By Anonymous on Thursday, October 30, 2003 - 03:48 am:
Dear all,
In my impurity estimation method validation , some impurity linearity plot passes through zero but in some is does not, though the R is more than 0.999 in all cases. What explanation should i assigne to this.
Pls help.
By Anonymous on Thursday, October 30, 2003 - 06:08 am:
You have the impurity being measured in the substance OR you have an interference with the impurity that elutes at the same time. It is possible that you have a small amount of curvature (non-linearity) of response, but without seeing the data I could not state so with confidence. Do you see anything in the blank chromatograms which matches the impurity? How much away from zero intercept are you? Within experimental error of the confidence level?
Rod
By Tim on Friday, October 31, 2003 - 04:11 am:
If you have a "blank" that has the same matrix as the impurity solutions and it has no visible impurity peaks, you could add a zero value to your plot.
Have you tried forcing the line through zero and see what effect that has on your R value?
You also have to consider the accuracy of impurity estimation - are your data points based on just one reading, multiple readings of different solutions, multiple readings of the same solution, the average of multiples? All can contribute to experimental error that will affect your linearity.
By Lima on Friday, October 31, 2003 - 04:43 am:
I've attended a seminar about method validation ( by USP )and they said linearity curve should not pass zero.R value is not enough to evaluate a linearity curve. You should consider about std. deviation as well.It should be close to zero.Also you may check if all values on the linearity line.
By Tom M on Friday, October 31, 2003 - 01:30 pm:
http://www.lcresources.com/discus/messages/3282/3214.html?WednesdayFebruary2620031201pm
By Anonymous on Friday, October 31, 2003 - 10:45 pm:
Hi,
some more detail ,
Linear regression fails to pass through zero within 95% confidence. Range of the y-intercept, at 95% confidence, is 0.3712 – 3.7495.
Is it really must to pass it through zero?? as we are working at LOQ level, allowed RSD is 10.
By Anonymous on Monday, November 3, 2003 - 12:34 am:
do you have linearity points below LOQ?
what happens if you discard these and calculate correlation and intercept again?
I think this could be justified by saying: linearity was effected from the poor repeatability of a solution below LOQ?
By Anonymous on Saturday, November 15, 2003 - 11:32 am:
Anonymous: I am working on impurities validation as well. Accuracy samples are duplicate spl at conc. 0.05%,0.1%,0.5%,1.0%, 1.5%,2.0% of working conc. which is 100 micro grams/ml. How would you prepare the samples? by preparing stock spl at 100 microgram /ml and diluting it to get the final concentration? Please suggest.
By Anirban Roy Chowdhury on Sunday, June 13, 2004 - 09:07 am:
If you don't have zero intercept then then it means that that at zero conc there is some absorbance(probably matrix).Try to compare the absorbance of placebo at same conc of the sample solution. it may be due to that.As we reach LOD levels the interferance of other compounds increase.
But if you are using pure compound for linearity as is usually the case. Then may be Lamber Beer law is not applicable at such low concentration or in other words there is some non linearity.