The dispensed material is staged for charge-in. One line in the batch record is blank: the second person’s verification of the material. The supervisor says the operator did the work correctly. The operator says a colleague was right there and looked at the label. Both are credible. Neither is a record.
In this article
The wrong item feels like proof because the story hangs together.
Reviewers are not usually careless here. Three things push the same way. The person speaking is credible. The account is coherent: everyone describes the same ordinary morning. And the pile of material grows. A printout, an access log and a statement look like three pieces of corroboration, so the file feels heavier even when none of it addresses the question that is open.
That is the habit to break. Counting items is not testing them. The question that matters is narrower: for each item, what claim can it carry, and what claim can it not?
If a reviewer can say, item by item, what is established and what is not, the barrier is gone. If the answer is “it all points the same way,” the reasoning has not started.
What each item can carry.
The table uses the synthetic case from the The Record in Practice: a solid oral dosage form, a component dispensed at step 14, and a blank line for the second person who verifies the material identification. Step 14 of this record requires that verification before dispensing. No real batch, product, person or site is involved.
| Item | What it can carry | What it cannot carry |
|---|---|---|
| Balance printout: 62.5 kg at 08:43, lot field reads ML-4471 | A weighing was recorded at a time, with that lot named. | That the container held that lot, or that anyone compared the container to the lot named. |
| Badge log: KP in the weighing room 08:20 to 09:05 | KP was present. | What KP did there. |
| Step 13 balance check, verified by KP, initialed 08:31 | KP signed a different check at 08:31. | That KP performed or signed the step 14 identification check, or when. |
| Operator recollection, unsigned, taken by the supervisor who proposed closing the item | A lead worth following. | An independent record. It is not made at the time, not attributable to a signed entry, and not independent of the proposal to close. |
| Label photograph, scan record or second printout (none located yet) | It could identify the container, the lot and the time. | Nothing yet. It has not been found. |
| KP’s account, taken now by someone independent | Whether KP was the intended verifier, who compared what against what, and when. Independent and corroborated, it can help establish what happened. | A record made at the time. It is a new statement made today, and the file should say so. |
Three questions sort the rows. Does it identify the specific event? A printout of a weighing identifies a weighing, not an identity check. Does it name who did it? A badge log names who entered, not who verified. Was it made at the time? A recollection taken after the item was flagged is made later, by someone who is not independent of the proposed closure.
The conclusion you can write while the check is not established.
Quality does not have to choose between “the check happened” and “the check did not happen.” It can write what the evidence supports and stop there.
The sentence avoids two errors. It does not say the check was performed, because nothing in the file records it. It does not say the check was skipped, because nothing in the file shows that either. It names what is known and leaves the open question visibly open, where the next reviewer will see it.
Keep the question open, and make it specific.
An open question is not a stalled one. The questions that remain here are specific and answerable:
- Ask KP, independently. Was KP the intended verifier? What did KP compare, against what, and was it before or after the material was dispensed?
- Look for a record made at the time. A label photograph, a scan record, a second printout.
- Check the material while it can still be checked. If the dispensed material has not yet been charged, hold the container and re-check its label against the lot named in step 14. That does not replace the missed check. It bears on whether the right material was used.
What the file should not do is close the question for convenience. A late-entered verification signature would document a conclusion the evidence does not yet support. An accurate, dated note that the line was found blank is a different thing. Retraining would answer a question about behavior before Quality knows what happened. And proceeding because later testing will confirm the material swaps one piece of evidence for another: identity testing of the lot speaks to the lot, not to which container was dispensed.
What the decision record should contain
The Four Lenses keep the response from running ahead of the evidence. Applied to the synthetic case, a record that preserves the reasoning might read as follows. This is a fictional example, not a template for a real investigation.
| Lens | What the record contains | Question before closing |
|---|---|---|
| EvidenceWhat was known? | Step 14 initialed at 08:44, verification line blank. Balance printout at 08:43 naming the lot. Badge log placing KP in the room. KP’s step 13 sign-off at 08:31. An unsigned recollection taken by the supervisor who proposed closing. | Which items identify the event, name the person and were made at the time? |
| JudgmentHow was the evidence evaluated? | The items are consistent with a check. None records one. Performance of the check is not established. | Is the conclusion limited to what each item carries, rather than to the closure that was proposed first? |
| RiskWhat remained uncertain? | Whether the container held the lot named, what KP compared and when, and whether the material has been used. | What is still unknown, and what could it affect? |
| AuthorizationWhat was decided? | The discrepancy stays open under the site’s procedure. The specific evidence is gathered. The decision on the batch is held until the impact is assessed. | Is the authorizing person deciding on a conclusion the evidence supports? |
What the record should explain
A blank verification line is evidence that the record stopped answering a question. People who were present can help answer it. They cannot be the answer unless what they say is tied to a specific event, to a named person and to a time. A later account can help establish what happened when it is independent and corroborated, but the file must say it was made later. A Quality reviewer should be able to show, item by item, which of those the file supports.
Make the review answer four questions.
- What does each item identify? The specific event, or only something near it.
- Who does each item name? The person who did the work, or only someone who was present.
- When was each item made? At the time, or after the question was raised.
- What conclusion does that support? Write it, and write what remains open.
For the step before this one, read why not every record gap is a writing problem. For what a complete record may still leave unexplained, read objective evidence and the GMP decision record.
Regulatory context and sources
For finished pharmaceuticals, 21 CFR 211.101(c) and (d) call for a second person to examine each container of component dispensed, including that the containers are properly identified, and to verify each component added. 21 CFR 211.188(b)(11) requires the batch record to identify the persons performing and directly supervising or checking each significant step. 21 CFR 211.100(b) requires written production and process control procedures to be followed and documented at the time of performance, and any deviation from them to be recorded and justified.
21 CFR 211.192 requires any unexplained discrepancy to be thoroughly investigated, with the investigation extended to other batches that may have been associated with it, and the conclusions and followup recorded. 21 CFR 211.22(a) gives the quality control unit the authority to review production records to assure that no errors have occurred or, if errors have occurred, that they have been fully investigated.
Source context: Regulatory requirements: 21 CFR 211.22; 21 CFR 211.100, Written procedures; deviations; 21 CFR 211.101, Charge-in of components; 21 CFR 211.188, Batch production and control records; and 21 CFR 211.192, Production record review. The case is the synthetic case used in the ComplianceWorxs The Record in Practice; the record example and the Four Lenses review are CW editorial examples, not regulatory mandates. This article has not been reviewed by a practitioner for regulatory sufficiency. ComplianceWorxs does not provide regulatory or legal advice and does not judge whether any record or corrective action is adequate.