Objective: The goal of this study was to determine the prevalence of bacteremia in pediatric patients with community-acquired pneumonia (CAP) at our institution and to test the effectiveness of newly developed guidelines for obtaining blood cultures.
Methods: Using recent literature and local expert opinion, institutional guidelines for obtaining blood cultures in pediatric patients with CAP were developed. A retrospective chart review of children treated in the emergency department or admitted for CAP from January 2010 through June 2011 was conducted. Demographic and clinical data were collected, including results of blood cultures. χ2 tests assessed for variables associated with bacteremia, whether a blood culture was obtained, and if the decision to obtain a culture was appropriate based on our guidelines.
Results: The study included 330 patients; 155 (47%) blood cultures were obtained in our patient population. Five cultures were true-positive findings, making the prevalence of bacteremia 3.2% in patients with blood cultures and 1.5% in all patients. All 5 true positive results met criteria for blood culture based on our guidelines. Applying our guidelines retrospectively, the decision to obtain a blood culture met criteria in 55% of the cases. Bivariate analysis showed that patients discharged from the emergency department had higher rates of guideline-appropriate decisions than patients admitted. Radiographic findings were associated with making a guideline-appropriate decision regarding blood culture.
Conclusions: Instituting local guidelines that limit the frequency of obtaining blood cultures in pediatric patients with CAP is likely to capture any patient with bacteremia. This study suggests that blood cultures may not need to be routinely obtained in all patients admitted to the hospital with CAP.
Pneumonia in pediatric patients is a significant illness in the United States and worldwide. After the widespread use of the Haemophilus influenzae type B vaccine and pneumococcal conjugate vaccine, the prevalence of bacteremia in pediatric patients with pneumonia is reported to be 1% to 3%.1–5 Nevertheless, blood cultures are obtained in up to 33% of these patients,4 and false-positive rates range from 0.8% to 7.3%.6–9 The utility of positive blood culture results, especially in pediatric patients, remains unclear. Several adult studies have found that positive culture results rarely alter clinical management. Some pediatric literature supports the limited clinical utility of positive blood culture results,1,10 whereas another study saw a change in management in >80% of patients with a positive result on blood culture.4 By obtaining blood cultures in only high-risk patients and certain clinical scenarios, the added costs and length of hospitalization associated with false-positive findings could be reduced. Adult studies have estimated that a false-positive blood culture result adds ∼$6000 to hospitalization costs and 4 to 8 days to the length of stay.11,12 One pediatric retrospective chart review analyzed 2971 patients aged <3 years who had a documented fever and a blood culture obtained in an emergency department (ED) setting. They estimated a net cost of >$34 000 to reevaluate children managed as outpatients who had a false-positive result on blood culture.13 Furthermore, false-positive blood culture results lead to broadened antibiotic coverage, often with vancomycin, subsequently contributing to increased antibiotic resistance.14
In 2011, the Infectious Diseases Society of America (IDSA) published guidelines for the clinical management of community-acquired pneumonia (CAP) in both inpatient and outpatient settings.15 In terms of blood cultures in the outpatient setting, the IDSA states that blood cultures should not be routinely obtained in nontoxic, fully immunized children. However, the IDSA guidelines recommend that blood cultures should be obtained in children who fail to demonstrate improvement after the start of appropriate antibiotic therapy or who require hospitalization for moderate to severe CAP, especially complicated pneumonia.
The purpose of the current study was to determine the prevalence of bacteremia in patients with CAP at our institution and to test the effectiveness of newly developed guidelines for obtaining blood cultures in pediatric patients with CAP through retrospective chart review before institutional guideline adoption.
This study was a collaborative effort among many specialties at the Medical University of South Carolina (MUSC), including pediatric emergency medicine, hospitalists, and infectious diseases. The local guidelines for obtaining blood cultures in pediatric patients with CAP were developed by using recent literature, as well as local expert opinion, and included: (1) febrile and age <6 months or febrile with delayed immunizations; (2) patients who are immunocompromised; (3) chronic medical conditions predisposing to severe or recurrent pneumonia; (4) hospitalization <14 days before the diagnosis of pneumonia; (5) toxic-appearing or requiring PICU admission; (6) radiographic concern for an effusion, empyema, or abscess; and (7) central line in place. After development of these guidelines, a retrospective chart review was conducted on the children discharged from the ED or hospital from January 2010 through June 2011 to determine bacteremia prevalence and to test the clinical effectiveness of the guidelines before institutional adoption.
Demographic and clinical data were collected on children discharged from the pediatric ED or the Children’s Hospital at MUSC during the study period. International Classification of Diseases, Ninth Revision codes (480–486, 510–511, and 513) were used to identify patients who had a primary diagnosis of pneumonia. Relevant medical history, immunization status, initial physical examination findings (including general appearance, respiratory rate, and oxygen saturation), white blood cell (WBC) count, chest radiograph findings, and blood culture results were recorded. For chest radiographs, the finalized reads by the attending radiologists were documented when available. For blood cultures, pathogenic bacteria were defined as Streptococcus pneumoniae, Staphylococcus aureus, group A β-hemolytic streptococci, and Haemophilus influenzae; contaminants were defined as coagulase-negative Staphylococcus species, α-hemolytic streptococci, Micrococcus species, and Corynebacterium species.4 The decision to obtain blood cultures was documented as either appropriate or inappropriate based on our guidelines. Bivariate analysis was used to determine what clinical characteristics were associated with obtaining blood cultures, making guideline-appropriate decisions, and bacteremia. The institutional review board at MUSC approved this study.
In total, there were 330 patients in the study, with a mean age of 4.9 years (range: 0–18 years). General demographic and clinical data are given in Table 1. Regarding disposition, 40% were discharged from the ED, 27% were admitted from the pediatric ED, and 33% were admitted from an outside facility or primary care provider. A focal infiltrate was seen on 70.5% of the chest radiographs, and evidence of an effusion or empyema was documented for 9% of the radiographs. In our patient population, 71 patients (21.5%) had a WBC count >15 000 ×103/mL.
A total of 155 (47%) blood cultures were obtained in our patient population. Of the local guidelines, the most common criteria these patients met were radiographic evidence of an effusion, empyema, or abscess, and toxic-appearing or PICU admission (Table 2). Of the blood cultures obtained, 55% met criteria and 45% did not meet criteria based on our guidelines. A total of 145 (94%) blood cultures had negative results, and 10 cultures had positive results. Five cultures were true-positive findings, making the prevalence of bacteremia 3.2% in patients with blood cultures and 1.5% in all patients. However, 50% of positive blood culture results were found to be contaminants. Of the patients with an effusion or empyema, 46% underwent pleural fluid analysis. Pleural fluid culture findings were positive in 3 patients (23%), of whom 2 also had a positive blood culture result (Table 3).
There was no bounce-back admission in which bacteremia was subsequently identified. One patient was initially discharged from the pediatric ED on oral antibiotics without a blood culture obtained and returned with worsening symptoms. Repeat chest radiograph demonstrated a new effusion, but the blood culture result was negative. In fact, all patients with bacteremia were found to have a positive blood culture result during their initial management before any treatment was initiated.
Of the 5 cases of bacteremia, there were 3 cases of S pneumoniae, 1 case of E coli, and 1 case of group A Streptococcus (Table 3) infection. Not only were all 5 patients treated as inpatients from their initial presentation, but all patients with bacteremia also had radiographic evidence of an effusion or empyema. In addition, all 5 patients were admitted to the PICU and/or described as being septic, thereby meeting institutional criteria. However, none of the patients with bacteremia had sickle cell disease or other chronic medical conditions, were immunocompromised, were recently hospitalized, or had a central line in place. All patients who had bacteremia were initially treated with vancomycin, but this decision was likely based on radiographic findings and severity of illness because the blood culture result had not yet been determined to be positive. However, only 1 patient with a false-positive blood culture result received vancomycin but was then discharged on cefdinir after being hospitalized for only 2 days.
Bivariate analysis revealed statistically significant associations between patient age, disposition, chest radiograph findings, and whether a blood culture was obtained. Patients discharged from the ED were less likely to have a blood culture obtained, and patients with an effusion or empyema were more likely to have a blood culture obtained. Disposition and chest radiograph findings were also significantly associated with whether a guideline-appropriate decision was made. Leukocytosis, defined as a WBC count >15 000 ×103/mL, was significantly associated with bacteremia (Table 4).
In this study, the prevalence of bacteremia in patients with CAP was 3.2%, which is similar to results of recently published studies.1–5 All 5 cases of bacteremia had radiographic evidence of an effusion or empyema and therefore warranted a blood culture based on our institutional guidelines. Four patients were admitted to the PICU and described as toxic-appearing; another patient was described as being toxic without requiring PICU admission. Although all 5 patients were treated as inpatients and had similar radiographic findings, there was no other unifying criterion that was met by all patients.
Despite the low prevalence of bacteremia, blood cultures are routinely obtained as part of the initial evaluation of patients admitted for CAP. The 2011 IDSA guidelines15 do not recommend routine use of blood cultures in fully immunized, otherwise healthy children diagnosed with CAP but do recommend blood cultures in patients admitted to the hospital with moderate to severe CAP. We propose that not all patients admitted for CAP are “moderately to severely ill” and therefore do not warrant routine blood cultures. Specifically, well-appearing patients without radiographic evidence of an effusion or empyema who are admitted primarily for mild hypoxia do not need a blood culture obtained. Furthermore, although our guidelines do not explicitly address patients who have failed to improve with appropriate outpatient therapy, there were zero cases of patients in our study presenting after appropriate outpatient antibiotic therapy who were subsequently found to have bacteremia.
The overall low prevalence of bacteremia in this patient population has been established, yet the clinical utility of positive blood culture results remains unclear. In a retrospective cohort study of 355 immunocompetent adults admitted for CAP, Corbo et al9 found that management was changed by clinical status in 49% of patients and by blood culture results in 5% of patients. Hickey et al1 performed a retrospective chart review of 939 pediatric patients with radiographic evidence of pneumonia; 409 blood cultures were obtained. They found no change in the management of any of the 11 pediatric patients with pneumonia and a true positive blood culture result. In addition, Leonard and Beattie10 studied 1159 blood cultures in nearly 16 000 patients evaluated at a pediatric accident and ED over 7 months. Of 26 true positive results on blood culture, only 5 significantly influenced clinical management. Furthermore, the possibility of false-positive results can lead to repeat testing and prolonged hospitalization.11,12 More specifically, Sard et al13 performed an 8-year retrospective chart review of 2971 patients aged <3 years who presented to the ED with a documented fever and had a blood culture obtained. They found that about one-third of children managed as outpatients who had false-positive blood culture results underwent reevaluation, to include an ED visit for repeat blood culture and possibly further testing and parenteral antibiotics. The net cost of reevaluating these patients was estimated to be >$34 000. On the contrary, Shah et al4 found that management was changed in 5 of 6 patients with a true positive result on blood culture. However, the argument could be made that appropriate decisions regarding antibiotics could have been determined based on clinical status alone. In our study, most patients were started empirically on ceftriaxone and discharged from the hospital on amoxicillin or amoxicillin-clavulanate, and most patients either did not have a blood culture performed or had a negative blood culture result. Of the 5 patients with bacteremia, changes in antibiotics were based on blood culture results in 2 patients and on pleural fluid culture results in 2 patients. The rationale to change antibiotics was unclear in the final patient with bacteremia based on our retrospective chart review.
This study has several limitations. First, because blood cultures were not obtained for all patients, the true prevalence of bacteremia in this patient population could not be determined. However, no patients were found to have bacteremia on subsequent evaluations and hospitalizations. Although the contamination rate was elevated at 3.2% in our study, 50% of positive blood culture results were contaminants. This study was performed only on patients at our institution over an 18-month span. Therefore, these results may not be generalizable to other hospitals and institutions. In addition, we had no way of knowing if patients discharged from the pediatric ED were subsequently admitted to another hospital.
This study suggests that instituting local guidelines to limit the frequency of blood cultures in patients with pneumonia will likely capture all patients with clinically significant bacteremia. This study supports the recent recommendations that blood cultures do not need to be routinely obtained in all patients with CAP. For future quality improvement efforts surrounding guideline implementation, retrospective chart review can be a useful way to test the predicted clinical effectiveness before institutional implementation and may encourage guideline adoption. Future studies to examine the clinical utility of positive blood culture results could strengthen the argument against routine blood cultures. In addition, a cost-effectiveness analysis of the financial effect of reducing the number of blood cultures, including decreased length of stay and decreased laboratory costs, could be helpful.
FINANCIAL DISCLOSURE: The authors have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
- community-acquired pneumonia
- emergency department
- Infectious Diseases Society of America
- Medical University of South Carolina
- white blood cell
- Ramsey BW,
- Marcuse EK,
- Foy HM,
- et al
- Corbo J,
- Friedman B,
- Bijur P,
- Gallagher EJ
- Leonard P,
- Beattie TF
- Souvenir D,
- Anderson DE Jr.,
- Palpant S,
- et al
- Bradley JS,
- Byington CL,
- Shah SS,
- et al
- Copyright © 2013 by the American Academy of Pediatrics