Summary
The objectives of this project are to develop a risk assessment model that will take knowledge from multiple sources and research studies into account to determine where and when the risk of a contamination event is increased. Input will be sought from the scientific literature as well as experts in the field. Data will be distilled into online risk maps that point out where and when to concentrate efforts to mitigate known risks. The tool will help growers to be more pro-active in managing risk from foodborne pathogen intrusion into their fields and may lead to an even safer supply of the nutritious products grown in California.
Technical Abstract
We are proposing to develop a quantitative microbial risk assessment model that will enable produce growers to more easily assess their specific risk so that appropriate intervention measures may be taken when they are likely most effective and necessary. Such a new approach would distill the current knowledge into a spatio-temporal risk assessment tool supplemented with expert opinion where data may be missing. We will start with focus group meetings of researchers, extension specialists, representatives of industry, and governmental agencies who will provide feedback on data sources to include into the model, and who will point out which information may still be unavailable but desirable to include. The goal of the focus group meetings is to verify that appropriate pathways, model assumptions and input parameters are used for the risk model and to identify any additional information sources that may have been missed by the research team. Further, current knowledge gaps will be discussed to ensure acknowledgement of risk model limitations. The focus group will be updated on project progress twice yearly and further feedback sought for the duration of the project. The envisioned spatial-explicit quantitative microbiological risk assessment model will integrate qualitative and quantitative data that are translated into probability distributions that will be used to estimate the risk of contamination under various scenarios. We combine scenario tree modeling with high-resolution spatial information, which will enable quantification of risks at a fine spatial scale allowing producers to identify their specific risk. The final step in the process will be to develop a user-friendly web-based decision support tool based on the programming language R and R package Shiny that enables us to design interactive web apps. Together, the project will create a framework for the implementation of best practices for risk mitigation for fresh produce fields under different agricultural, climatic and environmental system scenarios. We envision this tool to contribute to safer produce production while allowing for co-existence with livestock commodities in a safe and sustainable fashion.
Research Objectives
The project aims to develop a risk assessment framework for the implementation of best practices for risk mitigation for fresh produce contamination under different agricultural scenarios. 4
Objective 1: Focus-groups -listening sessions
• Focus groups with researchers, extension specialists, representatives of industry, and governmental agencies will be conducted for collection and discussion of model inputs.
Objective 2: Quantitative risk assessment to evaluate the risk of fresh produce contamination under various agricultural, climatic and environmental system scenarios
• A quantitative microbial risk assessment model will be developed to estimate the probability of the presence/absence of pathogenic E. coli under diverse epidemiological scenarios The risk assessment models will predict the risk of produce contamination in relation to: 1) proximity to grazing cattle (different distances and stocking density), 2) watershed microbial quality, 3) weather events (wet, floods, drought, wind, etc.), 4) adjacent land use (CAFOs, pastures, rangeland, natural barriers) and 5) presence of various types of wildlife.
Objective 3: Decision support tool: Online user-friendly platform
• A web-based platform based of the findings of Objectives 1 and 2 will be developed, using R and shiny, to help produce growers and stakeholders better understand the pathogen dynamics under different agricultural scenarios. This user-friendly tool will support decision-making in order to prevent the contamination of fresh produce.
Findings & Recommendations
We developed a framework for risk assessment under various agricultural settings, climatic, and environmental scenarios in the Salinas Valley, CA, that integrated multiple sources of information, including evidence from the current body of literature, expert opinion, and public databases using a hybrid stochastic and spatially explicit risk assessment model approach.
This framework was integrated into a user-friendly interface dashboard that is publicly available to help growers, the produce industry, and other stakeholders manage risks associated with contamination of leafy greens with pathogenic E. coli via different pathways. These pathways include: 1) wildlife intrusion, 2) animal feeding operations (AFOs) and concentrated animal feeding operations (CAFOs), 3) grazing cattle, 4) other sources (proximity to compost facilities, diversified small-scale farms [DSSF], and hobby farms), 5) irrigation water, and 6) flood events.
We established ongoing partnerships with the industry that led to a series of discussions that helped shape the model through a participatory modeling approach. Moreover, these interactions developed into a fully structured expert opinion elicitation process to generate data to populate some of the parameters for the risk assessment for which only scarce data were available.
By following a participatory modeling approach, not only were we able to use the stakeholders’ feedback to improve the risk assessment model and its assumptions, but we also increased the usefulness of the tool and the engagement of its potential users. In other words, stakeholders were active participants in the co-design and beta-testing of the tool, which will directly increase usability and operational value of the developed risk assessment.
Future work focused on data collection and integration of relevant data sources would further strengthen the model, increase its quantitative rigor, and support continued refinement for industry use. Potential improvements include individual model customization using grower-specific data, without compromising data ownership, liability, or confidentiality.
The sensitivity analysis indicated that the initial probability of contamination is the parameter with the greatest influence on model outcomes. Incorporating additional data, such as baseline contamination data from on-farm testing, would help reduce uncertainty and enable more precise risk estimates. Furthermore, as new evidence becomes available in the scientific literature, continued model refinement will be possible using the current framework.
Regardless of these limitations, the model represents an important step towards improving food safety and provides a valuable decision-support tool for farmers, stakeholders, and the produce industry overall.