Exploration For Influencing Factors Of Hemorrhagic Fever With Renal Syndrome Based On Spatial Panel Data Model

Hemorrhagic fever with renal syndrome(HFRS),a rodent-borne natural focal disease caused by hantavirus(HV).HV is widely distributed all over the world,and a variety of new HV are constantly being discovered.The global HV infections showed an upward trend of fluctuation.HFRS is widespread in China and pose a serious threat to human health.From 2004 to 2019,a total of 209,209 HFRS cases and 1,855 deaths were reported in mainland China,indicating that the epidemic prevention and control situation is still very serious.Shandong Province is one of the key epidemic areas of HFRS in China.In recent years,the number of cases in Shandong has been in the front rank among all provinces in China.HFRS has become a major public health problem in Shandong Province.The occurrence,transmission and prevalence of HFRS involve multiple stages,and many factors,such as ecological environment,socioeconomic status and healthcare systems,can affect the epidemic of HFRS.HFRS data based on the infectious disease surveillance system includes not only the cross section data of different time points,but also the time series data of each region.The analysis of the spatiotemporal data can provide scientific reference for epidemic prevention and control.With further development of spatial econometrics,spatial panel data models,which can process spatially dependent data and also allow for consideration of spatial heterogeneity,are becoming useful tools for analyzing spatiotemporal data,especially infectious disease data.Based on panel data models,spatial panel data models control the spatial dependence and heterogeneity of data by incorporating the spatial interaction of cross-sectional dimension,and then discuss the direct effect and spatial spillover effect of influencing factors on the risk of disease.Therefore,in this study,methods of spatiotemporal statistical analysis were used to reveal spatiotemporal patterns of HFRS,spatial econometric models were used to estimate the impact of socioeconomic and environmental factors on the incidence of HFRS.These results could provide theoretical basis for HFRS prevention and control and relevant policy making.Methods:The data of HFRS cases,socioeconomic factors and ecological environmental factors from 2010 to 2019 in Shandong Province at the county scale were collected.Descriptive analysis were conducted to analyze the epidemiological characteristic of HFRS.Using spatial autocorrelation and spatiotemporal scanning analysis,the spatiotemporal pattern of HFRS incidence in Shandong Province was explored,and the areas and periods of high-incidence clusters were detected.We also estimated the relative risk of the aggregation areas.The mixed model,panel data model and spatial panel data model were constructed to analyze the influence of socioeconomic and ecological environmental factors on the annual HFRS incidence.Comparing goodness-of-fit of models,we selected the best model to further analyze whether there is spatial spillover phenomenon.Furthermore,the monthly data were modeled to analyze the direct effect and spatial spillover effect of meteorological factors on the incidence of HFRS under the premise of considering the lag effect of meteorological factors.Results:1.From 2010 to 2019,there were 12,185 HFRS cases in Shandong Province,including 156 deaths,with a fatality rate of 1.28%,The highest incidence was 1.87/100,000 in 2013,and the lowest was 0.72/100,000 in 2019.The prominent peak of HFRS is from October to December and a low peak is from April to June.As for spatial distribution,counties with high incidence mainly concentrated in central and southeastern regions.Among all cases,the 31-70 years age group accounted for the highest proportion(81.01%).The number of male patients was higher than that of female patients,with a sex ratio of 2.64:1.Farmers accounted for 84.14%of all cases,followed by workers(5.38%)and students(2.39%).2.The incidence of HFRS in Shandong Province has spatial autocorrelation,with the global Moran’s Iindex ranging from 0.161 to 0.575 each year.Spatiotemporal scanning analysis showed that there was a most likely cluster including Qingdao,Zibo,Dongying,Weifang,Rizhao and Linyi City,with the aggregation time from October 2012 to March 2015(RR=4.73,P<0.001).The secondary clusters were Jinxiang County from December 2015 to May 2018(RR=6.91,P<0.001),Jiaxiang County from December 2012 to May 2015(RR=5.50,P<0.001)and Pingyin County from March 2013 to June 2014(RR=3.92,P=0.013).3.After adjusting spatial individual effect,spatial autocorrelation and temporal autocorrelation,the spatial panel data model showed that annual HFRS incidence was negatively related with population density(β=-0.595,P<0.001),and relative humidity(β=0.046,P=0.006)was positively correlated with HFRS incidence in dynamic spatial panel data model;The spatial panel data model was fitted to the monthly data of high and small peak periods,we found that the monthly incidence of HFRS in each county was positively correlated with the incidence in adjacent areas(spatial autoregressive coefficient δ=0.155,P<0.001),and average wind speed from April to June affected the incidence in the large peak period(β=0.470,P<0.05);For small peak period,the spatial spillover effect of average wind speed from March to May can significantly impact HFRS incidence of neighboring counties(P<0.05).In the analysis of different time scales,the goodness-of-fit of the model in the order was spatial panel data model>panel data model>mixed model.Conclusions:1.The trend of HFRS in Shandong Province is similar to the disease status of historical epidemic cycles.At present,the epidemic situation is stable at a low level.The incidence showed obvious seasonal distribution characteristics,and middle-aged and elderly farmers are more susceptible to the infection.2.The distribution of HFRS cases in Shandong Province showed spatial autocorrelation and spatiotemporal aggregation.The high-risk areas are mainly concentrated in the central and southeastern regions of Shandong Province and some counties in the southwest.3.The goodness-of-fit of spatial panel data model is superior to classical mixed model and panel data model,and it can explore spatial spillover effects of influencing factors.The spatial panel data model provide a reference for the spatiotemporal analysis of HFRS monitoring data,which contribute to the formulation and implementation of epidemic prevention and control policies.