Writing a CAPA that actually closes

A customer audit or a regulatory inspection ends and the observations arrive. The site responds. Weeks later the response comes back marked inadequate, and now there is a second problem on top of the first: the authority or the customer has evidence that your investigation process does not work.

The corrective action is rarely what fails. These three things are.

1. The root cause is a restatement of the observation

Observation: Batch record for batch 24-0117 had two entries corrected without the correction being dated and initialled.

Root cause: The operator did not date and initial the correction.

That is not a root cause. It is the observation written in the past tense.

The question that has not been asked is why — and it has to be asked until the answer is something the site can change. Was the operator trained on good documentation practice? Was the training assessed for effectiveness or just attended? Does the batch record have space to make a correction properly? Does the line supervisor review entries at a point where the error could still be caught? Is the production schedule such that documentation happens at the end of a shift, from memory?

Each of those has a different corrective action. Retraining the operator addresses none of them.

2. The correction is presented as the corrective action

These are separate things and reviewers check for both.

• Correction — what you did about this instance. The record was reconciled, the batch was assessed, product impact was evaluated.
• Corrective action — what you changed so that the cause cannot produce the same failure again.
• Preventive action — where else the same cause could be operating, and what you did there.

A response that says “the operator was retrained and the record was corrected” contains a correction and a training event. It contains no corrective action, because nothing about the system changed. If the same operator is replaced next month, the site is back where it started.

3. There is no effectiveness check, or it is not a check

The commitment that closes a CAPA is the one that demonstrates the fix worked. That means:

• A measurable indicator. Documentation error rate per batch record reviewed, not “improved compliance”.
• A defined period. Reviewed across the next twenty batches, or for three months — something with an end date.
• A pre-stated threshold. What result means it worked, and what result means it did not.
• A stated consequence. What happens if the threshold is not met, i.e. the CAPA reopens and escalates.

“Effectiveness will be monitored during routine review” is not an effectiveness check. It is a plan to look at things.

The scope question everyone skips

The single most common cause of a repeat observation is that the original CAPA was scoped to the batch that was inspected.

If the cause was a batch record design that gives no room for a proper correction, that design is probably in use across every product on that line. If the cause was that effectiveness of documentation training was never assessed, that applies to every trained person on site.

Ask explicitly: where else does this cause exist? Write the answer into the response, even where the answer is “nowhere else, and here is why”. A reviewer who can see that you asked the question is far more likely to accept the conclusion.

A response that reads well

Structure it so a reviewer can follow it without reconstruction:

1. The observation, quoted as written.
2. Immediate correction and product impact assessment, with the batches affected.
3. Investigation — what was examined, what was ruled out and why.
4. Root cause, stated as a system condition.
5. Corrective action, with owner and date.
6. Preventive action across the assessed scope.
7. Effectiveness check: indicator, period, threshold, consequence.
8. Attachments referenced by number.

Nothing here is difficult. It is simply the discipline of answering the question that was asked, rather than the question that would be easiest to close.