Masterstudiengang "Drug Regulatory Affairs"

Master-Thesis

Development of a rationale for deriving acceptance criteria for the validation of analytical methods in the pharmaceutical quality control

Dr. Katrin Neubert (Abschlußjahr: 2007)

Language: English

The type of method (HPLC, ELISA), its area of application (assay, examination of purity, release, stability testing), as well as the specification and variability of production of the finished medicinal product are relevant for the acceptance limits of the characteristics of a method's parameters.

Established parameters for the description of the performance of an analytical method are specificity, linearity, accuracy, precision, and, for purity methods, additionally the limit of detection, limit of quantification, and reporting level. As part of the examinations for robustness, it is shown to which variabilities of the operative method's parameters it yields reliable analytical results. As a scale for the precision, the variance of the individual results is generally given. Fundamental for the knowledge of the acceptable variabilities of analytical results during routine work is, however, the confidence interval of the variance.

The upper limit of the confidence interval of the variance can be derived directly from the variability of the production and specification. The acceptable variance for the individual analyses that results from the acceptable confidence interval for the variance is directly dependent on the sample size. An unrealistically high sample size for routine work results very quickly from this purely mathematical approach.

Mathematical considerations, in particular model calculations and simulations, are suitable to improve the understanding of relationships that can be described mathematically.
The band width remaining for the analytical variability results from that required by production. A production that demands a bandwidth of ±5% requires an analytical method that is only possible theoretically with a variability of "0."

In the interest of a routine, smooth release code of practice, the available bandwidth should be divided up in isolated cases properly between the production and the analytical technique. Based on this, the required sample size can then be calculated using the conventional variance.

Pages: 43