Friday, April 1, 2011

Overdiagnosed: What do we do when we're the one

Update, April 7 - Gail Collins' NY Times Op-Ed describes her head-spinning reaction to changes in recommendations about hormone therapy for menopausal women . . . you really have to be your own advocate.

Back in January, I mentioned the NY Times review of Overdiagnosed: Making People Sick in the Pursuit of Health by Drs. H. Gilbert Welch, Lisa M. Schwartz, and Steven Woloshin (Beacon Press, 2011, 191 pages). Now I've had a chance to read the book, and it is fascinating.

As a metrics person, I have often argued that it is important to look at people in large groups. You can accurately predict what will happen to some subset of them: ie, when the City of New York reports that 63% of children in its public schools will graduate within four years of entry, I believe them. (Well, sort of. See this New York Times story about some recent recalculations.) The trouble is, there are rarely ways to predict ahead of time which person will end up in which group. Generally, more children from higher income families graduate within four years, but not all of them do. And you can't tell, at the beginning of ninth grade, which kid will be in which group. This is one kind of problem when we're considering social or public services. It's another entirely when we're talking medical screening and health care.

Welch and his co-authors make a compelling case that new and improved diagnostic tests -- more sensitive blood tests, better scanners, decoding the human genome -- have meant that we are able to see more abnormalities and treat emerging conditions earlier. Sometimes, as in the case of high blood pressure, it's a good idea to treat an asymptomatic patient. But, Welch and Co. argue, in many cases we're treating people with mild or no symptoms. We've made an assumption that if treating early is good, than treating earlier must be better. The problem arises because if we've found an abnormality early -- before it's symptomatic --  there's no way of telling whether it will be nothing, or it will kill you.

As they go on to show in a number of contexts, including MRIs, CT scans, PSA testing, mammograms, and genetic screening, the more we test, or see on a scan, the more likely we are to find an abnormality. And when we see an abnormality, we are, for a variety of societal and cultural reasons, likely to treat it. But we forget, they point out, that the treatment may be worse than the abnormality. Or it may have side effects that set off a cascade of poor outcomes. Or, and this is hard to take in, we may have treated a cancer that was never going to grow, or not fast enough to kill the victim. Or, and this is even harder, we may be diagnosing a fast-moving, aggressive cancer earlier that was going to kill the victim in the same amount of time, or perhaps just a little bit less, than it would have without any treatment.

Who wants to take the chance, when a screen turns up a possible cancer, that it's slow growing? We are acculturated to fight cancer with all we can. But the stories Welch and his co-authors provide, of harm done by what may have been unnecessary treatment, are compelling. As are the mounting costs of health care. What's the thinking person to do? I don't actually want to be treated for a cancer that's not going to kill me, but it's hard to turn down treatment against medical advice. Fortunately, Welch, Schwartz, and Woloshin offer some complex rules, and some common-sense ones, to bear in mind:
  • To grossly oversimplify, cancer is a product of genes and environment. You can't control your genes, or even your entire environment, but you can control some important aspects of it. Don't smoke. Eat well. Get exercise.
  • Symptoms are important. They are a signal something is wrong and are the best predictor of serious problems. So in the absence of symptoms, stay skeptical. Keep asking questions.
  • Remember that while early is good, earlier is not necessarily better, and there may be costs and risks to treatment.

      No comments:

      Popular Posts