Utility of Bayesian networks in QMRA-based evaluation of risk reduction options for recycled water.

ABSTRACT

BACKGROUND: Quantitative microbial risk assessment (QMRA), the current method of choice for evaluating human health risks associated with disease-causing microorganisms, is often constrained by issues such as availability of required data, and inability to incorporate the multitude of factors influencing risk. Bayesian networks (BNs), with their ability to handle data paucity, combine quantitative and qualitative information including expert opinions, and ability to offer a systems approach to characterisation of complexity, are increasingly recognised as a powerful, flexible tool that overcomes these limitations. OBJECTIVES: We present a QMRA expressed as a Bayesian network (BN) in a wastewater reuse context, with the objective of demonstrating the utility of the BN method in health risk assessments, particularly for evaluating a range of exposure and risk mitigation scenarios. As a case study, we examine the risk of norovirus infection associated with wastewater-irrigated lettuce. METHODS: A Bayesian network was developed following a QMRA approach, using published data, and reviewed by domain experts using a participatory process. DISCUSSION: Employment of a BN facilitated rapid scenario evaluations, risk minimisation, and predictive comparisons. The BN supported exploration of conditions required for optimal outcomes, as well as investigation of the effect on the reporting nodes of changes in 'upstream' conditions. A significant finding was the indication that if maximum post-treatment risk mitigation measures were implemented, there was a high probability (0.84) of a low risk of infection regardless of fluctuations in other variables, including norovirus concentration in treated wastewater. CONCLUSION: BNs are useful in situations where insufficient empirical data exist to satisfy QMRA requirements and they are exceptionally suited to the integration of risk assessment and risk management in the QMRA context. They allow a comprehensive visual appraisal of major influences in exposure pathways, and rapid interactive risk assessment in multifaceted water reuse scenarios.