In total, 2,769 patients were recruited in the ARENA registry. Mean age was 72.5 ± 10.9 years and 62.6% of participants were male (Table 1). The majority of patients were diagnosed with paroxysmal AF (Table 1). A concomitant cardiac condition was present in 76.6% of patients. Hypertension was the most the most common cardiac co-morbidity, followed by coronary artery disease (Table 1). Median left ventricular ejection fraction (LVEF) was preserved in the majority of cases (Table 1). Mean left atrial diameter was 45.0 ± 7.2 mm. Additional cardiovascular risk factors were common, in particular arterial hypertension, diabetes, and renal failure (Table 1). Mean CHA2DS2-VASc score was 3.6 ± 1.7 and mean HAS-BLED score was 2.1 ± 1.1. Prior stroke or transient ischemic attack (TIA) had been diagnosed in 13.7% of patients. Anticoagulation therapy was prescribed in 88.4% of cases at baseline, using NOACs in the majority of cases (66.9%). In 20.0% of participants vitamin K antagonists (VKA) were prescribed. Antiplatelet agents were present in 17.0% of patients at recruitment, and in 11.0% as a combination therapy in either a dual combination or triple combination scheme with OAC therapy. Rivaroxaban (26.3%) and apixaban (25.1%) were the most frequently employed NOACs at baseline in the overall cohort, followed by dabigatran (11.9%) and edoxaban (3.7%). Beta-blockers were prescribed in 75.4% of cases, and digitalis in 11.3%. A rhythm control strategy with cardioversion, antiarrhythmic medical therapy, or catheter ablation was chosen or had been attempted in 1016 (42.7%) patients, whereas rate control was documented as the primary strategy in 1339 (56.3%) patients. Patients in whom a rhythm control concept was followed were younger (71.5 ± 9.6 years vs. 75.5 ± 9.2 years, P < 0.001), less often affected by some cardiac co-morbidities (coronary artery disease: 37.6% vs. 45.8%, P < 0.001; dilated cardiomyopathy: 11.0% vs. 8.1%, P = 0.015), and less often had a history of stroke or TIA (11.0% vs. 17.4%, P < 0.001). We cannot exclude that some patients in whom rhythm control therapies were documented had later been switched to rate control, e.g., due to previous unsuccessful rhythm control attempts. However, patient characteristics in the subgroups of rate or rhythm control are consistent with those of patients predominantly assigned to the respective therapeutic strategy in everyday clinical practice. Pharmacological antiarrhythmic therapy was present in 3.6% of patients with class I antiarrhythmic drugs and in 6.5% with class III antiarrhythmic agents. Nearly one in five patients had undergone catheter ablation of AF and nearly a third of the cohort had undergone cardioversion (Table 1). The majority of patients described absent (43.5% EHRA I) or mild (43.1% EHRA II) AF-associated symptoms. Shortness of breath during everyday activity was present in the majority of cases (36.5% NYHA II; 26.2% NYHA III/IV). The quality of life reflected in the EQ-5D-5L score was 0.89 [Q1: 0.77; Q3: 0.97]. Subjectively perceived health status was mildly reduced according to VAS score (70 [Q1: 50; Q3: 80]) at recruitment.

Patients recruited during the intervention period were older than patients in the control period and were more often affected by cardiac co-morbidities (Table 1). The most structural cardiac conditions conditions were coronary artery disease and heart failure. However, mean left ventricular systolic function was not statistically different between groups in cases with documented echocardiography at baseline (Table 1). Sex distribution was similar in both groups. More patients in the intervention period were diagnosed with paroxysmal AF which was the most common type of AF in both groups. First diagnosis of AF was more common in the control period (Table 1). There was no statistically significant difference in common cardiac risk factors such as smoking, hypertension, diabetes mellitus, or renal failure. However, median glomerular filtration rate was slightly lower in the intervention group (Table 1). Additionally, more patients in the intervention group had a history of cerebral thromboembolic complications (Table 1). The mean CHA2DS2-VASc and HAS-BLED scores in the intervention group were higher (Table 1). A CHA2DS2-VASc score of ≥ 2 was recorded in 89.5% of patients in the intervention group and 83.9% of patients in the control group (P < 0.001). AF-related symptoms were more pronounced in the control group (Table 1). The quality of life at baseline as reflected in the EQ-5D-5L-index 2 weeks after recruitment was not different between the two groups (control group: 0.89 [Q1 = 0.77; Q3 = 0.96]; intervention group: 0.89 [Q1 = 0.78; Q3 = 0.97]). However, with respect to specific interrogation dedicated to anxiety or cardiophobia, more patients in the control period stated that they perceived nervousness, fear of acute cardiac events, and avoided activity leading to increase in heart rate, whereas patients in the intervention group were less often anxious with respect to heart-related symptoms or physical activity (Supplemental Table 1). More patients in the intervention phase had undergone medical or pharmacological cardioversion or device implantation, whereas there was no significant difference in the rates of previous AF ablation (Table 1). At baseline, there was no statistically significant difference in oral anticoagulation rates (Table 2 and Fig. 1). More patients received NOACs in the intervention group (Table 2 and Fig. 1). Apixaban was prescribed more often in the intervention group (control group: 22.9%, intervention group: 27.7%, P = 0.004, Fig. 1). Interventional therapies for stroke prevention by LAA occluder were not statistically different between groups (control period: 0.6%, intervention period: 0.7%, P = 0.62) at baseline. Concomitant antiplatelet therapy was prescribed in a minority of patients in both groups (Table 2). At baseline, beta-blockers and antiarrhythmic medication were more often prescribed in the control group (Table 2).

Oral anticoagulation in the control vs. intervention group. NOACs constituted the preferred agent for OAC therapy in both subgroups with a decline in use of vitamin K antagonists over the course of the observation period. Other therapies for stroke prevention include subcutaneous heparin administration, which was applied in a minority of cases. Rates of OAC prescription did not differ significantly between the subgroups at baseline (P = 0.63). At follow-up, there was a higher adherence to OAC-therapy in the intervention group in comparison to the control group. VKA vitamin k antagonist ** = P < 0.01

Follow-up data was procured until at a median of 665 days [Q1: 568 days; Q3: 1041 days] after recruitment in 93.9% of baseline survivors. In 16.5%, only information on vital status was available from local registration offices. In the control group, follow-up data on 1282 patients (93.7%) were available after 706 days [Q1: 516 days; Q3:1176 days] and in 1266 (94.1%) participants in the intervention group with a median follow-up duration of 630 days [Q1: 605 days; Q3: 995 days]. Apart from vital status, further information on patient status and clinical events was available in 61.9% of patients of the overall cohort. In patients lost to follow-up, reasons for unavailability of further clinical end points included death (31.8%), withdrawal of consent for further follow-up interrogations (24.6%), failure to contact the patient and information only available from the local registration office (23.5%), information only available from relatives of the patients (3.4%), or other personal reasons for non-response (16.8%) without significant difference between the groups (loss to follow-up for additional end points control group: 38.3%; intervention group: 37.9%, P = 0.84).

Estimated 1-year-mortality rate was 7.6% (CI 6.3–9.2%) in the control group and 7.2% (CI 5.9–8.8%) in the intervention group (P = 0.68, Fig. 2). Kaplan–Meier estimates of combined end points of death, myocardial infraction, stroke (MACCE) (control group: 8.2% [CI 6.8–9.8%]; intervention group 7.9% [CI 6.6–9.5%], P = 0.81) and death, myocardial infarction, stroke, and major bleeding (control group: 8.5% [CI 7.1–10.2%]; intervention group 8.5% [CI 7.1–10.1%], P = 0.92) were not statistically different between groups.

One-year survival in the control vs. intervention group. Kaplan–Meier analysis of all-cause mortality. There was no statistically significant difference between the two groups

At follow up, there was no significant difference in stroke, transient ischemic attack, or myocardial infarction between groups (Fig. 3A). Severe bleeding complications were rare events and showed no statistically significant difference between subgroups (Fig. 3A). More moderate bleeding complications were recorded in the intervention group (control group: 2.9%, intervention group: 5.2%; P = 0.021). Re-hospitalization rates during follow-up did not differ significantly between the control group and the intervention group (Fig. 3B). However, AF-related re-hospitalization was more common in the control group, whereas non-cardiovascular re-hospitalization rates were elevated in the intervention group (Fig. 3B). As to the number of outpatient medical contacts during the follow-up period, there was no statistical difference between the two groups (control group: 1.8 ± 1.9 visits per quarter of the year; intervention period: 1.8 ± 1.6 visits, P = 0.28).

Clinical follow-up in the control vs. intervention group. A Thrombembolic and bleeding complications at long-term follow-up. Severe adverse events were rare and without statistically significant difference between the two groups. TIA transient ischemic attack. B Hospitalization rates during follow-up. Overall hospitalization rates (left columns) and the respective underlying diagnoses leading to hospitalization are shown. Patients in the control group were more often affected by AF-related hospitalization, whereas patients in the intervention group were more often hospitalized for non-cardiac reasons. AF atrial fibrillation, CV cardiovascular, FU follow-up

In the control group, more cardioversions (control group: 9.1%, intervention group: 5.3%, P = 0.002) and catheter ablations of AF had been performed during follow-up (control group: 12.0%, intervention group: 5.1%, P < 0.001). Regarding the severity of AF-related symptoms, there was no difference between the two groups (EHRA III or IV in the control group: 5.8%, intervention group: 4.9%, P = 0.46, Fig. 4A). However, in comparison to baseline, an improvement in AF-related symptoms of survivors reflected in EHRA states could be detected in both groups (signed-rank test P < 0.001).

Cardiac symptoms and health-related psychological burden at follow-up. A AF-related symptoms characterized by EHRA states (left columns) and physical capacity classified by NYHA states (right column) in the control vs. intervention group at follow-up. Symptoms showed no significant difference between the two groups. EHRA European Heart Rhythm Association; NYHA New York Heart Association. B Patient-reported cardiac anxiety and nervousness in the control vs. intervention group. In both subgroups, comparable levels of anxiety were reported

About one in ten patients was member in an AF-related self-help group (control group: 13.1%, intervention group: 11.6%, P = 0.39). Quality of life as measured by the EQ-5D-5L-index at long-term follow-up was not different between the two groups at follow-up (control group: 0.91 [Q1 = 078; Q3 = 0.97], intervention group: 0.88 [Q1 = 0.75; Q3 = 0.97], P = 0.086). Specific interviews regarding anxiety and cardiophobia revealed no statistically significant difference between groups at follow-up, with a reduction in AF-related anxiety in comparison to baseline interrogation (Supplemental Table S1, Fig. 4B).

When interviewed regarding the role of the ARENA project with respect to the individual health awareness, 24.6% of patients in the control group and 22.2% of patients in the intervention group reported that the ARENA project had influenced their health perception (P = 0.27).

At long-term follow-up, more patients in the intervention group received oral anticoagulation therapy according to available prescription regimens (Table 2 and Fig. 1). In comparison to the baseline visit, there was a more pronounced decrease in OAC prescription in the control group in comparison to the intervention group (Table 2). In both groups, the use of VKA decreased in comparison to baseline, and NOACs were the preferred choice for OAC therapy (Table 2 and Fig. 1). Prescription rates of antiplatelet agents decreased in both groups (Table 2).

At follow-up, the rates of beta-blocker prescription showed no significant change in both the control group and the intervention group, and were not different between the two groups (Table 2). Also, rates of antiarrhythmic medical therapy were similar to baseline in the control group and intervention group, and not statistically different between the two groups at follow-up (Table 2).

To exclude patients with a temporary indication for OAC at baseline, e.g., after cardioversion with low CHA2DS2-VASc score, we performed a subgroup analysis of patients with only class I or class IIa indication for permanent OAC therapy. Data on anticoagulation regimens were available in 718 patients with permanent OAC indication in the control group and in 710 patients in the intervention group. In this subgroup analysis, the rates of oral anticoagulation were not significantly different between the groups at follow-up (control group: 87.5%, intervention group: 90.1%, P = 0.11). Monotherapy with antiplatelet agent not in combination with any OAC was present in 3.8% of patients in the control group and 4.8% of patients in the intervention group with a class I/IIa indication for OAC (P = 0.36). Rates of severe thromboembolic or bleeding complications in this subgroup analysis during follow-up were comparable to the overall cohort (stroke in the control group: 1.1% vs. intervention group: 1.2%, P = 0.94; TIA in the control group: 1.7% vs. intervention group: 1.5%, P = 0.75; myocardial infarction in the control group: 0.8% vs. intervention group: 1.6%, P = 0.18; severe bleeding complication in the control group: 1.4% vs. intervention group: 2.0%, P = 0.38). In analogy to the overall cohort, re-hospitalization rates were similar between the control and intervention group (control group: 50.6%, intervention group: 50.9%, P = 0.92), with a higher proportion of AF-related re-hospitalizations in the control group (control group: 31.1%, intervention group: 18.7%, P < 0.001).