Background/aim Findings regarding air pollution and of autism spectrum disorder (ASD) are inconsistent. A recent publication testing associations between exposure to NO2 and autism used a distributed lag model with ASD data and weekly NO2 exposures from Israel in order to identify windows of vulnerability. When mutually adjusted, prenatal associations reached a negative peak around week 15 of gestation.

Methods A directed acyclic graph (DAG) was built to represent the causal structure and the underlying assumptions needed for live birth bias to create biassed negative associations between prenatal exposure to air pollution and risk of ASD.

Results Live-birth bias could arise from the fact that ASD can only be assessed in live-born children, and many pregnancies are lost and do not end in a live birth. This inevitable selection of only live births into the analysis may lead to bias of the observed association from the actual causal association if a) air pollution is a risk factor for pregnancy loss (this assumption is supported by some literature); and b) there are other factors (‘U’, likely unmeasured, even unknown) that influence both pregnancy loss and ASD. A possible ‘U’ is prenatal stress, which is implicated in both pregnancy loss and ASD. Selection of live births opens the backdoor path ASD <-- U -->Pregnancy Loss <-- Air Pollution, associates air pollution and ASD and biases the causal association in question.

Conclusion Live-birth bias can create a negative association between air pollution and ASD. This bias has implications for all air pollution-ASD studies, and it may also be relevant to other neurodevelopmental conditions.