Screening Of Indicators And Establishment Of Models Of Risk Warning In Venous Thromboembolism

Venous Thromboembolism(VTE)has insidious onset and poor prognosis,and its incidence is increasing year by year,which seriously endangers patients' life and increases medical cost.The etiology of VTE is complex and clinical tools for assessment are numerous.Early identification of the risk and accurate assessment have become research hotspots in the field of medicine.Previous studies believe that VTE is induced by the coaction of multiple links and factors.New risk factors of VTE keep emerging in recent years especially with the social industrialization and population aging.VTE is a serious complication of inpatients.Their effective data in EMR(electronic medical record)can be used for automatic early warning of VTE,which will promptly suggest the medical staff to take corresponding effective measures to reduce the rate of sudden death and disability of VTE high-risk groups.That is of great significance.Therefore,VTE risk warning indicators need to be further analyzed and evaluated,and the establishment of VTE risk warning model will be a quick and effective method to realize VTE automatic warning based on EMR.On the basis of case review,this study used Logistic regression analysis to screen out new risk warning indicators and build VTE risk warning model,discussed the independent prediction role of each risk warning indicator in the occurrence of VTE,and established a new assessment and warning method.In this way,the primary prevention and accurate treatment of VTE can be realized,and the clinical outcomes can be finally changed to reduce the medical cost and economic burden,so as to further guarantee the life of VTE high-risk groups.Objective1.To understand the incidence and risk warning indicators of VTE of inpatients.2.To screen out VTE independent risk warning indicators through single and multiple factors analysis,and build preliminary VTE risk warning model.3.To evaluate and externally verify the early warning model,and discuss the clinical application value of the VTE risk early warning model,so as to provide scientific basis for the realization of clinical automatic VTE risk early warning.Method1.VTE patients,who meet the requirements of the study and have been discharged from a Third-grade general hospital in Jiangsu Province from January 1,2013 to December 31,2016,were selected to analyze the incidence of VTE and related indicators of VTE patients in a single center.2.Using the method of retrospective case-control study in investigation,VTE patients who met the inclusion and exclusion criteria were selected as VTE group,and non-VTE patients who matched the general data of department,age,gender and other conditions were selected as a control group at frequency of 1:1.The VTE risk warning indicators were analyzed by single factor analysis.The P<0.3 indicators in single factor analysis were included in the follow-up multifactor analysis.The independent risk warning indicators of VTE were screened by multifactor Logistic regression analysis.3.The preliminary construction of VTE risk warning model was based on independent risk warning indicators screened by Logistic regression analysis,and truncated values of screening VTE were obtained and an evaluation of the model was carried out.The validity of Caprini risk assessment scale and VTE risk warning model was compared through ROC curve analysis.Validation data set was established,and truncation value of VTE risk warning model was used to evaluate the inspection efficiency of the validation data set for VTE patients.Results1.According to the inclusion and exclusion criteria of the study,257 cases of VTE in 914 hospitalized patients from January 1,2013 to December 31,2016 were randomly sampled as the VTE group of the follow-up study.The incidence of VTE in single-center inatients was about 2.85‰.Among them,PE was 1.20‰ and DVT was 1.91‰.The number of male and female VTE patients was 440(48.14%)and 474(51.86%)respectively.The number of PE,DVT and PE patients complicated with DVT was 301(32.93%),530(57.99%)and 83(9.08%).The number of medical and surgical VTE patients was 324(35.45%)and 590(64.55%).According to the distribution of departments,the order was orthopaedics(330,36.11%),general surgery(162,17.72%),respiratory medicine(156,17.07%),interventional radiology(82,8.97%),cardiology(48,5.25%)and general practice department(42,4.60%).2.41 variables were analyzed by single factor analysis.Among them,15 variables with statistical significance P<0.05 were hospitalization time X4(P=0.003),payment mode Xs(P=0.017),cough X6(P=0.019),expectoration X7(P=0.002),pleural chest pain X10(P<0.001),cyanosis X11(P=0.011),chest tightness and shortness of breath X15(P<0.001),unexplained syncope X16(P<0.001),unilateral lower extremity pain X18(P<0.001),hypertension X23(P=0.006),thrombin time X31(P=0.028),fibrinogen degradation product X34(P<0.001),D-dimer X36(P<0.001),number of white blood cells X39(P<0.001),Caprini score X41(P<0.001).3.28 variables of P<0.3 in univariate analysis were included into subsequent multivariate logistic regression analysis.The results showed that pleural chest pain X10(P<0.001),post-exhaustion shortness of breath X14(P=0.045),chest tightness X15(P<0.001),unilateral lower extremity pain X18(P<0.001),smoking X25(P=0.005),fibrinogen degradation product X34(P<0.001),Caprini score X41(P=0.004)were seven unique factors.Finally,risk warning indicators were selected to participate in the construction of VTE risk warning model.4.ROC curve was used to evaluate the validity of Caprini risk assessment scale and VTE risk warning model,AUC(95%CI),truncation value(95%CI),accuracy,Youden index(95%),sensitivity and specificity were 0.596(0.552,0.638),>5(>4,>5),61.3%,0.226(0.167,0.290),26.07%,96.50%,and 0.960(0.940,0.976),>0.438(>0.263,>0.504),92.2%,0.844(0.789,0.879),92.61%,91.83%,respectively.The difference between the two curves was statistically significant(Z=14.521,P<0.0001).In Validation dataset data,the correct rate and the Youden index of the VTE screening using the VTE risk warning model were 81.8%and 62.5%,respectively.Conclusion1.The incidence of VTE was higher in clinical inpatients(about 285 cases/100,000 persons),among which the incidence of DVT was higher than PE,VTE was more common in elderly patients,the numbers of male and female patients were similar,the number of surgical patients was higher than that of medical patients.2.In clinical routine work,Caprini score is an independent risk early-warning index for VTE occurrence.In addition,6 variables such as pleural chest pain,fatigue shortness of breath,chest tightness,unilateral lower limb pain,smoking,fibrinogen degradation products are independent risk early-warning indicators for VTE.3.VTE risk warning model is more effective than Caprini risk assessment scale in testing VTE.External validation shows that VTE risk warning model has a high-test efficiency for VTE screening.