Time valuation of historical outbreak attribution data.

Area of Science:

• Food safety
• Epidemiology
• Statistical modeling

Background:

• Accurate foodborne illness attribution is crucial for public health decisions and food safety.
• Current methods often lack precision due to limited data, leading to uncertainty in estimating illness proportions linked to specific food commodities.
• Existing statistical approaches for combining data over time are often data-intensive or lack theoretical grounding.

Purpose of the Study:

• To develop and validate a novel statistical methodology for estimating foodborne illness attribution fractions.
• To address the uncertainty associated with limited annual outbreak data by leveraging historical information.
• To provide a more robust and biologically plausible approach for food safety decision-making.

Main Methods:

• Introduction of a new estimation strategy that progressively down-weights historical data based on deviations from a Poisson process.
• Application of the methodology to estimate attribution fractions for Salmonella and Escherichia coli O157:H7 in common food commodities.
• Comparison of the new method's estimates against two alternative statistical estimators.

Main Results:

• The proposed method offers a statistically sound way to "borrow strength" from previous years' data without over-reliance.
• Estimates for Salmonella and E. coli O157:H7 attribution fractions were generated for various food commodities.
• Performance evaluation indicated the new method's potential to provide more reliable estimates compared to alternatives.

Conclusions:

• The developed estimation strategy provides a valuable tool for improving the accuracy of foodborne illness attribution.
• This method enhances the ability to prioritize food safety interventions and monitor progress.
• The approach offers a statistically rigorous and adaptable framework for analyzing outbreak data over time.