Hylke Salverda

158 Chapter 9 SpO2 was implied in a post-hoc analysis of the BOOST-II UK, where infants that died had a lower median SpO2 when compared to survivors in their own group (Died: 90% in the 85%-89% group, 94% in the 91%-95% group. Survived: 94% in 85%-89% group, 95% in 91-95% group). 32 Although chapter 3 provides causality through a randomised crossover study on which algorithm performs best, it should be noted that it remains unclear how generalisable the results are for a preterm infant’s full stay in the NICU. Different clinical conditions, which are certain to arise during the long stay of a very preterm infants, may demand different strategies or configurations. For example, a more responsive strategy may be more appropriate when apnoea of prematurity occurs frequently, which increases in frequency at a later postnatal age,33 whereas an infant with chronic lung disease may benefit from a slower weaning strategy. Moreover, in the first days after birth when respiratory insufficiency is often related to a surfactant deficiency, it will be necessary for an automated oxygen controller to quickly adapt when exogenous surfactant is administered, as this likely leads to a change in oxygen requirement. An algorithm which is programmed to be resistant to large changes could provide too much oxygen leading to hyperoxia. In chapter 4 we further investigated the differences between algorithms during the admission, throughout the entire range of postnatal age and postmenstrual age, by comparing oxygenation data of infants treated with OxyGenie or CLiO2. In this chapter we report on data collected from the six years we either used CLiO2 or OxyGenie as standard of care. We observed that with the Oxygenie better control of oxygenation was carried throughout all postnatal ages on the NICU. Infants treated with OxyGenie had spent significantly lower proportions of time spent in hyperoxia and hypoxia while the average FiO2 and group characteristics were not significantly different. Combined, the studies in chapter 3 and 4 provide clear evidence that for better control of oxygenation in the NICU, OxyGenie should be preferred over CLiO2. When we included periods where no supplemental oxygen was given, the difference in achieved target range time was smaller, but still significantly different. Both controllers achieved a very high proportion of time within target range (OxyGenie 92.5%, CLiO2 90.2%) when considering all pulse-oximetry data (i.e. periods with and periods without supplemental oxygen). As a result, one could question the clinical relevance of the observed difference between CLiO2 and OxyGenie. However, it is likely that oxygenation-related morbidity and mortality mostly has its genesis in days of respiratory instability. For example, a higher incidence of retinopathy of prematurity is found to be associated with intermittent hypoxia34 as well as hyperoxia,35 both