In this study, 339 patients (154 men, 185 women) with a median age of 74 years (range 23-97) admitted to the Stroke Unit, Department of Neurology in 1986, have been followed up for 14 years. The diagnoses were intra-cerebral hemorrhage (ICH, 30, 8.8%), cardioembolic cerebral infarction (CE, 71, 20.9%), lacunar infarction (LI, 47, 13.9%) and atherosclerotic cerebral infarction (ACI, 191, 56.3%). The cumulative probabilities of recurrent stroke rates at 1-, 5- and 10-year follow-ups were 13.5% (95% confidence interval, CI, 9.6-17.4), 38.7% (95% CI 32.6-44.8) and 53.9% (95% CI 46.7-61.1). According to Cox proportional hazard regression analysis, age, severity of stroke, previous stroke and systolic blood pressure are each of importance in predicting recurrent stroke. During the observation period, 290 patients (85.5%) died. The mortality rate of 24.5% during the first year was 4.5 times higher compared to the normal population of the same age and gender. Patients with LI had lower mortality rates compared to ICH by the log rank test (p =0.0275), to CE (p =0.000) and to ACI (p =0.049). Thirty-nine percent of all vascular deaths after the first year were caused by recurrent strokes. Fatal index/recurrent stroke occurred statistically more frequently in the CE group versus the non-CE one (p =0.005). Cox proportional hazard regression analysis indicated that age, severity of stroke, previous stroke, heart failure and fasting blood glucose exceeding 6 mmol/l or history of diabetes were each predictors of mortality. In conclusion, this study has shown the worse outcomes for all subtypes of stroke compared to the normal population and also clearly pointed out independent predictors of recurrent stroke or death at the time of diagnosis. Copyright
2001. Vol. 12, no 3, 171-180 p.