A new drug that promises to selectively treat heart failure in African-Americans is up for approval from the US Food and Drug Administration. But the controversy that surrounds BiDil, which would be the first drug marketed to a specific racial group, illustrates that, among scientists, there is little consensus on the usefulness of race in genetic studies.

How important is race? In medicine, the issues are not black and white, scientists say.

BiDil appears to be ineffective in whites, but in blacks, researchers reported that the drug reduces mortality by 43% (N. Engl. J. Med. 351, 2049–2057; 2004). But the problem with this approach is that it seems to confirm a genetic difference between races where there may be none, says David B. Goldstein, a geneticist at University College London.

Researchers who receive federal funds are required to include people of varying self-reported racial and ethnic groups in their studies. But race or ethnicity is “a blurry proxy” that reflects environmental and cultural, as well as genetic, differences, says Francis Collins, director of the US National Human Genome Research Institute in Bethesda. About 99.9% of the genome is identical from person to person and most variation occurs within racial and ethnic groups, not between them. Nonetheless, some differences in health have been found to align with self-reported race.

Most scientists agree that race—or, perhaps, a sociodemographic group—is often an important surrogate for environmental, cultural and even genetic factors underlying disease predisposition and outcomes. “Race isn't the ending point, it's the starting point,” says Neil Risch, professor of genetics at Stanford University.

For both social and scientific reasons, however, race is a troublesome word, says Shomarka O.Y. Keita, senior research associate at Howard University's National Human Genome Center. Socially, Keita notes, the term is associated with racism; scientifically, it has been used to describe 'subspecies'—a genetic divergence far too broad to describe the minor genetic variations found between people. “Sociodemographic groups of Americans can and should be studied,” Keita says, “but they are not subspecies or isolated breeding populations; they are not races.”

Sociodemographic groups of Americans can and should be studied... but they are not subspecies or isolated breeding populations; they are not races. Shomarka O.Y. Keita, Howard University's National Human Genome Center.

Still, it doesn't make sense to get rid of the term race because, like sex and age, race can provide valuable information, some researchers say. Because scientists often find what they look for, however, it is important to track socioeconomic status, environmental exposures, housing and access to care even in genetic studies.

Alexandra Shields, associate professor at the Health Policy Institute at Georgetown University, urges caution in using race as a variable. “There's a danger in looking for causal explanations from genes when the differences are almost certainly more powerfully related to environmental factors,” says Shields. In the January issue of the American Psychologist, she and her colleagues will argue that although race may be appropriate to use during recruitment, “self-identified race should no longer be used in genetics analyses.” Instead, geneticists should test people to genetically determine the continental ancestry of their grandparents, she says.

Keita also suggests replacing racial-ethnic categories with ancestry and collecting information on where people live, early life history and environment. For instance, environmental exposures that have not yet been identified might appear genetic. Socioeconomic differences between mothers in different groups might also result in different fetal environments, which might affect adult health.

Some research journals request that researchers specify ancestral and environmental information so that the data can be grouped flexibly for medical geneticists, epidemiologists and biological anthropologists. “We can't abandon race because it might be our only clue to a problem,” Collins says. “But it's not enough on its own.”