Author + information
- Received June 27, 2017
- Revision received August 24, 2017
- Accepted September 7, 2017
- Published online February 19, 2018.
- Melissa Leung, MBBS, BSc(med), MBiostat, PhDa,b,
- Philippe J. van Rosendael, MDa,
- Rachid Abou, MDa,
- Nina Ajmone Marsan, MD, PhDa,
- Dominic Y. Leung, MBBS, PhDb,
- Victoria Delgado, MD, PhDa and
- Jeroen J. Bax, MD, PhDa,∗ ()
- aDepartment of Cardiology, Leiden University Medical Centre, Leiden, the Netherlands
- bDepartment of Cardiology, Ingham Institute at Liverpool Hospital, University of New South Wales, Sydney, Australia
- ↵∗Address for correspondence:
Dr. Jeroen J. Bax, Department of Cardiology, Leiden University Medical Center, Albinusdreef 2, 2333 ZA Leiden, the Netherlands.
Objectives This study sought to investigate the prognostic implications of the clinical subtype of atrial fibrillation (AF): paroxysmal or persistent.
Background Underlying structural abnormalities of the left atrium may be responsible for the initial clinical presentation of AF in either paroxysmal or persistent form, yet the prognostic implications of the clinical subtype on presentation are unknown.
Methods Over a median of 7 years, 1,773 patients (age 64 ± 12 years, 74% males) with nonvalvular AF with index presentations for paroxysmal or persistent AF were followed for the occurrence of all-cause mortality. Clinical information including cardiovascular risk factors, comorbid diseases associated with AF, and CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 [double weight], diabetes, stroke [double weight], vascular disease, age 65–74, and sex category [female]) score was collected and analyzed.
Results In this study, 1,005 patients (57%) had persistent AF. Eighty patients (10%) with paroxysmal AF and 174 patients (17%) with persistent AF died during the follow-up period. Persistent AF compared with paroxysmal AF upon initial AF diagnosis was independently associated with worse survival independent of the CHA2DS2-VASc score and other high-risk cardiovascular risk factors (hazard ratio: 1.24; 95% confidence interval: 1.11 to 1.38).
Conclusions In patients with nonvalvular AF, persistent AF compared with paroxysmal AF upon first diagnosis is independently associated with increased mortality.
Atrial fibrillation (AF) is the leading cause of cardiovascular disease worldwide (1), with a prevalence increasing in both developed and developing countries (2). AF is an independent predictor of all-cause mortality, associated with a 2-fold adjusted increase in death (3). The cardiovascular deaths associated with AF result from progressive heart failure and sudden death (4). A recent large observational study demonstrated a bidirectional relationship between AF and heart failure, with approximately 50% of heart failure patients developing incident AF, and one-third of AF patients experiencing incident heart failure (5). It has remained unclear whether AF is associated with mortality independent of the coexisting conditions with which it is often observed (3). Studies examining the prognostic implications of clinical subtype in a large general population of patients with AF have been limited by inclusion of prevalent AF (6–9), small numbers of AF patients (7,8,10,11), or a lack of clarification of the independent impact of AF on mortality in multivariate or time-dependent analyses (6,8–11). In patients with AF, the use of beta-blockers and statins has been associated with a better prognosis, and the presence of myocardial dysfunction, heart failure, and concomitant chronic kidney disease have been associated with a worse prognosis (12–20). It remains unclear whether the clinical subtype of AF—paroxysmal or persistent—has any impact on mortality. We hypothesized that the clinical subtype of AF may have a differential impact on all-cause mortality.
Patients in this study were identified from a database of all patients who were subsequently referred to our tertiary referral center for electrical cardioversion for AF from June 2005 until January 2015. The first admission (including visits to the emergency department or hospital admission) with AF for all such patients was identified. The AF subtype was then classified, and the clinical details were collected based on chart review from the time of their first admission. AF was diagnosed in accordance with the European Society of Cardiology guidelines for the management of AF (21).
The patients were classified into 2 groups: paroxysmal AF or persistent AF. Paroxysmal AF was defined as self-terminating AF, usually within 48 h, with paroxysms continuing for up to a maximum of 7 days. Persistent AF was defined as recurrent AF that was not self-terminating, with an episode lasting >7 days or requiring termination by either electrical or pharmacological cardioversion. Permanent AF was defined as AF with the accepted rhythm and rhythm control strategies no longer pursued. The patients with persistent or permanent AF were grouped together for the purposes of the present study. For both the paroxysmal and persistent AF groups the patients were required to have had no prior documentation of AF by clinical examination, electrocardiography, or device monitoring. Patients with a history of palpitations before the initial diagnosis of AF, mechanical heart valves, or mitral stenosis were excluded. A total of 1,773 patients were included in the present study and then followed-up with the index AF as time point 0.
Clinical information including demographic data, medications, cardiac risk factors (hypertension, diabetes mellitus, dyslipidemia, smoking history), comorbid medical conditions such as coronary artery, pulmonary, and thyroid disease, and history of heart failure were collected. The CHA2DS2-VASc score (congestive heart failure, hypertension, age ≥75 [double weight], diabetes, stroke [double weight], vascular disease, age 65–74, and sex category [female]) was additionally calculated for each patient based on a point system in which 2 points were assigned for a history of stroke or transient ischemic attack, or age ≥75 years; and 1 point each was assigned for age 65–74 years, a history of hypertension, diabetes, cardiac failure, or vascular disease (myocardial infarction, peripheral artery disease, complex aortic plaque), and female sex (22). Laboratory parameters, such as serum creatinine, total cholesterol, triglycerides, and thyroid hormone levels, were recorded. Left ventricular ejection fraction was measured from two-dimensional echocardiographic images on apical 4- and 2-chamber views using Simpson’s biplane method of discs (23). The Institutional Review Board approved this retrospective analysis of clinically acquired data and waived the need for written informed consent.
Clinical endpoint and follow-up
Patients were followed by electronic chart review for the clinical endpoint of all-cause mortality. All-cause mortality was defined as death from any cause. Deceased patients were identified from the hospital medical records (EPD-Vision and EZIS, Leiden University Medical Centre, Leiden, the Netherlands). This information system is linked to the national death registry and updated on a monthly basis.
Continuous variables are presented as mean ± SD for normally distributed variables, and median (interquartile difference) for non-Gaussian variables. Continuous variables were compared with the unpaired Student’s t-test, and Wilcoxon rank sum test, as appropriate. Categorical data are summarized as frequencies and percentages, and are compared using the chi-square test. Kaplan-Meier survival curves were constructed for the primary endpoint of all-cause mortality, with log-rank testing for statistical significance between strata. Multivariable Cox regression analysis was performed to identify the factors independently associated with all-cause mortality. Hazard ratios (HRs) and 95% confidence intervals (CIs) are provided.
Clinical and echocardiographic parameters were chosen a priori based on biological plausibility and published studies (5,12–14,24). Case elimination was used for missing data, whereby a patient was excluded from analyses when that variable was used if data were missing for a particular variable. In addition, propensity analysis was performed creating two groups of patients matched for age, sex, body mass index, CHA2DS2-VASc score, and left ventricular ejection fraction (Online Appendix, Table 1). A 2-tailed p < 0.05 was considered statistically significant. Statistical analyses were performed using STATA version 12 (STATACorp, College Station, Texas).
A total of 1,773 patients with nonvalvular AF were followed over a median of 7.0 years (interquartile difference 7.1 years). The mean age of the population was 64 ± 12 years, with 1,312 males (74%).
There were 1,005 patients (57%) in the study population with persistent AF. Patients with persistent AF were older, had a larger body mass index, a higher prevalence of hypertension, moderate or severe left-sided valvular heart disease and pulmonary disease, worse left ventricular ejection fraction, and greater use of angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers and beta-blockers compared with the patients who had paroxysmal AF (Table 1). In addition, patients with persistent AF had a higher median CHA2DS2-VASc score and greater use of anticoagulants compared with the patients who had paroxysmal AF. No other differences were observed between the patients with persistent AF and those with paroxysmal AF. The requirement for subsequent electrical cardioversion or radiofrequency catheter ablation were not different between the patients with paroxysmal compared with persistent AF, respectively.
A total of 254 patients (14%) met the primary endpoint of all-cause mortality. The baseline clinical characteristics of the patients according to mortality status at the end of the follow-up period are presented in Table 2. The patients who died were older, with a higher prevalence of smoking, dyslipidemia, and diabetes mellitus compared with the patients who survived. The prevalence of hyperthyroidism was similar between the two groups; however, the rates of pulmonary disease, coronary artery disease, heart failure, stroke, significant left-sided valvular heart disease, and implantable cardiac defibrillator use were higher in those who died. However, the patients who died had higher use of anticoagulants, angiotensin-converting enzyme inhibitors, or angiotensin II receptor blockers, amiodarone, mineralocorticoid receptor antagonists, and statins, and lesser use of flecainide compared with their counterparts. The use of other antiarrhythmic drugs was similar. Furthermore, use of diuretics was higher in the patients who died, reflecting the higher prevalence of heart failure in this group. In regard to laboratory parameters, serum creatinine levels were higher while the hemoglobin and total cholesterol levels were lower in those who died.
Comparison between AF subtypes
Significantly more patients with persistent AF than paroxysmal AF died during the follow-up period (17% vs. 10%, respectively; p < 0.001). To estimate the excess mortality attributable to AF clinical subtype, we performed multivariable Cox regression analyses with factors known to influence mortality in patients with AF, including the CHA2DS2-VASc score, incorporated as covariates in the model (Table 3). The presence of persistent AF compared with paroxysmal AF upon first AF diagnosis was independently associated with worse survival independent of the CHA2DS2-VASc score (HR: 1.24; 95% CI: 1.11 to 1.38).
When we evaluated the association between the type of AF and all-cause mortality in a propensity analysis, the patients with persistent AF had higher cumulative event rates as compared with the patients with paroxysmal AF (Figure 1). On multivariable analysis, the type of AF was still independently associated with all-cause mortality (persistent AF vs. paroxysmal AF; HR: 2.19; 95% CI: 1.072 to 4.476; p = 0.032) (Online Appendix, Table 2).
A secondary analysis explored the differential impact of persistent compared with permanent AF on survival, excluding the patients with permanent AF (n = 9), in view of a previous finding that permanent AF was associated with higher mortality than paroxysmal AF (25). Persistent AF compared with paroxysmal AF remained an independent correlate of worse survival in a multivariable model controlling for the same covariates (HR: 1.54; 95% CI: 1.13 to 2.08; p = 0.006) (Online Appendix, Table 3).
The present study demonstrated that the prognostic value of AF clinical subtype—persistent compared with paroxysmal AF—in a large population with nonvalvular AF upon first diagnosis was independently associated with worse survival, after adjusting for factors known to influence mortality.
AF clinical subtype and prognostic indicators
The value of AF subtype for predicting cerebrovascular events has previously been shown in multiple studies of patients with implantable cardiac devices (26–30). Recently, in a small study of patients with AF undergoing transcatheter aortic valve implantation, the presence of persistent AF was associated with a higher cumulative risk of stroke or death at 2 years compared with the presence of paroxysmal AF or sinus rhythm (31). In a Swedish registry of 155,071 patients with acute myocardial infarction, the composite outcome of all-cause mortality, myocardial infarction, or ischemic stroke at 90-days was compared in four subtypes of AF (new onset AF with sinus rhythm at discharge, new onset AF with AF at discharge, paroxysmal AF, and chronic AF) with the patients in sinus rhythm (32). The investigators demonstrated that all subtypes of AF were associated with a higher risk of the composite endpoint when these patients were compared with the patients in sinus rhythm, but there were no major differences between the AF subtypes.
Previous studies examining the prognostic implications of AF subtypes (paroxysmal versus persistent) have mainly focused on stroke and other adverse cardiovascular events. The limited studies that have examined the relationship between AF clinical subtype and mortality have differed in their inclusion of prevalent rather than incident AF (6–9), their small numbers of AF patients (7,8,10,11), their exclusion of patients with AF at baseline, or their lack of adjustment for known high-risk parameters such as the CHA2DS2-VASc score in multivariate or time-dependent analyses (6,8–11). Our present study demonstrates a relationship of AF subtype with increased mortality in patients with incident AF after adjustment for known high-risk parameters.
The clinical scores CHA2DS2-VASc and CHADS2 (congestive heart failure, hypertension, age ≥75, diabetes, stroke [double weight]) have been widely used in stroke risk stratification. More recently, there has been growing interest in examining the prognostic value of the CHA2DS2-VASc score in patients without AF. MOST (Mode Selection Trial) examined the predictive value of CHA2DS2-VASc and CHADS2 scores for the individual and composite outcomes of death or stroke in 2,010 patients with pacemakers for sinus node dysfunction (33). Each 1 point increase in the CHA2DS2-VASc or CHADS2 scores was associated with an increase in the HR for the combined endpoint of death or stroke, and was independent of a previous history of AF.
A proposed explanation for the relationship of CHA2DS2-VASc and CHADS2 scores with cardiovascular events and mortality independent of AF is that the individual components of these scores (e.g., diabetes, hypertension, coronary artery disease, and heart failure) constitute risk markers for vascular events that include, but are not limited to, stroke. Patients who died in our cohort had higher CHA2DS2-VASc scores and an increased prevalence of diabetes and heart failure. Furthermore, our study demonstrated that the diagnosis of persistent as opposed to paroxysmal AF as a worse prognostic marker is independent of the risk measured by the CHA2DS2-VASc score.
The demonstration of higher mortality in patients with persistent AF compared with paroxysmal AF at the onset of diagnosis is supported by the findings that patients with persistent AF showed distinct electrophysiological properties and AF drivers indicative of diffuse biatrial substrate disease (34). The subsequent management of our patients was not controlled, yet we demonstrated the independent prognostic value of the AF subtype at first presentation. Together with the similar requirements for subsequent electrical cardioversion and radiofrequency catheter ablation, this may suggest that the prognostic implications of persistent versus paroxysmal AF may be independent of any subsequent management of AF.
Our study was performed retrospectively and was designed to examine whether the clinical subtype of AF upon initial diagnosis—as distinct from the lifetime burden of AF after diagnosis—bears any significance for prognosis. In view of this, no monitoring of patients by Holter or device (loop recorder, pacemaker, or implantable cardiac defibrillator) was performed after the initial diagnosis of AF. The impact of the latter, in particular with frequent, recurrent, short-lived paroxysms, on mortality would also be of interest in future prospectively designed studies. However, in practice, most patients with AF do not receive electronic device monitoring, so the burden of disease is largely determined by patient history and clinical presentation. Our study explores this from a practical perspective.
Clinical data were collected from the patient’s initial presentation with AF to our institution, not necessarily during the patient’s admission for cardioversion. Although these patients eventually were referred for electrical cardioversion, it may not have occurred during their initial presentation; the patients may have spontaneously reverted before having the procedure performed. Hence, our patient population is representative of a general population of AF patients presenting at the hospital, with the exception of AF patients who did not require cardioversion during the follow-up period.
In a large population of patients with nonvalvular AF, the presence of persistent AF compared with paroxysmal AF upon first diagnosis was independently associated with increased mortality.
COMPETENCY IN MEDICAL KNOWLEDGE: Atrial fibrillation is an independent predictor of all-cause mortality. The cardiovascular deaths that AF is associated with result from progressive heart failure and sudden death.
TRANSLATIONAL OUTLOOK: In patients with nonvalvular AF, persistent AF compared with paroxysmal AF upon first diagnosis is independently associated with increased mortality and is incremental to the risk measured by the CHA2DS2-VASc score. Together with the similar requirement for subsequent electrical cardioversion and radiofrequency catheter ablation in these patients, this may suggest that the prognostic implications of persistent versus paroxysmal AF may be independent of any subsequent management of AF.
The Department of Cardiology at Leiden University Medical Center has received unrestricted research grants from Biotronik, Boston Scientific, Medtronic, and Edwards Lifesciences. Dr. M. Leung has received a Pfizer Investigator Initiated Research Grant. Dr. Delgado has received speaker fees from Abbott Vascular. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- atrial fibrillation
- congestive heart failure, hypertension, age ≥75, diabetes, stroke (double weight)
- congestive heart failure, hypertension, age ≥75 (double weight), diabetes, stroke (double weight), vascular disease, age 65–74, and sex category (female)
- confidence interval
- hazard ratio
- Received June 27, 2017.
- Revision received August 24, 2017.
- Accepted September 7, 2017.
- 2018 American College of Cardiology Foundation
- Chugh S.S.,
- Havmoeller R.,
- Narayanan K.,
- et al.
- Benjamin E.J.,
- Wolf P.A.,
- D’Agostino R.B.,
- Silbershatz H.,
- Kannel W.B.,
- Levy D.
- Santhanakrishnan R.,
- Wang N.,
- Larson M.G.,
- et al.
- Kitchin A.H.,
- Milne J.S.
- Marijon E.,
- Le Heuzey J.-Y.,
- Connolly S.,
- et al.
- Nielsen P.B.,
- Larsen T.B.,
- Gorst-Rasmussen A.,
- Skjoth F.,
- Lip G.Y.
- Ntaios G.,
- Papavasileiou V.,
- Makaritsis K.,
- et al.
- Alonso A.,
- Lopez F.L.,
- Matsushita K.,
- et al.
- Ananthapanyasut W.,
- Napan S.,
- Rudolph E.H.,
- et al.
- Tanaka A.,
- Inaguma D.,
- Shinjo H.,
- Murata M.,
- Takeda A.
- Winkelmayer W.C.,
- Patrick A.R.,
- Liu J.,
- Brookhart M.A.,
- Setoguchi S.
- Baber U.,
- Howard V.J.,
- Halperin J.L.,
- et al.
- Glotzer T.V.,
- Hellkamp A.S.,
- Zimmerman J.,
- et al.
- Capucci A.,
- Santini M.,
- Padeletti L.,
- et al.
- Glotzer T.V.,
- Daoud E.G.,
- Wyse D.G.,
- et al.
- Boriani G.,
- Glotzer T.V.,
- Santini M.,
- et al.
- Shaul A.A.,
- Kornowski R.,
- Bental T.,
- et al.
- Batra G.,
- Svennblad B.,
- Held C.,
- et al.
- Lim H.S.,
- Denis A.,
- Middeldorp M.E.,
- et al.