Probabilistic risk calculation for chemical mixtures: environmental risk of pharmaceuticals under future scenarios

Preparing for tomorrow’s environmental issues requires understanding of how risks will evolve with time. Current regulatory models of environmental risk assessment of pharmaceuticals make a conservative prediction of present risk, without considering interactions with global change or presenting uncertainty transparently. In this paper, we present a prototype object-oriented Bayesian network for the prediction of risk for 6 pharmaceuticals across 36 spatial, temporal, population growth and infrastructure scenarios. We compare individual and combined distributions of Risk Quotients across the scenarios. Our results suggest that risk posed would be greatest in rural regions, especially under larger population growth scenarios, but that improved wastewater treatment infrastructure could mitigate risk. We demonstrate the added value of a joint probability of risk threshold exceedance approach, to summed Risk Quotients. With this prototype, we have developed a large-scale probabilistic model and shown its value in forecasting risk, including via an alternative approach to combining individual risks.