Lot Traceability and Laboratory Recordkeeping

The Lot as the Unit of Identity

Traceability, in the general quality-systems sense, is the ability to trace the history, application, or location of an object (ISO 9000:2015). In a laboratory handling physical materials, that ability is not distributed evenly across every possible descriptor. A compound name identifies chemistry. A catalog designation identifies a product line. A shipment identifies a logistics event. None of these identifies the specific quantity of matter that sat in a specific vessel and produced a specific analytical result. That is the work the lot identifier does, and nothing else in the record does it.

The regulatory definitions are worth reading closely because they are precise in a way that casual usage is not. Under 21 CFR 210.3(b)(2), a batch is a specific quantity of a drug or other material intended to have uniform character and quality within specified limits, produced according to a single manufacturing order during the same cycle of manufacture. Under 210.3(b)(10), a lot is a batch, or a specific identified portion of a batch, having that same uniformity within specified limits. The third definition is the operative one: 210.3(b)(11) defines a lot number, control number, or batch number as any distinctive combination of letters, numbers, or symbols from which the complete history of manufacture, processing, packing, holding, and distribution of that batch or lot can be determined.

Note what that last definition does and does not constrain. It says nothing about format — no required length, no checksum, no prescribed encoding of date or line. It defines the identifier entirely by function: an identifier is a lot number if and only if a complete history can be recovered from it. A string printed on a label that leads nowhere satisfies the typography of a lot number without satisfying the definition. This functional framing is the single most useful idea in the whole subject, because it converts an abstract expectation into something a reader can test.

Materials outside the scope of pharmaceutical manufacturing regulation inherit the convention without inheriting the enforcement. The consequence is that lot boundaries are set by whoever produced the material, and a receiving laboratory cannot redefine them after the fact. Two containers bearing the same lot identifier are, by assertion, from one production event and one analytical characterization. Two containers of the same compound bearing different lot identifiers are two distinct materials for every purpose that matters analytically: impurity profile, residual solvent content, water content, counter-ion stoichiometry, and physical form can all differ between lots of nominally identical chemistry. Results attach to lots, not to names.

What Fails When Lot Identity Is Unreliable

Three distinct capabilities fail, and they fail independently, so a laboratory can lose one without immediately noticing the loss of the others. The first is attribution: the inability to state which material produced a given result. Once the link is broken, an analytical value becomes a number about an unspecified thing. It can still be reported, which is precisely the danger — nothing about the number looks wrong.

The second is investigation. When a result falls outside expectation, the first structural question is whether the cause lies in the material or in the method. Answering it requires partitioning: re-examining other work performed on the same lot, and work performed on other lots by the same method. Reliable lot identity makes both partitions available and bounds the investigation to a finite set of affected records. Unreliable lot identity leaves the anomaly unbounded. Every prior result becomes potentially implicated and none can be cleared, which in practice means the investigation either expands past the point of feasibility or is quietly abandoned.

The third is reproducibility, which is really the first two extended over time and across laboratories. Freedman, Cockburn and Simcoe estimated that roughly half of United States preclinical research spending goes to work that is not reproducible, and their decomposition of contributing causes placed biological reagents and reference materials as the largest single category, ahead of study design, data analysis, and laboratory protocols. Material identity is not a peripheral bookkeeping concern within the reproducibility problem; by that accounting it is the largest part of it.

Two general checks follow directly from the functional definition in 210.3(b)(11). First, a lot identifier should be specific to one production event of one material; if the same identifier string appears on chemically distinct materials, it is not resolving to a single batch, and whatever function it is serving, it is not the function the definition describes. Second, a lot identifier should resolve to a corresponding analytical record carrying the same identifier, since otherwise the complete history the definition requires cannot be recovered from it. Both are verification steps a laboratory applies to its incoming materials as a matter of routine, in the same spirit as verifying a balance calibration before weighing.

The Receipt Record: Where the Chain Begins

The receipt record is the point at which an external identifier is admitted into an internal system, and every subsequent link depends on the fidelity of that admission. It should be created at receipt, not reconstructed later from the container, because the container will eventually be relabeled, consumed, or discarded, and because a record written after the fact records what someone remembers rather than what occurred.

A sufficient receipt record generally captures:

• Date and time of receipt, and the identity of the person who received the material.
• The source as designated on the accompanying documentation, and any reference number that links back to it.
• The material designation exactly as it appears on the label, transcribed verbatim rather than normalized to house nomenclature.
• The lot identifier exactly as printed, including case, punctuation, and leading zeros.
• Container count, container type, and nominal contents per container.
• The condition of containers and closures on arrival, and the condition of the outer packaging, including any indicators of transit conditions.
• Any accompanying documentation, identified by its own document number and date, and where it has been filed.
• The internal accession identifier assigned by the receiving laboratory.
• The location to which the material was assigned, and the nominal conditions of that location as recorded.
• Any discrepancy between what was expected and what arrived, described rather than merely flagged.

Two cautions apply to that list. The internal accession identifier is an addition to the lot identifier, never a replacement for it; the mapping between the two must be one-to-one and permanently retrievable, or the laboratory has simply substituted its own opaque string for someone else’s. And transcription should be verbatim even where the printed identifier looks malformed, because normalizing it destroys the evidence of the malformation. A photograph of the label attached to the receipt record costs nothing and resolves a large fraction of later disputes about what was actually written.

Discrepancies deserve their own discipline. A container that arrives with a lot identifier different from the one on its accompanying paperwork is a recordable event whether or not it is later resolved. Recording the discrepancy and its resolution is what makes the eventual resolution meaningful; silently correcting the record to the value someone believes is right converts a documented anomaly into an undocumented one.

Chain of Custody

Chain of custody is the principle that at every moment from receipt to final disposition, accountability for a material rests with either an identified person or a defined, access-controlled location, and that every transition between those states is a recorded event. The test of an unbroken chain is not that the record is long but that it contains no unaccounted intervals. A gap is not a neutral absence of information; it is an interval during which the material’s identity and condition are unverified, and no later reconstruction repairs it. A reconstruction is an assertion about the past, and it should be labeled as one.

Four event types account for most of what happens to a material in a laboratory: transfer between people, relocation between storage locations, subdivision of a container into subordinate containers, and final disposition through consumption or disposal. Each requires the same four facts — who, when, what, and how much — and subdivision requires one more.

That additional fact is parentage. When material from a received container is subdivided, each subordinate container must carry both its own identifier and the identifier of the lot it came from. A subordinate container labeled only with a compound name has severed the chain at the moment of subdivision, and every result generated from it thereafter is attributable to a compound but not to a lot. A two-part label — parent lot identifier plus a serial for the subordinate container — preserves the link at trivial cost, and the subdivision event itself should appear in the record with the same who-when-what-how-much detail as any transfer.

Periodic reconciliation closes the loop. Comparing quantity received against quantity documented as consumed, transferred, or disposed of, and against quantity remaining, produces a check on the custody record as a whole. An unexplained discrepancy is not primarily an inventory problem. It indicates that at least one custody event went unrecorded, which means the record is known to be incomplete and the extent of the incompleteness is unknown.

The Notebook and Its Electronic Equivalents

The receipt record and the custody record are inert unless the lot identifier appears again in the experimental record at the point where the material was actually used at the bench. This is the most commonly broken link in an otherwise disciplined system: a laboratory that documents receipt meticulously and then writes only a compound name in the notebook has recorded the material’s arrival and its work separately, with nothing joining them. The lot identifier belongs in the experimental entry itself, alongside the quantity taken and the container it was taken from.

Paper practice is well established and remains defensible. A bound, pre-paginated book; permanent ink; entries made as the work proceeds rather than transcribed later from loose sheets; no pages removed; corrections made by a single strike-through that leaves the original legible, initialed and dated, with a reason where the reason is not obvious; unused space cancelled so that later insertion is visibly impossible; and, where the work warrants it, a witness signature from someone who understood what was done. None of this is ceremonial. Each rule exists to make a specific class of undetectable alteration detectable.

Electronic systems must reproduce those same guarantees through different mechanisms. Under 21 CFR 11.10(e), that means secure, computer-generated, time-stamped audit trails that independently record the date and time of operator entries and actions creating, modifying, or deleting records, retained at least as long as the records themselves; EudraLex Volume 4 Annex 11 sets comparable expectations for computerised systems. The practical requirements are unglamorous: individual accounts with no shared credentials, because an audit trail that attributes an action to a shared login attributes it to no one; controlled system clocks; restricted ability to alter time stamps; verified backup and restore rather than assumed backup; and audit trail review conducted as a scheduled activity rather than merely enabled as a feature.

The subtlest question in electronic recordkeeping is what constitutes the original record. FDA’s 2018 data integrity guidance draws the distinction between static and dynamic records: a printed chromatogram is a static representation, but if the underlying data file can be reprocessed with different integration parameters, the dynamic file — together with its method and processing parameters — is the original, and the printout alone is an incomplete record. The same logic covers the lot identifier itself. When it is stored as a metadata field in an instrument data system, that field is part of the record, subject to the same audit trail expectations as the result it qualifies.

ALCOA and ALCOA+

The ALCOA mnemonic originated inside FDA in the early 1990s as a training device for investigators conducting Good Laboratory Practice inspections. Stan Woollen, who coined it, later described its origin in a 2010 article, and it has since been absorbed into essentially every major data integrity guidance. Its durability comes from being a review checklist rather than a compliance framework: five properties a record either has or does not.

Attributable means the record identifies who generated it and, where relevant, who reviewed it. Legible means it can be read and understood for as long as it must be retained, which for electronic records is a question about formats and readers, not handwriting. Contemporaneous means it was made at the time of the activity — the property that entries transcribed from scrap paper at the end of a session specifically lack. Original means the record is the first capture of the observation, or a verified true copy of it, which is where the static-versus-dynamic distinction applies. Accurate means it correctly reflects what occurred, including corrections that are themselves visible.

Applied to lot traceability, each letter has a concrete referent. Attributable: the receipt entry names the receiver. Contemporaneous: the entry was made at receipt, not reconstructed from the container weeks later. Original: the identifier was transcribed from the container itself rather than copied from a secondary document. Accurate: it matches the container character for character. Legible: it will still be readable when someone re-examines the work.

The ALCOA+ extension adds four properties that address the record set rather than the individual record: complete, consistent, enduring, and available. Complete means nothing has been omitted, including failed runs and repeated analyses. Consistent means the sequence of events is coherent and chronological across the whole record. Enduring means the record survives on media that will outlast the retention period. Available means it can actually be retrieved for review throughout that period. These additions appear in convergent form across MHRA’s GXP data integrity guidance (2018), WHO Technical Report Series 996 Annex 5 (2016), PIC/S PI 041-1 (2021), and OECD’s GLP data integrity advisory document No. 22 (2021). The convergence is worth noting: these bodies regulate different activities and arrived at substantially the same list.

Retention, Archiving, and Why Traceability Underpins Reproducibility

There is no single retention period that applies to all laboratory records, and a laboratory that adopts one without identifying which obligation it is satisfying has usually chosen it arbitrarily. Several anchors exist. Under 21 CFR 211.180(a), production, control, and distribution records associated with a batch are retained at least one year past that batch’s expiration date, or three years after distribution for certain OTC products exempt from expiration dating. For work funded by United States federal awards, 2 CFR 200.334 requires record retention for three years from submission of the final financial report, extended until any litigation, claim, or audit involving the records is resolved. ISO/IEC 17025:2017 takes a different approach for testing and calibration laboratories, requiring the laboratory to define and control retention times for its technical records rather than prescribing a number. Journal policies, institutional research data policies, patent considerations, and litigation holds can each impose longer periods, and the longest applicable obligation governs.

Archiving is not backup, and conflating them is a common failure. Backup guards against loss of a live system. Archiving preserves a fixed record, with its metadata and audit trail, for a defined period under controlled access. A backup of a database that is later restored into a newer application version may recover the values while degrading the audit trail — which is to say, it may recover the results while losing the attribution that made them records rather than numbers. The ALCOA+ criterion of endurance is fundamentally about this: media obsolescence, proprietary format obsolescence, and migration are foreseeable events over a multi-year retention period, and migration procedures should be validated to demonstrate that attribution, time stamps, and lot linkage survive the move.

The reproducibility argument closes the circle. In Nature’s 2016 survey of more than 1,500 researchers, a large majority reported having failed to reproduce another scientist’s experiment, and a majority reported failing to reproduce their own. Many contributing causes are methodological, but a substantial share reduces to an identity question: whether the material in the second laboratory was the same material as in the first. A written method that specifies a compound name, a manufacturer, and nothing further is reproducible only up to the between-lot variation of that material, and that variation is precisely what the method does not report.

What lot traceability provides, in the end, is not certainty about a material. It provides the ability to ask a specific question later and get a specific answer: this result came from this quantity of material, received on this date, from this source, characterized by this analytical record, held under these conditions, handled by these people. Any of those facts may later prove to be wrong. The point of the record is that it can be checked, and an unverifiable claim and a verified one are different kinds of object even when they happen to say the same thing.

References

1. U.S. Code of Federal Regulations, Title 21, Part 210, §210.3(b)(2), (b)(10), (b)(11) — definitions of “batch,” “lot,” and “lot number, control number, or batch number.” (21 CFR 210.3)
2. U.S. Code of Federal Regulations, Title 21, Part 211, Subpart J, §211.180 — General requirements (records retention). (21 CFR 211.180)
3. U.S. Code of Federal Regulations, Title 21, Part 11 — Electronic Records; Electronic Signatures; §11.10(e) (audit trails). (21 CFR Part 11)
4. U.S. Code of Federal Regulations, Title 2, §200.334 — Record retention requirements (Uniform Administrative Requirements for Federal Awards). (2 CFR 200.334)
5. U.S. Food and Drug Administration. Data Integrity and Compliance With Drug CGMP: Questions and Answers — Guidance for Industry. December 2018.
6. Medicines and Healthcare products Regulatory Agency (UK). ‘GXP’ Data Integrity Guidance and Definitions, Revision 1. March 2018.
7. World Health Organization. Guidance on good data and record management practices. WHO Technical Report Series No. 996, Annex 5. 2016.
8. Pharmaceutical Inspection Co-operation Scheme. Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments, PI 041-1. 1 July 2021.
9. OECD. Advisory Document of the Working Party on Good Laboratory Practice on GLP Data Integrity. OECD Series on Principles of Good Laboratory Practice and Compliance Monitoring, No. 22, ENV/CBC/MONO(2021)26. 2021.
10. European Commission. EudraLex Volume 4, Good Manufacturing Practice Guidelines, Annex 11: Computerised Systems. 2011.
11. ISO/IEC 17025:2017, General requirements for the competence of testing and calibration laboratories — clause 7.5 (technical records) and clause 8.4 (control of records, including retention times).
12. ISO 9000:2015, Quality management systems — Fundamentals and vocabulary (definition of traceability as the ability to trace the history, application or location of an object). Clause number not asserted; sources differ.
13. Woollen SW. Data Quality and the Origin of ALCOA. The Compass (newsletter of the Southern Regional Chapter, Society of Quality Assurance), Summer 2010. Article verified; exact volume and page numbers not confirmed.
14. Freedman LP, Cockburn IM, Simcoe TS. The Economics of Reproducibility in Preclinical Research. PLoS Biology. 2015;13(6):e1002165. DOI: 10.1371/journal.pbio.1002165
15. Baker M. 1,500 scientists lift the lid on reproducibility. Nature. 2016;533(7604):452-454. DOI: 10.1038/533452a