What are the main research findings during the last 5 years that have changed my approach to clinical practice?
- Correspondence to Professor Christian F Poets, Department of Neonatology, Tuebingen University Hospital, Tuebingen D-72076, Germany;
Contributors CFP was involved in collecting the manuscripts reviewed in this article and in writing it.
- Accepted 20 July 2011
- Published Online First 24 August 2011
When asked to address the above question, findings that appeared to be among the most relevant included (1) interventions in the delivery room directed at supporting the physiological transition from intrauterine to extrauterine life rather than actively intervening in it; (2) recent data suggesting that keeping extremely low-gestational age neonates at a pulse oximeter saturation (SpO2) of 91–95% would increase their chances of survival compared with aiming for lower SpO2 values; (3) using caffeine citrate in infants <1250 g with apnoea of prematurity improves neurodevelopmental outcome; (4) injecting antivascular epithelial growth factor into the vitreous seems to be an effective treatment for retinopathy of prematurity and (5) moderate hypothermia for perinatal hypoxic-ischaemic encephalopathy increases the likelihood of survival without neurological impairment. Here, data that support these recent changes in approach will be presented and discussed.
Supporting the postnatal adaptation
For decades, the prevailing medical view was that ensuring a good transition from intrauterine to extrauterine life requires a highly active approach, as evident from the term ‘resuscitation’ used to summarise the interventions applied during this process. It was the important work of Saugstad et al1 and Vento et al2 that reminded us that less may be more, that is, that the neonate will stabilise faster if kept in room air instead of 100% oxygen immediately after birth and, more importantly, that neonatal mortality may be reduced by using this approach.3
Now it is one thing to accept this for a term neonate, but what about preterms? Here, the data are less clear, although these infants are even more vulnerable to oxidative stress, and might thus benefit more from restricted oxygen use. Four studies reported on using oxygen blenders during the transition from intrauterine to extrauterine life. The first two were randomised controlled trials (RCTs), one randomising 42 infants ≤28 weeks requiring ‘active resuscitation’ to an initial fraction of inspired oxygen (FiO2) of 0.3 or 0.9, which was increased every 60–90 s by 0.1, if heart rate was <100/min, or decreased by 0.1, if oximeter saturation (SpO2) was >85%. No differences were found between groups in relation to the type of ventilation administered on admission to neonatal intensive care unit, or in time to attain clinical stabilisation; mean ‘final’ FiO2 was 0.45 in both groups.4 The second RCT randomised 41 infants ≤32 weeks ‘requiring resuscitation’ to an initial FiO2 of 1.0 or 0.21, again with predefined changes being allowed, depending on SpO2 and heart rate. Every infant in the lower O2 group had received an increase in FiO2 by 3 min of birth, but mean FiO2 was 0.6 versus 1.0 at 3 min after birth, and SpO2 was significantly lower throughout the first 10 min, while heart rate did not differ between groups.5
The other two studies performed a longitudinal comparison of SpO2 and clinical data before and after a change in unit policy from 100% to blended oxygenaccording to the newborn's SpO2. Dawson et al compared 20 infants receiving 100% O2 with 106 initially receiving room air. Of the latter infants, 92% received some supplemental O2 by 5 min after birth, with SpO2 having risen to a mean of 81% by 6 min. In the first 10 min, 80% versus 55% of infants had an SpO2≥95%.6 The last study compared changes in partial arterial O2 pressure (PaO2) before and after a change in unit guidelines from 100% O2 to an FiO2 titration aiming at an SpO2 of 85–95%. This change in policy resulted in more infants maintaining their PaO2 below 80 mm Hg (44% vs 70%) without any adverse effects.7 Thus, despite limitations pertaining to study methodology, it seems from these data that the use of pulse oximetry and oxygen blenders immediately after birth will permit stabilisation of preterm infants at least as well as administration of 100% O2, while better avoiding hyperoxaemia. Although such a policy has been suggested already in neonatal guidelines,8 it should be noted that clinically relevant outcomes, using a rigorous study design, have still to be investigated for the approach in this population.
Along the same lines, there is also increasing evidence to suggest that stabilising the preterm infant using continuous positive airway pressure (CPAP) may be a better option than intubation immediately after birth. The classic experiments by Björklund et al9 in newborn rabbits, showing that five large-volume breaths administered to a surfactant-deficient lung are sufficient to induce long-lasting inflammatory changes, have led to several RCTs comparing stabilisation via CPAP administration versus routine intubation and mechanical ventilation after birth.
The first of these trials randomised 610 infants of 25–28-week gestational age (GA) to CPAP or intubation and ventilation at 5 min after birth; primary outcome was death or persistent oxygen requirement at 36-week GA.10 Here, 33.9% of infants randomised to CPAP died or developed bronchopulmonary dysplasia (BPD), compared with 38.9% of those being intubated immediately after birth. In half of the infants (54%) in the CPAP group, intubation and mechanical ventilation could be avoided throughout the first 5 days, so that surfactant use was also halved. The incidence of pneumothorax, however, was 9% in the CPAP group, compared with 3% in the intubation group (p<0.001), while the median duration of intubation and mechanical ventilation was 3 days (IQR 0–11) in the CPAP group versus 4 days (1–14) in the control group (p<0.001).
A similar question was asked in the SUPPORT trial, where 1316 infants of 24–27-week GA were randomly assigned to CPAP or intubation and surfactant treatment within 1 h of birth.11 Again, the primary outcome was death or oxygen requirement at 36-week GA, and this did not differ significantly between groups, being 47.8% in the CPAP and 51.0% in the surfactant group (RR 0.95, 95% CI 0.85 to 1.05). With regard to secondary outcomes, infants on CPAP were ventilated for a median of 10 (IQR 2–32) versus 13 (2–36) days (p<0.03), were less likely to require postnatal corticosteroid therapy (7.2% vs 13.2%, RR 0.57, 95% CI 0.41 to 0.78, p<0.001), and had no increased rate of any other adverse outcomes (eg, severe necrotising enterocolitis (NEC), retinopathy of prematurity (ROP) or intraventricular haemorrhage (IVH)). In contrast to the Continuous Positive Airway Pressure or Intubation at Birth (COIN)trial,10 there was no significant difference in the number of infants developing pneumothorax or any air leak in the first 14 days (6.8% vs 7.4%), which may be related to the fact that infants randomised to CPAP in the COIN trial received a Positive End-Expiratory Pressure (PEEP)of 8 cm H2O and were intubated at an FiO2>0.60, whereas PEEP was 5 cm H2O in the SUPPORT trial and infants were intubated at an FiO2>0.50.
A third study randomised 208 infants of 25–28-week GA, who had not been intubated immediately after birth, to either CPAP or intubation followed by prophylactic surfactant within 30 min of birth. Primary outcome was the need for mechanical ventilation within 5 days of birth. Similar proportions of infants (33.0% vs 31.4%) met this outcome, and there were again no differences in secondary outcomes. In fact, the proportions of infants breathing room air at 36 weeks' postmenstrual age were surprisingly similar (78.6% vs 78.1%).12
Taken together, these data demonstrate that a policy of early stabilisation in the delivery room using CPAP will result in similar, if not better, short-term outcomes compared with a policy of immediate intubation and mechanical ventilation (plus surfactant). This is the case even in infants born at less than 28–29-week GA, provided that somewhat liberal failure criteria are chosen to allow infants needing surfactant receiving this early to reduce the risk of pneumothorax. Recent data show that an FiO2 threshold of 0.45 for intubation in infants with Respiratory Distress Syndrome (RDS), initially managed on CPAP, will shorten the time to surfactant delivery without a significant increase in intubation rate.13 Although there are no data yet on the longer term safety of CPAP as the primary mode of respiratory support in extremely low-gestational age neonates (ELGANs), the current body of evidence suggests that a policy of selective intubation and mechanical ventilation may result in better outcomes. Results from a study soon to be published in full suggest that the proportion of infants receiving mechanical ventilation can be further reduced by instilling surfactant via a thin catheter inserted into the trachea during CPAP therapy.14
Avoiding low oxygen SATURATIONS in extremely preterm infants
For many years, data from observational studies seemed to support a more restrictive use of oxygen in ELGANs. In fact, one such study, comparing outcomes in five centres with four different oxygen target ranges, reported that ROP rates were lowest in the centre with the lowest target, while mortality and rates of cerebral palsy were similar.15 Thus, it came somewhat as a surprise when the first of four large RCTs on the effects of different target ranges for SpO2 in ELGANs reported its neonatal outcomes. Study oximeters had been manipulated so that they displayed SpO2 values that were either 3% higher or lower than actual values, yielding groups with target ranges of 85–89% and 91–95% SpO2, respectively. Primary outcome parameter was a composite of threshold ROP or death before discharge. While this outcome was similar between groups, death before discharge occurred more frequently in the group targeting 85–89% SpO2 (19.9% vs 16.2%, RR 1.27, 95% CI 1.01 to 1.60).16 Six months later, the UK, Australian and New Zealand BOOST II trials, designed to apply the same target ranges, stopped patient recruitment after they also detected a significantly higher mortality risk in their infants randomised to the low oxygen group. Here, oximeters had a software upgrade while studies were ongoing that resulted in a clearer separation between the high and the low oxygen group (for details, see Stenson et al17). A joint safety analysis of survival to 36 weeks' postmenstrual age, pooling 2315 infants from these two studies (1055 recruited after the change in algorithm), found a 65% higher mortality risk for infants randomised to the low oxygen group after the change in algorithm (115/527 vs 70/528 deaths, RR 1.65, 95% CI 1.09 to 2.49), which had not been seen with the old algorithm. As a result, both studies closed recruitment. The fourth, however, also had data analysed by its Data Monitoring Committee (DMC), which saw no reason to report data from an interim analysis.
Many questions remain unanswered following publication of these data, for example, why a group difference was found in the BOOST II studies only after a change in pulse oximeter algorithm, while it was already found with the old algorithm in the SUPPORT study. Also, morbidity data are yet very limited. On the other hand, given the magnitude of the difference in mortality, the data currently available make it difficult to continue recommending SpO2 target ranges much below 90% in ELGANs, although we do not know at the moment what the exact optimum target range for SpO2 should be.
Methylxanthines are used for apnoea of prematurity (AOP) for more than 50 years, but studies performed until 5 years ago had been underpowered to assess neurodevelopment. Thus, it is most commendable that Schmidt et al enrolled over 2000 infants in a large RCT, the Caffeine for Apnea of Prematurity study. Caffeine (or placebo) was started during the first 10 days after birth in infants of 500–1250 g birth weight until no longer considered necessary for AOP treatment.18 Mechanical ventilation, CPAP and oxygen could all be discontinued approximately 1 week earlier in infants treated with caffeine. In the caffeine group, there was also a 40% lower risk of BPD (36% vs 47%; OR 0.6, 95% CI 0.5 to 0.8), a 30% lower risk of developing a symptomatic patent ductus arteriosus (OR 0.7, 95% CI 0.5 to 0.8), and a 40% reduction in the risk of developing ROP >stage 3 or requiring treatment for ROP (5% vs 8%; OR 0.6, 95% CI 0.4 to 0.9). Most important, however, death or disability at 18 months corrected age, the primary outcome, was reduced by 23% (OR 0.77, 95% CI 0.64 to 0.93). This benefit was particularly strong for cerebral palsy: 4.4% versus 7.3% of infants had this outcome (RR 0.58, 95% CI 0.39 to 0.87).19
In subgroup analyses, the effect of caffeine on the primary outcome was found to be restricted to infants requiring respiratory support at randomisation.20 Interestingly, the reduced duration of the need for ventilatory support was only evident in those who were randomised within their first 3 days of life. Caffeine appears, at the moment, to be the only drug identified that will reduce cerebral palsy, at least in infants <1250 g birth weight. Given that this effect appears to be restricted to those receiving ventilatory support at treatment onset, it is most likely mediated via a reduced need for mechanical ventilation rather than a reduced rate of bradycardia and desaturation.
Antivascular epithelial growth factor treatment for threshold ROP
One concern with putting the recent data on the potentially beneficial effect of higher oxygen levels into practice is that it may result in more infants developing ROP. In this regard, publication of the first RCTs on the effects of antivascular epithelial growth factor (antiVEGF) treatment for ROP is particularly timely. In this study, 150 infants (300 eyes) with stage 3+ ROP plus disease in zones I or II were randomised to either intravitreal bevacizumab (0.625 mg) or conventional laser therapy. Primary outcome was recurrence of ROP requiring treatment before 54 weeks postmenstrual age. ROP recurred in 4 infants in the bevacizumab group and 19 in the laser-therapy group (4% vs 22% of eyes treated, p=0.002).21 Subgroup analysis, however, showed significant treatment effects only for zone I ROP, not for zone II.
VEGF is needed, among other things, for neuronal survival and lung development. Systemic (endogenous) VEGF levels are still almost fully suppressed 1 month after a single intraocular bevacizumab injection, clearly raising concern about potential long-term side effects of this new therapy. It is clearly too early, at the moment, to advocate it as a replacement for laser therapy in stage 3+ ROP, but it certainly raises hopes that this disease may soon lose some of its threat.
Hypothermia treatment for hypoxic-ischaemic encephalopathy
Asphyxia contributes 23% to the 4 million neonatal deaths that occur annually worldwide.22 Until recently, no proven method of preventing or at least mitigating the hypoxic damage was available, but evidence from experimental studies suggested that cellular injury occurs only some time after the hypoxic insult, leaving a window of opportunity for preventing brain damage after the hypoxic insult had occurred. Hypothermia is one such intervention, but despite encouraging results from several RCTs, an expert panel convened by the National Institutes of Child Health and Human Development concluded as recently as 2006 that existing evidence was insufficient to make hypothermia standard of care for hypoxic-ischaemic encephalopathy (HIE).23
In late 2010, however, a meta-analysis was carried out of data from 10 RCTs, encompassing 767 infants with moderate to severe neonatal encephalopathy who received hypothermia (core temperature, 33–34°C) for 72 h or standard care (with three trials reporting 18-month outcome data). This study found that therapeutic hypothermia reduced the combined rate of death and severe disability at 18 months by one fifth (RR 0.81, 95% CI 0.71 to 0.93, number needed to treat (NNT) 9 (5–25)). Expressed differently, therapeutic hypothermia increased the odds of survival with normal neurological function by 53% (RR 1.53, 95% CI 1.22 to 1.93, p<0.001; NNT 8 (5–17)). A subgroup analysis on whether it is possible to define patients who are too severely affected by hypoxic damage to benefit from hypothermia treatment showed that outcome was significantly improved by hypothermia in moderately, but not severely, affected infants.24 The latter finding, however, is in contrast to data from an even more recent trial on hypothermia, enrolling 129 neonates, with 77 classified as suffering from severe HIE. Here, the odds of death or severe disability at 18–21 months in the hypothermia vs control group were 0.21 (0.09–0.54) in the total group and 0.17 (0.05–0.57) in the more severely affected subgroup.25 With just one study supporting the use of hypothermia in severely affected neonates with HIE, it remains an individual decision whether or not to offer this treatment to such infants. To summarise, in infants with HIE, moderate hypothermia is associated with a consistent reduction in death and neurological impairment at 18 months and should be applied, within 6 h of birth, to all full-term neonates showing signs of moderate (and potentially also more severe) HIE. Whether this also holds true for preterm or near-term neonates and whether the beneficial effect of hypothermia can be further enhanced by combining hypothermia treatment with other strategies such as xenon insufflation of their inspired air26 awaits results from further studies, some of which are currently ongoing.
Although the last 5 years have brought astonishing progress in neonatal care, we need more data on new approaches in many areas, including those on a more individualised, development-orientated care provision and on better ways to ensure optimal nutrition and growth of ELGANs. There are equally important recent findings that have not been mentioned up to this point in this review. These include data from studies on the use of probiotics for NEC prevention,27 ,28 consequences to be drawn from the conflicting data on the use of nitric oxide for BPD prevention (nicely reviewed in Sosenko and Bancalari29 and Donohue et al30) or on the increasing evidence that volume-controlled ventilation may be preferable to pressure-controlled ventilation in infants with RDS.31 It is no coincidence that a common theme of the recent progress summarised here is that most is based on data from large RCTs, and this should continue to be the route by which progress is made in the treatment of this extremely vulnerable patient group.
CFP is grateful to Dirk Bassler, MD, MSciEpi, for reviewing this manuscript and to Derek Stebbens, MA, for linguistic assistance.
Competing interests None.
Provenance and peer review Commissioned; internally peer reviewed.