Internal Audit Findings: A Safety-Critical Out-of-Calibration Tool and a CAPA That Closed Itself

Priority 1 - finding (2), major. It is the only one with a known product risk. The measurement resource for a safety-critical characteristic was outside its calibration system for five months, so there is no valid evidence that 3,100 brackets were torqued correctly.

Containment, in order:

• Quarantine remaining stock in-house and identify the exact affected population by date code, work order, and machine - traceability first, since without it containment is guesswork.
• Notify the customer in writing within 24 hours, per contract. Do this before the investigation is finished; the contract requires notification of the condition, not a completed root cause.
• Re-verify a sample of the affected brackets with a calibrated torque wrench or by destructive/non-destructive torque audit, and disposition the stock (rework, re-torque, scrap, or concession) based on the results. Where re-verification is impossible, treat the product as suspect.
• Check the previous calibration result and the wrench's drift history to bound the error; a torque wrench that failed high gives over-torqued joints and stripped threads, one that failed low gives under-torque. The failure mode, not just the date, determines the risk.

Priority 2 - findings (1) and (3), major when taken together. Missing initials alone would normally be a minor records issue. Here it is a repeat of an unresolved problem and the earlier CAPA closed with no root cause and no effectiveness check, which makes it a systemic breakdown of the corrective action process rather than a clerical lapse.

Priority 3 - management review gap, minor to major depending on your registrar. The input requirements for management review were not met; standing alone it is a minor system weakness, but it explains why findings 1 and 3 persisted.

Guard against the common error of grading by how bad the paperwork looks. The safety-critical calibration escape is major because of the product, not because of the paperwork.

Why the existing CAPA fails: "Operators retrained. Closed." contains no root cause, no action beyond a training event, no product risk assessment, and no verification that anything changed. Training is an action that addresses a presumed cause (knowledge), and here the presumption was never tested. It also repeats, which is itself evidence the cause was not found.

Root cause analysis - investigate before choosing an action. Candidate causes to test:

• Form design: the initials block is on the back of the traveler, or the field is not adjacent to the measurement it certifies.
• Workflow: records are completed at end of shift from memory, and the shift ends at the same time for all inspectors.
• Records are paper and physically move between benches; some sheets are completed retroactively.
• Staffing: one inspector covers two lines during breaks, so sign-off is skipped under time pressure. Use a structured method (5 Whys on a specific record, or a fishbone grouped by method/people/machine/measurement) and verify the cause with data - e.g. are the six missing records all from one shift, one part family, or one week?

Action (example, contingent on cause): move sign-off into the electronic inspection screen with a mandatory field that blocks saving the record until initials and timestamp are captured; if paper must stay, redesign the form so the field sits beside the measurement and add a supervisor check at end of shift.

Effectiveness verification, written into the CAPA before closure:

• Metric: missing initials per 100 inspection records, current 6/20 = 30%.
• Sample: audit 30 records per month for 3 consecutive months.
• Pass: zero missing entries across all 90 records.
• Checked by: quality manager, not the person who performed the action; verified 90 days after implementation.

Also required: an assessment of whether any product was released without valid inspection records (risk of undetected nonconformity), and a note linking back to the earlier ineffective CAPA.

What is missing procedurally: audit results and corrective action status are required inputs to management review, and the review is required to produce decisions and actions on improvement and resource needs. A memo is not a management review. Leaving this unaddressed converts a single audit finding into a system finding that will be raised again at the next surveillance audit, and it removes the only forum where the caliper of the three findings can be allocated budget.

What should happen:

1. An extraordinary management review with the audit report, CAPA status, the calibration escape, and customer feedback on the agenda - minuted, with named owners and dates.
2. A decision on the calibration system as a system: recall list reviewed for other overdue items, interval strategy changed to drift-based, and a resource assigned.
3. A review of all open CAPAs for the same defect - closed with training only, or closed without verification - since this one is unlikely to be unique.
4. A defined follow-up audit date to verify the actions, before the surveillance audit.

The business case, for a plant manager who responds to cost:

• One customer notification on a safety-critical characteristic can trigger a controlled-shipping requirement, a supplier quality audit, and in the worst case a recall of the bracket assembly. The cost of one recall dwarfs the cost of a torque audit on 3,100 parts.
• The repeated finding means the first corrective action's cost produced no benefit; the same problem will be paid for again.
• A major nonconformity at surveillance puts the certificate at risk, and the certificate is a condition of several customer contracts.

Frame the request as a choice between a bounded internal cost now and an unbounded external one later, with the 24-hour notification obligation as the concrete trigger date.