Jennifer Taher1, Jake Cosme1, Brian A Renley2, David J Daghfal2, Paul M Yip3. 1. Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, ON, Canada. 2. Diagnostics Division, Abbott Laboratories, Abbott Park, IL, USA. 3. Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, ON, Canada; Department of Clinical Biochemistry, University Health Network, Toronto, ON, Canada. Electronic address: paul.yip@utoronto.ca.
Abstract
OBJECTIVES: The Abbott Alinity family of chemistry and immunoassay systems recently launched with early adopters contributing imprecision and bias data, which was consolidated to assess the performance of Alinity assays across multiple sites using the Sigma metric. Multi-site Sigma metrics were determined for 3 ion-selective electrodes, 12 photometric assays, and 3 immunoassays across 11 independent laboratory sites in 9 countries. METHODS: Total allowable error (TEa) goals followed a previously defined hierarchy that used CLIA as the primary goal. Bias was calculated against the Abbott ARCHITECT system using Passing-Bablok regression analysis using individual site data or pooled aggregate data. Sigma metrics were calculated as (%TEa - |% bias|)/%CV. For individual-site analysis, the Sigma metrics for each assay were compared using the individual-site and the pooled biases. For multi-site analysis, the average CV and the pooled bias were used to generate a Pooled Sigma metric encompassing the global performance for a given assay. RESULTS: A total of 97 individual-site and 18 Pooled Sigma metrics were calculated for available assays. Individual Sigma metrics varied across sites, with 90% of assays performing 4 Sigma or higher, and 17 of 18 Pooled Sigma metrics indicated performance greater than 4 Sigma. Sigma metrics were significantly improved in 16 assays when using pooled bias rather than individual-site bias. CONCLUSIONS: This multi-center study applies a novel application of Sigma metrics to the first Alinity users and reveals analytical performance of greater than 4 Sigma for vast majority of assays. Laboratories with limited resources can leverage larger data sets for Pooled Sigma metric analysis, providing a tool to assess the consistency of analytical performance from multiple sites.
OBJECTIVES: The Abbott Alinity family of chemistry and immunoassay systems recently launched with early adopters contributing imprecision and bias data, which was consolidated to assess the performance of Alinity assays across multiple sites using the Sigma metric. Multi-site Sigma metrics were determined for 3 ion-selective electrodes, 12 photometric assays, and 3 immunoassays across 11 independent laboratory sites in 9 countries. METHODS: Total allowable error (TEa) goals followed a previously defined hierarchy that used CLIA as the primary goal. Bias was calculated against the Abbott ARCHITECT system using Passing-Bablok regression analysis using individual site data or pooled aggregate data. Sigma metrics were calculated as (%TEa - |% bias|)/%CV. For individual-site analysis, the Sigma metrics for each assay were compared using the individual-site and the pooled biases. For multi-site analysis, the average CV and the pooled bias were used to generate a Pooled Sigma metric encompassing the global performance for a given assay. RESULTS: A total of 97 individual-site and 18 Pooled Sigma metrics were calculated for available assays. Individual Sigma metrics varied across sites, with 90% of assays performing 4 Sigma or higher, and 17 of 18 Pooled Sigma metrics indicated performance greater than 4 Sigma. Sigma metrics were significantly improved in 16 assays when using pooled bias rather than individual-site bias. CONCLUSIONS: This multi-center study applies a novel application of Sigma metrics to the first Alinity users and reveals analytical performance of greater than 4 Sigma for vast majority of assays. Laboratories with limited resources can leverage larger data sets for Pooled Sigma metric analysis, providing a tool to assess the consistency of analytical performance from multiple sites.
Authors: Rosanna Inzitari; Sebastian Vencken; David Daghfal; Karl McAuley; Anthony McDermott; Marie Galligan; Peter Doran Journal: Pract Lab Med Date: 2021-03-26