Monday, November 23, 2009

On Mammograms

While this is in theory a political blog, you may or may not be aware that I'm actually far more interested in public health. These two interests are tied together in a lot of important ways. Sometimes this is a good thing, as public health concerns can (and absolutely should) inform political decisions.

Other times, this is a bad thing. Public health is based on science, whereas politics is rarely even based in reality. But since the two are tied, public health recommendations often have political ramifications, and sometimes lead to political controversies. That's what we're seeing right now with the USPSTF's recent recommendations that screening mammography is not indicated for women under age 50.

It's worth noting a few things up front. First, this is only about screening mammograms. That is, mammograms performed on healthy women with no symptoms of breast cancer. If someone has a lump, they should have a mammogram no matter what age they are, and this recommendation doesn't change that. These recommendations also don't apply to women at increased risk, such as those with a strong family history of breast cancer. They're only for healthy, low-risk women.

Second, this is hardly a new topic. It's been debated for years and years, and while I didn't expect recommendations like this right now, I'm not a bit surprised that this is the recommendation. It's where the science points, and the science has been pointing that way for quite awhile.

Third, the recommendation to not begin screening mammograms until age 50 is pretty commonplace in Europe already, and has worked well in those countries. It's not like the US is doing something out of the ordinary, it's just joining much of the rest of the industrialized world.

Those elements out of the way, let's look at a pretty terrible editorial written by Martin Schram (a syndicated columnist who thinks H1N1 is called H1V1, so you know he's an expert on health matters), and published in the Sentinel. It's a load of made-up numbers, pointless fearmongering, and bad arguments. So let's take it apart.
A federal task force of experts had indeed reversed years of government guidance on breast cancer exams -- announcing that most women younger than 50 should not get "routine" screening mammograms every two years, after all. Nor should they examine their own breasts for lumps.
Yeah, the USPSTF is not a federal task force. Here's what they are, according to their own website:
The U.S. Preventive Services Task Force (USPSTF), first convened by the U.S. Public Health Service in 1984, and since 1998 sponsored by the Agency for Healthcare Research and Quality (AHRQ), is the leading independent panel of private-sector experts in prevention and primary care. The USPSTF conducts rigorous, impartial assessments of the scientific evidence for the effectiveness of a broad range of clinical preventive services, including screening, counseling, and preventive medications. Its recommendations are considered the "gold standard" for clinical preventive services.
In fact, the USPSTF's recommendations include this disclaimer:
Disclaimer: Recommendations made by the USPSTF are independent of the U.S. government. They should not be construed as an official position of the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services.
Pretty straightforward, that. The AHRQ is a federal agency, the USPSTF is an independent body that gets some funding from them. Maybe this is just splitting hairs, but Schram is not starting off well. He continues:
The reason given by the U.S. Preventive Services Task Force, a panel chosen and funded by the federal government [Unicow: Was this clause stolen from the Wikipedia article?], is that these examinations are no longer considered statistically effective. But it sounds as if they really mean the costly mammograms are not cost-effective. The exams save women's lives, the panel acknowledged -- but just not as many of them ages 40 to 49 as women older than 50.
Schram is hearing things. He may think he knows that they "really mean" something about cost-effectiveness, but the head of the panel disagrees:
But Ned Calonge, who chairs the 16-member panel, defended the recommendations and denied that cost or the debate over health-care reform played any role in the decision. "Cost just isn't a consideration when the task force deliberates," said Calonge, who is also the chief medical officer for the Colorado Department of Public Health and Environment.
That's as it should be. The panel is concerned with the evidence, not the cost. Which isn't to say that cost concerns don't matter, merely that they're not part of the deliberations of the USPSTF, and their report only mentions financial costs very briefly and in passing, not as an element of the recommendations. There are other groups who deal with cost, the USPSTF deals with evidence only.

Now is where things get ugly.
Here's a close look at the statistical mind-bending that the task force pursued to arrive at its stunning conclusions:

The task force concluded that the risk of breast cancer in women age 40 to 49 is very low -- and there is a significant risk that a mammogram can produce false-positive results that can lead to unnecessary biopsies that can be disfiguring. For every 1,000 women who are screened annually starting at 40 years old, according to the model studied by the task force, a mere 0.7 women would be saved from dying of breast cancer. An estimated 470 other women would receive a false-positive result and 33 would be given unnecessary biopsies.
These numbers are a little weird right off. I'm not quite sure where he got them, because the recommendation statement gives the "number needed to invite for screening" to prevent one breast cancer death in the 39-49 age group as 1904 (CI, 929 to 6378), which is closer to 0.5 than 0.7 per 1,000 women. Maybe Schram has a different source I'm not seeing, or maybe he's just playing games with numbers. I certainly can't think of any other reason he'd talk about "0.7 women" instead of giving the more-understandable "it takes X screenings to prevent one death" measure.

But let's just assume that wherever Schram got his figures is correct, and maybe just measuring something slightly different than what I was looking at. What's not at all correct is what Schram then does with the figures. "Statistical mind-bending" doesn't begin to cover it.
But here is another way journalists can crunch the numbers to breathe measurable lives into that task force's 0.7 statistic: The last census showed there are 22,617,241 women in the United States aged 40 to 49. Now ask: What if all the women in America age 40 to 49 were given annual mammograms?

According to the task force's model, 15,832 American women would be saved from death by breast cancer. Yes, by that model 10,629,990 women might receive false-positive results and suffer the anxiety that surely would cause. And 746,361 women in America age 40 to 49 might undergo unnecessary biopsies. But all of those women would still be alive.
No, Martin, this is not another way journalists can crunch the numbers, unless they're trying to be deceptive. The task force's model tells you no such thing.

See, you can't take statistics that say that in a sample population it took 1903 screenings to result in one life saved and then go forwards and apply that (with ridiculous specificity) to an imaginary scenario in which every woman in her 40s is screened. It just doesn't work that way.

We have an estimate of what happens now, there's no basis for applying it to an imaginary wonderland in which everyone is screened. The simple act of screening everyone would likely change those figures. Furthermore, it's a blatant attempt to serve up higher numbers. Why not report on the number of women that age who we already know benefited? Because it's not going to be nearly as high as the stuff that Schram just makes up, of course.

At this point, Schram asks an odd question:
So, isn't the real question whether women should retain the option of deciding for themselves whether they want to undertake a mammogram that could save their lives but could also cause the anxiety of false-positives and perhaps an unnecessary biopsy? Is that the choice that the women should make for themselves? Or should it be made for them by their health insurance plan?
Well yeah, of course they should retain that option. Pretty much everyone has said that. And nobody's pushing for health insurances to stop covering mammograms. Not the USPSTF, not the government, not even any insurance companies (as far as I can see). Schram claims that insurances will stop covering mammograms, but provides nothing to back up this claim, and it seems pretty unlikely to me.

Regardless, even if insurances do stop covering them (which I don't think they should, or will), that doesn't change the facts as reported by the USPSTF. And that's where the ire is being directed, for reasons that I can't understand. The facts are what they are, the recommendations are in accordance with the facts. They may not be convenient or popular, but they're well-grounded.

The real issue here shouldn't be whether the USPSTF put out good recommendations or not. In fact, that isn't the issue. Very few people have argued with the evidence supporting these recommendations, because it's quite solid. The argument seems to be that it's somehow wrong to even make recommendations, or that it's misogynistic/socialist/rationing/whatever to even make them.

It's not. They're recommendations. You can take them or leave them. Medicare and Medicaid are going to pay for mammograms the same way they currently do, as will most private insurances. And that's fine. Women who want them can get them, and that's fine too. Hopefully this will help women in this age group to put a bit more thought into whether they need them or not, which is never a bad thing.

The fact of the matter is, screening mammograms in the 40-49 age group have little impact on survival, and do have real downsides. False-positives are no laughing matter, and they're quite common. Overdiagnosis (treating something that was found on a screening mammogram but would never become clinically problematic) can cause real harm. On the other hand, mammograms can save lives, even in this age group. It's a balancing act, for sure.

Finally, there's one quote from another Sentinel story that's bugging me. It's from Leominster native Kathy DiRusso, who leads a team that raises money for breast cancer research:
The results of the study seem to say the costs and false positives of administering mammograms do not justify the number of cases found in young women, but DiRusso contested that.

"To me, one life saved justifies the cost," she said. "If it was (task force members') lives, maybe they'd think differently."
Not if they're doing their jobs right, Kathy. The evidence points strongly in this direction. If you don't like it, then point out where it's wrong. Don't unfairly paint the USPSTF as somehow not caring about cancer patients. To do so is idiotic, and for all you know there may be members of the task force who have had their lives saved this way.

But while that was irritating, it's the "one life saved justifies the cost" that's really horrible. You hear this argument all the time, almost always used when evidence doesn't support what they want supported. Abstinence-only educators talk about "if it can keep one kid from getting pregnant it's worth the cost," D.A.R.E. proponents say "if it can keep one kid from using drugs it's worth the cost." I say that's all bullshit.

It's a stupid argument, especially since the "cost" we're talking about with screening mammography isn't in dollars, it's in quality of life. Does one life saved justify it if the cost is ten women who needlessly underwent radical mastectomies? How about if it's a thousand women undergoing biopsies? How about ten thousand women suffering needless stress and anxiety due to a false positive? Or, if you really want to talk about financial costs, how about balancing the cost of paying for mostly-unnecessary screenings versus spending that same money on finding better treatments? Unfortunately, there's only so much money to go around, and you can't pay for screenings without detracting somewhat from treatment and researching cures.

Saving lives is unquestionably a good thing, but saving one life does not justify any possible cost. These recommendations are trying to balance the cost (in quality of life) versus the benefit (in lives saved), and that's always going to be tricky.

By all means go ahead and argue that the benefit outweighs the cost, but let's not do it by pretending that even the tiniest benefit outweighs significant costs, and let's not do it by mangling statistics. Better yet, how about we just leave the politics out of this one?