The Problem With Predicting Uncommon Events: The Critical Effect of Prevalence in Test Performance.

We used the Bayesian model to show the relationship between prevalence and the test's negative and positive predictive value. We used the above principle to understand the utility of biomarkers for acute rejection under different pretest probability of rejection. Given the test's sensitivity and specificity, the disease prevalence affects the predictive value of the test; the clinical decision to get any test should be considered while understanding the prevalence of disease and cost, risks, benefits of the tests, and available alternatives.