Abstract
BACKGROUND: Specific risk predictor scores of intracranial hemorrhage (ICH) risk in Asian subjects are lacking. We determined the incidence rate and predictors of ICH in patients with non-valvular atrial fibrillation (AF). METHODS: A prospective nationwide registry of patients with AF was conducted from 27 hospitals in Thailand. The adjudicated primary outcome was the development of ICH during follow-up. Multivariable Cox proportional hazard model was performed to identify the independent predictors for ICH. A predictive model for ICH risk was developed and validated by bootstrap, calibration plot, C-statistics, and decision curve analysis using our own data. RESULTS: We studied a total of 3405 patients (mean age 67.8 years; 58.2% male) with an average follow-up duration of 31.8 ± 8.7 months, during which ICH developed in 70 patients (2.06%). The incidence rate of ICH was 0.78 (0.61-0.98) per 100 person-years. Predictors of ICH were chosen from the theory-driven approaches in combination with the results of the univariable analysis. The predictive risk model had a c-index of 0.717 (0.702-0.732) with good calibration, internal validation, and clinical usefulness using decision curve analysis. The probability of ICH at 3 years for an individual patient derived from the prediction model was compared with the probability derived from HAS-BLED score by using the C-statistics. The ICH probability from the COOL-AF model was superior to the HAS-BLED score in the prediction of ICH. CONCLUSION: The incidence rate of ICH was 0.78 (0.61-0.98) per 100 person-years. Predictors of ICH were older age, male sex, nonsmoking, renal replacement therapy, and use of oral anticoagulants.