Two presentations, given at the The European Respiratory Society (ERS) International Congress, report on links between air pollution and both increased risk of infant death and decreased lung function in children.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“These certainly aren’t the first pieces of research to provide evidence of an association between exposure to polluted air in the womb or in early life, and various kinds of ill health in babies and children. The two new studies do add what might be important details, though as usual with conference presentations when we haven’t seen the full research report and it hasn’t been through peer, review, it’s pretty hard to be definite about what they have and haven’t shown.
“There are three particular issues about this kind of research where it’s important to know the details of what was done and how. What were the populations studied? Well, that’s reasonably clear even from the brief reports – both sets of data sources are large and respectable, and also, they’re from the UK so there’s not much doubt about whether their findings could apply here. Was the exposure to air pollution measured appropriately? This gets harder to assess. Most people don’t have air pollution monitoring equipment outside their house, and both these studies appear to have used mathematical and statistical models to get measures of pollution on a suitably fine scale. This modelling can’t be 100% accurate but it’s usually pretty good, of average at least – but we don’t have the details. And there’s the problem of confounding. People who live in areas of low air pollution will differ from those who live in more polluted areas in very many ways, and it’s possible that these other differences could be the actual cause of differences in infant mortality or in children’s’ lung function. So, we can’t be sure, from any one observational study like these, that pollution exposure is the cause of differences the observed differences in health or in death rates. Researchers can and do make statistical adjustments to allow for these other differences, as far as they can – but the detail of how the adjustments are done does matter, and no such adjustments can be made for factors on which the researchers have no data. These brief reports don’t say much about such adjustments. The Kotecha et al. study of infant mortality says they adjusted for deprivation, birthweight, the mother’s age, the baby’s sex, and for multiple births such as twins. All these things are known to affect infant mortality – but I don’t yet know how the adjustments were done. No adjustment seems to have been made for whether the mother smoked during pregnancy, probably because the researchers did not have relevant data. There’s evidence of associations between maternal smoking and infant mortality from previous studies, and if mothers in the areas of higher pollution were more likely to smoke, that might be an explanation of part of the new findings. The other study (Hansell et al.) says that they adjusted for potential confounders, but the brief report doesn’t say which ones (though it must have included the child’s sex, the mother’s education level, and whether she smoked during pregnancy).
“Because of these issues, and mainly we haven’t yet got full, peer-reviewed, research reports, I find it hard to weigh up exactly how to interpret these new findings. I suspect, and hope, that this really is useful, high-quality research, but the fact that they are observational studies means that we really can’t conclude that the air pollution is the main or only cause of the differences in health that they found. That said, there’s now a long list of research studies, mostly observational, showing associations and correlations between air pollution levels and the health of children and babies, and I for one would be very surprised if in the end it’s shown that air pollution plays little or no role in causing the health effects. But the details are still far from clear.
“Some of the statements about risk of death to babies, from the Kotecha et al. study, really do sound alarming – a 20-50% increased risk of death in the most polluted areas compared to the least polluted. That’s certainly worth taking into account – but these are relative measures (based on odds ratios), and they also don’t take into account the adjustment for potential confounders. The adjusted figures are given in the research abstract. The largest relative effect there is a 21% difference, for sulphur dioxide and neonatal deaths (that is, deaths of babies in their first 28 days of life). In the UK, such deaths are very uncommon these days, so even a 21% change does not correspond to large numbers of deaths. In England and Wales in 2016, just 27 out of every 10,000 babies, who were born alive, died in the first 28 days of life. Moving that figure by 21% would change the total by 5 or 6 deaths out of 10,000, if the 21% change applied to everyone – but it couldn’t, because most people don’t live in the areas of high pollution, so that the 21% difference doesn’t apply to them. (I could give more useful figures if I had the full research report.) These results are also subject to some statistical uncertainty, but again we don’t know how much because it isn’t stated in the brief reports. All the other adjusted associations were smaller, and some weren’t statistically significant, so that it’s plausible they are due to nothing more than random chance. I’m not saying that the results aren’t important, because you can’t avoid the pollution if you live in a high-pollution area, but it’s important to see them in the context of the very low rates of neonatal and infant mortality in this country.”
Dr Graham Wheeler, Senior Statistician, University College London, said:
“There is a large and growing body of evidence showing that exposure to air pollutants can have adverse effects on one’s health, and also reduce life expectancy in the long term. Whilst the conclusions of these studies are similar to previous research, these are not published results that have been peer-reviewed and should be treated with some caution.
“For example, in the study on infant mortality, no margins of error have been given around these estimates, so we don’t know how precise the results given actually are.
“Also, no absolute numbers have been given, so we don’t know whether a 20% increase in the risk of death translates to a small or large increase in the actual number of deaths.
“Furthermore, after adjusting for factors like birthweight, deprivation score and maternal age, the reported increases in risk of death drop by a huge amount. It may be that other factors that were not accounted for in this work may play some part in further explaining the differences in infant mortality.
“As the authors state, understanding the biological mechanisms through which air pollution affects foetal health and infant mortality will help determine the most appropriate strategies to ensure everyone has access to clean air.
“Please note: I’m not a pollution/environmental health expert but am mainly commenting on the stats.”
Dr Penny Woods, Chief Executive, British Lung Foundation, said:
“These two studies add to the ever-growing body of evidence showing the terrible harm air pollution is doing to our country’s lungs.
“Dr Hansell’s study mirrors others showing air pollution can stunt the growth of children’s lungs – leaving them at risk of a lifetime of ill health.
“The evidence linking infant mortality and air pollution is still emerging but Dr Kotecha’s study, which suggests babies born in the UK’s most-polluted areas see a significantly increased risk of death, should be a wake-up call to government.
“Both these studies concentrate on the damage done by air pollution to some of the most vulnerable members of our society – the very young.
“Our children can’t protect their own lungs. There is limited action the public can take to reduce their family’s exposure to air pollution. The government must do their job and take action to tackle this public health crisis.
“This would include writing World Health Organisation limits on pollutants into UK law through a new Environment Act.
“And local authorities in some of the most-polluted cities, such as Manchester, Liverpool and Birmingham, must be supported to bring in charging Clean Air Zones which charge polluting cars. These will help clean up the country’s toxic air in the shortest time possible and better protect everyone’s lungs.”
‘Effects of air pollution on all cause neonatal and post-neonatal mortality: population based study’ by Sarah Kotecha et al and ‘Prenatal, early-life and childhood exposure to air pollution and lung function in the UK Avon Longitudinal Study of Parents and Children (ALSPAC) cohort’, by Anna Hansell et al will be presented at the European Respiratory Society International Congress on Sunday 29th September.
There is no paper as this is not published work.
Prof Kevin McConway: Prof McConway is a member of the SMC Advisory Committee, but his quote above is in his capacity as a professional statistician.
Dr Graham Wheeler: “I am employed by UCL, am a Fellow, Chartered Statistician and Statistical Ambassador of the Royal Statistical Society, and a voluntary research committee member for Chiltern Music Therapy, a not-for-profit organisation providing music therapy services. I have previously received honoraria from Novametrics Consulting Ltd.”
Dr Penny Woods: “None.”