Impact assessment of emission management strategies of the pharmaceuticals Metformin and Metoprolol to the aquatic environment using Bayesian networks

The issue of pharmaceuticals in the environment has caused increasing concern in the recent years and various strategies have been proposed to tackle this problem. This work describes a Bayesian network (BN)-based socio-ecological impact assessment of a set of measures aimed at reducing the entry of pharmaceuticals in the aquatic environment. The measures investigated were selected across three sectors: public health market, environmental politics and drug design innovation. The BN model was developed for two drugs, Metformin and Metoprolol, and it models the distribution of the Predicted Environmental Concentration (PEC) values as a function of different measures. Results show that the sensitivity of the PEC for the two drugs to the measures investigated reflects the distinct drug characteristics, suggesting that in order to ensure the successful reduction of a broad range of substances, a spectrum of measures targeting the entire lifecycle of a pharmaceutical should be implemented. Furthermore, evaluation of two scenarios reflecting different emission management strategies highlights that the integrated implementation of a comprehensive set of measures across the three sectors results in a more extensive reduction of the contamination. Finally, the BN provides an initial forecasting tool to model the PEC of a drug as a function of a combination of measures in a context-specific manner and possible adaptations of the model are proposed.