Bloodstream infections (BSIs) are a major cause of morbidity and mortality in children.1–5 Severe infections were responsible for a quarter of the deaths of children in intensive care units (ICUs) in Australia and New Zealand in the last 12 years (2002–2013).5 The clinical diagnosis of BSI in children can be challenging due to varied presentation depending on age, site of infection, causative organism and underlying comorbidities.4 The 3 most important factors that influence bacteremia epidemiology include patient age, vaccination coverage and exposure to invasive medical procedures.4 There has been a shift in the epidemiology of bacteremia, with an increasing incidence of healthcare-associated (HCA) BSI caused by Staphylococcus aureus and Enterobacterales.4 There is increasing concern regarding antibiotic resistance rates among Enterobacterales, S. aureus and Streptococcus pneumoniae, partly driven by inappropriate antibiotic use in primary care settings and changes in day care practices.4 Knowledge of local epidemiology is important to guide empiric antibiotic guidelines.1 We conducted this retrospective review of BSI episodes at The Children’s Hospital at Westmead (CHW) to document the major causes of pediatric BSI in various age groups at our hospital and to assess appropriateness of empiric antibiotic therapy.
We reviewed all children with laboratory-confirmed BSI identified at the CHW, Sydney, Australia, during a 3-year period between January 2014 and December 2016. The CHW is a 300-bed specialist pediatric hospital and is the largest pediatric hospital in Australia. All positive blood cultures during the study period were identified from the laboratory electronic database. Ethics approval was obtained.
In our study, bacteremia was defined as a positive blood culture with a significant pathogen deemed to be responsible for clinical illness.1 Isolates commonly considered commensals were included if they were cultured on >1 occasion within a 48-hour period. If the patient had only 1 positive culture with a commensal organism collected from a peripheral site it was regarded as a contaminant and if it was collected from a central venous access device (CVAD) it was considered an isolate of uncertain significance. Primary BSI was defined as a laboratory-confirmed BSI that was not associated with any clinically apparent infection at another body site, and secondary BSI was defined as BSI associated with a tissue or organ infection at another body site.6 We defined central-line associated bloodstream infection as a significant BSI with no other apparent primary site of infection that occurred in a patient who either had a CVAD in situ or had a CVAD removed within 48 hours of the BSI diagnosis.6 Community-onset BSI was defined as the identification of a significant pathogen in a blood culture taken within 48 hours of admission, and if the blood culture was collected >48 hours after admission then it was regarded as a hospital-onset (HO) BSI.2,6 HCA BSI was defined as an episode of BSI in a patient who had a CVAD in situ, or a patient known to and admitted under a subspecialty team (hospital admissions of newly presenting patients are routinely admitted under general medicine at our institution). If an episode was community-onset and not HCA, it was considered community-associated (CA). An episode of BSI included any subsequent cultures of the same organism within 14 days of the first positive blood culture; repeat positive cultures in this period were not considered a new episode.6 All positive blood cultures from stillborn babies were excluded from this review. We also excluded positive blood cultures collected from patients admitted with fabricated or induced illness by carers.
Blood cultures were processed using the BacT/ALERT automated microbial detection system (Biomerieux) by the microbiology department at CHW. Isolates were identified using VITEK MS automated mass spectrometry microbial identification system using Matrix Assisted Laser Desorption Ionization Time-of-Flight technology. Susceptibility testing was carried out in accordance with Clinical and Laboratory Standards Institute methods and breakpoints.
Between January 2014 and December 2016, there were 2283 positive blood cultures from 47,368 blood cultures submitted to the CHW microbiology laboratory. We excluded 1292 positive blood cultures from the analysis (Fig. 1); 1027 of these were probable contaminants, representing 45% of positive blood cultures and 2.1% of all blood cultures collected. Overall 465 clinically significant episodes of BSI were reported from 391 patients. Bacteremia accounted for 4.8 per 1000 admissions during the study period. The characteristics of the 465 episodes of BSI are summarized in Table 1. The median age of the onset of bacteremia was 43.9 (0.2–223.0) months and the interquartile range of age for children managed for BSI was 9.9–104.5 months; 257 (54.4%) were men. Of the 465 BSI episodes, 131 (28.2%) episodes were community-onset community-associated (CO-CA), 187 (40.2%) were community-onset healthcare-associated and 147 (31.6%) were HO. Patients managed by hematology and oncology teams comprised more than one-third of patients with comorbidities. More than half of the patients with a BSI were children with a CVAD in situ.
Of the clinically significant BSI episodes, 243 (52.3%) were caused by Gram-positive bacteria (S. aureus was the most frequent with 198 episodes); 198 (42.6%) were caused by Gram-negative bacteria (Escherichia coli was the most frequent); 18 (3.9%) were polymicrobial BSIs; and 18 (3.9%) were caused by yeast (Table 1). All BSI episodes caused by coagulase-negative staphylococci and yeast were in patients with a CVAD.
Primary BSI was more common in neonates (<28 days old) and secondary BSI was more common in children >28 days old. The most frequent sites of infection among secondary BSI were urinary tract infection in 28- to 365-day-old children and osteoarticular infection in children >1 year of age (Fig. 2). BSI caused by Streptococcus agalactiae was the most common cause of BSI in neonates. E. coli was the most common organism in children 28–365 days old and S. aureus was the most common in all children >1 year of age reflecting the prevalence of urinary tract infection and osteoarticular infections respectively (Fig. 3). S. aureus and Streptococcus pyogenes were the most common organisms causing BSI in patients with skin and soft tissue infection. Streptococcus pneumoniae was most common in patients admitted with pneumonia Fig. 4.
Among the patients with CO-CA BSI, secondary BSI (88.5%) was much more frequent than primary BSI (11.5%). Osteoarticular infections (33.6%) followed by urinary tract infection (17.6%) were then most common sites of infection in secondary BSI. S. aureus (39.7%), E. coli (16.8%), Streptococcus pyogenes (15.3%), Salmonella species (11.5%) and Streptococcus pneumoniae (10.7%) were the most prevalent organisms causing CO-CA BSI (Table 1).
The healthcare burden and morbidity of CO-CA BSI in previously healthy children was considerable. Forty-two percent (55 of 131) of these episodes required surgical intervention, abscess drainage or debridement of the infected site or CVAD insertion for prolonged courses of antibiotics, and 10 episodes (7.6%) required patient admission to the ICU for cardiorespiratory support. Despite significant morbidity, no BSI attributable mortality was reported among CO-CA BSI at 30 days in this cohort. Overall, 30-day mortality was 2.8 % (13 episodes) of all cases managed for BSI, all with HCA BSI and having preexisting comorbidities.
Of the 10 children with community-onset BSI managed in ICU, all were infants <2 years of age. Three patients with CA-methicillin resistant S. aureus (CA-MRSA) and multi-focal tissue disease; 3 with invasive pneumococcal disease (2 non-vaccine serotype).
All of the clinically significant isolates collected from positive blood cultures (including duplicates) underwent antimicrobial susceptibility testing (Table 2). Of the 15 Streptococcus pneumoniae isolates tested, 13 (86.6%) showed reduced susceptibility or resistance to penicillin and 20% showed reduced susceptibility to third-generation cephalosporins. About one-fifth (22 of115) of the S. aureus isolates were methicillin-resistant S. aureus (MRSA). The majority of E. coli isolates were susceptible to third-generation cephalosporins (63 of 69; 91.3%) and gentamicin (62 of 69; 89.9%).
This is the largest study of pediatric bacteremia from Australia. More than 70% of all episodes of BSI were either community-onset healthcare-associated or HO BSI. This may reflect an increase in CVAD use in patients with comorbidities and increase in the ambulatory and outpatient care for such children, a trend we have previously observed at our institution among episodes of S. aureus bacteremia.7 Similarly, a study from United Kingdom showed that bacteremia in children presenting to emergency department is increasingly healthcare-associated and resistant to recommended empiric antibiotics.2 This emphasizes the need to continue to optimize infection control strategies and CVAD care in hospital as well as in the community including a focus on parent/carer education regarding CVAD care.
We identified a contamination rate (false positive blood culture) consistent with that described in other studies. An acceptable contamination rate in established guidelines is 2%–4% of blood cultures collected. Blood culture contamination is a considerable problem that results in additional costs because of further testing, extended hospital stay and unnecessary antibiotic use.1 It is important for hospitals to monitor contamination rates of blood cultures and develop and implement strategies to reduce the impact of contamination.1,8
S. aureus and E. coli were the most common causes of BSI in our cohort, a finding similar to that described in other pediatric centers from Europe and North America. This reflects the change in the epidemiology of BSI associated with increased vaccination coverage in high-income countries.1,4,9 Australia has one of the world’s most comprehensive, publically funded national immunization programs with high coverage (>90%) in almost all locations. National vaccination of children against Haemophilus influenzae type B was introduced in 1993, for Neisseria meningitidis serogroup C in 2003, and for Streptococcus pneumoniae (pneumococcal conjugate vaccine) in 2005.1,10 Before the introduction of comprehensive vaccination against these encapsulated bacteria, previous Australian studies showed that H. influenzae type b accounted for 65% of community-acquired BSI among children <5 years of age in 198711 and Streptococcus pneumoniae was the leading cause (21.6 %) of BSI in Queensland between 2001 and 2010 with a decreasing trend observed in the final few years after the introduction of Prevenar 7.1 In our study, H. influenzae type b BSI was extremely uncommon (2 of 465, 0.4%) and Streptococcus pneumoniae BSI was also infrequent (26; 5.6%). We did not identify any cases of culture-proven N. meningitidis BSI during the study period; however, this period 4 children were positive for N. meningitidis using polymerase chain reaction (PCR) analysis of blood, and this may reflect the timing of blood cultures for this organism relative to antimicrobial therapy, or differences in the sensitivity and specificity of blood culture and PCR for invasive meningococcal disease. PCR was not used routinely for any other pathogens. A multicenter retrospective cohort study from multiple ICUs in Australia and New Zealand also reported a significant decrease in meningococcal and pneumococcal infections in the era of national vaccination program against these organisms.5
Although the epidemiology of pediatric BSI may vary with geographical location, we think the causative pathogens in CO-CA secondary BSI in this cohort will be representative of that in other high-income countries and even in resource limited settings with well implemented vaccination programs. It is evident from the literature that S. aureus and Streptococcus pyogenes are the most causes of osteoarticular infection and skin and soft tissue infection. In addition, E. coli is recognized as the most common cause of urosepsis across all ages. However, the epidemiology of antimicrobial resistance of the prevalent organisms causing BSI will be location dependent. We have shown a stable rate of methicillin resistance among CA S. aureus of about 20% since 2006 at the CHW, similar to the rate published by the Australian Group on Antimicrobial Resistance in 2008 and slightly higher than that described by Dr McMullan and colleagues when they studied >1000 cases of Staphylococcus bacteremia from Australia and New Zealand, reporting that 13.2 % of S. aureus isolates were MRSA.12,13 There was a low rate of extended-spectrum beta-lactamase resistance in E. coli in blood stream isolates at our hospital.
We note that all the children with CO-CA BSI who required intensive care admission had Gram-positive sepsis; 3 patients with MRSA and 1 with penicillin-resistant Streptococcus pneumoniae. Our local empiric antibiotic guidelines recommend giving cefotaxime, vancomycin and gentamicin. In a national study from Australian and New Zealand pediatric ICUs, approximately half of the patients admitted with severe infection were previously healthy and the most common pathogens responsible were N. meningitidis, S. aureus, Streptococcus pyogenes, E.coli and Streptococcus pneumoniae.5
Our study has several limitations. First, we classified BSI in all patients admitted under a subspecialty team as healthcare-associated BSI. It is possible that some CA infections may have been misclassified as healthcare-associated, leading to underestimation of CA BSI in this cohort; given admission practices at our hospital, we think this possible misclassification bias is small. Second, we considered all single positive cultures with a common commensal organism collected from patients with CVADs to be of uncertain significance. Many of these patients did not have another culture collected within 48 hours and before commencing antibiotics. This may have resulted in an underestimation of central-line associated bloodstream infection frequency and overestimated the culture contamination rate. However, we thought it is important to retain the specificity of the definition for clinically significant BSI in patients with a CVAD in situ. Third, this was a laboratory-based study and all the patients managed for BSI were identified by positive blood cultures done at our hospital. Children with culture-negative BSI/sepsis and patients with culture-proven BSI where the blood culture was collected in a different center before transfer would not have been identified in this study which may have resulted in underestimation of the total rate of BSI managed in our hospital. We think it is unlikely that the overall clinical and microbiologic spectrum of disease would be affected. Fourth, we note that the susceptibility testing results were extracted as an aggregate for all blood cultures, rather than per episode, and the results we have presented may include duplicate isolates. However, we believe the study provides a representative overview of blood stream infection managed at a specialist referral pediatric hospital in Australia.
Our study shows the impact of a comprehensive national vaccination program against selected encapsulated bacteria, emphasizes the burden of osteoarticular infection and urinary tract infection in CO-CA BSI and reveals the rising importance of CVADs in pediatric BSI.
We would like to acknowledge all of the microbiology laboratory staff for their hard work to get us the microorganisms profile and Ms. Archana Karlekar for her assistance obtaining the data from the laboratory information system.
1. Er J, Wallis P, Maloney S, et al. Paediatric bacteraemias in tropical Australia. J Paediatr Child Health. 2015;51:437–442.
2. Irwin AD, Drew RJ, Marshall P, et al. Etiology of childhood bacteremia and timely antibiotics administration in the emergency department. Pediatrics. 2015;135:635–642.
3. Agyeman PKA, Schlapbach LJ, Giannoni E, et al. Swiss Pediatric Sepsis Study. Epidemiology of blood culture-proven bacterial sepsis in children in Switzerland: a population-based cohort study. Lancet Child Adolesc Health. 2017;1:124–133.
4. Pai S, Enoch DA, Aliyu SH. Bacteremia in children: epidemiology, clinical diagnosis and antibiotic treatment. Expert Rev Anti Infect Ther. 2015;13:1073–1088.
5. Schlapbach LJ, Straney L, Alexander J, et al. ANZICS Paediatric Study Group. Mortality related to invasive infections, sepsis, and septic shock in critically ill children in Australia and New Zealand, 2002-13: a multicentre retrospective cohort study. Lancet Infect Dis. 2015;15:46–54.
6. Prevention CfDCa. Bloodstream Infection Event (Central Line-Associated Bloodstream Infection and Non-central Line Associated Bloodstream Infection). In: National Healthcare Safety Network (NHSN). Patient Safety Component Manual. 2018:1–38. Available at: https://www.cdc.gov.pdfs
7. Roediger JC, Outhred AC, Shadbolt B, et al. Paediatric Staphylococcus aureus
bacteraemia: a single-centre retrospective cohort. J Paediatr Child Health. 2017;53:180–186.
8. Weddle G, Jackson MA, Selvarangan R. Reducing blood culture contamination in a pediatric emergency department. Pediatr Emerg Care. 2011;27:179–181.
9. Biondi E, Evans R, Mischler M, et al. Epidemiology of bacteremia in febrile infants in the United States. Pediatrics. 2013;132:990–996.
10. Australian Technical Advisory Group on Immunization (ATAGI). The Australian Immunization Handbook. 2018. 10th ed. Canberra: Australian Government Department of Health; Available at: immunisationhandbook.health.gov.au
11. McIntyre PB, Tilse MH, O’Callaghan M, et al. Blood cultures in hospitalized children. Med J Aust. 1987;147:485–489.
12. Britton PN, Andresen DN. Paediatric community-associated Staphylococcus aureus
: a retrospective cohort study. J Paediatr Child Health. 2013;49:754–759.
13. McMullan BJ, Bowen A, Blyth CC, et al. Epidemiology and mortality of Staphylococcus aureus
Bacteremia in Australian and New Zealand children. JAMA Pediatr. 2016;170:979–986.