CAPA effectiveness check: why does a closed CAPA come back?

A CAPA is closed, the evidence is filed and the same finding turns up again at the next audit. Practitioners are regularly discussing that situation on Reddit, and we follow those conversations to share what is being discussed and add to it.

This piece gives quality managers at UK companies a way to define and test effectiveness before closure. It covers how to write the criterion when the CAPA is opened, how to test why the problem went unnoticed and what evidence shows recurrence has stopped. It includes one worked example with a defined recurrence window.

A CAPA effectiveness check is the planned test of whether a corrective or preventive action has stopped the problem it was opened for. It is separate from confirming that the action was carried out.

Action completed, action verified and effectiveness confirmed are three different results

Closed CAPAs come back when these three results are recorded as one tick. Each needs its own evidence and its own date.

• Action completed: the task was done. The procedure was revised, the template was changed or the training was delivered. The evidence is a dated record of the task.
• Action verified: the action was checked against what it was meant to do. The revised template produces the right output or the trained staff can perform the task. The evidence is a test or an observation of the change itself.
• Effectiveness confirmed: the problem has not recurred over a defined window, measured against a criterion set at the start. The evidence is data collected after the change has been running.

The order matters. Verification can happen within days of completion. Effectiveness cannot be confirmed until the window has run. A CAPA closed at the first or second stage has not yet been tested against the thing that brings findings back.

Write the effectiveness criterion when the CAPA is opened

A criterion written at closure is shaped by what the team already knows happened. A criterion written at opening sets the test before the result is known. A usable criterion answers three questions in plain terms.

What will be measured

Name one measurable signal tied to the original finding. It might be a count of repeat findings, a rejection rate, a number of late calibrations or the percentage of records with a missing field. Avoid writing "no recurrence" on its own. The record should also say where a recurrence would show up: which log, which report or which audit.

Over what window

The window is the period the measure must stay clear. Set it from how often the process runs and how often its conditions change. A daily process shows a pattern sooner than a quarterly one. Include at least one cycle of whatever triggered the original problem, such as a changeover, a shift pattern or a supplier delivery. The length is a choice for the organisation and the CAPA owner to make, and the reasoning belongs in the record.

Who confirms it

Name the reviewer and the review date when the CAPA opens. The reviewer should not be the person who owns the action. A quality manager from another area or a process owner with no stake in the fix looking successful will read thin evidence more critically.

Test how the original problem went unnoticed

A CAPA usually concentrates on why the problem happened. The effectiveness check has to answer a second question: why did the controls already in place not catch it? A fix that removes the cause but leaves the same blind spot means the next problem can go undetected for just as long.

The effectiveness check must also test why existing controls missed it. ASQ’s eight disciplines (8D) guide, a structured problem-solving method, includes steps to verify root causes, verify the effectiveness of corrective action and prevent recurrence. It also asks the team to identify why the problem went unnoticed when it occurred.

In practice, that means testing the detection point itself. Run a sample of records through the same inspection, or introduce a known error and see whether the control catches it. If it does not, the CAPA needs a detection action as well as a cause action, and the effectiveness criterion should include the result of that test.

A worked example: a recurrence window with a defined success criterion

The example below is illustrative and uses invented figures. A UK distribution site found wrong lot codes on 4 of 600 outgoing pallet labels in one quarter. The root cause was manual keying of lot codes at the label printer. The existing control was a despatch operator comparing the label to the paperwork by eye, which missed transposed digits.

The actions were a label template that pulls the lot code from the stock system and a scanner check at despatch that compares the label barcode to the order. Completed: 3 March. Verified: 17 March, when a test run of 50 labels matched the stock record and the scanner check was observed on three shifts.

The criterion, written when the CAPA opened, reads as one sentence: zero lot code mismatches across all outgoing pallets for 90 days from 17 March, measured from the weekly scanner exception report, plus a monthly seeded test in which a deliberately wrong label must be rejected 10 times out of 10. The despatch supervisor collects the data. A quality manager from another site confirms the result on 15 June.

The site chose 90 days to cover three monthly stock cycles. Another site could reasonably choose a different window. If one mismatch appears at any point, the CAPA stays open and the root cause is reviewed. If a seeded error passes the scanner once, the detection action has failed even when no real mismatch occurs. If both measures are met, effectiveness is confirmed on 15 June and the CAPA closes with the data attached. That gives three dated results: completed on 3 March, verified on 17 March and effective on 15 June.

What evidence shows recurrence has stopped

Evidence that recurrence has stopped is data collected after the action and set against the criterion written at the start. A signed statement that the team believes the problem is fixed is not evidence.

Four kinds of evidence carry weight. The first is the measured result for the full window, with the source report attached. The second is the result of the detection test, showing the control now catches the error. The third is a search of other records for the same root cause, such as complaints, audit findings and nonconformances in other areas or sites, because a recurrence can appear in a different place under a different title. The fourth is the reviewer’s decision, with their name and date.

A missed criterion is a result and not a failure of the process. The CAPA stays open, the root cause is reviewed and the criterion is kept as written, so the second attempt is judged against the same test.

Corrective and preventive action software for UK companies: what to confirm before closing a CAPA

Before closure, a quality manager can run a short check. Does the record hold the criterion, the window, the data source and the named reviewer? Are completion, verification and effectiveness dated separately? Has the detection test been run? Has a search for related records been done?

Ideagen’s post on why corrective actions fail to stick makes the point that actions should not be closed on implementation alone, and it covers the wider process gaps for readers who want them.

Corrective and preventive action software supports this check when it enforces the sequence. It should hold the criterion fields as required entries at opening, keep completed, verified and effective as separate statuses, schedule a review task for the named reviewer at the end of the window and link CAPAs that share a root cause. A spreadsheet or an email thread can hold the same fields. The difference is whether anything stops closure until they are filled.

Five closure habits that bring CAPAs back and how to replace them

What gets closed too early	What to set at opening	What the quality manager gets
The CAPA closes the day the revised procedure is signed.	Record completed, verified and effective as separate statuses with separate dates.	A closure record that shows which of the three has actually happened.
The criterion says "no recurrence" with no window.	State the measure, the window and the data source when the CAPA opens.	A pass or fail result at review that nobody has to interpret.
The action owner signs off their own effectiveness.	Name an independent reviewer and a review date at opening.	A check that does not rely on the person who wrote the fix.
The same finding returns in another area and is logged as new.	Search related records for the same root cause across areas and sites.	Earlier sight of a root cause that reaches beyond the original scope.
The fix targets the cause and the failed detection point is never examined.	Test the control that missed the problem with a sample or a seeded error.	Confidence that a repeat would be caught at source.

A CAPA is closed when the criterion is met, not when the action is done

Closure is a decision that the data makes. Write the criterion at opening with the measure, the window and the reviewer. Test why the existing controls missed the problem. Keep completed, verified and effective as separate results. The window is the organisation’s choice, and the CAPA stays open until the result is in.

To continue the discussion with other practitioners, join Ideagen’s Reddit community.