Accuracy and Precision of the CoaguChek S versus Laboratory INRs in a Clinic

Abstract

BACKGROUND:The CoaguChek S is the next-generation coagulation monitor for measuring the international normalized ratio (INR) that replaces the CoaguChek device. Studies are lacking comparing the CoaguChek S with local laboratory INR assessment to ensure its accuracy and precision for monitoring patients on anticoagulation.OBJECTIVE:To evaluate accuracy, precision, and technical ease-of-use of the CoaguChek S compared with laboratory measurements.METHODS:Accuracy was evaluated in 101 patients by parallel assessment of INRs (CoaguChek S and laboratory); precision was evaluated in 31 patients using duplicate INRs from CoaguChek S and laboratory and from liquid quality controls. Accuracy was determined using orthogonal regression, Bland—Altman plot, and clinical applicability (INRs discrepant in categorization of INR goal and resulting in different therapeutic decisions). Precision was examined by comparing mean difference ± SD between repeated INRs from CoaguChek S and laboratory, coefficient of variation (CV), and coefficient of repeatability (CR). The influence of low and elevated INRs on accuracy and precision was also examined. To assess ease-of-use of the monitor, the number of technical errors was recorded.RESULTS:The CoaguChek S significantly correlated to laboratory measurement (r = 0.93); 16.7% of INRs resulted in discrepant categorization and 24.5% would have required a different therapeutic plan. The CV and CR compared well between CoaguChek S and laboratory (6% vs. 4.9%; 0.455 vs. 0.346, respectively). When subgroups of INR values <4.0 and <3.0 were evaluated, the precision improved with both methods. Precision, based on liquid quality controls, was good (CV 4.6% = low-level; 3.3% = high-level). The CoaguChek S was found to have an error rate of 1.8%.CONCLUSIONS:The CoaguChek S is an accurate and precise alternative to laboratory assessment of the INR at values <4.0; it is an efficient device with a low likelihood of errors during testing.