Search This Blog

Loading...

Follow by Email

Wednesday, April 2, 2014

Mammograms are not as awesome as we said they were: Damage control articles published in JAMA

A few weeks ago I wrote a blog which addressed a newly released study of the effectiveness of mammograms. This article, in the the British Medical Journal, looked at women who were followed over a 25 year period as part of a Canadian study, found that there was no evidence that mammograms reduced deaths from breast cancer. Most women who will die of breast cancer do so regardless of whether that cancer was diagnosed by a mammogram, and the few who are saved because their cancers were diagnosed earlier than they would have been if only clinical exam were used may have been canceled out by the far more significant number of women who were harmed by overdiagnosis, that is being diagnosed with a breast cancer that would never have troubled them had they not had mammogram screening.

I was concerned that this didn't set off a huge discussion among doctors and all of the rest of us about whether we should really continue to do regular screening mammograms. Our lives, health and billions of dollars rest partially on the answer to this question. The fact that it is so important is probably why the conversation has been so slow to start. Not doing the tremendous number of screening mammograms we now do would be very destabilizing since hospitals and surgeons and radiologists depend heavily on this revenue stream. Women's health centers at hospitals are primarily about mammograms and breast cancer diagnosis and treatment. These employ nurses and administrative assistants and social workers and counselors. There are ongoing and yearly campaigns designed to get women to get mammograms. A major change in policy has the potential to free up a great deal of financial and human resources, but at the same time jobs will be lost and budgets broken. If we accept that mammograms have lead to substantial overdiagnosis, this will have a huge emotional impact on women who have been diagnosed with breast cancer as a result of mammogram screening.

So it is not surprising that the response to this Canadian study was measured.

Today two articles were published in the Journal of the American Medical Association one addressing mammograms in general and the other mammograms in women over the age of 74. These articles have been published less than two months after the BMJ article and are already getting lots of press and lots of discussion. The article about the overall risks and benefits of mammography is a review of multiple studies, including the BMJ article, and is really interesting to read. It delicately steps around some strong evidence that mammograms have no particular value in saving womens' lives and comes up with numbers that nevertheless make preventative mammogram screening look unattractive. The conclusions are that mammograms do reduce breast cancer deaths, but on the order of 1-50 per 10,000 women screened for 10 years, depending on age. A total of around 300 will be diagnosed with breast cancer and up to a third of these cancers will be overdiagnosed, resulting in women presumably being treated with radiation, surgery and chemotherapy for tumors that would not have caused harm. Six-thousand of the 10,000 screened with mammography, fully 60%, will be called back during this 10 years for abnormal mammograms that will need further workup, including more imaging and biopsies. Still, the conclusion is that mammograms reduce breast cancer mortality, but only a little bit.

But what about the Canadian study published in February? This looked at women aged 40-59 who were screened with either clinical examination (examination of the breast by a trained health care provider) or mammograms plus clinical examination over the 5 year study period and then followed for 25 years. The only difference in the groups at 25 years is that the mammogram group had more breast cancer. There was no mortality difference. Did the Canadian women in the BMJ study play catch-up and get mammograms after the study was done and so reap all of the mortality benefit? If so, they appear to have also avoided a certain amount of overdiagnosis by taking their 5 year holiday. There are many ways to study the efficacy of mammogram screening, and none of the many studies that have been analyzed and meta-analyzed was really able to do a gold standard approach. Because of the fact that we have embraced mammogram screening as our standard of care, we have not done the definitive study.  Ideally we would compare a group of women who were denied access to mammograms for 25 years and only received clinical breast exams or breast self examination to a group who had mammograms at varying frequencies, say every 1-4 years, along with their clinical breast exams. No such experiment has been done so we rely on evidence gleaned from huge populations over many years but with less than ideal designs.

The other article in today's JAMA tackles the question of mammogram screening in patients older than 74. No actual studies have been done on this population and many countries stop recommending mammography for patients aged 70-75 years of age. In the US a significant proportion of women getting mammograms are over 74. Today's article concludes, based on extrapolating the data we have based on younger patients' data, that there would be a mortality benefit of mammogram screening if the women in question were expected to live 10 or more years. That is such a can of worms. I have very warm and respectful relationships with many older women who are my patients and I find it very difficult to admit to myself, much less them, that I expect they will die before 10 years elapse. That also assumes that I have any reasonable idea. The most important predictor of being alive tomorrow is being alive today, which is the message I like patients to take with them.

In a delightful juxtaposition, Dr. Mary Tinetti, an academic geriatrician, wrote an article about how extrapolating benefits of interventions from younger to older patients is often inaccurate. She doesn't address the mammogram issue, but her article is well placed. We really don't know what good or harm mammogram screening will do for our patients beyond the ages we have studied. It does appear, however, that overdiagnosis increases with age (see the BMJ article), so more women in their final decades will be diagnosed with breast cancers that would never have caused them harm if we continue to subject them to screening mammograms. I have watched my elderly patients suffer through radiation and chemotherapy and it is a lousy way to enjoy retirement.

So this kerfluffle about whether to do mammograms or not is a really big deal, and there is much damage control going on. It is interesting to look at this from the sidelines, and I am sure there is much that I am not perceiving. Many well meaning people are highly invested in the prevention of breast cancer and much good work is being done. Mammography is definitely not going away. It is a reasonable way to detect breast cancer, and detecting breast cancer comes before treating it, and we have improved tremendously in our ability to effectively treat and cure it. Mammogram screening for patients at high risk yields a whole different set of numbers than what I have quoted. There may also be ways to augment mammogram screening with other testing to make sure that the breast cancers we treat actually need to be treated. It is past time, though, that we question the wisdom of pushing for regular mammograms in unselected women over the age of 50.

1 comment:

Jamie Derrick said...

very useful and well written... thank you for pulling the facts together. There is another issue that I'm aware of as a psychologist and that is the stress that is added by doing a yearly mammogram. Women in my practice worry about it before the test, are stressed during the test, and then worry until the results come back. Some postpone the whole thing but then worry they are being foolish. In a society that generates fear and the addresses fear by selling people something --- well, this seems relevant to the discussion at hand also. The 'intangible' costs matter also.