2.6 Epidemiological relevance of the network
To analyse the causality of the direct pathway transmission, we used the k-test, comparing the observed connectivity between cases with the expected distribution of the k-statistic, which is the average number of observed cases, occurring within one step of an infected case in the network (k=1) (VanderWaal et al., 2016) . We assessed whether it was possible that the pattern of observed disease outbreaks (cases) could be the result of network transmission, thus evaluating itsepidemiological relevance . We used a network of premises specifying the status of their nodes (not infected=0, infected=1) for every year and the cumulative period. We then compared the observed k-statistic with its permuted distribution, where case locations are randomly reassigned within the network through 1,000 iterations.
Cleaning of the datasets was performed to homogenize premises classification, and remove inconsistencies, also match outbreaks with movements between databases. We computed all analyses using R V.3.6.3 (https://cran.r-project.org/) and software gephi v0.9.2 (https://gephi.org/).