Abstract

Background

Intestinal infectious diseases (IIDs) can impact the growth and development of children and weaken adults. This study aimed to establish a spatial panel model to analyze the relationship between factors such as population, economy and health resources, and the incidence of common IIDs. The objective was to provide a scientific basis for the formulation diseases prevention measures.

Methods

Data on monthly reported cases of IIDs in each district and county of Zhejiang Province were collected from 2011 to 2021. The spatial distribution trend was plotted, and nine factors related to population, economy and health resources were selected for analysis. A spatial panel model was developed to identify statistically significant spatial patterns of influencing factors (P鈥�&濒迟;鈥�0.05).

Results

The results revealed that each type of IIDs exhibited a certain level of clustering. Each IIDs had a significant radiation effect, HEV (b鈥�=鈥�0.28, P鈥�<鈥�0.05), bacillary dysentery (b鈥�=鈥�0.38, P鈥�<鈥�0.05), typhoid (b鈥�=鈥�0.36, P鈥�<鈥�0.05), other infectious diarrheas (OIDs) (b鈥�=鈥�0.28, P鈥�<鈥�0.05) and hand, foot and mouth disease (HFMD) (b鈥�=鈥�0.39, P鈥�<鈥�0.05), indicating that regions with high morbidity rates spread to neighboring areas. Among the population characteristics, density of population acted as a protective factor for bacillary dysentery (b=-1.81, P鈥�<鈥�0.05), sex ratio acted as a protective factor for HFMD (b=-0.07, P鈥�<鈥�0.05), and aging rate increased the risk of OIDs (b鈥�=鈥�2.39, P鈥�&濒迟;鈥�0.05). Urbanization ratio posed a hazard factor for bacillary dysentery (b鈥�=鈥�5.17, P鈥�<鈥�0.05) and OIDs (b鈥�=鈥�0.64, P鈥�<鈥�0.05) while serving as a protective factor for typhoid (b=-1.61, P鈥�<鈥�0.05) and HFMD (b=-0.39, P鈥�&濒迟;鈥�0.05). Per capita GDP was a risk factor for typhoid (b鈥�=鈥�0.54, P鈥�<鈥�0.05), but acted as a protective factor for OIDs (b=-0.45, P鈥�<鈥�0.05) and HFMD (b=-0.27, P鈥�&濒迟;鈥�0.05). Additionally, the subsistence allowances ratio was a risk factor for HEV (b鈥�=鈥�0.24, P鈥�&濒迟;鈥�0.05).

Conclusion

The incidence of IIDs in Zhejiang Province exhibited a certain degree of clustering, with major hotspots identified in Hangzhou, Shaoxing, and Jinhua. It would be essential to consider the spillover effects from neighboring regions and implement targeted measures to enhance disease prevention based on regional development.

Background

Statutory infectious diseases of intestinal infectious diseases (IIDs) in China include cholera, poliomyelitis, typhoid and paratyphoid (TAP), viral hepatitis E (HEV), viral hepatitis unspecified, bacillary dysentery, hand, foot and mouth disease (HFMD) and other infectious diarrheas (OIDs) [1]. These diseases have a negative impact on children鈥檚 growth and development, reduce the physical strength of adults, result in labor force loss, and pose risks to people鈥檚 health and social development [2]. Recent studies indicated a significant decrease in the incidence of IIDs in China due to specific prevention and control measures. However, there still exist spatially and temporally clusters of these diseases [3]. The Yangtze River Delta region, which experiences relatively high economic development and attracts a large number of migrant populations, has emerged as an area with a high incidence of IIDs [4, 5]. Zhejiang Province exhibits a significant incidence of IIDs, with 248,000 reported cases and an incidence rate of 384.10/100,000, thereby ranking highest among all notifiable IIDs within the province [6, 7]. Meanwhile OIDs and HFMD are the most common infectious diseases [8]. These results underscore the ongoing significance of IIDs as a critical public health concern in Zhejiang Province.

Previous studies on IIDs largely focused on epidemiological descriptions and spatio-temporal aggregation analysis [9, 10]. With the application of spatio-temporal data analysis in public health [11, 12], panel data model has proven effective in addressing the issue of non-independent spatial unit data. Compared with panel data analysis, spatial panel data analysis integrates time series and cross-sectional studies, utilizing the spatial properties of infectious disease data analysis. This model considers primary aggregation and diffusion, taking into account for the interplay between factors [13,14,15]. By implementing effective control measures for heterogeneity and autocorrelation, the estimation results can be improved in terms of accuracy, as suggested in previous studies [16, 17]. In this study, a spatial panel data model incorporating population, economic and health resource data at the district and county level was used to analyze the influence factors of IIDs in Zhejiang Province from 2011 to 2021. The study aimed to provide a scientific basis for relevant departments in formulating and improving preventive measures.

Methods

Data sources

The annual reported cases of IIDs from 2011 to 2021 in the 89 districts and counties were collected from the Zhejiang Provincial Center for Disease Control and Prevention. Cholera and poliomyelitis were excluded from the analysis due to their extremely low incidence. Due to data availability, HEV, typhoid and bacillary dysentery of IIDs under category B, OIDs and HFMD of IIDs under category C were included in this study. The study area includes all districts and counties except for Longgang District, which was established in 2020 and subsequently merged into Cangnan District for calculation purposes.

The variables including population, density, sex ratio, aging rate, GDP, per capita GDP, subsistence allowances ratio, urbanization ratio, beds, and medical practitioners data were collected from Statistical Yearbooks and Statistical Communique of National Economic and Social Development.

Electronic maps of Zhejiang Province were obtained from the National Catalogue Service for Geographic Information ().

Descriptive statistical analysis

The explanatory variable for this study is the annual incidence for each IIDs, along represent the number of IIDs per 100,000 residents and were calculated as

$$\:\begin{array}{c}\begin{array}{c}Incidence\:rate=\frac{number\:of\:new\:cases\:during\:a\:specific\:time\:period}{total\:population\:at\:risk\:during\:that\:time\:period}\times\:\text{100,000}\end{array}\end{array}$$

(1)

The numerator is the number of new cases, and the denominator is the total population at risk, which is the number of people who were susceptible to the disease or condition at the beginning of the time period. This formula gives the number of new cases per 100,000 people at risk during the time period. It鈥檚 important to note that incidence rate is different from prevalence, which is the number of existing cases of a disease or condition in a population at a given point in time. Incidence and variables maps were generated for each district and county using ArcGIS 10.8.1 with shapefile data.

Standard deviation ellipse

The Standard deviational ellipse (SDE) [18] was a crucial method for analyzing spatial data, providing insights into the overall characteristics of ingredient distribution from multiple perspectives. The ellipse鈥檚 center indicated the mean center of element distribution. The long axis indicated the main trend direction, the length reflected the degree of dispersion. The short axis showed the range of element distribution. This method offered a comprehensive understanding of spatial patterns, making it valuable for data analysis. ArcGIS10.8.1 was used in the analysis of the standard deviational ellipse.

$$\:\begin{array}{c}\begin{array}{c}{\sigma\:}_{x}=\sqrt{2}\sqrt{\frac{\sum\:_{i=1}^{n}\:{\left(\stackrel{\sim}{{x}_{i}}\text{cos}\theta\:-\stackrel{\sim}{{y}_{i}}\text{sin}\theta\:\right)}^{2}}{n}}\end{array}\end{array}$$

(2)

$$\:\begin{array}{c}\begin{array}{c}{\sigma\:}_{y}=\sqrt{2}\sqrt{\frac{\sum\:_{i=1}^{n}\:{\left(\stackrel{\sim}{{x}_{i}}\text{sin}\theta\:+\stackrel{\sim}{{y}_{i}}\text{cos}\theta\:\right)}^{2}}{n}}\end{array}\end{array}$$

(3)

Spatial panel data model

To ensure the independence of the explanatory variables, a test for multicollinearity and variance inflation factor (VIF) was conducted. A VIF value of less than 10 indicated the absence of multicollinearity.

The data of this study were both temporal and spatial, forming panel data. Compared to the general regression model, the spatial model fully considered the autocorrelation of spatial data, leading to better parameter estimation validity.

The traditional panel data model lacked spatial correlation by overlooking individual spatial effects [19]:

$$\:\begin{array}{c}\begin{array}{c}{y}_{it}={X}_{it}^{{\prime\:}}\beta\:+{\mu\:}_{i}+{\epsilon\:}_{it}\end{array}\end{array}$$

(4)

where $\:i=\text{1,2},\cdots\:,n$ denotes individuals in the cross-section; $\:t=\text{1,2},\cdots\:,t\:$indicates different time points; $\:{y}_{it}$ is the dependent variable at the time point $\:t$ of cross-section $\:i$; $\:X$ is the explanatory variable; $\:\beta\:$ is the regression coefficient; $\:{\mu\:}_{i}$ is the individual spatial effect, which controls for all background variables that do not vary over time; $\:{\epsilon\:}_{it}$ is a random error term with mean 0 and variance $\:{\sigma\:}^{2}$ and satisfying an independent normal distribution.

Spatial panel data mainly imported spatial lagged term or spatial error terms in traditional panel model [20, 21]. Those models were known as the Spatial Lag model (SLM) and Spatial Error model (SEM) respectively. Another model imported these two terms simultaneously was the Spatial Durbin model (SDM). The SLM considered the potential spatial correlation among dependent variables, while the SEM assumed that the spatial correlations arise from neglected variables [22]. Spatial autocorrelation was presented in SDM not only between adjacent study regions for the dependent variables, but also for the explanatory variable across neighboring regions. The structure of these models was defined as follows:

SLM

$$\:\begin{array}{c}\begin{array}{c}{y}_{it}=\delta\:{\sum\:}_{j=1}^{n}{W}_{ij}{y}_{jt}+{X}_{it}^{{\prime\:}}\beta\:+{\mu\:}_{i}+{\epsilon\:}_{it}\end{array}\end{array}$$

(5)

where $\:i$銆�$\:t$銆�$\:X$銆�$\:\beta\:$銆�$\:{\epsilon\:}_{it}$ are the same as in the model (4); $\:\delta\:$ is the spatial autoregressive coefficient, which indicates the existence of spatial spillover between adjacent spatial individuals, and its magnitude reflects the degree of spatial interactions; $\:{W}_{ij}$ is the $\:n\times\:n$ spatial weight matrix (SWM) [23]; $\:{W}_{ij}$ is its element, and $\:j$ represents a region different from $\:i$.

SEM

$$\:\begin{array}{c}\begin{array}{c}{y}_{it}={X}_{it}^{{\prime\:}}\beta\:+{\mu\:}_{i}+{\phi\:}_{it}\end{array}\end{array}$$

(6)

$$\:\begin{array}{c}\begin{array}{c}{\phi\:}_{it}=\rho\:\sum\:_{j=1}^{n}{W}_{ij}{\phi\:}_{jt}+{\mathcal{E}}_{it}\end{array}\end{array}$$

(7)

where $\:i$, $\:t$, $\:X$, $\:\beta\:$, $\:{\epsilon\:}_{it}$, $\:\delta\:$, $\:{W}_{ij}$ are the same as in the model (4) and (5); $\:{\phi\:}_{it}$ is the spatial error term; $\:\rho\:$ is the spatial autocorrelation coefficient, the magnitude of which reflects the degree of spatial correlation between the regression residuals.

SDM

$$\:\begin{array}{c}\begin{array}{c}{y}_{it}=\delta\:{\sum\:}_{j=1}^{n}{W}_{ij}{y}_{jt}+{X}_{it}^{{\prime\:}}\beta\:+{\mu\:}_{i}+{\phi\:}_{it}\end{array}\end{array}$$

(8)

$$\:\begin{array}{c}\begin{array}{c}{\phi\:}_{it}=\rho\:\sum\:_{j=1}^{n}{W}_{ij}{\phi\:}_{jt}+{\mathcal{E}}_{it}\end{array}\end{array}$$

(9)

where $\:\beta\:$ is the regression coefficient; $\:{\epsilon\:}_{it}$ is a random error term; $\:\delta\:$ is the spatial autoregressive coefficient; $\:{W}_{ij}$ is the SWM; $\:{\phi\:}_{it}$ is the spatial error term; $\:\rho\:$ is the spatial autocorrelation coefficient; have the same meaning as in the (4), (5), (6) and (7).

The Moran鈥檚 I [23] measured the level of similarity in attribute values between neighboring spatial elements. It could be used to assess the presence of signification spatial dependence in the residuals after regression using an Ordinary least square (OLS) model.

The Hausman test [24] and the LR joint significance test [25] were used to determine whether the spatial panel model should have spatial fixed effects, time-period fixed effects, or two-way fixed effects. The Lagrange Multiplier test (LM test) and the Robust Lagrange Multiplier test (rLM test) [26, 27] were used to determine the presence of spatial correlation. The initial assumption of the LM-lag and rLM-lag test was that there is no spatial lag term. The initial assumption of the LM-error and rLM-error test was that there is no spatial error term. If one LM test showed a significant effect and another indicated an insignificant effect, the significant formal spatial effects model should be used. If both the LM test indicated that the significant effects, then SDM should be chosen [28].

Post hoc test named the Wald test and Likelihood ratio (LR) test were used to enhance the robustness of the model [21]. These were used to determine whether the optimal SDM should be degraded to SLM or SEM [29]. The original hypothesis of the post hoc test assumed that the SDM could be degraded. If the result of the LM test or the rLM test were conflict with the posthoc test, Elhorst [21] thought that the SDM should be retained as it was better applied to the general situation.

Due to the import of the SWM in the SDM, the regression coefficients could not be directly used to explain the effect of the independent variables on the incidence. LeSage and others [30] decomposed the effect into direct effect and indirect effect using partial differential equation. The direct effect indicated the average influence of the independent variable on the region. Indirect effects reflected the average impact of the explanatory variable on the dependent variable in other neighboring regions. It accounted for the feedback effect on the dependent variable in the neighboring regions and its subsequence influence on the dependent variable in the current region.

The spatial adjacency weight matrix was generated with GeoDa 1.18.0 software. For spatial panel data model analysis, MATLAB R2022a software was utilized. The spatial panel data model code referenced the Econometrics Toolbox by James P. Lesage ( ). This study considers statistically significant if P is less than 0.05.

Results

The incidence of IIDs

The overall annual incidence of IIDs in Zhejiang Province was showed in Fig.听1. The results showed that the average annual incidence of HFMD was the highest and the typhoid was the lowest.

Fig. 1

Line graph of annual incidence rates of five intestinal infectious diseases in Zhejiang Province, 2011鈥�2021

From 2011 to 2021, a total of 21,728 cases of HEV were reported in Zhejiang Province, with an average annual incidence of 4.03/100,000. The highest incidence was in 2011 (5.58/100,000) and the lowest in 2020(2.50/100,000). Spatially, HEV cases were mainly clustered in Hangzhou City and the northwest. Qujiang District has consistently shown a high incidence of HEV(Fig.听2).

Fig. 2

The distribution of听reported incidence of HEV in Zhejiang Province during 2010鈥�2021

For bacillary dysentery, a total of 28,112 cases were reported, with an average annual incidence of 5.25/100,000. The highest incidence was observed in 2011 (12.17/100,000), while the lowest was recorded in 2021 (2.78/100,000). Although the incidence of bacillary dysentery decreased over the years, it remained high in Hangzhou City (Fig.听3).

Fig. 3

The distribution of听reported incidence of bacillary dysentery in Zhejiang Province during 2011鈥�2021

In the case of typhoid, there were 3,293 reported cases, and the overall incidence was low and exhibited a trend of decline over the years. The highest incidence occurred in 2011 (1.04/100,000), whereas the lowest was seen in 2021 (0.15/100,000). Before 2017, typhoid cases were primarily clustered in Ningbo City in the northeast and Wenzhou City in the south (Fig.听4).

Fig. 4

The distribution of听reported incidence of typhoid in Zhejiang Province during 2011鈥�2021

As for OIDs, a total of 2,279,514 cases were reported, with an average incidence of 223.00/100,000. The highest incidence occurred in 2017 (270.86/100,000), while the lowest was observed in 2020 (162.26/100,000). Zhuji District exhibited a consistently high incidence, with additional clusters detected in Hangzhou City and Jiaxing City (Fig.听5).

Fig. 5

The distribution of听reported incidence of OIDs in Zhejiang Province during 2011鈥�2021

HFMD accounted for 1,481,248 reported cases, with an average incidence of 274.38/100,000. The highest incidence occurred in 2018 (494.97/100,000), while the lowest was recorded in 2020 (114.38/100,000). HFMD cases were mainly clustered in Ningbo City, Wenzhou City, Hangzhou City, and Lishui City (Fig.听6).

Fig. 6

The distribution of听reported incidence of HFMD in Zhejiang Province during 2011鈥�2021

Spatial evolutionary trends in morbidity of IIDs

The spatial aggregation characteristics and evolutionary trends of IIDs in Zhejiang Province were analyzed using the center of gravity trajectory at six specific time points: 2011, 2013, 2015, 2017, 2019, and 2021.

The center of gravity for HEV incidence was primarily concentrated in Zhuji District and Yiwu District, with an overall eastward shift. HEV incidence was widespread and concentrated in Jinhua City, Shaoxing City, Hangzhou City, Lishui City and western Ningbo City. The distributing followed a north-south direction, exhibiting a more dispersed pattern (Fig.听7).

Fig. 7

Migration trend of HEV in Zhejiang Province from 2011 to 2021

Regarding bacillary dysentery, the incidence range gradually decreased, suggesting fewer regions with high incidence. The center of gravity shifted northwest from Zhuji District to Fuyang District, following a north-south direction with a more dispersed incidence (Fig.听8).

Fig. 8

Migration trend of bacillary dysentery in Zhejiang Province from 2011 to 2021

For typhoid, the center of gravity migrated from northeast Jinhua to Shaoxing City. The incidence range expanded, and the clustering changed from south-north to northeast-southwest becoming more concentrated incidence (Fig.听9).

Fig. 9

Migration trend of Typhoid in Zhejiang Province from 2011 to 2021

The center of OIDs gravity was concentrated in Zhuji District and Shaoxing City, with a more clustered incidence and no noticeable change in the trend of concentrated incidence (Fig.听10).

Fig. 10

Migration trend of OIDs in Zhejiang Province from 2011 to 2021

As for HFMD, the center of gravity initially moved northwards from Panan District and eventually back to Panan District. The incidence range was more extensive, and the clusters shifted northwards and then southwards (Fig.听11).

Fig. 11

Migration trend of HFMD in Zhejiang Province from 2011 to 2021

Multicollinearity test

In order to address potential heteroscedasticity, a square root inverse sine transformation was applied to the explanatory variables of component ratio, while a natural logarithm transformation was performed on the remaining variables. The basic statistical information of the explanatory variables was shown in Table听1. The distribution of the annual average of explanatory variables across districts and counties was shown in Fig.听12. Variables with VIF greater than 10 were excluded, resulting in a model with nine covariates: Density of population, Sex ratio, Aging ratio, GDP, Per capita GDP, Subsistence allowances ratio, Urbanization ratio, Number of beds per thousand population, Number of medical practitioners (assistants) per thousand population. The results of VIF were shown in Table听1.

Table 1 Basic information of variables and results of multicollinearity test

Fig. 12

Distribution of explanatory variables by district and county in Zhejiang Province

Spatial correlation test and model selection

The original hypothesis of 鈥渘o spatial autocorrelation鈥� was rejected at the significance level of 0.05 for HEV (Moran鈥檚 I鈥�=鈥�0.34, P鈥�<鈥�0.05), Bacillary dysentery (Moran鈥檚 I鈥�=鈥�51, P鈥�<鈥�0.05), Typhoid (Moran鈥檚 I鈥�=鈥�0.31, P鈥�<鈥�0.05), OIDs (Moran鈥檚 I鈥�=鈥�0.42, P鈥�<鈥�0.05), HFMD (Moran鈥檚 I鈥�=鈥�0.49, P鈥�&濒迟;鈥�0.05). Thus, introducing spatial autocorrelation using a spatial panel model should be considered, requiring the introduction of SWM.

According to the Hausman and LR test, the significant joint of spatial fixed effects were observed for each IIDs: HEV (LR鈥�=鈥�842.47, P鈥�<鈥�0.05), bacillary dysentery (LR鈥�=鈥�1257.28, P鈥�<鈥�0.05), typhoid (LR鈥�=鈥�860.86, P鈥�<鈥�0.05), HFMD (LR鈥�=鈥�793.28, P鈥�<鈥�0.05), and OIDs (LR鈥�=鈥�1769.77, P鈥�&濒迟;鈥�0.05). These results rejected the original hypothesis that the association of spatial fixed effects was not significant. Similarly, the significance joint of time-period fixed effects was observed for each IIDs: HEV (LR鈥�=鈥�71.80, P鈥�<鈥�0.05), bacillary dysentery (LR鈥�=鈥�153.36, P鈥�<鈥�0.05), typhoid (LR鈥�=鈥�30.83, P鈥�<鈥�0.05), HFMD (LR鈥�=鈥�512.85, P鈥�<鈥�0.05), and OIDs (LR鈥�=鈥�139.26, P鈥�&濒迟;鈥�0.05). These results rejected the original hypothesis that the association of time-period fixed-effect was not significant. Combining these two results showed that a spatial panel data model with both spatial and time-period fixation needs to be considered for each IIDs.

Both the LM test and rLM test rejected the original hypothesis at the significance level of 0.05. additionally, both the Wald test and LR test rejected the original hypotheses. Consequently, it is recommended to use the SDM for all analyses of IIDs. The results of model selection were showed in Table听2.

Table 2 Results of spatial correlation test and model selection

SDM with dual fixation in time space and effect decomposition

In the model, nine explanatory variables that passed the multicollinearity test were included and analyzed by incorporating both spatial lag and spatial error terms. Table听3 showed the results of SDM with dual fixation in time space and Effect decomposition.

Table 3 Spatial panel data model of IIDs

The results of HEV indicated a good model fit (R²鈥�=鈥�0.71). The results demonstrated that both the subsistence allowances ratio and the urbanization ratio significantly impact the incidence rate. The higher subsistence allowances ratio (b鈥�=鈥�0.23, P鈥�<鈥�0.05) in the local area corresponded to a higher incidence of HEV. The urbanization ratio in neighboring districts and counties (b鈥�=鈥�3.68, P鈥�<鈥�0.05) indirectly contributed to an increased incidence rate of HEV. Combining the direct and indirect effects, the overall effect of urbanization ratio (b鈥�=鈥�3.32, P鈥�<鈥�0.05) was positive. The results of SDM and effect decomposition for HEV were showed in Supplementary Table S1.

For bacillary dysentery, results revealed a high model fit (R²鈥�=鈥�0.89). The local density of population (b=-1.81, P鈥�<鈥�0.05) and the number of beds per thousand population (b=-1.14, P鈥�<鈥�0.05) were found to decrease the incidence. Additionally, the per capita GDP (b鈥�=鈥�11.24, P鈥�<鈥�0.05) and the urbanization ratio (b鈥�=鈥�5.17, P鈥�<鈥�0.05) in the neighboring area had a positive effect. Specific results were shown in Supplementary Table S2.

The results of typhoid indicated a good model fit (R²鈥�=鈥�0.69). The per capita GDP of the region (b鈥�=鈥�0.54, P鈥�<鈥�0.05) would increase the incidence, and the per capita GDP of the neighboring region (b鈥�=鈥�0.84, P鈥�<鈥�0.05) had a positive spillover effect on the local incidence. Combining the direct and indirect effects, the per capita GDP (b鈥�=鈥�1.39, P鈥�<鈥�0.05) showed an overall risk effect. The results of SDM and effect decomposition were showed in Supplementary Table S3.

The results for OIDs were shown in, with a high model fit (R²鈥�=鈥�0.89). The aging rate (b鈥�=鈥�2.39, P鈥�<鈥�0.05) and urbanization ratio (b鈥�=鈥�0.64, P鈥�<鈥�0.05) of the region increased the incidence, whereas the sex ratio (b=-0.01, P鈥�<鈥�0.05), the per capita GDP (b=-0.45, P鈥�<鈥�0.05) and number of beds per thousand population (b=-0.28, P鈥�<鈥�0.05) decreased the incidence. The per capita GDP (b=-0.50, P鈥�<鈥�0.05) and number of beds per thousand population (b=-0.27, P鈥�<鈥�0.05) of neighboring regions had a negative spillover effect on incidence in the region. Detailed results were showed in Supplementary Table S4.

The results of HFMD fitting were shown, with the good model fit (R²鈥�=鈥�0.76). The sex ratio (b=-0.07, P鈥�<鈥�0.05), per capita GDP (b=-0.27, P鈥�<鈥�0.05), and urbanization ratio (b鈥�=鈥�0.39, P鈥�<鈥�0.05) of the local area would reduce the incidence of HFMD. The per capita GDP (b=-0.66, P鈥�<鈥�0.05) and sex ratio (b=-0.06, P鈥�<鈥�0.05) of neighboring regions had a negative spillover effect on the incidence of HFMD in the region. Specific results were shown in Supplementary Table S5.

The spatial effect coefficients (W*deP.var.) for each IIDs were found to be statistically significant (P鈥�<鈥�0.05), which indicated that the incidence in the region would be affected by the incidence in neighboring regions. The coefficients indicated a positive spillover effect in the incidence of IIDs: HEV (b鈥�=鈥�0.28, P鈥�<鈥�0.05), bacillary dysentery (b鈥�=鈥�0.38, P鈥�<鈥�0.05), typhoid (b鈥�=鈥�0.36, P鈥�<鈥�0.05), OIDs (b鈥�=鈥�0.28, P鈥�<鈥�0.05) and HFMD (b鈥�=鈥�0.40, P鈥�&濒迟;鈥�0.05).

Discussion

In this study, we established eleven-year panel data of district and county, explored the migration trend, social influencing factors, and spillover effect of IIDs.

The results showed that IIDs in Zhejiang Province have consistently decreased. This decline may be attributed to the continuous enhancement of infectious disease prevention and control capabilities, improvements of the living conditions in urban and rural residents, and the implementation of immunization planning strategies [31,32,33,34]. In 2020, the incidence of HEV, OIDs, and HFMD were at their lowest. This could potentially be attributed to the preventive measures taken during the COVID-19 pandemic [35]. The empirical findings revealed that the incidence of IIDs were affected not only by local factors but also by the neighboring districts and counties, indicating a spatial spillover phenomenon. This indicated that there was a clustering pattern in the incidence of IIDs, which aligns with previous studies [3, 36].

The incidence of IIDs was influenced by the economic and urban development of districts and counties, but it had different effects on different diseases. HFMD primarily transmitted through contact, showed a higher cluster in Jinhua and Shaoxing, regions with slightly poorer economic development. This concentration suggested that higher urbanization and better economic conditions lead to lower incidence, aligning with Li Liang鈥檚 study in Jiangsu Province. It could be relatively poorer overall hygiene conditions and inadequate hand-washing habits in rural regions. Additionally, HEV, bacillary dysentery, typhoid, and OIDs transmitted through water and food had varying economic impacts. Hangzhou and Ningbo, with high per capita GDP and urbanization rate, showed a concentration of bacillary dysentery and OIDs, consistent with previous studies [37]. However, this pattern contradicted the global incidence trend of infectious diseases, where regions with better economic development typically had a lower incidence. On the one hand, in rural regions, bacillary dysentery and OIDs would not receive proper attention or accurate diagnosed, often being labeled as acute gastroenteritis or diarrhea during consultation [37]. On the other hand, the distribution of medical resources was unbalanced, resulting in better healthcare conditions in economically developed regions compared to less developed ones. Consequently, patients increasingly seek treatment at tertiary hospitals in economically regions, leading to potentially higher actual incidence in economically disadvantaged regions than what is currently reported. The cluster of typhoid in Shaoxing, where the urbanization ratio is low but the per capita GDP is relatively high, and the spillover effect of the per capita GDP of neighboring regions may also increase the incidence, which was consistent with the study of Zhejiang province by Hua Gu [38]. It would be related to the fact that regions with high per capita incomes attract more migrants. Regions with high urbanization and economic prosperity tended to have lower incidence, possibly due to better living habits and healthcare conditions. HEV exhibited a higher incidence in regions with higher subsistence allowances ratio. Studies suggested the main influencing factor for HEV in China has shifted from water supply and sanitation conditions to the degree of economic development [39, 40], thanks to the introduction of the HEV239 Vaccine and improved health resources. Regions with a significant proportion of residents relying on subsistence allowances lag behind in economic development and required improved health environment.

The study also investigated the association between demographic characteristics and IIDs. We found that regions with higher density had the lower incidence of bacillary dysentery, which was consistent with the findings of Xiangxue Zhang et al. [41] in the Beijing-Tianjin-Tangshan regions. These indicated that higher populations facilitated accelerated the spread of the disease. Moreover, the higher sex ratio, indicating the higher proportion of males in the region, was linked to increased incidence of HFMD, which was consistent with the study by Xinguang Yin et al. [42]. They suggested that this could be attributed to women having more frequently contact with children at home and work, thereby increasing their exposure to HFMD patients and elevating the disease risk. Additionally, the higher aging ratio corresponded to the higher incidence of OIDs, which was in line with previous studies [43, 44]. This may be due to OIDs having a wider variety of pathogens and the impact of aging on the immune system of the host. However, it is worth noticing that the susceptible population of OIDs mainly concentrated on young children and adolescents [45, 46]. It is important to acknowledge that our study did not involve the analysis of different age groups due to the data accessibility, which is a limitation.

IIDs were also influenced by regional health resources. The higher number of beds per thousand population corresponded to the lower incidence of bacillary dysentery and OIDs. Moreover, regions with high per capita economic income and abundant medical and health resources exhibited lower IIDs incidence. This indicated that improved literacy among residents regarding infectious disease prevention and control had also contributed to the decline in the incidence of infectious diseases to some extent [47].

The economic development of neighboring regions also affected the incidence of local infectious diseases. Li Xiaoping et al. [36] analyzed the relationship between regional transmission of infectious diseases based on the perspective of spatial spillover. They pointed out that the increase in the urbanization level of the eastern region, where the degree of urbanization is higher, was conducive to the inhibition of the spread of infectious diseases in the region, also inhibited the spillover effect of infectious diseases in the inter-regional region.

There were some limitations in this study. Firstly, it鈥檚 crucial to consider the issue of surveillance data proposed by Haochen Wu [37] when examining the relationship between the economy and infectious diseases. The actual number of cases could be higher than reported cases. In view of the problem of underreporting of epidemiological data, relevant studies have proposed how to correct it [48, 49]. Besides, we did not analyze meteorological factors on the incidence of IIDs. There have been numerous studies investigating the influence of meteorological factors on the incidence of IIDs [50,51,52,53], highlighting not only the relationship between environmental factors and diseases but also the interaction among the various factors. Furthermore, some factors exhibit a turning point concerning their relationship with infectious diseases, when reached a certain level may no longer have a suppressive effect. These aspects necessitate further in-depth research.

Conclusion

From 2011 to 2021, the incidence of IIDs in Zhejiang Province exhibited a certain degree of clustering, with major hotspots identified in Hangzhou, Shaoxing, and Jinhua. High levels of urbanization rate as protective factors against HEV and Typhoid, yet pose as risk factors for Bacillary Dysentery and IIDs under category C. When examining the influence of social factors on these diseases, it is essential to consider the spillover effects from neighboring regions and implement targeted measures to enhance disease prevention based on regional development.