Salmonella Brandenburg in sheep meat in New Zealand : preliminary studies to support a risk assessment approach : a thesis presented in partial fulfillment of the requirements for the degree of Masters of Veterinary Sciences in Veterinary Public Health at Massey University, Palmerston North, New Zealand

Abstract

Abortion and death of ewes caused by a particular strain of Salmonella Brandenburg is an animal disease problem that is unique to the South Island of New Zealand. Like most Salmonella serovars, this organism is zoonotic and has caused cases in occupationally exposed people. As Salmonella are primarily recognised as agents of foodborne disease, the potential for foodborne transmission must be acknowledged, although human cases attributed to consumption of sheep meat have not yet been reported. Salmonella Brandenburg has an additional concern for New Zealand’s sheep meat industry owing to the possibility that contamination of sheep meat products could compromise market access. In 1995, the Sanitary Phytosanitary Agreement of the World Trade Organisation specified that scientific risk analysis was required before countries could refuse to import animal or plant materials on the basis of risks to animal, plant, or human health. This thesis presents initial microbiological studies of the prevalence and concentration of Salmonella Brandenburg on sheep meat carcasses that were conducted in conjunction with other projects designed to address the Salmonella Brandenburg issue using a modern risk assessment approach. The microbiological studies (Chapters 3 and 4) are preceded by two introductory discussions that provide the context for the project. Chapter 1 presents an overview of national and international regulatory approaches to food safety, foodborne diseases and protection of consumer health relevant to meat and meat products. A selective review of literature on Salmonella focuses on Salmonella in sheep and on aspects most relevant to food safety. Chapter 2 summarises information on published quantitative microbiological risk assessments (QRA) conducted using the guidelines developed by the Codex Alimentarius Commission to apply QRA to microbiological foodborne hazards. A conceptual framework is presented for developing a QRA for Salmonella Brandenburg in sheep meat that covers all sectors of the food supply chain from animal production to the point of consumption. Following the precedent of previous QRA efforts, the food supply chain is divided into a series of five modules: animal production; transport and lairage; slaughter and processing; retail and distribution; and consumer. For each module, key outputs (prevalence and concentration of Salmonella in animals or product at various points in the supply chain), and their likely determinants, are identified. The specific objective of the microbiological studies conducted was to estimate the prevalence and i concentration of Salmonella on sheep carcasses from animals originating from farms that had experienced Salmonella Brandenburg disease and other farms from the same region that had no history of this disease. Prior to undertaking the field studies, it was necessary to conduct some methodological studies to evaluate the effect of sample handling procedures on the results obtained with quantitative bacteriology. Chapter 3 presents three controlled laboratory experiments with swab samples taken from meat contaminated experimentally with the epidemic strain of Salmonella Brandenburg. The Most Probably Number (MPN) method was used to quantify counts of Salmonella Brandenburg per 100cm2 area of meat swabbed. In each experiment, control samples were processed immediately, and treatment samples were subjected to different periods and conditions of storage. Treatments were chosen to emulate anticipated conditions that would be required for the field studies due to logistic constraints. The three storage protocols evaluated were: Experiment 1: Storage of swabs diluted in buffered peptone water (BPW) for 48h at 40C Experiment 2: Storage of swabs diluted in BPW for 5 days at 40C Experiment 3: Storage of swabs for 24h at 40C before dilution in BPW, followed by storage for a further 48h at 40C. Differences in counts between control and treatment samples were not tested statistically, owing to the small samples sizes, but were numerically less than one log difference in all experiments. In 2 of the 3 experiments, counts for stored samples were in fact numerically greater than for samples processed immediately. These results suggested that carcass swabs contaminated with Salmonella could be stored under the specified conditions without affecting the results of quantitative bacteriology using the MPN method. Chapter 4 presents a study undertaken to obtain initial qualitative and quantitative estimates of the presence of Salmonella organisms on sheep carcasses sampled at 3 points in the processing chain (i.e. slaughter floor, cooler, and boning room). Slaughtered sheep (ewes and lambs) were sourced from six farms in the Central Otago/Southland region of the South Island where Salmonella Brandenburg disease is endemic. Three farms (case farms) were selected based on the occurrence of an outbreak of Salmonella Brandenburg ii disease during the spring of 2000. Three non-case farms from the same region were also sampled. As the disease epidemics are temporally clustered in July and August, well before lambs are sent for slaughter, sampling was replicated after an interval of approximately 2 months to assess likely temporal variation in risk of carcass contamination. For comparative purposes, samples from sheep carcasses were also collected from 6 groups of sheep slaughtered at 2 plants in the North Island where salmonellosis due to Salmonella Brandenburg infection in sheep has not been reported. A total of 1417 carcasses were sampled in the study and initially tested by BAX® test. Of these, 1214 samples were sourced from the 3 case and 3 non-case farms supplying the South Island plant. The remaining 203 carcasses were sampled at the 2 North Island plants. A total of 138 (11.3%) of the 1214 samples collected in the South Island plant tested positive for the presence of Salmonella Brandenburg. No positive findings were obtained from the samples collected in the North Island plants. The vast majority (130 or 94%) of the 138 positive samples was obtained in the first period of sampling, indicating a substantial decline in risk of carcass contamination in the period between the first and second sampling. These findings indicated that the prevalence of carcass contamination with Salmonella Brandenburg was markedly elevated in the region where sheep flocks experienced abortion outbreaks caused by the organism. Although clinical Salmonella Brandenburg enteric disease has not been reported in lambs, the first sampling revealed that overall prevalence of contamination was higher (33%) for lamb carcasses than ewe carcasses (10%) from the same farms. While the prevalence of lamb carcass contamination was comparable for both case and non-case farms, the prevalence of ewe carcass contamination was strongly clustered and only 2 samples were positive from non-case farms. Estimates of the prevalence of contamination were influenced by the location of sampling carcasses (e.g. slaughter floor, cooler), but estimates of bacterial numbers on positive carcasses were generally similar regardless of class of stock, time of sampling, or sampling location in the plant. No positive samples were obtained from swabs of primary cuts in the boning room. Collectively these findings suggest that the emergence of Salmonella Brandenburg infection of sheep in the South Island may have considerable implications for product safety and public health. A strong case can be made for more research to better characterise the potential risks and to explore potential risk mitigation strategies. While the data obtained in this study have provided valuable insights into several important aspects of the issue, due to logistic and other constraints they have considerable shortcomings with respect to the requirements of the formal QRA. These shortcomings were discussed and evaluated in terms of representativeness and suitability for quantitative risk assessment. Chapter 5 presents an extension of the conceptual framework for a QRA outlined in Chapter 2, by integrating the data obtained from the bacteriological study, as well as data from other sources. Major data gaps are identified and suggestions are presented with respect to options for ongoing research to advance understanding and management of Salmonella Brandenburg in New Zealand sheep meat. More extensive and representative surveys are required to obtain more reliable data on farm, and within-farm, prevalence of infection as well as more extensive and representative longitudinal studies of the prevalence and concentration of the organism during slaughter and processing. It is considered that more systematic surveys at the time of apparent highest risk would be a more reliable means of assessing potential exposure of consumers than predictive microbiology.