Using the US National Air Toxics Assessment to benchmark the USEtox inhalation-mediated carcinogenic impacts of air emissions

Regional air quality models generate a high-resolution prediction of pollutant concentrations using sophisticated treatments of atmospheric fate and transport processes, speciation, and chemistry. By comparison, the multi-media box modeling underpinning the USEtox model is far simpler but allows for broad coverage of hazardous air pollutants. The purpose of the present article is to benchmark USEtox results via inhalation against a state-of-the-science regulatory model and investigate methodological areas that drive differences in results. The 2014 US National Air Toxics Assessment (NATA) used a linked dispersion-regional air quality model to evaluate human health risk from emissions of air toxics, a prescribed list of hazardous air pollutants that can cause cancer and severe human health effects. We compare the NATA results for human health risk from carcinogenic emissions to characterization factors generated by running the USEtox model on the same starting emissions inventory, namely, the US National Emissions Inventory (NEI), using the North American model parameterization. We have also decomposed characterization factors into intake fractions and effect factors. The separate comparison analysis allows us to trace the driving force for cancer risk differences. Statistical analysis is applied to determine the significance of pollutant characteristics in driving differences in results, including molecular weight, vapor pressure, octanol–water partitioning, compartment half-lives in air, and chemical composition. Comparative analysis shows a good concordance for carcinogenic impacts between the two models, with a factor of ~10 difference. The difference in effect factors is ~1.4, which is within an expected range. However, intake fraction is the major cause of cancer risk divergence. Intake fractions in NATA tend to have large variations due to the high-resolution air dispersion model and including secondary pollutants. NATA and USEtox results have consistent levels of statistical significance across the pollutant characteristics evaluated, with the greatest level of significance (p < 0.05) for molecular weight and vapor pressure, which strongly influence atmospheric fate. Formaldehyde is the dominant contributor to the total burden of diseases with 5000 to 20,000 cases of cancer per year. It is critical to benchmark LCIA results against current state-of-the-science models in order to ensure that LCA results can be reliably used for decision support. The USEtox model was not designed to mimic the complexity of full-scale air quality modeling, but still manages to produce a relative scaling of carcinogenic impacts that is consistent with NATA. Further regionalization, calibration, or inclusion of additional chemical transformations within the USEtox model may yield further performance improvements.