Impact Of Meteorological Factors On Other Infectious Diarrhea And Forecasting And Early Warning

BackgroundClimate change has caused the changes in the temporal and spatial characteristics of climate-sensitive diseases,thereby increasing the burden of some climate-sensitive diseases,so it is considered to be one of the important factors threatening human health.In the face of a changing climate environment,a key measure to maintain and enhance human health is to determine the impact of meteorological factors on sensitive infectious diseases and their future evolution trends.In recent years,the incidence of some climate-sensitive diseases is still high in China,which has attracted a large number of attention.In-depth study of the impact of meteorological factors on sensitive infectious diseases and early warning models can provide important reference for relevant departments to formulate health policies and improve the ability to face changing climate conditions.In China,in order to facilitate the management of infectious diarrhea,diarrhea other than cholera,dysentery,typhoid and paratyphoid is defined as other infectious diarrhea according to the prevalence and severity of infectious diarrhea.The incidence of other infectious diarrhea ranks second among category C notifiable infectious diseases,which poses a considerable challenge to the medical service of our country.Previous literature focused on description of the three distribution characteristics of other infectious diarrhea,ignoring the temporal and spatial trends of its characteristic.Several studies have shown that other infectious diarrhea is a climate-sensitive infectious disease,which is closely related to climate change.Previous studies have shown that other infectious diarrhea is clustered in time and space,but most of studies have only analyzed the effect of meteorological factors on other infectious diarrhea from a time perspective,lacking analysis from a spatiotemporal perspective.At the same time,there are still lacking studies to assess the attributable risk of meteorological factors on other infectious diarrhea from the perspective.In addition,due to different immunity and adaptability of different populations to climate change,the prevalence of other infectious diarrhea varies among different populations.However,few studies have explored the effect of meteorological factors on other infectious diarrhea in subgroups such as sex,age and occupation.Accurate predictions can predetermine the occurrence of the disease,and the use of intervention measures in advance can reduce the adverse effects of the disease.Although there are some prediction studies of other infectious diarrhea based on meteorological factors,most of the studies are limited to a certain county or city.And the prediction time is short,usually a few weeks to one or two months,and there is no evaluation on the applicability of multiple forecasting models using the same meteorological-disease dataset.In terms of early warning,the early warning response system of other infectious diarrhea issues warning signals through routine reported data in China,which is unable to warn the diseases that have not yet occurred,so it is less feasible in practical work.Objectives(1)To understand the changing trend of spatiotemporal distribution of the incidence of other infectious diarrhea in China from 2004 to 2019,and to understand the distribution characteristics of other infectious diarrhea from 2014 to 2019.(2)To study the spatiotemporal risks and attributable risks of meteorological factors to other infectious diarrhea,and to identify vulnerable groups.(3)Establish forecasting models based on meteorological factors in cities with a high incidence of infectious diarrhea and evaluate their applicability,and use the optimal forecasting model to establish an early warning model in Beijing.MethodsFrom 2004 to 2019,the annual incidence data of other infectious diarrhea in 31 provinciallevel administrative regions were obtained from the Public Health Science Data Center in mainland China.The daily data of other infectious diarrhea in 334 prefecture-level city and 4 municipalities across the country from 2014 to 2019 came from the National Notifiable Infectious Diseases Reporting Information System.Meteorological data came from the China Meteorological Data Sharing Service Network.Demographic data was obtained from the Statistical Yearbook of each city.In order to avoid the instability of the results,316 cities with the number of other infectious diarrhea greater than the 5th percentile(P5)were screened as the study area.Joinpoint regression model and hotspot analysis were used to identify the temporal and spatial trend of other infectious diarrhea in China from 2004 to 2019,and the distribution characteristics of other infectious diarrhea from 2014 to 2019 were described using descriptive epidemiological methods.The Bayesian spatiotemporal model was used to explore the spatiotemporal risk effect of meteorological factors on other infectious diarrhea on the national scale.In addition,the risk trend of other infectious diarrhea was described from 2014 to 2019 in China,and 316 cities were divided into three categories:hot cities,cold cities,non-cold and non-hot cities according to the posterior distribution probability of the risk.The attributable risk assessment under the framework of the distributed lag non-linear model was used to calculate the attributable risk of other infectious diarrhea caused by meteorological factors,and then identify vulnerable populations by attributable risk.Cities with the largest number of other infectious diarrhea were selected to establish five forecasting models from the 10 meteorological geographic divisions in China(without Tibet),including seasonal autoregressive integrated moving average with external regressors,generalized additive model,random forest,support vector machine and boosted regression tree(BRT).The coefficient of determination(R2),root mean square error(RMSE),and mean absolute error(MAE)were used to evaluate their prediction accuracy and applicability.The city with the highest number of other infectious diarrhea was selected to establish an early warning model using the model with the most applicability.We set the P75 of the historical weekly incidence of other infectious diarrhea as the early warning threshold,and an early warning signal will be issued when the number of predicted cases exceeds this threshold.Sensitivity,specificity,Youden index and area under the ROC curve(AUC)were used to evaluate early warning models.Results(1)From 2004 to 2019,the incidence of other infectious diarrhea continued to increase at an annual rate of 7.7%in China.Except for Beijing,the incidence of other provincial administrative regions showed an upward trend.The hotspots of other infectious diarrhea gradually increased,and the confidence level of hotspot gathering area gradually increased.From 2014 to 2019,a total of 5 844 066 cases of other infectious diarrhea were reported in 338 cities of China,with a sex ratio of 1.26:1.Other infectious diarrhea occurred in all age groups,and the incidence was mainly concentrated in people under 5 years old(59.47%).And children have the highest incidence(58.66%)among occupational population.In terms of time distribution,other infectious diarrhea has obvious seasonal peaks during summer and autumn,and winter;in terms of spatial distribution,the number of other infectious diarrhea in BeijingTianjin-Hebei region,Chongqing,Shanghai and its surrounding cities were relatively high.(2)The results of the Bayesian spatiotemporal model showed that for every 1℃ increase in monthly average temperature,the risk of other infectious diarrhea in the whole population increased by 0.7%,and 95%confidence interval(CI)was 0.5%-0.9%.The monthly average relative humidity increase every 1%,the risk of other infectious diarrhea in the whole population reduced by 0.3%(95%CI:-0.6%～-0.1%).The results of the attributable risk assessment showed that men,people under 5 years old,and children had the highest attributable risk in gender,age,and occupational subgroups,respectively.(3)The evaluation results of the five forecasting models showed that the BRT model has the strongest applicability.The average values of R2,RMSE and MAE in the test set are 0.75,41.79 and 31.73 in 10 high-incidence cities,respectively.Further,the BRT model was used to establish early warning models for other infectious diarrhea in Beijing,which has the largest number of cases.When P75 of the actual number of cases was used as the early warning threshold,the sensitivity of the early warning model on the test set was 92.31%and the specificity was 97.44%.The Youden index was 0.897 and the AUC value was 0.976.The warning effects good,and the warning effect is stable when the warning threshold is adjusted.Conclusions(1)From 2004 to 2019,the incidence of other infectious diarrhea showed an upward trend in other provincial administrative regions except Beijing,and the hotspot gathering area also gradually increased.Other infectious diarrhea have two seasonal peaks,occurred during summer and autumn,and winter every year,respectively.(2)Higher average temperatures increase the risk of other infectious diarrhoea,and men,people under 5 years old and children are vulnerable groups exposed to average temperature.Relative humidity is negatively correlated with other infectious diarrhea.(3)The BRT model has high forecast accuracy and good applicability,and other infectious diarrhea early warning models established based on BRT have good early warning ability.