Value of Blood Cultures in the Management of Children Hospitalized with Community-Acquired Pneumonia

Background and objectives Current guidelines for the management of community-acquired pneumonia (CAP) in children recommend obtaining a blood culture for children with moderate to severe pneumonia; yet, there is no guidance to assess the severity of the disease. Thus, a blood culture is obtained for the majority of children admitted with CAP, regardless of the severity of their symptoms. The study was designed to investigate and identify the prevalence of bacteremia in pediatric patients hospitalized with CAP and to evaluate the clinical and laboratory variables associated with bacteremia. Methods We conducted a medical record review of children aged from two months to 18 years diagnosed with CAP between January 1, 2013, and December 31, 2017, at our two urban tertiary centers. We used binary logistic regression analysis and chi-square tests to look at factors associated with blood culture positivity. Results A total of 464 patients were admitted with CAP. Blood cultures were obtained in 357 (76.9%) patients; 23 patients had repeated cultures. Fifteen patients had positive cultures: 5/380 (1.3%) were considered true positive results and 10/380 (2.6%) were considered contaminants. Intensive care unit (ICU) admission (OR 5.6 with 95% CI (1- 31), p<0.03), toxic appearance (OR 12.8 with 95% CI (1.3-125), p<0.01), and significantly elevated C-reactive protein (CRP) (>300 mg/L (p<0.01) were associated with bacteremia. Conclusion The prevalence of bacteremia among children admitted for CAP is low. The use of routine blood cultures should be reserved for children with moderate to severe pneumonia. Further studies are required to better risk-stratify children with CAP.


Introduction
Community-acquired pneumonia (CAP) is one of the main causes of hospitalization of children in the United States, with an annual estimated cost exceeding one billion dollars [1][2]. A recent prospective multicenter study conducted by the Centers for Disease Control and Prevention (CDC) reported that the annual incidence of hospitalization for CAP is 15.7 cases per 10,000 children from January 1, 2010, to June 30, 2012 [3].
Viruses are by far the most common cause of CAP while bacteria account for only 15% of cases [3]. After the introduction of the vaccines against Streptococcus pneumoniae and Haemophilus influenza type B, the rates of CAP secondary to these invasive bacterial infections have decreased from 7.7% to 4% [4][5][6].
National guidelines for the management of infants and children with CAP, established in 2011 by the Infectious Diseases Society of America (IDSA), recommend "only obtaining blood culture in children with moderate and severe CAP" [7]. This statement, as mentioned in the guideline, fails to accurately define what would be considered "moderate to severe CAP." Past studies have reported a wide range in the prevalence of bacteremia in CAP (0.8%-17.4%) [8][9][10][11][12][13].
A meta-analysis published in 2015 on the role of blood culture in pediatric CAP found that studies focusing on patients with severe CAP had a higher prevalence of bacteremia when compared to studies that included non-severe CAP. Validated clinical prediction rules or stratification criteria to define or categorize the severity of pediatric CAP are not available [14]. In adults, a six-point scoring system based on the presence of confusion, uremia, respiratory rate, blood pressure, and age ≥ 65 years (CURB-65 (confusion, urea, respiratory rate, blood pressure, and 65 years of age or older) score) is widely adopted to define severe pneumonia [25][26][27]. Few studies have attempted to describe the correlation of clinical or radiological parameters like toxic appearance or presence of effusion with bacteremia secondary to CAP in children [28].
The main study objective was to identify the prevalence of bacteremia in pediatric patients hospitalized with CAP in two tertiary-level hospitals. Secondary objectives included: (1) to determine the prevalence of false-positive blood cultures, (2) to identify the microbiology and susceptibility patterns of true positive blood cultures, (3) to identify the clinical and laboratory variables associated with bacteremia in patients hospitalized with CAP, and (4) to determine the impact of blood culture results on antimicrobial therapy and length of hospital stay.

Materials And Methods
This is a medical record review of all patients admitted to New York City (NYC) Health and Hospitals/Kings County Hospital (KCH) and State University of New York, Downstate Medical Center (DMC) with a diagnosis of community-acquired pneumonia between January 1, 2013, and December 31, 2017. Both centers are tertiary-level hospitals located in Brooklyn, New York. The Institutional Review Board at KCH and DMC approved the study protocol.
We included children and adolescents aged two months to 18 years of age. Subjects were identified according to International Classification of Diseases, Ninth Revision (ICD-9) (480-488.1, 510-511, 513, Jan 1 2013-Sept 30 2015) and ICD-10 (J10-J18.9, Oct 1 2015-Dec 31 2017) codes for pneumonia. We excluded children and adolescents with a diagnosis of hospital-acquired pneumonia, defined as pneumonia diagnosed more than 48 hours from admission or less than two weeks after hospital discharge, as well as children living in chronic care facilities.
Charts were manually reviewed to obtain demographic, clinical, and laboratory data, which included relevant past medical history (previous history of pneumonia, chronic lung disease, asthma, immunodeficiency, sickle cell disease, cerebral palsy), recent use of antibiotics (defined as any patient who self-reported antibiotics use five days before hospitalization ), vaccination status (including influenza vaccine in the last season ), vital signs, general appearance, respiratory examination findings, relevant laboratory results (white blood cell count, neutrophil count, C-reactive protein), imaging studies (chest X-ray, ultrasound, tomography), blood culture (obtained in the first 48 hours upon admission), as well as results of other microbial testing, including nasopharyngeal aspirate and pleural fluid analysis (cell count, pH, protein, glucose, lactate dehydrogenase (LDH), culture). Complications encountered during the hospital course included admission or transfer to the pediatric intensive care unit, supplemental oxygen requirement, and ventilatory support. Length of stay was also recorded.
The distinction between true-positive vs. contaminated blood cultures was determined by the primary team taking care of the patient and later reviewed by two of the authors (AY and EFM).

Results
We collected data from 464 children. The median age was three years (IQR 2-6) and 51.9% were female. A total of 162 (34.9%) had no significant past medical history and 446 (96%) were upto-date on their immunizations, as shown in Table 1. Table 1 outlines the demographic and clinical characteristics of our population.   Three out of five bacteremic patients were admitted to the pediatric intensive care unit (PICU). Two were admitted directly from the emergency department (ED) to the PICU and were treated aggressively with broad-spectrum antibiotics. One patient was initially treated with ampicillin in the inpatient unit but deteriorated clinically and was transferred to the PICU, where antibiotic treatment was advanced to ceftriaxone and vancomycin.

Demographic and Clinical Characteristics
Two patients who were admitted to the inpatient unit were started on ampicillin and continued to show improvement on the same antibiotic. We found a significant association between PICU admission (OR 5.6 and 95% CI (1.0-31), p<0.03) and ill appearance on presentation (OR 12.8 and 95% CI (1.3-125), p<0.01) with a true-positive blood culture.  Finally, in regard to the radiological and laboratory investigations. Of 19 (4%) patients with evidence of effusion or necrotizing pneumonia on chest radiographs, three out of the 19 patients had positive blood cultures; and out of these three, only one patient underwent pleural fluid tapping. Two additional patients who underwent pleural tapping had negative blood cultures, and none of these three patients who undergone tapping had positive pleural fluid cultures. The radiographic evidence of effusion or necrotizing pneumonia was associated with bacteremia (P<0.11) while the binary logistics regression analysis showed that significantly elevated CRP (> 300 mg/L) was associated with a true-positive blood culture (p<0.01).

Discussion
Bacteremia is an unusual complication of CAP in hospitalized children. Our study suggests that approximately 1% of admitted children with CAP are bacteremic. Penicillin-susceptible S. pneumoniae was the most common isolated organism (60%), which is consistent with recently published studies [15]. Patients with bacteremia mostly had co-morbid conditions, were ill-appearing, or were admitted to the PICU. Our findings are consistent with current IDSA guidelines, which suggest that only children with moderate to severe CAP will benefit from obtaining a blood culture on admission.
Unlike some reports, we did not find a correlation between the radiological findings of effusion or necrotizing pneumonia and a truly positive blood culture. Heine  The impact in the management of the blood culture results was limited in our study. Two patients were admitted directly to the ICU while a third one was transferred from the inpatient unit after clinical deterioration. The results from the blood culture allowed for accurate diagnosis in all of them and appropriate antibiotic de-escalation at least in two of the patients.
For the two other patients admitted to the inpatient unit, these results led to a therapeutic dilemma with subsequent prolongation of hospital stay for both of them and unnecessary broadening of antibiotics for one of them, even though the reported organism was penicillinsusceptible S. pneumoniae.
The prevalence of blood culture contaminants in this study was 2.6%, which is similar to prior reports [28]. Furthermore, a positive blood culture was twice more likely to be a contaminant than a true pathogen. Despite the high likelihood of reported growth in blood culture to be a contaminant, determining the possibility of true bacteremia is still challenging for physicians and impacts negatively on patient care. In this study, five of the 10 patients with false-positive cultures had repeated cultures, even though the treating physician's impression was in favor of a contaminant organism.
This study has several limitations. Our research is a retrospective chart review that is more prone to chart abstraction biases, especially personal data such as ill appearance. This study was done in our two urban academic medical centers and its results may not be generalizable to other institutions. Finally, we do not have data on whether children discharged from our two facilities may have been readmitted to other hospitals.

Conclusions
The rate of positive blood cultures in children admitted with CAP is low, and its impact on clinical management is limited. Children that were ill-appearing, admitted to the intensive care unit, or with significantly elevated CRP levels were more likely to have bacteremia secondary to CAP. The current recommendation of obtaining a blood culture for children with moderate to severe CAP, although correct, needs to be clarified. Further studies are needed to better riskstratify children with CAP, thus providing a targeted approach for obtaining blood cultures,