Submitted to: Journal of Food Protection
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/1/2006
Publication Date: 12/12/2006
Citation: Oscar, T.P. Development, evaluation and verification of a tertiary model for growth of salmonella typhimurium on cooked chicken breast burger portions in the absence of microbial competition. Journal of Food Protection. 68(12): 2606-2613. Interpretive Summary: Incorporation of complex models that predict the growth of human disease-causing bacteria in food into user-friendly computer software applications is important for the routine use of the models in the food industry to help assess food safety. The current study was undertaken to develop a user-friendly computer model for predicting the growth of Salmonella on chicken breast meat that is mishandled after cooking. Predictions of the model were successfully verified against the data used to develop the model. A new method for evaluating the predictions of models, the acceptable prediction zone method, was developed. The new method of performance evaluation overcomes limitations of current methods and provides users of the model with greater confidence that the predictions are reliable. However, the current model was developed in the absence of other microorganisms and thus, likely provides conservative predictions of the risk of Salmonella infection from cooked chicken that has been improperly handled. Research is needed to develop models that consider the important effect of microbial competition on the growth of Salmonella so that better predictions can be made that prevent the destruction of safe food and ensure the disposal of unsafe food.
Technical Abstract: Incorporation of predictive models for growth of food pathogens into user-friendly computer software applications (tertiary modeling) is important for the routine use of tertiary models in the food industry to help assess food safety. The objectives of the current study were to develop a tertiary model for growth of Salmonella Typhimurium, to develop a method for quantifying model performance and to evaluate the ability of the tertiary model to predict the data used to develop it (verification). Kinetic data for high-density (4.8 log CFU/g) growth of a single strain of Salmonella Typhimurium (ATCC 14028) on cooked ground chicken breast burger portions (one-g) incubated at 8 to 47 degrees C in the absence of microbial competition were fit to a logistic with delay primary model to determine lag time (LT), maximum specific growth rate (SGR) and maximum population density (MPD). Secondary models for LT, SGR and MPD as a function of temperature were combined in a computer spreadsheet with the primary model to create a tertiary model that predicted the growth of S. Typhimurium as a function of time and temperature. Ability of the tertiary model to reverse the modeling process and predict the data used to develop it was evaluated using an acceptable prediction zone method where the log cycle difference (Delta) between observed density and density predicted by the tertiary model was determined for 433 prediction cases obtained from 30 growth curves. The proportion of Delta (pDelta) in an acceptable prediction zone from a Delta of -0.5 (fail-safe) to 0.25 (fail-dangerous) log CFU/g was used to quantify model performance. Models with pDelta < 0.70 were considered to provide predictions with acceptable accuracy and bias. An acceptable pDelta of 0.79 was obtained for the tertiary model predictions, whereas an acceptable pDelta of 0.95 was obtained for the primary model predictions of the same data. Although combining the primary and secondary models to create the tertiary model resulted in a loss of model performance, the tertiary model provided predictions with acceptable accuracy and bias and thus, was successfully verified.