4.9. Summary

Throughout this chapter we have proposed ways to understand, estimate and interpret various criteria required for assessing the validity of an analytical method. These criteria must be reported and the code used to compute the criteria must be documented according to best practices. Regardless of the complexity of computations and models, the objective of an analytical method (are the measurement errors acceptable?) should never be forgotten and should remain the primary focus. The information needed to make a decision is contained in the measurement error profile. All performance criteria—linearity, accuracy, precision, limits, measurement errors—can be assessed using a graphical profile and can be easily understood and interpreted by an analyst.

Another important remark related to the decision is that, if the tolerance intervals are within the acceptance limits, all of the required criteria are guaranteed to be met. The opposite, however, is not true, i.e., even when all of the performance criteria are satisfied, the measurement errors will not necessarily be acceptable. Although it is common to assume that good methods will always produce good results and most regulatory documents were written in this spirit, it is important to remember that only the opposite statement holds true: good results can only be obtained with a good method.

4.9.1. Terminology

Analytical procedure (or method or assay). Written procedure which describes all the means and the operating procedures required to perform the analysis of the analyte. That is, field of application, principle and/or reactions, definitions, reactants, equipments, operating procedures, expression of results, suitability tests, test reports.

Analyte or activity. The analyte (or activity when relevant) is the matter of the analytical procedure. The analyte is a physical entity (e.g., water activity), chemical entity (e.g., active substance alone or in a pharmaceutical formulation, total lipids, aspartame, lead) or biological entity (e.g., atpmetric activity). In the case of quantitative analytical procedures of farm and food products, "analyte" is equivalent to "measurand".

Matrix. All constituents of the laboratory sample other than the analyte. A type of matrix is defined as a group of materials or of products considered by the analyst as having a homogeneous behavior with regards to the analytical method used.

Blank. Test performed in the absence of the matrix (blank reactant) or in a matrix without analyte (blank matrix). By extension, the instrument response in the absence of the analyte is used (blank instrumental).

Accepted reference value. An accepted reference value is a value used as a reference, agreed for a comparison and derived from:

• A theoretical or established value, based on scientific reasons.

• An assigned or certified value, based on experimental data from a national or international organization.

• A consensus value, based on a collaborative experimental work.

• The mathematical expectation of the (measurable) quantity, in case where the previous points are not applicable, i.e., the mean of a specified population of determinations (cf. NF ISO 5725-1).

Calibration standard (or calibration sample). Calibration standards are samples of known concentrations, with or without matrix, that allow drawing the calibration curve.

Validation standard (or validation sample). Validation standards are samples reconstituted in the matrix or in any other reference material with true values set by consensus and used to validate the analytical procedure.