Proficiency Testing Z-Score: Laboratory Capability Assessment Online
What Is a Z-Score?
Proficiency testing (PT) assesses a laboratory's testing capability with the Z-score: Z = (x - X)/sigma, where x is the laboratory result, X the assigned value (consensus or reference value) and sigma the standard deviation for proficiency assessment. The Z-score standardizes results of different items, enabling comparison of deviation across programs and rounds - the most widely used capability indicator in laboratory accreditation and surveillance. It lets test items with different units be compared on one scale, so laboratories can track their own performance longitudinally.
Judgement Criteria
Per ISO 13528: |Z| of 2 or less is satisfactory; between 2 and 3 is questionable (a problem signal requiring investigation); 3 or more is unsatisfactory (corrective action needed and accreditation may be affected). The tool applies this automatically and supports summarizing Z-scores across rounds and items to spot systematic-bias trends. Questionable results should be investigated within the current reporting period; unsatisfactory results trigger the corrective-action process.
Where Do the Assigned Value and SD Come From?
The assigned value can come from a reference laboratory (certified reference value), a consensus value (robust mean, median) or a formulation value; the proficiency-assessment SD can use a robust standard deviation (e.g., MAD, Qn) or a prescribed value. The tool accepts direct result-plus-assigned-value-plus-SD input and can also compute the robust mean and robust SD automatically from a set of results. Robust statistics are more stable with small samples and outliers and are the ISO 13528-recommended statistical method.
How to Use It
Enter your result with the organizer's assigned value and SD, compute the Z-score per item, compare against the criteria, and when unsatisfactory start the root-cause analysis (personnel, equipment, method, calibration) and participate in retesting. Note that with few participating laboratories, robust statistics are more reliable because outliers heavily influence the ordinary mean. Handle PT samples within the routine testing workflow so the results represent the laboratory's normal performance.