Survival of Escherichia coli O157:H7 in Acidified Foods: A predictive Modeling Approach

Since the initial outbreak reported in 1982, there have been numerous cases of illness associated with Escherichia coli O157:H7. Epidemiological reports estimate that 20,000 cases of E. coli related illness are seen each year. A study of the O157:H7 outbreaks in the United States from 1982-2002 revealed that food-borne related illness accounts for roughly 50% of the reported cases [Emerging Infectious Diseases 11:4 2005 pp. 603-609]; thus the survival of E. coli O157:H7 in food systems is of great concern.;Predictive microbiology aims to create and evaluate appropriate mathematical models for describing microbial growth and survival in food systems. In this dissertation, mathematical models are developed and assessed for their ability to characterize the survival of E. coli O157:H7 in acidic environments similar to those exhibited in acidified food products. The classic Weibull model was used as a primary model to characterize the linear and nonlinear killing kinetics of E. coli exposed to acid and salt. Analysis of the nonlinear survival and the intracellular physiology revealed a significant correlation between the killing kinetics and the intracellular pH. A secondary linear model was additionally able to describe the time needed to achieve a 5-log reduction in viable cell counts; however this secondary model underestimated the times to achieve a 5-log reduction in validation experiments.;A mechanistic model of E. coli acid inhibition based on chemical composition of the intracellular matrix was derived. This model predicts the change in intracellular pH when cells are subjected to increasing concentrations of acetic acid. The survival of E. coli in solutions of different salts was also studied and an in-depth model comparison is presented here. Finally, a modeling consideration of the use of a nonlinear, autonomous differential equation or a linear, non-autonomous one for Gompertzian growth and inactivation is examined. While for a single dataset with a fixed initial condition yields an indistinguishable curve, the models can be distinguished if a second initial condition is used in a follow up experiment. Tumor growth and E. coli survival datasets were used to illustrate these issues. The modeling and experimental approaches presented in this dissertation reveal possible methods for controlling the survival of E. coli in low-pH (pH < 4.6) environments.