Nitrogen Oxides (NOx/NO2) Risk Assessment and Respiratory Disease

RoseyFields conducts NO2 and NOx risk assessments using the following methodology. This document describes the technical and regulatory basis for how RoseyFields’ environmental scientists and risk assessors evaluate ambient and source-specific NO2/NOx exposures, the federal regulatory framework that governs the assessment of this chemical group, and how the resulting values and analyses are applied in litigation support, regulatory compliance, and exposure reconstruction work.

1. Chemical Identity, Regulatory Background, and Exposure Sources

1.1 Chemical Identity

Nitrogen dioxide (NO2; CASRN 10102-44-0) is a reddish-brown, reactive gas and one of a family of highly reactive gases collectively referred to as oxides of nitrogen (NOx), which also includes nitric oxide (NO) and other nitrogen-oxygen compounds. NO2 forms rapidly in the atmosphere through the oxidation of NO, primarily by reaction with ozone and other atmospheric oxidants. Because NO is converted to NO2 on a timescale of minutes to hours in ambient air, NO2 is used by EPA as the indicator species for the broader NOx group in setting and monitoring compliance with air quality standards.

NO2 does not exist in a stable form in aqueous solution; it reacts essentially instantaneously with water to form nitric and nitrous acids. This chemical behavior has direct regulatory consequences discussed in Section 3 below: EPA’s Integrated Risk Information System (IRIS) does not carry an oral toxicity value for NO2, and the chemical is evaluated exclusively as an inhalation hazard.

1.2 Regulatory Background

NO2/NOx is one of six “criteria pollutants” for which EPA has established National Ambient Air Quality Standards (NAAQS) under Section 109 of the Clean Air Act. EPA first promulgated a primary and secondary annual NO2 standard of 53 parts per billion (ppb) in 1971. That standard was reviewed and retained without revision in 1985 and 1996. In 2010, following its most recent full toxicological and epidemiological review at that time, EPA added a short-term (1-hour) primary standard of 100 ppb, set as the three-year average of the 98th percentile of the annual distribution of daily maximum 1-hour concentrations, while retaining the 53 ppb annual standard. EPA reviewed the primary standard again in 2018 and most recently completed action on the secondary (welfare-based) standard in 2024; in both instances, the existing standards were retained without revision. The 2010 rule and EPA’s addition of the 1-hour standard were upheld by the D.C. Circuit in American Petroleum Institute v. EPA, 684 F.3d 1342 (D.C. Cir. 2012), cert. denied, 133 S. Ct. 1724 (2013).

Because NO2/NOx is regulated primarily through this NAAQS mechanism rather than through a chemical-specific IRIS toxicity value, RoseyFields’ methodology for this chemical group necessarily differs in structure from methodologies used for chemicals with an EPA-derived reference dose (RfD), reference concentration (RfC), or cancer slope factor. Section 3 explains this distinction and the framework RoseyFields applies in its place.

1.3 Exposure Sources and Pathways

NO2 and NOx are combustion byproducts, formed when fuel is burned at high temperature in the presence of atmospheric nitrogen and oxygen. Relevant sources encountered in RoseyFields’ casework include:

• On-road and off-road mobile sources — diesel and gasoline engines, including trucks, locomotives, and construction equipment;
• Stationary combustion sources — industrial boilers, process heaters, turbines, incinerators, and flares;
• Electric utility generation — coal-, oil-, and gas-fired power plants;
• Residential and commercial combustion — natural gas appliances, including unvented or poorly vented cooking and heating equipment; and
• Secondary formation — NOx emitted from any of the above sources acts as a precursor to ground-level ozone and secondary fine particulate matter (PM2.5), extending its relevance beyond direct NO2 inhalation exposure.

Inhalation is the only exposure pathway of toxicological concern; because NO2 reacts with water essentially instantaneously, dermal and oral/ingestion pathways are not evaluated. Because ambient NO2 concentrations are dominated by traffic and near-source combustion, near-road and near-facility populations, and populations living or working in enclosed spaces with gas combustion appliances, generally experience the highest exposures and are a recurring focus in RoseyFields’ exposure reconstructions.

2. Regulatory Framework Applied: Why NO2/NOx Is Not Assessed Using an RfD/RfC/UF Architecture

For most non-carcinogenic chemicals, EPA (and RoseyFields, following EPA’s approach) derives a point of departure (POD) from a single critical study, applies a series of uncertainty factors (UFs) to that POD, and arrives at an RfD or RfC that is then compared to estimated exposure to calculate a hazard quotient (HQ). NO2 is a documented exception to this paradigm, and RoseyFields’ methodology reflects that exception rather than forcing NO2 into a framework EPA itself has not applied to this chemical:

• No RfD. EPA withdrew the oral RfD for NO2 from IRIS in 1994. The withdrawal was not a data-quality decision but a chemistry-based one: NO2 does not exist per se in water because it reacts instantaneously with water to form nitric and nitrous acids. Toxicity values for the resulting nitrate and nitrite ions are addressed separately on their own IRIS files and are outside the scope of an NO2/NOx assessment.
• No RfC. EPA has stated that an inhalation RfC will not be derived for NO2 because a NAAQS is already available and serves that risk-management function. IRIS defers entirely to the NAAQS process for this chemical.
• No IRIS carcinogenicity classification. IRIS lists the carcinogenicity assessment for NO2 as “not evaluated.”

Because IRIS explicitly channels the health-protective analysis of NO2/NOx into the NAAQS process, RoseyFields’ methodology substitutes EPA’s NAAQS-development framework for the POD-selection, PK-modeling, and UF-derivation steps used with an RfD/RfC chemical. That framework has three components, developed sequentially for each periodic NAAQS review:

• Integrated Science Assessment (ISA) — a comprehensive synthesis of the epidemiological, controlled human exposure, and toxicological literature, in which EPA reaches formal causal determinations (e.g., “causal,” “likely to be causal,” “suggestive but not sufficient,” “inadequate,” or “not likely to be causal”) for each health endpoint and exposure duration;
• Risk and Exposure Assessment (REA) — a quantitative analysis of population exposures and health risks at current and alternative standard levels, using ambient monitoring data, exposure modeling, and concentration-response functions drawn from the key epidemiological studies identified in the ISA; and
• Policy Assessment (PA) — a staff evaluation that bridges the science in the ISA/REA to the policy judgments the EPA Administrator must make under the Clean Air Act’s requirement that primary standards protect public health with an adequate margin of safety.

The most recent NO2 ISA (EPA/600/R-15/068, finalized January 2016, addressing the primary/health-based standard) found a causal or likely-causal relationship between short-term NO2 exposure and respiratory effects, particularly asthma exacerbation, and found long-term NO2 exposure likely causal for respiratory effects including the development of asthma in children. It classified NO2’s relationship to cardiovascular and metabolic effects, birth outcomes, and (for long-term exposure) cancer as “suggestive, but not sufficient to infer a causal relationship” — a lower confidence tier than the causal/likely-causal findings for respiratory endpoints, and one RoseyFields treats accordingly in any case-specific weight-of-evidence discussion.

3. “Point of Departure” Equivalent: The Epidemiological and Controlled Human Exposure Evidence Base

Rather than a single critical study and POD, the current primary NO2 NAAQS rests on a weight-of-evidence synthesis across (a) controlled human exposure (chamber) studies of asthmatic volunteers and (b) large population-based epidemiological studies, principally time-series and case-crossover analyses of emergency department visits and hospital admissions for asthma and other respiratory outcomes. RoseyFields relies on the same category of studies EPA identified as key evidence in the ISA and REA when it needs to characterize concentration-response relationships for NO2 in litigation or regulatory work, rather than treating any single study as a POD in the RfD/RfC sense.

Representative studies in this evidence base, several of which RoseyFields has cited directly in expert work, include:

• Zheng et al. (2021), a systematic review and meta-analysis of short-term ozone, NO2, and SO2 exposure and asthma-related emergency department visits and hospital admissions, which found the certainty of evidence for an association with 24-hour NO2 exposure to be high — among the more recent and most frequently cited quantitative syntheses of the NO2-asthma relationship;
• Peel et al. (2005), a large time-series analysis of ambient air pollution, including NO2, and respiratory emergency department visits in Atlanta, Georgia, which is among the U.S. multi-pollutant time-series studies EPA drew on in developing NO2 concentration-response functions for its regulatory impact and risk/exposure analyses; and
• Schildcrout et al. (2006), an eight-city panel analysis of ambient air pollution, including NO2, and asthma exacerbation in children with asthma, another study identified by EPA as part of the key evidence base for the asthma-exacerbation endpoint underlying the current NO2 NAAQS.

These studies, and others like them summarized in the ISA, are epidemiological rather than experimental-animal studies, so the RfD/RfC methodology’s cross-species extrapolation and internal-versus-external dose modeling steps — ordinarily required to translate an animal POD to an equivalent human exposure — are not part of the NO2/NOx methodology. Human epidemiological and controlled human exposure data provide direct human concentration-response information, and EPA’s REA converts that evidence into risk estimates at ambient (external, human) concentrations without an interspecies dosimetry step. RoseyFields follows the same approach: where a case requires characterizing dose-response, RoseyFields relies on human-based concentration-response functions drawn from this literature rather than constructing an animal-to-human PK/PBPK extrapolation, which EPA itself did not use in deriving the NO2 NAAQS.

4. Uncertainty Factors and the NAAQS “Adequate Margin of Safety” Standard

For RfD/RfC chemicals, uncertainty is expressed through discrete, multiplicative uncertainty factors (typically interspecies, intraspecies, subchronic-to-chronic, LOAEL-to-NOAEL, and database-completeness factors) applied directly to the POD. Because NO2’s standard is not derived from a single-study POD, this UF architecture is not part of EPA’s NO2 methodology, and RoseyFields does not apply UF values to a POD for this chemical group; doing so would misrepresent both EPA’s own approach and the nature of the underlying evidence.

Uncertainty and protectiveness are instead addressed by the Clean Air Act’s statutory standard itself. Section 109(b)(1) requires primary NAAQS to be set at a level “requisite to protect the public health” with “an adequate margin of safety.” EPA operationalizes this margin of safety qualitatively and quantitatively across several dimensions considered together in the ISA/REA/PA process, rather than as sequential numeric multipliers:

• Selection of at-risk and susceptible populations (principally children and individuals with asthma) as the population for which protection is evaluated, rather than a healthy general-population average;
• Use of the full distribution of ambient exposure and risk (via the REA’s exposure and risk modeling) rather than a single central-tendency estimate;
• Consideration of the coherence and consistency of effect estimates across multiple independent epidemiological studies and study designs, and across controlled human exposure data, in place of a single-study NOAEL/LOAEL determination; and
• The averaging-time and statistical-form design of the standard itself (e.g., the 98th-percentile form of the 1-hour standard, rather than a simple maximum or mean), which is itself a built-in protectiveness mechanism against short-term peak exposures that a chronic-only standard would not capture.

When RoseyFields characterizes uncertainty in an NO2/NOx assessment, it addresses these same dimensions narratively and, where the case supports it, quantitatively (e.g., sensitivity analyses on modeled versus monitored concentrations, or on which epidemiological concentration-response function is applied) rather than presenting a table of UF values, since no such table exists in EPA’s own NO2 methodology.

5. The NAAQS Values RoseyFields Applies as the Health-Protective Benchmark

In place of a derived RfD/RfC number, RoseyFields uses the current primary NAAQS for NO2 as the health-protective benchmark against which modeled or monitored concentrations are compared:

• Annual standard: 53 ppb (0.053 ppm), annual arithmetic mean, first established in 1971 and retained in every subsequent review; and
• 1-hour standard: 100 ppb, expressed as the three-year average of the 98th percentile of the annual distribution of daily maximum 1-hour concentrations, established in 2010 and retained in the 2018 review.

Both are enforceable, health-based, national standards that EPA has determined — after the ISA/REA/PA process described above — are requisite to protect public health, including at-risk populations, with an adequate margin of safety. RoseyFields treats these values as functionally analogous to an RfC for purposes of screening-level comparison, while documenting for the record that they are derived through the NAAQS process rather than the RfD/RfC/UF process, since the distinction matters for how a value is defended in litigation or regulatory review.

6. Carcinogenicity

IRIS does not carry a carcinogenicity classification for NO2; the assessment is listed as “not evaluated,” and no cancer slope factor or inhalation unit risk value exists in IRIS for this chemical. The 2016 ISA characterized the evidence for a relationship between long-term NO2 exposure and cancer as “suggestive, but not sufficient to infer a causal relationship” — a category below “likely to be causal” and well below “causal.” RoseyFields does not apply a quantitative excess lifetime cancer risk (ELCR) calculation for direct NO2 exposure in its assessments, consistent with the absence of an EPA-derived unit risk value and the ISA’s own characterization of the cancer evidence as inconclusive.

Where a matter involves combustion sources that emit NOx alongside other pollutants with established cancer potency values (for example, polycyclic aromatic hydrocarbons or specific metals in a stack emissions profile), RoseyFields addresses those pollutants’ cancer risk separately under their own IRIS or California OEHHA toxicity values; NOx itself is not carried forward into that calculation as a carcinogen. RoseyFields does, however, account for NOx’s role as an ozone and secondary PM2.5 precursor when a case involves regional or cumulative air quality impacts, since ozone and PM2.5 carry their own, separately derived health-effects evidence bases and, for PM2.5, quantitative risk relationships that are distinct from the NO2 analysis itself.

7. Application in RoseyFields’ Practice

7.1 Screening and Exceedance Analysis

Because no RfC/HQ framework applies, RoseyFields’ standard screening step for NO2/NOx is a direct comparison of modeled or monitored ambient concentrations to the applicable NAAQS form (annual mean versus 53 ppb; 98th-percentile 1-hour value versus 100 ppb), rather than a hazard quotient (modeled or measured concentration ÷ RfC). Where a site or source is subject to a state or federal air permit, RoseyFields also compares results to any applicable state ambient air quality standard or permit limit, which for NO2 is typically set equal to or more stringent than the NAAQS.

7.2 Dispersion Modeling and Source Attribution

For source-specific matters, RoseyFields uses EPA’s AERMOD dispersion model to estimate ground-level NO2 concentrations attributable to a specific facility, stack, or fleet of mobile sources, incorporating stack parameters, emission rates, meteorological data, and terrain, and applying EPA’s Tier 1–3 NO2 conversion methodologies (e.g., the Ambient Ratio Method or Ozone Limiting Method) to convert modeled NOx emissions to the NO2 fraction relevant to the NAAQS comparison. Modeled concentrations are added to representative background monitoring data before comparison to the NAAQS, consistent with EPA’s modeling guidance. This modeling output supports source attribution and apportionment questions common in litigation — for example, distinguishing a specific defendant’s contribution to ambient NO2 from regional background and other sources.

7.3 Exposure and Dose Reconstruction

In historical exposure reconstructions, RoseyFields combines available monitoring data, dispersion modeling, and, where relevant, indoor combustion-source estimates (for example, unvented gas appliance use) to reconstruct a plaintiff’s or population’s historical NO2 exposure profile. Because inhalation is the only relevant pathway, this reconstruction focuses on ambient and microenvironmental air concentrations and time-activity patterns rather than dietary or dermal contact pathways.

7.4 Litigation and Regulatory Support

In litigation and regulatory contexts, RoseyFields’ NO2/NOx work typically supports: (a) rebuttal or affirmative expert opinions on whether a facility’s emissions caused or contributed to a NAAQS exceedance; (b) evaluation of a defendant’s compliance history against permit limits and the NAAQS; (c) apportionment analyses distinguishing a specific source’s contribution from background and third-party sources; and (d) characterization of the strength and consistency of the epidemiological evidence linking NO2 exposure to the health outcomes at issue in a case, drawing on the same ISA-identified study base described in Section 3. Because the NAAQS is an established, judicially-tested regulatory benchmark (see American Petroleum Institute v. EPA, discussed in Section 1.2), RoseyFields’ reliance on it, rather than on a constructed RfD/RfC-style value that EPA itself declined to derive, is generally more defensible on Daubert-type challenges to methodology than an ad hoc toxicity value would be.

8. Key Uncertainties and Sensitive Populations

RoseyFields’ NO2/NOx assessments account for the following recurring sources of uncertainty and populations of heightened concern:

• Co-pollutant confounding — NO2 is emitted alongside PM2.5, ozone precursors, and other traffic- or combustion-related pollutants, and epidemiological effect estimates for NO2 can be difficult to fully separate from these co-pollutants; the 2016 ISA specifically notes that recent studies strengthen evidence that short-term respiratory effects of NO2 are independent of other traffic-related pollutants, but the degree of independence remains an active area of study and a frequent point of dispute in litigation.
• Exposure measurement error — ambient monitors may not fully capture near-source, near-road, or indoor microenvironmental concentrations relevant to a specific plaintiff population, which is why RoseyFields pairs monitoring data with AERMOD-based dispersion modeling wherever site-specific exposure reconstruction is required.
• Susceptible and at-risk populations — children, older adults, and individuals with pre-existing asthma or other respiratory disease are consistently identified in the ISA as experiencing disproportionate risk from a given ambient NO2 concentration; RoseyFields flags these populations specifically when a case involves schools, residential areas near roadways, or occupational cohorts with elevated baseline respiratory disease prevalence.
• Indoor and near-road microenvironments — because unvented gas combustion appliances and near-roadway locations can produce short-term concentrations well above area-wide ambient levels, RoseyFields evaluates whether case-specific exposure occurred predominantly outdoors, indoors, or near a major roadway before selecting the appropriate exposure metric.
• Evolving causal determinations — EPA’s causal-determination framework is periodically revised as the epidemiological and toxicological literature grows (for example, the upgrade of several endpoints between the 2008 and 2016 ISAs); RoseyFields tracks the current ISA and any newer peer-reviewed literature published since the last EPA review when a case requires characterizing the present state of the science rather than the state of the science as of a specific EPA rulemaking date.

References