Study explores risks of chemical exposure from household products

According to the Centers for Disease Control and Prevention, exposure to chemicals can cause negative health effects. Building on previous research identifying chemicals in consumer products, SwRI and EPA also analyzed how rubber, plastic, apparel, upholstery and fabric samples responded to environmental factors such as hot cars or clothing.

The research was published in Environmental Science and Technology magazine, examined four years of data captured through advanced chromatography, suspect screening, untargeted analysis and Highlight™, a machine learning method developed by SwRI. Rather than screening samples for a single known compound, this approach allows scientists to identify, characterize, and evaluate large libraries of chemicals through suspect screening analysis. This method identified 88,795 unique chemical signatures and 1,883 compound groups from 13 analyzed batches.

“Highlight uses machine learning algorithms for fast pattern matching to speed up workflow,” said William Watson, a research engineer in SwRI’s Intelligent Systems Division and lead author of the study.

Another aim of the research is to advance the field of exposomes, which explores how exposure to chemicals from environmental, dietary, lifestyle and other sources throughout life affects human health. Describing chemicals in household items and common sources of exposure may aid future biomonitoring efforts.

“Consumer products are made up of more than just one chemical. Think of it as a mixed bag of related chemicals,” said Kristin Favela, Ph.D., a researcher in SwRI’s Department of Chemistry and Chemical Engineering. “We wanted to determine whether the chemicals in the samples were ’emissionable’ or ‘extractable’ to understand the extent and likelihood of human exposure to the chemicals.”

SwRI exposes clothing, upholstery, fabric, rubber and plastic samples to two different heat settings and solvent strengths. The researchers wanted to determine whether the test samples would release chemical vapors that could be inhaled in indoor environments (such as a hot car) or while being worn. The study also explored whether the chemicals could be extracted to better understand real-world exposure risks, such as when children chew household items.

“In addition to helping to deepen our understanding of the risks to the public from exposure to chemicals, this study demonstrates our ability to retrospectively analyze and understand old data sets using machine learning and highlighting results,” Watson said.

Using EPA’s toxicity prediction program ToxCast, the team performed additional analysis and interpretation to predict risk based on human exposure and available bioactivity data. Of the 88 identified extractable and emissible chemicals, 66 have available ToxCast data, and the majority of ToxCast in vitro assay data (92%, an average of 441 assays per chemical) indicate that there is no concentration-dependent active. No data are available for the other 22 chemicals. However, at higher concentrations, BKF, a synthetic antioxidant used to stabilize plastics and rubber, did show adverse effects at exposures up to 42.3 mg/kg/day. This research may help advance a screening model that can predict the emission activity of household products.