CAPA effectiveness checks: how to prove the problem was solved

A CAPA is not effective because every action was completed on time. It is effective when objective evidence shows that the identified causes are controlled and the intended quality outcome is being sustained.

This distinction matters. A revised SOP can be approved, a damaged component replaced and every employee trained while the original problem remains capable of recurring. Action completion answers, “Did we do what we planned?” An effectiveness check answers, “Did the plan work?”

ICH Q10 expects CAPA effectiveness to be evaluated. EU GMP Chapter 1 requires the effectiveness of corrective and preventive actions to be monitored and assessed in line with quality-risk-management principles. An effectiveness check is therefore part of the CAPA, not optional paperwork after it.

What is a CAPA effectiveness check?

A CAPA effectiveness check is a planned, documented evaluation of whether implemented actions achieved their defined objectives without creating unacceptable new risks. It should confirm, as applicable, that:

• The immediate condition was corrected.
• The verified direct, contributing and systemic causes were addressed.
• The same event has not recurred within a meaningful number of opportunities.
• The improved control is being followed and performs as intended.
• Related products, processes, equipment or departments were appropriately covered.
• The change did not introduce adverse or unintended effects.
• The remaining risk is understood and acceptable.

1. Design the check when the CAPA is approved

Effectiveness cannot be judged consistently if the success criteria are invented after results are known. The check should be designed when the action plan is approved and refined through change control if implementation changes the expected outcome.

A complete effectiveness plan should specify:

The plan should be proportionate. A low-risk administrative correction may require a focused record review. A CAPA addressing a potential cross-contamination mechanism may require qualification, multiple production campaigns, environmental or cleaning data and continued trend review.

2. Define criteria that measure the intended outcome

Strong criteria are specific, measurable where practicable, linked to the cause and sensitive enough to detect failure. They measure the performance of the corrected system—not merely the activity used to change it.

Training completion and SOP approval remain implementation evidence. They may be reviewed, but they cannot independently demonstrate sustained performance.

Measure more than recurrence where necessary

“No recurrence” can be useful when events are frequent enough and reliably detected. It is weak when the event is rare, detection is poor or only one process opportunity has occurred. Combine measures where one indicator would create false confidence.

Useful measures may include:

• Recurrence rate for the same event or cause category
• Process or product-performance data against a justified baseline
• Error, rejection, deviation, complaint or alarm-response rates
• Direct observation of correct practice across shifts
• Competency demonstration under normal working conditions
• Targeted audit or self-inspection results
• Preventive-maintenance, calibration or system-challenge results
• Trend, capability or control-chart behaviour where statistically appropriate
• Absence of adverse effects in a related quality attribute

3. Select a review period based on opportunities—not habit

A standard 30-, 60- or 90-day period may be administratively simple, but elapsed time alone does not prove that the corrected control was challenged. The period should capture enough relevant opportunities for the failure to recur if the CAPA were ineffective.

Consider:

• Batch or activity frequency
• Expected recurrence frequency before the CAPA
• Process variability and time required for the change to stabilise
• Product, patient and compliance risk
• Detection sensitivity and data availability
• Different shifts, operators, products, suppliers or equipment configurations
• Campaign manufacture, shutdowns and seasonal environmental conditions
• Complaint, stability or distribution data that emerge slowly

Define both a minimum number of opportunities and a minimum elapsed period when necessary. For example: “ten consecutive campaigns, including at least two night-shift operations, over not less than eight weeks.”

4. Use evidence that is relevant, reliable and traceable

Evidence should relate directly to the expected outcome and come from controlled or otherwise reliable sources. The reviewer should be able to trace the conclusion back to the records examined.

Direct performance evidence

• Batch, laboratory, equipment, maintenance or environmental records
• Electronic audit trails, alarms, access logs or workflow timestamps
• In-process and finished-product trends
• Challenge tests, qualification or verification results
• Direct observation and competency-assessment records
• Complaints, returns, deviations and rejected-batch data

Implementation evidence

• Approved revised procedures and forms
• Completed engineering or software changes
• Training and competence records
• Updated maintenance, calibration or supplier controls
• Change-control completion and qualification approval

Both may be needed, but implementation evidence should not be presented as performance evidence. A screenshot showing that an electronic field was added proves configuration; completed records and exception data show whether the field controls the process.

Choose the sample before seeing the results

Define the population, sampling method and exclusions in advance. Samples should represent the risk: shifts, products, operators, equipment, suppliers and event types relevant to the cause. Selecting only readily available or known-good records biases the conclusion.

There is no universal GMP sample size for CAPA checks. The quantity should be justified by risk, frequency, variability, detectability and the strength of other evidence. Statistical methods can help where data volume and the decision justify them, but an arbitrary percentage does not become scientific simply because it is written in an SOP.

5. Assign an objective reviewer

The CAPA owner can assemble the evidence, but the effectiveness conclusion should receive independent Quality review. Higher-risk or technically complex CAPAs may need engineering, validation, microbiology, IT, regulatory or process experts.

The reviewer should confirm that the original problem and causes were correctly represented, criteria were not weakened after seeing the data, exclusions were justified, adverse trends were considered and the evidence supports the decision.

6. Make one of four honest closure decisions

“Inconclusive” should not become a route to indefinite administrative closure. Explain why evidence is insufficient, what will be collected next, who is responsible and how risk is controlled meanwhile.

If the CAPA is ineffective, do not merely repeat the same action. Reconsider the original investigation, causal assumptions, scope and product impact. A failed effectiveness check is new quality information.

Composite field examples

The following anonymised composite scenarios reflect situations commonly encountered in pharmaceutical operations. They are provided to demonstrate the decision logic, not to prescribe universal acceptance limits.

Example 1: repeated missing times in batch records

Problem: Equipment-cleaning steps were signed, but completion times were repeatedly omitted. The initial proposal was to retrain all operators and verify 100% training completion.

Cause and action: The investigation found that the form grouped one time field beneath four separate activities, while the SOP required a time for each activity. The form was redesigned, document review instructions were clarified and operators and reviewers demonstrated use of the revised record.

Effectiveness plan: Quality sampled 15 records from each of four shifts over three months and observed one cleaning activity per shift. The criteria required every critical cleaning step to be traceable, at least 98% completion of required time fields, no repeat deviation from the same form-design cause and reviewer detection of any planted training-example omission.

Decision: One early noncritical omission occurred in the first week, followed by sustained compliant performance. Because the predefined criterion allowed up to 2% noncritical omissions and all other criteria were met, the CAPA was considered effective. The early omission remained documented; it was not deleted from the assessment.

Example 2: tablet-weight variation after a maintenance failure

Problem: Intermittent low tablet weights were linked to movement in a feeder-drive locking component. The maintenance instruction did not include a defined locking-integrity check.

Actions: The affected component was replaced; similar presses were inspected; the preventive-maintenance instruction, spare-part standard and technician competency check were updated; and process-alarm capability was reviewed.

Effectiveness plan: Evidence covered the next ten qualifying batches on the affected press over not less than eight weeks, the next planned maintenance inspection and a targeted check of comparable presses. Criteria included no recurrence attributable to feeder-drive movement, acceptable tablet-weight trend behaviour, completion of the defined torque and integrity checks, and no adverse increase in compression stops or rejects.

Decision: Product data met requirements, but the first scheduled maintenance record lacked the required measured value. The equipment outcome appeared satisfactory, yet the systemic maintenance control was not reliably implemented. The CAPA was classified as partially effective, the maintenance workflow was strengthened and a further effectiveness period was required.

Example 3: delayed response to a warehouse temperature alarm

Problem: A high-temperature alarm generated overnight was acknowledged after 47 minutes. The escalation contact list was outdated and responsibility during security handover was unclear.

Actions: Escalation roles and contacts were revised, the system was configured for secondary notification, handover checks were added and response drills were performed.

Effectiveness plan: Monthly challenge tests covered weekday, weekend, day and night conditions for six months, including the hot season. Criteria required delivery to primary and secondary contacts, acknowledgement within the approved time, documented escalation where the first contact did not respond, and no missed handover verification.

Decision: The first four challenges passed. The fifth exposed a weekend delay caused by an uncommunicated roster change. The CAPA was ineffective because the same escalation vulnerability remained, even though no commercial batch had experienced another temperature excursion. The failure was valuable evidence that the system needed a roster-independent escalation control.

Common effectiveness-check mistakes

• Defining success only as completion of training, SOP revision or equipment repair
• Writing criteria after reviewing the outcome
• Using “no recurrence” when there were too few opportunities or weak detection
• Applying the same 30-, 60- or 90-day period to every CAPA
• Selecting a convenient sample that excludes difficult shifts, products or equipment
• Changing acceptance criteria when the original target is missed
• Ignoring related indicators that worsened after the action
• Reviewing only the department’s summary instead of traceable source data
• Allowing the CAPA owner to approve their own effectiveness conclusion without Quality challenge
• Extending inconclusive checks repeatedly without a risk-based endpoint
• Closing a partially effective CAPA without controlling the residual weakness
• Failing to reopen the investigation when evidence contradicts the original root cause

The practical effectiveness-check sequence

1. Restate the original problem, affected scope and supported causes.
2. Map each CAPA action to the cause or risk it is intended to control.
3. Define the expected process or quality outcome.
4. Select direct measures and supporting indicators.
5. Record the baseline and predefined acceptance criteria.
6. Define the population, representative sample and exclusions.
7. Set a review period based on sufficient process opportunities and risk.
8. Identify controlled evidence sources, reviewer and escalation rule.
9. Confirm implementation before beginning the performance period.
10. Collect and evaluate all relevant results, including failures and unintended effects.
11. Classify the outcome as effective, partially effective, ineffective or inconclusive.
12. Document the rationale, residual risk and required follow-up.
13. Feed the outcome into deviation trends, management review and continual improvement.

Educational notice: This article provides general educational information. Manufacturers should follow their approved procedures and apply product and process knowledge, quality-risk management, marketing-authorisation commitments and the requirements of their applicable regulatory authorities.