Complete batch record before review, not during it

The record does not always reach review blank. Often it arrives almost complete: the equipment used in one step is not identified, a weighing was entered at the end of the shift, a verification signature was left for later, an in-process result was copied by hand from a printout. Each gap looks small. Together they leave quality review reconstructing what happened instead of checking what was recorded.

A complete record is not the same as a reviewed record. Review confirms that the record is correct. It should not be the point at which the record becomes complete. When review starts completing the record, the batch may be approved on information reconstructed after execution.

What the rule asks for, and when

21 CFR 211.188 requires batch production and control records to include "complete information relating to the production and control of each batch." The list includes documentation that each significant step was accomplished, with dates, identity of individual major equipment and lines used, identification of each batch of component or in-process material, weights and measures, in-process and laboratory control results, inspection of the packaging and labeling area before and after use, actual yield and percentage of theoretical yield, any sampling performed, and identification of the persons performing and directly supervising or checking each significant step.

The expected content comes from the master record. 21 CFR 211.186 defines what the master production and control record must contain, and 211.188(a) requires the batch record to carry an accurate reproduction of the appropriate master record, checked for accuracy, dated and signed. If the master record does not provide for a field, the batch record is unlikely to carry it.

The timing comes from 21 CFR 211.192: all production and control records must be reviewed and approved by the quality control unit "before a batch is released or distributed," and any unexplained discrepancy must be thoroughly investigated, extending to other batches that may be associated.

In Brazil, ANVISA RDC 658/2022, Art. 124, states that records must be made or completed whenever an action is taken, so that all significant manufacturing activities are traceable. Art. 125 requires every change to be signed and dated and to keep the original entry readable. For computerized systems, IN 134/2022, Art. 32, states: "Para os registros que dão suporte à liberação dos lotes, deve ser possível gerar impressões indicando se algum dos dados foi alterado desde a sua inserção original" (records that support batch release must allow printouts showing whether any data has been changed since original entry); and Art. 33 calls for a risk-based audit trail of GMP-relevant changes and deletions.

Both readings point the same way: the record is made when the step happens, and batch record review and approval happen on a record that is already complete.

Where records tend to be incomplete

In practice the gaps repeat in a few places:

• Equipment not identified. The step is recorded, but not the equipment or line it ran on. Without it, the batch cannot be tied to that equipment's cleaning, calibration or maintenance status.
• Value entered later. The weight or parameter was written on scrap paper and moved to the record at the end of the shift. The value may be right, but it is no longer contemporaneous.
• Verification without a verifier. The step requires a second person, and the verification field is empty or was signed later by someone who did not witness execution.
• Transcribed result. The in-process or laboratory result was copied from the printout or instrument, and the batch record does not point to the original data.
• Deviation not linked. Something happened during the batch, but the record does not reference the investigation that 211.188(b)(12) requires.

None of these depends on bad faith. They tend to appear when the record accepts being filled in later.

What FDA warning letters show about review

Two recent warning letters, listed in the references, show the effect on review.

A July 2026 letter [ref. 5] to an over-the-counter drug manufacturer in India notes, among other points, "multiple discrepancies between the source data in your analytical data books and the final values reported in your batch manufacturing records or certificates of analysis" and states that the quality unit "signed off on these records without ensuring the consistency and accuracy of the data." The same letter says the firm "failed to make the required entries in your equipment logbooks to reflect the use of multiple instruments for the specific batch release tests performed."

In an April 2026 letter [ref. 4] to a finished pharmaceuticals manufacturer, under 21 CFR 211.192, FDA states: "Your QU failed to perform investigations into OOS results obtained during trial injections." Under 21 CFR 211.160(a), it cites the lack of a written procedure to review audit trails and raw analytical data "to ensure data reliability for batch release."

In both cases review took place, but on a record that did not contain, or did not link to the original of, everything that was generated. Data integrity on the shop floor is covered in ALCOA+ on the floor.

Warning or hard stop

The distinction that weighs most in system design is between warning and blocking.

A warning reports that a field is empty and lets the step continue. A hard stop prevents the step from being completed, or the record from moving to review, until the required field is filled in by someone with current training on the SOP for that step.

Where the electronic batch record requires equipment identification before a step starts, checks it against equipment status, captures the instrument value with operator and time, and does not let the record move to review with a required field empty or a deviation unlinked, review checks a complete record. That does not guarantee compliance on its own, nor replace quality unit review. What changes is that the gap shows up during execution, while it can still be corrected with the original information, and not at review. With the record complete at execution, review can focus on exceptions, as described in review by exception.

Too many hard stops also have a cost: if every field halts operations, the shift starts working around the system. It is worth setting hard stops through risk assessment, based on the steps the master record treats as critical, and leaving the rest as warnings.

A question for your next batch

In the last batch record reviewed, how many corrections did review make, and how many of them could the system have required during execution?

If most corrections are equipment identification, verification signatures or values entered later, review is completing the record, and approval rests on that reconstruction.

References

1. eCFR — 21 CFR 211.186, Master production and control records
2. eCFR — 21 CFR 211.188, Batch production and control records
3. eCFR — 21 CFR 211.192, Production record review
4. FDA — Warning letter, Ava Inc., 04/14/2026
5. FDA — Warning letter, Dabur India Limited, 07/24/2026
6. ANVISA — RDC 658/2022, Good Manufacturing Practices for medicines (Portuguese)
7. ANVISA — IN 134/2022, computerized systems (Portuguese)