The application of advanced statistical approaches to investigate the epidemiology and improve the surveillance of non-Typhoid salmonellosis associated with Salmonella Heidelberg and Salmonella Typhimurium in Ontario
In this thesis, I investigated the epidemiology of non-typhoid salmonellosis in Ontario in 2015, with a specific focus on Salmonella Typhimurium and S. Heidelberg. Data for each reported human case of S. Typhimurium and S. Heidelberg in Ontario in 2015 were analyzed through a series of research projects. Overall thesis objectives were to compare risk factors associated with each serotype, and to explore clusters and potential outbreaks in space, time, and space-time. Where relevant, human case data were combined with population and agricultural census data, and data for licensed meat plants (abattoirs) in Ontario. The application of various statistical models was explored to further understand the epidemiology of each of these serotypes, and to identify potential improvements to current processes for surveillance, cluster detection, and public health case and outbreak investigations.
In comparing the epidemiology of these serotypes, several key findings were identified: 1) Using a case-case study design, consumption of sprouts was found to increase the odds of infection due to S. Heidelberg, relative to S. Typhimurium. Conversely, recent travel or contact with reptiles each increased the odds of infection with S. Typhimurium, relative to S. Heidelberg.
2) Using scan statistics, clusters of S. Heidelberg and S. Typhimurium were identified in space, time, and space-time, independent of molecular laboratory data. Clusters were validated and potential outbreaks identified using molecular and risk factor data.
3) Using scan statistics and focused spatial tests, clusters of S. Heidelberg were identified around several meat plants. Risk factor and molecular laboratory data provided evidence in support of implicated meat plants as a source of exposure.
4) Using mixed regression models, rates of S. Typhimurium and S. Heidelberg were found to be associated with agricultural and socioeconomic variables such as agricultural animal density, the proportion of married individuals, and labour participation, each of which may influence animal exposure and food consumption patterns, impacting the risk of exposure to Salmonella. This thesis demonstrates how identification and consideration of serotype-specific differences for Salmonella, and the use of geospatial methods for cluster detection, can be used to optimize and inform public health surveillance and disease prevention efforts.