Abstract
Background: The pervasive use of digital devices, particularly among adolescents, has transformed bedtime routines — a point starkly highlighted by Netflix’s CEO in 2017: ‘We compete with sleep, and we’re winning.’ Existing research, largely cross-sectional, suggests a significant negative association with sleep but falls short of establishing causal inference due to design limitations. Most studies neglect key factors such as the specific timing of night-time screen use and night-to-night variability, which likely leads to inaccuracies and confounding. The need for more objective and reliable measures is evident, and the scarcity of longitudinal repeated-measures studies underscores the necessity for more robust study designs to better understand the size of the association.
Objectives: The primary aim of the Bedtime Electronic Device (BED) study and this thesis was to describe the influence of bedtime screen use on sleep among adolescents aged 11 to 14 years. The specific objectives were to develop a reliable coding protocol to objectively measure screen use from video camera footage that details device types and specific behaviours (Chapter 4), apply this protocol to describe the before-bed and bedtime screen habits of adolescents (Chapter 5) and evaluate the impact of screen time duration and activity type on sleep parameters within this age group (Chapter 6).
Methods: The BED study used a 7-night repeated-measures longitudinal design to investigate the impact of screen time on sleep in 83 adolescents aged 11 to 14 years in Dunedin, New Zealand. The data were collected from March to December 2021, and coding and analysis were performed from 2022 to 2023. Participants wore PatrolEyes DV7 video cameras on four non-consecutive nights that recorded data from two hours before bedtime until bedtime. Stationary video cameras in the participants’ bedrooms captured screen time between bedtime and before shuteye time. Sleep outcomes were measured using AX3 accelerometers over one week. A coding protocol described in Chapter 4 was developed and assessed for inter-rater reliability. Four coders independently coded 600 minutes of footage and classified device types (eight categories) and screen activities (nine categories) using a separate sample of participants (n = 14, 11 to 14 years). Inter-rater reliability was assessed with weighted Cohen’s kappa (κ). Descriptive screen time and sleep data, including means and standard deviation, are presented in Chapter 5. Nightly variations in screen use and sleep were analysed at the between- and within-person level with standard deviations and coefficients in Chapter 6. A mixed-effects regression model, which considered participants as random effects, explored the screen use and sleep relationship, which adjusted for weekend effects.
Results: Overall reliability of the full coding protocol was excellent (κ = ≥0.8). Coder agreement ranged from 92% to 99% across 27 to 107 different instances of screen use in the dataset (Chapter 4). In Chapter 5, descriptive data from 83 participants (mean age 12.3 years) captured through video camera footage from 344 nights revealed that 99% of participants engaged in screen use within the two hours before bedtime time-frame for a mean duration of 62 minutes. During the time between getting into bed and trying to go to sleep, 53% of participants used screens for an average of 45 minutes, and approximately 30% continued for 18 minutes after their initial attempt at sleep (shuteye time). The primary screen activities included watching, gaming and multitasking on devices such as phones, laptops and televisions. In Chapter 6, screen use in the two hours before bed was significantly associated with an average delay of 31 minutes in sleep onset and 27 minutes in sleep offset compared to screen-free evenings. No significant associations were found with other before bedtime sleep outcomes. Larger effect sizes were observed for screen use between bedtime and shuteye time: for every additional 10 minutes of screen use, there was an average increase of 40 minutes in sleep latency and a 35-minute delay in sleep onset. Passive screen use was associated with a significantly earlier bedtime by an average of 21 minutes. Interactive screen use and social media significantly increased sleep latency by an average of 41 and 54 minutes, respectively. Multitasking across devices led to a significant average decrease in total sleep time by 35 minutes compared to nights without such activity.
Conclusion: This study provides a more refined and objective view of how various types of screen usage influence the sleep of young adolescents, which contests the prevalent view that screen time before bedtime is invariably detrimental to sleep. There is a compelling case for more nuanced sleep hygiene guidelines to emphasise the reduction of specific screen time during the time when adolescents are in bed — especially interactive and multitasking screen activities. A practical amendment to current sleep hygiene guidelines might be to focus on use types during bedtime and avoiding interactive activities in this period; thus, this offers health practitioners, parents and adolescents more pragmatic and achievable strategies to improve sleep hygiene. Additional research is warranted to determine the optimal cessation time for screen use to better support adolescents during their bedtime routine.