📱 New: Reports sent automatically on WhatsApp.
Back to Blog
Lab Management

Quality Control Best Practices: Westgard Rules Made Simple

ML
Mithla Lab Team
Lab Operations
April 10, 202510 min read

Westgard rules are the most widely used statistical framework for deciding whether a run of quality control results means your analyzer is working correctly — or whether you need to stop and investigate before releasing patient results. The rules themselves are simple once you see them applied; the hard part is applying them consistently, every run, without relying on someone remembering to check.

The building blocks: mean and standard deviation

Every Westgard rule is built around your QC material's established mean and standard deviation (SD), typically plotted on a Levey-Jennings chart. Each new QC result is compared against how many SDs it falls from the mean. The rules just define which patterns of deviation are acceptable and which mean "stop and investigate."

The core rules

  • 1_2s — a warning rule, not a rejection rule: one QC result outside ±2SD. This flags a run for closer attention but doesn't automatically reject it on its own.
  • 1_3s — one QC result outside ±3SD. This is a rejection rule — the run is out of control and should not be released.
  • 2_2s — two consecutive QC results (same level, or two levels in the same run) both outside ±2SD on the same side of the mean. Rejection rule; usually points to a systematic shift.
  • R_4s — the range between two QC results within the same run exceeds 4SD (e.g., one is +2SD and the other is −2SD). Flags random error.
  • 4_1s — four consecutive QC results (across runs) outside ±1SD on the same side. Flags a slow, systematic drift.
  • 10x — ten consecutive QC results on the same side of the mean, regardless of distance. Another systematic-error signal, often the first one to catch a subtle calibration drift.

Which rules should actually trigger rejection?

Most labs don't run all of these as hard rejection rules — that produces too many false rejections and QC fatigue. A common, well-tested combination (the "Westgard multirule" approach) uses 1_3s, 2_2s, R_4s, and 4_1s as rejection rules, with 1_2s as a warning-only trigger that prompts a look at the other rules rather than an automatic stop. The right combination for your lab depends on your analyte's clinical risk tolerance and how tightly controlled your process already is — this is a judgment call your lab director should own, not a one-size-fits-all default.

Where automation actually helps

Applying these rules by eye, across every analyte and every shift, doesn't scale — and it's exactly the kind of repetitive pattern-matching that's easy to get wrong when you're busy. What LIMS-level QC automation should do:

  • Flag rule violations automatically the moment a QC result is entered, before it's possible to proceed to reporting patient results on that run.
  • Keep a running Levey-Jennings view per analyte per QC level, without someone manually charting it.
  • Require a documented corrective action before a flagged run can be cleared — not just a silent override.
  • Preserve the full QC history and any corrective-action notes for audit and accreditation review.

The part that's easy to skip: documentation

A rule violation without a documented corrective action is a compliance gap waiting to be found during an audit. Whatever system you use, make sure every flagged run has a recorded reason and resolution — recalibrated, reagent replaced, repeat confirmed within range — attached to it permanently, not just a note in someone's notebook.

Westgard rules aren't complicated math — they're a discipline. The value comes from applying them the same way, every single run, which is exactly the kind of consistency that's worth building into your system rather than your staff's memory.