Analysis of Ingredient Lists to Quantitatively Characterize Chemicals in Consumer Products
Isaacs, K., K. Phillips, J. Wambaugh, AND P. Price. Analysis of Ingredient Lists to Quantitatively Characterize Chemicals in Consumer Products. SOT Annual Meeting 2016, New Orleans, LA, March 13 - 17, 2016.
Presented at SOT Annual Meeting 2016 March 13-17 New Orleans
The EPA’s ExpoCast program is developing high throughput (HT) approaches to generate the needed exposure estimates to compare against HT bioactivity data generated from the US inter-agency Tox21 and the US EPA ToxCast programs. Assessing such exposures for the thousands of chemicals in consumer products requires data on product composition. This is a challenge since quantitative product composition data are rarely available. We developed methods to predict the weight fractions of chemicals in consumer products from weight fraction-ordered chemical ingredient lists, and curated a library of such lists from online manufacturer and retailer sites. The probabilistic model predicts weight fraction as a function of the total number of reported ingredients, the rank of the ingredient in the list, the minimum weight fraction for which ingredients were reported, and the total weight fraction of unreported ingredients. Weight fractions predicted by the model compared very well to available quantitative weight fraction data obtained from Material Safety Data Sheets for products with 3-8 ingredients. Lists were located from the online sources for 5148 products containing 8422 unique ingredient names. A total of 1100 of these names could be located in EPA’s HT chemical database (DSSTox), and linked to 864 unique Chemical Abstract Service Registration Numbers (392 of which were in the Tox21 chemical library). Weight fractions were estimated for these 864 CASRN. Using a recently-developed EPA database of chemical function, the functions associated with the largest numbers of chemicals in this dataset were surfactants (N=278), skin conditioners (N=258), fragrances (N=251), solvents (N=149), and colorants (N=143). Functions associated with the highest mean predicted weight fractions were fillers (0.67), lubricants (0.51), solvents (0.45), surfactants (0.44), and cleansers (0.43). These ingredient data sources and weight-fraction prediction models advance our ability to parameterize empirical and mechanistic exposure models for chemicals in consumer products in support of risk-based decision-making. This abstract does not necessarily reflect U.S. EPA policy.