Effects of cardiac rehabilitation on atrial fibrillation recurrence, mortality, hospitalization, and exercise capacity: a systematic review and meta-analysis

Abstract

Background

Cardiac rehabilitation (CR) may benefit patients with atrial fibrillation (AF) in areas such as exercise capacity. However, evidence regarding its impact on AF recurrence, a key clinical outcome, remains inconsistent. This systematic review and meta-analysis of randomized controlled trials (RCTs) aimed to evaluate the effects of CR on AF recurrence, all-cause mortality, hospitalization, and exercise capacity in patients with AF.

Methods

We systematically searched the Cochrane Library, PubMed, and Embase from January 1, 1980, to April 20, 2026. Two reviewers independently screened studies, extracted data, and assessed risk of bias. A total of 1,550 patients from 11 RCTs were included in the meta-analysis based on the predefined inclusion and exclusion criteria.

Results

Pooled analysis demonstrated a significant reduction in AF recurrence among patients receiving CR compared with controls (risk ratio (RR) 0.77, 95% confidence interval (CI) [0.67–0.89], p = 0.0003, I² = 20%). However, this effect was not statistically significant in the exploratory subgroup of patients post-ablation (RR 0.86, 95% CI [0.69–1.07], p = 0.17, I² = 0%). Exercise capacity, measured via the 6-minute walking test, significantly improved (weighted mean difference (WMD) 32.22, 95% CI [21.22–43.23], p < 0.00001, I² = 0%). No significant differences were observed in all-cause mortality (RR 1.04, 95% CI [0.74–1.46], p = 0.81, I² = 0%), hospitalization rates (RR 1.00, 95% CI [0.85–1.19], p = 0.97, I² = 0%), or the composite outcome of mortality and hospitalization (RR 1.01, 95% CI [0.90–1.14], p = 0.83, I² = 0%).

Conclusions

CR is associated with a reduction in AF recurrence in the overall AF population, though this benefit was not observed in the exploratory subgroup of patients post-ablation. It significantly improves exercise capacity but does not appear to affect mortality or hospitalization rates. Further high-quality randomized trials are needed to evaluate the benefits and potential risks of CR across different types of AF.

Introduction

The global burden of atrial fibrillation (AF), the most common sustained arrhythmia, continues to grow, reflected in its rising worldwide incidence and prevalence (Linz et al., 2024). AF is associated with an elevated risk of death, stroke, and peripheral embolism, which substantially impairs patients’ physical and mental health and compromises their quality of life (Schnabel et al., 2015; Sagris et al., 2021). AF is associated with multiple risks, including a 2-fold increased risk of myocardial infarction, a 5-fold increased risk of stroke and heart failure, as well as dementia and cognitive decline. Coexistence of the aforementioned conditions is associated with a higher mortality compared to each condition alone (Linz et al., 2024). The global prevalence of AF has doubled from 1990 to 2019, reaching 59.7 million cases in 2019 (Li et al., 2022). More recent estimates from 2021 indicate that there were approximately 4.48 million new cases, 8.36 million disability-adjusted life years, and 0.34 million deaths attributable to AF or atrial flutter (AFL) (Cheng et al., 2024).

Cardiac rehabilitation (CR), typically incorporating exercise training and risk factor management, has been demonstrated to reduce arrhythmia duration in patients with paroxysmal and persistent AF. For individuals with chronic AF, CR can alleviate arrhythmia-related symptoms by lowering the resting ventricular rate (Robaye et al., 2020).

A large body of evidence supports that CR is beneficial for patients with AF (Wagner et al., 2018; Alharbi et al., 2019; De With et al., 2019; Kato et al., 2019; Nourmohammadi et al., 2019; Hegbom et al., 2006; Malmo et al., 2016; Osbak et al., 2011; Pippa et al., 2007; Wahlstrom et al., 2017; Zeren et al., 2016; Keteyian et al., 2019; Garnvik et al., 2020; Wu et al., 2022; Leggio et al., 2021; Aoyama et al., 2021; Misra et al., 2023; Buckley et al., 2021b; Buckley et al., 2021a; Cai et al., 2022; AbuElkhair et al., 2023; Ashburner et al., 2024; Mills et al., 2020). A 2024 Cochrane (Buckley et al., 2024) review found that exercise-based CR programmes targeted at AF patients greatly improves AF-specific measures including reduced AF recurrence, symptom burden, and episode duration, as well as the mental components of HRQoL. However, the impact of CR specifically on AF recurrence, particularly in key subgroups such as post-ablation patients, warrants further detailed examination.

There has been increasing awareness of CR as a treatment for AF patients. However, it remains controversial whether CR has a positive effect on AF burden. Findings reported by Pogosova et al. (2023) showed no significant difference between the CR groups and the control groups in AF recurrence while Elliott et al. (2023) observed that CR reduced the risk of AF recurrence.

To address the current controversy and to build upon existing meta-analyses, we conducted this systematic review and meta-analysis to specifically evaluate the impact of CR on AF recurrence, with comprehensive exploratory subgroup analyses. We also aimed to evaluate its effects on other critical outcomes of patients with AF, including mortality, hospitalization, and exercise capacity.

Methods

Search strategy and selection criteria

The meta-analysis was performed in accordance with the preferred reporting items for systematic reviews and meta-analysis (PRISMA) guidelines (Page et al., 2021). A systematic literature search was conducted in PubMed, EMBASE and the Cochrane Library from January 1, 1980 to April 20, 2026. Key search words included the following: ‘risk factor’ OR ‘cardiac rehabilitation’ OR ‘cardiovascular rehabilitation’ OR ‘targeted therapy’ OR ‘exercise’ OR ‘sport’ OR ‘train’ AND ‘atrial fibrillation’. In addition, the retrieved previous Cochrane Reviews were carefully reviewed to identify their included literature one by one, and articles that met the following criteria were included in our statistics.

Study registration and protocol amendments

The protocol for this systematic review was prospectively registered on the International Platform of Registered Systematic Review and Meta-analysis Protocols (INPLASY) on June, 2020 (Registration number: INPLASY202060003). In accordance with best practice for living reviews or to capture emerging evidence, the search strategy was updated on April 20, 2026, and the INPLASY record was amended accordingly. No other modifications were made to the pre-specified PICOS criteria (Population, Intervention, Comparison, Outcomes, Study design).

The inclusion criteria for our analysis were as follows: (i) Randomized controlled trials (RCTs); (ii) AF patients; (iii) A CR intervention group was allocated to perform rehabilitation programme (The rehabilitation programme must have included an exercise training component. It may also have included a risk factors management component); (iv) The control group received usual care or instruction to continue their regular exercise habits or low-enough-intensity exercise with only little improvement in fitness was expected. (v) The primary outcome: the number of patients with AF recurrence; the secondary outcomes: all-cause mortality and hospitalization, and 6-minute walking test. The exclusion criteria were as follows: (i) animal test; (ii) no study outcome indicators and unpublished studies. The definition and management of AF (including prior interventions such as cardioversion or antiarrhythmic drug therapy in the non-ablation population) were based on contemporary guidelines, primarily those of the European Society of Cardiology (ESC).

Data extraction and quality assessment

The retrieved records were imported into reference management software (e.g., EndNote). Two reviewers (T Shen and C Shi) independently screened the titles and abstracts of all records against the inclusion criteria. The full texts of potentially eligible studies were then retrieved and assessed for eligibility by the same two reviewers independently. Any disagreements at either stage were resolved through discussion or by consulting a third reviewer (YM Jiang). The process of selecting studies was documented using a PRISMA flow diagram.

Two reviewers (T Shen and C Shi) extracted the following information for each study independently: (1) author, year of publication and study design; (2) AF patient demographic and clinical characteristics; (3) intervention characteristics; (4) the number of patients with AF recurrence, or weighted mean difference (WMD) and standard deviation (SD) (or standard error (SE), p value or 95% confidence interval (CI); (5) characteristics of assessment methodology for AF recurrence; (6) reporting of adverse events and intervention compliance. Any discrepancies in extracted data were resolved through consensus or by adjudication from a third reviewer (YM Jiang).

Study quality was assessed using the Cochrane Collaboration tool by two researchers (T Shen and C Shi) independently. If there were discrepancies, a third researcher (YM Jiang) would take part in the discussion to make the final assessment decision. The Cochrane Collaboration tool assessed the risk of bias in the following domains: selection bias, performance bias, detection bias, attrition bias, reporting bias and other biases. Each domain was classified as ‘low risk’, ‘unclear’, or ‘high risk’.

To address the potential issue of duplicate data from multiple reports of the same study, we implemented the following procedure: (1) during the screening phase, we compared trial registration numbers, author lists, institutions, sample sizes, and intervention details to identify potential duplicate reports; (2) if multiple publications from the same trial were identified, they were considered as a single study; (3) data were then extracted from the publication with the most comprehensive outcome reporting or the longest follow-up period to avoid duplication bias.

Statistical analysis

The statistical methods for this meta-analysis were performed according to our pre-registered protocol (INPLASY202060003) (Shen et al., 2020). In brief, analyses were conducted using RevMan V5.3 (The Nordic Cochrane Centre, Copenhagen, Denmark) and R (version 4.3.1; R Core Team, 2023). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. For binary and continuous outcomes, we calculated risk ratios (RRs) and weighted mean differences (WMDs), respectively. Heterogeneity was quantified using the I² statistic, with I² > 50% indicating substantial heterogeneity, leading to the use of a random-effects inverse variance model; otherwise, a fixed-effect model was applied. Given the limited number of studies (n = 3) reporting exercise capacity, a random-effects model was employed for this outcome to provide a more conservative estimate. We judged the statistical significance based on 5% level of significance and reported pooled mean results with 95% CIs. To explore potential sources of heterogeneity, exploratory subgroup analyses were conducted based on whether studies specifically enrolled post-ablation patients or a broader AF population not restricted by ablation status (termed “non-ablation-specific” subgroup for brevity). Visual inspection of funnel plots was used to assess the risk of publication bias.

Results

Search results and study selection

A total of 8,792 records were identified through the pre-specified database searches up to April 20, 2026, and 2,959 duplicates were removed. Among the remaining 5,833 records, 32 full-text articles were assessed for eligibility. After full-text assessment, 11 studies (Kato et al., 2019; Pogosova et al., 2023; Elliott et al., 2023; Risom et al., 2016; Rienstra et al., 2018; Skielboe et al., 2017; Luo et al., 2017; Abed et al., 2013; Baek et al., 2019; Gessler et al., 2021; Bubnova et al., 2022) were included in the final meta-analysis (Fig. 1).

Figure 1. Study flow diagram.

Baseline characteristics of participants

The number of participants in each study ranged from 47 to 382. For the intervention and the control groups, there were 788 and 762 participants respectively. Duration of follow-up ranged from 12 weeks to 2.6 years (Table 1). Detailed characteristics of each included study are provided in Table S1.

Table 1. Baseline characteristics of included studies.

Study (Publication year)	Study country	Duration of follow-up	Population	Study design	Intervention group	Control group	Diagnostic mode	Number	Intervention group		Control group
									Events	Total (M/F)	Events	Total (M/F)
Abed et al. (2013)	Australia	1.25 years	AF patients	RCT	Cardiac rehabilitation (exercise and risk factors management) ET:3/w 20 min, LI exercise∼1–2/w LI exercise + MI exercise	Exercise advice	7-day Holter	81	9	42 (Not reported)	22	39 (Not reported)
Risom et al. (2016)	Denmark	0.5 year	AF patients	RCT	Cardiac rehabilitation (exercise) ET:3/w, 12w, CT	Usual care	Clinical evaluation	210	3	105 (74/31)	3	105 (77/28)
Luo et al. (2017)	USA	2.6 years	AF and HF patients	RCT	Cardiac rehabilitation (exercise) ET: supervise: 3/w, 30 min, 12w home-based: 120 min/w, 2 year	Usual care	ECG	382	21	193 (Not reported)	17	189 (Not reported)
Skielboe et al. (2017)	Denmark	0.25 year	AF patients	RCT	Cardiac rehabilitation (exercise) ET: 2/w, 60 min, 12w, 80% of MPE	2/w, 60 min, 12w, 50% of MPE	ECG	70	13	37 (22/15)	17	33 (19/14)
Rienstra et al. (2018)	Netherlands, UK	1 year	AF and HF patients	RCT	Cardiac rehabilitation (exercise and risk factors management) ET:2-3/w, 20–30 min highest intensity that allows conversation	Causal treatment	Holter monitoring	245	30	119 (94/25)	47	126 (99/27)
Kato et al. (2019)	Japan	0.5 year	AF patients	RCT	Cardiac rehabilitation (exercise) ET: supervise: 1–2/w 60 min edurance and resistance training unsupervise: 2–3/w 30 min MIT	Continue usual lifesyle	ECG	59	6	28 (20/8)	8	31 (28/3)
Baek et al. (2019)	South Korea	12 ± 7 months	AF patients	RCT	Cardiac rehabilitation (18 times, No specifics on methodology)	Usual care	Not stated	68	2	23 (Not reported)	7	45 (Not reported)
Gessler et al. (2021)	Germany	1 year	AF patients	RCT	Cardiac rehabilitation (weight reduction programme and exercise: 2/m, 6 months)	Usual care	ILR	133	44	67 (43/24)	43	66 (41/25)
Bubnova et al. (2022)	Russian	12 months	AF patients	RCT	Cardiac rehabilitation (exercise) ET: 45 min/d	Usual care	Holter monitoring	47	6	24 (24/0)	9	23 (23/0)
Pogosova et al. (2023)	Russian	1 year	AF patients	RCT	Cardiac rehabilitation (Consultancy and remote support: Not specified)	Continue usual lifesyle	Clinical evaluation	135	28	90 (48/42)	19	45 (22/23)
Elliott et al. (2023)	Australia	12 months	AF patients	RCT	Cardiac rehabilitation (exercise) ET: 210 min/W, LI exercise+ MI exercise	Usual care	Holter monitoring+ECG	120	36	60 (35/25)	48	60 (34/26)

Notes.

Abbreviation

M/F

male/female

AF

atrial fibrillation

RCT

randomized controlled trial

ET

exercise therapy

LI

low intensity

MI

moderate intensity

CT

continue training

HF

heart failure

ECG

electrocardiogram

MPE

maximum perceived exertion

MIT

moderate intensity training

ILR

implantable loop recorder

Quality assessment

Using the Cochrane Collaboration Tool, detailed information of quality assessment is shown in (Figs. 2A and 2B).

Figure 2. Risk of bias summary.

Outcomes

AF recurrence

Evidence from the pooled analysis of eleven studies showed 198 AF recurrences in 788 CR participants and 240 AF recurrences in 762 control participants. The pooled RR for AF recurrence was significantly lower in the CR group than in the control group (RR 0.77, 95% CI [0.67–0.89], p = 0.0003, I² = 20%) (Fig. 3).

Figure 3. Atrial fibrillation recurrence.

This figure compares the risk ratio (RR) for AF recurrence between the cardiac rehabilitation (CR) group and the control group. Pooled analysis demonstrated a significant reduction in AF recurrence among patients receiving CR compared with controls (RR 0.77, 95% CI [0.67–0.89], p = 0.0003). Heterogeneity was low across studies (I² = 20%), indicating consistent findings.

Among the 11 studies, six were in the post-ablation population and the rest were in the non-ablation-specific population. Exploratory subgroup analyses showed that there was no significant impact of CR on the risk of AF recurrence in the post-ablation population (RR 0.86, 95% CI [0.69–1.07], p = 0.17, I² = 0%). There was a significant decrease in RR between the CR groups and the control groups in the non-ablation-specific population (RR 0.72, 95% CI [0.59–0.86], p = 0.0005, I² = 42%) (Fig. 4). The subgroup analysis should be interpreted as exploratory.

Figure 4. Exploratory subgroup analyses of atrial fibrillation recurrence.

This exploratory subgroup analysis stratified studies by whether they enrolled post-ablation patients or a broader AF population not restricted by ablation status. In the post-ablation subgroup, CR did not significantly reduce AF recurrence (RR 0.86, 95% CI [0.69–1.07], p = 0.17, I² = 0%). In the non-ablation-specific subgroup, CR significantly reduced AF recurrence (RR 0.72, 95% CI [0.59–0.86], p = 0.0005, I² = 42%).

Mortality and hospitalization

Four studies reported all-cause mortality and five studies reported hospitalization. There was no significant impact of CR on the risk of all-cause mortality (RR 1.04, 95% CI [0.74–1.46], p = 0.81, I² = 0%), hospitalization (RR 1.00, 95% CI [0.85–1.19], p = 0.97, I² = 0%), or the composite of mortality and hospitalization (RR 1.01, 95% CI [0.90–1.14], p = 0.83, I² = 0%) (Fig. 5).

Figure 5. Mortality and hospitalization.

This figure shows the pooled effects of CR on all-cause mortality (four studies), hospitalization (five studies), and the composite outcome of mortality and hospitalization (five studies). No significant differences were observed between CR and control groups for any of these outcomes (mortality: RR 1.04, 95% CI [0.74–1.46]; hospitalization: RR 1.00, 95% CI [0.85–1.19]; composite: RR 1.01, 95% CI [0.90–1.14]). Heterogeneity was 0% for all analyses.

Exercise capacity

Three studies reported the 6-minute walking test. Pooled analysis using a random-effects model showed a significant improvement in exercise capacity in intervention versus control participants (WMD 32.22, 95% CI [21.22–43.23], p < 0.00001, I² = 0%) (Fig. 6).

Figure 6. 6-minute walking test.

Three studies reported the 6-minute walking test. Pooled analysis using a random-effects model showed a significant improvement in exercise capacity in CR participants compared with controls (weighted mean difference (WMD) 32.22 m, 95% CI [21.22–43.23], p < 0.00001). Heterogeneity was low (I² = 0%).

Publication bias

The Begg’s funnel plot showed a symmetric distribution (Fig. 7). Begg’s test for publication bias showed no significant correlation between log-transformed relative risks and their standard errors (Kendall’s τ = −0.15, z = −0.62, p = 0.53), indicating no statistical evidence of publication bias.

Figure 7. Funnel plot for publication bias.

Discussion

This systematic review identified eleven randomized clinical trials, with a total of 1,550 patients that compared CR with usual care control. Ten trials complied with the European Society of Cardiology recommendation for physical activity for secondary prevention (Piepoli et al., 2014), and the other one were unspecified. We found evidence from pooled estimates that CR could lead to improvements in AF recurrence and exercise capacity but no difference mortality and hospitalization. A key and novel finding of our analysis was that the effect of improving AF recurrence was heterogeneous, and the improvement was not statistically significant in the post-ablation population. Meta-regression analysis confirmed low statistical heterogeneity among the included studies.

From a clinical perspective, it is important to distinguish between statistical significance and clinical importance. The observed reduction in AF recurrence (RR 0.77) is statistically significant and suggests a potential clinical benefit, particularly in non-ablation-specific patients. the heterogeneity in patient status prior to CR precisely reflects the real-world clinical pathway for managing symptomatic, non-permanent AF. Patients may present in sinus rhythm, undergo cardioversion, or in some cases, even begin rehabilitation while in AF. Our pooled analysis indicates that CR provides clinical benefit across these common starting points, which enhances the generalizability and practical relevance of our findings. However, the absolute risk reduction and number needed to treat depend on the individual patient’s profile and the clinical context. Conversely, the lack of effect on mortality and hospitalization, while statistically clear, does not preclude potential benefits in patient-centered outcomes such as symptom burden and quality of life.

The World Health Organization (WHO) defines CR as a comprehensive set of activities and interventions aimed at enabling patients with cardiovascular disease to achieve optimal physical, mental, and social well-being, thereby allowing them to reintegrate into society and lead an active life through their own efforts (Alnozha et al., 1993). CR aims to restore health for people with AF or those who have been treated for AF, mainly through regular exercise. Whether CR reduces the risk of AF recurrence, comparing inconsistent findings in previous studies, we searched for more accurate conclusions and possible reasons by pooling and comparing analysis.

In contrast to previous meta-analyses, including the recent 2024 Cochrane review (Buckley et al., 2024), which broadly summarized the benefits of CR across multiple outcomes in AF, our study provides a more focused and nuanced analysis. To our knowledge, this is one of the most comprehensive meta-analyses specifically addressing CR’s effect on AF recurrence, with detailed exploratory subgroup analyses. The underlying mechanisms of CR’s improvement in AF patients may include various aspects. Firstly, CR modifies sedentary lifestyles and improves cardiovascular function. CR can improve left ventricular diastolic function (Kato et al., 2019; Bubnova et al., 2022), lower resting heart rate (Bubnova et al., 2022), increase peak oxygen consumption (VO2peak) (Kato et al., 2019; Rienstra et al., 2018; Skielboe et al., 2017; Luo et al., 2017), and promote systemic circulation. These effects collectively reduce left atrial pressure and decrease AF triggers. Secondly, CR targets the mitigation of cardiovascular risk factors—including obesity, high blood pressure, and dyslipidemia—which in turn alleviates the heart’s workload (Pogosova et al., 2023; Gessler et al., 2021; Bubnova et al., 2022). Thirdly, CR can reduce inflammation, improve endothelial function and attenuate atrial remodelling (Abed et al., 2013; Kim et al., 2023; Qin et al., 2021; Buckley, Lip & Thijssen, 2020). Finally, CR may improve autonomic nervous balance (Abidi et al., 2023) and alleviate anxiety or depression (Knapen et al., 2015; Stubbs et al., 2017; Lange & Herrmann-Lingen, 2007; Yu et al., 2012; Franklin et al., 2021; Schuch et al., 2016; Schnabel et al., 2013), thereby potentially reducing the risk of AF recurrence.

Several factors may explain the differential treatment effects observed between the post-ablation and non-ablation-specific populations: Firstly, the strong inherent intervention effect of ablation itself: Radiofrequency ablation directly addresses the electrophysiological basis of AF by isolating pulmonary veins or eliminating abnormal foci (Tzeis et al., 2024). If the procedure is successful, the recurrence rate is already significantly reduced, making the additional benefits of CR statistically difficult to demonstrate. Secondly, mismatch between the mechanisms of CR and post-ablation needs: CR primarily reduces AF recurrence through indirect effects such as improving cardiovascular function, controlling risk factors (e.g., hypertension, obesity), and mitigating inflammation. However, the core reasons for post-ablation recurrence may be procedure-related factors such as residual non-pulmonary vein foci, early recurrence due to excessive inflammatory response (Wei et al., 2022) and late recurrence due to electrical reconnection of previously isolated pulmonary veins and development of atrial fibrosis (Erhard, Metzner & Fink, 2022)—factors that CR has limited influence on. Additionally, while CR may reduce AF by modulating autonomic nervous balance, ablation itself (particularly its denervation effect) may already partially overlap with this mechanism (Benali et al., 2025). Thirdly, time window and assessment period considerations: The first three months following ablation (the blanking period) constitute a high-risk window for AF recurrence (Choi et al., 2020; Wood et al., 2017). However, the therapeutic effects of CR typically require a longer duration (e.g., ≥6 months) to become apparent, meaning short-term studies with limited follow-up are unlikely to detect significant differences. Moreover, post-ablation patients typically undergo intensive monitoring (e.g., implantable loop recorders Gessler et al., 2021), allowing earlier detection of asymptomatic recurrences, whereas potential benefits in the CR group (e.g., symptom reduction) may not be captured in the statistics. Fourthly, other possible factors: Post-ablation patients may adhere more strictly to antiarrhythmic or anticoagulant therapy, offsetting some of the benefits of CR. Additionally, shortly after ablation, patients may experience reduced physical recovery capacity and impaired cardiac functional reserve, along with potential discomfort or insufficient awareness of the importance of rehabilitation, which could hinder the optimal implementation and efficacy of CR measures. Variations in exercise mode, intensity, and individual physiological characteristics similarly affect the outcomes of CR.

In addition, it is likely that CR’s improvement to exercise capacity can be attribute to improve in the heart rate response and peripheral oxygen extraction (Smarz et al., 2021), and positive changes in oxygen delivery, vasculature, peripheral tissues and inflammation (Pinckard, Baskin & Stanford, 2019; Olver, Ferguson & Laughlin, 2015; Tao et al., 2015; Stanford & Goodyear, 2016; Lundby & Jacobs, 2016; Vega et al., 2017; Kasapis & Thompson, 2005). Exercise training is associated with favorable cardiac, vascular and skeletal muscle function improvements that result in increased convective and diffusive O2 transport, leg blood flow and O2 extraction in patients (Tucker et al., 2018).

The practical relevance of this study lies in its support for the integrated, bio-psycho-social management of AF, aligning with contemporary guidelines. The 2024 ESC guidelines for AF advocate the Co-morbidity, Avoid stroke, Reduce symptoms, Evaluation (CARE) approach, emphasizing comprehensive risk factor management (Van Gelder et al., 2024). By establishing individualized lifestyle management, CR promotes key outcomes—including symptom control, reduced AF burden, and improved exercise capacity—via appropriately dosed training (Langheim, 2023), directly targeting potentially reversible AF drivers such as obesity, physical inactivity, and hypertension (Chung et al., 2020).

These findings offer direct implications for refining clinical pathways. Firstly, regarding assessment strategies, our results support systematically evaluating eligibility for CR early in the care pathway for symptomatic AF, especially in the non-ablation-specific population. Secondly, for non-pharmacological interventions, CR should be considered a foundational therapy alongside pharmacotherapy and procedures. Thirdly, in terms of patient involvement, the demonstrated benefits highlight the need for structured education to enhance engagement and adherence. Finally, for long-term care models, embedding structured CR into routine follow-up facilitates a holistic, patient-centered approach, consistent with the CARE paradigm.

While CR does not significantly reduce AF recurrence in the short term among post-ablation patients, it confers measurable benefits in terms of enhanced exercise tolerance, reduced AF burden, and alleviation of symptoms (De With et al., 2019; Kato et al., 2019; Bubnova et al., 2022).

Future research should consider an integrated bio-psycho-social perspective to optimize AF management. Specific directions include: (1) developing multidimensional risk assessment tools that incorporate psychological and social determinants of health; (2) investigating the effect of combined psychological interventions (e.g., cognitive-behavioral therapy) and CR on AF symptom burden and quality of life; (3) designing structured, patient-centered rehabilitation pathways tailored to different AF phenotypes (e.g., paroxysmal vs. persistent, post-ablation vs. non-ablation); and (4) evaluating the long-term impact of CR on hard endpoints such as stroke, heart failure, and cardiovascular mortality in larger, pragmatic trials.

Earlier implementation of CR is associated with superior clinical outcomes. While complete prevention of AF recurrence may not be achievable, consistent rehabilitation engagement demonstrates significant reductions in arrhythmia episode frequency, symptomatic burden, and may potentially lower the risks of stroke, heart failure, and other cardiovascular complications.

The advantage of our investigation was that the included studies were RCTs. The large participant populations and varying geographic locations contributed to a diverse and a comprehensive set of data. In addition, heterogeneity was systematically investigated through planned exploratory subgroup analyses and meta-regression. The low statistical heterogeneity (I²) observed across most analyses strengthens the reliability of our pooled estimates. However, we acknowledge several limitations. Firstly, significant heterogeneity existed in the cardiac rhythm status of patients at the onset of the rehabilitation intervention across studies (e.g., in sinus rhythm, in AF, or having recently undergone cardioversion). This lack of a uniform baseline clinical state may influence the estimated effect size of CR and complicates a direct, mechanistic interpretation of its effect on the arrhythmic substrate. Secondly, the baseline characteristics of patients and follow-up duration in included studies were heterogeneous, which may affect the comparability and interpretation of long-term outcomes. Thirdly, the included studies exhibited considerable heterogeneity in sample size (ranging from 47 to 382 participants) and follow-up duration (ranging from 12 weeks to 2.6 years), which may affect the comparability and interpretation of long-term outcomes. Fourthly, significant heterogeneity existed in the CR intervention protocols across studies. While all interventions included an exercise training component, they varied considerably in exercise type, intensity, frequency, duration per session, and overall intervention period. This heterogeneity in CR ‘dose’ and composition, while reflecting real-world clinical practice, complicates the interpretation of mechanistic pathways. Fifthly, several studies reported insufficient details on randomization, allocation concealment, and blinding for proper bias assessment; additionally, the post-ablation subgroup analysis included only six studies with a limited sample size, and we did not search grey literature—all of which may introduce potential bias and limit the generalizability of our findings. Sixthly, differences in trials included the design, analysis, outcome definitions (such as the way in which AF recurrence was determined) and to what extent these differences affected the results were unclear. Finally, included trials all recruited highly selected study populations. In two of the trials, the participants had AF and HF. In two of the trials, participants were required to have a body mass index (BMI) greater than 27 and 30, respectively. In one of the trials, all patients had early persistent AF and mild to moderate HF, and consented to undergo electrical cardioversion three weeks after enrollment. Another trial required the patients to have persistent AF and no AF recurrence during blanking period after catheter ablation. These factors potentially limited the applicability of this review to the broader group of AF patients.

Future research should address several key limitations identified in this review. First, larger prospective studies with standardized CR protocols are needed to enable meaningful comparisons across trials and to establish optimal exercise prescriptions for AF patients. Second, future meta-analyses should include larger sample sizes of post-ablation patients to adequately assess the potential benefits of CR in this population, and incorporate grey literature to reduce publication bias. Third, subgroup analyses by AF type (paroxysmal vs. persistent) are warranted to determine whether treatment effects vary across different clinical phenotypes. Fourth, dose–response analyses should be conducted to identify the optimal intensity, frequency, and duration of exercise training for maximizing clinical benefits. Finally, further trials should evaluate the impact of CR on hard clinical endpoints such as stroke, thromboembolic complications, cardiovascular death, and hospitalization, with longer follow-up periods to capture long-term outcomes.

Conclusion

Evidence from the included RCTs indicates that CR is associated with a significant reduction in AF recurrence in the overall AF population, but not in the exploratory subgroup of post-ablation patients. CR significantly improved exercise capacity, but had no significant effect on mortality and hospitalization. Further high-quality research is needed to investigate the effects of CR in AF patients, particularly to identify which subpopulations (e.g., specific types of AF, timing post-ablation) derive the greatest benefit from CR interventions.

Clinical message

The evidence suggested that CR demonstrated efficacy in preventing AF recurrence; however, this effect was not statistically significant in the post-ablation cohort. While CR significantly improved exercise capacity, it showed no significant impact on mortality or hospitalization rates. Future high-quality randomized trials are needed to evaluate the benefits and potential risks of CR for adults with different types of AF.

Supplemental Information

Supplemental Information 2. Raw Data.

Dataset exported from Cochrane Review Manager (RevMan 5.3).

Funding Statement

This work was supported by the Shanghai Municipal Health Commission Research Project (No. 20234Y0186), 2023 Annual Clinical Research Cultivation Project of Tongji Hospital of Shanghai (No. ITJ(QN)2304), Shanghai Hospital Development Center Foundation—Shanghai Municipal Hospital Rehabilitation Medicine Specialty Alliance (No. SHDC22023304), Shanghai Health System Key Supportive Discipline Project (No. 2023ZDFC0302) and 2023 Annual Clinical Research Training Project of Tongji Hospital of Shanghai (No. ITJ(ZD)2302). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.