Creatinine QC trending after a new reagent lot
On Monday morning a chemistry analyser is running two levels of quality control material for creatinine. Level 1 (mean 1.00 mg/dL, SD 0.05) gives 1.04 and 1.02 mg/dL. Level 2 (mean 4.00 mg/dL, SD 0.10) gives 4.31 and 4.33 mg/dL today; the level 2 results on the preceding three days were 4.05, 4.12, and 4.18 mg/dL. A new reagent lot was loaded and calibrated on Friday afternoon, and the calibration was accepted. Twenty-two patient creatinine results have already been reported this morning; their mean is roughly 15% higher than last week for the same wards. Every other analyte on the instrument is within its control limits.
1. Apply the quality control rules. Can the patient results from this run be reported?
Read the worked answer
No - the run is out of control.
Level 2 today: 4.31 is (+3.1 SD) and 4.33 is (+3.3 SD) from a mean of 4.00 with SD 0.10. Two consecutive control results exceed 2 SD on the same side of the mean, which is a 2-2s violation and a rejection rule. It is not the 1-2s warning rule.
There is a second, independent signal: 4.05, 4.12, 4.18, 4.31, 4.33 over five days is a monotonic drift, breaching the trend/4-1s logic. Level 1 passing is what makes this insidious - a bias that scales with concentration does not usually show at the low level, and staff who check only the normal level will miss it.
Reporting rules follow: results generated while QC is out of control are not reportable. Stop, investigate, and only release once the fault is corrected and both levels are acceptable.
2. What about the twenty-two results already sent to the wards? Set out your reasoning, not just an action.
Read the worked answer
They are suspect and must be handled as a possible systematic bias, not left alone.
Evidence: the shift is concentration-dependent, it began around the time of the new reagent lot, and the patient population mean is 15% higher for the same wards. Together these point to a reagent- or calibration-related positive bias rather than a genuine change in twenty-two patients.
Steps:
- Tell the laboratory director or pathologist on duty immediately; a look-back is their call but you must put the information in front of them.
- Identify the affected window - everything since the Friday calibration, not just this morning.
- Re-calibrate with a fresh vial and re-assay the stored patient samples from the affected period.
- Compare old and new values; if the difference is clinically material (a creatinine of 1.8 becoming 2.1 changes a dose or a stage), issue corrected reports and telephone the wards.
- Do not quietly overwrite results. Corrected reports are dated, versioned, and explained.
3. Where would you look for the root cause, and what would you check first?
Read the worked answer
Work backwards from the most likely to the least.
First: compare the lots. Assay the same patient samples and a set of calibrators with the previous (known-good) reagent and the new lot on the same instrument. A split-sample comparison that shows a consistent positive offset on the new lot is near-conclusive. Verify the new lot's calibrator values were entered correctly and that the calibrator lot matches the reagent lot.
Second: reagent handling. Was the new lot shipped or stored outside 2–8 °C? Is the on-board pack beyond its stability after opening? Evaporation on the analyser and a partially used pack concentrate the reagent.
Third: the instrument. Probe, syringe, and cuvette condition; a failing dispense would normally affect other analytes too, and they are fine, which argues for the reagent.
Whatever the finding, file a nonconformance, quarantine the suspect lot, and record the investigation. Do not simply repeat the control until it passes - that converts a real bias into a permanent, invisible one.