Maternal antidepressant use and the risk of autism

Autism spectrum disorder (ASD), and the factors that may cause it, are frequently discussed in the news. There have been reports in the media over the last week suggesting that maternal use of antidepressants during pregnancy ‘doubles the risk of autism’. The coverage generated highlights the need to follow our top tip of ‘Don’t stop at the headline’. What does ‘doubling’ the risk actually mean here, and are there alternative explanations that could account for any increased risk?

The study has been comprehensively described by the NHS Choices team and we would recommend anyone wanting to find out more about the specifics to read their thorough description and analysis of the study. We will focus on the media coverage which, fortunately, was mostly accurate. The exception to this was use of sensationalist headlines that, out of context, could raise fear and alarm in pregnant mothers taking anti-depressants. Both the Telegraph and Daily Mail ran with headlines that suggested that taking anti-depressants during pregnancy ‘doubles the risk’ of having a child with autism. When we get into the nuts and bolts of what this means it becomes clear that these headlines are unnecessarily sensationalist.

There are two key contextual issues to bear in mind here. Firstly, the ‘doubling’ that is referred to – terminology which will cause alarm – is numerically a very small risk. The diagnosis of autism in the overall sample was less than 1% (0.72% to be exact) – ‘doubling’ the risk raised the diagnosis rate to 1.2%, reflecting not even one extra child per 100. While the Daily Mail did refer to ‘doubling the risk’ in their headline, fortunately they were very clear in emphasizing the context of this in their sub-headlines. They pointed out that the absolute risk ‘remains tiny’ and provided the figure of 1.2% of children who developed autism whose mothers had taken anti-depressants. Clearly a very small proportion of children whose mothers take anti-depressants during pregnancy develop autism – almost 99% of those who take this medication do not go on to have a child with autism. It is good to see this emphasized in headings within reporting, albeit only as sub-headings.

While the researchers did control for maternal depression there may be other factors – whether genetic or environmental – that predispose a mother to depression and her child to autism. So the factors causing this small increased risk may not be anything to do with the actual anti-depressant treatment itself – something we call confounding variables in scientific research. The Huffington Post summed this up nicely – “the key problem is that women who take a medication when they are pregnant have a reason for taking it. Blaming the outcome on the medication, without considering the underlying disease, is like saying that umbrellas cause flooding. Taking away the umbrellas – “not treating the rain” – does not prevent flooding, it just means that people get soaked“. We gave similar examples in our posts on assessing risk and associations and causal relationships. Within the context of this anti-depressant study, more research is needed to identify the factors responsible for the small increased risk observed.

Boukhris, T. et al. (2015). Antidepressant Use During Pregnancy and the Risk of Autism Spectrum Disorder in Children. JAMA Pediatrics. DOI: 10.1001/jamapediatrics.2015.3356