EU GMP Chapter 4 asks that any alteration is signed and dated, leaves the original readable and records a reason where appropriate. Correction fluid, overwritten digits and blacked-out entries all break that rule and ALCOA+. Engineering and site teams need it in their induction as much as QA.
Grade A must meet ISO 5, 3,520 particles of 0.5 µm or larger per cubic metre, both at rest and in operation. An FFU array is therefore judged with operators and machines in the zone: uniform unidirectional flow, leak-tested H14 filters and a smoke study run with the real interventions.
EU GMP Annex 1 (2022) lists significant modification to the HVAC system among the reasons to repeat an aseptic process simulation. Changes to airflow pattern, pressure regime or filter housings in aseptic areas should reach QA with an engineering note before the work, and the requalification should be planned with the shutdown.
Classification spreads sampling points evenly by floor area; environmental monitoring should go where contamination would reach open product first. Place particle probes and settle plates from the smoke study, the intervention list and a documented risk assessment, keep them clear of the flow, cover the last-to-recover spots, and reference the evidence behind each point.
Two independent long-lead paths, each 80% likely on time, give a 64% chance that both arrive for qualification. A qualitative FMEA register manages risks one at a time; a Monte Carlo schedule analysis adds them up and gives P50 and P80 dates. It earns its cost where paths converge on a fixed date.
The revised EU GMP Annex 1 has applied since 25 August 2023, with point 8.123 following in 2024. A site still planning its gap assessment is years behind. Most findings land in the CCS, for example a pressure alarm delay set at commissioning that nobody can justify.
EU GMP Annex 11 periodic evaluation asks whether a computerized system is still valid and compliant. Most of the evidence already exists in change control, deviation and incident records and the upgrade history. Review first, then retest only the functions the review shows are affected.
Compressed air and nitrogen that drive product pigs, blanket vessels or purge tube fillers touch the product. Give each use point its own ISO 8573-1 class for particles, water and oil, sample after the final filter rather than at the compressor, and test viable counts to ISO 8573-7.
Grade B at rest is held to 3,520 particles ≥0.5 µm and 29 ≥5 µm per m³. Annex 1 (2022) defines at rest for GMP grades as HVAC running, main equipment installed and standing by, and no personnel, reached after a clean-up period of under 20 minutes as guidance. The protocol should record that condition.
EU GMP Annex 15 section 3 asks for design compliance with GMP to be demonstrated and documented, with URS requirements verified during DQ. Meeting minutes rarely do that. A DQ report traces every URS line to a drawing, calculation or datasheet at a stated revision, with a verdict and a signature.
Check the cascade room by room: doors should open into the higher-pressure room, BMS setpoints should match the drawing, EMS alarm limits and delays should match the CCS, and each pressure transmitter should measure the rooms the drawing names. Record every difference before deciding which side is wrong.
An SOP register should record, for each SOP, the owner, version, review date and every document it refers to. Humiley's template covers 536 SOPs across 20 departments. The cross-reference column tells a change control which documents a revision touches, and review dates keep superseded versions out of use.