Prediction of paroxysmal atrial fibrillation based on 24-hour Holter electrocardiographic monitoring

BACKGROUND: Predicting and detecting paroxysmal atrial fibrillation at an early stage is a key priority for preventing cardioembolic complications. AIM: This study aimed to develop a predictive tool for paroxysmal atrial fibrillation in patients with sinus rhythm. MATERIAL AND METHODS: A single-center case-control study was conducted involving 6630 patients. The main group comprised 97 individuals with newly diagnosed paroxysmal atrial fibrillation. The control group included 99 patients without atrial fibrillation, matched for anthropometric and comorbidity parameters. Standard laboratory and instrumental methods were used. During 24-hour Holter ECG monitoring, the following parameters were analyzed: sex, age, monitoring duration, and rhythm driver. Special attention was given to early ectopic beats of the “P on T” and “R on T” types. Differences were considered statistically significant at p ≤ 0.05. RESULTS: Holter ECG parameters (extrasystole, ectopy, and paroxysmal tachycardia) were significantly more frequent and had higher values in the main group. A specific type of early atrial extrasystole (“P on T”) was observed in 97.9% of the main group, compared with 4.0% of the control group (odds ratio, 8461.648; 95% CI, 382.1983–187336). The number of supraventricular extrasystoles (isolated, paired, grouped) was significantly higher in the main group. Interval durations were also significantly longer in this group. No significant differences between groups were found in the frequency of ventricular extrasystoles or ST-segment depression. A logistic regression model was developed based on the most significant predictors: sex; number of atrial and atrioventricular supraventricular extrasystoles; number of isolated and paired ventricular extrasystoles; presence of allorhythmia in ventricular ectopy; and presence of early “P on T” ectopic beats. The model yielded an area under the curve (AUC) of 0.996, with an optimal risk threshold of 0.5 and prediction accuracy of 97.45%. CONCLUSION: A predictive tool for paroxysmal atrial fibrillation in patients with sinus rhythm was developed, demonstrating high predictive performance.