Skip to main content
NeuroWiki
Back to Trials

Clinical Question

PFO and stroke: anticoagulate or antiplatelet?

Synthesises 4 trialsWhen the PFO is not closed. No adequately powered trial has ever tested this; every dataset is a subgroup

Clinical Synthesis

No adequately powered trial has ever compared anticoagulation with antiplatelet therapy in PFO-associated stroke: every dataset is a subgroup or an underpowered arm. A real signal favouring anticoagulation does appear in pooled subgroups, and it decayed as trials accumulated, while the bleeding excess it causes reproduced.

Bottom line

Default to antiplatelet therapy, and know why. The efficacy signal favouring anticoagulation is fragile: the pooled odds ratio moved from 0.48 to 0.70 when one more trial was added to the same pool. The bleeding excess is not fragile: it reproduces across independent analyses (RR 1.69 and RR 1.57) and points the other way. Anticoagulation is not wrong, and the AAN 2020 advisory (Statement 3a, Level C) treats the two as equally acceptable rather than endorsing aspirin; anticoagulation becomes the preferred choice when there is an independent indication, meaning a defined thrombophilia, an unprovoked DVT, or an unprovoked pulmonary embolism (Statement 3b, Level B). Before applying any of this, ask whether the PFO is plausibly causal at all: use the RoPE score to frame that question, and when the PFO looks incidental, spend the effort on finding the real mechanism rather than on the antithrombotic choice.

Sources

WARSS· 2001PICSS· 2002Rivaroxaban or aspirin for patent foramen ovale and embolic stroke of undetermined source· 2018NAVIGATE ESUS· 2018RE-SPECT ESUS· 2019Dabigatran or Aspirin After Embolic Stroke of Undetermined Source in Patients With Patent Foramen Ovale· 2021Registry record, recruiting as of 2026-07-31· 2026Clinical question 2 (antithrombotic choice); Statements 3a (Level C) and 3b (Level B)· 2020Weak recommendation for anticoagulant therapy over antiplatelet therapy (low-quality evidence)· 20189 RCTs, 15,451 participants; PFO-status subgroup interaction· 20257 ESUS RCTs, 14,804 patients; PFO stratum· 2024Anticoagulation versus antiplatelet therapy for secondary stroke prevention in patients with embolic stroke of undetermined source· 2022Oral anticoagulation versus antiplatelet therapy in patients with patent foramen ovale and cryptogenic stroke· 2022Closure, Anticoagulation, or Antiplatelet Therapy for Cryptogenic Stroke With Patent Foramen Ovale· 2018

Trials in this question· 4

PICSS2002
Non-prespecified subgroup of a substudy; low-intensity warfarin did not beat aspirin.
Subgroup onlyHR 0.52 (0.16 to 1.67)
CLOSE2017
The anticoagulation arm was a parallel comparison CLOSE was not powered for: 3 of 187 versus 7 of 174.
Underpowered armHR 0.44 (0.11 to 1.48)
NAVIGATE-ESUS2018
Prespecified PFO subgroup at 45% power, in a trial stopped for futility and bleeding.
UnderpoweredHR 0.54 (0.22 to 1.36)
RE-SPECT ESUS2021
PFO subgroup null; interaction p=0.83, and the hazard ratio is an AAN extraction.
No interactionHR 0.88 (0.45 to 1.71)

Score this at the bedside

RoPE Score

PFO-attributable fraction in cryptogenic stroke.

Related questions

PFO closure for cryptogenic stroke?

From the early trials that missed their primary endpoint (CLOSURE-I, PC, RESPECT 2013) to the positive 2017 cluster and DEFENSE-PFO; AF excess is the trade-off

Should I anticoagulate this patient?

AF timing, ESUS, and recurrence prevention

Was this useful?

NeuroWiki is a clinical reference. It does not substitute for your clinical judgment, current guidelines, or your institution's protocol. Verify before acting. Do not enter patient names, MRNs, or dates of birth. Privacy Policy