Prioritising pharmaceuticals for environmental risk assessment: Towards adequate and feasible first-tier selection

The presence of pharmaceuticals in the aquatic environment, and the concerns for negative effects on aquatic organisms, has gained increasing attention over the last years. As ecotoxicity data are lacking for most active pharmaceutical ingredients (APIs), it is important to identify strategies to prioritise APIs for ecotoxicity testing and environmental monitoring. We have used nine previously proposed prioritisation schemes, both risk- and hazard-based, to rank 582 APIs. The similarities and differences in overall ranking results and input data were compared. Moreover, we analysed how well the methods ranked seven relatively well-studied APIs. It is concluded that the hazard-based methods were more successful in correctly ranking the well-studied APIs, but the fish plasma model, which includes human pharmacological data, also showed a high success rate. The results of the analyses show that the input data availability vary significantly; some data, such as logP, are available for most API while information about environmental concentrations and bioconcentration are still scarce. The results also suggest that the exposure estimates in risk-based methods need to be improved and that the inclusion of effect measures at first-tier prioritisation might underestimate risks. It is proposed that in order to develop an adequate prioritisation scheme, improved data on exposure such as degradation and sewage treatment removal and bioconcentration ability should be further considered. The use of ATC codes may also be useful for the development of a prioritisation scheme that includes the mode of action of pharmaceuticals and, to some extent, mixture effects.

► The accuracy of nine prioritisation schemes is analysed.
► Exposure estimates based only on sales statistics is highly misleading.
► Inadequate effect measures in first-tier prioritisation underestimate risks.
► Without proper exposure and effect data, hazard-based methods outperform risk-based.