Abstract
Background: Children who sleep less are at increased risk of developing obesity. Although it appears that children may be more susceptible to overeating, the mechanism(s) that may lead children to eating more when they are tired remain unconfirmed. Furthermore, the role that digital screens may play in mediating the relationship between sleep and obesity is yet to be established, which may be partly due to the difficulty of objectively measuring screen use across the day. Therefore, the aims of this thesis are to explore whether getting less sleep impacts problematic eating behaviours and screen use in children, and to develop a methodology to objectively compare all day screen use in children.
Methods: 105 children from Dunedin New Zealand took part in the DREAM study, a randomised cross-over trial that asked children to go to bed one hour earlier each night for one week (sleep extension), and one hour later each night for one week (sleep restriction). At the end of the week, the caregivers were asked to complete an adapted version of the Child Eating Behaviour Questionnaire (CEBQ), which asks about eight problematic eating behaviours over the past 7 days. Child participants completed a desire to eat (image based) computerized questionnaire at the end of each week to assess whether mild levels of sleep loss affected children’s susceptibility to wanting non-core foods. Energy intake from 24-hour diet recalls and an eating in the absence of hunger experiment were used to validate the desire to eat questionnaire and the CEBQ for short term use in children. To objectively measure changes in screen use between the sleep conditions, the child participants were asked to put on a wearable camera that took photos every 2 seconds for two days of each intervention week (day 3 and day 7). Total recreational screen use was calculated for the before school, after school and weekend periods in participants who were adherent to wearing the cameras during both intervention weeks. A coding protocol was developed to allow comparison of screen use measures using different coding rules, and exploratory analyses were undertaken to see how different coding assumptions and intervals of image capture affect estimates of total screen use.
Results: Children slept 37.1 (95% CI -46.0, -28.1) minutes less per night during sleep restriction compared with extension. Validation analyses showed that behaviours from the adapted CEBQ were associated with objective measures of energy intake, indicating that the adapted version appeared appropriate for measuring short-term eating behaviours in children. Intention to treat analyses showed no significant differences in their desire to eat non-core foods (mean difference, 95% CI: -0.9; -3.0, 1.1) when sleep was restricted. Children engaged in more emotional overeating (0.12; 0.01, 0.24) and undereating (0.15; 0.03, 0.27) when tired, with exploratory analyses revealing different findings by weight status; participants who were overweight had decreased satiety responsiveness (-0.30; -0.48.-0.12), and increased eating speed (0.27; 0.10, 0.44), while normal weight children engaged in more emotional undereating (0.24; 0.11, 0.38) and food fussiness (0.15; 0.03, 0.28). In those participants who were adherent to wearing the cameras (n=49), there was an increase in screen use during the week of sleep restriction compared to sleep extension, within the after-school period only (18.7 to 66.3 minutes), with a higher proportion (1.2%-6.0%) of their day spent using screens during this period. No differences in screen use were found over the weekend, although participants who engaged in certain screen use activities, including gaming, tended to do this more on the weekend days in the week of sleep restriction. Analysis of screen use revealed differences in estimates of screen use when using different rules to code screen use from wearable camera images, with assumptions around blocked images resulting in median estimates of screen use increasing by 56.4 minutes (26.5, 87.7), compared to no assumptions. No significant differences were found when analysing data at different intervals between images, including up to 60 seconds, when compared to 2 second intervals.
Conclusions: This thesis shows that even relatively small decreases in sleep duration of around 40 minutes per night can increase emotional eating behaviour in children, which may partly explain why children appear to eat more when they are tired. Digital screen use may play a role in the sleep-obesity relationship, with children appearing to spend more time engaged with screens after school when tired. This research highlighted the challenges of measuring screen use using wearable cameras, and future research should consider how to reduce researcher burden, while considering how analysis techniques can impact estimates of screen use in children. To fully understand whether sleep may be an effective strategy for improving obesity outcomes in children, further high-quality research to improve children’s sleep over the long term is needed, while considering the mediating roles of eating behaviour and screen use.