Atrial fibrillation (AF) is one of the most common arrhythmias in older patients, affecting nearly 33 million individuals worldwide (Benjamin et al., 1998). Patients with AF have a 1.5-fold higher risk of mortality and 2.5-fold higher risk of stroke than those without AF (Go et al., 2001). Long-term anticoagulant therapy is essential for reducing the risk of thromboembolic events in these patients. Liver disease, caused by hepatitis B or C virus infection, alcohol-related liver disease, and metabolic dysfunction-associated fatty liver disease, leads to hepatic impairment and abnormal coagulation, contributing to a global public health burden (Asrani et al., 2019; Xiao et al., 2025). Particularly in patients with cirrhosis, coagulation balance is often severely disrupted, elevating the risk of both venous thromboembolism (VTE) and bleeding (Tripodi and Mannucci, 2011). Therefore, clinical guidelines and expert consensus recommend anticoagulation in patients with AF and liver disease to prevent ischemic stroke or systemic embolism (Joglar et al., 2023; Husted et al., 2014).

The most commonly used oral anticoagulants include vitamin K antagonists (VKAs) and direct oral anticoagulants (DOACs). DOACs are favored over VKAs owing to their predictable pharmacokinetics, fewer food and drug interactions, and the absence of a need for regular therapeutic drug monitoring (Kirchhof et al., 2016; Chen et al., 2020). Although the safety and effectiveness of DOACs have been established in patients with AF and cancer or chronic kidney disease, their use in patients with liver disease remains controversial because such patients are frequently excluded from randomized controlled trials (RCTs) (Khorana et al., 2017; Jones et al., 2024; Schrag et al., 2023; Hydes et al., 2023). Previous meta-analyses have suggested that DOACs may be effective and safe alternatives to VKAs in patients with AF and liver disease (Chen et al., 2022). However, conflicting findings have been reported. Mort et al. reported that the risk of spontaneous bleeding was elevated in patients with liver disease receiving DOACs (Mort et al., 2021), while Song et al. found increased readmission rates associated with DOAC use compared to warfarin (Song et al., 2024). Moreover, most studies have neither compared individual DOACs nor evaluated the effects of different dosage regimens.

Given the complex balance between bleeding and thrombotic risks in patients with AF and liver disease, further evidence is needed to guide clinical decision-making (Senzolo and Garcia-Pagan, 2023). A recent meta-analysis by Miranda Maria et al. (2025) specifically examined patients with both AF and liver cirrhosis and found that DOACs were associated with lower risks of major bleeding, gastrointestinal (GI) bleeding, and all-cause mortality than VKAs, with no significant difference in thromboembolic events. However, that study focused exclusively on patients with cirrhosis and did not perform subgroup analyses by geographic region or follow-up duration, nor did it assess differences among individual DOACs or compare different dosage regimens. Additionally, their study did not apply the GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) framework to assess the certainty of evidence (Miranda Maria et al., 2025).

To address these limitations, we conducted a comprehensive systematic review and meta-analysis that incorporated real-world data from studies on patients with AF and liver disease. In addition to evaluating the effectiveness and safety of DOACs versus VKAs, we performed subgroup analyses by geographic region, follow-up duration, DOAC type (apixaban vs rivaroxaban), and dose (standard vs. low-dose). We also applied the GRADE framework to assess the quality of the evidence. These enhancements aim to provide more detailed and clinically actionable guidance on anticoagulation strategies for this high-risk patient population.

This systematic review and meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Supplementary Table S1) (Schulman and Kearon, 2005). The study was prospectively registered with PROSPERO (registration ID CRD42024623387).

For patients with deep vein thrombosis (DVT)/pulmonary embolism (PE) and liver disease or cirrhosis with portal vein thrombosis (PVT), relevant guidelines recommend DOACs to prevent thrombosis (European Association for the Study of the Liver, 2022; Northup et al., 2021). However, AF management guidelines do not provide specific recommendations for oral anticoagulants (OACs) selection in patients with liver disease (Joglar et al., 2023). Although high-quality RCTs are lacking, recent large-scale real-world data provide evidence supporting OACs selection. Previous meta-analyses suggest that DOACs and VKAs have comparable effectiveness in preventing thromboembolism, with DOACs associated with lower bleeding risk (Chen et al., 2022; Zhao et al., 2023; Li et al., 2023), but there is also a meta-analysis indicating that the IS/SE risk of DOACs is lower than that of VKAs (Huang et al., 2021). Therefore, the results of different meta-analyses are controversial. These earlier analyses often included small sample sizes and primarily focused on comparisons between DOACs and VKAs without exploring heterogeneity across patient subgroups. With the emergence of real-world data, updated meta-analyses are necessary to reflect broader populations and nuanced treatment considerations (Douros et al., 2024; Simon et al., 2024; Lawal et al., 2023). To address these research gaps, our study incorporated recent evidence and performed subgroup analyses by region and follow-up duration. We also evaluated the effectiveness and safety of different DOAC types and dosing regimens, contributing to a more refined understanding of treatment options.

Our findings showed that, in patients with liver disease and AF, DOACs were superior to VKAs in reducing major bleeding, intracranial bleeding, GI bleeding, and all-cause mortality. No significant differences were observed between groups in terms of IS/SE. These results are consistent with studies in patients with AF without liver disease, reinforcing the notion that DOACs offer a more favorable net clinical benefit than warfarin (Carnicelli et al., 2022). Most retrospective studies lack detailed clinical data on liver disease severity, including ascites, hepatic encephalopathy. And the ischemia and bleeding scores of patients with AF. Additionally, the absence of international normalized ratio (INR) data prevents assessment of the anticoagulant effect of VKAs. These limitations likely contributed to the high heterogeneity observed in the results.

In the United States, the use of DOACs in cirrhotic patients with AF increased substantially between 2012 and 2019 (from 18.7% to 77.6%) (Simon et al., 2023; Pereira Portela et al., 2024). This trend has outpaced the quality of available evidence, highlighting a disconnect between practice and data. In our cirrhosis subgroup analysis, the risks of major, GI, and intracranial bleeding were significantly lower with DOACs than with VKAs, whereas no significant differences were found in IS/SE or all-cause mortality. Nisly SA et al. reported no significant difference between DOACs and traditional anticoagulation in patients with cirrhosis (Nisly et al., 2021), while recent large-scale retrospective studies suggest that DOACs are safer. In our updated meta-analysis, DOACs demonstrated greater safety than VKAs, consistent with the findings of Miranda Maria et al. (2025). Furthermore, the present study confirmed that DOACs outperform VKAs in terms of composite effectiveness and safety outcomes.

Ethnic and regional variations may influence both liver disease etiology and thromboembolic risk. For example, thromboembolic events are more frequent in Asian populations than in non-Asian populations (Romiti et al., 2023). Moreover, alcoholic liver disease and nonalcoholic fatty liver disease are more prevalent in Europe and the Americas, whereas viral hepatitis predominates in Asia (Shepard et al., 2005; Blachier et al., 2013; Younossi et al., 2023). Our regional subgroup analysis demonstrated that the safety profile of DOACs is consistently better than that of VKAs in North America, Asia, and Europe. However, in European patients with cirrhosis, DOACs and VKAs showed comparable effectiveness and safety, although this conclusion was based on a limited number of studies and requires further validation. DOACs are particularly effective in reducing GI bleeding in Asian patients and intracranial bleeding in American patients. Since AF patients with liver disease often require long-term anticoagulation, we also analyzed outcomes according to follow-up duration. When follow-up was < 1 year, GI bleeding risk was similar between the groups; however, with follow-up beyond 1 year, DOACs significantly outperformed VKAs in reducing GI bleeding. These findings suggest that DOACs offer sustained benefits.

All oral anticoagulants undergo hepatic metabolism to varying extents (Speed et al., 2023), which raises concerns about increased drug exposure and bleeding risk in patients with liver impairment. To mitigate this, clinicians often reduce DOAC doses, although this may compromise their effectiveness. Our analysis found no significant difference in thrombotic or bleeding risks between low- and regular-dose DOACs. However, regular-dose DOACs are associated with lower all-cause mortality, suggesting a potential clinical advantage. This finding aligns with previous research (Chong et al., 2021; Cho et al., 2019), although it is important to consider that patients receiving lower doses tend to be older and have more comorbidities, which may underestimate the true benefits of standard dosing.

Different DOACs have demonstrated varied clinical outcomes (Ray et al., 2021; Dawwas et al., 2022). However, the optimal DOACs for patients with liver disease or cirrhosis remains unclear. Our results suggest that apixaban may be safer than rivaroxaban, with lower risks of major and GI bleeding, while both agents showed comparable risks of IS/SE, intracranial bleeding, and all-cause mortality. Based on pharmacokinetic studies, Frost et al. reported similar apixaban exposure between patients with mild-to-moderate hepatic dysfunction (Child-Pugh A/B) and healthy controls (Frost et al., 2021), whereas Kubitza et al. found that rivaroxaban exposure in patients with moderate hepatic impairment was more than twice that observed in patients with mild impairment or normal liver function (Kubitza et al., 2013). In addition, multiple studies have reported a higher risk of GI bleeding with rivaroxaban in patients with AF without liver disease (Ingason et al., 2021; Ray et al., 2021). These results suggest a lower risk of bleeding with apixaban in the clinical setting. Nonetheless, apixaban and rivaroxaban have demonstrated similar effectiveness and safety in patients with cirrhosis. This may be attributed to the higher biliary excretion of apixaban than that of rivaroxaban, resulting in greater hepatic accumulation and potential toxicity in severe liver dysfunction (Douros et al., 2024; Graff and Harder, 2013). This study updates the meta-analysis on the effectiveness and safety of DOACs in patients with liver disease and AF, offering reference for the advancement of guidelines or consensus. In addition, our results provide accurate suggestions for clinical administration in DOACs selection. Future research should prioritize personalized anticoagulation strategies based on liver disease severity, bleeding/thrombosis risk, and drug interactions. Further investigation into other DOACs, such as dabigatran and edoxaban, is also needed. Multi-center RCTs are essential to strengthen the current evidence base.

Our study has several limitations. First, most of the included studies were observational, RCTs are required to confirm these findings. Second, despite the use of multivariable adjustment methods, residual confounding factors cannot be ruled out. Notably, data on concomitant antiplatelet therapy, such as aspirin or P2Y12 inhibitors, which could influence bleeding risk, are often unavailable. Third, the severity of liver disease (e.g., Child–Turcotte–Pugh classification) has not been consistently reported, limiting stratified analyses. Fourth, adherence to long-term anticoagulant therapy and attainment of INR targets with VKA treatment could not be assessed in most studies (e.g., Time in Therapeutic Range (TTR) for patients treated with VKAs). Fifth, due to the limitation of the included literature, some secondary outcomes, including readmission risk, were not assessed. Finally, due to limited data, we were unable to evaluate dabigatran or edoxaban, and comparisons among DOACs were restricted to apixaban and rivaroxaban.

In patients with liver disease, DOACs were as effective as VKAs and demonstrated superior safety, including those with cirrhosis. DOACs consistently showed better outcomes across regions, except in European patients with cirrhosis, in whom no difference was observed. A longer follow-up was associated with a reduced risk of GI bleeding with DOACs. Although thrombotic and bleeding risks were similar between dosing regimens, Regular-dose DOACs reduced all-cause mortality. Apixaban has a better safety profile than rivaroxaban in liver disease, supporting its preferential use in individualized anticoagulation strategies. However, there was no difference in the effectiveness and safety of rivaroxaban and apixaban in patients with cirrhosis.