Impact of meteorological factors on the incidence of hand, foot, and mouth disease caused by major enterovirus pathogens: a distributed lag nonlinear analysis in Fujian Province, China Authors: Qiuping Fan, Yunfan Zhang, Yuanzhen Zhang, Yi Chen, Guimei Cai, Yuxian Tang, Yongqiang Yu, Zuquan Weng, Wei Yuan, Naipeng Kan, Xiaojing Nie Published: 2025 ABSTRACT ObjectiveInvestigating the impact of meteorological factors on the incidence of the main pathogens of hand, foot, and mouth disease (HFMD) in Fujian Province, China, through spatiotemporal feature analysis and the construction of a distributed lag non-linear model (DLNM).MethodsHFMD and its major pathogens enterovirus 71 (EV71) and coxsackie virus A16 (CV-A16) data from January 2017 to December 2020 in Fujian Province were obtained through the statutory infectious disease reporting surveillance system of China, and meteorological data were obtained from the national scientific meteorological data center of China. Correlation analysis was used to select meteorological variables. DLNM was constructed to investigate the impact of meteorological factors on the prevalence of EV71 and CV-A16.ResultsDuring the study period, a total of 261,434 confirmed cases of HFMD were reported in Fujian Province, with an annual incidence rate ranging from 67.62 to 243.57 per 100,000 population. The temporal distribution indicated a bimodal pattern of HFMD cases from 2017 to 2019. In 2020, HFMD cases and incidence rate significantly decreased. Spatial distribution analysis revealed that infections were predominantly concentrated in the southwestern region of Fujian, and a decreasing trend in incidence rate with increasing latitude. Correlation analysis indicated that the highest temperature (Tmax), lowest temperature (Tmin), mean temperature (MT), precipitation (PRE) and relative humidity (RHU) were positively correlated with infections, while atmospheric pressure (PRS) was negatively correlated. Analysis results from the DLNM showed that the relative risk of EV71 and CV-A16 associated with MT across various prefecture-level cities in Fujian exhibited an inverted “U” shape or a linear relationship, with a “harvesting” phenomenon observed at low MT levels.ConclusionThe impact of meteorological factors on the incidence of EV71 and CV-A16 exhibits spatial heterogeneity. Low MT can promote viral epidemics in the short term.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12879-025-12388-y. Introduction In the context of global warming, glacier melting, and the increasing frequency of extreme weather events, there is growing concern about the potential threats of climate change to human health. Among these concerns, the impact of meteorological factors on human infectious diseases has become a research hotspot both domestically and internationally in recent years [1–4]. Hand, foot, and mouth disease (HFMD) is an acute infectious disease primarily caused by enterovirus 71 (EV71) and coxsackie virus A16 (CV-A16) [5, 6]. Its incidence rate ranks in the top three among Class C infectious diseases in China. Some cases can progress to encephalitis, meningitis, encephalomyelitis, myocarditis, pulmonary edema, and even death, making it a serious public health safety issue worldwide [7]. Therefore, exploring the impact of meteorological factors on the incidence of the main pathogens of HFMD will provide effective reference for the prevention and control of HFMD. EV71 is the primary pathogen responsible for severe cases and fatalities of HFMD. It is a small, positive-sense, single-stranded RNA virus that was first isolated from children with viral encephalitis in California, USA, in 1969 [1]. Currently, EV71 is predominantly circulating in the Asia-Pacific region, with sporadic cases or localized small-scale outbreaks in Europe, the Americas, Australia, and Africa [2]. In 1997, an outbreak of HFMD caused by EV71 occurred in Malaysia, resulting in the deaths of 41 children [3]. The following year, a large-scale epidemic of HFMD emerged in China, with 405 severe cases and 75 fatalities reported, with EV71 being the main pathogen [4]. In 2010, China experienced another large-scale HFMD outbreak associated with EV71, with over 1.7 million cases, 27,000 of which presented severe neurological complications, and the death toll reached 905 [5]. Additionally, EV71 outbreaks have been reported in other Asia-Pacific countries, including Russia, South Korea, Thailand, and the Philippines [6–9]. CV-A16, another major pathogen of HFMD, is a single-stranded positive-sense RNA virus that was first successfully isolated in South Africa in 1951 [10]. In 1957, Robinson first reported HFMD caused by CV-A16 [11], and since then, infections with CV-A16 have been reported in multiple countries and regions [12–14]. Research by Ang et al. found that the HFMD outbreaks in Singapore in 2002, 2005, and 2007 were mainly caused by CV-A16 [15]. In 2005, among 764 children hospitalized with HFMD at a pediatric hospital in Vietnam, 53.8% of the specimens were positive for enteroviruses, with CV-A16 accounting for 52.1% [16]. Epidemics of varying magnitudes involving CV-A16 have also occurred in Beijing, Shanghai, and Guangzhou, China [17, 18]. Additionally, Xu et al. reported three cases with severe complications following CV-A16 infection, including two cases of brainstem encephalitis and one case of acute flaccid paralysis [19]. Given the high pathogenic potential of EV71 and CV-A16 and their high prevalence in the Asia-Pacific region, investigating the incidence characteristics of EV71 and CV-A16 is conducive to disease prevention and control. In recent years, there has been growing concern about the impact of meteorological factors on human health. Among the 375 known infectious diseases in humans, 58% have seen an increase in pathogenicity and extended epidemic periods due to climate change, primarily related to the alteration of environments in which pathogens survive and reproduce [20]. Meteorological factors also play a significant role in the transmission of HFMD pathogens; for instance, in subtropical and temperate regions, the incidence of HFMD is positively correlated with temperature, rainfall, relative humidity, and hours of sunshine, and negatively correlated with average atmospheric pressure [21]. This may be related to the climatic conditions in these regions being more conducive to the survival and reproduction of EV71 and CV-A16 [21]. The study by Chen B et al. suggested that the periodic epidemic pattern of HFMD is basically consistent with the variation pattern of meteorological factors. Further research revealed that the prevalence of EV71 is positively correlated with temperature and rainfall, especially when the temperature ranges from 20℃ to 25℃, the incidence of EV71 increases significantly [22]. In contrast, the prevalence of CV-A16 is positively correlated with relative humidity. Additionally, another study found that the sunshine duration in high-incidence areas of HFMD is obviously lower than that in non-high-incidence areas, indicating that sunshine duration may be an important meteorological factor for HFMD early warning [23]. Therefore, exploring the impact of meteorological factors on HFMD will provide a theoretical basis for the early prevention of HFMD and effectively reduce the impact of climate change or extreme weather on human health. Currently, infectious disease research models are primarily divided into two categories. One category is based on linear assumptions, predicting future events based on the early epidemic characteristics of infectious diseases. However, the relationship between diseases and influencing factors may present non-linear relationships such as “J”-shaped or inverted “V”-shaped, thus limiting the application of this model. The other category of models uses factors affecting disease transmission as independent variables and establishes models based on the correlation between disease occurrence and these variables, such as the Distributed Lag Nonlinear Model (DLNM) [24]. After the introduction of the Cross Basis Function (CB) and Poisson regression, DLNM can assess the impact of factors on event outcomes from both the lag and exposure dimensions, better fitting the lag effects of exposure factors and the non-linear relationship between exposure-response [25–28]. Scholars have used DLNM to explore the impact of meteorological factors on the incidence of HFMD, but due to the significant differences in climatic conditions across regions, the study results vary [29–32]. Guo Z et al. found that the cumulative effect of temperature on HFMD incidence risk presents an inverted “U”-shaped relationship, with the incidence risk peaking at 25℃ [29]. Deng X et al. also suggested that the cumulative effect of temperature on HFMD incidence risk shows an inverted “U”-shaped relationship, but its peak occurs at 23.8℃ [30]. Additionally, other studies have indicated that the relationship between the two can also be an inverted “J”-shaped or “M”-shaped relationship [31, 32]. A study in Yangzhou City of China showed that both low humidity and high humidity environments have lag effects on HFMD incidence. Within a 14-day lag period, the overall effect curve presents an inverted “U”-shaped pattern [29]. In contrast, a study in Guangzhou City suggested that low humidity environments have an acute promoting effect on HFMD incidence [31]. Although high humidity environments have a lag effect, their effect duration is longer, and the cumulative incidence risk increases with the rise of humidity [31]. Furthermore, the impacts of other meteorological factors such as atmospheric pressure, sunshine duration, rainfall, and temperature difference on HFMD vary across different regions [33–35].These discrepancies highlight the need for region-specific analyses, as climatic conditions, topographic features, and population dynamics collectively modulate the relationship between meteorological factors and HFMD transmission. As Fujian Province in China is located in a subtropical region with a warm and humid climate, the incidence of HFMD remains high. However, the complex mountainous and hilly terrain leads to significant differences in climatic conditions between prefecture-level cities, which can affect the incidence of HFMD to varying degrees. Thus, this study aims to explore the impact of meteorological factors on the incidence of HFMD caused by the main pathogens EV71 and CV-A16 in various prefecture-level cities of Fujian Province using the DLNM, providing a reference for the prevention of HFMD. Materials and methods Study sites Fujian Province is located between 23°31′N and 28°18′N latitude, and 115°50′E and 120°43′E longitude, situated on the southeast coast of China. The province administers 9 prefecture-level cities (Fuzhou, Longyan, Nanping, Ningde, Putian, Quanzhou, Sanming, Xiamen, Zhangzhou), with a total land area of 124,000 square kilometers and a maritime area of 136,000 square kilometers. By the end of 2022, the permanent population of Fujian Province was 41.88 million, with a birth rate of 7.07‰, a death rate of 6.52‰, and a natural growth rate of 0.55‰. Close to the Tropic of Cancer, Fujian Province has a warm and humid subtropical maritime monsoon climate, shaped by the Wuyi Mountains to the northwest, which block the invasion of cold air from the north. This climate is characterized by strong monsoon circulation, short winters and long summers, abundant water resources, and frequent disastrous weather. Additionally, the complex terrain of mountains and hills in Fujian Province leads to significant climatic variations between different prefecture-level cities. Meteorological data Daily meteorological data are sourced from the China National Science Meteorological Data Center (https://data.cma.cn/), which includes Maximum temperature (Tmax), Minimum temperature (Tmin), Mean temperature (MT), Precipitation (PRE), Air pressure (PRS), Relative humidity (RHU), and Maximum wind speed (MWS). The collected raw data are screened. Since daily meteorological data must correspond one-to-one with daily incidence data, and there are few outliers, it is not considered to handle outliers by deletion. In cases where daily meteorological values are missing, the values from the previous day are used to replace the missing values. HFMD case data From January 1, 2017, to December 31, 2020, the monthly incidence and incidence rates of HFMD in Fujian Province were obtained from the statutory infectious disease reporting surveillance system of the Centers for Disease Control and Prevention (CDC) in China (https://www.chinacdc.cn/), and the daily incidence data for EV71 and CV-A16 were obtained from the statutory infectious disease reporting surveillance system of the Fujian Provincial CDC (http://www.fjcdc.com.cn/). All of the above are government reports, and all personal data are anonymized. The clinical diagnostic criteria for HFMD are in accordance with the regulations set by the Ministry of Health of China [36]. The inclusion criteria were as follows: (1) The child’s residence was Fuzhou City, Longyan City, Nanping City, Ningde City, Putian City, Quanzhou City, Sanming City, Xiamen City, or Zhangzhou City; (2) The onset time was from January 1, 2017 to December 31, 2020; (3) The diagnosis of HFMD cases complied with the diagnostic criteria specified in the Diagnostic and Treatment Guidelines for Hand, Foot, and Mouth Disease (2018 Edition) [36]; (4) The clinical specimens of HFMD cases (including pharyngeal swabs, anal swabs, feces, blister fluid, blood, and cerebrospinal fluid, etc.) met at least one of the following conditions: (1) EV71 or CoxA16 was isolated from clinical specimens, or specific nucleic acid detection for EV71 or CoxA16 was positive; (2) Serum IgM antibodies against EV71 or CoxA16 in the acute phase were positive; (3) Neutralizing antibodies against EV71 or CoxA16 in the convalescent phase were more than 4 times higher than those in the acute phase [36]. The collected raw data were screened, duplicates were checked using report numbers, cases with an interval of more than 10 days between the onset date and diagnosis date were removed, and for cases with unclear onset dates, the diagnosis date was used for calculation. Database establishment and quality control Firstly, HFMD is a key infectious disease under surveillance in China, and it is strictly reported through the national statutory infectious disease reporting system in accordance with the management and direct reporting requirements for infectious disease outbreaks, covering various levels and types of medical and health institutions, ensuring that the data and information are authentic and reliable. Secondly, for case data, strict inclusion criteria are followed, and the data are uniformly processed using Excel 2019 software. The accuracy, authenticity, and logical consistency of the data are verified, and errors and missing information are corrected and supplemented by repeatedly checking the data; for meteorological data, errors, missing values, and extreme values are checked and processed by repeatedly querying the official website data. Subsequently, a time-series database is established, with “case report date” as the single variable for frequency counting, to obtain the daily incidence of EV71 and CV-A16 infection cases from January 1, 2017, to December 31, 2020. If there are no cases on a particular day, the observed value for that day’s case count is recorded as 0. Finally, the daily incidence data are matched with daily meteorological data to establish a MySQL database, forming time-series files for statistical analysis. Statistical methods Descriptive epidemiological methods were employed to characterize and analyze the characteristics of HFMD cases and EV71 and CV-A16 infections in Fujian Province from January 1, 2017, to December 31, 2020. This analysis revealed the temporal distribution of HFMD and the spatial distribution of EV71 and CV-A16 infections across various prefecture-level cities in Fujian Province, identifying the epidemic seasons and high-incidence areas. Statistical tables were constructed to describe the trends in meteorological data across prefecture-level cities in Fujian Province. For quantitative variables, the central tendency and dispersion were described using the mean ± standard deviation (x̅±s) when the data were normally distributed, and using the median and percentiles (Px) when the data were not normally distributed. The distribution of daily meteorological data during the study period was described using mean ± standard deviation and percentiles P0, P10, P25, P50, P75, P90, P100, with values rounded to one decimal place. The Spearman correlation analysis module in SPSS Statistics 26.0 software was used to test the correlations between variables. Meteorological variables significantly correlated (P < 0.05) with EV71 and CV-A16 cases were included in the subsequent phase of the study. To avoid the impact of multicollinearity on the model, variables with a correlation coefficient r > 0.7 were considered to have a high degree of correlation and could not be included in the same model simultaneously, while variables with r < 0.7 could be included in the model together. Model construction The epidemics of EV71 and CV-A16 exhibit pronounced seasonal and cyclical patterns and are significantly correlated with various meteorological factors. Consequently, this study controlled for potential confounding factors such as meteorological factors not included in the model, seasonal trends, day of the week, and holiday effects, in addition to building the model. The process of model construction is as follows: Firstly, the “dlnm,” “ggplot2,” “readxl,” and “splines” packages in RStudio 2023.12.0 software were applied for model construction and data analysis. A DLNM was established, with the basic expression as follows: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}Log[E(Yt)]&=\beta+cb(MT)+cb(PHU) +cb(PRS)\cr&+ns(confounding factors, df)\cr&+ns(time, df)+dow+holiday\end{aligned}$$\end{document} In the above formula, t represents the number of observed days, E(Yt) denotes the estimated number of infections on day t, β is the constant term, and the Log function is chosen as the link function; cb(MT), cb(RHU), and cb(PRS) represent the cross-basis for mean temperature, relative humidity, and air pressure, respectively, reflecting the impact of meteorological factors on the daily number of EV71 and CV-A16 infection cases at different lags; ns stands for natural cubic splines, which can fit the non-linear relationship between meteorological factors and disease; time is the time variable; df represents the degrees of freedom; dow and holiday are considered as confounding factors; the reference values for the cross-basis are set to the median of the respective meteorological factor. Degrees of freedom for both dimensions ranging from 3 to 5 and exhaustively combined. By fitting all possible parameter combinations and calculating their Akaike Information Criterion (AIC) values, the model with the smallest AIC was ultimately selected as the optimal model. To more comprehensively observe the long-term effects of meteorological factors on EV71 and CV-A16 infections, the lag period is set to 30 days. Finally, the lag selection in the original DLNM is altered to verify the robustness of the DLNM. Results Epidemiological characteristics of HFMD in Fujian Province from 2017 to 2020 Temporal distribution From January 1, 2017, to December 31, 2020, a total of 261,434 confirmed cases of HFMD were reported in Fujian Province. The annual number of HFMD cases from 2017 to 2020 were 88,036, 95,260, 51,274, and 26,864, respectively, with annual incidence rates of 227.25/100,000, 243.57/100,000, 130.10/100,000, and 67.62/100,000, respectively. Among laboratory-confirmed cases, CV-A16 accounts for 73.98% and EV71 accounts for 26.02%. HFMD cases occurred monthly from 2017 to 2019, with a bimodal distribution of annual cases, peaking in June each year and with a secondary peak in September to October, while the lowest incidence was observed in February, indicating a distinct seasonal epidemic pattern. In 2020, there was a significant decrease in the number of HFMD cases and incidence rates, with the lowest monthly case count at only 75, peaking in October, and no apparent bimodal distribution was observed (Fig. 1). Fig. 1Monthly incidence of HFMD in Fujian Province, 2017–2020. a represents 2017, b represents 2018, c represents 2019, d represents 2020, X-axis: month, Y-axis: number of incidence cases (left) and incidence rate (right), blue color represents the number of incidence cases (persons) and orange color represents the incidence rate (1/100,000). e represents the overall incidence in 2017–2020, X-axis: month, Y-axis: number of incidence cases (persons), blue represents 2017, red represents 2018, yellow represents 2019 and green represents 2020 Monthly incidence of HFMD in Fujian Province, 2017–2020. a represents 2017, b represents 2018, c represents 2019, d represents 2020, X-axis: month, Y-axis: number of incidence cases (left) and incidence rate (right), blue color represents the number of incidence cases (persons) and orange color represents the incidence rate (1/100,000). e represents the overall incidence in 2017–2020, X-axis: month, Y-axis: number of incidence cases (persons), blue represents 2017, red represents 2018, yellow represents 2019 and green represents 2020 Spatial distribution From January 1, 2017, to December 31, 2020, EV71 and CV-A16 infection cases were reported in all prefecture-level cities of Fujian Province annually. Longyan City reported the highest cumulative number of cases, accounting for 23% of the total cases in the province, followed by Zhangzhou City (16%), and Fuzhou City ranked third (11%). Putian City reported the fewest cumulative cases, representing only 2%. EV71 and CV-A16 infection cases were predominantly concentrated in the southwestern region of Fujian Province, with Longyan City being the most prominent, and the incidence rate showed a decreasing trend with increasing latitude (Fig. 2). Fig. 2Spatial distribution of EV71 and CV-A16 infections in each urban area of Fujian Province from 2017 to 2020. This figure aims to illustrate the geographical clustering characteristics of the two infections. The darker color indicates a higher proportion of cases in each urban area relative to the total number of EV71 and CV-A16 infections in Fujian Province during the study period Spatial distribution of EV71 and CV-A16 infections in each urban area of Fujian Province from 2017 to 2020. This figure aims to illustrate the geographical clustering characteristics of the two infections. The darker color indicates a higher proportion of cases in each urban area relative to the total number of EV71 and CV-A16 infections in Fujian Province during the study period Correlation analysis between EV71 and CV-A16 infection cases and meteorological factors The meteorological data collected in this study contained only 5 abnormal data points. Analysis of the meteorological data revealed that EV71 and CV-A16 infections in Fujian Province were associated with Tmax, Tmin, MT, PRE, PRS, and RHU (P < 0.05). MWS is only positively correlated with EV71 and CV-A16 infections in Zhangzhou City (r > 0, P < 0.05), with no significant correlation in other prefecture-level cities (P > 0.05) (Supplementary Material Tables 1–18). Previous studies have suggested that the impact of PRE on HFMD is not significant [33], hence this study did not include PRE in the DLNM. This study found significant multicollinearity among Tmax, Tmin, and MT (r > 0.7, P < 0.01), and therefore did not include them simultaneously in the model to avoid the impact of multicollinearity on the results. Ultimately, this study included MT, RHU, and PRS in the DLNM to further assess the impact of meteorological factors on the prevalence of EV71 and CV-A16, and established time-series plots (Fig. 3). As shown in Fig. 3, MT and PRS exhibit distinct periodicity, with their cyclical fluctuations in opposite directions; although the periodicity of RHU is not apparent, RHU in all prefecture-level cities is generally above 60%. Fig. 3Time series map of meteorological factors in urban areas of Fujian Province in 2017–2020. a represents Fuzhou city, b represents Longyan city, c represents Nanping city, d represents Ningde city, e represents Putian city, f represents Quanzhou city, g represents Sanming city, h represents Xiamen city, and i represents Zhangzhou city. Blue color represents air pressure (hPa), green color represents relative humidity (%), and brown color represents mean temperature (℃). The high similarity among subplots reflects the spatial consistency of meteorological changes across Fujian Province. Despite the overall similarity, distinct patterns are observable: (1) MT and PRS exhibit pronounced and opposite periodic fluctuations. (2) RHU lacks obvious periodicity but maintains a consistent high level (> 60%) across all prefecture-level cities throughout the study period Time series map of meteorological factors in urban areas of Fujian Province in 2017–2020. a represents Fuzhou city, b represents Longyan city, c represents Nanping city, d represents Ningde city, e represents Putian city, f represents Quanzhou city, g represents Sanming city, h represents Xiamen city, and i represents Zhangzhou city. Blue color represents air pressure (hPa), green color represents relative humidity (%), and brown color represents mean temperature (℃). The high similarity among subplots reflects the spatial consistency of meteorological changes across Fujian Province. Despite the overall similarity, distinct patterns are observable: (1) MT and PRS exhibit pronounced and opposite periodic fluctuations. (2) RHU lacks obvious periodicity but maintains a consistent high level (> 60%) across all prefecture-level cities throughout the study period Construct a DLNM to describe the impact of meteorological factors on the prevalence of EV71 and CV-A16 viruses in prefecture-level cities in Fujian Province Using the median values of meteorological data as reference points, this study investigates the impact of meteorological factors on the prevalence of EV71 and CV-A16 in various prefecture-level cities of Fujian Province within a 30-day lag period. MT model From the three-dimensional curve plots (Fig. 4Aa-Ia) and contour plots (Fig. 4Ab-Ib) of MT for various prefecture-level cities in Fujian Province, it can be observed that in Fuzhou, Longyan, Nanping, Ningde, Quanzhou, and Sanming, both low and high temperatures can increase the risk of EV71 and CV-A16 infections (RR > 1) within a short lag period, with the acute effect of low temperatures being greater than that of high temperatures. The maximum single-day lag effect of RR is reached at 3.9 °C with a 0-day lag in Fuzhou, 4.7 °C with a 0-day lag in Longyan, 4.4 °C with a 0-day lag in Ningde, 4.7 °C with a 15-day lag in Putian, 6.5 °C with a 0-day lag in Quanzhou, 2.9 °C with a 0-day lag in Sanming, and 33.5 °C with a 0-day lag in Xiamen. The overall effect plots of MT for various prefecture-level cities in Fujian Province (Fig. 4Ac-Ic) can exhibit an inverted “U” shape or a linear relationship. Specifically, the overall effect plots for Fuzhou, Ningde, Quanzhou, and Sanming cities show an inverted “U” relationship, divided into two segments. When MT is below 21.5 °C or above 25.5 °C (Fuzhou), below 21.0 °C or above 31.1 °C (Ningde), below 17.4 °C or above 23.4 °C (Quanzhou), and below 21.4 °C or above 27.3 °C (Sanming), MT has a protective effect on EV71 and CV-A16 infections (RR < 1). When MT is within the aforementioned temperature ranges, the risk of infection increases (RR > 1), and the infection risk reaches a peak at specific temperatures. The overall effect plot for Zhangzhou city also shows an inverted “U” shape, but except for an MT of 23.8 °C (peak RR), all other RR values are < 1, suggesting that within a 30-day lag period, the impact of MT on EV71 and CV-A16 infections is not limited. In Longyan, Nanping, Putian, and Xiamen cities, as MT increases, the RR values also increase, reaching a peak (RR > 1) at specific temperatures. Fig. 4Three-dimensional curve graphs (a), contour maps (b), and overall effect curve graphs (c) of the infection risks of MT, EV71 and CV-A16 in various prefecture-level cities of Fujian Province. A: Fuzhou City, B: Longyan City, C: Nanping City; D: Ningde City; E: Putian City; F: Quanzhou City; G: Sanming City; H: Xiamen City; I: Zhangzhou City. Take Fuzhou City as an example: (a) Three-dimensional plot of the relative risk of EV71 and CV-A16 on mean temperature during a 30-day lag period in Fuzhou city. X-axis: mean temperature (℃), Y-axis: relative risk, Z-axis: lag days (d). (b) Contour plot of the relative risk of EV71 and CV-A16 onset by mean temperature during a 30-day lag period in Fuzhou city. X-axis: mean temperature (℃), Y-axis: lag days (d). The blue color represents RR < 1, the white color represents RR = 1, the red color represents RR > 1. (c) Overall effect plot of mean temperature on the risk of EV71 and CV-A16 onset during a 30-day lag period in Fuzhou city. X-axis: mean temperature (℃), Y-axis: relative risk, reference value of 21.5℃. The RR value peaked when the temperature is 23.5℃ (RR = 1.070, 95% CI: 0.853–1.343). When the temperature is 25.5 °C, RR = 1.000 (95% CI: 0.621–1.612) Three-dimensional curve graphs (a), contour maps (b), and overall effect curve graphs (c) of the infection risks of MT, EV71 and CV-A16 in various prefecture-level cities of Fujian Province. A: Fuzhou City, B: Longyan City, C: Nanping City; D: Ningde City; E: Putian City; F: Quanzhou City; G: Sanming City; H: Xiamen City; I: Zhangzhou City. Take Fuzhou City as an example: (a) Three-dimensional plot of the relative risk of EV71 and CV-A16 on mean temperature during a 30-day lag period in Fuzhou city. X-axis: mean temperature (℃), Y-axis: relative risk, Z-axis: lag days (d). (b) Contour plot of the relative risk of EV71 and CV-A16 onset by mean temperature during a 30-day lag period in Fuzhou city. X-axis: mean temperature (℃), Y-axis: lag days (d). The blue color represents RR < 1, the white color represents RR = 1, the red color represents RR > 1. (c) Overall effect plot of mean temperature on the risk of EV71 and CV-A16 onset during a 30-day lag period in Fuzhou city. X-axis: mean temperature (℃), Y-axis: relative risk, reference value of 21.5℃. The RR value peaked when the temperature is 23.5℃ (RR = 1.070, 95% CI: 0.853–1.343). When the temperature is 25.5 °C, RR = 1.000 (95% CI: 0.621–1.612) RHU model From the three-dimensional curve plots (Fig. 5Aa-Ia) and contour plots (Fig. 5Ab-Ib) of relative humidity (RHU) for various prefecture-level cities in Fujian Province, it can be observed that in Fuzhou, Longyan, Ningde, Quanzhou, and Sanming cities, both low and high humidity environments exhibit varying degrees of lag in their impact on EV71 and CV-A16 infections, with both conditions increasing the risk of infection (RR > 1). In Nanping, Putian, and Xiamen cities, the impact of low and high humidity environments on EV71 and CV-A16 infections is not limited (RR < 1). In Zhangzhou city, high humidity increases the risk of EV71 and CV-A16 infections (RR > 1) after a lag of 26 days, while the impact of low humidity on infection is not limited (RR < 1). The overall effect plots of RHU for various prefecture-level cities in Fujian Province (Fig. 5Ac-Ic) can exhibit a “U” shape, inverted “U” shape, “L” shape, or linear relationship. Specifically, the overall effect plots for Fuzhou, Longyan, and Sanming cities show a “U” shape, with RR peaking at 35.0% (Fuzhou), 22.3% (Longyan), and 33.4% (Sanming) RHU. The overall effect plots for Nanping, Putian, Xiamen, and Zhangzhou cities show an inverted “U” shape, with RR peaking at 78.7% (Nanping), 72.8%–77.0% (Putian), 76.5% (Xiamen), and 75.0% (Zhangzhou) RHU. The overall effect plot for Quanzhou city shows an “L” shape, with RR peaking at 20.6% RHU. The overall effect plot for Ningde city shows a linear relationship, with RR values increasing as RHU increases, and RHU > 76.7% being a risk factor for EV71 and CV-A16 infections (RR > 1), with RR peaking at 100.0% RHU. Fig. 5Three-dimensional curve graphs (a), contour maps (b), and overall effect curve graphs (c) of the infection risks of RHU, EV71 and CV-A16 in various prefecture-level cities of Fujian Province. A: Fuzhou City, B: Longyan City, C: Nanping City; D: Ningde City; E: Putian City; F: Quanzhou City; G: Sanming City; H: Xiamen City; I: Zhangzhou City. Take Fuzhou City as an example: (a) Three-dimensional plot of the relative risk of EV71 and CV-A16 on relative humidity during a 30-day lag period in Fuzhou city. X-axis: relative humidity (%), Y-axis: relative risk, Z-axis: lag days (d). (b) Contour plot of the relative risk of EV71 and CV-A16 onset by relative humidity during a 30-day lag period in Fuzhou city. X-axis: relative humidity (%), Y-axis: lag days (d). The blue color represents RR < 1, the white color represents RR = 1, the red color represents RR > 1. (c) Overall effect plot of relative humidity on the risk of EV71 and CV-A16 onset during a 30-day lag period in Fuzhou city. X-axis: relative humidity (%), Y-axis: relative risk, reference value of 72.3%. The RR value peaked when the relative humidity is 35.0% (RR = 12.519, 95% CI: 0.106-1472.843) Three-dimensional curve graphs (a), contour maps (b), and overall effect curve graphs (c) of the infection risks of RHU, EV71 and CV-A16 in various prefecture-level cities of Fujian Province. A: Fuzhou City, B: Longyan City, C: Nanping City; D: Ningde City; E: Putian City; F: Quanzhou City; G: Sanming City; H: Xiamen City; I: Zhangzhou City. Take Fuzhou City as an example: (a) Three-dimensional plot of the relative risk of EV71 and CV-A16 on relative humidity during a 30-day lag period in Fuzhou city. X-axis: relative humidity (%), Y-axis: relative risk, Z-axis: lag days (d). (b) Contour plot of the relative risk of EV71 and CV-A16 onset by relative humidity during a 30-day lag period in Fuzhou city. X-axis: relative humidity (%), Y-axis: lag days (d). The blue color represents RR < 1, the white color represents RR = 1, the red color represents RR > 1. (c) Overall effect plot of relative humidity on the risk of EV71 and CV-A16 onset during a 30-day lag period in Fuzhou city. X-axis: relative humidity (%), Y-axis: relative risk, reference value of 72.3%. The RR value peaked when the relative humidity is 35.0% (RR = 12.519, 95% CI: 0.106-1472.843) PRS model From the three-dimensional curve plots (Fig. 6Aa-Ia) and contour plots (Fig. 6Ab-Ib) of atmospheric pressure (PRS) for various prefecture-level cities in Fujian Province, it can be observed that Fuzhou, Quanzhou, and Zhangzhou cities all show that high pressure increases the risk of infection in the short term (RR > 1), while the impact of low pressure on infection is not limitde (RR < 1) or exhibits a lag effect. Longyan, Nanping, Ningde, and Sanming cities all show that high pressure increases the risk of infection in the short term (RR > 1), while the impact of low pressure on infection is not limited (RR < 1). Putian city shows that low pressure increases the risk of infection in the short term (RR > 1), while the impact of high pressure on infection is not limited (RR < 1), and Xiamen city shows that within a specific PRS range (981.0 hPa to 996.0 hPa), it increases the risk of infection in the short term (RR > 1), with other PRS values having no limited impact on infection. The overall effect plots of PRS for various prefecture-level cities in Fujian Province (Fig. 6Ac-Ic) generally exhibit an inverted “U” shape and linear relationships. The overall effect plots for Fuzhou, Nanping, Ningde, Sanming, Xiamen, and Zhangzhou cities show an inverted “U” shape, with the risk of infection peaking at PRS values of 1005.5 hPa (Fuzhou), 986.1 hPa (Nanping), 1007.0 hPa (Ningde), 988.4 hPa (Sanming), 994.8 hPa (Xiamen), and 1011.2 hPa (Zhangzhou), and the risk of infection decreasing at lower or higher PRS values. The overall effect plots for Longyan and Quanzhou cities exhibit linear relationships, with RR values decreasing or increasing as PRS increases, and PRS < 971.3 hPa (Longyan) and PRS > 1010.3 hPa (Quanzhou) being risk factors for EV71 and CV-A16 infections (RR > 1). Fig. 6Three-dimensional curve graphs (a), contour maps (b), and overall effect curve graphs (c) of the infection risks of PRS, EV71 and CV-A16 in various prefecture-level cities of Fujian Province. A: Fuzhou City, B: Longyan City, C: Nanping City; D: Ningde City; E: Putian City; F: Quanzhou City; G: Sanming City; H: Xiamen City; I: Zhangzhou City. Take Fuzhou City as an example: (a) Three-dimensional plot of the relative risk of EV71 and CV-A16 on air pressure during a 30-day lag period in Fuzhou city. X-axis: air pressure (hPa), Y-axis: relative risk, Z-axis: lag days(d). (b) Contour plot of the relative risk of EV71 and CV-A16 onset by air pressure during a 30-day lag period in Fuzhou city. X-axis: air pressure (hPa), Y-axis: lag days (d). The blue color represents RR < 1, the white color represents RR = 1, the red color renresents RR > 1. (c) Overall effect plot of air pressure on the risk of EV71 and CV-A16 onset during a 30-day lag period in Fuzhou city. X-axis: air pressure (hPa), Y-axis: relative risk, reference value of 1005.5 hPa. The RR value peaked when the air pressure is 1005.5hPa (RR = 1.000, 95% CI: 0) Three-dimensional curve graphs (a), contour maps (b), and overall effect curve graphs (c) of the infection risks of PRS, EV71 and CV-A16 in various prefecture-level cities of Fujian Province. A: Fuzhou City, B: Longyan City, C: Nanping City; D: Ningde City; E: Putian City; F: Quanzhou City; G: Sanming City; H: Xiamen City; I: Zhangzhou City. Take Fuzhou City as an example: (a) Three-dimensional plot of the relative risk of EV71 and CV-A16 on air pressure during a 30-day lag period in Fuzhou city. X-axis: air pressure (hPa), Y-axis: relative risk, Z-axis: lag days(d). (b) Contour plot of the relative risk of EV71 and CV-A16 onset by air pressure during a 30-day lag period in Fuzhou city. X-axis: air pressure (hPa), Y-axis: lag days (d). The blue color represents RR < 1, the white color represents RR = 1, the red color renresents RR > 1. (c) Overall effect plot of air pressure on the risk of EV71 and CV-A16 onset during a 30-day lag period in Fuzhou city. X-axis: air pressure (hPa), Y-axis: relative risk, reference value of 1005.5 hPa. The RR value peaked when the air pressure is 1005.5hPa (RR = 1.000, 95% CI: 0) Sensitivity analysis By adjusting model parameters to test the stability of each DLNM, we set the lag days for the MT, RHU, and PRS models for each prefecture-level city to 29 d, 30 d, and 31 d, respectively. The shapes of the three-dimensional curve plots showed no significant changes after altering the lag days (Supplementary Material Figs. 1 to 3), indicating that the DLNM used in this study are robust and can be used to explain the nonlinear relationship between meteorological factors and EV71 and CV-A16 infections in various prefecture-level cities of Fujian Province. Discussion HFMD is a common acute infectious disease in children, with the main pathogens being EV71 and CV-A16 [36–38]. In recent years, the incidence rate of HFMD in Fujian Province has been second only to viral hepatitis [33], imposing a significant healthcare burden on society. Studies suggest that climatic conditions are closely associated with the occurrence of HFMD [39]. Located on the southeast coast of China, Fujian Province experiences a warm and humid subtropical monsoon climate, and its complex terrain of mountains and hills results in considerable climatic variations among different prefecture-level cities. Therefore, investigating the impact of meteorological factors on the prevalence of EV71 and CV-A16 in Fujian Province will provide a theoretical basis for the establishment of an HFMD early warning system in the province. Firstly, we collected monthly incidence and incidence rate data of HFMD in Fujian Province from January 2017 to December 2020 through the CDC’s statutory infectious disease reporting surveillance system. The results showed that a total of 261,434 confirmed cases of HFMD were reported in Fujian Province from 2017 to 2020, with an annual incidence rate ranging from 67.62 to 243.57 per 100,000, and an average annual incidence rate of 167.13 per 100,000, which is higher than the national average during the same period (125.22 per 100,000). Among all laboratory-confirmed cases, CV-A16 accounts for 73.98% and EV71 accounts for 26.02%, indicating that CV-A16 was the predominant pathogen during the study period. This is consistent with previous studies, which have shown that CV-A16 has caused the most HFMD outbreaks in China since 2011 [40]. Further analysis revealed that HFMD in Fujian Province has a distinct seasonal epidemic pattern, with a peak in June each year and a secondary peak in September to October, consistent with the bimodal distribution characteristics of HFMD in southern China, but different from the unimodal distribution in northern China, which may be related to climatic differences and other factors [37, 41, 42]. Compared to 2017–2019, the number of HFMD cases and incidence rates dropped significantly in 2020, a phenomenon that may be associated with the COVID-19 pandemic. During the pandemic, measures such as wearing masks, frequent hand washing, and reducing outings were taken in China to control the spread of COVID-19, which also greatly reduced the occurrence of HFMD [43]. Notably, October 2020 was the peak month for HFMD cases that year, and similar observations were made by studies conducted by Zhu et al. and Cui et al. [33, 44]. Since children are the main susceptible population for HFMD, we speculate that this phenomenon may be related to increased environmental exposure following the resumption of school in September 2020 [36, 45]. The above results indicate that June is the peak month for HFMD in Fujian Province. To effectively control and reduce the spread of the disease, measures such as wearing masks and strictly enforcing hand hygiene to interrupt the transmission routes can be taken. HFMD not only exhibits seasonal prevalence but also geographical clustering [38]. To further investigate the spatial distribution characteristics of HFMD in Fujian Province, we analyzed the daily incidence data of the main pathogens of HFMD (EV71 and CV-A16) in various prefecture-level cities of Fujian Province from January 1, 2017, to December 31, 2020. We found that the infection rates of EV71 and CV-A16 decreased with increasing latitude, with lower latitude areas in Fujian Province having higher incidence rates. Specifically, Longyan City reported the highest cumulative number of cases, accounting for 23% of the total provincial cases. This is similar to the findings of Lee et al. and Xing et al. [37, 46]. In summary, Longyan City, located in the low-latitude area of our province, has a relatively high incidence rate. More stringent surveillance and control measures should be implemented during the HFMD epidemic season to reduce its transmission and thereby lower the incidence rate. To further quantify the impact of climatic factors on the prevalence of EV71 and CV-A16, we collected meteorological data from various prefecture-level cities in Fujian Province from 2017 to 2020 and conducted a correlation analysis. The results showed that Tmax, Tmin, MT, PRE, and RHU were positively correlated with the infections of EV71 and CV-A16 (r > 0, P < 0.05), while PRS was negatively correlated with the infections of EV71 and CV-A16 (r < 0, P < 0.05), suggesting that high temperature and humidity can promote the prevalence of EV71 and CV-A16. Hu et al.used a weighted regression model to analyze the influencing factors of HFMD incidence in children under nine years old, and the results indicated a significant correlation between Tmax, Tmin, MT, PRE, RHU, and the incidence of HFMD [47]. Li et al., using a negative binomial multivariate regression approach, also observed similar results [48]. Our study results are consistent with the above two studies, which may be related to the climatic conditions for the survival and reproduction of EV [21]. This study also found that MWS was only positively correlated with the infections of EV71 and CV-A16 in Zhangzhou City (r > 0, P < 0.05), with no significant correlation in other prefecture-level cities (P > 0.05). Other epidemiological studies on HFMD have also found a positive correlation between MWS and the incidence of HFMD [33, 49], while some researchers have found a negative correlation between this meteorological factor and the incidence of HFMD [50]. Therefore, the impact of MWS on the incidence of HFMD is still unclear at present. Temperature is the most important factor affecting the transmission of HFMD [51]. The three-dimensional curve plots and contour plots from the DLNM of MT suggest that in most areas of Fujian Province, the relative risk (RR) reaches its maximum single-day lag effect at the lowest MT, and there is a “harvesting” phenomenon for infections by EV71 and CV-A16 at low temperatures, indicating a short-term promotional effect on the transmission of EV71 and CV-A16. This is consistent with the findings from Guangdong Province [52]. The “harvesting” phenomenon refers to the occurrence of disease in a subset of susceptible individuals over a short period, leading to a temporary surge in incidence followed by a gradual decline as the number of recently exposed susceptible individuals diminishes. Since this phenomenon mainly affects susceptible populations, the outcome is not recurrent. We speculate that the reasons for this phenomenon include: (1) Children’s immune systems are not fully developed and are easily affected by weather changes, thus increasing the risk of children being infected by EV71 and CV-A16 when temperatures drop [38]; (2) Low temperatures may lead to increased indoor aggregation of children, which enhances close contact and facilitates virus transmission; (3) Low-temperature environments are not conducive to the survival and reproduction of EV71 and CV-A16. As MT decreases, the activity of EV gradually weakens, so EV71 and CV-A16 do not cause long-term repeated infections in low-temperature environments [10]. The overall effect plots show that in Fuzhou, Ningde, Quanzhou, and Sanming cities, the cumulative RR values for EV71 and CV-A16 infections related to MT exhibit an inverted “U” shape relationship, with RR decreasing as temperature increases beyond a certain threshold. Wang J et al. reached the same conclusion, while in Longyan, Nanping, Putian, and Xiamen cities, RR values increase with temperature, peaking at Tmax [53]. This is consistent with the findings of Zhu et al., but differs from the inverted “J” and “M” shape relationships observed in other regions [31–33]. The discrepancies in nonlinear lagged exposure-response relationships between this study and previous research may be attributed to multiple factors: (1) Fujian’s subtropical monsoon climate differs from the temperate or arid climates of other study areas, and other meteorological conditions that modulate virus transmission; (2) Fujian’s mountainous and hilly terrain creates microclimatic variations across prefecture-level cities, which may alter the net effect of temperature on HFMD incidence. RHU and PRS also play significant roles in the transmission of HFMD. The DLNM results for RHU suggest that its impact on the prevalence of EV71 and CV-A16 across various prefecture-level cities in Fujian Province exhibits distinct spatial heterogeneity, with the overall effect plots showing either a “U” shape or an inverted “U” shape relationship. Some studies have found that RHU can have an inverted “U” shaped relationship or a “U” shaped relationship with the cumulative RR of HFMD, while other researchers have observed a “J” shaped relationship between the two [54, 55]. Although our statistical analysis suggests a negative correlation between PRS and the prevalence of HFMD, the DLNM results for PRS vary across different prefecture-level cities in Fujian Province, making it difficult to generalize the impact of PRS on the prevalence of EV71 and CV-A16. Moreover, the limited data available on the influence of PRS on the incidence of HFMD constrains further validation of our findings. In summary, HFMD in Fujian Province exhibits seasonal prevalence and geographical clustering, with a peak in incidence in June and a secondary peak in September to October, and a relatively higher incidence in low-latitude areas. The impact of meteorological factors on the prevalence of EV71 and CV-A16 shows spatial heterogeneity. Among the meteorological factors closely associated with the prevalence of EV71 and CV-A16, low MT has a short-term promotional effect on the infection, and high RHU is a risk factor for EV71 and CV-A16 infection. However, the transmission of HFMD is influenced by a combination of factors, including meteorological factors as well as host susceptibility, population density, economic development, etc., and the impact of these factors on the prevalence of HFMD warrants further investigation. Conclusions Meteorological factors exhibit distinct spatial heterogeneity in their impact on the prevalence of EV71 and CV-A16, with low MT promoting viral epidemics in the short term. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1