Optimal timing of oral anticoagulation initiation in patients with acute ischaemic stroke and atrial fibrillation: a comprehensive meta-analysis and systematic review.

The optimal timing for initiating direct oral anticoagulants (DOACs) for secondary stroke prevention in patients with atrial fibrillation and acute ischaemic stroke remains controversial due to concerns about haemorrhagic transformation. This study aimed to analyse the efficacy and safety of early versus late DOAC initiation. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, a systematic review was conducted, searching major databases (PubMed, Embase, Cochrane Library and ClinicalTrials.gov) up to May 2024. A total of 11 studies were identified, comprising nine cohort studies (75.5% weight) and two randomised controlled trials (RCTs) (24.5% weight), involving 13 020 participants. The early DOAC group (mean initiation 3.5±1.29 days) included 6250 participants, while the late group (5.7±1.25 days) had 6770 participants. Outcome measures included recurrent ischaemic stroke (RIS), intracranial haemorrhage (ICH), systemic embolism, major haemorrhage (MH), non-major haemorrhage (NMH) and all-cause mortality. Statistical analysis using the Cochrane Review Manager calculated ORs and 95% CIs via the Mantel-Haenszel random effects model. This pooled meta-analysis revealed that the early DOAC group had lower rates of RIS (2.2% vs 2.9%, OR 0.72, 95% CI 0.52 to 0.98, p=0.04, I²=40%) and ICH (0.51% vs 0.93%, OR 0.45, 95% CI 0.29 to 0.70, p<0.05, I²=0%) compared with the late DOAC group. Subgroup analysis of RCTs and cohort studies showed reduced RIS and ICH risks in the early DOAC group, with moderate heterogeneity. In the sensitivity analysis, the early group (<4 days) had a lower risk of RIS compared with the late group (>4 days) without a statistically significant impact on ICH. No significant differences in MH, NMH, systemic embolism or all-cause mortality were observed between either group; however, a limited number of RCTs and moderate heterogeneity weakened the conclusions. Additional RCTs are needed to provide more definitive insights.