Detection of spatial, temporal and space-time Salmonella Heidelberg and Salmonella Typhimurium clusters in Ontario in 2015, and comparisons to known outbreaks.

Purpose: Salmonellosis is one of several reportable diseases in Ontario (ON). Two or more cases of the same serotype that are linked to a common exposure or related to one another in time and/or space are considered a potential outbreak. While laboratory data can help to determine the molecular relatedness of cases, results may take up to several weeks. This study aimed to assess the utility of the retrospective spatial scan statistic in detecting clusters of Salmonella Heidelberg and Salmonella Typhimurium cases using data from ON in 2015. Identified clusters were validated by laboratory data (where available) to determine whether identified clusters were likely outbreaks.

Methods: Data representing the location of each reported S. Heidelberg or S. Typhimurium case in 2015, responsible serotype and symptom onset date were exported to SaTScan for retrospective spatial, temporal, and space-time analyses using the spatial scan statistic with Bernoulli models and a space-time permutation model. Analyses were performed with and without those cases linked to known outbreaks. Laboratory subtyping data (i.e. pulsed field gel electrophoresis (PFGE) and/or phage type) and food and environmental exposure information (e.g. travel, animal contact, poultry and other food item consumption) were used to explore the relatedness of cases within identified clusters.

Results: Spatial, temporal and space-time analyses identified a known outbreak of S. Heidelberg in 2015 (n = 9 cases) and a previously unidentified cluster of S. Heidelberg cases. Most cases (94%) within a cluster detected via a space-time permutation model of S. Heidelberg cases shared an identical PFGE pattern and appeared to represent a true outbreak.

Conclusions: The spatial scan statistic, and particularly the space-time permutation model, could assist in outbreak identification before laboratory data are available, allowing for faster cluster identification and implementation of control measures.