Update: see also this story.
The U.S. Preventive Services Task force (whatever that is), in the Annals of Internal Medicine, yesterday advised women not to begin mammogram screenings for breast cancer until 50, and then conducting them only every other year and not annually, as had been recommended previously.
It’s about time. But why the change?
Docs have finally woke up to the effects of false positives. Mammograms are far from perfect test instruments, and a false positive is when the mammogram says you have cancer when you truly do not. These happen all the time with mammograms, especially in younger women who are not at special risk (no family history of cancer, no smoking, etc.). That is, mammograms are far too often saying that women have cancer when they truly do not. And when that happens, that costs something.
The worst false positive cost for many women—though it might not seem like it until it happens to them—is the stress and anxiety that follows a positive mammogram. “Oh, no! I have cancer!” is what most women think after they are told that “The mammogram found something.” Days, even weeks can go by until further tests can be taken—biopsies, CT scans, X-rays, and so forth which carry their own risk of infection and radiation exposure. Meanwhile, many women are consumed with fear that this is it!
But it is almost always not it. A standard calculation, done in every introductory statistics class, shows that the actual probability that you have cancer given the mammogram “found something” is only about 9%—only 9%!. That’s how crappy mammograms really are (for healthy, not-at-risk females).
What other costs are there? (from the paper)
False-positive results are common with mammography and can cause anxiety and lead to additional imaging studies and invasive procedures (such as biopsy or fine-needle aspiration)…Biopsies may occur as a consequence of false-positive mammography results…Anxiety, distress, and other psychosocial effects can exist with abnormal mammography results…Other potential harms, such as pain caused by the procedure, exist…
Overdiagnosis can occur when screening detects early-stage invasive breast cancer or DCIS [ductal carcinoma in situ] in a woman, typically older, who is likely to die from another cause before the breast cancer would be clinically detected. Overdiagnosis can also occur in younger women if a detected DCIS or other early-stage lesion never progresses to invasive cancer…Similarly, unnecessary earlier treatment can occur at any age when screening detects a slower-growing cancer that would have eventually become clinically apparent but would never have caused death.
Radiation exposure may increase the risk for breast cancer, but usually only at much higher doses than those used in mammography, although regular mammography could contribute to cumulative radiation doses from additional imaging for other reasons.
How to make a decision whether to have a mammogram
Regular readers will know that I have written about this subject often. For those who are mathematically inclined, I have a formal paper on this that appeared in Biometrics a couple of years back. A PDF is available here.
I also have a Decision Calculator that you can play with. This allows you to input your own false positive and false negative costs and discover whether mammograms are worth it. The example data built in is for healthy (American) women, aged 40 to 60-ish, undergoing their first mammogram screening. Women who have additional risk factors (like family history of cancer, who smoke, etc.) should consult their doctors.
I point out that the Naive-o-gram—a device which proclaims every women does not have cancer—beats the success rate of accurate diagnoses of professional mammograms (for the healthy population of which we speak). And therefore, the Naive-o-gram is a viable alternative to a mammogram. In other words, don’t bother with a mammogram unless your costs (of false positives and negatives) are very different than most women’s.
What distresses me about the operatives of the right (and the left) is their penchant for grasping onto every bit of news that bobs up and then finding a way to interpret it negatively. I am already hearing reports that these new mammogram guidelines are the result of Obama’s new health care system. “Rationing!” goes the cry. “They are preparing us for limited availability of care now!” John Gambling on WOR radio (710 AM, NYC) is saying these things as I write.
Nonsense, of course. But we can suppose it’s only a matter of time before conspiracies of the left are touted. Drug companies—no, insurance companies—are trying to find a way to reduce costs and they don’t want to pay for all those mammograms for women!
Wait and see if I’m right.
Update: see also this story.
A close friend worked on a mammography research project in Britain almost 30 years ago. When she left the project, they were inclining to the conclusion that mass mammography was a bad idea, largely for the reasons you outline above. An added problem was that the women most at risk were those least likely to volunteer for mammography. A few years ago I read a very good summary of the issue, referring to that research project and others, in Gerd Gigerenzer’s book on “Risk”. I then tried to explain the problem to a group of colleagues, all of whom had been subjected to heaps of Higher Education, but they seemed to dismiss the concept of screening being deleterious as some sort of capitalist-fascist plot.
Since I only have a very small chance of having a dog in this fight, (the chance of my moving from the UK to the US times the chance of my getting breast cancer*), I’m probably fairly neutral in opining that the timing of this announcement was unfortunate.
By all means have a debate on the merits of Breast Cancer screening, or ‘preventative’ medicine in general – the debate is doubtless overdue – but in the current climate when the Obamacare, (or is it PelosiCare), legislation is high on the agenda, it may be hard to separate the signal from the noise. Whilst I would favour personal choice, coupled with personal responsibility, on the matter, the dearth of unbiased opinion, (except in rare cases like this site), would make that choice difficult for most people.
* The risk of a man developing breast cancer over his lifetime is about 1/1000
This is why I keep reading this blog. Best post on the net for this subject. How do we get to the point were the decision is would you pay for this out of your one pocket? High Deductible and Health Savings Accounts.
In general, I agree with you about over-testing, but I think that Joe Triscari’s makes a good comment on your previous post.
Rather than focusing just on costs, doesn’t it make sense to take into account overall quality of life (QOL)? If I give relative quality of life ratings in the mammogram example (adapted from your ratings):
(PosTest,Presence of Cancer)=-150 [cancer is pretty bad]
(NegTest,Presence)=-210 [cancer worse if we think we don’t have it]
(NoTest,Presence)=-190 [not quite as bad as (NegTest,Presence) ?]
(PosTest,Absence)=-70 [needless anxiety]
(NegTest,Absence)=8 [peace of mind?]
then I get an expected 1.216 QOL by taking the test, and -1.52 QOL by not taking it. So it seems to make sense to take the test. Essentially, even a small positive value for (NegTest,Absence) results in an improved quality of life because the associated probability (.922) is so high.
You seem to assume that the correct comparator is test vs no test as opposed to “test for A” versus “test for B” versus “engage in behavior C” versus …? (Test for B could be Matt’s “wait until you fall into a higher risk category.”)
Also, aren’t you simply restating the precautionary principle – via the numbers you assign? The precautionary principle also treats undesirable events in isolation, i.e., no trade offs.
You didn’t have to wait for long. Heard on the Ron Owens program in the Bay Area at 9 AM that this is a way for insurance companies to save money.
There seem to be a lot of men commenting here. I don’t see a calculation that is very important for me when deciding if I should (at 46 years old) get a mammogram. Add this into the QOL assessment: Every woman I know who’s had the test says it’s excruciatingly painful to squish your breast flat. Women would rather risk breast cancer than have this test.
I’m “fortunate” enough to have access to all the news channels at work, and I was blessed to see a bunch of different stories on this issue.
What was most amazing was how all the radiologists were saying that they are against less testing.
Ari, Gigerenzer is very clear. Screening mammography is recommended for their patients by German radiologists, but not for themselves or their spouses.
Kind of like opthamologists (why does Firefox think this isn’t a real word?) who perform Lasik but don’t get it done?
There’s a financial consideration in these things?
Ari, possibly because it’s spelt Ophthalmology?
Sorry, I’m a congenital smartarse.
Gigerenzer’s book is very good. Highly recommended.
Ari: 0, Kevin 1
I concur, the risk of false positives in tests, especially where mass screening is used is not yet being subjected to sufficient scrutiny, perhaps because people like to think better safe than sorry and politicians like to be on the side of GOOD. Well it must be good mustn’t it?
Older generations might have felt that sufficient unto the day was the evil thereof and going looking for problems was an invitation for them to turn up and bother you: and the right time to deal with them was when they did. This was also a common view amongst many UK doctors of the 1950’s: why worry the patient about by what might be when there wasn’t much effective treatment anyway?
My late father was an eminent medical man of his day, Fellow of the British Medical Association and all that and when serious statistical anaylsis of drug trials, the double blind etc. became commonplace he was intensely interested and turned his very able mind to the large scale survival data in certain types of carcinomas which was for the first time becoming available partly due to the National Health Service which compiled them aross the whole country.
His mathematical knowledge was quite good, he knew all about basic statistics but hey when you have a son at university studying physics and mathematics and with access to computer core time, remember that? why not rope him in? And he discovered something interesting.
Now at that time the Ethical Committee of the General Medical Council, of which he was a member, did not allow any comparative experiment to test different treatments for grave conditions, except in very unusual circumstances, and even then doctor and patient had to seek approval from the local ethical committee. I don’t know how it works today.
But since different doctors had different views as to the best treatment and the patients too, it did allow compilation of the anonymous data as to the success rate of the different treatments elected. Most importantly in the case of breast cancer at that time many women refused all but palliative treatment.
So he had available data of sufficently large numbers of the survival of patients which he divided into three categories, those who had radical treatment, those who had minimal treatment, then known as lumpectomy, and those who had none.
Unfortunately the data wes deficient back then because there was no way of knowing when the patient presented with symptoms, which made comparisons of survival rates over time, life expectancy, a bit problematic and so made it hard to determine optimal treatment, nowadays called the gold standard. As a result the methods were changed so the doctor recorded his assessment of how the patient presented, and thus how advanced the disease was. My late father never lived to see that data: which I suspect probably look pretty crude to modern eyes.
What he did spot, and published, was that although the survival rate across all three groups looked pretty like a normal distribution it was slightly skewed, there was always a long tail, between 1 and 2% of long term survivors, and these appeared to have a death rate with age close to normal life expectancy of that time.
His view was, and I thought then he was right and indeed still do, that these long term survivors had been misdiagnosed and never had the disease in the first place.
Of course all that was fifty years ago, today in some ways we are far more advanced, or as the gentleman said ‘When I was young medicine was cheap. safe and didn’t do much, today it is expensive, dangerous and often works.’
Well sometimes the magic works and sometimes it doesn’t. But even today there is still a problem with the medical profession and its offshoots over statistical knowledge.
In part I think that speciality has created highly knowledgeable technicians who do not consider the patient in more general terms as an old fashioned physician would do. In the UK the profession has tried to address this with the idea of having a more general doctor in charge of overall patient management.
But in part I think it also is that medicine and quackery are and remain closely entwined, and I do not mean to insult the medical profession, it is simply that every day they deal with people who often expect more of them than they know they can do: and that colours their view.
The fact is that people like to believe in miracles, and the charlatans and mountebanks have always played on this. And any doctor, even today has something of that in his make up: it goes with the territory as I believe you Americans say.
Make no mistake I believe in modern medicine. The latest quackery from diets to fashionable herbal remedies, to the wisdom of the Ancients whether pyramidic or otherwise, all pass me by.
But then of course my problem is I could have been a pretty good charlatan myself, I never did become one, stage illusion is as far as ever I went down that road, but I can still watch them in action and admire their skill and art.
Amen, sister. Horrifying. Especially if it leaves a mark.
Not quite. There’s more to it when taking into account costs for correct tests (I have another paper on this). Plus, what Bernie said. But, unless your screen name is completely misleading, I’d recommend not taking the test no matter what.
Haha! I love it! I am now officially a political prognosticator.
Radiologists not in favor of fewer referrals and fewer billing hours. As you say. Shocking.
Aren’t you a sweetheart.
My dear wife had regular Mammograms because she was in the “at risk” category (family history of cancer). A Mammogram showed something and a subsequent biopsy confirmed it. It has been operated on and we are all very happy with the outcome.
I guess there is always an exception to the rule.
My wife is now encouraging our daughters to undergo regular Mammograms from age 40. One of them is a health professional (Pharmacist) and agrees.
Having Cancer in the immediate family is sobering and makes you focus on lifes priorities.
Clearly, those who have family history of cancer are not in the same category as healthy, otherwise risk-free young women. Those with a family history have a much greater risk and therefore more frequent checking makes sense.
I now hear a doc regretting that “patient advocates” were not represented in the Annals study. This is an unfortunate request. The call should go out for those who had suffered from false positives, true negatives, etc. Not just those who had the disease. You cannot just hear from the patients, you need to hear from the non-patients as well.
This is such an emotional matter that some people do not want to hear anything which says that more testing is not needed. I have heard from a radiologist who says she regularly turns back women who are not even 30 years old. They pine for the mammogram “so they can be sure.” This happens so often that it should tell you that emotions run very high, and that some people will be unwilling to listen to any counter arguments.
Men do get breast cancer, though it is rare.
Of course, nobody recommends mammogram screening for men. I suppose this indicates that people do understand that at some point the cost of screening outweighs the benefit.
I am a breast cancer survivor. I’m not writing to tell my story or opinion on this matter. However, I am highly interested in this debate considering it is a personal issue. While researching everyone’s views and looking for information I came across Dr. Len Lichtenfeld’s blog. He is a doctor of the American Cancer Society. Today Dr Lichtenfeld is hosting an online chat at noon ET on his blog. I suggest you all check it out. May clear up a few questions and what not. Here’s the link
Steve, I expect you’re right that at some point “people” do understand, but I’d expect that most only understand in financial terms, which is usually interpreted as the state saving money at the patient’s expense. Convincing patients that for their own good it may be better not to screen is another matter.
So let the patients and the insurance companies figure all this out on their own, as it is a financial and statistical matter, and get us physicians out of the loop on saying yes or no for SCREENING (vs diagnostic) mammograms. Diagnostic mammos, by the way, are when you have discovered a palpable mass and need to see what it is … cyst, tumor, or whatever.
Eliminate asking us to say no and risk all the malpractice liability when we are inevitably sued under the current tort system when a denied patient shows up with a malignancy.
Why should any sane physician take such a risk if the patient insists?