Non-optimal temperature-attributable mortality and morbidity burden by cause, age and sex under climate and population change scenarios: a nationwide modelling study in Japan

Abstract

Background: Future temperature effects on mortality and morbidity may differ. However, studies comparing projected future temperature-attributable mortality and morbidity in the same setting are limited. Moreover, these studies did not consider future population change, human adaptation, and the variations in subpopulation susceptibility. Thus, we simultaneously projected the temperature-related mortality and morbidity by cause, age, and sex under population change, and human adaptation scenarios in Japan, a super-ageing society.

Methods: We used daily mean temperatures, mortality, and emergency ambulance dispatch (a sensitive indicator for morbidity) in 47 prefectures of Japan from 2015 to 2019 as the reference for future projections. Future mortality and morbidity were generated at prefecture level using four shared socioeconomic pathway (SSP) scenarios considering population changes. We calculated future temperature-related mortality and morbidity by combining baseline values with future temperatures and existing temperature risk functions by cause (all-cause, circulatory, respiratory), age (<65 years, ≥65 years), and sex under various climate change and SSP scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5). Full human adaptation was simulated based on empirical evidence using a fixed percentile of minimum mortality or morbidity temperature (MMT), while no adaptation was simulated with a fixed absolute MMT.

Findings: A future temporal decline in mortality burden attributable to non-optimal temperatures was observed, driven by greater cold-related deaths than heat-related deaths. In contrast, temperature-related morbidity increased over time, which was primarily driven by heat. In the 2050s and 2090s, under a moderate scenario, there are 83.69 (95% empirical confidence interval [eCI] 38.32-124.97) and 77.31 (95% eCI 36.84-114.47) all-cause deaths per 100,000 population, while there are 345.07 (95% eCI 258.31-438.66) and 379.62 (95% eCI 271.45-509.05) all-cause morbidity associated with non-optimal temperatures. These trends were largely consistent across causes, age, and sex groups. Future heat-attributable health burden is projected to increase substantially, with spatiotemporal variations and is particularly pronounced among individuals ≥65 y and males. Full human adaptation could yield a decreasing temperature-attributable mortality and morbidity in line with a decreasing population.

Interpretation: Our findings could support the development of targeted mitigation and adaptation strategies to address future heat-related impacts effectively. This includes improved healthcare allocations for ambulance dispatch and hospital preventive measures during heat periods, particularly custom-tailored to address specific health outcomes and vulnerable subpopulations.

Funding: Japan Science and Technology Agency and Environmental Restoration and Conservation Agency and Ministry of the Environment of Japan.

Keywords: Attributable risk; Climate change; Morbidity; Mortality; Non-optimal temperature; Population ageing.

Fig. 1

Temporal trends in projected temperature increase by climate change scenarios and projected population change by shared socio-economic pathway (SSP), relative to baseline period (2010–19) in Japan. Please note that the scale of each figure is distinct. Corresponding numeric data are presented inTables S3 and S5.

Fig. 2

Temporal trends in non-optimal temperature-attributable excess mortality rate (per 100,000 population) by cause and age groups under alternative climate and population change scenarios during future period. Please note that the scale of each figure is distinct. The vertical lines are 95% empirical confidence intervals (eCIs) for the total temperature-attributable mortality rate. Corresponding numeric data are presented inTable S10.

Fig. 3

Temporal trends in non-optimal temperature-attributable morbidity rate (per 100,000 population) by cause and age groups under alternative climate and population change scenarios during future period. Please note that the scale of each figure is distinct. The vertical lines are 95% empirical confidence intervals (eCIs) for the total temperature-attributable morbidity rate. Corresponding numeric data are presented inTable S13.

Fig. 4

Decadal trends of heat-attributable morbidity and morbidity rate (per 100,000 population) by cause and age groups at country level under alternative climate and population change scenarios. Please note that the scale of each figure is distinct. Corresponding numeric data are presented inTables S10 and S13.

Fig. 5

Fold changes in point estimates of heat-attributable mortality and morbidity rate by causes in general population for 2090–99 compared to 2010–19 under SSP2-4.5 and corresponding population scenario for 47 prefectures in Japan. Please note that the scale of each map is distinct. Corresponding numeric data are presented inTables S14–S16.