Dealing with Variability in Industry Risk Assessments to Support Safe Product Design

Making decisions on the safety of food products under uncertainty and based on variable data/information may undermine confidence in the decision. This has traditionally led to the application of deterministic calculations with conservative assumptions (to account for uncertainty and variability) in the food industry when establishing the basis for safety of new products. Performance criteria and related metrics based on such conservative approach, have been in existence for a long time and provided a good safety record. In some circumstances, though, probabilistic approaches offer advantages for innovative product design, as they enable the quantification of the relative likelihood of alternative outcomes, thus better informing decision making under uncertainty and variability. Sources of uncertainty and variability may come from product/process parameters, microbial responses to control measures, and prevalence/level of microbial contaminants, among others. Methodologies used for articulation of uncertainty and variability in safety assessments are well established (e.g., Monte Carlo simulation, sensitivity and scenario analyses techniques). This presentation illustrates how a combination of deterministic and probabilistic approaches can be used to ensure products are safe by design in a fit-for-purpose manner. A case study related to the validation of a continuous thermal sterilisation processing line for low-acid liquid foods will be used to illustrate application in industry. It describes how process and microbiological data can be integrated iteratively with microbial inactivation and thermal process models in a Monte Carlo simulation framework to define a safe operating space. The suitable application and interpretation of these methods requires specialist knowledge and training, and therefore are unlikely to be the ‘norm’ for industry-based risk assessors. However, they are extremely valuable in increasing confidence in decision making for complex product designs where the decision would otherwise be hindered if such methods were not available to industry.