Fellow Yale School of Medicine New Haven, Connecticut, United States
Background: Adult literature has found repeat blood cultures (RC) in cases of bloodstream infection (BSI) offer little utility in most cases unless specific organisms or patient characteristics are present. There is much less data in pediatrics. Objective: Our objective is to determine patient and microbial variables associated with persistent BSI, defined as growth of organism more than 48 hours after an index culture. Design/Methods: We conducted a cross-sectional study of children < 18 years within our health system from August 1, 2016 through December 31, 2021 who had a positive blood culture. We excluded children admitted to the neonatal intensive care unit. Patient data included the age, location of the blood draw, presence of a central venous line (CVL), immune status, structural cardiac disease, levels of inflammatory markers, and evidence of concomitant infection. Organism data included the name and antimicrobial resistance pattern. Our primary outcome was persistent BSI. Bivariate analysis was conducted and multivariable logistic regression analysis is in progress. Results: Index blood cultures from 414 children grew 570 organisms. Of these, 506 (88.8%) had a RC drawn >48 hours after the index culture and 68 (11.9%) were positive (p < 0.001). Index cultures drawn in the oncology setting were associated with the highest frequency of persistent BSI at 16/79 (20.3%) while initial blood cultures drawn in the emergency department setting were less likely to have persistent BSI (26/331, 7.9%) (p < 0.003). Organisms with greater than 20% persistence included Candida sp. (5/7, 71.4%), S. aureus (28/85, 32.9%), P. aeruginosa (4/17, 23.5%), and other non-fermenting Gram-negative rods (3/12, 25%). Beta-hemolytic streptococcal species (n=24) and Pneumococcus (n=21) had no positive RCs (p < 0.001). CVLs were associated with persistent BSI (40/251, 15.9%) compared with no CVL (28/319, 8.8%) (p < 0.001). Age, immune status, structural cardiac disease, concomitant infection, and inflammatory markers were not associated with persistent BSI.
Conclusion(s): Children with Candida sp., S.aureus, and non-fermenting Gram-negative rods including P. aeruginosa, were more likely to have persistent BSI. There were no persistent BSIs for Pneumococcus or beta-hemolytic Streptococcal species. The incidence of persistent BSI is higher in cultures drawn on the oncology floor and lower for those from the emergency department. There is no evidence that age significantly contributes to the frequency of persistent BSI. These data suggest the type of organism and presence of CVL are important in determining the need for RC.
