Slow enrolment

Currently, newborns receive 100% supplemental oxygen by free flow when they remain cyanotic despite demonstrating regular respiratory effort. Resuscitating infants with continuous positive airway pressure (CPAP) in room air may improve oxygen saturations more quickly than providing FFO2 because of its ability to establish functional residual capacity in the lungs. Our primary hypothesis is that in this blinded, randomized control trial, more infants (≥35 weeks gestation) resuscitated with CPAP in room air will have an oxygen saturation ≥80% at five minutes of age compared to infants resuscitated with the 50% FFO2.

INTERVENTION Resuscitation will proceed as per standard of care and infants are only to receive respiratory support if they remain cyanotic despite 30 seconds of spontaneous regular respirations. All infants will have a pulse oximetry probe placed on the right hand (pre-ductal position) immediately after birth. Alarm limits will not be set. The pulse oximetry probe will first be placed on the infant and then attached to the cable connected to the pulse oximeter as this has been shown to result in a shorter time to detecting a good quality signal. The pulse oximeter will be set at "maximum" sensitivity with an averaging time of 2 seconds.

Randomization will occur at the time of delivery. Infants that meet the entry criteria will be randomized to resuscitation with one of two Neopuff CPAP circuits:

GROUP 1 (21% CPAP) - infants will receive CPAP with 21% oxygen at 6 cm H2O continuously for at least 5 minutes.

GROUP 2 (50% FFO2) - infants will receive the equivalent of 100% FFO2 for at least 5 minutes. Note that infants receiving 100% FFO2 actually inspire approximately 50% oxygen because of entrained room air. To allow us to maintain blinding of the intervention, infants in the FFO2 group will actually receive 50% oxygen provided via a T-piece infant resuscitator that has been modified so as not to deliver any pressure to the airways.

The FiO2 (21% or 50%) will be set by the study investigator attending the delivery. An anaesthesia monitor (Datex Ohmeda) will be connected in-line between the Neopuff T-piece and the face mask to record airway pressures, end tidal carbon dioxide levels and concentration of inspired oxygen (to confirm that infant are receiving either 21% or 50% oxygen).

A treatment failure for both groups will be defined as any of the following:

1. Infant stops making a spontaneous respiratory effort
2. Infant fails to achieve an oxygen saturation of 75% by 3 minutes of age or 80% by 5 minutes of age. These values are near the lower limit of normal for healthy infants transitioning in room air.
3. Heart rate <60 beats per minute

In the event of a treatment failure, resuscitation will proceed with intermittent PPV delivered with 100% oxygen using a flow-inflating bag and mask system which is available on all of our resuscitation warmers. We have chosen to switch to a bag and mask system in this instance so as not to unblind the study (see section on blinding). Chest compressions will be started after 30 seconds of tactile stimulation and positive pressure ventilation have failed to increase the HR to >60 bpm. We recognize that there is no data to suggest that providing 100% oxygen is the best approach in treating bradycardia in the preterm newborn. However, it is the current standard of care. None of the preterm infants enrolled in our room air resuscitation study have required chest compressions.

If the pulse oximeter does not register a good signal, resuscitation will continue at the current inspired O2 concentration as long as the HR >100 bpm. This also applies to the period of time immediately following birth prior to the POM giving a clear reading. In our experience, we are able to reliably obtain a good-quality pulse oximetry signal within 30 seconds of probe application in newborn infants using the method suggested by O'Donnel and colleagues.

Blood gases will be drawn within 30 to 60 minutes of admission to the NICU as per routine practice for infants admitted for respiratory support. All intubated infants will have arterial access established via either an umbilical artery, left radial artery, right or left posterior tibial arteries (all post-ductal in location) or via the right radial artery (pre-ductal location). Infants without arterial access will not have blood gas data collected given the limited utility of PaO2 estimates from capillary or venous samples.

RANDOMIZATION AND CONCEALEMENT Block randomization will be used with random block sizes of 2 and 4 using a computer generated randomization schedule to increase the likelihood of equal enrolment into each group. The randomization procedure will occur on site and treatment assignment (21% CPAP, 50% FFO2) will be kept in opaque (manila envelopes with foil inserts), sequentially numbered, sealed envelopes in a locked drawer in the NICU. Infants will be randomized immediately prior to birth since randomizing infants following clinical assessment would delay the onset of resuscitative procedures. It is recognized that some infants who are randomized will not require respiratory support. While central and computer randomization have definite advantages over the use of opaque envelopes, their use in a resuscitation study is difficult given the time constraints. Treatment assignment will be concealed from everyone apart from the study investigator attending the delivery. The resuscitation team, biostatistician, data collector and clinical team providing ongoing care for the infant in the NICU will be blinded to the intervention. This will be achieved by randomizing infants to either a modified or standard T-piece infant resuscitator circuit. The circuit for the 50% FFO2 group will be altered so that pressure is not generated when the mask is held against the infant's face. The circuit for the 21% CPAP group will be unaltered and hence will generate CPAP when held against the face. It is not possible to detect if a unit has been modified without dismantling it. The circuit pressure gauge will also be covered. The study investigator will not be blinded so that he/she can appropriately set the FiO2 to either 21% or 50%.

We have a custom mobile research cart which shields the oxygen blender, pulse oximetry monitor, T-piece infant resuscitation unit and the anaesthesia monitor from the view of the resuscitation team

SAMPLE SIZE CALCULATION The sample size estimate is based on the primary outcome of proportion of infants reaching an oxygen saturation of ≥80% by 5 minutes of age. A sample size of 138 infants (69 in each group) would be required to detect a difference of 20% (71% to 91%) with 80% power at the 5% level of significance (two-tailed). The sample size estimate for the secondary outcome measure of proportion of infants reaching an oxygen saturation of ≥90% by 10 minutes of age is 214 (107 in each group). This will have 80% power to detect a difference of 20%(43% to 63%) between two adjacent means at the 5% level of significance (two-tailed test).

We plan to enroll 220 infants (110 in each group) to provide adequate power to address both our primary and secondary outcomes.

DATA ANALYSIS An intention to treat principal will be used so that infants receiving respiratory support will be analyzed in the group to which they were randomized. Exclusion after randomization will occur only for those infants that would not have qualified to receive 50% FFO2 as per NRP guidelines. A comparison of the proportion of infants achieving the target saturations at 5 and 10 minutes of age will be performed using the Chi-squared test.

Other clinical parameters that are continuous variables will be analyzed using t-tests where the distributions are approximately normal. These variables include duration of respiratory support in NICU, length of stay in NICU and PaO2 at time of NICU admission. If normalizing transformations cannot be found or if censored observations occur, the time variables will be analyzed using survival analysis techniques. Binary outcomes (occurrence of air leaks, IVH, death, NICU admission) will be compared using the Chi Square test. Apgar scores at 1, 5 and 10 minutes are ordinal variables and will be compared using the Mann-Whitney U test (Wilcoxon rank sum test). An interim analysis is not planned.

Inclusion Criteria:

1. Inborn
2. ≥35 weeks gestational age
3. Neonatal resuscitation team present at delivery
4. Central cyanosis despite 60 seconds of spontaneous respirations in room air. Sustained respirations are defined as: spontaneous breathing that is regular in rhythm and sufficient to maintain a heart rate ≥100 beats per minute such that provision of artificial tidal volume breathing is not required.

Exclusion Criteria:

1. Lethal anomalies
2. Cyanotic congenital heart disease
3. Known hemoglobinopathy
4. Risk of persistent pulmonary hypertension (eg, meconium aspiration, conditions associated with pulmonary hypoplasia such as prolonged)
5. Known anatomical lung abnormalities
6. Conditions where CPAP is contraindicated (eg, congenital diaphragmatic hernia). Mode of delivery will not be a criterion for inclusion or exclusion.