Chromatography Forum

Dear all,
if someone can help me with this problem with residual solvents...
I know that pharmacopoeia allows RSD for replicate standard injections of 15%.
If I have a sample- first probe of 10 ppm and the second probe 17 ppm, it is more than 15% allowed (limit is 100 ppm).
My question is-what is RSD allowed for two samples? Considering that system suitability is OK (RSD of replicate standard injections below 15%).
How do I treat RSD when it comes to residual solvents?
How big can be the difference between 2 samples (RSD allowed)?
Thanks

Dear all,
if someone can help me with this problem with residual solvents...
I know that pharmacopoeia allows RSD for replicate standard injections of 15%.
If I have a sample- first probe of 10 ppm and the second probe 17 ppm, it is more than 15% allowed (limit is 100 ppm).
My question is-what is RSD allowed for two samples? Considering that system suitability is OK (RSD of replicate standard injections below 15%).
How do I treat RSD when it comes to residual solvents?
How big can be the difference between 2 samples (RSD allowed)?
Thanks

You can calculate RSD of the two variables (10 & 17) using excel. Then keep one variable constant, eg 1st variable, calculate 2nd so as the RSD meet the given value.

If you have two samplings of the same sample ( or am I misunderstanding your post, ) then you cannot do a RSD with only two data points.

If would be a very homogeneous sample to have an acceptable RSD at such low levels.

If you have reproducible values from multiple injections of the same sample then your data can be good, it just shows that your bulk sample is not perfectly homogeneous.

I hope I did not misunderstand your post.

Rodney George

Thanks for your answers.
My problem is generally with RSD of two samples. I have just put one example. Another example-one probe 1000 ppm, and the second probe 1200 ppm. Should I consider my results OK or is there some problem because the difference of 200 ppm between the probes. Limit is 5000 ppm for this solvent (maximum allowed concentration).
So, how do I know how big the difference between probes can be? For head space, i do 2 probes and i calculate the average.
With liquid injection (residual solvents), I do 2 injections per probe (2 probes)-it means 4 results and i calculate the average.

The issue that regulators will address is the reproducibility of your method. You use the word 'probe' as I would use the words 'test result'.

You should demonstrate that when you do multiple probes of one sample preparation your results are within a defined limit. For example, if you prepared a solution of dissolved drug that was known to contain 1000 ppm of methanol and you tested this single solution three times the RSD of your test results should be within say.... 10%.

This demonstrates that your method has a suitable reproducibility. Now if you were to prepare three different solutions of an untested drug sample and you performed a single test (probe) on each of the three solutions and got results with a mean of 2000 ppm that gave a RSD less than 10% you would have a high confidence level that the measured level is accurate and that your sample is homogeneous.

But if your results from the three probes had a mean of 2000 ppm but the RSD of the results was 30% then you would conclude that your sample was not homogeneous or your method of sample preparation was not reproducible.

Going back to the beginning, if, when you tested a single solution three times and got results with a RSD of 30%, then your method itself is not reproducible.

You have to show your test method is reproducible, your preparation of a probe solution is reproducible, and the bulk sample is homogeneous for any meaningful results to be valid.

best wishes,

Rodney George
consultant

I am not sure of the validity of averaging results from two different techniques for a final result, I don't think many regulators would accept that as a valid average.

As stated above, the RSD of a set of 2 in not valid statistical measurement, a minimum of three values using the same technique is required for an RSD value.

How reproducible are the values from liquid injection? If those values are within 5% and your headspace values are not within 15%, there is a issue with your headspace method. If this is the case, there are many threads on this forum dealling with headspace analysis.