Sedentary time among primary school children in south-west Germany: amounts and correlates

ST was assessed by a multi-sensor device (Actiheart®; CamNtech, Cambridge, UK), which was validated for assessing PA in children [36]. The sensor was worn for 24 h a day for six consecutive days. It was fitted to the child’s chest at school by trained staff. Parents were instructed how to re-fit the device in case of detaching. Recordings had to be available for at least 10 h per day including at least 1 day of the weekend and two weekdays [37]. First and last day of the recording were excluded from the analysis. Heart rate in beats per minute (bpm) and one-dimensional bodily acceleration in counts per minute (cpm) were recorded using 15 s epochs. Energy expenditure was calculated using the branched model approach [36] with Actiheart®‘s captive software [38], expressed in metabolic equivalents (MET = (REE + AEE)/REE). Individual resting metabolic rate was calculated on the basis of the Schofield equation [39]. The standard definition of energy expenditure ≤1.5 MET was used as a threshold to identify total ST [25]. Additionally, PA levels were classified conventionally into LPA >1.5 to <3.0 MET and MVPA ≥3.0 MET [40]. Individual daily sleeping time was subtracted from daily assessed recording time to quantify waking time being sedentary (= total ST). Therefore, individual sleeping time was identified for every analysed day by an obvious increase and drop of heart rate for the point of awakening and falling asleep, respectively. Two experts independently set time point of awakening and falling asleep when recordings were complete and clear. In case of disagreement, a third expert was involved. Subsequently, total ST per day was calculated as: mean total ST = [(mean ST weekday × 5) + (mean ST weekend day × 2)]/7.

Anthropometric data were assessed at school by trained staff using standardized procedures [41]. Height was measured to the nearest 0.1 cm using a stadiometer and weight to the nearest 0.05 kg (Seca 213 and Seca 862, respectively, Seca Weighing and Measuring Systems Hamburg, Germany). BMI was calculated (kg/m²) and classified into percentiles according to the German definition by Kromeyer-Hauschild et al. [42]. Calculated BMI percentiles (BMIPCT) were categorized into underweight (≤10 percentiles), normal weight (>10 to ≤90 percentiles), overweight (>90 to ≤97 percentiles) and obese (>97 percentiles) for children’s weight status. Child related (gender, age, migration background) and family related factors (parental education and household income) were assessed by parental questionnaire. The level of parental education was assessed by the highest educational level of parents or of a single parent based on CASMIN levels (primary, secondary and tertiary) [43]. Household income was assessed according to the seven categories of the Winkler-Index [44]. Migration background was defined as 1) having at least one parent who was born abroad or 2) a parent speaking to their child in a foreign language during the first 3 years of the child’s life. Group differences were investigated separately for gender, day, migration background, weight status, and activity level. According to the definition, children not reaching 60 min of MVPA daily were classified as inactive [25].

In contrast to the existing assumption that only a weak association of ST and PA exists [28], higher amounts of ST among inactive children (MVPA < 60 min/day) as well as relatively high negative correlations of ST with MVPA and LPA were found in this study. Stronger associations may result from objective assessment in comparison to studies collecting subjective data [28]. Studies also used different thresholds and methods to measure PA, which possibly also caused the variety of previous findings. Moreover, the very low age range may lead to a stronger correlation. Even though, similar to our results, the large cross-sectional STOPP study (n = 1538) reported a very high negative between-subjects correlation (r = −0.837) of objectively assessed ST and MVPA among 6 to 10 year old Swedish. Wrist-worn accelerometers (Actiwatch®) were used to assess children’s activity and energy expenditure was calculated in MET. The same definition for sedentary behaviour of energy expenditure <1.5 MET was used, while 332 cpm were set as cut-off point for calculating ST [49]. Also supporting this relationship, a longitudinal study with 9 and 12 year old children found that a higher decline in MPVA (β: −1.66) was significantly associated with a higher increase of ST, while attendance at sport clubs was significantly associated with a smaller increase of ST (β: −2.04). Accelerometers (ActiGraph® GT1M) and a cut-off point of <100 cpm was used to assess ST of 365 children in northeast England [29]. Even though this association still needs to be further investigated, the current findings indicate that more MVPA might have a reducing effect on total ST in children. As children’s activity behaviour is modifiable, children’s ST should be reduced while MVPA should be increased to at least 60 min a day or more, as recommended [33]. Though, in order to achieve this common goal, parental support is necessary. It has been shown that if parents are active in sports clubs, there is a greater likelihood that their children are also physically active in sports clubs [50]. In addition, a review concluded out of four studies (with objective and subjective assessment), that among 4 to 12 year old children a positive connection of boys’ overall PA with parental PA exists [51], pointing at the importance of parents being active role models. Therefore, interventions promoting more PA and less ST should not only target children but also their parents.

Further, this research shows that children do not seem to compensate ST of school lessons at the weekend. Rather, they are even more sedentary at the weekend, which was also found among 9 to 10 year old children of the UK [19]. Supporting this, relatively high ST of 41% to 51% in available spare time, i.e. overall after-school sedentary time measured objectively, among 5 to 18 year old’s was reported [15]. Children and adolescents spent most of their spare time with watching television (20.4%) and with non-screen based sedentary behaviour (57%) including social sedentary behaviour, motorized transport, homework and reading [15]. Preferred leisure time activities in German children are similar, more or less sedentary activities, such as watching television, playing computer games or meeting friends [52]. In contrast to our findings, Nilsson et al. [20] reported less ST at weekends and in leisure time compared to school time using accelerometers to assess ST (<100 cpm) in 9 and 15 year old Europeans, but no German children were included. This again indicates that findings can vary for countries and samples, as explained previously. Besides preferences of activities, there are many other factors which may influence after-school ST such as location or being alone, e.g. in children being in after-school care almost 10% less ST was shown [15]. Additionally, previous studies also reported higher amounts of MVPA at weekdays [20, 32] as well as reaching the PA guideline more often at weekdays [32] than at weekends. As children usually spend (much leisure) time at weekends with their parents or family, this again highlights the need of parents to engage (with) their children in more PA especially during leisure time. This seems important because, as mentioned before, parental PA has been shown to be a key factor to increase children’s PA [50].

Furthermore, in this sample household income was found to be related to ST, with a tendency towards less sedentary children in families with a higher household income. One reason for less ST among these children might be that their parents are more prepared to pay for (expensive) sports equipment or sports club fees. However, the correlation was no longer evident after separating analysis by gender, indicating a weak influence. Contrary, one previous study of Pulsford et al. [53] with a large and representative sample of school children in the UK (n = 629), aged 10 to 11 years, found no association between ST and household income, where ST also was objectively measured and defined as less than 100 cpm. However, not only household income but also parental education might be associated with children’s ST, which was also investigated in this study. Pulsford and colleagues [53] found a weak inverse association of parental education with objectively assessed ST before school time, while boys spent 11.82 min more time being sedentary in and after school if their parents came from a higher educational background. Another study reported that parental education level predicted self-reported screen time among adolescent girls (mean age 12.8 years) [54]. However, no correlation of parental education level with ST was found in this study with a slightly younger sample. Because income and education are often investigated jointly in socio-economic status [27, 31], evidence on these two single factors seems insufficient and partly inconsistent [31], wherefore no conclusion can be drawn yet. Although this study did not confirm this, in the current literature there is a tendency towards a positive association of school related ST and a negative association of screen-based ST with higher parental education [53], while income might be weakly negatively correlated to ST. Thus, for a clear statement further research is necessary.

One of the non-modifiable determinants which are often linked to ST is migration background, which was also investigated in this study. In this sample, the proportion of children with migration background is similar to a representative national sample, although more girls than boys had a migration background [55]. Here, almost 20 additional minutes of ST could be explained by migration background. This was also confirmed by a recent study among 1943 preschool children in the Netherlands, which found that sedentary behaviour was significantly higher among 8 to 9 year old children having a migration background than in the Dutch population [56]. Moreover, previous research shows that among children with migration background, lower PA levels were found, while they had fewer memberships in sport clubs and less accessibility to active toys [55, 56], pointing at less overall PA among children with migration background. Also, since screen time is a part of ST, it should also be mentioned that in Germany, the highest percentage of television or computer consumption of at least 3 h daily was reported among children with migration background [55].

In this sample, the prevalence of overweight and obese children is approximately 6 points under the national sample of 7 to 10 year old German children (15%) [34]. In contrast to previous investigations on the association of weight status with ST [1‐5, 31], in this study, overweight was not associated with ST. Besides the lower prevalence, this might result from the fact that activity energy expenditure of overweight and obese children is possibly overestimated if expressed in MET [57]. Because these children have to carry their fat mass as additional weight, their performance is less than in normal weight children at the same work load. Paradoxically, more weight results in higher energy expenditure and therefore higher assessed activity. Consequently, sedentary behaviour is misclassified into higher energy levels. Additionally, higher resting metabolic rates in younger children and in children with normal BMI percentiles (<85 percentile) were reported, also leading to higher MET for the same activity [46]. But to get suitable individual energy expenditure the calculation of resting metabolism is very important. Energy expenditure expressed in MET is more adequate in children if individual resting metabolic rates were used [47]. However, differences in resting metabolism can also cause variance of PA levels and ST, which is why different calculation of energy expenditure should be investigated in further studies with children.

Moreover, confirming previous findings [4, 27, 29, 31, 48], in this sample higher ST was found among girls indicating that they prefer certain sedentary activities in comparison to boys; Reading books, doing crafts or painting as well as listening to music or using the internet are more popular spare time activities among German girls, while German boys more often prefer sports or playing outside [52]. Further, also the association of ST with MVPA in this study potentially explains a part of the gender difference in sedentary behaviour. Previous studies have shown, that boys engage in more PA during the entire day [32, 48], as well as during different segments of the day e.g. during school recess or PE lessons [58‐60]. Additionally, boys reach the WHO activity guideline more often than girls [32].

To summarize, ST differed not only between gender, but was also shown to be higher among inactive children (50% of this sample) and children with migration background (25% of this sample), while ST in children of parents with high household income was by tendency lower. Since ST is linked to different health-related factors these findings indicate that those children are more prone to sustain potential health consequences in their later lives [12‐14]. Therefore, they should especially be considered in interventions, but further research considering combined risk groups is necessary. Additionally, interventions might be (more) effective if targeting children’s leisure time activity, especially at weekends and encouraging parents to support their children’s activity.

This is one of the few – if not the only – studies investigating objectively assessed total ST in German primary school children during awake time. Using individual resting metabolic rates for the calculation of children’s energy expenditure should be considered as a significant strength of this study in addition to the objectively measured ST and the individual identification of sleeping hours. However, the results need to be interpreted with caution since there are also some limitations to this research. Children’s amounts of ST could possibly be misinterpreted due to the analysis of ST on the basis of a branched model approach and expressing energy expenditure in MET (especially in overweight and obese children), as previously explained in detail. Further, there was no possibility to additionally assess children’s activities, e.g. by diary during the day. Therefore, a more detailed conclusion on children’s activities during ST is not possible which could be useful to explain the here found relatively low amounts of ST. Moreover, as children’s bouts of MVPA were reported to last between 3 and 6 s [61], the shortest recording epoch available was chosen. Longer epochs might miss even more movements and further overestimate children’s sedentary time. Thus, it is possible that not every activity is assessed comprehensively enough. Furthermore, study participation was voluntary for schools, parents and children, which might have led to a selection bias. However, this study has investigated ST objectively for a relatively large sample size which should be considered as a very meaningful strength besides the previous mentioned strengths.