Hazard grouping framework for antibiotic disinfection by-products in treated wastewater samples: an innovative approach combining (Q)SAR predictions, social choice function, and multi-criteria analysis

A combination of (Q)SAR analysis and the ELECTRE multi-criteria decision-making approach was adopted to classify antibiotics and their disinfection by-products (DBPs) generated during chlorination in four wastewater treatment plants (WWTPs) in Spain, in terms of environmental and human health hazards. A purpose-built database including eight antibiotics and 104 DBPs was updated using in silico (Q)SAR models to predict seven endpoints related to environmental fate (mobility, biodegradability, STP removal, and PBT) and human health (carcinogenicity, mutagenicity, and estrogen receptor binding). Endpoint weights were determined using the Borda count method, based on the judgment of an international panel of experts. The ELECTRE method then integrated the weighted endpoints, grouping compounds according to aggregate hazard. The results showed that 23% of the DBPs formed by chlorination were predicted with higher aggregate hazard than parent compounds, comparing individual DBPs but considering all endpoints evaluated. Application of the hybrid framework to real monitoring data revealed that WWTPs with fewer identified DBPs did not necessarily present lower aggregate hazard. The evaluation of the WWTP studied leads to different results when considering the number or hazard of DBPs, demonstrating the importance of combining strategies. This framework provides a computational screening tool for discerning aggregate DBP-related hazards in WWTPs. However, further experimental confirmation is needed.