Molecular biomarker analysis — Determination of the performance characteristics of qualitative measurement methods and validation of methods
Introduction: Qualitative (binary) analytical methods (e.g. applied to screening tests) for use in the analysis of food or food products (including seeds of food crops) with the purpose of demonstrating the presence/absence of a given measurand in a sample should provide objective evidence that they are adequate for their intended use. A validated test method is much preferred over one that has not undergone studies to determine its accuracy and reliability for its specific purpose. These methods that yield a binary result (yes/no, positive/negative, etc.) are referred to as “qualitative” or “binary” methods. As with quantitative methods, qualitative method performance has to be characterized with respect to the concentration of the measurand. However, only two conditions are indicated in the result: either the measurand is detected (a positive result) or it is not detected (a negative result). While internationally recognized guidelines (e.g. ISO 5725-2, References [7] and [16]) have been produced over the years to harmonize the validation of quantitative analytical methods, no consensus is yet available among stakeholders on a practical implementation of the performance criteria approach to the validation of qualitative methods for use in food and food products. Conceptual approaches for validating qualitative methods classically focused on parameters such as sensitivity, selectivity, false positive rate and false negative rate, based on detection/non-detection of the measurand in the te
Scope: This document specifies methods that yield a binary result and are used for the determination in food or food products (including seeds of food crops) of the presence of molecular biomarkers. These methods are typically used where the measurand is expected to be present in very small amounts and concentrations at the limit of detection (LOD). Methods are validated in terms of the probability of detection (POD) and of the precision of the POD. They do not rely on the concept of false positive/false-negative results, or the concept of LOD. However, inferences about the precision of the classical LOD can be made. This document describes the extent of method validation. The annexes provide different statistical models that can be considered depending on the analytical method, structure of data and statistical experience. This document does not apply to quantitative methods that are used to make a detection decision by comparing the value of a response to a cut-off value using a quantitative method, where the methods are validated by using quantitative statistics on the responses. This document also does not apply to microbiological test methods, starch, essential oils or quantitative methods.