Objective Current guidelines endorse the use of non-vitamin K antagonist oral anticoagulants (NOACs) in patients with atrial fibrillation (AF). However, little is known about their safety and efficacy in valvular heart disease (VHD). Similarly, there is a paucity of data regarding NOACs use in patients with a bioprosthetic heart valve (BPHV). We, therefore, performed a network meta-analysis in the subgroups of VHD and meta-analysis in patients with a BPHV.

Methods PubMed, Cochrane and Embase were searched for randomised controlled trials. Summary effects were estimated by the random-effects model. The outcomes of interest were a stroke or systemic embolisation (SSE), myocardial infarction (MI), all-cause mortality, major adverse cardiac events, major bleeding and intracranial haemorrhage (ICH).

Results In patients with VHD, rivaroxaban was associated with more ICH and major bleeding than other NOACs, while edoxaban 30 mg was associated with least major bleeding. Data combining all NOACs showed a significant reduction in SSE, MI and ICH (0.70, [0.57 to 0.85; p<0.001]; 0.70 [0.50 to 0.99; p<0.002]; and 0.46 [0.24 to 0.86; p<0.01], respectively). Analysis of 280 patients with AF and a BPHV showed similar outcomes with NOACs and warfarin.

Conclusions NOACs performed better than warfarin for a reduction in SSE, MI and ICH in patients with VHD. Individually NOACs performed similarly to each other except for an increased risk of ICH and major bleeding with rivaroxaban and a reduced risk of major bleeding with edoxaban 30 mg. In patients with a BPHV, results with NOACs seem similar to those with warfarin and this needs to be further explored in larger studies.