Risk Factors And Prediction Model Of Acute Ischemic Stroke Recurrence Based On Propensity Score Matching

Objective:Propensity score matching method was used to explore the risk factors of acute ischemic stroke recurrence,and the prediction model of line graph was constructed to predict the risk of acute ischemic stroke recurrence,so as to provide clinical decision-making basis.Methods:In this study,patients with acute ischemic stroke who were first diagnosed and discharged from Shanxi Cardiovascular Disease Hospital and Taiyuan Central Hospital from July 2020 to July 2021 were collected as research objects,and the included subjects were followed up for one year.After the follow-up,Lost follow-up due to die or other causes,during the follow-up period,the final selected patients were grouped according to whether new cerebral infarction occurred.The baseline covariates(potential confounders)of the two groups were matched using the propensity score matching method,so that the baseline covariates of the two groups were balanced.After matching,the two groups of patients were compared,and the risk factors with statistical differences were included in the multivariate logistic regression model to explore the risk factors of acute ischemic stroke recurrence.A line graph prediction model based on the matched multivariate logistic regression results was constructed to predict the recurrence risk of acute ischemic stroke.The predictive effect of the model was evaluated by receiver operating curve(ROC curve),clinical decision curve(DCA),calibration curve and other indicators.Results:A total of 949 patients with acute ischemic stroke were included in this study,among which 36 patients with incomplete information due to loss of follow-up or other reasons were excluded.Finally,913 patients with first-episode acute ischemic stroke were included.After 1-year follow-up,159 patients in the relapsed group and754 patients in the non-relapsed group,respectively,with a 1-year recurrence rate of17.4%.After matching propensity scores at 1:1 ratio,159 pairs of balanced samples between groups were obtained,among which 9 pairs were accurately matched and150 pairs were fuzzy matched.logistic regression showed that hypertension(OR =2.387,P = 0.018,95%CI: 1.162~4.900),diabetes mellitus(OR = 2.649,P = 0.017,95%CI: 1.189~5.899),total cholesterol(OR = 0.023,P < 0.05,95%CI: 0.007~0.070),triglyceride(OR = 2.791,P < 0.05,95%CI: 1.912~4.076),low density lipoprotein(OR = 0.793,P = 0.034,95%CI: 0.639~0.983),homocysteine(OR = 0.957,P = 0.001,95%CI: 0.933-0.982)and D-dimer(OR = 2.647,P < 0.024,95%CI: 1.137-6.162)were risk factors for recurrent acute ischemic stroke.Based on this,the prediction model of line graph is constructed,the prediction model of line graph was constructed.The results showed that the AUC value was 0.867(95%CI: 0.827-0.908),the Hosmer Lemeshow goodness of fit test was used to evaluate the calibration and showed that x2= 6.546,P = 0.586.The calibration curve of the model was basically consistent with the ideal curve.The results of the clinical decision curve showed that the model had certain clinical practicability.Conclusion:Hypertension,diabetes mellitus,the level of cholesterol,triglyceride,low density lipoprotein,homocysteine and D-dimer were risk factors for recurrence of acute ischemic stroke within 1 year.The prediction model based on the matching multivariate logistic regression results has good predictive ability and clinical practicability for the recurrence of acute ischemic stroke.