Outliers as a Cause of False Cardiac Troponin Results: Investigating the Robustness of 4 Contemporary Assays

Abstract

It is important that cardiac troponin be measured accurately with a robust method to limit false results with potentially adverse clinical outcomes. In this study, we characterized the robustness of 4 analytical platforms by measuring the outlier rate between duplicate results. We measured cardiac troponin concurrently in duplicate with 4 analyzers on 2391 samples. The outliers were detected from the difference between duplicate results and by calculating a z value: z = (result 1 - result 2) ÷ √(SD1(est)² + SD2(est)²), with z > 3.48 identifying outliers with a probability of 0.0005. The outlier rates were as follows: Abbott Architect i2000SR STAT Troponin-I, 0.10% (0.01%-0.19%); Beckman Coulter Access2 Enhanced AccuTnI, 0.44% (0.25%-0.63%); Roche Cobas e601 TroponinT hs, 0.06% (0.00%-0.13%); and Siemens ADVIA Centaur XP TnI-Ultra, 0.10% (0.01%-0.19%). The occurrence of outliers was higher than statistically expected on all platforms except the Cobas e601 (χ² = 2.7; P = 0.10). A conservative approach with a constant 10% CV and z > 5.0 identified outliers with clear clinical impact and resulted in outlier rates of 0.11% (0.02%-0.20%) with the Architect i2000SR STAT Troponin-I, 0.36% (0.19%-0.53%) with the Access2 Enhanced AccuTnI, 0.02% (0.00%-0.06%) with the Cobas e601 TroponinT hs, and 0.06% (0.00%-0.13%) with the ADVIA Centaur XP TnI-Ultra. Outliers occurred on all analytical platforms, at different rates. Clinicians should be made aware by their laboratory colleagues of the existence of outliers and the rate at which they occur.