An ethical permit from the Swedish Ethical Review Authority has been approved with the registration number: (2022-02255-01).

This retrospective, single-centre, register-based cohort study included patients from the Swedish registry of congenital heart disease (SWEDCON) (19). The SWEDCON register is a National Quality Registry frequently used for research projects due to its high quality and good follow-up data. It consists of subregisters for adult patients with congenital heart defects, children with congenital heart defects, paediatric cardiothoracic surgery, foetal register and a register for cardiomyopathies. Patients who underwent transcatheter PFO closure between March 1st 2017, and June 30th 2020, at the Karolinska University Hospital, Stockholm, were included for analysis. Cryptogenic stroke was the indication for all PFO closures in this study.

In total, 262 patients (98 females, 164 males) were included, of which 261 received a GSO device. Mean age at intervention was 46.3 years (SD ± 10.4), mean BMI 25.5 (SD ± 4.2). Cardiovascular risk factors such as arterial hypertension were noted in 32 patients (12.2%), Diabetes mellitus in 3 patients (1.1%), hypercholesterolemia in 29 patients (11%) and smoking in 11 patients (4%), see Table 1.

All patients underwent TTE and TOE with administration of agitated saline as a diagnostic evaluation after cryptogenic stroke to verify the PFO diagnosis. Before the intervention, all patients were evaluated at a multidisciplinary team conference by an interventional cardiologist, neurologist, clinical physiologist and neuroradiologist, where the decision to proceed with percutaneous PFO closure was made.

All interventions were performed in a monoplane catheterisation laboratory using conscious sedation and local anaesthesia. Femoral venous access was obtained in all patients using a micropuncture technique with ultrasound guidance to minimise puncture-related complications. Two ipsilateral punctures were performed to allow access for the PFO-closure device and ICE catheter for intraprocedural imaging. An 11 Fr introducer was inserted into the femoral vein for device delivery. An ICE catheter (Abbot View Flex, Abbott Vascular, CA, USA) was introduced through a 9 Fr introducer and placed in the right atrium. Right heart catheterisation and the PFO closure were performed according to IFU.

After the procedure, patients were observed in the cardiology ward or daycare facility for 4–6 h, including a bedside TTE, to confirm the position of the device in situ and the absence of pericardial effusion. Patients were kept on strict bed rest for the first hour, followed by mobilisation. The antiplatelet regimen used for most patients was lifelong acetylsalicylic acid (160 mg once daily) combined with clopidogrel (75 mg once daily) for six months. Depending on the patient's comorbidity and preexisting therapy, anticoagulation (e.g., apixaban or coumadin) was continued with life-long acetylsalicylic acid.

Access site closure was performed at the sole discretion of the treating interventionist, using preclosure with either the Perclose ProGlide™ system (Abbott Vascular, Santa Clara, California, USA) for each femoral venous puncture site or a figure-of-eight suture see Figure 1.

Baseline and periprocedural characteristics were retrospectively collected from patient charts. Bleeding or puncture-related complications were analysed following the Bleeding Academic Research Consortium Definition for Bleeding (BARC) criteria (20).

Categorical variables were analysed using logistic regression models and described as proportions (percentages). Continuous variables were analysed using linear regression models and expressed as mean values with standard deviations (SDs). In all analyses, P-values < 0.05 were considered statistically significant. All data obtained were registered in Microsoft Excel® (Microsoft, Redmond, Washington, USA). Analyses were performed using STATA software (version 16.1 Stata Corp., College Station, Texas, USA).

SDD was achieved in 246 (93.9%) patients. Among the patients with delayed discharge (i.e., no SDD), the following complications were registered: vascular complications (n = 3), late procedure starting time (n = 4), pain in the groin (n = 2), epigastric pain (n = 1), drainage of pericardial effusion (n = 2), post-procedural atrial fibrillation (n = 1), post-procedural desaturation (n = 1), post-procedural pulmonary complications (n = 1), and personal reasons (n = 1). The reasons for post-procedural readmissions and emergency room (ER) visits were atrial fibrillation (arrhythmia n = 16, 6.1%), fever (n = 4, 1.5%) or chest pain/pain at the access site (n = 4, 1.5%). One patient was operated on due to an AV fistula five weeks post-PFO closure.

The Perclose ProGlide™ system was used for access site closure in 166 patients, five (3%) of whom suffered vascular complications. Three had minor venous bleeding after application of the Perclose ProGlide™ and were treated with additional figure-of-eight sutures in the catheterisation laboratory. Technical failure at deployment of the ProGlide™ system occurred in one patient, and the access site was closed with a figure-of-eight suture instead. One patient was readmitted to the hospital due to vascular complications and underwent surgery due to an arteriovenous (AV) fistula five weeks post-PFO closure. No other patient had to be readmitted due to vascular complications. There were two type 3b (cardiac tamponade) bleeding complications and eight type 1 bleeding complications, according to BARC (20). There were no differences in complications between patients who received the ProGlide™ device and those who did not (Table 3).

All patients in this study underwent PFO closure with conscious sedation and general anaesthesia, which can facilitate early discharge. Postoperative nausea and vomiting (PONV) is a common adverse effect of general anaesthesia affecting up to 80% of patients (24) and may prolong hospital stay. The primary risk factor for PONV is the use of volatile anaesthetics. Performing the intervention with conscious sedation and local anaesthesia facilitates SDD by reducing the rate of PONV. An argued advantage of general anaesthesia with intubation is the possibility of performing TOE, but it can prolong procedural duration. Other options are TOE without general anaesthesia and/or with only a brief TOE exam at the end of the procedure (18).

At our institution, ICE and fluoroscopy are the standard protocols for imaging of the atrial septum, enabling procedures without TOE and general anaesthesia.

The perioperative imaging guided by ICE rather than TOE reduces the necessity for general anaesthesia and lessens the risk of associated complications (25, 26). Hence, the use of ICE during all the procedures in this study possibly contributed to the low rate of complications and high rate of SDD. This is supported by Blusztein et al., presenting that ICE and avoiding general anaesthesia were highly and significantly associated with SDD (23). They reported SDD in 382/554 patients (68.9%), with ICE used in 503/554 patients (90.7%), ICE and TOE were used in another three patients, and the remaining patients were treated using fluoroscopy or TOE, concluding that SDD was more common in patients with ICE.

In a retrospective study by Barker et al., SDD was achieved in 456/467 patients (97.6%) (22). Similar to our study, patients underwent PFO closure using local anaesthesia and conscious sedation. Periprocedural imaging in the study by Barker et al. was done using ICE and fluoroscopy in 86/467 patients (18.4%); the remaining 381 patients underwent closure with fluoroscopy alone.

High success rates of SDD rates were seen in the study by Barker et al. (22), in our study and in the subgroup of patients with local anaesthesia in the study by Bluzstein et al (23), suggesting that local anaesthesia is associated with SDD.

An extensive overview discussing the advantages and disadvantages of ICE for PFO closure has been published by Egidy Assenza (27). Another approach was described by Achim et al., using fluoroscopy as imaging during the procedure with a short TOE control before device release to assess device positioning and stability. This approach lead to significantly shorter procedural length in the fluoroscopy-guided group with no difference in procedural complications, including death, major bleeding, device dislodgement, stroke or clinically relevant peripheral embolisation between the two groups (p = .99) (28).

Applying a VCD such as the ProGlide™ Perclose system can further increase the possibility and safety of early discharge. Sekhar et al. reported the device to be safe and effective for femoral artery closure after heart catheterisation before early discharge (29). In our study, there was no difference overall in SDD rate between the patients who received the ProGlide™ device and those who did not receive the ProGlide™ device.

Late vascular problems were rarely seen in our study. One patient (0.38%) was diagnosed with an AV fistula one week after the procedure and underwent vascular surgery five weeks after the initial procedure.

Arrhythmia was the most common complication seen after the PFO closures in this study. The most encountered arrhythmias after PFO closure are known to be atrial tachycardias such as supraventricular tachycardia (SVT), atrial fibrillation (AFib) and atrial flutter (Af) (REF). Readmission rates are mainly driven by atrial fibrillation (AFib) requiring treatment, as demonstrated in this study and supported by Abrahamyan et al. (30). There is a low risk of clinically significant atrioventricular disturbances such as complete atrioventricular block (AVB) requiring pacemaker implantation (30). The risk of SVT requiring medical therapy or electrical cardioversion was reported to be around 1.5% in a large study by Szkutnik (31); late AVB requiring pacemaker was seen in 2/739 patients (0.27%) after 4.3 years respectively 1.5 years after Amplatzer device implantation. This reflects the results in our study, where acute onset AF delayed hospital discharge in one patient and caused ER visits in another sixteen patients.

This study is mainly limited by the retrospective and observational design, which is susceptible to confounding. There was no randomisation to either SDD or VCD, which complicates the interpretation of the result.