Modelling Survival of Salmonella at Low aw
The presence of Salmonella in shelf stable foods with low aw has been a recurring problem for the food industry for decades, resulting in numerous disease outbreaks and product recalls. Salmonella survives in these products for months or even years. Its survival at low aw is influenced by temperature and product formulation.
Research has focused on determining how formulations of foods such as peanut spreads, powdered infant formula, and chocolate influence survival of Salmonella but, because different formulations usually impose more than one extrinsic factor affecting cell viability, conclusions concerning survival cannot be readily generalized. We have insufficient quantitative information defining the impact of specific components in low-aw foods on survival of Salmonella.
The ultimate goal of the research reported here is to develop models for predicting survival of Salmonella in low-aw foods as affected by aw, temperature, and food composition. Such models could aid in the development of safer product formulations and more accurate risk assessments.
We have completed modeling two compositional factors, i.e., water mobility and salt content. The model covers aw 0.19 to 0.54 and 25 to 80 °C for water mobility, and 70 to 80 °C for salt content up to 17%. Our approach was to use a model system based on whey protein that has been modified to change one compositional factor at a time.
We found that neither salt content nor water mobility influence survival independent of the effects of aw and temperature. Inactivation rates increase as aw and temperature increase. In addition, because of a tailing effect, inactivation deviates from log-linear kinetics as temperature increases. As a result, data fit better to the nonlinear Weibull model as compared to the log-linear model.
A predictive model based on aw and temperature was validated using market-purchased low-aw foods. The model provided good predictions for the survival of Salmonella in low-fat foods such as wheat flour and powdered milk (12% discrepancy value and -3% bias), but predictions were not accurate for foods with a higher fat content such as 12% fat cocoa powder and 12% fat peanut meal (50% discrepancy value and -9% bias). When only fat-free foods were included in the validation model, 88% of the predictions were in the fail-safe range.
This model can be accessed on Combase. We are presently modifying the model to account for fat and sugar content.