The matrix approach for determining intermediate precision

Written by Dr. Janet Thode Posted in Method validation

An important aspect that - depending on the type of method - is evaluated during method validations according to the ICH Q2(R1) guideline is precision. This includes repeatability as well as intermediate precision (inter-assay precision).

Intermediate precision examines the variability of the results, which are obtained under changing conditions common in the laboratory. This includes different laboratory technicians who are performing the method, different days on which the method is done and / or different instruments that are used for analysis (e.g. 2 different photometers). To examine these different aspects, you can either consider each aspect individually or you can apply the so-called "matrix approach" (experimental design) for intermediate precision, which kills all aspects with one stone ;-)

The matrix approach is illustrated in the following table:

Experiment Operator Day Instrument *
1  1  1 1
2  2  1 2
3  1  2 1
4  2  2 2
5  1  3 2
6  2  3 1
* 2 different photometers / HPLC systems / etc.

It consists of 6 experiments, in which 2 technicians carry out the experiments over 3 days using 2 different instruments. The sample is analyzed at 100% target concentration. For evaluation of the intermediate precision matrix approach, mean, standard deviation (SD) and relative standard deviation (RSD) are calculated and finally the relative standard deviation is compared to a previously defined acceptance criterion (e.g. RSD must be ≤ x%).

If you tweak the matrix approach mentioned above a little and focus on 6 independent analyses taking place on 6 different days and also include another factor (such as different batches of HPLC columns), the result is a Kojima design or Japanese NIHS design designated evaluation scheme for intermediate precision, which was proposed by Shigeo Kojima in 2002 [1]. It can be illustrated in tabular form as follows:

Independent experiment / day
1 2 3 4 5 6
Analyst 1 1 1 2 2 2
Equipment 1 2 1 2 1 2
Column 1 2 2 2 1 1

Today, from a scientific point of view, risk-based approaches, as described by Borman et al. in 2019 [2], might probably be a more reasonable choice than such generic pre-QbD approaches ????

 

References

[1]   Kojima S. (2002). Evaluation of intermediate precision in the validation of analytical procedures for drugs, Pharm. Tech. Japan, Vol. 18 (5):695-704.

[2]   Borman P.J., Chatfield M.C. (2019). Risk-Based Intermediate Precision Studies for Analytical Procedure Validation, Pharmaceutical Technology Regulatory Sourcebook eBook 2, 12-22