Society For Risk Analysis Annual Meeting 2012

Advancing Analysis

Session Schedule & Abstracts

* Disclaimer: All presentations represent the views of the authors, and not the organizations that support their research. Please apply the standard disclaimer that any opinions, findings, and conclusions or recommendations in abstracts, posters, and presentations at the meeting are those of the authors and do not necessarily reflect the views of any other organization or agency. Meeting attendees and authors should be aware that this disclaimer is intended to apply to all abstracts contained in this document. Authors who wish to emphasize this disclaimer should do so in their presentation or poster. In an effort to make the abstracts as concise as possible and easy for meeting participants to read, the abstracts have been formatted such that they exclude references to papers, affiliations, and/or funding sources. Authors who wish to provide attendees with this information should do so in their presentation or poster.

Common abbreviations

Symposium: Challenges Associated with Using Human Surveillance Data for Foodbourne Illness Source Attribution

Room: Pacific Concourse H   10:30 AM - 12 PM

Chair(s): Dana Cole, Michael Batz

Sponsored by MRASG

Foodborne diseases continue to be an important cause of morbidity and mortality in the United States. An important part of establishing and monitoring progress toward targeted food safety goals is to identify the most important sources of foodborne illness. Data collected by human illness surveillance systems of the Centers for Disease Control and Prevention (CDC) have been used to estimate the number of illnesses attributable to various sources for over a decade. This symposium will present some of the challenges associated with using human surveillance data to estimate foodborne illness source attribution and interdisciplinary approaches to address them.

M2-F.1  10:30  A Jungle of Surveillance Numbers, Attribution Methods, and Uncertainties. Cole D*; Centers for Disease Control and Prevention

Abstract: The Centers for Disease Control and Prevention (CDC) has used data from human illness surveillance and specially-designed studies to estimate the number of illnesses attributable to various sources for over a decade. Data from foodborne outbreak investigations have always been an important source of information, because many of these investigations determine the specific food associated with illnesses. However, most foodborne illness is not associated with reported outbreaks, and some causes of foodborne illness are never or rarely associated with foodborne outbreaks. To estimate important sources of sporadic illness, case-control studies have been used. These studies are associated with several sources of uncertainty, including correlated and nested variable structures, missing data, and several sources of bias. Recently available methods may help to address these challenges and refine currently available estimates.

M2-F.2  10:50  The 400 Pound Gorilla: Missing Data and Bias. Hoekstra M*; Centers for Disease Control and Prevention

Abstract: When data are missing from analytic data sets, a series of questions arises regarding what to do to ensure that analysis proceeds toward results that are minimally biased and maximally precise. Combine this challenge with the presence of measurement error and things start to get complicated. In this talk we discuss these questions in the context of two methods for estimating the relative importance of different causes of foodborne illness: case-control studies of sporadic illness and disease outbreak analysis.

M2-F.3  11:10  Estimating the Lion’s Share: Novel Methods for Attributing Sporadic Foodborne Illnesses. Gu W*; Atlanta Research and Education Foundation

Abstract: The majority of foodborne illnesses are sporadic (non outbreak-related) in which food sources cannot be determined for individual patients. Case-control studies offer a glimpse of etiologic patterns by using classification models to identify significant exposures. Conventional analyses of case-control data can be hampered by numerous interrelated exposures and missing data. Tree-based models, e.g., random forest, provide an important alternative for identifying complex causal pathways in the data from exposures to illness. Combined with counterfactual modeling of exposure removal from the study population, these powerful tools can be used for estimation of population attributable fractions. Another line of research taps Bayesian modeling to borrow information from other sources to reduce uncertainties.

M2-F.4  11:30  Cutting Through The Thicket: Evaluating the Applicability of Outbreak-based Attribution. Batz MB*, Hoffmann SA, Morris JG; University of Florida

Abstract: Prioritizing scarce food safety resources in accordance with risk starts with the question: which pathogens, in which foods, cause the most impact on public health. One approach to answering it involves estimating disease incidence, estimating costs of illness and impacts on quality of life, and attributing illnesses to foods. Foodborne illness source attribution presents analytical challenges because no single approach is sufficient and because the most comprehensive source of information – CDC’s national database of reported foodborne outbreaks – has significant limitations. In our rankings of pathogen-food combinations, we developed a new approach for attribution that combines information from outbreak data and an expert elicitation. This approach evaluates the applicability of outbreak-based estimates to attribute foodborne disease incidence to food categories. For each pathogen, we evaluated five characteristics: density of outbreak data (number of outbreaks per year), representativeness of outbreaks to overall incidence (ratio of estimated national disease incidence as estimated by CDC to average annual number of reported outbreak cases), deviation between experts and outbreaks (sum of the mean of the squared difference between experts’ estimates and outbreak estimates), expert agreement as measure of quality of available scientific evidence (mean standard deviation across experts’ individual estimates), and comparison to available case-control studies. Based on the totality of the evidence, we determined outbreak-based attribution was insufficient for Campylobacter, Cryptosporidium parvum, Toxoplasma gondii, and Yersinia enterocolitica. In our comparative risk assessment, we used expert-based attribution estimates for these four pathogens, and outbreak-based attribution for others. Other pathogens for which outbreak-based attribution is lacking include Listeria monocytogenes, non-O157 Shiga-toxin producing E. coli, Shigella, and Cyclospora.

[back to schedule]