Background
Non-valvular atrial fibrillation (NVAF) is the most common arrhythmia associated with severe thromboembolic events. Stroke caused by NVAF often can cause higher immobility and mortality than other stroke risk factors [
1]. Furthermore, the prevalence of atrial fibrillation (AF) is on the rise across the world especially in Asia [
2], where there is a large and rapidly ageing population. Therefore, stroke prevention is crucially important for Asian patients with NVAF [
3]. Although it has been used to prevent stroke for years, warfarin is still underused and under-dosed in Asian patients, and the quality of international normalized ratio (INR) control is substandard in Asia compared with its western counterpart [
4,
5]. This may be due to the fact that Asian patients are more sensitive to warfarin with a narrow INR range, accompanied by a higher risk of hemorrhagic complications [
6]. As a result, low-intensity warfarin is often prescribed in clinical practice which may contribute to the increasing risk of embolism.
Different from warfarin, novel oral anticoagulants (NOACs) are lower risk and are excellent in preventing stroke and limiting haemorrhage especially intracranial haemorrhage [
7,
8]. Moreover, a recent meta-analysis has suggested that NOACs are more effective and safer among Asian patients than non-Asians in terms of complications such as stroke or systemic embolism (SSE) and major bleeding [
9], which indicates that Asian patients would largely benefit from the development of NOACs. Furthermore, the usage of NOACs has constantly increased while aspirin prescription has gradually decreased among Asians in recent years [
10]. Hence, the rise of NOACs has revolutionized the field of NVAF management for Asian patients [
11].
Until now, several NOACs (dabigatran, rivaroxaban, apixaban and edoxaban) have been approved in Asian countries, and it is still unclear that which NOAC is more effective and safer. Although direct or indirect comparisons between NOACs have been published, these trails mainly focus on western countries and results from Asia are considerably limited. Considering the ethnic and regional differences, it is important to assess the most favourable oral anticoagulation for Asian patients. Randomized controlled trial (RCT) represents the gold standard for evaluating the clinical effectiveness of an intervention. However, results from real-world studies could perfectly complement RCTs due to the RCT’s ability to reflect a genuine clinical practice and sample larger populations [
12]. Comparison based on high-quality real-world studies is an alternative option in the absence of large RCTs.
Hence, the objective of this study is to conduct a network meta-analysis (NMA) comparing the clinical efficacy and safety of several NOACs in clinical practice among Asian patients with NVAF.
Methods
Search strategy
The literature search was implemented by two investigators in Embase, Medline, and the Cochrane Library from the databases’ inceptions through March 1, 2019. Search terms included atrial fibrillation, apixaban, dabigatran, rivaroxaban, edoxaban, real world, observational studies, and registry studies. The reference lists was also screened for included studies and relevant reviews to increase the sample size of literature reviewed. Only articles published in English were selected in this study for quality control. The detailed search strategies were presented in Additional file
4: Table S1.
Study selection
The population, interventions, comparisons, outcomes, and study design (PICOS) were used to define the eligibility criteria. The inclusion criteria were as follows: (1) Asian patients with NVAF; (2) treatments with NOACs (apixaban, dabigatran, rivaroxaban or edoxaban) for stroke prevention; (3) real-world studies including prospective or retrospective cohort studies; (4) adjusted hazard ratio (HR) using propensity score matching (PSM) or multivariate analysis. The following studies were excluded: (1) Asian patients with valvular AF or non-Asian patients; (2) treatment with aspirin (±clopidogrel); (3) non-full-text or studies not published in English; (4) case-control studies or cross-section studies.
Two investigators reviewed literature separately and evaluated the included articles by pre-specified selection criteria. Disagreements were resolved through discussions between the two investigators or consulting with a third investigator.
Two investigators independently extracted the effective information regarding the study design, treatments, patient characteristics, the number of enrolled patients, follow-up duration, and outcomes including SSE, ischemic strokes, all-cause deaths, major bleedings, and intracranial haemorrhages.
Quality assessment
Assessment of the risk of bias for included studies was performed using the Newcastle-Ottawa Quality Assessment Scale (NOS) which is recommended by the Cochrane Collaboration for observational studies [
13,
14]. The NOS has eight items within three domains: selection (representativeness), comparability (due to design or analysis), and outcomes (assessment and follow-up). A study scores one star for the satisfaction of each criterion, with the exception of the comparability domain (design or analysis), where a maximum of two stars can be awarded. In this study, publications that achieved eight or more stars on the NOS were considered as high quality, and moderate quality was defined as the achievement of six to seven starts, and less than six starts were considered low quality.
Statistical analysis
Network meta-analysis was conducted using the mvmeta software package in STATA14 software. NMA synthesizes data from a network of trials and provides an integration of direct evidence with indirect evidence producing a relative ranking of all treatments [
15], which is able to provide a solution to the present challenge of few head to head comparisons between different NOACs in Asians.
Adjusted HR for primary and secondary outcomes were estimated in this NMA. Primary outcomes included stroke or systemic embolism and major bleeding; four secondary outcomes were ischemic strokes, all-cause deaths, major bleedings, and intracranial haemorrhages. The hierarchy of the treatments was performed using the surface under the cumulative ranking (SUCRA) curve, where a larger SUCRA value represented a better rank of the treatment [
16].
Statistical heterogeneity will be assessed with P values and I2 statistics (percentage of total variation across studies due to heterogeneity). An I2 value over 50% indicates substantial heterogeneity while an I2 value under 50% indicates low or moderate heterogeneity. Either a random-effect or fixed-effect model was adopted based on the result of heterogeneity analysis. Network inconsistency was evaluated by an inconsistency plot to examine the differences between direct and indirect evidence. Potential publication bias was assessed through visual inspection of funnel plots.
Sensitivity analysis
Although potential confounders in these studies included in this NMA were adjusted using PSM or multivariate analysis, it is plausible that there are more confounding variables in real-word studies, especially in low-quality studies which included relatively smaller populations with inadequate comparability. Hence, a sensitivity analysis was conducted by excluding studies with low quality to improve the reliability of results from this NMA.
Discussion
Despite the numerous meta-analysis or NMA conducted in the field, the direct or indirect comparisons between NOACs focusing on Asian patients remain elusive. Through the NMA, we overcame the barrier of lacking head to head evidence in Asian patients and conducted an indirect comparison of several NOACs among patients with non-valvular atrial fibrillation. The present analysis suggested that apixaban, dabigatran, and rivaroxaban were superior to warfarin in reducing the risks of stroke and haemorrhage. Comparing with dabigatran, apixaban was associated with a lower risk of ischemic stroke, but no statistically significant difference was observed in terms of clinical safety. Moreover, we found that rivaroxaban was associated with a higher risk of major bleeding and ischemic stroke when compared with apixaban. However, there was no statistically significant difference in both safety and efficacy outcomes between rivaroxaban and dabigatran.
A recent study [
9] suggested that NOACs as a whole were superior to warfarin in lowering stroke and bleeding risks. The present analysis comparing three NOACs (apixaban, rivaroxaban and dabigatran) with warfarin, further suggested that apixaban, rivaroxaban, and dabigatran were safer and more effective in stroke prevention among Asians. In comparison with warfarin, apixaban can lower the risk of ischemic stroke by 61% (Fig.
3) and all-cause death rate by 71% (Fig.
6). Apixaban, rivaroxaban, and dabigatran also cause less bleeding compared to warfarin, which may be related to the high risk of warfarin-induced intracranial bleeding and the low quality of INR control in Asia [
33,
34].
Previous studies [
4,
35‐
37], which mainly included the studies from western countries, suggest no significance in efficacy between apixaban and dabigatran, but less major bleeding risk arises from the apixaban usage. However, the present analysis suggests that apixaban provides an additional 41% reduction on ischemic stroke risk (Fig.
3) without showing an improved safe outcome, which may be partly due to the commonly low-dose usage of dabigatran in clinical practice among Asian patients [
18,
22]. In agreement with previous studies [
4,
36‐
38], apixaban had less major bleeding complications compared with rivaroxaban (Fig.
4). Furthermore, the present analysis demonstrates that rivaroxaban can increase the chance of ischemic stroke when compared to apixaban among Asians patients. However, due to an insufficient study sample size selected for this analysis, further research is required in order to draw a more conclusive result.
Evidence from RCTs is usually considered more reliable than real-world studies such as in the following studies. Nevertheless, performing an RCT requires strict eligibility criteria within a relatively small patient population, which limits the generality of results to be used as a clinical guide [
39]. Real-world studies can estimate a much broader population not limited to age and other diseases, while being closer to a clinical practice [
40]. Thus, real-word studies are considered a potential alternative to complement the evidence from RCTs [
41]. Furthermore, propensity score matching (PSM) is widely performed in recent real-world studies to balance the distribution of biases and confounders between groups in order to achieve the purpose of simulating random assignment [
12]. The majority of studies in this NMA performed PSM or multivariate analysis to maximally eliminate the influence of confounding factors. Although the results of this analysis are somewhat different from previous investigations, the findings still reflect the practical clinical benefits of oral anticoagulants among Asians patients. Moreover, the results from clinical practice could provide a new perspective on the use of NOACs for Asians.
In this analysis, we found that low-dose NOACs, especially dabigatran, were more widely used in clinical practice among Asian patients [
18,
22], although a recent study [
42] suggests that standard dose NOACs are more effective in stroke prevention without increasing risks of bleeding for Asians. However, considering the increasing complexity and variety of clinical practices, such as serious comorbidities and advanced age, it is understandable that low-dose oral anticoagulants are more likely to be prescribed by clinicians in order to avoid severe bleeding complications [
5,
6,
43]. A recent observational study from Korea suggested that both dose dabigatran displayed similar efficacy outcomes; moreover, dabigatran 110 mg performed better than dabigatran 150 mg with regards to lowering bleeding risk [
44]. At present analysis, we also found that although most of the Asian patients in this NMA received low-dose dabigatran (110 mg), dabigatran can effectively minimize the danger of ischemic stroke and major bleeding compared with warfarin.
Several limitations are present in this analysis. Firstly, the initial aim of this analysis was to compare the differences between four NOACs with each other, including apixaban, dabigatran, rivaroxaban, and edoxaban, but the studies that compared edoxaban with other anticoagulants for NVAF in the East Asia-Pacific region is limited. Thus, this analysis mainly studied the differences among apixaban, dabigatran, and rivaroxaban. Secondly, this NMA merely evaluated the mixed dose oral anticoagulants and the difference between regular and reduced dose was not tested mainly due to the lack of studies which analyzed the different doses of NOACs separately in Asia [
25,
12]. Therefore, further studies are expected to be conducted to comprehensively understand the impact of dosage of oral anticoagulants on Asians. Thirdly, the studies included in this analysis were mainly from East Asian and Southeast Asian countries and regions, namely Taiwan, Japan, and Korea. Studies from the other parts of Asia, especially larger populated countries such as India and Pakistan, were missing due to the language barrier since only English literature were selected for this study. Nevertheless, considering the great quality control of publications in English journals due to the peer-review process, the bias is likely to be minimal.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.