Impact of London's road traffic air and noise pollution on birth weight: retrospective population based cohort studyBMJ 2017; 359 doi: https://doi.org/10.1136/bmj.j5299 (Published 05 December 2017) Cite this as: BMJ 2017;359:j5299
All rapid responses
Re: Impact of London's road traffic air and noise pollution on birth weight: retrospective population based cohort study
We would like to commend Smith et al. (2017) on an insightful study into the impacts of air and noise pollution on foetal growth. This study adds a new perspective which can be added to the emerging evidence base around the impacts of urban environments on public health. We have critically appraised the study and would like to draw attention to some areas that we feel would benefit from some clarity.
It can be acknowledged that the authors attempted to correct for confounders. However, birth weight is complex and has a wide range of determinates that were not adequately considered. The causation of those confounders that were measured will also be assessed using the Bradford-Hill guidelines for confounding(Bruce et al. 2009).
In addition to the confounders that were modelled in the study, the other key determinates that are known to impact on birth weight include: previous stillbirth, previous small for gestational age (SGA) baby, vigorous daily exercise, maternal or paternal SGA, chronic hypertension, diabetes, renal impairment, antiphospholipid syndrome and recreational drug use(WHO 2014; RCOG 2014). We suspect that there were issues around cost and ethics, which meant that these could not be measured directly. Moreover, we acknowledge that deprivation measures may have been used as a proxy to account for many of the confounders not directly measured. Unfortunately, it cannot accurately account for all of them.
Furthermore, the measures that were used may not accurately determine exposure to confounders. For example, the use of census data as a measure for tobacco consumption unlikely to accurately determine smoking rates during pregnancy, as the data was taken in 2011, after the study period, and used tobacco expenditure as a measure of consumption rather than actual smoking rates during pregnancy. When all these factors are taken into consideration, it further undermines our confidence in the strength of the conclusions that can be drawn from the measured outcomes.
When considering causation using the Bradford-Hill guidelines, it can be confidently assumed that the exposure of air and noise pollution preceded the onset of disease, as these were measured during the pregnancy and the outcome of these exposures were measured at birth.
However, the strength of association was less certain. We were less confident than the authors about the impact of the air pollutant exposures on foetal growth. Smith et al. have attributed exposure to some air pollutants (NO2, NOx, PM2.5 traffic exhaust, PM2.5 traffic non-exhaust, PM2.5, and PM10) during pregnancy to have significantly affected low birth weight. When considering the odds of low birth weight after exposure to each air pollutant the odds ratios were found to be narrow and at or close to 1.00(Smith et al. 2017, p.13). With such a large cohort population we would expect a more statistically significant result to be confident of the effect.
As result of the weak strength of association and the incomplete multiple regression analysis, we wondered whether there was actual independence of confounding. It was stated in the study that the authors did not adjust for multiple testing, but there was not a clear explanation for why this was the case. Therefore, we are not clear whether this could increase the chance of Type I error.
We are encouraged that the results of this study are consistent with a number of international studies, which highlights that it is more likely that a genuine causal association exists, even in different populations. Having said this, this study only focuses on a population with a relatively high exposure to air and noise pollution. It would have been useful to have compared the results to areas of low levels of exposure to establish whether there would be a correspondingly low rate of foetal growth abnormalities in areas of low air and noise pollution; this would also help to understand if there is indeed a dose-response relationship.
Lastly, although there is emerging evidence for a biological model and some animal model studies, it is still not clear how air pollution may cause foetal growth abnormalities(Slama et al. 2008; Clemente et al. 2015; van den Hooven et al. 2012; Veras et al. 2008). Further research in this area would help corroborate the causation found in population studies.
As a result of these observations, we suggest that a small prospective cohort study of a smaller population may be able to add confidence to this study. Using smaller numbers would allow for more in-depth data collection, perhaps using medical records during pregnancy to better control for confounders. Moreover, a series of studies could be undertaken collaboratively, in a number of urban and rural areas around the UK to get a more representative sample for the country as a whole and to try and establish a dose-response relationship.
Bruce, N., Pope, D. and Stanistreet, D. (2009). Cohort studies. In Quantitative methods for health research: a practical interactive guide to epidemiology and statistics. Chichester: Wiley, pp. 193–256.
Clemente, D.B.P. et al. (2015). Prenatal Ambient Air Pollution, Placental Mitochondrial DNA Content, and Birth Weight in the INMA (Spain) and ENVIRONAGE (Belgium) Birth Cohorts. Environmental Health Perspectives, 124(5). [online]. Available from: http://ehp.niehs.nih.gov/1408981 [Accessed March 7, 2018].
van den Hooven, E.H. et al. (2012). Air Pollution Exposure and Markers of Placental Growth and Function: The Generation R Study. Environmental Health Perspectives. [online]. Available from: http://ehp.niehs.nih.gov/1204918 [Accessed March 7, 2018].
RCOG. (2014). The Investigation and Management of the Small–for–Gestational–Age Fetus. 27 Sussex Pl, Marylebone, London NW1 4RG: Royal College of Obstetricians and Gynaecologists. [online]. Available from: https://www.rcog.org.uk/globalassets/documents/guidelines/gtg_31.pdf [Accessed May 2, 2018].
Slama, R. et al. (2008). Meeting Report: Atmospheric Pollution and Human Reproduction. Environmental Health Perspectives, 116(6), pp.791–798.
Smith, R.B. et al. (2017). Impact of London’s road traffic air and noise pollution on birth weight: retrospective population based cohort study. BMJ, p.j5299.
Veras, M.M. et al. (2008). Particulate Urban Air Pollution Affects the Functional Morphology of Mouse Placenta1. Biology of Reproduction, 79(3), pp.578–584.
WHO. (2014). Low Birth Weight Policy Brief. Avenue Appia 20, CH-1211 Geneva 27, Switzerland: Department of Nutrition for Health and Development, World Health Organisation. [online]. Available from: http://apps.who.int/iris/bitstream/10665/149020/2/WHO_NMH_NHD_14.5_eng.p... [Accessed May 2, 2018].
Competing interests: No competing interests
Smith el al must be commended for clearly demonstrating that air pollution from road traffic in London adversely affects fetal growth.(1) The study adjusted for confounding variables and the estimate of a 3% population attributable factor (the proportional reduction in population morbidity that would occur if exposure was reduced to zero) is merely the tip of the iceberg in terms of public health challenges. Indeed, the most frequent avoidable causes of low fetal growth and preterm birth are associated with the social gradient and they interact: smoking, pollution, poor education, poor nutrition, poor housing, single motherhood… and more - the list goes on. Accordingly, in such a quagmire, no adjustment method can reliably calculate the population attributable factor for each cause separately, and what would be the point? The response must be comprehensive but Marmot’s salient call to reduce unfair distribution of resources and conditions of everyday life continues to remain almost ignored.(2)
In the UK, as in France, around 60,000 babies are born prematurely each year as over 1 in 10 pregnancies ends in preterm birth: half after spontaneous onset of labour and half after provider-initiated intervention. This is a major issue as gestational age is highly associated with neonatal mortality and with short- and long-term morbidities. Healthcare professionals must apply good evidence when intervening in individual cases, but also lobby for universal social justice. Who better to advocate for the future children destined for poor health from before their births? Otherwise, like the daughters of Danaus, they will be condemned for eternity to fill an ever-draining barrel with water carried in sieves.(3).
1 Smith RS, Fecht D, Gulliver J et al. Impact of London's road traffic air and noise pollution on birth weight: retrospective population based cohort study. BMJ 2017;359:j5299
2 Marmot M. Commission on Social Determinants of Health. Achieving health equity: from root causes to fair outcomes. Lancet 2007;370:1153-63.
3 Daughters of Danaus. Available at https://en.wikipedia.org/wiki/Daughters_of_Danaus
Competing interests: No competing interests