Stroke prevention is the most important task of treating patients with non-valvular atrial fibrillation (AF). The most effective way for prevention of ischemic stroke in AF includes indefinite (life-long) use of oral anticoagulants. Long-term use of anticoagulants is associated with increased risk of bleeding. Therefore, maintenance of an optimal ratio of efficiency and safety in each individual patient is of particular importance, when the expected benefit (reduction in risk of ischemic stroke and thromboembolic complications in the blood vessels beyond the central nervous system) significantly exceeds the risk of clinically significant bleeding (especially intracranial and major extracranial bleeding). Analysis of data obtained in routine clinical practice (various kinds of registers and databases) allows to evaluate the efficacy and safety of oral anticoagulation in non-valvular AF in a wider range of patients. This approach has significant limitations, including less reliable the procedure of patient selection and analysis of outcomes, and nonrandomized (and usually retrospective) character of comparisons without certainty in the comparability of selected patients by severity criteria that can influence results. In clinical practice, physician makes a choice of oral anticoagulant, and a reason for that selection usually remains unclear. The situation is complicated by the fact that doses used in practice are often different from those studied in controlled clinical trials and recommended for widespread use. All this makes to interpret the results obtained with sufficient care, and to avoid categorical judgments and firm conclusions.

Keywords: NOAC, bleeding, oral anticoagulants, warfarin, stroke, atrial fibrillation