https://orcid.org/0000-0002-1155-8443
Figures
Abstract
While meteorological and socioeconomic factors are well-documented modifiers of spatial heterogeneity in temperature–hand, foot, and mouth disease (HFMD) associations, substantial unexplained heterogeneity remains. This study investigates underexplored environmental modifiers—including extreme temperature events (heat waves and cold spells), air pollution, and normalized difference vegetation index (NDVI)—by analyzing 484,928 HFMD cases among children under six years old in western China (2013–2019) using distributed lag nonlinear models and meta-regression. We found that cold spells (accounted for 3.84% of the spatial heterogeneity attributable above the baseline level), PM2.5 (3.06%), heat waves (2.72%), PM10 (2.08%), NDVI (1.57%) and O3 (0.78%) were statistically significant modifiers of spatial heterogeneity in the temperature-HFMD associations. Further analysis of PM2.5 components identified nitrate (1.78%) and ammonium (1.58%) as additional modifiers. Among these, cold spells, heat waves, PM2.5, and NDVI were the primary contributors. Specifically, the relative risk (RR) of HFMD at the 95th temperature percentile reached 3.17 (95% CI: 2.39–4.19) in frequent heat waves regions and 3.05 (2.35–3.95) in high-PM2.5 areas. Conversely, regions with low cold-spell frequency also exhibited increased temperature-related risk (RR = 3.13, 2.40–4.07) at the same temperature, as did low NDVI regions (RR = 2.16, 1.79–2.59). Spatial cluster analysis further revealed that the central and northeastern regions exhibited higher temperature-related HFMD risks compared to the southwestern region. These pronounced spatial modification effects challenge the generalizability of single-site study and highlight the importance of region-specific public health strategies that integrate early warning systems for extreme temperatures, air pollution mitigation, and locally adapted greening interventions.
Author summary
Hand, foot, and mouth disease (HFMD) is a common illness among young children, especially in Asia. Although temperature is known to influence the risk of HFMD, the relationship between temperature and disease varies from place to place. In this study, we analyzed over 480,000 HFMD cases in children under six years old across 88 counties in western China. We explored how regional environmental conditions—such as heat waves, cold spells, air pollution, and green space—might explain these differences. We found that areas with more heat waves or higher levels of air pollution had greater HFMD risk during hot weather. In contrast, regions with more green space or frequent cold spells showed weaker temperature effects. We also used clustering methods to group regions with similar environmental patterns and found that some areas were more vulnerable to temperature-related HFMD than others. These results show that local environmental conditions play an important role in shaping disease risk and suggest that region-specific prevention strategies may help reduce HFMD in children.
Citation: Sun J, Yao G, Gu J, Tang H, Wu Y, Chang Y, et al. (2026) Spatial heterogeneity in the temperature–hand, foot, and mouth disease association among children: A multicounty time-series study in western China. PLoS Negl Trop Dis 20(1): e0013801. https://doi.org/10.1371/journal.pntd.0013801
Editor: David Safronetz, Public Health Agency of Canada, CANADA
Received: May 23, 2025; Accepted: November 26, 2025; Published: January 2, 2026
Copyright: © 2026 Sun et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The HFMD case data used in this study were obtained from the Chinese Center for Disease Control and Prevention (China CDC) (http://www.chinacdc.cn/en/) under official approval. As these data are subject to access restrictions, they are not publicly available. Researchers interested in accessing the HFMD data may contact the China CDC via phone (+86-010-58900412) or email (data@chinacdc.cn) to obtain permission through the same procedure. The environmental data used in this study are publicly available. Meteorological data can be accessed from the China Meteorological Data Sharing Service (http://data.cma.cn/site/index.html), and air pollution data are available from the China Pollutants in the Atmosphere (CHAP) dataset (https://weijing-rs.github.io/product.html). The environmental data relevant to this study are provided in the Supporting information files (S1 Data and S2 Data).
Funding: This work was supported by the National Natural Science Foundation of China (82103947 to ZD), the Science and Technology Program of Guangzhou, China (202206080003 and 2024A04J4623 to ZD), and the Fundamental Research Funds for the Central Universities, Sun Yat-sen University (23qnpy105 to ZD). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Hand, foot and mouth disease (HFMD) is a highly prevalent infectious disease of childhood caused by enteroviruses and characterized primarily by fever, oral ulcers and vesicular rashes [1]. Although most cases are mild and self-limiting, severe complications such as encephalitis and cardiopulmonary failure can occur [2]. Since the late 1990s, recurrent large-scale outbreaks have caused a significant burden of disease throughout the Asia-Pacific region [3,4]. Across eight high-burden Asia-Pacific countries and regions, including mainland China, HFMD contributes to an estimated 96,900 age-adjusted DALYs annually [5]. In mainland China, HFMD has been notifiable since 2008, with over 22 million cases and 3,500 deaths reported between 2008 and 2019 [6,7]. In the absence of targeted antiviral treatment [8], strengthening prevention remains a public health priority. Understanding the factors that influence HFMD transmission is therefore essential for effective control strategies.
Ambient temperature is an important environmental determinant of HFMD risk and has been extensively studied. A systematic review and meta-analysis of 11 studies from five countries confirmed a significant positive association between temperature and HFMD incidence [3]. However, substantial variation in effect estimates was observed both between and within countries, as reported in previous studies [9–11]. These inconsistencies arise from heterogeneity in population demographics, exposure definitions, statistical models, and study designs. In addition to methodological factors, recent research has increasingly focused on the role of meteorological and socioeconomic variables in shaping the spatial heterogeneity of temperature-HFMD associations [11–13]. For example, a multicenter study of 143 Chinese cities found that factors such as temperature, relative humidity, atmospheric pressure, sunshine duration, and GDP per capita explained only 0.01% to 2.0% of the observed spatial variation [11]. Although these factors partially account for the heterogeneity, they remain insufficient to fully explain regional disparities, suggesting the presence of additional, unmeasured environmental modifiers.
Regional temperature significantly influences the spatial heterogeneity of HFMD-environment relationships [11–13]. Extreme temperature events (ETEs), including heat waves and cold spells, have gained increasing attention due to their severe health impacts. While the exact biological mechanisms remain unclear, extreme temperatures disrupt thermoregulation and trigger systemic stress responses such as cardiovascular dysfunction, fluid imbalance, and inflammation [14]. These responses can severely compromise human health. Due to immature thermoregulatory and immune systems, children are more susceptible to extreme temperatures, increasing their risk of HFMD [15,16]. Moreover, while extreme temperatures may accelerate viral inactivation, they can also promote indoor crowding, which increases the risk of HFMD transmission [17]. However, the role of extreme temperatures in driving HFMD spatial heterogeneity remains unclear. In addition to temperature, air pollution may modify the temperature–HFMD relationship. Accumulating evidence links pollutants such as PM2.5, PM10, and O3 to increased HFMD risk [18,19]. The small particle size and large surface area of PM2.5 and PM10 enable them to absorb toxic substances and penetrate the alveoli, inducing oxidative stress and inflammatory responses that impair antiviral defenses [20,21]. Additionally, these particles may act as carriers for enteric viruses, increasing airborne exposure and facilitating HFMD transmission [22,23]. O3, a potent oxidizing agent, can damage respiratory mucosal barriers and impair host immunity, further increasing susceptibility to HFMD, although it can also inactivate viruses [24,25]. Despite these insights, the influence of air pollution on HFMD spatial heterogeneity has received limited attention. Given its ubiquity and potential synergistic effects with temperature, air pollution may contribute to regional variations in the temperature-HFMD relationship.
In addition, traditional economic factors, such as population density and GDP per capita, have been shown to be closely associated with HFMD [11,12]. Nighttime light index and Normalized Difference Vegetation Index (NDVI) are modern indicators that correlate strongly with these economic factors. Nighttime light not only serves as a proxy for economic activity but also reflects urbanization levels, energy consumption, and human activity intensity [26]. NDVI, an indicator of regional green space, may influence HFMD risk by modifying behavioral exposures and physiological recovery [27]. While these indicators offer novel insights, their specific roles as modifiers of the temperature-HFMD relationship remains limited, especially in high-burden regions such as Guizhou Province that face a distinct scarcity of epidemiological evidence regarding multiple environmental exposures [28,29].
To address these gaps, this study introduces ETEs, PM2.5, PM10, O3, nighttime light, and NDVI to broaden the scope of regional heterogeneity research. We applied a Distributed Lag Nonlinear Model (DLNM) combined with meta-regression to analyze the HFMD time-series data from 88 counties in Guizhou Province in western China. The aim was to quantify the modifying effects of these environmental factors on the temperature-HFMD relationship. Additionally, spatial clustering analysis based on these environmental modifiers was performed to identify distinct regional patterns in temperature-HFMD associations.
Materials and methods
Ethics statement
The HFMD data used in this study was obtained from the Guizhou Provincial Center for Disease Control and Prevention. The dataset did not contain any personally identifiable information, such as names, identification numbers, or contact details. No individual-level interventions were conducted, and there was no direct contact with participants. This study was conducted at the county level using spatially aggregated data and did not involve any individual-level analysis or operations. In accordance with established ethical guidelines and practices for public health surveillance research, formal ethical approval and informed consent were not required.
Study area and population
This study analyzed 484,928 HFMD cases among children under six years old in Guizhou Province, China, from 2013 to 2019, using data from the China Disease Prevention and Control Information System. Guizhou, located in western China (103°36′–109°35′E, 24°37′–29°13′N), covers 176,200 km2, with 88 counties and a population of 38.56 million as of November 1, 2020. Daily case counts were aggregated based on the first date of symptom onset for each county to support spatio-temporal analysis.
Map data
The Guizhou Province base map was sourced from the National Platform for Common GeoSpatial Information Services (Tianditu platform; https://cloudcenter.tianditu.gov.cn/administrativeDivision) in Chinese, under the Map Review Approval Number: GS (2024) 0650.
Meteorological data
Meteorological data were sourced from the China Meteorological Data Sharing Service (http://data.cma.cn/site/index.html), including daily measurements of temperature, air pressure, precipitation, relative humidity, wind speed, and sunshine duration from 31 national surface meteorological stations in Guizhou Province. Spatial interpolation of station data to 88 county-level administrative units across the province was performed using the nearest neighbor matching method to ensure geographical accuracy.
Following previous research practices, this study extends the analysis by treating ETEs as regional environmental factors [11,12]. Specifically, heat waves and cold spells were examined [30,31], with heatwave days defined as consecutive days (D ≥ 3 days) with daily maximum temperature (Tmax) above the 95th percentile threshold (Tmax ≥ 95th percentile), whereas cold spell days were identified as sustained periods (D ≥ 3 days) with daily minimum temperature (Tmin) below the 5th percentile threshold (Tmin ≤ 5th percentile). Compared to singular temperature metrics, ETEs provide a multidimensional characterization of event intensity, frequency, and duration [14]. Binary indicators (1 for event days, 0 otherwise) were used to quantify ETE occurrences, with cumulative counts of heatwave and cold spell days subsequently analyzed at the county level [32].
Other County-level data
Air pollution data were derived from the China Air Pollutants Dataset (CHAP) (https://weijing-rs.github.io/product.html), which provides daily PM2.5 and PM10 concentrations at 1 km resolution, and O3 concentrations at 10 km resolution. These data were generated using the Spatial-Temporal Extremely Random Tree model, which has been validated for accuracy and reliability [33–35]. In addition, data on selected PM2.5 components—including ammonium ( ), nitrate (
), black carbon (BC), and organic matter (OM)—were obtained from the TAP dataset (http://tapdata.org.cn/?page_id=129), provided at 10 km resolution. These constituents, although not exhaustive, represent key components relevant to health and environmental studies.
Environmental indicators of urbanization, including nighttime light and Normalized Difference Vegetation Index (NDVI), were used to represent human activity and green space, respectively [36,37]. Nighttime light data were obtained from VIIRS on board the Suomi NPP satellite, provided by the Earth Observation Group (EOG), with annual mean illuminance (1 km × 1 km resolution) after quality control to remove interference (e.g., auroras, biomass burning). NDVI data representing vegetation cover was obtained from the NASA MODIS13Q1 dataset, with values ranging from -1 to 1 and updated every 16 days at 1 km resolution. Both indicators were aggregated into county-level annual averages using GIS tools in R.
Finally, county-level socio-economic data (population density and GDP per capita) were obtained from the Guizhou Provincial Statistical Yearbook and included as covariates to account for spatial heterogeneity in development and demographics.
Study design and statistical analysis
The analysis consisted of three stages. First, county-level temperature-HFMD relationships were estimated. Second, a meta-regression analysis was performed to identify environmental factors contributing to spatial heterogeneity. Finally, cluster analysis was applied to group counties with similar heterogeneous characteristics to further explore their grouping patterns.
County-level nonlinear and lagged analysis
We utilized quasi-Poisson regression combined with DLNMs [9]; this framework utilized a two-dimensional exposure-lag-response model to characterise the relationship between temperature and HFMD, using a cross-basis function within the DLNM framework [38]. The equation represents the applied specific model for temperature exposure:
where Yt is the daily count of HFMD cases on day t, following a quasi-Poisson distribution to account for overdispersion; ut is the mean of the quasi-Poisson distribution, representing the expected number of HFMD cases on day t. To capture the non-linear exposure-response and lag-response associations, natural cubic splines (ns) with 3 degrees of freedom (df) were applied to both dimensions. The degrees of freedom were selected based on the model fit evaluated by the Quasi Akaike Information Criterion (QAIC) [39], as detailed in Supplementary Materials. To thoroughly investigate the lag structure of the temperature effect, a maximum lag of 14 days was chosen, taking into account the incubation and duration of HFMD [11,12]. To account for long-term trends and seasonal variations, natural cubic splines with 8 df per year were applied to calendar time. Moving averages of meteorological variables (relative humidity, wind speed, sunshine duration, air pressure) over the same lag period were incorporated to control for potential time-varying confounders. In addition, day of the week (dow) and holiday indicators (including school vacations and national public holidays) were included to further adjust for temporal patterns.
Meta-regression analysis of spatial heterogeneity
In the second stage, a meta-regression approach was applied to pool county-specific temperature-HFMD associations, aiming to identify environmental factors driving spatial heterogeneity [40]. Each environmental factor was first assessed individually for its contribution to heterogeneity. Factors showing significant associations were subsequently entered into a multivariate meta-regression model using a stepwise procedure. As population density and GDP per capita are well-documented determinants of infectious disease dynamics [41], these were treated as fixed covariates throughout the model selection process. To mitigate multicollinearity, we screened regional variables by excluding variable pairs with a correlation coefficient greater than 0.8 [42]. In addition to this pairwise analysis, we specifically evaluated the four key fixed covariates—nighttime light, NDVI, population density, and GDP per capita—using the Variance Inflation Factor (VIF), applying a conservative threshold of 2.5 to identify potential issues [43]. The final set of predictors was determined based on model fit assessed by the Akaike Information Criterion (AIC) and significance tested using the multivariate likelihood ratio (LR) test. Residual heterogeneity was evaluated using the I2 statistic and Cochran’s Q test [39].
Spatial clustering analysis
Based on the identified environmental drivers of heterogeneity, hierarchical cluster analysis was performed to group counties into distinct clusters with similar heterogeneity profiles. Subsequent cluster-specific analyses were conducted to identify homogeneous regions with similar environmental and socioeconomic profiles. First, variables were selected based on prior heterogeneity analysis and multicollinearity assessment. Following Z-score standardization of all continuous variables, model-based clustering was conducted using the Mclust package in R, which evaluates 14 distinct covariance parameterizations ranging from spherical to general (ellipsoidal) covariance structures. The optimal number of clusters and the corresponding covariance model were determined by maximizing the Bayesian Information Criterion (BIC). After model-based clustering, hierarchical cluster analysis was performed to group counties into distinct clusters with similar heterogeneity profiles. This clustering approach provided additional insights into spatially clustered patterns of vulnerability. For comprehensive details regarding the methodology and steps involved, refer to Supporting Information S1 Text.
Sensitivity analysis
To assess the robustness of our findings, sensitivity analyses were conducted by varying the df for the exposure-response, lag-response, and meteorological covariates (relative humidity, wind speed, sunshine duration, and air pressure) from 3 to 5. Furthermore, to evaluate the influence of heat wave days and cold spell days definitions, we applied alternative definitions: heat wave was defined as daily maximum temperature (Tmax, °C) ≥ 95th percentile for at least 2 consecutive days, while cold spell was defined as daily minimum temperature (Tmin, °C) ≤ 5th percentile for at least 2 consecutive days. These alternative definitions were incorporated into the meta-regression framework.
All analyses were conducted in R 4.2.1, using “data.table” and “dplyr” for data processing, “dlnm” for distributed lag non-linear modeling, “mvmeta” for meta-analysis, “sp” and “sf” for spatial analysis, “raster” for raster data, and “ggplot2” for visualization.
Results
Descriptive characteristics of HFMD cases and environmental exposures
Table 1 and Fig 1 present the descriptive statistics of HFMD cases and environmental factors across 88 counties in Guizhou Province from 2013 to 2019, highlighting spatial variation. A total of 484,928 HFMD cases were reported among children under six years old, with a median of 2,846 cases per county (range: 326–44,096). Population density varied considerably, ranging from 53 to 10,575 persons per km2, while GDP per capita ranged from CNY 16,251 to CNY 116,523. Air pollution levels also varied considerably, with median concentrations of 34.12 µg/m3 for PM2.5, 54.17 µg/m3 for PM10, and 75.52 µg/m3 for O3. In addition, important PM2.5 components such as ,
, BC, and OM showed substantial differences across counties (Table 1). Indicators of environmental urbanization were also highly variable, with a median nighttime light intensity of 52.30 nW/cm2/sr and a median NDVI of 0.54 (Table 1). A full list of the 88 counties included in the study is provided in the Supporting Information (S1 Table). In addition, the temporal variation of six environmental variables across Guizhou Province from 2013 to 2019 is presented in the Supporting Information (S1 Fig).
The base layers of the map were obtained from the National Platform for Common GeoSpatial Information Services via the Tianditu platform (in Chinese) (https://cloudcenter.tianditu.gov.cn/administrativeDivision).
Nonlinear temperature-HFMD association and spatial heterogeneity
As shown in Fig 2, the pooled exposure-response curve between temperature and HFMD exhibits a characteristic J-shaped pattern. Compared to the median temperature, the relative risk (RR) of HFMD at the 5th percentile temperature (-1.4 °C) was 0.85 (95% confidence interval [CI]: 0.74, 0.98), whereas at the 95th percentile temperature (27.3 °C), the RR increased sharply to 2.01 (95% CI: 1.68, 2.41). The dashed lines indicate the county-specific estimates, showing considerable variation in the temperature-HFMD associations across the 88 counties, suggesting the presence of spatial heterogeneity. Lag-response relationships are shown in Supporting Information (S2 Fig).
Note: 1) All estimates are referenced to the median temperature (12.9 °C). 2) The two vertical black lines represent the 5th (1.4 °C) and 95th (27.3 °C) percentiles of the temperature distribution. 3) The gray band represents the 95% confidence interval for the pooled effect estimate. 4) The gray dashed lines represent county-specific estimates of temperature-HFMD associations.
Environmental determinants of spatial heterogeneity in the temperature–HFMD association
Table 2 presents the multivariate meta-regression results quantifying the spatial heterogeneity in the temperature-HFMD associations. The intercept-only model indicated moderate heterogeneity (I2 = 50.08%). Six environmental variables were identified as significant contributors to spatial heterogeneity in the temperature-HFMD association. These included cold spells (p < 0.001, ΔI2 = 3.84%, representing the proportion of spatial heterogeneity above the baseline level), PM2.5 (p = 0.001, ΔI2 = 3.06%), heat waves (p = 0.001, ΔI2 = 2.72%), PM10 (p = 0.020, ΔI2 = 2.08%), NDVI (p = 0.025, ΔI2 = 1.57%), and O3 (p = 0.041, ΔI2 = 0.78%). Notably, the concentration difference between PM10 and PM2.5 was not statistically significant (p = 0.416), suggesting limited relevance in explaining spatial heterogeneity. Given the strong correlation between PM2.5 and PM10 (r = 0.97) (S3 Fig), the observed PM10 effect may largely reflect the contribution of PM2.5. Taken together, cold spells, heat waves, PM2.5, and NDVI emerged as the main contributors to spatial heterogeneity. In addition, further analysis of four PM2.5 components identified (p = 0.026, ΔI2 = 1.78%) and
(p = 0.028, ΔI2 = 1.58%) as additional modifying factors. To identify the optimal combination of factors explaining spatial heterogeneity, we employed a stepwise approach to build multivariate meta-regression models (Table 2). Although a full model including all significant modifiers was tested, it did not improve explanatory power, resulting in a slightly lower ΔI2 of 5.07%. In contrast, the optimal model, selected on the basis of both the Akaike Information Criterion (AIC) and the proportion of heterogeneity explained, included four variables: population density, GDP per capita, cold spells, and PM2.5. These variables together accounted for 6.20% of the spatial heterogeneity. The Wald test confirmed the statistical significance of these factors (p-values < 0.001), highlighting their key role in shaping regional variation in temperature-related HFMD risk.
In addition, this phase of the analysis primarily focuses on the correlations between nighttime light, NDVI, and key socioeconomic indicators, including population density and GDP per capita. The correlation matrix (S3 Fig) revealed that, among the four variables, NDVI and population density exhibited the highest correlation (r = -0.62), which remains below the |0.8| threshold commonly used to indicate strong correlation. Consistent with this, VIF values for all four variables were below 2.5 (S7 Table), indicating low multicollinearity.
Modifying effects of regional factors on temperature-related HFMD risk
Fig 3 illustrates the modifying effects of eight regional factors—heat waves, cold spells, PM2.5, PM10, O3, NDVI, and two PM2.5 components ( and
)—on the temperature-HFMD association. These modifiers showed relatively stable effects at low temperatures but either amplified or suppressed HFMD risk at higher temperatures. For example, regions with more frequent heat waves (the 90th percentile) exhibited a substantially higher HFMD risk under high-temperature conditions. At the 95th percentile of temperature in these regions, the RR of HFMD was 3.17 (95% CI: 2.39-4.19), whereas in regions with infrequent heat waves (the 10th percentile), the RR was 1.23 (95% CI: 1.03-1.63) (Fig 3a; S2 Table). A similar modifying effect was observed for PM2.5. At the 95th percentile of temperature in PM2.5 high-exposure regions (the 90th percentile), the RR of HFMD was 3.05 (95% CI: 2.35-3.95), compared with 1.39 (95% CI: 1.10-1.77) in regions with lower PM2.5 concentrations (the 10th percentile) (Fig 3c; S2 Table). Comparable modification patterns in temperature-HFMD relationship were also identified for PM10 and two PM2.5 components (Fig 3d, 3g, and 3h; S2 Table). Conversely, cold spells demonstrated a protective modifying effect on the temperature-HFMD relationship. Under the same high temperature condition (the 95th percentile), regions with more frequent cold spells (the 90th percentile) had an RR of HFMD of 1.27 (95% CI: 1.07-1.63), compared to 3.13 (95% CI: 2.40-4.07) in regions with fewer cold spells (the 10th percentile) (Fig 3b; S2 Table). A similar inverse pattern was also evident for NDVI and O3. For NDVI, at the 95th percentile of temperature in regions with low NDVI (the 10th percentile) showed a higher HFMD RR of 2.16 (1.79-2.59) compared to 1.59 (1.21-2.08) in regions with higher NDVI (the 90th percentile) (Fig 3f). Details of results for O3 and the lag-response relationships of these environmental modifiers are provided in the Supporting Information (S2 Table and S4 Fig).
Notes: 1) The curves represent the temperature-HFMD associations at the 10th (blue line) and 90th (red line) percentiles of each spatial environmental factor. 2) The reference temperature is 12.9 °C. 3) Shaded areas indicate 95% confidence intervals.
Cluster analysis of regional heterogeneity in temperature–HFMD associations
Our study identified 3 distinct clusters among the 88 counties using 14 clustering algorithms, with the optimal classification determined by the Bayesian Information Criterion (S5 Fig). Fig 4 illustrates the temperature-HFMD associations across clusters, highlighting substantial regional heterogeneity. Analysis of variance (ANOVA) results (S4 Table, S6 Fig) consistently yielded p-values < 0.05, indicating significant differences in environmental characteristics between the clusters. In the central region (Cluster 1; Fig 4a), HFMD risk increased sharply with increasing temperature, reaching a RR of 2.47 (1.98-3.09) at the 95th percentile temperature (26.2 °C). A similar upward trend was observed in the northeastern region (Cluster 3; Fig 4c), with a peak RR of 2.07 (1.57-2.74) at the 95th percentile (28.6 °C). By contrast, the south-western region (Cluster 2; Fig 4b) showed an S-shaped exposure-response relationship with no clear monotonic pattern. At the 95th percentile (25.8 °C), the RR was 1.27 (0.98-1.64), indicating a nonsignificant association. A full list of the 88 counties classified into the three clusters is provided in Supporting Information (S3 Table).
The base layers of the map were obtained from the National Platform for Common GeoSpatial Information Services via the Tianditu platform (in Chinese) (https://cloudcenter.tianditu.gov.cn/administrativeDivision). Notes: Gray shaded areas indicate 95% confidence intervals, and gray dashed lines represent county-level estimates.
Sensitivity analyses
Sensitivity analyses confirmed the robustness of the model, with QAIC values ranging from 6587.50 to 6605.00 across different df configurations. The optimal model used 3 df for exposure-response, lag-response and meteorological covariates (S5 Table). In addition, alternative definitions of heat wave days and cold spell days were used. The heterogeneity analysis based on alternative definitions remained statistically significant, supporting the robustness of the primary findings despite small variations in heterogeneity (S6 Table).
Discussion
Our study identified eight statistically significant effect modifiers of the temperature–HFMD association: heat waves, cold spells, PM2.5, PM10, NDVI, O3, and two PM2.5 components ( and
). To the best of our knowledge, evidence on these spatial effect modifiers has not been previously reported in the literature. Among them, cold spells, heat waves, PM2.5, and NDVI were the main contributors to spatial heterogeneity. Specifically, heat waves and PM2.5 positively modified the association, increasing the risk of HFMD under high temperature conditions. In contrast, cold spells, NDVI, and O₃ negatively modified the association, attenuating the temperature-related risk. Further spatial cluster analysis revealed three distinct regional patterns of the temperature-HFMD association, reinforcing the spatially modifying role of these environmental factors across regions.
In studies of spatial heterogeneity in infectious diseases like HFMD, which are influenced by multiple complex factors, the proportion of heterogeneity explained by each individual factor is typically limited. For instance, key meteorological variables such as temperature, rainfall, and relative humidity usually explain 1% to 3% of the spatial variation in disease incidence, even after considering socioeconomic factors [11,12,28]. However, this relatively modest explanatory power does not diminish the value of these factors, as they can still provide meaningful insights when statistically significant. In contrast, this study focuses on ETEs (cold spells and heat waves), which are expected to have a greater potential impact on human health, particularly on children, compared to average temperature. The explanatory power we found for these extreme events (3.82% for cold spells, 2.72% for heat waves) is, respectively, higher than or close to the upper limit of the range typically reported for conventional meteorological variables (e.g., 1–3%), which aligns with theoretical expectations [11]. Furthermore, the inclusion of air pollutants (PM2.5: 3.06%, PM10: 2.08%) and NDVI (1.57%) significantly increased the explanatory power, surpassing the explanatory values of many conventional meteorological variables in previous studies. Even though O3 (0.78%) showed lower explanatory power, it remained statistically significant. In this study, the four factors—cold spells, PM2.5, population density, and GDP per capita—collectively explain 6.2% of the heterogeneity. This cumulative explanatory power not only significantly exceeds the level of single factors in previous studies but also reflects the combined effect of multiple significant factors. Specifically, the individual explanatory power of cold spells (3.82%) and PM2.5 (3.06%) both surpass the typical range of 1%-3% for conventional meteorological variables reported in prior research. Population density and GDP per capita, meanwhile, are well-recognized key factors influencing regional heterogeneity in existing studies. Thus, these four factors can be identified as the main contributors to the spatial heterogeneity of HFMD.
Building on the discussion of the combined effects of multiple factors driving spatial heterogeneity in HFMD, this study reveals that heat waves and cold spells independently explain the spatial heterogeneity in temperature-related HFMD risk, significantly surpassing the explanatory power of average temperature [11,12]. This underscores the critical role of ETEs as key drivers of regional heterogeneity in the temperature-HFMD association. The stronger contribution of ETEs can be mechanistically supported by children’s unique biological and behavioral vulnerabilities, particularly regarding heat waves. Biologically, extreme heat accelerates thermoregulation (e.g., sweating and skin perfusion), leading to dehydration and increased cardiac output [14]. When thermoregulation fails, core body temperature rises, triggering systemic inflammation, oxidative stress, endothelial dysfunction and direct cytotoxicity, all of which can impair immune responses and increase susceptibility to viral infections [44,45]. These effects are of particular concern in children, whose thermoregulatory systems and immune defenses are still developing. Behaviorally, heatwaves may alter caregiving routines and increase children’s exposure to communal environments such as swimming pools or other public facilities [17], thereby increasing opportunities for person-to-person contact and exposure to enterovirus-contaminated water [46,47]. In regions with frequent cold spells, the absence of an increased risk of HFMD under high temperature conditions may be due to behavioral adaptation. Residents in these regions are more sensitive to temperature changes and tend to adopt precautionary behaviors during heat events, such as reducing outdoor activity and adopting protective measures. These adaptive behaviors are likely to contribute to the reduced risk of HFMD observed during periods of high temperature in these regions.
In addition to temperature-related factors, ambient air pollution may further modify the temperature-HFMD association and contribute to regional disparities. Our results underscore the important role of air pollution—particularly PM2.5, PM10, O3, and two PM2.5 components ( and
)—in shaping the spatial heterogeneity of this association. Specifically, the strength of the temperature-HFMD relationship varied with regional PM2.5 levels, with markedly higher HFMD risk under high-temperature conditions observed in areas with elevated PM2.5 concentrations. Although direct evidence on PM2.5 as a modifier of the temperature-HFMD association remains limited, several studies have reported significant associations between PM2.5 exposure and HFMD incidence. For example, a study conducted in Shijiazhuang [48], a city with severe PM2.5 pollution, found an increased risk of HFMD when PM2.5 concentrations ranged from 76 to 200 μg/m3, whereas a protective effect was observed at extremely high concentrations. These results are not directly comparable with our findings because of the much lower PM2.5 concentrations in our study area (29.12-44.68 μg/m3). In contrast, a study conducted in Guilin [25], a geographically adjacent area, reported a positive association between PM2.5 and HFMD, with an estimated risk ratio of 1.01 (95% CI: 0.70-1.46), partially supporting the plausibility of our findings. Two plausible mechanisms may explain this relationship: first, enteroviruses may adsorb to particulate matter, increasing the atmospheric stability and long-range transmission potential of these viruses [49]; second, PM2.5 may impair host defenses by inducing oxidative stress, promoting inflammation and even causing genetic damage, thereby increasing susceptibility to infection [50,51]. Although synergistic effects between PM2.5 and high temperatures on HFMD have not been directly confirmed, previous studies have shown that co-exposure to high PM2.5 levels and elevated temperatures can exacerbate health risks [52], which indirectly supports our findings. Moreover, the observed modification of the temperature-HFMD association by PM2.5 suggests a potential synergistic effect with high temperature. To test this hypothesis, future studies could explicitly model the interaction between PM2.5 and heat waves. Beyond association, analytical approaches such as Structural Equation Modeling could be applied to elucidate the causal pathways linking temperature, air pollution, and HFMD risk. In contrast, O3 accounted for only 0.78% of the observed heterogeneity, suggesting a relatively limited role in modifying the temperature-HFMD relationship at the regional scale. However, unlike PM2.5, the modifying effect of O3 showed an inverse pattern - regions with lower O3 levels exhibited a stronger association between high temperatures and HFMD risk. This finding is consistent with experimental studies demonstrating that higher O3 concentrations can reduce viral survival and replication [53], as well as epidemiological evidence showing a negative association between ambient O3 levels and HFMD incidence [25]. Although the modifying effect of O3 as a regional factor appears to be weak, these results highlight the need for further mechanistic and epidemiological studies to explore the potential interactions between O3 and temperature in shaping enterovirus transmission under real-world environmental conditions.
To further explore the contribution of PM2.5 to spatial heterogeneity in the temperature-HFMD relationship, we examined the effects of four specific PM2.5 components, among which and
emerged as key contributors. Both are secondary inorganic aerosols commonly derived from atmospheric reactions involving NOx and ammonia, which originate from traffic emissions, agricultural activities, and industrial processes. Their potential mechanisms are similar to those of total PM2.5, including increased oxidative stress, immune dysfunction, and their role as carriers of enteric viruses [54,55]. However, these two components explained only part of the observed heterogeneity, suggesting that other PM2.5 constituents—such as organic carbon, elemental carbon, and trace metals—may further contribute. Future studies should characterize these components and quantify their individual and synergistic effects on the temperature-HFMD association.
In addition to ETEs and air pollutants, our study included broader environmental indicators of urbanization, extending previous research on spatial heterogeneity. While earlier studies mainly emphasized socio-economic factors such as per capita GDP, we introduced two novel indicators: NDVI, which reflects green space coverage, and nighttime light, which captures urban economic activity. Among them, NDVI emerged as a statistically significant modifier of the temperature-HFMD associational novel finding in our study. Consistent with cold spells and O₃, NDVI showed a negative modifying effect. Temperature-related HFMD risk was more pronounced in regions with low NDVI, whereas the association was attenuated in greener areas, suggesting a potential buffering role of vegetation against heat-related HFMD risk. Although NDVI has not previously been examined as a modifier in this context, our finding is consistent with previous evidence linking greater green space to reduced HFMD incidence [56]. Moreover, this protective role of vegetation appears to extend beyond HFMD. For instance, a multi-country study across 49 low- and middle-income countries reported that greater greenspace access was associated with lower prevalence of acute respiratory infections, fever, and diarrhea among children under five [57]. Similarly, a nationwide analysis of 266 Chinese cities found a negative association between NDVI and COVID-19 incidence [58], suggesting a general protective role of vegetation against infectious diseases. Mechanistically, this effect may operate through multiple pathways relevant to HFMD transmission. Environmentally, vegetation can moderate local microclimates via evapotranspiration, which may reduce enterovirus stability and human exposure [59]. Concurrently, green space attenuates air pollutants (e.g., PM2.5, O3), limiting exposure to virus-laden aerosols and mitigating pollutant-induced inflammation that could facilitate infection [60]. Behaviorally, access to greenspace may encourage outdoor activity in less crowded, better-ventilated settings, reducing close-contact transmission, while also enhancing immune and psychological resilience [61]. Future studies could explore the causal pathways through which NDVI influences HFMD transmission, or its potential interactions with temperature and air pollutants in modulating disease risk.
This study has several important strengths that enhance the validity of its findings. Most importantly, it incorporates complete population-level data covering all reported HFMD cases in Guizhou Province (n ≈ 500,000), providing substantial statistical power and minimizing selection bias. Furthermore, it advances current knowledge by systematically examining spatial variation in the temperature-HFMD association across different environmental contexts, providing novel insights into potential effect modification by ecological and pollution-related factors. Nevertheless, several methodological limitations need to be considered. First, as an ecological time-series study, the results are inherently limited in inferring individual-level causality. However, the large sample size combined with sensitivity analyses supports the robustness of the results. Second, the use of area-level environmental data to estimate exposure may lead to non-differential misclassification, which tends to attenuate effect estimates towards the null. Third, due to the lack of surveillance data on enterovirus subtypes (e.g., EV71, CA16) associated with HFMD, this study did not analyze the spatial heterogeneity of these viral subtypes. Different subtypes may vary in sensitivity to temperature and other environmental conditions, which could influence the observed heterogeneity in the temperature-HFMD relationship. This limitation is common in current HFMD spatial heterogeneity research, so caution is needed when extrapolating these findings to regions with distinct viral subtype distributions. Additionally, the study period was necessarily limited to 2013–2019, both because of the availability of standardized air pollution data and to minimize potential confounding from behavioral changes, health care disruptions and environmental changes during the COVID-19 pandemic. Although China’s HFMD surveillance system may underreport mild cases, such bias is unlikely to significantly affect internal comparisons, as time-series methods rely on relative variation within consistent reporting systems and adjust for seasonality and secular trends. Notably, the degree of heterogeneity observed in our study is consistent with previous findings [11,12] and supports the robustness of our estimates. However, the limited explanatory power of the measured modifiers highlights the complex, multifactorial nature of environmental health interactions. Future research should examine additional contributors, such as PM2.5 metal constituents, behavioral adaptation, regional access to health care, and viral strain variability.
Conclusion
This study identified eight environmental modifiers contributing to the spatial heterogeneity in the association between temperature and HFMD, with heat waves, cold spells, PM2.5, and NDVI being the primary contributors. These findings provide new evidence on how environmental factors drive spatial variation in temperature–HFMD associations and deepen the understanding of environmental determinants of disease risk. The results demonstrate that temperature-related HFMD risks are substantially modified by regional environmental contexts, challenging the generalizability of single-site study. These insights also underscore the need for regional, tailored prevention strategies, including context-adapted greening interventions, to mitigate HFMD and other environmentally mediated infectious diseases.
Supporting information
S2 Data. Multicounty aggregated dataset for meta-analysis.
https://doi.org/10.1371/journal.pntd.0013801.s002
(PDF)
S1 Table. List of 88 counties in Guizhou province included in the study.
https://doi.org/10.1371/journal.pntd.0013801.s003
(DOCX)
S2 Table. Relative risk analysis of heterogeneous environmental factors in the temperature-HFMD associations.
https://doi.org/10.1371/journal.pntd.0013801.s004
(DOCX)
S3 Table. A list of 88 counties in Guizhou Province divided into three clustering types.
https://doi.org/10.1371/journal.pntd.0013801.s005
(DOCX)
S4 Table. Statistical description of the three clusters and ANOVA.
https://doi.org/10.1371/journal.pntd.0013801.s006
(DOCX)
S5 Table. Sensitivity analysis comparing different definitions of heat wave and cold spell in meta-analysis.
https://doi.org/10.1371/journal.pntd.0013801.s007
(DOCX)
S6 Table. Sensitivity analysis of differently defined heat and cold spell in meta-analysis.
https://doi.org/10.1371/journal.pntd.0013801.s008
(DOCX)
S7 Table. Multicollinearity diagnostics based on Variance Inflation Factor (VIF).
https://doi.org/10.1371/journal.pntd.0013801.s009
(DOCX)
S1 Fig. Time series plot of six meteorological factors in Guizhou Province.
https://doi.org/10.1371/journal.pntd.0013801.s010
(TIF)
S2 Fig. Lag plots of county-level and overall estimates of temperature-HFMD associations.
https://doi.org/10.1371/journal.pntd.0013801.s011
(TIF)
S3 Fig. Correlation matrix of regional environmental factors.
https://doi.org/10.1371/journal.pntd.0013801.s012
(TIF)
S4 Fig. Lag plots of environmental modifiers contributing to spatial heterogeneity in temperature-HFMD associations.
https://doi.org/10.1371/journal.pntd.0013801.s013
(TIF)
S6 Fig. Lags of three clustered areas in the temperature–HFMD associations.
https://doi.org/10.1371/journal.pntd.0013801.s015
(TIF)
Acknowledgments
The authors thank the Guizhou Center for Disease Control and Prevention for providing the data of HFMD cases.
References
- 1. Zhu P, Ji W, Li D, Li Z, Chen Y, Dai B, et al. Current status of hand-foot-and-mouth disease. J Biomed Sci. 2023;30(1):15. pmid:36829162
- 2. Chen Y, Ji W, Duan G, Feng H, Zhang Y, Chen S, et al. Childhood hand, foot and mouth disease sequelae cohort study in Henan, China: cohort profile. BMJ Open. 2025;15(1):e083958. pmid:39788784
- 3. Cheng Q, Bai L, Zhang Y, Zhang H, Wang S, Xie M, et al. Ambient temperature, humidity and hand, foot, and mouth disease: A systematic review and meta-analysis. Sci Total Environ. 2018;625:828–36. pmid:29306826
- 4. Lee MHP, Chong YM, Tay CG, Koh MT, Chem YK, Noordin N, et al. Detection of enteroviruses during a 2018 hand, foot and mouth disease outbreak in Malaysia. Trop Biomed. 2021;38(1):150–3. pmid:33797539
- 5. Koh WM, Badaruddin H, La H, Chen MI-C, Cook AR. Severity and burden of hand, foot and mouth disease in Asia: a modelling study. BMJ Glob Health. 2018;3(1):e000442. pmid:29564154
- 6. Zhang W, Rui J, Cheng X, Deng B, Zhang H, Huang L, et al. Transmissibility of hand, foot, and mouth disease in 97 counties of China. Sci Rep. 2022;12(1):4103.
- 7. Yi X, Chang Z, Zhao X, Ma Y, Liu F, Xiao X. The temporal characteristics of the lag-response relationship and related key time points between ambient temperature and hand, foot and mouth disease: A multicity study from mainland China. Sci Total Environ. 2020;749:141679. pmid:32836135
- 8. Li Z, Ji W, Chen S, Duan G, Jin Y. Hand, foot, and mouth disease challenges and its antiviral therapeutics. Vaccines. 2023;11(3).
- 9. Zhang R, Lin Z, Guo Z, Chang Z, Niu R, Wang Y, et al. Daily mean temperature and HFMD: risk assessment and attributable fraction identification in Ningbo China. J Expo Sci Environ Epidemiol. 2021;31(4):664–71. pmid:33547422
- 10. Deng J, Gao X, Xiao C, Xu S, Ma Y, Yang J, et al. Association between diurnal temperature range and outpatient visits for hand, foot, and mouth disease in Hefei, China: a distributed lag nonlinear analysis. Environ Sci Pollut Res Int. 2020;27(28):35618–25. pmid:32613503
- 11. Xiao X, Gasparrini A, Huang J, Liao Q, Liu F, Yin F, et al. The exposure-response relationship between temperature and childhood hand, foot and mouth disease: A multicity study from mainland China. Environ Int. 2017;100:102–9. pmid:28069250
- 12. Deng X, Chen Z, Zou Y, He Y, Chen S, Wang Q, et al. The effect of daily mean temperature on hand, foot and mouth disease and the source of regional heterogeneity in Chongqing, China, 2010-2019. Environ Health Prev Med. 2022;27:47. pmid:36517013
- 13. Bo Z, Ma Y, Chang Z, Zhang T, Liu F, Zhao X, et al. The spatial heterogeneity of the associations between relative humidity and pediatric hand, foot and mouth disease: Evidence from a nation-wide multicity study from mainland China. Sci Total Environ. 2020;707:136103. pmid:31874401
- 14. Xu R, Huang S, Shi C, Wang R, Liu T, Li Y, et al. Extreme temperature events, fine particulate matter, and myocardial infarction mortality. Circulation. 2023;148(4):312–23.
- 15. Principi N, Campana BR, Argentiero A, Fainardi V, Esposito S. The Influence of Heat on Pediatric and Perinatal Health: Risks, Evidence, and Future Directions. J Clin Med. 2025;14(4):1123. pmid:40004654
- 16. Brimicombe C, Wieser K, Monthaler T, Jackson D, De Bont J, Chersich MF, et al. Effects of ambient heat exposure on risk of all-cause mortality in children younger than 5 years in Africa: a pooled time-series analysis. Lancet Planet Health. 2024;8(9):e640–6. pmid:39128471
- 17. Luan G, Liu S, Zhang W, Zhai L, Zhang Y, Sun L, et al. Estimating the influence of high temperature on hand, foot, and mouth disease incidence in China. Environ Sci Pollut Res Int. 2023;30(1):1477–84. pmid:35915310
- 18. Hu Q, Zhao F, Yang Y, Lu S, Xie Y, Geng J, et al. Association between ambient air pollution and risk of hand, foot, and mouth disease: a systematic review and meta-analysis. BMC Public Health. 2025;25(1):2663. pmid:40770629
- 19. Qian J, Luo C, Lv Q, Liu Y, Zhang T, Yin F, et al. Associations between ambient air pollutants and childhood hand, foot, and mouth disease in Sichuan, China: a spatiotemporal study. Sci Rep. 2023;13(1):3993. pmid:36899026
- 20. Lakey PSJ, Berkemeier T, Tong H, Arangio AM, Lucas K, Pöschl U, et al. Chemical exposure-response relationship between air pollutants and reactive oxygen species in the human respiratory tract. Sci Rep. 2016;6:32916. pmid:27605301
- 21. Feng S, Gao D, Liao F, Zhou F, Wang X. The health effects of ambient PM2.5 and potential mechanisms. Ecotoxicol Environ Saf. 2016;128:67–74. pmid:26896893
- 22. Zhu Y, Xie J, Huang F, Cao L. Association between short-term exposure to air pollution and COVID-19 infection: Evidence from China. Sci Total Environ. 2020;727:138704. pmid:32315904
- 23. Hao J, Yang Z, Huang S, Yang W, Zhu Z, Tian L, et al. The association between short-term exposure to ambient air pollution and the incidence of mumps in Wuhan, China: A time-series study. Environ Res. 2019;177:108660. pmid:31445438
- 24. Wiegman CH, Li F, Ryffel B, Togbe D, Chung KF. Oxidative Stress in Ozone-Induced Chronic Lung Inflammation and Emphysema: A Facet of Chronic Obstructive Pulmonary Disease. Front Immunol. 2020;11.
- 25. Yu G, Li Y, Cai J, Yu D, Tang J, Zhai W, et al. Short-term effects of meteorological factors and air pollution on childhood hand-foot-mouth disease in Guilin, China. Sci Total Environ. 2019;646:460–70. pmid:30056233
- 26. Schumann GJP, Mellander C, Lobo J, Stolarick K, Matheson Z. Night-Time Light Data: A Good Proxy Measure for Economic Activity? PLoS One. 2015;10(10).
- 27. Du Z, Yang B, Jalaludin B, Knibbs L, Yu S, Dong G, et al. Association of neighborhood greenness with severity of hand, foot, and mouth disease. BMC Public Health. 2022;22(1):38. pmid:34991526
- 28. Fan C, Liu F, Zhao X, Ma Y, Yang F, Chang Z, et al. An alternative comprehensive index to quantify the interactive effect of temperature and relative humidity on hand, foot and mouth disease: A two-stage time series study including 143 cities in mainland China. Sci Total Environ. 2020;740:140106. pmid:32927545
- 29. Gordon CJ, Johnstone AFM, Aydin C. Thermal stress and toxicity. Compr Physiol. 2014;4(3):995–1016. pmid:24944028
- 30. Chen J, Yang J, Zhou M, Yin P, Wang B, Liu J, et al. Cold spell and mortality in 31 Chinese capital cities: Definitions, vulnerability and implications. Environ Int. 2019;128:271–8. pmid:31071590
- 31. Yang J, Yin P, Sun J, Wang B, Zhou M, Li M, et al. Heatwave and mortality in 31 major Chinese cities: Definition, vulnerability and implications. Sci Total Environ. 2019;649:695–702. pmid:30176480
- 32. Xu R, Sun H, Zhong Z, Zheng Y, Liu T, Li Y, et al. Ozone, Heat Wave, and Cardiovascular Disease Mortality: A Population-Based Case-Crossover Study. Environ Sci Technol. 2024;58(1):171–81. pmid:38100468
- 33. Wei J, Li Z, Li K, Dickerson RR, Pinker RT, Wang J, et al. Full-coverage mapping and spatiotemporal variations of ground-level ozone (O3) pollution from 2013 to 2020 across China. Remote Sens Environ. 2022;270:112775.
- 34. Wei J, Li Z, Xue W, Sun L, Fan T, Liu L, et al. The ChinaHighPM10 dataset: generation, validation, and spatiotemporal variations from 2015 to 2019 across China. Environ Int. 2021;146.
- 35. Wei J, Li Z, Cribb M, Huang W, Xue W, Sun L, et al. Improved 1 km resolution PM 2.5 estimates across China using enhanced space–time extremely randomized trees. Atmos Chem Phys. 2020;20(6):3273–89.
- 36. Zheng Q, Seto KC, Zhou Y, You S, Weng Q. Nighttime light remote sensing for urban applications: Progress, challenges, and prospects. ISPRS J Photogram Remote Sens. 2023;202:125–41.
- 37. Zhao A, Tian X, Jin Z, Zhang A. The imprint of urbanization on vegetation in the ecologically fragile area: A case study from China’s Loess Plateau. Ecol Indicat. 2023;154:110791.
- 38. Gasparrini A. Modeling exposure-lag-response associations with distributed lag non-linear models. Stat Med. 2014;33(5):881–99. pmid:24027094
- 39. Gasparrini A, Armstrong B, Kenward MG. Multivariate meta-analysis for non-linear and other multi-parameter associations. Stat Med. 2012;31(29):3821–39. pmid:22807043
- 40. Gasparrini A, Armstrong B. Reducing and meta-analysing estimates from distributed lag non-linear models. BMC Med Res Methodol. 2013;13:1. pmid:23297754
- 41. Li C-H, Mao J-J, Wu Y-J, Zhang B, Zhuang X, Qin G, et al. Combined impacts of environmental and socioeconomic covariates on HFMD risk in China: A spatiotemporal heterogeneous perspective. PLoS Negl Trop Dis. 2023;17(5):e0011286. pmid:37205641
- 42. Mason CH, Perreault WD. Collinearity, Power, and Interpretation of Multiple Regression Analysis. J Market Res. 1991;28(3):268–76.
- 43. Johnston R, Jones K, Manley D. Confounding and collinearity in regression analysis: a cautionary tale and an alternative procedure, illustrated by studies of British voting behaviour. Qual Quant. 2018;52(4):1957–76. pmid:29937587
- 44. Ebi KL, Capon A, Berry P, Broderick C, de Dear R, Havenith G, et al. Hot weather and heat extremes: health risks. Lancet. 2021;398(10301):698–708. pmid:34419205
- 45. Basagaña X, Sartini C, Barrera-Gómez J, Dadvand P, Cunillera J, Ostro B, et al. Heat waves and cause-specific mortality at all ages. Epidemiology. 2011;22(6):765–72. pmid:21968768
- 46. Salo RJ, Cliver DO. Effect of acid pH, salts, and temperature on the infectivity and physical integrity of enteroviruses. Arch Virol. 1976;52(4):269–82. pmid:13766
- 47. Rzezutka A, Cook N. Survival of human enteric viruses in the environment and food. FEMS Microbiol Rev. 2004;28(4):441–53. pmid:15374660
- 48. Liu R, Cai J, Guo W, Guo W, Wang W, Yan L, et al. Effects of temperature and PM2.5 on the incidence of hand, foot, and mouth in a heavily polluted area, Shijiazhuang, China. Environ Sci Pollut Res Int. 2022;29(8):11801–14. pmid:34550518
- 49. Nor NSM, Yip CW, Ibrahim N, Jaafar MH, Rashid ZZ, Mustafa N, et al. Particulate matter (PM2.5) as a potential SARS-CoV-2 carrier. Sci Rep. 2021;11(1):2508. pmid:33510270
- 50. Chen H, Oliver BG, Pant A, Olivera A, Poronnik P, Pollock CA, et al. Effects of air pollution on human health - Mechanistic evidence suggested by in vitro and in vivo modelling. Environ Res. 2022;212(Pt C):113378. pmid:35525290
- 51. Feng Y, Castro E, Wei Y, Jin T, Qiu X, Dominici F, et al. Long-term exposure to ambient PM2.5, particulate constituents and hospital admissions from non-respiratory infection. Nat Commun. 2024;15(1):1518. pmid:38374182
- 52. Zhou L, Wang Y, Wang Q, Ding Z, Jin H, Zhang T, et al. The interactive effects of extreme temperatures and PM2.5 pollution on mortalities in Jiangsu Province, China. Sci Rep. 2023;13(1):9479. pmid:37301905
- 53. Lin Y-C, Juan H-C, Cheng Y-C. Ozone exposure in the culture medium inhibits enterovirus 71 virus replication and modulates cytokine production in rhabdomyosarcoma cells. Antiviral Res. 2007;76(3):241–51. pmid:17764758
- 54. He G, Pan Y, Tanaka T. Air pollution and the airborne diseases: Evidence from China and Japan. J Environ Econ Manag. 2025;130:103117.
- 55. Qi Y, Chen Y, Yan X, Liu W, Ma L, Liu Y, et al. Co-Exposure of Ambient Particulate Matter and Airborne Transmission Pathogens: The Impairment of the Upper Respiratory Systems. Environ Sci Technol. 2022;56(22):15892–901. pmid:36240448
- 56. Du Z, Lawrence WR, Zhang W, Zhang D, Yu S, Hao Y. Bayesian spatiotemporal analysis for association of environmental factors with hand, foot, and mouth disease in Guangdong, China. Sci Rep. 2018;8(1):15147. pmid:30310172
- 57. Zhang Y-D, Odo DB, Li J-X, Hu L-X, Qiu H-L, Xie Y-T, et al. Greenspace and burden of infectious illnesses among children in 49 low- and middle-income countries. Cell Reports Sustainability. 2024;1(8):100150.
- 58. Peng W, Dong Y, Tian M, Yuan J, Kan H, Jia X, et al. City-level greenness exposure is associated with COVID-19 incidence in China. Environ Res. 2022;209:112871. pmid:35123969
- 59. Xiao X, Zhang L, Xiong Y, Jiang J, Xu A. Influence of spatial characteristics of green spaces on microclimate in Suzhou Industrial Park of China. Sci Rep. 2022;12(1):9121. pmid:35650391
- 60. Fu J, Fu H, Zhu C, Sun Y, Cao H. Assessing the health risk impacts of urban green spaces on air pollution - Evidence from 31 China’s provinces. Ecol Ind. 2024;159:111725.
- 61. Markevych I, Schoierer J, Hartig T, Chudnovsky A, Hystad P, Dzhambov AM, et al. Exploring pathways linking greenspace to health: Theoretical and methodological guidance. Environ Res. 2017;158:301–17. pmid:28672128