Laboratory data archiving: originals, readability and retention

The file still exists, but the program that read it no longer works. The laboratory therefore has the bytes, not necessarily a usable record. Archiving must allow the activity to be found and interpreted even when people, instruments and applications change.

The initial decision concerns the complete record to retain, not simply the most convenient format. Identify content, relationships, required period, responsibilities and reading tools. Availability must be demonstrable throughout retention, not postponed until the next inspection.

1. Define the record and its purpose

For each process, identify the elements documenting activities and decisions: original data, relevant metadata, methods and versions, processing, audit trails, approvals and links to the sample or batch. Specify the authoritative record’s location and how separate components are reassembled.

Assign data ownership, archive responsibilities and retrieval-request arrangements. If an external provider is involved, clarify access, retrieval and evidence availability after the contract ends. A shared folder without an index or responsibilities does not demonstrate a controlled archive.

Operational backup protects recovery capability; the archive manages retention over time. Terminology matters: FDA Data Integrity guidance Q1(e) uses “backup” in the context of §211.68(b) for a true copy maintained throughout the required period. A temporary rotating copy alone does not meet that purpose.

2. Preserve content and meaning, not only appearance

A static record presents fixed information; a dynamic record permits interaction with its content. A chromatogram image may show the result without allowing examination of signals, parameters or processing history needed for review.

Do not automatically convert every dataset to PDF. A static format may be appropriate when it is the original or a suitable true copy of the record; it does not indiscriminately replace dynamic content. Assess the functions and metadata needed to reconstruct the activity and demonstrate their preservation.

A verified copy requires a controlled process, comparison criteria and evidence of the outcome. A format’s name does not guarantee completeness. Even native-format retention needs a reading plan: “original retained” and “original accessible” are separate checks.

3. Build a plan by record type

This matrix is an original editorial model. Complete the period field with the applicable requirement, the event starting the period and review responsibility; the references shown are not automatic deadlines.

EU GMP Chapter 4 distinguishes batch documentation from other documents: §§4.11–4.12 do not justify one number of years for the whole laboratory. Consider product, market, dossier, supported activity and specific constraints. Record any disposal holds for investigations or other applicable obligations.

4. Design reading, search and access

Inventory format, reader version, compatible operating system, licences, media and other dependencies. For encrypted data, key custody and recovery capability must last as long as the need to read. No format alone promises perpetual readability.

Define useful searches: sample, batch, study, instrument, method and time interval. Verify that an authorised user can trace a result to relevant components without memorising old paths. Distinguish archive consultation from data modification.

For review and inspection, prepare display and export arrangements preserving context and relationships. A readable printout can help, but does not remove the need for necessary electronic elements. The link between signatures and records must remain interpretable in the consultation environment.

5. Simulated case: the old viewer will not open

During a periodic check, a spectral file is found in the archive, but its historical reader will not start on the new operating system. The PDF report is available, yet cannot expose all content required for the investigation. Retrieval is therefore incomplete.

The team preserves the original, records the problem and identifies version, dependencies and supported alternatives. It assesses a compatible controlled reading environment or a verified conversion or migration. It does not indiscriminately install old software in the production laboratory or delete the file because a PDF exists.

Testing compares content, metadata, relationships and required review functions using representative examples, documenting losses or differences. Only an approved decision supported by evidence establishes the retention path. If an element remains unreadable, its limitation and impact remain declared.

6. Review the archive until obligations end

Plan checks of retrievability, readability and integrity at justified intervals. Also check after relevant software, hardware or service changes. Select examples including older versions and complex relationships, not just the easiest file to open.

Record retrieval time and method, results, exceptions and actions. Monitor obsolescence, contract expiry and availability of expertise. Keeping a licence does not demonstrate that the environment still works.

At the end of the period, verify that every obligation has ended and no relevant hold exists. Disposal decisions must be authorised and traceable under procedure, considering copies and linked components. System retirement does not automatically authorise record deletion; this article neither performs nor directs deletion on real systems.

Sources and status — checked 2 October 2026. EU GMP Chapter 4, January 2011, §§4.10–4.12; Annex 11, January 2011, §17; FDA Data Integrity, final nonbinding guidance, December 2018, Q1(d–e), Q9–10; 21 CFR Part 11, §§11.10(b–c), within its scope; PIC/S PI 041-1, 1 July 2021, §9.9. Matrix and case are examples to adapt to the applicable context.