Foodborne Proportion of Gastrointestinal Illness: Estimates from a Canadian Expert Elicitation Survey

Abstract

The study used a structured expert elicitation survey to derive estimates of the foodborne attributable proportion for nine illnesses caused by enteric pathogens in Canada. It was based on a similar study conducted in the United States and focused on Campylobacter, Escherichia coli O157:H7, Listeria monocytogenes, nontyphoidal Salmonella enterica, Shigella spp., Vibrio spp., Yersinia enterocolitica, Cryptosporidium parvum, and Norwalk-like virus. For each pathogen, experts were asked to provide their best estimate and low and high limits for the proportion of foodborne illness relative to total cases. In addition, they provided background information with regard to food safety experience, including self-evaluated expertise for each pathogen on a 5-point scale. A snowball approach was used to identify 152 experts within Canada. The experts' background details were summarized using descriptive statistics. Factor analysis was used to determine whether the variability in best estimates was related to self-assessed level of expertise or other background information. Cluster analysis followed by beta function fitting was undertaken on best estimates from experts who self-evaluated their expertise 3 or higher. In parallel, Monte Carlo resampling was run using triangular distributions based on each expert's best estimate and its limits. Sixty-six experts encompassing various academic backgrounds, fields of expertise, and experiences relevant to food safety provided usable data. Considerable variation between experts in their estimated foodborne attributable proportions was observed over all diseases, without any relationship to the expert's background. Uncertainty about their estimate (measured by the low and high limits) varied between experts and between pathogens as well. Both cluster analysis and Monte Carlo resampling clearly indicated disagreement between experts for Campylobacter, E. coli O157, L. monocytogenes, Salmonella, Vibrio, and Y. enterocolitica. In the absence of more reliable estimates, the observed discrepancy between experts must be explored and understood before one can judge which opinion is the best.

Food-Specific Attribution of Selected Gastrointestinal Illnesses: Estimates from a Canadian Expert Elicitation Survey

“…In our earlier report on survey responses for foodborne proportion of illness (Ravel et al, 2010), only experts who selfassessed their pathogen expertise as 3 ( = medium, some direct experience but wide reading) or higher were used in the analysis. Although we recognized that some self-assessments may have been inflated, there was no correlation between expertise level (3, 4, or 5) and estimates of foodborne proportion of illness.…”

“…The tool for the Canadian survey was adapted from the work of Hoffmann et al (2006) with permission of the authors and it is described in Ravel et al (2010). In the first part of the survey, respondents provided details on their academic training and work experience and a self-assessment of their expertise with respect to each pathogen and each food group.…”

“…The data collection was stopped in early December 2008 after three reminders emailed to the nonrespondents. More details are reported in the work of Ravel et al (2010). …”

“…A secondary objective was to characterize uncertainty in the estimates as well as the level of agreement among experts on food source attribution. The first part of this study focused on the proportion of cases attributed to all food sources (i.e., via foodborne route) and the analysis was presented in an earlier article (Ravel et al, 2010). This second article reports the experts' attributions of illness to specific food sources and compares these opinions by pathogen-food combinations.…”

Enteric outbreak surveillance in British Columbia, 2009-2013

“…European data demonstrated that eggs and egg products were the most important food vehicles for Salmonella (10). Food-specific attribution has been done previously in Canada twice using expert elicitation and once using outbreak data (16,17,18). All three identified poultry meat as the most likely source of Salmonella infections.…”