Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia in clinical practice, with rising global prevalence that increases markedly with age [1]. As a progressive disease, AF increases the risk of ischemic stroke by five-fold [2], and is associated with an elevated risk of heart failure, cognitive impairment, myocardial infarction, and sudden cardiac death [3], posing serious threats to patients’ health and quality of life while imposing substantial burdens on socioeconomic systems and healthcare services. It is reported that about one-third of AF patients are asymptomatic, especially in the early stages of recognition, while the clinical outcomes of asymptomatic AF patients show no significant difference compared to symptomatic patients [4]. Therefore, identifying patients at risk for AF is crucial for its early prevention, diagnosis, and treatment.

Several risk schemes have been proposed to predict new-onset atrial fibrillation (NAF), including the Cohorts for Heart and Aging Research in Genomic Epidemiology-Atrial Fibrillation (CHARGE-AF) score [5], the Mayo AF score [6], and the HARMS2-AF score, using hypertension, age, raised body mass index (BMI), male sex, sleep apnea, smoking, and alcohol for assessment [7]. However, these scoring schemes were derived in European and American populations, and may not be fully applicable to Asian patients. The onset and maintenance of AF depend on electrical and structural remodeling of the atria [8]. Previous studies have shown that electrocardiographic P-wave indices can reflect underlying atrial remodeling [9]. The P-wave dispersion (Pd) represents the degree of variation in P-wave duration (PWD) across the 12 surface leads of the electrocardiogram (ECG) in sinus rhythm, reflecting the presence of spatially heterogeneous electrical activity within the atria. Multiple studies have demonstrated that increased Pd is independently associated with the occurrence of paroxysmal atrial fibrillation (PAF) [10, 11]. However, there is insufficient evidence to support an association between Pd and NAF in population-based observational studies. In this study, we conducted a retrospective analysis of the most recent sinus rhythm Pd within the preceding year in patients with NAF, using a simple, cost-effective, noninvasive, and widely accessible 12-lead surface ECG. The aim was to investigate the association between Pd and NAF within one year.

This study revealed that patients who developed NAF exhibited significantly higher Pd values on sinus rhythm ECGs during the preceding year compared to controls. Following comprehensive adjustment for potential confounders, Pd maintained a significant association with NAF within one year. Moreover, the GAM analysis demonstrated a nonlinear positive correlation between Pd and NAF risk.

AF represents a growing global health challenge with complex pathophysiology. Both non-modifiable factors (such as advanced age, genetic predisposition, and sex differences) and modifiable factors (such as HBP, DM, smoking, obesity, physical inactivity, and obstructive sleep apnea) contribute to increased AF susceptibility. Inflammation, oxidative stress, autonomic nervous system dysregulation, and activation of the renin-angiotensin-aldosterone system (RAAS) can induce both electrical and structural remodeling of the atria, thereby promoting both the initiation and perpetuation of AF [8]. Early risk prediction, screening, diagnosis, and individualized management are important for patients to reduce complications and improve quality of life. Therefore, population-based studies to explore the predictive value of noninvasive diagnostic techniques for NAF have long been of interest. The CHARGE-AF scheme [5] was an AF prediction model developed using individual-level data from 3 large cohorts in the United States (the Atherosclerosis Risk in Communities [ARIC] study, the Cardiovascular Health Study [CHS], and the Framingham Heart Study [FHS]), and validated in two European cohorts. The predictive performance of the simple and enhanced scores was comparable (AUC = 0.765 vs 0.767). The Mayo AF score [6] was an AF model developed based on seven recognized clinical risk factors. For patients with AF risk scores of 1, 2, 3, 4, 5 and above, the OR values for subsequent diagnosis of AF were 3.05, 12.9, 22.8, 34.0, 48.0, respectively, and AUC = 0.812. The HARMS2-AF score [7] was a novel lifestyle risk scoring scheme developed and validated using data from the UK Biobank (UKB) and the FHS. Over a 5-year follow-up period, the AUC was 0.782 in the UKB cohort and 0.757 in the FHS validation cohort. The above scoring schemes demonstrate good predictive efficacy, but their applicability to Asian populations remains to be validated due to differences in ethnicity and lifestyle. In addition to simple risk scoring models constructed from cardiovascular risk factors associated with AF, machine learning prediction models driven by artificial intelligence have also been rapidly advancing in recent years. Lubitz et al. [14] demonstrated that wearable devices might facilitate identifying individuals with undiagnosed AF. Raghunath et al. [15] used a deep network learning approach based on artificial intelligence techniques to predict NAF from resting 12-lead ECG. However, the high demands on signals, data, and decision models may limit their widespread clinical application.

As is well known, the simpler an AF risk assessment tool is, the easier it becomes to identify high-risk patients early on, thereby facilitating further monitoring and diagnosis of AF. This study employed a conventional 12-lead ECG, where the P-wave represents the depolarization of the left and right atria. The P-wave indices could reflect underlying atrial structure, size and electrical activity [16]. The regional delay in atrial depolarization might produce different PWD, as surface P-waves at various locations might be affected to varying degrees by regional variations in atrial activation time. This regional hypothesis explaining variations in PWD intervals was termed Pd [17]. Previous studies have shown that Pd has been recognized as a specific indicator of atrial abnormalities at the electrophysiological and anatomical levels [18]. Marks et al. [19] showed that Pd $>$40 ms was found to be the only independent correlate predicting the risk of AF by following 178 patients with cryptogenic stroke using implantable loop recorders for one year. A recent study of patients with early-onset hypertension identified Pd as a potential electrophysiologic parameter for predicting NAF in this population [20]. In agreement with prior findings, this study showed that increased Pd in sinus rhythm was associated with the risk of NAF within one year.

Pd reflects the prolongation of intra-atrial and interatrial conduction time, as well as nonuniform propagation of sinus impulses within the atria [21]. An increase in Pd signifies a shortened atrial effective refractory period and heightened spatial heterogeneity, accompanied by a loss of the normal heart rate dependence and adaptability. Slowed heterogeneous conduction and increased dispersion of refractoriness within atrial tissue render it highly susceptible to reentry, which can progress to AF in severe cases. A study [22] has reported that atrial dilation can cause atrial myofibrils to become stretched and deformed, leading to their uneven distribution. This subsequently results in non-uniform conduction of atrial depolarization, creating conditions conducive to the unidirectional block and areas of slow conduction required for AF initiation. Therefore, this study incorporated the relevant LAD as a covariate in the regression model. After comprehensive adjustment for confounding factors, the results demonstrated that Pd remained significantly associated with NAF and could serve as an independent predictor. The association between Pd and NAF may help clinicians identify high-risk populations and provides reference for clinical prevention and treatment of AF.

This study has several limitations. Firstly, it was a retrospective study derived from a single center. However, both the cases and controls were selected from a well-defined cohort, which minimized the potential for selection bias. Secondly, although potential confounding factors were adjusted for during the analysis, residual confounding factors may still exist, as is the case with any observational study. Finally, this study cannot yet be compared with other scoring schemes in terms of predictive efficacy for NAF. As mentioned in the discussion section, comparisons between different studies are inherently limited due to variations in study design and study populations. Therefore, future retrospective cohort or prospective studies may incorporate Pd into models such as CHARGE-AF to enhance diagnostic performance.

In summary, this study demonstrated that increased ECG Pd in sinus rhythm correlated with NAF within one year. After fully adjusting for confounding factors, Pd remained significantly associated with NAF and served as an independent predictor of this risk. This finding may assist clinicians in implementing more frequent follow-up examinations or long-term monitoring for high-risk patients, thereby facilitating early detection of AF.

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

MX designed the study and made significant contributions to data collection and analysis. CW, YM, PY and YL collected data and conducted the study. CW contributed to acquisition and interpretation of data and drafted the manuscript. All authors contributed to the editorial changes to the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The study was carried out in accordance with the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of the Third Affiliated Hospital of Soochow University (Protocol No. [2022] Tech-018). Since this was a retrospective study, the ethics committee of our hospital had waived informed consent.

We would like to express our sincere gratitude to everyone who provided assistance and advice for this article, which will be helpful for our later study.

This study was supported by Top Talent of Changzhou “The 14th Five-Year Plan” High-Level Health Talents Training Project (2022260).

The authors declare no conflict of interest.