A novel model developed for quantitative microbial risk assessment in the pork food chain : a dissertation presented in partial fullfilment [sic] of the requirements for the degree of Doctor of Philosophy at Massey University, Institute of Veterinary, Animal and Biomedical Sciences, Massey University, Palmerston North, New Zealand

Abstract

Food-borne diseases contribute substantially to morbidity and mortality rates worldwide. The deleterious impact of these diseases on human health, concurrent with the associated socioeconomic cost has led to an increased demand for the production of safe food globally. Consequently, agencies such as the World Health Organization (WHO) and the Food and Agriculture Organization (FAO) have resolved to address this issue. In this vein, scientific, risk-based approaches which facilitate estimation of the probability of disease occurrence, the magnitude of the disease and efficacious control measures have been recommended for use internationally. Many pathogens have been implicated as aetiological agents of food-borne disease. The WHO has identified non-typhoidal Salmonella, Escherichia coli and thermophilic Campylobacter as zoonotic food-borne pathogens of greatest importance. These pathogens can be transmitted to humans through pork consumption. This thesis therefore proposes a suite of novel, mechanistic, semi-stochastic, quantitative, modular process risk models describing the propagation of these three pathogens from the live pig at the abattoir, to pork chops sold at retail. The model is developed for use in risk-based, quantitative microbial exposure assessments in New Zealand and can be employed to explore different intervention strategies targeted at mitigating contamination levels of these pathogens on pork chops. The models comprise multiple, coupled, differential and difference equations. These equations explicitly describe bacterial growth, inactivation, removal, cross-contamination and food partitioning occurring in continuous and discrete time in abattoirs and at retail. Distributions of pathogen numbers on the surface of carcasses, and prevalence levels are output by the models at different stages of abattoir processing and pork chop production. Both dressed pork carcases exiting abattoirs in New Zealand and pork chops at retail are predicted to contain low surface contamination levels of the pathogens under consideration, while a small percentage is estimated to be highly contaminated. Median contamination levels on dressed pork exiting the abattoir are predicted to be less than one cfu/cm2. Generally, there are large reductions in surface bacterial numbers for all three organisms from the time the live pig enters the abattoir, to sale of the pork chop at retail. The introduction of a second singeing procedure immediately postevisceration in the abattoir is predicted by our models, to be an effective mitigation strativ egy, with estimated reductions in median pathogen levels of 100%. This control measure is considered to be more effective than coverage of the anal region of the pig during evisceration. This latter mitigation strategy was predicted to result in 10% – 44% reduction of median pathogen contamination levels. At retail, pork chops are also estimated to contain low numbers of these pathogens. Therefore handling of the raw pork chop soon after purchase from retail outlets may be associated with a low risk of contracting salmonellosis, colibacillosis and campylobacteriosis. This risk can be further reduced by placing pork chops in a blast chiller for 12 hours prior to display. When this mitigation strategy was modelled the outputs indicated a 15% – 61% reduction in the maximum pathogen levels on pork chops, 44 – 100% reduction in the 10th – 90th range and 14% – 50% reduction in pathogen prevalence levels. Detailed investigation revealed the limitations of a specific modelling approach. We determined that the population-based modelling approach is not an appropriate alternative to the individual-based modelling approach when there is a large disparity in contamination levels between processed carcasses. Therefore the former technique should not be used in the presence of large heterogeneity with respect to the number of bacteria on the food unit of interest, or when bacterial populations input into the model are described with large variances. This thesis demonstrates the application of a suite of novel risk models in the pork food chain. We propose use in quantitative microbial exposure assessments. The applicability of these models is not only limited to the pork chain or to the above mentioned pathogens, but by modification of parameters, the entire model, or portions thereof can be extrapolated to other animal species undergoing similar abattoir procedures with pathogens of analogous epidemiological patterns. Finally the information provided by the models can be instrumental in assisting risk managers in their decision-making and policy development undertakings and provide guidance to effectively and strategically funnel limited resources.