Respiratory disease remains a major cause of morbidity and mortality in preterm infants, and both mortality and morbidity from bronchopulmonary dysplasia (BPD) remain high, particularly in extremely preterm infants. Experimental work has suggested that BPD is promoted by mechanical ventilation,¹ and recent studies suggest that treatment soon after delivery with early nasal continuous positive airway pressure (nCPAP) was apparently associated with less lung damage compared with that in controls with intermittent positive pressure ventilation.^2,3 Early nCPAP might be beneficial for infants, but many of the data are anecdotal and the evidence base is incomplete.^4–8

Any treatment that aims to ameliorate BPD will have to be effective in the extremely preterm population, in whom the disease is most common and severe, and probably be suitable for use very soon after birth when many of the major predisposing factors such as chorioamnionitis,⁹ surfactant deficiency^10,11 and increased lung water content¹² are present at birth. Variables around the time of birth such as the need for positive pressure ventilation may also help to predict which infants could benefit most from early nCPAP.¹³ However, clinicians may be deterred from instituting nCPAP soon after delivery in very immature infants in the belief that these patients lack the ability to sustain spontaneous ventilation and are likely to need reintubation.

Inadequate information on the ability of very preterm infants to maintain appropriate gas exchange on nCPAP soon after delivery could impede the assessment and implementation of nCPAP as a useful treatment for preterm infants. Some studies have looked at the ability of preterm infants to sustain nCPAP without undergoing reintubation for further mechanical ventilation,^4,5,14–16 but none has specifically examined the chance of reintubation in this vulnerable group, where prophylactic surfactant is often given before extubation to nCPAP.

In recent years, like many other institutions, the Neonatal Intensive Care Unit, Queen Charlotte’s and Chelsea Hospital London, UK, has developed a culture that aims at early extubation for all preterm infants, and we have gathered information on the natural history and the probability of reintubation for infants born at <27 weeks’ gestation who were established on nCPAP in the first week after delivery.

PATIENTS AND METHODS

Data collection

During 2001, we instituted a change in practice whereby we moved from mechanical ventilation being the prime means of respiratory support to nCPAP. We had routinely given prophylactic surfactant to extremely preterm infants since 1996, and this practice was continued. nCPAP after prophylactic surfactant rapidly became an established practice in the neonatal unit. A retrospective, observational study was conducted over a period of 1 year between November 2002 and October 2003 to review this method of respiratory support. We examined the clinical records of all infants admitted to the Neonatal Intensive Care Unit, Queen Charlotte’s and Chelsea Hospital, at <27 weeks’ gestation, as estimated from the date of the mothers last menstrual period (LMP) or by early ultrasound imaging where this date was uncertain.

Standardised data were collected on all infants who were admitted to the neonatal intensive care unit within 24 h of delivery, recording gestational age at birth, birth weight, sex, antenatal corticosteroid and postnatal surfactant treatment, and mode of ventilation on postnatal days 1–7 as well as on day 28 and week 36 of post menstrual age. The timing and reason for all reintubations were noted. Early outcome data were collected on the number of infants receiving increased fraction of inspired oxygen (Fio₂) at 36 weeks’ post menstrual age. Postmenstrual age was defined as the age in weeks, starting from the first day of the LMP when the date of LMP was known or by early ultrasound imaging where this date was uncertain.

Early management

All inborn babies were placed in a plastic bag immediately at birth to minimise heat and fluid loss and then electively intubated. Curosurf (Chiesi, Cheadle, UK), as a single bolus dose of 120 mg, was given within 10 min of delivery. Initial ventilation was provided with a Neopuff Infant Resuscitator (Fisher and Paykel, Berkshire, UK) with a positive end expiratory pressure (PEEP) of 5 cm H₂O. We aimed to keep the peak inspiratory pressure (PIP) low at about 16 cm H₂O, or if unable to achieve this, the lowest PIP to maintain appropriate chest movement. An inspiratory time of 0.5 s, rate of 40/min and an initial Fio₂ of 0.3 were advised. Oxygen requirements were monitored by pulse oximetry with arterial oxygen saturation (Spo₂) alarm limits set at 85–92%. All babies were transferred from the delivery room to the neonatal unit on conventional mechanical ventilation, in a transport incubator. They were placed on an open radiant heat cot in a “humidification tent” with convention ventilation continued using the SLE 2000 ventilator. The recommended ventilation criteria were the lowest PIP to maintain adequate but not excessive chest movement, inspiratory time of 0.5 s, Fio₂ to maintain Spo₂ 88–90% (alarm limit set at 85–92%) with a PEEP of 5 cm H₂O. The rate was determined after an initial capillary blood gas to maintain the pH >7.2 with partial pressure of alveolar carbon dioxide (Paco₂ between 5.5 and 8 kPa (45–60 mm Hg).

The infants were then assessed for extubation. To be suitable for extubation, the infant had to be haemodynamically stable and should have shown some spontaneous respiration. The PIP pressures had to be relatively low at ⩽18 cm H₂O and the Fio₂ <0.3 to maintain Spo₂ 88–90% (alarm limit set at 85–92%) with a PEEP of 5 cm H₂O. An acceptable capillary blood gas had a pH >7.22 with Paco₂ <8 kPa (60 mm Hg). Those judged clinically suitable for extubation were nursed prone; while they were prone peripheral venous access was obtained and infants then prepared for extubation to nCPAP. Those judged clinically not immediately suitable for extubation had umbilical arterial and venous catheters sited and were reassessed frequently to see whether extubation would be possible. It was not the practice to wean the ventilator rate in those infants judged suitable for extubation postnatally; such infants were extubated from a rate of 30–40 beats/min. For those who required a more prolonged period of ventilation, the rate was reduced to 20–35 beats/min before extubation. The rate was never reduced < 20 beats/min. Around the time of extubation, infants were intravenously given a loading dose of caffeine citrate (caffeine base 25 mg/kg in two equal doses 1 h apart); a maintenance dose (caffeine base 6 mg/kg once daily) was continued if extubation was successful. Caffeine levels were not monitored during the study period.

All infants received nCPAP via nasal prongs from the Infant Flow Driver (EME, Brighton, UK) with a Fisher and Paykel 850 humidifier, and were nursed while in the prone position, with every effort made to ensure an effective seal and good audible air entry. nCPAP pressure was initially set at 6–7 cm H₂O and quickly increased to a maximum of 9 cm H₂O if chest retractions, increasing tachypnoea or increasing oxygen requirement were observed. Fio₂ was adjusted to maintain Spo₂ 88–90% (alarm limit set at 85–92%).

The infants continued on nCPAP as long as there was an adequate respiratory drive, the criteria for which included Fio₂ <0.5, pH >7.20, with no limit set for Paco₂ provided the pH was maintained. Failure of nCPAP requiring reintubation was characterised by recurrent apnoea and bradycardia not responding to stimulation and positioning, respiratory failure with a pH persistently <7.20 or Fio₂>0.5 to maintain acceptable saturations and, finally, sepsis. Sepsis was defined as a clinical picture with bacterial growth seen on blood culture, change in C reactive protein levels and blood film, or a clinical picture with a change in C reactive protein levels and blood film in the absence of bacterial growth seen on blood culture. The infants were then reintubated and ventilated with the least support to achieve adequate gas exchange and chest inflation, as described earlier.

nCPAP was reduced gradually, with increasing time spent on low-flow nasal cannula oxygen delivered by a low-flow oxygen meter (10–50 cm³) without humidification if the baby had Fio₂ <0.25 on nCPAP. Additionally, there needed to be a good respiratory effort with no chest retractions, and no apnoea and bradycardia that required active management with positive-pressure T-piece ventilation.

Patent ductus arteriosus was managed conservatively. Routine echocardiography and prophylactic treatment were not undertaken. If a baby repeatedly failed extubation to nCPAP for no reason other than a large duct with bidirectional flow seen on an echocardiogram, then treatment with ibuprofen was considered. Surgical ligation was reserved for cases where the condition of the baby worsened on ventilation and all other causes were excluded.

All infants born outside the Queen Charlotte’s and Chelsea Hospital were transferred on conventional ventilation; all had received surfactant within 30 min of birth but not always prophylactically. Once admitted to the neonatal unit in this hospital, they were managed as described earlier.

Statistical analysis

These retrospective observational data were analysed using a bayesian approach to make a point estimate (with 95% confidence limits) of the risk of reintubation for an individual infant.¹⁷

We estimated the probability of successful extubation by calculating the β density according to the simplified function given by Berry¹⁷:

β (a,b) = p^a(1−p)^b

where p, population success proportion; a, number of successful extubations; and b, number of failed extubations.

The population success proportion p is any value between 0 and 1, and represents the possible combinations of success and failure at extubation. If all infants fail extubation, p = 0; and if all succeed on CPAP, p = 1. If p = 0.5, then half the infants extubated succeed in remaining on CPAP.

Evaluating the function creates the β density, which is a model of the probability of success and failure, analogous to the more familiar model of a linear relationship given by the formula y = mx+c. β density curves have several useful properties as models of probability. For large values of a and b, β density curves approximate a gaussian distribution, and thus a mean and 95% confidence limit can be meaningfully calculated. As a and b increase, the curves become progressively narrower and, thus, the confidence limits become narrower. If a>b (the number of successes >the number of failures), the mean value of the curve moves to the right, and the mean value becomes higher.

The mean value of β density calculated using a and b is the predictive probability, which formally estimates the chance that the next time a trial is made it will be successful. In the context of this dataset, predictive probability estimates the probability that the next infant established on CPAP using the practices implemented in this study will escape reintubation.

Estimation of the predictive probability (with the 95% confidence limits for the estimate) has the advantages that it provides a probability estimate that relates directly to an individual case (indeed, the next infant extubated under the same conditions as the dataset was collected) and is of direct relevance to daily clinical practice. A more complete description of this bayesian approach to predicting the outcome of individual cases is given by Berry.¹⁷

This observational study is not randomised and did not set out to test any hypothesis about the effect of early CPAP on respiratory outcome. However, there may be some interest in an exploratory examination of the relationship between extubation at 48 h after delivery and respiratory outcome at 36 weeks of gestation, and we analysed this using Fisher’s exact test.

RESULTS

Over this 1-year period, 60 infants born at <27 weeks’ gestation were admitted to the neonatal intensive care unit Queen Charlotte’s and Chelsea Hospital. Seven infants were excluded from further study as they were admitted 24 h after delivery. Table 1 gives the demographic details of the infants.

The survival rate in infants born at <24 weeks’ gestation was 25%, and 50% in those born at 24–24.8 weeks. Of those babies successfully extubated at 48 h, 88% (23/26) survived to discharge. Figure 1 shows the natural history of infants during the first week after delivery; infants receiving mechanical ventilation, those extubated and the cumulative number of deaths are shown for each day. Extubation to nCPAP was attempted in 21 infants on day 1, seven of whom needed reintubation. Table 2 describes the number of extubations in the first 7 days after delivery, with the predictive probability (with 95% confidence limits) of successful establishment of nCPAP during that day. For postnatal days 3 and 7, the numbers are too small to allow calculation of reliable probabilities.

The youngest infant to be sustained on nCPAP after extubation on day 1 was born at 23.8 weeks. In all, 25% of infants born at <24 weeks’ gestation were extubated within the first 7 days. No infant with a birth weight <660 g was successfully established on early nCPAP.

Analysis of early outcome data showed that 18 infants were receiving oxygen at 36 weeks’ corrected gestational age. Table 3 relates ventilation status at 48 h of age to outcome at 36 weeks’ postmenstrual age. Fisher’s exact test shows that there is a significant difference between the groups (p = 0.019); however, as this study is observational, retrospective and not randomised, this implicit hypothesis testing should be treated with caution.

DISCUSSION

The results of our study over 1 year show that most infants born at <27 weeks’ gestation could be treated with nCPAP during the first week of life without requiring reintubation during that time; many for whom this was not possible were extremely unwell and died. By postnatal day 2, more than half the surviving infants were breathing spontaneously on nCPAP.

Only data related to the first week are presented in detail, as this is the period of greatest interest, because any treatment with the potential to reduce BPD must be effective during this time. A review of our records shows that during this period all cases of reintubation were due to failed nCPAP, whereas after 7 days one third were due to other factors, such as sepsis, necrotising enterocolitis or the need for surgery. As these events could have occurred equally in babies receiving mechanical ventilation, the results from the first 7 days give a more meaningful picture of the ability of these infants to sustain themselves with nCPAP support.

This study has limitations. As a retrospective observational study with no randomisation, it reflects a particular practice. Potentially relevant features include a specialist obstetric service experienced in preterm labour and delivery, high incidence of antenatal steroid and prophylactic surfactant treatment, early use of caffeine, routine early use of parenteral nutrition with early trophic breast milk feeds, skilled nurses experienced in the use of nCPAP soon after delivery, and detailed attention to positioning, with the infant lying prone and at an angle of about 30°. However, the data show that in routine tertiary neonatal practice, it is possible to use nCPAP soon after delivery in a high proportion of very immature infants.

The numbers of infants in the study are too small to make a precise assessment of the chances at different gestational ages or birth weights. However, no baby weighing <660 g sustained nCPAP without requiring reintubation, whereas 25% of infants born at 23 or 24 weeks’ gestation were successful in doing so. This suggests that diaphragmatic and intercostal muscle bulk may be more important than the maturity of the respiratory drive in sustaining ventilation on nCPAP, and this possibility deserves further investigation.

In conclusion, early nCPAP treatment seems to be possible and safe in infants between 23 and 27 weeks’ gestation. Further studies of the use and value of nCPAP in this age group are justified.