What the inspector wrote
An assay result of 88.2% (specification 95.0 to 105.0%) for batch 7702 was obtained on 3 May. The sample was re-prepared and re-injected the same day, giving 97.1%, which was reported on the certificate of analysis. No laboratory investigation was opened until 9 May, after the batch had been certified. The original result was not referenced in the batch record.
Cited against EU GMP Chapter 6, 6.35 and 6.36; PIC/S PE 009 Part I, Chapter 6
Why it was cited
This is the classic 'testing into compliance' pattern and regulators treat it as a data integrity issue, not a laboratory error. Once a failing result exists, the only acceptable next step is a documented Phase I investigation. Retesting first, then investigating, means the retest was chosen to make the problem go away.
Reporting only the passing result compounds it: the batch record now misrepresents what the laboratory knew at release.
What would have prevented it
- A procedure that makes any retest or re-preparation after a failing result impossible without an open, numbered investigation.
- Analyst training that treats an OOS as a stop signal, with supervisors reinforcing it rather than pushing for a result.
- Audit trail review that specifically looks for re-injections and re-processing on the same sample within a short window.
Written from the deficiency categories published by EU and UK regulators. Details are illustrative and do not describe a specific named site.