Screen-based media use clusters are related to other activity behaviours and health indicators in adolescents

The health benefits of being physically active in childhood and adolescence are well established [1], but it is now emerging that sedentary behaviour at this age may also impact on health. Sedentary behaviour refers to any waking behaviour characterized by an energy expenditure ≤1.5 METs while in a sitting or reclining posture [2]. Spending more than 2 hours per day in sedentary behaviours has been associated with unfavourable body composition, decreased fitness, lowered scores for self-esteem and pro-social behaviour and decreased academic achievement in school-aged children and youth [3]. Whilst being sedentary used to be regarded as the ‘opposite’ end of a physical activity continuum, they are now conceptualised as separate constructs as the behaviours are not mutually exclusive: children can be highly active and highly sedentary [4, 5], or not very active but also not very sedentary [6].

Sedentary behaviour is a broad construct encompassing a range of behaviours [2]. Whilst television viewing time was widely used as a proxy measure for all sedentary behaviour in the past [7, 8], over recent decades there has been a substantial increase in other sedentary activities, especially those that are screen-based such as using computers (for schoolwork, games and social networking) and playing electronic games [9]. Together these screen based media (SBM) occupy a considerable portion of young peoples’ discretionary leisure time, with recent estimates ranging from anywhere between 4 and 7 hours per day for an adolescent [10].

It is becoming apparent that the various types of SBM may be independently associated with some developmental and health risks and outcomes. For example, in a large study of Finnish adolescents [11] general computer use and playing electronic games were related to neck pain but television viewing was not. Similarly, Gopinath et al. [12] found that television viewing and playing electronic games were positively associated with diastolic blood pressure in 12-year-old children, whereas no association was found for computer use. In a national study of US children and adolescents (aged 6–17 years), television viewing was associated with poorer general health as well as poorer social/emotional health whereas reported time spent on the computer, specific use not defined, was not [13]. Babey et al. [14] also observed significantly different socio-economic and environmental correlates for television viewing compared with computer use in US adolescents. Treating SBM as a single construct may thus result in important links with health and development being missed [15]. However, even separating SBM into television viewing, computer use and electronic game use may not be sufficiently specific for understanding some relationships. Within computer use there is evidence for differential effects depending on the type of computer use. For example, computer use for gaming has been correlated with poor academic outcome [16] however computer use for communication and educational activities has been associated with better academic outcomes [17] and more favourable estimates of psychosocial stress [18].

Reducing SBM use in leisure time is regarded as a promising strategy to help prevent obesity and promote activity [19] and numerous interventions are investigating the best way to achieve this [20]. However to effectively target SBM interventions, it is essential to understand who is using what type of SBM, and also to establish whether certain SBM are associated with distinct health risks. Organising people into clusters, based on similar behaviours, may be useful to enhance the understanding of health links, identify intervention targets and tailor interventions [6, 21, 22].

Cluster analysis has been conducted using physical activity and sedentary behaviour exposures and has shown that children can be both physically active and engaged in substantial sedentary behaviours [4, 23, 24]. Recently, Ferrar et al. [25], identified over 19 studies in the past 10 years that have assessed clusters of time use patterns in adolescents. They found consistent cluster and cluster correlate patterns amongst the diverse studies, supporting the notion that this is a useful and important approach to identifying groups of people for intervention. Many of these clusters involved elements of screen time, often involving clusters which exhibited both high screen times and high levels of physical activity — the so-called “techno-actives. Indeed, Olds et al. [23] in their survey of South Australian children aged 9–14 years, found females classed as “sporties” had high SBM use and females classed as “screenies” were more active than those classed as “socialisers”, who were engaged in predominantly sedentary social activity. Nelson et al. [24], using data from the National Longitudinal Study of Adolescent Health, identified seven groups of adolescents aged 14–16 years, based on physical activity and sedentary behaviours– ranging from inactive adolescents who spent a lot of time watching television and playing games to those who did little physical activity and watched little television. Jago et al. [4] identified three clusters in British children aged 10–11 years; those who were highly active and not sedentary (more likely to be males), those who were low active and moderately sedentary (more likely to be females) and those who were both highly active and highly sedentary. Clusters, or classes, of children have also been identified using latent class analysis (LCA) [26], a model-based approach which offers several analytic advantages over standard cluster analysis [6, 21]. LCA has been used successfully to identify groups of young individuals according to health risk behaviours including physical inactivity [21, 27] and overall sedentary behaviour [21, 22].

However, there have been no studies published to date that have used either standard cluster analysis or LCA to identify groups based solely on their SBM use, nor that have differentiated the type of computer use. Given the emerging evidence of the importance of sedentary exposure to health, a better understanding of how SBM behaviours cluster is needed. The aim of this paper therefore was to establish firstly whether distinct clusters of SBM activity exist, and secondly to examine the relationship of any identified clusters with other sedentary and physical activity behaviours and with physical and mental health indicators.

This is the first study, to our knowledge, to examine whether adolescents can be grouped solely on their SBM behaviour. SBM was highly prevalent amongst these adolescents, with estimates of 27 hours/week comparable with prior findings [10]. Three defining clusters of SBM were observed: C1 ‘instrumental computer users’ who were defined by computer use for social communication, email, but also had high exposure to general computer use; C2 ‘multi-modal e-gamers’ who had both high console game use and high computer game use; and C3 ‘computer e-gamers’ who had high computer game use only. Television viewing was high in all three clusters. The findings confirm the ability to categorise adolescents by aspects of their time use, which Ferrar et al. [29] recently demonstrated across a range of time use studies from around the world. Population estimates indicated that whilst one of the clusters was exclusively male (C2 ‘multi-modal e-gamers’), two mixed gender clusters were identified, suggesting that the differences in SBM use are more than just a gender issue. Indeed although C1 and C2 had similar overall SBM exposure, they had different computer and TV use, highlighting the importance of more specific measures of computer use and of not treating SBM as a single construct. The SBM clusters identified in this study showed specific patterns of screen use behaviours that could be useful to assist in targeting of interventions aimed at reducing screen time.

The SBM clusters were shown to differ in how they correlated with other activity/ sedentary behaviours. Multi-modal e-gamer males (C2), who were distinguishable by their high time on both console games and computer games, recorded lower overall weekly steps and lower time in - MVPA than males from the other two clusters. Lower reported exercise and sport levels have also been observed in other high TV/ high video games clusters [24], though high screen time and high physical activity levels can also co-exist [4]. However screen time is often not broken down into specific activities and clustering by specific SBM activities may therefore offer some advantage. Whilst it is not known whether targeting gaming for these males would result in more activity, knowing this association could be important for helping shape interventions aimed at either activity or screen-time. Indeed it was also the C2 males that also recorded the highest TV time. Phone use was also different for the males in the ‘instrumental computer user’ cluster (C1), though perhaps this is not surprising as males who use email more may be more socially communicative in general than males who are low email users. Social clusters, demonstrating a myriad of social activities, have been identified by others [49]. Phone use was high for the females of both clusters and was therefore not able to distinguish between clusters. Females are known to spend more time on social activities more than males [50] and it was not surprising that the majority were female in the ‘instrumental computer user’ cluster, that was noted for its higher email use.

SBM clusters were also shown to relate to physical health indicators. Whilst the evidence was weak, there was a suggestion that the multi-modal e-gamers, those spending the highest time on computer and console games, were at risk of poor back muscle endurance, which may be reflective of the amount of time spent sitting in a slumped position. Computer use is widely recognised to negatively impact on neck and back posture and muscle activity [51] which could partly explain this. It was therefore surprising that there was no association between cluster grouping and neck and back pain, though previous studies looking at computer use and pain in adolescents have produced mixed findings [52, 53]. It may be that it is not just the amount of computer use, but the pattern of use and extent of ergonomic mismatch between user and computer which leads to pain [54]. No associations were found between cluster groups and BMI in our study. Whilst Jago et al. [4] also found no difference in BMI between their groups which differed markedly in both physical activity and sedentary behaviours, other cluster research suggests a more consistent pattern of higher SBM activity with higher levels of overweight [25]. Overall, the activity and health profile tended to be better for those in the ‘instrumental computer user’ cluster, suggesting intervention priority should be given to targeting of the ‘multi-modal e-gamers’ and computer e-gamers’.

Psychosocial health indicators were also correlated with SBM cluster type. The ‘instrumental computer users’ females had marginally better self-efficacy than the females who spent more time on computer games. Strong relationships with peers are known to be important for adolescent self-efficacy [55]. Social networking, characterised by email use in this study (but more likely to be use of social media sites such as facebook, twitter etc. in future studies) might be reflecting this connection and management of social networks [56]. Further, the ‘instrumental computer user’ males who were less involved in computer games and more involved in emailing and phone use reported better mood, which again could be a reflection of a better social peer network. These data suggest that involving these adolescents’ social networks may be an important avenue for future SBM focussed intervention work, as has been highlighted by others in physical activity research [57]. As expected perhaps, C1 males came from higher socio-economic areas which is consistent with the research showing higher levels of child and adolescent computer game use amongst families from lower incomes [58]. Socioeconomic status appears to be related to a number of aspects of SBM related to what technology is used and for what purposes [58]. Again, the different cluster characteristics suggest clusters may help inform better-targeted interventions, in this case using social networks to target C1 adolescents and targeting computer game use in lower SES families.

SBM use by adolescents did cluster and these clusters related differently to other activity/sedentary behaviours and physical and psychosocial health indicators. It is clear that SBM use is not a single construct and future research needs to take consideration of the types of computer use and electronic game use in addition to TV viewing if it intends to understand the impact SBM has on health. Cluster analysis offers an attractive method to simplify this complexity and allows exploration of relationships with markers of health. Cluster analysis also offers an attractive approach to assist with the targeting interventions aiming to decrease SBM use or increase MVPA.