Construction And Verification Of Risk Prediction Model Of Central Venous Catheter Occlusion

Objective:1.Investigate and analyze the risk factors of CVC occlusion to provide reference for the construction of prediction model;2.Construct a CVC occlusion risk prediction model and verify and evaluate it.Methods:Selected inpatients in the intensive care unit,general surgery,neurosurgery,neurology,gastroenterology,and geriatrics department in a tertiary A general hospital from March to May 2020,and inpatient maintenance of the CVC as the research object.A "Questionnaire on Related Factors of Central Venous Catheter occlusion" was drawn up.According to the questionnaire,the subjects were followed up and observed from the day the catheter was inserted,and relevant data of the patients were collected.Use SPSS20.0 and Med Calc software for data analysis and processing.According to the catheter insertion time,they were divided into modeling group and verification group according to 3:1.Taking the influencing factors of the occurrence of catheter occlusion as the independent variable,and whether the occurrence of catheter occlusion is the dependent variable,a single factor analysis of each factor was performed.Multivariate Logistic regression analysis was performed on the variables with statistical significance in the analysis results,and the independent risk factors affecting CVC occlusion were screened and determined.The logical scoring method is used to assign values to risk factors and the quartile method is used to grade the degree of risk,and a risk prediction model for CVC occlusion is initially constructed.The Hosmer-Lemeshow goodness-of-fit test and receiver operating characteristic curve(ROC)were used to evaluate the established CVC occlusion prediction model in the modeling group and the verification group,respectively.Results:A total of 440 subjects(330 in the modeling group and 110 in the verification group)were included in this study,and a total of 86(61 in the modeling group and 25 in the verification group)patients had catheter occlusions,the incidence rate was 19.54%,of which complete occlusions In 12 cases,74 cases were partially blocked.The results of univariate analysis showed that there were statistically significant differences in age,assessment of self-care ability,history of thrombosis,history of central venous catheterization,hypertension,diabetes,platelet count,D-dimer,blood products,and parenteral nutrition(P<0.05)Multivariate logistic regression analysis showed that the history of thrombosis(OR=8.894,95%CI: 3.118 ～ 25.371),diabetes(OR=2.133,95%CI: 1.013 ～ 4.488),D-dimer(OR=2.800,95 %CI: 1.400～5.599),parenteral nutrition(OR=2.304,95%CI:1.161～4.572)are independent risk factors for CVC occlusion(P<0.05).Logical scoring method was used to assign risk factors to a CVC occlusion r isk prediction model: Y=3×thrombosis history+1×diabetes+1×D-dimer(≥0.5mg/L)+1×parenteral nutrition.According to the model,the risk score ranges from 0 to 6 points.According to the quartile method to classify the degree of risk,the following results are obtained: 0 is low risk,1 to 3 is medium risk,and 4 to 6 is high risk.Examination of the occurrence of catheter occlusion in the two groups showed that the area under the ROC curve(AUC)of the modeling group was 0.745(95%CI: 0.694～0.791),the specificity was 73.98%,and the sensitivity was 65.57%;Hosmer-Lemeshow test It shows that =9.411,P=0.309,and the predicted probability is 87.00%.In the verification group,25 cases of catheter occlusion(22.72%),the area under the ROC curve(AUC)was 0.750(95%CI: 0.658 ～ 0.827),the specificity was 75.29%,and the sensitivity was 68.00%;Hosmer-Lemeshow test showed that,14.263,P=0.075,the predicted probability is 84.50%.Conclusion:The risk prediction model constructed in this study includes four independent risk factors: thrombosis history,diabetes,D-dimer,and parenteral nutrition.The established model assigns values to each independent risk factor and grades the degree of risk.The assessment content and method are simple.It has been verified that this model has good predictive ability and can provide a reference for the screening of high-risk patients with clinical CVC occlusion.The research object of this study only comes from a tertiary A hospital with a relatively small sample size.The collection of some factors affecting CVC occlusion is incomplete,and it will be further revised and improved in the future.