Graphic representation of numerically calculated predictive values: an easily comprehended method of evaluating diagnostic tests.

Abstract

The authors present a brief BASIC-language computer program that calculates positive and negative predictive values from the known sensitivity and specificity of a test over a range of disease prevalences and produces easily comprehended graphs of these relationships using commonly employed graphics software. These graphs permit rapid appreciation of the effect of prior probability on predictive values. They may facilitate selection of the appropriate test for the anticipated setting and avoidance of the unexpected inaccuracies that often result when tests are chosen on the basis of sensitivity and specificity alone.