Super non-critical write-up of an extremely unconvincing study, both contributing to misinformation on causes of autism: Autism and DDT: What one million pregnancies can — and can’t — reveal nature.com/articles/d4158…
On the headline alone: it wasn't an analysis of a million pregnancies - it was a search through a million pregnancies to find 778 with autism, which were compared to "matched" 778 without. So, what is presented as a huge sample was actually TINY!
Both the primary article and the write-up by @NatureNews state that exposure to high levels of DDT increased risk of autism by 32%! That sounds enormous! Of course, it's a *relative* increase - from about 1% to about 1.32% - also TINY
Moreover, that finding rests on one data point (in a tiny sample) - not exactly robust. And, as the authors (almost) concede, it would not survive correction for multiple tests (as another chemical was also tested for association)
Even if were robust (which it isn't) and even if were a substantial effect (which it really isn't) this is an observational, correlation study. Causation cannot be inferred.
So, we have findings that at best should be considered as very preliminary data presented as a solid and important finding in a flagship journal (and many other places), perpetuating the idea of environmental toxins as the cause of an "autism epidemic"
There is no such epidemic and this paper certainly does not provide strong evidence for DDT as an environmental risk factor
...ends, in exasperation
I forgot to mention the meta-problem of publication bias in fields looking for "effects", especially with small scale studies. Would this study have been published if negative?
It's hard to overestimate the damage that massive uncritical news coverage of these kinds of stories does. All the public will see is the headline and the misleading statistic: DDT causes 32% risk in autism!
This feeds into idea that autism is somehow caused by environmental toxins. It's red meat for conspiracy theorists and anti-vaxxers. This kind of stuff puts lives at risk.
A related post: Autism, epidemiology, and the public perception of evidence wiringthebrain.com/2014/11/autism… 
