In an analysis of Norwegian mammography screening program data, Kalager et al report in the current issue of the Annals of Internal Medicine that the estimated rate of cancer overdiagnosis (detection of cancer at screening that otherwise would not have been detected in a person’s lifetime) ranges from 15 percent to 25 percent, translating to 6 to 10 women overdiagnosed and only 1 death prevented for every 2500 women invited to be screened every 2 years for 10 years, based on data with an average of 2.2 years of follow-up.  

Unfortunately, as with the authors’ initial publication in the New England Journal of Medicine, their analysis is flawed, leading to an overestimate of overdiagnosis and thus invalid conclusions about the value of mammography.  The truth is that the Norwegian screening program is too early in its inception, and there is too little follow-up, to draw conclusions about the effectiveness and side effects of mammography with any measurable confidence.  Fortunately, there are randomized controlled trials with long-term follow-up, and national programs that introduced population-based screening much sooner than Norway, from which there are much better and more trustworthy estimates of mammography screening performance.  

For example, using 29-year-follow-up data from the Two-County Trial in Sweden, it is estimated that there are 7.5 to 10 deaths prevented among 2500 women invited to be screened every 2 to 2.5 years for 10 years.  Estimates derived from evaluation of the data from the UK mammography screening program, also with about 30 years of follow-up, are that 2 to 2.5 lives are saved for every overdiagnosed case.  The more robust, more mature, and more carefully analyzed data not only prove the life-saving benefits of mammography screening, but also indicate a much smaller frequency of harms (including overdiagnosis) than what is suggested by Kalager et al.