There is no reliable evidence that regular mammograms reduce the risk of dying of breast cancer in women of any age, according to Danish researchers who performed an analysis of all of the major studies that have addressed the issue.
That surprising conclusion by members of the Cochrane Collaboration, a respected international body that assesses medical evidence, contradicts the recommendations of many medical groups in the United States and Canada. There has been consensus among such groups that regular mammograms can prevent some breast cancer deaths in women over 50, and most of the recent debate in this country has centered on the use of the X-ray tests for women in their forties.
The analysis by Ole Olsen and Peter C. Gotzsche of the Nordic Cochrane Center in Copenhagen concluded that of seven large mammography trials involving half a million women, three were of poor scientific quality and two were so flawed that the researchers discounted their results.
"No study did it completely well," Olsen said.
The report, published in Saturday's issue of the Lancet and on the journal's Web site, faults some studies for failing to ensure that women were assigned randomly either to get mammograms or not. (Random assignment ensures that participants in the groups being compared are similar in age and other factors.) It also raises questions about whether researchers in some studies were biased in determining whether deaths of some study participants were caused by breast cancer or by other diseases.
The two most scientifically sound studies, which the authors called medium quality, found no reduction in breast cancer deaths among women who had regular mammograms. In contrast, the three studies considered poor quality reported, on average, a 32 percent reduction in breast cancer mortality. None of the five found that having mammograms reduced overall mortality.
The benefits and risks of mammography have been passionately debated for years. "I am going to guess . . . that [the Danish researchers'] perspective is in the minority," said Barnett S. Kramer, a senior medical scientist in the National Cancer Institute's cancer prevention division. However, "the Cochrane group and the Cochrane method are well respected."
One reason there has been such intense disagreement, researchers said, is that even in favorable studies, the benefit of mammograms in reducing cancer deaths appears modest at best. "That's what makes it so difficult and leaves additional room for passion and difference of opinion," Kramer said.
"As far as screening tests go, Pap tests [for cervical cancer] and flexible sigmoidoscopies for colon cancer . . . are much better than a mammogram," said Karla Kerlikowske of the University of California at San Francisco. "And people hate that."
Olsen said he and his co-author examined how causes of death were determined in the studies because they were puzzled that some studies found a reduction in breast cancer mortality, yet no reduction in overall mortality.
"One possibility is that some [deaths] that would have been labeled as breast cancer deaths are relabeled as something else," he said. "If we look at the data that we can get access to, it seems as if there is a tendency for doctors to more rarely write 'breast cancer' on the death certificates of women that have been screened [with mammography] but more often on those of women who have not been screened," he said. "But the total number of deaths is the same."
The Danish analysis also found that women in the studies who got mammograms received more aggressive treatment for breast cancer, including about 30 percent more lumpectomies and mastectomies, than women who did not receive the test.
Maryann Napoli, associate director of the Center for Medical Consumers in New York, said that finding, coupled with the conclusion that mammography doesn't prevent breast cancer deaths, indicated that "mammography causes more harm than good."
"I think it should change the debate," she said. "We have been sold a bill of goods about the idea that early detection saves lives."
Kramer of the National Cancer Institute said some researchers would like to try to resolve the conflicting findings of the large mammography studies by amassing all the "raw data" from the studies and reanalyzing them according to uniform rules. So far, he added, some of the researchers in charge of the studies have refused to hand over the data.
Olsen said he thought the only way to determine conclusively whether mammography is beneficial would be to conduct a giant international study and measure its impact on overall mortality. But other researchers doubted that such a project is possible, because of widespread faith in the test's benefits among doctors and the public.
"You'd have to randomize people to mammography," said Mary B. Barton, an instructor of ambulatory care and prevention at Harvard Medical School. "You'd have to keep mammography away from the control group. No one here is going to sign up for that."
