Etude Critique des Différentes Approches de Validation des Méthodes Analytiques

The analytical methods’ results play a crucial role in many areas on the decisions made on their basis, such as determining the quality and quantity of active ingredients of drug products, the quality of food or in assessing pharmacokinetic studies or bioavailability and bioequivalence. The reliability of analytical results is paramount in this context, as it should certainly meet the needs of end users, in order to ensure the reliability of the conclusions taken with them. To attain this reliability, the validation of analytical methods is therefore of chief importance. Validation of analytical methods is one of the essential steps in the entire process of demonstrating reliability of analytical results that will be generated by these methods in their routine use. In addition, the interpretation and evaluation of test results during this step should be based on a statistical basis. Otherwise, the importance and usefulness of the analytical results for decision making is irrelevant. We are acquainted to several analytical validation approaches in both the literature and in various regulations’ text. The main objective of this thesis is to compare these various approaches in theory and in practice, in order to determine the reliability of the decisions taken by each of them. To achieve this goal, we first redefined the aim of any quantitative analytical method and of their validation. We then conducted a comparative study of various regulations by identifying errors and confusions included in these documents and in particular their practical implications when assessing the validity of a method. Given these findings, a new approach for assessing the validity of quantitative analytical methods have been studied and enlarged a statistical point of view. It is based on a statistical methodology using a « beta;-expectation » tolerance interval that is incorporated into final decision-making tool called accuracy profile. This profile ensures that a specified proportion of the future results that will be provided by the method during its routine use is well included in the acceptance limits fixed a priori according to the users needs. For this purpose, a practical comparative study between this approach and different conventional approaches existing was performed. Then an illustration of the potential universality and versatility of this methodology has been made through its application of different analytical methods encountered in the pharmaceutical field. Also a new capability index has been proposed that is suitable for the assessment of an analytical method capability in the context of its validation. Finally, a study of certain practices encountered in laboratories was made, which involves the use of placebos as validation standards instead of actual samples from the manufacturing routine.