Geographical variation in pharmacological prescription

Promoting rational drug administration in treatments is one of the most important issues in Public Health. Bayesian hierarchical models are a very useful tool for incorporating geographical information into the analysis of pharmacological prescription data. They allow the mapping of spatial components which express the trend of geographical variation. In addition, these models are able to deal with uncertainty in a sequential way through prior distributions on parameters and hyperparameters. Bayes’ theorem combines all types of information and provides the posterior distribution which is computed through Markov Chain Monte Carlo (MCMC) simulation methods.Simulated data for pharmacological prescription corresponding to people with a diagnosis of degenerative osteoarthritis have been analyzed. Specifically, the number of prescriptions and pharmaceutical costs per patient have been evaluated as well as its relationship with gender and age. Geographical variation between different administrative units is also introduced and discussed.