Non-typhoidal Salmonella are major human foodborne pathogens causing gastroenteritis. The consumption of pork products is considered as a risk factor for human salmonellosis. Numerous researchers are interested in the epidemiology and ecology of italic>Salmonella along the pork production chain because of the recognition of its importance for developing control strategies at pre- and post-harvest stages. However, the application of various types of epidemiologic data to establish useful approaches for Salmonella interventions still has a great potential for development. In this dissertation, our long-term goal is to explore some methodological approaches for addressing the pre- and post-harvest Salmonella interventions. This dissertation is organized as follows. The first chapter is a brief discussion of the current knowledge about the Salmonella epidemiology from pigs to pork products and potential approaches for determining pre- and post-harvest Salmonella control points. The second and third chapters elucidate the epidemiology of Salmonella in pork by evaluating the predictive ability of sub-iliac lymph nodes and the tracing ability of amplified fragment length polymorphism (AFLP) to determine the farm source of the Salmonella contamination at abattoir. Chapter 4 illustrates the application of a quantitative risk assessment approach to assess the impact of deep tissue lymph nodes on Salmonella contamination in ground pork.

In Chapter 2, we tested the hypothesis that sub-iliac lymph nodes could be used as a marker for farms with a high prevalence of Salmonella. Salmonella were recovered from 3.4% of 1,490 farm fecal samples and only 0.06% of 1,739 sub-iliac lymph nodes, which were from matched pigs and carcasses. Of 21 farms, 71.4% had at least one positive fecal sample, and 4.2% (of 24 farms) had at least one positive sub-iliac lymph node. The low detection rate of Salmonella in sub-iliac lymph nodes (0.06%) limits their usefulness as dependable pre-harvest predictor of Salmonella contamination originating on farm (3.4%).

In Chapter 3, we aimed to evaluate the association of molecular methods of Salmonella typing with units of swine production. To assess interventions utilizing molecular tools, knowledge of the production unit that a molecular tool is capable of differentiating is needed. The approach was to use analysis of molecular variance (AMOVA) based on amplified fragment length polymorphism (AFLP) data to describe genetic variance between Salmonella isolates three hierarchal production units, i.e., farms, harvest cohorts within farms and pigs within harvest cohorts. The results of AMOVA found that 27.2% of overall genetic variation was due to variance among farms (P-value = 0.003), which indicated that AFLP was able to differentiate Salmonella isolates of swine source epidemiologically between the farms. A permutation test using AFLP data provided evidence that on-farm and at-abattoir Salmonella from pigs of the same farms were more related than from different farms (P-value = 0.04), which suggested AFLP was able to link the abattoir contamination to the farm origin.

Chapter 4 employed a quantitative risk assessment model to evaluate the contribution of deep tissue lymph nodes (DTLNs) to ground pork Salmonella contamination in the United States. A scenario analysis was conducted to compare the differences in the probability of Salmonella contaminated ground pork when three input parameters related to Salmonella contamination in DTLNs were modified ( 1) probability of a swine carcass with Salmonella positive DTLNs, 2) Salmonella concentration in DTLNs, and 3) weight of DTLNs from a single carcass contributing to ground pork). The scenario analysis showed when the input parameter, probability of swine carcasses with Salmonella positive DTLNs, was changed from baseline value to half of the value and then zero, the estimated mean probability of Salmonella contaminated ground pork changed from 8.3% to 8.1% and 7.8%, respectively. Similar minor changes were observed when the other two input parameters were changed. In contrast, the estimated probability of Salmonella contaminated ground pork significantly increased when the parameter related to the amount of contaminated surface was increased. However, the impact of DTLNs was very stable through different levels of surface area contamination. These findings indicated that in our proposed model of how lymph nodes were incorporated into ground pork production, the presence of Salmonella in DTLNs didn't have an impact on Salmonella contamination in ground pork compared to other intervention strategies such as mitigation of Salmonella on carcass surface. The intervention of DTLNs at processing plants might not be able to effectively reduce the Salmonella contamination in ground pork.