Background During anticoagulation for prevention of stroke in patients with non-valvular atrial fibrillation (NVAF), bleeding is the most serious complication. In Western countries, the incidences of major bleeding and intracranial hemorrhages with low-dose warfarin are known to occur at a rate of 0.4-1.3% and 0.2% per year, respectively. The purpose of this study was to investigate the incidence and risk factors for major bleeding related with warfarin therapy in Japanese patients with NVAF. Methods and Results From August 2004 to July 2005, 667 NVAF patients treated with warfarin for NVAF were followed-up. The target prothrombin time-international normalized ratio (PT-INR) value was set at 1.6-2.6 (low-dose warfarin). The exposure on warfarin was 503 patient-years (average PT-INR 2.00 ±0.40). During the follow-up period, 12 major bleeding complications occurred (2.38% per patient-year), which included 3 intracranial hemorrhages (0.60% per patient-year). Among the patients' characteristics, average PT-INR ≥2.27 during the study was identified as an independent risk factor for major bleeding. Conclusions The incidence of major bleeding and intracranial hemorrhages in Japanese NVAF patients with low-dose warfarin therapy was 2.38% and 0.60% per patient-year, respectively, which is higher than in Westerners. (Circ J 2007; 71: 761 - 765)
Prognosis of Japanese patients with atrial fibrillation (AF) should be determined in a prospective cohort study. A prospective cohort of The Shinken Database 2004 comprised details on all the new patients who visited The Cardiovascular Institute Hospital in 2004 (n=2,412), which included 286 AF patients (11.8%, 205 males, 64.1+/-12.3 years, 165 paroxysmal). In this Japanese cohort of AF patients, the prevalence of congestive heart failure (CHF) was low compared with that in Westerners. Total mortality and cardiovascular morbidity of these patients were also quite low, 1.7% and 11.2% at 1 year, respectively. However, patients with CHF exhibited higher mortality (7.3%), and hospitalization for CHF increased the rate to 22.6%. There was no significant difference in mortality between paroxysmal and persistent AF. Multiple Cox-hazard regression analysis identified hospitalization at initial visit and lack of anticoagulation as independent predictors for mortality, and existence of CHF as the only independent predictor for cardiovascular morbidity. CONCLUSIONS The present study identified the mortality (1.7%) and morbidity (11.2%) of Japanese AF patients during 1 year after initial visit. Hospitalization for CHF and lack of anticoagulation were independent predictors for mortality, suggesting important roles of co-morbidities of Japanese AF patients.
Non-linear models by machine learning may identify different risk factors with different weighting in comparison to conventional linear models. The analyses were performed in 15,933 patients included in the Shinken Database (SD) 2004–2014 (n = 22,022) for whom baseline data of blood sampling and ultrasound cardiogram and follow-up data at 2 years were available. Using non-linear models with machine learning software, 118 risk factors and their weighting of risk for all-cause mortality, heart failure (HF), acute coronary syndrome (ACS), ischemic stroke (IS), and intracranial hemorrhage (ICH) were identified, where the top two risk factors were albumin/hemoglobin, left ventricular ejection fraction/history of HF, history of ACS/anti-platelet use, history of IS/deceleration time, and history of ICH/warfarin use. The areas under the curve of the developed models for each event were 0.900, 0.912, 0.879, 0.758, and 0.753, respectively. Here, we described our experience with the development of models for predicting cardiovascular prognosis by machine learning. Machine learning could identify risk predicting models with good predictive capability and good discrimination of the risk impact.