Use of robust ZB and ZW to evaluate proficiency testing data

Area of Science:

• Clinical Chemistry
• Laboratory Medicine
• Biostatistics

Background:

• Traditional PT programs often use mean-based targets, which are sensitive to outliers.
• This method can lead to inaccurate performance assessments when data includes extreme values.

Purpose of the Study:

• To evaluate the performance of Jaffe and enzymatic creatinine assays.
• To compare the effectiveness of robust z-scores versus traditional methods for outlier detection in PT programs.

Main Methods:

• Creatinine data were grouped by Jaffe and enzymatic methods.
• Normality testing was performed on standardized sum and difference data before and after outlier removal.
• Laboratory performance was assessed using ZB and ZW scores, with defined acceptability criteria.

Main Results:

• Jaffe and enzymatic methods yielded non-comparable results; initial data were not normally distributed.
• Outlier removal resulted in normally distributed data, improving assessment reliability.
• Acceptable performance rates were higher with the Jaffe method (90.2% between, 86.0% within labs) compared to the enzymatic method (82.6% between, 82.4% within labs).

Conclusions:

• Robust z-scores (ZB and ZW) are more reliable for assessing laboratory performance in PT programs due to their reduced sensitivity to outliers.
• The findings support the use of robust z-scores for more accurate and dependable evaluations in clinical chemistry.