Check a control sequence for Westgard flags

Start with an assigned mean and SD

Enter the established mean and a positive SD for one control level, followed by its results in measurement order. The checker converts each result to a standardised distance, z = (result − mean) / SD. Both the mean and SD must use the same measurement units as the control results. Do not combine raw results from different concentration levels in this single sequence, and do not recalculate the mean from a suspect run to make its flags disappear.

The entered sequence is treated as consecutive observations. Missing results, changes in control lot or changes in the measurement procedure can break the clinical meaning of that continuity even though the arithmetic still works. Use decimal points for fractional results and check that the order matches your worksheet. The checker cannot infer run boundaries, elapsed time or which measurements belong to the same analytical batch.

Understand the six checks

A 1-2s warning occurs when one result is beyond two SD but not beyond three SD. A result beyond three SD receives a 1-3s rejection flag instead of an additional 1-2s warning. Two consecutive results beyond two SD on the same side trigger 2-2s. Four consecutive results beyond one SD on the same side trigger 4-1s, while ten consecutive results strictly on the same side of the mean trigger 10x.

For R-4s, this implementation checks adjacent observations for one result above +2 SD and the other below −2 SD. It does not flag every pair whose numerical separation exceeds four SD. More importantly, an adjacent pair across separate runs is not automatically a valid within-run R-4s assessment. Where your procedure applies R-4s to different control levels within a run, review those measurements separately using the required run structure.

All comparisons are strict. A result exactly at +2 SD does not satisfy a condition requiring a value above +2 SD. A result exactly on the mean does not extend a 10x sequence. Overlapping sequences can generate more than one flag, so the count of flags is not the count of independent analytical failures.

Decide what happens next

A warning prompts review, not an automatic conclusion that the analytical system has failed. Rejection flags require action under the laboratory’s authorised procedure. Consider control preparation, reagent integrity, calibration, maintenance, environmental changes and transcription errors. Do not repeatedly measure controls until a passing result appears without understanding and documenting the original event.

Record the investigation, corrective action, repeat testing and decision about affected patient results in the laboratory’s records. A clean checker output only means that these six implemented patterns were not detected in the supplied sequence. It does not prove acceptable analytical performance, adequate control frequency or absence of clinically important error. Select the rule set and review window using the assay’s performance and clinical risk.

Assigned mean = 100; SD = 2. Results: 101, 98, 104.6, 104.8, 99, 100, 107, 101, 97, 95.4.

The z values are 0.5, −1, 2.3, 2.4, −0.5, 0, 3.5, 0.5, −1.5, −2.3. There are 1-2s warnings on results 3, 4 and 10; a 2-2s rejection on results 3–4; and a 1-3s rejection on result 7. There is no R-4s, 4-1s or 10x flag. Result numbers start at 1.