Sensitivity and specificity, are not the only metrics we use for diagnostic tests. Another pair of metrics valuable in evaluating the performance of diagnostic tests, consists of the positive predictive value and the negative predictive value. Let's keep talking about Mary, the 50-year old woman I've mentioned before, who has done a mammography to assess whether she has breast cancer or not. As we discussed, when comparing the true disease status with the result of the diagnostic test, we can have four possible combinations. True Positive, False Positive, False Negative and True Negative. You have already seen how to calculate sensitivity and specificity using the true disease status as the denominator. What if we use the test results as a denominator? The proportion of positive tests that correctly identified diseased individuals, is called Positive Predictive Value. In other words, positive predictive value is the proportion of True Positives among all positive results. The proportion of negative results that correctly identify non-diseased individuals is called Negative Predictive Value. An equivalent expression is that negative predictive value is the proportion of true negatives among all negative results. These definitions reflect a population-based view of the diagnostic test results. Another way to look at it, is from Mary's perspective. If a test result is positive, the positive predictive value is an expression of the probability that she has breast cancer. If a test result is negative, the probability that she does not have cancer is equal to the negative predictive value of the mammography. You have learned earlier, that sensitivity and specificity are characteristics of the test itself, and do not vary according to the prevalence of the disease in the population. This is not true for positive and negative predictive values, they heavily depend on the prevalence of the disease. For instance, even an excellent diagnostic tests, with 100 percent sensitivity, and 99.9 percent specificity, can yield a positive predictive value as low as 50 percent, when the disease is very rare with a prevalence of 0.1 percent. A positive predictive value of 50 percent means that half of the positive tests are wrong, which is a pretty terrible outcome. The same test when applied to a population where the disease prevalence is 10 percent, yields a positive predictive value of 99 percent. Positive and negative predictive values provide information about the effectiveness of a test within a specific context. They also help you interpret the result from the point of view of the individual who undertook the test, which is critical both in clinical work and in policy decisions.