The Research Of PICC-associated Thrombosis Risk Prediction Model In Patients With Cancer

Objective:Based on the meta-analysis,this study obtained PICC-related thrombotic risk factors and their combined effect values in tumor patients,and constructed a risk assessment model for predicting PICC-associated thrombosis in patients with tumors which can help to identify patients with high risk of PICC-related thrombosis.For this condition,the venous access device selection and health education work of PICC-related thrombus will be carried out to help reduce the prevalence of PICC-related thrombosis in patients with cancer,control medical costs,and promote the health of patients.Methods:First,the meta-analysis was used to analyze the literature on the risk factors associated with PICC-related thrombosis in patients with tumors.The ORi and the average exposure rate Pi of PICC-associated thrombotic risk factors were obtained in tumor patients.Then,based on the results obtained by meta-analysis,a logistic regression model was constructed,R software was used to simulate the binomial distribution function to generate the corresponding random data set,and the risk grade partitioning node of the model was determined and the patients are divided into three levels of risk including low,medium,and high risk levels.Finally,the ROC curve was fitted with the incidence data of the actual population,and the prediction performance of the logistic regression model was verified.Result:Thirty-three studies were included in this meta-analysis.Thirteen risk factors of PICC-associated thrombosis were included in the prediction model,including gender,obesity,pathological type,clinical stage,diabetes,history of thrombosis,D-dimer elevation,History of catheterization,cephalic catheterization,5Fr catheter,catheter tip position,history of chemotherapy,use of anticoagulant drugs,combined OR values were 2.25,4.32,3.95,4.21,2.86,2.76,2.69,2.4,3.53,4,5.04,1.51,3.06,respectively.The risk prediction model based on meta-analysis wasLogit?P?=?+0.811x₁+1.463x₂+1.374x₃+1.437x₄+1.051x₅+1.015x₆+0.99x₇+0.875x₈+1.261x₉+1.386x₁₀+1.617x₁₁+0.412x₁₂+1.118x₁₃.Based on local incidence of PICC-associated thrombosis in tumor patients,Logistic regression model was Logit?P?=-6.59+0.811x₁+1.463x₂+1.374x₃+1.437x₄+1.051x₅+1.015x₆+0.99x₇+0.875x₈+1.261x₉+1.386x₁₀+1.617x₁₁+0.412x₁₂+1.118x₁₃.Among them,x₁,x₂,x₃,...x₁₃3 represent gender,obesity,pathological type,clinical stage,diabetes,thrombosis history,D-dimer elevation,catheterization history,cephalic vein catheterization,5Fr catheter,catheter tip Location,history of chemotherapy,use of anticoagulant drugs.According to the trend of the probability of simulation results,the patient with incidence probability P?0.216 was the low-risk group,0.216<P?0.524 was the middle-risk group,and P>0.524 was the high-risk group.The area under the ROC curve and the 95%CI was 0.731?0.662,0.799?,the sensitivity was 0.958,and the specificity was 0.496.Conclusion:In this study,the risk factors of PICC-associated thrombus prediction models in tumor patients were comprehensively evaluated by meta-analysis,and a risk prediction model was constructed based on this.The results of this study can be used to distinguish the risk level of PICC-related thrombosis in tumor patients.The model has been validated.It can also provide a reference for the risk prediction of PICC-related thrombosis in tumor patients,as well as the selection of venous access devices and the individualized prevention of PICC-related thrombosis.