Is this a valid approach? (Calibration)

[Pear]

Hi,

I have an HPLC method that is using the calibration points of 0.5, 1, 2, 4, 6 and 10ppm.

My R2 is 0.999 and accuracy and precision at 10 and 4ppm are 99.9% and <1%RSD.

My problem is with my 0.5ppm value. My %RSD is still very good but the accuracy jumps to 110% with independent 0.5ppm standards; even ones made from the same stock solution as my calibration 0.5ppm.

I don't want to lose accuracy down at the 0.5ppm as this is where a lot of test samples will be but unfortunately my calibration range must be up to 10ppm as this is the upper limit for the test samples.

Is it a valid approach to make a calibration range of say 0.5-2 ppm and analyse my samples. If any sample is above my highest calibration point of 2ppm, I will re-analyse the sample with a new calibration range of 2-10ppm? If every sample is below the 2ppm calibration limit...I can just state that they pass as they're obviously below 10ppm?

99.99% of samples for this method will sit around 0.5 - 1ppm so it seems silly to lose accuracy at this level as the upper calibration points seem to be causing this inaccuracy?

Thank you for any help.

[tom jupille]

What happens if you add another calibration point below 0.5 (e.g., 0.2)?

[Pear]

What happens if you add another calibration point below 0.5 (e.g., 0.2)?

Hi Tom,

I had thought about rejigging my calibration series from:
0.5, 1, 2, 4, 6 and 10

to

0.25, 0.5, 1, 2, 6 and 10

6 and 10 are the most important as these are the "warning and action" limit i.e. 6ppm is a warning things are getting high and 10ppm is fail.

But I thought if I was seeing 0.5 getting reduced to 110%~ accuracy, 0.25% may be even worse as if I changed this ,during method validation I wouldn't be doing accuracy and precision on 0.5ppm as I am just now, but 0.25ppm as my lowest new calibration point.

[Multidimensional]

To be valid, a calibration range should cover and go beyond (~ 20%) the expected range needed. It appears that your current calibration range does not extend far enough for the ACTUAL samples. Extend the complete calibration range and include enough standards to insure reproducible and accurate results at ALL of the concentrations that are part of the "Calibration".

[DR]

UV detection for LC is generally good for about 7 orders of magnitude for dynamic range.
You're flirting with 2.
Try larger injection volumes or slightly less diluted solutions. Either should shore up that low end of your curve.

[wss]

What's your detection limit? If I have a wide sort of concentration range, I usually do a couple of points per order of magnitude. One recent method I reviewed had a similar looking cal that the analysts had tacked a low cal point onto. I revised it to 0.01, 0.05, 0.1, 0.5, 1, 5, 10. You could jump up a bit depending on your LOD and do 0.1, 0.5, 1, 5, 10, 50, 100 or similar. Easier to make up standards that way too.

But I think you at least need to add another cal point below, maybe 0.1 or 0.2.

[anionman]

What happens if you take the 10 ppm out of the calibration? I used to run fluoride from 0.1 ppm to 8 ppm but the curve gets a bit non-linear over 4 ppm and it would hurt recovery at 0.1 ppm. I could solve it by using a quadratic calibration but that is not allowed by the method so I only calibrate to 4 ppm now. Do you weight the calibration? That helps recovery at the low levels a lot. The lab where I work has always had a ± 15% acceptance criteria for our reporting limit standards because it has always been hard to see the lower levels.