It’s a holiday week. But when this morning’s paper delivered yet another op-ed by Dr. H. Gilbert Welch, citing (and breaking an embargo on) yet another, misleading and manipulative two-author analysis of breast cancer screening by him and one other scientist, I thought it worth documenting some concerns.
I’ll start by mentioning that Dr. Welch and I seem to agree on one point – that women should have access to information so they might make reasoned decisions about breast cancer screening. He refers, also, to controversy among professionals about the relative benefits and harms of screening mammography. That there is debate is incontrovertible. No argument there.
The problem is that educated, middle-aged women are being nudged, and frightened, or even charmed into not going for mammography. Nudged, by papers like the one in JAMA today, which acknowledges controversy about statistics and then suggests a falsely low range for lives saved per number of women who get screened. Frightened, by headlines that highlight the risks of overdiagnosis, a statistical concept. If a woman finds out she has an early-stage breast tumor, she and her doctor can (and should) actively decide how much therapy she should have based on the molecular subtype of her tumor, stage and other factors. Just because you find a Stage 0 or small tumor by screening, doesn’t mean you have to over-treat it. If medical education were what it should be, there would be little or no overtreatment because doctors would discuss appropriate options with women and not advise them to have too much therapy. And charmed, yes – by the false notion that breast cancer is often nothing to worry about, that in many cases it can be let alone. That it might just disappear.
I am not aware of a single pathology-documented, published case of a breast tumor going away on its own. Yes, there are slow-growing tumors that may not do harm. But those tend to occur in older women. Those cases are, in general, irrelevant to discussions of breast cancer screening in women between the ages of 40 and 60 or so. What matters most in assessing screening benefits is the number of life-years saved, which is potentially huge for women in this age bracket, and quality of life changes due to the intervention, as assessed over decades.
For today, I’ll point to just a few issues in the JAMA paper. The authors state that among 1000 U.S. women age 50 years who are screened annually for a decade, “490 to 670 will have 1 false alarm.” But as detailed in Table 2 of their paper, it turns out the range for women who undergo false-positive biopsies is far lower: between approximately 50 and 100 per thousand women, depending on the age group and study from which the authors draw the “data.” What that means, according to the numbers they’ve culled from studies of non-specialized radiologists, is that only 1 in 10 women would undergo a breast biopsy, and not have cancer, per decade of screening. So the numbers of false positives involving biopsy are not so high.
Most of the false positives are callbacks for additional imaging. Welch and his colleague talk about frequency and anxiety produced by “false alarms.” They go as far as to cite studies documenting that “anxiety may persist for at least 3 years and produce psychological morbidity…” But if women appreciated the data to support that, in most cases – approximately 85 percent – breast cancer can be removed and metastatic disease avoided, over the long haul, by early detection, most of us, and certain anyone making decisions based on reason, wouldn’t mind the follow-up and worrying about irregularities noted on a screening test. Most of us can handle the emotional aspects, and uncertainty, of screening over the course of a few days. To suggest otherwise is patronizing.
Years ago, breast cancer screening was widely considered an act of empowerment, a way for women to take control of their bodies, and to avoid the disfiguring and sadly lethal effects of late-stage breast cancer, besides the potential need for treatment until the end of life. Now, mammography is more accurate and involves less radiation than ever before. Women might be demanding universal access to better, state-of-the-art facilities, rather than shying away from the test.
As for those women who do get called for needle breast biopsies, I say that’s not such an onerous prospect. What’s key is that the procedure be done under local anesthesia, under imaging (typically ultrasound) guidance in an office by a skilled radiologist. The sample should be reviewed by a well-trained breast pathologist, and molecular studies evaluated in a central lab that routinely runs those kinds of tests.
Finally, in the end of today’s op-ed, Welch suggests that the way to reduce uncertainties about breast cancer screening is to carry out costly and somehow randomized clinical trials to see how much and how often screening is needed to demonstrate a survival benefit. But, as his tone suggests, I suspect he doesn’t really favor investment in those clinical trials.
The fact is, I don’t either, at least not for mammography at this point in the U.S. As I and others have pointed out, it takes 15 – 20 years of follow-up in a trial to demonstrate that screening and early detection reduce breast cancer deaths. In North America, the availability of mammography correlates with a reduction in mortality from breast cancer by over a third. He and others have attributed improvements in survival to better treatments. I and others would suggest that while therapy has improved quite a bit since 1985, the greatest benefit derives from most women avoiding the need for life-long treatment by having small tumors found and removed before they’ve spread. This applies in over 80 percent of invasive cases. The survival boost is from the combination, with early detection playing a significant (large) role in the equation.
Why I don’t support starting new randomized trials for mammography, besides that they’d be costly and hard to carry out, is that we can’t wait 20 years to know how best and often to screen women. Rather, it would be better to spend those theoretical research dollars in finding how to prevent the disease. If in 20 years breast cancer is less common, as we all hope will be the case, and true positives are rare, screening of the population won’t be needed. (If breast cancer rates do climb, Bayes’ theorem would support screening, because the positive predictive value of the test would, unfortunately, be higher.) Either way, by 2034 the technology would have improved, or we might have a valid alternative to mammography for screening, and so the studies would be, again, out of date.
It would be better to spend what resources we invest in mammography on improving the quality of screening facilities, now, so that women who decide to go for the procedure can, at least, know that it’s being performed with modern equipment and by doctors and technicians who are capable of state-of-the-art procedures involving the lowest level of radiation exposure possible, careful reading of the images, and application of sonography to further examine the appearance of women with dense breasts, when needed.
All for now.
I wish all my readers a happy and healthy 2014,