Gastrointestinal bleeding in the hospitalized patient: a case-control study to assess risk factors, causes, and outcome.

PURPOSE: To determine the risk factors, etiology, and outcome of clinically important gastrointestinal bleeding that occurs after hospital admission (nosocomial gastrointestinal bleeding). PATIENTS AND METHODS: Cases consisted of consecutive patients who developed gastrointestinal bleeding more than 24 hours after admission to the hospital. Cases were compared with two control populations: a set of hospitalized patients without gastrointestinal bleeding matched with cases for age, gender, and length of stay; and all patients admitted to the hospital with clinically important gastroduodenal ulcer bleeding during the study period. Case and controls were compared with respect to risk factors for gastrointestinal bleeding and outcomes. Data were obtained through a comprehensive review of medical records.
RESULTS: Clinically important nosocomial gastrointestinal bleeding occurred in 67 inpatients after a mean hospital length of stay of 14 +/- 10 days. The majority (64%) of the patients were not hospitalized in the intensive care unit at the onset of the bleeding. Seventy-two percent of the patients who developed bleeding had been receiving some form of bleeding prophylaxis. In a multivariate analysis, a prior intensive care unit stay (odds ratio 2.5; 95% confidence interval 1.0 to 6.1; P <0.05) and mechanical ventilation (OR 3.4; 95% CI 1.1 to 10.7; P = 0.03) were independent risk factors for the onset of bleeding. Nosocomial gastrointestinal bleeding was associated with poor outcome, with an associated mortality of 34%. Duodenal ulcer disease was the most common source of nosocomial gastrointestinal bleeding, accounting for 36% of cases overall. Nosocomial ulcer bleeders were less likely to have a previous history of ulcer disease (13% versus 50%; P <0.05) Helicobacter pylori infection (14% versus 62%; P <0.0001), chronic active gastritis (29% versus 91%; P <0.0001), or to be taking NSAIDs (48% versus 68%; P <0.08) than patients admitted to the hospital with ulcer bleeding.
CONCLUSIONS: Gastrointestinal bleeding remains an important complication of hospitalization, with a high associated mortality. Our current approaches to prevention of this complication are imperfect. Bleeding tends to occur after a prolonged hospital stay and is more likely to occur in patients with more severe underlying illnesses. Duodenal ulcer disease is the most common source of this bleeding. Nosocomial gastroduodenal ulcer disease is distinct in etiology from the ulcer disease that occurs in outpatients.