The formula
P(A|B) = P(B|A) · P(A) ÷ P(B)
What it means
The answer depends on how many people have the condition before taking the test at all. If few do, the healthy are so many that even a small rate of false positives produces more false positives than true ones. That is why an excellent test on a rare condition returns, for most of its positives, healthy people.
How to work it out by hand
- Count how many of every thousand have the condition: that is the prevalence
- Of those, how many test positive: that is the sensitivity
- Of the healthy ones, how many also test positive: those are the false positives
- Divide the true positives by all the positives
What is worth knowing
The way to grasp it without formulas is to count people rather than juggle percentages. Of a thousand people at 1 % prevalence, ten have it. The test catches nearly all of them, say nine or ten. But of the nine hundred and ninety healthy ones, 5 % of false positives is nearly fifty more people. Total: about sixty positives, of whom only ten really have it. One in six. That is why rare conditions are not screened for en masse, and why a positive is almost always confirmed with a second, different test.