OBJECTIVES: To assess cost-effectiveness of direct oral anticoagulants (DOACs) compared with vitamin K antagonists (VKAs) for stroke prevention in atrial fibrillation (AF) by pooling incremental net benefits (INBs).
DESIGN: Systematic review and meta-analysis.
SETTING: We searched PubMed, Scopus and Centre for Evaluation of Value and Risks in Health Registry from inception to December 2019.
PARTICIPANTS: Patients with AF.
MAIN OUTCOME MEASURES: The INB was defined as a difference of incremental effectiveness multiplied by willing to pay threshold minus the incremental cost; a positive INB indicated favour treatment. These INBs were pooled (stratified by level of country income, perspective, time-horizon, model types) with a random-effects model if heterogeneity existed, otherwise a fixed effects model was applied. Heterogeneity was assessed using Q test and I2 statistic. Risk of bias was assessed using the economic evaluations bias (ECOBIAS) checklist.
RESULTS: A total of 100 eligible economic evaluation studies (224 comparisons) were included. For high-income countries (HICs) from a third-party payer (TPP) perspective, the pooled INBs for DOAC versus VKA pairs were significantly cost-effective with INBs (95% CI) of $6632 ($2961.67 to $10 303.72; I2=59.9%), $6353.24 ($4076.03 to $8630.45; I2=0%), $7664.58 ($2979.79 to $12 349.37; I2=0%) and $8573.07 ($1877.05 to $15 269.09; I2=0%) for dabigatran, apixaban, rivaroxaban and edoxaban relative to VKA, respectively but only dabigatran was significantly cost-effective from societal perspective (SP) with an INB of $11 746.96 ($2429.34 to $21 064.59; I2=52.4%). The pooled INBs of all comparisons for upper-middle income countries (UMICs) were not significantly cost-effective. The ECOBIAS checklist indicated that risk of bias was mostly low for most items with the exception of five items which should be less influenced on pooling INBs.
CONCLUSIONS: Our meta-analysis provides comprehensive economic evidence that allows policy makers to generalise cost-effectiveness data to their local context. All DOACs may be cost-effective compared with VKA in HICs with TPP perspective. The pooling results produced moderate to high heterogeneity particularly in UMICs. Further studies are required to inform UMICs with SP.