Miller's study, published in 2014, is a randomized controlled trial, which corresponds to the highest quality criteria for population-based studies. Data are analyzed from groups whose subjects are randomly selected and then compared.
Here, the study involves 90,000 women, 45,000 with screening, 45,000 without screening. In fact the trials (NBSS 1 and 2 , National Breast screening studys ) were conducted in Canada in the 1980s with women screened annually for 5 years with annual mammography and clinical examination, and then followed up for 10 years. Here Miller proposes a re-evaluation after 25 years of follow-up for these two groups.
What are the conclusions?
1°-No difference in mortality between the two groups (mortality = number of deaths in relation to the total number of people screened).
2° Survival rates are identical, regardless of tumor stage(survival = number of deaths in relation to the number of cancers diagnosed)
3° 22% over-diagnosis
4° No difference between the two groups in the rate of fatal cancers.
More precisely, Miller finds 22% overdiagnosis, or 1 overdiagnosis (and thus overtreatment) for every 424 women who received mammography screening, for a zero benefit regarding the reduction of mortality from breast cancer.
The criticisms that have been made against Miller have been varied. First, it was argued that there could have been contamination of both groups because of the length of follow-up. In fact, some of the follow-ups stop after 7 to 10 years, which limits two drawbacks: some women in the non-screening group could still have had a mammogram one year or the other, while some women in the screening group could have "missed" a year of mammography. Waiting another 20 years would dilute or blur the results.
-First, the effect of non-compliance in the screening group and contamination in the non-screening group will rather lead to an underestimation of over-diagnosis.
-Secondly, the detractors of these studies argued that it would take a very long time to see the effectiveness of screening, as it would only be over a very long period of time that the danger of undetected cancers in the non-screened group would be seen. But here, even after 25 years, we still do not see this famous " dormant cancer " finally appearing, and no excess mortality of women who are not screened, perhaps because dormant cancer does not exist...
Miller was also criticized for not being representative of the French system, which screens every two years and begins at age 50 (whereas the Canadian trials targeted women aged 40-59). However, in the United States there was a debate about starting screening at 40 years of age.
It is clear that mammography was prematurely marketed to us as the ideal way to reduce the danger of cancer, particularly the killer one.
Ref : Miller AB, Wall C, Baines CJ, Sun P, To T, Narod SA. Twenty five years follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial. The BMJ. 2014 Feb 11;348:g366