Catheter ablation for atrial fibrillation is associated with lower incidence of heart failure and death

Abstract

Aims

Catheter ablation for atrial fibrillation (CAF) improves symptoms, but whether CAF improves outcome is less clear. The purpose of this study was to investigate whether CAF is associated with improved outcome in atrial fibrillation (AF) patients with previous direct current (DC) cardioversion.

Methods and results

We performed a nationwide cohort study including all patients who underwent their 1st direct current cardioversion for AF in the period 2003–15 (N = 25 439). End points were all-cause death, cardiovascular death, stroke/thromboembolism, and incident heart failure (HF). Catheter ablation for AF was treated as a time-varying covariate and the association with outcome was assessed using Cox regression. We also constructed a propensity-matched cohort and assessed the association between CAF and outcome. Median follow-up was 5.3 years (inter-quartile range 3.0–8.7 years). A total of 3509 patients (13.8%) underwent CAF during the study period. Following adjustment for age, gender, comorbidities, medications, educational level, household income, and CHA2DS2VASc score, CAF was associated with reduced risks of all-cause death, cardiovascular death, and incident HF [all-cause death: hazard ratio (HR) 0.69, P < 0.001; cardiovascular death: HR 0.68, P = 0.003; incident HF: HR 0.76, P = 0.011]. Catheter ablation for AF was not associated with a reduced risk of stroke/thromboembolism. These results were replicated in a propensity-matched cohort.

Conclusion

In AF patients with a prior DC cardioversion, CAF was associated with a reduced risk of all-cause and cardiovascular death. This may be due to a reduced risk of HF.

Introduction

Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia.¹ Since AF increases the risk of stroke, heart failure (HF), and death, the disease poses a major threat to public health.² Whether the maintenance of sinus rhythm in AF improves outcome has been widely debated. Currently, it is accepted that pharmacological rhythm control offers no benefit over rate control strategies.^3,4 In the AFFIRM trial, pharmacological rhythm control did not improve survival nor did it reduce the risk of stroke.⁵ However, a post-hoc analysis of AFFIRM suggested that the maintenance of sinus rhythm was associated with improved outcome, but that this benefit may have been offset by the antiarrhythmic medication used in the rhythm control arm.⁶ It is well known that several antiarrhythmic medications have serious side effects, and some may even increase mortality.⁷ The irregular ventricular rate and the loss of atrial filling caused by AF has been shown to significantly increase pulmonary capillary wedge pressure to levels observed in diastolic dysfunction.⁸ Raised left ventricular filling pressure is a hallmark of HF.⁹ Furthermore, permanent AF is associated with a significantly higher risk of incident HF when compared to persistent AF, even after adjustment for common risk factors.¹⁰ Hence, it is possible that the maintenance of sinus rhythm may reduce the rate of left ventricular deterioration in AF and lead to improved survival.

Trials have shown that catheter ablation for AF (CAF) is superior to antiarrhythmic therapy in reducing AF recurrence and maintaining sinus rhythm.^11,12 Recently, the CASTLE-AF trial showed that

CAF significantly reduces rates of death and hospitalization in HF patients with AF when compared with medical therapy.¹³ The CABANA trial randomized patients with persistent or paroxysmal AF to either catheter ablation (N = 1108) or drug therapy (N = 1096) for rate and rhythm control and found that CAF did not reduce the incidence of death, disabling stroke, cardiac arrest, or major bleeding.¹⁴ However, the trial experienced a high degree of crossover from the drug arm to the CAF arm (27.5%) and the trial did not assess the incidence of HF. For now, CAF remains recommended in only symptomatic patients with paroxysmal or persistent AF.^3,15 Observational studies have indicated that CAF may improve survival and overall prognosis in AF.^16–19 However, these studies have either been small or have not evaluated all of the major AF-associated complications (death, stroke, and HF). Hence, in daily clinical practice, the question of ablation vs. no ablation may still be up for debate. This study sought to estimate the effect of CAF on long-term rates of the major AF-associated complications in a nationwide cohort of AF patients following their first direct current (DC) cardioversion for AF.

Methods

Data availability and ethics

The authors were granted full access to anonymized raw data from nationwide administrative registries by Statistics Denmark (Central Authority on Danish Statistics). According to Danish law, informed consent and approval by a local ethics committee is not required for registry-based studies.

Data sources

In Denmark, all citizens are assigned a unique personal identification number (PIN). The PIN is used to accurately link information at the individual level both within and between the administrative registers maintained by the comprehensive Danish Civil Registration System.²⁰ The Danish Civil Registration System collects extensive information regarding vital status, healthcare utilization, income status, and educational level. All Danish citizens have equal access to free government-paid healthcare, including both primary and hospital care, irrespective of social status and financial means. In this study, information was collected from five population registries using the PIN: (i) The Danish National Patient Registry,²¹ which holds information on all hospital contacts, procedures, and diagnoses (including outpatient visits) for all Danish citizens since 1977. (ii) The Danish National Prescription Registry, which holds information on all prescriptions filled at Danish pharmacies since 1995. This registry is used by pharmacies so that they may be reimbursed by the government for medications dispensed and is therefore accurate.²² Patients are partially reimbursed for medications purchased. (iii) The Danish National Population Registry,²⁰ from which information regarding date of birth, sex, emigration status, household income, and education level was retrieved. (iv) The Danish National Cause of Death Registry,²³ which was used to collect information on vital status and cause of death. An overview of the registries used for the study can be found in the Supplementary Material online.

Study sample

In Denmark, CAF is indicated for symptomatic paroxysmal or persistent AF. It is typically offered to symptomatic patients if efforts to remain in sinus rhythm has been attempted with DC cardioversion and medication. In this study, we attempted to include only patients with an indication for CAF in order to reduce the degree of confounding by indication and healthy user bias. In prior registry-based studies, patients who underwent CAF were matched or compared with other AF patients without regard for atrial fibrillation type or symptom status, and no attempts were made to restrict the AF populations to consider only patients with paroxysmal or persistent AF.^17,19,24,25 Patients with paroxysmal and persistent AF have lower stroke rates, a more favourable prognosis, and are typically younger than patients with permanent AF.²⁶ Hence, if patients who undergo CAF are compared with patients with permanent AF, it is likely that the results will be confounded by indication (healthy user bias), because permanent AF carries a worse prognosis.^10,27 Thus, to identify patients with symptomatic persistent AF, we identified all external DC cardioversions performed for AF in Denmark in the period 1 January 2000 to 1 January 2016 (Figure 1) (procedure code: BFFA0). To be considered, patients were required to survive for at least 30 days following their DC cardioversion and have a diagnosis of AF prior to their DC cardioversion or an AF diagnosis associated with the DC cardioversion (Figure 1). Patients with prevalent valvular disease were excluded (Figure 1). Patients with a history of aortic of mitral valve surgery were excluded (Figure 1). Adding this inclusion criteria to the study made our groups more comparable regarding age and comorbidities, making confounding by indication (healthy user bias) less of an issue. Since CAF for AF was not routinely performed in Denmark before 2003, the inclusion period was restricted to 1 January 2003 to 1 January 2015.

Catheter ablation for atrial fibrillation

In this study, the exposure of interest was CAF for AF. Information on whether a patient underwent CAF for AF during the follow-up period was retrieved from the Danish National Patient Registry (procedure code: BFFB04). This definition of CAF has been previously validated and yielded a sensitivity of 97% for identifying ‘first-ever’ CAF patients.²⁸

Baseline comorbidity and medication

Baseline comorbidities were retrieved from the Danish National Patient Registry, and diagnoses were considered obsolete if the last diagnosis code was recorded more than 5 years prior to inclusion. All diagnoses from the Danish National Patient Register included in the Charlson comorbidity index have been validated with positive predictive values of 96–100%.²⁹ Baseline medications were retrieved from the Danish National Prescription Registry. Medication use was defined as one or more claimed prescriptions in the 6 months prior to inclusion or in the 30 days following inclusion. For all patients, comorbidities and medications were reassessed and updated annually during the follow-up period and included as time-varying covariates in the full cohort analysis. For details, we refer to the Statistics section. Finally, we assessed the proportion of patients who received anticoagulant therapy in the period 6–12 months after their first CAF procedure. In this analysis, patients were considered to receive anticoagulation therapy if they had claimed at least one prescription for a vitamin K antagonist or a novel oral anticoagulant during the period 6–12 months after their first CAF procedure.

Study end points

The end points of this study were all-cause death, cardiovascular death, stroke/thromboembolism, and incident HF. Cardiovascular death was defined as cause of death coded as ICD-10: I00–I99. Information regarding vital status was retrieved from the Danish National Cause of Death Registry. Data on stroke/thromboembolism (coded as ICD-10: DI63, DI64, and DI74) and incident HF (coded as ICD-10: DI50, DI110, DI130, and DI132) were collected from the Danish National Patient Registry. For HF, a 3-month blanking period following the 1st CAF was introduced.¹⁵ In current CAF guidelines, a 3-month blanking period in which recurrence of AF is not considered sign of treatment failure is recommended, since such recurrences may be related to short-term atrial inflammatory changes and autonomic dysregulation following the CAF procedure.^15,30 Patients were followed from time of inclusion until emigration, death, or the end of the follow-up period on 1 January 2016.

Statistics

STATA version 14 and SAS version 9.4 were used for statistical analysis. The Student’s t-test or the Wilcoxon Rank-Sum test was used to compare continuous variables between groups at baseline as appropriate. The χ² test was used to compare categorical variables between groups at baseline. Patients who emigrated from Denmark during the follow-up period were censored from the analysis (n = 94). In this study, patients experienced the exposure of interest (CAF for AF) at some point during the follow-up period. Hence, to avoid survival bias, CAF was treated as a time-varying covariate in time-dependent Cox regression.^31,32 Patients who underwent CAF contributed with time at risk to the non-CAF group up until the date of CAF, at which point they were reclassified to contribute with time at risk to the CAF group for the remainder of the study period. Since some patients received more than one CAF for AF during the study period, we adjusted all models for the number of ablations received during follow-up, included as a time-varying covariate. In the full cohort analyses, all comorbidities and medications from Table 1 were reassessed and updated annually (every 365 days following inclusion into the study) to adjust for changes in comorbidity or medication status. Thus, these variables were also included as time-varying covariates. Finally, to consolidate the strength of our findings, we also constructed a propensity-matched cohort. Patients who underwent CAF were matched 1:1 with patients who did not undergo CAF during the study period on the propensity to undergo CAF within 2 years of follow-up using the STATA package psmatch2.³³ The propensity to undergo CAF within 2 years of follow-up was calculated using a multivariable logistic regression model including all variables from Table 1 assessed at baseline with the addition of inclusion date.

Results

Study cohort

We identified 25 439 patients without prevalent valvular disease or a history of aortic or mitral valve surgery who underwent their first DC cardioversion for AF in the period 1 January 2003 to 1 January 2015 (Figure 1). Follow-up was near-complete (>99%) with a median follow-up time of 5.3 years (inter-quartile range 3.0–8.7 years). During the follow-up period, 3509 patients (13.8%) underwent CAF. The median time from inclusion to CAF was 1.3 years (inter-quartile range 0.6–2.8 years). Of those who underwent CAF during the study period, 41.6% received two ablations, 13% received three ablations, and only 3% received four ablations during the study period. Out of those who underwent CAF, 2697 patients (77%) received anticoagulation therapy in the 6–12 months after their first CAF (36 patients died within 1 year after their first CAF procedure and were excluded from this analysis). Table 1 displays the baseline characteristics of patients who underwent CAF during the study period in comparison to patients who did not undergo CAF. Patients who underwent CAF were younger and less likely to be male (Table 1). They had higher household income and were more likely to have a long education (Table 1). They were less likely to have significant comorbidity and used fewer medications as compared to patients who did not undergo CAF during the study period (Table 1). In the propensity-matched cohort, most baseline covariates were well balanced between patients who underwent CAF during the study period and patients who did not (Table 1).

Death

During the study period, 4943 (20%) died from all causes while 3199 (13%) died from cardiovascular causes. The crude rates of death were higher in the non-CAF group (Supplementary material online, Table S4). In a multivariable time-dependent Cox regression model adjusted for all variables in Table 1 with the addition of inclusion year, calendar date of CAF, and the number of ablations received, CAF was associated with a 31% reduction in the risk of all-cause death and a 32% reduction in the risk of cardiovascular death (P < 0.001 and P = 0.003, respectively) (Figure 2). We obtained similar results when considering only the propensity-matched cohort (Figure 2).

Stroke/thromboembolism

A total of 1722 patients suffered stroke/thromboembolism during the study period. The crude rates of stroke/thromboembolism were higher in the non-CAF group as compared to the CAF group (Supplementary material online, Table S4). In a time-dependent Cox regression model adjusted for all variables in Table 1 with the addition of inclusion year, calendar date of CAF, and the number of ablations received, CAF was not associated with a significantly reduced risk of stroke/thromboembolism (Figure 2). Similarly, there was no association between CAF and the risk of stroke/thromboembolism when considering only the propensity-matched cohort (Figure 2).

Heart failure

During follow-up, 2616 (13%) patients developed incident HF [in these analyses, patients with HF at baseline (N = 4657) were excluded]. The crude incidence rate of HF was higher in the non-CAF group (Supplementary material online, Table S4). In a time-dependent Cox regression model adjusted for all variables in Table 1 with the addition of inclusion year, calendar date of CAF, and the number of ablations received, CAF was associated with an ∼24% reduced risk of incident HF (Figure 2). These results were replicated in the propensity-matched cohort (Figure 2). In the full cohort, we also assessed the association between CAF and mortality in patients free from HF at baseline. In patients without HF at baseline, following adjustment for all variables in Table 1 with the addition of inclusion year, calendar date of CAF, and the number of ablations received, CAF was associated with a significantly reduced risk of both all-cause and cardiovascular death [all-cause death: hazard ratio (HR) 0.74, 95% confidence interval (CI) 0.59–0.92, P = 0.007] (cardiovascular death: HR 0.73, 95% CI 0.55–0.99, P = 0.042). These results were replicated in the propensity-matched cohort (all-cause death: HR 0.68, 95% CI 0.50–0.94, P = 0.019) (cardiovascular death: HR 0.59, 95% CI 0.38–0.91, P = 0.018).

Discussion

Our study suggests that CAF may improve survival in patients with persistent AF. This beneficial effect of CAF on survival may be due to a reduced risk of HF.

Catheter ablation for atrial fibrillation and survival

Approximately 15 years ago, the AFFIRM trial demonstrated that pharmacologic rhythm control is not superior to rate control in improving survival in AF.⁵ However, it is well known that antiarrhythmic drugs carry significant risks of side effects, and some may even increase mortality.⁷ Hence, it has been suggested that the adverse effects of pharmacologic rhythm control may offset the benefits of sinus rhythm.⁷ This notion is supported by a post-hoc analysis of the AFFIRM trial, in which sinus rhythm was independently associated with reduced cardiovascular mortality, while the use of antiarrhythmic drugs was associated with a higher risk of death when the analysis was adjusted for sinus rhythm as a time-varying covariate.⁶ Likewise, the ATHENA trial showed that dronedarone reduced the incidence of hospitalization and death from cardiovascular causes.³⁴ In our study, CAF was significantly associated with reduced rates of all-cause and cardiovascular death, and it was also associated with lower rates of incident HF. This poses the question: if sinus rhythm can be maintained without significant adverse or toxic effects (such as those associated with antiarrhythmic therapy), can it then improve survival and overall prognosis in AF? The results of our study, and those of several other observational studies, suggest that CAF may in fact improve survival in AF.^16–19,35 For instance, in an observational study of 1171 consecutive patients with symptomatic AF who underwent CAF or rhythm control with antiarrhythmic medication, Pappone et al.¹⁶ found that CAF was associated with an ∼50% reduced risk of all-cause death. Likewise, in another observational study of 3058 patients with paroxysmal or persistent AF, Ghanbari et al.¹⁸ reported a significant association between CAF and a reduced risk of cardiovascular death. Although most studies have suggested a beneficial effect of CAF on survival, in a study of 846 CAF patients matched to 11 324 AF patients using nationwide health data, Chang et al.²⁴ found no association between CAF and a reduced risk of death after adjustment for confounding. However, Chang et al. used nationwide data from Taiwan, and the patients who underwent CAF were significantly younger (mean age 51 years vs. 56 years in our study), making direct comparison with our results difficult. Overall, our results are in accordance with previous observational studies suggesting a beneficial effect of CAF on survival and prognosis in AF.

Catheter ablation for atrial fibrillation and stroke

With AF, much emphasis is put on the risk of stroke. It is well known that stroke risk in AF is highly dependent on anticoagulation therapy. It has previously been shown that, depending on risk profile, 50–70% of CAF patients in Denmark remain on anticoagulation therapy at 1 year following CAF.²⁵ Accordingly, in our study, 77% of CAF patients received anticoagulation therapy during the 6–12 months following their first CAF. This is not surprising, since current AF guidelines recommend to follow standard anticoagulation guidelines regardless of maintenance of sinus rhythm following CAF.³ Hence, this may explain why we found no association between CAF and a reduced or increased risk of stroke in our study, since the large majority of CAF patients remained on anticoagulation therapy. It follows that the association between CAF and a reduced risk of cardiovascular death found in this study cannot stem from a reduced risk of stroke. Our results are similar to results reported by Pappone et al.¹⁶ and Ghanbari et al.¹⁸: they both studied patients with symptomatic paroxysmal or persistent AF, both studies found an association with a reduced risk of cardiovascular death and no significant association between CAF and a reduced risk of stroke. Importantly, these results are in conflict with several other register-based studies which all report an association with a reduced risk of stroke following CAF,^17,19,24 including a registry-based study from Denmark.²⁵ However, these studies all differ from ours in one important aspect: in every study, patients who underwent CAF were matched or compared to other AF patients without regard for atrial fibrillation type or symptom status. No attempts were made to restrict the AF populations to consider only patients with paroxysmal or persistent AF.^17,19,24,25 In our study, however, we considered only patients who had undergone their first DC cardioversion for AF. As a result, it is likely that we included primarily patients with symptomatic persistent AF for whom CAF may be indicated.^3,15 This is important, because patients with paroxysmal and persistent AF have lower stroke rates and a more favourable prognosis when compared to patients with permanent AF.²⁶ Hence, if patients who undergo CAF are compared with patients with permanent AF, it is likely that the results will be confounded by indication (healthy user bias), because permanent AF carries a worse prognosis.¹⁰ Even though confounding was addressed through statistical adjustment or matching in each of the studies,^17,19,24,25 it is questionable whether such procedures were truly capable of creating comparable groups if no information about atrial fibrillation type (permanent or persistent/paroxysmal) was available, and therefore residual confounding remains a concern. These differences make direct comparisons with our results difficult. Finally, another potential explanation for the lack of association between CAF and a reduced risk of stroke observed in this study may be the notion of AF-associated atrial cardiomyopathy.³⁶ AF is associated with left atrial structural remodelling, fibrosis, endothelial dysfunction, and inflammation which all contribute to a pro-thrombotic state.^37–39 Interestingly, it has been suggested that it may be the underlying atrial cardiomyopathy found in patients with AF and not necessarily the AF rhythm itself which facilities the majority of stroke risk observed in this patient population.^37–39 Although more research is needed before any conclusions can be drawn, this hypothesis is interesting and concurs with the results of our study.

Catheter ablation for atrial fibrillation and heart failure

We found that CAF was significantly associated with a reduced risk of incident HF. It is well known that AF is a strong independent risk factor for HF,⁴⁰ and CAF has been shown to improve outcome in AF patients with HF.¹³ HF develops in up to 40% of patients with AF,⁴¹ and HF is responsible for 15–30% of all deaths in patients with AF.⁴² Hypothetically, since there was no association between CAF and a reduced risk of stroke, the reduced risk of cardiovascular death associated with CAF may stem from a reduced risk of HF. Few studies have assessed the risk of HF following CAF. However, Pappone et al.¹⁶ found that CAF was associated with a significantly reduced risk of adverse cardiovascular events during follow-up, most of which were HF events. Prolonged tachycardia is well known to cause tachycardia-induced cardiomyopathy, and if left untreated, may lead to HF with reduced ejection fraction.⁴³ Consequently, AF can cause HF if ventricular rate is not controlled,⁴⁴ and even though adequate pharmacological rate control is administered, the overall heart rate in AF rhythm is likely to be higher than it would be in sinus rhythm. Hence, the higher heart rate observed in AF patients may directly contribute to an increased risk of HF. Furthermore, there has been evidence to suggest that the irregularity of AF rhythm itself, independently of the ventricular rate response, may be detrimental to cardiac function: invasive hemodynamic data has demonstrated that the irregular ventricular rate and the loss of atrial filling caused by AF significantly increase pulmonary capillary wedge pressure to levels observed in diastolic dysfunction.⁸ Raised left ventricular filling pressure is a hallmark of HF.⁹ In support of these findings, a recent trial assessing the effect of CAF in patients with systolic HF indicates that CAF may improve survival and reduce hospitalization rates in HF patients.¹³ Given these considerations and the results of our study, it is possible that CAF may reduce the risk of HF.

Catheter ablation for atrial fibrillation and CABANA

The CABANA trial randomized patients with persistent or paroxysmal AF to either CAF (N = 1108) or drug therapy (N = 1096) for rate and rhythm control.¹⁴ Results indicate that CAF may not be superior to drug therapy for reducing the incidence of a composite outcome consisting of death, disabling stroke, serious bleeding, or cardiac arrest at 5 years following inclusion.⁴⁵ Although the investigators did find that CAF significantly reduced the rate of a composite consisting of death or cardiovascular hospitalization,⁴⁵ the results of CABANA appear to be in conflict with those of the present study. However, there are several important differences between CABANA and our study which make direct comparisons difficult: (i) In CABANA, the median age in the ablated arm was 68 years, while in our study the mean age of patients in the ablated group was 58 years. Furthermore, to be considered for inclusion into CABANA, patients had to be ≥65 years or have at least one cardiovascular risk factor. In our study, no such inclusion criteria were imposed. In CABANA, 10.6% of patients in the ablated arm had suffered at least one cerebrovascular accident or transient ischaemic attack prior to inclusion, while this was only the case for 3.5% of ablated patients in our study. Thus, these data suggest that ablated patients in CABANA were older and more comorbid as compared to ablated patients in our study. It is possible that CAF and the maintenance of sinus rhythm may be more beneficial in younger, healthier patients. (ii) CABANA was a single-blinded trial and this may have contributed to the high rate of drug to ablation crossover reported in the study (27.5%).⁴⁵ Although the study was powered to allow for a maximum of 25–30% crossover,⁴⁶ this could have affected the significance level of the results, and a sham-controlled trial could be considered to assess the true effect of CAF on outcome. (iii) In CABANA, both academic and non-academic clinical centres were included, and sites were only required to have at least a 100-patient experience with CAF.⁴⁶ In Denmark, CAF is only performed at highly specialized university hospitals receiving a high volume of patients ensuring a very high level of experience and skill. These differences in operator criteria may also have contributed to the differences in results observed between our studies. Finally, it must be acknowledged that our study was observational and may therefore be still be subject to residual confounding despite the rigorous statistical methods applied in this study. Thus, our study indicates that CAF may improve outcome in AF by reducing the incidence of HF, and that it may be beneficial to also focus on HF prevention and not solely stroke prevention in future trials assessing the effect of CAF for AF.

Limitations

In this study, we did not have access to specific information regarding the type of AF. Hence, in order to restrict our study population to include mainly patients with symptomatic persistent AF for whom CAF is indicated, we restricted our inclusion criteria to consider only AF patients who had undergone their first DC cardioversion for AF. Unfortunately, we were not able to accurately identify patients with paroxysmal AF in the registers. However, although it is unlikely, we cannot exclude the possibility that some patients with permanent AF were included into the study under the present inclusion criteria. Another limitation is the lack of information on whether patients remained in sinus rhythm following CAF. However, a lack of this information would tend to underestimate the benefits of CAF, since patients who underwent CAF but did not benefit from this treatment or had frequent relapses would be included in the CAF group. Also, the time from atrial fibrillation diagnosis until inclusion into the study differed between patients who underwent catheter ablation (CAF) during the study period and patients who did not undergo CAF, even in the propensity-matched cohort. Consequently, we adjusted our full model and our propensity-matched model for this variable to account for any potential confounding. However, we realize that it is not possible to completely rule out that this difference may indicate important differences between the two groups. Furthermore, for the non-death outcomes (stroke/thromboembolism and HF), we used data from the Danish National Patient Registry, which holds all information and diagnoses recorded by all Danish hospitals coded in ICD 10 since 1995.²¹ This means that receiving a diagnosis of stroke/thromboembolism or HF necessitates some type of contact with a hospital. This hospital contact may be directly related to the outcome (for instance admittance due to newly discovered HF or stroke/thromboembolism) or it may be unrelated and the outcome may be discovered through patient history or examination/assessment. This constitutes a limitation of the present study, since we do not know the follow-up regimens of the patients undergoing CAF as compared to the patients who did not undergo CAF. However, this method of follow-up assessment allows for complete follow-up on all patients (in this study, >99%), and the register is widely used for follow-up assessment in epidemiological studies due to its accuracy and high validity.^21,47 Unfortunately, because this was a register-based study, it was not possible to collect information on rhythm status during follow-up (sinus rhythm vs. AF rhythm). Finally, our study was observational and therefore unable to prove causation. Even though we used two different statistical methods to ensure robustness of our results (propensity matching and full cohort analysis with statistical adjustment), it is possible that at least part of our results may be explained by confounding by indication. Ideally, our results should be replicated in a clinical trial setting, and whether CAF reduces the incidence of HF in AF patients can only be proven using a randomized controlled trial design. Nevertheless, our results suggest that it may be beneficial to also focus on HF prevention and not solely on stroke prevention in future trials assessing the effect of CAF for AF.

Conclusion

In AF patients with a prior DC cardioversion, CAF was associated with a reduced risk of all-cause and cardiovascular death. This may be due to a reduced risk of HF.

Funding

DM was supported by a scholarship from the Danish Heart Foundation (‘Hjerteforeningen’, Grant: 17-R115-A7498-22057) during the preparation of this manuscript. TB-S was supported by the Fondsbørsvekselerer Henry Hansen og Hustrus Hovedlegat 2016. The sponsors had no role in the study design, data collection, data analysis, data interpretation, or writing of the article.

Conflict of interest: J.H.S. reports grants, personal fees, and other from Medtronic, grants and personal fees from Biotronik, personal fees from Boehringer Ingelheim, grants from Gilead, outside the submitted work. S.D.S. has received research grants from Alnylam, Amgen, AstraZeneca, Bellerophon, BMS, Celladon, Cytokinetics, Gilead, GSK, Ionis, Lone Star Heart, Mesoblast, MyoKardia, NIH/NHLBI, Novartis, Sanofi Pasteur, and Theracos and has consulted for Akros, Alnylam, Amgen, AstraZeneca, Bayer, BMS, Cardior, Corvia, Cytokinetics, Gilead, GSK, Ironwood, Merck, Novartis, Roche, Takeda, Theracos, Quantum Genetics, Cardurion, AoBiome, Janssen, and Cardiac Dimensions. All other authors declared no conflict of interest.

References