Comparative study of accuracy and clinical agreement of the CoaguChek XS portable device versus standard laboratory practice in unexperienced patients.

The objective of the study was to compare the accuracy and clinical agreement of the CoaguChek XS versus the standard laboratory practice. Forty-one patients on long-term anticoagulation with acenocumarol without previous experience in self-monitoring participated to obtain 218 pairs of data. Several methods for comparative statistics were applied to assess the possible disagreements between techniques as well as a range of previously published criteria of clinical agreement and the very recently described error-grid for INR comparison that we partially modify. The mean age was 52.1 and the indications for oral anticoagulation were prosthetic valves (36.59%), atrial fibrillation (34.15%), venous thromboembolic disease (21.95%) and others (7.31%) with a target range of 2-3 INR units (63.4%) or 2.5-3.5 (36.6%). Analyzing the whole series of data, the Pearsons rho correlation coefficient for precision between methods was 0.95 and the C(b) bias correction factor for accuracy 0.99 with a minimal bias of 0.1 INR units between methods applying the Bland-Altman plot. The linear regression procedure described by Passing and Bablok showed a minimal deviation from the best-fit line and a slope of 0.90. The mean of the absolute relative differences was 7% which is in the "very good" range of agreement. No results were found in the clinically "dangerous" D zone of the error-grids with 99% of data in the clinically irrelevant and low relevant areas A and B. In this study self-management with the CoaguChek XS was clinically safe and reliable.