The Screening Myth
TORONTO – Breast-cancer screening has long been viewed as one of the most important tools for reducing mortality from the disease. That is why recent doubts about its effectiveness – intensified by the publication in February of the 25-year follow-up to the Canadian National Breast Screening Study – have come as such a shock. How can breast-cancer screening, which facilitates early detection of the disease, not prevent deaths from it?
Understanding screening’s limitations requires, first and foremost, an understanding of the process. A mammogram (x-ray of the breast) is administered to ostensibly healthy people to detect unsuspected disease. If any abnormalities are found, a diagnostic test is conducted to confirm the presence of the disease. If the results are positive, treatment begins.
The first limitation of breast-cancer screening is obvious: where effective diagnosis and treatment are not available, screening cannot have any impact. But there is more to the issue – namely, whether screening ultimately fulfills its intended purpose of reducing breast-cancer mortality rates.
There have been several attempts to evaluate, through observational studies, the impact of screening in the general population. But such studies, which are based on comparisons between uncontrolled groups, tend to produce biased results. None has been able to account for differences in treatment. More important, none has found a reduction in advanced breast cancer in the screened groups – a requirement to deem screening effective.
With observational studies producing inadequate results, scientists turned to a procedure developed years ago to evaluate new tuberculosis drugs: randomized screening trials. Such trials begin with the selection of women at risk of developing breast cancer, and randomly allocate them to be screened or not.
Any woman in either group who develops breast cancer is treated to the fullest extent possible, following the closest possible treatment plan to other participants in the trial (taking into consideration the stage of the disease at diagnosis). Researchers follow the women’s progress over many years, with the goal of comparing the two groups’ mortality rates.
In North America, where, for ethical reasons, randomization requires informed consent, only two such trials have been conducted. The first wasin New York in the 1960s, using annual mammograms and breast examinations for the screened group.
The second, the Canadian National Breast Screening Study, which began in the 1980s, also used annual mammograms and breast examinations for the screened group. But, in the control group, women aged 50-59 also received annual clinical breast examinations, and women aged 40-49 were given a single breast examination.
In Europe, one trial began in the United Kingdom in the 1970s, again using breast examinations and mammography for the screened group. A more recent British trial, performed on women aged 39-41, used only mammography, as did the largest of all, the Swedish two-county trial. Though the trial results have been inconsistent, the conclusion is clear: breast-cancer screening does not reduce mortality rates.
The trials that have shown a significant decrease in mortality are usually older, and other factors have compromised the results. For example, the Swedish two-county trial was randomized by clusters of population groups, and the comparability between the women in the screened and control groups was not confirmed. Moreover, the trial was conducted before modern breast-cancer treatments were freely available.
More recent trials, such as that conducted in the UK, show at most a non-significant benefit from screening. The results of the 25-year follow-up of the Canadian trial – which can be considered the most accurate, given that the women were carefully characterized according to risk factors and had access to the most advanced treatments – are even more damning.
Beyond showing no benefit from mammography screening, the Canadian trial highlighted one of the practice’s major disadvantages: over-diagnosis. Indeed, 22% of those in the screened group were over-diagnosed, meaning that the cancers found would not have become clinically detectable during the woman’s lifetime. Add to that the in situ cancers – early-stage cancers that have not spread to neighboring tissue – and the number rises to 50%.
To be sure, over-diagnosis is inevitable with good screening tests. The real problem is the over-treatment that follows, owing to the lack of reliable tests to differentiate observable lesions from those that actually require medical intervention.
Many of the women who believe that their lives were saved by mammography screenings actually had so-called “IDLE” cancers (indolent lesions of epithelial origin), which would not have progressed before they died of other causes. Meanwhile, the very aggressive cancers, which are most likely to prove fatal, tend to present as “interval” cancers that appear between screenings. In the deadliest cases, regular screenings would not make much difference.
Efforts to develop new, more sensitive screening tests are thus likely to do more harm than good, as they increase the rate of over-diagnosis, without improving outcomes among women with aggressive cancers. What breast-cancer patients really need are tests that identify the risk that the disease actually poses, as well as new, lifesaving treatments. It is time to recognize that breast-cancer screening does not save lives – and to focus on the strategies that will.
Anthony B. Miller is Professor Emeritus at the Dalla Lana School of Public Health, University of Toronto.