A Bayesian Method for Exposure Prevalence Comparison During Foodborne Disease Outbreak Investigations.

Area of Science:

• Epidemiology
• Statistics
• Food Safety

Background:

• Foodborne disease outbreaks require timely source identification for public health.
• Current investigations often use one-sample binomial tests to compare case exposure to general population data.
• This method has limitations in accounting for uncertainty in reference population exposure estimates.

Purpose of the Study:

• To introduce and evaluate a Bayesian alternative to the one-sample binomial test for foodborne outbreak investigations.
• To assess the utility of the Bayesian approach in accounting for uncertainty in exposure prevalence.
• To compare the performance of the Bayesian method with the traditional binomial test using a real-world outbreak scenario.

Main Methods:

• A Bayesian statistical model was developed to estimate exposure prevalence.
• The model was applied to a 2020 Escherichia coli O157:H7 outbreak linked to leafy greens.
• Exposure prevalence in cases was compared with 2018-2019 FoodNet Population Survey data.
• Prospective simulations were conducted at multiple time points during the investigation.

Main Results:

• The Bayesian approach generated posterior probabilities of increased leafy green consumption prevalence as more case-patients were interviewed.
• High probabilities (>0.70) were observed for specific leafy green items before the traditional binomial test reached statistical significance.
• The Bayesian method demonstrated increased sensitivity in detecting associations earlier in the investigation.

Conclusions:

• A Bayesian approach for assessing exposure prevalence in foodborne outbreaks can be more effective than the standard one-sample binomial test.
• This method enhances the ability to identify outbreak vehicles by better handling exposure uncertainty.
• The findings suggest a potential improvement in the speed and accuracy of foodborne outbreak investigations.