Assessing the role of socio-economic factors in shaping the temperature-mortality exposure-response relationship in China

Non-optimal temperatures significantly influence public health. However, the role of socio-economic factors in modulating health risks associated with non-optimal temperatures varies geographically and among different populations. Thus, the meteorological, air quality, health data, and socio-economic indicators were obtained from 23 districts in North and 48 districts in East China, respectively. Employing a two-stage meta-analysis, the exposure-response relationship was constructed for temperature against mortality from non-accidental causes, cardiovascular and cerebrovascular diseases, and respiratory illnesses. Furthermore, a non-linear spline regression was applied to assess the impact of socio-economic indicators on the exposure-response relationship and predicted future risks under various Shared Socioeconomic Pathways. The results revealed that the influence of socio-economic factors on the exposure-response curve showed heterogeneity in East China and North China. In North China, the shape of the exposure-response curve changed greatly under different socio-economic levels, while it remained similar in East China. In East China, the relative risk of heat and cold exposure was reduced in regions with high GDP, high levels of public finance, good medical services, and a low proportion of the elderly population. Specifically, the risk of non-accidental deaths due to heat shows a nearly linear negative correlation with per capita GDP in East China, with a decrease of the relative risk by 0.075 for every 10 thousand yuan increase in per capita GDP. Future projections indicate that population aging plays a decisive role in shaping the exposure-response curves. Although economic growth can reduce the risk of heat-related mortality, the combined effect of population aging and economic increase results in steeper exposure-response curves in both hot and cold temperature ranges in the future. In conclusion, although spatial variations in relative risk changes still exist, enhancing the adaptive capacity of populations can mitigate health risks associated with future climate change.


Introduction
The report of the Sixth Session of the Intergovernmental Panel on Climate Change states that with further global warming, every region is projected to increasingly experience a change in climatic impact-drivers of temperature extremes (high confidence) (IPCC 2023).Numerous epidemiological studies have revealed that both high and low temperatures are associated with an increased risk of mortality in populations (WHO 2020, Shao et al 2021, Ribas et al 2023).Climate change has significant impacts on human health (Yin et al 2019, Yang et al 2021, Wang et al 2021a, Sun et al 2022, Yan et al 2022).Thus, accurately estimating the health risks for populations under future climate change is highly important for developing effective mitigation and adaptation strategies for climate change.
The key to estimating health risks under future climate change lies in establishing the exposureresponse relationship between temperature and health outcomes (Vicedo-cabrera et al 2019, Sun et al 2021).Environmental epidemiologists have extensively established the relationship between nonoptimal temperature exposure and various health outcomes at regional levels (Burkart et al 2021, Li et al 2021, He et al 2022, LIU et al 2022b, Bai et al 2023, Momtazmanesh et al 2023).It is worth noting that each geographical area exhibits its distinctive and individualized exposure-response curve between temperature and mortality (Zeng et al 2016).In addition, under different Shared Socio-economic Pathways (SSPs) and climate change scenarios in the future, socio-economic level development and urbanization level will be different (IPCC 2023).These differences can potentially modify the original relationship between temperature and mortality (Hu et al 2022, Zhang et al 2023a).To accurately assess the health impacts of climate change in future scenarios, it becomes imperative to consider not only historical exposure-response relationships but also the influence of geographical and socio-economic factors (Lay et al 2021).It is also worth mentioning that socioeconomic factors are expected to become the most influential factors shaping the changes in exposureresponse relationships under climate change.
However, the geographical and populationspecific variations in the role of socio-economic factors in modulating health risks associated with non-optimal temperatures have yet to be fully explored (Milojevic et al 2016, Xing et al 2020, Liu et al 2022a, Sun et al 2024).The modification effect of socio-economic factors exhibits spatial heterogeneity, which can be observed in two aspects (Huang et al 2014, Huang et al 2015, Gosling et al 2017).Firstly, the socio-economic factors affecting the relative risk of non-optimal temperature vary across regions, categorized as global and regional variables.Global variables significantly influence the temperature-death exposure-response relationship in all regions, while regional variables only impact it in certain regions.Secondly, the intensity and degree of changes in urbanization levels have varying impacts on the relative risk of non-optimal temperatures.A few studies have explored the spatial heterogeneity of the relationship between socio-economic factors and non-optimal relative risk (Janko et al 2019).For instance, it revealed that age, education level, and socio-economic status were global variables affecting heat-related deaths, while thermal environment, income level, vegetation coverage, and population types are regional variables, and their influence on adverse temperature death is different across different regions (Song et al 2021, Wang et al 2021c).Based on the analysis of Chinese district and county data, it was also found that under different urbanization levels, the importance of the heat vulnerability index is different (Wang et al 2021b).
China is faced with significant health risks from climate change, with varying socio-economic levels among different population groups.These disparities contribute to differences in adaptive capacity and health risks of the population.Therefore, this study selected two typical regions in China (East China and North China), representing different climate zones.By analyzing the non-linear relationship between the exposure-response relationship at county levels and socio-economic factors, this study aims to explore the impact of population, aging, medical services, and economy on the non-optimal temperature exposure-response relationship in climate change in the two regions, and to complete the estimates of the exposure-response relationships.The results provide valuable data support to project the estimation of the health impacts of climate change and serve as a basis for formulating future urban adaptation strategies.

Data sources
Two regions were selected as study areas (figure S1 in the supplementary material), including 23 and 48 districts/counties in North China and East China, respectively.Meteorological data, including daily average temperatures and relative humidity, were obtained from the weather stations in each district/county (table S1 in supplementary material).Daily mean air pollutants, including fine particulate matter (PM 2.5 ), coarse particulate matter (PM 10 ), sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), carbon monoxide (CO), and ozone (O 3 ) were obtained from China National Environmental Monitoring Centre (table S1).Daily non-accidental mortality (code: A00-R99), cardiovascular and cerebrovascular diseases mortality (code: I00-R99), and respiratory disease mortality (code: J00-R99) from 1 January 2010 to 31 December 2016 were obtained from the Chinese Center for Disease Control and Prevention.Socioeconomic data, including the population, the proportion of the urban population, the proportion of the elderly population (aged above 60 years), the number of beds in health care facilities, industrial output, public finance revenue, gross regional product, number of middle school students for each district/county were obtained from the local statistical yearbook and the 6th China national population census.

Establishment of the pooled exposure relationship between temperature and mortality in East and North China
A two-stage fitting method was employed to investigate the exposure-response relationships in North China and East China (Gasparrini et al 2012, Gasparrini and Armstrong 2013, Yang et al 2021).In the first stage, a distributed lag model was utilized to establish the relationship between daily mean temperature exposure and daily mortality in different counties.
where Y i,t is the observed number of non-accidental deaths, cardiovascular and cerebrovascular diseases, and respiratory disease mortality deaths at district i on day t (t = 1, 2, …, 2557); α is the model intercept; s(t; β) represents a natural cubic spline with seven degrees of freedom (df) per year to adjust for seasonality and the time trend of mortality.The day of the week (DOW) was included as a categorical variable.A cross-basis term f (T obs ; θ) generated by DLNM was used to capture the non-linear and delayed effects of temperature on mortality.The 14 day lag term was fitted with an equally-spaced log-values function with 3 df for a natural cubic spline.The temperature term was defined based on the 10th, 75th, and 90th percentiles with 3 df for a natural cubic spline.Based on previous studies and the AIC criterion, the parameters were determined (Yang et al 2021).The center temperature was determined by identifying the temperature associated with the minimum relative risk in the exposure-response curve.Notably, air pollutants' data prior to 2013 were not available, hence, they were not accounted for in the final model.However, subsequent testing revealed that air pollutants had minimal impact on the exposure-response relationship between temperature and mortality.
In the second stage, the regression coefficients θi and associated estimated covariances matrices S i obtained from the first stage were utilized as dependent variables.A model for random-effect multivariate meta-analysis can be written as follows: where, φ is the unknown between-study covariance matrix; θ can be interpreted as the populationaverage outcome parameters, defining the pooled exposure-response association in North China and East China, respectively.

Identification of the effects of socio-economic demographic factors on non-optimal temperature risk
Based on the regression coefficients θi and associated estimated covariances matrices S i , a meta-analysis was conducted to establish a non-linear response relationship between the θ values and socio-economic factors.
where µ i represents the random effects, ) means that the relationship between θ i and socio-economic factors was modeled using a natural cubic spline with 4 df.Thus, the exposure-response relationship under different socio-economic levels in the two regions was estimated.Based on the exposure-response curves derived for different socio-economic demographic levels, the average relative risks in the high-temperature and low-temperature segments were determined.Specifically, the exposure-response relationship between temperature and total non-accidental mortality was predicted at various socio-economic levels, ranging from the 5th to 100th percentiles, using a meta non-linear fitting model.Subsequently, based on the minimum relative risk observed in North and East China, the temperature range was divided into high and low segments.Finally, utilizing the exposure-response curve, relative risks were calculated for different temperature values with an interval of 0.1 • C. The average relative risks in the hightemperature and low-temperature segments were then obtained.

Prediction of exposure-response curves under future scenarios
Under the SSPs global framework, along with the Cobb-Douglas production model, county-level economic statistics were predicted (Jing et al 2022).
Based on the data from the Sixth National Population Census as a reference, the age structure of the population in future scenarios was calculated using demographic parameters and migration (Zhang et al 2023b), and then the degree of population aging (the proportion of people aged 60 and above) was obtained based on the dataset.Based on the nonlinear relationship between per capita GDP levels and exposure response, the shape of future exposure response under scenarios where only GDP levels vary was initially projected.Subsequently, by incorporating population aging data, the shape of future exposure response under different SSP scenarios that account for both population aging and GDP level enhancement was derived.This discrepancy can be attributed to multiple factors, such as the increase in public awareness and preparedness, improvements in human adaptability, and advancements in living environment conditions.This phenomenon has also been revealed by previous studies.For instance, a time-series data analysis study by Wang et al (2023) showed that there was a reduced susceptibility for cold spells both in the temperate and cold climatic zone.On one hand, people living in areas with more cold spells and hot wave events, which might help the residents establish physiological acclimatization and emotional adaptation to cold spells and hot waves.On the other hand, the increased adaption capacity to extreme low and high temperatures was also owing to the increased improved living conditions (CNBS 2020).There is a higher prevalence of indoor heating in North China than in East China.In rural areas, individuals in North China have the option to use coal, gas, and electric heating equipment, while urban areas have established central heating systems.Although the impact of air conditioning on human health is controversial (Cao et al 2013), the higher prevalence of air conditioning usage among individuals in East China than that in North China might also reduce the mortality during high temperatures to some extent.The disparity in relative risk between the two regions further confirms that longterm acclimation enables individuals to reduce their relative risk of exposure to non-optimal temperatures through adaptive measures.

Impact of socio-economic factors on the exposure-response relationship between temperature and mortality in the two regions
In this study, a meta-analysis and spline fitting were conducted to examine the non-linear relationship between exposure-response curves and socioeconomic and geographical factors in East China and North China.As is shown in figures 1 and 2, there exist distinct variations in the impact of these factors on the exposure-response relationship between East China and North China.The exposure-response curve in North China exhibits a varying shape with the alteration of socio-economic factors, and anomalously low values are observed in the low temperature range at certain factor levels.Conversely, the shape of the exposure-response curve in East China remains unchanged with variations in social and economic factors, consistently displaying a U-shape.It shows that in North China, the relative risk occurred at an unusually low value when temperature was below 0 • C in areas in which the urbanization, the per capita GDP, the number of beds, and public finance revenue were low, showing a decreasing trend with the decrease in temperature.This phenomenon may be attributed to the relatively stronger adaptability of the North China region to cold temperatures, while also indicating the presence of imbalanced development in climate adaptability across different regions (Anderson and Bell 2009, Ng et al 2014, Jingesi et al 2023).People in North China exhibit the habit of heating in winter (Zheng and Bu 2018), which causes exposure misclassification and increases the uncertainty in relative risk identification.Meanwhile, the sample size was relatively small when constructing the exposure relationship with county data, which further led to a wider confidence interval in the exposure curve.
A certain regularity was observed in the change of exposure-response curves as they vary with the levels of socio-economic factors in East China.As shown in figure 3 and S3, most socio-economic factors, including the urbanization level, the number of beds per capita in health care facilities, the public finance per capita, and the proportion of young people had similar influences on the average relative risk of low and high temperature in East China.The influence of the urbanization level on the average relative risk shows similar non-linear characteristics during both cold and hot temperatures, with the relative risk firstly increasing and then decreasing with the increase in the urbanization level.The highest relative risk occurs at the urbanization rate of 50% ∼ 60%.This might be attributed to the development characteristic of urbanization.At the early stage of urbanization, rapid economic development and poor healthcare facilities were not conducive to people's resistance to adverse weather.As urbanization levels continue to rise, there has been an increase in the installation rate of heating equipment and air conditioning in urban areas.This, coupled with heightened public awareness of risks, has resulted in a reduction in people's exposure risks and improvements in healthcare.Consequently, the relative risk of cold and hot temperatures on inhabitants has decreased.
With the increase in the population, the relative risks of cold and hot temperatures in East China showed a similar trend of increasing and then decreasing.This trend is in line with the urbanization rate, illustrating the change in the ability of population aggregation to protect humans from adverse climatic conditions in the process of urbanization.The increased number of beds per capita in healthcare facilities was associated with a decrease in the relative risk of the population during both cold and hot temperatures.This demonstrates that the improvement in medical facilities is the key to improving the ability of the population to withstand adverse weather conditions (Kruk et al 2018, Chang et al 2023).The influence of industrial output value on the relative risk of non-optimal temperature was a V-shaped curve, which initially decreased and later increased with the increase in industrial output value.This indicates that moderate industrialization has a positive effect on improving people's ability to withstand severe weather events.Both under-industrialization and over-industrialization will lead to an increase in mortality risk.Additionally, the effect of the proportion of young people (students) on relative risk was found to be non-linear, indicating that young individuals were not particularly susceptible to mortality associated with non-optimal temperatures.Latitude has an increasing influence on the relative risk of high temperatures.Due to the small temperature difference among districts in East China, the influence of geographical location on the relative risk of nonoptimal temperature may be related to the difference in economic development levels in different regions.
The important indicators to measure the level of urban economic development are per capita GDP and local financial level.With the increase in public financial revenue and per capita GDP, the average relative risk of non-optimal temperatures gradually decreases.As illustrated in figure 3, the increase in public financial revenue and increase in per capita GDP have an approximately linear impact on reducing the relative risk of hot temperatures.With the increase in per capita GDP by 10 thousand yuan (RMB), the average relative risk of hot temperatures decreased by 0.075.The improvement in local public financial revenue can reduce the people's relative risk by building more public infrastructure to reduce people's exposure and improving localized public medical resources (Xiangzhao and Yanrui 2021).
There are multiple mechanisms how the improvement of per capita GDP level could reduce population health risks during non-optimal temperature (Smith 2023).On the one hand, people with high income are more inclined to and have the opportunity to choose more comfortable working and living environments, reducing exposure to high and low temperatures, which reduces the health risks of the population (Kjellstrom et al 2007).On the other hand, residents with better economic level have higher accessibility of medical resources (Du et al 2022), which further reduces the mortality during non-optimal temperature In addition, in areas with high per capita GDP, the population usually has a good nutritional level and physical fitness, a high cognition level of adverse temperature risk, and a good grasp of scientific and correct preventive measure (Smith 2023)., which also reduces the health risk of the population.
Since the elderly population was sensitive to nonoptimal temperatures, the average relative risk caused by hot temperatures increased with the aging level, which is consistent with previous studies (Baccini et al 2008, Lou et al 2021).As shown in figure 3, with the increase of 10% in the proportion of the elderly, the average relative risk of hot temperatures increased by 0.17.However, as depicted in figure 3(s), the average relative risk caused by low temperatures did not increase monotonously with the increase in the aging level, but firstly decreased and later increased.When the proportion of the elderly population (aged more than 60 years) was about 14.4%, the lowest average relative risk was observed.The reason for this phenomenon was that the influence of aging on low temperature risk has a certain threshold.When the degree of aging is not severe, the influence of aging on the relative risk will be influenced by covariates, such as the medical level.
Furthermore, the effects of per capita GDP on the exposure-response curves for different health endpoints in East China were compared.With the increase in per capita GDP level, the risk of high-temperature exposure for non-accidental death (figure 3), cardiovascular and cerebrovascular mortality (figure 4(a)), and respiratory disease mortality (figure 4(b)) significantly reduced.For nonaccidental mortality, the average relative risk of heat exposure peaked at the lowest per capita GDP level (1.8 thousand yuan) with a relative risk of 1.25.At the lowest per capita GDP level, the average relative risk of heat exposure for respiratory disease mortality was 1.49, while that of cardiovascular disease mortality was 1.34.Compared with the other health endpoints, the mortality risk for respiratory diseases shows the largest variation across different levels of per capita GDP, illustrating that respiratory diseases are more sensitive to economic levels (Foley et al 2019).
In North China, the widespread use of indoor heating equipment during winter introduces exposure misclassification when relying on stations' meteorological data to assess people's exposure to cold temperatures.The popularity of heating equipment is higher in regions with a higher level of regional economic development, leading to a greater likelihood of exposure misclassification.This, combined with potential statistical errors in health data, contributes to significant error and uncertainty in constructing exposure-response curves at the district/county level.Consequently, it becomes more challenging to identify the relationship between non-optimal health risks and socio-economic factors in North China.The change in relative risk of non-optimal temperatures in North China was not as obvious and significant as that in East China with the change in influencing factors, and the trend identification has great uncertainty, especially for the cold temperatures.The relative risks of non-optimal temperature under the varying levels are illustrated in figures S4 and S5.Those with a reasonable explanation of variation include the population and the proportion of the urban population.With the increase in population, the relative risk of cold temperatures initially increased and later decreased, while the relative risk of hot temperatures decreased after exceeding a certain threshold.This also illustrates the role of population aggregation and urbanization in protecting humans from adverse climatic conditions.

Future estimates of exposure-response relationship curves
Since there is a significant correlation between exposure-response curves and per capita GDP levels in East China, future exposure-response curves were projected based on per capita GDP levels.When considering only per capita GDP levels, the exposureresponse curves at each 10 year interval in East China under different climate change scenarios (SSP1, SSP2, SSP3, SSP4, and SSP5) are illustrated in figure 5.It is evident that as economic levels increase, there is an enhancement in the population's capacity to withstand high-temperature risks.Consequently, noticeable reductions in relative risks of high-temperature exposure were noted from 2030 to 2100 across SSP1, SSP2, SSP3, and SSP4 scenarios at equivalent temperature levels.However, improvements in the population's resilience to low-temperature risks were less pronounced.Notably, under the SSP5 scenario, characterized by a substantial increase in per capita GDP level, significant errors arose from extrapolation using existing models, leading to larger discrepancies in the exposure-response curve.Except for SSP5, SSP1 exhibits the highest rate of GDP growth among the scenarios.In the context of the SSP1 scenario, the relative risks at 34.8 • C decreased from 1.44 (95%CI: 1.33-1.56)to 1.37 (95%CI: 1.20-1.56),1.23 (95%CI: 1.04-1.44),and 1.04 (95%CI: 0.6-1.70)during 2030, 2050, and 2100, respectively.
The GDP development levels for both SSP2 and SSP4 scenarios demonstrate consistent patterns.In the context of the SSP2 scenario, the relative risks of exposure to a high temperature of 34.8 • C decreased from 1.44 to 1.38 (95%CI: 1.21-1.56),1.25 (95%CI: 1.07-1.47),and 1.08 (95%CI:0.74-1.57)during 2030, 2050, and 2100, respectively.Similarly, under the SSP4 scenario, the relative risks of exposure to a high temperature of 34.8 • C decreased from 1.44 to 1.36 (95%CI: 1.20-1.56),1.23 (95%CI: 1.05-1.49),and 1.10 (95%CI: 0.79-1.52)during 2030, 2050, and 2100, respectively.In contrast, the SSP3 scenario exhibits a slower rate of GDP development, resulting in less-pronounced reductions in relative risks of high temperature exposure.During 2030, 2050, and 2100, the relative risks of exposure to a high temperature of 34.8 • C decreased from an initial value of 1.44 (95%CI: 1.33-1.56)to values of approximately 1.40 (95%CI: 1.26-1.57),1.32 (95%CI: 1.13-1.53),and 1.27 (95%CI: 1.08-1.48),respectively under SSP3 scenario.These demonstrate that economic development plays a crucial role in mitigating the risks associated with high-temperature exposure induced by climate change.However, it is important to note that the present study solely focused on estimating exposure risks within the current temperature range in East China, as significant errors may arise from extrapolation calculations using statistical models.It is also worth noting that future temperature increases resulting from climate change are expected to surpass the existing maximum, thereby exacerbating the associated risks.Consequently, the complex interplay between escalating temperatures and mitigating effect caused by economic growth will jointly influence future population heat exposure risks, therefore necessitating further exploration and identification.Although, economic development can significantly enhance the adaptability of the population, the changing proportion of the elderly population in China will also alter the exposure-response curve of the population in the future.Using population projections from Tsinghua University, Beijing, China, the future trend of the share of the population aged 60 years and above was projected in East China.Considering the rapid progression of population aging in the future, it was projected that by 2050, the proportion of individuals aged 60 and above in East China will reach 44%, exceeding the predicted range of existing models.Therefore, the present study only estimated the exposure-response curve for East China in 2030 when the population aging is projected to reach approximately 29%.By comparing with figure 5 (where only economic levels change), we observe a more pronounced variation in the exposure-response curve.It is evident that the level of aging exerts a more significant influence than the level of GDP in shaping the exposure-response curves in the near future.As shown in figure 6, it was found that the relative risk of heat and cold exposure will increase by 2030.
The relative risk of humans at a high temperature of 34.8 • C will rise from 1.44 (95%CI: 1.33-1.56)to 2.34 (95%CI: 0.83-6.60) in East China.The relative risk of humans at a low temperature of −6.3 • C will rise from 1.82 (95%CI: 1.64-2.02)to 2.38 (95%CI: 0.55-10.27) in East China.This indicates that in the future, there will be an exacerbation of overall population vulnerability, leading to heightened sensitivity towards both heat and cold exposures.This is consistent with the results of previous studies (Xing et

Conclusions
This study analyzed the relationship between the temperature exposure-response curve and socioeconomic factors using a two-stage meta-analysis and non-linear spline regressions.In addition, the study predicted the future temperature exposure-response curve under different SSPs with fixed and changing aging levels.The results show that compared to East China, North China demonstrates a lower relative risk during low temperatures and a higher relative risk during high temperatures.In North China, the shape of the exposure-response curve changed greatly under different social and socioeconomic levels.In East China, with the change in social and economic factors, the change in the exposure-response relationship shows a certain regularity.In the early stage of urbanization, population aggregation did not reduce the average relative risk of cold and heat exposure, and the average relative risk of adverse temperature showed a downward trend following urbanization to a certain extent.The improvement in the per capita GDP level, per capita fiscal income level, and medical and health level has played an important role in enhancing the population's resistance to the risk of adverse temperatures.Specifically, the average relative risk of heat exposure is approximately linear with the increase in per capita GDP, with a decrease in the relative risk by 0.075 for every 10 thousand yuan increase in per capita GDP.Comparing the impact on different mortality rates, the variation in respiratory disease risk was most pronounced at different per capita GDP levels, indicating its sensitivity to economic development.It was found that the level of aging exerts a more significant influence than the level of GDP in shaping the exposure-response curves in the near future.In the future, when the aging level intensifies, the overall exposure response curve of the population will become steeper in both the cold and the hot zones.When population aging was held constant, socio-economic development could mitigate heat exposure to some extent, indicating the effectiveness of climate adaptation.

Figure 1 .
Figure 1.The exposure-response curves of non-accidental deaths in North China at different levels of socio-economic factors.The different colored lines represent the exposure response curves for different factors with low, medium and high level, respectively.Proportion of The Elderly represents the proportion of the people aged above 60 years; Population represents the registered population of each district; Bed occupancy ratio and Proportion of students represent the ratio of the number of beds in health care facilities and the number of middle school students to the registered population; Industrial output per capita, Public Finance per capita, and per capita GDP represent the industrial output, Public Finance, and GDP to the registered population; Proportion of The Elderly and Proportion of Urban population were from the Sixth National Population Census.Population, Number of beds in health care facilities, Industrial output, Public Finance, GDP, and number of middle school students were from the 2015 Statistical Yearbook.

Figure 2 .
Figure 2. The exposure-response curves of non-accidental deaths in East China at different levels of socio-economic factors.The definitions used in this figure are the same as in figure 1

Figure 3 .
Figure 3. Variation in relative risk of hot temperature exposure for non-accidental deaths in East China with different socio-economic factor levels.

Figure 4 .
Figure 4. Change in average relative risk of heat exposure for cardiovascular disease mortality (a) and respiratory disease mortality (b) with increasing per capita GDP.

Figure 5 .
Figure 5. Future exposure-response curves in East China under different climate change scenarios (SSP1, SSP2, SSP3, SSP4, and SSP5) with the current proportion of the elderly.
al 2022, Cole et al 2023, He et al 2023).Cole et al (2023) revealed that the exclusion of any projected changes in demographics leads to an underestimation of health burdens by an average of 64%.Studies based on age-group projections also have confirmed this phenomenon.He et al (2023) indicated that the rise in the aged population (⩾65) would substantially amplify the excess deaths related to cold spells (increase by 101.1% in the 2050s and 146.2% in the 2090s).Xing et al (2022) also predicted that only intensified aging can increase future temperaturerelated CVD mortality by 48.8%-325.9% in Beijing, China.

Figure 6 .
Figure 6.Exposure-response curves in East China under the SSP1 scenario in 2030 with the changing proportion of the elderly.

3.1. Comparison of exposure-response relationship based on temperature and temperature quantile in the two regions
Based on the exposure-response associations, the minimum mortality temperature for North China and East China was estimated to be 21.1 and 23.2 • C, the average relative risk of cold exposure in East China was 0.114 higher than that in North China.Conversely, in the high temperature range, the relative risk for individuals in North China shows a sharper increase with increasing temperature compared to those in East China.At 23.2-33.6 • C, the average relative risk of heat exposure in North China was 0.128 higher than that in East China.For instance, when the temperature reached 33 • C, the relative risks were 1.60 (95%CI: [1.38,1.87])and 1.31 (95%CI: [1.24,1.40]) in North China and East China, respectively.