Improved Outcomes of Extremely Premature Outborn Infants: Effects of Strategic Changes in Perinatal and Retrieval Services
OBJECTIVE. The goal was to evaluate the impact of statewide coordinated changes in perinatal support and retrieval services on the outcomes of extremely premature births occurring outside perinatal centers in the state of New South Wales, Australia.
METHODS. The intervention included additional, network-coordinated, perinatal telephone advice to optimize in utero transfers and centralization of the neonatal retrieval system, with preferential admission of retrieved infants (outborn infants) to perinatal centers instead of freestanding pediatric hospitals, from the middle of 1995. Population birth and NICU admission cohorts of infants of 23 to 28 weeks of gestation were studied. Outcomes of epoch 1 (1992 to the middle of 1995; 1778 births and 1100 NICU admissions) were compared with those of epoch 2 (1997–2002; 3099 births and 2100 NICU admissions), after an 18-month washout period.
RESULTS. There were 25% fewer nontertiary hospital live births (19.7% vs 14.9%) and more prenatal steroid use. Despite an 11.4% average annual increase in NICU admissions between the 2 epochs, fewer infants were outborn (12.0% vs 9.3%) and outborn mortality rates decreased significantly (39.4% vs 25.1%), particularly for those between 27 and 28 weeks of gestation. The overall improvement was equivalent to 1 extra survivor per 16 New South Wales births. There were also significantly fewer serious outcome morbidities in outborn infants during epoch 2, over the improvements in inborn infants.
CONCLUSIONS. Statewide coordinated strategies in reducing nontertiary hospital births and optimizing transport of outborn infants to perinatal centers have improved considerably the outcomes of extremely premature infants. These findings have vital implications for health outcomes and resource planning.
Transferring women at risk of very preterm birth to perinatal centers for delivery decreases the neonatal mortality rate considerably.1–4 Infants born outside such centers (outborn infants) are at increased risk of developing many major neonatal complications1–5 and long-term neurodevelopmental disabilities,5–7 possibly because of suboptimal resuscitation at birth or poor access to specialist support.5,8 After birth, outborn infants also may be compromised by the speed and efficiency with which eventual transfer to an appropriate tertiary center is accomplished. In addition, the type of tertiary facility to which the infants are transferred may influence outcomes. For example, Shah et al9 demonstrated that Canadian, outborn, premature (<32 weeks) infants admitted to perinatal centers had a lower risk of death, compared with those admitted to freestanding pediatric hospitals (adjusted odds ratio [OR]: 2.25; 95% confidence interval [CI]: 1.20–4.20); staff and management policies in such facilities may not be equipped or accustomed to the needs of premature infants, because of low patient volumes.
The state of New South Wales (NSW) in Australia is geographically vast. With a total land area of 809443 km2, NSW is ∼25% larger than Texas (696241 km2). It has a population of only 6.7 million, however, with >4 million concentrated in the coastal and metropolitan areas of Sydney, and 24% of births occur in remote facilities or rural centers.10 The annual birth rate in NSW is ∼87000 births per year, with most infants (69000 births per year; 79%) being delivered in nontertiary hospitals.11
The NSW Pregnancy and Newborn Services Network (PSN) was established in 1991 to oversee the planning, coordination, and regionalization of perinatal services. There are 6 perinatal centers and 2 pediatric hospitals in Sydney and 1 perinatal center in Newcastle (150 km north of Sydney). An electronic NICU bed-state bulletin board was introduced in 1992 to facilitate transfers and NICU data collection for clinical audits. Before 1995, outborn NSW infants were transferred preferentially to 1 of 2 freestanding pediatric hospitals in Sydney. Each of the 2 pediatric hospitals and the perinatal center in Newcastle at that time maintained an individually staffed and locally coordinated transport service.
In 1995, PSN coordinated the development of 3 major changes within NSW, the aims of which were to reduce the number of outborn infants and to reduce the incidence of death and morbidities associated commonly with these high-risk deliveries. Firstly, the transport services from the 2 freestanding pediatric hospitals in Sydney were combined and centralized to form a single, independent, statewide service, the NSW Newborn and Pediatric Emergency Transport Service (NETS). This service covered the entire state of NSW. The scope and infrastructure of NETS have been described elsewhere.12
Secondly, because it was though that perinatal centers were better equipped and more experienced in the care of tiny infants, infants at <32 weeks of gestation were transported preferentially to perinatal centers. Two of the perinatal centers, as well as the pediatric hospitals, have the capacity to offer pediatric surgery. The NICU directors agreed to this concept in 1994, during the NETS amalgamation planning. Some of these destination changes, although they become evident in 1994, became more widely recognized in 1995, as part of the NETS destination advice guidelines that were disseminated to referring pediatricians.
Finally, the Perinatal Advice Line (PAL) was established. This is a statewide, fetal/maternal specialist, telephone service that provides assistance to community hospital obstetricians; it aims to encourage, to coordinate, and to optimize the clinical conditions of in utero transfers to tertiary centers. The PAL is activated whenever there are uncertainties regarding the management of high-risk pregnancies or the safety of potential transfers.
This study was designed to assess the effects of these changes on the outcomes of outborn infants in the state of NSW. The study period was divided into 2 epochs, namely, January 1992 to June 1995, before implementation of the 3 changes described above (epoch 1, or the precentralization epoch), and January 1997 to December 2002, after implementation of those changes (epoch 2) and an 18-month washout period. We hypothesized that the outcomes of outborn, high-risk, extremely premature infants, especially the mortality and morbidity rates of extremely preterm (23–28 weeks of gestation) infants, would be improved after implementation of those strategic changes.
Data Source and Patient Population
Data were obtained from the Neonatal Intensive Care Units' (NICUS) Data Collection for all infants at 23 to 28 weeks of gestation who were born between January 1, 1992, and June 30, 1995 (epoch 1), or between January 1, 1997, and December 31, 2002 (epoch 2), and who were admitted to 1 of the 9 units. Mechanical ventilation is not administered by any of the intermediate care nurseries within NSW. Data from nontertiary maternity hospitals are not included in this study.
The NSW Midwives Data Collection provides baseline population data for all births between 23 and 28 weeks of gestation in NSW hospitals. Births occurring in hospitals in the Australian Capital Territory (ACT) (∼4500 per year) are not part of the NSW Midwives Data Collection. The ACT lies within the boundary of NSW, and the NICU in the Canberra Hospital of the ACT is a member of the regional NICU network data collection. However, to provide consistency for analysis within the study hypothesis, only hospital births and outcomes from NICUs in NSW were analyzed; there were very few cross-boundary outborn admissions between ACT and NSW (10 infants in 7 years after the Canberra Hospital joined the NICUS Data Collection in 1995).
Operationally, after 1992, NICUS data were collected prospectively and collated within each NICU by a designated data manager, usually with a professional background in neonatal nursing. The data manager uses standardized published definitions13 to compile information in a central database. The accuracy of NICUS Data Collection data was validated in 1999.13
This study was restricted to infants born between 23 weeks 0 days and 28 weeks 6 days of gestation. Outborn infants were defined as those who were not born in any of the 7 perinatal centers in NSW. Infants who died before or during retrieval were not included for analysis, because such data were not collected during the first epoch.
Analyses were performed with SPSS 11.5.0 (SPSS, Chicago, IL). The χ2 test and t test were used where appropriate. Multivariate analysis was performed by means of multiple logistic regression analyses with stepwise elimination based on likelihood ratio,14,15 with entry and removal criteria of P < .05 and P > .10, respectively. We used the difference-in-differences analysis technique16 to differentiate the effects of the strategic changes on the outborn mortality rate and other outcome variables from the overall downward trend in neonatal mortality rates over time. The statistical significance of the differences was calculated by using the SEs of the differences. The level of statistical significance for all analyses was set at P < .05, using 2-tailed comparisons. The significance level was not changed when multiple comparisons were performed.17
Extremely Premature Births in NSW Hospitals and Their NICU Admissions
There was a small increase in the number of extremely premature births annually in NSW hospitals, with a total of 1778 births at 23 to 28 weeks of gestation (∼508 births per year) during epoch 1 and 3099 similar births (∼517 births per year, a 1.7% increase) during epoch 2 (Table 1). The stillbirth rate was marginally (2.1%) lower. There was a significant reduction in the number of extremely premature births occurring in nontertiary hospitals from epoch 1 to epoch 2 (30.1% and 24.6%, respectively; P < .001), which was mostly attributable to a 25% decrease in the proportion of nontertiary hospital live births (19.7% vs 14.9% of all live births; P < .001), of which a disproportionate 83% were of 23 to 25 weeks of gestation (Table 1).
The number of infants admitted to NICUs also increased (by an average yearly rate of 11.4%, from 314 infants per year to 350 infants per year) between the 2 epochs. The majority of infants were inborn, and thus the proportion of outborn NICU admissions decreased significantly (from 12.0% to 9.3%; P < .001). The changes in nontertiary hospital births and outborn admissions from 1992 to 2002 are summarized in Fig 1. The temporal relationships with the formation of the statewide network (PSN), NETS, and PAL and the introduction of the National Health and Medical Research Council guidelines on premature births,18 emphasizing in utero transfers, are also illustrated.
Mortality Rate Comparisons
The mortality rate was reduced significantly in the outborn group from epoch 1 to epoch 2 (from 39.4% to 25.1%; P = .007), and this difference was accounted for mostly by a reduction in the mortality rate of infants between 27 and 28 weeks of gestation (P < .001) (Table 1). With inborn infants showing a similar trend, there was significant improvement in the overall mortality rate between the 2 epochs (27.1% vs 23.7%; P = .035). There were 801 extremely preterm NICU survivors (45.1% of 1778 NSW births) in epoch 1 and 1602 survivors (51.7% of 3099 births) in epoch 2. This represents a 6.6% increase in the NICU survival rate (OR: 1.305; 95% CI: 1.161–1.467), equivalent to 1 extra NICU survivor per 16 NSW births (95% CI: 10–27 births), all stillbirths and live births included. The temporal changes in mortality rates and NICU survival rates (with respect to total births) from 1992 to 2002 are illustrated in Fig 2. The reduction of outborn mortality rates and the net increase in NICU survivors per total NSW births coincided with the interventions of NETS and PAL. The early shifting of nontertiary hospital live births (predominantly of 23–25 weeks) to tertiary hospital births during epoch 1 did not correlate with an increasing number of NICU survivors within epoch 1.
Perinatal Characteristics of NICU Infants and Outcome Comparisons
Maternal and delivery characteristics for all patients are shown in Table 2. Significantly fewer outborn infants were admitted to freestanding pediatric hospitals in epoch 2 (34.8% and 2.1% in epoch 1 and epoch 2, respectively; P < .001). The majority of extremely premature outborn infants with congenital malformations (total of 21 infants during both epochs) were also admitted to perinatal centers, instead of pediatric hospitals. Three were admitted to freestanding pediatric hospitals and none died. Four of the 18 infants admitted to perinatal centers died (2 infants of 25 weeks, of 550 g and 815 g, 1 infant of 26 weeks and 810 g, and 1 infant of 27 weeks and 950 g).
Table 3 compares morbidity data between the infants admitted in the 2 epochs. In general, the outcomes for outborn infants improved between the 2 epochs, but these infants continued to have significantly higher morbidity rates, compared with inborn infants, especially with regard to severe (grade 3 or 4) IVH (30.6% of outborn infants vs 9.3% of inborn infants in epoch 1; P < .001; 19.4% vs 10.0% in epoch 2; P = .002) and radiologically or surgically proven necrotizing enterocolitis (17.4% vs 4.5% in epoch 1; P < .001; 7.2% vs 1.7% in epoch 2; P < .001).
Multivariate logistic regression was performed to determine factors predictive of death (Table 4). This model showed a significant reduction in the mortality rate of infants of 27 to 28 weeks of gestation in epoch 2 (adjusted OR: 0.171; 95% CI: 0.067–0.434) and a higher risk of death for infants admitted to children's hospitals (adjusted OR: 3.006; 95% CI: 1.277–7.074). Prenatal steroid therapy was not a significant predictor in this model.
The basic concept of the difference-in-differences approach is to study the impact of strategic changes (intervention) by comparing the performance of the outborn group (treatment group), relative to the inborn group (control group), before and after intervention. The inborn mortality rate decreased by 1.9% (SE: 1.7%) from before intervention to after intervention, whereas the outborn mortality rate decreased by 14.3% (SE: 5.3%) for the same period (Table 1). This gives a difference-in-differences of 12.4% (SE: 3.7%; P < .001), which is equivalent to a relative improvement of 31.4% in the outborn mortality rate. Stratification of this difference-in-differences according to gestational age showed significant relative improvement only for infants between 27 and 28 weeks of gestation (relative improvement of 62.5%; P < .001).
Difference-in-differences analyses showed significant improvement in most neonatal morbidity rates for outborn infants during epoch 2, above that of inborn infants (Table 3). However, there were more chronic lung diseases in epoch 2, although less postnatal steroid treatment and home oxygen therapy.
Previous studies showed unequivocally that the prognosis for extremely premature infants is considerably better when their birth and immediate care are conducted at tertiary perinatal centers.1,8,19–21 However, in geographically vast areas such as the state of NSW in Australia, high-risk births inevitably occur in nontertiary institutions, despite committed attempts to transfer at-risk women to perinatal centers before delivery. Therefore, the dramatic improvements in the mortality and morbidity rates of outborn infants, as demonstrated by our study, were most likely attributable to multiple factors, such as widespread changes in referral coordination, perinatal advice, and neonatal transport services, and resulted in almost-simultaneous reductions in the numbers of stillbirths and nontertiary hospital births and improvements in the conditions of both inborn and outborn infants who ultimately received intensive care.
Our findings are particularly important because the rate of extremely premature births, both locally11,22 and nationally,23 will most likely continue to increase in the near future, because of widespread increases in maternal age and in the use of assisted conception. This study showed a striking 25% reduction in the number of nontertiary hospital live births, a concurrent and significant reduction in the outborn mortality rate (from 39.4% to 25.1%), and a substantial decrease in the overall NICU mortality rate (from 27.1% to 23.7%). This result, when taken together with the 2.0% reduction in the number of stillbirths, equates to an increase in the number of NICU survivors per total births of 6.6% in epoch 2, or an equivalent of 1 extra survivor for every 16 extremely premature births (all stillbirths and live births combined).
The outborn infants who were admitted eventually to NICUs might have been a biased and favored group, because they might have been more robust than infants who were not selected by local practitioners for transfer.24 In the current study, only approximately one half of nontertiary hospital live births were offered NICU care in both epochs, compared with almost all of the live inborn infants. Imminent labor or a pessimistic regional obstetric attitude, as shown in our national survey,25 might have delayed or prevented in utero or ex utero transfers, and nontransferred infants would most certainly have died, because NSW has no facilities to care for extremely preterm infants outside the 9 perinatal urban units. Unfortunately, data on nontertiary neonatal deaths are not available; therefore, this confounding factor cannot be assessed currently.
However, these issues do not detract from the fact that outborn mortality rates, particularly for infants of 27 to 28 weeks, improved dramatically during the 2 study epochs. The formation of a centralized statewide neonatal transport service (NETS), with its multiple, rapid-response, transport teams, most likely played a major role. McNamara et al26 showed that the presence of a skilled transport team at high-risk deliveries improved the quality of neonatal resuscitation significantly. This finding was confirmed by Hood et al,27 who showed that outborn infants who were not transferred to a NICU by a neonatal transport team had a 60% greater mortality rate. Of note, more outborn mothers received prenatal steroid therapy in epoch 2. This practice might also have been a surrogate marker of improved peripheral hospital access to expert advice through the PAL network.
Furthermore, the transport policy that delegates preference for the transfer of premature infants to perinatal centers likely would have improved outcomes, because perinatal centers are considerably more accustomed to managing prematurity-associated problems (because of greater patient load and appropriately streamlined clinical practices) than are freestanding pediatric hospitals. This was demonstrated in Canada by Shah et al,9 who found that outborn premature infants who were admitted to perinatal centers had lower risks of death and morbidities.
It is indisputable that outcome improvements were likely to have been influenced considerably by concomitant advances in neonatal preterm care. It is interesting to note, however, that the reduction in outborn mortality rates was extremely rapid and was evident even during the 18-month washout period (as shown in Fig 2); the speed with which these changes came about is likely attributable to local strategic implementation of the neonatal retrieval service and perinatal advice. The reduction in the number of nontertiary hospital births during epoch 1 did not translate effectively into instantaneously increased numbers of NICU survivors. This lack of apparent effectiveness in epoch 1 could be attributable to a very high inherent mortality rate in this gestational age range and also to less well-coordinated perinatal advice and retrieval. Indeed, in the difference-in-differences analyses, morbidities such as intraventricular hemorrhage and necrotizing enterocolitis showed improvement that exceeded that for inborn infants. This is vitally important, because these morbidities have all been implicated in the development of adverse neurodevelopmental outcomes. The higher rate of chronic lung disease is likely related to the increased survival rate.
Whether our findings can compare directly with those for other countries is difficult to ascertain, because NSW has a much lower outborn rate than many other countries, such as the United States, where the Vermont Oxford Network data showed significantly higher (albeit decreasing) outborn rates for very low birth weight admissions from 1991 (20.5%) to 1999 (16.1%).28 Despite this, the benefits of regionalizing perinatal services for premature infants have been demonstrated repeatedly,27 and the NSW strategies described in this article might have worldwide implications in developed and developing countries. For example, it would be interesting to see whether our results could be duplicated in the Canadian network with destination preference changes.
Unfortunately, there were unavoidable weaknesses and gaps in our data. For example, we did not have detailed information on the quality of perinatal care, resuscitation or treatment at delivery, or stabilization procedures before and during transport at community hospitals. The Midwives Data Collection, which is usually completed at the time of discharge of the mother from hospital, could not provide reliable data on ultimate nontertiary hospital mortality rates. Information on whether the infant died before retrieval was not available for the first epoch, to allow for comparisons. These and other unmeasured factors might have confounded our findings.
Our findings demonstrated a significant improvement in the overall outcomes of extremely premature births after the development of a centralized neonatal transport system, almost-universal transfer of these high-risk infants to perinatal centers, and ready accessibility of tertiary care obstetric advice. These results have considerable implications for regionalization planning and resource utilization for perinatal services.
We thank the directors, the NICUS members, and the audit officers of all tertiary units for supporting this collaborative study, as follows: NSW PNS, Dr David Henderson-Smart (director); NICUS, Barbara Bajuk (coordinator), Trina Vincent; NETS, Dr Andrew Berry (state medical director), the Canberra Hospital, Dr Graham Reynolds (director), Dr Alison Kent, John Edwards; John Hunter Hospital, Dr Chris Wake (director), Lynne Cruden; Royal Prince Alfred Hospital, Dr Nick Evans (director), Dr Phil Beeby, Shelley Reid; Liverpool Hospital, Dr Robert Guaran (director), Dr Ian Callander, Dr Jacqueline Stack, Kathryn Medlin, Sara Wilson; Nepean Hospital, Dr Lynne Downe (director), Mee Fong Chin; the Children's Hospital at Westmead, Dr Robert Halliday (director), Dr Nadia Badawi, Karen Walker; Royal North Shore Hospital, Dr Tushar Bhuta (director), Dr Martin Kluckow, Vicki Gallimore; Sydney Children's Hospital, Dr Barry Duffy (director), Dr Gary Williams, Janelle Young, Denise Georgakopoulos; Westmead Hospital, Dr William Tarnow-Mordi, Dr Marilyn Rochefort (director), Jane Baird; Royal Hospital for Women, Dr Kei Lui (director), Dr Paul Garvey, Dianne Cameron. We thank Tara Stevermuer, MAppStat, Centre for Health Service Development, University of Wollongong (Wollongong, Australia) for critical review of the statistic models used. We also thank the infants and their families and the nursery, obstetric, and medical records staff members of the obstetric and children's hospitals in NSW and the ACT.
- Accepted July 27, 2006.
- Address correspondence to Kei Lui, MBBS, FRACP, MD, Department of Newborn Care, Royal Hospital for Women, Barker St, Randwick, NSW 2031, Australia. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Simpson JM, Evans N, Gibbard RW, Heuchan AM, Henderson-Smart DJ. Analysing differences in clinical outcomes between hospitals. Qual Saf Health Care.2003;12 :257– 262
- Truffert P, Goujard J, Dehan M, Vodovar M, Breart G. Outborn status with a medical neonatal transport service and survival without disability at two years: a population-based cohort survey of newborns of less than 33 weeks of gestation. Eur J Obstet Gynecol Reprod Biol.1998;79 :13– 18
- ↵Saigal S, Rosenbaum P, Hattersley B, Milner R. Decreased disability rate among 3-year-old survivors weighing 501 to 1000 grams at birth and born to residents of a geographically defined region from 1981 to 1984 compared with 1977 to 1980. J Pediatr.1989;114 :839– 846
- ↵Rashid A, Bhuta T, Berry A. A regionalised transport service: the way ahead? Arch Dis Child.1999;80 :488– 492
- ↵Bajuk B. Validation of the Neonatal Intensive Care Units' Data Collection. In: Proceedings of the 5th Annual Congress of the Perinatal Society of Australia and New Zealand 2001. Canberra, Australia: Perinatal Society of Australia and New Zealand; 2002:55
- ↵Kleinbaum DG. Logistic Regression: A Self-Learning Text. New York, NY: Springer-Verlag; 1994
- ↵Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: Wiley; 1989
- ↵Perneger TV. What's wrong with Bonferroni adjustments. BMJ.1998;316 :1236– 1238
- ↵National Health and Medical Research Council. Clinical Practice Guidelines: Care Around Preterm Birth. Canberra, Australia: National Health and Medical Research Council; 1997
- ↵Laws P, Sullivan E. Australia's Mothers and Babies 2002. Sydney, Australia: National Perinatal Statistics Unit, Australian Institute of Health and Welfare; 2004. Available at: www.npsu.unsw.edu.au/AMB%202002.pdf . Accessed March 20, 2006
- ↵Abdel-Latif ME, Bajuk B, Oei J, Vincent T, Sutton L, Lui K. Does rural or urban residence make a difference to neonatal outcome in premature birth? A regional study in Australia. Arch Dis Child Fetal Neonatal Ed.2006;91 :F251—F256
- ↵Horbar JD, Badger GJ, Carpenter JH, et al. Trends in mortality and morbidity for very low birth weight infants 1991–1999. Pediatrics.2002;110 :143– 151
- Copyright © 2006 by the American Academy of Pediatrics