Do greenspaces really reduce heat health impacts? Evidence for different vegetation types and distance-based greenspace exposure

Song, Jinglu; Gasparrini, Antoni; Wei, Di; Lu, Yi; Hu, Kejia; Fischer, Thomas B.; Nieuwenhuijsen, Mark J.

doi:http://dx.doi.org/10.1016/j.envint.2024.108950

Do greenspaces really reduce heat health impacts? Evidence for different vegetation types and distance-based greenspace exposure

License

Citation

Song J, Gasparrini A, Wei D, Lu Y, Hu K, Fischer TB, et al. Do greenspaces really reduce heat health impacts? Evidence for different vegetation types and distance-based greenspace exposure. Environ Int. 2024 Sep;191:108950. DOI: 10.1016/j.envint.2024.108950

Abstract

Background: While vegetation type, population density and proximity to greenspaces have been linked to human health, what type and location of greenspace matter most have remained unclear. In this context, there are question marks over investment-style metrics. Objectives: This paper aims at establishing what vegetation type may matter most in modifying heat-mortality associations, and what the optimal buffer distances of total and specific types of greenspace exposure associated with reduced heat-related mortality risks are. Methods: We conducted small-area analyses using daily mortality data for 286 Territory Planning Units (TPUs) across Hong Kong and 1 × 1 km gridded air temperature data for the summer months (2005-2018). Using a case time series design, we examined effect modifications of total and specific types of greenspaces, as well as population-weighted exposure at varying buffer distances (200-4000 m). We tested the significance of effect modifications by comparing relative risks (RRs) between the lowest and highest quartiles of each greenspace exposure metric; and explored the strength of effect modifications by calculating the ratio of RRs. Results: Forests, unlike grasslands, showed significant effect modifications on heat-mortality associations, with RRs rising from 0.98 (95 %CI: 0.92,1.05) to 1.06 (1.03, 1.10) for the highest to lowest quartiles (p-value = 0.037) The optimal distances associated with the most apparent effects were around 1 km for population-weighted exposure, with the ratio of RRs being 1.424 (1.038,1.954) for NDVI, 1.191 (1.004,1.413) for total greenspace, and 1.227 (1.024,1.470) for forests. A marked difference was observed in terms of the paired area-level and optimal distance-based exposure to total greenspace and forests under extreme heat (p-values < 0.05). Discussion: Our findings suggest that greenspace, particularly nearby forests, may significantly mitigate heat-related mortality risks.