Predictors of in-hospital mortality in patients with non-variceal upper gastrointestinal bleeding

OBJECTIVE: to determine the independent predictors of in-hospital death of Hispanic patients with nonvariceal upper gastrointestinal bleeding (NVUGB). EXPERIMENTAL DESIGN: prospective and observational trial. PATIENTS: in a period between 2000 and 2009, all patients with NVUGB admitted to our hospital were studied. Demographical and clinical characteristics, endoscopic findings and laboratory tests were evaluated χ² and Mann-Whitney U analyses were per-formed for comparisons, and binary logistic regression was employed to identify independent predictors of in-hospital mortality. RESULTS: 1,067 patients were included, 65% male with a mean age of 58.8 years. Mean number of comorbidities per patient was 1.6 ± 0.76. The most frequent cause of bleeding were gastric and duodenal ulcers (55.4%); 278 patients (25.8%) received endoscopic treatment of which 69.1% had combined therapy. Rebleeding occurred in 36 patients (3.4%) of which 50% died. In-hospital mortality was 10.2%, of which only 3.1% was associated to bleeding. When comparing causes of death among patients with and without comorbidities, only hypovolemic shock was found significative (48.3 vs. 25%; p = 0.020). Binary logistic regression found that the number of comorbidities, Rockall scale score; serum albumin < 2.6 g/dL on admission; rebleeding and length of hospital stay were independent risk factors of in-hospital mortality. CONCLUSION: the number of comorbidities, the Rockall scales core, an albumin level < 2.6 g/dL, the presence of rebleeding and hospital stay were predictors of in-hospital mortality in patients with NVUGB.