false negative observations
480 true negative observations
400 persons with X
600 persons without X
Positive predictive value = —a— = -true positives (360)-x 100 = 75%
Thus, 3 out of 4 of the persons with positive observations really have the disease, and 1 out of 4 does not.
By a similar calculation, you can determine the probability that a negative observation is a true negative. The results here are reasonably reassuring to the involved patient:
Negative predictive value = —d— = —true negatives (480)— x 100 = 92%
As prevalence of the disease in a population diminishes, however, the predictive value of a positive observation diminishes remarkably, while the predictive value of a negative observation rises further. In Example 2, in a second population, B, of 1000 people, only 1% have disease X. Now there are only 10 cases of X and 990 people without X. If this population is screened with the same observation, which has a 90% sensitivity and an 80% specificity, here are the results:
Example 2. Prevalence of Disease X = 1%
Disease X Present Absent
Was this article helpful?