Objectives To use national laboratory surveillance data to determine whether pathogens responsible for neonatal bacteraemia were sensitive to nationally recommended antibiotic regimens.
Design All reports of neonatal bacteraemia received by the Health Protection Agency's voluntary surveillance scheme in England and Wales from January 2006 until March 2008, were extracted from the database. Organisms were ranked by frequency, and proportions susceptible to antimicrobials recommended for empirical treatment of neonatal sepsis were determined.
Results There were 1516 reports of bacteraemia for neonates <48 h old (early-onset) and 3482 reports for neonates 2–28 days old (late-onset). For early-onset bacteraemia, group B streptococcus (GBS) was the most frequent pathogen (31%) followed by coagulase-negative staphylococci (CoNS; 22%), non-pyogenic streptococci (9%) and Escherichia coli (9%). For late-onset bacteraemia, CoNS were isolated most frequently (45%), followed by Staphylococcus aureus (13%), Enterobacteriaceae (9%), E coli (7%) and GBS (7%). More than 94% of organisms (early-onset) were susceptible to regimens involving combinations of penicillin with either gentamicin or amoxicillin, amoxicillin combined with cefotaxime or cefotaxime monotherapy. More than 95% of organisms (late-onset) were susceptible to gentamicin with either flucloxacillin or amoxicillin and amoxicillin with cefotaxime, but only 79% were susceptible to cefotaxime monotherapy.
Conclusions Current guidelines for empirical therapy in neonates with sepsis are appropriate. However, gentamicin-based regimens should be used in preference to cefotaxime-based treatments, because of lower levels of susceptibility to cefotaxime and the need to avoid exerting selective pressure for resistance. Surveillance data linked to clinical data should further inform rational antibiotic prescribing in neonatal units.
Statistics from Altmetric.com
Neonatal septicaemia remains a major cause of mortality and morbidity despite advances in perinatal and neonatal care.1 In neonates, early warning symptoms of septicaemia are often minimal, but the clinical course may be fulminant. Therefore, antimicrobial treatment of neonates with suspected sepsis must start without delay. As microbiological and antimicrobial susceptibility results are not immediately available, initial antimicrobial treatment is usually empirical with the aim of being effective against the most likely pathogens. In order to guide empirical prescribing, it is crucial to monitor changes in the pattern of causative organisms and their antimicrobial susceptibility profiles.
What is already known on this topic
▶ Knowledge of both the common pathogens causing septicaemia in neonates and their antimicrobial susceptibility is essential in order to select appropriate antimicrobial treatment.
▶ Empirical antimicrobial treatment is increasingly compromised by antimicrobial resistance.
What this study adds
▶ The current pattern of bacteria isolated from blood cultures taken from neonates in England and Wales is described.
▶ National guidelines for empirical antimicrobial therapy in neonates with sepsis appear appropriate.
▶ However, cefotaxime-based treatments should not be recommended, because of lower susceptibility levels and the need to avoid exerting selective pressure.
The main routes of infection for a neonate are perinatal vertical transmission and postnatal exposure to organisms usually associated with nosocomial infections. Perinatal vertical transmission manifests itself in early-onset infection during the first 2 days after birth, the most common pathogen being group B streptococcus (GBS). Following the introduction of universal screening for maternal GBS colonisation and adoption of widespread intrapartum antibiotic prophylaxis (IAP) in the USA in the mid-1990s, changes in the distribution of pathogens from predominantly gram-positive organisms to gram-negative Enterobacteriaceae, particularly Escherichia coli, were reported.2,–,5 However, since many factors impact on the maternal vaginal flora (age, ethnic group, underlying medical condition, community and intrapartum antibiotic treatment), the reason for this shift in causative organisms is uncertain. Recent studies in the UK report national incidence rates of GBS infection in infants similar to those seen in the USA despite the lack of screening for antenatal GBS carriage and routine IAP use.6
In contrast, late-onset infections in neonates are usually nosocomial and become clinically evident more than 2 days after birth. Selection pressure due to intensive use of antimicrobials in neonatal units (NNUs) has lead to changes in the antimicrobial susceptibility patterns of organisms associated with nosocomial infections. Changes in the epidemiology of nosocomial infections and outbreaks of methicillin-resistant Staphylococcus aureus, ampicillin-resistant E coli, vancomycin-resistant enterococci and multi-resistant gram-negative organisms are increasingly being reported from NNUs.7,–,10 This is a major concern, especially since resistant gram-negative bacteria make the choice of empirical therapy more difficult and result in a poorer prognosis for the treated patient.11
Guidance for health professionals on prescribing medicines to children in Britain is provided by the British National Formulary for Children (BNF-C).12 However, a recent review of antibiotic policies in British and Irish NNUs found adherence to the BNF-C guidelines in only 69% of NNUs with widespread use of other antimicrobial regimes.13 In the present study, we used national laboratory surveillance data to determine whether the pathogens responsible for neonatal bacteraemia were sensitive to the antibiotic treatment recommended by the BNF-C and other commonly used antimicrobial regimes.
We extracted all reports of neonatal bacteraemia to the Health Protection Agency's (HPA) voluntary surveillance scheme in England and Wales from January 2006 until March 2008. Voluntary reporting to the HPA's LabBase2 surveillance database is continuous and captures routine microbiological information on bacteraemia from 90% of the hospital laboratories in England and Wales.14 A case of neonatal bacteraemia was defined as isolation of bacteria from blood culture from an infant younger than 29 days. Following the BNF-C's classification, early-onset infection was defined as presenting in neonates less than <48 h old and late-onset infection as occurring in neonates 2–28 days old. Bacteraemia reports included the neonate's date of birth, sex and hospital, the sample date, microorganism isolated, antimicrobial susceptibility and the reporting laboratory. Reports for the same strain from the same neonate within a 14-day period were removed. We grouped organisms according to whether they were gram-positive or gram-negative. Varying species of coagulase-negative staphylococci (CoNS) were combined, as were non-pyogenic streptococci, Enterobacteriaceae other than E coli (including Klebsiella spp., Enterobacter spp., Morganella spp., Kluyvera spp., Citrobacter spp., Pantoea spp., Proteus spp., Salmonella spp., Serratia spp. and Yersinia enterocolitica) and different species of Enterococcus, Pseudomonas, Corynebacterium and Acinetobacter. Pathogen groups were ranked according to frequency and we analysed their susceptibility to antimicrobials, or combinations of antimicrobials that are currently recommended for treating suspected neonatal septicaemia, based on antibiotic sensitivities reported by the laboratory. Organisms were regarded as susceptible if they were reported as being susceptible to either one or both of the antibiotics tested. Overall estimates of the proportion of pathogens susceptible to different treatment regimens are shown with and without CoNS, which rarely cause fulminant infections, are often due to contamination and are usually treated based on antibiotic sensitivities. Microsoft Office Access 2003, Microsoft Office Excel 2003 and STATA v 10.1 were employed for data handling and analysis.
Reports of isolates from neonatal bacteraemia
Between January 2006 and March 2008, the surveillance scheme received 1516 reports of bacteria isolated from blood cultures taken from neonates <48 h old (early-onset) and 3482 reports for neonates 2–28 days old (late-onset). Fifty-one species of bacteria were reported for both age groups. The ranked distributions for the 20 most frequently reported organisms are presented for the two age groups in tables 1 and 2. The majority of bacteraemias were attributed to gram-positive organisms (82% in early-onset and 81% in late-onset bacteraemias). For early-onset bacteraemias, GBS (31%) were the most frequent pathogens, followed by CoNS (22%) and non-pyogenic streptococci (9%). E coli was the most frequently reported gram-negative organism (9%). For late-onset bacteraemias, CoNS accounted for the majority (45%) of isolates reported. Other common organisms were S aureus (13%) and Enterococcus spp. (7%). The most frequent gram-negative bacteria were Enterobacteriaceae other than E coli (9%).
Antimicrobial susceptibility of bacterial isolates
Among early-onset bacteraemias, the antimicrobial susceptibility of all isolates with known test results was 94% to the combination of penicillin+gentamicin, 100% to amoxicillin+cefotaxime, 98% to amoxicillin+penicillin and 96% to monotherapy with cefotaxime. After exclusion of CoNS results, susceptibility to the antibiotic regimes was the same or higher (table 3). The antimicrobial profiles for bacteria isolated from late-onset bacteraemias show that susceptibility was 69% to the combination flucloxacillin+gentamicin, 93% to amoxicillin+cefotaxime, 96% to amoxicillin+gentamicin and 78% to cefotaxime as a single agent. Susceptibility to the antimicrobial regimes was the same or higher after exclusion of results for CoNS (table 4).
We reviewed national laboratory surveillance data to guide empirical prescribing for neonatal septicaemia. These data show that the aetiology of neonatal bacteraemia is diverse and varies by age, but the reported causative organisms are consistent with previous studies on pathogens causing bacteraemia in neonates.15,–,18 Gram-positive organisms caused the majority of neonatal bacteraemias. Only 18% of the reported early-onset bacteraemias were due to gram-negative organisms, with E coli being the most frequently reported. For late-onset bacteraemias, 19% of reports indicated a gram-negative organism, with Enterobacteriaceae other than E coli most frequently reported.
The predominance of gram-negative infections in the neonatal care setting in the USA described in the very low-weight preterm infant population may reflect the implementation of universal GBS screening and use of IAP which has resulted in a significant decline in early-onset group B streptococcal infection in the USA.3 19 Routine screening for antenatal GBS carriage is not recommended in the UK and IAP use is considered only in the presence of known risk factors.20 21 The continued predominance of GBS as the main pathogen reported as early-onset infection in our study thus probably reflects the difference in IAP use between countries. As we have no data on birth weight/gestation we cannot provide a direct comparison with these US data. The decrease in GBS reported for late-onset bacteraemias compared to reports for early-onset bacteraemia demonstrates the shift from vertical transmission as the cause of neonatal sepsis during the first days of life to hospital-associated infections.22 23
CoNS accounted for 22% of early-onset bacteraemia and nearly half of late-onset bacteraemia. CoNS are often regarded as contaminants, sometimes being excluded from analyses.3 16 However, one study found that more than half of late-onset CoNS-positive neonatal bacteraemias were considered to be true infections.24 We were not able to determine the clinical significance of CoNS and other organisms of uncertain pathogenicity in our dataset due to lack of clinical parameters and have included these isolates in our overall analysis. However, as CoNS is predominantly a nosocomial pathogen, it is likely that the apparent early-onset CoNS bacteraemias predominantly reflect contamination rather than true infection. Even when isolated from late-onset bacteraemia, it is argued that CoNS is not a significant cause of mortality in neonates and does therefore not warrant specific empirical antibiotic therapy.25 Presenting the analysis with and without CoNS allowed consideration of this viewpoint.
The overall susceptibility of blood culture isolates (excluding CoNS) to the antibiotics commonly used in British and Irish NNUs showed low resistance rates for most regimes. For early-onset bacteraemia, more than 94% of organisms were susceptible to regimens involving combinations of penicillin with either gentamicin or amoxicillin, amoxicillin combined with cefotaxime or cefotaxime monotherapy. For late-onset bacteraemia, more than 95% of organisms were susceptible to gentamicin with either flucloxacillin or amoxicillin and amoxicillin with cefotaxime, but only 79% were susceptible to cefotaxime monotherapy.
Despite an overall susceptibility of ≥93% for amoxicillin+cefotaxime, only 75% of the Enterobacteriaceae other than E coli and 46% of the Pseudomonas spp. from late-onset bacteraemias were susceptible to this combination. This is of particular concern since these pathogens are associated with significant morbidity and mortality in neonates and 11% of late-onset bacteraemias were due to these organisms in our study. Furthermore, cefotaxime is generally not considered to be effective clinically against enterococci, Acinetobacter spp. and Listeria monocytogenes, organisms which ranked among the 20 most commonly reported pathogens in our study.26 Moreover, the frequent use of third-generation cephalosporins (eg, cefotaxime) has been shown to drive the development of resistant bacterial pathogens on neonatal intensive care units and has been linked to outbreaks caused by extended-spectrum β-lactamase-producing bacteria.27,–,29 de Man et al as well as other teams provided evidence that a regime avoiding cefotaxime decreases the incidence of resistant gram-negative bacteria on NNUs.28 30 Minimising antimicrobial resistance is particularly important in the NNU setting where the newborn patients have a low grade of endogenous flora and will inevitably be colonised by nosocomial strains.
There were two limitations to our study. Due to the lack of clinical parameters we were not able to determine whether isolation of organisms from blood reflected true infection or contamination. Also, it was not possible to correlate in vitro susceptibility testing with in vivo clinical effectiveness. Future work aimed at addressing these issues might include linkage of LabBase2data to clinical and prescribing data. Additionally, the low frequency of susceptibility tests for some antimicrobial/organism combinationslimited the interpretations of results. In particular, cefotaxime susceptibility was reported for only 4–7% of gram-positive organisms and for 37–39% of gram-negative organisms. This is probably due to laboratories choosing their panel of antibiotics for susceptibility testing based on the initial identification of the organism rather than the guidance for empirical prescribing. For example, cefotaxime is not included in national surveillance of the antibiotic susceptibility of staphylococci, reflecting the fact that it is not routinely tested.31
This analysis of national laboratory surveillance data has shown that current guidelines for empirical antimicrobial therapy in neonates with sepsis appear appropriate, given the results from susceptibility testing of causative organisms. However, cefotaxime-based treatments should not be recommended for empirical treatment of suspected neonatal sepsis in the UK, because of the lower levels of susceptibility in important causative organisms and the need to avoid exerting selective pressure particularly on gram-negative bacteria. We would suggest that NNUs switch to gentamicin-based regimens rather than cefotaxime-based regimens. More robust surveillance data, linked to clinical and treatment data, could further inform rational antibiotic prescribing based on microbiological findings as well as the type of patients affected.
The authors thank the microbiology laboratories across the country for their considerable efforts in reporting to LabBase2. The authors also thank our colleagues in the iCAP Group (Improving Antibiotic Prescribing in Primary care): Professor Ian Wong, The School of Pharmacy; Dr Sonia Saxena, Imperial College London; Dr Paul Long, The School of Pharmacy; Dr Alessandro Porta, St George's Hospital; Helen Bird, St George's Hospital; Ruthie Birger, Imperial College London; Saida Mehonic; Shama Wagle, BNF Publications; Joanna Murray, Imperial College London; Elizabeth Koshy, Imperial College London; Yingfen Hsia, The School of Pharmacy. We also thank Theresa Lamagni for her helpful advice.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
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.