Fourteen Unsigned Batch Records Before a Quality Audit
An internal audit of 60 batch records found 14 with no second-person verification signature. Of those 14, nine carry initials in different ink from the rest of the record, dated two to five days after batch release; three of those nine also show a verification date earlier than the completion date of the step being verified. The records were flagged three days before a client quality audit. The area manager proposes to have the verifiers sign the 14 records now, on the basis that the verification did happen and was simply not written down, and to log the finding as a documentation formatting issue so that it does not trigger a formal deviation.
1. Classify the findings and explain which proposed response is defensible and which is not.
Read the worked answer
There are three findings here, not one, and they carry different severities.
(a) Fourteen records with missing second-person verification - a documentation and compliance failure against the procedure.
(b) Nine records with initials in different ink added after release - a record-integrity question. From the record alone you cannot tell whether this is late documentation of work genuinely done or an after-the-fact signature. That ambiguity is the finding, and it does not resolve in the organisation's favour by assumption.
(c) Three records with a verification date earlier than the completion of the step verified - the record as written cannot be true. This is the strongest integrity indicator and it converts the question from was the signature late to is the record reliable.
Signing the 14 now is the one action that guarantees the worst outcome: it creates a document that appears contemporaneous when it is not, turning a documentation error into a falsified record. The defensible response is to leave the originals untouched, secure the 14 records, and add a dated, signed, attributed late entry that is explicitly labelled as a late entry stating what was done, when, by whom, and why it was not recorded at the time. The nine plus three must go into a formal deviation and investigation. Reclassifying them as a formatting issue removes the trigger for that investigation, which is the actual failure - the missing initials are recoverable, the decision not to look is not.
2. What root cause and corrective action would you expect the investigation to reach?
Read the worked answer
Start by testing whether the problem clusters - by shift, by product, by verifier, by day of week. The three impossible-date records will cluster, and that clustering is what distinguishes a systemic timing problem from individual behaviour. Test the obvious systemic explanations before reaching for individual ones.
Likely systemic candidates: second-person verification is not physically possible at the point of work (the verifier is not on site or not on shift when the step completes); the record is paper and initials are accepted as a substitute for a system timestamp; end-of-shift catch-up signing is tolerated because there is no real-time capture; overnight and weekend coverage lacks a qualified verifier. Each of these produces late signatures without anyone intending to misrepresent anything.
Corrective actions follow from whichever root cause survives. Concurrent electronic verification with role-based sign-off and a system timestamp removes the ink and date problem entirely and prevents late entry by design. System-enforced segregation of duties prevents self-verification. Training is not a CAPA unless the root cause is genuinely knowledge, and here the pattern suggests it is not.
Effectiveness check: 100% record review for three months, target of zero late entries, plus a periodic data-integrity audit that samples for signature-to-timestamp consistency. Extend the scope check to other batches handled by the same verifiers and the same shift pattern, because the 60-record sample bounds nothing.
3. What is reported to the client and the regulator, and who decides?
Read the worked answer
Notification of the deviation and the investigation goes to the client under the quality agreement's deviation and data-integrity provisions, and the decision to notify - like the decision on regulatory notification - belongs to the quality authority with release responsibility, not the area manager. That decision, and its rationale, should itself be documented; an undocumented decision not to notify is indistinguishable from a decision not to look.
Regulatory notification depends on product, jurisdiction, and whether any unverified step affects release, so it cannot be settled inside the audit meeting. What can be stated is the assessment date, the criteria being applied, and who owns the conclusion.
At the client audit, say what is true: the finding exists, it is under formal investigation, the records are secured, containment is in place, and no conclusion on release impact will be asserted until the investigation determines which steps were unverified. Do not claim the batch is unaffected before the scope check supports it, and do not offer the formatting reclassification the area manager suggested.
Removing or softening the finding in the audit trail is itself the reportable act. An audit that finds 14 unsigned records finds a fixable problem; an audit that later discovers the finding was downgraded finds a governance failure, and that is what gets escalated to the client's own quality authority.