Control of anticoagulation with vitamin K antagonists: overestimation of median time in therapeutic range when assessed by linear interpolation

Abstract

Patients receiving vitamin K-antagonists are monitored by regular assessment of the International Normalized Ratio (INR). There are two popular methods for therapeutic control of anticoagulation in patient groups: 1) Time in Therapeutic Range (TTR) assessed by linear interpolation of successive INR measurements; 2) the cross-sectional proportion (CSP) of all patients' last INRs within range. The purpose of the present study is to compare the two methods using data from 53 Dutch Thrombosis Centres and to develop a semi-quantitative model for TTR based on different types of INR change. Different groups of around 400,000 patients in four consecutive years were evaluated: patients in the induction phase, short-term, long-term, low-target range, high-target range, receiving either acenocoumarol or phenprocoumon, and performing self-management. Each Centre provided TTR and CSP results for each patient group. TTR and CSP were compared using the Wilcoxon signed-rank test. Separately, we analysed the relationship between consecutive INR results regarding in or out of range and their frequency of occurrence in patients of two different cohorts. Good correlation was observed between TTR and CSP (correlation coefficient 0.694-0.950 in low-target range). In long-term acenocoumarol patients (low-target range) the median TTR was significantly higher than CSP (80.0 % and 78.7 %, respectively; p<0.001). In long-term phenprocoumon patients (low-target range) there was no significant difference between median TTR (83.0 %) and median CSP (82.6 %). In conclusion, the correlation between TTR assessed by linear interpolation and CSP was good. TTR assessed by linear interpolation was higher than CSP in patients on acenocoumarol.

Control of anticoagulation with VKAs: overestimation of median TTR when assessed by linear extrapolation: A comment

“…In the absence of a gold standard, their conclusion that one method overestimates is equally justified as the conclusion that the other method consistently underestimates. This is particularly the case since the four hypothetical scenarios that were presented are not the whole spectrum of scenarios with their respective probabilities (i. e. when both measurements are within the therapeutic range (see Figure 1A in [1]), it seems more likely that time between was spent within than outside the range).…”

Control of anticoagulation with VKAs: overestimation of median TTR when assessed by linear interpolation: Reply

“…We thank Drs. Rosendaal and Cannegieter for their critical comments on our contribution (1,2). We compared the Time in Therapeutic Range (TTR), assessed by linear interpolation, to the cross-sectional proportion (CSP) of all patients' last International Normalised Ratio (INR), also known as the cross-section-of-the-files method.…”