Race and ethnicity are applied in inappropriate and even harmful ways in biomedical research, the National Academies of Science, Engineering, and Medicine said in a report issued Wednesday, calling on scientists, research funders, and publishers to transform the way they use — and don’t use — the categories in research.
“The current use of race and ethnicity in research is sometimes seen as an exercise in checking boxes or a matter of using the ‘right’ labels rather than understanding why they are used,” the report reads. Improving their use in biomedical research will require “intentionality at every step of the research process and understanding the nuances involved in doing so.”
Race and ethnicity is a fraught topic, said report committee chair M. Roy Wilson, president emeritus of Wayne State University. “It’s personal, it differs by time and by country, and there is no uniform, standard definition that has stood the test of time,” he said in an interview with STAT.
Since the nation’s racial uprising in 2020, discussions of how and why race is used have roiled medical science, with many saying research often misrepresents racial groups as differing biologically when they do not. Such mistaken notions about biology in medicine, for example believing Black people have thicker skin and don’t feel pain, have led to systematic discrimination and generations of harm caused by poorer medical care, experts say.
Use of race, and the implications and harm of the way it’s used, has long been discussed within social science, but those discussions have been more recent — and sometimes painful — for those in medicine and biomedical research.
The 271-page report, sponsored by the Doris Duke Foundation and the Burroughs Wellcome Fund and assembled by a committee over the course of a year, leans into the complexity and nuance of the topic, issuing nine detailed recommendations to the biomedical research community in order to make ethical, contextual, and scientifically sound decisions about how and whether to use race and ethnicity.
The National Institutes of Health requires that its funded research reports racial and ethnic categories established by the Office of Management and Budget in 1997 — the same categories used in the U.S. Census — to understand and encourage the diversity of study participants. Categories such as Black, Asian, or Hispanic are meant to be used as social attributes, not scientific or biological ones, those standards state. But over time, “their use in research contributes to their conflation with biological meaning and importance,” the report reads.
As a result, clinical research often inappropriately uses racial and ethnic categories as proxies for factors like genetic variants and environmental exposures. A year and a half ago, another NASEM report, commissioned by the NIH, called for an overhaul in the way genetics researchers use those same racial and ethnic labels, saying they perpetuate “a pervasive misconception that humans can be grouped into discrete, innate biological categories.” When those concepts make their way to clinical algorithms — tools used by doctors every day to make decisions about their patients’ care — they can result in harm.
In one of its recommendations, the new report aims to encourage research that uncovers the underlying factors that drive correlations between race, ethnicity, and the incidence of certain diseases and health outcomes, such as income or exposure to pollutants. Too often, said Wilson, those questions go uninterrogated. “There’s a smattering of it that’s starting to happen, but not nearly as many studies as on biological characteristics in different races. Every journal has dozens of them on a monthly basis,” said Wilson. “But very few talk about other variables that could be more relevant.”
Other recommendations call for researchers to provide operational definitions for race and ethnicity in their studies, rationale for using race and ethnicity in research, and transparency in their methods.
“Requiring researchers to provide a scientific rationale for their use of race and ethnicity is something that is specific and new and is not being done,” said Shazia Siddique, a health systems researcher and gastroenterologist at the University of Pennsylvania who led a review by the Agency for Healthcare Research and Quality on the impact of health care algorithms on racial and ethnic disparities. That recommendation in particular could have an impact on patients, “because it has so many downstream consequences to how we then implement that research,” she said, including in clinical guidelines and decision-making tools.
The report addressed those algorithms and said simply removing race would not resolve issues, but recommended that new tools be developed with more transparency and have their performance reported across different racial and ethnic groups. It noted that such calculators were based on biomedical research that often had deeply rooted bias.
Developers of those tools often “start to point fingers downstream at the people who are collecting data, and then the researchers who are publishing the data,” said Siddique. “Ultimately, I think it takes all of us to make change.”
Still, none of the report’s recommendations mean that race and ethnicity data should go uncollected, the report emphasizes.
“Using race and ethnicity can be appropriate, but it’s context dependent,” said committee member Ruqaiijah Yearby, who teaches health law at Moritz College of Law at The Ohio State University. “If researchers are going to use race and ethnicity, they should be clear about why, and if they are using it to study health disparities, they should think about including other causes of that,” such as racism or social determinants of health, she said.
Identifying the usefulness of race and ethnicity will require careful, case-by-case analysis from researchers. “In order to better serve society, our biomedical research methods must better represent what race means in people’s lives,” wrote Wilson in a preface to the report.
To do so, one recommendation focuses on how to include multiracial and multiethnic people, who have often been left out of research because they did not fit cleanly into established categories. “How do you characterize me? I don’t know,” said Wilson. “I’ve always identified as Black, but my mother is Japanese.”
OMB recently released revisions to its race and ethnicity categories, adding the category Middle Eastern or North African and encouraging multiple choice selection; agencies including the NIH will submit plans to incorporate those changes by September 28, 2025 — a move that will influence how race and ethnicity is collected in future research. But Wilson emphasized that mandated race and ethnicity collection to meet recruitment goals should be considered separately from the categories collected for analytic purposes, which should be “as granular as the statistics will allow you to be.” Many categories are seen as problematic; for example, the lumping of Asians — nearly 60% of the world’s population — into one category has served to mask health disparities within some smaller groups.
Another recommendation encourages researchers to partner more with the communities they study. Reaching out earlier in the process, said Yearby, would improve research by helping refine questions and theories. “If you leave it until the end, until recruitment, you miss a lot of issues that could help with your hypothesis,” she said.
The responsibility of using race and ethnicity in better ways lies not only with researchers, but the entire biomedical ecosystem, said Yearby. Journal editors and research funders need to play a part by giving researchers consistent guidelines on the best use of race and ethnicity, and then enforcing those guidelines, the report recommended. “That’s an area that really hasn’t been fleshed out very well,” said Siddique. A lot of people say that journals and funders like the NIH should be doing this, but what we should be doing is not totally clear.”
One concrete recommendation from the report encourages funders to extend research timelines to allow scientists to involve the community in their research. Many community-based researchers have said they have struggled to receive grants because of the longer time frame their research requires and said their work was often rejected by leading medical journals as not being rigorous enough. “This report says community-based participatory research is not separate from the biomedical research that is happening,” said Yearby. “You should work with the community you are studying.”
Researchers need resources to be able to accomplish all of this, noted Wilson — and the committee was mindful of placing the burden solely on the scientists. “There’s responsibility on the part of sponsors and funders of research to appropriately fund biomedical research that looks deeper in the issues,” said Wilson, “and not just do whatever is most convenient.”