| Objective:To explore the risk factors of rebleeding after esophageal and gastric var iceal bleeding(EGVB)endoscopic tissue glue combined with polycinol injection,and establish a predictive model by Logistic regression.Predictive value of rebleeding in patients with EGVB was compared by self-built model,classic Child-Pugh score and MELD score,providing a reference for guiding the prediction of rebleeding af ter EGVB endoscopic tissue glue combined with polycinol injection.Methods:495 patients in the Gastroenterology Department of Affiliated Hospital of Z unyi Medical University who were treated with endoscopic treatment of EGVB in c irrhosis from January 1st of 2017 to December 31 st of 2018 were selected.The ab ove patients were randomly divided into Training group(340 cases)and Validation group(155 cases)according to the ratio of 7:3.The patient-related information,incl uding name,gender,etiology,HE,ascites,combined with HCC or not,TBIL(μmol/L/mg/L),ALB(g/L),Na+(mmol/L),Glu(mmol/L),Ca+(mmol/L),PT(s),internati onal standardized ratio,Cr(mg/dl),HGB(g/L),PLT(* 109/L),tissue gel(dosage ml),polycinol(dosage ml),using blood products or not,anti-infection or not,were c ollected and Child-Pugh scores and MELD scores were calculated based on relevant data.Quantitative data that meet the normal distribution were indicated by(`x±s),the comparison between the two groups was analyzed by the independent sample t test.Non-normal data was expressed by M(P25,P75),and the comparison between the two groups was analyzed by M-H U test.Chi-square test or Fisher’s exact pro bability test was used to test qualitative data.Logistic regression model was mainly used for binary classification variables.In this paper,patients were classified by n o bleeding=0 and bleeding=1.Stepwise of multivariate Logistic regression analysiswas used to analyze the independent risk factors for rebleeding after treatment and the model with minimum AIC value was selected as the prediction model.And the ROC curve evaluation model was adopted.Results: Gender(0=male,1=female),etiology(0=bile or alcoholic,1=others),HGB,Ca+,surgical method(tissue glue+polycinol as control,2=sclerosis/polycinol,3=tiss ue glu+polycinol+metal clip),use blood products or not(0=no,1=yes),anti-infection o r not(0=no,1=yes)were included in the model,where gender(female),surgical metho d2(tissue glue /polycinol)and use blood products or not(yes)were risk factors fo r rebleeding.Etiology(other),HGB,Ca+,surgical method 3(tissue glue+polycinol+Metal clip),and anti-infection or not(yes)were protective factors for rebleeding.R=5.29+0.53gender(0=male,1=female)-0.09etiology(0=bile or alcoholic,1=other)-0.02 H GB-2.12Ca++0.43 surgical method2(0=other,1=tissue glue /polycinol)-0.87 surgical meth od3(0=other,1=tissue glue+polycinol +Metal clip)+1.20 using blood products or not(0=no,1=yes)-0.75anti-infection or not(0=no,1=yes).And by R software,a nomogram was produced.The higher the total score,the higher the risk of rebleeding.ROC was used to compare self-built model,Child-Pugh score and MELD score.Child-Pugh score,MELD score and AUC of self-built model in Validation group were 0.5325,0.5032 and 0.7074,respectively.The self-built model can have a better prediction o f rebleeding.Conclusion:This study established the Predictive Model of Rebleeding after Endoscopic Treatment of Esophageal and Gastric Variceal(EGVB)in Cirrhosis,which can be evaluated by ROC and can be used as a risk forecasting model of rebleeding after endoscopic treatment of EGVB in cirrhosis.R=5.29+0.53gender(0=male,1=female)-0.09etiology(0=bile or alcoholic,1=other)-0.02HGB-2.12Ca++0.43 surgical method2(0=other,1=tissue glue /polycinol)-0.87 surgical method3(0=other,1=tissue glue+polycinol +Metal clip)+1.20 using blood products or n ot(0=no,1=yes)-0.75anti-infection or not(0=no,1=yes)Predictive Model of Rebleeding after Endoscopic Treatment of Esophageal and Gastric Variceal Bleeding(EGVB)in Cirrhosis can predict the risk of rebleeding in patients with cirrhosis treated with EGVB endoscopic therapy... |
PDF Full Text Request