Egyptian Board Clinical Pathology: EQA Failure and a QC Shift After a Reagent Lot Change

EQA Failure and a QC Shift After a Reagent Lot Change

The biochemistry laboratory of a 400-bed hospital holds ISO 15189 accreditation and measures HbA1c by cation-exchange HPLC on two identical instruments, running about 60 patient samples per day. In the latest national EQA cycle the result for one of the two survey samples was 7.8% against an assigned value of 6.9%, giving a z-score of 3.4; the second sample was acceptable (z = 0.6). The internal QC file shows that since a new reagent lot was introduced 12 days ago, the low-level control has run at a mean of 5.9% against a manufacturer target of 5.6% (previously 5.5%), with the last six points all above +2SD but below +3SD; the high-level control is unchanged. Both analysers share the same reagent lot but have separate columns, one of which was replaced six weeks ago. The laboratory manager suggests simply re-running the EQA sample, noting that 'it is within 0.9% and clinicians won't care', and that patient reporting should continue because the QC is still within 3SD.

1. Is the manager's proposal acceptable? Justify your answer with reference to quality-control rules and to the clinical significance of the bias.

Read the worked answer

No - the proposal breaches both QC rules and accreditation requirements.

QC rule analysis

Six consecutive low-level points above +2SD is a shift; the standard Westgard multirule set rejects on 2 of 3 consecutive points beyond 2SD (2₂ₛ across runs), and certainly on 4₁ₛ or 10ₓ trends. By any of these rules the run should have been rejected and patient results withheld from the point at which the shift was established, not released because no single point crossed 3SD.
A reagent lot change preceding the shift makes the reagent the prime suspect and means the manufacturer's assigned values for the new lot must be verified by the laboratory.
A single acceptable EQA sample does not cancel a failing one; EQA performance is judged per sample and per analyte across the cycle.

Clinical significance of the bias

A positive bias of about +0.3% HbA1c is small for an individual but changes classification near decision limits: a patient at 6.4% could be reported as 6.7% (diagnostic threshold 6.5%) and a patient at 7.9% as 8.2% (target thresholds, insulin initiation, driving/occupational criteria).
The bias is not random; it is systematic and will apply to every result until corrected - which is exactly the pattern that harms patients at scale.
The 0.9% EQA discrepancy (z = 3.4) exceeds the acceptable total error for HbA1c and, at minimum, requires a formal nonconformity.

Immediate actions

Stop reporting HbA1c (or report with an explicit caveat and clinician notification) until the bias is explained.
Quarantine the reagent lot and switch back to the previous, verified lot if available.
Review every patient result reported since day 1 of the new lot; identify patients with results near decision thresholds for repeat testing.
Notify the EQA provider and the accreditation body according to their timelines; open a nonconformity.

2. Design the root-cause investigation. What experiments and records would distinguish reagent lot, calibration, instrument/column and sample-handling causes?

Read the worked answer

Step 1 - Verify the facts

Retrieve the EQA sample handling record: receipt, storage, reconstitution, time on the instrument, which analyser, whether it was run in singleton, and whether the result was transcribed correctly. A transcription or reconstitution error is a common and easily excluded cause.
Retrieve lot-to-lot verification records for the new reagent lot, the calibrator lot and its expiry, column installation date and cumulative injections, buffer lot, temperature logs and maintenance records for both analysers.

Step 2 - Localise the bias

Compare the two analysers on the same fresh patient samples (n ≥ 20, spanning 5–12%): if both are biased identically, the shared component (reagent/calibrator) is implicated; if only one, suspect its column or hydraulics.
Re-run the low-level control on the old and new reagent lots on both instruments.
Test the EQA material with an independent method (boronate affinity HPLC, enzymatic or IFCC reference method) to establish whether the survey sample itself is commutable or has a matrix effect.
Run a lot-to-lot comparison per CLSI EP26-type protocol (≥ 20 patient samples across the range, duplicate measurements, bias and 95% CI against the acceptance limit).
Check calibrator recovery: a new reagent lot may require recalibration; verify with fresh calibrator and check the calibration curve parameters against previous.

Step 3 - Rank the causes

Shared reagent lot across both analysers, with one control level affected and the other not, points to a lot-specific matrix/calibration issue at low concentration - a classic pattern for ion-exchange HbA1c reagents.
Column age on one analyser is a secondary hypothesis, testable by swapping columns or comparing injection counts.
Sample-related causes (storage, temperature, EDTA tube type, sample age) are excluded by running fresh samples under controlled conditions.

Step 4 - Document

Record the experimental design, data, statistical analysis and conclusion in the CAPA file; the investigation itself is auditable evidence.

3. State the corrective and preventive actions, how effectiveness would be verified, and what must be communicated and to whom.

Read the worked answer

Corrective actions

Revert to the previous verified reagent lot or obtain a replacement lot from the manufacturer with written confirmation of performance.
Recalibrate and re-establish control means and SDs after the change; update the Levey-Jennings charts and the QC target sheet.
Repeat or recall patient samples reported during the affected period, prioritising those with results within 0.3% of a clinical decision limit; issue amended reports where indicated.
Raise a complaint/nonconformity with the reagent manufacturer and return retained vials for investigation.
Repeat the failed EQA sample only as part of the investigation, not as a substitute for it, and record both results.

Preventive actions

Mandatory user verification of every new reagent and calibrator lot before routine use, with defined acceptance limits and documented sign-off.
Strengthen review of QC rules at the bench: explicit rejection rules (2 of 3 beyond 2SD, 4₁ₛ, 10ₓ) and a documented rule that a shift triggers a hold on reporting, not a wait for a 3SD breach.
Staff training and competency assessment on multirule interpretation and on the change-control procedure.
Review of the EQA failure investigation with the quality manager and inclusion in the management review and annual quality report.

Effectiveness verification

Acceptable performance on the next two EQA cycles for HbA1c.
QC within established limits for at least 3 months and on the next lot change.
Concordance between the two analysers within a predefined bias limit, audited monthly for 6 months.

Communication

The treating clinicians (via a laboratory bulletin describing the bias and which patients need retesting), the quality manager and laboratory director, the EQA provider, the accreditation body (nonconformity and corrective action within the required timeframe), the manufacturer, and - through the governance route - the clinical risk/patient safety committee. The whole pathway should be handled as a nonconformity and corrective action under ISO 15189:2022 requirements for ensuring validity of examination results (clause 7.7) and nonconformities/corrective actions (clause 8.7).