Leisure Time Computer Use and Overweight Development in Young Adults – A Prospective Study

Sara Thomée; Lauren Lissner; Mats Hagberg; Anna Grimby-Ekman

Disclosures

BMC Public Health. 2015;15(839) 

In This Article

Discussion

There were clear cross-sectional and prospective associations between computer gaming and overweight (BMI ≥ 25) among the young women in this study, even after adjusting for demographic (age, occupation) and life style (physical activity, sleep, social support, total computer time) factors. For the young men, only cross-sectional associations could be detected. Leisure time computer use for communicating was not to the same extent related to overweight; spending more than 2 h daily for emailing and chatting was cross-sectionally related to overweight only among the women. No clear prospective associations were found for emailing/chatting and overweight or increased change in BMI in either sex.

The results are partly in line with earlier studies finding a relationship between screen time and overweight or BMI in children and adults (e.g.[7–20,28]). However, to our knowledge there are only few studies that have specifically examined computer use as a risk factor for overweight in adults, and only cross-sectional associations seem to have been found previously.[18–20] The present study suggests that the content of the computer use can be of importance, as time spent on computer gaming appeared to be more connected to weight gain than time spent on emailing or chatting. In this regard, Kautiainen et al.[12] found associations between time spent on the computer for emailing, writing and surfing the internet and overweight, but not for time spent on digital games, which is partly inconsistent with our study. However, the Kautiainen study was cross-sectional and in a younger age group (14–18 years). That the increased change in BMI from baseline to 5-year follow-up in relation to computer gaming was seen mainly among the normal weight women and not among the overweight, is contrary to the results of Falbe et al.[8] and Mitchell et al.,[9] where associations were stronger among the overweight.

The prospective results seen only in the women in the present study possibly indicate gender differences. In the study of Finnish adults, Heinonen et al.[20] found cross-sectional associations between computer time and BMI and waist circumference in the females only. Gender differences have also been found in some of the studies on children or adolescents. For example, in the mentioned study by Kautiainen et al.[12] the association of computer use for emailing, writing and surfing the internet and overweight was statistically significant only in the girls. Furthermore, Falbe et al.[8] found longitudinal associations between digital game playing and increased BMI in girls only. However, in the longitudinal study by Altenberg et al.[13] computer time predicted changes in BMI in the boys, and not in the girls.

Using the internet for communication and social networking has been considered to be a more common activity among females compared to males.[3,29] In our study population, the reported amount of leisure time spent on emailing and chatting was about the same in both sexes, which is in accordance with Swedish internet statistics from 2013.[3] But, there are apparent gender differences in time spent on computer gaming. A literature review on gender differences in online gaming[6] showed many similarities in males and females' motivations to play online games, but males tend to play more action and simulation games, while females play more logic and skills training games. We have no information about game content in our study.

Possible Mechanisms

But how is it that women gamers in particular seemed to be vulnerable to weight gain in our study? One of the potential mechanisms for weight gain in connection to screen time is the sedentary nature of the activity, i.e. sitting behaviors and low energy expenditure. In Vandelanotte et al.,[19] time spent by the computer was associated also with increased time spent on other sedentary activities. Interestingly, on the national public health level in Sweden, the increased use of computers the past decades, and thus the inferred increase of sedentary activities, has been paralleled with an actual increase in reported leisure time physical activity.[30] In our data, the high gamers reported lower levels of physical activity compared to the others. While this also applied to the men, it was especially pronounced among the women gamers. However, the associations between time spent on gaming and overweight were significant even after adjusting for level of physical activity. Moreover, an additional stratified regression analysis showed that gaming was associated with increased change in BMI in all levels of physical activity in the women (data not shown). It is plausible that regular physical exercise a few times per week does not compensate for physical inactivity the rest of the week. It is also possible that physical inactivity by the computer is especially detrimental to women. In a study by Scheers et al.[31] showing associations between BMI and decreased physical activity levels in both men and women (measured with actigraphs), the duration of sedentary bouts and number of breaks in sedentary time was related to BMI mainly in women. It is possible that our female gamers are subject to longer bouts of inactivity than the men even if the total time by the computer is the same. In this line of reasoning, a possible explanation for why the time spent on emailing/chatting was not to the same extent related to overweight, could be that these communication activities not necessarily imply long durations of physical inactivity even if the total time spent is the same, and thus, maybe a higher energy expenditure.

On the other hand, there is evidence that diet is the most important factor for weight gain.[32,33] Screen time has been associated with a less healthy diet including higher consumption of energy dense snacks and drinks and lower consumption of fruits and vegetables, among children, adolescents and adults.[15,34,35] Computer gaming has been suggested to entail a lower energy intake in combination with a slightly higher energy expenditure, compared to TV viewing, as both hands may be busy using the controls (e.g.[16]). Also, TV viewing to a larger extent implies exposure to advertisements for fat and sugary foods.[15,16,36] But the question is then if there are gender differences in energy intake while at the computer? Unfortunately, we have no data on diet in our study. However, there is some evidence that gender differences may exist in this regard; in the systematic review by Pearson et al.[34] of dietary intake and sedentary behavior, about half of those studies that examined gender differences observed them. The conclusion was that the associations between sedentary behavior and diet were more consistent for females than for males. Thus, diet may be an underlying issue in our population.

Sleep is another possible mediator between computer use and overweight, for example shown by Arora et al..[14] Screen activities may interfere with sleep[37–39] and short sleep is associated with overweight and obesity.[40–42] There were no major differences in reported sleep between the men and women in our study. We have previously reported[43] that the "high email/chatters" as well as the "high gamers" of both sexes more often reported having sleep problems than the "low" groups (prevalence ratios in the range of 1.3–1.4 after adjusting for relationship status, educational level, and occupation). Moreover, the "high email/chatters" of both sexes had an increased prospective risk to have developed sleep problems after one year (prospective PRs 1.9 and 1.7, for men and women, respectively).[37] However, there may be gender differences in the association between sleep and weight.[41] Self-reported short sleep duration was associated with increased weight in the men but not in the women in a study of young adults by Meyer et al.,[42] while self-reported sleep problems (falling asleep or staying asleep) was valid for the women and not for the men.

Another aspect that needs to be addressed is that computer gaming is much less common among women. This raises the question if the women gamers possibly are a more select group than the male gamers. Apart from the fact that women play less than men, women seem to take up gaming later in life, and the average female (online) gamer is older than the male.[6] Due to the limited age span of our study population we could not investigate age-related gender differences. As mentioned earlier, the female gamers seemed to have lower levels of physical activity. The high gaming women also reported being subject to lower social support in private life, and thirty percent were neither in work nor in school. In a previous study in the same cohort, we found that the female computer gamers more often reported stress and depressive symptoms,[43] and they had a prospective risk of developing depressive symptoms.[37] There is a reciprocal relationship between obesity and depression, i.e. obesity can lead to depression, but depression can also lead to obesity in some women.[44] Altogether, these women may be subject to several health-related risk factors.

It should also be considered that it is possible that a more detailed categorization of the exposure variables, with higher cut-offs than >2 h per day, may have indicated intensive gaming to be a risk factor for overweight development also in the men.

Methodological Considerations

The strengths of this study include the prospective design with follow-ups after one and five years, and a fairly large study group from a population-based sample. However, there are also several limitations that should be considered. All variables except for age and sex are based on self-reported data. Self-reported BMI is known to be underestimated, mostly because of the underreporting of weight,[45,46] and this may bias the results in unknown ways. Another concern in relation to BMI is that height was only asked for in the baseline questionnaire. BMIs could be overestimated at the follow-ups, due to the fact that men may still be growing at the age of 20–24, although, the proportion of men still growing in this group is probably small.

Further, the validity of self-reported computer use may be questioned,[47,48] implying recall difficulties and recall bias. There may also be potential misclassifications of the two main exposure variables because of them not being mutually exclusive: chatting is sometimes part of computer gaming. Both are included in the total computer use variable and some illogical reports are seen in Table 1, which puts light on the limitations of self-reported data. Further, we assessed only two types of leisure time computer use; gaming and communicating, and thus fail to examine other potential leisure time computer activities. These two were chosen because they were the dominating leisure pastimes by the computer that emerged in a qualitative interview study about computer use and potential mental health effects.[49] It should also be pointed out that the data collection started in 2007, which is prior to the broad use of social media applications such as Facebook, Twitter, etc. Moreover, we do not examine gaming and communicating on other devices. It can be questioned if it is relevant to single out computers as a specific exposure, as technological development gives us a variety of devices for similar activities. For example, in the computer gaming reference group (i.e. no computer gaming), there may be participants who are heavy gamers but on other types of consoles, which hypothetically would entail the same type of consequences. Thus, this type of "misclassification" may dilute effects. Furthermore, the baseline data was collected before the widespread use of smartphones and the mobile internet, but there are probably participants who handled emails and chatting via a mobile phone at the time of the data collection. This may also be a source of misclassification. In order to keep up with developments, the questionnaire items were changed from only concerning computer use to also including smartphones and tablets at the 5-year follow-up in 2012. Although this is a reasonable modernization, it limits the comparability of the exposure data from baseline to the 5 year follow-up. Regardless, it was inevitable for the present study to fail to take into account the diversity of technologies and applications that may today be present. Exposure assessment that is applicable in longitudinal studies is a challenge when studying the modern technologies. Future studies could probably be strengthened by the use of objective exposure measures. It also seems relevant with more detailed assessments of the factors involved in the potential mechanisms. Among the limitations of our study, is the lack of information about, for example, time spent on other sedentary activities, and diet.

The study group confers some limitations due to selection biases. There was a low response rate at baseline and attrition to the follow-ups. A healthy selection can be expected. Women were overrepresented in the study group, and a non-respondent analysis at baseline showed that also native Swedes were overrepresented.[22] Although we adjusted for some demographic and lifestyle factors, only baseline data were used for this, and there may be other confounding factors that are unaccounted for. For example, socioeconomic position (SEP) has been seen to have an inverse relationship with overweight in several European countries, including Sweden.[50] We lack a relevant marker for SEP in the study. Educational level, which is often used as a marker for SEP, was considered to be of low relevance in this young adult group. Naturally, in the course of the five years of study, educational levels increased, and a higher proportion of the participants worked, rather than studied, at the 5 year follow-up. Finally, carrying out separate analyses for men and women eliminated confounding due to sex, but the low response-rate, especially among the men, suggests that caution should be used when generalizing the results to a general population of (Swedish) young adults.

Implications

We have identified a new risk group for overweight development: young adult female computer gamers. The estimated additional increase of 1.33 BMI-units in the 5-year follow-up period for those who gamed >2 h per day at baseline, would for a woman of average height and weight in the cohort, correspond to an additional weight gain of 3.7 kg. The results should warrant attention for further research to confirm or mitigate the results. As overweight and obesity bring about negative consequences for health and quality of life, especially in females,[51,52] this could be an important group to target for public health preventive strategies. It seems necessary to investigate more closely which mechanisms can be at play in the suggested weight development, in order to develop relevant interventions and direct the right actions.

processing....