Validation of quantitative analytical methods by means of the accuracy profile

Analysts are now very familiar with method validation and, over the last 10 or 15 years, have acquired a good deal of experience in this field. But why has – method validation become such an important concern for analysts –? The answer lies in the development of quality assurance in laboratories.

The very principles of quality assurance are well known and codified in very general standards published by the International Organization for Standardization (ISO). They emphasize that the quality of products or services supplied by a company – such as test results – is intended to satisfy a customer's needs. Quality assurance is anything that can individually or collectively influence the result of an analysis.

Conceptually, it is the set of measures to be taken to ensure that the results supplied will be of the quality required for the end-user's intended use. In practice, depending on the laboratory's field of activity, each manager can draw on a number of standards that provide more specific guidelines for organizing quality assurance. Today, there are three main standards governing the life of analytical laboratories:

1. The ISO/CEI 17025 standard, which leads to accreditation, i.e. verification of a laboratory's competence and ability to carry out analyses; this is a very general standard, of interest to many types of laboratory;

2. Good Laboratory Practice (GLP) applies to laboratories involved in the development of any new chemical molecule with potential effects on human health or the environment, such as drugs, pesticides, etc. It is a mandatory requirement for pharmaceutical laboratories;

3. The ISO 15189 standard, which supplements the "Guide de bonne exécution des analyses" (GBEA) designed specifically for medical laboratories and specifies how they can be accredited.

All these texts assert the need to validate methods, but without explaining how to go about it. For this reason, numerous guides, recommendations, industry standards and guidelines have been published to help analysts carry out validation studies. Unfortunately, they often contain inconsistencies, both in the vocabulary used and in the statistical methods proposed, and experience shows that the procedures generally proposed pose a problem when it comes to making the final decision. So the practical question of how to validate a method remains a matter of debate.

This article is a proposal for a universal, more coherent and operational approach based on a single tool: the accuracy profile. It applies to quantitative analysis methods only.