S. Medical Licensing Examination. “I have to do it so I can become the doctor I want to be and dismantle racism embedded in this system,” he says.

People doing the work hope to tap into the anti-racist energy that surged in the United States in summer 2020 following the killings of Breonna Taylor and George Floyd. Yet broader interest in racial justice may already be waning, as a recent Pew Research Center survey has suggested.

“At Harvard, we feel so ready for change,” says Amen. “Unfortunately, the momentum has died down. I worry that in six months when we look back, we’ll have missed a beautiful opportunity to impact Black student lives on campus.”

Change must be embraced by those in power if it is to last rather than return to the norm, “which is to avoid talking about racism in medicine except in closed circles,” says Reid.

When the student group White Coats for Black Lives convened HMS community members in June for 8 minutes and 46 seconds of silent kneeling to honor George Floyd, Reid was less impressed by the show of solidarity than concerned that action would stop there.

“The pictures of people kneeling were lovely,” she says, “but what I really care about is what changes when you stand up.”

Stephanie Dutchen is a science writer in the HMS Office of Communications and External Relations.

Illustration and animation: Jori Bolton. Images: Kelly Davidson (Reid and Amen); John Soares (Jones)