Two situations with the same operating characteristics (sensitivity and specificity) of the test, but a different prevalence of the condition in the population.

In the first example, the prevalence is 40% in the population, thus 400 of these 1,000 people have the condition.  Using the sensitivity shows that 90% of these 400, thus 360 people are correctly and 40 incorrectly identified as having the condition.  Using specificity information shows that 90% of 600, thus 540 are correctly and 60 incorrectly identified as not having the condition.  The predictive value of a positive test result in this situation is 86%.  In other words, a positive test truly identifies about 9 out of 10 cases and in the 10th the test is wrong.

In the second example, the prevalence is much lower.  Using the same test with the same operating characteristics, a positive test is correct in only half of the cases:  the test is virtually useless in this situation.

To top