Study On Scrub Typhus Risk And Risk Prediction Based On The Driving Effect Of Rat Density And Meteorological Factors

Background:Guangzhou,a high-incidence areas and key points in epidemic prevention and control of scrub typhus,is located in the subtropical and coastal zone with a warm and humid climate.It facilitates the disease propagation due to abundant plant species and wildlife.With the increasing number of scrub typhus cases and the expansion of epidemic area,it is necessary to understand the epidemic characteristics and spatiotemporal epidemic characteristics of scrub typhus and construct the risk prediction model,which will provide guideline for early warning and policy making of scrub typhus in the future.Objective:To understand the epidemiological characteristics of scrub typhus including the temporal,demographic distributions and spatial-temporal cluster analysis;To explore the correlation between the monthly incidence of scrub typhus and rat density,meteorological and environmental factors in Guangzhou;To construct prediction models of scrub typhus with random forest regression by integrating rat density,meteorological and environmental factors as predictive factors.Methods:1.Monthly scrub typhus surveillance data from 2006-2019 were collected at the Chinese National Communicable Disease Surveillance Network.Descriptive epidemiology were used to analyze the epidemiological characteristics of scrub typhus case;Global spatial autocorrelation analysis,partial autocorrelation analysis and spatiotemporal cluster analysis were conducted to detected the spatiotemporal clusters of scrub typhus occurrence.2.Monthly rat density surveillance data from 2006-2019 were collected at Guangzhou Center for Disease Control and Prevention.Monthly meteorological data from Guangzhou during2006-2019 across China were obtained from the China Metaorological Data Sharing Service System.Population density,Average arable land area and Green coverage area during 2006-2019were collected at Guangzhou Statistics Bureau.The Cross-correlation analysis was used to judged the the lag time between related effected factors.Generalized additional model was used to explore the correlation between the monthly incidence of scrub typhus and rat density,meteorological and environmental factors.3.Monthly scrub typhus cases,monthly rat density,monthly meteorological factors data,population density,Average arable land area and Green coverage area during 2006-2019 as predictive factors was applied to construct Random forest prediction models of scrub typhus,which can predict the risk of scrub typhus and screen the importance of predictive factors.The Microsoft Office 2010,Arc GIS 10.7,Sa TScan 9.1,and R 3.6.3 were used in this study.Results1.There were totally 11,120 cases reported in Guangzhou during 2006 to 2019 with the average annual incidence rate of 4.50/10 million.The incidence of scrub typhus has increasing yearly..The ratio of occurrence between men and women is 0.96:1 with statisticallysignificance.The majority of cases occurred in the age group of 50-70 years.Themost common occupation of the infected patients was farmer.The case distribution of scrub typhus werespatiotemporal cluster in Guangzhou.LISA analysis showed hotspots(High-High)was primarily located in Conghua district.Space-time scan statistic demonstrated there are four clusters and the primary clusters for high incidence of scrub typhus were found at Conghua and ZengchengDistricts.2.Generalized additional model shows that the incidence of scrub typhus is positively associated with the lagging effect of rat density,monthly maximum temperature,monthly minimum temperature,monthly relative humidity and negatively correlated with monthly mean air pressure and monthly mean wind speed.3.The predictive factors including monthly mean air pressure,monthly minimum temperature,monthly average arable land area,monthly rat density,green coverage area,monthly maximum temperature monthly population density with the optimum lag period as the predictive factors to constructed the random forest regression model,which possesses asuperior predictive ability.R~2reached 84.17%,RMSE reached 23.79。ConclusionThe annual incidence of scrub typhus has increasing yearly during the 14-year study period.Guangzhou had demonstrate apparent spatial distribution differences.The prevention and control strategies of high hotspots and high-risk groups should be intensified.Eliminating both mites and rodent in advance can effectively control the development of scrub typhus.The random forest model can early predicted the risk of scrub typhus.