Background: The optimal choice of initial antibiotic therapy for patients with high-risk febrile neutropenia (FN) in children is unclear and varies by the institution on the basis of local antibiograms and epidemiology of specific pathogens. The authors evaluated the appropriateness of antibiotics for the empiric treatment of FN in pediatric patients with cancer in our institution on the basis of changes in the epidemiology of organisms isolated from blood cultures (BCx).
Methods: The authors conducted a retrospective medical record review of pediatric patients who received any oncology care (including patients with cancer and patients who had stem cell transplant) at University of Chicago Medicine Comer Children's Hospitals (March 2009 to December 2016) with a diagnosis of FN who had at least 1 BCx obtained. They reviewed pathogens isolated from BCx and determined whether they were pathogens or contaminants using the Infectious Diseases Society of America (IDSA) guidelines and the team's decision to treat. They investigated the microbiologic spectrum and susceptibility patterns of pathogens causing bacteremia in pediatric FN and whether the empiric therapy chosen may have affected clinical outcomes.
Results: A total of 667 FN episodes were identified in 268 patients. BCx were negative in 497 (74.5%) and were determined to be contaminants in 27 (4%). In 143 episodes (21.5%), the BCx were positive for a pathogenic species. Polymicrobial bacteremia was identified in 25 episodes; a total of 176 pathogens were isolated. The majority of pathogens (95/176, 54%) were Gram-positive (GP), whereas 64 of 162 (36%) were Gram-negative (GN), 5 were fungal, and 4 were mycobacterial. The most common GP pathogens were viridans group streptococci (VGS) (n=34, 19.3%), coagulase-negative staphylococci (n=25, 14%), and methicillin-susceptible Staphylococcus aureus (n=12, 6.8%). Of aerobic GN bacilli, 15 (8.5%) were AmpC producers and 3 (1.7%) carried extended-spectrum beta-lactamases. There was no increase in the prevalence of multidrug-resistant GN isolates during the study period. Patients with VGS and multidrug-resistant GN bacteremia were more likely to be admitted to the pediatric intensive care unit [odds ratio (OR), 3.24; P=0.017; and OR, 2.8; P=0.07, respectively]. There were trends toward a higher prevalence of GP pathogens causing bacteremia and the emergence of VGS with decreased penicillin sensitivity. The prevalence of bacteremia with VGS was higher in acute myelogenous leukemia and neuroblastoma (OR, 2.3; P<0.01) than in patients with other solid tumors.
Conclusions: Empiric antibiotic treatment should be tailored to patients' risk for VGS and multidrug-resistant organisms. Individual hospitals should monitor the pathogens causing FN among patients with cancer to guide choice of empiric therapy.