Comparative efficacy and safety of warfarin care bundles and novel oral anticoagulants in patients with atrial fibrillation: a systematic review and network meta-analysis

Warfarin care bundles (e.g. genotype-guided warfarin dosing, patient’s self-testing [PST] or patient’s self-management [PSM] and left atrial appendage closure) are based on the concept of combining several interventions to improve anticoagulation care. NOACs are also introduced for stroke prevention in atrial fibrillation (SPAF). However, these interventions have not been compared in head-to-head trials yet. We did a network meta-analysis based on a systematic review of randomized controlled trials comparing anticoagulant interventions for SPAF. Studies comparing these interventions in adults, whether administered alone or as care bundles were included in the analyses. The primary efficacy outcome was stroke and the primary safety outcome was major bleeding. Thirty-seven studies, involving 100,142 patients were assessed. Compared to usual care, PSM significantly reduced the risk of stroke (risk ratio [RR] 0.24, 95% CI 0.08–0.68). For major bleeding, edoxaban 60 mg (0.80, 0.71–0.90), edoxaban 30 mg (0.48, 0.42–0.56), and dabigatran 110 mg (0.81, 0.71–0.94) significantly reduced the risk of major bleeding compared with usual warfarin care. Cluster rank plot incorporating stroke and major bleeding outcomes indicates that some warfarin care bundles perform as well as NOACs. Both interventions are therefore viable options to be considered for SPAF. Additional studies including head-to-head trials and cost-effectiveness evaluation are still warranted.


S4 Appendix: Matching of Major Bleeding Definitions
In order to minimize the heterogeneity on the definitions of major bleeding across the included studies, major bleeding events were matched against the standardized bleeding end-point definitions adopted by the Bleeding Academic Research Consortium (BARC). 39 Major bleeding outcome was collected based on BARC3-5 criteria because this bleeding definition was standardized to include both clinical and laboratory information. Furthermore, BARC definition was the most updated bleeding definition.
We directly collected major bleeding outcome from any studies reported outcome as BARC definition. If not, the compatibility criteria were considered. The compatible definitions must be standardized based on BARC 3-5 criteria and they must not contain any lower severity of bleeding. For instance, ISTH major bleeding definition (fatal bleeding, symptomatic bleeding in a critical area or organ such as intracranial, reduction of haemoglobin 2g/dL or transfusion of two or more units of whole blood) could be categorized into BARC type 3A (reduction of haemoglobin 3 to <5g/Dl or transfusion with overt bleeding), type 3B (reduction of haemoglobin 5g/dL), type 3c (intracranial haemorrhage), and type 5 (fatal bleeding). Therefore, ISTH major bleeding was a compatible definition and ISTH major bleeding outcomes could be collected into our analysis. In contrast, a non-official definition such as 'intracranial bleeding and/or requiring blood transfusion or surgical intervention and/or hospitalization' could not be categorized as compatible definition due to the inclusion of hospitalization criteria which was unclear in severity and might not be compatible with BARC 3-5.

S9 Appendix: Results of Network Meta-analysis
Intervention options are in order of their efficacy or safety ranking. Estimates are presented as risk ratios (RR) and 95% confidence intervals. Treatments are ordered by rankings for each outcome.
Comparisons between interventions should be read from column to row for each outcome (row intervention is reference). Risk ratios less than 1 favour the column-defining treatment. To obtain risks ratios for comparisons in the opposite direction, reciprocals should be taken. Significant results are in bold and underlined. Blue-colour box represents that unfavourable outcome was decreased. In contrast, red-colour box represents that unfavourable outcome was increased.

eTable 9.1: Results of network meta-analysis of interventions options on stroke or systemic embolism
The following table shows the effect sizes (risk ratio) and the rank order (SUCRA ranks) compared to usual warfarin care (UC).

: Results of network meta-analysis of interventions options on intracranial bleeding
The following table shows the effect sizes (risk ratio) and the rank order (SUCRA ranks) compared to usual warfarin care (UC).

.7: Results of network meta-analysis of interventions options on gastrointestinal bleeding
The following table shows the effect sizes (risk ratio) and the rank order (SUCRA ranks) compared to usual warfarin care (UC).

.8: Results of network meta-analysis of interventions options on myocardial infarction
The following table shows the effect sizes (risk ratio) and the rank order (SUCRA ranks) compared to usual warfarin care (UC).

1: Sensitivity analyses for the risk of stroke or systemic embolism with intervention options of different study characteristics
The following table shows the effect sizes (risk ratio) and the rank order (SUCRA ranks) compared to usual warfarin care (UC) before (main analysis) and after subgroup analyses.

2: Sensitivity analyses for the risk of major bleeding with interventions options with different major bleeding definitions
The following table shows the effect sizes (risk ratio) and the rank order (SUCRA ranks) compared to usual warfarin care (UC) before (standard analysis) and after sensitivity analyses.

3: Sensitivity analyses for the risk of major bleeding with intervention options of different study characteristics
The following table shows the effect sizes (risk ratio) and the rank order (SUCRA ranks) compared to usual warfarin care (UC) before (main analysis) and after subgroup analyses.

4: Sensitivity analyses for the risk of all-cause mortality with intervention options of different study characteristics
The following table shows the effect sizes (risk ratio) and the rank order (SUCRA ranks) compared to usual warfarin care (UC) before (main analysis) and after subgroup analyses.