The benefits and drawbacks of mammography screening continue to rumble on. In October 2012 an independent review panel published their findings on breast cancer screening in the United Kingdom, and concluded that, despite overdiagnosis, screening using mammography saves lives. However, a recent observational study of 30 years' worth of data gathered in the United States and published in the New England Journal of Medicine questioned the usefulness of mammography screening in light of high rates of overdiagnosis.
In the UK review, the panel decided that data from randomized trials provide the best estimate for addressing the benefits and risks of mammography screening, despite the caveats that these are old trials and that the methodology used in some of these trials might have introduced bias. Their meta-analysis of 11 randomized trials led them to conclude that in women invited for screening compared with controls (women not invited for screening) the relative risk reduction of dying from breast cancer was 20% (relative risk 0.80 (95% confidence interval (CI) 0.73–0.89)). Moreover, in terms of absolute benefit the panel calculated that for every 10,000 women invited for screening between the ages of 50 and 70, 43 breast cancer deaths would be prevented. What about overdiagnosis (women whose breast cancers are unlikely to cause their death during their natural life span)? The panel looked at three of the randomized trials because only these trials followed the control group of women for 5 or more years after the trial and did not offer them screening. From the point of view of a woman (aged 50–70 years) invited for screening every 3 years over a 20-year period and diagnosed with breast cancer during this time, the panel estimate that the likelihood of overdiagnosis would be 19%. Overall, the panel concluded that for every breast cancer death prevented by mammography screening, about three women will be overdiagnosed and unnecessarily treated. Although the UK panel also looked at data from observational studies, they decided that such data do not provide a robust approach for estimating overdiagnosis.
