How To Interpret Studies On Screen Time And Eating BehaviourThursday, April 28, 2016
Much of the research on the contribution of screen time, sedentariness, food consumption and other factors comes from cross-sectional or longitudinal studies, where researchers essentially describe correlations and statistical “effect sizes”.
To be at all meaningful, analyses in such studies need to be adjusted for known (or at least likely) confounders (or at least the confounders that happen to available).
No matter how you turn and wind the data, such studies by definition cannot prove causality or (even less likely) predict the outcome of actual intervention studies.
Nevertheless, such studies can be helpful in generating hypotheses.
Thus, for example, I read with interest the recent paper by Lei Shang and colleagues from the University of Laval, Quebec, Canada, published in Preventive Medicine Reports.
The researchers looked at cross-sectional data on 630 Canadian children aged 8-10 years with at least one obese biological parent.
While the overall median daily screen time was about 2.2 hours, longer screen time was associated with higher intake of energy (74 kcal) and lower intake of vegetables & fruit (- 0.3 serving/1000 kcal).
This unhealthy “effect” of screen time on diet appeared even stronger among children with overweight.
Thus, there is no doubt that the study shows that,
“Screen time is associated with less desirable food choices, particularly in overweight children.”
The question of course remains whether or not this relationship is actual causal or in other words, does watching more television lead to an unhealthier diet (I am guessing no one assumes that eating an unhealthier diet leads to more TV watching).
Unfortunately, this is not a question that can be answered by this type of research.
Nor, is this type of research likely to predict whether or not reducing screen time will get the kids to eat better.
Indeed, it doesn’t take a lot of imagination to come up with other explanations for these findings that would not require any assumption of a causal link between eating behaviours and television watching.
For one, TV watching could simply be a surrogate measure for parenting style – perhaps parents that let their kids watch a lot of TV are also less concerned about the food they eat.
And, for all we know, reducing TV time may (e.g. by cutting the kids off from TV – or cutting the parents off from a convenient babysitter) in the end make the kids eating behaviours even worse.
Who knows – that’s exactly the point – who knows?
To be fair, the authors are entirely aware of the limitations of such studies:
“This study was cross-sectional, so no causal inference could be made and the possible mechanism is not clear. Although our data collection strictly followed the detailed manual procedure to guarantee the quality control (QUALITY Cohort Technical Documents, 2011), potential bias and errors may still exist in those self-reported questionnaires. A number of potential confounding factors have been adjusted in the regression models, but the results may still be confounded by other known and unknown factors.”
So, while the findings may well fit into the “narrative” of sedentariness -> unhealthy diets -> obesity, we must remain cautious in not overinterpreting findings from these type of studies or jumping to conclusions regarding policies or other interventions.