Analysis On Epidemic Characters Of Influenza In Chongqing From 2004 To 2006 And Applications Of Arima Model On Predictive Incidence Of Infectious Disease

Objective: (1) To explore the epidemic characters of influenza in ChongQing and provide scientific basis for the prevention of influenza. (2) To develop the prevention policy of infectious diseases scientifically , the time series method ( ARIMA model) is used to analyze and forecast the dynamic trend of influenza and diarrhea .Method: (1) Collected and analyzed the data of epidemiology and pathogen of influenza-like illness(ILI) and outbreaks of influenza in ChongQing from 2004 to 2006. (2) Sample of incidence rates of influenza and diarrhea are taken out from 2002 to 2006 , SPSS is used to analyze and forecast,Q statistic is used to verify the applicability of the model.Results: (1) The peaks of ILI% were Apr and Oct of 2004 , Apr,Aug and Nov of 2005, Apr and Jul of 2006 . The peaks of the isolation rate of influenza virus were appeared a little later than the peaks of ILI%. In 2005 and 2006 , the peak of ILI% which showed in summer were more high than that in spring , but it did not isolate more virus in summer . (2) Influenza has showed clearly seasonal character in ChongQing , the type of strains changed in Aug of 2004,Feb and Sep of 2005,Feb and Jun of 2006, the peaks of ILI% followed one or two month later . (3) The outbreak of ILI or influenza occurred during all the year, which increased in Spring and Autumn; most of the outbreak occurred at Primary and Middle schools, especially in ruler district; the mainly type of outbreak was A, H1N1 and H3N2 were alternately appeared. (4) The ACF chart of influenza in ChongQing did not showed its seasonal character, ARIMA model only can be used to predict influenza in 3 months, it has no practicality sense. The ACF chart of diarrhea in ChongQing showed its fixed seasonal character, the multiple seasonal ARIMA model can be used to forecast for diarrhea incidence with high prediction precision. With 48 data , the R2 of the model is 0.9733 in 6 months and is 0.8574 in 12 months, 36 and 42 data also can be used for building ARIMA model ,the R2 of them were lower than that of 48 data.Conclusion: (1) Two peaks of influenza were shown in ChongQing, which were in Spring and Autumn, so it was suggested that prevention measures should be enhanced in Spring and Autumn and the time of get influenza vaccine were in Feb,Mar or in Sep,Oct annually. The diagnoses of respiratory disease should be strengthened in summer to avoided misdiagnosing. The change of the type of strains around Feb and Aug could need paying more attention, It should try to isolate virus in order to found the changes of the types of strains around Feb and Aug. The prevention of outbreaks of influenza should be emphasized on Primary and Middle schools, when the mainly type changed, we must guard against the occurrence of outbreaks. (2) ARIMA model could be used to analyze the seasonal characters of diseases, the effect of predict for disease incidence were different, it showed better for those disease which has fixed incidence.The data for build ARIMA model can be less than 50, but with more data, the effect was better. The effect of predict would debased with time extended, so it suggested that new data should be entered for building model constantly.