Corrective and Preventive Action

Use the terms deliberately

Containment prevents suspect product, materials or equipment from moving while the event is assessed. Correction restores an immediate condition, such as replacing a failed probe or resetting a controlled process. Product disposition decides what happens to the affected lot. Corrective action addresses the cause of the event. Preventive action applies justified learning before the same mechanism creates another event elsewhere. Combining these into one vague action hides incomplete work.

Control the affected scope

Identify the credible time window, lots, raw materials, process locations and distributed product. Place physical and electronic holds that remain effective across storage and dispatch systems. Reconcile quantities so held, sampled, reworked, destroyed and released product account for the original lot. Expand or narrow the scope only with recorded evidence. An unaffected-looking package is not outside scope merely because the defect is not visible.

Make a supported disposition. Release, rework, diversion, relabelling, extended hold or destruction must follow the actual hazard, defect, legal route and supporting evidence. Rework needs a defined process, identity and verification; it cannot be an improvised attempt to make measurements pass. Consider whether the proposed treatment controls the whole affected volume and whether it creates new quality, safety or labelling consequences. Keep the product decision separate from the repair or procedural change.

Select action from causal evidence

Match the action to the demonstrated mechanism. Repair or redesign equipment where condition or access failed; revise a process where its limits or sequence were inadequate; strengthen supplier controls where incoming variation was causal; and train where knowledge or execution was genuinely deficient. Retraining alone is weak when the system contained unclear instructions, poor interfaces, inaccessible equipment, excessive workload or ineffective verification. Address detection failure as well as occurrence.

Control change and unintended effects

Changes to formulation, curing agents, temperatures, humidity, airflow, time, equipment, software, packaging or sampling can alter the validated process. Define approval, scientific support, trial conditions, version control and rollback before implementation. Do not solve a quality fault by weakening a safety control. Manufacturer instructions, protected specifications and jurisdictional requirements remain controlling where applicable. Temporary repairs require defined limits and replacement dates.

Assign and document the action

Record the event, affected scope, immediate controls, causal conclusion, selected action, owner, due date, approvals, verification method and closure authority. Preserve actual completion evidence rather than a tick-box assertion. Link maintenance orders, revised procedures, training records, product disposition, laboratory results and monitoring data to one event identity. Late actions and interim risk controls should remain visible.

Verify effectiveness

Define in advance what result would show the action worked, over what period and under which comparable conditions. Use measures matched to the cause: fault frequency, positional distribution, actual process values, calibration results, maintenance findings, environmental samples or audit observations. One acceptable batch may confirm restoration but not sustained prevention. An intermittent event requires a long enough observation window to make recurrence detectable.

Extend learning proportionately

Review comparable products, lines, chambers, suppliers and sites for the same mechanism. Preventive action should follow evidence rather than automatically copying every change facility-wide. A control appropriate to one fermented sausage may be unsuitable for a cooked product or protected specification. Trend actions by cause and recurrence, and reopen those whose predicted improvement did not occur. Closure means effectiveness is demonstrated, not simply that the assigned task is complete.

Handle temporary controls

When permanent correction cannot be completed immediately, define an interim control with scope, owner, start date, monitoring, escalation limit and expiry. Increased inspection or testing can reduce risk temporarily but may be weaker than restoring the failed preventive control. Confirm that staffing, instruments and review frequency make the interim measure workable. Reauthorise it explicitly if the repair date changes; an undocumented temporary practice easily becomes the uncontrolled normal process.

Manage overdue and unsuccessful actions

An overdue action is a current risk condition, not only an administrative metric. Review whether containment and interim controls remain adequate, whether production should continue and who accepts the residual risk. If effectiveness criteria fail, reopen the causal analysis rather than extending the closure date automatically. The original cause may be incomplete, the action may not have been implemented as designed or the process may have changed after implementation.

Use recurrence as evidence

A repeated fault provides evidence about the earlier action. Compare mechanism, position, materials, equipment and timing rather than grouping events only by symptom name. A visually similar defect can have a different cause; a different symptom can arise from the same failing control. Link the records and test the earlier prediction. Recurrence after a claimed effective action requires documented review of both the causal conclusion and the verification method.

Close with accountable approval

Closure approval should confirm that product quantities reconcile, required notifications are complete, controlled documents and training are current, permanent actions are installed and effectiveness evidence meets the predefined rule. Separate technical completion from final approval where risk justifies independent review. Preserve the closure basis and any residual limitations. The date a task was marked complete is not the date the system proved effective unless the evidence supports both.

Preserve the baseline after change

Once the revised control is approved, establish its new baseline using actual process and fault data. Record the effective formulation, procedure, software, equipment condition and monitoring arrangement so later drift can be distinguished from normal variation. Compare performance with the pre-action period using consistent definitions. If the action changes how faults are detected, interpret apparent increases or decreases carefully because the measurement system, not only the process, may have changed.

Audit the complete action chain

Periodic review should trace selected events from detection through hold, quantity reconciliation, disposition, causal analysis, implementation and effectiveness. This reveals actions that appear closed in a summary register but lack product evidence or sustained verification. Review recurring causes, overdue actions and repeated temporary controls together. The objective is not paperwork completeness alone; it is to test whether the system reliably protects product and changes the conditions that created the event. Findings should be assigned, dated, reviewed and retained as part of the same accountable improvement system.