ABSTRACT
BACKGROUND & AIMS. For upper gastrointestinal bleeding (UGIB), guidelines recommend pharmacological treatment before endoscopy. Therefore, it is important to establish an early diagnosis of the variceal or non-variceal source of bleeding. This study aims to analyze the clinical and laboratory parameters which are predictors of the UGIB etiology, and to develop a score for predicting variceal or non-variceal bleeding. METHODS. This study comprised patients presenting to the emergency department of a tertiary care center with UGIB, throughout a 1-year period. Clinical, ultrasound data and laboratory parameters were noted. RESULTS. Of the 517 patients with UGIB, 29.8% had variceal and 70.2% non-variceal bleeding. Six factors were associated with variceal hemorrhage: cirrhosis (OR=10.74, 95% CI: 3.50-32.94, p<0.001), history of variceal hemorrhage (OR=13.11, 95%CI: 3.09-55.57, p<0.001), ascites (OR=4.41, 95% CI: 1.74-11.16, p=0.002), thrombocytopenia (OR=2.77, 95% CI: 1.18-6.50, p=0.01), elevated INR (OR=4.77, 95% CI:1.47-15.42, p=0.009) and elevated bilirubin levels (OR=2.43, 95% CI:1.01-5.84, p=0.04). Two factors were associated with non-variceal bleeding: the use of NSAIDs (OR=0.32, 95%CI: 0.13-0.83, p=0.01) and of anticoagulants (OR=0.04, 95%CI: 0.00-0.89, p=0.04). A prediction score for UGIB etiology was designed based on this model. We calculated a cutoff value of 0.968, higher values being predictive of variceal bleeding. Positive predictive value (PPV) and negative predictive value (NPV) were: 82.7% and 97%, respectively. The score was validated prospectively in another group of 162 patients: PPV and NPV were 72.7% and 95.3%, respectively. CONCLUSIONS. Several factors were identified as predictors for the etiology of UGIB. Due to its high PPV and NPV, our UGIB etiology score might be useful in predicting variceal bleeding and could assist in the selection of pharmacological therapy before endoscopy.
