Timing of anticoagulation after ischaemic stroke and atrial fibrillation: better swift than sorry

Wendy C Ziai, Georgios Tsivgoulis

Lancet Available online 23 June 2025

https://doi.org/10.1016/S0140-6736(25)00676-2

Non-valvular atrial fibrillation accounts for 13–26% of acute ischaemic strokes.¹ Direct oral anticoagulants (DOACs) are highly effective in reducing the risk of ischaemic stroke with significantly lower rates of intracranial haemorrhage compared with vitamin K antagonists.² Once an ischaemic stroke has occurred, the timing of resuming anticoagulation therapy in patients with atrial fibrillation has long been questioned; the initiation of such therapies has traditionally been delayed due to concern for haemorrhagic transformation of ischaemic stroke territory as well as intra-cranial and extra-cranial bleeding complications. Both observational and clinical trial data suggest that early DOAC treatment might be safe,³ although four randomised controlled clinical trials (TIMING, ELAN, OPTIMAS, and START) have individually been unable to show superiority of early versus delayed initiation.⁴ Current international guidelines therefore do not provide specific recommendations about the best time to initiate DOACs in this setting.5, 6

In The Lancet, Hakim-Moulay Dehbi and colleagues⁴ report the results of the CATALYST study, an individual patient data meta-analysis (IPDMA) comparing early initiation of a DOAC within 4 days after stroke onset versus later initiation after 5 days or more. The study included 5441 participants from four randomised controlled clinical trials (mean age 77·7 years, 45·4% women, median National Institutes of Health Stroke Scale [NIHSS] score at admission 5 points [IQR 3–10]). Study participants were predominantly European with smaller proportions of participants from Asia, the Middle East, Africa, and the USA. The authors report a reduction of the primary outcome, a composite of recurrent ischaemic stroke, symptomatic intracerebral haemorrhage, and unclassified stroke within 30 days in the early DOAC compared with later DOAC initiation group (odds ratio [OR] 0·70, 95% CI 0·50–0·98, p=0·04). These benefits were consistent across key prespecified subgroups including initial clinical stroke severity, reperfusion treatment, and previous use of oral anticoagulants. Most of these benefits were due to reduced risk of recurrent ischaemic stroke at day 30 (OR 0·66, 0·45–0·96, p=0·03), but not at day 90, when a numerical reduction in favour of early DOAC initiation was not significant (OR 0·88, 0·65–1·19, p=0·40). Symptomatic intracerebral haemorrhage event rate was low (<0·5%) and did not differ between groups at day 30 or 90.

The authors should be commended on a successful trial data collaboration and performing a well planned, prospectively designed, and robustly executed IPDMA, which reinforces current trends in clinical management for the type of stroke patients enrolled in these trials. CATALYST excluded 91 patients for data-related issues (eg, absence of consent and loss to follow-up) and no baseline imbalances were observed between the two groups, which is an improvement over observational data. The event rate for the primary outcome was less than 5% and differed by less than 1% between the early and late groups (2·12% vs 3·02%). This result is not too surprising given the 4-day difference in median time from symptom onset to DOAC initiation (3·0 vs 7·2 days). However, most new ischaemic events occurred in the first week, emphasising the upfront risk of early atrial fibrillation associated stroke recurrence without anticoagulation, which ranges from 0·5% to 1·3% per day for the first 2 weeks.⁷ The rates of the ischaemic stroke outcomes increased only slightly more at 90 days than at 30 days, and this increase was more substantial for the early start group. Haemorrhage rates were not increased, suggesting no excessive risk associated with early anticoagulation through that period.

The study generally avoided patients with larger cerebral infarcts who are at greater risk for haemorrhagic transformation with early anticoagulation, which compromises the generalisability of the findings.8, 9 More specifically, patients with an NIHSS-score of more than 15 points comprised only 13% of the population and CATALYST removed major and high-risk strokes from ELAN (approximately 23% of the enrolled individuals),¹⁰which accounted for infarct size on imaging to assign DOAC start times (all assigned >5 days) and minor strokes from ELAN (approximately 37% of enrolled individuals); these participants were all enrolled before day 5. START included only patients with mild-to-moderate strokes.¹¹ TIMING and OPTIMAS used single cutoff timepoints for early versus late DOAC administration initiation. No patients with symptomatic haemorrhagic transformation (parenchymal haematoma type 2) were included. Therefore, for the patients with a severe stroke or with parenchymal haematoma type 2 in whom clinicians previously might have delayed anticoagulation up to 2 weeks per current guidelines,5, 6CATALYST cannot further qualify the recommendation.

Other limitations include insufficient access to as-treated data, limiting conclusions regarding DOAC dosage and compliance especially in the event of procedures such as gastrostomy. Most patients in each trial were on apixaban, with smaller proportions on other DOACs: dabigatran, edoxaban, and rivaroxaban.

The results of CATALYST provide the most definitive evidence supporting a policy of early DOAC initiation within the first 4 days after mild-to-moderate acute ischaemic stroke. Still, the study was underpowered based on a post-hoc power calculation. However, when death was added to the primary outcome, the point estimate for effect of early versus later DOAC increased only slightly and was no longer significant (OR 0·81, 95% CI 0·64–1·01).

For patients with an increased bleeding risk, we await a planned analysis (initially in OPTIMAS, to be expanded to CATALYST) of imaging data on haemorrhagic transformation, extent of infarction, and severe small vessel disease on baseline imaging. Even then, adherence to strict predefined early or late initiation thresholds might oversimplify the complexity of this decision, which often requires assessment of other comorbidities including advanced age, polypharmacy, and frailty when delayed DOAC therapy might derive increased benefit.12, 13

Despite the limitations of this IPDMA, it supports the evidence for early initiation of oral anticoagulation in atrial fibrillation-associated acute ischaemic stroke, especially in patients with smaller infarct size and not at risk for haemorrhagic transformation: better swift than sorry.