Risk assessment of chemicals and their mixtures are hindered by scarcity and inconsistencies between different environmental exposure limits

In chemical risk assessment, measured or modelled environmental concentrations are compared to environmental exposure limits (EELs), such as Predicted No Effect Concentrations (PNECs) or hazardous concentrations for 5% of species (HC05s) derived from species sensitivity distributions (SSDs). However, for many chemicals the EELs include large uncertainties or, in the worst case, the necessary data for their estimation are completely missing. This makes the assessment of chemical risks and any subsequent implementation of management strategies challenging. In this study we analyzed the uncertainty of EELs and its impact on chemical risk assessment. First, we compared three individual EEL datasets, two primarily based on experimental data and one based on computational predictions. The comparison demonstrates large disagreements between EEL data sources, with experimentally derived EELs differing by more than seven orders of magnitude. In a case-study, based on the predicted emissions of 2005 chemicals, we showed that these uncertainties lead to significantly different risk assessment outcomes, including large differences in the magnitude of the total risk, risk driver identification, and the ranking of use categories as risk contributors. We also show that the large data-gaps in EEL datasets cannot be covered by commonly used computational approaches (QSARs). We conclude that an expanded framework for interpreting risk characterization outcomes is needed. We also argue that the large data-gaps present in ecotoxicological data need to be addressed in order to achieve the European zero pollution vision as the growing emphasis on ambient exposures will further increase the demand for accurate and well-established EELs.

Evolution of the sewage treatment plant model SimpleTreat: use of realistic biodegradability tests in probabilistic model simulations

Given the large number of chemicals under regulatory scrutiny, models play a crucial role in the screening phase of the environmental risk assessment. The sewage treatment plant (STP) model SimpleTreat 3.1 is routinely applied as part of the European Union System for the Evaluation of Substances to estimate the fate and elimination of organic chemicals discharged via sewage. SimpleTreat estimates tend to be conservative and therefore only useful for lower-tier assessments. A probabilistic version of SimpleTreat was built on the updated version of the model (SimpleTreat 3.2, presented in a parallel article in this issue), embracing likeliest as well as worst-case conditions in a statistically robust way. Probabilistic parameters representing the variability of sewage characteristics, STP design, and operational parameters were based on actual STP conditions for activated sludge plants in Europe. An evaluation study was carried out for 4 chemicals with distinct sorption and biodegradability profiles: tonalide, triclosan, trimethoprim, and linear alkylbenzene sulfonate. Simulations incorporated information on biodegradability simulation studies with activated sludge (OECD 314B and OECD 303A tests). Good agreement for both median values and variability ranges was observed between model estimates and monitoring data. The uncertainty analysis highlighted the importance of refined data on partitioning and biodegradability in activated sludge to achieve realistic estimates. The study indicates that the best strategy to refine the exposure assessment of down-the-drain chemicals is by integrating higher-tier laboratory data with probabilistic STP simulations and, if possible, by comparing them with monitoring data for validation.

Evolution of the sewage treatment plant model SimpleTreat: applicability domain and data requirements

SimpleTreat 3.1 is the sewage treatment plant (STP) model implemented in the European Union (EU) framework for the environmental risk assessment of chemicals. The model was originally designed for neutral hydrophobic chemicals, whereas many substances currently under regulatory scrutiny, are ionizable at environmental pH. Although the model has been adapted to describe ionization (SimpleTreat 3.1), the fate of organic ions is limited to the unbound aqueous phase, which seriously restricts the applicability domain. New regressions were implemented to estimate the sludge-water partition coefficient normalized to organic carbon (KOC ) of monovalent acids and bases from the octanol-water partition coefficient (KOW ), the dissociation constant (pKa) and the pH. We evaluated the updated model (SimpleTreat 3.2) with 10 test chemicals by comparing predictions with monitoring data collected from the literature. Test chemicals were specifically selected to challenge the applicability domain and to cover a wide range of functionality and physical-chemical properties. Although predicted effluent concentrations are generally conservative, SimpleTreat 3.2 provides reasonable estimates for use in lower-tier risk assessment for neutral and monovalent ionizable chemicals. The accuracy of the new KOC regressions is acceptable for monovalent acid but is lower for bases, for which measured sludge KOC is highly recommended. Measured KOC are also recommended for ionic surfactants and necessary for organic ligands, which may limit the applicability of SimpleTreat using a basic input data set. The conservative nature of model estimates reflects the default worst case, non-numerical parameterization of biodegradation rates and the assumption that biodegradation is limited to the unbound aqueous phase. The potential of refining the description of biodegradation using higher tier simulation tests is explored in a parallel article (Franco et al. this issue).