Module 9 2021

24/03/2021

Assay design & stats analysis

Statistical methods used for analysis of assays are typically based on the assumptions that responses are: • Normally distributed - Data can be log transformed to attain normality - USP 1033 “Relative potency measurements are typically nearly log normally distributed” • Homogenous - Bias in the assay can be detected to certain extent • Independent - More often than not – assay dilutions are pseudoreplicates rather than independent

The Organisation for Professionals in Regulatory Affairs

27

27

Assay design & stats analysis

Factors that are important in any assay are: • Precision – closeness of agreement between a series of measurements • Accuracy – closeness of agreement between the value found its Reference.

inaccurate imprecise

inaccurate precise

accurate precise

accurate imprecise

The Organisation for Professionals in Regulatory Affairs

28

28

14