Article Text

Download PDFPDF

Early surgery and neurodevelopmental outcomes of children born extremely preterm
  1. Rodney W Hunt1,2,3,4,
  2. Leah M Hickey1,
  3. Alice C Burnett1,3,4,5,
  4. Peter J Anderson3,4,
  5. Jeanie Ling Yoong Cheong2,3,4,5,
  6. Lex W Doyle5,6
  7. for the Victorian Infant Collaborative Study group
  1. 1 Department of Neonatal Medicine, Royal Children’s Hospital, Melbourne, Australia
  2. 2 Department of Clinical Science, Murdoch Childrens Research Institute, Melbourne, Australia
  3. 3 Department of Paediatrics, University of Melbourne, Melbourne, Australia
  4. 4 Victorian Infant Brain Studies, Murdoch Childrens Research Institute, Melbourne, Australia
  5. 5 Neonatal Services, Royal Women’s Hospital, Melbourne, Australia
  6. 6 Department of Obstetrics & Gynaecology, University of Melbourne, Melbourne, Australia
  1. Correspondence to Dr Rodney W Hunt, Department of Neonatal Medicine, Royal Children’s Hospital, 50 Flemington Road, Parkville, VIC 3052, Australia; rod.hunt{at}


Objectives To (1) compare the neurodevelopmental outcomes at 8 years of age of children born extremely preterm (EP) who underwent surgical procedures during the course of their initial hospital admission with those who did not and (2) compare the outcomes across eras, from 1991 to 2005.

Design Prospective observational cohort studies conducted over three different eras (1991-1992, 1997 and 2005). Surviving EP children, who required surgical intervention during the primary hospitalisation, were assessed for general intelligence (IQ) and neurosensory status at 8 years of age. Major neurosensory disability comprised any of moderate/severe cerebral palsy, IQ less than -2 SD relative to term controls, blindness or deafness.

Results Overall, 29% (161/546) of survivors had surgery during the newborn period, with similar rates in each era. Follow-up rates at 8 years were high (91%; 499/546), and 17% (86/499) of survivors assessed had a major neurosensory disability. Rates of major neurosensory disability were substantially higher in the surgical group (33%; 52/158) compared with those who did not have surgery (10%; 34/341) (OR 4.28, 95% CI 2.61 to 7.03). Rates of disability in the surgical group did not improve over time. After adjustment for relevant confounders, no specific surgical procedure was associated with increased risk of disability.

Implications and relevance Major neurosensory disability at 8 years was higher in children born EP who underwent surgery during their initial hospital admission compared with those who did not. The rates of major neurosensory disability in the surgical cohort are not improving over time.

  • neonatology
  • neurodevelopment
  • epidemiology
  • paediatric surgery

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?

  • Approximately 20%–30% of EP survivors will require surgery during their first hospital admission.

  • When assessed in early childhood, these children have been shown to have an increased risk of neurodevelopmental impairment.

What this study adds?

  • Rates of surgical exposure for EP survivors have been stable over the past 25 years at around 30%.

  • Surgery in the neonatal period was associated with poorer neurodevelopmental outcome at 8 years, although the risk was not associated with any particular surgical procedure.

  • Over three cohorts of extremely preterm infants, born between 1991 and 2005, the association between surgical exposure and impaired neurodevelopmental outcome has not improved.


As survival rates for extremely preterm (EP) infants (22–27 completed weeks' gestation) have improved, attention has turned to morbidity among survivors, specifically identification of risk factors for adverse neurodevelopment.1 Neonatal surgery has been identified as one risk factor in a number of studies, regardless of gestation at birth.2–4 Other non-Victorian studies reporting increased adverse outcomes for preterm infants who had surgery in the neonatal period have only reported outcomes up to 5 years of age.5 6 In addition, previous studies have not investigated adverse associations between surgery and developmental outcomes over decades.

The objective of the current study was to compare neurodevelopmental outcomes at 8 years of age in regional cohorts of children born EP in three distinct eras (1991–1992, 1997 and 2005) who underwent surgical procedures during their initial hospital admission compared with those who required no surgery. In addition, we aimed to compare the outcomes of the 2005 cohort, with earlier eras to determine if outcomes had changed over time.


The setting for this study was the state of Victoria, Australia. All four tertiary neonatal units in Victoria collaborated with government data collection agencies since the late 1970s to obtain population-based data on long-term outcomes for the smallest and most immature survivors in the state, initially for those of birth weight <1000 g and from the 1990s also for those born <28 weeks.2

All EP live births (22–27 completed weeks’ gestation), free of lethal anomalies, were recruited from three distinct eras, that is, 1991–1992 (24 months), 1997 (12 months) and 2005 (12 months). Controls comprised children of normal birth weight (>2499 g), matched for expected date of birth of a preterm child, sex, maternal private health insurance status and mother’s country of birth (English speaking or not). Some of the 8 year outcomes from these cohorts have been described elsewhere.7–11

Perinatal data were collected at the time of recruitment and prospectively during the course of the study. Gestation at birth was confirmed by an early obstetric ultrasound, that is, before 20 weeks, or by menstrual history if early ultrasound was not available. Birth weight z-scores were computed relative to the British Growth Reference,12 as this is the only reference with z-scores available for gestation ages under 40 weeks. Surgery during the primary hospitalisation and the type of surgery were recorded, among other perinatal variables.

Sociodemographic details collected included age of the mother at the birth of the child, years of maternal education (dichotomised into lower and higher around the median years of schooling), social class (based on the occupation of the major income earner in the family) and categorisation of lower (unskilled or unemployed) or higher (semi-skilled, skilled or professional), and the primary language spoken at home (multilingual vs English only).

Main outcome measures

At 8 years of age, participants were assessed by paediatricians and psychologists who were unaware of group allocation. Details of the 8 year assessments have been reported elsewhere.11 Cerebral palsy (CP) was diagnosed in children with abnormal tone and loss of motor function, and its severity was determined by a functional classification (1991–1992 cohort) or the Gross Motor Function Classification Scale (GMFCS)13 (1997 and 2005 cohorts). Blindness was defined as having visual acuity <6/60 in the better eye, and deafness comprised a hearing impairment requiring amplification or a cochlear implant or worse. Cognitive ability was assessed using the Wechsler Intelligence Scale for Children–Third Edition for the 1991–1992 cohort,14 the WISC-IV for the 1997 cohort15 and the Differential Ability Scales–Second Edition16 for the 2005 cohort. To standardise the IQ scores from the different assessments that were administered, z-scores of general intelligence (IQ) were computed relative to the mean (SD 15) scores for the controls for each cohort, weighted for the distribution of social risk variables within each cohort. Children too impaired to complete the cognitive tests were assigned an IQ z-score of −4 SD. Major neurosensory disability comprised any of moderate/severe CP (unable to walk or walking with considerable difficulty, with or without appliances, or GMFCS levels 2–5), blindness, deafness or an IQ less than −2 SD. Characteristics of the control groups have been reported previously.17

Statistical analyses

Data were analysed using STATA V.14.1.18 Group differences were compared using generalised estimating equations, with robust SEs to allow for clustering of children because of multiple births.19 Where the models would not converge, logistic regression models were used. To assess the independent effect of neonatal surgery on 8-year outcomes, analyses were repeated adjusting for potential confounding perinatal and sociodemographic variables including outborn status, antenatal corticosteroids, gestational age at birth, sex, birth weight z-score, patent ductus arteriosus (PDA), exogenous surfactant, necrotising enterocolitis (NEC), grade 3 or 4 intraventricular haemorrhage (IVH), cystic periventricular leucomalacia (PVL), postnatal corticosteroids, oxygen dependency at 36 weeks, age of the child when assessed at 8 years, age of the mother at the birth of the child, lower maternal education, lower social class and multilingual households. To determine trends over time, the 2005 cohort was compared with both the 1991–1992 and 1997 cohorts, using an interaction term between group and era in the analyses. Comparisons are presented as ORs or mean differences, both with 95% CI.

The studies have been approved by the human research ethics committees at the Royal Women’s Hospital, the Mercy Hospital for Women, Monash Medical Centre and The Royal Children’s Hospital, Melbourne. Written informed consent was obtained from the parents of controls and for the 2005 EP cohort; follow-up was considered routine clinical care for EP children in earlier cohorts.


Overall, 546 infants born 22–27 weeks' gestation survived to 8 years of age. The rates of surgery in survivors were similar in the three eras (1991–1992: 28% (63/225), 1997: 34% (51/151) and 2005: 34% (57/170)). Among those who had surgery, the most common operations were inguinal hernia repair (n=60; 37%), ligation of a PDA (n=55; 34%), treatment of retinopathy of prematurity (ROP; n=27; 17%; 12 treated with cryotherapy, 13 treated with laser and one with both), bowel surgery (n=26; 16%) and surgery for ventricular dilatation (n=11; 7%; 10 with Rickham’s reservoir, of whom six went on to a ventriculoperitoneal (VP) shunt, and one with VP shunt only).

Children who had surgery were more immature and lighter at birth and were more often male, and overall, they were sicker, with more receiving surfactant, having a PDA, NEC, grade 3 or 4 IVH, cystic PVL or being oxygen dependent at 36 weeks (table 1). Sociodemographic variables, other perinatal variables and age at assessment did not differ substantially between the groups (table 1).

Table 1

Participant characteristics contrasted between extremely preterm children who received surgery in the newborn period and those who did not

Overall, 91% (499/546) of survivors were assessed for disability at 8 years. There was no difference between those who were assessed and not assessed in either the surgical or non-surgical groups, with the exception that those assessed within the surgical subgroup were less mature but had higher birth weight z-scores than those who were not assessed (online suppleme ntary file 1). Of those assessed, 17% (86/499) had a major disability at 8 years; 64 children had an IQ less than −2 SD, 33 had moderate or severe CP, 11 were deaf and 5 were blind. Rates of major disability were higher in those who had surgery compared with those who did not, within each era (table 2). The rates of major disability in the surgical groups were similar across eras (table 2).

Supplementary file 1

Table 2

Rates of major disability at 8 years of age in the surgical and non-surgical groups

Rates of major disability were increased in those who had surgery among all the common types of operations; however, the evidence was weaker for an association with bowel surgery (table 3). The unadjusted OR was highest for ventricular drainage surgery, but few infants had this surgery. After adjustment for perinatal and social variables, the evidence for the associations of major disability with surgery and its subtypes weakened, and only the association with any surgery remained significantly associated with major disability (table 3). In multivariable models, there was little evidence that the risk of major disability was independently related to being outborn, antenatal corticosteroid administration, gestational age at birth, birth weight z-score, exogenous surfactant, PDA, NEC, oxygen dependency at 36 weeks, corrected age of the child when assessed, mother’s age at birth of the child or lower maternal education. Moreover, there were no differences in the rates of disability across the eras (1991–1992, 18%; 1997, 15%; 2005, 18%). However, the risk of major disability was significantly higher in males; those with postnatal corticosteroids, grade 3 or 4 intraventricular haemorrhage or cystic periventricular leucomalacia; those in multilingual families; and those with lower social class (online supplementary table 2).

Table 3

Major disability among surgery subtypes

Of the two the most common impairments leading to major disability, low IQ or CP, any surgery was independently related to IQ scores less than −2 SD (adjusted OR 2.38, 95% CI 1.16 to 4.89, p=0.018) and to moderate or severe CP (adjusted OR 4.02, 95% CI 1.14 to 14.2, p=0.031), after adjustment for confounding variables. Inguinal hernia surgery was not independently related to either IQ scores less than −2 SD (adjusted OR 1.75, 95% CI 0.75 to 4.09, p=0.20) or to moderate or severe CP (adjusted OR 1.36; 95% CI 0.32 to 5.84, p=0.68). PDA ligation was not independently related to IQ scores less than −2 SD (adjusted OR 2.22, 95% CI 0.88 to 5.60, p=0.09) or to moderate or severe CP (adjusted OR 0.29, 95% CI 0.04 to 2.28, p=0.24). There were too few subjects with deafness or blindness to warrant multivariable analyses; however, four of the five blind children had been treated with either cryotherapy or laser.


Approximately one-third of surviving infants born EP had a surgical procedure during their primary hospitalisation after birth. For infants who underwent any surgery, the rates of major disability at 8 years were substantially increased compared with the rate among non-surgery-exposed controls even after adjustment for confounders. Having reported previously that EP children who underwent surgery were at increased risk for impaired neurodevelopment at 5 years of age compared with those who did not,6 our findings extend the knowledge that impairment persists until at least 8 years of age and that this risk is not diminishing over time.

In the current study, the power to detect associations between neurodevelopmental impairment and different types of surgery was low given the relatively small sample sizes for individual procedures. Prior to adjustment for postnatal confounders, PDA ligation appeared to be independently associated with neurodevelopmental impairment as has previously been reported in the literature.20–22 However, Weisz et al reported results of another retrospective cohort study in which postnatal confounders of various neonatal morbidities were adjusted for, motivated by awareness that previous retrospective studies failed to make such statistical adjustments. By minimising survival bias in the ligated group, they found no difference in outcomes of death, disability or neurodevelopmental impairment between those EP infants treated medically or surgically for PDA.23 When a similar analytic model was applied in our study, surgical management of PDA was no longer associated with poor neurodevelopment. Foglia and Schmidt advised ongoing caution for anything other than restrained surgical ligation of PDA, suggesting that the best path forward in the balancing of bias between surgically and medically treated PDA is the conduct of a prospective randomised controlled trial comparing the two therapies head to head.24

Interestingly, bowel surgery did not confer an independent increased risk of major neurosensory disability, suggesting that the systemic inflammatory response associated with conditions requiring bowel surgery, such as NEC, was not mediating adverse outcome as strongly as others’ have observed.25–27 The lack of an independent adverse effect of neurosurgery, particularly given the precursor of direct cerebral injury, is probably explained by the lack of power, given that only 11 participants had such surgery over the three eras described.

Over the period of observation of the current study, reductions in mortality have been reported for term born infants with specific congenital anomalies requiring surgery, like congenital diaphragmatic hernia.28 29 However, despite significant advances in perinatal care, such as the administration of antenatal corticosteroids, and the use of exogenous surfactant, which reduce morbidities, including PDA, IVH or NEC that commonly result in referral for surgery,30 31 the need for surgical intervention for the complications of prematurity has not decreased over time. In our study, it was the sickest and smallest babies who developed morbidities requiring surgery, and perhaps, these advances are least likely to protect the most fragile.

Disentangling the factors related to surgical exposure in preterm infants that might contribute to adverse outcomes is difficult, with a number of possible mechanisms for cerebral injury postulated. These include increased severity of illness for those infants who require surgery as a consequence of prematurity, exposure to potentially neurotoxic anaesthetic drugs,32 cardiovascular instability during anaesthesia resulting in impaired cerebral circulation33 and injury to the immature oligodendrocytes by a systemic inflammatory response associated with surgery. It may be argued following the large General Anesthesia compared to Spinal anesthesia (GAS) trial, which reported no difference in neurodevelopment at 2 years of age in infants who underwent surgery with general compared with regional anaesthesia, that the anaesthetic agents themselves do not explain the adverse effects of surgery seen in the current study.34 However, the surgeries in the GAS trial were conducted at approximately term corrected age and infants born before 27 weeks’ gestation were excluded. Neurobiologically, it is reasonable to postulate that the most immature babies are the most vulnerable to the neurotoxic effects of anaesthesia and surgery. A limitation of the current study is that we have no data on any of these variables, apart from gestational age, adjustment for which did not alter any conclusions.

Major strengths of this study are the high rates of follow-up of sequential geographical cohorts, prospectively recruited over three distinct time periods. What is clear, both from the current study and from a meta-analysis of developmental outcomes for infants requiring non-cardiac surgery,35 is that infants exposed to surgery early in life are at increased risk for neurosensory impairment and that this risk has not reduced over time despite improvements in perinatal care. Surgery early in life is performed in life-saving circumstances or to mitigate risk of impairment. The persistence of the effect over time suggests a lack of appreciation of the long-term risks of early surgery, with a consequent lack of awareness about the need for neuroprotection for these at-risk patients. With survival rates now high, attention to long-term morbidity is critical. Investigation into the neurobiological substrates of later disability has begun. The anaesthetic community has embraced their potential role as mediators of poor outcome enthusiastically and continues to explore improvements in clinical care.36 It is imperative now that the neonatal surgical community seriously investigates how risk can be reduced through the employment of novel neuroprotective strategies alongside further improvements in perinatal care.

A limitation of the current study is that the data were collected in an era when there was little awareness of components of the risk that may be conferred by exposure to surgery early in life. Specifically, no data were collected pertaining to anaesthetic exposure and the details of agents used, although almost all of the procedures, with the exception of ROP surgery, were performed under general rather than regional anaesthesia. We also did not record the exact age of the child when they had their surgery. Most operations were performed in operating theatres, with the exception of laser therapy for ROP, and some duct ligations in more recent eras.

There are at least two important reasons why it is necessary to know if surgery is related to adverse outcomes. The first reason is that we need to know if surgery remains an association with adverse neurodevelopment over a time of rapid changes in perinatal and neonatal care because we need to try to disentangle the role, if any, that surgery may be having on the high rates of adverse outcomes for EP children. The best way to establish if any operation is causing harm is a randomised controlled trial (RCT). Although RCTs are difficult in neonatal intensive care, particularly since surgery is deemed to be urgent and life-saving in many cases, it is still possible to consider RCTs for some of the more common operations. One of the the most common operations is ligation of the ductus arteriosus, which is rarely performed for life-saving reasons and which is eminently suitable for a proper RCT comparing surgery with no surgery and with long-term neurodevelopmental outcome as the primary endpoint. The second reason is to be able to identify the highest risk infants for early developmental care and more intensive follow-up in situations where resources for such care are limited. Early intervention over the first months of life at home, either starting in the nursery or after discharge, improves long-term neurodevelopmental outcomes in high-risk infants.37


EP infants who require surgery prior to discharge remain at significant risk of adverse neurodevelopmental outcome, and this risk persists at least into mid-childhood. Any surgery, whether minor or life-saving, early in life, confers an increased risk of long-term disability for infants born EP. Surgery in the neonatal period is only performed under circumstances of necessity to preserve life or vital functions; however, attention must now be directed toward neuroprotective strategies for this vulnerable group.



  • RWH contributed equally.

  • Contributors RWH: conception and design of the study, data interpretation, drafting and revising the article and approval of the final manuscript as submitted. LMH: conception and design of the study, data interpretation, drafting and revising the article and approval of the final manuscript as submitted. ACB: conception and design of the study, data interpretation, drafting and revising the article and approval of the final manuscript as submitted. PJA: conception and design of the study, data interpretation, drafting and revising the article and approval of the final manuscript as submitted. JLYC: conception and design of the study, data interpretation, drafting and revising the article and approval of the final manuscript as submitted. LWD: conception and design of the study, obtaining funding, data analysis and interpretation, drafting and revising the article and approval of the final manuscript as submitted.

  • Funding Supported by grants from the National Health and Medical Research Council of Australia (Centre of Clinical Research Excellence #546519; Centre of Research Excellence #1060733; Early Career Fellowship #1053787 to JC; Senior Research Fellowship #1081288 to PJA) and the Victorian Government’s Operational Infrastructure Support Programme.

  • Competing interests None declared.

  • Patient consent Parental/guardian consent obtained.

  • Ethics approval Human Research Ethics Committee, The Royal Children’s Hospital, Melbourne.

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

  • Collaborators Members of the Victorian Infant Collaborative Study Group Convenor: Jeanie Cheong.1,2,3,4 Collaborators (in alphabetical order): Peter Anderson,2,4,5 Alice Burnett,2,4,5,6 Catherine Callanan,4 Elizabeth Carse,7 Margaret P. Charlton,7 Noni Davis,4 Lex W Doyle,1,2,3,4,5 Julianne Duff,4 Leah Hickey,6 Esther Hutchinson,4,6 Marie Hayes,7 Elaine Kelly,4,8 Katherine J Lee,9 Marion McDonald,4 Gillian Opie,8 Gehan Roberts,2,4,5,10 Amanda Williamson8. (1) Neonatal Services, Royal Women’s Hospital, Melbourne, Australia; (2) Victorian Infant Brain Studies, Murdoch Childrens Research Institute, Melbourne, Australia; (3) Department of Obstetrics & Gynaecology, University of Melbourne, Melbourne, Australia; (4) Premature Infant Follow-up Program, Royal Women’s Hospital, Melbourne, Australia; (5) Department of Paediatrics, University of Melbourne, Melbourne, Australia; (6) Neonatal Medicine, The Royal Children’s Hospital, Melbourne, Australia; (7) Newborn Services, Monash Medical Centre, Melbourne, Australia; (8) Neonatal Services, Mercy Hospital for Women, Melbourne, Australia; (9) Clinical Epidemiology and Biostatistics, Murdoch Childrens Research Institute, Melbourne, Australia; (10) Centre for Community and Child Health, The Royal Children’s Hospital, Melbourne, Australia.