A woman who has a mammogram that shows she has dense breasts—more tissue than normal—is usually considered to be at a higher risk of breast cancer than other women. That’s because the denser tissue makes it harder for radiologists to detect breast cancer in mammograms. Plus, the extra tissue itself is a risk factor.

But according to a new study published in the journal Annals of Internal Medicine, radiologists read breast scans differently. One doctor might see a scan and flag it as showing breasts with dense tissue, while another doctor might look at the same scan and think it looks normal.

This variation can be a problem for women, who might get conflicting information from different doctors (or the same doctor and different scans from year to year). About half of all states in the U.S. require healthcare providers to warn women if mammogram results indicate dense breasts while recommending possible further testing. The U.S. Food & Drug Administration (FDA) is also considering a similar policy. But depending on who read a patient's mammogram results, the warning may not be as helpful as it was intended to be.

“These legislative and regulatory initiatives have generated controversy because of the large number of women affected and the lack of evidence or consensus in the medical community with regard to appropriate supplemental screening strategies for women with dense breasts,” the study authors wrote.

The study looked at mammogram readings from radiologists who had analyzed at least 500 different mammograms between 2011 and 2013 at 30 different facilities. Ultimately, that included nearly 217,000 mammograms of more than 145,000 women between 40 and 89, interpreted by 83 radiologists.

In total, nearly 37 percent of all these mammograms were marked as being of dense breasts. But not every radiologist had similar rates, and not every women’s scans were judged similarly year to year. Some radiologists marked only about 6 percent of the scans they read as dense, while others saw dense tissue in nearly 85 percent of the mammograms they interpreted. At the top, 25 percent of radiologists rated about 51 percent of patients as having dense breasts, while at the bottom, 25 percent of radiologists rated only about 29 percent of their patients as having dense breasts.

More than 17 percent of women who had annual mammograms read by different radiologists received discordant results. Even when mammograms were read by the same radiologists from year to year, 10 percent of women had discordant readings. Mostly, the difference came when the breasts were considered dense upon first reading and then considered not dense after a second scan.

These disparate readings were found across geographical and other demographic lines.

Ultimately, this finding means that for a warning of breast density to be useful, it could be necessary to create a standardization of what constitutes a dense breast. The study authors argued that the designation of risk is not helpful if it is not the same across variables—certain instances of breast cancer risk could be undervalued and overly cautious errant readings could cause unnecessary worry or extra tests.