Sensitivity and specificity belong to the test. PPV and NPV belong to the population it is used in. Change the prevalence and see how the meaning of a result shifts.
| Disease | No disease | Total | |
|---|---|---|---|
| Test + | |||
| Test − | |||
| Total | 10,000 |
The same test, applied in every possible setting. The dashed line marks the current prevalence.
A ruler that applies Bayes’ theorem: a straight line from pre-test probability through the likelihood ratio lands on the post-test probability.
How to read it. Mark the pre-test probability on the left axis and the test’s likelihood ratio on the middle axis. Extend the line to the right axis to read the post-test probability.
Why a straight line works. All three axes use log-odds. Because post-test odds = pre-test odds × LR, on a log scale that becomes addition: logit(post) = logit(pre) + ln(LR). The middle axis sits halfway between the other two, so a straight line does the sum for you.
Rule of thumb. LR+ above 10 or LR− below 0.1 usually changes clinical decisions. LRs between 0.5 and 2 barely move the probability.