The discrimination threshold: A selection criterion for analytical methods based on measurement uncertainty - Application to animal stress studies

Abstract

Many biological studies seek to confirm the effect of a treatment on the levels of chemical markers in biological tissues. This requires choosing the analytical method best able to detect the difference between basal levels and those found after treatment. We propose a new approach to calculate a criterion we call the 'discrimination threshold' (DT), and we applied it to an analytical method that we developed to determine cortisol in cattle plasma to detect stress. DT is derived from the measurement uncertainty (MU), and combines the variabilities of both the analytical method and the marker. The uncertainty of the analytical method comes from the method validation study. The marker variability (here cortisol) is modeled from a literature review. The graphical representation of DT allows estimating the applicability of the method. Our analytical method is shown to perform well when the difference in blood cortisol exceeds 18 ng. mL^-1.

Keywords: Accuracy profile; Bovine cortisol; Discrimination threshold; ELISA test; Measurement uncertainty; Selection criteria.