Salmonella in Swine in Portugal: Characterization of Risk Factors and Modelling Infectio

Salmonellosis is one of the major causes of food-borne disease in the world. The EU Regulation (EU Regulation No 2160/2003) requires its Member States (MS) to implement control measures in order to reduce the prevalence in food production species, including pigs. To set the reduction target each MS carried out baseline surveys to estimate the Salmonella spp. prevalence in some food production animals. In pigs, a baseline study at abattoir level (collection of lymph nodes of pigs slaughtered) and another at herd level (collection of pen faecal samples of breeding pigs) were performed. During these crosssectional studies, information regarding herd management practices and potential risk factors linked to Salmonella was collected. Data from these baseline studies in Portugal was used in this thesis for a risk factor analysis where some risk factors were highlighted to be linked with increased risk at abattoir or at farm level. The study at abattoir level identified the following risk factors: abattoir region and sample collection time. Region of the herd, size of the herd (in terms of sows), management of breeding boars, source of semen, rodents control, number of animals per pen, breeding sector room, and source of feed were identified as influential risk factors in the herd level study.;Salmonella serotypes were divided in two groups: serotype Typhimurium and S. Typhimurium-like strains with the antigenic formula: 1,4,[5],12:i:-., and other serotypes. A categorical risk factor model was implemented to assess whether the risk factors were the same between the groups of serotypes. For the breeding pigs dataset, the group "Typhimurium" was associated with the stock density (number of breeding pigs and number of pigs per pen), the characteristics of the pig (age of breeding sows), and the source of semen. On the other hand, the group "other serotypes" was associated with region of the herd, source of semen, control of rodents, breeding sector room and source of feed. The risk factors for group Typhimurium suggest a contagious pattern and the risk factors for other serotypes appear to be related to environmental factors.;Each European MS should ideally implement control programmes to reduce the prevalence of Salmonella spp.. However in practice, the control of this agent in the swine sector has proved to be difficult and expensive at farm level, so the evaluation of the efficiency of control strategies for this agent has become an important and stringent issue. With this aim in mind we developed a stochastic model which simulates the agent spread inside a farrow-to-finish herd which can be used to test control measures in terms of costbenefit. Some preparatory work was performed to estimate the transmission parameters to be used in the simulation model using data from a published longitudinal study which followed S. Typhimurium infected cohorts. Our model allowed for sensitivity and specificity of the tests used in the longitudinal study to be included, as well as for unobserved cohort effects and time-dependent effects. The simulation model tried to mimic what is happening in a herd, in terms of management practice and was linked with an infection model which simulates the infections states for each sow/pig. The parameters which most influenced the infection state of sows at farrowing/suckling were: the transmission rate from susceptible (S) to infectious (I), from I to carrier (R), and from R to S; when applied to the sow-compartment. On the other hand, the parameters which most influenced the infection state of pigs for slaughter (fattening pigs) were: the transmission rate from S to I and the transmission rate from I to R, when applied to the pig-compartment; the transmission rate from R to S applied to sows at gestation and the piglets' protective factor. Several control measures can be recommended to influence these parameters in an attempt to control the proportion of infectious animals. The simulation model potentially allows quantification of cost-benefit control measures if linked to an economic model. The simulation model is flexible enough to introduce changes in the parameter distributions or values if future research and legislation so require. At the same time the model can be adapted to different types of production (e.g. breeding units, finisher units) as it was built in a compartmental way.