The Research Of Non-variceal Upper Gastrointestinal Bleeding Risk Factors And The Evaluation Of Four Scoring Systems In Predicting Clinical Outcomes

BackgroundUpper gastrointestinal bleeding(UGIB)is a common and emergent disease.Although endoscopy and medication treatment have improved,the mortality of UGIB has not decreased significantly.At present,a variety of NVUGIB scoring systems have been reported for different clinical outcomes.However,the prediction ability of studies shows difference.The GBS and FRS are widely used to predict mortality and other outcomes worldwide and recommended for guideline.AIMS65 score and PNED are developed as novel scoring systems in recent decades,and there are only few studies about these scores in China.Therefore,we compare the ability of four scoring systems in predicting clinical outcomes.Meanwhile,we evaluate the risk factors of different clinical outcomes.Objective1.To evaluate the ability of prediction for rebleeding,mortality,clinical intervention and the optimal thresholds of clinical outcomes with GBS?AIMS65?FRS and PNED scoring systems.2.To study the risk factors for different clinical outcomes,making a foundation for the new scoring systems in the future.Method1.The study included 394 NVUGIB patients in Tianjin Medical University General Hospital from January 2015 and December 2016.Date was retrospectively analyzed and calculated using four scoring systems for each patient.The area under the receiver-operating characteristic curve(AUROC)and the cut-off values were calculated for different scoring systems and outcomes including rebleeding,mortality and clinical intervention to assess the efficiency.2.Using logistic regression analysis to identify the risk factors for clinical outcomes.Result1.Peptic ulcer,upper gastrointestinal tract malignancy and the erosive lesions were the main causes of NVUGIB in order,accounting for 60.6%,13.7%,11.2%,respectively.Of the 394 patients,the males were 3.2 times as greater as the females.They were with a mean age of 60.9±14.9 years old and a mean length stay in hospital of 9.6±5.1 days.Overall,9.1% experienced rebleeding and 4.1% died in hospital.And48.0% needed at least one clinical intervention,of whom 41.3% needed blood transfusion,9.1% needed endoscopic hemostasis,6.6% needed surgery and 3.3%needed interventional therapy.2.The mean scores in patients with rebleeding,death or clinical intervention were higher than the others without adverse outcomes.3.The PNED score had the highest discriminative ability compared with GBS,AIMS65 score and FRS at predicting in-hospital mortality(AUROC 0.933 vs 0.809 vs0.813 vs 0.809,p<0.05),and the other three scoring systems were without difference.The ability of predicting rebleeding of GBS and FRS were superior to AIMS65score(AUROC 0.715 vs 0.702 vs 0,597,p<0.05).There were no significant differences in predicting clinical intervention with the GBS,FRS and AIMS65(AUROC 0.656vs0.613 vs 0.620,p>0.05).However,they were not the optimal scores due to the lower AUROC.The GBS score can identify an optimal score for rebleeding,mortality and clinical intervention with more than 7,more than 9 and more than 7.Moreover,a AIMS65 score of more than 0,a FRS score of more than 5 and a PNED score of more than 3 were best at predicting mortality.As to rebleeding,the AIMS65 score of more than 1 and the FRS score of more than 4 were the optimal cut-off values.The AIMS65 score of more than 0 and a FRS score of more than 3 were the optimal cut-off values at predicting clinical intervention..4.Hemoglobin,albumin,PTINR and blood urea were associated with rebleeding.The independent predictors for rebleeding were PTINR,hemoglobin and albumin.Rebleeding,red blood cells transfusion,the age more than 65,hemoglobin,albumin,PTINR and blood urea were associated with NVUGIB mortality.Among them,PTINR and blood urea were the independent risk factors for predicting mortality.The age more than 65,hemoglobin,albumin and blood urea were associated with clinical intervention.Hemoglobin and albumin were the independent risk factors.Conclusion1.Peptic ulcer,upper gastrointestinal tract malignancy and the erosive lesions were still the main causes of NVUGIB in order.2.The PNED score is an available score for predicting mortality,which is superior to GBS,AIMS65 and FRS score.The ability of predicting rebleeding of GBS and FRS were superior to AIMS65 score.GBS,AIMS65 and FRS were not the optimal scores for predicting clinical intervention.And the optimal cut-off values for four scoring systems need further study.3.The independent predictors for rebleeding were PTINR,hemoglobin and albumin.PTINR and blood urea were the independent risk factors for predicting mortality.Hemoglobin and albumin were the independent risk factors for clinical intervention.