What this case supports, what depends on local policy, and where continuous improvement picks up

This case supports a clear method: state the event specifically, correct the immediate problem, trace contributing factors with a proportionate tool, name the system cause rather than stopping at 'staff error,' choose the strongest action the risk supports, and define a measurable check before you close the record. It does not support skipping proportionality. CAP and CLSI QMS11 both expect the laboratory to scale investigation depth to possible patient impact, and not every nonconforming event, a single short-filled tube caught and redrawn before testing, for example, needs a full Five Whys chain, a fishbone, and a barrier analysis. Save that depth for events with real or possible patient impact, a recurring pattern, or a regulatory or accreditation trigger.

Several pieces of this case are explicitly local policy and will differ by laboratory: the exact reference interval and H-index action limit for potassium on your analyzer, the laboratory's own severity, occurrence, and detectability scoring scale for FMEA, the threshold that decides which events get a full RCA versus a brief correction-only record, which role has authority to close a corrective action, the laboratory's specific quality-indicator set and alert thresholds, and the scope of the change-control procedure itself. State your own laboratory's values for these before applying this method to a real event.

One nonconforming event and its corrective action is a single loop. The Institute for Healthcare Improvement's Model for Improvement frames sustained gains through repeated Plan-Do-Study-Act (PDSA) cycles: state an aim, define a measure, propose a change, test it at small scale, then decide to adopt, adapt, or abandon before expanding, a 'PDSA ramp' rather than a single big-bang rollout. CLSI QMS11 pushes this further by recommending that a laboratory trend nonconforming events by phase of testing, department, error type, severity, cause, instrument, and shift, and feed significant trends into management review. A single LIS hold fixes one gap; watching the trend across all flagged-result categories over the next year tells you whether the underlying change-control weakness has actually closed or only moved to a different analyte.

When the next flagged result gets released and someone reaches for 'retrain the tech' as the fix, ask what engineered control would have made that release harder to do by accident, and ask what number you will check in 30, 60, and 90 days to know whether the answer actually worked.