Hylke Salverda

156 Chapter 9 rule-based (chapter 2 figure 1), resembling an if-this-then-do-that mechanism; proportional-integral-derivative control (chapter 2 figure 2), a mathematical combination combining the current, past and future oxygenation; and adaptive control (chapter 2 figure 3), which tailors the algorithms response to the individual infant, for example to the severity of lung disease in the infant. Besides using these components in different degrees, algorithms also differ in the promptness of their response and what is targeted (i.e. the middle or the limits of the target range). All commercially available algorithms were demonstrated to achieve higher proportions of time within target range when compared to manual titration, however which algorithm is most effective is unknown, as is the effect on clinical outcome. Comparing these algorithms purely based on the available literature is difficult, considering a variation in choice of target range, pulse oximeter settings, ventilator mechanics, choice of inclusion/exclusion criteria, modes of respiratory support and the aims of the studies. Direct ‘head-to-head’ comparisons of AOCs are required, so clinicians know what to expect when using a specific automated oxygen controller. The aim of this thesis was to evaluate outcomes after using automated oxygen control. We set out to describe currently available devices and their differences, we will start to discuss our data on oxygenation in the NICU while using automated oxygen control, and will thereafter discuss clinical and long-term outcome. Effectivity of automated oxygen control algorithms on oxygenation of preterm infants in the NICU The preferred way to compare two automated oxygen control algorithms would be to compare the algorithms within the same patients at the same time. In this way, the only changing condition would be the ventilator with the built-in automated oxygen controller, while most other aspects remain equal. This thesis describes the first comparative study between automated oxygen controllers within the same infants. The lack of evidence in this regard may be explained by the difficulties met when performing such a study. First of all, researchers (temporarily) need to have a running stock of two different brands of ventilators, including disposable materials required to use these machines. Staff would need to be competent to work with both ventilators, including not only bedside staff but also supportive staff such as technicians. Secondly, actual execution of the study includes switching between ventilators in (sometimes invasively) supported babies. Preparing and switching ventilators increases the workload for already busy bedside staff. Even though it is unlikely a patient would notice a change in ventilator, parents may be reluctant