Radiologists aren’t the only ones criticizing the new mammogram study

Supporters of the most recent paper from the Canadian National Breast Screening Study (CNBSS) falsely contend that only radiologists are criticizing this study. This simply is not true. The Canadian study flaws have been well documented for decades.

Robert E. Tarone at the National Cancer Institute (who isn’t a radiologist) wrote in 1995 that there was a statistically significant excess of advanced cancers that were allocated to the mammography group. The World Health Organization long ago excluded the CNBSS from its analyses of screening mammography’s impact of breast cancer mortality. In a recent interview with CNN, the American Cancer Society echoed similar methodological concerns to those raised by American College of Radiology, Society of Breast imaging and others. Breast cancer groups, such as Breastcancer.org, have criticized this study and warned against following the author’s recommendations. CNN’s Dr. Sanjay Gupta detailed the problems with applying such an old study (conducted in the 1980s) to today and recommended regular mammograms for women ages 40 and older.

The study was not blindly randomized. Women were examined by a doctor or nurse before being assigned to the control group (which did not get mammograms) or the group that was screened.  This means that the doctors or nurses who may have felt a lump in a woman’s breast, and/or lumps in her armpit that might be lymph nodes to which the cancer spread, were faced with the choice of putting a woman that they strongly suspected had cancer into the group that wouldn’t be screened.

Since these women were assigned from open lists, how many local coordinators, out of common decency, acted to save these women’s lives? The coordinators should never have been put in that position. Study authors contend that local staff did not affect randomization. However, the Canadian government inquiry confirmed that a local coordinator was, in fact, fired because study organizers felt that this person had done so. Leaders of the inquiry admit that they did not talk to local coordinators because they felt coordinators “would have been unlikely to admit” affecting randomization. Coordinators were also not permitted to be interviewed by anyone else. So, how is this question somehow settled?

The CNBSS survival figures also support healthy skepticism. The 5-year survival rate in the CNBSS unscreened group was over 90 percent. At that time in Canada, the 5-year survival rate was 75 percent. This kind of discordance again raises questions about randomization. As the American College of Obstetricians and Gynecologists noted in their press release on the BMJ study, the CNBSS was the only one of many large randomized trials to show no mortality reduction from screening. The CNBSS is also the only one in which organizers knew which women likely had cancers prior to allocation.

The study’s own reference physicist stated in peer-reviewed published studies that the study’s “[mammography] quality was far below the state of the art, even for that time.” Many of the facilities in the trial, which took place in the 1980s, used older mammography machines. Many of the technologists who performed the mammograms across Canada were not properly trained on how to position the woman to include as much breast tissue as possible, so some cancers that could have been detected on the mammogram were missed.

These deficiencies more than explain why only 32 percent of the cancers found in the Canadian study were detected by mammography when most large trials show that mammography detects over 60 percent of cancers and why the cancers found by mammography were more advanced (larger) than those found by mammograms in other trials. Conversely, the false negative rate in the trial (when cancer is present, but missed on a mammogram), was actually worse in the Canadian study than in those done in the 1960s and 1970s.

Criticisms of the CNBSS are well founded and have been well known for 20 years. These factors are why medical professionals worldwide have rejected the trial’s conclusions. Mudslinging doesn’t change the facts. However, it does add to confusion among women regarding when to be screened for breast cancer and may ultimately cost lives.

Geraldine McGinty is chair, commission on economics, American College of Radiology.

Comments are moderated before they are published. Please read the comment policy.

  • Patient Kit

    Thank you for this. You raise enough good questions to support my own healthy skepticism about this study. For now, I’m trusting my own gut instincts about this and opting to continue to get regular mammogram screenings and I would recommend that my younger sisters do the same. I just hope this study doesn’t give health insurance companies and government a way to stop paying for mammograms. I still remember the fight to get insurance to start paying for mammograms. To me, the question of whether or not mammography screening is useful and good for most women will always be haunted by this hardcore financial fact: It would be much cheaper to only treat women with advanced breast cancer than to screen all women on a regular basis and treat early cancers. I’m not ignoring the threat of false negatives and positives, but I want women to have the option to take that risk. And I don’t want questionable science to hand insurance the out that I suspect they would love to have.

    • Lisa

      Kit, if insurance companies were looking for a way to sop paying for screening mammograms, the USPSTF recommendations that women begin having screening mammograms at age 50, would have allowed them to stop paying for younger woman’s screening mammograms. I think the right to screening mammograms beginning at age 40 won’t be attacked, in part due to the pro-mammogram lobbies.

  • Kristy Sokoloski

    The fact that over 60% of the cancers are found by mammography as indicated by large scale studies tells me that this number is still not impressive. If they would have said that mammography can detect 90% to 95% of the cancers then I would say interesting. This study regardless of how old it is and others like it that were done in the past and even now in the present just continues to show why much of Medicine is opinion based. And with every study there’s always room for bias to occur regardless of the criteria used to design the studies. And they are also based on the opinion of the authors conducting the studies. Which is why I don’t put a lot of weight in to them although I find some of them interesting reading.

  • Lisa

    The US Preventative Services Task Force is updating its recommendations on breast cancer screening. I suspect their findings will basically the same as their 2009 recommendations: that women with average risks have biennial screening mammograms between the ages of 50 to 74.

    There were debates about the value of mammograms before the recent paper on the CNBSS was published. I think it is quite unlikely that this study will change anything. What the paper has done is given those who are in favor of screening mammograms a platform to spread their message.

  • Markus

    I am a little amused that this article “Radiologists aren’t the only…” is authored by a person connected to the radiology world.
    On a serious note, what articles should one read? The international consensus group referenced in this article does not give screening mammography a glowing endorsement. They estimate, note the word estimate, a 5-10% reduction in mortality using what seems to me a best case scenario, and they say that they are awaiting long term outcomes since they acknowledge that most studies are of too short a duration for a disease that actually does have a long mortalit

  • Andrea Borondy Kitts

    We had a 3 hour discussion in my UConn Chronic Disease Control class last week about the recent British Medical Journal (BMJ) paper on the 25 year follow up to the Canadian National Breast Screening Study. In attendance were 2 cancer epidemiologists, one a breast cancer expert, and an oncologist with breast cancer expertise. In conclusion, the consensus among the experts was the study did not provide enough credible evidence to change the recommendation for breast cancer screening in the US.

    Key concerns raised with this study:

    In the mammography group 2/3 of the breast cancers found were palpable. This is not consistent with clinical breast screening experience where most breast cancers found with mammography are too small to feel.

    Screening was only for 5 years. Most breast cancers are slow growing so may not have screened long enough to capture the difference in mortality. The risk for breast cancer increases with age, so each passing year after the screening stopped the women had a higher risk of getting breast cancer. The number of breast cancers found in the mammography and non mammography arms were the same after the 5 year screening period was stopped.

    Although mortality did not differ between the 2 groups, the cancers found in the mammography group were smaller and less were lymph node positive.

    The study is predicated on universal access to adjuvant therapy. Canada has a National Health System so all women have access to care. This is not the case in the US.

    Other discussion points:

    There are 22 million women in the US between the ages of 40 – 49. Not screening this population would save a lot of money.

    Need to have better genetic markers for who will develop breast cancer to risk stratify the population to screen

    My “take-aways” from the discussion

    Large screening clinical trials/studies are hard to do right especially with decades of follow-up; often new technology and new treatments can overtake results

    Policy can be influenced by strong advocacy sometimes resulting in non-evidence based recommendations

    Even the experts don’t fully understand the results and don’t know how to apply to population recommendations

Most Popular