423 - A Multicenter Cohort Study of Neonatal Late-Onset Positive Cultures: Center Variation in Treatment

Publication Number: 2411.423

Alvaro Zevallos, Childrens Hospital of Philadelphia, Arlington, VA, United States; Sagori Mukhopadhyay, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Erica C.. Prochaska, Johns Hopkins University School of Medicine, Baltimore, MD, United States; Michael T. Favara, ChristianaCare, Media, PA, United States; Kristin Weimer, Duke University School of Medicine, Durham, NC, United States; Margaret Gilfillan, St. Christopher's Hospital for Children, Philadelphia, PA, United States; Kathryn Ziegler, Sidney Kimmel Medical College at Thomas Jefferson University, Abington, PA, United States; Andrew Berenz, Rush Medical College of Rush University Medical Center, Chicago, IL, United States; Sarah Khan, McMaster University Michael G. DeGroote School of Medicine, Hamilton, ON, Canada; Joern-Hendrik Weitkamp, Vanderbilt University School of Medicine, Nashville, TN, United States; Pablo J. Sanchez, Nationwide Children's Hospital -OSU, Columbus, OH, United States; Katia C. Halabi, Nationwide Children's Hospital, Morristown, NJ, United States; Robert L. Schelonka, Oregon Health & Science University School of Medicine, Portland, OR, United States; Mark R. Conaway, University of Virginia School of Medicine, Charlottesville, VA, United States; David A. Kaufman, University of Virginia School of Medicine, Charlottesville, VA, United States; Dustin D. Flannery, Children's Hospital of Philadelphia, Wynnewood, PA, United States

Background: The microbiology of neonatal late-onset sepsis (LOS) defines disease patterns and informs empiric therapy. Variation in LOS microbiology across centers may reflect differences in patient risk, infection prevention, or contamination classification practices. Limited data describe how positive cultures are categorized as infection versus contamination in NICUs, hindering understanding of center-level variation.

Objective: To describe center-level variation in microbiology and treatment classification of late-onset positive blood, CSF, urine, and respiratory cultures across 12 NICUs.

Design/Methods: Multicenter cohort of all infants admitted to 12 NICUs in the United States and Canada from March 2018 to July 2022. All positive cultures >72 hours after birth were included. Cultures were classified as treated if the infant received ≥5 days of antibiotics, died, or transferred before course completion; cultures not treated and not associated with death or transfer were classified as contaminated. Multiple positive cultures were counted as new when a new organism was identified or a repeat organism occurred >10 days later. Organisms were grouped as coagulase-negative staphylococci (CoNS), other Gram-positive, Gram-negative, fungal, polymicrobial, or unidentified. Pathogen group was the primary variable for assessing center-level variation.

Results: Among 46,516 infants contributing 1,127,320 patient-days, 3,292 positive cultures were identified among 2,468 infants (Table). Of these, 2,356 (71.8%) were treated and 927 (28.2%) contaminated. The proportion treated varied by source-highest for blood/CSF (85%) and lowest for urine (61%)-and by site (33-100%). Among treated cultures, CoNS and Gram-negatives were the leading groups. CoNS accounted for 40% of treated blood/CSF cultures, varying from 13-73% across centers, whereas Gram-negatives comprised 25% overall (8-47%). There was wide inter-center heterogeneity in both microbiologic composition and infection incidence per 1,000 patient-days, with up to four-fold differences between centers (Figure 1). Among contaminants, CoNS predominated, but the relative contribution of each group differed by site and source, suggesting variation in classification or culture practices (Figure 2).

Conclusion(s): Across 12 NICUs, the proportion of late-onset positive cultures treated as infection versus contamination and the microbiologic composition of both groups varied markedly by site. These findings suggest that local epidemiology and classification practices contribute to observed variation and represent potential targets for improving diagnostic stewardship in NICUs.

Table: Characteristics of Late-Onset Positive Cultures and Infection Classification Across 12 NICUs


Figure 1: Center-Level Composition and Incidence of Treated Late-Onset Infections Across 12 NICUs
Legend: Stacked bars show proportional distribution of pathogen groups among treated cultures for each center, stratified by source (blood/CSF, urine, respiratory). Circles denote site-specific incidence rates (per 1,000 patient-days).

Figure 2: Center-Level Composition and Incidence of Contaminated Late-Onset Positive Cultures Across 12 NICUs
Legend: Stacked bars show proportional distribution of pathogen groups among contaminants by center and source. Circles denote contamination rates (per 1,000 patient-days).