Health vulnerability denotes the sensitivity, exposure, and adaptation capability of individuals or groups to climate change, reflecting the responsiveness of health outcomes to climate-related dangers and the efficacy of coping strategies (26). The key elements of developing a health vulnerability assessment system is the thorough integration of multidimensional influencing elements and the precision of quantification. Researchers, both nationally and globally, have investigated diverse methodologies for assessing health vulnerability, while the primary dimensions of evaluation frameworks frequently vary. First, a common method for developing indicator systems is based on dimensional composition. Researchers generally consider aspects such as physiological health, environmental factors, and socioeconomic circumstances (27–30). Although these studies include essential aspects such as physiological health, environmental conditions, and socioeconomic status, the complexity and subjectivity of the conclusions may stem from the weighting and interplay of various dimensions. Secondly, regarding measurement tools and methodologies, several academics have developed vulnerability assessment frameworks by considering health determinants and health outcomes as two essential components (31, 32). These frameworks utilize diverse methodologies, including analysis of variance (ANOVA) (31), empirical orthogonal function (EOF) analysis (32), equal weighting methods (33), principal component analysis (PCA) (33, 34), and dimensionality reduction statistical methods (35), among others. Certain scholars have performed evaluations about the origins of health impact exposure, highlighting the detrimental effects of these sources on individual health. These studies rigorously analyze potential health risks linked to exposure sources by examining criteria such as disaster intensity, geographical extent, frequency, and duration. The objective is to investigate the correlation between exposure sources and health outcomes, encompassing morbidity (32, 36), death (31, 36), and individual health status (33). Prevalent methodologies encompass expert assessment (26), case study approaches (36, 37), and more ways. These assessment approaches are relevant across different spatial scales, covering a broad range of subjects, including regions (33, 38), communities (39), and cities (40). The evaluations encompass multiple forms of health impact exposure, including air pollution (41), flooding (33, 34), elevated temperatures (34, 36), and increasing sea levels (37).

Prior research has established a basis for the development of health vulnerability assessment systems; nevertheless, there remains room for further exploration. Primarily, regarding the design of health indicators, the majority of experts have concentrated mostly on physiological wellness (26, 31–37). This research, however, comprehensively examines physiological health, psychological health, and health habits. Both physical and mental health substantially affect vulnerability, whereas health habits determine long-term health consequences. At the macro level, researchers generally take into account geographical or habitat conditions regarding environmental elements. Geographic location influences healthcare accessibility, and environmental quality directly impacts health. Secondly, regarding research perspective, this study concentrates on the migrant worker demographic. Prior research has examined vulnerability from multiple viewpoints, such as the older adult (42–45) and children (46–49), or have examined migrant workers’ health vulnerability. Chinese migrant workers constitute a unique demographic, arising from particular socioeconomic and policy contexts with significant contributions to social development. Nonetheless, scholarly investigation on this cohort remains very nascent. Thirdly, regarding data authority and generalizability, a substantial number of previous studies have depended on questionnaires or interviews administered by research teams focusing on particular regions or populations (26, 36, 37). Although these research provide useful insights, their findings frequently lack generalizability. This study, different from this, utilizes authoritative public databases from nationwide surveys done by reputable institutions, hence augmenting the universality and trustworthiness of the research findings.

The concept of the “migrant worker” emerged in the early stages of China’s economic reforms as a response to surplus rural labor, reflecting the migration patterns within China’s urban–rural dual structure (50). Current studies on the correlation between temperature variations and the well-being of migrant workers predominantly emphasize the effects of elevated temperatures (51). Many migrant workers, mainly in the construction sector, are subjected to elevated temperatures for extended durations, resulting in heat stress, profuse sweating, weariness, headaches, and diminished work efficiency and social interaction (52). Exposure to elevated temperatures heightens the risk of heat-related ailments, such as heat exhaustion, dehydration, and potentially heatstroke (53). Prolonged exposure to high temperatures can also heighten the risk of acquiring respiratory and cardiovascular problems (54, 55). Moreover, increasing temperatures have been demonstrated to expedite the transmission of some diseases, broadening their geographical range (56). Certain research have investigated the effects of cold temperatures on the health of migratory laborers. Migrant workers in the fishing business encounter increased risks of injury and disease in colder environments (57). During the COVID-19 pandemic, reduced temperatures heightened the vulnerability of migrant workers to the virus within healthcare institutions, as they regularly operated on the frontlines, often interacting with patients and their close contacts (58). The majority of migratory laborers either lack the awareness to acquire or opt not to obtain formal work injury and unemployment insurance to conserve funds, thus, when in instances of health issues resulting from temperature variations, compensation is severely limited (59). This leads to an inability to finance medical care, exacerbating their physical health situation. Moreover, substantial temperature variations intensify the psychological strain on migrant laborers, in addition to their direct effects on physical health. Temperature fluctuations, coupled with occupational stress, frequently result in anxiety and tension, hence heightening the likelihood of mental health disorders (60). Although current research has explored the impact of temperature changes on the health of migrant workers, it predominantly concentrates on certain diseases or psychological stressors. Research focused on the health vulnerabilities of migrant workers concerning temperature variations is limited. Therefore, this paper takes temperature changes as the focal point, specifically examining their impact on the health vulnerability of migrant workers, filling this gap in the literature. This study additionally examines the mechanisms by which temperature change influence the health vulnerability of migrant workers by incorporating psychological burden as a mediating variable, thereby developing an impact mechanism model grounded in relevant theoretical frameworks.

Multiple research have underscored the substantial effects of temperature change on both physical and mental health, affected by many factors (4, 61). The mechanisms governing body temperature indicate that mental health may be affected by environmental temperatures, as specific neurotransmitters, including biogenic amines, are involved in mood regulation and thermoregulation (62). Increasing temperatures facilitate the dissemination of vector-borne diseases (63) and aquatic illnesses (64), while simultaneously promoting the proliferation of fungi and molds, hence elevating the incidence of respiratory and asthma-related conditions. Extreme temperatures, whether elevated or reduced, have adverse impacts on both the body and the mind. Besides their physical impact, severe temperatures provide significant psychological difficulties. Increased temperatures, specifically, have been demonstrated to heighten psychological distress, whilst decreased temperatures frequently mitigate certain adverse mental conditions (65). Temperature fluctuations may intensify risk factors for established mental health issues (66), with extended exposure to elevated temperatures linked to increased psychological stress and mental tiredness (67). The impacts are particularly significant among migratory workers, who frequently endure unstable working conditions with restricted access to suitable housing, adequate rest intervals, or opportune psychiatric assistance (53). The interplay of environmental stresses and socioeconomic pressure amplifies psychological vulnerability, worsening mental health conditions (68). Vulnerability is often a matter of perception, as different populations experience and respond to temperature changes differently, leading to varying levels of psychological burden (69). The stress induced by temperature changes compromises the immune system, rendering migrant workers more vulnerable to sickness (70). Furthermore, the psychological burden of coping with extreme temperatures frequently leads to maladaptive coping mechanisms, like heightened alcohol intake or unhealthy eating practices, which exacerbate general health deterioration (71). This research empirically investigates the effects of temperature change on the health vulnerability of migratory workers and explores the underlying mechanisms. The objective is to elucidate the mechanisms by which temperature fluctuations affect health vulnerability, so establishing a scientific foundation for the development of pertinent policies.

The health disparities between older and newer generations of migrant workers have been well documented (72). Temperature has also emerged as an important factor influencing these differences, as research demonstrates that fluctuations in temperature exposure during pregnancy can variably affect the health of subsequent generations, with elevated temperatures exerting a more adverse impact on birth outcomes (73–75). Firstly, significant intergenerational differences exist in the physical health of migrant workers. Research indicates that the older cohort of migrant workers, who migrated to urban areas in the 1980s and 1990s, encountered more unfavorable working circumstances, possessed little health knowledge, and had insufficient access to appropriate healthcare facilities. In comparison to the general population, the absence of prompt healthcare or the neglect to disclose health concerns frequently intensified their health vulnerability (76). The newer generation of migrant workers, who have relocated to urban areas, generally possess elevated educational qualifications and enhanced access to health-related information, so contributing to their overall health improvement (77). Besides physical health, notable intergenerational disparities are also seen in the psychological well-being of migratory workers. The newer generation of migrant workers faces unique challenges due to the effects of China’s economic reforms and the one-child policy (78). Factors include diminished social support, reduced marriage rates, a disparity between expectations and reality, and challenges in urban integration have rendered the psychological well-being of the younger generation more susceptible than that of their older counterparts (79–81).

In summary, whereas current research has made significant advancements and laid a solid foundation for this study, several research gaps remain to be addressed.

Firstly, the correlation between temperature changes and the health vulnerability of migrant workers remains little investigated. This research improves the assessment of health vulnerability by utilizing micro-level data and developing a health vulnerability index grounded in the social-ecological model, building upon prior studies. Health vulnerability is intricately linked to an individual’s social background and lifestyle choices. Our concept encompasses various health dimensions, including physical health, mental health, habits, social support, and environmental exposure, situating individual health within a wider social and ecological framework. This methodology uncovers the intricate relationships between health and social-ecological factors (82, 83), facilitating a more refined evaluation of the effects of temperature variations on health.

Secondly, this work presents a novel research perspective. Migrant workers merit attention for their substantial contributions to socio-economic development; nonetheless, their socioeconomic position is relatively low. While certain researchers have examined the health vulnerabilities of migrant workers, these studies have primarily focused on international migrant workers. However, within the context of different countries, Chinese migrant workers possess distinct traits and qualities compared to global migrant labor populations. Research on migrant laborers from foreign nations is not universally relevant to China’s distinct situation. This research addresses such deficiency.

Lastly, this research examines both the immediate effects of temperature change on health vulnerability and the underlying mechanisms involved. By investigating how social, economic, and environmental conditions interact with temperature changes, the study provides fresh insights into the pathways through which climate impacts migrant workers’ health vulnerability. Given the significant intergenerational differences between older and newer generations of migrant workers, the study further examines how temperature change differentially affect the health vulnerability of these two groups.

The individual sample data for this study originates from the 2020 China Family Panel Studies (CFPS) conducted by the China Social Science Survey Center at Peking University. Meteorological data is obtained from the National Meteorological Science Data Center of China, which provides daily data on fundamental meteorological elements from more than 2,400 stations nationwide, encompassing temperature, air pressure, relative humidity, precipitation, wind direction and velocity, and sunlight. This study utilizes county-level administrative codes from the CFPS data to identify respondents’ actual residences and use GIS technology to physically correlate meteorological station data with respondents’ locations using latitude and longitude coordinates. In covered by multiple meteorological stations, the average data from these stations is utilized to more precisely represent the regional climate features. Utilizing the methodologies of Currie and Neidell (84), Deschênes and Greenstone (85), and Schlenker and Walker (86), we additionally employed the Inverse Distance Weighting (IDW) technique to interpolate daily meteorological data, thereby improving the continuity and comprehensiveness of the dataset, specifically mitigating potential biases stemming from the irregular distribution of meteorological stations. Furthermore, we consolidated the daily meteorological data into seasonal data and matched it to the city-year CFPS data to maintain consistency across the temporal dimension. We standardized the geographic coding of the CFPS and meteorological datasets to ensure data consistency, thereafter conducting data cleaning and verification. Missing and outlier values were omitted, and the matching logic was verified using random sampling. Ultimately, by integrating macroeconomic data, we acquired 3,930 valid observations at the individual-city level, so ensuring the correctness and scientific rigor of the data.

This study utilizes cross-sectional data and employs a multivariable regression model to examine the effects of temperature change on the health vulnerability of migratory workers. This model accounts for both individual and city-level attributes, with standard errors clustered at the individual level. The multivariable regression model is formulated as follows: (1) H V u , i = α 0 + β 1 T e m p u , i + δ v X u , i + e u , i

In Equation 1, the subscript u denotes the city, while i signifies the individual sample. HV represents the dependent variable. X u , i signifies the control variables, encompassing gender, age, years of education, marriage, family size, health status, sunshine duration, precipitation, average wind speed, dew point temperature, atmospheric pressure, regional GDP, industrial structure, and the number of health and social welfare workers per 10,000 population. e u , i denotes the random disturbance term, α 0 signifies the constant term, β 1 represents the coefficient for the explanatory variable, and δ v indicates the coefficients of the control variables.

Table 2 displays the descriptive statistics of the variables. The findings indicate that the mean value of HV is 0.3726, accompanied by a standard deviation of 0.2423, signifying considerable variability in individual health vulnerability within the sample. The mean temperatures for spring, summer, autumn, and winter are 14.77°C, 24.26°C, 14.07°C, and 2.12°C, respectively. The winter temperature exhibits the highest standard deviation, indicating the greatest regional variability in winter temperatures. Gen and Mar are binary variables, with means of 0.359 and 0.866, respectively, signifying that 35.9% of the sample is male and 86.6% is married. The mean age is 48.89 years, while the average years of schooling is 8.43 years. The mean FS is 3.65, accompanied by a substantial standard deviation, indicating variability in family sizes within the sample. The average Health is 0.706, suggesting that the majority of persons in the sample self-identify as healthy. Environmental data, including sunshine duration, precipitation, wind speed, dew point temperature, and atmospheric pressure, have been normalized to have averages around 0 and standard deviations near 1. In terms of economic variables, the average regional GDP is 16.69, the mean IS value is 0.60, and HSWW per 10 k has a mean value of 3.55, reflecting the variation in economic and social security levels across different cities.

To evaluate the model’s robustness, we conducted the Pesaran cross-sectional dependence test (Pesaran CD Test) on both the baseline regression model and the mechanism verification model to determine the presence of cross-sectional dependence among the sample units. The p-values for the three models were 0.15, 0.21, and 0.22 (Table 3). According to the null hypothesis of the Pesaran test (H0: no cross-sectional dependency) and the p-values obtained, we fail to reject the null hypothesis, suggesting the absence of significant cross-sectional dependence in the models. This suggests that there is no significant cross-sectional association in the health vulnerability of migratory workers within the modeling framework of the present study. The random disturbances between different individuals are relatively independent. Consequently, the outcomes from our regression model can utilize two-way fixed effects regression estimates, which possess high robustness and scientific validity, without necessitating further model adjustments to resolve cross-sectional dependence concerns.

The baseline regression results (Table 3, Column 1) indicate that increasing spring and summer temperatures significantly enhance health vulnerability, with coefficients of 0.0958 and 0.0556, respectively, both significant at the 1% level. The effect of spring temperature increases is more pronounced. Winter temperature increases exert a notable positive influence on health vulnerability, but to a lesser extent, with a coefficient of 0.0169, significant at the 1% level. Conversely, increasing fall temperatures have a substantial negative correlation with health vulnerability, with a coefficient of −0.1960, which is significant at the 1% level, suggesting that higher fall temperatures markedly diminish health vulnerability. We also account for many additional characteristics that may influence the health susceptibility of migratory workers, including gender, age, years of education, marital status, family size, health status, and sunshine duration. Gender and age are significant at the 1% level, indicating disparities in health vulnerability among migrant workers of varying genders and ages.

The mechanism testing revealed that psychological strain partially mediates the association between temperature fluctuations and the health vulnerability of migrant workers. In model (3) (Table 3, Column 3), the coefficient for the influence of psychological burden on health vulnerability is 0.0514, significant at the 1% level, signifying that an increase in psychological burden correlates with an increase in health vulnerability. In model (2) (Table 3, Column 2), the coefficient for the impact of increasing spring temperatures on psychological burden is 0.2822, significant at the 1% level, indicating that for each unit increase in spring temperature, psychological load rises by approximately 0.282 units. The coefficients for summer, fall, and winter temperatures on psychological burden are −0.0342, −0.2196, and − 0.1856, respectively, and are statistically significant at the 5, 1, and 1% levels. The results indicate that temperature increases over these three seasons alleviate psychological strain, with the greatest significant relief observed with rising temperatures in the fall. Furthermore, sunshine length exerts a substantial positive influence on health vulnerability, evidenced by a coefficient of 0.2701, indicating that extended sunshine durations correlate with increased health vulnerability. This may be associated with heightened stress or other health complications resulting from extended exposure to sunshine.

This study performed a heterogeneity analysis between the new and old generations of migrant workers, with the new generation being individuals born in 1980 or later, and the old generation consisting of those born prior to 1980 (121). The “post-1980 migrant workers,” born in the 1980s and 1990s, were classified according to the pivotal historical setting of China’s reform and opening-up, which commenced approximately at that period, is also known as the contemporary industrial workers and rural builders. The living experiences of this generation significantly diverge from those of their parents, influenced by swift economic advancement, more global engagement, the implementation of the one-child policy, and the proliferation of the internet and globalization. The “new generation of migrant workers” denotes a subset of the “post-80s” cohort, characterized by their distinct life experiences—lacking substantial agricultural background, reluctant to reside permanently in rural locales, yet finding it difficult to change their rural identity and encountering challenges in assimilating into urban society while enduring uncomfortable living conditions (122). Their experience of migrating to cities contrasts differs from that of the previous migrant workers who relocated to cities during the initial stage of China’s reform. The current generation has unique demographic compositions, social attitudes, identity dimensions, and work-life aspirations (123). Table 4 displays the outcomes of the heterogeneity test. The findings indicate that the coefficients representing the effect of increasing spring temperatures on the health vulnerability of both new and old generation migrant workers are 0.0444 and 0.1249, respectively, with both being statistically significant at the 1% level. This suggests that increasing spring temperatures markedly elevate health susceptibility for both cohorts, with a diminished effect on the younger generation. The coefficients indicating the impact of increasing summer temperatures on health vulnerability are 0.0320 for the younger generation and 0.0175 for the older generation, both statistically significant at the 1% level. This indicates that increasing summer temperatures intensify health risk, particularly affecting the new generation migrant workers more significantly. In the autumn, the coefficients indicating the effect of increasing temperatures on health vulnerability are −0.1030 for the younger generation and − 0.1451 for the older generation, both statistically significant at the 1% level. This suggests that increasing fall temperatures reduce health vulnerability, particularly benefiting older migratory workers. In the case of winter temperatures, the coefficients are negative for both groups, but only statistically significant for the new generation, with a coefficient of −0.0204 at the 10% level. This suggests that rising winter temperatures have a mild alleviating effect on the health vulnerability of the new generation, though the effect is relatively weak.

Table 5 displays the outcomes of the robustness tests. To validate the empirical findings, two kind of robustness checks were performed: (1) Replacing the Dependent Variable: In the robustness assessments, health vulnerability was determined utilizing the principal component analysis (PCA) technique. The findings demonstrate that spring, summer, and winter temperatures have a notably positive impact, whereas fall temperatures dramatically diminish health vulnerability. These data indicate that the influence of seasonal temperature variations on health vulnerability remains consistent even after altering the dependent variable. (2) Modifying the Model: The second robustness test utilized a multilevel model to consider both individual-level and city-level influences. In contrast to the baseline regression model, the multilevel approach is more suitable for managing hierarchical data structures, such as the layered relationships between cities and individuals. It encapsulates the interplay between individual and urban attributes, yielding more precise estimates and accounting for correlations among individuals at the same level, thus minimizing estimation errors (124). The findings indicate that the impacts of spring, summer, and winter temperatures are no longer substantial, whereas autumn temperature persists in exerting a major negative influence. Control factors, including gender, age, and family size, exhibit comparable levels of significance in both models. Variables such as health status and cumulative precipitation maintain consistent signals and significance across both robustness assessments, hence affirming the validity of the initial findings.