Purpose: We will report the results of an expert elicitation designed to provide globally consistent, and regionally specific food attribution estimates for 11 major pathogens. These attribution estimates were designed for use in the WHO work on the global burden of foodborne disease.
Methods: Expert elicitation is used because of a lack of consistent conventional data sources in many lower income regions. The study uses Cooke’s Classical expert elicitation method because it provides a transparent, reproducible method of measuring widespread tendencies toward bias and toward over or understating objective uncertainty and uses this information in aggregating expert responses. Median food attribution estimates and 95% uncertainty bounds were developed for 154 pathogen/subregion pairs.
Results: Performance-weighted aggregate estimates provided more statistically accurate estimates than equal-weighted aggregates without great sacrifice of informativeness. Exposure to between 2 and 3 foods typically account for over 80 percent of total foodborne cases for specific pathogen/subregion pairs. The role of specific foods in causing foodborne illness due to specific pathogen varies across subregions for some, but not all pathogens. Uncertainty intervals are generally wide, as would be expected given scant data collection and research on food attribution in many regions of the world.
Significance: These estimates also provide a means of comparing the relative roles of different foods in causing major foodborne illnesses around the world. These results suggest that for many pathogens and subregions, it may be advisable to target prevention efforts on specific food exposure routes.