Chromatography Forum

delete thread

Does this happen for every level of the calibration for the drugs that are affected ? If not, which levels are affected ? How does the linearity (r-squared) look ?

Peter

I second Peter Apps' questions. It sounds to me like the calibration is not quite linear. But I'd want to know more details before really making a guess as to what's going on.

It depends on what regulations you're under, but some different things you try do are limit the calibration range (if your calibration is 1ppm - 100ppm, but all samples are below 20ppm, get rid of all cal points above 20), change the weighting of the calibration curve to 1/x or 1/x^2, or simply use a curve that fits better than linear. As I'm in a research environment, I just go with the latter and pick the curve that gives the best fit. In my view, that's the most accurate, and better than messing with data until you can slap a linear curve on everything.

As an aside, sometimes a good R^2 value can be misleading. You can have a curve that appears to be linear at first glance and gives a good R^2, but is very off at low levels. Very relevant if you are doing any trace analysis where your samples tend to be at the low end of the cal.

THANKS FOR for the replies, I was able to get help from outside the forum to solve issues on the problem compounds.

Hello,

I suppose you're using the esi probe... could you tell me what source setting did you choose for drug of abuse analysis?

thanks

when I run my curves, the calculated conc for the calibrators is significantly less than the the theoretical conc.

This sounds very much like either you are not using the right model for your calibration curve, or that something (another analyte, instability, binding to walls of autosampler vial, etc.) interferes with the analysis of the problematic drug.

Try running a calibration curve with only the troublesome drug. Check residuals to make sure it is linear throughout. R^2 is nearly worthless for this if your calibration range is >2 orders of magnitude. If it isn't linear, you need to figure out why.