• BACKGROUND AND PURPOSE
    • Several risk scores have been developed to predict the stroke risk after transient ischemic attack (TIA). However, the validation of these scores in different cohorts is still limited. The objective of this study was to elucidate whether these scores were able to predict short-term and long-term risks of stroke in patients with TIA.
  • METHODS
    • From the Fukuoka Stroke Registry, 693 patients with TIA were followed up for 3 years. Multivariable-adjusted Cox proportional hazards model was used to assess the hazard ratio of risk factors for stroke. The discriminatory ability of each risk score for incident stroke was estimated by using C-statistics and continuous net reclassification improvement.
  • RESULTS
    • The multivariable-adjusted Cox proportional hazards model revealed that dual TIA and carotid stenosis were both significant predictors for stroke after TIA, whereas abnormal diffusion-weighted image was not. ABCD3 (C-statistics 0.61) and ABCD3-I (C-statistics 0.66) scores improved the short-term predictive ability for stroke (at 7 days) compared with the ABCD2 score (C-statistics 0.54). Addition of intracranial arterial stenosis (at 3 years, continuous net reclassification improvement 30.5%; P<0.01) and exclusion of abnormal diffusion-weighted imaging (at 3 years, continuous net reclassification improvement 24.0%; P<0.05) further improved the predictive ability for stroke risk until 3 years after TIA.
  • CONCLUSIONS
    • The present study demonstrates that ABCD3 and ABCD3-I scores are superior to the ABCD2 score for the prediction of subsequent stroke in patients with TIA. Addition of neuroimaging in the ABCD3 score may enable prediction of long-term stroke risk after TIA.