Projecting future temperature-related mortality using annual time series data: An example from Hong Kong

Background: Previous studies projecting future temperature-related mortality under climate change have mostly used short-term temperature-mortality associations based on daily time series data. The present study aimed to project mortality under different Representative Concentration Pathways (RCPs) in 21st century in Hong Kong by using analysis of annual data during 1976-2018. Methods: We employed a degree-days approach, calculating the sum of daily degrees above or below certain temperature threshold within a relevant historical year. The yearly age-standardized mortality rates (ASMRs) were regressed on annual hot and cold degree-days in quasi-Poisson generalized additive models to assess the exposure-response function that was subsequently used to calculate future changes in ASMR. The projection was performed without and with certain human adaptation assumed. Results: ASMRs were projected to have net increases under RCPs 4.5, 6.0, and 8.5, with increased mortality attributable to excess hot days exceeding decreases attributable to excess cold days. The average net changes under RCP8.5 was estimated to be 0.12%, 12.44%, 38.99%, and 89.25% during 2030s, 2050s, 2070s, and 2090s, respectively. Higher projected ASMRs were estimated for those aged over 75 years and for cardiovascular deaths. When human adaptation was considered, slope reduction alone under RCP4.5 and 6.0 and all adaptation assumptions under RCP8.5 might still not offset its corresponding adverse impact. Conclusions: The projected decreases in cold-related mortality do not compensate for projected increases in heat-related mortality in Hong Kong. Better public adaptations strategies are warranted for coping with the adverse health impacts of climate change on a local scale.