Randomized trial compares Bayesian models to map dengue incidence in Indonesia, suggesting targeted public health interventions.
Key Points
This research aims to compare Bayesian spatial conditional autoregressive models for mapping dengue cases in Indonesia.
Modeled dengue counts at the province level for 2023 using Bayesian spatial Poisson models.
Used average annual temperature and public health workforce as predictors.
Fitted models with BYM and Leroux priors, employing Markov chain Monte Carlo techniques. Departure from spatial independence was determined using Moran's I.
Temperature showed a risk ratio of RR=0.90 (95% CrI: 0.76 to 1.07) in the BYM model, indicating lower risk.
Workforce density had a risk ratio of RR=1.05 (95% CrI: 1.03 to 1.07), suggesting higher risk associated with more health workers.
The BYM model showed a marginally better fit compared to the Leroux prior. Risk mapping indicated higher dengue burden in Kalimantan and eastern regions.