Risk and Exposure Assessment to Support the
Review of the NC>2 Primary National Ambient
Air Quality Standard: Second Draft
-------
EPA-452/P-08-004a
August 2008
Risk and Exposure Assessment to Support the
Review of the NC>2 Primary National Ambient
Air Quality Standard: Second Draft
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina
-------
Disclaimer
This draft document has been prepared by staff from the Ambient Standards Group, Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency. Any opinions,
findings, conclusions, or recommendations are those of the authors and do not necessarily reflect
the views of the EPA. This document is being circulated to obtain review and comment from the
Clean Air Scientific Advisory Committee (CASAC) and the general public. Comments on this
draft document should be addressed to Scott Jenkins, U.S. Environmental Protection Agency,
Office of Air Quality Planning and Standards, C504-06, Research Triangle Park, North Carolina
27711 (email: Jenkins.scott@epa.gov).
-------
Table of Contents
List of Tables iv
List of Figures vi
List of Acronyms/Abbreviation vii
1. INTRODUCTION 1
1.1 OVERVIEW 1
1.2 HISTORY 5
1.2.1 History of the NO 2 NAAQS. 5
1.2.2 Health Evidence from Previous Review 6
1.2.3 Assessment from Previous Review 7
1.3 SCOPE OF THE RISK AND EXPOSURE ASSESSMENT FOR THE CURRENT REVIEW 7
2. SOURCES, AMBIENT LEVELS, AND EXPOSURES 9
2.1 SOURCES OF NO2 9
2.2 AMBIENT LEVELS OF NO2 9
2.3 EXPOSURE TO NO2 10
2.3.1 Overview 10
2.3.2 Uncertainty Associated with the Ambient NO2 Monitoring Method 11
2.3.3 Uncertainty Associated with Ambient Levels as a Surrogate for Exposure 12
3. AT RISK POPULATIONS 13
3.1 OVERVIEW 13
3.2 SUSCEPTIBILITY: PRE-EXISTING DISEASE 13
3.3 SUSCEPTIBILITY: AGE 14
3.4 SUSCEPTIBILITY: GENETICS 15
3.5 SUSCEPTIBILITY: GENDER 16
3.6 VULNERABILITY: PROXIMITY (ON OR NEAR) TO ROAD WAYS 16
3.7 VULNERABILITY: SOCIOECONOMIC STATUS 16
3.8 NUMBER OF SUSCEPTIBLE/VULNERABLE INDIVIDUALS 17
4. HEALTH EFFECTS 18
4.1 INTRODUCTION 18
4.2 ADVERSE RESPIRATORY EFFECTS FOLLOWING SHORT-TERM EXPOSURES 18
4.2.1 Overview 18
4.2.2 Respiratory Emergency Department Visits and Hospitalizations 19
4.2.3 Respiratory Symptoms 20
4.2.4 Lung Host Defenses and Immunity 22
4.2.5 Airway Hyperresponsiveness 23
4.2.6 Airway Inflammation 26
4.2.7 Lung Function 26
4.2.8 Conclusions and Coherence of Evidence for Short-Term Respiratory Effects 27
4.3 OTHER ADVERSE EFFECTS FOLLOWING SHORT-TERM EXPOSURES 28
4.4 AD VERSE EFFECTS FOLLOWING LONG-TERM EXPOSURES 29
4.4.1 Respiratory Morbidity 29
4.4.2 Mortality 30
4.4.3 Other Long-Term Effects 31
August 2008 - Draft
-------
4.5 RELEVANCE OF SPECIFIC HEALTH EFFECTS TO THE NO2 RISK CHARACTERIZATION 31
4.5.1 Overview 31
4.5.2 Epidemiology 32
4.5.3 Controlled Human Exposure Studies 34
5. IDENTIFICATION OF POTENTIAL ALTERNATIVE STANDARDS FOR ANALYSIS 37
5.1 INTRODUCTION 37
5.2 INDICATOR 37
5.3 AVERAGING TIME 37
5.4 FORM 39
5.5 LEVEL 40
6. OVERVIEW OF APPROACHES TO ASSESSING EXPOSURES AND RISKS 45
6.1 INTRODUCTION 45
6.2 SIMULATING THE CURRENT AND ALTERNATIVE STANDARDS 47
6.2.1 Adjustment of Ambient Air Quality 48
6.2.2 Adjustment of Potential Health Effect Benchmark Levels 50
7. AMBIENT AIR QUALITY ASSESSMENT AND HEALTH RISK CHARACTERIZATION 51
7.1 OVERVIEW 51
7.2 APPROACH 53
7.2.1 Air Quality Data Screen 53
7.2.2 Selection of Locations for Air Quality Analysis 55
7.2.3 Estimation ofOn-Road Concentrations using Ambient Concentrations 56
7.3 AIR QUALITY AND HEALTH RISK CHARACTERIZATION RESULTS 59
7.3.1 Ambient Air Quality (As Is) 59
7.3.2 On-Road Concentrations Derived From Ambient Air Quality (As Is) 67
7.3.3 Ambient Air Quality Adjusted to Just Meet the Current and Alternative Standards 72
7.3.4 On-Road Concentrations Derived From Ambient Air Quality Adjusted to Just Meet the Current and
Alternative Standards 82
7.4 UNCERTAINTY ANALYSIS 90
7.4.1 Air Quality Data 90
7.4.2 Measurement Technique for Ambient NO2 91
7.4.3 Temporal Representation 91
7.4.4 Spatial Representation 92
7.4.5 Air Quality Adjustment Procedure 92
7.4.6 On-Road Concentration Simulation 94
7.4.7 Health Benchmark 96
8. EXPOSURE ASSESSMENT AND HEALTH RISK CHARACTERIZATION 98
9. CHARACTERIZATION OF HEALTH RISKS USING DATA FROM EPIDEMIOLOGICAL
STUDIES 99
9.1 INTRODUCTION 99
9.2 GENERAL APPROACH 100
9.3 AIR QUALITY INFORMATION 104
9.4 CONCENTRATION-RESPONSE FUNCTIONS 105
9.5 BASELINE HEALTH EFFECTS INCIDENCE DATA 107
9.6 ADDRESSING UNCERTAINTY AND VARIABILITY 108
9.7 RISK ESTIMATES FOR EMERGENCY DEPARTMENT VISITS Ill
10. REFERENCES 117
APPENDICES
APPENDIX A - Supplement to the NO2 Air Quality Characterization
APPENDIX B - Supplement to the NO2 Exposure Assessment
APPENDIX C - Nitrogen Dioxide Health Risk Assessment for Atlanta, GA
August 2008 - Draft
-------
List of Tables
Number Page
Table 4-1. Fraction of nitrogen dioxide-exposed asthmatics with increased nonspecific airway
hyperresponsiveness 24
Table 7-1. Counts of complete site-years of NO2 monitoring data 54
Table 7-2. Locations selected for Tier INO2 Air Quality Characterization, associated
abbreviations, and values of selection criteria 56
Table 7-3. Monitoring site-years and annual average NO2 concentrations for two monitoring
periods, historic and recent air quality data (as is) using monitors sited >100 m of a
major road 60
Table 7-4. Monitoring site-years and annual average NO2 concentrations for two monitoring
periods, historic and recent air quality data (as is) using monitors sited <100 m of a
major road 61
Table 7-5 Number of exceedances of short-term (1-hour) potential health effect benchmark
levels in a year, 1995-2000 historic NO2 air quality (as is) using monitors sited >100
m of a major road 63
Table 7-6 Number of exceedances of short-term (1-hour) potential health effect benchmark
levels in a year, 2001-2006 recent NO2 air quality (as is) using monitors sited >100 m
of a major road 64
Table 7-7. Number of exceedances of short-term (1-hour) potential health effect benchmark
levels in a year, 1995-2000 historic NO2 air quality (as is) using monitors sited <100
m of a major road 65
Table 7-8. Number of exceedances of short-term (1-hour) potential health effect benchmark
levels in a year, 2001-2006 recent NO2 air quality (as is) using monitors sited <100 m
of a major road 66
Table 7-9. Estimated annual average on-road NO2 concentrations for two monitoring periods,
historic and recent air quality data (as is) 69
Table 7-10. Estimated number of exceedances of short-term (1-hour) potential health effect
benchmark levels in a year on-roads, 1995-2000 historic NO2 air quality (as is) 70
Table 7-11. Estimated number of exceedances of short-term (1-hour) potential health effect
benchmark levels in a year on-roads, 2001-2006 recent NO2 air quality (as is) 71
Table 7-12 Estimated annual mean NO2 concentration and the number of exceedances of 1-hour
NO2 concentration levels, using 2001-2003 air quality adjusted to just meeting a 1-
hour 100 ppb 98th percentile alternative standard, monitoring locations sited > 100 m
of a major road 78
Table 7-13. Estimated annual mean NO2 concentration and the number of exceedances of 1-hour
NO2 concentration levels, using 2001-2003 air quality adjusted to just meeting a 1-
hour 100 ppb 98th percentile alternative standard, monitoring locations sited < 100 m
of a major road 79
Table 7-14. Estimated mean number of exceedances of 100 ppb 1-hour NO2 concentrations,
using 2001-2003 air quality as is and that adjusted to just meeting the current and
alternative standards (98th percentile) for monitoring locations sited > 100 m and <
100 m of a major road 80
August 2008 - Draft iii
-------
Table 7-15. Estimated mean number of exceedances of 150 ppb 1-hour NO2 concentrations,
using 2001-2003 air quality as is and that adjusted to just meeting the current and
alternative standards (98th percentile) for monitoring locations sited > 100 m and <
100 m of a major road 81
Table 7-16. Estimated annual mean NO2 concentration and the number of exceedances of 1-hour
NO2 concentration levels on-roads, using 2001-2003 air quality adjusted to just
meeting a 1-hour 100 ppb 98th percentile alternative standard 87
Table 7-17. Estimated mean number of exceedances of 100 ppb 1-hour NO2 concentrations on-
roads, using air quality as is and that adjusted to just meeting the current and
alternative standards (98th percentile) 88
Table 7-18. Estimated mean number of exceedances of 150 ppb 1-hour NO2 concentrations on-
roads, using air quality as is and that adjusted to just meeting the current and
alternative standards (98th percentile) 89
Table 7-19. Summary of qualitative uncertainty analysis for the air quality and health risk
characterization 97
Table 9-1. Estimated Incidence of Respiratory ED Visits Associated with "As Is" NO2
Concentrations and NO2 Concentrations that Just Meet Alternative Standards in
Atlanta, GA, Based on Adjusting 2005 NO2 Concentrations 112
Table 9-2. Estimated Incidence of Respiratory ED Visits Associated with "As Is" NO2
Concentrations and NO2 Concentrations that Just Meet Alternative Standards in
Atlanta, GA, Based on Adjusting 2006 NO2 Concentrations 113
Table 9-3. Estimated Incidence of Respiratory ED Visits Associated with "As Is" NO2
Concentrations and NO2 Concentrations that Just Meet Alternative Standards in
Atlanta, GA, Based on Adjusting 2007 NO2 Concentrations 114
August 2008 - Draft iv
-------
List of Figures
Number Page
Figure 1-1. Overview of the analyses described in this document and their interconnections 4
Figure 5-1. NO2 effect estimates (95% CI) for ED visits/HA and associated 1-h daily maximum
NO2 levels (98th and 99th percentile values in boxes) 42
Figure 5-2. NO2 effect estimates for respiratory symptoms and associated 1-h daily maximum
NO2 levels (98th and 99th percentile values in boxes) 43
Figure 7-1. Estimated mean number of exceedances of selected 1-hour potential health effect
benchmark levels, using recent air quality adjusted to just meeting the current annual
standard (0.053 ppm) 75
Figure 7-2. Estimated number of exceedances of potential health effect benchmarks (100 ppb,
top; 200 ppb, bottom) in Chicago given just meeting alternative 1-hour standard levels
(98th percentile, left; and 99th percentile, right) using recent air quality data from
monitors sited < 100 m of a major road and sited >100 m of major roads 76
Figure 7-3. Estimated number of exceedances of 200 ppb in four locations (Phoenix, Los
Angeles, Philadelphia, and St. Louis) given just meeting alternative 1-hour 98th
percentile standard levels using recent air quality data from monitors sited < 100 m of
a major road and sited >100 m of major roads 77
Figure 7-4. Estimated mean number of exceedances of selected 1-hour potential health effect
benchmark levels on-roads, using 2001-2003 air quality adjusted to just meeting the
current annual standard (0.053 ppm) 84
Figure 7-5. Estimated number of exceedances of potential health effect benchmarks (100 ppb,
top; 200 ppb, bottom) on-roads in Chicago given just meeting alternative 1-hour
standard levels (98th percentile, left; and 99l percentile, right) using recent air quality
data 85
Figure 7-6. Estimated number of exceedances of 200 ppb in-roads in four locations (Phoenix,
Los Angeles, Philadelphia, and St. Louis) given just meeting alternative 1-hour 98th
percentile standard levels using recent air quality data 86
Figure 9-1. Major components of nitrogen dioxide health risk assessment for emergency
department visits 102
August 2008 - Draft v
-------
List of Acronyms/Abbreviations
AADT Annual average daily traffic
A/C Air conditioning
AER Air exchange rate
AERMOD American Meteorological Society (AMS)/EPA Regulatory Model
AHS American Housing Survey
APEX EPA's Air Pollutants Exposure model, version 4
ANOVA One-way analysis of variance
AQS EPA's Air Quality System
AS Asthma symptoms
BRFSS Behavioral Risk Factor Surveillance System
C Cough
CAA Clean Air Act
CAMD EPA's Clean Air Markets Division
CASAC Clean Air Scientific Advisory Committee
CDC Centers for Disease Control
CHAD EPA's Consolidated Human Activity Database
CHF Congestive Heart Failure
Clev/Cinn Cleveland and Cincinnati, Ohio
CMSA Consolidated metropolitan statistical area
CO Carbon monoxide
COPD Chonic Obstructive Pulmonary Disease
COV Coefficient of Variation
C-R Concentration-Response
CTPP Census Transportation Planning Package
DVRPC Delaware Valley Regional Planning Council
EDR Emergency department visits for respiratory disease
EDA Emergency department visits for asthma
ED AC Emergency department visits for asthma - children
HAAC Hospital admissions for asthma - children
ER Emergency room
EPA United States Environmental Protection Agency
EOC Exposure of Concern
GM Geometric mean
GSD Geometric standard deviation
GST Glutathione S-transferase (e.g., GSTM1, GSTP1, GSTT1)
h Hour
HNO3 Nitric acid
HONO Nitrous acid
ID Identification
ISA Integrated Science Assessment
ISH Integrated Surface Hourly Database
km Kilometer
August 2008 - Draft
VI
-------
L95
LA
m
max
ME
med
MI
min
MSA
NAAQS
NAICS
NCEA
NEI
NEM
NCDC
NHAPS
NHIS
NO2
NOX
NO3'
NWS
NYC
NYDOH
03
OAQPS
OR
ORD
ORIS
POC
ppb
PEN
PM
ppm
PRB
PROX
PVMRM
RECS
RIU
RR
SAS
SB
SEP
SIC
SD
se
TDM
Lower limit of the 95th confidence interval
Los Angeles, California
Meter
Maximum
Microenvironment
Median
Myocardial Infarction
Minimum
Metropolitan statistical area
National Ambient Air Quality Standards
North American Industrial Classification System
National Center for Environmental Assessment
National Emissions Inventory
NAAQS Exposure Model
National Climatic Data Center
National Human Activity Pattern Study
National Health Interview Survey
Nitrogen dioxide
Oxides of nitrogen
Nitrate ion
National Weather Service
New York City
New York Department of Health
Ozone
Office of Air Quality Planning and Standards
Odds ratio
Office of Research and Development
Office of Regulatory Information Systems identification code
Parameter occurrence code
Parts per billion
Penetration factor
Particulate matter
Parts per million
Policy-Relevant Background
Proximity factor
Plume Volume Molar Ratio Method
Residential Energy Consumption Survey
Rescue inhaler use
Relative risk
Statistical Analysis Software
Shortness of breath
Social-economic position
Standard Industrial Code
Standard deviation
Standard error
Travel Demand Modeling
August 2008 - Draft
vn
-------
tpy Tons per year
TRIM EPA's Total Risk Integrated Methodology
U95 Upper limit of the 95th confidence interval
US DOT United States Department of Transportation
US EPA United States Environmental Protection Agency
USGS United States Geological Survey
VMT Vehicle miles traveled
W Wheeze
August 2008 - Draft viii
-------
August 2008 - Draft ix
-------
1. INTRODUCTION
1.1 OVERVIEW
3 The U.S. Environmental Protection Agency (EPA) is conducting a review of the national
4 ambient air quality standards (NAAQS) for nitrogen dioxide (NO2). Sections 108 and 109 of the
5 Clean Air Act (The Act) govern the establishment and periodic review of the air quality criteria
6 and the NAAQS. These standards are established for pollutants that may reasonably be
7 anticipated to endanger public health or welfare, and whose presence in the ambient air results
8 from numerous or diverse mobile or stationary sources. The NAAQS are based on air quality
9 criteria, which reflect the latest scientific knowledge useful in indicating the kind and extent of
10 identifiable effects on public health or welfare that may be expected from the presence of the
11 pollutant in ambient air. The EPA Administrator promulgates and periodically reviews primary
12 (health-based) and secondary (welfare-based) NAAQS for such pollutants. Based on periodic
13 reviews of the air quality criteria and standards, the Administrator makes revisions in the criteria
14 and standards and promulgates any new standards as may be appropriate. The Act also requires
15 that an independent scientific review committee advise the Administrator as part of this NAAQS
16 review process, a function now performed by the Clean Air Scientific Advisory Committee
17 (CAS AC).
18 The Agency has recently made a number of changes to the process for reviewing the
19 NAAQS (described at http://www.epa.gov/ttn/naaqs/). In making these changes, the Agency
20 consulted with CAS AC. This new process, which is being applied to the current review of the
21 NC>2 NAAQS, contains four major components. Each of these components, as they relate to the
22 review of the NC>2 primary NAAQS, is described below.
23 The first of these components is an integrated review plan. This plan presents the
24 schedule for the review, the process for conducting the review, and the key policy-relevant
25 science issues that will guide the review. The integrated review plan for this review of the NC>2
26 primary NAAQS is presented in the Integrated Review Plan for the Primary National Ambient
27 Air Quality Standard for Nitrogen Dioxide (EPA, 2007a). The policy-relevant questions
28 identified in this document to guide the review are:
August 2008 - Draft
-------
1 • Has new information altered the scientific support for the occurrence of health effects
2 following short- and/or long-term exposure to levels of nitrogen oxides (NOX) found in
3 the ambient air?
4 • What do recent studies focused on the near-roadway environment tell us about health
5 effects of NOX?
6 • At what levels of NOX exposure do health effects of concern occur?
7 • Has new information altered conclusions from previous reviews regarding the plausibility
8 of adverse health effects caused by NOX exposure?
9 • To what extent have important uncertainties identified in the last review been reduced
10 and/or have new uncertainties emerged?
11 • What are the air quality relationships between short-term and long-term exposures
12 toNOx?
13 Additional questions will become relevant if the evidence suggests that revision of the current
14 standard might be appropriate. These questions are:
15 • Is there evidence for the occurrence of adverse health effects at levels of NOX lower than
16 those observed previously? If so, at what levels and what are the important uncertainties
17 associated with that evidence?
18 • Do exposure estimates suggest that exposures of concern for NOx-induced health effects
19 will occur with current ambient levels of NC>2 or with levels that just meet current, or
20 potential alternative, standards? If so, are these exposures of sufficient magnitude such
21 that the health effects might reasonably be judged to be important from a public health
22 perspective? What are the important uncertainties associated with these exposure
23 estimates?
24 • Do the evidence, the air quality assessment, and the risk/exposure assessment provide
25 support for considering different standard indicators or averaging times?
26 • What range of levels is supported by the evidence, the air quality assessment, and the
27 risk/exposure assessments? What are the uncertainties and limitations in the evidence
28 and the assessments?
August 2008 - Draft
-------
1 • What is the range of forms supported by the evidence, the air quality assessment, and the
2 exposure/risk assessments? What are the uncertainties and limitations in the evidence
3 and the assessments?
4 The second component of the review process is a science assessment. A concise
5 synthesis of the most policy-relevant science has been compiled into the Integrated Science
6 Assessment (ISA). The ISA is supported by a series of annexes that contain more detailed
7 information about the scientific literature. The ISA to support this review of the NC>2 primary
8 NAAQS is presented in the Integrated Science Assessment for Oxides of Nitrogen - Health
9 Criteria, henceforth referred to as the ISA (EPA, 2008a).
10 The third component of the review process is a risk and exposure assessment, the second
11 draft of which is described in this document. The purpose of this draft document is to
12 communicate EPA's assessment of exposures and risks associated with ambient NC>2. This
13 second draft of the risk and exposure assessment develops estimates of human exposures and
14 risks associated with current ambient levels of NC>2, with levels that just meet the current
15 standard, and with levels that just meet potential alternative standards. Figure 1-1 (below)
16 presents a schematic overview of the analyses described in this document and how those
17 analyses fit together. Each of the steps highlighted in Figure 1-1 is described in more detail in
18 subsequent sections of this document.
19
August 2008 - Draft
-------
in ISA
2
3
4
5
6
7
Identification of
benchmark
*!
€ p
a I
c ^
= ,
t
Qualitative characterization of
controlled human exposure
studies
Qualitative
characterization of U.S.
epidemiology studies
Air quality
in
U.S.
Inform city
selection
Exposure
characterization in
city/cities
i 5. S;
identify
alternative
standards
risk
assessment
Compare NO2 levels with
potential health benchmark
levels
Compare N02
exposures to
health benchmark
Output: of per
that NO2
Output; and
of
to N02 that
levels
Oytpuf; of
and of
for
on
Air Quality
on
Exposure
on
Epidemiology
Risk-based
considerations to
Inform standard setting
Figure 1-1. Overview of the analyses described in this document and their interconnections
The results of the risk and exposure assessment will be considered alongside the health evidence,
as evaluated in the final ISA, to inform the policy assessment and rulemaking process (see
below). The draft plan for conducting the risk and exposure assessment to support the NC>2
primary NAAQS is presented in the Nitrogen Dioxide Health Assessment Plan: Scope and
Methods for Exposure and Risk Assessment, henceforth referred to as the Health Assessment
8 Plan (EPA, 2007b). The first draft of the risk and exposure assessment is presented in Risk and
9 Exposure Assessment to Support the Review of the 7V02 Primary National Ambient Air Quality
10 Standard: First Draft (EPA, 2008b).
August 2008 - Draft
-------
1 The fourth component of the process is the policy assessment and rulemaking. The
2 Agency's views on policy options will be published in the Federal Register as an advance notice
3 of proposed rulemaking (ANPR). This policy assessment will address the adequacy of the
4 current standard and of any potential alternative standards, which will be defined in terms of
5 indicator, averaging time, form,1 and level. To accomplish this, the policy assessment will
6 consider the results of the final risk and exposure assessment as well as the scientific evidence
7 (including evidence from the epidemiologic, controlled human exposure, and animal
8 toxicological literatures) evaluated in the ISA. Taking into consideration CASAC advice and
9 recommendations, as well as public comment on the ANPR, the Agency will publish a proposed
10 rule, to be followed by a public comment period. Taking into account comments received on the
11 proposed rule, the Agency will issue a final rule to complete the rulemaking process.
12 1.2 HISTORY
13 1.2.1 History of the NO2 NAAQS
14 On April 30, 1971, EPA promulgated identical primary and secondary NAAQS for NC>2
15 under section 109 of the Act. The standards were set at 0.053 parts per million (ppm), annual
16 average (36 FR 8186). In 1982, EPA published Air Quality Criteria for Oxides of Nitrogen
17 (EPA, 1982), which updated the scientific criteria upon which the initial NC>2 standards were
18 based. On February 23, 1984, EPA proposed to retain these standards (49 FR 6866). After
19 taking into account public comments, EPA published the final decision to retain these standards
20 on June 19, 1985 (50 FR 25532).
21 On July 22, 1987, EPA announced that it was undertaking plans to revise the 1982 air
22 quality criteria (52 FR 27580). In November 1991, EPA released an updated draft air quality
23 criteria document for CASAC and public review and comment (56 FR 59285). The draft
24 document provided a comprehensive assessment of the available scientific and technical
25 information on health and welfare effects associated with NO2 and other oxides of nitrogen. The
26 CASAC reviewed the draft document at a meeting held on July 1, 1993 and concluded in a
27 closure letter to the Administrator that the document "provides a scientifically balanced and
28 defensible summary of current knowledge of the effects of this pollutant and provides an
1 The "form" of a standard defines the air quality statistic that is to be compared to the level of the standard
in determining whether an area attains the standard.
August 2008 - Draft 5
-------
1 adequate basis for EPA to make a decision as to the appropriate NAAQS for NO2" (Wolff,
2 1993). The Air Quality Criteria Document for the Oxides of Nitrogen was then finalized (EPA,
3 1993).
4 The EPA also prepared a Staff Paper that summarized an air quality assessment for NO2
5 conducted by the Agency (McCurdy, 1994), summarized and integrated the key studies and
6 scientific evidence contained in the revised air quality criteria document, and identified the
7 critical elements to be considered in the review of the NO2 NAAQS. The CASAC reviewed two
8 drafts of the Staff Paper and concluded in a closure letter to the Administrator (Wolff, 1995) that
9 the document provided a "scientifically adequate basis for regulatory decisions on nitrogen
10 dioxide." In September of 1995, EPA finalized the Staff Paper entitled, "Review of the National
11 Ambient Air Quality Standards for Nitrogen Dioxide: Assessment of Scientific and Technical
12 Information" (EPA, 1995).
13 In October 1995, the Administrator announced her proposed decision not to revise either
14 the primary or secondary NAAQS for NO2 (60 FR 52874; October 11, 1995). A year later, the
15 Administrator made a final determination not to revise the NAAQS for NO2 after careful
16 evaluation of the comments received on the proposal (61 FR 52852, October 8, 1996). The level
17 for both the existing primary and secondary NAAQS for NO2 is 0.053 parts per million (ppm)
18 (100 micrograms per cubic meter of air [|j,g/m3]), annual arithmetic average, calculated as the
19 arithmetic mean of the l-hourNO2 concentrations.
20 1.2.2 Health Evidence from Previous Review
21 The prior Air Quality Criteria Document (AQCD) for Oxides of Nitrogen (EPA, 1993)
22 concluded that there were two key health effects of greatest concern at ambient or near-ambient
23 levels of NO2, increased airway responsiveness in asthmatic individuals after short-term
24 exposures and increased occurrence of respiratory illness in children with longer-term exposures.
25 Evidence also was found for increased risk of emphysema, but this was of maj or concern only
26 with exposures to levels of NO2 much higher than then-current ambient levels. The evidence
27 regarding airway responsiveness was drawn largely from controlled human exposure studies.
28 The evidence for respiratory illness was drawn from epidemiologic studies that reported
29 associations between respiratory symptoms and indoor exposures to NO2 in people living in
30 homes with gas stoves. The biological plausibility of the epidemiologic results was supported by
August 2008 - Draft 6
-------
1 toxicological studies that detected changes in lung host defenses following NO2 exposure.
2 Subpopulations considered potentially more susceptible to the effects of NO2 included
3 individuals with preexisting respiratory disease, children, and the elderly.
4 1.2.3 Assessment from Previous Review
5 In the previous review of the NO2 NAAQS, risks were characterized by comparing
6 ambient monitoring data, which was used as a surrogate for exposure, with potential health
7 benchmark levels identified from controlled human exposure studies. At the time of the review,
8 a few studies indicated the possibility for adverse health effects due to short-term (e.g., 1-hour)
9 exposures between 0.20 ppm and 0.30 ppm NO2. Therefore, the focus of the assessment was on
10 the potential for short-term (i.e., 1-hour) exposures to NO2 levels above potential health
11 benchmarks in this range. The assessment used monitoring data from the years 1988-1992 and
12 screened for sites with one or more hourly exceedances of potential short-term health effect
13 benchmarks. Predictive models were then constructed to relate the frequency of hourly
14 concentrations above short-term health effect benchmarks to a range of annual average
15 concentrations, including the current standard. Based on the results of this analysis, both
16 CAS AC (Wolff, 1995) and the Administrator (60 FR 52874) concluded that the minimal
17 occurrence of short-term peak concentrations at or above a potential health effect benchmark of
18 0.20 ppm (1-h average) indicated that the existing annual standard would provide adequate
19 health protection against short-term exposures. This conclusion was instrumental in providing
20 the rationale for the decision in the last review to retain the existing annual standard.
21 1.3 SCOPE OF THE RISK AND EXPOSURE ASSESSMENT FOR THE
22 CURRENT REVIEW
23 NOX include multiple gaseous (e.g., NO2, NO, HONO) and particulate (e.g., nitrate)
24 species. As discussed in the integrated review plan (2007a), the current review of the NO2
25 NAAQS will focus on the gaseous species of NOX and will not consider health effects directly
26 associated with particulate species of NOX. Of the gaseous species, EPA has historically
27 determined it appropriate to specify the indicator of the standard in terms of NO2 because the
28 majority of the information regarding health effects and exposures is for NO2. The current ISA
August 2008 - Draft
-------
1 (EPA, 2008a) has found this to be the case and, therefore, NC>2 will be used as the indicator for
2 the gaseous NOX in the risk and exposure assessments described in this document.
August 2008 - Draft
-------
i 2. SOURCES, AMBIENT LEVELS, AND EXPOSURES
2 2.1 SOURCES OF NO2
3 Ambient levels of NC>2 are the product of both direct NC>2 emissions and emissions of
4 other NOX (e.g, NO), which can then be converted to NC>2 (for a more detailed discussion see the
5 ISA, section 2.2). Nationally, anthopogenic sources account for approximately 87% of total NOX
6 emissions. Mobile sources (both on-road and off-road) account for about 60% of total
7 anthopogenic emissions of NOX, while stationary sources (e.g., electrical utilities and industry)
8 account for the remainder (annex table 2.6-1). Highway vehicles represent the major mobile source
9 component. In the United States, approximately half the mobile source emissions are contributed by
10 diesel engines and half are emitted by gasoline-fueled vehicles and other sources (annex section
11 2.6.2 and Table 2.6-1). Apart from these anthopogenic sources, there are also natural sources of
12 NOX including microbial activity in soils, lightning, and wildfires (ISA, section 2.2.1 and annex
13 section 2.6.2).
14 2.2 AMBIENT LEVELS OF NO2
15 According to monitoring data, nationwide levels of ambient NC>2 (annual average)
16 decreased 41% between 1980 and 2006 (ISA, Figure 2.4-15). Between 2003 and 2005, national
17 mean concentrations of NC>2 were about 15 ppb for averaging periods ranging from a day to a
18 year. The average daily maximum hourly NC>2 concentrations were approximately 30 ppb.
19 These values are about twice as high as the 24-h averages. The highest maximum hourly
20 concentrations (-200 ppb) between 2003 and 2005 are more than a factor often higher than the
21 mean hourly or 24-h concentrations (ISA, Figure 2.4-13). The highest levels of NC>2 in the
22 United States can be found in and around Los Angeles, in the Midwest, and in the Northeast.
23 Nitrogen dioxide is monitored mainly in large urban areas and, therefore, data from the
24 NC>2 monitoring network is generally more representative of urban areas than rural areas. Levels
25 in non-urban areas can be estimated with modeling. Model-based estimates indicate that NC>2
26 levels in many non-urban areas of the United States are less than 1 ppb. Levels in these areas
27 can approach policy-relevant background concentrations, which are those concentrations that
28 would occur in the United States in the absence of anthopogenic emissions in continental North
August 2008 - Draft
-------
1 America (defined here as the United States, Canada, and Mexico). For NC>2, policy-relevant
2 background concentrations are estimated to range from 0.1 ppb to 0.3 ppb (ISA, section 2.4.6).
3 Ambient levels of NO2 exhibit both seasonal and diurnal variation. In southern cities,
4 such as Atlanta, higher concentrations are found during winter, consistent with the lowest mixing
5 layer heights being found during that time of the year. Lower concentrations are found during
6 summer, consistent with higher mixing layer heights and increased rates of photochemical
7 oxidation of NC>2. For cities in the Midwest and Northeast, such as Chicago and New York City,
8 higher levels tend to be found from late winter to early spring with lower levels occurring from
9 summer though the fall. In Los Angeles the highest levels tend to occur from autumn though
10 early winter and the lowest levels from spring though early summer. Mean and peak
11 concentrations in winter can be up to a factor of two larger than in the summer at sites in Los
12 Angeles. In terms of daily variability, NC>2 levels typically peak during the morning rush hours.
13 Monitor siting plays a key role in evaluating diurnal variability as monitors located further away
14 from traffic will show cycles that are less pronounced over the course of a day than monitors
15 located closer to traffic.
16 2.3 EXPOSURE TO NO2
17 2.3.1 Overview
18 Human exposure to an airborne pollutant can be characterized by contact between a
19 person and the pollutant at a specific concentration for a specified period of time (ISA, section
20 2.5.1). The integrated exposure of a person to a given pollutant is the time-weighted average of
21 the exposures over all time intervals for all microenvironments in which the individual spends
22 time. Microenvironments in which people are exposed to air pollutants such as NC>2 typically
23 include residential indoor environments and other indoor locations, near-traffic outdoor
24 environments and other outdoor locations, and in vehicles (ISA, Figure 2.5-1).
25 There is a large amount of variability in the time that individuals spend in different
26 microenvironments, but on average people spend the majority of their time (about 87%) indoors.
27 Most of this time is spent at home with less time spent in an office/workplace or other indoor
28 locations (ISA, Figure 2.5-1). On average, people spend about 8% of their time outdoors and 6%
29 of their time in vehicles. Significant variability surrounds each of these broad estimates,
30 particularly when considering influential personal attributes such as age or gender; when
August 2008 - Draft 10
-------
1 accounting for daily, weekly, or seasonal factors influencing personal behavior; or when
2 characterizing individual variability in time spent in various locations (McCurdy and Graham,
3 2003; Graham and McCurdy, 2004). Typically, the time spent outdoors or in vehicles could vary
4 by 100% or more depending on which of these influential factors are considered. One potential
5 consequence of this is that exposure misclassification can result when total human exposure is
6 not disaggregated between relevant microenvironments and the variability in time spent in these
7 locations is not taken into consideration. Such misclassification, which can occur in
8 epidemiologic studies that rely on ambient pollutant levels as a surrogate for exposure to ambient
9 NO2, may obscure the true relationship between ambient air pollutant exposures and health
10 outcomes. Sections 2.3.2 and 2.3.3 (below) discuss in more detail sources of NO2 exposure
11 misclassification that are relevant for the current review of the NC>2 NAAQS.
12 2.3.2 Uncertainty Associated with the Ambient NOi Monitoring Method
13 The current approach to monitoring ambient NC>2 can introduce uncertainty into exposure
14 estimates. For example, the method for estimating ambient NC>2 levels (i.e., subtraction of NO
15 from a measure of total NOX) is subject to interference by NOX oxidation products. Limited
16 evidence suggests that these compounds result in an overestimate of NO2 levels by roughly 20 to
17 25% at typical ambient levels. Smaller relative errors are estimated to occur in measurements
18 taken near strong NOX sources since most of the mass emitted as NO or NO2 would not yet have
19 been further oxidized. Relatively larger errors appear in locations more distant from strong local
20 NOX sources. Additionally, many NO2 monitors are elevated above ground level in the cores of
21 large cities. Because most sources of NO2 are near ground level, this produces a gradient of NO2
22 with higher levels near ground level and lower levels being detected at the elevated monitor.
23 One comparison has found an average of a 2.5-fold increase in NO2 concentration measured at 4
24 meters above the ground compared to 15 meters above the ground. Levels are likely even higher
25 at elevations below 4 meters (ISA, section 2.5.3.3). Another source of uncertainty in exposure
26 estimates can result from monitor location. NO2 monitors are sited for compliance with air
27 quality standards rather than for capturing small-scale variability in NO2 concentrations near
28 sources such as roadway traffic. Significant gradients in NO2 concentrations near roadways have
29 been observed in several studies, and NO2 concentrations have been found to be correlated with
30 distance from roadway and traffic volume (ISA, section 2.5.3.2).
August 2008 - Draft 11
-------
1 2.3.3 Uncertainty Associated with Ambient Levels as a Surrogate for Exposure
2 Many epidemiologic studies rely on measures of ambient NC>2 concentrations as
3 surrogates for personal exposure to ambient NC>2. Results have been mixed regarding the
4 appropriateness of using ambient levels of NC>2 as a surrogate for personal exposures to ambient
5 NC>2. Studies examining the association between ambient NC>2 and personal exposure to NC>2
6 have generated mixed results due to 1) the prevalence of indoor sources of NO2; 2) the spatial
7 heterogeneity of NC>2 in study areas; 3) the seasonal and geographic variability in the infiltration
8 of ambient NC>2; 4) differences in the time spent in different microenvironments; and 5)
9 differences in study design. As a result, some researchers have concluded that ambient NO2 may
10 be a reasonable proxy for personal exposure, while others have noted that caution must be
11 exercised (ISA, section 2.5.9). However, this source of exposure error is not expected to change
12 the principal conclusions from NC>2 epidemiologic studies (see chapter 4 of this document) since
13 it generally tends to reduce, rather than increase, effect estimates (ISA, section 5.2.2).
August 2008 - Draft 12
-------
3. AT RISK POPULATIONS
2 3.1 OVERVIEW
3 Specific subpopulations are at increased risk for suffering NCVrelated health effects. This
4 could occur because they are affected by lower levels of NC>2 than the general population
5 (susceptibility), because they experience a larger health impact than the general population to a
6 given level of exposure (susceptibility), and/or because they are exposed to higher levels of NC>2
7 than the general population (vulnerability). The term susceptibility generally encompasses
8 innate (e.g., genetic or developmental) and/or acquired (e.g., age or disease) factors that make
9 individuals more likely to experience effects with exposure to pollutants. Given the likely
10 heterogeneity of individual responses to air pollution, the severity of health effects experienced
11 by a susceptible subgroup may be much greater than that experienced by the population at large.
12 Factors that may influence susceptibility to the effects of air pollution include age (e.g., infants,
13 children, elderly); gender; race/ethnicity; genetic factors; and pre-existing disease/condition (e.g.,
14 obesity, diabetes, respiratory disease (e.g., asthma, chonic obstructive pulmonary disease
15 (COPD)), cardiovascular disease, airway hyperresponsiveness, respiratory infection, adverse
16 birth outcome) (ISA, sections 4.3.1, 4.3.5, and 5.3.2.8). In addition, some population groups are
17 vulnerable to pollution-related effects because their air pollution exposures are higher than those
18 of the general population. Factors that may influence vulnerability to the effects of air pollution
19 include socioeconomic status, education level, air conditioning use, proximity to roadways,
20 geographic location, level of physical activity, and work environment (e.g., indoor versus
21 outdoor) (ISA, section 4.3.5). The ISA discusses factors that can confer susceptibility and/or
22 vulnerability to air pollution with most of the discussion devoted to factors for which NC>2-
23 specific evidence exists (ISA, section 4.3). These factors are discussed in more detail below.
24 3.2 SUSCEPTIBILITY: PRE-EXISTING DISEASE
25 A number of health conditions are believed to put individuals at greater risk for adverse
26 events following exposure to air pollution. In general, these include asthma, COPD, respiratory
27 infection, conduction disorders, congestive heart failure (CHF), diabetes, past myocardial
28 infarction (MI), obesity, coronary artery disease, low birth weight/prematurity, and hypertension
August 2008 - Draft 13
-------
1 (ISA, sections 4.3.1, 4.3.5, and 5.3.2.9). In addition to these conditions, epidemiologic evidence
2 indicates that individuals with bronchial or airway hyperresponsiveness, as determined by
3 methacholine provocation, may be at increased risk for experiencing respiratory symptoms (ISA,
4 section 4.3.1). In considering NO2 specifically, the ISA evaluates studies on asthmatics,
5 individuals with cardiopulmonary disease, and diabetics (ISA, sections 4.3.1.1 and 4.3.1.2).
6 These groups are discussed in more detail below.
7 Epidemiologic and controlled human exposure studies, supported by animal toxicology
8 studies, have provided evidence for associations between NO2 exposure and respiratory effects in
9 asthmatics (ISA, section 4.3.1.1). The ISA found evidence from epidemiologic studies for an
10 association between ambient NO2 and children's hospital admissions, emergency department
11 (ED) visits, and calls to doctors for asthma. NO2 levels were associated with aggravation of
12 asthma effects that include symptoms, medication use, and lung function. Time-series studies
13 also demonstrated a relationship in children between hospital admissions or ED visits for asthma
14 and ambient NO2 levels, even after adjusting for co-pollutants such as particulate matter (PM)
15 and carbon monoxide (CO) (ISA, section 4.3.1.1). Important evidence was also available from
16 epidemiologic studies of indoor NO2 exposures. Recent studies have shown associations with
17 asthma attacks and severity of virus-induced asthma (ISA, section 4.3.1.1). In addition, in
18 controlled human exposure studies, airway hyperresponsiveness in asthmatics appeared to be the
19 most sensitive indicator of response to NO2 (ISA, section 4.3.1.1).
20 Compared to asthma, less evidence is available to support cardiovascular disease as a
21 mediator of susceptibility to NO2. However, recent epidemiologic studies report that individuals
22 with preexisting conditions (e.g., including diabetes, CHF, prior MI) may be at increased risk for
23 adverse cardiac health events associated with ambient NO2 concentrations (ISA, section 4.3.1.2).
24 There is only limited supporting evidence from clinical or toxicological studies on potential
25 susceptibility to NO2 in persons with cardiovascular disease (ISA, section 4.3.1.2).
26 3.3 SUSCEPTIBILITY: AGE
27 The ISA identifies both children (i.e., <18 years of age) and older adults (i.e., >65 years
28 of age) as groups that are potentially more susceptible than the general population to the health
29 effects associated with ambient NO2 concentrations (ISA, section 4.3.2). The ISA found
30 evidence that associations of NO2 with respiratory ED visits and hospitalizations were stronger
August 2008 - Draft 14
-------
1 among children and older adults, though not all studies agreed on this issue (ISA, section 4.3.2).
2 In addition, long-term exposure studies suggest effects in children that include impaired lung
3 function growth, increased respiratory symptoms and infections, and onset of asthma (ISA,
4 section 3.4 and 4.3.2). In some studies, associations between NC>2 and hospitalizations or ED
5 visits for CVD have been observed in elderly populations. Among studies that observed positive
6 associations between NC>2 and mortality, a comparison indicated that, in general, the elderly
7 population was more susceptible than the non-elderly population to NC>2 effects (ISA, section
8 4.3.2).
9 3.4 SUSCEPTIBILITY: GENETICS
10 As noted in the ISA (section 4.3.4), genetic factors related to health outcomes and
11 ambient pollutant exposures merit consideration. Several criteria must be satisfied in selecting
12 and establishing useful links between polymorphisms in candidate genes and adverse respiratory
13 effects. First, the product of the candidate gene must be significantly involved in the
14 pathogenesis of the adverse effect of interest. Second, polymorphisms in the gene must produce
15 a functional change in either the protein product or in the level of expression of the protein.
16 Third, in epidemiologic studies, the issue of confounding by other environmental exposures must
17 be carefully considered (ISA, section 4.3.4).
18 Investigation of genetic susceptibility to NC>2 effects has focused on the glutathione S-
19 tranferase (GST) gene. Several GST genes have common, functionally-important alleles that
20 affect host defense in the lung (ISA, section 4.3.4). GST genes are inducible by electrophilic
21 species (e.g., reactive oxygen species) and individuals with genotypes that result in enzymes with
22 reduced or absent peroxide activity are likely to have reduced defenses against oxidative insult.
23 This could potentially result in increased susceptibility to inhaled oxidants and radicals.
24 However, data on genetic susceptibility to NO2 are only beginning to emerge and, while it
25 remains plausible that there are genetic factors that can influence health responses to NC>2, the
26 few available studies do not provide specific support for genetic susceptibility to NC>2 exposure
27 (ISA, section 4.3.4).
August 2008 - Draft 15
-------
1 3.5 SUSCEPTIBILITY: GENDER
2 As reported in the ISA, a limited number of NC>2 studies have stratified results by gender.
3 The results of these studies were mixed, and the ISA does not draw conclusions regarding the
4 potential for gender to confer susceptibility to the effects of NC>2 (ISA, section 4.3.3).
5 3.6 VULNERABILITY: PROXIMITY (ON OR NEAR) TO ROADWAYS
6 The ISA includes discussion of vulnerable populations that experience increased NC>2
7 exposures on or near roadways (ISA, section 4.3.6). Large gradients in NOx concentrations near
8 roadways lead to increased exposures for individuals residing, working, or attending school in
9 the vicinity. Many studies find that indoor, personal, and outdoor NC>2 levels are strongly
10 associated with proximity to traffic or to traffic density (ISA, section 4.3.6). Due to high air
11 exchange rates, NC>2 levels inside a vehicle could rapidly approach levels outside the vehicle
12 during commuting (ISA, section 4.3.6). Mean in-vehicle NC>2 levels are between 2 and 3 times
13 ambient levels measured at fixed sites nearby (ISA, section 4.3.6). Therefore, individuals with
14 occupations that require them to be in traffic or close to traffic (e.g., bus and taxi drivers,
15 highway patrol officers, toll collectors) and individuals with long commutes could be exposed to
16 relatively high levels of NC>2 compared to ambient levels. Due to the high peak exposures while
17 driving, total personal exposure could be underestimated if exposures while commuting are not
18 considered.
19 3.7 VULNERABILITY: SOCIOECONOMIC STATUS
20 The ISA discusses evidence that socioeconomic status (SES) modifies the effects of air
21 pollution (section 4.3.6). Many recent studies examined modification by SES indicators on the
22 association between mortality and PM or other indices such as traffic density, distance to
23 roadway, or a general air pollution index (ISA, section 4.3.6). SES modification of NC>2
24 associations has been examined in fewer studies. For example, in a study conducted in Seoul,
25 South Korea, community-level SES indicators modified the association of air pollution with ED
26 visits for asthma. Of the five criteria air pollutants evaluated, NC>2 showed the strongest
27 association in lower SES districts compared to high SES districts (Kim et al., 2007). In addition,
28 Clougherty et al. (2007) evaluated exposure to violence (a chonic stressor) as a modifier of the
29 effect of traffic-related air pollutants, including NC>2, on childhood asthma. The authors reported
August 2008 - Draft 16
-------
1 an elevated risk of asthma with a 4.3-ppb increase in NO2 exposure solely among children with
2 above-median exposure to violence in their neighborhoods. Although these recent studies have
3 evaluated the impact of SES on vulnerability to NO2, they are too few in number to draw
4 definitive conclusions (ISA, section 5.3.2.8).
5 3.8 NUMBER OF SUSCEPTIBLE/VULNERABLE INDIVIDUALS
6 The population potentially affected by NO2 is large. A considerable fraction of the
7 population resides, works, or attends school near major roadways, and these individuals are
8 likely to have increased exposure to NO2 (ISA, section 4.4). Based on data from the American
9 Housing Survey, approximately 36 million individuals live within 300 feet (-90 meters) of a
10 four-lane highway, railroad, or airport (ISA, section 4.4). Furthermore, in California, 2.3% of
11 schools with a total enrollment of more than 150,000 students were located within -500 feet of
12 high-traffic roads, with a higher proportion of non-white and economically disadvantaged
13 students attending those schools (ISA, section 4.4). Of this population, those with physiological
14 susceptibility will have even greater risks of health effects related to NO2. In the United States,
15 approximately 10% of adults and 13% of children have been diagnosed with asthma, and 6% of
16 adults have been diagnosed with COPD (ISA, section 4.4). The prevalence and severity of
17 asthma is higher among certain ethnic or racial groups such as Puerto Ricans, American Indians,
18 Alaskan Natives, and African Americans (ISA, section 4.4). Furthermore, a higher prevalence of
19 asthma among persons of lower SES and an excess burden of asthma hospitalizations and
20 mortality in minority and inner-city communities have been observed (ISA, section 4.4). In
21 addition, population groups based on age also comprise substantial segments of the population
22 that may be potentially at risk for NO2-related health impacts. Based on U.S. census data from
23 2000, about 72.3 million (26%) of the U.S. population are under 18 years of age, 18.3 million
24 (7.4%) are under 5 years of age, and 35 million (12%) are 65 years of age or older. Hence, large
25 proportions of the U.S. population are in age groups that are likely to have increased
26 susceptibility and vulnerability for health effects from ambient NO2 exposure. The considerable
27 size of the population groups at risk indicates that exposure to ambient NO2 could have a
28 significant impact on public health in the United States.
August 2008 - Draft 17
-------
4. HEALTH EFFECTS
2 4.1 INTRODUCTION
3 The ISA, along with its associated annexes, provides a comprehensive review and
4 assessment of the scientific evidence related to the health effects associated with NC>2 exposures.
5 For these health effects, the ISA characterizes judgments about causality with a hierarchy (for
6 discussion see ISA, section 1.3) that contains the following five levels.
7 • Sufficient to infer a causal relationship
8 • Sufficient to infer a likely causal relationship (i.e., more likely than not)
9 • Suggestive but not sufficient to infer a causal relationship
10 • Inadequate to infer the presence or absence of a causal relationship
11 • Suggestive of no causal relationship
12 Judgments about causality are informed by a series of criteria that are based on those set forth by
13 Sir Austin Bradford Hill in 1965 (ISA, table 1.3-1). These criteria include strength of the
14 observed association, availability of experimental evidence, consistency of the observed
15 association, biological plausibility, coherence of the evidence, temporal relationship of the
16 observed association, and the presence of an exposure-response relationship. The judgments of
17 the ISA, along with the rationale supporting those judgments, are presented below.
18 4.2 ADVERSE RESPIRATORY EFFECTS FOLLOWING SHORT-TERM
19 EXPOSURES
20 4.2.1 Overview
21 The ISA concludes that, taken together, recent studies provide scientific evidence that is
22 sufficient to infer a likely causal relationship between short-term NC>2 exposure and adverse
23 effects on the respiratory system (ISA, section 5.3.2.1). This finding is supported by the large
24 body of recent epidemiologic evidence as well as findings from human and animal experimental
25 studies. These epidemiologic and experimental studies encompass a number of endpoints
26 including ED visits and hospitalizations, respiratory symptoms, airway hyperresponsiveness,
27 airway inflammation, and lung function. Effect estimates from epidemiologic studies conducted
28 in the United States and Canada generally indicate a 2-20% increase in risks for ED visits and
August 2008 - Draft 18
-------
1 hospital admissions and higher risks for respiratory symptoms (ISA, section 5.4). The findings
2 relevant to these endpoints, which provide the rationale to support the judgment of a likely causal
3 relationship, are described in more detail below.
4 4.2.2 Respiratory Emergency Department Visits and Hospitalizations
5 Epidemiologic evidence exists for positive associations of short-term ambient NC>2
6 concentrations below the current NAAQS with increased numbers of ED visits and hospital
7 admissions for respiratory causes, especially asthma (ISA, section 5.3.2.1). Total respiratory
8 causes for ED visits and hospitalizations typically include asthma, bronchitis and emphysema
9 (collectively referred to as COPD), pneumonia, upper and lower respiratory infections, and other
10 minor categories. Temporal associations between ED visits or hospital admissions for respiratory
11 diseases and ambient levels of NO2 have been the subject of over 50 peer-reviewed research
12 publications since the last review of the NO2 NAAQS. These studies have examined morbidity
13 in different age groups and have often utilized multi-pollutant models to evaluate potential
14 confounding effects of co-pollutants. Associations are particularly consistent among children
15 and older adults (65+ years) when all respiratory outcomes are analyzed together (ISA, figures
16 3.1-8 and 3.1-9) and among children and subjects of all ages for asthma admissions (ISA, figures
17 3.1-12 and 3.1-13). When examined with co-pollutant models, associations of NC>2 with
18 respiratory ED visits and hospital admissions were generally robust and independent of the
19 effects of co-pollutants (ISA, figures 3.1-10 and 3.1-11). The plausibility and coherence of these
20 effects are supported by experimental (i.e., toxicologic and controlled human exposure) studies
21 that evaluate host defense and immune system changes, airway inflammation, and airway
22 responsiveness (see subsequent sections of this document and ISA, section 5.3.2.1).
23 Of the ED visit and hospital admission studies reviewed in the ISA, 6 key studies were
24 conducted in the United States (ISA, table 5.4-1). Of these 6 studies, 4 evaluated associations
25 with NO2 using multi-pollutant models (Peel et al., 2005 and Tolbert et al., 2007 in Atlanta; New
26 York Department of Health (NYDOH), 2006 and Ito et al., 2007 in New York City) while 2
27 studies used only single pollutant models (Linn et al., 2000; Jaffe et al., 2003). In the study by
28 Peel and colleagues, investigators evaluated ED visits among all ages in Atlanta, GA during the
29 period of 1993 to 2000. Using single pollutant models, the authors reported a 2.4% (95% CI:
30 0.9, 4.1) increase in respiratory ED visits associated with a 30-ppb increase in 1-h max NO2
August 2008 - Draft 19
-------
1 concentrations. For asthma visits, a 4.1% (95% CI: 0.8%, 7.6%) increase was detected in
2 individuals 2 to 18 years of age. Tolbert and colleagues reanalyzed these data with 4 additional
3 years of information and found essentially similar results in single pollutant models (2.0%
4 increase, 95% CI: 0.5, 3.3). This same study found that the associations were positive, but not
5 statistically-significant, in multi-pollutant models that included PMio or ozone (Os). In the study
6 conducted by the New York Department of Health, investigators evaluated asthma ED visits in
7 Bronx and Manhattan, New York over the period of January, 1999 to November, 2000. In
8 Bronx, the authors found a 6% (95% CI: 1%-10%) increase in visits per 20 ppb increase in 24-h
9 average concentrations of NO2 and a 7% increase in visits per 30 ppb increase in daily 1-h
10 maximum concentrations. These effects were not statistically-significant in 2-pollutant models
11 that included PM2.5 or SO2. In Manhattan, the authors found non-significant decreases (3% for
12 24-h and a 2% for daily 1-h maximum) in ED visits associated with increasing NC>2. In the study
13 by Ito and colleagues, investigators evaluated ED visits for asthma in New York City during the
14 years 1999 to 2002. The authors found a 12 % (95% CI: 7%, 15%) increase in risk per 20 ppb
15 increase in 24-h ambient NC>2. Risk estimates were robust and remained statistically significant
16 in multi-pollutant models that included PM2.5, 63, CO, and 862.
17 4.2.3 Respiratory Symptoms
18 Evidence for associations between NC>2 and respiratory symptoms is derived primarily
19 from the epidemiologic literature, although the experimental evidence for airway inflammation
20 and immune system effects (described in the ISA, section 3.1 and summarized in subsequent
21 sections of this document) does provide some plausibility and coherence for the epidemiologic
22 results (ISA, section 5.3.2.1). Consistent evidence has been observed for an association of
23 respiratory effects with indoor and personal NC>2 exposures in children (ISA, sections 3.1.5.1 and
24 5.3.2.1) and with ambient levels of NC>2 as measured by community monitors (ISA, sections
25 3.1.4.2 and 5.3.2.1, see Figure 3.1-6). In the results of multi-pollutant models, NC>2 associations
26 in multicity studies are generally robust to adjustment for co-pollutants including O3, CO, and
27 PMio (ISA, section 5.3.2.1 and Figure 3.1-7). Specific studies of respiratory symptoms are
28 discussed in more detail below.
29
30
August 2008 - Draft 20
-------
1 Studies of Ambient NO 2
2 Epidemiologic studies using community ambient monitors have found associations
3 between ambient NO2 concentrations and respiratory symptoms (ISA, sections 3.1.4.2 and
4 5.3.2.1, Figure 3.1-6) in cities where NO2 concentrations were within the range of 24-h average
5 concentrations observed in recent years. Several studies have been published since the last
6 review of the NO2 NAAQS including single-city studies (e.g., Ostro et al., 2001; Delfino et al.,
7 2002) and multi-city studies in urban areas covering the continental United States and southern
8 Ontario (Schwartz et al., 1994; Mortimer et al., 2002; Schildcrout et al., 2006). The multi-city
9 studies are discussed in more detail below.
10 Schwartz el at (1994) studied 1,844 schoolchildren, followed for 1 year, as part of the Six
11 Cities Study that included the cities of Watertown, MA, Baltimore, MD, Kingston-Harriman,
12 TN, Steubenville, OH, Topeka, KS, and Portage, WI. Respiratory symptoms were recorded
13 daily. The authors reported a significant association between 4-day mean NO2 levels and
14 incidence of cough among all children in single-pollutant models, with an odds ratio (OR) of
15 1.61 (95% CI: 1.08, 2.43) standardized to a 20-ppb increase in NO2. The incidence of cough
16 increased up to approximately mean NO2 levels (-13 ppb) (p = 0.01), after which no further
17 increase was observed. The significant association between cough and 4-day mean NO2 level
18 remained unchanged in models that included O3 but lost statistical significance in two-pollutant
19 models that included PMio (OR = 1.37 [95% CI: 0.88, 2.13]) or SO2 (OR = 1.42 [95% CI: 0.90,
20 2.28]).
21 Mortimer et al. (2002) studied the risk of asthma symptoms among 864 asthmatic
22 children in New York City, NY, Baltimore, MD, Washington, DC, Cleveland, OH, Detroit, MI,
23 St Louis, MO, and Chicago, IL. Subjects were followed daily for four 2-week periods over the
24 course of nine months with morning and evening asthma symptoms and peak flow recorded.
25 The greatest effect was observed for morning symptoms using a 6-day moving average, with a
26 reported OR of 1.48 (95% CI: 1.02, 2.16) per 20 ppb increase in NO2. Although the magnitudes
27 of effect estimates were generally robust in multi-pollutant models that included O3 (OR for 20-
28 ppb increase in NO2 = 1.40 [95% CI: 0.93, 2.09]), O3 and SO2 (OR for NO2 = 1.31 [95% CI:
29 0.87, 2.09]), or O3, SO2, and PMio (OR for NO2 = 1.45 [95% CI: 0.63, 3.34]), they were not
30 statistically-significant.
August 2008 - Draft 21
-------
1 Schildcrout et al. (2006) investigated the association between ambient NO2 and
2 respiratory symptoms and rescue inhaler use as part of the CAMP study. The study reported on
3 990 asthmatic children living within 50 miles of an NC>2 monitor in Boston, MA, Baltimore, MD,
4 Toronto, ON, St. Louis, MO, Denver, CO, Albuquerque, NM, or San Diego, CA. Symptoms and
5 use of rescue medication were recorded daily, resulting in each subject having an approximate
6 average of two months of data. The authors reported the strongest association between NO2 and
7 increased risk of cough for a 2-day lag, with an OR of 1.09 (95% CI: 1.03, 1.15) for each 20-ppb
8 increase in NO2 occurring 2 days before measurement. Multi-pollutant models that included CO,
9 PMio, or SO2 produced similar results (ISA, Figure 3.1-5, panel A). Additionally, increased NO2
10 exposure was associated with increased use of rescue medication, with the strongest association
11 for a 2-day lag, both for single- and multi-pollutant models (e.g., for an increase of 20-ppb NO2
12 in the single-pollutant model, the RR for increased inhaler usage was 1.05 (95% CI: 1.01, 1.09).
13 Studies of Indoor NO2
14 Evidence supporting increased respiratory morbidity following NO2 exposures is also
15 found in studies of indoor NO2 (ISA, section 3.1.4.1). For example, in a randomized
16 intervention study in Australia (Pilotto et al., 2004), students attending schools that switched out
17 unvented gas heaters, a major source of indoor NO2, experienced a decrease in both levels of
18 NO2 and in respiratory symptoms (e.g., difficulty breathing, chest tightness, and asthma attacks)
19 compared to students in schools that did not switch out unvented gas heaters (ISA, section
20 3.1.4.1). An earlier indoor study by Pilotto and colleagues (1997) also found that students in
21 classrooms with higher levels of NO2 had higher rates of respiratory symptoms (e.g., sore thoat,
22 cold) and absenteeism than students in classrooms with lower levels of NO2. This study detected
23 a significant concentration-response relationship, strengthening the argument that NO2 is
24 causally related to respiratory morbidity. A number of other indoor studies conducted in homes
25 have also detected significant associations between indoor NO2 and respiratory symptoms (ISA,
26 section 3.1.4.1).
27 4.2.4 Lung Host Defenses and Immunity
28 Impaired host-defense systems and increased risk of susceptibility to both viral and
29 bacterial infections after NO2 exposures have been observed in epidemiologic, controlled human
30 exposure, and animal toxicological studies (ISA, section 3.1.1 and 5.3.2.1). A recent
August 2008 - Draft 22
-------
1 epidemiologic study (Chauhan et al., 2003) provides evidence that increased personal exposure
2 to NO2 worsened virus-associated symptoms and decreased lung function in children with
3 asthma. The limited evidence from controlled human exposure studies indicates that NO2 may
4 increase susceptibility to injury by subsequent viral challenge at exposures of as low as 0.6 ppm
5 for 3 hours in healthy adults (Frampton et al., 2002). Toxicological studies have shown that lung
6 host defenses, including mucociliary clearance and immune cell function, are sensitive to NO2
7 exposure, with effects observed at concentrations of less than 1 ppm (ISA, section 3.1.7). When
8 taken together, epidemiologic and experimental studies linking NO2 exposure with viral illnesses
9 provide coherent and consistent evidence that NO2 exposure can result in lung host defense or
10 immune system effects (ISA, sections 3.1.7 and 5.3.2.1). This group of outcomes also provides
11 some plausibility for other respiratory system effects. For example, effects on ciliary action
12 (clearance) or immune cell function (i.e. macrophage phagocytosis) could lead to the type of
13 outcomes assessed in epidemiologic studies, including respiratory illness or respiratory
14 symptoms (ISA, section 5.3.2.1).
15 4.2.5 Airway Hyperresponsiveness
16 In acute exacerbations of asthma, bronchial smooth muscle contraction occurs quickly to
17 narrow the airway in response to exposure to various stimuli including allergens or irritants.
18 Bronchoconstriction is the dominant physiological event leading to clinical symptoms and
19 interference with airflow (National Heart, Lung, and Blood Institute, 2007). Inhaled pollutants
20 such as NO2 may enhance the inherent responsiveness of the airway to a challenge by allergens
21 or nonspecific agents (ISA, section 3.1.3). In the laboratory, airway responses can be measured
22 by assessing changes in pulmonary function (e.g., decline in FEVi) or changes in the
23 inflammatory response (e.g., using markers in bronchoalveolar lavage (BAL) fluid or induced
24 sputum) (ISA, section 3.1.3).
25 The ISA (section 5.3.2.1) draws two broad conclusions regarding airway responsiveness
26 following NO2 exposure. First, the ISA concludes that NO2 exposure may enhance the
27 sensitivity to allergen-induced decrements in lung function and increase the allergen-induced
28 airway inflammatory response at exposures as low as 0.26 ppm NO2 for 30 minutes (ISA, section
29 5.3.2.1 and Figure 3.1-2). Second, exposure to NO2 has been found to enhance the inherent
30 responsiveness of the airway to subsequent nonspecific challenges in controlled human exposure
August 2008 - Draft 23
-------
1 studies (section 3.1.4.2). In general, small but significant increases in nonspecific airway
2 responsiveness were observed in the range of 0.2 to 0.3 ppm NC>2 for 30-minute exposures and at
3 0.1 ppm NC>2 for 60-minute exposures in asthmatics. This enhanced airway responsiveness
4 could have important clinical implications for asthmatics since transient increases in airway
5 responsiveness following NC>2 exposure have the potential to increase symptoms and worsen
6 asthma control (ISA, section 5.4). In addition, the ISA sites the controlled human exposure
7 literature on airway hyperresponsiveness as being supportive of the epidemiologic evidence on
8 respiratory morbidity (ISA, section 5.4). Because studies on airway hyperresponsiveness have
9 been used to identify potential health effect benchmark values and to inform the identification of
10 potential alternative standards for evaluation (see sections 4.5 and 5 of this document), more
11 detail is provided below on the specific studies that form the basis for the conclusions in the ISA
12 regarding this endpoint.
13 Folinsbee (1992) conducted a meta-analysis using individual level data from 19 clinical
14 NC>2 exposure studies measuring airway responsiveness in asthmatics (ISA, section 3.1.3.2).
15 These studies included NC>2 exposure levels between 0.1 ppm and 1.0 ppm and most of them
16 used nonspecific bronchoconstricting agents such as methacholine, carbachol, histamine, or cold
17 air. The largest effects were observed for subjects at rest. Among subjects exposed at rest, 76%
18 experienced increased airway responsiveness following exposure to NC>2 levels between 0.2 and
19 0.3 ppm. Results from an update of this meta-analysis (results combined only from nonspecific
20 responsiveness studies) are presented in the ISA (Table 3.1-3) and in Table 4-1 below.
21
22
23
24
Table 4-1. Fraction of nitrogen dioxide-exposed asthmatics with increased nonspecific
airway hyperresponsiveness
NO2 ppm
ALL EXPOSURES
EXPOSURE WITH EXERCISE
EXPOSURE AT REST
25
Values are the fraction of asthmatics (out of the total number of individuals in parenthesis)
having an increase in airway responsiveness following NC>2 versus air exposure. See table 3.1-3
in the ISA for more detail. B indicates p < 0.05 and c indicates p < 0.01.
August 2008 - Draft
24
-------
1 As noted in Table 4-1, when exposed at rest 66% of subjects experienced an increase in
2 airway responsiveness following exposure to 0.1 ppm NC>2, 67% of subjects experienced an
3 increase in airway responsiveness following exposure to NC>2 concentrations between 0.1 and
4 0.15 ppm (inclusively), 75% of subjects experienced an increase in airway responsiveness
5 following exposure to NC>2 concentrations between 0.2 and 0.3 ppm (inclusively), and 73% of
6 subjects experienced an increase in airway responsiveness following exposure to NC>2
7 concentrations above 0.3 ppm. Effects of NC>2 exposure on the direction of airway
8 responsiveness are statistically-significant at all of these levels. Because this meta-analysis
9 evaluates only the direction of the change in airway responsiveness, it is not possible to discern
10 the magnitude of the change from these data. However, the results do suggest that short-term
11 exposures to NO2 at near-ambient levels (<0.3 ppm) can alter airway responsiveness in people
12 with mild asthma (ISA, section 3.1.3.2).
13 Several studies published since the last review address the question of whether low-level
14 exposures to NC>2 enhance the response to specific allergen challenge in mild asthmatics (ISA,
15 section 3.1.3.1). These recent studies suggest that NC>2 may enhance the sensitivity to allergen-
16 induced decrements in lung function and increase the allergen-induced airway inflammatory
17 response. Strand et al. (1997) demonstrated that single 30-minute exposures to 0.26-ppm NCh
18 increased the late phase response to allergen challenge 4 hours after exposure, as measured by
19 changes in lung function. In a separate study (Strand et al., 1998), 4 daily repeated exposures to
20 0.26-ppm NO2 for 30 minutes increased both the early and late-phase responses to allergen, as
21 measured by changes in lung function. Barck et al. (2002) used the same exposure and challenge
22 protocol in the earlier Strand study (0.26 ppm for 30 min, with allergen challenge 4 hours after
23 exposure), and performed BAL 19 hours after the allergen challenge to determine NC>2 effects on
24 the allergen-induced inflammatory response. Compared with air followed by allergen, NC>2
25 followed by allergen caused an increase in the BAL recovery of polymorphonuclear (PMN) cells
26 and eosinophil cationic protein (ECP) as well as a reduction in total BAL fluid volume and cell
27 viability. ECP is released by degranulating eosinophils, is toxic to respiratory epithelial cells,
28 and is thought to play a role in the pathogenesis of airway injury in asthma. Subsequently, Barck
29 et al. (2005) exposed 18 mild asthmatics to air or 0.26 ppm NC>2 for 15 minutes on day 1,
30 followed by two 15 minute exposures separated by 1 hour on day 2, with allergen challenge after
31 exposures on both days 1 and 2. Sputum was induced before exposure on day 1 and after
August 2008 - Draft 25
-------
1 exposures (morning of day 3). Compared to air plus allergen, NC>2 plus allergen resulted in
2 increased levels of ECP in both sputum and blood and increased myeloperoxidase levels in
3 blood. All exposures in these studies (Barck et al., 2002, 2005; Strand et al., 1997, 1998) used
4 subjects at rest. They used an adequate number of subjects, included air control exposures,
5 randomized exposure order, and separated exposures by at least 2 weeks. Together, they indicate
6 the possibility for effects on allergen responsiveness in some asthmatics following brief
7 exposures to 0.26 ppm NC>2. However, other recent studies have failed to find effects using
8 similar, but not identical, approaches (ISA, section 3.1.3.1). The differing findings may relate in
9 part to differences in timing of the allergen challenge, the use of multiple versus single-dose
10 allergen challenge, the use of BAL versus sputum induction, exercise versus rest during
11 exposure, and differences in subject susceptibility (ISA, section 3.1.3.1).
12 4.2.6 Airway Inflammation
13 Effects of NC>2 on airway inflammation have been observed in controlled human
14 exposure and animal toxicological studies at higher than ambient levels (0.4-5 ppm). The few
15 available epidemiologic studies were suggestive of an association between ambient NC>2
16 concentrations and inflammatory response in the airway in children, though the associations
17 were inconsistent in the adult populations examined (ISA, section 3.1.2 and 5.3.2.1). Controlled
18 human exposure studies provide evidence for increased airway inflammation at NC>2
19 concentrations of <2.0 ppm. The onset of inflammatory responses in healthy subjects appears to
20 be between 100 and 200 ppm-minutes, i.e., 1 ppm for 2 to 3 hours (ISA, Figure 3.1-1). Increases
21 in biological markers of inflammation were not observed consistently in healthy animals at levels
22 of less than 5 ppm; however, increased susceptibility to NC>2 concentrations of as low as 0.4 ppm
23 was observed when lung vitamin C was reduced (by diet) to levels that were <50% of normal.
24 These data provide some evidence for biological plausibility and one potential mechanism for
25 other respiratory effects, such as exacerbation of asthma symptoms and increased ED visits for
26 asthma (ISA, section 5.3.2.1).
27 4.2.7 Lung Function
28 Recent epidemiologic studies that examined the association between ambient NC>2
29 concentrations and lung function in children and adults generally produced inconsistent results
30 (ISA, sections 3.1.5.1 and 5.3.2.1). Controlled human exposure studies generally did not find
August 2008 - Draft 26
-------
1 direct effects of NO2 on lung function in healthy adults at levels as high as 4.0 ppm (ISA, section
2 5.3.2.1). For asthmatics, the direct effects of NO2 on lung function also have been inconsistent at
3 exposure concentrations of less than 1 ppm NO2.
4 4.2.8 Conclusions and Coherence of Evidence for Short-Term Respiratory Effects
5 As noted previously, the ISA concludes that the findings of epidemiologic, controlled
6 human exposure, and animal toxicological studies provide evidence that is sufficient to infer a
7 likely causal relationship for respiratory effects following short-term NO2 exposure (ISA,
8 sections 3.1.7 and 5.3.2.1). The ISA (section 5.4) concludes that the strongest evidence for an
9 association between NO2 exposure and adverse human health effects comes from epidemiologic
10 studies of respiratory symptoms, ED visits, and hospital admissions. These studies include panel
11 and field studies, studies that control for the effects of co-occurring pollutants, and studies
12 conducted in areas where the whole distribution of ambient 24-h average NO2 concentrations
13 was below the current NAAQS level of 0.053 ppm (53 ppb) (annual average). The effect
14 estimates from the U.S. and Canadian studies generally indicate a 2-20% increase in risks for ED
15 visits and hospital admissions. Risks associated with respiratory symptoms are generally higher
16 (ISA, section 5.4).
17 Overall, the epidemiologic evidence for respiratory effects can be characterized as
18 consistent, in that associations are reported in studies conducted in numerous locations with a
19 variety of methodological approaches. Considering this large body of epidemiologic studies
20 alone, the findings are also coherent in the sense that the studies report associations with
21 respiratory health outcomes that are logically linked together. In addition, a number of these
22 associations are statistically-significant, particularly the more precise effect estimates (ISA,
23 section 5.3.2.1). These epidemiologic studies are supported by evidence from toxicological and
24 controlled human exposure studies, particularly by controlled human exposure studies that
25 evaluate airway hyperresponsiveness in asthmatic individuals (ISA, section 5.4). Together, the
26 epidemiologic and experimental data sets form a plausible, consistent, and coherent description
27 of a relationship between NO2 exposures and an array of adverse health effects that range from
28 the onset of respiratory symptoms to hospital admission.
29 However, as noted in the ISA (section 5.4), it is difficult to determine "the extent to
30 which NO2 is independently associated with respiratory effects or if NO2 is a marker for the
August 2008 - Draft 27
-------
1 effects of another traffic-related pollutant or mix of pollutants." On-road vehicle exhaust
2 emissions are a nearly ubiquitous source of combustion pollutant mixtures that include NOX and
3 can be an important contributor to NO2 levels in near-road locations. Although this complicates
4 the efforts to quantify specific NO2-related health effects, the evidence summarized in the ISA
5 indicates that NC>2 associations generally remain robust in multi-pollutant models and supports a
6 direct effect of short-term NC>2 exposure on respiratory morbidity at ambient concentrations
7 below the current NAAQS level. The robustness of epidemiologic findings to adjustment for co-
8 pollutants, coupled with data from animal and human experimental studies, support the
9 determination that the relationship between NC>2 and respiratory morbidity is likely causal, while
10 still recognizing the relationship between NC>2 and other traffic related pollutants.
11 4.3 OTHER ADVERSE EFFECTS FOLLOWING SHORT-TERM
12 EXPOSURES
13 The ISA concludes that the epidemiologic evidence is suggestive but not sufficient to
14 infer a causal relationship between short-term exposure to NO2 and all-cause and
15 cardiopulmonary-related mortality (ISA, section 5.3.2.3). Results from several large U.S. and
16 European multi-city studies and a meta-analysis study indicate positive associations between
17 ambient NC>2 concentrations and the risk of all-cause (nonaccidental) mortality, with effect
18 estimates ranging from 0.5 to 3.6% excess risk in mortality per standardized increment (20 ppb
19 for 24-h averaging time, 30 ppb for 1-h averaging time) (ISA, section 3.3.1, Figure 3.3-2, section
20 5.3.2.3). In general, the NC>2 effect estimates were robust to adjustment for co-pollutants. Both
21 cardiovascular and respiratory mortality have been associated with increased NC>2 concentrations
22 in epidemiologic studies (ISA, Figure 3.3-3); however, similar associations were observed for
23 other pollutants, including PM and SCh. The range of risk estimates for excess mortality is
24 generally smaller than that for other pollutants such as PM. In addition, while NC>2 exposure,
25 alone or in conjunction with other pollutants, may contribute to increased mortality, evaluation
26 of the specificity of this effect is difficult. Clinical studies showing hematologic effects and
27 animal toxicological studies showing biochemical, lung host defense, permeability, and
28 inflammation changes with short-term exposures to NC>2 provide limited evidence of plausible
29 pathways by which risks of mortality may be increased, but no coherent picture is evident at this
30 time (ISA, section 5.3.2.3).
August 2008 - Draft 28
-------
1 The ISA concludes that the available evidence on cardiovascular health effects following
2 short-term exposure to NC>2 is inadequate to infer the presence or absence of a causal relationship
3 at this time (ISA, section 5.3.2.2). Evidence from epidemiologic studies of heart rate variability,
4 repolarization changes, and cardiac rhythm disorders among heart patients with ischemic cardiac
5 disease are inconsistent (ISA, section 5.3.2.2). In most studies, associations with PM were found
6 to be similar or stronger than associations with NC>2. Generally positive associations between
7 ambient NC>2 concentrations and hospital admissions or ED visits for cardiovascular disease have
8 been reported in single-pollutant models (ISA, section 5.3.2.2); however, most of these effect
9 estimate values were diminished in multi-pollutant models that also contained CO and PM
10 indices (ISA, section 5.3.2.2). Mechanistic evidence of a role for NC^in the development of
11 cardiovascular diseases from studies of biomarkers of inflammation, cell adhesion, coagulation,
12 and thombosis is lacking (ISA, section 5.3.2.2). Furthermore, the effects of NC>2 on various
13 hematological parameters in animals are inconsistent and, thus, provide little biological
14 plausibility for effects of NC>2 on the cardiovascular system (ISA, section 5.3.2.2).
15 4.4 ADVERSE EFFECTS FOLLOWING LONG-TERM EXPOSURES
16 4.4.1 Respiratory Morbidity
17 The ISA concludes that overall, the epidemiologic and experimental evidence is
18 suggestive but not sufficient to infer a causal relationship between long-term NC>2 exposure and
19 respiratory morbidity (ISA, section 5.3.2.4). The available database evaluating the relationship
20 between respiratory illness in children and long-term exposures to NC>2 has increased since the
21 last review of the NC>2 NAAQS. A number of epidemiologic studies have examined the effects
22 of long-term exposure to NC>2 and reported positive associations with decrements in lung
23 function and partially irreversible decrements in lung function growth (ISA, section 3.4.1, figures
24 3.4-1 and 3.4-2). Specifically, results from the California-based Children's Health Study, which
25 evaluated NC>2 exposures in children over an 8-year period, demonstrated deficits in lung
26 function growth (Gauderman et al., 2004). This effect has also been observed in Mexico City,
27 Mexico (Rojas-Martinez et al., 2007a,b) and in Oslo, Norway (Oftedal et al., 2008), with
28 decrements ranging from 1 to 17.5 ml per 20- ppb increase in annual NO2 concentration. Similar
29 associations have been found for PM, Os, and proximity to traffic (<500 m), though these studies
30 did not report the results of co-pollutant models. The high correlation among traffic-related
August 2008 - Draft 29
-------
1 pollutants makes it difficult to accurately estimate independent effects in these long-term
2 exposure studies (ISA, section 5.3.2.4). With regard to asthma incidence and long-term NC>2,
3 two major cohort studies, the Children's Health Study (Gauderman et al., 2005) and a birth
4 cohort study in the Netherlands (Brauer et al., 2007), observed significant associations.
5 However, several other studies failed to find consistent associations between long-term NC>2
6 exposure and asthma outcomes (ISA, section 5.3.2.4). Similarly, epidemiologic studies
7 conducted in the United States and Europe have produced inconsistent results regarding an
8 association between long-term exposure to NC>2 and respiratory symptoms (ISA, sections 3.4.3
9 and 5.3.2.4). While some positive associations were noted, a large number of symptom
10 outcomes were examined and the results across specific outcomes were inconsistent (ISA,
11 section 5.3.2.4).
12 Animal toxicological studies may provide biological plausibility for the chonic effects of
13 NC>2 that have been observed in epidemiologic studies (ISA, sections 3.4.5 and 5.3.2.4). The
14 main biochemical targets of NC>2 exposure appear to be antioxidants, membrane polyunsaturated
15 fatty acids, and thiol groups. NC>2 effects include changes in oxidant/antioxidant homeostasis
16 and chemical alterations of lipids and proteins. Lipid peroxidation has been observed at NC>2
17 exposures as low as 0.04 ppm for 9 months and at exposures of 1.2 ppm for 1 week, suggesting
18 lower effect thesholds with longer durations of exposure. Other studies showed decreases in
19 formation of key arachidonic acid metabolites in AMs following NC>2 exposures of 0.5 ppm.
20 NO2 has been shown to increase collagen synthesis rates at concentrations as low as 0.5 ppm.
21 This could indicate increased total lung collagen, which is associated with pulmonary fibrosis, or
22 increased collagen turnover, which is associated with remodeling of lung connective tissue.
23 Morphological effects following chonic NC>2 exposures have been identified in animal studies
24 that link to these increases in collagen synthesis and may provide plausibility for the deficits in
25 lung function growth described in epidemiologic studies (ISA, section 3.4.5).
26 4.4.2 Mortality
27 The ISA concludes that the epidemiologic evidence is inadequate to infer the presence or
28 absence of a causal relationship between long-term exposure to NC>2 and mortality (ISA, section
29 5.3.2.6). In the United States and European cohort studies examining the relationship between
30 long-term exposure to NC>2 and mortality, results have been inconsistent (ISA, section 5.3.2.6).
August 2008 - Draft 30
-------
1 Further, when associations were suggested, they were not specific to NC>2 but also implicated PM
2 and other traffic indicators. The relatively high correlations reported between NC>2 and PM
3 indices make it difficult to interpret these observed associations at this time (ISA, section
4 5.3.2.6).
5 4.4.3 Other Long-Term Effects
6 The ISA concludes that the available epidemiologic and toxicological evidence is
7 inadequate to infer the presence or absence of a causal relationship for carcinogenic,
8 cardiovascular, and reproductive and developmental effects related to long-term NCh exposure
9 (ISA, section 5.3.2.5). Epidemiologic studies conducted in Europe have shown an association
10 between long-term NC>2 exposure and increased incidence of cancer (ISA, section 5.3.2.5).
11 However, the animal toxicological studies have provided no clear evidence that NC>2 acts as a
12 carcinogen (ISA, section 5.3.2.5). The very limited epidemiologic and toxicological evidence
13 does not suggest that long-term exposure to NC>2 has cardiovascular effects (ISA, section
14 5.3.2.5). The epidemiologic evidence is not consistent for associations between NC>2 exposure
15 and growth retardation; however, some evidence is accumulating for effects on preterm delivery
16 (ISA, section 5.3.2.5). Scant animal evidence supports a weak association between NC>2
17 exposure and adverse birth outcomes and provides little mechanistic information or biological
18 plausibility for the epidemiologic findings.
19 4.5 RELEVANCE OF SPECIFIC HEALTH EFFECTS TO THE NO2 RISK
20 CHARACTERIZATION
21 4.5.1 Overview
22 As described previously, the ISA characterizes judgments about causality with a hierarchy
23 (for discussion see ISA, section 1.3) that contains the following five levels.
24 • Sufficient to infer a causal relationship
25 • Sufficient to infer a likely causal relationship (i.e., more likely than not)
26 • Suggestive but not sufficient to infer a causal relationship
27 • Inadequate to infer the presence or absence of a causal relationship
28 • Suggestive of no causal relationship
August 2008 - Draft 31
-------
1 For purposes of the quantitative characterization of NC>2 health risks, staff has judged it
2 appropriate to focus on endpoints for which the ISA concludes that the available evidence is
3 sufficient to infer either a causal or a likely causal relationship. The only endpoint meeting
4 either of these criteria is respiratory morbidity following short-term NC>2 exposure. The ISA
5 (section 5.4) concludes that the "epidemiologic, controlled human exposure and animal
6 toxicologic studies provided evidence that short-term NC>2 exposures can result in adverse
7 impacts to public health at current ambient concentrations (mean 24-h avg concentrations
8 ranging from 3-70 ppb [Table 5.4-1]). In particular, a set of coherent and consistent respiratory
9 health outcomes were associated with short-term NC>2 exposures including exacerbated asthma
10 and other respiratory symptoms, increased airway hyperresponsiveness, inflammation, impaired
11 host defense, aggravated viral infections, and increased ED visits and hospital admissions."
12 Therefore, for purposes of characterizing health risks associated with NC>2, we have focused on
13 respiratory morbidity endpoints that have been associated with short-term NC>2 exposures. Other
14 endpoints (e.g., long-term effects) will be considered as part of the evidence-based evaluation of
15 potential alternative standards during the rulemaking stage of the NAAQS review. In evaluating
16 the appropriateness of specific endpoints for use in the NC>2 risk characterization, we have
17 considered both epidemiologic and controlled human exposure studies.
18 4.5.2 Epidemiology
19 The ISA characterizes the epidemiologic evidence for respiratory effects as consistent, in
20 that associations are reported in studies conducted in numerous locations and with a variety of
21 methodological approaches (ISA, section 5.3.2.1). The findings are also coherent in the sense
22 that the studies report associations with respiratory health outcomes that are logically linked
23 together (ISA, section 5.3.2.1). When the epidemiologic literature is considered as a whole,
24 there are generally positive associations between NC>2 and respiratory symptoms, hospitalization,
25 and ED visits. A number of these associations are statistically significant, particularly the more
26 precise effect estimates (ISA, section 5.3.2.1). However, the ISA (section 5.4) offers the
27 following caveat to consider when interpreting the epidemiologic results: "It is difficult to
28 determine from these new studies the extent to which NC>2 is independently associated with
29 respiratory effects or if NC>2 is a marker for the effects of another traffic-related pollutant or mix
30 of pollutants (see Section 5.2.2 for more details on exposure issues). A factor contributing to
August 2008 - Draft 32
-------
1 uncertainty in estimating the NCVrelated effect from epidemiologic studies is that NC>2 is a
2 component of a complex air pollution mixture from traffic related sources that include CO and
3 various forms of PM." These caveats should be considered when interpreting a quantitative NO2
4 risk estimate based on the epidemiology literature. Despite these uncertainties, the ISA (section
5 5.4) concludes that, "Although this complicates the efforts to disentangle specific NO2-related
6 health effects, the evidence summarized in this assessment indicates that NC>2 associations
7 generally remain robust in multi-pollutant models and supports a direct effect of short-term NC>2
8 exposure on respiratory morbidity at ambient concentrations below the current NAAQS. The
9 robustness of epidemiologic findings to adjustment for copollutants, coupled with data from
10 animal and human experimental studies, support a determination that the relationship between
11 NO2 and respiratory morbidity is likely causal, while still recognizing the relationship between
12 NC>2 and other traffic related pollutants." Therefore, epidemiologic studies have been judged to
13 be an appropriate basis for a quantitative assessment of the risks associated with ambient NC>2.
14 When selecting specific epidemiologic studies on which to base the risk assessment, staff
15 has considered several factors. First, we have judged that studies conducted in the United States
16 are preferable to those conducted outside the United States given the potential for effect
17 estimates to be impacted by factors such as the ambient pollutant mix, the placement of
18 monitors, activity patterns of the population, and characteristics of the healthcare system.
19 Second, we judged that studies of ambient NO2 are preferable to those of indoor NC>2 given that
20 studies of indoor NC>2 focus on exposures in locations with indoor sources of NC>2. These indoor
21 sources can result in exposure patterns, NC>2 levels, and co-pollutants that are different from
22 those typically associated with ambient NC>2. Third, we judged it appropriate to focus on studies
23 of ED visits and hospital admissions. When compared to studies of respiratory symptoms, the
24 public health significance of ED visits and hospital admissions are less ambiguous (e.g., because
25 of the potential disconnect between health outcomes and subjective symptom ratings). In
26 addition, baseline incidence data are more readily available for these endpoints. Finally, we
27 judged it appropriate to focus on studies that evaluated NC>2 health effect associations using both
28 single- and multi-pollutant models. Taking these factors into consideration, we have chosen to
29 focus on the studies by Peel and colleagues (2005) and by Tolbert and colleagues (2007) in
30 Atlanta, Georgia. The epidemiology-based risk assessment is described in more detail in
31 subsequent sections of this document.
August 2008 - Draft 33
-------
1 4.5.3 Controlled Human Exposure Studies
2 Controlled human exposure studies have addressed the consequences of short-term (e.g.,
3 30-minutes to several hours) NC>2 exposures for a number of health endpoints including airway
4 responsiveness, host defense and immunity, inflammation, and lung function (ISA, section 3.1).
5 In identifying health endpoints from controlled human exposure studies on which to focus the
6 characterization of NC>2 health risks, staff judges it appropriate to focus on endpoints that occur
7 at or near ambient levels of NC>2 and endpoints that are of clinical significance. With regard to
8 the NC>2 levels at which different effects have been documented, the ISA concludes that 1) in
9 asthmatics NO2 may increase the allergen-induced airway inflammatory response at exposures as
10 low as 0.26-ppm for 30 min (ISA, Figure 3.1-2) and NC>2 exposures between 0.2 and 0.3 ppm for
11 30 minutes or 0.1 ppm for 60-minutes can result in small but significant increases in nonspecific
12 airway responsiveness (ISA, section 5.3.2.1); 2) limited evidence indicates that NC>2 may
13 increase susceptibility to injury by subsequent viral challenge following exposures of 0.6-1.5
14 ppm for 3 hours; 3) evidence exists for increased airway inflammation at NC>2 concentrations less
15 than 2.0 ppm; and 4) the direct effects of NC>2 on lung function in asthmatics have been
16 inconsistent at exposure concentrations below 1 ppm (ISA, section 5.3.2.1). The ISA notes that
17 epidemiologic studies have reported health effects associations in areas reporting maximum
18 ambient concentrations from 100 to 300 ppb (ISA, Tables 5.3-2 and 5.3-3). Therefore, of the
19 health effects caused by NC>2 in controlled human exposure studies, the only effect identified by
20 the ISA to occur at or near ambient levels is airway hyperresponsiveness in asthmatics.
21 The airway response can vary dramatically between individuals, ranging from mild to
22 severe and spanning orders of magnitude (ISA, section 4.3.1.1). When discussing the clinical
23 significance of NCVrelated airway hyperresponsiveness, the ISA concludes that "transient
24 increases in airway responsiveness following NC>2 exposure have the potential to increase
25 symptoms and worsen asthma control" (ISA, sections 3.1.3 and 5.4). That this effect could have
26 public health implications is suggested by the large size of the asthmatic population in the United
27 States (see above and ISA, Table 4.4-1). In addition, NC>2 effects on airway responsiveness are
28 part of the body of experimental evidence that provides plausibility and coherence for the effects
29 observed on hospital admissions and ED visits in epidemiologic studies (ISA, section 5.3.2.1).
30 Therefore, although studies on several of the endpoints evaluated in controlled human exposure
31 studies provide qualitative support for the ability of NC>2 to cause adverse effects on respiratory
August 2008 - Draft 34
-------
1 health, the focus for purpose of quantifying risks associated with ambient NO2 is airway
2 responsiveness (see below).
3 Because many of the studies of airway responsiveness evaluate only a single level of NO2
4 and because of methodological differences between the studies, staff has judged that the data are
5 not sufficient to derive an exposure-response relationship in the range of interest. Therefore, the
6 most appropriate approach to characterizing risks based on the controlled human exposure
7 evidence for airway responsiveness is to compare estimated NO2 air quality and exposure levels
8 with potential health effect benchmark levels. Estimates of hourly peak air quality
9 concentrations and personal exposures to ambient NO2 concentrations at and above specified
10 potential health effect benchmark levels provide some perspective on the potential public health
11 impacts of NO2 exposure. Staff recognizes that there is high inter-individual variability in NO2-
12 induced effects on airway responsiveness such that only a subset of asthmatic individuals
13 exposed at and above a given benchmark level may actually be expected to experience an
14 adverse effect.
15 To identify potential health effect benchmarks, staff has relied on the ISA's evaluation
16 of the NO2 human exposures studies. Controlled human exposure studies involving allergen
17 challenge in asthmatics suggest that NO2 exposure may enhance the sensitivity to allergen-
18 induced decrements in lung function and increase the allergen-induced airway inflammatory
19 response at exposures as low as 0.26-ppm NO2for 30 min (ISA, Figure 3.1-2 and section
20 5.3.2.1). Exposure to NO2 also has been found to enhance the inherent responsiveness of the
21 airway to subsequent nonspecific challenges (ISA, section 5.3.2.1). In asthmatics, small but
22 significant increases in nonspecific airway responsiveness have been observed in the range of 0.2
23 to 0.3 ppm NO2 for 30 minute exposures and at 0.1 ppm NO2 for 1-h exposures (ISA, section
24 5.3.2.1). Therefore, for the risk characterization, staff judges that 1-h NO2 levels in this range
25 are appropriate to consider as potential health benchmarks for comparison to air quality levels
26 and exposure estimates. To characterize health risks with respect to this range, potential health
27 effect benchmark values of 0.10 ppm, 0.20 ppm, 0.25 ppm, and 0.30 ppm have been employed to
28 reflect the lower- middle- and upper-end of the range identified in the ISA as levels at which
29 controlled human exposure studies have provided evidence for the occurrence of NO2-induced
30 airway hyperresponsiveness.
August 2008 - Draft 35
-------
1 In choosing this range, we recognize that uncertainties exist regarding the percentage of
2 asthmatics expected to experience an increase in responsiveness following NC>2 exposure and in
3 the clinical implications of such an increase. A meta-analysis presented in the ISA (see Table 4-
4 1 above) suggests that between 66% and 75% of asthmatics may experience an increase in
5 airway responsiveness following short-term NC>2 exposures in the range of 0.1 to 0.3 ppm.
6 However, this meta-analysis provides information only on the direction of the NC>2 effect and not
7 on its magnitude. In addition, the NC>2 controlled human exposure studies of airway
8 responsiveness have focused primarily on mild asthmatics. It is possible that more severely
9 affected asthmatics could experience a more severe response following NC>2 exposures in this
10 range. It is also possible that they could experience a response at lower levels of NC>2 than the
11 mild asthmatics included in these studies. However, even considering these uncertainties, staff
12 judges that the identified range of concentrations is sufficient to provide some perspective on the
13 potential public health impacts of NC>2 exposures, especially when the results of the risk
14 characterization based on airway responsiveness are considered in conjunction with the risk
15 assessment based on the epidemiology literature.
16
17
August 2008 - Draft 36
-------
i 5. IDENTIFICATION OF POTENTIAL ALTERNATIVE
2 STANDARDS FOR ANALYSIS
3 5.1 INTRODUCTION
4 The primary goals of the NO2 risk and exposure assessment described in this draft
5 document are to estimate short-term exposures and potential human health risks associated with
6 1) recent levels of ambient NO2; 2) NO2 levels associated with just meeting the current standard;
7 and 3) NO2 levels associated with just meeting potential alternative standards. This section
8 identifies potential alternative standards in terms of indicator, averaging time, form, and level
9 and provides the rationale that was used to select them.
10 5.2 INDICATOR
11 The NOX include multiple gaseous (e.g., NO2, NO) and particulate (e.g., nitrate) species.
12 In considering the appropriateness of different indicators, we note that the health effects
13 associated with particulate species of NOX have been considered within the context of the health
14 effects of ambient particles in the Agency's review of the NAAQS for PM. Thus, as discussed in
15 the integrated review plan (2007a), the current review of the NO2 NAAQS is focused on the
16 gaseous species of NOX and will not consider health effects directly associated with particulate
17 species of NOX. Of the gaseous species, EPA has historically determined it appropriate to
18 specify the indicator of the standard in terms of NO2 because the majority of the information
19 regarding health effects and exposures is for NO2. The final ISA has found that this continues to
20 be the case and, therefore, staff believes that NO2 remains the most appropriate indicator.
21 5.3 AVERAGING TIME
22 The current annual standard for NO2 was originally set in 1971 based on epidemiologic
23 studies that supported a link between adverse respiratory effects and long-term exposure to low-
24 levels of NO2. Although the quantitative basis for the annual averaging time was later called into
25 question (60 FR 52876), the annual standard was retained in the most recent review (60 FR
26 52876) for two key reasons. First, the evidence showing the most serious health effects
27 associated with long-term exposures (e.g., emphysematous-like alterations in the lung and
28 increased susceptibility to infection) came from animal studies conducted at concentrations well
August 2008 - Draft 37
-------
1 above those permitted in the ambient air by the annual standard. Second, an air quality
2 assessment conducted by EPA concluded that areas that meet the annual standard would be
3 unlikely to experience short-term peaks above levels that had been shown in controlled human
4 exposure studies to impact endpoints of potential concern (i.e., airway responsiveness).
5 The issue of averaging time will be reconsidered in the current review. As described
6 above, the ISA concludes that, when taken together, "recent studies provide scientific evidence
7 that NO2 is associated with a range of respiratory effects and is sufficient to infer a likely causal
8 relationship between short-term NO2 exposure and adverse effects on the respiratory system"
9 (ISA, section 5.3.2.1). This conclusion is based, in part, on the observation that a number of
10 epidemiologic studies have detected positive associations between short-term (e.g., 1-h, 24-h)
11 NO2 concentrations and health effects. Many of these studies have been conducted in locations
12 where long-term ambient levels of NO2 are well below the current annual standard. As a result,
13 staff has concluded that it is appropriate to consider alternative averaging times for their ability
14 to protect against health effects associated with short-term NO2 levels and/or exposures.
15 In contrast to the conclusion in the ISA concerning respiratory morbidity associated with
16 short-term exposures to NO2, the ISA concludes that the "evidence examining the effect of long-
17 term exposure to NO2 on respiratory morbidity is suggestive but not sufficient to infer a causal
18 relationship" (ISA, section 5.3.2.4). In addition, the ISA concludes that the available evidence
19 for the effect of long-term exposure to NO2 on other health outcomes (i.e., mortality, cancer,
20 cardiovascular effects, reproductive and developmental effects) is "inadequate to infer the
21 presence or absence of a causal relationship" (ISA, sections 5.3.2.5 and 5.3.2.6). As a result,
22 staff has not considered alternative long-term standards in the current assessment.
23 In considering appropriate short-term averaging times, staff has considered evidence from
24 both experimental and epidemiologic studies. New evidence from controlled human exposure
25 studies generally evaluates exposures between 30 minutes and 3 hours while epidemiologic
26 studies have used different short-term averaging periods, most commonly 1-h and 24-h (ISA,
27 section 3.1). A few epidemiologic studies have considered both 1-h and 24-h averaging times,
28 allowing comparisons to be made. The ISA reports that such comparisons failed to reveal
29 differences between effect estimates based on a 1-h averaging time versus those based on a 24-h
30 averaging time (ISA, section 5.3.2.7). Therefore, the ISA concludes that it is not possible to
31 discern whether effects observed in epidemiologic studies are attributable to average daily (or
August 2008 - Draft 38
-------
1 multiday) concentrations (24-h avg) or high, peak exposures (1-h max) (ISA, section 5.3.2.7). In
2 addition, the ISA concludes that experimental studies in both animals and humans provide
3 evidence that NO2 exposures from less than 1 hour up to 3 hours can result in respiratory effects
4 (section 5.3.2.7). Given that the epidemiologic evidence does not provide clear guidance in
5 choosing between 1-h and 24-h averaging times, and given that the experimental literature
6 provides support for the occurrence of effects following exposures of shorter duration than 24
7 hours (e.g., 1-h), staff has chosen to evaluate standards with a 1-h averaging time.
8 5.4 FORM
9 In evaluating alternative forms for the primary standard, staff recognizes that it is important
10 to have a form that 1) reflects the health risks posed by elevated NO2 concentrations and 2)
11 achieves a balance between limiting the occurrence of peak concentrations and providing a stable
12 and robust regulatory target. Consistent with judgments made in recent reviews of the PM (71
13 FR 61144) and O3 (73 FR 16436) NAAQS, staff judges that a concentration-based form for the
14 NO2 standard would better reflect health risks and would provide greater stability than a form
15 based on expected exceedances. A concentration-based form gives proportionally greater weight
16 to hours when concentrations are well above the level of the standard than to hours when the
17 concentrations are just above the standard, while an expected exceedance form would give the
18 same weight to an hour that just exceeds the standard as to an hour that greatly exceeds the
19 standard. Therefore, a concentration-based form better reflects the health risks posed by elevated
20 NO2 concentrations and, in developing potential alternative standards for consideration, we have
21 focused on standards with concentration-based forms. The most recent review of the PM
22 NAAQS (completed in 2006) judged that using a 98th percentile form averaged over 3 years
23 provides an appropriate balance between limiting the occurrence of peak concentrations and
24 providing a stable regulatory target (71 FR 61144). In that review, staff also considered other
25 forms within the range of the 95th to the 99th percentiles. In making recommendations regarding
26 the form, staff considered the impact on risk of different forms, the year-to-year stability in the
27 air quality statistic, and the extent to which different forms of the standard would allow different
28 numbers of days per year to be above the level of the standard in areas that achieve the standard.
29 Based on these considerations, staff recommended either a 98th percentile form or a 99th
30 percentile form. We have made similar judgments in identifying an appropriate range of forms
August 2008 - Draft 39
-------
1 for potential alternative NO2 standards. As a result of these judgments, we have determined it
2 appropriate to consider both the 98th and 99th percentile NO2 concentrations averaged over 3
3 years. We have judged that these percentiles, when combined with the range of alternatives
4 identified for the level of the standard (see below), offer a sufficient range of options to balance
5 the objective of providing a stable regulatory target against the objective of limiting the
6 occurrence of peak concentrations.
7 5.5 LEVEL
8 In developing an approach to formulating an appropriate range of NO2 levels for analysis,
9 staff has taken into account several considerations including the following. First, since the last
10 review of the NO2 NAAQS, a large number of published epidemiologic studies have evaluated
11 associations between respiratory morbidity and short-term levels of ambient NO2. In general,
12 these studies report positive associations and a number of these associations are statistically-
13 significant. The ISA notes that many of these studies have been conducted in locations where
14 ambient levels of NO2 are well below the level of the current NAAQS (ISA, section 5.3.2.1).
15 Second, controlled human exposure studies have detected effects of NO2 exposure on several
16 health endpoints. Of these, only airway hyperresponsiveness is associated with exposures to
17 NO2 concentrations at or near ambient levels. In fact, the NO2 exposure levels associated with
18 increased airway responsiveness overlap the maximum ambient NO2 concentrations in some
19 locations where associations with respiratory effects have been detected. Third, limitations in
20 both epidemiologic studies (e.g., confounding by co-pollutants) and controlled human exposure
21 studies (e.g., most sensitive populations likely not evaluated) suggest that an appropriate
22 approach to identifying levels for potential alternative standards is to consider both types of
23 studies. Therefore, to determine the levels that should be evaluated, staff has relied on both key
24 epidemiologic studies conducted in the United States that evaluate associations between short-
25 term levels of NO2 and respiratory morbidity (symptoms, hospital admissions, ED visits) and on
26 controlled human exposure studies that evaluate airway hyperresponsiveness following NO2
27 exposure. Figures 5-1 and 5-2 below show standardized effect estimates3 and the 98th and 99th
28 percentile concentrations of daily 1-h maximum NO2 for locations and time periods that
3 The effect estimates presented in figures 5-1 and 5-2 are for those endpoints included in figure 5.3-1 and table 5.4-
1 of the IS A.
August 2008 - Draft 40
-------
1 correspond to key U.S. epidemiologic studies identified in the ISA (see table 5.4-1 in ISA for a
2 list of key studies).
3 Of the key U.S. epidemiologic studies included in figures 5-1 and 5-2, the highest 1-h
4 NC>2 concentrations were detected in the two studies conducted in Los Angeles (Linn et al.,
5 2000; Ostro et al., 2001). For these studies, the 98th and 99th percentile 1-h daily maximum
6 concentrations of NC>2 overlap levels that the ISA concludes are associated with increased airway
7 responsiveness in controlled human exposure studies (ISA, section 5.3.2.1). Therefore, staff
8 judges that the combination of the epidemiologic studies by Linn et al. (2000) and Ostro et al.
9 (2001), as well as the meta-analysis (Folinsbee, 1992; ISA, table 3.1-3; table 4-1 of this
10 document) of controlled human exposure studies on airway responsiveness, provide an
11 appropriate basis for identifying the upper end of the range of standard levels to be considered.
12 Given that the ISA concludes that significant increases in airway responsiveness are associated
13 with short-term exposures to NO2 at 0.2 to 0.3 ppm and given that the epidemiologic studies by
14 Linn et al. (2000) and Ostro et al. (2001) are associated with 98th and 99th percentile 1-h daily
15 maximum NO2 levels that are just below (Linn et al., 2000) and just above (Ostro et al., 2001)
16 0.2 ppm (see figures 1 and 2 below), staff judges that an appropriate upper end of the range of
17 potential standard levels is a daily maximum 1-h NO2 concentration of 0.20 ppm.
August 2008 - Draft 41
-------
20
15
1-hoyr
244ioyr effect
EDAA
JC
m
10
EOAC
EORA
EDRA EDAA
-10 -|
-15 -]
-20 !
iDAA
EDAC
EDAC
1-h 98 0
1-h99. 0
178
197
HAAC
200? Pee! 2008 2006
NYC
Legend:
EDRA = Emergency department visits for respiratory disease -
EDAA = Emergency department visits for asthma — all ages
EDAC = Emergency department visits for asthma - children
HAAC = Hospital admissions for asthma - children
(to 200"
NYC
- all ages
V V
Jaffe 2003
Cte¥/Cino
Linn
LA
NYOOH 2006
NYC
Figure 5-1. NO2 effect estimates4 (95% CI) for ED visits/HA and associated 1-h daily
•>th
,th
maximum NOi levels (98 and 99 percentile values in boxes )
4Effect estimates presented in figures 5-1 and 5-2 are from single pollutant models only. The studies by Tolbert et
al, (2007); Peel et al, (2005); NYDOH (2006); Ito et al., (2007); and Delfino et al. (2002) also evaluated multi-
pollutant models. NO2 effect estimates retained statistical-significance in the study by Ito, but not in the other
studies.
5 Authors of relevant U.S. and Canadian studies were contacted and air quality statistics from the study monitor that
recorded the highest NO2 levels were requested. In cases where authors provided 1-hour daily maximum air quality
statistics, this information is presented in figures 1 and 2 (studies by Tolbert, Peel, NYDOH, Delfino). In one case
(study by Ito) authors provided 24-hour air quality data, but identified a specific monitor in AQS. We used AQS to
reconstruct the 1-hour daily maximum air quality for that monitor during the time period of the study. In three cases
(studies by Jaffe, Linn, Ostro), we were not able to identify appropriate statistics from the information provided by
the authors and the authors did not provide monitor identification information. In these cases, we attempted to
reconstruct the air quality data set for the location and time of the study using EPA's Air Quality System (AQS).
We have not yet received air quality information from any of the Canadian authors contacted and we were unable to
reconstruct the air quality data sets for the Canadian studies. Therefore, for purposes of identifying levels of
potential alternative standards, our analysis was based on these key U.S. studies.
August 2008 - Draft
42
-------
1-hour effect estimates
4-hour effect estimate
24-hour effect
2
O
4
5
6
7
8
9
10
130 -•
110 --
90 --
I 70
w
M
X
iu
*»*
c
o
50 -
30 --
2 10 -•
o
GL
,10 ..
-30
AS
AS
AS
AS
MS
V l Y J
Alpine, CA LA
Legend:
AS =
W =
C = Cough
MS = Morning symptoms
"We do not 1-h i8ih and 99th N02 levels for several of the U.S. respiratory
in 5.4-1 of the ISA. of ISA. 5.4-1} suggests that 24-h NOZ in the
by by and are lower than the 24-hour levels in U.S.
24-h in the study by Linn are to 24-h reported in other U.S. studies, and 1-h in
the study by are 1-h maximum reported in other U.S. studies. Such not
for the by it is the only study that reports 4-hour NO,
Figure 5-2. NOi effect estimates for respiratory symptoms and associated 1-h daily
•>th
,th
maximum NOi levels (98 and 99 percentile values in boxes)
In identifying additional standard levels that should be analyzed, staff has considered that
1) health effect associations in epidemiologic studies are observed in locations with 1-h daily
maximum levels of NC>2 below 0.2 ppm (i.e., 99th percentile levels in several studies are close to
0.1 ppm); 2) controlled human exposure studies that evaluate the ability of NO2 to elicit airway
hyperresponsiveness have assessed mild asthmatics and more severely affected asthmatics could
experience increased airway responsiveness at lower levels of NO2 than observed in these
studies; and 3) a meta-analysis presented in the ISA (see Table 4-1) detects statistically-
August 2008 - Draft
43
-------
1 significant effects on the direction of airway responsiveness following short-term NC>2 exposures
2 as low as 0.1 ppm. As a result of these considerations, staff judges that it would be appropriate
3 to consider additional standard levels that provide a margin of safety relative to 0.20 ppm.
4 Therefore, we will also consider daily maximum 1-h NC>2 standard levels of 0.10 ppm and 0.15
5 ppm.
6 In identifying the lower end of the range of standards that will be analyzed, staff has
7 considered the fact that the study by Delfmo et al., (2002) provides evidence for associations
8 between short-term ambient NC>2 concentrations and respiratory morbidity in a location where
9 the 98th and 99th percentile concentrations of the 1-h daily maximum levels of NC>2 were well
10 below 0.1 ppm (Delfmo et al., 2002). This study detects associations between 1-h and 8-h (only
11 8-h associations were statistically-significant) levels of NO2 and asthma symptoms in a location
12 where the 98th and 99th percentile 1-h daily maximum NC>2 concentrations were 0.050 and 0.053
13 ppm, respectively. The 8-h effect estimate in this study remained positive, but became
14 statistically non-significant, in a two-pollutant model that also included PMio. Staff judges that it
15 is appropriate to base the lower end of the range of alternative standard levels on this study by
16 Delfmo et al. (2002). Therefore, we will also consider a 1-h daily maximum standard level of
17 0.050 ppm.
August 2008 - Draft 44
-------
i 6. OVERVIEW OF APPROACHES TO ASSESSING EXPOSURES
2 AND RISKS
3
4 6.1 INTRODUCTION
5 The purpose of the assessments described in this document is to characterize exposures
6 and risks associated with recent ambient levels of NC>2, with levels associated with just meeting
7 the current NC>2 NAAQS, and with levels associated with just meeting potential alternative
8 standards (see chapter 5 of this document for discussion of potential alternative standards). To
9 characterize health risks, we have employed thee approaches. With each approach, we have
10 characterized health risks associated with the air quality scenarios of interest (i.e., recent air
11 quality unadjusted, air quality adjusted to simulate just meeting the current standard, and air
12 quality adjusted to simulate just meeting potential alternative standards). In the first approach,
13 NC>2 air quality levels have been compared to potential health effect benchmark values derived
14 from the controlled human exposure literature. In the second approach, modeled estimates of
15 actual exposures have been compared to potential health effect benchmarks. In the third
16 approach, exposure-response relationships from epidemiologic studies have been used to
17 estimate health impacts. An overview of the approaches to characterizing health risks is
18 provided below and each approach is described in more detail in subsequent sections of this
19 document and the associated appendices.
20 In the first approach, we have compared NC>2 air quality with potential health effect
21 benchmark levels for NC>2. Scenario-driven air quality analyses have been performed using
22 ambient NC>2 concentrations for the years 1995 though 2006. With this approach, NC>2 air
23 quality serves as a surrogate for exposure. All U.S. monitoring sites where NC>2 data have been
24 collected are represented by this analysis and, as such, the results generated are considered a
25 broad characterization of national air quality and human exposures that might be associated with
26 these concentrations. An advantage of this approach is its relative simplicity; however, there is
27 uncertainty associated with the assumption that NO2 air quality can serve as an adequate
28 surrogate for exposure to ambient NC>2. Actual exposures might be influenced by factors not
29 considered by this approach, such as the spatial and temporal variability in human activities.
August 2008 - Draft 45
-------
1 In the second approach, we have used an inhalation exposure model to generate more
2 realistic estimates of personal exposures. Estimates of personal exposure have been compared to
3 potential NO2 health benchmark levels. For this exposure analysis, a probabilistic approach was
4 used to model individual exposures considering the time people spend in different
5 microenvironments and the variable NO2 concentrations that occur within these
6 microenvironments across time, space, and microenvironment type. This approach to assessing
7 exposures was more resource intensive than using ambient levels as a surrogate for exposure;
8 therefore, staff has included the analysis of only one specific location in the U.S. (Atlanta
9 MSA)6. Although the geographic scope of this analysis is restricted, the approach provides
10 realistic estimates of NO2 exposures, particularly those exposures associated with important
11 emission sources of NOX and NO2, and serves to complement to the broad air quality
12 characterization.
13 For the characterization of risks in both the air quality analysis and the exposure
14 modeling analysis described above, staff has used a range of short-term potential health effect
15 benchmarks. The levels of potential benchmarks are based on NO2 exposure levels that have
16 been associated with increased airway responsiveness in asthmatics in controlled human
17 exposure studies (ISA, section 5.3.2.1; see above for discussion). Benchmark values of 100,
18 150, 200, 250, and 300 ppb have been compared to both NO2 air quality levels and to estimates
19 of NO2 exposure. When NO2 air quality is used as a surrogate for exposure, the output of the
20 analysis is an estimate of the number of times per year specific locations experience 1-h levels of
21 NO2 that exceed a particular benchmark. When personal exposures are simulated, the output of
22 the analysis is an estimate of the number of individuals at risk for experiencing daily maximum
23 1-h levels of NO2 of ambient origin that exceed a particular benchmark. An advantage of using
24 potential health effect benchmark levels to characterize health risks is that the effects observed in
25 controlled human exposure studies clearly result from NO2 exposure. This is in contrast to
26 health effects associated with NO2 in epidemiologic studies, which may also be associated with
27 pollutants that co-occur with NO2 in the ambient air. Thus, when using epidemiologic studies as
28 the basis for risk characterization, the unique contribution of NO2 to a particular health effect
6 In the document titled Risk and Exposure Assessment to Support the Review of the NO2 Primary National Ambient
Air Quality Standard: First Draft, we have presented the results of an exposure analysis for Philadelphia. Based on
CASAC comments received on that exposure analysis, we have refined our approach and applied those refinements
to the Atlanta analysis presented in this document. The original Philadelphia analysis is presented in the appendix to
this document, but has not been modified since the first draft.
August 2008 - Draft 46
-------
1 may be difficult to quantify. A disadvantage of the potential benchmark approach is that the
2 magnitude of the NC>2 effect on airway responsiveness can vary considerably from individual to
3 individual and not all asthmatics would be expected to respond to the same levels of NO2
4 exposure. Therefore, the public health impacts of NO2-induced airway hyperresponsiveness are
5 difficult to quantify.
6 In the third approach, we have estimated respiratory ED visits as a function of ambient
7 levels of NC>2 measured at a fixed-site monitor representing ambient air quality for an urban area.
8 In this approach, concentration-response functions are derived from NC>2 epidemiologic studies
9 and are used to estimate the impact of ambient levels of NC>2, as measured at a fixed-site
10 monitor, on ED visits. By focusing on a different health endpoint from the first two approaches
11 described above, this epidemiology-based approach provides additional perspective on the
12 potential public health impacts of NC>2. Relative to the approaches that use controlled human
13 exposure studies, this approach to characterizing health risks has several advantages. For
14 example, the public health significance of the effect in question (i.e., ED visits) is less
15 ambiguous in terms of its impact on an individual than in the case of airway
16 hyperresponsiveness. In addition, the concentration-response relationship reflects real-world
17 levels of NC>2 and co-pollutants present in ambient air. However, a disadvantage of this
18 approach is the ambiguity and complexity associated with quantifying the contribution of NC>2 to
19 the reported health impacts relative to the contributions of co-occurring pollutants.
20 6.2 SIMULATING THE CURRENT AND ALTERNATIVE STANDARDS
21 A primary goal of this draft of the risk and exposure assessments is to evaluate the ability
22 of the current NC>2 standard (0.053 ppm annual average) and potential alternative standards (see
23 chapter 5 of this document) to protect public health. In order to evaluate the ability of a specific
24 standard to protect public health, NO2 concentrations need to be adjusted such that they simulate
25 levels of NC>2 that just meet that standard. For example, all areas of the United States currently
26 have ambient NC>2 levels below the current annual standard. Therefore, to simulate just meeting
27 the current annual standard, NC>2 air quality levels must be rolled upward. Similarly, to simulate
28 a potential standard that is below current air quality levels, those current levels must be rolled
29 downward. This process of adjusting air quality to simulate just meeting a specific standard is
30 described in more detail below. For purposes of illustration, the adjustment to simulate just
August 2008 - Draft 47
-------
1 meeting the current standard is described. However, adjustments to simulate just meeting the
2 potential alternative standards have been accomplished using the same proportional approach.
3 6.2.1 Adjustment of Ambient Air Quality
4 Based on the form of the current standard and observed trends in ambient monitoring,
5 ambient NC>2 concentrations were proportionally rolled-up at each location using the maximum
6 annual average concentration that occurred in each year. While annual average concentrations
7 have declined significantly over the time period of analysis, the variability in the concentrations,
8 both the annual average and 1-hour concentrations, have remained relatively constant (see
9 section 7 in Appendix A for details). Therefore, proportional adjustment factors F for each
10 location (/') and year (/) were derived by the following:
11
12 Fi]=53/Cm^ equation (6-1)
13
14 where,
15
16 Fy• = Adjustment factor (unitless)
17 Cmax,ij = Maximum annual average NO2 concentration at a monitor in a location/'and
18 yeary (ppb)
19
20 In these cases where staff simulated a proportional roll-up in ambient NC>2 concentrations
21 using equation (6-1), it is assumed that the current temporal and spatial distribution of air
22 concentrations (as characterized by the current air quality data) is maintained and increased NOX
23 emissions contribute to increased NC>2 concentrations, with the highest monitor (in terms of
24 annual averages) being adjusted so that it just meets the current 0.053 ppm annual average
25 standard. Values for each air quality adjustment factor used for each location evaluated in the air
26 quality and risk characterization are given in Appendix A (section 7.2). For each location and
27 calendar year, all the hourly concentrations in a location were multiplied by the same constant
28 value F to make the highest annual mean equal to 53 ppb for that location and year. For
29 example, of several monitors measuring NC>2 in Boston for year 1995, the maximum annual
30 mean concentration was 30.5 ppb, giving an adjustment factor ofF = 53/30.5 = 1.74 for that
31 year. All hourly concentrations measured at all monitoring sites in that location would then be
32 multiplied by 1.74, resulting in an upward scaling of hourly NC>2 concentrations for that year.
33 Therefore, one monitoring site in Boston for year 1995 would have an annual average
August 2008 - Draft 48
-------
1 concentration of 0.053 ppm, while all other monitoring sites would have an annual average
2 concentration below that value, although still proportionally scaled up by 1.74. Then, using the
3 adjusted hourly concentrations to simulate just meeting the current standard, the metrics of
4 interest (e.g., annual mean NC>2 concentration, the number of potential health effect benchmark
5 exceedances) were estimated for each site-year.
6 Proportional adjustment factors were also derived considering the form, averaging time,
7 and levels of the potential alternative standards under consideration. Discussion regarding the
8 staff selection of each of these components is provided in chapter 5 of this document. The 98th
9 and 99th percentile 1-hour NC>2 concentrations averaged across three years of monitoring were
10 used in calculating the adjustment factors at each of four levels as follows:
11
12 F^ = S/C%1^ equation (6-2)
13
14 where,
15
16 Fy • = Adjustment factor (unitless)
17 S = Alternative standard level (50, 100, 150, 200 ppb 1-hour concentration)
18 C%ue,ij = Maximum 98th or 99th percentile 1-hour NC>2 concentration averaged across
19 three years at a monitor in location /' (ppb)
20
21 As described above for adjustments made in simulating just meeting the current standard,
22 it is assumed that the current temporal and spatial distribution of air concentrations (as
23 characterized by the current air quality data) is maintained and increased NOX emissions
24 contribute to increased NC>2 concentrations, with the highest monitor (in terms of the 3 -year
25 average at the 98th or 99th percentile) being adjusted so that it just meets the level of the
26 particular 1-hour alternative standard. Since the alternative standard levels range from 50 ppb
27 through 200 ppb, both proportional roll-up and roll-backs were used to adjust the 1-hour
28 concentrations. The values for each air quality adjustment factor used for each location
29 evaluated in the air quality and risk characterization are given in Appendix A, section 7.2. Only
30 the more recent air quality data were used and separated into two 3 -year periods, 2001-2003 and
3 1 2004-2006. The 1-hour concentrations were adjusted in a similar manner described above for
32 just meeting the current standard, however, due to the form of the standard, only one factor was
33 derived for each 3 -year period, rather than one factor for each calendar year as was done with
34 just meeting the current standard.
August 2008 - Draft 49
-------
1 6.2.2 Adjustment of Potential Health Effect Benchmark Levels
2 Rather than proportionally modify the air quality concentrations used for input to the
3 exposure model, a proportional roll-down of the potential health effect benchmark level was
4 performed. This was done to reduce the processing time associated with the exposure modeling
5 simulations since there were tens of thousands of receptors modeled in each location. In
6 addition, because the adjustment is proportional, the application of a roll-down of the selected
7 benchmark level is mathematically equivalent to a proportional roll-up of the air quality
8 concentrations. The same approach used in the air quality adjustment described above was used
9 in the exposure modeling to scale the benchmark levels downward to simulate just meeting the
10 current standard. For example, an adjustment factor of 2.27 was determined for Atlanta for year
11 2001 to simulate ambient concentrations just meeting the current standard, based on a maximum
12 predicted annual average NC>2 concentration of 23.3 ppb for a modeled receptor placed at an
13 ambient monitoring location. Therefore, the 1-hour potential health effect benchmark levels of
14 100, 200, and 300 ppb were proportionally rolled-down to 44, 88, and 132 ppb, respectively for
15 year 2001. This procedure was applied for each year within each location where an exposure
16 modeling was performed to simulate just meeting the current standard. Additional details
17 regarding derivation of the adjusted benchmark levels are given in chapter 8 of this document.
18
August 2008 - Draft 50
-------
i 7. AMBIENT AIR QUALITY ASSESSMENT AND HEALTH RISK
2 CHARACTERIZATION
4 7.1 OVERVIEW
5 Ambient monitoring data for each of the years 1995 through 2006 were used in this
6 analysis to characterize NC>2 air quality across the U.S. This air quality data, as well as other
7 NC>2 concentrations derived from ambient levels, were used as a surrogate to estimate potential
8 human exposure. While an individual ambient monitor measures NC>2 concentrations at a
9 stationary location, the monitor may well represent the concentrations that persons residing
10 nearby are exposed to. The extrapolation of ambient monitor concentration to personal exposure
11 will be dependent upon the spatial distribution of important emission sources, the siting of the
12 ambient monitors, and consideration of places that persons visit. It is within this context that the
13 approach for evaluating the ambient NC>2 air quality was designed.
14 Based on the health effects information from the human clinical and epidemiological
15 studies, the averaging time of interest for the air quality characterization was 1-hour, with
16 concentration levels ranging from between 100 and 300 ppb. Since the current standard is based
17 on annual average levels of NC>2 while the most definitive health effects evidence is associated
18 with short-term (i.e., 30-minute to 1-hour, or one to several day) exposures, the air quality
19 analysis required the development of a model that relates annual average and short-term levels of
20 NC>2. To characterize this relationship and to estimate the number of exceedances of the
21 potential health effect benchmarks in specific locations, several possible models were explored
22 (i.e., exponential regression, logistic regression, a regression assuming a Poisson distribution,
23 and an empirical model). An empirical model, employing the annual average and hourly
24 concentrations, was chosen to avoid some of the difficulties in extrapolating outside the range of
25 the data. In addition, an empirical model could be used for any averaging time of interest. A
26 detailed discussion justifying the selection of this approach is provided in Appendix A, section 6.
27 The available NC>2 air quality were first divided into two year-groups; one contained data
28 from years 1995-2000, representing an historical data set; the other contained the monitoring
August 2008 - Draft 51
-------
1 years 2001-2006, representing recent ambient monitoring. Each of these monitoring year-groups
2 were evaluated considering the NC>2 concentrations as they were reported and representing the
3 conditions at that time (termed in this assessment "as is"). This served as the first air quality
4 scenario, with the results within each year-group separated by monitor distance from a major
5 road (either <100 m or >100 m). The ambient monitor data were categorized in this manner to
6 account for the potential influence of vehicle emissions on concentrations measured at the
7 monitors within close proximity to roadways. There is potential for different concentration
8 levels measured at each of these locations (i.e., near-road versus away from road) and thus
9 potentially different exposure concentrations experienced by those persons spending time in
10 these locations. A second scenario used the as is ambient monitoring data obtained from
11 monitors sited >100 m from a major road and a simplified on-road simulation approach to
12 estimate on-road NC>2 concentrations for each of the year-groups. This scenario was developed
13 by recognizing that vehicles are important emission sources of NOX and NC>2 and that people
14 spend time inside vehicles on roads.
15 Two additional scenarios followed in similar fashion to the as is air quality analysis,
16 however these scenarios considered the ambient NC>2 concentrations simulated to just meeting
17 the current standard of 0.053 ppm annual average and each of the alternative 1-hour standards of
18 50, 100, 150, and 200 ppb.7 Due to the form of the alternative standards considered here (98th
19 and 99th percentiles average over 3 years), the recent ambient monitoring data set was divided
20 into two three-year periods, 2001-2003 and 2004-2006. Thus, the air quality characterization
21 results are separated into two broad analyses, one using air quality as is and the other where air
22 quality was adjusted to just meeting the current and alternative standards. Within both of these
23 analyses, an additional simulation was performed to estimate NC>2 concentrations on roads. The
24 first scenario described above is the only scenario that uses purely measurement data. Each of
25 the other scenarios either uses a simulation procedure to estimate on-road concentrations
26 (scenario 2), concentrations that just meet a particular standard level (scenario 3), or both
27 (scenario4).
28 Since all of the NO2 ambient monitoring sites are represented by this analysis, the results
29 are considered a broad characterization of national air quality and potential human exposures that
7 Originally, the historic data was evaluated using concentrations as is and for just meeting the current standard. The
potential alternative standards were not evaluated using the 1995-2000 air quality. Results for evaluating air quality
just meeting the current annual average standard using the historic data set are provided in Appendix A section 9.
August 2008 - Draft 52
-------
1 might be associated with these scenario-driven concentrations. The output of this air quality
2 characterization was used to estimate the number of times per year specific locations experience
3 levels of NC>2 that may cause adverse health effects in susceptible individuals. Each location that
4 was evaluated contained one to several monitors operating for a few to several years, generating
5 a number of site-years of data. The number of site-years in a location were used to generate a
6 distribution of two exposure and risk characterization metrics; the annual average concentrations
7 and the numbers of exceedances that did (observed data) or could occur (simulated data) in a
8 year for that location. The mean and median values were reported to represent the central
9 tendency of each metric for the four scenarios in each air quality year-group, while the minimum
10 value served to represent the lower bound. Since there were either multiple site-years or
11 numerous simulations performed at each location using all available site-years of data, results for
12 the upper percentiles included the 95th, 98th and 99th percentiles of the distribution.
13 7.2 APPROACH
14 There were three broad steps to allow for the characterization of the air quality. The first
15 step involved collecting, compiling, and screening the ambient air quality data collected since the
16 prior review in 1995. A screening of the data followed to ensure consistency with the NC>2
17 NAAQS requirements. Then, criteria based on the current standard and the potential health
18 effect benchmark levels were used to identify specific locations for analysis using descriptive
19 statistical analysis of the screened data set. All other monitoring data not identified by the
20 selected criteria were grouped into one of two non-specific categories. These locations (both the
21 specific and non-specific) served as the geographic centers of the analysis, where application of
22 the empirical model was done to estimate concentrations and exceedances of potential health
23 effect benchmark levels. In addition to the use of the ambient concentrations (as is) and ambient
24 concentrations just meeting the current and alternative standard levels, on-road concentrations
25 were estimated in this air quality characterization to approximate the potential exposure and risk
26 metrics associated with these concentrations.
27 7.2.1 Air Quality Data Screen
28 NC>2 air quality data and associated documentation from the years 1995 through 2006
29 were downloaded from EPA's Air Quality System (AQS) for this purpose (EPA, 2007c, d). A
30 site was defined by the state, county, site code, and parameter occurrence code (POC), which
August 2008 - Draft 53
-------
1 gives a 10-digit monitor ID code. As required by the NC>2 NAAQS, a valid year of monitoring
2 data is needed to calculate the annual average concentration. A valid year at a monitoring site
3 was comprised of 75% of valid days in a year, with at least 18 hourly measurements for a valid
4 day (thus at least 274 or 275 valid days depending on presence of a leap year and a minimum of
5 4,932 or 4,950 hours). This served as the screening criterion for data used in the analysis.
6 Site-years of data are the total numbers of years the collective monitors in a location were
7 in operation. Of a total of 5,243 site-years of data in the entire NC>2 1-hour concentration
8 database, 1,039 site-years did not meet the above criterion and were excluded from any further
9 analyses. In addition, since shorter term average concentrations are of interest, the remaining
10 site-years of data were further screened for 75% completeness on hourly measures in a year (i.e.,
11 containing a minimum of 6,570 or 6,588, depending on presence of a leap year). Twenty-seven
12 additional site-years were excluded, resulting in 4,177 complete site-years in the analytical
13 database. Table 7-1 provides a summary of the site-years included in the analysis, relative to
14 those excluded, by location and by two site-year groups.8 The air quality data from AQS were
15 separated into these two groups, one representing historic data (1995-2000) and the other
16 representing more recent data (2001-2006) to represent temporal variability in NC>2
17 concentrations within each location. The selection of locations was a companion analysis to the
18 screening, however, it is discussed in a separate section.
19
20 Table 7-1. Counts of complete site-years of NO2 monitoring data.
21
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington
Atlanta
Colorado Springs
El Paso
Com
1995-2000
58
47
11
26
12
193
24
93
46
69
24
26
14
Number of
plete
2001-2006
47
36
11
10
12
177
20
81
39
66
29
0
30
Site-Years
Incorr
1995-2000
16
20
2
10
4
16
1
12
6
21
5
4
11
i plete
2001-2006
34
22
2
4
1
19
4
24
8
18
1
4
0
Site-^
% Cor
1995-2000
78%
70%
85%
72%
75%
92%
96%
89%
88%
77%
83%
87%
56%
fears
iplete
2001-2006
58%
62%
85%
71%
92%
90%
83%
77%
83%
79%
97%
0%
100%
14 of 18 named locations and the 2 grouped locations contained enough data to be considered valid for year 2006.
August 2008 - Draft
54
-------
Location
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other CMSA
Not MSA
Total
Com
1995-2000
6
16
22
6
56
1135
200
Number of
plete
2001-2006
4
35
27
6
43
1177
243
4177
Site-Years
Incorr
1995-2000
0
4
8
0
3
249
112
i plete
2001-2006
2
9
25
0
9
235
141
1066
Site-^
% Cor
1995-2000
100%
80%
73%
100%
95%
82%
64%
fears
iplete
2001-2006
67%
80%
52%
100%
83%
83%
63%
80%
2 7.2.2 Selection of Locations for Air Quality Analysis
3 Criteria were established for selecting sites with high annual means and/or frequent
4 exceedances of potential health effect benchmarks. Selected locations were those that had a
5 maximum annual mean NC>2 level at a particular monitor greater than or equal to 25.7 ppb, which
6 represents the 90th percentile across all locations and site-years, and/or had at least one reported
7 1-hour NO2 level greater than or equal to 200 ppb, the lowest level of the potential health effect
8 benchmarks. A location in this context would include a geographic area that encompasses more
9 than a single air quality monitor (e.g., particular city, metropolitan statistical area (MSA), or
10 consolidated metropolitan statistical area or CMSA). First, all monitors were identified as either
11 belonging to a CMSA, a MSA, or neither. Then, locations of interest were identified through
12 statistical analysis of the ambient NO2 air quality data for each site within a location.
13 Fourteen locations met both selection criteria and an additional four met at least one of
14 the criteria (see Table 7-2).9 In addition to these 18 specific locations, the remaining sites were
15 grouped into two broad location groupings. The Other CMSA location contains all the other sites
16 that are in MS As or CMS As but are not in any of the 18 specified locations. The Not MSA
17 location contains all the sites that are not in an MSA or CMSA. The final database for analysis
18 included air quality data from a total of 205 monitors within the named locations, 331 monitors
19 in the Other CMSA group, and 92 monitors in the Not MSA group.
20
21
9 New Haven, CT, while meeting both criteria, did not have any recent exceedances of 200 ppb and contained one of
the lowest maximum concentration-to-mean ratios, therefore was not separated out as a specific location for
analysis.
August 2008 - Draft
55
-------
1 Table 7-2. Locations selected for Tier I NO2 Air Quality Characterization, associated
2 abbreviations, and values of selection criteria.
Location
Type1 Code Description Abbreviation
CMSA*
CMSA
CMSA*
CMSA*
CMSA*
CMSA*
CMSA
CMSA*
CMSA*
CMSA*
MSA*
MSA*
MSA*
MSA
MSA*
MSA*
MSA
MSA*
MSA/CMSA
-
1122
1602
1692
2082
2162
4472
4992
5602
6162
8872
0520
1720
2320
3600
4120
6200
6520
7040
-
-
Boston-Worcester-Lawrence,
MA-NH-ME-CT
Chicago-Gary-Kenosha, IL-IN-
Wl
Cleveland-Akron, OH
Denver-Boulder-Greeley, CO
Detroit-Ann Arbor-Flint, Ml
Los Angeles-Riverside-Orange
County, CA
Miami-Fort Lauderdale, FL
New York-Northern New
Jersey-Long Island, NY-NJ-CT-
PA
Philadelphia-Wilmington-
Atlantic City, PA-NJ-DE-MD
Washington-Baltimore, DC-MD-
VA-WV
Atlanta, GA
Colorado Springs, CO
El Paso,TX
Jacksonville, FL
Las Vegas,NV-AZ
Phoenix-Mesa,AZ
Provo-Orem,UT
St, Louis, MO-IL
Other MSA/CMSA
Other Not MSA
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington DC
Atlanta
Colorado Springs
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other CMSA
Not MSA
Maximum # of
Exceedances
of 200 ppb
1
0
1
2
12
5
3
3
3
2
1
69
2
2
11
37
0
8
10
2
Maximum
Annual
Mean
(ppb)
31.1
33.6
28.1
36.8
25.9
50.6
16.8
42.2
34.0
27.2
26.6
34.8
35.1
15.9
27.1
40.5
28.9
27.2
31.9
19.7
1 CMSA is consolidated metropolitan statistical area; MSA is metropolitan statistical area according to the
1 999 Office of Management and Budget definitions (January 28, 2002 revision).
* Indicates locations that satisfied both the annual average and exceedance criteria.
5 7.2.3 Estimation of On-Road Concentrations using Ambient Concentrations
6 Since mobile sources can account for a large part of personal exposures to ambient NC>2
7 in some individuals, the potential impact of roadway levels of NC>2 was evaluated. A strong
8 relationship has been reported between NC>2 levels measured on roadways and NC>2 measured at
9 increasing distance from the road. This relationship has been described previously (e.g., Cape et
10 al., 2004) using an exponential decay equation of the form:
August 2008 - Draft
56
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Cx=Cb + Cve^
where,
equation (7-1)
Cx = NO2 concentration at a given distance (x) from a roadway (ppb)
Cb = NO2 concentration (ppb) at a distance from a roadway
by road or non-road source emissions.
, not directly influenced
Cv = NO2 concentration contribution from vehicles on a roadway (ppb)
k = Rate constant describing NO2 combined formation/decay with perpendicular
distance from roadway (meters'1)
x = Distance from roadway (meters)
Based on the findings of several researchers, much of the decline
with distance from the road has been shown to occur within the first few
90% within 10 meter distance), returning to near ambient levels between
(Rodes and Holland, 1981; Bell and Ashenden, 1997; Gilbert et al., 2003
in NO2 concentrations
meters (approximately
200 to 500 meters
;Pleijeletal., 2004).
At a distance of 0 meters, referred to here as on-road, the equation reduces to the sum of the non-
source influenced NO2 concentration and the concentration contribution expected from vehicle
emissions on the roadway using
Cr=Ca(l+m)
where,
Cr = 1-hour on-road NO2 concentration (ppb)
equation (7-2)
Ca = 1-hour ambient monitoring NO2 concentration (ppb) either as is or modified
to just meet the current standard
m = Modification factor derived from estimates of Cv/Cb (from equation (7-1))
and assuming that Ca = Cb.10
10 Note that Ca differs from Cb since Ca may include the influence of on-road as well as non-road sources. However,
it is expected that for most monitors the influence of on-road emissions is minimal so that Ca = Cfr
August 2008 - Draft 57
-------
1 To estimate on-road NO2 levels as a function of the level recorded at ambient monitors
2 and the distance of those monitors from a roadway, empirical data from published scientific
3 literature were used. A literature review was conducted to identify published studies containing
4 NO2 concentrations on roadways and at varying distances from roadways. Relevant data
5 identified from this literature review were used to estimate m (equation 7-1) generating a
6 distribution of values for use in estimating on-road concentrations. See Appendix A, section 8
7 for a detailed explanation of derivation of the on-road modification factors and the literature
8 sources used.
9 Theoretically, NO2 concentrations can increase at a distance from the road due to
10 chemical interaction of NOX with 63, the magnitude of which can be driven by certain
11 meteorological conditions (e.g., wind direction). As such, the maximum NO2 concentration may
12 not occur on the road but at a distance from the road. However, there are two important
13 components of this estimation procedure that need consideration. First, the relationship
14 developed from peer-reviewed NO2 roadway and near-road measurement studies was used to
15 estimate NO2 concentrations that occur on the road and not used to estimate NO2 concentrations
16 at a distance from the road. If this does occur in a location, the ambient monitors located within
17 100 m of a road would capture this potential effect, where such monitors are available. Second,
18 since there is potential for monitors that are sited near roadways to be influenced by vehicle
19 emissions and equation (7-2) assumes the ambient concentration is approximating NO2
20 concentrations not directly influenced by the roadway, the monitors within 100 m were not used
21 for calculating the on-road concentrations. The uncertainty regarding these issues and potential
22 effect on exposure estimates are discussed in section 7.4.
23 To estimate NO2 levels on roadways, each monitoring site was randomly assigned one
24 on-road factor (m) for summer months and one for non-summer months from the derived
25 empirical distribution. On-road factors were assigned randomly because we expect the empirical
26 relationship between Cv and Cb to vary from place to place and we do not have sufficient
27 information to match specific ratios with specific locations. Hourly NO2 levels were estimated
28 for each site-year of data in a location using equation (7-2) and the randomly assigned on-road
29 modification factors. The process was simulated 100 times for each site-year of hourly data. For
30 example, the Boston CMSA location had 210 random selections from the on-road distributions
31 applied independently to the total site-years of data (105). Following 100 simulations, a total of
August 2008 - Draft 58
-------
1 10,500 site-years of data were generated using this procedure (along with 21,000 randomly
2 assigned on-road values selected from the appropriate empirical distribution).
3 Simulated on-road NC>2 concentrations were then used to generate concentration
4 distributions for the annual average concentrations and distributions for the number of
5 exceedances of short-term potential health effect benchmark levels. Mean and median values are
6 reported to represent the central tendency of each parameter estimate. Since there were multiple
7 site-years and numerous simulations performed at each location using all valid site-years of data,
8 results for the upper percentiles were expanded to the 95th, 98th and 99th percentiles of the
9 distribution. In using the Boston CMSA data as an example for years 1995-2000, 5800 site years
10 of on-road concentration hourly data were simulated, and both the annual average concentration
11 and numbers of exceedances of potential health effect benchmark levels were calculated. The
12 95th, 98th and 99th percentiles were the 5510th, the 5684th, and the 5742nd highest values,
13 respectively, of the 5800 calculated and ranked values. Roadways with high vehicle densities are
14 likely better represented by on-road concentration estimates at the upper tails of the distribution.
15 7.3 AIR QUALITY AND HEALTH RISK CHARACTERIZATION
16 RESULTS
17 7.3.1 Ambient Air Quality (As Is)
18 As described earlier, this first scenario analyzing the as is air quality is based purely on
19 the measurement data. The air quality data obtained from AQS were separated into two year-
20 groups, one representing historic data (1995-2000) and the other representing more recent data
21 (2001-2006). Detailed descriptive statistics regarding concentration distributions for particular
22 locations, monitoring sites, and specific monitoring years are provided in the Appendix A,
23 section 5. A summary of the descriptive statistics for the annual average ambient NO2
24 concentrations at each selected location is provided in Tables 7-3 and 7-4 for monitors sited
25 >100 m and < 100 m from a major road, respectively. None of the locations contained a
26 measured exceedance of the current standard of 0.053 ppm at any monitor. The highest observed
27 annual average concentrations were measured in Los Angeles and Phoenix during the historic
28 monitoring period and considering the monitors >100 m from a major road. There were a fewer
29 number of locations with monitors sited < 100 m of a major road, however in most of the
30 locations where comparative monitoring data were available, the annual average concentrations
August 2008 - Draft 59
-------
4
5
6
7
8
9
10
11
12
13
14
15
were greater at the monitors within 100 m of a major road (in 23 of 27 possible location/year-
group combinations). Four locations (Denver, Los Angeles, Phoenix, St. Louis) contained
higher concentrations at the more distant monitors for one year-group when compared with the
monitors within 100 m. Where concentrations were greater at the near road monitors, the
concentrations were on average about 20-25% higher when compared with the more distant
monitors in each corresponding location, regardless of year-group. A comparison of the year-
group of data within each monitor site-group indicates that the more recent monitoring
concentrations were lower, on average by about 13-15%. These average trends in concentration
across year-group and monitor site group were generally observed across all percentiles of the
distribution.
Table 7-3. Monitoring site-years and annual average NO2 concentrations for two monitoring
periods, historic and recent air quality data (as is) using monitors sited >100 m of a
major road.
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington DC
Atlanta
Colorado Springs
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other CMSA
Not MSA
1995-2000
Site-
Years
18
28
5
7
12
92
9
47
35
33
24
25
8
6
8
14
6
18
1135
200
Annual Mean (ppb) 1
mean min med p95 p98 p99
18
20
19
22
19
27
9
24
21
18
14
16
19
15
10
30
24
17
14
8
5
9
17
15
12
6
9
11
15
9
5
7
14
14
3
26
23
5
1
0
18
22
20
23
19
28
9
26
20
19
15
17
18
15
6
29
24
19
14
7
25
27
21
26
26
40
10
35
33
25
25
24
23
16
24
34
24
21
24
16
25
28
21
26
26
46
10
36
33
26
27
35
23
16
24
34
24
21
26
19
25
28
21
26
26
46
10
36
33
26
27
35
23
16
24
34
24
21
28
19
2001-2006
Site-
Years
14
17
3
5
12
105
10
48
26
35
29
-
24
4
27
14
6
13
1177
243
Annual Mean (ppb) 1
mean min med p95 p98 p99
9
21
18
21
19
20
8
20
19
17
12
-
15
14
10
25
24
16
12
7
5
16
17
18
14
5
7
10
14
7
3
-
8
13
1
21
21
12
1
1
9
19
17
21
19
20
8
19
18
18
14
-
16
14
7
24
23
16
12
6
12
28
19
26
23
33
10
28
28
24
19
-
18
15
22
29
29
21
20
14
12
28
19
26
23
34
10
31
28
25
23
-
18
15
22
29
29
21
22
16
12
28
19
26
23
36
10
31
28
25
23
-
18
15
22
29
29
21
24
16
1 The mean is the sum of the annual means for each monitor in a particular location divided by the number of
site-years across the monitoring period. The min, med, p95, p98, p99 represent the minimum, median, 95th, 98th,
and 99th percentiles of the distribution for the annual mean.
2 Colorado Springs monitoring data were collected as part of short-term study completed in September 2001 ,
therefore there are no 2001-2006 data.
August 2008 - Draft
60
-------
1
2
3
4
Table 7-4. Monitoring site-years and annual average NO2 concentrations for two monitoring
periods, historic and recent air quality data (as is) using monitors sited <100 m of a
major road.
Location
Boston
Chicago
Cleveland
Denver
Los Angeles
Miami
New York
Philadelphia
Washington DC
Colorado Springs
El Paso
Las Vegas
Phoenix
St, Louis
1995-2000
Site-
Years
40
19
6
19
101
15
46
11
36
1
6
8
8
38
Annual Mean (ppb) 1
mean min med p95 p98 p99
18
29
26
14
25
11
31
30
23
18
29
19
31
18
6
22
23
6
4
6
22
26
13
18
23
7
24
9
20
31
27
9
23
9
29
29
23
18
29
25
30
19
31
34
28
35
45
17
42
34
27
18
35
27
40
26
31
34
28
35
46
17
42
34
27
18
35
27
40
27
31
34
28
35
46
17
42
34
27
18
35
27
40
27
2001-2006
Site-
Years
33
19
8
5
72
10
33
13
31
6
8
13
30
Annual Mean (ppb) 1
mean min med p95 p98 p99
18
27
20
31
25
10
29
23
20
18
15
25
15
7
18
14
27
4
6
18
18
13
13
3
11
8
18
28
22
29
27
10
28
24
20
19
19
24
15
25
32
24
37
37
16
40
30
26
22
23
37
23
30
32
24
37
40
16
40
30
26
22
23
37
25
30
32
24
37
41
16
40
30
26
22
23
37
25
1 The mean is the sum of the annual means for each monitor in a particular location divided by the number of
site-years across the monitoring period. The min, med, p95, p98, p99 represent the minimum, median, 95th, 98th,
and 99th percentiles of the distribution for the annual mean.
2 Colorado Springs monitoring data were collected as part of short-term study completed in September 2001 ,
therefore there are no 2001-2006 data.
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
The estimated number of exceedances of four potential health effect benchmark levels
(150, 200, 250, and 300 ppb NO2 for 1-hr) is shown in Tables 7-5 and 7-6 for the historic and
recent ambient monitoring data, respectively, and where the monitors were sited > 100 m from a
major road. The number of exceedances of each benchmark were summed for the year at each
monitor; a single monitor value of 10 could represent ten 1-hr exceedances that occurred in one
day, 10 exceedances in 10 days, or some combination of multiple hours or days that totaled 10
exceedances for the year. In general, the number of benchmark exceedances was low across all
locations and considering both year-groups of the as is air quality. The average number of
exceedances of the lowest 1-hour concentration level of 150 ppb across each location was
typically none or one. Considering that there are 8,760 hours in a year, this many exceedances
amounts to a small fraction of the year (0.01%) containing an exceedance of the potential health
effect benchmark level. For locations with greater than 1 yearly average exceedance, the
numbers were primarily driven by a single site-year of data. For example, the Colorado Springs
mean is 3 exceedances per year for the years 1995-2000; however, this mean was driven by a
August 2008 - Draft
61
-------
1 single site-year that contained 69 exceedances of 200 ppb. That particular monitor (ID
2 0804160181) does not appear to have any unusual attributes (e.g., the closest major road is
3 beyond a distance of 160 meters and the closest stationary source emitting > 5 tons per year (tpy)
4 is at a distance > 4 km) except that a power generating utility (NAICS code 221112) located 7.2
5 km from the monitor has estimated emissions of 4,205 tpy. It is not known at this time whether
6 this particular facility is influencing the observed concentration exceedances at this specific
7 monitoring site. Similarly, Detroit contained the largest number of excedances of 200 ppb (a
8 maximum of 12) for as is air quality data from years 2001-2006 (Table 7-6). Again, all of those
9 exceedances occurred at one monitor (ID 2616300192) during one year (2002). The number of
10 exceedances of higher potential benchmark concentration levels at each of the locations was less
11 than that observed at the 200 ppb level. Most locations had no exceedances of 250 or 300 ppb,
12 with higher numbers confined to the same aforementioned cities where exceedances of 200 ppb
13 were observed.
14 When considering the historic data and monitors sited within 100 m of a major road
15 (Table 7-7), a few locations contained exceedances of the potential health effect benchmark
16 levels, driven mainly by observations from one or two monitors. For example, in Phoenix a
17 single year from one monitor (ID 0401330031) was responsible for all observed exceedances of
18 200 ppb. This monitor is located 78 m from a major road along with 10 stationary sources
19 located within 10 km of this monitor, 9 of which contained estimated emissions of less than 60
20 tpy (one source emitted 272 tpy, see Appendix A, section 4). It is not known if observed
21 exceedances of 200 ppb at this monitor are a result of proximity of major roads or the stationary
22 sources. There were fewer locations with observed exceedances of the benchmark levels at the
23 monitors sited within 100 m of a major road considering the more recent as is air quality. Eleven
24 of thirteen total locations contained an average of zero exceedances of the 150 ppb benchmark
25 level (Table 7-8).
August 2008 - Draft 62
-------
2 Table 7-5 Number of exceedances of short-term (1-hour) potential health effect benchmark levels in a year, 1995-2000 historic NO2 air
3 quality (as is) using monitors sited >100 m of a major road.
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington
DC
Atlanta
Colorado
Springs
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other
CMSA
Not MSA
Exceedances of 150 ppb 1
mean
0
0
0
1
1
3
0
0
0
0
0
8
0
0
0
0
0
1
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
1
0
0
4
10
22
0
0
0
1
1
47
0
0
0
2
0
12
0
0
P98
1
0
0
4
10
42
0
3
10
2
1
143
0
0
0
2
0
12
0
2
p99
1
0
0
4
10
44
0
3
10
2
1
143
0
0
0
2
0
12
1
4
Exceedances of 200 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
3
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
2
3
2
0
0
0
1
0
3
0
0
0
0
0
8
0
0
p98
0
0
0
2
3
2
0
0
3
2
1
69
0
0
0
0
0
8
0
0
p99
0
0
0
2
3
4
0
0
3
2
1
69
0
0
0
0
0
8
0
1
Exceedances of 250 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
0
1
0
0
0
0
1
0
0
0
0
0
0
0
4
0
0
p98
0
0
0
0
1
1
0
0
0
1
0
23
0
0
0
0
0
4
0
0
p99
0
0
0
0
1
2
0
0
0
1
0
23
0
0
0
0
0
4
0
0
Exceedances of 300 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p98
0
0
0
0
1
0
0
0
0
0
0
4
0
0
0
0
0
0
0
0
p99
0
0
0
0
1
1
0
0
0
0
0
4
0
0
0
0
0
0
0
0
Notes:
1 The mean number of exceedances represents the number of exceedances occurring at all monitors in a particular location divided by the number of site-
years across the monitoring period. The min, med, p95, p98, and p99 represent the minimum, median, 95 h, 98th, and 99th percentiles of the distribution for
he number of exceedances in any one year within the monitoring period.
August 2008 - Draft
63
-------
1
2
3
4
Table 7-6 Number of exceedances of short-term (1-hour) potential health effect benchmark levels in a year, 2001-2006 recent NO2 air
quality (as is) using monitors sited >100 m of a major road.
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington
DC
Atlanta
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other
CMSA
Not MSA
Exceedances of 150 ppb 1
mean
0
0
0
0
2
0
0
0
0
0
0
0
2
0
0
7
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
p95
0
0
0
0
16
0
0
0
0
0
0
0
6
0
0
39
0
0
0
P98
0
0
0
0
16
1
0
0
1
0
1
1
6
0
0
39
0
0
1
p99
0
0
0
0
16
1
0
0
1
0
1
1
6
0
0
39
0
0
2
Exceedances of 200 ppb 1
mean
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
p95
0
0
0
0
12
0
0
0
0
0
0
0
2
0
0
0
0
0
0
p98
0
0
0
0
12
0
0
0
1
0
0
0
2
0
0
0
0
0
0
p99
0
0
0
0
12
0
0
0
1
0
0
0
2
0
0
0
0
0
1
Exceedances of 250 ppb 1
mean
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
0
8
0
0
0
0
0
0
0
1
0
0
0
0
0
0
p98
0
0
0
0
8
0
0
0
1
0
0
0
1
0
0
0
0
0
0
p99
0
0
0
0
8
0
0
0
1
0
0
0
1
0
0
0
0
0
1
Exceedances of 300 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
0
5
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p98
0
0
0
0
5
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p99
0
0
0
0
5
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Notes:
1 The mean number of exceedances represents the number of exceedances occurring at all monitors in a particular location divided by the number of site-
years across the monitoring period. The min, med, p95, p98, and p99 represent the minimum, median, 95 , 98th, and 99th percentiles of the distribution for
:he number of exceedances in any one year within the monitoring period.
2 Colorado Springs monitoring data were collected as part of short-term study completed in September 2001 , therefore there are no 2001 -2006 data.
5
6
7
August 2008 - Draft
64
-------
1
2
3
4
Table 7-7. Number of exceedances of short-term (1-hour) potential health effect benchmark levels in a year, 1995-2000 historic NO2 air
quality (as is) using monitors sited <100 m of a major road.
Location
Boston
Chicago
Cleveland
Denver
Los Angeles
Miami
New York
Philadelphia
Washington
DC
Colorado
Springs
El Paso
Las Vegas
Phoenix
St, Louis
Exceedances of 150 ppb 1
mean
0
0
2
0
2
0
0
0
0
0
2
1
27
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
1
0
1
0
p95
0
0
9
6
11
3
2
1
0
0
7
11
147
0
P98
1
0
9
6
18
3
3
1
1
0
7
11
147
0
p99
1
0
9
6
33
3
3
1
1
0
7
11
147
0
Exceedances of 200 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
1
5
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
1
1
1
1
0
0
0
0
2
11
37
0
p98
1
0
1
1
2
1
3
0
0
0
2
11
37
0
p99
1
0
1
1
2
1
3
0
0
0
2
11
37
0
Exceedances of 250 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
1
1
0
0
0
0
0
0
0
3
3
0
p98
0
0
1
1
0
0
0
0
0
0
0
3
3
0
p99
0
0
1
1
0
0
0
0
0
0
0
3
3
0
Exceedances of 300 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
0
0
0
0
0
0
0
0
3
0
0
p98
0
0
0
0
0
0
0
0
0
0
0
3
0
0
p99
0
0
0
0
0
0
0
0
0
0
0
3
0
0
Notes:
1 The mean number of exceedances represents the number of exceedances occurring at all monitors in a particular location divided by the number of site-
years across the monitoring period. The min, med, p95, p98, and p99 represent the minimum, median, 95 , 98th, and 99th percentiles of the distribution for
he number of exceedances in any one year within the monitoring period.
August 2008 - Draft
65
-------
2 Table 7-8. Number of exceedances of short-term (1-hour) potential health effect benchmark levels in a year, 2001-2006 recent NO2 air
3 quality (as is) using monitors sited <100 m of a major road.
Location
Boston
Chicago
Cleveland
Denver
Los Angeles
Miami
New York
Philadelphia
Washington
DC
El Paso
Las Vegas
Phoenix
St, Louis
Exceedances of 150 ppb 1
mean
0
0
0
1
0
1
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
1
0
0
0
0
0
0
0
0
0
p95
0
0
1
1
2
5
1
0
0
0
0
0
0
P98
0
0
1
1
2
5
1
0
0
0
0
0
0
p99
0
0
1
1
6
5
1
0
0
0
0
0
0
Exceedances of 200 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
0
0
3
0
0
0
0
0
0
0
p98
0
0
0
0
1
3
0
0
0
0
0
0
0
p99
0
0
0
0
1
3
0
0
0
0
0
0
0
Exceedances of 250 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
0
0
3
0
0
0
0
0
0
0
p98
0
0
0
0
1
3
0
0
0
0
0
0
0
p99
0
0
0
0
1
3
0
0
0
0
0
0
0
Exceedances of 300 ppb 1
mean
0
0
0
0
0
0
0
0
0
0
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
0
0
0
3
0
0
0
0
0
0
0
p98
0
0
0
0
0
3
0
0
0
0
0
0
0
p99
0
0
0
0
0
3
0
0
0
0
0
0
0
Notes:
1 The mean number of exceedances represents the number of exceedances occurring at all monitors in a particular location divided by the number of site-
wears across the monitoring period. The min, med, p95, p98, and p99 represent the minimum, median, 95h, 98th, and 99th percentiles of the distribution for
[the number of exceedances in any one year within the monitoring period.
August 2008 - Draft
66
-------
1 7.3.2 On-Road Concentrations Derived From Ambient Air Quality (As Is)
2 Descriptive statistics for estimated on-road NO2 concentrations are presented in Table 7-
3 9. These estimated on-road concentrations were generated using the simulation procedure
4 described above (section 7.2.3) and represent the second scenario. For the 18 named locations,
5 the calculation only used monitors sited at a distance > 100 m of a major road. The two grouped
6 locations (i.e., "Other CMSA" and "Not MSA") did not have estimated monitor distances to
7 major roads therefore all monitoring data available were used to estimate the distribution of on-
8 road NO2 concentrations.
9 The simulated on-road annual average NO2 concentrations are, on average, a factor of 1.8
10 higher than their respective ambient levels. This falls within the range of ratios reported in the
11 ISA (about 2-fold higher concentrations on roads) (ISA, section 2.5.4). Los Angeles, New York,
12 and Phoenix were predicted to have the highest on-road NO2 levels. This is a direct result of
13 these locations already containing some of the highest "as-is" NO2 concentrations prior to the
14 on-road simulation (see Table 7-3).
15 The median of the simulated concentration estimates for Los Angeles were compared
16 with NO2 measurements provided by Westerdahl et al. (2005) for arterial roads and freeways in
17 the same general location during spring 2003. Although the averaging time is not exactly the
18 same, comparison of the medians is judged to be appropriate.11 The estimated median on-road
19 concentration for 2001-2006 is 36 ppb which falls within the range of 31 ppb to 55 ppb identified
20 by Westerdahl et al. (2005).
21 On average, most locations are predicted to have fewer than 10 exceedances per year for
22 the 200 ppb potential health effect benchmark while the median frequency of exceedances in
23 most locations is estimated to be 1 or less per year (Tables 7-10 and 7-11). When considering
24 the lower 1-hour benchmark of 150 ppb, most locations (17 out of 20) were estimated to have
25 less than 50/year, on average. There are generally fewer predicted mean exceedances of the
26 potential health effect benchmark levels when considering recent air quality compared with the
27 historic air quality. Areas with a relatively high number of estimated exceedances (e.g., Provo)
28 are likely influenced by the presence of a small number of monitors and one or a few exceptional
11 Table 10 considers annual average of hourly measurements while Westerdahl et al. (2005) reported between 2 to 4
hour average concentrations. Over time, the mean of 2-4 hour averages will be similar to the mean of hourly
concentrations, with the main difference being in the variability (and hence the various percentiles of the
distribution outside the central tendency).
August 2008 - Draft 67
-------
1 site-years where there were unusually high concentrations at the upper percentiles of the
2 concentration distribution.
3 The upper percentiles for estimated number of exceedances of the 150 ppb, 1-hr average
4 level in most locations using the historic ambient monitoring data was between 100 and 300 per
5 year, while a few locations were estimated to contain up to a several hundred exceedances (e.g.,
6 Los Angeles, New York, and Phoenix). There were lower numbers of estimated exceedances
7 considering the 2001-2006 air quality compared with the historic data, with most locations
8 containing under 200 estimated exceedances of 150 ppb per year at the 98th and 99th percentiles.
9 As expected, the frequency of benchmark exceedances at all locations was lower when
10 considering any of the higher benchmark levels (i.e., 200, 250, 300 ppb, 1-hr average) compared
11 with 150 ppb.
12 The number of predicted benchmark exceedances across large urban areas may be used to
13 broadly represent particular locations within those types of areas. For example, Chicago, New
14 York, and Los Angeles are large CMS As, have several monitoring sites, and have a large number
15 of roadways. Each of these locations was estimated to have, on average, about 10 exceedances
16 of 200 ppb per year on-roads. Assuming that the on-road exceedances distribution generated
17 from the existing monitoring is proportionally representing the distribution of roadways within
18 each location, about one-half of the roads in these areas would not have any on-road
19 concentrations in excess of 200 ppb. This is because the median value for exceedances of 200
20 ppb in most locations was estimated as zero. However, Tables 7-10 and 7-11 indicate that there
21 is also a possibility of tens to just over a hundred exceedances of 200 ppb in a year as an upper
22 bound estimate on certain roads/sites in a particular year.
23
August 2008 - Draft 68
-------
1
2
Table 7-9. Estimated annual average on-road NO2 concentrations for two monitoring periods,
historic and recent air quality data (as is).
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington
Atlanta
Colorado
Springs2
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other CMSA
Not MSA
1995-2000
Site-
Years
1800
2800
500
700
1200
9200
900
4700
3500
3300
2400
2500
800
600
800
1400
600
1800
113500
20000
Annual Mean (ppb) 1
mean min med p95 p98 p99
32
37
35
39
35
50
17
43
39
33
26
30
34
28
17
54
43
31
26
14
7
11
22
19
15
8
11
14
19
12
6
9
17
18
4
33
29
7
1
0
32
39
34
38
34
49
17
42
36
33
25
30
33
27
11
52
42
33
25
12
51
59
47
55
52
83
23
73
63
53
49
52
49
37
45
75
58
47
47
31
55
63
49
58
57
91
25
78
73
58
57
64
54
39
50
78
62
50
53
35
57
66
53
62
59
97
26
83
77
61
60
73
57
41
55
80
64
52
57
39
2001-2006
Site-
Years
1400
1700
300
500
1200
10500
1000
4800
2600
3500
2900
2400
400
2700
1400
600
1300
117700
24300
Annual Mean (ppb) 1
mean min med p95 p98 p99
16
37
32
39
34
37
15
35
34
31
21
26
25
18
45
43
30
21
12
7
20
22
23
18
6
9
12
18
9
4
10
17
2
26
26
16
1
1
16
35
32
38
34
36
14
34
32
31
23
26
25
13
43
41
29
21
11
25
57
42
54
47
63
21
56
52
51
40
39
34
43
63
61
41
39
27
28
64
43
61
52
72
24
62
60
56
43
43
36
46
70
69
46
45
31
29
66
45
62
54
77
24
66
64
59
47
43
37
50
72
70
49
48
33
1 The mean is the sum of the annual means for each monitor in a particular location divided by the number of site-
years across the monitoring period. The min, med, p95, p98, p99 represent the minimum, median, 95th, 98th, and
99th percentiles of the distribution for the annual mean.
2 Colorado Springs monitoring data were collected as part of short-term study completed in September 2001 ,
therefore there are no 2001-2006 data.
4
5
August 2008 - Draft
69
-------
2 Table 7-10. Estimated number of exceedances of short-term (1-hour) potential health effect benchmark levels in a year on-roads, 1995-
3 2000 historic NO2 air quality (as is).
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington
DC
Atlanta
Colorado
Springs
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St, Louis
Other
MSA/CMSA
Other Not
MSA
Exceedances of 150 ppb 1
mean
11
39
15
48
39
166
3
63
25
21
24
45
21
3
14
104
21
14
10
2
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
1
2
1
17
19
54
0
8
2
1
1
0
8
0
0
31
0
0
0
0
p95
79
212
108
185
158
738
13
397
124
128
160
267
96
13
95
447
112
74
55
11
P98
106
338
130
230
207
1023
27
560
311
208
271
447
141
30
294
630
195
121
109
31
p99
125
385
146
288
270
1268
27
685
369
240
357
626
149
36
306
670
245
132
168
55
Exceedances of 200 ppb 1
mean
1
7
2
8
10
43
0
13
4
3
4
21
4
0
2
14
2
2
1
1
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
4
2
6
0
0
0
0
0
0
0
0
0
2
0
0
0
0
p95
9
41
19
36
48
213
2
92
20
20
31
171
20
1
5
65
9
15
6
2
p98
20
97
27
46
72
348
4
155
45
39
57
264
31
2
34
89
33
25
18
7
p99
24
118
31
53
86
508
5
212
63
56
87
325
39
4
36
102
34
28
32
14
Exceedances of 250 ppb 1
mean
0
1
0
2
4
12
0
3
1
1
1
12
1
0
0
2
0
1
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
1
6
1
10
21
63
0
21
4
2
3
111
5
0
0
13
0
10
1
1
p98
4
23
5
12
34
118
0
44
11
8
11
183
7
1
6
21
1
13
3
2
p99
7
30
5
15
35
188
1
55
15
11
21
219
8
1
6
27
4
14
6
4
Exceedances of 300 ppb 1
mean
0
0
0
1
2
4
0
1
0
0
0
7
0
0
0
1
0
1
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
1
4
14
17
0
4
0
1
1
55
0
0
0
3
0
7
0
0
p98
1
3
1
6
21
39
0
10
5
2
1
121
2
0
0
6
0
11
1
1
p99
1
7
1
7
26
68
0
14
7
3
2
160
2
0
0
11
0
13
2
2
Notes:
1 The mean number of exceedances represents the number of exceedances occurring at all monitors in a particular location divided by the number of site-
years across the monitoring period. The min, med, p95, p98, and p99 represent the minimum, median, 95 , 98th, and 99th percentiles of the distribution for
he number of exceedances in any one year within the monitoring period.
August 2008 - Draft
70
-------
2 Table 7-11. Estimated number of exceedances of short-term (1-hour) potential health effect benchmark levels in a year on-roads, 2001-
3 2006 recent NO2 air quality (as is).
Location2
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington
DC
Atlanta
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St, Louis
Other
MSA/CMSA
Other Not
MSA
Exceedances of 150 ppb 1
mean
0
24
14
41
20
42
1
21
12
11
8
6
7
9
37
117
7
4
1
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
1
3
6
3
4
0
1
1
0
0
0
2
0
2
1
0
0
0
p95
1
160
79
171
116
227
4
129
62
81
52
34
29
39
184
658
48
17
4
P98
2
211
89
270
149
405
9
210
110
130
101
45
53
169
302
702
84
44
14
p99
10
337
89
379
171
546
16
280
211
141
121
54
53
205
350
703
102
76
27
Exceedances of 200 ppb 1
mean
0
4
2
4
5
7
0
3
1
1
1
1
3
1
3
70
1
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
p95
0
17
16
25
29
37
0
22
5
7
8
4
15
3
14
547
3
1
1
p98
0
44
23
40
44
87
1
45
12
14
16
8
23
14
28
662
10
5
4
p99
1
69
23
53
45
129
2
72
30
21
25
9
24
15
44
662
14
10
8
Exceedances of 250 ppb 1
mean
0
0
0
0
2
1
0
1
0
0
0
0
2
0
0
33
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
1
4
3
16
7
0
3
1
0
1
1
8
0
1
234
0
0
0
p98
0
5
5
6
22
20
0
10
1
1
3
1
15
0
3
606
2
1
2
p99
0
10
6
7
28
28
0
16
7
2
6
1
15
2
4
612
2
1
3
Exceedances of 300 ppb 1
mean
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
13
0
0
0
min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
med
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
p95
0
0
2
0
13
1
0
0
0
0
0
0
5
0
0
3
0
0
0
p98
0
1
3
1
14
3
0
1
1
0
1
0
8
0
0
423
0
0
1
p99
0
1
3
1
21
10
0
2
1
0
2
0
8
0
0
435
1
0
2
Notes:
1 The mean number of exceedances represents the number of exceedances occurring at all monitors in a particular location divided by the number of site-
years across the monitoring period. The min, med, p95, p98, and p99 represent the minimum, median, 95 , 98th, and 99th percentiles of the distribution for
:he number of exceedances in any one year within the monitoring period.
2 Colorado Springs monitoring data were collected as part of short-term study completed in September 2001 , therefore there are no 2001 -2006 data.
August 2008 - Draft
71
-------
1 7.3.3 Ambient Air Quality Adjusted to Just Meet the Current and Alternative
2 Standards
3 As described in section 6.2, each of the current and alternative standards were evaluated
4 using the more recent air quality data set (i.e., 2001-2006). Analysis results are presented for a
5 few selected locations, potential health effect benchmarks, and alternative standard levels, since
6 there were a total of 10 air quality scenarios (8 alternative standards, the current standard, and as
7 is), for each year group of data (2001-2003 and 2004-2006), for each of the monitor groups
8 (<100m and >100 m), and evaluated at 5 potential health effect benchmark levels (100, 150, 200,
9 250, 300 ppb 1-hour). All of the results for each location are provided in Appendix A, section 9,
10 much of which is summarized here in a series of key figures.
11 Figure 7-1 illustrates the estimated mean number of exceedances of the lowest
12 concentration levels (i.e., 100, 150, and 200 ppb) for each year-group air quality data adjusted to
13 just meeting the current annual average standard. The number of estimated exceedances of 100
14 ppb generally ranges from tens to several hundred, with subtle differences in the estimates for
15 each year-group and monitor siting category. For many of the locations, estimated number of
16 exceedances of 100 ppb are slightly higher for the 2004-2006 year-group when compared with
17 the 2001-2003 year-group, and the monitors sited at >100 m from a major road contained more
18 estimated exceedances that the monitors sited within 100 m of a major road. The estimated
19 number of exceedances of 150 and 200 ppb were much lower, for most locations the average
20 number of exceedances was under 100. Trends noted for these concentration levels were
21 consistent with that estimated for the 100 ppb level, with the lowest number of estimated
22 exceedances of 150 and 200 ppb associated with the 2001-2003 air quality for monitors < 100 m
23 of a major road. Note however that thirty-two of the 63 possible year-group and monitor-site
24 data combinations at the 19 locations did not have any exceedances of the 200 ppb level.
25 Figure 7-2 presents the mean estimated number of exceedances when considering the air
26 quality adjusted to just meeting the potential alternative standard levels, using Chicago as an
27 example to illustrate the relationship between the two forms of the standard. The trends in the
28 results presented for Chicago that apply to the other locations with a few exceptions. As
29 expected, the estimated number of exceedances is lower for a 99th percentile form compared with
30 each corresponding level using the 98th percentile form of alternative standard. In general, the
August 2008 Draft 72
-------
1 number of estimated exceedances of the potential health effect benchmark levels at monitoring
2 sites < 100 m from a major road is greater than the numbers estimated for monitors sited > 100 m
3 from a major road. This is what one would expect given the greater potential for vehicle
4 emissions influencing ambient concentrations at near road monitors. As expected, the number of
5 exceedances of the potential health effect benchmark levels decreases with decreasing alternative
6 standard level. Regardless of year-group or monitoring group, an alternative standard level of
7 100 ppb tended to reduce the number of estimated exceedances to either a few to none.
8 Figure 7-3 presents mean estimated number of exceedances of the 200 ppb concentration
9 level for a few additional locations, Phoenix, Los Angeles, Philadelphia, and St. Louis. Again,
10 there are trends in these results that are consistent with that reported for the Chicago results, with
11 few exceptions. For example, in St. Louis the estimated number of exceedances at monitors
12 located > 100 m from a major road were greater than those estimated using the monitoring sites
13 < 100 m from a major road. Also note that there were variable results when comparing year-
14 groups across the different locations within the monitor site-group; sometimes the year 2001-
15 2003 contained greater numbers of exceedances when compared with 2004-2006, other time not.
16 However, the alternative standard level of 100 ppb at either percentile consistently reduced the
17 number of benchmark exceedances.
18 Tables 7-12 and 7-13 summarize the annual mean concentrations and estimated number
19 of exceedances given 2001-2003 air quality adjusted that just meets the 1-hour 100 ppb 98th
20 percentile standard at monitors sited > 100 m and < 100 m from a major road, respectively. The
21 tables provide a more comprehensive comparison of the numbers of exceedances of the complete
22 range of potential health effect benchmarks for each of the locations, as well as providing upper
23 percentile estimates for each of the parameters. These particular results are provided to describe
24 trends within a given standard level, similar results are expected with differing year-group. The
25 complete results for all of the standard levels, including the observed number of exceedances (as
26 is air quality) provided in Appendix A, section 9. Most locations contained a mean of less than
27 100 exceedances of the 100 ppb concentration level, with upper percentile estimates ranging
28 from the tens to a few hundred. These results are comparably less than those estimated using air
29 quality adjusted to just meeting the current standard (Figure 7-1). At potential health effect
30 benchmark levels above 100 ppb, there were few estimated exceedances, particularly at and
31 above the 200 ppb level, considering both the mean and the upper percentiles.
August 2008 Draft 73
-------
1 Tables 7-14 summarizes the observed and estimated mean numbers of exceedances of
2 100 ppb using the 2001-2003 air quality as is and adjusted to just meeting the current standard
3 and the potential alternative 98th percentile standards at each location. The number of
4 exceedances for the as is air quality generally fell within the number of exceedances estimated
5 using alternative 1-hour 98th percentile standards of 50 ppb and 100 ppb at each location. When
6 the air quality was adjusted to just meeting the current annual average standard, the estimated
7 number of exceedances fell within the range of that estimated using the alternative 1-hour 98th
8 percentile standards of 100 ppb and 150 ppb at each location. In a similar manner, Table 7-15
9 summarizes the observed and estimated mean numbers of exceedances of 150 ppb 1-hour at each
10 location. The number of exceedances using as is air quality in each location was most similar to
11 that estimated using the alternative 1-hour 98th percentile standard of 50 ppb, while estimates
12 using the air quality adjusted to just meeting the current standard again fell within the range of
13 estimated numbers of exceedance using the alternative 1-hour 98th percentile standards of 100
14 ppb and 150 ppb at each location.
August 2008 Draft 74
-------
Monitors Sited >= 100 m from a Major Road
Monitors Sited < 100 m from a Major Road
2
3
4
5
6
1
8
9
10
12
13
14
16
17
18
19
20
21
22
23
24
^ {/) 1 £.\J\J -
| =5"g 1100
is >| 1000
| 1 S 900
00 = (0
o ° •£ 800
i- .± ai
j| < t 700
| o 0 600
c 2 ~ 500
S § g 400
-a « ^ 300
100
B> 150
• >200
_ n
>< U -t i i
fl)
o > •§ 1 000
.X 75 ,2 900
OQ = OT
•SO- 800
jj 5 | 700
| o 0 600
^ 4 1 500
J|l 400
TI « ~ 300
ra c ^ 200
£100
@> 150
• >200
_ n
ppb
ppb
ppb
1
L n
n
L, n IL
L
S|
i
1
, n
r
ppb
ppb
ppb
No data
i n
in 1 I
1
i
n
i
1
n
L n [L
-
-
-
_
n> 100 ppb
B> 150 ppb
• > 200 ppb
.
IL IL
8
n ^
• IL 5 fl
r
In
,n 11
-
-
-
-
-
n> 100 ppb
M> 150 ppb
• > 200 ppb
n
i
i
^ n
*
Ln
i
L
»
o
I
-£5 £3 £3
n-c n •§ •§
s[] i i
No data
L
8 8 8
I] 5 55
S SI
ill
g
Location
i
Location
® o
o
Figure 7-1. Estimated mean number of exceedances of selected 1-hour potential health effect benchmark levels, using recent air quality
adjusted to just meeting the current annual standard (0.053 ppm). (Top row contains 2001-2003 air quality, bottom row contains 2004-2006 air
quality. Left column contains monitors sited > 100 m of a major road, right column contains monitors sited < 100 m of a road.)
August 2008 Draft
75
-------
Q.
Q.
O
O
*"" l_
t is
O o>
>-
8 "
' I
01 £
x »1
111
Q.
Q.
*- IS
O 01
-
I?
li
01
I
2800
2400
2000
1600
1200
800
400
140
120
100 --
80
60
40 --
20
-•-2001-2003, <100m
-a-2001-2003, >=100m
• - 2004-2006, <100m
e- 2004-2006, >=100m
-•—2001-2003, <100m
-B—2001-2003, >=100m
-•- 2004-2006, <100m
e- 2004-2006, >=100m
-•—2001-2003, <100m
-Q- 2001 -2003, >=100m
• - 2004-2006, <100m
e- 2004-2006, >=100m
-•-2001-2003, <100m
-B—2001-2003, >=100m
• - 2004-2006, <100m
e- 2004-2006, >=100m
50 100 150 200
98th Percentile Alternative 1-hour Standard Level (ppb)
250 50 100 150 200
99th Percentile Alternative 1-hour Standard Level (ppb)
250
Figure 7-2. Estimated number of exceedances of potential health effect benchmarks (100 ppb, top; 200 ppb, bottom) in
Chicago given just meeting alternative 1-hour standard levels (98th percentile, left; and 99th percentile, right) using recent air
quality data from monitors sited < 100 m of a major road and sited >100 m of major roads.
August 2008 Draft
16
-------
-Phoenix-
-Los Angeles-
140
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
8 is
C <1>
TO ^—
o ro
« c
o .-
X i-
£ §
2 *
«i.
Q.
^3 £^
!§
E CM
4-1
V)
LU
O
tn
o ro
o) m
« c
HJ =
n- O
o .=
ti
D
1 £
+j ^3
I CM
I/)
LU
120
100 --
80
60
40 --
20
140
120 --
100 --
80
60
40 --
20
-•-2001-2003, <100m
-B-2001-2003, >=100m
-«- 2004-2006, <100m
-o- 2004-2006, >=100m
-•-2001-2003, <100m
-B-2001-2003, >=100m
«- 2004-2006, <100m
e- 2004-2006, >=100m
-Philadelphia-
-St. Louis-
-•-2001-2003, <100m
-a-2001-2003, >=100m
• - 2004-2006, <100m
e- 2004-2006, >=100m
-•-2001-2003, <100m
-B-2001-2003, >=100m
• - 2004-2006, <100m
-«- 2004-2006, >=100m
0 50 100 150 200 250 0 50 100 150 200
98th Percentile Alternative 1-hour Standard Level (ppb) 98th Percentile Alternative 1-hour Standard Level (ppb)
Figure 7-3. Estimated number of exceedances of 200 ppb in four locations (Phoenix, Los Angeles, Philadelphia, and St. Louis) given just meeting
alternative 1-hour 98th percentile standard levels using recent air quality data from monitors sited < 100 m of a major road and sited >100 m of major
roads.
250
August 2008 Draft
77
-------
1
2
3
4
5
Table 7-12 Estimated annual mean NO2 concentration and the number of exceedances of 1-hour NO2 concentration levels, using 2001-
2003 air quality adjusted to just meeting a 1-hour 100 ppb 98th percentile alternative standard, monitoring locations sited > 100 m of a
major road.
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington
DC
Atlanta
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other
MSA/CMSA
Other Not
MSA
Site-
Years
6
9
3
2
6
51
6
26
14
18
14
12
2
16
5
3
9
612
127
Annual Mean (ppb)
Mean
18
33
35
35
40
26
22
29
34
34
22
28
36
19
39
47
35
16
13
Min
10
26
34
32
36
6
17
16
25
16
7
20
36
4
32
43
29
1
2
Med
21
31
34
35
39
28
23
27
32
39
27
30
36
14
42
48
34
17
12
p99
22
43
36
38
45
43
26
45
50
46
41
34
37
41
43
49
41
31
33
Number of Exceedances of 1-Hour Level
>100 ppb
Mean
4
37
72
58
146
21
85
19
58
93
61
50
160
37
66
175
82
2
9
Min
0
1
49
54
88
0
5
0
4
0
0
13
124
0
8
66
6
0
0
Med
2
17
75
58
140
9
43
9
33
71
17
40
160
2
91
206
32
0
0
p99
18
160
92
61
217
112
243
89
244
274
335
94
195
172
115
253
214
24
180
>150 ppb
Mean
0
1
2
2
18
1
6
0
1
3
3
2
10
0
0
1
2
0
1
Min
0
0
1
1
1
0
0
0
0
0
0
0
4
0
0
0
0
0
0
Med
0
0
2
2
7
0
4
0
1
1
0
1
10
0
0
0
0
0
0
p99
2
5
3
2
47
13
18
4
3
10
23
10
15
3
0
2
9
3
25
> 200 ppb
Mean
0
0
0
1
8
0
1
0
0
0
0
0
1
0
0
0
0
0
1
Min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Med
0
0
0
1
3
0
0
0
0
0
0
0
1
0
0
0
0
0
0
p99
0
1
1
1
30
5
2
0
1
0
3
1
2
0
0
0
1
0
7
> 250 ppb
Mean
0
0
0
0
5
0
0
0
0
0
0
0
1
0
0
0
0
0
0
Min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Med
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
0
0
p99
0
0
0
0
25
2
0
0
1
0
1
1
2
0
0
0
0
0
6
> 300 ppb
Mean
0
0
0
0
3
0
0
0
0
0
0
0
1
0
0
0
0
0
0
Min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Med
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
0
0
p99
0
0
0
0
15
0
0
0
1
0
1
0
2
0
0
0
0
0
1
August 2008 Draft
78
-------
1
2
3
4
5
6
Table 7-13. Estimated annual mean NO2 concentration and the number of exceedances of 1-hour NO2 concentration levels, using 2001-
2003 air quality adjusted to just meeting a 1-hour 100 ppb 98th percentile alternative standard, monitoring locations sited <
100 m of a major road.
Location1
Boston
Chicago
Cleveland
Denver
Los Angeles
Miami
New York
Philadelphia
Washington
DC
El Paso
Las Vegas
Phoenix
St. Louis
Site-
Years
19
10
3
2
44
6
20
7
14
3
6
5
17
Annual Mean NO2
(ppb)
Mean
34
42
44
53
30
26
43
43
39
39
26
44
31
Min
13
34
42
52
5
15
30
33
26
37
6
31
17
Med
39
45
43
53
31
25
41
42
42
40
28
50
33
p99
57
50
46
55
48
40
58
53
48
40
42
54
49
Number of Exceedances of 1-Hour Level
>100 ppb
Mean
67
120
165
171
40
103
74
92
92
158
89
105
46
Min
0
20
127
104
0
34
4
14
0
117
0
1
0
Med
44
112
144
171
25
81
50
67
87
131
81
135
25
p99
221
267
224
237
160
252
277
230
197
226
196
201
202
>150 ppb
Mean
2
4
8
17
1
4
2
2
1
13
2
1
2
Min
0
0
5
8
0
0
0
0
0
5
0
0
0
Med
0
1
6
17
0
1
0
2
0
16
0
0
0
p99
8
37
12
26
8
17
18
3
6
17
12
3
11
> 200 ppb
Mean
0
1
0
5
0
0
0
0
0
0
0
0
0
Min
0
0
0
1
0
0
0
0
0
0
0
0
0
Med
0
0
0
5
0
0
0
0
0
0
0
0
0
p99
1
7
0
8
3
2
2
0
1
0
0
0
1
> 250 ppb
Mean
0
0
0
0
0
0
0
0
0
0
0
0
0
Min
0
0
0
0
0
0
0
0
0
0
0
0
0
Med
0
0
0
0
0
0
0
0
0
0
0
0
0
p99
0
0
0
0
1
0
0
0
0
0
0
0
0
> 300 ppb
Mean
0
0
0
0
0
0
0
0
0
0
0
0
0
Min
0
0
0
0
0
0
0
0
0
0
0
0
0
Med
0
0
0
0
0
0
0
0
0
0
0
0
0
p99
0
0
0
0
1
0
0
0
0
0
0
0
0
1 Detroit, Atlanta, and Provo did not have any monitors sited within 1 00 m of a major road. The Other CMSA/MSA and Other Not MSA locations did not
have estimated distances of monitors to major roads.
August 2008 Draft
79
-------
1
2
3
4
5
Table 7-14. Estimated mean number of exceedances of 100 ppb 1-hour NO2 concentrations, using 2001-2003 air quality as is and that
adjusted to just meeting the current and alternative standards (98th percentile) for monitoring locations sited > 100 m and <
100 m of a major road.
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington DC
Atlanta
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other MSA/CMSA
Other Not MSA
Sites >= 100 m of a major road
As is
0
1
0
2
9
7
0
1
0
0
1
0
1
0
0
0
0
0
1
Current
std
8
71
233
525
438
63
438
23
95
228
434
385
732
260
91
512
223
48
121
Alternative 1-hour 98tn percentile
standard
50
0
0
0
1
8
0
1
0
0
0
0
0
1
0
0
0
0
0
1
100
4
37
72
58
146
21
85
19
58
93
61
50
160
37
66
175
82
2
9
150
163
525
674
932
1058
241
454
331
111
896
429
622
821
533
1064
2187
798
42
77
200
546
1568
1707
2318
2461
914
1044
1299
2041
1974
924
1553
1770
1152
2582
3660
1941
240
284
Sites < 100 m of a major road
As is
0
4
0
19
13
0
3
0
0
2
0
2
0
Current
std
119
194
491
152
113
546
67
146
232
768
543
133
141
Alternative 1-hour 98tn percentile
standard
50
0
1
0
5
0
0
0
0
0
0
0
0
0
100
67
120
165
171
40
103
74
92
92
158
89
105
46
150
812
1075
1241
1836
403
566
999
1278
1061
1112
1038
1681
570
200
1863
2721
2865
4161
1403
1214
2837
2873
2476
2330
1825
3238
1687
August 2008 Draft
80
-------
1
2
3
4
5
Table 7-15. Estimated mean number of exceedances of 150 ppb 1-hour NO2 concentrations, using 2001-2003 air quality as is and that
adjusted to just meeting the current and alternative standards (98th percentile) for monitoring locations sited > 100 m and <
100 m of a major road.
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington DC
Atlanta
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other MSA/CMSA
Other Not MSA
Sites >= 100 m of a major road
As is
0
0
0
0
3
0
0
0
0
0
0
0
1
0
0
0
0
0
0
Current
std
0
2
11
62
45
4
76
1
2
10
62
25
134
10
0
5
11
2
14
Alternative 1-hour 98tn percentile
standard
50
0
0
0
0
3
0
0
0
0
0
0
0
1
0
0
0
0
0
0
100
0
1
2
2
18
1
6
0
1
3
3
2
10
0
0
1
2
0
1
150
4
37
72
58
146
21
85
19
58
93
61
50
160
37
66
175
82
2
9
200
56
301
398
465
664
129
315
177
399
514
266
322
585
288
617
1476
470
19
43
Sites < 100 m of a major road
As is
0
0
0
1
0
0
0
0
0
0
0
0
0
Current
std
4
8
34
16
6
86
2
4
7
79
22
2
6
Alternative 1-hour 98tn percentile
standard
50
0
0
0
0
0
0
0
0
0
0
0
0
0
100
2
4
8
17
1
4
2
2
1
13
2
1
2
150
67
120
165
171
40
103
74
92
92
158
89
105
46
200
431
660
768
1015
225
401
589
679
589
686
615
996
309
August 2008 Draft
81
-------
1 7.3.4 On-Road Concentrations Derived From Ambient Air Quality Adjusted to Just
2 Meet the Current and Alternative Standards
3 Just as was done with the as is air quality data, on-road NC>2 concentrations were
4 estimated using the air quality adjusted to just meeting the current and alternative standard and
5 the approach described in section 7.2.3. The analysis was performed using the more recent air
6 quality separated into two year-groups (2001-2003 and 2004-2006) based on the form of the
7 potential alternative standards (i.e., a 3-year average). Results are presented in a manner
8 consistent with section 7.3.3, whereby the number of exceedances of the potential benchmark
9 levels were estimated. However, for the sake of brevity only key figures and tables are provided
10 here. The complete results for the estimated on-road concentrations and numbers of benchmark
11 exceedances are provided in Appendix A, section 9.
12 Figure 7-4 illustrates the estimated mean number of exceedances of the 100, 150, and 200
13 ppb levels on-roads, given 2001-2003 air quality adjusted to just meeting the current annual
14 average standard. Most locations contained an average of hundreds to thousands of estimated
15 exceedances of 100 ppb, much greater than those estimated using either the ambient monitors
16 sited < 100 m or > 100 m of a major road (Figure 7-1). The estimated number of exceedances of
17 the 150 and 200 ppb levels were also higher on-roads, most locations were estimated to contain
18 several hundred exceedances of 150 ppb and a few hundred exceedances of 200 ppb using air
19 quality concentrations adjusted to just meeting the current standard.
20 The effect of the potential alternative standards on the estimated on-road NC>2
21 concentrations was also analyzed at each of the locations. Figure 7-5 illustrates each of the
22 standard levels (50, 100, 150, and 200 ppb 1-hour) and the two forms (98th and 99th percentiles)
23 investigated, again using Chicago as an example to describe observed trends. The trends
24 observed in Figure 7-2 and described in section 7.3.3 are similar to that observed here, albeit
25 with greater numbers of exceedances estimated on-roads compared with those estimated for
26 monitors near-roads or sited at a distance from major roads. Estimated numbers of
27 concentrations above 100 ppb are several hundred to a thousand considering a standard level of
28 100 ppb (either perecentile), however exceedances of 200 ppb are estimated to be under one
29 hundred.
August 2008 Draft 82
-------
1 Similar numbers of exceedances on-roads were estimated at other locations using air
2 quality adjusted to just meeting the potential alternative standards. Figure 7-6 illustrates the
3 estimated number of exceedances of 200 ppb at four selected locations as an example, Phoenix,
4 Los Angeles, Philadelphia, and St. Louis, using a 98th percentile form of a 1-hour standard. The
5 number of concentrations above 200 ppb is similar at each of the locations (including Chicago),
6 particularly when comparing the 100 ppb standard level, ranging from tens to just under 100.
7 Table 7-16 presents a more comprehensive comparison at this particular standard level (98th
8 percentile at 100 ppb) using 2001-2003 adjusted air quality at each of the locations. For most
9 locations, the estimated mean number of exceedances of 200 ppb on-roads was 100 or less, with
10 upper percentiles estimated to number about one to several hundreds of exceedances. The mean
11 number of exceedances of 250 and 300 ppb were less, ranging from a few to tens of occurrences
12 in a year.
13 Tables 7-17 and 7-18 summarizes the observed and estimated mean numbers of
14 exceedances of 100 and 150 ppb on-roads, respectively, using all the recent air quality as is and
15 that adjusted to just meeting the current standard and the potential alternative 98th percentile
16 standards at each location. Trends for the as is air quality and that adjusted to just meeting the
17 current followed similar trends to that observed for the monitors sited > 100 m and < 100 m of a
18 major road (see Tables 7-14 and 7-15, for the 2001-2003 air quality). The estimated number of
19 exceedances on-roads using the as is data fell within the range of estimates provided by the
20 alternative 1-hour 98th percentile standards of 50 and 100 ppb, while the estimated on-road
21 exceedances of 150 ppb fell within the range of provided by the 100 and 150 ppb alternative
22 standards.
August 2008 Draft 83
-------
3
4
•o
| II -o 3000 -
S = $ 2400
??!
1 < 5 1800 -
Ego
3 0 .C
Z CM *^
_ — U>
i 5 = 1200 -
o o -K
g CM g
•o 10" S
^ 0
1
D> 100 ppb
• S 150 ppb
• >200 ppb
i_
I
1
"1
I
1
LU COTJ'-^W-E
O 0) C ^ O (1) t
Tn
-------
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
0 4000 -
•§ «
g £ 3500 -
o n
U .E 3000 -
"5 5
0 5 2500 -
^ O 2000 -
ra -Q
0 g- 1500 -
o
J> ? 1000 -
^ 50° "
UJ
0 -
1i!UU -,
0 1 1 00
I i 100°
g >• 900
x 5
UJ .E 800
"5 5
o ° 70°
E ^ 600
z i 500
| g; 400
T> 8 300
v CM
| "5 200
S 10°
0
: /"
-•-2001-2003 /
-•-2004-2006 / /*
: / /
/ /
m /
/ /
: / /
- / /
/ /
v/
: //*
: //
•^
r - „„ ^ x
/ 7.
//
; / /
\ / /
B-^^^^ 1 1 1
-•-2001-2003
-•- 2004-2006 /
/ •
/ /
- / /
: / /
/ /
-/ /
- / /
/ /
'•-- //
/ /
/ /
/ /
/ /
/ /
; jS/
: //^
w^
: -B-2001-2003
: -•- 2004-2006
: •
: / »
: / /
/ /
; X^/
'' ///^
0
50 100 150 200 250 0 50 100 150 200
98th Percentile Alternative Standard Level (ppb) 99th Percentile Alternative Standard Level (ppb)
Figure 7-5. Estimated number of exceedances of potential health effect benchmarks (100 ppb, top; 200 ppb, bottom) on-roads in Chicago
250
given just meeting alternative 1-hour standard levels (98th percentile, left; and 99th percentile, right) using recent air quality data.
August 2008 Draft
85
-------
-Phoenix-
-Los Angeles-
2 o 1800
as |=
A ~° S 1600
5 8 ^ 140°
6 £!
7 2 o 120°
8 c ^ 1000
9 |d
10 =z 80°
11 8|
11 S °- 600
12 -og
13 $£ 400
14 E
}7 5 200
15 uj
1 A n
10 OH
17
|
-•-2001-2003 //
-•-2004-2006 /
/
/
/
/
,f
/
/
/
: -•- 2001-2003
-•- 2004-2006
:
: /*
: /
: / ,•
/ //
: / /
: y /
\ __^^
0 50 100 150 200 25
18 -Philadelphia- -St. Louis-
20 »
01 o 1800
Zl c
22 •§ S 1600 -
OO OJ ^"
2.3 o ra
24 uj = 140°
25 o | 1200_
26 ||
27 E ^ 1000 -
98 19
Z0 c z 800
29 ^ -°
^^ d) ^
30 ^£ eoo
31 "S S
S u_ 400
32 E °
33 % 200
34 " o
-•-2001-2003
-•-2004-2006
P
//
//
/?
^^
-^-^—^-^
-•-2001-2003
-•- 2004-2006
/ /
/ /
/ /
^ /
/,/
/ ,/
///
^^^^=4^
35 '
JJ 0 50 100 150 200 250 0 50 100 150 200 25
37
98th Percentile Alternative Standard Level (ppb)
98th Percentile Alternative Standard Level (ppb)
Figure 7-6. Estimated number of exceedances of 200 ppb in-roads in four locations (Phoenix, Los Angeles, Philadelphia, and
»th
St. Louis) given just meeting alternative 1-hour 98 percentile standard levels using recent air quality data.
August 2008 Draft
86
-------
1 Table 7-16. Estimated annual mean NO2 concentration and the number of exceedances of 1-hour NO2 concentration levels on-roads,
2 using 2001-2003 air quality adjusted to just meeting a 1-hour 100 ppb 98th percentile alternative standard.
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington
DC
Atlanta
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other
MSA/CMSA
Other Not MSA
Site-
Years
600
900
300
200
600
5100
600
2600
1400
1800
1400
1200
200
1600
500
300
900
61200
12700
Annual Mean NO2
(ppb)
Mean
33
60
63
63
72
48
40
52
63
62
39
51
66
35
71
85
63
30
24
Min
13
33
43
40
46
7
22
20
32
20
9
24
46
5
41
55
36
1
3
Med
34
58
62
60
69
47
39
49
58
63
42
50
65
25
69
82
61
29
21
p99
57
104
88
96
110
96
62
105
116
117
93
82
93
94
112
127
99
65
67
Number of Exceedances of 1-Hour Level
>100 ppb
Mean
411
1197
1306
1589
1793
701
820
906
1509
1445
704
1097
1374
839
1876
2950
1441
188
202
Min
1
44
254
265
419
0
56
0
52
1
0
62
451
0
77
664
93
0
0
Med
302
951
1224
1395
1670
450
771
661
1288
1305
470
988
1312
272
1820
2998
1321
52
33
p99
1511
4002
2727
3446
3929
3357
2054
3630
4554
4550
3040
2693
2842
3736
4400
5067
3589
1555
1700
>150 ppb
Mean
66
283
327
383
516
142
251
171
343
401
191
256
422
232
462
913
366
24
38
Min
0
0
33
11
37
0
1
0
0
0
0
2
25
0
2
19
0
0
0
Med
12
138
256
237
377
43
164
65
171
183
44
154
370
37
278
715
227
1
2
p99
541
1564
1003
1621
1748
1145
1215
1310
2045
2317
1556
1302
1185
2062
2165
3311
1766
358
564
>200 ppb
Mean
8
71
92
92
157
31
80
37
82
107
53
57
121
61
100
227
91
4
9
Min
0
0
0
0
1
0
0
0
0
0
0
0
3
0
0
1
0
0
0
Med
0
14
44
19
100
3
30
6
18
20
3
21
74
2
11
83
26
0
0
p99
90
641
393
608
629
374
647
412
706
828
624
403
491
687
769
1512
663
89
154
> 250 ppb
Mean
1
21
30
23
61
7
24
11
23
32
16
15
34
19
19
60
25
1
3
Min
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Med
0
1
7
3
31
0
4
0
1
1
0
4
16
0
0
4
3
0
0
p99
21
291
176
217
312
117
291
181
350
297
225
107
189
328
156
401
243
18
59
> 300 ppb
Mean
0
7
12
6
29
2
8
4
7
10
5
4
11
5
4
19
8
0
1
Min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Med
0
0
1
1
7
0
0
0
0
0
0
0
5
0
0
0
0
0
0
p99
9
118
85
54
162
36
118
74
153
135
91
34
61
132
43
178
113
4
27
August 2008 Draft
87
-------
2 Table 7-17. Estimated mean number of exceedances of 100 ppb 1-hour NO2 concentrations on-roads, using air quality as is and that
3 adjusted to just meeting the current and alternative standards (98th percentile).
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington DC
Atlanta
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other MSA/CMSA
Other Not MSA
2001 -2003 Air Quality
As is
12
252
103
403
185
414
21
205
161
156
98
85
34
88
527
241
91
54
9
Current
std
455
1478
2065
2384
2779
1170
1680
900
1788
1941
1572
2053
2790
1347
1932
3555
2057
804
748
Alternative 1-hour 98tn percentile
standard
50
8
71
92
92
157
31
80
37
82
107
53
57
121
61
100
227
91
4
9
100
411
1197
1306
1589
1793
701
820
906
1509
1445
704
1097
1374
839
1876
2950
1441
188
202
150
1172
2918
2996
3064
3642
2081
1743
2430
3340
3041
1550
2353
2916
1605
3841
4716
3129
760
610
200
1865
4311
4402
3801
4863
3258
2504
3598
4566
4247
2296
3215
4086
2143
4880
5567
4483
1451
1078
2004-2006 Air Quality
As is
5
151
294
81
177
17
168
87
80
59
67
45
55
353
394
50
32
10
Current
std
462
1357
2163
2835
1184
1487
1050
1914
1697
1665
2324
2755
1206
2309
2971
1785
886
737
Alternative 1-hour 98tn percentile
standard
50
8
42
181
170
50
47
51
72
75
43
78
131
61
83
195
55
11
9
100
372
934
1971
1834
984
586
1072
1381
1202
673
1200
1280
767
1909
678
1055
359
197
150
1045
2546
3235
3440
2390
1284
2596
2992
2575
1487
2426
2673
1416
3807
1995
2434
1101
590
200
1735
3841
3842
4552
3366
1863
3774
4127
3639
2200
3214
3839
1932
4812
3195
3650
1859
1008
August 2008 Draft
88
-------
2 Table 7-18. Estimated mean number of exceedances of 150 ppb 1-hour NO2 concentrations on-roads, using air quality as is and that
3 adjusted to just meeting the current and alternative standards (98th percentile).
Location
Boston
Chicago
Cleveland
Denver
Detroit
Los Angeles
Miami
New York
Philadelphia
Washington DC
Atlanta
El Paso
Jacksonville
Las Vegas
Phoenix
Provo
St. Louis
Other MSA/CMSA
Other Not MSA
2001 -2003 Air Quality
As is
0
33
14
51
34
67
1
24
17
17
12
7
3
10
57
21
8
5
1
Current
std
79
395
677
999
1079
295
761
178
472
656
714
820
1295
583
503
1452
683
203
269
Alternative 1-hour 98tn percentile
standard
50
0
7
12
6
29
2
8
4
7
10
5
4
11
5
4
19
8
0
1
100
66
283
327
383
516
142
251
171
343
401
191
256
422
232
462
913
366
24
38
150
411
1197
1306
1589
1793
701
820
906
1509
1445
704
1097
1374
839
1876
2950
1441
188
202
200
922
2362
2440
2692
3112
1607
1457
1944
2790
2574
1275
1993
2412
1393
3300
4282
2627
540
470
2004-2006 Air Quality
As is
0
15
34
6
18
1
17
5
5
5
5
11
7
25
214
4
2
1
Current
std
91
335
761
1263
296
745
226
623
587
803
1114
1329
561
640
1360
647
265
274
Alternative 1-hour 98tn percentile
standard
50
0
2
16
16
3
4
4
4
5
3
6
25
9
3
71
4
1
1
100
64
190
626
581
220
170
231
325
316
174
317
394
227
436
266
249
63
41
150
372
934
1971
1834
984
586
1072
1381
1202
673
1200
1280
767
1909
678
1055
359
197
200
819
1996
2922
2966
1944
1066
2129
2504
2150
1221
2069
2245
1225
3305
1526
1991
839
439
August 2008 Draft
89
-------
1 7.4 UNCERTAINTY ANALYSIS
2 This uncertainty analysis first identifies the sources of the assessment that do or do not contribute
3 to uncertainty, and provide a rationale for why this is the case. A qualitative evaluation follows
4 for the types and components of uncertainty, resulting in a summary describing, for each source
5 of uncertainty, the direction of influence the uncertainty may have on the surrogate exposure
6 estimates. This bias direction indicates how the source of uncertainty is judged to influence
7 estimated concentrations, either the concentrations are likely "over-" or "under-estimated". In
8 the instance where two or more types or components of uncertainty result in offsetting direction
9 of influence, the uncertainty was judged as "both". "Unknown" was assigned where there was
10 no evidence reviewed to judge the uncertainty associated with the source. Table 7-19 provides a
11 summary of the sources of uncertainty identified in the air quality characterization and the
12 overall judged bias of each.
13 7.4.1 Air Quality Data
14 One basic assumption is that the AQS NO2 air quality data used are quality assured
15 already. Reported concentrations contain only valid measures, since values with quality
16 limitations are either removed or flagged. There is likely no selective bias in retention of data
17 that is not of reasonable quality, it is assumed that selection of high concentration poor quality
18 data would be just as likely as low concentration data of poor quality. Given the numbers of
19 measurements used for this analysis, it is likely that even if a few low quality data are present in
20 the data set, they would not have any significant effect on the results presented here. Therefore,
21 the air quality data and database used likely contributes minimally to uncertainty. Temporally,
22 the data are hourly measurements and appropriately account for variability in concentrations that
23 are commonly observed for NO2 and by definition are representative of an entire year. In
24 addition, having more than one monitor does account for some of the spatial variability in a
25 particular location. However, the degree of representativeness of the monitoring data used in this
26 analysis can be evaluated from several perspectives, one of which is how well the temporal and
27 spatial variability are represented. In particular, missing hourly measurements at a monitor may
28 introduce bias (if different periods within a year or different years have different numbers of
29 measured values) and increase the uncertainty. Furthermore, the spatial representativeness will
30 be poor if the monitoring network is not dense enough to resolve the spatial variability (causing
August 2008 Draft 90
-------
1 increased uncertainty) or if the monitors are not evenly distributed (causing a bias). Additional
2 uncertainty regarding temporal and spatial representation by the monitors is expanded below.
3 7.4.2 Measurement Technique for Ambient NOi
4 One source of uncertainty for NC>2 air quality data is due to interference with other
5 oxidized nitrogen compounds. The ISA points out positive interference, commonly from HNOs,
6 of up to 50%, particularly during the afternoon hours, resulting in overestimation of
7 concentrations. Also, negative vertical gradients exist for monitors (2.5 times higher at 4 meter
8 vs. 15 meter vertical siting (ISA, section 2.5.3.3), thus monitors positioned on rooftops may
9 underestimate exposures. Only 7 of the 17712 monitors in the named locations contained
10 monitoring heights of 15 meters or greater, with nearly 60% at 4 meters or less height, and 80%
11 at 5 meters or less in height. Not accounting for this potential vertical gradient in NO2
12 concentrations may generate underestimates of exceedances for some sites, however the overall
13 impact of inferences made for the locations included in this assessment is likely minimal since
14 most monitors are sited at less than 4-5 meters in vertical height. In addition, the relationship at
15 heights below 4 meters is uncertain (e.g., a breathing height of 2 meters is commonly used) and
16 therefore would add an unknown bias to the estimated NO2 concentrations above a benchmark
17 when used as a surrogate for human exposure.
18 7.4.3 Temporal Representation
19 Data are valid hourly measures and are of similar temporal scale as identified health
20 effect benchmark concentrations. There are frequent missing values within a given valid year
21 which contribute to the uncertainty as well as introducing a possible bias if some seasons, day
22 types (e.g., weekday/weekend), or time of the day (e.g., night or day) are not equally represented.
23 Since a 75 percent daily and hourly completeness rule was applied, some of these uncertainties
24 and biases were reduced in these analyses. Data were not interpolated in the analysis. Similarly,
25 there may be bias and uncertainty if the years monitored vary significantly between locations.
26 Although monitoring locations within a region do change over time, the NO2 network has been
27 reasonably stable over the 1995-2006 period, particularly at locations with larger monitoring
28 networks, so the impact to uncertainty is expected to be minimal regarding the bias direction. It
12 28 monitors did not have height reported (therefore, 177 + 28 = 205 total number of monitors in named locations)
August 2008 Draft 91
-------
1 should also be noted that use of the older data in some of the analyses here carries the
2 assumption that the sources present at that time are the same as current sources, adding
3 uncertainty to results if this is not the case. Separating the data into two 6-year groups (historic
4 and recent for the as is evaluation) and two further subsets of the more recent air quality (2001-
5 2003 and 2004-2006) before analysis reduces the potential impact from changes in national- or
6 location-specific source influences and is judged to have a minimal bias.
7 7.4.4 Spatial Representation
8 Relative to the physical area, there are only a small number of monitors in each location.
9 Since most locations have sparse siting, the monitoring data are assumed to be spatially
10 representative of the locations analyzed here. This includes areas between the ambient monitors
11 that may or may not be influenced by similar local sources of NC>2. For these reasons the
12 uncertainty and bias due to the spatial network may be moderate, although the monitoring
13 network design should have addressed these issues within the available resources and other
14 monitoring constraints. Bias would be most prevalent in locations with the fewest monitors,
15 although the direction being largely unknown. In addition, the air quality characterization used
16 all monitors meeting the 75 percent completeness criteria, without taking into account the
17 monitoring objectives or land use for the monitors. Thus, there will be some lack of spatial
18 representation and uncertainty due to the inclusion/exclusion of some monitors that are very near
19 local sources (including mobile sources) resulting in both over- or under- estimations.
20 7.4.5 Air Quality Adjustment Procedure
21 There is uncertainty in the air quality adjustment procedures due to the uncertainty of the
22 true relationship between the adjusted concentrations and the as is air quality. The adjustment
23 factors used for the current and alternative standards each assumed that all hourly concentrations
24 will change proportionately. However, the impact of the adjustment on the estimated
25 concentrations is a function of the particular form and level of the standard simulated and,
26 depending on whether concentrations are adjusted upwards or downward, will vary.
27 Different sources have different temporal emission profiles, so that equally applied
28 changes to the concentrations at the ambient monitors to simulate hypothetical changes in
29 emissions may not correspond well with all portions of the concentration distribution. When
30 adjusting concentrations upward to just meeting the current standard, the proportional adjustment
August 2008 Draft 92
-------
1 used an equivalent multiplicative factor for all portions of the concentration distribution, the
2 upper tails were treated the same as the area of central tendency. This may not necessarily
3 reflect changes in an overall emissions profile that may result from, for example, an increase in
4 the number of sources in a location. It is possible that while the mean concentration measured at
5 an ambient monitor may increase with an increase in the sources affecting concentrations
6 measured at the monitor, the tails of the distribution might not have a proportional increase.
7 Adjusting the ambient concentrations upwards to simulate the alternative standards also carries a
8 similar degree of uncertainty however the multiplicative factors are derived from the upper
9 percentiles of the 1-hour concentrations and applied to the 1-hour concentrations equally. In
10 each of these instances of adjusting the concentrations upwards, there may be an associated over-
11 estimation in the concentrations at the upper tails of the distributions, leading to over-estimation
12 in the numbers of exceedances. In adjusting concentrations downward (e.g., the alternative
13 standard level of 50 ppb 1-hour, 99th percentile), the use of a proportional multiplicative
14 adjustment derived from and applied to the upper tails of the concentration distribution may
15 better represent what might occur to emissions with added source controls. However it is likely
16 that the mean concentrations and lower percentiles of the distribution are under-estimated.
17 Similarly, emission changes that would affect the concentrations at the design monitor
18 containing the highest concentration (annual mean, 98th or 99th percentile 1-hour) may not
19 necessarily impact lower concentration sites proportionately. This could result in
20 overestimations in the number of exceedances at lower concentration sites within a location,
21 however it is likely to be minimal given that the greatest numbers of exceedances typically were
22 measured at the monitoring sites with the highest concentrations within the location (Appendix
23 A, section 7). This bias would be less in locations containing several monitors, such as Boston,
24 New York, or Los Angeles. Universal application of the proportional simulation approach at
25 each of the locations was done for consistency and was designed to preserve the inherent
26 variability in the concentration profile. A few locations were noted that may have an exceptional
27 number of estimated exceedances as a result of the air quality adjustment approach, particularly
28 those locations with few monitoring sites that contained very low concentrations and/or atypical
29 variability in hourly concentrations. These few locations (e.g., Miami, Jacksonville, Provo) may
30 contain overestimations at the upper tails of the concentration distribution, leading to bias in
August 2008 Draft 93
-------
1 estimated number of exceedances at both the upper percentiles and the mean when using the air
2 quality simulated to just meet the current standards.
3 7.4.6 On-Road Concentration Simulation
4 On-road and ambient monitoring NC>2 concentrations have been shown to be correlated
5 significantly on a temporal basis (e.g., Cape et al., 2004) and motor vehicles are a significant
6 emission source of NOX, providing support for estimating on-road concentrations using ambient
7 monitoring data. The relationship used in this analysis to estimate on-road NC>2 concentrations
8 was derived from data collected in measurement studies containing mostly long-term averaging
9 times, typically 14-days or greater in duration (e.g., Roorda-Knape, 1998; Pleijel et al., 2004;
10 Cape et al, 2004), although one study was conducted over a one-hour time averaging period
11 (Rodes and Holland, 1981). This is considered appropriate in this analysis to estimate on-road
12 hourly concentrations from hourly ambient measures, assuming a direct relationship exists
13 between the short-term peaks to time-averaged concentrations (e.g., hourly on-road NC>2
14 concentrations are correlated with 24-hour averages). While this should not impact the overall
15 contribution relationship between vehicles and ambient concentrations on roads, the decay
16 constant k will differ for shorter averaging times. The on-road concentration estimation also
17 assumes that concentration changes that occur on-road and at the monitor are simultaneous (i.e.,
18 within the hour time period of estimation). Since time-activity patterns of individuals are not
19 considered in this analysis, there is no bias in the number of estimated exceedances. The long-
20 term data used to develop the algorithm used were likely collected over variable meteorological
21 conditions (e.g., shifting wind direction) and other influential attributes (e.g., rate of
22 transformation of NO to NC>2 during the daytime versus nighttime hours) than would be observed
23 across shorter time periods. This could result in either over- or under-estimations of
24 concentrations, depending on the time of day. The variability in NC>2 concentration within an
25 hour was also not considered in this analysis, that is, the on-road concentration at a given site
26 will likely vary during the 1-hour time period. If considering personal exposures to individuals
27 within vehicles that are traveling on a road, it is likely that their exposure concentrations would
28 also vary due to differing roadway concentrations. This could also result in either over- or
29 under-estimations of concentrations, depending on the duration of travel and type of road
30 traveled on.
August 2008 Draft 94
-------
1 On-road concentrations were not modified in this analysis to account for in-vehicle
2 penetration and decay. Therefore, in-vehicle concentrations would be overestimated if using the
3 on-road concentrations as a surrogate, given that reactive pollutants (e.g., PM^.s) tend to have a
4 lower indoor/outdoor (I/O) concentration ratio (Rodes et al., 1998). Chan and Chung (2003)
5 report mean (I/O) ratios of NO2 for a few roadways and driving conditions in Hong Kong. On
6 highways and urban streets, the value is centered about 0.6 to 1.0, indicating decay of NO2 as it
7 enters the vehicle.
8 At locations where traffic counts are very low (e.g., on the order of hundreds/day) the on-
9 road contribution has been shown to be negligible (Bell and Ashenden, 1997; Cape et al., 2004),
10 therefore any monitors sited in rural areas with minimal traffic volumes may result in small
11 overestimations of NO2 concentrations using equation (7-2) at these locations. Monitors sited
12 within 100 m of the roadway were not used in the calculation of on-road concentrations due to
13 the possibility of these monitors already accounting for notable impact from vehicle emissions
14 (e.g., Beckerman et al., 2008), thus controlling for a double-counting of on-road concentrations.
15 However, there is potential for influence by non-road source emissions on the measured
16 concentrations at the monitors used (> 100 m froma major road), contrary to an assumption that
17 there is an absence of direct source influence (only mobile sources were controlled for by
18 selecting monitors these monitors). Therefore, at certain monitors directly affected by emissions
19 from non-road sources, the simulated on-road concentrations may be over-estimated. Another
20 source of uncertainty in the spatial heterogeneity of NO2 concentrations regards the presence of
21 street canyons on roadways. These localized areas may be subject to highly variable
22 concentrations within a short span of a road, often defined by the presence of man-made
23 structures, such as buildings, on both sides of the road. A comparison of street canyon measured
24 NOX concentrations with those measured at a reference site (termed background) indicate that
25 there is about a factor of 2.3 difference in the concentrations (Ghenu et. al, 2007). Vardoulakis
26 et al. (2004) reported mean NO2 concentrations at a major intersection can be a factor of about
27 2.1 times greater than on-road concentrations measured at a few hundred meters distance within
28 a street canyon.13 Because these factors are within the range of simulation factors used here in
29 estimating the on-road concentration, i.e., ranging from a factor of 1.2 to 3.7 times the ambient
13 Ambient concentrations at a site not influenced by mobile sources were not reported in this Vardoulakis et al.
(2004).
August 2008 Draft 95
-------
1 concentrations, it is likely that some of the estimated on-road concentrations are similar in
2 magnitude to those found in street canyons. In addition, NOX is primarily emitted as NO (e.g.,
3 Heeb et al., 2008; Shorter et al., 2005), with substantial secondary formation due predominantly
4 to NO + Os -^ NO2 + O2. Numerous studies have demonstrated the Os reduction that occurs
5 near major roads, reflecting the transfer of odd oxygen to NO to form NO2, a process that can
6 impact NO2 concentrations both on- and downwind of the road. Some studies report NO2
7 concentrations increasing just downwind of roadways and that are inversely correlated with Os
8 (e.g., Beckerman et al., 2008), suggesting that peak concentration of NO2 may not always occur
9 on the road, but at a distance downwind. Uncertainty regarding where the peak concentration
10 occurs (on-road or at a distance from the road) in combination with the form of the exponential
11 model used to estimate the on-road concentrations (the highest concentration occurs at zero
12 distance from road) may also lead to overestimation in the number of exceedances.
13 Another source of uncertainty is the extent to which the near-road study locations used to
14 derive the on-road simulation factors represent the locations in these analyses. The on-road and
15 near-road data were collected in a few locations, most of them outside of the United States. The
16 source mixes (i.e., the vehicle fleet) in study locations may not be representative of the U.S. fleet.
17 Without detailed information characterizing the emissions patterns for the on-road study areas,
18 there was no attempt to match the air quality characterization locations to specific on-road study
19 areas, which might have improved the precision of the estimates. However, since concentration
20 ratios were selected randomly from all the near-road studies and applied to each monitor
21 individually, and since we estimated overall minimum and upper bounds using multiple
22 simulations, the analysis provides a reasonable lower and upper bound estimates of the number
23 of exceedances.
24 7.4.7 Health Benchmark
25 The choice of potential health effect benchmarks, and the use of those benchmarks to
26 assess risks, can introduce uncertainty into the risk assessment. For example, the potential health
27 effect benchmarks used were based on studies where volunteers were exposed to NO2 for
28 varying lengths of time. Typically, the NO2 exposure durations were between 30 minutes and 2
29 hours. This introduces some uncertainty into the characterization of risk, which compared the
30 potential health effect benchmarks to estimates of exposure over a 1-hour time period. Use of a
August 2008 Draft 96
-------
4
5
6
7
8
9
10
11
12
13
1-hour averaging time could over- or under-estimate risks. In addition, the human exposure
studies evaluated airways responsiveness in mild asthmatics. For ethical reasons, more severely
affected asthmatics and asthmatic children were not included in these studies. Severe asthmatics
and/or asthmatic children may be more susceptible than mildly asthmatic adults to the effects of
NC>2 exposure. Therefore, the potential health effect benchmarks based on these studies could
underestimate risks in populations with greater susceptibility.
Table 7-19. Summary of qualitative uncertainty analysis for the air quality and health risk
characterization.
Source
Air Quality Data
Ambient Measurement
Temporal Representation
Spatial Representation
Air Quality Adjustment
On-Road Simulation
Health Benchmarks
Type
Database quality
Interference
Vertical siting
No Extrapolation < 4m
Scale
Missing data
Years monitored
Source changes
Scale
Monitor objectives
Temporal scale
Spatial scale
Temporal scale
Decay
Spatial scale
Model used
Non US studies used
Averaging time
Susceptibility
Bias Direction
both
over
under
unknown
none
both
both
over
unknown
both
over
over
both
over
over
over
unknown
unknown
under
Notes:
Bias Direction: indicates the direction the source of uncertainty is judged to
influence either the concentration or risk estimates
August 2008 Draft
97
-------
i 8. EXPOSURE ASSESSMENT AND HEALTH RISK
2 CHARACTERIZATION
August 2008 Draft 98
-------
i 9. CHARACTERIZATION OF HEALTH RISKS USING DATA
2 FROM EPIDEMIOLOGICAL STUDIES
3
4 9.1 INTRODUCTION
5 As mentioned above in chapter 6, in response to advice received from the CASAC NC>2 Panel on
6 the 1st draft REA, we have conducted a focused quantitative risk assessment in which estimates
7 of respiratory ED visits as a function of ambient levels of NC>2 have been developed for a single
8 urban area (i.e., the Atlanta MSA). In this approach, concentration-response functions are
9 derived from NC>2 epidemiological studies and are used in conjunction with ambient air quality
10 data representing alternative air quality scenarios and baseline incidence data to estimate the
11 impact of ambient levels of NC>2 on ED visits associated with these air quality scenarios. The
12 purpose for the current risk assessment is to present an illustrative case study that provides
13 information on the magnitude and potential changes in NCVrelated public health impacts
14 associated with recent air quality and alternative air quality scenarios simulating attainment of
15 the current and alternative NC>2 standards. Chapters 4 and 5 of this document provide additional
16 qualitative assessment of the epidemiological evidence most relevant to characterizing NO2-
17 related health effects in the United States including respiratory-related ED visits as well as other
18 health endpoints. As described in chapter 1, the Agency's views on policy options addressing
19 the adequacy of the current standard and alternative standards that takes into consideration both
20 the final results of the risk assessment discussed in this chapter, as well as the air quality and
21 exposure assessments presented in chapters 7 and 8, and the scientific evidence evaluated in the
22 ISA will be presented in the next step of the NAAQS-review process in an ANPR published in
23 the Federal Register.
24 Previous reviews of the NC>2 primary NAAQS, completed in 1985 and 1996, did not
25 include quantitative health risk assessments. Thus, the risk assessment described in this
26 document builds upon the methodology and lessons learned from the risk assessment work
27 conducted for the recently concluded PM and 63 NAAQS reviews (Abt Associates, 2005; Abt
28 Associates, 2007). Many of the same methodological issues are present in conducting a risk
29 assessment for each of these criteria air pollutants where epidemiological studies provided the
30 basis for the concentration-response relationships used in the quantitative risk assessment.
August 2008 Draft 99
-------
1 The NO2 health risk assessment described in this chapter estimates the incidence of
2 respiratory-related ED visits associated with short-term exposures to NO2 under recent ("as is")
3 air quality levels, upon just meeting the current NO2 standard of 0.053 ppm annual average, and
4 upon just meeting several potential alternative NO2 primary NAAQS in the Atlanta MSA.14 As
5 discussed in more detail in chapter 6 above, staff has elected to evaluate daily maximum 1-h
6 standard levels of 0.05, 0.10, 0.15, and 0.20 ppm using both 98th and 99th percentile forms and
7 averaged over a thee-year period.15 The risk assessment is intended as a tool that, together with
8 other information on this health endpoint and other health effects evaluated in the final ISA and
9 discussed elsewhere in this document, can aid the Administrator in judging whether the current
10 primary standard protects public health with an adequate margin of safety, or whether revisions
11 to the standard are appropriate.
12 Section 9.2 describes the general approach used to conduct the risk assessment for ED
13 visits. Sections 9.3, 9.4, and 9.5 discuss in more detail the three types of inputs required to
14 conduct the assessment. Section 9.6 presents a discussion of uncertainties and variability and
15 section 9.7 presents a summary of results from the assessment and key observations.
16 9.2 GENERAL APPROACH
17 The general approach used for the NO2-related ED risk assessment is dictated by the fact
18 that it is based on concentration-response functions which have been estimated in
19 epidemiological studies evaluated in the final ISA. Since these studies estimate concentration-
20 response functions using ambient air quality data from fixed-site, population-oriented monitors,
21 the appropriate application of these functions in a risk assessment similarly requires the use of
22 ambient air quality data at fixed-site, population-oriented monitors. In order to estimate the
23 incidence of respiratory-related ED visits associated with recent air quality conditions in a set of
24 counties attributable to ambient NO2 exposures, as well as the change in incidence of this health
25 effect in that set of counties corresponding to a given simulated change in NO2 levels
26 representing just meeting the current or alternative 1-h daily maximum NO2 standards, the
27 following thee elements are required:
14 The current NO2 standard refers to a two-year period and requires that the annual average NO2 level be less than
or equal to 0.053 ppm in each of the two years.
15 As an example, for the alternative standards using the 98th percentile form, the standard is met when the average
of the annual 98th percentile daily maximum 1 -hour concentrations for a 3 -year period is at or below the specified
standard level.
August 2008 Draft 100
-------
1 • Air quality information including: (1) "as is" air quality data for NC>2 from
2 ambient monitors in the assessment location, and (2) "as is" concentrations adjusted
3 to reflect patterns of air quality estimated to occur under a simulation where the
4 area's air quality is adjusted to just meet the specified standard. (These air quality
5 inputs are discussed in more detail in section 6.2 of this document).
6
7 • Concentration-response functions which provide an estimate of the relationship
8 between the health endpoint of interest and ambient NC>2 concentrations.
9
10
11 • Baseline health effects incidence. The baseline incidence of the health effect in
12 the assessment location in the target year is the incidence corresponding to "as is"
13 NC>2 levels in that location in that year.
14
15 Figure 9-1 provides a broad schematic depicting the role of these components in the NC>2
16 risk assessment. Each of the key components (i.e., air quality information, estimated
17 concentration-response functions, and baseline incidence) is discussed below, highlighting those
18 points at which judgments have been made.
19 These inputs are combined to estimate health effect incidence changes associated with
20 specified changes in NC>2 levels. Although some epidemiological studies have estimated linear
21 or logistic concentration-response functions, by far the most common form, and the form
22 relevant for the epidemiological study used in the current risk assessment is the exponential (or
23 log-linear) form:
24
25 y = Beftc, (Equation 9-1)
26
27 where x is the ambient NC>2 level, y is the incidence of the health endpoint of interest at NC>2
28 level x, ft is the coefficient of ambient NC>2 concentration (describing the extent of change iny
29 with a unit change in x), and B is the incidence at x=0, i.e., when there is no ambient NC>2. The
30 relationship between a specified ambient NC>2 level, XQ, for example, and the incidence of a given
31 health endpoint associated with that level (denoted as.yo) is then
32
33 y0=Be^. (Equation 9-2)
34
35 If we let XQ denote the baseline (upper) NC>2 level, and xj denote the lower NC>2 level, and
36 y0andyi denote the corresponding incidences of the health effect, we can derive the following
August 2008 Draft 101
-------
1
2
Air Quality
for
Recent ("As is")
NO2
Air Quality Adjustment
Procedures
Current
Proposed
Changes in
Distribution of
NO2 Air
Quality
: Concentration-Response
Human
Studies
-
Response
Relationships
Estimates of
Incidence or
and
Data
Risk
Risk Estimates:
« Recent Air
Quality
« Current
Standard
* Alternative
Standards
Figure 9-1. Major components of nitrogen dioxide health risk assessment for emergency
department visits.
4
5
6
7
August 2008 Draft
102
-------
1 relationship between the change in x, Ax= (XQ- xj), and the corresponding change in;;, Ay, from
2 equation (9-1)16:
O
4 ky = (y0-yl) = y0[\-e-flAx]. (Equation 9-3)
5
6 Alternatively, the difference in health effects incidence can be calculated indirectly using
7 relative risk. Relative risk (RR) is a measure commonly used by epidemiologists to characterize
8 the comparative health effects associated with a particular air quality comparison. The risk of
9 ED visits for respiratory illness at ambient NC>2 level XQ relative to the risk of ED visits for
10 respiratory illness at ambient NC>2 level xi, for example, may be characterized by the ratio of the
11 two rates: the rate of ED visits for respiratory illness among individuals when the ambient NC>2
12 level is x0 and the rate of ED visits for respiratory illness among (otherwise identical) individuals
13 when the ambient NC>2 level is xj. This is the RR for ED visits for respiratory illness associated
14 with the difference between the two ambient NC>2 levels, XQ and xj. Given a concentration-
15 response function of the form shown in equation (9-1) and a particular difference in ambient NC>2
16 levels, Ax, the RR associated with that difference in ambient NC>2, denoted as RR-Ax, is equal to
17 e|3Ax. The difference in health effects incidence, Ay, corresponding to a given difference in
18 ambient NC>2 levels, Ax, can then be calculated based on this RR-Ax as
19
20 Ay = (y0-yi) = y0[l - (l/RR^)]. (Equation 9-4)
21
22 Equations (9-3) and (9-4) are simply alternative ways of expressing the relationship
23 between a given difference in ambient NC>2 levels, Ax > 0, and the corresponding difference in
24 health effects incidence, Ay. These health impact equations are the key equations that combine
25 air quality information, concentration-response function information, and baseline health effects
26 incidence information to estimate health risks related to changes in ambient NC>2 concentrations.
If Ax<0 -i.e., if Ax= (x!-x0)-then the relationship between Ax and Ay can be shown to be
Ay = (yl - y0) = y0 [epta - 1]. If Ax < 0, Ay will similarly be negative. However, the magnitude of Ay will be the
same whether Ax > 0 or Ax < 0 - i.e., the absolute value of Ay does not depend on which equation is used.
August 2008 Draft 103
-------
1 9.3 AIR QUALITY INFORMATION
2 As illustrated in Figure 9-1, and noted earlier, air quality information required to conduct
3 the NC>2 risk assessment includes (1) recent air quality data for NC>2 from a suitable monitor for
4 the assessment location and (2) air quality adjustment procedures to modify the recent data to
5 simulate air quality data just meeting the current annual and potential alternative 1-h daily
6 maximum standards. The approach used to adjust air quality data to simulate meeting specified
7 standards is discussed above in section 6.2.
8 In the first part of the risk assessment, we estimate the incidence of the health effect
9 associated with "as is" levels of NC>2 (or equivalently, the change in health effect incidence, Ay,
10 associated with a change in NC>2 concentrations from "as is" levels of NC>2 to 0 ppb). In the
11 second part, we estimate the incidence of the health effect associated with NC>2 concentrations
12 simulated to just meet a specified standard (i.e., the current NC>2 standard of 0.053 ppm annual
13 average as well as each of potential alternative 1-h daily maximum standards).
14 To estimate the incidence of a health effect associated with "as is" NC>2 levels in a
15 location, we need a time series of hourly "as is" NC>2 concentrations for that location. We have
16 used monitor data from the Georgia Tech monitor (monitor id =131210048), the monitor that
17 was used in Tolbert et al. (2007), the epidemiological study from which we obtained the
18 concentration-response functions (see section 9.4 below). Complete hourly data were available
19 on over 93 percent of the days - 348 days in 2005, 345 days in 2006, and 340 days in 2007.
20 Missing NC>2 concentrations were filled in, as described in section 3.5 of Appendix C.
21 Because Tolbert et al. (2007) estimated a relationship between daily respiratory-related
22 ED visits and the 3-day moving average (i.e., NC>2 levels on the same day, the previous day, and
23 the day before that) of the daily 1-h maximum NC>2 concentrations, we calculated the 3-day
24 moving average of the daily 1-h maximum NO2 concentrations at the monitor to provide the air
25 quality input to the risk assessment.
26 The calculations for the second part of the risk assessment, in which we estimated risks
27 associated with NC>2 levels simulated to just meet the current annual standard and potential
28 alternative 1-h daily maximum standards were done analogously, using the monitor-specific
29 series of adjusted daily maximum hourly concentrations rather than the monitor-specific series of
30 "as is" daily maximum hourly concentrations.
August 2008 Draft 104
-------
1 9.4 CONCENTRATION-RESPONSE FUNCTIONS
2 As indicated in Figure 9-1, another key component in the risk assessment model is the set
3 of concentration-response functions which provide estimates of the relationship between the
4 health endpoint of interest and ambient NO2 concentrations. As discussed above, the health
5 endpoint of interest for this focused quantitative risk assessment is respiratory-related ED visits.
6 As discussed in sections 4.2.2 and 4.5.2 several community epidemiological studies have been
7 conducted in the U.S. that examined the relationship between NO2 and other air pollutants and
8 increased ED visits either for all respiratory causes or for asthma-related visits. Figure 5-1 in
9 this document summarizes the single pollutant model effect estimates from these studies. As
10 discussed in section 4.5.2, staff has considered several factors in selecting the urban area and
11 epidemiological studies upon which the current risk assessment is based. First, we have judged
12 that studies conducted in the United States are preferable to those conducted outside the United
13 States given the potential for effect estimates to be impacted by factors such as the ambient
14 pollutant mix, the placement of monitors, activity patterns of the population, and characteristics
15 of the healthcare system. Second, we judged that studies of ambient NO2 are preferable to those
16 of indoor NO2 given that studies of indoor NO2 focus on exposures in locations with indoor
17 sources of NO2. These indoor sources can result in exposure patterns, NO2 levels, and co-
18 pollutants that are different from those typically associated with ambient NO2. Third, we judged
19 it appropriate to focus on studies of ED visits. When compared to studies of respiratory
20 symptoms, the public health significance of ED visits is less ambiguous for the individuals
21 affected. In addition, baseline incidence data are more readily available for these endpoints.
22 Finally, we judged it appropriate to focus on studies that evaluated NO2 health effect associations
23 using both single- and multi-pollutant models. Taking these factors into consideration, we have
24 chosen to focus on the studies by Tolbert and colleagues (2007) in Atlanta, Georgia that address
25 ED visits for respiratory causes as a case study to illustrate the magnitude and changes in
26 estimated NO2-related risks for this endpoint for various air quality scenarios.
27 Tolbert et al. (2007) estimated concentration-response functions using both single
28 pollutant models (i.e., where NO2 was the only pollutant entered into the health effects model)
29 and multi-pollutant models (i.e., where one or two co-pollutants (PMio, 63, CO) were entered
30 into the health effects model). To the extent that any of the co-pollutants present in the ambient
31 air may have contributed to the health effects attributed to NO2 in single pollutant models, risks
August 2008 Draft 105
-------
1 attributed to NO2 might be overestimated where concentration-response functions are based on
2 single pollutant models. However, if co-pollutants are highly correlated with NO2, their
3 inclusion in an NO2 health effects model can lead to misleading conclusions in identifying a
4 specific causal pollutant. When collinearity exists, inclusion of multiple pollutants in models
5 often produces unstable and statistically insignificant effect estimates for both NO2 and the co-
6 pollutants. Given that single and multi-pollutant models each have both potential advantages and
7 disadvantages, with neither type clearly preferable over the other in all cases, we report risk
8 estimates based on both single- and multi-pollutant models in the NO2 risk assessment.
9 All of the models in Tolbert et al. (2007) used a 3-day moving average of pollution levels
10 (i.e., the average of 0-, 1-, and 2-day lags), so the issue of which of several different lag
11 structures to select does not arise. The issue of how well a given lag structure captures the actual
12 relationship between the pollutant and the health effect, however, is still relevant. Models in
13 which the pollutant-related incidence on a given day depends only on same-day or previous-day
14 pollutant concentration (or some variant of those, such as a two- or thee-day average
15 concentration) necessarily assume that the longer pattern of pollutant levels preceding the
16 pollutant concentration on a given day does not affect incidence of the health effect on that day.
17 To the extent that a pollutant-related health effect on a given day is affected by pollutant
18 concentrations over a longer period of time, then these models would be mis-specified, and this
19 mis-specification would affect the predictions of daily incidence based on the model. The extent
20 to which short-term NO2 exposure studies may not capture the possible impact of long-term
21 exposures to NO2 is unknown. A number of epidemiologic studies have examined the effects of
22 long-term exposure to NO2 and observed associations with decrements in lung function and
23 partially irreversible decrements in lung function growth. The final ISA (EPA, 2008a)
24 concludes, however, that "overall, the epidemiological evidence was suggestive but not sufficient
25 to infer a causal relationship between long-term NO2 exposure and respiratory morbidity" (ISA,
26 section 3.4). Currently, there is insufficient information to adequately adjust for the potential
27 impact of longer-term exposure on respiratory ED visits associated with NO2 exposures, if any,
28 and this uncertainty should be kept in mind as one considers the results from the short-term
29 exposure NO2 risk assessment.
August 2008 Draft 106
-------
1 9.5 BASELINE HEALTH EFFECTS INCIDENCE DATA
2 As illustrated in Equation 9-1, the most common health risk model based on air pollution
3 epidemiological studies expresses the reduction in health risk (Ay) associated with a given
4 reduction in NC>2 concentrations (Ax) as a percentage of the baseline incidence (y). To
5 accurately assess the impact of changes in NC>2 air quality on health risk in a given urban area,
6 information on the baseline incidence of health effects in that location is therefore needed. For
7 this assessment, baseline incidence is the incidence under recent ("as is") air quality conditions.
8 We obtained annual estimates of the baseline incidence of respiratory ED visits in
9 Atlanta, GA via personal communication with the authors of the study conducted in the Atlanta
10 area (Tolbert, 2007). Tolbert et al. (2007) notes that there are 42 hospitals with emergency
11 departments in the 20-county Atlanta MSA. Of these, 41 were able to provide incidence data for
12 at least part of the study period (1993 - 2004). For purposes of the NC>2 risk assessment, we
13 need incidences for the years of the risk assessment (2005 - 2007). Assuming that baseline
14 incidence of respiratory ED visits does not change appreciably in the span of a few years, we
15 have used the incidence of respiratory ED visits for the most recent year (i.e., 2004) in the
16 Tolbert et al. study, which was 121,818 respiratory ED visits.17 Because this baseline incidence
17 estimate is based on 36 hospitals, rather than the total 42 hospitals in Atlanta, this will be an
18 underestimate of baseline incidence. This is a source of downward bias in our estimates of NO2-
19 related risk.
20 Average daily baseline incidences, necessary for short-term daily concentration-response
21 functions, were calculated by dividing the annual incidence by the number of days in the year for
22 which the baseline incidences were obtained. To the extent that NC>2 affects health, however,
23 actual incidence rates would be expected to be somewhat higher than average on days with high
24 NO2 concentrations; using an average daily incidence would therefore result in underestimating
25 the changes in incidence on such days. Similarly, actual incidence rates would be expected to be
26 somewhat lower than average on days with low NC>2 concentrations; using an average daily
27 incidence would, therefore, result in overestimating the changes in incidence on low NC>2 days.
28 Both effects would be expected to be small, however, and should largely cancel one another out.
17 The specific definition of "respiratory-related" emergency department visits used in Tolbert et al. (2007) included
visits with the following respiratory illnesses as the primary diagnosis (specified by ICD-9 diagnostic codes):
asthma (493, 786.07, and 786.09), COPD (491, 492, and 496), upper respiratory illness (460 - 465, 460.0, and 477),
pneumonia (480 - 486), and bronchiolitis (466.1, 466.11, and 466.19).
August 2008 Draft 107
-------
1 9.6 ADDRESSING UNCERTAINTY AND VARIABILITY
2 An important issue associated with any population health risk assessment is the
3 characterization of uncertainties and variability. Uncertainty refers to the lack of knowledge
4 regarding both the actual values of model input variables (parameter uncertainty) and the
5 physical systems or relationships (model uncertainty - e.g., the shape of the concentration-
6 response functions). In any risk assessment, uncertainty is, ideally, reduced to the maximum
7 extent possible, but significant uncertainty often remains. It can be reduced by improved
8 measurement and improved model formulation. In addition, the degree of uncertainty can be
9 characterized, sometimes quantitatively. For example, for the NCh risk assessment the statistical
10 uncertainty surrounding the estimated NC>2 coefficients in the concentration-response functions is
11 reflected in the confidence intervals provided for the risk estimates presented in this chapter and
12 in Appendix C. Additional uncertainties are discussed briefly below and in more detail in
13 Appendix C.
14 Variability refers to the heterogeneity in a population or variable of interest that is
15 inherent and cannot be reduced though further research. The current risk assessment for Atlanta
16 is based on locations-specific inputs (i.e., air quality data, baseline incidence data, and
17 concentration-response functions are for the Atlanta MSA). Variability in air quality data is
18 considered to some extent by the inclusion of thee years of data. Temporal variability is more
19 difficult to address, because the risk assessment focuses on some unspecified time in the future
20 when a given standard is just being met. To minimize the degree to which values of inputs to the
21 analysis may be different from the values of those inputs at that unspecified time: we have used
22 recent input data - for example, air quality data for the period 2005-2007 and baseline incidence
23 data for 2004. However, future changes in these inputs have not been predicted (e.g., future
24 population levels or changes in baseline incidence).
25 A number of important sources of uncertainty have been addressed qualitatively. Section
26 3.8 in Appendix C discusses in greater detail the uncertainties and variability present in the
27 health risk assessment. The following is a brief discussion of the major sources of uncertainty
28 and variability in the risk assessment and how they are dealt with or considered in the risk
29 assessment:
30 • Causality. There is uncertainty about whether the association between NC>2 and ED
31 visits actually reflects a causal relationship. Our judgment, drawing on the
August 2008 Draft 108
-------
1 conclusions in the ISA and as discussed in more detail in chapter 4, is that there is, at
2 a minimum, a likely causal relationship with either short-term NC>2 itself or with NC>2
3 serving as an indicator for itself and other components of ambient air associated with
4 combustion processes.
5 • Empirically estimated concentration-response relationships. In estimating the
6 concentration-response relationships, there are uncertainties: (1) surrounding
7 estimates of NO2 coefficients in concentration-response functions used in the
8 assessment, (2) concerning the specification of the concentration-response model
9 (including the shape of the relationships) and whether or not a population threshold or
10 non-linear relationship exists within the range of concentrations examined in the
11 studies, and (3) concerning the possible role of co-pollutants. The uncertainty
12 resulting from the statistical uncertainty associated with the estimated NO2 coefficient
13 in the concentration-response function has been characterized by confidence intervals
14 reflecting sample size. These confidence intervals do not reflect the uncertainties
15 related to the concentration-response functions, such as whether or not the model
16 used in the epidemiological study is the correct model form. With respect to
17 uncertainties about model form and whether or not a population threshold exists, the
18 available epidemiological studies neither support nor refute the existence of
19 thresholds at the population level. Concerning the possible role of co-pollutants in
20 the Tolbert et al. (2007) study, NO2 was only moderately correlated with the other
21 pollutants considered (i.e., PMio, Os) that produced the concentration-response
22 functions that have been used in the risk assessment, although it was fairly highly
23 correlated (r = 0.7) with CO. When a study, such as Tolbert et al. (2007) is conducted
24 in a single location, the problem of possible confounding is particularly difficult.
25 Single-pollutant models, which omit co-pollutants, may produce overestimates of the
26 NC>2 effect, if some of the effects are really due to one or more of the other pollutants.
27 On the other hand, effect estimates based on a multi-pollutant model can be uncertain
28 and even result in statistically insignificant estimates where there is a true
29 relationship, if the co-pollutants included in the model are highly correlated with
30 NO2. As a result of these considerations, we report risk estimates based on both the
31 single- and multi-pollutant models from Tolbert et al. (2007). It should be noted that
32 use of a concentration-response relationship based on an epidemiological study
33 conducted in the same location for this risk assessment reduces some potential
34 uncertainties since it does not involve extrapolation of the relationship across
35 different geographic areas with different population characteristics, land uses, source
36 mixtures and other factors.
37 • Adequacy of ambient NO? monitors as surrogate for population exposure. The
38 Tolbert et al. (2007) study used ambient concentrations at fixed-site monitors to
39 represent ambient exposure and for several reasons this may or may not provide a
40 good representation of ambient NO2 exposure for the population. The final ISA
41 identifies the following thee components to exposure measurement error: (1) the use
42 of average population rather than individual exposure data; (2) the difference between
43 average personal ambient exposure and ambient concentrations at central monitoring
44 sites; and (3) the difference between true and measured ambient concentrations (final
45 ISA, section 1.3.2, p. 1-5). While a concentration-response function may understate
46 the effect of personal exposure to NO2 on the incidence of a health effect, it will give
August 2008 Draft 109
-------
1 an unbiased estimate of the effect of ambient concentrations on the incidence of the
2 health effect, if the ambient concentrations at monitoring stations provide an unbiased
3 estimate of the ambient concentrations to which the population is exposed. If NC>2 is
4 the causal agent, the understatement of the impact of personal exposures is not a
5 concern, since NC>2 NAAQS are expressed in terms of ambient, not personal
6 exposure, levels. However, if NC>2 is not the causal agent, and the effects are due to
7 confounding copollutants or other factors, then reducing ambient NC>2 levels might
8 not result in the estimated reductions in the health effects.
9 • Adjustment of air quality distributions to simulate just meeting the current annual
10 standard and alternative 98th and 99th percentile daily maximum 1-h standards. The
11 current annual standard and many of the alternative 1-h standards analyzed in the
12 current risk assessment requires an upward adjustment of recent ambient NO2 levels.
13 In adjusting air quality to simulate just meeting these standards, we have assumed that
14 the overall shape of the distribution of 1-h and 24-h concentrations would not change.
15 While we believe this is a reasonable assumption in the absence of evidence
16 supporting a change in the distribution, we recognize this as an important additional
17 uncertainty, especially for those scenarios where considerable upward adjustment is
18 required to simulate just meeting some of the standards.
19 • Baseline incidence. There are uncertainties related to the baseline incidence
20 including: (1) the extent to which baseline incidence varies between the year used in
21 the assessment (i.e., 2004) and some unspecified future year when air quality is
22 adjusted to simulate just meeting the current and alternative standards; (2) the extent
23 to which baseline incidence is underestimated because only 36 of the 42 emergency
24 departments provided baseline incidence for the study in 2004; (3) the use of annual
25 incidence date to develop daily baseline incidence; and (4) the extent to which
26 Atlanta area residents visited emergency departments outside of the Atlanta MSA.
27 As noted previously, the use of the available baseline incidence for 2004 results in
28 some underestimation of the risk for the Atlanta MSA since data were only available
29 from 36 of the 42 emergency departments for that year (i.e., about 14% of emergency
30 departments were not included). Concerning the use of annual baseline incidence to
31 estimate daily incidence, to the extent that NC>2 affects health, actual incidence would
32 be expected to be somewhat higher than average on days with high NO2
33 concentrations and using an average daily incidence would result in underestimating
34 the changes in incidence on such days. Similarly, actual incidence would be expected
35 to be somewhat lower on days with low NC>2 concentrations and using an average
36 daily incidence would result in overestimating the changes in incidence on such days.
37 Both of these effects would be expected to be small and should largely cancel each
38 other out. With respect to the last uncertainty, we consider this to be a relatively
39 minor uncertainty since most ED visits are likely to be made to the closest emergency
40 department available, which, for residents of the Atlanta MSA are likely to be within
41 that MSA. The baseline incidence data has not been adjusted for any future changes
42 such as aging of the population over time or possible changes in ED visits due to
43 increased in-migration of younger individuals.
August 2008 Draft 110
-------
1 9.7 RISK ESTIMATES FOR EMERGENCY DEPARTMENT VISITS
2 In this section, we present risk estimates associated with several air quality scenarios,
3 including thee recent years of air quality as represented by 2005, 2006, and 2007 monitoring
4 data. In addition, risk estimates are presented for a hypothetical scenario, where air quality from
5 2006 and 2007 is adjusted upward to simulate just meeting the current annual NC>2 standard, and
6 for scenarios where the thee year period (2005-2007) is adjusted (either up or down) to simulate
7 just meeting potential alternative 98th and 99th percentile daily maximum 1-h standards. As
8 discussed previously in chapter 5, potential alternative 1-h standards with levels set at 0.05, 0.10,
9 0.15, and 0.20 have been included in the risk assessment.
10 Throughout this section and Appendix C the uncertainty surrounding risk estimates
11 resulting from the statistical uncertainty of the NC>2 coefficients in the concentration-response
12 functions used is characterized by ninety-five percent confidence intervals around estimates of
13 incidence, incidence per 100,000 population, and percent of total incidence that is NCVrelated.
14 In some cases, the lower bound of a confidence interval falls below zero. This does not imply
15 that additional exposure to NC>2 has a beneficial effect but only that the estimated coefficient in
16 the concentration-response function was not statistically significantly different from zero. Lack
17 of statistical significance could reflect insufficient statistical power to detect a relationship that
18 exists or could reflect that no relationship exists.
19 Tables 9-1, 9-2, and 9-3 present the risk estimates for NO2-related ED visits associated
20 with recent air quality (2005, 2006, and 2007, respectively). Table 9-1 for 2005 also includes
21 risk estimates for just meeting several alternative 1-h daily maximum standards based on
22 adjusting 2005-2007 air quality data to simulate just meeting these alternative standards.
23 Similarly, Tables 9-2 and 9-3 include risk estimates associated with just meeting these same
24 alternative 1-h standards, as well as risk estimates associated with a simulation where air quality
25 is adjusted upward to represent just meeting the current 0.053 ppm annual NC>2 standard. Since
26 attainment of the current annual standard is based on the most recent two year period, risk
27 estimates for the annual standard are only included in the tables based on 2006 and 2007 air
28 quality.
29 In Table 9-1, and similarly in Tables 9-2 and 9-3, the first row of incidence estimates is
30 based on a single pollutant model (i.e., NC>2 only) and results in the largest estimates for MV
31 related respiratory ED visits. The next three rows present risk estimates based on two pollutant
August 2008 Draft 111
-------
1
2
3
Table 9-1. Estimated Incidence of Respiratory ED Visits Associated with "As Is" NO2 Concentrations and NO2 Concentrations that Just
Meet Alternative Standards in Atlanta, GA, Based on Adjusting 2005 NO2 Concentrations.*
Other
Pollutants in
Model
none
CO
03
PM10
PM10, 03
Incidence of Respiratory Emergency Department Visits Associated with "As is" N02 Concentrations and N02 Concentrations that Just Meet
Alternative Standards**
"as is"
3600
(1900-5300)
3100
(1000-5100)
1800
(-100-3700)
1300
(-700 - 3300)
700
(-1400-2800)
Atternative 98th percentile 1-hr daily maximum standards
(ppm)
0.05***
2600
(1400-3800)
2200
(700 - 3600)
1300
(-100-2600)
900
(-500 - 2300)
500
(-1000-2000)
0.1
5100
(2700 - 7400)
4300
(1500-7200)
2600
(-100-5200)
1800
(-1000-4600)
1000
(-2000 - 4000)
0.15
7500
(4100-10900)
6400
(2200-10500)
3900
(-200 - 7700)
2700
(-1600-6800)
1600
(-3000 - 5900)
0.2
9900
(5400-14300)
8500
(2900-13800)
5100
(-200-10200)
3600
(-2100-9000)
2100
(-4000 - 7800)
Alternative 99th percentile 1-hr daily maximum standards
(ppm)
0.05
2400
(1300-3500)
2000
(700 - 3400)
1200
(-100-2500)
800
(-500 - 2200)
500
(-900-1900)
0.1
4700
(2500 - 6900)
4000
(1400-6700)
2400
(-100-4900)
1700
(-1000-4300)
1000
(-1800-3700)
0.15
7000
(3800-10200)
6000
(2000 - 9800)
3600
(-200 - 7200)
2500
(-1500-6400)
1500
(-2800 - 5500)
0.2
9300
(5000-13300)
7900
(2700-12900)
4800
(-200 - 9500)
3400
(-1900-8400)
1900
(-3700 - 7300)
*Estimated incidences of respiratory emergency department visits are based on the concentration-response functions estimated in Tolbert et al. (2007) [results
corresponding to Figure 2 in Tolbert et al. (2007) were obtained via personal communication with P. Tolbert]. All models use a 3-day moving average of the
daily 1-hr, maximum N02 concentration and apply to all ages.
^Incidence was quantified down to 0 ppb. Incidences are rounded to the nearest 100.
***Alternative 1-hr daily maximum standards are characterized by a concentration of m ppm and an nth percentile, requiring that the average of the 3 annual
nth percentile 1-hr daily maxima over a 3-year period be at or below m ppm.
Note: Numbers in parentheses are 95% confidence intervals based on statistical uncertainty surrounding the N02 coefficient.
August 2008 - Draft
112
-------
1
2
3
Table 9-2. Estimated Incidence of Respiratory ED Visits Associated with "As Is" NO2 Concentrations and NO2 Concentrations that Just
Meet Alternative Standards in Atlanta, GA, Based on Adjusting 2006 NO2 Concentrations.*
Other
Pollutants In
Model
none
CO
03
PM10
PMio, 03
Incidence of Respiratory Emergency Department Visits Associated with "As is" N02 Concentrations and N02 Concentrations that Just Meet the Current and
Alternative Standards**
"as is"
3800
(2000 - 5500)
3200
(1100-5300)
1900
(-100-3900)
1300
(-800 - 3400)
800
(-1500-2900)
current annual
standard
10900
(5900-15700)
9400
(3200-15200)
5600
(-300-11200)
4000
(-2300 - 9900)
2300
(-4400 - 8600)
Alternative 98th percentile 1-hr daily maximum standards
(ppm)
0.05***
2700
(1400-3900)
2300
(800 - 3800)
1400
(-100-2700)
900
(-500 - 2400)
500
(-1000-2100)
0.1
5300
(2800 - 7700)
4500
(1500-7400)
2700
(-100-5400)
1900
(-1100-4800)
1100
(-2100-4100)
0.15
7800
(4200-11300)
6700
(2300-11000)
4000
(-200 - 8000)
2800
(-1600-7100)
1600
(-3100-6200)
0.2
10300
(5600-14800)
8800
(3000-14400)
5300
(-200-10600)
3700
(-2200 - 9400)
2200
(-4200-8100)
Alternative 99th percentile 1-hr daily maximum standards
(ppm)
0.05
2500
(1300-3600)
2100
(700 - 3500)
1300
(-100-2600)
900
(-500 - 2300)
500
(-1000-1900)
0.1
4900
(2600 - 7200)
4200
(1400-6900)
2500
(-100-5100)
1800
(-1000-4500)
1000
(-1900-3900)
0.15
7300
(3900-10600)
6200
(2100-10200)
3700
(-200 - 7500)
2600
(-1500-6600)
1500
(-2900 - 5700)
0.2
9600
(5200-13900)
8200
(2800-13400)
4900
(-200 - 9900)
3500
(-2000 - 8700)
2000
(-3900 - 7600)
*Estimated incidences of respiratory emergency department visits are based on the concentration-response functions estimated in Tolbert et al. (2007) [results corresponding
to Figure 2 in Tolbert et al. (2007) were obtained via personal communication with P. Tolbert]. All models use a 3-day moving average of the daily 1 -hr. maximum N02
concentration and apply to all ages.
"Incidence was quantified down to 0 ppb. Incidences are rounded to the nearest 100.
***Alternative 1 -hr daily maximum standards are characterized by a concentration of m ppm and an nth percentile, requiring that the average of the 3 annual nth percentile 1 -
hr daily maxima over a 3-year period be at or below m ppm.
Note: Numbers in parentheses are 95% confidence intervals based on statistical uncertainty surrounding the N02 coefficient.
August 2008 - Draft
113
-------
1
2
3
Table 9-3. Estimated Incidence of Respiratory ED Visits Associated with "As Is" NO2 Concentrations and NO2 Concentrations that Just
Meet Alternative Standards in Atlanta, GA, Based on Adjusting 2007 NO2 Concentrations.*
Other
Pollutants In
Model
none
CO
03
PM10
PM10, 03
Incidence of Respiratory Emergency Department Visits Associated with "As Is" N02 Concentrations and N02 Concentrations that Just Meet the Current and
Alternative Standards**
"as is"
3400
(1800-4900)
2900
(1000-4800)
1700
(-100-3500)
1200
(-700 - 3000)
700
(-1300-2600)
current annual
standard
9800
(5300-14200)
8400
(2900-13700)
5100
(-200-10100)
3600
(-2100-8900)
2100
(-4000 - 7800)
Alternative 98th percentile 1-hr daily maximum standards
(ppm)
0.05***
2400
(1300-3500)
2000
(700 - 3400)
1200
(-100-2500)
800
(-500 - 2200)
500
(-900-1900)
0.1
4700
(2500 - 6900)
4000
(1300-6700)
2400
(-100-4900)
1700
(-1000-4300)
1000
(-1800-3700)
0.15
7000
(3800-10200)
6000
(2000 - 9900)
3600
(-200 - 7200)
2500
(-1500-6400)
1500
(-2800 - 5500)
0.2
9300
(5000-13400)
7900
(2700-12900)
4800
(-200 - 9500)
3400
(-1900-8400)
1900
(-3700 - 7300)
Alternative 99th percentile 1-hr daily maximum standards
(ppm)
0.05
2200
(1200-3300)
1900
(600 - 3200)
1100
(-100-2300)
800
(-400 - 2000)
500
(-900-1700)
0.1
4400
(2400 - 6400)
3800
(1300-6200)
2200
(-100-4500)
1600
(-900 - 4000)
900
(-1700-3500)
0.15
6500
(3500 - 9500)
5600
(1900-9200)
3300
(-200 - 6700)
2400
(-1400-5900)
1400
(-2600-5100)
0.2
8600
(4700-12500)
7400
(2500-12100)
4400
(-200 - 8900)
3100
(-1800-7800)
1800
(-3400 - 6800)
*Estimated incidences of respiratory emergency department visits are based on the concentration-response functions estimated in Tolbert et al. (2007) [results corresponding
to Figure 2 in Tolbert et al. (2007) were obtained via personal communication with P. Tolbert]. All models use a 3-day moving average of the daily 1 -hr. maximum N02
concentration and apply to all ages.
"Incidence was quantified down to 0 ppb. Incidences are rounded to the nearest 100.
"""Alternative 1 -hr daily maximum standards are characterized by a concentration of m ppm and an nth percentile, requiring that the average of the 3 annual nth percentile 1 -
hr daily maxima over a 3-year period be at or below m ppm.
Note: Numbers in parentheses are 95% confidence intervals based on statistical uncertainty surrounding the N02 coefficient.
August 2008 - Draft
114
-------
1 models (i.e., NC>2 + CO, NC>2 + 63, NO2 + PMio). The last row presents risk estimates based on
2 a three pollutant model (i.e., NC>2 + PMio + 63). As noted above in this chapter, effect estimates
3 based on a multi-pollutant model can be uncertain and even result in statistically insignificant
4 estimates where there is a true relationship, if the co-pollutants included in the model are highly
5 correlated with NC>2. The negative lower bounds of the confidence intervals for many of the risk
6 estimates based on multi-pollutant models is the result of this problem and staff do not view this
7 as suggesting any health beneficial effect of increasing NC>2 exposure levels.
8 Tables 4-4, 4-5, and 4-6 in Appendix C present these same risk estimates expressed in
9 terms of incidence per 100,000 general population in the Atlanta MSA based on recent air
10 quality and simulating just meeting alternative standards based on 2005, 2006, an 2007 air
11 quality data. Finally, Tables 4-7, 4-8, and 4-9 in Appendix C present these same risk estimates
12 in terms of percent of total incidence of ED visits for the Atlanta MSA based on the same three
13 years of air quality data.
14 Key Observations
15 Presented below are key observations resulting from the respiratory-related ED visits risk
16 assessment:
17 • Respiratory-related ED visits estimated to result from exposures to NC>2 were
18 estimated for a single urban area (i.e., Atlanta) for several recent years of air
19 quality (2005-2007) and for air quality adjusted to simulate just meeting the
20 current annual NO2 standard and several alternative 1-hour daily maximum NO2
21 standards. While we would expect some differences in estimated NCVrelated ED
22 respiratory visits across different locations due to differences in populations, land
23 use patterns, access to medical facilities, co-pollutants and other factors affecting
24 exposure and the concentration-response relationships, we believe that the risk
25 estimates do provide a useful perspective on the likely overall magnitude and
26 pattern of ED visits associated with various NC>2 air quality scenarios in urban
27 areas within the U.S.
28 • The largest risk estimates were associated with single-pollutant NC>2
29 concentration-response functions based on the effect estimates reported in Tolbert
30 et al. (2007). Risk estimates based on various co-pollutant models with Os, CO,
31 and PMio resulted in significant reduction in the risk estimates, often by a factor
32 of two or greater and resulted in much wider confidence intervals.
33 • The only standards that resulted in a reduction in risk estimates from the baseline
34 of recent air quality for the three year period examined were the 98th and 99th
35 percentile 1-hour daily maximum standards set at the level of 0.05 ppm.
36 • The impact of changing the level of the alternative 1-hour daily maximum
37 standards is substantially greater than the impact of changing from a 98th to a
August 2008 Draft 115
-------
1 99th percentile standard. For example, changing from a 98th percentile 1-hour
2 daily maximum standard based on 0.05 ppm to one based on 0.1 ppm reduces the
3 estimated incidence of respiratory-related ED visits in Atlanta by about 49 percent
4 in 2007 (from 4700 to 2400); however, changing from a 98th percentile 1-hour
5 daily maximum standard based on 0.05 ppm to a 99th percentile 1-hour daily
6 maximum standard based on 0.05 reduces the incidence in 2007 by only about 8
7 percent (from 2400 to 2200).
8 • The overall pattern of risk estimates is similar across the three years examined.
9 For the three years examined, there was not significant year-to-year variability in
10 the risk estimates.
11 • Important uncertainties and limitations associated with the risk assessment which
12 were discussed above in section 9.6 and which should be kept in mind as one
13 considers the quantitative risk estimates include:
14 - uncertainty about the extent to which the associations between NC>2 and ED
15 visits for respiratory causes actually reflect causal relationships;
16 - statistical uncertainty due to sampling error which is characterized in the
17 assessment;
18 - uncertainties associated with the air quality adjustment procedure that was
19 used to simulate just meeting the current annual and several alternative 1-h
20 daily maximum standards;
21 -uncertainties associated with the estimated baseline incidence for ED
22 respiratory visits;
23 - uncertainties related to how changes in population, activity patterns, air
24 quality, and other factors over time might impact the risk estimates;
25 - there is uncertainty about the extent to which the risk estimates presented
26 for the Atlanta urban area are representative of other urban locations in the
27 U.S..
August 2008 Draft 116
-------
i 10. References
2
3 AHS. (2003a). American Housing Survey for 2003. Available at:
4 kH|)l//ww^^
5
6 AHS. (2003b). Source and Accuracy Statement for the 2003 AHS-N Data Chart. Available at:
7 http;//www,,,cCTMS.,gQY^^
9 Barck, C, Sandstrom T, Lundahl J, Hallden G, Svartengren M, Strand V, Rak S, Bylin G.
10 (2002). Ambient level of NC>2 augments the inflammatory response to inhaled allergen in
11 asthmatics. RespirMed. 96:901'-917.
12
13 Barck, C, Lundahl J, Hallden G, Bylin G. (2005). Brief exposures to NC>2 augment the allergic
14 inflammation in asthmatics. Environ Res. 97:58-66.
15
16 Beckerman, B, Jerrett M, Brook JR, Verma DK, Arain MA, Finkelstein MM. (2008).
17 Correlation of nitrogen dioxide with other traffic pollutants near a major expressway.
18 AtmosEnviron. 42:275-290.
19
20 Bell, S and Ashenden TW. (1997). Spatial and temporal variation in nitrogen dioxide pollution
21 adjacent to rural roads. Water Air SoilPollut. 95:87-98.
22
23 Brauer, M, Hoek G, Smit HA, De Jongste JC, Gerritsen J, Postma DS, Kerkhof M, Brunekreef B.
24 (2007). Air pollution and development of asthma, allergy and infections in a birth cohort.
25 Eur. Respir. J. 29:879-888.
26
27 Bylin, G, Lindvall T, Rehn T, Sundin B. (1985) Effects of short-term exposure to ambient
28 nitrogen dioxide concentrations on human bronchial reactivity and lung function. Eur. J.
29 Respir. Dis. 66:205-217.
30
31 Cape, JN, Tang YS, van Dijk N, Love L, Sutton MA, Palmer SCF. (2004). Concentrations of
32 ammonia and nitrogen dioxide at roadside verges, and their contribution to nitrogen
33 deposition. Environ Pollut. 132:469-478.
34
35 CARB. (2001). Indoor air quality: residential cooking exposures. Final report. California Air
36 Resources Board, Sacramento, California. Available at:
37 ktt]ii//wwwj||^^
38
39 Chan, AT and Chung MW. (2003). Indoor-outdoor air quality relationships in vehicle: effect of
40 driving environment and ventilation modes. Atmos Environ. 37:3795-3808.
August 2008 Draft 117
-------
1 Chauhan, AJ, Inskip HM, Linaker CH, Smith S, Schreiber J, Johnston SL, Holgate ST (2003).
2 Personal exposure to nitrogen dioxide (NO2) and the severity of virus-induced asthma in
3 children. Lancet 361:1939-1944.
4
5 Clougherty, JE, Levy JI, Kubzansky LD, Ryan PB, Suglia SF, Canner MJ, and Wright RJ.
6 (2007). Synergistic effects of traffic-related air pollution and exposure to violence on
7 urban asthma etiology. Environ. Health Perspect. 115:1140-1146.
8
9 Delfino, RJ, Gone H, Linn WS, Pellizzari ED, Hu Y. (2003). Asthma symptoms in Hispanic
10 children and daily ambient exposures to toxic and criteria air pollutants. Environ. Health
11 Perspect. 111:647-656.
12
13 Delfino, RJ, Zeiger RS, Seltzer JM, Street DH, McLaren CE. (2002). Association of asthma
14 symptoms with peak particulate air pollution and effect modification by anti-
15 inflammatory medication use. Environ. Health Perspect. 110:A607-A617.
16
17 Devalia, JL, Rusznak C, Herdman MJ, Trigg CJ, Tarraf H, Davies RJ. (1994) Effect of nitrogen
18 dioxide and sulphur dioxide on airway response of mild asthmatic patients to allergen
19 inhalation. Lancet 344:1668-1671.
20
21 Dietert, RR, Etzel RA, Chen D, Halonen M, Holladay SD, Jarabek AM, Landreth K, Peden DB,
22 Pinkerton K, Smialowicz RJ, Zoetis T. (2000). Workshop to identify critical window of
23 exposure for children's health: immune and respiratory systems work group summary.
24 Environ Health Perspect. Suppl. 108(3):483-490.
25
26 EPA. (1982). Air Quality Criteria for Oxides of Nitrogen. National Center for Environmental
27 Assessment, Research Triangle Park, NC. EPA-600/8-82-026. Available at
28 hH|)i/M|Mb^^
29
30 EPA. (1993). Air Quality Criteria Document for the Oxides of Nitrogen. National Center for
31 Environmental Assessment, Research Triangle Park, NC. EPA-600/8-91/049F.
32 Available at: htt|r//cFjMib±^
33
34 EPA. (1995). Review of the National Ambient Air Quality Standards for Nitrogen Dioxide:
35 Assessment of Scientific and Technical Information. OAQPS Staff Paper. Office of Air
36 Quality Planning and Standards, Research Triangle Park, NC. EPA-452/R-95-005.
37 September 1995.
38
39 EPA. (2002). Consolidated Human Activities Database (CHAD) Users Guide. Database and
40 documentation available at: http://www.epa.gov/chadnet 1 /.
41
August 2008 Draft 118
-------
1 EPA. (2004). AERMOD: Description of Model Formulation. Office of Air Quality Planning
2 and Standards, Research Triangle Park, NC. EPA-454/R-03-004. Available at:
3 http://www.epa.gov/scramOO l/7thconf/aermod/aermod_mfd.pdf.
4
5 EPA. (2006a). Total Risk Integrated Methodology (TRIM) - Air Pollutants Exposure Model
6 Documentation (TRIM.Expo / APEX, Version 4) Volume I: User's Guide. Office of Air
7 Quality Planning and Standards, Research Triangle Park, NC. June 2006. Available at:
8 http ://www. epa. gov/ttn/fera/human_apex.html.
9
10 EPA. (2006b). Total Risk Integrated Methodology (TRIM) - Air Pollutants Exposure Model
11 Documentation (TRIM.Expo / APEX, Version 4) Volume II: Technical Support
12 Document. Office of Air Quality Planning and Standards, Research Triangle Park, NC.
13 Available at: http://www.epa.gov/ttn/fera/human_apex.html.
14
15 EPA. (2007a). Plan for Review of the Primary National Ambient Air Quality Standard for
16 Nitrogen Dioxide. Available at:
17 http://www.epa.gov/ttn/naaq s/standards/nox/s_nox_cr_pd.html.
18
19 EPA. (2007b). Nitrogen Dioxide Health Assessment Plan: Scope and Methods for Exposure
20 and Risk Assessment. Office of Air Quality Planning and Standards, Research Triangle
21 Park, NC. Available at: http://www.epa.gov/ttn/naaqs/standards/nox/s_nox_cr_pd.html.
22
23 EPA. (2007c). US EPA Air Quality System (AQS). Download Detailed AQS Data. Available
24 at: httpi//wawjga..jQ^^^
25
26 EPA(2007d). Field Guide to Air Quality Data (vl.0.0). February 28, 2007. Available at:
27 hltp;//_w_w_w_^
28
29 EPA. (2007e). 2002 National Emissions Inventory Data & Documentation. Office of Air
30 Quality Planning and Standards, Research Triangle Park, NC. Available at:
31 hltp;//wwwji^
32
33 EPA. (2007f). Clean Air Markets - Data and Maps. Office of Atmospheric Programs,
34 Washington, DC. Emissions Prepackaged Data Sets. Available at:
35 http://camddataandmaps.epa.gov/gdm/index.cfm?fuseaction=emissions. wizard.
36
37 EPA. (2007g). Ozone Population Exposure Analysis for Selected Urban Areas (July 2007).
38 Office of Air Quality Planning and Standards, Research Triangle Park, NC. EPA-452/R-
39 07-010. Available at: http://epa.gov/ttn/naaqs/standards/ozone/s_o3_cr_td.html.
40
41 EPA. (2007h). Review of the National Ambient Air Quality Standards for Ozone: Assessment
42 of Scientific and Technical Information. OAQPS Staff paper. Office of Air Quality
August 2008 Draft 119
-------
1 Planning and Standards, Research Triangle Park, NC. EPA-452/R-07-007a. Available
2 at: http://epa.gOv/ttn/naaqs/standards/ozone/s o3 cr sp.html.
4 EPA. (2008a). ISA for Oxides of Nitrogen - Health Criteria. National Center for
5 Environmental Assessment, Research Triangle Park, NC. Available at:
6 http ://cfpub. epa.gov/ncea/cfm/recordisplay. cfm?deid= 194645.
7
8 EPA (2008b). Risk and Exposure Assessment to Support the Review of the NO2 Primary
9 National Ambient Air Quality Standard: First Draft. Office of Air Quality Planning and
10 Standards, Research Triangle Park, NC.
11
12 Finlayson-Pitts, BJ and Pitts IN. (2000). Chemistry of the Upper and Lower Atmosphere.
13 Academic Press, San Diego CA. Page 17.
14
15 Folinsbee, LJ. (1992). Does nitrogen dioxide exposure increase airways responsiveness?
16 ToxicollndHealth. 8:273-283.
17
18 Gauderman, W J, Avol E, Gilliland F, Vora H, Thomas D, Berhane K, McConnell R, Kuenzli N,
19 Lurmann F, Rappaport E, Margolis H, Bates D, Peters J. (2004). The effect of air
20 pollution on lung development from 10 to 18 years of age. N. Engl. J. Med. 351:1057-
21 1067.
22
23 Gauderman, WJ, Avol E, Lurmann F, Kuenzli N, Gilliland F, Peters J, McConnell R. (2005).
24 Childhood asthma and exposure to traffic and nitrogen dioxide. Epidemiology 16:737-
25 743.
26
27 Ghenu, A, Rosant J-M, Sini J-F. (2007). Dispersion of pollutants and estimation of emissions in
28 a street canyon in Rouen, France. Environmental Modeling & Software. 23:314-321.
29
30 Gilbert, NL, Woodhouse S, Stieb DM, Brook JR. (2003). Ambient nitrogen dioxide and
31 distance from a major highway. Sci Total Environ. 312:43-46.
32
33 Graham, SE and McCurdy T. (2004). Developing meaningful cohorts for human exposure
34 models. J Expos Anal Environ Epidemiol. 14(1)23-43.
35
36 Heeb, NV, Saxer CJ, Forss AM, Bruhlmann S. (2008). Trends of NO-, NO2-, and NH3-
37 emissions from gasoline-fueled Euro-3- to Euro-4-passenger cars. Atmos Environ.
38 42:2543-2554
39
August 2008 Draft 120
-------
1 Ito, K. (2007). Association between doarse particles and asthma emergency department (ED)
2 visits in New York City. Presented at: American Thoracic Society international
3 conference; San Francisco, CA.
4
5 Jaffe, DH, Singer ME, Rimm AA. (2003). Air pollution and emergency department visits for
6 asthma among Ohio Medicaid recipients, 1991-1996. Environ. Res. 91:21-28.
7
8 Jenkins, HS, Devalia JL, Mister RL, Bevan AM, Rusznak C, Davies RJ. (1999). The effect of
9 exposure to ozone and nitrogen dioxide on the airway response of atopic asthmatics to
10 inhaled allergen: dose- and time-dependent effects. Am. J. Respir. Crit. CareMed.
11 160:33-39.
12
13 Johnson, T, Mihlan G, LaPointe J, Fletcher K, Capel J. (1999). Estimation Of Carbon
14 Monoxide Exposures and Associated Carboxyhemoglobin Levels In Denver Residents
15 Using pNEM/CO (version 2.0). Prepared by ICF Kaiser Consulting Group for the U.S.
16 Environmental Protection Agency under Contract No. 68-D6-0064, March 1999.
17
18 Jorres, R, Magnussen H. (1990). Airways response of asthmatics after a 30 min exposure, at
19 resting ventilation, to 0.25 ppm NO2 or 0.5 ppm SO2. Eur. Respir. J. 3:132-137.
20
21 Jorres, R, Magnussen H. (1991). Effect of 0.25 ppm nitrogen dioxide on the airway response to
22 methacholine in asymptomatic asthmatic patients. Lung 169:77-85.
23
24 Jorres, R, Nowak D, Grimminger F, Seeger W, Oldigs M, Magnussen H. (1995). The effect of 1
25 ppm nitrogen dioxide on bronchoalveolar lavage cells and inflammatory mediators in
26 normal and asthmatic subjects. Eur Respir J. 8:416-424.
27
28 Kim, S-Y, O'Neill M, Lee J-T, Cho Y, Kim J, and Kim H. (2007). Air pollution, socioeconomic
29 position, and emergency hospital visits for asthma in Seoul, Korea. Int. Arch. Occup.
30 Environ. Health 80:701-710.
31
32 Koontz, MD, Mehegan LL, and Nagda NL. (1992). Distribution and Use of Cooking Appliances
33 That Can Affect Indoor Air Quality, Report No. GRI-93/0013. Gas Research Institute,
34 Chicago, IL.
35
36 Langstaff, JE. (2007). OAQPS Staff Memorandum to Ozone NAAQS Review Docket (OAR-
37 2005-0172). Subject: Analysis of Uncertainty in Ozone Population Exposure Modeling.
38 [January 31,2007]. Available at:
39 httpj//www,,e^^
40
August 2008 Draft 121
-------
1 Linn, WS, Shamoo DA, Anderson KR, Peng RC, Avol EL, Hackney JD, Gong, H Jr. (1996).
2 Short-term air pollution exposures and responses in Los Angeles area schoolchildren. J.
3 Exposure Anal. Environ. Epidemiol. 6:449-472.
4
5 Linn, WS, Szlachcic Y, Gong H, Kinney PL, Berhane KT. (2000). Air pollution and daily
6 hospital admissions in metropolitan Los Angeles. Environ. Health Perspect. 108:427-
7 434.
8
9 McCurdy, TR. (1994). Analysis of high 1 hour NC>2 values and associated annual averages
10 using 1988-1992 data. Office of Air Quality Planning and Standards, Research Triangle
11 Park, NC. Available in Docket A-93-06.
12
13 McCurdy, T, Glen G, Smith L, Lakkadi Y. (2000). The National Exposure Research
14 Laboratory's Consolidated Human Activity Database. JExp Anal Environ Epidemiol.
15 10: 566-578.
16
17 McCurdy T and Graham SE. (2003). Using human activity data in exposure models: analysis of
18 discriminating factors. J Expos Anal Environ Epidemiol. 13:294-317.
19
20 Mohsenin, V. (1987) Airway responses to nitrogen dioxide in asthmatic subjects. J. Toxicol.
21 Environ. Health 22:371-380.
22
23 Mortimer, KM, Neas LM, Dockery DW, Redline S, Tager IB. (2002). The effect of air
24 pollution on inner-city children with asthma. EurRespirJ. 19:699-705.
25
26 National Cooperative Highway Research Program (NCHRP) (1997). Recommended Procedure
27 for Long-Range Transportation Planning and Sketch Planning, NCHRP Report 387,
28 National Academy Press. 151 pp., ISBN No: 0-309-060-58-3.
29
30 National Heart, Lung, and Blood Institute (NHLBI). (2007) Expert panel report 3: guidelines for
31 the diagnosis and management of asthma. Bethesda, MD: U.S. Department of Health and
32 Human Services, National Institutes of Health, National Asthma Education and
33 Prevention Program; Nffl Publication No. 07-4051.
34
35 New York Department of Health. (2006). A study of ambient air contaminants and asthma in
36 New York City, Final Report Part B: Air contaminants and emergency department visits
37 for asthma in the Bronx and Manhattan. Prepared for: The U.S. Department of Health
38 and Human Services, Agency for Toxic Substance and Disease Registry.
39
40 Oftedal, B, Brunekreef B, Nystad W, Madsen C, Walker SE, Nafstad P. (2008). Residential outdoor
41 air pollution and lung function in schoolchildren. Epidemiology 19:129-137.
August 2008 Draft 122
-------
1
2 Ostro, B, Lipsett M, Mann J, Braxton-Owens H, White M. (2001). Air pollution and
3 exacerbation of asthma in African-American children in Los Angeles. Epidemiology
4 12:200-208.
5
6 Peel, JL, Tolbert PE, Klein M, Metzger KB, Flanders WD, Knox T, Mulholland JA, Ryan PB,
7 Frumkin H. (2005). Ambient air pollution and respiratory emergency department visits.
8 Epidemiology. 16:164-174.
10 Persily, A, Gorfain J, Brunner G. (2005). Ventilation design and performance in U.S. office
11 buildings. ASHRAE Journal. April 2005, 30-35.
12
13 Pilotto, LS, Douglas RM, Attewell RG, Wilson SR (1997) Respiratory effects associated with
14 indoor nitrogen dioxide exposure in children. Int. J. Epidemiol. 26: 788-796.
15
16 Pilotto, LS, Nitschke M, Smith BJ, Pisaniello D, Ruffm RE, McElroy HJ, Martin J, Killer JE.
17 (2004). Randomized controlled trial of unflued gas heater replacement on respiratory
18 health of asthmatic schoolchildren. IntJEpidemiol. 33:208-214.
19
20 Pleijel, H, Karlsson GP, Gerdin EB. (2004). On the logarithmic relationship between NO2
21 concentration and the distance from a highroad. Sci Total Environ. 332:261-264.
22
23 Rodes, C, Sheldon L, Whitaker D, Clayton A, Fitzgerald K, Flanagan J, DiGenova F, Hering S,
24 Frazier C. (1998). Measuring Concentrations of Selected Air Pollutants Inside
25 California Vehicles. California Environmental Protection Agency, Air Resources Board.
26 Final Report, December 1998.
27
28 Rodes, CE and Holland DM. (1981). Variations of NO, NO2 and Os concentrations downwind
29 of a Los Angeles freeway. Atmos Environ. 15:243-250.
30
31 Roger, LJ, Horstman DH, McDonnell W, Kehrl H, Ives PJ, Seal E, Chapman R, Massaro E.
32 (1990). Pulmonary function, airway responsiveness, and respiratory symptoms in
33 asthmatics following exercise inNO2. Toxicol. Ind. Health 6:155-171.
34
35 Romieu, I, Ramirez-Aguilar M., Sienra-Monge JJ, Moreno-Macias, H, Del Rio-Navarro BE,
36 David G, Marzec J, Hernandez-Avila M, London S. (2006). GSTM1 and GSTP1 and
37 respiratory health in asthmatic children exposed to ozone. Eur. Respir. J. 28:953-959.
38
August 2008 Draft 123
-------
1 Rojas-Martinez, R, Perez-Padilla R, Olaiz-Fernandez G, Mendoza-Alvarado L, Moreno-Macias
2 H, Fortoul T, McDonnell W, Loomis D, Romieu I. (2007a) Lung function growth in
3 children with long-term exposure to air pollutants in Mexico City. Am. J. Respir. Crit.
4 CareMed. 176:377-384.
5
6 Rojas-Martinez, R, Perez-Padilla R, Olaiz-Fernandez G, Mendoza-Alvarado L, Moreno-Macias
7 H, Fortoul T, McDonnell W, Loomis D, Romieu I. (2007b) Lung function growth in
8 children with long-term exposure to air pollutants in Mexico City. Online data
9 supplement. Available: http://ajrccm.atsjournals.Org/cgi/data/176/4/377/DCl/l [3
10 October, 2007].
11
12 Roorda-Knape, MC, Janssen NAH, De Hartog JJ, Van Vliet PHN, Harssema H, Brunekreef B.
13 (1998). Air Pollution from traffic in city districts near major roadways. Atmos Environ.
14 32(11)1921-1930.
15
16 Rubenstein, I, Bigby BG, Reiss TF, Boushey HA Jr. (1990) Short-term exposure to 0.3 ppm
17 nitrogen dioxide does not potentiate airway responsiveness to sulfur dioxide in asthmatic
18 subjects. Am. Rev. Respir. Dis. 141:381-385.
19
20 Schildcrout, JS, Sheppard L, Lumley T, Slaughter JC, Koenig JQ, Shapiro GG. (2006).
21 Ambient air pollution and asthma exacerbations in children: an eight-city analysis. Am J
22 Epidemiol. 164:505-517.
23
24 Schwartz, J, Dockery DW, Neas LM, Wypij D, Ware JH, Spengler JD, Koutrakis P, Speizer FE,
25 Ferris BG, Jr. (1994). Acute effects of summer air pollution on respiratory symptom
26 reporting in children. Am JRespir Crit Care Med. 150:1234-1242.
27
28 Shorter, JH, Herndon S, Zahniser MS, Nelson DD, Wormhoudt J, Demerjian KL, Kolb CE.
29 (2005). Real-Time Measurements of Nitrogen Oxide Emissions from In-Use New York
30 City Transit Buses Using a Chase Vehicle. Environ Sci Technol. 39:7991-8000.
31
32 Spicer, CW, Kenny DV, Ward GF, Billick IH. (1993). Transformations, lifetimes, and sources
33 of NO2, HONO, and HNO3 in indoor environments. JAWMA. 43(11): 1479-1485.
34
35 Strand, V, Salomonsson P, Lundahl J, Bylin G. (1996). Immediate and delayed effects of
36 nitrogen dioxide exposure at an ambient level on bronchial responsiveness to histamine in
3 7 subj ects with asthma. Eur Respir J. 9:73 3 -740.
38
39 Strand, V, Rak S, Svartengren M, Bylin G. (1997). Nitrogen dioxide exposure enhances
40 asthmatic reaction to inhaled allergen in subjects with asthma. Am J Respir Crit Care
41 Med. 155:881-887.
August 2008 Draft 124
-------
2 Strand, V, Svartengren M, Rak S, Barck C, Bylin G. (1998). Repeated exposure to an ambient
3 level of NO2 enhances asthmatic response to nonsymptomatic allergen dose. Eur Respir
4 J. 12:6-12.
5
6 Tolbert, PE, Klein M, Peel JL, Sarnat SE, Sarnat JA. (2007). Multipollutant modeling issues in
7 a study of ambient air quality and emergency department visits in Atlanta. J Expos Sci
8 Environ Epidemiol. 17S2:S29-35.
9
10 Tunnicliffe, WS, Burge PS, Ayres JG. (1994) Effect of domestic concentrations of nitrogen
11 dioxide on airway responses to inhaled allergen in asthmatic patients. Lancet 344:1733-
12 1736.
13
14 US DOT. (2007). Part 3-The Journey To Work files. Bureau of Transportation Statistics
15 (BTS). Available at: httg^/transtats1bts±goy/.
16
17 Vardoulakis, S, Gonzalez-Flesca N, Fisher BEA, Pericleous K. (2004). Spatial variability of air
18 pollution in the vicinity of a permanent monitoring station in central Paris. Atmos.
19 Environ. 39:2725-2736.
20
21 Westerdahl, D, Fruin S, Sax T, Fine PM, Sioutas C. (2005). Mobile platform measurements of
22 ultrafme particles and associated pollutant concentrations on freeways and residential
23 streets in Los Angeles. Atmos Environ. 39:3597-3610.
24
25 Witten, A, Solomon C, Abbritti E, Arjomandi M, Zhai W, Kleinman M, Balmes J. (2005).
26 Effects of nitrogen dioxide on allergic airway responses in subjects with asthma. J.
27 Occup. Environ. Med. 47:1250-1259.
28
29 Wolff, GT. (1993). Letter to EPA Administrator Carol Browner: "CASAC Closure on the Air
30 Quality Criteria Document for Oxides of Nitrogen." EPA-SAB-CASAC-LTR-93-015,
31 September 30.
32
33 Wolff, GT. (1995). Letter to EPA Administrator Carol Browner: "CASAC Review of the Staff
34 Paper for the Review of the National Ambient Air Quality Standards for Nitrogen
35 Dioxide: Assessment of Scientific and Technical Information." EPA-SAB-CASAC-
36 LTR-95-004, August 22.
37
38 Yao, X, Lau NT, Chan CK, Fang M. (2005). The use of tunnel concentration profile data to
39 determine the ratio of NO2/NOX directly emitted from vehicles. Atmos Chem Phys Discuss.
40 5:12723-12740. Available at: http://www.atmos-chem-phys-discuss.net/5/12723/2005/acpd-
41 5-12723-2005.pdf.
August 2008 Draft 125
-------
United States Office of Air Quality Planning and Standards EPA-452/P-08-004a
Environmental Protection Air Quality Strategies and Standards Division August 2008
Agency Research Triangle Park, NC
------- |