Article Text

Download PDFPDF

Early onset neonatal sepsis: diagnostic dilemmas and practical management
  1. A R Bedford Russell,
  2. R Kumar
  1. Neonatal Unit, Birmingham Women's NICU, Birmingham, UK
  1. Correspondence to Dr A R Bedford Russell, Neonatal Unit, Birmingham Women's NICU, Mindelsohn Way, Birmingham B15 2TG, UK; Alison.bedfordrussell{at}nhs.net

Abstract

Early onset neonatal sepsis is persistently associated with poor outcomes, and incites clinical practice based on the fear of missing a treatable infection in a timely fashion. Unnecessary exposure to antibiotics is also hazardous. Diagnostic dilemmas are discussed in this review, and suggestions offered for practical management while awaiting a more rapidly available ‘gold standard’ test; in an ideal world, this test would be 100% sensitive and 100% specific for the presence of organisms.

  • Neonatology
  • Evidence Based Medicine
  • Infectious Diseases
  • Immunology
  • Microbiology

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.

Introduction

Culture-proven early onset neonatal infection is confirmed in <1%1 of admissions to neonatal units, yet accounts for up to 16%2 of all neonatal mortality and contributes to significant morbidity. Culture-negative infection occurs more frequently.

Fear of infection results in the majority of newborns who are given antibiotics, receiving them unnecessarily.3 The hazards are beyond the fact that antibiotic therapy drives antibiotic resistance. Antibiotics also disrupt maternal and the newborn's faecal flora, and in so doing, disrupt normal development of the nascent immune system.4 Neonatologists are constrained by the availability of insufficiently sensitive and specific microbiological diagnostic tools, which would allow more judicious antibiotic prescribing.

In recognition of the number of dilemmas regarding diagnosis and management of early onset neonatal infection which contribute to variations in practice, the UK National Institute for Health and Care Excellence (NICE) commissioned a clinical guideline development group (CGDG), consisting of a wide range of stakeholders, to systematically review all available evidence and develop a guideline which would prioritise the treatment of sick newborn babies and use antibiotics more selectively.5 Recently published evidence has strengthened the original NICE CGDG reccomendations.6

Definition, incidence and causes of early onset neonatal sepsis

Early onset neonatal sepsis (EONS) is variably defined as infection occurring within 48 h1–72 h2 of birth.

The incidence is around 0.901 (UK) to 0.982 (USA) per 1000 live births, and 0.9% of all neonatal admissions.1 In all studies, the risk of sepsis and mortality increases with decreasing gestational age and birth weight,7 increasing to approximately 1% in babies 401–1500 g,2 and is highest in babies <1000 g, with 26% of all admissions <1000 g experiencing one or more episodes of infection during their neonatal stay.1

Group B streptococcus (GBS) and Escherichia coli are the most common causative organisms of EONS, excluding coagulase-negative staphylococci (CoNS). CoNS are frequent skin contaminants accounting for 20%1 (UK) to 24%2 (USA) of positive cultures. EONS secondary to GBS is defined as infection at <7 days7 and may present as a fulminating septicaemic illness, often complicated by pneumonia. Some centres consider all EONS CoNS isolates as true infections; others consider all as contaminants. Rigorous blood culture drawing technique is key. Because of the wide variation in interpretation, many studies exclude cases whereby a single culture is positive for CoNS2 or only include such cases when deemed clinically relevant.1 GBS accounts for 43%2 (USA)–58%1 (UK) and E coli for 18%1 (UK) to 29%2 (USA) of EONS bacteraemias, respectively. Listeria and Staphylococcus aureus (methicillin-sensitive staphylococcus aureus) are infrequently isolated. In an American study,2 73% infants with GBS isolates were term, and 81% with E coli were preterm. In the UK, while the absolute number of term babies with GBS EONS is higher than for preterm babies, spontaneous preterm delivery alone represents a risk factor for infection with the incidence of GBS EONS in babies <1500 g being 4/1000 versus 0.38/1000 for term babies.

The incidence of EONS secondary to GBS has not reduced in the UK as it has in the USA2 and Australasia1 ,8 subsequent to the introduction of universal antenatal screening for GBS at 35–37 weeks gestation, and intrapartum antibiotic prophylaxis (IAP) is offered to those women who are GBS-colonised, at the onset of labour.

In the UK, the National Screening Committee (UKNSC) has controversially not recommended a universal maternal GBS screening programme. This is currently under review. The Royal College of Obstetricians Green Top guideline9 recommends a risk-factor-based strategy to prevent GBS. IAP is, however, recommended for prevention of GBS infection if a woman is found to be colonised with GBS when incidentally tested, or has had a previous baby with invasive GBS disease, or as part of broad-spectrum antibiotic therapy if she is febrile in labour, or has evidence of chorioamnionitis.

Antenatal risk factors for GBS infection can be identified in up to 60% of cases and include preterm delivery (<37 weeks), prolonged rupture of membranes (PROM ≥18 h) or known genital carriage of GBS during pregnancy.7 GBS can also cross intact membranes, and the absence of PROM does not negate the risk of GBS infection: while membranes were ruptured for >18 h in 44% of babies with culture-proven GBS EONS, 56% with GBS EONS were not born following PROM. Successive studies have demonstrated that a risk-factor-based approach to prevention of GBS is limited, with frequent missed opportunities for prevention of GBS sepsis when risk factors are present.10 ,11

Clinical indicators and risk factors for EONS

Risk factors contribute to a clinical diagnosis but are insufficiently robust to be reliable in making a diagnosis of EONS. A number of risk factor scores have been devised for asymptomatic neonates at risk of EON,12 none of which has been shown to be sufficiently robust for widespread use. NICE CGDG noted the lack of good quality evidence to guide management of neonatal early onset sepsis. In producing best practice guidance on management, NICE CGDG reached consensus view to stratify risk factors and clinical indicators of EONS5 attributing ‘Red Flags’ to indicators that should prompt a high level of concern (box 1). Any baby with one Red Flag indicator or risk factor, or two or more ‘Non-red Flag’ indicators (box 2) should be promptly assessed for infection and treated with antibiotics without delay. The guideline evidence has recently been updated to reflect publication of relevant new information,6 none of which has any potential impact on the original guidance (box 3). It is notable that while epidural anaesthesia may be a risk factor for early onset neonatal fever, it is not associated with a higher incidence of EONS. Hence, it is suggested that it is unnecessary to investigate febrile offspring of mothers who have had epidurals, unless fever persists or the baby has other signs or risk factors for neonatal sepsis.6

Box 1

National Institute for Health and Care Excellence Red Flag risk factors and clinical indicators for early onset neonatal infection (EONS) which should prompt a high level of concern of EONS 5

Maternal and neonatal risk factors

▸ Parenteral antibiotic treatment to woman for confirmed or suspected invasive bacterial infection at any time during labour or in the 24 h periods before and after the birth.

▸ Suspected or confirmed infection in another baby in the case of a multiple pregnancy.

Neonatal clinical indicators

▸ Respiratory distress starting >4 h after birth.

▸ Seizures.

▸ Need for mechanical ventilation in a term baby.

▸ Signs of shock.

▸ Any baby with one Red Flag indicator or risk factor, should be promptly assessed for infection and treated with antibiotics without delay.

Box 2

National Institute for Health and Care Excellence ‘non-Red Flag’ risk factors for, and indicators of, early onset neonatal sepsis

Maternal

▸ Invasive group B streptococcal infection in a previous baby.

▸ Maternal group B streptococcal colonisation, bacteriuria or infection in the current pregnancy.

▸ Prelabour rupture of membranes.

▸ Preterm birth following spontaneous labour (before 37 weeks gestation).

▸ Suspected or confirmed rupture of membranes for more than 18 h in a preterm birth.

▸ Intrapartum fever higher than 38°C, or confirmed or suspected chorioamnionitis.

Neonatal clinical indicators

▸ Altered behaviour or responsiveness.

▸ Altered muscle tone (eg, floppiness).

▸ Feeding difficulties (eg, feed refusal).

▸ Feed intolerance, including vomiting, excessive gastric aspirates and abdominal distension.

▸ Abnormal heart rate (bradycardia or tachycardia).

▸ Signs of respiratory distress.

▸ Hypoxia (eg, central cyanosis or reduced oxygen saturation level).

▸ Jaundice within 24 h of birth.

▸ Apnoea.

▸ Signs of neonatal encephalopathy.

▸ Need for cardiopulmonary resuscitation.

▸ Need for mechanical ventilation in a preterm baby.

▸ Persistent fetal circulation (persistent pulmonary hypertension).

▸ Temperature abnormality (lower than 36°C or higher than 38°C) unexplained by environmental factors.

▸ Unexplained excessive bleeding, thrombocytopenia or abnormal coagulation (International Normalised Ratio >2.0).

▸ Oliguria persisting beyond 24 h after birth.

▸ Altered glucose homeostasis (hypoglycaemia or hyperglycaemia).

▸ Metabolic acidosis (base deficit of 10 mmol/L or greater).

▸ Local signs of infection (eg, affecting the skin or eye).

▸ Any baby with two or more ‘non-red Flag’ indicators should be promptly assessed for infection and treated with antibiotics without delay.

Box 3

Summary of key points regarding new evidence reviewed for National Institute for Health and Care Excellence Antibiotics for early onset neonatal infection guideline,6 with no potential impact on guidance

Key Points

Risk factor for infection and clinical indicators of possible infection

▸ Epidural analgesia may be a risk factor for early onset neonatal fever, irrespective of intrapartum fever, but is not associated with a higher incidence of neonatal infection.

Investigations before starting antibiotics in the baby

▸ Full blood count indices may not be sufficiently sensitive to rule out early onset infection in neonates.

▸ Serum procalcitonin concentration: the heterogeneity of available evidence on use of procalcitonin concentrations, prevents conclusions regarding the value of this marker.

Antibiotics for suspected infection

▸ Gentamicin dosing 5 mg/kg every 36–48 h according to gentamicin level at 22 h can achieve effective and safe gentamicin levels in very preterm as well as term babies.

Duration of antibiotic treatment

▸ Serial C-reactive protein (CRP) measurements Antibiotic treatment can safely be stopped at 48 h in culture-negative very low birth weight infants who have CRP concentrations <10 mg/L at presentation and at 48 h.

More recently, pulse oximetry has gained momentum as a screening tool to detect illnesses including EONS in addition to congenital heart disease, which the ‘PulseOx study’ was originally designed to investigate.13 Of 208 babies who ‘failed’ initial pulse oximetry, 55 had pneumonia (with radiological features and raised C-reactive protein/CRP), two had culture-proven GBS sepsis and 28 had raised CRPs with clinical signs suggesting culture-negative sepsis. Pulse oximetry is currently being considered by the UKNSC for inclusion as a screening test adjunctive to the newborn and infant physical examination.

Diagnosis of EONS

Early diagnosis of EONS is challenging because clinical characteristics are non-specific and difficult to differentiate from those of non-infectious aetiologies. The repertoire of ancillary laboratory tests is also limited and not always reliable.14 Over 95% of babies treated with antibiotics for suspected infection ultimately prove to have no evidence of infection.3

Microbial culture

The ‘gold standard’ for diagnosing infection is, historically, a positive blood or cerebrospinal fluid (CSF) culture with a minimum reporting delay of 36–48 h. Microbial cultures suffer from low sensitivity and specificity: a negative blood culture result is almost inevitable for a large proportion of blood cultures because of the submission of inadequate volumes of blood, and only one culture being drawn.15 ,16 A false positive culture of CoNS is common,1 making diagnosis of EONS using the historical gold standard, a challenge.17

To avoid false positive cultures, blood for culture should be drawn from a freshly punctured blood vessel using a strict aseptic technique and a closed system. The skin disinfectant should be left to dry for at least 1 min to insure maximal killing of skin organisms. The common practice of using an open system (insertion of cannula from which blood is aspirated with a separate syringe and needle placed in the hub of the cannula), risks the dilemma of how to interpret a false positive culture result. There is also a risk of false positives if blood is drawn from an indwelling vascular device.17

Positive cultures from sites such as the umbilicus, groin, ear, nose, throat, pharynx and rectum and gastric aspirates are informative about colonisation, but are of limited value in diagnosis.5 Colonisation of babies without clinical signs of infection does not warrant antibiotic treatment. The same applies to a GBS-positive maternal vaginal swab, which indicates colonisation but not necessarily invasive infection unless the baby has symptoms and signs of infection.5

The ideal diagnostic test would be rapid, sensitive, specific and not affected by maternal antibiotic therapy. Reliable diagnostic techniques may better inform antibiotic management of the newborn, and negative results enable clinicians to have confidence in prescribing antibiotics for shorter periods of time or not at all.

Molecular assays

Advances in molecular microbiology have provided new molecular assays based on PCR, which amplifies specific target regions in the microbial genome, and helps to detect specific bacterial proteins in body fluids and swabs. The advantages include the speed with which results become available, and the ability to tailor antimicrobial therapy to those results.

A recent systematic review and meta-analysis18 assessed whether molecular assays have sufficient sensitivity (>0.98) and specificity (>0.95) to replace microbial cultures in the diagnosis of neonatal sepsis. Although real-time PCR and broad-range conventional PCR amplification methods had higher sensitivity (0.9; 95% CI 0.78 to 0.95) and specificity (0.96; 95% CI 0.94 to 0.97), than other assays, molecular assays still do not have sufficient sensitivity to replace microbial cultures in the diagnosis of neonatal sepsis but may perform well as ‘add-on’ tests.

A variety of technological advances are needed before PCR can replace conventional culture, including: improved recovery of micro-organisms in whole-blood extractions, increased assay sensitivity, simpler testing platforms that could easily be run 24 h a day, and more assays to detect antibiotic-resistant genes, so reducing reliance on culture-based protocols for antimicrobial susceptibility testing.19

Lumbar puncture

A lumbar puncture (LP) should be performed to obtain CSF prior to starting antibiotics if it is thought safe to do so, and there is a strong clinical suspicion of infection, or there are clinical symptoms or signs suggestive of meningitis.5 Antibiotic therapy should not be delayed in order to perform an LP.

In a retrospective case review of infants who developed meningitis in the first 72 h of life, selective criteria for performing an LP would have delayed or missed the diagnosis of meningitis in up to 37% of those evaluated. Those with meningitis included preterm babies with respiratory distress syndrome (RDS), asymptomatic term infants with positive blood cultures as well as term infants with no neurological signs or symptoms, and negative blood cultures.20 Positive CSF cultures, despite negative blood cultures, have also been reported for very low birthweight neonates, without specific neurological clinical manifestations, undergoing screens for suspected late-onset sepsis, and support a lower threshold for performing an LP as part of a late-onset sepsis screen.20

Pronounced variation in performance of LP for EONS, even when adjusting for clinical conditions that would prompt LP, have been reported in the USA21 ,22 and may also be a feature in the UK in spite of NICE guidance.

Interpretation of CSF microscopy

Interpretation of neonatal CSF parameters is difficult if white blood cell (WBC) counts are marginally raised or the tap is traumatic. An upper limit of CSF WBC count >21/mm3 gives 79% sensitivity and 81% specificity for the diagnosis of meningitis.21 CSF WBC counts and protein levels are higher and decline more slowly with postnatal age in preterm infants compared with term infants.23 The various formulas and ratios applied to traumatic taps to compare observed with predicted WBC counts or protein levels in CSF samples are unreliable. Adjustment merely results in a loss of sensitivity, with marginal gain in specificity, and tends to ‘over-correct’ the result.24 In suspicious clinical cases, the only course is to repeat the LP after 24–48 h.

The role of chest radiograph

Pneumonia is a common presentation of EONS,7 and may be missed if a chest radiograph is not performed. GBS pneumonia mimics RDS,25 and should be considered if a baby with radiographic appearances of RDS, is disproportionately sick. NICE guidelines5 do not give a directive on the role of chest radiography as part of a screen for EONS, however, it is notable that even in older children pneumonia may be present with limited clinical signs, and there is significant added value of chest radiography in the diagnosis of pneumonia.26

Evaluation of diagnostic methods

Leucocyte indexes and C-reactive protein

CRP concentrations may be normal in the early stages of infection,3 and values are subject to physiological variation during the first few days of life limiting the use of single values. Following antigenic stimulation, it takes at least 12 h for a CRP level to become raised;27 serial measurements in the first 24–48 h of symptoms increases the test sensitivity, and normal CRP values during this period have a 99% negative predicted value for diagnosis of infection.28 The CRP should be measured at presentation and again after 18–24 h in order to facilitate decisions regarding LP (if initial CRP >10 mg/L), and decision making at 36 h, regarding duration of antibiotic therapy.5

Studies including two systematic reviews of the likelihood ratios for leucocyte indexes and CRP to predict sepsis have all concluded that there is too much heterogeneity within the studies and no such ideal test or combination of tests.29 ,30 Leucopenia and neutropenia, as well as high immature to total neutrophil ratio (I:T ratio) are undoubtedly associated with increasing odds of infection (ORs 5.38, 6.84 and 7.90, respectively); however, the test sensitivities for detection of sepsis are low.31 The combination of two normal I:T ratios within 24 h and a negative blood culture has been suggested as indicative of a non-infected neonate, and may be a contributory marker to assist with the decision to stop antibiotics.32 Having reviewed the more recent evidence, NICE evidence update advisory group (NICE EUAG) does not recommend full blood count for the diagnosis of EONS as the indices are insufficiently sensitive to exclude EONS6 (box 2).

Serum procalcitonin

Procalcitonin (PCT) is the prehormone of calcitonin, normally secreted by thyroid C-cells, and rises within 3–6 h of exposure to infection. It has been suggested that elevated serum PCT concentrations are more sensitive and specific in the differentiation between neonatal infection and inflammation than CRP and may also differentiate between bacterial and viral infection. In a meta-analysis of 16 studies (1959 neonates), the sensitivity and specificity for diagnosing infection were 81% and 79%, respectively.33 Limitations of the evidence included variation in definition of neonatal sepsis, differences in age and gestation of neonates and a high level of statistical heterogeneity among studies analysed. Hence, NICE EUAG concluded that the value of PCT in diagnosis of infection requires further research, and PCT concentration cannot be recommended within the current guideline (box 3).

Cytokine profiles and neutrophil/monocyte adhesion molecules

Multiple cytokines, for example, interleukins 6, 8 and 10, and tumour necrosis factor-α, and leucocyte adhesion molecules, for example, CD64, CD11b, have been studied for diagnosis of neonatal sepsis. All lack sufficient sensitivity and specificity to be recommended as diagnostic tools for EONS.14

Antimicrobial therapy

Antibiotic therapy drives antibiotic resistance and also alters the types of colonising microbial flora, especially in the gut leading to skewing of immune development. E coli EONS has, at best, remained stable in the USA, but approximately 82% of E coli isolates are resistant to amoxicillin or gentamicin in preterm infants.2 Because gut flora drive the nascent immune system, peripartum antibiotic exposure is increasingly recognised as a major driver of immune dysregulation, resulting in an increasing incidence of atopy and asthma in childhood.4

Antibiotic therapy should be stopped after 36 h if cultures are negative, if two CRP measurements are negative5 and if there are no further signs of infection. Alternatively, when there is a suspicion that clinical progress is suboptimal, consideration should always be given to an empiric change of antibiotic therapy to include a broader spectrum of pathogens.

A multiple logistic regression analysis revealed that empiric treatment failure of EONS could be predicted at 24 h using the following variables: need for vasoactive treatment (OR 2.83 (1.21 to 6.66)); WBC <5000 or >20 000 per mm3 on day 1 (2.51 (1.09 to 5.81)); I:T ratio >0.2 on day 1 (2.79 (1.10 to 7.11)) and platelet count per 10 000 mm3 increase on day 1 (0.92 (0.86 to 0.98)).34 Such analyses should be validated in other datasets but have the potential to improve neonatal outcome by ensuring that appropriate antibiotics are used as early as possible. Once a bacterium is identified, the antibiotic regimen should be targeted appropriately.

An initial gentamicin dose of 5 mg/kg regardless of gestation has been recommended by NICE CGDG5 but not universally adopted by UK neonatologists, who have concerns regarding renal clearance of gentamicin in very immature babies. If a second dose is required, it is recommended that this should be given at 36 h, with a trough level immediately prior to the second dose, and adjustment of the dose to achieve a trough concentration of <2 mg/L. A retrospective analysis in babies <28 weeks, demonstrated that giving 5 mg/kg every 36–48 h achieves safe and effective peak and trough gentamicin concentrations, according to levels taken at 22 h.35 This evidence is reassuring and supports NICE recommendations.6

Antimicrobial stewardship

Antimicrobial stewardship (AMS) refers to the coordinated interventions to prescribe and measure the most appropriate, pathogen-specific, narrow-spectrum drug regimen, dose, duration of therapy and route of administration.36 Department of health guidance provides an outline of evidence-based AMS in the secondary healthcare setting promoting a ‘Start Smart—Then Focus’ approach for all antibiotic prescriptions (box 4). Elements of an AMS should also include an assessment of acute-trust AMS activities using the self-assessment toolkit, a trust AMS management team/committee, ward-focused antimicrobial teams and evidence-based antimicrobial-prescribing guidelines. Some recommendations are not applicable to the neonatal context, for example, changing intravenous to oral antibiotics at the nearest opportunity, however, neonatal-specific recommendations have previously been published in this journal37 along with broader guidelines on managing and preventing Gram-negative infections in neonatal units.38

Box 4

Antimicrobial Stewardship Programme: ‘Start Smart—Then Focus’36

Start Smart

  • Do not start antibiotics in the absence of clinical infection.

  • Use local guidelines to initiate prompt effective antibiotic treatment if there is evidence/suspicion of infection.

  • Document on drug chart and in medical records: clinical indication, duration of review date, route and dose.

  • Obtain cultures first.

Then Focus

  • Review the clinical diagnosis and continuing need for antibiotics by 48 h and make a clear plan of action: the ‘Antimicrobial Prescribing Decision’.

  • The Antimicrobial Prescribing Decisions relevant to neonates are: Stop, Change or Continue.

  • It is essential to document the review and subsequent decision in the medical records.

The tension between the need to investigate a baby at risk of EONS and to prescribe antibiotics promptly, versus the fact that most newborn babies given antibiotics do not have infection but become exposed to the risks of antimicrobial therapy, has been recognised by NICE. Exposure to antibiotics must be minimised safely. NICE Quality Standards Advisory Committee and Project Team are developing quality standards with the aim of reducing infant mortality, reducing admissions and readmissions to neonatal units, maternity and neonatal length of stay and improving neonatal neurological and auditory development. Implementation of NICE antibiotics for neonatal infection quality standards are anticipated to shape future neonatal antibiotic prescribing practice from birth to 28 days of life in the UK, and likely beyond.

Summary

The main diagnostic dilemmas in managing EONS centre on the lack of a rapid and reliable test, which is 100% specific and sensitive for presence of microbes in sterile sites. Recent evidence has been reviewed and does not alter NICE guidelines, which were developed using robust methodology and which should be used to manage babies at risk and with clinical indicators of EONS. All neonatal units must adopt an AMS programme; while antibiotics can be life saving, they are by no means harmless to those who do not need them.

References

View Abstract

Footnotes

  • Contributors ARBR wrote the first and final drafts. RK conducted an extensive literature review and updated the appropriate sections.

  • Competing interests None.

  • Provenance and peer review Commissioned; externally peer reviewed.

Linked Articles

  • Fantoms
    Ben Stenson