Testing of APIs for NDSRI

I am searching reference document for implementation of skip testing for the NDSRI, if results of NDSRI is above 10% of specification limit and below 30% of specification limit.
If someone can help to provide reference / strategy for the same, it will be great help.

Dear @SachinG,

You can find the strategy in EMA Q&A Q9. Is that right?


Dear Sir,
Thanks for your reply. The information provided in the Q&A is valuable, however I am searching to establish some strategy by that we can decide:

how many batches testing to be done before going for skip testing and
at what frequency ( i.e after how many batches we should perform testing of NDSRI in API).


The EMA guideline indicates that you should test 3 batches or 10% of the yearly production. Whichever number is the highest.

Now, for the skip testing strategy itself, it should be considered the same approach mentioned previously as this varies depending on the yearly forecast.

Finally, for the frecuency, that is a QA decision. Should you test at the end of the year, or beggining, mid, end of each year? You need to balance the decision. For example, if you test at the end of the year and for some reason one batch was above the 30%, I think you will need to re-test all the batches produced in that year and from that moment onwards do routine testing. To again do skip testing you will need to modify your process to have an steady trend < 30% and provide quite a lot of data backing your statement.

If one batch is above the 100%, that would be tricky and will need some regulator to be involved for further decisions.


Dear Diego,
Thanks for your response. I could now make my strategy.