A-101 Is That Unexpected? Education on Handling Real World Complications with Tightly Controlled Analytes During Method Evaluation

Abstract

Abstract Background A common activity in clinical laboratories is performance evaluations. Given the importance for a new analyzer being put into clinical use, this exercise typically requires extensive preparation with significant time and resources invested to bring on new instrumentation. During early adoption, studies examined real-world performance and comparability of the Atellica® CI Analyzer (Atellica CI) to clinically used instruments. Precision, linearity, and method comparison (MC) studies, conducted in accordance with CLSI protocols, were completed at several locations installing new instruments to examine the analytical performance of common clinical laboratory assays. This work examines the impact of using only unmodified specimens to verify performance during method comparison for tightly controlled biological analytes like electrolytes and illustrates the impact of the tested sample range on regression analysis. Methods Testing was completed using unmodified, de-identified, remnant patient serum specimens. The MC studies for these electrolytes were completed over multiple days using single replicates from at least 40 individual specimens, on two instruments, the Atellica CI and the Dimension® EXL in accordance with CLSI EP09-3A. The effect of sample range on regression was demonstrated through data restriction and inclusion of modified sample results within the data analysis. Data analysis was completed using Analyse-it for Microsoft Excel (version 6.15.4). Results Representative method comparison data for sodium, chloride, and potassium found slope differences were more pronounced when sample concentrations were more limited. Individual assay slopes were determined using Deming linear regression model and comparisons between restricted samples show the effects of sample range on regression and correlation parameters. Conclusions Understanding the relationship between sample range and regression analysis is helpful to illustrate an important fundamental concept of method comparison. These examples illustrate the importance in covering as much of the analytical measuring interval as possible and the impact on regression analysis when samples are narrowly grouped.