174 Chapter 10 %, p<0.05). There was no significant difference in time with SpO2 <80% (0.5 (IQR 0.1-1.0) % vs 0.2 (IQR 0.1-0.4) %, p=0.061). Long-lasting SpO2 deviations occurred less frequently during OxyGenie control. The OxyGenie control algorithm was more effective in keeping the oxygen saturation within TR and preventing hyperoxaemia, and equally effective in preventing hypoxaemia (SpO2<80%), albeit at the cost of a small increase in mild hypoxaemia. Chapter 4 is a continuation on comparing these two different automated oxygen control devices in preterm infants on time spent in different oxygen saturation ranges. Contrary to chapter 3, the entire stay in the NICU was investigated in a retrospective cohort study. Preterm infants (OxyGenie 75 infants, CLiO2 111 infants) born under 30 weeks of gestation receiving at least 72 hours of supplemental oxygen during respiratory support between October 2015 and November 2020 were studied. Inspired oxygen concentration was titrated by the OxyGenie controller between February 2019 and November 2020 and the CLiO2 controller between October 2015 and December 2018 as standard of care. Time spent within the SpO2 TR was higher during OxyGenie control (median 71.5 [IQR 64.6–77.0] % vs 51.3 [47.3–58.5] %, p<0.001). Infants under OxyGenie control spent less time in hypoxic and hyperoxic ranges (SpO2 <80%: 0.7 [0.4–1.4] % vs 1.2 [0.7–2.3] %, p<0.001; SpO2 >98%: 1.0 [0.5-2.4] % vs 4.0 [2.0-7.9] %, p<0.001). Both groups received a similar fraction of inspiratory oxygen (29.5 [28.0 – 33.2] % vs 29.6 [27.7-32.1] %, p=non-significant). Again, oxygen saturation targeting was better in the Oxygenie cohort, which resulted in less hypoxia and hyperoxia. To conclude this part, we report what the effect is of using one-per-second or one-perminute data in chapter 5. Large amounts of data are collected in neonatal intensive care units which could be used for research. It is unclear whether this data, usually sampled at a lower frequency, is sufficient for retrospective studies. We investigated what to expect when using one-per-minute data for descriptive statistics. One-persecond fraction of inspiratory oxygen and oxygen saturation data was processed to one-per-minute data and compared on average, standard deviation, target range time, hypoxia, days of supplemental oxygen, and missing signal. Outcomes calculated from data recordings (one-per-minute=92, one-per-second=92) showed very little to no difference. Neither did sub-analyses of recordings under 100 and 200 hours. In this study descriptive statistics of one-per-minute data were comparable to one-persecond and could be used for retrospective analyses. Comparable, routinely collected one-per-minute data could be used to develop algorithms or finding associations retrospectively.
RkJQdWJsaXNoZXIy MTk4NDMw