Using Multinomial and Space-Time Permutation Models to Understand the Epidemiology of Infectious Bronchitis in California Between 2008 and 2012.

Although infectious bronchitis virus (IBV) has been described as one of the most economically important viral respiratory diseases in poultry, there are few analyses of outbreaks that use spatial statistics. In order to better understand how the different genotypes of IBV behave spatially and temporally, we used geographic information system-based mapping coupled with spatial and spatial-temporal statistics to identify statistically significant clustering of multiple strains of infectious bronchitis (IB) between 2008 and 2012 in California. Specifically, space-time permutation and multinomial models were used to identify spatial and spatial-temporal clusters of various genotypes of IBV. Using time permutations (i.e., windows) spanning days to years, we identified three statistically significant ( P < 0.05) clusters. In contrast, multinomial models identified two statistically significant spatial-temporal clusters and one statistically significant spatial cluster. When comparing the space-time permutation and multinomial models against each other, we identified spatial and temporal overlap in two of the three statistically significant clusters. From a practical perspective, multinomial clustering approaches may be advantageous for studying IB because the model allows the different genotypes of IB to be independent nominal variables, thereby allowing for a more detailed spatial analysis. To that point, based on their risk ratios, the genotypes classified as vaccine-related were identified as the most significant contributor to two of the three mutinomial clusters. Additionally, statistically significant clusters were mapped and layered on a hot-spot analysis of commercial poultry farm density in order to qualitatively assess the relationship between farm density and clusters of IBV. Results showed that one of the three space-time permutations and one of the three multinomial clusters were spatially centered near the highest density farm areas, as determined by the hot-spot analysis.