A new study, in The BMJ, examines the association between exposure to air pollution from road traffic in London during pregnancy and an increased risk of low birth weight babies born at full term.
A roundup accompanied this analysis.
Title, Date of Publication & Journal
‘Impact of London’s road traffic air and noise pollution on birth weight: retrospective population based cohort study’ by Toledano et al.
Published: Tuesday 5th 2017
Study’s main claims – and are they supported by the data
The paper does not demonstrate that “air pollution from road traffic in London is adversely affecting fetal growth”.
This is an observational study and, as such, cannot establish a causal link. Some important confounders (e.g. smoking) have not been properly taken into account.
The claim that there is little evidence for an independent effect of traffic-related noise on birth weight outcomes appears to be largely supported by the study.
This is a large study in which air pollution and traffic noise has been carefully modelled. These were not measured directly, but the modelling used by the authors was reasonably accurate.
The authors adjusted for some important confounders such as maternal age, area-level deprivation, area-level expenditure on tobacco (as a proxy for smoking), and birth registration type (as a proxy for marital status). But this will not properly take account of confounding, and may mean the main conclusion – that pollution is adversely affecting fetal growth – is not correct.
Smoking in pregnancy is one of the top risk factors for low birth weight. It is plausible that mothers in deprived areas are more likely to smoke AND be exposed to pollution – so smoking is an important potential confounder here. But the study only uses an approximation for smoking and doesn’t look at whether individual mothers smoked.
Similarly, individual-level variability in deprivation could also be important as it could affect birth weight through other key measures (e.g. a mother’s BMI).
When the authors adjusted for confounders it reduced the odds ratios quite substantially; it is likely that accurate adjustment for things like smoking would reduce them even further. As the odds ratios are already very small in magnitude, it is possible that the observed associations are entirely due to residual confounding.
The study did not measure or model individual mothers’ exposure to pollution. In other words, we only know the levels of pollution in the local areas – not how much each person was actually exposed to.
Residual confounding – confounding that has not been adjusted for, either because certain factors have not been measured, or because they have been measured inaccurately.
Odds ratio – a measure of relative risk. An odds ratio of 1 means no effect.
* ‘Impact of London’s road traffic air and noise pollution on birth weight: retrospective population based cohort study’ by Rachel B Smith et al. published in The BMJ at on Tuesday 5 December 2017.
Before The Headlines is a service provided to the SMC by volunteer statisticians: members of the Royal Statistical Society (RSS), Statisticians in the Pharmaceutical Industry (PSI) and experienced statisticians in academia and research. A list of contributors, including affiliations, is available here.