| ObjectiveIn this project,patients with unprovoked venous thromboembolism(VTE)who have stopped using anticoagulant drugs are studied to understand the relevant predictors of disease recurrence.By constructing a binary Logistic regression model,the model is externally verified to establish the unprovoked VTE recurrence prediction model,in order to better identify the high-risk population of recurrence,provide targeted interventions for patients,provide reference for clinicians to adjust anticoagulation programs,and provide new ideas for the development of domestic VTE-related research.Methods1.By searching publicly published domestic and foreign studies on the risk factors for recurrence of patients with unprovoked venous thromboembolism,describe the research progress in this field at home and abroad,and screen possible risk factors.Based on the domestic clinical situation,we design the a questionnaire.2.The convenience sampling method was used to select 200 patients with unprovoked venous thromboembolism recurrence in the First Affiliated Hospital of Zhengzhou University from September 2017 to August 2019 as the recurrence group,and 200 patients without recurrence as the control group,and relevant data were collected.After sorting out the relevant data,use IBM SPSS Statistics 21.0 for statistical processing.The patient data are statistically described using mean ± standard deviation,frequency and percentage to meet the continuous variables with normal distribution and uniform variance.Two independent sample t-tests are used for single Factor analysis.The chi-square test was used for statistical inference for binary and unordered multi-categorical variable data,and the non-parametric test was used for ordered multi-categorical variable data.The variables with P<0.05 were included in the binary Logistic regression analysis,and the entry level was set to 0.05,and the elimination level was set to 0.10 to obtain the Logistic regression model.HosmerLemeshow is used to test the goodness of fit of the prediction model,and the area under the receiver operating characteristic curve is used to evaluate the discrimination of the model.The closer the area under the receiver operating characteristic curve is to 1,the better the discrimination of the model.Develop a predictive model.The optimal critical value is calculated according to the Youden index,Youden index=sensitivity+specificity-1,and the corresponding value of the largest Youden index is the optimal critical value.3.To verify the prediction performance of the model,the convenience sampling method was used to select 166 patients with a history of unprovoked venous thromboembolism who attended the First Affiliated Hospital of Zhengzhou University from September 2019 to August 2020.At the same time,the patient’s imaging results and disease diagnosis are collected at the same time as the gold standard for judging whether the patient has recurrence under actual conditions.Calculate the probability of recurrence through the predictive model,compare it with the actual situation,draw the receiver operating characteristic curve,use the area under the receiver operating characteristic curve to evaluate the discrimination of the model,and obtain the accuracy,sensitivity and specificity of the recurrence prediction model based on the data of the four grid table.Results1.We design the a questionnaire which mainly includes gender,age,body mass index,family history,smoking history,drinking history,type of first venous thromboembolism,post thrombotic syndrome,and laboratory test results(blood Routine and coagulation function tests)and other indicators.2.A univariate analysis of the relevant data of 400 patients in the modeling group and 166 patients in the verification group showed that the difference was not statistically significant(P>0.05).3.A univariate analysis of the data of 400 patients in the modeling group showed that the following variables may be the influencing factors for the recurrence of unprovoked venous thromboembolism:age(Z=21.129,P<0.001),gender(χ2=4.002,P=0.045),family history(χ2=26.639,P<0.001),smoking history(Z=8.489,P=0.005),first venous thromboembolism type(χ2=16.448,P<0.001),post thrombotic syndrome(χ2=63.473,P<0.001),fibrinogen degradation products(t=-4.516,P<0.001),D-Dimer(t=-6.490,P<0.001).4.Perform collinearity analysis on the statistically significant independent variables in the univariate analysis,and the variance expansion coefficients are all<3.Incorporating the above independent variables into Logistic regression for multivariate analysis,it is concluded that the first venous thromboembolism is pulmonary thromboembolism(P=0.001,OR=2.733,95%Cl:1.534～4.869),the first venous thromboembolism was pulmonary thromboembolism with deep vein thrombosis of the lower extremities(P=0.007,OR=2.947,95%Cl:1.338～6.489),Male patients(P=0.008,OR=2.450,95%Cl:1.269～4.729),post thrombotic syndrome(P<0.001,OR=4.345,95%Cl:2.489～7.583),D-dimer(P<0.001,OR=3.770,95%CI:2.391～5.944).5.Based on the results of multivariate analysis,the Logistic regression model for predicting the recurrence of unprovoked venous thromboembolism was preliminarily determined:Logit P=-4.456+1.005*the type of first venous thromboembolism is pulmonary thromboembolism+1.081*the type of first venous thromboembolism is deep vein thrombosis and pulmonary thromboembolism+0.896*male+1.469*post thrombotic syndrome+1.385*D-dimer,except that D-dimer is a numerical variable,the above independent variable assignment methods are all(0=No,1=Yes).6.Verification of the prediction model for the recurrence of unprovoked venous thromboembolism:The goodness of fit in the modeling group was tested by HosmerLemeshow,χ2=13.813,v=8,P=0.087 and the null hypothesis was not rejected;Area under the working characteristic curve=0.841(95%CI:0.803～0.880)the best critical value=0.580.The validation group model validates the area under the receiver operating characteristic curve=0.740(95%CI:0.649～0.830).ConclusionThe first venous thromboembolism is pulmonary thromboembolism,first venous thromboembolism is pulmonary thromboembolism with deep vein thrombosis,male,post thrombotic syndrome,D-dimer is the risk factors for recurrence of patients with unprovoked venous thromboembolism.The Logistic regression model including the above 5 risk factors has good conformity and discrimination validity.It can provide evidence for clinical screening of high-risk populations with recurrence of unprovoked venous thromboembolism.It is convenient for medical staff to intervene early,which can reduce the recurrence rate of such patients. |
PDF Full Text Request