Food Protection Trends - September 2009 - 571

GENERAL INTEREST PAPER Re-engineering the United States Food Safety System MATTHIEU COUTURIER1 and NANCY LEVESON2 Technology and Policy Program and 2Engineering Systems Division and Aeronautics and Astronautics Dept., Massachusetts Institute of Technology, Room 33–334, 77 Massachusetts Ave., Cambridge, MA 02139-4301, USA 1 SUMMARY The current food safety system is broken, with a patchwork of surveillance systems and over 15 agencies in charge of food safety; this was made clear in February with the Peanut Corporation of America Salmonella contamination. In this article, we describe a new approach to risk management that can potentially support reengineering the United States food safety system. The model, based on systems theory, departs from the traditional chain-of-events models and uses a systems engineering approach to tackle the problem. INTRODUCTION Every year, one in four Americans will suffer from food poisoning, according to the US Centers for Disease Control and Prevention (CDC) (5). In recent years, bagged spinach, green onions, hot peppers and tomatoes were recalled because of hepatitis A, Escherichia coli O157:H7, and Salmonella contamination. At the beginning of the year, Salmonellatainted peanut butter products manufactured by the Peanut Corporation of America (PCA) in Georgia killed nine people, sickened an estimated 22,000 and forced manufacturers to recall over 3,000 products (4, 6, 13). The current food safety system, with a patchwork of surveillance systems and over 15 agencies in charge of food safety (2), was designed for a much simpler and local food supply chain and is overwhelmed in this new environment. Four major federal agencies (FDA, USDA, EPA and DHS) and a myriad of state agencies are in charge of inspections, standards, regulation and certification of the US food supply chain. This makes for a very complex system in which the different agencies act independently and with potentially overlapping mandates. Parts of the food supply chain can fall through the cracks of those agencies and go unmonitored and unregulated. Furthermore, the agencies, both at the state and federal levels, are underfunded and do not have sufficient resources to conduct health inspections of local plants, much less inspect foreign production plants. New regulation and new funding are long overdue to help protect the health of the American public. The existing system is outdated, puts the public at risk, and goes against the long-term financial interests of food manufacturers (16). Similarly, the monitoring system in charge of detecting foodborne illnesses is very slow to react and is not designed to properly handle food contamination at the national or international level. Foodborne illness can spread all across the country—the peanut butter Salmonella contamination affected people in 43 different states and in Canada (15). It can originate either from within the country or from abroad—in 2008, a Salmonella outbreak resulted from contaminated jalapeño peppers from Mexico (3). It typically takes two weeks between the time someone is diagnosed with an illness and the time the test result is submitted to federal officials. At the same time, the food supply chain is so complex that it is often hard to trace the problem back to the contamination source. The Minnesota Department of Health, known as one of the best in the nation, followed several wrong leads before being able to track down the peanut butter problem, thanks to jars of peanut butter found in a nursing home (8). Clearly something needs to be done to fix these problems, but the question is, What? A recent report provided nineteen recommendations for strengthening the system (18). Sometimes, however, intervening in complex systems leads to similar or even worse problems through unintended consequences. Standard risk management engineering techniques include building and analyzing models to understand the sources of risk and to evaluate potential changes meant to reduce risk; however, those techniques have had limited applicability to these types of problems because the tools were created for man-made engineering artifacts in which the assumptions do not match those of more complex, sociotechnical systems like the food and public health systems. New engineering risk management approaches that do work on complex social systems, however, are applicable. In this article, we describe a new approach to risk management that can potentially support re-engineering the US food safety system. The first step in the reengineering process is to model and analyze the current safety control structure. The models can then be used to generate and evaluate potential changes and improvements. MODelINg The CURReNT US FOOD SAFeTY SYSTeM Traditionally, engineering safety and loss techniques are based on a model of causality that assumes that losses occur because of chains of directly-related failure events. For example, the owner of a peanut factory ships peanuts that have failed tests for contamination, the peanuts are used in commercial products, and customers get sick. A root cause is assessed, which is usually some event along the chain. In the example, the root cause assessed might be the actions of SEPTEMBER 2009 | FOOD PROTECTION TRENDS 571

Food Protection Trends - September 2009

Table of Contents for the Digital Edition of Food Protection Trends - September 2009