Background: Bacteremia is commonplace in patients undergoing hemodialysis since the vascular access site is a ready source of infection. Mortality is notably high. However, uncertainties exist with respect to therapy including indications for surgical removal of vascular access site and duration of therapy. We therefore conducted a large-scale collaborative study of bacteremia in hemodialysis patients in six US academic medical centers to define the epidemiology of such infections and to address issues of management.
Patients and methods: We conducted a prospective observational study over 2 years. Severity of illness at onset of bacteremia was defined by objective criteria. Patients were followed for 90 days to assess late complications including endocarditis and mortality. Univariate and multivariate analyses were used to assess risk factors for mortality.
Results: Patients experiencing 127 consecutive episodes of bacteremia were enrolled. The most common cause of bacteremia was Staphylococcus aureus (31%), followed by aerobic gram-negative bacilli (28%) and coagulase-negative staphylococci (13%). Polymicrobial bacteremia occurred in 6% of patients. The most frequent focus of infection was the access site for hemodialysis, although urinary tract, gastrointestinal tract and lung were also implicated. Aerobic gram-negative bacilli and enterococci usually originated from the urinary tract. S. aureus was significantly more likely to cause infection of the access site than other bacteria (p = 0.0001). S. aureus endocarditis was diagnosed in two patients who were receiving antibiotic therapy for S. aureus bacteremia. Removal of the infected access site (shunt, fistula, catheter) was performed for 86% of the patients (95% of the intravenous catheters and 80% of the arteriovenous fistulas/shunts). Overall mortality was 33% at 90 days and was significantly associated with severity of illness at onset of antibiotic therapy and age >60 years. Mortality was not significantly different in patients undergoing surgical removal of infected access site versus those treated with antibiotics alone.
Conclusion: When S. aureus was isolated from the blood, the access site was the most frequent source. Surgical removal of the access site did not have a notable impact on mortality. Until a randomized trial proves otherwise, it appears that surgical removal of the access site can be individualized. Selected patients who are less severely ill (based on objective criteria) can maintain their hemodialysis access site and be treated with 2 weeks of antibiotic therapy.