In recent months, various professional organizations have called for reducing the use of routine medical screening tests (PSA screening for prostate cancer; mammogram for breast cancer; etc.). These guidelines are informed by statistical analyses.
But a lot of people find the idea of less screening counter-intuitive, even unpalatable. For example, Skip Lockwood, President of the Project to End Prostrate Cancer, lamented:
The whole concept that you would do anything to reduce the amount of information you have does not make sense to me.
There is indeed something of a paradox. What's wrong with screening everybody? I address this issue in Chapter 4 ("Timid Testers / Magic Lassos") of Numbers Rule Your World.
Here's the short version:
Diagnosis tests are not foolproof: some proportion of those testing positive will in fact not have the ailment while the tests will fail to detect some of those who have the ailment. False positives lead to over-diagnosis, unnecessary procedures, and potential harm due to side effects. False negatives give people a false sense of security, and if detected too late, something like cancer may have become too advanced to treat.
By definition, most patients who take a screening test are healthy. Dr. Ablin told us, for instance, only 3% of American men die from prostate cancer. If 97% are not at risk, even a tiny false positive rate, when multiplied by millions of test-takers, will result in a boatload of false positives.
But no test can distinguish between a false positive and a true positive (if we know who is a true positive, we don't need a test, do we?) So these tests produce a boatload of positive results, only a small portion of which are true positives. This is why false positives are a huge problem.
Since a high proportion of those who test positive do not have the disease, what the screening test gives us is unreliable information. And having unreliable information can be worse than having no information.
When a screening test is first rolled out, it is usually recommended for higher-risk populations. If the screened population is subsequently augmented, the newly screenable patients have a lower risk of disease than those meeting the first criterion. In other words, healthy patients are disproportionately added into the screening pool, and a good proportion of these will receive positive test results erroneously. This is why more targeted screening works better than broad-based screening.
Because prostate cancer is even rarer in young men than in older men, if young men are routinely screened, almost all the positive results will be false positives. This is why statisticians want such tests to be more targeted, less broad-based.
This is a complex issue and this is as short as I can make it. For a fuller discussion, read Chapter 4.
I also highly recommend Dr. Ablin's article in the NYT: he focuses on another shoe to drop concerning PSA tests, that is, evidence that PSA screening has no health benefits at all, and that the amount of PSA is not highly correlated with the presence of prostate cancer.
About $3 billion are spent on PSA screening tests annually in the States.
1. "Education should accompany prostate screening, new guidelines say", Thomas H. Maugh II, Los Angeles Times, Mar 4 2010.
2. "The Great Prostate Mistake", Richard J. Ablin, New York Times, Mar 10 2010.