Projections of Temperature-Attributed Mortality under Climate Change Scenarios: an Analysis of 395 Locations in 15 Countries

Antonio Gasparrini,Yuming Guo,Francesco Sera,Swarna Khare,Clare Heaviside,Aurelio Tobias,Masahiro Hashizume,Eric Lavigne,Antonella Zanobetti,Joel Schwartz,Daniel Oudin Astrom,Bertil Forsberg,Paola Michelozzi,Matteo Scortichini,Xerxes Seposo,Yue-Liang Leon Guo,Chang-fu Wu,Haidong Kan,Tran Ngoc Dang,Do Van Dung,Micheline de Sousa Zanotti Stagliorio Coelho,Paulo Hilario Nascimento Saldiva,Shilu Tong,Yasushi Honda,Ho Kim,Sotiris Vardoulakis,Shakoor Hajat,Ben Armstrong

DOI: https://doi.org/10.1289/isee.2016.3831

2016-01-01

Abstract:Introduction Several studies provide future projections of the temperature-related mortality in single cities or countries, but few offer a systematic assessment across climates and populations using flexible models separating heat and cold contributions. Methods Two-stage analysis using time series data from 395 locations in 15 countries (Australia, Brazil, Canada, China, Italy, Japan, Philippines, South Korea, Spain, Sweden, Taiwan, Thailand, UK, USA, Vietnam) in 1985-2012. Temperature-mortality associations were estimated using a distributed lag non-linear model with 21 days of lag. Attributable risk was computed for non-optimal temperature and for cold and heat separately. The impact was determined using historical data, and, assuming no adaptive or population changes, for 2020s and 2050s using daily time series projected in low and high emission scenarios by 5 climate models. Results On average across scenarios, cold-attributable mortality would decrease from 7.11% in 1985-2012 to 5.93% in 2020s and to 5.13% in 2050s. Heat would increase from 0.40% to 0.94% and to 1.84% in the same periods. These figures mask differences across populations. Countries in temperate climates would experience a strong reduction in cold-attributed deaths (from 7.46% in 1985-2012 to 5.74% in 2050s), counterbalanced by an increase due to heat (from 0.37% to 1.49%). In (sub)tropical countries, reduction due to cold (from 3.04% to 1.49%) would be exceeded by increase in heat-attributed deaths (from 0.75% to 3.94%). Conclusions This large multi-country epidemiological analysis indicates that climate change can alter temperature-related mortality substantially, with differential patterns between cold and heat. Without adaptive response, (sub)tropical countries would experience a much higher heat-attributed mortality, while the decrease in cold-attributed deaths would have a large impact in temperate climates. The analysis will be extended to account for adaptation and demographic changes.