Never order screening tests on populations with low prevalence of your target disease.
Suppose the prevalence of disease “Q” is 10% in Population X, & 0.1% in Population Y.
Suppose the Specificity of the diagnostic test for “Q” is 99% (very high, compared to most of our tests)
Test 1,000 persons in Population X:
- 100 test Positive because they have “Q” [10% prevalence]
- 10 test False-Positives (99% test specificity = 1% false-positive)
- So out of 110 Positives, 91% are true-positive, 9% false-positive
Test 1,000 persons in Population Y:
- 1 tests Positive because of “Q” [0.1% prevalence]
- 10 test False-Positives (99% specificity of test)
- So out of 11 Positives, 91% are False !!!!!
Therefore, NEVER screen persons with low likelihood of having the disease, because the vast majority of your “positive” results will be false! Maybe your initial screening test was cheap & easy, but subsequent work-up for all the false-positives may be expensive & invasive. In the 1970’s, some big companies performed screening treadmill tests for executives (white males, but otherwise young & asymptomatic). Treadmills have relatively low specificity. All the (invariably false-) “positives” went to cath (with maybe a 1 in 100 mortality back then???).
And what’s a test? All those “Have You Ever Had…” review-of-systems lists we give to asymptomatic patients in the waiting room. It’s more than a paper & pencil, because all the “positives” generate time & effort [detracting from more important concerns], & maybe other [worse] tests.
In terms of tuberculin testing (PPD / IGRA), a false-positive generates 9 months of INH Tx, with attendant risk of (potentialy fatal) drug-induced hepatitis. Worth the risk (approx. 1 in 200 for the hepatitis if <35 y.o.) for possibly true-positives (i.e. at-risk patients), but not for the many false-positives among persons with no risk factors for Latent TB Infection.