RESUMO
In response to the first human outbreak (January May 2005) of Saint Louis encephalitis (SLE) virus in Córdoba province, Argentina, we developed an environmental SLE virus risk map for the capital, i.e. Córdoba city. The aim was to provide a map capable of detecting macro-environmental factors associated with the spatial distribution of SLE cases, based on remotely sensed data and a geographical information system. Vegetation, soil brightness, humidity status, distances to water-bodies and areas covered by vegetation were assessed based on pre-outbreak images provided by the Landsat 5TM satellite. A strong inverse relationship between the number of humans infected by SLEV and distance to high-vigor vegetation was noted. A statistical non-hierarchic decision tree model was constructed, based on environmental variables representing the areas surrounding patient residences. From this point of view, 18% of the city could be classified as being at high risk for SLEV infection, while 34% carried a low risk, or none at all. Taking the whole 2005 epidemic into account, 80% of the cases came from areas classified by the model as medium-high or high risk. Almost 46% of the cases were registered in high-risk areas, while there were no cases (0%) in areas affirmed as risk free.
Assuntos
Árvores de Decisões , Encefalite de St. Louis/epidemiologia , Meio Ambiente , Sistemas de Informação Geográfica , Animais , Argentina/epidemiologia , Surtos de Doenças , Encefalite de St. Louis/transmissão , Humanos , Insetos Vetores , Medição de Risco , Estações do AnoRESUMO
Forecasting models were developed for predicting Aedes aegypti larval indices in an endemic area for dengue (cities of Tartagal and Orán, northwestern Argentina), based on the Breteau and House indices and environmental variables considered with and without time lags. Descriptive models were first developed for each city and each index by multiple linear regressions, followed by a regional model including both cities together. Finally, two forecasting regional models (FRM) were developed and evaluated. FRM2 for the Breteau index and House index fit the data significantly better than FRMI. An evaluation of these models showed a higher correlation FRM1 than for FRM2 for the Breteau index (r = 0.83 and 0.62 for 3 months; r = 0.86 and 0.67 for 45 days) and the House index (r = 0.85 and 0.79 for 3 months; r = 0.79 and 0.74 for 45 days). Early warning based on these forecasting models can assist health authorities to improve vector control.