EPA
United States
Environmental
Protection Agency
Carbon Monoxide National Ambient
Air Quality Standards: Scope and
Methods Plan for Health Risk and
Exposure Assessment
April 2009
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
EPA-452/R-09-004
April 2009
Carbon Monoxide National Ambient Air Quality Standards:
Scope and Methods Plan for Health Risk and Exposure
Assessment
U.S. Environmental Protection Agency
Office of Air and Radiation
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Ambient Standards Group
Research Triangle Park, NC 27711
11
-------
Disclaimer
This Carbon Monoxide National Ambient Air Quality Standards: Scope and Methods Plan for
Health Risk and Exposure Assessment (also referred to in this document as the Plan) has been
prepared by the Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency (EPA). Any opinions, findings, conclusions, or recommendations are those of the
authors and do not necessarily reflect the views of EPA. This document is being circulated to
facilitate consultation with the Clean Air Scientific Advisory Committee (CASAC) and for
public comment. Comments on this document should be addressed to Dr. Ines Pagan (email:
pagan.ines @epa.gov), U.S. Environmental Protection Agency, Office of Air Quality Planning
and Standards, C504-6, Research Triangle Park, North Carolina 27711.
11
-------
Table of Contents
1. Introduction 1
1.1. Chronology of the CO NAAQS Reviews 2
1.2. Overview of the Exposure/Dose Assessment analysis in Prior Reviews 4
2. Health effects and Approach to Risk Characterization 6
2.1. Health Effects and Effects Levels 6
2.2. Approach to Risk Characterization for Cardiovascular-Related Health Effects
Observed in Controlled Human Exposure Studies 8
2.3. Approach for Risk Characterization for Cardiovascular-related Health Effects
Reported in Epidemiological Studies 11
3. Scope and Approach for Population Exposure/Dose Analysis 13
3.1. Previous Assessments 13
3.2. The Exposure/Dose Modeling 15
3.2.1.Improvements to Algorithms 18
3.2.2.Improvements to Model Input Data 19
3.2.3.The COHb Model 20
3.3. Current Approach for Exposure and Dose Modeling 20
3.3.1.Specification of Microenvironments 21
3.3.2.Population Demographics 22
3.3.3. Commuting 23
3.3.4.Ambient CO Concentrations 23
3.3.5. Air Quality Adjustment to Meet Standards 25
3.3.6.Indoor Sources 26
3.4. Characterization of Uncertainty and Variability 26
3.4.1.Addressing Variability 27
3.4.2.Uncertainty Characterization - Qualitative Assessment 28
3.4.3.Uncertainty Characterization - Quantitative Analysis 29
4. Schedule of key milestones 31
5. References 33
Appendix A. Sensitivity Measures A-l
List of Figures
Figure 2-1. The Effect of CO Exposure on Time to Onset of Angina in Controlled Human
Exposure Studies 9
Figure 3-1. Conceptual Model and Data Flow of APEX 16
in
-------
List of Tables
Table 3-1. Microenvironments to be Modeled 22
Table 3-2. Estimated Values of the Parameter M in Equation 1 25
Table 4-1. Key Milestones for the Risk and Exposure Assessments 32
IV
-------
List of Acronyms/Abbreviations
APEX EPA's Air Pollutants Exposure model
AQCD Air Quality Criteria Document
AQS EPA's Air Quality System
CAA Clean Air Act
CASAC Clean Air Scientific Advisory Committee
CFK Coburn-Forster-Kane equation
CHAD EPA's Consolidated Human Activity Database
CO Carbon monoxide
COHb Carboxyhemoglobin
CTPP Census Transportation Planning Package
EPA United States Environmental Protection Agency
GC Gas chromatography
Hb Hemoglobin
hr Hour
ISA Integrated Science Assessment
IRP Integrated Review Plan
MET Metabolic equivalents by activity
NAAQS National ambient air quality standards
NCEA National Center for Environmental Assessment
NEI National Emissions Inventory
NEM NAAQS Exposure Model
NCDC National Climatic Data Center
NRC National Research Council
NWS National Weather Service
OAQPS Office of Air Quality Planning and Standards
OAR Office of Air and Radiation
ORD Office of Research and Development
O2 Oxygen
pNEM Probabilistic NAAQS Exposure Model
ppb Parts per billion
ppm Parts per million
PRB Policy-relevant background
RE A Risk and Exposure Assessment
SD Standard deviation
TRIM EPA's Total Risk Integrated Methodology
-------
1. INTRODUCTION
The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
the carbon monoxide (CO) national ambient air quality standards (NAAQS). EPA's overall plan
and schedule for this CO NAAQS review are presented in the Plan for Review of the Carbon
Monoxide National Ambient Air Quality Standards (US EPA, 2008b), or Integrated Review Plan
(IRP). This IRP (US EPA, 2008b) outlines the Clean Air Act (CAA) requirements related to the
establishment and periodic review of the NAAQS and the process and schedule for conducting
the current CO NAAQS review. It presents the key policy-relevant issues to be addressed in this
review as a series of policy-relevant questions that will frame our approach to determining
whether the current primary NAAQS for CO should be retained or revised. l The IRP also
discusses two key components in the NAAQS review process - an Integrated Science
Assessment (ISA) and a Risk and Exposure Assessment (REA) - in addition to the policy
assessment and rulemaking components that complete the review.
The ISA, prepared by EPA's Office of Research and Development (ORD), National
Center for Environmental Assessment (NCEA), provides a critical assessment of the latest
available policy-relevant scientific information upon which the NAAQS are to be based. The
ISA critically evaluates and integrates scientific information on the health and welfare effects
associated with exposure to CO in ambient air. At this time, a first draft of the ISA and related
Annexes (US EPA, 2009) has been released for CASAC review and public comment at an
upcoming meeting (scheduled for May 12-13, 2009). The REA, to be prepared by EPA's Office
of Air and Radiation (OAR), Office of Air Quality Planning and Standards (OAQPS), will draw
from the information assessed in the ISA. The REA will focus on a quantitative assessment of
exposure and dose metrics that are relevant to health effects of concern. The REA will include,
as appropriate, quantitative estimates of human exposures and risks associated with recent
ambient levels of CO, with levels simulated to just meet the current standards, and with levels
simulated to just meet possible alternative standards.
1 This plan will generally refer to the review of the primary standards for CO because there is currently no
secondary NAAQS for CO to review. However, the scope of EPA's review will include consideration of whether,
based on the revised air quality criteria for CO, it is appropriate to propose a new secondary standard.
-------
This document describes the scope and methods planned for use in conducting the human
health risk and exposure assessments to support the review of the primary (health-based) CO
NAAQS. Since this document is being prepared early in the review process, prior to CASAC
and public review of the first draft ISA (US EPA, 2009), it is appropriately general in nature.
Nonetheless, it is intended to provide enough specificity to facilitate consultation with CASAC,
as well as for public comment, in order to obtain advice on the overall scope, approaches, and
key issues in advance of conducting analyses and presenting results in the first draft REA. The
first draft ISA (US EPA, 2009) was used as the basis for the development of the approaches
described below. This includes information on atmospheric chemistry, source emissions, air
quality, human exposure, dosimetry and pharmacokinetics, and related health effects. CASAC
consultation on this planning document coincides with its review of the first draft ISA (US EPA,
2009). CASAC and public comments on this document will be taken into consideration in the
development of the first draft REA, the preparation of which will coincide with and draw from
the second draft ISA. The second draft REA will draw on the final ISA and will reflect
consideration of CASAC and public comments on the first draft REA. The final REA will take
into consideration CASAC and public comments on the second draft REA. The final ISA and
final REA will inform the policy assessment and rulemaking steps that will lead to proposed and
final decisions on the CO NAAQS.
This introductory chapter includes a chronological description of events that mark the
most significant milestones in the CO NAAQS reviews that have been conducted since 1971.
Chapter 2 presents and overview of health effects evidence relevant to the planned assessments
and the basic approach to risk characterization in this plan. Chapter 3 presents the approach for
the planned exposure/dose analysis and for characterizing uncertainty and variability in the
analysis. The schedule for completing these assessments is presented in Chapter 4.
1.1. Chronology of the CO NAAQS Reviews
On April 30, 1971, EPA promulgated identical primary and secondary NAAQS for CO,
under section 109 of the Act, set at 9 parts per million (ppm), 8-hr average and 35 ppm, 1-hr
average, neither to be exceeded more than once per year (36 FR 8186). In 1979, EPA published
the Air Quality Criteria Document for Carbon Monoxide (1979 AQCD) (US EPA, 1979a), which
updated the scientific criteria upon which the initial CO standards were based. A Staff Paper
-------
(US EPA, 1979b) was prepared and, along with the 1979 AQCD, served as the basis for the
development of the proposed rulemaking (45 FR 55066) published on August 18, 1980. Delays
due to uncertainties regarding the scientific basis for the final decision resulted in EPA's
announcing a second public comment period (47 FR 26407). Following substantial
reexamination of the scientific data, EPA prepared an Addendum to the 1979 AQCD (US EPA,
1984a) and an updated Staff Paper (US EPA, 1984b). Following review by CAS AC, EPA
announced its final decision (50 FR 37484) not to revise the existing primary standard and to
revoke the secondary standard for CO on September 13, 1985, due to a lack of evidence of direct
effects on public welfare at ambient concentrations.
In 1987, EPA initiated action to revise the criteria for CO and released a revised AQCD
for CASAC and public review. In a "closure letter" (McClellan, 1991) sent to the Administrator,
the CASAC concluded that the 1991 AQCD (US EPA, 1991) ". . . provides a scientifically
balanced and defensible summary of current knowledge of the effects of this pollutant and
provides an adequate basis for the EPA to make a decision as to the appropriate primary NAAQS
for CO." A revised Staff Paper was subsequently reviewed by CASAC and the public, and in a
"closure letter" (McClellan, 1992) sent to the Administrator, it was stated that"... a standard of
the present form and with a numerical value similar to that of the present standard would be
supported by the present scientific data on health effects of exposure to carbon monoxide."
Based on the 1991 AQCD (US EPA, 1991) and staff conclusions and recommendations
contained in the revised Staff Paper (US EPA, 1992), the Administrator announced the final
decision (59 FR 38906) on August 1, 1994, that revision of the primary NAAQS for CO was not
appropriate at that time. Thus, the primary standards were retained at 9 parts per million (ppm),
8-hr average and 35 ppm, 1-hr average, neither to be exceeded more than once per year.
In 1997, revisions to the AQCD (US EPA, 1991) for the CO NAAQS were initiated and a
workshop was held in September 1998 to review and discuss material to be contained in the
revised AQCD. On June 9, 1999, CASAC held a public meeting to review the first draft AQCD
and to provide a consultation on a draft exposure analysis methodology document. CASAC
Panel members provided comments and suggestions for the exposure analysis methodology,
including improvements for modeling indoor sources and ventilation rates, and calling on EPA to
do more to address the overall uncertainly in the exposure/dose model. Comments from CASAC
Panel members and the public on the AQCD were considered in a second draft AQCD, which
-------
was reviewed at a CAS AC meeting, held on November 18, 1999. After revision of the second
draft AQCD, the final 2000 AQCD (US EPA, 2000) was released in August 2000. EPA put on
hold the NAAQS review when Congress requested that the National Research Council (NRC)
review the impact of meteorology and topography on ambient CO concentrations in high altitude
and extreme cold regions of the U.S. In response, the NRC convened the Committee on Carbon
Monoxide Episodes in Meteorological and Topographical Problem Areas, which focused on
Fairbanks, Alaska as a case-study. A final report, "Managing Carbon Monoxide Pollution in
Meteorological and Topographical Problem Areas," was published in 2003 (NRC, 2003) and
offered a wide range of recommendations regarding management of CO air pollution, cold start
emissions standards, oxygenated fuels, and CO monitoring. Following completion of this NRC
report, EPA did not conduct rulemaking to complete the review.
EPA initiated the current review of the NAAQS for CO on September 13, 2007, with a
call for information from the public (72 FR 52369) requesting the submission of recent scientific
information on specified topics. A workshop was held on January 28-29, 2008 (73 FR 2490) to
discuss policy-relevant scientific and technical information to inform EPA's planning for the CO
NAAQS review. Following the workshop, a draft of EPA's Integrated Review Plan (IRP) "Plan
for Review of the National Ambient Air Quality Standards for Carbon Monoxide" (US EPA,
2008a) was made available in March 2008 for public comment and was discussed by the
CASAC via a publicly accessible teleconference consultation on April 8, 2008 (73 FR 12998).
EPA made the final plan available in August 2008 (US EPA, 2008b). In November 2008, EPA
held an authors' teleconference with invited scientific experts to discuss preliminary draft
materials prepared during the ongoing development of the CO ISA and its supplementary
Annexes. The first draft ISA (US EPA, 2009) for CO was made available for public review on
March 12, 2009, and will be reviewed by CASAC concurrently with this Scope and Methods
Plan at a meeting scheduled for May 12 and 13, 2009.
1.2. Overview of the Exposure/Dose Assessment analysis in Prior Reviews
Reviews of the CO NAAQS completed in 1985 and 1994 did not include quantitative
health risk assessments. Rather, these reviews included analysis of exposure to ambient CO and
associated internal dose in terms of carboxyhemoglobin (COHb) levels which were used to
characterize risks in selected urban study areas. These prior risk characterizations compared the
-------
numbers of at-risk individuals and percent of the at-risk population exceeding several potential
health effect benchmarks, expressed in terms of COHb levels. This characterization was based
on COHb levels observed in several controlled human exposure studies reporting aggravation of
angina associated with short-term (< 8-hr) CO exposures.
Although the EPA did not complete the review initiated in 1997, OAQPS continued work
on the CO exposure assessment to further develop the exposure assessment modeling component
of the Total Risk Integrated Methodology (TRIM) system. A draft technical report (Johnson et
al., 2000) was produced documenting the application of the CO exposure and dose modeling
methodology for two study areas (Denver and Los Angeles). This report was subjected to an
external peer review by three exposure modeling experts convened by Science Applications
International Corporation (SAIC, 2001).
The methods used in this previous assessment, described below in chapter 3, form the
bases for the analysis planned for this review. The planned analysis will also incorporate
improvements made to the exposure model since the previous assessment.
-------
2. HEALTH EFFECTS AND APPROACH TO RISK
CHARACTERIZATION
The overall scope and approach to risk characterization for the current CO NAAQS
review builds upon the methodology and analyses conducted in prior reviews of the CO
standards. A brief summary of the health effects evidence and our provisional judgments about
health effects endpoints to be included in the risk characterization is presented below in section
2.1, based on information in the first draft ISA. We note that the first draft REA will be
informed by CASAC and public comment on this plan and review of the first draft ISA (US
EPA, 2009), in addition to the information and evaluation contained in the second draft ISA and
associated Annexes. The basic approach to risk characterization described in this plan reflects
the availability of data from two different types of health studies: controlled human exposure
studies and epidemiologic studies. Our plan to address the range of effects related to CO
ambient exposures evaluated in the first draft ISA (US EPA, 2009) is organized based on the
health effects supported by these two types of studies and is discussed below in sections 2.2 and
2.3, respectively.
2.1. Health Effects and Effects Levels
The mechanism of toxicity believed to be associated with health effects of greatest
concern from CO exposure is hypoxia induced by elevated COHb levels. The primary exchange
route for CO to human tissues is through the lungs. Although CO is a naturally occurring
chemical in blood, being produced endogenously by normal catabolic processes, blood COHb
levels do not often exceed 0.5 to 0.7 percent in healthy individuals unless exogenous CO is
inhaled. Some individuals with high endogenous CO production can have COHb levels of 1.0 to
1.5 percent (e.g., people with anemia). Exogenous CO diffuses across the alveoli in the
pulmonary region of the lung, entering the blood where it immediately binds with hemoglobin
(Hb) to form COHb. Most healthy individuals can physiologically compensate for the resulting
reduction in tissue oxygen (©2) levels (e.g. through increased blood flow, blood vessel dilation)
although the effect of reduced maximal exercise capacity has been reported in healthy persons
even at low COHb levels (-3%). However, reduced delivery of O2 is of heightened concern for
individuals with ischemic heart diseases, since they have an already compromised O2 delivery
system to the heart muscle, which puts them at increased risk if exposed to CO.
-------
The first draft ISA (US EPA, 2009) indicates that the integrative synthesis of the
available evidence from controlled human exposure, epidemiologic, and toxicological studies
suggests that a causal relationship is likely to exist between relevant short-term CO exposures
and cardiovascular morbidity. The "most compelling evidence of a CO-induced effect on the
cardiovascular system" comes from controlled exposure studies in humans with coronary artery
disease (US EPA, 2009, sections 2.3.1 and 5.2). These studies, described in more detail in the
1991 and 2000 AQCDs, "demonstrate consistent decreases in the time to onset of exercise-
induced angina and S-T segment changes at COHb levels ranging from 3-6 percent, with one
multicenter study reporting similar effects at COHb levels as low as 2.4 percent" (US EPA,
2009, p.2-6). The first draft ISA (US EPA, 2009) also indicates that there are no human clinical
studies published since the 2000 AQCD evaluating the effects of CO exposures resulting in
COHb levels lower than 2.4 percent (US EPA, 2009, p.2-6).
A number of epidemiologic studies published since the 2000 AQCD and evaluated in the
first draft ISA (US EPA, 2009) report associations between ambient CO concentrations and
increased emergency department visits and hospital admissions for individuals suffering from
cardiovascular disease. The 2000 AQCD concluded that epidemiologic studies provided
evidence that short-term variations in ambient CO concentrations were associated with daily
hospital admissions for heart disease. The first draft ISA (US EPA, 2009) builds on this
conclusion. All but one of the recent epidemiologic studies was conducted in locations where
the entire distribution of monitored CO concentrations were below the level of the current CO
NAAQS (for details on the recent epidemiologic studies see US EPA, 2009, Table 5-7). In
discussing the epidemiologic evidence, the first draft ISA (US EPA, 2009, p.5-45) notes that "it
is difficult to determine from this group of studies the extent to which CO is independently
associated with cardiovascular disease outcomes or if CO is a marker for the effects of another
traffic-related pollutant or mix of pollutants." While acknowledging that this "complicates the
efforts to disentangle specific CO-related health effects" the first draft ISA (US EPA, 2009)
notes that CO associations generally remain robust in copollutant models, that the specific
endpoints are coherent with human clinical and toxicologic evidence from studies conducted at
higher concentrations, and that these considerations "support a direct effect of short-term CO
exposure on cardiovascular morbidity at ambient concentrations below the current NAAQS
level."
-------
2.2. Approach to Risk Characterization for Cardiovascular-Related Health Effects
Observed in Controlled Human Exposure Studies
As discussed above, there are a number of controlled human exposure studies reporting
reduced time to onset of angina and other clinical cardiovascular measures in moderately
exercising angina subjects who received short-term exposures to CO. As shown in Figure 2-1,
there are statistically significant group mean responses, measured in terms of reduced time to
onset of exercise-induced angina, observed in the range of 3 to 6 percent COHb (measured by
CO-oximeter) in subjects with coronary artery disease. However, there is no clear pattern across
the different studies with respect to the magnitude of the decreased time to onset of angina versus
dose level. It is important to note that the results presented in Figure 2-1 have been compiled
from individual studies with different study design and methodology (e.g., different exercise
duration, methods used to measure COHb with different levels of accuracy, subject populations);
therefore, comparisons must be interpreted with caution. In addition, these studies do not
address the fraction of the population experiencing a specified health effect at various dose
levels. Thus, based on information in the first draft ISA (US EPA, 2009), there does not appear
at this time to be sufficient controlled human exposure data to support the development of
quantitative dose-response relationships which would be required in order to conduct a
quantitative risk assessment for this health endpoint.
-------
CO ^w
I25-
1
P 15-
c
^ 10-
CO
(D
b 5-
(J) ^
D_
_m
I Anderson et al.
(1973) B
1
Kleinman
et al. T T
T (1998)^ j
Klpinman \ ~T Allred et al. I
i • T _L (1 989a b 1
MQRQ'> T P 1991) ~ X Adams et al.
^lyoyj I Allred et al. W nQ88^
(1989a,b, T
1991) i Shepsetal.
T (1987)
__
Percent COHb
Figure 2-1. The Effect of CO Exposure on Time to Onset of Angina in Controlled Human
Exposure Studies
For comparison across studies, data are presented as mean percent differences in COHb
levels (CO-oximeter measurement) between air- and CO-exposure days for individual
subjects calculated from each study. Bars indicate calculated standard errors of the mean.
Source: Adapted from the 2000 Air Quality Criteria for Carbon Monoxide (US EPA, 2000)
Similar to the approach used in prior CO NAAQS reviews, we plan to estimate CO
exposures and resulting doses (i.e., COHb levels) and characterize risk for the population with
cardiovascular disease in two urban study areas associated with CO levels representing recent air
quality and air quality adjusted to simulate just meeting the current CO NAAQS and any
potential alternative standards under consideration. Risk will be characterized using a potential
health effect benchmark level approach. More specifically, we will estimate the number and
percent of the population with cardiovascular disease that would exceed potential health effect
-------
benchmark levels, derived from the evaluation of the controlled human exposure studies
summarized above and specified in terms of COHb levels, upon just meeting various CO air
quality scenarios.
Potential health benchmark values to be used in the planned risk characterization linked
to the exposure/dose analyses will be derived solely based on the controlled human exposure
literature. This is primarily because CO concentrations reported in controlled human exposure
studies represent actual personal exposures rather than concentrations measured at fixed site
ambient monitors. In addition, controlled human exposure studies can examine the health effects
of CO in the absence of co-pollutants that can confound results in epidemiologic analyses; thus,
health effects observed in controlled human exposure studies can confidently be attributed to a
defined COHb dose level associated with ambient CO exposures.
In identifying the potential health effect benchmark levels, staff plan to use 2.0, 2.5 and
3.0 percent COHb for the risk characterization for the effects observed in cardiovascular disease
patients reported in a number of controlled human exposure studies. This range captures the
lowest adverse effect levels reported in most of the controlled human exposure studies reporting
effects of CO on individuals with angina. While most of the early studies used CO-oximeters to
measure COHb levels, later studies used gas chromatography (GC) as the method of
measurement, which is widely considered to be more accurate. Comparisons summarized in the
1991 AQCD (US EPA, 1991) between CO-oximeter and GC measurements of COHb found that
the CO-oximeter measurements could be either higher or lower, depending on the specific
instrument and the measurement range, and were particularly variable at low COHb levels (<
5%).
As discussed in section 3.2, the calculation of dose (blood levels of COHb) for the
exposure/dose modeling planned for this assessment is based on the well-established Coburn-
Forster-Kane (CFK) equation (Coburn et al., 1965). We recognize that COHb estimates from the
exposure/dose modeling are more appropriately compared to COHb levels measured using GC.
The range of potential health effects benchmarks that we plan to use extends lower than the
range where controlled human exposure studies reported CO-related health effects (i.e., 3-6
percent COHb with one multicenter study reporting effects at 2.4% COHb using GC) to take into
consideration both the uncertainty about the actual COHb levels experienced in the controlled
human exposure studies due to the use of different measurement methods and that these studies
10
-------
did not include individuals with more severe cardiovascular disease who may respond at lower
COHb levels relative to the subjects tested. Based on this consideration, staff believes that 2.0,
2.5, and 3.0 percent COHb are appropriate values for potential health effect benchmark levels to
be included in the current CO risk characterization to address concerns about cardiovascular
effects observed in a number of controlled human exposure studies.
The exposure and dose estimation will be conducted using EPA's APEX model (see
section 3 for additional details). Counts will be estimated for the number of people and the total
number of occurrences for which various potential health effect COHb benchmark levels are
exceeded. We selected Denver and Los Angeles areas as exposure modeling areas of interest
because they (1) have been included in prior CO NAAQS exposure assessments and thus serve
as an important connection with past assessments, (2) they have historically had the highest
elevated CO ambient concentrations among urban areas in the US, (3) Denver represents a high
altitude city and the interaction of CO and high altitude is of interest, and (4) they are in the top
four of the 10 urban areas evaluated in the ISA with at least 75 percent data completeness with
respect to the maximum and 99th percentile 1- and 8-hr daily maximum CO concentrations.
2.3. Approach for Risk Characterization for Cardiovascular-related Health Effects
Reported in Epidemiological Studies
In deciding whether or not to conduct a quantitative risk assessment for cardiovascular
morbidity based on associations reported in community epidemiologic studies, we plan to take
into account the following considerations: (1) whether the weight of the evidence supports
conducting a quantitative assessment for specific health endpoints, (2) whether the data needed
to conduct such quantitative assessments are available, (3) the anticipated utility of results to
inform decisions on the adequacy of the current CO NAAQS and to provide insights related to
potential alternative standards, and (4) whether there is adequate time to complete such
assessments under the current court-ordered schedule.
As noted above, the first draft ISA (US EPA, 2009) evaluates epidemiologic findings
from a group of studies, many of which were conducted since the 2000 CO AQCD (US EPA,
2000) that observed associations between ambient CO concentrations and increases in
emergency department visits and hospital admissions for cardiovascular effects. All but one of
11
-------
the recent epidemiologic studies were conducted in locations where the entire distribution of
monitored CO concentrations was below the level of the current CO NAAQS. As noted
previously in this document, uncertainty in evaluating the epidemiological evidence, specifically,
whether the effects reported at relatively low ambient CO concentrations in these studies are
causally related to CO or whether ambient CO levels are serving as a surrogate for one or more
components of the overall traffic-related air pollutant mixture may preclude the use of this
evidence in a quantitative risk assessment. Moreover, there are concerns about whether
measurement error in epidemiological studies utilizing fixed site monitors, which are potentially
a poor representation of personal exposures to CO that vary spatially and temporally, can
influence the observed association between CO and cardiovascular effects. In staffs view
whether the results of co-pollutant models (US EPA, 2009, Figure 5.5) provide sufficient
evidence to support conducting a quantitative risk assessment for CO effects at ambient levels
warrants further consideration in consultation with CASAC prior to EPA deciding whether to
conduct an epidemiologically based quantitative risk assessment for cardiovascular-related
hospital admissions or emergency department visits.
12
-------
3. SCOPE AND APPROACH FOR POPULATION EXPOSURE/DOSE
ANALYSIS
3.1. Previous Assessments
The model used for exposure analysis was pNEM/CO (probabilistic NEM applied to
CO), a version of the CO NAAQS Exposure Model (NEM) that incorporated Monte Carlo
sampling and multiple runs, or realizations, of the model. The model outputs of interest were
estimates of the number of person-days of exposure to various CO levels for various scenarios
for adults with cardiovascular disease in Denver. Estimates also were made of the percentage of
the cardiovascular heart disease population in Denver2 that would exceed selected COHb levels
one or more times per year under different scenarios. The estimates of COHb were derived by
applying a modified version of the CFK differential equation that estimates COHb levels from
CO exposure as a function of time and physiological and environmental factors (e.g., blood
volume, altitude, endogenous CO production rate).
The analysis indicated that if the current 8-hr standard were just met, the proportion of
the nonsmoking population with cardiovascular disease experiencing exposures at or above
9 ppm for 8 hrs decreased by an order of magnitude or more from existing CO levels, down to
less than 1 percent of the total person-days in that population. Likewise, meeting the current 8-hr
standard reduced the proportion of the nonsmoking cardiovascular-disease population person
days at or above COHb levels of concern by an order of magnitude or more relative to existing
CO levels. Upon meeting the 8-hr standard, EPA estimated that less than 0.1 percent of the
nonsmoking cardiovascular-disease population would experience a COHb level of about 2.1
percent. A smaller percentage of the at-risk population was estimated to exceed higher COHb
percentages. The analysis also took into account that certain indoor sources (e.g., passive
smoking, gas stove usage) contributed to total CO exposure but could not be effectively
mitigated by setting more stringent ambient air quality standards.
Additional exposure analyses were planned in 1999 using the Denver and Los Angeles
areas to provide estimates of CO exposures and resultant COHb levels for adults with
2 It was estimated in the 1992 exposure analysis that there were about 36,800 non-smoking adults in Denver with
diagnosed or undiagnosed (silent) ischemia.
13
-------
cardiovascular disease in these two urban areas. Denver was included in the planned analyses
for comparison purposes because it was the only city included in the exposure analysis
conducted in the previous review and it was one of the areas with the highest ambient CO levels
in the country. In addition, Denver was one of a few areas where a personal CO exposure study
had been conducted. After an initial review of the methodology, EPA also planned to conduct
the analyses for Los Angeles for several reasons: (1) it presented the largest potential public
health burden due to its relatively higher ambient CO levels and potential population exposure;
(2) an extensive monitoring network was available; and (3) availability of a study in Los Angeles
of personal and indoor CO concentrations that potentially could be used to evaluate the model.
The primary target population in the 1999 analysis was adults with cardiovascular
disease, as it was in the 1992 analysis. The 1999 analysis initially focused on several scenarios:
(1) current air quality (1995 for Denver); and (2) the presence of indoor sources (gas
stoves/ovens and passive smoking) versus ambient air without indoor sources. The analyses
were intended to provide a basis for assessing protection afforded by the current CO standards
and preliminary insight into the relative impact indoor sources may have on total exposure. The
model selected to estimate population exposure was an updated version of pNEM/CO that was
used in the 1992 Denver analysis, with the major outputs of interest being estimates of the
number and percentage of person-days of exposure to various CO levels and the number and
percentage of person-hrs and people exceeding various COHb levels. Only the 8-hr NAAQS
was planned for evaluation because previous analyses indicated that it was the controlling
standard for attainment.
A draft exposure analysis report (Johnson et al., 1999) applying the updated exposure model only
to the Denver area was provided to the CASAC CO Panel and made available for public review
in March 1999 for the purpose of obtaining scientific and public input on the proposed
methodology. The CASAC CO Panel conducted a consultation on the methodology for the
analysis on June 10, 1999. The CASAC Panel members provided a number of specific
suggestions, including improving the algorithm for estimating inside vehicle exposures,
differentiating between electronic and gas pilot lights for stoves, and using alveolar instead of
inhaled ventilation rates in the physiological model. Since that time, these and other
improvements to the model have been made, as described in the next section. The CASAC Panel
members also suggested that the exposure analysis include additional information to address the
14
-------
overall uncertainty in the model. This plan addresses this issue in section 3.4. As noted in
section 1.1, subsequent to the CASAC consultation on the 1999 draft CO exposure methodology
report, a draft technical report (Johnson et al., 2000) was produced documenting the application
of the CO exposure and dose modeling methodology for Denver and Los Angeles. This report
was developed as part of the EPA's efforts to improve its exposure modeling tools.3 As
described in more detail in sections 3.2 and 3.3 below, the planned exposure/dose assessment
builds on the 1999 and 2000 CO exposure/dose modeling efforts for Denver and Los Angeles.
3.2. The Exposure/Dose Modeling
EPA's Air Pollutants Exposure (APEX) model (also referred to as the Total Risk
Integrated Methodology/Exposure (TREVI.Expo) model) will be used in this analysis for the
estimation of population exposures and resulting dose due to ambient CO levels. The EPA has
developed APEX as a tool for estimating human population exposure to criteria and air toxic
pollutants. APEX serves as the human inhalation exposure model within the Total Risk
Integrated Methodology (TRIM) framework (Richmond et al., 2002; US EPA 2003).
Figure 3-1 provides a schematic overview of the APEX model. APEX simulates the
movement of individuals through time and space and their exposure to a given pollutant in
indoor, outdoor, and in-vehicle microenvironments. The model stochastically generates
simulated individuals using census-derived probability distributions for demographic
characteristics. The population demographics are from the 2000 Census at the tract level, and a
national commuting database based on 2000 Census data provides home-to-work commuting
flows between tracts. Any number of simulated individuals can be modeled, and collectively
they represent a random sample of the study area population.
APEX has a flexible approach for modeling microenvironmental concentrations, where
the user can define the microenvironments to be modeled and their characteristics. Typical
indoor microenvironments include residences, schools, and offices. Outdoor microenvironments
include near roadways, bus stops, and playgrounds. Inside cars, trucks, and mass transit vehicles
are microenvironments which are classified separately from indoors and outdoors.
3 This draft report was subjected to an external peer review by 3 exposure modeling experts convened by Science
Applications International Corporation (SAIC, 2001).
15
-------
Figure 3-1. Conceptual Model and Data Flow of APEX
16
-------
APEX calculates the concentration in the microenvironment associated with each event in an
individual's activity pattern to obtain event-specific exposures.
The concentrations in each microenvironment are calculated using either a factors or
mass-balance approach, and the user specifies the probability distributions of the parameters that
go into the concentration calculations (e.g., indoor-outdoor air exchange rates). These
distributions can depend on the values of other variables in the model. For example, the
distribution of air exchange rates in a home, office, or car depends on the type of heating and air
conditioning present, which are also stochastic inputs to the model. The user can choose to keep
the value of a stochastic parameter constant for the entire simulation (e.g., house volume), or can
specify that a new value shall be drawn hourly, daily, or seasonally from specified distributions.
APEX also allows the user to specify diurnal, weekly, or seasonal patterns for various
microenvironmental parameters.
APEX was derived from the probabilistic NAAQS Exposure Model (pNEM) used in
prior CO NAAQS exposure assessments as described in section 3.1. Since that time the model
has been restructured, improved, and expanded to reflect conceptual advances in the science of
exposure modeling and newer input data available for the model. A user's guide and technical
support document describe the APEX model in detail (US EPA, 2008c,d). As discussed below,
key improvements to algorithms include:
• replacement of the cohort approach with a probabilistic sampling approach focused on
individuals,
• development of a flexible method for specifying distributions for probabilistic
microenvironment parameters and other model inputs,
• enhancement with a new approach for construction of longitudinal activity patterns for
simulated persons, and
• accounting for fatigue and oxygen debt after exercise in the calculation of ventilation
rates.
Further major improvements to data input to the model are discussed below, which focus on:
• residential air exchange rates, and
• census and commuting data.
17
-------
3.2.1. Improvements to Algorithms
The pNEM was based on cohorts of people, with each cohort treated collectively as a
subgroup. APEX models individuals rather than cohorts, which allows APEX to address both
intra- and inter-individual variability in human activities, inhalation rates, and dose uptake rates.
The model randomly selects a sample of hypothetical individuals in an actual population
database and simulates each individual's movements through time and space (e.g., at home, in
vehicles) to estimate their exposure to the pollutant.
APEX simulates the variability in the factors affecting exposure, which is important for
assessment of the distribution of population exposures and dose. It incorporates stochastic
processes representing the natural variability of personal profile characteristics, activity patterns,
and microenvironment parameters. APEX has been developed to provide the user with a large
degree of flexibility in specifying the distributions for modeling these stochastic processes.
There are 15 parametric distributions and a percentile-based distribution that can be used, which
can be selected conditionally according to the values of other variables or sampled values from
other distributions.
A key issue in exposure modeling is the construction of year-long activity sequences for
individuals based on a cross-sectional activity data base of 24-hour records. The human activity
data will be drawn from the most recent version of the Consolidated Human Activity Database
(CHAD) (McCurdy et al., 2000; EPA, 2002), developed and maintained by the Office of
Research and Development's (ORD) National Exposure Research Laboratory (NERL). The
CHAD includes data from several surveys covering specific time periods at city, state, and
national levels, with varying degrees of representativeness, providing more than 28,000 diary-
days of activity data (compared to about 17,000 in the previous modeling effort). The typical
subject in the time/activity studies in CHAD provided less than two days of diary data. For this
reason, the construction of a season-long activity sequence for each individual requires some
combination of repeating the same data from one subject and using data from multiple subjects.
An appropriate approach should adequately account for the day-to-day and week-to-week
repetition of activities common to individuals while maintaining realistic variability between
individuals. The method in APEX for creating longitudinal diaries was designed to capture the
18
-------
tendency of individuals to repeat activities, based on reproducing realistic variation in a key
diary variable, which is a user-selected function of diary variables (Glen et al., 2008).
In addition to exposure, APEX models breathing rates based on the physiology of each
individual and the activities performed. For each activity type in CHAD, a distribution is
provided for a corresponding metabolic energy of work ratio, MET ratio (McCurdy, 2000). The
MET ratio is a ratio of the rate of energy consumption for non-rest activities as compared to the
resting rate of energy consumption. The MET ratios have less interpersonal variation than do the
absolute energy expenditures. Based on age and gender, the resting metabolic rate, along with
other physiological variables are determined for each individual as part of their anthropometric
characteristics. Because the MET ratios are sampled independently from distributions for each
diary event, it may be possible to produce time-series of MET ratios that are physiologically
unrealistic. APEX employs a MET adjustment algorithm based on a modeled oxygen deficit to
prevent such overestimation of MET and breathing rates. The relationship between the oxygen
deficit and the applied limits on MET ratios are nonlinear and are derived from published data on
work capacity and oxygen consumption (Isaacs et al., 2008).
3.2.2. Improvements to Model Input Data
Distributions of air exchange rates (AERs) for the indoor microenvironments will be
developed using data from several studies, comprising a total of more than 6,000 AER
measurements (EPA, 2007, Appendix A). We plan to develop distributions of AERs for the two
study areas based on stratification of the data by season and presence or absence of an air
conditioner, as well as by geographic location.
To ensure that individual's daily activities are accurately represented within APEX, it is
important to integrate working patterns into the assessment. The APEX tract-level commuting
data are derived from the 2000 Census and collected as part of the Census Transportation
Planning Package (CTPP). CTPP contains tabulations by place of residence, place of work, and
the flows between the residence and work. These data are available from the U.S. Department of
Transportation, Bureau of Transportation Statistics (U.S. Department of Transportation and U.S.
Census Bureau, 2000). This database was not available for the previous modeling effort.
19
-------
3.2.3. The COHb Model
Since COHb levels are a biomarker for the health effects of ambient-level exposures to
CO and are used as an bioindicator of CO exposure, the focus of the exposure/dose assessment is
on estimating COHb levels which can be related to potential health effect benchmarks as
described in section 2.1. Therefore, the relationship between COHb and exposure to ambient CO
levels is critical to the characterization of health risks associated with various CO air quality
scenarios.
Dose (blood levels of COHb) is calculated based on the estimated exposures, estimated alveolar
ventilation rate, and other physiological parameters for each simulated individual. The
calculation of dose is based on the well-established CFK equation (Coburn et al., 1965), and has
been used for many years by EPA and others to model COHb formation (see for example,
Richmond and Johnson, 2000). This is a mechanistic model that uses physical and physiological
processes and an understanding of biological processes to predict COHb levels resulting from
CO exposures. The literature discusses linear and non-linear forms of the CFK equation. The
linear form is an approximation that allows an explicit solution, but is not accurate under all
conditions. The non-linear form is considered to be more physiologically correct and is the one
implemented in APEX, taking into account exertion level and a variety of physiological
parameters (e.g., lung diffusivity, endogenous CO production rate, Hb level, blood volume) (US
EPA, 2008b). The CFK model is well accepted and has been evaluated using measured CO and
COHb and has been shown to provide a good approximation to the COHb level at a steady level
of inhaled exogenous CO (US EPA, 2009, page 4-2; US EPA, 2000, section 5.5.1).
3.3. Current Approach for Exposure and Dose Modeling
The exposure/dose assessment for the current review of the primary CO NAAQS is
designed to estimate human exposures and dose and to characterize the potential health risks that
are associated with recent ambient levels, with ambient levels that just meet the existing
standard, and with ambient levels that just meet alternative standards that may be under
consideration. The exposure/dose assessment draws upon the information presented in the draft
ISA and its Annexes, the previous AQCDs for CO (US EPA, 1991, 2000), and the 1999 and
2000 draft exposure analysis reports (Johnson et al., 1999, 2000) and subsequent improvements
to the exposure model in developing the CO exposure/dose assessment. This includes
20
-------
information on atmospheric chemistry, air quality, human exposure, formation of blood COHb
levels, and health effects of concern. The goals of the CO exposure/dose assessment are: (1) to
develop estimates of blood COHb levels in sensitive populations resulting from exposure to CO
for various CO air quality scenarios noted above; (2) to estimate the number of people and the
total number of occurrences for which potential health effect COHb benchmark levels are
exceeded for various CO air quality scenarios noted above; and (3) to identify key assumptions
and uncertainties in the exposure and dose estimates.
The general approach is to estimate population exposures to ambient CO in two urban
areas in the U.S., Denver, CO and Los Angeles, CA. These areas were selected since they have
been previously modeled for CO exposures using the pNEM model. These two urban areas
continue to be in the top urban areas evaluated in the first draft ISA with relatively complete data
with respect to daily maximum 1- and 8-hr CO concentration levels. These two urban areas
continue to also exhibit relatively higher CO ambient concentrations than other urban areas as
shown in the draft ISA.
In the first draft REA, exposure and dose estimates for the general population and the
population with cardiovascular disease for these areas will be generated for recent CO levels,
based on a recent 3-year period, and for levels adjusted to just meet the current NAAQS.
Exposures and dose would be modeled for any potential alternative CO standards in the second
draft REA, based on adjusting the air quality data. An exposure analysis technical support
document with a detailed description of the methodology and results will accompany the draft
REA.
3.3.1. Specification of Microenvironments
The first step in the exposure/dose assessment will be to update the parameter
distributions from the 2000 application for the mass balance and factors approaches for the
calculation of microenvironmental concentrations in APEX. For this modeling application, we
propose modeling the microenvironments listed in Table 3-1.
21
-------
Table 3-1. Microenvironments to be Modeled
Microenvironment Method
Indoors - residence mass balance
Indoors - restaurants, bars mass balance
Indoors - schools mass balance
Indoors - day care centers (commercial) mass balance
Indoors - other (e.g., offices, shopping) mass balance
or factors
Outdoors - bus stop factors
Outdoors - near road factors
Outdoors - other (e.g., playgrounds, parks) factors
In vehicle - cars and light trucks mass balance
or factors
In vehicle - heavy trucks mass balance
or factors
In vehicle - school buses mass balance
or factors
In vehicle - mass transit vehicles factors
We plan to calculate the concentrations in each microenvironment using either a factors
or mass-balance approach, depending upon data availability, with probability distributions for
the input parameters used in the calculations (e.g., indoor-outdoor air exchange rates) supplied as
inputs to the model. These distributions represent the variability of parameters, and can vary
spatially and can be set up to depend on the values of other variables in the model.
3.3.2. Population Demographics
We plan to obtain tract-level population counts from the 2000 Census of Population and
Housing Summary File I.4 Summary File 1 (SF 1) contains the 100-percent data, which is the
information compiled from the questions asked of all people and about every housing unit. In
the 2000 U.S. Census, estimates of employment were developed by census tract.5 The file input
to APEX will be broken down by gender and age group, so that each gender/age group
combination is given an employment probability fraction (ranging from 0 to 1) within each
census tract. The age groupings in this file are: 16-19, 20-21, 22-24, 25-29, 30-34, 35-44, 45-54,
4 http://www.census.gov/prod/cen2000/doc/sfl.pdf
5 Employment data from the 2000 Census can be found on the U.S. Census web site:
http://www.census.gov/population/www/cen2000/phc-t28.html (Employment Status: 2000- Supplemental Tables).
22
-------
55-59, 60-61, 62-64, 65-69, 70-74, and greater than 75 years of age. Children under 16 years of
age will be assumed to be not employed.
3.3.3. Commuting
As part of the population demographics inputs, it is important to integrate commuting
patterns into the assessment for workers. In addition to using estimates of employment by tract,
APEX also incorporates home-to-work commuting data. We plan to use the national commuting
database provided with APEX in this analysis. Commuting data were derived from the 2000
Census and were collected as part of the Census Transportation Planning Package (CTPP) (U.S.
DOT, 2000).6 The data used to generate APEX inputs were taken from the "Part 3-The Journey
To Work" files. These files contain counts of individuals commuting from home-to-work
locations at a number of geographic scales. These data have been processed to calculate
fractions for each tract-to-tract flow to create the national commuting data distributed with
APEX. This database contains commuting data for each of the 50 states and Washington, D.C.
This data set does not differentiate people that work at home from those that commute within
their home tract.
3.3.4. Ambient CO Concentrations
We plan to estimate ambient CO concentrations using the same methodology used in the
previous CO exposure modeling analysis and update the data underlying the methodology to the
extent that more recent data are available.
In the previous pNEM/CO application (see Johnson et al., 2000), outdoor concentrations
were estimated from the relationship
C mdt = Mm Lmd Tmdt Cdt ' (1)
where dis a monitor index, m is a microenvironment index, tis a time index (hourly), and
'district
-------
concentrations for microenvironments in the geographic district d are estimated using
concentrations measured at monitor d.
C°mdt = the estimated outdoor CO concentration with respect to microenvironment m in district
d at time t,
Mm = constant multiplier (> 0) specific to microenvironment m (Table 3-2),
Lmd = location multiplier randomly selected from a lognormal distribution with geometric
mean equal to 1.0 and geometric standard deviation equal to 1.52, for each
microenvironment m, and district d (held constant for all hours),
Tmdt = time-of-day multiplier randomly selected from a lognormal distribution with
geometric mean equal to 1.0 and geometric standard deviation equal to 1.63, for each
microenvironment m, district d, time t, and
Cdt = the CO concentration measured at monitor d at time t.
The development of equation 1 was based on data from a California residential exposure
study conducted by Wilson, Colome, and Tian (1995) together with data provided by the Denver
Personal Monitoring Study (Akland et al, 1985; Johnson, 1984). The parametric distributions for
L and T were estimated by analyzing data obtained from the California residential study, which
measured 10-minute CO concentrations outside 156 residences, 70 from Los Angeles and 86
from San Diego This yielded a database of 6,330 hourly average concentrations, which were
compared with the hourly CO concentrations measured simultaneously at the nearest fixed-site
monitor.
The values of the microenvironment-specific parameter M were estimated by an analysis
of data from the Denver Study. During this study, each of approximately 450 subjects carried a
personal exposure monitor (PEM) for two 24-hour periods. Each PEM measured CO
concentration continuously. The PEM readings were averaged by exposure event such that each
event was associated with a single microenvironment and a single hour. Event durations ranged
from one minute to one hour. The microenvironment assigned to each PEM reading was
determined from entries made in a real-time diary carried by the subject.
Equation 1 estimates the outdoor CO concentration associated with a particular
microenvironment m, even when the microenvironment is an indoor location. Few of the
outdoor PEM values reported by the Denver study could be reliably associated with particular
indoor microenvironments. Consequently, a simplified procedure was employed for estimating
24
-------
the values of M, assuming that the mean of the indoor PEM values for each indoor
microenvironment was approximately equal to the mean of the outdoor concentration associated
with that indoor microenvironment.
Table 3-2. Estimated Values of the Parameter M in Equation 1
Microenvironment M
Indoors Residence 1.034
Indoors Auto service station 2.970
Indoors Health care facility, School, Church, 0.989
Manufacturing facility
Indoors Other locations 1.213
Outdoors Near road, Bicycle, Motorcycle 1.607
Outdoor Indoor parking garage, Outdoor parking garage, 2.970
Outdoor parking lot, Outdoor service station
Outdoors Other locations 1.436
Vehicle Automobile, Truck, Bus, Train, Subway, Other 3.020
vehicle
Source: Johnson et al., 2000.
3.3.5. Air Quality Adjustment to Meet Standards
There are many possible ways to create characterizations of air quality to represent
scenarios "just meeting" the current and any potential alternative CO standards. Previous
reviews have used a proportional adjustment method which decreased CO levels on all days by
the same percentage at all monitors in a given area. The percentage amount of adjustment is just
enough so that neither the 1-hr nor the 8-hr levels of the standards under consideration are
exceeded. Generally, the amount of adjustment required to just meet the 1-hr and 8-hr levels will
not be the same, so, in practice, this procedure brings the design value for one of the two
averaging times to be equal to the level of the corresponding standard, while the design value of
the other averaging time would be reduced to a level below the standard. In this review, we will
again evaluate the appropriateness of the proportional adjustment approach by comparing it with
historical changes in distributions of CO concentrations in selected locations. In this review, the
25
-------
required adjustment will result in an increase of CO levels to simulate just meeting the current
standards.
3.3.6. Indoor Sources
The indoor sources of CO that will be modeled are emissions from gas stoves for cooking
and the contribution of passive smoking indoors ("environmental tobacco smoke," or ETS). A
review of the scientific literature for other indoor sources (e.g., kerosene space heaters, gas space
heaters, wood stoves, fireplaces, and attached garages) will be conducted to ascertain whether
emissions from these sources can be adequately characterized by available data. As in prior
assessments, we plan to present CO exposure and dose both including and excluding indoor
sources. The primary focus for the CO NAAQS review is on the ambient contribution to
exposures and dose.
The resulting exposure analysis will provide estimates of CO exposure and their
associated COHb levels for population living in the selected urban areas. It is recognized that
there are significant uncertainties associated with the resulting exposure and dose estimates, and
these uncertainties must be taken into account in assessing the utility of the exposure analysis.
The next section discusses the planned approach for assessing these uncertainties
3.4. Characterization of Uncertainty and Variability
An important issue associated with any population exposure and/or dose assessment is
the characterization of uncertainty and variability.
Variability refers to the heterogeneity in a population or variable of interest that is
inherent and cannot be reduced through further research. This variability may be due to
differences in population (e.g., age distribution), population activities that affect exposure to CO
(e.g., proximity to roadways), levels of CO, and/or other factors that vary either within or across
urban areas.
Uncertainty refers to the lack of knowledge regarding both the actual values of model
input variables (parameter uncertainty) and the physical systems or relationships (model
uncertainty - e.g., the relationship between ambient CO concentrations and CO concentrations
measured at fixed site monitors). In any exposure analysis, uncertainty is, ideally, reduced to
the maximum extent possible, through improved measurement of key parameters and ongoing
26
-------
model refinement. However, significant uncertainty often remains and emphasis is then placed
on characterizing the nature of that uncertainty and its impact on exposure and dose estimates.
The characterization of uncertainty can include both qualitative and quantitative analyses, the
latter requiring more detailed information. The goal for addressing variability is to incorporate
the sources of variability into the model to ensure that the estimates of exposure and dose reflect
the variability of exposure and dose across the study population. Our approach to variability is
outlined in section 3.4.1.
The planned approach for evaluating uncertainty is adapted from guidelines outlining
how to conduct a qualitative uncertainty characterization for exposure assessment (WHO, 2008).
First, the key sources of the assessment that contribute to uncertainty are identified, and the
rationale for why they are included is discussed. Second, a qualitative characterization for the
types and components of uncertainty is carried out using sensitivity analysis. This results in a
summary describing, for each source of uncertainty, an indication of the potential bias direction,
and an assignment of the uncertainty to low, medium, and high categories. Third, a limited
quantitative assessment of uncertainty is performed for selected sources of uncertainty. Our
qualitative and quantitative approaches to characterizing uncertainty are addressed in sections
3.4.2 and 3.4.3.
3.4.1. Addressing Variability
APEX has been designed to enable incorporation of variability of almost all of the input
data and parameters, including the physiological parameters which are important input
parameters to the CFK equation. As a result, APEX addresses much of the variability in the
exposure and dose estimates resulting from the variability of the factors affecting human
exposure and dose. The following model inputs and parameters have probability distributions
representing variability:
• Population - Random samples from Census tracts, by age, gender, race
• Activity patterns - Stratified samples from CHAD
• Commuting - Random samples from Census tracts
• Employment - Random samples from Census tracts, by age and gender
• Ambient pollutant concentrations (spatial and temporal variability)
• Ambient meteorological data (spatial and temporal variability)
27
-------
• Physiology relevant to estimating alveolar ventilation rate: body mass, resting metabolic
rate, maximum level of sustained metabolic activity, oxygen uptake per unit of energy
expended, and metabolic equivalents by activity (METS)
• Physiology relevant to estimating COHb levels: blood volume, lung diffusivity,
endogenous CO production rate, and amount of Hb in the blood
• Coefficients of regression equations for ventilation rates
• Coefficients of regression equations for resting metabolic rates
• Coefficients of regression equations for body surface area
• Coefficients of regression equations for height
• Air exchange rates
• Decay and deposition rates
• Penetration factors
• Proximity factors
• Volumes of microenvironments
• Indoor source emission rates
• Air conditioning prevalence rates
3.4.2. Uncertainty Characterization - Qualitative Assessment
We plan to include a qualitative discussion of uncertainty in the exposure and dose
assessment, starting with identification and description of key sources of uncertainty, and a list of
secondary sources of uncertainty. This will be followed with an analysis of the sensitivity of the
model output (estimated distributions of exposure and dose) to each of the individual factors
(input data and parameters) contributing to uncertainty. A local sensitivity analysis will be
performed, varying the one at a time factors by five percent of their nominal values and running
APEX with all other parameters at their nominal values. Since most of the inputs to APEX are
specified as parametric distributions (of variability), it is the parameters of these distributions
that will be varied. In a few cases, correlation between parameters may exist where pairs of
parameters may have to be simultaneously varied if the value of one parameter constrains the
other. The sensitivity and elasticity7 will be calculated and tornado graphs will be used to
present the results of this sensitivity analysis. These graphs are particularly useful in illustrating
which factors have a potentially higher impact on the exposure and dose results and how all the
factors rank as to influencing those results and in which direction (positively or negatively).
This local analysis will be supplemented with a global sensitivity analysis, where the
sensitivity involves the study of the exposure model behavior over the entire range of exposure
See Appendix A for definitions of Sensitivity and Elasticity.
28
-------
parameter variation and investigating their impact on the overall result. We will calculate the
sensitivity score8 for model inputs for which we can identify ranges of potential variation. The
advantage of the sensitivity score is that it differentiates between precise (well-known) inputs
with high sensitivity and imprecise inputs with high sensitivity: this measure will be larger if the
input is less precisely known (keeping sensitivity the same). These are the inputs with the
potential for large impacts on model output uncertainty. This will allow for their ranking in
order to prioritize our focus on the most important ones.
These analyses of model sensitivity will support a qualitative discussion of uncertainties
and a qualitative assessment of the particular sources of uncertainly in terms of their potential
impact on exposure and dose levels using "high," "medium," and "low" designations.
3.4.3. Uncertainty Characterization - Quantitative Analysis
The primary difficulty in performing quantitative uncertainty analysis is how to
appropriately characterize the uncertainties of the model inputs and formulation when faced with
limited data. Information about the variability of model inputs or the variability and uncertainty
combined is often available, but it is usually difficult to estimate the uncertainty separately from
the variability.
Based on previous analyses of uncertainties of exposure modeling (Langstaff, 2007), we
expect the following to be influential sources of uncertainty associated with modeling CO
population exposure and dose:
• In representing the significant spatial and temporal gradients in ambient CO
concentrations relative to fixed-site CO concentrations, in particular, CO concentrations
near roadways.
• In portraying behavior (activity patterns and energy expenditures) related to CO exposure
and dose (e.g., amount of time spent in high CO microenvironments, outdoor activities).
There are not much activity data in CHAD that are relatively recent and based on surveys
in Denver and Los Angeles. Therefore the activity data used in this modeling effort may
not be representative of the cities and time periods to be modeled.
See Appendix A.
29
-------
• In estimating air exchange rates indoors and in vehicles. AERs are the most important
determinant in the relationship between outdoor concentrations and the concentrations in
indoor and in-vehicle microenvironments.
The sensitivity analyses described above will be carried out for each of these sources of
uncertainty. The algorithm for estimating outdoor concentrations, described in section 3.3, will
be evaluated by comparing its predictions with monitored CO concentrations. In the absence of
the data required to support a comprehensive 2-dimensional probabilistic uncertainty analysis
that treats both uncertainty and variability of all of the model inputs, a limited 2-dimensional
probabilistic analysis will be conducted to support quantitative characterization of uncertainty.
For the factors judged to have a major impact on the overall uncertainty of the estimated
exposure and dose, a literature search will be conducted to identify ranges of uncertainty for each
of the factors and the limited quantitative probabilistic analysis will be carried out for these
model inputs.
30
-------
4. SCHEDULE OF KEY MILESTONES
Table 4-1 lists the key milestones for the Risk and Exposure Assessment (REA) that are
planned as part of the current CO NAAQS review. Consultation with the CAS AC CO Panel is
scheduled for May 12-13, 2009 to obtain review of the first draft Integrated Science Assessment
(US EPA, 2009) and to obtain input on the plans to conduct quantitative assessments. EPA staff
will then proceed to develop estimates of CO exposures and resulting doses (i.e., COHb levels)
for the population with cardiovascular disease in two urban study areas associated with CO
levels representing recent air quality and air quality adjusted to simulate just meeting the current
CO NAAQS, and any potential alternative standards under consideration. These estimates and
the methodologies used will be presented in the first draft CO REA. CASAC and public
comments on this plan will be taken into consideration in the development of the second draft
REA, the preparation of which will coincide and draw from the second draft ISA. The first draft
REA is scheduled to be released for CASAC and public review on October 29, 2009. EPA will
receive comments on this draft document from the CASAC and the general public at a meeting
planned for November 2009. The second draft REA will draw on the final ISA and will reflect
consideration of CASAC and public comments on the first draft REA. The second draft REA
will include assessments for just meeting potential alternative standards. We plan to release the
second draft REA in February 2010 for review by CASAC and the general public at a meeting
that is planned for March 2010. Staff will consider these review comments and prepare a final
REA, to be completed by May 28, 2010. The final REA will reflect consideration of CASAC
and public comments on the second draft REA. The final ISA and final REA will inform the
policy assessment and rulemaking steps that will lead to a final decision of the CO NAAQS.
This schedule is based on a court-ordered schedule that governs the completion of the review
(See Communities/or a Better Environment v. EPA, No. 07-3678, N.D. Cal., May 5, 2008),
which requires EPA to sign proposed and final rules by October 28, 2010 and May 13, 2011,
respectively.
-------
Table 4-1. Key Milestones for the Risk and Exposure Assessments
Milestones
First draft CO ISA
Release CO REA Scope and Methods Plan
CASAC/public review and meeting on first draft CO ISA
CASAC consultation on CO REA Scope and Methods Plan
Release second draft CO ISA
Release first draft CO REA
CASAC/public review and meeting on second draft CO ISA and first
draft REA
Final CO ISA
Release second draft of the CO REA
CASAC/public review and meeting on second draft CO REA
Final CO REA
Date
March 12, 2009*
April 2009
May 12-13, 2009
May 12-13, 2009
September 2009
October 29, 2009*
November 2009
January 29, 2010*
February 20 10
March 20 10
May 28, 2010*
* Court-ordered deadline dates.
32
-------
5. REFERENCES
Adams KF, Koch G, Chatterjee B, Goldstein GM, O'Neil JJ, Bromberg PA, Sheps DS. (1988).
Acute elevation of blood carboxyhemoglobin to 6% impairs exercise performance and
aggravates symptoms in patients with ischemic heart disease. J Am Coll Cardiol 12: 900-
909.
Akland G, Hartwell TD, Johnson TR, Whitmore RW. (1985). Measuring Human Exposure to
Carbon Monoxide in Washington, D.C., and Denver, Colorado, during the Winter of
1982-1983, Environmental Science and Technology. 19:911-918.
Allred EN, Bleecker ER, Chaitman BR, Dahms TE, Gottlieb SO, Hackney JD, Pagano M,
Selvester RH, Walden SM, Warren J. (1989). Short-term effects of carbon monoxide
exposure on the exercise performance of subjects with coronary artery disease. N Engl J
Med321: 1426-1432.
Allred EN, Bleecker ER, Chaitman BR, Dahms TE, Gottlieb SO, Hackney JD, Pagano M,
Selvester RH, Walden SM, Warren J. (1991). Effects of carbon monoxide on myocardial
ischemia. Environ Health Perspect 91: 89-132.
Anderson EW, Andelman RJ, Strauch JM, Fortuin NJ, Knelson JH. (1973). Effect of low-level
carbon monoxide exposure on onset and duration of angina pectoris. A study in ten
patients with ischemic heart disease. Ann Intern Med 79: 46-50.
Coburn R, Forster R, Kane P. 1965. Consideration of the physiological variables that determine
the blood carboxyhemoglobin concentrations in man. J. CLin. Invest. 44:1899-1910.
Glen G, Smith L, Isaacs K, McCurdy T, Langstaff J. (2008). A new method of longitudinal
diary assembly for human exposure modeling. Journal of Exposure Science and
Environmental Epidemiology 18, 299-311.
Graham S E and McCurdy T. (2004). Developing meaningful cohorts for human exposure
models. J Expo Anal Environ Epidemiol. 14: 23-43.
Isaacs K, Glen G, McCurdy T, Smith L. (2008). Modeling energy expenditure and oxygen
consumption in human exposure models: accounting for fatigue and EPOC. Journal of
Exposure Science and Environmental Epidemiology 18, 289-298.
Johnson T. (1984). A Study of Personal Exposure to Carbon Monoxide in Denver, Colorado.
Report No. EPA-600/54-84-014, U. S. Environmental Protection Agency, Research
Triangle Park, North Carolina.
Johnson T, Capel J, McCoy M. (1996a). Estimation of Ozone Exposures Experienced by Urban
Residents Using a Probabilistic Version of NEM and 1990 Population Data. Prepared by
International Technology Air Quality Services for Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Contract no. 63-D-30094. April 1996. Available at:
http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_pr_td.html.
Johnson T, Capel J, McCoy M, Warnasch J. (1996b). Estimation of Ozone Exposures
Experienced by Outdoor Children in Nine Urban Areas Using a Probabilistic Version of
NEM. Prepared by International Technology Air Quality Services for Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency, Research
Triangle Park, NC. Contract no. 63-D-30094. April 1996. Available at:
http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_pr_td.html.
Johnson T, Mihlan, GLaPointe, J, Fletcher K, Capel J. (1999). Estimation of Carbon Monoxide
Exposures and Associated Carboxyhemoglobin levels in Denver Residents Using
33
-------
pNEM/CO (Version 2.0) prepared by ICF Kaiser Consulting Group for U.S. EPA, Office
of Air Quality Planning and Standards, under Contract No. 68-D6-0064, WANos. 1-19
and 2-24.
Johnson T, Mihlan G, LaPointe J, Fletcher K, Capel J. (2000). Estimation of Carbon Monoxide
Exposures and Associated Carboxyhemoglobin levels in Denver Residents Using
pNEM/CO (Version 2.1), prepared by ICF Kaiser Consulting Group for U.S. EPA, Office
of Air Quality Planning and Standards, under Contract No. 68-D6-0064, WANos. 1-19,
2-24, 2-30, and 3-3, Docket EPA-HQ-OAR-2008-0015-DRAFT-0009.
Kleinman M T, Davidson DM, Vandagriff RB, Caiozzo VJ, Whittenberger JL.(1989). Effects of
short-term exposure to carbon monoxide in subjects with coronary artery disease. Arch.
Environ. Health 44: 361-369.
Kleinman MT, Leaf DA, Kelly E, Caiozzo V, Osann K, O'Niell T. (1998). Urban angina in the
mountains: effects of carbon monoxide and mild hypoxemia on subjects with chronic
stable angina. Arch. Environ.Health 53: 388-397.
Langstaff JE. (2007). OAQPS Staff Memorandum to Ozone NAAQS Review Docket (EPA-HQ-
OAR-2005-0172). Subject: Analysis of Uncertainty in Ozone Population Exposure
Modeling. [January 31, 2007]. Available at:
http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_cr_td.html
McClellan RO, Chairman, Clean Air Scientific Advisory Committee. (1991). Letter to William
K. Reilly, EPA Administrator, July 17, 1991.
McClellan RO, Chairman, Clean Air Scientific Advisory Committee. (1992). Letter to William
K. Reilly, EPA Administrator, August 1, 1992.
McCurdy T. (2000). "Conceptual basis for multi-route intake dose modeling using an energy
expenditure approach." J. Exposure Anal. Environ. Epidemiol. 10:1-12.
McCurdy T, Glen G, Smith L, Lakkadi Y. (2000). The National Exposure Research Laboratory's
Consolidated Human Activity Database. J Expo Anal Environ Epidemiol. 10: 566-578.
Morgan and Henrion. (1990). Uncertainty: A Guide to Dealing with Uncertainty in Qualitative
Risk and Policy Analysis. Cambridge University Press.
National Research Council. (2003). Managing Carbon Monoxide Pollution in Meteorological
and Topographical Problem Areas. Washington, D.C. The National Academies Press.
Richmond H and Johnson T. (2000). Estimating COHb levels in exposed populations.
Appendix C in Johnson et al. (2000), Estimation of Carbon Monoxide Exposures and
Associated Carboxyhemoglobin levels in Denver Residents Using pNEM/CO (Version
2.1).
SAIC. (2001). Memo to Harvey Richmmond, EPA, Technical Peer Review (including reviewers
comments) of "Estimation of Carbon Monoxide Exposures and Associated
Carboxyhemoglobin Levels for Residents of Denver and Los Angeles Using pNEM/CO
(version 2.1)", Docket EPA-HQ-OAR-2008-0015.
Sheps DS, Adams KF Jr., Bromberg PA, Goldstein GM, O'Neil JJ, Horstman D, Koch G. (1987).
Lack of effect of low levels of Carboxyhemoglobin on cardiovascular function in patients
with ischemic heart disease. Arch.Environ. Health 42: 108-116.
US EPA (1979a). Air Quality Criteria for Carbon Monoxide. Research Triangle Park, NC: Office
of Health and Environmental Assessment, Environmental Criteria and Assessment
Office; report no. EPA-600/8-79-022.
34
-------
US EPA (1979b). Assessment of Adverse Health Effects from Carbon Monoxide and
Implications for Possible Modifications of the Standard. Office of Air Quality Planning
and Standards. Research Triangle Park, NC.
US EPA (1984a). Revised Evaluation of Health Effects Associated with Carbon Monoxide
Exposure: An Addendum to the 1979 EPA Air Quality Criteria Document for Carbon
Monoxide. Office of Health and Environmental Assessment, Environmental Criteria and
Assessment Office; report ro. EPA-600/8-83-033F. Research Triangle Park, NC
US EPA (1984b). Review of the NAAQS for Carbon Monoxide: Reassessment of Scientific and
Technical Information. Office of Air Quality Planning and Standards, report no. EPA-
450/584-904. Research Triangle Park. NC.
US EPA. (1991). Air Quality Criteria for Carbon Monoxide. Research Triangle Park, NC:
Office of Health and Environmental Assessment, Environmental Criteria and Assessment
Office; report no. EPA/600/8-90/045F.
US EPA (1992). Review of the National Ambient Air Quality Standards for Carbon Monoxide:
Assessment of Scientific and Technical Information. Office of Air Quality Planning and
Standards Staff Paper, report no.EPA/452/R-92-004.
US EPA (1999). Total Risk Integrated Methodology - TRIM.Expo Technical Support
Document, External Review Draft, November 1999. Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC. Report
no. EPA-453/D-99-001. Available:
http://www.epa.gov/ttn/fera/trim_fate.html#1999historical.
US EPA (2000). Air Quality Criteria for Carbon Monoxide. National Center for Environmental
Assessment, Office of Research and Development, U.S. Environmental Protection
Agency, Research Triangle Park, NC 27711; report no. EPA/600/P-99/001F. June 2000.
Available: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=l8163.
US EPA (2003). Consolidated Human Activities Database (CHAD) Users Guide. Database and
documentation available at: http://www.epa.gov/chadnetl/
US EPA (2007). Ozone Population Exposure Analysis for Selected Urban Areas. Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency, Research
Triangle Park, NC. report no. EPA-452/R-07-010. July 2007. Available:
http: //www. epa. gov/ttn/naaqs/standards/ozone/s_o3 _cr_td. html.
US EPA (2008a). Draft Integrated Review Plan for the National Ambient Air Quality Standards
for Carbon Monoxide. U.S. Environmental Protection Agency, Research Triangle Park,
NC, report no. EPA/452D-08-001. Availiable at:
http://www.epa.gov/ttn/naaqs/standards/co/s_co_cr_pd.html.
US EPA (2008b). Integrated Review Plan for the National Ambient Air Quality Standards for
Carbon Monoxide. U.S. Environmental Protection Agency, Research Triangle Park, NC,
report no. EPA/452R-08-005. Availiable at:
http://www.epa.gov/ttn/naaqs/standards/co/s_co_cr_pd.html.
US EPA (2008c). Total Risk Integrated Methodology (TRIM) - Air Pollutants Exposure Model
Documentation (TRIM.Expo / APEX, Version 4) Volume I: User's Guide. Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency, Research
Triangle Park, NC. report no. EPA-452/B-08-00la. Available:
http ://www. epa.gov/ttn/fera/human_apex.html.
US EPA (2008d). Total Risk Integrated Methodology (TRIM) - Air Pollutants Exposure Model
Documentation (TRIM.Expo / APEX, Version 4) Volume II: Technical Support
35
-------
Document. Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, NC. report no. EPA-452/B-08-001b. Available:
http ://www. epa.gov/ttn/fera/human_apex.html.
US EPA (2009). Draft Integrated Science Assessment for Carbon Monoxide - Health Criteria
(Draft Report). U.S. Environmental Protection Agency, Research Triangle Park, NC,
report no. EPA/600/R-08/071. Available at:
http://www.epa.gov/ttn/naaqs/standards/co/s_co_cr_isa.html.
WHO (2008). Harmonization Project Document No. 6. Part 1: Guidance document on
characterizing and communicating uncertainty in exposure assessment. Available at:
http://www.who.int/ipcs/methods/harmonization/areas/exposure/en/.
Wilson AL, Colome SD, Tian Y. (1993). "California Residential Indoor Air Quality Study,
Volume 1, Methodology and Descriptive Statistics, Appendix." Integrated
Environmental Services, Irvine, California.
36
-------
United States Office of Air Quality Planning and Standards Publication No. EPA 452/R-09-004
Environmental Protection Air Quality Strategies and Standards Division April 2009
Agency Research Triangle Park, NC
Postal information in this section where appropriate.
------- |