Individual Treatment Effect Estimation of 2 Doses of Dabigatran on Stroke and Major Bleeding in Atrial Fibrillation

BACKGROUND: We aimed to estimate absolute benefit and harm from treatment with dabigatran in individual patients with atrial fibrillation, and to select the optimal dose for each individual. METHODS: We derived and validated a prediction model for ischemic stroke/systemic embolism and major bleeding in patients with atrial fibrillation from the 3 treatment arms of the RE-LY trial (Randomized Evaluation of Long-Term Anticoagulation Therapy With Dabigatran Etexilate) (n=11 955 in derivation cohort, n=6158 in validation cohort). Readily available patient characteristics were included in Fine and Gray competing risk models (sex, age, smoking, antiplatelet drugs, previous vascular disease, diabetes mellitus, blood pressure, estimated glomerular filtration rate, and hemoglobin). Five-year risks for ischemic stroke/systemic embolism and major bleeding were estimated without anticoagulation therapy, and compared with high- and low-dose dabigatran. RESULTS: Model calibration was good, and discrimination was adequate with a c-statistic of 0.65 (95% CI, 0.62-0.70) for ischemic stroke/systemic embolism and 0.69 (95% CI, 0.66-0.71) for major bleeding. The 5-year absolute risk reduction for ischemic stroke/systemic embolism with dabigatran 150 mg twice daily ranged from <10% in 20% of patients to >25% in 14% of patients, and the 5-year absolute risk increase for major bleeding ranged from <5% in 53% of patients to 15% to 20% in 1% of patients. Comparing high-dose to low-dose dabigatran, the net benefit (absolute risk reduction minus absolute risk increase) was positive for 46% of patients. CONCLUSIONS: The absolute treatment benefits and harms of dabigatran in atrial fibrillation can be estimated based on readily available patient characteristics. Such treatment effect estimations can be used for shared decision making before starting dabigatran treatment and to determine the optimal dose. CLINICAL TRIAL REGISTRATION: URL: https://www.clinicaltrials.gov . Unique identifier: NCT00262600.