The establishment of ischemic stroke recurrence prediction model and its application value

Abstract

Objective To establish a prediction model for 3-years recurrence after initial ischemic stroke by Cox proportional hazards regression and individual prognostic Index(PI)equation, and to evaluate its application value and external reality. Methods The inpatients with cerebral ischemic stroke hospitalized in Neurology Department in North China University of Science and Technology Affiliated Hospital were finally internalized between January 2013 and December 2013.Follow-up study on recurrence was carried out between January 2016 and December 2016.The recurrence prediction model was constructed by the Cox proportional hazards regression model.During January 2016 and December 2016, data of patients with ischemic stroke were prospectively continuously collected.And PI equation was used to verify its external reality in ischemic stroke patients. Results A total of 184 cases had stroke recurrence during the follow-up period.The Cox proportional hazards regression model analysis showed that age(RR=1.303, 95%CI: 1.019-1.666), history of heart disease(RR=1.788, 95%CI: 1.127-2.836), hypertension(RR=1.897, 95%CI: 1.097-3.280), diabetes(RR=1.674, 95%CI: 1.015-2.760)and total cholesterol (RR=2.136, 95%CI: 1.396-3.266) were the independent risk factors for stroke recurrence.The established recurrence model was correlated with individual PI equation, which was PI=0.265X1+ 0.581X2+ 0.640X3+ 0.515X4+ 0.759X5.By the validation study of PI equation to predict stroke recurrence among patients admitted later, the sensitivity was 0.719, specificity was 0.769, and accuracy was 0.800. Conclusions Age, history of heart disease, hypertension, diabetes, and total cholesterol are independent risk factors for recurrence of ischemic stroke.And the PI for predicting stroke recurrence within 3 years after initial stroke is successfully established, which is good and helpful for predicting ischemic stroke recurrence. Key words: Stroke; Risk factors; Cox's regression model; Prognosis index