Paroxysmal atrial fibrillation prediction score based on 24-hours electrocardiogram-monitoring data

Abstract

Abstract Background Paroxysmal atrial fibrillation may be an incidental finding during routine 24-hour Holter electrocardiogram monitoring. Prediction and early detection of atrial fibrillation paroxysms is one of the priorities for the prevention of cardioembolic complications. Aim To create a method for early diagnosis of paroxysmal atrial fibrillation in patients with sinus rhythm. Materials and methods A single-center case-control study was conducted involving 6630 patients. The main group included 97 people with newly diagnosed paroxysmal atrial fibrillation; the control group included 99 patients without atrial fibrillation. Both groups were equivalent in anthropometric data and comorbidity. We used standard laboratory and instrumental methods. In 24-hours Holter electrocardiogram monitoring, the following parameters were analyzed: gender; age; observation time; main pacemaker; analysis of supraventricular and ventricular ectopic activity; the presence of pauses, blockades; ST segment; QT intervals; heart rate variability. We paid attention to the presence of early extrasystoles types "P on T" and "R on T". Results A logistic regression equation "Arfa" has been developed, which includes: gender, number of atrial and atrioventricular supraventricular extrasystoles, number of single and paired ventricular extrasystoles, allorhythmias during ventricular extrasystoles, presence or absence of early "P to T" extrasystoles. Pr = 1 / (1 + EXP ( - ( - 3,65982 + 0,000144 * Atrial ES + 0,000354 * Atrioventricular ES + 0,000842 * Single ventricular ES - 0,01613 * Couple ventricular ES - 0,002863 * Ventricular allorhythmias - 1,596521 * Gender m + 9,043299 * «Р on Т» ES))). Where: ES - extrasystoles, gender f = 0, m = 1; «P on T» ES = 1, absence of «P on T» ES = 0. AUC=0,996, and the optimal value of the risk coefficient for paroxysmal AF, providing the best separation of predicted outcomes, is 0,5 with a prediction accuracy of 97,45%. Conclusions The risk of developing paroxysmal atrial fibrillation in patients with sinus rhythm using "Arfa" regression: Pr>0,5 – high risk of developing paroxysmal atrial fibrillation. We recommend treatment of extrasystoles with further monitoring of 24-hour Holter electrocardiogram. If Pr<0,5 – low risk of developing paroxysmal atrial fibrillation.Confusion plotROC-curve of prognosis model “Arfa”