Atrial fibrillation (AF) and heart failure (HF) are highly prevalent cardiovascular disorders associated with substantial morbidity, mortality, and healthcare resource utilization (1, 2). Concurrent AF and HF result in significantly worse prognoses compared to either condition alone. With global population aging, the incidence and prevalence of both conditions are progressively increasing (3, 4). The frequent coexistence of AF and HF is attributed to shared pathophysiological mechanisms and risk factors, exacerbating clinical outcomes when they overlap (5, 6). Clinical management complexity markedly increases among patients diagnosed simultaneously with HF and AF (HF + AF). On one hand, HF patients exhibit heightened sensitivity to fluctuations in heart rate and hemodynamics (7); on the other hand, persistent or recurrent AF exacerbates cardiac dysfunction, establishing a detrimental cycle of disease progression (8, 9). Consequently, optimization of therapeutic approaches for this high-risk cohort remains a significant clinical challenge.

Currently, therapeutic strategies available for managing HF + AF primarily encompass catheter ablation, rhythm control (RhC), rate control (RC), and combined rhythm and rate control (Rh + RC). CA frequently utilizes irrigated-tip radiofrequency ablation catheters, designed to cool the catheter-tissue interface to achieve deeper, well-demarcated lesions and to allow the application of higher energy, thus minimizing thromboembolic complications during left atrial procedures (10). CA seeks to restore and maintain sinus rhythm, demonstrating significant symptomatic improvement and enhanced exercise capacity (11). Although AF notably increases stroke risk and mortality, the capacity of CA to mitigate these risks and improve long-term survival remains inconclusive. Nonetheless, large observational series have reported notably low post-ablation stroke incidences (12).

RhC principally involves antiarrhythmic drugs (AADs) (such as amiodarone or sotalol) to achieve and maintain sinus rhythm. However, this strategy frequently yields suboptimal efficacy, as pharmacological interventions alone do not fully address the underlying electrophysiological substrate initiating and perpetuating AF (13). Additionally, AADs are associated with significant proarrhythmic potential and high discontinuation rates driven by adverse events (14, 15).

RC, primarily utilizing beta-blockers, calcium channel blockers, or digoxin, targets ventricular rate reduction to alleviate the cardiac load of HF. Previous evidence from major trials indicated that pharmacological rhythm control alone offered no clear survival advantage over rate control in patients with AF (16). Conversely, Rh + RC strategies have emerged as supplementary or alternative therapeutic options when monotherapies are insufficient (17). Nevertheless, the comparative effectiveness and appropriate patient selection criteria for these therapies remain unclear: while some data favor CA or RhC for enhanced cardiac functional outcomes, other evidence suggests RC or combined modalities provide similar efficacy in specific patient subpopulations. Consequently, balancing therapeutic efficacy, safety profiles, procedural complexity, and patient tolerance remains a critical yet unresolved clinical issue (18).

Currently available randomized controlled trials (RCTs) and traditional meta-analyses typically offer limited or direct pairwise comparisons (e.g., CA vs. medical therapy for short-term endpoints, or RhC vs. RC strategies for rhythm maintenance and ventricular rate control), resulting in narrow evidence bases that limit comprehensive comparative assessment across CA, RhC, RC, and Rh + RC strategies within an integrated analytical framework.

Addressing this knowledge gap, the present investigation employs a comprehensive network meta-analysis (NMA) to systematically evaluate and comparatively assess the efficacy and safety profiles of CA, RhC, RC, and Rh + RC interventions in patients with HF complicated by AF. Leveraging robust direct and indirect comparisons among high-quality RCT data, our analysis aims to provide clinicians with an evidence-based, detailed framework for therapeutic decision-making. Ultimately, this study endeavors to advance personalized clinical management of HF + AF patients, promoting optimal and evidence-driven selection of therapeutic interventions tailored to specific patient populations.

This Bayesian network meta-analysis, encompassing 19 RCTs and 3,471 patients, comprehensively compared the efficacy and safety of four therapeutic strategies—Rh, CA, RC, and RhC—in the treatment of AF with concurrent HF. Among these, RhC consistently demonstrated favorable outcomes, ranking highest for improving LVEF and reducing all-cause mortality. RC produced the greatest reduction in BNP, with RhC also performing well in this domain. Although the absolute improvement in LVEF with RhC was modest, even small gains have been associated with better NYHA class and exercise tolerance, indicating clinically meaningful benefits. RhC also showed the lowest incidence of adverse events, followed by Rh, suggesting good tolerability. These findings support the hypothesis that combining Rh and RC may yield synergistic effects, enhancing both cardiac function and patient survival.

Although the improvement in LVEF with CA was statistically significant, the absolute change—ranging from 0.28 to 0.62—remained modest. Nevertheless, such improvements have been linked to enhanced functional capacity, improved NYHA class, and better long-term prognosis in HF patients. Reductions in BNP observed with CA, particularly when compared to rate control and Rh + RC strategies, suggest improved myocardial stress profiles and potential reductions in the risk of decompensation and hospitalization. Thus, CA appears to confer both biomarker and clinical stability benefits.

CA demonstrated consistent superiority across multiple outcomes, including LVEF improvement, BNP reduction, and decreased rates of all-cause mortality, HF hospitalization, and AF recurrence. These results are consistent with prior conventional meta-analyses (35), which highlighted CA’s favorable effects on cardiac function and prognosis. Unlike earlier studies focused on direct comparisons, our network meta-analysis employed a broader evidence base, allowing simultaneous assessment of multiple strategies. This approach strengthens the comparative foundation for recommending CA while clarifying the clinical role of pharmacological alternatives.

These findings are reinforced by time-to-event analyses from pivotal trials. In the CABANA-HF cohort (20), catheter ablation reduced all-cause mortality by 18% (HR 0.82, P = 0.003) over 4 years, aligning with our pooled OR of 0.58 for CA vs. Rh + RC. This consistency across analytic methods strengthens the evidence for CA’s survival benefit.

CA and RhC also led to significant improvements in MLHFQ scores, reflecting enhanced patient-reported quality of life. Improvements in fatigue, physical function, and emotional wellbeing are particularly relevant in HF management. Notably, CA reduced all-cause mortality with an odds ratio of 0.58 compared to Rh + RC. Assuming a baseline mortality of 20%, this yields a number needed to treat (NNT) of approximately 10, suggesting a meaningful survival benefit.

Our findings are in line with the CASA-AF randomized trial series, particularly the studies by Haldar et al. (36) and Boyalla et al. (37), which compared CA and thoracoscopic surgical ablation (SA) in patients with long-standing persistent atrial fibrillation (LSPAF). Although these studies were not included in our analysis due to differences in intervention types, they reported that CA achieved comparable or superior outcomes in rhythm control, symptom relief, and cost-effectiveness, further supporting the utility of CA in complex AF populations.

While RhC and RC yielded intermediate benefits across several outcomes, they consistently underperformed relative to CA on composite endpoints. RhC also ranked lower in safety assessments, and RC did not demonstrate a clear advantage in tolerability. These findings underscore the need to tailor treatment choices to individual risk profiles and therapeutic objectives.

Heart failure-related hospitalizations represent a major burden for both patients and healthcare systems. The significant reduction in hospitalization rates observed with CA suggests its potential not only for rhythm control but also for stabilizing disease progression. This finding is consistent with results from CASA-AF trials by Haldar et al. (36) and Boyalla et al. (37), which reported improvements in quality-adjusted life years and symptom relief following CA.

Although CA and RhC showed trends toward reduced AF recurrence, statistical significance was not always achieved. Nonetheless, rhythm control remains clinically valuable. Maintaining sinus rhythm can alleviate AF-related symptoms, reduce thromboembolic risk, and enhance cardiac efficiency—particularly beneficial in patients with concurrent HF.

Despite CA’s superiority in clinical outcomes, it did not rank highest in safety metrics, highlighting the importance of careful patient selection, especially for those with multiple comorbidities. The CASA-AF program, including the health economic evaluations by Boyalla et al. (37) and Haldar et al. (38), demonstrated that CA was associated with fewer complications, shorter hospital stays, and superior cost-effectiveness compared to surgical ablation. Although our analysis did not reveal statistically significant differences in adverse event rates, SUCRA rankings suggested a slight safety edge for Rh + RC, indicating potential utility in high-risk groups.

A major limitation of our study is the absence of subgroup analyses based on HF phenotype (HFrEF, HFmrEF, HFpEF). Given the distinct pathophysiological mechanisms and differential treatment responses of these subtypes, this omission may obscure relevant heterogeneity. Unfortunately, most included RCTs lacked sufficient stratified data, precluding meaningful subgroup evaluation. Additionally, patients with HFmrEF were underrepresented, and tripartite classification was rarely reported. Future trials should address this gap by incorporating phenotype-specific data to guide more precise, individualized treatment recommendations.

A key strength of our analysis lies in its broad comparative scope, integrating both pharmacologic and invasive treatment strategies. By leveraging a Bayesian network meta-analytic approach, we synthesized a wide range of evidence to inform clinical decision-making and personalized care planning.

Our study employed rigorous methodological safeguards, including risk of bias assessment using ROB 2.0, GRADE grading of evidence certainty, and publication bias assessment via funnel plots. Most critical outcomes—such as all-cause mortality, hospitalization, and LVEF—were rated as having moderate to high certainty, supporting the credibility and clinical relevance of our conclusions.

Nonetheless, several limitations warrant consideration. First, the small number of direct head-to-head RCTs reduces confidence in certain indirect comparisons. Second, our inability to perform HF phenotype-specific analyses limits the applicability of findings to diverse patient populations. Third, most included trials lacked placebo control arms, constraining interpretation of absolute efficacy. Finally, the Bayesian network meta-analysis was conducted using Stata due to its accessibility and functional adequacy; however, it does not support reporting of heterogeneity indices (τ², I²) or convergence diagnostics (e.g., trace plots, Gelman–Rubin statistics). Although we assessed consistency and robustness via SUCRA, the absence of formal MCMC diagnostics remains a methodological limitation. Future studies should consider dedicated Bayesian platforms like WinBUGS or JAGS to enhance analytic rigor.

Although we intended to present subgroup forest plots for key outcomes such as cardiovascular death, all-cause mortality, stroke, and heart failure hospitalization, we found that these endpoints were not uniformly reported across the included studies. Most trials only provided data on all-cause mortality and HF-related hospitalization, while cardiovascular death and stroke were either absent, inconsistently defined, or reported without sufficient subgroup stratification. This limitation restricted our ability to generate comprehensive and statistically valid forest plots for these outcomes.

Notably, only a few high-quality RCTs—such as CABANA (19, 20). Di Biase et al. (24) and MacDonald et al. (26)—provided a more complete set of safety and efficacy endpoints. However, the limited number of such studies and the heterogeneity in their definitions made subgroup synthesis and network connectivity infeasible for all four outcomes. Future clinical trials should strive for standardized endpoint reporting, especially for safety events like stroke and cardiovascular mortality, to support more robust subgroup analyses in network meta-analyses.

We echo recent consensus from the COMET initiative calling for standardized outcome reporting in AF-HF trials. Future studies should prioritize: Stratification by HF phenotype (HFrEF/HFpEF); Clear differentiation of stroke subtypes; Separate reporting of cardiovascular vs. non-cardiovascular death.

Although safety events such as major bleeding and stroke are of critical importance, they were inconsistently reported across the included studies. Only a limited number of trials provided event-level safety data with numerical details. Moreover, there was heterogeneity in definitions and follow-up periods. Therefore, we summarized available safety data in Supplementary Table 5 for reference but did not include these outcomes in the quantitative network meta-analysis due to the risk of biased estimation. We encourage future trials to adopt standardized reporting of adverse events to facilitate more comprehensive comparative evaluations.

This network meta-analysis provides robust evidence supporting the clinical superiority of CA over other evaluated strategies across critical outcomes, including enhanced cardiac function, reduced mortality, hospitalization, and AF recurrence among HF patients with concomitant AF. RhC and RC remain viable therapeutic alternatives for selected outcomes, while Rh + RC strategies may confer specific benefits in certain patient subsets, particularly in quality of life improvement. Clinical decision-making should integrate patient-specific comorbidities, procedural risks, and long-term outcomes data to optimize individualized care. Finally, our findings highlight the necessity for future large-scale, rigorously designed RCTs incorporating detailed subgroup analyses to further refine and enhance personalized treatment strategies for HF patients complicated by AF.