Original Contribution ED predictors of upper gastrointestinal tract bleeding in patients without hematemesis B Michael D. Witting MD, MS a,b, * , Laurence Magder PhD a , Alan E. Heins MD b,c , Amal Mattu MD b , Carlos A. Granja MD b , Mona Baumgarten PhD a a Department of Epidemiology and Preventive Medicine, University of Maryland School of Medicine, Baltimore, MD 21201, USA b Division of Emergency Medicine, University of Maryland School of Medicine, Baltimore, MD 21201, USA c Department of Emergency Medicine, University of South Alabama, Mobile, AL 36617, USA Received 19 October 2005; revised 9 November 2005; accepted 9 November 2005 Abstract Objectives: In patients with gastrointestinal (GI) tract bleeding, the bleeding source is uncertain in the absence of hematemesis. We sought to identify clinical variables predictive of an upper GI bleed- ing source. Methods: This retrospective cohort study involved patients admitted via the ED for GI tract bleeding without hematemesis, who underwent confirmatory testing. We used logistic regression analysis to identify clinical variables independently associated with an upper GI source. Results: Among 325 patients, odds ratios for the strongest predictors were as follows: black stool, 16.6 (95% confidence interval [CI], 7.7-35.7); age less than 50 years, 8.4 (95% CI, 3.2-22.1); and blood urea nitrogen/creatinine ratio 30 or greater, 10.0 (95% CI, 4.0-25.6). Seven (5%) of 151 with none of these factors had an upper GI tract bleed, versus 63 (93%) of 68 with 2 or 3 factors. Conclusion: Black stool, age less than 50 years, and blood urea nitrogen/creatinine ratio of 30 or greater independently predict an upper GI tract bleeding source. D 2006 Elsevier Inc. All rights reserved. 1. Introduction Upper gastrointestinal (GI) tract bleeding has an annual incidence of roughly 90 episodes per 100 000 population, and lower GI tract bleeding has an estimated annual inci- dence of 20 per 100 000 [1]. As the population ages, the number of lower GI tract bleeding admissions will in- crease, because hospitalization for lower GI tract bleeding is 200 times more frequent in the elderly than in the young [2]. Emergency physicians make several decisions based on their prediction of the bleeding source—upper versus lower—but the bleeding source, which is obvious in patients with hematemesis, is usually unclear in the absence of hematemesis [3]. The source of bleeding may influence the selection of consultant and the timing of consultation. If a patient has upper GI tract bleeding, a gastroenterol- 0735-6757/$ – see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.ajem.2005.11.005 Results were presented at the Society for Academic Emergency Medicine Annual Meeting; May 2004; Orlando, Fla. B This project was funded by an intramural research award from the University of Maryland School of Medicine. * Corresponding author. Division of Emergency Medicine, University of Maryland School of Medicine, Baltimore, MD 21201, USA. Tel.: +1 443 310 2002; fax: +1 410 545 5167. E-mail address: mwitt001@umaryland.edu (M.D. Witting). American Journal of Emergency Medicine (2006) 24, 280–285 www.elsevier.com/locate/ajem