Batch release by the quality unit: the signature has to show what was reviewed

Batch release by the quality unit: the signature has to show what was reviewed

In batch release, the failure mode is usually not the missing signature. It is usually the signature that is there, on time, on a batch record package the quality unit (QU) could not review in full: a deviation still open, an out-of-specification (OOS) result without a completed investigation, instrument data that was not retained. In the record, the batch looks released with due care. In practice, the decision was made on part of the evidence.

When this shows up in an inspection, it is rarely treated as late paperwork. It is treated as a failure of QU oversight. And an electronic release system does not change that conclusion on its own: it only helps if the signature stays tied to the evidence the QU actually saw.

What the regulation asks of the QU

One distinction worth making early: releasing a batch is not an administrative step at the end of production. Under 21 CFR 211.22(a), the quality control unit has the responsibility and authority to approve or reject the product, and the authority to review production records to assure that no errors have occurred or, if errors have occurred, that they have been fully investigated. 211.22(d) adds that these responsibilities and procedures must be in writing and followed.

211.192 spells out what goes into that review: all production and control records, including those for packaging and labeling, are reviewed and approved by the quality unit before the batch is released or distributed. Any unexplained discrepancy, or any failure of the batch to meet its specifications, is thoroughly investigated, including other batches that may have been affected, and the investigation is documented in writing.

And 211.188 defines what has to exist for that review to be possible: the batch production and control record with complete information, including identification of the persons performing and directly supervising or checking each significant step, and the investigations made.

Read together, the three sections say something simple: the QU can only release what it was able to review, and it can only review what the batch record kept in full.

Where release tends to lose the evidence

Recent FDA warning letters repeat a pattern. Three examples from 2025 and 2026 help separate the mechanisms.

Record reviewed without being complete. In an October 2025 letter [ref. 1], FDA notes that the QU did not ensure review of production and control records for completeness and accuracy (211.192), written procedures for investigating discrepancies, change control, or periodic product review. What is missing there is not one more document. It is the review that should support the decision on the batch.

Release with an investigation pending. In a November 2025 letter [ref. 2], the QU failed to adequately review production records before release: OOS results without investigation, reprocessing, and label reconciliation failures. These are three events that should hold the decision until resolved, and they stayed outside what the QU looked at.

Data that was not kept to be reviewed. In a May 2026 letter [ref. 3], data from an instrument was deleted automatically when the computer was shut down, so there was no record that the QU had reviewed that data. The same letter cites the instrument password taped to the computer, the absence of an audit trail, and the lack of a procedure to retain electronic records. Here the failure starts before the QU: the data never reached the batch record package.

In the last two letters, FDA asks, among the corrective actions, for a complete and final review of each batch and its related information before the QU disposition decision. It is the same requirement as 211.192, stated as an action plan.

What the system has to ensure before the signature

As I read it, an electronic batch release workflow only helps the QU if it has three properties.

1. Completeness check before the signature. The system brings together what the decision depends on: the electronic batch record (EBR) with every step executed and checked, deviations and investigations linked to the batch and closed (or with a formal justification to proceed), quality control results, environmental monitoring data where applicable, material and label reconciliation. If a required item is open, the release signature is not available. The decision to proceed with an open item exists, but as a controlled exception path, with the responsible person and the justification recorded.

2. Separation of roles and second-person review. Whoever performed a step is not the one who checks it, and whoever checked production is not the one who releases the batch for the QU. 211.188 already requires identification of the persons performing and directly supervising or checking each significant step. The system needs to prevent the same user from holding both roles on the same record, not just record it when it happens.

3. Signature with meaning and with what was seen. Under 21 CFR 11.50, a signed electronic record shows the printed name of the signer, the date and time, and the meaning of the signature (review, approval, responsibility), also in any human-readable form (screen or printout). That is the minimum. What makes the signature useful in an inspection is being tied to the exact version of the package reviewed: which deviations, which results, which version of the batch record. If the package changes after the signature, the change shows in the audit trail and the release goes back for review.

None of this replaces a QU with the authority and resources to reject a batch. The system makes visible what was reviewed; the decision remains with the QU.

When release only looks rigorous

The quieter failure mode is not the batch released with an explicit open item. It is the release that looks rigorous on the metrics. Signals worth watching:

• the time from end of production to batch release drops, while the number of open deviations per batch rises
• deviations closed on the same day as batch release, more often than in the rest of the month
• batch review done on a consolidated report, without direct access to the original instrument data
• execution and check signatures by the same user on critical steps, even if only a few
• records changed after release without the release being redone

In these cases, release is faster because the review got shallower, not because the process improved.

Before saying release has evidence

• Is the batch release signature blocked while a deviation or investigation linked to the batch is open, or does only a warning appear?
• For a batch from last month, can you show exactly which documents and data the QU had in front of it when it signed?
• Does the system prevent the same person from performing and checking the same critical step?
• Is the instrument data that feeds the decision retained with an audit trail, or does the QU review a summary?
• Does a change to the record after release reopen the release?

At your plant, for the last batch released, did the QU sign on the complete package or on what was ready at the time?

References

• [ref. 1] FDA — Warning letter to a finished pharmaceuticals manufacturer, 10/30/2025
• [ref. 2] FDA — Warning letter to a finished pharmaceuticals manufacturer, 11/12/2025
• [ref. 3] FDA — Warning letter to a finished pharmaceuticals manufacturer, 05/14/2026
• 21 CFR 211.22, 211.188 and 211.192 (current text on eCFR); 21 CFR 11.50.