Analysis Of Risk Factors For Rebleeding Of Ⅱa And Ⅱb Peptic Ulcers Without Endoscopic Treatment And Establishment Of Predictive Models

Objective: To analyze the risk factors of early rebleeding in Forrest Ⅱa and Ⅱb patients without endoscopic treatment,to construct a new risk factor prediction model,and to evaluate the ability of AIMS65 scoring system,Glasgow Blatchford(GBS)scoring system,Rockall(RS)scoring system to predict early rebleeding in Forrest Ⅱa and Ⅱb patients without endoscopic therapy.Methods: A retrospective analysis is made of inpatients who were admitted to the Third Xiangya Hospital of Central South University from January 2013 to December 2020 and are diagnosed as peptic ulcer by gastroscopy,with Forrest grade Ⅱa or Ⅱb,and don’t receive endoscopic treatment.A total of 110 cases is concluded.According to whether rebleed within 3 months after successful hemostasis,they are divided into a rebleeding group(17 cases)and a non-bleeding group(93 cases).The independent risk factors of early rebleeding are analyzed,and a risk factor prediction model is established.The AIMS65,GBS,and RS scores of the patients are calculated respectively,the receiver operating characteristic curve is drawn,and the area under the working curve is calculated.Ability to treat early rebleeding in Forrest Ⅱa and Ⅱb patients.Results:1.A total of 99 patients are collected in this study,including87 males and 23 females,aged(57.04±16.880)years,and 17 are rebleeding within 3 months.2.Univariate analysis show that there are statistically significant differences in age,NSAIDS use,antiplatelet drug use,shock,fibrinogen concentration,hemoglobin value,and albumin(P<0.05).The significant indicators of univariate analysis are included in the multivariate logistic regression analysis.The results show that the fibrinogen concentration and hemoglobin are statistically significant(P<0.05),suggesting that the fibrinogen concentration is a risk factor for rebleeding,and the hemoglobin concentration is a protective factor for rebleeding.3.Constructing a prediction model with two meaningful indicators in multivariate logistic regression analysis,and drawing the ROC curve of the prediction model.The area under the curve AUC is 0.956,95%CI(0.904～1.009)(P<0.05),indicating the model has better predictive ability.4.The AIMS65,GBS and Rockall scores are significantly different between the rebleeding group and the non-bleeding group(P<0.05).The rebleeding group has a higher score than the non-bleeding group.5.The AUC of AIMS65,GBS,and RS scoring systems for evaluating rebleeding are0.705(95% CI 0.555～0.856)、0.800(95% CI 0.691～0.909)、0.759(95%CI 0.630 ～0.889),respectively.The sensitivity of GBS score is the highest,at 0.765.Conclusions: 1.Fibrinogen concentration is a risk factor for rebleeding,and the hemoglobin concentration is a protective factor for rebleeding.The prediction model established according to the logistic regression analysis equation can more accurately predict the risk of rebleeding.2.Among the three scoring systems,the GBS score has the highest predictive value for rebleeding.