A Bayesian probabilistic framework for next-generation chemical risk assessment: The case of PFOA in crops irrigated with treated wastewater

Poly-and perfluoroalkyl substances (PFAS) are persistent organic pollutants ubiquitously detected across aquatic matrices, representing a significant concern for both human health and the resilience of aquatic ecosystems. Continuously released by wastewater treatment plants, there is an urgent need to develop innovative and adaptable methodologies for their risk assessment in both retrospective and prospective ways, aiming to describe uncertainty to produce a reliable risk estimate. Current methods for quantitative health risk assessment for PFAS are mostly deterministic and case-specific, with very few studies available, primarily due to limited knowledge about fate, transport, and toxicity, as well as a lack of suitable data necessitating lab-and time-intensive pilot experiments. Here, a Bayesian network-based risk assessment approach is proposed to model health risk associated with perfluorooctanoic acid through consumption of salad irrigated with reclaimed wastewater. This method, built upon evidence from the literature and experts knowledge, accounts for health risk assessment across conventional exposure scenarios, providing a valuable decision-making and risk management tool for water utilities. Its flexibility and universality allow for application even in data-scarce environments, with robustness improving as new evidence becomes available. The proposed risk model is intended to describe the multi-barrier approach perspective, facilitating forecasting and risk mitigation throughout the supply chain, and supporting efficient water management and reuse. In conclusion, the validated model was applied to data collected during a one-year monitoring campaign at two wastewater treatment plants, revealing a low risk of PFOA exposure.