Mammograms have doubled the number of early-stage breast cancer cases detected in the U.S, but have not done much to reduce the number of deaths from advanced breast cancer, according to a new analysis of 30 years of data.

In addition, as many as one-third of all newly diagnosed breast cancer patients in the U.S. are overdiagnosed as a result of mammogram screening. Patients are considered "overdiagnosed" when data suggest they were treated, often at an early stage, for a cancer that would never have harmed them.

Mammograms have contributed to the overdiagnosis of as many as 1.3 million women in the past 30 years, the study shows. Because this finding is based on statistics of early- and late-stage cancers, researchers have no way of identifying which women were overdiagnosed.

"Overdiagnosis of breast cancer is a greater problem in the U.S. than is generally appreciated," said study researcher Dr. Archie Bleyer, of the Oregon Health and Science University. "I was surprised by the magnitude of the problem."

Another recent study from Norway came to a similar conclusion, estimating that 15 to 20 percent of breast cancer cases are overdiagnosed.

There are varying U.S. guidelines for breast cancer screening. The American Cancer Society recommends annual mammograms for women beginning at age 40. The U.S. Preventive Services Task Force recommends mammograms every two years for women between ages 50 and 74. The reason for the 10-year difference is that the USPSTF says there is a lack of evidence that mammograms have more benefits than harms for women in their 40s.

The potential harms of screening include creating unnecessary worry over false-positive results and conducting unneeded biopsies. The harms of overdiagnosis include treating women with drugs or radiation — which have their own side effects — for cancers that would never have affected their health.

In the new study, researchers compiled data from the National Institutes of Health on women 40 and older who underwent mammograms between 1976 and 2008, and compared the numbers to the rate of breast cancer among women younger than 40, a group the researchers assumed were not getting mammograms.

The analysis showed that the rate of early breast cancer in women over 40 doubled between 1976 and 2008, as routine mammogram screening was widely adopted. Meanwhile, the rate of early breast cancer cases increased for women under 40 by a tiny, but statistically significant, 0.25 percent yearly.

The data also showed that advanced breast cancer cases among women over 40 declined only slightly.

"This analysis suggests that the effect of mammography screening on declines in late-stage breast cancer and mortality is modest," said Dr. Rachel Ballard-Barbash, a National Cancer Institute researcher who was not involved in the study.

The study shows that mammograms have had only a small effect in preventing the progression to late-stage disease, Ballard-Barbash said. In other words, the idea that routinely screening women will catch cancers in their early, rather than late and less-treatable stages, was only weakly supported by the study.

The study did show that women who are screened have reduced rates of cancers that have spread regionally, to sites near the breast such as the lymph nodes.

While death rates decreased by 28 percent over the study period, this is likely due to better breast cancer treatments, the researchers said.

"There have been advances in treatment that this analysis suggests have had a more substantial effect [than screening]," said Ballard-Barbash, who noted that the study did not examine actual trends in treatment.

"Overdiagnosis remains an area that deserves discussion and analysis to help in balance the pros and cons of screening — not just for breast but for other cancers as well," said Diana Buist,, of the Group Health Research Institute in Washington.

While the researchers generally made reasonable assumptions, assuming that mammograms don't start until after age 40 for most women may not always be correct and could have affected the results, Buist said. Another gap in the analysis was the exclusion of unstaged cancer cases, many of which are advanced, she added.

"The next step is to better inform women and their care providers of the relative benefits and potential harms of [mammograms] that include but are not limited to overdiagnosis," Bleyer said. "They have a choice in breast cancer screening."

This study is published Thursday (Nov. 22) in the New England Journal of Medicine.