The Canadian National Breast Screen-ing Study (NBSS) should not influence decisions about screening mammography, neither for individuals nor on a policy level. The recent British Medical Journal publication of the latest update on the NBSS is not new research. It was because of major problems with its design and execution that in 2002 the World Health Organization excluded the NBSS from analysis of the impact of screening mammography on mortality from breast cancer.
The biggest flaw of the NBSS was corruption of the randomization process. When women volunteered for the study, but before they were assigned to the control or study group, they had a thorough clinical breast exam. It is clear that women found to have breast lumps were selectively put in the mammography group.
The assignment was supposed to be done without knowledge of the result of the physical exam, but the names were written on open lists, making it possible for the staff to leave blank lines onto which they could then write the names of the women with lumps. The NBSS is the only mammography trial where more women died in the mammogram group than in the control group, not because mammograms don’t work, but because more women with cancer were PUT into the mammogram group. Of women who died within 7 years because of advanced cancer, eight were in the mammogram group and only one was in the control group.
This wouldn’t have happened if the randomization process had been blinded.
Contamination of the control group was significant. In the control group, 17% of women aged 50 to 59 and 26% of women aged 40 to 49 had mammograms outside the trial. And some of them would have had cancer detected and treated but they were still counted in the control group. This added to the appearance that the death rate was similar between the two groups.
Mammography in this trial was poor quality; they used secondhand equipment to save money. The mammography unit used in the Vancouver Centre of the NBSS was 10 years old at the start of the trial. The false negative rate in the NBSS was worse than in studies done in the 1960s and 1970s.
In the NBSS, the average size of the cancers detected by mammography was 19 mm, only 2 mm smaller than those detected by clinical breast exam. Compare that to our screening program in British Columbia where 65% of cancers detected at screening are less than or equal to 15 mm, and 76% are node negative.
The Canadian Task Force on Preventive Health Care was unduly influenced by this trial. They balanced what they considered the benefits and harms of mammography and concluded that it should be done less frequently. By including the NBSS in their meta-analysis they mathematically reduced the demonstrated mortality reduction.
The Task Force overstated the harms: They considered it a significant harm to make women nervous when called back from screening and for some to have unnecessary biopsies. The vast majority of these are needle biopsies done with local anesthetic. Most women would choose those short-lived harms over being denied access to screening.
If women heed the bad advice of these authors we will soon see an increase in the average size of breast cancers, the rate of axillary metastasis, and the death rate.
—Paula Gordon, OBC, MD, FRCPCClinical Professor, Department of Radiology, UBC