Accuracy and precision of a portable anticoagulation monitor in a clinical setting

Abstract

Background: Office-based anticoagulation monitors offer significant advantages in convenience, yet their performance has been inadequately characterized.

Methods: We characterized the performance of a portable anticoagulation monitoring system with respect to precision and agreement with a reference laboratory. Eighty-five patients from a university outpatient anticoagulation clinic provided 143 whole blood sample pairs for evaluating agreement between the monitor and the laboratory. Fifty-four patients each provided a second pair of samples for assessing the monitor's precision, and 23 pairs of measurements from the reference laboratory were used for assessing the laboratory's precision. Anticoagulation was measured using International Normalized Ratio (INR) values. Agreement between monitor and laboratory was evaluated as the difference between paired measurements. Precision was calculated as the within-patient standard deviation based on paired values.

Results: Within the range of 2.0 to 3.0 INR units, the monitor yielded values that were up to 0.3 units higher on average than the laboratory values. Within the range of greater than 3.0 to 4.5 INR units, the monitor yielded values that were up to 0.5 units lower on average than the laboratory values. Seventy-five percent of paired monitor and laboratory values were within 0.7 INR units; 90% were within 0.9 units. Within-patient standard deviation was 0.23 units for the monitor and 0.19 units for the laboratory.

Conclusions: The monitor differed systematically from the laboratory and was moderately less precise. The magnitude of these effects was not great, however, and accuracy was best at around INR = 3.0, the border between low and high therapeutic ranges. The clinic-based monitor is useful for patients requiring frequent surveillance of anticoagulation status.