Abstract
BACKGROUND: Observational data have reported associations between patent foramen ovale (PFO), cryptogenic stroke and migraine. However, randomized trials of PFO closure do not demonstrate a clear benefit either because the underlying association is weaker than previously suggested or because the trials were underpowered. In order to resolve the apparent discrepancy between observational data and randomized trials, we investigated associations between (1) migraine and ischemic stroke, (2) PFO and ischemic stroke, and (3) PFO and migraine. METHODS: Eligibility criteria were consistent; including all studies with specifically defined exposures and outcomes unrestricted by language. We focused on studies at lowest risk of bias by stratifying analyses based on methodological design and quantified associations using fixed-effects meta-analysis models. RESULTS: We included 37 studies of 7,686 identified. Compared to reports in the literature as a whole, studies with population-based comparators showed weaker associations between migraine with aura and cryptogenic ischemic stroke in younger women (OR 1.4; 95% CI 0.9-2.0; 1 study), PFO and ischemic stroke (HR 1.6; 95 CI 1.0-2.5; 2 studies; OR 1.3; 95% CI 0.9-1.9; 3 studies), or PFO and migraine (OR 1.0; 95% CI 0.6-1.6; 1 study). It was not possible to look for interactions or effect modifiers. These results are limited by sources of bias within individual studies. CONCLUSIONS: The overall pairwise associations between PFO, cryptogenic ischemic stroke and migraine do not strongly suggest a causal role for PFO. Ongoing randomized trials of PFO closure may need larger numbers of participants to detect an overall beneficial effect.