Women with breast implants have nearly 40% greater risk of dying from breast cancer: study
Source: NY Daily News
A new study in the British Medical Journal examines 12 previous studies of women diagnosed with breast cancer published since 1993. Researchers found women with cosmetic implants had a 38% greater risk of dying from breast cancer than those without, and a 26% greater risk of being diagnosed at a later stage, when cancer may be more difficult to treat.
The implants aren't causing cancer, evidence suggests -- rather, having them inside the breast may make suspicious growths more difficult to spot during routine mammogram screenings.
Breast implants are radio-opaque, researchers wrote, and may hide some of the breast tissue from sight.
.......
"Although it shows reduced breast cancer survival rates among women who had breast implants at the time of diagnosis, the findings are not conclusive and more research is definitely needed before we can fully understand the effect of breast implants on survival," Dr. Caitlin Palframan of the U.K.-based charity Breakthrough Breast Cancer told The Guardian, adding that women should make sure their health practitioners know they have implants prior to getting a mammogram.
Read more: http://www.nydailynews.com/life-style/health/women-breast-implants-greater-risk-dying-breast-cancer-study-article-1.1332000
longship
(40,416 posts)This article is horribly misleading. It fails to account for the base rate of breast cancer and whether this increase is relative or absolute. Most likely it is relative, meaning it would be a 38% increase in the percentage of breast cancer deaths. But such an increase could be a very small increase in actual cases if the base rate is small. In fact, depending on how the study was done, its size, and other parameters, a 38% increase of relative risk may not even be significant.
This article is extraordinarily bad reporting even if this statistic is significant, which I doubt. It exploits people's math ignorance, specifically their ignorance of statistical significance and the difference between an absolute difference and a relative difference. The former speaks of cases; the latter of percentages. There's a huge difference between the two.
There just isn't enough information in this article to determine whether these findings are important. Plus, a single study is rarely enough to establish a causal link.
I would relax about this one until it has a chance to be vetted by other researchers.
La Lioness Priyanka
(53,866 posts)sometimes the article is really misleading about the actual study
its not implying causality, mostly saying that the implants makes cancer harder to detect
longship
(40,416 posts)No. Didn't read the study.
Almost certainly this is a retrospective study, and as you correctly point out, it is not normally possible to assign causation from such studies.
But, no matter what, my complaint stands. I would bet that this is relative risk which may or may not be significant.
I am bandwidth limited here. Maybe somebody can provide more information on the underlying study and comments from other researchers.
Thanks for responding.
La Lioness Priyanka
(53,866 posts)longship
(40,416 posts)The study is based on Cochrane reviews (usually reliable) and meta-analysis (sometimes questionable, depending on selection biases and statistical methods).
It is up to qualified people to vett and comment on this meta-analysis, which I am not. This may be important, but even the conclusions of the paper state clearly that more research is needed.
I am no epidemiologist, so I cannot comment on this particular study. But I can comment on the misuse of statistics in the article cited in the OP. That was the purpose of my original response here.
The take-away is that caution should be advised when reading any science reportage in the general press. Sometimes it is sensationalized, as is clearly the case here. I'll stand by my original post on that issue.
marshall
(6,665 posts)Found here: http://www.bmj.com/content/346/bmj.f2399
Keep in mind that this is a study of studies. The researchers describe the process by which they identified twelve studies spread out over a number of years that met specific criteria for what they deemed accuracy and usefulness. Their actual study is to aggregate and evaluate all of the studies.
cosmicone
(11,014 posts)If the base rate is 1:1000 in the population as a whole and they study 4,000 women (2,000 with implants and 2,000 without), if the ones without breast implants get 2 cases (1:1000) and the one with breast implants gets 3 cases, that would be a 50% increase but since it is only one extra case in an inadequate sample of 4,000, this percentage is meaningless.
Very irresponsible indeed.
longship
(40,416 posts)Certainly not statistically meaningful.
Science journalism fail here.
La Lioness Priyanka
(53,866 posts)it really makes no sense to me.
La Lioness Priyanka
(53,866 posts)A casual read of the study shows that in no way was this as simplistic as the article suggests.
This was not irresponsible or science fail, it was a responsible meta-analysis that clearly passed peer review, with statistically sig results.
this is not a case of the scientists being irresponsible but of readers not understanding the difference between the actual study, and a newspapers coverage of said study.
cosmicone
(11,014 posts)and an attempt to grab headlines. The original article is fine and honest.
The original article is here: http://www.bmj.com/content/346/bmj.f2399.pdf%2Bhtml
La Lioness Priyanka
(53,866 posts)longship
(40,416 posts)Sorry for not making myself clear.
I have now looked at the study and read parts of it, and scanned the rest. It may be okay, but I am no expert in medicine, let alone the subject of this particular paper. It may be good, but I'll leave other epidemiologists to determine that.
Thank you for your respectful replies and questions.
zazen
(2,978 posts)Well, I couldn't swear to this, but in reading all of this literature as a layperson (with BC) for seven years, it seems that larger breasts tend to have more fatty tissue in the mix, making mammograms easier to read. Breast density perhaps has the highest correlation with BC incidence, more than family history, even (though it may overlap). (Forgive the non-scientific language here.) It's not just that it obscures reading mammograms: it's the type of tissue, leading more often to what docs call "busy breasts."
So more women with smaller breasts (congenitally) and higher density might be overrepresented among women with implants.
Also, although the article touches on sociological/demographic issues (that women who can afford cosmetic surgery will tend to be screened more than those who can't), it doesn't question whether certain ethnic groups (at least in North America) tend toward breast implants more than others. If those groups already have higher risk factors, then they may be overrepresented in this sample.
Sunlei
(22,651 posts)and prone to cancers in constantly compressed tissues. Not even to mention surgery scar areas as totally changed tissues or some of the implant materials, the outer covers that are known carcinogenic.
RainDog
(28,784 posts)is why I love DU.
thanks, all of you.