A critical meta-analysis of predicted no effect concentrations for antimicrobial resistance selection in the environment

•Environmental risk assessment of antimicrobials is needed to reduce AMR selection.•Semi-systematic searches found selective concentrations, collated into a database.•PNECR interpretation facilitated through appraisal of approaches used.•Until further review, a PNECR of 0.01 µg/L for all antibiotics could be considered.•Future research directions are provided. Antimicrobial resistance (AMR) is one of the greatest threats to human health with a growing body of evidence demonstrating that selection for AMR can occur at environmental antimicrobial concentrations. Understanding the concentrations at which selection for resistance may occur is critical to help inform environmental risk assessments and highlight where mitigation strategies are required. A variety of experimental and data approaches have been used to determine these concentrations. However, there is minimal standardisation of existing approaches and no consensus on the relative merits of different methods. We conducted a semi-systematic literature review to collect and critically appraise available minimal selective concentration (MSC) and predicted no effect concentration for resistance (PNECR) data and the approaches used to derive them. There were 21 relevant articles providing 331 selective concentrations, ranging from 0.00087 µg/L (ciprofloxacin) to 2000 µg/L (carbenicillin). Meta-analyses of these data found that selective concentrations are highly compound-dependent, and only a subset of all antimicrobials have been the focus of most of the research. The variety of approaches that have been used, knowledge gaps and future research priorities were identified, as well as recommendations for those considering the selective risks of antimicrobials in the environment. [Display omitted]