The Application of Multi-Criteria Decision Analysis to Improve the Reliability of Chemical Hazard Assessments

Most decisions aimed at improving sustainability require trade-offs and the balance of priorities. Multi-Criteria Decision Analysis (MCDA) methods can be used to facilitate assessment of these trade-offs. Chemical Hazard... [ view full abstract ]

Most decisions aimed at improving sustainability require trade-offs and the balance of priorities. Multi-Criteria Decision Analysis (MCDA) methods can be used to facilitate assessment of these trade-offs. Chemical Hazard Assessment (CHA) is one strategy for improving sustainability, with a particular focus on reducing the inherent hazard due to chemicals in products. CHA inherently requires addressing trade-offs in priorities, since CHA accounts for various aspects of human health, physical, and environmental hazards. Furthermore, CHA requires the collection of data on potential chemical hazard, but there are significant challenges due to data gap and quality issues. For instance, the data needed to effectively conduct a CHA is most frequently derived from animal testing, yet clearly not every chemical can be extensively tested. Also, there is a gap between the assessment stage and the data collection stage because most of the commonly used data sources only contain classification results without the details on the chemical testing, which makes it difficult to determine the data quality. In this study, our purpose was to use MCDA, not to evaluate trade-offs between chemical hazard traits in CHA, but rather to evaluate the available data sources relative to their data quality and reliability, thereby identifying which data sources are best suited for use in CHA. The framework we used for the CHA was GreenScreenÒ for Safer Chemicals v1.3. Sixteen data sources were evaluated and classified. These data sources included the Globally Harmonized System of Classification and Labeling in Japan (GHS- Japan); Sigma Aldrich material safety data sheets (MSDSs); GESTIS; Collaborative on Health and the Environment, Toxicant and Disease Database (CHE); Occupational Toxicants and MAK values (MAK); The National Institute for Occupational Safety and Health Pocket Guide (NIOSH); California Proposition 65 (Prop 65); PAN Pesticides Databse (PAN); PubChem; The Toxicology Data Network (TOXNET); The Organization for Economic Cooperation and Development eChemPortal (eChemPortal); International Chemical Secretariat Substitute it Now List (ChemSec SIN); The Endocrine Disruption Exchange (TEDX); EPI-Suite^TM; Danish (Q)SAR Database ((Q)SAR) and VEGA. The evaluation and classification was conducted on the basis of six criteria: reliability, adequacy, transparency, volume, accessibility and ease of use. We then applied two MCDA methodologies – Multi-Attribute Value Theory (MAVT) and Technique for Order Preference by Similarity to the Ideal Solution (TOPSIS) – to integrate the classification results. Through the use of DecernsMCDA DE software, we were able to conduct the weighting sensitivity analysis to understand the influence of each classification criterion on the output result. The evaluation and classification of data quality and reliability improve the transparency and reliability of CHA, and the MCDA provides guidance on which data sources are best suited for the assessment.