Predictors of a variceal source among patients presenting with upper gastrointestinal bleeding

BACKGROUND: Patients with upper gastrointestinal bleeding (UGIB) require an early, tailored approach best guided by knowledge of the bleeding lesion, especially a variceal versus a nonvariceal source. OBJECTIVE: To identify, by investigating a large national registry, variables that would be predictive of a variceal origin of UGIB using clinical parameters before endoscopic evaluation. METHODS: A retrospective study was conducted in 21 Canadian hospitals during the period from January 2004 until the end of May 2005. Consecutive charts for hospitalized patients with a primary or secondary discharge diagnosis of UGIB were reviewed. Data regarding demographics, including historical, physical examination, initial laboratory investigations, endoscopic and pharmacological therapies administered, as well as clinical outcomes, were collected. Multivariable logistic regression modelling was performed to identify clinical predictors of a variceal source of bleeding. RESULTS: The patient population included 2020 patients (mean [+ SD] age 66.3 +/- 16.4 years; 38.4% female). Overall, 215 (10.6%) were found to be bleeding from upper gastrointestinal varices. Among 26 patient characteristics, variables predicting a variceal source of bleeding included history of liver disease (OR 6.36 [95% CI 3.59 to 11.3]), excessive alcohol use (OR 2.28 [95% CI 1.37 to 3.77]), hematemesis (OR 2.65 [95% CI 1.61 to 4.36]), hematochezia (OR 3.02 [95% CI 1.46 to 6.22]) and stigmata of chronic liver disease (OR 2.49 [95% CI 1.46 to 4.25]). Patients treated with antithrombotic therapy were more likely to experience other causes of hemorrhage (OR 0.44 [95% CI 0.35 to 0.78]). CONCLUSION: Presenting historical and physical examination data, and initial laboratory tests carry significant predictive ability in discriminating variceal versus nonvariceal sources of bleeding.