Included studies
A total of 1852 unique records were identified from various data sources for the review of stroke prevention in AF ().
The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow chart for review of stroke prevention in AF.
Twenty-three completed eligible RCTs were identified for inclusion in the review, with a total of 41 associated references for these trials.95–135 No ongoing trials were identified. A summary of the characteristics of the 23 trials is presented in . Twenty of the trials were multicentre; two trials were each conducted in two centres; and one trial was conducted in one centre. The majority of the multicentre trials were conducted across several countries in North and South America, Europe, Asia, Russia and Israel, Australia and South Africa. The two-centre trials were conducted in one country: one in China and the other in Denmark. The single-centre trial was conducted in Denmark. Sixteen of the trials were Phase III studies and seven were Phase II studies. The number of patients randomised across the 23 trials ranged from 75 to 21,105, with a total of 94,656 patients, of whom 97% (91,333) were from the Phase III studies. Thirteen studies: six Phase III and seven Phase II studies examined a NOAC. Four studies examined edoxaban, three each examined apixaban and dabigatran, two examined rivaroxaban and one examined betrixaban.
Characteristics of 23 included randomised trials in stroke prevention in AF
Eligibility criteria for patient participation were similar across studies: all patients having non-valvular AF, whether new or existing, and including paroxysmal, persistent or permanent types. Diagnosis of AF was predominantly by electrocardiography. In a few cases, Holter recording, pacemaker or other intracardiac recording was used. The mean age of included patients was reported in only 61% of the studies and this ranged from 63.3 to 81.5 years. The percentage of male patients was reported in 78% of the studies, and this varied significantly across the studies, ranging from 44.9% to 82.9%. Mean BMI was not often reported and ranged from 24.4 to 30.5 kg/m2. Percentage of patients with previous stroke, hypertension and chronic heart failure varied significantly across the studies, ranging from 5% to 63.8%, from 38% to 93.7%, and from 0% to 100%, respectively. Bleeding risk among patients was assessed predominantly with the CHADS2 scoring system.
Warfarin was examined in all but two of the 23 of the included studies, against a NOAC in 12 studies, and against aspirin in nine studies. Standard intensity warfarin (INR 2–3) was examined by all of the studies, although in a few studies the warfarin arm was a mixture of low intensity (INR < 2) and standard intensity, in unknown proportions. Across all of the studies, mean TTR for warfarin ranged from 45.1% to 83% of the treatment duration. One study98 compared both low intensity warfarin (INR < 2) and standard intensity (INR 2.5–3.5) dicoumarol with aspirin, but the mean TTR was not reported for the standard intensity dicoumarol arm. The doses of NOACs we examined were edoxaban 30 mg, 45 mg, and 60 mg od and 30 mg and 60 mg bd; apixaban 2.5 mg and 5 mg bd; dabigatran 50 mg, 110 mg, 150 mg and 300 mg bd; rivaroxaban 15 mg and 20 mg od; and betrixaban 40 mg, 60 mg and 80 mg od. Examined aspirin dosages ranged from 75 mg to 325 mg od.
Treatment duration in the edoxaban and dabigatran studies was predominantly 3 months, although one study reported mean treatment durations of 24 months and another reported a median treatment duration of 29.8 months. Mean treatment duration for apixaban studies ranged from 13.1 to 21.6 months and one study reported 3 months’ treatment duration. The two studies on rivaroxaban reported 30 months’ treatment duration and a mean treatment duration of 19.4 months, respectively. Mean treatment duration 4.9 months was reported in the betrixaban study. Treatment duration was similar for each comparator in almost all the NOAC studies. Reported efficacy and safety outcome types were similar across studies and these were reported at the end of the treatment periods. All 23 studies reported data on stroke, 15 studies reported data on MI, 18 studies reported data on major bleeding, 12 studies reported data on CRB, and 18 studies reported data on all-cause mortality. Fifteen of the 23 studies, including all the NOAC studies, were sponsored by pharmaceutical companies. Six studies were funded by grants from medical research bodies although two of these grants contained contributions from a pharmaceutical company. Sponsor detail was not reported in two studies. In most of the pharmaceutical company sponsored studies, the sponsor(s) had influence on the study design, data management and analysis.
Results on clinical effectiveness and safety
The 27 different interventions considered in the 23 trials are listed in –, which show the number of patients analysed and the number of outcome events for each outcome reported in each trial. We performed NMAs for seven outcomes: stroke or SE, ischaemic stroke, MI, major bleeding, CRB, intracranial bleeding and all-cause mortality. Arms that were considered not to provide any evidence of interest to inform health decisions in the UK were excluded from the analyses. Specifically, we excluded the warfarin arm with INR range 1.6–2 from the AF-VKA-ASA-CHINA trial,122 the warfarin arm with INR range of < 2 from PATAF,98 the placebo arm from AFASAK,95 and the two warfarin arms with a fixed daily dose from AFASAK II.97
List of distinct interventions examined by included randomised trials in stroke prevention in AF
Safety outcomes reported by 23 included randomised trials in stroke prevention in AF: number of events for each outcome in each trial
Efficacy outcomes reported by 23 included randomised trials in stroke prevention in AF: number of events for each outcome in each trial
We defined two independent nodes for warfarin interventions, labelled as ‘warfarin (INR 2–3)’ and ‘warfarin (INR 3–4)’, respectively. The first of these formed the reference treatment across all networks in the AF review. We included in ‘warfarin (INR 2–3)’ the trials with a therapeutic INR range of 2–3 (e.g. ACTIVE W,100 AFASAK95), as well as some interventions with an INR range of 2.5–3.5 (AF-EDOX-VKA-ASIA115 and PATAF98) or 2.0–4.5 (SPAF II96). In some trials the INR range for some of patients in the warfarin arm was subtherapeutic (< 2.0), so that the total INR range was 1.6–3.0. These interventions were excluded from the main analysis, but merged with the INR 2–3 node in a sensitivity analysis. As a consequence, there were three two-arm trials (J-ROCKET AF,120 Chinese ATAFS99 and AF-ASA-VKA-CHINA135) that were included only in sensitivity analyses.
We also defined two independent nodes for antiplatelet interventions (‘aspirin’ or ‘aspirin plus clopidogrel’), using the cut-off point of 150 mg, with the understanding that daily doses above that were appropriate for stroke prevention in AF, whereas lower doses were appropriate for secondary prevention of cardiovascular events. The dose range considered in the AVERROES trial105,116,117,121 (81–324 mg od) was much wider than in any other trial, and we included this intervention in the lower-dose node (< 150 mg od) because some patients from that study had received a low daily dose. As a sensitivity analysis, we excluded the AVERROES trial105,116,117,121 from the network. Finally, our main analysis used a binomial model, assuming equal follow-up times across arms within trials and ignoring some variations in how results were reported. We undertook a separate analysis for all outcomes taking into account the differences in duration of follow-up within and between trials, and the differences in the definition of event used across trials (e.g. total number of events vs. first events only).
Results are presented as follows for each of the six outcomes. First, we provide network plots to illustrate the comparisons of interventions made in the different trials. Second, we illustrate the risk-of-bias assessments specific to the outcome for each trial included in the network. Third, we present results tables for each intervention compared with the reference treatment (warfarin with a target INR range of 2–3). Fourth, we present results tables for pairwise comparisons among licensed doses of the NOACs. For both sets of results tables, posterior median ORs and 95% credible intervals from Bayesian fixed-effects analyses are shown, although we refer to the latter as CIs for convenience. In these tables we present results separately for any available direct evidence, for any indirect comparisons that can be made (excluding the direct evidence) and for the NMA (which combines the direct and the indirect evidence). Comparisons from the NMA with a ratio between interval limits of > 9 were considered ‘imprecisely estimated’ and are presented at the bottom of each table (note that calculation of indirect evidence was not undertaken for imprecisely estimated comparisons). A summary of results across outcomes is provided at the end, in the form of a ‘rankogram’, which illustrates the probability that each treatment is best, second best, and so on, for each outcome. Last, forest plots of all contributing data, with ORs calculated using standard frequentist methods, are included in Appendix 2.
Stroke or systemic embolism
Sixteen studies reported the number of stroke or SE events, and the other seven trials reported the number of stroke events, so that the resulting network was based on data from all 23 trials, comparing a total of 26 interventions (). There were 3217 stroke or SE events. Twenty studies were included in the main analysis, with the remaining three included only in sensitivity analyses. The thicker lines joining interventions, which mainly correspond with comparisons between licensed doses of NOACs and warfarin (INR 2–3) represent the larger (mainly Phase III) trials. Similarly, the larger green circles represent the interventions to which the largest number of patients were randomised. Importantly, there were no direct comparisons between different NOACs, although there were numerous comparisons between different doses of the same NOAC in mainly Phase II trials, and some such comparisons in larger trials. Therefore, comparisons between the effects of different NOACs need to be inferred from the network (indirect evidence).
Network plot for stroke or SE (stroke prevention in AF). a, Excluded interventions that were included in sensitivity analyses.
shows risk-of-bias judgements for studies reporting stroke or SE. The studies were at mixed risks of bias: there were concerns about lack of blinding of participants for most trials, and about lack of allocation concealment and blinding of outcome assessment in some.
Included trials and risk-of-bias assessment for stroke or SE (stroke prevention in AF)
, which shows comparisons of licensed doses with warfarin (INR 2–3), suggests that both low- and high-dose antiplatelet drugs increase the risk of stroke or SE compared with warfarin (INR 2–3). Among NOACs, there was some evidence that apixaban [5 mg bd (bd)], dabigatran (150 mg bd), edoxaban (60 mg od) and rivaroxaban (20 mg od) reduce the risk of stroke or SE compared with warfarin (INR 2–3). Most other comparisons were imprecisely estimated. Comparisons among licensed doses of NOACs were almost all based on indirect evidence (). Among the comparisons that were not classified as imprecisely estimated, there was some evidence that edoxaban (60 mg od) and rivaroxaban (20 mg od) increase the risk of stroke or SE compared with dabigatran (150 mg bd).
Results for stroke or SE (stroke prevention in AF): comparisons with warfarin (INR 2–3)
Results for stroke or SE (stroke prevention in AF): NOACs (licensed doses only)
Results from a supplementary analysis taking into account the differences in duration of follow-up within and between trials, and the differences in the definition of event used across trials (e.g. total number of events vs. first events only), are presented in and . They are very similar to those for ORs.
Results for stroke or SE (stroke prevention in AF): comparisons with warfarin (INR 2–3) – sensitivity analysis using HRs instead of ORs
Results for stroke or SE (stroke prevention in AF): NOACs (licensed doses only): sensitivity analysis using HRs instead of ORs
As a post hoc sensitivity analysis, we fitted a fixed-effects meta-regression model using the mean TTR for warfarin patients (see ) as a covariate and the mean log-odds ratio (log-OR) from each pairwise comparison (with warfarin as the reference category) as the response variable. There was little evidence of effect modification due to mean TTR (estimated coefficient 0.0021 with 95% CI −0.07 to 0.08 per 1% increase). The model fit indices were very similar with and without the covariate.
Ischaemic stroke
Fourteen studies reported on 2228 ischaemic stroke events, leading to a connected network comparing a total of 15 interventions (). Twelve studies were included in the main analysis, with the remaining two included only in sensitivity analyses. The studies were at mixed risks of bias (). There were concerns about lack of blinding of participants for most trials, and about lack of allocation concealment and blinding of outcome assessment in one trial (AF-ASA-VKA-CHINA,135 only included in sensitivity analyses due to implementation of warfarin within non-standard INR range).
Network plot for ischaemic stroke (stroke prevention in AF). a, Excluded interventions that were included in sensitivity analyses.
Included trials and risk-of-bias assessment for ischaemic stroke (stroke prevention in AF)
, which shows comparisons of all interventions with warfarin (INR 2–3), suggests that both low- and high-dose antiplatelets increase the risk of ischaemic stroke compared with warfarin (INR 2–3). Among NOACs, there was some evidence that dabigatran (150 mg bd) reduces the risk of ischaemic stroke compared with warfarin, whereas edoxaban (30 mg od) increases that risk. There was little evidence that the risk of ischaemic stroke differed between licensed doses of NOACs ().
Results for ischaemic stroke (stroke prevention in AF): comparisons with warfarin (INR 2–3)
Results for ischaemic stroke (stroke prevention in AF): NOACs (licensed doses only)
In a sensitivity analysis to take into account the differences in duration of follow-up, NMA results were as presented in and , and show very similar results.
Results for ischaemic stroke (stroke prevention in AF): comparisons with warfarin (INR 2–3) – sensitivity analysis using HRs instead of ORs
Results for ischaemic stroke (stroke prevention in AF): NOACs (licensed doses only) – sensitivity analysis using HRs instead of ORs
Myocardial infarction
A total of 15 studies reported 1334 MI events, leading to a network of 16 interventions (). Thirteen studies were included in the main analysis, with the other two included only in sensitivity analyses. The studies were at mixed risks of bias (). There were concerns about lack of blinding of participants for most trials, and about lack of allocation concealment and blinding of outcome assessment in some.
Network plot for MI (stroke prevention in AF). a, Excluded interventions that were included in sensitivity analyses.
Included trials and risk-of-bias assessment for MI (stroke prevention in AF)
shows weak evidence that dabigatran (110 mg bd), dabigatran (150 mg bd) and edoxaban (30 mg od) increase the risk of MI compared with warfarin (INR 2–3), and weak evidence that rivaroxaban (20 mg od) decreases risk of MI compared with warfarin (INR 2–3). None of the interventions was superior or inferior to warfarin (INR 2–3). The pairwise comparisons of licensed NOACs, presented in , show weak evidence that dabigatran (150 mg bd) increases the risk of MI compared with apixaban (5 mg bd), and evidence that rivaroxaban (20 mg od) reduces the risk of MI compared with dabigatran (150 mg bd). Results were similar in a sensitivity analysis, taking into account the differences in duration of follow-up within and between trials, and the differences in the definition of event used across trials (e.g. total number of events vs. first events only) ( and ).
Results for MI (stroke prevention in AF): comparisons with warfarin (INR 2–3)
Results for MI (stroke prevention in AF): NOACs (licensed doses only)
Results for MI (stroke prevention in AF): comparisons with warfarin (INR 2–3): sensitivity analysis using HRs instead of ORs
Results for MI (stroke prevention in AF): NOACs (licensed doses only): sensitivity analysis using HRs instead of ORs
Major bleeding
Eighteen studies reported 4314 major bleeding events, leading to a network of 24 interventions (). Seventeen studies were included in the main analysis, with the remaining study included only in sensitivity analyses. These studies were at mixed risks of bias (). There were concerns about lack of blinding of participants for most trials, and about lack of allocation concealment and blinding of outcome assessment in some.
Network plot for major bleeding (stroke prevention in AF). a, Excluded interventions that were included in sensitivity analyses.
Included trials and risk-of-bias assessment for major bleeding (stroke prevention in AF)
There was weak evidence that antiplatelet therapy (< 150 mg od) reduced major bleeding compared with warfarin (INR 2–3). There was evidence that apixaban (5 mg bd), dabigatran (110 mg bd), edoxaban (30 mg od) and edoxaban (60 mg od) reduced major bleeding risk compared with warfarin (INR 2–3) (). Comparisons among licensed doses of NOACs, presented in , suggest that dabigatran (150 mg bd) increases risk of major bleeding compared with apixaban (5 mg bd), whereas rivaroxaban (20 mg od) increases risk of major bleeding compared with apixaban (5 mg bd) and edoxaban (60 mg od).
Results for major bleeding (stroke prevention in AF): comparisons with warfarin (INR 2–3)
Results for major bleeding (stroke prevention in AF): NOACs (licensed doses only)
In a sensitivity analysis to take into account the differences in duration of follow-up, NMA results were as presented in and , and show very similar results. Another sensitivity analysis involved fitting a fixed-effects meta-regression model using the mean TTR for warfarin patients (see ) as a covariate and the mean log-OR from each pairwise comparison (with warfarin as the reference category) as the response variable. We found no evidence of an effect modification according to mean TTR (estimated coefficient 0.04 with 95% CI −0.03 to 0.12 per 1% increase). The model fit indices yielded almost identical values for the models with and without the covariate.
Results for major bleeding (stroke prevention in AF): comparisons with warfarin (INR 2–3): sensitivity analysis using HRs instead of ORs
Results for major bleeding (stroke prevention in AF): NOACs (licensed doses only) – sensitivity analysis using HRs instead of ORs
Clinically relevant bleeding
Twelve studies reported 9556 CRB events, leading to a network of 23 interventions (). Eleven studies were included in the main analysis, with the remaining study included only in sensitivity analyses. These studies were at mixed risks of bias (), the concerns being due to lack of blinding of participants for most trials.
Network plot for CRB (stroke prevention in AF). a, Excluded interventions that were included in sensitivity analyses.
Included trials and risk-of-bias assessment for CRB (stroke prevention in AF)
Results presented in suggest that antiplatelet therapy (< 150 mg od) reduces CRB compared with warfarin (INR 2–3). Note that the licensed dose for of antiplatelet therapy for AF is ≥ 150 mg od: no studies provided data for that dose for CRB. Among NOACs, there was evidence that apixaban (5 mg bd), edoxaban (30 mg od) and edoxaban (60 mg od) reduce CRB compared with warfarin (INR 2–3). However, edoxaban (30 mg bd) and edoxaban (60 mg bd) increased CRB compared with warfarin (INR 2–3). Among licensed NOACs (), there was evidence that edoxaban (60 mg od) and rivaroxaban (20 mg od) increase CRB compared with apixaban (5 mg bd) and that rivaroxaban (20 mg od) increased CRB compared with edoxaban (60 mg od).
Results for CRB (stroke prevention in AF): comparisons with warfarin (INR 2–3)
Results for CRB (stroke prevention in AF): NOACs (licensed doses only)
Supplementary NMAs of HRs rather than ORs show very similar results ( and ).
Results for CRB (stroke prevention in AF): comparisons with warfarin (INR 2–3): sensitivity analysis using HRs instead of ORs
Results for CRB (stroke prevention in AF): NOACs (licensed doses only): sensitivity analysis using HRs instead of ORs
Intracranial bleeding
Eight studies reported a total of 757 intracranial bleeds, leading to a network of 10 interventions (). Seven trials were included in the primary analysis, with the remaining study included only in sensitivity analyses. These studies were at mixed risks of bias (), the concerns being due to lack of blinding of participants and, in one study, lack of blinding of outcome assessment.
Network plot for intracranial bleeding (stroke prevention in AF). a, Excluded interventions that were included in sensitivity analyses.
Included trials and risk-of-bias assessment for intracranial bleeding (stroke prevention in AF)
There was strong evidence that apixaban (5 mg bd), dabigatran (110 mg bd), dabigatran (150 mg bd), edoxaban (30 mg od), edoxaban (60 mg od) and rivaroxaban (20 mg od) reduced risk of intracranial bleeding compared with warfarin (INR 2–3) (). For each of these NOAC doses except rivaroxaban (20 mg od), the estimated reduction in risk was > 50%. There was weak evidence that risk of intracranial bleeding was increased for rivaroxaban (20 mg od) compared with apixaban (5 mg bd), dabigatran (150 mg bd) and edoxaban (60 mg od) (). Analysing HRs rather than ORs led to similar results ( and ).
Results for intracranial bleeding (stroke prevention in AF): comparisons with warfarin (INR 2–3)
Results for intracranial bleeding (stroke prevention in AF): NOACs (licensed doses only)
Results for intracranial bleeding (stroke prevention in AF): comparisons with warfarin (INR 2–3) – sensitivity analysis using HRs instead of ORs
Results for intracranial bleeding (stroke prevention in AF): NOACs (licensed doses only) – sensitivity analysis using HRs instead of ORs
All-cause mortality
Eighteen studies reported 6479 all-cause mortality events, leading to a network of fifteen interventions (). Fifteen studies were included in the primary analysis, with the remaining three studies included in sensitivity analyses. These studies were at mixed risks of bias (). There were concerns about lack of blinding of participants for most trials, and about lack of allocation concealment and blinding of outcome assessment in some studies.
Network plot for all-cause mortality (stroke prevention in AF). a, Excluded interventions that were included in sensitivity analyses.
Included trials and risk-of-bias assessment for all-cause mortality (stroke prevention in AF)
suggests that all NOAC doses with comparisons that were not imprecisely estimated [apixaban (5 mg bd), dabigatran (110 mg bd), dabigatran (150 mg bd), edoxaban (30 mg od), edoxaban (60 mg od) and rivaroxaban (20 mg od)] were associated with a reduced risk of all-cause mortality compared with warfarin (INR 2–3). There was little evidence that the risk of all-cause mortality differed between licensed doses of NOACs (). Analysing HRs rather than ORs produced similar results ( and ).
Results for all-cause mortality (stroke prevention in AF): comparisons with warfarin (INR 2–3)
Results for all-cause mortality (stroke prevention in AF): NOACs (licensed doses only)
Results for all-cause mortality (stroke prevention in AF): comparisons with warfarin (INR 2–3): sensitivity analysis using HRs instead of ORs
Results for all-cause mortality (stroke prevention in AF): NOACs (licensed doses only): sensitivity analysis using HRs instead of ORs
Summary of results and ranking of interventions
Results from NMAs suggest that a number of the licensed doses of NOACs reduce the risk of the outcomes stroke or SE, major bleeding, CRB, intracranial bleeding and all-cause mortality compared with the reference treatment, warfarin (INR 2–3). There was evidence that edoxaban increased CRB compared with warfarin (INR 2–3). Risk of MI appeared higher for some NOACs than for warfarin (INR 2–3). Comparisons for some licensed NOAC doses, such as apixaban (2.5 mg bd) and betrixaban (40 mg od), could not be estimated precisely.
Several studies conducted in Asian countries considered a lower INR range for warfarin interventions in elderly patients. We excluded these from the main analysis but included them (merged with the reference treatment, warfarin INR 2–3) as a second sensitivity analysis for each outcome. This allowed us to incorporate a non-licensed dose of rivaroxaban (15 mg od) that was included in the J-ROCKET AF trial,106,112,123,129 showing a reduced risk of stroke compared with warfarin (INR 1.6–3), with a median OR of 0.49 (95% CI 0.24 to 0.99). Apart from this, results (available on request) showed the same trends as described above.
The dose range for the antiplatelet arm in the AVERROES trial105,116,117,121 was unusually wide (81–324 mg od). Because some of the patients had received a dose that was below standard, it was decided to merge it with the antiplatelets (< 150 mg od) node for the primary analysis. In a further sensitivity analysis for each outcome, this trial105,116,117,121 was excluded. Again, the results (available from the authors) were not substantially different from those presented above. With regard to model appraisal, we did not identify any instance of lack of convergence among the Markov chains, poor model fit or inconsistency. Few of the comparisons were replicated across studies; when there were multiple estimates we did not find evidence of statistical heterogeneity.
Rankograms plotting the probability that each of the licensed interventions for AF is ranked best, second best, and so on, for preventing each outcome, are displayed in . The non-NOAC interventions (warfarin, INR 2–3) and antiplatelet therapy (aspirin/clopidogrel, ≥ 150 mg od) were ranked worst for stroke or SE and ischaemic stroke and were not among the best three interventions for any of the outcomes. Warfarin (INR 2–3) was also ranked as the worst intervention to reduce the risk of intracranial bleeding. Among the licensed NOACs, apixaban (5 mg bd) was ranked as among the best interventions for major bleeding, intracranial bleeding, all-cause mortality, stroke or SE, ischaemic stroke and MI. Edoxaban (60 mg od) was ranked second for major bleeding and all-cause mortality. Except for all-cause mortality and MI, outcomes for rivaroxaban (20 mg od) were ranked less highly than those for apixaban (5 mg bd), dabigatran (150 mg bd) and edoxaban (60 mg od).
Rankogram for licensed interventions examined in stroke prevention in AF. IC, intracranial.
