Hylke Salverda

72 Chapter 4 during their entire admission. Considering the results from our cross-over study14, we hypothesized that OxyGenie is more effective in maintaining SpO2 within the target range during respiratory support of preterm infants. Methods Study setting We retrospectively retrieved prospectively collected data of all patients born at 2429 weeks gestation, admitted to the NICU of the LUMC between October 2015 and November 2020. Our centre is a tertiary-level perinatal centre in Leiden, the Netherlands, and we have an average of 100 infants born under 30 weeks of gestation per annum. In the Netherlands, no ethical approval is required for anonymised studies with patient data collected for standard care. The Medical Ethical Review Committee of Leiden Den Haag Delft provided a statement of no objection for obtaining and publishing the anonymised data. Infants were excluded if they: were admitted >24 hours after birth, had major congenital abnormalities, or received less than 72 hours of respiratory support. A minimum of 72 hours was chosen to exclude unrepresentative extreme scores from infants transferredwithin days after birthwith little respiratory support, or infantswho died shortly after birth. In both situations the impact of automated oxygen titration would likely be negligible. The AVEA cohort consisted of infants admitted between October 18th 2015 (three months after implementation of CLiO 2 into standard care) to December 1st 2018, the SLE cohort consisted of infants born between February 1st 2019 and November 1st 2020, allowing for a wash-out period of two months. Respiratory support Both the OxyGenie and the CLiO2 algorithm change the FiO2 automatically according to the measured SpO2, where generally larger deviations from the SpO2 target range lead to larger changes in FiO2. Both are adaptive in the sense that when a patients’ average supplemental oxygen requirement is higher, adjustments in FiO2 will also be larger. One difference lies in the exact way the adjustment in FiO2 is calculated: both OxyGenie and CLiO2 base the magnitude of adjustment on how far out of the target range the current SpO2 is, and the trajectory of recent values, but OxyGenie also takes past values into account (by addition of an integral term, the sum of past differences between desired and measured SpO2). More detail on the function of these algorithms can be found elsewhere.15