Prophylactic antibiotics in cirrhotics with upper gastrointestinal hemorrhage: a prospective, controlled trial.

Background: Infections are a frequent complication in cirrhotics, and gastrointestinal bleeding may increase the infection rate. Nonabsorbable antibiotics or quinolone have been employed to decrease the incidence of infection. Since most of these studies were performed in western countries, it is still unclear whether this holds true in our Taiwan cirrhotic patients. Thus we conducted this study using a different formula of antibiotics to evaluate the efficacy of reducing infection rates in cirrhotics with upper gastrointestinal bleeding.

Methods: From July 1999 to August 2000, all cirrhotic in-patients presenting with upper gastrointestinal bleeding but without infection were enrolled. The patients should not have received antibiotics within 2 weeks before admission and should have expected life expectance more than 7 days. Eligible patients who had received endoscopy within 12 hours of hospitalization were randomly allocated into 2 groups. Group A received intravenous cefazolin 1 gm every 8 hours started before endoscopy. After 3 days of prophylactic parenteral antibiotics, antibiotics were shifted to oral cephalexin of 500 mg every 6 hours for 4 days. Group B served as control subjects. All patients received chest X-ray, blood and urine cultures, and ascites culture and sputum culture if ascites and sputum were found. Patients were excluded when initial blood, urine or ascites culture was positive for bacterial growth.

Results: Ninety-seven patients were included. Group A was comprised of 47 patients and Group B comprised of 50 patients. There was no significant difference in age, sex, Child-Pugh's score or initial hemoglobin between the 2 groups. Proved infection developed in 6 patients of Group B. By contrast, no proved infection was found in Group A. Three organisms belonged to gram-negative bacilli and 3 organisms were gram-positive cocci. The incidence of proved infection during hospitalization was 0% in Group A and 12.0% in Group B (p = 0.027). If possible infection cases (patient's body temperature more than 38 degrees C for more than 2 days) were included, the infection rate was 6.4% in Group A and 26% in Group B (p = 0.013). Infection-related mortality occurred in 2 patients in Group B, but none in Group A.

Conclusions: Our prophylactic antibiotic treatment proved safe and effective in reducing the infection rate in cirrhotics with upper gastrointestinal bleeding.