Correcting a batch record: why systematically sending it back to production is counterproductive

During a batch record review, a signature is missing on a somewhat critical step. The record goes back to production. It comes back the next day with the signature, carefully added, without the slightest annotation. As if everything had been perfect from the start. How many invisible corrections has this record undergone before landing on Quality’s desk?

“I send it back so they see their mistake”: the method that teaches nobody anything

If sending records back to production really helped teams improve, the error rate would be very low and right first time excellent. That’s generally not the case.

Take a rounding error that doesn’t change whether the result is compliant. Corrected straight away, it takes two minutes. Sent back to production, it takes five minutes to handle, five to ten minutes for the person to get back into the subject, and five minutes to check the correction: about ten times longer. Not counting the days of waiting, even more if the team concerned works weekends or on rotation. And often, it’s the team leader who makes the correction, to save time.

Instead, keep a log of errors. It brings out design flaws in the record, and if it’s often the same person, a conversation with them will be far more useful than a hasty correction.

“I can’t sign on their behalf”: ethics on the surface

If you have the skills to check a batch record, you have the skills to correct what is obvious, provided you have the information. The nuance matters: you can’t add a data item that nobody is sure of. In that case, it’s a potential quality risk, and it goes through the deviation management system.

But is it really more responsible to ask someone to sign, three weeks later, for an activity they no longer remember? Do you remember what you ate five days ago?

What about data integrity?

Everyone is trained in ALCOA principles, and then a signature is allowed to appear as if by magic, with no date or explanation. The message sent is clear: filling in the record in real time is optional. Yet the purpose of a batch record isn’t to make people believe everything was perfect, but to guarantee the traceability of activities. Every correction must therefore be annotated: when, why, by whom.

Correct intelligently, not systematically

• Correct what you are sure of: a calculation error, a date that can be traced, a check that can be made from the raw data.
• If in doubt, talk before sending it back: a short message to the person concerned often avoids an unnecessary round trip.
• Record all after-the-fact corrections: annotated, dated and signed.
• Set clear rules for direct corrections: minor, traceable and justified corrections.
• Manage it: number of records sent back per week, processing time, recurring errors.

Keep it in perspective: excellence, not perfection

Finding three missing data items out of two hundred is less than 1.5% of omissions. That’s already a good score. Fewer send-backs means fewer records waiting, smoother release, and more time for what matters: understanding where errors come from and simplifying the record so they don’t happen again.