Multicriteria Decision Analysis Characterization of Chemical Hazard Assessment Data Sources

ABSTRACT Chemical hazard assessment (CHA), which aims to investigate the inherent hazard potential of chemicals, has been developed with the purpose of promoting safer consumer products. Despite the increasing use of CHA in recent years, finding adequate and reliable toxicity data required for CHA is still challenging due to issues regarding data completeness and data quality. Also, collecting data from primary toxicity reports or literature can be time consuming, which promotes the use of secondary data sources instead. In this study, we evaluate and characterize numerous secondary data sources on the basis of 5 performance attributes: reliability, adequacy, transparency, volume, and ease of use. We use GreenScreen for Safer Chemicals v1.4 as the CHA framework, which defines the endpoints of interest used in this analysis. We focused upon 34 data sources that reflect 3 types of secondary data: chemical‐oriented data sources, hazard‐trait–oriented data sources, and predictive data sources. To integrate and analyze the evaluation results, we applied 2 multicriteria decision analysis (MCDA) methodologies: multiattribute utility theory (MAUT) and stochastic multiobjective acceptability analysis (SMAA). Overall, the findings in this research program allow us to explore the relative importance of performance criteria and the data source quality for effectively conducting CHA. Integr Environ Assess Manag 2019;00:1–14. © 2019 SETAC Key Points The evaluation and characterization of secondary data sources through multicriteria decision analysis (MCDA) provide an innovative quantitative approach on guiding how to choose and evaluate data sources for chemical hazard assessment. Through MCDA analysis, the chemical‐oriented data sources exhibit better performance compared to hazard‐trait–oriented data sources and predictive data sources for chemical hazard assessment. The sensitivity analysis in multiattribute utility theory (MAUT) indicates the final score for each data source will vary by changing the weighting factor for each performance attribute, and some common trends can be captured for the 3 different types of data sources. The applications of MAUT and stochastic multiobjective acceptability analysis (SMAA) are complementary for making decisions on which data sources are more suitable for chemical hazard assessment and which performance attributes are of greatest importance.