BAMRA: Bayesian approaches to microbial risk assessment

Summary

Regulators often need to make decisions regarding relative risks within a particular food chain which cannot be observed directly: e.g. would human food-borne illness be reduced by the introduction of a particular intervention into pasteurised milk production?  Simulation and mathematical modelling have become important tools to support MRA (Microbial Risk Assessment) in making such decisions.

The project investigates whether involving stakeholders in assessing potential interventions and their likely effects would improve or detract from the overall findings.  This involves experimentally comparing the effects of including and excluding stakeholders in a series of workshops aimed at dealing with a hypothetical food chain risk.  All the stakeholders will contribute mainly to the assessment and incorporation of value judgements into the decision making, but some will also be expert in the ‘science’ that is related to understanding the risk and there will be a need to elicit their ‘best’ estimates and uncertainties related to the quantitative risk modelling. 

Obtaining unbiased and representative judgements from experts and stakeholders is notoriously difficult and we propose to employ formal ‘expert opinion’ elicitation techniques, developed in the context of Bayesian statistics, which seek to minimise these biases. Methods for Bayesian analysis of computer code outputs (BACCO) will be used to analyse the MRA models and for the propagation of uncertainty. These methods are significantly more efficient than Monte-Carlo methods and will allow for a more detailed analysis of the likely effects of interventions and the associated uncertainty. 

The output from the models will be fed back into the workshops to inform decision making on the interventions.  Finally, the groups will formulate how they will communicate their decisions and the nature of the underlying risk to the public, and the efficacy of their communications will be tested.

Project partners

  • Dr Helen Clough, University of Liverpool
  • Prof. Simon French, Manchester University Business School
  • Dr Andy Hart, Central Science Laboratory, York
  • Dr Jeremy Oakley, University of Sheffield

Researchers

  • Prof Richard Shepherd
  • Martina Petkov

Funding