Recontamination in food processing : quantitative modelling for risk assessment

Every year at least 1.5 to 6% of the Dutch population suffers from foodborne illnesses. This may result in symptoms like vomiting or diarrhoea but can in some cases also lead to death. Processes like pasteurisation or sterilisation reduce the number of pathogenic bacteria in food products. Food safety is further controlled by implementing systems like Good Manufacturing Practices (GMP) and Hazard Analysis Critical Control Points (HACCP). Although these preventive measures do improve food safety, the presence of pathogenic bacteria can never be completely eliminated. Therefore, there is always a probability of becoming ill after consuming a food product. This probability can be estimated using Microbiological Risk Asessment (MRA).For one element of MRA, the exposure assessment, the food chain is evaluated from the raw materials to the final product to estimate the level of pathogens in a food product at the moment of consumption. Predictive modelling techniques are combined with process engineering models to evaluate the effect of food processing steps on microbial survival and food safety. The same approach can be used to evaluate processes for other factors than microbiological hazards as was demonstrated in the evaluation of food processing on the accumulation of salts in a potato product. This approach proved to be useful to obtain quantitative information of the process and find critical points in the system.In MRA, usually, only growth or inactivation in the different process steps is considered and re-introduction of bacteria during the production process is not taken into account. However, when pathogens re-enter the product after an inactivation step, this can cause foodborne illness. It is therefore relevant to quantify this recontamination so that it can be incorporated in MRA. A literature review, however, revealed that not many available recontamination models are directly applicable to the food industry.The objective of this thesis was therefore to quantify recontamination using predictive modelling techniques so that it can be incorporated in MRA. Two possible recontamination routes were investigated: recontamination via formation of biofilms (layers of bacteria and their extracellular products attached to a surface) and recontamination via aerial transfer (e.g. through the existence of aerosols).Recontamination via biofilms was modelled using mass balances and growth kinetics. Parameters for the biofilm model were obtained in biofilm experiments in which both biofilm formation and the release of cells into the flowing liquid were measured in time for up to 9 days. Staphylococcus aureus was chosen as a model pathogen and silicon tubing was used as testing material. The experiments were performed in duplicate for different flow conditions (Reynolds = 3.2, 32 and 170). It was shown that at higher Reynolds numbers, the biofilm developed faster and the desorption rate increased accordingly.As a case study, the developed biofilm model was applied in an example process system: the production of an acid-based spread. Different scenarios were evaluated, which demonstrated that the developed biofilm model indeed can be incorporated in an exposure assessment to determine the level of pathogens at the moment of consumption and to investigate whether biofilm formation is important or not. It was shown that when biofilms have been formed prior to a heating step, high levels of S. aureus can be reached locally, which can cause production of enterotoxins. These toxins are not inactivated in the heating step and can thus cause foodborne illness. Therefore, regular cleaning and a good hygienic design are a prerequisite to prevent foodborne diseases due to such biofilm formation.The second route that was quantified was recontamination via the air. Data on the number of airborne micro-organisms were collected from literature and industries. The settling velocities of different micro-organisms were calculated for different products by combining the data on aerial concentrations with sedimentation counts. Statistical analyses were performed to clarify the effect of different products and seasons on the number of airborne micro-organisms and the settling velocity. For both bacteria and moulds three significantly different product categories with regard to the level of airborne organisms were identified. The statistical distribution in these categories was described by a lognormal distribution. The settling velocity did not depend on the product, the season of sampling, or the type of micro-organism and could be described by a lognormal distribution as well. The probability of recontamination via the air was estimated using the number of bacteria in the air, the settling velocity, and the exposed area and time of the product.The effect of aerial recontamination was investigated in the production of smoked salmon as a case study. Using worst-case assumptions, it was shown that once salmon is contaminated with Listeria monocytogenes via the air, this pathogen can grow out to high numbers in the product. Therefore, either the frequency of recontamination or possibilities for growth should be reduced to improve safety of the product. Based on this exposure assessment, it was shown that smoked salmon should be stored for less than 2 weeks at 5 °C to control illnesses due to aerial recontamination of the food.Overall, the work described in this thesis contributes to the reduction in the lack of recontamination models by quantifying biofilm formation and air recontamination. The application of the models in two example production processes showed that they can be incorporated in MRA to determine its importance and estimate the probability of becoming ill after consumption of the food item. However, other recontamination routes should be quantified as well and more data are necessary to validate and improve the models. Reliable recontamination models can be used together with growth and inactivation models to obtain a more reliable outcome of a MRA