The regulatory challenge of chemicals in the environment: Toxicity testing, risk assessment, and decision‐making models

ABSTRACT Environmental assessment for chemicals relies on models of fate, exposure, toxicity, risk, and impacts. Together, these models should provide scientific support for regulatory risk management decision‐making, assuming that progress through the data‐information‐knowledge‐wisdom (DIKW) hierarchy is both appropriate and sufficient. Improving existing regulatory processes necessitates continuing enhancement of interpretation and evaluation of key data for use in decision‐making schemes, including ecotoxicity testing data, physical‐chemical properties, and environmental fate processes. Yet, as environmental objectives also increase in scope and sophistication to encompass a safe chemical economy, testing, risk assessment, and decision‐making are subject to additional complexity due to the ongoing interaction between science and policy models. Problems associated with existing design and implementation choices in science and policy have both limited needed development beyond chemo‐centric environmental risk assessment modeling and constrained needed improvements in environmental decision‐making. Without a thorough understanding of either the scientific foundations or the disparate evaluation processes for validation, quality, and relevance, this results in complex technical and philosophical problems that increase costs and decrease productivity. Both over‐ and under‐management of chemicals are consequences of failure to validate key model assumptions, unjustified standardized views on data selection, and inordinate reification (i.e., abstract concepts are wrongly treated as facts). HighlightsEnvironmental assessment of chemicals relies on models of fate, exposure, toxicity, and risk.as environmental protection objectives increase in sophistication regulatory processes increase in complexity.Increased complexity results in increased costs and decreased productivity.Key issues include unvalidated model assumptions, unjustified views on data selection, and inordinate reification.Improvements needed: increased rigor in model design, data to information development, and validity/relevance determination.