Article Text

Original research
Sex-specific differences in adverse outcome events among patients with atrial fibrillation
  1. Simone Evers-Dörpfeld1,2,
  2. Stefanie Aeschbacher1,2,
  3. Elisa Hennings1,2,
  4. Ceylan Eken1,2,
  5. Michael Coslovsky1,2,3,
  6. Nicolas Rodondi4,5,
  7. Jürg H Beer6,
  8. Giorgio Moschovitis7,
  9. Peter Ammann8,
  10. Richard Kobza9,
  11. Selinda Ceylan1,2,
  12. Melina Krempke1,2,
  13. Christine S Meyer-Zürn1,2,
  14. Elisavet Moutzouri4,5,
  15. Anne Springer1,2,
  16. Christian Sticherling1,2,
  17. Leo H Bonati10,
  18. Stefan Osswald1,2,
  19. Michael Kuehne1,2,
  20. David Conen11
  21. for the Swiss-AF Investigators
  1. 1 Department of Medicine, Cardiology Division, University Hospital Basel, Basel, Switzerland
  2. 2 Cardiovascular Research Institute, University Hospital Basel, Basel, Switzerland
  3. 3 Department of Clinical Research, Clinical Trial Unit, University Hospital Basel, Switzerland
  4. 4 University of Bern Institute of Primary Health Care, Bern, Switzerland
  5. 5 Department of General Internal Medicine, Inselspital University Hospital Bern, Bern, Switzerland
  6. 6 Department of Medicine, Baden Cantonal Hospital, Baden, Switzerland
  7. 7 Division of Cardiology, Ospedale Regionale di Lugano-Civico e Italiano, Lugano, Switzerland
  8. 8 Department of Cardiology, Kantonsspital Sankt Gallen, Sankt Gallen, Switzerland
  9. 9 Department of Cardiology, Luzerner Kantonsspital, Luzern, Switzerland
  10. 10 Department of Neurology and Stroke Center, University Hospital Basel, Basel, Switzerland
  11. 11 Population Health Research Institute, McMaster University, Hamilton, Ontario, Canada
  1. Correspondence to Dr David Conen, Population Health Research Institute, McMaster University, Hamilton, ON L8S 4L8, Canada; David.Conen{at}


Objective To assess whether women with atrial fibrillation (AF) have a higher risk of adverse events than men during long-term follow-up since controversial data have been published.

Methods In the context of two very similar observational multicentre cohort studies, we prospectively followed 3894 patients (28% women) with previously documented AF for a median of 4.02 (3.00–5.83) years. The primary outcome was a composite of ischaemic stroke, myocardial infarction and cardiovascular death. Secondary outcomes included the individual components of the composite outcome, hospitalisation for heart failure, major and clinically relevant non-major bleeding, stroke or systemic embolism and non-cardiovascular death.

Results Mean age was 73.1 years in women vs 70.8 years in men. The incidence of the primary endpoint in women versus men was 2.46 vs 3.24 per 100 patient-years, respectively (adjusted HR (aHR) 0.74, 95% CI 0.58 to 0.94; p=0.01). Women died less frequently from cardiovascular (aHR 0.57, 95% CI 0.41 to 0.78; p<0.001) and non-cardiovascular causes (aHR 0.68, 95% CI 0.47 to 0.98; p=0.04). There were no significant sex-specific differences in stroke (incidence 1.05 vs 1.00; aHR 1.02, 95% CI 0.70 to 1.49, p=0.93), myocardial infarction (incidence 0.67 vs 0.72; aHR 0.98, 95% CI 0.61 to 1.57, p=0.94), major and clinically relevant non-major bleeding (incidence 4.51 vs 4.34; aHR 0.95, 95% CI 0.79 to 1.15, p=0.63) or heart failure hospitalisation (incidence 3.28 vs 3.07; aHR 1.06, 95% CI 0.85 to 1.32, p=0.60).

Conclusion In this large study of patients with established AF, women had a lower risk of death than men, but there were no sex-specific differences in other adverse outcomes.

  • atrial fibrillation
  • outcome assessment
  • healthcare
  • risk factors
  • stroke
  • epidemiology

Data availability statement

No data are available.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Patients with atrial fibrillation (AF) face an increased risk of death and other adverse events, including stroke, heart failure and cognitive dysfunction.1–4 Whether or not women with AF have a higher risk of adverse outcomes than men is controversial.5–10 In the RACE II Study, which included 403 men and 211 women with permanent AF, women had a higher incidence of a cardiovascular composite outcome, but also presented with more risk factors than men.6 The risk of adverse events between men and women was not statistically significant after adjusting for these risk factors, indicating that sex-related differences in event rates may be in part driven by differences in risk factors.6 In contrast, findings from a much larger registry suggested that women had a higher risk of stroke and systemic embolism than men even after multivariable adjustment.8 However, follow-up duration was only 1 year.

Therefore, more data from large studies with long-term follow-up are needed to better understand whether there are sex-specific differences in adverse events among patients with AF, and whether these associations are independent of differences in comorbidities and risk factors. For this purpose, we investigated sex-specific differences in adverse clinical outcomes among 3894 contemporary patients with AF followed for up to 10 years.


Study design and participants

For this analysis, we combined data from two ongoing observational prospective multicentre studies with very similar methodology: Swiss Atrial Fibrillation Study (Swiss-AF) and Basel Atrial Fibrillation Study (Beat-AF). We previously published details about study methodology.11 12 In brief, Swiss-AF enrolled 2415 patients with documented AF between 2014 and 2017 across 14 centres throughout Switzerland. Recruitment of the Beat-AF Study took place between 2010 and 2014, and 1553 patients were included across seven centres in Switzerland.

In both cohorts, patients were required to have an age ≥65 years with previously documented AF. A convenience sample of patients <65 years was enrolled in both cohorts, to assess the effects of AF on individuals in the active workforce. Main exclusion criteria were the inability to sign informed consent, any acute illness within the past 4 weeks and any secondary forms of AF (eg, postoperative AF or sepsis related AF). Patients participating in Beat-AF were not allowed to take part in Swiss-AF. Of 3968 study patients enrolled in both studies, 7 were removed from Beat-AF due to accidental enrolment in both cohorts, and 67 participants were excluded because they had no follow-up information, resulting in 3894 (98%) patients included in this analysis (figure 1).

Figure 1

Flow chart. BEAT-AF, Basel Atrial Fibrillation Study; SWISS-AF, Swiss Atrial Fibrillation Study.

Study variables

The same standardised questionnaires were used in both cohorts to assess personal and medical information during yearly follow-up visits. Health perception was determined using a Visual Analogue Scale ranging from 0 to 100, with 0 being the worst and 100 being the best imaginable health status. Patients provided information about AF-related symptoms, such as palpitations, weakness, dyspnoea, chest pain, dizziness, fatigue, exercise intolerance or syncope. Smoking status was categorised into current smokers and current non-smokers (past or never smokers). Body weight and height were measured using calibrated devices and body mass index (BMI) was calculated as weight in kilograms divided by height in metres squared. The mean of three consecutive blood pressure measurements obtained in a supine position was used in this analysis. AF type was classified by the local investigators into paroxysmal, persistent and permanent AF according to the 2010 AF guidelines of the European Society of Cardiology.13 We classified rhythm control medications according to the Vaughan-Williams classification.14

Follow-up and clinical outcome events

All Beat-AF patients had a face-to-face study visit at baseline. Subsequent follow‐up information was collected by mailed questionnaires and phone interviews. In Swiss‐AF, yearly follow-up study visits were done in person.

The primary endpoint for this study consisted of a composite of ischaemic stroke, myocardial infarction and cardiovascular death. Secondary outcomes included cardiovascular and non-cardiovascular death, stroke or systemic embolism, myocardial infarction, hospitalisation for heart failure, and major and clinically relevant non-major bleeding. Event definitions are provided in online supplemental table S1. We collected detailed information for every clinical outcome detected. All events were confirmed by two independent reviewers.

Supplemental material

Statistical analysis

Baseline characteristics were stratified by sex. Continuous data are presented as numbers (percentages) and means±SD or medians (IQR) and compared using t-tests or Mann-Whitney U tests, as appropriate. Categorical variables were compared using Χ2 tests.

Person-years of follow-up were calculated as the time from study enrolment until the first occurrence of an outcome event, last follow-up visit, death, drop-out or loss to follow-up. Cox proportional hazards models were constructed to investigate the associations of sex with clinical outcome events, and to adjust for differences in comorbidities and risk factors. Results were presented as HRs and 95% CIs. In a first step, all models were adjusted for age. In a subsequent step, these models were further adjusted for education, BMI, smoking status, alcohol intake, oral anticoagulation, AF type (paroxysmal vs non-paroxysmal), rhythm control medication, and a history of AF intervention therapy, hypertension, diabetes mellitus, heart failure, coronary heart disease, stroke/transient ischaemic attack, sleep apnoea syndrome or renal failure. A third model additionally adjusted for health perception and AF-related symptoms. The proportional hazards assumption was checked for each model using Schönfeld residuals, and no violations were detected.

We conducted prespecified subgroup analyses for the primary endpoint according to age (≥75 vs <75 years), oral anticoagulation, AF type (paroxysmal vs non-paroxysmal), history of stroke, history of diabetes and history of coronary artery disease (CAD). We included multiplicative interaction terms in separate non-stratified models to test for a subgroup effect. All interaction tests used the same p value threshold as indicated below, and they were considered to be hypothesis generating. All statistical analyses were performed using R (V.1.3.1073). A p value of <0.05 was considered to indicate statistical significance.

Patient and public involvement

Patients and the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research.


Baseline characteristics of the 3894 patients stratified by sex are presented in table 1. Of all patients, 1095 (28%) participants were women and the mean age of the overall population was 71.4 years. Compared with men, women were older (73±9 vs 71±10 years, p<0.001), had a lower health perception (69±19 vs 73±18, p<0.001), reported more often AF-related symptoms (78% vs 62%, p<0.001) and had higher systolic blood pressure levels (137±20 mm Hg vs 133±18 mm Hg, p<0.001). In contrast, men more often had diabetes mellitus (18% vs 11%, p<0.001) and CAD (32% vs 15%, p<0.001).

Table 1

Baseline characteristics stratified by sex

During a median follow-up time of 4.02 (IQR 3.00–5.83) years, we confirmed 502 primary outcome events. Incidence rates per 100 patient-years were 2.46 in women and 3.24 in men (table 2, figure 2). After multivariable adjustment, the HR for female sex was 0.74 (95% CI 0.58 to 0.94, p=0.01). The absolute and relative risks of all secondary outcomes are presented in table 2. The incidence rates of stroke/systemic embolism in women and men were 1.05 and 1.00 per 100 person-years, respectively. In multivariable models, we found no evidence that sex was associated with stroke/systemic embolism (adjusted HR 1.02; 95% CI 0.70 to 1.49, p=0.93). As shown in table 2, we found no evidence that sex was associated with myocardial infarction, major and clinically relevant non-major bleeding, or heart failure hospitalisations.

Figure 2

Cumulative incidence of the primary composite outcome stratified by sex.The composite outcome was defined as time to first occurrence of ischaemic stroke, myocardial infarction or cardiovascular death. HR adjusted for age.

Table 2

Associations of sex with primary and secondary outcome events

Overall, 505 all-cause deaths occurred, of which 313 (62%) were of cardiovascular origin. In women versus men, the incidence rate per 100 person-years was 2.28 vs 3.22 for all-cause death, 1.27 vs 2.05 for cardiovascular death and 0.99 vs 1.17 for non-cardiovascular death. The multivariable-adjusted HRs for female sex were 0.62 (95% CI 0.49 to 0.79, p<0.001) for all-cause death, 0.57 (95% CI 0.41 to 0.78, p<0.001) for cardiovascular death and 0.68 (95% CI 0.47 to 0.98, p=0.04) for non-cardiovascular death (table 2).

There were no statistically significant interactions across all subgroups analysed, as shown in table 3, with one potential exception. We observed a potentially stronger sex-specific effect in patients taking oral anticoagulation compared with those not taking oral anticoagulation at baseline (p for interaction=0.05).

Table 3

Associations of female sex with the composite outcome across various subgroups


Several important findings emerged from this prospective analysis of sex-specific differences in patients with AF. First, there were many differences in baseline characteristics between men and women. While women suffered more often from AF-related symptoms and lower health perception, men had more cardiovascular comorbidities and risk factors. Second, women had a significantly lower risk of having a primary outcome event than men. Third, this reduced risk was due to an increased risk of cardiovascular and non-cardiovascular mortality in men. In contrast, men and women had a similar risk of stroke/systemic embolism, myocardial infarction, heart failure hospitalisation or bleeding.

Women and men in our study had different baseline characteristics. Women were on average 2 years older, had more often paroxysmal AF and higher blood pressure levels. Those findings are comparable with other studies.6 8 As reported previously, rhythm control therapies such as pulmonary vein isolation, electrocardioversion or radiofrequency ablations were less likely in women despite a higher prevalence of AF-related symptoms6 10 15 and a lower health perception.10 15 16 These important differences need to be taken into account in the interpretation of sex-specific analyses. Despite multivariable adjustment, some residual confounding likely persists.

Women had a lower risk of the primary composite outcome than men. This association was driven by a lower risk of death in women. After multivariable adjustment, both cardiovascular and non-cardiovascular deaths were significantly lower in women. Our findings are similar to those from the ORBIT-AF registry,10 but differ from findings of the RACE II and GARFIELD-AF Studies, where no significant sex-specific differences in mortality were observed.6 8 A meta-analysis found that women with AF had higher risks of all-cause and cardiovascular mortality than men,5 whereas another meta-analysis published more recently found no sex-specific differences for cardiovascular death, all-cause death and major bleeding.17 The differential associations of men and women with mortality across different studies are likely multifactorial and reflect differences in baseline characteristics, different settings as well as somewhat different eligibility criteria. It is possible that improved treatment of women in more recent studies has reduced some of the sex-specific differences in earlier studies. A possible explanation why men experienced more fatal events in our cohort could be that they had a higher burden of CAD, heart failure and other comorbidities than women, leading to more severe events in men with a higher case fatality rate. Alternatively, it is also possible that the higher rate of sudden death in men compared with women may reduce the number of potentially lifesaving hospital admissions in men.18 While we have taken into account differences in risk factors and comorbidities, the observed differences in mortality may reflect the survival advantage in women seen in population-based statistics in Switzerland19 and elsewhere.20 21

On the other hand, women did not have a higher risk of stroke or systemic embolism. These results are in contrast to various previous studies,8 10 but in line with data from the RACE II Study that described similar stroke rates in men and women after multivariable adjustments.6 Women in our study were well anticoagulated (85% vs 84% men) and the age difference was only 2 years. These differences were more extreme in the ORBIT-AF registry (women 4 years older, anticoagulation rate 10% lower in women) and in the GARFIELD-AF registry (women 4 years older, women were more likely to have subtherapeutic anticoagulation). Our results are therefore consistent with the fact that widespread oral anticoagulation translates into lower stroke rates in both men and women.22–24

We found no significant associations of female sex with the occurrence of heart failure hospitalisation, myocardial infarction, and major and clinically relevant non-major bleeding. Recent observational studies regarding sex-specific differences in patients with AF showed similar results,6 8–10 but meta-analyses presented conflicting results.5 17 Data from the ORBIT-AF registry after 2 years of follow-up were in line with our data and did not find significant differences between men and women.10 Most previous studies reported similar event rates for bleeding in women and men,6 8 while some studies reported higher rates of heart failure hospitalisation and myocardial infarction in men compared with women during their 1 year of follow-up.9 Again, residual confounding and intrinsic differences between men and women could explain some of these study-specific differences.

Strengths of our study include the large sample size of 3894 well-characterised patients with AF with up to 10 years of prospective follow-up and high anticoagulation rates. Some potential limitations also need to be considered. The present data cannot reveal causal mechanisms given the observational nature of our study. Participants in our cohort were mostly white and enrolled in the Swiss health insurance system. It is unclear to what extent our data are generalisable to other population settings. Women constituted 28% of all participants and are therefore under-represented, similar to previous studies.6 25 26 The relatively wide CIs for the outcomes of stroke/systemic embolism and myocardial infarction suggest that our study may have been underpowered to detect small sex-specific differences for these events.

In conclusion, in a contemporary population of patients with AF with high anticoagulation rates, men died more often from both cardiovascular and non-cardiovascular causes than women. However, we observed no differences in the occurrence of cardiovascular events, including stroke or systemic embolism, myocardial infarction, hospitalisation for heart failure or bleeding.

Key messages

What is already known on this subject?

  • Controversial data have been published about whether women with atrial fibrillation (AF) have a higher risk of adverse events than men during long-term follow-up.

What might this study add?

  • Our study provides long-term follow-up data up to 10 years in a large contemporary sample of patients with AF. Our results show that in well-anticoagulated patients with established AF, women had a lower risk of death than men, but no sex-specific differences in risk of stroke of other adverse outcome events were observed.

How might this impact on clinical practice?

  • Our data emphasise the importance of anticoagulation in both women and men given the low risk of stroke in our data.

Data availability statement

No data are available.

Ethics statements

Patient consent for publication

Ethics approval

All participants signed a written informed consent. Both studies were conducted in accordance with the Declaration of Helsinki and approved by the lead ethics committee EKNZ (Ethikkommission Nordwest- und Zentralschweiz).


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • SE-D and SA are joint first authors.

  • Presented at This paper has previously been presented by SE-D as an abstract at the Joint Annual Meeting of the Swiss Society of Cardiology (SSC) and the Swiss Society of Cardiac and Thoracic Vascular Surgery (SSCC) 9–11 June 2021.

  • Collaborators All Swiss-AF and Beat-AF investigators are listed in the supplement.

  • Contributors Conception and design—DC, SE-D, SA, MK and SO. Analysis and interpretation of the data—SA, SE-D, EH, DC and MC; all authors provided input in the interpretation of the data. Drafting of the manuscript—SE-D, SA and DC. Critical revision for important intellectual content—all authors. Final approval of the manuscript—all authors. Guarantor—DC.

  • Funding The Beat-AF Study was supported by the Swiss National Science Foundation (grant number PP00P3_159322), the Swiss Heart Foundation, the University of Basel, Boehringer Ingelheim, SanofiAventis, Merck Sharp & Dome, Bayer, Daiichi-Sankyo and Pfizer/Bristol-Myers Squibb. The Swiss-AF cohort study is supported by grants of the Swiss National Science Foundation (grant numbers 33CS30_148474 and 33CS30_177520), the Foundation for Cardiovascular Research Basel and the University of Basel.

  • Disclaimer None of the funders had any role in manuscript preparation or in the decision to submit the manuscript for publication.

  • Competing interests CSM-Z reports a research grant from Medtronic and speaker fees from Vifor Pharma and Novartis. DC has received consultant/speaker fees from Servier Canada and Roche Diagnostics, outside of the submitted work. GM has received consultant fees for taking part in advisory boards from Novartis, AstraZeneca, Bayer and Böhringer Ingelheim, outside of the submitted work. JHB reports grants from the Swiss National Foundation of Science, the Swiss Heart Foundation, grants from Bayer, lecture fees from Sanofi Aventis and Amgen, to the institution outside the submitted work. LHB received grants from the Swiss National Science Foundation (PBBSB-116873, 33CM30-124119, 32003B-156658; Bern, Switzerland), the Swiss Heart Foundation (Bern, Switzerland, and the University of Basel (Basel, Switzerland). LHB has received an unrestricted research grant from AstraZeneca, and consultancy or advisory board fees or speaker’s honoraria from Amgen, Bayer, Bristol-Myers Squibb, and Claret Medical, and travel grants from AstraZeneca and Bayer. MK reports personal fees from Bayer, personal fees from Böhringer Ingelheim, personal fees from Pfizer BMS, personal fees from Daiichi Sankyo, personal fees from Medtronic, personal fees from Biotronik, personal fees from Boston Scientific, personal fees from Johnson&Johnson, grants from Bayer, grants from Pfizer BMS, grants from Boston Scientific and grants from Daiichi Sankyo. He is supported by the Swiss National Science Foundation (32473B_176178) and the Swiss Heart Foundation. NR received a grant from the Swiss Heart Foundation. RK receives institutional grants from Abbott, Biosense-Webster, Boston-Scientific, Biotronik, Medtronic and Sis-Medical.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.