Construction And Validation Of Complication Prediction Model For Carotid Artery Stenting

Purpose:To explore the risk factors of long-term complications after carotid artery stenting（CAS）,construct and verify the risk prediction model.Methods:The patients who were hospitalized in the Department of Neurology of Sichuan Provincial People’s Hospital from January 2017 to December 2021 were retrospectively studied.They were diagnosed as carotid stenosis by digital subtraction angiography（DSA）,treated with CAS for the first time,rechecked DSA or computed tomography angiography（CTA）one year after operation.The baseline data,laboratory data,postoperative complications,body mass index（BMI）and antiplatelet drugs were collected.SPSS 23.0 software was used for data analysis.According to the order of review time,they were divided into modeling group and verification group,The ratio of patients in the two groups was 3:1.Multivariate logistic regression analysis was carried out with the long-term complications after CAS as the dependent variable and the factors affecting the long-term complications after CAS as the independent variable.The logical scoring method was used to assign the selected risk factors,the quartile method was used to explore the relationship between the degree of risk and complications,and the risk prediction model of long-term complications after CAS was constructed.Hosmer lemeshow goodness of fit test and receiver operating characteristic curve（ROC）were used to verify the model.Results:Among 313 subjects,73 cases had long-term complications,with an incidence of 23.32%.Among them,tia22 cases（7.03%）,6 cases of minor stroke（1.92%）,2 cases of major stroke（0.64%）,2 cases of myocardial infarction（0.64%）,6cases of death（1.92%）,and 35 cases of restenosis（11.18%）.Multivariate logistic regression analysis showed that smoking history（OR=5.965,95%CI:2.527～14.079）,high LDL（OR=6.290,95%CI:1.828～21.645）,Abnormal BMI（OR=2.298,95%CI:1.023～5.163）and failure to take antiplatelet drugs in accordance with doctors’instructions（OR=65.656,95%CI:12.052～357.683）were independent risk factors for long-term complications after CAS（P<0.05）.The risk prediction model of long-term complications after CAS was constructed as follows:Y=5×Antiplatelet drugs taken without doctor’s advice+2×High LDL（>3.3 mmol/L）+2×Smoking history+1×BMI was abnormal（<18.5 or>24）.According to the model,the score range of long-term complications after CAS is 0～9 points.The area under ROC curve（AUC）of the modeling group was 0.841（95%CI:0.779～0.903）,the specificity was 88.50%,and the sensitivity was 60.40%;Hosmer lemeshow test shows that x~2=3.637,P=0.820,and the prediction probability is 83.8%.The AUC of the validation group was 0.816（95%CI:0.706～0.926）,the specificity was 56.90%,and the sensitivity was 95.00%;Hosmer lemeshow test shows that x~2=12.196,P=0.094,and the prediction probability is 78.2%.Conclusion:Smoking history,high LDL,Abnormal BMI and non-compliance with antiplatelet drugs were significantly correlated with long-term complications after CAS.The risk prediction model of long-term complications after CAS constructed in this study has certain prediction ability,and can provide reference for the prevention of long-term complications after CAS.