Risk Ranking for Public Decision Making

Michael Batz

Resources for the Future

It is the goal of the Food Safety Research Consortium (FSRC) to move towards a food safety system built on science- and risk-based decision making. It is our goal to provide policy makers and stakeholders with the analytical tools necessary in such a system. As a first step, researchers at the University of Maryland School of Medicine (primarily, J. Glenn Morris Jr.) and Resources for the Future (primarily, Michael Taylor, Michael Batz, Alan Krupnick, Sandra Hoffmann) developed the Foodborne Illness Risk Ranking Model (FIRRM) to identify, quantify, and compare the public health impact of the most important microbiological food hazards.

A "top down" Monte Carlo simulation model based on epidemiological approaches and data, FIRRM focuses on 28 foodborne pathogens and their pathways across a comprehensive range of food categories, and produces rankings of food-pathogen combinations. There are three modules in the model: 1) estimation of incidence of each pathogen based on public health surveillance data; 2) economic costs and loss of Quality Adjusted Life Years (QALYs) due to the symptoms, severities, medical treatments (physician visits, hospitalization), fatalities, recovery periods, and chronic sequelae associated with pathogen-specific illnesses; and 3) the attribution of illnesses from pathogens to food-pathogen combinations, based on outbreak data, expert elicitation, risk assessments, case-control studies, and other data. Resulting food-pathogen combinations can be ranked by five measures of public health impact: cases, hospitalizations, deaths, economic costs, and QALYs. Valuation is currently completed for only four pathogens (Salmonella, Campylobacter, E. coli O157:H7, and Listeria monocytogenes), although estimates for remaining FoodNet pathogens, Norovirus, and Toxoplasma are currently underway. The most critical data gap is in food attribution: outbreak data has problems that make it insufficient as a solitary source, and so we attempted to gather expert judgment and other data as alternatives and supplements. These data show large uncertainties in which foods are associated with illness.

Although current results from FIRRM are only preliminary, we are updating many data sources, addressing uncertainty issues attempting to fill data gaps, and developing a web interface. The second version of FIRRM should be completed in October 2006. The model was developed in Analytica, which has a graphical point-and-click user interface; it is likely not "easy to use" for new users, but the eventual web interface should simplify interaction and allow users to organize and save results. The second draft should provide robust rankings useful for broad policy decision-making, such as resource allocation decisions. It will likely only provide partial answers, however, as we still expect to encounter data quality issues.