We prospectively studied 5 1 consecutive bacteremic patients with chronic liver disease in order to evaluate their clinical presentation and to assess the relationship of various clinical parameters to mortality. Forty-two patients had alcoholic liver disease and 40 were in Class C, by the Pugh modification of Child's criteria.
Soft tissue infections were the most common source of bacteremia, followed by pneumonia, spontaneous bacterial peritonitis and urinary tract infection. Gram positive organisms were isolated in 69% of cases, and Gram negative ones in 31%. In nine patients, no source of bacteremia was detected. Leukocytosis occurred in 59% of patients and bandemia in only 41%.
Although appropriate antibiotic therapy was begun in all cases on admission, 17 patients (33%) died in the hospital. Of 38 clinical parameters evaluated, multivariate analysis revealed that the three variables contributing the most independent information toward predicting in-hospital mortality were the absence of a history of fever, an elevated serum creatinine and marked leukocytosis. Improved understanding of the pathophysiologic relationship between these parameters and patient outcome may enable us to improve the therapy of bacteremic patients with chronic liver disease.
The predisposition of patients with serious liver disease to bacteremia is well documented (1-8). However, the clinical features and course of bacteremic patients with hepatocellular disease has not been prospectively evaluated. We therefore undertook such a study with two goals: (i) to evaluate the clinical presentation and hospital course of bacteremic individuals with chronic liver disease, and (ii) to assess the utility of various parameters in predicting mortality.
Methods
When a set of blood cultures is drawn in our hospital, 20 ml of blood is divided into two evacuated bottles. One contains 90 ml of Trypticase soy broth with yeast (Calscott, Carson, Calif.), used for aerobic culture.
Microbiologic Methods.