| ObjectivesIn order to better prevent and control of hand foot and mouth disease,protect the preschool children,improve the prevention and control measures,formulate more reasonable strategies,we performed researches on epidemic characteristics,risk factors,distribution of pathogens,incidence trend prediction and outbreak simulation of hand,foot and mouth disease in Yichang city,Hubei Province,China.Methods1.Software program GIS was used to plot morbidity distribution of county level.Software program Geocoding was used to position the latitude and longitude for all the reported cases,and then density analysis was performed to plot kernal density and point density of the distribution of these cases and their pathogens.2.Baseline field investigation on risk factors of individuals and kindergartens was conducted by specialists of local diseases control and prevention center.Software program Epidata was used to construct database,then student's t test and multiple stepwise logistic regression analysis were performed to calculate odds ratios and their 95%confidence intervals,then use the results to select risk factors which were statistical significance.3.Based on monthly incidence rates of hand foot and mouth disease from January 2008 to December 2016,linear prediction series and residual series were generated by using optimal ARIMA model.Then new residual series were generated by NARNN model based on the previous residual series which had been generated by ARIMA model.The final result was a compound result of ARIMA-NARNN model.4.Three statuses of incidence rate?[0,7.5?,[7.5,10),[10,??),per 100,000)has been selected according with characteristics of monthly incidence rates from 2008 to 2016.Then markov chain has been built to calculate transition probability of each month in year 2017,and predict the statuses of incidence rate.5.By using software program Matlab,SIR model was established by propagation dynamics method.combining with the global sensitivity and uncertainty analysis.we could grasp the influence between parameters,and could realize the outbreak simulation process that assist of judging the outbreak tendency.Results1.Epidemiological characteristics:A total of 37988 cases were reported from 2008 to 2016.The reported incidence rate was present serration,high in even number years and low in odd.Two peaks could be seen from April to June and November to December all through the year.Kruskal-Wallis test results showed statistical significance in towns alongside the Yangtze River between towns which not alongside?x?2=117.93,P<0.0001?.Point density analysis showed that cases integrated in Xiling.Yiling and Yidu.The difference between gender?X?2=30.90,P<0.0001?.categories?X?2=52.99,P<0.0001?and age?X?2=4.06,P=0.0440?were statistically significant.A total of 3162 cases were tested positive,with 2226 cases of EV71 infection,accounting for 70.40%of the laboratory test cases;454 cases of CoxA16 infection,accounting for 14.36%;482 cases of other enterovirus infection,accounting for 15.24%.Kernal density of EV71 positive cases mainly concentrated in downtown area of Yichang city and Xiaoxita streets in Yiling district.The distribution of CoxA16 was wider than EV71 that all county-level administrative region tend to integrate together,mainly concentrated in downtown area of Yichang city and Gufu town,Xingshan County.Other intestinal virus cases were mainly concentrated in Maopin town,Zigui County and Longzhouping town,Chang Yang County.2.Individual risk factors analysis:living in the lower layer?1?3?was the protective factor?OR=0.698?.Resident place with a public toilet around was a risk factor?O=1.294?.Get vaccine information through traditional media?OR=2.785?,hand,foot and mouth disease knowledge awareness of the caregiver who scored of 0 points was a risk factor for not been infected?OR=2.535?,artificial feeding?OR=1.837?,the kids who was in good health?OR=3.573?were risk factors for not been infected.The kids who participated after-school classes,can defecated independently,had a family history,had a history of allergies,share drinking glasses,bite fingers,has been to long distance bus or the train station and supermarkets or the shopping mall before ten days were the protective factors of the one who inclined not to be infected,their odds ratios were 0.648,0.473,0.315,0.568,0.670,0.566,0.492 and 0.802,respectively.3.Risk factor of groups:the ratio between teacher and the children that higher than 1:5,the ratio range from 1:5 to 1:7,kindergarten which with a school infirmary and a school bus were the protective factors of low incidence rate areas,the odds ratios were 0.079,0.193,0.175 and 0.101.4.Prediction and simulation:The optimal model been decided on goodness of fit?BIC=3.022?and Ljung-Box statistics?P=0.334?was ARIMA?1,0,5?X?1,1,0?12.The results predicted by ARIMA model contains certain differences between the actual values,but corresponded to the basic trend.The optimal model with hidden layer of neural number 15 and time delay number 5 was been adopted by MSE and correlation coefficient?R=0.89251?.The hybrid model of ARIMA and NARNN also showed an overall downward trend of hand foot and mouth disease.The results of markov chain model showed that monthly incidence rates of year 2017 could be staying low.Uncertainty analysis showed that the proportion of outbreak situation sustained would be 98.75%,78.75%and 45.78%while contact coefficience distributed by k???2,7?,k???0.5,10?and k???0.2,15?at ideal condition set,respectively.If we do not carry out any prevention and control measures,after about 30 unit intervals the number of infected cases would quickly climbed to the peak.By the influence of strengthening prevention and control measures,the peak would turn to delay,and the peak value declined rapidly.While at the condition considering group influence factors,the proportion would be 98.44%,80.31%and 45.31%,respectively.The measures would have tiny influence on the peak time but on the peak value.The more stringent measures make the smaller the peak value.Conclusions1.Obviously,central tendency on geographical positions and periodicity of the onsets of hand foot and mouth disease in Yichang city were remarkable.2.Cases of hand foot and mouth disease were been profoundly influenced by large amounts of complicated factors,such as individual differences,environmental factors,behavior factors and cognitive factors which constitute the cause set.3.Risk factors of group of hand,foot and mouth disease might be influenced by key nodes such as teachers,school infirmary,school bus and public toilets and so on.4.Overall,there was an overall downward trend of hand foot and mouth disease in 2017 in Yichang city,the incidence rate would be stay in a low level.Outbreak simulation showed that group factor and the opportunity of disease prevention and control had larger influence on epidemic trend,but contact could not explain all the transmission.ARIMA model could be used while data quality is high.Hybrid model would be better when data quality is unclear.Markov chain model could be used for short-term trend prediction of acute infectious diseases when interval partition is correct.GSUA combined with dynamic propagation would help understanding and grasping epidemic characteristics,key nodes of control and epidemic trend.provided useful information for prevention and control.Innovation1.A better understanding of epidemiological characteristics and central tendency of HFMD by using GIS method?2.A combination investigations of risk factors of groups and individuals were performed to compensate the shortage.3.A comprehensive and thorough discuss of the occurrence and development trend of HFMD was carried out by prediction and simulation from the macro and micro perspective,respectively.Limitation1.Futher follow-up study would better explained the results.2.It should be more cautious for us to realize the differences between social,economical and cultural about the results extropolation.3.Laboratory test on serotype and genotypic subgroup would provided More scientific evidences for selecting control strategies. |
PDF Full Text Request