The study investigated longitudinal associations between SB and 1) dietary behaviors and 2) leisure-time PA over a period of 20 months. Findings indicated that changes in TV/DVD use and computer/game use during this period were associated with unhealthy changes in dietary behaviors; BL TV/DVD use and computer/game use were also independently associated with undesirable changes in dietary behaviors. The association between TV/DVD use and leisure-time PA was inverse, but not substantive. No significant association was found between change in computer/game use and change in leisure-time PA.
The associations between SB and dietary behaviors found in this study are in line with those of a recent systematic review of mainly cross-sectional studies [13
], including a European cross-national cross-sectional study with representative samples from different countries including Norway which found TV viewing to be positively associated with the consumption of sweets and soft drinks and inversely related to the consumption of vegetables [33
]. The few longitudinal studies included in the review also indicated an inverse relationship between SB and the intake vegetables [18
]; and a positive association between TV viewing and the consumption of food items such as snacks and sugar sweetened beverages [35
], which is in line with the findings of this study. This study adds to the existing literature by extending these findings to the use of computer/games. Inverse associations between TV use and consumption of fruits were also documented in the aforementioned studies. However, in the present study, this association was non-significant, which might be related, among other things, to the fact that the change in fruit consumption over 20 months was very small, making it difficult to find associations; baseline TV/DVD use was however inversely related to change in the consumption of fruits.
Although the associations between SB and dietary behaviors documented in this study are small, they have likely been attenuated by measurement errors; and a consistent pattern of unfavorable associations with several dietary behaviors investigated was found. In addition, SB at age 11 was also independently associated with the change in the dietary behaviors investigated over 20 months. The specific mechanisms behind the association between SB and dietary behaviors cannot be identified from this study, and only hypotheses can be put forward. One possibility is the exposure to advertisements which influences children’s consumption patterns [8
], including through a higher responsiveness to food advertising among children with higher media exposure [8
]. There is a ban on advertisements aimed at children during children’s TV programs in Norway. However, children’s programs broadcasted from outside of Norway do contain advertisements; the children might also watch adult-targeted TV programs containing food adverts. The children might also be exposed to advertisements on the internet. The other possibility is that parents provide some types of foods that they themselves associate with screen time to their children, starting from a young age. According to the habit theory, once habits are formed, automatic triggering of subsequent behavior occurs upon exposure to environmental cues which normally precede the action [36
]. Hence, if screen-based SB have been repeatedly accompanied by consumption of unhealthy food items, these SB might become automatic cues to such dietary habits [36
]. In a study among adolescents in the Netherlands, Kremers et al. found that more time spent on TV viewing is associated with a stronger habit for soft drink intake, and found that a quarter of the variance in habit strength of soft drink consumption could be explained by habit strength of TV use [37
]. de Bruijn and van den Putte similarly found that a strong habit towards TV viewing was positively related with soft drink consumption [38
]. Provision of healthy foods during screen time by parents might help disrupt such automatic activation of unhealthy eating habits. The current findings imply that a decrease in SB might result in favorable changes in dietary behaviors, although these changes might be small.
The study lends support to the findings of several studies, mainly cross-sectional, pointing to a lack of significant association between SB and leisure-time PA [21
]. The findings are however not supportive of those of other longitudinal and cross-sectional studies indicating inverse associations between PA and SB [14
]. The weak longitudinal association between TV/DVD use and leisure-time PA might however indicate a possibility for some displacement of PA by TV/DVD use, although this association is unlikely to be clinically relevant. Measurement errors have however likely attenuated this association. Previous studies have similarly found TV viewing to be more strongly or more consistently inversely associated with PA compared to other SB such as game use and computer use [16
]. As stated by Marshal et al., the total amount of time per day engaged in different SB is inevitably prohibitive of PA [2
]; however it is possible that not all types of PA are related to SB. In a moderator analysis in their review paper, Marshal et al. found that only vigorous PA was inversely related to TV viewing. This finding could be real, but it could also be related to the fact that vigorous PA is more easily recalled than moderate PA, making it easier to find associations [2
It has been suggested that the link between SB and overweight/obesity is more likely to be due to associations with dietary behaviors than to associations with PA [13
]. The results of this study support this suggestion.
Limitations and strengths of the study
An important methodological issue to be taken into consideration when interpreting the results of this study is the regression dilution bias, which is a statistical phenomenon whereby random measurement errors in independent variables attenuate regression coefficients [39
]. As random measurement errors in the independent variables used in this study, i.e. TV/DVD use and computer/game use exist, the weak associations found are most probably higher than documented. Such measurement errors can also shift results to the null [39
The changes in dietary behaviors and leisure-time PA over 20 months documented in this study were rather small, which might make it difficult to find significant associations [32
]. Information about amount of consumption was not available except for soft drink with sugar consumption, therefore only frequency measures were used, and adjustment for total energy intake was not possible. Finally, although temporal relationships could be assessed in this study, its observational nature does not allow for causal inferences to be made. A randomized controlled trial that alters the SB would make it possible to get a strong evidence of causality. In addition, although there is no reason to believe that a change in dietary behaviors would lead to a change in SB, it is possible that a change in PA leads to a change in SB and not vice versa. Finally, the study is restricted to a single geographical area, making generalizability limited.
The study has several strengths. Most evidence about the associations investigated in this study has to date been derived from cross-sectional studies. However, demonstrating that a change in an independent variable predicts a change in a dependent variable provides a stronger evidence of causality compared to using an independent variable measured at one point in time [40
]. Adjustment for characteristics of children empirically and theoretically associated with the behaviors investigated was another strength of this study. SB other than TV use, as well as several dietary behaviors were also included. The sample size was relatively large, and the rate of retention in the study was high.