SummaryAs non-valvular atrial fibrillation (AF) brings a risk of stroke, oral anticoagulants (OAC) are recommended. In ‘real world’ clinical practice, many patients (who may be, or perceived to be, intolerant of OACs) are either untreated or are treated with anti-platelet agents. We hypothesised that edoxaban has a better net clinical benefit (NCB, balancing the reduction in stroke risk vs increased risk of haemorrhage) than no treatment or anti-platelet agents. We performed a network meta-analysis of published data from 24 studies of 203,394 AF patients to indirectly compare edoxaban with aspirin alone, aspirin plus clopidogrel, and placebo. Edoxaban 30 mg once daily significantly reduced the risk of all stroke, ischaemic stroke and mortality compared to placebo and aspirin. Compared to aspirin plus clopidogrel, there was a lower risk of intra-cranial haemorrhage (ICH). Edoxaban 60 mg once-daily had a reduced risk of any stroke and systemic embolism compared to placebo, aspirin, and aspirin plus clopidogrel. Mortality rates for both edoxaban doses were estimated to be lower compared to any anti-platelet, and significantly lower compared to placebo. With overall reduced risk of ischemic stroke and ICH, both edoxaban doses bring a NCB of mean (SD) 1.68 (0.15) saved events per 100 patients per year compared to anti-platelet drugs in a clinical trial population. The NCB was demonstrated to be lower, at 0.77 (0.12) events saved (p< 0.01) when modeled to data from a ‘real world’ cohort of AF patients. In conclusion, edoxaban is likely to provide even better protection from stroke and ICH than placebo, aspirin alone, or aspirin plus clopidogrel in both clinical trial populations and unselected community populations. Both edoxaban doses would also bring a positive NCB compared to anti-platelet drugs or placebo/non-treatment based on ‘real world’ data.Note: The review process for this paper was fully handled by Christian Weber, Editor in Chief.
P3447Mortality and stroke prevention in atrial fibrillation; network meta-analysis of real world data
