Regresar

Spatio-temporal hierarchical models for mapping relative risks of dengue in the Municipality of Girardot, Aragua State, Venezuela

Abstract:

Hierarchical Bayesian space-time models have been used in the mapping of disease, studies of environmental pollution and industrial pollution, among many others. Under this methodology, the data is associated with point in a locality E and an instant in time t. The aim of this work is to model the relative risk of dengue in Girardot Municipality, Aragua State, Venezuela, during the epidemic period 2009. In that sense, we propose three models. First, a binomial model that measures the variability in the count of occurrence of the disease in the parishes of the municipality. A second model includes the binomial model as a first hierarchical level, plus a second level which introduces the spatial effect, the temporal effect and space-time interaction. Finally, a third spatial model that follows a Poisson model at the first level of hierarchy for the number of cases, and in the second level of hierarchy relates the relative risk associated with covariates through the logarithm function over a random effect. Data were collected for weeks and classified according to the parishes of the municipality. The Deviance Information Criterion (DIC) was used to select the best model. The Poisson model was best suited to represent the relative risk of contracting dengue in the area under study, showing that high-risk patterns were found in the parishes located in the south and southwest of the Girardot municipality, some of them bordering the lake of Valencia.