Particulate Matter (PM) Risk Assessment and Cardiopulmonary Disease

RoseyFields conducts particulate matter risk assessments using the following methodology. Before presenting it, RoseyFields flags an important structural point: particulate matter is not assessed under EPA’s single-chemical IRIS framework (the framework that produces a reference dose, reference concentration, or cancer slope factor derived from an animal point of departure and a set of numeric uncertainty factors). PM is a “criteria pollutant” regulated under the Clean Air Act through National Ambient Air Quality Standards (NAAQS), informed by an Integrated Science Assessment (ISA) that uses a five-level, weight-of-evidence causality framework built primarily on human epidemiological cohort evidence rather than an animal-derived point of departure. RoseyFields’ methodology below follows the user’s nine-part outline, but sections 3 through 7 are adapted to reflect the actual EPA framework that governs PM, rather than forcing PM into a POD/uncertainty-factor/RfD-RfC/CSF structure that does not exist for this pollutant. Where a requested element (e.g., a numeric uncertainty factor table, or a cancer slope factor) simply is not part of how PM is regulated, RoseyFields says so explicitly rather than inventing one.

1. Chemical Identity, Regulatory Background, and Exposure Sources

Particulate matter is not a single chemical but a heterogeneous mixture of solid and liquid particles suspended in air, classified by aerodynamic size rather than by a single chemical formula. The size fractions relevant to regulation and to RoseyFields’ assessments are PM2.5 (particles with a nominal aerodynamic diameter of 2.5 micrometers (µm) or less, “fine” particles), PM10 (particles 10 µm or less), PM10-2.5 (“thoracic coarse” particles, the fraction between 2.5 and 10 µm), and ultrafine particles (UFPs, generally under 0.1 µm), which are measured but not currently subject to a separate NAAQS.

PM is both primary (directly emitted) and secondary (formed in the atmosphere from gaseous precursors including sulfur dioxide, oxides of nitrogen, ammonia, and volatile organic compounds). Within urban environments, primary PM2.5 emissions arise from motor vehicle exhaust and brake/tire wear, industrial combustion, residential and commercial fuel combustion including wood burning, and cooking, while secondary PM2.5 formed from SO2 and NOx precursors accounts for the majority of PM2.5 mass in many locations. PM10-2.5 is almost entirely primary in origin and is dominated by crustal (soil/dust) material, sea salt, and biological material, while UFPs arise both from direct emission (chiefly motor vehicles, especially near roadways) and from atmospheric new-particle formation. Human exposure is overwhelmingly by inhalation; PM is not assessed for oral or dermal exposure pathways in the regulatory framework, and RoseyFields’ assessments accordingly focus exclusively on ambient and personal air exposure.

Regulation of PM proceeds under Clean Air Act Sections 108 and 109, which direct EPA to periodically review the air quality criteria for PM (compiled in the ISA) and to set and review primary (health-based) and secondary (welfare-based) NAAQS. As of RoseyFields’ most recent review, the operative primary standards are an annual PM2.5 standard of 9.0 µg/m3 (annual mean, averaged over 3 years), a 24-hour PM2.5 standard of 35 µg/m3 (98th percentile, averaged over 3 years), and a 24-hour PM10 standard of 150 µg/m3 (not to be exceeded more than once per year on average over 3 years); the secondary annual PM2.5 standard is 15.0 µg/m3. The annual primary PM2.5 standard was strengthened from 12.0 to 9.0 µg/m3 in a 2024 EPA rulemaking, which has been the subject of ongoing litigation and agency reconsideration; RoseyFields checks the current NAAQS table before finalizing any case-specific assessment, because this value is actively contested and subject to change. Unlike an IRIS RfD or RfC, a NAAQS is not framed as a toxicological “safe dose” derived from a single critical study; it is a regulatory standard set by the EPA Administrator to protect public health with an adequate margin of safety, informed by both the ISA’s causality determinations and a separate quantitative risk and exposure assessment.

2. Point-of-Departure Selection: Why the IRIS Framework Does Not Apply, and What EPA Uses Instead

RoseyFields does not identify, and EPA’s PM ISA does not derive, a single critical animal study, NOAEL, or LOAEL functioning as a “point of departure” for PM in the sense used for IRIS chemical assessments. Three structural features of PM explain why:

PM is assessed for evidence of a no-threshold, continuous concentration-response relationship, not a NOAEL. For short-term and long-term PM2.5 exposure, the epidemiological evidence synthesized in the ISA does not identify a population-level threshold below which effects clearly do not occur; the ISA instead evaluates the shape of the concentration-response (C-R) relationship across the observed range of ambient concentrations, including at and below current NAAQS levels. This is fundamentally different from the NOAEL/LOAEL paradigm, which presumes a threshold exists and seeks to locate it in a single controlled study.

The primary evidence base is human epidemiological cohort data, not animal toxicology. Rather than selecting one principal animal study, EPA’s ISA integrates very large prospective human cohort studies as its central evidentiary pillar for the best-supported PM2.5 health outcomes, corroborated by controlled human exposure and animal toxicological studies that establish biological plausibility and mechanism. The most influential of these cohorts include the extended Harvard Six Cities cohort (Lepeule et al., 2012), the American Cancer Society Cancer Prevention Study II (CPS-II) cohort analyzed for cardiopulmonary mortality and lung cancer (Pope et al., 2002), a shape-of-the-concentration-response analysis of the same cohort (Pope et al., 2009), an extended CPS-II never-smoker analysis for lung cancer (Turner et al., 2011), and a national Medicare-population time-series analysis of short-term PM2.5 exposure and mortality (Di et al., 2017a). RoseyFields relies on this same body of large-cohort epidemiology, rather than an animal point of departure, as the evidentiary basis for its PM risk characterizations.

EPA uses a five-level causality determination, not a single POD, to characterize the evidence for each outcome. The ISA classifies the evidence for each PM size fraction, exposure duration, and health outcome combination into one of five categories: (1) causal relationship; (2) likely to be causal relationship; (3) suggestive of, but not sufficient to infer, a causal relationship; (4) inadequate to infer the presence or absence of a causal relationship; or (5) not likely to be a causal relationship. RoseyFields treats this causality determination, rather than a numeric POD, as the analog starting point for a PM risk assessment: it identifies which health outcomes have a sufficiently well-established, well-replicated evidentiary basis to support quantitative risk characterization at all. For PM2.5, the current ISA determines a causal relationship for short-term and long-term cardiovascular effects and for both short-term and long-term mortality; a likely to be causal relationship for short-term and long-term respiratory effects, long-term nervous system effects, and pregnancy/birth outcomes; and a suggestive, but not sufficient to infer, causal relationship for cancer, short-term nervous system effects, and male/female reproduction and fertility. For PM10-2.5, only short-term respiratory and short-term cardiovascular effects reach “suggestive” or higher; most other outcome categories for PM10-2.5 and for UFPs remain at “suggestive” or “inadequate.”

RoseyFields’ practical consequence of this framework is that the strength of quantitative risk characterization RoseyFields can defensibly offer varies by outcome: mortality and cardiovascular disease (causal) support the most robust quantitative concentration-response analysis; respiratory disease (likely causal) supports strong but slightly more qualified quantitative analysis; and cancer, reproductive, and nervous system effects (suggestive) support only a qualitative or heavily caveated quantitative discussion.

3. Dosimetric Modeling: Deposition and Internal Dose, Not Animal-to-Human Extrapolation of a Toxicity Value

Although there is no POD to dosimetrically adjust, PM dosimetry science plays an important, differently-purposed role in RoseyFields’ methodology: it establishes the biological plausibility linking ambient (external) PM concentrations to internal deposited and retained dose, which supports the causality determinations described above and informs which size fractions are of greatest concern for which target tissues.

Particle deposition in the respiratory tract is governed by particle size, with larger PM10-2.5 particles depositing predominantly in the extrathoracic and tracheobronchial regions and smaller PM2.5 and UFP particles able to penetrate to, and deposit in, the alveolar region. EPA’s dosimetry modeling uses the Multiple-Path Particle Dosimetry (MPPD) model to estimate species-specific and region-specific deposition fractions across the respiratory tract for a given particle size distribution, and to support cross-species comparison of deposited dose in animal toxicological studies (Miller et al., 2016). A small fraction of deposited, poorly soluble ultrafine particles (generally under 200 nm) has been shown to translocate from the lung into systemic circulation and distribute to other organs, providing a plausible mechanistic pathway for the extrapulmonary (e.g., cardiovascular, nervous system) effects associated with PM exposure; the fraction translocating is generally under roughly 0.2% for particles in the 5-200 nm range, though it may be higher for the smallest particles in that range. RoseyFields relies on this deposition and translocation science to support biological plausibility arguments in a case, rather than to numerically adjust a toxicity value, because no single toxicity value analogous to an RfC exists for PM to adjust.

On the exposure side, rather than an animal-to-human internal dose adjustment, RoseyFields’ “dose conversion” work for PM is exposure modeling: converting ambient monitoring or modeled air concentrations into an individually-relevant exposure estimate. EPA’s ISA describes a hierarchy of exposure estimation approaches of increasing sophistication, from simple fixed-site monitor assignment, through land-use regression and chemical transport models, to hybrid models that combine chemical transport model output, satellite-derived aerosol optical depth, ground-based monitoring, and land-use variables to produce fine-scale (often 1 km or better) predicted PM2.5 surfaces with strong out-of-sample cross-validation performance. Indoor infiltration (the fraction of ambient PM that penetrates indoors, denoted Finf) is an additional adjustment RoseyFields applies where appropriate, since Finf is consistently below 1 and varies by particle size, season, and building characteristics, and time-series analyses have shown that ignoring infiltration produces attenuated (biased toward the null) health-effect estimates relative to models that account for it.

4. Uncertainty Factor Derivation: Why Numeric UFs Are Not Applied, and What Replaces Them

RoseyFields does not apply, and EPA’s PM framework does not derive, the interspecies, intraspecies, subchronic-to-chronic, LOAEL-to-NOAEL, or database-completeness uncertainty factors used in IRIS chemical assessments, because there is no single animal POD to which such factors would be applied. This is not an oversight; it reflects the fact that the primary PM evidence base is human epidemiological data collected across large, diverse populations at ambient exposure concentrations, which avoids the interspecies and route-to-route extrapolation uncertainties that numeric UFs are designed to address. In place of numeric UFs, RoseyFields’ methodology addresses analogous sources of uncertainty using the qualitative and quantitative tools EPA itself applies in the ISA and in its companion quantitative risk and exposure assessments:

In place of interspecies/intraspecies uncertainty: RoseyFields evaluates whether an epidemiological association is consistent across multiple independent human cohorts, geographic regions, and exposure-assessment methods (a form of direct human replication that substitutes for animal-to-human extrapolation), and separately reviews the populations and lifestages the ISA identifies as being at increased risk of a PM-related health effect (Section 9 below), which functions similarly to (but is evidentiarily stronger than) a default intraspecies/sensitive-subpopulation UF.

In place of subchronic-to-chronic extrapolation: Because the major PM cohort studies (e.g., the extended Harvard Six Cities follow-up in Lepeule et al., 2012, and the CPS-II cohort in Pope et al., 2002 and related publications) directly measure long-term (years to decades) mortality and morbidity outcomes in relation to long-term average exposure, no subchronic-to-chronic extrapolation is needed for long-term effects; RoseyFields treats short-term (days) and long-term (months to years) PM2.5 exposure as two separate, independently evidenced exposure regimes, each with its own causality determination and its own concentration-response literature, rather than extrapolating from one to the other.

In place of a LOAEL-to-NOAEL adjustment: Because the epidemiological evidence generally does not establish a population threshold (a no-effect level) for PM2.5, particularly for mortality and cardiovascular disease, RoseyFields does not apply a LOAEL-to-NOAEL-type adjustment; instead, RoseyFields’ quantitative work characterizes risk as a continuous function of exposure concentration using a concentration-response coefficient (see Section 6), consistent with EPA’s own approach in its quantitative risk assessments.

In place of a database uncertainty factor: RoseyFields addresses database completeness qualitatively, by relying on the ISA’s own causality determination (Section 2) as an integrated judgment of the overall strength, consistency, and coherence of the evidence base for each outcome, and by separately considering established sources of study-level uncertainty specific to PM epidemiology: potential copollutant confounding (since PM is never encountered as an isolated single-component exposure), exposure measurement error (attenuating associations toward the null in most cases, per the ISA’s own assessment), and averaging-time and lag-structure choices in the underlying studies.

5. Derivation of the “RfD”/”RfC”: No Such Value Exists for PM — the NAAQS Is the Operative Standard

RoseyFields confirms that EPA has not derived, and does not maintain, an oral reference dose or inhalation reference concentration for particulate matter in the sense used for IRIS chemicals such as hydrogen sulfide. There is no PM entry in the IRIS database at all; PM is a criteria pollutant regulated exclusively through the NAAQS process described in Section 1. RoseyFields treats the current NAAQS values as the operative regulatory benchmarks in place of an RfD/RfC:

• Annual primary PM2.5 standard: 9.0 µg/m3 (annual mean, 3-year average)
• 24-hour primary PM2.5 standard: 35 µg/m3 (98th percentile, 3-year average)
• 24-hour primary PM10 standard: 150 µg/m3 (not to be exceeded more than once per year on average over 3 years)
• Annual secondary PM2.5 standard: 15.0 µg/m3

RoseyFields emphasizes an important conceptual distinction for any report or testimony: an RfD/RfC is framed as an estimate of an exposure “likely to be without an appreciable risk of deleterious effects” to even sensitive populations, essentially a health-protective individual-exposure benchmark; a NAAQS is a population-level ambient air quality standard, set by the EPA Administrator considering both the health science (from the ISA) and a quantitative risk and exposure assessment (which estimates population-level health risk, including excess mortality and morbidity risk, at varying possible standard levels), and is required by the Clean Air Act to protect public health with an adequate margin of safety, not to represent a concentration with zero risk. Because the epidemiological evidence for PM2.5 mortality and cardiovascular disease does not establish a clear population threshold, EPA’s own risk assessments have found that some risk is estimated to remain even at levels at or below the current NAAQS; this is a materially different regulatory and scientific posture than an IRIS RfC, and RoseyFields does not characterize the PM2.5 NAAQS to a court or regulator as a toxicological no-effect level.

6. Carcinogenicity Classification: Suggestive Under EPA’s ISA Framework; Group 1 Under IARC — No EPA-Derived CSF/IUR Exists

RoseyFields distinguishes two separate carcinogenicity classifications that are sometimes conflated in litigation and public discussion, and does not treat them as interchangeable:

EPA’s PM ISA causality determination for cancer. Under the ISA’s five-level framework, the current (2019) determination for long-term PM2.5 exposure and cancer is “suggestive of, but not sufficient to infer, a causal relationship” — a step below “likely to be causal” and well below “causal.” This reflects a body of supportive but not fully consistent epidemiological evidence, including cohort findings of increased lung cancer incidence and mortality associated with long-term PM2.5 exposure (Pope et al., 2002; Turner et al., 2011), alongside meta-analytic and mechanistic uncertainty that has kept the overall determination at the “suggestive” level. Because the determination is not “causal” or “likely to be causal,” EPA has not derived, and RoseyFields does not apply, an EPA-derived inhalation unit risk (IUR) or cancer slope factor (CSF) for PM under this framework, in contrast to the H2S assessment (which has no cancer value for the separate reason that the animal/human database was found inadequate).

IARC’s classification of outdoor air pollution and PM. In a separate process, the International Agency for Research on Cancer convened an expert working group in 2013 that classified outdoor air pollution, and PM specifically as a major component of outdoor air pollution, as “carcinogenic to humans” (IARC Group 1), based on sufficient evidence of carcinogenicity in humans and experimental animals and strong mechanistic support (Loomis et al., 2013). RoseyFields notes this classification is scientifically significant and is often invoked in litigation, but flags for any report or testimony that: (a) IARC’s Group 1 classification is a hazard identification (whether the agent can cause cancer under some circumstance), not a dose-response or potency assessment, and IARC does not itself derive a CSF or IUR; and (b) IARC’s classification technically applies to “outdoor air pollution” as a complex mixture and to PM as a major identified component of that mixture, not to a PM CSF specific to a given exposure scenario.

Practical consequence for quantitative cancer risk assessment. Because no EPA-derived CSF or IUR exists for PM, RoseyFields does not calculate an excess lifetime cancer risk for PM exposure using the standard IRIS-style methodology (exposure concentration multiplied by a chemical-specific IUR). Where a case requires some quantitative characterization of PM-associated cancer risk, RoseyFields instead relies on epidemiological relative-risk or hazard-ratio estimates from the cohort literature (e.g., the lung cancer hazard ratios reported in Turner et al., 2011, and the lung cancer/cardiopulmonary mortality relative risks in Pope et al., 2002) applied to a case-specific baseline cancer incidence rate, and clearly labels this as an epidemiologically-derived excess risk estimate rather than an IRIS-style CSF-based calculation, given the different assumptions and evidentiary bases underlying each approach.

7. How RoseyFields’ Risk Assessors Apply These Values in Practice

Standard-based screening in place of a Hazard Quotient. Because PM has no RfD/RfC, RoseyFields does not calculate a Hazard Quotient (exposure ÷ RfC) for PM. In its place, RoseyFields performs a NAAQS-comparison screen: reconstructing a case-specific ambient or personal PM2.5/PM10 exposure estimate (see below) and comparing it to the applicable annual and 24-hour NAAQS values, to characterize whether a location or time period was in attainment or nonattainment, and by how much, over the relevant exposure window. This is a regulatory compliance metric rather than a toxicological safety margin, and RoseyFields’ reports state that distinction explicitly.

Excess risk (not excess lifetime cancer risk in the IRIS sense) using concentration-response functions. For the causal and likely-to-be-causal outcomes (Section 2) — principally mortality, cardiovascular disease, and respiratory disease — RoseyFields calculates excess risk using epidemiologically-derived concentration-response (C-R) coefficients from the cohort literature (e.g., the hazard ratios per unit increase in long-term PM2.5 concentration reported in Lepeule et al., 2012, Pope et al., 2009, and Di et al., 2017a), applied to a case- or population-specific baseline incidence/mortality rate and exposure increment, following the same general log-linear or Cox-proportional-hazards C-R function structure EPA itself uses in its quantitative risk and exposure assessments supporting NAAQS reviews. This produces an estimated number or rate of excess health events attributable to a specified PM exposure increment, which is conceptually and mathematically distinct from an IRIS excess lifetime cancer risk calculation (exposure × IUR), even though both are sometimes loosely referred to as “excess risk.”

Exposure and dose reconstruction. RoseyFields reconstructs case-specific PM exposure using the same hierarchy of methods described in the ISA (Section 3): fixed-site regulatory monitor data where a monitor is well-sited relative to the exposure location, hybrid land-use-regression/chemical-transport-model/satellite products where finer spatial resolution is needed, and, for occupational or near-source scenarios, dispersion modeling or personal/microenvironmental monitoring data where available. RoseyFields applies an infiltration adjustment for indoor exposure time and documents the averaging time (24-hour versus annual) used, since the two NAAQS metrics and the two exposure-duration causality determinations are not interchangeable.

Litigation and regulatory support applications. RoseyFields uses this framework to support: (a) Clean Air Act citizen-suit and enforcement matters and NAAQS attainment/nonattainment disputes, where the NAAQS-comparison screen is directly dispositive of a regulatory compliance question; (b) toxic tort and nuisance matters alleging PM-related respiratory, cardiovascular, or mortality harms, where the causal/likely-causal determination and cohort-derived C-R functions support general causation, subject to the same specific-causation requirements (dose reconstruction, temporality, differential diagnosis) RoseyFields applies to any chemical; (c) property value and welfare-effects matters, drawing on the ISA’s separate causal determination for visibility impairment; and (d) rebuttal of opposing expert reports that misapply an RfD/RfC/CSF-style methodology to PM, that treat NAAQS attainment as equivalent to a finding of no health risk, or that treat the “suggestive” cancer causality determination as equivalent to a “causal” finding or an EPA-derived cancer potency value.

8. Key Uncertainties and Sensitive Populations

RoseyFields’ case-specific PM assessments account for the following uncertainties and sensitive-population considerations, drawn directly from the ISA’s own treatment of these issues:

Copollutant confounding. Because PM is never encountered as an isolated exposure — ambient PM co-occurs with ozone, NOx, SO2, and other traffic- and combustion-related pollutants — RoseyFields reviews whether the epidemiological studies relied upon in a given case used copollutant (multipollutant) models to assess whether the PM association is independent, and discloses this as a source of residual uncertainty, particularly for outcome categories at the “suggestive” causality level.

Exposure measurement error. As described in Section 3, exposure error in PM epidemiology tends to attenuate (bias toward the null) observed health-effect associations, though bias in either direction is possible; RoseyFields treats this as a reason that published C-R coefficients may, if anything, understate true effects, rather than as a reason to discount the epidemiological evidence base generally.

PM component and source heterogeneity. Because PM2.5 mass is a composite of many chemically and physically distinct components (sulfate, nitrate, organic and black carbon, crustal material, metals) arising from different sources, and the ISA’s evidence on whether specific components or sources are more strongly associated with health effects than PM2.5 mass overall remains an active area of only partially resolved science, RoseyFields does not assume that a case-specific source’s PM emissions carry the same potency as PM2.5 mass in general without additional source-apportionment analysis.

Populations and lifestages at increased risk. The ISA identifies several groups the evidence indicates may be at increased risk of a PM-related health effect: individuals with pre-existing cardiovascular disease; individuals with pre-existing respiratory disease (asthma, COPD); children, whose lungs are still developing and who have higher ventilation rates relative to body size; older adults; individuals of lower socioeconomic status; and populations residing near major roadways, who experience elevated exposure to traffic-related PM including UFPs. RoseyFields applies these susceptibility factors both when selecting the appropriate cohort-derived concentration-response coefficient for a case (using, where available, subgroup-specific estimates) and when evaluating whether a specific plaintiff’s underlying health status is consistent with heightened PM susceptibility as part of a specific-causation analysis.

The “suggestive” outcome categories require additional caution. For cancer, short-term nervous system effects, and reproductive/fertility outcomes, RoseyFields treats general causation as considerably less well established than for mortality, cardiovascular disease, or respiratory disease, and does not extend the quantitative C-R methodology described in Section 7 to these outcomes without substantial additional case-specific literature review and appropriately qualified opinions.

Actively evolving regulatory standard. Because the annual PM2.5 NAAQS has changed twice in the past decade (2012 revision to 12.0 µg/m3; 2024 revision to 9.0 µg/m3) and remains subject to ongoing litigation and potential reconsideration, RoseyFields confirms the currently effective NAAQS value at the time of any case-specific analysis rather than relying on a previously cited figure, and discloses in any report which NAAQS vintage was in effect during the relevant exposure period, since a location’s attainment status can differ depending on which standard applies.

Bibliography