Large clinical laboratories often rely on multiple chemistry analyzers. However, when a new analyzer is introduced, the laboratory must establish whether the old and new methods are comparable and can be used interchangeably. In this study, we compared the newly introduced Atellica CH930 chemistry analyzer with the already established Architect ci4100 and Cobas 6000 c501 from our laboratory. Patient samples were randomly selected from daily routine testing and a total of 22 analytes were investigated. Total error (TEobs) between test (Atellica) and comparative (Architect and Cobas) methods was calculated at relevant medical decision levels (MDL). For demonstrative purposes, the assessment of method comparability was based on three different criteria: allowable total error (TEa) derived from biological variation (BV), CLIA proficiency testing criteria for acceptable analytical performance, and CLIA-calculated Sigma metrics. These sets of analytical performance specifications were also compared, and their strengths and limitations are discussed in this paper. Performance of Atellica CH930 against Architect ci4100 was acceptable or nearly acceptable at 82%, 95%, and 64% of the 22 investigated MDLs across 9 analytes, according to BV-TEa, CLIA-TEa, and CLIA-calculated Sigma metrics, respectively. Similarly, performance of Atellica CH930 against Cobas 6000 c501 was acceptable or nearly acceptable at 61%, 93%, and 63% of the 54 investigated MDLs across 22 analytes, according to BV-TEa, CLIATEa, and CLIA-calculated Sigma metrics, respectively. However, method comparability should not be evaluated by a “one size fits all” approach as some analytes require different criteria of acceptability, ideally based on medically allowable error and clinical outcome.

No abstract