EQA
When EQA Goes Wrong: A Corrective Action Playbook for PT Failures
A proficiency testing failure is not, by itself, a disaster. Every lab fails a PT challenge eventually — methods have bad days, lots go off, samples get mishandled. What separates a lab that recovers cleanly from one that ends up in regulatory trouble is what happens in the four weeks after the failure.
The failure is a data point. The corrective action is the evidence — and a sloppy corrective action is what turns a single failed analyte into a finding, a repeat failure, or worse. This is a playbook: what to do, in what order, and how to document it so the response is as defensible as the rest of your quality system.
First: don’t panic, and don’t recalibrate yet
The instinct after an unacceptable result is to do something — recalibrate, swap the reagent lot, re-run the sample. Resist it until you understand the failure, because acting before you know the cause does two harmful things: it destroys evidence (you can’t investigate a state you’ve already changed), and it risks “fixing” something that was never broken while the real cause persists.
The one thing that is urgent is the clock. Accreditation bodies expect PT results reviewed and acted on within a defined window — CAP’s is 30 days; ISO 15189 and UKAS schemes run shorter (commonly 14–21 days). The corrective action process needs to start, and be documented as having started, well inside that window. We covered the scheme-by-scheme review windows in CAP vs CLIA proficiency testing.
Step 1: Confirm it’s a real failure
Before investigating the method, rule out the cheap explanations. A surprising share of “failures” are administrative:
- Clerical/transcription error — the right result entered in the wrong field, a units mix-up, a decimal slip.
- Sample handling — the PT sample was reconstituted wrong, stored wrong, or run differently from patient samples.
- Provider/grading issue — a withdrawn challenge, an ungraded analyte, a peer group too small to produce a stable consensus.
If the failure is genuinely clerical or the challenge was withdrawn, document that as the finding with evidence — it’s a legitimate, defensible outcome. But “it was a clerical error” must be proven, not asserted; an auditor who sees that explanation with nothing behind it reads it as a lab dodging a real investigation.
Step 2: Root cause — go past “technologist error”
If the failure is real, the corrective action lives or dies on the root cause. The single most common weakness in PT corrective actions is stopping at “technologist error — retrained”. It’s almost never the real root cause, and it’s the explanation auditors trust least, because it can’t recur into a fix — you can’t verify that a person is now permanently careful.
Push past the first answer. A simple framework:
| Candidate cause | What to check |
|---|---|
| Calibration | Was the instrument in calibration on the test date? Any drift in recent calibration history? |
| Reagent / lot | New lot around the failure date? Lot-to-lot verification done? Storage intact? |
| Maintenance | Was scheduled maintenance current? Any function-check anomalies? |
| Method / setup | Correct method, correct settings, correct sample type? |
| QC at the time | What were the in-house controls doing on and around the test date? |
| Sample-specific | PT-sample matrix effect, reconstitution, stability? |
Your calibration and EQA history are the investigation’s best tools here — a z-score that had been drifting upward for three rounds (see understanding z-scores in EQA) points at gradual calibration or reagent drift, while a sudden jump points at a discrete change like a lot or a recalibration. The trend tells you where to look.
Step 3: Run the corrective action through a real workflow
A corrective action isn’t a paragraph in a logbook — it’s a tracked process with states, owners, and dates. A defensible workflow moves through clear stages:
- Open — the failure is logged, an owner assigned, a target resolution date set.
- In progress — root-cause investigation underway, with findings recorded as they emerge.
- Closed — root cause identified, corrective action taken, preventive action defined to stop recurrence.
- Verified — and this is the stage labs skip — evidence that the corrective action worked.
Each stage is dated and attributable. “Open → verified” with timestamps is the story an auditor wants: a problem found, investigated, fixed, and proven fixed, all inside the review window.
Step 4: Effectiveness verification — the step that’s actually required
Closing a corrective action means you did something. Verifying effectiveness means you proved it worked. These are different, and accreditation standards increasingly expect the second.
Effectiveness verification ties the corrective action to a subsequent acceptable result. You recalibrated and replaced the reagent lot? The next PT round (or a repeat challenge, or sustained in-control QC over a defined period) coming back acceptable on that analyte is the evidence the fix held. Without it, you have a corrective action that might have worked and a method that might fail again next round — which, under CAP and CLIA, is exactly how a single failure becomes a consecutive failure and escalates toward a reportable, test-authority-threatening event.
Step 5: Preventive action and the audit trail
Corrective action fixes this failure. Preventive action stops the class of failure. If the root cause was a missed lot-to-lot verification, the preventive action is a process change — lot verification becomes a gated step, not an optional one. If it was calibration drift caught too late, the preventive action might be a tighter calibration interval or active z-score trend monitoring.
Throughout, attach evidence: the calibration record from the investigation, the new-lot verification, the recalibration certificate, the subsequent acceptable result. File integrity matters here as everywhere — checksummed evidence that demonstrably hasn’t been altered since upload is what makes the CA record stand up two years later.
The independent-reviewer safeguard
One structural control worth building in: under stricter accreditation regimes, the person who entered the failing result shouldn’t be the sole person who closes out the corrective action and signs off the round. Separating “did the work” from “reviewed and closed it” catches the case where the same blind spot that caused the failure also waves through an inadequate fix. It’s the same independent-review principle that underpins good document approval.
The playbook, condensed
When an unacceptable PT result lands:
- Start the clock — log it; you’re inside a 30-day (CAP) or shorter window.
- Confirm it’s real — rule out clerical, sample-handling, and provider/grading explanations, with evidence.
- Find the true root cause — work the checklist; never stop at “technologist error”. Ask what QC was doing.
- Run the CA through its states — open → in progress → closed → verified, dated and owned.
- Verify effectiveness — tie the fix to a subsequent acceptable result, not just to “we recalibrated”.
- Add preventive action — change the process so this class of failure can’t recur.
- Keep the evidence linked and checksummed — the record is the deliverable.
A lab that runs this playbook turns its worst EQA day into its strongest piece of audit evidence: proof that the quality system works — that it catches problems, investigates them honestly, fixes them, and confirms the fix. That’s worth more to an inspector than a wall of passing results, because it’s the thing passing results can never demonstrate.
Related reading:
