Background
Adolescents spend more than 60% of their daily waking hours sedentary [
1]. Studies have found that higher levels of sedentary time are associated with more depressive symptoms, a less favourable weight status and decreased physical fitness among adolescents [
2,
3]. Nevertheless, the evidence regarding the relation between sedentary time and health outcomes among adolescents is less consistent compared with adults [
4,
5]. Next to the total volume of sedentary time per day, more prolonged sitting may also have a negative impact on adolescents’ health. Some studies found that more prolonged sitting is associated with increased cardiometabolic risk factors, a less favourable weight status and decreased physical fitness among youth [
6,
7]; however, other studies found limited or no evidence for an association with health above and beyond the association of total sitting time [
8,
9]. Nevertheless, sedentary behaviour guidelines across the world recommend to regularly break up prolonged sitting in addition to reducing the total sitting time [
10].
Sleep behaviour has received increased attention in health promotion research as well. Although there are currently no international sleep guidelines, 14- to 17-year-old adolescents are recommended to sleep between 8 and 10 h per night and younger adolescents even between 9 and 11 h per night by paediatric sleep experts from the American National Sleep Foundation [
11]. However, the proportion of adolescents meeting this recommendation ranges from 32% to 86% on schooldays, and 79% to 92% on non-schooldays [
12]. In addition, these prevalence rates might even be an overestimation, as sleep onset latency (i.e., lying awake in bed before falling asleep) was not considered when calculating the sleep duration [
13]. Insufficient sleep is associated with negative health outcomes in adolescents on the short and long term, including depressive symptoms, poor perceived mental health, increased likelihood of substance use [
14,
15], and increased risk of developing cardiovascular disease, hypertension, type 2 diabetes and obesity [
16].
Previous studies have examined the interrelatedness between both behaviours and suggest a bidirectional, negative association between sedentary time and sleep behaviour [
17,
18]. These studies have applied and reported on a within-subjects analysis approach in order to reveal the acute effects of engaging more or less in a behaviour on a specific day or during a specific night [
19]. Currently, there are two studies that have been conducted in adolescents. A study among 15-year-old adolescents found that on days with more sedentary time, adolescents had later sleep onset, shorter sleep duration, but higher sleep maintenance efficiency (i.e. the minutes of actual sleep between sleep onset and offset divided by the sleep duration interval, expressed in percentages) [
20]. In the opposite direction, nights with longer sleep duration and later sleep offset, but also nights with a later sleep onset, predicted less sedentary time the day after. Another study in adolescents from the 6th and 8th grade also observed a bidirectional, negative association between sedentary time, screen-time and sleep duration, although the association varied by the specific sedentary and sleep health parameter used and by the time period of a day (e.g. sitting time during school hours vs. after school hours) [
21]. In both studies sedentary time was measured using hip-worn accelerometers. Although hip-worn accelerometers are reliable and valid tools to measure adolescents’ physical (in)activity [
22], they are unable to capture body posture [
23]. Using physically inactive time as an indicator for sedentary time (i.e. time spent in a sitting position) might lead to mistakenly including standing [
23,
24]. To more accurately assess sedentary time, it is highly recommended to use thigh-worn accelerometers [
25]. Finally, no accelerometer-based studies among adolescents have examined the bidirectional association between prolonged sedentary time and sleep, although it is important to see whether the relation with sleep duration is different (e.g. stronger).
The aim of this study was to investigate the bidirectional association between objectively measured sedentary time and prolonged sedentary time with sleep duration in 12- to 14-year-old adolescents from Flanders, Belgium, using both a within- and between-subjects analysis approach. As mentioned before, the within-subjects analysis reveals the immediate effects of engaging more or less in a certain health behaviour, whereas the between-subjects analysis generates general knowledge on behavioural differences between individuals [
19]. To unravel the association between the behaviours, it has been recommended to apply both approaches [
26]. Our hypothesis was to find bidirectional, negative associations between (prolonged) sedentary time and sleep for the between- and within-subjects analyses.
Discussion
The aim of this study was to investigate bidirectional associations between (prolonged) sedentary time and sleep duration in 12- to 14-year old adolescents from Flanders, Belgium. Within-subject results showed that adolescents who sit more than they normally do on a specific school day, sleep less the following night. In addition, adolescents who engage in more prolonged sitting than they normally do on a specific school day, also sleep less the following night, although the association was only borderline significant.
Although various mechanisms to explain the acute effect of physical activity on sleep have already been suggested [
35], there is not much evidence on the mechanisms explaining the acute effect of sitting on sleep. Therefore, we suggest some hypotheses below, but future research should verify them. A first possible explanation could be that specific factors mediate the association between more (prolonged) sitting and less sleep. For example, high levels of sedentary time and engaging in long sedentary bouts are related to negative emotions in adolescents, such as feeling unwell or stress [
36]. Experiencing negative emotions has in turn an impact on adolescents’ sleep onset, resulting in a short sleep duration [
37]. Another example is that more sitting and more prolonged sitting (i.e. less breaks in sitting time) on a given day are both associated with less energy expenditure [
38]. This could result in feeling less sleepy in the evening and not feeling the urge to go to bed [
20]. This might especially be detrimental for adolescents, as biologically they already have an evening circadian phase preference for later bed times [
39]. So in terms of sleeping sufficiently during the night, an important and useful strategy for adolescents might thus be to not only limit but also to break up the time spent sedentary the preceding day. Another explanation for the acute effect could be that the additional sedentary time of the adolescents on a specific day is spent in the form of screen-time activities during the evening period [
40]. This would imply that it is not the sitting behaviour itself that is important for sleeping sufficiently the following night, but rather the activity that is done while sitting. Indeed, screen-time activities are an important source of entertainment and might therefore push adolescents to postpone their bed time [
41]. This results in a shorter sleep duration, as rise time is mostly fixed on schooldays. Additionally, such screen-time activities can involve long bouts of sedentary time, such as binge watching, which explains the specific association between prolonged sitting and sleep duration. Extra analyses conducted as part of this study confirm that it is the additional (prolonged) sedentary time during non-school hours that is significantly related to a shorter sleep duration, and not the additional (prolonged) sedentary time during school hours (see Additional file
3). This is in line with the study of Kim and colleagues [
21] in which more sedentary time during the after-school period, and not during the school period, was associated with a shorter sleep duration the following night among young adolescents. This means that when promoting the reduction and breaking up of sedentary time during the day, there should be a focus on the non-school hours.
Unexpectedly, the association was not found in the other direction: sleep duration during the night was not significantly related to more sedentary time or to more time in long sedentary bouts on the following day. However, other studies found the association in both directions (e.g. [
17,
20]), suggesting our results do not completely align. A possible reason why sleep on a given night did not affect sedentary time the next day in our study, might be that we only collected data on school days. Previous research has attributed some of the within-variability among adolescents to the different sleeping patterns between weekdays and weekend days (i.e. catching up the loss of sleep from the school week during the weekend) [
42], suggesting that there might be less within-variability between the different weekdays. In addition, our study sample included relatively young adolescents, and adolescents’ age has shown to be a positive predictor of the within-variability in total sleep duration [
42,
43]. However, one of the studies that found an association between sleep duration on a given night and sedentary time the following day also only included school days in young adolescents, suggesting other factors might also be relevant to explain the difference between study findings. The difference in measurement method could play a role here, as previous studies have used hip-worn accelerometers to measure both sedentary time as well as sleep [
17,
20]. However, hip-worn accelerometers cannot accurately distinguish standing from sitting [
23,
24], suggesting our estimates of time spent sitting are more accurate than the estimates of other studies. For sleep, it is less clear if hip- or wrist-worn devices are preferred, although sleep has been validated mainly with wrist-worn devices [
44]. Hip-worn devices have been shown to overestimate sleep duration compared to wrist-worn devices [
45], however, others have stated that hip-worn devices are superior to assess sleep duration [
46].
In contrast to the within-subjects results, the between-subjects analyses revealed a bidirectional association. Adolescents who, on average, sleep less, have higher levels of sedentary time and spend more time in prolonged sitting. It was also found that adolescents who, on average, have higher levels of sedentary time or spend more time in prolonged sitting, have a shorter sleep duration. These two modifiable health behaviours are thus in general interrelated, which means future interventions focusing on reducing or regularly breaking up sedentary time could improve adolescents’ sleeping behaviour on a population level, and vice versa. This can be important for age-specific obesity prevention programmes, as sedentary time and sleep have been identified as two factors influencing its development [
2,
16]. This would imply that the health effect of population-based interventions focusing on one behaviour (sedentary time or sleep) might be reinforced because of a concurrent effect on the other behaviour. However, it must be acknowledged that effect sizes were relatively small.
A first strength of this study is that both within-subjects as between-subjects analyses were conducted in order to unravel the association between sleep and sedentary time in adolescents. Between-subjects analyses inform us about inter-individual differences or in other words, the general interrelatedness of sedentary time and sleep duration (i.e. on a trait level), whereas within-subjects analyses inform us about intra-individual differences or the immediate effects of sitting/sleeping more or less than usual (i.e. on a state level) [
19]. Another strength is the inclusion of spending time in sedentary bouts of at least 30 min as a proxy for prolonged sitting on a day. Previous studies have only investigated relationships between levels of sedentary time and sleep [
17,
20,
21], and it has been advocated to examine the associations between breaks in sedentary time and sleep as well [
26]. The current study adds to the literature by showing that not only the total time spent sedentary, but also how it is accumulated is important for adolescents’ sleep. A final strength is the use of device-based instruments to objectively measure sitting time, the time spent sitting in bouts of at least 30 min and sleep duration. Moreover, we used a thigh-worn accelerometer, which is more accurate to estimate people’s sedentary time compared to a hip-worn accelerometer [
25]. However, the disadvantage is that there is no information on the specific sedentary activities performed, suggesting future research should combine different measurement methods. Study limitations include the convenience sampling method and specific focus on secondary schools offering general education, limiting the representativeness of our findings. Another limitation is that we only had data on school days because of the nature of the larger school-based intervention study from which data were used. A final limitation is that we did not have day-to-day data on adolescents’ sleep quality. Although sleep duration and sleep quality are inherently associated with each other, it has been advocated that sleep quality is the most important health parameter [
47,
48]. Future studies could thus use ecological momentary assessment methods to assess adolescents’ sleep quality.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.