About the calculation of assay or impurity

[ghant]

Dear all,
generally, we conduct the test of assay or impurity, we will prepare two samples per batch and two or one injection per sample, then calculate the average of two data and report it. Is there a accept criteria of precision for the two data which used to calculate average. You know we can use RSD% if there are no less than three data, but how to estimate the precision of only two data. I‘d appreciate it if anyone can share his view on this topic.

[Blazer]

If you're looking for a reference to regulation that dictates this, there isn't one. It all comes down to your company's policy. In your case, I suppose you could put a requirement in to calculate the % difference between the mean assay results of the two preps. If they are within x% or x.x% of one another, you can accept the data as valid.

[Fernando]

Dear ghant

In the case of the assay I prepare two standard solutions injecting 5 times std number 1 and 3 times number 2. The confrontation between them has to be in the range 98,0% to 102,0 %.
In pararalel I prepare two samples, with 2 injections each. Then I calculate for each sample the per cent difference. Maximun 2,0 % (of area).
After the calculations the percent difference between the two samples must not be more than 2,0 %.
I agree with Blazer that there is no regulation.

Good luck.

Fernando

[gtma]

Duplicate injections of the same solution is not common these days, HPLCs are very robust. For assay, at least two replicates are recommended. For impurities, one replicate should be sufficient for area % approach but at least two for external std approach.