66 - Leukopenia and Neutropenia with and without Procalcitonin as Predictors of Serious Bacterial Infections in Young Febrile Infants

Poster Number: 66
Publication Number: 66.105

Andrew T. Krack, Children's Hospital Colorado, Highlands Ranch, CO, United States; Michelle Eckerle, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States; Prashant Mahajan, University of Michigan Medical School, Ann Arbor, MI, United States; Octavio Ramilo, Nationwide Children's Hospital, Columbus, OH, United States; John M. VanBuren, University of Utah, Salt Lake City, UT, United States; Russell K. Banks, University of Utah, Farmington, UT, United States; Charlie Casper, University of Utah School of Medicine, Draper, UT, United States; David Schnadower, University of Cincinnati College of Medicine, Cincinnati, OH, United States; Nathan Kuppermann, University of California, Davis, School of Medicine, Sacramento, CA, United States

Background: Prediction rules for risk stratification of febrile infants ≤60 days for serious bacterial infections (SBIs) that include serum procalcitonin (PCT) and C-reactive protein (CRP) are highly accurate. However, these tests are not universally available. Prior data on leukopenia and neutropenia as predictors of SBI are incomplete and need further investigation, particularly in well-appearing infants and given increasingly available novel biomarkers such as PCT.  

Objective: We sought to determine the test accuracy of leukopenia and neutropenia in identifying non-critically-ill febrile infants ≤60 days with SBIs with and without the addition of PCT.  

Design/Methods: This was a secondary analysis of a prospective observational cohort of 7407 febrile infants <60 days who presented to one of 26 PECARN site EDs between December 2008- May 2013 and June 2016 -April 2019. Eligibility was defined as age ≤60 days, documented or historical fevers ≥38°C within 24 hours of presentation, and SBI evaluation including a blood culture. We evaluated the risk of SBI in patients with leukopenia (WBC count ≤5000 cells/µL) and neutropenia (ANC ≤1000 cells/µL), respectively, and the incremental impact of PCT values when available. We performed multivariable logistic regression with SBI as the outcome, using clinical and laboratory predictors, stratified by the WBC count and ANC, respectively.

Results: Of the 6865 evaluated infants, 671 (10%) had SBIs, including 150 (2%) with bacteremia and/or bacterial meningitis; PCT was available in 3098. The frequency of SBIs was 4.7% (28/593) in infants with leukopenia, but highest (30%, 6/20) in those with WBC counts of 0-2500 cells/µL (Table 1). Conversely, frequency of SBIs increased as ANC increased (Table 2). In multivariable logistic regression analyses, a WBC count of 0-2500 cells/µL was associated with increased odds of SBIs, both in models that included (n=3064, OR 32.08; 95% CI 6.36-136.64, p< 0.001) and did not include PCT (n=6758, OR 23.77; 95% CI 6.05-74.09, p< 0.001). Notably, however, in all patients with WBC counts ≤5000 cells/µL who had bacteremia and who had PCT performed, PCT was >0.5ng/mL.Conclusion(s): In febrile infants <60 days, leukopenia between 0-2500 cells/µL, but not neutropenia, was associated with SBIs. However, in the cohort in whom PCT was available, all patients with bacteremia and/or bacterial meningitis who had WBC counts ≤5000 cells/µL also had elevated PCT levels. This suggests that leukopenia be considered a risk factor for SBI in well-appearing febrile young infants only when PCT is unavailable.
Table 1. Clinical variable summaries by WBC Count categories (complete cohort, PCT not required)
Table 2. Clinical variable summaries by ANC categories (complete cohort, PCT not required)