Correction factor to improve agreement between point-of-care and laboratory International Normalized Ratio values

Abstract

Purpose. Results of a research project to quantify and improve the accuracy of point-of-care (POC) International Normalized Ratio (INR) values are reported. Methods. The accuracy of POC INR values relative to laboratorymeasured INR values was retrospectively assessed in a cohort of patients with same-day INR determinations by both methods. Univariate linear regression was performed to derive a correction factor for POC INR values of >3; this correction factor was validated in a second cohort. Results. In the derivation cohort (259 patients and 344 paired INR results), agreement of POC values with corresponding laboratory INR values at two specified thresholds (±15% and ±25%) was 51.2% and 66.6%, respectively; for POC INR values of >3 (n = 205), agreement was lower (24.9% and 44.9%, respectively). Univariate linear regression yielded a coefficient of 0.77 (95% confidence interval, 0.76-0.79; p < 0.001). Applying a correction factor of 0.8 to POC INR values in a validation cohort (169 patients and 209 paired INR values) significantly improved the accuracy of POC INR values of >3 relative to laboratory values (from 7% to 71.1% at the lower threshold and from 23.5% to 88.8% at the higher threshold, p < 0.0001 for both comparisons). Conclusion. Agreement between POC and laboratory INR results in one institution was poor, especially when POC INR values exceeded 3. Application of an institution-specific correction factor to POC INR values of >3 improved agreement with laboratory INR results but would not have significantly reduced differences in protocol-based warfarin dosage adjustments.