How racism skewed estimates of heart disease in women

Growing up as a multiracial person in the United States, Alexis Reeves was no stranger to the impact of racial discrimination. She spent summer vacations with her father’s family in Pennsylvania, where her Black grandfather had worked as a sharecropper. Nearly everyone on that side of her family had high blood pressure, took cholesterol medication, or had experienced some form of heart disease.

The reason, Reeves later discovered, was a phenomenon called “weathering.” Just as water dripping on stone erodes rock over time, researchers have found the continual stress of racial discrimination wears down the bodies of minority populations .

Weathering doesn’t just hurt people—it hurts the science that could help them. People of color often develop diabetes and heart disease at earlier ages than white populations, excluding them from research studies that track individuals over time.

So in a new study, Reeves decided to rejigger data from a long-running analysis of women’s health: the Study of Women’s Health Across the Nation (SWAN). Since 1994, the U.S. - based project has been tracking the health middle-age and older women as they get older. But it had been leaving people out.

Reeves and colleagues made that discovery when they analyzed a data set of people who had been assessed for their eligibility to participate in SWAN. Between 1995 and 1997, SWAN researchers screened nearly 16,000 women ages 42 to 55 to observe how midlife hormonal changes affected various aspects of their health. Anyone who was likely to undergo the hormonal changes associated with menopause could participate. But individuals who were pregnant, had their ovaries and uterus removed, or those who had already experienced menopause, were not included. Because women of color tend to age earlier because of weathering, they were more likely to have already experienced menopause, and therefore be excluded from the study. The final subset included more than 6500 eligible women—roughly 3300 of whom participated in the study.

For the new analysis, Reeves and her colleagues examined the roughly 9000 women who had been excluded in the previous iteration of the study because these restrictions didn’t account for weathering. Then the researchers effectively “enrolled” these women in SWAN, matching up their data with similar participants who were included in the study in order to see how their health trajectories compared with those of the previous cohort.

The findings were surprising. Including these individuals lowered the average age at which women—regardless of race—experienced heart disease, hypertension, and diabetes by nearly 20 years. Black and Hispanic women who had been excluded because of menopause were also the most likely to already have diabetes or hypertension at the start of the study. Nearly one-quarter of all the newly included women had high blood pressure when the study began. Insulin resistance, a precursor to diabetes, began 11 years earlier, and eheart disease began about 5 years earlier in Black and Hispanic women than the original SWAN indicated, the team reports this week in JAMA Network Open .

Studies such as SWAN often focus on the healthiest participants so they can observe when diseases set in with age. But leaving out those who already have the condition being studied hurts everyone—not just people of color, Reeves says. “I wasn’t expecting to see how much the selection [of study participants] changed these estimates,” Reeves says. “We’re overestimating the timing of onset of these diseases for everyone in SWAN, regardless of race. It basically amounts to only telling part of the story.”

It’s an “eye-opening” finding, says Jennifer Deal, an epidemiologist at Johns Hopkins University who was not involved with the work. This bias in selecting study participants is important and often overlooked, she says.

“It’s a very clever use of statistical methods to try to address these biases in health disparities research,” adds Kim Lind, an epidemiologist who studies chronic diseases at the University of Arizona, also not involved. “The differences in this study are both statistically and clinically significant.”

The new study’s findings are in line with previous work, which has found that people of color experience metabolic diseases at younger ages . A recent analysis of national data reported that, on average, white people are diagnosed with hypertension by age 47. The average age at diagnosis for Black people was 42, and Hispanic individuals 43. The chronic stress of social discrimination , environmental exposures , income inequality , and other aspects of systemic racism are thought to contribute to this disparity.

SWAN’s original investigators had long wondered about how accelerated aging among people of color might have biased who was included in the study, says epidemiologist Carrie Karvonen-Gutierrez, who leads the project’s University of Michigan, Ann Arbor, site and also co-authored the new work. She says she was “shocked” to see that despite SWAN’s broad criteria for including women, the study’s previous results were still strongly biased.

The results could impact how clinicians evaluate metabolic disease based on a person’s age and gender, Lind says. And they’re probably an underestimate, she adds, because the new analysis only compared women with similar education levels.

Deal says she hopes the work won’t just force scientists to re-evaluate old studies, but to think more about the ones they’re currently working on. “The best way to avoid this bias,” she says, “is to think very carefully about how we design our studies.”

Reporting for this story was supported by a fellowship from the Gerontological Society of America, the Journalists Network on Generations, and the Silver Century Foundation.