Eye-opening article explains mammography debacle

 Blogged here with full permission of the author, Cornelia J. Baines MD, MSc, FACE

Professor Emerita, Dalla Lana School of Public Health, Ontario, Canada.

Published on Canadian Women's Health Network (http://www.cwhn.ca)

Unpacking the great mammography debate

December 18, 2012.

Unpacking the great mammography debate


By Cornelia J. Baines

Since the late 1980s people have argued heatedly about the usefulness of mammography screening, especially in women aged 40 to 49. This unceasing controversy continues. Burgeoning new research in the first decade of 2000 clearly revealed that therapy was outweighing the impact of mammography in terms of mortality reduction. But the defence of screening persists. In September 2012, the Journal of Medical Screening [3] (JMS) issued a supplement reporting that, based on European data, screening benefits were unquestionable. I document the flaws underlying this conclusion, but these flaws are more easily understood when framed by the controversy’s history. So I outline events up to 2000 and then critique the JMS message.

Since 2000, many downsides of screening have been widely reported. Twenty-one hundred (2,100) women aged 40 to 49 must be screened every two years for 11 years to avoid one breast cancer death. Of these, 700 women endure false-positive screens leading to unnecessary diagnostic work-up and anxiety, and ten to 15 will be over-diagnosed and receive unnecessary breast cancer treatment. (Having breast cancer, I am appalled that anyone might undergo unnecessary therapy.) Importantly, contrary to expectations, screening has not reduced subsequent incidence of advanced cancers, a prerequisite for successful screening.

The good news is that even in jurisdictions without screening programs, breast cancer death rates have declined to the same extent as observed in screened populations, with declines even in women in their 30s to whom screening is not offered. The message is clear: breast cancer treatment has radically improved over the last few decades leaving little opportunity for screening to help. And downsides to screening are incontrovertible.

Why the continuing defence of screening? Let’s look at the history.

In the beginning, that is, in the last half of the 20th century, randomized controlled trials evaluated breast cancer screening. Women in a specific age group were either randomized to mammography or to be “controls” with no mammography. Of all the trials, two received much attention.

Trial A used signed informed consent for both groups. Trial B assumed consent with attendance for screening but controls were “un-informed.”

Trial A randomized individuals. Trial B randomized clusters of women.

Trial A compared mammography to no intervention in women aged 40 to 49 and mammography plus clinical breast examination to clinical breast examination alone in women aged 50 to 59. Trial B compared mammography to no intervention.

In Trial A, 100% of participants complied at first screen. Not so for Trial B.

Trial A annually collected detailed demographic and medical information from both groups, for the trial’s duration. Trial B recorded only age at entry.

Trial A offered four to five rounds of screening. Trial B only two to four.

Trial A used two-view mammography; Trial B used single-view.

Trial A screened women every 12 months; Trial B every 24 to 33 months.

Trial A at first screen had smaller cancers and a higher cancer detection rate compared to Trial B.

Trial A over time published consistent numbers regarding trial outcomes such as cancers and deaths; only in 2011 did Trial B finalize its numbers.

Trial A audited mammography externally based on a randomized stratified sample. Trial B described no audit.

Trial A externally audited all tissue biopsies. Trial B did not.

Trial A externally reviewed causes of deaths for confirmed or probable breast cancer cases. Trial B reported their procedure 26 years after first publication.

Now it came to pass that Trial B reported that mammography hugely reduced breast cancer deaths in women aged 40 to 74 and the world rejoiced. Paradoxically, Trial B also revealed a 26% excess of breast cancer deaths in screened women aged 40 to 49; that was ignored. Trial A reported no screening benefit for women aged 50 to 59 and the world was sore aggrieved. Excess breast cancer deaths in screened women aged 40 to 49 (as in Trial B) was loudly condemned. Trial A (The Canadian National Breast Screening Study) was found to be flawed, while Trial B (The Swedish Two-County Trial) was praised and magnified. Forth went Two-County prophet Tabár with Mammography Education Incorporated [4] proclaiming “mammography saves lives” for nigh unto 30 years. Revered he preached in upscale hotels at touristic destinations for substantial fees, and new disciples spread his gospel. Corporations like GE Healthcare and Siemens (that make mammography units) gave their benediction, donating vast sums to the American College of Radiology to promulgate the prophet’s message. And so great was women’s fear of breast cancer, they too became converts steadfastly believing in screening mammography. Others following the new gospel were surgeons and pathologists, for lo their work-load increased!

Evidence from all the trials comparing mammography to no mammography reveals a 15% reduction in breast cancer mortality for women aged 50 to 69. But North American women aged 40 to 49 insisted that they too be screened. Radiologists even persuaded the U.S. Senate that these women deserved no less, in spite of unbiased, expert opinion to the contrary. Those not endorsing screening for women under 50 included the Canadian Task Force on Periodic Health Examination, the Canadian Workshop Group (Canadian Cancer Society, Department of National Health and Welfare and the National Cancer Institute of Canada), the U.S. Preventive Services Task Force, the American College of Physicians, the U.K. Forrest Report, the International Union Against Cancer, the European Group for Breast Cancer Screening, the New Zealand Cancer Society, the British Medical Association, and national policies in Finland, Denmark and Holland. Such dissent was heresy. The public agreed the best available version of the truth came only from radiologists and their acolytes.

Despite the wave of more recent research showing fewer benefits than previously thought from screening, not surprisingly, screening advocates are strenuously defending themselves, most recently in a supplement to the Journal of Medical Screening (JMS), describing findings from the Euroscreen Working Group (ESWG). Where controversy prevails, the messengers as well as the message require scrutiny. JMS has a low impact factor of only 1.69, compared to the Journal of the National Cancer Institute (13.757), Annals of Internal Medicine (16.7), New England Journal of Medicine (51), Lancet (33.797) and BMJ (14.093). What could explain the choice of a low impact journal?

The JMS’ pro-screening message was praised in AuntMinnieMobile [5], an online entity offering "news, educational links, … and information for radiologists, technologists, administrators and medical imaging professionals.” AuntMinnieMobile is sponsored "in part" by Phillips Health Care, manufacturers of imaging systems.

“The researchers found that mammography had a mortality reduction of 48% among women who were screened. What's more, the study found that the number of women whose lives were saved outnumbered by two to one the number who were ‘over-diagnosed,’ or who had cancers detected that might never cause symptoms or threaten their lives” (Letter from the Editor, Devil 13/09/12).

With these results, the editor declared, the breast screening controversy was finally over!

However, according to objective expert panels, breast cancer screening does not remotely achieve a 48% mortality reduction and the number of over-diagnosed women vastly exceeds breast cancer deaths avoided. Why the huge divergence between ESWG findings and panels like the Canadian Task Force? Simply put, screen advocates dominate both ESWG and JMS; objective panels include a range of expertise and minimal opportunity for conflict of interest.

Scrutiny of JMS is revealing.

  • The JMS editor (ND) has for decades co-authored articles with Tabár (Two-County Trial).
  • SD, a JMS editorial board member co-authored many papers with Tabár, was ESWG Coordinator, and the supplement’s Guest Editor and Review Co-ordinator, authoring two of eight articles.
  • SM, a JMS board and ESWG member authored three articles.
  • Board member AKH wrote the editorial.
  • ESWG member (MB) acted as a Review Co-ordinator and authored four articles.

The overlapping roles are clear.

Turning finally and briefly to the JMS message, what are the flaws?

  • Omitted from the text are studies demonstrating that breast cancer mortality has declined in the absence of screening, even in the too-young-to-be-screened.
  • Over-diagnosis is a much larger problem than claimed in JMS. A recent report in the high impact New England Journal of Medicine shows that more than a million U.S. women have been over-diagnosed in the last 30 years because of screening.
  • Totally overlooked is the 20% increased risk of mastectomy for screen-detected cancers.
  • Screening benefits are inflated. Long recognized as biased and inappropriate for evaluating screening, case-control studies that falsely inflate benefit are included in the ESWG analysis.
  • Incidence-based-mortality [7] (IBM) partitions mortality by screening and further inflates benefit. “One must be cautious in interpreting the results (based on IBM), since factors like lead-time bias can influence these analyses, whereas they generally will not influence overall death certificate mortality.”
  • Lead-time (earlier detection extending survival time backwards) and over-diagnosis both spuriously enhance benefit.
  • The JMS reports a 48% breast cancer mortality reduction in women attending screening, comparing attenders at lower risk of death (healthy screenee effect) inappropriately with non-attenders, at higher risk of death. A valid comparison requires two similar populations, one offered screening and the other not, each including attenders and non-attenders.
  • The potential for conflict of interest is evident.

Women have a choice: they can rejoice in the message they want to hear or they can consider whether the message is serving interests other than their own.

Cornelia Baines, MD, MSc, FACE, was co-principle investigator and deputy director of the Canadian National Breast Screening Study, a major trial of breast screening that enlisted 90,000 women across Canada in the 80s. She has closely followed screening outcomes research since then. Baines has also been involved in research respecting silicone breast implants and multiple chemical sensitivity. She is Professor Emerita at the Dalla Lana School of Public Health in Toronto.