Data Sources and Methods Used for Attribution
It may seem like it should be straightforward to determine the food sources responsible for illnesses, but it is more complicated than you might think. This is because it is usually not possible to know which food made an individual person sick, or to know if a food was responsible for the illness. People rarely know what made them ill, and it can be difficult or impossible to figure out. When a group of people become ill at the same time in a foodborne outbreak, investigation may sometimes determine which food was responsible, providing a direct link between the foodborne illnesses and a food.
Yet, even in an outbreak it is difficult to determine the contaminated food. This is because:
- People may not know or remember everything they ate;
- Many ingredients went into the dish; and
- Not everyone who ate the same contaminated food becomes ill.
Many different kinds of infections can spread through food. Some infections, like salmonellosis, are common causes of outbreaks, so we understand their sources. Other infections, like toxoplasmosis, do not cause outbreaks, and thus are not reflected in the outbreak data. For these sporadic (non-outbreak) illnesses, it is usually impossible to know what food or other exposure was the source without conducting a special study.Top of Page
There is no single data source or analytic method that is best for estimating attribution for all agents. CDC uses several data sources and combinations of methods to estimate the number of illnesses associated with each type of food. Some are highly specific for one type of infection or one type of food, and others are more general. For example, by comparing the characteristics of Salmonella found in foods and animals with those found in people, we can attribute Salmonella infections back to specific animal sources. The best method may differ for different agents and foods.
- Outbreaks: Outbreak data provide information about the role of various foods in causing illnesses. However, outbreaks contribute only a small proportion (less than 5 percent) of lab-confirmed foodborne illnesses. Only using outbreak data assumes that the food sources are the same for outbreak-related illnesses and the more common sporadic (non-outbreak) illnesses, which is more accurate for some food-pathogen combinations than others.
- The National Outbreak Reporting System (NORS) collects data electronically from state, local, and territorial public health agencies and includes reports of all outbreaks of enteric (intestinal) illness transmitted by food, water, animal contact, or by person-to-person contact that they investigated.
- CDC regularly publishes summaries of foodborne disease outbreaks reported through NORS.
- Sporadic illnesses: Most health departments do not collect detailed data about exposures of individual people with illnesses, such as salmonellosis, that are often foodborne. Because most people do not know what made them sick, illnesses reported to a health department usually do not reveal the likely source. CDC and health departments do intensive studies that compare the exposures of people with sporadic cases of a particular infection, such as E. coli O157 1, with the exposures of other people who are not ill.
- By comparing exposures reported by the two groups, we can determine which foods likely are responsible for some of the illnesses. This approach is called a case-control study.
- Although case-control studies allow us to fill in knowledge gaps, these efforts use many resources and depend on people remembering what they ate.
- Pathogen subtype: Pathogens of the same type are not exactly alike. In the laboratory, small differences between pathogens can be used to subtype them. The subtype is a marker, like a fingerprint, that sometimes can help distinguish strains from different sources.
- Investigators can compare the subtypes of pathogens isolated from animals and foods with the subtypes isolated from infected humans. They can combine this information with data on how often people consume certain foods, and then use mathematical models to estimate the number of illnesses associated with each source.
- CDC and the US Department of Agriculture's Food Safety and Inspection Service (FSIS) recently collaborated to adapt a pathogen-subtype model for Salmonella infections to US data 2.
2 Guo C, Hoekstra RM, Schroeder CM, Pires SM, Ong KL, Hartnett E, Naugle A, Harman J, Bennett P, Cieslak P, Scallan E, Rose B, Holt KG, Kissler B, Mbandi E, Roodsari R, Angulo FJ, Cole D. Application of Bayesian techniques to model the burden of human salmonellosis attributable to U.S. food commodities at the point of processing: adaptation of a Danish model. Foodborne Pathog Dis. 2011 Apr;8(4):509-16. doi: 10.1089/fpd.2010.0714. Epub 2011 Jan 16.
- Page last reviewed: January 8, 2014
- Page last updated: January 8, 2014
- Content source: