CAPA FDA: Requirements, Records and Inspection Preparation

This guide helps quality teams review a CAPA procedure or assemble the evidence behind an individual investigation. It covers applicability, data sources, causal reasoning, action records, effectiveness, management review and a practical record worksheet. The checklists are review aids, not a guarantee of a particular inspection outcome.

The primary sources for the opening distinction are 21 CFR 211.192, FDA's ICH Q10 guidance, and FDA's QMSR overview.

For submission-facing work, keep the link between the quality record and any regulatory assessment explicit. The CAPA-to-eCTD guide and quality records for regulatory submissions address that handoff. A CAPA record is not automatically a submission-ready document or proof that a regulatory filing is unnecessary.

Establish which framework applies

Do not select an obligation from the phrase “CAPA FDA” alone. Identify the product, regulated activity, market, relevant record and period being assessed. An older inspection record and a current procedure can require different historical context.

For a drug failure covered by section 211.192, the investigation duty applies whether or not the batch has been distributed. A threshold for opening a separate CAPA record does not cancel the investigation. Review the work performed, not just the name of the form used.

For devices, FDA identifies February 2, 2026 as the QMSR effective date and incorporates ISO 13485:2016. Its current inspection information also describes the replacement of QSIT by Compliance Program 7382.850. Use the current FDA QMSR information when reviewing a present-day procedure. Historical citations to 820.100 can remain relevant to older findings when their date and context are clear.

For biologics or mixed operations, identify additional product-specific requirements rather than treating a broad product label as a complete applicability decision. Record unresolved questions, their owner and the evidence needed to answer them. A missing classification should produce a review question, not an assumed framework.

Distinguish the event, correction and causal action

A deviation or nonconformance record can contain a substantial investigation. It is misleading to assume deviations address only an immediate cause while CAPAs alone investigate systemic conditions. Organizations structure records differently; the required work and evidence still need to be traceable.

FDA's September 2006 Quality Systems Approach guidance, sections III.D and IV.D.4–5, distinguishes remedial correction, corrective action and preventive action. Use that nonbinding terminology guidance alongside the current applicability assessment.

For example, correcting an erroneous entry does not explain why the wrong field was used. Retraining on an ambiguous form may leave the design problem intact. Conversely, reviewing a new form before use can identify and address that potential problem without waiting for a failure.

Review seven connected areas of the evidence

The following areas organize a practical review. They do not reproduce the former seven subsections of 820.100 and are not presented as a universal statutory checklist. For an operational walkthrough, use the CAPA process guide.

1. Quality signals and escalation decisions

Check whether the quality system captures relevant information from production, laboratory results, complaints, audits, returned material, suppliers, equipment and process monitoring. Include sources appropriate to the operation; a drug site and a device service organization will not have identical data streams.

Review the procedure's escalation criteria together with examples of how they were applied. A threshold can help manage work, but it should not postpone an investigation already required for a covered event. Keep a documented basis for the escalation decision and the next responsible person.

When reviewing a regulatory observation, identify the actual concern, product context, response request and work needed. Do not infer a universal response timeline or a mandatory separate CAPA form solely from the fact that an observation exists.

2. Investigation scope and causal reasoning

Trace the investigation from the factual problem statement to the conclusion. Check the expected condition, observed result, timeline, affected identifiers and evidence supporting the assessed extent.

Tools such as Five Whys, fishbone analysis, fault trees, comparisons of affected and unaffected conditions, and FMEA can organize different questions. Calling a tool “FDA-accepted” without a relevant source does not establish the adequacy of an investigation. A diagram or score also does not replace evidence.

For each material conclusion, ask what supports it, what contradicts it and what alternatives were considered. Determine whether the proposed explanation accounts for the event pattern. If “human error” is the conclusion, examine what is known about instructions, interfaces, workload, training and available controls.

An unresolved root cause should remain identified as uncertain. Record the remaining hypotheses, the risk response and further work. Do not require a definitive statement when the records support only a range of explanations.

Review related batches, products and processes where the evidence indicates a connection. For covered drug investigations, the broader scope is expressly addressed by section 211.192, linked above. Keep the reason for including or excluding an area visible in the record.

3. Action selection and planning

Check that each action responds to the supported cause or identified potential problem. A generic instruction to “retrain,” “review procedures” or “improve maintenance” does not explain what will change or how its effect can be assessed.

A practical action plan identifies the exact work, owner, justified due date, dependencies, interim controls, implementation evidence and intended outcome. Separate tasks with different owners rather than making one person nominally responsible for an undefined cross-functional effort.

Consider whether the action creates another risk or changes a validated condition, specification, supplier arrangement or regulatory commitment. Use the relevant change-control assessment. The same type of CAPA can lead to different changes depending on the actual cause and operating conditions.

If a proposed action addresses only a symptom, return to the causal explanation. For example, replacing a failed part can restore equipment operation, while understanding and controlling the failure mechanism may require additional work. State which problem each action is intended to solve.

4. Implementation and controlled records

Confirm that the approved action became the intended operating condition. Evidence can include an approved procedure revision, equipment change record, applicable qualification or validation results, controlled distribution and personnel assessment.

Check the exact version and scope. A signed training record for the old procedure does not demonstrate preparation for the revised task. A change request marked complete does not show that obsolete instructions have been removed from the point of use.

Retain traceable links among the CAPA, change control, training, technical records and approvals. If the organization changes the action plan during implementation, preserve the reason and the relationship between versions.

For electronic records, assess the scope of 21 CFR 11.1 and applicable controls. Neither the purchase of software nor a “compliant” vendor label establishes that the configured workflow and records meet every requirement.

5. Effectiveness evidence

The central question is whether the evidence supports the intended outcome. Implementation and effectiveness are related but distinct. An SOP approval may prove that a document changed; it does not, by itself, show that the original failure mechanism is controlled.

Select the approach for the failure mode and risk. A focused task observation, process study, record review, trend analysis or audit may answer different questions. Where an independent reviewer is useful, assign one through the quality process; this guide does not assert that a particular staffing arrangement is universally required.

The cited CAPA guidance does not establish fixed 30–90-day monitoring or a 90–180-day total lifecycle for all cases. A long waiting period can still provide weak evidence if the relevant process rarely runs. Conversely, evaluate the actual evidence rather than assuming that a particular calendar interval automatically proves or disproves effectiveness.

6. Information flow and management decisions

Review how the people responsible for affected activities receive the information needed to act. An action plan can fail when a revised procedure is approved but the operating team continues to use the earlier version or a supplier never receives the relevant change.

Use management review to address unresolved risk, resources, repeated mechanisms and constraints on completion. Identify the decision, responsible person and followup. A dashboard that reports overdue actions without assigning a response may not help resolve them.

For pharmaceutical quality systems, ICH Q10 section 3.2.2 and Table II, linked above, discuss CAPA in the lifecycle framework. Map recommendations and applicable requirements to the actual process; do not describe all pharmaceutical management activities using former device subsection numbers.

7. Closure, unresolved work and trend feedback

Examine what the closure decision means in the organization's procedure. Check whether implementation, effectiveness, product impact and followup were evaluated and whether the conclusion follows from the records.

If additional work remains in a separate record, identify the dependency, owner and risk control. Administrative closure should not make unresolved work disappear from oversight. Retain unsuccessful checks and reopen or escalate through the approved process when the evidence requires further work.

Feed the resulting information back into monitoring and procedure review. Search for recurring failure modes rather than only matching CAPA titles. A new record number can hide the recurrence of the same underlying condition.

Worked review: a repeated investigation after training

This is an illustrative review exercise, not a reported FDA case or inspection finding.

A packaging team opens a new record for entries made in the wrong form field. The reviewer finds an earlier investigation with a similar description. That record concluded “operator error,” documented retraining and closed after no further errors were reported during a quiet production period.

The first review question is whether the events share a mechanism. Compare form versions, instructions, operators' task conditions and production activity. Do not assume the newer event is identical simply because its title matches, and do not assume it is unrelated because it has a different record number.

Suppose the new evidence identifies two confusing field labels and an instruction using an obsolete name. The earlier training did not change those conditions. A suitable plan might assess affected records and product, correct the form/instruction, control the versions in use and define a task-relevant assessment. Those are illustrative actions that still need technical and quality review.

Now vary the evidence:

• The revised process has not been used: document the lack of observation opportunities and the next justified assessment. Do not count inactivity as successful routine operation.
• The process was used, but errors were no longer recorded consistently: resolve the measurement gap before claiming an improved trend.
• The intended process was used under relevant conditions and the planned criteria were met: evaluate the evidence and retain the basis and limits of the conclusion.
• The error recurred: reassess the explanation, implementation and control. Adding another training task without examining those branches may repeat the earlier failure.

The example preserves three separate judgments: whether the proposed cause is supported, whether the action was implemented, and whether the outcome supports effectiveness. A single “closed” status does not answer all three.

A practical CAPA record worksheet

Use this structure as a review aid. Adapt it to the applicable requirements and controlled procedure; it is not an FDA form or a certification of compliance.

Test retrieval, not just the presence of links. Open the referenced record and check its version and relevance. If an attachment is missing or access is limited to a departed employee, assign recovery or access work rather than marking the evidence complete.

Determine retention from the actual product and record requirements and the approved schedule. A CAPA record can link several record types with different obligations. This article does not supply a universal retention period for all of them.

Timelines and metrics without invented FDA benchmarks

Plan dates from risk, complexity, dependencies and interim controls. Record why a proposed extension is justified, what remains unresolved and who approves the next step. A fixed day-count does not explain the urgency or adequacy of every investigation.

The following metrics can support internal review. Their usefulness depends on defined populations, periods and classifications.

If calculating a rate, define the numerator and denominator consistently. With no eligible records, report that the rate is not applicable. Do not infer perfect performance from an empty period. Internal targets such as an average age below 90 days or an effectiveness rate above 90% need an organizational rationale; they are not supplied here as FDA benchmarks.

Use inspection information with the right context

An inspection observation, warning letter, guidance document and regulation have different roles. Do not convert a passage from one firm's historical finding into a universal procedural rule. Identify its product scope, date, facts and the obligation it discusses before using it in training or a checklist.

Likewise, a link to an observation database does not substantiate a “top ten” CAPA ranking. A frequency claim needs a stated population, period, denominator and classification method. This guide uses practical review categories rather than presenting unsupported enforcement statistics.

Before an inspection, select records that exercise different inputs, failure modes and stages. Trace them from original signal to current status. Review open work, unsuccessful effectiveness checks, recurring conditions, scope decisions and the ability to retrieve supporting records. Do not assume an inspector will request a fixed number of CAPAs or a universal lookback period.

For a current device inspection, use FDA's current QMSR information linked above. For a pharmaceutical procedure, keep the relevant drug CGMP obligations and guidance references explicit. A historical QSIT citation alone is insufficient to establish a current inspection approach.

Next steps

Choose one current procedure and one investigation record. Establish the applicable framework, use the worksheet to locate the evidence, and record gaps with owners and justified next actions. Test whether the documented conclusion would still hold if the process had little activity or the measurement approach changed.

Use the FDA CAPA requirements guide for an applicability review and the CAPA process guide for the operational sequence. For submission-facing changes, retain the connection to the underlying quality records and the appropriate regulatory assessment.