Spatial Endemic Situation And Forecast For Schistosomiasis In Hubei Province

Objectives1. To analyze the epidemic situation and spatial distribution of schistosomiasis and oncomelania in Hubei Province for the purpose of exploring high-risk areas and providing guidance on schistosomiasis control.2. To analyze the relationship between snail distribution and climate for the purpose of determining the main influencingfactor of climate and providing scientific evidence to control the snail population.3. To evaluate the short-term change trend of the prevalence of Schistosoma japonicum infection in humans and bovines for the purpose of providing a theoretical basis for schistosomiasis forecasting and early warning, as well as to provide a simple and feasible method for the short-term forecasting of S. japonicum infection prevalence.Methods1. We retrospectively collected data on endemic schistosomiasis in humans, bovines, and snails from the county base from2008to2012. The data on endemic schistosomiasis were matched to Geographic Information System (GIS) geospatial databases to constitute the spatial analysis database of schistosorniasis. The index of patients and infected bovines, infection rate of human and bovines, snail area, and percentage of snail area were observed using descriptive analysis method. The spatial aggregation of S. japonicum infection and snail distribution were analyzed by global spatial autocorrelation using Moran’s I index and local spatial autocorrelation of the Getis-Ord Gi*index.2. Snail data from18counties in Hubei province in2009were collected to calculate the average density of living snails and snail S. japonicum infection. Relevant information on climatic factors was also collected. The average densities of live snails and S. japonicum infection rate converted to the square root were used as response variables. The mean annual temperature, average annual minimum temperature, average annual relative humidity, average annual rainfall, and average annual sunshine hours were used as independent variables. The relationship between the average density of living snails,S. japonicum infection, and climatic factors was analyzed by global and local spatial regression models.3. The time-series Auto regression integrated moving average (ARIMA) model was constructed to forecast the short-term change trend of S. japonicum infection prevalence in human and bovines in Hubei province from1987to2013.Results1. Spatial characteristics of endemic schistosomiasis in Hubei provinceThe number of schistosomiasis patients and infection rate of S. japonicum were both decreased from2008to2012. The global autocorrelation analysis results on the infection rate of S. japonicum for five years were statistically significant (Moran’s I>0, P<0.01), which suggested that spatial agglomerations were present in the global autocorrelation analysis results of S. japonicum infection from2008to2012. Local autocorrelation analysis showed that the number of highly aggregated areas ranged from eight to eleven within the five-year analysis period. The highly aggregated areas were mainly distributed in eight counties, namely, Jingzhou district, Shashi district, Jiangling county, Gongan county, Shishou city, Jianli County, Honghu city, and Chibi city.The livestock examined were mainly bovines found around the endemic areas of Hubei province. The number of infected bovines and infection rate of S. japonicum were both decreased. The global autocorrelation analysis results of the infection rate of S. japonicum within five years showed that results were statistically significant (Moran’s I>0, P<0.01), whichsuggested that spatial agglomerations were present in the bovines with S. japonicum infection from2008to2012. Local autocorrelation analysis showed that the number of highly aggregated areas ranged from six to nine in within the five-year period. The highly aggregated areas were mainly distributed in six counties, namely, Shashi district, Jiangling county, Gongan county, Shishou city, Jianli county, and Honghu city.The snail areas decreased slightly but not significantly from2008to2012. The percentage of snail areas initially increased and subsequently declined. However, the change was minimal. The global autocorrelation analysis results of the percentage of snail areaswithin five years were statistically significant (Moran’s I>0, P<0.01), which suggestedthe presence of spatial agglomerations in the percentage of snail areas from2008to2012. Local autocorrelation analysis showed that the number of highly aggregated areas was distributed in eight counties, namely Jiangling county, Gongan county, Shishou city, Jianli county, Honghu city, Xiantao, Hannan district, and Jiayu county.2. Analysis on snail distribution and climatic factorsThe least squares regression (OLS) model of the average density of live snails and climatic factors was performed, and the residuals of multiple linear regression model were subjected to spatial autocorrelation (Moran’s I=0.2780, P=0.000). Therefore, analysis using only the general linear regression analysis is not sufficient and further application of spatial regression model is required. Lagrange multiplier statistic results suggest that the spatial lag model (SLM) should be employed. The spatial parameter p of SLM model was significant (p=-0.4365, P<0.05). The R2and LIK values of the SLM model were larger than that of OLS. Moreover, AIC and SC values were smaller than that of OLS, which indicated that the SLM model significantly was better than the OLS model. The SLM model results showed that the average density of live snails was positively correlated with annual average temperature and rainfall (P<0.05). The average density of live snails was also positively correlated with the average annual minimum temperature and annual average relative humidity, negatively correlated with annual average sunshine hours. However, all correlations were not statistically significant (P>0.05). Geographically weighted regression (GWR) was further used for local spatial regression analysis. The results showed that AIC values of the GWR model (9.2714) was smaller than that of the OLS (15.9756) and SLM (12.0415) models, indicating that GWR model fitted better. Parameters estimated by the GWR model were not the same in each study area and changed within a certain range.The OLS model of snail S. japonicum infection and climatic factors was utilized and the residuals of multiple linear regression model were subjected to spatial autocorrelation (Moran’s I=0.1828, P=0.000). Therefore, further application of spatial regression model is required, and the results obtained using Lagrange multiplier statistics suggested that SLM model should be employed. The spatial parameter p of the SLM model was significant (p=-0.1515, P<0.05). The R2and LIK values of the SLM model were larger than that of OLS. Moreover, the AIC and SC values were smaller than that of the OLS model, which indicated that the SLM model fitted better than the OLS model. The SLM model results showed that snail S. japonicum infection were positively correlated with annual average temperature (P<0.05). The snail S. japonicum infection were also positively correlated with the average annual minimum temperature, annual rainfall and relative humidity and negatively correlated with annual average sunshine hours. However, all correlations were not statistically significant (P>0.05). GWR was used for local spatial regression analysis. The results showed that AIC values of GWR model (4.0395) was smaller than that of the OLS (7.8222) and SLM (7.5764), indicating that GWR model fitted better. Parameters estimated by the GWR model changed within a certain range in each study area, which showed spatial heterogeneity.3. Time-series ARIMA modelwas used to forecast the prevalence of S. japonicum infectionTime-series ARIMA model was well fitted to the data on S. japonicum infection prevalence. The actual value of schistosomiasis infection and the prevalence of S. japonicum infection in human and bovines were all in the95%confidence interval of its predicted values. The prediction results showed that the prevalence of S. japonicum infection in human and bovines would continue to decrease slightly in the next five years.Conclusion1. Global autocorrelation analysis identified the schistosomiasis endemic clustering in Hubei province on the global view. Local autocorrelation analysis can further detect the locations with positive spatial clustering. The spatial clustering proved to be significant to the endemic schistosomiasis of human and bovines, snail distribution, and high cluster areas that were mainly distributed in the Yangtze River Basin of Jianghan Plain area. Given that Hubei province reached the schistosomiasis control criterion, the schistosomiasis endemic decreased and the accumulation scope changed; however, the snail endemic was also minimally altered.2. When comparing the data of snails on spatial autocorrelation, heterogeneity, and climate factors, the results derived from spatial regression analysis were superior to those of classical linear regression analysis. The major climatic factors that affected live snail average density were annual average temperature and rainfall. The main climate factor that influenced snail S. japonicum infection was annual average temperature.3. The time-series ARIMA model showed high-quality prediction accuracy. The time-series ARIMA model can be used for the short-term forecasting of schistosomiasis according to the actual needs of schistosomiasis control.