International Journal of Cardiology 221 (2016) 10561060

Contents lists available at ScienceDirect

International Journal of Cardiology

journal homepage: www.elsevier.com/locate/ijcard

Review

Incidence of atrial brillation is associated with age and gender in

subjects practicing physical exercise: A meta-analysis and
meta-regression analysis
Natale Daniele Brunetti a,,1, Francesco Santoro a,b,1, Michele Correale c,1, Luisa De Gennaro d,1,
Giusy Conte a,1, Matteo Di Biase a,1
a
Department of Medical and Surgical Sciences, University of Foggia, Italy
b
Sankt Georg Asklepios Klinik, Hamburg, Germany
c
Cardiologia Universitaria, Ospedali Riuniti, Foggia, Italy
d
U.O. Cardiologia, Ospedale San Paolo, Bari, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The link between physical activity and the risk of atrial brillation (AF) remains controversial. We therefore
Received 25 May 2016 sought to further assess by a meta-analysis whether increased levels of physical activity may increase the risk
Accepted 8 July 2016 of AF.
Available online 09 July 2016
In October 2015, a PubMed research was conducted for studies that investigated this topic. We identied 11
relevant studies with a total of 81,787 participants. The pooled analysis did not show an increased risk of AF in
Keywords:
Atrial brillation
subjects practicing physical activity (odds ratio (OR) = 0.92, 95% C.I. = 0.841.01, p = 0.077, I2 = 90%).
Physical exercise However, given the observed large heterogeneity among studies, a subgroup analysis was performed in order to
Sport identify possible variables inuencing the risk of AF. Signicantly higher risk of AF in subjects with reported phys-
Meta-analysis ical activity was found in studies enrolling exclusively male participants (OR = 7.49, 95% C.I. = 3.1219.01,
Meta-regression p b 0.001, I2 = 0%) and subjects younger than 54 years (OR 5.30, 95% C.I. = 3.438.20, p b 0.001, I2 = 1.7%),
while results were opposite in studies enrolling male and/or female participants OR = 0.89 (95% C.I. = 0.81
0.97, p = 0.01) and subjects older than 54 years (OR = 0.84, 95% C.I. = 0.760.92, p b 0.001). A reverse correla-
tion was also found at meta-regression analysis between age and OR of AF (p = 0.047).
In conclusion, in our meta-analysis, there is a non-signicant trend toward lower risk of AF in subjects practicing
physical activity. The risk seems higher in male subjects. A reverse correlation between age and risk of AF seems
to be evident.
2016 Elsevier Ireland Ltd. All rights reserved.

1. Introduction rates of incidence of atrial brillation (AF) [4]. In a study by Karjalainen

Observational studies report an association between physical exer-
cise and a lower incidence of cardiovascular disease [1]; the hypothe-
sized mechanism responsible for such positive effect is thought to be
the modulation of physical exercise on principal cardiovascular risk fac-
tors and markers of inammation [2,3]. Nevertheless, some studies
raised the doubt that physical exercise may be linked with higher
Corresponding author at: Cardiology Department, University of Foggia, Viale Pinto n.1,
71100 Foggia, Italy.
E-mail address: natale.brunetti@unifg.it (N.D. Brunetti).
1
The institution where work was performed: Cardiology Department, University of
Foggia, Foggia, Italy.
et al., intense sport activity was associated with a 5-fold increased risk of
AF [5]. In another study by Molina et al., endurance sport practice (mar-
athon running) was associated with an 8.8-fold risk of incident AF in the
multivariate age- and blood pressure-adjusted Cox regression analysis
[6].
Meta-analysis studies yielded contrasting results. A rst meta-
analysis by Abdulla and Nielsen showed a higher risk of AF in athletes
compared with controls (odds ratio (OR) = 5.3) [7]. On the other
hand, more recent meta-analyses did not conrm such results and
found comparable incidence of AF in larger populations of subjects
practicing physical exercise, regardless of the intensity of such exercise
[8,9].
On the base of such data, we therefore aimed to further investigate in
a more comprehensive meta-analysis study the link between physical

http://dx.doi.org/10.1016/j.ijcard.2016.07.133
0167-5273/ 2016 Elsevier Ireland Ltd. All rights reserved.
 N.D. Brunetti et al. / International Journal of Cardiology 221 (2016) 10561060
exercise and AF, seeking to identify possible bias responsible for such
contrasting evidence.
2. Methods
2.1. Eligibility criteria
Studies assessing the risk of developing AF in subjects practicing physical exercise or
sport activity were selected for the meta-analysis. There was no strict denition of physical
activity and also studies including athletes and subjects practicing sport activity were not
excluded. Moreover, studies including subjects practicing exercise which had a compara-
tor group with less physical activity which would allow risk estimates to be calculated
were also included in the study. Studies were required to assess or follow-up participants
for AF. There was no restriction based on study design (e.g., retrospective, prospective),
study cohort and language of study report.
2.2. Search strategy
We searched PubMed with no date or language restriction in October 2015 using the
search terms atrial brillation AND physical exercise OR physical activity OR sport
OR athlete AND 0001/01/01[PDat]: 2015/09/01[PDat] AND 0001/01/01[PDat]:
2015/09/01[PDat]. We also checked the bibliographies of potentially relevant studies
and reviews for additional studies.
2.3. Study selection and data extraction
Two reviewers (NDB and GC) screened all titles and abstracts for studies that met the
inclusion criteria and excluded any articles that did not clearly fulll the selection criteria.
Full reports of potentially relevant studies were retrieved and studied and nal decisions
on inclusion or exclusion were made. We then independently double extracted data
from included studies on study design, study date, sample size, participant age, gender,
physical activity groups, follow-up, and results. We aimed to contact authors for clarica-
tion if there were any uncertainties.
2.4. Data analysis
We used Stata statistical software to conduct a xed effects meta-analysis using in-
verse variance method for pooled OR. Where possible, we chose to pool adjusted risk es-
timates from the primary studies; otherwise, we used raw outcome data to yield
unadjusted risk estimates. Furthermore, where multiple groups were reported, we chose
to pool the extreme (high intensity) or the lower exercise (low intensity) groups (i.e.,
the highest intensity or highest frequency of exercise compared to controls) because
this would be likely to improve the chance of detecting any potential association. We
planned to perform subgroup analysis based on mean age and gender of populations
enrolled in the studies.
Fig. 1. Studies selection ow-chart.
1057
2.5. Statistical heterogeneity
Statistical heterogeneity was assessed using I2 statistic, with I2 values of 30%60%
representing a moderate level of heterogeneity. If I2 was above 60% for the pooled analysis,
we sought to explore sources of heterogeneity in subgroups of studies.
We also assessed publication bias by plotting the OR against its standard error in a
funnel plot diagram and by the Egger regression asymmetry test.
2.6. Meta-regression analysis
A nal meta-regression analysis was performed with OR for developing AF as depen-
dent variable and mean age of patients enrolled in the studies as predictor.
3. Results
3.1. Study design and participant characteristics
A total of 11 studies were included in this meta-analysis, which in-
cluded 1 post hoc analysis of randomized controlled trials [10], 5 cohort
studies [5,1113], and 5 casecontrol studies [4,6,1417] with 81,787
subjects considered for the nal pooled analysis. The number of study
participants ranged from 57 to 57,000, and their mean ages ranged
from 43 years to 73 years. The process of study selection is shown in
Fig. 1.
The results of the meta-analysis using xed effects model are shown
in Fig. 2. The overall risk of AF using a xed effects model was not signif-
icantly lower in subjects practicing physical exercise than in controls:
OR = 0.92 (95% C.I. = 0.841.01, p = 0.077).
For heterogeneity, p b 0.0001 and I2 = 89.6% were signicant, indi-
cating fairly non-homogenous studies with signicant variation in
between studies, Egger's linear regression test was signicant (Egger
bias = 1.50 (95% CI = 0.292.80, p = 0.0193), and funnel plot showed
(Fig. 3) a large asymmetry and therefore a publication bias.
Subgroup analysis was therefore performed considering main
data retrievable from studies. When considering 4 studies enrolling
exclusively male gender participants [5,6,14,17], the OR of AF associ-
ated with physical exercise was signicantly higher (OR = 7.49, 95%
C.I. = 3.1219.01, p b 0.001, I2 = 0%) (Fig. 2). Results were different
when considering studies enrolling also female gender participants
1058 N.D. Brunetti et al. / International Journal of Cardiology 221 (2016) 10561060

Fig. 2. Physical exercise and risk of atrial brillation by gender.

[4,1013,15,16] with an OR = 0.89 (95% C.I. = 0.810.97, p = 0.01,
I2 = 91%).
Differences were also found when considering studies with
mean age of participants b 54 years [46,15] (OR = 5.30 (95%
C.I. = 3.438.20, p b 0.001, I 2 = 1.73%) (Fig. 4) vs those with a
mean age 54 years [1014, 16 17] (OR = 0.84, 95% C.I. = 0.76
0.92, p b 0.001).
A reverse linear correlation was found at meta-regression analysis
considering the mean age of subjects enrolled and the OR of developing
AF at follow-up (p = 0.047, Fig. 5).

Fig. 3. Physical activity and risk of atrial brillation by age.

 N.D. Brunetti et al. / International Journal of Cardiology 221 (2016) 10561060
Fig. 4. Funnel plot showing signicant asymmetry in distribution of studies.
4. Discussion
We showed in this study that physical activity may be associated
with a trend toward a lower risk of AF. Evidence, however, is inconclu-
sive, given a large heterogeneity in studies available on this topic and
consequent contrasting results in different populations.
Reasons for such contrasting data are unclear and may be barely hy-
pothesized. Statistical heterogeneity attributable to methodological dif-
ferences in study design, population, type of assessment of physical
activity or sports activity, ascertainment of AF, and timing of such eval-
uations may be partly considered responsible for controversial results.
Studies with positive associations between physical activity and AF
may be likely related to smaller sample sizes or poor methodological
quality.
Also the exact quantication of physical exercise is extremely poor
and variable among studies; some studies were based on self-reported
questionnaires, largely biased by individual perception; although fre-
quency and duration of exercise can be objective, the intensity of phys-
ical activity is more difcult to quantify.
Subgroup analysis, however, seems to suggest a possible bias played
by gender and age; the risk of developing AF seems to be increased in
studies enrolling subjects younger and exclusively male, while the link
between AF and physical exercise seems to be less tight with older
Fig. 5. Meta-regression analysis showing the link between mean age of study participants
and risk of atrial brillation at follow-up (p b 0.05).
1059
subjects and female populations (female, mixed male/female popula-
tions). However, studies assessing men and women separately did not
found that signicant difference for either gender [11].
We also showed a linear reverse correlation between mean age of
subjects enrolled and risk of developing AF. Drca et al. showed as AF is
associated with different levels of physical activity in different ages in
men [18]; data from the Swedish National Inpatient Register seem to
support the concept that leisure-time exercise at younger age is associ-
ated with increased risk of AF. Lower risk of AF in older subjects is partly
explained by higher rates of incidence of AF in the elderly, regardless of
physical activity.
Our ndings are actually in line with previous reviews by Ofman [8]
and Muller-Riemenschneider [19], which did not show signicant asso-
ciations between physical activity and increased risk of AF.
On the other hand, a review by Nielsen et al. including ten studies
concluded that long-term vigorous physical training is associated with
increased risk of AF [20], conrming another meta-analysis by the
same authors [7].
The mechanisms by which the risk of AF may be increased with
physical activity may possibly include increased parasympathetic tone,
left atrial enlargement, left ventricular hypertrophy and left ventricular
dilatation [21]. Increased vagal tone following regular physical exercise
may reduce atrial refractoriness and consequently lead to AF [22].
Nevertheless, studies supporting a possible association between AF
and physical activity in general are casecontrol in design and are of
poor methodological quality. Evidence coming from such studies is
therefore conditioned by important methodological bias.
The issue is therefore still matter of debate; however, we do believe
that our results may provide possible keys (gender and age differences)
to explain controversy in meta-analysis results. Surely, larger obser-
vational studies or even randomized studies (not easy to perform,
indeed) are warranted to nally clarify the point and assess the
real risk of AF associated with physical exercise and sport, or even
age and gender.
5. Conclusions
In our meta-analysis study, there is a non-signicant trend toward
lower risk of AF in subjects practicing physical activity. The risk of devel-
oping AF seems to be signicantly higher in the subgroup of studies en-
rolling exclusively male and younger subjects. A reverse correlation
between age and risk of AF in subjects practicing physical exercise
may be hypothesized.
6. Limitations
This meta-analysis was mostly performed on retrospective case
control studies; several studies enrolled an extremely reduced number
of patients or an extremely selected kind of athletes. Clinical character-
istics were not accessible for all studies. The extreme heterogeneity of
studies does not allow conclusive results.
Disclosures
No conict of interest to disclose.
References
[1] P.D. Thompson, D. Buchner, I.L. Pina, G.J. Balady, M.A. Williams, B.H. Marcus, K. Berra,
S.N. Blair, F. Costa, B. Franklin, G.F. Fletcher, N.F. Gordon, R.R. Pate, B.L. Rodriguez,
A.K. Yancey, N.K. Wenger, American Heart Association Council on Clinical Cardiology
Subcommittee on Exercise, Rehabilitation, and Prevention, American Heart Associa-
tion Council on Nutrition, Physical Activity, and Metabolism Subcommittee on
Physical Activity, Exercise and physical activity in the prevention and treatment of
atherosclerotic cardiovascular disease, Circulation 107 (2003) 31093116.
[2] S. Mora, N. Cook, J.E. Buring, P.M. Ridker, I.M. Lee, Physical activity and reduced risk
of cardiovascular events: potential mediating mechanisms, Circulation 116 (2007)
21102118.
1060 N.D. Brunetti et al. / International Journal of Cardiology 221 (2016) 10561060
[3] P.K. Myint, R.N. Luben, N.J. Wareham, A.A. Welch, S.A. Bingham, K.T. Khaw, Physical
activity and brinogen concentration in 23,201 men and women in the EPIC-Norfolk
population-based study, Atherosclerosis 198 (2008) 419425.
[4] R. Elosua, A. Arquer, L. Mont, A. Sambola, L. Molina, E. Garcia-Moran, J. Brugada, J.
Marrugat, Sports practice and the risk of lone AF: a casecontrol study, Int. J. Cardiol.
108 (2006) 332337.
[5] J. Karjalainen, U.M. Kujala, J. Kaprio, S. Sarna, M. Vittasalo, Lone AF in vigorously
exercising middle aged men: a casecontrol study, BMJ 316 (1998) 17841785.
[6] L. Molina, L. Mont, J. Marrugat, A. Berruezo, J. Brugada, J. Bruguera, C. Rebato, R.
Elosua, Long-term endurance sport practice increases the incidence of lone AF in
men: a follow up study, Europace 10 (2008) 618623.
[7] J. Abdulla, J.R. Nielsen, Is the risk of atrial brillation higher in athletes than in the
general population? A systematic review and meta-analysis, Europace 11 (2009)
11561159.
[8] P. Ofman, O. Khawaja, C.R. Rahilly-Tierney, A. Peralta, P. Hoffmeister, M.R. Reynolds,
J.M. Gaziano, L. Djousse, Regular physical activity and risk of atrial brillation: a
systematic review and meta-analysis, Circ. Arrhythm. Electrophysiol. 6 (2013)
252256.
[9] C.S. Kwok, S.G. Anderson, P.K. Myint, M.A. Mamas, Y.K. Loke, Physical activity and in-
cidence of atrial brillation: a systematic review and meta-analysis, Int. J. Cardiol.
177 (2014) 467476.
[10] B.M. Everett, D. Conen, J.E. Buring, M.V. Moorthy, I.M. Lee, C.M. Albert, Physical
activity and the risk of incident AF in women, Circ. Cardiovasc. Qual. Outcomes 4
(2011) 321327.
[11] L. Frost, P. Frost, P. Vestergaard, Work related physical activity and risk of a hospital
discharge diagnosis of AF or utter: the Danish Diet, Cancer, and Health Study,
Occup. Environ. Med. 62 (2005) 4953.
[12] H. Heidbuchel, W. Anne, R. Willems, B. Adriaenssens, F. Van de Werf, H. Ector,
Endurance sports is a risk factor for AF after ablation for atrial utter, Int. J. Cardiol.
107 (2006) 6772.
[13] D. Mozaffarian, C.D. Furberg, B.M. Psaty, D. Siscovick, Physical activity and incident
AF in older adults, Circulation 118 (2008) 800807.
[14] S. Baldesberger, U. Bauersfeld, R. Candinas, B. Seifert, M. Zuber, M. Ritter, R. Jenni, E.
Oechslin, P. Luthi, C. Scharf, B. Marti, C.H. Attenhofer Jost, Sinus node disease and ar-
rhythmias in the long term follow-up of former professional cyclist, Eur. Heart J. 29
(2008) 7178.
[15] L. Mont, D. Tamborero, R. Elosua, I. Molina, B. Coll-Vinent, S.M. SitgesM, B. Vidal, A.
Scalise, A. Tejeira, A. Berruezo, J. Brugada, GIRAFA (Grup Integrat de Recerca en
Fibril-laci Auricular) Investigators, Physical activity, height, and left atrial size are
independent risk factors for lone AF in middle-aged healthy individuals, Europace
10 (2008) 1520.
[16] M. Myrstad, J. Grimsmo, A.K. Gulsvik, A.H. Ranhoff, Self-reported AF in still active old
cross-country skiersthe Birkebeiner aging study, Eur. Heart J. 32 (2011) 182.
[17] F. van Buuren, K.P. Mellwig, L. Faber, C. Prinz, A. Fruend, J.B. Dahm, T. Kottmann, N.
Bogunovic, D. Horstkotte, T. Butz, C. Langer, The occurrence of AF in former top-level
handball players above the age of 50, Acta Cardiol. 67 (2012) 213220.
[18] N. Drca, A. Wolk, M. Jensen-Urstad, S.C. Larsson, Atrial brillation is associated with
different levels of physical activity levels in different ages in men, Heart 100 (2014)
10371042.
[19] F. Muller-Riemenschneider, F. Andersohn, S. Ernst, S.N. Willich, Association of
physical activity and AF, J. Phys. Act. Health 9 (2012) 605616.
[20] J.R. Nielson, K. Wachtell, J. Abdulla, The relationship between physical activity and
risk of AFa systematic review and meta-analysis, J. Atrial Fibrillation 5 (2013)
2027.
[21] A. Aizer, J.M. Gaziano, N.R. Cook, J.E. Manson, J.E. Buring, C.M. Albert, Relation of
vigorous exercise to risk of AF, Am. J. Cardiol. 103 (2009) 15721577.
[22] R. Alessie, M. Nusynowitz, J. Abildskow, G. Moe, Non-uniform distribution of vagal
effects on the atrial refractory period, Am. J. Phys. 194 (1958) 406410.