Residential greenness and allergic outcomes
In line with our observations, the reported associations for NDVI and asthma were heterogeneous and, in some cases, non-linear. Like in our study, medium but not high NDVI increased the risk of asthma in 7,040 Taiwanese children. 26 Also in line with us, exposure to medium but not high greenness was associated with asthma in the Euro-Siberian climatic region of a Spanish multicentre study of 2,472 children. 27 Nevertheless, high greenness was protective against wheezing in the same region of the same study.27 Higher NDVI in the 100 m buffer, but not in larger buffers, was related to increased asthma risk in 1,489 Kaunas children, and this association was stronger in children living further away from parks. 28 We observed opposite tendencies. Like our protective association between the high NDVI tertile and asthma, two other urban studies – one in 1,915 Chicago children29 and another one in 187 Turin children30 – reported protective greenness effects on asthma. Higher greenness was also related to lower asthma risk in a large New Zealand study of 49,956 children. 31 Nevertheless, an even larger study (n = 65,000) from the Vancouver Metropolitan Area failed to detect such an association. 32 Null findings between NDVI and asthma were also reported in 3,178 Spanish children33 and 5,643 children from Suzhou, China.34
We did not observe any association between NDVI and AR and the association between NDVI and aeroallergen sensitization was inconsistent. This is in line with the aforementioned studies by Tischer et al., 27 Dadvand et al. 33 and Li et al. 34 In addition, Gernes et al.35 reported null findings between NDVI and rhinitis and atopic sensitization in 425 children from the USA. Our two previous multi-centre studies, whose study populations were distinct from the one used in this analysis, reported differential associations with AR, nose and eye symptoms and aeroallergen sensitization, as is mentioned in the Introduction. 11,12 Interestingly, effect modification by moving did not reveal any consistent trends in Fuertes et al.,12 in line with our present analysis. Unlike our stratified results by air pollutants, no pattern was observed in analyses stratified by NO2 subgroups in these two studies. 11,12