The Spatiotemporal Epidemiology of Influenza in Thailand
This study is based on the assumption that seasonal influenza in tropical region correlates with climatic, geographical, and socio-economic factors. There are very few studies addressing seasonality and geographical distribution of influenza epidemic in Thailand, not to mention, study on association between influenza incidence and its contributing factors. The main objectives of this study are to 1) identify seasonality, spatial dependency, and spatiotemporal correlation of seasonal influenza in Thailand, 2) identify climatic, geographical, and socio-economic factors associated with influenza incidence in Thailand, and 3) develop comprehensive prediction models of influenza incidence in Thailand. The results of the study suggest that there is obvious seasonal pattern of influenza incidence in Thailand. The peak of the epidemic occurs in rainy season (June-July). The bottom of the epidemic happens in the middle of summer (April) and in winter (December). ArcGIS, geographical information system software, was used in this study to identify spatial dependency. The epidemic displays spatial dependency as there were 127 districts out of 928 districts having significantly higher incidence of influenza than neighboring districts, while 21 districts showing significantly lower incidence than neighboring districts. Rainfall, temperature, humidity, altitude, population, and household incomes showed significant correlations with influenza incidence in Thailand. Population is the most important contributing factor. However, correlations of all factors are weak to very weak. A GWR model was developed to predict influenza incidence. The variables included in the model are accumulated rainfall, and district population. The correlation coefficient of the model was 0.369. And a series of GWR models for each month of the year were also developed. The correlation coefficients of the models suggested that the monthly GWR models predict the incidence better than the general GWR model only in the months with high incidence of influenza (June-August). According to the finding of this study, we know that when and where influenza vaccination should be provide in order to control annual spread of seasonal influenza in Thailand. Although, the GWR model developed in this study is not a good prediction model, but it provides common ground for further study on seasonal influenza in Thailand and the region.