Acute upper-gastrointestinal bleeding (UGIB) is a commonly encountered cause of admission in emergency department (ED). Early risk stratification allows appropriate therapy that may be helpful to advance the patient's morbidity and mortality. Investigators hypothesized that early RDW levels may have an independent, linear relationship with recurrent or massive bleeding in UGIB patients.

This is a retrospective clinical study including the UGI patients in ED. All data will be collected through the electrical medical records. The major endpoints are the high risk patients and 30-days mortality which defined cases of including any of one among the death, re-bleeding, and received the intervention (such as blood transfusion, endoscopic therapy and operation).

Investigators assessed the relationship between initial RDW level and high risk patients and 30-days mortality. In addition, investigators compare the dicrimination power for predicting outcomes between the UGIB scoring system and RDW level.

Univariate and multivariate logistic regression testing and a Cox hazard regression model were used to determine the factors associated with outcome variables. In addition, investigators constructed receiver operating characteristic curves, and the areas under the curves and confidence intervals were calculated to compare the discriminatory power for outcomes.

Inclusion Criteria:

• Age over 18 years
• Diagnosis as Upper gastrointestinal bleeding
• Check RDW when patients visited at emergency department

Exclusion Criteria:

• Paediatrics
• Follow up loss
• Data could not be acquired

Adult patients who visited to emergency department due to Upper gastrointestinal bleeding