Bayesian space–time analysis of Echinococcus multilocularis-infections in foxes

A total of 26,220 foxes that were hunted or found dead in Thuringia, Germany, between 1990 and 2009 were examined for infection with Echinococcus multilocularis, the causative agent of human alveolar echinococcosis, and 6853 animals were found infected. The available data on the foxes including the location (local community; district) and the date of hunting/death were analyzed using a hierarchical Bayesian space–time model. The distribution of the model parameters and their variability was estimated on the basis of the sample size, the number of cases per spatial unit and time interval, and an adjacency matrix of the municipalities using a Markov Chain Monte Carlo simulation technique to assess the spatial and temporal changes in the distribution of the parasite. The model used to evaluate the data is widely applicable and can be applied to analyse data sets with gaps and variable sample sizes per spatial and temporal unit. In the study area, the prevalence of E. multilocularis increased from 11.9% (95% confidence interval 9.9–14.0%) in 1990 to 42.0% (39.1–44.1%) in 2005. While the infection was present in foxes only in the north-western parts of Thuringia in 1990, it had spread over the entire state by 2004. These results demand increased vigilance for human alveolar echinococcosis in Thuringia.