Perhaps there is a simple ansawer to this...

I am currently validating a qualitative GC-MS screen method and am facing a question that troubles me. The LOD is determined under repetition conditions on one day. Firstly by estimating 3*noise and then bu runnnig a spiked sample x number of times. However, the precision is also determined with a study over days, and I am afraid that the GC-MS fluctuates som much over days that the precision will be quite large compared to the LOD. Is this really OK?!

Thanks :-)

I am currently validating a method for GC-FID, and eventhough i don't use an MS, i am very (TOO) familiar with those terms.

If you fear the difference is big, then measure both?

I use the "aantoonbaarheidsgrens" (don't know it in english) which is determined as 3times the STD Deviation of spiked samples, 1 sample a day (reproducability), minimum 5 samples.

Depending on the source of your statistic, LOD will be defined with repetition conditions or reproducability conditions. In school i saw it as repetition, now i use reproducability.

And i don't want to trouble you, but eventhough both systems are allowed, the "1 sample a day" system is preferred, because this gives a better representation of what the real value is.

But now to answer your question: In my method the difference is acceptable.
With repetition i get 0,25µg/ml, with reproducability i get 0,44µg/ml. Seems a big difference, yet if you consider that my linear range for this method starts at 4,27µg/ml and the method is most often used to check whether there is more then 50µg/ml, then you know that difference doesn't really matter.

So basically i'm saying: compare your LOD to what values you have in your linear range and what you are values you need to determine.

Hi

Thankyou for your very fine answer. I was kind of thinking that the measure including the day to day variation to determine the LOD would be the best. I am however surprised that the difference is so small in your case. PErhaps the FID is a little more "stable" than the MS-detector, which can be sensitive to contaminants but also the inlet due to our dirty samples. I must be sure to get extremes of apparature conditions in the validation, in order to get the correct limit of detection and values for the system suitability.

Mogens

Could be an FID is more stabile, don't really know that.

But in my method the reproducability is normal, the repetition value on the other hand is a bit high, according to VITO.be it should be closer to 0.2 µg/ml

Good luck ;)

Tom

correction: closer to 0.1 ;)