Developing risk-based food safety management

The drive for risk-based food safety management, systems and control has spread world-wide in recent decades. Since the term is still internationally undefined, its use and implementation vary, producing different realizations. In this Ph.D. thesis, microbiological risk assessment (MRA) was investigated as a basis for risk-based food safety management, which was defined as ‘food safety management based on risk assessment in order to achieve an appropriate level of protection (ALOP)’. Governments are responsible for commissioning MRAs and also for setting food safety targets up to a certain point, but the practical management measures that need to be in place in order to achieve the targets are to be addressed by the operators. On the plant level, food safety is usually managed through regulation, quality assurance systems and a hazard analysis and critical control point (HACCP) programme with its prerequisites. In Finland, food safety management on the food plant level is implemented through an HACCP-like regulated system termed an ownchecking (OC) programme. A quantitative microbiological risk assessment (QMRA) was conducted on salmonella in the beef production chain according to the official standards of the Codex Alimentarius Commission (Codex Alimentarius), and utilized in determining the food safety metrics for beef production. The Finnish Salmonella Control Programme (FSCP) and the main official interventions due to it were examined in the light of risk-based food safety management. The targets set for beef processing plants by the government were converted into quantitative limits, and the results of salmonella monitoring included in the FSCP were examined by the QMRA. The goal of the FSCP was declared in 1994 to ‘maintain the present salmonella situation’, which was considered to refer to the salmonella incidence in humans at that time, and also the de facto ALOP. The requirement for a maximum salmonella prevalence of 1% at defined stages of the beef production chain was embodied in the FSCP. This statement was considered to convey performance objectives (PO) for the aforementioned stages. According to the QMRA, the de facto ALOP was achieved in the referred year 1999, and even the true prevalence levels in the FSCP were estimated to be clearly under the set PO limits with 95% credibility. However, the PO limits were set too high for the de facto ALOP to be maintained in practice. If the salmonella prevalence reached the PO limit of 1% or values near it, the public health risk would increase and overrun the de facto ALOP. The QMRA produced in this work has for the first time provided the possibility to quantitatively asses the relationships between targets set in the FSCP and their impact on public health. At present, imports of beef and beef-derived foods may impose on Finnish consumers a significantly greater exposure than domestic products. If their salmonella prevalence or their share of the foods consumed in Finland increase, the number of human cases could rapidly rise. The models for the QMRA were mainly Bayesian hierarchical models using Markov chain Monte Carlo (MCMC) techniques, which was found to be a flexible and appropriate method for this type of complex modelling. The resulting distributions were also regarded as an advantage compared to the results from models developed with the deterministic approach, because the presentation of results included the extent of the uncertainty, and also in this manner better illustrated the actual operational environment. Based on an inquiry, the personnel in food processing plants had a positive attitude towards food safety management systems, but the knowledge, training and involvement of those employees directly operating on the site with these systems were discovered to be deficient. Therefore, a generic semi-quantitative hygiene risk assessment model, Hygram®, was developed for small and medium-sized food enterprises to offer assistance in understanding, training, and, first of all, detecting the critical steps of the processes, and thereby to contribute to the development of their own-checking systems towards risk-based food safety management. Hygram® was not considered a risk-based tool as such, but whenever the critical limits of the process have been defined as equal to a risk assessment, Hygram® can be used as a risk-based management tool. It can also serve as a tool for systematic hazard analysis and CCP detection when establishing a food safety management system. To conclude, the development of risk-based food safety management is a process in which risk assessment is an essential tool. Scientific, technical, psychological and resource-bound barriers need to be overcome in order to put risk-based management systems into practice. This study showed that QMRA can be valuable in national risk management decision making, although few QMRAs are currently available. Appropriate tools for practical risk management decision making on the industrial level, such as Hygram®, need to be further developed.