Is PM More Toxic Than the Sum of Its Parts? Risk-Assessment Toxicity Factorsvs.PM-Mortality “Effect Functions”

Publisher Website
Google Scholar
Abstract
Epidemiology studies of populations living in areas with good air quality report correlations between levels of ambient particulate matter (PM) and mortality rates. These associations occur at low PM concentrations that are below current air quality standards. Can such concentrations cause mortality, given the toxicity of PM chemical constituents? We examined chemical-specific, dose-response data typically used in U.S. EPA human health risk assessments. These assessments rely on established, no-effect thresholds for noncancer health endpoints. We found that chemicals identified as constituents of ambient PM are present at concentrations considerably below the regulatory thresholds used in risk assessment (i.e., below the RfCs and RfDs that identify levels for which no adverse health effects are anticipated). From the perspective of risk assessment, exposure to the concentrations of chemicals in ambient PM (e.g., sulfate, nitrate, and elemental carbon) cannot be expected to cause death. Hence, the health effects attributed to ambient PM in “regulatory impact analyses” appear to be at odds with what would be predicted from a standard U.S. EPA health-risk assessment for PM chemicals. Four possible resolutions of this paradox are that (1) the mixtures of chemicals present in ambient PM are vastly more toxic than the sum of individual components, (2) small portions of the general population are vastly more sensitive to certain ambient PM chemicals than reflected in U.S. EPA toxicity factors, (3) the toxicity of ambient PM is unrelated to its chemical constituents, or (4) PM mass concentration is not the causal factor in the reported associations. The associations may arise because ambient PM concentrations (1) are a surrogate for unmeasured copollutants (e.g., HAPs), (2) covary with confounding factors that cannot be fully controlled (e.g., weather, demographics), or (3) covary with unmeasured (e.g., societal, behavioral, or stress) factors.