Simple Risk Factors To Predict Urgent Endoscopy In Non-variceal Upper Gastrointestinal Bleeding

Objectives: The goal of this study is to evaluate how to predict high-risk nonvariceal upper gastrointestinal bleeding(NVUGIB) for urgent endoscopic examination and hemostasis pre-endoscopically.Methods: A total of 569 NVUGIB patients between Match 2011 and January 2015 were retrospectively studied. The clinical characteristics and laboratory data were statistically analyzed. The severity of NVUGIB was based on high-risk NVUGIB(Forrest I-IIb), and low-risk NVUGIB(Forrest IIc and III). By univariate analysis 、 logistic regression and receiver operating characteristic(ROC) curve, simple risk score systems were derived which predicted patients’ risks of potentially needing endoscopic intervention to control bleeding.Results: By single factor analysis and Binary Logistic regression analysis,our found that red blood, red blood, shock, hemoglobin and urea nitrogen can predict high-risk NVUGIB risk factors. In order to determine the cutoff point of hemoglobin, urea nitrogen,using ROC curve analysis, the results showed that the Hb≤ 75 g/L, BUN≥8.5 mmol/L can predict high-risk NVUGIB.Thus, the red hematemesis, red stool, shock, Hb≤ 75 g/L,BUN≥8.5 mmol/L was choose to be the risk factor of our new Risk score systems. At last,we choose these five factors( Hb≤75 g/L, red hematemesis, red stool, shock, BUN≥8.5mmol/L) as the risk factors to predict high risk NVUGIB.In order to assess the value of these index, we use following indication(1)Hb≤75g/L,(2)red hematemesis,(3)red stool,(4)shock,(5)BUN≥8.5 mmol/L to draw the ROCcurve, the ROC curve analysis outcome as follow:(1)、(2)、(3)、(4)、(5) the sensitivity of47.5%, 50.8%, 14.2%, 10.9%, 14.2%, specificity of 76.7%, 85.5%, 96.4%, 97.9%, 49.6%( table 4.2). Because the fomer index(1)、(2)、(3)、(4) has high specificity, low sensitivityto predict high-risk patients easily but it’s easy to lead misdiagnosis. While the last index(5) has low sensitivity and high specificity, so it’s easy to lead missed diagnosis. Thus wecannot use the single factor to predict the high risk NVUGIB.In order to predict high risk NVUGIB well before the endoscopy, we combine these indications differently. Firstly, we define the color of hematemesis and melena was scored 1 when it was red hematemesis and stool(dark red, bright red or containing red clot), and scored 0 when presented as other presentations. Then we combined(1)(2)(3),(1)(2)(5),(1)(2)(3)(5),(1)(2)(3)(4)(5) as Score 1, Score2, Score 3, Score4,respectively.Above four scoring system and Blatchford risk score system for ROC curve,which found four risk prediction score system sensitivity and specificity were higher than Blatchford risk score, so the four scoring system can well distinguish between high and low risk NVUGIB patients.In these four systems, The highest area of ROC curves was 0.746, with sensitivity0.675, specificity 0.733 in score 4 when the cut-off point was 1.5. And the score 1 is the lowest of these system, its area under the ROC curve、sensitivity and specificity was 0.725,0.782, 0.623, respectively. Although the risk predictive Score 1 is poor of others, it contains lest index which is easy to calculate, easy to member and feasible to application in the clinical, so that we choose Score 1 as the risk predictive system to predict high-risk before endoscopy at last.Conclusion:Simple risk score systems can predict high-risk NVUGIB well which need emergency endoscopy or endoscopic intervention to control bleeding. While the practicability should be continue to confirm.