The High Mortality of Gastrointestinal Bleeding in HIV-Seropositive Patients: A Multivariate Analysis of Risk Factors and Warning Signs of Mortality in 50 Consecutive Patients

The patterns of which human immunodeficiency virus (HIV)-seropositive patients underwent endoscopy for gastrointestinal bleeding at a university hospital were analyzed in 50 consecutive patients admitted from July 1984 through December 1989, and criteria were developed as to which patients are most likely to benefit from endoscopy. Analyzed patient data included the medical records, follow-up until July 1990 obtained by telephone questionnaire in 46 patients, and autopsy findings in the 11 patients undergoing autopsy. Thirty-seven percent of the patients did not undergo endoscopic or radiographic examinations indicated to determine the cause of bleeding. The adequacy of the evaluation was not related to race, intravenous drug abuse, homosexuality, hemophilia, the diagnosis of known AIDS, or being a public patient. In 21 of the 28 cases in which the cause of bleeding was determined, the diagnosed lesions had a specific, effective therapy. The mortality from gastrointestinal bleeding was 39.0%, compared with 8.3% in 48 controls without known HIV infection (p less than 0.001 by Fisher's exact test, odds ratio = 7.0, odds ratio confidence interval = 5.0-9.7). Statistically significant independent predictors of mortality included leukocytosis, concurrent major diseases, intravenous drug abuse, transfusion of 5 or more units of packed erythrocytes, and the presence of a bloody nasogastric aspirate or hematemesis (Wilk's lambda statistic = 0.369, p less than 0.0001). In particular, 10 of 11 patients (89%) with two or more concurrent major diseases died, whereas only three of 24 patients (13%) with no concurrent diseases died during the hospitalization. We conclude a large percentage of HIV-seropositive patients did not undergo a diagnostic evaluation for gastrointestinal bleeding at a university hospital, and there was no discernible rational pattern as to which patients underwent endoscopy. Endoscopy is an important and indicated procedure in HIV-seropositive patients with no or one concomitant major illness. HIV-seropositive patients with gastrointestinal bleeding and two or more concomitant major illnesses have an exceedingly poor prognosis, and are less likely to benefit from invasive diagnostic tests and aggressive therapy.