Hylke Salverda

162 Chapter 9 management system, precluding human error and recall bias. For clinical outcome we were able to have two independent researchers check all electronic patient records, in which data is collected prospectively as part of standard care. While the current standard definition of BPD 46 in chapter 6 and chapter 8 has been used, this definition does not take into account the use of AOC. The general consensus is that during a day supplemental oxygen should be given for at least 12 hours to be counted towards the 28 days required for the diagnosis of BPD. During automated oxygen control the administered fraction of oxygen may only intermittently be above 0.21 in a 24-hour period, and this may not be predictive of BPD, for example when these brief moments are linked to apnoeic events. Depending on what criteria are used to define BPD, significantly more infants would be classified as having BPD. Thus, the standard BPD definition may be unsuitable when AOC is used as standard care. Loss to follow up is unfortunately common in follow-up research and may lead to bias.47 In chapter 7 there was a relatively high rate of missing data due to loss to follow-up (pre-AOC 6.9%, post-AOC 10.6%). The majority of missing children were transferred to another NICU in the neonatal period and had subsequent followup there, therefore we expect them to be missing at random and not related to neurodevelopmental outcome. However, children lost to follow-up may be under treatment in a special care facility and therefore not missing at random. Parents may be less inclined to present their child for follow-up when they already receive regular tests in such a facility. To prevent biased results due to missing such children, we requested data for all children tested elsewhere. The strength of this study was that the children are always tested by trained professionals as part of a standardized national follow-up programme, improving the repeatability and reliability of the assessment of neurodevelopmental outcome.