Our data showed that adolescent girls who participated in cultural activities had less central adiposity eleven years later compared to girls who did not participate in such activities. However, participation in social leisure time activities predicted increased central adiposity in both girls and boys. In adolescent boys particularly, participation in social leisure time activities predicted higher overall adiposity at follow-up. Additionally, highly culturally active adolescents seemed to be better protected against the effect of obesity-susceptibility genes when measured in young adulthood.
Strengths of the present study are a general population sample and the 11 year follow-up of adolescents into adulthood (Additional file 1
: Tables S2 and S3). The great mobility in this age group and consequently loss of follow-up participants was a challenge. However, comparison of baseline data between participants and non-participants in the follow-up showed no essential differences, suggesting no systematic selection bias (data not shown). Socio-economy might have an important influence on participation in cultural and social activities, though adjusting for the socio-economic status in adulthood did not change the results.
Regression to the mean (RTM) in the two subsequent measurements may occur. In this way, the RTM-effect may have decreased the observed association. Although, a certain possible measurement error was likely to behave in the opposite direction, the adiposity measures were very homogenous in the study sample and were not expected to affect the associations substantially.
We repeated the analyses including only the 1123 adolescents with normal weight at baseline in the analyses, because the effect of environmental influences are suggested to be larger in individuals with phenotypic extremes during adolescence [37
] and might bias our results. Additionally, overweight in adolescents tends to track into adulthood [42
] and it is more likely that environmental effects might be easier detectable in normal weight adolescents without onset of obesity. The results might indicate that the favourable effect of cultural participation in adolescence is more preventive against overweight and less a remedy for becoming normal weight again.
A possible limitation might be that we did not include physical activity as a component in the index of social activity. We were mostly interested in the “social” part of the social activities, although we cannot exclude some overlap with physical activity in our definition of social activities. Therefore, we adjusted for the level of physical activity.
The effect of cultural participation
There might be several explanations for the inverse effect of participation in cultural activities on adiposity measures. Firstly, participation in cultural activities is found to be associated with an overall regular lifestyle and healthy behaviour [43
]. Secondly, relaxing leisure time activities like listening to music, reading and TV watching seem to be strong predictors of coping with stress, as suggested by Iwasaki et al. [44
] and thereby counteract stress-induced central obesity. Thirdly, Lajunen et al. [16
] argues that adolescents cannot eat while playing music, keeping the energy input and output in balance.
Girls were obviously more engaged in cultural activities compared to boys, and it seemed that only in girls the participation in cultural activities was protective against overweight. This gender difference may be due to factors like different food choices, body satisfaction, and physical activity , as well as differences in participation in cultural activities [45
Our findings might possibly be related to some biological pathways, demonstrated in animal experiments. Rodents who were exposed to enriched environments where they were offered exercises both mentally (i.e. toys, change in food locations, more complex housing) and physically tended to switch from a white fat phenotype to a brown fat phenotype, more likely to burn fat depots [46
]. Furthermore, diet induced obesity has been shown to be prevented through the genetic and environmental activation of hypothalamic-adipocyte axis [8
]. These effects were specifically noticed in rodents exposed to an enriched environment compared to rodents exposed only to physical exercise, like running. Histological studies show that these mechanisms may also be relevant in humans [47
Cultural activities, as defined in this study, were sitting activities. There is no unanimous conclusion on the negative effects of the different sedentary activities per se on development of overweight [43
]. Although there is evidence that TV viewing is a risk factor for development of obesity, being more important in children than adolescents [48
], the pathways of the association are unclear [39
]. We have, therefore, repeated the analyses without the component “Watching TV or a video” as a part of the index of cultural activities (Additional file 1
: Table S6). The exclusion of “Watching TV or a video” from the cultural activity index did not change the results substantially. Specifically, excluding the component TV viewing from the index of cultural activities weakened the effect estimates slightly, but the confidence intervals remained nearly unchanged. “Watching TV or a video” was individually also inversely associated with central body fat estimates in girls in discordance with other studies [49
]. TV viewing may in our population counteract stress in concordance to Iwasaki et al. [44
]. Another explanation may be that TV viewing seemed not to displace physical activity in our population. Biddle et al. has stated that TV viewing may be reflective of other unhealthy behaviours [51
], but this does not seem to be the case for the girls in our study. Also, the relationship between TV viewing and obesity has been suggested to be influenced by other confounders like snacking, psycho-pathologies [39
], socio-economic status [51
] and family structure [48
] which may not be applicable to the girls in our population. Finally, Norwegian adolescents seemed to consume less TV during dinner than peers in other European countries [49
The effect of social activities
Consistent with the findings of Christakis et al. [10
], but in discordance with the study of Lajunen et al. [16
], our data suggest a positive association between adolescent participation in social
activities and overweight in young adulthood. The social activities as used in the study of Lajunen comprised not solely social activities. Like Barabasi et al. [52
], we might suggest that the social network may be a factor in the spread of obesity through social ties. Some studies [6
] proposed psychosocial pathways as one possible mechanism in the beneficial effects of social participation on psychological health in adults. Others [4
], on the contrary, found that the relative risk of mortality increased equal to the extent of the social network. Likewise, our data indicate negative influence on health measures, i.e. adiposity measures. One can imagine that in adolescents, a psychological susceptible period with diverse psychosocial influences and with peers as important reference persons, it might be difficult to sustain a good self-concept. Adolescents without good self-concept might change easily their lifestyle to the unhealthier one of their friends.
Modification of obesity-susceptibility genetic markers
There is growing evidence that the 32 loci robustly associated in large-scale meta-analyses of GWAS with BMI in adults [25
], also affect BMI in childhood and adolescence [31
]. Heritability of BMI is estimated to vary from 0.58 among 11
year-old girls and boys to 0.83 and 0.74 respectively in 17
year-old boys and girls. Also, the genetic contribution to BMI seems to be strong during adolescence [56
]. Though, the known combined genetic effect on BMI is modest, estimated to only 2–3
% of the genetic variation in BMI.
In concordance to other studies [57
] we preferred to use a genetic predisposition score rather than a single locus because a genetic predisposition score explains also a greater genetic variance. This approach may be preferable when demonstrating an interaction between genetic susceptibility and a life style factor. Additionally, multiple allele scores increase statistical power and may represent a more generalizable influence of genetic obesity susceptibility compared with analyses of individual variants.
For estimating the cumulative effect of the SNPs combined we calculated a genetic predisposition score (GPS) by summing the BMI-increasing alleles across the 12 SNPs [35
]. We did not weight the risk alleles on basis of their individual effect sizes because no well-accepted effect sizes were available for each of the SNPs. It is previously demonstrated that weighting of risk alleles may have limited effect [59
Lajunen et al. [56
] also indicated that common environmental factors had an effect in early adolescence. According to Haberstick et al. [60
] the role of environment compared to genetic influence may be more profound in girls than in boys.
Our data revealed a marked divergence between girls and boys in the distribution of body fat. Participation in social activities was in girls more related to central adiposity and in boys to overall adiposity. This is in congruence with other studies [5
]. The divergence between girls and boys may be due to biology which is expressed in differences in distribution, storage and metabolic processing of body fat. Physical activity in contrary to cultural activities seems not to attenuate the effect of obesity-susceptibility genetic variants on adiposity estimates in adolescents [61
Eventually, environmental influences like psychosocial stress may override genetically driven factors [62
] and increase particularly the central adipose measures in girls, who are more susceptible for stress-related pathology [63
]. Another reason could be that the effects of genetic influences could be sex-limited due to different neuro-endrocrine functioning and fat metabolism [64
]. As a consequence, an important question is whether this difference in fat distribution might be an indication of greater health risk in girls [64
Further, from previous studies we know that the MC4R
has a role in the food intake, thermogenesis and loco motor activity [65
]. The well-known FTO
has possible a role in the HPA-axis and influences body composition through energy expenditure [31
]. Thus, a pathway that affects body composition through the central or distal nervous system might be possible. This pathway we have hypothesized also to be important for the effect of leisure time activity.