Quality Control and Quality Assurance
Quality control and quality assurance covers the statistical and regulatory machinery that tells you whether a reported result is trustworthy. The exam tests Westgard rule application, control statistics, the difference between calibration and calibration verification, proficiency testing conduct, and documentation of corrective action.
Control statistics
Establish the mean and SD from at least 20 data points collected over 20 days. CV = (SD/mean) x 100. The Levey-Jennings chart plots control values against ±1, ±2, and ±3 SD limits. Controls must bracket medical decision points; a single normal-level control cannot detect error at a high clinical cutoff.
Westgard multirule interpretation
- 1-2s: warning only - inspect the other levels, do not reject on this rule alone.
- 1-3s: reject; random error, or a very large systematic error.
- 2-2s: two consecutive controls exceed the same ±2SD limit; reject; systematic error.
- R-4s: within-run range between two controls exceeds 4 SD; reject; random error.
- 4-1s: four consecutive values exceed the same ±1SD limit; reject; systematic error (shift).
- 10x: ten consecutive values fall on one side of the mean; reject; systematic error. A shift is abrupt and persistent; a trend drifts in one direction and suggests reagent aging, lamp deterioration, or electrode drift.
Calibration, verification, and PT
Calibration defines the signal-to-concentration relationship using calibrators. Calibration verification confirms the reportable range using materials independent of the calibrators, and is required after calibration, after major maintenance, and on schedule. Proficiency testing is external and must be handled exactly like a patient sample; results are never shared between laboratories, and a failure triggers documented root-cause analysis.
Worked example
Level 1 returns at +2.6 SD and level 2 at +0.4 SD. The single excursion is a 1-2s warning, so patient results may be released. Two hours later both levels exceed +2SD: this is 2-2s, a systematic error, so results are held, the reagent lot and calibration are inspected, and the run is repeated after correction.
Traps that cost marks
- Treating 1-2s as a rejection rule.
- Releasing patient results before resolving a 1-3s.
- Using an expired control or the manufacturer's assayed mean instead of your own established mean.
- Accepting a slow trend as normal drift rather than investigating it.
- Failing to document corrective action for a rejected run.