A-010 Systemwide Implementation of High Sensitivity Troponin T in a Multi-center Hospital Set Up with Instrument Variability: The Baystate Health Experience

Abstract

Abstract Background High-sensitivity cardiac troponin (hs-cTn) testing helps in rapid risk stratification in patients with acute coronary syndromes resulting in early identification of low-risk patients eligible for early discharge. A high sensitivity troponin test can measure the 99th percentile upper reference limit with an analytical imprecision ≤10% and it can measure cTn above the limit of detection in ≥50% of healthy subjects. We carried out method validation of the Elecsys Gen 5 hsTnT assay with variable instrument capacities like Roche cobas e801, e601, and e411 in our health system. The analytical variability of these instruments was studied and standardized to implement the Gen 5 hsTnT assay systemwide for clinical use in four hospitals. Methods The cTnT control samples with low, medium, and high concentrations were used for precision studies (N = 20). The calibration materials at five levels ranging from 8 ng/L to 10 000 ng/L were used for evaluating linearity, accuracy, and analytical measurement range. The observed and assigned values were compared using linear regression analysis. Heparinized plasma samples were tested using the Gen 5 hsTnT assay. Method comparisons were carried out on different Roche cobas analyzers in each hospital splitting remnant specimens. Analyzer e801 performance was compared with a reference laboratory. Simultaneously, the e601 and e411 analyzers were compared with the e801. Statistical analysis was carried out using EP Evaluator and GraphPad Prism. Results The Roche e801 demonstrated imprecision of 0.5%–2.2% (N = 20), the e601 had imprecision of 0.4%–1.4% (N = 20) and the e411 had 0.8%–4.8% (N = 20) across the three levels of control samples tested. These were well below the expected imprecision criteria of ≤10% for troponins. Deming regression analysis of the 5th GEN assays showed linear relationship across platforms: y (e801 5th GEN) = 1.007x (Reference Lab 5th GEN) + 50.65, R = 0.9899 with positive bias of 8% (N = 45); y (e601 5th GEN) = 0.929x(e801 5th GEN) + (−1.426), R = 0.9995 with negative bias of 7% (N = 48); y (e411 5th GEN) = 0.942x(e801 5th GEN) + (−7.044), R = 1.000 with negative bias of 7% (N = 49). The accuracy, reportable range, and linearity of Gen 5 hsTnT were tested on e801, e611 and e411 analyzers over a measured range of 8–10 000 ng/L. Allowable systematic error (SEa) was 1.5 ng/L (conc) or 5.0% (e801); 1.5 ng/L (conc) or 7.5% (e611 and e411). The maximum deviation for a mean recovery from 100% was 2.4% (e801), 2.7% (e601) and 6.2% (e411). The 5 of 5 mean recoveries were accurate within the SEa for all the three analyzers. The 15 out of 15 results were accurate within the allowable total error (TEa) of 3 ng/L (conc) or 10.0% (e801), 3 ng/L (conc) or 15.0% (e601 and e411). The three analyzers passed reportable range and accuracy tests, and the results were linear. Conclusion The TEa was determined from instrument specific imprecision, systemwide medical decision limits and parameters of rule out criteria. The standardization of analytical performance characteristics is important in a large health system with instrument variability and provision of transfer of care across health systems.