Survival rates for extremely preterm (EPT; < 28 weeks) or extremely low-birthweight (ELBW; < 1000 g) babies have improved dramatically since the introduction of modern neonatal intensive care approximately four decades ago. In the state of Victoria, Australia, survival rates for ELBW infants improved from approximately one in four in the late 1970s to three in four in the late 1990s.1 There have been many reasons for the improving survival rate, including the increased use of antenatal corticosteroid therapy2 and surfactant therapy after birth.3 However, a major reason has been the increased willingness to offer intensive perinatal and neonatal care to EPT/ELBW babies.4 Despite the rapidly improving survival rates, there has been no accompanying improvement in longterm neurological impairments or disabilities, which remain much more common in EPT/ELBW infants than in term-born or normal-birthweight controls.5,6 There are, however, some recent advances that are known to improve long-term neurological outcome for preterm infants, including magnesium sulphate given to women likely to deliver very preterm, caffeine therapy in infants < 1251 g birthweight, docosahexaenoic acid (DHA) supplementation to milk feeds in babies < 33 weeks' gestation and early developmental interventions for preterm infants after discharge home. These will be discussed in turn.
Antenatal magnesium sulphate therapy
Theoretically, magnesium sulphate might be neuroprotective because of effects on cellular metabolism, cell death or injury or blood flow to the brain.7 Some,8,9 but not all,10–13 observational studies in humans also suggest a neuroprotective effect for magnesium sulphate. The experimental and observational human data triggered several randomized controlled trials (RCTs) of antenatal magnesium sulphate therapy designed specifically to be neuroprotective for the fetus. Data from these RCTs have been systematically reviewed and analysed, first in The Cochrane Library,14 and subsequently elsewhere.15
The first RCT was a two-arm study by Mittendorf et al.16 in which a total of 149 women in active labour at 25–33 weeks' gestation were enrolled from October 1995 to January 1997 at a single US centre. Those with active labour and cervical dilatation ≤ 4 cm were considered candidates for tocolysis with magnesium sulphate (the ‘tocolytic’ arm); they were randomly allocated to receive magnesium sulphate as a 4-g bolus followed by 2–3 g/hour maintenance (n = 46 women, 55 babies), or an alternative tocolytic (nonblinded) (n = 46 women, 51 babies). The remainder (with cervical dilatation > 4 cm) were considered the ‘neuroprotective’ arm of the study and were randomly allocated to either a 4-g magnesium sulphate bolus (n = 29 women, 30 babies) or saline placebo (blinded) (n = 28 women, 29 babies). The study was stopped prematurely because of concerns about higher paediatric mortality in the magnesium group.17 Children were assessed for cerebral palsy at 18 months corrected age.16
In the Australasian Collaborative Trial of Magnesium Sulphate (ACTOMgSO4) study by Crowther et al.,18 a total of 1062 women who were less than 30 weeks pregnant and in whom birth was anticipated within 24 hours were enrolled from February 1996 to September 2000 in Australia and New Zealand. Women were randomly allocated to either intravenous magnesium sulphate (n = 535 women, 629 live babies) or an identical volume of saline placebo (n = 527 women, 626 live babies). The magnesium sulphate dose was 4 g over 20 minutes, followed by 1 g/hour for up to 24 hours or until birth, whichever came first. There were no repeat courses of treatment. The major end point was survival free of cerebral palsy at 2 years of age corrected for prematurity. Substantial gross motor dysfunction was defined as level II or higher on the Gross Motor Function Classification System (GMFCS).19
In the PREMAG trial by Marret et al.,20 a total of 573 women whose birth was planned or expected within 24 hours were enrolled at 18 collaborating centres in France, but data from only 13 centres (564 women) were included in the final report. Women were randomly allocated to either intravenous magnesium sulphate 4 g or an equal volume of saline placebo over 30 minutes. The major end point of the study was white matter injury to the infant diagnosed by cranial ultrasonography. However, follow-up data for surviving children to 2 years of age were also reported.21,22 Substantial gross motor dysfunction in this study was defined as moderate or severe impairment.
In the most recent multicentre trial from the USA, by Rouse et al.,23 known as the BEAM (Beneficial Effects of Antenatal Magnesium Sulphate) study, 2241 women 24–31 weeks pregnant with singletons or twins and at high risk of delivery were randomized to receive a 6-g loading dose and 2 g/hour maintenance infusion of either magnesium sulphate or identical placebo. The majority of women (87%) were eligible because of premature rupture of the membranes, rather than because of spontaneous premature labour, as magnesium sulphate within the previous 12 hours was an exclusion criterion and magnesium sulphate is a common tocolytic in the USA. The major end point was survival free of moderate or severe cerebral palsy at 2 years of age. Cerebral palsy was diagnosed by certified paediatricians or paediatric neurologists, and severity was based on the Gross Motor Function Classification System (GMFCS);19 substantial gross motor dysfunction was defined GMFCS level II or higher.
The major outcomes from a meta-analysis of the data14,15 from these four RCTs are as follows. Antenatal magnesium sulphate treatment had no overall significant effect on total paediatric mortality [risk ratio (RR) 0.95; 95% confidence interval (CI) 0.80 to 1.12; P = 0.53; four trials; 4446 infants]. There was a significant reduction in the rate of cerebral palsy from antenatal magnesium sulphate treatment (risk difference −0.019; 95% CI −0.032 to −0.006; P = 0.006; four trials; 4446 infants). Overall, the number needed to treat to prevent one case of cerebral palsy was 63 (95% CI 43 to 155), assuming a baseline risk of 5% for cerebral palsy in the control group. There was also a significant reduction in cerebral palsy in the BEAM study23 (RR 0.59; 95% CI 0.40 to 0.85). There was a significant reduction in the combined outcome of paediatric mortality or cerebral palsy (risk difference −0.024; 95% CI −0.046 to −0.003; P = 0.027; four trials; 4446 infants). The number needed to treat to achieve an extra survivor free of cerebral palsy was 42 (95% CI 24 to 346), assuming a baseline risk of 17% for death or cerebral palsy in the control group.
Substantial gross motor dysfunction was significantly less common with magnesium, both overall (RR 0.60; 95% CI 0.43 to 0.83; P = 0.002) and in two individual studies (ACTOMgSO418: RR 0.53; 95% CI 0.30 to 0.92; BEAM23: RR 0.56; 95% CI 0.33 to 0.95). There was an overall marginal reduction in the rate of death or substantial gross motor dysfunction with magnesium (RR 0.84; 95% CI 0.71 to 1.00; P = 0.05) and in the ACTOMgSO4 study18 (RR 0.74; 95% CI 0.59 to 0.93).
There is little doubt that antenatal magnesium sulphate therapy given to women at risk of preterm birth is a neuroprotective against motor disorders in the preterm fetus. It reduces the rates of cerebral palsy and substantial gross motor dysfunction in early childhood, and also the rates of the combined outcomes of death or cerebral palsy and death or substantial gross motor dysfunction. Importantly, there are no obvious harmful paediatric effects; in particular the studies found little evidence of respiratory depression caused by magnesium sulphate – rates of low Apgar scores at five minutes and of ongoing respiratory support were similar in the treatment and placebo groups.14
There are well-described maternal side-effects, such as hypotension and tachycardia, which occur frequency with antenatal magnesium sulphate.14 However, these persist for only a short time after ceasing treatment. More serious maternal effects, such as death or cardiac arrest, are rare at the doses used in the RCTs described in the Cochrane review.14
Caffeine therapy for apnoea of prematurity
Methylxanthines as a group have been used to treat or prevent apnoea of prematurity for more than three decades, but more recently caffeine has become the dominant drug used for this purpose, because it needs to be given only once per day and it has fewer side-effects than other methylxanthines. However, there have been concerns that caffeine and other methylxanthines might be harmful because they inhibit adenosine receptors, and adenosine preserves brain ATP levels and protects brain cells during experimental hypoxia and ischaemia.24,25 To evaluate the long-term risks and benefits of caffeine, we conducted a large, multicentre, international, placebo-controlled RCT of caffeine in infants of birthweight < 1251 g starting in the first 10 days of life. The study was funded by the Canadian Institutes of Health Research and the National Health and Medical Research Council of Australia. A total of 2006 infants were enrolled between October 1999 and October 2004 and randomly assigned to receive caffeine citrate or normal saline placebo. A loading dose of 20 mg of caffeine citrate per kilogram body weight was followed by a daily maintenance dose of 5–10 mg per kilogram. The median duration of therapy was just over 5 weeks.
During the primary hospitalization, caffeine therapy reduced the rates of symptomatic patent ductus arteriosus requiring either medication [caffeine 29.3% vs. placebo 38.1%; adjusted odds ratio (AOR) 0.67; 95% CI 0.55 to 0.81; P < 0.001) or surgery (caffeine 4.5% vs. placebo 12.6%; AOR 0.32; 95% CI 0.22 to 0.45; P < 0.001), bronchopulmonary dysplasia (defined as requiring oxygen therapy at 36 weeks; caffeine 36.3% vs. placebo 46.9%; AOR 0.63; 95% CI 0.52 to 0.76; P < 0.001) and the need for post-natal corticosteroid therapy (caffeine 14.4% vs. placebo 20.2%; P < 0.001). Caffeine also shortened the durations of assisted ventilation via an endotracheal tube and the use of any positive airway pressure or of oxygen therapy by approximately one week.26,27
At follow-up at 18–21 months of age, 377 out of 937 (40.2%) infants assigned to caffeine had died or survived with a neurodevelopmental disability (any of blindness, deafness, cerebral palsy or developmental delay), compared with 431 of the 932 infants (46.2%) assigned to placebo (AOR 0.77; 95% CI 0.64 to 0.93; P = 0.008).27 Caffeine significantly reduced the incidence of cerebral palsy (4.4% vs. 7.3%; AOR 0.58; 95% CI 0.39 to 0.87; P = 0.009) and of cognitive delay (33.8% vs. 38.3%; AOR 0.81; 95% CI 0.66 to 0.99; P = 0.04). At a subsequent follow-up at 5 years of age, the neurological benefits of caffeine had diminished and the reduction in the rate of adverse neurological outcomes was no longer statistically significant (caffeine 21.1% vs. placebo 24.8%; AOR 0.82; 95% CI 0.65 to 1.03; P = 0.09).28
Exactly how caffeine might improve neurological outcome is not clear. In a side study at the main centre involved in the caffeine RCT in Australia, we were able to undertake magnetic resonance imaging at term in 70 infants who were enrolled in the main study. We found little difference in brain volumes that might explain the cognitive benefit of caffeine, but we were able to detect microstructural changes in the white matter of the cerebral cortex that were consistent with advanced maturation of the brain related to caffeine therapy, suggesting that caffeine has a direct effect on the brain, rather than just simply reducing other morbidities in the newborn period as a mechanism for improving neurological outcomes.29
Docosahexaenoic acid supplementation of milk feeds
One of the reasons why very preterm infants have higher rates of adverse cognitive outcomes than term-born infants might be that preterm infants lack an omega-3 fatty acid, called docosahexaenoic acid (DHA), which has an essential role in brain development and function. Infants born preterm miss out on the intrauterine supply of DHA at a time when the most rapid accumulation of DHA occurs and when brain growth is at its greatest.
A multicentre RCT in preterm infants born before 33 weeks' gestation was designed in Australia to assess the effect of supplementing feeds with 1% of dietary fats as DHA, a level equivalent to the amount that babies would receive if they had remained in the uterus, compared with usually dietary amounts of DHA, equivalent to 0.2–0.3% of dietary fats as DHA, with the major outcome being developmental outcome at 18 months corrected age.30 Stratification was by centre, birthweight (< 1250 g and ≥ 1250 g) and infant sex. The intervention started soon after gastrointestinal feeds were commenced, and continued until the child was 40 weeks of post-menstrual age.
Between 2001 and 2005, 657 infants were enrolled in the trial; 93.5% (n = 614) completed the 18-month follow-up. Although the overall increase in the Mental Development Index on the Bayley Scales of Infant Development, Second Edition, with DHA supplementation did not reach statistical significance (mean difference 1.9, 95% CI −1.0 to 4.7), fewer infants in the high-DHA group than in the control group exhibited mildly delayed (18.5% vs. 25.9%, respectively; P = 0.05) or significantly delayed (4.7% vs. 10.1%, respectively; P = 0.02) mental development. Among the < 1250 g birthweight subgroup, Mental Developmental Index scores were higher in infants fed a high-DHA diet than among those in the control group (mean difference 5.1; 95% CI 0.5 to 9.6; P = 0.03) and fewer infants had mildly delayed mental development (16% vs. 32.6%; P = 0.01). Girls fed a high-DHA diet had higher Mental Developmental Index scores than girls fed the standard-DHA diet (mean difference 4.5; 95% CI 0.4 to 8.7; P = 0.03), and this translated to fewer girls with mildly delayed (9.9% vs. 25%; P = 0.01) or significantly delayed (1.4% vs. 9.7%; P = 0.01) mental development. The Mental Development Index scores of boys did not differ between groups. The children in the high-DHA group showed no obvious increase in adverse health outcomes, suggesting that the new treatment is safe in the short term.
Early developmental interventions after discharge home
Early developmental interventions have been proposed to help reduce the rate of adverse long-term outcomes in very preterm or tiny infants. A number of trials of early intervention have been reported; in some the intervention has been started while the baby is still in the nursery, and in others intervention commenced after discharge home, but the effectiveness of early developmental intervention programmes for preterm infants post hospital discharge is unclear when individual trials of early intervention are considered. One way to resolve the issue is to consider all trials together in a systematic review and meta-analysis of data.
In a recent systematic review of RCTs and quasi-RCTs of early developmental intervention programmes post discharge from hospital for preterm infants, 16 trials, enrolling 2379 infants, were identified, six of which were RCTs.31 There was considerable variability in the depth and breadth of the interventions; some were focused on infant development, some were focused on the parent–infant relationship and some on both. After meta-analysis of the data from the reported studies, the intervention improved cognitive outcomes at infant age [1–3 years; eight trials; 1444 infants; developmental quotient (DQ); standardized mean difference (SMD) 0.46 SD; 95% CI 0.36 to 0.57; P < 0.0001) and at preschool age [3–4 years; three trials; 1006 infants; intelligence quotient (IQ); SMD 0.46 SD; 95% CI 0.33 to 0.59; P < 0.0001]. However, this effect was not sustained at school age (5–17 years; three trials; 1111 infants; IQ; SMD 0.02 SD; 95% CI −0.10 to 0.14; P = 0.71). There was substantial heterogeneity between studies in cognitive outcomes at infant and school ages. When only studies with the highest methodological quality were considered, the heterogeneity was reduced and the benefits of early intervention on cognitive outcomes in infancy and at preschool age were maintained. The size of the effect, an almost 0.5 SD improvement in cognitive scores, was clearly clinically important, suggesting that these were true treatment effects of early intervention on cognitive outcomes. That the benefits seem to be limited to children prior to school age could be explained by the fact that other influences, particularly the effects of the family and environment, are more likely than events in the first year of life to be influencing important later cognitive development. There was little evidence of an effect of early intervention on motor outcomes, including on rates of cerebral palsy, or on the results of standardized motor assessments at any age, but only two studies reported outcomes beyond 2 years of age.
As the majority of studies included in the systematic review involved infants born in the 1980s and the infants tended to be relatively mature at birth in some studies, we decided to complete our own RCT of early intervention after discharge home in infants born at < 30 weeks' gestational age.32 In our trial, the intervention comprised nine home visits by a psychologist and a physiotherapist over the first year after discharge home, focusing not only on infant development and the parent–infant relationship, but also on parental mental health. A total of 61 infants were randomly allocated to intervention, and 59 to standard follow-up care, between 2005 and 2007. At 2 years of age the intervention improved cognitive outcome by only 0.27 SD, which was not statistically significant (95% CI −0.10 to 0.63; P = 0.20) compared with the mean treatment benefit of 0.46 SD in the systematic review. On the other hand, infant behaviour was substantially improved by the intervention, as was parental mental health, with fewer symptoms of anxiety or depression in the main caregiver.32 The longer-term effects of our intervention programme on both the child and family need to be determined.
What has been happening to long-term outcomes for extremely preterm babies on a regional basis?
Because survival rates of extremely preterm infants cannot increase indefinitely, the emphasis of perinatal and neonatal intensive care must be on reducing the rates of neurosensory disability rates in survivors, which have remained too high relative to term-born control infants over a long period of time.5 In the state of Victoria, Australia, we have been able to assess long-term outcomes in all infants born at < 28 weeks' gestational age in three discrete cohorts, comprising the calendar years 1991–92,33 199734 and 2005.5 There were no major changes in perinatal or neonatal intensive care over this time, but caffeine became the methylxanthine of choice by 2005, replacing theophylline, which had been used through the 1990s. The other therapeutic change was the use of post-natal corticosteroid therapy to prevent or treat bronchopulmonary dysplasia.
Rates of post-natal corticosteroid therapy and the doses used were high in the 1990s, but both were lower by 2005. The change in practice was triggered by reports of adverse long-term outcomes in children who had been exposed to post-natal corticosteroids, and was augmented by authoritative statements which appeared in the early 2000s declaring that post-natal corticosteroids should be severely limited outside RCTs.35 Despite more than 40 RCTS of post-natal corticosteroid therapy, the role of corticosteroids in clinical practice was uncertain. After systematically reviewing the data available up to the early 2000s, we were able to demonstrate that, if the risk of a ventilator-dependent infant developing bronchopulmonary dysplasia was very high (> 65%), then the effect of post-natal corticosteroids would be most likely to improve the chance of the baby surviving free of cerebral palsy.36 Although we were thwarted in our attempts to clarify the issue further in a RCT of low-dose dexamethasone, the most commonly used corticosteroid in ventilatordependent EPT/ELBW infants, we were able to show that such a low dose, approximately one-tenth of that used in the most common regimen throughout the trials of the 1990s, was still able to lead to successful extubation of very high-risk infants, without any obvious short- or long-term effects.37,38 By 2005, the starting dose of corticosteroids most widely used in clinical practice in Victoria was the dose used in our RCT, and the total dose received by those few infants who were ultimately treated was approximately one-quarter of that used in the 1990s.
With widespread treatment with caffeine and the reduction in post-natal corticosteroids in 2005 compared with the 1990s, it was anticipated that the long-term outcome for infants born at < 28 weeks' gestation who survived in Victoria would be improved. At 2 years of age, 163 of 172 (95%) of consecutive survivors born at < 28 weeks were assessed at 2 years corrected age.5 Severe developmental delay (a developmental quotient less than −3 SD compared with the mean and SD of term-born control infants) was substantially reduced, to 3.7%, compared with 14.8% in survivors from 1997 (relative risk 0.25; 95% CI 0.10 to 0.60; P < 0.001). Severe neurosensory disability was also substantially reduced, to 3.7%, compared with 15.4% in survivors from 1997 (relative risk 0.24; 95% CI 0.10 to 0.57; P < 0.001). There were no substantial changes over time in rates of cerebral palsy, blindness or deafness.
As survival rates for EPT infants reach a ceiling, we must focus on improving long-term quality of life for survivors and, in particular, reduce the excessive rates of neurosensory disabilities that occur in preterm survivors compared with term-born children. There are signs that this is happening in the state of Victoria, with reductions in the rate of severe developmental delay and neurosensory disability in survivors born in 2005 compared with those born in the 1990s. This has occurred in part because of the introduction of caffeine therapy, which is known to improve long-term cognitive and motor outcomes in preterm infants. Also, a more rational use of post-natal corticosteroids in 2005 has possibly contributed. Several other promising therapies that are now known to improve long-term outcomes include antenatal magnesium sulphate, supplementation of feeds with DHA and early intervention after discharge home. As these therapies are introduced into clinical practice, their uptake and effects on long-term outcomes must be monitored in future geographical cohorts of EPT infants.