Objectives To compare mortality and serious morbidity rates between outborn and inborn livebirths at 22–27 weeks' gestation.
Design Population-based cohort study.
Setting Victoria, Australia.
Patients Livebirths at 22–27 weeks' gestation free of major malformations in 2010–2011.
Interventions Outcome data for outborn (born outside a tertiary perinatal centre) infants compared with inborn (born in a tertiary perinatal centre) infants were analysed by logistic regression, adjusted for gestational age, birth weight and sex.
Main outcome measures Infant mortality and serious morbidity rates to hospital discharge.
Results 541 livebirths free of major malformations were recorded. By 1 year, 49 (58%) outborns and 140 (31%) inborns died (adjusted OR (aOR) 2.78, 95% CI 1.52 to 5.09, p=0.001). In total, 445 infants were admitted to neonatal intensive care unit (NICU); 93 died by 1 year (14/49 outborns and 79/396 inborns), (aOR 1.75, 95% CI 0.87 to 3.55, p=0.12). There were no significant differences in rates of necrotising enterocolitis, intraventricular haemorrhage, retinopathy of prematurity, bronchopulmonary dysplasia (BPD) or the combined outcome of death or BPD in outborn infants compared with inborn infants. Outborns had an increased risk of cystic periventricular leukomalacia (cPVL) compared with inborns (12.2% vs 2.8%, respectively; aOR 5.34, 95% CI 1.84 to 15.54, p=0.002).
Conclusions Mortality rates remained higher for outborn livebirths at 22–27 weeks' gestation compared with inborn peers in 2010–2011. Outborn infants admitted to NICU did not have substantially different rates of mortality or serious morbidity compared with inborns, with the exception of cPVL. Longer-term health consequences of outborn birth before 28 weeks' gestation need to be determined.
- Extremely preterm
Statistics from Altmetric.com
What is already known on this topic?
Outborn extremely preterm (EPT) infants have an increased risk of mortality compared with inborn peers.
Birth in a tertiary perinatal centre is recommended for births <28 weeks' gestation because of the survival and quality of survival advantages.
The incidence of outborn livebirths <28 weeks' gestation has been increasing in Victoria in the past decade.
What this study adds?
EPT outborn infants are less likely to be admitted to a neonatal intensive care unit (NICU) following livebirth compared with EPT inborn infants.
Short-term outcomes have improved for EPT outborn infants admitted to NICU compared with previous cohorts in Victoria.
The presence of a dedicated neonatal retrieval team at EPT births may reduce mortality and serious morbidity risk, especially in low-level non-tertiary hospitals.
Extremely preterm (EPT) birth <28 weeks' gestation occurs in fewer than 1% of pregnancies in Australia1 but remains a leading cause of neonatal mortality. In 2010–2011, 50% of neonatal deaths were attributed to spontaneous preterm birth.1 ,2 Benchmarking outcome data for Australian-born infants with international cohorts can be challenging, as wide variations in mortality and morbidity are reported, even from countries with similar systems of regionalised perinatal care.3–7 Such variations have been attributed to differing legal requirements for reporting births at the lower limits of viability,8 ,9 differing ethical views and practices regarding resuscitation and neonatal intensive care admission for births <25 weeks3 ,10–14 and differing inclusion criteria. As parent counselling and decision making about providing or withholding intensive perinatal interventions is influenced by such statistics, regularly updating mortality and morbidity data for EPT infants is essential.
Disparities in mortality risk for outborn compared with inborn EPT and extremely low birthweight infants have been reported in Australia since the late 1970s15–17 and more recently in international population-based cohorts.18–22 In Victoria, Australia, the gap between outborn and inborn mortality risks narrowed in the early 2000s; however, in 2006–2009, the gap widened again. Moreover, while inborn mortality declined, from 29% in 2001–2005 to 25% in 2006–2009, outborn infant mortality rose, from 51% to 58%.23 Given these trends, an updated investigation of outcomes for these infants was warranted.
Accurate reporting of mortality requires the inclusion of all livebirths, including livebirths who die without being admitted to a neonatal intensive care unit (NICU).9 ,24 Our aims were to report rates of infant mortality and serious morbidity to hospital discharge for all livebirths at 22–27 weeks' gestation in a geographically defined population, comparing infants born outside of tertiary perinatal centres (‘outborn’) with infants born in tertiary centres (‘inborn’).
We conducted a population-based cohort study of all livebirths at 22+0 to 27+6 weeks' gestation in the state of Victoria, Australia, from 1 January 2010 to 31 December 2011. Livebirths in any of the 70 hospitals with Level 1 or Level 2 obstetric and special care nursery (SCN) services were coded as outborn. Livebirths in non-obstetric hospitals and births before arrival at a hospital were also coded as outborn. Livebirths in all three tertiary perinatal centres with NICUs were coded as inborn. Terminations of pregnancy were excluded. Morbidity data were collected to the time of hospital discharge, including after back transfer to a Level 1 or 2 SCN, and mortality data to 1 year of age. Livebirths with major congenital malformations and chromosomal abnormalities who died were excluded from these analyses as birthplace (outborn vs inborn) would not have altered their outcome. Major malformations were coded according to the International Classification of Diseases, Tenth Revision,25 and the Australian and New Zealand Neonatal Death Classification System.26
Morbidity was defined as necrotising entercolitis (NEC; any stage), intraventricular haemorrhage (IVH; all grades), cystic periventricular leukomalacia (cPVL) and retinopathy of prematurity (ROP; all stages) according to international criteria.27–29 Bronchopulmonary dysplasia (BPD) was defined as requiring supplemental oxygen and/or respiratory support (intermittent mandatory ventilation, continuous positive airway pressure or high flow) at 36 weeks' postmenstrual age.30 Rates of survival to discharge home without severe morbidity were also reported. Severe morbidity was defined as NEC (stage 2 or 3), IVH (grade 3 or 4), cPVL, ROP requiring laser or BPD requiring home oxygen.
Data on all livebirths and infant deaths were sourced from the Consultative Council on Obstetric and Paediatric Mortality and Morbidity (CCOPMM), which obtains data prospectively on all births in the state. Multiple data sources, including the state-wide Paediatric Infant Perinatal Emergency Retrieval service (PIPER) and obstetric databases at all three tertiary perinatal centres, were used to cross-reference CCOPMM data and to obtain prospectively collected perinatal data. Mortality and morbidity data were obtained from the Australian and New Zealand Neonatal Network (ANZNN) and NICU databases at the tertiary NICUs. Data were entered into EpiData (Data Management and Analysis System, 2002, Odense, Denmark) after 1-year follow-up had been completed.
Data were imported into STATA (V.14.0, StataCorp, College Station, Texas, USA) for analysis. Standard deviations (SDs) for birth weight (Z-scores) were calculated for each infant.
Morbidity to hospital discharge and mortality data comparing outborns with inborns were analysed by logistic regression, adjusted for gestational age, birth weight (Z-scores) and sex. Adjusted odds ratios (aORs), 95% CIs and p values were obtained. Perinatal risk factors independently associated with infant mortality were identified using a multiple logistic regression model.
In 2010–2011, 796 births were recorded at 22–27 weeks' gestation in Victoria, excluding terminations of pregnancy. Of these, 577 (72.5%) were liveborn, including 36 with major congenital malformations who subsequently died. Of 541 livebirths free of major malformations, 84 (15.5%) were outborn. The outcomes of all 577 livebirths are shown in figure 1.
Overall, 445 (82%) livebirths free of major malformations were admitted to NICU. The proportion of NICU admissions increased with increasing gestational age, from 5% (n=2/40) at 22 weeks to 97% (n=129/133) at 27 weeks. Overall, 58% (49/84) of outborn livebirths were admitted to NICU, compared with 87% (396/457) of inborns. Excluding the 22-week infants, 68% (48/71) of outborns and 92% (395/430) of inborns were admitted to NICU. Differences in admission rates were most evident at 23 weeks’ gestation (11% outborn vs 39% inborn) and at 24 weeks' gestation (67% vs 87%). All infants not admitted to NICU died.
Perinatal and infant characteristics of NICU admissions are shown in table 1. Higher proportions of mothers of outborns were teenagers, multigravida and had a previous preterm birth compared with mothers of inborns. They were less likely to have prelabour rupture of the membranes or to have received antenatal corticosteroids.
The majority of outborns (78%) admitted to NICU were born in a Level 2 hospital, with only 8% born in a Level 1 hospital and 2% in a non-obstetric hospital. These infants were transferred to a tertiary NICU by a dedicated neonatal retrieval team from PIPER, all within 12 hours of birth. Ambulance paramedics transferred six infants born before arrival (BBA) at hospital to a tertiary centre NICU, as PIPER does not attend out-of-hospital births.
Mortality rates in all outborn and inborn livebirths
Mortality by outborn/inborn status, gestational age, time of death and NICU admission status is shown in table 2.
Overall, 189/541 (35%) livebirths died by 1 year. Mortality was 100% at 22 weeks, 80% at 23 weeks, 51% at 24 weeks, 26% at 25 weeks, 14% at 26 weeks and 13.5% at 27 weeks. Of the 189 deaths, 96 (51%) were livebirths not admitted to NICU.
For livebirths admitted to NICU, infant mortality was 100% at 22 weeks, 31% at 23 weeks, 42% at 24 weeks, 25% at 25 weeks, 13% at 26 weeks and 11% at 27 weeks. In total, 93/445 (21%) infants admitted to NICU died. There were 64 neonatal deaths in NICU and 29 postneonatal infant deaths, including eight postdischarge home.
Mortality risk according to outborn/inborn birth status
Overall mortality risk was 2.8 times higher for outborn compared with inborn livebirths. By 1 year, 49/84 (58%) outborns compared with 140/457 (31%) inborns died (aOR 2.78, 95% CI 1.52 to 5.09, p=0.001). Even excluding 22-week infants from the analysis, the odds of mortality for outborn 23–27 week infants remained 2.7 times that of inborn peers (aOR 2.70, 95% CI 1.49 to 4.92, p<0.001).
When we compared infant mortality risk in outborns versus inborns admitted to NICU, we found no difference between the two groups. By 1 year, 14/49 (29%) outborn infants compared with 79/396 (20%) inborn infants died. There were no substantial differences in the odds of neonatal death (aOR 2.20, (95% CI 0.95 to 5.07), p=0.065), death before discharge (aOR 1.68, (95% CI 0.77 to 3.70), p=0.19) or death before 1 year of age after NICU admission (aOR 1.75 95% CI 0.87 to 3.55, p=0.12).
In infants admitted to NICU, risk factors independently associated with death before hospital discharge were diminishing gestational age (p<0.001) or birth weight (p=0.001) and male sex (p=0.04). Outborn birth, no antenatal corticosteroid exposure, being a multiple and vaginal birth were not independently associated with mortality.
Outborn infants retrieved by a dedicated neonatal retrieval team and transferred to a tertiary NICU had lower overall infant mortality rates (9/43, 21%) compared with infants transferred to a tertiary NICU by ambulance paramedics following out-of-hospital birth (5/6, 83%). Compared with birth in a tertiary centre, out-of-hospital birth was associated with more than an 11-fold increased risk of mortality (aOR 11.50, 95% CI 1.21 to 109.3, p=0.03).
Rates of serious morbidity to death or discharge home for infants admitted to NICU are shown in table 3. There were no significant differences in rates of NEC (any stage), IVH (any grade), ROP, the combined outcome of death or ROP, BPD, the combined outcome of death or BPD, discharge home on oxygen or length of stay in outborn infants compared with inborn infants admitted to NICU. However, outborns had an increased risk of PVL compared with inborns.
Morbidities to hospital discharge for 360 survivors are shown in table 4. Overall, 242/360 (67%) NICU survivors (45% of livebirths) were discharged home without severe morbidity, with rates rising with gestational age.
In our study of 22–27 weeks' gestation livebirths in Victoria in 2010–2011, birth outside a tertiary centre was associated with a 2.8-fold increased risk of infant mortality compared with birth in a tertiary centre. Overall, 58% of outborns died; the same proportion as in 2006–2009.23 Furthermore, the risk of mortality for outborn livebirths compared with inborn livebirths was marginally higher than we reported for the 2006–2009 cohort in Victoria (aOR 2.78 in 2010–2011 vs aOR 2.66 in 2006–2009).23
We speculated higher overall proportions of 22-week outborn livebirths and deaths might explain the differences in mortality risk, given tertiary clinicians in Victoria are reluctant to accept in utero transfers at 22 weeks in the absence of maternal indications. However, this did not prove to be true. More than two-thirds of 22-week livebirths were inborn and infant mortality at 22 weeks' gestation was 100%, irrespective of birthplace. However, only two infants were admitted to NICU. Our NICU admission and survival data for 22-week livebirths were similar to France (EPIPAGE-2, 2011 cohort)31 but lower than in the USA (2006–2011)12 or Japan (2003–2005).5
Moreover, even without 22-week infants in our analysis, the odds of mortality for outborn 23–27 week infants were still nearly 3 times that of inborn peers.
After 22 weeks' gestation, significant differences in mortality risk were most evident at lower gestational ages (23–24 weeks) with higher rates of outborn deaths at the birth hospital and lower rates of NICU admission compared with inborn livebirths. Insufficient time to administer antenatal corticosteroids, suboptimal resuscitation and lack of equipment and staff expertise have been identified as risk factors contributing to mortality in these infants.17 In this study, resuscitation data for outborn livebirths were only available for those referred to PIPER and not for 31 of 35 outborn livebirths who died at the birth hospital without being referred. Due to confidentiality of the data provided, it is not possible to obtain information on decision making or other aspects of care for outborn livebirths who were not referred to PIPER. Only four infants who died during resuscitation attempts were referred. The high rate of outborn deaths without referral warrants further investigation.
Differing attitudes and practices with regard to resuscitation and provision of neonatal intensive care for infants born in the ‘grey zone’ may have played a role.11 ,12 ,32 When EPT infants are perceived to have poor survival chances and perinatal interventions, including in utero transfer, are withheld, high mortality risk becomes a self-fulfilling prophecy.10 We have shown that many perinatal healthcare providers in Victoria underestimate survival and overestimate major neurosensory disability risk in EPT outborn infants.33 However, without knowledge of the decision-making process for each outborn birth, we can only hypothesise that opportunities to transfer in utero or request PIPER transfer may have been disregarded because outcome was perceived to be poor.
In this study, only 58% of outborn livebirths were admitted to a tertiary NICU. However, of those admitted, 78% survived to 1 year. This was the highest survival rate reported in Victoria in 30 years.15 ,16 We found no difference in mortality risk comparing outborns with inborns admitted to NICU. However, we acknowledge our study was underpowered, with only 49 outborns admitted to NICU. We also found outborn infants admitted to NICU had similar risks of serious morbidity compared with inborns, with the exception of PVL, even after adjusting for perinatal risk factors. The comparable rates of BPD and the combined outcome of death and/or BPD were unexpected, given the low antenatal corticosteroid exposure rates in the outborns (6% complete course). The significantly higher risk of PVL in outborns was concerning and warrants further investigation.
We attribute the 30-year high in survival rates in outborn NICU-admitted infants to a number of factors. In Victoria, EPT infants are preferentially admitted to one of three tertiary perinatal centre NICUs over the paediatric hospital NICU. In this cohort, 94% of outborns were admitted to a perinatal centre NICU, with only 6% admitted to the paediatric hospital NICU. As there are only four NICUs for the entire state, patient volumes in the perinatal centre NICUs are in the order of 120–200 admissions at 22–27 weeks per year. Preferential admission of EPT infants to a perinatal centre NICU with high volume of admissions is known to be associated with improved outcomes.3 ,34
The majority of outborns transferred to a tertiary NICU after birth (78%) were born in Level 2 hospitals. These hospitals have paediatricians on staff and facilities to provide short-term ventilation support until the retrieval team arrives. Arguably, these infants had access to human and equipment resources not available in lower level hospitals, resulting in improved quality of resuscitation and stabilisation. To illustrate, 50% of infants referred to PIPER from Level 2 hospitals were intubated before PIPER arrived and placed onto an infant ventilator while awaiting retrieval. In Level 1 hospitals, there is no infant ventilator for stabilisation. General practitioners (GPs) or GP obstetricians provide obstetric and neonatal care. With limited exposure to EPT infants and limited equipment to stabilise these infants, resuscitation is less likely to be successful.
In Victoria, PIPER provides neonatal retrieval services for the entire state. In an attempt to further reduce mortality and morbidity associated with outborn birth, PIPER aims to be present at births <29 weeks in non-tertiary hospitals to resuscitate and stabilise the infant whenever feasible. Of the 43 infants retrieved by PIPER, a team was present at 17 (39%) births. More importantly, a team was requested and was present at three of four births in Level 1 hospitals, where there was no paediatrician on staff. Improved quality of resuscitation is reported when dedicated neonatal retrieval teams resuscitate outborn infants in the delivery room.35 PIPER provided intensive care interventions such as surfactant administration at the birth hospital (42/43 infants), respiratory function monitoring and blood gas monitoring at the birth hospital and during transport. No infant died during transport to the receiving NICU. The combination of all these factors is likely to have contributed to reducing mortality risk in these infants in 2010–2011.
In spite of positive outcomes for outborns transferred to a tertiary centre after birth by PIPER, we caution against complacency in efforts to transfer high-risk pregnancies in utero. The survival advantage associated with tertiary centre birth should not be overlooked, irrespective of the ability to successfully transfer these infants ex utero. In this cohort, infants born in tertiary centres and admitted to NICU following in utero transfer had mortality rates similar to inborn booked tertiary centre births (19% and 18%, respectively). In utero transfers accounted for 60% of inborn births.
The high mortality rate in BBAs transferred to a tertiary centre by ambulance paramedics was expected. Unlike the French model,36 PIPER does not attend births in residential homes. In this cohort, six BBAs born at 22, 23, 24, 25 (×2) and 26 weeks, respectively, were admitted to NICU and only one survived. Given the limitations of ambulances equipped primarily to carry adults and older children, morbidities including severe hypothermia and hypoglycaemia associated with BBA birth are common,37 and improving outcomes in these infants remains a challenge.
As we were unable to find a contemporaneous study reporting outborn and inborn morbidity and mortality, we compared outcomes of all EPT livebirths in Victoria in 2010–2011 with two international population-based cohorts: EPIPAGE-2 (France, 2011) and EPICure-2 (England, 2006). Overall, 35% of 22–27 week livebirths and 19% of infants admitted to NICU in Victoria died before NICU discharge. In France in 2011, 39% of 22–27 week livebirths and 26% of infants admitted to NICU died before discharge.31 We were only able to compare serious morbidity rates in these two cohorts when reporting criteria were identical. We found rates of severe NEC (stage 2 or 3) were 3.3% in Victoria versus 5.2% in France, severe IVH (grade 3 or 4) were 5% versus 11%, respectively, and cPVL 3.3% versus 2.5%, respectively. We could not compare severe BPD or survival without major morbidity because reporting criteria differed between the studies.
We also compared outcomes with 22–26 week livebirths in England4 (EPICure-2 cohort), acknowledging these births were in 2006. Despite centralisation of perinatal care in England in the early 2000s, 43.6% of the 2006 EPICure-2 cohort were outborn, compared with 16.4% of 22–26 week livebirths in Victoria in 2010–2011. Overall mortality for all 22–26 week livebirths in England was 49% compared with 40% in Victoria. In England, 38% of 22–26 week NICU admissions died before discharge, whereas our rate was 23%. However, higher NICU admission rates were reported in England (83% compared with 77% in Victoria). Survival without major morbidity could not be compared because reporting criteria differed between studies.
Strengths and limitations
The strength of our study is that it was population based, comprising all livebirths in the state over a finite and contemporary time, including livebirths that died and were not admitted to NICU. Unlike single-centre studies and those only reporting outcome data for inborn births or for infants admitted to NICU, our study is not subject to selection bias and true mortality rates are reported.24 Furthermore, our study was unique as few studies report morbidity and mortality data comparing outborn and inborn infants from population-based cohorts. We also reported deaths post-discharge, up to 1 year after birth.
Our study was limited by the inability to differentiate between outborns who died at the birth hospital despite resuscitation attempts from deaths following a decision to limit intensive interventions. Given 37% of outborn infants died at the birth hospital without being referred to PIPER, an exploration of the decision-making process regarding provision of perinatal interventions for these infants would have been valuable.
Infant mortality rates remain higher for outborn compared with inborn 22–27 week livebirths in Victoria as they have done for more than three decades. However, short-term outcomes were improved for outborn infants offered neonatal intensive care. Longer-term health consequences from outborn birth before 28 weeks' gestation need to be determined.
The authors wish to thank Dr Mary-Ann Davey at the Clinical Council's Unit, Health Service Programs Branch, Victorian Department of Health and Human Services, for providing the denominators for all 22–27 week livebirths and infant deaths in Victoria in 2010–2011. The authors express their gratitude to Kaye Bawden, Tania Fletcher, Esther Wong, Jo Brooks and Lynne Rigg for their assistance extracting data at the tertiary centres, Karen Watson for assisting with data extraction from PIPER transport records and Rae Crossman, Karen Watson and Sarah-Pia Carron for their assistance entering data into EpiData.
Correction notice This paper has been amended since it was published Online First. An error was introduced into the legend of table 1 by the production office. The words ‘livebirths at’ were inadvertently introduced. These words have now been deleted.
Twitter Follow Rosemarie Boland at @roseandtoby
Contributors RAB designed the study, wrote the ethics, collected, cleaned and analysed the data and wrote the draft manuscript. LWD, JAD and PGD supervised and contributed to all stages of the study design and statistical analysis and edited the manuscript. The content has not been reproduced from another source, other than the points of text that are referenced as part of the literature review. Each author has reviewed the manuscript and approved the submission of this revised version. The authors take full responsibility for the manuscript.
Funding RAB was the recipient of the Felix Meyer Faculty Research Scholarship from the Faculty of Medicine, Dentistry and Health Sciences at the University of Melbourne. Her PhD research was also supported with funding from a National Health and Medical Research Council (NHMRC) Centre for Research Excellence in Newborn Medicine at the Murdoch Childrens Research Institute (ID 1060733). RAB currently holds a Postdoctoral Fellowship, funded by the Murdoch Childrens Research Institute. JAD and PGD are funded by the NHMRC: JAD holds an Early Career Fellowship (ID 1059111); PGD holds a Practitioner Fellowship (ID 1012686).
Competing interests None declared.
Ethics approval The following Human Research and Ethics Committees approved the study: (1) CCOPMM. Victorian Government Department of Health and Human Services. Study ID: RR13-31. (2) Royal Children's Hospital, Parkville, Victoria, including the PIPER service. ID 31124A. (3) Mercy Health, Heidelberg, Victoria. Study ID: R11/28. (4) Southern Health: Clayton, Victoria. Study ID: 11320Q. (5) Royal Women's Hospital, Parkville, Victoria. Study ID: 11/06.
Provenance and peer review Not commissioned; externally peer reviewed.
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.