Minimal number of runs for validation

[EH]

I do this for API impurity method:

Linearity: PQL to 120% of the impurity limit (5 points)
Accuracy: Quantitate the linearity points against an external std or covered by Intermediate Precision
Intermediate Precision: 2 analysts, Triplicate prep at each level (PQL, 100%, 120%)
Specificity: forced degradation unless you have known impurities
Robustness: column temp, flow rate, mobile phase organic content
Solution Stability: std and sample (3 time points)

[ntruong]

Hi Bart,
Yes, the linearity curve should consist of five levels from 1/2 LOQ to 120% of impurity limit (one measurement and single injection for each level). Why overkill? To determine RRF, take the slope of this curve and divide by the slope of main compound (5 levels from 50% to 150%). Report one decimal place for RRF value more than one (1.2, 2.5, etc) and two for any RRF value less than one (0.96, 0.83, etc.) per USP report requirement.
Hi EH,
Do you characterize the API impurity? If so, you could spike the limit concentration into the 100% level of main compound and ensure that all peaks are well resolved. Why force-deg?[/quote]