Almost every major regulatory action of the last decade has cited data integrity. Not because organizations set out to falsify records, but because their systems allowed records to be created, changed or discarded without a reliable, attributable trace. A data integrity program is therefore not a documentation exercise — it is an engineering and governance discipline applied to every piece of information a regulator may rely upon.
This whitepaper sets out a framework that multi-site manufacturers can implement incrementally: establish governance, map data lifecycles, translate ALCOA+ into controls, and then prove those controls work through continuous monitoring rather than pre-inspection remediation.
Why data integrity programs fail
Programs rarely fail because a principle was unknown. They fail because ownership was ambiguous, because hybrid paper-and-system processes left unmanaged gaps, or because controls were designed for the validated system while the real work happened in a spreadsheet beside it.
- No single accountable owner for data across quality, manufacturing, laboratory and IT.
- Uncontrolled spreadsheets performing GxP calculations outside the validated estate.
- Shared logins and generic accounts that break attributability at the source.
- Audit trails enabled but never reviewed, or reviewed without documented rationale.
- Local site practices that diverge silently from the corporate procedure.
The ALCOA+ control framework
ALCOA+ is widely quoted and rarely operationalised. Each attribute should map to a specific, testable system or procedural control with defined evidence.
| Attribute | Control expectation | Evidence in an inspection |
|---|---|---|
| Attributable | Unique user identity, no shared accounts, e-signature binding | User access review, signature manifest |
| Legible | Human-readable records and durable export formats | Retrieved record from archive |
| Contemporaneous | System-enforced timestamps from a synchronised source | Time source configuration, audit trail |
| Original | Raw data preserved with defined true-copy process | Instrument raw file plus verified copy |
| Accurate | Validated calculations, second-person or system verification | Validation report, review record |
| Complete | All data retained including repeats, aborts and reprocessing | Audit trail showing full history |
| Consistent | Sequenced events with no ability to backdate | Chronological audit trail extract |
| Enduring | Retention aligned to product lifecycle plus regulation | Retention schedule and archive test |
| Available | Retrievable within inspection timelines | Timed retrieval demonstration |
Mapping the GxP data lifecycle
Before controls can be assessed, the data itself must be inventoried. A lifecycle map records where data originates, every system it traverses, who may change it, and how it is finally retained and retrieved.
GxP data lifecycle
Generation
Instrument, operator entry or interfaced system creates the record.
Processing
Calculations, integrations and transformations under version control.
Review
Risk-based audit trail and second-person review with documented rationale.
Reporting
Release decisions and regulatory submissions built on approved data.
Retention
Archived in readable form for the full statutory retention period.
Retrieval
Demonstrable recovery within inspection response timelines.
Governance that holds under pressure
Governance should be light enough to operate weekly and strong enough to escalate. In practice this means a named data owner per domain, a standing review forum, and a single register of data integrity risks with treatment plans and due dates.
- 1Appoint accountable data owners across QA, manufacturing, laboratory and IT.
- 2Maintain a live inventory of GxP systems, interfaces and spreadsheets.
- 3Score each data flow for criticality, detectability and existing control strength.
- 4Define audit trail review frequency proportional to that risk score.
- 5Track findings, CAPAs and effectiveness checks in one auditable system.
78%
of observations trace to attributability or audit trail gaps
3x
faster inspection response with a maintained data inventory
< 24h
target retrieval time for archived GxP records
Risk-based audit trail review
Reviewing every audit trail entry is neither feasible nor expected. What is expected is a defensible rationale: which events matter, how often they are examined, who performs the review, and what happens when an anomaly is found.
- Define critical event types — result modification, method change, reprocessing, deletion, clock change.
- Automate exception filtering so reviewers see signal rather than volume.
- Record the review, the reviewer and the outcome as a controlled record.
- Escalate anomalies into the deviation process rather than resolving them informally.
A control you cannot evidence is a control you do not have.
Technology enablers
A unified platform removes whole categories of risk by design: identity is centralised, audit trails are native rather than optional, and quality, laboratory and manufacturing records share one lineage. Where a single platform is not yet realistic, focus first on eliminating uncontrolled spreadsheets and shared accounts — these two issues account for a disproportionate share of findings.
Building inspection readiness as a steady state
Organizations that inspect well do not prepare for inspections; they operate in a permanently evidenced state. Dashboards show overdue audit trail reviews, open data integrity risks, access review status and archive retrieval test results. Nothing has to be assembled, because everything is already current.