Using Multiple Sources of Surveillance Data to Attribute Foodborne Illnesses

BACKGROUND:  Estimating the number of human illnesses that are due to specific food sources is necessary to determine the interventions needed to reduce illness and to measure progress toward public health goals. Estimates of foodborne illness source attribution are used for many purposes, including informing strategic planning and risk-based decision-making, and estimating the benefits and impact of interventions. In 2011, the Centers for Disease Control and Prevention (CDC), the Food and Drug Administration (FDA), and the Food Safety and Inspection Service (FSIS) partnered to form the Interagency Food Safety Analytics Collaboration (IFSAC), with the current focus of estimating the source attribution of infections associated with specific foods and settings.

METHODS:  Data from outbreak investigations reported by States provide the primary link between foods and illnesses, and are commonly used for source attribution models. Surveillance data provided by active FoodNet surveillance are essential to track changes in incidence over time and the proportion of illnesses that are sporadic or travel-related. Special studies, such as case-control studies, provide information on both food and non-food exposures associated with illnesses. National surveillance data on laboratory-confirmed infections can be combined with data on food contamination (FDA and FSIS regulatory testing) and food consumption to develop different types of source attribution models. Using these surveillance data and different analytic approaches, IFSAC is working to better estimate how many illnesses are attributable to each food source.

RESULTS: Several projects are providing their first deliverables. Using outbreak data and new methods, IFSAC is updating source attribution fractions for four priority foodborne pathogens (Campylobacter, E. coli O157, Listeria, and Salmonella). A Bayesian model was developed to estimate the expected number of Salmonella illnesses attributable to specific food categories, using food consumption data and pathogen isolation data from regulatory testing. Finally, a source attribution model that blended estimates obtained from a case-control study of laboratory-confirmed infections with estimates from outbreak data was also developed to attribute Salmonella serotype Enteritidis foodborne infections to shell eggs (40%) and other major sources. 

CONCLUSIONS:  Human surveillance data, as well as surveillance data on food contamination and food consumption, are crucial to efforts to improve food safety and track progress toward health goals. Multiple approaches using data from different surveillance systems are being used to attribute illnesses to specific food and non-food sources.