Linearity Validation - Best Practice?

[Rob Burgess]

When validating a HPLC method for linearity what is the usual best practice? Is it to simply inject the lowest conc. followed by the next highest conc., followed by the next highest conc. etc. or is it better (as I have done in the past) to randomise the order of injection of your linearity solutions?

[bert]

Sometimes, simple can be the best..... We inject QC samples at LOQ, medium and highest concentration level and calculate the difference between nominal and calculated concentrations. These shoulld fall within acceptable limits, pending your application. Furthermore, evaluate the difference between nominal and back-calculated concentrations of the calibration samples.

regards

[MK]

I've done duplicates by first going from low to high and then back from high to low. This challenges the method for etc. carryover.

[DR]

I usually start at ~150% of what I'd expect for assay samples & go down through LOQ & LOD.

[BenNC]

When you have that wide of a range (LOQ <0.1%? - 150%) do you still have an even distribution of points? We have recently been discussing the validity of the typical least squares linear regression when estimating the line for a wide range with uneven distribution of points.

Any thoughts?

Thanks,

[Consumer Products Guy]

I make a standard solution then take different aliquots to volume, to cover the expected range in the samples plus leeway. We take at least five dilutions (and a blank). At times we've gone high enough concentrations to document deviation from linearity.