March 17, 2014

expert reaction to two studies investigating children’s screen time and health

Two studies in JAMA Pediatrics  investigated children’s use of electronic media; one reported associations between screen time and some measures of child emotional wellbeing, the other between maternal monitoring and body mass index (BMI) in children.

Comments on study by T Hinkley et al looking at screen time and emotional wellbeing in children*

Prof David Spiegelhalter, Winton Professor of the Public Understanding of Risk, University of Cambridge, said:

“From this study, it is impossible to estimate a child’s increased risk from an hour of extra screen time, or even if there is any effect at all.   Most of the measures show no evidence of an effect, some show marginal increased risk but we would always expect some results like this by chance alone, given the number of outcomes measured: indeed some outcomes suggest extra screen time is beneficial.  Any impact of increased screen time in pre-school children will be complex and multi-factorial, and studies like this are not capable of determining causation.”

Prof Dorothy Bishop, Professor of Developmental Neuropsychology, University of Oxford, said:

“This study looked at emotional well-being, self-esteem, and family and social functioning of children in relation to the amount of time spent on watching TV or using a computer, measured two years earlier. Out of 48 analyses, only one showed evidence of poorer outcomes in both boys and girl, after taking into account socio-emotional characteristics at the start of the study and background variables. This was a slight increase in impaired family functioning associated with number of hours spent watching TV on a weekday. There were no consistent effects for boys and girls in relation to hours spent using e-games/computers, and measures of self-esteem, emotional well-being or social functioning.”

Dr Andrea Danese, Senior Lecturer and Consultant in Child & Adolescent Psychiatry, King’s College London’s Institute of Psychiatry, said:

“The paper addresses a broad and intriguing question: is technology, namely something we built to improve our lives and entertain us, actually detrimental for children’s well-being? In particular, the paper tested the effects of electronic media use (i.e., TV, video games, computer use) on children’s emotions and social interactions both within and outside the family.

“The study was well designed. It was large, involving more than 3,500 two to six-year-old children from eight European countries. It was longitudinal, allowing authors to test whether electronic media was linked to changes in children’s well-being over time. It used well-validated measures.

“The study results, however, were puzzling. First, electronic media, on average, exerted only small effects on children’s well-being. Even in the presence of ‘statistically significant’ effects, the extent to which these effects are clinically meaningful remains arguable.

“Second, there was wide variation in how much electronic media affected well-being in different children. In other words, whether or not electronic media affected well-being is likely to depend on several other characteristics of the child and the family. Finally, the effects were inconsistent between boys and girls and between different measures of well-being.

“Therefore, based on these results, it does not appear that electronic media is generally detrimental for children’s well-being.”

Prof Lynne Murray, Research Professor in Developmental Psychopathology, University of Reading, said:

“These findings are in line with those of several studies in North America (US and Canada) on TV viewing – especially for the under twos. Like any population study that doesn’t include an intervention (this report excludes intervention families) inferring causal direction is problematic, and there are always unmeasured factors that could account for the associations (e.g. do only some kinds of families allow their children to engage in a lot of screen time?). The researchers try to account for these, but one can never be certain that something else not measured hasn’t been operating. Having said that, and as mentioned before, the results are entirely consistent with other, very well-controlled, studies, some of which HAVE included quasi intervention components – so, although it might make for uncomfortable reading, it’s important to take the message on board. But further important questions remain – what KIND of screen time is most influential; can materials be produced that do NOT have such negative consequences; and if children DO engage in screen time, can parents mitigate the harmful effects?  For example, there is evidence that sitting with the child, and discussing the material, can be of help to children in processing the content.”

Professor Sonia Livingstone, Professor of Social Psychology, London School of Economics, said:

“This seems a conscientiously conducted study on a substantial scale in an important area, as concern about the possible adverse effects of very young children’s screen time rises. But most of the effects tested were not supported by the findings – in other words, there was no evidence that spending more time with computers or television had adverse effects on peer problems, self-esteem or social functioning. The findings that girls who play more computer games have some emotional problems, and that children who watch more television experience more problematic family function is interesting but remains unexplained by the study. Since only some of the findings expected were substantiated by the study, and the authors seem not to have an explanation for why some but not other findings were obtained, there is clearly more research needed to understand what is really going on here – we don’t yet know the causal direction, or the mechanisms at work, if any, in accounting for such findings.”

Prof Patrick Wolfe, Professor of Statistics, UCL, said:

“The authors report a 1.2- to 2.0-fold increase in poorer well-being outcomes for each additional hour of electronic media use.  These figures refer to the inferred increase in the likelihood of being at risk for different outcomes (such as emotional problems or poor family functioning), associated with each additional usage hour. 

“However, it’s important to not simply focus exclusively on these numbers and lose sight of the broader context of the study.  Here’s why:

“First, to understand these figures a bit better, let’s take a closer look at Table 3 in the paper*.  Here the authors report these inferred likelihood increases as odds ratios, with 1 being even odds.  As can be seen in Table 3, putative changes in likelihood (i.e. odds ratios) were estimated under a variety of different modeling scenarios and with respect to different types and times of media usage, for a total of 96 different combinations of models, categories, and outcome types.

“However, the inferred likelihoods are not significantly different from 1 (i.e., even odds) in most (nearly 90% of) categories the authors analyzed.  Because of this, I think the results should be interpreted with due caution. Specifically, the authors report 11 combinations in which the estimated odds ratios differ significantly from 1; these 11 estimated odds ratios vary from 1.2-2.0 (where, as a reminder, 1.0 represents even odds).  However, looking again at Table 3, the reported upper range of 2.0 is really quite different from the median of all 11 heightened odds ratio estimates, which is 1.3, and the mean of all 11, which is 1.42.

“Thus it is important that we not interpret these results as saying something stronger than they are.  Of the 96 investigated combinations, only 11 yielded odds ratios significantly greater than 1 (even odds).  Of these 11, half were between 1 and 1.3.

“Second, stepping back and thinking through the study more broadly, it’s important not to confuse symptoms and causes here – electronic media usage may well be coupled to other things the authors did not study (which in turn might be influencing well-being, which in turn might be mediating media habits!).

“Overall, it is interesting to look for an effect here – but I think if there is a linkage, then it’s probably related to quite a few other more important factors.  There’s probably a lot more going on than any simple direct link, and as the authors themselves say, ‘Further research is required to identify potential mechanisms of this association’.”

Comments on study by S Tiberio et al looking at screen time and Body Mass Index in children^ǂ

Dr Esther van Sluijs, Group Leader, MRC Epidemiology Unit & Centre for Diet and Activity Research (CEDAR), University of Cambridge, said:

“This study adds to a growing body of evidence on the importance of parenting practices for the development of a healthy weight during childhood. Further research is now needed into what determines when and how much parents monitor their child’s media use and how it can be changed.

“There is little evidence available on how we can help parents to effectively change their children’s health behaviours and this should be a priority moving forward.”

Dr Gavin Sandercock, Senior Lecturer in Clinical Physiology, University of Essex, said:

“We know that obesity runs in families and this paper shows that the parents of children who get fatter from 5 – 10 years old are less likely to have rules around TV and other media use (screen time).

“In particular, the study reports that that mums who don’t monitor their young children’s screen time may be putting them at risk of becoming overweight or obese. Children whose screen time wasn’t monitored watched more TV and got fatter more quickly than those living in families with rules around screen time.

“This offers a mechanism to link screen time and fatness and suggests that inactivity comes before obesity in a child’s life. The results go against the idea that fatness leads to lower activity levels and suggest that the UK should adopt guideline daily limits on TV and screen time as seen in the US or Australia.

“The screen time reported in this study was 1h 45min per day which is much lower than the values for children in our own studies – within which 40% of boys and 20% of girls exceed 2h per day.”

Prof Patrick Wolfe, Professor of Statistics, UCL, said:

“Overall this is a well described and responsibly reported study, and its limitations are well documented by the authors.

“I think the toughest thing to disambiguate here is the (hidden) linkage between media monitoring and other parental characteristics – perhaps closely-monitoring parents are more likely to be engaged overall, and therefore it’s natural to ask if the link the authors report is in fact direct, or simply a consequence of some other parenting choices.  (We also know that some of the other variables in the study, such as activity participation, are linked to socioeconomic strata.)  As the authors note in their concluding sentence, ‘future research must identify additional explanatory variables’.

“There is also the question of which population, precisely, is under study – since fathers were originally selected for the study cohort based on risks associated with their geography.

“Finally, while child BMI was directly measured, parents reported both monitoring habits and child media time – this has the potential to introduce response bias (under/over-reporting), which in turn could be coupled with the BMI of the child in question.  This has the potential to distort the reported results.”

*‘Early childhood electronic media use as a predictor of poorer well-being: a prospective cohort study’ by Trina Hinkley et al. published in JAMA Paediatrics on Monday 17 March 2014.

^ǂ‘Parental monitoring of children’s media consumption: the long-term influences on Body Mass Index in children’ by Stacey S. Tiberio et al.  published in JAMA Paediatrics on Monday 17 March 2014.