New Mammography Study

From the New York Times Article

A mammography study involving 90,000 women over 25 years has been released.  Several stark conclusions:

1) Death rates from breast cancer were the same in women who got mammograms and those who did not.

2) One in five breast cancers found and treated did not pose a threat to women’s health

3) There is no benefit to finding breast cancers before then can be felt.

4) Studies showing a survival benefit for mammograms may lag behind improved treatments and may not follow standards for clinical research.

Dr. Susan Love commented on this article on facebook.  “One reason that mammography doesn’t add much is that early detection turns out to be less important than biology ie aggressive tumors are worse than less aggressive tumors!”

For years, the drumbeat has been “early detection saves lives.” Study after study as well as anecdotal evidence from my own life shows this isn’t true. Recently I was involved in a flap on Facebook with someone who said that breast cancer survival rates are soaring.

Are they really soaring, I asked?

Yes, you often hear that 90+% of women diagnosed with breast cancer are still alive in 5 years. How much of that statistic is attributable to improved screening technology? What I mean by that — with improved detection in digital mammography, a radiologist might detect a tumor before it is palpable.  A lot of breast cancer grows slowly, so it might be years earlier.  She might survive the five years and pump up that statistic, but if the biology is such that our current medicines can’t stop her specific type of breast cancer, she will go on to die eventually. Maybe in 8 years instead of 4 thanks to early detection.

No, that doesn’t mean that early detection bought her four more years. It means that the cancer was discovered four years before it would have been without mammography and without impacting the outcome. You’ll note that this study found no survival benefit to catching the tumor before it’s palpable.

In other words, we can find a cancer years earlier, but if its biology defies our available treatment methods, what good does that do?  We need to focus on the biology of the tumor, not on early detection.

The article says this about DCIS:

If the researchers also included a precancerous condition called ductal carcinoma in situ, the overdiagnosis rate would be closer to one in three cancers, said Dr. Anthony B. Miller of the University of Toronto, the lead author of the paper. Ductal carcinoma in situ, or D.C.I.S., is found only with mammography, is confined to the milk duct and may or may not break out into the breast. But it is usually treated with surgery, including mastectomy, or removal of the breast.

We need to focus on the reality of this disease, not on what we want it to be. Yes, I know it’s scarier and means we have far less control over our health than we like to believe we do.  But if we really want to change the outcomes, we need to start dealing in facts. One in five is unacceptable. One in three is unacceptable.

The risks of over-treatment are real, folks.  Consider just my case: I don’t have routine scans to check for cancer recurrence.  Since I was diagnosed at age 41 and treated with every technology possible, my doctor believes that radiation from regular scans over the years could eventually cause another cancer in my body.

And that doesn’t even begin to address the physical, emotional, and spiritual toll that chemotherapy, lymph node removal, a bilateral mastectomy, and 26 doses of radiation took from me.  And I needed all that to survive my aggressive Stage III cancer.

As an insider, the idea that women are getting all this treatment without needing it is simply obscene.  All because it serves the status quo.

A friend of mine just sent me one of her favorite inspirational quotes:   “Do the best you can until you know better. Then when you know better, do better.” – Maya Angelou

Twenty years ago, we started down this path of early detection because we thought it would save lives.  Turns out it doesn’t, so now it’s time to do better.

Organizations like Komen, associated most closely with this disease in the public’s mind, need to turn this big pink ship around.  This is not to say that their decades of awareness have been a complete waste of time, but it’s time to move on.  From the article:

Dr. Kalager, an epidemiologist and screening researcher at the University of Oslo and the Harvard School of Public Health, said there was a reason the results were unlike those of earlier studies. With better treatments, like tamoxifen, it was less important to find cancers early. Also, she said, women in the Canadian study were aware of breast cancer and its dangers, unlike women in earlier studies who were more likely to ignore lumps.


“It might be possible that mammography screening would work if you don’t have any awareness of the disease,” she said.

7 thoughts on “New Mammography Study”

  1. Seems to me that the very first sentence of the actual paper, “Regular mammography screening is done to reduce mortality from breast cancer,” is contestable or at least incomplete. Similar to any medical assessment, MS can also be seen as helping to extend life beyond which it would have been otherwise. When viewed thru such a lens, a 25 yr mortality metric is unlikely to capture important possible outcomes. Also, am I reading this wrong or did the ‘mammography group’ receive MS (in addition to ‘manual’ breast exams) only in first five yrs, while the ‘control group’ continued to receive breast exams throughout the evaluation period? Hopefully I am mistaken, as such a design introduces obvious problems in comparing treatment and nontreatment effects past 5 yrs.

    1. Matt,

      Mammography is indeed done to reduce mortality. It’s goal is to catch breast cancer early and according to an increasingly obsolete paradigm, the earlier you catch a cancer, the more likely it is that you can cure it. Emerging science shows that this simply isn’t how it works — whether you will live or die of breast cancer is a function of the individual biology of the disease rather than the stage. This study affirms that.

      As far as the study design goes, there appears to be a 5 year screening period. During that time, one group got clinical exams and screening mammograms, the other group got clinical exams only. All those people diagnosed during the screening period were followed for another 20 years. The death rate in the two groups over the total 25 year period was the same.

      So this is looking at screening methods, not treatment.


      1. When a deadly disease has no general cure, then it is difficult to understand how detection methods of any kind can be construed as reducing mortality. Formal studies showing no difference in long term death rates regardless of treatment path seem wasteful. Isn’t such a result obvious?

        On the other hand, some treatment paths may be more effective than others in prolonging life. And for some people, extra months or years are worth a fortune. It seems, therefore, that an interesting empirical question is how much longer, if any, breast cancer patients live if they are detected by MS. It appears from the paper that ~25% more breast cancer cases were detected in the MS group than in the ‘control’ group during the initial 5 yr screening window (see T1 666 vs 524–a difference that is almost certainly highly significant given the structure of the data), implying that MS generally leads to earlier detection and treatment. Might this prolong life and if so, how much longer?

        If MS screening was done only for 1st 5 yrs and then both groups continued to receive manual exams (or any other assessment/treatment for that matter) for the next 20, then hopefully you see the problem in drawing clean conclusions from this study. One obvious question is whether mortality results would have differed if the MS group continued yearly screenings throughout the entire 25 period–particularly since the initial 5 yr window indicates higher detection rate w MS.

        1. It’s generally accepted (although not well tracked) that 20% of all women diagnosed with breast cancer will go on to develop metastatic breast cancer (the deadly kind, stage 4). So while it’s true that you can never really know you’ve been cured, everyone is trying to get in the 80% who will go on to die of something else.

          The theory has been that you will improve your likelihood of being in that 80% if you catch your cancer when it’s teeny tiny, before it can be palpated. It makes intuitive sense that cancer starts with one bad cell and then two, etc, so the earlier you can catch it, the better it is for you. Over the last couple of decades, there is mounting evidence that this isn’t true, or at least that it isn’t complete. People initially diagnosed with an early stage breast cancer can still die. Anecdotal but an example, my friend Rachel was initially diagnosed in her early 30s with Stage 2. Not by a mammogram, but by her own sense that something wasn’t right and constant pushing for tests with her doctors. She had the recommended treatments, but a few years later it came back as metastatic cancer and she died eight years after her original diagnosis.

          There is something wrong with this early detection idea.

          In this study, only people diagnosed in that five year period were tracked. My reading of it — if you were diagnosed in year 6 or 7, you are not included in these mortality numbers, so whether a woman is screened or not after year 5 doesn’t matter.

          This study does not discuss treatment options, only method of initial detection.

          The fact that more women were diagnosed but the same percentage died is an indication of over-diagnosis, not of the success of mammograms. We all have cancer cells floating around in our bodies. That doesn’t mean we will die of cancer. What we need to learn is what makes some tumors deadly and others a non-existential threat. Right now, we detect early cancers and our only option is to treat them like they are the 20% because we have no way to differentiate.

          Treatments are harsh too – I don’t want anyone going through what I went through unless it’s necessary. A friend likened it to this — you find out you have high cholesterol so the doc tells you you’ve got to have bypass surgery because that high cholesterol might someday kill you. Ridiculous in the framework of heart health, but pretty much the standard of care in breast cancer.

          So the argument here is not “ban mammography.” It is to stop clinging to it as if it is a life-saver, because whether your cancer is large enough to find in a clinical exam, found by the patient, or found in a mammogram, this study shows that your outcome will be the same. That’s what needs to be better understood — why, regardless of stage a diagnosis, some cancers metastasize and some don’t. From there, we can have a rational discussion about treatment.

        2. Thx for the helpful insight. Given aims of the study, would certainly make more sense to track only detections in 1st 5 yr screens. What confuses me is the inclusion of subsequent 6-25 yr detection/mortality data in the analysis (e.g., Table 3, Figs 1, 2). Clouds things, at least for me.

          Still would have been interesting to have reported/compared time from detection until death for those who did die during the 25 yr window. Maybe MS adds time but not enough that it is not blotted out by a blunt long term binary (alive/dead) metric. Of course, maybe that’s the substance of a separate paper and earning another publication for the research team…

          General rule in process mgt is that when processes are not well understood, then upfront inspection data can assist decision makers in issues involving subsequent remedial action. The essential question seems to be whether MS provides useful info for subsequent decision-making, and at what cost.

          For studies that show no relationship between inspection findings and long term binary process outcomes, it does not necessarily follow that the inspection has no utility.


    I wish you’d post the video discussing the differences between mortality and survivability.

    One line in your piece really resonated. You mentioned people did not want to cede control. I have been mulling over the stiff resistance and women insisting that mammograms saved their lives contrary to the study results. It is simply too scary to admit that early detection doesn’t save lives.

    I have metastatic breast cancer and a neighbor upon learning it actually said “but breast cancer is always curable when detected early, didn’t you get mammograms”? I coolly told her that I found a lump and my mammo was negative. But the implication that I am at fault for having breast cancer goes back to control. We can control this. Except when we cannot. Cancer doesn’t do what we want it to no matter if we insist on annual mammograms (despite research demonstrating how imprecise and unnecessary), how positive we remain, how much money we raise for awareness, or how much tumeric or garlic we ingest.

    Mammograms don’t save lives. There, I said it. Let’s move on to research that focuses on which tumors grow aggressively and which do not. Yes, I’m the choir and I’m just repeating what you wrote. But after reading a comment again saying “mammogram saved my life…” I had to write.

    Thanks, your writing is clear eyed to me.

Comments are closed.