ICH guidline ask for the justification of presentation of new drug substance, to do so we should follow the ICH Q1(E) guideline,
This guideline is destinated to evaluate data stability
indeed, this guideline don't give many informations about the way of using statistical tests to discuss self life of the drug
my question is : there is much test like ANOVA and others statistical test which we could use to discuss self life of drug, but the use of those tests required much atention
when and where we should use those tests to verify most important change and how to justify the variability following two condition : accelrated and long term condition
Will try to find the answer. Currently I am teaching stability study to b.pharm students.
Hi,
in my opinion the matter in ICH Q1(E) is not _testing_ but _estimation_, that is not evaluating p-values but building prediction (or tolerance) intervals. Be aware they are _not_ confidence intervals (if needed googlize "tolerance interval").
A possible way to a solution could be :
1 - check the shape of the evolution of concentration of drug with time to choose an appropriate relation between concentration and time;
2 - perform an appropriate (according to 1) regression of concentration over time;
3 - assess prediction (or tolerance) limits of the estimated line/curve for a required/desired confidence level - and again _not_ the usual confidence limits;
3.5 - a one sided limit could do the job but I would not recommend;
4 - compute the dates of intersections of the prediction limits with the accepted level of concentration of the drug - usually graphical assessment will do the job, exact computation could be cumbersome;
5 - the lower bound of this last interval is the desired shelf life assessment.
I can give you more details on the techniques in R, but data mandatory.
Room temperature data and accelerated are treated the same.
Best.
Dear Jean Maccario
thank you for your fruitful answer
Indeed am working too on estimating measurement uncertainty based on tolerance interval Iam too interesting if you can help to do with R
There is another thing: when it's possible to justify the shelf life of a drug.
For example, if we would like to follow the logigram as stated in the Annex A of ICH Q (1E), we began all the time with the accelerated result, however the accélrated result couldn't give a realistic estimate of the concentration variation with time, due to quantity (fewer) results (following the ICH Q(1A)
Regards
I would be glad to help but, as said, some data are needed.
Relative amounts of active drug are enough, real names unnecessary.
You may send privately by a message inside RG, but to keep it public will probably be of some interest to other people.
We will need at least three time points, the fewer you have, the wider intervals (confidence or tolerance) will be.
Best.
Hi, in additions of nice comments above, I think it will be nice for you, if you can read newest Guidance's of the USP 41;
please read , and for evaluating the selectivity and statistic of your work.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis