Abstract
Dengue fever, and especially the life-threatening form - DHF is an infectious mosquitoborne disease that places a heavy burden on public health systems in Malaysia as well as on most of the tropical countries around the world. Various environmental factors such as rainfall, temperature, living conditions, demography structure domestic waste management and population distribution are important in determining the mosquito survival and reproduction. A geostatistical modelling, analysis and mapping approach has been utilised in this research to understand the correlation between dengue fever prevalence, population distribution and meteorological factor, and the characteristics of space-time clusters in the Johor State. By supplementing GIS with geostatistical analysis and space-time permutation scan statistic tools, the spatial variation of dengue incidence can be mapped. Geographical weighted regression (GWR) analysis has revealed a strong (R2= 0.87) positive spatial association between dengue fever prevalence and population distribution in the Johor State. The dengue prevalence is expected to be higher in densely populated urban area, such as in Johor Bahru: however, there is a “rule” change in the Johor Bahru sub-district due to the positive impact from a dengue control and prevention programme. GWR analysis has also identified that ten to 14 days of accumulative rainfall is sufficient to support the mosquito breeding cycle and the dengue virus incubation period (vector + host) in the Johor Bahru district is 15 days. Space-time clusters showed that dengue transmission is a contagious type as the spacetime extent is limited at 200m and 20 days and mainly involved household transmission. Results from this study reveal the ability of an augmented GIS surveillance system by incorporating the disease epidemiology and a geostatistical approach to provide reliable information for infectious disease management, control and surveillance. This research is the first study that has utilised GWR in infectious vector-borne disease, especially the attempt to “spatialise” the time in Hypothesis 2. In addition, it is also the first study which makes use of spatial-scan statistic permutation model to study the characteristics of dengue fever space-time clusters.