Risk assessment for Clostridium botulinum: a network approach

Gary C. Barker, Nicola L. C. Talbot, Mike W. Peck

Research output: Contribution to journalArticlepeer-review

27 Citations (Scopus)


The construction and implementation of a mathematical framework for the representation of the hazards that arise from Clostridium botulinum growth, and toxin production, in food are described. Botulism has been recognised as a serious foodborne illness for over a century and, more recently, has become the subject of increased concern due to changing processing and consumption patterns associated with foods. In this respect quantitative risk assessment has an increasingly important role to play in assisting risk management and ensuring the safety of minimally processed foods and foods with extended shelf life. Bayesian Belief Networks are a type of expert system that integrates a graphical, flow diagram like, representation of a hazard domain with a powerful technique for combining probabilities. This technique facilitates the accumulation of understanding and experience, for particular hazard domains, into computer tools that can be used to inspect risks and account for decisions. Analysis of the hazards associated with foodborne botulism involves Belief Network components that represent contamination processes, thermal death kinetics for spores, germination and growth of cells, toxin production and patterns of consumer behaviour, etc. These developments are discussed and three important aspects of the food safety information supply, complexity, dependency and uncertainty highlighted. The benefits associated with a Bayesian view of food safety assessment are illustrated by a Belief Network representation which supports, and prioritises, decisions and actions that (a) minimise the chances and extent of detrimental events and (b) maximise opportunities for awareness and control.
Original languageEnglish
Pages (from-to)167-175
Number of pages9
JournalInternational Biodeterioration & Biodegradation
Issue number3-4
Publication statusPublished - Oct 2002

Cite this