The Logistic Regression Predict Model Of Portal Venous System Thrombosis In Cirrhotic Potal Hypertensive Patients After Splenectomy

Objective:To investigate the relevant risk factors of portal venous system thrombosis(PVT) in cirrhotic potal hypertensive patients after splenectomy and to establish a Logistic regression predict model,its value would be evaluated on predicting PVT of these patients.Methods:A retrospective study of the clinical features including hypersplenism and/or oesophagogastric varices of 119 cirrhotic portal hypertensive patients who received splenectomy in our hospital from Jan.2009 to Dec.2013 were enrolled, to divide into two groups using the standard of turning out PVT two weeks after splenectomy,using one-way analysis and multiple Logistic regression analysis towards the relevalnt peri-operative risk factors to sort out the independent risk factors, Based on the multivariate analysis results, we draw relevant ROC curve, Logistic regression predict model was established and its value would be evaluated on predicting PVT of these patients.Results:Logistic regression analysis showed that three independent risk factors affecting PVT two weeks after splenectomy,which were platelet count(PLT), mean platelet volume(MPV) and D-Dimer,the data of which came from five to seven days after splenectomy,two independent protect factors were usage of anticoagulation therapy (UAT) and usage of reducing portal pressure therapy (URPT), Logistic regression predict model established according to the above factors was expressed as Logit P =-9.165+0.664×PLT (×1011/L)+0.413×MPV (fL)+0.662×D-Dimer (mg/L)-1.674 ×UAT(Yes=1,No=0)-1.518×URPT(Yes=1,No=0), and the cut off value was-1.14,the area under the receiver operating characteristic curve(AUROC) and the accuracy were 0.865 and 84.03%, the cut off value of PLT、MPV、D-Dimer were 4.42×1011/L、13.30fL、 2.55mg/L, respectively.Conclusion:Our study concluded:when PLT exceeding 4.42×1011/L、MPV exceeding 13.30fL and D-Dimer exceeding 2.55mg/L from five to seven days after splenectomy would increase the risk of turning out PVT,UAT and URPT could lessen the risk of turning out PVT,The Logistic regression predict model was established in this study to predict PVT two weeks after splenectomy with a high sensitivity,specificity and accuracy,it provided mathematical model and cut off value for predicting PVT after splenectomy, it could help doctors to choose reasonable treatment methods in perioperative period, as for a further comment, to verify the accuracy had important clinical significance and value.