Relevance of Predictive Value

The concept of predictive value helps us with interpretation of a particular test result in a particular patient. We can look at both negative and positive predictive values, but positive predictive value is usually more helpful. Positive predictive value of a test result is the proportion of positive test results that are true positives. In other words, positive predictive value tells us the probability that the patient actually has the disease if our test is positive. The positive predictive value decreases when looking for a disease in a low prevalence population. Looking for a disease in a low prevalence population is commonly referred to as ‘looking for a needle in a haystack’. A real life example of this is the PKU example above - you get swamped by false positives, even though your test has a very high specificity. Certain diseases like respiratory syncytial virus (RSV) are looked for more effectively during high prevalence season rather than in low prevalence season. Let’s go back to the numerical example we used for calculating predictive value. Note how Positive Predictive Value increases when you test in a high prevalence population.

Another example of the importance of prevalence, if you did a Rapid Strep Test (RST) on all children in a school, there would be many more false positives because the prevalence of disease is very low in children who feel well enough to go to school. False positives are also high for RST when it is performed during seasons (e.g. summer) when there is a high prevalence of viral pharyngitis and a low prevalence of Strep pharyngitis. An RST performed only on symptomatic children in a season when Strep pharyngitis is common would have fewer false positives. To increase the positive predictive value of a screening test, a program could target those at high risk of developing the disease, based on considerations such as demographic factors, medical history, or occupation. For example, mammograms are recommended for women over the age of forty years because that is a population with a higher prevalence of breast cancer. These examples illustrate the importance of prevalence in determining the predictive value of a given test result, but sensitivity and specificity are still important. The more sensitive a test, the less likely an individual with a negative test will have the disease and thus the greater the negative predictive value. The more specific the test, the less likely an individual with a positive test will be free from disease and the greater the positive predictive value. Diagnostic tests performed to confirm a suspected diagnosis must have a higher specificity and so a higher positive predictive value as compared to a screening test. Negative predictive value tells us the probability that the person does not have the condition if the test is negative. This parametric looks at the likelihood that a person will not have the disease when looking only at those people who test negative.