![](https://assets.reachrx.ai/images/10.1038%E2%98%8Ds41390-025-04707-5/_page_0_Picture_1.jpeg) ![](https://assets.reachrx.ai/images/10.1038%E2%98%8Ds41390-025-04707-5/_page_0_Picture_2.jpeg) # CLINICAL RESEARCH ARTICLE OPE ![](https://assets.reachrx.ai/images/10.1038%E2%98%8Ds41390-025-04707-5/_page_0_Figure_4.jpeg) # Antibiotic exposure in culture-negative preterm infants: a 10-year single-centre study Cheryl Anne Mackay (1)^{1,2 ™}, Elizabeth A. Nathan³, Michelle Claire Porter⁴, Damber Shrestha¹, Rolland Kohan^1,2 and Tobias Strunk^1,2,5 © The Author(s) 2026 **BACKGROUND:** Antibiotic exposure in neonatal intensive care units (NICU) is high. This study describes antibiotic use in very preterm infants and examines the association between duration of exposure and outcomes in blood culture negative (CN) infants. **METHODS:** Infants <32 weeks' gestation admitted between January 2012 and June 2022 were included in this retrospective cohort study. Data were extracted from electronic databases. Antibiotic exposure was calculated as duration of treatment (DOT) and antibiotic utilisation rate (AUR) and compared with neonatal outcomes including mortality, late onset sepsis (LOS), necrotising enterocolitis (NEC), chronic lung disease (CLD), severe retinopathy of prematurity (ROP) and/or severe brain injury. **RESULTS:** There were 3235 CN infants included in the study; 1601 (49.5%) received antibiotics for $\leq$ 2 days of which 266 (8.2%) received no antibiotics; 841 (26.0%) received antibiotics for $\geq$ 5 days. DOT decreased from 78.0 to 61.9 per 1000 and AUR from 0.07 (IQR 0.04–0.11) to 0.05 (IQR 0.03–0.10) from 2012 to 2022. Higher AUR and/or prolonged antibiotic exposure was associated with increased mortality, brain injury, NEC, ROP, LOS, and CLD. **CONCLUSION:** Antibiotics are critical for infants with sepsis but can cause harm in those without. Strategies to reduce antibiotic exposure are needed to improve preterm infant outcomes. Pediatric Research; https://doi.org/10.1038/s41390-025-04707-5 ## IMPACT. - Prolonged antibiotic exposure is common in culture-negative, very preterm infants. - Although antibiotics are critical for infants with culture-positive sepsis, they can cause harm in those who are culture-negative. - This study adds to the small pool of evidence examining antibiotic use and its association with increased morbidity and mortality in very preterm infants. - The study findings will impact antibiotic prescribing practices significantly and result in strategies to reduce antibiotic exposure in these at-risk infants. #### INTRODUCTION Antibiotics are the most commonly prescribed medications in neonatal intensive care units (NICU) with approximately 60% of infants in NICUs receiving antibiotics in the first three days of admission.^1,2 Empiric antibiotic treatment is particularly high in preterm infants and is inversely proportional to gestational age. Approximately 3% of infants ≥ 34 weeks' gestation receive antibiotics during admission, compared with ~80% of those < 32 weeks' and ~98% of those < 28 weeks'.^3-5 Often, prolonged courses of antibiotics are administered despite negative cultures.⁵ Prompt antibiotic therapy is appropriate for infants at increased sepsis risk, and for culture-confirmed sepsis appropriate agent choice and treatment duration are essential. However, antibiotic therapy confers no benefit in culture-negative (CN) infants but has been associated with harm beyond driving antimicrobial resistance. Early and/or prolonged antibiotic exposure disrupts the neonatal microbiome at a critical phase of immune development and is associated with worse outcomes in very preterm infants (death, late-onset sepsis (LOS), necrotizing enterocolitis (NEC), chronic lung disease (CLD) and poor growth).^6-9 It also has longer-term associations with metabolic, immune-mediated and inflammatory conditions in childhood.^7,10 The higher risk of systemic infection in very preterm infants, predominantly due to distinct qualities of the immature innate and adaptive immune systems, invasive procedures they require and prolonged hospital stays, is associated with significant morbidity and mortality and may drive initiation and/or prolonged antibiotic therapy. Against this background, we examined a large single-center cohort of inborn infants < 32 weeks' gestation to describe temporal trends in antibiotic utilization (including duration of therapy) and to evaluate the association between antibiotic exposure and mortality and major neonatal morbidities ¹Neonatology, Child and Adolescent Health Service, King Edward Memorial Hospital, Subiaco, WA, Australia. ²Division of Paediatrics, School of Medicine, University of Western Australia, Crawley, WA, Australia. ³Division of Obstetrics and Gynaecology, School of Medicine, University of Western Australia, Crawley, WA, Australia. ⁴Microbiology Department, PathWest Laboratory Services, Nedlands, WA, Australia. ⁵Wesfarmers' Centre for Vaccines and Infectious Diseases, The Kids Research Institute, Nedlands, WA, Australia. ^©email: cherylanne.mackay@uwa.edu.au Received: 2 June 2025 Revised: 28 September 2025 Accepted: 12 November 2025 Published online: 10 January 2026 (LOS, NEC, CLD, severe retinopathy of prematurity (ROP) and/or severe brain injury) to better inform antibiotic prescribing practices and optimize infant outcomes. Contemporary data addressing these questions are scarce globally and, to our knowledge, unknown in Australian cohorts. #### **METHODS** ## **Study Design and Participants** A retrospective cohort study was conducted at King Edward Memorial Hospital (KEMH) in Perth. All infants < 32 weeks gestation admitted to the KEMH NICU between January 1, 2012, and June 30, 2022, were included from birth to discharge or transfer out of the unit, which could occur up to 44 weeks' corrected for gestational age (CGA). Infants with major congenital anomalies were excluded due to higher sepsis risk and antibiotic exposure. Infants not born at KEMH and those with blood culture-positive early-onset sepsis (EOS) were excluded from the outset. Those with blood culture-positive LOS, NEC, and/or SIP were excluded from the date of diagnosis, as they would justifiably have been commenced on antibiotics as per best practice. All infants with culture-positive cerebrospinal fluid (CSF) had a positive blood culture with the same organism and were only included once to avoid duplication. Infants with missing data were also excluded. Maternal, neonatal, and follow-up data were retrieved from a unit-specific electronic neonatal database using Microsoft Access (2010)®. The study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology for Newborn Infection (STROBE-NI) guideline. Is It was approved by the Child and Adolescent Health Service (CAHS) Governance, Evidence, Knowledge and Outcomes (GEKO) Triage Committee (000471). ## **Study Setting** KEMH is the only tertiary perinatal centre in Western Australia (WA) and covers a geographical area of more than 2.5 million $\rm km^2$ with a total population of 2.4 M–2.8 M during the study period. There were on average 5772 births and 2228 neonatal admissions annually during the study period. The average annual stillbirth, preterm birth and neonatal mortality rates in Western Australia were 6.6/1000 births, $\sim\!9\%$ and 1.6/1000 livebirths, and facility-specific (KEMH) rates were 2.5/1000 births, $\sim\!12\%$ and 5.1/1000 livebirths respectively at the time. KEMH is a public hospital that provides comprehensive obstetric care and routinely offers resuscitation and intensive care for infants from 23 weeks' gestation onwards. Antenatal antibiotics are administered for prevention of neonatal *Streptococcus agalactiae* infection using screening-and risk-based criteria, preterm and/or prolonged rupture of membranes >18 h, and clinical evidence of chorioamnionitis. Recommendations for blood collection for culture include a minimum 0.5 ml in infants <28 weeks' gestation and 1 ml in infants ≥28 weeks'. Empiric treatment for sepsis in the neonate includes penicillin and gentamycin for EOS and vancomycin and gentamycin for LOS. Quality improvement strategies, such as early introduction and faster advancement of enteral feeds, topical coconut oil for skin integrity, decreased central lines, and infection prevention and control measures, were introduced during the study period but there were no significant changes in diagnostic and treatment practices. ## **Definitions of Variables** Data was collected and defined according to the Australia and New Zealand Neonatal Network (ANZNN) guidelines.¹⁴ Study variables were defined as follows: - (i) Preeclampsia was defined according to the Society of Obstetric Medicine of Australia and New Zealand guideline for hypertension in pregnancy at the time.¹⁵ - (ii) Gestational diabetes was defined according to the Australasian Diabetes in Pregnancy Society's Consensus Guideline at the time. ¹⁶ - (iii) Chorioamnionitis diagnosed based on clinical features of maternal fever (≥ 38 °C), clinical signs such as uterine tenderness and/or purulent amniotic fluid, and/or laboratory evidence of infection. Not all patients had histological evidence of chorioamnionitis. - (iv) Antenatal antibiotics included any systemic antibiotics administered to the mother within 48 h of birth. - (v) Antenatal steroids were considered given if at least one dose was given at least 24 h before birth. - (vi) GA was assessed on maternal history and obstetric ultrasound at <20 weeks gestation or, in the absence of these, by clinical assessment as per the Dubowitz and Ballard score.¹⁷ - (vii) Neonatal antibiotics included only intravenous antibiotics administered during the neonatal admission. - (viii) Chronic lung disease (CLD) was defined as supplemental oxygen requirement at 36 weeks corrected gestational age (CGA)¹⁴ - (ix) NEC was diagnosed using Bell's criteria and only NEC ≥ Stage 2 included.¹⁸ - (x) Neonatal sepsis was defined by positive blood or CSF culture with a bacterial or fungal pathogen AND antibiotic treatment ≥5 days. EOS was defined as sepsis occurring at <48 h of life and lateonset sepsis LOS as that occurring ≥48 h after birth.¹⁴ Typically non-pathogenic organisms (contaminants) were pre-specified and included Coagulase negative staphylococci (CONS), Streptococcus viridans, Streptococcus bovis, Micrococcus species, Corynebacterium and Bacillus cereus. These were only considered pathogenic if the patient had CRP > 20 mg/L within 48 hours of the positive blood culture AND received antibiotic treatment for ≥5 days.¹⁴ - (xi) Severe retinopathy of prematurity (ROP) included stages 3 to 5; aggressive ROP characterised by posterior location, prominent neovascularisation and plus disease, and rapid progression of neovascularisation; and/or ROP requiring treatment.¹⁴ - (xii) Severe brain injury was defined as Grade 3 or 4 intraventricular haemorrhage (IVH) as defined by Papile or cystic periventricular leukomalacia (PVL) detected on cranial ultrasound or MRI brain prior to discharge. 19,20 #### **Outcome measures** Antibiotic exposure was counted in calendar days regardless of the number of individual agents and doses given on that day. Exposure was described as none, ≤ 2 days, 3 to 4 days, and prolonged (≥ 5 days). Consecutive and non-consecutive days of antibiotic exposure were considered in the analysis to differentiate the impact of short courses of empiric antibiotic treatment for possible sepsis, discontinued early in response to negative cultures, compared to prolonged consecutive courses of antibiotics in CN infants. The antibiotic utilisation rate (AUR) was defined as the number of days an infant was exposed to 1 or more antimicrobial agents divided by the total length of hospital stay in days. Duration of treatment (DOT) was calculated as total antibiotic days per 1000 inpatient days. Antimicrobial exposure as prophylaxis for fungal or urinary tract infections was not included Neonatal outcomes included pre-discharge mortality and/or major neonatal morbidity including LOS, NEC ≥ stage 2, CLD, severe ROP and/or severe brain injury until discharge from the neonatal unit. ### Statistical analysis Maternal and neonatal characteristics were compared between infants with antibiotic exposure $\leq 2$ days, 3-4 days, and $\geq 5$ days. The prevalence of outcomes for infants without any antibiotic exposure was low and these infants were combined with the $\leq 2$ d exposure reference group for analyses. One-way ANOVA or Kruskal-Wallis tests were used for parametric and non-parametric continuous outcomes, and Chi-square tests for categorical outcomes. Durations of ventilation data, central lines and length of stay were summarised using Kaplan-Meier survival estimates and compared between groups using log-rank tests. Deaths were censored for survival analyses. Temporal trends associated with annual AUR over the study period were analysed using negative binomial regression due to overdispersion in the data and reported as incidence rate ratios (IRR) and 95% confidence intervals. Associations between AUR, or prolonged antibiotic exposure, and binary neonatal outcomes were analysed using logistic regression and reported as odds ratios and 95% confidence intervals. Length of hospital stay was analysed using Cox proportional hazards modelling with deaths censored in analysis and reported as hazard ratios and 95% confidence intervals. The effect of consecutive compared with non-consecutive antibiotic exposure and neonatal outcomes was analysed. All models were adjusted for gestational age, sex, birthweight z-score and year of birth as these were considered potential confounding factors for neonatal outcomes. All tests were two-sided and *p*-values < 0.05 were considered significant. Stata version 17 (College Station, TX; StatCorp LLC) was used for all analyses. SPRINGER NATURE Pediatric Research ![](https://assets.reachrx.ai/images/10.1038%E2%98%8Ds41390-025-04707-5/_page_2_Figure_2.jpeg) Fig. 1 Description of case inclusion in the study. Flow chart describing neonates <32 weeks' gestation who were included in the study and those with major congenital anomalies, early-onset sepsis (EOS), missing data, late-onset sepsis (LOS), necrotising enterocolitis (NEC) and spontaneous intestinal perforation (SIP) who were not included. # RESULTS # Demographic characteristics There were 3310 neonates <32 weeks gestation admitted to the KEMH NICU during the study period of which 3235 met the inclusion criteria; 2969 infants received antibiotics of any duration, 841 received prolonged antibiotics ( ≥ 5 days), and only 266 remained free of antibiotic exposure. In this cohort, there were 173 deaths, 11 in the first 12 h of life, 86 between 12 h and 7 days and 78 after 7 days of life; 267 infants developed LOS (n = 226) and/or NEC/SIP (n = 57) and were censored from the date of diagnosis (Fig. 1). Table [1](#page-3-0) describes maternal and neonatal characteristics described according to duration of antibiotic exposure. Mothers of infants exposed to prolonged antibiotics ( ≥ 5 days) were younger, more likely to be Aboriginal or Torres Strait Islander (ATSI), have chorioamnionitis, be treated with antibiotics antenatally, and birth vaginally compared to those with antibiotic exposure ≤2 days. They were less likely to have diabetes and/or multiple births. Neonates exposed to prolonged antibiotics were of lower gestational age and birth weight and were more likely to have a 5-minute Apgar score <7, require longer duration of ventilation and/ or CPAP, and have central lines in place during their admission. # Antibiotic exposure Timing and duration of antibiotic exposure. Overall, DOT was 71.4 per 1000 inpatient days and infants received a median of 2 days antibiotics (IQR 2–5) per patient (Supplementary Table S1A). The majority (2654, 89.4%) were treated for suspected EOS and 689 (23.2%) were treated for suspected LOS (Supplementary Table S1A). There were 1601 (49.5%) infants who received antibiotics for total ≤2 days (ie no antibiotics or <2 days antibiotics), 793 (24.5%) for 3–4 days and 841 (26.0%) for ≥5 days (Table [1](#page-3-0)). Antibiotic exposure ≥5 days occurred more often in CN (n = 749) than culture-positive infants (n = 226). Trends in antibiotic exposure over time and by gestational age. DOT decreased from 78.0 to 61.9 per 1000 inpatient days and AUR from a median of 0.07 (IQR 0.04–0.11) to 0.05 (IQR 0.03–0.10) from 2012 to 2022 (Supplementary Table S1A). The mean estimate for AUR decreased for all infants <32 weeks combined from 0.08 (95% CI 0.07–0.09) to 0.07 (95% CI 0.06–0.07) (IRR 0.98, 95% CI 0.97–0.99; p < 0.001), and also for the 28 to <32 week gestational age group (n = 1967) from 0.08 (95% CI 0.07–0.09) to 0.07 (95% CI 0.06–0.08) (IRR 0.98, 95% CI 0.96–0.99, p < 0.001) from 2012 to 2021 (Fig. [2](#page-4-0) and Supplementary Table S1B). Pediatric Research **Table 1.** Maternal and neonatal characteristics of infants by days of antibiotic exposure | | | Antibiotic exposure | (days) | | | |----------------------------------|----------------------|---------------------|-------------------|-------------------|----------------| | | All (n = 3235) N (%) | ≤ 2 N = 1601 | 3-4 N = 793 | ≥5 N = 841 | p ^a | | Maternal characteristics | | | | | | | Maternal age (years) (mean, sd) | 30.0 (6.0) | 30.5 (5.9) | 30.1 (5.8) | 29.2 (6.2) | < 0.001 | | Ethnicity (n,%) | | | | | | | Caucasian | 2039 (63.0) | 1034 (64.6) | 509 (64.2) | 496 (59.0) | <0.001 | | ATSI | 431 (13.3) | 178 (11.1) | 103 (13.0) | 150 (17.8) | | | Other | 765 (23.7) | 389 (24.3) | 181 (22.8) | 195 (23.2) | | | Preeclampsia (n,%) | 558 (17.2) | 293 (18.3) | 130 (16.4) | 135 (16.1) | 0.288 | | Gestational diabetes (n,%) | 365 (11.3) | 216 (13.5) | 90 (11.3) | 59 (7.0) | <0.001 | | Multiple birth (n,%) | 891 (27.5) | 524 (32.7) | 216 (27.2) | 151 (18.0) | <0.001 | | Chorioamnionitis (n,%) | 423 (13.1) | 131 (8.2) | 111 (14.0) | 181 (21.5) | <0.001 | | Antenatal antibiotics (n,%) | 1230 (38.0) | 550 (34.4) | 314 (39.6) | 366 (43.5) | <0.001 | | Antenatal steroids (n,%) | 2147 (66.5) | 1049 (65.7) | 527 (66.6) | 571 (68.1) | 0.498 | | Mode of delivery (n,%) | | | | | | | Caesarean section | 1979 (61.2) | 1031 (64.4) | 459 (57.9) | 489 (58.1) | <0.001 | | Vaginal | 1251 (38.7) | 566 (35.4) | 334 (42.1) | 351 (41.7) | | | Not stated ^b | 4 (0.1) | 3 (0.2) | 0 (0.0) | 1 (0.1) | | | Inborn (n,%) | 2939 (90.9) | 1480 (92.4) | 695 (87.6) | 764 (90.8) | <0.001 | | Neonatal characteristics | | | | | | | Gestational age (weeks) | | | | | | | Median (IQR) | 29.3 (27.0–30.9) | 30.0 (28.3–31.1) | 29.3 (27.1–30.9) | 27.1 (25.1–29.3) | <0.001 | | <28 weeks (n, %) | 1097 (33.9) | 347 (21.7) | 257 (32.4) | 493 (58.6) | <0.001 | | Birthweight | | | | | | | Weight (g) (median, IQR) | 1205 (907–1518) | 1332 (1055–1580) | 1210 (945–1528) | 916 (710–1260) | <0.001 | | Z-score (median, IQR) | 0.12 (-0.53-0.73) | 0.11 (-0.45-0.74) | 0.20 (-0.48-0.76) | 0.06 (-0.67-0.69) | <0.001 | | Male sex (n, %) | 1787 (55.2) | 851 (53.2) | 442 (55.7) | 494 (58.7) | 0.029 | | Apgar<7 at 5 min ( n , %) | 612 (18.9) | 233 (14.6) | 139 (17.5) | 240 (28.6) | <0.001 | | Ventilation | | | | | | | Total (n, %) | 2145 (66.3) | 868 (54.2) | 548 (69.1) | 729 (86.7) | <0.001 | | Duration (hours) (median, IQR) | 24 (10–146) | 14 (7–43) | 20 (10–89) | 129 (21–619) | <0.001 | | CPAP | | | | | | | Total (n, %) | 2977 (92.0) | 1446 (90.3) | 743 (93.7) | 788 (93.7) | 0.002 | | Duration (hours) (mendian, IQR) | 249 (67–926) | 130 (43–507) | 245 (68–914) | 906 (290–1250) | <0.001 | | Central lines | | | | | | | Total (n, %) | 1152 (35.6) | 331 (20.7) | 285 (35.9) | 536 (63.7) | < 0.00 | | Duration (days) (median, IQR) | 93 (58–128) | 84 (46–111) | 83 (46–122) | 105 (73–140) | < 0.00 | | | | | | | | LOS late onset sepsis, NEC necrotising enterocolitis, SIP spontaneous intestinal perforation, ATSI Aboriginal or Torres Strait Islander, Antenatal antibiotics included any antibiotics administered within 48 h of birth; Antenatal steroids included at least one dose of steroids ≥24 h before delivery; Ventilation: Any form of invasive ventilation via endotracheal tube, CPAP non-invasive continuous positive airway pressure ventilation by nasal prongs. There was no change in AUR for infants born <26 weeks (n=398) and 26 to <28 weeks (n=489) gestational age (Fig. 2 and Supplementary Table S1B). The proportion of infants <32 weeks receiving prolonged antibiotics decreased from 101 (35.7%) to 28 (21.1%) over the study period (adjusted IRR 0.94, 95% CI 0.91–0.96, p < 0.001) (Supplementary Table S1A). AUR was inversely proportional to gestational age with a median of 0.31 (IQR 0.07–1.00) at 22 weeks compared with 0.07 (IQR 0.05–0.12) at 31 weeks gestation (Fig. 3). Neonatal outcomes associated with antibiotic exposure in infants without sepsis, NEC and/or SIP. Higher AUR in CN infants was associated with increased mortality, significant brain injury, LOS, NEC, and the composite of any adverse outcome (Table 2A and Fig. 4). Using antibiotic duration $\leq 2$ days as reference, antibiotic duration $\geq 5$ days was associated with increased mortality, ROP, NEC, CLD, longer hospital stays, and the composite of any adverse outcome (Table 2B and Fig. 4). These findings were significant for $\geq 5$ consecutive days of antibiotics but not for non-consecutive days for all outcomes except length of hospital stay (Supplementary Table S2). ## **DISCUSSION** In this study, antibiotic exposure in infants <32 weeks' GA without sepsis, NEC and/or SIP decreased from 2012 to 2022. However, SPRINGER NATURE Pediatric Research ^<sup>ap-value represents the comparison of infants with antibiotic exposure ≤2 d, 3–4 d and ≥5 d; ^b Not stated' category was not included in the statistical test. ![](https://assets.reachrx.ai/images/10.1038%E2%98%8Ds41390-025-04707-5/_page_4_Figure_2.jpeg) Fig. 2 Temporal trends in antibiotic utilisation rate (AUR) in infants <32 weeks' gestational age (GA) without sepsis, necrotising enterocolitis (NEC) or spontaneous intestinal perforation (SIP). Line graphs of the change in AUR from 2012 to 2021 in (a) infants <26 weeks' GA, (b) infants 26 to <28 weeks' GA, (c) infants 28 to <32 weeks' GA and (d) all infants <32 weeks' GA without sepsis, NEC or SIP. Data are presented as estimated incidence rates (dots) and 95% confidence intervals (error bars) derived from binomial regression models adjusted for gestational age, birth weight z-score, sex and year of birth. Data for 2022 was incomplete and therefore excluded from this analysis. ![](https://assets.reachrx.ai/images/10.1038%E2%98%8Ds41390-025-04707-5/_page_4_Figure_4.jpeg) Fig. 3 Antibiotic use by gestational age in infants <32 weeks' gestational age without sepsis, necrotising enterocolitis (NEC) and/or spontaneous intestinal perforation (SIP). Line graph of median antibiotic utilisation rate (AUR) by gestational age at birth in weeks. antibiotic therapy for CN infants continued to occur. At the unit level (AUR), greater use aligned with higher odds of mortality, LOS, NEC and significant brain injury (composite likewise), with no adjusted association for CLD or severe ROP. At the patient level, longer courses were associated with higher odds of CLD and severe ROP, among other outcomes. Several differences were noted in maternal and neonatal characteristics between infants with and without prolonged antibiotic exposure. Most of these parameters, including maternal chorioamnionitis, lower gestational age and birth weight, lower Apgar scores, mechanical ventilation and central lines, have been associated with higher risk of neonatal sepsis and may have resulted in reluctance to stop antibiotics despite negative cultures. ^21–23 Blood culture is considered the gold standard in diagnosis of neonatal sepsis but, despite poor performance characteristics, ancillary biomarkers are commonly used in Pediatric Research SPRINGER NATURE Table 2. Mortality in infants who survived >12 hours and morbidity in infants surviving >7 days according to A) median antibiotic utilisation rate and B) duration of antibiotic exposure | A. | | | | | |-------------------------------------|------------------------------------|-----------------------------|----------------------------|---------| | | AUR (Median) | Unadjusted OR 95% CI | Adjusted OR 95% CI | pa | | Mortality | 0.67 | 1.97 (1.84–2.11) | 1.81 (1.68–1.95) | < 0.001 | | Length of stay (days) (median, IQR) | - | 0.92 (0.89–0.95)β | 1.03 (1.00–1.06)β | 0.057 | | LOS | 0.24 | 1.65 (1.55–1.77) | 1.52 (1.42–1.64) | < 0.001 | | NEC ≥Grade 2 | 0.17 | 1.39 (1.27–1.51) | 1.22 (1.10–1.35) | < 0.001 | | CLDb | 0.07 | 1.24 (1.17–1.32) | 1.06 (0.97–1.16) | 0.178 | | Severe ROP > Stage 3c | 0.09 | 1.12 (1.02–1.22) | 1.06 (0.99–1.13) | 0.119 | | Significant brain injury | 0.08 | 1.31 (1.22–1.41) | 1.18 (1.09–1.29) | < 0.001 | | Any adverse outcomesd | 0.08 | 1.43 (1.34–1.52) | 1.32 (1.21–1.44) | < 0.001 | | B. | | | | | | | exposure ≥ 5
Antibiotic
days | Unadjusted OR/HR 95%
CIe | Adjusted OR/
HRae95% CI | p | | Mortality | 53 (6.3) | 1.44 (0.99–2.10) | 1.45 (1.19–1.76) | < 0.001 | | Length of stay (days) (median, IQR) | 82 (53–114) | 1.96 (1.77–2.18) β | 1.27 (1.15–1.41) β | < 0.001 | | LOS | 74 (8.9) | 1.74 (1.25–2.41) | 1.19 (1.00–1.43) | 0.057 | | NEC ≥Grade 2 | 22 (2.7) | 1.79 (0.99–3.23) | 1.44 (1.10–1.89) | 0.008 | | CLDb | 406 (51.1) | 7.88 (6.39–9.72) | 2.54 (1.94–3.33) | < 0.001 | | Severe ROP > Stage 3c | 101 (13.3) | 7.90 (4.89–12.76) | 1.95 (1.14–3.36) | 0.015 | | Significant brain injury | 59 (7.1) | 3.38 (2.20–5.20) | 1.21 (0.99–1.46) | 0.058 | | Any adverse outcomesd | 484 (58.4) | 6.37 (5.27–7.70) | 1.96 (1.52–2.52) | < 0.001 | AUR antibiotic utilisation rate, OR odds ratio, HR hazard ratio, CI confidence interval, LOS late onset sepsis, NEC necrotising enterocolitis, CLD chronic lung disease, ROP retinopathy of prematurity; Significant brain injury: IVH ≥Grade 3 and/or cystic PVL, β Reported as hazards ratio. ap-value represents the unadjusted and adjusted effects on outcomes of infants for a 10% increase in AUR. Adjustments were made for gestational age, birthweight z-score, sex and year of birth. Hazard ratios are interpreted as higher is worse (i.e.,longer duration of hospital stay has a higher hazard ratio). bExcludes deaths before 36 weeks corrected gestational age. cAssessed on infants <31 weeks gestation and/or birthweight <1250 g. dComposite outcome of mortality, LOS, NEC, CLD, significant brain injury and severe ROP. eRelative to antibiotics ≤2 days. ![](https://assets.reachrx.ai/images/10.1038%E2%98%8Ds41390-025-04707-5/_page_6_Figure_2.jpeg) Fig. 4 Estimated probability of adverse outcome according to the antibiotic utilisation rate (AUR). Line graph of estimated probability of mortality (a) and any composite outcome (b) according to AUR. The composite outcome included mortality and/or any major morbidity including late-onset sepsis, necrotising enterocolitis ≥ stage 2, chronic lung disease, severe retinopathy of prematurity and/or severe brain injury. Figures represent the predicted probabilities derived from logistic regression models adjusted for gestational age, birth weight z-score, sex and year of birth. The grey shaded area represents the 95% confidence interval for the predicted probability. addition and typically result in prolonged antibiotic exposure without evidence for improved outcomes.²² Most neonates in the study received antibiotics for suspected EOS based on obstetric risk factors and postnatal clinical presentation, but treatment was often continued despite sterile cultures.²⁴ This prolonged exposure may be in response to elevated inflammatory markers and/or a clinically unwell baby with poor response to treatment. However, decreased use of unproven ancillary biomarkers can reduce antibiotic exposure without an increase in adverse outcomes in the first 7 days of life, and alternative diagnoses to sepsis should be considered in the unwell infant to reduce unnecessary antibiotic treatment.^25,26 Recent approaches to the diagnosis and management of EOS in preterm infants, based on risk-stratification and deimplementation of biomarker use, are being increasingly recognized as both appropriate and safe.^24–26 To avoid unnecessary antibiotic days, continuation decisions should prioritize cultures and clinical trajectory rather than CRP or other non-specific biomarkers. We observed a substantial reduction in antimicrobial exposure from 78.0 to 61.9 DOT per 1000 inpatient days corresponding with decreasing LOS incidence over the same period.²⁷ The lower incidence of LOS was most likely related to implementation of quality improvement strategies during the study period including early introduction and faster advancement of enteral feeds, topical coconut oil for skin integrity, decreased central lines, improvements in venous line insertion and maintenance, and hand hygiene.²⁷ A greater decrease in antibiotic consumption from 100 to 40 DOT per 1000 inpatient days has been reported in infants <32 weeks in Norway, highlighting that further progress is feasible.⁴ Future opportunities include improvement of antimicrobial stewardship (AMS) interventions, improving blood culture technique to reduce false positives, optimising inoculum volume to increase yield, prompt discontinuation of antibiotics when cultures are negative, and identification of novel biomarkers with a high negative predictive value.^28–30 We found higher odds of adverse neonatal outcomes associated with longer antibiotic exposure in CN infants. Higher AUR and/or prolonged exposure were associated with increased mortality, brain injury, NEC, LOS, CLD, longer hospital stays, and the composite of any adverse outcome, with a dose-dependent effect. This association is well described and likely related to disruption of the neonatal microbiome. ^6,7,31–35 Colonization with commensal organisms is protective, preventing colonization with pathogenic organisms which can subsequently cause sepsis and/ or NEC.^7,36 Mucosal microbiota critical for instruction of the developing immune system effect immunological alterations in the gut, development of immune tolerance and competence, and inflammatory balance.^7,36,37 Although the neonatal microbiome may recover following shorter courses of antimicrobials, prolonged exposure has been associated with permanent perturbations of the microbiome and long term health effects including obesity, asthma, atopy, inflammatory bowel disease, diabetes and multiple sclerosis.^7,10,36,38 The adverse effects of prolonged exposure had a stronger association with consecutive rather than non-consecutive antibiotic days. Although any antibiotic exposure disrupts the intestinal microbiome, longer exposure is associated with extended dysbiosis, slower recovery and greater antimicrobial resistance. It is possible that short, non-consecutive courses of antibiotics give the intestinal microbiome a chance to recover between treatment episodes, resulting in less detrimental effect, emphasizing the importance of timely cessation of antibiotic therapy. So, Sonsidering >90% of blood cultures are positive by 36 h, this would be a reasonable time point to stop antibiotics in CN infants to reduce unnecessary antibiotic exposure. This study had several strengths, including the large sample size of a contemporaneous cohort, data extraction from a well-established electronic database, and rigorous data analysis. However, the study also had limitations inherent to its monocentric and retrospective nature which limit generalizability. There was also the potential for confounding by indication, with sicker infants receiving longer courses of antibiotics, despite statistical modelling. Inclusion of laboratory data, such as ancillary biomarkers, and an estimate of illness severity, may have added value but were beyond the scope of the study. In conclusion, antibiotics are critical therapy for infants with sepsis but can cause harm in those without. Potential future strategies to reduce antibiotic exposure include implementing evidence-based antimicrobial stewardship programs, identification of and withholding antibiotics in infants at low infection risk, optimization of blood culture collection technique and volume, and identification of novel biomarkers with high predictive values. 30,43 #### **DATA AVAILABILITY** All data generated or analysed during this study are included in this article. Further enquiries can be directed to the corresponding author. Pediatric Research SPRINGER NATURE ## REFERENCES - 1. Al-Turkait, A., Szatkowski, L., Choonara, I. & Ojha, S. Review of Drug Utilization Studies in Neonatal Units: A Global Perspective. Int. J. Environ. Res. Public Health 17, 5669 (2020). - 2. Boverman, G. et al. Neonatal ICU Antibiotic Use Trends within an Integrated Delivery Network. Antimicrob. Resist Infect. Control 11, 21 (2022). - 3. Dimopoulou, V. et al. Antibiotic Exposure for Culture-Negative Early-Onset Sepsis in Late-Preterm and Term Newborns: An International Study. Pediatr. Res. 97, 1629–1635 (2025). - 4. Huncikova, Z. et al. Variation in Antibiotic Consumption in Very Preterm Infants-a 10 Year Population-Based Study. J. Antimicrob. Chemother. 79, 143–150 (2024). - 5. Wagstaff, J. S. et al. Antibiotic Treatment of Suspected and Confirmed Neonatal Sepsis within 28 Days of Birth: A Retrospective Analysis. Front Pharm. 10, 1191 (2019). - 6. Vatne, A. et al. Early Empirical Antibiotics and Adverse Clinical Outcomes in Infants Born Very Preterm: A Population-Based Cohort. J. Pediatr. 253, 107–114.e105 (2023). - 7. Cotten, C. M. Adverse Consequences of Neonatal Antibiotic Exposure. Curr. Opin. Pediatr. 28, 141–149 (2016). - 8. Mukhopadhyay, S., Sengupta, S. & Puopolo, K. M. Challenges and Opportunities for Antibiotic Stewardship among Preterm Infants. Arch. Dis. Child Fetal Neonatal Ed. 104, F327–f332 (2019). - 9. Strobel, K. M. et al. Effect of Early Antibiotic Exposure on Necrotizing Enterocolitis and Growth in Extremely Preterm Infants. Pediatr. Res. 98, 137–143 (2025). - 10. Sarkar, A., Yoo, J. Y., Valeria Ozorio Dutra, S., Morgan, K. H. & Groer, M. The Association between Early-Life Gut Microbiota and Long-Term Health and Diseases. J. Clin. Med. 10, 459 (2021). - 11. Strunk, T., Molloy, E. J., Mishra, A. & Bhutta, Z. A. Neonatal Bacterial Sepsis. Lancet 404, 277–293 (2024). - 12. Chughtai, A. A. et al. Association between Congenital Anomalies and Late-Onset Bacterial Infections in Neonates Admitted to Neonatal Intensive Care Units in Australia and New Zealand: A Population-Based Cohort Study. Neonatology 122, 95–105 (2025). - 13. Fitchett, E. J. A. et al. Strengthening the Reporting of Observational Studies in Epidemiology for Newborn Infection (Strobe-Ni): An Extension of the Strobe Statement for Neonatal Infection Research. Lancet Infect. Dis. 16, e202–e213 (2016). - 14. Australia and New Zealand Neonatal Network (ANZNN). ANZNN Data Dictionary 2023, (2022). - 15. Lowe, S. A. et al. SOMANZ Guidelines for the Management of Hypertensive Disorders of Pregnancy 2014. Aust. N. Z. J. Obstet. Gynaecol. 55, e1–e29 (2015). - 16. Nankervis, A. et al. McElduff A for the Australasian Diabetes in Pregnancy Society ADIPS Consensus Guidelines for the Testing and Diagnosis of Gestational Diabetes Mellitus in Australia, [https://www.adips.org/downloads/](https://www.adips.org/downloads/adipsconsensusguidelinesgdm-03.05.13versionacceptedfinal.pdf) [adipsconsensusguidelinesgdm-03.05.13versionaccepted](https://www.adips.org/downloads/adipsconsensusguidelinesgdm-03.05.13versionacceptedfinal.pdf)final.pdf (2014). - 17. Ballard, J. L. et al. New Ballard Score, Expanded to Include Extremely Premature Infants. J. Pediatr. 119, 417–423 (1991). - 18. Kliegman, R. M. & Walsh, M. C. Neonatal Necrotizing Enterocolitis: Pathogenesis, Classification, and Spectrum of Illness. Curr. Probl. Pediatr. 17, 213–288 (1987). - 19. de Vries, L. S., Eken, P. & Dubowitz, L. M. The Spectrum of Leukomalacia Using Cranial Ultrasound. Behav. Brain Res 49, 1–6 (1992). - 20. Papile, L. A., Burstein, J., Burstein, R. & Koffler, H. Incidence and Evolution of Subependymal and Intraventricular Hemorrhage: A Study of Infants with Birth Weights Less Than 1,500 Gm. J. Pediatr. 92, 529–534 (1978). - 21. Araújo, B. C. & Guimarães, H. Risk Factors for Neonatal Sepsis: An Overview. J. Pediatr. Neonatal Individualized Med. (JPNIM 9, e090206 (2020). - 22. De Rose, D. U. et al. Stop in Time: How to Reduce Unnecessary Antibiotics in Newborns with Late-Onset Sepsis in Neonatal Intensive Care. Trop. Med. Infect. Dis. 9, 63 (2024). - 23. Murthy, S., Godinho, M. A., Guddattu, V., Lewis, L. E. S. & Nair, N. S. Risk Factors of Neonatal Sepsis in India: A Systematic Review and Meta-Analysis. PLoS One 14, e0215683 (2019). - 24. Mukhopadhyay, S. et al. Variation in Sepsis Evaluation across a National Network of Nurseries. Pediatrics 139, e20162845 (2017). - 25. Barboza, A. Z. et al. Trends in C-Reactive Protein Use in Early-Onset Sepsis Evaluations and Associated Antibiotic Use. J. Pediatr. 273, 114153 (2024). - 26. Gannon, J., Strunk, T., Friesen, N. & Saw, C. Neonatal Early-Onset Sepsis Calculator: Impact on Antibiotic Use in a Level Ii Neonatal Unit in Western Australia. Pediatr. Neonatol. 65, 71–75 (2024). - 27. Mackay, C. A. et al. Epidemiology and Outcomes of Neonatal Sepsis: Experience from a Tertiary Australian NICU. Neonatology 121, 703–714 (2024). - 28. Gustavsson, L., Lindquist, S., Elfvin, A., Hentz, E. & Studahl, M. Reduced Antibiotic Use in Extremely Preterm Infants with an Antimicrobial Stewardship Intervention. BMJ Paediatr. Open 4, e000872 (2020). - 29. Mascarenhas, D., Ho, M. S. P., Ting, J. & Shah, P. S. Antimicrobial Stewardship Programs in Neonates: A Meta-Analysis. Pediatrics 153, e2023065091 (2024). - 30. Rajar, P. et al. Antibiotic Stewardship in Premature Infants: A Systematic Review. Neonatology 117, 673–686 (2020). - 31. Cantey, J. B., Pyle, A. K., Wozniak, P. S., Hynan, L. S. & Sánchez, P. J. Early Antibiotic Exposure and Adverse Outcomes in Preterm, Very Low Birth Weight Infants. J. Pediatr. 203, 62–67 (2018). - 32. Cotten, C. M. et al. Prolonged Duration of Initial Empirical Antibiotic Treatment Is Associated with Increased Rates of Necrotizing Enterocolitis and Death for Extremely Low Birth Weight Infants. Pediatrics 123, 58–66 (2009). - 33. Dierikx, T. H. et al. Association between Duration of Early Empiric Antibiotics and Necrotizing Enterocolitis and Late-Onset Sepsis in Preterm Infants: A Multicenter Cohort Study. Eur. J. Pediatr. 181, 3715–3724 (2022). - 34. Shi, W. et al. Early Antibiotic Exposure and Bronchopulmonary Dysplasia in Very Preterm Infants at Low Risk of Early-Onset Sepsis. JAMA Netw. Open 7, e2418831 (2024). - 35. Ting, J. Y. et al. Association between Antibiotic Use and Neonatal Mortality and Morbidities in Very Low-Birth-Weight Infants without Culture-Proven Sepsis or Necrotizing Enterocolitis. JAMA Pediatr. 170, 1181–1187 (2016). - 36. Zeissig, S. & Blumberg, R. S. Life at the Beginning: Perturbation of the Microbiota by Antibiotics in Early Life and Its Role in Health and Disease. Nat. Immunol. 15, 307–310 (2014). - 37. Carl, M. A. et al. Sepsis from the Gut: The Enteric Habitat of Bacteria That Cause Late-Onset Neonatal Bloodstream Infections. Clin. Infect. Dis. 58, 1211–1218 (2014). - 38. Fortmann, I. et al. Perinatal Antibiotic Exposure and Respiratory Outcomes in Children Born Preterm. JAMA Netw. Open 8, e259647–e259647 (2025). - 39. Zwittink, R. D. et al. Association between Duration of Intravenous Antibiotic Administration and Early-Life Microbiota Development in Late-Preterm Infants. Eur. J. Clin. Microbiol Infect. Dis. 37, 475–483 (2018). - 40. Huggard, D. et al. Time to Positivity (Ttp) of Neonatal Blood Cultures: A Trend Analysis over a Decade from Ireland. J. Matern Fetal Neonatal Med 34, 780–786 (2021). - 41. Kuzniewicz, M. W., Mukhopadhyay, S., Li, S., Walsh, E. M. & Puopolo, K. M. Time to Positivity of Neonatal Blood Cultures for Early-Onset Sepsis. Pediatr. Infect. Dis. J. 39, 634–640 (2020). - 42. Mukhopadhyay, S. et al. Time to Positivity of Blood Cultures in Neonatal Late-Onset Bacteraemia. Arch. Dis. Child Fetal Neonatal Ed. 107, 583–588 (2022). - 43. Stocker, M. et al. Less Is More: Antibiotics at the Beginning of Life. Nat. Commun. 14, 2423 (2023). # AUTHOR CONTRIBUTIONS MCA: investigation, data curation, resources, writing – original draft and editing; NE: data curation, formal analysis, resources, software, writing – review and editing; PM: conceptualization, writing – review and editing; SD: data curation, resources, software, writing – review and editing; KR: data curation, resources, writing – review and editing; ST: conceptualization, investigation, data curation, supervision, writing – review and editing. ## FUNDING Open Access funding enabled and organized by CAUL and its Member Institutions. # COMPETING INTERESTS The authors declare no competing interests. # ETHICAL APPROVAL This study protocol was reviewed and approved by the CAHS GEKO Triage Committee, approval number 000471. The need for informed consent was waived by the CAHS GEKO Triage Committee. # ADDITIONAL INFORMATION Supplementary information The online version contains supplementary material available at . Correspondence and requests for materials should be addressed to Cheryl Anne Mackay. Reprints and permission information is available at [http://www.nature.com/](http://www.nature.com/reprints) [reprints](http://www.nature.com/reprints) Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Pediatric Research Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit [http://](http://creativecommons.org/licenses/by/4.0/) [creativecommons.org/licenses/by/4.0/.](http://creativecommons.org/licenses/by/4.0/) © The Author(s) 2026 Pediatric Research