T3-02 Development of an Interactive Modeling Tool to Predict the Risks Associated with Contaminated Fresh-cut Lettuce in Canadian Distribution Systems

Monday, July 23, 2012: 1:45 PM
Room 553 (Rhode Island Convention Center)
Sebastien Villeneuve, Agriculture and Agri-Food Canada, StHyacinthe, Canada
Leila Hashemi Beni, Agriculture and Agri-Food Canada, StHyacinthe, Canada
Kevin Cote, Agriculture and Agri-Food Canada, StHyacinthe, Canada
Denyse LeBlanc, Agriculture and Agri-Food Canada, Kentville, Canada
Aamir Fazil, Public Health Agency of Canada, Guelph, Canada
Ainsley Otten, Public Health Agency of Canada, Guelph, Canada
Robin McKellar, Agriculture and Agri-Food Canada, Ottawa, Canada
Pascal Delaquis, Agriculture and Agri-Food Canada, Summerland, Canada
Introduction: The assessment of risks associated with the contamination of fresh-cut lettuce with pathogens such as Escherichia coli O157:H7 would benefit from the means to map the spatial distribution of contaminated product over time and predict potential public health impacts.

Purpose: The objective of this work was to develop a modeling tool capable of 1) mapping the distribution of contaminated fresh-cut lettuce from processing to retail; 2) predicting pathogen levels in packaged lettuce; and 3) estimating consumer accessibility and risk over time.

Methods: A relational database was constructed to store relevant data including: geographical coordinates of Canadian processors, distribution centres, retail outlets; product volumes, time and product temperature at each stage; transportation routes and census data. ArcGIS™ Tools (Network Analyst, Spatial Analyst) were used to calculate transit times and to quantify population accessibility for each retail outlet. The overall risk model incorporating a predictive equation for E. coli O157:H7 in packaged lettuce was designed with Arena™ software.  Visual Basic (VA) was used to develop an interface to dynamically display the distribution of contaminated product and the predicted risk index.

Results: The performance of the tool was demonstrated for a scenario whereby 3600 kg of fresh-cut lettuce contaminated with 1 x 103 CFU/g E. coli 0157:H7 were processed in one plant that supplies two retail chains in one Canadian province. Product distribution and populations most at risk of purchasing contaminated product were measured over time and results were mapped. The model predicted a twelve-day outbreak due to contamination in 3888 of 710,978 retail packages sold in the province. The average prevalence of contaminated packages in implicated retail outlets varied between 9.8 to 14.5%, depending on day.

Significance: The ability to model the spread and impact of contaminated products will enable more accurate estimates of food safety risks and improved identification of vulnerabilities in fresh-cut produce distribution systems.