A record can be retrievable and still fail ALCOA

4.2.5 is a filing test, not a trust test

Clause 4.2.5 requires controlled records to remain legible, readily identifiable, and retrievable. All three are properties of storage and indexing: can the record be found, read without special effort, and matched to the right device, batch, or event. None of the three asks whether the data the record reports is actually accurate, or whether it was captured by the person, and at the time, it claims to have been. A record can satisfy all three and still describe something that didn't happen the way the record says it did.

ALCOA asks where the data came from

FDA's Data Integrity and Compliance With Drug CGMP Questions and Answers, finalized in December 2018, builds its own expectations around data that is attributable, legible, contemporaneous, original, and accurate. A record has to be traceable to the specific person who created it, generated at the time the activity it describes actually happened, kept as the original record or a verified true copy, and correct. A file that was backdated, or signed by someone who didn't perform the work it documents, fails most of those tests while still being perfectly legible, identifiable, and sitting exactly where the retention schedule says it should be.

MHRA adds four more letters for a reason

The UK's ‘GXP’ Data Integrity Guidance and Definitions, Revision 1, published by the MHRA in March 2018, extends the same five attributes into what practice now widely calls ALCOA+: the data also has to be complete, consistent, enduring, and available — a whole data set, internally consistent with itself, preserved for its full required retention period, and genuinely available for review rather than merely retrievable in principle. A record that's legible, identifiable, and retrievable under Clause 4.2.5, but missing the one data point that would let a reviewer actually reproduce its stated result, has passed the narrower test and failed the broader one in the same breath.

Where electronic records make the gap wider

An electronic record raises the stakes on exactly the attributes 4.2.5 never names. Contemporaneous capture depends on a system timestamp nobody can quietly edit after the fact; attributable depends on a login tied to one specific person rather than a shared device account; original depends on an audit trail that shows every change, not a document that simply looks finished. Part 11's own controls — audit trails, access controls, identity-bound electronic signatures — exist to make those attributes checkable on an electronic record the way they'd already be checkable on a paper page carrying a wet signature and a handwritten date. A retention policy built only around 4.2.5's three words treats an electronic record as compliant the moment it opens, without asking whether what opens is still the same thing that was originally recorded.

Device records inherit the same test, guidance or not

FDA's own data-integrity guidance speaks to drug CGMP under 21 CFR Parts 210 and 211, not to device recordkeeping directly — there's no equivalent device-specific guidance naming ALCOA by letter. That gap in jurisdiction doesn't make the underlying question go away for a device manufacturer's own records. A complaint file, a calibration log, or a batch release record kept under ISO 13485 faces the identical exposure: a reviewer asking whether an entry is attributable and contemporaneous isn't asking a drug-CGMP question, they're asking whether the record can be trusted, and Clause 4.2.5's own three words were never built to answer that on their own.

Where this meets the file

A record-integrity checklist that scores a controlled record against ALCOA's attributes separately from the retrievability test Clause 4.2.5 already sets, rather than treating one as a stand-in for the other, is previewed alongside the launch catalog's other quality-system templates. If your program draws that line differently, the shelf takes that correction directly.