C-sections and autism

The observation that autism prevalence has been increasing, beyond what would be explained by improved diagnosis alone, has launched the quest to identify the factors responsible. In spite of an increasing list of such factors, few can actually stand up to serious scrutiny. However, the media rarely miss a chance to report any potential risk factor, most of the time based on inconclusive studies, which can then have dramatic effects on health choices. The most extreme example was the suggested link between MMR vaccinations and autism, alimenting the anti-vaccination movement and leading to measles outbreaks. The list of putative risk factors includes mercury, pollutants, parenting strategies, dairy products and video games. Emily Willingham, scientist and blogger, has reviewed up to 50 factors, inclduing relevant references, in this caustic post, and Research The Headlines recently covered a story suggesting siblings born closer together are at higher risk for autism.

The latest addition to the list is baby delivery by caesarean section, or C-section, following a recent publication in The Journal of Child Psychology and Psychiatry led by Eileen Curran, based at Department of Obstetrics and Gynaecology at the University College Cork. The research was a meta-analysis of previously published data and reports a pretty minimal (and inconclusive) increased risk of autism and ADHD for C-sections, which requires further investigation. But media headlines are portraying a completely different story: Babies born by Caesarean section are 23% more likely to develop autism, says The Independent, and others like The Irish Times and ITV News have a similar tone. That actually sounds like quite a dramatic (and scary) effect. Being Italian I cannot avoid thinking that, if I were to believe such a figure, in my country of origin where C-sections have the highest frequency in Europe (almost 40% compared to an average of 25%) autism too should be more prevalent. In fairness, most of the media reports explain in the main body of the text that the effect is minimal (and the Daily Mail has a warning in their piece title) and it is unclear whether there is a causal link. So why such dramatic headlines? Where is the confusion?

Autism prevalence is about 1% in the population. An increased 23% risk means an increased 23% on the 1% starting point, which in reality means a very small effect. This is quite a common mistake, that has immediately prompted reactions through social media (see @wiringthebrain for comments). Ben Goldacre had previously discussed this misunderstanding of risk, and suggested that risk is better explained by natural numbers than frequencies – we describe this method in one of our recent How to “Research the Headlines” posts.

So, let’s have a look to the numbers in the study. Well, it is not immediately possible to interpret the provided numbers. The most interesting result is derived by the analysis of 13 studies which, when pooled together, produced an odds ratio (OR) of 1.23 for autism associated with C-section. OR are statistical values used to indicated whether an exposure (or risk factor) has an effect on a phenotype (such as a disease or other outcome, for example). OR = 1 means no effect; OR > 1 means increased risk; OR < 1 means reduced risk. With this value in mind we can work out some numbers. The prevalence of autism is about 1% which suggests that about 10 children in every 1000 will develop the condition. Based on the results of the meta-analysis, we should expect a 23% increase in the baseline prevalence of autism for children born by C-section, which would translate to about 12 children in every 1000. This is not a big increase and certainly not enough to indicate C-sections represent a major risk factor for autism as some of the headlines want us to believe. Furthermore, there is no evidence that C-sections are a causal factor for autism. There are many different reasons C-sections are carried out so, providing the suggested association is genuine, we might also expect that some factors contributing to autism may also contribute to the need for delivery by C-section.

The meta-analysis, therefore, does not show any substantial increased risk of developing autism in children delivered by C-sections, yet the headlines we have seen this week have the potential to be dangerous. In most cases, C-sections are practiced in the best interest of mothers and their babies and could be life-saving. We really do not want to see mothers opting out and putting themselves and their babies at risk because of unfounded and alarmist information.

A press release for a study that does not report particularly strong results was unnecessary; the headlines that followed have been irresponsible and potentially dangerous.

Curran E.A. et al. (2014). Research Review: Birth by caesarean section and development of autism spectrum disorder and attention-deficit/hyperactivity disorder: a systematic review and meta-analysis.. Journal of Child Psychology and Psychiatry. DOI: 10.1111/jcpp.12351