Identifying Root Causes in Food Safety Incidents
A customer reports a piece of hard plastic in a salad pot. By the time the complaint reaches quality, the line has been cleaned twice, a shift has changed, and the packaging from that day has gone. Whatever caused the fragment is now much harder to find — not because your team lacks skill, but because evidence disappears fast during an incident. Identifying root causes well is mostly about working an investigation in the right order while the trail is still warm.
Focus here is the investigation workflow: how you move from an incident report to a cause you can defend with evidence. Methods such as 5 Whys and fishbone diagrams support that workflow, but sequence and discipline matter more than any single tool.
Secure the scene and the facts first
Your first job is containment and preservation, in parallel. Hold product that may be affected, using traceability to set the window from the last known good state. Keep the reported item, its packaging, and any fragments exactly as received; photograph them before handling changes them. Record who reported the issue, when, and in what words, before retelling smooths away details.
Preserve process evidence before routine activity destroys it. Pull monitoring records, maintenance logs, and changeover sheets for the window in question. Ask for damaged tools, rejected packaging, and removed parts to be kept rather than binned. In the salad pot case, a cracked ingredient crate or a damaged depositor guard found during the next clean may be the single most important clue — if nobody threw it away.
Build a timeline of the incident
Assemble events in time order: production start, ingredient batches used, changeovers, breaks, maintenance interventions, checks and their results, and the point where the problem was or could have been created and detected. Gaps in a timeline are findings in themselves. A three-hour stretch with no foreign body check, or a maintenance visit with no record, tells you where to dig.
Interview people as close to the event as you can, separately where accounts may differ. Ask what they saw, did, and used — not what they think went wrong. Operators often hold the decisive detail: a noise, a workaround, a part that "always sticks." Treat early accounts as data, and protect the people giving them; an investigation that hunts for someone to blame will receive silence next time.
Evidence to gather as standard
- The complaint item, retained samples, and packaging from the affected window.
- Monitoring, cleaning, maintenance, and training records for the shifts involved.
- Equipment condition reports and photographs taken before repair or deep cleaning.
- Supplier and raw material information for batches in use.
- Accounts from operators, supervisors, and engineers, recorded promptly.
Separate symptoms, direct causes, and root causes
Work in layers. A symptom is what was observed: plastic in a finished pot. A direct cause is the immediate mechanism: a fragment detached from a specific crate during decanting. A root cause is the system condition that allowed it: crates are not inspected for damage at goods-in, damaged crates have no quarantine route, and decanting staff have no instruction to check. Actions aimed only at the direct cause — bin that crate — leave the system ready to produce the next fragment from the next damaged crate.
Test each candidate cause against the evidence. Does it explain all the facts, including timing and why detection failed? Can you reproduce or physically verify it — matching fragment material to the crate, for example? Did similar incidents share it? A cause that explains only part of the timeline is probably a contributing factor; keep it, label it honestly, and keep digging for what sits beneath it.
Know where investigations usually go wrong
Rushed investigations settle on the first plausible story, especially when that story requires no spending. Watch for causes chosen because they are comfortable: individual carelessness, a one-off supplier fault with no evidence sent to the supplier, or a training gap nobody tested. Watch also for investigation by committee opinion, where the loudest voice wins and dissenting evidence is never recorded.
Detection deserves its own line of enquiry. Ask not only how the hazard entered, but why your checks did not catch it. Sieves, magnets, metal detectors, visual checks, and label verifications exist to provide a second chance; if both creation and detection failed, your CAPA may need actions on both sides. Many repeat incidents are detection failures wearing the disguise of new problems.
Turn the confirmed cause into action
Hand over a cause statement an action planner can use: specific, evidenced, and layered. "Damaged crates entered production because goods-in inspection has no damage criterion or quarantine process for returnable packaging (root cause); a fragment detached during decanting on the date in question (direct cause); visual checks at packing did not detect it because fragments match product colour and no foreign body check covers decanted ingredients (detection gap)."
Record also what you ruled out and why. Ruled-out causes — with the test or record that excluded them — stop the case being reopened on rumour later, and show an auditor the investigation had depth. Close the investigation phase by agreeing the cause with the people who own the process, not just within quality; an owner who does not accept the cause will not deliver the fix.
Key points
- Contain product and preserve physical and documentary evidence before routine work destroys it.
- Build a timeline; gaps in records or checks are findings in themselves.
- Interview promptly, ask about observations rather than opinions, and protect people who speak up.
- Separate symptom, direct cause, and root cause — act on the system condition, not only the mechanism.
- Test candidate causes against all the facts, and investigate why detection failed as well as why the hazard occurred.
- Write a specific, evidenced cause statement and record what you ruled out.
