Efficacy of exercise intervention for atrial fibrillation: a systematic review and meta-analysis Authors: Qiaoyan Ma, Lizhong Wang Published: 2025 ABSTRACT BackgroundThis meta-analysis examined the efficacy of exercise interventions for atrial fibrillation (AF).MethodsPubMed, Embase, and the Cochrane Library databases were searched for potentially eligible randomized controlled trials (RCTs) from inception to Nov 2024. Patients were diagnosed with AF. The intervention group received exercise management. The control group received standard AF management. The outcomes were the Short Form 36 Health Survey Questionnaire (SF-36), heart rate at rest (HR), 6-min walk test (6MWT), peak oxygen uptake (VO2 peak), and the changes in VO2 peak. The RCTs were evaluated according to the Cochrane risk bias tool.ResultsSixteen RCTs (593 and 556 participants in the exercise and control groups) were included. Regarding the SF-36, the exercise intervention did not improve the general health (WMD = 4.10, 95%CI: -3.17, 11.37, P = 0.269; I2 = 86.4%, Pheterogeneity<0.001), physical function (PF) (WMD = 7.50, 95%CI: -5.78, 22.77, P = 0.268; I2 = 98.8%, Pheterogeneity<0.001), and vitality (WMD = 4.45, 95%CI: -1.80, 10.69, P = 0.163; I2 = 80.0%, Pheterogeneity<0.001) scores. The exercise intervention did not improve the HR at rest (WMD=-1.71, 95%CI: -5.23, 1.82, P = 0.343; I2 = 61.6%, Pheterogeneity=0.016). The exercise intervention improved the 6MWT (WMD = 66.27, 95%CI: 16.49, 116.04, P = 0.009; I2 = 77.4%, Pheterogeneity=0.001), VO2 peak (WMD = 3.08, 95%CI: 0.88, 5.29, P = 0.006; I2 = 63.3%, Pheterogeneity=0.018) and changes in VO2 peak (WMD = 3.78, 95%CI: 2.16, 5.39, P < 0.001; I2 = 0.0%, Pheterogeneity=0.615) in patients with AF. No publication bias was observed. The sensitivity analyses suggested that the results were robust.ConclusionsIn patients with AF, an exercise intervention appears to improve the 6MWT and VO2 peak, but not resting HR and overall health-related quality of life. However, subgroup analyses indicated that exercise significantly improved the vitality (VT) subscore of the SF-36 in specific patient populations, suggesting benefits in certain clinical contexts. Future studies should prioritize head-to-head comparisons of exercise modalities across AF subtypes, standardized adherence monitoring, and longitudinal registry-linked designs to assess clinical progression and real-world applicability.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12872-025-05378-0. Background Atrial fibrillation (AF), characterized by uncoordinated atrial activation and an irregular ventricular response, is a prevalent supraventricular tachyarrhythmia [1–4]. The 2019 AHA/ACC/HRS Guideline further classifies AF into paroxysmal (self-terminating within 7 days), persistent (continuous beyond 7 days), long-standing persistent (continuous > 12 months), and permanent (accepted by patient and clinician) [5]. It affects approximately 1%−2% of the general population in developed countries and poses a significant health burden, associated with a 5-fold increased risk of stroke and a 2-fold increased risk of all-cause mortality [6, 7]. Current management strategies include rate control, rhythm control, cardioversion, ablation, and thromboembolic prophylaxis, which often involve invasive procedures or pharmacotherapy with inherent limitations [1–4]. Exercise training presents a promising non-pharmacological adjunct, with potential benefits including symptom improvement, antiarrhythmic effects in paroxysmal AF, and possibly protective effects against AF development [8–10]. However, the efficacy of exercise across the AF spectrum - including symptom control in paroxysmal AF, preventing progression to persistent AF, and managing AF with heart failure comorbidity - remains unclear. This ambiguity arises from pathophysiological distinctions between these scenarios and conflicting reports on exercise outcomes [9, 11]. Prior syntheses yield inconsistent results. While one meta-analysis found no significant impact of exercise on health-related quality of life (QoL) in AF patients and suggested a potential link between high-intensity endurance exercise and increased AF prevalence [9], others reported improved physical exercise capacity (assessed via 6-minute walk test [6MWT] or peak oxygen uptake [VO2 peak]) [12], enhanced left ventricular ejection fraction (LVEF), specific SF-36 subscores, and reduced body mass index (BMI) [13]. While recent meta-analyses, such as the work by Buckley et al. [14]., have established the effects of exercise on outcomes like AF recurrence, a comprehensive synthesis focusing on a broader spectrum of patient-centered functional outcomes (e.g., specific QoL domains, changes in VO₂ peak) and physiological metrics across diverse exercise modalities remains needed. Therefore, this study aims to provide a contemporary, multidimensional evaluation of exercise interventions in AF, with a specific focus on mechanistic and functional outcomes that complement existing evidence. Methods Study design All steps of this meta-analysis were performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines [15, 16]. After elaborating on the research parameters using the PICOS principle [17], the literature was searched, and the retrieved papers were screened based on the inclusion and exclusion criteria. This study has been prospectively registered in PROSPERO (registration number: CRD420251115389). Eligibility criteria The inclusion criteria were (1) Population (P): Adult patients diagnosed with AF, regardless of clinical subtype or stage; (2) Intervention (I): Exercise-based interventions, including but not limited to aerobic training, anaerobic exercise, combined exercise programs, exercise-based cardiac rehabilitation, Qigong, and yoga; (3) Comparison (C): Conventional AF treatment (e.g., standard pharmacological management), physical activity on prescription, or moderate-intensity exercise programs; (4) Outcomes (O): Studies reporting at least one of the following outcomes were included: Health-related quality of life measures (e.g., SF-36 domains: general health [GH], physical function [PF], vitality [VT]); Resting heart rate (HR);6-minute walk test (6MWT); Peak oxygen consumption (VO₂ peak); Changes in VO₂ peak from baseline; To ensure conceptual and methodological homogeneity for quality of life assessment, only studies utilizing the SF-36 health survey were included for this outcome. Data from other QoL instruments (e.g., MLHFQ, EQ-5D) were not pooled in the analysis. (5) Study Design: Only randomized controlled trials (RCTs) were included. The exclusion criteria were (1) conference abstract, case report, meta-analysis, review, animal study, or protocol; (2) language not English; (3) full text could not be obtained; (4) no data available. Search strategy PubMed, Embase, and the Cochrane Library databases were searched for potentially eligible studies from inception to November 2024, using the MeSH terms of ‘Atrial Fibrillation’ and ‘Exercise’ as well as relevant key words. Detailed search strategies are provided in Supplementary Table 1. The selection and inclusion of the studies were performed in two stages by two independent reviewers (** and **). This process included the analysis of titles/abstracts followed by the full texts. Disagreements were resolved by a third reviewer (**). Data extraction and quality assessment The retrieved data included the names of the authors, year, country, patient characteristics (male percentage and age), and study characteristics (sample size, intervention, training period, session duration, number of sessions per week, control, and exercise intensity) and outcomes (SF-36, HR at rest, 6MWT, VO2 peak, and the changes in VO2 peak). The RCTs were evaluated according to the Cochrane risk bias tool [18]. The quality assessment was independently performed in duplicate by two investigators (** and **), and any discrepancies were resolved through discussion with a third reviewer (**). Statistical analysis All analyses were performed using STATA SE 14.0 (StataCorp, College Station, TX, USA). Weighted mean differences (WMDs) and the corresponding 95% confidence intervals (CIs) were used to compare the outcomes. Statistical heterogeneity among the included studies was calculated using Cochran’s Q-test and the I2 index. An I2 >50% indicated high heterogeneity. The random-effects model was used when high heterogeneity was present among studies; otherwise, the fixed-effects model was applied [18]. P-values < 0.05 were considered statistically significant. Potential publication bias was assessed using Begg’s and Egger’s tests. In addition, sensitivity analyses were performed to identify the effects of the individual study on the pooled results and test the reliability of the results. Results Identification of the eligible studies Figure 1 presents the study selection process. A thorough reading of the full text was conducted before making these selections. A total of 1619 records were identified from the databases, but 652 were removed before the screening, among the 967 records left. After applying predetermined criteria for inclusion and exclusion, a total of 951 articles were eliminated (86 were meeting abstracts, 134 were reviews, 8 were meta-analyses, 63 were other ineligible articles), leaving only 16 for the meta-analysis. Fig. 1Identification of eligible studies Identification of eligible studies Study characteristics and quality assessment Table 1 presents the characteristics of the included studies. The 16 studies included 593 participants in the exercise group and 556 participants in the control group. Eleven studies were from Europe [9, 19–28], two from the Middle East [29, 30], one from Japan [31], one from China [32], and one from North America [33]. There were eight RCTs exercise interventions that involved aerobics [9, 21, 23, 27–29, 32, 33], while two involved exercise combination [20, 31], two involved cardiac rehabilitation program [19, 22], two involved Yoga [25, 26], one anaerobic and one Qigong [24]. Table 2 presents the quality evaluation of the RCTs according to the Cochrane tool. Eight RCTs scored 6 points, and eight RCTs scored 7 points. Table 1Characteristics of the included studiesAuthor, YearCountrySample size (n)InterventionTrainingperiodSessiondurationNumber ofsessionsper weekControlMale%Age(M/sd)ExerciseintensityOutcomeBorland,2020Sweden46/50Physiotherapist-led exercise-based cardiac rehabilitation3 months60 min2 timesPhysical activity on prescription73.91%/68%74 ± 4/74 ± 6RPE of 13–15SF-36, HR restCai,2021China49/48Aerobics12 weeks65 min5 timesstandard rehabilitation treatment63.3%/66.7557 ± 11/57 ± 9NKVO2peak [ml/(min × kg)]Christle,2018Germany35/35Combination6 months60 min1 timemoderate-intensity exercise63%/60%69 ± 9/71 ± 860%−70%VO2peakHR restJoensen,2019Denmark28/24Aerobics12 weeks60 min2 timesAF usual treatment61%/71562.2 ± 10/60.2 ± 8.970% of maxexercisecapacityAF-QoL-186MWTKato,2019Japan28/31Combination6 months30 min2–3 timesUsual care for AF71%/90%67 ± 10/65 ± 840%−60% of max exercise capacity6MWTHR restMalmo,2016Norway26/25Aerobics12 weeks45 min3 timesAF usual treatment77%/88%56 ± 8/62 ± 985–95% of max HRVO2 peakSF-36HR restNguyen,2020Netherlands80/39cardiac rehabilitation program8 weeksNK2–3 timesUnsuccessful physical active patients83%/72%65 ± 8/64 ± 93–6 METsSF-36Nourmohammadi,2019Iran25/25Aerobics8 weeks60 min2 timesRoutine care program40%/52%57.2 ± 7.4/59.9 ± 7.570–80% of maxHRSF-36Osbak,2012Denmark24/23Aerobics12 weeks60 min3 timesAF conventionaltreatment75%/74%69.5 ± 7.3/70.9 ± 8.370% of maxexercisecapacity6MWTHR restPippa,2007Italy22/21Qigong16 weeks90 min2 timesUsual care63.6%/76.2%68.3 ± 7.2/67.8 ± 9.1NK6MWT, VO2 peakWahlstrom,2017Sweden33/36Yoga12 weeks60 min1 timeUsual care48%/72%64 ± 7/63 ± 8NKHR restWahlström,2020Sweden38/41Yoga12 weeks60 min1 timeUsual care53%/51%65 ± 9/63 ± 10NKSF-36HR restZeren,2016Turkey17/16Anaerobic12 weeks30 min3 timesStandardmedicaltreatment only47%/56%66.18 ± 8.76/67.06 ± 6.3930% of max HR6MWT, VO2 peakTasuku 2024UK13/10HIIT12 weeks60 min2 timesmoderate-intensity exercise77%/60%67 ± 4/71 ± 9HR within 67–95% HRpeak or RPE between 12 and 16VO2 peakKim 2023USA30/31Anaerobic1 yearNK3 timesUsual care76.7%/61.3%65.3 ± 4/62.4 ± 5.485–95% HRpeakSF-36,VO2 peakSeifert 2024Germany99/101Anaerobic1 yearNKNKUsual care63.6%/69.3%68 ± 6.67/65 ± 7.04NKSF-36RPE Rating of perceived exertion, AF Atrial fibrillation, HR at rest Heart rate at rest, 6MWT 6-min walk test, m, VO2 peak Peak oxygen uptake, SF-36 Short Form 36 Health Survey Questionnaire Table 1Characteristics of the included studiesAuthor, YearCountrySample size (n)InterventionTrainingperiodSessiondurationNumber ofsessionsper weekControlMale%Age(M/sd)ExerciseintensityOutcomeBorland,2020Sweden46/50Physiotherapist-led exercise-based cardiac rehabilitation3 months60 min2 timesPhysical activity on prescription73.91%/68%74 ± 4/74 ± 6RPE of 13–15SF-36, HR restCai,2021China49/48Aerobics12 weeks65 min5 timesstandard rehabilitation treatment63.3%/66.7557 ± 11/57 ± 9NKVO2peak [ml/(min × kg)]Christle,2018Germany35/35Combination6 months60 min1 timemoderate-intensity exercise63%/60%69 ± 9/71 ± 860%−70%VO2peakHR restJoensen,2019Denmark28/24Aerobics12 weeks60 min2 timesAF usual treatment61%/71562.2 ± 10/60.2 ± 8.970% of maxexercisecapacityAF-QoL-186MWTKato,2019Japan28/31Combination6 months30 min2–3 timesUsual care for AF71%/90%67 ± 10/65 ± 840%−60% of max exercise capacity6MWTHR restMalmo,2016Norway26/25Aerobics12 weeks45 min3 timesAF usual treatment77%/88%56 ± 8/62 ± 985–95% of max HRVO2 peakSF-36HR restNguyen,2020Netherlands80/39cardiac rehabilitation program8 weeksNK2–3 timesUnsuccessful physical active patients83%/72%65 ± 8/64 ± 93–6 METsSF-36Nourmohammadi,2019Iran25/25Aerobics8 weeks60 min2 timesRoutine care program40%/52%57.2 ± 7.4/59.9 ± 7.570–80% of maxHRSF-36Osbak,2012Denmark24/23Aerobics12 weeks60 min3 timesAF conventionaltreatment75%/74%69.5 ± 7.3/70.9 ± 8.370% of maxexercisecapacity6MWTHR restPippa,2007Italy22/21Qigong16 weeks90 min2 timesUsual care63.6%/76.2%68.3 ± 7.2/67.8 ± 9.1NK6MWT, VO2 peakWahlstrom,2017Sweden33/36Yoga12 weeks60 min1 timeUsual care48%/72%64 ± 7/63 ± 8NKHR restWahlström,2020Sweden38/41Yoga12 weeks60 min1 timeUsual care53%/51%65 ± 9/63 ± 10NKSF-36HR restZeren,2016Turkey17/16Anaerobic12 weeks30 min3 timesStandardmedicaltreatment only47%/56%66.18 ± 8.76/67.06 ± 6.3930% of max HR6MWT, VO2 peakTasuku 2024UK13/10HIIT12 weeks60 min2 timesmoderate-intensity exercise77%/60%67 ± 4/71 ± 9HR within 67–95% HRpeak or RPE between 12 and 16VO2 peakKim 2023USA30/31Anaerobic1 yearNK3 timesUsual care76.7%/61.3%65.3 ± 4/62.4 ± 5.485–95% HRpeakSF-36,VO2 peakSeifert 2024Germany99/101Anaerobic1 yearNKNKUsual care63.6%/69.3%68 ± 6.67/65 ± 7.04NKSF-36RPE Rating of perceived exertion, AF Atrial fibrillation, HR at rest Heart rate at rest, 6MWT 6-min walk test, m, VO2 peak Peak oxygen uptake, SF-36 Short Form 36 Health Survey Questionnaire Characteristics of the included studies 70% of max exercise capacity AF-QoL-18 6MWT 6MWT HR rest VO2 peak SF-36 HR rest 70–80% of max HR AF conventional treatment 70% of max exercise capacity 6MWT HR rest SF-36 HR rest Standard medical treatment only RPE Rating of perceived exertion, AF Atrial fibrillation, HR at rest Heart rate at rest, 6MWT 6-min walk test, m, VO2 peak Peak oxygen uptake, SF-36 Short Form 36 Health Survey Questionnaire Table 2Quality evaluation for Cochrane toolStudyRandom sequence generation (selection bias)Allocation concealment (selection bias)Blinding of participants and peraonnel (performance bias)Blinding of outcome assessmnet (detection bias)Incomplete outcome data (attrition bias)Selective reporting (reporting bias)Other biasTotal quality scoresBorland, 2020*******7Cai, 2021*******7Christle, 2018******6Joensen, 2019*******7Kato, 2019*******7Malmo, 2016******6Nguyen, 2020******6Nourmohammadi, 2019******6Osbak, 2012******6Pippa, 2007******6Wahlstrom, 2017******6Wahlström, 2020******6Zeren, 2016*******7Tasuku 2024*******7Kim 2023*******7Seifert 2024*******7 Table 2Quality evaluation for Cochrane toolStudyRandom sequence generation (selection bias)Allocation concealment (selection bias)Blinding of participants and peraonnel (performance bias)Blinding of outcome assessmnet (detection bias)Incomplete outcome data (attrition bias)Selective reporting (reporting bias)Other biasTotal quality scoresBorland, 2020*******7Cai, 2021*******7Christle, 2018******6Joensen, 2019*******7Kato, 2019*******7Malmo, 2016******6Nguyen, 2020******6Nourmohammadi, 2019******6Osbak, 2012******6Pippa, 2007******6Wahlstrom, 2017******6Wahlström, 2020******6Zeren, 2016*******7Tasuku 2024*******7Kim 2023*******7Seifert 2024*******7 Quality evaluation for Cochrane tool Quality of life (SF-36) Seven studies used the SF-36 scores as outcomes [9, 19, 22, 25, 27, 29, 33]. The exercise intervention did not improve the GH (WMD = 4.10, 95%CI: −3.17, 11.37, P = 0.269; I2 = 86.4%, Pheterogeneity<0.001), PF (WMD = 7.50, 95%CI: −5.78, 22.77, P = 0.268; I2 = 98.8%, Pheterogeneity<0.001), and VT (WMD = 4.45, 95%CI: −1.80, 10.69, P = 0.163; I2 = 80.0%, Pheterogeneity<0.001) scores (Fig. 2A, B and C). Subgroup analyses by AF subtype showed insignificant difference on GH between groups in paroxysmal/persistent AF subgroup (WMD = 6.59, 95%CI: −1.00, 14.18, P = 0.089; I2 = 81.2%, Pheterogeneity=0.005) and non-paroxysmal/persistent AF subgroup (WMD = 0.29, 95%CI: −10.76, 11.34, P = 0.959; I2 = 73.6%, Pheterogeneity=0.052) (Figure S1). Results of subgroup analyses on control subtype showed insignificant difference on GH between intervention and control group in standard rehabilitation treatment subgroup (WMD=−0.28, 95%CI: −10.06, 9.51, P = 0.956; I2 = 68.8%, Pheterogeneity=0.073) and medical treatment + Observation subgroup (WMD = 6.99, 95%CI: −0.47, 14.44, P = 0.066; I2 = 79.5%, Pheterogeneity=0.008) (Figure S2). Results of subgroup analyses on AF subtype showed insignificant difference on PF between intervention and control group in paroxysmal/persistent AF subgroup (WMD = 10.92, 95%CI: −10.99, 32.83, P = 0.329; I2 = 98.3%, Pheterogeneity<0.001) and non-paroxysmal/persistent AF subgroup (WMD = 3.74, 95%CI: −1.73, 9.20, P = 0.180; I2 = 73.2%, Pheterogeneity=0.011) (Figure S3). Results of subgroup analyses on control subtype showed insignificant difference on PF between intervention and control group in standard rehabilitation treatment subgroup (WMD = 1.00, 95%CI: −3.69, 5.69, P = 0.676; I2 = 0, Pheterogeneity=1.000) and medical treatment + Observation subgroup (WMD = 10.08, 95%CI: −6.38, 26.53, P = 0.230; I2 = 99.2%, Pheterogeneity<0.001) (Figure S4). Results of subgroup analyses on trial duration showed insignificant difference on PF between intervention and control group in ≤ 3-month subgroup (WMD = 7.12, 95%CI: −9.29, 23.53, P = 0.395; I2 = 98.1%, Pheterogeneity<0.001) and > 3-month subgroup (WMD = 8.02, 95%CI: −8.19, 24.23, P = 0.332; I2 = 90.1%, Pheterogeneity=0.001) (Figure S5). Fig. 2Forest plots of the Short Form 36 Health Survey Questionnaire (SF-36). A General health (GH). B Physical function (PF). C Vitality (VT) Forest plots of the Short Form 36 Health Survey Questionnaire (SF-36). A General health (GH). B Physical function (PF). C Vitality (VT) Although the overall analysis showed no significant improvement in SF-36 scores, subgroup analyses revealed nuanced effects. Notably, exercise interventions led to a significant improvement in VT subscore for the paroxysmal/persistent AF subgroup (WMD = 7.28, 95%CI: 1.73, 12.82, P = 0.010; I²=68.3%) and in the subgroup where the control was medical treatment plus observation (WMD = 8.28, 95%CI: 3.28, 13.28, P = 0.001; I²=51.4%) (Figures S6, S7). These findings suggest that the benefits of exercise on QoL may be specific to certain patient populations and clinical settings. HR at rest. Seven studies included HR at rest as an outcome [9, 19, 20, 23, 25, 26, 31]. Through the meta-analysis we found that the exercise intervention did not improve the HR at rest (WMD=−1.71, 95%CI: −5.23, 1.82, P = 0.343; I2 = 61.6%, Pheterogeneity=0.016) (Fig. 3). Results of subgroup analyses on AF subtype showed insignificant difference on HR between intervention and control group in paroxysmal/persistent AF subgroup (WMD=−3.52, 95%CI: −7.55, 0.52, P = 0.088; I2 = 50.3%, Pheterogeneity=0.110) and non-paroxysmal/persistent AF subgroup (WMD = 1.12, 95%CI: −5.57, 7.81, P = 0.743; I2 = 71.5%, Pheterogeneity=0.030) (Figure S8). Results of subgroup analyses on control subtype showed insignificant difference on HR between intervention and control group in standard rehabilitation treatment subgroup (WMD=−0.83, 95%CI: −6.15, 4.49, P = 0.760; I2 = 76.5%, Pheterogeneity=0.005) and medical treatment + Observation subgroup (WMD=−3.24, 95%CI: −6.99, 0.52, P = 0.091; I2 = 0, Pheterogeneity=0.016) (Figure S9). Results of subgroup analyses on trial duration showed insignificant difference on HR between intervention and control group in ≤ 3-month subgroup (WMD=−1.38, 95%CI: −6.71, 3.95, P = 0.612; I2 = 71.0%, Pheterogeneity=0.008) and >3-month subgroup (WMD = 3.00, 95%CI: −6.44, 0.44, P = 0.087; I2 = 0, Pheterogeneity=1.000) (Figure S10). Fig. 3Forest plot of the heart rate at rest Forest plot of the heart rate at rest 6MWT Five RCTs examined the effect of exercise on the 6MWT [21, 23, 24, 30, 31] and found that the exercise intervention improved the 6MWT in patients with AF (WMD = 66.27, 95%CI: 16.49, 116.04, P = 0.009; I2 = 77.4%, Pheterogeneity=0.001) (Fig. 4). Results of subgroup analyses on trial duration showed insignificant difference on 6MWT between intervention and control group in ≤ 3-month subgroup (WMD = 53.39, 95%CI: −7.65, 114.42, P = 0.086; I2 = 76.6%, Pheterogeneity=0.014) and >3-month subgroup (WMD = 88.92, 95%CI: −28.61, 206.44, P = 0.138; I2 = 88.0%, Pheterogeneity=0.004) (Figure S11). Fig. 4Forest plot of the 6-min walk test (6MWT) Forest plot of the 6-min walk test (6MWT) VO2 peak Six studies examined VO2 peak [9, 21, 28, 31–33] and showed that the exercise intervention improved the VO2 peak in patients with AF (WMD = 3.08, 95%CI: 0.88, 5.29, P = 0.006; I2 = 63.3%, Pheterogeneity=0.018) (Fig. 5). Results of subgroup analyses on AF subtype showed insignificant difference on VO2 peak between intervention and control group in paroxysmal/persistent AF subgroup (WMD = 1.34, 95%CI: −0.40, 3.08, P = 0.132; I2 = 0, Pheterogeneity=0.501) but significant difference in non-paroxysmal/persistent AF subgroup (WMD = 5.36, 95%CI: 3.68, 7.04, P < 0.001; I2 = 0, Pheterogeneity=0.442) (Figure S12). Results of subgroup analyses on control subtype showed significant difference on VO2 peak between intervention and control group in standard rehabilitation treatment subgroup (WMD = 3.67, 95%CI: 1.47, 5.88, P = 0.001; I2 = 60.6%, Pheterogeneity=0.038) but insignificant difference in medical treatment + Observation subgroup (WMD=−0.70, 95%CI: −5.20, 3.80, P = 0.761; I2 = NA, Pheterogeneity=NA) (Figure S13). Results of subgroup analyses on trial duration showed insignificant difference on VO2 peak between intervention and control group in ≤ 3-month subgroup (WMD = 2.24, 95%CI: −0.58, 5.07, P = 0.119; I2 = 45.0%, Pheterogeneity=0.142) and >3-month subgroup (WMD = 4.24, 95%CI: −0.27, 8.75, P = 0.065; I2 = 86.7%, Pheterogeneity=0.006) (Figure S14). Fig. 5Forest plot of the peak oxygen uptake (VO2 peak) Forest plot of the peak oxygen uptake (VO2 peak) Changes in VO2 peak Two studies examined the changes in VO2 peak [9, 32] and showed that the exercise intervention improved the VO2 peak in patients with AF (WMD = 3.78, 95%CI: 2.16, 5.39, P < 0.001; I2 = 0.0%, Pheterogeneity=0.615) (Fig. 6). Fig. 6Forest plot of the changes in peak oxygen uptake (VO2 peak) Forest plot of the changes in peak oxygen uptake (VO2 peak) Publication bias No publication bias was observed when considering HR at rest (Begg’s test, P = 0.548; Egger’s test, P = 0.397) (Figure S15A), SF-36 GH score (Begg’s test, P = 1.000; Egger’s test, P = 0.073) (Figure S15B), SF-36 PF score (Begg’s test, P = 0.368; Egger’s test, P = 0.460) (Figure S15C), SF-36 VT score showed asymmetry (Begg’s test, P = 0.806; Egger’s test, P = 0.056) (Figure S15D), 6MWT (Begg’s test, P = 0.462; Egger’s test, P = 0.105) (Figure S15E), VO2 peak (Begg’s test, P = 1.000; Egger’s test, P = 0.568) (Figure S15F), and changes in VO2 peak (Begg’s test, P = 1.000) (Figure S15G). Sensitivity analysis The sensitivity analyses suggest that the pooled results were robust (Figure S16). Discussion This meta-analysis of 16 RCTs demonstrates that exercise interventions significantly improve functional capacity in patients with AF, evidenced by gains in the 6MWT and VO₂ peak, while overall effects on resting heart rate and health-related quality of life were not apparent. Although previous meta-analyses on the same topic have been published [34, 35], the current meta-analysis encompassed a broader range of health outcomes for AF patients, including heart rate at rest, SF-36 scores (GH, PF, and VT), 6MWT, VO2 peak, and changes in VO2 peak. This comprehensive approach allows for a more holistic understanding of how exercise interventions affect various aspects of AF patient health. Moreover, this study benefited from its rigorous inclusion criteria, which specifically required RCTs focusing on different types of exercise interventions such as aerobic interval training, yoga, Qigong, and anaerobic exercises. This meticulous selection process ensures a high level of evidence and reduces bias, providing a robust foundation for its conclusions. The inclusion of diverse exercise forms and detailed subgroup analyses offered deeper insights into which specific types of exercise were most beneficial for particular AF subtypes or intervention settings. Moreover, while prior syntheses [9, 12, 13], including the recent and rigorous work by Buckley et al. [36], have contributed significantly to the field, our study offers several distinct and complementary insights. First, our search extended through November 2024, thereby capturing several recent RCTs not included in previous meta-analyses. Second, beyond a global QoL score or a limited set of outcomes, we evaluated a wider spectrum of clinically relevant endpoints, including resting heart rate, changes in VO₂ peak, and multiple SF-36 subdomains to dissect specific dimensions of QoL. Third, we performed detailed subgroup analyses across AF subtypes, control interventions, and trial durations to identify potential modifiers of the treatment effect, offering nuanced evidence for personalized exercise prescription. Finally, our analysis encompassed a diverse range of exercise modalities, providing preliminary insights into their relative effectiveness. 6MWT and VO2 peak are two parameters of exercise capacity. Risom et al. [12] analyzed six RCTs and concluded that the exercise intervention did not improve the health-related quality of life but improved the surrogates of exercise capacity. The meta-analysis by Kato et al. [13] included five RCTs and showed that the exercise intervention in AF patients improved the exercise capacity, left ventricular ejection fraction (LVEF), body mass index (BMI), and the GH and VT subscores of the SF-36. However, we observed no significant changes in vitality, physical function, or general health subscores – possibly due to heterogeneity in QoL assessment tools and AF subtypes. Therefore, the fact that exercise intervention in AF patients improves exercise capacity appears to make consensus among the available meta-analyses, while the impact on the AF outcomes appears to be less clear. Still, exercise intervention in the general population improves health-related outcomes, quality of life, and mortality, with minimal safety concerns [37, 38]. Clinically, the 66 m 6MWT gain exceeds the minimal clinically important difference (MCID) of 30–35 m for cardiac patients [39], suggesting meaningful functional improvement. The 3.08 mL/kg/min VO₂ peak increase similarly approaches gains seen in heart failure rehabilitation [40], supporting exercise integration into AF management. On the other hand, AF is a potentially life-threatening condition, and it remains unclear whether the beneficial effects of exercise observed in the general population can be applied to patients with AF. Still, in the present meta-analysis, one of the included studies reported an improvement in HR at rest after the exercise intervention [19]. That study applied a 3-month physiotherapist-led cardiac rehabilitation program compared with physical activity on prescription [19]. The use of a formal, vigorous, and rigorous cardiac rehabilitation program might result in better outcomes than other interventions. Future studies and meta-analyses could focus on this type of intervention in patients with AF. QoL is a subjective parameter of the patients’ perception of the impact of a condition on their quality of life [41]. Many tools exist for the assessment of the quality of life [41–44]. The selection of a specific tool can yield different outcomes. The SF-36 is a commonly used tool for the assessment of the health-related quality of life. Still, Joensen et al. [21] used the AF-QoL-18 tool instead of the SF-36 and reported improvements in the SF-18 scores with the exercise intervention. The AF-QoL-18 tool is specifically validated for the quality of life in patients with AF [45]. Therefore, the selection of quality of life tools made for AF might yield different results. The high heterogeneity observed in the overall QoL analysis (I² >80%) warrants cautious interpretation. It is critical to note that this heterogeneity does not stem from pooling different measurement scales, as our analysis was strictly limited to the SF-36 instrument. Instead, it likely reflects the significant clinical and methodological diversity across trials, including different AF subtypes, control group designs, and exercise protocols. The fact that heterogeneity was substantially lower in several of our subgroup analyses supports this interpretation and underscores the importance of considering these moderating factors. As in a previous meta-analysis, Risom et al. [12] could not examine the impact of the exercise intervention on mortality or serious cardiovascular events in patients with AF. Such outcomes necessitate long follow-ups or cohort studies, which were not included in the present meta-analysis. A retrospective study suggested that exercise-based cardiac rehabilitation decreased the risks of all-cause mortality, rehospitalization, and stroke over a follow-up of 18 months [46]. A study suggested an inverse dose-response relationship between exercise and AF and its complications, but the benefits might be lost in men exercising at high levels [47], which could provide hints for the design of future trials. Although, this meta-analysis included 14 high-quality RCTs provides reliable evidence on the identification efficacy of exercise interventions for AF. However, heterogeneity was observed regarding different exercise type, duration, and intensity, and studies are still necessary with specific interventions that might be more promising. Especially, differences between various intervention methods for physical activity are significant, as indicated by the study conducted by Malmo, Nes (9). Their research demonstrates the beneficial effects of aerobic interval training (AIT) on AF duration, AF symptoms, cardiac function, exercise capacity, quality of life, and overall patient outcomes. These findings highlight the importance of further developing exercise interventions tailored to meet the specific needs of AF patients. Of note, in the subgroup analyses of this meta-analysis, several positive outcomes were observed. Notably, exercise interventions demonstrated a significant improvement in the VT subscore of the SF-36 for the paroxysmal/persistent AF subgroup, although no significant difference was found in the non-paroxysmal/persistent AF subgroup. Furthermore, there was a significant difference in VT between the intervention and control groups within the medical treatment plus observation subgroup. Additionally, improvements in VO2 peak were significantly greater among patients with non-paroxysmal/persistent/disabling AF compared to those with paroxysmal/persistent/disabling AF. Also, standard rehabilitation treatment subgroups showed significant improvements in VO2 peak. Cardiac rehabilitation for patients with atrial fibrillation is designed to provide comprehensive patient support by optimizing clinical stability and therapeutic management, prescribing and facilitating physical activity, offering education on lifestyle-related risk factors and psychosocial stressors, and conducting regular outcome assessments [48]. These findings suggest that tailored exercise interventions may offer particular benefits for specific AF subtypes and when integrated into standard rehabilitation programs. Moreover, subgroup analyses in this study suggest that more intensive, supervised, and structured exercise regimens-particularly aerobic and combined training within rehabilitation settings - are highly promising for AF management, especially in patients with persistent forms of the arrhythmia. Future clinical protocols may prioritize these approaches for targeted patient care. Our findings on the benefits of exercise for functional capacity should be interpreted within the contemporary management landscape of AF, where catheter ablation plays a pivotal role. As highlighted by the reviewer, ablation is highly effective for restoring sinus rhythm, reducing drug burden, and improving quality of life [49, 50]. A recent comparative study by Matteucci et al. highlighted that the choice of ablation energy source (with pulsed-field ablation showing superiority) and anesthesia strategy significantly influences patient-reported quality of life and symptom perception post-ablation [47, 51]. This underscores that technical procedural aspects can have a profound impact on the patient’s lived experience. Furthermore, the benefits of restoring sinus rhythm may extend into unexpected domains of quality of life. A compelling analysis by La Fazia et al. revealed that successful catheter ablation was strongly associated with a reduction in the need for medications for erectile dysfunction, suggesting a significant improvement in sexual health - an important yet often overlooked aspect of overall well-being in AF patients [52]. Limitations There were several limitations of this meta-analysis: (1) The heterogeneity of the results shown by HR AT REST, SF-36 GH, SF-36 PF, SF-36 VT, 6MWT may be due to the limited number of research studies included and the subjective factors present in the research content. Therefore, it is recommended to accumulate more research and conduct subgroup research to reduce heterogeneity. (2) Funnel plots for the SF-36 GH, SF-36 PF, and SF-36 VT score showed asymmetry. Although Begg’s test did not reveal any significant publication bias for any of the three scores, Egger’s regression asymmetry tests confirmed the presence of a significant publication bias only for SF-36 PF score (P = 0.001). This is also reflected in sensitive analysis. Only six RCTs examined the effect of exercise intervention on the SF-36 PF score. Due to the limited number of primary studies available, we opted not to exclude the research. Then we conducted an examination while carefully considering its heterogeneity. (3) The quality of the meta-analysis is dependent on the quality of the studies it includes, as it inherits their limitations and weaknesses. The number of studies included in the analysis was limited, and they varied in terms of their outcomes and definitions, resulting in smaller numbers of studies for each pooled analysis. Some studies had small sample sizes, which may have led to an overestimation of treatment effects. (4) The RCTs included were diverse in terms of exercise type, intervention duration, and intensity, and some interventions may be more effective than others. Our research results can be negatively impacted by the included studies used different outcomes and measurement tools (5) In order to gain a deeper understanding of the impact of rehabilitation programs, it is imperative that future research places greater emphasis on adherence. This will enable researchers to determine whether any lack of effectiveness is attributable to poor adherence to the program [53]. (6) Generalizability is tempered by geographical limitations (11/16 studies European), variable exercise modalities (aerobic, yoga, qigong), and underrepresentation of persistent AF cohorts. Outcomes may be less transferable to healthcare systems without structured cardiac rehabilitation infrastructure. Future studies should prioritize head-to-head comparisons of high-intensity interval training versus moderate continuous training across AF subtypes using validated AF-specific quality-of-life tools, trials in underrepresented populations, and longitudinal investigations of exercise effects on AF progression/hospitalization via registry-linked designs. (7) Limited to the data extracted in this study, important clinical questions including stroke, myocardial infarction, hospitalization and death were not addressed, more trials with longer follow-up are needed to verify these issues. Conclusions This meta-analysis of RCTs suggests that in patients with AF, an exercise intervention appears to improve exercise-related outcomes like the 6MWT and VO2 peak, but not the HR at rest and the health-related quality of life. Future research should directly compare exercise modalities in different AF subtypes, monitor adherence consistently, and use registry-linked designs to assess long-term outcomes and real-world effectiveness. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3. Supplementary Material 4. Supplementary Material 4. Supplementary Material 5. Supplementary Material 5. Supplementary Material 6. Supplementary Material 6. Supplementary Material 7. Supplementary Material 7. Supplementary Material 8. Supplementary Material 8. Supplementary Material 9. Supplementary Material 9. Supplementary Material 10. Supplementary Material 10. Supplementary Material 11. Supplementary Material 11. Supplementary Material 12. Supplementary Material 12. Supplementary Material 13. Supplementary Material 13. Supplementary Material 14. Supplementary Material 14. Supplementary Material 15. Supplementary Material 15. Supplementary Material 16. Supplementary Material 16. Supplementary Material 17. Supplementary Material 17. Supplementary Material 18. Supplementary Material 18.