&EPA
United! S*K
Enviraimerilfll PiutmBmi
Agamy
Policy Assessment for the Review of the
Carbon Monoxide National Ambient Air
Quality Standards
External Review Draft
-------
EPA-452/P-10-005
March 2010
Policy Assessment for the Review of the Carbon Monoxide
National Ambient Air Quality Standards
External Review Draft
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Research Triangle Park, North Carolina
-------
DISCLAIMER
This document has been reviewed by the Office of Air Quality Planning and Standards (OAQPS), U.S.
Environmental Protection Agency (EPA), and approved for publication. This OAQPS Staff Paper contains the
conclusions and recommendations of the staff of the OAQPS and does not necessarily reflect the views of the
Agency. Mention of trade names or commercial products is not intended to constitute endorsement or
recommendation for use.
-------
TABLE OF CONTENTS
LIST OF FIGURES ii
LIST OF TABLES ii
1 INTRODUCTION 1-1
1.1 PURPOSE 1-1
1.2 BACKGROUND 1-2
1.2.1 Legislative requirements 1-2
1.2.2 Previous Reviews 1-3
1.2.3 The Current Review 1-5
1.3 CURRENT AIR QUALITY 1-6
1.3.1 Sources to ambient air 1-6
1.3.2 Ambient Monitoring Network 1-7
1.3.3 Ambient Monitoring Concentrations 1-9
1.4 GENERAL APPROACH AND ORGANIZATION OF THE DOCUMENT .1-11
REFERENCES 1-12
2 REVIEW OF THE PRIMARY STANDARDS FOR CARBON MONOXIDE 2-1
2.1 APPROACH 2-1
2.1.1 Approach Used in the Previous Review 2-2
2.1.2 Approach for the Current Review 2-6
2.2 ADEQUACY OF THE CURRENT STANDARD 2-7
2.2.1 Evidence-based Considerations 2-8
2.2.2 Exposure/Risk-based Considerations 2-35
2.2.3 Initial Staff Conclusions on Adequacy of the Current Standards 2-45
2.3 CONSIDERATION OF ALTERNATIVE STANDARDS 2-46
2.3.1 Indicator 2-46
2.3.2 Averaging Time 2-47
2.3.3 Form and Level 2-48
2.3.3.1 Potential Alternative Forms 2-50
2.3.3.2 Potential Alternative Levels 2-50
2.3.4 Initial Staff Conclusions on Alternative Standards 2-56
2.4 KEY UNCERTAINTIES AND AREAS FOR FUTURE RESEARCH
AND DATA COLLECTION 2-57
REFERENCES 2-59
3 CONSIDERATION OF A SECONDARY STANDARD FOR CO 3-1
3.1 CONSIDERATION IN PREVIOUS REVIEWS 3-1
March, 2010 i Draft - Do Not Quote or Cite
-------
3.2 CONSIDERATION OF EVIDENCE AVAILABLE IN THE CURRENT
REVIEW 3-1
3.3 STAFF CONCLUSIONS FOR THE CURRENT REVIEW 3-3
REFERENCES 3-5
LIST OF FIGURES
Figure 1-1 Trends in anthropogenic CO emissions in the U.S. by source category for
1990 and 1996-2002. Source: U.S. EPA (2008c) 1-7
Figure 1-2 Trends in CO concentration (second maximum 8-hour average) in the U.S.,
1990-2008 1-10
Figure 1-3 Trends in CO concentration (second maximum 1-hour average) in the U.S.,
1990-2008 1-11
Figure 2-1 A conceptual model of CO source-to-health outcome pathway 2-15
Figure 2-2 Components of CO source-to-outcome conceptual model measured in
controlled human exposure and epidemiological studies discussed in this
document 2-15
LIST OF TABLES
Table 2-1 Information from U.S. epidemiological studies of hospital admissions for
ischemic heart disease (ISA, Figure 5-2), cardiovascular disease
(ISA, Figure 5-5) and congestive heart failure (ISA, Figure 5-4) and
on air quality during the study periods 2-13
Table 2-2 Carboxyhemoglobin levels and reported effects in CAD patients and healthy
adults 2-24
Tabble 2-3 Air quality information for key U.S. epidemiological studies of hospital
admissions for coronary heart disease (ISA, Figure 5-2) 2-29
Table 2-4 Air quality information for key U.S. epidemiological studies of hospital
admissions for CVD (ISA, Figure 5-5) 2-30
Table 2-5 Portion of simulated at-risk population with maximum end-of-hour COHb
levels at or above the indicated level under air quality conditions simulated to
just meet the current standards and under "as is" (2006) conditions. COHb
levels reflect endogenous levels and ambient CO exposures 2-41
Table 2-6 Portion of simulated at-risk population with maximum end-of-hour COHb
levels below the indicated COHb level under alternative levels for the 8-hour
standard for the current form and a 99th percentile daily maximum form
2-52
March, 2010
Draft - Do Not Quote or Cite
-------
Table 2-7 Portion of simulated at-risk population with maximum end-of-hour COHb
levels below the indicated COHb level under alternative levels for the 1-hour
standard for the current form and a 99th percentile daily maximum
form 2-55
March, 2010 iii Draft - Do Not Quote or Cite
-------
1 1 INTRODUCTION
2 1.1 PURPOSE
3 The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
4 the carbon monoxide (CO) national ambient air quality standards (NAAQS). The overall plan
5 and schedule for this review were presented in the Plan for Review of the National Ambient Air
6 Quality Standards for Carbon Monoxide (IRP; USEPA, 2008b). * The IRP identified key policy-
7 relevant issues to be addressed in this review as a series of questions that frame our consideration
8 of whether the current NAAQS for CO should be retained or revised.
9 This Policy Assessment (PA), prepared by staff in EPA's Office of Air Quality Planning
10 and Standards (OAQPS), is intended to help "bridge the gap" between the relevant scientific
11 information and assessments and the judgments required of the EPA Administrator in
12 determining whether, and if so, how it is appropriate to revise the NAAQS for CO. This draft
13 PA presents factors relevant to EPA's review of the current primary (health-based) standards and
14 consideration of a secondary (welfare-based) standard. It focuses on both evidence- and risk-
15 based information in evaluating the adequacy of the current CO NAAQS and in identifying
16 potential alternative standards for consideration. In this draft PA, we consider the scientific and
17 technical information available in this review as assessed in the Integrated Science Assessment
18 for Carbon Monoxide (henceforth referred to as the ISA, USEPA, 2010a), prepared by EPA's
19 National Center for Environmental Assessment (NCEA), and the Risk and Exposure Assessment
20 to Support the Review of the Carbon Monoxide Primary National Ambient Air Quality
21 Standards: Second External Review Draft (henceforth referred to as the draft Risk and Exposure
22 Assessment document or draft REA, USEPA, 201 Ob). In so doing, we focus on information that
23 is most pertinent to evaluating the basic elements of NAAQS: indicator,2 averaging time, form,3
24 and level. These elements, which together serve to define each standard, must be considered
25 collectively in evaluating the health and welfare protection afforded by the PM standards.
26 While this draft PA should be of use to all parties interested in the CO NAAQS review, it
27 is written with an expectation that the reader has some familiarity with the technical discussions
28 contained in the ISA (USEPA, 201 Oa) and the second draft REA (USEPA, 201 Ob).
1 As described below in section 1.2.3, the schedule for this review is governed by a court order.
2 The "indicator" of a standard defines the chemical species or mixture that is to be measured in
determining whether an area attains the standard.
3 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.
March 2010 1-1 Draft - Do Not Quote or Cite
-------
1 1.2 BACKGROUND
2 1.2.1 Legislative requirements
3 Two sections of the Clean Air Act (Act) govern the establishment and revision of the
4 NAAQS. Section 108 (42 U.S.C. 7408) directs the Administrator to identify certain pollutants
5 that meet specified criteria, including emissions which "may reasonably be anticipated to
6 endanger public health and welfare" and whose presence "in the ambient air results from
7 numerous or diverse mobile or stationary sources" and to issue air quality criteria for them. Air
8 quality criteria are to "accurately reflect the latest scientific knowledge useful in indicating the
9 kind and extent of identifiable effects on public health or welfare which may be expected from
10 the presence of [a] pollutant in ambient air . . ."
11 Section 109 (42 U.S.C. 7409) directs the Administrator to propose and promulgate
12 "primary" and "secondary" NAAQS for pollutants listed under section 108. Section 109(b)(l)
13 defines a primary standard as one "the attainment and maintenance of which in the judgment of
14 the Administrator, based on such criteria and allowing an adequate margin of safety, are requisite
15 to protect the public health."4 A secondary standard, as defined in Section 109(b)(2), must
16 "specify a level of air quality the attainment and maintenance of which, in the judgment of the
17 Administrator, based on such criteria, is requisite to protect the public welfare from any known
18 or anticipated adverse effects associated with the presence of [the] pollutant in the ambient air."5
19 The requirement that primary standards include an adequate margin of safety was
20 intended to address uncertainties associated with inconclusive scientific and technical
21 information available at the time of standard setting. It was also intended to provide a reasonable
22 degree of protection against hazards that research has not yet identified. Lead Industries
23 Association v. EPA, 647 F.2d 1130, 1154 (D.C. Cir 1980), cert, denied, 449 U.S. 1042 (1980);
24 American Petroleum Institute v. Costle, 665 F.2d 1176, 1186 (D.C. Cir. 1981), cert, denied, 455
25 U.S. 1034 (1982). Both kinds of uncertainties are components of the risk associated with
26 pollution at levels below those at which human health effects can be said to occur with
27 reasonable scientific certainty. Thus, in selecting primary standards that include an adequate
28 margin of safety, the Administrator is seeking not only to prevent pollution levels that have been
4 The legislative history of section 109 indicates that a primary standard is to be set at the "maximum
permissible ambient air level.. .which will protect the health of any [sensitive] group of the population," and that for
this purpose "reference should be made to a representative sample of persons comprising the sensitive group rather
than a single person in such group." S. Rep. No.91-1196, 91st Cong., Sess. 10 (1970)
5 Welfare effects as defined in section 302(h) (42U.S.C. 7602(h)) include, but are not limited to, "effects in
soils, water, crops, vegetation, man-made materials, animals, wildlife, weather, visibility and climate, damage to and
deterioration of property, and hazards to transportation, as well as effect on economic values on personal comfort
and well-being."
March 2010 1-2 Draft - Do Not Quote or Cite
-------
1 demonstrated to be harmful but also to prevent lower pollutant levels that may pose an
2 unacceptable risk of harm, even if the risk is not precisely identified as to nature or degree.
3 In selecting a margin of safety, the EPA considers such factors as the nature and severity
4 of the health effects involved, the size of the sensitive population(s) at risk, and the kind and
5 degree of the uncertainties that must be addressed. The selection of any particular approach to
6 providing an adequate margin of safety is a policy choice left specifically to the Administrator's
7 judgment. Lead Industries Association v. EPA, supra, 647 F.2d at 1161-62.
8 In setting standards that are "requisite" to protect public health and welfare, as provided
9 in section 109(b), EPA's task is to establish standards that are neither more nor less stringent
10 than necessary for these purposes. In so doing, EPA may not consider the costs of implementing
11 the standards. See generally Whitman v. American Trucking Associations, 531 U.S. 457, 465-
12 472, 475-76 (2001).
13 Section 109(b)(l) of the Act requires that "not later than December 31, 1980, and at 5-
14 year intervals thereafter, the Administrator shall complete a thorough review of the criteria
15 published under section 108 and the national ambient air quality standards . . . and shall make
16 such revisions in such criteria and standards and promulgate such new standards as may be
17 appropriate . . . ." Section 109(d)(2) requires that an independent scientific review committee
18 "shall complete a review of the criteria . . . and the national primary and secondary ambient air
19 quality standards . . . and shall recommend to the Administrator any new . . . standards and
20 revisions of existing criteria and standards as may be appropriate . . . ." Since the early 1980's,
21 this independent review function has been performed by the Clean Air Scientific Advisory
22 Committee (CASAC) of EPA's Science Advisory Board.
23 1.2.2 Previous Reviews
24 EPA initially established NAAQS for CO, under section 109 of the Act, on April 30,
25 1971. The primary standards were established to protect against the occurrence of
26 carboxyhemoglobin levels in human blood associated with health effects of concern. The
27 standards were set at 9 parts per million (ppm), as an 8-hour average and 35 ppm, as a 1-hour
28 average, neither to be exceeded more than once per year (36 FR 8186). In the 1971 decision, the
29 Administrator judged that attainment of these standards would provide protection of public
30 health with an adequate margin of safety and would also protect against known and anticipated
31 adverse effects on public welfare, and accordingly setting the secondary (welfare-based)
32 standards identical to the primary (health-based) standards.
33 In 1985, EPA concluded its first periodic review of the criteria and standards for CO (50
34 FR 37484). In that review, EPA updated the scientific criteria upon which the initial CO
35 standards were based through the publication of the 1979 Air Quality Criteria Document for
March 2010 1-3 Draft - Do Not Quote or Cite
-------
1 Carbon Monoxide (AQCD; USEPA, 1979a) and prepared a Staff Paper (USEPA, 1979b), which,
2 along with the 1979 AQCD, served as the basis for the development of the notice of proposed
3 rulemaking which was published on August 18, 1980 (45 FR 55066). Delays due to
4 uncertainties regarding the scientific basis for the final decision resulted in EPA's announcing a
5 second public comment period (47 FR 26407). Following substantial reexamination of the
6 scientific data, EPA prepared an Addendum to the 1979 AQCD (USEPA, 1984a) and an updated
7 Staff Paper (USEPA, 1984b). Following review by CAS AC (McClellan, 1991, 1992), EPA
8 announced its decision not to revise the existing primary standard and to revoke the secondary
9 standard for CO on September 13, 1985, due to a lack of evidence of direct effects on public
10 welfare at ambient concentrations (50 FR 37484).6
11 On August 1, 1994, EPA concluded its second periodic review of the criteria and
12 standards for CO by deciding that revisions to the CO NAAQS were not warranted at that time
13 (59 FR 38906). This decision reflected EPA's review of relevant scientific information
14 assembled since the last review, as contained in the 1991 AQCD (USEPA, 1991) and the 1992
15 Staff Paper (USEPA, 1992). Thus, the primary standards were retained at 9 parts per million
16 (ppm), 8-hr average and 35 ppm, 1-hr average, neither to be exceeded more than once per year
17 (59 FR 38906).
18 The next periodic review was initiated in 1997, and a workshop was held in September
19 1998 to review and discuss material to be contained in the AQCD. On June 9, 1999, CAS AC
20 held a public meeting to review the first draft AQCD and to provide a consultation on a draft
21 exposure analysis methodology document. Comments from CASAC Panel members and the
22 public on the AQCD were considered in a second draft AQCD, which was reviewed at a CASAC
23 meeting, held on November 18, 1999. After revision of the second draft AQCD, the final 2000
24 AQCD (U.S. EPA, 2000) was released in August 2000. EPA put this review on hold when
25 Congress requested that the National Research Council (NRC) review the impact of meteorology
26 and topography on ambient CO concentrations in high altitude and extreme cold regions of the
27 U.S. In response, the NRC convened the Committee on Carbon Monoxide Episodes in
28 Meteorological and Topographical Problem Areas, which focused on Fairbanks, Alaska as a
29 case-study. A final report, "Managing Carbon Monoxide Pollution in Meteorological and
30 Topographical Problem Areas," was published in 2003 (NRC, 2003) and offered a wide range of
31 recommendations regarding management of CO air pollution, cold start emissions standards,
32 oxygenated fuels, and CO monitoring. Following completion of this NRC report, EPA did not
33 conduct rulemaking to complete the review.
6 EPA concluded in 1985 that "no standards appear to be requisite to protect the public welfare from any
known or anticipated adverse effects from ambient CO exposures" (50 FR 37494).
March 2010 1-4 Draft - Do Not Quote or Cite
-------
1 1.2.3 The Current Review
2 On September 13, 2007, EPA issued a call for information from the public (72 FR 52369)
3 requesting the submission of recent scientific information on specified topics. A workshop was
4 held on January 28-29, 2008 (73 FR 2490) to discuss policy-relevant scientific and technical
5 information to inform EPA's planning for the CO NAAQS review. Following the workshop, a
6 draft IRP (USEPA, 2008a) was made available in March 2008 for public comment and was
7 discussed by the CASAC via a publicly accessible teleconference consultation on April 8, 2008
8 (73 FR 12998). EPA made the final IRP available in August 2008 (USEPA, 2008b).
9 In preparing the CO ISA, NCEA held an authors' teleconference in November 2008 with
10 invited scientific experts to discuss preliminary draft materials prepared as part of the ongoing
11 development of the CO ISA and its supplementary Annexes. The first draft ISA (USEPA,
12 2009a) for CO was made available for public review on March 12, 2009 (74 FR 10734) and
13 reviewed by CASAC at a meeting held on May 12-13, 2009 (74 FR 15265). A second draft ISA
14 (USEPA, 2009b) was released for CASAC and public review on September 23, 2009 (74 FR
15 48536), and it was reviewed by CASAC at a meeting held on November 16-17, 2009 (74 FR
16 54042). The final ISA was released in January 2010 (USEPA, 201 Oa).
17 In preparing the REA, OAQPS first released a draft planning document, the draft Scope
18 and Methods Plan, for consultation with CASAC and public review at the CASAC meeting held
19 on May 12-13, 2009. Taking into consideration CASAC and public comments on the draft Plan,
20 OAQPS staff conducted a risk/exposure assessment, and released for CASAC review and public
21 comment a first draft REA (USEPA, 2009c), which was reviewed at the CASAC meeting held
22 on November 16-17, 2009. A second draft REA was released for CASAC review and public
23 comment in February 2010, and will be reviewed at an upcoming CASAC meeting scheduled for
24 March 22-23, 2010. Drawing from information in the final CO ISA and the second draft REA,
25 this draft PA is now being released for CASAC review and public comment at the same meeting.
26 The schedule for completion of this review is governed by a court order resolving a
27 lawsuit filed in March 2003 by a group of plaintiffs representing two regional environmental
28 organizations.7 The court order that governs this review, entered by the court on November 14,
29 2008, provides that EPA will meet the following schedule in completing this review: a final
30 risk/exposure assessment is to be completed by May 28, 2010 and notices of proposed and final
31 rulemaking are to be signed by October 28, 2010 and May 13, 2011, respectively
7 The lawsuit alleged that EPA had failed to perform its mandatory duty, under section 109(d)(l), of
completing the current review within the period provided by statute. American Lung Association v. Whitman (No.
1:03CV00778, D.D.C. 2003).
March 2010 1-5 Draft - Do Not Quote or Cite
-------
1 1.3 CURRENT AIR QUALITY
2 This section provides a general overview of the current air quality conditions to provide
3 context for this consideration of the current standards for carbon monoxide. A more
4 comprehensive discussion of air quality information is provided in the ISA (ISA, sections 3.2
5 and 3.4), and a more detailed discussion of aspects particularly relevant to the exposure
6 assessment is provided in the second draft REA (REA, chapter 3).
7 1.3.1 Sources to ambient air
8 Carbon monoxide in ambient air is formed primarily by the incomplete combustion of
9 carbon-containing fuels and by photochemical reactions in the atmosphere. As a result of the
10 combustion conditions, CO emissions from large fossil-fueled power plants are typically very
11 low because optimized fuel consumption conditions make boiler combustion highly efficient. In
12 contrast, internal combustion engines used in many mobile sources have widely varying
13 operating conditions. Therefore, higher and more varying CO formation results from the
14 operation of these mobile sources (ISA, section 3.2). In the National Emissions Inventory for
15 2002, CO emissions from on-road vehicles accounted for approximately half of the total CO
16 emissions by individual source sectors in the U.S. (ISA, Figure 3-1.8 As with previous reviews
17 of the CO NAAQS, mobile sources continue to be a significant source sector for CO in ambient
18 air.
19 As discussed in the ISA, the spatial and temporal patterns of ambient CO concentrations
20 are heavily influenced by the patterns associated with mobile source emissions (ISA, chapter 3).
21 In metropolitan areas of the U.S., due to their relatively greater motor vehicle density, the
22 contribution from mobile sources has been estimated to be as high as 75% of all ambient CO
23 emissions in the 2002 National Emissions Inventory (ISA, p. 3-2). For example, emissions in
24 urban Denver County, Colorado originating from on-road mobile sources are estimated to be
25 about 71% of total CO emissions (ISA, section 3.2), compared with 20% for rural Garfield
26 County, Colorado9 (ISA, section 3, Figure 3-6). When considering all mobile sources
27 nationwide (non-road and on-road combined), the contribution to total ambient CO emissions
8 EPA compiles CO emissions estimates for the U.S. in the National Emissions Inventory (NEI). The 2002
NEI provides the most recent publicly available CO emissions estimates for the U.S. that meet EPA's data quality
assurance objectives. Estimates come from various sources and different data sources use different data collection
methods, most of which are based on engineering calculations and estimates rather than measurements. Although
these estimates are generated using well-established approaches, uncertainties are inherent in the emission factors
and models used to represent sources for which emissions have not been directly measured. Uncertainties vary by
source category, season and region (ISA, section 3.2.1).
9 The 2002 National Emissions Inventory estimate for on-road emissions in Garfield is 20,000 tons, and the
total emissions from all sources is estimated to be 98,831 (99K) tons. Thus, in this example the on-road vehicles
accounts for 20.2% of the total emissions (ISA, section 3, figure 3-6).
March 2010 1-6 Draft - Do Not Quote or Cite
-------
1 nationally is over 80%. Again using Denver County as an example, all mobile sources are
2 estimated to contribute about 98% of total CO emissions.
3 National-scale anthropogenic CO emissions have decreased 35% between 1990 and 2002
4 (Figure 1-1). Nearly all the national-level CO reductions since 1990 are the result of emission
5 reductions in on-road vehicles (ISA, Figure 3.-2).
6 Figure 1-1. Trends in anthropogenic CO emissions in the U.S. by source category for
7 1990 and 1996-2002. Source: U.S. EPA (2008c)
eo
a
160
140
o 120
4—l
I 1°°
§, 80
60
60
40
20
0
o
'co
*2
£
UJ
Fuel combustion
Other industrial
On-road vehicles
Nonroad vehicles and
*90
*96 '97 *98 '99 '00 '01 '02
Year
9 1.3.2 Ambient Monitoring Network
10 Ambient CO concentrations are measured by monitoring networks that are operated by
11 state and local monitoring agencies in the U.S., and are funded in part by the EPA. The main
12 network providing ambient data for use in comparison to the NAAQS is the State and Local Air
13 Monitoring Stations (SLAMS) network. CO monitors are typically sited to reflect one of the
14 following spatial scales:
15 • Microscale: Data represent concentrations within a 100 m radius of the monitor. For
16 CO, microscale monitors are sited 2 - 10 m from a roadway. Measurements are
17 intended to represent the near-road or street canyon environment.
18 • Middle scale: Data represent concentrations averaged over areas defined by 100 - 500
19 m radii. Measurements are intended to represent several city blocks.
March 2010
1-7
Draft - Do Not Quote or Cite
-------
1 • Neighborhood scale: Data represent concentrations averaged over areas defined by 0.5
2 - 4.0 km radii. Measurements are intended to represent extended portions of a city.
3 In addition to monitoring required for determining compliance with the NAAQS, the
4 EPA is currently in the process of implementing plans for a new network of multi-pollutant
5 stations called NCore that is intended to meet multiple monitoring objectives (ISA, p. 3-21). The
6 complete NCore network, required to be fully implemented by January 1, 2011, will consist of
7 approximately 63 urban and 20 rural stations and will include some existing SLAMS sites that
8 have been modified to include additional measurements. The majority of NCore stations will be
9 sited, however, to represent neighborhood, urban, and regional scales, consistent with the NCore
10 network design objective of representing exposure expected across urban and rural areas in
11 locations that are not dominated by local sources.
12 To promote uniform enforcement of the air quality standards set forth under the CAA,
13 EPA has established federal reference methods (FRMs) for ambient air sample collection and
14 analysis. Measurements for determinations of NAAQS compliance must be made with FRMs or
15 methods designated as equivalent (federal equivalent methods, FEMs). Given the levels of the
16 CO NAAQS (35 ppm, 1-hour; 9 ppm, 8-hour), a lower detectable limit (LDL) on the order of 1.0
17 ppm is well below the NAAQS levels and is therefore sufficient for demonstration of
18 compliance.10 More than 95% of monitors in use during 2005-2007 have LDL of 0.5 ppm (ISA,
19 Appendix A, Table A-8). With ambient CO levels now routinely near or below 1 ppm, however,
20 a large percentage of the measurements are below the LDL of conventional monitors,
21 contributing greater uncertainty in a larger portion of the distribution of monitoring data (ISA,
22 section 3.4.1). For example, more than half of the dataset of nationally reported hourly data for
23 2005-2007 analyzed in the ISA fell below the reported LDL of 0.5 ppm (ISA, p. 3-56). To
24 reduce the uncertainty at these lower concentrations, a new generation of ambient CO monitors
25 has been designed that provides improved sensitivity for measurements at or below the typical
26 ambient CO levels measured in most urban and all rural locations. The number of active
27 monitors employing such sensitive methods is increasing, primarily in association with the
28 implementation of the NCore network.n The extent to which these more sensitive monitors
29 become integrated into non-NCore SLAMS stations, however, will depend on the availability of
30 funding for states to replace existing legacy CO monitors as well as the possibility that
10 Among the 13 approved FRMs in use in the SLAMS network for which data were reported to EPA's Air
Quality System (AQS) between 2005 and 2009, nine are "legacy" methods with a federal method detection limit
(MDL) listed as 0.5 ppm.
11 For example, four approved FRMs are newer, more sensitive methods with a federal MDL of 0.02 ppm
and a growing body of ambient data from more sensitive CO instruments is becoming available. Testing performed
by EPA on several such CO monitors in 2005 and 2006 demonstrated MDLs of approximately 0.017 - 0.018 ppm
(17 - 18 ppb), slightly below the stated LDL of 0.02 - 0.04 ppm (ISA, section 3.4.1).
March 2010 1-8 Draft - Do Not Quote or Cite
-------
1 monitoring requirements for CO might either encourage or require such technological
2 improvements.
3 1.3.3 Ambient Monitoring Concentrations
4 Nearly 400 ambient monitoring stations report continuous hourly averages of CO
5 concentrations across the U.S. None of monitoring sites with extensive records12 for 2007 in 243
6 counties reported a second-highest 1-hour CO concentration above 35 ppm, the level of the
7 current 1-hour NAAQS, while sites in two counties reported second-highest 1-hour CO
8 concentrations between 15.1 and 35.0 ppm. Sites in five counties reported second-highest 8-hour
9 CO concentration of 5.0 ppm or higher (ISA, section 3.5.1.1).
10 For the most recent period for which air quality status relative to the CO NAAQS has
11 been analyzed (2007-2008), all areas of the U.S. meet both CO NAAQS.13 In the previous two
12 periods (2005-2006 and 2006-2007), one area, (Jefferson County, Alabama) has failed to meet
13 the 8-hour standard. Further, one area of the country is designated in non-attainment with the
14 CO NAAQS, however, air quality in that area currently meets the standards.
15 The current levels of ambient CO across the U.S. reflect the steady declines in ambient
16 concentrations that have occurred over the past several years. Both the 2nd highest 1-hour and
17 8-hour concentrations have significantly declined since the last review (Figures 1-2 and 1-3). At
18 the set of sites across the U.S. that have been continuously monitored since 1990 the average 2nd
19 highest 8-hour and 1-hour concentrations have declined by nearly 70%.
12 During the period 2005-2007, 291 out of 376 monitors sited in 243 different counties, cities or
municipalities, met the following dataset completeness criteria: 75% of the hours in a day, 75% of days in a
calendar quarter and 3 complete quarters for 3 years (ISA, section 3.4.2.2). An exception was made for monitors in
U.S. EPA Region 10 for which two rather than three complete quarters were considered to meet the criteria (ISA, p.
3-20).
13 The air quality status in areas monitored relative to the CO NAAQS is provided at
http://www.epa.gov/air/airtrends/values.html.
March 2010 1-9 Draft - Do Not Quote or Cite
-------
1 Figure 1-2. Trends in CO concentration (second maximum 8-hour average) in the U.S.,
2 1990-2008. The white line indicates average across the sites; ninety percent of
3 sites have concentrations below the top line, while ten percent of sites have
4 concentrations below the bottom line.
CO Air Quality, 1990 - 2008
(Based on Annual 2nd Maximum 8-hour average)
National Trend Based on 206 sites
16
14
10
90% of sites have concentrations below this line
NAAQS = 9 ppm
Average among all sites
u
o
O
10% of sites have concentrations below this line
1111111111222222222
012345
1234567
March 2010
1-10
Draft - Do Not Quote or Cite
-------
2
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
Figure 1-3. Trends in CO concentration (second maximum 1-hour average) in the U.S.,
1990-2008. The white line indicates average across the sites; ninety percent of
sites have concentrations below the top line, while ten percent of sites have
concentrations below the bottom line.
CO Air Quality, 1990 - 2008
(Based on Annual 2nd Maximum 1-hour average)
National Trend Based on 206 sites
t~. 30
Q. _
0,25
NAAQS = 35 ppm
90% of sites have concentrations below this line
Average among all sites
10% of sites have concentrations below tnis line
12345
234587
1.4 GENERAL APPROACH AND ORGANIZATION OF THE DOCUMENT
This draft PA includes staffs evaluation of the policy implications of the scientific
evidence reviewed in the ISA and the results of quantitative analyses based on that evidence.
Taken together, this information informs initial staff conclusions and the identification of a range
of policy options for consideration in addressing public health and welfare effects associated
with ambient CO.
Following this introductory chapter, chapter 2 focuses on review of the primary standards
for CO, presenting background information on the rationale for previous reviews and the
approach followed in the current review. Chapter 2 discusses the adequacy of the current
standards, taking into account evidence- and risk-based considerations, and includes initial staff
conclusions on adequacy. Chapter 2 also discusses potential alternative standards for
consideration, focusing on indicator, averaging time, form, and level, and includes initial staff
conclusions on alternative standards for consideration. Finally, chapter 3 discusses information
relevant to staffs consideration of a secondary standard for CO.
March 2010
1-11
Draft - Do Not Quote or Cite
-------
1 REFERENCES
2 McClellan RO, Chairman, Clean Air Scientific Advisory Committee. (1991). Letter to William K. Reilly, EPA
3 Administrator, July 17, 1991.
4 McClellan RO, Chairman, Clean Air Scientific Advisory Committee. (1992). Letter to William K. Reilly, EPA
5 Administrator, August 1, 1992.
6 National Research Council (2003). Managing Carbon Monoxide Pollution in Meteorological and Topographical
7 Problem Areas. Washington, D.C. The National Academies Press
8 U.S. EPA (1979a). Air Quality Criteria for Carbon Monoxide. Research Triangle Park, NC: Office of Health and
9 Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA-600/8-79-022.
10 U.S. EPA (1979b). Assessment of Adverse Health Effects from Carbon Monoxide and Implications for Possible
11 Modifications of the Standard. Office of Air Quality Planning and Standards, Research Triangle Park, NC.
12 U.S. EPA (1984a). Revised Evaluation of Health Effects Associated with Carbon Monoxide Exposure: An
13 Addendum to the 1979 EPA Air Quality Criteria Document for Carbon Monoxide. Office of Health and
14 Environmental Assessment, Environmental Criteria and Assessment Office; report ro. EPA-600/8-83-033F.
15 Research Triangle Park, NC
16 U.S. EPA (1984b). Review of the NAAQS for Carbon Monoxide: Reassessment of Scientific and Technical
17 Information. Office of Air Quality Planning and Standards, report no. EPA-450/584-904. Research
18 Triangle Park. NC.
19 U.S. EPA. (1991). Air Quality Criteria for Carbon Monoxide. Research Triangle Park, NC: Office of Health and
20 Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA/600/8-90/045F.
21 U.S. EPA (1992). Review of the National Ambient Air Quality Standards for Carbon Monoxide: Assessment of
22 Scientific and Technical Information. Office of Air Quality Planning and Standards Staff Paper, report
23 no.EPA/452/R-92-004..
24 U.S. EPA (2000). Air Quality Criteria for Carbon Monoxide. National Center for Environmental Assessment, Office
25 of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711;
26 report no. EPA/600/P-99/00IF. June 2000. Available at:
27 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=18163.
28 U.S. EPA (2008a). Draft Integrated Review Plan for the National Ambient Air Quality Standards for Carbon
29 Monoxide. U.S. Environmental Protection Agency, Research Triangle Park, NC, report no. EPA/452D-08-
30 001. Availiable at: http://www.epa.gov/ttn/naaqs/standards/co/s_co_cr_pd.html.
31 U.S. EPA (2008b). Plan for Review of the National Ambient Air Quality Standards for Carbon Monoxide. U.S.
32 Environmental Protection Agency, Research Triangle Park, NC, report no. EPA/452R-08-005. Also known
33 as Integrated Review Plan. Availiable at: http://www.epa.gov/ttn/naaqs/standards/co/s_co_cr_pd.html.
34 U.S. EPA (2008c). U.S. EPA's 2008 Report on the Environment (Final Report). U.S. Environmental Protection
35 Agency. DC.
36 U.S. EPA (2009a) Integrated Science Assessment for Carbon Monoxide, First External Review Draft. U.S.
37 Environmental Protection Agency, Research Triangle Park, NC, report no. EPA/600/R-08/071. Available
38 at: http://www.epa.gov/ttn/naaqs/standards/co/s_co_cr_isa.html
March 2010 1-12 Draft - Do Not Quote or Cite
-------
1 U.S. EPA (2009b) Integrated Science Assessment for Carbon Monoxide, Second External Review Draft, U.S
2 Environmental Protection Agency, Research Triangle Park, NC, report no. EPA/600/R-09/019B.
3 Available at: http://www.epa.gOv/ttn/naaqs/standards/co/s cocrisa.html
4 U.S. EPA (2009c) Risk and Exposure Assessment to Support the Review of the Carbon Monoxide Primary National
5 Ambient Air Quality Standards, First External Review Draft, U. S Environmental Protection Agency,
6 Research Triangle Park, NC, report no. EPA/600/R-09/019B. Available at:
7 http://www.epa.gOv/ttn/naaqs/standards/co/s co crisa.html
8 U.S. EPA (2010a) Integrated Science Assessment for Carbon Monoxide, National Center for Environmental
9 Assessment, Research Triangle Park, NC. EPA/600/R-09/019F. Available at:
10 http://www.epa.gOv/ttn/naaqs/standards/co/s co crisa.html
11 U.S. EPA (20lOb) Risk and Exposure Assessment to Support the Review of the Carbon Monoxide Primary National
12 Ambient Air Quality Standards, Second External Review Draft, U.S Environmental Protection Agency,
13 Research Triangle Park, NC, report no. EPA/600/R-09/019B. Available at:
14 http://www.epa.gOv/ttn/naaqs/standards/co/s co cr rea.html
March 2010 1-13 Draft - Do Not Quote or Cite
-------
1 2 REVIEW OF THE PRIMARY STANDARDS FOR CARBON
2 MONOXIDE
3 This chapter presents initial staff conclusions for consideration in deciding whether the
4 existing primary standards for carbon monoxide (CO) should be revised and, if so, what
5 revisions are appropriate. The current primary CO standards include a 1-hour and an 8-hour
6 standard to protect public health from exposure to CO. In initially evaluating the current primary
7 NAAQS we have reviewed, and as appropriate, updated, a series of key policy-relevant issues
8 initially presented in the Plan for Review of the National Ambient Air Quality Standards for
9 Carbon Monoxide (USEPA, 2008a, section 3.1). The answers to these policy-relevant questions
10 will inform decisions on whether, and if so, how to revise the primary standards for CO.
11 Following a background section regarding considerations in the previous review, the
12 discussion in this chapter focuses on two central issues related to: (1) the adequacy of the current
13 CO standards and (2) what potential alternative standards, if any, should be considered in this
14 review. Within each of these broad areas, a series of questions are addressed. The four basic
15 elements of the NAAQS (e.g., indicator, averaging time, level, and form) will be considered
16 collectively in evaluating the health protection afforded by the current or any alternative
17 standards considered.
18 2.1 APPROACH
19 For the purposes of this draft Policy Assessment (PA), staff has drawn from EPA's
20 assessment and integrated synthesis of the scientific evidence presented in the Integrated Science
21 Assessment for Carbon Monoxide (USEPA, 2010a; henceforth referred to as the ISA) and 2000
22 Air Quality Criteria Document for Carbon Monoxide (USEPA, 2000; henceforth referred to as
23 the 2000 AQCD) and on the analyses presented in the second draft Risk and Exposure
24 Assessment (USEPA, 201 Ob; henceforth referred to as the draft REA). The evidence-based
25 discussions presented in this chapter draw upon evidence from epidemiologic studies, controlled
26 human exposure studies, and toxicological studies evaluating short- or long-term exposures to
27 CO, as discussed in chapter 5 of the ISA, with supporting information related to dosimetry and
28 potential mode of action evidence as presented in chapters 4 and 5 of the ISA, respectively, as
29 well as the integration of evidence across disciplines presented in chapter 2 of the ISA. The
30 exposure/risk-based discussions have drawn from the quantitative health analyses for CO
31 presented in the draft REA. Together the evidence-based and risk-based considerations have
32 informed our conclusions related to the adequacy of the current CO standards and alternative
33 standards that are supported by the currently available scientific evidence.
March 2010 2-1 Draft - Do Not Quote or Cite
-------
1 In presenting a range of primary standard options for consideration, we note that the final
2 decision is largely a public health policy judgment. A final decision must draw upon scientific
3 information and analyses about health effects and risks, as well as judgments about how to deal
4 with the range of uncertainties that are inherent in the scientific evidence and analyses. Our
5 approach to informing these judgments, discussed more fully below, is based on the recognition
6 that the available health effects evidence generally reflects a continuum, consisting of ambient
7 levels at which scientists generally agree that health effects are likely to occur, through lower
8 levels at which the likelihood and magnitude of the response become increasingly uncertain.
9 This approach is consistent with the requirements of the NAAQS provisions of the Act and with
10 how EPA and the courts have historically interpreted the Act. These provisions require the
11 Administrator to establish primary standards that, in the Administrator's judgment, are requisite
12 to protect public health with an adequate margin of safely. In so doing, the Administrator seeks
13 to establish standards that are neither more nor less stringent than necessary for this purpose.
14 The Act does not require that primary standards be set at a zero-risk level, but rather at a level
15 that avoids unacceptable risks to public health.
16 This section includes background information on the approach used in the previous
17 review of the CO standards (section 2.1.1) and also a discussion of the approach for the current
18 review (section 2.1.2).
19 2.1.1 Approach Used in the Previous Review
20 The current primary standards for CO are set at 9 parts per million (ppm) as an 8-hour
21 average and 35 ppm as a 1-hour average, neither to be exceeded more than once per year. These
22 standards were initially set in 1971to protect against the occurrence of carboxyhemoglobin
23 (COHb) levels that may be associated with effects of concern (36 FR 8186). Reviews of these
24 standards in the 1980s and early 1990s, respectively, identified additional evidence regarding
25 ambient CO, COHb levels and health effects (USEPA, 1984a, 1984b; USEPA, 1991; USEPA,
26 1992; McClellan, 1991,1992). Assessment of the evidence in those reviews led the EPA to
27 retain the existing standards without revision (59 FR 38906).
28 The 1994 decision to retain the primary standards without revision was based on the
29 evidence published through 1990 and reviewed in the 1991 AQCD (USEPA, 1991), the 1992
30 Staff Paper assessment of the policy-relevant information contained in the AQCD and the
31 quantitative exposure assessment (USEPA, 1992), and the advice and recommendations of
32 CAS AC (McClellan 1991, 1992). At that time, as at the time of the prior NAAQS review (50
33 FR 37484), COHb levels in blood were recognized as the most useful estimates of exogenous
34 CO exposures and to serve as the best biomarker of CO toxicity for ambient-level exposures to
March 2010 2-2 Draft - Do Not Quote or Cite
-------
1 CO. Consequently, COHb levels were used as the indicator of health effects in the identification
2 of health effect levels of concern for CO (59 FR 38909).
3 In reviewing the standards in 1994 the Administrator first recognized the need to
4 determine the COHb levels of concern "taking into account a large and diverse health effects
5 database". The more uncertain and less quantifiable evidence was taken into account to identify
6 the lower end of this range to provide an adequate margin of safety for effects of clear concern.
7 To consider ambient CO concentrations likely to result in COHb levels of concern, a model
8 solution to the Coburn-Forster-Kane (CFK) differential equations was employed in the analysis
9 of CO exposures expected to occur under air quality scenarios related to just meeting the current
10 8-hour CO NAAQS, the controlling standard (USEPA, 1992).1 Key considerations in this
11 approach are described below
12 Carboxyhemoglobin Levels of Concern and Margin of Safety
13 The assessment of the science that was presented in the 1991 AQCD (USEPA, 1991)
14 indicated that CO is associated with effects in the cardiovascular system, central nervous system
15 (CNS), and the developing fetus. Additionally, factors recognized as having potential interaction
16 with CO that may alter health effects included exposures to other pollutants, some drugs and
17 some environmental factors, such as altitude. Cardiovascular effects of CO, as measured by
18 decreased time to onset of angina and to onset of significant electrocardiogram (ECG) ST-
19 segment depression2 were judged by the Administrator to be "the health effects of greater
20 concern, which clearly had been associated with CO exposures at levels observed in ambient air"
21 (59 FR 38913).
22 Based on the consistent findings of response in patients with coronary artery disease3
23 across the controlled human exposure evidence (Adams et al., 1988; Allred et al., 1989a, 1989b,
24 1991; Anderson et al., 1973; Kleinman et al., 1989, 1998; Sheps et al., 1987) and discussions of
25 adverse health consequences in the 1991 AQCD and the 1992 Staff Paper, at the CAS AC
1 Air quality analyses of CO levels in the U.S. consistently demonstrate that attainment of the 8-hour
standard results in 1-hour maximum concentrations well below the corresponding 1-hour standard.
2 The ST-segment is a portion of the electrocardiogram, depression of which is an indication of insufficient
oxygen supply to the heart muscle tissue (myocardial ischemia). Myocardial ischemia can result in chest pain
(angina pectoris) or such characteristic changes in ECGs or both. In individuals with coronary artery disease, it
tends to occur at specific levels of exercise. The duration of exercise required to demonstrate chest pain and/or a 1-
mm change in the ST segment of the ECG were key measurements in the multicenter study by Allred et al (1989a,
1989b, 1991).
3 Coronary artery disease (CAD), often also called coronary heart disease or ischemic heart disease is a
category of cardiovascular disease associated with narrowed heart arteries. Individuals with this disease have
myocardial ischemia, which occurs when the heart muscle receives insufficient oxygen delivered by the blood.
Exercise-induced angina pectoris (chest pain) occurs in many of them. Among all patients with diagnosed CAD, the
predominant type of ischemia, as identified by ST segment depression, is asymptomatic (i.e., silent). Patients who
experience angina typically have additional ischemic episodes that are asymptomatic (2000 AQCD, section 7.7.2.1).
In addition to such chronic conditions, CAD can include myocardial infarction (ISA, p. 5-24).
March 2010 2-3 Draft - Do Not Quote or Cite
-------
1 meetings and in the July 1991 CAS AC letter, the Administrator concluded that "CO exposures
2 resulting in COHb levels of 2.9-3.0 percent (CO-Ox) or higher in persons with heart disease have
3 the potential to increase the risk of decreased time to onset of angina pain and ST-segment
4 depression" (59 FR 38913).4 Two of the five key studies were given particular emphasis in the
5 1991 AQCD to indicate the basis for conclusions regarding lowest observed-effect levels of
6 COHb in patients with exercise-induced ischemia, in terms of measured COHb and its
7 representation in terms of increase from baseline COHb on the order of 1.5 to 2.2% (USEPA,
8 1991, pp. 1-11 to 1-12; Allredetal., 1989a, 1989b, 1991; Anderson et al., 1973).
9 While EPA and CAS AC recognized the existence of a range of views among health
10 professionals on the clinical significance of the responses observed in the clinical studies,
11 CASAC noted that the dominant view was that they should be considered "adverse or harbinger
12 of adverse effect" (McClellan, 1991) and EPA recognized that it was "important that standards
13 be set to appropriately reduce the risk of ambient exposures which produce COHb levels that
14 could induce such potentially adverse effects" (59 FR 38913) as those occurring at COHb levels
15 of 2.9-3.0% (CO-oximeter) (59 FR 38913; USEPA, 1991, p. 1-12; USEPA, 1992, pp. 20-22). In
16 further considering additional results from the controlled human exposure evidence as well as
17 other aspects of the available evidence and uncertainties regarding modeling estimates of COHb
18 formation and human exposure to COHb levels in the population associated with attainment of a
19 given CO NAAQS, the Administrator recognized the need to extend the range of COHb levels
20 for consideration in evaluating whether the current CO standards provide an adequate margin of
21 safely to those falling between 2.0 to 2.9 (59 FR 38913). Factors considered in recognizing this
22 margin of safety included the following (59 FR 38913).
23 • Uncertainty associated with the clinical importance of cardiovascular effects associated
24 with exposures to CO that resulted in COHb levels of 2 to 3 percent. Although
25 recognizing the possibility that there is no threshold for these effects even at lower
26 COHb levels, the health significance of the small changes in ST-segment depression
27 observed was considered somewhat minor.
28 • Findings of short-term reduction in maximal work capacity measured in trained
29 athletes exposed to CO at levels resulting in COHb levels of 2.3 to 7 percent.
30 • The potential that the most sensitive individuals have not been studied, the limited
31 information regarding the effects of ambient CO in the developing fetus, and concern
32 about visitors to high altitudes, individuals with anemia or respiratory disease, or the
33 elderly.
4 Based on matched measurements available for CO-Ox and gas chromatography in this range, staff note
that CO-oximetry measurements of 2.9 to 3.0 percent COHb appear to correspond to GC measurements on the order
of 2% (Allredetal., 1991).
March 2010 2-4 Draft - Do Not Quote or Cite
-------
1 • Potential for short term peak CO exposures to be responsible for impairments
2 (impairment of visual perception, sensorimotor performance, vigilance or other CNS
3 effects) which could be a matter of concern for complex activities such as driving a car,
4 although these effects had not been demonstrated to be caused by CO concentrations in
5 ambient air.
6 • Concern based on limited evidence for individuals exposed to CO concurrently with
7 drugs (e.g., alcohol), during heat stress, or co-exposure to other pollutants.
8 • Large uncertainties remaining regarding modeling COHb formation and estimating
9 human exposure to CO which could lead to overestimation of COHb levels in the
10 population associated with attainment of a given CO NAAQS.
11 • Uncertainty associated with COHb measurements made using CO-Ox which may not
12 reflect COHb levels in angina patients studied, thereby creating uncertainty in
13 establishing a lowest effects level for CO.
14 Based on these considerations of the evidence, the Administrator identified a range of
15 COHb levels of concern extending from 2.9% at the upper end down to 2% at the lower end and
16 concluded that "evaluation of the adequacy of the current standard should focus on reducing the
17 number of individuals with cardiovascular disease from being exposed to CO levels in the
18 ambient air that would result in COHb levels of 2.1 percent." She additionally concluded that
19 standards that "protect against COHb levels at the lower end of the range should provide an
20 adequate margin of safety against effects of uncertain occurrence, as well as those of clear
21 concern that have been associated with COHb levels in the upper-end of the range" (59 FR
22 48914).
23 Estimation of Population Exposures
24 To estimate CO exposures and resulting COHb levels that might be expected under air
25 quality conditions that just met the current standards, an analysis of exposure and associated
26 internal dose in terms of COHb levels in the population of interest in the city of Denver,
27 Colorado was performed (59 FR 38906; USEPA, 1992). That analysis indicated that if the 9
28 ppm 8-hour standard were just met, the proportion of the nonsmoking population with
29 cardiovascular disease experiencing a daily maximum 8-hour exposure at or above 9 ppm for 8
30 hours decreased by an order of magnitude or more as compared to the proportion under then-
31 existing CO levels, down to less than 0.1 percent of the total person-days in that population.
32 More specifically, upon meeting the 8-hour standard, EPA estimated that less than 0.1% of the
33 nonsmoking cardiovascular-disease population would experience a COHb level greater than or
34 equal to 2.1%. A smaller percentage of the at-risk population was estimated to exceed higher
March 2010 2-5 Draft - Do Not Quote or Cite
-------
1 COHb levels.5 Based on these estimates, the Administrator concluded that "relatively few
2 people of the cardiovascular sensitive population group analyzed will experience COHb levels >
3 2.1 percent when exposed to CO levels in absence of indoor sources". The analysis also took
4 into account that certain indoor sources (e.g., passive smoking, gas stove usage) contributed to
5 total CO exposure and EPA recognized that such sources may be of concern for such high risk
6 groups as individuals with cardiovascular disease, pregnant women, and their unborn children
7 but concluded that "the contribution of indoor sources cannot be effectively mitigated by
8 ambient air quality standards" (59 FR38914).
9 Decision Regarding Adequacy of the Standards
10 Based on consideration of the evidence and the quantitative results of the exposure
11 assessment, the Administrator concluded that revisions of the current primary standards for CO
12 were not appropriate at that time (59 FR 38914). The Administrator additionally concluded that
13 both averaging times for the primary standards, 1-hour and 8-hour, be retained. The 1-hour and
14 8-hour averaging times were first chosen when EPA promulgated the primary NAAQS for CO in
15 1971. The selection of the 8-hour averaging time was based on the following: (a) most
16 individuals' COHb levels appeared to approach equilibrium after 8 hours of exposure, (b) the 8-
17 hour time period corresponded to the blocks of time when people were often exposed in a
18 particular location or activity (e.g., working or sleeping), and (c) judgment that this provided a
19 good indicator for tracking continuous exposures during any 24-hour period. The 1-hour
20 averaging time was selected as better representing a time period of interest to short-term CO
21 exposure and providing protection from effects which might be encountered from very short
22 duration peak exposures in the urban environment (59 FR 38914).
23 2.1.2 Approach for the Current Review
24 To evaluate whether it is appropriate to consider retaining the current primary CO
25 standards, or whether consideration of revisions is appropriate, we adopted an approach in this
26 review that builds upon the general approach used in the last review and reflects the broader
27 body of evidence and information now available. As summarized above, the Administrator's
28 decisions in the previous review were based on an integration of information on health effects
29 associated with exposure to ambient CO; expert judgment on the adversity of such effects on
30 individuals; and policy judgments as to when the standard is requisite to protect public health
31 with an adequate margin of safety, which were informed by air quality and related analyses,
5 In the 1992 assessment, the person-days (number of persons multiplied by the number of days per year
exposed) and person-hours (number of persons multiplied by the number of hours per year exposed) were the
reported exposure metrics. Upon meeting the 8-hour standard, it was estimated that less than 0.1% of the total
person-days simulated for the nonsmoking cardiovascular-disease population were associated with a maximum
COHb level greater than or equal to 2.1% (USEPA, 1992; Johnson et al., 1992).
March 2010 2-6 Draft - Do Not Quote or Cite
-------
1 quantitative exposure and risk assessments when possible, and qualitative assessment of impacts
2 that could not be quantified.
3 In conducting this assessment, we draw on the current evidence and quantitative
4 assessments of exposure pertaining to the public health risk of ambient CO. In considering the
5 scientific and technical information, we consider both the information available at the time of the
6 last review and information newly available since the last review, including the current ISA and
7 the 2000 AQCD (USEPA, 2010; USEPA, 2000), as well as current and preceding quantitative
8 exposure/risk assessments (USEPA 201 Ob). As was the case at the time of the last review, the
9 best characterized health effect associated with CO levels of concern is hypoxia (reduced oxygen
10 availability) induced by increased COHb levels in blood (ISA, section 5.1.2). Accordingly, CO
11 is of particular concern for those with impaired cardiovascular systems, and the most compelling
12 evidence of cardiovascular effects is that from a series of controlled human exposure studies
13 among exercising individuals with coronary heart disease (CHD) also referred to as coronary
14 artery disease (CAD) (ISA, sections 5.2.4 and 5.2.6). Additionally available in this review are a
15 number of epidemiological studies that investigated the association of cardiovascular disease-
16 related health outcomes with concentrations of CO at ambient monitors. To inform our review
17 of the ambient standards, we performed a quantitative exposure and dose modeling analysis that
18 estimated COHb levels associated with different air quality conditions in simulated at-risk
19 populations in two U.S. cities. Thus, in developing conclusions in this review as discussed in
20 sections 2.2 and 2.3 below, we have taken into account both evidence-based and exposure/risk-
21 based considerations framed by a series of key policy-relevant questions.
22 2.2 ADEQUACY OF THE CURRENT STANDARD
23 In considering the adequacy of the current CO standards, the overarching question we
24 consider is:
25 • Does the currently available scientific evidence- and exposure/risk-based
26 information, as reflected in the ISA and draft REA, support or call into question
27 the adequacy of the protection afforded by the current CO standards?
28 To assist us in interpreting the currently available scientific evidence and the results of
29 recent quantitative exposure/risk analyses to address this question, we have focused on a series
30 of more specific questions, posed within sections 2.2.1 and 2.2.2 below. In considering the
31 scientific and technical information, we consider both the information available at the time of the
32 last review and information newly available since the last review which has been critically
33 analyzed and characterized in the 2000 AQCD and more recently in the ISA.
March 2010 2-7 Draft - Do Not Quote or Cite
-------
1 2.2.1 Evidence-based Considerations
2 In considering the evidence with regard to the issue of adequacy of the current standard,
3 we address a series of questions beginning with the health effects associated with CO exposure,
4 followed by the use of COHb levels as the indicator of CO exposures and biomarker for
5 characterizing the potential for health effects associated with exposures to ambient CO, and then
6 the identification of the populations most susceptible to the effects of CO. We next consider the
7 evidence regarding the levels of CO in ambient air associated with health effects and the
8 important uncertainties associated with the evidence.
9 • Does the current evidence alter our conclusions from the previous review
10 regarding the health effects associated with exposure to CO?
11 The current evidence continues to support our conclusions from the previous review
12 regarding key health effects associated with CO exposure. The best characterized effect of CO
13 continues to be related to the binding of CO to blood Hb to form increased levels of COHb (ISA,
14 sections 4.1 and 5.1.2) and the primary focus is on associated cardiovascular effects (ISA,
15 section 5.2). In the scientific assessment for the current review, a likely causal relationship is
16 judged to exist between relevant CO exposures6 and cardiovascular effects (ISA, section 2.5.1)
17 which is similar to conclusions in the last review. The evidence for effects on the central
18 nervous system, birth outcomes and developmental effects, and respiratory effects, while in some
19 cases expanded from that which was available at the time of the last review, is judged to be
20 suggestive of a causal relationship with relevant CO exposures (ISA, section 2.5). These overall
21 findings, additional details of which are described below, are consistent with and extend in some
22 ways conclusions drawn from the health effects evidence in the last review.
23 The long-standing body of evidence that has established many aspects of the biological
24 effects of CO continues to contribute to our understanding of the health effects of ambient CO.
25 Upon inhalation, CO diffuses through the respiratory zone (alveoli) to the blood where it binds to
26 Hb, forming COHb. This binding to reduced iron in heme proteins and the alteration of their
27 function is the common mechanism underlying biological responses to CO. Accordingly,
28 inhaled CO elicits various health effects through binding to, and associated alteration of the
29 function of, a number of heme-containing molecules, mainly Hb (see e.g., ISA, section 4.1). The
30 best characterized health effect associated with CO levels of concern is hypoxia (reduced O2
31 availability) induced by increased COHb levels in blood and decreased oxygen availability to
32 critical tissues and organs, specifically the heart (ISA, section 5.1.2). Thus, people with a
33 cardiovascular disease that affects their ability to compensate for this effect (e.g., through
6 Relevant CO exposures are defined in the ISA as "generally within one or two orders of magnitude of
ambient CO concentrations" (ISA, section 2.5).
March 2010 2-8 Draft - Do Not Quote or Cite
-------
1 vasodilation or increased cardiac output) are a key population at risk from CO exposures.7 The
2 dose metric most commonly used as a bioindicator of exposure and health risk from CO is the
3 level of COHb in the blood (USEPA, 1991; 2002; 2010a).
4 The body of health effects evidence for CO has grown considerably since the previous
5 review with the addition of numerous epidemiological and toxicological studies (ISA; 2000
6 AQCD). This evidence provides additional detail and support to our prior understanding of CO
7 effects and population susceptibility. For example, the currently available evidence expands on
8 potential nonhypoxic mechanisms for CO effects, although the importance of such mechanisms
9 at environmentally-relevant CO exposures is unclear. Most notably, the current evidence
10 includes much expanded epidemiological evidence that provides support for previous
11 conclusions regarding cardiovascular disease-related susceptibility and indications of air quality
12 conditions that may be associated with ambient CO-related risk.
13 In the most recent analysis of the health evidence EPA has developed a framework to
14 characterize the evidence as to likelihood of causal relationships between exposure to ambient
15 CO and specific health effects (ISA, chapter 1). The clearest evidence is available for
16 cardiovascular effects, such that the major conclusion drawn in the ISA regarding the critical
17 analysis of all available data on health effects of CO including the clinical and epidemiological
18 evidence is that "Given the consistent and coherent evidence from epidemiologic and human
19 clinical studies, along with biological plausibility provided by CO's role in limiting Q^
20 availability, it is concluded that a causal relationship is likely to exist between relevant[8] short-
21 term CO exposures and cardiovascular morbidity" (ISA, p. 2-6, section 2.5.1). The evidence is
22 judged suggestive of causal relationships between relevant short- and long-term CO exposures
23 and CNS effects, birth outcomes and developmental effects following long-term exposure,
24 respiratory morbidity following short-term exposure, and mortality following short-term
25 exposure (ISA, section 2.5, Table 2-1). It was concluded that there is not likely to be a causal
26 relationship between relevant long-term CO exposures and mortality (ISA, Table 2-1).
27 Similar to the previous review, results from controlled human exposure studies of
28 individuals with coronary artery disease (Adams et al., 1988; Allred et al., 1989a, 1989b, 1991;
29 Anderson et al., 1973; Kleinman et al., 1989, 1998; Sheps et al., 1987;) are the "most compelling
30 evidence of CO-induced effects on the cardiovascular system" (ISA, Section 5.2). Additionally,
7 As described in discussing susceptibility below, cardiovascular disease comprises many types of medical
disorders, including heart disease (e.g., CHD), cerebrovascular disease (e.g., stroke), hypertension (high blood
pressure), and peripheral vascular diseases and heart disease patients often have markedly reduced circulatory
capacity and reduced ability to compensate for increased circulatory demands during exercise and other stress (2000
AQCD, p. 7-7).
8 Relevant CO exposures are defined in the ISA as "generally within one or two orders of magnitude of
ambient CO concentrations" (ISA, section 2.5).
March 2010 2-9 Draft - Do Not Quote or Cite
-------
1 the use of an internal dose metric, COHb, adds to the strength of the findings in these controlled
2 exposure studies. As a group, these studies demonstrate the role of CO in increasing the
3 susceptibility of people with CAD to incidents of exercise-associated myocardial ischemia.
4 Among these controlled human exposure studies, the ISA places principal emphasis on
5 the study of coronary artery disease patients by Allred et al. (1989a, 1989b, 1991) which was key
6 to considerations from the previous review for the following reasons: 1) dose-response
7 relationships was observed; 2) effects were observed at the lowest COHb levels tested (mean of
8 2-2.4% COHb following experimental CO exposure), providing no evidence of a threshold; 3)
9 objective measures of myocardial ischemia (ST-segment depression) were assessed, as well as
10 the subjective measure of decreased time to induction of angina; 4) measurements were taken
11 both by gas chromatography (GC), which provides a more accurate measurement of COHb blood
12 levels9, and by CO-Ox; 5) a large number of subjects were used; 6) a strict subject selection
13 protocol was employed to include only CAD patients with reproducible exercise-induced
14 angina.; and 7) the study was conducted at multiple laboratories around the U.S. This study
15 evaluated changes in time to exercise-induced onset of markers of myocardial ischemia resulting
16 from two short CO exposures targeted to result in mean study subject COHb levels of 2% and
17 4%, respectively (ISA, section 5.2.4). In this study, subjects (n=63) on three separate occasions
18 underwent an initial graded exercise treadmill test, followed by 50 to 70-minute exposures under
19 resting conditions to room air CO concentrations or CO concentrations targeted for each subject
20 to achieve blood COHb levels of 2% and 4%. The exposures were to average CO concentrations
21 of 0.7 ppm (room air concentration range 0-2 ppm), 117 ppm (range 42-202 ppm) and 253 ppm
22 (range 143-357 ppm). After the 50- to 70-min exposures, subjects underwent a second graded
23 exercise treadmill test, and the percent change in time to onset of angina and time to ST endpoint
24 between the first and second exercise tests was determined. For the two CO exposures, the
25 average post-exposure COHb concentrations were reported as 2.4% and 4.7%, and the
26 subsequent post-exercise average COHb concentrations were reported as 2.0% and 3.9%.10
9 As stated in the ISA, the gas chromatographic technique for measuring COHb levels "is known to be
more accurate than spectrophotometric measurements, particularly for samples containing COHb concentrations <
5%" (ISA, p. 5-41). CO-oximetry is a spectrophotometric method commonly used to rapidly provide approximate
concentrations of COHb during controlled exposures (ISA, p. 5-41). At the low concentrations of COHb (<5%)
more relevant to exposures to ambient CO, co-oximeters are reported to overestimate COHb levels compared to GC
measurements, while at higher concentrations, this method is reported to produce underestimates (ISA, p.4-18).
10 While the COHb blood level for each subject during the exercise tests was intermediate between the
post-exposure and subsequent post-exercise measurements (e.g., mean 2.0-2.4% and 3.9-4.7%), the study authors
noted that the measurements at the end of the exercise test represented the COHb concentrations at the approximate
time of onset of myocardial ischemia as indicated by angina and ST segment changes. The corresponding ranges of
CO-Ox measurements for the two exposures were 2.7-3.2% and 4.7-5.6%. In this document, we refer to the GC-
measured mean of 2.0% or 2.0-2.4% for the COHb levels resulting from the lower experimental CO exposure.
March 2010 2-10 Draft - Do Not Quote or Cite
-------
1 Relative to clean-air exposure that resulted in a mean COHb level of 0.6% (post-
2 exercise), exposure to CO resulting in post-exercise mean COHb concentrations of 2.0% and
3 3.9% were observed to decrease the exercise time required to induce ST-segment depression by
4 5.1 (p=0.01) and 12.1% (p<0.001), respectively. These changes were well correlated with the
5 onset of exercise-induced angina, the time to which was shortened by 4.2% (p=0.027) and 7.1%
6 (p=0.002), respectively, for the two experimental CO exposures (Allred et al., 1989a, 1989b,
7 1991).u As at the time of the last review, while ST-segment depression is recognized as an
8 indicator of myocardial ischemia, the exact physiological significance of the observed changes is
9 unclear (ISA, p. 5-48).
10 No human clinical studies have been specifically designed to evaluate the effect of
11 controlled short-term exposures to CO resulting in COHb levels lower than a study mean of 2%
12 (ISA, section 5.2.6). However, an important finding of the multi -lab oratory study was the dose-
13 response relationship (discussed further in addressing a subsequent question below) observed
14 between COHb and the markers of myocardial ischemia, with effects observed at the lowest
15 increases in COHb tested, without evidence of a measurable threshold effect. As reported by the
16 authors, the results comparing "the effects of increasing COHb from baseline levels (0.6%) to 2
17 and 3.9% COHb showed that each produced further changes in objective ECG measures of
18 ischemia" implying that "small increments in COHb could adversely affect myocardial function
19 and produce ischemia" (Allred et al., 1989b; Allred et al., 1991).
20 The epidemiological evidence has expanded considerably since the last review including
21 numerous additional studies that are coherent with the evidence on markers of myocardial
22 ischemia from controlled human studies of CAD patients (ISA, section 2.7). The most recent
23 series of epidemiological studies have evaluated the associations between ambient concentrations
24 of multiple pollutants (i.e. PM2.5, NO2, SO2, Os and CO) at fixed-site monitors and increases in
25 emergency department visits and hospital admissions for specific cardiovascular health outcomes
26 including ischemic heart disease (IHD), myocardial infarction (MI), congestive heart failure
27 (CHF), and cardiovascular diseases (CVD) as a whole (Bell et al., 2009; Koken et al., 2003; Linn
28 et al., 2000; Mann et al., 2002; Metzger et al., 2004; Peel et al., 2007; Symons et al., 2006;
29 Tolbert et al., 2007; Wellenius et al., 2005). As noted by the ISA, "[s]tudies of hospital
30 admissions and ED visits for IHD provide the strongest [epidemiological] evidence of ambient
31 CO being associated with adverse CVD outcomes" (ISA, p. 5-40, section 5.2.3; Linn et al., 2000;
11 Another indicator measured in the study was the combination of heart rate and systolic blood pressure
which provides a clinical index of the work of the heart and myocardial oxygen consumption, since heart rate and
blood pressure are major determinants of myocardial oxygen consumption (Allred et al., 1991). A decrease in
oxygen to the myocardium would be expected to be paralleled by ischemia at lower heart rate and systolic blood
pressure. This heart rate-systolic blood pressure indicator at the time to ST-endpoint was decreased by 4.4% at the
3.9% COHb dose level and by a nonstatistically-significant, smaller amount at the 2.0% COHb dose level.
March 2010 2-11 Draft - Do Not Quote or Cite
-------
1 Mann et al., 2002;Metzger et al., 2004; Peel et al., 2007). With regard to studies for other
2 measures of cardiovascular morbidity, the ISA notes that "[tjhough not as consistent as the IHD
3 effects, the effects for all CVD hospital admissions (which include IHD admissions) and CHF
4 hospital admissions also provide evidence for an association of cardiovascular outcomes and
5 ambient CO concentrations" (ISA, section 5.2.3; Bell et al., 2009; Tolbert et al., 2007).12
6 The geographic areas in the studies for IHD and all CVD outcomes (which is inclusive of
7 IHD) include Atlanta, urban areas of California and a group of 126 urban counties (Table 2-1
8 below; ISA, Figure 2-1). In these studies, different multiple-year periods were assessed and the
9 air quality statistics analyzed varied from 1 hour to 24 hours. Together the individual study
10 periods spanned the period from 1988 through 2005. Further, the studies encompass a range of
11 air quality conditions, such that during the time periods of study for these outcomes, only the
12 three Atlanta studies (Metzger et al., 2004; Peel et al., 2007; Tolbert et al., 2007) (and the studies
13 of CHF in Pittsburgh and Baltimore)13 did not include years in which ambient CO concentrations
14 exceeded the 8-hour standard (Table 2-1).
15
16
17
12 As summarized further in response to a subsequent question below, the ISA notes the difficulty in
determining from this group of studies on cardiovascular disease outcomes the extent to which CO is independently
associated with CVD outcomes as compared to CO as a marker for the effects of another traffic-related pollutant or
mix of pollutants, although the ISA concludes that the evidence provides support for a direct effect of short-term
ambient CO exposure on CVD morbidity (ISA, pp. 5-40 to 5-41).
13 The study by Symons et al (2006) in Baltimore focused on an 8-month period from April to December,
2002 (ISA, pp. 5-31, 5-32, C-21).
March 2010 2-12 Draft - Do Not Quote or Cite
-------
1 Table 2-1. Information from U.S. epidemiological studies of hospital admissions for ischemic heart disease (ISA, Figure 5-2),
2 cardiovascular disease (ISA, Figure 5-5) and congestive heart failure (ISA, Figure 5-4) and on air quality during
3 the study periods.
Study Reference
Study Area
Study Time
Period
Health
Outcome
Risk Estimate
(confidence interval)
Estimates are
standardized within
averaging times}
Air Quality
Metric
Analyzed
with Health
Outcome
CO concentration in
study areas, in terms
of air quality metric
analyzed (ppm)
Mean or
median
(study)
99th
percentile
(AQS)
Range of design values
(ppm) for study areas
during study period
1-hour
8-hour
U.S. studies of hospital admissions for coronary heart disease (drawn from ISA, Figure 5-2).
Metzgeretal., 2004
Peel etal., 2007
Mann etal., 2002
Linn etal., 2000
Atlanta
Atlanta
CA
CA
1993-2000
1993-2000
1988-1995
1992-1995
IHD
IHD
IHD
Ml
1.016(0.999-1.034)
1.0136(1.0053-1.0220)
1.02(1.01-1.03)
1-hr daily max
1-hr daily max
8-hr daily max
24-hr average
1.52
1.8
2.07
1.5
5.5-5.9
5.5 - 5.9
1.3-15.9
1.1-8.3
4.3-16.3
4.3-16.3
16.5-32
16.5-25
3.2-5.3
3.2 - 5.3
11.6-23.4
11.6-16.4
U. S. studies of hospital admissions for cardiovascular disease (drawn from ISA, Figure 5-5)
Bell etal., 2009
Tolbert etal., 2007
Linn etal., 2000
126 urban counties
Atlanta
CA
1999-2005
1993-2004
1992-1995
CVD 65+
CVD
CVD
1.0096(1.0079-1.0112)
1.017(1.008-1.027)
1.012(1.005-1.020)
1-hr daily max
1-hr daily max
24-hr average
1.32
1.6
1.5
1.2-22.1
5.3 - 5.4
1.1-8.3
17.5-33.5
3.5-16.3
16.5-25
8.8-24.3
2.5-5.3
11.6-16.4
U. S. studies of hospital admissions for congestive heart failure (drawn from ISA, Figure 5-4)
Linn etal., 2000
Mann etal. ,2002
Symons et al., 2006
Wellenius et al., 2005
Metzgeretal. ,2004
Peel etal., 2007
Koken etal. ,2003
CA
CA
Baltimore
Pittsburgh
Atlanta
Atlanta
Denver
1992-1995
1988-1995
2002
1987-1999
1993-2000
1993-2000
1993-1997
CHF
IHD+CHF
CHF
CHF
CHF
CHF
CHF
1.012(1.005-1.020)
1.0304(1.0135-1.0475)
1.08(0.40-2.99)
1.0843(1.0614-1.1078)
1.01(0.99-1.03)
1.019(0.997-1.041)
1.181(1.002-1.393)
24-hr average
8-hr daily max
8-hr daily max
24-hr average
1-hr daily max
1-hr daily max
24-hr average
1.5
2.07
0.4
1.03
1.52
1.8
0.9
1.1-8.3
1.3-15.9
2.3
1.6-3.9
5.5-5.9
5.5 - 5.9
1.3-2.0
16.5-25
16.5-32
4.4-9.6
3.1-5.9
4.3-16.3
4.3-16.3
8.2-18.23
11.6-16.4
11.6-23.4
2.6-3.0
1.3-8.8
3.2-5.3
3.2-5.3
4.7-10.43
Table legend. Information is from (a) studies that provide the strongest and most consistent weight of evidence from the epidemiological literature, and (b) additional supportive epidemiological studies. Presented are risk
estimates (unadjusted for other pollutants) for hospital admissions for ischemic heart disease (IHD), myocardial infarction (Ml), cardiovascular disease (CVD; Bell 2009 included only subjects 65 years old and above)
outcomes, and congestive heart failure (CHF), with the study reported mean/median for the ambient CO concentration metric analyzed by the study authors. The 99th percentile CO concentrations for the three studied
metrics (1-, 8-hr daily max, or 24-hr average) were derived from AQS data for monitors in study areas during study time periods. Design values are CO concentrations in the statistical form of the standard; presented is the
range of 2nd maximum 1 -hour and 8-hour average concentrations at highest monitor in each study area across years (from AQS for the areas and time periods of the studies). 'Estimates standardized for 1 ppm increase in
CO concentration for 1-hour metric, 0.75 ppm for 8-hour metric and 0.5 ppm for 24-hour metric. Lag period (e.g., 0, 1,0-1, 0-2) varies by study. 2These values for Metzgeretal., 2004 and Bell etal., 2009 are the median.
3These design values reflect full year of data, inclusive of study months (July-Aug).
March 2010
2-13
Draft - Do Not Quote or Cite
-------
1 Additional epidemiological studies have been conducted on associations of ambient CO
2 with other cardiovascular effects. For example, preliminary evidence of a link between exposure
3 to CO and alteration of blood markers of coagulation and inflammation in individuals with CAD
4 or CVD has been provided by a few well conducted and informative studies (ISA, Table 5-6;
5 Delfino et al., 2008; Liao et al., 2005). As noted by the ISA, however, further studies are
6 warranted to investigate the role of these markers in prothrombotic events and their possible
7 contribution to the pathophysiology of CO-induced aggravation of ischemic heart disease (ISA,
8 section 5.2.1.8). Other epidemiological studies (including field and or panel studies) also
9 provide some evidence of a link between CO exposure and heart rate and heart rate variability
10 (ISA, section 5.2.1.1). With regard to the two of three studies reporting a positive association
11 with heart rate, the ISA concluded that "further research is warranted" to corroborate the results,
12 while the larger number of studies for heart rate variability parameters is characterized as mixed
13 (ISA, p. 5-15). Additionally, of the two studies of electrocardiogram changes indicative of
14 ischemic events (ISA, section 5.2.1.2), one found no association and, in the other study, unlike
15 the association with black carbon in that study, the association with CO did not remain
16 significant in multipollutant models (ISA, p. 5-16). A limited number of epidemiological studies
17 (Bell et al., 2009; Linn et al., 2000) have investigated hospital admissions for stroke (including
18 both hemorrhagic and ischemic forms) and generally report small or no associations with
19 ambient CO concentrations (ISA, section 5.2.1.9, Table 5-8 and Figure 5-3).
20 Our consideration of these epidemiological studies here reflects our understanding of CO
21 mechanisms and effects which is informed by the long-standing body of evidence for CO
22 summarized above. Figure 2-1 below presents a conceptual model of the pathway from CO
23 exposures to these effects. Figure 2-2 separately shows elements of this conceptual model
24 investigated in clinical studies in contrast to those in the epidemiological studies discussed
25 above. More specifically, the clinical studies provide linkages between directly measured
26 controlled short-term CO exposures and COHb levels in CAD patients and between those COHb
27 levels and effects in that study group, specifically, markers of myocardial ischemia. In contrast,
28 the epidemiological studies report associations between ambient CO concentrations at fixed-site
29 monitors and cardiovascular-related health outcomes (i.e., hospital admissions and emergency
30 room visits). However, our ability to integrate the evidence from these epidemiological studies
31 with our knowledge of CO-related effects based on the clinical evidence is limited in a few
32 aspects. Most particularly, we lack information on the actual CO exposures and associated
33 internal COHb levels that may have elicited the population response observed in the
34 epidemiological studies, including the extent to which the response may be driven by CO
35 exposures other than the ambient CO concentrations. Moreover, it is also unknown how those
36 unmeasured exposures relate to the exposure concentrations of the clinical studies (e.g., 1 hour at
March 2010 2-14 Draft - Do Not Quote or Cite
-------
1 approximately 50-200 ppm). We are also limited in our understanding of the specific
2 relationship between the changes in indicators of myocardial ischemia observed in the controlled
3 human exposure studies and the hospital admissions assessed in the epidemiological studies. In
4 consideration of the epidemiological studies for cardiovascular outcomes in light of the larger
5 body of evidence, the ISA notes that the "known role of CO in limiting C>2 availability lends
6 biological plausibility to ischemia related health outcomes following CO exposure", providing
7 coherence between the two sets of findings. The ISA additionally recognizes a lack of clarity as
8 to whether "the small changes in COHb associated with ambient CO exposures results in
9 substantially reduced O2 delivery to tissues" (ISA, p. 5-48).
10
11
Ambient CO
Nonambient
CO
^
' Exposure
r
>
Endogenous
r.n
-
Internal
Dose
(e.g., COHb)
-
Effects
(e.g., markers
of myocardial
Ischemia)
— »
Health
Outcomes
(e.g., hospital
admissions)
Figure 2-1. A conceptual model of CO source-to-health outcome pathway.
12
13
14
15
d CO Exposure
Baseline CO
(endogenous
+ background)
-
Internal
Dose
(COHb)
Clinical Studies (CAD Patients)
Effects
(markers
of myocardial
Ischemia)
Ambient CO
— ^
Epidemiological Studies
Health
Outcomes
(e.g., IHD
hospital
admissions)
Figure 2-2. Components of CO source-to-outcome conceptual model measured in
controlled human exposure and epidemiological studies discussed in this
document.
March 2010
2-15
Draft - Do Not Quote or Cite
-------
1 Toxicological evidence presented in the current review provides further evidence of the
2 effects of CO in the cardiovascular system (e.g., aortic injury, microvascular permeability,
3 vascular remodeling, ventricular hypertrophy) occurring at laboratory animal CO exposure levels
4 ranging from 35 to 200 ppm (ISA, section 5.2.5).
5 At the time of the last review, there was evidence for effects other than cardiovascular
6 morbidity, including neurological, respiratory and developmental effects. With regard to
7 neurological effects, acute exposures to CO have long been known to induce CNS effects such as
8 those observed with CO poisoning, although limited equivocal evidence available at the time of
9 the last review included indications of some neurobehavioral effects to result from CO exposures
10 resulting in a range of 5-20% COHb (2000 AQCD, section 6.3.2). No additional clinical or
11 epidemiological studies are now available that investigated such effects of CO at ambient levels
12 (ISA, section 5.3). With regard to potential effects of CO on birth outcomes and developmental
13 effects, the currently available evidence includes limited but suggestive epidemiologic evidence
14 for a CO-induced effect on preterm-birth, birth defects, decrease in birth weight, other measures
15 of fetal growth, and infant mortality (ISA, section 5.4.3). The available animal toxicological
16 studies provide some support and coherence for these birth and developmental outcomes,
17 although a clear understanding of the mechanisms underlying potential reproductive and
18 developmental effects is still lacking (ISA, section 2.5.3). There is limited evidence of an
19 association between short-term exposure to CO and respiratory-related outcomes, although only
20 preliminary evidence is available regarding a mechanism that could provide plausibility for CO-
21 induced effects. Thus, while there is some additional evidence on neurological, respiratory and
22 developmental effects, it remains limited.
23 In summary, rather than altering our conclusions from the previous review, the current
24 evidence provides continued support and some additional strength to our previous conclusions
25 regarding the health effects associated with exposure to CO and continues to indicate
26 cardiovascular effects, particularly effects related to the role of CO in limiting oxygen
27 availability, as those of greatest concern at low exposures.
28 • Does the current evidence continue to support a focus on COHb levels as the most
29 useful indicator of CO exposures and the best biomarker to characterize potential
30 for health effects associated with exposures to ambient CO? Or does the current
31 evidence provide support for a focus on alternate dose indicators to characterize
32 potential for health effects?
33 As discussed in both the 2000 AQCD (USEPA, 2000) and the ISA, the best characterized
34 mechanism of action of CO is tissue hypoxia caused by binding of CO to hemoglobin to form
35 COHb. Increasing levels of COHb with subsequent decrease in oxygen availability for organs
36 and tissues are of concern in people with pre-existing heart disease who have compromised
March 2010 2-16 Draft - Do Not Quote or Cite
-------
1 compensatory mechanisms (e.g., arteries unable to vasodilate, reduced capacity to increase
2 cardiac output) such as those affected with coronary heart disease. The integrative review of
3 health effects of CO indicates that "the clearest evidence indicates that individuals with coronary
4 artery disease are most susceptible to an increase in CO-induced health effects" (ISA, section
5 5.7.8) and the evidence, including that from clinical studies described in addressing the previous
6 question (regarding health effects associated with exposure to CO), continues to support levels of
7 COHb as the most useful indicator of CO exposure that is related to the health effects of CO of
8 major concern.
9 Alternative mechanisms of CO-induced effects primarily associated with CO's ability to
10 bind heme-containing proteins other than hemoglobin and myoglobin, and involving a wide
11 range of cellular tissues and CO concentrations, have been described in the 2000 AQCD
12 (USEPA, 2000, section 5.8) and in the ISA (ISA, section 5.1.3). The ISA notes that collectively
13 the older toxicological studies demonstrated that exposure to high concentrations of CO resulted
14 in altered functions of heme proteins other than myoglobin and hemoglobin which have the
15 potential to interfere with basic cell and molecular processes that could lead to dysfunction
16 and/or disease. More recent toxicological studies described in the ISA, conducted at CO
17 exposure levels ranging from 35 to 100 ppm have provided evidence of non-hypoxic
18 mechanisms involving binding of CO to reduced iron in heme proteins by which CO may be
19 causing cellular injury or toxicity (ISA, section 2.5). The ISA notes that these mechanisms may
20 or may not be interrelated and include alteration of nitric oxide signaling, inhibition of
21 cytochrome C oxidase, heme loss from protein, disruption of iron homeostasis and alteration of
22 cellular redox status. Less understood mechanisms noted in the ISA are alteration in ion channel
23 activity and modulation of protein kinases signaling pathways. The ISA notes that "CO may be
24 responsible for a continuum of effects from cell signaling to adaptive responses to cellular injury,
25 depending on intracellular concentrations of CO, heme proteins and molecules which modulate
26 CO binding to heme proteins" (ISA, section 5.1.3.3).
27 Although there is evidence of additional biomarkers potentially associated with effects of
28 CO, new research is needed to further understand these biomarkers and their linkage to CO-
29 induced effects in at-risk populations. Therefore, the evidence indicates that COHb continues to
30 be the most useful and well supported indicator of CO exposures and the best biomarker to
31 characterize the potential for health effects associated with exposures to ambient CO.
March 2010 2-17 Draft - Do Not Quote or Cite
-------
1 • Does the current evidence alter our understanding of populations that are
2 particularly susceptible to CO exposures? Is there new evidence that suggest
3 additional susceptible populations that should be given increased focus in this
4 review?
5 The term susceptibility has been used to recognize populations that have a greater
6 likelihood of experiencing effects related to ambient CO exposure (ISA, section 5.7). Thus,
7 susceptible populations are at greater risk of CO effects and are referred to as at-risk in the
8 discussion here. This increased likelihood of response to CO can potentially result from many
9 factors, including pre-existing medical disorders or disease state, lifestage, gender, lifestyle or
10 increased exposures (ISA, section 5.7).
11 The current evidence, while much expanded in a number of ways, continues to support
12 our conclusions from the previous review regarding the key and best-characterized at-risk
13 population for exposure to ambient CO. Based on the evidence from clinical studies also
14 considered in the last review, with which the now much-expanded epidemiological evidence
15 base is coherent, the population with pre-existing cardiovascular disease associated with
16 limitation in oxygen availability continues to be the best characterized population at risk of
17 adverse CO-induced effects, with coronary artery disease (CAD) recognized as "the most
18 important susceptibility characteristic for increased risk due to CO exposure" (ISA, section
19 2.6.1). An important factor determining the increased susceptibility of this population is their
20 inability to compensate for the reduction in oxygen levels due to an already compromised
21 cardiovascular system. Individuals with a healthy cardiovascular system (i.e., with healthy
22 coronary arteries) have operative physiologic compensatory mechanisms (e.g., increased blood
23 flow and oxygen extraction) for CO-induced hypoxia and are unlikely to be at increased risk of
24 CO-induced effects (ISA, p. 2-10).14 In addition, the high oxygen consumption of the heart,
25 together with the inability to compensate for the hypoxic effects of CO make the cardiac muscle
26 of a person suffering with coronary artery disease a critical target for the hypoxic effects of CO.
27 In the current review, recognition of susceptibility of the population with pre-existing
28 cardiovascular disease, such as CAD, is supported by the expanded epidemiological database,
29 which includes a number of studies reporting significant increases in hospital admissions for
30 IHD, angina and MI in relation to CO exposures (ISA, section 2.7). Further support is provided
31 by epidemiologic studies (Mann et al., 2002; and Peel et al., 2007) of increased hospital
32 admissions and emergency department visits for IHD among individuals with secondary
14 The other well-studied individuals at the time of the last review were healthy male adults that
experienced decreased exercise duration at similar COHb levels during short term maximal exercise. This
population was of lesser concern since it represented a smaller sensitive group, and potentially limited to individuals
that would engage in vigorous exercise such as competing athletes (1991 AQCD, section 10.3.2).
March 2010 2-18 Draft - Do Not Quote or Cite
-------
1 diagnoses for other cardiovascular outcomes including arrhythmia and congestive heart failure
2 (ISA, section 5.7), and toxicological studies reporting altered cardiac outcomes in animal models
3 of cardiovascular disease (ISA, section 5.2.1.9).
4 Cardiovascular disease comprises many types of medical disorders, including heart
5 disease, cerebrovascular disease (e.g., stroke), hypertension (high blood pressure), and peripheral
6 vascular diseases. Heart disease, in turn, comprises several types of disorders, including
7 ischemic heart disease (CHD or CAD, myocardial infarction, angina), congestive heart failure,
8 and disturbances in cardiac rhythm (2000 AQCD, section 7.7.2.1). Types of cardiovascular
9 disease other than those discussed above may also contribute to increased susceptibility to the
10 adverse effects of low levels of CO (ISA, section 5.7.1.1). For example, some evidence with
11 regard to other types of cardiovascular disease such as congestive heart failure, arrhythmia, and
12 non-specific cardiovascular disease, although more limited for peripheral vascular and
13 cerebrovascular disease, indicates that "the continuous nature of the progression of CAD and its
14 close relationship with other forms of cardiovascular disease suggest that a larger population
15 than just those individuals with a prior diagnosis of CAD may be susceptible to health effects
16 from CO exposure" (ISA, p. 5-117).
17 Although there was little experimental data available at the time of the last review to
18 adequately characterize specific health effects of CO at ambient levels for other potentially at-
19 risk populations, several other populations were identified as being potentially more at risk of
20 CO-induced effects due to a number of factors. These factors include pre-existing diseases that
21 could inherently decrease oxygen availability to tissues, lifestage vulnerabilities (e.g., fetuses, the
22 elderly), gender, lifestyle, medications or alterations in the environment (e.g., increased altitude).
23 This is consistent with the ISA conclusions in the current review which recognize other
24 populations that may be potentially susceptible to the effects of CO as continuing to include:
25 those with other pre-existing diseases that may have already limited O2 availability or increased
26 COHb production or levels, such as people with obstructive lung diseases, diabetes and anemia;
27 older adults and fetuses during critical phases of development; commuters and those living near
28 heavily traveled roadways; visitors to high-altitude locations; and people ingesting medications
29 and other substances that enhance endogenous or metabolic CO production.
30 Preliminary evidence from epidemiological, controlled human exposure, and
31 toxicological studies indicate that people with obstructive lung disease (e.g., COPD patients with
32 underlying hypoxia, asthmatics) may be a potentially susceptible population (ISA, section
33 5.7.1.2). Overall, the few available epidemiological studies have reported weak, positive
34 associations between ambient CO and CVD hospital admissions for individuals with underlying
35 COPD. Additionally, a controlled human exposure study of individuals with COPD reported
36 that two patients experienced COPD exacerbation and a slight anti-inflammatory effect during
March 2010 2-19 Draft - Do Not Quote or Cite
-------
1 CO exposures of 100-125 ppm for 2 hours (ISA, section 5.5.1.2). Other epidemiological studies
2 (ISA, section 5.5.2.2) have reported weak associations in asthmatics, which constitute another
3 population that can experience exercise-induced airflow limitation. Preliminary evidence was
4 also shown in one animal toxicological study (Ohio et al., 2008), indicating mild pulmonary
5 inflammation upon exposure to 50 ppm CO.
6 With regard to other potentially at-risk populations, there is also limited epidemiological
7 data showing associations between CO exposures and diabetes (ISA, section 5.7.1.3).
8 Epidemiologic studies (Pereira Filho et al., 2008; Zanobetti and Schwartz, 2001) provide
9 suggestive evidence that CVD patients with diabetes may be at greater risk of emergency
10 department visits and hospital admissions than those without diabetes (ISA, section 5.7.1.3).
11 Inferences may also be drawn from results from panel studies for individuals with metabolic
12 syndrome15 that observed associations between short-term exposures to CO and changes in heart
13 rate variability parameters (ISA, section 5.2.1.1) and from a toxicological study providing
14 evidence of vascular dysfunction associated with increases in endogenous CO in an animal
15 model of metabolic syndrome (ISA, section 5.7.1.3). People with anemia who have reduced
16 oxygen-carrying capacity and/or higher baseline COHb levels are an additional population
17 considered to be potentially susceptible to the hypoxic effects of CO. However, there are no
18 controlled human studies or epidemiological studies that have specifically examined the CO-
19 related health effects in individuals with anemia (ISA, section 2.6.1, p. 2-11).
20 Older adults (65+) have been considered as a potentially susceptible population to the
21 effects of CO, primarily due to the increased prevalence of cardiovascular disease among this
22 population when compared to all age groups or lifestages. There is limited epidemiologic
23 evidence showing associations between short-term exposure to CO and increases in IHD or
24 myocardial infarction (MI) hospital admissions among older adults as compared to all age groups
25 or younger adults. The combination of this limited epidemiological data and the fact that older
26 adults have a higher prevalence of CAD than the general population, indicates that older adults
27 are a potentially susceptible population for increased health effects due to CO exposure (ISA,
28 section 5.7.2.1).
29 The developing fetus has been a population considered to be at potential risk of CO-
30 induced effects due to the higher affinity of their hemoglobin for CO when compared to that of
31 adults. Limited evidence indicates an association of CO exposure during early pregnancy with
32 pre-term births and birth defects (ISA, p. 2-8). Toxicological studies reporting effects in
33 laboratory animals including decrements in birth weight, reduced prenatal growth, and effects on
34 the heart provide support to the epidemiological evidence.
15 These patients share risk factors with diabetics; see ISA section 5.7.1.3.
March 2010 2-20 Draft - Do Not Quote or Cite
-------
1 Gender has also been considered as a possible source for susceptibility. However, the
2 evidence is inconclusive based on the limited epidemiological data, and the gender-specific
3 variability in CO endogenous production (ISA, section 5.7.3). Increased altitude and physical
4 activity through their effect on CO uptake and elimination has also been considered to potentially
5 influence susceptibility. For example, residents of low altitudes visiting high altitudes,
6 especially the elderly and those with CAD (Leaf and Kleinman, 1996; Kleinman et al., 1998),
7 may be at greater risk from added effects of ambient CO than adapted residents (ISA, p. 2-12).
8 Other less certain susceptibility factors have been considered, such as those pertaining to the
9 taking of medications that may alter CO production (ISA, p. 2-12). Lastly, people experiencing
10 increased CO exposures, such as those spending relatively greater amounts of time in
11 microenvironments with relatively higher levels of ambient CO have been considered to be
12 potentially at risk to CO-induced effects (ISA, section 5.7.6).
13 As we recognize the potential susceptibility of the populations identified above, we also
14 note the lack of information on specific COHb levels that may be associated with health effects
15 in these other groups as well as a way to relate the specific evidence available for the CAD
16 population to these other populations.
17 In summary, the current evidence continues to support the identification of people with
18 cardiovascular disease as having susceptibility to CO-induced health effects (ISA, 2-12), with
19 those having CAD as the best characterized population with increased susceptibility to CO-
20 induced health effects (ISA, sections 5.7.1.1 and 5.7.8). An important susceptibility
21 consideration for this population is the inability to compensate for CO-induced hypoxia since
22 individuals with CAD have an already compromised cardiovascular system. Included in this
23 susceptible population are those with angina pectoris (cardiac chest pain), those who have
24 experienced a heart attack, and those with silent ischemia or undiagnosed IHD (AHA, 2003).
25 People with other cardiovascular diseases, particularly heart diseases, are also at risk of CO-
26 induced health effects. We also recognize other populations potentially susceptible to CO-
27 induced effects, most particularly those with other pre-existing diseases that may have already
28 limited oxygen availability, increased COHb levels, or increased endogenous CO production,
29 such as people with obstructive lung diseases, diabetes and anemia; however, information
30 characterizing susceptibility is limited.
March 2010 2-21 Draft - Do Not Quote or Cite
-------
1 • Does the current evidence alter our conclusions from the previous review
2 regarding the levels of CO in ambient air associated with health effects?
3 At the time of the last review, EPA's conclusions regarding concentrations of CO in
4 ambient air that might be associated with risk of health effects were drawn from the combined
5 consideration of the evidence of COHb levels for which cardiovascular effects of concern had
6 been reported and the results of an exposure and dose modeling assessment (59 FR 38906). As
7 described in more detail in section 2.1.1 above, the key effects judged to be associated with CO
8 exposures resulting from concentrations observed in ambient air were cardiovascular effects, as
9 measured by decreased time to onset of exercise-induced angina and to onset of ECG ST-
10 segment depression (59 FR 38913).
11 Levels of COHb that have been associated with different types of effects in clinical
12 studies are summarized in Table 2-2 below. At the time of the last review, decreases in time to
13 onset of exercise-induced angina (a symptom of myocardial ischemia) had been documented in
14 multiple studies at post-exposure COHb levels ranging from 2.9 to 5.9% (CO-Ox), which
15 represented incremental increases of 1.5-4.4% COHb from baseline (Adams et al., 1988; Allred
16 et al., 1989a, 1989b, 1991; Anderson et al., 1973; Kleinman et al., 1989, 1998; Sheps et al.,
17 1987). The matched measurements available from Allred et al. (1989a, 1989b, 1991) of CO-Ox
18 and gas chromatography, the method generally recognized to be the more accurate for COHb
19 levels below 5% (ISA, section 5.2.4), indicate that CO-Ox measurements of 2.9 to 3.0% COHb
20 generally correspond to GC measurements on the order of 2-2.4%.16 Other groups for which
21 clinical studies have reported CO-related effects at slightly higher COHb levels include effects in
22 subjects with cardiac arrhythmias and effects on exercise duration and maximal aerobic capacity
23 in healthy adults. Among the studies of myocardial ischemia indicators in patients with CAD,
24 the two studies involving the lowest experimental CO exposures (which resulted in average
25 increases in COHb of about 1.5% over pre-exposure baseline) were Anderson et al. (1973)17 and
26 the more recent Allred et al. (1989a, 1989b, 1991) to which we give primary attention in this
27 review (discussed in more detail above). Neither study provided evidence of a measurable
28 threshold at the lowest experimental CO exposures and associated COHb levels assessed (mean
29 of 2.0-2.4% COHb, GC). Allred et al. (1989a, 1989b, 1991) further reported a dose-response
30 relationship between the increased COHb levels and the response of the assessed indicators of
31 myocardial ischemia (Allred et al., 1989a, 1989b, 1991). While this evidence informed our
16 In the lower CO exposure group, the post-exposure mean COHb was 3.21% by CO-Ox and 2.38% by
GC, while the post-exercise mean COHb was 2.65% by CO-Ox and 2.00% by GC (Allred et al., 1989a, 1989b,
1991).
17 The study by Anderson et al. (1973) did not use GC to measure COHb levels, and reported reduced
exercise duration due to increased chest pain at CO exposures resulting in 2.9% COHb (CO-Ox), representing a
1.6% increase in average COHb levels over baseline.
March 2010 2-22 Draft - Do Not Quote or Cite
-------
1 conclusions regarding COHb levels associated with health effects, the CO exposure
2 concentrations employed in the studies to achieve these COHb levels were substantially above
3 ambient concentrations. Thus, an exposure and dose assessment was performed to consider the
4 COHb levels that might be attained as a result of exposures to ambient CO allowed under the
5 current NAAQS, as described in section 2.1.1 above.
6 Since the time of the last review, there have been no new controlled human exposure
7 studies specifically designed to evaluate the effects of CO exposure in susceptible populations at
8 COHb levels below 2%. Thus, similar to the last review, the multilaboratory study by Allred et
9 al.(1989a, 1989b, 1991) continues to be the study that has evaluated cardiovascular effects of
10 greatest concern (i.e., reduced time to exercise-induced myocardial ischemia as indicated by
11 ECG ST-segment changes and angina) at the lowest tested COHb levels (ISA, section 2.7). This
12 is currently considered the most rigorous and well designed study, presenting the most sensitive
13 analysis methods (GC used in addition to CO-Ox) to quantify COHb blood levels. Key findings
14 from that study with regard to levels of CO associated with health effects include the following:
15 • Increases in COHb to mean levels of 2.0% and 3.9% from baseline (0.6%) significantly
16 reduced exercise time required to induce markers of myocardial ischemia in CAD
17 patients.18
18 • The associated dose-response relationship between incremental changes in COHb and
19 change in time to myocardial ischemia in CAD patients indicates a 1.9% and 3.9%
20 reduction in time to onset of exercise-induced angina and ST-segment change,
21 respectively, per 1% increase in COHb concentration without evidence of a measurable
22 threshold.
23 Since the last review, there have been numerous improvements to the exposure and
24 COHb models that we employ to estimate exposure and dose for the current review. The results
25 of modeling using these improved tools in the current review and our conclusions based on it
26 with regard to the expectation for COHb levels of concern to occur in the at-risk population
27 under differing air quality conditions are described in section 2.2.2 below.
18 Relative to baseline (0.6% mean COHb), mean COHb levels of 2.0-2.4% and 3.9-4.7% were associated
with decreases in time to onset of exercise-induced ST-segment changes of 5.1% (p=0.01) and 12.1%(p<0.001),
respectively. Similarly, the time to onset of exercise-induced angina was shortened by 4.2% (p=0.027) and 7.1%
(p=0.002) for these two mean levels of COHb (Allred et al., 1989a, 1989b. 1991).
March 2010 2-23 Draft - Do Not Quote or Cite
-------
1 Table 2-2. Carboxyhemoglobin levels and reported effects in CAD patients and healthy
2 adults.
Study
Population
Patients with
coronary
artery
disease*
Healthy
adults
Effect
Reduction in time to exercise-induced onset of
myocardial ischemia
5. 1 % decrease in time to ST- segment
change, 4.2% decrease in time to angina
12.1% decrease in time to ST- segment
change, 7.1% decrease in time to angina
Reduction in exercise duration prior to angina
and/or ST segment changes
Exercise-induced arrhythmia
Reduction in exercise duration and maximal
aerobic capacity
Equivocal behavioral effects (hand eye
coordination, vigilance, continuous performance
of critical tasks)
%COHb
(mean)
GC
2.0-2.4
3.9-4.7
NA"
NA
NA
NA
CO-Ox
2.7-3.2
4.7-5.6
2.8-5.2
6
3-20
5
%COHb,
increase
over
baseline
1.4-1.7
3.3-4.0
1.5-4.2
Study
Allredetal, 1989a,
1989b, 1991
Adams etal., 1988;
Anderson etal.,
1973; Kleinmanet
al., 1989; Kleinman
etal., 1998;Sheps
etal., 1987
Shepset. al., 1990,
1991
Drinkwater et al.,
1974;Ekblomand
Huot, 1972;
Horvath, 1975;
Raven etal.,
1974a, 1974b
Benignus et. al.,
1987, 1990a,
1990b;Putzet. al.,
1976, 1979
* All studies involved subjects with reproducible exercise-induced angina.
** NA = not available.
3
4 Since the time of the last review, a number of additional epidemiological studies have
5 been conducted of health outcome associations with different ambient CO metrics (e.g., ISA,
6 section 5.2). The majority of these epidemiological studies, conducted in selected urban areas of
7 the U.S. or Canada, were designed to evaluate the effects of multiple pollutants (i.e. PM2.5, NC>2,
8 SC>2, 63 and CO). As discussed above, a number of these studies have reported associations
9 between concentrations of CO measured at fixed-site ambient monitors and cardiovascular
10 endpoints other than stroke, particularly hospitalizations and emergency department visits for
11 specific cardiovascular health outcomes including ischemic heart disease (IHD), congestive heart
12 failure (CHF) and cardiovascular diseases (CVD) as a whole (Bell et al., 2009; Koken et al.,
13 2003; Linn et al., 2000; Mann et al., 2002; Metzger et al., 2004; Peel et al., 2007; Symons et al.,
March 2010
2-24
Draft - Do Not Quote or Cite
-------
1 2006; Tolbert et al., 2007; Wellenius et al., 2005). In general, these studies report positive
2 associations and a number of these associations are statistically significant (ISA, sections 5.2.3
3 and 5.2.1.9; Table 2-1 above).
4 In considering these epidemiological studies in the case of CO, we recognize that there is
5 coherence between the available clinical and epidemiological evidence with regard to the health
6 effects of CO in the cardiovascular system (primarily for ischemic related events). As noted
7 above, however, we also recognize several gaps between the two lines of evidence which
8 complicate their integration, particularly with regard to ambient exposures and ambient
9 concentrations associated with health effects of concern. As discussed in the ISA, the
10 epidemiological studies reported associations with emergency department visits and/or hospital
11 admissions for IHD and other cardiovascular disease-related outcomes that are plausibly related
12 to the effects on physiological indicators of myocardial ischemia (e.g., ST-segment changes)
13 demonstrated in the controlled human exposure studies, providing coherence between the two
14 sets of findings. However, there is appreciable uncertainty with regard to the specific
15 relationship between these two types of measures of cardiovascular morbidity. Furthermore, the
16 clinical studies document relationships between controlled elevated exposures to CO and specific
17 levels of an internal dose metric, COHb, which elicited specific responses in CAD patients. This
18 informs our interpretation of the associations we observe in the epidemiological studies, although
19 COHb has not been monitored in the epidemiological studies and the personal CO exposure
20 concentrations eliciting the outcomes observed, including the contributions from ambient and
21 nonambient CO, are unknown (as illustrated in Figure 2-2 above).19 Further, to achieve their
22 experimental target levels of COHb, the clinical studies used short (approximately 1-hour)
23 exposures to concentrations of CO that were substantially higher than current commonly
24 occurring ambient concentrations (ISA, section 2.7). The relationship of the clinical exposure
25 concentrations to exposure concentrations occurring in the populations studied in the
26 epidemiological studies is unknown, but the clinical exposures may be higher. For example,
27 individual exposure concentrations for the lower clinical exposure ranged from 42 to 202 ppm
28 (Allred et al., 1989b). Experimental studies have not been conducted at COHb levels that might
29 be elicited in response to exposures to lower CO concentrations such as those within the ranges
30 of current maximum ambient concentrations (ISA, section 2.7) or closer to personal exposure
31 concentrations more commonly expected currently (e.g., ISA, Figure 3-45). Gaps in our
32 understanding of the role of nonambient CO exposures and their contribution to COHb levels
19 As recognized in the ISA in consideration of the epidemiological studies, the known role of CO in
limiting O2 availability lends biological plausibility to ischemia-related health outcomes following CO exposures,
although "it is not clear whether the small changes in COHb associated with ambient CO exposures results in
substantially reduced O2 delivery to tissues" (ISA, p. 5-48).
March 2010 2-25 Draft - Do Not Quote or Cite
-------
1 also complicates our ability to disentangle ambient concentrations that may be related to health
2 effects.
3 In considering these epidemiological studies in the case of CO, we recognize several
4 uncertainties that complicate the quantitative interpretation with regard to ambient concentrations
5 that might be eliciting the reported health outcomes. These uncertainties relate to conclusions
6 regarding CO as the pollutant eliciting the effect reported in these studies and to our
7 understanding of the ambient CO concentrations (and total concentrations) to which study
8 subjects demonstrating effects are exposed.
9 • With regard to the first, a major uncertainty relates to the difficulty in determining the
10 extent to which ambient CO is independently associated with cardiovascular effects or
11 if CO at ambient levels is acting as a surrogate for the effects of another traffic-related
12 pollutant or mixture of pollutants (ISA, section 5.2.3). As noted in the ISA, in
13 interpreting the epidemiological evidence for cardiovascular morbidity "[i]t is difficult
14 to determine from this group of studies the extent to which CO is independently
15 associated with CVD outcomes or if CO is a marker for the effects of another traffic-
16 related pollutant or mix of pollutants. On-road vehicle exhaust emissions are a nearly
17 ubiquitous source of combustion pollutant mixtures that include CO and can be an
18 important contributor to CO in near-road locations" (ISA, p. 5-40 - 5-41). We also
19 note CAS AC's recognition of the potential for co-pollutants to pose confounding to be
20 "particularly problematic for CO" and the need to give consideration to the possibility
21 of CO serving as a surrogate for a mixture of fossil- fuel-combustion-related pollutants
22 (Brain and Samet, 2010).
23 • To identify the CO-specific effect, and assess and control for potential confounding of
24 CO-associated health effects by other pollutants for which measurements are available,
25 two-pollutant regression models are a commonly used statistical method (ISA, section
26 1.6.3). Although CO associations may, in some studies, be slightly attenuated in
27 models adjusting for one other combustion-related pollutant, they generally remain
28 robust (Figures 5-6 and 5-7). However, there are several statistical issues influencing
29 results generated using multi-pollutant models, such as models that force all pollutants
30 to follow the same lag structure, the inclusion of correlated but noncausal variables,
31 and the potential omission of etiologically relevant co-pollutants. Although data for
32 co-pollutant modeling are available for some of the other criteria pollutants, data are
33 not available for additional combustion-related pollutants, which may or may not
34 include all pollutants that elicit the health effects (ISA, section 2.5.1). In light of the
35 uncertainty (referenced in discussion of the question below) regarding the biological
36 plausibility of CO-related effects at low ambient concentrations, it is possible that the
37 CO risk estimates ascribed to measured ambient CO levels in epidemiological studies
38 is instead another pollutant that is highly correlated with CO, but was not measured.
39 • With regard to the second, the presence of exposure error and/or measurement error
40 from reliance on fixed-site ambient CO monitors, a large portion of which are
41 measuring CO concentrations at or below detection limits can result in the over or
42 underestimation of the effects of the poorly measured pollutant. This limitation can
43 add uncertainty to the association between ambient concentrations of CO and health
March 2010 2-26 Draft - Do Not Quote or Cite
-------
1 effects estimates (ISA, p. 3-91) and constrain their interpretation with regard to CO
2 exposure levels (Brain and Samet, 2010).
3 We have considered here the epidemiological studies in the U.S. for ischemia-related
4 outcomes. In Table 2-2, we focused on the U.S. studies of CHD hospital admissions for CHD
5 (i.e., IHD and MI), that provided the strongest and most consistent weight of evidence (ISA,
6 Figure 5-2, pp. 5-24 to 5-26). In Table 2-3, we present the U.S. studies for the broader category
7 of all CVD hospital admissions that also contributed to the ISA's causality determination (ISA,
8 Figure 5-5, pp. 5-33 to 5-36). In focusing on these studies, we did not include studies of
9 associations with CHF, a chronic condition for which there are multiple causes and for which the
10 evidence is less clear.20 We have focused here on studies for IHD, MI and all CVD outcomes
11 (which are inclusive of IHD), which we consider to be key due to the more direct conceptual
12 linkages that can be drawn between the these outcomes and the myocardial ischemia effects of
13 the CAD clinical studies.
14 Of particular interest in addressing the question posed here is information pertaining to
15 CO concentrations in ambient air that meet the current standard. While, the full set of these
16 epidemiological studies, including those reporting associations with ambient CO concentrations
17 under conditions when the current standards were not met, provide support to the previous
18 evidence regarding cardiovascular effects of CO, it is the studies involving air quality conditions
19 in which the current standards were met that are most informative within the context of
20 considering the adequacy of the current standards.21 Accordingly, in considering these
21 epidemiological studies, we have considered the air quality conditions during the periods of
22 study.
23 The areas of focus across these studies of IHD and CVD outcomes (inclusive of IHD)
24 include Atlanta, urban areas of California and a group of 126 urban counties22 and the multi-year
25 time periods for these studies together span the period from 1988 through 2005 (Tables 2-3 and
26 2-4 below). During the time periods of study, only the three Atlanta studies (Metzger et al.,
27 2004; Peel et al., 2007; Tolbert et al., 2007) did not include years in which ambient CO
28 concentrations exceeded the 8-hour standard (Tables 2-3 and 2-4). Based on data reported to
29 AQS for the general areas studied, ambient CO concentrations during a portion of the time
30 periods studied in the other three of these analyses exceeded the level of the current 8-hour
31 NAAQS (Bell et al., 2007; Linn et al., 2000; Mann et al., 2002). Among these three studies, all
20 We note that two of the studies listed in Tables 2-3 and 2-4 also reported positive associations with CHF
(ISA, pp. 5-31 to 5-33). Additionally, there are three other studies of CHF associations shown in Table 2-1 above,
two of which were statistically significant, with one of those with significant results focused on an area in which the
current 8-hour standard was just met (Koken et al., 2003; Symons et al., 2006; Wellenius et al., 2005).
21 As at the time of the last review, the 8-hour standard continues to be the controlling standard.
22 The 126 counties span some 40 states, including the District of Columbia.
March 2010 2-27 Draft - Do Not Quote or Cite
-------
1 of which reported statistically significant risk estimates, the highest 8-hour design values ranged
2 from 11.6-23.4 ppm, as compared to the 8-hour standard of 9 ppm.23 In the Atlanta area, which
3 was the focus for the other three studies (Peel et al., 2007; Metzger et al., 2004; Tolbert et al.,
4 2007), the second highest non-overlapping 8-hour average CO concentration in a year (i.e., the
5 design value for the 8-hour standard) ranged from 3.2 up to 5.3 ppm over the combined 12 years
6 of study. One of these three studies (Tolbert et al., 2007) reported a statistically significant risk
7 estimate for cardiovascular hospital visits associated with CO concentrations, while the other two
8 did not. In the Atlanta area during the period studied, the 99th percentile daily maximum 8-hour
9 CO concentration in a year ranged up to 4.9 ppm, while the yearly 99th percentile daily
10 maximum 1-hour CO concentration ranged up 7.9 ppm. The studies in which CO concentrations
11 exceeded the current NAAQS during some portion of the study period accordingly are less
12 informative to our overarching question regarding the adequacy of the protection afforded by the
13 current standard. We note that studies reporting associations with IHD or total CVD outcomes
14 for which air quality data reported to AQS do not indicate that the current 8-hour standard was
15 exceeded during the study period are limited to those for Atlanta.
23 Staff also note that the national-scale study of 126 urban counties by Bell et al. (2009) included a subset
analysis that evaluated associations for all cardiovascular outcomes, adjusted for NO2, after restricting the dataset to
days with a 1-hour daily maximum CO concentration less than 35 ppm, which is the level of the 1-h standard
although its form is the 2nd maximum in a year. This subset analysis found a statistically significant effect estimate
identical to the estimated risk with the full data set. Further subset analyses by Bell et al. in this study using data
below cutoff values of 1-10 ppm all found positive, statistically significant associations (with adjustment for NO2),
although confidence intervals increased at the lower cutoff values and the proportion of values near and below the
limit of detection grew larger.
March 2010 2-28 Draft - Do Not Quote or Cite
-------
1 Table 2-3. Air quality information for key U.S. epidemiological studies of hospital admissions for coronary heart disease
2 (ISA, Figure 5-2).
Study
Metzgeretal, 2004
Peel etal., 2007
Mann etal, 2002
Linn etal., 2000
Area
Atlanta
Atlanta
Atlanta
Air
Quality
Data
(from 3
monitors
in 20
counties)
CA
California
Air
Quality
Data
(from 31
monitors
in 4
counties)
CA
Study
Period
1993-2000
1993-2000
1993
1994
1995
1996
1997
1998
1999
2000
1988-1995
1988
1989
1990
1991
1992
1993
1994
1995
1992-1995
Health
Outcome
IHD
IHD
IHD
Ml
1-hour Daily Maximum
(ppm)
Mean
1.5*1
1.8*
99th
percentile
5.5-5.9
5.5-5.9
6.1
6.3-6.6
5.8-7.9
5.3-6.1
5-5.4
4.8-4.9
4.9-5.4
4.9-5.4
2-22
4.0-31.0
2.0-22.0
4.0-22.0
2.0-21
2.0-21.0
2.0-19.0
2.1-14.5
1.5-11.6
8-hour Daily Maximum
(ppm)
Mean
2.07*
99th
percentile
4.4
4.3-4.9
4.3-4.9
3.5-3.6
3.1-4.2
2.8-3.9
3-3.8
3.0-3.8
1.3-15.9
1.4-14.6
2.0-10.1
1.6-11.7
2-10.7
1.0-8.1
1.1-7.6
1.1-10.2
1-6.7
24-hour Average
(ppm)
Mean
1.5*
99th
percentile
2.4
2.7-3.3
2.4-3.2
2.0-2.7
2.3-2.4
2.1-2.4
2.0-2.3
1.5-1.8
1.2-10.9
1.7-6.9
1.3-7.8
2.0-8.7
1.0-9.6
1.1-8.0
1.1-8.9
1.0-7.6
1.1-8.3
Design Values (ppm)
for Study Areas
During Study Period
1-hour
4.6-16.3
4.6-16.3
6.5
7.1
16.3
6.6
5.8
5
6.3
4.6
16.5-32
32
28
23
29
25
20
20.8
16.5
16.5-25
8-hour
3.2-5.3
3.2-5.3
4.9
5.3
5.2
3.8
4.3
4.1
4.1
3.2
11.6-23.4
23.4
18.3
15.9
15.7
16.4
13.8
16
11.6
11.6-16.4
Table legend. This table presents information from key studies finding associations with coronary heart disease (ISA, section 5.2.3, Figure 5-2). The health outcomes evaluated include hospital
admissions for ischemic heart disease (IHD) and myocardial infarction (Ml). The 99th percentile (in ppm) of the CO concentration for the specific study metrics (i.e.,1-, 8-hour daily maximum, or
24-hour average) were derived from AQS data for monitors in the study areas during the study period. Design values are values of CO concentration in the statistical form of the standard;
presented here is the range of 2nd highest 1 -hour and 8-hour average concentrations per year at the highest monitor in each study area (obtained from the AQS database for the areas and time
periods of the studies).
* Risk estimates and CO concentrations for the study metric as reported by the study authors.
1This value for Metzger et al. (2004) is the median.
March 2010
2-29
Draft - Do Not Quote or Cite
-------
1 Table 2-4. Air quality information for key U.S. epidemiological studies of hospital admissions for CVD (ISA, Figure 5-5).
Study
Bell etal, 2009
Tolbertetal, 2007
Linn etal., 2000
Area
126 urban
counties
Atlanta
Atlanta Air Quality
Data (3 monitors
in 20 counties)
CA
Study Period
1999-2005
1993-2004
Health
Outcome
CVD 65+
CVD
1-hc
ME
mean
1.3*
1.6*
)ur Daily
iximum
ppm)
99th
percentile
1.2-22.1
5.3-5.4
8-hour Daily
Maximum (ppm)
mean
99th
percentile
24-hour Average
(ppm)
mean
99th
percentile
Design Values (ppm)
for Study Areas
During Study Period
1-hour
17.5-33.5
3.5-16.3
8-hour
8.8-24.3
2.5-5.3
For 1993-2000, see Table 2-2 above
2001
2002
2003
2004
1992-1995
CVD
3.8-4.1
4.2-4.3
2.6-3.3
2.9
2.4-3.0
2.2-2.5
2-2.1
2
1.5*
1.8-1.9
1.5-1.6
1.2-1.7
1.5
1.1-8.3
5.3
4.6
3.5
4.6
16.5-25
4.1
3.6
2.5
2.5
11.6-16.4
Table legend. This table presents information from key studies finding associations with CVD (ISA , section 5.2.3, Figure 5-5). The health outcomes evaluated in these studies
include hospital admissions for cardiovascular disease (CVD; Bell et al (2009) only included subjects 65 years old and above), with the associated risk estimates, unadjusted for
other pollutants. The 99th percentile (in ppm) of the CO concentration for the specific study metrics (i.e.,1-, 8-hour daily maximum, or 24-hour average) were derived from AQS data
for monitors in the study areas during the study period. Design values are values of CO concentration in the statistical form of the standard; presented here is the range of 2nd
highest 1-hour and 8-hour average concentrations per year at the highest monitor in each study area (obtained from the AQS database for the areas and time periods of the
studies).
* CO concentrations for the study metric as reported by the study authors.
March 2010
2-30
Draft - Do Not Quote or Cite
-------
1 In summary, although there is no new evidence regarding lower COHb levels associated
2 with effects, the evidence is much expanded with regard to epidemiological24 analyses of
3 ambient monitor concentrations, which observed associations between specific and overall
4 cardiovascular-related outcomes and ambient CO measurements. With regard to extending our
5 understanding of effects occurring below levels of CO observed in the clinical studies, however,
6 the epidemiological evidence for CO is somewhat limited. The epidemiological evidence lacks
7 measurements of COHb or personal exposure concentrations that would facilitate integration
8 with the clinical data. Furthermore, as discussed above, the full epidemiological evidence base
9 for IHD outcomes or CVD outcomes as a whole includes a number of studies involving
10 conditions in which the current standard was not met. Though these studies are informative to
11 consideration of the relationship of health effects to the full range of ambient CO concentrations,
12 they are less useful to informing our conclusions regarding adequacy of the current standards.
13 Consideration of the smaller set of studies under conditions where the current standards were met
14 is considered to better inform our consideration of the adequacy of the standards or conditions of
15 lower ambient concentrations. Among the few studies for IHD or CVD outcomes under
16 conditions where the current standards were met, one study reported statistical significance for
17 IHD or CVD outcomes, and all of these studies are limited to a single study area (i.e. Atlanta).
18 These limitations lessen the weight we give these studies for our purposes here in making
19 interpretations with regard to ambient concentrations of CO that may be eliciting health
20 outcomes, particularly in light of the availability of strong evidence from controlled human
21 exposure studies. Therefore, with regard to this question of CO concentrations we have
22 primarily focused our consideration of the epidemiological studies on the extent to which this
23 evidence is consistent with and generally supportive of conclusions drawn from the combined
24 consideration of the controlled human exposure evidence with estimates from the exposure and
25 dose assessment (section 2.2.2 below). As in the previous review, we believe the integration of
26 the controlled human exposure evidence with the exposure and dose estimates will be most
27 important to informing conclusions regarding ambient CO concentrations of public health
28 concern.
29 • To what extent have important uncertainties identified in the last review been
30 reduced and/or have new uncertainties emerged?
31 Since the time of the last review, some important uncertainties have been reduced, some
32 still remain and others associated with newly available evidence have been identified. A range
24 Few epidemiological studies that had investigated the relationship between CO exposure and ischemic
heart disease were available at the time of the last review (USEPA, 1991, section 10.3.3).
March 2010 2-31 Draft - Do Not Quote or Cite
-------
1 of important uncertainties were identified in a number of areas at the time of the last review (59
2 FR 38913, USEPA, 1992), including:
3 • The adverse nature and significance of the small changes in ST-segment depression
4 identified at the lowest COHb levels investigated in the controlled human exposure
5 studies, and the magnitude of risk associated with such changes for specific health
6 outcomes, such as myocardial infarction, or of slight but cumulative myocardial
7 damage, among other possibilities.
8 • The extent to which COHb measurements made using CO-Ox do not reflect COHb
9 levels in angina patients studied and the potential for as yet unidentified health effects
10 at COHb levels below 2%.
11 • The potential for short-term peak CO exposures to contribute to CNS effects which
12 might affect individual's performance of complex activities such as driving a car or to
13 contribute to other effects of concern.
14 • Effects of ambient CO on potentially susceptible populations other than those with
15 cardiovascular disease, including the developing fetus.
16 • Modeling of COHb formation associated with exposures to ambient CO under different
17 air quality conditions, including those associated with attainment of different NAAQS.
18 As discussed below, some of these uncertainties have been reduced, while some still remain.
19 The CO-induced effects considered of concern at the time of the last review were reduced
20 time to exercise-induced angina and ST-segment depression in patients suffering from coronary
21 artery disease as a result of increases in COHb associated with short CO exposures. These
22 effects had been well documented in multiple studies, and it was recognized that the majority of
23 cardiologists at the time believed that recurrent exercise-induced angina was associated with
24 substantial risk of precipitating myocardial infarction, fatal arrhythmia, or slight but cumulative
25 myocardial damage (USEPA, 1992, p. 22; 59 FR 38911; Basan, 1990; 1991 AQCD). As at the
26 time of the last review, although ST-segment depression is a recognized indicator of myocardial
27 ischemia, the exact physiological significance of the observed changes among individuals with
28 CAD is unclear (ISA, p. 48).
29 In interpreting the study results at the time of the last review, EPA recognized uncertainty
30 in the COHb measurements made using CO-Ox and associated uncertainty in establishing a
31 lowest effects level for CO (USEPA, 1992, p. 31). A multicenter then-recent study (Allred et al.,
32 1989a, 1989b, 1991) was of great importance at that time for several reasons including the large
33 number of subjects used, the rigorous protocol used for subject selection, the use of the most
34 accurate method to measure blood COHb levels and the finding of a dose-response relationship
35 between COHb levels and the ischemic events evaluated in the study. This study reported
36 changes in post-exercise ST-segment depression and reduced time to onset of exercise-induced
37 angina as a result of increases in COHb from a mean baseline of 0.6% to mean levels of 2% and
March 2010 2-32 Draft - Do Not Quote or Cite
-------
1 3.9% (ISA, section 5.2.4; (Allred et al., 1989a,1989b, 1991). In the current review of the
2 evidence related to cardiovascular effects associated with CO exposure, the we place primary
3 emphasis on the findings of Allred et al. (1989a, 1989b, 1991) recognizing the superior quality of
4 the study, both in terms of the rigorous study design as well as the sensitivity of the analytical
5 methods used in determining COHb concentrations (ISA, Section 2.7). No additional clinical
6 studies are available that evaluate responses to lower COHb levels in the cardiovascular-disease
7 population, and uncertainties still remain in determining specific and quantitative relationships
8 between the CO-induced effects in these studies and the increased risk of specific health
9 outcomes. Further, with regard to then unidentified effects at lower COHb levels, no studies
10 have identified other effects on the CAD population or on other populations at lower exposures
11 (ISA, sections 5.2.2).
12 The last review recognized a variety of neurobehavioral effects associated with CO
13 exposure, including changes in visual perception, hearing, motor performance and vigilance
14 among other measures of neurobehavioral performance based on a series of studies conducted
15 from the mid 1960's through the early 1990's (1991 AQCD). Since these effects were observed
16 at exposures to CO resulting in COHb levels ranging from 5-20%, and were poorly understood at
17 the time (1991 AQCD), the review focused on cardiovascular effects, which had been observed
18 at COHb levels below 5% and consequently for which a focus would also provide adequate
19 protection against potential adverse neurobehavioral effects. No new human clinical studies
20 have evaluated CNS or behavioral effects of exposure to CO (ISA, section 5.3.1). However,
21 given the drastic reduction in CO ambient concentrations, the occurrence of these effects in
22 response to ambient CO would be expected to be rare within the current population. Thus, our
23 uncertainty with regard to the potential for such effects to be associated with current ambient CO
24 exposures is reduced.
25 At the time of the last review, the developing fetus was recognized as having the potential
26 to be particularly vulnerable to the effects of CO exposures because human fetal development
27 often occurs at or near critical tissue oxygenation levels (Longo, 1977; 59 FR 38911; USEPA,
28 1992, p. 25). The human data at the time of the last review consisted of a limited set of studies
29 primarily considering the effects of maternal cigarette smoke in the developing fetus, many of
30 them conducted at very high exposure levels (500-1000 ppm). The effects observed included
31 spontaneous abortion and death due to decreased birth weight. Additional animal toxicological
32 studies, also conducted at CO concentrations higher than the ambient CO levels at the time,
33 provided suggestive evidence of fetal mortality, teratogenicity, reduced body weight,
34 morphological changes, altered cardiovascular development and neurochemical changes. Since
35 the last review, the epidemiologic and toxicological evidence of effects on birth and
36 developmental outcomes has somewhat expanded, although the available evidence is still
March 2010 2-33 Draft - Do Not Quote or Cite
-------
1 considered limited with regard to effects on preterm birth, birth defects, decreases in birth
2 weight, measures of fetal growth, and infant mortality. Although there are animal toxicological
3 studies providing support and coherence for those effects, the understanding of the mechanisms
4 underlying reproductive and developmental effects is still lacking (ISA, section 5.4.1). Thus,
5 although the evidence continues to "suggest[s] that critical developmental phases may be
6 characterized by enhanced sensitivity to CO exposure" (ISA, p. 2-11), evidence is still lacking on
7 the potential for adverse effects on birth and developmental outcomes at CO exposures near
8 those associated with current levels of ambient CO. As a result uncertainty still remains in this
9 area.
10 Numerous improvements have been made over the last decade that have reduced the
11 uncertainties associated with the models used to estimate COHb levels resulting from ambient
12 CO exposures under different air quality conditions, including those associated with just meeting
13 the current CO NAAQS (draft REA, section 4.3). This progression in exposure model
14 development has led to the Agency's currently used model (APEX4.3) that has an enhanced
15 capacity to estimate population CO exposures and more accurately predicts COHb levels in
16 persons exposed to CO. Our application of APEX4.3 to the current situation, using updated data
17 and new algorithms to estimate exposures and doses experienced by individuals, better
18 represents the variability in population exposure and COHb dose levels. However, while APEX
19 4.3 is greatly improved when compared with previously used exposure models, its application is
20 still limited with regard to data to inform our understanding of spatial relationships in ambient
21 CO concentrations and within microenvironments of particular interest. Further information
22 regarding model improvements and remaining exposure modeling uncertainties are described in
23 section 2.2.2 below.
24 In the current review, we continue to recognize coronary artery disease as the most
25 important susceptibility characteristic for increased risk due to CO exposure (ISA, p. 2-10).
26 While a much expanded epidemiologic database inclusive of studies showing associations
27 between ambient CO monitor concentrations and increases in emergency room visits and
28 hospitalizations for ischemic heart disease events support this conclusion, a variety of associated
29 uncertainties complicate the interpretation of these studies. Although CO associations generally
30 remain robust in copollutant models (ISA, Figures 5-6 and 5-7, section 2.7), disentangling a CO-
31 specific effect at low ambient concentrations from other pollutants remains of concern,
32 particularly for the quantitative interpretation of these studies. The potential for CO to act as a
33 surrogate for another combustion related pollutant included in a copollutant statistical model
34 "complicates quantitative interpretation of the effect estimates reported in epidemiologic studies"
March 2010 2-34 Draft - Do Not Quote or Cite
-------
1 (Tolbert 2007; ISA, p. 2-16).25 Moreover, as recognized by CAS AC, "the problem of co-
2 pollutants serving as potential confounders is particularly problematic for CO" and "a better
3 understanding of the possible role of co-pollutants is relevant to regulation ... on the health
4 effects of CO" (Brain and Samet, 2010). Error in the assessment of personal CO exposure
5 through the use of fixed site-monitors measuring CO levels at or below the limit of detection of
6 the monitor is a new source of uncertainty since the prior review due to an overall decrease in
7 ambient CO concentrations. This error may lead to an over or underestimation of CO health risk
8 estimates. Thus, some uncertainties associated with the epidemiological evidence "complicate
9 the quantitative interpretation of the epidemiologic findings, particularly regarding the biological
10 plausibility of health effects occurring at COHb levels resulting from exposures to the ambient
11 CO concentrations" assessed in these studies (ISA, p. 2-17).
12 In summary, some important uncertainties from the last review have been reduced,
13 including those associated with concerns for ambient levels of CO to pose neurobehavioral risks.
14 A variety of uncertainties still remain including the adverse nature and significance of the small
15 changes in ST-segment depression identified at the lowest COHb levels investigated, and the
16 magnitude of associated risk of specific health outcomes, as well as the potential for as yet
17 unidentified health effects at COHb levels below 2%. Additionally, although the evidence base
18 is somewhat expanded with regard to the potential for CO effects on the developing fetus,
19 uncertainties remain with regard to this potential susceptibility. Our exposure and dose models
20 have improved giving us increased confidence in their estimates. We additionally recognize that
21 the expanded body of epidemiological evidence includes its own set of uncertainties which
22 complicates its interpretation.
23 2.2.2 Exposure/Risk-based Considerations
24 Our consideration of the scientific evidence in the current review, as at the time of the
25 last review (summarized in section 2.1.1 above), is informed by results from a quantitative
26 analysis of estimated population exposure and resultant COHb levels. As in our consideration of
27 the evidence in section 2.2.1 above, we have organized the discussion that follows here around a
28 set of key questions to assist us in drawing from the draft assessment of CO exposure and
29 resultant COHb levels for at-risk populations living in two cities under current air quality
25 In interpreting the epidemiological evidence for cardiovascular morbidity the ISA notes that it "is
difficult to determine from this group of studies the extent to which CO is independently associated with CVD
outcomes or if CO is a marker for the effects of another traffic-related pollutant or mix of pollutants. On-road
vehicle exhaust emissions are a nearly ubiquitous source of combustion pollutant mixtures that include CO and can
be an important contributor to CO in near-road locations. Although this complicates the efforts to disentangle
specific CO-related health effects, the evidence indicates that CO associations generally remain robust in copollutant
models and supports a direct effect of short-term ambient CO exposure on CVD morbidity." (ISA, pp. 5-40 to 5-41).
March 2010 2-3 5 Draft - Do Not Quote or Cite
-------
1 conditions and conditions simulated to just meet the current CO standards. These questions are
2 intended to inform consideration of the following overarching question.
3 Prior to addressing the series of questions below, we provide a description of the
4 modeling tools used in and the design of the assessment, the magnitude of COHb levels. We
5 then consider the question regarding the size of the simulated at-risk population estimated to
6 experience COHb levels of interest, followed by the identification of key uncertainties associated
7 with our assessment of exposure and dose with regard to drawing conclusions as to the adequacy
8 of the protection afforded by the current CO standards. We then consider the exposure and dose
9 estimates from the quantitative assessment with regard to implications as to the extent of the
10 population provided protection from COHb levels of concern under the current standards to help
11 inform judgments about the extent to which such estimates may be judged to be important from a
12 public health perspective.
13 In the draft assessment conducted for this review, described in detail in the draft REA and
14 summarized here, we have estimated CO exposure and associated COHb levels in simulated
15 populations in two urban study areas in Denver and Los Angeles, in which current CO
16 concentrations are below the current standards. We selected these areas because: (1) both cities
17 have been included in prior CO NAAQS exposure assessments and thus serve as an important
18 connection with past assessments; (2) historically, they have generally had the highest CO
19 ambient concentrations among urban areas in the U.S.; and (3) Denver is at high altitude and
20 represents an important risk scenario due to the potential increased susceptibility to CO exposure
21 associated with high altitudes. In addition, of 10 urban areas across the U.S. selected for detailed
22 analysis in the ISA and having monitors meeting a 75% completeness criteria, the two locations
23 were ranked first (Los Angeles) and second (Denver) regarding the percentage of elderly
24 population within 5, 10, and 15 km of monitor locations, and ranked first (Los Angeles) and fifth
25 (Denver) regarding number of 1- and 8-hour daily maximum CO concentration measurements
26 (ISA, section 3.5.1.1).
27 Estimates were developed for exposures associated with current "as is" conditions (2006
28 air quality) and also for higher ambient CO concentrations associated with air quality conditions
29 simulated to just meet the current 8-hour standard.26 In considering the adequacy of the current
30 standard, it is important to note that over the last few years, the standard has been met throughout
31 the country with few exceptions.27 Although we consider it unlikely that air concentrations in
32 many urban areas across the U.S. that are currently well below the current standard would
26 As discussed elsewhere, the 8-hour standard is the controlling standard for CO.
27 As described in section 1.3.3 above, in the most recent period analyzed (2007-2008), all areas of the U.S.
met both CO NAAQS. In both of the previous periods (2005-2006 and 2006-2007), one area did not meet the 8-
hour standard. Further, one area of the country is designated in non-attainment with the CO NAAQS although air
quality in that area has met the standards in the past three periods.
March 2010 2-36 Draft - Do Not Quote or Cite
-------
1 increase to just meet the standard, we also recognize the potential for CO concentrations in some
2 areas currently below the standard to increase to just meet the standard. Accordingly, we have
3 simulated conditions of increased CO concentrations that just meet the current standard in the
4 two study areas. In this scenario, we note there is uncertainty associated with simulating this
5 hypothetical profile of higher CO concentrations that just meet the current standard.
6 The exposure and dose modeling, presented in detail in the draft REA, relied on EPA's
7 Air Pollutant Exposure model (APEX4.3), which estimates human exposure using a stochastic,
8 event-based microenvironmental approach. This model has a history of application, evaluation,
9 and progressive model development in estimating human exposure and dose for several NAAQS
10 reviews, including CO, Os, NO2, and SO2. As mentioned above, we have made major changes to
11 the exposure modeling approach used in the last CO review. The prior review relied on
12 estimated population exposure and dose generated from pNEM, a model that 1) employed a
13 cohort-based approach,28 2) relied on a limited set of activity pattern data (approximately 3,600
14 person-days) 3) used four broadly defined categories to estimate breathing rates (i.e., slow-
15 sleeping, slow-awake, medium, and fast), and implemented a geodesic distance range
16 methodology to approximate workplace commutes (Johnson et al., 1992; USEPA, 1992). Each
17 of these approaches used by pNEM, while appropriate given the data available at the time would
18 tend to limit the ability to accurately model the lower and upper tails of the population exposure
19 distribution. In contrast, APEX4.3 includes new algorithms to 1) simulate activity sequences and
20 exposure profiles for individuals, 2) estimate activity-specific minute-by-minute oxygen
21 consumption and breathing rates, 3) address spatial variability in home and work-tract ambient
22 concentrations for commuters, and 4) estimate event-based microenvironmental concentrations.
23 APEX also uses additional data available from recent activity pattern surveys (CHAD now has
24 about 34,000 person-days of data) and uses the most recent US census data to represent
25 population demographics and home-to-workplace commutes. Modeling the prevalence of
26 exposures or dose in a population of interest (such as individuals with coronary heart disease)
27 has also been enhanced in APEX with the addition of output tables presenting statistics for
28 identified subpopulations and age groups (or life stages), using the updated census population
29 demographic data, and including options for approaches used to estimate microenvironmental
30 concentration.
When using the cohort approach, each cohort is assumed to contain persons with identical exposures
during the specified exposure period. Thus, variability in exposure will be attributed to differences in how the
cohorts are defined, not necessarily reflecting differences in how individuals might be exposed in a population. A
total of 420 cohorts were used to estimate population exposure based on selected demographic information (11
groups using age, gender, work status), residential location, work location, and presence of indoor gas stoves
(Johnson, etal., 1992; USEPA, 1992).
March 2010 2-37 Draft - Do Not Quote or Cite
-------
1 APEX probabilistically generates a sample of hypothetical individuals from an actual
2 population database and simulates each individual's movements through time and space (e.g.,
3 indoors at home, inside vehicles) to estimate his or her exposure to a pollutant. Population
4 characteristics are taken into account to represent the demographics of each study area. With
5 regard to age and gender demographics for the simulated at-risk population (adults with
6 diagnosed and undiagnosed CAD), we augmented estimates of the prevalence of CAD provided
7 by the National Health Interview Survey with estimates of undiagnosed ischemia that were based
8 on two assumptions: (1) there are 3.5 million persons in U.S. with undiagnosed IHD (drawn
9 from an estimate by the American Heart Association) and (2) persons with undiagnosed IHD are
10 distributed within the population in the same manner as persons with diagnosed IHD (draft REA,
11 section 5.5.2). Based on exposure concentrations, minute-by-minute activity levels, and
12 physiological characteristics of the simulated person (see draft REA, chapters 4 and 5), APEX
13 estimates the level of COHb in the blood at the end of each hour based on solution to the
14 nonlinear Coburn-Forster-Kane equation (draft REA, section 4.4.7).
15 APEX simulations performed for this review have focused on exposures to ambient CO
16 occurring in eight microenvironments, with no additional contribution from indoor CO sources.
17 We note that, where present, indoor sources, including gas stoves and tobacco smoke can also be
18 important contributors to total CO exposure and may be of concern for at-risk populations.
19 Some assessments performed previously have included modeling simulations both with and
20 without certain indoor sources and these assessments provide context for the assessment of
21 ambient CO exposure and dose. For example, a 2000 exposure/dose assessment indicated that
22 the impact of such sources can be substantial on the portion of the population experiencing
23 higher exposures and COHb levels (Johnson et al., 2000). In the last review it was noted that
24 while these indoor sources were shown to contribute to total CO exposure they would not be
25 effectively mitigated by setting more stringent ambient air quality standards (59 FR 38914). In
26 the current review, model simulations have not included indoor sources.
27 As discussed in the previous section (Section 2.2.1), people with cardiovascular disease
28 are the population of primary focus in this review. Further, based on the available evidence in
29 the current review, coronary artery disease, also known as coronary heart disease is the "most
30 important susceptibility characteristic for increased risk due to CO exposure" (ISA, p. 2-11).
31 Controlled human exposure studies have provided specific quantitative COHb dose-response
32 information for this specific population with regard to effects on markers of myocardial
33 ischemia. In identifying COHb levels associated with specific effects for CAD patients, we have
34 given primary focus to the multi-laboratory study in which COHb was analyzed by the more
35 accurate GC method (Allred et al., 1989a, 1989b, 1991). Specifically, this study has documented
36 reductions in times to exercise-induced change in the ST-segment of an electrocardiogram and to
March 2010 2-3 8 Draft - Do Not Quote or Cite
-------
1 exercise-induced onset of angina, observed in response to CO exposures involving subjects with
2 pre-existing CAD, as described in section 2.2.1 above.
3 Studies have not been designed to evaluate similar effects at CO exposure eliciting COHb
4 levels below this range. In addition, these studies do not address the fraction of the population
5 experiencing a specified health effect at various dose levels. These aspects of the evidence
6 contributed to EPA's conclusion that there are currently insufficient controlled human exposure
7 data to support the development of quantitative dose-response relationships which would be
8 required in order to conduct a quantitative risk assessment for this health endpoint, rather than a
9 benchmark level approach. Further, based on the current evidence with regard to quantitative
10 information of COHb levels and association with specific health effects, the primary target
11 population for purposes of the quantitative assessment is adults with CHD (also known as
12 ischemic heart disease IHD or CAD), both diagnosed and undiagnosed.29 Evidence
13 characterizing the nature of specific health effects of CO in other populations is limited and does
14 not include specific COHb levels related to health effects in those groups. As a result, the
15 quantitative assessment does not develop separate quantitative dose estimates for populations
16 other than those with CAD.
17 We have reviewed COHb estimates developed in the draft REA for the simulated at-risk
18 population with attention to both COHb in absolute terms and based on consideration of the
19 contribution associated with ambient CO exposures. For purposes of the review of COHb in
20 absolute terms, we identified benchmark levels of 1.5%, 2.0%, 2.5% and 3% COHb. Selection
21 of this range of benchmark levels is based on consideration of the evidence from controlled
22 human clinical studies of CAD patients (discussed in section 2.2.1 above). This range includes
23 the range of COHb levels identified as levels of concern in the review completed in 1994 (2.0 to
24 2.9%) and the level given particular focus (2.1%), as described in section 2.1.1 above (USEPA,
25 1992; 59 FR 48914). The range of benchmark levels identified here extends below the lowest
26 mean COHb level (e.g., 2.0% post-exercise in Allred et al., 1989b) resulting from controlled
27 increased exposure to CO in the clinical evidence. The extension of this range reflects a number
28 of considerations, including: (1) comments from the CASAC CO panel on the draft Analysis
29 Plan (Brain, 2009); (2) consideration of the uncertainties regarding the actual COHb levels
30 experienced in the controlled human exposure studies; (3) that these studies did not include
31 individuals with most severe cardiovascular disease30; (4) the lack of studies that have evaluated
29 As described in sectionl .2 above, this is the same population group that was the focus of the CO NAAQS
exposure/dose assessments conducted previously (e.g., US EPA, 1992; Johnson et al., 2000).
30 Although the CAD patients evaluated in the controlled human exposure study by Allred et al. (1989a,
1989b, 1991) are not necessarily representative of the most sensitive population, the level of disease in these
individuals ranged from moderate to severe, with the majority either having a history of myocardial infarction or
having > 70% occlusion of one or more of the coronary arteries (ISA, p. 5-43).
March 2010 2-39 Draft - Do Not Quote or Cite
-------
1 effects of experimentally controlled short-term CO exposures resulting in mean COHb levels
2 below 2.0-2.4%; and (5) the lack of evidence of a threshold at the increased COHb levels
3 evaluated. We note that CASAC comments on the first draft REA recommended the addition of
4 a benchmark at 1% COHb and we have presented results for this COHb level in the second draft
5 REA. Given that this level overlaps with the upper part of the range of endogenous levels in
6 healthy individuals as characterized in the ISA (ISA, p. 2-6) and is within the upper part of the
7 range of baseline COHb levels in the study by Allred et al (1989b, Appendix B), however, we
8 considered that it may not be appropriate to place weight on it as a potential health effects
9 benchmark level and accordingly has not focused on dose estimates for it in the discussion
10 below. Additionally, in considering contributions to COHb associated with ambient CO
11 exposures, we recognize the findings of Allred et al (1989a, 1989b, 1991) with regard to a dose-
12 response relationship for decreases in time to exercise-induced angina and ST-segment change
13 per 1% increase in COHb concentration in persons with pre-existing CAD (ISA, section 5.2.4;
14 Allred et al., 1989a, 1989b, 1991). In considering COHb estimates from the quantitative
15 assessment below, we take note of estimates of endogenous COHb levels from the quantitative
16 exposure and dose analysis which indicated that in the absence of external CO sources,
17 approximately 1.5% of the simulated at-risk population in one of the two populations was
18 estimated to experience maximum end-of-hour endogenous COHb levels at or above 1% (draft
19 REA, section 6).
20 The benchmark levels identified are used to interpret COHb levels estimated to occur in
21 the modeled population in response to exposures to ambient CO in different air quality scenarios
22 in light of the evidence for cardiovascular effects in individuals with CHD when exposed to CO.
23 Specifically, in considering the exposure and dose estimates with regard to consideration of the
24 adequacy of the current standards here, we focus on estimates of the portion of this at-risk
25 population estimated to experience maximum end-of-hour COHb levels above these
26 benchmarks, and on information concerning the COHb increment above endogenous levels
27 associated with ambient CO exposures.
28 • What is the magnitude of at-risk population COHb levels estimated to occur in
29 areas simulated to just meet the current CO standards? What portion of the at-
30 risk population is estimated to experience maximum COHb levels above levels of
31 potential health concern?
32 In addressing this question, we consider the population COHb estimates provided by the
33 draft REA (draft REA, section 6.2) and summarized in Tables 2-5 and 2-6 below. These tables
34 include estimates associated with current "as is" conditions and conditions simulated to just meet
35 the current standard. In considering these estimates, we particularly focus on the extent to which
36 the current standards provide protection to the simulated at-risk population from COHb levels of
March 2010 2-40 Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
potential concern, by comparing the estimated levels in the population to the potential health
effect benchmarks described above. As described above, the draft REA estimates of COHb
reflect the impact of exposures to ambient CO in addition to individual endogenous levels (draft
REA, section 6.4).
Under "as is" conditions in the two study areas, fewer than 0.1 % and 2% of the
simulated at-risk population are estimated to experience an end-of-hour COHb concentration at
or above 2% and 1.5%, respectively, with the results for the Los Angeles study area including
slightly larger portions of the population above these benchmarks than that estimated for the
Denver study area. Under conditions with higher ambient CO concentrations that just meet the
current 8-hour standard, the portion of the simulated at-risk population estimated to experience
these levels of COHb is greater. But for these simulations, the relationship between the two
study areas is reversed with the Denver study area having higher COHb levels. In the Denver
study area 3.4% of the simulated at-risk population is estimated to experience a maximum end-
of-hour COHb level at or above 2%, while 19% of the population is estimated to experience a
maximum COHb level at or above 1.5%. For the Los Angeles study area, less than 1% and 5%
of the population was estimated to experience maximum end-of-hour COHb levels at or above
2% and 1.5%, respectively (Table 2-5).
Table 2-5. Portion of simulated at-risk population with maximum end-of-hour COHb
levels at or above the indicated level under air quality conditions simulated to
just meet the current standards and under "as is" (2006) conditions. COHb
levels reflect endogenous levels and ambient CO exposures.
Maximum
end-of-
hour
COHb
Level
>3%
>2.5%
>2%
>1.5%
Percentage (%) of Simulated At-risk Population
Just meeting current standards
Los Angeles
<0.1
<0.1
0.5
4.6
Denver
0.2
0.6
3.4
19.1
"as is" conditions
Los Angeles
0.0
<0.1
1.5
Denver
0.0
0.6
* <0.1 is used to represent nonzero estimates below 0.1%
22
March 2010
2-41
Draft - Do Not Quote or Cite
-------
1 • What are the key uncertainties associated with our exposure and dose estimates,
2 including those of particular significance with regard to drawing conclusions as to
3 the adequacy of the protection afforded by the current CO standards?
4 In considering the uncertainties associated with the quantitative estimates of exposure
5 and dose from the draft REA, we relied on an approach intended to identify and compare the
6 relative impact that important sources of uncertainty may have on the estimated potential health
7 effect endpoints (i.e., estimates of the maximum end-of-hour COHb levels in the simulated at-
8 risk population). The approach used (as described in section 7.2 of the second draft REA) was
9 developed using guidelines outlining how to conduct a qualitative uncertainty characterization
10 (WHO, 2008) and applied in the most recent NO2 (US EPA, 2008b) and SO2 NAAQS reviews
11 (US EPA, 2009). We employed the mainly qualitative approach given the extremely limited data
12 available to inform probabilistic analyses. The qualitative approach used varies from that of
13 WHO (2008) in that a greater focus of the characterization performed was placed on evaluating
14 the direction and the magnitude of the uncertainty; that is, qualitatively rating how the source of
15 uncertainty, in the presence of alternative information, may affect the estimated exposures and
16 health risk results. In addition and consistent with the WHO (2008) guidance, we discuss the
17 uncertainty in the knowledge-base (e.g., the accuracy of the data used, acknowledgement of data
18 gaps) and decisions made where possible (e.g., selection of particular model forms), though
19 qualitative ratings were assigned only to uncertainty regarding the knowledge-base.
20 In the characterization of uncertainty for the current analysis, we identified and evaluated
21 fourteen separate sources of uncertainty associated with four main components of the
22 assessment:
23 • Ambient monitor CO concentrations
24 - database quality, spatial and temporal representation, and missing data
25 substitution;
26 • Adjustment of air quality to simulate just meeting the current and potential alternative
27 standards
28 - historical data used, proportional approach used;
29 • APEX inputs and algorithms
30 - population database, activity pattern database, longitudinal profile algorithm,
31 meteorological data, micrenvironmental algorithm and input data, commuting
32 algorithm, CHD prevalence, physiological factors; and,
33 • Potential health effects benchmark levels for the simulated at-risk population.
34
March 2010 2-42 Draft - Do Not Quote or Cite
-------
1 By comparing judgments made regarding the magnitude and direction of influence the
2 identified sources have on estimated exposure concentrations and dose levels and the existing
3 uncertainties in the knowledge-base, we identified six sources (i.e., the spatial and temporal
4 representation, historical data used, activity pattern database, longitudinal profile algorithm,
5 microenvironmental algorithm and input data, and physiological factors) that remain as the most
6 important uncertainties in this assessment. Taking into consideration improvements in the model
7 algorithms and data since the last review, and having identified and characterized these
8 uncertainties here, we conclude that the estimates associated with the current analysis, at a
9 minimum, better reflect the full distribution of exposures and dose as compared to results from
10 the 1992 analysis. When considering the overall quality of the current exposure modeling
11 approach, the algorithms, and input data used, alongside the identified limitations and
12 uncertainties, we conclude that the quantitative assessment provides reasonable estimates of CO
13 exposure and COHb dose for the simulated population the assessment is intended to represent
14 (i.e., the population residing within the urban core of each study area). We note somewhat
15 greater uncertainty in characterizing of the ends of the distribution of population exposures.
16 • To what extent are the estimates of at-risk population COHb levels under
17 conditions just meeting the current CO standards important from a public health
18 perspective?
19 In considering the public health implications of the quantitative dose estimates, we first
20 note the size of the at-risk population simulated in the draft REA, people with CAD, both
21 diagnosed and undiagnosed. This population is recognized as that most susceptible to increased
22 risk of CO-induced health effects (ISA, sections 5.7.1.1 and 5.7.8). The 2007 estimate from the
23 National Health Interview Survey (NHIS) performed by the U.S. Centers for Disease Control of
24 the size of the U.S. population with coronary heart disease, angina pectoris (cardiac chest pain)
25 or who have experienced a heart attack (ISA, Table 5-26) is 13.7 million people (ISA, pp.5-117).
26 Further, there are estimated to be three to four million additional people with silent ischemia or
27 undiagnosed IHD (AHA, 2003). In combination this represents a large population represented
28 by the draft REA analyses and for which the COHb benchmarks described above (based on
29 studies of CAD patients) are relevant, that is more susceptible to ambient CO exposure when
30 compared to the general population (ISA, section 5.7).
31 As discussed above, the draft REA estimates that 0.5 percent of the Los Angeles and
32 3.4% of the Denver study CHD populations may experience maximum end-of-hour COHb levels
33 of the same magnitude as the average COHb level resulting for the lower experimental exposure
34 group in the study by Allred et al. (1989a, 1989b, 1991) which resulted in a statistically
35 significant reduction in the time to exercise-induced markers of myocardial ischemia (angina and
36 ECG ST-segment change) of 4-5% on average. While recognizing that the exposure resulting in
March 2010 2-43 Draft - Do Not Quote or Cite
-------
1 the 2% average COHb level associated with this result was significantly over background (e.g.,
2 approximately 1 hour at 50-200 ppm CO as compared to 0-2 ppm clinic air concentrations), we
3 also note the dose-response relationship was summarized by the authors as indicating decreases
4 of roughly 1.9% in time to exercise-induced angina and 3.9% in time to exercise-induced ST-
5 segment change per 1% increase in COHb concentration in persons with pre-existing CAD, as
6 discussed in section 2.2.1 above (ISA, section 5.2.4; Allred et al., 1989a,1989b, 1991).31 As
7 noted in the ISA, the physiological significance of the observed ST-segment changes among
8 individuals with CAD is unclear and variability in severity of disease among individuals with
9 CAD is likely to influence the critical level of COHb which leads to adverse cardiovascular
10 effects (ISA, p. 2-6).
11 In addition to the population with diagnosed and undiagnosed CAD, we note implications
12 with regard to risk to people with other forms of CVD, which we recognize as potentially
13 susceptible (ISA, section 2.6.1). For example, NHIS estimates for 2007 indicate there are
14 approximately 12 million additional people with other types of heart disease (ISA, Table 5-26).
15 Within this broad group, implications are more significant for those included in this population
16 for which their disease state affects their ability to compensate for the hypoxia-related effects of
17 CO (ISA, section 4.4.4).
18 Other potentially susceptible populations to the effects of CO include people with chronic
19 obstructive pulmonary disease, diabetes and anemia, as well as older adults and fetuses during
20 critical phases of development (as discussed in section 2.2.1 above). In considering potential
21 impacts on such populations, however, we recognize limitations in the evidence in informing us
22 with regard to the specific effects of CO on these populations and, importantly, the exposures
23 and associated COHb levels associated with such effects.
24 In summary, we recognize that conclusions regarding the extent to which the estimates of
25 exposure and COHb associated with just meeting the current standard are important from a
26 public health perspective depend in part on public health policy judgments about the significance
27 of effects at the COHb benchmark levels considered and judgments about the level of public
28 health protection with an adequate margin of safety. For example, we note that these estimates
29 indicate that the current 8-hour standard limits substantially the percent of the population
30 expected to experience maximum COHb levels at or above the higher benchmarks, while
31 limiting much less so the percentage of the population with maximum COHb levels above the
31 The dose-response relationship was derived as the average of the regressions of the individual study
subject data for changes in time to onset of the monitored measures of ischemia across their range of COHb levels
(from their baseline COHb to their two higher COHb levels resulting from the two experimental CO exposures).
The clean-air exposure, post-exercise (baseline) COHb levels in the individual study subjects ranged from 0.2% to
1.1%, their post-exercise COHb levels for the lower experimental CO exposure ranged from 1.0 to 3.0% and their
post-exercise COHb levels for the higher experimental CO exposure ranged from 2.3 to 5.1% (Allred et al., 1989b).
March 2010 2-44 Draft - Do Not Quote or Cite
-------
1 lowest benchmarks. Thus, we initially conclude that these estimates provide support for a range
2 of views regarding the importance of these estimates form a public health perspective.
3 2.2.3 Initial Staff Conclusions on Adequacy of the Current Standards
4 In considering the adequacy of the current standard, staff gives great weight to the long-
5 standing body of evidence for CO, augmented in some aspects since the last review, that has
6 established: the common mechanism of CO health effects as involving binding to reduced iron
7 in heme proteins and the alteration of their functioning; the important role of hypoxia (reduced
8 oxygen availability) induced by increased COHb blood levels in eliciting health effects; the use
9 of COHb as the bioindicator and dose metric for evaluating CO exposure and potential for health
10 effects; and, people with cardiovascular disease as a key population at risk from low CO
11 exposures (ISA; 2000 AQCD; 1991 AQCD). We additionally recognize the expansion of the
12 epidemiological evidence that provides coherence with our conclusions regarding cardiovascular
13 disease-related susceptibility and may (recognizing various uncertainties associated with the
14 evidence in the particular case of CO) provide indications of air quality conditions associated
15 with ambient CO-related risk, most particularly in areas during periods which included
16 exceedance of the current 8-hour standard (ISA, section 5.2.1).
17 As at the time of the last review, we give weight to COHb estimates developed from
18 modeling exposures to ambient CO under conditions simulated to just meet the current,
19 controlling, 8-hour standard. These COHb estimates and the incremental contribution from
20 ambient CO exposures are considered in light of the clinical studies. These estimates indicate
21 that some 1-3 percent of the simulated at-risk population are expected to experience maximum
22 COHb levels of a magnitude for which association has been demonstrated with markers of
23 myocardial ischemia and larger portions of the population may experience maximum COHb
24 levels at or above the lowest benchmark considered. Further, in considering public health
25 significance of such estimates, while recognizing uncertainty in the clinical significance
26 associated with the smaller effects observed at the lower COHb levels, we note the substantial
27 size of the at-risk population in the U.S. and the existence of other potentially susceptible
28 populations for which evidence is more limited.
29 In summary, we draw initial conclusions with regard to the adequacy of the current
30 standard from both the evidence and from the exposure and dose assessment, in light of related
31 limitations and uncertainties. We initially conclude that the combined consideration of the body
32 of evidence and the quantitative exposure and dose estimates provide support for a standard at
33 least as protective as the current standard. Further, we recognize that conclusions regarding the
34 adequacy of the current standard depend in part on public health policy judgments about the
35 significance of effects at the COHb benchmark levels considered and judgments about the level
March 2010 2-45 Draft - Do Not Quote or Cite
-------
1 of public health protection with an adequate margin of safety. Thus, we initially conclude that
2 the combined consideration of the evidence and quantitative estimates described above may be
3 viewed as providing support for either retaining or revising the current 8-hour standard.
4 The extent to which the current standard is judged to be adequate or not depends on a
5 variety of factors, inclusive of technical considerations, science policy judgments and public
6 health policy judgments. These include public health policy judgments on the appropriate COHb
7 benchmark levels on which to place weight; judgments on the clinical significance of the effects
8 that have been observed at those levels; uncertainty associated with interpretation of the
9 epidemiological evidence as providing information on ambient CO as distinct from information
10 on the mixture of pollutants associated with traffic, and the weight to place on the
11 epidemiological studies for purposes of considering adequacy of the current standards, given that
12 few are available (in one city) during periods under air quality conditions that did not exceed the
13 current standard; and the weight to place on the results of the draft exposure assessment relative
14 to those from past assessments, recognizing the implementation of a new modeling approach and
15 input data.
16 Staffs initial conclusions with regard to elements of revised primary standards for CO
17 that may be appropriate to consider are discussed in the following sections. These conclusions
18 are drawn from consideration focused on the 8-hour standard in recognition of its role in
19 controlling ambient air quality conditions for CO. We give more specific consideration to each
20 of the two CO primary standards in the discussion of alternative standards in section 2.3 below.
21 2.3 CONSIDERATION OF ALTERNATIVE STANDARDS
22 To the extent that the information available in this review suggests that revision of the
23 current standards is appropriate to consider, staff has considered whether the available body of
24 evidence supports consideration of options that are different from the current standard, as
25 articulated by the following overarching question:
26 • To what extent does the currently available scientific evidence- and exposure/risk-
27 based information, as reflected in the ISA and draft REA, support consideration
28 of alternatives to the current CO standards to provide increased protection from
29 ambient CO exposures?
30 To assist us in interpreting the currently available scientific evidence and the results of
31 recent quantitative exposure/risk analyses to address this question, we have focused on a series
32 of more specific questions in sections 2.3.1 and 2.3.2 below. In considering the scientific and
33 technical information, we consider both the information available at the time of the last review
34 and information newly available since the last review which has been critically analyzed and
35 characterized in the 2000 AQCD and more recently in the ISA. Specifically, we consider how
March 2010 2-46 Draft - Do Not Quote or Cite
-------
1 the currently available scientific evidence informs decisions regarding the basic elements of the
2 NAAQS: indicator, averaging time, level and form. Considerations with regard to indicator and
3 averaging time are presented in sections 2.3.1 and 2.3.2. Form and level are discussed in section
4 2.3.3.
5 2.3.1 Indicator
6 The indicator for carbon monoxide is carbon monoxide as measured by the federal
7 reference methods and equivalent methods. Unlike several other criteria pollutants, there are not
8 multiple compounds or size fractions of carbon monoxide. Federal reference methods are
9 available that can effectively measure carbon monoxide, and thus we have not identified any
10 basis for considering an indicator other than carbon monoxide for the standard.
11 2.3.2 Averaging Time
12 In considering potential alternative averaging times other than the current 1- and 8-hour
13 averaging times, we consider the following question:
14 • Do health effects evidence and air quality/exposure assessments provide support
15 for considering different exposure indices or averaging times?
16 The averaging times for the current standards are 1-hour (35 ppm) and 8-hours (9 ppm).
17 These averaging times were first chosen when EPA promulgated the primary NAAQS for CO in
18 1971 (36 FR 8186), and were retained in subsequent reviews (1980, 45 FR 55066; 1985, 50 FR
19 37485; and 1994, 59 FR 38906).
20 The 8-hour averaging time was selected primarily based on the rationale that most
21 individuals achieve steady-state blood levels of COHb after approximately 8 hours of continuous
22 CO exposure (USEPA, 1979; AQCD 2000, section 7.4; ISA, section 4.3.2.2). Another
23 consideration was that the 8-hour time-frame represented a good basis for tracking continuous
24 exposures during any 24-hour period, recognizing that people may be exposed in approximately
25 8-hour blocks of time (e.g., working or sleeping).
26 The 1-hour averaging time primarily reflected consideration of the potential impact of
27 exposures of 1 hour or shorter, such as very short duration peak exposures in urban environments
28 (59 FR 38914). Much of the health evidence, however, is linked to exposures of 1 hour or
29 longer, such as the controlled human studies demonstrating aggravation of myocardial ischemia
30 after exposure of individuals with coronary artery disease to CO for approximately 1-2 hours.
31 Further, based on consideration of air quality data patterns it was recognized that attainment of
32 the 8-hour standard would tend to also limit short-term peaks to levels below those of concern
33 (45 FR 55077).
March 2010 2-47 Draft - Do Not Quote or Cite
-------
1 In considering whether the information available in this review supports consideration for
2 different averaging times for the CO standards, we note that the available information is
3 generally consistent with and supportive of the conclusions reached in the previous reviews to
4 retain the 1- and 8-hour averaging times.
5 • Controlled human exposure studies, with which epidemiological evidence is coherent,
6 provide the clearest evidence of short-term CO-induced effects and reflect exposures of
7 1-2 hours (ISA, sections 5.2, 5.2.6.1).
8 • Epidemiological studies provide evidence based on analyses of 1-, 8-, and 24-hour
9 averaging times (ISA, section 5.2.6).
10 • Most individuals would be expected to achieve steady-state levels of blood COHb after
11 6-8 hours of continuous exposure to a fixed CO concentration. Therefore, 8 hours may
12 reasonably be considered closely representative of longer continuous exposures (2000
13 AQCD, section 7.4).
14 • The 1-hour averaging time is considered to provide protection from the occurrence of
15 shorter term peak exposure concentrations of ambient CO that may be encountered in
16 some environments, and from COHb levels that may result from such exposures (2000
17 AQCD, section 7.4; ISA, section 2.4.1).
18 Taken together, we initially conclude that the information available in the current review
19 continues to support the 1- and 8-hour averaging times and does not provide information that
20 would lead to consideration of standards with alternative averaging times.
21 2.3.3 Form and Level
22 In considering potential alternative forms and levels for the standards, we focused
23 primarily on form and level for the 8-hour standard, while also considering form and level for the
24 1-hour standard. In so doing, we addressed the following overarching question:
25 • What is the range of alternative levels and forms for the standard that are
26 supported by the health effects evidence and air quality/exposure assessments,
27 and what are the uncertainties and limitations in that health effects evidence and
28 air quality/exposure assessments?
29 In this section, we focus on general considerations with regard to form in section 2.3.2.1.
30 We then address specific considerations for both level and form in section 2.3.2.2, first with
31 regard to the 8-hour standard and then with regard to the 1-hour standard.
32 2.3.3.1 Potential Alternative Forms
33 When evaluating alternative forms in conjunction with specific levels, staff considers the
34 adequacy of the public health protection provided by the combination of level and form to be the
35 foremost consideration. In addition, we recognize that it is preferable to have a form that is
36 reasonably stable and relatively insulated from the impacts of extreme meteorological events or
March 2010 2-48 Draft - Do Not Quote or Cite
-------
1 other rare, transitory impacts on air quality. A standard set with a high degree of instability could
2 have the effect of reducing public health protection because shifts in and out of attainment could
3 disrupt an area's ongoing implementation plans and associated control programs.
4 Therefore, consistent with recent reviews of the Os, PM and NC>2 NAAQS, we have
5 focused here on a concentration-based form (as compared with an expected or allowable
6 exceedance-based form) and are considering this with regard to an average over 3 years. As
7 noted in the review of the Os NAAQS (USEPA, 2007), concentration-based forms32 may better
8 reflect pollutant-associated health risks than forms based on expected exceedances because
9 concentration-based forms give proportionally greater weight to periods of time when pollutant
10 concentrations are well above the level of the standard than to times when the concentrations are
11 just above the standard, while an expected exceedance form would give the same weight to
12 periods of time with concentrations that just exceed the standard as to times when concentrations
13 greatly exceed the standard. Concentration-based forms also provide greater regulatory stability
14 than a form based on allowing only a single expected exceedance.
15 In considering specific concentration-based forms on which to focus the current review,
16 we have initially considered such forms as have been considered in other NAAQS reviews,
17 including the 98th and 99th percentile forms averaged over three years, which were considered in
18 the most recent PM and NO2 NAAQS reviews (USEPA, 2005; USEPA, 2008b).33 We note that
19 a 99th percentile form for an 8-hour daily maximum standard would correspond to the fourth
20 highest daily maximum 8-hour average concentration in a year. We have initially considered
21 that a 98th percentile form or forms of 99th percentile or fourth highest daily maximum would
22 provide appropriate balances between limiting peak CO concentrations and providing a stable
23 regulatory target.
24 When considering results of the second draft REA as they relate to standard form, we
25 note that a decision on form must be made in conjunction with selection of a particular standard
26 level. The primary emphasis in such a decision will be on the level of public health protection
27 provided by the combination of form and level. The current form of both the 8-hour and 1-hour
28 standards is a maximum, not to be exceeded more than once per year. In the section below we
29 consider the results of the second draft REA for different combinations of level with both a 99th
30 percentile concentration-based form and the current "exceeded only once per year" form.
32 The term "concentration-based" forms is used here to refer to forms based on a concentration associated
with a targeted point on the distribution of pollutant concentrations, as distinguished from a "not to be exceeded" or
"to be exceeded no more than x times in a year' form.
33 Given the limited time frame for this review, only one alternative form (for the 8-hour and 1-hour
standards) has been evaluated in the REA, such that we do not have the results of the quantitative assessment in
terms of a 98th percentile form available to us at this time.
March 2010 2-49 Draft - Do Not Quote or Cite
-------
1 2.3.3.2 Potential Alternative Levels
2 In considering potential alternative standard levels with regard to their potential to
3 provide greater protection than that afforded by the current standard against CO-related adverse
4 health effects, we have taken into account scientific evidence from both experimental and
5 epidemiologic studies, as well as the uncertainties and limitations in that evidence, and the
6 quantitative estimates of exposure and dose provided by the second draft REA. We note that the
7 scientific evidence and quantitative assessment can provide insights into alternative standard
8 levels only within the context of specific averaging times and forms. Therefore, this section
9 considers the evidence and quantitative analysis as they relate to alternative levels particular to
10 different forms and averaging times. The primary focus of staffs initial considerations in this
11 draft Policy Assessment is on the 8-hour standard because it has long been the controlling
12 standard for air quality.
13 As an initial matter, we recognize that indoor sources of CO can be important
14 determinants of population exposures to CO and to population distributions of COHb including
15 maximum COHb levels. For some portions of the population these sources may be the primary
16 influences on external exposure-related COHb levels. However, we take note of the conclusions
17 drawn in the previous review that the contribution of indoor sources to individual exposures and
18 associated COHb levels cannot be effectively mitigated by ambient air quality standards (e.g., 59
19 FR 38914) and so focus here on COHb levels resulting from ambient CO exposures. At the
20 same time, we additionally recognize people whose exposures to nonambient CO result in
21 elevated COHb levels, and particularly those whose COHb levels may be appreciably influenced
22 by additional ambient CO exposures. Findings of previous quantitative assessments in which
23 some indoor sources were represented (e.g., gas stoves and tobacco smoke), however, indicated
24 that indoor source contributions to population maximum COHb levels were substantial in
25 comparison to ambient contributions. While we are limited with regard to information regarding
26 CO emissions from indoor sources today and how they may differ from the time of the 2000
27 assessment, we note that ambient contributions have notably declined and indoor source
28 contributions from some sources may also have declined. Thus, we have no firm basis to
29 conclude a different role for indoor sources today with regard to contribution to population CO
30 exposure and COHb levels.
31 We believe that the integration of the health effects evidence with the exposure and dose
32 estimates will be particularly important to informing conclusions regarding standard levels and
33 forms considered to provide protection of public health with an adequate margin of safety. With
34 regard to the scientific evidence, as discussed in section 2.2.1 above, we have given principal
35 emphasis to findings of the multilaboratory controlled human exposure study of individuals with
36 pre-existing CAD (Allred et al., 1989a, 1989b, 1991). In this study, controlled experimental
March 2010 2-50 Draft - Do Not Quote or Cite
-------
1 exposures to CO that resulted in a post-exercise study mean COHb level as low as 2.0% relative
2 to clean-air exposures that resulted in a mean COHb level of 0.6% (post-exercise) were
3 associated with myocardial ischemia-related effects (including, specifically, reduced time to
4 exercise-induced angina and ST-segment change). A dose-response analysis of the study subject
5 specific data, in which individual regressions of study subject responses at baseline COHb and at
6 the two increased COHb levels were averaged, indicated decreases of roughly 1.9% in time to
7 exercise-induced angina and 3.9% in time to exercise-induced ST-segment change per 1%
8 increase in COHb concentration, with no evidence of a measurable threshold at the controlled
9 exposures evaluated (ISA, section 5.2.4; Allred et al., 1989a, 1989b, 1991). For our purposes
10 here, we additionally have considered the epidemiological studies with regard to the extent to
11 which this evidence is consistent with and generally supportive of conclusions drawn from the
12 combined consideration of the controlled human exposure evidence with estimates from the
13 exposure and dose assessment. We take note of the epidemiological studies showing positive
14 associations with IHD and CHD health outcomes described in section 2.2.1 above, three of
15 which were based in a study area in which the current standard was met, the association for one
16 of which was statistically significant. In our consideration of the evidence, we recognize
17 CAS AC advice with regard to aspects of the epidemiological evidence that are particularly
18 problematic for CO and the weight to be given to the well-conducted clinical studies (Brain and
19 Samet, 2010).
20 In considering health risk implications for the simulated at-risk population associated
21 with dose estimates from the quantitative exposure/dose assessment we compared the estimates
22 to potential health effects benchmark levels of 1.5%, 2.0%, 2.5% and 3% COHb (as described in
23 section 2.2.2 above). In considering the population estimates of maximum end-of-hour COHb
24 levels from the draft REA for the purposes of considering the level for the 8-hour standard that
25 provides appropriate public health protection, we have focused on the results in terms of
26 percentage of simulated at-risk population in each study area estimated to experience maximum
27 end-of-hour COHb levels below each of the two lower potential health effects benchmarks and
28 an intermediate COHb level. Based on analysis of the air quality data used in the draft REA to
29 identify the equivalent levels for alternative forms, the five different air quality conditions
30 simulated for each of the two study areas provide information on an array of alternative levels for
31 the current and an alternative form (Table 2-6).
32 In considering the estimates provided by the draft REA (summarized in Table 2-6 below),
33 we note that differences between the two study areas with regard to the simulated at-risk
34 population distributions of maximum end-of-hour COHb levels (noted in the section 6.2 of the
35 draft REA) lead to differences in the proportion of the population protected from maximum end-
36 of-hour COHb levels at or above the potential health effect benchmarks being considered here.
March 2010 2-51 Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
For example, the percentage of the simulated at-risk population estimated to experience
maximum COHb levels below 2% is 99.5% or greater in the Los Angeles study area for all
simulated levels for either form, while the percentage of the simulated population in Denver
estimated to experience maximum end-of-hour COHb levels below 2% is less than 99%
(approximately 96.6%) for the highest alternative standard level simulated for either form.
Alternatively, the percentage of the simulated population estimated to experience maximum
COHb levels below 1.5% is more than 95% in the Los Angeles study area and more than 80% in
the Denver study area for all simulated levels for either form. Further, the percentage of the
simulated at-risk population estimated to experience maximum COHb levels below an
intermediate COHb level of 1.8%, is more than 99% in the Los Angeles study area and more
than 98% in the Denver study area for all simulated levels for either form.
Table 2-6. Portion of simulated at-risk population with maximum end-of-hour COHb
levels below the indicated COHb level under alternative levels for the 8-hour
standard for the current form and a 99th percentile daily maximum form.
Form
Second Highest Non-
overlapping 8-hour
Concentration
99th Percentile of 8-
hour Daily Maximum
Concentration
Level
9.4
6.5
5.7
5.6
5.4
3.1
8.2
7.2
5.7
5.1
5.0
4.7
4.3
4.1
2.8
< 2 pe
CO
Los
Angeles
99.5
99.9
99.9
99.9*
99.9
99.5
99.9
99.9*
99.9
99.9
rcent
Hb
Denver
96.6
99.2
99.6
99.6
100.0*
96.6
99.2
99.6
99.6
100.0*
< 1 .8 p<
CO
Los
Angeles
99.7
99.8
99.6*
99.8
99.7
99.6*
99.8
99.8
s rcent
Hb
Denver
98.1
98.9
99.2
100.0*
98.1
98.9
99.2
100.0*
<1.5p
CO
Los
Angeles
95.4
98.6
99.1
98.5*
99.4
95.4
98.6
98.5*
99.1
99.4
e rcent
Hb
Denver
80.9
92.9
95.8
96.2
99.4*
80.9
92.9
95.8
96.2
99.4*
* Asterisks indicate simulations based on "as is" (2006) air quality conditions.
15
16
17
18
19
Taking into account the differences between the simulated at-risk populations in the two
study areas, recognizing considerations regarding maximum COHb benchmarks of 1.5 to 2% and
above, and being mindful of the support provided by the epidemiological studies, we consider
the draft REA estimates of maximum end-of-hour COHb levels with regard to level of protection
March 2010
2-52
Draft - Do Not Quote or Cite
-------
1 estimated to be provided by the different alternative levels for the 8-hour standard (Table 2-6
2 below). In further considering the weight to place on this information, we are mindful of several
3 key aspects of the evidence (described in more detail in section 2.2.1 above):
4 • Uncertainty in the clinical significance of the effects observed in the controlled human
5 exposure studies of CAD patients.
6 • The lack of evidence for a measurable threshold effect for the effects observed in the
7 human clinical studies and the lack of studies that have evaluated effects of
8 experimentally controlled short-term CO exposures of individuals with CAD that
9 resulted in study mean COHb levels below 2%.
10 • Uncertainty associated with interpretation of COHb levels estimated in the quantitative
11 assessment to result from simulated CO exposure concentrations much lower than the
12 experimental CO exposure concentrations used in the controlled human studies to
13 increase subject COHb levels to COHb study targets.
14 • The lack of studies of COHb levels associated with health effects in other potentially
15 susceptible populations.
16 • Limitations associated with the epidemiological studies for CO which complicate our
17 use for quantitative purposes.
18
19 Staff find that identification of the range of alternative standards that may be appropriate
20 to consider differs based on the weight placed on different aspects of the evidence and on
21 different aspects of the quantitative dose estimates, as well as on public health policy decisions
22 regarding the significance of the effects considered, the appropriate COHb benchmark on which
23 to focus in considering maximum end-of-hour COHb levels, and the targeted proportion of the
24 at-risk population. For example, focusing on the 2.0% COHb benchmark and considering a
25 target for the at-risk population of 95 or 96% across the two study areas would indicate support
26 for retaining the current level of the 8-hour standard. That is, the draft REA results in Table 2-6
27 indicate that under the current standard, the maximum end-of-hour COHb levels for at least 96%
28 of the at-risk populations in the two study areas would be expected to fall below the 2%
29 benchmark.34 Alternatively, consideration of a higher population target of 99% while focusing
30 on the 2% COHb benchmark, would lead to consideration of a level of approximately 5 to 6
31 ppm, as would a population target of 98% below the 1.8% COHb benchmark, while a focus on a
32 COHb benchmark of 1.5% with a target of 95% for the at-risk population would indicate support
33 for consideration of a level somewhat below 5 ppm. On the other hand, consideration of a target
34 level of 99% in combination with a focus on the 1.5% COHb benchmark would indicate support
35 for a somewhat lower level, of about 3 ppm.
34 Results are not currently available from the second draft REA for the percentage of the simulated at-risk
population with end-of-hour COHb levels below 1.8%.
March 2010 2-53 Draft - Do Not Quote or Cite
-------
1 Thus, we note that placing weight on the exposure assessment results for the population
2 target of 95% or 96% for a benchmark level of 2% COHb, in combination with a recognition of
3 limitations in the epidemiological evidence for CO with regard to its use in drawing quantitative
4 conclusions regarding a level for the standard, would provide a basis for considering retaining
5 the current 8-hour standard. Alternatively, placing weight on the three epidemiological studies
6 for the location in which the current standard was met, the association in one of which
7 statistically significant, and consideration of the ambient CO concentrations in that study
8 (Atlanta, 99th percentile 8-hour daily maximum ranging up to 4.9 ppm during the study period)
9 would lead to consideration of the lower part of the range identified above.
10 While the focus in this draft Policy Assessment is on consideration of the controlling 8-
11 hour standard, we have also considered the rationale for the existing 1-hour standard as generally
12 providing protection from effects that might result from short peak exposures over this shorter
13 exposure duration (as described in section 2.3.1 above). We find value in retaining this standard
14 for that purpose, recognizing that it may provide such protection independent from the type of
15 protection provided by the 8-hour standard.
16 We note that the results of the quantitative exposure/dose analysis also provides estimates
17 for the ten model simulations (described in the draft REA) in terms of potential alternative levels
18 for a 1-hour standard either of current form or a revised 99th percentile daily maximum form
19 (Table 2-7 below). These estimates indicate the extent to which levels for the 1-hour standard
20 that are appreciably lower than the current level might influence population maximum end-of-
21 hour COHb levels in the situation where the 8-hour standard was of a magnitude such that it no
22 longer was the controlling force for ambient CO concentrations. From these estimates it can be
23 seen that an appreciable reduction in the level for the 1-hour standard would be required to
24 achieve an impact on population COHb levels comparable to those achieved by the current 8-
25 hour standard or levels only somewhat lower. For example, to achieve a similar distribution of
26 expected COHb levels as the current 8-hour standard, the results from the draft REA indicate that
27 a level of approximately 15 ppm would be required for the 1-hour standards, either in terms of
28 the current form or of a 99th percentile daily maximum form (Table 2-7 below).
March 2010 2-54 Draft - Do Not Quote or Cite
-------
1 Table 2-7. Portion of simulated at-risk population with maximum end-of-hour COHb
2 levels below the indicated COHb level under alternative levels for the 1-hour
3 standard for the current form and a 99th percentile daily maximum form.
4
5
6
7
8
9
10
11
12
13
14
15
Form
Second Highest 1-
hour concentration
99th percentile of
1-hour Daily
Maximum
Concentration
Level
16.2
11.8
11.2
9.7
9.3
8.2
8.1
7.2
6.8
4.6
13.3
11.6
9.2
8.0
7.7
7.4
7.1
6.7
4.5
<2 pe
CO
Los
Angeles
99.5
99.9*
99.9
99.9
99.9
99.5
99.9
99.9*
99.9
99.9
rcent
Hb
Denver
96.6
99.2
99.6
99.6
100.0*
96.6
99.2
99.6
99.6
100.0*
< 1.8 p<
CO
Los
Angeles
100.0
99.6*
99.7
99.8
99.8
99.7
99.6*
99.8
99.8
3 rcent
Hb
Denver
98.1
98.9
99.2
100.0*
98.1
98.9
99.2
100.0*
< 1.5 p<
CO
Los
Angeles
95.4
98.5*
98.6
99.1
99.4
95.4
98.6
98.5*
99.1
99.4
3 rcent
Hb
Denver
80.9
92.9
95.8
96.2
99.4*
80.9
92.9
95.8
96.2
99.4*
* Asterisks indicate simulations based on "as is" (2006) air quality conditions.
With regard to the current 1-hour standard, staff has identified several policy options in
this draft Policy Assessment that may be appropriate to consider. First, we note that the current
1-hour standard may be considered to provide protection from effects which might be
encountered from infrequent, very short duration peak ambient concentrations. With regard to
such a purpose for the 1-hour standard independent of a focus on COHb levels as illustrated in
Table 2-7 above, we have not identified information that might provide for an alternative level
and form. Alternatively, while recognizing that the 8-hour standard has traditionally been the
controlling standard, we have also considered alternative levels for the 1-hour standard drawing
from the results of the draft REA in similar manner to the discussion above for the 8-hour
standard. In considering the alternatives for public health policy judgments identified above for
the 8-hour standard, we note the results of the draft REA with regard to alternative levels for the
March 2010
2-55
Draft - Do Not Quote or Cite
-------
1 1-hour standard might be concluded to indicate support for a range from approximately 15 ppm
2 at the upper end to approximately 5 ppm at the lower end, combined with a 99th percentile form.
3 Further, in light of the role of the 8-hour standard as the controlling standard, and
4 depending on judgments made regarding additional protection that may be provided by the 1-
5 hour standard from infrequent, very short duration peak ambient concentrations, we note that it
6 may also be appropriate to consider revoking the 1-hour standard in conjunction with revision of
7 the 8-hour standard to provide increased protection from maximum COHb levels that may result
8 from ambient CO concentrations.
9 2.3.4 Initial Staff Conclusions on Alternative Standards
10 Initial staff conclusions for consideration in making decisions on the primary standards
11 for CO, together with supporting conclusions from sections 2.2.3 and 2.3.2 above, are briefly
12 summarized below. In drawing these initial conclusions, we are mindful that the Act requires
13 standards to be set that, in the Administrator's judgment, are requisite to protect public health
14 with an adequate margin of safety, such that the standards are to be neither more nor less
15 stringent than necessary. Thus, the Act does not require that NAAQS be set at zero-risk levels,
16 but rather at levels that avoid unacceptable risks to public health.
17 (1) Staff initially concludes that the combined consideration of the body of evidence and
18 the quantitative exposure and dose estimates provide support for standards at least as
19 protective as the current suite of standards to provide appropriate public health
20 protection for susceptible populations, including most particularly individuals with
21 cardiovascular disease, against effects of CO in exacerbating conditions of reduced
22 oxygen availability to the heart.
23 (2) Staff also initially concludes that, depending on public health policy judgments about
24 protecting public health with an adequate margin of safety, the information available
25 in this review supports consideration of either retaining or revising the current suite of
26 standards, with any such revisions focused on providing increased public health
27 protection.
28 (3) With regard to the indicator for the CO standards, staff initially concludes that it is
29 appropriate to continue to use measurements of CO in accordance with federal
30 reference methods as the indicator to address effects associated with exposure to CO.
31 (4) With regard to the current 8-hour standard, staff initially concludes that it is
32 appropriate to consider the following range of policy options:
33 a. Retaining the 8-hour averaging time.
34 b. Retaining or revising the level of the standard, with consideration given to
35 levels within the range from the current level of 9 ppm at the upper end to
March 2010 2-56 Draft - Do Not Quote or Cite
-------
1 approximately 3 ppm at the lower end to provide protection from maximum
2 COHb levels that may result from ambient CO exposures.
3 c. In conjunction with consideration of revising the level, it is also appropriate to
4 consider revising the form of the standard to a 99th percentile or fourth-highest
5 daily maximum form, including consideration of such a form that is the
6 average of this concentration across three years.
7 (5) With regard to the current 1-hour standard, staff initially concludes that it is
8 appropriate to consider the following range of policy options:
9 a. Retaining the current 1-hour standard to provide protection from effects which
10 might be encountered from infrequent, very short duration peak ambient
11 concentrations.
12 b. Revising the standard, with consideration given to levels within the range
13 from approximately 15 ppm at the upper end to approximately 5 ppm at the
14 lower end, combined with a 99th percentile or fourth-highest daily maximum
15 form, to provide protection from maximum COHb levels that may result from
16 ambient CO exposures. Consideration of a revised level for the 1-hour
17 standard within this range would be intended to provide the same protection as
18 revision of the 8-hour standard within the range identified in item (4)b above.
19 c. In conjunction with consideration of revising the level of the 8-hour standard
20 to a lower level, it is also appropriate to consider revoking the 1-hour standard
21 in recognition of the increased protection provided by a revised 8-hour
22 standard from maximum COHb levels that may result from ambient CO
23 exposures.
24 2.4 KEY UNCERTAINTIES AND AREAS FOR FUTURE RESEARCH AND
25 DATA COLLECTION
26 Staff believes it is important to highlight key uncertainties associated with establishing
27 ambient standards for CO. Such key uncertainties and recommendations for health-related
28 research, model development, and data gathering are outlined below. In some cases, research in
29 these areas can go beyond aiding standard setting to aiding in the development of more efficient
30 and effective control strategies. We note, however, that a full set of research recommendations
31 to meet standards implementation and strategy development needs is beyond the scope of this
32 discussion. We have initially identified the following key uncertainties, research questions and
33 data gaps that have been highlighted in this review of the health-based primary standards.
34 • A critical aspect of our consideration of the evidence and the quantitative dose
35 estimates is our interpretation of the controlled human clinical studies. Expansion of
March 2010 2-57 Draft - Do Not Quote or Cite
-------
1 this evidence base in several areas would reduce uncertainty in our interpretation for
2 purposes of risk characterization.
3 o Additional information relevant to health effects of short-term increases in CO
4 exposure resulting in COHb levels below 2% in individuals with CAD.
5 o Additional research to clarify the clinical significance of the sensitive markers
6 of myocardial ischemia observed in controlled human exposure studies of
7 CAD patients.
8 o Research to reduce uncertainty associated with interpretation of COHb levels
9 estimated in the quantitative assessment to result from simulated CO exposure
10 concentrations much lower than the experimental CO exposure concentrations
11 used in the controlled human studies to increase subject COHb levels to
12 COHb study targets.
13 • Existing clinical data correlates CO-induced effects (i.e., cardiovascular ischemia
14 observed at specific COHb levels) to COHb levels resulting from short-term, high CO
15 exposure concentrations. However, there is lack of studies evaluating effects of CO, in
16 terms of COHb levels, at lower ambient concentrations and/or longer term exposures
17 periods.
18 • Further research to characterize CO effects for other potentially susceptibility
19 populations, including populations with preexisting diseases other than CAD that limit
20 O2 availability, such as anemia and obstructive respiratory diseases; populations at
21 certain lifestages (i.e., older adults and fetuses at critical stages of development);
22 populations with elevated baseline COHb due to increased CO production, such as
23 individuals with diabetes;, and visitors to high altitude.
24 • Our interpretation of the epidemiological evidence for CO would benefit from a better
25 understanding of the possible role of co-pollutants.
26 • Our future assessments would benefit from research on studies evaluating effects of
27 CO linked to biomarkers other than COHb and research on mechanisms of CO-
28 induced effects other that those associated with limiting O2 availability at ambient CO
29 levels.
March 2010 2-58 Draft - Do Not Quote or Cite
-------
1 REFERENCES
2 Adams, K.F.; Koch, G.; Chatterjee, B.; Goldstein, G.M.; O'Neil, J.J.; Bromberg, P.A.; Sheps, D.S.; McAllister, S.;
3 Price, C.J.; Bissette, J. (1988) Acute elevation of blood carboxyhemoglobin to 6% impairs exercise
4 performance and aggravates symptoms in patients with ischemic heart disease. J. Am. Coll. Cardiol. 12:
5 900-909
6 Allred, E.N.; Bleecker, E.R.; Chaitman, B.R.; Dahms, T.E.; Gottlieb, S.O.; Hackney, J.D.; Pagano, M; Selvester,
7 R.H.; Walden, S.M.; Warren, J. (1989a) Short-term effects of carbon monoxide exposure on the exercise
8 performance of subjects with coronary artery disease. N. Engl. J. Med. 321: 1426-1432.
9 Allred, E.N.; Bleecker, E.R.; Chaitman, B.R.; Dahms, T.E.; Gottlieb, S.O.; Hackney, J.D.; Hayes, D.; Pagano, M.;
10 Selvester, R.H.; Walden, S.M.; Warren, J. (1989b) Acute effects of carbon monoxide exposure on
11 individuals with coronary artery disease. Cambridge, MA: Health Effects Institute; research report no. 25.
12 Allred E. N.; Bleecker, E. R.; Chaitman, B. R.; Dahms, T. E.; Gottlieb, S. O.; Hackney, J. D.; Pagano, M.; Selvester,
13 R. H.; Walden, S. M.; Warren, J. (1991) Effects of carbon monoxide on myocardial ischemia. Environ.
14 Health Perspect. 91:89-132.
15 AHA. (2003). Heart and Stroke Facts. American Heart Association, Dallas, TX. Available at:
16 http://www.americanheart.org/downloadable/heart/1056719919740HSFacts2003text.pdf
17 Anderson EW, Andelman RJ, Strauch JM, Fortuin NJ and Knelson, JH (1973). Effect of low level carbon monoxide
18 exposure on onset and duration of angina pectoris. Annals of Internal Medicine 79: 46-50.
19 Basan M.M. (1990) Letter to the Editor. N. Engl J Med 332: 272.
20 Bell ML; Peng RD; Dominici F; Samet JM. (2009). Emergency admissions for cardiovascular disease and ambient
21 levels of carbon monoxide: Results for 126 U.S. urban counties, 1999-2005. Circulation, 120:949-955.
22 Benignus V.A., Muller, K.E., Barton, C.N., Prah, J.K., 1987, Effect of low level carbon monoxide on compensatory
23 tracking and event monitoring, Neurotoxicol Teratol 9:227-234.
24 Benignus V.A.,Muller, Ke.e.,Smith, M.M, Pieper, K.S., Prah, J.D.1990a Compensatory tracking in humans with
25 elevated carboxyhemogloin. Neurotoxicol Teratol 12: 105-110.
26 Benignus V.A.,Muller, Ke.e.,Mallot, C.M, J.D. 1990b Dose-effects functions for carboxyhemoglobin and behavior.
27 Neurotoxicol Teratol 12:111-118.
28 Brain JD. (2009). Letter from Dr. Joseph Brain to Administrator Lisa Jackson. Re: Consultation on EPA's Carbon
29 Monoxide National Ambient Air Quality Standards: Scope and Methods Plan for Health Risk and Exposure
30 Assessment. CASAC-09-012. July 14, 2009
31 Brain JD and Samet JM (2010). Letter from Drs. J.D. Brain and J.M. Samet to Administrator Lisa Jackson. Re:
32 Review of the EPA's Integrated Science Assessment for Carbon Monoxide (Second External Review
33 Draft). EPA-CASAC-10-005. January 20, 2010.
34 Drinkwater, B. L.; Raven, P. B.; Horvath, S. M.; Gliner, J. A.; Ruhling, R. O.; Bolduan, N. W.; Taguchi, S. (1974)
35 Air pollution, exercise, and heat stress. Arch. Environ. Health 28: 177-181.
36 Ekblom, B.; Huot, R. (1972) Response to submaximal and maximal exercise at different levels of
37 carboxyhemioglobin. Acts Physiol. Scand. 86; 474-482.
March 2010 2-59 Draft - Do Not Quote or Cite
-------
1 Ohio et al., 2008 Ohio AJ; Stonehuerner JG; Dailey LA; Richards JH; Madden MD; Deng Z; Nguyen NB;
2 Callaghan KD; Yang F; Piantadosi CA (2008). Carbon monoxide reversibly alters iron homeostasis and
3 respiratory epithelial cell function.. Am J Respir Cell Mol Biol, 38: 715-723.
4 Horvath S. M. (1975) Influence of carbon monoxide on cardiac dynamics in nrmal and cardiovascular stressed
5 animals. Santa Barbars, CA: University of California, Institute on Environmental Stress; final report on
6 grant ARB-2006. Available from: NTIS, Springfield, VA; PB-254821.
7 Johnson T, Capel J, Paul R, Wijnberg L. (1992). Estimation of Carbon Monoxide Exposure and associated
8 Carboxyhemoglobin levels in Denver Residents Using a Probabalistic verion of NEM, prepared by
9 International Technology for U.S. EPA, Office of Air Quality Planning and Standards, Durham, NC,
10 Contract No. 68-DO-0062
11 Johnson T, Mihlan G, LaPointe J, Fletcher K, Capel J. (2000). Estimation of Carbon Monoxide Exposures and
12 Associated Carboxyhemoglobin Levels for Residents of Denver and Los Angeles Using pNEM/CO
13 (Version 2.1). Report prepared by ICF Consulting and TRJ Environmental, Inc., under EPA Contract No.
14 68-D6-0064. U.S. Environmental Protection Agency, Research Triangle Park, North Carolina. Available
15 at: http://www.epa.gov/ttn/fera/human_related.html. June 2000.
16 KleinmanMT, DavidsonDM, Vandagriff RB, Caiozzo VJ, Whittenberger JL.(1989). Effects of short-term
17 exposure to carbon monoxide in subjects with coronary artery disease. Arch. Environ. Health 44: 361-369.
18 Kleinman MT, Leaf DA, Kelly E, Caiozzo V, Osann K, O'Niell T. (1998). Urban angina in the mountains: effects of
19 carbon monoxide and mild hypoxemia on subjects with chronic stable angina. Arch. Environ.Health 53:
20 388-397.
21 Koken PJM; Piver WT; Ye F; Elixhauser A; Olsen LM; Portier CJ. (2003). Temperature, air pollution, and
22 hospitalization for cardiovascular diseases among elderly people in Denver. Environ Health Perspect, 111:
23 1312-1317.
24 Leaf, D. A.; Kleinman, M. T. (1996a) Urban ectopy in the mountains: carbon monoxide exposure at high altitude.
25 Arch. Environ. Health 51: 283-290.
26 Linn WS; Szlachcic Y; Gong H Jr; Kinney PL; Berhane KT. (2000). Air pollution and daily hospital admissions in
27 metropolitan Los Angeles. Environ Health Perspect, 108: 427-434.
28 Longo, L. D. (1977) The biological effects of carbon monoxide on the pregnant woman, fetus, and newborn infant.
29 Am. J. Obstet. Gunacol. 129: 69-103.
30 Mann JK; Tager IB; Lurmann F; Segal M; Quesenberry CP Jr; Lugg MM; Shan J; Van den Eeden SK. (2002). Air
31 pollution and hospital admissions for ischemic heart disease in persons with congestive heart failure or
32 arrhythmia. Environ Health Perspect, 110: 1247-1252.
33 McClellan RO, Chairman, Clean Air Scientific Advisory Committee. (1991). Letter to William K. Reilly, EPA
34 Administrator, July 17, 1991.
3 5 McClellan RO, Chairman, Clean Air Scientific Advisory Committee. (1992). Letter to William K. Reilly, EPA
36 Administrator, August 1, 1992.
37 Metzger KB; Tolbert PE; Klein M; Peel JL; Flanders WD; Todd KH; Mulholland JA; Ryan PB; Frumkin H. (2004).
3 8 Ambient air pollution and cardiovascular emergency department visits. Epidemiology, 15: 46-56.
39 Peel JL; Metzger KB; Klein M; Flanders WD; Mulholland JA; Tolbert PE. (2007). Ambient air pollution and
40 cardiovascular emergency department visits in potentially sensitive groups. Am J Epidemiol, 165: 625-633.
March 2010 2-60 Draft - Do Not Quote or Cite
-------
1 Putz, V. R.; Johnson, B. L.; Setzer, J. V. (1976) Effects of CO on vigilance performance effects of low level carbon
2 monoxide on divided attention, pitch discrimination, and the auditory eve pot tial. Cincinnati, OH: U. S.
3 Department of Health, Education, and Welfare National Institute for Occupational Safety and Health;
4 report no. NIOSH-77-124. Available from: NTIS, Springfield, VA; PB-274219.
5 Putz, V. R.; Johnson, B. L; Setzer, J. V. (1979) A comparative study of the effects of carbon monoxide and
6 methylene chloride on human performance. J. Environ. Pathol. Toxicol. 2: 97-112.
7 Raven, P. B. Drinkwater, B. L.; Horvath, S. M; Ruhling, R. O.; Gliner, J. A.; Sutton, J. C.; Bolduan, N. W. (1974a)
8 Age, smoking habits, heat stress, and their interactive effects with carbon monoxide and
9 peroxyacetylnitrate on man's aerobic power. Int. J. Biometeorol. 18: 222-232.
10 Raven, P. B.; Drinkwater, B. L.; Ruhling, R. O.; Taguchi, S.; Gliner, J.; Horvath, S. M. (1974b) Effect of carbon
11 monoxide and peroxyacetyl itrate on man's maximal aerobic capacity. J. Appl. Physiol. 36: 288-293.
12 Sheps, D. S.; Adams, K. F., Jr.; Bromberg, P. A.; Goldstein, G. M.; O'Neil, J. J.; Horstman, D.; Koch, G. (1987)
13 Lack of effect of low levels of carboxyhemoglobin on cardiovascular function in patients with ischemic
14 heart disease. Arch. Environ. Health 42: 108-116.
15 Sheps, D.S.; Herbst, M.C.; Hinderliter, A.L.; Adams, K.F.; Ekelund, L.G.; O'Neil, J.J.; Goldstein, G.M.; Bromberg,
16 P.A.; Dalton, J.L.; Ballenger, M.N.; Davis, S.M.; Koch, G.G. (1990) Production of arrythmias by elevated
17 carboxyhemoglobin in patients with coronary artery disease. Ann. Intern. Med. 113: 343-351.
18 Symons JM; Wang L; Guallar E; Howell E; Dominici F; Schwab M; Ange BA; Samet J; Ondov J; Harrison D; Geyh
19 A. (2006). A case-crossover study of fine paniculate matter air pollution and onset of congestive heart
20 failure symptom exacerbation leading to hospitalization. Am J Epidemiol, 164: 421-433.
21 Tolbert PE; Klein M; Peel JL; Sarnat SE; Sarnat JA. (2007). Multipollutant modeling issues in a study of ambient air
22 quality and emergency department visits in Atlanta. J Expo Sci Environ Epidemiol, 17: S29-S35.
23 U.S. EPA (1979). Air Quality Criteria for Carbon Monoxide. Research Triangle Park, NC: Office of Health and
24 Environmental Assessment, Environmental Criteria and Assessment Office, report no. EPA/600/8-79-022.
25 U.S. EPA (1984a). Revised Evaluation of Health Effects Associated with Carbon Monoxide Exposure: An
26 Addendum to the 1979 EPA Air Quality Criteria Document for Carbon Monoxide. Office of Health and
27 Environmental Assessment, Environmental Criteria and Assessment Office, Research Triangle Park, NC.
28 Report no. EPA-600/8-83-033F.
29 U.S. EPA (1984b). Review of the NAAQS for Carbon Monoxide: Reassessment of Scientific and Technical
30 Information. Office of Air Quality Planning and Standards, Research Triangle Park. NC. Report no. EPA-
31 450/584-904.
32 U.S. EPA. (1991). Air Quality Criteria for Carbon Monoxide. Office of Health and Environmental Assessment,
3 3 Research Triangle Park, NC. EPA/600/8-90/045F.; .Available at:
34 http://www.epa.gov/ttn/naaqs/standards/co/s_co_pr.html
35 U.S. EPA (1992). Review of the National Ambient Air Quality Standards for Carbon Monoxide: Assessment of
36 Scientific and Technical Information. OAQPS Staff Paper. Office of Air Quality Planning and Standards,
37 Research Triangle Park, NC. EPA/452/R-92-004.
38 U.S. EPA (2000). Air Quality Criteria for Carbon Monoxide. National Center for Environmental Assessment,
3 9 Office of Research and Development, Research Triangle Park, NC. EPA/600/P-99/00 IF. Available at:
40 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=18163.
March 2010 2-61 Draft - Do Not Quote or Cite
-------
1 U.S. EPA (2005). Review of the National Ambient Air Quality Standards for Paniculate Matter: Policy Assessment
2 of Scientific and Technical Information. Office of Air Quality Planning and Standards. OAQPS Staff
3 Paper. Office of Air Quality Planning and Standards, Research Triangle Park, NC. EPA-452/R-05-005a.
4 U.S. EPA (2007) Review of the National Ambient Air Quality Standards for Ozone: Policy Assessment of Scientific
5 and Technical Information OAQPS Staff Paper. Office of Air Quality Planning and Standards, Research
6 Triangle Park, NC. EPA-452/R-07-007.
7 US EPA (2008a). Plan for Review of the National Ambient Air Quality Standards for Carbon Monoxide. U.S.
8 Environmental Protection Agency, Research Triangle Park, NC, report no. EPA/452R-08-005. Also known
9 as Integrated Review Plan. Available at: http://www.epa.gOv/ttn/naaqs/standards/co/s co cr_pd.html.
10 U.S. EPA (2008b) Risk and Exposure Assessment to Support the Review of the NO2 Primary National Ambient Air
11 Quality Standard. Office of Air Quality Planning and Standards, Research Triangle Park, NC. EPA-452/R-
12 08-008a.
13 US EPA (2009). Risk and Exposure Assessment to Support the Review of the SO2 Primary National Ambient Air
14 Quality Standard. EPA-452/R-09-007. August 2009. Available
15 athttp://www.epa.gov/ttn/naaqs/standards/so2/data/200908SO2REAFinalReport.pdf.
16 U.S. EPA (20lOa) Integrated Science Assessment for Carbon Monoxide, National Center for Environmental
17 Assessment, Research Triangle Park, NC. EPA/600/R-09/019F. Available at:
18 http://www.epa.gOv/ttn/naaqs/standards/co/s co crisa.html
19 U.S. EPA (20 lOb) Risk and Exposure Assessment to Support the Review of the Carbon Monoxide Primary National
20 Ambient Air Quality Standards, Second External Review Draft, U.S Environmental Protection Agency,
21 Research Triangle Park, NC, report no. EPA/600/R-09/019B. Available at:
22 http://www.epa.gOv/ttn/naaqs/standards/co/s co crrea.html
23 Wellenius GA; Bateson TF; Mittleman MA; Schwartz J. (2005). Paniculate air pollution and the rate of
24 hospitalization for congestive heart failure among medicare beneficiaries in Pittsburgh, Pennsylvania. Am J
25 Epidemiol, 161: 1030-1036.
26 WHO (2008). Harmonization Project Document No. 6. Part 1: Guidance document on characterizing and
27 communicating uncertainty in exposure assessment. Available at:
28 http://www.who.int/ipcs/methods/harmonization/areas/exposure/en/.
29 Zanobetti A; Schwartz J (2001). Are diabetics more susceptible to the health effects of airborne particles? Am J
30 Respir Crit Care Med, 164: 831-833.
March 2010 2-62 Draft - Do Not Quote or Cite
-------
1 3 CONSIDERATION OF A SECONDARY STANDARD FOR CO
2 This chapter focuses on the key policy-relevant issues related to the review of welfare-
3 related effects of CO. Under Section 109(b) of the Clean Air Act, a secondary standard is to be
4 established at a level "requisite to protect the public welfare from any known or anticipated
5 adverse effects associated with the presence of the pollutant in ambient air." Section 302(h) of
6 the Act defines effects on welfare in part as "effects on soils, water, crops, vegetation, man-made
7 materials, animals, weather, visibility, and climate."
8 Specifically, this chapter first summarizes the history of EPA's consideration of
9 secondary standards for CO. Considerations is then given to the evidence currently available for
10 welfare effects to inform decisions in this review as to whether, and if so how, to establish
11 secondary standards for CO based on public welfare considerations. Initial staff conclusions are
12 based on the assessment and integrative synthesis of the scientific evidence presented in the ISA
13 (USEPA, 2010), building on the evidence described in the 2000 AQCD (USEPA, 2000).
14 3.1 CONSIDERATION IN PREVIOUS REVIEWS
15 With the establishment of the first NAAQS for CO in 1971, secondary standards were set
16 identical to the primary standards. CO was not shown to produce detrimental effects on certain
17 higher plants at levels below 100 ppm. The only significant welfare effect identified for CO
18 levels possibly approaching those in ambient air was inhibition of nitrogen fixation by
19 microorganisms in the root nodules of legumes associated with CO levels of 100 ppm for one
20 month (National Air Pollution Control Administration, 1970). In the first review of the CO
21 NAAQS, the threshold level for plant effects was recognized to occur well above ambient CO
22 levels, such that vegetation damage as a result of CO in ambient air was concluded to be very
23 unlikely (50 FR 37494). As a result, EPA concluded that the evidence did not support
24 maintaining a secondary standard for CO, as welfare-related effects had not been documented to
25 occur at ambient concentrations (50 FR 37494). Bsed on that conclusion, EPA rescinded the
26 secondary standard. In the most recent review of CO, which was completed in 1994, EPA again
27 concluded there was insufficient evidence of welfare effects occurring at or near ambient levels
28 to support setting a secondary NAAQS (59 FR 38906). That review did not consider climate-
29 related effects.
30 3.2 CONSIDERATION OF EVIDENCE AVAILABLE IN THE CURRENT
31 REVIEW
32 To evaluate whether consideration of establishment of a secondary standard for CO is
33 appropriate, we adopted an approach in this review that builds upon the general approach used in
March 2010 3-1 Draft-Do Not Quote or Cite
-------
1 the last review and reflects the broader body of evidence and information now available. In
2 developing conclusions in this review below, staff has taken into account the following
3 overarching question:
4 • Does the currently available scientific information provide support for
5 considering the establishment of a secondary standard for CO?
6 In considering this overarching question, we first note that the extensive literature search
7 performed for the current review did not identify any evidence of ecological effects of CO
8 unrelated to climate-related effects, at or near ambient levels (ISA, section 1.3 p. 1-3). However,
9 ambient CO has been associated with welfare effects related to climate (ISA, section 3.3).
10 Climate-related effects of CO were considered for the first time in the 2000 AQCD. The greater
11 focus on climate in the current ISA relative to the 2000 AQCD reflects comments from CASAC
12 and increased attention to the role of CO in climate forcing (Brain and Samet, 2009). Based on
13 the current evidence, the ISA has concluded that "a causal relationship exists between current
14 atmospheric concentrations of CO and effects on climate" (ISA, section 2.2). Accordingly, the
15 following discussion focuses on climate-related effects of CO in addressing the question posed
16 above.
17 Recently available information does not alter the current well-established understanding
18 of the role of urban and regional CO in continental and global-scale chemistry, as outlined in the
19 2000 AQCD. CO is a weak direct contributor to greenhouse warming. The most significant
20 effects on climate result indirectly from CO chemistry, related to the role of CO as a major
21 atmospheric sink for hydroxyl groups (OH). Increased concentrations of CO can lead to
22 increased concentrations of other gases whose loss processes also involve OH chemistry. Some
23 of these gases, such as methane (CH4) and ozone (Os), contribute to the greenhouse effect
24 directly while others deplete stratospheric Os (ISA, section 3.3 p. 3-11).
25 Advances in modeling and measurement have improved our understanding of the relative
26 contribution of CO to climate forcing. This link between ambient CO and climate is based on
27 both direct radiative forcing (RF) of CO estimated at 0.024 W/m2 by Sinha and Toumi (1996)
28 and indirect effects of CO on climate through CH4, Os and carbon dioxide (CO2) (Forster et al.
29 2007). The combined RF for these indirect effects of CO was estimated by the
30 Intergovernmental Panel on Climate Change (IPCC) to be -0.2 W/m2 over the period 1750-2005
31 (Forster et al., 2007) with more than one-half of the forcing attributed to O3 formation.
32 Ambient CO, by itself, does not significantly absorb the Earth's longwave radiation and
33 warm the atmosphere and surface (ISA, section 2.2). Rather, the effects of CO emissions on
34 climate are felt indirectly because emissions of CO lead to increased ambient concentrations of
35 two important greenhouse gases, CH4 and O3. This increase results in contributions to RF that
36 are significant (Forster et al. 2007). However, it is highly problematic to evaluate the indirect
March 2010 3-2 Draft-Do Not Quote or Cite
-------
1 effects of CO on climate due to the spatial heterogeneity in its emissions and concentrations and
2 due to the localized chemical interdependences involving CO, CFL;, and Os, which also depend
3 on levels of NOx. (ISA section 3.3).
4 Most climate model simulations are based on global scale scenarios and have a high
5 degree of uncertainty associated with short-lived climate forcers such as CO (ISA, section 3.3. p.
6 3-16). These models may fail to consider the local variations in climate forcing due to emissions
7 sources and local meteorological patterns (ISA, section 3.3 p. 3-16).
8 With regard to time period, ambient CO emissions vary on an hourly basis and exhibit
9 seasonal and weekday/weekend fluctuations (ISA, section 3.5). This episodic emission pattern,
10 coupled with the relatively short atmospheric lifetime of CO, has limited the ability to
11 characterize welfare-related exposures to this pollutant. CO is classified as a short-lived climate
12 forcing agent, prompting CO emission reductions to be considered as a possible strategy to
13 mitigate effects of global warming. It is possible to compute individual contributions to RF of
14 CO from separate emissions sectors, although uncertainty in these estimates has not been
15 quantified (ISA, section 3.3 p. 3-13).
16 Additional sources of uncertainty in the current assessment include the chemical
17 interdependences of CO and other climate forcing pollutants and the complexity of CO
18 atmospheric chemistry (ISA, section 3.2.2). Measurement of and techniques for assessing
19 climate forcing are improving, but estimates of RF still have -50% uncertainty (ISA, section 3.3,
20 p. 3-13). Large regional variations in CO concentrations also contribute to the uncertainties in
21 the RF from CO and other trace gases (ISA, section 2.2 p. 3-12).
22 3.3 STAFF CONCLUSIONS FOR THE CURRENT REVIEW
23 In considering whether the currently available scientific information indicates the need
24 for a secondary standard for CO, we have reached the following initial conclusions:
25
26 (1) With respect to non-climate welfare effects, including ecological effects and impacts
27 to vegetation, there is no currently available scientific information that supports a CO
28 secondary standard.
29
30 (2) With respect to climate-related effects, we note that there is evidence of climate
31 forcing effects associated with CO (ISA, sections 2.2 and 3.3). There are weak direct,
32 and stronger, but primarily highly variable, indirect continental and regional climate
33 forcing effects from CO. However, the available information provides no basis for
34 estimating how localized changes in the temporal, spatial and composition patterns of
35 ambient CO likely to occur across the US with (or without) a secondary standard
March 2010 3-3 Draft-Do Not Quote or Cite
-------
1 would affect local, regional, or nationwide changes in climate. For these reasons, it is
2 currently not feasible to conduct an analysis for the purpose of considering a CO
3 secondary standard based on climate considerations. Moreover, more than half of the
4 indirect forcing effects of CO have been estimated to occur through Os formation, and
5 effects of Os are more appropriately considered in the context of the review of the Os
6 NAAQS, rather than in this CO NAAQS review. Based on these considerations, staff
7 initially concludes there is insufficient information at this time to support the
8 consideration of a secondary NAAQS based on CO effects on climate processes.
9
March 2010 3-4 Draft-Do Not Quote or Cite
-------
1 REFERENCES
2
3 Brain J. and Samet J. (2009) Clean Air Scientific Advisory Committee (CASAC) Carbon Monoxide Review Panel's
4 Review on EPA's Integrated Science Assessment for Carbon Monoxide (First External Review Draft)
5 Memorandum to Lisa Jackson, EPA Administrator, from Drs. Joseph Brain and Jonathan Samet. June.
6 Available at:
7 http://vosemite.epa.gov/sab/sabproduct.nsf/264cbl227d55e02c85257402007446a4/90eleb5977789a4f8525
8 746e006275a8!OpenDocument&TableRow=2.3#2.
9 Forster P; Ramaswamy V; Artaxo P; Berntsen T; Betts R; Fahey DW; Haywood J; Lean J; Lowe DC; Myhre G;
10 Nganga J; Prinn R; Raga G; Schultz M; Van Dorland R. (2007) Changes in atmospheric constituents and in
11 radiative forcing, Chapter 2. In Solomon S, Qin D; Manning M; Chen Z; Marquis M; Averyt KB; Tignor
12 M; Miller HL (Ed.),IPCC Fourth Assessment Report (AR4): Climate Change 2007: Working Group I
13 Report: The Physical Science Basis, (pp. 129-234). Cambridge, U.K. and New York, NY:
14 Intergovernmental Panel on Climate Change; Cambridge University Press.
15 National Air Pollution Control Administration. (1970) Air Quality Criteria for Carbon Monoxide. Washington, DC:
16 U.S. Department of Health, Education and Welfare, Public Health Service; report no. NAPCA-PUB-AP-
17 62. Available from: NTIS, Springfield VA; PB-190261
18 Sinha A and Toumi R. (1996) A comparison of climate forcings due to chloroflurocarbons and carbon monoxide.
19 Geophys Res Lett, 23: 65-68.
20 U.S. Environmental Protection Agency. (1991) Air Quality Criteria for Carbon Monoxide, Research Triangle Park,
21 NC, report no.EPA/600/8-90/045F.
22 U.S. Environmental Protection Agency. (2000) Air Quality Criteria for Carbon Monoxide. National Center for
23 Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency,
24 Research Triangle Park, NC, report no. EPA/600/P-99/00IF. June 2000. Available at:
25 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=18163.
26 U.S. Environmental Protection Agency. (2010) Integrated Science Assessment for Carbon Monoxide. U.S.
27 Environmental Protection Agency, Research Triangle Park, NC, report no. EPA/600/R-09/019F. Available
28 at: http://www.epa.gOv/ttn/naaqs/standards/co/s co cr isa.html.
29
30
31
32
33
34
35
36
37
March 2010 3-5 Draft-Do Not Quote or Cite
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-452/P-10-005
Environmental Protection Health and Environmental Impacts Division March 2010
Agency Research Triangle Park, NC
------- |