#atrial-fibrillation #ischemic-stroke #anticoagulation #DOAC #randomized-controlled-trial

Early versus Later Anticoagulation for Stroke with Atrial Fibrillation (ELAN)

Authors, Journal, Affiliations, Type, DOI

Authors: Urs Fischer, Masatoshi Koga, Daniel Strbian, Mattia Branca, Stefanie Abend, Sven Trelle, Maurizio Paciaroni, Götz Thomalla, Patrik Michel, Krassen Nedeltchev, Leo H. Bonati, George Ntaios, Thomas Gattringer, Else-Charlotte Sandset, Peter Kelly, Robin Lemmens, P.N. Sylaja, Diana Aguiar de Sousa, Natan M. Bornstein, Zuzana Gdovinova, Takeshi Yoshimoto, Marjaana Tiainen, Helen Thomas, Manju Krishnan, Gek C. Shim, Christoph Gumbinger, Jochen Vehoff, Liqun Zhang, Kosuke Matsuzono, Espen Kristoffersen, Philippe Desfontaines, Peter Vanacker, Angelika Alonso, Yusuke Yakushiji, Caterina Kulyk, Dimitri Hemelsoet, Sven Poli, Ana Paiva Nunes, Nicoletta Caracciolo, Peter Slade, Jelle Demeestere, Alexander Salerno, Markus Kneihsl, Timo Kahles, Daria Giudici, Kanta Tanaka, Silja Räty, Rea Hidalgo, David J. Werring, Martina Göldlin, Marcel Arnold, Cecilia Ferrari, Seraina Beyeler, Christian Fung, Bruno J. Weder, Turgut Tatlisumak, Sabine Fenzl, Beata Rezny-Kasprzak, Arsany Hakim, Georgia Salanti, Claudio Bassetti, Jan Gralla, David J. Seiffge, Thomas Horvath, Jesse Dawson, for the ELAN Investigators
Journal: New England Journal of Medicine, 2023;388(26):2411-2421 (published online May 24, 2023; updated June 29, 2023)
Affiliations: University Hospital Basel and University Hospital Bern, Switzerland (coordinating centres), with 103 sites across Europe, the Middle East and Asia
Type: Investigator-initiated, multicentre, open-label, assessor-blinded randomised controlled trial. Funded by the Swiss National Science Foundation, the Swiss Heart Foundation, the UK Stroke Association and the Japanese National Cerebral and Cardiovascular Center Intramural Research Fund; no industry involvement.
DOI: https://doi.org/10.1056/NEJMoa2303048

Overview

ELAN randomised 2,013 patients with atrial fibrillation and acute ischaemic stroke to early or later initiation of a direct oral anticoagulant, using infarct size on imaging rather than the NIHSS to set the timing windows. At 30 days the primary composite of recurrent ischaemic stroke, systemic embolism, major extracranial bleeding, symptomatic intracranial haemorrhage or vascular death occurred in 2.9% with early treatment and 4.1% with later treatment, a risk difference of −1.18 percentage points (95% CI −2.84 to 0.47). The trial deliberately tested no statistical hypothesis: the aim was a precise estimate, and the authors' interpretation is that early initiation can be supported on the basis of a confidence interval extending from a 2.8-point reduction to a 0.5-point increase in risk. It is the largest randomised evidence on anticoagulation timing after AF-related stroke and the primary source behind the current Class IIa recommendation for early initiation.

Keywords

Atrial fibrillation; ischaemic stroke; direct oral anticoagulants; DOAC; anticoagulation timing; haemorrhagic transformation; secondary prevention; randomised controlled trial

Key Takeaways

Background and Rationale

DOACs reduce stroke and systemic embolism in AF, but the optimal time to start one after an acute ischaemic stroke was undefined: early initiation risks intracranial haemorrhage into infarcted tissue, later initiation risks early recurrence. Both risks are highest in the first days.
Guideline practice rested on consensus and on the observation that haemorrhagic transformation is related to infarct size — crystallised as the "1-3-6-12-day rule" (anticoagulation at 1, 3, 6 and 12 days after TIA and minor, moderate and severe stroke respectively), used in many countries.
Prior randomised data were small or selection-biased; imaging-based risk stratification had been proposed but not tested prospectively at scale.

Methods

Design: investigator-initiated, open-label trial with outcome assessors unaware of assignment, at 103 stroke centres in 15 countries (Europe, Middle East, Asia), November 2017 to September 2022.
Population: adults with acute ischaemic stroke and permanent, persistent or paroxysmal non-valvular AF; stroke confirmed by CT/MRI. Intravenous thrombolysis and thrombectomy before randomisation were allowed; therapeutic anticoagulation at stroke onset was not. Petechial haemorrhage within the infarct was permitted; confluent parenchymal haematoma or remote intracranial haemorrhage was excluded.
Severity by imaging, not NIHSS: minor = infarct ≤1.5 cm; moderate = infarct in a cortical superficial branch territory of the MCA, ACA or PCA; major = larger territorial infarct, or brainstem/cerebellar infarct >1.5 cm.
Intervention: any DOAC licensed in the trial country at the appropriate dose. Early arm — within 48 hours for minor or moderate stroke, day 6 or 7 for major stroke. Later arm — day 3 or 4 for minor, day 6 or 7 for moderate, day 12, 13 or 14 for major stroke (i.e. the 1-3-6-12 rule).
Primary outcome: composite of recurrent ischaemic stroke, systemic embolism, major extracranial bleeding, symptomatic intracranial haemorrhage or vascular death within 30 days of randomisation.
Analysis: no superiority, inferiority or non-inferiority hypothesis was tested. The sample size was chosen to give a 95% CI at least 2.0 percentage points wide around an expected 5.0% versus 4.5% event rate. Modified intention-to-treat after excluding 19 randomised participants (13 at a site closed for Good Clinical Practice non-adherence, 5 who declined post hoc consent, 1 randomised twice).

Results

Participants

36,643 screened; 2,032 randomised; 2,013 in the modified ITT population (1,006 early, 1,007 later).
Median age 77 years (IQR 70–84); 45% female; median NIHSS 5 at admission and 3 at randomisation; median CHA₂DS₂-VASc 5 (IQR 4–6); 88.5–89.3% had pre-stroke mRS 0–2.
Infarct size: 37% minor, 40% moderate, 23% major in both groups; 39.7% received thrombolysis and 21–24% thrombectomy before randomisation; about 50% were on aspirin at screening.
Treatment started per protocol in 949 early and 935 later participants; a further 49 and 56 respectively started outside the planned window.

Primary Outcome

29/1,006 (2.9%) early versus 41/1,007 (4.1%) later; adjusted OR 0.70 (95% CI 0.44–1.14); risk difference −1.18 percentage points (95% CI −2.84 to 0.47).
Primary-outcome data available for 98%; missing outcomes imputed (22 early, 16 later).
Post hoc bootstrap analysis: 98% probability that early treatment increases the 30-day primary-event risk by no more than 0.5 percentage points.

Components at 30 Days

Outcome	Early (n=984)	Later (n=991)	Adjusted risk difference (95% CI)
Primary composite	2.9%	4.1%	−1.18 (−2.84 to 0.47)
Recurrent ischaemic stroke	1.4%	2.5%	−1.14 (−2.41 to 0.13); OR 0.57 (0.29–1.07)
Systemic embolism	0.4%	0.9%	−0.55 (−1.34 to 0.23); OR 0.48 (0.14–1.42)
Symptomatic intracranial haemorrhage	0.2%	0.2%	0.01 (−0.52 to 0.53); OR 1.02 (0.16–6.59)
Major extracranial bleeding	0.3%	0.5%	−0.25 (−0.90 to 0.41); OR 0.63 (0.15–2.38)
Vascular death	1.1%	1.0%	0.13 (−0.84 to 1.09); OR 1.12 (0.47–2.65)

Ninety-Day Outcomes (not prespecified as a secondary outcome — exploratory)

Primary composite 3.7% versus 5.6%; OR 0.65 (95% CI 0.42–0.99); adjusted risk difference −1.92 (−3.82 to −0.02).
Recurrent ischaemic stroke 1.9% versus 3.1%; OR 0.60 (0.33–1.06).
Symptomatic intracranial haemorrhage 0.2% in both groups; OR 1.00 (0.15–6.45).
Death from any cause 4.5% versus 4.8%; OR 0.93 (0.61–1.43). mRS ≤2 at 90 days 66.6% versus 65.8%.
Any serious adverse event at 90 days 13.9% versus 15.8%.

Subgroups and Sensitivity Analyses

No apparent heterogeneity across prespecified subgroups (age, infarct size, NIHSS, site), but the trial was not powered for subgroup analysis and confidence intervals were not adjusted for multiplicity.
Per-protocol and alternative-model sensitivity analyses were consistent with the main result.

Discussion

Early initiation "can therefore be supported if indicated or if desired"; the outcome rates rose only slightly between 30 and 90 days, suggesting no excess hazard from early anticoagulation over that period.
The trial's imaging-based severity classification is presented as the reason symptomatic intracranial haemorrhage stayed at ~0.2% in both arms. The authors call for confirmation using NIHSS-based definitions and for testing whether major stroke can be anticoagulated earlier than day 6.
Contrast with TIMING (Sweden; terminated early after 888 patients; generally similar results) and OPTIMAS (UK; ongoing at the time): ELAN randomised within 48 hours in minor/moderate stroke and compared against the widely used 1-3-6-12 rule.

Limitations of the document

No hypothesis was tested. The trial was powered for precision, not for non-inferiority or superiority; the primary confidence interval crosses zero, and the favourable reading rests on the width of the interval and the post hoc probability statement rather than on a formal test.
Open-label treatment, mitigated only by blinded outcome assessment and independent adjudication.
Low median NIHSS at randomisation (3) — a milder population than many stroke cohorts; the results may not transfer to severe deficits.
Patients on therapeutic anticoagulation at baseline were excluded, so the trial does not address re-starting in patients already anticoagulated who had a stroke.
Parenchymal haematoma (Heidelberg PH1/PH2) at randomisation was excluded, so the safety of early anticoagulation in that group is unknown.
Stroke severity was not centrally adjudicated; classification relied on site investigators using a standardised visual rating scheme.
Predominantly European, White population; no ethnicity data collected; extrapolation to other populations is uncertain.
Subgroup analyses are underpowered and not adjusted for multiplicity.
Published 2023; the guideline recommendation it informs is Class IIa, reflecting that a single open-label trial with no formal hypothesis test is not definitive.

Key Concepts Mentioned

Key Entities Mentioned

Wiki Pages Updated

log.md carries the ingest entry.