Exploring historical Canadian foodborne outbreak data sets for human illness attribution.

Human illness attribution has been recently recognized as an important tool to better inform food safety decisions. Analysis of outbreak data sets has been used for that purpose. This study was conducted to explore the usefulness of three comprehensive Canadian foodborne outbreak data sets covering 30 years for estimating food attribution in cases of gastrointestinal illness, providing Canadian food attribution estimates from a historical perspective. Information concerning the microbiological etiology and food vehicles recorded for each outbreak was standardized between the data sets. The agent-food vehicle combinations were described and analyzed for changes over time by using multiple correspondence analysis. Overall, 6,908 foodborne outbreaks were available for three decades (1976 through 2005), but the agent and the food vehicle were identified in only 2,107 of these outbreaks. Differences between the data sets were found in the distribution of the cause, the vehicle, and the location or size of the outbreaks. Multiple correspondence analysis revealed an association between Clostridium botulinum and wild meat and between C. botulinum and seafood. This analysis also highlighted changes in food attribution over time and generated the most up-to-date food attribution values for salmonellosis (29% of cases associated with produce, 15% with poultry, and 15% with meat other than poultry, pork, and beef), campylobacteriosis (56% of cases associated with poultry and 22% with dairy products other than fluid milk), and Escherichia coli infection (37% of cases associated with beef, 23% with cooked multi-ingredient dishes, and 11% with meat other than beef, poultry, and pork). Because of the inherent limitations of this approach, only the main findings should be considered for policy making. The use of other human illness attribution approaches may provide further clarification.