The investigators aim to measure the effect of targeting premature babies to a slightly higher oxygen saturation target range (92-97%) than routinely used, for a brief period, to plan a future larger study of the effect of this on clinical outcomes.

It is still unclear exactly what levels of oxygen premature babies need - both too little or too much oxygen in the first weeks after birth may be harmful. Previous studies used saturation monitoring (SpO2), where a small probe shines light through the skin and calculates how much oxygen is carried in the blood. These studies demonstrated using an SpO2 range of 91-95% rather than 85-89% was associated with more babies surviving and fewer babies suffering from a bowel condition called necrotising enterocolitis (NEC). However, targeting oxygen higher increased the number of infants who needed treatment for an eye condition called retinopathy of prematurity (ROP).

It is possible an SpO2 range higher than 91-95% would be associated with even better survival. It is also possible that a higher range might not improve survival but could increase the need for ROP treatment.

Infants born at less than 29 weeks gestation, greater than 48 hours of age and receiving supplementary oxygen would be eligible for inclusion. The study is at the Royal Infirmary of Edinburgh. Total study time is 12 hours for each infant (6 hours at the standard 90-95% range used in our unit, and 6 hours at 92-97%). It is a crossover study with infants acting as their own controls.

Based on previous research the investigators are confident these oxygen levels will not be dangerously high. To provide an additional measure of oxygen the investigators will also use a transcutaneous monitor for the 12 hour study period, which fastens gently to the skin and measures oxygen and carbon dioxide levels on the skin surface.

This study is a single centre randomised crossover study. Infants born at less than 29 weeks gestation, greater than 48 hours of age and receiving supplementary oxygen would be eligible for inclusion. The study is at the Royal Infirmary of Edinburgh. Total study time is 12 hours for each infant (6 hours at the standard 90-95% range used in our unit, and 6 hours at 92-97%). It is a crossover study with infants acting as their own controls.

A complete list of all monitoring that will be carried out is shown below:

1. SpO2 monitoring
2. TcPO2 monitoring
3. Heart rate monitoring (used to validate SpO2 readings as described below)
4. Arterial gas sampling (only if conducted by the direct care team as part of the routine care of the infant will these be recorded. No extra blood samples will be taken as part of the study)

Each infant will be monitored simultaneously with an SpO2 monitor and TcPo2 monitor, with data recording starting from the time the transcutaneous monitor finishes calibrating. SpO2 readings will be downloaded directly from the multiparameter patient monitor. SpO2 will be measured using a Siemens Infinity SC7000 multiparameter monitor. This monitor uses Siemens' Oxisure pulse oximetry technology and Nellcor Oximax saturation probes and incorporates ECG to reduce motion artefact. To reduce the influence of artefact further, data from an infants chest leads (recording heart rate) will also be downloaded. If there is a greater than 10 beats per minute difference between heart rate measured from the pulse oximeter and heart rate measured from the chest leads, SpO2 for that time point will be disregarded.

TcPO2 will be measured using a SenTec Digital Monitoring System with OxiVent sensor. TcPO2 is calculated by dynamic fluorescence quenching which measures oxygen molecules present in the vicinity of a fluorescent dye incorporated within the sensor surface. The sensor is operated at a constant temperature of 43 degrees Celsius. Control of sensor temperature and application duration are designed to meet all applicable standards and this monitoring device is used routinely in many neonatal units, with transcutaneous monitoring part of the monitoring equipment we use in selected infants in our unit and being transported by our local neonatal transport team. Temperature is supervised by two independent circuits, as well as by the monitor firmware.

Transcutaneous data will be transferred contemporaneously to a bedside PC via the proprietary software V-STATS (with V-CareNeT) version 4.01. We will ensure the time on the monitor is synchronised with the time displayed on the electronic patient record system recording SpO2 and heart rate. V-STATS software can identify any spurious spikes in TcPO2 caused by air bubbles and these identified segments of data will be discarded. Artefact has not significantly affected readings during the routine use of this monitor previously on our neonatal unit and with our neonatal transport team. The site of the transcutaneous probe will be rotated on each infant every 2 hours.

SpO2, TcPO2 and heart rate data will be recorded every second. The time of any arterial oxygen samples taken routinely during the study will be recorded from the unit Radiometer ABL800 FLEX blood gas analyser.

Inclusion Criteria:

• Infants born at less than 29 weeks gestation
• Infants greater than 48 hours of age
• Infants who are receiving supplementary oxygen

Exclusion Criteria:

• Congenital anomalies that would affect oxygenation (eg. cardiac defects, congenital diaphragmatic hernia)

Infants will be randomised to either higher (92-97%) or lower (90-95%) oxygen saturation target range, and then cross-over to the alternative range after 6 hours of monitoring (with 12 hours in total study time). The study is randomised but not blinded.