Linear range

[jtreacy]

This might seem a very basic question but...............
If you've validated your method is linear- say between 5 and 1000ppm. You run some calibrants - say between 10 and 100ppm. Your samples come out at say 7.5ppm or 120ppm - is it Ok to quote these figures given that you know your method is linear over that range?
I've had some difficulty with some QA people over this.

John

[MK]

What I have seen, it is acceptable to use single point standards in the middle of a proven linear range. I think it is more important to repeat the standard after 4 or 6 samples than to run multiple points.

I think you should be fine.

[Johneg]

If you have proven that the method is linear, then a single point calibration is suitable. For the best accuracy, it is better for the single point concentration to be similar to your expected sample concentration.

I agree, that replication of your calibration injection is important to confirm that your system is stable throughout the run.

[jtreacy]

Actually I'd never do a single point calibration. What if the intercept was non-zero? I was just talking about a multi-point calibration where the samples were just outside the calibration but not the validated linear range. I've had difficulty explaining that to QA people. But I suppose if they understood science they'd never have gone into QA in the first placed would they? The SOP wielding fascists!
John

[tom jupille]

At the risk of sounding pedantic, tell your QA people that results within the validated range are, by definition, valid.

That said, I am always more comfortable with interpolated than with extrapolated values. If I were regularly getting high or low values, I'd add an additional calibratant or two to bracket the wider range.

[jtreacy]

That approach probably best describes what I do.
Apologies to our brothers and sisters in QA for my strong remarks yesterday........... It must have been the beer!
John