Predictors of mortality in patients admitted to hospital for acute upper gastrointestinal hemorrhage.

Abstract

We wanted to identify features of prognostic significance in patients admitted to hospital because of acute upper gastrointestinal (UGI) hemorrhage. A prospective, longitudinal study of 321 consecutive cases admitted during 1988-91 was carried out. The relative risk of mortality associated with each of the background, laboratory, and endoscopic features and the hospital course was calculated. Multiple stepwise logistic regression was used to define factors independently associated with mortality. Two models were evaluated, the first based on the data at presentation (history, physical findings, initial laboratory data) and the second based on the first, plus the endoscopic and follow-up data. The overall mortality was 7.8%. At presentation the features associated with a significantly (p < 0.05) increased risk of mortality were (adjusted odds ratios in parentheses) age > or = 75 years (11.2), a history of cancer (12.1), blood in the gastric aspirate (9.6), and a systolic blood pressure < or = 90 mm Hg (6.4). The overall predictors of mortality were age > or = 75 years (12.7), blood in the gastric aspirate (18.9), serum creatinine level > or = 150 mumol/l (14.8), increased serum aminotransferase level (20.2), and persistent or recurrent bleeding (57.3). In patients admitted to hospital because of UGI hemorrhage the prognosis depends on age, underlying diseases, hemodynamic status, and the persistence or recurrence of bleeding. The causes of bleeding were not relevant to the prognosis.