Prevalence, Risk Factors, and Outcomes of Bacteremic Pneumonia in Children

Abstract

Background: Previous studies examining bacteremia in hospitalized children with pneumonia are limited by incomplete culture data. We sought to determine characteristics of children with bacteremic pneumonia using data from a large prospective study with systematic blood culturing.

Methods: Children <18 years hospitalized with pneumonia and enrolled in the multicenter Etiology of Pneumonia in the Community study between January 2010 and June 2012 were eligible. Bivariate comparisons were used to identify factors associated with bacteremia. Associations between bacteremia and clinical outcomes were assessed by using Cox proportional hazards regression for length of stay and logistic regression for ICU admission and invasive mechanical ventilation or shock.

Results: Blood cultures were obtained in 2143 (91%) of 2358 children; 46 (2.2%) had bacteremia. The most common pathogens were Streptococcus pneumoniae (n = 23, 50%), Staphylococcus aureus (n = 6, 13%), and Streptococcus pyogenes (n = 4, 9%). Characteristics associated with bacteremia included male sex, parapneumonic effusion, lack of chest indrawing or wheezing, and no previous receipt of antibiotics. Children with bacteremia had longer lengths of stay (median: 5.8 vs 2.8 days; adjusted hazard ratio: 0.79 [0.73-0.86]) and increased odds of ICU admission (43% vs 21%; adjusted odds ratio: 5.21 [3.82-6.84]) and invasive mechanical ventilation or shock (30% vs 8%; adjusted odds ratio: 5.28 [2.41-11.57]).

Conclusions: Bacteremia was uncommonly detected in this large multicenter cohort of children hospitalized with community-acquired pneumonia but was associated with severe disease. S pneumoniae was detected most often. Blood culture was of low yield in general but may have greater use in those with parapneumonic effusion and ICU admission.

FIGURE 1

Hospital LOS for children hospitalized with pneumonia with and without bacteremia. Time to discharge was modeled by using Cox proportional hazards regression while controlling for demographic variables, number of comorbidities, household smoke exposure, any antibiotics before blood culture, temperature, PF ratio, altered mental status, chest indrawing, asymmetric breath sounds, wheezing, white blood cell count, infiltrate pattern, parapneumonic effusion, and ICU admission.