Modeling of dengue incidence in Paraiba State, Brazil, using distributed lag models

The literature provides several statistical models to explain dengue incidence. However, there is disagreement as to the real contribution of models based on climate factors or models based on relative variables to combat the vector, because their variability fails to provide a satisfactory statistical explanation for the behavior of dengue incidence. Distributed lag models (DLM) posit that a variable response Y will be explained by the presence of a variable X at the same moment in time t and in the previous moments (t-1,...). The current study presents the use of DLM in dengue modeling. Among the several models tested, two displayed apparently interesting results. A DLM using rainfall was not validated statistically. Another model using the number of counties (municipalities) with dengue presented valid and satisfactory results. This DLM was also feasible from the perspective of State Health Departments, allowing the development of a model with interesting statistical results and good accuracy, using a single data source.