Abstract: Objective To study on the risk factors and to construct a noninvasive prediction model for upper gastrointestinal bleeding (UGB) in patients with hepatitis B related cirrhosis.Methods The clinical data of 255 patients with hepatitis B related cirrhosis admitted to the Affiliated Hospital of Xuzhou Medical University from January 2017 to June 2019 were collected. They were devided into two groups and studied by retrospective case-control method. 119 Patients with UGB were in the experimental group and 136 patients without UGB were in the control group. A non-invasive prediction model for UGB in patients with hepatitis B related cirrhosis was established through Logistic regression analysis. The receiver operating characteristic curve (ROC curve) of the prediction model with various indicators were drawn and the area under the curve (AUC) was calculated. Results It was shown that hepatocellular carcinoma, albumin, fibrinogen, leukocyte count, esophageal and gastric varices, platelet/spleen length were independent influencing factors for UGB in the model (P<0.05). The sensitivity and specificity of the risk factors-based model for screening UGB in hepatitis B related cirrhosis were 80.7% and 86.8%, respectively. The area under the ROC curve (AUC) was 0.904 (0.867 ~ 0.941), which was better than any other single index.Conclusion The established non-invasive prediction model based on six independent influencing factors in this study has good accuracy for UGB, which warrants further verification and model tuning.

Key words: Hepatitis B related cirrhosis, Upper gastrointestinal bleeding, Risk factors, Prediction model