Last reviewed 5 November 2019
While weight of evidence (WoE) has been used in the practice of law for several centuries, with respect to the prioritisation and risk assessment of chemicals, definitions and practices vary in complexity, a new report aims to address that.
No 311 in the Organisation for Economic Cooperation and Development’s series on testing and assessment is intended to provide universal guiding principles that should be considered when developing or augmenting systematic approaches to WoE for chemical evaluation and key elements to formulating a systematic approach to WoE.
The principles include:
a hypothesis which involves a clear formulation and statement of the problem for which evidence is needed and possible alternative hypotheses
be systematic and comprehensive in design by documenting a step-wise procedure integrating all evidence and indicating how evidence was collected, evaluated and weighed
include a treatment of uncertainty arising from available data (knowns) and data and/or knowledge gaps (unknowns)
consider the potential for bias during collection, evaluation and weighing of evidence
be transparent by including clear documentation to assist the communication of WoE decisions so that they can be understood, reproduced, supported or questioned by all interested parties.
The key elements contain the necessary steps that should be taken to determine the overall strength of evidence to answer a hypothesis question and include:
problem formulation (hypothesis development)
evidence collection (establish lines of evidence and knowledge gaps)
evidence evaluation (determine data reliability, uncertainty and relevance)
evidence weighing (assign weight to evidence)
evidence integration and reporting (examine evidence coherence and impact of uncertainty).
The OECD says that if the principles and elements described in this document are considered, a consistent, clear and transparent delivery of evidence can follow allowing all stakeholders to understand decision-making including potential for unreasonable bias.