CAPA Effectiveness Check

An effectiveness check is a documented, criteria-based verification — performed after the corrective and/or preventive action has been implemented and after enough time has passed for recurrence to have manifested — that the CAPA achieved its intended effect. It is not the verification that the action was performed (that is implementation verification). It is the verification that the action worked.

The distinction matters. Updating an SOP, training operators on the new revision and signing the training records is implementation. Confirming that the failure mode the SOP was meant to prevent has not recurred in the following 90 days across the relevant production volume is effectiveness.

• 21 CFR 820.100(a)(4) — verifying or validating the corrective and preventive action to ensure that such action is effective and does not adversely affect the finished device.
• 21 CFR 820.100(a)(7) — ensuring that information related to quality problems or nonconforming product is disseminated to those directly responsible.
• ISO 13485:2016 §8.5.2(d) and §8.5.3(d) — reviewing the effectiveness of the corrective / preventive action taken.
• ICH Q10 §3.2.2 — the CAPA system should include an evaluation of the effectiveness of the actions taken.
• EU GMP Chapter 1 §1.4(xiv) — corrective and preventive actions should be evaluated for their effectiveness.
• FDA QMSR (effective 2 February 2026) — incorporates ISO 13485:2016 by reference, including §8.5.2/§8.5.3.

1. CAPAs are closed at implementation. The team finishes the action items and closes the record — but no one returns 60 / 90 / 180 days later to check whether the problem recurred.
2. Criteria are subjective. 'Operators trained' is not a criterion for whether the deviation pattern has stopped. 'Zero recurrence of the OOS over the next 6 months across at least 30 batches' is.
3. Measurement period is undefined or too short. Closing a recurrence check after one batch tells you almost nothing about whether the root cause is fixed.
4. Verification of action is treated as effectiveness. Auditors hear 'we trained 27 operators and have signed records' and ask 'and did the deviation rate go down?' — the gap between the answers is the finding.
5. Effectiveness is checked against the symptom, not the root cause. A CAPA on a labelling deviation might be 'effective' on labelling errors but ineffective on the underlying batch-release-pressure root cause.
6. Effectiveness checks are performed but the result is not documented as a closure decision — the file shows the data was gathered but not what conclusion was drawn.

Good criteria share four properties: they are quantitative or objectively verifiable; they have a defined measurement period and population; they are anchored to the root cause; and they were defined before the action was implemented (not retrofitted at closure).

The measurement period must be long enough for recurrence to plausibly manifest. For a daily-occurring deviation, 30 days might be sufficient. For a deviation that occurs once per quarter, 30 days is meaningless and the measurement period must be at least 6-12 months. The period should be justified in the CAPA record at the moment the action is defined.

It is acceptable to close the CAPA at implementation and run the effectiveness check as a follow-up task with its own record, provided the original CAPA explicitly defines the criterion, the period and the responsible owner. What is not acceptable is to close the CAPA at implementation with no commitment to measure effectiveness.

An effectiveness check that fails is a positive outcome of the quality system. It tells you the action did not address the root cause — usually because the RCA was incomplete. The correct response is to reopen the CAPA (or open a new one), redo or extend the RCA, define a new action, and run a new effectiveness check. Closing the failed CAPA as 'effective' because you do not want to reopen it is fraud.

Preventive actions are harder to measure because there is no recurrence to count — the failure has not happened. Criteria for preventive actions are typically leading indicators: the absence of the precursor condition the preventive action was meant to eliminate, a trend metric that the action was meant to improve, or a process-capability index that the action was meant to raise. The discipline is the same: define the criterion, the period and the data source in advance.

• Defining the criterion at closure rather than at action definition.
• Treating implementation verification ('operators trained') as effectiveness.
• Measurement period too short for the failure frequency.
• Effectiveness check checks the symptom, not the root cause.
• Failed effectiveness check quietly closed as 'effective with caveats'.
• No data source identified — the check becomes anecdotal.
• Repeated CAPAs on the same root cause without anyone noticing that the previous effectiveness check should have caught it.