| Background and Objective:The rate of recurrence after ischemic st roke is high,and the morbidity and mortality of recurrent strokes are high er than index stroke.Therefore,the purpose of this study was to construct and validate a prognostic risk prediction model which can individually pr edict the risk of recurrence within 1 year in patients with acute first-episo de non-cardiac ischemic stroke(NCIS).In order to clear the key populati ons for recurrence of the patients in clinical work.Furthermore,we can pr ovide a better individualized secondary prevention for patients and reduce the rate of recurrence.Methods:(1)Research objects and data: A total of 511 patients with acute first-episode NCIS who met the inclusion and exclusion criteria fro m the Third Xiangya Hospital of Central South University were retrospec tively recruited from January 2019 to December 2019.The demographic characteristics,basic clinical data,laboratory and imaging data of the pati ents were collected and followed up for 1 year.Patients were divided into a recurrence group and a non-recurrence group according to whether the s troke recurrence occurred.(2)Construction of the prediction model: In order to construct and v alidate the predictive model,they were assigned to the developing set and validating set in a proportion of 7: 3 according to the date of admission.Univariate logistic regression analysis was used to screen possible predict ors,and multivariate logistic regression analysis was used to construct a p rediction model and draw a nomogram.(3)Validation of the prediction model: We respectively use receiver operating characteristic curve and area under the curve(AUC),Hosmer-L emeshow test and its calibration plots,and decision curve analysis(DCA)to evaluate and validate the discrimination,calibration,and clinical applic ability of the predictive model in the development and validation setting.Results:(1)Research objects: 439 patients who met the criteria were included,and the 1-year stroke recurrence rate was 10.93%(N=48).(2)Construction of the prediction model: The patient data from Janu ary 1,2019 to October 5,2019 was used as the developing set(307 cases),and the patient data from October 6,2019 to December 31,2019 was use d as the validating set(132 cases).Univariate Logistic regression analysis showed that age≥70 years old,baseline SBP≥140mm Hg,hypertension,d iabetes,coronary heart disease,WBC>7.35*10^9/L,N%>64.65%,N>4.39*10^9/L,NLR>3.20*10^9/L,GLY>6.1mmol/L,multiple infarcts and int racranial large vessel stenosis were possible predictors of 1-year stroke re currence in patients with acute first-episode NCIS.The results of multivar iate Logistic regression analysis showed that age≥70 years(OR=2.679;95%CI: 1.134-6.333;P=0.025),multiple infarcts(OR=6.762;95%CI: 2.863-15.972;P=0.000),hypertension(OR=4.106;95%: 1.106-15.235;P=0.035),diabetes(OR=2.500;95%CI: 1.042-6.003;P=0.040)and WBC>7.35*10^9/L(OR=5.518;95%CI:2.090～14.565;P=0.001)were independ ent risk factors for 1-year stroke recurrence in patients with acute first-epi sode NCIS(P<0.05).The five independent risk factors were used to cons tructing prediction model and nomogram.(3)Validation of the prediction model: The results of area under the ROC curve of the prediction model respectively were 0.836(95%CI: 0.753-0.920)and 0.746(95%CI: 0.746)in the development and validation sett ing.The results of Hosmer-Lemeshow test respectively showed ? 2=16.53,P=0.0853>0.05、 ? 2=29.04,P=0.0012<0.05 in the development and v alidation setting.The results of DCA showed the predictive model had cli nical applicability when a threshold probability of >10%.Conclusion:This study constructed and validated a predictive mode l for 1-year stroke recurrence in acute first-time NCIS patients combining demographic characteristics,clinical risk factors,inflammatory factors an d imaging indicators.The model has good discrimination,accuracy and cl inical applicability,but the accuracy in validation set is not good. |
PDF Full Text Request