Comparing patient-level and site-level anticoagulation control as predictors of adverse events

Introduction Percent time in therapeutic range (TTR) is increasingly used to summarize anticoagulation control over time among patients receiving warfarin. Higher TTR improves outcomes of care, but studies have varied regarding whether TTR is best summarized as center-based percent time in therapeutic range (cTTR) or as individual percent time in therapeutic range (iTTR). Our aim was to compare cTTR to iTTR in predicting ischemic stroke, major hemorrhage, and all-cause mortality. Materials and methods Veterans Health Administration data of 57,281 patients receiving warfarin therapy were included. iTTR was calculated using linear interpolation. Each site's mean TTR was calculated, and the cTTR was assigned to all patients at that site. We used Cox proportional hazards to examine cTTR and iTTR as predictors of major hemorrhage, ischemic stroke, and all-cause mortality. Results Comparing worst to best quartiles of INR control, cTTR was not a statistically significant predictor of major hemorrhage or ischemic stroke, hazard ratios (HR) were 1.02 (95% confidence interval [CI] 0.93-1.11) and 1.00 (95% CI: 0.88-1.13), respectively. cTTR was a weak predictor of all-cause mortality (HR: 1.14, 95% CI: 1.07-1.22). iTTR predicted major hemorrhage (HR: 1.79, 95% CI: 1.63-1.96), ischemic stroke (HR: 1.91, 95% CI: 1.67-2.19), and all-cause mortality (HR: 2.20, 95% CI: 2.05-2.35). Conclusion iTTR significantly predicted risk of major hemorrhage, ischemic stroke, and all-cause mortality. cTTR was a weak predictor of all-cause mortality. Though cTTR may be a better target for site-level quality improvement efforts, iTTR may be a more suitable measure for use in comparative effectiveness research.