Introduction

Atrial fibrillation (AF) is the most common arrhythmias and increases the risk of thromboembolic events [1]. The use of oral anticoagulant therapy remains the mainstay to prevent the AF-related stroke. The application of non-vitamin K antagonist oral anticoagulants (NOACs, including dabigatran, rivaroxaban, apixaban, and edoxaban) has generally increased due to their advantages over warfarin in terms of the reduced frequent monitoring of anticoagulant activity, and fewer drug-drug or drug-food interactions. Prior pivotal randomized clinical trials (RCTs) [2,3,4,5] and meta-analyses [6,7,8] have consistently showed that NOACs have relative efficacy and safety alternatives to warfarin in worldwide patients with AF. In addition, comparisons of the efficacy and safety between NOAC and NOAC have indicated that apixaban has the most favorable safety profiles [9].

With the rapid population ageing, AF patients in Asia suffer higher risks of stroke and intracranial bleeding than patients in other regions. As such, oral anticoagulation therapy for AF is particularly important in Asia. Two prior meta-analyses including the sub-analyses of NOAC trials suggest that the use of NOACs (standard dose in particular) is non-inferior to warfarin use in Asian patients with AF [10, 11]. However, the results of RCTs may be not necessarily consistent with those of real-world studies. More recently, several observational studies in Asia have compared the safety and efficacy of NOACs versus warfarin. In addition, the efficacy and safety profiles between NOAC and NOAC remain unclear, leaving physicians with difficulties in decision-making regarding the choice of NOACs. Therefore, we performed a meta-analysis to compare the efficacy and safety outcomes of NOACs in Asian patients with AF from the real-world settings.

Methods

We preformed this systematic review and meta-analysis according to the guidance from the Cochrane Handbook for Systematic Reviews, [12] and reported the results of this meta-analysis based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) recommendations [13].

Literature search

We systematically searched the electronic databases of PubMed and Embase to identify eligible observational studies from inception to June 2019. The search terms were applied as follows: atrial fibrillation OR atrial flutter AND non-vitamin K antagonists OR direct oral anticoagulants OR new oral anticoagulants OR novel oral anticoagulants OR dabigatran OR rivaroxaban OR apixaban OR edoxaban AND vitamin-K antagonists OR warfarin. We also searched the reference lists of included studies, or related reviews, editorials, and letters for additional reports. No linguistic restriction was applied in this study.

Study eligibility criteria

Observational studies were included according to the following criteria: (1) study population: Asian patients with non-valvular AF receiving at least one NOAC compared with warfarin, or compared with other NOACs; (2) interventions: any NOAC (dabigatran, rivaroxaban, apixaban, or edoxaban) and warfarin; (3) outcomes: studies reported at least one of the efficacy (stroke or systemic embolism [SSE], ischemic stroke, all-cause death) or safety (major bleeding, intracranial hemorrhage [ICH], and gastrointestinal [GI] bleeding) outcomes; and (4) effect estimates: odds ratios (ORs) and 95% confidence intervals (CIs).

We excluded studies according to the following criteria: (1) AF patients with certain interventions (e.g., cardioversion, coronary interventions, catheter ablation, or left-atrial appendage closure); (2) AF patients with certain diseases such as coronary artery disease, liver disease, diabetes, or cancer; (3) several forms of publications such as reviews, case reports, editorials, letters, or abstracts; (4) studies that reported the results by combining different NOACs; (5) if studies used the same data source, we selected the study with the longest study period.

Study selection and data extraction

All studies retrieved from the literature search were screened by two reviewers (Z.B.X and Y.Z) independently. Any disagreement or uncertainty was resolved by consensus or discussion with a third author (W.G.Z). We evaluated the potentially available studies according to titles, abstracts, and full-texts, sequentially. For each included study, we extracted information about study characteristics, patient demographics, anticoagulants, follow-up time, and outcomes.

Quality assessment

The Newcastle-Ottawa Scale (NOS) item were used to evaluate the study quality of observational studies. The NOS tool involved 3 domains with a total of 9 points: the selection of cohorts (0–4 points), the comparability of cohorts (0–2 points), and the assessment of the outcome (0–3 points) [14]. In this study, an NOS score of ≥ 6 points indicated a moderate-to-high quality, and an NOS score of < 6 points indicated a low quality [15].

Statistical analysis

In this study, we assessed the efficacy and safety of any NOAC (dabigatran, rivaroxaban, apixaban, or edoxaban) versus warfarin, and compared the efficacy and safety between NOAC and NOAC. The statistical analyses were performed using the Review Manager (RevMan) version 5.3 software (The Cochrane Collaboration 2014, Nordic Cochrane Centre Copenhagen, Denmark).

We collected the sample size and number of events in each group, and calculated the ORs and 95% CIs. If the number of events was unavailable, the expected number of events was calculated using the event rates: event number = (total patient number)×(event rate [per 100 patient years])×(follow-up time [years]) [16]. The Cochrane Q test and I2 index were used to measure the statistical heterogeneity, where P < 0.1 and I2 > 50% indicated a substantial heterogeneity, respectively. Given the heterogeneity inherent across the included studies, we should draw a relatively conservative conclusion based on the results of a random-effects model. According to the Cochrane handbook, the publication bias was assessed by using the funnel plots of the reported effect estimates. When we assessed the efficacy and safety of NOACs versus warfarin, the following sensitivity analyses were performed: (1) applying a fixed-effects model to re-perform the analysis; (2) only including the studies that did not use the method of propensity score matching; (3) only including studies with the NOS score of ≤ 7 points; (4) only including studies with the follow-up time of < 1 year; and (5) deleting the study with the largest statistical weight in the pooled analysis.

Results

Study selection

As shown in Supplemental Figure 1, the literature search retrieved 8,003 records from the PubMed and Embase databases. After the title and abstract screenings, the 65 remaining studies underwent full-text assessment. Thereinto, 47 studies did not meet with the pre-defined inclusion criteria (Supplemental Table 1): (1) clinical trials (n = 17); (2) study participants had a significant overlap (n = 12); (3) AF patients with certain comorbid diseases (n = 9); and (4) studies presented the results by combining different NOACs (n = 9). Finally, a total of 18 observational studies [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34] were included in this meta-analysis.

The main characteristics of the included studies are shown in Supplemental Table 2. Eleven studies applied the method of propensity score matching to balance patient characteristics between study groups. All the included studies had a moderate-to-high quality with an NOS score of > 6 points.

Efficacy and safety between NOACs and warfarin

Dabigatran versus warfarin

As shown in Table 1 and Fig. 1, compared with warfarin use, the use of dabigatran was associated with lower risks of SSE (OR = 0.78, 95% CI 0.71–0.85), IS (OR = 0.34, 95% CI 0.17–0.67), all-cause death (OR = 0.51, 95% CI 0.27–0.96), major bleeding (OR = 0.56, 95% CI 0.43–0.73), ICH (OR = 0.33, 95% CI 0.22–0.48), and GI bleeding (OR = 0.60, 95% CI 0.38–0.93).

Table 1 Efficacy and safety of NOACs versus warfarin for stroke prevention in Asian patients with AF
Full size table
Fig. 1
figure 1

Forest plot for comparing the efficacy (a) and safety (b) outcomes of dabigatran with warfarin in Asian patients with AF. AF, atrial fibrillation; SSE, stroke or systemic embolism; IS, ischemic stroke; ICH, intracranial hemorrhage; GI, gastrointestinal; CI, confidence interval

Full size image

Rivaroxaban versus warfarin

As presented in Table 1 and Fig. 2, compared with warfarin, rivaroxaban significantly reduced the risks of SSE (OR = 0.74, 95% CI 0.68–0.82), IS (OR = 0.53, 95% CI 0.31–0.91), all-cause death (OR = 0.41, 95% CI 0.26–0.66), major bleeding (OR = 0.54, 95% CI 0.44–0.67), ICH (OR = 0.35, 95% CI 0.19–0.67) and GI bleeding (OR = 0.59, 95% CI 0.46–0.77).

Fig. 2
figure 2

Forest plot for comparing the efficacy (a) and safety (b) outcomes of rivaroxaban with warfarin in Asian patients with AF. AF, atrial fibrillation; SSE, stroke or systemic embolism; IS, ischemic stroke; ICH, intracranial hemorrhage; GI, gastrointestinal; CI, confidence interval

Full size image

Apixaban versus warfarin

In comparison with warfarin, apixaban was associated with decreased risks of all-cause death (OR = 0.27, 95% CI 0.09–0.82), major bleeding (OR = 0.41, 95% CI 0.35–0.48), ICH (OR = 0.29, 95% CI 0.23–0.38) and GI bleeding (OR = 0.20, 95% CI 0.14–0.28), but showed marginally significant reductions in SSE (OR=0.70, 95%CI 0.49-1.01) and IS (OR=0.41, 95%CI 0.16-1.02) (Table 1 and Figure 3).

Fig. 3
figure 3

Forest plot for comparing the efficacy (a) and safety (b) outcomes of apixaban with warfarin in Asian patients with AF. AF, atrial fibrillation; SSE, stroke or systemic embolism; IS, ischemic stroke; ICH, intracranial hemorrhage; GI, gastrointestinal; CI, confidence interval

Full size image

Edoxaban versus warfarin

As shown in Table 1 and Supplemental Figure 2, edoxaban versus warfarin showed significant reductions in SSE (OR = 0.29, 95% CI 0.22–0.39) [19], IS (OR = 0.29, 95% CI 0.23–0.36) [19, 22], all-cause death (OR = 0.26, 95% CI 0.20–0.34) [22], major bleeding (OR = 0.19, 95% CI 0.14–0.25) [19, 22], ICH (OR = 0.14, 95% CI 0.09–0.23) [19, 22] and GI bleeding (OR = 0.17, 95% CI 0.07–0.43) [19, 22].

Sensitivity analysis

As shown in Supplemental Table 3, all the sensitivity analyses produced similar results to the main analyses, suggesting that compared with warfarin, all NOACs had lower or similar rates of thromboembolic and bleeding events.

Efficacy and safety between NOAC and NOAC

Dabigatran versus rivaroxaban

As shown in Table 2 and Supplemental Figure 3, there were no differences between dabigatran and rivaroxaban for the efficacy and safety outcomes including SSE (OR = 1.08, 95% CI 0.98–1.19), IS (OR = 1.04, 95% CI 0.94–1.15), all-cause death (OR = 0.99, 95% CI 0.62–1.57), major bleeding (OR = 1.11, 95% CI 0.68–1.79), ICH (OR = 0.96, 95% CI 0.80–1.15) and GI bleeding (OR = 1.15, 95% CI 0.95–1.41).

Table 2 Efficacy and safety between NOACs for stroke prevention in Asian patients with AF
Full size table

Dabigatran versus apixaban

In comparison with apixaban, dabigatran was significantly associated with increased risks of IS (OR = 1.38, 95% CI 1.18–1.61), all-cause death (OR = 1.51, 95% CI 1.17–1.95), and GI bleeding (OR = 3.20, 95% CI 2.27–4.51), but had no statistical differences in SSE (OR = 1.16, 95% CI 0.71–1.87), major bleeding (OR = 1.02, 95% CI 0.43–2.44), and ICH (OR = 1.30, 95% CI 0.77–2.21) (Table 2 and Supplemental Figure 4).

Rivaroxaban versus apixaban

Rivaroxaban had significantly elevated risks of SSE (OR = 1.31, 95% CI 1.13–1.51), IS (OR = 1.31, 95% CI 1.13–1.52), ICH (OR = 1.45, 95% CI 1.11–1.90), and GI bleeding (OR = 2.75, 95% CI 1.89–3.99), but showed similar rates of all-cause death (OR = 1.48, 95% CI 0.76–2.88) and major bleeding (OR = 1.34, 95% CI 0.72–2.52) as for apixaban (Table 2 and Supplemental Figure 5).

Of note, no syntheses conducted for outcomes regarding edoxaban versus other NOACs due to the insufficient studies.

Publication bias

As shown in Supplemental Figures 6–11, there were no obvious publication biases inspected by the funnel plots.

Discussion

In the real-world Asian patients with AF, our present meta-analysis first showed that (1) compared with warfarin, all NOACs reduced the risks of bleeding events and all-cause death, whereas dabigatran, rivaroxaban, and edoxaban were associated with decreased risks of SSE and IS; and (2) apixaban versus dabigatran was associated with decreased risks of IS and GI bleeding; and apixaban versus rivaroxaban was associated with reduced risks of SSE, IS, ICH, and GI bleeding. The real-world evidences from Asian patients with AF suggested that NOACs were non-inferior to warfarin, and apixaban might be a better option available compared with dabigatran or rivaroxaban.

Previously, Wang and colleagues [11] performed a meta-analysis by including Asian patients with AF from the NOAC trials for dabigatran (Randomized Evaluation of Long-Term Anticoagulation Therapy; RE-LY), rivaroxaban (Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation; ROCKET AF), rivaroxaban (Japanese-Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation; J-ROCKET AF), apixaban (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation; ARISTOTLE), and edoxaban (Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation-Thrombolysis in Myocardial Infarction 48-ENGAGE AF; TIMI 48). In their analyses, compared with warfarin, standard-dose NOACs are associated with lower risks of both efficacy (SSE and all-cause death) and safety (major bleeding and ICH) outcomes, whereas low-dose NOACs have a favorable safety profile with reductions in major bleeding and ICH. Subsequently, 2017 consensus on stroke prevention in Asians with AF recommends standard-dose NOACs as the first choice [35]. Of note, a limitation in the study of Wang et al. [11] was that the results were driven by combining different NOACs. In addition, whether NOACs in real-world daily practice would show the efficacy and safety differences from participants in the NOAC trials remains unclear. Our current real-world data support the findings of Wang et al. [11] and strengthen the validity of NOACs in Asia. The different types of NOACs were analyzed separately in our study, suggesting that dabigatran, rivaroxaban, and edoxaban were associated with lower risks of both efficacy and safety profiles compared with warfarin. Consistent with the sub-analysis of the ARISTOTLE trial [36], the real-world data also showed that the use of apixaban was safer than warfarin use with reductions in the bleeding risks in Asia. Of note, only two included studies [19, 22] reported the effectiveness and safety of edoxaban in real-world practice, and therefore, our data in relation to edoxaban have to be interpreted cautiously. With the rapid population ageing, data from observational studies would help overcome the under-use of anticoagulation therapy in Asia.

Among the included studies, a total of 11, 6, 3, and 2 studies reported that more than 50% of patients received reduced or low dose of dabigatran, rivaroxaban, apixaban, and edoxaban, respectively (Supplemental Table 4). In the study of Jeong et al. [18], 51.5% of patients received rivaroxaban 15 mg, but the use of both rivaroxaban 15 mg/day and 20 mg/day had similar efficacy and lower risks of major bleeding compared with warfarin use. Chan et al. [19] reported that reduced-dose NOACs (edoxaban 15–30 mg/day, apixaban 2.5 mg bid, rivaroxaban 10–15 mg/day, and dabigatran 110 mg bid) were associated with lower rates of thromboembolism and bleeding events compared with warfarin. In our meta-analysis, the subgroup analysis based on the NOAC dose could not be performed due to the limiting data.

Direct or indirect comparisons of NOACs for efficacy and safety profiles in worldwide AF patients have been reported in several studies [6, 9, 37], suggesting a better safety profile (i.e., reduced risk of major bleeding) but similar efficacy profile when comparing the use of apixaban to dabigatran or rivaroxaban. To the best of our knowledge, we first compared the efficacy and safety outcomes between NOAC and NOAC in Asian patients with AF. Our results supported apixaban as the most favorable NOAC, mainly manifesting as reduced rates of IS and GI bleeding for apixaban versus dabigatran, and lower risks of SSE, IS, ICH, and GI bleeding for apixaban versus rivaroxaban. These data might help clinicians better understand and guide the choice of anticoagulants in Asian patients with AF. However, the efficacy and safety between edoxaban with other NOACs could not be performed due to the insufficient studies. Further study with head-to-head direct comparisons would confirm the efficacy and safety differences between NOAC and NOAC. Given there are still no such RCTs currently, the real-world data with comparative assessment between NOAC and NOAC might help clinicians in decision-making for stroke prevention in Asia.

Limitations

Our meta-analysis had several limitations. First, the quality of anticoagulation activity in warfarin users and the adherence or persistence to NOACs were not considered. Second, based on the real-world data, the residual confounders should be considered when interpreting our findings. Third, 6 included studies were from Japan, 4 studies from Korea, 5 studies from China, 1 study from Singapore, 1 study from Malaysia, and 1 study from Israel. As such, 15 of 18 studies were from Japan, Korea, and China, limiting the generalization to all Asians with AF. Finally, the numbers of studies included for quantitative syntheses of the outcomes of interest were relatively small. As such, we could not perform the subgroup analyses to explore the source of heterogeneity across the included studies.

Conclusions

Based on current real-world evidences, NOACs were non-inferior to warfarin for AF stroke prevention in Asia. Apixaban might represent a better option compared with dabigatran or rivaroxaban.