Combination of spatial clustering methods using weighted average voting for spatial epidemiology

Abstract

The methods of spatial clustering analyze the phenomenon under study, identifying the significant and not significant clusters, which when used individually do not exactly reflect the reality of the phenomenon studied. However, with the combination of the methods it becomes possible to obtain better results. The objective of this work was to perform a combination of methods of spatial clustering, by using weighted average voting rule, for identification of municipalities in the state of Paraiba more vulnerable to the dengue fever. For methodology application, dengue fever cases in the state of Paraiba-Brazil in the year of 2011 were used. The spatial Scan statistic, Getis-Ord, Besag-Newell methods combined by the weighted average voting rule were used in order to obtain a final map with the classification of each municipality according to “priority municipalities”, “transition municipalities” (which can become priority or not) and “non-priority”. This method allowed the visualization of the spatial distribution of the dengue fever in all municipalities of Paraiba, allowing to identify vulnerable municipalities to the dengue fever. The levels of priority can help managers for decisions concerning the specific characteristics of each municipality.