Microbiological risk assessment of food : A stepwise quantitative risk assessment as a tool in the production of microbiologically safe food

Food safety is a prerequisite for food products, since consumers trust on buying safe foods. Food producers cannot gain direct profits from controlling food safety, instead they have much to lose if their products turn out to be unsafe.Food safety management systems, such as the Hazard Analysis Critical Control Point (HACCP) system, have gained much interest in the past years. Momentary, food safety is often managed for a large part on qualitative grounds. A quantitative approach of food safety management is useful by objective analyses, and can be attained by quantitative risk assessment (QRA). This thesis describes a method for stepwise microbial QRA for food products in general, and various steps of QRA.The first step in QRA is hazard identification, which is qualitative. An identification procedure for foodborne microbial hazards has been developed, and implemented as an expert system. The procedure is product oriented; it selects microbial hazards related to a specific food product. The hazard identification consists of three levels of detail. First, it selects the most obvious hazards for a product, based on reported foodborne outbreaks in the past. Second, in more detailed analysis, hazards are selected based on reported presence of pathogens in the ingredients of the product. Finally, comprehensive hazard identification can be performed for selection of unexpected hazards. In case of selection of many hazards, knowledge rules support the user in selecting the most relevant hazards for a product, making the procedure interactive.After hazard identification, exposure assessment is performed as part of QRA. Modelling microbial growth and inactivation is an important aspect of exposure assessment. Many predictive models have been developed in the past years, varying from general and simple models to specific and advanced models. Since no model is able to accurately predict microbial responses under all conditions, it is sensible to start with simple models and obtain order of magnitude estimates. If relevant, more accurate estimates can best be gained by comparing various models. It was shown that advanced models not necessary result in better estimates. In other words, the virtues of simple models were shown for both rough and detailed exposure assessments.Estimation of the extent of inactivation under various conditions is also part of exposure assessment. For inactivation by irradiation, we studied the quantitatively most relevant factors for the irradiation parameter D ₁₀ . A data analysis of 539 D ₁₀ values from the literature resulted in a first classification of D ₁₀ in spores and vegetative bacteria, with spores having significantly higher D ₁₀ values. Further analysis confirmed extreme high resistance of various vegetative bacteria. The categorisation of quantitatively important factors into separate D ₁₀ categories is a useful tool in designing and evaluating irradiation processes.Next to hazard identification and exposure assessment, hazard characterisation and risk characterisation are the third and fourth aspects of QRA. These four aspects have been integrated in a stepwise approach for QRA; the SIEFE model. SIEFE is an acronym for Stepwise and Interactive Evaluation of Food safety by an Expert system. The main goal of the SIEFE model is obtaining quantitative insight into food production processes. The stepwise approach starts roughly and semi-quantitatively, to find risk-determining phenomena. These phenomena can then be studied more accurately in a second level of detail. Non-relevant aspects can be omitted in this level, simplifying the complex problem of microbial food safety assessment. A third level of detail can be used for even more detailed analyses, for example stochastic description of parameters.The SIEFE model providing quantitative insight into food production processes has been shown by application of the SIEFE model to two example products. In addition, this was confirmed by a comparison of the SIEFE model to another approach of microbial QRA from the literature. Transparent risk assessment was shown to be a powerful tool in decision-making, even if not all necessary quantitative information is available