Mortality in atrial fibrillation. Is it changing?

Abstract

Patients with atrial fibrillation experience higher mortality rates than those without this condition. Recent studies have explored whether mortality rates in atrial fibrillation patients and the overall impact of atrial fibrillation on mortality has changed. Overall, mortality in atrial fibrillation has decreased over the last few decades, with no strong differences between men and women. These improvements could be caused by advances in preventing thromboembolic complications of atrial fibrillation or better management of comorbidities in these patients. Understanding the mechanisms for these changes and developing novel approaches to improve survival in AF patients are areas deserving of future research.

INTRODUCTION

Atrial fibrillation (AF) is a common arrhythmia, affecting over 30 million worldwide [1]. In the US alone, more than 5 million Americans have AF, with a prevalence of 8% in those 65 years of age and older [2, 3]. Between one in three and one in five persons are expected to develop AF during their lifetime [4]. Health care spending associated to the treatment of AF and its complications in the US has been estimated at over $28 billion annually [5].

In addition to being highly prevalent, AF increases risk of adverse outcomes. Compared to individuals without the arrhythmia, AF patients experience higher rates of death, stroke, dementia, heart failure, and other cardiovascular diseases [6]. A landmark publication from the Framingham Heart Study showed that mortality rates were 50–90% higher in participants with AF compared to those without AF [7]. Numerous subsequent publications have confirmed this association [8–11]. The last two decades have brought significant breakthroughs in the treatment of patients with AF, from the demonstrated efficacy of oral anticoagultion to prevent stroke and reduce mortality [12], to the successful use of pulmonary vein isolation for the treatment of selected AF patients [13]. A relevant question is whether these advances have resulted in improvements in the survival of AF patients over this time period.

In this review, we summarize the literature reporting mortality trends in patients with AF, with a special emphasis on publications over the last five years. We start by briefly describing studies that have demonstrated higher mortality rates in patients with AF, follow with a detailed presentation of studies of mortality trends in AF, and finish with a discussion of potential explanations for the observed trends and the implications for clinical practice and future research.

IMPACT OF ATRIAL FIBRILLATION ON MORTALITY

Several community-based prospective studies have shown that individuals diagnosed with AF experience higher mortality rates than those without AF, independently of demographic and clinical factors. The Framingham Heart Study was one of the first to report this association in an analysis including 5,209 participants followed for 40 years. Mortality rates in women and men who developed AF were 90% and 50% higher, respectively, than in those who did not [7]. A similar pattern was observed among 15,406 participants in the Renfrew/Paisley study, with mortality rates 120% and 50% higher in women and men who developed AF compared to those without AF [11].

Comparable associations have been reported in more recent studies. Among 34,722 women participating in the Women’s Health Study, mortality rates were more than double in those who developed AF compared to those who did not [10]. The impact of AF was particularly strong for cardiovascular disease mortality, with a fourfold increase in risk associated with AF. Similarly, the Atherosclerosis Risk in Communities Study, which followed a racially diverse cohort of 15,080 individuals for more than 20 years, reported a doubling in mortality rates in those who were diagnosed with AF, compared to AF-free individuals, with a similar magnitude for Whites and Blacks [9]. Results from selected cohort studies are presented in the figure.

Figure.

Forest plot presenting the association of atrial fibrillation with mortality in selected cohort studies as rate ratios and 95% confidence intervals. ARIC: Atherosclerosis Risk in Communities Study. CI: Confidence interval. RR: rate ratio.

TEMPORAL TRENDS IN ATRIAL FIBRILLATION-ASSOCIATED MORTALITY

Based on the published literature, there is no question that being diagnosed with AF results in increased mortality rates. With advancements in the management of AF and prevention of its thromboembolic complications, it is reasonable to hope that the impact of AF on mortality may have been blunted. Several recent studies have explored this question by examining trends in mortality rates among persons with AF.

The Framingham Heart Study, a community-based prospective cohort, has systematically ascertained and adjudicated diagnoses of AF among its study participants since 1958. Using this information, study investigators have been able to assess mortality rates after AF in the 50-year period between 1958 and 2007 [14]. During these five decades, age-adjusted 5-year mortality in AF decreased from 55% in 1958–1967 to 39% in 1998–2007 for women, and from 53% in 1958–1967 to 37% in 1998–2007 for men. In models adjusted for demographic and clinical risk factors, mortality rates decreased by 25% (95% confidence interval, −3%, 46%) between 1958–1967 and 1998–2007, with a significant trend over time.

In contrast to results from the Framingham Heart Study, an assessment of mortality trends among residents of Olmsted County, Minnesota, between 2000 and 2010 did not report improvements in mortality over that time period [8]. This study identified 3,344 patients with incident AF in the entire community. During a mean follow-up of 4.6 years, 1,615 patients died (48%). Compared to the general population, mortality rates were 19 times higher during the first 30 days after diagnosis, 4 times higher in days 31–90 post diagnosis, and 50% higher in days 91–365. Mortality in AF was not different from the general populations after 1 year. During the study period, mortality remained relatively constant both for the 90-day post-diagnosis period and among 90-day survivors, with adjusted hazard ratios of mortality of 0.96 (95% confidence interval, 0.71, 1.32) and 1.05 (95% confidence interval, 0.85, 1.31) for 2010 versus 2000, respectively. A study period shorter than that for the Framingham Heart Study may be responsible for the inconsistency of results.

Similar to the Olmsted County study, an analysis of electronic health records from patients in the Geisinger Health System, in central and northeast Pennsylvania, for the period 2004–2016 did not identify improvements in survival after AF diagnosis [15]. On the contrary, mortality increased slightly over the study period. Mortality following AF diagnosis was 8% at 3 months, 11% at 6 months, 16% at 1 year, 30% at 3 years, and 41% at 5 years, with annual relative increases in mortality of 2–3% [15]. Temporal changes in the case-mix of AF patients diagnoses in this healthcare system could account for the increased mortality.

A population-based study conducted in Western Australia evaluated mortality after incident AF, identified from hospitalizations, in the period 1995–2010 [16]. Over these 16 years, 57,552 incident AF events occurred. The overall 30-day, 1-year and 3-year mortality after AF diagnosis was 7%, 17% and 27%, respectively. Mortality decreased modestly but significantly over time, from 7% to 6% (30-day mortality), 18% to 15% (1-year mortality) and 29% to 24% (3-year mortality). After multivariable modeling, hazard ratios (95% confidence intervals) of 30-day, 1-year and 3-year mortality comparing the period 2007–2010 to 1995–1998 were 0.91 (0.83, 1.00), 0.86 (0.81, 0.91) and 0.82 (0.78, 0.86), respectively.

Using a similar approach linking countrywide registries, investigators evaluated mortality trends after an AF diagnosis in Denmark for the 30-year period between 1983 and 2012 [17]. This analysis identified 312,420 persons with a first-time hospitalization for AF. The 5-year mortality risk decreased linearly from 58% in 1983–1987 to 48% in 2008–2012 After adjustment for age, sex and individuals’ comorbidities, the hazard ratio of death after AF diagnosis in 2008–2012 was 0.62 (95% confidence interval 0.61–0.63) compared to the 1983–1987 period.

Finally, an analysis of the United Kingdom-based Clinical Practice Research Datalink evaluated mortality trends in AF in the period 1998–2010 [18]. The Clinical Practice Research Datalink is a validated primary care database linked to hospitalizations. Patients in this database can be considered representative of the population of the United Kingdom. During the study period, the 1-year mortality was 9% in women and 11% in men diagnosed with AF, with higher mortality in older (24% in those 85 years and older) than younger individuals (1% in those 18–39 years of age). Mortality trends were evaluated independently in patients 55 to 74 years old and older patients. Reductions in mortality were observed for the younger group (3% relative rate reduction per year, 95% confidence interval 1%, 5%), but not in the older (0% relative rate reduction per year, 95% confidence interval −1%, 1%).

In summary, most studies, but not all, published in the last five years suggest that mortality after receiving an AF diagnosis has slightly declined over the preceding decades across different populations. Mortality reductions seem to extend to both men and women, though some evidence suggests that the change may be less manifest in older patients. Available studies in the United States are inconsistent and have had limited geographic and demographic representation and, therefore, it is unclear whether reduced mortality has occurred across different areas and in diverse racial and ethnic groups. The Table summarizes the characteristics and results of recent studies evaluating mortality trends in AF.

Table.

Study characteristics and main results of recent studies evaluating trends in mortality rates among patients with atrial fibrillation.

Study	Period	Location	No. of AF cases	Main findings
Framingham Heart Study [14]	1958–2007	Framingham, MA, USA	1,544	5-yr mortality rate: Women: 55% (1958–67) to 39% (1998–2007) Men: 53% (1958–67) to 37% (1998–2007)
Chamberlain et al. [8]	2000–2010	Olmsted Co, MN, USA	3,344	Risk of death did not differ over the study period (HR 0.96, 95% CI, 0.71–1.32 for 2010 versus 2000)
Geisinger Health System	2004–2016	Pennsylvania, USA	21,133	Small relative increase in mortality. Annual relative risk (95%CI) was 1.03 (1.02–1.04) for 3-month survival, 1.02 (1.01–1.03) for 1-year survival, and 1.02 (1.01–1.03) for 5-year survival
Briffa et al. [16]	1995–2010	Western Australia	57,552	30-day mortality: 7% to 6% 1-yr mortality: 18% to 15% 3-year mortality: 29% to 24%
Schmidt et al. [17]	1983–2012	Denmark	312,420	5-yr mortality: 58% (1983–87) to 48% (2008–12)
Clinical Practice Research Datalink [18]	1998–2010	United Kingdom	57,818	Mortality decreased in persons ages 55–74 (IRR 0.97, 95%CI 0.95–0.99 per year) but not in persons 75+ (IRR 1.00, 95%CI 0.99–1.01 per year)

CI: Confidence interval. IRR: Incidence rate ratio. HR: hazard ratio

WHAT CAN EXPLAIN TEMPORAL CHANGES IN MORTALITY?

The improved survival after AF diagnosis could be driven by a plethora of factors, including advances in the prevention of stroke and other AF-related complications, a change in the case-mix of AF patients being diagnosed over time, and better control of comorbidities and mortality risk factors in AF patients. Oral anticoagulation in patients with AF has demonstrated efficacy in the reduction of stroke and overall mortality [12]. More recently, direct oral anticoagulants have shown to reduce mortality compared to warfarin in patients with AF [19]. Consequently, guidelines for AF treatment have recommended oral anticoagulation in most patients with AF for some time now [20, 21], with more recent updates to the guidelines favoring the use of the CHA₂DS₂-VASc score over CHADS₂ for stroke prediction, removing the indication for aspirin use as an antithrombotic agent given lack of efficacy, recommending the use of direct oral anticoagulants, which may facilitate compliance and have better outcomes, and supporting the use of nonpharmacological approaches to stroke prevention, including percutaneous occlusion of the left atrial appendage [22, 23]. As a result, the proportion of AF patients being initiated on oral anticoagulation, either warfarin or one of the direct oral anticoagulants, has increased steadily. In Denmark, among 153,682 persons older than 30 newly diagnosed with AF, proportion initiating warfarin in the 60-day period after diagnosis increased from 14% in 1996 to 41% in 2011 [24]. These changes were accompanied by reductions in 1-year risk of thromboembolism in AF patients over that same period, from 3.9% in 1996 to 2.6% in 2011, with steeper reductions in patients initiating oral anticoagulation. A similar analysis of newly diagnosed AF patients in the US Medicare population between 1992 and 2010 reported increases in warfarin use from 29% to 63% in men, 31% to 59% in women, 27% to 61% in whites, 19% to 52% in blacks, and 19% to 49% in patients of other race [25]. Similar increases in oral anticoagulation use have been reported in other regions, including Quebec (period 1998 to 2006) [26], Norway (period 2010 to 2015) [27], and Germany (period 2005 to 2014) [28]. The impact on mortality trends of increased rates of oral anticoagulation in general and the growing use of direct oral anticoagulants, however, has not been quantified.

Another major change in the management of patients with AF has been the growing availability and use of catheter ablation of AF predominantly through pulmonary vein isolation [29]. AF ablation has demonstrated effectiveness for restoring and maintaining sinus rhythm and improving quality of life [30]. However, evidence of AF ablation having a beneficial impact on cardiovascular endpoints and death is lacking. For example, the CABANA (Catheter Ablation verses Anti-arrhythmic Drug Therapy for Atrial Fibrillation) trial did not show superiority of ablation versus drug therapy for a combined primary outcome including death, disabling stroke, severe bleeding or cardiac arrest [13]. In contrast, the CASTLE-AF (Catheter Ablation versus Standard Conventional Therapy in Patients with Left Ventricular Dysfunction and Atrial Fibrillation) trial reported lower mortality associated with ablation in patients with AF and heart failure with reduced ejection fraction [31]. All in all, though AF ablation may be responsible for lower mortality in some patient subgroups, it is unlikely to explain trends in mortality among AF patients.

Changes in the average severity of AF diagnoses may partly explain improved survival if AF is diagnosed in earlier stages, among younger individuals, or in those with fewer comorbidities. No evidence suggests this is the case. Among patients with AF enrolled in Medicare in the United States, prevalence of comorbidities actually increased in the period 1993–2007 [3]. In Olmsted County, Minnesota, the average CHADS₂ score of newly diagnosed AF patients did not change between 2000 and 2010 [8]. With the extended use of wearable devices able to diagnose AF and the push for AF screening programs, the future case mix of new AF may change, which will have to be considered in evaluating mortality trends.

Improvements in the management of comorbidities among AF patients could have also led to better survival. Control of most traditional cardiovascular risk factors, such as hypertension, hypercholesterolemia, and smoking, has improved over the last decades [32]. Even though the prevalence of obesity is increasing in most settings, its impact on mortality among AF patients is controversial [33]. These changes may have resulted in reduced mortality rates particularly from cardiovascular causes, though there is no published evidence that these improvements in risk factor control are also occurring in AF patients.

Finally, changes in AF mortality should be assessed in the context of changes in mortality in patients with other cardiovascular diseases. significant improvements in survival over time have been described for patients with heart failure [34], stroke [35], and myocardial infarction [36]. Given the interrelated aspects of these conditions, by sharing cardiovascular risk factors and by acting as risk factors or triggers for each other, improved survival in one condition is likely to have a positive impact on the other.

IMPLICATIONS FOR CLINICAL PRACTICE AND FUTURE RESEARCH

Despite mortality rates in patients with AF improving over the last few decades, mortality in AF patients remains higher than in persons without AF [37]. This enduring gap needs to provide impetus for the development of novel interventions to prevent premature death in AF. At the same time, mortality is not the only outcome of relevance to AF patients. Preventing stroke, heart failure, and avoidable hospitalizations, as well as improving health-related quality of life and other patient-reported outcomes, is key [38]. A comprehensive assessment of trends in AF prognosis requires evaluation of multiple relevant clinical outcomes.

Better survival in patients with AF could be achieved by improved oral anticoagulation, extending its use and increasing adherence [39], by better control of cardiovascular risk factors in the context of an integrated approach to patient management [40], and by greater attention to the management of multimorbidity, which is highly prevalent in patients with AF and is associated with adverse outcomes [41, 42].

Persistent high mortality rates despite recent improvements also highlight the need to emphasize primary prevention of AF. To date, however, there are no established interventions specifically developed for the primary prevention of AF [43]. Prevention and control of cardiovascular risk factors, particularly hypertension and obesity, are likely to reduce AF risk, but this has not been tested in trials with AF as a primary endpoint. Addressing this knowledge and practice gap should be a priority in AF research over the coming years [44].

Lastly, even if published studies suggest that mortality after AF diagnosis has improved in developed countries (United States, Europe, Australia), we lack information on whether improvements have occurred in other world regions and across different sociodemographic groups. For example, studies in the US have been limited geographically, including mostly white individuals. We lack information on trends in AF-related mortality across racial and ethnic groups, as well as across diverse socioeconomic backgrounds, even though racial and socioeconomic disparities in AF treatment and outcomes have been described repeatedly [45–47].

CONCLUSION

In summary, AF is unequivocally associated with increased mortality. Recent trends, however, suggest that mortality rates can be improving among AF patients. Understanding the mechanisms for these changes and developing novel approaches to improve survival in AF patients are areas deserving of future research.