Working the reagent-lot event through the plan

Return to the 10:25 result. Level 1 read 2.04 units, inside its 1.90 to 2.10 range. Level 2 read 8.63 units, above its 8.40 unit upper limit, flagged high. The failure mode entry for this test system already named reagent degradation or lot-related bias, with transport temperature excursion or an unverified new lot as the likely cause, and positive bias in released patient results as the effect. Existing controls listed for that entry were receipt-temperature review, the internal analyzer check, daily external QC, lot verification, and maintenance records.

Detectability is where this event is instructive. The internal electronic check passed at both 07:00 and 10:25; it checks instrument function, not reagent chemistry, so it was never going to catch a lot-related bias. Level 1, tested at a lower concentration, stayed in range. Level 2, tested at a higher concentration, is the layer that actually caught the shift. That is the practical meaning of detectability in a risk entry: not whether a control exists, but whether the specific control in front of you had a real chance to catch this specific failure.

The disposition follows directly from 42 CFR 493.1256 and 493.1289. Results from the affected run are not released. Testing is held. The laboratory investigates reagent storage and lot verification using its corrective-action procedure. It evaluates patient results from the unacceptable run and from every run back to the last acceptable control run for possible impact, because an unacceptable control does not tell you exactly when the process started drifting, only that it had drifted by 10:25.

This event also feeds back into quality assessment, not just corrective action on this one run. If the investigation shows that the existing control frequency, the shipping-temperature check, or the lot-verification step did not actually control this risk, the risk assessment and the QCP need to be reassessed, not just the one run corrected. A plan that is never revisited after a real failure is not doing ongoing quality assessment; it is only doing incident response. When a control catches something, ask not only what corrective action the failed run needs, but whether the plan that let the failure develop still matches the actual risk.