When the QC Chart Says Stop: A Shift in Sodium
A clinical chemistry laboratory runs sodium, potassium, and chloride on a single analyser. Over the past 7 days the daily two-level quality control (QC) results for sodium at the normal level have been: 139, 140, 139, 141, 140, 139, 140 mmol/L, with a mean of 139.7 and a standard deviation (SD) of 0.76. The manufacturer's target is 140 mmol/L with an allowable SD of 1.0 mmol/L. The Levey-Jennings chart shows all points within ±2 SD, but the last 7 points are all on one side of the mean. The analyser passed calibration 8 days ago. The laboratory is accredited and must report patient results. The supervisor is on leave.
1. Apply the appropriate Westgard rules to these data. Is the run acceptable? Justify your answer.
Read the worked answer
Westgard evaluation:
- 1-2s: no point exceeds ±2 SD → not violated.
- 1-3s: no point exceeds ±3 SD → not violated.
- 2-2s: no two consecutive points exceed 2 SD on the same side → not violated.
- R-4s: no range violation → not violated.
- 7T (seven-trend): seven consecutive points on the same side of the mean → violated.
Conclusion: the run is not acceptable. Although individual points are within limits, the sustained one-sided trend indicates a systematic error (bias), not random error. The most likely causes are a calibration drift, a reagent lot change, a deteriorating electrode, or a change in the control material. Reporting patient results without investigation risks a systematic bias across all sodium results.
2. What is the difference between random and systematic error, and what actions would you take before releasing patient results?
Read the worked answer
Random vs systematic error:
- Random error scatters points around the mean; detected by 1-2s/1-3s and increased imprecision. Causes: pipetting variability, bubbles, electrical noise.
- Systematic error shifts the mean; detected by 2-2s, 4-1s, 7T, 10x. Causes: calibration bias, reagent lot change, electrode drift, control lot change, temperature shift.
Actions before release:
- Hold patient results and place the analyser in a 'results held' status.
- Repeat the QC on a fresh aliquot to exclude a control-handling error.
- Check the QC chart and maintenance logs for a recent event (calibration 8 days ago, reagent lot change, electrode replacement).
- Recalibrate and re-run QC; if the trend persists, replace the sodium electrode and check reagent integrity.
- Review patient results from the affected period for implausible values and consider re-testing.
- Document the non-conformance, root cause, and corrective action in the QMS.
- Notify the supervisor/clinical lead and, if results were already released, issue amended reports.
3. How would you calculate the coefficient of variation (CV) for this QC data, and is the method performing within acceptable limits?
Read the worked answer
CV calculation: CV = (SD / mean) × 100 = (0.76 / 139.7) × 100 ≈ 0.54%.
Interpretation:
- The imprecision is excellent (CV 0.54%), well within typical analytical goals for sodium (often <1–2%).
- However, the mean (139.7) is 0.3 mmol/L below the target (140), and the 7-point trend indicates a small but persistent bias.
- The manufacturer's allowable SD (1.0) is not exceeded, so the method is precise; the problem is accuracy/bias, which the CV does not capture.
Teaching point: a good CV does not guarantee a good result. Always assess bias (mean vs target) as well as imprecision, and use trend rules to detect drift before it breaches limits. If the bias exceeds the total allowable error budget, recalibrate even if the CV is acceptable.