Since 1992, scientists have known that black women have a lower chance of breast cancer survival at the time they’re diagnosed. Since 2002, scientists have known that women of color are more likely to die from breast cancer than white women, and that black women are more likely to have more aggressive cases and 33 percent more likely to die compared to white women.

However, the U.S. Preventive Services Task Force (USPSTF) has always had a one-size-fits all recommended age to begin annual screenings for breast cancer (50 years of age since 2016). This number is based on the median age of all breast cancer diagnoses regardless of race. Now, according to a new meta-analysis published in JAMA Surgery, that age is too late for women of color.

A group of researchers from Massachusetts General Hospital and Harvard Medical School examined 37 years of breast cancer cases in 747,763 women between 40 and 73 years old. They found that the median age of black, Hispanic, and Asian women diagnoses with breast cancer tended to be earlier—at ages 56, 55, and 55 respectively—compared to white women, whose median age at diagnosis was 59.

“If people follow the USPTF guidelines and do not screen patients until age 50, then minority patients would be harmed,” Sahael Stapleton, one of the co-authors of the analysis, said. “It has nothing to do with the care delivery. It has to do with our guidelines being biased.”

The study recommends that in order to correct the bias in these guidelines, “screening ages would need to decrease to 47 years for black, 46 years for Hispanic, and 47 years for Asian patients.”

Scientific research has a long history of excluding women and people of color from clinical trials, especially in the case of cancer. In fact, a 2004 study found that people of color and women are less likely to enroll in cancer research than white males, and that the participation rates of people of color was actually declining.

According to Stapleton, addressing the bias against women and people of color has been a longstanding, unaddressed challenge in medical science.

“We as a society have focused a lot of attention on improving cultural competency at the care delivery point. But some of the problems that we are facing may be happening at the scientific research stage,” Stapleton said. “Flawed science hurts more people than flawed care.”

That’s not to say that access to care doesn’t have a significant effect on health outcomes. According to a study published in 1995, although black women were found to be diagnosed at a later stage of cancer than white women, the timing and history of mammograms accounted for less than 10% of that time difference. But we don’t know how many women have been diagnosed unnecessarily late, and how many lives have been lost because of it.