July 17, 2018

expert reaction to association of digital media use and ADHD symptoms among teenagers

A new study, published in JAMA, investigates whether the frequency of using digital media among 15-16 year olds without significant ADHD symptoms is associated with subsequent occurrence of ADHD symptoms during a 24-month follow-up.

Prof Andy Przybylski, Associate Professor and Director of Research at the Oxford Internet Institute, University of Oxford, said:

“This is an interesting exploratory study that is re-analysing data drawn from teenagers enrolled in a longitudinal study being conducted in Los Angeles in the United States. Strengths of the paper include a relatively large sample and careful documentation of the sample, and a relatively circumspect interpretation of the correlations the researchers report. These are key things missing from many similar studies which generate this kind of buzz. There are three reasons why we should be cautious about this study.

“First, though the analyses are done well they demonstrate a very small correlation between digital media use and non-clinical measure of the ADHD symptoms. This means the study is a proof of concept that tells us we need very large samples when we design future studies because the possible effects are extremely small.

“Second, the study doesn’t measure either digital media use or ADHD directly. For both the study relies on survey responses provided by the student in question. It is not clear if teachers or parents would rate the children similarly or if the self-reported measure of digital screen use is correlated with either actual behaviour or higher quality survey items.

“Finally, because this was an exploratory study, instead of a registered or confirmatory study, the results must be understood as tentative. To the authors’ credit they use a statistical technique to adjust for multiple tests but this could have easily been a paper about depression (a reported control variable but could have originally been an outcome). It’s possible that the evidence for depression, a popular hypothesis, was not strong enough and so only ADHD symptoms was a publishable outcome. Pre-registered open research practices are needed to know for sure.”

Dr Jessica Agnew-Blais, MRC Postdoctoral Research Fellow, Institute of Psychiatry, Psychology & Neuroscience, King’s College London, said:

“A strength of this study is that it reflects modern-day, multifaceted digital media use, given its comprehensive assessment of activities including video chatting, reading blogs, and liking or sharing other’s social media pictures and posts. This study notes the volatility of ADHD symptom patterns in adolescence, which is an important topic for future research to more fully explore.

“The authors note that the association between digital media use and elevated ADHD symptoms is of modest size, and that when examining a continuous measure of ADHD symptoms over time while adjusting for baseline ADHD symptom level, digital media use was only associated with a small change in ADHD symptom score over two years. Whether such a small change would have any clinically meaningful implications remains to be seen.

“The authors only assess ADHD symptom frequency, so it remains unclear whether relatively small changes in ADHD symptom frequency over these two years are disruptive or impairing in everyday life.

“It is worth noting that over 80% of students reported high frequency use of digital media, and the vast majority of these students do not have elevated ADHD symptoms.

“The authors are (rightfully) cautious not to over-interpret their finding of an association between digital media use and ADHD symptoms given the possibility of potential residual confounding and the modest size of the association.”

Prof Kevin McConway, Emeritus Professor of Applied Statistics at The Open University, said:

“I really can’t get excited about this study’s findings, even though in statistical terms, what the researchers have done appears careful and competent. It’s appealing to think that two aspects of modern life that have emerged in recent years – ADHD in adolescents and frequent use of digital media like social networking, music streaming and texting – might be linked in some way. But, just because it’s appealing, that doesn’t mean it’s true, and the research report makes it clear there are many reasons for doubt. The research raises more questions than it answers, I’d say. And in any case it’s difficult to get too excited about an association that the researchers explicitly report to be ‘modest’.

“It looks impressive at first glance that thousands of students were involved – but the key number that matters for assessing the strength of evidence in studies like this is really the number who get the disorder in question, ADHD symptoms in this case, and that is in the low hundreds. But, more importantly, any association between ADHD symptoms and digital media use in these students can’t directly be taken as evidence of an association in wider groups of adolescents. That’s because it isn’t even clear that the students involved were typical of students in Los Angeles where they lived (and nobody would claim that the inhabitants of Los Angeles are typical of everywhere in the US, let alone in other countries). Their schools had to be invited to take part, and one of the reasons for inviting them was proximity to the researchers. Then the students had to consent to take part, their parents had to consent as well, and the students had to provide enough of the relevant data over a 24-month period. Of the 4100 students that were eligible to take part in these schools, under 2600 ended up providing useable data. (About 200 were left out because they already reported having ADHD symptoms on entering the study, but most were left out for other reasons.) There’s no way of telling how typical these students were of all students in LA high schools, let alone anywhere else. So maybe there’s an association that could be observed more widely than just in these students, but we can’t by any means be sure of that.

“Even if there really is an association, a study like this can tell us very little about the reasons for the association – that is, about what actually causes what. Again, the research report, rightly, makes that clear and explicit. We certainly don’t know that frequent use of digital media causes ADHD. That remains a possibility, but there are many other possibilities that this study can’t rule out. Maybe some other factor independently caused some students to use social networks, stream music and text more frequently than most, and also caused those students to be more likely to report ADHD-like symptoms later. In other words, perhaps the frequent digital media use and the reported ADHD symptoms are just two aspects of some wider issue, so that if somehow the students in question were prevented from using the media so often, they would still show the ADHD symptoms. Statistical adjustments can be made to try to allow for possible confounding factors like this, and these researchers did make some such adjustments, but they can’t allow for things they didn’t measure. Indeed there are other possibilities too. The research report mentions reverse causation – that is, the possibility that the frequent media use might be caused by ADHD symptoms, rather than the other way round. That sounds odd, given that the reports of ADHD symptoms occurred after the reports of media use, but it can’t be ruled out because the times when the students report things might not always match the times when they first occurred. Indeed, because everything in the study is based on the students’ own reports of their uses of digital media and there possible ADHD symptoms, it’s difficult to be sure that there would be similar associations if we had some objective measures of their media use and proper clinical diagnoses of ADHD. For all kinds of reasons, what people report about themselves doesn’t always match a more objective measure of what’s going on.”

* ‘Association of Digital Media Use With Subsequent Symptoms of Attention-Deficit/ Hyperactivity Disorder Among Adolescents’ by Ra et al. will be published in JAMA on Tuesday 17 July.

Declared interests

Prof Andy Przybylski: “I have no conflicts of interest to disclose.”

Dr Jessica Agnew-Blais: no conflicts of interest.

Prof Kevin McConway: “Kevin McConway is a member of the Advisory Committee of the Science Media Centre”.