Pediatrics and Neonatology 65 (2024) 71e75

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: http://www.pediatr-neonatol.com

Original Article

Neonatal early-onset sepsis calculator:

Impact on antibiotic use in a level II neonatal
unit in Western Australia
John Gannon a,*, Tobias Strunk b, Noel Friesen a, Chia Saw a

a
Department of Paediatrics, SJOG Midland Hospital, Australia
b
King Edward Memorial Hospital, Australia

Received Nov 23, 2022; received in revised form Mar 28, 2023; accepted Apr 20, 2023
Available online 24 August 2023

Key Words Abstract Background: Overuse of empirical intravenous antibiotics in neonates in high-
antibiotic income countries (HICs) is well documented. The Kaiser Permanente neonatal early-onset
stewardship; sepsis (EOS) calculator is an evidence-based sepsis risk assessment tool that has demonstrated
early-onset sepsis potential to reduce antibiotic usage in this population. The incidence of early-onset sepsis in
calculator; most HICs is 0.4e0.8 per 1000 live births. The objective was to evaluate the calculator’s impact
Kaiser permanente; on antibiotic rates and length of stay in a regional level II Special Care Nursery.
neonatal infection; Methods: A single-centre retrospective cohort study compared antibiotic administration rates
neonatal sepsis in the first 72 h in neonates
35 weeks gestation born during two 6-month periods in 2019 (pre-
EOS calculator) and 2021 (post-EOS calculator). Electronic and paper case records were ac-
cessed to capture data. Continuous data were summarised using mean and standard deviation,
and categorical data were summarized using frequency distributions. There were 951 (2019)
and 1129 (2021) infants born during the study periods.
Results: Following implementation of the calculator, antibiotic exposure decreased from
13.7% to 4.7% of all neonates without reported negative outcomes. Mean length of stay for ne-
onates born across the two periods decreased from 2.38 to 2.13 days. Indications for antibiotic
use shifted more towards clinical condition and away from obstetric risk factors. There were
no culture-proven cases of sepsis or readmissions with EOS in either period.
Conclusion: Implementation of the EOS calculator significantly reduced exposure to antibi-
otics, without adverse outcomes.
Copyright ª 2023, Taiwan Pediatric Association. Published by Elsevier Taiwan LLC. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/
by-nc-nd/4.0/).

* Corresponding author.
E-mail address: john.gannon@childrenshealthireland.ie (J. Gannon).

https://doi.org/10.1016/j.pedneo.2023.04.010
1875-9572/Copyright ª 2023, Taiwan Pediatric Association. Published by Elsevier Taiwan LLC. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 J. Gannon, T. Strunk, N. Friesen et al.

1. Introduction clinical condition of the neonate, the algorithm generates a

patient-specific estimate of EOS risk and suggests clinical
management.
1.1. Background A previous retrospective observational study examining
births in this center identified that the proportion of neo-
The World Health Organization 2021 report estimated nates receiving antibiotics for suspected EOS could poten-
between 1.3 million and 3.9 million cases of neonatal tially be reduced substantially from 13.4% to 3.9%, if the
sepsis annually, leading to 400,000e700,000 deaths.1 It calculator were utilized.21 The calculator was introduced to
projects that deaths could be reduced by 84% through our level II special care nursery (SCN) on 1st February 2021
adequate use of intrapartum antibiotics and early effec- with the aim of safely reducing use of IV antibiotics in the
tive diagnosis and treatment of sepsis.1,2 Early-onset neonatal population. Neonatal staff had undergone EOS
sepsis (EOS) is defined as infection occurring in the first calculator training prior to this date to ensure compliance
72 h of life.1 The most commonly used EOS diagnosis in- and correct usage. This retrospective cohort study was
volves a symptomatic neonate with a positive blood or later carried out to evaluate if previously projected re-
cerebrospinal fluid culture.3 A meta-analysis by Fleisch- ductions in antibiotic use had occurred following the cal-
mann et al. estimated EOS mortality to be 17.6%.2 Group B culator’s introduction.
Streptococcus (GBS) is the most common causative agent,
and Escherichia coli is the most common cause of 1.2. Objectives
mortality.4 Vertical transmission contributes the
majority of infections.4 Risk factors for EOS include pre-
The primary objective was to compare rates of antibiotic
maturity, maternal GBS colonization, chorioamnionitis,
use and length of hospital admission for all live births
35
prolonged rupture of membranes and inadequate
weeks gestation between two 6-month periods before and
intrapartum antibiotics.5 Manifestations of EOS can vary
after the calculator’s introduction. Secondary objectives
from temperature instability, irritability and lethargy, to
were to compare mean length of stay in hospital and SCN
severe respiratory distress, meningitis and shock.6 Symp-
for patients treated with antibiotics, and to compare
toms may be non-specific in the early stages of illness and
readmissions with EOS and mortality across the two
there is no biomarker to definitively rule in sepsis; thus
periods.
definitive diagnosis remains challenging.3,7 Clinical suspi-
cion often leads to empirical antibiotic therapy in unin-
fected neonates, and tendency towards overtreatment 2. Methods
due to the potentially severe consequences of missed
diagnosis.5
2.1. Study design and setting
With increased GBS screening in pregnancy and use of
intrapartum antibiotics, particularly in high-income coun-
This is a single-centre retrospective pre/post cohort study
tries (HICs), rates of neonatal sepsis have declined signifi-
in a regional hospital over two separate 6-month periods.
cantly in recent decades.8 Incidence of culture-proven EOS
Local incidence of EOS in Western Australia is estimated to
in most HICs is 0.4e0.8 per 1000 live births.9e11 Simonsen
be 0.5/1000 live births.21 Period 1 births were between 1st
et al. reported in 2014 that while incidence in the United
February and 31st July 2019. This served as the baseline
States was between 0.77 and 1 per 1000 live births, neo-
period, with neonatal sepsis risk algorithm based on a local
nates were being investigated for sepsis at rates of 7e13%.4
adaptation of the American Academy of Pediatrics guide-
Intravenous (IV) antibiotics administration rates for neo-
lines.22 Period 2 births were from 1st February e 31st July
nates have been documented at over 100 times higher than
2021, following the introduction of the EOS calculator to
the actual incidence of EOS.12e14
the center. Periods of 6 months were chosen based on an
Antibiotic over-administration poses risks of drug
expected sample size of around 1000 births in each.
toxicity and bacterial resistance and is associated with a
Exclusion criteria were <35 weeks gestational age (GA),
range of complications. Immediate risks include invasive
critically unwell neonates requiring transfer to tertiary
candidiasis (due to selective pressure), and long-term risks
units, and indications for antibiotic therapy outside of
include immune-mediated conditions (due to alterations in
manifestations of EOS. Electronic and paper case records
host microbiome).15e18 Other consequences include sepa-
for mother and baby were accessed to capture required
ration of baby from parents, difficulties in establishing
data points. Personal information data were de-identified.
breastfeeding, painful blood tests and cannulations,
This paper is reported using the STROBE (Strengthening
greater likelihood of invasive procedures and intensive care
the Reporting of Observational Studies in Epidemiology)
unit admission, and increased time spent in hospital with
guideline.23 Ethics approval was granted by the St. John of
higher healthcare costs.19,20
God Health Care Human Research Ethics Committee
The Kaiser Permanente neonatal EOS calculator is a free
(reference number 1690).
online multivariate tool designed to guide decision-making
relating to testing and treatment of neonates for EOS,
based on objective data available at the time of birth. The 2.2. Statistical methods
physician inputs five risk factors to the calculator (gesta-
tional age, highest maternal temperature, duration of Continuous data were summarized using mean, range and
rupture of membranes in hours, maternal GBS status, and standard deviation. Categorical data were summarized
use of intrapartum antibiotics). Also taking into account the using frequency distributions. Data on hospital and SCN

72
 Pediatrics and Neonatology 65 (2024) 71e75

length of stay were extracted from coding department 4. Results are summarized in Table 1
records. Comparisons of antibiotic usage and length of
stay between the study periods were analyzed using 4.1. Discussion
Fisher Exact test and student’s unpaired t-test.
The EOS calculator was developed in 2016 and several
studies have highlighted its contribution to a decline in
3. Results antibiotic administration in the neonatal population.14,24e27
Our study demonstrated a reduction in antibiotic treatment
from 13.7% to 4.7% for neonates
35 weeks GA born in the
3.1. Pre-EOS calculator (period 1) center (OR 0.311, 95% CI: 0.223 e 0.435, p < 0.00001).
There were no culture-proven cases of sepsis and no re-
There were 951 live births
35 weeks. According to the ported adverse outcomes in any of the 183 neonates on
conventional EOS guidelines, 130 (13.7%) received IV anti- antibiotics across both periods. These findings are consis-
biotics in the first 72 h. Eight patients were excluded; seven tent with predicted effects based on our previous retro-
because of transfer to tertiary center, and one due to spective study21 and with another Western Australian
alternative indication for treatment (lower limb swelling). observational study from a tertiary neonatal unit.14
113 neonates were treated in the first 24 h, 14 on day two Mean length of stay for all neonates
35 weeks born
and three on day three. Out of 130 neonates, 59 (45.4%) across the two periods reduced from 2.38 days to 2.13 days
were admitted to SCN and 71 were observed in SCN for 6 h following implementation of the EOS calculator, in line with
only. Main indication for treatment was maternal risk fac- other studies.20,28,29 Achten et al. for example, demon-
tor(s) in 41 cases (31.5%), and clinical condition in 82 cases strated an overall reduction in mean length of stay from 3.48
(63.1%). Mean length of stay for all live births was 2.38 days to 3.27 days across two pre-post cohorts.28 Reduced
days. For neonates on IV antibiotics, mean hospital length length of stay can lead to lower healthcare costs and fewer
of stay was 4.40 days and mean SCN length of stay was 2.12 hospital-related complications and adverse events.6
days. Mean length of stay for the neonates treated with anti-
biotics increased from Period 1 to Period 2, from 4.40 to
5.77 days (hospital), and from 2.12 to 3.83 days (SCN). This
3.2. Post-EOS calculator (period 2) increase is likely due to fewer ‘clinically well’ neonates
being brought to SCN and treated with antibiotics based on
There were 1129 live births
35 weeks. Using the EOS maternal risk factor(s), after the introduction of the
calculator, 53 (4.7%) received IV antibiotics in the first 72 h, calculator. As per hospital policy in this center, all neonates
with all therapy initiated within the first 24 h. This excludes started on antibiotics are monitored for a minimum of 6 h in
12 neonates for transfer to tertiary center. Out of 53 neo- SCN, then transferred to postnatal ward or kept in SCN
nates, 39 (73.6%) were admitted to SCN and 14 were depending on clinical condition. Only 59/130 (45.4%) in
observed for 6 h only. Main indication for treatment was Period 1 were admitted to SCN following 6 h of observation,
maternal risk factor(s) in 10 cases (18.9%) and clinical whereas in Period 2 there were 39/53 (73.6%) who
condition in 42 cases (79.2%). Mean length of stay for all live remained in SCN requiring extra support. In analyzing the
births was 2.13 days. For neonates on IV antibiotics, mean data, those monitored for 6 h only were counted as 1 day of
hospital length of stay was 5.77 days and mean SCN length SCN admission and, as there was a higher of proportion of
of stay was 3.83 days. these cases in Period 1, this resulted in a shorter mean
There were no cases of culture-proven cases of sepsis or length of stay for neonates on antibiotics in this period.
deaths in either period. There were no readmissions with This comparison shows that prior to the calculator’s intro-
missed cases of EOS in either period. duction, a higher proportion of neonates were treated for
suspected EOS, who required no extra support after
observation and were clinically well enough to return to the
3.3. Primary outcomes ward. A longer average SCN length of stay for Period 2
supports the position that the calculator is identifying more
Rate of antibiotic use across the two periods demonstrated a ‘clinically unwell’ neonates who would receive longer IV
reduction from 13.7% to 4.7% (OR 0.311, 95% CI: 0.223 e antibiotic courses and SCN admissions.
0.435, p < 0.00001). Mean length of stay for all live This change is also evidenced by the main indications for
births across the two periods decreased from 2.38 days to treatment as documented in patient notes. Neonates in
2.13 days. Period 1 were more likely than those in Period 2 to have
maternal risk factor(s) as the main reason for antibiotics
(31.5% vs. 18.9%). By far the most common indication for
3.4. Secondary outcomes this was maternal fever. Conversely, neonates in Period 2
were more often treated for clinical condition (e.g., res-
For neonates treated with antibiotics, mean hospital length piratory distress) than their counterparts in Period 1 (79.2%
of stay increased from 4.40 to 5.77 days, and mean SCN vs. 63.1%).
length of stay increased from 2.12 to 3.83 days. There were The EOS calculator offers an evidence-based individual
no deaths or readmissions with EOS in either period. patient risk estimate to help guide physicians in managing

73
 J. Gannon, T. Strunk, N. Friesen et al.

Table 1 Summary of study results.

Period 1 (2019) Period 2 (2021) Statistical analysis
Total live births
35 weeks 951 1129
Hospital length of stay (days) Mean 2.38 Mean 2.13 Student’s unpaired t-test
Range 0e14 Range 0e23 p-value 0.0008
Culture-proven sepsis 0 0
Readmissions with EOS 0 0
Deaths 0 0
Antibiotics within 72 h 130 (13.7%) 53 (4.7%) Fisher exact test
p-value <0.00001
Mean birth weight 3390 g [SD 466] 3251 g [SD 619]
Mean gestational age 38 weeks þ 6 days 38 weeks þ 4 days
SCN admission status Fisher exact test
p-value 0.0006
SCN admission 59 (45.4%) 39 (73.6%)
SCN observation for 6 h 71 (54.6%) 14 (26.4%)
Age at initiation of antibiotic
treatment
<24 h 113 53
24e48 h 14 0
>48 h 3 0
Length of stay for neonates treated for suspected EOS
Hospital length of stay (days) Mean 4.40 Mean 5.77 Student’s unpaired t-test
Range 1e14 Range 1e23 p-value 0.0057
SCN length of stay (days) Mean 2.12 Mean 3.83 Student’s unpaired t-test
Range 1e11 Range 1e15 p-value 0.0003
Main indication for treating as EOS N [ 130 N [ 53
Risk factors 41 (31.5%) 10 (18.9%)
Maternal fever 39 8
GBS with inadequate antibiotics 0 2
Maternal raised CRP 1 0
Meconium-stained liquor 1 0
Clinical condition 82 (63.1%) 42 (79.2%)
Respiratory distress 30 28
Oxygen desaturation 9 6
Hypoglycaemia 19 4
Temperature instability 18 2
Omphalitis 0 1
Metabolic acidosis/cord acidosis 4 1
Non-blanching rash 1 0
Hypotonia 1 0
Unspecified 7 (5.4%) 1 (1.9%)
CRP Z C-reactive protein; EOS Z early onset sepsis; GBS Z Group B streptococcus; SCN Z special care nursery.

neonates with suspected sepsis. This study has demon- neonates who received antibiotics in the first 72 h of life.
strated a reduction in antibiotic rates from 13.7% to 4.7% The retrospective analysis only included neonates who
with no deaths or adverse outcomes, representing an could be cared for in our level II SCN, excluding critically
improvement in antimicrobial stewardship. Mean length of unwell neonates who required transfer to tertiary centers.
stay decreased from 2.38 days to 2.13 days across the two Data are not available on outcomes of neonates who were
6-month periods, which can lead to reduced nosocomial transferred to tertiary centers, whether treated or un-
complications and healthcare costs. Mean hospital and SCN treated for suspected EOS.
length of stay increased for the neonates on IV antibiotics
from Period 1 to Period 2, with greater numbers treated for 5. Conclusion
clinical condition rather than risk factors.
Implementation of the neonatal EOS calculator reduces the
4.2. Limitations rate of antibiotic overexposure in neonates. The authors
recommend that other centers consider adding the EOS
This study examined a small sample size, with 2080 live calculator to clinical practice in neonatal units. Prior to
births across both periods. Analysis was restricted to introducing the tool, it is advisable to conduct a

74
 Pediatrics and Neonatology 65 (2024) 71e75

prospective implementation study to evaluate potential 15. Cotten CM. Adverse consequences of neonatal antibiotic
impact. Further research is recommended to reduce anti- exposure. Curr Opin Pediatr 2016;28:141e9.
biotic exposure and duration in culture-negative neonates. 16. Shaw SY, Blanchard JF, Bernstein CN. Association between the
use of antibiotics in the first year of life and pediatric in-
flammatory bowel disease. Am J Gastroenterol 2010;105:
Declaration of competing interest 2687e92.
17. Wernroth ML, Fall K, Svennblad B, Ludvigsson JF, Sjölander A,
The authors have no conflicts of interest relevant to this Almqvist C, et al. Early childhood antibiotic treatment for
article and no funding was received for conducting this otitis media and other respiratory tract infections is associated
study. with risk of type 1 diabetes: a nationwide register-based study
with sibling analysis. Diabetes Care 2020;43:991e9.
18. Ben-Ami R, Olshtain-Pops K, Krieger M, Oren I, Bishara J,
References Dan M, et al. Antibiotic exposure as a risk factor for flucona-
zole-resistant Candida bloodstream infection. Antimicrob
1. Global report on the epidemiology and burden of sepsis: cur- Agents Chemother 2012;56:2518e23.
rent evidence, identifying gaps and future directions, 2020, 19. Benaim EH, Upadhyay K, Talati AJ. Comparison of institutional
World Health Organization; Geneva, Available at https://apps. guidelines with established early onset sepsis risk calculator in
who.int/iris/handle/10665/334216. Assessed March 28, 2023. reducing antibiotic use in an inner-city NICU in US. J Glob
2. Fleischmann C, Reichert F, Cassini A, Horner R, Harder T, Antimicrob Resist 2020;21:124e9.
Markwart R, et al. Global incidence and mortality of neonatal 20. Bridges M, Pesek E, McRae M, Chabra S. Use of an early
sepsis: a systematic review and meta-analysis. Arch Dis Child onset-sepsis calculator to decrease unnecessary NICU admis-
2021;106:745e52. sions and increase exclusive breastfeeding. J Obstet Gynecol
3. Hofer N, Zacharias E, Müller W, Resch B. An update on the use Neonatal Nurs 2019;48:372e82.
of C-reactive protein in early-onset neonatal sepsis: current 21. Saw C, Kulasekaran K, Fernando DT, Tan D, Dharmapuri R,
insights and new tasks. Neonatology 2012;102:25e36. Bulsara MK, Friesen ND. Retrospective cohort study of neonatal
4. Simonsen KA, Anderson-Berry AL, Delair SF, Davies HD. Early- early onset of sepsis and the role of the EOS calculator in a
onset neonatal sepsis. Clin Microbiol Rev 2014;27:21e47. level II nursery. Pediatr Neonatol 2021;62:512e21.
5. Good PI, Hooven TA. Evaluating newborns at risk for early- 22. Brady MT, Polin RA. Prevention and management of infants
onset sepsis. Pediatr Clin North Am 2019;66:321e31. with suspected or proven neonatal sepsis. Pediatrics 2013;132:
6. Deshmukh M, Mehta S, Patole S. Sepsis calculator for neonatal 166e8.
early onset sepsis - a systematic review and meta-analysis. J 23. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC,
Matern Fetal Neonatal Med 2021;34:1832e40. Vandenbroucke JP. The Strengthening the Reporting of
7. Shah BA, Padbury JF. Neonatal sepsis: an old problem with new Observational Studies in Epidemiology (STROBE) statement:
insights. Virulence 2014;5:170e8. guidelines for reporting observational studies. Lancet 2007;
8. Bizzarro MJ, Raskind C, Baltimore RS, Gallagher PG. Seventy- 370:1453e7.
five years of neonatal sepsis at Yale: 1928e2003. Pediatrics 24. Huseynova R, Bin Mahmoud L, Hamad Aljobair F, Huseynov O,
2005;116:595e602. Career H, Jaganathan PP, et al. Use of early-onset sepsis risk
9. Stoll BJ, Hansen NI, Sánchez PJ, Faix RG, Poindexter BB, Van calculator for neonates
34 weeks in a large tertiary neonatal
Meurs KP, et al. Early onset neonatal sepsis: the burden of centre, Saudi Arabia. Cureus 2021;13:e14620.
group B Streptococcal and E. coli disease continues. Pediatrics 25. Akangire G, Simpson E, Weiner J, Noel-MacDonnell J,
2011;127:817e26. Petrikin J, Sheehan M. Implementation of the neonatal
10. Schrag SJ, Farley MM, Petit S, Reingold A, Weston EJ, Pondo T, sepsis calculator in early-onset sepsis and maternal cho-
et al. Epidemiology of invasive early-onset neonatal sepsis, rioamnionitis. Adv Neonatal Care 2020;20:25e32.
2005 to 2014. Pediatrics 2016;138:e20162013. 26. Achten NB, Klingenberg C, Benitz WE, Stocker M,
11. Klingenberg C, Kornelisse RF, Buonocore G, Maier RF, Schlapbach LJ, Giannoni E, et al. Association of use of the
Stocker M. Culture-negative early-onset neonatal sepsis - at neonatal early-onset sepsis calculator with reduction in anti-
the crossroad between efficient sepsis care and antimicrobial biotic therapy and safety: a systematic review and meta-
stewardship. Front Pediatr 2018;6:285. analysis. JAMA Pediatr 2019;173:1032e40.
12. Fjalstad JW, Stensvold HJ, Bergseng H, Simonsen GS, 27. Kuzniewicz MW, Puopolo KM, Fischer A, Walsh EM, Li S,
Salvesen B, Rønnestad AE, et al. Early-onset sepsis and anti- Newman TB, et al. A quantitative, risk-based approach to t
biotic exposure in term infants: a nationwide population-based he management of neonatal early-onset sepsis. JAMA Pediatr
study in Norway. Pediatr Infect Dis J 2016;35:1e6. 2017;171:365e71.
13. Kuzniewicz MW, Walsh EM, Li S, Fischer A, Escobar GJ. Devel- 28. Achten NB, Visser DH, Tromp E, Groot W, van Goudoever JB,
opment and implementation of an early-onset sepsis calculator Plötz FB. Early onset sepsis calculator implementation is
to guide antibiotic management in late preterm and term ne- associated with reduced healthcare utilization and financial
onates. Jt Comm J Qual Patient Saf 2016;42:232e9. costs in late preterm and term newborns. Eur J Pediatr 2020;
14. Strunk T, Buchiboyina A, Sharp M, Nathan E, Doherty D, 179:727e34.
Patole S. Implementation of the neonatal sepsis calculator in 29. Cuoco MP, Parbhoo D, D’Aprano A. Introduction of the neonatal
an Australian tertiary perinatal centre. Neonatology 2018;113: sepsis calculator at a low-dependency special care nursery in
379e82. Australia. J Matern Fetal Neonatal Med 2022;35:7532e5.

75