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

Congenital anomalies (CAs) with structural or functional conditions present at birth can have significant long-term impacts on physical, intellectual, and social wellbeing. Globally, around 8 million neonates are born with one or more CAs every year (3–6% of total births), and of those, around 240,000 die during the neonatal period [1, 2]. The most common major anomalies are congenital cardiac defects, neural tube defects, and trisomy 21 [3]. The National Congenital Anomalies Data Collection (NCADC) collects data from various sources across Australia. NCADC identified over 8,900 (3%) neonates with a CA in Australia in 2016, and of them, one-third (29%) had multiple anomalies [4]. Some neonates with CAs are admitted to neonatal (NICUs) or paediatric intensive care units for supportive care and/or surgery, and these neonates may be at higher risks of systemic complications [5].

Infections are among the leading causes of mortality in NICUs [6]. While definitions vary, neonatal infection are broadly classified into early-onset (pathogens vertically acquired before or during and onset <48 h from birth) and late-onset sepsis (LOS) (horizontally acquired, onset >48 h). Both gestational age and birth weight are inversely related to infection risk, and infections are experienced by up to 25% of infants born <28 weeks gestation and/or of extremely low birth weight (LBW) [7‒9].

NICU managements such as parental nutrition, mechanical ventilation, and central venous lines also increase the risk of sepsis, particularly if prolonged [10‒13]. However, data available regarding the risk of infection associated with CA are inconclusive [6, 14, 15], which may be related to small samples of CA infants evaluated in previous reports. This study aimed to examine the association of congenital abnormalities and bacterial LOS in a large population-based cohort of neonates admitted to NICUs in Australia and New Zealand. We hypothesised that babies with congenital abnormalities may have higher risk of developing bacterial LOS.

This was a population-based cohort study using Australian and New Zealand Neonatal Network (ANZNN) data of neonates [16]. ANZNN is a collaborative network, established in 1994, to monitor the mortality and morbidity of high-risk neonates admitted to NICUs across Australia and New Zealand. Maternal and neonatal data are entered into a centralised database by trained audit officers at each participating hospital [16]. All neonates admitted to a participating unit during the first 28 days of life who meet one or more of the following criteria are included in the ANZNN database: (1) <32 weeks gestation at birth or (2) <1,500 grams at birth or (3) received assisted ventilation or (4) received major surgery (involved opening a body cavity) or (5) received therapeutic hypothermia [16]. Criteria 3, 4, and 5 are set to include term (FT) and moderately preterm (MPT) neonates who require critical care interventions while excluding other unwell neonates admitted only for reasons of close monitoring, investigations, or not needing these high-level interventions. All very preterm (VPT, <32 weeks), MPT (32–36 weeks), and FT (37–41 weeks) neonates born from 2007 to 2017 and registered in the ANZNN database were included in this study. Neonates delivered post-term (>41 weeks) predominantly admitted for perinatal complications were not included in the study.

Confirmed infection is defined in the ANZNN database as “isolation of an organism from at least one blood or CSF culture or PCR identification in CSF and, after consideration of the clinical and laboratory evidence, a decision is made to treat with antibiotics with therapeutic intent against this organism” [17, 18]. We only included bacterial LOS, defined as “the presence of at least one episode of blood or CSF bacterial infection with initial symptoms occurring from 48 h after birth.” Major surgery was defined as “baby had surgery which involved opening a body cavity during the admission.”

CAs are defined as “structural abnormalities (including deformations) present at birth and diagnosed prior to separation from care (discharge home).” CAs were further categorised into single and multiple CAs. Minor skin anomalies, such as lymphangioma, hemangioma, non-neoplastic naevus, and other specified congenital malformations, were excluded. Major surgery is defined as “baby had surgery which involved opening a body cavity during this admission” [16]. Other study variables were the age of the mother, assisted reproductive treatment, plurality, gender of baby, premature rupture of membranes, mode of birth, birth year, birth weight.

Since prematurity is an important risk factor for infections and admission to NICUs, data were stratified by VPT (<32 weeks), MPT (32–36 weeks), and FT (37–41 weeks) neonates. Demographic/clinical characteristics of neonates and mothers and rates of LOS were compared among gestational age groups in neonates with and without CA. We fitted log-binomial model under generalised linear model framework to estimate relative risk (RR) of bacterial infection, for babies with CAs. In order to adjust for potential confounders while keeping a parsimonious model, we adopted backward elimination approach. We fitted the univariable generalised linear model with all potential confounders. All variables with a p value ≤0.1 in the univariable model were included in the initial multivariable model. Then we removed variables that did not have any confounding effect, that is, could not make meaningful (roughly 10%) change in the effect measure with the main exposure variable. If a variable turned out to be a confounder, it was kept in the model. Variables that were not confounders but had a significant (p < 0.05) association with the outcome variable in the multivariable model, or had clinical significance, were also kept. Unadjusted and adjusted RRs with their 95% confidence intervals (95% CIs) were presented in tables. Major surgery during a NICU admission was identified as an effect modifier for infections in NICU; therefore, stratified analysis was also performed by surgery status. Given the differences in outcome of a coagulase-negative staphylococci (CONS) sepsis versus other bacterial pathogens, an additional analysis was also conducted for neonates who has at least one episode of CONS infection during NICU stay. Data analyses were performed using SAS 9.4.

This study was approved by the UNSW Human Research Advisory Panel (HREAP) G Health, Medical, Community and Social Ethic Committee (#HC190419) and the ANZNN committee. ANZNN is a collaborative network of NICUs in Australia and New Zealand to monitor and improve the care of high-risk neonates. The data collection is a clinical quality registry for clinical audit, benchmarking, and outcome feedback to individual unit. Opt-out consent for data collection is approved by regional or institutional ethics for the storage and amalgamation of de-identified clinical data.

Of 102,808 neonates included in the study, 38,720 (37.7%) were born VPT; 33,699, (32.8%) MPT; and 30,389 (29.6%) at FT. Demographic and clinical characteristics of neonates and mothers are presented in Tables 1 and 2. Males were over-represented at 58%, and around two-thirds were of extremely LBW (<1,000 grams) or LBW (1,000–2,499 g). Major surgery was performed in 9,725 (9.5%) neonates, more than half of whom were born at FT.

During the study period, 8,789 (8.5%) neonates with one or more CAs were admitted to NICUs, and the majority (56.2%) were born at FT. Rates of CAs were 3.4%, 7.5%, and 16.2% in very VPT, MPT, and FT neonates, respectively (Table 1). CAs of heart and circulatory system (44%, 3,879/8,789), digestive system (24%, 2,107/8,789), and musculoskeletal system (23%, 2,062/8,789) were most common (online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000540276). Figure 1 shows trends of CAs in neonates born from 2007 to 2017 and admitted to all NICU in Australia and New Zealand according to ICD-10 classification. Overall, NICU admissions of neonates with CAs increased over the past 10 years, particularly for CAs of the heart and circulatory system (Fig. 1).

Neonates at higher gestational age groups were more likely to have CA than those in lower gestational age groups. For the non-surgery group, mean gestational age was 34.6 (±4.5 weeks) in CA infants compared to 33 weeks (±4.5 weeks) in babies without CA (p value < 0.001). This gestation difference was also present for neonates with and without CA who needed surgery (36.7 weeks vs. 33.2 weeks, p value <0.001, respectively).

For the overall population, the incidence of at least one bacterial LOS during the study period was 5.3%, with a higher rate among VPT neonates (11.1%), compared to MPT (1.8%) and FT neonate (1.8%). Infection rates decreased from 5.8% in 2007 to 3.7% in 2017 (Fig. 2). Most neonates had a single episode of LOS (81.6%, 4,464/5,472), while 14.4% (790) had two episodes and 4% (218) had three or more episodes of LOS. Of the total 6,760 episodes of bacterial LOS during the study period, Staphylococcus species were the most commonly isolated pathogens (66.6%, 4,491/6,760). Most Staphylococcus infections (3,842/4,491, 85.5%) were CONS. A total of 3,417 neonates had at least one episode of CONS during the study period. Other common bacterial infections included Streptococcus spp., Escherichia coli, Enterobacter spp., Enterococcus spp., Klebsiella spp., and Haemophilus spp.

The bacterial LOS rate in neonates with CA was 8.2% (720/8,789) and 5.1% in neonates without CA (4,752/94,019) (adjusted RR [aRR] 1.67, 95% CI: 1.45–1.92). Rates of LOS and RRs adjusted to confounding factors were significantly higher in the neonates with CA than those without CA: VPT (17.8 vs. 10.9%, aRR 1.33, 95% CI: 1.12–1.56), MPT (8.7% vs. 1.3%, aRR 4.37, 95% CI: 3.17–6.02), and FT (5.4% vs. 1.1%, aRR 2.83, 95% CI: 1.94–4.11) (see Table 3). Neonates with CAs had higher rates of CONS infections than other infections (online suppl. Table S2).

As surgery was identified as an effect modifier for infections, additional analysis was performed by surgery status. After stratification for the need for major surgical intervention for the overall population, bacterial LOS was reported in 14.2% (1,380/9,725) of neonates who required major surgery compared to 4.4% (4,092/93,083) in neonates who did not require major surgery (p < 0.001). Among neonates who did not have major surgery, rates of LOS and adjusted infection risks for neonates with CA were consistently higher compared to those without CA across all 3 GA groups: VPT (14.3% vs. 9.6%; aRR 1.32, 95% CI: 1.11–1.57), MPT (4% vs. 0.9%; aRR 4.45, 95% CI: 3.23–6.14), and FT (2.1% vs. 0.7%; aRR 2.87, 95% CI: 1.97–4.18), respectively. Overall, the risk of LOS was 71% higher in neonates with CA compared to those without CA (aRR 1.71, 95% CI: 1.48–1.96). After adjusting for confounders including birth weight, CAs were not associated with LOS for neonates who needed surgery overall or in any GA group (Table 3). Among VPT infants needing surgery, there were significant gestation differences between those with or without CA. Those 1,698 infants without CA were of lower gestation (mean GA 26 vs. 28 week, respectively, p < 0.001) and birth weight (mean BW 940 g vs. 1,148 g, p < 0.001) than those 372 infants with CA and had a higher crude rate of LOS. In determining if the presence of multiple CAs were associated with higher risk of bacterial LOS across all gestational age groups, results showed that multiple CAs further increased these risks, particularly within the moderate and FT gestation groups (Table 4).

To our knowledge, this is the largest study to demonstrate the higher risk of bacterial LOS in a diverse cohort of 8,789 CA neonates who recently received NICU care. After adjusting for the relatively higher gestation of CA infants and other confounding factors, we continue to find higher adjusted LOS risk in VPT, MPT, and FT neonates with CA than their counterparts. In a stratified analysis by surgery, our study showed an association between CA and LOS in preterm, MPT, and FT neonates, only in neonates with CA who did not require major surgery.

Overall, the bacterial LOS rate was 5.3% in our study with a higher rate among VPT neonates (11.1%). Previous studies show that around 10–30% of neonates admitted to NICUs experience at least one episode of LOS before discharge [8, 19, 20]. However, there has previously been limited evidence around the relationship between CAs and bacterial LOS. Murthy et al. [14] examined rates of bloodstream infection and urinary tract infection in 1,085 infants with congenital diaphragmatic hernia. Of 1,085 infants, 8.3% had bloodstream infections and 6% had urinary tract infections. In multivariate analyses, lower birth weight, low Apgar score, receipt of extracorporeal membrane oxygenation, patch repair of congenital diaphragmatic hernia, and other CAs were associated with the composite outcome of death and infections in NICU. We also found higher rates of infection in 8,789 neonates with a diverse array of CAs. In a study of LOS infants in Taiwan, CAs independently increased the mortality risk in neonates with LOS (OR, 4.12; 95% CI: 1.60–10.60), but there was no LOS control group in this study [19]. A study in Egypt also did not find any association between CAs and infections; however, it was a much smaller cross-sectional study, including just over 1,000 neonates from 6 NICUs [6]. In contrast, we used a cohort study design with a control group and the sample size of over 100,000 infants, which was 100 times larger to detect differences.

Higher rates of infection in neonates with CAs may be due to the nature of the anomaly, using catheters and central lines, parenteral nutrition, mechanical ventilation, performing other invasive and non-invasive procedures, and nutritional deficiencies [14]. Some neonates with major congenital malformations may have associated immune dysfunction, which may put them at higher risk of developing infections [21, 22]. Like other studies [15], Staphylococcus was the most frequently isolated pathogen in our study, predominantly CONS. While our study was not able to delineate the primary source of all infections, a higher proportion of CONS may suggest that interventions such as central line associated infections are a significant contributor to the high rate of infections observed in neonates with CAs.

Although, the main aim of this study was to examine the association of congenital abnormalities and LOS, major surgery was identified as an effect modifier for infections in NICU. In stratified analysis by surgery status, CAs were associated with LOS in only the non-surgery group in our study, and we did not find any association between CAs and infections in neonates who had undergone surgery. Neonates with CA who had major surgery do not have an increased risk of infection compared to neonates without CA who had major surgery is unexpected. This observation requires further investigation. It is likely that surgery confers such a high risk of infection in of itself as a risk factor that it dilutes any observed effect of CAs on infection risk. It is also possible that neonates in the group who had surgery were admitted for reasons other than their CAs, such as prematurity or complications of prematurity, which may have in themselves increased their infection risk. That may explain that the difference was not significant in preterm and MPT neonates after adjusted for gestational age and other risk factors.

The main strength of this study is a large dataset, which is representative of all tertiary NICUs in Australia and New Zealand. Moreover, we only used clinically valid data on CAs and bacterial LOS as this information was taken from verified hospital records. There are imitations of this study as well. First, we included babies admitted to NICU who met ANZNN registration criteria of needing respiratory support or needing operative surgery within the first 28 days of life that some had CAs and majority of them were born at FT. In contrast, rates of CAs were 4 times higher for preterm in the NCADC database [4, 23]. This may be due to the reason that we did not include babies with CAs who were admitted to neonatal units for investigation and assessment but who only needed respiratory support or surgery after the first 28 days of life or did not need at all. Some of these infants may also be admitted to a paediatric intensive care unit. Therefore, this study has a selection bias towards those who were particularly unwell that required active intervention within the first 4 weeks of life. Nonetheless, among admitted NICU neonates, who require critical care intervention in first 28 days, neonates born with CA were more vulnerable to infection than those without CA. We could not adjust for duration of parenteral nutrition, mechanical ventilation, central-line insertion, and blood transfusion prior to LOS, which may cause an increase in NICU infections [11, 12, 15] and in turn could result in a further prolonged duration of critical care support. Some of these variables of risk exposure prior to LOS were not available in our dataset to explore cause-effect relationship. Finally, we included all neonates with structural CAs in the ANZNN database and only excluded some minor CAs. It is possible that the risk of infection may differ across the spectrum of major CAs included. We note there is no standard definition for major CAs. Development of a consistent international framework to define the severity of CAs may allow us to better understand and address risks of additional morbidities such as infection for these infants in the future.

This large population-based cohort study showed that neonates with CAs may be at higher risk of developing nosocomial bacterial LOS during NICU stay. LOS rates were higher in neonates who required major surgery compared to neonate neonates who did not require major surgery; however, CAs were associated with an increased risk of LOS among only in non-surgical neonate. The higher rate of infections in surgical neonates with CAs is at present unexplained and suggests that the infection risk for this group is modified by factors beyond surgery itself. Despite advances in neonatal care the world ever has led to an increase in survival rate of infants, nosocomial infections remain a serious challenge. Given their higher infection risk, infants with CAs require effective prevention efforts to avoid infections complications such as high morbidity and mortality, prolonged admission in NICUs, necrotising enterocolitis, and neurodevelopmental impairment. Existing infection prevention practices should be strictly adhered to, including proper hand hygiene, use of personal protective equipment, maximal sterile precautions, and adequate care and prompt removal of unnecessary catheters. Optimisation of infection prevention strategies for neonate with CAs should be explored.

The authors thank all the Advisory Council Members of the ANZNN (*denotes ANZNN Executive): Australia: Scott Morris (Flinders Medical Centre, SA, Australia), Peter Schmidt, Manbir Chauhan* (Gold Coast University Hospital, QLD, Australia), Larissa Korostenski (John Hunter Children’s Hospital, NSW, Australia), Mary Sharp, Steven Resnick, Rebecca Thomas, Tobias Strunk* (King Edward Memorial and Perth Children’s Hospitals, WA, Australia), Jacqueline Stack (Liverpool Hospital, NSW, Australia), Pita Birch, Tori Oliver* (Mater Mother’s Hospital, QLD, Australia), Dan Casalaz, Jim Holberton* (Mercy Hospital for Women, VIC, Australia), Alice Stewart, Rod Hunt*, Kenneth Tan* (Monash Medical Centre, VIC, Australia), Lucy Cooke (Neonatal Retrieval Emergency Service Southern Queensland, QLD, Australia), Lyn Downe (Nepean Hospital, NSW, Australia), Jonathan Davis (Newborn Emergency Transport Service, WA, Australia), Michael Stewart (Paediatric Infant Perinatal Emergency Retrieval, VIC, Australia), Andrew Berry (Newborn and Paediatric Emergency Transport Service, NSW, Australia), Leah Hickey (Royal Children’s Hospital, VIC, Australia), Mantho Kgosiemang (Royal Darwin Hospital, NT, Australia), Tony De Paoli, Naomi Spotswood* (Royal Hobart Hospital, TAS, Australia), Srinivas Bolisetty, Kei Lui* (Royal Hospital for Women, NSW, Australia), Eveline Staub (Royal North Shore Hospital, NSW, Australia), Mark Greenhalgh, (Royal Prince Alfred Hospital, NSW, Australia), Pieter Koorts (Royal Brisbane and Women’s Hospital, QLD, Australia), Sue Jacobs (Royal Women’s Hospital, VIC, Australia), Amy Keir* (SAAS MedSTAR Kids, SA, Australia), Clare Collins (Sunshine Hospital, VIC, Australia), Andrew Numa (Sydney Children’s Hospital, NSW, Australia), Hazel Carlisle (the Canberra Hospital, ACT, Australia), Nadia Badawi, Himanshu Popat (the Children’s Hospital at Westmead, NSW, Australia), Gary Alcock (Townsville University Hospital, QLD, Australia), Melissa Luig* (Westmead Hospital, NSW, Australia), Alison Kent (Women’s & Children’s Hospital, SA, Australia); New Zealand: Bronwyn Dixon (Christchurch Women’s Hospital), Brian Darlow (Christchurch School of Medicine), Peter Fowlie, Andrew Kelly (Dunedin Hospital), Guy Bloomfield (Middlemore Hospital), Mariam Buksh, Malcolm Battin* (Auckland City Hospital), Jutta van den Boom (Waikato Hospital), Helen Miller, Angelica Allermo-Fletcher, Claire Jacobs* (Wellington Women’s Hospital); other: Victor Samuel Rajadurai* (KK Women’s and Children’s Hospital, Singapore), Simon Lam (Prince of Wales Hospital, Hong Kong). We also wish to thank ANZNN Executives that are not members of hospitals contributing data: Georgina Chambers* (National Perinatal Epidemiology and Statistics Unit, University of New South Wales), David Barker* (Whangarei Hospital, New Zealand), Anjali Dhawan* (Blacktown Hospital, NSW, Australia), Natalie Merida* (consumer), Denise Harrison* (ACNN). We also thank the members of the ANZNN Scientific Research Review Committee: Kenneth Tan, Nicola Austin, Barbara Bajuk, Luke Jardine, Michael Meyer, Tim Schindler, Michael Stark, Tobias Strunk, Javeed Travadi.

This study was approved by the UNSW Human Research Advisory Panel (HREAP) G: Health, Medical, Community and Social Ethic Committee (#HC190419). Opt-out informed consent protocol was used for use of participant data for research purposes. This consent procedure was reviewed and approved by the Australian and New Zealand Neonatal Network (ANZNN) committee. ANZNN is a collaborative network of NICUs in Australia and New Zealand to monitor and improve the care of high-risk neonates. The data collection is a clinical quality registry for clinical audit, benchmarking, and outcome feedback to the individual unit.

There is no conflict of interest declared by any author.

There was no funding involved in this study.

Abrar Ahmad Chughtai: conceptualisation and design of the study, data analysis, and preparation of the first draft of manuscript. Kei Lui: conceptualisation and design of the study, data analysis, and review/editing of the manuscript. Naomi Spotswood, Tobias Strunk, Trisha Parmar, Tim Schindler, Himanshu Popat, and Sharon Sue Wen Chow: contributed to data analysis and review/editing of the manuscript. All the authors reviewed the final version and gave approval of the version to be published.

All data have been presented in this study. Further enquiries can be directed to the corresponding author.