Hello,
In USP 233, Spike Recovery experiments are required to evaluate Accuracy of Instrumental Methods and the acceptance criteria is 70-150%. How was this range decided? Why not 69-151%? If I have a Spike Recovery outside of the acceptance criteria range by 1%, how significant is this?
Similarly, the Relative Standard Deviation (RSD) of the concentration of 6 replicates run on Day 1 should be no more than (NMT) 20%, whereas the RSD of the concentration of 12 replicates across Day 1 and Day 2 should be NMT 25%. Who decided the acceptance criteria and what is the justification for the published criteria?
Thank you in advance
I don't know exactly "who" wrote USP233, but I suspect the answer in general is related to the Horwitz Trumpet. See The curious Horwitz function (rsc.org):
https://www.rsc.org/images/horwitz-function-technical-brief-17_tcm18-214859.pdf
I expect that for the elements of interest in USP233 (most likely Pb, Hg, Cd, As and so on), the typical concentrations are so low that a recovery between -50% and+50% is deemed acceptable, based on many accumulated tests of compounds and elements at similar concentrations in different matrices in the past.
There is probably a (simple) statistical test to determine how probable any particular recovery (-51%recovery vs 50% etc) is from the expected mean (a Z-test perhaps?), but I'm not the person to answer that with certainty.
As to the %RSD for 1 day vs 2 days, it is probably a statistical calculation where for n=6 20% is a 95% Confidence Interval, and for n=12 over n=2 days, the interval is expanded, but not in a linear fashion.
If anyone with more statistical knowledge can explain, I'm sure it would be welcomed.
Re-reading the Horwitz function paper and the USP233 articles, I don't know why the USP has an asymmetrical recovery (-30% and +50%), while the general form of the "trumpet" is drawn with equal + and - recoveries. I think it might rest with the statisticians to explain that fully.
- Badges
- Science topic
- Similar topics
- Human-Computer Interaction
- Collaboration