Human Health Risk Assessment for Ambient Nitrogen Dioxide
This Human Health Risk Assessment for Ambient Nitrogen Dioxide (NO2) is a comprehensive review of the most relevant health- and exposure-related science for this air pollutant, prepared by the Air Quality Assessment Section of Health Canada. This review is intended to support the development of Canadian Ambient Air Quality Standards (CAAQS) for NO2; these standards are one of the elements of the federal-provincial-territorial Air Quality Management System.
Nitrogen oxides (NOx) are emitted predominantly from combustion sources. Most emissions of NOx are as nitric oxide (which is rapidly converted to NO2), along with lesser quantities of NO2 itself. Based on the National Pollutant Release Inventory (NPRI), in 2011 the major ambient releases of NOx in Canada were from mobile sources (50% of total emissions), mostly from off-road and on-road diesel engines. Substantial amounts were also emitted from industrial sources (30%), the majority from upstream oil and gas. Lesser quantities were released by the non-industrial category (12%, most of this from combustion-generated electrical power), and by natural sources (7%, the majority from microbial activity in fertilized agricultural soils). The trend in NOx emissions as reported to the NPRI from 1985 to 2011 has been generally downward for mobile, non-industrial, and natural sources, whereas industrial emissions have increased over the same time period, largely because of emissions from the upstream oil and gas sector.
Information on ambient NO2 concentrations in Canada is provided primarily by the National Air Pollution Surveillance (NAPS) network of monitoring stations. Data from this network indicate that NO2 levels display marked variations in space and time on several scales, often reflecting the important influence of traffic emissions on exposure.
With respect to spatial variation, the highest concentrations of ambient NO2 occur at transportation- and potentially industrial source-influenced sites. NO2 levels at other NAPS site types are lower and appear proportional to the degree of urbanicity, probably a function of the parallel changes in emissions from mobile sources, residential heating, and other population-related sources. Various concentration metrics (daily 1-hour maximum (1-h max), 24-hour average (24-h avg), and annual average) demonstrate similar relationships in this regard. There is also large spatial variability in NO2 in relation to markers of traffic emissions, including distance from roads, traffic volumes, and road length.
Concerning temporal variation, both daily 1-h max and annual avg ambient NO2 concentrations at various NAPS site types nationwide decreased steadily between 1997 and 2011, attributable to NOx-specific regulatory controls on the mobile sector and fossil-fueled electric power generation. All site types also exhibited a common pattern by season, with wintertime maxima and summertime minima, the latter consistent with increased mixing heights and photochemical oxidation of NO2 and decreased emissions from residential heating compared with winter. Concentrations of ambient NO2 also vary throughout the day, with two peak concentrations corresponding to morning and afternoon/evening rush hours. NO2 levels on the weekend are generally lower than those on weekdays and the diurnal peaks are shorter on weekends, likely as a combined result of reduced traffic (especially diesel truck traffic) and the lack of rush hour traffic on weekends.
Exposure to NO2 from Ambient Sources
The entire population is exposed to NO2 originating from ambient sources, both when people are outdoors and when they are in indoor environments into which ambient NO2 has infiltrated. As they go through their day, some people also spend time in locations that have higher NO2 concentrations as a result of releases from non-ambient sources (e.g. indoors in homes with gas stoves).
This assessment is being conducted to support the development of an ambient standard for NO2, and is based in large part on the extensive epidemiological evidence linking ambient concentrations of NO2 to a wide range of health effects. In this context, a key issue is the ability of NO2 concentrations measured by the monitoring network to serve as an indicator of personal exposure to NO2 of ambient origin, as opposed to the total personal exposure to NO2 from all sources that is measured in most exposure assessment studies.
Studies of the relationship between personal exposures to NO2 and concentrations measured by ambient monitoring networks have generally shown positive and often statistically significant correlations or regressions between short-term ambient concentrations and total personal exposures. Usually the ambient component of personal exposure to air pollutants is not directly measurable, but the total personal exposure can be regarded as the personal exposure of ambient origin if there are no indoor sources. In those studies where indoor sources of NO2 were absent, the correlation of personal exposure with ambient concentrations was moderate to strong, and was increased 2- to 3-fold compared with that observed in the presence of indoor sources. In addition, the association between total personal exposures and ambient NO2 was greater in the warm season (when people are generally more exposed to ambient air pollutants because building infiltration and time spent outdoors are greater) in a number of studies. These findings were confirmed in a recent meta-analysis of a large number of exposure assessment studies.
Overall, the results of these studies indicate that, although the concentrations measured by the ambient monitoring network may not account for differences between individuals in exposure to NO2 of ambient origin, they appear to be a reasonable surrogate for exposure at a population level. In addition, day-to-day variations in exposure of the population to NO2 of ambient origin are likely to track changes in the concentrations measured at a central site/sites. It is these variations over time and the ability to represent population average personal exposure, rather than the absolute magnitude of the exposure itself, that are the basis for the associations between ambient NO2 levels and the health effects reported in short-term epidemiological studies. Therefore, ambient concentrations are a useful and appropriate exposure measure for epidemiological studies of the health effects of NO2 air pollution.
Traffic as a Source of Exposure to NO2
Vehicle emissions are an important source of NO2 in urban environments. Large horizontal gradients in NO2 concentrations near major roadways have been observed in a number of studies; levels on or near roads or in vehicle cabins were several times greater than urban background levels in these studies. Traffic variables are also often significant predictors of ambient NO2 in land use regression models, and of personal exposure to NO2. Near-road NO2 also displays a negative vertical gradient, with concentrations being increased nearer to the road surface.
Localized emission from roadway sources leads to variability in NO2 concentrations that is not captured by the existing regional air quality monitoring network. This variation affects population-level exposure estimates and adds exposure measurement error to epidemiology studies that rely on ambient concentrations as indicators of exposure. Elevated concentrations of NO2 on or near roadways also increase the exposure of anyone who spends substantial amounts of time in such locations.
Correlations of Other Pollutants with Ambient NO2
The associations between ambient NO2 and co-pollutants released from the same sources need to be considered in interpreting the results of epidemiological studies of NO2-related health effects. Causal attribution to NO2 is challenging because epidemiological associations can potentially reflect correlations with other pollutants rather than true causal association with NO2. In most studies, ambient levels of NO2 are moderately to strongly correlated with traffic-related pollutants such as carbon monoxide (CO) and fine particulate matter (PM2.5), and less so with pollutants that are more regional in nature (e.g. ozone (O3)) or that originate from other sources (e.g. sulphur dioxide (SO2)). Correlations are also moderate between personal NO2 and ambient or personal exposure to other combustion-related pollutants within urban areas, most notably CO, PM2.5, polycyclic aromatic hydrocarbons, certain volatile organic compounds such as benzene and 1,3-butadiene, and PM constituents such as elemental carbon (EC) and organic carbon (OC).
General Approach to Assessing Weight of Evidence for Health Effects
The health effects of NO2 have been extensively examined in a very large number of studies, including epidemiological studies of health effects associated with NO2, controlled human exposure studies in volunteers exposed to NO2 in experimental chamber studies, and toxicology studies of animals exposed to NO2 in the laboratory.
In this assessment, epidemiological studies of ambient NO2 have been weighted more heavily than animal toxicological or controlled human exposure studies for several reasons: 1) epidemiological studies provide the most direct approach for assessing the health effects of "real world" complex mixtures of air pollutants to which people are exposed; 2) human populations are highly heterogeneous as compared with laboratory animal populations and encompass a large range of susceptibilities, disease/illness status and exposures; and 3) no species extrapolation is necessary.
However, the results from animal toxicological studies and especially controlled human exposure studies are still quite relevant and shed light on results from epidemiological studies, particularly with respect to pathophysiological mechanisms underlying observed effects.
In addition, the epidemiology studies are observational rather than experimental, and hence there can be uncertainty as to whether the effects reported in the epidemiology studies are in fact due to ambient NO2 alone. The NO2 may be a marker (in whole or in part) for other air pollutants, or the observed association may even be the result of some other factor.
To evaluate the weight of evidence that the epidemiological associations between health outcomes and ambient NO2 are causal, it is necessary to examine the various lines of evidence in combination and to assess the collective evidence using established criteria for causal determination. In this assessment, the evidence for various categories of health outcomes is reviewed in an integrated fashion, by reporting together the findings from the available epidemiological, controlled human exposure, and/or animal toxicological studies. This collective evidence is then evaluated for various categories of health outcomes in light of considerations that have traditionally been used to form judgments as to whether the observed associations are causal; likely to be causal; suggestive, but not sufficient to infer a causal relationship; etc.
These considerations include:
- the strength of association, including the magnitude and precision of the risk estimates and their statistical significance;
- the robustness of the associations to model specifications and adjustment for potential confounders such as weather, temporal trends, and co-occurring pollutants;
- the consistency of reported associations across studies and study designs conducted by different researchers in different locations and times;
- the coherence of the relationship between exposure to NO2 and related endpoints within and across animal toxicology, controlled human exposure, and various types of epidemiological studies; and
- the biological plausibility of the associations in light of what is known regarding NO2 dosimetry and the types of effects observed and the associated potential mechanisms of action, based largely on animal toxicology and controlled human exposure studies.
Short-term Respiratory Effects
In short-term controlled studies of asthmatic adults, exposure to near-ambient levels of NO2 elicited a range of adverse respiratory effects, including decreased lung function, increased airway hyperresponsiveness (AHR), and airway inflammation. Most of these effects, as well as increases in asthma-related respiratory symptoms, were also associated with ambient NO2 in epidemiological studies of asthmatic children. Respiratory symptoms in asthmatic children were also related to indoor NO2 in several epidemiological studies, and interventions to reduce NO2 from gas appliances in classrooms decreased respiratory symptoms. The mechanisms by which these effects occur have been investigated in both humans and animals and provide biologically plausible pathways for these effects.
Ambient NO2 concentrations were significantly and independently associated with increased respiratory and asthma hospitalizations and asthma emergency room visits (ERVs) in numerous population-based epidemiology studies. These findings are strongly coherent with the experimental and epidemiological evidence for lung function decrements, increased respiratory symptoms, airway inflammation, and increased AHR in asthmatics, and they provide an indication of the public health impacts at a population level arising from the effects on the airways seen in experimental and epidemiological studies.
Thus several lines of evidence indicate that ambient NO2 is associated with asthma exacerbations. The epidemiological associations with asthma-related endpoints exhibit strength of association, consistency, robustness, and coherence. In conjunction with the experimental findings in animals and humans, the overall evidence indicates that there is a causal relationship between short-term exposure to ambient NO2 at current levels and increased asthma-related morbidity (including airway inflammation and AHR, increases in respiratory symptoms, and asthma hospitalizations and ERVs).
Short-term Cardiovascular Effects
In population-based epidemiological studies, there were consistent and significant associations of ambient NO2 with increased cardiovascular mortality, hospitalizations, and ERVs, but these morbidity outcomes were often also related to other pollutants. In addition, the NO2-related risks were often attenuated by adjustment for co-pollutants or no co-pollutant models were conducted.
In some panel studies and controlled human exposure studies, there were NO2-related decreases in heart rate variability, changes in ventricular repolarization, and increases in inflammatory and/or coagulatory biomarkers. However, the findings in this small dataset were somewhat inconsistent, and the spectrum of NO2-related cardiovascular effects has not been well characterized.
Given the questions surrounding the independence of the NO2-related effects and the limited supporting data, the overall evidence is suggestive, but not sufficient to infer a causal relationship between short-term exposure to ambient NO2 and cardiovascular effects.
Mortality Related to Short-term Exposure
In numerous epidemiological studies of various designs, short-term ambient NO2 was independently associated with increases in total non-accidental, cardiopulmonary, cardiovascular, and respiratory mortality. These associations were observed in cities from various regions of the world, encompassing different climatic regimes, pollutant mixes, and socioeconomic conditions. However, the coherence of the epidemiological findings with respect to NO2-related morbidity that could give rise to mortality from cardiovascular and respiratory causes is somewhat limited, though there is evidence for several elements in the sequences of events that could give rise to increased cardiovascular and respiratory mortality. There are also indications of plausible (albeit non-specific) mechanisms of action for mortality from major cardiovascular and respiratory causes (myocardial infarction, chronic obstructive pulmonary disease (COPD), and respiratory infections).
Overall, the associations with total non-accidental, cardiopulmonary, and to a lesser extent cardiovascular and respiratory mortality display strength of association, consistency, and robustness, but lack some aspects of coherence; it is concluded that there is likely a causal relationship between short-term exposure to ambient NO2 at current levels and these categories of mortality.
Long-term Respiratory Morbidity
In epidemiological studies, long-term exposure to ambient NO2 was associated with adverse respiratory effects, especially in children, including reduced measures of lung function and reduced lung function growth. In children, several cohort studies also showed relationships between long-term exposure to NO2 and the development of asthma and/or allergic responses. Long-term exposure to NO2 levels appears to increase the incidence of asthma in adults as well. However, some uncertainty remains about the possible role of other co-occurring pollutants in the NO2-related respiratory effects.
The epidemiological associations with respiratory health endpoints exhibit consistency, strength of association, and coherence across disciplines, as well as some indication of robustness and biological plausibility. However, considering the questions surrounding the possible role of co-pollutants, the overall evidence indicates that there is likely a causal relationship between long-term exposures to current levels of ambient NO2/NOx and respiratory effects related to the development of asthma or allergic-related disease.
Other Long-term Effects
Overall, the limited available evidence is suggestive, but not sufficient to infer a causal relationship between long-term exposure to ambient NO2 and each of cardiovascular effects, cancer and related effects, mortality, and reproductive and developmental endpoints. For each of these, there are significant ambient NO2-related associations in some epidemiology studies, but the database is lacking in a number of respects, and more research is needed.
A number of other emerging NO2-related effects warrant further examination, including those on the central nervous system and on other morbidity outcomes (diabetes, appendicitis, inflammatory bowel disease, otitis media, osteoporosis and rheumatoid arthritis) to determine whether such effects are consistently observed and occur at relevant concentrations. The emerging evidence that polymorphisms in some genes can influence the association between air pollutant exposures and morbidity effects indicates a potential role for such genetic effect modification in some at-risk populations, and additional research in this area would contribute to a fuller understanding of these gene-environment interactions.
Potential Confounding by Co-pollutants
The presence of other co-pollutants, especially those that arise from the same sources as NO2, such as traffic, complicates the interpretation of the results of epidemiological studies of NO2-related health effects. It also makes causal attribution to NO2 challenging because epidemiological associations can potentially reflect correlations with other pollutants rather than true causal association with NO2. For those health effects for which the weight of evidence is relatively strong (i.e. causal or likely to be causal), NO2-related risks were generally robust to adjustment for co-pollutants. This was observed most often in models with common air pollutants including PM10, O3, and SO2. However, effect estimates for NO2 were also generally not sensitive to adjustment for traffic-related pollutants including CO, PM2.5, and (in a small number of studies) NO, ultrafine particles, EC/BC or particulate metals, though traffic-related pollutants have not been extensively studied in this regard.
Subgroups with Increased Sensitivity or Exposure to Ambient NO2
Individuals with certain pre-existing diseases appear to be sensitive to exposure to ambient NO2. Several lines of evidence from controlled human exposure and epidemiological studies indicate that asthmatics are a susceptible subgroup. There is some evidence (albeit more limited than for asthma) indicating that people with COPD also appear to be more sensitive to NO2.
Age is also clearly related to susceptibility. The results of epidemiological studies indicate that children, especially asthmatics, are more at risk of respiratory health outcomes from both short- and long-term exposure to NO2. Older adults appear to be more sensitive to short-term effects of NO2 on respiratory hospital admissions, ERVs and other medical visits, as well as all-cause and respiratory mortality. Older adults also had increased risks for cardiovascular mortality and morbidity in epidemiological studies.
Concentrations of ambient NO2 are increased near local sources, especially in on-road, near-road, and in-vehicle microenvironments for roadways with heavy traffic. People who spend substantial amounts of time in such locations can have elevated exposures to NO2. These would include people who spend a long time in vehicles commuting or during the course of their work (e.g. truck drivers), who work or commute in proximity to major roadways (e.g. roadway construction workers, cyclists), or who reside, work, attend school, etc. in buildings near such roadways.
People engaged in vigorous physical activity would also inhale greater amounts of NO2.
Implications of Exposure Measurement Error
The relationship between ambient concentrations and personal exposure to NO2 of ambient origin will vary as a result of the influence of a number of factors, including spatial and temporal variability in NO2 concentrations, time-activity patterns, building ventilation, and perhaps measurement artifacts and analytical methods. The influence of these factors results in exposure measurement error and potential bias in the risk estimates of epidemiology studies that are based on ambient concentrations. The bias can be either upward or downward, though it is expected to most often underestimate risks and make it more difficult to detect a health effect. Several studies performed in Atlanta, GA, investigated the potential bias from using fixed area monitors on the resulting estimates of short-term risk for cardiovascular disease ERVs. Their results indicated that the spatial heterogeneity of air pollutants was a much greater source of measurement error than instrument imprecision. For NO2, most results suggested that this measurement error markedly attenuated the risk estimates, sometimes even resulting in a loss of statistical significance.
Therefore, this source of uncertainty should not change one of the principal conclusions of this assessment, based largely on epidemiological studies, that several categories of adverse health effects are consistently and independently associated with ambient NO2 concentrations.
Public Health Impacts
The effects associated with NO2 have been observed in epidemiological studies in Canada and in other countries at NO2 concentrations that occur in Canada, well below existing ambient air quality objectives and standards. For those health outcomes for which the weight of evidence and statistical power are greatest (i.e. mortality, respiratory/asthma hospitalizations and asthma-related ERVs for short-term ambient NO2 exposure, and respiratory morbidity for long-term ambient NO2 exposure), the mean or median ambient levels at which effects are observed overlap those measured at all NAPS site types, ranging from non urban to transportation- and potentially industrial source-influenced sites. Therefore, adverse health effects in epidemiological studies are occurring at ambient NO2 concentrations that are commonly experienced in Canada.
In most of the studies that examined the shape of the concentration-response relationship for short-term NO2-related mortality or medical visits, there was an approximately linear relationship, with no clear evidence of a threshold. Overall, the current evidence indicates that if a general population threshold exists for the health effects of NO2, it is likely to be near the lower limit of ambient NO2 concentrations. Consequently, the available evidence indicates that any increment in concentrations of ambient NO2 presents an increased risk for serious health effects, up to and including mortality.
Although the risks for ambient NO2-related health effects are relatively small by traditional epidemiological standards, the entire population is exposed, and the subpopulations that have increased sensitivity or exposure to NO2 (including children, older adults, individuals with asthma or COPD, people engaged in vigorous physical activity, and those spending substantial amounts of time near major roadways) comprise a considerable proportion of the population. In addition, the health impacts that have been the focus of most assessments, including mortality, hospitalizations, and ERVs, represent serious outcomes. Further, these are just the "tip of the iceberg" in the pyramid of health effects associated with ambient NO2, and the unmeasured morbidity also has important public health impacts and costs. As a result, the public health impacts of ambient NO2 are substantial and are expected to remain important as the population ages and the pool of older adults increases, especially given the higher underlying death and disease rates in this age group.
With respect to the durations of exposure that are associated with health effects, the types of health effects, the estimated risks for these effects, and the consistency of the findings are much the same for daily 1-h max and 24-h avg ambient NO2. There is also some indication of the same kinds of health effects for other sub-daily averages (e.g. 3-h). These similarities are not unexpected, given that these various short-term exposure metrics are highly correlated with one another. In addition, the overlap between ambient levels in Canada and the concentrations associated with health effects in the epidemiological studies is similar for daily 1-h max, 24-h avg, and even annual or longer-term average NO2. In short, the information on health effects associated with ambient NO2 does not itself provide strong support for any one short-term exposure metric over the other as the basis for the form of the CAAQS. However, the differences between the types of health effects that are related to short-term versus long-term ambient NO2 suggest that standards are needed for each of these durations to protect against the associated health effects.
Support for Development of New CAAQS
This risk assessment was conducted to inform the development of new CAAQS for NO2 to replace the current National Ambient Air Quality Objectives (NAAQOs). It is recommended that new CAAQS be developed for ambient NO2 with consideration of the following key conclusions from the health risk assessment:
- there is strong evidence that ambient NO2 causes both short-term and long-term respiratory effects, and short-term mortality, as well as suggestive evidence linking it to a wide range of other adverse health outcomes;
- these effects have been observed in epidemiological studies at NO2 concentrations that commonly occur in Canada, well below the levels of existing NAAQOs and other ambient standards, such as provincial/territorial guidelines and the US National Ambient Air Quality Standards;
- in studies examining the shape of the concentration-response curve, there is an approximately linear relationship between ambient NO2 concentrations and health effects, with no clear evidence of a threshold; hence, based on the balance of the evidence it should be assumed that any increment in levels of ambient NO2 presents an increased risk for health effects, up to and including mortality;
- the health evidence supports the establishment of both short-term and long-term standards to protect against the full suite of health effects associated with ambient NO2.
To obtain an electronic copy of the document, Human Health Risk Assessment for Ambient Nitrogen Dioxide, please contact AIR@hc-sc.gc.ca.
- Date modified: