Many Medical Screening Tests Wasteful
Posted on December 29, 2016
Screening tests are very often bad medicine for the same reason that dragnets are generally bad policing, which is the same reason why drug testing is generally bad for society. It’s called false positives.
People have it in their heads that scientific tests are perfect. That is never true. Every test for a discrete condition comes with two nonzero error rates: false positives and false negatives.
False negatives are somewhat bad. That means you missed a disease that was actually present. However there will be other chances to detect it.
False positives are usually worse. That means you think there is a disease when there isn’t.
Best case scenario for a false positive: you do a second test that is more accurate and more expensive and more time consuming. Usually you find out it was all a mistake, after a scary period of worrying about it.
Worst case scenario: the second test scores another false positive, so you go ahead and have an unneeded operation. Worst of the worst:maybe you needlessly die under the knife–another patient killed by screening.
Maybe you think it is worth the risk, because lots of times you catch a killer disease early on and cure it. Think again.
Here comes the hooker: any disease you are screening for is actually kind of rare. Let’s say 1 in 1000 women of age 50 have undetected but serious breast cancer, and the screening test has a 10% false positive rate. Then a single screening of 2000 women leads to 200 false positives. Suppose there is a false negative rate of 50% (the reality is far worse). Then the same screening leads to 1 actual cancer detected. That means that 200 woman have to undergo unnecessary procedures such as biopsies or worse. In some cases there will be unnecessary operations. Meanwhile the one true cancer will probably be treated successfully–but that could have happened even without the screening. (Moreover, a large share of detected cancers would never have actually killed the host, leading to a lot of unnecessary operations.)
After 10 years of annual screening, a majority of women will experience false positives followed by a certain amount of unnecessary grief and iatrogenic risk.
Similar comments apply to PSA tests for prostate cancer.
Screening has a place when there are other risk factors. Far too often it is a poorly justified procedure.
There is another kind of routine testing that resembles screening but is typically different in three respects:
1. It doesn’t detect a discrete condition, but rather makes a measurement that has a wide range, so that small measurement errors are not important.
2. The measurement directly gets at a possible disease, rather than a correlated symptom.
3. The test is relative simple and inexpensive and can be repeated often.
Examples include blood pressure, blood sugar levels, cholesterol levels, obesity index.
Also, confirmation tests are another unrelated case, because they don’t suffer from the same kind of lopsided error quantities you get from screening an entire population. Thus you might have strep-throat-like symptoms and take a strep test and get a false positive and then take some unnecessary antibiotics. However only a tiny share of the population ever gets into that situation because we limit strep tests to people who already appear to have a step throat.