It is 6 a.m., the chemistry analyzer has just run its morning controls, and the glucose Level 2 is sitting just outside 2 SD. Do you release patient results? Rerun the control? Call the supervisor? The Westgard rules exist precisely for this moment. They turn “this looks off” into a consistent, defensible decision.
The basics: the Levey-Jennings chart
Each control level has an established mean and standard deviation (SD), ideally calculated from your own laboratory’s data (at least 20 runs) rather than copied from the package insert. Plotted on a Levey-Jennings chart, about 95% of results from a stable method fall within ±2 SD and about 99.7% within ±3 SD. A result outside those limits is not automatically an error. It is a signal to apply the rules.
The rules, one by one
- 1-2s (warning). One control beyond ±2 SD. On its own this is a warning, not a rejection. It tells you to check the other rules before accepting the run. Used as a rejection rule, it would wrongly reject a large share of good runs.
- 1-3s (reject). One control beyond ±3 SD. This mainly catches random error: a bubble in the sample probe, a clot, a mislabeled control cup, an electrical spike.
- 2-2s (reject). Two consecutive controls beyond the same 2 SD limit, either both levels in one run or the same level in two consecutive runs. This points to systematic error: a calibration shift, a new reagent lot, a deteriorating lamp.
- R-4s (reject). In one run, one control is above +2 SD and another is below −2 SD, a range of more than 4 SD. This signals random error, because the two controls moved in opposite directions.
- 4-1s (reject). Four consecutive results beyond the same 1 SD limit. A small but persistent systematic shift.
- 10x (reject). Ten consecutive results on the same side of the mean. Another systematic pattern, often a slow drift. Some laboratories use 8x or 12x versions depending on how many controls they run.
Why the type of error matters
Knowing which rule failed tells you where to look.
| Error type | Rules that catch it | Usual suspects |
|---|---|---|
| Random | 1-3s, R-4s | Pipetting, bubbles, clots, sample mix-ups, power fluctuations |
| Systematic | 2-2s, 4-1s, 10x | Calibration, reagent lot or storage, instrument temperature, lamp or electrode wear |
What to do when a rule fails
- Stop. Do not release patient results from the affected run.
- Identify the rule and the error type.
- Investigate the likely cause. Check reagents, calibration date, maintenance logs, control handling.
- Fix the cause, then rerun the controls.
- Assess the impact. If the problem may have affected results already released, re-test patients since the last acceptable run, as your SOP requires.
- Document everything, including the corrective action.
Rerunning a control until it passes is not quality control. It is hoping.
The most common bad habit in busy labs is simply repeating a failed control without investigating. If the second run passes, the original problem may still be there, just hidden by chance.
Making the rules work for your method
The classic multirule (1-3s/2-2s/R-4s/4-1s/10x with 1-2s as a warning) is a good default. Westgard’s Sigma rules go a step further: methods with excellent performance (Sigma 6) may need only a 1-3s rule with two controls per run, while weaker methods need the full multirule and more control measurements. Choosing rules based on how well each method actually performs avoids both excessive false rejections and missed errors. CLSI C24 is the consensus guideline for designing a QC strategy.
References
- Westgard QC, “Westgard Rules” and Westgard Sigma Rules: westgard.com
- Westgard QC, “50 ways to misuse Westgard Rules”: westgard.com
- Bio-Rad Unity Real Time documentation, Westgard rules: qcnet.com
- Practical application of Westgard Sigma rules with run size, PMC: ncbi.nlm.nih.gov
