Abstract
Although commonly associated with cardiovascular disease or other medical conditions, atrial fibrillation may also occur in individuals without any known underlying conditions. This manifestation of atrial fibrillation has been linked to extensive and long-term exercise, as prolonged endurance exercise has shown to increase prevalence and risk of atrial fibrillation. In contrast, more modest physical activity is associated with a decreased risk of atrial fibrillation, and current research indicates a J-shaped association between atrial fibrillation and the broad range of physical activity and exercise. This has led to the hypothesis that the mechanisms underlying an increased risk of atrial fibrillation with intensive exercise are different from those underlying a reduced risk with moderate physical activity, possibly linked to distinctive characteristics of the population under study. High volumes of exercise over many years performed by lean, healthy endurance trained athletes may lead to cardiac (patho)physiological alterations involving the autonomic nervous system and remodelling of the heart. The mechanisms underlying a reduced risk of atrial fibrillation with light and moderate physical activity may involve a distinctive pathway, as physical activity can potentially reduce the risk of atrial fibrillation through favourable effects on cardiovascular risk factors.
Introduction
Atrial fibrillation is the most common cardiac arrhythmia, affecting more than six million Europeans.1 Atrial fibrillation is associated with a five-fold increased risk of stroke, a three-fold increased risk of heart failure and a two-fold increased risk of myocardial infarction and death,2,3 which necessitates preventive measures. Although most commonly associated with cardiovascular disease or other medical conditions, atrial fibrillation may also occur in individuals without any underlying known medical conditions, which is often called lone or idiopathic atrial fibrillation.1 This latter manifestation of atrial fibrillation has been linked to extensive and long-term endurance exercise,4 but knowledge of exercise, physical activity and risk of atrial fibrillation is still limited and rather ambiguous. Elite athletes with years of systematic exercise and non-elite athletes with a high volume of endurance exercise show increased prevalence and risk of atrial fibrillation.5,6 On the other hand, studies of the general population indicate that light-to-moderate physical activity may be protective against atrial fibrillation.7–11 In this paper, we discuss possible mechanisms underlying these seemingly contradictory associations, suggesting there may be at least two distinct mechanisms underlying the associations between physical activity and atrial fibrillation. Moreover, this study adds to the existing literature by aiming to identify whether study characteristics can explain some of the diversity in existing findings.
Atrial fibrillation risk in populations with different levels of physical activity and exercise
Current and former elite athletes
Research on exercise at professional elite level involves elite athletes with years of high-volume exercise, and Wernhart and Halle12 have defined elite athletes as ‘individuals performing at a competitive level with a high exercise performance’, characterized by a high maximal oxygen uptake (VO2max) and years of systematic training and a large metabolic turnover. The body of research on atrial fibrillation in elite athletes can typically be classified according to current and former elite athletes, the latter constituted by former athletes, mostly in endurance sport, aged >45 years.
One of the first studies on exercise and atrial fibrillation reported a 9% prevalence of atrial fibrillation in young elite athletes, substantially higher than the age-matched population prevalence of atrial fibrillation.13 Another study on athletes competing at Olympic, national and regional levels could not replicate these findings, reporting a prevalence of atrial fibrillation < 1% in these athletes14 (Table 1). Studies of former elite athletes demonstrate a higher prevalence of atrial fibrillation compared with the general population prevalence,5,15,16 although not totally consistently17 (Table 2). A meta-analysis including six case–control studies concluded that the risk of developing atrial fibrillation was five times higher in athletes than in controls (mean age 51 years, 93% men).18
Prevalence and risk of atrial fibrillation in former and current elite athletes.
| Publication year, authors | Study population | Men (n) | Women (n) | Age (years) | Follow-up (years) | AF diagnosis | PA level/dose | Prevalence (%) or RR (95% CI) of AF, athletes vs. control group |
|---|---|---|---|---|---|---|---|---|
| 1998 Karjalainen et al.5 | Veteran orienteers vs. healthy controls, Finland | 262 athletes, 373 controls | 0 | Mean 47/49 | 10 | Self- reported | High level for many years | RR athletes vs. controls: OR 5.5 (1.3–24.4) |
| 1998 Furlanello et al.13 | Young elite athletes, Italy | 122 athletes | 24 athletes | Mean 24 | Cross- sectional | ECG | Elite athletes | Athletes 9% |
| 2005 Pelliccia et al.14 | Highly trained athletes of 38 different sports, Italy | 1298 athletes | 479 athletes | Mean 24 | Cross- sectional | ECG | Highly trained athletes | Athletes < 1% |
| 2008 Baldesberger et al.15 | Former professional cyclists vs. golfers, Switzerland | 62 athletes, 62 controls | 0 | Mean 66 | 38 | ECG | Former professional cyclists | Cyclists 10%, controls 0%, p = 0.03 |
| 2009 Bjørnstad et al.17 | Former elite athletes, Norway | 15 athletes | 15 | Mean 24 | 15 | ECG | Former endurance elite athletes | 0, no control group |
| 2012 van Buuren et al.16 | Former top-level handball players vs. healthy controls, Germany | 33 athletes, 24 controls | 0 | Mean 57 | Cross- sectional | ECG | >20 years with 18 h/week | Former athletes 30%, controls 0% |
| Publication year, authors | Study population | Men (n) | Women (n) | Age (years) | Follow-up (years) | AF diagnosis | PA level/dose | Prevalence (%) or RR (95% CI) of AF, athletes vs. control group |
|---|---|---|---|---|---|---|---|---|
| 1998 Karjalainen et al.5 | Veteran orienteers vs. healthy controls, Finland | 262 athletes, 373 controls | 0 | Mean 47/49 | 10 | Self- reported | High level for many years | RR athletes vs. controls: OR 5.5 (1.3–24.4) |
| 1998 Furlanello et al.13 | Young elite athletes, Italy | 122 athletes | 24 athletes | Mean 24 | Cross- sectional | ECG | Elite athletes | Athletes 9% |
| 2005 Pelliccia et al.14 | Highly trained athletes of 38 different sports, Italy | 1298 athletes | 479 athletes | Mean 24 | Cross- sectional | ECG | Highly trained athletes | Athletes < 1% |
| 2008 Baldesberger et al.15 | Former professional cyclists vs. golfers, Switzerland | 62 athletes, 62 controls | 0 | Mean 66 | 38 | ECG | Former professional cyclists | Cyclists 10%, controls 0%, p = 0.03 |
| 2009 Bjørnstad et al.17 | Former elite athletes, Norway | 15 athletes | 15 | Mean 24 | 15 | ECG | Former endurance elite athletes | 0, no control group |
| 2012 van Buuren et al.16 | Former top-level handball players vs. healthy controls, Germany | 33 athletes, 24 controls | 0 | Mean 57 | Cross- sectional | ECG | >20 years with 18 h/week | Former athletes 30%, controls 0% |
AF: atrial fibrillation; PA: physical activity; RR: relative risk; CI: confidence interval; OR: odds ratio; ECG: electrocardiography.
Prevalence and risk of atrial fibrillation in former and current elite athletes.
| Publication year, authors | Study population | Men (n) | Women (n) | Age (years) | Follow-up (years) | AF diagnosis | PA level/dose | Prevalence (%) or RR (95% CI) of AF, athletes vs. control group |
|---|---|---|---|---|---|---|---|---|
| 1998 Karjalainen et al.5 | Veteran orienteers vs. healthy controls, Finland | 262 athletes, 373 controls | 0 | Mean 47/49 | 10 | Self- reported | High level for many years | RR athletes vs. controls: OR 5.5 (1.3–24.4) |
| 1998 Furlanello et al.13 | Young elite athletes, Italy | 122 athletes | 24 athletes | Mean 24 | Cross- sectional | ECG | Elite athletes | Athletes 9% |
| 2005 Pelliccia et al.14 | Highly trained athletes of 38 different sports, Italy | 1298 athletes | 479 athletes | Mean 24 | Cross- sectional | ECG | Highly trained athletes | Athletes < 1% |
| 2008 Baldesberger et al.15 | Former professional cyclists vs. golfers, Switzerland | 62 athletes, 62 controls | 0 | Mean 66 | 38 | ECG | Former professional cyclists | Cyclists 10%, controls 0%, p = 0.03 |
| 2009 Bjørnstad et al.17 | Former elite athletes, Norway | 15 athletes | 15 | Mean 24 | 15 | ECG | Former endurance elite athletes | 0, no control group |
| 2012 van Buuren et al.16 | Former top-level handball players vs. healthy controls, Germany | 33 athletes, 24 controls | 0 | Mean 57 | Cross- sectional | ECG | >20 years with 18 h/week | Former athletes 30%, controls 0% |
| Publication year, authors | Study population | Men (n) | Women (n) | Age (years) | Follow-up (years) | AF diagnosis | PA level/dose | Prevalence (%) or RR (95% CI) of AF, athletes vs. control group |
|---|---|---|---|---|---|---|---|---|
| 1998 Karjalainen et al.5 | Veteran orienteers vs. healthy controls, Finland | 262 athletes, 373 controls | 0 | Mean 47/49 | 10 | Self- reported | High level for many years | RR athletes vs. controls: OR 5.5 (1.3–24.4) |
| 1998 Furlanello et al.13 | Young elite athletes, Italy | 122 athletes | 24 athletes | Mean 24 | Cross- sectional | ECG | Elite athletes | Athletes 9% |
| 2005 Pelliccia et al.14 | Highly trained athletes of 38 different sports, Italy | 1298 athletes | 479 athletes | Mean 24 | Cross- sectional | ECG | Highly trained athletes | Athletes < 1% |
| 2008 Baldesberger et al.15 | Former professional cyclists vs. golfers, Switzerland | 62 athletes, 62 controls | 0 | Mean 66 | 38 | ECG | Former professional cyclists | Cyclists 10%, controls 0%, p = 0.03 |
| 2009 Bjørnstad et al.17 | Former elite athletes, Norway | 15 athletes | 15 | Mean 24 | 15 | ECG | Former endurance elite athletes | 0, no control group |
| 2012 van Buuren et al.16 | Former top-level handball players vs. healthy controls, Germany | 33 athletes, 24 controls | 0 | Mean 57 | Cross- sectional | ECG | >20 years with 18 h/week | Former athletes 30%, controls 0% |
AF: atrial fibrillation; PA: physical activity; RR: relative risk; CI: confidence interval; OR: odds ratio; ECG: electrocardiography.
Risk of atrial fibrillation in non-elite endurance athletes.
| Publication year, authors | Study population | Men (n) | Women (n) | Baseline age (years) | Follow-up (years) | AF diagnosis | PA type, level, dose | Prevalence (%) or RR (95% CI) of AF, athletes vs. control group | Shape of relationship |
|---|---|---|---|---|---|---|---|---|---|
| 2008 Molina et al.6 | Marathon runners vs. sedentary controls, Spain | 252 athletes, 305 controls | 0 | Mean 45, range 20–60 | 11 | ECG, lone AF | Total PA | Marathon runners vs. controls: HR 8.8 (1.6–61.29) | |
| 2010 Grimsmo et al.26 | Former Birkebeineren participants, Norway | 78 | 0 | Range 54–92 at follow-up | 30 | ECG, lone AF | Cross-country skiing | Athletes 12.8% | |
| 2013 Andersen et al.27 | Vasaloppet participants, Sweden | 47,477 | 5278 | Mean 38 | 16 | Swedish In-Patient Register | Cross-country skiing, number finished races, finishing time | ≥5 vs. 1 race: HR 1.29 (1.04–1.61). Finishing time: HR 1.20 (0.93–1.55) | Increased risk for each race |
| 2014 Myrstad et al.24 | Birkebeineren participants and the Tromsø Study, Norway | 2277 | 0 | Range 65–90 | Cross- sectional | Self-reported | Sedentary to vigorous, long-term endurance training | Birkebeiner vs. sedentary: OR 1.9 (1.14–3.18). Moderate vs. sedentary: lower risk, NS | J-shape |
| 2014 Myrstad et al.25 | Birkebeineren participants and the Oslo Health Study, Norway | 3545 | 0 | Range 53–92 | Cross- sectional | ECG, self- reported | Sedentary to vigorous, long-term endurance training | Per 10 years’ experience: OR 1.16 (1.06–1.29) | AF risk increases with years of endurance training |
| 2015 Myrstad et al.23 | Birkebeineren participants and the Oslo Health Study, Norway | 0 | 1679 | Range 53–75 | Cross- sectional | Self-reported | Sedentary to vigorous, long-term endurance training | ≥40 years’ endurance training vs. never: OR 2.18 (0.94–5.06) | J-shape |
| Publication year, authors | Study population | Men (n) | Women (n) | Baseline age (years) | Follow-up (years) | AF diagnosis | PA type, level, dose | Prevalence (%) or RR (95% CI) of AF, athletes vs. control group | Shape of relationship |
|---|---|---|---|---|---|---|---|---|---|
| 2008 Molina et al.6 | Marathon runners vs. sedentary controls, Spain | 252 athletes, 305 controls | 0 | Mean 45, range 20–60 | 11 | ECG, lone AF | Total PA | Marathon runners vs. controls: HR 8.8 (1.6–61.29) | |
| 2010 Grimsmo et al.26 | Former Birkebeineren participants, Norway | 78 | 0 | Range 54–92 at follow-up | 30 | ECG, lone AF | Cross-country skiing | Athletes 12.8% | |
| 2013 Andersen et al.27 | Vasaloppet participants, Sweden | 47,477 | 5278 | Mean 38 | 16 | Swedish In-Patient Register | Cross-country skiing, number finished races, finishing time | ≥5 vs. 1 race: HR 1.29 (1.04–1.61). Finishing time: HR 1.20 (0.93–1.55) | Increased risk for each race |
| 2014 Myrstad et al.24 | Birkebeineren participants and the Tromsø Study, Norway | 2277 | 0 | Range 65–90 | Cross- sectional | Self-reported | Sedentary to vigorous, long-term endurance training | Birkebeiner vs. sedentary: OR 1.9 (1.14–3.18). Moderate vs. sedentary: lower risk, NS | J-shape |
| 2014 Myrstad et al.25 | Birkebeineren participants and the Oslo Health Study, Norway | 3545 | 0 | Range 53–92 | Cross- sectional | ECG, self- reported | Sedentary to vigorous, long-term endurance training | Per 10 years’ experience: OR 1.16 (1.06–1.29) | AF risk increases with years of endurance training |
| 2015 Myrstad et al.23 | Birkebeineren participants and the Oslo Health Study, Norway | 0 | 1679 | Range 53–75 | Cross- sectional | Self-reported | Sedentary to vigorous, long-term endurance training | ≥40 years’ endurance training vs. never: OR 2.18 (0.94–5.06) | J-shape |
AF: atrial fibrillation; PA: physical activity; RR: relative risk; CI: confidence interval; ECG: electrocardiography; HR: hazard ratio; OR: odds ratio; NS: not significant at p < 0.05
Risk of atrial fibrillation in non-elite endurance athletes.
| Publication year, authors | Study population | Men (n) | Women (n) | Baseline age (years) | Follow-up (years) | AF diagnosis | PA type, level, dose | Prevalence (%) or RR (95% CI) of AF, athletes vs. control group | Shape of relationship |
|---|---|---|---|---|---|---|---|---|---|
| 2008 Molina et al.6 | Marathon runners vs. sedentary controls, Spain | 252 athletes, 305 controls | 0 | Mean 45, range 20–60 | 11 | ECG, lone AF | Total PA | Marathon runners vs. controls: HR 8.8 (1.6–61.29) | |
| 2010 Grimsmo et al.26 | Former Birkebeineren participants, Norway | 78 | 0 | Range 54–92 at follow-up | 30 | ECG, lone AF | Cross-country skiing | Athletes 12.8% | |
| 2013 Andersen et al.27 | Vasaloppet participants, Sweden | 47,477 | 5278 | Mean 38 | 16 | Swedish In-Patient Register | Cross-country skiing, number finished races, finishing time | ≥5 vs. 1 race: HR 1.29 (1.04–1.61). Finishing time: HR 1.20 (0.93–1.55) | Increased risk for each race |
| 2014 Myrstad et al.24 | Birkebeineren participants and the Tromsø Study, Norway | 2277 | 0 | Range 65–90 | Cross- sectional | Self-reported | Sedentary to vigorous, long-term endurance training | Birkebeiner vs. sedentary: OR 1.9 (1.14–3.18). Moderate vs. sedentary: lower risk, NS | J-shape |
| 2014 Myrstad et al.25 | Birkebeineren participants and the Oslo Health Study, Norway | 3545 | 0 | Range 53–92 | Cross- sectional | ECG, self- reported | Sedentary to vigorous, long-term endurance training | Per 10 years’ experience: OR 1.16 (1.06–1.29) | AF risk increases with years of endurance training |
| 2015 Myrstad et al.23 | Birkebeineren participants and the Oslo Health Study, Norway | 0 | 1679 | Range 53–75 | Cross- sectional | Self-reported | Sedentary to vigorous, long-term endurance training | ≥40 years’ endurance training vs. never: OR 2.18 (0.94–5.06) | J-shape |
| Publication year, authors | Study population | Men (n) | Women (n) | Baseline age (years) | Follow-up (years) | AF diagnosis | PA type, level, dose | Prevalence (%) or RR (95% CI) of AF, athletes vs. control group | Shape of relationship |
|---|---|---|---|---|---|---|---|---|---|
| 2008 Molina et al.6 | Marathon runners vs. sedentary controls, Spain | 252 athletes, 305 controls | 0 | Mean 45, range 20–60 | 11 | ECG, lone AF | Total PA | Marathon runners vs. controls: HR 8.8 (1.6–61.29) | |
| 2010 Grimsmo et al.26 | Former Birkebeineren participants, Norway | 78 | 0 | Range 54–92 at follow-up | 30 | ECG, lone AF | Cross-country skiing | Athletes 12.8% | |
| 2013 Andersen et al.27 | Vasaloppet participants, Sweden | 47,477 | 5278 | Mean 38 | 16 | Swedish In-Patient Register | Cross-country skiing, number finished races, finishing time | ≥5 vs. 1 race: HR 1.29 (1.04–1.61). Finishing time: HR 1.20 (0.93–1.55) | Increased risk for each race |
| 2014 Myrstad et al.24 | Birkebeineren participants and the Tromsø Study, Norway | 2277 | 0 | Range 65–90 | Cross- sectional | Self-reported | Sedentary to vigorous, long-term endurance training | Birkebeiner vs. sedentary: OR 1.9 (1.14–3.18). Moderate vs. sedentary: lower risk, NS | J-shape |
| 2014 Myrstad et al.25 | Birkebeineren participants and the Oslo Health Study, Norway | 3545 | 0 | Range 53–92 | Cross- sectional | ECG, self- reported | Sedentary to vigorous, long-term endurance training | Per 10 years’ experience: OR 1.16 (1.06–1.29) | AF risk increases with years of endurance training |
| 2015 Myrstad et al.23 | Birkebeineren participants and the Oslo Health Study, Norway | 0 | 1679 | Range 53–75 | Cross- sectional | Self-reported | Sedentary to vigorous, long-term endurance training | ≥40 years’ endurance training vs. never: OR 2.18 (0.94–5.06) | J-shape |
AF: atrial fibrillation; PA: physical activity; RR: relative risk; CI: confidence interval; ECG: electrocardiography; HR: hazard ratio; OR: odds ratio; NS: not significant at p < 0.05
However, the external validity of these findings is difficult to assess due to a number of methodological limitations, and some authors have concluded that the evidence for an association between vigorous exercise and atrial fibrillation is weak.19–21 Many of the studies have a cross-sectional design and are lacking a sedentary control group, the number of participants and atrial fibrillation cases is limited, and participants are highly selected by involving mainly men competing at elite level.5,6,13,15,16 Most studies did not adjust for potential confounders, and the lifestyle of an elite athlete may be different from that of the general population in many aspects, which could possibly obscure unadjusted findings. Moreover, athletes may be more aware of their body and sensitive to symptoms, possibly resulting in more diagnoses of atrial fibrillation. Only the studies of Pelliccia et al.14 and Furlanello et al.13 include women.
Endurance exercise on a non-professional level
Endurance exercise aims to improve aerobic capacity, often measured as VO2max. In this paper, we distinguish elite athletes from non-elite athletes. Although the latter population is non-professional, many still perform at a high level and have been exposed to high volumes of exercise over many years. Typically, these non-professional elite athletes have full-time job and family commitments in addition to extensive exercise. In a Danish report, athletes who exercise with high intensity almost daily and > 6 h/week constitute < 5% of the total population.22
Studies of these non-professional athletes with a high endurance exercise volume for many years may further illuminate the issue (Table 3). Myrstad et al. assessed the prevalence of self-reported atrial fibrillation in male Norwegian veteran skiers participating in the 54-km Birkebeiner cross-country ski race and in the general population.23–25 The authors showed that the prevalence of self-reported atrial fibrillation after exclusion of participants with coronary heart disease was 13.0% in the veteran skiers and 9.8% in the general population. Further analyses showed that endurance sport practice gave a significant added risk of atrial fibrillation corresponding to an odds ratio of 1.90.24 Moreover, in a study of 3545 Norwegian men, Myrstad et al.25 showed that cumulative years of regular endurance exercise were associated with a 16% significant gradually increased risk of atrial fibrillation per 10 years of exercise. The increased risk was significant both among skiers and among the men from the general population (20% per 10 years of exercise). Regular endurance exercise was also associated with a gradually increased risk of atrial flutter.25 These findings correspond with another study following skiers participating in the Birkebeiner cross-country ski race for 30 years, showing a high prevalence of atrial fibrillation (12.8%) among the skiers.26 These results are supported by a study of participants of the 90 km cross-country race Vasaloppet, following the participants from 1989 to 2005, using the Swedish In-Patient Register to ascertain atrial fibrillation.27 Those who finished ≥ 5 races had 29% higher risk of atrial fibrillation than those who completed one race. Risk of atrial fibrillation was non-significantly increased in those who had the fastest relative finishing time.
Physical activity and risk of atrial fibrillation in the general population.
| Publication year, authors | Study population | Men (n) | Women (n) | Baseline age (years) | Follow-up (years, mean) | AF diagnosis | PA type, level, dose | RR (95% CI), moderate vs. sedentary PA | RR (95% CI), high vs. sedentary PA | Shape of relationship between PA and AF |
|---|---|---|---|---|---|---|---|---|---|---|
| 2008 Mozaffarian et al.8 | Cardiovascular Health Study, USA | 2287 | 3159 | Mean 73, range > 65 | 12 | ECG, hospital discharge | Leisure-time PA, quintiles, intensity, walking distance/pace | Walking distance, pace: RR 0.78 (0.65–0.94) PA quintile III vs. I: 0.75 (0.61–0.90) | PA quintile V vs. I: RR 0.64 (0.52–0.79). Intensity, high vs. none: RR 0.87 (0.64–1.19) | Walking: inverse, linear, p < 0.001. PA: inverse, linear, p < 0.001. Intensity: U-shape |
| 2009 Aizer et al.34 | Physicians’ Health Study, USA | 16921 | 0 | Range 40–84 | 12 | Self-reported | Vigorous exercise, days/week | 1–2 vs. <1 day/week: NS | 5–7 vs. <1 day/week: RR 1.2 (0.99–1.36) | Positive, linear association, p = 0.04 |
| 2011 Everett et al.30 | Women’s Health Study, USA | 0 | 34759 | Mean 54, range 49–59 | 14 | ECG, medical journal | Leisure-time PA, MET-h/week | Middle vs. low PA: NS | High vs. low PA: NS | NS |
| 2013 Williams and Franklin11 | National Runners’ Health Study II/National Walkers’ Health Study, UK | 19,044 | 27,763 | Mean 44–59 | 6 | Self-reported | Walking and running, MET-h/day | 2–3 vs. > 1 MET-h/day: RR 0.7, p < 0.05 | >3 vs. < 1 MET-h/day: RR 0.6, p < 0.001 | Inverse, linear |
| 2013 Thelle et al.33 | CONOR cohort, Norway | 147,462 | 162,078 | Range 40–45 | 5 | Norwegian Prescription Database, lone AF | Leisure and work PA, four levels | Moderate vs. sedentary: NS | Vigorous vs. sedentary: RR men 3.14 (2.17–4.54), women NS | Men: flipped L, increasing, p < 0.001 |
| 2014 Huxley et al.10 | Atherosclerosis Risk in Communities Study, USA | 6445 | 7774 | Mean 54, range 45–64 | 22 | ECG, hospital discharge | Leisure and work PA, MET-min/week | Middle vs. low PA: NS | High vs. low PA:HR 0.89 (0.80–1.00) | Inverse, linear, p = 0.06 |
| 2014 Azarbal et al.28 | Women’s Health Initiative Observational Study, USA | 0 | 81,317 | Range 50–79 | 11 | Medical records or Medicare | Walking, recreational, strenuous exercise, MET-h/week | 3–9 MET-h/week vs. none: NS | >9 MET-h/week vs. none: RR 0.90 (0.85–0.96) | Inverse, linear, p = 0.003 |
| 2014 Drca et al.29 | Swedish men | 44,410 | 0 | Mean 60, range 45–79 | 12 | Swedish Inpatient Register | Walking/bicycling; light-to-moderate. Exercise: moderate-to-high | Walking > 60 min/day vs. almost never: RR 0.87 (0.77–0.97) | Exercise > 5 h/week vs. almost never: RR 1.17 (1.03–1.32) | Walking: inverse, linear, p = 0.03. Exercise: positive, linear, p = 0.01 |
| 2014 Knuiman et al.31 | Busselton Health Study, Australia | 1861 | 2406 | Range 25–84 | 16 | Hospital admission | Doing any vigorous each week | Not assessed | Vigorous vs. none: HR 0.80 (0.62–1.03) | Not assessed |
| 2015 Drca et al.9 | Swedish mammography cohort | 0 | 36,513 | 60 (49–83) | 12 | Swedish Inpatient Register | Walking/bicycling: light-to-moderate. Exercise: moderate-to-high | Walking, ≥ 40 min/day vs. almost never: RR 0.81 (0.72–0.92) | Exercise > 5 vs. < 1 h/week: RR 0.85 (0.75–0.95) | Walking: inverse, linear, p > 0.001. Exercise: inverse, linear, p = 0.02 |
| 2015 Bapat et al.32 | Multi-ethnic study of atherosclerosis, USA | 2722 | 3071 | 62 (45–84) | 7 | Medical records or Medicare | Total intentional exercise, vigorous exercise, MET-min/week | Lower or middle tertile vs. none PA: NS | Upper tertile PA: NS | HR for AF increased with total intentional PA |
| 2015 Woodward et al.36 | Taupo bicycle study, New Zealand | 1874 | 725 | ≥16 | 7 | Hospital discharge | Cyclists vs. national cohort, h/week, km/week | No difference cyclists vs. general population | ||
| 2016 Morseth et al.7 | Tromsø Study, Norway | 10,300 | 10,184 | Mean 39 (men), 37 (women) | 20 | ECG | Leisure and work PA, four levels | Moderate vs. sedentary: HR 0.81 (0.68–0.96) | Vigorous vs. sedentary: NS | J-shape |
| 2016 Skielboe et al.35 | 7258 | 8560 | Range 20–93 | 20 | ECG | Leisure and occupational PA, four levels | Leisure-time PA: NS Occupational PA: NS | Leisure-time PA: NS Occupational PA: High vs. low PA: HR 1.21 (1.02–1.43) Very high vs. low PA: HR 1.39 (1.03–1.88) | Leisure-time PA: NS Occupational PA: J-shaped |
| Publication year, authors | Study population | Men (n) | Women (n) | Baseline age (years) | Follow-up (years, mean) | AF diagnosis | PA type, level, dose | RR (95% CI), moderate vs. sedentary PA | RR (95% CI), high vs. sedentary PA | Shape of relationship between PA and AF |
|---|---|---|---|---|---|---|---|---|---|---|
| 2008 Mozaffarian et al.8 | Cardiovascular Health Study, USA | 2287 | 3159 | Mean 73, range > 65 | 12 | ECG, hospital discharge | Leisure-time PA, quintiles, intensity, walking distance/pace | Walking distance, pace: RR 0.78 (0.65–0.94) PA quintile III vs. I: 0.75 (0.61–0.90) | PA quintile V vs. I: RR 0.64 (0.52–0.79). Intensity, high vs. none: RR 0.87 (0.64–1.19) | Walking: inverse, linear, p < 0.001. PA: inverse, linear, p < 0.001. Intensity: U-shape |
| 2009 Aizer et al.34 | Physicians’ Health Study, USA | 16921 | 0 | Range 40–84 | 12 | Self-reported | Vigorous exercise, days/week | 1–2 vs. <1 day/week: NS | 5–7 vs. <1 day/week: RR 1.2 (0.99–1.36) | Positive, linear association, p = 0.04 |
| 2011 Everett et al.30 | Women’s Health Study, USA | 0 | 34759 | Mean 54, range 49–59 | 14 | ECG, medical journal | Leisure-time PA, MET-h/week | Middle vs. low PA: NS | High vs. low PA: NS | NS |
| 2013 Williams and Franklin11 | National Runners’ Health Study II/National Walkers’ Health Study, UK | 19,044 | 27,763 | Mean 44–59 | 6 | Self-reported | Walking and running, MET-h/day | 2–3 vs. > 1 MET-h/day: RR 0.7, p < 0.05 | >3 vs. < 1 MET-h/day: RR 0.6, p < 0.001 | Inverse, linear |
| 2013 Thelle et al.33 | CONOR cohort, Norway | 147,462 | 162,078 | Range 40–45 | 5 | Norwegian Prescription Database, lone AF | Leisure and work PA, four levels | Moderate vs. sedentary: NS | Vigorous vs. sedentary: RR men 3.14 (2.17–4.54), women NS | Men: flipped L, increasing, p < 0.001 |
| 2014 Huxley et al.10 | Atherosclerosis Risk in Communities Study, USA | 6445 | 7774 | Mean 54, range 45–64 | 22 | ECG, hospital discharge | Leisure and work PA, MET-min/week | Middle vs. low PA: NS | High vs. low PA:HR 0.89 (0.80–1.00) | Inverse, linear, p = 0.06 |
| 2014 Azarbal et al.28 | Women’s Health Initiative Observational Study, USA | 0 | 81,317 | Range 50–79 | 11 | Medical records or Medicare | Walking, recreational, strenuous exercise, MET-h/week | 3–9 MET-h/week vs. none: NS | >9 MET-h/week vs. none: RR 0.90 (0.85–0.96) | Inverse, linear, p = 0.003 |
| 2014 Drca et al.29 | Swedish men | 44,410 | 0 | Mean 60, range 45–79 | 12 | Swedish Inpatient Register | Walking/bicycling; light-to-moderate. Exercise: moderate-to-high | Walking > 60 min/day vs. almost never: RR 0.87 (0.77–0.97) | Exercise > 5 h/week vs. almost never: RR 1.17 (1.03–1.32) | Walking: inverse, linear, p = 0.03. Exercise: positive, linear, p = 0.01 |
| 2014 Knuiman et al.31 | Busselton Health Study, Australia | 1861 | 2406 | Range 25–84 | 16 | Hospital admission | Doing any vigorous each week | Not assessed | Vigorous vs. none: HR 0.80 (0.62–1.03) | Not assessed |
| 2015 Drca et al.9 | Swedish mammography cohort | 0 | 36,513 | 60 (49–83) | 12 | Swedish Inpatient Register | Walking/bicycling: light-to-moderate. Exercise: moderate-to-high | Walking, ≥ 40 min/day vs. almost never: RR 0.81 (0.72–0.92) | Exercise > 5 vs. < 1 h/week: RR 0.85 (0.75–0.95) | Walking: inverse, linear, p > 0.001. Exercise: inverse, linear, p = 0.02 |
| 2015 Bapat et al.32 | Multi-ethnic study of atherosclerosis, USA | 2722 | 3071 | 62 (45–84) | 7 | Medical records or Medicare | Total intentional exercise, vigorous exercise, MET-min/week | Lower or middle tertile vs. none PA: NS | Upper tertile PA: NS | HR for AF increased with total intentional PA |
| 2015 Woodward et al.36 | Taupo bicycle study, New Zealand | 1874 | 725 | ≥16 | 7 | Hospital discharge | Cyclists vs. national cohort, h/week, km/week | No difference cyclists vs. general population | ||
| 2016 Morseth et al.7 | Tromsø Study, Norway | 10,300 | 10,184 | Mean 39 (men), 37 (women) | 20 | ECG | Leisure and work PA, four levels | Moderate vs. sedentary: HR 0.81 (0.68–0.96) | Vigorous vs. sedentary: NS | J-shape |
| 2016 Skielboe et al.35 | 7258 | 8560 | Range 20–93 | 20 | ECG | Leisure and occupational PA, four levels | Leisure-time PA: NS Occupational PA: NS | Leisure-time PA: NS Occupational PA: High vs. low PA: HR 1.21 (1.02–1.43) Very high vs. low PA: HR 1.39 (1.03–1.88) | Leisure-time PA: NS Occupational PA: J-shaped |
AF: atrial fibrillation; PA: physical activity; RR: relative risk; CI: confidence interval; ECG: electrocardiography; NS: not significant at p < 0.05; MET: metabolic equivalent of task; HR: hazard ratio.
Physical activity and risk of atrial fibrillation in the general population.
| Publication year, authors | Study population | Men (n) | Women (n) | Baseline age (years) | Follow-up (years, mean) | AF diagnosis | PA type, level, dose | RR (95% CI), moderate vs. sedentary PA | RR (95% CI), high vs. sedentary PA | Shape of relationship between PA and AF |
|---|---|---|---|---|---|---|---|---|---|---|
| 2008 Mozaffarian et al.8 | Cardiovascular Health Study, USA | 2287 | 3159 | Mean 73, range > 65 | 12 | ECG, hospital discharge | Leisure-time PA, quintiles, intensity, walking distance/pace | Walking distance, pace: RR 0.78 (0.65–0.94) PA quintile III vs. I: 0.75 (0.61–0.90) | PA quintile V vs. I: RR 0.64 (0.52–0.79). Intensity, high vs. none: RR 0.87 (0.64–1.19) | Walking: inverse, linear, p < 0.001. PA: inverse, linear, p < 0.001. Intensity: U-shape |
| 2009 Aizer et al.34 | Physicians’ Health Study, USA | 16921 | 0 | Range 40–84 | 12 | Self-reported | Vigorous exercise, days/week | 1–2 vs. <1 day/week: NS | 5–7 vs. <1 day/week: RR 1.2 (0.99–1.36) | Positive, linear association, p = 0.04 |
| 2011 Everett et al.30 | Women’s Health Study, USA | 0 | 34759 | Mean 54, range 49–59 | 14 | ECG, medical journal | Leisure-time PA, MET-h/week | Middle vs. low PA: NS | High vs. low PA: NS | NS |
| 2013 Williams and Franklin11 | National Runners’ Health Study II/National Walkers’ Health Study, UK | 19,044 | 27,763 | Mean 44–59 | 6 | Self-reported | Walking and running, MET-h/day | 2–3 vs. > 1 MET-h/day: RR 0.7, p < 0.05 | >3 vs. < 1 MET-h/day: RR 0.6, p < 0.001 | Inverse, linear |
| 2013 Thelle et al.33 | CONOR cohort, Norway | 147,462 | 162,078 | Range 40–45 | 5 | Norwegian Prescription Database, lone AF | Leisure and work PA, four levels | Moderate vs. sedentary: NS | Vigorous vs. sedentary: RR men 3.14 (2.17–4.54), women NS | Men: flipped L, increasing, p < 0.001 |
| 2014 Huxley et al.10 | Atherosclerosis Risk in Communities Study, USA | 6445 | 7774 | Mean 54, range 45–64 | 22 | ECG, hospital discharge | Leisure and work PA, MET-min/week | Middle vs. low PA: NS | High vs. low PA:HR 0.89 (0.80–1.00) | Inverse, linear, p = 0.06 |
| 2014 Azarbal et al.28 | Women’s Health Initiative Observational Study, USA | 0 | 81,317 | Range 50–79 | 11 | Medical records or Medicare | Walking, recreational, strenuous exercise, MET-h/week | 3–9 MET-h/week vs. none: NS | >9 MET-h/week vs. none: RR 0.90 (0.85–0.96) | Inverse, linear, p = 0.003 |
| 2014 Drca et al.29 | Swedish men | 44,410 | 0 | Mean 60, range 45–79 | 12 | Swedish Inpatient Register | Walking/bicycling; light-to-moderate. Exercise: moderate-to-high | Walking > 60 min/day vs. almost never: RR 0.87 (0.77–0.97) | Exercise > 5 h/week vs. almost never: RR 1.17 (1.03–1.32) | Walking: inverse, linear, p = 0.03. Exercise: positive, linear, p = 0.01 |
| 2014 Knuiman et al.31 | Busselton Health Study, Australia | 1861 | 2406 | Range 25–84 | 16 | Hospital admission | Doing any vigorous each week | Not assessed | Vigorous vs. none: HR 0.80 (0.62–1.03) | Not assessed |
| 2015 Drca et al.9 | Swedish mammography cohort | 0 | 36,513 | 60 (49–83) | 12 | Swedish Inpatient Register | Walking/bicycling: light-to-moderate. Exercise: moderate-to-high | Walking, ≥ 40 min/day vs. almost never: RR 0.81 (0.72–0.92) | Exercise > 5 vs. < 1 h/week: RR 0.85 (0.75–0.95) | Walking: inverse, linear, p > 0.001. Exercise: inverse, linear, p = 0.02 |
| 2015 Bapat et al.32 | Multi-ethnic study of atherosclerosis, USA | 2722 | 3071 | 62 (45–84) | 7 | Medical records or Medicare | Total intentional exercise, vigorous exercise, MET-min/week | Lower or middle tertile vs. none PA: NS | Upper tertile PA: NS | HR for AF increased with total intentional PA |
| 2015 Woodward et al.36 | Taupo bicycle study, New Zealand | 1874 | 725 | ≥16 | 7 | Hospital discharge | Cyclists vs. national cohort, h/week, km/week | No difference cyclists vs. general population | ||
| 2016 Morseth et al.7 | Tromsø Study, Norway | 10,300 | 10,184 | Mean 39 (men), 37 (women) | 20 | ECG | Leisure and work PA, four levels | Moderate vs. sedentary: HR 0.81 (0.68–0.96) | Vigorous vs. sedentary: NS | J-shape |
| 2016 Skielboe et al.35 | 7258 | 8560 | Range 20–93 | 20 | ECG | Leisure and occupational PA, four levels | Leisure-time PA: NS Occupational PA: NS | Leisure-time PA: NS Occupational PA: High vs. low PA: HR 1.21 (1.02–1.43) Very high vs. low PA: HR 1.39 (1.03–1.88) | Leisure-time PA: NS Occupational PA: J-shaped |
| Publication year, authors | Study population | Men (n) | Women (n) | Baseline age (years) | Follow-up (years, mean) | AF diagnosis | PA type, level, dose | RR (95% CI), moderate vs. sedentary PA | RR (95% CI), high vs. sedentary PA | Shape of relationship between PA and AF |
|---|---|---|---|---|---|---|---|---|---|---|
| 2008 Mozaffarian et al.8 | Cardiovascular Health Study, USA | 2287 | 3159 | Mean 73, range > 65 | 12 | ECG, hospital discharge | Leisure-time PA, quintiles, intensity, walking distance/pace | Walking distance, pace: RR 0.78 (0.65–0.94) PA quintile III vs. I: 0.75 (0.61–0.90) | PA quintile V vs. I: RR 0.64 (0.52–0.79). Intensity, high vs. none: RR 0.87 (0.64–1.19) | Walking: inverse, linear, p < 0.001. PA: inverse, linear, p < 0.001. Intensity: U-shape |
| 2009 Aizer et al.34 | Physicians’ Health Study, USA | 16921 | 0 | Range 40–84 | 12 | Self-reported | Vigorous exercise, days/week | 1–2 vs. <1 day/week: NS | 5–7 vs. <1 day/week: RR 1.2 (0.99–1.36) | Positive, linear association, p = 0.04 |
| 2011 Everett et al.30 | Women’s Health Study, USA | 0 | 34759 | Mean 54, range 49–59 | 14 | ECG, medical journal | Leisure-time PA, MET-h/week | Middle vs. low PA: NS | High vs. low PA: NS | NS |
| 2013 Williams and Franklin11 | National Runners’ Health Study II/National Walkers’ Health Study, UK | 19,044 | 27,763 | Mean 44–59 | 6 | Self-reported | Walking and running, MET-h/day | 2–3 vs. > 1 MET-h/day: RR 0.7, p < 0.05 | >3 vs. < 1 MET-h/day: RR 0.6, p < 0.001 | Inverse, linear |
| 2013 Thelle et al.33 | CONOR cohort, Norway | 147,462 | 162,078 | Range 40–45 | 5 | Norwegian Prescription Database, lone AF | Leisure and work PA, four levels | Moderate vs. sedentary: NS | Vigorous vs. sedentary: RR men 3.14 (2.17–4.54), women NS | Men: flipped L, increasing, p < 0.001 |
| 2014 Huxley et al.10 | Atherosclerosis Risk in Communities Study, USA | 6445 | 7774 | Mean 54, range 45–64 | 22 | ECG, hospital discharge | Leisure and work PA, MET-min/week | Middle vs. low PA: NS | High vs. low PA:HR 0.89 (0.80–1.00) | Inverse, linear, p = 0.06 |
| 2014 Azarbal et al.28 | Women’s Health Initiative Observational Study, USA | 0 | 81,317 | Range 50–79 | 11 | Medical records or Medicare | Walking, recreational, strenuous exercise, MET-h/week | 3–9 MET-h/week vs. none: NS | >9 MET-h/week vs. none: RR 0.90 (0.85–0.96) | Inverse, linear, p = 0.003 |
| 2014 Drca et al.29 | Swedish men | 44,410 | 0 | Mean 60, range 45–79 | 12 | Swedish Inpatient Register | Walking/bicycling; light-to-moderate. Exercise: moderate-to-high | Walking > 60 min/day vs. almost never: RR 0.87 (0.77–0.97) | Exercise > 5 h/week vs. almost never: RR 1.17 (1.03–1.32) | Walking: inverse, linear, p = 0.03. Exercise: positive, linear, p = 0.01 |
| 2014 Knuiman et al.31 | Busselton Health Study, Australia | 1861 | 2406 | Range 25–84 | 16 | Hospital admission | Doing any vigorous each week | Not assessed | Vigorous vs. none: HR 0.80 (0.62–1.03) | Not assessed |
| 2015 Drca et al.9 | Swedish mammography cohort | 0 | 36,513 | 60 (49–83) | 12 | Swedish Inpatient Register | Walking/bicycling: light-to-moderate. Exercise: moderate-to-high | Walking, ≥ 40 min/day vs. almost never: RR 0.81 (0.72–0.92) | Exercise > 5 vs. < 1 h/week: RR 0.85 (0.75–0.95) | Walking: inverse, linear, p > 0.001. Exercise: inverse, linear, p = 0.02 |
| 2015 Bapat et al.32 | Multi-ethnic study of atherosclerosis, USA | 2722 | 3071 | 62 (45–84) | 7 | Medical records or Medicare | Total intentional exercise, vigorous exercise, MET-min/week | Lower or middle tertile vs. none PA: NS | Upper tertile PA: NS | HR for AF increased with total intentional PA |
| 2015 Woodward et al.36 | Taupo bicycle study, New Zealand | 1874 | 725 | ≥16 | 7 | Hospital discharge | Cyclists vs. national cohort, h/week, km/week | No difference cyclists vs. general population | ||
| 2016 Morseth et al.7 | Tromsø Study, Norway | 10,300 | 10,184 | Mean 39 (men), 37 (women) | 20 | ECG | Leisure and work PA, four levels | Moderate vs. sedentary: HR 0.81 (0.68–0.96) | Vigorous vs. sedentary: NS | J-shape |
| 2016 Skielboe et al.35 | 7258 | 8560 | Range 20–93 | 20 | ECG | Leisure and occupational PA, four levels | Leisure-time PA: NS Occupational PA: NS | Leisure-time PA: NS Occupational PA: High vs. low PA: HR 1.21 (1.02–1.43) Very high vs. low PA: HR 1.39 (1.03–1.88) | Leisure-time PA: NS Occupational PA: J-shaped |
AF: atrial fibrillation; PA: physical activity; RR: relative risk; CI: confidence interval; ECG: electrocardiography; NS: not significant at p < 0.05; MET: metabolic equivalent of task; HR: hazard ratio.
Female athletes
The association between exercise and atrial fibrillation in women has also been examined in a cohort of 278 female veteran cross-country skiers with a mean age of 62 years. The prevalence of self-reported atrial fibrillation was 8% in female skiers, and women who had exercised for ≥ 40 years had a two-fold increased risk of atrial fibrillation compared with women who had never exercised regularly.23 These findings, along with results from Andersen et al.,27 indicate that prolonged endurance exercise might cause atrial fibrillation also among women.
The general population
While intensive endurance exercise seems to increase the risk of atrial fibrillation, the effect of lower levels of physical activity on atrial fibrillation risk seems ambivalent. Several population-based, prospective studies show a linear, decreasing trend in atrial fibrillation risk with increasing leisure-time physical activity,10,28 and walking or running time, pace or distance8,9,11,29 (Table 3). Some studies show a similar non-significant association.30–32 Overall, all of these studies show a linear trend towards lower atrial fibrillation risk with increasing physical activity, even at the highest physical activity level.
In contrast, some studies report a higher risk of atrial fibrillation among the most vigorous physically active individuals compared with the least active (Table 3). Thelle et al.33 found that vigorous physical activity was associated with a higher risk of lone atrial fibrillation, defined by a flecainide or sotalol prescription, in men but not in women aged 40–45 years. Aizer et al.34 observed that vigorous exercise 5–7 days/week showed a higher risk of atrial fibrillation compared with no vigorous activity. Morseth et al.7 found a J-shaped association between physical activity level and atrial fibrillation risk in a general population of men and women aged 30–67 years at baseline, although vigorous physical activity non-significantly increased the risk of atrial fibrillation compared with inactive individuals. Myrstad et al.24 confirmed the J-shaped relationship between physical activity and atrial fibrillation by combining data from a general population and endurance trained cross-country skiers. One of the few studies on occupational physical activity and atrial fibrillation showed an increased risk of atrial fibrillation with high physical activity, whereas leisure-time physical activity in this study was not related to risk of atrial fibrillation.35
The majority of studies showed a trend towards a lower atrial fibrillation risk with light and moderate physical activity, although significant only in a few studies,7,8,11 and no studies showed an increased risk with light or moderate physical activity (Table 3). The volume of physical activity necessary to decrease the risk of atrial fibrillation seems low. For example, Morseth et al.7 showed that the next lowest physical activity level, involving recreational walking, cycling or other forms of activity > 4 h/week, showed a 20% reduced risk of atrial fibrillation. Similarly, > 405 kcal per week of leisure-time physical activity reduced the atrial fibrillation risk by 25%,8 and walking at least 20 min/day seems to be protective against atrial fibrillation29 compared with the least active individuals.
Studies of physical activity and atrial fibrillation risk in the general population embrace large variations in physical activity levels, from vigorous exercise several days per week to mostly sedentary individuals, and although all studies used questionnaires to measure physical activity, the inquiries and categorizing vary widely. Depending on measurement instruments, prevalence of physical activity may vary largely between studies.
We therefore examined whether the results might vary according to study characteristics.
Gender
The majority of studies included both men and women, although three of the studies showing a linear, inverse relationship between physical activity and atrial fibrillation risk included women only.9,28,30 Two of the three studies on higher atrial fibrillation risk with high physical activity included both genders;7,33 however, two of these studies had too low power to examine vigorous physical activity and atrial fibrillation in women.7,33 Thus, it may seem premature to conclude that gender differences could explain these variations in results. However, a recent review concluded that both total physical activity and vigorous physical activity increased the risk of atrial fibrillation in men and decreased the risk in women, suggesting different risk patterns of atrial fibrillation with physical activity in men and women.37,38
Age and populations
Age ranges are rather similar across studies, as the majority of studies have included a baseline age range from 45 years and above. Although some studies have restricted recruitment to certain populations, such as physicians34 or runners and walkers,11 differences in population characteristics do not seem to explain the variation in results. We are not aware on any studies on physical activity, exercise and atrial fibrillation in non-Caucasians, but atrial fibrillation incidence has shown to be lower in people of Hispanic, African and Chinese origin than in Caucasians,39 and given that the higher incidence of atrial fibrillation in Caucasians may be related to factors such as larger left atrial diameter,40 studies on physical activity, exercise and atrial fibrillation in multi-ethnic populations are needed.
Measurements methods and physical activity levels
Studies have used varying definitions of physical activity and exercise, which could possibly influence the findings. Morseth et al.,7 Myrstad et al.24 and Thelle et al.33 used the same question on leisure-time physical activity, with four physical activity levels, ranging from mostly sedentary to vigorous exercise, defined as participation in hard training or sports competitions, regularly and several times per week.7,33 These studies showed an increased atrial fibrillation risk with vigorous exercise, although significant in two of the studies.24,33 Aizer et al.34 based their analyses on regular engagement in exercise vigorous enough to work up a sweat, and found an increased atrial fibrillation risk among those who exercised vigorously more than 4 days/week.
Studies showing a decreasing risk of atrial fibrillation with increasing physical activity8,9,11,28 define the highest physical activity as activity beyond a certain limit, mainly defined by energy expenditure (metabolic equivalent of task (MET) or kcal), which is not directly comparable to the above-mentioned studies. However, the highest level of physical activity in studies showing a reduced atrial fibrillation risk was defined as energy expenditure > 9 MET-h/week,28 > 3.4 MET-h/day,11 or > 1840 kcal/week as the highest category,8 which could translate into more modest exercise than in the studies showing an increased atrial fibrillation risk with vigorous exercise.7,24,33,34 However, although various definitions of physical activity levels may partly explain these ambiguous findings with vigorous exercise, the usefulness of physical activity quantity calculations is limited with self-reported physical activity and exercise. All existing studies rely on self-reported physical activity, which is prone to recall bias41 and overestimation,42 which may also vary with measurement instruments, physical activity levels and populations. Therefore, we look forward to studies using more objective and accurate measures to quantify volume and patterns of physical activity in relation to atrial fibrillation.
Mechanisms of atrial fibrillation and physical activity
Taken together, current research indicates a J-shaped relationship between physical activity and risk of atrial fibrillation in men (Figure 1). This J-shaped relationship has led to the hypothesis that different mechanisms could explain the higher risk of atrial fibrillation with prolonged endurance exercise on one hand, and the lower risk with moderate physical activity on the other hand43,44 (Figure 1).
Shape of the association between risk of atrial fibrillation and physical activity level, and proposed underlying mechanisms.
AF: atrial fibrillation
Potential mechanisms explaining an increased atrial fibrillation risk with endurance exercise
Cardiac adaptations to vigorous exercise include increased vagal tone, lower resting heart rate, and increased stroke volume, chamber dilatation and hypertrophy, better systolic and diastolic function, modified metabolism and electric characteristics.45–47 These adaptations, often expressed as athlete’s heart, are assumed to be physiological, reversible adaptations to the increased demands during exercise.48 However, the alterations have been shown to possibly increase the risk of atrial fibrillation, thereby contributing to the pathophysiological mechanisms linked to the increased atrial fibrillation risk with intensive and long-term endurance exercise.49 Several mechanisms have been proposed, but the exact mechanisms behind the observed increase in atrial fibrillation risk with intensive exercise remain speculative.
Autonomic activation
The autonomic nervous system seems to play a part in both initiating and maintaining atrial fibrillation,50–52 and canine experiments have shown that autonomic nerve stimulation in the atrium and pulmonary vein initiates arrhythmias.50 Both increased parasympathetic and increased sympathetic nervous system activity has been observed to facilitate atrial fibrillation in athletes, as atrial fibrillation may be triggered during exercise (adrenergically induced atrial fibrillation) and rest (vagally induced atrial fibrillation).53,54 Furlanello et al.13 showed that among young athletes with paroxysmal atrial fibrillation, atrial fibrillation occurred during exercise in more than 60% of the athletes. Mont et al.54 showed that vagally induced atrial fibrillation was more common in physically active (57%) than in sedentary atrial fibrillation patients (18%).
Prolonged exercise may lead to long-term changes in autonomic activation, which could expose athletes to increased risk of atrial fibrillation later in life.46 This may be particularly linked to increased vagal tone, as vagally induced atrial fibrillation seems to be more apparent in older athletes.53 Regular endurance exercise leads to increased vagal tone,46 and low resting heart rate is common among endurance athletes.26 Increased vagal activity reduces the atrial refractory period, facilitating re-entry, which may trigger atrial fibrillation.55 In a study of non-elite male endurance athletes (mean age 42 years), lifetime training volume > 4500 h was associated with prolongation of signal-averaged P-wave duration, higher parasympathetic tone and more frequent premature atrial contractions compared with lifetime training < 1500 h.46 In exercising rats, prevalence of atrial fibrillation increased after 16 weeks of daily exercise,47 and the exercising rats had significantly reduced resting heart rate and increased parasympathetic tone after 16 weeks of exercise, compared with sedentary rats. However, resting heart rate and vagal tone were normalized in the exercising rats after four weeks of detraining.47
Atrial dilatation
Left atrial size is a risk factor for atrial fibrillation,56 and cross-sectional studies show enlargements of the left atrium in athletes,14,46,57–63 which is also shown in experimental animal models.64 Longitudinal studies confirm left atrial remodelling after a period of high-volume exercise.65-67 Only a few studies have linked exercise and atrial dimensions to atrial fibrillation. A case–control study68 showed that atrial fibrillation patients had higher cumulated physical activity levels and larger left atrial dimensions than controls. In marathon runners6 and cross-country skiers, large left atrial dimension is associated with a higher risk of atrial fibrillation.26 However, a study of young, competitive athletes showed enlarged left atrial dimensions but low prevalence of atrial fibrillation (1%).14 Furthermore, atrial remodelling has shown to be reversible after detraining.13,64,69
Atrial dilatation due to long-term endurance training has been linked to increased volume and pressure overload during exercise. In animal experiments, raising atrial pressure, which induces atrial dilatation, leads to a shortening of the atrial refractory period and increases vulnerability to atrial fibrillation.70 Volume and pressure overload could therefore potentially trigger atrial fibrillation during exercise, and it is also speculated that repeated stretch of the atrial wall during extensive exercise over many years may lead to atrial enlargement, stretch-induced microtrauma, inflammation and fibrosis, which are potential substrates for atrial fibrillation.4,71,72
Atrial ectopy
Ectopic beats are an important trigger of atrial fibrillation, and some studies have shown that supraventricular premature beats are more common in athletes,46,57,73 suggesting that increased atrial ectopy could be responsible for the increased atrial fibrillation risk in athletes.74 However, a third study could not confirm these results.15
Fibrosis
Some animal studies and a few human studies have elucidated the role of fibrosis in exercise-induced atrial fibrillation.64,75–77 In rats, 16 weeks of exercise showed collagen deposition and increased fibrosis marker expression in the atria and ventricles, with an increased susceptibility to arrhythmia.64
In humans, studies of veteran athletes showed an increase in markers of myocardial fibrosis75 and a high prevalence of myocardial fibrosis shown on cardiac magnetic resonance imaging76 compared with sedentary controls. Similarly, Breuckmann et al.77 showed a three times higher prevalence of myocardial fibrosis in non-elite marathon runners than in controls. Whether this increased prevalence of fibrosis in athletes facilitates atrial fibrillation is uncertain.
Inflammation
Swanson78 postulated a hypothesis on inflammation as a possible mechanism linking exercise and atrial fibrillation. Long-term, intensive exercise has been reported to generate a systemic inflammatory response,79 which in turn may lead to development of atrial fibrillation, as C-reactive protein and Interleukin-6 levels can be elevated in atrial fibrillation patients.80–83 This is further supported by a study of elite mountain marathon runners showing transient acute inflammation, as well as conduction delay in the atria and increased atrial wall tension, after a mountain marathon race,84 and the authors speculated that these changes may lead to atrial remodelling and increased risk of atrial fibrillation.
Gastro-oesophageal acid reflux
Based on results from separate literature taken together, Swanson85 hypothesizes that vigorous exercise may induce gastro-oesophageal acid reflux,86–88 which in turn has been shown to increase the risk of atrial fibrillation by 39% in a large population study of 163,627 participants.89 This hypothesis needs further elaboration.
Performance enhancing drugs
Performance enhancing supplements such as anabolic steroids and stimulants may have an arrhythmogenic effect.90 Only a few case reports indicating increased atrial fibrillation risk with anabolic steroids have been reported,49 and due to the unknown extent of use of illegal performance enhancing drugs, the effects of supplements on atrial fibrillation risk are largely unknown.
Genetic predisposition
With a ∼10% prevalence of atrial fibrillation in most athlete studies, most athletes do not develop atrial fibrillation, introducing the question of whether some people are predisposed to atrial fibrillation. To date, a genome-wide association study (GWAS) has identified 26 loci that are associated with atrial fibrillation.91 The clinical significance of these GWAS variants and their ability to predict mechanisms of atrial fibrillation development are yet unknown, and the atrial fibrillation risk prediction only slightly improved by adding genetic information.92 To the best of the authors’ knowledge, no study has examined the association between genetic predisposition, exercise and atrial fibrillation.
Potential mechanisms explaining the lower atrial fibrillation risk with light–moderate physical activity
Cardiac adaptations to exercise are generally assumed to be beneficial, although vigorous physical activity, particularly prolonged endurance exercise, may lead to cardiac ‘overadaptation’ and (patho)physiological changes, thereby increasing the risk of atrial fibrillation. To what extent these adaptations also occur with light and moderate physical activity is not known, as there is only a handful studies on cardiac adaptations to more moderate physical activity. Dawes et al.93 examined self-reported physical activity in relation to cardiac remodelling in a healthy adult population. Using cardiac magnetic resonance imaging, the authors found that increasing physical activity levels were associated with greater left ventricular mass, ventricular end-diastolic volume, ventricular stroke volume, and slightly lower right and left ventricular ejection fractions. The study of Dawes et al. suggests that cardiac alterations occur with moderate physical activity or exercise between three and five hours per week, and the effects of physical activity on cardiac structure were greater than that of systolic blood pressure.93
In the Framingham Heart Study,94 which included middle-aged adults, higher levels of moderate–vigorous physical activity, although well below the doses of athletes, were associated with larger left atrial size and lower arterial stiffness. A recent study of 4342 elderly reported that higher levels of physical activity were associated with improved diastolic and systolic function but no clinically significant associations between physical activity and left atrial dimensions.95
The few studies on cardiac adaptations suggest that also moderate physical activity can lead to improved cardiac function. However, these alterations are probably modest and beneficial to the development of atrial fibrillation, as moderate physical activity is consistently associated with a reduced atrial fibrillation risk. Instead, some studies indicate that the lower atrial fibrillation risk may be linked to cardiovascular risk factors.8,10,30
Cardiovascular risk factors such as obesity, hypertension and type 2 diabetes mellitus increase the risk of atrial fibrillation,96–98 and physical activity modifies these cardiovascular risk factors,99 for example by improving weight, glucose and lipid control, and endothelial function, and lowering resting heart rate and blood pressure.100 Everett et al.30 and Mozaffarian et al.8 both showed that adjustment for risk factors such as body mass index (BMI), diabetes, hypertension, hyperlipidaemia and cardiovascular disease attenuated the association between physical activity and atrial fibrillation. Mozaffarian et al.8 discussed that the lower atrial fibrillation risk with increasing physical activity and walking in their study may be mediated in part by risk factors such as BMI, blood pressure, glucose, cholesterol and C-reactive protein levels. In the ARIC study,10 physical activity attenuated the increased atrial fibrillation risk associated with overweight and obesity, although only in men.
On the other hand, Drca et al.9 found that the inverse association between physical activity and risk of atrial fibrillation in women did not change with adjustment for risk factors such as BMI, hypertension, diabetes and cardiovascular disease.
Conclusion
Current research indicates a J-shaped relationship between physical activity, exercise and atrial fibrillation, but the findings need to be interpreted in view of methodological limitations. Existing research points to a wide range of underlying mechanisms, of which most remain to be established. Although still speculative, current research seems to substantiate the hypothesis that the mechanisms underlying an increased risk of atrial fibrillation with intensive exercise are different from those underlying a reduced risk with moderate physical activity, possibly linked to distinctive characteristics of the population under study. High volumes of exercise over many years are performed by lean, healthy endurance trained athletes, responding to intensive exercise by cardiac (patho)physiological alterations. The mechanisms likely involve autonomic nervous system influence, remodelling of the heart, and trigger areas in pulmonary veins and atria. Genetic polymorphisms are likely to a play a role, but evidence is lacking. The mechanisms underlying a reduced risk of atrial fibrillation with light–moderate physical activity may involve a distinctive pathway, as physical activity can potentially reduce the risk of atrial fibrillation through favourable effects on cardiovascular risk factors.
Author contribution
DST, BM and MLL contributed to the conception or design of the work. BM and MM contributed to the acquisition, analysis or interpretation of data for the work. BM drafted the manuscript. All authors critically revised the manuscript. All gave final approval and agree to be accountable for all aspects of work ensuring integrity and accuracy.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is supported by the Northern Norway Regional Health Authority.
Comments