What the inspector wrote
Change control CC-24-0387 replaced the coating pan exhaust fan with a unit of higher capacity. The impact assessment recorded 'no impact' for validation, regulatory, cleaning, stability and training, with no rationale, and was signed by the requesting engineer and QA only. Validation, regulatory affairs and production were not consulted. Subsequent deviation DEV-25-0041 attributed a coating uniformity drift to altered drying conditions resulting from the increased exhaust rate.
Why it was cited
Annex 15 11.3 requires quality risk management to evaluate the potential impact on quality, validation, regulatory status and other systems, and ICH Q10 expects the evaluation to draw on the relevant expertise. A row of 'no' ticks from two people, one of whom wanted the change, is not an evaluation.
The later deviation proves the point: the impact existed, was foreseeable by anyone who knew the process validation, and was not foreseen because nobody who knew was asked.
What would have prevented it
- Require each impact area to be answered by the function that owns it, with a reference to what was checked (protocol number, dossier section, agreement clause) for every 'no' as well as every 'yes'.
- Give QA reviewers an explicit instruction to reject assessments with unsupported 'no impact' answers.
- For major changes, hold a change review meeting with minutes rather than routing a form.