Method validation calculation of the Decision Limit

[carol]

Dear All,

I am trying to validate a new method for substances with LMR and I would like to calculate the CCα and de CCβ. However, I keep funding different ways of doing it.
According to the Decision 2002/657 states “The corresponding concentration at the value of the decision limit plus 1,64 times the standard deviation of the within-laboratory reproducibility equals the detection capability “ How do I calculated the standard deviation of the within-laboratory reproducibility?? Because I have prepared six replicates of samples at three different concentration 0.5*MRL,1*MRL and 1.5*MRL.
Any comment or help will be very useful,
Thanks in advance
Carol

[James_Ball]

U.S. EPA used a similar calculation where you do the replicate injections at a level not more than 5x the expected detection limit, taking the replicates through the complete sample preparation process. You then take the Standard Deviation of those injections * Student-t value for n-1 injections. I normally use 7 replicates so my calculation is Standard Deviation * 3.14 = MDL(Method Detection Limit). If the MDL calculates to less than 1/10 of the spiking level of the replicates then it should be performed again at a lower spiking level.

The problem I have encountered with this method is that it usually gives me results that are too low and if I inject a blank spiked at that level I can not see it. I like to double check by using an Instrument Detection Limit which is finding the amount that when injected will give a Signal to Noise ratio of 3-5, then combine that with a preparation efficiency where I take a higher level spiked blank, take it through the full preparation process and determine a percent recovery. If you then take Instrument Detection Limit / (Percent Recovery/100) you will get an accurate idea of just how little you can detect from the sample. This isn't a method that is published or accepted but it will work as a good check to your statistically calculated detection limit.