Hylke Salverda

160 Chapter 9 not demonstrate a possible harmful effect of using AOC as standard care. Although there has been no report of an adverse effect in any of the trials comparing automated oxygen titration with manual titration, concerns for masking clinical deterioration have frequently been expressed.23, 37 On the contrary, after training of clinical staff in our unit, assessment of basal oxygen requirement combined with the magnitude and frequency of interventions by the AOC device was used as an additional, objective indicator of clinical status. Although it has become clear that AOC greatly reduces hypoxia and hyperoxia, which are known to be injurious, there are several (unmeasured) confounding factors that can influence outcome in very preterm infants until hospital discharge. Preterm infants may experience many other potentially harmful sequelae before birth, during their admission (e.g. sepsis, intraventricular haemorrhage) and after discharge, on which automated oxygen control is unlikely to have influence. Also, as demonstrated in the oxygenation studies of this thesis, the gain in time within target range was mostly attributable to a reduction in mild hypoxia and hyperoxia, and we have no data on the duration of hypoxic events. A lack of reducing (long lasting) hypoxic events could have reduced the effect on clinical outcome. In addition, neonatal care is a rapidly changing field with frequent changes to standard care, which may influence outcomes of cohort studies in either direction, and may have negated the effect of automated oxygen control. For example, we also changed the lower limit of the target range to 90% instead of 85%. While we used large cohorts for the observational studies in this thesis, it is likely that an appropriately powered RCT is needed to measure an effect on important clinical outcomes. Currently, the FiO2-C trial -a large multicentre trial aiming to include 2340 patients- randomises between automated oxygen control or manual titration during the entire admission, and will compare the effect on clinical and neurodevelopmental outcome at 24 months of corrected age.38 There are four different AOCs allowed for the study, which will likely not be equally effective at maintaining SpO2 within TR. We demonstrated in this thesis that there can be a large difference in impact between AOCs. While using all available AOCs in a trial has likely been a pragmatic choice, the AOC with the best oxygenation control will have the largest treatment effect. Indeed, an example of the difference in impact by AOCs is given in chapter 8, where the OxyGenie group developed less morbidity compared to the CLiO2 group in a matched cohort study. Significantly fewer infants received treatment for ROP, infants received less intensive respiratory support and, although there were more supplemental oxygen days, the duration of stay in the NICU was shorter. Other short-