Purpose: The study aimed to compare published semi-quantitative risk ranking approaches with an alternative quantitative approach that includes log transformation (either with or without binning) to score individual food-hazard pairs across the ranking criteria, using fresh produce as a model system.
Methods: Data from literature were used to define scoring bins for ranking criteria used in a published risk ranking model. 10,000 food-pathogen pairs were randomly generated from uniform distributions over realistic ranges of the criteria using standard risk assessment methods to define a reference set, and these random variables were then transformed and aggregated according to the different ranking methods. The semi-quantitative method used bins to assign each criterion to an arbitrarily defined number, and the alternative methods used log transformed risk scores on a scale between 0 and 1 with or without binning. Individual criteria scores were then summed to derive a final risk score for each produce-pathogen pair. The ranking methods were compared to the reference set using scattergrams and Kendall’s rank correlation coefficient.
Results: The alternative quantitative methods had markedly higher correlation coefficients than those of the semi-quantitative method. The log transformation without binning provides the best ranking relative to the reference method, and the log-transformation with binning performs almost the same. The results indicated that use of a quantitative model allows for a higher resolution and reduction in the loss of information and better alignment with sound mathematical principles.
Significance: A fully quantitative risk ranking method provides a useful approach to prioritize produce-pathogen pairs and support risk-based decision making.