Time-updated body-weight change and clinical outcomes in anticoagulated patients with atrial fibrillation: a multicentre retrospective cohort study

Baseline body mass index is static, whereas longitudinal weight change may reflect evolving vulnerability during anticoagulated follow-up. We examined associations between time-updated percent body-weight change and outcomes in anticoagulated patients with atrial fibrillation. We studied patients with atrial fibrillation initiating warfarin or a non-vitamin K antagonist oral anticoagulant at four South Korean hospitals. The main cohort required outpatient baseline weight within 180 days before or on oral anticoagulant initiation and at least one later outpatient weight. Percent body-weight change was updated using outpatient measurements. The primary outcome was a composite of thromboembolism, major bleeding, or all-cause death. Time-dependent Cox models used natural splines, centre stratification, and robust standard errors. Among 3,878 patients, median age was 71.0 years and 41% were women. There were 349 primary outcome events, corresponding to 7.19 events per 100 person-years. Weight change showed a nonlinear association with the primary outcome. Compared with no change, adjusted hazard ratios with 95% confidence intervals were 2.04 (1.45–2.87) for 10% loss, 1.47 (1.20–1.81) for 5% loss, 1.26 (1.08–1.47) for 5% gain, and 1.57 (1.17–2.12) for 10% gain. Associations were strongest for death, present for major bleeding, and not clear for thromboembolism. In anticoagulated patients with atrial fibrillation, dynamic body-weight change was nonlinearly associated with outcomes, mainly through death and major bleeding. Repeated outpatient weight measurements may provide a simple marker of evolving clinical risk. Not applicable.