Title, Date of Publication & Journal‘

Association of prenatal exposure to air pollution with autism spectrum disorder’ by Lief Pagalan et al. was published in JAMA Pediatrics at 16:00 UK time on Monday 19 November 2018.

Study’s main claims – and are they supported by the data

The authors suggest that their study shows that women living in areas with high nitric oxide pollution are more likely to have children with autism spectrum disorder (ASD) even after allowing for a series of factors such as family income.  This is a large study but the results are only marginally significant.  As such the study is of interest but is certainly not headline news.

The authors say that their findings “suggest that reducing exposure to NO for pregnant women may be associated with a reduction in ASD incidence” – however, given this study is observational and cannot prove whether one thing caused the other, we cannot know whether reduction in exposure would lead to a reduction in incidence.

In fact, the apparent association with ASD in this study is similar for all three measures of air pollution (particulate matter with a diameter less than 2.5 μm (PM2.5), nitrogen dioxide (NO2), and nitric oxide (NO)) with very similar size of effect, but only nitric oxide crept over the line of statistical significance.  I am not convinced the effect of NO is meaningfully different from the effect of the other two pollutants.  In any case, the effect size is small and as this is observational we can’t know whether it’s a causal relationship.

The authors note that previous studies in the US have indicated closeness to motorways and air pollution as possible triggers as suggested in this study, whereas three studies in Europe have not confirmed any association between air pollution and ASD (although possibly due to weaknesses in those studies) – so the picture is not clear.  ASD is thought to result from a complex combination of genes and other research suggests it is most likely triggered or complicated by several environmental factors.

Strengths/Limitations

Strengths:

The study is relatively large and well organised.

Limitations:

No correction is made for carrying out three significance tests.  Once corrected the results may no longer be statistically significant.

The study analysis does not allow for the fact that mothers with several children appear several times in the data once for each child.  This cluster effect will exaggerate the statistical significance level which is already close to a nominal level of 5% (i.e. close to not being statistically significant anyway).

As with all such cohort studies, association does not imply causation – the study cannot show that it is air pollutants that cause an increased risk of ASD.  Access to diagnosis and closeness to polluted areas may well both be associated with mediating factors such as wealth, which have not been allowed for fully in the analysis.

Glossary

ASD = autism spectrum disorder.

Any specific expertise relevant to studied paper (beyond statistical)?

None.