Start with four questions
Closing a deviation record does not establish that the underlying problem has been solved. A useful investigation tells a clear story that someone who was not there can follow.
- What happened? Define the event and the unmet requirement.
- Why did it happen? Establish causes supported by evidence.
- What was affected? Assess actual and credible potential impact.
- What needs to change? Choose proportionate actions and determine how to evaluate effectiveness.
Fictional example
A blank verification field
During batch record review, Quality finds a required second-person verification field left blank. The operator says they forgot to obtain the signature.
Was the verification performed but not documented, or was it never performed? Those situations may have different product-quality implications. The blank field establishes a documentation gap; it does not, by itself, answer what occurred.
One hypothesis is that an interruption exposed a workflow that relied on memory. That is a possibility to test—not an established root cause.
A practical investigation approach
The following approach summarizes the episode. These are practice suggestions, not a prescribed regulatory sequence. Follow the procedures and requirements applicable to your product and quality system.
1. Contain and define
Protect the product and process as appropriate to the event. This may involve a hold, pausing an activity, notifying the right people, or preserving evidence. Define the unmet requirement, time, location, affected records or batches, known facts, and open questions.
2. Build the evidence and timeline
Review relevant batch records, procedure revisions, training history, equipment information, and electronic records. Check for similar events. Separate observations from hypotheses and look for evidence that could contradict the explanation you favor.
3. Interview to understand
Ask personnel to walk through the work. Explore interruptions, clarity of instructions, equipment behavior, and availability of necessary people. Corroborate important recollections with objective records where possible.
4. Look beyond the individual
An individual lapse can be real without being the complete explanation. Examine process design, procedures, workload, tools, equipment, and supervision. Treat each potential contributor as a hypothesis until evidence supports it.
5. Test the causal explanation
Five Whys and fishbone diagrams can organize possibilities. A completed worksheet is not proof. Ask what supports or contradicts each explanation, and avoid forcing a complex event into a single predetermined cause.
6. Assess impact and scope
Consider product quality, patient safety, data reliability, and compliance. Explain the scope selected, including related batches or operations where relevant. Acknowledge uncertainty and describe how it is managed; an unsupported statement of no impact is not a reasoned assessment.
7. Match action to cause—and check the result
A correction addresses an immediate condition; corrective action addresses the cause of an identified problem; preventive action addresses a potential problem. Whether a formal CAPA is appropriate should follow the applicable quality system, risk, and findings. Retraining alone may leave a verified workflow weakness unresolved.
Define how effectiveness will be assessed. Finishing the action and demonstrating that it worked are different things.
Before you close the investigation
Use these prompts to challenge the reasoning in a draft record:
- Can a reader distinguish established facts, hypotheses, and unresolved questions?
- Does the evidence support the causal explanation, including any system contributors?
- Does the impact assessment explain the scope and remaining uncertainty?
- Do the actions address the findings, with a meaningful way to assess effectiveness?
The aim is an accurate explanation and a defensible decision—not simply a completed form.
Primary source · United States
Where the regulation fits
For drug products within its scope, 21 CFR 211.192, Production record review, requires thorough investigation of unexplained discrepancies and specification failures, including potentially associated batches and products. It also requires a written investigation record with conclusions and follow-up.
The regulation provides the requirement; the practical approach above illustrates ways to support a sound investigation. It does not prescribe Five Whys, a fishbone diagram, or one universal CAPA response. Applicability depends on the product, jurisdiction, and quality system.
Keep asking why.
Continue the conversation with Quality Unpacked. Keep any public examples general and leave out confidential details.
Watch the full episode ↗