PREDICTORS OF ATRIAL FIBRILLATION IN PATIENTS WITH ISCHEMIC STROKE OF UNDETERMINED ETIOLOGY

Aim To identify a complex of predictors and to create a mathematical model for prognosis of atrial fibrillation (AF) in patients with ischemic stroke of undetermined etiology. Material and methods The study included 981 patients with ischemic stroke. Effects of the following factors were evaluated: gender, a history of stroke, a history of thromboembolism, presence of diabetes mellitus, grade of arterial hypertension, functional class (FC) of chronic heart failure (CHF), age, data of blood biochemistry, and data of coagulogram. The prognostic model was constructed using the binary logistic regression. The value of area under the ROC curve for the proposed prognostic model was calculated. Results The main predictors of AF in patients with ischemic stroke of undetermined etiology were CHF FC, a history of stroke, age, gender, values of cholesterol and prothrombin index, which were included into the final prognostic model. The sensitivity of the developed model was 83.5% and the specificity was 85.596. The area under the ROC curve corresponding to the interrelation between the prognosis of AF and the regression function value was 0.921 +/- 0.012 with 95% confidence interval: 0.898-0.944. Conclusion According to the results of the study, the probability of AF in patients with ischemic stroke increased with CHF progression, recurrent stroke, older age, female gender, and reduced prothrombin index and cholesterol level.