Predictive model development and lag phase characterisation for applications in the meat industry

Foods of bovine origin have been implicated as the principle vehicle in disease outbreaks of pathogenic Escherichia coli, hence the increasing interest in understanding the behaviour of this pathogen on carcasses during processing and handling. In the absence of rapid and noninvasive methods to determine the microbiological safety of meat carcasses, predictive modelling has been suggested as a strategy to estimate the consequences of processing and handling procedures on the fate and numbers of microorganisms. This thesis reports two contributions in realising the potential of predictive modelling for use in the food industry: development and extensive validation of a growth model for E. coli and advances in characterisation of bacterial lag phase duration. A modified square-root model for predicting E. coli growth was developed and its reliability assessed against a variety of data. The bias and accuracy factors formed the basis of the model evaluations and provided an objective summary of model performance. The model performed well when compared to data generated in liquid growth media, ground beef and for data collated from the literature. The model performed as well as, or better than other published models for E. co/i growth. In particular, the model predicted growth in meat and meat products better than other models. Evaluation of predictive models in industry is considered the most rigorous test of model performance. The deliberate introduction of potential pathogens in abattoirs is precluded in Australia, thus the E. coli growth model could not be validated by trials on carcasses during normal commercial processing. Instead, a predictive growth model for a surrogate organism, Klebsiella oxytoca, was developed and evaluated against data for the growth of K. oxytoca on carcasses during normal chilling operations. Those studies suggest that predictive modelling can be used to predict the average changes in numbers of bacteria on a carcass resulting from temperature and water activity changes caused by air chilling processes. Lag times have long been considered an uncontrollable variable in food microbiology. Studies were undertaken to describe the effects of environment and physiological history of the cell on lag times. Abrupt temperature, pH and osmotic shifts of cultures were found to induce lag phases in a variety of foodborne bacteria, highlighting the prospect of inducing lags by manipulating the rate and extent of change of environmental conditions. Variability in bacterial lag times was reduced by using the concept of relative lag times or "generation time equivalents", i.e. the ratio of lag time to generation time (RLT). The physiological history of the cell, including growth phase and habituation, affect the magnitude of the RLT response. In general, environmental downshifts induce larger RLTs than equivalent upshifts. These observations support the hypothesis that lag time can be understood in terms of the amount of work to be done to adjust to new environmental conditions and the rate at which that work is done. The results in this thesis demonstrate that careful interpretation of RLT responses under very stressful environmental conditions is required due to potential changes in growth curve shapes. Additionally a normal physiological range for water activity is proposed. Characterisation of bacterial lag times using RLT simplifies their inclusion in growth predictions, thus increasing the utility of predictive models. Results in this thesis support those of Ross (1999) who observed a common pattern of distribution of relative lag times for a wide range of species across a wide range of conditions in the range of 4 to 6 generation time equivalents