Anticoagulation in patients with non-valvular atrial fibrillation: an evaluation of stability and early factors that predict longer-term stability on warfarin in a large UK population

Objective:To determine the proportion of patients with non-valvular atrial fibrillation (NVAF) treated with warfarin that achieved a 6-month period within the target INR range (stability). To then evaluate any associations between stability and outcome and to determine whether stability can be predicted by clinical factors at an early stage in warfarin treatment. Methods: This study was a record linkage study in 1513 patients with NVAF treated with warfarin for a minimum of 6-months, carried out in a large UK population. The main outcome measures were stability (defined as six months within the target INR range [2.0-3.0]), thromboembolic and bleeding event rates and mortality. Secondary outcome measures were the predictive value of baseline characteristics and other treatment variables. Results: Stability was achieved in 52% of the study group. Standardised mean survival was significantly higher in the group who achieved stability (Delta = 16.91 months, p < 0.001) with a hazard ratio of 4.36 (p < 0.001). The stable group had a lower rate of both thromboembolic events (0.8% vs. 2.3% per patient year) and bleeds recorded on inpatient diagnoses (0.4% vs. 1.2% per patient year). Failure to achieve stable control was associated with age (Odds Ratio [OR] 1.011 (95% Confidence Interval [CI] 1.001-1.021)) and morbidity at baseline (OR 1.015; 95% Cl 1,007-1.022). An increase in mean time between visits (OR 0.939; 95% Cl 0.926-0.952) and the percentage time in range (OR 0.889; 95% Cl 0.879-0.900) was associated with a decrease in the chance of instability. Greater variability in INR was also associated with a failure to achieve stability (OR 1.518; 95% Cl 1.427-1.615). Receiver Operator Characteristic (ROC) analysis using data from the first three months of treatment demonstrated good discrimination of stability using age and morbidity at baseline and percentage time in range and frequency of visits during the first three months of treatment (area under curve [AUC] 0.780; standard error [SE] 0.012; 95% Cl 0.757-0.803). Conclusions: Many patients never achieved a period of 6-months stability and were at increased risk of thromboembolic events and bleeds. Age, morbidity at baseline and variability of INR control in the first three months could be used to predict instability using warfarin. This study infers that patients should be treated more intensively in the early stages of warfarinisation in order to improve outcome.