Document Type


Publication Date


Published In

American Journal of Medical Research


Diagnostic testing, Criterion standard test, Statistics, Receiver operating characteristic, FDA, Net economic benefits


Given a certain technology or procedure for diagnostic testing, different cutoff points produce different sensitivity and specificity rates. The cutoff point that generates highest sensitivity and specificity establishes the Criterion Standard Test (otherwise known as the Gold Standard Test). If, subject to good reason, a new testing technology or procedure emerges, the optimum cutoff point associated with it may generate higher sensitivity and specificity and thus a new improved Criterion Standard Test. Various cutoff selection methodologies have been proposed, all based on Euclidean geometry, involving the so-called Receiver Operating Characteristic (ROC) curve. Our purpose in this paper is to recommend a new selection methodology based on the P-Value associated with the well-known Pearson’s chi-squared test (χ2) – the conventional test utilized when testing for dependence between state of nature (disease present or not present) and evidence (test positive or negative measures). Using a hypothetical numerical example, we demonstrate that the cutoff point associated with the lowest P-Value of the Pearson’s chi-squared test is the one that maximizes sensitivity and specificity, or overall accuracy, thus establishing the Criterion Standard Test. Although the best geometric method (sums of squares) and the proposed method are equally effective in selecting the optimum cutoff point, only the proposed new procedure selects based on statistical significance. Additionally, we propose a simple theoretical benefits / costs linear setting to discuss the importance of net benefits associated with testing accuracy and reference harmful as well as beneficial testing cases found in various literature sources.




The definitive version of record can be found online at:


© 2017 Addleton Academic Publishers