Multirules look for different patterns
Rules may flag a single extreme point, repeated results on one side of the mean, ranges across controls, or sequential values beyond selected standard-deviation bands. Combining rules aims to improve error detection while managing false rejection.
A rule label such as 1-3s is incomplete without the charted quantity, baseline calculation, distribution, sequence definition, treatment of multiple controls, and procedural response.
Read the rule name as a pattern, not a verdict
The first number commonly describes how many control observations participate, and the subscript describes the standard-deviation boundary or pattern. Exact implementations differ, so the software and procedure must publish the sequence logic and whether a rule is a warning or rejection signal.
| Common notation | Pattern commonly represented | Typical role to evaluate |
|---|---|---|
| 1_2s | One observation beyond ±2 SD | Warning or trigger for another rule |
| 1_3s | One observation beyond ±3 SD | Large isolated departure |
| 2_2s | Two consecutive observations beyond the same ±2 SD side | Possible systematic shift |
| R_4s | Two control results in the same run differ by more than 4 SD, often one above +2 SD and one below −2 SD | Possible increased random error |
| 4_1s | Four consecutive observations beyond the same ±1 SD side | Persistent shift |
| 10_x | Ten consecutive observations on one side of the mean | Sustained bias or shift |
Bioassay results are not automatically clinical QC measurements
Potency methods may have nonlinear fitted results, few runs, changing control or reference lots, heterogeneous products, unequal uncertainty, censored estimates, and method-version changes. Those conditions can violate a simple fixed-mean, fixed-SD interpretation.
Decide whether to chart raw control response, log potency, reference midpoint, normalized recovery, or another stable quantity. Do not apply rules to product samples that intentionally vary and call the variation process drift.
Simulate and back-test the proposed rules
Using representative historical or designed data, estimate how often each rule signals under stable behavior and under plausible shifts or variance increases. Review whether the response burden is operationally tolerable and scientifically useful.
Avoid tuning rules on the same incidents used to claim performance without independent evaluation. Version rule definitions and baseline populations.
Define the response before relying on a flag
A rule signal should show the rule, points involved, observed values, baseline and limits, run context, and date. It should not silently change run disposition.
Keep rules configurable and versioned, and preserve links to the source runs, annotations, investigation, and review outcome. The approved procedure—not the chart color—determines the operational response.
Frequently asked questions
Which Westgard rules should a bioassay use?
There is no universal set. Select and justify rules based on the charted quantity, baseline, error modes, false-signal rate, and procedural response.
Does a Westgard signal invalidate the current potency result?
Not by itself. The method and quality procedure define investigation and disposition; the signal supplies monitoring evidence.