Long-term incidence and risk factors of atrial fibrillation after percutaneous patent foramen ovale occlusion: a retrospective cohort study Authors: Oscar Rommens, Maarten Blondeel, Mathias Maes, Marie-Paule Goidts, Christophe Garweg, Els Troost, Peter Haemers, Tomas Robyns, Alexander Van De Bruaene, Joris Ector, Rik Willems, Werner Budts, Pieter De Meester, Bert Vandenberk Published: 2026 ABSTRACT AbstractBackgroundPercutaneous patent foramen ovale (PFO) closure is an established intervention for secondary prevention of cryptogenic stroke. Increasing evidence indicates that the procedure is associated with a risk of new-onset atrial fibrillation (AF). This study aimed to describe the temporal distribution and clinical factors associated with new-onset AF following percutaneous PFO closure.MethodsPatients aged ≥16 years undergoing percutaneous PFO closure between 1999 and 2021 at a tertiary care centre were retrospectively included. New-onset AF was identified through systematic review of electronic health records. Descriptives are presented as median with IQR. Time-to-event analysis was performed using cumulative incidence functions and Cox proportional hazards regression.ResultsAmong 867 patients (median age 51 years (42–58), 47% female), PFO closure was performed for secondary stroke prevention in 86%. During a median follow-up of 11.7 years (6.3–16.8), new-onset AF occurred in 97 patients (11.2%), corresponding to an incidence rate of 1.0 per 100 patient-years. AF diagnoses clustered early after closure, with 26.8% occurring within the first 3 months. In multivariable analysis, higher age at closure (HR 1.05 per year; 95% CI 1.03 to 1.07), larger occluder device size (HR 1.09 per mm; 95% CI 1.03 to 1.16) and any moderate or severe valvular regurgitation (HR 3.70; 95% CI 1.65 to 8.33) were independently associated with new-onset AF.ConclusionsNew-onset AF is common after PFO closure and shows early postprocedural clustering. Increasing age, larger occluder size and valvular regurgitation are independently associated with AF occurrence, underscoring the need for focused rhythm surveillance after closure.Keywords: Atrial Fibrillation; Heart Defects, Congenital; Heart Septal Defects, Atrial WHAT IS ALREADY KNOWN ON THIS TOPIC WHAT THIS STUDY ADDS In this single-centre retrospective cohort of 867 patients with a median follow-up of 11.7 years, new-onset atrial fibrillation occurred in 11.2%, with approximately one-quarter of cases diagnosed within 3 months after closure. Older age, larger occluder device size and moderate or severe valvular regurgitation were independently associated with new-onset atrial fibrillation, while residual shunting was associated with subsequent arterial embolic events. In this single-centre retrospective cohort of 867 patients with a median follow-up of 11.7 years, new-onset atrial fibrillation occurred in 11.2%, with approximately one-quarter of cases diagnosed within 3 months after closure. Older age, larger occluder device size and moderate or severe valvular regurgitation were independently associated with new-onset atrial fibrillation, while residual shunting was associated with subsequent arterial embolic events. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These findings support focused rhythm surveillance after patent foramen ovale closure, particularly in patients with older age, larger devices or structural cardiac abnormalities. Future studies should combine systematic rhythm monitoring with detailed anatomical and clinical phenotyping to distinguish transient procedure-related atrial fibrillation from atrial fibrillation reflecting an underlying atrial substrate. These findings support focused rhythm surveillance after patent foramen ovale closure, particularly in patients with older age, larger devices or structural cardiac abnormalities. Future studies should combine systematic rhythm monitoring with detailed anatomical and clinical phenotyping to distinguish transient procedure-related atrial fibrillation from atrial fibrillation reflecting an underlying atrial substrate. Introduction The presence of a patent foramen ovale (PFO), a remnant of fetal circulation, is seen in approximately 25% of adults.1 This persistent connection between the systemic and pulmonary circulation is associated with increased risk of several cardiovascular complications, such as cryptogenic stroke or systemic emboli.1 2 Clinical trials have shown benefit of transcatheter PFO closure in the prevention of recurrent stroke and prevention of decompression sickness.3–7 Therefore, guidelines recommend transcatheter PFO closure in selected patients considered at risk.1 8 Since the introduction of the procedure, it is known that percutaneous PFO closure may contribute to new-onset atrial fibrillation (AF), especially shortly after the procedure. This in itself is directly associated with an increased risk of transient ischaemic attack (TIA) and stroke, creating a competing risk when considering PFO closure.9 10 While the underlying pathophysiology remains largely unknown, inflammation and atrial stretch due to device implantation are considered to be drivers for, mainly short-term, new-onset AF.11 12 More recent data also suggest a higher incidence of late-onset AF. This relationship, however, is less well established.13 14 Several studies and meta-analyses have tried to estimate the incidence of new-onset AF following PFO closure, with incidence rates varying widely depending on the duration of follow-up, diagnostic approaches used and criteria for defining AF.9 12 15 This study aims to assess the incidence of new-onset AF following transcatheter PFO closure in a real-world setting and to identify clinical variables associated with the development of new-onset AF using healthcare records of all patients who underwent PFO closure since the inception of the transcatheter PFO closure programme in a tertiary care centre. Methods Patient selection All patients aged ≥16 years who underwent percutaneous PFO closure in a tertiary care centre from 1999 through 2021 were included in a single-centre, retrospective registry. PFO closure was performed for established clinical indications in accordance with the clinical guidelines at the time of the procedure. Patients with a diagnosis of AF or atrial flutter prior to the procedure, patients without follow-up clinical records, patients who underwent cardiac transplant, and patients with an ASD were excluded (figure 1). Cumulative incidence curves for AF. (A) New-onset AF from time of PFO closure. (B) New-onset AF by size of the closure device. The cumulative number at risk was 858 rather than 867 because device size information was unavailable for 9 patients (1.0%). AF, atrial fibrillation; PFO, patent foramen ovale. Data acquisition Demographic, clinical and imaging data, as well as data related to the procedure were collected by reviewing the electronic medical records. Transthoracic echocardiography (TTE) and transoesophageal echocardiography (TEE) were performed according to contemporary guideline recommendations for comprehensive cardiac and valvular assessment. Valve morphology and function were evaluated using standard quantitative and qualitative parameters.16–18 The presence of a PFO, interatrial shunting and atrial septal morphology and mobility were assessed using contrast-enhanced TEE at the time of diagnosis and again during the PFO closure procedure with procedural guiding TEE. Echocardiography reports were stored in the electronic medical record in a structured format, allowing systematic extraction of predefined variables for analysis. When multiple TTEs or TEEs were performed, the exam closest to the PFO occlusion with adequate reporting of echocardiographic parameters was used as the index exam. Preprocedural AF or atrial flutter was defined by documentation in physician reports, medical history or by electrocardiographic evidence on 12-lead ECG or Holter monitoring prior to the procedure. In the context of secondary stroke prevention, 24-term or long-term Holter monitoring up to 30 days was routinely performed prior to PFO closure. Data related to the PFO closure included the indication, size (right atrial disk diameter) and type of the occlusion device. All procedural related parameters were stored in an institutional registry using a predefined template. All post-procedural medical reports were reviewed systematically in relation to the prespecified relevant clinical parameters. Information on the presence of a residual shunt at any point after the procedure was extracted, as was detailed information on the clinical events of interest defined below. Clinical follow-up and endpoints Follow-up ranged from the time of PFO closure until either occurrence of all-cause mortality or last known follow-up on or before 31 May 2025. Patients were followed in routine clinical practice with evaluations conducted at predefined, fixed intervals. Endpoints after device implantation include (1) first AF diagnosis, (2) first stroke or TIA and (3) any first arterial embolism. AF was defined as a diagnosis of new-onset AF or atrial flutter by the treating physician, detected on ECG, telemetry or Holter, starting from time of PFO closure. The AF endpoint was further divided into early AF, defined as atrial arrhythmia diagnosed within the first 3 months following the procedure, and overall AF. For patients who developed early AF, recurrences beyond 3 months following PFO closure were also assessed. Statistical analysis All continuous variables showed a non-normal distribution using Kolmogorov-Smirnov testing for normality and are presented as median with IQR. Categorical variables are presented with absolute numbers and percentages. P values listed for patient demographics are the result of Mann-Whitney U testing for continuous variables and χ² for categorical variables. Endpoints were analysed using time-to-event methods. Incidences are reported as a percentage per year together with the 95% CI and presented using cumulative incidence functions (CIFs) with log-rank testing if relevant. Univariate and multivariate Cox proportional hazard regression models were applied for the clinical endpoints reporting HRs with 95% CI. Variables that were statistically significant in univariate analysis were included in a multivariate forward stepwise model (entry p<0.05, removal p>0.10). Proportional hazard assumptions were tested using Schoenfeld residuals, and variables violating this assumption were stratified accordingly. Collinearity was assessed using a covariance matrix. When the correlation coefficient was greater than 0.4, the variable that resulted in the lowest Harrell’s C-index was excluded from the multivariate analysis. Univariate Cox regression analyses are available as online supplemental tables 2,3. Device occluder size was dichotomised based on the Youden index for new-onset AF as outcome. P values <0.05 were considered significant. Missing data were handled by listwise deletion. Statistical analyses and visualisation were performed using Stata V.18 (Stata) and GraphPad Prism V.10 (GraphPad Software, La Jolla, California, USA). Results Cohort selection and characteristics From 1999 to 2021, a total of 948 transcatheter PFO closure procedures were registered. Of these, 81 (8.5%) were excluded (online supplemental figure 1), resulting in 867 (91.5%) patients included in the present analysis. Patient characteristics are outlined in table 1. The median age was 51 years (IQR 42–58) and 404 (46.6%) were female. Secondary prevention for stroke or TIA was the most frequent indication in 793 patients (91.5%), followed by peripheral embolisms (n=38, 4.4%), decompression sickness (n=10, 1.2%), desaturation (n=7, 0.8%), migraine (n=6, 0.7%) and other (n=13, 1.5%). Cardiovascular risk factors were prevalent with a median BMI of 25 kg/m2 (IQR 23–28), 323 (37.4%) patients had a history of smoking, 239 (27.7%) patients had arterial hypertension and 472 (54.4%) patients had dyslipidaemia. A mobile interatrial septum was present in 333 patients (38.8%), with 255 patients having a documented aneurysmal interatrial septum (29.7%). Transcatheter PFO closure was mostly performed using Occlutech (Occlutech, Jena, Germany) (42.3%) and Cardia PFO occluders (Cardia, Eagan, Minnesota, USA) (28.9%). The median occluder size of the right-sided disk or umbrella was 25 mm (IQR 25–29). Table 1Baseline patient characteristicsAll(N=867)No AF(N=770)AF(N=97)P valueFemale sex, no. (%)404 (46.6)363 (47.1)41 (42.3)0.364Age, years (IQR)51 (42–58)50 (42–57)56 (49–62)<0.001Body mass index, kg/m² (IQR)25 (23–28)25 (23–28)26 (23–29)0.066Estimated GFR, mL/min (IQR)92(78–104)94(80–105)80(71–93)<0.001Medical history, no. (%) Stroke or TIA748 (86.3)664 (86.2)84 (86.6)0.922 Diabetes mellitus51 (5.9)45 (5.8)6 (6.2)0.893 Ischaemic cardiomyopathy26 (3.0)20 (2.6)6 (6.2)0.051 Smoking history323 (37.3)285 (37.0)38 (39.2)0.678 Hypertension239 (27.6)202 (26.2)37 (38.1)0.013 Dyslipidaemia472 (54.4)416 (54.0)56 (57.7)0.490Medication, no. (%) ACE inhibitor, ARB or ARNI145 (16.7)126 (16.4)19 (19.6)0.423 Beta-blocker108 (12.5)85 (11.0)23 (23.7)<0.001 Calcium-channel blocker70 (8.1)61 (7.9)9 (9.3)0.644 Mineralocorticoid receptor antagonist15 (1.7)10 (1.3)5 (5.2)0.006 Thiazide diuretic55 (6.3)43 (5.6)12 (12.4)0.010 Loop diuretic8 (0.9)7 (0.9)1 (1.0)0.906 Statin353 (40.8)316 (41.1)37 (38.1)0.578Echocardiography findings, no. (%) Atrial septal aneurysm255 (29.7)217 (28.4)38 (40.0)0.020 Mobile interatrial septum333 (38.8)304 (39.8)29 (30.5)0.0079Atrioventricular valve regurgitation<0.001 None287 (33.1)272 (35.3)15 (15.5) Mild531 (61.3)458 (59.5)73 (75.3) Moderate or severe49 (5.7)40 (5.2)9 (9.3)Aortic valve regurgitation0.035 None750 (88.9)671 (89.5)79 (84.0) Mild82 (9.7)71 (9.5)11 (11.7) Moderate or severe12 (1.4)8 (1.1)4 (4.3) Right-ventricular dilation14 (1.7)12 (1.6)2 (2.2)0.721 Left-ventricular dilation16 (1.9)14 (1.9)2 (2.2)0.863 Atrial dilation152 (17.5)127 (16.5)25 (25.8)0.024 Device size, mm (IQR)25 (25–29)25 (25–26)25 (25–30)<0.001 Residual shunt, no. (%)65 (7.5)55 (7.2)10 (10.3)0.268 Mortality, no. (%)85 (9.8)73 (9.5)12 (12.4)0.367 Follow-up, year (IQR)11.7(6.3–16.8)11.1(6.0–16.2)16.0(12.0–19.8)<0.001AF, atrial fibrillation; ARB, angiotensin–receptor blocker; ARNI, angiotensin receptor–neprilysin inhibitor; GFR, glomerular filtration rate; TIA, transient ischaemic attack. Baseline patient characteristics AF, atrial fibrillation; ARB, angiotensin–receptor blocker; ARNI, angiotensin receptor–neprilysin inhibitor; GFR, glomerular filtration rate; TIA, transient ischaemic attack. New-onset AF following PFO closure During a median follow-up of 11.7 years (IQR 6.3–16.8), new-onset AF occurred in 97 patients (11.2%) at an incidence rate of 1.0 per 100 patient-years (95% CI 0.8 to 1.3). The CIF (figure 1A) demonstrates an early peak in AF occurrence as 26 (26.8%) developed AF within the first 3 months following closure with an incidence rate of 12.8 per 100 patient-years (95% CI 8.5 to 18.4). Of these 26, 8 (30.7%) had AF recurrences beyond the first 3 months. Following the first 3 months, the incidence rate of new-onset AF dropped to 0.8 per 100 patient-years (95% CI 0.6 to 1.0). The results of a multivariable Cox regression for new-onset AF are shown in table 2. Older age at the time of closure (HR 1.05/year, 95% CI 1.03 to 1.07; p<0.001), a larger occluder device size (HR 1.09/mm increase in size, 95% CI 1.03 to 1.16; p=0.004) and mild (HR 1.91, 95% CI 1.05 to 3.49; p=0.035) and moderate or severe valve regurgitation at baseline (HR 3.70, 95% CI 1.65 to 8.33; p=0.002) were significantly associated with a higher incidence of new-onset AF. Based on the Youden index, device occluder size was dichotomised at 27 mm and the CIF (figure 1B) shows both early and late new-onset AF divergence. Following a forward stepwise approach, dilation of the left ventricle (HR 4.68, 95% CI 1.10 to 19.85; p=0.036) was the only predictor of new-onset AF within 3 months following PFO closure. There was no significant difference in AF incidence between different types of PFO occluder devices (p=0.423). Table 2Multivariable Cox regressions for new-onset atrial fibrillationOverallDuring first 3 monthsHR (95% CI)P valueHR (95% CI)P valueAge (per year)1.05(1.03 to 1.07)<0.001Device size (per mm)1.09(1.03 to 1.16)0.004Any valve regurgitation NoneReference Mild1.91(1.05 to 3.49)0.035 Moderate/severe3.70(1.65 to 8.33)0.002Left ventricular dilation4.68(1.10 to 19.85)0.036 Multivariable Cox regressions for new-onset atrial fibrillation Arterial embolism following PFO closure During follow-up, 59 patients (6.8%) experienced an ischaemic stroke or TIA after PFO closure (figure 2A; incidence rate 0.6 per 100 patient-years, 95% CI 0.5 to 0.8). Acute myocardial infarction occurred in 21 patients (2.4%, incidence rate 0.2 per 100 patient-years, 95% CI 0.1 to 0.3), while 3 patients (0.3%) experienced other arterial embolic events. The overall incidence rate for any arterial embolism was 0.8 events per 100 patient-years (95% CI 0.7 to 1.0). Residual shunting was documented in 65 patients (7.5%) at some point during follow-up. Cumulative incidence curves for outcomes. (A) Stroke or TIA and any arterial embolism after PFO occlusion. (B) Any arterial embolism after PFO occlusion by presence of residual shunting. PFO, patent foramen ovale; TIA, transient ischaemic attack. In multivariate Cox regression (table 3), new arterial embolism after PFO closure was associated with ischaemic cardiomyopathy (HR 3.08, 95% CI 1.38 to 6.91), arterial hypertension (HR 1.94, 95% CI 1.21 to 3.09), statin use (HR 1.22, 95% CI 1.07 to 1.38) and documentation of a residual shunt after closure (figure 2B, HR 2.83, 95% CI 1.55 to 5.16). Better renal function (HR 0.98 per mL/min increase in eGFR, 95% CI 0.97 to 0.99) and female gender (HR 0.53, 95% CI 0.33 to 0.86) were associated with a lower incidence of arterial embolism after PFO closure. Table 3Multivariable Cox regression for any arterial embolism following PFO closureHR (95% CI)P valueeGFR (per mL/min)0.98 (0.97 to 0.99)0.004Female gender0.53 (0.33 to 0.86)0.010Ischaemic cardiomyopathy3.08 (1.38 to 6.91)0.006Arterial hypertension1.94 (1.21 to 3.09)0.006Residual shunt2.83 (1.55 to 5.16)0.001Statin1.22 (1.07 to 1.38)0.002eGFR, estimated glomerular filtration rate; PFO, patent foramen ovale. Multivariable Cox regression for any arterial embolism following PFO closure eGFR, estimated glomerular filtration rate; PFO, patent foramen ovale. Discussion In this large, single-centre, retrospective cohort study with a median follow-up of 11.7 years, new-onset AF occurred in 11.2% of patients after transcatheter PFO closure, corresponding to an incidence rate of 1.0 per 100 patient-years, with an early peak within the first 3 months following the procedure. Older age at the time of closure, larger occluder size, and the presence of moderate to severe valvular regurgitation were independently associated with new-onset AF. Left ventricular dilation was the only variable associated with new-onset AF occurring within the first 3 months. Arterial embolic events, including recurrent ischaemic stroke or TIA, occurred during follow-up at an incidence rate of 0.8 per 100 patient-years. Besides conventional cardiovascular risk factors, such as arterial hypertension, the documentation of residual shunting after closure was associated with a higher incidence of arterial embolisms. Reported incidences of new-onset AF after PFO closure vary substantially across randomised trials, registries, and meta-analyses, largely depending on follow-up duration and rhythm monitoring intensity. In a comprehensive meta-analysis including 45 studies and over 17 000 patients, Oliva et al9 reported a pooled AF incidence rate of 1.3 per 100 patient-years (95% CI 1.0 to 1.8) with most events occurring early after closure. Randomised trials of PFO closure for secondary stroke prevention demonstrated an increased incidence of AF in closure arms compared with medical therapy. In the CLOSE trial, Mas et al4 reported new-onset AF or atrial flutter in 4.6% of patients in the PFO closure group versus 0.9% in the antiplatelet-only group (p=0.02), with 90.9% of events occurring within 1 month. Similarly, in the REDUCE trial, Søndergaard et al6 reported AF or atrial flutter in 6.6% of patients undergoing PFO closure versus 0.4% in the antiplatelet-only group (p<0.001), with 83% detected within 45 days. However, in the RESPECT trial, Saver et al5 observed seven periprocedural AF events, all resolving before discharge, and no significant difference in AF incidence beyond the periprocedural period between closure and medical therapy (0.48 vs 0.34 per 100 patient-years; HR 1.47, 95% CI 0.64 to 3.37). Of note, these randomised trials4–6 reported residual shunting in 5.5%–7.0% and recurrent ischaemic stroke or TIA rates in the range of 0.4–0.6 per 100 patient-years despite closure. These findings indicate that embolic risk is reduced but not abolished and that occult or subclinical AF may contribute to residual events during follow-up with underdiagnosed atherosclerosis as a possible contributor. The association between residual shunting and arterial embolism after PFO closure has been well described in both retrospective and prospective studies.19–21 The challenge in interpreting these rates lies in the absence of a control population. First, the higher prevalence of new-onset AF in our cohort in comparison to previous trials is likely explained by the relatively higher age of patients at PFO closure, longer follow-up time, close clinical follow-up and better representation of risk in a real-world setting. Second, population-based cohorts report substantially lower age-specific AF incidence rates, estimated at approximately 0.11 per 100 patient-years in individuals aged 55–59 years in the Rotterdam Study22 and 0.09 per 100 patient-years in adults aged 40–49 years in the MORGAM Consortium,23 indicating that AF incidence after PFO closure exceeds expected background rates. To assess the absolute risk of AF added by the closure procedure, a matched control group with PFO, but without closure is required. The nationwide Danish cohort study14 included 817 patients undergoing PFO closure, 1224 patients with a PFO diagnosis without closure and 8170 matched population controls. They reported a 5-year risk of AF of 7.8% (95% CI 5.5 to 10.0) in the closure cohort, compared with 3.1% (95% CI 2.0 to 4.2) in the PFO diagnosis cohort and 1.2% (95% CI 0.8 to 1.6) in the matched population cohort. The excess AF risk after PFO closure was almost entirely driven by early events, with 78% of AF diagnoses occurring within the first 3 months, corresponding to an adjusted HR of 2.3 (95% CI 1.3 to 4.0) versus the PFO diagnosis cohort and 51 (95% CI 21 to 125) versus the matched population cohort during this early period. Beyond 3 months, AF risk was no longer increased compared with the PFO diagnosis cohort (adjusted HR 0.7, 95% CI 0.3 to 1.7), while a modestly elevated risk persisted relative to the general population (adjusted HR 2.5, 95% CI 1.2 to 5.0). Importantly, AF ascertainment in this registry relied on clinically recognised AF diagnoses. In contrast, a recent prospective cohort using implantable cardiac monitors both before and after PFO closure detected AF in 27.6% of patients at 1 year, with 19.3% occurring within 30 days, despite exclusion of pre-existing AF through a median of 8.6 months of preprocedural monitoring.24 Notably, only 5 of 32 patients (15.6%) with AF were symptomatic, underscoring that the majority of post-closure AF episodes are clinically silent and likely underdetected by routine follow-up strategies, such as 24-hour Holters. Given the inherent connection between the presence of a PFO and the increased risk for AF, patients with a PFO might have a higher risk for subclinical AF in the setting of a thromboembolic event, underscoring the need for thorough and prolonged rhythm monitoring in this population. The mechanisms underlying AF after PFO closure are likely multifactorial and temporally distinct, involving an interaction between procedural triggers and pre-existing atrial vulnerability. So far, the evidence supports a higher intrinsic susceptibility to AF in patients with PFO, where transcatheter closure may act as an additional trigger that predominantly precipitates arrhythmia in the early postprocedural phase.9 14 However, longer-term AF likely reflects progression of an underlying atrial substrate rather than a transient procedural effect. Early postprocedural AF has been attributed to acute atrial stretch, localised septal irritation and inflammatory responses induced by device implantation, which may transiently increase AF vulnerability. Experimental and clinical studies of septal interventions and intracardiac device implantation have demonstrated acute electrical instability related to mechano-electrical coupling, autonomic imbalance and localised atrial injury, supporting these mechanisms as plausible contributors to early AF occurrence.11 However, in our cohort we failed to identify more than one clinical predictor of early AF after PFO closure. Importantly, this observation should be interpreted in the context of outcome ascertainment based on routine clinical follow-up rather than systematic rhythm monitoring. This implies that left ventricular dilation was associated with clinically relevant AF rather than the full spectrum of subclinical arrhythmia. Consequently, the absence of strong early predictors does not exclude the presence of transient, mechanistically driven AF episodes that remain clinically silent. In contrast, late-onset or recurrent AF appears less readily explained by transient procedural effects alone. The association between larger occluder size and AF, in the absence of an association with early AF, suggests a potential role for longer-term mechanical interaction with the atrial septum, altered atrial geometry or progressive atrial remodelling. In addition to mechanical considerations, device size selection is influenced by septal anatomy features which may reflect a more advanced or vulnerable atrial phenotype, including larger PFO dimensions, longer tunnel length, atrial septal aneurysm or increased septal mobility.25 Early-onset AF might be an early sign of this vulnerability, since 30.7% of patients in our population with an early episode of AF have a recurrent episode beyond 3 months after the procedure. The association with moderate to severe valvular regurgitation in our cohort further supports a substrate-based mechanism. Chronic volume overload of the atria may result in elevated left atrial pressures, promoting atrial dilation and structural remodelling. In addition, aortic regurgitation is frequently accompanied by aortic root dilation, which may alter atrial septal geometry and potentially increase mechanical interaction or stretch at the site of the PFO occluder, thereby facilitating arrhythmogenesis.26 Furthermore, the higher prevalence of AF in patients with PFO without closure compared with matched controls raises the hypothesis that PFO may represent a marker of an underlying atrial cardiomyopathy rather than a purely incidental anatomical finding.14 This concept is further supported by emerging genetic and developmental data suggesting shared pathways influencing atrial septal formation and atrial electrophysiological properties.27 While definitive genotype–phenotype correlations between PFO and AF susceptibility have not yet been established, genome-wide association studies have identified loci associated with atrial structure and electrical remodelling that may plausibly contribute to both conditions. Addressing these open research gaps will require prospective trials combining systematic rhythm monitoring with advanced structural, functional, electrocardiographic and genetic phenotyping, to better delineate the mechanisms of AF in patients with PFO and to optimise patient selection and long-term outcomes after PFO occlusion. Limitations The retrospective nature of this study presents inherent limitations. Data collection was dependent on the accuracy of data collected at the time of PFO closure and during follow-up. For example, 4.3% of patients eligible for inclusion could not be assessed for the development of AF due to lack of follow-up data. The single-centre design of this study may limit the generalisability of the findings. The presence of preprocedural AF was ruled out using Holter monitoring prior to the procedure in most of the patients. Nevertheless, more stringent rhythm assessment with long-term rhythm monitoring might have detected more arrhythmias before PFO closure. Follow-up differed for included patients, depending on referring cardiologists and patient preference, allowing for heterogeneity in duration and intensity of follow-up. Conclusions This study shows an elevated incidence of new-onset AF following PFO closure in a real-world setting, especially within the first 3 months after the procedure, highlighting the need for close rhythm monitoring. Older age, larger occluder devices and moderate to severe valvular regurgitation are associated with higher risk of new-onset AF after PFO closure. Late onset of new AF after PFO closure might indicate an underlying mechanism that is not fully explained by stretching of the atrial septum or irritation due to the device. Our findings support closer rhythm surveillance during the periprocedural period and the need for further research regarding the intrinsic relationship between PFO presence and susceptibility to atrial arrhythmias. Supplementary material Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This study involves human participants and was approved by the local ethics committee of UZ Leuven (S67305). The need for informed consent was waived due to the retrospective study design. Data availability statement Data are available on reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available on reasonable request.