A-089 Assessment of differential proficiency testing performance in common clinical chemistry analytes

Abstract Background The robust measurement of analytes can be influenced by a large array of consideration including method type, analyte stability, and the complexity of the procedure. It was of interest to assess whether performance on proficiency testing was significantly different across some of the primary clinical chemistry analytes. Methods RCPAQAP survey proficiency testing results were obtained for 18 participating laboratories beginning from the first survey of 2019. The data was filtered to include analytes which were measured by at least 3 laboratories in at least 3 unique survey cycles. Only analytes with greater than or equal to 44 unique proficiency testing results were included. A total of 16 laboratories and 33 analytes were used for further assessment. A chi-square test was used to compare total amounts of pass/fails across each analyte. Standardized residuals were calculated to identify which analytes deviated most from the predicted amount of pass/fails based on overall analyte pass/fail counts. An ANOVA test was used to assess the average pass rates of the analytes. This was done by calculating the overall pass-rate of each laboratory for each analyte and treating these as individual data points. After determining overall performance, analytes with standardized residuals greater than 5 from expected pass/fail amounts were selected for further assessment. The selected analyte’s pass-rates were paired by laboratory and compared to all other analytes, using a one-sided paired Mann-Whitney u test. Additionally, chi-square results were obtained comparing the pass/fail amounts of the selected analyte compared to all other analytes. Results The pass/fail amounts were not equivalent across all targets, based on chi-square (p-value = 3.3x10^-86) and the pass-rates were not equivalent based on the ANOVA (p-value = 1.97x10^-5). The analytes with increased pass amounts, based on standardized residuals, included Triglyceride (7.27), Blood Urea Nitrogen (6.82), and Gamma-Glutamyl Transferase (5.16). The analytes with increased failure amounts were Chloride (8.81), HCG-Quantitative (7.12), Troponin-I (6.88), and Cortisol (5.22). Based upon the paired Mann-Whitney u test, four of the seven markers remained significant (p-values less than .05) when pass-rates were compared to all other analytes. Troponin-I (p-value =.031; n=5 laboratories) and Chloride (p-value=.023; n=13) showed decreased performance whereas BUN (p-value = .012; n=15) and Triglycerides (p-value less than .001; n=16) showed increased performance. Based on chi-square results all seven markers assessed showed significantly different (p-value less than .001) pass/fail amounts when compared to the pass/fail amounts in all other analytes. Conclusion With this study we have presented evidence that the robust measurement of a target can be analyte specific. Two analytes, BUN and Triglycerides, showed significantly greater proficiency testing performance compared to all other analytes, independent of laboratory. Two analytes, Troponin-I and Chloride, showed significantly decreased performance. These findings may suggest that overall improvements could be made to the analyte methodologies for Troponin-I and Chloride, given their critical role in the diagnosis and treatment of patients. Future studies on why BUN and triglycerides can be measured so robustly may elucidate how to enhance analyte measurement methodologies in general.