Research in context
Evidence before this study
Randomised controlled trials have shown that oral anticoagulation reduces the high risk of systemic embolism by almost two-thirds for patients with atrial fibrillation despite doubling their low risk of major bleeding. However, these trials excluded patients with intracranial haemorrhage. We searched the Cochrane Central Register of Controlled Trials, MEDLINE Ovid (from 1946), Embase Ovid (from 1974), online registers of clinical trials, and bibliographies of relevant publications on June 11, 2021, with no language restrictions (for search terms see appendix pp 3–5). We found one completed randomised feasibility study involving 30 patients (NASPAF-ICH, NCT02998905) and one completed randomised phase 2 trial involving 101 patients (APACHE-AF, NCT02565693) that compared the effects of oral anticoagulation versus antiplatelet therapy for participants with atrial fibrillation after intracerebral haemorrhage; these trials were inconclusive about clinical outcomes. Meta-analyses of observational studies of patients with atrial fibrillation and intracranial haemorrhage mostly found associations between oral anticoagulation and reduced risks of major ischaemic vascular events, but no significant change in the risk of recurrent major haemorrhagic vascular events.
Added value of this study
The Start or STop Anticoagulants Randomised Trial (SoSTART) is, to our knowledge, the largest randomised controlled trial to date to compare the effects of starting versus avoiding oral anticoagulation for atrial fibrillation after intracranial haemorrhage. Participants allocated to start oral anticoagulation had more intracranial haemorrhage recurrences, but our prespecified margin for declaring non-inferiority was not met (p=0·152). However, non-significant results for our three composite secondary outcomes suggest that starting oral anticoagulation might be superior to avoiding oral anticoagulation for preventing any symptomatic major vascular event.
Implications of all the available evidence
Further randomised trials are justified to investigate the non-inferiority of the effects of oral anticoagulation on major bleeding for patients with atrial fibrillation after intracranial haemorrhage or whether oral anticoagulation might be superior for preventing symptomatic major vascular events (especially those that are fatal or disabling). Clinicians should embed ongoing randomised controlled trials that are addressing this problem in their clinical practice so that these trials and the COCROACH planned individual participant data meta-analysis are adequately powered to provide definitive evidence.