Quantitative microbial process model to estimate the risk of salmonellosis from consuming pork and pork products

Salmonellosis is the second largest cause of foodborne illness in Australia, with rates of salmonellosis increasing over recent time. Outbreaks of foodborne illness attributed to pork products are most often associated with Salmonella. Pork industry risk managers, therefore, commissioned this study to develop robust risk assessment models to quantify the risks of salmonellosis to Australian consumers from pork burgers and moisture-infused pork, with the aim of informing risk managers in their decisions. Pork burgers are currently being marketed in Australia as an alternate serving suggestion for pork mince, while moisture-infused pork is also now being promoted as an alternative low-fat product with increased juiciness to offset the effects of reduced intramuscular fat. These products, however, may present increased risk compared to other pork products because of the potential internalisation of pathogens. The risk assessments followed the approaches and guidelines recommended by FAO/WHO and Codex for microbiological risk assessment and used a stochastic modelling approach. An extensive review of literature relevant to these risk assessments identified many appropriate data sources, including Salmonella prevalence after boning, storage conditions during retail storage, consumer transport and domestic storage; and dose-response models for Salmonella. The lack of predictive models for thermal inactivation of Salmonella spp. in pork burger patties and the growth of Salmonella in moisture-infused pork, as well as the Salmonella concentration after boning, endpoint cooking temperatures for pork products, and the effect of the infusion process, were all identified as data gaps. Accordingly, novel data on the behaviour of Salmonella in pork products were also developed and incorporated into the modelling. Thermal inactivation of Salmonella in pork burgers was investigated by cooking pork patties made from pork mince with either 'Low' or 'Regular' levels of fat. Mince was inoculated with several strains of Salmonella and cooked to one of seven internal endpoint temperatures (46, 49, 52, 55, 58, 61, 64 °C) and Salmonella enumerated, with survival described by a generalised logistic regression model. The fat content of pork mince influenced Salmonella survival significantly (p = 0:043), with increasing fat levels resulting in increased Salmonella survival, though this effect became negligible as the internal endpoint temperature approached 62 °C. Survival of Salmonella serovars did not differ significantly (p = 0:86 and 0:10 for the intercept and slope of the fitted logistic regression curve, respectively). The growth of Salmonella in moisture-infused pork as a function of temperature was assessed by injecting intact pork loin steaks with a cocktail of Salmonella serovars. The steaks were cut into strips and incubated at nominal temperatures of 8, 15, 20, 25 or 30 °C, and Salmonella enumerated at various time intervals. Primary and secondary growth models were fitted to these observations using both one-step and two-step procedures to develop predictive growth models. The fits obtained using both procedures were compared with those obtained from ComBase data for meat products and were found to be similar, with the one-step method in closer agreement with the ComBase fits than the two-step fits. These results indicate that Salmonella growth is not significantly retarded by moisture infusion brines. Stochastic risk assessment models were constructed to estimate the changes in Salmonella prevalence and concentration, between retail and consumption, with the probability of illness per meal estimated. These models were constructed in the statistical programming language 'R' using the 'mc2d' package, with models constructed in two dimensions, allowing variability and uncertainty to be separated, quantified and assessed independently. This separation is not currently implemented in commercial stochastic modelling software packages. The mean risk of illness from the consumption of a pork burger was estimated at 1:54 x 10-8 (95% uncertainty credible interval 7:2 x 10-10, 4:96 x 10-8) and for a moisture-infused pork steak at 4:12 x 10-8 (95% uncertainty credible interval 9:85 x 10-9, 7:75 x 10-8). Sensitivity analysis was performed in both risk assessments using Spearman rank correlation, which identified the low prevalence and concentration of Salmonella on pork in Australia and the high temperatures to which Australian consumers cook pork as having the greatest influence on the low risk experienced by consumers. Hypothetical scenarios were also investigated to determine their effect on risk, including changes to mean retail temperatures and consumer pork 'doneness' preferences. The work described in this thesis has produced predictive models for Salmonella inactivation in pork burgers and growth in moisture-infused pork, providing novel information on factors that influence these processes. The risk assessments of salmonellosis from consumption of pork burgers and moisture-infused pork in Australia provide valuable information to industry on these risks, which will inform risk-based decisions, including industry risk management strategies, for pork products in Australia.