Relative risk estimation of dengue disease in Bandung, Indonesia, using poisson-gamma and bym models considering the severity level

Abstract

Recently, dengue as one of the most dangerous diseases in the world has attracted more attention due to its soaring infection cases. One method to estimate the relative risks of dengue transmission commonly used is through the statistics approach. Dengue cases of all severity levels spread rapidly in every district in Bandung, Indonesia every month. There are two different severity levels of dengue disease: the early-stage known as Dengue Fever (DF) and the severe-stage manifested as Dengue Hemorrhagic Fever (DHF) and Dengue Shock Syndrome (DSS). This research investigates the early stage, the severe stage, and the combination of both stages. The non-spatial Poisson-gamma model and spatial Besag, York, and Mollie (BYM) model are applied to estimate the relative risks in each district in Bandung every month. These two models are chosen to analyze whether there is a spatial effect in dengue transmission in particular critical area. This research will use 2013’s data from St. Borromeus hospital, one of the reputable hospitals in Bandung. The results show that the implementation of non-spatial Poisson-gamma and spatial BYM models does not depict a significant difference in the result of the relative risk estimation of dengue transmission in Bandung. The Deviance Information Criterion (DIC) diagnostic indicates that non-spatial model is better than the spatial model. Therefore, it can be concluded that there is no spatial effect in dengue transmission in Bandung. It means that dengue transmission in Bandung is not affected by neighboring areas. This analysis is also applicable to every stage estimated, both for the early-stage as well as the severe-stage.