Deaths Attributable to Anomalous Temperature: A Generalizable Metric for the Health Impact of Global Warming
Hengyi Liu,Mingkun Tong,Fuyu Guo,Qiyue Nie,Jiwei Li,Pengfei Li,Tong Zhu,Tao Xue
DOI: https://doi.org/10.1016/j.envint.2022.107520
IF: 11.8
2022-01-01
Environment International
Abstract:The U-shaped association between health outcomes and ambient temperatures has been extensively investigated. However, such analyses cannot fully estimate the mortality burden of climate change because the features of the association (e.g., minimum mortality temperature) vary with human adaptation; thus, they are not generalizable to different locations. In this study, we assumed that humans could adapt to regular temperature variations; and thus examined the all-cause mortality attributable to temperature anomaly (TA), an indicator widely utilized in climate science to measure irregular temperature fluctuations, across 115 cities in the United States (US). We first used quasi-Poisson regressions to obtain the city-specific TA-mortality associations, then used meta-regression to pool these city-specific estimates. Finally, we calculated the number of TA-related deaths using the uniform pooled association, then compared it to the estimates from city-specific associations, which had been controlled for adaptation. Meta-regression showed a U-shaped TA-mortality association, centered at a TA near 0. According to the pooled association, 0.579 % (95 % confidence interval [CI]: 0.465–0.681 %), 0.394 % (95 % CI: 0.332–0.451 %), and 0.185 % (95 % CI: 0.107–0.254 %) of all-cause deaths were attributable to all anomalous temperatures (TA ≠ 0), anomalous heat (TA > 0), and anomalous cold (TA < 0), respectively. At the city level, heat-related deaths estimated from the pooled association were in good agreement with heat-related deaths estimated from the city-specific associations (R2 = 0.84). However, the cold-related deaths estimated from the two methods showed a weaker correlation (R2 = 0.07). Our findings suggest that TA constitutes a generalizable indicator that can uniformly evaluate deaths attributable to anomalous heat in distinct geographical locations.