Article Text

Comparison of the management recommendations of the Kaiser Permanente neonatal early-onset sepsis risk calculator (SRC) with NICE guideline CG149 in infants ≥34 weeks’ gestation who developed early-onset sepsis
  1. Rachel Morris1,
  2. Steve Jones2,
  3. Sujoy Banerjee1,
  4. Andrew Collinson3,
  5. Hannah Hagan4,
  6. Hannah Walsh3,
  7. Graham Thornton2,
  8. Ian Barnard5,
  9. Chris Warren3,
  10. Jennifer Reid4,
  11. Alison Busfield4,
  12. Jean Matthes1
  1. 1 Neonatal Intensive Care, Singleton Hospital, Swansea, Wales, UK
  2. 2 Paediatrics, Royal United Hospitals Bath NHS Foundation Trust, Bath, UK
  3. 3 Paediatrics, Royal Cornwall Hospitals NHS Trust, Truro, UK
  4. 4 Paediatrics, Royal Devon and Exeter NHS Foundation Trust, Exeter, Devon, UK
  5. 5 Neonatal Medicine, Glan Clwyd Hospital, Rhyl, Wales, UK
  1. Correspondence to Dr Rachel Morris, Neonatal Intensive Care, Singleton Hospital, Swansea, Wales CF14 4XW, UK; rmmorris19{at}


Objective To compare the management recommendations of the Kaiser Permanente neonatal early-onset sepsis risk calculator (SRC) with National Institute for Health and Care Excellence (NICE) guideline CG149 in infants ≥34 weeks’ gestation who developed early-onset sepsis (EOS).

Design Retrospective multicentre study.

Setting Five maternity services in South West of England and Wales.

Patients 70 infants with EOS (<72 hours) confirmed on blood or cerebrospinal fluid culture.

Methods Retrospective virtual application of NICE and SRC through review of maternal and neonatal notes.

Main outcome measure The number of infants recommended antibiotics by 4 hours of birth.

Results The incidence of EOS ≥34 weeks was 0.5/1000 live births. Within 4 hours of birth, antibiotics were recommended for 39 infants (55.7%) with NICE, compared with 27 (38.6%) with SRC. The 12 infants advised early treatment by NICE but not SRC remained well, only one showing transient mild symptoms after 4 hours. Another four babies received antibiotics by 4 hours outside NICE and SRC guidance. The remaining 27 infants (38.6%) received antibiotics when symptomatic after 4 hours. Only one infant who was unwell from birth, died. Eighty-one per cent of all EOS infants were treated for clinical reasons rather than for risk factors alone.

Conclusion While both tools were poor in identifying EOS within 4 hours, NICE was superior to SRC in identifying asymptomatic cases. Currently, four out of five EOS have symptoms at first identification, the majority of whom present within 24 hours of birth. Antibiotic stewardship programmes using SRC should include enhanced observation for infants currently treated within NICE guidance.

  • sepsis risk calculator
  • neonatal
  • early onset neonatal sepsis
  • NICE Guideline CG149

Statistics from

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?

  • Sepsis risk calculator (SRC), used on many units in the USA, has safely reduced the number of babies receiving antibiotics.

  • In the UK, there is a huge potential to reduce the number of babies treated unnecessarily with antibiotics.

What this study adds?

  • By 4 hours of birth, SRC recommended antibiotics for 38.6%, while the National Institute for Health and Care Excellence recommended antibiotics for 55.7% of infants with early-onset sepsis (EOS).

  • Eighty-one per cent of infants with EOS are currently identified by clinical examination when they become unwell.

  • Enhanced recording of observations and education of parents and staff should supplement the SRC if introduced in UK.


Early-onset neonatal sepsis (EOS) occurs in approximately 0.7 per 1000 live births1 in the UK. The most common causative organism is Group B Streptococcus (GBS). Two initiatives influencing management are the National Institute for Health and Care Excellence (NICE) guideline CG149 (see online supplementary appendix 2—advising investigation and antibiotic administration to babies with one red flag or two risk factors),2 3 and the Royal College of Obstetricians and Gynaecologists guidance on intrapartum antibiotic prophylaxis,4 the latter using a risk-based approach rather than universal screening for maternal GBS. While these guidelines undoubtedly save lives, many healthy mothers and babies receive antibiotics to prevent a very infrequent condition, although with significant morbidity and mortality.5

Treating newborn infants unnecessarily with antibiotics does not represent prudent healthcare. It exposes babies to unnecessary invasive procedures, risks related to antibiotic toxicity and side effects from drug errors, increases maternal anxiety, affects bonding and reduces initiation of breast feeding.6 Altering the baby’s microbiome may affect the delicate programming of both innate and adaptive immune systems7–11 and may contribute to serious lifetime sequelae.12–19 Furthermore, unnecessary use of antibiotics may promote the development of antibiotic resistance.20 21

In the USA, use of the Kaiser Permanente sepsis risk calculator (SRC)22 reduced the number of babies exposed to antibiotic prophylaxis. Publications including a recent meta-analysis show the safety of this approach in the USA.23 24 However, healthcare systems differ and there is no published study to inform if this approach is safe in the UK, or if additional resources or system enhancements are required to implement this tool. SRC is a freely available web-based tool ( well validated in babies ≥34 weeks’ gestation in the USA. It calculates the risk of EOS using a Bayesian approach.25–27 Starting with the incidence of EOS ≥34 weeks in the general population, the risk is modified using continuous variables—gestation, highest maternal intrapartum temperature, time in hours of membrane rupture prior to birth, and discontinuous variables such as maternal carriage of GBS (positive, negative or unknown), and maternal antibiotic use during labour. This generates the prior or birth risk. The clinical condition of the infant—well appearing, equivocal or clinical illness—is incorporated to generate a posterior risk. While the sepsis risk score (SRS) is evidence based, the management recommendations (normal care, enhanced observations, blood culture with enhanced observation or blood culture with antibiotics) are a consensus opinion of the Kaiser Permanente clinicians.

The aim of this study was to compare recommendations of NICE guidelines with SRC up to 4 hours after birth in infants with confirmed EOS through retrospective application of both tools. This outcome would inform a wider quality improvement initiative to reduce antibiotic use.

Patients and methods

Considering our project was a review of clinical notes, with virtual application of clinical tools and no actual change of practice, the institutional joint scientific research committees classified this as non-research/quality improvement, not requiring ethical approval.

Participating neonatal units identified all positive blood or cerebrospinal fluid (CSF) cultures taken within the first 72 hours in infants ≥34 weeks’ gestation, through their microbiology and other perinatal databases. Inclusion criteria were positive blood or CSF culture with an organism known to cause EOS and treated with at least 5 days of antibiotics, or intention to treat if the baby died. Cultures growing coagulase-negative staphylococcus or other skin commensals were likely contaminants and excluded. The units were at liberty to go back as far as their data allowed, as long as it had a complete set of all eligible babies.

Individual unit and cohort rates of EOS were calculated for infants ≥34 weeks’ gestation. Units had the option to use their own background EOS incidence, rates available on the SRC calculator at the time of study were between 0.2 and 0.6 per 1000, or they could choose the overall study cohort rate of 0.5 per 1000 (table 1). Data extracted from the maternal and neonatal notes included gestation, highest maternal antepartum temperature, rupture of membranes duration (ROM), maternal GBS status, and type and duration of intrapartum antibiotics. GBS carriage contributes only 2.3% of the variance to the SRS26 and the option of inputting ‘unknown’ allows use in UK without altering validity. Using specific guidance (online supplementary appendix 1), these data were entered into the SRC to generate the prior risk. The clinical state of the infant (well, equivocal, clinical illness) was then entered to compute a posterior risk which generated the management recommendations.

Table 1

Incidence of early-onset sepsis in babies born ≥34 weeks’ gestation

We chose to compare the tools at 4 hours, to accommodate categorisation to the ‘equivocal’ clinical state on the SRC that could take 2–4 hours of observation.

The SRC management recommendation related to the computed posterior risk: ≥3/1000—empiric antibiotics; 1.54–3/1000—blood culture and monitor vital signs every 4 hours for 24 hours; ≥0.65 to <1.54/1000—monitor vital signs for 24 hours; <0.65/1000—routine vitals only. Where the calculator directed ‘strongly advise antibiotics’ or ‘consider antibiotics’ we regarded this as ‘empiric antibiotics’ for our comparison with NICE.

To assess the NICE recommendations, we recorded maternal and neonatal red flags and risk factors. We made a slight modification to the NICE guidance. In many cases, time of onset of labour was unknown to enable calculation of duration of prelabour rupture of membranes. We pragmatically used ROM ≥24 hours (in term infants) as a proxy similar to common clinical practice. The local assessors wrote a narrative, the organism identified, and the infant’s outcome to discharge. The data were independently assessed by two central assessors who checked for any discrepancies, the accuracy of SRS calculations and the management recommendations by SRC and NICE. Analysis was undertaken on Excel 2016 (Microsoft) and SPSS V.25 (IBM). The correct recommendations of NICE and SRC were used for analysis, rather than the actual intervention, to reduce bias from inaccurate interpretation of guidelines.


Five neonatal units including an intensive care unit, two local units and two special care units (table 1) participated in the study over a variable period. Seventy-two babies of gestation ≥34 weeks had culture proven EOS, out of a birth population of 142 333 of the same gestation range. 87.5% of pathogens were GBS (table 2). Two cases were excluded, one record could not be located and the other had insufficient recorded information. The remaining 70 cases with complete information were included in the analysis. The only death was a baby with septicaemia and meningitis symptomatic from birth. Table 3 shows the patient characteristics.

Table 2

Organisms identified

Table 3

Infant characteristics

Both NICE and SRC recommended antibiotics for 27 babies within the first 4 hours. One was clinically well. All others had signs or symptoms consistent with sepsis. Most were on continuous positive airway pressure or other respiratory support. Three received therapeutic cooling.

NICE recommended antibiotics in the first 4 hours to another 12 babies, whereas SRC did not. Of these, six mothers were feverish in labour (>38°C) while eight received intrapartum antibiotics. These babies, treated for NICE red flags or risk factors, were clinically well. Only one developed transient respiratory symptoms. In no case did SRC recommend giving antibiotics where NICE did not.

If the tools were rigorously applied, 31 babies should not have received antibiotics within the first 4 hours by either tool. Nevertheless, four did receive antibiotics. One had a congenital abnormality with exposed organs, while three had respiratory concerns, which settled quickly. NICE guideline creep, clinical nervousness or other unrecorded concerns accounted for these. Age at presentation for the remainder was as follows: 12 infants symptomatic between >4 and 8 hours, 6 infants between >8 and 12 hours, 6 infants between >12 and 24 hours, and the remaining 3 infants >24 hours (presenting at 36, 44 and 52 hours).

NICE advises ‘For babies with no red flags and only one risk factor consider whether it is safe to withhold antibiotics or necessary to monitor the baby’s vital signs and clinical condition’. If we consider this as directive, 11 babies were eligible for regular clinical observations by NICE, while the remaining 20 infants with no risk factors were designated ‘normal care’ and could not be distinguished until symptomatic after 4 hours.

Of those babies not treated by SRC in the first 4 hours, SRC advised enhanced observations and blood cultures in three and enhanced observations alone in a further four cases by 4 hours of age.

Figure 1 shows the distribution of the prior and posterior SRS by 4 hours in four groups—all infants, infants recommended antibiotics by NICE, those by SRC and a final group where neither tool recommended antibiotics.

Figure 1

Box and whisker plot showing the distribution of prior and posterior sepsis risk scores by 4 hours in four groups: (A) all infants in the study, (B) infants advised antibiotics as per NICE, (C) infants advised antibiotics as per SRC, (D) infants advised antibiotics by neither. Each box depicts the IQR, the line within the box depicts the median, whereas the lower and upper margins depict the 25th and 75th percentiles, respectively. The whiskers extend 1.5× the IQR from the median. The circles (o) represent the outliers (1.5 to ≤3 times the IQR from the median) and the asterisk (*) plots the extreme outlier >3× the IQR from the median. NICE, National Institute for Health and Care Excellence; SRC, sepsis risk calculator.


Our study is the first to report the implications of applying SRC exclusively to a UK cohort of infants with confirmed EOS derived from a large population of 142 000 infants of gestation ≥34 weeks spanning 2008–2017.

Published studies on SRC have used various designs,28–34 usually to show the potential reduction in antibiotic use. Very few studies, including those recently published from the UK, had large enough cohorts to establish safety of SRC for a condition as infrequent as EOS.28 35 The studies that address this, including a meta-analysis that showed no increase in mortality and readmissions comparing the SRC with conventional approach, mostly derive their data from the USA.23 24

There are significant differences between the UK healthcare set-up and clinical practice in the USA. As well as universal screening for GBS in pregnancy in the USA, there are differences in hospital stay, electronic health record documentation and schedule of observations for the newborn.

Our results show that approximately 81% of babies who develop EOS present with symptoms. Within 4 hours, both tools are poor at predicting which babies will develop sepsis later, for at least two out of five babies (44.3%), this currently relies on clinical skill and vigilance rather than screening criteria. More infants destined to develop EOS are identified and treated by 4 hours of age by NICE than SRC. This absolute difference (17.1%) is due to NICE detecting 12 asymptomatic babies in this period. A recent meta-analysis of sensitivity of SRC compared with NICE in detecting 75 culture-positive EOS in the first 72 hours reported the probability of ‘missed’ cases as 0.19 (95% CI 0.11 to 0.29).36 However, the ‘missed’ cases were not necessarily missed, but represented infants where empiric antibiotics were not initiated in the asymptomatic phase following initial evaluation.36

A sensitivity analysis of an intervention without reference to its specificity fails to put the scale of the issue into perspective. Applying 16% reported use of antibiotics28 to our study cohort, approximately 22 720 infants would have received antibiotics and associated interventions. Considering a projected 50-74% reduction in antibioitic use with SRC, the trade off in our cohort for prophylacically treating 12 asymptomatic infants with EOS was the unnecessary treatment of approximately 11386-16852 infants. Transient bacteraemia without illness is recognised and may be over-represented as ‘sepsis’ in retrospective analysis. Assuming these infants would become symptomatic, they would simply add to the pool of infants with EOS currently identified only through routine clinical vigilance. The value of predictive tools is in detection of asymptomatic babies in the first few hours after birth; less so if four out of five babies have symptoms when first identified. The incidence of EOS is falling in developed countries including the UK. The risk-based approach advocated by NICE to administer antibiotics to asymptomatic babies may be outdated and increase lifelong risk.37–40 Replacing prophylactic antibiotic use with frequent clinical observations and treatment based on early symptoms may represent prudent care.

To analyse this in more detail, of the 43 babies whom SRC did not detect <4 hours (as opposed to 31 babies on NICE), SRC advised blood culture for 3, enhanced observations for 4 and ‘normal care’ for 36 infants. Similarly, NICE would have assigned 11 babies for enhanced observation and 20 to ‘normal care’. We should however appreciate that ‘normal care’ directed by SRC is significantly more enhanced and does not equate to normal care in the UK. Babies deemed ‘normal’ at birth in the UK may receive no or very few recorded observations and may go home before 12 hours of age. ‘Normal care’ in Pennsylvania hospitals23 includes vital signs every hour for 3 hours after transfer to the postpartum floor, followed by vital signs every 4 hours until 12 hours old and every 12 hours thereafter. Babies with SRS from 0.7 to 1.49 receive hourly observations for 3 hours and then four hourly until 36 hours of age (KM Puopolo, personal communication). Babies born by vaginal delivery typically go home at 36–48 hours or if born by C-section, at 72–96 hours of age. Similar systems are in place within Kaiser Permanente hospitals (A Fischer, personal communication, Permanente Medical Group). Irrespective of which tool we use, but particularly if we introduce SRC in the UK, we will also need a robust system of observation for babies.

The strength of our study is a large birth cohort, a multicentre approach with different risk profile of cases and a complete microbiology data set. However, there are limitations. As a retrospective study, the reviewers were dependent on the quality of the documentation, and not able to ascertain if undocumented concerns were present. In the absence of electronic documentation, one cannot be certain if documentation reflected the actual start and/or identification of symptoms. We were unable to study in detail the two tools after 4 hours, because of infrequent clinical observations. Figure 1 demonstrates that the prior SRS (based on risk factor alone) is low in EOS babies irrespective of the clinical tool applied in contrast to the posterior SRS following incorporation of clinical condition. Nevertheless, those infants not identified by either tool also had low posterior SRS at 4 hours and were undetected until symptomatic. This confirms that no tool should over-ride clinical judgement and if clinically suspected of sepsis, infants should undergo blood culture and prompt treatment with antibiotics.

Various units have generated their own specific SRC guidelines using a number of different approaches. Dhudasia et al 23 used an SRS threshold of 1.54/1000 to start empiric antibiotics. In our cohort, five babies had posterior SRS between 1.54 and 3. The recommendations were ‘consider antibiotics’ in three and ‘blood culture and 4 hourly vitals’ in the remaining two. If pragmatically, we treat both these groups, our limited data indicate no need to lower the threshold for empiric antibiotics from 3/1000.

We believe that a quality improvement activity on antibiotic stewardship must be introduced cautiously balancing the many benefits of reducing antibiotics and hospital stay with unintended consequence of missed sepsis. Based on the findings of Goel et al 28 and our analysis of the data in this paper, we have proposed the following cautious approach of implementing SRC in Wales, and this is currently under evaluation. If, shortly after birth, no NICE red flag or risk factor is present, and the infant is well, then normal care is assigned (no change from current practice). Any infant with a NICE risk factor or red flag is referred to the neonatal team for evaluation using SRC. Within this cohort, asymptomatic, low-risk infants with SRS <0.65 are observed in hospital for at least 24 hours while antibiotics are withheld. Observations are hourly for the first 2 hours, then two hourly until 12 hours, then four hourly. The observation period extends for medium-risk asymptomatic infants (SRS 0.65–1.54) or those with low risk who developed transient symptoms to 36 hours. The threshold for prophylactic antibiotics is >1.54 (Philadelphia adaptation), in contrast to the original SRC threshold >3. A blood culture entails taking at least 1 mL blood41 42 to increase the sensitivity of early culture positivity. All babies recommended for blood culture alone should be commenced antibiotics, administered within 1 hour of intention to treat. In addition, we recommend prophylactic treatment for infants whose previous sibling had a history of culture proven invasive GBS sepsis and their mother had not received adequate intrapartum prophylaxis. Each family receives an information leaflet raising awareness of symptoms and outlining the reasons for observation or antibiotic treatment. The above approach will ensure enhanced observations replace antibiotic treatment for most infants currently treated. This new approach will not change the management of babies without recognisable NICE red flags or risk factors, who are indistinguishable from normal babies until they become symptomatic. The outcome of this approach may inform future guidelines on management of EOS in neonates.


While both tools were poor in identifying EOS within 4 hours, NICE was superior to SRC in identifying asymptomatic cases. Currently four out of five EOS have symptoms when identified, and the majority present within 24 hours. The introduction of SRC in the UK has the potential to reduce antibiotic use. In newborn infants with risk factors, high-quality observations up to 24–36 hours, stratified by risk, with improved staff and parent awareness of EOS should accompany SRC use. This proposal needs further evaluation in large prospective studies.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Contributors Study design and planning, application for whether ethics approval was necessary: JM, SJ. Data collection (maternal and neonatal data): JM, RM, SJ, GT, CW, AC, HW, HH, JR, IB, AB. Statistical analysis and interpretation: JM, SB, RM. Composition of manuscript: JM, RM, SB, SJ. Final manuscript was reviewed by all authors. JM is the overall guarantor for the content of this article.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

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

  • Data availability statement Data are available upon reasonable request. Individual participant data that underlie the results reported in this article, after deidentification, are available immediately following publication to researchers who provide a methodologically sound proposal. Proposals should be directed to the corresponding author RM at