Chromatography Forum

All,

When validating an HPLC method for impurities, is there a general industry standard for agreement between two analysts for ruggedness?

For most of the methods I've worked on, we've set agreement to +/- 10%, but this is based on an impurity with a decent chromophore spiked at about 0.1% w/w.

I now have a method where the impurity limit is much lower than normal and the impurity in question has a very low RRF (peak areas ~1000) . Is it reasonable to widen my acceptance criteria for agreement of have we been locked into 10% because of history? My concern is that with such a low response, any difference would be greatly magnified.

My first thought here would be to use a more sensitive method but those of you in the industry know that sometimes we have little lattitude to change this...

I thank you in advance for a quick response, deadlines are looming.

Jeff

Two conflicting arguments...

The acceptance criteria are set out in the validation protocol. If the validation meets these criteria then the method can be deemed robust. As long as the criteria are laid out in the protocol and those criteria are met then the method passes validation. The question is what numbers do you stick in the protocol.

Or...

A method validation robustness test is there to prove the method is robust and the data you get is accurate and precise. If the method is not robust (ie you have greater than 10% between analysts) then it cannot be considered validated and hence the data obtained from that method is no good.

Thanks for your response. I'm looking at this from a "What numbers to put in the protocol" standpoint. I'm not sure the method will pass 10% because of the low peak areas involved and I wonder if 20% is acceptable in this case? Or is it unreasonable to set acceptance criteria that wide?

GoldLeader,

The short answer is: yes, you can use acceptance criteria other than +/- 10% especially when dealing with impurities.

You should choose the acceptance criteria from a scientific and a practical standpoint. Of course, your QA group will have the final OK on it.

You have options. One of them being to use an absolute value rather than a relative one. You can have an acceptance criteria of +/- 10%, relative. You could also chose to have +/0.02% w/w absolute.

0.10 %w/w +/10% relative gives a range of 0.09 to 0.11 %w/w

0.10 %w/w +/- 0.02% absolute gives a range of 0.08 to 0.12 %w/w

Regards,
Dan