Evaluating Intervention Strategies to Reduce Contamination of Fresh Produce at Farm: Using Field Data to Improve the Predictive Capabilities of a “Virtual Laboratory”

Introduction: Fresh produce can become contaminated due to contact with different contamination sources such as irrigation water, soil amendment, wild and domestic animals, and workers, among others. Given the wide range of contamination sources, there is a need to systematically characterize the contamination potential and compare the efficacy of different interventions to reduce contamination and, hence, risk of illness.

Purpose: The purpose of this project was to develop and demonstrate a “virtual laboratory” that could (1) support the investigation of “contamination scenarios”, (2) allow for the comparison of different interventions, and (3) offer a risk-based sampling approach for microbiological contamination in the growing field.

Methods: Pilot studies evaluated Escherichia coli O157:H7 contamination of romaine lettuce and Salmonella spp. contamination of fresh market tomatoes during the production and harvest stages. An Agent-Based Modeling framework was used to predict the contamination prevalence and levels in the growing field. Input values were derived using data from literature review, expert judgment, and data generated in a field trial experiment in California’s Salinas Valley to support contamination transfer rates associated with different events (e.g., water splash from animal feces).

Results: The notional results from the hypothetical case studies suggested that contamination levels could be significantly reduced by limiting wild animal access to the growing field, and assigning sufficient buffer zones between the field and the neighboring cattle farm. Furthermore, the application of a risk-based sampling approach indicated that contaminated units (e.g., lettuce heads and tomato plants) could be identified with a higher probably than standard “Z” sampling patterns.

Significance: This methodology offers a transparent, practical, and robust modeling approach with which to evaluate the efficacy of different mitigation options and identify areas in the growing field for targeted sampling activities.