Purpose: To expand the C. botulinum food safety predictive model to consider the interactive effect of moisture, pH, potassium-based replacements for NaCl and DSP, fat, and sorbate.
Methods: Eighty formulations were identified using a central composite design targeting seven factors (50-60% moisture, pH 5.4-6.2, 0-0.2% sorbic acid, 10-30% fat, 1.7-2.4% NaCl, 0.8-1.6% DSP, and 0-50% potassium replacement for sodium salt). Treatments were inoculated with 3-log proteolytic C. botulinum spores (10-strain mixture) per gram, hot-filled into sterile vials, and stored anaerobically at 27°C. Samples (5/interval) were assayed at 0, 1, 2, 3, 4, 8.5, 17.5, 26, 40 and 56 weeks for presence of botulinum toxin using the mouse bioassay. A parametric survival model was fit to the censored time to toxin data. The model can predict both failure probability at specified times and time to toxicity with specified failure probability.
Results: All linear, quadratic and pairwise effects were considered for model fit. As hypothesized, the effects of pH, moisture, DSP, NaCl and sorbate were significant (P<0.001). Fat level and potassium-replacement were significant at P <0.017 and 0.053. The model is conservative, consistently predicting failure for toxic samples although it does predict failure for some samples that were not toxic. Comparison with previously collected challenge study data confirms that the model predictions are valid only for combinations within the ranges tested.
Significance: This research adds the factors of salt reduction, fat, and sorbate to the model predicting the botulinum safety of process cheese products. Additional study is required to expand the model to lower moisture and higher phosphate-emulsifier concentrations.