The burden of ambient temperature on years of life lost: A multi-community analysis in Hubei, China

Background: Compared with death rates, years of life lost (YLL) has been widely used as a more informative indicator to quantify the burden of premature death. In the context of global climate change, existing evidence linking ambient temperatures and YLL was very scarce across the globe. Methods: Daily mortality and meteorological data during 2009-2012 were obtained from 12 communities across Hubei Province in central China. A two-stage approach was used for statistical analysis. At the first stage, a generalized linear regression combined with distributed lag non-linear model was applied to estimate community-specific temperature-YLL associations. A second-stage multivariable meta-analysis was then conducted to pool the community-specific estimates of temperature-related effects on YLL. Results: A pooled J- or U-shaped association was observed between ambient temperature and YLL due to different mortality categories. Heat effects occurred immediately and only persisted for several days, whereas cold effects were delayed and much longer-lasting. At the provincial level, heat effect (per 1 degrees C increase from 75th to 99th percentile of temperature) at lag 0-2 days and cold effect (per 1 degrees C decrease from 25th to 1st percentile of temperature) at lag 0-21 days was associated with an increase of 1.91% (95% CI: 0.83, 3.00) and 5.09% (2.79, 7.40) in YLL due to non-accidental deaths, respectively. Much greater effect estimates of cold than heat were also observed for other mortality-specific YLLs (except for respiratory mortality). Heat effects on YLL were higher for males and the youth, while cold effects were greater for females and the elderly. Additionally, relatively stronger associations between heat, cold and YLL were consistently observed in low-educated persons. Conclusions: This multi-community study strengthened the evidence that both cold and hot temperatures were associated with increased years of life lost. Our findings may have important implications for better understanding the burden of premature death related to temperature extremes. (C) 2017 Elsevier B.V. All rights reserved.