Perhaps unfortunately timed given the reliance we have on screens these days, JAMA Pediatrics has come out with an article entitled “Association of Early-Life Social and Digital Media Experiences With Development of Autism Spectrum Disorder–Like Symptoms.” The title doesn’t leave much to the imagination and the paper pretty much says what you’d expect. It focuses on little kids, and says that kids who are exposed to more screens at age 12 months are more likely to have autism-like symptoms at the age of 2. This is an observational study: it just compares kids with more screen exposure to those with less.
The title is careful to use the word “Association” rather than “Causal Link” but I will say the paper is much, much less careful. They conclude by saying this is a modifiable risk — meaning that screen time exposure is something we should be looking to change to reduce autism — a statement which effectively implies causality. An accompanying editorial in the journal doubles down on this.
So how convincing is the data? Honestly, I would say not very. Allow me to list two grievances!
- There are huge differences across income groups in autism symptoms and diagnosis; we know there are differences in screen viewing (although they do not show this) and their measures of income and socioeconomic status are not perfect. Together this equals bias in estimates
- The results are actually decidedly mixed. Autism diagnosis is not associated significantly with screen time at either 12 or 18 months. Autism symptom measures are associated with screen time at 12 months but not 18 months. So, really, they’re only picking up a significant result in one of their four comparisons.
This is only two complaints but, really, they are big ones. I applaud the efforts to get new data on these questions, but let’s try to remind ourselves it is an association, not a causal argument.