Key Points
Question
What is the incidence of atrial fibrillation in a large cohort of elite athletes and its association with potential risk factors?
Findings
In this 20-year study of 6813 athletes (35.0% women), with most sports disciplines represented, only 21 athletes (1 woman) had atrial fibrillation. Increasing values of age, years of competition, and left atrial anteroposterior diameter were the main factors associated with higher atrial fibrillation risk.
Meaning
The incidence of atrial fibrillation is low among young elite athletes, but potential contributors (eg, atrial dimensions) need to be monitored.
Abstract
Importance
Vigorous exercise (particularly endurance sports) might increase the risk of atrial fibrillation (AF), but there is variability in the reported frequency of this condition among elite athletes. The underlying pathophysiologic source remains unclear.
Objective
To determine AF incidence in a large cohort of elite athletes and its association with potential risk factors.
Design, Setting, and Participants
Retrospective observational cohort of all Spanish athletes (N = 6813) referred to a single center was used to determine AF incidence from January 1, 1997, to December 31, 2017, and cross-sectional analysis was conducted to compare athletes with and without reported AF. The cohort covered most sports disciplines, and the studied athletes were on national teams and competed in major international events. Cardiologists responsible for echocardiographic assessment were not blinded to the condition (AF or no AF) of the athletes.
Exposures
All participants underwent at least 1 cardiologic evaluation, including assessment at time of AF diagnosis in those with this condition.
Main Outcomes and Measures
Diagnosis of AF based on resting and/or exercise electrocardiogram, and/or 24-hour Holter monitoring and echocardiography-assessed atrial dimensions.
Results
A total of 6813 Spanish elite athletes (2385 [35.0%] women) were referred for cardiac evaluation during the study period. Mean (SD) age was 22 (7) years, and mean (SD) time of competition was 8 (5) years. Only 21 athletes (1 woman), participating in different types of sports, had AF (ie, paroxysmal [n = 18], persistent [n = 1], or long-standing persistent [n = 2]) during the 20-year study. In multivariate analysis, increasing values of age (odds ratio [OR], 1.07; 95% CI, 1.00-1.14), years of competition (OR, 1.14; 95% CI, 1.07-1.22), and left atrial anteroposterior diameter (OR, 1.21; 95% CI, 1.10-1.32) were associated with higher AF risk.
Conclusions and Relevance
The incidence of AF is low among young Spanish elite athletes, even when considering only endurance athletes. Yet, potential contributors (particularly atrial remodeling) need to be monitored.
This study examines the incidence of atrial fibrillation in a cohort of elite athletes.
Introduction
Moderate-intensity physical exercise decreases the risk of cardiovascular disease and cardiac events through various mechanisms, including cardioprotection against ischemia1 and a decreased risk of malignant arrhythmias.2 However, the potential benefits of very high exercise loads are more questionable, with strenuous exercise, particularly endurance sports, potentially increasing the risk of atrial fibrillation (AF).3,4 There is, nevertheless, variability in the reported frequency of AF among athletes.5 The pathophysiologic factors associated with this condition remain unclear, with training-induced left atrial (LA) remodeling thought of as a potential contributor.5 We therefore studied AF incidence in a large cohort of elite athletes and its association with potential risk factors, such as atrial dimension.
Methods
We followed a single-center, retrospective, observational cohort design to determine AF incidence in all Spanish athletes who were referred for 1 or more evaluation (including M-mode and 2-dimensional echocardiography) in the Department of Cardiology, Sports Medicine Center, Spanish Sports Health Protection Agency, Madrid, Spain, from January 1, 1997, to December 31, 2017, and meeting the criteria of being on national teams and competing in major international events. All athletes with AF were evaluated on occurrence of this condition. The protocol was approved by the local ethics committee (Universidad Europea Miguel de Cervantes-Comité de Ética de la Investigación) and conformed to the Declaration of Helsinki.6 The participants provided oral consent.
We also performed a cross-sectional comparison of the characteristics (including echocardiography) of athletes showing or not showing AF. To this end, we used the data obtained on AF diagnosis (AF athletes) and those corresponding to the last available assessment (non-AF athletes). We performed a within-participant longitudinal comparison of baseline echocardiographic characteristics and those corresponding to AF occurrence in athletes in the AF group with complete data from 2 or more serial evaluations.
Diagnosis of AF was based on resting and/or exercise electrocardiogram, and/or 24-hour Holter monitoring. Regarding the latter, to screen for potential AF cases, all athletes with 1 or more potentially suspicious cardiac signs (resting heart rate <40 beats/min, frequent extrasystoles in resting/exercise electrocardiogram) underwent 24-hour Holter monitoring during the study period (n = 1505 assessments). All echocardiography evaluations (Philips Sonos 7500 system; Advanced Diagnostics) were conducted to determine right superoinferior diameter (RA.Sid), LA.SId, LA anteroposterior diameter (LA.APd), and sphericity index (LA.APd/LA.SId ×100). Cardiologists responsible for echocardiographic assessment were not blinded to the condition (AF or no AF) of the athletes.
Statistical Analysis
Fisher exact and unpaired, 2-tailed t tests for categorical and continuous variables, respectively, were used to compare the group of athletes with AF and their non-AF peers. A nonparametric Mann-Whitney test was used to compare baseline echocardiography data and those obtained on AF occurrence in athletes showing AF and having complete echocardiography data from 2 or more evaluations. Logistic regression analyses (univariate and multivariate, stepwise forward) were performed to determine the association between risk of AF and potential contributing factors (age, years of competition, participation in endurance sports, atrial dimensions). Receiver operating characteristic curves were calculated to determine potential cutoff points for factors associated with AF risk. The level of statistical significance was set at P < .05, and all analyses were performed with the Stata Statistical Software Package for Macintosh, version 13 (StataCorp).
Results
The cohort comprised 6813 athletes (mean [SD] age, 22 [7] years, with a mean [SD] time of 8 [5] years of competition, and included 2385 [35.0%] women) participating in most sports disciplines, including 28% endurance (eg, running, triathlons), 47% mixed (eg, team sports), and 25% power/sprint sports (eg, sprint/throwing/jumping track and field events, weightlifting) (Table 1). The mean (SD) total number of evaluations per athlete was 3 (3), with 4306 (63.2%) individuals in the cohort undergoing 2 or more complete evaluations. The mean (SD) number of evaluations in the AF group was 3 (1), with 9 athletes evaluated only once.
Table 1. Main Sports and Demographic Characteristics by Group.
| Sport Discipline | Athletes With AF, Mean (SD) (n = 21) | Athletes Without AF, Mean (SD) (n = 6792) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No. | Age, y | BP, mm Hg | Height, cm | Weight, kg | BSA | No. | Age, y | BP, mm Hg | Height, cm | Weight, kg | BSA | |
| Aerobics | 0 | NA | NA | NA | NA | NA | 252 | 24 (5) | 117 (11)/67 (6) | 172 (13) | 74 (18) | 1.9 (0.3) |
| Archery | 0 | NA | NA | NA | NA | NA | 210 | 22 (7) | 119 (13)/64 (8) | 174 (15) | 72 (19) | 1.8 (0.3) |
| Badminton | 0 | NA | NA | NA | NA | NA | 110 | 22 (7) | 119 (12)/64 (7) | 171 (12) | 77 (86) | 1.8 (0.6) |
| Basketball | 5 | 35 (9) | 128 (10)/70 (9) | 183 (12) | 84.5 (16.3) | 2.3 (0.1) | 391 | 22 (7) | 119 (13)/64 (8) | 180 (15) | 75 (15) | 1.8 (0.3) |
| Beach volleyball | 0 | NA | NA | NA | NA | NA | 46 | 22 (8) | 114 (10)/67 (6) | 176 (16) | 66 (20) | 1.8 (0.3) |
| Boxing | 0 | NA | NA | NA | NA | NA | 136 | 22 (6) | 119 (11)/64 (8) | 174 (10) | 68 (15) | 1.8 (0.2) |
| Canoeing | 3 | 41 (11) | 128 (8)/75 (0) | 178 (2) | 97 (7) | 2.2 (0.1) | 276 | 21 (6) | 119 (11)/65 (8) | 173 (13) | 70 (17) | 1.8 (0.2) |
| Cycling | 0 | NA | NA | NA | NA | NA | 210 | 21 (6) | 119 (12)/65 (8) | 172 (19) | 66 (15) | 1.8 (0.3) |
| Motor driving | 0 | NA | NA | NA | NA | NA | 42 | 20 (5) | 121 (8)/66 (4) | 174 (6) | 66 (3) | 1.8 (0.01) |
| Fencing | 0 | NA | NA | NA | NA | NA | 249 | 22 (7) | 125 (9)/64 (7) | 175 (11) | 69 (15) | 1.8 (0.2) |
| Golf | 1a | 18 | 125 (70) | 184 | 87 | 2.1 | 217 | 22 (7) | 118 (12)/64 (9) | 172 (23) | 69 (14) | 1.8 (0.3) |
| Gymnastics | 0 | NA | NA | NA | NA | NA | 232 | 22 (7) | 119 (11)/65 (8) | 175 (11) | 73 (52) | 1.9 (0.4) |
| Handball | 0 | NA | NA | NA | NA | NA | 175 | 22 (6) | 120 (13)/70 (8) | 175 (11) | 70 (14) | 1.8 (0.2) |
| Field Hockey | 0 | NA | NA | NA | NA | NA | 210 | 22 (6) | 119 (12)/64 (8) | 174 (11) | 69 (16) | 1.8 (0.2) |
| Jet ski | 0 | NA | NA | NA | NA | NA | 38 | 23 (11) | 120 (17)/66 (9) | 174 (9) | 70 (25) | 1.8 (0.3) |
| Judo | 0 | NA | NA | NA | NA | NA | 503 | 22 (6) | 119 (12)/64 (8) | 174 (12) | 69 (16) | 1.8 (0.3) |
| Karate | 0 | NA | NA | NA | NA | NA | 114 | 23 (7) | 120 (12)/64 (8) | 174 (11) | 68 (17) | 1.8 (0.3) |
| Kickboxing | 0 | NA | NA | NA | NA | NA | 55 | 19 (6) | 121 (12)/64 (8) | 172 (12) | 65 (11) | 1.8 (0.2) |
| Motorcycling | 0 | NA | NA | NA | NA | 51 | 21 (8) | 123 (12)/69 (9) | 168 (11) | 59 (15) | 1.7 (0.3) | |
| Mountaineering/trail running | 1b | 56 | 135 (80) | 189 | 97 | 2.3 | 51 | 22 (4) | 125 (6)/60 (10) | 177 (12) | 71 (8) | 1.9 (0.2) |
| Paddle | 0 | NA | NA | NA | NA | NA | 105 | 22 (6) | 117 (17)/66 (7) | 168 (23) | 67 (16) | 1.7 (0.3) |
| Rowing | 1 | 42 | 125 (75) | 191 | 88 | 2.2 | 141 | 22 (8) | 118 (12)/62 (8) | 175 (13) | 62 (8) | 1.8 (0.2) |
| Rugby | 0 | NA | NA | NA | NA | NA | 201 | 22 (6) | 121 (12)/66 (9) | 175 (10) | 69 (15) | 1.8 (0.2) |
| Sailing | 0 | NA | NA | NA | NA | NA | 75 | 20 (4) | 114 (10)/64 (8) | 175 (10) | 69 (15) | 1.8 (0.2) |
| Skating | 0 | NA | NA | NA | NA | NA | 64 | 23 (7) | 121 (13)/63 (7) | 173 (7) | 72 (16) | 1.9 (0.2) |
| Skiing (alpine) | 0 | NA | NA | NA | NA | NA | 115 | 22 (7) | 118 (11)/64 (8) | 175 (12) | 72 (43) | 1.8 (0.4) |
| Snowboard | 0 | NA | NA | NA | NA | NA | 44 | 21 (6) | 118 (8)/68 (6) | 176 (13) | 65 (12) | 1.8 (0.2) |
| Soccer (outdoor) | 2 | 42 (21) | 130 (65) | 182 (3) | 83 (1) | 2.0 (0.0) | 314 | 24 (8) | 117 (11)/65 (8) | 174 (10) | 70 (13) | 1.8 (0.2) |
| Soccer (indoor) | 0 | NA | NA | NA | NA | NA | 95 | 24 (8) | 117 (11)/65 (8) | 174 (10) | 70 (13) | 1.8 (0.2) |
| Squash | 0 | NA | NA | NA | NA | NA | 45 | 19 (3) | 116 (12)/59 (5) | 179 (13) | 71 (15) | 1.9 (0.3) |
| Swimming | 0 | NA | NA | NA | NA | NA | 112 | 21 (6) | 120 (12)/65 (7) | 181 (10) | 72 (19) | 1.9 (0.4) |
| Taekwondo | 0 | NA | NA | NA | NA | NA | 113 | 21 (6) | 120 (10)/64 (8) | 173 (17) | 70 (19) | 1.9 (0.3) |
| Table tennis | 0 | NA | NA | NA | NA | NA | 75 | 22 (8) | 120 (10)/65 (6) | 172 (13) | 65 (17) | 1.8 (0.3) |
| Tennis | 0 | NA | NA | NA | NA | NA | 80 | 19 (5) | 121 (11)/65 (7) | 175 (15) | 67 (15) | 1.8 (0.3) |
| Track and field (long-distance running) | 4 | 43 (16) | 125 (9)/75 (7) | 175 ± 4 | 75.3 ± 4.1 | 1.9 ± 0.1 | 362 | 21 (5) | 118 (11)/65 (8) | 173 (18) | 63 (19) | 1.7 (0.3) |
| Track and field (middle-distance running) | 1 | 45 | 125 (75) | 180 | 82 | 2.0 | 192 | 22 (6) | 119 (11)/63 (8) | 173 (13) | 69 (18) | 1.8 (0.2) |
| Track and field (sprint/throwing/jumping events) | 1 | 45 | 130 (75) | 185 | 79 | 2.0 | 414 | 22 (7) | 119 (14)/65 (8) | 173 (14) | 70 (17) | 1.8 (0.2) |
| Triathlon | 2 | 38 (1.4) | 115 (14)/73 (4) | 174 (15) | 70 (24) | 1.8 (0.4) | 95 | 22 (6) | 120 (11)/64 (8) | 174 (14) | 70 (18) | 1.8 (0.3) |
| Volleyball | 0 | NA | NA | NA | NA | NA | 43 | 22 (6) | 119 (13)/65 (10) | 175 (11) | 67 (11) | 1.8 (0.2) |
| Water polo | 0 | NA | NA | NA | NA | NA | 102 | 21 (5) | 122 (12)/68 (10) | 173 (11) | 67 (15) | 1.8 (0.2) |
| Weightlifting | 0 | NA | NA | NA | NA | NA | 190 | 22 (7) | 120 (12)/66 (9) | 174 (13) | 77 (20) | 1.9 (0.3) |
| Wrestling | 0 | NA | NA | NA | NA | NA | 196 | 22 (7) | 120 (12)/66 (9) | 174 (13) | 71 (20) | 1.8 (0.3) |
| Wushu | 0 | NA | NA | NA | NA | NA | 51 | 21 (5) | 119 (11)/63 (7) | 176 (10) | 69 (13) | 1.8 (0.2) |
Abbreviations: AF, atrial fibrillation; BP, blood pressure; BSA, body surface area; NA, not applicable.
Atrial fibrillation associated with thyroiditis.
Sole woman in the AF cohort.
Only 21 athletes (1 woman), participating in distance running, basketball (n = 5 each), canoeing (n = 3), triathlon, soccer (n = 2 each), and rowing, mountaineering, golf, or track and field sprint events (n = 1 each), had AF (ie, paroxysmal [n = 18], persistent [n = 1], or long-standing persistent [n = 2]) during the 20-year study period. Diagnosis was based on resting (n = 13) and/or exercise electrocardiogram (n = 5), and/or 24-hour Holter monitoring (n = 15). Except for 1 athlete, all of the cases were symptomatic, usually in the form of palpitations at rest or after exertion with or without reported fatigue or performance decrements. One case was not attributable to sports practice (thyroiditis related) and no case has led to major clinical consequences.
Athletes with reported AF showed higher (all P < .001) values than their non-AF peers for male sex (100% vs 65%, respectively), mean (SD) age (38 [13] vs 22 [6] years), years of competition (21 [11] vs 8 [5]), RA.SId (59 [8] vs 52 [6] mm), LA.Sid (57 [9] vs 51 [6] mm), LA.APd (43 [9] vs 34 [5] mm), and sphericity index (77% [13%] vs 67% [9%]). The proportion of endurance athletes was also higher in the AF group (11 [52.4%] vs 1725 [25.4%], P = .003). Logistic regression results are presented in Table 2. Increasing values of age, years of competition (athletes competing in masters categories), and LA.APd were associated with higher AF risk in multivariate analysis.
Table 2. Associations With AF Risk in Spanish Elite Athletes.
| Variable | Logistic Regression, OR (95% CI) | Association Cutoff Value | Receiver Operator Curves for Cutoff Values | |||
|---|---|---|---|---|---|---|
| Univariate | Multivariate, Stepwise Forward | AUC (95% CI) | Sensitivity, Specificity, % | P Value | ||
| Sex (male) | 11.02 (1.48-82.23)a | NS | ||||
| Age | 1.18 (1.14-1.22)b | 1.07 (1.00-1.14)c | 27 y | 0.87 (0.76-0.97) | 85, 81 | <.001 |
| Years of competition | 1.15 (1.11-1.19)b | 1.14 (1.07-1.22)b | 14 y | 0.86 (0.76-0.96) | 80, 86 | <.001 |
| Endurance sport | 2.58 (1.10-6.11)c | NS | ||||
| RA.Sid | 1.20 (1.13-1.29)b | NS | 55 mm | 0.77 (0.65-0.89) | 75, 72 | <.001 |
| LA.Sid | 1.16 (1.09-1.25)b | NS | 54 mm | 0.70 (0.56-0.84) | 71, 70 | .001 |
| LA.APd | 1.32 (1.22-1.43)b | 1.21 (1.10-1.32)b | 41 mm | 0.83 (0.71-0.95) | 70, 92 | <.001 |
| Left atrial sphericity indexd | 1.10 (1.06-1.14)b | NS | 79% | 0.72 (0.59-0.85) | 50, 89 | .001 |
Abbreviations: AF, atrial fibrillation; AUC, area under the curve; LA.APd, left atrial anteroposterior diameter; LA.SId, left atrial superoinferior diameter; NS, nonsignificant in the model; OR, odds ratio; RA.Sid, right superoinferior diameter.
P =.02
P < .001.
P = .03.
LA.APd/LA.SId × 100.
In athletes in the AF group with 2 or more evaluations (n = 12), no significant differences were found between baseline assessment and the assessment corresponding to AF occurrence for mean (SD) values of RA.SId (58 [10] vs 61 [7] mm, respectively, P = .07), LA.SId (57 [6] vs 59 [7] mm, P = .45), LA.APd (43 [6] vs 45 [9] mm, P = .30), and sphericity index (77% [13%] vs 77% [12%], P = .85).
Discussion
The incidence of AF among young Spanish elite athletes was low (0.3%) as expected in young people, and nearly identical to that reported in previous research with young elite athletes (ie, 0.3% in Italian athletes aged 24 years on average).7 The fact that the reported prevalence of AF among athletes is variable, ranging from 0.3% to 12.8%, likely reflects methodologic biases, such as differences between studies in the participants’ age (which was lower in our participants than in most research in the field), training status, or sports specialty.8
Increased arterial pressure during exercise, together with atrial wall stretching, such as that potentially induced by long-term exposure to strenuous endurance exercise, might lead to microtrauma, inflammation, and fibrosis, which are arrhythmogenic.5 Regarding LA enlargement, our data are in overall agreement with those from a healthy, nonathletic Spanish population,9 suggesting that LA remodeling toward sphericity (ie, greater enlargement of the anteroposterior relative to the longitudinal diameter) is associated with an increased AF risk. In athletes, LA enlargement might increase proportionally with lifetime training hours or within competition level,5 with previous research7 showing that approximately 20% of young competitive athletes (mean age, 24 years) have an LA.APd greater than 40 mm vs 10% in the present study. In this regard, although increased LA size in athletes is overall considered a physiologic, nonpathologic cardiac adaptation,5 our data suggest that follow-up of LA dimensions might be useful in young athletes with high LA.APd values.
Limitations and Strengths
Our findings are limited by the retrospective design and the population studied, which included only athletes of Spanish nationality. Further research is needed with 3-dimensional cardiac imaging techniques, allowing for better geometric LA characterization. In addition, we might have underestimated AF incidence by missing some silent cases or those of athletes unwilling to recognize their condition, despite relying on several diagnostic methods (resting and exercise electrocardiography and 24-Holter monitoring). In this regard, the best method of AF screening remains a matter of debate, with AF being more infrequent, especially in young adults, and difficult to identify than other cardiovascular diseases (eg, hypertension) where screening is routinely applied.10 Furthermore, given the often paroxysmal nature of the arrhythmia, especially in otherwise healthy athletes, such as those studied herein, any screening method that uses a single rhythm snapshot, irrespective of the technique used, is unlikely to detect individuals with AF types other than persistent or high-burden paroxysmal AF.11
Strengths of our study are the large elite athletes cohort, including many women, who have been rarely studied in research on sports-related AF and numerous sports disciplines, and identification of cutoff values for potential contributors to AF risk (notably, LA.Apd ≥41 mm).
Conclusions
In contrast with previous research on more veteran athletes, particularly endurance athletes, in whom AF might occur more frequently than in the age-matched general population,8 the incidence of AF is low among young elite athletes, even when considering only those engaging in endurance sports. However, potential contributors, particularly atrial remodeling, need to be monitored. Future research might also determine the potential long-term consequences of AF occurrence in young athletes.
References
- 1.Thijssen DHJ, Redington A, George KP, Hopman MTE, Jones H. Association of exercise preconditioning with immediate cardioprotection: a review. JAMA Cardiol. 2018;3(2):169-176. doi: 10.1001/jamacardio.2017.4495 [DOI] [PubMed] [Google Scholar]
- 2.Fiuza-Luces C, Santos-Lozano A, Joyner M, et al. . Exercise benefits in cardiovascular disease: beyond attenuation of traditional risk factors [published online August 16, 2018]. Nat Rev Cardiol. doi: 10.1038/s41569-018-0065-1 [DOI] [PubMed] [Google Scholar]
- 3.Thompson PD. Physical fitness, physical activity, exercise training, and atrial fibrillation: first the good news, then the bad. J Am Coll Cardiol. 2015;66(9):997-999. doi: 10.1016/j.jacc.2015.07.003 [DOI] [PubMed] [Google Scholar]
- 4.Du X, Dong J, Ma C. Is atrial fibrillation a preventable disease? J Am Coll Cardiol. 2017;69(15):1968-1982. doi: 10.1016/j.jacc.2017.02.020 [DOI] [PubMed] [Google Scholar]
- 5.Sanchis-Gomar F, Perez-Quilis C, Lippi G, et al. . Atrial fibrillation in highly trained endurance athletes—description of a syndrome. Int J Cardiol. 2017;226:11-20. doi: 10.1016/j.ijcard.2016.10.047 [DOI] [PubMed] [Google Scholar]
- 6.World Medical Association World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):2191-2194. [DOI] [PubMed] [Google Scholar]
- 7.Pelliccia A, Maron BJ, Di Paolo FM, et al. . Prevalence and clinical significance of left atrial remodeling in competitive athletes. J Am Coll Cardiol. 2005;46(4):690-696. doi: 10.1016/j.jacc.2005.04.052 [DOI] [PubMed] [Google Scholar]
- 8.Wernhart S, Halle M. Atrial fibrillation and long-term sports practice: epidemiology and mechanisms. Clin Res Cardiol. 2015;104(5):369-379. doi: 10.1007/s00392-014-0805-0 [DOI] [PubMed] [Google Scholar]
- 9.Bisbal F, Alarcón F, Ferrero-de-Loma-Osorio A, et al. . Left atrial geometry and outcome of atrial fibrillation ablation: results from the multicentre LAGO-AF study. Eur Heart J Cardiovasc Imaging. 2018;19(9):1002-1009. doi: 10.1093/ehjci/jey060 [DOI] [PubMed] [Google Scholar]
- 10.Passman R, Piccini J. Electrocardiography screening for atrial fibrillation: we can do better [published online August 7, 2018]. JAMA Cardiol. doi: 10.1001/jamacardio.2018.220010 [DOI] [PubMed] [Google Scholar]
- 11.Charitos EI, Stierle U, Ziegler PD, et al. . A comprehensive evaluation of rhythm monitoring strategies for the detection of atrial fibrillation recurrence: insights from 647 continuously monitored patients and implications for monitoring after therapeutic interventions. Circulation. 2012;126(7):806-814. doi: 10.1161/CIRCULATIONAHA.112.098079 [DOI] [PubMed] [Google Scholar]