Abstract
It is important to have appropriate models for the surveillance and control of mosquito-borne diseases, such as dengue fever (OF). These models need to be based on appropriate temporal and spatial scales. The aim of this study was to illustrate the impact of different temporal and spatial scales on DF control decisions. We applied the Getis-Ord Gi* statistic at different temporal and spatial scales to examine the local level of spatial clusters at these scales in order to identify and visualize areas where numbers of adult female Aedes mosquitoes were extreme and geographically homogenous. The modeled hotspot areas were different, depending on whether they were modeled on weekly, monthly or yearly aggregated data. A similar result was found when using different spatial scales for modeling, with different scales giving different hotspot regions. For 2006, the highest risk areas (18 districts) were mostly identified in the central districts with a high rate of similarity (95%) compared to the highest risk areas (19) identified in the averaged five-year period model. Knowledge of appropriate temporal and spatial scales can provide an opportunity to specify the health burden of OF and its vector within the hotspots, as well as set a platform that can help to pursue further investigations into associated factors responsible for increased disease risk based on different temporal and spatial scales. (C) 2012 Elsevier B.V. All rights reserved.