ABSTRACT
Remote sensing and geographic information system [GIS] technologies were used to discriminate between 130 villages in the Nile Delta, at high and low risk for filariasis, as defined by microfilarial prevalence. Landsat Thematic Mapper [TM] data were digitally processed to generate a map of land-cover as well as spectral indices such as NDVI and moisture index. A Tasseled Cap transformation was also carried out on the TM data which produced 3 more indices [brightness, greenness and wetness]. GIS functions were used to extract information on land-cover and spectral indices within 1 km buffers around the study villages. The relationship between satellite data and prevalence was investigated using discriminant analysis. The analysis indicated that the most important landscape elements associated with prevalence were water and marginal vegetation, while wetness and moisture index were the most important indices. Discriminant functions generated for these variables were able to predict correctly 80% and 74% of high and low prevalence villages, respectively, with an overall accuracy of 77%. The present approach provided a promising tool for regional filariasis surveillance and helps direct control efforts
Subject(s)
Filariasis/transmission , Information Systems , Microfilariae/pathogenicity , Epidemiologic MethodsABSTRACT
Geographic information system [GIS] was used to analyze the spatial distribution of filariasis in the Nile Delta. The study involved 201 villages belonging to Giza, Qalyoubia, Monoufia, Gharbia, and Daqahlia governorates. Villages with similar microfilarial [mf] prevalence rates were observed to cluster within 1-2 km distance, then clustering started to decrease significantly with distance up to 5 km [Pearson correlation coefficient = -0.98]. The likelihood of negative and high prevalence villages being contiguous was very low [1.8%, n = 612 village-pairs] indicating homogeneity in disease processes within the defined spatial scales. Of the villages located within 2 km from the main Nile branches [n = 46], 95% exhibited low prevalence. In addition, the spatial pattern of mf prevalence was shown to be negatively associated with annual rainfall and relative humidity, while it was positively associated with annual daily temperature. Average mf prevalence in warmer, relatively drier areas receiving 25 mm of rain was significantly higher [3.9%] than that in less warmer but more humid areas receiving 50 mm of rain [1.6%] [P <0.0001]. Based on the results of the present study, GIS was used to generate a filariasis risk map that could be used by health authorities to efficiently direct surveillance and control efforts. This investigation identified some of the factors underlying filariasis spatial pattern, quantified clustering and demonstrated the potential of GIS application in vector-borne disease epidemiology