Variability in Anticoagulation Control Predicts Thromboembolism After Mechanical Cardiac Valve Replacement: A 23-Year Population-Based Study

OBJECTIVE: To assess optimal control of blood anticoagulation to maximize antithrombotic protection after mechanical cardiac valve replacement. DESIGN: A population-based study of 96 patients with a mean follow-up of 7.7 years (range, 1 month to 23 years) was performed in Olmsted County, Minnesota, and 10,301 prothrombin time (PT) ratios were determined after mechanical heart valve replacement. MATERIAL AND METHODS: PT ratios were analyzed in a new time-dependent Cox proportional-hazards model by defining an algorithm for comparing variability in PT ratios at each month of follow-up and relating these to thromboembolic events. The new method was compared with several conventional time-independent definitions. RESULTS: During 740 person-years of follow-up, 19 of 96 patients (20%) had 27 thromboembolic events. Of these 19 patients, 8 (42%) had events within 3 months after valve replacement. Freedom from any thromboembolic event was 72% at 15 years. The event rate was high (7.5% per year) during high variability and low (0.9% per year) during low variability in the PT ratio. This relationship was lost when time dependence was removed. More PT ratios were less than 1.5 during high (27%) than during low (19%) variability. Several conventional definitions of adequacy of anticoagulation that averaged PT ratios before a thromboembolic event or throughout follow-up or that compared the proportion of PT ratios above or below a fixed ratio did not define or only partially defined different thromboembolic risks. CONCLUSION: Periods of high and low variability of PT ratios define high and low risk of thromboembolism, respectively.