
USFDA 483 Torrent Indrad facility in India (FEI 3005029956); inspection from 8 April 2019 to 16 April 2019 by FDA inspector Lata Mathew, FDA inspector Jogy George, FDA inspector Zhao Wang, with four observations. One of the observation was on deficient Out of specification (OOS) investigation of API. The key points included not considering a redilution study, deficient solution stability study and not considering average of all passing results though all samples were retested after invalidating initial results.
Out of specification investigation for API products is deficient.
OOS investigation number OOS/IN/A/ST/17/013 was initiated on 03/06/17 to probe OOS results generated during assay testing at 12M timepoint (250C/60%RH and 300C/75%RH).
As part of hypothesis (Phase 1) a redilution study was not considered using the diluent allowed by the method to rule out any dilution error. Additionally a service engineer for the instrument confirmed that there was no instrument failure during initial or hypothesis testing. However instrument error was still suspected and substantiated by intentionally pumping air bubble into the system and the standard injections yielded results higher than the specification limits. Furthermore since the method validation for solution stability was deficiently established (only for short duration) a second set of samples was analysed in duplicate to invalidate the original failing results.
Passing results of other batches from initial sequence were discarded and retested results were reported without a valid justification. No consideration was given to report the average of all passing results when definitive root cause is not identified.
OOS investigations should be comprehensive, not to be hurried through with preconceived notions of errors / causes. The OOS investigation should give the impression that an investigation process was followed and from this possible causes were shortlisted / arrived at. Many QC labs practice long sequences cubbing many samples for higher throughput of analysis, but this can be counterproductive when errors / discrepancies happen in even one of the samples in a sequence.
Leave a Comment
You must be logged in to post a comment.