So the group, the U.S. Preventive Services Task Force, started what it thought would be a straightforward job: gathering the newest science and asking about the benefits and risks of breast cancer screening, the best time to start and how often women should be screened.
The group ended up recommending that most women forgo routine mammograms in their 40s and test every other year instead of every year.
The response was swift and angry. Professional groups, like the American College of Radiology, advocacy groups, like the American Cancer Society, and politicians said the guidelines would deprive women of a life-saving test. And some said the guidelines were politically motivated to save money.
Panel members have been taken aback by the response. Their work seemed almost mundane, they say, just an effort to gather and evaluate the best possible evidence.
The task force, a 16-member panel of experts appointed by the Department of Health and Human Services, began its work as usual. It went to an academic center, in this case the Evidence-Based Practice Center at the Oregon Health and Science University, and asked for an extensive review of all the relevant papers published on breast cancer screening, including ones used in the last review. At that time, the task force recommended routine screening starting at 40, saying that there were benefits although they became greater as age increased. The Oregon group had done similar reviews for the panel, including a review for the 2002 guidelines.
This time, the panel hoped that it could get missing pieces of the puzzle. New studies allowed scientists to zero in on benefits and harms for women in their 40s and to evaluate with far more certainty not just whether women should be screened but also how often.
The Oregon scientists began by combing the literature. By November 2007, the researchers, led by Dr. Heidi D. Nelson, a professor of medicine, medical informatics and clinical epidemiology at the university, had finished its review and sent its work to 15 outside scientists for review, then sent it to the panel. Finally, the researchers were ready to make their first full presentation to the panel members.
Part of that evidence, which Dr. Nelson’s group included, was new results from a huge study in England of mammograms for women in their 40s. This study, published in 2006, compared 54,000 women offered mammograms starting at age 40 with 107,00 women the same age who were not offered them. Previous studies of women in their 40s had them starting at various times in that decade of their lives and so were less useful.
But the British study saw only a small decline in the breast cancer death rate after 10 years, and it was not statistically significant, meaning it could have occurred by chance. Previous studies also failed to find a statistically significant effect for women in their 40s.
Dr. Nelson’s group did a new analysis combining all the studies. By adding up all the small benefits, the researchers concluded that there was a slight benefit of screening, a statistically significant 15 percent decline in the death rate from breast cancer for women in their 40s.
That means, they said, that 1,900 women ages 40 to 50 must be screened to prevent one death from breast cancer up to 20 years later. At the same time, even with the screening, five deaths would have occurred anyway, probably because many of those cancers grew so fast that no matter how early they were found it was impossible to cure the women. So in the end, 1 out of 6 deaths would have been prevented.
The task force wanted more information. What about the harms of screenings for women in their 40s, it asked?
One harm is excess tests, like biopsies. But there was not much published data.
Dr. Nelson’s group drew upon a National Cancer Institute database of eight million mammograms in the United States telling what sort of mammogram — digital or film — the women got, when they got it, and whether they had follow-up tests. Analyzing those data, she concluded that women in their 40s have about a 10 percent chance of a false positive and a 1 percent chance of having a biopsy each time they have a mammogram. While those risks are small, they gain more significance when weighed against the relatively small risk of cancer for women in their 40s — a risk of 1.5 per of 1,000 women.
The serious harm, panel members said, is overdiagnosis, finding cancers that are better off not being found.
In 2002, when the group last reviewed breast cancer screening studies, the idea of overdiagnosis was not well formed. It has been hard for many people, even scientists, to believe that some cancers start then stop or even regress. But researchers all over the world have been finding overdiagnosis in studies of all sorts of cancers.
Dr. Barnett Kramer of the National Institutes of Health, who was not part of the panel, described overdiagnosis as “pure harm” because it means that women are treated with measures like chemotherapy, radiation and surgery for tumors that do not need treating.
Dr. Russell Harris, who was a panel member and is a professor of medicine at the University of North Carolina, said overdiagnosis “became an issue in discussions and it was not an issue before.”
One way of looking at cancer is as three different diseases, Dr. Harris said. One type grows so fast that early diagnosis is futile. Another grows so slowly it does not need to be found early to be cured. And as many as a quarter of those slowing-growing cancers would never be noticed in a woman’s lifetime. Cancers in the third group can be cured if they are caught early. But, Dr. Harris said, at least with breast cancers, that third group makes up only 15 percent of the deadly cancers.
The panel struggled to get an estimate of overdiagnosis, but could not get a precise figure.
“Harms at least need to be discussed,” said Dr. Timothy Wilt, a panel member and a professor of medicine at the University of Minnesota. “More tests and more treatment isn’t necessarily a better health outcome.” Dr. Wilt estimated that as many as 30 percent of cancers found by mammograms in women in their 40s were overdiagnosed.
The panel wanted more, though. It wanted a better handle on overdiagnosis at all ages, precise and accurate figures. And it had another question that had not been answered in 2002: What is the best interval between mammograms?
Clinical trials did not answer that question because they did not directly compare screening intervals.
In January 2008, the panel commissioned six independent groups to do statistical modeling and get the answers.
Overdiagnosis proved impossible to model, said Donald Berry, a statistician at MD Anderson Cancer Center who was one of the modelers. Estimates of overdiagnosis ranged from 6 percent to 50 percent of breast cancers, and he was not comfortable putting a precise figure on the problem.
Screening intervals were a different story. There, the modelers agreed, the answer was clear. Their analyses concluded that mammograms every two years give the nearly the same benefit as annual ones but confer half the risk of harms. The analysis involved national data on how many women got mammograms and how many did not, at what interval, and what happened subsequently — how many women got cancer and how many women died of cancer.
The modelers gave their paper to the panel but did not know if they would use the analysis. Dr. Berry suspected the group would say pretty much what it said in 2002, not daring to so forcefully embrace the new findings, which he said clearly indicated that there is almost no benefit to screening women in their 40s and that women can be screened every two years instead of annually.
But the task force took the new research to heart.
“I was pleasantly surprised,” Dr. Berry said.