[Introduction]: In the past few years, food safety has gained increased interest in the industry and the government. This resulted in the introduction of GMP and HACCP to guarantee food safety. However, food production is not a static situation: changes in the food chain (e.g. other raw materials, distribution channels etc.) will occur and influence food safety. Quantitative insight is required to evaluate the effect of these changes. Process Risk Modelling (PRM) is a quantitative microbial risk assessment (QRMA) method, which helps with this. PRM estimates the risk to the consumer as a function of process parameters in the food chain (Figure 1). By changing the values of the process parameters, the effect of, for instance, a new technology can be evaluated. [The current QRMA-studies have a number of flaws]: - The studies do not separate variability and uncertainty of the process parameters. Suppose that a model indicates that the temperature of the refrigerator determines the final risk for a large part. For instance, a model uses a range from 2 to 10°C for the temperature of a refrigerator. This can imply that the temperature indeed varies so much (variability) or that the temperature is constant on a certain value, which is known to lie between 2 and 10°C (uncertainty). This difference is important with regard to future decisions. When the temperature is uncertain, it is advisable to do further study to determine the value of the temperature more precisely. When the temperature has a large variability, it is advisable to put more effort into temperature control. - Criteria for model selection are lacking. Each study uses different methods for data-collection and data-combination, different models for microbial growth etc. - The studies examine specific pathogen/product-combinations (for instance, Campylobacter spp. in broiler chickens). The studies evaluate risk management options without considering the effects on other organisms (e.g. Salmonella) and on the sensorial quality of the products. - The majority of the studies adopt a farm-to-fork approach, but often the primary production phase is not considered due to a lack of data. A more logical starting point would be to specify a certain acceptable risk to the consumer, and reason backwards to how the risk is influenced. [Aim]: Development of a general method for modelling the hazards in a food chain, which includes variability as well as uncertainty, and validation of this methodology with a reference chain. [Method]: - Analysis of existing studies - Collection of missing data - Exploration of the performance of different modelling methods (Point-estimates, Monte Carlo simulations, Bayesian statistics etc.) - Development of methods and criteria for risk assessments of food chains - Apply methodology for case-study salmonella on chicken
|