David Newman, from Mt. Sinai in New York, has one of the most insightful and interesting podcasts in Emergency Medicine: SMARTEM. Here’s an excerpt from one of his podcasts which talks about diagnostic testing. It’s 30 minutes that is really worth your time. I couldn’t have said it better, so listen to the man himself.
Axiom I: There is no such thing as a perfect test.
All tests are imperfect. Every test is sometimes wrong. This could be a false positive or a false negative. Each of these has implications. A false positive CT for PE commits someone to anti-coagulation. The CT itself has its risks: allergy, renal injury, cancer, etc.
Axiom II: Context trumps results
You can take a highly accurate test, but it in the wrong context and it will be useless (or worse – make you do something bad). The example he gives is testing a man for pregnancy. You get one of two options: either you find out he’s not pregnant (you don’t need this test), or you get a positive result and it misleads you (are you searching now for something that isn’t there?).
You need to understand the context in which you are applying the test! Or, don’t order things unless you know why.
Axiom III: There is a test threshold
You need a context in which you think a test is appropriate. Pregnancy testing in a man is an inappropriate context. This is a prevalence of disease at which time it makes sense to order a test. If there’s a 0% chance of having a disease for which you’re testing, it doesn’t make sense to order the test. You don’t need to order a test when the prevalence of disease is
- …below the test threshold: ordering a pregnancy test in a man. There is no chance he’s pregnant. You are below the test threshold.
- …above the test threshold: a man shot in the abdomen and hypotensive doesn’t need a CT scan of the belly. Why? The prevalence of disease (ie, badness in the belly requiring a trip to the OR) is too high. You know something bad is going on in the belly. There’s an 80-90-100% chance of having an operative lesion. They need to go to the OR. The test will just delay treatment.
You do need to get a test when it is in between the lower and upper test thresholds, the “sweet spot.” This is the right context – where a test can actually help you (and your patient).
When you are outside the test threshold, you may be more likely to harm them then help them with a test. All this being said, sometimes you just have to order a test that isn’t indicated. Try admitting a patient without a CBC.
For the med student rotation, each of you come up with a test we commonly use and a scenario which is below and above the test threshold. Bring those examples to lecture.