The effect modification of greenspace and impervious surface on the heat-mortality association: Differences by the dissimilarity index

Recent studies have shown that heat-mortality risk differs by level of greenspace and impervious surface. However, these studies do not consider both green spaces and impervious surfaces simultaneously, and further did not fully consider community- and individual-level characteristics. In this study we explored the effect modification of greenspace and impervious surface on the association between heat and mortality and how it differs by race/ethnicity dissimilarity index levels in North Carolina, USA. We aggregated datasets for greenspace, impervious surface estimates, temperature, and mortality for 1275 census tracts for North Carolina, USA, for 2000 to 2016 for 5 warm months (May to September). We used distributed lag non-linear models to estimate the heat-mortality relationship in each census tract. Heat-mortality relative risk (RR) was higher for census tracts with low greenspace than high greenspace (RR comparing risk at 99th temperature and minimum mortality temperature: 1.08 (1.02, 1.15) for low greenspace and 0.97 (0.87, 1.08) for high greenspace). Heat-mortality RR was higher for tracts with high impervious surface than low impervious surface (1.04 (1.00, 1.09) for high impervious surface and 0.94 (0.84, 1.05) for low impervious surface). Census tracts with high dissimilarity value and low greenspace had the highest heat-mortality risk compared to the tracts with high dissimilarity value with and high greenspace (1.13 (1.02, 1.24) for high dissimilarity index and 0.97 (0.86, 1.09) for low dissimilarity index). Communities with low greenspace or high impervious surfaces had higher heat-mortality associations, and this effect modification was higher for high race/ethnicity dissimilarity regions.