PM2.5 (Fine Particulate Matter) Risk Assessment and Cardiopulmonary Disease

RoseyFields conducts PM2.5 risk assessments using the following methodology: this page is based on the Agency for Toxic Substances and Disease Registry’s 2024 Guidance for Inhalation Exposures to Particulate Matter and on EPA’s 2019 Integrated Science Assessment for Particulate Matter and its 2022 Supplement, which together provide the current federal scientific and regulatory framework for evaluating PM2.5. A threshold methodological note is necessary before proceeding: unlike the discrete chemicals RoseyFields typically evaluates through EPA’s IRIS program (with a derived oral reference dose, inhalation reference concentration, and/or cancer slope factor), PM2.5 has no IRIS toxicity value of any kind. ATSDR’s guidance states this explicitly: “ATSDR does not have a Minimal Risk Level (MRL) value for PM.” PM2.5 is instead regulated as one of six “criteria pollutants” under the Clean Air Act’s National Ambient Air Quality Standards (NAAQS) program, and its health-protective comparison values are derived directly from human epidemiological concentration-response relationships rather than from an animal point of departure adjusted by default uncertainty factors. Accordingly, this page follows the structure RoseyFields uses for all chemical-specific methodology pages, but several sections (Sections 3–7) necessarily describe a different underlying scientific process than they would for an IRIS chemical, and the deviations are identified explicitly at each step. Citations to specific findings and numeric values below are to the underlying peer-reviewed epidemiological studies rather than to the ATSDR or EPA documents themselves.

1. Chemical Identity, Regulatory Background, and Exposure Sources

PM2.5 (“fine particulate matter”) is not a single chemical substance but an operationally defined size fraction: particles with a mean aerodynamic diameter of 2.5 micrometers (μm) or less. As such, PM2.5 is a highly heterogeneous mixture whose chemical composition varies by time, region, meteorology, and emission source, and can include sulfates, nitrates, ammonium, organic carbon, elemental (black) carbon, and trace metals, among other constituents. PM2.5 is emitted directly (“primary” PM2.5) from combustion sources — including coal-fired power plants, industrial facilities, motor vehicles (particularly diesel engines), wildfires, and residential wood burning — and is also formed indirectly in the atmosphere (“secondary” PM2.5) through chemical reactions involving precursor gases such as sulfur dioxide (SO2), nitrogen oxides (NOx), ammonia (NH3), and volatile organic compounds. Because secondary formation can occur well downwind of the original emission source, and because fine particles remain suspended in the atmosphere and travel much greater distances than coarser particles, PM2.5 concentrations at a given location often reflect a regional mixture of local and long-range-transported sources rather than a single attributable emitter.

PM2.5 is one of six criteria pollutants for which EPA has established NAAQS under Section 109 of the Clean Air Act, alongside PM10 (thoracic particles, ≤ 10 μm), ozone, carbon monoxide, sulfur dioxide, nitrogen dioxide, and lead. On February 7, 2024, EPA finalized a revision strengthening the primary (health-based) annual PM2.5 NAAQS from 12.0 to 9.0 μg/m3, while retaining the primary 24-hour PM2.5 standard at 35 μg/m3, the secondary (welfare-based) annual standard at 15.0 μg/m3, and the secondary 24-hour standard at 35 μg/m3. Unlike an IRIS RfD or RfC, which is derived once by EPA’s Office of Research and Development and applied nationally as a fixed health-based benchmark until formally revised, a NAAQS is a national ambient air concentration limit tied to a statutory process: the Clean Air Act requires EPA to review the underlying science for each criteria pollutant at least every 5 years and revise the standard as necessary to protect public health with an adequate margin of safety, meaning the numeric PM2.5 benchmark used in regulatory and litigation contexts changes over time as new epidemiological evidence accumulates, in a way that IRIS toxicity values for most industrial chemicals do not.

The exclusive route of concern for PM2.5 risk assessment is inhalation; unlike many IRIS chemicals, PM2.5 is not evaluated for oral or dermal exposure pathways, because ambient particulate exposure occurs through breathing rather than through incidental ingestion of contaminated soil or dermal contact. Within the respiratory tract, particle deposition and the resulting health effects depend heavily on particle size: PM2.5 (and its ultrafine particle subset, particles ≤ 0.1 μm) can penetrate deep into the lower respiratory tract and alveolar region and, for the smallest particles, translocate into systemic circulation, whereas coarser PM10-2.5 particles are deposited predominantly in the upper airway. This size-dependent deposition pattern is a principal reason PM2.5 is regulated separately from, and generally found to pose greater health risk than, the coarser PM10-2.5 fraction, and is discussed further in Section 4.

2. Point-of-Departure Analog: Selection of Key Epidemiological Studies

For a conventional IRIS chemical, Section 2 of this methodology template would identify the single critical animal or human study and specific adverse effect level from which a point of departure (POD) is derived. PM2.5 risk assessment does not use this model. Because PM2.5 epidemiological studies to date have not identified a population-level threshold concentration below which adverse health effects cease to occur — effects have been observed at progressively lower ambient concentrations as study populations, exposure-assessment resolution, and statistical power have improved — EPA’s ISA and NAAQS review process do not select a single “critical study” or POD in the IRIS sense. Instead, the scientific foundation for both the NAAQS and ATSDR’s screening comparison values rests on the cumulative, converging evidence from a series of very large, long-running prospective cohort studies, each contributing a concentration-response (C-R) function — an estimated relative risk or hazard ratio per unit increase in PM2.5 concentration — rather than a discrete no-effect dose.

The scientific foundation for PM2.5 standard-setting is built substantially on three cohort research programs. The Harvard Six Cities Study, an original prospective cohort of approximately 8,111 adults followed from the mid-1970s, first established a statistically significant association between fine particulate air pollution and total mortality across six U.S. cities with differing pollution levels (Dockery et al., 1993); extended follow-up of the same cohort through 2009, with improved exposure assessment and adjustment for individual risk factors, confirmed and refined this association, finding an approximately 14% increase in all-cause mortality per 10 μg/m3 increase in long-term average PM2.5 exposure (Lepeule et al., 2012). The American Cancer Society Cancer Prevention Study II (ACS CPS-II) cohort, comprising approximately 1.2 million adults enrolled beginning in 1982, found that long-term PM2.5 exposure was associated with increased all-cause, cardiopulmonary, and lung cancer mortality, with each 10 μg/m3 increase in PM2.5 associated with an approximately 4%, 6%, and 8% increase in these respective mortality categories (Pope et al., 2002). Following an independent Health Effects Institute reanalysis that confirmed the validity of the original Six Cities and ACS findings using the investigators’ original data (Krewski et al., 2000), a subsequent extended and spatially refined analysis of the ACS cohort with 18 years of follow-up further strengthened confidence in these associations and specifically identified ischemic heart disease as the cardiovascular cause of death most strongly and consistently associated with long-term PM2.5 exposure (Krewski et al., 2009).

A methodologically distinct and more recent line of evidence, directly relevant to setting a health-protective standard at low ambient concentrations, comes from open cohort studies of the full U.S. Medicare population. Di et al. (2017) constructed an open cohort of all 60,925,443 Medicare beneficiaries in the continental United States from 2000 through 2012 (460,310,521 person-years of follow-up), using satellite-derived, ZIP-code-resolution PM2.5 exposure estimates, and found that mortality risk increased with PM2.5 exposure even when the analysis was restricted to person-years with annual exposure below 12 μg/m3 (the NAAQS in effect at the time of the study), with no evidence of a concentration threshold below which risk leveled off; a companion analysis using the same cohort’s hospital admission records found that increasing exposure from below 8 μg/m3 to above 8 μg/m3 was associated with a 15% increase in all-cause admission hazard, an 18% increase in circulatory admissions, and a 21% increase in respiratory admissions, even among enrollees whose exposure never exceeded 12 μg/m3. This finding — that PM2.5-associated risk continues below the lowest concentrations previously studied, without evidence of a threshold — is the central scientific fact that both explains why PM2.5 cannot be assigned a conventional NOAEL-based POD and, together with the studies above, directly supported EPA’s 2024 decision to lower the annual NAAQS to 9.0 μg/m3 (Section 6).

3. Dosimetric Considerations (No Animal-to-Human Dose Conversion)

Section 3 of this template ordinarily addresses animal-to-human and internal-to-external dose conversion — methodology that is not applicable to PM2.5 in the way it is for IRIS chemicals. Because the epidemiological studies underlying PM2.5 risk assessment (Section 2) measure human health outcomes directly in relation to human ambient air exposure, no interspecies extrapolation, allometric scaling, or physiologically based pharmacokinetic modeling is performed or required; the concentration-response functions described above are already expressed in human-relevant units (risk per μg/m3 of ambient PM2.5) and require no cross-species conversion before use. Animal toxicological and controlled human exposure studies do exist in the PM2.5 database, but they serve a different, subordinate function than in IRIS chemical assessments: rather than providing the quantitative point of departure, they provide mechanistic and biological-plausibility support for the epidemiological associations (for example, controlled human exposure studies demonstrating PM2.5-associated changes in heart rate variability, vasoconstriction, and blood hemostatic markers over exposure durations of 1–22 hours), consistent with the EPA ISA’s evidence-integration framework, which weighs epidemiological, controlled human exposure, and animal toxicological evidence together to reach a qualitative causality determination (Section 7) rather than using the toxicological data to derive a quantitative POD.

The one dosimetric concept genuinely analogous to IRIS chemical assessment is regional respiratory tract deposition modeling, which is used qualitatively (rather than to derive a quantitative human equivalent concentration) to explain why PM2.5 is regulated separately from, and is generally considered more health-hazardous per unit mass than, the coarser PM10-2.5 fraction: particles in the PM2.5 size range and smaller are deposited more efficiently in the tracheobronchial and alveolar regions of the lower respiratory tract, where they can persist, translocate across the alveolar-capillary barrier into systemic circulation, and drive the systemic inflammatory and oxidative stress pathways believed to mediate cardiovascular effects, whereas coarser PM10-2.5 particles are deposited predominantly in the upper airway and cleared more readily by mucociliary action. This size-dependent deposition pattern is why ATSDR’s and EPA’s assessment frameworks treat PM2.5 and PM10-2.5 as distinct exposure metrics with separate comparison values (Section 6), rather than combining them into a single PM10 dosimetric category.

4. Uncertainty Characterization (No Traditional Uncertainty Factors Applied)

Section 4 of this template ordinarily reports the specific interspecies, intraspecies, subchronic-to-chronic, LOAEL-to-NOAEL, and database-completeness uncertainty factors (UFA, UFH, UFS, UFL, UFD) applied to derive an IRIS RfD or RfC. None of these default multiplicative uncertainty factors is applied in PM2.5 risk assessment, and RoseyFields’ risk assessors should not attempt to construct or apply them by analogy. This is not an oversight in the regulatory record; it reflects a structural difference in methodology. The IRIS UF system exists specifically to bridge gaps between an animal study and human risk (interspecies), between a study population and the general human population (intraspecies), between a subchronic study and lifetime exposure (subchronic-to-chronic), between a LOAEL and an inferred NOAEL, and to account for missing toxicological studies (database completeness). Because PM2.5’s scientific foundation is built on very large, long-duration human cohort studies (Section 2) that directly measure the general human population, including sensitive subgroups, over exposure periods extending to a decade or more, and because the standard-setting and screening processes described below are not built around identifying and dividing down from a no-effect level, most of the specific gaps the UF system is designed to bridge do not arise in the same form for PM2.5.

Uncertainty in PM2.5 risk assessment is instead characterized and addressed through different mechanisms specific to the epidemiological and regulatory context: (1) statistical uncertainty in each cohort study’s concentration-response function is expressed and reported directly as a confidence interval around the relative risk or hazard ratio estimate (e.g., the 95% CI of 3–10% reported by Di et al. (2017) for the association between crossing the 12 μg/m3 threshold and increased all-cause hospital admission), rather than folded into a single point-estimate POD; (2) uncertainty regarding the shape of the concentration-response relationship at low concentrations, and whether it remains linear down to the lowest observed exposures, is addressed by EPA’s ISA and Policy Assessment through explicit review of the available low-concentration cohort evidence (Section 2) and sensitivity analyses restricting cohorts to below-NAAQS exposure levels, rather than through a default LOAEL-to-NOAEL factor; (3) potential confounding by co-pollutants, weather, socioeconomic status, and smoking is addressed at the level of individual study design and statistical adjustment (and is a central focus of EPA’s per-study confidence evaluation within the ISA’s systematic review process) rather than through a database-completeness-type default factor; and (4) the statutory NAAQS process itself builds a margin of safety directly into the standard-setting decision, as required by Clean Air Act Section 109(b)(1), which requires that primary standards be set at a level “requisite to protect the public health” with “an adequate margin of safety” — a policy-level safety margin conceptually related to, but not calculated the same way as, the multiplicative UF approach used for IRIS chemicals.

ATSDR’s screening framework, described further in Section 6, addresses uncertainty by a different and more direct route still: rather than deriving its own PM-specific comparison value with associated uncertainty factors, ATSDR adopts the World Health Organization’s Air Quality Guidelines (AQGs) as its screening comparison values, explicitly on the basis that AQGs represent “the most health-protective comparison value available” for PM2.5, reflecting a policy choice to use the most conservative available benchmark rather than to independently derive and adjust a value using the standard ATSDR Minimal Risk Level uncertainty-factor methodology applied to other chemicals.

5. Derivation of Health-Based Comparison Values (in Place of an RfD/RfC)

5.1 ATSDR Screening Comparison Values: WHO Air Quality Guidelines

In the absence of an ATSDR-derived Minimal Risk Level, ATSDR’s 2024 guidance directs health assessors to screen PM2.5 (and PM10) data against the World Health Organization’s 2021 Global Air Quality Guidelines (AQGs), selected because they represent the most health-protective PM comparison values currently available from any agency. For PM2.5, the applicable ATSDR screening values are 5 μg/m3 (annual average) and 15 μg/m3 (24-hour average); for PM10, the corresponding values are 15 μg/m3 (annual average) and 45 μg/m3 (24-hour average). These AQGs were derived by WHO through its own independent, global systematic review of the epidemiological cohort literature (substantially overlapping with, but not identical to, the studies EPA’s ISA relies upon), rather than through the toxicokinetic-model/uncertainty-factor methodology ATSDR applies to derive Minimal Risk Levels for other chemicals; no multiplicative uncertainty factors are layered onto the WHO AQG values before ATSDR applies them for screening. Acute (≤24-hour) exposures are screened against the 24-hour AQG, and chronic exposures are screened against the annual AQG; ATSDR guidance further specifies that where only sub-daily (e.g., 1-hour) monitoring data are available, the 24-hour AQG may still be applied as a conservative screening benchmark, because the underlying epidemiological literature has observed effects at concentrations at or below the 24-hour AQG using averaging periods as short as 1–22 hours in controlled-exposure and panel studies.

5.2 EPA’s National Ambient Air Quality Standards

EPA’s own regulatory benchmark for PM2.5, the NAAQS, is set through a different process from either an ATSDR MRL or an IRIS RfD/RfC: rather than dividing an experimentally derived point of departure by default uncertainty factors, EPA’s Office of Air and Radiation, informed by the ISA’s causality and concentration-response evidence synthesis (Section 2) and by quantitative risk and exposure assessments in a companion Policy Assessment, together with the advice of the independent Clean Air Scientific Advisory Committee (CASAC), selects a standard level and averaging time that the EPA Administrator judges requisite to protect public health with an adequate margin of safety, as required by Clean Air Act Section 109(b)(1). As finalized in February 2024, the current primary (health-based) PM2.5 NAAQS is 9.0 μg/m3 (annual average, evaluated as a 3-year average of annual mean concentrations) and 35 μg/m3 (24-hour average, evaluated as the 3-year average of the 98th percentile of daily concentrations); the annual standard was lowered from its prior 2012 level of 12.0 μg/m3 specifically in response to the low-concentration cohort evidence described in Section 2, including the Di et al. (2017) Medicare cohort finding of continued risk below 12 μg/m3.

5.3 Relationship Between the Two Value Sets

RoseyFields’ risk assessors should understand the WHO AQGs and the U.S. NAAQS as related but distinct benchmarks serving different purposes, both grounded in essentially the same body of underlying cohort evidence, but reflecting different institutional processes and, in the WHO’s case, an explicitly more health-protective (lower) numeric target: the WHO AQG annual value (5 μg/m3) is approximately half the current U.S. NAAQS annual value (9.0 μg/m3). ATSDR’s use of the more protective WHO value for public health screening, while EPA continues to use the NAAQS for regulatory attainment and enforcement purposes, is intentional and should be reflected accurately in any RoseyFields risk characterization: a site or exposure scenario can be simultaneously in attainment of the U.S. NAAQS while exceeding the WHO AQG, and ATSDR’s own health assessment guidance treats the latter, more protective comparison as the appropriate screening benchmark for evaluating potential health hazard, independent of NAAQS attainment status.

6. Carcinogenicity: Causality Determination in Place of a Cancer Slope Factor

PM2.5 has no EPA-derived oral cancer slope factor (CSF) or inhalation unit risk (IUR) of the kind IRIS provides for chemicals formally classified under EPA’s 2005 Guidelines for Carcinogen Risk Assessment. Because PM2.5 is regulated as a criteria pollutant under the NAAQS framework rather than assessed individually under the IRIS carcinogen risk assessment process, EPA has not conducted the quantitative dose-response modeling (point-of-departure identification, low-dose extrapolation) that would be required to generate a CSF or IUR, and no such value should be applied in RoseyFields’ PM2.5 cancer risk calculations.

What does exist is a qualitative causality determination: the 2019 ISA and its 2022 Supplement concluded that the evidence supports a “likely to be causal” relationship between long-term PM2.5 exposure and cancer, specifically lung cancer incidence and mortality, based on a body of cohort evidence including the ACS CPS-II findings that each 10 μg/m3 increase in long-term PM2.5 exposure was associated with an approximately 8% increase in lung cancer mortality (Pope et al., 2002), findings that were confirmed and refined in the extended, spatially resolved reanalysis of the same cohort (Krewski et al., 2009), together with supporting mechanistic evidence of PM2.5-associated oxidative stress, systemic inflammation, and genotoxic potential. Independently, the International Agency for Research on Cancer (IARC) has classified outdoor air pollution as a whole, and particulate matter specifically as a major component of that mixture, as Group 1, carcinogenic to humans, based substantially on lung cancer evidence. RoseyFields’ litigation and regulatory work involving PM2.5 and cancer outcomes should rely on this qualitative causality determination, the specific cohort-derived relative risk estimates described above (appropriately adjusted for the individual’s or population’s estimated cumulative exposure relative to the exposure ranges reflected in the underlying cohorts), and the IARC classification, rather than on any quantitative slope factor, which does not exist for this pollutant and should not be estimated or implied by analogy to other IRIS carcinogens.

7. Application in Practice: Screening, Risk Characterization, and Litigation Support

7.1 Data Averaging and Screening Against Comparison Values

Consistent with ATSDR’s three-step process (data averaging, screening, and data evaluation), RoseyFields first averages available PM2.5 monitoring data to the appropriate averaging period — 1-hour, 24-hour, or annual — using the most highly resolved underlying data available, then screens the averaged concentrations against the applicable WHO AQG (Section 5.1): 24-hour or shorter-duration data against the 15 μg/m3 24-hour AQG, and annual data against the 5 μg/m3 annual AQG. Because no numeric hazard quotient methodology (dividing exposure by an RfD/RfC) applies to PM2.5, screening is a direct concentration-to-guideline comparison rather than a calculated ratio; RoseyFields characterizes the result in terms of whether, and to what degree and frequency, monitored or modeled concentrations exceed the applicable AQG, and separately evaluates attainment status against the current NAAQS for any claims requiring assessment of regulatory compliance rather than health-protectiveness.

7.2 Contextualizing Site Data Against National and Regional Trends

Because PM2.5 is a ubiquitous regional and even continental-scale pollutant rather than a site-specific contaminant in the way most IRIS chemicals are, RoseyFields routinely contextualizes site- or case-specific PM2.5 data against EPA’s national monitoring network data (published annually in EPA’s “Our Nation’s Air” trends reports) and against background, upwind, and regional monitors, to help distinguish a source’s specific contribution from broader regional air quality. This contextualization is particularly important for exposure and dose reconstruction in litigation, where establishing the incremental contribution of a specific defendant’s emissions against a substantial regional and background PM2.5 concentration — nationally, the median annual average PM2.5 concentration across U.S. trend sites was approximately 7.7 μg/m3 in 2022, already above the current WHO AQG even before considering any site-specific source — is frequently a central, and contested, element of both general and specific causation analysis.

7.3 Health Impact and Risk Characterization

For quantitative risk characterization beyond simple screening, RoseyFields applies the concentration-response functions described in Section 2 directly — for example, using the Di et al. (2017) or Pope et al. (2002) hazard ratios per 10 μg/m3 increment, appropriately scaled to the estimated incremental PM2.5 exposure attributable to a specific source or scenario, together with applicable baseline mortality or morbidity rates for the affected population — an approach directly analogous to the health impact function methodology (e.g., as implemented in EPA’s BenMAP tool) that EPA itself uses to estimate the public health benefits of NAAQS revisions and other air quality regulations in its own regulatory impact analyses. This C-R-function-based approach, rather than an HQ/HI or excess-lifetime-cancer-risk calculation built on an RfD/RfC/CSF, is the technically appropriate and defensible method for quantifying PM2.5-attributable health risk in both regulatory and litigation contexts, and RoseyFields’ risk assessors should decline requests to force PM2.5 exposure data into an HQ or CSF-based risk calculation framework borrowed from IRIS chemical practice, since doing so would misrepresent both the available science and the methodology actually relied upon by EPA and ATSDR.

7.4 Application in Litigation and Regulatory Support

In litigation and regulatory matters, RoseyFields’ PM2.5 work typically supports general causation opinions by relying on the ISA’s causality determinations (Section 2 and Section 6) together with the specific, large-cohort concentration-response evidence underlying them — evidence that, unlike many IRIS chemicals, is derived overwhelmingly from human data at real-world exposure levels rather than extrapolated from animal bioassays, which can be a substantial evidentiary advantage in establishing that PM2.5 is capable of causing the alleged health effect at the exposure levels at issue. Specific causation analysis requires reconstructing the individual’s or population’s estimated PM2.5 exposure (using ambient monitoring data, dispersion or land-use regression modeling, or, where available, personal exposure monitoring or satellite-derived exposure surfaces of the kind used in Di et al. (2017)), comparing that exposure to the ranges and durations reflected in the underlying cohort studies, and conducting differential-etiology analysis addressing co-pollutant exposure, smoking, and other established risk factors for the outcome at issue. Because PM2.5 exposure is nearly universal and background concentrations are rarely negligible (Section 7.2), specific causation and apportionment analysis for PM2.5 claims typically requires more careful quantitative treatment of incremental versus background exposure than is typical for site-specific IRIS chemical contamination, where background exposure is more often negligible by comparison.

8. Key Uncertainties and Sensitive Populations

8.1 Absence of an Identified Threshold

The most significant scientific uncertainty in PM2.5 risk assessment, and the one with the greatest practical consequence for RoseyFields’ work, is that no population-level threshold of effect has been identified: as discussed in Section 2, the Di et al. (2017) Medicare cohort found continuing excess risk even when analysis was restricted to concentrations below the NAAQS in effect at the time, and no subsequent study has identified a concentration floor below which PM2.5-associated risk disappears. This means that, unlike most IRIS chemicals evaluated against an RfD/RfC (where exposure below the reference value is treated as presenting no appreciable risk), a finding that PM2.5 exposure is below the current NAAQS or even below the more protective WHO AQG does not equate to a finding of zero or negligible risk — it reflects only that the exposure is below the level judged, through a policy process balancing health protection against other statutory and practical considerations, to be the current regulatory or guideline benchmark. RoseyFields’ risk characterizations should state this distinction explicitly rather than implying that sub-NAAQS or sub-AQG exposure is risk-free.

8.2 Exposure Misclassification

Because PM2.5 concentrations vary substantially with location, meteorology, and season, and because most large cohort studies (Section 2) rely on ambient monitoring data, satellite-derived exposure surfaces, or ZIP-code/residential-address-level exposure assignment rather than true personal exposure monitoring, some degree of exposure misclassification is inherent in essentially the entire PM2.5 epidemiological evidence base; this nondifferential misclassification would generally be expected to bias observed associations toward the null rather than to overstate risk, providing some reassurance regarding the direction (though not the precision) of the reported concentration-response functions. RoseyFields’ case-specific exposure reconstruction work should account for this limitation explicitly, particularly where personal or near-source exposure is expected to differ materially from the ambient or residential-address exposure metrics used in the underlying cohort studies.

8.3 Co-Pollutant Confounding and Mixture Effects

PM2.5 rarely occurs in isolation from other combustion-related air pollutants (ozone, nitrogen dioxide, sulfur dioxide, carbon monoxide), and disentangling PM2.5-specific effects from those of co-occurring pollutants is an ongoing methodological challenge addressed with varying degrees of rigor across the underlying cohort literature, through statistical co-pollutant adjustment, two-pollutant models, and, in more recent work, causal-inference methods (e.g., instrumental variable and propensity-score approaches) intended to more directly isolate PM2.5’s independent contribution. As with the mixture-attribution issue RoseyFields addresses for other pollutant classes (e.g., PAHs, discussed in RoseyFields’ companion benzo[a]pyrene methodology page), PM2.5-specific risk estimates should be understood as estimates of PM2.5’s contribution within a real-world, multi-pollutant exposure context rather than as fully isolated single-pollutant effects.

8.4 Sensitive and Highly Sensitive Populations

ATSDR’s guidance formally defines “sensitive” population subgroups for PM exposure purposes as pregnant women, children, and older adults (age 65 and older), and defines a broader “highly sensitive” category encompassing members of these groups, or anyone in the general population, who has a pre-existing respiratory condition (e.g., asthma, chronic obstructive pulmonary disease) or cardiovascular disease. In addition, populations of lower socioeconomic status have been identified as facing both elevated PM2.5 exposure and elevated health risk from a given level of exposure, reflecting greater urban and near-roadway residential exposure, reduced healthcare access, and higher baseline prevalence of respiratory and cardiovascular disease (Pratt et al., 2015); EPA’s own analyses have similarly identified PM2.5 exposure and associated health-risk disparities by race and ethnicity, with Black populations in the United States found to bear disproportionately higher PM2.5 exposure burdens. RoseyFields’ case-specific risk assessments should identify whether any sensitive or highly sensitive subgroup membership applies to the individual or population at issue, since the applicable comparison values and underlying cohort risk estimates described in this methodology page are derived from and intended to be protective of general population risk, and case-specific risk for a sensitive or highly sensitive individual may be materially higher than the general-population estimate would suggest.

8.5 Structural Uncertainty From the Absence of a Traditional Toxicity Value

Finally, RoseyFields’ risk assessors should recognize that the absence of an IRIS-style RfD, RfC, CSF, or IUR for PM2.5 is itself a source of practical uncertainty in litigation and regulatory contexts, distinct from the scientific uncertainties above: opposing experts, courts, and regulators accustomed to HQ/HI and excess-lifetime-cancer-risk frameworks may expect or request PM2.5 risk characterization in that familiar format, and RoseyFields’ risk assessors should be prepared to explain — as this methodology page does — why that framework does not apply to this pollutant, and why the concentration-response-function-based and NAAQS/AQG-screening-based approaches described in Sections 5 and 7 represent the scientifically and methodologically correct alternative, consistent with the approach EPA and ATSDR themselves use, rather than a departure from standard practice.

References