The optimal diagnostic evaluation for establishing the risk of stroke among patients presenting to the emergency department (ED) with a transient ischemic attack has not been determined. The objective of this review is to assess the ability of diffusion-weighted magnetic resonance imaging (MRI) to predict the short-term risk of stroke.

MEDLINE, EMBASE, and the Cochrane Library were queried to identify studies examining the use of diffusion-weighted MRI in patients with classically defined transient ischemic attack. The primary outcome measure was the rate of stroke at 48 hours. Two reviewers determined study eligibility and extracted data. Quality was assessed according to published recommendations for the design and reporting of prognostic studies.

One thousand six hundred ninety-six abstracts were identified and 35 articles underwent full-text review. Six cohort studies met the inclusion criteria but were limited by selection bias and differences in duration and completeness of follow-up. Results were not consistent across studies, with 5 reporting higher rates of stroke among diffusion-weighted MRI–positive patients, whereas 1 study reported higher rates in diffusion-weighted MRI–negative patients. Among the 4 studies (N=629 patients) reporting 48-hour outcomes, the risk of stroke ranged from 0% to 2.9% in patients with negative diffusion-weighted MRI findings compared with 0% to 9% among those with positive diffusion-weighted MRI results.

Studies of variable quality, consistency, and precision suggest that diffusion-weighted MRI may identify patients at sufficiently low risk to warrant ED discharge and close outpatient follow-up.