The study demonstrates that clinical-radiological causes and outcome of car
dio-embolic infarcts in a population-based study correspond to a well-ident
ified stroke pattern. Cardio-embolic infarcts was diagnosed in 882 cases (3
7.9%) of 2,330 consecutive first-ever stroke patients included in a prospec
tive population-based stroke registry over a 14-year period (1985-1997). Th
irty-three criteria out of 98 were introduced into a monovariate analysis a
nd the significant variable were introduced into a multivariate analysis to
identify significant criteria to define stroke patterns in cardio-embolic
infarction. Cardiac sources of embolus included atrial arrhythmia, valvular
heart disease (19%), and cardiac failure (18%). Patients with cardio-embol
ic infarction showed a significantly higher rate of female predominance (p
<0.001), history of ischemic heart disease (p <0.001), acute stroke onset (
p <0.05) headache (p <0.05) previous treatment by anti-platelets and anti-K
-vitamin (p <0.001), Wernicke aphasia (p <0.001), severe deficit (p <0.001)
and death (p <0.001). After a logistic procedure, female gender and ischem
ic heart disease were the two independent risk factors associated with card
io-embolic stroke. Cardio-embolic stroke is a specific subtype of stroke wi
th its own clinical, radiological, etiological and prognostic characteristi
cs. In the acute stage, it is necessary to identify quickly this type of st
roke because of severe prognosis and appropriate treatment.