general rule for spiked amount during recovery test

[s2008]

Hello All,

I have a question about how much we can spiked during we do recovery test. Based on my experience, spiking the equal or less of the amount of target analyte in relatively complicated sample matrix usually give me bad results ( low recovery and high variability ). Spiking a few times more amount of target analyte would provide good results. Do you have same experience and how do you handle this issue?

Thanks advance

[chromatographer1]

"relatively complicated sample matrix"

Ah, the magic words !

Yes, this can happen and I believe the cause is non-homogeneity of the sample in which you spike your analyte. Subtle differences in a complex sample can affect recovery.

If you take your sample and make it as homogeneous as possible you should be able to spike from detection levels to a large percentage without suffering inconsistency.

All this is stated in general terms of course, as Murphy is alive and well.

Rodney George

[JGK]

In any recovery experiment I would always Use three levels, a Low, at the target Level and High.

If you only spike at one level, how do you deal with values away from your target?

[wipeoutxl]

Another way: You can span a wide range of concentration of spike, better if you make independent replicates, plot added vs found. If your method is bias-free, then you should get a linear curve with slope=1 and intercept=0. You can verify this statement by checking confidence interval of slope and intercept. Further, if you have a statistician/statistical package you can get an ellipsoid region for you slope and intercept, thus validating your bias-free hypothesis.

[Consumer Products Guy]

Not to intentionally hijack this thread, but for an EPA regulated material at 10 ppm active, we're soon to start a test method validation. For an active level this low, what would be the appropriate/approximate recovery levels for spiked product that would be considered acceptable? 85%? 90%? 95%? Surely, EPA couldn't expect the same precision and accuracy levels of anaytes present at levels about 1%????