Children’s sedentary behaviour: descriptive epidemiology and associations with objectively-measured sedentary time

Along with physical activity and nutrition, sedentary behaviour is thought to be a risk factor for poor health in both children and adults. A high level of sedentary behaviour has been associated with obesity [1] and is positively associated with insulin resistance and metabolic risk in children [2, 3]. Television (TV) viewing, in particular, has been associated with unfavourable body composition, reduced fitness, lower self-esteem and poorer academic performance [1]. Associations between health outcomes and other sedentary behaviours are however less clear [4]. Moreover, to what extent these associations are independent of physical activity remains uncertain, as a negative, albeit weak, association has previously been observed between physical activity and sedentary behaviour in children [5]. Nonetheless, it is recognised that children spend a significant proportion of their time being sedentary [6] and that time spent sedentary should be limited [7].

Sedentary behaviour is difficult to assess accurately due to its complex nature, with a variety of behaviours occurring at different times of day and in multiple locations [8, 9]. School-based behaviours are usually pre-determined with little flexibility, probably decreasing the variability in sedentary behaviour. A focus on leisure-time sedentary behaviour therefore appears more useful to inform health promotion. However, most sedentary measures focus on highly visible and prevalent behaviours, such as TV viewing and computer use [8, 10]. These screen-based behaviours may fail to fully capture the complexity and diversity of children’s leisure-time behaviour patterns [11, 12]. Project STIL (Sedentary Teenagers and Inactive Lifestyles) is one of the few studies that assessed a broad range of leisure-time sedentary behaviours in young people (13–16 years) [13, 14]. TV viewing, homework and motorised transport were amongst the five most common sedentary behaviours for both boys and girls. Other common behaviours among girls were sitting and talking, playing musical instruments and looking after pets [14], whereas playing computer/videogames and shopping/hanging out were common amongst boys [13].

Accelerometers are widely used in physical activity research and their applicability to the study of sedentariness is increasingly being explored [8]. Using a 100 counts per minute cut-point, waist worn accelerometry has demonstrated good agreement with activPAL for the assessment of sitting [15] and good to excellent classification accuracy for selected sedentary behaviours in children under controlled conditions [16]. In order to develop effective interventions aimed at reducing overall time children spend in sedentary behaviour, it is important to understand the types of behaviours children engage in and whether any of these are representative of overall time spent sedentary. This is of particular interest to TV viewing, which has been suggested as a key sedentary behaviour [17] and is commonly used as the sole marker of sedentary behaviour. Previous research indicates that TV viewing may be a reasonable marker of overall sedentary time in adult women [18], but not adolescents [11] or children [19].

The aims of this study therefore are 1) to describe the leisure-time sedentary behaviour of 9–10 year old British children, and 2) to identify markers of total sedentary time by studying associations between time spent in 12 sedentary behaviours and objectively-measured leisure-time spent sedentary.

A total of 1513 participants (73.3% of the study sample) were included in the analyses after excluding participants with incomplete or invalid accelerometer and YPAQ data. Those excluded from the analyses did not differ from those included on age, sex or BMI z-score, however they were more likely to be of lower SES (p = 0.03).

Weekly objectively-measured leisure-time sedentary time differed by sex (see Table 1) and SES (low (mean minutes ± SD): 2039.2 ± 319.84; middle: 2078.6 ± 302.5; high: 2125.8 ± 328.9; p-value for trend: <0.001).

British children aged 9–10 years engage in a variety of sedentary behaviours, with participation varying by sex and SES. Both sexes reported spending a substantial percentage of their leisure-time engaged in screen-based behaviours, with boys (57.0%) more so than girls (44.7%). Of the non-screen behaviours, most time was spent reading, followed by listening to music and sitting talking. A complex pattern of associations with objectively-measured sedentary time emerged, including both screen and non-screen based behaviours and negative as well as positive associations. However, the magnitude of these associations and the amount of variance in the outcome explained by these variables was too small to conclude that any single behaviour can be considered a meaningful marker of overall sedentary time.

Few studies have been undertaken describing the sedentary behaviours of children that included assessment of multiple screen and non-screen-based behaviours. We found that boys spent more time watching TV/videos, playing indoors with toys, playing board games/cards and playing videogames than girls. In contrast, girls generally spent more time in most non-screen behaviours. Consistent with our findings, other studies found that boys watch significantly more TV and played more videogames [13, 14, 28, 29] than girls and that girls spent more time talking on the phone than boys [13, 14]. A study in Chinese children also reported that primary school-aged girls spent more time than boys in activities such as extracurricular reading, writing and drawing [30]. Our observation that boys spent more time in screen-based and total sedentary behaviour than girls, but that girls spent more time in non-screen sedentary behaviours is consistent with findings in 9–16 year old Australians [29]. However, studies exploring sex differences in a range of sedentary behaviours, including non-screen activities, in children of this age are lacking. With specific information on what sedentary behaviours boys and girls do, it may be possible to target interventions more effectively.

We also studied differences by SES, which we defined using a composite score of parental education, car and house ownership. Whereas clear sex differences were observed for most behaviours, only some of the non-screen behaviours differed by SES. Children from a high SES background tended to report more time on non-screen sedentary behaviours although the overall reported time spent sedentary did not differ substantially between children from low and high SES backgrounds. Previous research has also reported higher engagement in non-screen sedentary behaviours among those from high SES backgrounds (as defined by parental education and household income), but also that those from low SES backgrounds spent more time in screen-based sedentary behaviour [29]. Using an area-based measure of deprivation, a recent study showed a similar pattern, but also that those from areas of lower deprivation (e.g. higher SES) tended to accumulate more sedentary time overall [28]. We observed similar patterns with the objectively-measured data. Research into SES differences in physical activity levels have also shown differences according to the type of SES measure used and the type of physical activity assessed [31], which may also partially explain the disparities in findings. Further research into the association between SES and sedentary behaviour is needed to disentangle what components of SES are important for children’s sedentary behaviour.

The common perception of TV viewing as an indicator of overall time spent sedentary was partially supported by this study, adding to a literature with mixed findings on this subject [11, 19, 32]. A significant association between TV viewing and accelerometer-assessed sedentary time was also observed in a recent European study [19]. However, similar to the current study, the authors concluded that the relationship was weak, as indicated by the magnitude of the association and the limited amount of variance explained. We added to the current evidence by investigating whether sedentary behaviours other than TV viewing might serve as a marker for overall sedentary time. When considering the behaviours as continuous variables, in addition to TV viewing, reporting more reading and less time playing board games or cards were associated with higher objectively-measured sedentary time. Given the skewed nature of the data and the observation that questionnaires are generally more suitable for ranking individuals [24], we also considered the behaviours as categorical variables. Here, a mix of positive and negative associations was again observed, with playing videogames emerging as the behaviour most strongly associated with objectively-measured sedentary time, particularly for boys. In both the continuous and categorical analyses, little additional variance in the outcome was explained over and above that of the confounders, even when multiple behaviours were retained in the final statistical model. Taken together, these results suggest that overall time spent sedentary, as measured by accelerometry, is unlikely to be captured accurately by focusing on a single behaviour.

To our knowledge, this is the first study to provide a descriptive epidemiology of a variety of sedentary behaviours in 9–10 year old children. It utilized both an objective and subjective measure of sedentary behaviour, assessed a wide range of screen and non-screen behaviours and considered differences by sex and SES. The large sample size and heterogeneity in location are key strengths of the SPEEDY study. Several limitations are acknowledged. Accelerometer and YPAQ data did not refer to the same week, with the YPAQ data referring to the week before the accelerometer was worn. Although this might partially explain the lack of associations observed, previous research has shown that children’s overall sedentary time is relatively stable day to day [33]. Although the sedentary items of the YPAQ questionnaire were deemed reliable, no data on the validity is available. However, the aim here was not to obtain a valid estimate of sedentary time per se, but to obtain information on the types of behaviours children engage in and to rank them in terms of participation; in general, questionnaires are an appropriate methodology to achieve this [24]. The SPEEDY data were collected in 2007 and we acknowledge that the media landscape has changed significantly since then, likely affecting the descriptive epidemiology provided. However, it is unlikely to affect the association with objectively-measured sedentary time. Lastly, differential drop-out was observed with children from low socioeconomic backgrounds less likely to provide valid data. The SPEEDY sample has also previously been shown to under represent obese children relative to the broader Norfolk population of this age, limiting the representativeness of the results shown.

Children engage in a wide variety of sedentary behaviours and participation varies by sex and SES. Interventions that target a single behaviour may be too narrow in focus to impact upon overall sedentary time. Moreover, careful consideration should be given to the selection of intervention targets because the mechanisms linking sedentary behaviour with health outcomes remain unclear [34, 35] and not all sedentary behaviours are associated with the same health risks [4]. This study adds to this debate by highlighting that participation in sedentary behaviours that may be considered more or less beneficial, socially or developmentally, varies across the population and individual behaviours may be positively or negatively associated with overall sedentary time. Future studies, both observational and experimental, should consider a wide range of sedentary behaviours and incorporate objective measures to quantify sedentary time where possible.