Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2026 Apr 16;16:17778. doi: 10.1038/s41598-026-48668-z

Humidity may amplify the temperature-related health risks in the context of climate change

Yiling He 1,#, Wei Dong 2,#, Liang Zhao 3,#, Huibin Wang 4, Wenjia Cai 5, Zhao Liu 6, Hualiang Lin 7, Wenjun Ma 8, Cunrui Huang 9,✉
PMCID: PMC13246766  PMID: 41992004

Abstract

Humidity and temperature may pose joint effects on human health. However, current temperature-related risk assessments generally focus on univariate temperature statistics, leading to inaccurate risk estimation. By combining temperature and humidity intensities as compound temperature-humidity events (hereinafter as compound events), we examined the effects of humidity in temperature-related health risks in the context of climate change in China using ambulance dispatch data, and further determined the high-risk types and thresholds of compound events. After assessing the health risks, compound events are defined as: i) warm and wet event: temperature ≥ 90th, humidity ≥ 55th and 70%; ii) warm and dry event: temperature ≥ 80th, humidity ≤ 10th; iii) cold and wet event: temperature ≤ 15th, humidity ≥ 70th and 70%; iv) cold and dry event: temperature ≤ 5th and humidity ≤ 25th. Results reveal that humidity can amplify the temperature-related health risks. Higher risks were found during cold and dry events and warm and wet events, posing risks as 1.102 (1.045–1.161) and 1.093 (1.068–1.118) respectively. Compound events mainly occurred in Southeast China during 1979–2019, while the frequency and gripped regions are projected to increase by 2100, and climate change will amplify these trends. Since humidity would exacerbate temperature-related health risks, therefore multiple meteorological parameters are needed in defining adverse weather conditions and conducting risk assessment accurately.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-48668-z.

Keywords: Climate change, Temperature, Humidity, Compound event, Health risk, Threshold, China

Subject terms: Climate sciences, Environmental sciences, Health care

Introduction

Extreme temperature exposures are known to contribute to an increase in mortality and morbidity, posing significant public health problems around the world1–8. For example, during the 1995 Chicago heatwave, there were over 700 excess deaths in a single day9. The well-known 2003 heatwaves led to 15,000 excess deaths in France alone10, and over 70,000 deaths across Europe11. The 2008 cold spell in fifteen provinces of southern China increased mortality by 43.8% (95% CI: 34.8% ~ 53.4%), resulting 148,279 excess deaths12. Previous studies have shown that both high and low humidity can pose adverse effects on human health13,14. Humidity will make temperatures feel hotter or colder than it actually is, as it affects evaporation from human skin. In the hot and humid environment, sweating capability of human body will be reduced, resulting in rising body temperature and physical discomfort. While during the cold weather, the chilling effect of atmospheric humid can lower the body temperature. Therefore, a deep investment of the role of humidity in the relationship between temperature and health risks would contribute to a more appropriate assessment of temperature-related health risks.

Various temperature indices have been developed to allow for the combination of temperature and humidity, such as humidex, apparent temperature (AT) or wet-bulb globe temperature (WBGT). These indices have been argued to do better than dry-bulb temperature in the health impacts assessment15–20, as they regard humidity as an integral component of environmental exposures rather than a confounding factor to be controlled in the model, thus incorporating the synergistic effects of temperature and humidity on human thermal perception. Bernard and Iheanacho showed that WBGT can more accurately predict health events related to occupational heat stress20. A study by Min et al. in South Korea also confirmed that temperature indices combined with humidity have a significantly higher correlation with heat-related mortality than single temperature indices17.

Although composite indices simplify analysis by integrating temperature and humidity into a single value, they mask differences in health risks under different temperature–humidity combinations. For example, the same WBGT value may arise from “high temperature and low humidity” and “low temperature and high humidity”, which are the two distinct environments with different mechanisms of impact and risk levels for human health. Moreover, with the process of climate change, future frequency and intensity of compound heat-humidity events are projected to escalate globally, resulting rising population exposure to the compound events21–26. Oversimplification of these complex exposures may result in underestimation of heat-related health risks. Thus, under the global warming with potentially increased compound events in the future, improved understanding of the variability of compound events and their health impacts has been highlighted.

To address the challenge of distinguishing health risk of different intensity of temperature and humidity exposures, we introduce a new definition of compound temperature-humidity event, which is selected from various combinations of temperature and humidity, thus to identify the role of humidity in temperature-related health risks. Specifically, we first assess the health risks under different compound events exposures, so as to identify the type and the thresholds of the events with high risks. We then evaluate that whether exposure to the compound events posing a higher risk than the temperature alone. We further investigate health impacts of compound temperature-humidity events over China based on current climate, and then estimate impacts during the historical and future periods under global warming scenarios. Results from this study show humidity would amplify the temperature-related health impacts, and therefore raising importance of using compound temperature indicators in the temperature-related health assessment.

Materials and methods

Study area

This study was focused on China, a climate change sensitive area with high population density, and rapid urbanization. Study was conducted in the urban districts of 13 cities, which includes Guangzhou, Shenzhen, Zhuhai, Hangzhou, Ningbo, Taizhou, Zhoushan, Lishui, Zhengzhou, Luoyang, Xuchang, Jinan, and Huairou district in Beijing. All of the selected areas are the densely populated in China. These cities are distributed across eastern, central and northern China, covering approximately 70% of the national population27. Furthermore, the study cities span three major climate zones (subtropical monsoon, temperate monsoon, temperate continental) and encompass various urbanization levels, including megacities, large cities, medium-sized cities.

Morbidity data

Emergency ambulance dispatches, the front-line of medical services, usually relate to acute health outcomes. The number of dispatches is considered indicative of the extent of people affected by hazardous weather, and is therefore selected as the health outcome in this study. Daily records of emergency ambulance dispatches in each city were obtained from the institutions such as the Centers for Disease Control (CDC) and local Ambulance Service Centers where collect ambulance dispatch information for the whole city, covering the longest time span of 2013–2019.

Climate data

Daily observations of mean temperatures and relative humidity were obtained from the China Meteorological Data Sharing Service System (http://data.cma.cn/en). To allow for a sensitivity analysis, air pollution data was obtained from the National Urban Air Quality Real-time Publishing Platform (http://106.37.208.233:20035). The air pollutants considered were particulate matter with an aerodynamic diameter less than 2.5 μm (PM2.5), ozone (O3), and sulfur dioxide (SO2).

Projections of daily mean temperature and relative humidity were obtained from 12 global climate models (GCMs), which are provided by the latest Coupled Model Intercomparison Project Phase 6 (CMIP6) (https://esgf-node.llnl.gov/search/cmip6/) (See Supplementary Table S1). To reduce uncertainties in future projections, we used a multi-model ensemble mean (MME) approach. As shown in previous studies, a sufficiently large CMIP6 ensemble (e.g., ≥ 10 models) effectively suppresses individual biases and internal variability, leading to more stable and robust climate outcomes28. These climate projections are driven by a new set of emissions and land use scenarios, which are rooted in socioeconomic pathways (shared socioeconomic pathways, SSPs) work in harmony with representative concentration pathways (RCPs) through shared policy assumptions to make the future scenarios more reasonable29,30. In this study, we considered three integrated scenarios, including SSP126 (combination of SSP1 and RCP2.6), SSP245 (combination of SSP2 and RCP4.5) and SSP585 (combination of SSP5 and RCP8.5), to represent the low, medium and high-end of future emissions and warming, respectively.

Observed climate data in China during 1979–2019 were obtained from ECMWF Reanalysis v5 (ERA5) project, which included temperature and dew-point temperature data (at 2 m above the land surface) at 0.25° × 0.25° spatial resolution. ERA5 is the fifth generation of ECMWF atmospheric reanalysis of the global climate, covering the period from 1950 to present31. With a high spatial resolution and a long, consistent record, ERA5 is a widely used and reliable reanalysis dataset. Evaluations show it accurately reproduces the spatiotemporal patterns of temperature and precipitation over China, including climatology, trends, and interannual variability32,33. Data was downloaded from the European Centre for Medium-Range Weather Forecasts (ECMWF, https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5 ). Relative humidity (at 2 m above the land surface) were then calculated by 2-m temperature and dew point temperature. In line with previous researches, we downscaled the spatial resolution of the climate data from 0.25° × 0.25° to 0.5° × 0.5° to ensure spatial consistency34–36. Meanwhile, we used the bilinear interpolation for regridding to minimize the loss of high-resolution information.

Population data

The population of 13 cities during the baseline period (2013–2019) was obtained from the Municipal Bureau of Statistics in each city. Historical population data in China during 1979 and 2000 was obtained from the inter-sectoral impact model intercomparison project (ISIMIP), which provide annual global population data for the years 1861–2005 on a 0.5° × 0.5°grid (https://www.isimip.org/gettingstarted/details/31/). And annual population data during 2001–2010 was generated from the gridded population of the world, version 4 (GPWv4), which is in line with the previous study6. Projections of future population changes in China based on SSP scenarios were obtained from Chen, et al.37. The main assumption for population growth in SSP scenarios is dependent on fertility, mortality, migration, and education rates.

Association between daily ambulance dispatches and daily mean temperature

To assess if unfavorable temperature would pose adverse health impacts on ambulance dispatches, a time-series quasi-Poisson regression analysis was used to estimate the city-specific ambulance dispatch risk (relative risk, RR) associated with daily mean temperature. A distributed lag non-linear model (DLNM) was used to examine delayed effects38. Consistent with previous studies, delayed effects were assessed for up to 7 days. The selection of degrees of freedom (df) was based on minimizing Akaike’s Information Criterion (AIC)39. Specifically, we used a natural cubic spline with 3 df for temperature and a natural cubic spline with 3 df for lag to capture nonlinear temperature effects and lagged effects respectively. A natural cubic spline with 3 df per year was used to control for the potential confounding effect of long-term trends. Another time-varying confounders and known predictors of ambulance dispatches were controlled in the model; these were relative humidity, air pollutants, day of week and seasonality.

Define the compound events with high risks

There is no standard compound events definition worldwide, although they are commonly defined as concurrence of extreme temperature event with extreme humidity event identified by relative thresholds. Compound temperature-humidity events were further specified by the intensity of temperature and humidity, as: i) compound high temperature with high humidity event (warm and wet event); ii) compound high temperature with low humidity event (warm and dry event); iii) compound low temperature with high humidity event (cold and wet event); iv) compound low temperature with low humidity event (cold and dry event). To derive a universal definition and to easily conduct spatial comparison for the entire China, here we utilized the percentile threshold definition for the event which is consistent with previous studies40,41. In the present study, a series of compound events were defined to figure out the thresholds of temperature and humidity with high risks and introduced as below.

On the first step, we classified compound events with refined groups. In line with previous studies, we set multi-gradient thresholds at 5th percentile intervals for temperature and relative humidity distributions to define high-risk events of varying severity42–44. By comparing the associated health risks across different thresholds, we identified the optimal definition criteria for such events and evaluated their health risks. Specifically, daily mean temperature and relative humidity of each city were divided from 5th percentile to 95th percentile of distributions, and 20 groups of temperature and humidity were obtained respectively. Then, 400 types of compound events were obtained by pairwise combination of humidity and temperature groups, presenting as a square grid with 20*20 grids. Estimating the distribution of humidity conditional on temperature provides flexibility for quantifying compound exposure by humidity and temperature simultaneously. This approach allows us to look at humidity at a given temperature or temperature at given humidity quantile.

On the second step, to estimate risks during compound events with different thresholds, a binary event variable was used for each day in the DLNM model. Days with compound event or individual event were categorized as 1 while a normal day was categorized as 0. Separate models were used to estimate risks of heat-related ambulance dispatches for each type of events. To obtain national-level pooled estimates separately for each type of compound events, the estimated city-specific associations (β) were pooled using a univariate meta-analysis. Heterogeneity was assessed through a multivariate extension of the I2 index and Cochran Q test, which quantifies the percentage of variability due to true differences across communities.

On the third step, we defined the compound temperature-humidity events in terms of high risk and extreme criteria. For the high-risk identification, we first drew a contour map with 20*20 grids to present the risks of compound events. The inflection point where health risks rise sharply and significantly is defined as the threshold for high-risk events. Then we drew a contour map with 20*20 grids to present the risks of compound events. The temperature and humidity were divided into high and low exposure levels at 50th percentile interval. This grid graph was then divided into four quadrants, representing four categories of compound events mentioned previously. For the extreme intensity classification, we connected the temperature and humidity corresponding to their 50th percentile in each quadrant, dividing four quadrants into outer and inner triangle that lie farther and closer to the grid center, respectively. The outer triangle is the aggregation of compound events that meet the extreme criterion. And finally, in the cluster of compound events that meeting the high-risk and extreme criteria, the temperature and humidity that closest to the 50th percentile of their distribution were defined as the thresholds of compound events with high risks.

To clarify whether the impact of compound events on ambulance dispatch is greater than that defined by a single temperature variable, we evaluate the risks of ambulance dispatches during compound events and temperature events with the same temperature thresholds.

Ambulance dispatches attributable to historical and future compound events

Ambulance dispatch numbers attributable (ANs) to each compound event per year during the historical period (1979–2019) and future (2020–2100) in the warming scenarios were estimated (See Eq. 1). Letter Inline graphic is the type of compound events (warm and wet, warm and dry, cold and wet, cold and dry events), letter Inline graphic is the historical periods and future warming scenarios, and letter c is the 13 cities in the baseline study. AF is the attributable fraction of ambulance dispatches for each compound event, basing on relative risks (RR) in the baseline model (See Eq. 2). ADC is the annual ambulance dispatch counts (See Eq. 3), ND is the number of days identified as the compound event, and POP is the number of populations during the historical, baseline and projected period.

graphic file with name d33e551.gif 1
graphic file with name d33e556.gif 2
graphic file with name d33e560.gif 3
graphic file with name d33e564.gif 4

Morbidities of ambulance dispatches in 13 cities were calculated based on Eq. 4. A fitted meta-analytical model was used to derive the best linear unbiased prediction of the overall morbidity in China.

Results

Patterns of ambulance dispatches and thermal characteristics

A total number of 2.46 million ambulance dispatches were recorded during the study periods (Table S2). Warmer and wetter weather was found in southern cities compared with cities in northern part of China. The relationship between mean temperature and ambulance dispatches followed a U-shaped curve (Figure S1). The optimum temperature (OT), corresponding to the lowest point in the exposure–response curve, was higher in the southern cities and lower in the northern cities. Increased risks were found during both hot and cold days, with higher risks during extreme temperatures.

Thresholds of high-risk compound temperature-humidity events

Pooled risks of ambulance dispatches during compound temperature-humidity events in China were presented in Fig. 1. Detailed risk estimates, the three-dimensional plot illustrating the temporal pattern of risk variations, and city-specific risk results are provided in the Supplementary Material (Figure S2-S4). As shown in Fig. 1 and Figure S3, the corresponding coverage of temperature and humidity distributions exhibits a substantial variation as the risks increases from 1 to 1.08. In contrast, minor fluctuation in temperature and humidity distributions result in a significant rise in health risks when RR ≥ 1.08. Accordingly, this study sets RR = 1.08 (corresponding to the 80th percentile of the risk distribution) as the threshold for high-risk events. As suggest by previous studies, the recommended wet condition was defined as relative humidity higher than 70%45,46. Thus considering the high-risk and extreme criteria, the thresholds of high-risk compound events were finally defined as follows: i) warm and wet event: temperature ≥ 90th, humidity ≥ 55th and 70%; ii) warm and dry event: temperature ≥ 80th, humidity ≤ 10th; iii) cold and wet event: temperature ≤ 15th, humidity ≥ 70th and 70%; iv) cold and dry event: temperature ≤ 5th and humidity ≤ 25th.

Fig. 1.

Fig. 1

Open in a new tab

Relative risks during compound temperature-humidity events defined by 400 types of temperature and humidity combinations in China.

Health risks of compound temperature-humidity events

City-specific and pooled risks of ambulance dispatches during compound temperature-humidity events were presented in Fig. 2. The highest effect estimate was observed during cold and dry event, followed by warm and wet event, and warm and dry event, while the lowest estimate was found during cold and wet event.

Fig. 2.

Fig. 2

Open in a new tab

Pooled effects of compound temperature-humidity events and effects of temperature with humidity controlled in China.

The pooled RRs with 95% CIs of compound events on ambulance dispatches were 1.102 (1.045–1.161), 1.093 (1.068–1.118), 1.091 (1.052–1.131) , and 1.081 (1.038–1.125) respectively. In addition, under the same temperature thresholds, the risks of compound temperature-humidity events were higher than that of single temperature events (Fig. 3). Results verified the amplified effects due to compound exposure.

Fig. 3.

Fig. 3

Open in a new tab

Age- and gender-specific ambulance dispatches during compound temperature-humidity events.

Gender and age-stratified risks of ambulance dispatches during compound events were presented in Fig. 3. Both males and females were sensitive to compound temperature-humidity events. When stratified by age group, the results shown that the elderly (people aged 60 and above) were sensitive to the compound temperature-humidity events. Specifically, during warm and wet events, highest risks were found among people aged 80 and above (RR = 1.118, 95% CI: 1.060–1.178), followed by people aged 60–79 years (RR = 1.116, 95% CI: 1.031–1.208). Similarly, during warm and dry events, higher risks were also found among the two elderly groups, with risks of 1.206 (95% CI: 1.105–1.315) and 1.104 (95% CI: 1.030–1.183), respectively. While during cold and wet events, people aged 18–44 years (RR = 1.075, 95% CI: 0.982–1.176) and people aged 45–59 (RR = 1.068, 95% CI: 1.026–1.112) were more vulnerable. People aged 80 and above, and people aged under 17 years were sensitive to the cold and wet events.

Historical and projected number and spatial distribution of compound events

In the past 30 years (1979–2019), the compound temperature-humidity events presented higher in the southeast and lower in the northwest of China, with hotspot in central China that suffered about 20–40 compound temperature-humidity events annually (Fig. 4). An increasing trend of the occurrence and spatial extension of compound events was detected in heat events, with hotspot in southeast of China, while the decreasing trend was found in cold events. In addition, under the higher emission scenarios (SSP245 and SSP585), spatial heterogeneity of the annual occurrence of compound events changes would be more significant (Fig. 4). For example, till the late twenty-first century, changes of compound events occurrence in most areas of China under SSP126 scenario is less than 10 per year. Under SSP585 scenario, the annual growth rate in Southeast China is more than 20 times per year, and even 40 times per year in some part of China, while most of Northwest China show a decreasing trend, and some areas will decrease by more than 10 times per year.

Fig. 4.

Fig. 4

Open in a new tab

Spatial distribution of the historical (1979–2019) annual number and future changes (2090–2100) of compound temperature-humidity events under SSP126, SSP245, and SSP585 warming scenarios in China. a) heat and wet event; b) heat and dry event; c) cold and wet event; d) cold and dry event; e) total number of four compound events.

Historical and projected distribution of attributable number of ambulance dispatches

In the past 30 years (1979–2019), population vulnerability of compound events varied among different districts in China (Fig. 5). Higher risks of ambulance dispatches attributable to compound events were observed in the East China, with about 5,000–15,000 ANs per year. Among the four compound events, the impact of ambulance dispatches due to cold and wet events has a wider spatial extension in China, followed by heat and wet events, and cold and dry events, while heat and dry events the least. With the process of climate change, future compound events are projected to bring about a wider spatial extension and a deeper influence of health impacts (Fig. 5). Significant increasing trends of heat and wet events and heat and dry events were detected, with additional 1,000–20,000 ANs per year compared with past 30 years. And these trends will be amplified by the higher emission scenarios. While the growth of ANs due to cold and wet events and cold and dry event exposure gradually decreased, among which southeast China showed a negative growth of ANs.

Fig. 5.

Fig. 5

Open in a new tab

Spatial patterns of the historical (1979–2019) annual number and future changes (2090–2100) of ambulance dispatch attributable to compound temperature-humidity events under SSP126, SSP245, and SSP585 warming scenarios in China. a) heat and wet event; b) heat and dry event; c) cold and wet event; d) cold and dry event; e) total ANs of four compound events.

Historical and projected total attributable number of ambulance dispatches

The estimated annual number of ambulance dispatch attributable to compound events (ANs) in China during 1979–2100 are shown in Fig. 6. Total number of annual ANs due to the compound events in China present an increasing trend during 1979–2029, and stabilize at the highest value during 2030–2060, while declining after 2060s. Compared with the low emission scenario (SSP126), under the middle and high emission scenarios (SSP245 and SSP585), the total ANs will increase by 42.4% and 33.3%, respectively at the end of the twenty-first century.

Fig. 6.

Fig. 6

Open in a new tab

Annual ambulance dispatch attributable to compound temperature-humidity events during 1979–2100 across China.

ANs related to cold events remain stable during 1979–2029, while a decreasing trend occurs after 2030s. Rising trend of ANs related to heat events was found from 1979 till the mid of the twenty-first century. When compared with SSP126 scenario, under the SSP585 warming scenario, ANs related to heat events (heat and wet event, and heat and dry event) will increase by 71.5%, while ANs related to cold events (cold and wet event, and cold and dry event) will decrease by nearly 4/5 (82.8%) at the end of the twenty-first century. Higher emission scenario will change the main cause of health risks. Specifically, during the past 30 years (1979–2000), ANs due to cold events was 1.5 times than that of heat events. By the end of the twenty-first century (2090s), ANs due to heat events will be 3.5–30 times as much as that due to cold effect under warming scenarios.

Sensitivity analysis

In the sensitivity analysis, we used generalized additive model (GAM) to examine association between meteorological exposure and health effects. Results showed a similar pattern in the exposure–response curve (Figure S5). In addition, we performed robustness checks by varying the maximum lag days (6—8 days) of compound events, and the degrees of freedom for the long-term trend (4—6 per year), and the results obtained were consistent with those from the main model (Table S3).

Discussion

In present study, the modification effects of humidity in temperature-related health impacts have been discerned in China. This is different from previous studies that merge several meteorological factors into a single indicator. Conversely, we assessed the health risks of temperature and humidity exposures in all combinations. The present study also identified the types and thresholds of compound temperature-humidity exposure with high risks. We further assessed the changes in the intensity and spatial distribution of risks due to compound events exposure during historical period (1981–2010) and future period (2020–2100) under different warming scenarios. Results provide a more robust and realistic projections of future heat-related health risks in a warming world. The analysis introduced in this study extends beyond the previous studies by highlighting the multiple meteorological parameters in defining adverse weather conditions and risk projections, which are distinct from those that using temperature alone to predict health impacts.

Our results suggest that the compound exposure of high-humidity and temperature poses higher risks than that of compound low-humidity and temperature events. One reason that high humidity makes hot or cold weather more unbearable is that increased relative humidity intensifies the perceived heat or cold, thereby reducing human temperature tolerance. Additionally, during heat and humid events, higher amount of water vapor will reduce the evaporation rate of skin moisture. As a result, it reduces the effectiveness of sweating in cooling the body, thus increasing the incidence of heat-related diseases47–50. Using global urban and rural weather station data, Zhang et al. showed that humid climates raise the wet-bulb temperature (Tw), a key indicator of heat stress, leading to an additional 2—6 dangerous heat-stress days each summer in humid regions51. Liang et al. also found higher morbidity risks occurred in humid-heat weather in Shanghai, China52. Since ambulance dispatch data does not include specific etiological information, future studies should incorporate etiology-specific data for further stratification.

Temperature-related impacts are often related to multiple drivers. However, the complexities are currently neglected with common heat-related health assessment, assuming that a single temperature value can adequately predicts health impacts53–56. Although an array of heat indices has been developed to account for the effects of multiple meteorological factors, such as temperature, humidity and wind57–59. However, these indices eventually merged into a single and new index, such as WBGT, AT. In that case, the risk of different combination of temperature and humidity cannot be distinguished. To be specific, if the value of WBGT during high temperature and low humidity event is the same as that during low temperature and high humidity, then the risks reported by WBGT would be the same, which would mask the health impacts of different weather exposures. Moreover, the threshold values of WBGT with high risks cannot be directly translated into specific temperature and humidity criteria. This limitation hinders the development of meteorological early warning systems and also prevents the public from taking preventive measures based on weather forecasts. Oversimplification of these complex responses can result in underestimate the risks of temperature-related health impacts as illustrated by examples in this research.

To identify the adverse meteorological conditions posed high risks on human health, this study comprehensively evaluated the health risks under different meteorological conditions, so as to determine the high-risk temperature and humidity thresholds. Conventional investigation of temperature-related health impacts generally set the threshold of adverse temperature empirically. For example, the intensity of extreme heat or heat wave events was 90th, 95th or 98th percentile of the temperature range. However, the present results reveal that the threshold of compound events is lower than that of extreme heat or heat wave events. The reason behind this might be that the combined effects of temperature and humidity may amplify the effect of temperature on health. Therefore, a combination of non-extreme temperature and humidity exposure will produce high risks.

The identified compound thresholds in the present study (heat ≥ 80th/90th percentile, cold ≤ 5th/15th percentile) are generally consistent with findings from single-factor meteorological studies. Traditional studies commonly define extreme temperature thresholds empirically; for instance, heat or heatwave events are often defined using 95th, 98th, or 99th percentiles of temperature60,61. Notably, our results indicate that compound event thresholds are lower, indicating that at the same temperature intense, compound temperature-humidity exposure yields higher risks than temperature alone, so the high-risk threshold for compound events can be identified at a less extreme temperature percentile, which is more sensitive to capturing the adverse health effects of compound meteorological conditions.

With the process of climate change, annual total number of ambulance dispatches attributable to compound events showed an inverted U-pattern that increase first and then decrease. This is because ANs caused by cold events will continuous to decline, while ANs caused by heat events will increase in the early and middle of the twenty-first century, and tend to be stable at the end of the twenty-first century. Due to the impacts of climate change, different emission scenarios will change the composition of ANs related to compound events62–64. Nearly a third of ANs are caused by cold events under the low emission scenario, while heat and humidity events take the majority under the higher emission scenarios. In addition, higher emission scenarios would exacerbate the spatial heterogeneity of effects of compound events on ambulance dispatches in China. Although concerns about changes in climate extremes over China have been widely raised over the past decades, no research to date have considered health risks of compound temperature-humidity events over China, which may have more severe impacts on human society than single climate extremes and lie a lot of uncertainty to reveal.

This study has several limitations. First, risk estimates were derived from 13 Chinese cities; while nationally representative to some extent, broader regional data (such as Northwest and Northeast China, high-altitude areas such as the Qinghai-Tibet Plateau) are needed to validate findings. Second, the thresholds of compound events with high-risks were based on morbidity, thus generalizing to mortality and other health outcomes requires caution. Third, heat stress is also affected by solar radiation, wind, human activity, and clothing, which were not included. Fourth, future projections relied on fixed urban boundaries without accounting for urban expansion, potentially introducing uncertainties. Moreover, long-term population adaptation was not explicitly considered. Higher-resolution data and integration of urban planning and adaptive capacity models are recommended for future research. Despite these limitations, this study provides a basis for understanding high-risk compound temperature-humidity events and supports further refined projections.

Considering adverse meteorological conditions as a compound event and decomposing them to discern the thresholds that trigger health impacts can improve risk modelling and assessment, thus providing a pathway to better understand the health impacts of climate change. Compound exposure of multiple meteorological factors exacerbates the adverse health impacts, with non-extreme temperature and humidity can pose high risks. With the ongoing climate change and population changes, the frequency and spatial extent of compound events are expected to increase, result in amplifying their impacts on health. Our findings support arguments for embedding more sophisticated human heat stress models into health impacts assessment and climate projections, thereby providing more robust and realistic projections of future heat-related health risks in a warming world.

Supplementary Information

Supplementary Information. (604.2KB, docx)

Author contributions

C.H. conceived the study. Y.H., W.D. and L.Z. conducted the study, performed the statistical analysis and drafted the manuscript. H.W. contributed to discussion. W.C., Z.L., H.L., and W.M. assisted with data acquisition and reviewed/edited manuscript. C.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Funding

This work was supported by the grant from Guangzhou Science and Technology Programme (No. 2025A04J4695), National Key R&D Program of China (2018YFA0606200).

Data availability

The ambulance dispatch data that support the findings of this study are available from institutions such as the Centers for Disease Control (CDC), and the Medical Information Centers. However, restrictions apply to the use of these data, which were used under license for the current study and thus are not publicly available. The data are, however, available upon request and with the permission of these institutions.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Yiling He, Wei Dong and Liang Zhao are contributed equally to this work.

References

  • 1.IPCC. Climate change 2022: impacts, adaptation, and vulnerability. Contribution of working group II to the sixth assessment report of the intergovernmental panel on climate change. (2022).
  • 2.Xu, J. et al. Effects of extreme heat on physiology, morbidity, and mortality under climate change: Mechanisms and clinical implications. BMJ (Clin. Res. Ed.)391, e084675. 10.1136/bmj-2025-084675 (2025). [DOI] [PubMed] [Google Scholar]
  • 3.Longden, T., Quilty, S., Haywood, P., Hunter, A. & Gruen, R. Heat-related mortality: An urgent need to recognise and record. Lancet Planet. Health4, e171. 10.1016/s2542-5196(20)30100-5 (2020). [DOI] [PubMed] [Google Scholar]
  • 4.Gasparrini, A. et al. Projections of temperature-related excess mortality under climate change scenarios. Lancet Planet. Health1, e360–e367. 10.1016/S2542-5196(17)30156-0 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Ebi, K. L. et al. Health risks of warming of 1.5 °C, 2 °C, and higher, above pre-industrial temperatures. Environ. Res. Lett.13, 063007. 10.1088/1748-9326/aac4bd (2018). [Google Scholar]
  • 6.Cai, W. et al. The 2020 China report of the Lancet Countdown on health and climate change. Lancet Public Health10.1016/s2468-2667(20)30256-5 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Watts, N. et al. The 2020 report of The Lancet Countdown on health and climate change: responding to converging crises. Lancet (London, England)10.1016/s0140-6736(20)32290-x (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Tao, J. et al. Quantifying localized heatwave impact on mortality: A multi-country modeling study in the Asia-Pacific region. Lancet Reg. Health West. Pac.61, 101653. 10.1016/j.lanwpc.2025.101653 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Semenza, J. C. et al. Heat-related deaths during the July 1995 Heat Wave in Chicago. N. Engl. J. Med.335, 84–90. 10.1056/nejm199607113350203 (1996). [DOI] [PubMed] [Google Scholar]
  • 10.Le Tertre, A. et al. Impact of the 2003 heatwave on all-cause mortality in 9 French cities. Epidemiology17, 75–79. 10.1097/01.ede.0000187650.36636.1f (2006). [DOI] [PubMed] [Google Scholar]
  • 11.Fouillet, A. et al. A predictive model relating daily fluctuations in summer temperatures and mortality rates. BMC Public Health7, 114. 10.1186/1471-2458-7-114 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Zhou, M. G. et al. Health impact of the 2008 cold spell on mortality in subtropical China: The climate and health impact national assessment study (CHINAs). Environ. Health13, 60. 10.1186/1476-069X-13-60 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Fang, W. et al. The joint and interaction effect of high temperature and humidity on mortality in China. Environ. Int.171, 107669. 10.1016/j.envint.2022.107669 (2023). [DOI] [PubMed] [Google Scholar]
  • 14.Mei, Y. et al. Associations and burdens of relative humidity with cause-specific mortality in three Chinese cities. Environ. Sci. Pollut. Res. Int.30, 3512–3526. 10.1007/s11356-022-22350-z (2023). [DOI] [PubMed] [Google Scholar]
  • 15.Hess, H. W. et al. Critical assessment of the recommended alert limit curves for occupational heat exposure. Am. J. Ind. Med.68, 1004–1012. 10.1002/ajim.70022 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Fard, P. et al. Developing a spatiotemporal repository of extreme heat and cold exposure in the United States for precision public health research. Environ. Sci. Technol.59, 19872–19884. 10.1021/acs.est.5c03658 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Min, J., Lee, W., Oh, J., Kang, C. & Ha, E. Comparing the role of relative and absolute humidity in heat-related mortality: A case time-series study in South Korea. Environ. Health Perspect.133, 67015. 10.1289/ehp15827 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Ho, H. C., Knudby, A., Xu, Y., Hodul, M. & Aminipouri, M. A comparison of urban heat islands mapped using skin temperature, air temperature, and apparent temperature (Humidex), for the greater Vancouver area. Sci. Total Environ.544, 929–938. 10.1016/j.scitotenv.2015.12.021 (2016). [DOI] [PubMed] [Google Scholar]
  • 19.Ahmadalipour, A. & Moradkhani, H. Escalating heat-stress mortality risk due to global warming in the Middle East and North Africa (MENA). Environ. Int.117, 215–225. 10.1016/j.envint.2018.05.014 (2018). [DOI] [PubMed] [Google Scholar]
  • 20.Bernard, T. E. & Iheanacho, I. Heat index and adjusted temperature as surrogates for wet bulb globe temperature to screen for occupational heat stress. J. Occup. Environ. Hyg.12, 323–333. 10.1080/15459624.2014.989365 (2015). [DOI] [PubMed] [Google Scholar]
  • 21.Tang, H. et al. Compound extreme events and health risks in China: A review. Atmos. Ocean. Sci. Lett.18, 100647. 10.1016/j.aosl.2025.100647 (2025). [Google Scholar]
  • 22.Yao, H. et al. Changes caused by human activities in the high health-risk hot-dry and hot-wet events in China. Commun. Earth Environ.5, 464. 10.1038/s43247-024-01625-y (2024). [Google Scholar]
  • 23.Han, T. et al. Seasonal prediction of midsummer compound heat-humidity events over Southeast China. Atmos. Res.308, 107541. 10.1016/j.atmosres.2024.107541 (2024). [Google Scholar]
  • 24.Wang, Y.-J. & Qin, D.-H. Influence of climate change and human activity on water resources in arid region of Northwest China: An overview. Adv. Clim. Change Res.8, 268–278. 10.1016/j.accre.2017.08.004 (2017). [Google Scholar]
  • 25.Yang, P., Xia, J., Zhan, C., Zhang, Y. & Hu, S. Discrete wavelet transform-based investigation into the variability of standardized precipitation index in Northwest China during 1960–2014. Theor. Appl. Climatol.132, 167–180. 10.1007/s00704-017-2063-3 (2018). [Google Scholar]
  • 26.Yuan, Q. et al. Spatio-temporal variation of the wet-dry conditions from 1961 to 2015 in China. Sci. China. Earth. Sci.60, 2041–2050. 10.1007/s11430-017-9097-1 (2017). [Google Scholar]
  • 27.Statistics, N. B. O. China statistical yearbook. China statistics press (2021).
  • 28.Lei, X. et al. Evaluation of CMIP6 models and multi-model ensemble for extreme precipitation over Arid Central Asia. Remote. Sens.15, 2376. 10.3390/rs15092376 (2023). [Google Scholar]
  • 29.O’Neill, B. C. et al. A new scenario framework for climate change research: The concept of shared socioeconomic pathways. Clim. Change.122, 387–400. 10.1007/s10584-013-0905-2 (2014). [Google Scholar]
  • 30.O’Neill, B. C. et al. The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6. Geosci. Model. Dev.9, 3461–3482. 10.5194/gmd-9-3461-2016 (2016). [Google Scholar]
  • 31.Hersbach, H. et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc.146, 1999–2049. 10.1002/qj.3803 (2020). [Google Scholar]
  • 32.Lin, J. et al. Outstanding performance of ERA5 reanalysis temperature in China since the 1950s and quantification of its abnormal error in 1965–1966. Q. J. R. Meteorol. Soc.10.1002/qj.5042 (2025). [Google Scholar]
  • 33.Jiao, D., Xu, N., Yang, F. & Xu, K. Evaluation of spatial-temporal variation performance of ERA5 precipitation data in China. Sci. Rep.11, 17956. 10.1038/s41598-021-97432-y (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Rai, M. et al. Future temperature-related mortality considering physiological and socioeconomic adaptation: A modelling framework. Lancet. Planet. Health.6, e784–e792. 10.1016/S2542-5196(22)00195-4 (2022). [DOI] [PubMed] [Google Scholar]
  • 35.Madaniyazi, L. et al. Seasonality of mortality under climate change: A multicountry projection study. Lancet. Planet. Health.8, e86–e94. 10.1016/S2542-5196(23)00269-3 (2024). [DOI] [PubMed] [Google Scholar]
  • 36.Chen, H. et al. Projections of heatwave-attributable mortality under climate change and future population scenarios in China. Lancet. Reg. Health. West. Pac.28, 100582. 10.1016/j.lanwpc.2022.100582 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Chen, Y. et al. Provincial and gridded population projection for China under shared socioeconomic pathways from 2010 to 2100. Sci Data7, 83. 10.1038/s41597-020-0421-y (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Gasparrini, A., Armstrong, B. & Kenward, M. G. Distributed lag non-linear models. Stat. Med.29, 2224–2234. 10.1002/sim.3940 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Armstrong, B. et al. The role of humidity in associations of high temperature with mortality: A multicountry, multicity study. Environ. Health Perspect.127, 97007. 10.1289/ehp5430 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Achebak, H., Devolder, D. & Ballester, J. Trends in temperature-related age-specific and sex-specific mortality from cardiovascular diseases in Spain: A national time-series analysis. Lancet Planet Health3, e297–e306. 10.1016/S2542-5196(19)30090-7 (2019). [DOI] [PubMed] [Google Scholar]
  • 41.Guo, Y. et al. Heat wave and mortality: A multicountry, multicommunity study. Environ. Health Perspect.125, 087006. 10.1289/ehp1026 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Li, J., Wang, S., Zhu, J., Wang, D. & Zhao, T. Accelerated shifts from heatwaves to heavy rainfall in a changing climate. npj Clim. Atmos. Sci.8, 214. 10.1038/s41612-025-01113-w (2025). [Google Scholar]
  • 43.Yin, Z. et al. Mortality risk associated with heatwave exposure based on daily maximum, minimum, and combined temperature thresholds. Eur. J. Public Health10.1093/eurpub/ckaf199 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Tao, J. et al. Quantifying localized heatwave impact on mortality: A multi-country modeling study in the Asia-Pacific region. Lancet Reg. Health West. Pac.10.1016/j.lanwpc.2025.101653 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Goad, N. & Gawkrodger, D. J. Ambient humidity and the skin: The impact of air humidity in healthy and diseased states. J. Eur. Acad. Dermatol. Venereol.30, 1285–1294. 10.1111/jdv.13707 (2016). [DOI] [PubMed] [Google Scholar]
  • 46.Wolkoff, P. Indoor air humidity, air quality, and health - An overview. Int. J. Hyg. Environ. Health221, 376–390. 10.1016/j.ijheh.2018.01.015 (2018). [DOI] [PubMed] [Google Scholar]
  • 47.Deng, P. et al. The role of high humidity on extreme-temperature-related mortality in Central China. Air Qual. Atmos. Health16, 2285–2295. 10.1007/s11869-023-01406-0 (2023). [Google Scholar]
  • 48.Sivaraj, S. et al. Heat, humidity and health impacts: How causal diagrams can help tell the complex story. Environ. Res. Lett.19, 074069. 10.1088/1748-9326/ad5a25 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Maddux, S. D., Yokley, T. R., Svoma, B. M. & Franciscus, R. G. Absolute humidity and the human nose: A reanalysis of climate zones and their influence on nasal form and function. Am. J. Phys. Anthropol.161, 309–320. 10.1002/ajpa.23032 (2016). [DOI] [PubMed] [Google Scholar]
  • 50.Engebretsen, K. A., Johansen, J. D., Kezic, S., Linneberg, A. & Thyssen, J. P. The effect of environmental humidity and temperature on skin barrier function and dermatitis. J. Eur. Acad. Dermatol. Venereol.30, 223–249. 10.1111/jdv.13301 (2016). [DOI] [PubMed] [Google Scholar]
  • 51.Zhang, K. et al. Increased heat risk in wet climate induced by urban humid heat. Nature617, 738–742. 10.1038/s41586-023-05911-1 (2023). [DOI] [PubMed] [Google Scholar]
  • 52.Liang, C. et al. The influence of humid heat on morbidity of megacity Shanghai in China. Environ. Int.183, 108424. 10.1016/j.envint.2024.108424 (2024). [DOI] [PubMed] [Google Scholar]
  • 53.Yang, J. et al. Heatwave and mortality in 31 major Chinese cities: Definition, vulnerability and implications. Sci. Total Environ.649, 695–702. 10.1016/j.scitotenv.2018.08.332 (2019). [DOI] [PubMed] [Google Scholar]
  • 54.Green, H. et al. Impact of heat on mortality and morbidity in low and middle income countries: A review of the epidemiological evidence and considerations for future research. Environ. Res.171, 80–91. 10.1016/j.envres.2019.01.010 (2019). [DOI] [PubMed] [Google Scholar]
  • 55.Yin, P. et al. The added effects of heatwaves on cause-specific mortality: A nationwide analysis in 272 Chinese cities. Environ. Int.121, 898–905. 10.1016/j.envint.2018.10.016 (2018). [DOI] [PubMed] [Google Scholar]
  • 56.Huang, C. et al. Mortality burden attributable to heatwaves in Thailand: A systematic assessment incorporating evidence-based lag structure. Environ. Int.121, 41–50. 10.1016/j.envint.2018.08.058 (2018). [DOI] [PubMed] [Google Scholar]
  • 57.Kamal, A., Fahim, A. K. F. & Shahid, S. Changes in wet bulb globe temperature and risk to heat-related hazards in Bangladesh. Sci. Rep.14, 10417. 10.1038/s41598-024-61138-8 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Bandiera, D. et al. Heat-related risk at Paris 2024: A proposal for classification and review of International Federations policies. Br. J. Sports Med.58, 860–869. 10.1136/bjsports-2024-108310 (2024). [DOI] [PubMed] [Google Scholar]
  • 59.Ueno, S., Hayano, D., Noguchi, E. & Aruga, T. Investigating age and regional effects on the relation between the incidence of heat-related ambulance transport and daily maximum temperature or WBGT. Environ. Health Prev. Med.26, 116. 10.1186/s12199-021-01034-z (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Gasparrini, A. et al. Temporal variation in heat-mortality associations: A multicountry study. Environ. Health Perspect.123, 1200–1207. 10.1289/ehp.1409070 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Ebi, K. L., Ogden, N. H., Semenza, J. C. & Woodward, A. Detecting and attributing health burdens to climate change. Environ. Health Perspect.125, 085004. 10.1289/EHP1509 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Ridder, N. N., Ukkola, A. M., Pitman, A. J. & Perkins-Kirkpatrick, S. E. Increased occurrence of high impact compound events under climate change. npj Clim. Atmos. Sci.5, 3. 10.1038/s41612-021-00224-4 (2022). [Google Scholar]
  • 63.Liu, Y. et al. Evaluation and future projection of compound extreme events in China using CMIP6 models. Clim. Change178, 11. 10.1007/s10584-024-03856-6 (2025). [Google Scholar]
  • 64.Zscheischler, J. et al. Future climate risk from compound events. Nat. Clim. Change.8, 469–477. 10.1038/s41558-018-0156-3 (2018). [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information. (604.2KB, docx)

Data Availability Statement

The ambulance dispatch data that support the findings of this study are available from institutions such as the Centers for Disease Control (CDC), and the Medical Information Centers. However, restrictions apply to the use of these data, which were used under license for the current study and thus are not publicly available. The data are, however, available upon request and with the permission of these institutions.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES