Associations between green space,air pollution and birthweight in Sydney Metropolitan Area,Australia

Growing number of evidence have reported that exposure to air pollution was associated with unfavourable birth outcomes while increased exposure to green spaces was associated with better birth outcomes. However, the effect of interactions between air pollution and green spaces on pregnancy outcomes remain unclear. Using the data on all the live births recorded in Sydney between 2016 January and 2017 December, we built up multilevel linear and logistic regression models with random intercepts for statistical area level 2 (SA2) to examine the association between residential levels of air pollution (NO2, PM2.5, SO2 and O3) and pregnancy outcomes including birthweight, low birthweight, and preterm birth. This was followed by assessment of potential effect modification by green space quantity by fitting 2-way interaction term between each air pollutants and green space quantity separately. Furthermore, building on the 2-way interaction term, we explored 3-way interactions by adding area level socio-economic status and population density. Higher levels of PM2.5, NO2 and SO2 were statistically significantly associated with decrease in birthweights (p < 0.05) in the adjusted models. We observed statistically significantly association between the lower risks of preterm birth and higher levels of NO2 in the adjusted models (p < 0.05). However, none of the other air pollutants were statistically significantly associated with the odds of low birthweight and preterm birth. In the adjusted models, green space quantity was statistically significantly associated with reduced odds of preterm birth (p < 0.05). The results for birthweights were in the same direction (p < 0.05) however, some of these associations did not remain statistically significant after adjusting the models for air pollution. In general, no associations were found for low birthweight. Statistically significant 2-way interactions between green space quantity and NO2, PM2.5 and SO2 levels indicated that association between moderate levels of NO2, PM2.5 and birthweight weakened in the greener areas (>20 %) especially in the affluent and densely populated areas while the effect of SO2 did not differ consistently across green space levels with benefit of exposure to the greenest areas (>40 %) was evident, yet not statistically significant. Our findings suggest that increasing green spaces in cities may help supporting a healthy start in life by decreasing harms of moderate levels of air pollution. Replication in different contexts and consideration of potentially contrasting results with different types of green space is warranted.