You are here

  • Home
  • Archive
  • Volume 99, Issue 6
  • Very low rates of culture-confirmed invasive bacterial infections in a prospective 3-year population-based surveillance in Southwest London

Article Text

Article menu

Download PDFPDF

Original article
Very low rates of culture-confirmed invasive bacterial infections in a prospective 3-year population-based surveillance in Southwest London
  1. Kirsty Le Doare1,2,
  2. Anna-Louise Nichols1,
  3. Helen Payne1,
  4. Rosy Wells1,
  5. Sonia Navidnia1,
  6. Gayle Appleby1,
  7. Elizabeth Calton1,
  8. Mike Sharland1,
  9. Shamez N Ladhani1,3,
  10. the CABIN Network
  1. 1Paediatric Infectious Diseases Research Group, St. George's Hospital, London, UK
  2. 2Wellcome Trust/Imperial Centre for Global Health Research, Imperial College, London, UK
  3. 3Immunisation Department, Public Health England, London, UK
  1. Correspondence to Dr Shamez N Ladhani, Immunisation Department, Public Health England, 61 Colindale Avenue, London NW9 5EQ, UK; shamez.ladhani{at}phe.gov.uk

Abstract

Objectives To estimate the incidence, clinical characteristics and risk factors for culture-confirmed invasive bacterial infections in England.

Design Prospective, observational, study of all children with positive blood and/or cerebrospinal fluid (CSF) culture over a 3-year period (2009–2011).

Setting All five hospitals within a geographically defined region in southwest London providing care for around 600 000 paediatric residents.

Patients Children aged 1 month to 15 years

Main outcome measures Rates of community-acquired and hospital-acquired invasive bacterial infections in healthy children and those with co-morbidities; pathogens by age group, risk group and clinical presentation.

Results During 2009–2011, 44 118 children had 46 039 admissions, equivalent to 26 admissions per 1000 children. Blood/CSF cultures were obtained during 44.7% of admissions, 7.4% were positive but only 504 were clinically significant, equivalent to 32.9% of positive blood/CSF cultures, 2.4% of all blood/CSF cultures and 1.1% of hospital admissions. The population incidence of culture-confirmed invasive bacterial infection was 28/100 000. One-third of infections were hospital acquired and, of the community-acquired infections, two-thirds occurred in children with pre-existing co-morbidities. In previously healthy children, therefore, the incidence of community-acquired invasive bacterial infection was only 6.4/100 000.

Conclusions Although infection was suspected in almost half the children admitted to hospital, a significant pathogen was cultured from blood or CSF in only 2.4%, mainly among children with pre-existing co-morbidities, who may require a more broad-spectrum empiric antibiotic regime compared to previously healthy children. Invasive bacterial infection in previously healthy children is now very rare. Improved strategies to manage low-risk febrile children are required.

https://doi.org/10.1136/archdischild-2013-305565

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    What is already known on this topic

    • Rates of serious bacterial infections in children have declined over the past 20 years, mainly through the national immunisation programme.

    • The UK has one of the highest childhood mortality rates among industrialised countries, with infections contributing to around 20% of childhood deaths.

    • Childhood hospital admission rates in England continue to rise, especially among infants.

    What this study adds

    • The hospital admission rate for children aged 1 month to 15 years was 26/1000 and 45% were for suspected infection; only 1.1% of all admissions and 2.4% of those with suspected infection had a significant pathogen grown from blood or cerebrospinal fluid.

    • Of the 504 significant invasive bacterial infections (annual incidence 28/100 000), 185 (37%) were hospital-acquired, 204 (40%) had community onset but occurred in children with known co-morbidities. Only 115 (23%) had a community onset and occurred in previously healthy children (6.4/100 000); incidence rates in the latter group are far below those for invasive Hib disease alone prior to routine vaccination.

    • The pathogens grown from children with community-onset infection and known co-morbidities were similar to those in children with hospital-acquired infection.

    Introduction

    The success of national immunisation programmes and the availability of effective antimicrobial therapies have led to significant declines in the burden of childhood infectious diseases in industrialised countries.1 At the same time, the number of childhood admissions to hospital, particularly for infection-related illnesses, has steadily increased to such an extent that more than a third of all infants are admitted to hospital in the first year of life.2 A major limitation of the current national surveillance programmes is the paucity of clinical and outcome data at a population level for children with invasive bacterial infections. In the literature, most population-based studies have focused on specific infectious diseases, mainly in relation to assessing the impact of national vaccination programmes. In 2009, therefore, a 3-year, population-based prospective surveillance was initiated to systematically collect information on the epidemiology, pathogens responsible, risk factors, clinical characteristics and outcome of culture-confirmed invasive bacterial infections in southwest London (SWL).

    Methods

    The web-based Childhood Acute Bacterial Infection Network (CABIN) surveillance project was developed for children aged 1 month to 15 years (inclusive) at five National Health Service (NHS) hospitals in SWL: St. George's Hospital NHS Trust, Croydon University Hospital, Epsom and St. Helier NHS Trust and Kingston Hospital. These hospitals provide medical care for an estimated 600 000 paediatric residents within a geographical region.3

    During 2009–2011, children with a positive blood and/or cerebrospinal fluid (CSF) culture at any of the five hospitals were identified through the local microbiology laboratory at monthly intervals. Neonates (aged less than one calendar month) were excluded as these infants are followed-up as part of a separate neonatal infection surveillance programme (http://www.neonin.org.uk).

    A standardised web-based questionnaire for each positive culture (http://www.cabin.sgul.ac.uk/) was completed by the lead clinician at each participating site using the case notes. Data were anonymised at source and included demographic details (gestational age at birth, age, gender), clinical information (co-morbidities, risk factors, acute presentation), laboratory parameters (white cell and neutrophil counts, C-reactive protein), microbiological data (pathogen, antibiotic sensitivity), management (antimicrobial therapy, intensive care admission), final diagnosis and outcome for each hospitalisation episode. Data were collated, verified and any missing/incorrect information or queries resolved by contacting the local lead. With modern culturing technologies, blood volumes collected are less problematic than previously and microbiological techniques are standardised as all participating hospitals operate BACTEC (Oxon.) blood culturing systems. For the duration of the study, the standard practice at all participating hospitals was to take at least 1–2 mL blood for culture from children suspected with infection prior to the commencement of antibiotic therapy.

    Definitions

    We pragmatically defined a positive blood/CSF culture to be clinically significant if the clinical presentation was consistent with infection caused by that pathogen and the child received antibiotic therapy directed towards that pathogen with or without additional interventions (eg, removal of an infected central line). Positive cultures from other sterile sites were not included unless accompanied by a positive blood/CSF culture. A hospital-acquired (HA) invasive bacterial infection was defined as a significant culture obtained >48 h after admission. Multiple positive cultures from the same patient with the same or different organism(s) isolated within 7 days were considered part of the same episode.

    Data analysis

    Population data were obtained from the 2011 UK National Census data (http://www.ons.gov.uk/ons/guide-method/census/2011/census-data/index.html). The total number of paediatric admissions, blood and CSF cultures taken by age, sex and month of admission were provided by participating hospitals. Anonymised clinical data from the secure web-based database were transferred to Microsoft Excel and Stata V.11 for final analysis. Children were categorised into three age-groups: infants (1–11 months), preschool (1–4 years) and older children (5–15 years). Data are mainly descriptive and summarised as proportions, with statistical significance assessed using the χ2 test or Fisher's exact test. Continuous data that did not follow a normal distribution are summarised as medians with IQRs and compared using the Mann–Whitney U test.

    Results

    During 2009–2011, a total of 44 118 children aged 1 month to 15 years had 46 039 admissions (overall admission rate, 2575/100 000 or 26/1000 admissions). Assuming that blood/CSF cultures were taken from children with suspected infection, almost half the admissions (20 578/46 039, 44.7%; 1151/100 000 population) were infection related. Overall and infection-related admission rates declined with age (table 1). The median age of children admitted with suspected infection was 30 (IQR, 3–72) months and 279 invasive bacterial infections (55.3%) occurred in males.

    View this table:
    Table 1

    Population rates of invasive bacterial infection by age 2009–2011 (episodes)

    Culture positivity

    A total of 18 366 blood cultures and 2212 lumbar punctures were performed in 18 982 children. Blood and CSF cultures were positive in 1442 (7.9%) and 88 (4.0%), respectively, and 482 (33.4%) blood and 22 (35%) CSF positive cultures were determined to be clinically significant. Overall, there were 504 (32.9%) invasive bacterial infection episodes in 375 children over the 3-year period, including 190 (39.6%) infants, 151 (25.3%) preschool and 163 (36.1%) older children. The pathogens causing invasive bacterial infections varied by age group (table 2), presence or absence of pre-existing underlying co-morbidity (table 2) and presence or absence of a central venous catheter (CVC). The majority of children (318/375, 84.8%) had a single admission, 10.4% (39/375) had two admissions and 4.8% (18/375) three or more admissions during the 3-year period.

    View this table:
    Table 2

    Significant pathogen episodes by presentation, age and comorbidity status 2009–2011

    Community-acquired invasive bacterial infections

    Almost two-thirds of invasive bacterial infection episodes (319/504; 63.3%) in 247 children were diagnosed at admission to hospital. Children with pre-existing co-morbidities represented 142/247 (57.5%) of community-acquired invasive bacterial infections and 124/128 (96.9%) of HA Infections (χ2=63.4; p<0.0001). Therefore, only 115 invasive bacterial infection episodes (22.8%) occurred in previously healthy children from the community (6/100 000 population), with the highest rates occurring in infants (38/100 000 population) (table 1).

    Community-acquired invasive bacterial infections in previously healthy children

    In infants with community-acquired invasive bacterial infections, the most common diagnoses were bacteraemia/septicaemia (19/44, 44.4%), meningitis (11/44, 25%) and urinary tract infections with septicaemia (12/44, 27.3%). Of the 45 significant pathogens identified in 44 infants, Escherichia coli (13/45, 28.9%), Group B Streptococcus (GBS) (10/45, 22.2%), Klebsiella pneumoniae (5/45, 11.1%) and Streptococcus pneumoniae (5/45, 11.1%) were responsible for almost three-quarters (table 2). In preschool children, bacteraemia/septicaemia (18/43, 41.9%) was the most common diagnosis followed by pneumonia (10/43, 23.3%) and gastrointestinal infections (5/43, 11.6%) with concomitant positive blood cultures. The most common pathogens in this age-group were S pneumoniae (12/44, 27.2%), Group A Streptococcus (7/44, 15.9%) and Salmonella species (6/44, 13.6%) (table 2). In older children, gastrointestinal infections (8/20, 40%) and osteomyelitis (6/20, 30%) with bacteraemia predominated, and the two main pathogens causing community-acquired infections were Staphylococcus aureus (8/26, 30.8%) and Salmonella sp. (7/26, 26.9%) (table 2). All the Salmonella infections occurred in September–October among travellers returning from India or Pakistan.

    Community-acquired invasive bacterial infection in children with co-morbidities

    Co-morbidities among the 131 children who developed community-acquired invasive bacterial infections varied by age and were similar to those with HA-invasive bacterial infections. Of note, 66.4% (87/131) of children in this group had a CVC, mainly for treatment of cancer (78/119, 65.5%), and coagulase-negative staphylococci (CoNS) were the most common cause. In infants, over a third (10/24, 42%) of whom had been born prematurely, the main co-morbidities were associated with prematurity and its complications. Most infants presented with septicaemia (9/24, 37.5%), CVC infection (7/24, 29.1%) or urinary tract infection with bacteraemia (5/24, 20.8%), with E coli (9/30, 30%) and GBS (4/30, 13.3%) being the most common pathogens. In preschool and older children, malignancy was the main co-morbidity (75/151, 49.7%) followed by neurological conditions (18/151, 11.9%), including three (3/18, 16.7%) with a ventriculoperitoneal shunt. More than half the community-acquired invasive bacterial infections episodes in children with co-morbidity (53.0%) occurred in those with a CVC and 45.5% (145/319) of community-acquired invasive bacterial infections were central line associated, mainly due to CONS (72/319 pathogens, 22.6%) (table 3).

    View this table:
    Table 3

    Significant pathogens by presence of central line and presentation community-acquired versus hospital-acquired serious bacterial infections 2009–2011

    Hospital-acquired invasive bacterial infections

    HA-invasive bacterial infections accounted for 36.7% (185/504) of all invasive bacterial infections and only four previously healthy children (8.7 per 100 000 admissions) developed HA-invasive bacterial infections over the 3-year period. Among the 131 children with co-morbidities who developed 181 HA-invasive bacterial infections episodes, 62.2% (115/185) were infants, mainly premature babies in neonatal intensive care units (NICU). Central line infections (49/93 episodes, 53%), caused mainly by CoNS (56/115 pathogens, 48.7%) predominated in this age-group, followed by septicaemia without focus (26/93, 28%) where the main pathogens were Enterococcus faecalis, GBS, Klebsiella pneumoniae and E coli. In 5–15-year-olds with co-morbidities, the majority of HA-invasive bacterial infections were central line related and most were associated with CoNS (table 2).

    Discussion

    We have established a minimum population-based estimate of childhood, culture-confirmed invasive bacterial infection and described the characteristics of children who developed invasive bacterial infection over a 3-year period. By conducting the study within a defined geographical region, we are able to provide robust estimates of culture-confirmed invasive bacterial infection in a large and ethnically diverse cohort, representing 15% of the total childhood population in England. Inclusion of all children with a positive blood/CSF culture in the surveillance also allowed us to collect extensive clinical and microbiological data.

    There are limited data published on the epidemiology of childhood invasive bacterial infections in countries with established national vaccination programmes. Several large studies have highlighted the low bacteraemia rates in the postvaccine era in children.4–6 In the UK and the Netherlands, for example, serious bacterial infections were identified in 12–25% of previously healthy children attending the Emergency Department. Bacteraemia and meningitis, however, represented only 1% of these infections.7 Although our cohort was restricted to only blood/CSF culture-confirmed cases representing the more severe end of childhood infection, we were able to collect detailed information on cases and responsible pathogens in different age groups and with different risk factors. Our strict inclusion criteria also allow us to compare rates with other population-based studies, including historical rates. In <5-year-olds, for example, our rates of 13.8/000 000 and 2.6/100 000 for culture-confirmed invasive bacterial infections and meningitis, respectively, are far lower than 36/100 000 and 25/100 000 for invasive Hib disease and Hib meningitis alone (which were both diagnosed through culture only), just before routine Hib vaccination was introduced in the UK in 1992.8

    Another important finding is that HA-invasive bacterial infection accounted for a third of all invasive bacterial infection and also two-thirds of children with community-acquired invasive bacterial infections had co-morbidities. HA infections are associated with prolonged hospitalisation, increased healthcare costs and significant long-term morbidity and mortality.9 In common with other reports, we found that HA-invasive bacterial infections were responsible for 60% of all invasive bacterial infection among premature infants. A previous study using data linkage by Blackburn and colleagues noted that approximately 50% of all invasive bacterial infection in infants under 1 year of age was HA.10 In addition, the Health Protection Agency recently reported an overall prevalence of HA infections in children of 5.4% (95% CI 3.9% to 7.5%) in England, with infants aged 1–23 months having the highest HA infection rates (8.2%) (http://www.hpa.org.uk/Publications/InfectiousDiseases/AntimicrobialAndHealthcareAssociatedInfections/1205HCAIEnglishPPSforhcaiandamu2011prelim/). Both studies indicate a worrying trend of rising HA infection rates in this population, which is exacerbated by increasing antimicrobial resistance among pathogens causing HA infections. Early recognition and prompt management is essential in children with suspected invasive bacterial infections in order to improve these outcomes. These include goal-directed therapy, bacterial cultures prior to commencing antibiotics and evidence-based broad spectrum empiric antibiotic use with de-escalation as soon as possible.11

    The remarkably low invasive bacterial infection rates in England, however, are in contrast to the rapidly rising trends in paediatric hospital admissions, predominantly for infection in infants. Since the early 2000's, the number of infants admitted to hospital has continued to rise year-on-year, such that 36% of infants had at least one hospital admission in the first year of life,2 ,12 with the majority of healthy infants admitted for less than 1 day,12 mainly for minor acute infections.13 The 2013 National Institute of Health and Care Excellence (NICE) guidance addresses the difficult task of identifying invasive bacterial infections in young children presenting to the emergency department with a febrile illness.14 However, accurate estimates of invasive bacterial infections in the UK are lacking. Rates of childhood invasive bacterial infections that were used to develop the recent NICE guidance, for example, are based on data from the 1980s and our data suggest they significantly overestimate the current population risk of invasive bacterial infections. If our estimate of 1% of hospital admissions is correct, this would have a major impact for policy-makers in the assumptions made about the investigations and management of children presenting with suspected infection.

    The finding that two-thirds of the community-acquired invasive bacterial infection episodes now occur in children with co-morbidities is also new. The risk factors (especially presence of a central line) and pathogens responsible for community-acquired infection in children with co-morbidities were similar to those with HA-invasive bacterial infection. While national empiric prescribing guidelines are in place for fever in children with malignancy, no such guidance exists for children with other co-morbidities. The British National Formulary for Children (BNF-C) currently recommends ceftriaxone for all children with suspected invasive bacterial infection but does not differentiate between previously healthy children and those with co-morbidities,15 even though the pathogens responsible and their antimicrobial susceptibility profile are very different. Further studies are needed to evaluate whether children with co-morbidities other than malignancy who become unwell in the community might benefit from more broad-spectrum empiric antibiotic therapy.

    Our study has limitations. First, we only included children with blood/CSF culture-confirmed invasive bacterial infection. However, the inclusion of cases with negative blood/CSF cultures but positive cultures from other sterile sites is unlikely to significantly increase the estimated rates. Similarly, local hospital microbiology departments generally do not perform blood/CSF PCR-testing for invasive bacterial infection. A national PCR-testing service is available, but only for meningococcal septicaemia/meningitis and for pneumococcal meningitis/empyema. Given that nearly all children with suspected invasive bacterial infection in England are hospitalised and have blood cultures taken prior to initiating empiric intravenous antibiotic therapy, there are unlikely to be many cases of PCR-positive, culture-negative cases in our cohort. During 2009–2011, for example, there were only nine PCR-positive tests for N meningitidis from 141 blood/CSF samples submitted for testing by all participating sites. Our case definition also does not include non-bacteraemic infections such as osteomyelitis, pneumonia or urinary tract infections, which may cause significant morbidity in children.16 ,17 In Australia, urinary tract infections and pneumonia were each responsible for 3.4% of almost 16 000 febrile illness episodes in children aged <5 years during 2004–200618 ,19 and the new NICE guidelines include urinalysis as part of the investigation algorithm for young children with a febrile illness.14

    Another limitation is that we were unable to compare population rates for invasive bacterial infections in healthy children and in those with co-morbidities because we do not have denominator data for childhood co-morbidity prevalence in SWL. A recent UK study estimated that 1.8% of children aged 2–15 years had co-morbidities associated with an increased risk of invasive pneumococcal disease in England.20 Even if the true number of children with invasive bacterial infection in our cohort were doubled and taking an exaggerated estimate of 5% childhood co-morbidity prevalence, the incidence of invasive bacterial infections in previously healthy children would still remain low at 13/100 000 population.

    Conclusions

    The CABIN study demonstrates the usefulness of combining clinical with epidemiological data in a small but well-defined geographical area and provides a framework for development of evidence-based interventions to prevent and manage childhood invasive bacterial infections. The low rates of invasive bacterial infections in previously healthy children highlight the long-term benefits of the national immunisation programme and other vaccines (such as the recently licensed meningococcal B and the group B streptococcal conjugate vaccine in late-phase clinical trials) are likely to also play an important role in the near future. Given the very low invasive bacterial infection rates in healthy children, new risk-based clinical algorithms need to be devised that include updated estimates of invasive bacterial infections and the possibility of short periods of observation in Paediatric Assessment Units (with or without investigations) to help differentiate non-serious from serious illness. Any such algorithm, however, will require objective validation through prospective studies, but could significantly reduce hospital admissions for low-risk children.

    Acknowledgments

    The authors would like to thank the microbiology and data management teams at all five hospitals and especially Janet Tolley (Epsom & St. Helier's Hospital), David Dalton and Oliver Schmidt (Croydon University Hospital).

    References

    1. ↵
      1. Sanders JW,
      2. Fuhrer GS,
      3. Johnson MD,
      4. et al
      . The epidemiological transition: the current status of infectious diseases in the developed world versus the developing world. Sci Prog 2008;91(Pt 1):1–37.
      OpenUrlCrossRefPubMed
    2. ↵
      1. Saxena S,
      2. Bottle A,
      3. Gilbert R,
      4. et al
      . Increasing short-stay unplanned hospital admissions among children in England; time trends analysis ‘97-'06. PLoS ONE 2009;4:e7484.
      OpenUrlCrossRefPubMed
    3. ↵
      Office of National Statistics (2012) Census, Population and Household Estimates for Wards and Output Areas in England and Wales. http://www.statistics.gov.uk/hub/index.html (accessed 22 Jul 2013).
    4. ↵
      1. Haddon RA,
      2. Barnett PL,
      3. Grimwood K,
      4. et al
      . Bacteraemia in febrile children presenting to a pae3iatric emergency department. Medical J Aust 1999;170:475–8.
      OpenUrl
    5. ↵
      1. Sard B,
      2. Bailey MC,
      3. Vinci R
      . An analysis of pediatric blood cultures in the postpneumococcal conjugate vaccine era in a community hospital emergency department. Pediatr Emerg Care 2006;22:295–300.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Bressan S,
      2. Berlese P,
      3. Mion T,
      4. et al
      . Bacteremia in feverish children presenting to the emergency department: a retrospective study and literature review. Acta Paediatr 2012;101:271–7.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Nijman RG,
      2. Vergouwe Y,
      3. Thompson M,
      4. et al
      . Clinical prediction model to aid emergency doctors managing febrile children at risk of serious bacterial infections: diagnostic study. BMJ 2013;346:f1706.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Booy R,
      2. Hodgson SA,
      3. Slack MP,
      4. et al
      . Invasive Haemophilus influenzae type b disease in the Oxford region (1985–91). Arch Dis Child 1993;69:225–8.
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Reed D,
      2. Kemmerly S
      . Infection control and prevention: a review of hospital-acquired infections and the economic implications. Oschsner J 2009;9:27–31.
      OpenUrl
    10. ↵
      1. Blackburn RH,
      2. Henderson K,
      3. Minaji M,
      4. et al
      . Exploring the epidemiology of hospital-acquired bloodstream infections in children in England (January 2009–March 2010) by linkage of national hospital admissions and microbiological databases. JPIDS 2012;1:284–92.
      OpenUrl
    11. ↵
      1. Dellinger RP,
      2. Levy MM,
      3. Rhodes A,
      4. et al
      . Surviving sepsis campaign: international guidelines for management of severe sepsis and septic shock: 2012. Crit Care Med 2013;41:580–637.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Gill PJ,
      2. Goldacre MJ,
      3. Mant D,
      4. et al
      . Increase in emergency admissions to hospital for children aged under 15 in England, 1999–2010: national database analysis. Arch Dis Child 2013;98:328–34.
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Koshy E,
      2. Murray J,
      3. Bottle A,
      4. et al
      . Significantly increasing hospital admissions for acute throat infections among children in England: is this related to tonsillectomy rates? Arch Dis Child 2012;97:1064–8.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      National Institute for Health and Care Excellence (NICE). Feverish illness in children—Assessment and initial management in children younger than 5 years. [CG160]. London: National Institute for Health and Care Excellence, 2012.
    15. ↵
      1. Ladhani S,
      2. Pebody RG,
      3. Ramsay ME,
      4. et al
      . Continuing impact of infectious diseases on childhood deaths in England and Wales, 2003–2005. Pediatr Infect Dis J 2010;29:310–13.
      OpenUrlPubMed
    16. ↵
      1. Shah TA,
      2. Meinzen-Derr J,
      3. Gratton T,
      4. et al
      . Hospital and neurodevelopmental outcomes of extremely low-birth-weight infants with necrotizing enterocolitis and spontaneous intestinal perforation. J Perinatol 2012;32:552–8.
      OpenUrlCrossRefPubMed
    17. ↵
      1. Myers AL,
      2. Hall M,
      3. Williams DJ,
      4. et al
      . Prevalence of bacteremia in hospitalized pediatric patients with community-acquired pneumonia. Pediatr Infect Dis J 2013;32:736–40.
      OpenUrlCrossRefPubMed
    18. ↵
      1. De S,
      2. Williams GJ,
      3. Hayen A,
      4. et al
      . Accuracy of the traffic light clinical decision rule for serious bacterial infections in young children with fever: a retrospective cohort study. BMJ 2013;346:f866.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Craig JC,
      2. Williams GJ,
      3. Jones M,
      4. et al
      . The accuracy of clinical symptoms and signs for the diagnosis of serious bacterial infection in young febrile children: prospective cohort study of 15 781 febrile illnesses. BMJ 2010;340:c1594.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. van Hoek AJ,
      2. Andrews N,
      3. Waight PA,
      4. et al
      . Effect of serotype on focus and mortality of invasive pneumococcal disease: coverage of different vaccines and insight into non-vaccine serotypes. PLoS ONE 2012;7:e39150.
      OpenUrlCrossRefPubMed

    Footnotes

    • Correction notice This paper has been amended since it was published Online First. In the abstract in the results section, first sentence, 24 admissions has been corrected to 26 admissions. In the second sentence of the results section of the abstract “7.4% of all blood/CSF cultures” has been corrected to “2.4%”. In the last sentence of the results section, 6.0/100 000 has been changed to 6.4/100 000. Also, on page 3 and in tables 2 and 3, Salmonella aureus has been corrected to Staphylococcus aureus.

    • Collaborators St George's Hospital: Adam Irwin, Laura Segal, Peter Riley; Croydon University Hospital: Jennifer Handforth; Epsom and St. Helier Hospital: Rim El Rifai; Kingston Hospital: Rowan Heath, Sue Luck.

    • Contributors KLD monitored data collection for the whole project, wrote the statistical analysis plan, cleaned and analysed the data, and drafted and revised the paper. A-LN and HP designed data collection tools and wrote the statistical analysis plan, RW monitored data collection at Kingston Hospital and contributed to the data analysis, SN monitored data collection at St. George's Hospital and contributed to the data analysis, GA and LC monitored data collection of oncology patients, contributed to data analysis and contributed to the drafting of the paper, SNL and MS initiated the collaborative project, developed the data collection tools and statistical analysis plan, implemented the data collection in all hospitals, and substantially contributed to the writing and revision of the paper.

    • Funding This Study was supported by the European Society for Paediatric Infectious Diseases (grant number SGA ALN).

    • Competing interests None.

    • Ethics approval The Oxfordshire Multicentre Research Ethics Committee (Reference: 11/SC/0054) granted ethical approval for the study.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement Additional data are available on request by contacting the corresponding author.