Positive predictive value vs Negative predictive value
How to tell Positive predictive value and Negative predictive value apart when a board question makes them look alike.
PPV asks disease given positive; NPV asks no disease given negative, and both shift with prevalence.
Side by side
Positive predictive value is the probability a patient has disease given a positive test and rises as disease prevalence rises.
Negative predictive value is the probability a patient lacks disease given a negative test and falls as disease prevalence rises.
Positive predictive value is the probability a patient has disease given a positive test and rises as disease prevalence rises.
Negative predictive value is the probability a patient lacks disease given a negative test and falls as disease prevalence rises.
PPV = TP/(TP+FP), the chance a positive test is truly disease; PPV rises as prevalence rises.
NPV = TN/(TN+FN), the chance a negative test is truly no disease; NPV rises as prevalence falls.
Watch it explained
The repair move
Before naming the answer, force the stem through the separator: PPV asks disease given positive; NPV asks no disease given negative, and both shift with prevalence.
Try a free question
CoreA new troponin assay is used to evaluate chest pain in an ED. In this population, the prevalence of acute MI is 25%. Of 1000 patients, 250 have MI, of whom 230 test positive (TP=230, FN=20). Of 750 without MI, 30 test positive (FP=30, TN=720). What is the POSITIVE PREDICTIVE VALUE (PPV) of the test in this population?
Show answer and explanation
Answer: A. PPV = TP/(TP+FP) = 230/(230+30) = 88.5%. PPV = TP / (TP + FP) = probability of disease given positive test. PPV depends on prevalence: at higher prevalence, more positives are true positives, so PPV rises. At lower prevalence, more positives are false positives, so PPV falls. Distinguish from sensitivity (intrinsic test characteristic).
Why the other choices are tempting
- B. Sensitivity is intrinsic to the test (TP/(TP+FN)).
- C. Specificity is intrinsic to the test (TN/(TN+FP)).
- D. NPV is the probability of no disease given negative test = TN/(TN+FN).
- E. Accuracy is overall correctness.
Flashcards
Tap a card to flip it.
What clue supports Positive predictive value?
Choose Positive predictive value when the stem gives this pattern: Positive predictive value is the probability a patient has disease given a positive test and rises as disease prevalence rises.
What clue supports Negative predictive value?
Choose Negative predictive value when the stem gives this contrasting pattern: Negative predictive value is the probability a patient lacks disease given a negative test and falls as disease prevalence rises.
Quick answers
What is the fastest way to tell Positive predictive value from Negative predictive value?
PPV asks disease given positive; NPV asks no disease given negative, and both shift with prevalence.
What is the key clue for Positive predictive value?
Positive predictive value is the probability a patient has disease given a positive test and rises as disease prevalence rises.
What is the key clue for Negative predictive value?
Negative predictive value is the probability a patient lacks disease given a negative test and falls as disease prevalence rises.