United States         Office of Research     EPA 600/P-99/001F
Environmental Protection   and Development      June 2000
Agency             Washington, DC 20460
Air Quality Criteria for
Carbon Monoxide

-------
                                EPA 600/P-99/001F
                                     June 2000
Air Quality  Criteria for
   Carbon Monoxide
  National Center for Environmental Assessment
     Office of Research and Development
     U.S. Environmental Protection Agency
      Research Triangle Park, NC 27711

-------
                                     Disclaimer

     This document has been reviewed in accordance with U.S. Environmental Protection Agency
policy and approved for publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.

-------
                                         Preface

     The U. S. Environmental Protection Agency (EPA) promulgates the National Ambient Air Quality
Standards (NAAQS) on the basis of an  up-to-date compilation of scientific knowledge  about the
relationship between various concentrations of ambient air pollutants and their adverse effects on man
and the environment. These air quality criteria are published in criteria documents. In 1970, the first
air quality criteria document for carbon monoxide (CO) was issued by the National Air Pollution Control
Administration, a predecessor of EPA.  On the basis of scientific information in that document, NAAQS
were promulgated for CO at levels of 9 ppm for an 8-h average and 35 ppm for a 1-h average. Periodic
scientific assessments of the published literature were completed by EPA in 1979 and, again, in 1984.
The last full-scale CO criteria document revision was published in 1991.  Although the air quality
criteria have changed over the past two decades, the NAAQS for CO have remained the same.  This
revised criteria document consolidates and updates the current scientific basis for another reevaluation
of the CO NAAQS in accordance with the provisions identified in Sections 108 and 109 of the Clean
Air Act.
     This document was prepared and reviewed by experts from state and federal government offices,
academia, and industry for use by EPA in support of decision making on potential public health risks
of CO; it describes the nature, sources, distribution, measurement, and concentrations of CO in both the
outdoor (ambient) and indoor environments and evaluates the latest data on the health effects in exposed
human populations. Although not intended to be an exhaustive literature review, this document is
intended to cover all pertinent literature through 1999.
     The National Center for Environmental Assessment—Research Triangle Park, NC, acknowledges
the contributions provided by the authors, contributors, and reviewers and the diligence of its staff and
contractors in the preparation of this document.
                                             in

-------

-------
                              Table of Contents
List of Tables	xi
List of Figures 	xiv
Authors, Contributors, and Reviewers	xix
U.S. Environmental Protection Agency, Science Advisory Board, Clean Air Scientific
    Advisory Committee, Carbon Monoxide Review Panel  	 xxvii
U.S. Environmental Protection Agency Project Team for Development of Air Quality
    Criteria for Carbon Monoxide	xxix

EXECUTIVE SUMMARY                                                      E-l

1. INTRODUCTION                                                          1-1
   1.1   LEGISLATIVE REQUIREMENTS	1-1
   1.2   REGULATORY BACKGROUND  	1-2
   1.3   RATIONALE FOR THE EXISTING CARBON MONOXIDE STANDARDS  .... 1-3
        1.3.1  Carboxyhemoglobin Levels of Concern	1-3
        1.3.2  Relationship Between Carbon Monoxide Exposure and
             Carboxyhemoglobin Levels  	1-4
        1.3.3  Estimating Population Exposure	1-5
        1.3.4  Decision on the Primary Standard  	1-5
   1.4   ISSUES OF CONCERN FOR THE CURRENT CRITERIA DEVELOPMENT  ... 1-5
        1.4.1  Sources and Emissions	1-5
        1.4.2  Atmospheric Chemistry 	1-6
        1.4.3  Global Cycle	1-6
        1.4.4  Measurement Technology	1-6
        1.4.5  Ambient Air Quality	1-6
        1.4.6  Indoor Emissions and Concentrations	1-6
        1.4.7  Exposure Assessment	1-7
        1.4.8  Mechanisms of Action 	1-7
        1.4.9  Health Effects	1-7
        1.4.10 Carbon Monoxide Interaction with Drugs 	1-8
        1.4.11 Subpopulations at Risk	1-8
   1.5   METHODS AND PROCEDURES FOR DOCUMENT PREPARATION	1-9
   1.6   ORGANIZATION AND CONTENT OF THE DOCUMENT	1-10
REFERENCES  	1-10

2. ANALYTICAL METHODS FOR MONITORING  CARBON MONOXIDE          2-1
   2.1   INTRODUCTION	2-1
   2.2   OVERVIEW OF TECHNIQUES FOR MEASUREMENT OF AMBIENT
        CARBON MONOXIDE	2-1
   2.3   GAS STANDARDS FOR CALIBRATION	2-2
        2.3.1  Pre-made Mixtures 	2-3
             2.3.1.1  Standard Reference Materials 	2-3

-------
                               Table of Contents
                                      (cont'd)
                                                                              Page
              2.3.1.2  National Institute of Standards and Technology Traceable
                      Reference Materials	2-4
              2.3.1.3  U.S. Environmental Protection Agency Protocol Gases  	2-5
              2.3.1.4  Dutch Bureau of Standards 	2-5
              2.3.1.5  Commercial Blends	2-5
        2.3.2  Laboratory Blended Mixtures	2-5
        2.3.3  Other Methods  	2-6
        2.3.4  Intercomparisons of Standards 	2-6
        2.3.5  Infrared Absorption	2-7
   2.4  MEASUREMENT IN AMBIENT AIR 	2-7
        2.4.1  Sampling System Components	2-7
        2.4.2  Quality Assurance  Procedures for Sampling  	2-7
        2.4.3  Sampling Schedules  	2-8
        2.4.4  Continuous Analysis	2-9
              2.4.4.1  Nondispersive Infrared Photometry	2-9
              2.4.4.2  Gas Chromatography-Flame lonization	2-10
              2.4.4.3  Mercury Liberation 	2-10
              2.4.4.4  Tunable Diode Laser Spectroscopy	2-11
              2.4.4.5  Resonance Fluorescence  	2-11
        2.4.5  Intercomparisons of Methods	2-11
        2.4.6  Other Methods of Analysis	2-11
   2.5  MEASUREMENT USING PERSONAL AND REMOTE MONITORS  	2-12
        2.5.1  Personal Monitors	2-12
        2.5.2  Remote Monitor 	2-12
   2.6  BIOLOGICAL MONITORING  	2-13
        2.6.1  Carboxyhemoglobin Measurements	2-13
        2.6.2  Breath Carbon Monoxide Measurements	2-14
        2.6.3  Relationships of Breath Carbon Monoxide to Blood
              Carboxyhemoglobin  	2-14
        2.6.4  Summary of the Relationship Between Biological Measurements
              of Carbon Monoxide	2-15
   2.7  SUMMARY 	2-16
REFERENCES  	2-17

3. SOURCES, EMISSIONS, AND CONCENTRATIONS OF CARBON
   MONOXIDE IN AMBIENT AND INDOOR AIR                                  3-1
   3.1  INTRODUCTION	3-1
   3.2  THE GLOBAL CYCLE OF CARBON MONOXIDE	3-1
        3.2.1  Global Background Concentrations of Carbon Monoxide	3-2
        3.2.2  Sources and Global Emissions Estimates of Carbon Monoxide	3-4
        3.2.3  The Atmospheric Chemistry of Carbon Monoxide	3-6
                                        VI

-------
                               Table of Contents
                                      (cont'd)
                                                                              Page
   3.3  NATIONWIDE CARBON MONOXIDE EMISSIONS ESTIMATES 	3-10
   3.4  CARBON MONOXIDE CONCENTRATIONS IN AMBIENT AIR 	3-16
        3.4.1  Nationwide Trends in Ambient Carbon Monoxide Concentrations	3-17
        3.4.2  Circadian Patterns in Carbon Monoxide Concentrations	3-20
        3.4.3  Characterization of the Spatial and Temporal Variability in Carbon
              Monoxide Concentrations in Selected U.S. Cities  	3-22
   3.5  SOURCES, EMISSIONS,  AND CONCENTRATIONS OF CARBON
        MONOXIDE IN INDOOR ENVIRONMENTS	3-39
        3.5.1  Sources and Emissions of Carbon Monoxide in Indoor Environments
              Prior to 1991 	3-39
        3.5.2  Combustion Sources and Estimated Emissions Rates  	3-40
              3.5.2.1  Gas Cooking Ranges, Ovens, and Furnaces	3-40
              3.5.2.2  Emissions from Unvented Space Heaters  	3-43
              3.5.2.3  Woodstoves and Fireplaces	3-44
              3.5.2.4  Environmental Tobacco Smoke	3-44
        3.5.3  Source-Related Concentrations of Carbon Monoxide in Indoor
              Environments Prior to 1991  	3-45
        3.5.4  Indoor Concentrations of Carbon Monoxide  	3-45
              3.5.4.1  Factors Affecting Carbon Monoxide Concentrations 	3-45
              3.5.4.2  Models for Carbon Monoxide Concentrations 	3-47
              3.5.4.3  Microenvironmental Monitoring Studies	3-48
   3.6  SUMMARY 	3-53
REFERENCES  	3-55
Appendix 3 A:  Spatial Correlation Coefficients for Carbon Monoxide 	  3A-1

4. POPULATION EXPOSURE TO CARBON MONOXIDE                         4-1
   4.1  INTRODUCTION	4-1
   4.2  BRIEF SUMMARY OF POPULATION EXPOSURE STUDIES PRIOR
        TO 1991  	4-2
        4.2.1  Sensitive Populations 	4-3
        4.2.2  Estimates of Population Exposure Based on Fixed-Site Monitors	4-3
        4.2.3  Surveys of Population Exposure Using Personal Monitors 	4-4
   4.3  POPULATION EXPOSURE MODELS	4-5
   4.4  SURVEY OF RECENT EXPOSURE STUDIES OF NONSMOKERS 	4-9
        4.4.1  Nonoccupational Exposures	4-9
              4.4.1.1  Exposure to Carbon Monoxide from Motor Vehicles	4-9
              4.4.1.2  Exposure to Carbon Monoxide in Recreational Vehicles 	4-13
              4.4.1.3  Residential Exposure to Carbon Monoxide	4-14
              4.4.1.4  Exposure to Carbon Monoxide at Commercial Facilities 	4-16
              4.4.1.5  Studies of Breath Carbon Monoxide in Populations:
                      The Effects of Exposure to Carbon Monoxide 	4-18
                                        VII

-------
                                Table of Contents
                                       (cont'd)
                                                                               Page
              4.4.1.6  Nonoccupational Exposure to Methylene Chloride	4-20
              4.4.1.7  Exposures to Carbon Monoxide from Passive Smoking	4-21
        4.4.2  Occupational Exposures	4-21
              4.4.2.1  Exposures to Carbon Monoxide in the Workplace  	4-21
              4.4.2.2  Exposures to Methylene Chloride in the Workplace	4-23
        4.4.3  Activity Pattern Studies  	4-24
              4.4.3.1  Activity Patterns of California Residents	4-24
              4.4.3.2  Activity Patterns of Children in Six States  	4-24
              4.4.3.3  A Comparative  Study Between California and the Nation  	4-26
              4.4.3.4  An English Study	4-27
              4.4.3.5  A Boston Study of Household Activities, Life Cycle, and
                      Role Allocation  	4-27
              4.43.6  The National Human Activity Pattern Survey  	4-27
   4.5  MAJOR FACTORS AFFECTING POPULATION EXPOSURE	4-28
        4.5.1  Federal Policies Affecting Transportation and Air Quality
              in Urban Areas  	4-28
        4.5.2  Federal and State Policies Affecting Temporal Trends in Exposure  	4-29
              4.5.2.1  Effects of Motor Vehicle Emission Standards on
                      Unintentional Death Rates	4-29
              4.5.2.2  Effects of Motor Vehicle Emission Standards on Passenger
                      Cabin Exposure  	4-31
        4.5.3  Social Changes Affecting Human Activity Patterns	4-33
   4.6  CONCLUSIONS	4-34
REFERENCES  	4-36

5. PHARMACOKINETICS AND MECHANISMS OF ACTION OF
   CARBON MONOXIDE                                                        5-1
   5.1  INTRODUCTION	5-1
   5.2  ABSORPTION, DISTRIBUTION, AND PULMONARY ELIMINATION  	5-2
        5.2.1  Pulmonary Uptake	5-2
              5.2.1.1  Mass Transfer of Carbon Monoxide 	5-2
              5.2.1.2  Effects of Dead Space and Ventilation/Perfusion Ratio  	5-2
              5.2.1.3  Lung Diffusion  of Carbon Monoxide  	5-3
        5.2.2  Tissue Uptake	5-4
              5.2.2.1  The Lung 	5-4
              5.2.2.2  The Blood	5-4
              5.2.2.3  Heart and Skeletal Muscle	5-5
              5.2.2.4  The Brain and Other Tissues	5-6
        5.2.3  Pulmonary and Tissue Elimination	5-6
   5.3  TISSUE PRODUCTION AND METABOLISM OF CARBON MONOXIDE	5-7
                                         VIII

-------
                               Table of Contents
                                      (cont'd)
                                                                             Page
   5.4  CONDITIONS AFFECTING CARBON MONOXIDE UPTAKE AND
        ELIMINATION  	5-8
        5.4.1  Environment and Activity	5-8
        5.4.2  Altitude	5-9
        5.4.3  Physical Characteristics  	5-10
        5.4.4  Health Status	5-11
   5.5  MODELING CARBOXYHEMOGLOBIN FORMATION	5-12
        5.5.1  The Coburn-Forster-Kane and Other Regression Models  	5-12
              5.5.1.1  Empirical Regression Models 	5-12
              5.5.1.2  Linear and Nonlinear Coburn-Forster-Kane Differential
                      Equations	5-12
              5.5.1.3  Confirmation Studies of Coburn-Forster-Kane Models	5-13
              5.5.1.4  Application of Coburn-Forster-Kane Models 	5-15
   5.6  INTRACELLULAR EFFECTS OF CARBON MONOXIDE	5-16
        5.6.1  Introduction	5-16
        5.6.2  Inhibition of Hemoprotein Function	5-16
        5.6.3  Free Radical Production  	5-17
        5.6.4  Stimulation of Guanylate Cyclase	5-18
   5.7  MECHANISMS OF CARBON MONOXIDE TOXICITY 	5-19
        5.7.1  Alterations in Blood Flow	5-19
        5.7.2  Mitochondrial Dysfunction and Altered Production of High-Energy
              Intermediates	5-19
        5.7.3  Vascular Insults Associated with Exposure to Carbon Monoxide	5-20
   5.8  OTHER EFFECTS OF CARBON MONOXIDE	5-21
   5.9  SUMMARY 	5-21
REFERENCES  	5-22

6. HEALTH EFFECTS OF EXPOSURE TO AMBIENT CARBON MONOXIDE       6-1
   6.1  INTRODUCTION	6-1
   6.2  CARDIOVASCULAR EFFECTS  	6-2
        6.2.1  Epidemiologic Studies of Cardiovascular and Other Disorders	6-3
              6.2.1.1  Introduction  	6-3
              6.2.1.2  Ambient Carbon Monoxide and Exacerbation of
                      Heart Disease	6-7
              6.2.1.3  Ambient Carbon Monoxide and Daily Mortality Counts  	6-21
              6.2.1.4  Ambient Carbon Monoxide and Frequency of
                      Respiratory Illness	6-28
              6.2.1.5  Ambient Carbon Monoxide and Low Birth Weight  	6-33
        6.2.2  Controlled Laboratory Studies  	6-38
   6.3  CENTRAL NERVOUS SYSTEM AND BEHAVIORAL EFFECTS	6-40
        6.3.1  Brain Oxygen Metabolism  	6-40
                                        IX

-------
                             Table of Contents
                                    (cont'd)
                                                                         Page
             6.3.1.1   Whole Brain	6-40
             6.3.1.2   Subregions of the Brain	6-41
       6.3.2  Behavioral Effects of Carbon Monoxide 	6-42
   6.4  DEVELOPMENTAL TOXICITY  	6-44
   6.5  ACUTE PULMONARY EFFECTS	6-45
   6.6  OTHER SYSTEMIC EFFECTS OF CARBON MONOXIDE 	6-45
   6.7  PHYSIOLOGIC RESPONSES TO CARBON MONOXIDE EXPOSURE	6-46
   6.8  COMBINED EXPOSURE OF CARBON MONOXIDE WITH OTHER
       POLLUTANTS, DRUGS, AND ENVIRONMENTAL FACTORS 	6-47
       6.8.1  High-Altitude Effects  	6-47
       6.8.2  Interaction with Drugs  	6-47
       6.8.3  Interaction with Other Air Pollutants and Environmental Factors	6-48
       6.8.4  Tobacco Smoke	6-49
   6.9  SUMMARY 	6-50
REFERENCES  	6-52

7.  INTEGRATIVE SUMMARY AND CONCLUSIONS                            7-1
   7.1  INTRODUCTION	7-1
   7.2  ENVIRONMENTAL SOURCES	7-2
   7.3  ENVIRONMENTAL CONCENTRATIONS	7-2
   7.4  CARBOXYHEMOGLOBIN LEVELS IN THE POPULATION	7-3
   7.5  MECHANISMS OF CARBON MONOXIDE ACTIVITY 	7-4
   7.6  HEALTH EFFECTS OF CARBON MONOXIDE	7-5
   7.7  SUBPOPULATIONS POTENTIALLY AT RISK FROM EXPOSURE
       TO AMBIENT CARBON MONOXIDE  	7-6
       7.7.1  Age, Gender, and Pregnancy as Risk Factors	7-6
       7.7.2  Preexisting Disease as a Risk Factor	7-7
             7.7.2.1   Subjects with Heart Disease	7-7
             7.7.2.2   Subjects with Other Vascular Diseases	7-8
             7.7.2.3   Subjects with Anemia and Other Hematologic Disorders	7-8
             7.7.2.4   Subjects with Obstructive Lung Disease	7-9
       7.7.3  Subpopulations at Risk from Combined Exposure to Carbon Monoxide
             and Other Chemical Substances	7-9
             7.7.3.1   Interactions with Drugs  	7-9
             7.7.3.2   Interactions with Other Chemical Substances in the
                     Environment	7-9
       7.7.4  Subpopulations Exposed to Carbon Monoxide at High Altitudes 	7-9
   7.8  CONCLUSIONS	7-10
REFERENCES  	7-10

APPENDIX A:  Abbreviations and Acronyms                                 A-l

-------
                                   List of Tables
Number                                                                         Page
1-1     National Ambient Air Quality Standards for Carbon Monoxide 	1-2
2-1     Performance Specifications for Automated Analytical Methods for
        Carbon Monoxide  	2-3
2-2     Suggested Performance Specifications for Monitoring Carbon Monoxide
        in Nonurban Environments	2-4
3-1     Summary of Major Sources and Sinks of Carbon Monoxide	3-2
3-2     Annual Global Carbon Monoxide Emissions Estimates	3-5
3-3     U.S. Carbon Monoxide Emissions 	3-11
3-4     Sites Not Meeting the 8-Hour Carbon Monoxide National Ambient Air Quality
        Standard, 1993 to 1997	3-18
3-5     Running-Average Exceedances of the 9-ppm 8-Hour Carbon Monoxide
        Standard, 1988, 1996, and 1997	3-21
3-6     Annual Orcadian Pattern of 8-Hour Average Carbon Monoxide Concentrations
        Culminating in Values Greater Than 9.5 ppm in Lynwood and Hawthorne, CA,
        During 1996  	3-22
3-7     Ranges in Average Carbon Monoxide Emission Rates for Residential Sources ... 3-41
3-8     Sources of Carbon Monoxide in the Indoor Environment 	3-42
3-9     Combustible Fuels in Homes in the United States in 1995  	3-47
3-10    Carbon Monoxide Descriptive Statistics for All Homes   	3-49
3-11    Carbon Monoxide Concentrations in Smoking and Nonsmoking Areas
        in Real-Life Situations 	3-52
3A-1    Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average
        Carbon Monoxide Data in the Denver Metropolitan Statistical Area	  3 A-2
3 A-2    Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average
        Carbon Monoxide Data in the Los Angeles Consolidated Metropolitan
        Statistical Area 	  3A-4
3 A-3    Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average
        Carbon Monoxide Data in the New York City Consolidated Metropolitan
        Statistical Area 	  3A-6
3 A-4    Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average
        Carbon Monoxide Data in the Phoenix Metropolitan Statistical Area	  3 A-8
3 A-5    Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average
        Carbon Monoxide Data in the Fairbanks Metropolitan Statistical Area	  3 A-9

                                         xi

-------
                                    List of Tables
                                        (cont'd)

Number                                                                           Page

4-1     Carbon Monoxide Concentrations in Selected Microenvironments of
        Denver, CO, 1982 and 1983 	4-6

4-2     Summary of Studies of In-Vehicle Exposure and Ambient Carbon Monoxide
        Concentrations, 1965 to 1992	4-11

4-3     Mean Breath Carbon Monoxide Levels and Sample Sizes Across Smoking
        Categories and Job Types	4-19

4-4     Studies of Occupational Exposures and Dosages	4-22

4-5     Time Spent in Different Microenvironments by Californians, 1987 to 1990	4-25

4-6     Percentage of Californians Who Use or Who Are in Proximity to Potential
        Sources of Either Carbon Monoxide or Methylene Chloride on a Given Day,
        1987 to 1990	4-26

4-7     Motor Vehicle Carbon Monoxide Emission Standards, Typical In-Vehicle
        Carbon Monoxide Exposures,  and Unintentional Carbon Monoxide-Related
        Death Rates in the United States	4-30

4-8     Typical Net Mean Carbon Monoxide Concentration Ranges by Travel Mode
        for Cities in Three Countries	4-32

6-1     Modeled Relative Risks of Interquartile Range Increases in Ambient Pollutant
        Concentrations for Daily Heart Disease Admissions in Persons >65 Years Old,
        Detroit, MI,  1986 to 1989	6-9

6-2     Modeled Effects of 5th- to 95th-Percentile Increments in Ambient Air Pollutant
        Concentrations on Daily Numbers of Cardiac and Respiratory Hospital
        Admissions in Single-Pollutant Models, by Season, Athens, Greece, 1998	6-11

6-3     Canadian City-Specific Relative Risks of a Change in Daily Maximum
        1-Hour Carbon Monoxide Levels from 1 ppm to 3 ppm for Congestive Heart
        Failure in the Elderly, Based on Random Effects Models and a Fixed Effect
        Analysis of Each City Separately, for Selected Model Specifications	6-13

6-4     Acute Myocardial Infarction:  One- and Two-Pollutant Models with Cool
        Season, London, UK, April  1,  1987, to March 31, 1994  	6-14

6-5     Modeled Percentage Increases in Hospital Admissions for Heart Disease,
        Associated with Interquartile Range Increases in Ambient PM10 and Carbon
        Monoxide, in Eight Locations, Across  These Locations, and in
        Tucson, AZ, 1988 to 1990   	6-18

6-6     Modeled Percentage Increases in Hospitalizations at Mean Pollutant
        Concentrations in Single- and  Multi-Pollutant Models,
        Toronto, Canada, 1980 to 1994  	6-20

                                          xii

-------
                                   List of Tables
                                       (cont'd)

Number                                                                          Page

6-7     Summary of Time Series Studies of Ambient Carbon Monoxide and
        Daily Frequency of Heart Disease Exacerbation 	6-22

6-8     Summary of Time Series Studies of Ambient Carbon Monoxide and
        Daily Mortality Counts	6-29

6-9     Summary of Time Series Studies of Ambient Carbon Monoxide and
        Daily Frequency of Respiratory Illness	6-34

6-10    Summary of Epidemiologic Studies of Ambient Carbon Monoxide and
        Low Birth Weight	6-37

6-11    Estimated Lowest-Observed-Effect Levels for Cardiovascular Effects of
        Exposure of Laboratory Animals to Carbon Monoxide	6-40

6-12    Key Health Effects of Carbon Monoxide Demonstrated by Controlled-Exposure
        Studies	6-50

7-1     Predicted Carbon Monoxide Exposures in the Population	7-4
                                         XIII

-------
                                   List of Figures

Number

2-1     Schematic diagram of gas filter correlation monitor for carbon monoxide 	2-9

2-2     The correlation between an end-tidal breath carbon monoxide concentration
        after a 10-second breathhold and blood carboxyhemoglobin levels expressed
        as individual data points, as well as mean, plus or minus standard deviation 	2-15

3-1     Latitudinal and seasonal variability in carbon monoxide concentrations
        obtained by the National Oceanic and Atmospheric Administration Climate
        Monitoring Diagnostics Laboratory monitoring network	3-3

3-2     Global background average carbon monoxide concentrations and growth
        rates for global background average carbon monoxide 	3-3

3-3     Locations of sites in the nationwide ambient carbon monoxide monitoring
        network, 1997	3-17

3-4     Nationwide composite average of the annual  second-highest 8-hour carbon
        monoxide concentrations, 1978 to 1997	3-20

3-5     Variability in the annual second-highest 8-hour carbon monoxide
        concentrations across all sites in the United States reporting at least
        8 years of data, 1988 to 1997  	3-20

3-6     Composite average of the annual second-highest 8-hour carbon monoxide
        concentrations for rural, suburban, and urban sites, 1988 to 1997	3-20

3-7     Diurnal variation of nationwide composite hourly average carbon monoxide
        concentrations for winter, 1987 to 1996	3-21

3-8     Map of Denver showing locations of carbon monoxide monitoring sites  	3-24

3-9     Average diurnal variation in carbon monoxide at the Denver-Broadway site
        for weekdays during the winter season	3-25

3-10    Monthly average diurnal variation in carbon monoxide at the
        Denver-Broadway site for weekdays from May 1986 through May 1987  	3-25

3-11    Monthly average diurnal variation in carbon monoxide at the
        Denver-Broadway site for weekdays from May 1995 through May 1996  	3-26

3-12    Central tendency statistics for the daily 8-hour maximum carbon monoxide
        concentration at the Denver-Broadway site during the winter season from
        1986 to 1995	3-26

3-13    Map of Los Angeles showing locations of carbon monoxide monitoring  sites  .... 3-27

3-14    Average diurnal variation in carbon monoxide at the Los Angeles-Lynwood
        site for weekdays during the winter season	3-28
                                          XIV

-------
                                   List of Figures
                                       (cont'd)

Number                                                                          Page

3-15    Central tendency statistics for the daily 8-hour maximum carbon monoxide
        concentration at the Los Angeles-Hawthorne site during the winter season
        from 1986 to 1995	3-28

3-16    Central tendency statistics for the daily 8-hour maximum carbon monoxide
        concentration at the Los Angeles-Barstow site during the winter season from
        1986 to 1995	3-28

3-17    Average diurnal variation in carbon monoxide at the Los Angeles-Hawthorne
        site for weekdays during the winter season	3-29

3-18    Average diurnal variation in carbon monoxide at the Los Angeles-El Toro
        site for weekdays during the winter season	3-29

3-19    Monthly average diurnal variation in carbon monoxide at the
        Los Angeles-Anaheim site for weekdays from May through May 1986 to
        1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996  	3-30

3-20    Map of New York showing locations of carbon monoxide monitoring sites	3-31

3-21    Monthly average diurnal variation in carbon monoxide at the
        New York-Flatbush site for weekdays from May through May 1986 to
        1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996  	3-32

3-22    Average diurnal variation in carbon monoxide at the New York-Manhattan
        site for weekdays during the winter season	3-32

3-23    Monthly average diurnal variation in carbon monoxide at the
        New York-Manhattan site for weekdays from May through May 1986 to
        1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996  	3-33

3-24    Monthly average diurnal variation in carbon monoxide at the
        New York-Morristown,  NJ, site for weekdays from May through May
        1986 to 1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996	3-33

3-25    Map of Phoenix showing locations of carbon monoxide monitoring sites	3-34

3-26    Monthly average diurnal variation in carbon monoxide at the
        Phoenix-Central site for weekdays from May through May 1986 to
        1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996  	3-35

3-27    Central tendency statistics for the daily 8-hour maximum carbon monoxide
        concentration at the Phoenix-East Butler site during the winter season from
        1986 to 1995	3-35

3-28    Monthly average diurnal variation in carbon monoxide at the Phoenix-West
        site for weekdays from May through May  1986 to 1987, 1989 to 1990,
        1992 to 1993, and 1995  to 1996	3-36

                                          xv

-------
                                   List of Figures
                                        (cont'd)

Number                                                                          Page

3-29    Monthly average diurnal variation in carbon monoxide at the Phoenix-South
        site for weekdays from May through May 1986 to 1987, 1989 to 1990,
        1992 to 1993, and 1995 to 1996	3-36

3-30    Map of Fairbanks showing locations of carbon monoxide monitoring sites  	3-37

3-31    Average diurnal variation in carbon monoxide at the Fairbanks-Federal
        Building site for weekdays during the winter season	3-38

3-32    Central tendency statistics for the daily 8-hour maximum carbon monoxide
        concentration at the Fairbanks-Federal Building site during the winter season
        from 1986 to 1995	3-38

3-33    Monthly average diurnal variation in carbon monoxide at the Fairbanks-
        Federal Building site for weekdays from October through March 1986 to
        1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996  	3-39

3-34    Percentage of U.S. households using unvented combustion heaters, by type
        of fuel, stratified by region 	3-46

3-35    Modeled indoor carbon monoxide concentration distributions in houses
        with only one indoor combustion pollutant source	3-48

3-36    Arithmetic mean carbon monoxide concentrations by presence or absence
        of combustion source	3-50

4-1     Conceptual overview of the probabilistic National Ambient Air Quality
        Standards Exposure Model	4-7

4-2     Observed versus simulated 8-hour daily maximum exposure for persons
        residing in homes with gas stoves in Denver, CO  	4-8

4-3     Observed versus simulated 8-hour daily maximum exposure for persons
        residing in homes without gas stoves in Denver, CO	4-8

4-4     Log-probability plot of the maximum 1-hour indoor minus outdoor carbon
        monoxide concentrations based on a random sample of 277 homes that
        used gas appliances in  California,  1991 and  1992  	4-15

4-5     Log-probability plot of the maximum 8-hour indoor minus outdoor carbon
        monoxide concentrations based on a random sample of 277 homes that
        used gas appliances in  California,  1991 and  1992  	4-15

4-6     Excess carbon monoxide concentrations in the exhaled air of nonsmoking
        control subjects, untreated asthmatics, and treated asthmatics	4-20

4-7     Trends in ambient carbon monoxide concentrations and in-vehicle carbon
        monoxide exposures, 1965 to 1992  	4-31
                                          XVI

-------
                                   List of Figures
                                        (cont'd)

Number                                                                           Page

5-1     Diagrammatic presentation of carbon monoxide uptake and elimination
        pathways and carbon monoxide body stores	5-2
5-2     Oxyhemoglobin dissociation curve of normal human blood, of blood
        containing 50% carboxyhemoglobin, and of blood with only 50%
        hemoglobin because of anemia	5-4
5-3     Plot of fractional sensitivities of selected variables versus time of exposure	5-14
6-1     Nonparametric smoothing of the association between ambient levels of carbon
        monoxide and hospital admissions for congestive heart failure among elderly
        people after adjustment for temperature, month, day of week, and year,
        1986 through 1989  	6-9

6-2     The relative risk associated with the exposure percentiles of carbon monoxide
        at specific temperature strata, based on results of the multi-pollutant and
        single-pollutant generalized linear models of hospital admissions for heart
        failure among the elderly in Chicago, IL, 1986 to 1989	6-16

6-3     The effect of carbon monoxide exposure on time to onset of angina	6-38
6-4     The relationship between carboxyhemoglobin and the cerebral metabolic
        rate for oxygen for goats and sheep  	6-41
6-5     The relationship between carboxyhemoglobin and behavior: effects in rats	6-43
6-6     The relationship between carboxyhemoglobin and behavior: effects in
        humans compared to rats 	6-43
7-1     Predicted carboxyhemoglobin levels resulting from 1- and 8-hour exposures
        to carbon monoxide at rest and with light exercise are based on the
        Coburn-Forster-Kane equation	7-3
7-2     Percentage breakdown of deaths from cardiovascular diseases in the
        United States	7-7

7-3     Estimated prevalence of cardiovascular disease by age and sex for the
        United States, 1988 to 1991  	7-8
                                          XVII

-------

-------
                     Authors, Contributors, and Reviewers

                              Chapter 1.  Introduction
Principal Author
Mr. James A. Raub—Project Manager and Coordinator for Health Effects
National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Contributors
Dr. Robert S. Chapman—Coordinator for Epidemiology Studies
National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Beverly M. Comfort—Coordinator for Indoor Air Emissions and Concentrations
National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. William G. Ewald—Coordinator for Measurement Methods
National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. David T. Mage—Coordinator for Population Exposure
National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph P. Pinto—Coordinator for Atmospheric Chemistry, Sources, and Emissions
National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Reviewers
Mr. Claire L. Barnett—New York Healthy Schools Network, 96 South Swan Street
Albany, NY 12210
Mr. Kelly M. Brown—Ford Motor Company, 17225 Federal Drive, Suite 145
Allen Park, MI 48101

Mr. Albert Donnay—MCS Referral & Resources, Inc., 508 Westgate Rd., Baltimore, MD 21229
Dr. Sandra E. Inkster—U.S. Consumer Product Safety Commission, 4330 East West Highway
Room 600, Bethesda, MD 02814

Dr. David J. McKee—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Nancy W. Newkirk—American Petroleum Institute, 1220 L Street, NW
Washington, DC 20005

Mr. Harvey M. Richmond—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

                                         xix

-------
                     Authors, Contributors, and Reviewers
                                       (cont'd)

           Chapter 2. Analytical Methods for Monitoring Carbon Monoxide

Principal Authors
Dr. Russell R. Dickerson—Department of Meteorology, The University of Maryland
College Park, MD 20742

Dr. David T. Mage—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Contributors
Mr. William G. Ewald—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph P. Pinto—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lance Wallace—National Exposure Research Laboratory
U.S. Environmental Protection Agency, Reston, VA 22092

Reviewers
Dr. Michael G. Apte—Indoor Environment Department
Lawrence Berkeley National Laboratory, Berkeley, CA 94720
Mr. Kelly M. Brown—Ford Motor Company, 17225 Federal Drive, Suite 145
Allen Park, MI 48101

Dr. Steven D. Colome—Integrated Environmental Services, Irvine, CA 92612-2935

Dr. Thomas E. Dahms—Department of Anesthesiology, School of Medicine
St. Louis University Medical  Center, St. Louis, MO  63110
Mr. Albert Donnay—MCS Referral & Resources, Inc., 508 Westgate Rd., Baltimore, MD 21229

Dr. Milan J. Hazucha—Department of Medicine
Center for Environmental Medicine and Lung Biology, The University of North Carolina
Chapel Hill, NC 27599
Dr. Michael T. Kleinman—Department of Community and Environmental Medicine
California College of Medicine, University of California, Irvine, CA 92697

Mr. Leon Langan—Langan Products, Inc., 2660 California Street, San Francisco, CA 94115
Dr. William A. McClenny—National Exposure Research Laboratory (MD-44)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Nancy W. Newkirk—American Petroleum Institute, 1220 L Street, NW
Washington, DC  20005
                                         xx

-------
                     Authors, Contributors, and Reviewers
                                       (cont'd)

Dr. Leonard Newman—Environmental Chemistry Division
Brookhaven National Laboratory, Upton, NY 11973
Dr. Paul Roberts—Sonoma Technology, Inc., Petaluma, CA  94954
               Chapter 3. Sources, Emissions, and Concentrations of
                    Carbon Monoxide in Ambient and Indoor Air

Principal Authors
Dr. Joseph P. Pinto—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711
Ms. Beverly M. Comfort—National Center for Environmental Assessment  (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Michael P. Zelenka—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Contributors
Mr. Warren P. Freas—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711
Dr. Alan H. Huber—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Reviewers
Dr. Michael G. Apte—Indoor Environment Department
Lawrence Berkeley National Laboratory, Berkeley, CA 94720
Dr. Irwin H. Billick—WEC Consulting, Ltd., Potomac, MD 20854
Mr. Kelly M. Brown—Ford Motor Company, 17225 Federal Drive, Suite 145
Allen Park, MI 48101

Mr. Tom Chappie—State of Alaska, Department of Environmental Conservation
555 Cordova Street, Anchorage, AK 99501
Dr. Steven D. Colome—Integrated Environmental Services, Irvine, CA 92612-2935

Mr. Albert Donnay—MCS Referral & Resources, Inc., 508 Westgate Rd., Baltimore, MD  21229
Dr. Peter G. Flachsbart—Department of Urban and Regional Planning
University of Hawaii atManoa, Honolulu, HI 96822

Dr. Lawrence J. Folinsbee—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711
                                        XXI

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)

Ms. Mary Good—Municipality of Anchorage, Department of Health and Human Services
825 L Street, P.O. Box 196650, Anchorage, AK 99519
Mr. Hank Hove—Fairbanks North Star Borough, 809 Pioneer Road, P.O. Box 71267
Fairbanks, AK 99707
Dr. Sandra E. Inkster—U.S. Consumer Product Safety Commission, 4330 East West Highway
Room 600, Bethesda, MD  02814

Dr. Kai-Shen Liu—Environmental Health Laboratory
California Department of Health Services, Berkeley, CA 94704
Mr. Thomas R. McCurdy—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Robert Morris—Department of Family Medicine
Tufts University School of Medicine, Boston,  MA  02111
Ms. Nancy W. Newkirk—American Petroleum Institute, 1220 L Street, NW
Washington,  DC  20005
Dr. Leonard Newman—Environmental Chemistry Division
Brookhaven National Laboratory, Upton, NY  11973

Dr. Paul Roberts—Sonoma Technology, Inc.,  Petaluma, CA 94954

Dr. Jed Waldman—California Department of Health Services, Berkeley, CA 94704
                Chapter 4.  Population Exposure to Carbon Monoxide

Principal Author
Dr. Peter G. Flachsbart—Department of Urban and Regional Planning
University of Hawaii atManoa, Honolulu, HI 96822

Contributor
Dr. David T. Mage—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Reviewers
Dr. Michael G. Apte—Indoor Environment Department
Lawrence Berkeley National Laboratory, Berkeley, CA 94720
Mr. C. Barnett—New York Healthy Schools Network, 96 South Swan Street
Albany, NY  12210
Dr. Irwin H. Billick—WEC Consulting, Ltd., Potomac, MD 20854
                                        XXII

-------
                     Authors, Contributors, and Reviewers
                                       (cont'd)

Mr. Kelly M. Brown—Ford Motor Company, 17225 Federal Drive, Suite 145
Allen Park, MI 48101

Dr. Steven D. Colome—Integrated Environmental Services, Irvine, CA 92612-2935
Mr. Albert Donnay—MCS Referral & Resources, Inc., 508 Westgate Rd., Baltimore, MD 21229

Dr. Milan J. Hazucha—Department of Medicine
Center for Environmental Medicine and Lung Biology, The University of North Carolina
Chapel Hill, NC 27599
Dr. Sandra E. Inkster—U.S. Consumer Product Safety Commission, 4330 East West Highway
Room 600, Bethesda, MD 02814

Dr. Michael T. Kleinman—Department of Community and Environmental Medicine
California College of Medicine, University of California, Irvine, CA 92697
Dr. Kai-Shen Liu—Environmental Health Laboratory
California Department of Health Services, Berkeley, CA 94704

Mr. Thomas R. McCurdy—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Robert Morris—Department of Family Medicine
Tufts University School of Medicine, Boston, MA 02111
Ms. Nancy W. Newkirk—American Petroleum Institute, 1220 L Street, NW
Washington, DC  20005
Mr. Harvey M. Richmond—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Jed Waldman—California Department of Health Services, Berkeley, CA  94704
                  Chapter 5.  Pharmacokinetics and Mechanisms of
                            Action of Carbon Monoxide
Principal Authors
Dr. Milan J. Hazucha—Department of Medicine
Center for Environmental Medicine and Lung Biology, The University of North Carolina
Chapel Hill, NC 27599
Dr. Stephen R. Thorn—Institute for Environmental Medicine and
Department of Emergency Medicine, University of Pennsylvania, Philadelphia, PA 19104
                                        XXIII

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
Reviewers
Mr. Claire L. Barnett—New York Healthy Schools Network, 96 South Swan Street
Albany, NY  12210

Mr. Kelly M. Brown—Ford Motor Company, 17225 Federal Drive, Suite 145
Allen Park, MI 48101

Dr. Steven D. Colome—Integrated Environmental Services, Irvine, CA 92612-2935

Dr. Thomas E. Dahms—Department of Anesthesiology
School of Medicine, St. Louis University Medical Center, St. Louis, MO 63110

Mr. Albert Donnay—MCS Referral & Resources, Inc., 508 Westgate Rd., Baltimore, MD 21229

Dr. Sandra E. Inkster—U.S. Consumer Product Safety Commission, 4330 East West Highway
Room 600, Bethesda, MD  02814

Dr. Michael T. Kleinman—Department of Community and Environmental Medicine
California College of Medicine, University of California, Irvine, CA 92697

Dr. James J. McGrath—Department of Physiology, School of Medicine
Texas Tech University Health Sciences Center, Lubbock, TX 79430

Ms. Nancy W. Newkirk—American Petroleum Institute, 1220 L Street, NW
Washington, DC  20005

Dr. Peter Tikuisis—Defence and Civil Institute of Environmental Medicine
1133 Sheppard Avenue, West, Toronto, Ontario, M3M 3B9, Canada
         Chapter 6.  Health Effects of Exposure to Ambient Carbon Monoxide

Principal Authors

Mr. James A. Raub—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Robert S. Chapman—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Vernon Benignus—National Health and Environmental Effects Research Laboratory (MD-58B)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Reviewers

Mr. Claire L. Barnett—New York Healthy Schools Network, 96 South Swan Street
Albany, NY  12210

Mr. Kelly M. Brown—Ford Motor Company, 17225 Federal Drive, Suite 145
Allen Park, MI 48101

                                         xxiv

-------
                      Authors, Contributors, and Reviewers
                                        (cont'd)

Dr. Steven D. Colome—Integrated Environmental Services, Irvine, CA 92612-2935

Dr. Thomas E. Dahms—Department of Anesthesiology, School of Medicine
St. Louis University Medical Center, St. Louis, MO  63110

Mr. Albert Donnay—MCS Referral & Resources, Inc., 508 Westgate Rd., Baltimore, MD 21229

Dr. Lawrence J. Folinsbee—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Milan J. Hazucha—Department of Medicine
Center for Environmental Medicine and Lung Biology, The University of North Carolina
Chapel Hill, NC 27599

Dr. Jon M. Heuss—Air Improvement Resources, Inc., 7355 Rickett Drive
Washington, MI 48094

Dr. Sandra E. Inkster—U.S. Consumer Product Safety Commission, 4330 East West Highway
Room 600, Bethesda, MD  02814

Dr. Michael T. Kleinman—Department of Community and Environmental Medicine
California College of Medicine, University of California, Irvine, CA 92697

Dr. Victor G. Laties—Environmental Medicine, University of Rochester Medical Center
School of Medicine and Dentistry, Rochester, NY 14642

Dr. James J. McGrath—Department of Physiology, School of Medicine
Texas Tech University Health Sciences Center, Lubbock, TX 79430

Dr. Robert Morris—Department of Family Medicine
Tufts University School of Medicine, Boston, MA 02111

Ms. Nancy W. Newkirk—American Petroleum Institute, 1220 L Street, NW
Washington, DC  20005
                   Chapter 7. Integrative Summary and Conclusions
Principal Author
Mr. James A. Raub—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Contributors

Dr. Robert S. Chapman—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Ms. Beverly M. Comfort—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
                                          xxv

-------
                      Authors, Contributors, and Reviewers
                                        (cont'd)

Mr. William G. Ewald—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. David T. Mage—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph P. Pinto—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Reviewers

Dr. Michael G. Apte—Indoor Environment Department
Lawrence Berkeley National Laboratory, Berkeley, CA 94720

Mr. Kelly M. Brown—Ford Motor Company, 17225 Federal Drive, Suite 145
Allen Park, MI 48101

Mr. Albert Donnay—MCS Referral & Resources, Inc., 508 Westgate Rd., Baltimore, MD 21229

Dr. Lawrence J. Folinsbee—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Sandra E. Inkster—U.S. Consumer Product Safety Commission, 4330 East West Highway
Room 600, Bethesda, MD  02814

Dr. James J. McGrath—Department of Physiology, School of Medicine
Texas Tech University Health Sciences  Center, Lubbock, TX 79430

Ms. Nancy W. Newkirk—American Petroleum Institute, 1220 L Street, NW
Washington, DC  20005

Dr. Stephen R.  Thorn—Institute for Environmental Medicine and
Department of Emergency Medicine, University of Pennsylvania, Philadelphia, PA  19104

Dr. Vanessa Vu—National Center for Environmental Assessment (8601D)
U.S. Environmental Protection Agency, 401 M St. SW, Washington, DC  20460

Dr. Jed Waldman—California Department of Health Services, Berkeley, CA 94704
                                         XXVI

-------
                      U.S. Environmental Protection Agency
                              Science Advisory Board
                     Clean Air Scientific Advisory Committee
                         Carbon Monoxide Review Panel
Chair
Dr. Joe Mauderly—Director of External Affairs, Senior Scientist, and Director of
National Environmental Respiratory Center, Lovelace Respiratory Research Institute
Albuquerque, NM 87108

Members

Mr. John Elston—Administrator, Office of Air Quality Management, State of New Jersey
Department of Environmental Protection and Energy, Trenton, NJ 08625

Dr. Philip K. Hopke—R.A. Plane Professor of Chemistry, Clarkson University
Potsdam, NY 13699

Dr. Eva J. Pell—Steimer Professor of Agriculture Sciences, Buckhout Laboratory
The Pennsylvania State University, University Park, PA 16802

Dr. Arthur C. Upton—Director, Independent Peer Review, CRESP
Environmental and Occupational Health Sciences Institute, Piscataway, NJ  08854

Dr. Sverre Vedal—Professor of Medicine, Vancouver General Hospital
Vancouver, BC Canada V57 3J5

Dr. Warren White—Senior Research Associate, Chemistry Department, Washington University
St. Louis, MO 63130

Consultants

Dr. Stephen M. Ayres—Director, International Health Programs
Virginia Commonwealth University/Medical College of Virginia, Richmond, VA 23284

Dr. Thomas E. Dahms—Professor and Director, Anesthesiology Research
Department of Anesthesiology, St. Louis University School of Medicine, St. Louis, MO 63110

Dr. Victor G. Laties—Professor Emeritus, Department of Environmental Medicine
University of Rochester Medical Center, Rochester, NY 14642

Dr. Brian Leaderer—Professor, Division of Environmental Health Sciences
Yale University  School of Medicine, New Haven, CT 06519

Dr. Lawrence D. Longo—Professor, School  of Medicine, Department of Physiology
Center for Perinatal Biology, Departments of Physiology and Gynecology and Obstetrics
School of Medicine, Loma Linda University, Loma Linda, CA 92354

Designated Federal Official

Mr. Robert Flaak—Designated Federal Officer and Team Leader, Committee Operations Staff
U.S.  Environmental Protection Agency, Science Advisory Board (1400), Washington, DC 20460

                                         xxvii

-------
                   U.S. Environmental Protection Agency
                          Science Advisory Board
                   Clean Air Scientific Advisory Committee
                       Carbon Monoxide Review Panel
                                    (cont'd)
Staff Assistant
Ms. Diana Pozun—Management Assistant, U.S. Environmental Protection Agency
Science Advisory Board (1400), Washington, DC 20460
                                    XXVIII

-------
                      U.S. Environmental Protection Agency
                         Project Team for Development of
                     Air Quality Criteria for Carbon Monoxide
Scientific Staff
Mr. James A. Raub—Project Manager and Coordinator for Health Effects
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Dr. Joseph P. Pinto—Coordinator for Atmospheric Chemistry, Sources, and Emissions
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Dr. Robert S. Chapman—Coordinator for Epidemiology Studies
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Mr. William G. Ewald—Coordinator for Measurement Methods
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Ms. Beverly M. Comfort—Coordinator for Indoor Air Emissions and Concentrations
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Dr. David T. Mage—Coordinator for Population Exposure
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Ms. Ellie Speh—Office Manager, Environmental Media Assessment Group
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Technical Support Staff

Mr. Douglas B. Fennell—Technical Information Specialist
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Ms. Diane H. Ray—Technical Information Manager
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Mr. Richard N. Wilson—Clerk
National Center for Environmental Assessment (MD-52), Research Triangle Park, NC  27711

Document Production Staff
Mr. John R. Barton—Document Processing Coordinator
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Ms. Diane G. Caudill—Graphic Artist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Ms. Yvonne A. Harrison—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Ms. Bettye B. Kirkland—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Mr. David E. Leonhard—Graphic Artist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

                                         xxix

-------
                     U.S. Environmental Protection Agency
                        Project Team for Development of
                    Air Quality Criteria for Carbon Monoxide
                                       (cont'd)

Ms. Carolyn T. Perry—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Ms. Veda E. Williams—Graphic Artist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Technical Reference Staff
Mr. R. David Belton—Reference Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Mr. John A. Bennett—Technical Information Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Mr. William C. Hardman—Reference Retrieval and Database Entry Clerk
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Ms. Sandra L. Hughey—Technical Information Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Mr. Jian Ping Yu—Reference Retrieval and Database Entry Clerk
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713
                                         XXX

-------
         AIR QUALITY CRITERIA  FOR CARBON  MONOXIDE
                           Executive  Summary
       The purpose of this document is to present air quality criteria for carbon monoxide (CO), in
accordance with Sections 108 and 109  of the Clean Air Act (CAA), that reflect the latest scientific
information useful in indicating the kind and extent of all identifiable effects on public health and welfare
that may be expected from the presence of CO in ambient air.  This document is an update of Air Quality
Criteria for Carbon Monoxide, published by the U.S. Environmental Protection Agency in 1991, and will
be used as the scientific basis for reevaluating the current  National Ambient Air Quality Standards
(NAAQS) for CO. This executive summary summarizes key findings from the present document.

Summary Findings

Monitoring
       Reliable methods are identified in Chapter 2 for monitoring CO concentrations in ambient air to
determine compliance with the NAAQS and the potential effects on overall air quality and for monitoring
the exposure of human populations to ambient CO.

• Several adequate techniques exist for highly reliable monitoring of CO to ensure compliance with the
 NAAQS.   The most reliable  method for continuous measurement of CO in ambient air is the
 nondispersive infrared optical transmission technique, the technique on which the U.S. Environmental
 Protection Agency-designated analytical reference methods  are based. One category of nondispersive
 infrared monitors, the gas filter correlation monitor, is still the single most widely used analyzer for fixed-
 site monitoring stations.
• Determining CO levels at many nonurban locations requires substantially better performance than that
 required to demonstrate compliance with the NAAQS.  Commercial  CO-monitoring instruments,
 sometimes with minor modifications, can meet the measurement needs for supplying useful data on the
 distribution and trends of ambient CO and for modeling photochemical smog in places where ambient
 levels are significantly below the NAAQS.
• There are commonly used and accepted procedures for generating CO measurement standards that are
 accurate to better than ±2% in the parts-per-million range and about ±10% in the range of concentrations
 found in the clean troposphere. Several CO measurement techniques have been intercompared and found
 reliable.

• Several electrochemical and passive sampling methods are available. These techniques are currently not
 equivalent to compliance monitoring methods but are useful for personal exposure studies and for
 measuring CO concentrations in indoor, outdoor, and in-transit microenvironments.
• Blood carboxyhemoglobin (COHb) level and CO concentration in exhaled breath are biological indicators
 of exogenous CO exposure and endogenous CO production.  Although the use of optical methods (e.g.,
 CO-Oximetry [CO-Ox]) is common for population sampling and clinical  analyses of COHb, gas
 chromatography is the method of choice for measuring low COHb levels (<5%) that are expected to occur

                                            E-1

-------
 with ambient CO exposures.  The measurement of CO in exhaled breath has practical advantages for
 population exposure sampling but has a greater potential for error in the estimation of COHb than does
 the direct measurement of COHb.

Global Tropospheric Chemistry
       Current information about the abundance and distribution, the nature of sources and sinks, and the
chemistry of CO in environments ranging from the global background to indoor air is summarized in
Chapter 3. The importance of CO for atmospheric chemistry also is discussed in this chapter.

• In nonurban areas, tropospheric CO has a significant role in affecting the oxidizing capacity of the earth's
  atmosphere. Reaction with CO is a principal process by which hydroxyl radicals are removed from the
  atmosphere. Reaction with hydroxyl radicals is also the primary process for removing many other man-
  made and natural compounds, including CO, from the atmosphere.

• Carbon monoxide is linked closely to the cycle of tropospheric ozone and participates in the formation
  of 20 to 40% of the ozone found in nonurban areas.  Ozone is an oxidant, a greenhouse gas, and a
  precursor of hydroxyl radicals. On balance, if CO increases, the net effect is to decrease hydroxyl
  radicals.
• Carbon monoxide is,  therefore, an intermediary in determining the future concentrations of many
  environmentally important trace gases. The future of methane, a greenhouse gas, cannot be evaluated
  adequately or predicted without an accurate understanding of the global CO budget, which  is not
  presently available. Similarly, predicting future concentrations of other environmentally important gases,
  such as the hydrochlorofluorocarbons that can deplete stratospheric ozone, depends on how well the CO
  budget is understood.

• Global background CO concentrations average about 120 and 40 ppb in remote marine areas  of the
  Northern and Southern Hemispheres, respectively, that are not affected by local sources. Results from
  flask and in situ monitoring stations show no discernible trend in CO levels from 1993 through 1997.

• The average lifetime of CO in the atmosphere is about 2 mo, longer at high latitudes and shorter at low
  latitudes.

• In addition to direct emissions from fossil fuel and biomass burning, CO is produced in the atmosphere
  by the photochemical oxidation of anthropogenic and biogenic hydrocarbons. Because of uncertainties
  in reaction kinetics, the identification of reaction products, and the effects of heterogeneous processes,
  the accuracy of estimates of photochemical sources of CO is limited.
• The global emissions of CO are about 2.3 x 109 metric tons per year, amounting to an annual source of
  about 1.0 x 109 metric tons of carbon in the atmosphere, compared with a global anthropogenic input of
  7.1 x IQ9 metric tons per year of carbon in carbon dioxide.  Estimates of individual CO sources are
  uncertain by a factor of two or more; however, the total production of CO is known to within 25%, based
  on its estimated rate of destruction because of reactions with hydroxyl radicals.

• Emissions from various  sources in developing countries  are likely to be very significant but are not
  known at present.

Regional and Urban Air Quality
       Emissions, concentrations, and effects of CO on air quality within the United States are discussed
in Chapter 3.
                                             E-2

-------
• Carbon monoxide plays  an  important role in  atmospheric photochemistry  in regional and urban
  environments. In urban areas, CO either can produce or destroy ozone, depending on the concentrations
  of nitrogen oxides and hydrocarbons. In numerical simulations of at least one urban air shed, CO was
  found to participate in the formation of 10 to 20% of the ozone found there.

• The nationwide average annual second-highest 8-h ambient CO concentration decreased from 10 ppm
  in 1978 to 4 ppm in 1997.

• On- and nonroad mobile  sources account for approximately 80% of the 1997 nationwide emissions
  inventory for CO.  Declines in ambient CO levels in the United States follow approximately the decline
  in motor vehicle emissions of CO. However, the relative importance of nonroad sources has increased
  over the past decade from 12.7% of total emissions in 1988 to 19.2% in 1997.
• There were 41 exceedances of the 8-h NAAQS for CO at 12 U.S. monitoring sites in 1997. These sites,
  in descending order, were located in Calexico  and Los Angeles-Long Beach, CA; Fairbanks, AK;
  Steubenville,  OH; El Paso, TX; and Phoenix, AZ.

• Median CO concentrations in the geographically diverse urban areas of Denver, CO; Los Angeles, New
  York, NY; Phoenix; and Fairbanks, have decreased from 1986 through 1995. However, the nature of the
  diurnal and seasonal variations has remained essentially the same.  These variations result largely from
  the  interaction among motor vehicle emissions, traffic patterns, and meteorological parameters, such as
  wind speed and mixing height.
• In general, the spatial distribution of CO within four of the five air sheds was highly heterogeneous. For
  instance, the  average  correlation between time  series of weekday 8-h average  maximum  CO
  concentrations at different monitoring sites in the Denver, New York, Los Angeles, and Phoenix urban
  areas ranged from 0.4 to 0.5. The corresponding average between-site correlation was 0.8 in Fairbanks.

Indoor Air Quality
      Indoor CO exposure may represent a significant portion of the total human exposure to CO.  The
sources, emissions, and concentrations of CO found in indoor microenvironments also are discussed in
Chapters.

• Carbon monoxide occurs indoors directly through emissions from various indoor combustion sources or
  indirectly as a result of infiltration or ventilation from outdoor sources. In the absence of indoor sources,
  average CO concentrations generally will equal those in the surrounding ambient environment.
• Emissions of CO  from the use of properly installed vented combustion appliances (e.g., gas and oil
  furnaces, gas water heaters and dryers) will not contaminate indoor air unless the units or venting systems
  are  malfunctioning.

• The maj or sources of CO in residential microenvironments are tobacco smoke, vehicle start-up and idling
  in attached garages, and unvented or improperly installed or malfunctioning  vented combustion
  appliances.  Factors affecting emissions of CO in the home include the type of source (e.g., gas
  appliances, woodstoves and  fireplaces, tobacco products), appliance design,  type of fuel used,  fuel
  consumption  rate, and source operating  condition.  Carbon monoxide concentrations in the  indoor
  environment will vary based on the source emission rate, use pattern, ambient CO concentration, air
  exchange rate, building volume, and air mixing within the indoor compartments.
• Carbon monoxide emissions from gas stoves depend on their use pattern, operating condition, and fuel
  consumption rate.  Ranges with standing pilot lights emit more CO than ranges with electronic pilot
  lights.   The contribution  of gas cooking stoves to CO  concentrations in the  indoor environment is
                                             E-3

-------
  expected to be negligible because of the intermittent nature of the stoves' use, unless gas stoves are used
  as a heat source.

• Carbon monoxide emissions from unvented space heaters vary as a function of unit design and operating
  condition, type  of fuel used and consumption rate, air currents near the space heater, and use pattern.
  Carbon monoxide concentrations in environments using space heaters depend on the type of space heater,
  emission rate, air exchange/infiltration rate, and frequency and duration of use.  Reported indoor CO
  concentrations are higher in homes using unvented space heaters as the primary source of heat.

• Woodstoves and fireplaces emit CO during fire start-ups and maintenance, through leaks in the stove or
  venting system, and from back drafting.  Carbon monoxide emissions are higher during the first stage
  of a fire because of increased fuel usage and lower combustion temperatures.
• Carbon monoxide emissions from tobacco smoke depend on the type of tobacco product (e.g., cigarette,
  cigar) and the degree to which tobacco is actively smoked.  Concentrations of CO from the use of tobacco
  products will exceed background concentrations, but will vary based on differences in ventilation, the
  number of cigarettes or cigars smoked, and the smoking rate.  For example, it is possible for cigar
  smokers to raise indoor CO concentrations to more than 9 ppm above ambient levels measured outside.

Population Exposure
       The reduction in automotive emissions brought about by the Clean Air Act have reduced in-traffic
CO exposures and traffic-related ambient CO concentrations well below those measured in the 1970s and
1980s. Chapter 4  describes the reduction over the past decade in human exposures to CO brought about
by the reduction in automotive emissions of CO. However, people still are exposed to CO at concentration
levels above the NAAQS in areas of high traffic density, and in indoor locations where tobacco is smoked
and where combustion devices (e.g., stoves, heaters) are not adequately vented.
• Fixed-site monitors often are used in urban areas to estimate the ambient concentrations to  which
  individuals in the surrounding areas may be  exposed.  These measurements tend to overestimate 8-h
  exposure values for people living in areas of lower traffic and underestimate the exposure of people
  living in areas of higher traffic.

• Neighborhood scale, fixed-site ambient CO monitoring may provide a reasonable estimate of the average
  CO exposures for people who live nearby and who are not exposed to tobacco smoke or other sources
  of CO in their homes and occupations.

• Nonsmokers exposed to tobacco smoke, heavy traffic fumes, and indoor sources of CO will have  higher
  body burdens of CO (COHb) than would be predicted from ambient data alone.

• Emission reductions in CO mandated by the Clean Air Act amendments have led to significant reductions
  in ambient CO  concentrations and lower traffic-related exposures to CO from motor vehicle  exhaust,
  suggesting that  estimates of current population exposure based on pre-1990 exposure studies may no
  longer apply. There currently is not a good estimate of CO exposure distribution for the population.

• Personal CO exposures that exceed the level of the NAAQS will still occur in some nonsmokers exposed
  to sources of CO not controlled by the Clean Air Act (e.g., recreational vehicles, garages, poorly vented
  or malfunctioning indoor combustion sources) or exposed in their occupations or hobbies to CO or to
  organic solvents that are metabolized to  CO (e.g., methylene chloride).
• Current CO exposure models adequately predict the average general population exposure  but still
  underpredict high CO exposures, indicating that further work is required to understand the activities and
  emissions associated with these higher exposures.
                                             E-4

-------
Pharmacokinetics and Mechanisms of Action
       The action of CO in the body and the factors influencing its uptake, distribution to vital tissues, and
elimination provide the foundation for measuring or predicting effects on organ function. In Chapter 5, the
basic principles of CO pharmacokinetics are reviewed, and the possible mechanisms for pathophysiologic
effects at the cellular level are discussed.
• A clear mechanism of action underlying the effects of low-level CO exposure is the decreased oxygen-
  carrying capacity of blood and subsequent interference with oxygen release at the tissue level that is
  caused by the binding of CO with hemoglobin, producing COHb. The resulting impaired delivery of
  oxygen can interfere with cellular respiration and cause tissue hypoxia. The critical tissues (e.g., brain,
  heart) of healthy subjects have intrinsic physiologic mechanisms (e.g., increased blood flow and oxygen
  extraction) to compensate for CO-induced hypoxia. In compromised subjects, or as CO levels increase,
  these compensatory mechanisms may be overwhelmed, and tissue hypoxia, combined with impaired
  tissue perfusion and systemic hypotension induced by hypoxia, may cause identifiable health effects.

• Carbon monoxide is produced endogenously  through heme degradation; metabolism of drugs; and
  degradation of unsaturated fatty acids, inhaled solvents, and other xenobiotics. High altitude and many
  medical disorders, especially anemias and inflammatory lung diseases, also may increase endogenous
  levels of CO.
• The amount of COHb formed from exogenous exposure is dependent on the CO concentration and
  duration of exposure, minute ventilation, lung diffusion capacity, and ambient pressure, as well as on the
  health  status and metabolism of the exposed individual.  The formation of COHb is reversible, but,
  because of a small blood-to-air CO pressure  gradient and tight binding  of CO to hemoglobin, the
  elimination half-time is quite long, varying from 2 to 6.5 h.

• The physical and physiological variables affecting the rate of COHb formation and elimination have been
  integrated into empirical and mathematical models for estimating COHb levels from different conditions
  of exposure. The nonlinear Coburn-Forster-Kane equation is the most widely used predictive model of
  COHb formation and still is considered the best all-around model for COHb prediction.
• Intracellular binding of CO to hemoproteins, particularly myoglobin found in heart and skeletal muscle,
  would be favored under conditions of low intracellular oxygen tension as COHb levels rise. The impact
  of ambient CO on intracellular CO uptake by myoglobin is not well understood.

• New investigations have expanded knowledge of the physiological effects of CO in two areas.  First,
  there is a growing recognition that CO may play  a role in normal neurotransmission and vasomotor
  control. Second,  there also is increased interest in the ability of CO to cause free-radical-mediated
  changes in tissues.  The impact of ambient CO on these processes and the roles they may have in
  pathophysiology are not yet well understood.

Health Effects
       Concerns about the potential health effects  of exposure to CO are addressed in Chapter 6 by
examining the published results of extensive controlled-exposure studies and more limited population-
exposure studies.  Emphasis is placed on the current  understanding of quantifiable health effects that are
likely to occur in humans at the low COHb levels (<5 %) that are predicted to result from typical ambient
CO exposures.

• Blood COHb levels are the best known indicators of potential health risk; however, the lowest-observed-
  effect levels depend on the method used for analysis. Gas chromatography (GC) generally is regarded
  as more accurate than CO-Ox for measuring low (<5%) COHb levels.


                                             E-5

-------
Maximal exercise duration and performance in healthy individuals has been shown to be reduced at
COHb levels of >2.3 and >4.3% (GC), respectively.  Performance decrements are small, however, and
likely to affect only competing athletes. No effects were observed during submaximal exercise in healthy
individuals at COHb levels as high as 15 to 20%.

Decreased exercise tolerance has been observed consistently in patients with coronary artery disease and
reproducible exercise-induced angina (chest pain) at COHb levels of 3 to 6% COHb (CO-Ox). The
indicators of myocardial ischemia during exercise, such as electrocardiographic changes and associated
chest pain, were statistically significant in one large multicenter clinical study at >2.4% COHb (GC) and
showed a dose-response relationship with increasing COHb.

An increase in the number and complexity of exercise-related arrhythmias (irregular heart beat) has been
observed at >6% COHb (CO-Ox) in some people with coronary artery disease and a high level of
baseline ectopy (a chronic arrhythmia) that may present increased risk of sudden death.
Some recent epidemiology studies are suggestive of community average ambient CO variations being
positively associated with fluctuations of indicators (e.g., cardiac-related hospital admissions) of heart
disease  exacerbation.  However, these findings are not considered conclusive because of questions
regarding (a) internal inconsistencies and coherence of the reported results within and across studies,
(b) the representativeness of the average ambient CO levels of spatially heterogenous ambient CO values
derived from fixed monitoring sites or of personal exposures that often include nonambient CO, and
(c) the biologic implausibility of any harmful effects occurring with the very small changes in COHb
levels (from near 0 up to about 1.0%) over typical baseline levels (about 0.5%) that would be expected
with the low average ambient CO levels (< 5.0 ppm, 1-h  daily max) evaluated in the epidemiology
studies.

Some epidemiologic studies  also suggest associations of  short-term ambient CO exposure with
nonaccidental daily mortality, the great majority of which occurs in people at least  65 years of age.
As above, the relative influences on these  associations of ambient and nonambient CO have not been
determined, and the possibility that CO is acting as a marker for other combustion-related pollutants
cannot be ruled out.

Laboratory animal studies indicate that acute CO poisoning can affect the growth and function of the
developing fetus.  Epidemiologic studies show a limited association between subchronic ambient CO
exposure and low birth weight; however, these studies are not conclusive.

Recent analyses indicate that significant behavioral  impairments in healthy individuals should not be
expected until COHb levels exceed 20%; however, mild  central nervous system effects have been
reported in the historical CO literature at COHb levels between 5 and 20%.
Ambient levels of CO are not known to have any direct effects on lung tissue. Observed epidemiologic
associations of short-term ambient CO levels with  daily respiratory illness frequency cannot yet be
interpreted with confidence.

Carbon  monoxide has the potential to interact with other stressors.  These include visitation to high
altitudes, especially for patients with coronary artery disease; use of psychoactive drugs or alcohol; use
of specific  medications, especially nitric oxide and calcium channel blockers; prolonged exposure to
heat; and exposure to other pollutants.
                                            E-6

-------
Subpopulations Potentially at Risk
       On the basis of monitored ambient CO concentrations and quantifiable CO concentration-response
relationships for health effects demonstrated in humans, the following conclusions are made in Chapter 7
regarding subpopulations potentially at risk from exposure to ambient CO.

• Young,  healthy  nonsmokers are not at immediate risk from ambient CO exposure because only
  limitations at maximal exercise performance have been demonstrated at the low COHb levels (<5%) that
  are predicted to result from ambient exposures.  Effects have not been demonstrated on healthy
  individuals performing submaximal exercise that is more typical of daily human activity.

• Patients with reproducible exercise-induced angina (chest pain) are a sensitive group within the general
  population that is at increased risk of experiencing decreased exercise tolerance because of exacerbation
  of cardiovascular symptoms at ambient or near-ambient CO-exposure concentrations that result in COHb
  levels of 2.4% (GC) or higher.
                                             E-7

-------
                                      CHAPTER 1
                                   Introduction
       This document is an update of Air Quality Criteria for Carbon Monoxide, published by the U.S.
Environmental Protection Agency (EPA) in 1991, and will serve as the basis for reevaluating the current
National Ambient Air Quality Standards (NAAQS) for carbon monoxide (CO) set in 1994.  Carbon
monoxide is one of six ubiquitous ambient air pollutants covered by the Federal Clean Air Act (CAA)
requiring an assessment of the latest scientific knowledge as a requisite step in the development of standards
to protect public health and welfare.  The present document is not intended as a complete and detailed
literature review, but it does summarize relevant key information from the previous 1991 document and
evaluates new information relevant to the CO NAAQS criteria development, based on pertinent published
literature available through 1999.
       Carbon monoxide, a trace constituent of the troposphere, is produced both by natural processes and
human activities. Because plants can both metabolize and produce CO, trace levels are considered a normal
constituent of the natural environment. Although ambient concentrations of CO in the vicinity of urban and
industrial areas can exceed global background levels, there are no reports of these currently measured levels
of CO producing any adverse effects on plants  or microorganisms.   Ambient  concentrations of CO,
however, may be detrimental to human health and welfare, depending on the levels that occur in areas
where humans live and work and on the susceptibility of exposed individuals to potentially adverse effects.
       This chapter presents a brief summary of the legislative and regulatory  history of the CO NAAQS
and the rationale for the existing standards, and it gives an overview of the issues, methods, and procedures
utilized in the preparation of the present document.
1.1  Legislative Requirements
       Two sections of the CAA govern the  establishment,  review, and revision  of the NAAQS.
Section 108 (U.S. Code, 1991) directs the Administrator of EPA to identify and issue air quality criteria for
pollutants that reasonably may be anticipated to endanger public health or welfare. These air quality criteria
are to reflect the latest scientific information useful in indicating the kind and extent of all identifiable
effects on public health or welfare that may be expected from the presence of the pollutant in ambient air.
       Section  109(a) of the CAA (U.S. Code, 1991) directs  the Administrator of EPA to propose and
promulgate primary and secondary NAAQS for pollutants identified under Section 108. Section 109(b)(l)
defines a primary standard as one that the attainment and maintenance of which, in the judgment of the
Administrator, based on the criteria and allowing for an adequate margin of safety, is requisite to protect
the public health.  The secondary standard, as defined in Section 109(b)(2), must specify a level of air
quality that the attainment and maintenance of which, in the judgment of the Administrator, based on the
criteria, is requisite to protect the public welfare from any known or anticipated adverse effects associated
with the presence of the pollutant in ambient air.
       Section 109(d) of the CAA (U.S. Code, 1991) requires periodic review and, if appropriate, revision
of existing criteria and standards. If, in the Administrator's judgment, EPA's review and revision of criteria
make  appropriate the proposal of new or revised standards,  such standards are to be revised and


                                             1-1

-------
promulgated in accordance with Section 109(b).  Alternatively, the Administrator may find that revision
of the standards is inappropriate and conclude the review by leaving the existing standards unchanged.
1.2  Regulatory Background
       On April 30, 1971, EPA promulgated identical primary and secondary NAAQS for CO at levels of
10 mg/m3 (9 ppm) for an 8-h average and 40 mg/m3 (35 ppm) for a 1-h average, not to be exceeded more
than once per year. The scientific basis for the primary standard, as described in the first criteria document
(National Air Pollution Control Administration, 1970), was a study suggesting that low levels of CO
exposure resulting in  carboxyhemoglobin (COHb) concentrations of 2 to 3% were associated with
neurobehavioral effects in exposed subjects (Beard and Wertheim, 1967).
       In accordance with Sections 108 and 109 of the CAA, EPA periodically has reviewed and revised
the criteria on which the existing NAAQS for CO (Table 1) are based. On August 18,1980, EPA proposed
certain changes in the standards based on scientific evidence reported in the revised criteria document for
CO (U. S. Environmental Protection Agency, 1979). Such evidence indicated that the Beard and Wertheim
(1967) study no longer should be considered as a sound scientific basis for the standard.  Additional medical
evidence accumulated since  1970, however, indicated that aggravation of angina  pectoris and other
cardiovascular diseases would occur at  COHb levels as low as 2.7 to 2.9%.  On August 18, 1980, EPA
proposed changes to the standard (Federal Register, 1980) based on the findings of the revised criteria. The
proposed changes included (1) retaining the 8-h primary  standard level of 9 ppm, (2) revising the  1-h
primary standard level from 3 5 ppm to 25 ppm, (3) revoking the existing secondary CO standards (because
no adverse welfare effects have been reported at or near ambient CO levels), (4) changing the form of the
primary standards from deterministic to statistical, and (5) adopting a daily interpretation for exceedances
of the primary standards, so that exceedances would be determined on the basis of the number of days on
which the  8- or 1-h average concentrations are above the standard levels.
	Table 1. National Ambient Air Quality Standards For Carbon Monoxide

 Date of Promulgation               Primary NAAQS                  Averaging Time

 August 1, 1994                     9 ppma (10 mg/m3)                      8-hb

	35ppma(40mg/m3)	l-hb

al ppm = 1.145 mg/m3, 1 mg/m3 = 0.873 ppm at 25 °C, 760 mm Hg.
bNot to be exceeded more than once per year.

Source: Federal Register (1994).
       The 1980 proposal was based in part on health studies conducted by Dr. Wilbert Aronow. In March
1983, EPA learned that the Food and Drug Administration (FDA) had raised serious questions regarding
the technical adequacy of several studies conducted by Dr. Aronow on experimental drugs, leading FDA
to reject use of the Aronow drug study data.  Therefore, EPA convened an expert committee to examine
the Aronow CO studies before any final decisions were made on the NAAQS for CO. In its report (Horvath
et al., 1983), the committee concluded that EPA shouldnot rely on Dr. Aronow's data because of concerns
regarding the research that substantially limited the validity and usefulness of the results.  An addendum
to the 1979 criteria document for CO (U.S.  Environmental Protection Agency, 1984) reevaluated the
                                             1-2

-------
scientific data concerning health effects associated with exposure to CO at or near ambient exposure levels
in light of the committee recommendations and taking into account findings reported subsequent to those
previously reviewed.  On  September 13, 1985, EPA issued  a  final notice (Federal Register, 1985)
announcing retention of the existing primary CO NAAQS and rescinding the secondary CO NAAQS.
       The criteria review process was initiated again on July 22, 1987, and notice of availability of the
revised draft criteria document was published in the Federal Register (Federal Register, 1990) on April 19,
1990. This draft document included discussion of several new studies of effects of CO on angina patients
that had been initiated in light of the controversy discussed above.  The  Clean Air Scientific Advisory
Committee (CAS AC) reviewed the draft criteria document at a public meeting held on April 30,1991.  The
EPA carefully considered comments  received from the public  and from CASAC in preparing the final
criteria document (U.S. Environmental Protection Agency, 1991). On July 17, 1991,  CASAC sent to the
EPA Administrator a "closure letter" outlining key issues and recommendations and indicating that the
document provided a scientifically balanced and defensible summary of the available knowledge of effects
of CO. A revised "staff paper" based on the scientific evidence was released for public review in February
1992, followed by two CASAC review meetings held on March 5 and on April 28, 1992.  The CASAC
came to closure on the final staff paper (U.S. Environmental Protection Agency, 1992) in a letter to the
Administrator dated August 11,1992, indicating that it provided a scientifically adequate basis for EPA to
make a regulatory decision on the appropriate primary NAAQS for CO. On August 1, 1994, EPA issued
a final decision (Federal Register, 1994) that revisions of the NAAQS for CO were not appropriate at that
time.
       In keeping with CAA requirements, EPA's National Center for Environmental Assessment again
periodically reviewed and revised the criteria for CO, as presented in this document.
1.3  Rationale for the Existing Carbon Monoxide Standards
       The following discussion describing the bases for the existing CO NAAQS set in 1994 has been
excerpted and adapted from "National Ambient Air Quality Standards for Carbon  Monoxide—Final
Decision" (Federal Register, 1994). The discussion includes the rationale for selection of the level and
averaging time for the NAAQS that would be protective of adverse effects  in the most sensitive
subpopulation and EPA's assessment that led to a decision not to revise the existing standards for CO.

1.3.1  Carboxyhemoglobin Levels of Concern
       In selecting the appropriate level and averaging time for the primary NAAQS for CO, the EPA
Administrator must first determine the COHb levels of concern, taking into account a large and diverse
health effects database. Based on the assessments provided in the criteria document (U.S. Environmental
Protection Agency, 1991) and in the staff paper (U.S. Environmental Protection Agency, 1992), judgments
were made to identify the most useful studies for establishing a range of COHb levels to be considered for
standard setting. In addition, the more uncertain or less quantifiable evidence was reviewed to determine
the lower end of the range that would provide an adequate margin of safety from effects of clear concern.
The following discussion summarizes the most critical considerations for the Administrator's 1994 decision
on the CO NAAQS.
       The Administrator of EPA concluded that cardiovascular effects, as measured  by decreased time
to onset of angina pain and by decreased time to onset of significant electrocardiogram (ECG) ST-segment
depression, were the health effects of greatest concern to be clearly associated with CO exposures at levels
observed in the ambient air. These effects  were demonstrated in angina patients at postexposure COHb
levels that were elevated to 2.9 to 5.9% (CO-Oximetry [CO-Ox] measurement), representing incremental
increases of 1.5 to 4.4% from baseline levels. Time to onset of significant ECG ST-segment change, which
is indicative of myocardial  ischemia in patients with documented coronary artery disease and a more

                                             1-3

-------
objective indicator of ischemia than angina pain, provided supportive evidence of health effects occurring
at exposures as low as 2.9 to 3.0% COHb (CO-Ox). The clinical importance of cardiovascular effects
associated with exposures to CO resulting in COHb levels less than 2.9% remains less certain and was
considered only in evaluating whether the current CO standards provide an adequate margin of safety.
       The Administrator of EPA also considered the following factors in evaluating the adequacy of the
current CO NAAQS.
• Short-term reductions in maximal work capacity were measured in trained athletes exposed to CO
  sufficient to produce COHb levels as low as 2.3%.
• The wide range of human susceptibility to CO exposures and ethical considerations in selecting subjects
  for experimental purposes, taken together, suggest that the most sensitive individuals have not been
  studied.
• Animal studies of developmental toxicity and human studies of the effects of maternal smoking provide
  evidence that exposures to high concentrations of CO can be detrimental to fetal development, although
  little is known about the effects of ambient CO exposures on the developing human fetus.
• Although little is known about the effects of CO on other potentially sensitive populations besides those
  with coronary artery disease, there is reason for concern about visitors to high altitudes, individuals with
  anemia or respiratory disease, and the elderly.
• Impairment of visual perception, sensorimotor performance, vigilance, and other central nervous system
  effects have not been demonstrated to be caused by CO concentrations commonly found in ambient air;
  however, short-term peak CO exposures may be responsible for impairments that could be a matter of
  concern for complex activities such as automobile driving.
• Limited evidence  suggests concern for individuals exposed to CO concurrently with drug use (e.g.,
  alcohol), heat stress, or coexposure to other pollutants.
• Large uncertainties remain regarding modeling COHb formation and estimating human exposure to CO
  that could lead to over- or underestimation of COHb levels associated with attainment of a given CO
  NAAQS in the population.
• Measurement of COHb made using the CO-Ox technique may not reflect the COHb levels in angina
  patients studied, thereby creating uncertainty in establishing a lowest effects level for CO.
The Administrator concluded that the lowest COHb level at which adverse effects have been demonstrated
in persons with angina is around 2.9 to 3.0%, representing an increase of 1.5% COHb above baseline when
using the CO-Ox to  measure COHb. These data serve to establish the upper end of the range of COHb
levels of concern.  Taking into account the above data uncertainties, the less significant health endpoints,
and less quantifiable data on other potentially sensitive groups, the lower end of the range was established
at 2.0% COHb.

1.3.2  Relationship Between Carbon Monoxide Exposure and Carboxyhemoglobin Levels
       To set ambient CO standards based on an assessment of health effects at various COHb levels, it is
necessary to estimate the ambient CO concentrations that are likely to result in COHb levels of concern.
The best all-around  model for predicting COHb levels is the Coburn, Foster, Kane (CFK) differential
equation (U.S. Environmental Protection Agency, 1991).  Baseline estimates of COHb levels expected to
be reached by nonsmokers exposed to various constant concentrations of CO can be determined by the CFK
equation (U.S. Environmental Protection Agency, 1992).  There are, however, two major uncertainties
involved in estimating COHb levels resulting from exposure to CO concentrations.  First, the large
distribution of physiological parameters used in the CFK equation across the population of interest is
sufficient to produce noticeable deviations in the COHb levels. Second, predictions based on exposure to
constant CO concentrations can  under-  or overestimate responses of individuals exposed to  widely
fluctuating CO levels that typically occur in the ambient environment.
                                             1-4

-------
1.3.3  Estimating Population Exposure
       The EPA review included an analysis of CO exposures expected to be experienced by residents of
Denver, CO, under air quality scenarios where the 8-h NAAQS is just attained. Although the exposure
analysis included passive smoking and gas stove CO emissions as indoor sources of CO, it did not include
other sources that may be of concern to high-risk groups (e.g., lawn equipment, woodstoves, fireplaces,
faulty furnaces).  The analysis indicated that, at the 8-h standard, fewer than 0.1% of the  nonsmoking
cardiovascular-disease population would experience a COHb level >2.1% (U.S. Environmental Protection
Agency, 1992).  A smaller population was estimated to exceed higher COHb percentages.

1.3.4  Decision on the Primary Standard
       Based on the exposure analysis results described above, the Administrator of EPA concluded that
relatively few people of the cardiovascular sensitive population group analyzed would experience COHb
levels >2.1% when exposed  to CO levels in the absence of indoor sources when the current ambient
standards were attained.  Although indoor sources of CO may be of concern to high-risk groups,  their
contribution cannot be effectively mitigated by ambient air quality standards.
       The Administrator of EPA also determined that both the 1-h and 8-h averaging times for CO were
valid because the 1 -h standard provided reasonable protection from health effects that might be encountered
from very short duration peak (acute) exposures in the urban environment, and the 8-h standard provided
a good indicator for tracking continuous exposures that occur during any 24-h period. The Administrator
concurred with staff recommendations (U.S. Environmental Protection Agency, 1992) that both averaging
times be retained for the primary CO  standards.
       For these reasons, the EPA Administrator determined under CAA Section 109(d)(l) that revisions
to the current 1-h (35 ppm) and 8-h (9 ppm) primary standards for CO were not appropriate at that time
(Federal Register, 1994).
1.4  Issues of Concern for the Current Criteria Development
       The following is abrief summary of key scientific issues that are addressed in this revised air quality
criteria document for CO. These issues are based on findings presented at symposia and workshops that
were convened to assess the current state of understanding of the sources, atmospheric cycle, and health
effects of CO and revised, as appropriate, by peer review comments received on earlier draft chapters of
this document.

1.4.1  Sources and Emissions
       Detailed  descriptions of the processes  forming CO during combustion were presented in  the
previous CO document.  These descriptions have been reviewed for accuracy in the revised document;
however, a  good deal of uncertainty exists  regarding the  correct values for CO emissions from
transportation sources. The term "transportation" includes both onroad and nonroad sources.  Onroad
sources  consist of automobiles, trucks, and buses. Nonroad sources consist of categories such as trains;
aircraft; boats and ships; and lawn, construction, recreational, logging, and agricultural equipment (see
Section  3.3). Emissions from transportation have been revised upward in the current emissions and trends
report (U. S. Environmental Protection Agency, 1996) from those used in the previous document. Emissions
estimates for CO from various sources are highly uncertain, especially those for transportation sources. The
potential of relatively newtechniques (e.g., inverse modeling) for testing and improving emissions estimates
needs to be evaluated.
                                             1-5

-------
1.4.2  Atmospheric Chemistry
       Much of the material discussed in the previous criteria document is already available in standard
textbooks and does not need to be reviewed. New information, however, is needed in this current review
regarding the chemistry of CO formation from the oxidation of methane and nonmethane hydrocarbons
(NMHCs).  For example, the fractional yields of CO resulting from the oxidation of NMHCs, especially
isoprene and monoterpenes, need to be established. The importance of CO for ozone formation in the urban
and nonurban atmosphere also needs to be highlighted.
       There are a number of ways to express the amount of a substance in the atmosphere. Perhaps the
most commonly used measure is concentration, which is the amount,  or mass, of a substance in a given
volume divided by that volume (e.g., moles per cubic meter in International System of Units [SI]).  Often
in the literature, however, quantities of gaseous substances are expressed as volume mixing ratios, such as
parts per million or parts per billion.  These terms are technically not "concentrations", but rather refer
properly to the molar mixing ratio of a substance (equivalent to volumetric mixing ratio for an ideal gas),
which is the ratio of the concentration of a substance to the concentration of all gaseous  components in a
given air volume (both in moles per cubic meter) (Seinfeld and Pandis, 1998).  Thus, mixing ratio is a mole
fraction that in SI units should be expressed as micromoles per mole for parts per million, and nanomoles
per mole for parts per billion.  Throughout this document, however,  mixing ratio will be referred to as
concentration of CO in parts per million or parts per billion because these terms have been extensively
referred to in the human exposure, toxicological, and  epidemiological literature and as the basis for CO
compliance monitoring for the NAAQS.

1.4.3  Global  Cycle
       Global trends in tropospheric CO concentrations declined from about 1988 to 1993 after several
years of annual increases, as determined by different networks of surface observations. Carbon monoxide
levels apparently have stabilized since 1993. The reasons for the changes in CO trends still need  to be
determined.

1.4.4  Measurement Technology
       The discussion on measurement methods for CO in the previous  document has been reviewed, older
methods have been removed, and newer methods for monitoring CO from various environmental sources
are presented.

1.4.5  Ambient Air Quality
       Because of the everchanging nature of atmospheric concentrations, levels in various environments
(rural, urban, and suburban) have been reanalyzed for different regions  of the United States. The temporal
variability of CO levels from daily to seasonal time scales also has been characterized. Relations between
urban concentrations of CO and regional and global  background levels also are examined, as well as
background levels of CO for use in different applications.

1.4.6  Indoor Emissions and Concentrations
       Indoor concentrations of CO are a function of outdoor concentrations, indoor sources, infiltration,
ventilation, and air mixing.  In the absence of indoor sources, concentrations of CO  in the indoor
environment are similar to those in ambient air; however, personal CO exposure studies have shown that
CO concentrations in  excess of 9 ppm  can occur in  certain indoor  and  in-transit microenvironments
associated with transportation sources that are not considered part of the ambient air. Unvented, improperly
installed, or poorly maintained combustion appliances, downdrafts during unstable weather conditions, and
depressurization from the operation  of exhaust systems and fireplaces also may contribute to potentially
                                             1-6

-------
high CO concentrations indoors. Further research is still needed, however, to determine the contribution
of nonambient sources to total human exposure to CO.

1.4.7  Exposure Assessment
       Compliance with the NAAQS is determined by measurements taken at fixed-site, ambient monitors,
yet exposure monitoring in the field and modeling studies indicate that individual personal CO exposures
are generally higher than ambient CO concentrations and often do not have a significant positive correlation
when compared to ambient CO concentrations measured by the fixed-site monitors alone. This is because
of the mobility of people and the spatial and temporal variability of CO concentrations across a given area.
The  nature of differences between fixed-site and personal monitoring results should be  given greater
attention, especially in regard to interpreting the results of epidemiology studies.
       Data from population field studies can be used to construct and test models of human exposure that
account for time and activity patterns known to affect exposure to CO.  New information from field
monitoring studies needs to be incorporated into exposure models to better capture the observed personal
exposure distributions, including the higher exposures found in the tail of exposure distribution.
       A unique feature of CO exposure is that the dose an individual receives can be estimated by
measuring COHb. The reader should note, however, that such exposure estimates are affected by the time
interval between peak CO exposure and blood sampling and by the use of any supplemental oxygen therapy.
It also has been shown that the method chosen for measurement of COHb can be a source of considerable
error, particularly at the low end of the CO dissociation curve, where COHb levels are <5%. The sensitivity
of COHb measurement techniques will, therefore, have an influence on the lowest-observed-effect level
(LOEL) for CO.  Gas chromatography (GC) is regarded as more accurate than CO-Ox for measuring the
lower COHb levels.

1.4.8  Mechanisms of Action
       The principle cause of CO toxicity is tissue hypoxia caused by CO binding to hemoglobin (Hb) and
failure of vasodilation to compensate for the reduced oxygen delivery.  Secondary mechanisms related to
intracellular uptake of CO have been the focus of recent research.  Current knowledge summarized in this
document suggests that the most likely protein other than Hb to be inhibited functionally at relevant levels
of COHb is myoglobin, found in heart and skeletal muscle. The extent of effects caused by CO molecules
in solution needs to be evaluated in relation to typical ambient CO exposures in the population.  Other
mechanisms of interest, which have not yet been demonstrated to occur at ambient CO levels, are cytotoxic
effects (e.g., vasomotor control, free oxygen radicals) independent of impaired oxygen delivery.

1.4.9  Health Effects
       There are many published studies on acute experimental and accidental exposures to CO; however,
there is not enough reliable information on chronic  exposures to low concentrations from either ambient
population-exposure studies or from occupational studies. Further work is needed, therefore, to determine
potential long-term exposures in the population and to develop reliable dose-response relationships for
at-risk groups. This information currently is missing  from the published literature. Some of the issues
associated with acute CO exposures are discussed below.

Cardiovascular Effects
       Maximal exercise duration is reduced in young, healthy, nonsmoking individuals at COHb levels
as low as 2.3% (GC), but this effect is small and would be of concern primarily for competing athletes.
Clinical studies on subj ects with reproducible exercise-induced angina have confirmed that adverse effects
occur with postexposure COHb levels as low as 2.4% (GC). Thus, aggravation of coronary artery disease
continues to provide the best scientific  basis in support for the  current (9-ppm, 8-h  and  35-ppm,  1-h)


                                             1-7

-------
NAAQS for CO.  More recent epidemiology studies in the United States, Canada, and Europe have
suggested that day-to-day variations in ambient CO concentrations are related to cardiovascular hospital
admissions and daily mortality, especially for individuals over 65 years of age. It is not clear, however,
if the observed association results from CO or from combustion-related particles, or, perhaps, from some
other, unmeasured pollutant exposure that covaries in time with CO.

Cerebrovascular Effects
       Carbon monoxide hypoxia increases cerebral blood flow in healthy subjects, even at very low
exposure levels. Behaviors that require sustained attention or performance are most sensitive to levels of
COHb >5%.  Disease or injury that  impairs compensatory increases in blood flow may increase the
probability of effects, but little is known about the susceptibility of compromised individuals to  ambient
levels of CO. Accidental exposures to high-level CO have been shown to cause neurological problems
weeks after recovery from the acute episode. It is not known, however, if these late neurological sequelae,
described as intellectual deterioration; memory impairment; and cerebral, cerebellar, and mid-brain damage,
result from long-term exposure to low ambient levels of CO.

Developmental Toxicity
       Relatively high CO exposures of 150 to 200 ppm during gestation, leading to approximately 15 to
25% COHb, produce reductions in birth weight, cardiomegaly,  delays in behavioral development, and
disruption in  cognitive function in newborn laboratory animals  of several species. Little data  exist on
humans exposed to CO for predicting a LOEL for developmental effects.  Studies relating human CO
exposures from ambient sources or cigarette smoking to reduced birth weight are potentially  relevant
because of the risk for developmental disorders; however, many  of these studies have not considered all
sources of CO and may be confounded by other variables (e.g., smoke components, maternal behavior,
nutrition, genetics).  Nevertheless, some health professionals have considered this evidence sufficient to
identify pregnant women, and the developing fetuses, as at risk to ambient levels of CO.

High-Altitude Effects
       There are relatively few reports on the effects of inhaling CO at high altitudes. Current knowledge
supports the possibility that the effects of hypoxic hypoxia and CO-hypoxia are at least additive. The
potential additive effects of CO exposure in sensitive individuals visiting at high altitudes need to be
considered.

1.4.10  Carbon Monoxide Interaction with Drugs
       There remains little direct information on the possible enhancement of CO toxicity by concomitant
illegal and prescription drug use or abuse; however, there are some data on psychoactive drugs that suggest
cause for concern.

1.4.11  Subpopulations at Risk
       On the basis of known effects described, heart disease patients with reproducible exercise-induced
angina appear to be best established as a sensitive group within the general population that is at increased
risk for experiencing health effects of concern at ambient or near-ambient CO exposure concentrations
resulting in COHb levels <5%.  Certain other groups are at potential risk from exposure to CO, but further
research is required to specify health effects associated with ambient or near-ambient CO exposures in these
probable risk groups.
                                              1-8

-------
1.5  Methods and  Procedures for Document Preparation
       The procedures that were followed for developing the revised criteria document for CO are different
from those that have been used for recent criteria documents.  For example, the previous CO criteria
document (U.S. Environmental Protection Agency, 1991) was a more comprehensive scientific review of
available information on the nature, sources, distribution, measurement, and concentrations of CO in the
environment and on the known and anticipated health effects that CO would have on at-risk population
groups.  In lieu of a comprehensive review of the literature, emphasis in the present criteria document has
been placed on the development of a concise summary  of  key information  and a more interpretative
discussion of the new scientific and technological data available since the previous criteria were evaluated.
The resulting document is more of an update, in accordance with recommendations made by CASAC.
       The main focus of this revised criteria document is on the evaluation and interpretation of more
recent air quality, human exposure, and health effects issues. Therefore, the techniques used to present this
information vary according to the state-of-science for the respective topics. For example, the analysis of
ambient air quality  is based on newly obtained air monitoring data and utilizes the previous analysis only
for showing trends over time. As a result of the relatively dramatic decrease in ambient CO concentrations,
population exposure to  ambient CO also has declined.  Human exposure studies conducted in the early
1980s and earlier distributions of COHb levels in the U.S. population that were relied on heavily in the
previous assessment are no longer relevant to the current picture of ambient CO exposure in the 1990s.
Thus, key information on population exposure must focus on the newer studies and on modeling results.
On the other hand, the health effects literature on CO has remained relatively static since the previous 1991
assessment, except for provocative publications on cellular mechanisms of CO action and on epidemiologic
associations of ambient CO with mortality and morbidity in the elderly population.  Newly published
studies on most of the other health outcomes reconfirm the conclusions made in the last document and are
incorporated into the previous summaries by reference only.
       Among early steps used in development of this revised document was the convening of symposia
or workshops to identify key scientific issues and to focus on selection of material to be included in the
document as the basis for the development of standard-setting criteria. Both EPA and non-EPA scientific
experts were utilized for this effort.
       First, an interdisciplinary scientific symposium was held in Portland, OR, in December 1997, to
assess current scientific understanding of the atmospheric cycle of CO, including its sources, sinks, and
distribution. Three main subject areas covered in the symposium relate to the distribution and spatial and
temporal variability of CO, the atmospheric budget of CO, and direct or indirect effects of CO on human
health.   Results from published symposium papers presented are included by reference in this revised
criteria document.
       Also, a mini workshop, jointly organized by EPA, the Gas Research Institute, and the Health Effects
Institute, was convened (April 24 and 25, 1998) in Chicago, IL, to provide expert scientific discussion on
the public health significance of exposures to low levels (<50 ppm) of CO. The three main topics covered
were human exposure patterns and trends in CO exposure, pharmacokinetics and mechanisms of action of
CO, and health effects. A summary of the workshop discussions and conclusions drawn from the meeting
were used by authors in preparation of draft criteria document chapters.
       Next, EPA  convened on September 17 and 18, 1998, a public peer-review workshop to define
further key issues, to review early drafts of the criteria document chapters, and to ascertain and discuss any
pertinent new literature. The authors of the draft chapters or sections of the document revised them on the
basis of the workshop recommendations. The revised chapters of the document then were incorporated into
the First External Review Draft of the document, which was released for public comment and was reviewed
by CASAC on June 9, 1999.  Necessary revisions were made in response to public comments and CASAC
recommendations before a Second External Review Draft of the criteria document was released in October
1999 and closure was reached on the document with CASAC at a November 18, 1999, public meeting.

                                             1-9

-------
1.6 Organization and Content of the Document
       This updated air quality criteria document for CO critically evaluates and assesses scientific
information on  air  quality, human  exposure,  and health effects  associated with exposure  to  the
concentrations found in the environment.  Emphasis has been placed on the development of a concise
review of key information and a more interpretative discussion of the new scientific and technological data
available since completion of the previous criteria document (U.S. Environmental Protection Agency,
1991).  The references cited in the document should be  reflective of the state of knowledge through 1999
on those issues most relevant to review of the NAAQS  for CO.
       To aid in the development of this document, concise summaries of the relevant published literature
and selective discussion of the literature have been undertaken. Studies that were presented in the previous
criteria document and whose data were judged to be significant because of their usefulness in deriving the
current NAAQS are discussed briefly  in the text. The reader, however, primarily is referred to the more
extensive discussion of these "key"  studies  in the previous document.  Other, older studies are discussed
in the text if they are open to reinterpretation because of newer data or are potentially useful  in deriving
revised standards for CO. Generally, only published information that has undergone scientific peer review
has been included in this revised criteria document.  However, some newer studies not yet published in the
open literature but meeting high standards of scientific  reporting also have been included in a few areas.
       The structure of the present document follows the general outline of the previous criteria document
(U.S. Environmental Protection Agency, 1991), especially for topics that have changed little since the last
criteria review. The resulting sequence of discussion should help the reader to find and contrast  similar
sections. There are, however, a few exceptions where  some topics have been consolidated into a single
chapter in order to present a more concise document.  The executive summary at the beginning of the
document provides a concise presentation of key information and conclusions from all subsequent chapters.
       The document begins with this introduction (Chapter 1), which provides the regulatory history of
CO and an understanding of the  scientific basis for the current CO NAAQS.  Information on analytical
methods for monitoring CO (Chapter 2) includes the measurement of CO in ambient (outdoor) and indoor
air, as  well as methods for measuring breath and blood CO levels in exposed individuals.  Chapter 3
provides information  on  the atmospheric  chemistry of  CO  and  typical sources,  emissions, and
concentrations found in the ambient and indoor environments—topics addressed in separate chapters of the
previous document.  The remaining chapters are similar to the previous  document, covering topics on
population exposure to CO (Chapter 4), pharmacokinetics and mechanisms of action (Chapter 5), and health
effects (Chapter 6). The final chapter (Chapter 7) provides an overall integrative summary of key findings
and an evaluation of subpopulations potentially at risk from exposure to CO.
References
Beard, R. R.; Wertheim, G. A. (1967) Behavioral impairment associated with small doses of carbon monoxide. Am. J. Public
      Health 57: 2012-2022.
Federal Register. (1980) Carbon monoxide; proposed revisions to the national ambient air quality standards: proposed rule. F. R.
      (August 18) 45: 55,066-55,084.
Federal Register. (1985) Review of the national ambient air quality standards for carbon monoxide; final rule. F. R. (September
      13)50:37,484-37,501.
Federal Register. (1990) Draft criteria document for carbon monoxide; notice of availability of external review draft. F. R. (April
      19)55: 14,858.
Federal Register. (1994) National ambient air quality standards for carbon monoxide—final decision. F. R. (August 1)
      59: 38,906-38,917.
Horvath, S. M.; Ayres, S. M.; Sheps, D. S.; Ware, J. (1983) [Letter to Dr. Lester Grant, including the peer-review committee report
      on Dr. Aronow's studies]. Washington, DC: U.S. Environmental Protection Agency, Central Docket Section; Docket no.
      OAQPS-79-7 IV. H.58.
                                               1-10

-------
National Air Pollution Control Administration. (1970) Air quality criteria for carbon monoxide. Washington, DC: U.S. Department
       of Health, Education, and Welfare, Public Health Service; report no. NAPCA-PUB-AP-62.
Seinfeld, J. H.; Pandis, S. N. (1998) Atmospheric chemistry and physics: from air pollution to climate change. New York, NY:
       John Wiley & Sons, Inc.
U.S. Code. (1991) Clean Air Act, §108, air quality criteria and control techniques, §109, national ambient air quality standards.
       U S. C. 42: §§7408-7409.
U. S. Environmental Protection Agency. (1979) Air quality criteria for carbon monoxide. Research Triangle Park, NC: Office of
       Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA-600/8-79-022.
U.S. Environmental Protection Agency. (1984) Revised evaluation of health effects associated with carbon monoxide exposure:
       an addendum to the 1979 EPA air quality criteria document for carbon monoxide. Research Triangle Park, NC: Office of
       Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA-600/8-83-033F.
U.S. Environmental Protection Agency. (1991) Air quality criteria for carbon monoxide. Research Triangle Park, NC: Office of
       Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA/600/8-90/045F.
U.S. Environmental Protection Agency. (1992) Review of the national ambient air quality standards for carbon monoxide: 1992
       reassessment of scientific and technical information. OAQPS staff paper. Research Triangle Park,  NC: Office of Air
       Quality Planning and Standards; report no. EPA-452/R-92-004.
U.S. Environmental Protection Agency. (1996) National air quality and emissions trends report, 1995. Research Triangle Park,
       NC: Office  of Air Quality Planning  and  Standards, Emissions  Monitoring and  Analysis Division; report no.
       EPA/454/R-96-005.
                                                      1-11

-------
                                    CHAPTER 2

              Analytical Methods for Monitoring
                            Carbon Monoxide
2.1  Introduction
       Investigations into relationships between ambient carbon monoxide (CO) levels and human health
outcomes and public health warnings of potentially harmful CO levels require accurate, precise, and
representative measurements of CO. Reliable measurement methods also are needed to evaluate the effects
of ambient CO on overall air quality. This chapter will review methods for monitoring CO in ambient air
for conditions ranging from clean continental environments to polluted urban ones. Biological methods for
monitoring the impact of ambient CO exposure on human populations also will be reviewed.
       To promote uniform enforcement of the air quality standards set forth under the Clean Air Act as
amended (U.S. Code, 1991), the U.S. Environmental Protection Agency (EPA) has established provisions
under which analytical methods can be designated as "reference" or "equivalent" methods (Code of Federal
Regulations,  199la).  Either a reference method or an equivalent method for air quality measurements is
required for acceptance of measurement data for National Ambient Air Quality Standards (NAAQS)
compliance. An equivalent method for monitoring CO can be so designated when the method is shown to
produce results equivalent to the approved reference monitoring method based on absorption of infrared
radiation from a nondispersed beam.
       The EPA-designated reference methods are automated methods utilizing the nondispersive infrared
(NDIR) technique, generally accepted as being the most reliable, continuous method for the measurement
of CO in ambient air. The official EPA reference methods (Code of Federal Regulations, 199 la) include
11 reference methods designated for use in  determining compliance for CO. Before a particular NDIR
instrument can be used in a reference method, it must be designated by the EPA as approved in terms of
manufacturer, model number, components, operating range, etc. Several NDIR instruments have been so
designated (Code of Federal Regulations, 1991a), including the gas filter correlation (GFC) technique,
which was developed through EPA-sponsored research (Burch et al., 1976). No equivalent method using
a principle other than NDIR has been designated for measuring CO in ambient air.
2.2  Overview of Techniques for Measurement of Ambient Carbon Monoxide
       The NDIR technique is an automated and continuous method that is based on the specific absorption
of infrared radiation by the CO molecule (Feldstein, 1967).  Most commercially  available analyzers
incorporate a gas filter to minimize interferences from other gases; they operate near atmospheric pressure,
and the most sensitive analyzers are able to detect minimum CO concentrations of about 0.02 ppm.
Interferences because of carbon dioxide (CO2) and water vapor can be dealt with so as not to affect the data
quality; a particle filter (Teflon® or nylon composition is recommended) and desiccant in the inlet line
improve reliability.  Nondispersive infrared analyzers are relatively insensitive to flow rate, require no wet
chemicals, are sensitive over wide concentration ranges, and have short response times. Nondispersive
                                           2-1

-------
infrared analyzers of the newer GFC type have overcome zero and span problems, as well as minor
problems caused by vibrations.
       A  more sensitive method for measuring low background levels is gas chromatography (GC)
(Bergman etal., 1975;Bruneretal, 1973;Dagnall etal, 1973; Porter and Volman, 1962;Feldstem, 1967;
Smith et al., 1975;  Swinnerton et al., 1968; Tesafik and Krejci, 1974).  This technique is an automated,
semicontinuous method where CO is separated from water, CO2, and hydrocarbons other than methane
(CH4) by a stripper column. Carbon monoxide and CH4 then are separated on an analytical column, and
the CO is passed through a catalytic reduction tube, where it is converted to CH4.  The CO (converted to
CH4) passes through a flame ionization detector (FID), and the resulting signal is proportional to the
concentration of CO in the air.  Mercury liberation detectors offer greater sensitivity and ease of operation
than FIDs.  (Section 2.4.4.3). These methods have no known interferences and can be used to measure
levels from 0.02 to  45 ppm.
       In  some environments, new technologies allow CO measurements to be made where it is difficult
to install  reference instruments.  Methods using electrochemistry, or integrated observations using
absorbents, allow measurements for personal monitoring exposure studies, for in-transit measurements, and
in areas where sites are being investigated and permanent installations  are not yet planned or available.
Focused studies of these developing methods should be encouraged to determine their precision, sensitivity
and stability because an understanding of the impact that CO has on human health requires the ability to
monitor under a wide variety of conditions.
       Whichever  method or instrument is used, it is essential that the  results be evaluated by frequent
calibration with samples of known composition (Commins et al., 1977; Goldstein, 1977; National Bureau
of Standards, 1975).  Chemical analyses can be relied on only after the analyst has achieved acceptable
accuracy in the analysis of such standard samples through an audit program.
       The performance specifications for automated CO analyzers currently in use are shown in Table 2-1.
The normal full-scale operating range for reference methods is 0 to 50 ppm (0 to  57 mg/m3).   Some
instruments offer higher ranges,  typically 0 to 100 ppm (0 to 115 mg/m3), or lower ranges such  as 0 to
20 ppm (0 to 23 mg/m3).  Fligher ranges up to  1,000 ppm (1,145 mg/m3) are used to measure CO
concentrations in vehicular tunnels and parking garages.
       Although CO is one of the criteria pollutants, it is also a precursor to ozone and a useful tracer of
combustion-derived pollutants (Carter, 1991; Ryan et al., 1998). These  additional roles for CO make its
detection at levels well below the NAAQS highly desirable. At many existing monitoring sites, the mixing
ratio is frequently below the lower detectable limit specified in Table 2-1.  Chemical Transport Models
(CTMs), developed to understand air pollution  and often required to  test  abatement strategies  for
photochemical smog, rely on accurate data for concentrations of source  gases including nitrogen oxides,
non-methane hydrocarbons, and CO. Boundary layer CO concentration ratios in urban areas are typically
100s of ppb (Seinfeld and Pandis, 1998; Moy et  al., 1994; Morales-Morales, 1998). A CO monitor with
precision of 500 ppb would be adequate to prove  compliance with the CO standard, but would not provide
adequate input data for CTMs.  This chapter, therefore, will review methods for measuring CO in ambient
air that provide sensitivity adequate to quantify the content of clean continental boundary layer air, that is
with uncertainty on the order of 10 ppb and having a detection limit around 50 ppb, in addition to methods
in current  use. Suggested performance specifications for monitoring CO in nonurban environments are
shown in Table  2-2.
2.3  Gas Standards for Calibration
       There are basically two different types of calibration gas mixtures:  (1) pre-made blends and
(2) mixtures prepared in the laboratory. Certain types of pre-made blends can be purchased with recognized
and accepted certification and traceability information. Other pre-made blends can be purchased without

                                             2-2

-------
 Table 2-1.  Performance Specifications for Automated Analytical Methods for Carbon Monoxide

 Range                                                     0 to 50 ppm (0 to 57 mg/m3)
 Noise                                                     0.5 ppm (0.6 mg/m3)
 Lower detectable limit                                     1.0 ppm (1.2 mg/m3)
 Interference equivalent
    Each interfering substance                               ±1.0 ppm (±1.2 mg/m3)
    Total interfering substances                              1.5 ppm (1.7 mg/m3)
 Zero drift
    12 h                                                    ±1.0 ppm (±1.2 mg/m3)
    24 h                                                    ±1.0 ppm (±1.2 mg/m3)
 Span drift, 24 h
    20% of upper range limit                                 ±10.0%
    80% of upper range limit                                 ± 2.5%
 Lag time                                                   lOmin
 Rise time                                                  5 min
 Fall time                                                   5 min
 Precision
    20% of upper range limit                                 0.5 ppm (0.6 mg/m3)
    80% of upper range limit	0.5 ppm (0.6 mg/m3)	

Definitions:
Range: Nominal minimum and maximum concentrations that a method is capable of measuring.
Noise:  The standard deviation about the mean of short duration deviations in output that are not caused by input
concentration changes.
Lower detectable limit: The minimum pollutant concentration that produces a signal of twice the noise level.
Interference equivalent: Positive or negative response caused by a substance other than the one measured.
Zero drift: The change in response to zero pollutant concentration during continuous unadjusted operation.
Span drift: The percent change in response to an upscale pollutant concentration during continuous unadjusted operation.
Lag time: The time interval between a step change in input concentration and the first observable corresponding change
in response.
Rise time: The time interval between initial response and 95% of final response.
Fall time: The time interval between initial response to a step decrease in concentration and 95% of final response.
Precision: Variation about the mean of repeated measurements of the same pollutant concentration, expressed as one
standard deviation about the mean.

Source: Code of Federal Regulations (1991a).
certification or with certification of limited acceptance.  There is no mechanism to provide accepted
certification for mixtures made in the laboratory. The EPA accepts only the first four types of gas mixtures
described below.

2.3.1  Pre-made Mixtures
2.3.1.1   Standard Reference Materials
       Calibration gas standards of CO in air (certified at levels of approximately 12, 23, and 46 mg/m3
or (10, 20, and 40 ppm, respectively) or in nitrogen (N2; 10 ppm to 13%) are obtainable from the Standard
Reference Material Program of the National Institute of Standards and Technology (NIST), formerly the


                                                2-3

-------
                                                   0 to 50 ppm (0 to 57 mg/m3)
                                                   0.05 ppm (0.06 mg/m3)
                                                    1.05 ppm (0.06 mg/m3)
0
     Table 2-2. Suggested Performance Specifications for Monitoring Carbon Monoxide in
	Nonurban Environments
 Range
 Noise
 Lower detectable limit
 Interference equivalent
   Each interfering substance
   Total interfering substances
 Zero drift
   12 h
   24 h
 Zero interval,3 maximum
 Span drift, 24 h
   20% of upper range limit
   80% of upper range limit
 Lag time
 Rise time
 Fall time
 Precision
   20% of upper range limit
   80% of upper range limit	
                                                   ±0.05 ppm (±0.06 mg/m3)
                                                   0.10 ppm (0.12 mg/m3)

                                                   ±0.1 ppm (±0.12 mg/m3)
                                                   ±0.1 ppm (±0.12 mg/m3)
                                                   Ih

                                                   ±5.0%
                                                   ±2%
                                                   1 min
                                                   5 min
                                                   5 min

                                                   0.2 ppm (0.24 mg/m3)
                                                   0.2 ppm (0.24 mg/m3)
aZero interval is the interval between measuring chemical zeros.

Source: Adapted from Code of Federal Regulations (1991a).
National Bureau of Standards, Gaithersburg, MD 20899.  These Standard Reference Materials (SRMs) are
supplied as compressed gas mixtures at about 135 bar (1,900 psi) in high-pressure aluminum cylinders
containing 800 L (28 ft3) of gas at standard temperature and pressure, dry (STPD) (National Bureau of
Standards, 1975; Guenther et al, 1996).  Each cylinder is supplied with a certificate stating concentration
and uncertainty.  The concentrations are certified to be accurate to ±1% relative to the stated values.
Because of the resources required for their certification,  SRMs are not intended for use as daily working
standards, but rather as primary standards against which  transfer standards can be calibrated.

2.3.1.2  National Institute of Standards and Technology Traceable Reference Materials
       Calibration gas standards of CO in air or N2, in the concentrations indicated above, are obtainable
from specialty gas companies.  Information as to  whether a specialty gas company supplies such mixtures
is obtainable from the specific company, or the information may be obtained from the Standard Reference
Material Program of NIST. These NIST Traceable Reference Materials (NTRMs) are purchased directly
from industry and are supplied as compressed gas mixtures at about  135 bar (1,900 psi) in high pressure
aluminum cylinders containing 4,000 L (140 ft3) of gas  at STPD.   Each cylinder is supplied with a
certificate stating concentration and uncertainty.  The concentrations  are certified to be accurate to within
±1% of the stated values (Guenther et al., 1996).
                                             2-4

-------
2.3.1.3  U.S. Environmental Protection Agency Protocol Gases
       Calibration gas standards of CO in air or CO in N2 at approximately the same concentrations as
SRMs and NTRMs can be purchased from specialty gas companies as EPA Protocol Gases. These gases
are blended and analyzed according to an EPA protocol document and are supplied as gas mixtures in high
pressure aluminum cylinders. These mixtures are supplied with certificates stating concentration and
uncertainty (U.S. Environmental Protection Agency, 1997).

2.3.1.4  Dutch Bureau of Standards
       Calibration gas standards of CO in air over a wide concentration range also can be purchased from
the Dutch Bureau of Standards, which is the Nederland Meetinstituut (MM) Holland (fax 31-15-261-2971).
These are Primary Reference Materials (PRMs) or Certified Reference Materials (CRMs). These Reference
Materials (PRMs or CRMs) are supplied as compressed gas mixtures at about 135 bar (1,900 psi) in high
pressure aluminum cylinders containing 800 L of gas at STPD. Each cylinder is supplied with a certificate
stating concentration and uncertainty.   The NIST  and  EPA  recognize the equivalency of specific NMi
standards with NIST standards on the strength of the NIST/NMi Declaration of Equivalency Document.

2.3.1.5  Commercial Blends
       Calibration gas mixtures of CO in air or N2 over a wide concentration range also can be purchased
commercially from many specialty gas companies. Some mixtures may have "certification" documentation
and some may not.  These mixtures can be ordered in cylinders of almost any size. Mild steel cylinders are
to be avoided (U.S. Environmental Protection Agency,  1991).
       The  nominal values for CO  concentration supplied  by the  vendor should be  verified by
intercomparison with an SRM or other validated standard sample. A three-way intercomparison has been
made among the NIST  SRM's, commercial gas blends, and an extensive set of standard gas mixtures
prepared by gravimetric blending  at EPA (Paulsell,  1976). Results of the comparison showed  that
commercial gas blends are within ±2% of the true value represented by a primary standard.  Another study
on commercial blends (Elwood,  1976) found poorer accuracy. To achieve compatible results in sample
analyses, different laboratories  should interchange and compare their respective working standards
frequently.

2.3.2  Laboratory Blended Mixtures
       Mixtures of CO in  almost any matrix gas can be blended in the laboratory.  One can start with
gaseous CO or mixtures of CO and dilute these to any concentration desired. The three common procedures
for blending mixtures into containers are the gravimetric (weighing) procedure, the manometric (pressure)
technique, and the volumetric method. One also can use dynamic dilution to prepare standards that are not
stored in containers but are used at the time of preparation. There are advantages and disadvantages to each
procedure, and one must evaluate the application, standards requirements, and laboratory equipment before
choosing the method of standards preparation.
       Standard samples of CO in air also can be prepared by flowing gas dilution techniques.  In a
versatile system designed for this purpose (Hughes et al., 1973), air at a pressure of about 0.7 to 7.0 bar
(about 10 to 100 psi) above ambient is first purified and dried by passage through cartridges of charcoal and
silica gel, then is passed through a sintered metal filter into a flow control and flowmeter system. The CO
(or a mixture of CO in air that is to be diluted further), also under pressure, is passed through a similar flow
control and flowmeter system.
       Dynamic dilution employed to make CO standards often relies on mass flow controllers.  When
performing a calibration with this technique, care should be taken to control the temperature and pressure
of the flow controllers. Investigations into the performance of several brands of mass flow controllers on
                                             2-5

-------
aircraft have revealed that, for large pressure changes, some instruments experience errors in the output well
beyond the specifications (Weinheimer and Ridley, 1990).

2.3.3 Other Methods
       Permeation tubes have been used for preparing standard mixtures of such pollutant gases as sulfur
dioxide and nitrogen dioxide (O'Keeffe and Ortman, 1966; Scaringelli etal., 1970). Permeation tubes are
not used routinely in the United States for making CO standard samples and are not recommended.  In the
permeation tube techniques, a sample of the pure gas under pressure is allowed to diffuse through a
calibrated partition at a defined rate  into a diluent gas stream to  give  a standard sample of known
composition.
       Another  possible way to liberate known amounts of CO  into  a diluent  gas is by thermal
decomposition of nickel tetracarbonyl [Ni(CO)4].  However, an attempt to use this as a gravimetric
calibration source showed that the relation between  CO  output and  weight loss of the Ni(CO)4 is
nonstoichiometric (Stedman et al, 1976).

2.3.4 Intercomparisons of Standards
       Initial efforts to establish the absolute uncertainty of CO standards and to put various research
groups around the world on the same scale revealed systematic errors in some of the standards. Careful
preparation of gas standards and repeated intercomparison of calibration gases and measurements on
ambient air since have led to general agreement within the international community  on both a reference
scale and on analytical methods. Calibration standards now generally agree to within 5%, and atmospheric
measurements made with a variety of analytical techniques agree to 10 ppb or better.
       The  National  Aeronautics  and  Space Administration (NASA), as  part  of the  Chemical
Instrumentation Test and Evaluation Project, intercompared a tunable diode laser spectroscopy (TDLS)
technique and several "grab"-sample gas chromatography-flame ionization detection (GC-FID) techniques
(Hoell et al., 1984, 1985).  Initial  results indicated a high degree of correlation  among the various
instruments, but agreement on the absolute concentration was only about 15%; differences were as large
as 38%. When the intercomparison was repeated (Hoell et al., 1987), calibration standards agreed within
95% confidence levels. Measurements of ambient air samples under actual field conditions demonstrated
agreement within experimental uncertainty (on the order of 10 ppb) for CO concentration ratios from 60 to
170 ppb.  When data from the various instruments were regressed, however, slopes  again differed from
unity by as much as 14%.
       Careful intercomparisons of calibration gases indicate that accurate and consistent standards can
be made. Hughes et al. (1991) compared primary gas standards of CO in N2 produced by NIST and the
National Physical Laboratory in the United Kingdom. These standards, prepared gravimetrically, contained
concentration ratios ranging from 10 ppm to 8%. In a blind intercomparison, the mean difference was
0.2%, well within the experimental uncertainty of the techniques. Novelli et al. (1991) gravimetrically
produced CO in zero air in the range of 25 to 1,000 ppb from both pure CO and a NIST SRM; they found
agreement to within 1%. Agreement with commercially available NIST-traceable standards was within 3%.
Reasonable consistency (6% or better) was found with standards used by Australian, German, Brazilian,
and several American institutions. One Australian standard was found to be 22% lower, although trouble
with this standard had been reported previously (Weeks et al., 1989). A reevaluation of the reference scale
in the range of nonurban ambient concentrations (Novelli et al., 1994) confirmed agreement to within 5%
or better for the National Oceanic and  Atmospheric Administration, NASA, and German groups.
       Intercomparisons of TDLS and NDIR GFC techniques (Poulida et al., 1991; Fried et al.,  1991)
indicated  agreement within experimental  uncertainty  (better than  10%  for typical tropospheric
concentrations of 100 to 1,000 ppb), when NIST-based standards were used to calibrate both instruments.
These experiments demonstrated good agreement in ambient and compressed air.  These results, as well

                                             2-6

-------
as results from spiking tests, indicated no significant interferences in either monitor. The intercomparisons
also established linearity for both techniques in the range from 100 ppm to 10 ppb.
       Recent standards normalization and intercomparisons of TDLS, mercury liberation, GC-FID, and
NDIR techniques are described by Novelli et al. (1998).  For concentration ratios down to the lowest
expected in the boundary layer, about 50 ppb, agreement among groups was typically better than 10 ppb;
for higher mixing ratios the typical agreement was about 5%.

2.3.5  Infrared Absorption
       The TDLS can provide an independent measurement of the concentration of a CO standard. Fried
et al. (1991) used the high-resolution transmission molecular absorption database for the line parameters
to calculate the concentration based on direct absorption. Their results agreed with a NIST-certified gas
standard to within 1.6%, well within the uncertainty of the absorption measurement.
2.4  Measurement in Ambient Air
       This section discusses several important aspects of the continuous and intermittent measurement
of CO in the atmosphere, including sampling techniques and schedules and recommended analytical
methods for CO measurement.

2.4.1  Sampling System Components
       Carbon monoxide monitoring requires a sample introduction system, an analyzer system, and a data
recording system. A sample introduction system consists of a sampling probe, an intake manifold, tubing,
and air movers.  This system is needed to collect the air sample from the atmosphere and to transport it to
the analyzer without altering the original concentration. It also may be used to introduce known gas
concentrations to periodically check the reliability of the analyzer output. Construction materials for the
sampling probe, intake manifold, and tubing should be tested to demonstrate that the test atmosphere
composition or concentration is not altered significantly.  It is recommended that sample introduction
systems be  fabricated from  borosilicate glass or fluorinated  ethylene propylene (Teflon®) if several
pollutants are to be monitored (Code of Federal Regulations, 1991b). However, in monitoring for CO only,
it has been reported (Wohlers et al., 1967) that no measurable pollutant losses were observed at the high
(>1 L/min)  sampling  flow  rates when  sampling systems  were constructed of tygon,  polypropylene,
polyvinylchloride,  aluminum, or stainless steel  piping.  The sample introduction  system should  be
constructed so that it presents no pressure drop to the analyzer. At low flow and low concentrations, such
operation may require validation.
       The analyzer system consists of the analyzer itself and any sample-preconditioning components that
may be necessary.  Sample preconditioning might require a moisture control system such as a Nafion®
drying tube to help minimize the false positive response of the analyzer (e.g., the NDIR analyzer) to water
vapor and a particulate filter to help protect the analyzer from clogging and possible chemical interference
caused by particulate buildup in the sample lines or analyzer inlet. The sample preconditioning system also
may include a flow metering and control device to control the sampling rate to  the analyzer.

2.4.2  Quality Assurance Procedures for Sampling
       The accuracy and validity of data collected from a CO monitoring system must be ensured through
a quality assurance program. Such a program consists of procedures for calibration, operational and
preventive maintenance, data handling, and auditing; the procedures should be documented fully in a quality
assurance program manual maintained by the monitoring organization.
       Calibration procedures consist of periodic multipoint primary calibration and secondary calibration,
both of which are prescribed to minimize systematic error.  Primary calibration involves the introduction

                                             2-7

-------
of test atmospheres of known concentration to an instrument in its normal mode of operation for the purpose
of producing a calibration curve.
       A calibration curve is derived from the analyzer response obtained by introducing several successive
test atmospheres of different known concentrations.  One recommended method for generating CO test
atmospheres is to use air containing no CO along with several known concentrations of CO in air or N2
contained in high-pressure gas cylinders and verified by NIST-certified SRMs wherever possible (Code of
Federal Regulations, 1991 a). The CO can be removed from an air stream by oxidation to CO2 on a catalyst
(Dickerson and Delany, 1988; Parrish et al, 1994).  The number of standard gas mixtures (cylinders)
necessary to establish a calibration curve depends on the nature of the analyzer output.  A multipoint
calibration  at five or six different  CO concentrations covering the operating range of the analyzer is
recommended by EPA (Code of Federal Regulations, 1991b; Federal Register, 1978).  Alternatively, the
multipoint calibration is accomplished by diluting a known high-concentration CO standard gas with zero
gas in a calibrated flow dilution system.
       Secondary calibration consists of a zero and upscale span of the analyzer.  This is recommended
to be performed daily (Federal Register, 1978). If the  analyzer response differs by more than 2% from the
certified  concentrations, then the analyzer  is adjusted  accordingly.  Complete records  of  secondary
calibrations should  be kept to aid in data reduction and for use in auditing.   For  high-sensitivity
measurement, hourly zeros and weekly calibrations are recommended.
       Specific criteria for data selection and several  instrument checks are available (Smith and Nelson,
1973). Data recording involves recording in a standard format for data storage, interchange of data with
other agencies, or data analysis.  Data analysis and interpretation usually include a mathematical  or
statistical analysis of air quality data and a subsequent  effort to interpret results in terms of exposure
patterns,  meteorological conditions, characteristics of emission sources, and  geographic and topographic
conditions.
       Auditing procedures consist of several quality control checks and subsequent error analyses to
estimate the accuracy and precision of air quality measurements. The quality control checks for CO include
data processing, control sample, and water vapor interference checks, all of which should be performed by
a qualified individual independent of the regular operator. The error analysis is a statistical evaluation of
the accuracy and precision of air quality data. Guidelines have been published by EPA (Smith and Nelson,
1973) for calculating an overall bias and standard deviation of errors  associated with data processing,
measurement of control samples, and water vapor interference,  from which the accuracy and precision of
CO measurements can be determined. Since January  1, 1983, all state and local agencies submitting data
to EPA must provide estimates of accuracy and precision of the CO  measurements based on primary and
secondary calibration records (Federal Register, 1978). The precision and accuracy audit results through
1985  indicate  that the  95% national probability limits  for precision  are ±9%,  and the 95% national
probability  limits for accuracy are within ±1.5% for all audit levels up to 85 ppm.  The results (accuracy)
for CO exceed comparable results for other criteria pollutants with national ambient  air quality standards
(Rhodes and Evans,  1987).

2.4.3  Sampling Schedules
       Carbon monoxide concentrations in the atmosphere exhibit  large temporal variations because of
changes  in the time and  rate that CO  is  emitted  by  different sources and because of changes  in
meteorological conditions that govern the amounts of transport and dilution that take place.  During a 1 -year
period, an urban CO station may monitor hourly concentrations of CO ranging from  below the minimum
detection limit to as high as 45 ppm (52  mg/m3). The NAAQS for CO are based on the second highest
1- and 8-h average concentrations; violations represent extreme events.  In order to  measure the highest
two values from the  distribution of 8,760  hourly values in a year, the "best" sampling schedule to employ
is continuous monitoring 24 h per day, 365 days per year.  Even so, continuous monitors rarely operate for

                                              2-8

-------
long periods without data losses because of malfunctions, upsets, and routine maintenance. Data losses of
5 to 10% are common.  Consequently, the data must be interpreted in terms of the likelihood that the
NAAQS were attained or exceeded. Statistical methods can be employed to interpret the results (Garbarz
etal,  1977;Larsen, 1971).
       Compliance with 1- and 8-h NAAQS requires continuous monitoring. Statistically valid sampling
could be performed on random  or  systematic  schedules, however, if annual averages or relative
concentration levels were of importance.  Most investigations of various sampling schedules have been
conducted for particulate air pollution data (Hunt, 1972 ;Ott and Mage, 1975 ;Phinney and Newman, 1972),
but the same schedules also could be used for CO monitoring. However, most instruments do not perform
reliably in intermittent sampling.

2.4.4  Continuous Analysis
2.4.4.1  Nondispersive Infrared  Photometry
       Carbon monoxide has a characteristic infrared absorption near 4.6 |^m.  The absorption of infrared
radiation by the CO molecule therefore can be used to measure CO concentration in the presence of other
gases.  The NDIR method is based on this principle.
       Nondispersive infrared systems have several  advantages. They are not sensitive to flow rate,  they
require no wet chemicals, they are reasonably independent of ambient air temperature changes, they are
sensitive over wide concentration ranges, and they have short response times. Further, NDIR systems  may
be operated by nontechnical  personnel.   Gas filter correlation spectroscopy analyzers are used most
frequently now in documenting compliance with ambient air standards.

Gas-Filter  Correlation Spectroscopy
       A GFC monitor (Burch et al.,  1976) has
the advantages of an NDIR instrument and the
additional  advantages  of smaller  size,  no
interference  from  CO2,  and  very   small
interference from water vapor. A top schematic
view of the  GFC monitor is shown in Figure 2-1,
showing the components of the optical path for
CO detection.   During operation,   air  flows
continuously through the sample cell.  Radiation
from the source  is directed by optical transfer
elements   through   the  two  main  optical
subsystems: (1) the rotating gas filter and (2) the
optical multipass (sample) cell. The beam exits
the sample  cell through interference filter (FC),
which limits the spectral passband to a few of the
strongest CO  absorption lines  in the 4.6-^m
region.  Detection of the transmitted radiation
occurs at the infrared detector (C).
       The gas correlation cell is constructed
with two compartments:  one compartment is
filled   with 0.5   atm  CO,   and  the  second
compartment is filled with pure N2.   Radiation
transmitted through the  CO is completely attenuated at wavelengths where CO absorbs strongly.  The
radiation transmitted through  the N2  is reduced by  coating the exit window of the cell with  a neutral
                       Cell Containing N2 Neutral Attentuator
          Chopper
                 Cell Containing
                 Carbon Monoxide
Figure 2-1.  Schematic diagram of GFC monitor for CO.
A = optical layout (M denotes mirror reflector, and L denotes
lens); B = detail of correlation cell.
Source: Chaney and McClenny (1977).
                                              2-9

-------
attenuator so that the amounts of radiation transmitted by the two cells are made approximately equal in
the passband that reaches the detector.
       In operation, radiation passes alternately through the two cells as they are rotated to establish a
signal modulation frequency. If CO is present in the sample, the radiation transmitted through the CO is
not appreciably changed, whereas that through the N2 cell is changed. This imbalance is linearly related
to CO concentrations in ambient air.

Enhanced Performance
       Although commercial CO monitors were designed to meet the performance specifications shown
in Table 2-1,  several instruments have the potential for much greater sensitivity.  Modifications of
commercially available NDIR monitors (Dickerson and Delany, 1988; Parrish et al., 1994) have been made
to enhance their performance, but the manufacturers have continued to improve instruments and offer
"high-sensitivity" options that could meet the requirements of monitoring clean continental air (i.e., a
detection limit of about 50 ppb and resolution of 10 ppb).
       The principal constraints on the lower detectable limits of commercially available NDIR CO
monitors are detector noise, water vapor interference, and drift in the background. Several methods have
been developed by researchers to improve detector noise, such as cooling the preamplifier and improving
the optics. More recent improvements made by the manufacturers, such as gold-coated mirrors and selected
infrared (IR) radiation detectors  have been effective in reducing detector noise.
       Water vapor produces a negative artifact such that a volume mixing ratio of 1% would reduce
apparent CO mixing ratio measurement by 50 ppb. This interference can be reduced to within tolerances
by drying the sample air with  a cold trap, desiccant, or drying tube (Dickerson and Delany, 1988).
Alternatively, the zero can be checked frequently enough so that changes in ambient humidity are unlikely
to produce a significant error (Parrish et al., 1994).
       The greatest source of potential error in monitoring CO in the 0.1-ppm range is background drift.
The stability of the instruments with respect to changes in calibration (span) is adequate, but the background
(zero) drifts on time scales of minutes to hours in response to, among other factors, instrument temperature.
This drift can be accounted for most easily by frequent chemical zeroing with a oxidizer that converts CO
to CO2.

2.4.4.2  Gas Chromatography-Flame lonization
       Carbon monoxide can be measured in either ambient air samples collected every few minutes or
in air from grab samples stored under pressure in inert canisters. Carbon monoxide in air samples is dried,
preconcentrated, reduced to methane, and detected by flame ionization (GC-FID) (Heidt, 1978; Greenberg
et al., 1984; Hoell et al., 1987). Uncertainty on the order of 10 ppb or 10% of the observation can be
obtained routinely.

2.4.4.3  Mercury Liberation
       This technique, involving reaction with hot mercuric oxide to produce elemental mercury vapor,
was developed early  this century (Moser and Schmid,  1914;Beckmanetal., 1948;McCulloughetal, 1947;
Mueller, 1954; Palanos, 1972; Robbins et al.,  1968) and is now  available  commercially (e.g., Trace
Analytical Inc., Menlo Park, CA). The method is temperature and pressure sensitive, and operation in the
continuous mode requires elimination of interferences from sulfur dioxide, hydrogen, and hydrocarbons
(Seiler et al., 1980).   Successful  continuous operation has been reported with response time on the order
of 20 s and detection limits near 20 ppb (Fishman et  al., 1980; Brunke et al., 1990).
       As a GC detector, mercury liberation  (GC-ML) offers high  sensitivity, without the interferences
inherent in continuous measurements (e.g., Novelli et al., 1991, 1998). Air samples are collected in glass
bottles and injected into a gas  chromatograph  with two  columns.  The CO is then detected with a


                                             2-10

-------
commercial mercuric oxide reduction detector (e.g., Trace Analytical Inc., Menlo Park, CA). The system
is linear from  10 ppb  to more than 1,000 ppb, has a detection limit below 10 ppb, and the reported
uncertainty is about 2%.

2.4.4.4  Tunable Diode Laser Spectroscopy
       Tunable diode lasers (TDLs) produce IR radiation with a line width that is narrow compared with
typical absorption lines of atmospheric trace gases.  Absorption of IR radiation by a single rotational line
in the 4.6-^m band can be exploited to measure CO with high precision and rapid response, and without
interferences; the sharp focus on a narrow spectral region provides great selectivity.  Air samples are
measured over open paths through the ambient air (Chaney et al, 1979) or by pulling air samples through
an orifice into a long-path cell maintained at a pressure well below ambient (Sachse et al., 1987; Fried et al.,
1991; Roths et  al. , 1 996). Radiation from a TDL is modulated over a very narrow wavelength region such
that absorption by CO  produces an alternating current signal.  The  background is measured by catalytic
oxidation of CO to CO2.
       Instruments based on TDLS are the fastest and most sensitive extant, with a typical detection limit
of a fewparts per billion and a response time of a few seconds. For long-term monitoring, the high cost and
need for a skilled operator on site are disadvantages.

2.4.4.5  Resonance Fluorescence
       Resonance fluorescence of CO in the vacuum ultraviolet has been used for a highly sensitive and
rapidly responding instrument (Volz and Kley, 1985; Gerbig et al., 1996).  Excitation is represented by the
following reaction:
                            CO(X1I)+hv^ COCX1!!).                            (2-1)
Atmospheric CO absorbs radiation in the 150-nm range from a radio frequency discharge lamp, and
fluorescence from the excited CO is detected by a photomultiplier tube. The lamp generates a plasma in
a continuous flow of CO2 in argon. Limits to the sensitivity of this instrument are set by interference from
water vapor, continuum Raman scattering by oxygen, and by drift in the lamp intensity. The pressure in
the fluorescence chamber must be maintained between 7 and 9 mbar air to balance  interference from
oxygen and the signal from CO.
       Recent improvements (Gerbig et al., 1999) have reduced the detection limit to 3 ppb for a response
time as short as a few seconds.  The high  sensitivity and small size of the instrument are desirable for
measurements from aircraft.  Before the instrument is practical for air pollution monitoring, its stability
must be improved.  As the  lamp window degrades, sensitivity is lost, such that,  after about 200 h of
operation, loss in the span of a factor of two can be expected.

2.4.5 Intercomparisons of Methods
       Several techniques (TDLS, NDIR/GFC, GC-FK), and GC-ML) have been evaluated in rigorous
intercomparisons under field  conditions.   For unpolluted tropospheric air, a number of instruments
employing different analytical principles consistently have measured concentration ratios that agree within
experimental uncertainty (Hoell et al., 1987; Fried et al., 1991; Poulida et al., 1991; Novelli et al., 1998).

2.4.6 Other Methods of Analysis
       Color changes induced by reaction of a solid or liquid date back to Haldane (1 897-1 898) and were
reviewed extensively in the previous criteria document. Examples include the colored  silver sol method,
the NIST colorimetric indicating gel, the length-of-stain indicator  tube,  and frontal analysis (U.S.
Environmental Protection Agency, 1991).


                                             2-11

-------
       More recently developed electrochemical  techniques show highly  improved resolution and
specificity (e.g., Langan, 1992; Lee et al, 1992a; Ott et al, 1995).  Electrochemical sensors operate by
measuring the current of a small fuel cell and, because of their reduced size and power requirements, have
been used extensively in exposure and indoor research studies (see Section 2.5).  Precision of 0.2 to 2 ppm
has been reported.  Further independent evaluation and intercomparison, followed by publication in the
reviewed scientific literature, is called for to  determine the sensitivity,  stability, and  selectivity of
electrochemical methods to establish equivalency to the NDIR instrument by EPA for use in compliance
monitoring.
2.5  Measurement Using Personal and Remote Monitors
2.5.1  Personal Monitors
       Monitors at fixed locations provide useful information on ambient CO concentrations and their
variability and trends, but such monitors cannot measure personal  exposure.  Information on personal
exposure, including home, in-transit, and work-related concentrations is needed for epidemiologic studies.
The previous criteria document (U.S. Environmental Protection Agency, 1991) reviewed the state of the
science of personal monitors as of about 1986.  Since that time, the devices have been further developed
and refined.
       One  technique involves an ion-exchange Y-type zeolite, with zinc ion as the adsorbent.  The
adsorbent is desorbed thermally, converted to methane, and analyzed using GC-FID (e.g., Lee et al.,
1992b,c; Lee and Yanagisawa, 1992,1995). Apte (1997) reviewed several of these devices and described
passive samplers based on transition metal compound color changes measured spectrochemically. The
method has  an interference with ethylene, inconsequential  in most  microenvironments, but provides
adequate performance (sensitivity  of 10 ppm/h and precision of ±20%  or better) for health studies.
Substantial work remains for most passive samplers on stability and response to change in temperature,
humidity,  and interferences.  These passive  techniques lack the  response  speed and sensitivity for
compliance ambient air monitoring or short time-scale personal monitoring, but they have been field tested
and found adequate for occupational CO exposure studies where a longer 8-h average CO value is measured
for comparison to an 8-h Occupational Safety and Health Administration standard (Apte et al., 1999).
       Numerous field studies on personal  exposure and microenvironmental sampling have been
conducted with electrochemical sensors (Akland et al., 1985; Ott et al., 1986; Wallace et al., 1988; Langan,
1992; Ott et al., 1994; Klepeis et al., 1999; McBride et al., 1999); some are described in Chapters 3 and 4.
Over the past 15 years, the electrochemical personal CO monitors have been much improved, and the latter
studies show the increased versatility of the electrochemical sensor for field sampling.  These studies
documented the spatial and temporal variability of CO concentrations in an urban area at locations away
from a central monitoring site and showed how the effects of sources of CO in microenvironments can add
a major increment to a person's CO exposure, as estimated only from compliance measurements of CO at
a central ambient monitoring station.

2.5.2  Remote Monitor
       Carbon monoxide emissions from vehicles can be measured rapidly  with a remote sensing IR
technique (Bishop et al., 1989; Bishop and Stedman, 1996) in which CO is measured near 4.3 /^m and CO2
at 3.6 /^m; a third wavelength is used as a reference for intensity. The instrument has been evaluated in a
double blind intercomparison with on-board NDIR, and the two methods agreed well within experimental
uncertainty (Lawson et al., 1990). Surveys conducted with this technique reveal that the majority of CO
is emitted by a minority of vehicles.  The method has been used to evaluate the efficacy of inspection and
maintenance programs and oxygenated fuels (Beaton et al., 1995; Stedman et al., 1997, 1998). The sum
                                            2-12

-------
of measurements indicates a general decrease in fleet-averaged CO emissions over the past decade (Bradley
etal., 1999).
2.6  Biological Monitoring
       A unique feature of CO exposure is that there is a biological marker for the recent dose that an
individual has received—the blood level of CO.  This level may be calculated by measuring  blood
carboxyhemoglobin (COHb) or by measuring CO in end-tidal exhaled breath after a standardized breathhold
maneuver, with a required correction for the background CO inhaled prior to a breathhold (Smith,  1977;
Wallace, 1983). The measurement methods for COHb and breath CO were reviewed extensively in the
previous criteria document (U. S. Environmental Protection Agency, 1991). This section provides an update
on advances in analytical methods for measuring blood COHb and breath CO that have been published in
the literature since the previous review.  New studies reporting breath CO or blood COHb in population
studies are discussed in Chapter 4, along with other new CO exposure assessments.

2.6.1  Carboxyhemoglobin Measurements
       Direct reading of COHb usually is performed in the clinical or hospital  setting through the  use of
a direct-reading spectrophotometer, such as a CO-Oximeter (CO-Ox). For clinical purposes, precision on
the order of ±1% COHb  is not of primary importance, because of the need to  differentiate between
conditions of low levels of COHb and the much higher levels of COHb that indicate treatment for CO
poisoning. The concern in this setting, for example, is to rapidly distinguish between 1 and 10% COHb,
not between 1  and 2% COHb.  Marshall et al. (1995) showed a wide range of threshold COHb values
(measured in the blood by CO-Ox, not estimated from breath CO) used to determine treatment in a sample
of 23 Boston, MA, area laboratories. For example, eight laboratories accepted values of 5 to 6% COHb as
normal in nonsmokers, a value that cannot be supported by the modern scientific literature. The authors
recommended the use of threshold limits of 3% COHb for nonsmokers and 10% COHb for smokers when
classifying subjects for treatment.
       The performance of the various early versions of the CO-Ox instruments for measuring blood COHb
was reviewed in Section 8.5 of the previous criteria document (U.S. Environmental Protection Agency,
1991).   These and later instruments, of different design from different manufacturers, used several
wavelengths of light for simultaneous measurement of hemoglobin (Hb), oxyhemoglobin (O2Hb), COHb,
and methemoglobin (Freeman and Steinke, 1993; Gong, 1995; Bailey et al., 1997). Vreman et al. (1993)
and Mahoney et al.  (1993) confirmed that considerable difficulties  were encountered for COHb
concentrations below 5% (a region with which most environmental studies of nonsmokers are concerned),
and the authors concluded that the CO-Ox is unreliable for environmental studies.  Some versions of the
CO-Ox also were found to be influenced by bilirubin and by fetal hemoglobin (Hb), presenting difficulty
in diagnosing newborn infants with j aundice (Vreman and Stevenson, 1994; Stevenson and Vreman, 1997).
Shepherd and McMahan (1996) present a highly detailed analysis of the causes and effects of oximeter
errors in blood gas analyses.
       Recent CO-Ox developments have been  a new six-wavelength instrument (Instrumentation
Laboratory, 1999) used by Kimmel et al. (1999) and a 128-wavelength instrument (Krarup, 1998) both of
which identify and correct for possible interferences. The latter instrument is still under formal independent
evaluation, and,  although peer-reviewed published results of comparison testing  are expected  to be
forthcoming shortly, the only article currently in press is in German, in a non-peer-reviewed journal
(Krarup, 1999). It is possible that comparison of the results on the same sample using  the new  multi-
wavelength instruments and older instruments with fewer wavelengths may show that the new instruments
measure lower COHb if they better correct for the positive interferences of various non-COHb species in
the blood, such as varying fractions of fetal hemoglobin and sulfhemoglobin. This would be consistent with

                                            2-13

-------
the report that some laboratories, as cited above, accepted 5 to 6% COHb from oximeter readings as normal
for nonsmokers.
       For a research study to relate health effects or breath CO to COHb, the method of choice is GC
analysis of the CO gas released from the blood when COHb is dissociated (U.S. Environmental Protection
Agency, 1991; Van Dam and Daenens, 1994; Lloyd and Rowe, 1999).  The reader, therefore, is alerted to
the difference between end-tidal breath CO to blood COHb relationships when the COHb is determined by
CO-Ox or GC. A calibration curve relating exhaled end-tidal breath CO to COHb should be based on a
standard breath-hold maneuver for the CO collection and the GC method of COHb analysis. It is beyond
the scope of this chapter to reanalyze the early COHb literature and estimate the effect of the possible
positive interferences that were not accounted for by the early CO-Ox instruments because each instrument
and its on-site calibration procedure would create a different bias that cannot be known with certainty.
However, in general, the levels of COHb associated with low levels of ambient CO exposure in field studies
may have been overestimated in the past.

2.6.2  Breath Carbon Monoxide Measurements
       Carbon monoxide in the breath can be measured by all  techniques used to measure ambient CO
concentrations, as described in the previous criteria document  (U.S. Environmental Protection Agency,
1991).  A common type of instrument in use for rapidly screening  large numbers of people for CO
exposures or measuring breath CO distributions is the electrochemical analyzer. The subject performs an
inhalation-breathhold maneuver and exhales through a mouthpiece into the instrument inlet.  The end-tidal
breath is retained for analysis, and the reading in parts per million of CO can be converted to COHb through
a calibration curve or nomogram provided with  the instrument.
       Vreman et al. (1993) presented evidence to show that a serious  positive interferent in the
electrochemical method (hydrogen gas) is  present in the exhaled breath of some persons as a result of
metabolism of certain foods. Because this could have affected many previous studies, including the very
large EPA studies in Washington, DC, and Denver,  CO (Akland et al.,  1985), it would be desirable to
determine the fraction of the population so affected. Because of the general decline of ambient CO, this
potential interference takes on more importance in any future studies, which must account for this problem
if employing electrochemical devices to measure breath CO.
       Lee et al. (1991) developed a TDLS system that was well suited for measuring low levels of CO in
breath. The system also can detect the abundance of isotopic CO (13C16O), with a preliminary finding of
a slight enrichment over atmospheric abundance in breath. Lee et al. (1994) employed  the instrument in
a study correlating breath CO and blood COHb in people living near Boulder, CO (described in Chapter 4).
       The passive CO sampler developed by Lee and Yanagisawa (1992, 1995) (see Section 2.5) has a
reusable sampling system that allows the collection of only the last 5 mL  of a breath expelled after breath
holding for 20 s, thus obtaining alveolar air undiluted by dead space air.  The sampler was unaffected by
humidity; however, the rather low efficiency of collection (50%) and the resulting fairly high detection limit
of 3.2 ppm may limit the utility of the sampler for environmental studies.

2.6.3  Relationships of Breath Carbon Monoxide to Blood Carboxyhemoglobin
       The end-tidal breath CO versus COHb relationships reviewed in the previous criteria document (see
Table 8-14 in U.S. Environmental Protection Agency,  1991) and in studies published in the literature since
then are often at variance because they use either a 10-, 15-, or 20-s breathhold step in the breath collection;
use either GC or CO-Ox for the blood COHb measurements; or  may not correct for the CO content of the
inhaled air. The use of a 20-s breathhold, as recommended by Jones et al.  (1958), with a  correction for the
CO content of the inhaled air (Smith, 1977; Wallace, 1983), would improve the reproducibility of the CO
breath measurements,  and  the use of GC would  improve the accuracy  of the corresponding COHb
measurement. The 20-s breathhold is preferable, because it maximizes the approach to equilibrium and

                                            2-14

-------
                                                10-
                                                       • Oh
                                                       • 1 h, 9 ppm CO
                                                       A 8 h, 9 ppm CO
                                                       r = 0.945
minimizes the magnitude of the required correction for CO in inhaled air.  Therefore, specific details
regarding the length of the breathhold, corrections for inhaled CO, and the method of COHb analysis should
be provided in the published discussions of studies of the CO-COHb relationship so that differences among
study results can be evaluated.
       One comprehensive review article on CO-COHb relationships (Vreman  et al.,  1995) discusses
physical and chemical properties, endogenous and exogenous sources of CO, body burden and elimination,
toxicity and treatment, clinical chemistry, measurement methods, and the relationship of CO and COHb
to bilirubin and jaundice in neonates.  A second, less comprehensive review from the same investigators
focuses on the production  of CO  and bilirubin in equal amounts  by heme degradation and  on the
physiological significance of CO as aneuronal messenger (Rodgers et al., 1994).
       Lee et al. (1994) performed a study of
CO-COHb relationships at altitude in Boulder.
A total of 13 nonsmoking adults were exposed
to 9 ppm CO for both  1 and 8 h.  Blood was
sampled  and end-tidal  breath samples were
taken after a 10-s breathhold.  Mean  COHb
values prior to exposure were 0.65% and, after
exposure for the 1 - and 8-h periods were 1.2 and
2.2%, respectively.  The corresponding mean
CO levels in the breath samples were about 2.4,
4.4, and 8.2 ppm (uncorrected for the ~0 ppm
ambient CO  in  inhaled air), respectively, as
shown in Figure 2-2. The slope of 3.65 ppm per
1% COHb saturation after a 10-s breathhold at
altitude is  somewhat  smaller than  previous
estimates of about 5 ppm CO per 1% COHb, but
the  previous  estimates  were based on  a 20-s
breathhold  near  sea level that maximizes  the
end-tidal CO  if the inhaled air had  a CO
concentration below the 20-s end-tidal breath
CO (Jones et al., 1958).
                                              Q.
                                              3  6H
                                              O
                                              o
                                              0)
                                              m
                                                 2-
                                                  0123
                                                             Blood Carboxyhemoglobin (%)

                                              Figure 2-2. The correlation between an end-tidal breath CO
                                              concentration after a 10-s  breathhold and blood  COHb
                                              levels expressed as individual data points, as well as mean
                                              ± standard deviation.  The  breath  concentration was not
                                              corrected for the concentration of CO in the inhaled air (Smith,
                                              1977; Wallace, 1983).
                                              Source:  Lee et al. (1991, 1994).
2.6.4  Summary of the Relationship Between Biological Measurements of
       Carbon Monoxide
       The use of CO-Ox to measure COHb provides useful information regarding values of COHb in
populations being studied for clinical diagnosis. However, the range of COHb values obtained with this
optical method for blood collected from nonsmokers is greater than that obtained from a split sample
analyzed for COHb by research laboratory GC.  Therefore, the greater potential exists with the CO-Ox for
having an incorrect absolute value for COHb, as well as an incorrectly broadened range of values, when
used in population studies. In addition, it is not clear exactly how sensitive the CO-Ox techniques are to
small changes in COHb at the low CO  end of the COHb dissociation curve. Interferences (e.g., from
variable levels of oxygen saturation of hemoglobin [O2Hb]) and nonlinear phenomena appear to have a very
significant influence on the COHb reading at low COHb concentrations in a sample, suggesting nonlinearity
or a disproportionality in the absorption  spectra of different species of Hb (e.g., HbA [adult],HbF [fetal],
HbS [sickle],  HbZH [Zurich]). Gas chromatography continues to be the method of choice for measuring
COHb in a research setting, although, with care, a CO-Ox can be specially calibrated by GC analysis of
calibration-standard blood samples prepared with low COHb concentrations (Allred et al., 1991).
                                             2-15

-------
       The measurement of exhaled breath CO has the advantages of ease, speed, precision (provided the
required correction for CO in the inhaled air is made), and greater subject acceptance than the invasive
measurement of blood COHb.  Breath CO measurement on randomly chosen people can be related to the
blood COHb by  use of an empirical relationship developed by  simultaneous measurements of COHb
(preferably by GC) and breath CO, using the identical procedure for the breath collection that is used in the
population study. The empirical relationships developed with different breath holding techniques will differ
from the theoretical Haldane equilibrium relationship for the reaction CO + O2Hb <->• O2 + COHb, which
depends on the ratio of adult- to fetal-hemoglobin (HbA:HbF).  This is because the Haldane relationship
is for in vitro static equilibrium, and the empirical end-tidal breath CO-blood COHb relationship is for an
in vivo dynamic equilibrium that depends on how long the breath is held and on the correction for the CO
in inhaled air.
2.7  Summary
       The review of the state of the science for this criteria document yields several major points
concerning analytical techniques for CO measurement.
       Several adequate techniques exist for highly reliable monitoring of CO to ensure compliance with
the NAAQS.  Determination of the actual mean ambient air concentration requires substantially better
performance than does the minimum required to demonstrate compliance with the NAAQS. Commercial
instruments, sometimes with minor modifications, can meet the measurement needs for supplying useful
data on the emission, distribution, and trends of ambient CO and for modeling photochemical smog.
       Use of enhanced instruments for monitoring of actual CO concentrations with reasonable precision
is  needed if CO levels in  clean continental air  outside  of urban environments are  to be quantified
adequately. Commonly used calibration standards and measurement techniques have in the past failed to
meet the criteria of precise measurement, but there is now general agreement on procedures for generating
standards with absolute  accuracy better than about 2% in the parts per million range and about 10% in the
range of mixing ratios found in the clean troposphere. Compressed air mixtures, traceable to NIST or NMi,
provide reliable means of precise calibration.
       The NDIR, GC-ML, GC-FTD, and TDL techniques have undergone careful evaluation with synthetic
air mixtures and ambient air, and are deemed reliable.  The methods were intercompared in both open and
blind studies with designated "disinterested, third-party" referees. Early problems were identified and
corrected, and the most recent intercomparisons indicate general agreement on calibration standards and
ambient air measurements over a broad range of concentrations.  New techniques should undergo the same
rigorous evaluation.
       Several new  electrochemical  and passive sampling methods have become available.  These
techniques are not yet equivalent to the NDIR method for compliance monitoring or precise enough for CO
measurements in background ambient air (<0.5 ppm CO) for inputs to CTM, but they are very useful for
personal  exposure and  indoor  research studies.  Further work  on the stability and  specificity of the
electrochemical methods to obtain EPA equivalency is warranted.
       The level of COHb in the blood may be determined directly by blood analysis or indirectly by
measuring CO in exhaled breath. The use of CO-Ox to measure COHb can provide useful information
regarding mean values in populations being studied or as an aid in clinical diagnosis. It has been shown,
however, that the range of values obtained with this optical method will be greater than that obtained with
other more accurate methods, especially at COHb levels <5%.  Gas chromatography continues to be the
method of choice for measuring COHb.
                                             2-16

-------
        The measurement of exhaled breath has the advantages of ease, speed, precision, and greater subj ect
acceptance  than measurement of blood COHb.   However, the  accuracy of the  breath  measurement
procedure and the validity of the in vitro Haldane relationship between breath and blood still remains in
question, especially at low environmental  CO concentrations.
References
Akland, G. G.; Hartwell, T. D.; Johnson, T. R.; Whitmore, R. W. (1985) Measuring human exposure to carbon monoxide in
       Washington, B.C., and Denver, Colorado, during the winter of 1982-1983. Environ. Sci. Technol. 19:911-918.
Allred, E. N.; Bleecker, E. R.; Chaitman, B. R; Dahms, T. E.; Gottlieb, S. O.; Hackney, J. D.; Pagano, M.; Selvester, R. H.;
       Walden, S. M.;  Warren, J. (1991) Effects of carbon monoxide on myocardial ischemia. Environ. Health Perspect.
       91:89-132.
Apte, M. G.  (1997) A population-based exposure assessment methodology for carbon monoxide: development of a carbon
       monoxide passive sampler and occupational dosimeter [dissertation]. Berkeley, CA: U. S. Department of Energy, Lawrence
       Berkeley National Laboratory, Environmental Energy Technologies Division; LBNL-40838.
Apte, M. G.; Cox, D. D.; Hammond, S. K.; Gundel, L. A. (1999) A new carbon monoxide occupational dosimeter: results from
       a worker exposure assessment survey. J. Exposure Anal. Environ. Epidemiol. 9: 546-559.
Bailey, S. R.; Russell, E. L.; Martinez, A. (1997) Evaluation of the AVOXimeter: precision, long-term stability, linearity, and use
       without heparin.  J. Clin. Monit. 13: 191-198.
Beaton, S. P.; Bishop, G. A.; Zhang, Y.; Ashbaugh, L. L.; Lawson, D. R.; Stedman, D. H. (1995) On-road vehicle emissions:
       regulations, costs, and benefits. Science (Washington, DC) 268: 991-993.
Beckman, A. O.; McCullough, J. D.; Crane, R. A. (1948) Microdetermination of carbon monoxide in air: a portable instrument.
       Anal. Chem. 20: 674-677.
Bergman, I.; Coleman, J. E.; Evans, D. (1975) A simple gas chromatograph with an electrochemical detector for the measurement
       of hydrogen and carbon monoxide in the parts per million range, applied to exhaled air. Chromatographia 8: 581-583.
Bishop, G. A.; Stedman, D. H. (1996) Measuring the emissions of passing cars. Ace. Chem. Res. 29: 489-495.
Bishop, G. A.; Starkey, J. R.; Ihlenfeldt, A.; Williams, W. J.; Stedman, D. H. (1989) IR long-path photometry: a remote sensing
       tool for automobile emissions. Anal. Chem. 61: 671A-677A.
Bradley, K. S.; Stedman, D. H.; Bishop, G. A. (1999) A global inventory of carbon monoxide emissions from motor vehicles.
       In: Khalil, M. A. K.; Pinto, J. P.; Shearer, M. J., eds. Carbon monoxide [special issue  of papers from an international
       conference; December 1997; Portland, OR]. Chemosphere Global Change Sci. 1: 65-72.
Bruner, F.; Ciccioli, P.; Rastelli, R. (1973) The determination of carbon monoxide in air in the parts per billion range by means
       of a helium detector. J. Chromatogr. 77: 125-129.
Brunke, E. -G.; Scheel, H. E.; Seiler, W. (1990) Trends of tropospheric CO, N2O, and CH4 as observed at Cape Point, South Africa.
       Atmos. Environ. 24A: 585-595.
Burch, D. E.; Gates, F. J.; Pembrook, J. D.  (1976)  Ambient carbon monoxide monitor. Research Triangle Park, NC: U.S.
       Environmental Protection Agency, Environmental Sciences Research Laboratory; report no. EPA-600/2-76-210.
Carter, W. P. L. (1991) Development of ozone reactivity scales for volatile organic compounds. Research Triangle Park,NC: U.S.
       Environmental  Protection Agency,  Atmospheric  Research and  Exposure  Assessment  Laboratory; report  no.
       EPA/600/3-91/050.
Chaney, L. W.; McClenny, W. A. (1977) Unique ambient carbon monoxide monitor based on gas filter correlation: performance
       and application. Environ. Sci. Technol. 11: 1186-1190.
Chaney, L. W.;Rickel, D. G.; Russwurm, G. M.; McClenny, W. A. (1979) Long-path laser monitor of carbon monoxide: system
       improvements. Appl.  Opt. 18: 3004-3009.
Code of Federal Regulations. (1991a) Ambient air monitoring reference and equivalent methods. C. F. R. 40: §53.
Code of Federal Regulations. (1991b) National primary and secondary ambient air quality standards. C. F. R. 40: §50.
Commins, B. T.; Berlin, A.; Langevin, M.; Peal, J. A. (1977) Intercomparison of measurement of carboxyhaemoglobin in different
       European laboratories and establishment of the methodology for the assessment of COHb levels in exposed populations.
       Luxembourg, Sweden: Commission of the European Communities, Health and Safety Directorate; doc. no. V/F/1315/77e.
Dagnall, R. M.; Johnson, D. J.; West, T. S. (1973) A method for the determination of carbon monoxide, carbon dioxide, nitrous
       oxide and sulphur dioxide in air by gas  chromatography  using an emissive helium plasma detector. Spectrosc. Lett.
       6: 87-95.
Dickerson, R. R.; Delany, A. C. (1988) Modification of a commercial gas filter correlation CO detector for enhanced sensitivity.
       J. Atmos. Oceanic Technol. 5: 424-431.
Elwood, J. H. (1976) A calibration procedural system and facility for gas turbine engine exhaust emission analysis. In: Calibration
       in air monitoring: a symposium; August 1975; Boulder, CO.  Philadelphia, PA: American Society for Testing and Materials;
       pp. 255-272; ASTM special technical publication no. 598.


                                                     2-17

-------
Federal  Register.  (1978) Air quality surveillance and data  reporting: proposed regulatory revisions.  F.  R.  (August 7)
      43: 34892-34934.
Feldstein, M. (1967) Methods for the determination of carbon monoxide. Prog. Chem. Toxicol. 3: 99-119.
Fishman, J.; Seiler, W.; Haagenson, P. (1980) Simultaneous presence of O3 and CO bands in the troposphere. Tellus 32: 456-463.
Freeman, G. L.; Steinke, J. M. (1993) Evaluation of two oximeters for use in cardiac catheterization laboratories. Catheterization
      Cardiovasc. Diagn. 30: 51-57.
Fried, A.; Henry, B.; Parrish, D. D.; Carpenter, J. R.; Buhr, M.  P. (1991) Intercomparison of tunable diode laser and gas filter
      correlation measurements of ambient carbon monoxide.  Atmos. Environ. 25A: 2277-2284.
Garbarz, J.-J.; Sperling, R. B.; Peache, M. A. (1977) Time-integrated urban carbon monoxide measurements. San Francisco, CA:
      Environmental Measurements, Inc.
Gerbig, C.; Kley, D.; Volz-Thomas, A.; Kent, J.; Dewey, K.; McKenna, D. S. (1996) Fast response resonance fluorescence CO
      measurements aboard the C-130: instrument characterization and measurements made during North Atlantic Regional
      Experiment 1993. J. Geophys. Res.  [Atmos.] 101: 29,229-29,238.
Gerbig, C.; Schmitgen, S.; Kley, D.; Volz-Thomas, A.; Dewey, K.; Haaks, D. (1999) An improved fast-response vacuum-UV
      resonance fluorescence CO instrument. J. Geophys. Res. [Atmos.] 104: 1699-1704.
Goldstein, G. M. (1977) COFIb measurements, results of discussion between Drs. A. Berlin, CEC and G. Goldstein, EPA on ECE
      doc V/F/1315/77e [letter to Mr. Jack Thompson]. Research Triangle Park, NC: U.S. Environmental Protection Agency;
      September 23.
Gong, A. K. (1995) Near-patient measurements of methemoglobin, oxygen saturation, and total hemoglobin: evaluation of a new
      instrument for adult and neonatal intensive care. Crit.  Care Med. 23: 193-201.
Greenberg, J. P.; Zimmerman, P. R.; Heidt, L.; Pollock, W. (1984) Hydrocarbon and carbon monoxide emissions frombiomass
      burning in Brazil. J. Geophys. Res.  [Atmos.] 89:  1350-1354.
Guenther, F. R.; Dorko, W. D.; Miller, W. R.; Rhoderick, G.  C. (1996) The NIST traceable reference material program for gas
      standards. Washington, DC: U. S. Department of Commerce, National Institute of Standards and Technology; NIST special
      publication 260-126.
Haldane, J. (1897-1898) Some improved methods of gas analysis. J. Physiol.  (London) 22: 465-480.
Heidt, L. E. (1978) Whole air collection and analysis. Atmos. Technol. (9): 3-8.
Hoell, J. M.; Gregory, G. L.; Carroll, M. A.; McFarland, M.; Ridley, B. A.; Davis, D. D.; Bradshaw, J.; Rodgers, M. O.; Torres,
      A. L.; Sachse, G. W.; Hill, G. F.; Condon, E. P.; Rasmussen, R. A.; Campbell, M. C.; Farmer, J. C.; Sheppard, J. C.; Wang,
      C. C.; Davis, L. I. (1984) An intercomparison of carbon monoxide, nitric oxide, and hydroxyl measurement techniques:
      overview of results. J. Geophys. Res. [Atmos.] 89: 11,819-11,825.
Hoell, J. M., Jr.; Gregory, G.  L.;  McDougal, D. S.; Sachse, G. W.; Hill,  G. F.; Condon, E. P.; Rasmussen, R. A. (1985)
      An intercomparison of carbon monoxide measurement techniques. J. Geophys. Res. [Atmos.] 90:  12,881-12,889.
Hoell, J. M., Jr.; Gregory, G. L.; McDougal, D. S.; Sachse, G.  W.; Hill, G. F.; Condon, E. P.; Rasmussen, R. A. (1987) Airborne
      intercomparison of carbon monoxide measurement techniques. J. Geophys. Res. [Atmos.] 92: 2009-2019.
Hughes, E. E.; Dorko, W. D.; Scheide, E.  P.; Hall, L. C.; Beilby, A. L.; Taylor, J. K. (1973) Gas generation systems for the
      evaluation of gas detecting devices. Washington, DC: U. S. Department of Commerce, National Bureau of Standards; report
      no. NBSIR 73-292.
Hughes, E. E.; Davenport, A. J.; Woods, P. T.; Zielinski, W. L., Jr. (1991) Intercomparison of a range of primary gas standards
      of carbon monoxide in nitrogen and  carbon dioxide  in nitrogen prepared by the National Institute of Standards and
      Technology and the National Physical Laboratory. Environ. Sci. Technol. 25: 671-676.
Hunt, W. F., Jr. (1972)  The  precision associated with the sampling frequency of log-normally distributed air pollutant
      measurements. J. AirPollut. Control Assoc. 22: 687-691.
Instrumentation Laboratory. (1999) IL 682 CO-Oximeter: ready when you are. Lexington, MA: Instrumentation Laboratory,
      Critical Care Technical Support Group.
Jones, R. H.; Ellicott, M. F.; Cadigan, J. B.; Gaensler, E. A. (1958) The relationship between alveolar and blood carbon monoxide
      concentrations during breathholding. J. Lab. Clin. Med.  51: 553-564.
Kimmel, E. C.; Carpenter, R. L.; Reboulet, J. E.; Still, K. R.  (1999) A physiological model for predicting carboxyhemoglobin
      formation from exposure to carbon monoxide in rats. J. Appl. Physiol. 86: 1977-1983.
Klepeis, N. E.; Ott, W. R.; Repace, J. L. (1999) The effect of  cigar smoking on indoor levels of carbon monoxide and particles.
      J. Exposure Anal. Environ. Epidemiol. 9: 622-635.
Krarup, T. (1998) The ABL™700 Oximeter:  new standards in interference detection. Copenhagen, Denmark: Radiometer Medical;
      bulletin no. 3.
Krarup, T. (1999) ABL 700 series [email to D. Mage]. Denmark, Copenhagen: Radiometer Medical; September 2.
Langan, L. (1992) Portability in measuring exposure to carbon monoxide. In: Measuring, understanding, and predicting exposures
      in the 21st century: proceedings of the conference; November 1991; Atlanta, GA. J. Exposure Anal. Environ. Epidemiol.
       l(suppl.  1): 223-239.
                                                      2-18

-------
Larsen, R. I. (1971) A mathematical model for relating air quality measurements to air quality standards. Research Triangle Park,
      NC: U.S. Environmental Protection Agency, Office of Air Programs; report no. AP-89.
Lawson, D. R.; Groblicki, P. J.; Stedman, D. H.; Bishop, G. A.; Guenther, P. L. (1990) Emissions from in-use motor vehicles in
      Los Angeles: a pilot study of remote sensing and the inspection and maintainence program. J. Air Waste Manage. Assoc.
      40: 1096-1105.
Lee, K.; Yanagisawa, Y. (1992) Development of a sampler for carbon monoxide in expired air. Presented at: 85th annual meeting
      and exhibition of the Air & Waste Management Association; June; Kansas City, MO. Pittsburgh, PA: Air & Waste
      Management Association; paper no. 92-145.03.
Lee,K.; Yanagisawa, Y.  (1995) Sampler for measurement of alveolar carbon monoxide. Environ. Sci. Technol.  29:104-107.
Lee, P. S.; Majkowski, R. F.; Perry, T. A. (1991) Tunable diode laser spectroscopy for isotope analysis  - detection of isotopic
      carbon monoxide in exhaled breath. IEEE Trans. Biomed. Eng. 38: 966-973.
Lee, K.; Yanagisawa, Y.; Spengler, J. D. (1992a) Comparison of carbon monoxide measurement methods. Presented at: 85th
      annual meeting & exhibition of the Air & Waste Management Association; June; Kansas City, MO; paper no. 92-80.04.
Lee, K.; Yanagisawa, Y.; Hishinuma, M.; Spengler, J. D.; Billick, I. H.  (1992b) A passive sampler for measurement of carbon
      monoxide using a solid adsorbent. Environ. Sci. Technol. 26: 697-702.
Lee, K.; Yanagisawa, Y.; Spengler, J. D.; Billick, I. H. (1992c) Measurement of personal carbon monoxide exposures by mailed
      passive sampler. J. Air Waste Manage. Assoc. 42: 1212-1213.
Lee, P. S.; Schreck, R. M.; Hare, B. A.; McGrath, J. J. (1994) Biomedical applications of tunable diode laser spectrometry:
      correlation between breath carbon monoxide and low level blood carboxyhemoglobin saturation. Ann. Biomed. Eng.
      22: 120-125.
Lloyd, W. G.; Rowe, D. R. (1999) Estimation of carboxyhemoglobin. Environ. Sci. Technol. 33: 782-785.
Mahoney, J. J.; Vreman, H. J.; Stevenson, D. K.; Van Kessel, A. L.  (1993) Measurement of carboxyhemoglobin and total
      hemoglobin by five specialized spectrophotometers (CO-oximeters) in comparison with reference methods. Clin. Chem.
      (Washington, DC) 39:  1693-1700.
Marshall,  M.  D.; Kales, S.  N.; Christiani, D. C.;  Goldman, R. H. (1995) Are reference intervals for carboxyhemoglobin
      appropriate? A survey of Boston area laboratories. Clin. Chem. (Washington, DC) 41: 1434-1438.
McBride, S. J.; Ferro, A. R.; Ott, W. R; Switzer, P.; Hildemann, L. M. (1999) Investigations of the proximity effect for pollutants
      in the indoor environment. J. Exposure Anal. Environ. Epidemiol. 9: 602-621.
McCullough, J. D.; Crane, R. A.; Beckman, A. O. (1947) Determination of carbon monoxide in air by use of red mercuric oxide.
      Anal. Chem. 19: 999-1002.
Morales-Morales, R. (1998) Carbon monoxide, ozone, and hydrocarbons in the Baltimore metropolitan area [dissertation]. College
      Park, MD: University of Maryland. Available from: University Microfilms International, Ann Arbor, MI; publication no.
      9908993.
Moser, L.; Schmid, O. (1914) Die Bestimmung des Kohlenoxyds durch gelbes quecksilberoxyd [The determination of carbon
      monoxide through yellow mercury oxide]. Z. Anal. Chem. 53: 217-233.
Moy, L. A.; Dickerson, R. R.; Ryan, W. F. (1994) Relationship between back trajectories and tropospheric trace gas concentrations
      in rural Virginia.  Atmos. Environ. 28: 2789-2800.
Mueller, R. H. (1954) A  supersensitive gas detector permits accurate detection of toxic or combustible gases in extremely low
      concentrations. Anal. Chem. 26: 39A-42A.
National Bureau of Standards. (1975) Catalog of NBS standard reference materials,  1975-76 edition. Washington, DC: U.S.
      Department of Commerce, National Bureau of Standards; NBS special publication no. 260.
Novelli, P. C.;  Elkins,  J. W.; Steele, L. P. (1991) The development and evaluation of a gravimetric reference scale  for
      measurements of atmospheric carbon monoxide. J. Geophys. Res. [Atmos.] 96: 13,109-13,121.
Novelli, P. C.; Collins, J. E., Jr.; Myers, R. C.; Sachse, G. W.; Scheel, H. E. (1994) Reevaluation of the NOAA/CMDL carbon
      monoxide reference  scale  and comparisons with CO reference  gases at  NASA-Langley and the Fraunhofer Institut.
      J. Geophys. Res.  [Atmos.]  99: 12,833-12,839.
Novelli, P. C.; Connors, V. S.; Reichle, H. G., Jr.; Anderson, B. E.; Brenninkmeijer, C. A. M.; Brunke, E. G.; Doddridge, B. G.;
      Kirchhoff, V. W. J. H.; Lam, K. S.; Masarie, K. A; Matsuo, T.; Parrish,  D. D.; Scheel, H. E.; Steele, L. P. (1998) An
      internally consistent set of globally distributed atmospheric carbon monoxide mixing ratios developed using results from
      an intercomparison of measurements. J. Geophys. Res. [Atmos.]  103: 19,285-19,293.
O'Keeffe, A. E.; Ortman, G.  C. (1966) Primary standards for trace gas analysis. Anal. Chem. 38: 760-763.
Ott, W.  R.; Mage, D. T. (1975) Random sampling as  an inexpensive means  for measuring average annual air pollutant
      concentrations in urban areas. Presented at: 68th annual meeting of the Air Pollution Control Association; June; Boston,
      MA. Pittsburgh, PA: Air Pollution Control Association; paper no. 75-14.3.
Ott, W. R.; Rodes, C. E.; Drago, R. J.; Williams, C.; Burmann, F. J. (1986) Automated data-logging personal exposure monitors
      for carbon monoxide. J. AirPollut. Control Assoc. 36: 883-887.
Ott, W.; Switzer, P.; Willits, N. (1994) Carbon monoxide exposures inside an automobile traveling on an urban arterial highway.
      J. Air Waste Manage. Assoc. 44:  1010-1018.
                                                      2-19

-------
Ott, W. R.; Vreman, H. J.; Switzer, P.; Stevenson, D. K. (1995) Evaluation of electrochemical monitors for measuring carbon
      monoxide concentrations in indoor, in-transit, and outdoor microenvironments. In: Measurement of toxic and related air
      pollutants: proceedings of an international symposium; May; Research Triangle Park, NC. Pittsburgh, PA: Air & Waste
      Management Association; pp. 172-177. (A&WMA publication VIP-50).
Palanos, P. N. (1972) A practical design for an  ambient carbon monoxide mercury  replacement analyzer. Anal. Instrum.
       10:  117-125.
Parrish, D. D.; Holloway, J.  S.; Fehsenfeld, F. C. (1994) Routine, continuous measurement of carbon monoxide with parts per
      billion precision. Environ. Sci. Technol. 28: 1615-1618.
Paulsell, C. D. (1976) Use of the National Bureau of Standards standard reference gases in mobile source emissions testing.
      In: Calibration in air monitoring: a symposium; August 1975; Boulder, CO. Philadelphia, PA:  American Society for
      Testing and Materials; pp. 232-245. (ASTM special technical publication 598).
Phinney, D. E.; Newman, J. E. (1972) The precision associated with the sampling frequencies of total particulate at Indianapolis,
      Indiana. J. AirPollut. Control Assoc.  22: 692-695.
Porter, K.; Volman, D. H. (1962) Flame ionization detection of carbon monoxide for gas chromatographic analysis. Anal. Chem.
      34: 748-749.
Poulida, O.; Dickerson, R. R.; Doddridge, B.  G.; Holland, J. Z.; Wardell, R. G.;  Watkins, J. G. (1991) Trace gas concentrations
      and meteorology in rural Virginia. 1. Ozone and carbon monoxide. J. Geophys. Res. [Atmos.] 96: 22,461-22,475.
Rhodes, R. C.;Evans,E. G. (1987) Precision and accuracy assessments for state and local air monitoring networks 1985. Research
      Triangle Park, NC: U.S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory; report no.
      EPA-600/4-87-003.
Robbins, R. C.; Borg, K. M.;  Robinson, E. (1968) Carbon monoxide in the atmosphere. J. AirPollut. Control Assoc. 18: 106-110.
Rodgers, P. A.; Vreman, H. J.; Dennery, P. A.; Stevenson, D. K. (1994) Sources of carbon monoxide (CO) in biological systems
      and applications of CO detection technologies. Semin. Perinatol. 18: 2-10.
Roths, J.;  Zenker, T.; Parchatka, U; Wienhold, F.  G.; Harris, G. W.  (1996)  Four-laser airborne infrared spectrometer for
      atmospheric trace gas measurements.  Appl. Opt. 35: 7075-7084.
Ryan, W. F.; Doddridge, B. G.; Dickerson, R. R.; Morales, R. M.; Hallock, K. A.; Roberts, P. T.; Blumenthal, D. L.; Anderson,
      J. A.; Civerolo, K. (1998) Pollutant transport during a regional O3 episode  in the mid-Atlantic states. J. Air Waste Manage.
      Assoc. 48: 786-797.
Sachse, G. W.; Hill, G. F.; Wade, L. O.; Perry, M. G. (1987) Fast-response, high-precision carbon monoxide sensor using a tunable
      diode laser absorption technique. J. Geophys. Res. [Atmos.] 92: 2071-2081.
Scaringelli, F. P.; Rosenberg, E.; Rehme, K.  A. (1970) Comparison of permeation devices and nitrite ion as standards for the
      colorimetric determination of nitrogen dioxide. Environ. Sci. Technol. 4: 924-929.
Seiler, W.; Giehl, H.; Roggendorf, P. (1980) Detection of carbon monoxide and hydrogen by conversion of mercury oxide to
      mercury vapor. Atmos. Technol. (12): 40-45.
Seinfeld, J. H.; Pandis, S. N. (1998) Atmospheric chemistry and physics: from air pollution to climate change. New York, NY:
      John Wiley & Sons,  Inc.
Shepherd, A. P.; McMahan,  C. A. (1996) Role of oximeter error in the diagnosis of shunts. Catheterization Cardiovasc. Diagn.
      37: 435-446.
Smith,N. J. (1977) End-expired airtechnic for determining occupational carbon monoxide exposure. J. Occup. Med. 19: 766-769.
Smith, F.; Nelson, A. C., Jr. (1973)  Guidelines for development of a quality assurance program: reference method for the
      continuous measurement of carbon monoxide  in the atmosphere. Research  Triangle Park, NC: U.S. Environmental
      Protection Agency, Quality Assurance and Environmental Monitoring Laboratory; report no. EPA-R4-73-028a.
Smith, R. G.; Bryan, R. J.; Feldstein, M.; Locke, D. C.; Warner, P. O. (1975) Tentative method for constant pressure volumetric
      gas analysis for O2, CO2, CO,  N2, hydrocarbons (ORSAT). Health Lab. Sci.  12: 177-181.
Stedman, D. H.; Kok, G.; Delumyea, R.; Alvord, H. H. (1976) Redundant calibration of nitric oxide, carbon monoxide, nitrogen
      dioxide, and ozone air pollution monitors by chemical and gravimetric techniques. In: Calibration in air monitoring: a
      symposium; August  1975; Boulder, CO. Philadelphia, PA: American Society for Testing and Materials; pp. 337-344.
      (ASTM special technical publication  598).
Stedman, D. H.; Bishop, G.  A.; Aldrete, P.;  Slott, R. S. (1997) On-road evaluation  of an automobile emission test program.
      Environ. Sci. Technol. 31: 927-931.
Stedman, D. H.;  Bishop, G. A.; Slott, R. S.  (1998) Repair avoidance and evaluating inspection and maintenance programs.
      Environ. Sci. Technol. 32: 1544-1545.
Stevenson, D. K.; Vreman, H. J. (1997) Carbon monoxide and bilirubin production in neonates. Pediatrics 100: 252-254.
Swinnerton, J. W.; Linnenbom, V. J.; Cheek, C. H. (1968) A  sensitive gas chromatographic method for determining carbon
      monoxide in seawater. Limnol. Oceanogr. 13: 193-195.
Tesafik, K.; Krejci, M. (1974) Chromatographic determination of carbon monoxide below the 1 ppm level. J. Chromatogr.
      91:539-544.
                                                      2-20

-------
U.S. Code. (1991) Clean Air Act, §108, air quality criteria and control techniques, §109, national ambient air quality standards.
       U. S. C. 42: §§7408-7409.
U.S. Environmental Protection Agency. (1991) Air quality criteria for carbon monoxide. Research Triangle Park, NC: Office of
       Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA/600/8-90/045F.
U.S. Environmental Protection Agency.  (1997) EPA traceability protocol for assay and certification of gaseous calibration
       standards.  Washington, DC:  National Center  for  Environmental  Research and  Quality Assurance;  report  no.
       EPA/600/R-97/121.
Van Dam, J.; Daenens, P. (1994) Microanalysis of carbon monoxide in blood by head-space capillary gas chromatography.
       J. Forensic Sci. 39: 473-478.
Volz, A.; Kley, D. (1985) A resonance-fluorescence instrument for the in-situ measurement of atmospheric carbon monoxide.
       J. Atmos. Chem. 2: 345-357.
Vreman,H. J.; Stevenson,D. K. (1994)Carboxyhemoglobin determined inneonatal blood with a CO-oximeter unaffected by fetal
       oxyhemoglobin. Clin. Chem. (Washington, DC) 40: 1522-1527.
Vreman, H. J.; Mahoney, J. J.; Stevenson, D. K. (1993) Electrochemical measurement of carbon monoxide in breath: interference
       by hydrogen. Atmos. Environ. Part A 27: 2193-2198.
Vreman, H. J.; Mahoney, J. J.; Stevenson, D. K. (1995) Carbon monoxide and carboxyhemoglobin. Adv. Pediatr. 42: 303-334.
Wallace, L. A. (1983) Carbon monoxide in air and breath of employees in an underground office. J. Air Pollut. Control Assoc.
       33:678-682.
Wallace, L.; Thomas, J.;Mage, D.; Ott, W. (1988) Comparison of breath CO, CO exposure, and Coburn model predictions in the
       U.S. EPA Washington-Denver (CO) study. Atmos. Environ. 22: 2183-2193.
Weeks, I. A.; Galbally, I. E.; Fraser, P. J.; Matthews, G. (1989) Comparison of the carbon monoxide standards used at Cape Grim
       and Aspendale. In: Forgan, B. W.; Ayres, G. P., eds. Baseline Atmosheric Program (Australia) 1987. Canberra, New South
       Wales: Department of Administrative Services, Bureau of Meteorology; pp. 21-25.
Weinheimer, A. J.; Ridley, B. A. (1990) A cautionary note on the use of some mass flow controllers. J. Geophys. Res. [Atmos.]
       95: 9817-9821.
Wohlers, H. C.; Newstein, H.; Daunis, D.  (1967) Carbon monoxide and sulfur dioxide adsorption on- and desorption from glass,
       plastic, and metal tubings. J. Air Pollut. Control Assoc. 17: 753-756.
                                                     2-21

-------
                                   CHAPTER 3

     Sources,  Emissions, and Concentrations of
     Carbon  Monoxide  in Ambient and Indoor Air
3.1  Introduction
      This chapter summarizes current information about the abundance and distribution, the nature of
sources and sinks, and the chemistry of carbon monoxide (CO) in various environments, ranging from the
global background to indoor air. Carbon monoxide is studied in these widely varied environments for
different reasons.  Sources such  as unvented, malfunctioning, or misused combustion  appliances,
combustion engines in garages or basements, and tobacco combustion can cause high concentrations of CO
to exist in  the indoor environment. Sources such as motor vehicles,  nonroad combustion engines or
vehicles, and biomass burning can cause high concentrations of CO in the outdoor environment.  In both
of these environments, CO is of direct concern because of the health effects that can result from human
exposure to these high concentrations. Human exposures to CO are discussed in Chapter 4, and possible
health effects are discussed in Chapter 6.
      Carbon monoxide in less polluted air is of interest because of its importance to atmospheric
chemistry.  Carbon monoxide can affect the formation of ozone (O3) and other photochemical oxidants in
the atmosphere.  Carbon monoxide strongly influences the abundance of hydroxyl radicals (OH), thus
affecting the global cycles of many biogenic and anthropogenic trace gases that affect the abundance of
stratospheric O3 and the energy budget of the atmosphere. Changes in CO concentrations, therefore, may
contribute to widespread changes in atmospheric chemistry and indirectly affect global climate. In this
chapter, the global scale aspects of CO are discussed first, and then the discussion proceeds to successively
smaller spatial scales. An overview of the major sources and sinks of CO and the resulting CO distribution
on a global basis and the importance of CO to tropospheric chemistry is presented in Section 3.2, followed
by a discussion of nationwide emissions of CO in Section  3.3.  Nationwide trends in ambient CO
concentrations and related discussions on CO air quality are presented in Section 3.4, and concentrations
and sources of CO in indoor environments are discussed in Section 3.5.
3.2  The Global Cycle of Carbon Monoxide
      The major sources and sinks of CO are summarized in Table 3-1. Examples of major activities
leading to the emissions of CO from each source category are shown in the second column of Table 3-1.
Many of these sources have natural components. As can be seen from Table 3-1, CO is produced as a
primary pollutant during the combustion of fossil  and biomass fuels. Vegetation also can emit CO directly
into the atmosphere as a metabolic by-product.
      Carbon monoxide is formed as an intermediate product during the photochemical oxidation of
methane and non-methane hydrocarbons (NMHCs) to carbon dioxide (CO2). Major sources of methane are
summarized in the second column of Table 3-1.  Likewise, major sources of NMHCs, whose oxidation
produces CO, are given. In addition, the photooxidation of organic matter in surface waters (oceans, lakes,
                                          3-1

-------
 	Table 3-1. Summary of Major Sources and Sinks of Carbon Monoxide	
  Sources and Sinks                                              Notes

  Sources

  Fossil fuel combustion              Transportation and coal, oil, and natural gas burning

  Biomass burning                   Agricultural clearing, wood and refuse burning, and forest fires3

  Methane (CH4) oxidation            Wetlands,3 agriculture (rice cultivation, animal husbandry, and biomass
                                  burning), landfills, coal mining, and natural gas and petroleum industry

  Non-methane hydrocarbon (NMHC)   Transportation (alkanes, alkenes, and aromatic compounds) and
  oxidation                         vegetation3 (isoprene and terpenes)

  Organic matter oxidation3            Humic and other organic substances in surface waters and soils

  Vegetation3                       Metabolic by-product

  Sinks

   Reaction with OH radicals          Hydroxyl radicals are ubiquitous scavengers of many atmospheric
                                  pollutants.

   Soil microorganisms3	Responsible microorganisms still need to be cataloged.	

  3Sources and sinks that have large natural components.
and rivers) and on the soil surface occurs. Carbon monoxide is lost primarily by reaction with atmospheric
OH radicals and by uptake by soil microorganisms.
       Carbon monoxide  concentrations and trends in the background  atmosphere are discussed
in Section 3.2.1. More detailed descriptions of the nature of individual sources of primary CO shown in
Table 3-1 and estimates of the strengths of these sources, along with similar material for nonchemical sinks
of CO, are given in Section 3.2.2. The atmospheric chemistry of CO, including the formation of secondary
CO, is discussed in Section 3.2.3.

3.2.1  Global Background Concentrations of Carbon Monoxide
       In common  usage, the term "background concentrations" refers to  concentrations observed in
remote areas relatively unaffected by local pollution sources.  However, several definitions of background
concentrations are possible (see Chapter 6, U.S. Environmental Protection Agency, 1996).  The  two
definitions chosen in that  document as being most relevant for regulatory purposes and for providing
corrections to assessments of the  health risks posed by exposure to CO  are based on estimates of
contributions from uncontrollable sources that can affect CO concentrations in the United States. The first
definition is the concentration resulting from anthropogenic and natural emissions outside North America,
and natural sources within North America. The second definition is the concentration resulting from global
natural sources. Because of long-range transport from anthropogenic source regions in North America, it
is impossible to obtain background concentrations defined above solely on the basis of direct measurement
in remote areas in North America. However, some inferences about what these concentrations may be can
be made with the help of numerical models and historical data.
       Surface measurements of CO concentrations are made routinely as part of the National Oceanic and
Atmospheric Administration's  Climate Monitoring Diagnostics Laboratory (NOAA/CMDL)  Global
Cooperative Air Sampling Network (e.g., Hofmann et al, 1996). Carbon monoxide flask samples are
collected weekly in flasks or continuously with in situ gas chromatographs at about 40 remote sites around
                                              3-2

-------
Figure 3-1.   Latitudinal and seasonal variability in CO
concentrations  obtained by the  NOAA/CMDL monitoring
network.
Source:  National Oceanic and Atmosphere Administration
(1999a).	
the world. These sites are located primarily in
the marine boundary layer, with a few located in
continental areas. The latitudinal and seasonal
variations in CO concentrations are summarized
in the three-dimensional diagram shown  in
Figure 3-1 (National Oceanic and Atmospheric
Administration,  1999a).  Annual average CO
concentrations  are about  120  ppb  in the
Northern Hemisphere and about 40 ppb in the
Southern Hemisphere. Seasonal maxima in CO
concentrations occur during late winter in both
hemispheres, and  minima  occur during late
summer, with about a factor  of two variation
between maximum and minimum  values.
Carbon  monoxide is  well  mixed  in  high
latitudes of both the Northern Hemisphere and
the Southern Hemisphere. A  steep gradient in
CO concentrations exists between about 30°
north (N) latitude and  about 10° south (S)
latitude.  Carbon monoxide concentrations range from a minimum of about 30 ppb during summer in the
Southern Hemisphere to about 200 ppb at high latitudes in the Northern Hemisphere during winter. Thus,
CO concentrations  in remote areas of the Northern Hemisphere are only a small fraction (~ 1 to 2%) of those
of concern to human health (as given by the National Ambient Air Quality Standards [NAAQS] for CO of
9 ppm for the second highest,  nonoverlapping 8-h average concentration).
       There were sufficient  data on tropospheric air quality trends to suggest that  CO concentrations
measured at global background sites were increasing at 1.2 ± 0.6% per year from 1981 to 1986, based on
data collected by the Oregon Graduate Institute (Khalil and Rasmussen, 1988a).  From 1987 until 1992,
global background concentrations of CO declined at  a rate of about  -2.6 ± 0.8% per year (Khalil and
Rasmussen, 1994), whereas Novelli et al. (1994) determined a rate of decrease in CO  of -6.5 ± 0.8% per
year  from 1990 to 1993,  and later  results
reported  by  NOAA/CMDL  indicate  that
background concentrations of CO continued to
decline, although at a lower rate of -2.6 ± 0.2%
per year from  1990 to  1995 (Novelli  et al.,
1998). Hypotheses to explain these observations
include reductions in fossil  fuel combustion
(Bakwin et al., 1994; Novelli et al., 1994; Khalil
and Rasmussen, 1994)  and  tropical biomass
burning (Yung et al., 1999). Possible increases
in tropospheric  OH concentrations  resulting
from enhanced transmission of solar ultraviolet
radiation caused by stratospheric O3  depletion
may have been an additional factor (Fuglestvedt
et al., 1994; Bekki et al., 1994). More recent
data for changes  in global  background  CO
concentrations are  shown in  Figure 3-2.  The
data from 1993 to 1997 do not show clearly any
stable upward or downward trend.
           125
         O
         O
            75
         8
               Global Average
               Global Growth Rate
              1990 1991 1992 1993 1994 1995 1996 1997

 Figure 3-2. Global background average CO concentrations
 (upper) and growth rates for global background average CO
 (lower).
 Source:  National Oceanic and Atmospheric Administration
 (1999b).	
3-3

-------
       Because direct measurements of sufficient precision for defining trends have come into use only
within the last 15 to 20 years, estimates of longer term trends in CO concentrations must come from indirect
means. Rinsland and Levine (1985) derived an increase in the mean tropospheric CO abundance of about
2% per year from 1950 to 1984, based on an examination of solar spectra captured on photographic plates
in Europe.  Column CO abundances obtained over Zvenigorod, Russia, from 1974 through 1997 have
increased by about 1% per year (Yurganov et al., 1999), whereas measurements obtained with a similar
technique over the Alps have indicated decreases in CO of -0.18 ± 0.16% per year from 1984 to 1995
compared with a change of -0.95 ± 0.32% per year from 1984 to  1993 (Mahieu et al., 1997).  The
difference arises mainly from a pronounced minimum during the second half of 1992 and 1993, depending
on the fitting function.  Thus, there is still considerable uncertainty in defining global trends for CO based
on differences in trends found in specific regions.
       Carbon monoxide concentrations measured in air bubbles trapped in the ice sheets of Greenland
and Antarctica have been used as proxies for CO concentrations in ambient air at the time the air bubbles
were sealed from the atmosphere (Haan et al., 1996). Carbon monoxide concentrations derived this way
for the preindustrial era (roughly corresponding to the year 1850, when anthropogenic activities should not
have influenced significantly the atmospheric composition) are about 90 ppb for the high-latitude Northern
Hemisphere and about 50 ppb for the high-latitude Southern Hemisphere. Some enhancement of Northern
Hemispheric values over Southern Hemispheric values during the preindustrial era is likely because of the
greater mass of vegetation that can emit NMHCs in the Northern Hemisphere.  However, it should be noted
that the CO in the trapped air bubbles also may result from the decomposition of organic compounds also
trapped in  the same air bubbles, and that it is  difficult to extract CO from the air bubbles without
contamination. Both factors tend to cause positive artifacts in the CO concentrations reported. In addition,
the Northern Hemispheric value derived from the ice cores is higher than that predicted by atmospheric
model studies of the preindustrial era that indicate CO concentrations of about 50 ppb (Thompson and
Cicerone, 1986; Pinto andKhalil, 1991; Thompson et al., 1993).

3.2.2  Sources and  Global Emissions Estimates of Carbon  Monoxide
       Global CO emission estimates are summarized in Table 3-2.  Motor vehicles contribute most of the
emissions from fossil fuel combustion on the global scale according to the entries by Logan et al. (1981)
and Dignon et al. (1998).   Bradley et al.  (1999) estimated global emissions from motor vehicles of
213 Tg/year in 1991 based on roadside remote sensing measurements around the world.  They  also
calculated a decrease of 17%  in global motor vehicle emissions from 1991 to 1995.  Their estimated
uncertainty in both figures is about 20%. Variables controlling the formation of CO during combustion of
any fuel are oxygen concentration, flame temperature, gas residence time at high temperature, and mixing
in the combustion zone. In general, increases in  all four factors result in lower amounts of CO produced
relative to CO2. Carbon monoxide is produced primarily during conditions of incomplete combustion.  The
estimates for fossil fuel emissions by stationary sources, shown in the footnotes to Table 3-2, do not include
significant contributions from power plants because fuels are burned with high efficiency in modern power
plants. Rather, they are based on estimates of CO emitted in small, hand-fired furnaces used for domestic
purposes (e.g., cooking, heating, water sterilization)  and in  inefficient  boilers  and furnaces used in
small-scale industrial operations. These latter sources are of significance only in eastern Europe and in
developing countries of Africa and Asia (especially China). However, it also should be noted that the
importance of these sources has been declining as energy needs are met increasingly by centralized power
plants.
       Biomass burning consists of wildfires and the burning of vegetation to clear new land for agriculture
and population resettlement; to control the growth of unwanted plants on pasture land; to manage forest
resources (prescribed burning); to dispose of agricultural and  domestic waste;  and as fuel for cooking,
heating, and water sterilization. Most wildfires may be ignited directly as the result of human activities with

                                             3-4

-------
                           Table 3-2. Annual Global Carbon Monoxide Emissions Estimates (in teragrams per year)
CO
(In
Allen et al.
(1996)
Sources
Fossil fuel combustion 329
Biomass burning 370
Natural NMHC oxidation 6 1 8
Anthropogenic NMHC —
oxidation
Methane oxidation 722
Oceans —
Soils —
Vegetation —
Total 2,039
Sinks
Soils
OH reaction
Total
Logan et al. Seiler and Conrad Pacyna and Graedel Dignon et al.
(1981) (1987) (1995) (1998)

450a 640 ± 200 440 ± 150 600b
655 1,000 ±600 700 ±200 600
560 900 ±500 800 ±400 300
90 — — 200
810 600 ±300 600 ±200 600
40 100 ±90 50 ±40 10
— — — 30
130 75 ±25 75 ±25 200
2,735 3,3 15 ±1,700 2,700 ± 1,000 2,500

300
2,300
2,600

(2.1)c
(2.7)
(2.3)
(2.3)
(2)


(4)
(1.5)

(3)
(1.4)

       Estimate includes 150 Tg/year from stationary sources.
       bEstimate includes 100 Tg/year from stationary sources.
       °Values in parentheses represent ratio of maximum to minimum estimate of source term.

-------
only a fraction (10 to 30%) initiated by lightning (Andreae, 1991). However, because of fire management
practices in which natural wildfires are suppressed, the buildup of fire fuels increases the susceptibility of
forests to more severe but less frequent fires in the future.  Thus, there is considerable uncertainty in
attributing the fraction of wildfire emissions to human activities because the emissions from naturally
occurring fires that would have been present in the absence of fire suppression practices would have to be
known. For these reasons, values given above for the average contribution from human activities given
above are likely to be upper limits. Biomass burning exhibits strong seasonality, with most biomass burned
during the local dry season. The smoldering phase of combustion yields higher emissions factors than the
flaming phase. Lobert et al. (1991) found,  in controlled combustion chamber experiments with a wide
variety of vegetation types, that, on average, 84% of CO was produced during the smoldering phase and
16% during the flaming phase of combustion. Smoldering conditions are more prevalent during the burning
of large pieces of vegetation, such as trees, compared with grasses. Nonetheless, most CO is produced in
the tropics by savanna burning (mainly in Africa), followed by burning forests, fuel wood, and agricultural
waste.  Less than 20% of the CO produced by biomass burning originates in middle and high latitudes,
where most wildfires also occur (Andreae, 1991).
       The other sources of CO shown in Table 3-2 all have large natural components. Carbon monoxide
may be evolved from the photodecomposition of organic matter in surface waters (such as oceans, rivers,
and lakes) and the soil surface.  Soils can act as a source or a sink for carbon monoxide, depending on soil
moisture, the intensity of sunlight reaching the soil surface, and soil temperature (e.g., Inman et al., 1971;
Conrad and Seiler, 1985).  Soil uptake of CO occurs because of anaerobic bacteria (Inman et al., 1971).
Emissions of CO from soils appear to occur by abiotic processes, such as thermodecomposition or
photodecomposition of organic matter. In general, warm and moist conditions found in most soils favor
CO uptake, whereas hot and dry conditions as found in deserts and some savannas favor the release of CO
(King,  1999). The value reported for soil emissions in Table  3-2 is based on very limited data, and hence
it is difficult even to assign uncertainty bounds (Conrad, 1996). Moxley and Cape (1997) hypothesized that
from 20 to 80% of CO in the stable nocturnal boundary layer (calculated heights between 40 and 220 m)
could have been depleted by soil microorganisms during transport inland 100 km from the Scottish coast.
       Estimates of the magnitude of the soil sink range from 250 to 640 Tg/year (Logan et al., 1981;
Cicerone, 1988), with a current "best" estimate of 300 Tg/year, with an uncertainty range of a factor of three
(Dignon et al., 1998).  More extensive field measurements, perhaps based on the eddy correlation technique
(Ritter et al., 1994), are needed to characterize the variability and the direction of the CO flux to the soil
surface. Most CO in the atmosphere is lost by its oxidation  to CO2 by OH radicals.  Reaction with OH
radicals accounts for a loss of 2,300  Tg/year, with an  uncertainty factor of 1.4 (Dignon  et al., 1998).
Because of large uncertainties  in individual sources  and sinks, the imbalance between sources (2,500
Tg/year) and sinks (2,600 Tg/year) is not significantly different from zero.  By using a mean value of 80%
for  biomass burning  resulting from human activity and a value of two-thirds for the fraction of CH4
produced by human activity (Tegart and Sheldon, 1993), it can be seen that approximately two-thirds of CO
is produced globally as the result of human activities.

3.2.3  The Atmospheric Chemistry of Carbon Monoxide
       Carbon monoxide is  produced by  the photooxidation  of CH4 and other organic compounds
(including NMHCs) in the atmosphere  and of organic molecules in surface waters and soils (Table 3-1 and
3-2). Estimates of CH4 emissions from the various source categories shown in Table 3-1 can be found in
the Intergovernmental Panel on Climate Change report (Tegart and Sheldon, 1993). Methane oxidation can
be summarized by the following sequence of reactions:
                                             3-6

-------
                                CH4 + OH - CH3 + H2O
                                CH3 + O2 (+M) - CH3O2 (+M)
                                CH3O2 + NO - CH3O + NO2
                                CH3O + O2 - CH2O + HO2
                                CH2O + hu - H2 + CO,
                           or   CH2O+hu-HCO+H,
                           or   CH2O + OH - HCO + H2O
                                HCO + O2 - CO + HO2,

where M is a mediator (e.g., nitrogen, molecular oxygen [O2], argon, CO2). The photolysis of formaldehyde
(CH2O) proceeds by two pathways, the first yields molecular hydrogen (H2) plus CO (55%), and the second
yields atomic hydrogen (H) plus the formyl radical (HCO) (45%), where the percentages are given for
overhead sun conditions (Rogers, 1990). Formyl radicals then react with O2 to form the hydroperoxy radical
(HO2) plus CO. In addition, the reaction of the methyl peroxy radical (CH3O2) with HO2 radicals, forming
methyl hydroperoxide (CH3OOH), needs to be considered,  especially in low nitrogen oxide  (NOX)
environments. The heterogeneous removal of soluble intermediate products, such as CH3OOH, CH2O, and
radicals, decreases the yield of CO from the oxidation of CH4.
       Although the oxidation of CH2O nearly always results in CO formation (except for the formation
of small quantities of formic acid in the reaction of CH2O with HO2), the oxidation  of acetaldehyde
(CH3CHO) does not always yield two CO molecules. The photolysis of CH3CHO also involves pathways
that produce molecules and radicals, namely CH4 + CO and the methyl radical (CH3) + HCO. Estimates
of the yield of CO from the photooxidation of CH4 and CH3 are subj ect to the same considerations outlined
above. The reaction of CH3CHO with OH radicals can yield acetyl radicals (CH3CO). The acetyl radicals
then will participate with O2 in atermolecular recombination reaction to form acetyl peroxy radicals, which
then can react with nitric oxide (NO) to  form CH3 + CO2 (or the acetyl peroxy radicals can react with
nitrogen dioxide [NO2] to form peroxyacetyl nitrate [PAN]). Thus, one of the carbon atoms can be oxidized
directly to CO2 without passing through CO. The yield of CO depends  on the OH concentration and the
photolysis rate of CH3CHO, as well as on the abundance of NO, as acetyl peroxy radicals also can react with
HO2 and other hydrogen-bearing radicals.
       Estimates of the yield of CO from the oxidation of more complex hydrocarbons require the
calculation of the yields of CH2O, CH3CHO, CH3CO, and analogous radicals from the oxidation of the
parent molecule.  Likewise, the  extent of heterogeneous removal of soluble intermediate products needs
to be considered in the oxidation of more complex hydrocarbons.  However,  in contrast to simple
hydrocarbons containing one or two carbon atoms, detailed kinetic information is lacking about the gas
phase oxidation pathways of many anthropogenic hydrocarbons (e.g., aromatic compounds, such as benzene
and toluene), biogenic hydrocarbons (e.g., isoprene, the monoterpenes), and their intermediate oxidation
products (e.g., epoxides, nitrates, carbonyl compounds).  As much as 30% of the carbon in hydrocarbons
in many urban areas is in the form of aromatic compounds (Grosjean and Fung, 1984; Seila et al, 1989).
Yet mass balance analyses performed on irradiated smog chamber mixtures of aromatic hydrocarbons
indicate that only about one-half of the carbon is in the form of compounds that can be identified. Reactions
that have condensible products,  such as those occurring during the oxidation of terpenes, also need to be
considered because these reactions produce secondary organic particulate matter, thereby reducing the
potential yield of CO.
       The yield of CO from the oxidation of CH4 is about 0.9, and it is about 0.4 from the oxidation of
ethane and propane, on a per carbon basis from estimates based on atmospheric model results (Kanakidou
et al., 1991).  Jacob and Wofsy (1990) estimated that 1 mole of CO is produced by the oxidation of 1 mole
                                             3-7

-------
of isoprene (corresponding to a conversion factor of 0.2 on a per carbon basis) for lowNOx concentrations.
For higher NOX concentrations, they estimated that 3 moles of CO are produced per mole of isoprene
oxidized (corresponding to a conversion factor of 0.6 on a per carbon basis). Isoprene accounts for most
of the CO  produced by the photochemical oxidation of NMHCs shown in Table 3-2.
       The major pathway for removal of CO from the atmosphere is the reaction of CO with OH radicals.
There have been numerous determinations of the rate coefficient for this reaction.  The most recent
evaluation of kinetics data for  use in atmospheric modeling (National  Aeronautics and Space
Administration, Panel for Data Evaluation, 1997) recommends a value of 1.5  x  10"13  (1 + 0.6 Patm)
cm3 molecules"1 s"1, with a value of 0 ± 300 K for E/R for the reaction

                                     OH + CO - Products.

This reaction proceeds through two channels.  The bimolecular channel yields H + CO2, whereas  the
addition channel leads to the formation of a carboxyl radical (HOCO). In the presence of O2, the HOCO
intermediate is converted to HO2 + CO2.  Therefore, for atmospheric purposes, the products of the reaction
OH + CO  can be taken to  be HO2  and  CO2.
       Estimates of OH radical concentrations can be used along with the rate coefficient given above to
calculate the lifetime of CO in the atmosphere. Measurements of OH radical concentrations in situ (Hard
et al., 1992; Mount and Williams, 1997; Poppe et al, 1994)  in the  lower troposphere show that their
concentrations are highly site specific and are highly variable in space and time.  Typical mid-latitude
noontime values during summer (when OH concentrations are at their highest values) range from about 5
to 10 x 106 OH/cm3 and are much lower during other times of the day and during other seasons. As a result,
it is difficult to derive average values that would be meaningful for  use in calculating the atmospheric
lifetime of long-lived species that react with OH radicals, based on direct measurements.  Models of the
atmospheric distribution of methyl chloroform (CH3CC13) have been used to derive diurnal and global
average OH concentrations for calculating the atmospheric lifetimes of long-lived species by Prinn et al.
(1992).  Average OH values they derived in this manner are about 8 x 105 OH/cm3. By further adjusting
the  OH fields derived  in a simulation  of  the CH3CC13 distribution, to  optimize the fit between  the
measurements and simulations of CH3CC13 concentrations, Krol et al. (1998) derived concentrations of
1.00 x 106 OH/cm3 in 1978 and 1.07 x  106 OH/cm3 in 1993. The resulting trend in OH values is estimated
to be 0.46  ± 0.6% year"1. Krol et al. (1998) also used a three dimensional model of atmospheric chemistry
to examine the sensitivity of the OH trend to stratospheric ozone depletion, decreases in  CO emissions,
increases in tropical water vapor, and NOX and CH4 emissions, and they derived an overall change of 6%
in OH concentrations from 1978 to 1993. However, it should be noted that many of the changes used as
input to the model calculations are highly  uncertain  and thus, the results should be viewed only as a
sensitivity study (Law, 1999). It also should be noted that Prinn et al. (1995) found little or no trend in OH
based on CH3CC13 data.
       The resulting globally averaged atmospheric lifetime of CO is then  approximately 2 mo.  Shorter
lifetimes are found in the tropics, whereas longer lifetimes are found at higher latitudes.  During winter at
high latitudes, CO is essentially inert. The CO lifetime is shorter than the characteristic time scale for
mixing between the hemispheres  (about 1 year), and  hence a large gradient in concentrations can exist
between the hemispheres (see Figure 3-1). In addition, the chemical lifetime of CO at high latitudes is long
enough to result in much smaller gradients between 30° latitude and the pole of either hemisphere.
However,  the lifetime of CO is much longer than typical residence times of CO in urban areas  (assuming
a diurnally averaged OH concentration of 3 x 106 OH/cm3 in urban areas) and in indoor environments,
where OH concentrations are expected to be orders of magnitude lower.
       Reaction with CO, is in turn,  the major reaction of OH radicals.  The reaction of CO with OH
radicals constitutes at least 50% of the tropospheric sink of OH radicals (e.g., Collins et al., 1997).  Thus,

                                             3-8

-------
changes in the abundance of CO could lead to changes in the abundance of a number of trace gases whose
major loss process involves reaction with OH radicals. These trace gases can absorb infrared radiation from
the earth's surface and contribute to the greenhouse effect (e.g., CH4) or can deplete stratospheric O3 (e.g.,
methyl  chloride  [CH3C1],  methyl  bromide  [CH3Br],  and hydrochlorofluorocarbons,  such  as
difluorochloromethane).  Because of the importance of CO in determining OH concentrations, interest has
focused on the possible effects of increases in anthropogenic CO emissions on the concentrations of gases
such as those listed above (Sze, 1977; Chameides et al,  1977; Thompson and Cicerone,  1986).  For
instance, Thompson and Cicerone (1986) found in numerical simulations, in which the CO mixing ratio at
the surface was allowed to increase by 1% per year from 1980 to 2000, while holding CH4 emissions
constant, that the mixing ratio of CH4 at the surface increased by about 12% (corresponding to a mean
increase of 0.56% per year), and that the mixing ratio of CH4 at the surface increased by about 30%
(corresponding to a mean increase of 1.3% per year) for an increase in surface CO mixing ratio of 2% per
year.  However, based on the trend results reported earlier, Briihl and Crutzen (1999) have examined the
consequences of decreases in CO concentrations for atmospheric chemistry. They found, by decreasing CO
emissions linearly by about 20% between 1990 and 2000, that the CO reductions could lead to a significant
decrease (-25%) in the growth rate of CH4 and even to a decrease in CH4 concentrations in the case where
CH4 emissions remain constant. An accurate knowledge of the sources and sinks of carbon monoxide in
the atmosphere is therefore necessary for assessing the effects of future increases in anthropogenic CO
emissions on the concentrations of the above-mentioned, radiatively and photochemically important trace
species. However, because of nonlinearities introduced into the calculation of OH radical concentrations
by short-lived NOX (e.g., Hameed et al., 1979), an accurate assessment of these  effects awaits the
development of three-dimensional chemistry and transport models incorporating the spatial variability of
NOX (e.g., Kanakidou and Crutzen,  1993).
       In the free troposphere,  in the absence of significant quantities of NMHCs, the effects of CO on
tropospheric O3  can be summarized as shown below.

               Atmospheric Reactions Leading to O3 Production
                          CO + OH -  CO2 + H
                          H + O2 (+M) - HO2 (+M)
                          HO2+NO-NO2 + OH
                          NO2 + hv -  O + NO
                          O  + O2 (+M) - O3 (+M)
               Net        CO + 2O9 - CO? + O
                                             3

               Atmospheric Reactions Leading to O3 Destruction

                          CO + OH -  CO2 + H
                          H + O2 (+M) - HO2 (+M)
                          HO2 + O3 -  OH + 2O2
               Net        CO + O3 - CO2 + O2

       The oxidation of CO by OH could lead to the production or destruction of O3, depending on the ratio
of NO to HO2 concentrations. Based on current values of rate coefficients for the reactions of HO2 with NO
and O3, in regions where NO concentrations are greater than about 10 ppt, the oxidation of CO leads to O3
formation, whereas, in areas where NO concentrations are less than about 10 ppt, the oxidation of CO leads
to O3 destruction. Nitric oxide concentrations less than 10 ppt typically are found over the tropical oceans
(Carroll et al., 1990).  A rough estimate of the fraction of O3 production resulting from CO in the remote


                                             3-9

-------
troposphere can be made by taking the overall rate of the reaction of CO with OH radicals and then
correcting for the fraction of HO2 radicals that do not react with NO, based on free radical balances
presented by Collins et al. (1997). This quantity (i.e., the rate of conversion of NO to NO2 by HO2 radicals
produced by the reaction of CO with OH radicals) represents 20 to 40% of the production of O3 on a global
basis.
       The effects of CO on O3 photochemistry in environments with abundant hydrocarbons (e.g., cities,
tropical rain forests) require a much more complex treatment that includes the competition for OH radicals
by CO and NMHCs and the effects of this competition on the overall budget of hydrogen-containing
radicals (i.e., OH, HO2).  In urban environments, reaction with OH radicals represents the major loss
process for NMHCs and initiates the sequence of further reactions leading to the formation of O3 and CO
itself. Detailed analyses of the radical balances (i.e., production and loss rates in each reaction) in urban
air chemistry models, as performed by Jeffries (1995), can give the amount of O3 formed because of the
reaction of CO. However, only a few such analyses have been performed. Jeffries (1995), presented the
results of a numerical simulation of an O3 episode in Atlanta, GA, on June 6, 1988, and found that reaction
with CO  constituted 33% of the loss of OH radicals. It was found, by tracking sources of various radicals
produced by the oxidation of volatile organic compounds (VOCs) and that oxidize NO to NO2, that CO
accounted for about 17.5% of the O3 formed in this example (compared to about 82.5% for VOCs).
Obviously, more analyses of this sort are needed to characterize regional differences in the importance of
CO in different cities in the United States, which may have very different combinations of CO, NMHC, and
NOX concentrations than those used in these examples. Because of nonlinearities in the production rate of
O3 involving each of the above species, caution should be exercised in attempting to estimate the effects
of variations in CO concentrations on O3 production rates in the case studies cited above.
3.3  Nationwide Carbon Monoxide Emissions Estimates
       Total estimated primary CO emissions in the United States for the period of 1988 through 1997,
summarized from the National Emissions Inventory Trends  database, are shown in Table 3-3 (U.S.
Environmental Protection Agency, 1998). These emissions are shown in the original units used in their
calculation (i.e., thousands of short tons per year) and with the same number of significant figures. A short
ton is equal to 2,000 Ib or 9.08 x 105 g. Table 3-3 shows that total CO emissions decreased by 24.7% from
1988 to 1997; however, the fractional contribution of transportation (the major source of CO both then and
now) remained relatively constant at 77%. The term "transportation" includes both onroad and nonroad
sources. Onroad sources consist of automobiles, trucks, and buses. Nonroad sources consist of categories
such as trains;  aircraft; boats; ships; and lawn, recreational, construction, logging, and agricultural
equipment. From 1988 to 1997, the contribution of onroad sources decreased from 61 to 57%, whereas the
contribution of nonroad sources increased from 13 to 19%.  In addition, there are several categories, such
as fuel consumption by electric utilities and industry, in which emissions have increased over the same
period. Total CO emissions for the United States were reported to be 66,189 thousand short tons for 1990,
the last year reported in the previous CO air quality criteria document. It can be seen from inspection of
Table 3-3 that the values for 1990 have been revised upward in the interim to 95,794 thousand short tons.
The upward revision in values for 1990 is primarily the result of changes in the methods for calculating
motor vehicle emissions. The MOBILES emissions factor model (U.S. Environmental Protection Agency,
1994) replaced the earlier MOBILE4.1 version (U.S. Environmental Protection Agency, 1991a).  The most
significant change was in using IM240 data to replace FTP testing of recruited in-use vehicles. This change
yielded significantly higher exhaust emissions, especially for older, higher mileage vehicles. In addition,
changes were made in methods for calculating inputs to the model (e.g., temperatures, operating mode) and
in the method for calculating vehicle miles traveled. Additional differences relate to the use of county-level
                                             3-10

-------
Table 3-3. U.S. Carbon Monoxide Emissions (thousands of short tons)
Source Category
Fuel Combustion Electrical Utility3
Fuel Combustion Industrial1"
Fuel Combustion Residential, Commercial,
Institutional0
Commercial/Institutional Coal
Commercial/Institutional Oil
Commercial/Institutional Gas
Miscellaneous Fuel Combustion (except residential)
Residential Wood (fireplaces and woodstoves)
Residential Other
Chemical and Allied Product Manufacturing11
Ferrous and Nonferrous Metal Processing6
Petroleum and Related Industries'
Other Industrial Processes5
Solvent Utilization
Storage and Transport
Waste Disposal and Recycling
Incineration
Conical wood burner
Municipal incinerator
Industrial
Commercial/institutional
1988
314
669
6,390
15
18
47
55
6,086
168
1,917
2,101
441
711
2
56
1,806
903
19
35
10
38
1989
321
672
6,450
15
17
49
55
6,161
153
1,925
2,132
436
716
2
55
1,747
876
19
35
9
39
1990
363
879
4,269
14
18
44
149
3,781
262
1,183
2,640
333
537
5
76
1,079
372
6
16
9
19
1991
349
920
4,587
14
17
44
141
4,090
281
1,127
2,571
345
548
5
28
1,116
392
7
17
10
20
1992
350
955
4,849
15
18
51
141
4,332
292
1,112
2,496
371
544
5
17
1,138
404
6
15
10
21
1993
363
1,043
4,181
15
18
53
143
3,679
274
1,093
2,536
371
594
5
51
1,248
497
6
14
87
21
1994
370
1,041
4,108
15
18
54
147
3,607
268
1,171
2,475
338
600
5
24
1,225
467
6
14
48
21
1995
372
1,056
4,506
15
19
54
145
3,999
273
1,223
2,380
348
624
6
25
1,185
432
6
15
10
21
1996
394
1,072
4,513
15
19
54
163
3,993
269
1,223
2,378
348
635
6
25
1,203
443
6
15
10
22
1997
406
1,110
3,301
16
19
56
168
2,278
264
1,287
2,465
364
663
6
26
1,242
467
6
16
11
23

-------
Table 3-3 fcont'd). U.S. Carbon Monoxide Emissions (thousands of short tons)
Source Category
Waste Disposal and Recycling (cont'd)
Residential
Other
Open Burning
Industrial
C ommercial/institutional
Residential
Other
Onroad Vehicles
Light-Duty Gas Vehicles and Motorcycles
Light-duty gas vehicles
Motorcycles
Light-Duty Gas Trucks
Heavy-Duty Gas Vehicles
Diesels
Heavy-duty diesel vehicles
Light-duty diesel vehicles
Nonroad Engines and Vehicles
Nonroad Gasoline
Recreational
Construction
1988

800
2
903
21
4
877
NA
71,081
45,553
45,367
186
17,133
7,072
1,322
1,290
32
14,698
12,464
318
401
1989

773
2
870
21
5
845
NA
66,050
42,234
42,047
187
15,940
6,506
1,369
1,336
34
14,820
12,537
321
398
1990

294
27
706
14
46
509
137
57,848
37,407
37,198
209
13,816
5,360
1,265
1,229
36
15,376
13,088
359
355
1991

312
26
722
14
18
516
144
62,074
40,267
40,089
177
15,014
5,459
1,334
1,298
36
15,368
13,065
365
329
1992

324
28
731
15
50
523
144
59,859
39,370
39,190
180
14,567
4,569
1,352
1,315
37
15,652
13,305
370
334
1993

340
29
749
15
52
529
153
60,202
39,163
38,973
190
15,196
4,476
1,367
1,328
38
15,828
13,454
374
348
1994

347
30
755
15
54
533
153
61,833
37,507
37,312
195
17,350
5,525
1,451
1,411
38
16,050
13,638
378
382
1995

351
29
750
15
52
536
147
54,106
33,701
33,500
200
14,829
4,123
1,453
1,412
39
16,271
13,805
382
393
1996

360
30
757
16
53
539
149
53,262
28,732
28,543
189
19,271
3,766
1,493
1,453
35
16,409
13,935
386
400
1997

380
31
772
16
55
545
156
50,257
27,036
26,847
189
18,364
3,349
1,508
1,468
35
16,755
14,242
389
423

-------
Table 3-3 fcont'd). U.S. Carbon Monoxide Emissions (thousands of short tons)
Source Category
Nonroad Engines and Vehicles (cont'd)
Industrial
Lawn and garden
Farm
Light commercial
Logging
Airport service
Recreational marine vessels
Nonroad Diesel
Recreational
Construction
Industrial
Lawn and garden
Farm
Light commercial
Logging
Airport service
Railway maintenance
Recreational marine vessels
Aircraft
Marine Vessels
Railroads
1988

1,207
5,866
92
3,219
31
144
1,185
1,129
3
634
150
23
176
44
58
35
2
4
931
56
118
1989

1,227
5,929
63
3,223
33
147
1,195
1,149
3
655
148
25
177
45
58
31
2
4
955
59
121
1990

1,387
6,501
213
2,428
32
116
1,698
1,180
3
677
146
27
178
46
58
38
2
4
904
83
121
1991

1,350
6,599
170
2,385
33
114
1,720
1,207
3
699
146
30
179
48
58
38
2
4
888
87
120
1992

1,374
6,684
199
2,453
34
118
1,739
1,236
3
721
147
32
180
49
57
38
2
5
901
85
125
1993

1,371
6,770
209
2,472
34
119
1,757
1,268
3
744
149
35
181
51
57
40
3
5
905
81
120
1994

1,404
6,823
175
2,551
36
121
1,769
1,300
3
766
152
38
183
53
56
41
3
5
915
82
114
1995

1,436
6,895
145
2,621
40
129
1,763
1,329
4
788
155
41
184
54
56
39
3
5
942
82
114
1996

1,446
6,949
150
2,658
41
131
1,775
1,330
3
789
156
44
182
56
52
40
3
5
949
82
112
1997

1,510
7,009
152
2,787
44
141
1,788
1,301
3
768
154
47
176
56
45
43
3
5
1,012
85
115

-------
                                      Table 3-3 fcont'd).  U.S. Carbon Monoxide Emissions (thousands of short tons)
Source Category
Miscellaneous
Other Combustion
Structural fires
Agricultural fires
Slash/prescribed burning
Forest wildfires
Other
Total All Sources
1988
15
15


4
10

,895
,895
242
612
,332
,709
NA
116,081
1989
8,153
8,153
242
571
4,332
3,009
NA
103,480
1990
11,208
11,207
164
415
4,668
5,928
32
95,794
1991
8,751
8,751
166
413
4,713
3,430
28
97,790
1992
7,052
7,052
168
421
4,760
1,674
30
94,400
1993
7,013
7,013
169
415
4,810
1,586
34
94,526
1994
9,614
9,613
170
441
4,860
4,114
28
98,854
1995
7,050
7,049
171
465
4,916
1,469
28
89,151
1996
9,463
9,462
142
475
4,955
3,863
27
90,929
1997
9,568
9,568
143
501
5,033
3,863
28
87,451
CO
Major Subcategories:
"Coal burning
bNatural gas burning
'Residential wood burning
"tarbon black manufacturing
Terrous metal production
'Refineries
gWood, paper, and pulp

Notes:
NA = not available. For several source categories, emissions either prior to or beginning with 1985 are not available at the more detailed level but are contained in the more aggregate
estimate. "Other" categories may contain emissions that could not be allocated accurately to specific source categories. To convert emissions to gigagrams (thousands of metric
tons), multiply the above values by 0.9072.

Source:  Adapted from U.S. Environmental Protection Agency (1998).

-------
statistics for vehicle registration, as well as the use of temperature data from individual counties. The value
of 6.2 x 107 short tons (56 Tg) shown in Table 3-3 for emissions from onroad vehicles for 1991 may be
compared to a value of 4.0 x 107 short tons (36 Tg) derived from remote sensing of vehicle exhausts for
1991 (Bradley et al., 1999).
       In addition, it should be noted that Table 3-3 does not include formation of secondary CO, such as
from the oxidation of isoprene. Annual emissions of isoprene in the contiguous United States are about
17.2 Tg/year (Pierce and Dudek, 1996).  A source of CO of 7.1 Tg/year can be calculated using the
conversion factor of 0.20 for carbon in isoprene to carbon in CO estimated by Jacob and Wofsy (1990).
This value would add about 9% to the estimated U.S. emissions for CO in 1995, shown in Table 3-3.  The
oxidation of anthropogenic and other natural NMHCs may supply an additional 2 to 3 Tg CO per year.
       A number of techniques, such as sampling in tunnels and the remote sensing of individual motor
vehicle emissions have been applied in the past several years at a number of locations throughout the United
States to test CO emissions estimates and to derive emissions factors (i.e.,  emissions per unit distance
traveled). Two major points have been realized on the basis of these studies:  first, that a small percentage
of motor vehicles are responsible for  most  of the emissions, and, second, that CO and hydrocarbon
emissions had been systematically underestimated by as much as a factor of two in emissions factor models.
As a result of these studies, a number of revisions have been made to emissions inventories.
       Roadside remote sensing data indicate that over  50% of CO and NMHC emissions are produced
by less than 10% of the vehicles (Lawson et al., 1990; Stephens and Cadle, 1991). These "superemitters"
are  typically older, poorly maintained vehicles.  Bishop and  Stedman (1996) also found that the most
important variables  governing CO emissions are fleet age and owner maintenance.  There are also a
surprising number of newer vehicles that are classified as superemitters.  Possible reasons are related to
tampering with emissions control systems to improve milage, the use of contaminated fuels that may
interfere with the proper operation of emissions control systems, and the lack of maintenance of emissions
control  equipment and the failure  of that equipment.   In addition to the above activities, so-called
"off-cycle" operations also can result in enhanced emissions  relative to those conditions for which
emissions testing usually is done. For example, rapid accelerations have been shown to increase emissions
relative  to less stressful driving modes.
       Roadside remote sensing of motor vehicle emissions has been used to evaluate the effectiveness of
inspection and maintenance programs in a number of locations (Zhang et al., 1996; Stedman et al., 1997,
1998). These studies generally have yielded disappointing results, indicating undetectable or smaller than
expected effects of inspection on vehicle emissions. Detailed analyses implicate behavioral responses, such
as shopping for a passing inspection (Bishop and Stedman, 1996) and reregistering nonconforming vehicles
in neighboring counties (Stedman et al., 1997, 1998). Roadside emissions  data also have been used to
evaluate the effects of reformulated fuels on emissions. Remote sensing (Bishop and Stedman, 1990) and
tunnel measurements (Gertler et al.,  1999) both indicate  fleet CO reductions in the 15 to 20% range.
       A comparison of emissions factors computed on the basis of tunnel measurements in Van Nuys, CA,
during the South Coast Air Quality  Study in 1987 with those  calculated by emissions inventory models
indicated that CO emissions were underpredicted by emissions models (i.e., the Emissions Factor 7C
[EMFAC7C] model, which is similar to MOBILE3) by a factor of 2.7, and hydrocarbon emissions were
underestimated by a factor of 3.8 (Ingalls et al., 1989; Pierson et al., 1990). However, in a reinterpretation
of the Van Nuys tunnel data using more recent versions of these models, Pollack et al. (1998) found that
emissions factors calculated using MOBILES a were a few percent greater than the ambient tunnel  data
indicated (21.3 versus 20.9 g/mi), compared with a factor of two difference using EMFAC7F (9.6 versus
20.9 g/mi).  Likewise, a comparison of emission factors computed on the basis of measurements in the Fort
McHenry, MD,  and Tuscarora, PA, tunnels with those calculated by emissions models (MOBILE4.1 and
MOBILES)  indicated that   both versions of the MOBILE model gave predictions within ±50% of
observations most of the time (Pierson et al., 1996). However, it should be noted that emissions in tunnels

                                             3-15

-------
arise from vehicles in warmed-up or hot-stabilized operation.  Cold and hot start emissions, which are
important components of the emissions inventory, generally are not present in tunnels and, thus, are not
evaluated in these studies.
       Comparisons of ambient air quality data with predictions of emissions factor models have been
made for conditions when ambient concentrations result primarily from local emissions with minimal
photochemical processing and minimal transport from locations with different source characteristics. The
optimal time to obtain such conditions is during the early morning, when ambient concentrations of CO,
non-methane organic  compounds (NMOCs), and NOX typically peak and are dominated by local mobile
source emissions (Fujita et al,  1992). These comparisons have been performed in California for the Los
Angeles Basin (Fujita et al., 1992), the San Joaquin Valley, and the San Francisco Bay area (Magliano et al.,
1993), and  for the Lake Michigan air quality region (Korc et al., 1993). A fairly consistent picture of
underpredictions of ambient CO to NOX and NMOC to NOX ratios by emissions factor models, after
allowing  for the effects of atmospheric processing and transport, emerged from these studies. In the Los
Angeles Basin study, ambient CO  to NOX ratios were  factors of 1.3 to 2.9 higher than corresponding
emissions inventory ratios during summer, and factors of 1.2 to 2.4 higher than predicted by emissions
models during fall.  In the San Joaquin Valley study, ambient CO to NOX ratios ranged from factors of 1.1
to 7.2 higher than predicted by emission models.  In the Lake Michigan area study, ambient CO to NOX
ratios ranged from factors of 1.7 to 4.7 higher than predicted by emissions models. However, more recent
comparisons between ambient and emission inventory CO to NOX ratio for Los Angeles and the San Joaquin
Valley show better agreement than in the previous studies mentioned above (Croes et al., 1996; Ipps and
Popejoy,  1998; Haste et al., 1998). These improvements have arisen largely through the process of model
development, evaluation, and further refinement.
       Stationary sources account for approximately 23% of nationwide CO emissions shown in Table 3-3.
Indoor sources represented in Table 3-3 by residential combustion of wood and other fuels account for only
about 3% of annually averaged, nationwide CO emissions. However, on a local basis where wood burning
is widespread, these sources can account for significant fractions of the CO present in ambient air.  Khalil
and Rasmussen (1988b,  1989) have shown that during the winter in Medford, OR, and in Olympia, WA,
the contribution of wood burning to CO concentrations was of comparable importance to automobiles.
Khalil and Rasmussen (1999) have found that biomass burning, which takes the form of agricultural burning
during the fall and residential wood burning during the winter, accounts for 20 to 40% of excess over
nonurban background CO concentrations.
3.4  Carbon Monoxide Concentrations in Ambient Air
       The U.S. Environmental Protection Agency's (EPA's) Aerometric Information Retrieval System
(AIRS) receives data from the National Air Monitoring Stations (NAMS) and the State and Local Air
Monitoring Stations (SLAMS). Current NAAQS define 1- and 8-h average concentrations that should not
be exceeded more than once per year.  The standards are met if the second-highest 1-h value is less than
or equal to 35 ppm (40 mg/m3), and the second-highest, nonoverlapping 8-h value is less than or equal to
9 ppm (10 mg/m3). Nationwide trends in ambient CO concentrations are presented in Section 3.4.1, diurnal
variations in ambient CO concentrations are presented in Section 3.4.2, and amore detailed characterization
of the spatial and temporal variability in ambient CO concentrations in selected urban areas is presented
in Section 3.4.3. The analyses in Section 3.4.3 were performed for the Denver, CO(Shadwicketal., 1997);
Los Angeles; New York City, NY; and Phoenix, AZ, Metropolitan  Statistical Areas (MSAs) (Shadwick
et al., 1997, 1998a,b,c) and for Fairbanks, AK.
                                            3-16

-------
                                          NAMS
                                          SLAMS
                                          Other
Figure 3-3.  Locations of sites in the nationwide ambient CO
monitoring network, 1997.
Source: U.S. Environmental Protection Agency (1998).
                                                    3.4.1  Nationwide Trends in Ambient
                                                          Carbon Monoxide Concentrations
                                                           In 1997, 538 monitoring sites reported
                                                    ambient CO air quality  data to EPA's AIRS.
                                                    Most  CO monitoring stations in the United
                                                    States are located in larger urban areas.  Figure
                                                    3-3  displays the geographic locations of the
                                                    monitoring sites reporting CO data to AIRS for
                                                    1997.  On the  map, the  sites are identified as
                                                    NAMS, SLAMS, or "Other". The NAMS were
                                                    established by EPA  to  ensure  a long-term
                                                    national   network  for  urban-area-oriented
                                                    ambient monitoring and to provide a systematic,
                                                    consistent database for air quality comparisons
                                                    and trends analysis. The  SLAMS allow state or
                                                    local governments to develop networks tailored
                                                    for their  specific  monitoring  needs.   These
NAMS and SLAMS sites conform to uniform criteria for monitorsiting, instrumentation, and quality
assurance.  "Other" monitors may be special purpose monitors, monitors at industrial sites, monitors on
tribal lands, etc. Although state and local air programs may require  extensive monitoring to document and
measure the local impacts of CO emissions, only two NAMS sites are required in urbanized areas with
populations greater than 500,000. Two categories of NAMS sites are required:  (1) peak concentration areas
(microscale), such as major traffic corridors, street canyons, and major arterial streets, and (2) areas with
high population and traffic densities (middle scale or neighborhood scale).
       Twenty-seven sites in  15 areas failed to meet the 8-h CO NAAQS in at least 1 year of the 5-year
period, 1993 to 1997. In 1996, only six of the sites shown in Figure 3-3 failed to meet the 8-h standard of
9 ppm, and none of the 538 monitoring sites exceeded the 1-h standard of 35 ppm.  The locations of these
27 sites and the second-highest 8-h CO concentrations and the number of exceedances by year are given
in Table 3-4.
       Figure 3 -4 shows the consistent, downward trend in the nationwide composite average of the annual
second-highest 8-h CO concentration during the 20-year period, 1978 through 1997. This statistic relates
directly to the averaging time and form of the current CO NAAQS and complies with the recommendations
of the Intra-Agency Task Force on Air Quality Indicators (U.S. Environmental Protection Agency, 1981).
The dashed curve in Figure 3-4 tracks the trend in the composite mean of the annual second-highest 8-h
average concentration for 184 monitoring sites that reported ambient air quality data in at least  17 of the
20 year period, 1978 to  1997.  All monitoring sites are weighted equally  when computing the nationwide
composite mean concentration. This selection criterion maximizes the number of sites available for trend
analyses.  This subset of sites yields good geographical coverage with sites  from  more than 90 cities in
39 states. Each year, site leases are lost, or sites are discontinued, and new sites come online; therefore, the
184 long-term-trend sites compose only one-third of the currently  active CO monitors. The solid line in
Figure  3-4  shows the trend in the composite  mean for a larger database of 368 sites that have reported
ambient CO monitoring  data in at least 8 of the past  10 years.   Missing annual second-highest  CO
concentration  data for the second through ninth years are estimated by linear interpolation  from the
surrounding years. Missing endpoints are replaced with the nearest valid year of data. This latter procedure
explains the discrepancy between the two curves in 1988. Specific computational details are described
elsewhere (U. S. Environmental Protection Agency, 1998). This larger data set permits the examination of
the intersite variability in peak CO concentrations. Figure 3-5 presents the 10th, 50th, and 90th percentile
concentrations and the composite mean concentrations across these 368 sites.  The 10th, 50th, and 90th CO
                                            3-17

-------
Table 3-4.  Sites Not Meeting the 8-Hour Carbon Monoxide National Ambient Air Quality Standard, 1993 to 1997
1993
Location
Anchorage, AK


Denver, CO
Detroit, MI
El Paso, TX

Fairbanks, AK


Flathead Co., MT
Calexico, CA

Jersey City, NJ
Las Vegas, NV/AZ
Los Angeles-Long Beach, CA


Airs Site ID
020200017
020200018
020200037
080310002
261630014
481410027
481410044
020900002
020900013
020900020
300290045
060250005
060250006
340175001
320030557
060371002
060371201
060371301
2nd
Maxa
7.7
9.7
9.9
10.4
5.6
7.4
10.6
10.1
9.6
9
n/d
n/d
n/d
7.6
9.9
8.1
8
13.8
No.
Exc.b
0
2
2
2
0
0
2
5
2
1
n/d
n/d
n/d
0
3
0
0
20
1994
2nd
Max
8.3
8.6
11
8.2
10.3
7.1
7.6
10.2
8.5
9.8
n/d
12.9
n/d
10.7
10.6
10.2
9.9
15.3
No.
Exc.
0
0
2
1
2
1
0
3
1
3
n/d
10
n/d
3
5
5
o
5
24
1995
2nd
Max
7.6
7.4
8.4
9.5
5.6
7.9
7.5
11.8
10.6
11.6
6.5
19.7
n/d
8.1
9.2
11
9.4
11.6
No.
Exc.
0
0
0
2
0
0
0
9
o
3
7
0
15
n/d
1
1
5
1
14
1996
2nd
Max
9.6
8.7
10.5
7.3
4.5
10.3
9.1
8.6
8.4
8.6
11.1
14.1
7.8
6.7
10.1
8.5
6.7
14.5
No.
Exc.
o
3
0
3
0
0
2
1
1
0
0
2
9
0
0
3
0
0
22
1997
2nd
Max
6.8
7.1
7
5.5
n/dc
7.9
7.2
12.1
10.8
10.6
4.9
16.7
9.6
6.7
6.3
7.2
7.7
15
No.
Exc.
0
0
0
0
n/d
1
1
3
2
4
0
12
2
0
0
0
1
12

-------












CO

CO



Table 3-4 (cont'd). Sites Not Meeting the 8-Hour Carbon Monoxide National Ambient Air Quality Standard,
1993 1994 1995 1996

Location Airs Site ID
060375001
Newark, NJ 340390003
Phoenix-Mesa, AZ 040130019
040130022
Provo-Orem, UT 490490004
Spokane, WA 530630040
530630044
Steubenville-Weirton, OH-WV 540290009

540290011
a Annual second-highest nonoverlapping 8-h average
b Number of exceedances of the 8-h CO NAAQS.
0 n/d = no data.
2nd
Maxa
9.6
6
8
n/d
9.6
9.8
11.8
8.3

9.4
No. 2nd No. 2nd No. 2nd No.
Exc.b Max Exc. Max Exc. Max Exc.
2 11.3 6 8.7 0 10.5 5
0 11.3 2 7.7 0 60
0 9.6 2 8.4 0 8.2 0
n/d n/d n/d 9.9 3 10 2
2 9.3 1 7.1 0 9.1 1
2 8.1 0 8.4 0 9 1
4 8.8 0 11.2 4 8.4 1
1 9.6 2 6 0 6.2 0

1 17.1 5 6.7 1 3.6 0
1993 to 1997
1997
2nd No.
Max Exc.
7.9 1
5.1 0
7 0
7.8 1
n/d n/d
6.3 0
n/d n/d
8.8 1

2.5 0
CO concentration.


Source: U.S. Environmental Protection Agency's Aerometric


Information Retrieval System (AIRS).




-------
E
Q- 10-
Q.
O 6-
O
                       Year
Figure 3-4.  Nationwide composite average of the annual
second-highest 8-h CO concentrations, 1978 to 1997.
Source: U.S. Environmental Protection Agency (1998).
                                                    15-
                                                  8
                                                   o
                                                   O  5
                                                   O
                                                   o
                                                         Key:
                                                               — 90th Percentile

                                                               — Mean
                                                               — Median
                                                               — 10th Percentile
                                                                                     368 Sites
                                                                                       NAAQS
                                                            88  89  9O 91 92 93 94  95 96 97
                                                                          Year
                                                  Figure 3-5. Variability in the annual second-highest 8-h
                                                  CO concentrations across all sites in the United States
                                                  reporting at least 8 years of data,  1988 to 1997.
                                                  Source:  U.S. Environmental Protection Agency (1998).
                       Year
                                                  concentrations  for  each  year  are  indicated,
                                                  respectively, by the bottom, middle, and top lines of
                                                  each box. For example, 10% of the 368 trends sites
                                                  reported 1988 second-highest 8-h CO concentrations
                                                  lower than the bottom of the first bar in Figure 3-5.
                                                  The yearly composite mean across all 368 sites is
                                                  indicated by the"x" in each bar. Figure 3-6 shows
                                                  trends in CO concentrations in each of the different
                                                  sampling environments (urban, suburban, and rural
                                                  sites).   As  can be  seen from Figure  3-6,  the
                                                  downward trend  in  ambient  CO concentrations
                                                  occurred at monitoring sites in urban, suburban, and
                                                  rural environments.  An interesting  feature of the
                                                  data shown in Figures 3-4 to 3-6 is the increase in
                                                  CO concentrations from 1993 to 1994, which is the
                                                  only year over year  increase except  for 1985 to
                                                  1986. The increase corresponds to an increase in
                                                  mobile source and wildfire emissions presented in
Table 3-3.  The decrease in ambient CO concentrations measured in populated areas over the past decade
also is reflected at least at one continental background site at Shenandoah National Park,  VA (Hallock-
Watersetal, 1999).

3.4.2  Circadian Patterns in Carbon Monoxide Concentrations
       The circadian variation in winter time, composite, hourly CO concentrations from 1987 through
1996  is shown in Figure  3-7 (Cohen and Iwamiya, 1998).  It can be seen that hourly mean  CO
concentrations peak during the morning rush hours (7 to 9 a.m.).  This peak results primarily from  CO
emitted into the relatively shallow morning boundary layer by motor vehicles (e.g., Fuj ita et al., 1992). The
CO concentrations decline towards mid-afternoon, as the height of the atmospheric mixing layer increases
and then increase again with the onset of the evening rush hour. Carbon monoxide concentrations fall off
less rapidly after the afternoon peak because the  mixing layer height decreases during evening and
Figure 3-6. Composite average of the annual second-
highest 8-h CO concentrations for rural, suburban, and
urban sites, 1988 to 1997.
Source: U.S. Environmental Protection Agency (1998).
                                             3-20

-------
                          Hour
                                                     nighttime.  There is a general decrease in CO
                                                     concentrations during the night because of a
                                                     lack  of  fresh  emissions  combined  with
                                                     processes such as mixing with CO-poor areas
                                                     and deposition to the surface. The downward
                                                     trend in CO concentrations from 1987 to 1996
                                                     is apparent for all times of the day.  Especially
                                                     notable is the  decease in 7 to  9 a.m. CO
                                                     concentrations, which is consistent with the
                                                     decrease in motor vehicle emissions that was
                                                     noted earlier for the same period.  During the
                                                     period from 1987 to 1996, the 24-h nationwide
                                                     composite average CO concentration decreased
                                                     from 2.0 to 1.2 ppm.
       The circadian pattern of 8-h average CO concentrations that exceed 9.5 ppm is somewhat different
from the hourly average concentration pattern shown in Figure 3-7.  One way to estimate the time of day
when 8-h average CO concentrations are likely to exceed 9.5 ppm is to total the numbers of these events
by hour of the day over the course of a year.  Hourly average CO  data were taken from AIRS to construct
running 8-h averages for 1996.
       In the previous CO criteria document (U.S. Environmental Protection Agency, 1991b), which
summarized ambient monitoring data through 1988, six stations in six major cities, with prominent patterns
of 8-h exceedances, were selected to demonstrate the variability in the circadian patterns of exceedances
using the above technique; some peaked in daylight hours and others in nighttime hours. Five of those six
monitors are still in operation; Table 3-5 summarizes and compares their 1988 record with 1996 data.
 Figure 3-7. Diurnal variation of nationwide composite hourly
 average CO concentrations for winter (December to February),
 1987to1996.
 Source: Cohen and Iwamiya (1998).	
  Table 3-5. Running-Average Exceedances of the 9-ppm 8-Hour Carbon Monoxide Standard,
                                      1988, 1996, and 1997
Location
Lynwood, CA
Hawthorne, CA
Las Vegas, NV
New York City, NY
Steubenville, OH
Spokane, WA
AIRS Site ID
060371301
060375001
320030557
360610081
390811012
530630040
1988
392
163
102
123
152
169
1996
110
19
3
NAa
0
1
1997
62
2
0
0
0
0
1997 Data Completeness
Jan. -Dec. 1997
Jan. -Dec. 1997
Jan. -Mar. 1997
Jan. -June, Oct.-Dec. 1997
Jan.-Sep. 1997
Jan. -Dec. 1997
 a NA = not available

 Note: Asof June 14,1999,the 1997 data for the Las Vegas, New York City, and Steubenville stations were incomplete. Based
 on 1996 data, Steubenville probably is not a problem; the New York City station is missing summer months, and the "zero"
 is probably warranted; and the Las Vegas station is missing the fall 1997 months; hence, the conclusion is indeterminate.
       Note that, in Table 3-5 and in the analysis of 1996 data shown in Table 3-6, a "running-average"
exceedance is defined as any hour that culminates in an 8-h average higher than 9.5 ppm. This definition
differs from that used in the construction of Table 3-4 in that the number of nonoverlapping exceedances
was used in Table 3-4.  A formal violation of the 8-h standard occurs when, in a given year, a second 8-h
                                              3-21

-------
   Table 3-6. Annual Circadian Pattern of
     8-Hour Average Carbon Monoxide
   Concentrations Culminating in Values
   Greater Than 9.5 ppm in Lynwood and
        Hawthorne, CA, During 1996
Ending of 8-h
Period
Midnight
1 a.m.
2a.m.
3 a.m.
4 a.m.a
5 a.m.
6 a.m.
7 a.m.
8 a.m.
9 a.m.
10 a.m.
11 a.m.
Noon
1 p.m.
2p.m.
3 p.m.
4p.m.
5 p.m.
6p.m.
7p.m.
8p.m.
9p.m.
10p.m.
11 p.m.
Total

Lynwood
9
9
10
11
8
7
8
7
8
8
6
3
2
1
1
0
0
0
0
0
1
1
4
6
110

Hawthorne
0
0
2
3
4
4
2
2
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
19
 Calibrations normally are done at 4 a.m., thus values are
  interpolated.
average exceeds 9.5 ppm but does not overlap the
first 8-h exceedance.  Exceedances culminating in
any hour are treated here because an individual's
cumulative exposure to a level greater than 9.5 ppm
could occur in any hour.
        At the Lynwood station, the running-average
exceedances  have  declined  from 392 in 1988 to
110 in 1996 (22 nonoverlapping exceedances); the
majority of exceedances in 1996 occurred in the hours
between midnight and sunrise, as they had in 1988.
They occurred in the months of January, February,
November, and December.
        At the Hawthorne station, running-average
exceedances have declined from 163 in 1988 to 19 in
1996 (five nonoverlapping exceedances).  These are
clustered around sunrise  when  dispersion is most
likely to be at a minimum. The exceedances at this
station also occur in the winter quarter.
        Only a small number of stations have several
running-average  exceedances;  however,   these
recurrent high concentrations are attributed to unusual
local situations. A prime example is the monitoring
station in  Calexico, which is several blocks away
from a major U.S.-Mexico border crossing and the
route leading to it; nine nonoverlapping exceedances
were recorded in 1996. Reportedly, there are often
long lines  of idling vehicles waiting to  cross the
border, including vehicles of Mexican registration
that  are not  equipped with the emission control
equipment  required on vehicles sold in the United
States. Such situations will need to be addressed on
a local, case-by-case basis.
3.4.3  Characterization of the Spatial and Temporal Variability in Carbon Monoxide
       Concentrations in Selected U.S. Cities
       The spatial and temporal variability of ambient carbon monoxide was characterized in four MSAs
in the continental United States (New York, Denver, Phoenix, and Los Angeles) and for Fairbanks. These
five urban areas were chosen to  characterize the spatial and temporal variability in CO in widely different
geographic regions.  New York City is characterized by urban canyons.  Denver is a rapidly growing,
high-altitude city.  Phoenix is  a rapidly growing city in an arid environment. Los Angeles is characterized
by emissions which are confined to a mountain basin.  Fairbanks is located in a mountain valley with a
much higher potential for air  stagnation than the continental U.S. cities. Each of these cities has been in
nonattainment of the 8-h NAAQS for CO at some time within the past 5 years. In addition, the four cities
in the continental Unites States have been the locations of studies either characterizing personal exposure
to CO or relating health outcomes to air pollution concentrations.
       Hourly average CO data obtained from EPA's AIRS were used to calculate running 8-h averages
for 1986 to 1996. Only valid hourly average values were used to compose the 8-h average. In the case that
less than six valid hourly average values were used to compose the 8-h average, the 8-h average was set to
                                             3-22

-------
a missing value.  The six valid hourly average values in an 8-h window corresponds to 75% data capture
in the 8-h window.
       The 24 running 8-h averages assigned to a day were used to compute the daily maximum 8-h
average. A daily maximum 8-h average was considered to be valid if at least 18 of the 8-h running averages
for the day were valid as described in the preceding paragraph. The 18 valid 8-h running averages in a day
corresponds to a 75% data capture. In the case that a valid daily maximum 8-h running average could not
be computed, a  missing value was assigned to the daily maximum 8-h average.  The summary statistics
were computed without regard to data capture. Summary statistics (aside from the total number of
observations) should be regarded as representative if at least 75% of the possible data values were valid.
Statistics on central tendency and correlation were tabulated for all of the sites in each urban area for both
the hourly and 8-h running averages. The statistics were analyzed by year, season, day of week, and hour
of the day.
       Figures 3-8 through 3-12 graphically represent air quality data from the Denver area CO monitoring
sites, then similar data are presented for CO monitoring sites in Los Angeles (Figures 3-13 through 3-19),
New York (Figures 3-20 through  3-24), and Phoenix (Figures 3-25 through 3-29) metropolitan areas.
Analysis of ambient CO data obtained in these four geographically diverse MS As has shown that urban CO
concentrations have decreased over the past 10 years.  However, there have been instances where the
downward trend has reversed itself on a year-to-year basis.  Although the number of violation days has
declined for these cities, and the seasonally averaged peak concentrations generally do not exceed 8 ppm,
at least one exceedance of 9 ppm for the maximum daily 8-h average for CO occurred in 1995/1996 (the
final year in this analysis) in all four of these cities.
       Data obtained from  different monitoring sites within a given MSA show a  large degree of
variability.  During 1996, for example, annual mean CO concentrations ranged from 0.4 to 1.5 ppm in the
Denver MSA, 0.4 to 3.2 ppm in the Los Angeles Consolidated Metropolitan Statistical Area (CMSA), 0.6
to 3.7 in the New York CMSA, and 0.7 to 3.4 in the Phoenix MSA. A more detailed analysis of the spatial
variability in CO concentrations in these four MSAs in the continental United States and in Fairbanks is
given in Appendix 3 A.  Carbon monoxide concentrations during the cold season (November through
February) range from 5 to 20% higher than the annual average in each MSA.  However, it should be noted
that, despite decreasing CO concentrations, the nature of the diurnal and seasonal variation observed at each
monitoring site has remained remarkably constant over the 10-year period covered in this analysis. At all
the  sites investigated here, it is clear that the diurnal and seasonal  variations in CO observed in these
metropolitan areas result largely from the interaction between motor vehicle emissions and meteorological
parameters that, at times,  can  be conducive to the  buildup of CO near the surface.   The  diurnal
concentration profiles in most cases show a very distinctive two-peaked structure for weekdays. The peaks
correspond to both the morning and evening rush hour commutes. Frequently, the morning peak is higher
than the evening peak at any given site because the height of the mixed layer is much lower during the
morning, thus inhibiting vertical mixing that would have diluted CO.  In the late afternoon and into early
evening, increased atmospheric turbulence resulting from solar heating raises the height of the mixed layer,
resulting in generally lower CO concentrations compared with those of the morning.
       Regional differences in atmospheric processes also may play a role in producing the nighttime
behavior of CO  observed at numerous sites in the Los Angeles and Phoenix MSAs compared with either
the nationwide composite average diurnal cycle of CO shown in Figure 3-7 or other locations, such as the
Denver or New York MSAs.  Colucci and Begeman (1969) suggested that the higher concentrations found
in Los Angeles than in Detroit, MI, or New York were caused, at least in part, by more frequent temperature
inversions and lower wind speeds in Los Angeles. In the Los Angeles and Phoenix metropolitan areas, CO
concentrations often remain until midnight at levels reached during the evening rush hour. Then, although
these  concentrations  gradually  diminish  throughout  the  night,  they  do not  drop to  the   low
afternoon concentrations (typically no more than 1 to 2 ppm and often less than that amount) before they

                                             3-23

-------
                                        Greeley
            BouIdeH-Marine
       Arvada
®
                           Welby  f/;:  "3=
                   3 ^Denier-Broadway
                      '"*"1---         J
  ^Denver-Juliafi
                        IJPiri¥er-Albion

                            Littleton
Population Density
    0 to 300 persons/km2
    301 to 1000 persons/km2
    1001 to 2000 persons/km2
!    2001 to 5000 persons/km2
    5001 to 10,000 persons/km2
    over 10,000 persons/km2
 0  CO Monitoring Sites
/\/ Counties
1~n States
    Interstates/Divided Highways
    Major Roads
    Water
Figure
10
V
3-8.
0

-------
                2,9 S6
                 1987
                  198S
                   1989
                    199O
                 Year-1.991.
                        1992
                         1993
                                      Color Scale  in  PPM
Figure 3-9.  Average diurnal variation in  CO at the Denver-Broadway site for weekdays during the winter season
(November through February). The abscissa shows the time of day from midnight to midnight, the ordinate shows years
from the winter of 1986-87 through  1995-96, and the z-axis shows CO concentration in parts per million.	
                 Month
                                         3456789

                                       Color  Scale in  PPM
Figure 3-10.  Monthly average diurnal variation in CO at the Denver-Broadway site for weekdays from May 1986 through
May 1987.  The abscissa shows the time of day,  the ordinate shows the month of the year, and the z-axis shows CO
concentration in parts per million.	
                                                  3-25

-------
                    Month
                               MaySS .*""" 3am
                                   mian
                                                    noon
                                                 9am
                                             6am     Hour-
                                   2345678

                                    Color- Scale  in PPM
Figure 3-11. Monthly average diurnal variation in CO at the Denver-Broadway site for weekdays from May 1995 through
May 1996.  The abscissa shows the time of day, the ordinate shows the month of the year, and the z-axis shows CO
concentration in parts per million.
                    1990  1991
                     YEAR
                    1990  1991
                     YEAR
                           1992  1993  1994  1995  1996
                                             -H-- P5C
                                             •••&••• P25
                                                       begin to increase again because of the morning
                                                       rush hour.  This pattern is shown quite well in
                                                       Figure 3-26, which depicts the seasonal diurnal
                                                       concentration profile for the Central Phoenix
                                                       site.  Comrie and Diem (1999) found that the
                                                       timing and strength of the nocturnal low-level
                                                       thermal   inversion  is   the   dominant
                                                       meteorological   factor   affecting   CO
                                                       concentrations in Phoenix.
                                                          In general, the highest values of ambient CO
                                                       were found during the wintertime (defined as
                                                       the months of November through February) in
                                                       all of the MSAs included here.  During colder
                                                       seasons, there is a higher incidence of enhanced
                                                       stability in  the atmospheric boundary layer,
                                                       inhibiting   vertical  mixing  (dilution)   of
                                                       emissions from the  surface.  There were a few
                                                       sites in the New York metropolitan area where
                                                       a wintertime peak in CO was not discernable;
                                                       the site  on Flatbush Avenue in  Brooklyn
(Figure 3-21) is an excellent example of this. It is not clear without further analysis what combination of
seasonal variations in emissions and meteorological parameters gave rise to this result.
        A map of Fairbanks showing the locations of the CO monitoring sites is shown in Figure 3-30.
Fairbanks is located in the Tanana Valley, in the interior of Alaska and exhibits a number of significant
differences in climatology from the four MSAs described previously.  Although the areas in the continental
Figure 3-12. Central tendency statistics for the daily 8-h max
CO concentration at the Denver-Broadway site  during the
winter season from 1986 to 1995. The top graph  shows box
plots (with  10,  25, 50, 75,  and 90 percentile values) for the
entire time series. Each circle (outlier) or diamond (extreme)
refers to an individual observation that is either three or four
standard deviations (SDs) from the mean, and the horizontal
line shows the current 8-h  NAAQS for CO.  The lower graph
again shows the 25, 50, and 75 percentile values  (P25, P50,
and P75, respectively) from the upper graph.
                                              3-26

-------
                                                         Barstow©
               ;i»l$^H^r^r""""S..!:...>  v
                                          ^«v6rs«we
      Population Density
           0 to 300 persons/km2
           301 to 1000 persons/km2
           1001 to 2000 persons/km2
      11111 2001 to 5000 persons/km2
           5001 to 10,000 persons/km2
           over 10,000 persons/km2
                                     0
     CO Monitoring Sites
     Counties
     States
     Interstates/Divided Highways
     Major Roads
     Water
      10  0  10  20 3040km
/Ubers Projection  parameters
aatiHn none    39 30 0.000
spheroid darke1866 45 30 0.000
         -96 00.000
         23 00.000
         0.00
         0.00
                                                     Produced November 15th, 1998
Figure 3-13.  Map of Los Angeles showing locations of CO monitoring sites.
                                    3-27

-------
                                       1995 ^T  3am
                                         midn
                                   0123456
                                    Color  Scale  in  PPM
                                                                            PPM
Figure 3-14. Average diurnal variation in CO at the Los Angeles-Lynwood site for weekdays during the winter season
(November through February).  The abscissa shows the time of day from midnight to midnight, the ordinate shows years
from the winter of 1986-87 through 1995-96, and the z-axis shows CO concentration in parts per million. _
    1986  1987  1988  1989  1990  1991  1992   1993   1994  1995

                     YEAR
                    •990  1991  1992  '993  1994  1995

                     YEAR
Figure 3-15. Central tendency statistics for the daily 8-h
max CO concentration at the Los Angeles-Hawthorne
site during the winter season from 1986 to 1995. The top
graph shows box plots (with 10, 25, 50, 75, and 90
percentile  values)  for the entire  time  series.   The
horizontal line shows  the current 8-h NAAQS for CO.
The lower graph  again  shows the 25, 50,  and 75
percentile values (P25, P50, and P75, respectively) from
the upper graph.	
                                                         I14
     1986  1987  1988  1989  1990  1991  1992  1993  1994  1995

                      YEAR
                                                          g  7
Figure 3-16. Central tendency statistics for the daily 8-h
max CO concentration at the Los Angeles-Barstow site
during the winter season from 1986 to 1995.  The top
graph  shows box plots (with 10,  25, 50,  75, and 90
percentile values) for the entire time series.  Each circle
(outlier) refers to an  individual observation that is three
SDs from the mean, and the horizontal line shows the
current 8-h NAAQS for CO. The lower graph again shows
the 25, 50, and 75 percentile values (P25, P50, and P75,
respectively) from the upper graph.
                                                    3-28

-------
                                                                           PPM
                                   0123456789  10
                                   Color  Scale  in  PEM


Figure 3-17. Average diurnal variation in CO at the Los Angeles-Hawthorne site for weekdays during the winter season
(November through February). The abscissa shows the time of day from midnight to midnight, the ordinate shows years
from the winter of 1986-87 through 1995-96, and the z-axis shows CO concentration in parts per million.	
                                         M  —'
                                      1995  ,  3am
                                         mian
                                                 6am
                                                    9am
01234567
 Color  Scale  in
                                                               8   9  10
Figure 3-18. Average diurnal variation in CO at the Los Angeles-El Tom site for weekdays during the winter season
(November through February).  The abscissa shows the time of day from midnight to midnight, the ordinate shows years
from the winter of 1986-87 through 1995-96, and the z-axis shows CO concentration in parts per million.
                                                   3-29

-------
                                                       1383- 1990

                                                                 -., 10
                                   0123456789 10
                                    Color  Seal© in  PPM
 F/gure 3-79. Monthly average diurnal variation in CO at the Los Angeles-Anaheim site for weekdays from May through
 May 1986 to 1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996. On each graph, the abscissa shows the time of day
 from midnight to midnight, the ordinate shows the month of the year, and the z-axis shows CO concentration in parts per
 million.
United States are subject to nocturnal inversions that are more pronounced during the winter season, the
low solar elevation, even at midday, particularly in December and January, result in nocturnal inversion
conditions that can persist 24 h/day in Fairbanks. The median midwinter inversion duration varied from
2 to 4 days in a study of the climatology of surface-based inversions at several sites in Alaska (Point
Barrow, Barter Island, and Kotzebue) and Northern Canada (Alert, Eureka, Mould Bay, Resolute, Sachs
Harbour, and Inuvik) (Bradley et al., 1992). Hence, intense inversions may continue during the hours of
maximum CO emissions (typically rush hour traffic). Downtown mixing heights as low as 10 m and wind
speeds less than 0.5 m/s  have been measured simultaneously  in Fairbanks.   The result is that even
nonindustrial areas of moderate size such as Fairbanks are subject to high CO levels (Bowling, 1986).
       A unique meteorological feature of extreme cold  weather climates like that at Fairbanks is the
formation of ice fog (Robinson and Bell,  1956).  Ice fogs are typically  present at temperatures below
-30 °C. Ice fogs in Fairbanks vary from 10 m to over 100 m in depth, which tends to deepen the mixing
layer. This increases dilution, resulting in lower CO concentrations.  One  study suggests that ice fog days
have only 40 to 50% of the CO concentrations seen on otherwise similar non-ice-fog days (Bowling, 1986).
       The average hourly diurnal CO concentration patterns during the winter months for the period from
1986 through 1995 are shown in Figure 3-31 for the Federal Building site. A downward trend in the hourly
average values is seen over the 10-year period. It is particularly interesting to note the shape of the diurnal
pattern in Figure 3-31. The daily CO concentration minimum occurs at approximately 6:00 a.m. This is
followed by a very rapid increase in CO levels that coincides with morning commuting activity. Then, at
approximately 9:00 a.m.,  the rate of change in CO concentrations remains positive but decreases in
magnitude (i.e., the concentrations are still increasing, but  at a lower rate). At about 4:00 p.m., the rate of
change in CO concentrations noticeably increases again. The increase usually  culminates with the daily
1-h average maximum CO value occurring approximately between 5:00 and 6:00 p.m. After the occurrence
of the daily CO maximum at approximately 6 p.m., the concentrations rapidly decrease. The decrease in
                                              3-30

-------
                  Hackensack


                    .'


             Pertiytoib
                 •...-Ereettold
     Population Density
          0 to 300 persons/km2
          301 to 1000 persons/km2
          1001 to 2000 persons/km2
         I 2001 to 5000 persons/km2
          5001 to 10,000 persons/km2
          over 10,000 persons/km2
                      0  CO Monitoring Sites
                    /\/ Counties
                    f~n States
                         Interstates/Divided Highways
                         Major Roads
                         Water
      10
10    20  30km
parameters
 09 30 o 000
 45300JOO
 -96 00.000
 23 00.000
 0.00
 0.00
                                                         Produced November 15th, 1998
Figure 3-20.  Map of New York showing locations of CO monitoring sites.
                                        3-31

-------
                                      1986- 1981
                                I  \>.\,  ,„. j	
                                I
                                |\ v/*;'#f&l'%&\
                             Hav        ' !'•->"••%•&'. i£ V-:") y   May
                                                No    *
                                             -f ^midi.
                                   K»y •"£;„, n«>n
                                      1992- 1993
                                                                          jdn
                                      0  1  2 ' 3  4  5	6	7	8	9	10
                                       Color  Scale in PBS

F/gure 3-21 Monthly average diurnal variation in CO at the New York-Flatbush site for weekdays from May through May
1986 to 1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996. On each graph, the abscissa shows the time of day from
midnight to midnight, the ordinate shows the month of the year, and the z-axis shows CO concentration in parts per million.
                                                        Hoar
                                    01234567
                                    Color  Scale  in
                                                                   9  10
Figure 3-22. Average diurnal variation in CO at the New York-Manhattan site for weekdays during the winter season
(November through February). The abscissa shows the time of day from midnight to midnight, the ordinate shows years
from the winter of 1986-87 through 1995-96, and the z-axis shows CO concentration in parts per million.	
                                                   3-32

-------
                                      1986- 1987
                                      1992- 1993
                                      01234567	8	1	10
                                       Color  Scale  in
F/gure 3-23. Monthly average diurnal variation in CO at the New York-Manhattan site for weekdays from May through
May 1986 to 1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996. On each graph, the abscissa shows the time of day
from midnight to midnight, the ordinate shows the month of the year, and the z-axis shows CO concentration in parts per
million.
                                                             1989- 1990
                                                             1995- 1996
                                      0123456  7	3	9	10
                                       Color  Scale  ir< PPM
Figure 3-24. Monthly average diurnal variation in CO at the New York-Morristown, NJ, site for weekdays from May through
May 1986 to 1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996.  On each graph, the abscissa shows the time of day
from midnight to midnight, the ordinate shows the month  of the year, and the z-axis on each graph shows CO
concentration in parts per million.
                                                  3-33

-------
                                      r OE

        Central PhlenWHI
                   South  Phoenix
                N. Miller Rd
                f
      Population Density
          0 to 300 persons/km2
          301 to 1000 persons/km2
          1001 to 2000  persons/km2
          2001 to 5000  persons/km2
          5001 to 10,000 persons/km2
          over 10,000 persons/km2
       ©
    CO Monitoring Sites
    Counties
    States
    Interstates/Divided Highways
    Major Roads
    Water
      10
10km
datum
units meters
spheroid darke1866
parameters
 29 30 0.000
 45300.000
 -96 00.000
 23 00.000
 0.00
 0.00
                                                  Produced November 15th, 1998
Figure 3-25. Map of Phoenix showing locations of CO monitoring sites.
                                   3-34

-------
                                   1987- 1988
                                                         1989- 1990
                                                                6 EKI
                                                                  midn
                                   1992- 1993
                                                         199S 1996
                                   0  1  23  4  56  7  8 9  10
                                    Color  Scale in PEM
Figure 3-26. Monthly average diurnal variation in CO at the Phoenix-Central site for weekdays from May through May
1986 to 1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996. On each graph, the abscissa shows the time of day from
midnight to midnight, the ordinate shows the month of the year, and the z-axis shows CO concentration in parts per million.
            5  §
                                               Median
                                               Outliers
                           1992  1993  1994 1995
                                                      CO  concentrations  continues  until  the
                                                      concentration minimum occurs at approximately
                                                      6:00 a.m., when, on weekdays, the cycle repeats.
                                                      This  diurnal CO pattern  indicates a situation
                                                      where there are bimodal peaks in concentration
                                                      but inadequate dilution between peaks to reduce
                                                      the CO concentration from one maxima to the
                                                      other. This diurnal pattern contrasts with those
                                                      found in the continental U.S. cities described
                                                      earlier (cf  Figures 3-9, 3-14, 3-17, and 3-18)
                                                      and, indeed, from the nationwide diurnal pattern
                                                      shown in Figure 3-7.
                                                            The central tendency statistics for the daily
                                                      maximum 8-h average CO concentrations for the
                                                      winter  season  from  1986   to  1995  for the
                                                      Fairbanks Federal Building  site  are shown  in
                                                      Figure 3 -32. The median and interquartile range
                                                      have remained relatively unchanged from about
                                                      1989 to 1995, the last complete winter season in
                                                      this  analysis.   Of  particular interest  is the
                                                      apparent upward trend in  the wintertime daily
maximum 8-h average CO concentrations after a brief period  of lower values from about 1990 to 1992.
Since then, however, the number and magnitude of daily maximum 8-h average CO concentrations over
9 ppm (the 8-h average NAAQS for CO) have increased.  This pattern again contrasts with that for the four
                    1990  1991
                     YEAR
Figure 3-27. Central tendency statistics for the daily 8-h max
CO concentration at the Phoenix-East Butler site during the
winter season from 1986 to 1995. The top graph shows box
plots (with  10,  25, 50, 75,  and 90 percentile values) for the
entire time  series. Each circle (outlier) refers to an individual
observation that is three SDs from the mean, and the
horizontal line  shows the current 8-h NAAQS for CO.  The
lower graph again shows the 25, 50, and 75 percentile values
(P25, P50,  and P75, respectively) from the upper graph.
                                               3-35

-------
                                      \, .„---•" noon
                                    May. .  6am
                                     mem
                                       0123  456789 10
                                        Color  Scale  in PPM
Figure 3-28. Monthly average diurnal variation in CO at the Phoenix-West site for weekdays from May through May 1986
to 1987, 1989 to 1990, 1992 to  1993, and 1995 to 1996. On each graph, the abscissa shows the time of day from
midnight to midnight, the ordinate shows the month of the year, and the z-axis shows CO concentration in parts per million.
                                       1987- 1988
                                                             1995- 1996
                                                          M&yi", Sara
                                                            tnian
                                                                     6 pn
                                       0123456789 10
                                        Color  Scale  in  PBM
Figure 3-29.  Monthly average diurnal variation in CO at the Phoenix-South site for weekdays from May through May 1986
to 1987, 1989 to 1990, 1992 to  1993, and 1995 to 1996.  On each graph,  the abscissa shows the time of day from
midnight to midnight, the ordinate shows the month of the year, and the z-axis shows CO concentration in parts per million.
                                                   3-36

-------
                                              o ••
                                        Federal Building
                                             srEE-02—
                                              _
                                    State Office Buiiaiig
                                        ««««« QTTTP 1 "5
                                  o
                          Hunter Elementary
                               '•  SITE 20  	
            Population Density
           10,000 or more  persons/km2
            5,000 to 10,000 persons/km!
            2,000 to 5,000  persons/km.2
              750 to 2,000  persons/km2
              100 to 750   persons/km2
                0 to 100   persons/km2
 *   CO Monitoring Sites
''"'   Counties
HI  States
/V   Major Roads
     Secondary Roads
     Water
     Rail
Figure 3-30. Map of Fairbanks showing locations of CO monitoring sites.
                                           3-37

-------
Figure 3-31. Average diurnal variation in CO at the Fairbanks-Federal Building site for weekdays during the winter season
(October through March). The abscissa shows the time of day from midnight to midnight, the ordinate shows years from
the winter of 1986-87 through 1995-96, and the z-axis shows CO concentration in parts per million.
               Federal Bldg, AIRS Site 020900002
             CO - Winter Daily Maximum 8 Hours Average For Weekdays
                                          igg   • Extremes
                      1990  1991  1992  1993
Figure 3-32. Central tendency statistics for the daily 8-h max
CO concentration at the Fairbanks-Federal Building site during
the winter season from 1986 to 1995. The top graph shows
box plots (with 10, 25, 50, 75, and 90percentile values) for the
entire time series. Each circle (outlier) or diamond (extreme)
refers to an individual observation that is either three or four
SDs from the mean, and the horizontal line shows the current
8-h NAAQS for CO. The lower graph again shows the 25, 50,
and 75 percentile values (P25, P50, and P75,  respectively)
from the upper graph.
MSAs  in   the   continental  United   States
considered   earlier.    At  these  MSAs,  the
maximum values have tended to track variations
in the median and the interquartile range.
     The monthly average diurnal pattern for
CO at the Federal Building site for weekdays
from October 1986 through March  1987 is
shown in Figure 3-33.  This figure shows the
5:00 to 6:00 p.m.  maximum in average CO
concentrations.   It also shows the continual
increase in  CO concentrations throughout the
day, starting with  the morning rush hour and
continuing unabated until the evening rush hour.
Similarity in the shape of the average monthly
diurnal CO patterns for  the three  sites  in
Fairbanks is to be expected, considering the
relatively high correlation coefficients between
each site combination. However, these patterns
are quite different from those found at the four
MSAs in the continental United States, which
tend   to   have   more  similarities   among
themselves.
                                                3-38

-------
 Figure 3-33. Monthly average diurnal variation in CO at the Fairbanks-Federal Building site for weekdays from October
 through March 1986 to 1987, 1989 to 1990, 1992 to 1993, and 1995 to 1996.  On each graph, the abscissa shows the
 time of day from midnight to midnight, the ordinate shows the month of the year, and the z-axis shows CO concentration
 in parts per million.	
3.5  Sources, Emissions, and Concentrations of Carbon Monoxide in Indoor
     Environments
       The general United States population spends a significant portion of time indoors. In recent years,
more emphasis has been placed on the evaluation of pollutant sources, emissions, and concentrations in
indoor environments to aid in the  evaluation of total human exposure.  It is particularly important to
evaluate carbon monoxide concentrations in indoor environments because indoor exposure may represent
a significant portion of the total CO exposure.
       The following sections include a discussion of sources, emissions, and concentrations of CO in
occupied enclosed environments. The discussion will primarily focus on residential environments; however,
a brief discussion of CO concentrations in other enclosed environments is also included. Emphasis is placed
on the evaluation of manufacturer-recommended uses of combustion appliances and consumer products and
the resulting CO emissions and concentrations. Accidental sources and concentrations and improper uses
of combustion appliances will be mentioned only briefly.  This section will discuss the results of recently
available studies but will only summarize those studies discussed in the previous criteria document (U.S.
Environmental Protection Agency,  1991b).

3.5.1   Sources and Emissions of Carbon Monoxide in Indoor Environments Prior to 1991
       Gas cooking stoves, unvented space heaters, tobacco combustion, wood-burning stoves, and
combustion engines represent sources of CO in indoor environments. Approximately 45% of the homes in
the United States used gas for cooking, drying clothes, and heating water in 1981. Emissions from burners
on gas cooking stoves were highly variable on the same cooking range and between gas ranges. Improperly
adjusted gas stoves (improper air-fuel ratio, often characterized by a yellow-tipping flame) could result in
a greater than fivefold increase in emissions over that of properly adjusted stoves (blue-flame).  However,
                                            3-39

-------
on the average, emissions were comparable for top burners, ovens, and pilot lights.  Vented gas dryers and
furnaces contributed a negligible amount of CO to the indoor environment (U.S. Environmental Protection
Agency, 1991b).
       Carbon monoxide emissions from unvented gas and kerosene space heaters were variable from
heater to heater.  Carbon monoxide emissions from unvented gas space heaters were higher for maltuned
units and varied with the method of emission testing. Among the types of unvented space heaters, emissions
were higher for the infrared gas space heaters than those for convective and catalytic unvented gas units.
For unvented kerosene space heaters, emissions were higher for the radiant heaters (U.S. Environmental
Protection Agency, 1991b).
       Limited information was available on CO emissions  from wood-burning stoves.  However,
nonairtight wood-burning stoves could contribute substantial amounts of CO into the indoor environment.
Carbon monoxide from cigarette combustion showed little variability  among  the brands but could be
substantial. In 1987, 29% of the United States population smoked.  Carbon monoxide emissions from two
cigarettes  smoked over an hour approached that of one range top burner operating under blue-flame
conditions. Ranges in average emissions for CO sources in the indoor environment are listed in Table 3-7
(U.S. Environmental Protection Agency, 1991b).

3.5.2  Combustion Sources and Estimated Emissions Rates
       Carbon monoxide occurs in indoor environments directly through emissions from various indoor
combustion sources or  indirectly as a result of infiltration or ventilation  from  outdoor sources.  In the
absence of an indoor source, CO concentrations will equal that of the ambient air. Nearby ambient sources
of CO can result in higher levels of CO indoors. Unvented, such as gas stoves and gas and kerosene space
heaters; vented, including furnaces, hot water heaters, fireplaces, and woodstoves;  and improperly vented
combustion appliances and consumer products represent the primary sources of CO emissions in the indoor
environment. Table 3-8 lists the various sources of CO in the indoor environment.  Emissions of CO from
use of combustion appliances will depend on several factors. These factors include the source (e.g., gas
cooking stoves, unvented space heaters, woodstoves, fireplaces), appliance design, type of fuel used, fuel
consumption rate, use pattern, and operating condition. Vented combustion appliances usually vent the
combustion by-products directly to the outside. However, if a flue, chimney, or vent leaks or is blocked,
the combustion by-products will be spilled into the indoor environment. Spillage of combustion by-products
also may occur during  conditions of depressurization.  Negative pressure exists when the outdoor air
pressure exceeds that in the indoor environment. Depressurization may develop during operation of kitchen
or bathroom exhaust fans, combustion appliances, and forced-air distribution  systems  and during fire
start-up.
       Two different approaches are used to evaluate CO  emissions for combustion appliances:   (1) the
direct or sampling-hood approach and (2) the mass-balance/chamber approach.  For details on these two
approaches, see U.S. Environmental Protection Agency (1991b).

3.5.2.1  Gas Cooking  Ranges, Ovens, and Furnaces
       Emissions of CO from gas top ranges will depend on the use pattern, operating condition, fuel
consumption rate, and air infiltration into the microenvironment.  A discussion on the use of gas appliances
in food preparation and air infiltration  appears in Section 3.5.4 on indoor CO concentrations. Average
annual household fuel consumption has been estimated at 5,000 ft3 for ranges with standing pilots (Johnson
et al, 1992). Menkedick et al. (1993) reported annual household fuel consumption of 2,180 ft3 (±890 ft3)
for burners, based on actual fuel consumption measurements taken on 103 gas ranges individually metered
over a 2-year period in Illinois. Fuel consumption for burner and standing pilots was 5,710 ft3 (±1,830 ft3).
The average fuel consumption also was affected by the age of the occupants (older adults used the range
                                             3-40

-------
Table 3-7. Ranges in Average Carbon Monoxide Emission Rates for Residential Sources
Fuel Fuel Consumption
Unit Type Type3 Rate (kJ/minb)
Gas Ranges0
Top burners NG —
"
Ovens NG —
—

Burner pilots NG —
Oven pilots NG —


Gas Space Heaters
Convective NG 131-784
P 353-660
Infrared NG 260-368
P 258
Catalytic NG 207
Kerosene Heaters
Convective 37-202
Radiant 85-168
Two-stage 132-182
Flame

Blue
Yellow-
tipping
Blue
Yellow-
tipping
Blue
Blue
Yellow-
tipping

Blue
Blue
Infrared
Infrared
Blue

	
"
—
Ranges in Average
Emission Rates (,ug/kJ)

15-215
92-197
12-257
53-62

28-56
209-322
40


3-33
16
45-69
45
9-14

4-60
27-173
9-54
Source

Himmel and DeWerth (1974)
Traynoretal. (1982)
Borrazzo et al. (1987)
Coteetal. (1974)
Moschandreas et al. (1985)
Fortmanetal. (1984)
Himmel and DeWerth (1974)
Coteetal. (1974)
Moschandreas et al. (1985)
Himmel and DeWerth (1974)
Traynoretal. (1982)
Borrazzo et al. (1987)
Fortmanetal. (1984)
Himmel and DeWerth (1974)

Himmel and DeWerth (1974)
Himmel and DeWerth (1974)
Moschandreas et al. (1985)


Traynoretal. (1984, 1985)
Moschandreas et al. (1985)
Thrasher and DeWerth (1979)
Zawackietal. (1984)
Traynoretal. (1984, 1985)
Traynoretal. (1984, 1985)
Moschandreas et al. (1985)
Traynoretal. (1984, 1985)
Moschandreas et al. (1985)

Leaderer(1982)
Traynoretal. (1983)
Moschandreas et al. (1985)
Leaderer(1982)
Traynoretal. (1983)
Moschandreas et al. (1985)
Traynoretal. (1983)
                                     3-41

-------
          Table 3-7 (cont'd).  Ranges in Average Carbon Monoxide Emission Rates for
                                        Residential Sources
Unit Type
Gas Dryer
Water Heater
Woodstoves and
Fireplaces
Tobacco Smoke
Cigarette
Fuel Fuel Consumption Ranges in Average
Type3 Rate (kJ/minb) Flame Emission Rates (/^tg/kJ)
NG — — 40-69
NG — — 25-77 ppm
— — — 0.08-2.18 g/h

— — 40-67 mg/cigaretted
Source
Moschandreas et al. (1985)
Belles etal. (1979)
Traynoretal. (1987)

National Research Council (1986)
Rickertetal. (1984)
aNG = natural gas, P = propane.
bOne kJ (kiloJoule) is the equivalent of 3.485 ft3 of natural gas.
°Fuel consumption rates not provided for most studies.
dMainstream and sidestream smoke.
Source
Table 3-8.  Sources of Carbon Monoxide in the Indoor Environment

                                               Comments
Outdoor (ambient air)
Gas cooking ranges
Gas space heaters
Kerosene space heaters
Environmental tobacco smoke
Fireplaces and woodstoves
Gas furnaces, clothes dryers,
and water heaters

Motor vehicles
               Carbon monoxide is produced as a primary pollutant during the combustion of
               fossil and biomass fuel and as a secondary gas in the photochemical oxidation of
               methane and other organic compounds in the atmosphere. Carbon monoxide
               enters indoor compartments through mechanical ventilation systems and
               infiltration through the building envelope.

               Emissions of CO from gas ranges depends on the use pattern, unit operating
               condition, and fuel consumption rate.  Gas ranges with standing pilots emit more
               CO than do units with electronic pilots. Poorly tuned burners emit more CO than
               well-tuned burners.

               Emissions of CO from gas space heaters are affected by the fuel type and
               consumption rate, type of burner (convective, radiant, or catalytic), operating
               condition, and duration of use.

               Emissions vary based on unit type (convective, radiant, or two-stage), operating
               condition, and duration of use.

               The majority of CO entering indoor compartments from the combustion of
               tobacco products is through sidestream smoke.

               Carbon monoxide is emitted during fire start-ups, leaks in stoves and pipes, and
               during backdrafting resulting from depressurization.

               Gas furnaces and dryers generally are vented and do not emit CO in the indoor
               environment unless the unit is malfunctioning.

               Operating motor vehicles in enclosed spaces  can be significant sources of CO in
               indoor environments.
                                                 3-42

-------
more frequently for preparing meals than did young adults), the amount of time spent at home, and the
presence of a standing pilot, and such consumption showed a seasonal trend. A recent study by Spicer and
Billick (1996) evaluated CO emissions from a gas stove-top burner, both with and without a load.  An
indoor monitoring study of 293 homes conducted by the Gas Research Institute, Pacific Gas and Electric,
and Southern California Gas showed increased CO emissions from the use of gas ranges with standing pilot
lights (Billick et al, 1984, 1996).  Study details are included in the section on indoor concentrations.
       Carbon monoxide emissions from vented gas furnaces and hot water heaters are generally negligible
(Borrazzo et al., 1987); however, emissions may vary based on the working condition and efficiency of the
unit. Dangerous levels of CO have been noted in cases where units malfunctioned or the venting system
leaked or was improperly installed.  Ryan and McCrillis (1994) evaluated CO emissions from two gas
furnaces; one furnace was an older model with an energy efficiency of 60 to 70%, and the other was a newer
furnace with an energy efficiency of 94%. The furnaces were operated for 10 min, then allowed to cool for
5 to 10 min.  The cycle was repeated 12 to 18 times during the course of each test.  The CO emission rate
was >1,000 //g/kJ for the older unit, compared with 6 //g/kJ for the newer, more efficient model. No recent
studies were found on the emission of CO from gas hot water heaters. However, Ault (1999) reported that
approximately 3% of the CO-related mortality and up to 3% of the nonfatal CO poisonings between 1992
and 1996 were caused by malfunctioning water heaters.

3.5.2.2  Emissions from Unvented Space Heaters
       Higher CO emissions have been reported for infrared  gas space heaters than for convective or
catalytic units. Other factors that may affect emissions from unvented space heaters include air circulation
near the heater, primary aeration, and air infiltration and exchange.
       Hedrick and Krug (1995) determined the emissions of CO from four different gas space heaters in
operation and the emissions from the pilot lights.  The study was conducted in  a 2,300  ft2,  1-story,
3-bedroom single-family dwelling with a full  basement.  Eight burner experiments and three pilot  light
experiments were conducted. The heaters were of four types:  (1) 10,000 BTU-h, blue-flame convective;
(2) 15,000 BTU-h (maximum) radiant-tile (infrared); (3) 14,000 BTU-h (maximum), fan-forced, blue-flame
convective; and (4) 16,000 BTU-h, perforated-tube convective. The heaters were operated for 8-h, followed
by a 15-h decay period.  The pilot studies were conducted over a 48-h period. The emission  rates for CO
varied from 8.7 //g/kJ for fan-forced models to 63.7 /^g/kJ for the infrared unit and to 200.31 yUg/kJ for the
perforated-tube type (CO emissions enhanced by leaky gas pressure regulating valve). Carbon monoxide
concentrations in the test house are discussed in Section 3.5.4 on CO concentrations. Spicer and Billick
(1996) reported CO emissions indexes of 19.1 and 28.7 /^g/kJ for a convective, blue-flame space heater.
An emission index of 44.1 //g/kJ was noted for a radiant unit.
       Fan et al. (1997) reported the average pollutant emission rates for a new portable gas stove, a used
kerosene radiant space heater, a kerosene lamp, an oil lamp, and several candles. Both 1-K-grade kerosene
and citronella patio torch fuel were used for the oil lamp test.   The lamp wick was tested at the normal
height (1 in.) and at high flame (2 in.). Four 7.6-cm diameter candles were burned together in each candle
test.  Both the lamp tests and the candle tests were conducted in a 0.15 m3 chamber.  Butane was used for
the portable gas stove and 1-K-grade kerosene was used in the kerosene space heater; both tests were
conducted in a room with a volume of 19 m3.  The tests were run for 30 min to 2 h,  with a 30-min decay
period. Estimates of the emission rates were done using a single-compartment mass  balance model.  The
CO emission rates for the portable gas stove and kerosene heater were 33.6 ± 15.0  and 226.7 ±100 mg/h,
respectively. Carbon monoxide emission rates of 8.2 ± 1.1, 7.1 ±0.8, and 4.7 ±3.0 mg/h were established
for the kerosene lamp,  oil lamp,  and candles, respectively. The height of the wick  did not affect the
emission of CO from either the kerosene lamp or the oil lamp.
       Miller and Hannigan (1999) reported the potential for high indoor concentrations of CO from the
use of unvented gas fireplaces. A series of tests were conducted in two homes in Denver with professionally


                                             3-43

-------
installed, accordingto the manufacturer's specifications, unventedgas fireplaces. Fireplace A was equipped
with an oxygen sensing system that switches off the gas supply if the oxygen level falls below the
manufacturer-defined safe level. Fireplace B was equipped with a thermoelectric safety gas control valve
that automatically shuts off all gas supply to the heater if enough fresh air is not available. The unvented
gas fireplaces were operated for 2 to 9 h at the low, medium, and high settings for fireplace A and the low
and high settings for fireplace B. Carbon monoxide emissions for fireplace A ranged from 250 to 9,000
mg/h. Emissions from fireplace B ranged from 2,600 to 5,000 mg/h. The highest emissions were noted
when fireplace A was operated at the medium setting and fireplace B at the high setting. The high emissions
were attributed to the high elevation of Denver, resulting in a lower ambient air pressure and a higher fuel-
to-air ratio.

3.5.2.3  Woodstoves and Fireplaces
       Carbon monoxide may enter the indoor environment during fire start-up and tending and through
leaks in the stove or venting system. Carbon monoxide emissions are higher during the first stage of a fire
because of increasing amounts of fuel being burned  and  inadequate temperature conditions.   Such
intermittent emissions makes it difficult to accurately determine CO emission rates. Mueller Associates
(1985) reported CO emission ranges of 0.07 to 0.375 g/h for wood heaters (stoves).  Carbon monoxide also
may enter the  indoor environment through  backdrafting when  the  natural draft  is overcome by
depressurization.  Depressurization generally  occurs during fire start-up, but also may  occur  during
operation of other equipment such as kitchen and bathroom exhaust fans, forced-air distribution systems,
and combustion appliances and because of outdoor conditions. Jaasma et al. (1995) conducted a study
designed to evaluate the effectiveness of custom-built glass doors for fireplaces in reducing CO emissions
under conditions of negative pressure. The glass doors decreased spillage of CO; however, decreasing the
leakiness of the glass doors did not always reduce CO spillage.  Tests with the glass doors closed had CO
emission rates of 2 to 36 g/h (highest concentrations represented leaking glass doors). Carbon monoxide
emissions on the order of 70 g/h were noted for glass-door-opened tests under negative pressure.
       Nagda et al. (1996) summarized the results of several studies on emissions of pollutants into living
compartments as  a result of house depressurization.   Carbon monoxide emissions were  found to be
insignificant.  Tiegs and Bighouse (1994) evaluated CO spillage from a woodstove under chamber and in-
house conditions.   Carbon monoxide leakage into the indoor  environment was noted from nonairtight
woodstoves during conditions of negative pressure.

3.5.2.4   Environmental Tobacco Smoke
       Carbon monoxide emissions from the combustion of tobacco occurs in the indoor environment when
smokers exhale the previously inhaled or mainstream smoke and from the emission of sidestream smoke
from smoldering tobacco products.  The amount of CO emitted will vary based on the type (e.g., cigarette,
cigar) and brand of tobacco product, the degree to which tobacco is actively smoked, and the amount of
smoke being absorbed by the lungs (Klepeis et al., 1996; Akbar-Khanzadeh and Greco, 1996). The majority
of the CO emissions are from sidestream smoke. Ott et al. (1992) reported a sidestream-to-mainstream ratio
of*3.
       The Federal Trade Commission compiled data on 933 varieties of cigarettes manufactured and sold
in the United  States in 1992.  These data were provided by the various cigarette manufacturers.  Carbon
monoxide emission rates for the brands of cigarettes reported ranged from <0.5 to 23.0 mg per cigarette
(cigarettes emitting 23.0 mg were unfiltered brands) (Federal Trade Commission, 1994). Klepeis et al.
(1995,  1996)  measured CO concentrations in airport smoking  lounges under real-life conditions.  They
estimated CO emissions to be 78 mg per cigarette (mainstream and sidestream) on the basis of an average
CO emission rate of 11.1 mg/min and a smoking duration of 7 min. An estimated total CO emission rate
of 81.2 mg for three cigarettes (mainstream and sidestream) was reported by Ott et al. (1992). Lofroth et al.


                                             3-44

-------
(1989) estimated a CO emission rate of 67 mg per cigarette (sidestream) based on a cigarette weight of
-1.2 g and a smoking duration of 12 min.  Large cigars emit substantially more CO than do cigarettes.
Emission rates of 82 to 200 mg CO/g (mass smoked; mainstream and sidestream) were reported by Klepeis
etal. (1999) (smoked by machine and by a person). Cigar mass ranged from 5.9 to 16.7 g, and the smoking
time was 7 to 40 min for the machine-smoked test and 78 and 90 min for the test measuring emissions from
a cigar being smoked by a person.

3.5.3  Source-Related Concentrations of Carbon Monoxide in  Indoor Environments Prior
       to 1991
       Microenvironments associated with motor vehicles usually result in the highest concentrations of
CO.  Carbon monoxide concentrations of up to 28 ppm were reported in indoor parking garages and indoor
environments associated with attached garages (Akland et al, 1985; Johnson et al., 1984; Wallace, 1983;
Flachsbart and Ott, 1984). Carbon monoxide concentrations inside moving vehicles can exceed the 8-h,
9 ppm and 1-h, 35 ppm NAAQS  for CO (Flachsbart et al., 1987; Chaney, 1978; Ziskmd et al., 1981) and
are generally higher in personal vehicles than in public transportation vehicles (Flachsbart et al., 1987;
Cortese and Spengler, 1976).
       Based on the intermittent use of gas cooking stoves, average long-term concentrations of CO are not
expected to be significant (Research Triangle Institute, 1990; Koontz and Nagda, 1987). However, short-
term peak concentrations of CO of 1.8 to 120 ppm have been reported from the use of gas cooking stoves
(Research Triangle Institute,  1990; Koontz and Nagda, 1987; Leaderer  et al., 1984; Moschandreas and
Zabransky, 1982; Sterling and Sterling, 1979).
       The use of unvented gas space heaters as primary heat sources is expected to result in higher long-
term concentrations of CO ranging from 0.26 to 9.49 ppm (mean) (McCarthy et al., 1987; Koontz and
Nagda, 1988). Peak CO concentrations from the use of unvented gas heaters are also generally higher than
unvented kerosene heaters and gas cooking stoves (Koontz and Nagda,  1987; Leaderer et al., 1984;
Davidson et al., 1987).
       Indoor concentrations of CO from the use of nonairtight wood-burning stoves can contribute as much
as 9 ppm to the average indoor CO concentration (Traynor et al., 1984).  Airtight stoves have been shown
to contribute from 0.1 to 2.0 ppm CO to the average CO background level (Humphreys et al., 1986; Traynor
etal., 1984)
       Concentrations of CO in environments with smoking is highly variable, depending on the type of
environment, number of cigarettes smoked, and the type and amount of ventilation. Peak CO concentrations
of 32 ppm (mechanical ventilation) and 41 ppm (natural ventilation) have been measured in automobiles
(Harke and Peters, 1974).  However, although cigarettes are expected  to contribute to the indoor CO
concentrations, the additions are not expected to be substantial except when heavy smoking occurs in small
spaces.

3.5.4  Indoor Concentrations of Carbon Monoxide
3.5.4.1  Factors Affecting Carbon Monoxide Concentrations
       A number of factors can affect indoor CO concentrations:  the presence of a source and its use
pattern, pollutant emission rate, ambient air concentrations, infiltration through the building envelope, air
exchange rate (AER), building volume, and air mixing within the indoor compartments.
       The major sources of CO in residential environments are unvented gas or kerosene appliances.
Carbon monoxide in the indoor  environment from vented combustion  appliances (furnaces, hot water
heaters, and gas clothes dryers) are generally negligible unless the unit is malfunctioning. Dangerous levels
of CO have been noted in cases where the venting system leaked or was improperly installed. Because gas
cooking ranges are used intermittently for cooking purposes, it is not likely that the use of gas ranges would
result in substantial increases in CO over long periods of time, except possibly in households where gas


                                            3-45

-------
cooking stoves have continuously burning pilots or are used improperly as a primary or secondary source
of heat.  Of 83.2 million adults surveyed that used gas stoves or ovens for cooking during the years 1988
to 1994, 7.7 million had used the stoves for supplemental heating at least one time during the previous year
(Slack and Heumann, 1997). Koontz et al. (1992) reported the results of a survey conducted in 1985 and
1991 designed to determine the prevalence of kitchen fans and the factors affecting their use and the impact
of other cooking appliances (e.g., microwave ovens, toaster ovens, hot plates) on the use of gas ranges for
cooking. The authors reported a 27% increase in the use of gas ranges without standing pilot lights between
1985 and 1991 and a 20% reduction in the use of both electric and gas stoves for cooking. Ninety-five
percent of the households surveyed reported having another form of cooking  appliance in addition to the
gas range, and, of this number, 55 to 65% reported using the stove less often.  There were, however, more
people using a gas range for purposes of supplemental heating than there were using electric ranges for that
purpose (11% versus 3.6%). Estimates of gas cooking stove usage range from 30 to 60 min/day. The use
of gas for cooking varies by location (Johnson, 1984;Hartwelletal, 1984; U.S.  Census Bureau, 1998,1999;
Wilson  et al., 1993). However, an estimated 40% of the households in the United States used gas for
cooking between 1985 and 1995 (Koontz et al., 1992; U.S. Census Bureau, 1998, 1999).
      The  use of unvented space heaters
represent a significant source of CO in indoor
environments. Data from the National  Health
and Nutrition Examination Survey estimated that
13.7 million adults used unvented combustion
space heaters between 1988 and 1994.  Based on
the information obtained in the  survey,  an
estimated 13.2% of the adult population in the
southern  United   States  used  unvented
combustion space heaters. An estimated 5.9% of
the adult population in the Midwest, 4.2% in the
Northeast, and 2.5% in the West used  unvented
space  heaters  (Figure   3-34)   (Slack  and
Heumann, 1997).   The U.S. Census Bureau
estimated   that    1,055,000   and
1,159,000 households used kerosene or another
liquid fuel  as primary  and secondary heating
fuels, respectively, in 1995 (U.S. Census Bureau,
1999).  The U.S. Environmental Protection Agency (1990) estimated that kerosene heaters are used
16.7 h/day in southern states as primary sources of heat,  and, in  regions where the heaters are used as
secondary heat  sources, estimated use ranges  from 2.6 to 10.7 h/day.  Information  on the number of
households using combustion appliances in the United States in 1995  appear in Table 3-9.
      The AER, the balance of the flow of air  in and out of a microenvironment, is based on the fraction
of air that enters the microenvironment through infiltration through unintentional openings in the building
envelope, natural ventilation through any designed opening in the building envelope (doors, windows), and
forced ventilation systems. Infiltration is the dominant mechanism for residential air  exchange. Forced
ventilation is typically the dominant mechanism for  air exchange in nonresidential buildings.  Natural
ventilation, airflow through doors and opened windows,  is seasonral (Koontz and Rector, 1995).  Air
exchange rates vary depending on the outside temperature, geographical location, type of cooking fuel used,
type  of heating system used, and building type (Colome et al., 1994).  Air exchange rates are generally
higher during the summer and lower during the winter months (Wilson et al., 1996; Murray and Burmaster,
1995; Colome etal, 1994; Research Triangle Institute, 1990). Air exchange rates were about l.Sh"1 during
           Northeast   Midwest
                            South
                                    West
Figure 3-34. Percentage of U.S. households using unvented
combustion heaters, by type of fuel, stratified by region (Third
National Health and Nutrition Examination  Survey, 1988 to
1994).
Source:  Slack and Heumann (1997).
                                             3-46

-------
              Table 3-9. Combustible Fuels in Homes in the United States in 1995
                                               Combustible Fuel Type
Heating"
Cookingb
Central Air
Conditioning0
Clothes Dryer*
Water Heater6
Piped Gas
49,203,000
(848,000)
35,001,000
2,971,000
15,998,000
50,558,000
Bottled Gas
4,251,000
(558,000)
4,217,000
—

3,239,000
Kerosene/Other
Fuel Oil Liquid Fuel
10,974,000 1,055,000
(451,000) (1,159,000)
— 301,000
— —

5,808,000 331,000
Wood
3,533,000
(7,949,000)
33,000
—

44,000
 "Based on 96,650,000 occupied housing units with heating fuel. Values in parenthesis represent use of fuel type as
  secondary heating fuel.
 bBased on 97,406,000 occupied housing units with cooking fuel.
 °Based on 46,577,000 occupied housing units with central air-conditioning.
 dBased on 70,756,000 occupied housing units with central air-conditioning.
 eBased on 97,522,000 occupied housing with hot piped water.

 Source: U.S. Census Bureau (1999).


warm weather and from 0.41  to 0.59 rT1 (arithmetic) during cold weather months. The one study  that
reported AER for all seasons, reported an AER of 0.76 IT1 (arithmetic) (Murray and Burmaster, 1995).
Homes with gas cooking stoves with standing pilots and gas wall furnaces had the highest AER. Homes
with gas stoves without pilots  had higher AER than electric stoves.  Also, homes with forced air furnaces
had higher AER than electric homes.
      Lagus Applied Technology, Inc. (1995) reported AERs for 49 nonresidential buildings (14 schools,
22 offices, and 13 retail establishments) in California. Average mean (median) AERs were 2.45 (2.24),
1.35 (1.09), and 2.22 (1.79) IT1 for schools, offices, and retail establishments, respectively.  Air infiltration
rates for 40 of the 49 buildings were 0.32, 0.31, and 1.12H"1 for schools, offices, and retail establishments,
respectively.  Air exchange rates for 40 nonresidential buildings in  Oregon and Washington (Turk et al,
1989) averaged 1.5 IT1 (mean  [median = 1.3 IT1]).  The geometric mean of the AERs for six garages  was
1.6 IT1 (Marr et al., 1998).  Park et al.  (1998) reported AERs for three stationary cars (cars varied by age)
under different ventilation conditions. Air exchange rates ranged from 1.0 to 3.0 h'1 for windows closed
and fan off, 13.3 to 23.5 h'1 for window opened and fan off, 1.8 to 3.7 h'1 for window closed and fan on
recirculation (two cars tested), and 36.2 to 47.5 h'1 for windows closed and fan on fresh air (one car tested).
An average AER of 13.1 h'1 was reported by Ott et al. (1992) for a station wagon moving at  20 mph with
the windows closed.

3.5.4.2  Models for Carbon  Monoxide Concentrations
      Indoor concentrations  of CO can be estimated using the mass-balance model. The mass-balance
model estimates the concentration of a pollutant over time.  The simplest form of the model is represented
by the following differential equation  for a perfectly mixed microenvironment and no air cleaner:
                                             3-47

-------
                             dC
                                IN
                                                                                          (3-1)
where dCm is the indoor pollutant concentration (mass/volume), t is time in hours, v is the air exchange rate,
CQUT is the outdoor pollutant concentration (mass/volume), V is the volume of the microenvironment, and
S is the indoor source emission rate. A more in-depth discussion of the mass-balance model may be found
in U.S. Environmental Protection Agency (1991b) and Nagda et al. (1987).
       Traynor et al. (1989) used a model to predict CO concentrations in residential environments for one
pollutant source.  Model inputs included  ambient air concentrations, source emission rates and usage
characteristics, compartment volume, AERs,
and  outside  temperatures.     The  model
combined the steady-state version of the mass-
balance model used in indoor  air  quality
studies, a source-usage model for space heating
appliances,  and  an  air  exchange  model.
A combination  of  the  Monte  Carlo  and
deterministic techniques was used to predict
indoor concentration distributions.  Based on
the modeled results, the use  of kerosene
heaters, unvented gas space heaters, and gas
ovens and ranges for heating produced the
highest  concentrations of CO  in  the indoor
environment (see Figure 3-35).  The findings
illustrated in Figure 3-35 are for only a limited
number of model runs, sources, and building
conditions.
              —t	1 Airtight v*odslovc

              	1 Gas Wall/Floor Furnao

              	1 Gas Boiler
                  CO Concentration (ppm)
Figure 3-35.  Modeled indoor CO concentration distributions in
houses with only one indoor combustion pollutant source.
Source:  Traynor et al. (1989).	
3.5.4.3  Microenvironmental Monitoring Studies
Residential Carbon Monoxide Concentrations Related to Indoor Sources
       The Gas Research Institute,  Pacific Gas and Electric Company, San Diego Gas and Electric
Company, and Southern California Gas Company initiated indoor monitoring of 293 randomly selected
homes in California. Monitoring was for a single 48-h period.  Carbon monoxide concentrations indoors
were reported to be closely associated with concentrations outdoors for most of the residences monitored.
However, 13 homes had CO concentrations above 9 ppm, and concentrations in one home exceeded 3 5 ppm.
Homes with gas ranges with standing pilot lights had higher CO concentrations than did homes with gas
ranges with electronic pilot lights or electric ranges. Homes with standing pilots had a 0.56-ppm increase
in net CO.  Indoor minus outdoor  CO concentrations for six averaging times were used to rank homes.
Using that criterion, 21 of the 293  homes studied were selected for case studies.  The higher CO seen in
these homes possibly was associated with occupant smoking, the use of gas stoves for heating purposes,
infiltration from attached garages, the type of heating system used (homes with gas wall furnaces had higher
CO), the building type and size (smaller multifamily homes had higher CO than larger single-family homes),
and more than  one CO source.  The average AER varied by type of heating system (wall furnaces >
forced-air > electric) and building type (multifamily units > single-family units) (Billick et al., 1984,1996;
Colome et al., 1994). The CO descriptive  statistics for homes in this study are given in Table 3-10.
       Research Triangle  Institute (1990) monitored CO  concentrations in 400 homes  for  3 days in
Suffolk and Onondaga Counties, NY. Homes in Suffolk County were monitored in January and February.
The selected homes in Onondaga County were monitored in February, March, and April. The average room
                                             3-48

-------











w

CD




Table 3-10.
All Homes

48-h
Arithmetic Mean 1.6
Standard Error 0.1
Mode 1.0
Percentiles
Maximum 12.9
95th 4.3
75th 1.8

50th 1.2

25th 0.7
5th 0.1
Minimum 0.0
Carbon Monoxide Descriptive Statistics for All Homes

Max
10-min
5.2
0.3
2.0

37.9
15.1
6.6

3.5

2.0
1.0
0.1
Indoor Average
Max
(number = 277; in parts per million)
Outdoor Average
Max Max
30-min Maxl-h Max 8-h 48-h 10-min 30-min Max 1-h Max 8-h
4.8 4.5 2.9 1.0
0.3 0.3 0.2 0.1
2.0 2.0 1.0 0.1

36.7 35.8 23.5 10.8
14.2 13.2 8.3 2.7
6.0 5.8 3.4 1.3

3.1 3.0 2.0 0.8

2.0 2.0 1.2 0.3
1.0 1.0 0.5 0.1
0.0 0.0 0.0 0.0
5.5 4.3 3.8 2.0
0.4 0.3 0.2 0.1
2.0 2.0 1.0 1.0

68.7 31.5 27.3 17.3
16.1 12.3 10.6 6.3
6.1 5.2 4.8 2.2

3.3 2.9 2.6 1.4

2.0 1.9 1.5 0.9
1.1 1.0 1.0 0.3
0.2 0.1 0.0 0.0
Source: Modified from Wilson et al. (1993).

-------
       Onondaga County Suffolk County
             Gas Stove
Suffolk County
 Kerosene
 Heaters
Onondaga County
 Woodstove/
 Fireplaces
 Figure 3-36. Arithmetic mean CO concentrations by presence
 or absence of combustion source.  Homes in Suffolk County
 were monitored in January and February and in Onondaga
 County in February, March, and April.
 Source:  Research Triangle Institute (1990).	
volume was assumed to be 50 m3.  The average
AER was 0.59 h"1.  Carbon monoxide monitors
were placed in the primary living  space  and
close to the source.  Approximately  half of the
homes used  gas  cooking stoves.  Kerosene
heaters had to be operated at least  3 h/day to
qualify as a CO source, and the woodstove or
fireplace had  to  be operated  an average of
2 h/day.   Any reported usage of gas stoves
qualified them as sources.  The  average  CO
concentration  in the primary living area  was
2.23 ±  0.17 ppm (results for 209 homes). Use of
both gas stoves and kerosene space heaters  was
associated  with increased CO.  Homes using
woodstoves or fireplaces had lower CO than did
homes without woodstoves or  fireplaces  (see
Figure 3-36).  An explanation for the finding of
higher CO  in homes without wood  stoves  and
fireplaces  was not  provided.    Lower  CO
concentrations may have been associated with an
increased air exchange rate. Also, CO emissions
for the use of wood stoves and fireplaces are intermittent and will generally only occur during fire start-up
and tending or through leaks in the stove or venting system.
       Hedrick and Krug (1995) reported CO concentrations from the use of unvented gas space heaters
and pilot lights in a test house in  Chicago, IL. The gas space heaters included blue-flame convective,
radiant-tile, fan-forced blue-flame  and perforated-tube convective units. Emission rates for the units are
discussed in Section 3.5.2.2. The  house was a 2,300 ft2, single-family, 3-bedroom dwelling with a full
basement.  Eight burner experiments and three pilot light experiments were conducted.  Heaters were
operated for 8  h followed by a 15-h decay period. The pilot studies were conducted over a 48-h period.
Fans were used to distribute emissions throughout the house, excluding the basement.  The highest CO
concentrations were seen with the  radiant-tile heater (13.4 ppm), and the lowest CO concentrations were
reported with the fan-forced unit (0.9 ppm). Two of the tests using the blue-flame convective units were
affected by gas leakage, resulting in  CO concentrations of 4.7 and 4.8 ppm.  The test not affected by the
leakage had a CO concentration of 2.7 ppm. The maximum CO concentration from use of the perforated-
tube convective heater was 31.9 ppm. Carbon monoxide concentrations during the pilot light experiments
did not exceed 2.0 ppm.
       Moderately high concentrations of CO have been reported in homes where unvented kerosene
heaters are in use.  Burton et al. (1990) monitored both the inside and outside of two mobile homes for
pollutants emitted from the operation of radiant and convective unvented kerosene space heaters. No other
sources of combustion by-products were in the homes. Heaters were operated from 4:00 to 9:00 p.m. daily.
Six  random sampling  periods  were conducted, three with heaters on and three with heaters off.
Measurements were made until 11:00 p.m. Average CO concentrations while the heaters were in operation
were 12 ppm for the convective heater and 4 ppm for the radiant heater; when the heaters were not in use,
average CO concentrations were 5 and 1 ppm, respectively. Ambient CO concentrations in the mobile home
park were reported to be negligible.
       Carbon monoxide concentrations in eight single-wide mobile homes (150 to 255 m3) were reported
by Mumford et al. (1990, 1991).  Convective kerosene heaters were used in four of the homes, three of the
homes used radiant heaters, and one home used a convective/radiant heater. Monitoring was conducted 2.6
3-50

-------
to 9.5 h/day (average 6.5 h/day) for 2 weeks with heaters on and 2 weeks with heaters off. Fuel consumption
rates ranged from 252 to 295 kJ/min for convective units, 105 to 168 kJ/min for radiant units, and 120
kJ/min for the convective/radiant unit. Average AERs were 0.47 h"1 with heater on and 0.48 h"1 with heater
off.  Monitoring began when heater use began and continued for 2 h after the heaters were turned off.
Carbon monoxide concentrations were above 9 ppm in four of the eight homes. In one home with a
convective heater, CO concentrations peaked at 51 ppm. The average CO concentration with heaters off
was 1.4 ppm.
       Williams et al. (1992) reported CO concentrations in eight all-electric mobile homes (each <100 m2)
from the use of kerosene heaters over a 6-day measurement period.  The space heaters were used for an
average of 4.5 h/day between 4:00 to 11:00 p.m. Measured CO in homes 0.5 h prior to kerosene heater use
ranged from 0 to 8ppm. Average CO during nonuse days was 1.4ppm±0.3.  Peak CO values ranged from
0.3 to 50.2 ppm. The 8-h average CO concentration in the homes was 7.4 ±1.4 ppm.  Peaks usually were
observed at the end of the combustion period.  The AER averaged 0.47 h"1 when the unit was in operation.
Homes with the radiant and multi-stage units had higher CO than homes with convective heaters. Average
1-h concentrations of CO for convective units ranged from 1.3 to 5.3 ppm, and from 1.1 to 28.3 ppm and
16.0 to 50.0 ppm for radiant and multi-stage units, respectively.
       A series of tests were conducted in two homes with professionally installed, according to the
manufacturer's specifications, unvented gas fireplaces in Denver.  Details on how the tests were conducted
and the CO emission rates are discussed in Section 3.5.2.2.  The highest concentration of CO, given as the
time-weighted average, was at the medium setting for one of the unvented gas fireplaces and at the high
setting for the other. Carbon monoxide concentrations ranged from 1.6 to 78 ppm for the first fireplace and
5.7 to 30 ppm for the second.  The high concentrations were attributed to the high elevation of Denver,
resulting in a lower ambient air pressure and a higher fuel-to-air ratio (Miller and Hannigan, 1999).

Carbon Monoxide Concentrations Related to Environmental Tobacco Smoke
       Carbon monoxide  concentrations in environments where smoking occurs exceed background CO
concentrations. The indoor concentrations will depend on the size of the indoor space, number of cigarettes
smoked, smoking rate, CO emission rate, differences in ventilation, and the ambient CO concentrations
(Turner etal., 1992).  The U.S. Centers for Disease Control and Prevent on reported an estimated 61 million
smokers in the United States in 1995, representing 29% of the population. The percentage of smokers
between 1994 and 1995 was unchanged. In 1994, an estimated 1.5 million American became daily smokers.
The estimated number of new smokers per year has not changed since  1980 (Tobacco Information and
Prevention Source, 1996).  The United States Department of  Agricultural estimated that 487 billion
cigarettes were sold in 1995 (Federal Trade Commission, 1997). An estimated 4.4 billion cigars were sold
in the United States in 1997,  compared to 3.8 billion in 1996 (Federal Trade Commission, 1999). Ott et al.
(1992, 1995) conducted a series of monitoring experiments in a one-story house during development of a
multi-compartment indoor mass-balance model to predict the pollutant concentrations from environmental
tobacco  smoke.  Smoking time ranged  from 6.5 to 9.5  min.   Carbon monoxide concentrations were
measured in three locations in the bedroom after three cigarettes were  smoked over  a 9-h  period.
Concentrations ranged from 0.4 to 0.6 ppm. The only ventilation was a partially opened window covered
with a shade (AER = 1.2  h"1). Klepeis et al.  (1995) reported a range of 0.41 to 1.2 ppm CO (average
0.75  ppm CO) in airport smoking lounges based on 10 sampling periods ranging from 60 to 146 min. The
average number of people smoking during the sampling periods ranged from 2.8 to 13.5. The room volumes
ranged from 238 to 803 m3, with AERs of 12.8 and 15.8 h"1. Holcomb (1993) reviewed the literature  on
tobacco smoke in various indoor environments and evaluated those data the authors identified as generated
under real-life conditions.  Carbon monoxide concentrations ranged from 0.1 to 10.2 ppm, depending  on
the indoor environment.   The results are outlined in Table 3-11.  Lofroth et al. (1989) reported CO
concentrations in a chamber tests for cigarettes smoked every 150 or 3 0 min for 4 h. The chamber volume


                                            3-51

-------
      Table 3-11.  Carbon Monoxide Concentrations in Smoking and Nonsmoking Areas in
     	Real-Life Situations (in parts per million)	
                                      Smoking
Nonsmoking
Diff. in Means
Category
Offices and
Public
Buildings
Restaurants
Taverns/Bars
Trains
Buses
Autos
No. of
Studies
13
5
2
2
1
1
Sample
Size
697
107
5
18
35
—
Mean
2.95
3.6
6.4
2.2
6.0
—
Range
0.1-8.7
0.4-9.0
—
1.0-5.2
3.7-10.2
—
Sample
Size Mean Range
275 2.99 0.7-4.0
— —
— —
10 1.30 0.5-2.9
— —
213 11.6 8.8-22.3
Smoking Minus
Nonsmoking
-0.04
—
—
0.90
—
—
 Source: Holcomb (1993).
was 13.6 m3, with a set AER of 3.55 h"1. The cigarette mass was 1.2 g, and mass smoked was 0.9 to 1.0 g.
Smoking duration was -12 min per cigarette. Carbon monoxide concentrations averaged 1.56 and 2.17 ppm
for the 30-min tests and 4.16 ppm for the 15-min test.
       Carbon monoxide concentrations  approaching 20 ppm were reported by Ott et al. (1992) in the
compartment of a moving vehicle after three cigarettes were smoked over a 60-min period.  The
vehicle traveled at 20 mi/h with the windows closed, and the air conditioned set at recirculation.  Carbon
monoxide concentrations reached almost 17 ppm after the first cigarette had been smoked. An averaged
CO concentration of 9 ppm, over a 200-min period, was reported by Klepeis et al. (1999) during a cigar
banquet. More than 100 cigars were smoked by approximately 30 people. Concentrations were reported
to range from 3 to 19 ppm under various smoking conditions.

Carbon Monoxide Concentrations Associated with Motor Vehicles
       In the United States, motor vehicles dominate total anthropogenic emissions of CO. Older vehicles
are likely to emit more CO than newer models. However, when the newer models with catalytic converters
are started, CO emissions are higher because of the rich fuel-air mixture to facilitate ignition and to improve
cold engine operation. Also, the catalytic converter is not warm enough to function efficiently (Marr et al.,
1998). Concentrations of CO inside a moving motor vehicle also can be affected by the CO being emitted
from other vehicles.  Emission rates for combustion vehicles are discussed in Section 3.3 of this chapter.
       Several  studies  have monitored the CO concentrations inside a moving vehicle under various
operating conditions.  Chan et al. (1989,  1991) evaluated CO, NOX, O3, and VOCs inside two  moving
vehicles(1983 and 1987 models). Tailpipe emissions were higher for the older model. Driving routes were
selected to represent three distinct traffic patterns: (1) urban traffic, (2) heavy traffic, and (3) rural traffic.
Inside ventilation was windows and vents closed and air-conditioning on, windows and vents closed and
fan on, or front windows half opened and vent and fan on. Average CO concentrations for 70 samples,
including both cars and all driving routes, were 11.3 ppm.  The in-vehicle concentrations were almost four
times higher than ambient CO (3.0 ppm). Carbon monoxide concentrations in the rural traffic pattern were
significantly less than either the urban or heavy traffic patterns. Carbon monoxide in urban and heavy traffic
                                             3-52

-------
patterns was not significantly different. The lowest CO concentrations were measured when the windows
were open; however, the concentration difference between the different ventilation modes was only 1.0 ppm.
Ott et al. (1994) reported an average CO concentration of 10.2 ppm inside a moving vehicle for 93 trips
under urban highway conditions at varying times of the day.  A more detailed discussion of this study
appears in the chapter on human exposure. Higher CO concentrations were reported both inside and outside
of moving vehicles in Sacramento and Los Angeles than those measured at roadside and ambient monitoring
stations. Carbon monoxide concentrations inside vehicles ranged from nondetectable to 6 ppm (Rodes et al.,
1998).
       Carbon monoxide emissions from the use of combustion engines in enclosed areas may produce
significant increases in CO in the environments where the engines are being operated. Reports from various
municipalities on activated CO alarms have identified a car idling in the garage as the source of CO in the
living compartment. Also, in a study of unintentional deaths in  California, of a reported 128 accidental
poisonings resulting from CO from combustion engine exhaust, 80 incidents were found to have occurred
in a garage or house, 38 in a vehicle parked outside, and 4 in vehicles while driving (Marr et al., 1998).
Kern et al.  (1990) measured CO concentrations in a detached garage  from operation of an  emissions-
controlled (catalytic reactor and oxygen sensor) and an emissions-uncontrolled vehicle (carbureted, without
a catalytic reactor). Two tests were conducted: the first with the garage door poorly sealed (3-in. crack)
and the second with the garage door sealed with rags. Carbon monoxide concentrations in the poorly sealed
garage reached 4,700 ppm for the uncontrolled car versus 2,000 ppm for the controlled car.  When the
garage was better sealed, CO concentrations, after 110 min  of operation, reached 8,400 ppm for the
uncontrolled vehicle versus  3,600 ppm for the controlled vehicle.
       Amendola and Hanes (1984) evaluated the concentration of CO in automotive repair shops based
on seasonal conditions and as a function of work environment size.  Monitoring was conducted in a small
service station (1 to 2 bays),  a large service station (>2 bays), and an automobile dealership. The 8-h time-
weighted average during warm weather ranged from 3.3 to 16.2ppm, 3.4 to 21.6 ppm, and 12.1 to 20.8 ppm
for the small and large service stations and the dealership, respectively; however,  the authors noted that CO
concentrations were affected by the type of ventilation used in the facility, volume and type of repairs, and
employee work habits, such  as minimizing engine run time.
3.6  Summary
       Carbon monoxide is produced by the incomplete combustion of burning fossil and biomass fuels.
Approximately 70% of the CO produced globally is the result of human activities.  Carbon monoxide in the
atmosphere is of both primary and secondary origin.  Secondary sources such as the photochemical
oxidation of CH4 and NMHCs account for almost one-half of the total source strength of CO.   The
uncertainty in estimates of the magnitudes of individual CO sources ranges from a factor of two to that of
three.
       Atmospheric CO concentrations in remote areas of the world have been increasing at the rate of
about 1 % per year throughout most of the industrial era.  This increase reflects the growth of anthropogenic
emissions from the combustion of fossil and biomass fuels and increased agriculture to feed the expanding
world population. However, CO concentrations decreased for several years from the late 1980s to the early
1990s.  The reasons for this decline are not clear, although several factors may be involved.  Since then,
there has been no clear trend in CO  concentrations.
       Carbon monoxide plays an important role in atmospheric chemistry because it is the major reactant
for OH radicals. Reaction with OH radicals is the loss mechanism for many trace gases that are responsible
for contributing to the greenhouse effect (e.g.,  CH4) and for depleting stratospheric O3 (e.g., CH3C1, and
CH3Br).  Thus, increases in CO concentrations can suppress OH radical concentrations and allow the
concentrations of these trace gases contributing to global-scale environmental problems to increase, even

                                             3-53

-------
if their emissions are constant.  Conversely, decreases in global average CO concentrations can stabilize or
even reverse the growth rates of the trace gases mentioned above.
       Carbon monoxide participates in the formation of 20 to 40% of O3 in the background or "clean"
troposphere and may be involved in urban O3 formation. One  study suggests that CO participated in the
formation of 10 to 20% of the O3 formed during a smog episode in Atlanta on June 6, 1988. Obviously, the
photochemistry must be examined in more cities before any more general statements about the importance
of CO in urban air chemistry can be made.
       Emissions from transportation dominate other sources of CO within the United States.  Residential
wood burning  may be an important source of CO in a number of urban areas.  The uncertainties in the
magnitudes of individual sources in the nationwide and worldwide emission inventories are comparable
(i.e., roughly a factor of between two and three).
       There has been a consistent decrease in the nationwide  annual second-highest maximum 8-h
composite average ambient CO concentration over the past 20 years, from about 11 ppm in 1977 to about
4ppmin 1996.  This improvement in CO quality occurred despite a 121% increase in vehicle miles traveled,
a 29% increase in population, and a 104% increase in gross domestic product in the United States over the
same period.  During the past 10 years, the composite  mean annual CO second-highest maximum 8-h
concentration decreased 37%  at 190 urban sites, 37%  at 142  suburban locations, and 48% at 10  rural
monitoring sites. Hourly average CO concentrations decreased from 2.0 ppm to 1.2 ppm over the past 10
years. Despite uncertainties in the calculations of CO emissions, the decline in ambient CO concentrations
in the United States reflects the controls placed on automotive emissions. These declines are seen clearly
in the trends in CO concentrations during times of day when CO concentrations result primarily from mobile
source emissions.
       The patterns and trends of observed CO reflect reductions in the CO emissions of the past 11 years.
However, it is important to note that the reported concentrations from the monitoring sites are representative
only of the air quality in their neighborhoods. Also, although personal exposure to CO from mobile sources
also should be decreasing, the CO values from the monitoring sites are not equivalent to personal exposures.
The same ratios of personal to monitored CO from past studies in urban areas with CO emissions dominated
by mobile sources may remain applicable today, but continued  validation is needed.
       Carbon monoxide occurs in the indoor environment directly as a result of the emissions from various
indoor combustion sources or indirectly as a result of infiltration or ventilation from outdoor sources.
Sources of CO in the indoor environment include vented (furnaces, hot water heaters, gas cooking stoves,
fireplaces and woodstoves) and unvented (kerosene and gas space heaters) combustion appliances, tobacco
smoke, and combustion engines.
       Carbon monoxide emissions and indoor concentrations from the use of combustion appliances are
highly variable. Emissions from combustion appliances are a function of the appliance design, combustion
efficiency and  operating condition, length and frequency of use, and the fuel type and consumption rate.
Concentrations of CO in the indoor environment will depend on the CO emission rate and use pattern of the
combustion appliance,  ambient air concentrations and infiltration through the building envelope, air
exchange rate,  building volume, and air mixing with the indoor environment. In the absence of an indoor
source,  CO concentrations generally will equal that of the surrounding ambient environment. Higher
concentrations are likely to occur in environments with multiple combustion sources.
       Unvented gas and kerosene space heaters generate the highest CO emissions and concentrations in
the indoor environment.   The highest CO concentrations likely will be found in homes  where unvented
space heaters are the primary heat source rather than in homes where unvented space heaters are used to
supplement another heat source.  However, with the decreasing usage of unvented space heaters, CO
emissions from this source likely will decrease.
       The use of well-maintained, energy-efficient gas stoves will result in only intermittent, small
increases in CO emissions and concentrations.  Gas ranges with standing pilots emit more CO than do  units

                                             3-54

-------
with electronic pilots; however, a steady decrease in the emission of CO from pilot lights likely will occur
with the replacement of gas cooking stoves with standing lights with models with electronic pilots. Also,
with the advent of other cooking appliances (e.g., microwaves, toaster ovens, heating plates), the use of
ranges in meal preparation is decreasing.  High indoor CO emissions and concentrations can occur when
gas cooking stoves are used improperly for heating.
        Carbon monoxide emissions and concentrations in the indoor environment from vented combustion
appliances (furnaces, hot water heaters, and gas clothes dryers) are generally negligible unless the unit is
malfunctioning.  Less efficient vented appliances will emit more CO than the more efficient appliances.
Dangerous levels of CO have been noted in cases where the unit was malfunctioning or the venting system
leaked or was improperly installed.
        Carbon monoxide generally is not emitted by airtight woodstoves and fireplaces, except during fire
start-up and maintenance, leaks in the venting system, and through backdrafting when the natural ventilation
is disrupted by depressurization. Depressurization generally occurs during fire start-up but also can occur
during operation of other equipment such as kitchen and bathroom exhaust fans, combustion  appliances,
and forced-air distribution systems and because of outdoor conditions.  Non-airtight stoves can emit a
significantly higher amount of CO in the indoor environment.
        Concentrations of CO in  environments  where  smoking occurs  can  exceed CO  background
concentrations.   Like combustion appliances, CO emissions and concentrations will depend on the CO
emission rate of the tobacco product, smoking rate, size of the indoor compartment, and air exchange rate.
The Center for Disease Control estimated that 29% of the United States population smoked in 1995.
        Carbon monoxide  emissions from the use of combustion engines in enclosed areas may produce
significant increases in CO in the environments where the engines are being operated. Reports from various
municipalities on activated CO alarms identified a car idling in the garage as the source of CO in the living
compartment. Also, in a study of unintentional deaths in California, of areported 128 accidental poisonings
resulting from CO from combustion engine exhaust, 80 incidents were found to have occurred in a garage
or house, 38 in a vehicle parked outside, and 4 in vehicles while driving.
References
Akbar-Khanzadeh, F.; Greco, T. M. (1996) Health and social concerns of restaurant/bar workers exposed to environmental tobacco
        smoke. Med. Lav. 87: 122-132.
Akland, G. G.; Hartwell, T. D.; Johnson, T. R.; Whitmore, R. W. (1985) Measuring human exposure to carbon monoxide in
        Washington, B.C., and Denver, Colorado, during the winter of 1982-1983. Environ. Sci. Technol. 19:911-918.
Allen, D. I; Kasibhatla, P.; Thompson, A. M.; Rood, R. B.; Doddridge, B. G.; Pickering, K. E.; Hudson, R. D.; Lin, S.-J. (1996)
        Transport-induced interannual variability of carbon monoxide determined using a chemistry and transport model.
        J. Geophys. Res. [Atmos.] 101: 28,655-28,669.
Amendola, A.  A.; Hanes, N. B.  (1984) Characterization of indoor carbon monoxide levels produced by the automobile.
        In: Berglund, B.; Lindvall, T.; Sundell, J., eds. Indoor air. Volume 4: chemical characterization and personal exposure.
        Proceedings of the 3rd international conference on indoor air quality and climate; August; Stockholm, Sweden.
        Stockholm, Sweden: Swedish Council for Building Research; pp. 97-102.
Andreae, M. O. (1991) Biomass burning: its history, use, and distribution and its impact on environmental quality and global
        climate. In: Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric implications. Cambridge,
        MA: MIT Press; pp.  1-21.
Ault, K. (1999) Non-fire carbon monoxide deaths and injuries associated with the use of consumer products: annual estimates.
        Bethesda, MD: U.S. Consumer Product Safety Commission; June.
Bakwin, P. S.; Tans, P. P.; Novelli, P. C.  (1994) Carbon monoxide budget in the Northern Hemisphere. Geophys. Res. Lett.
        21:433-436.
Bekki, S.; Law, K. S.; Pyle,  J. A. (1994) Effect  of ozone  depletion on atmospheric CH4 and CO concentrations. Nature
        371:595-597.
Belles, F. E.; DeWerth, D. W.; Himmel, R. L. (1979) NOx emissions of furnaces and water heaters predominant in the California
        market. Presented at: 72nd annual meeting of the Air Pollution Control Association; June; Cincinnati, OH. Pittsburgh,
        PA: Air Pollution Control Association; paper no. 79-60.1.


                                                3-55

-------
Billick, I.; Johnson, D.; Moschandreas, D.; Relwani, S. (1984) An investigation of operational factors that influence emission rates
        from gas appliances. In: Berglund, B.; Lindvall, T.; Sundell, J., eds. Indoor air. Volume 4: chemical characterization and
        personal exposure. Proceedings of the 3rd international conference on indoor air quality and climate; August; Stockholm,
        Sweden. Stockholm, Sweden: Swedish Council for Building Research; pp. 181-187.
Billick, I. H.; Behrens, D. W.; Becker, E. W.; Garrison, C. A.; Wilson, A. L.; Colome, S. D.; Tian, Y. (1996) California residential
        indoor air quality study: an overview. In: Proceedings of the 1995 international gas research conference; November 1995;
        Cannes, France. Rockville, ME): Government Institutes, Inc.; pp. 1707-1716.
Bishop, G. A.; Stedman, D. H. (1990) On-road carbon monoxide emission measurement comparisons for the 1988-1989 Colorado
        oxy-fuels program. Environ. Sci. Technol. 24: 843-847.
Bishop, G. A.; Stedman, D. H. (1996) Measuring the emissions of passing cars. Ace. Chem. Res. 29: 489-495.
Borrazzo, J. E.; Davidson, C. I; Hendrickson, C. T. (1987) A statistical analysis of published gas stove emission factors for CO,
        NO, andNO2. In: Seifert, B.; Esdorn, H.; Fischer, M.; Rilden, H.; Wegner, J., eds. Indoor air '87: proceedings of the 4th
        international conference on indoor air quality and climate. Volume 1: volatile organic compounds, combustion gases,
        particles and fibres, microbiological agents; August; Berlin, Federal Republic of Germany. Berlin, Federal Republic of
        Germany: Institute for Water,  Soil and Air Hygiene; pp. 316-320.
Bowling, S. A. (1986) Climatology of high-latitude air pollution as illustrated by Fairbanks and Anchorage, Alaska. J. Clim. Appl.
        Meteorol. 25: 22-34.
Bradley, R. S.; Keimig, F. T.; Diaz, H.  F. (1992) Climatology of surface-based inversions in the North American Arctic.
        J. Geophys. Res. [Atmos.]  97: 15,699-15,712.
Bradley, K. S.; Stedman, D. H.; Bishop, G. A. (1999) A global inventory of carbon monoxide emissions from motor vehicles.
        In: Khalil, M. A. K.; Pinto, J. P.; Shearer, M. J., eds. Carbon monoxide [special issue of papers from an international
        conference; December  1997; Portland, OR]. Chemosphere Global Change Sci.  1: 65-72.
Bruhl, C.; Crutzen, P. J. (1999) Reductions in the anthropogenic emissions of CO and their effect  on CH4. In: Khalil, M. A. K.;
        Pinto, J. P.; Shearer, M. J., eds. Carbon monoxide [special issue  of papers from an international conference; December
         1997; Portland, OR]. Chemosphere Global Change Sci.  1: 249-254.
Burton, R. M.; Seila, R. A.; Wilson, W. E.; Pahl, D. A.; Mumford, J. L.; Koutrakis, P. (1990) Characterization of kerosene heater
        emissions inside two mobile homes. In: Indoor air '90: precedings of the 5th international conference on indoor air quality
        and  climate, volume 2, characteristics of indoor air;  July-August; Toronto, ON,  Canada. Ottawa, ON, Canada:
        International Conference on Indoor Air Quality and Climate, Inc.; pp. 337-342.
Carroll, M. A.; Hastie, D. R; Ridley, B. A.; Rodgers, M.  O.; Torres, A. L.; Davis, D. D.; Bradshaw, J. D.; Sandholm, S. T.; Schiff,
        H. I; Karecki, D. R; Harris, G.  W.; Mackay, G. I; Gregory, G.  L.; Condon, E. P.; Trainer, M.; Hubler, G.; Montzka,
        D. D.; Madronich, S.; Albritton, D. L.; Singh, H. B.; Beck, S. M.; Shipham, M. C.; Bachmeier, A. S. (1990) Aircraft
        measurement of NOX over the eastern Pacific and continental United States and implications for ozone production.
        J. Geophys. Res. [Atmos.]  95: 10,205-10,233.
Chameides, W. L.;Liu, S. C.; Cicerone,R. J. (1977) Possible variations in atmospheric methane. J. Geophys. Res. 82: 1795-1798.
Chan, C.-C.; Ozkaynak, H.; Spengler, J. D.; Sheldon, L.; Nelson, W.; Wallace, L. (1989) Commuter's exposure to volatile organic
        compounds, ozone, carbon monoxide, and nitrogen dioxide. Presented at: 82nd annual meeting of the Air and Waste
        Management Association; June; Anaheim, CA. Pittsburgh, PA: Air and Waste Management Association; paper no.
        89-34A.4.
Chan, C.-C.; Ozkaynak, H.; Spengler, J. D.; Sheldon, L.  (1991) Driver exposure to volatile organic compounds, CO, ozone and
        NO2 under different driving conditions. Environ.  Sci. Technol. 25: 964-972.
Chaney, L. W. (1978) Carbon monoxide automobile emissions measured from the interior of a traveling automobile. Science
        (Washington, DC) 199: 1203-1204.
Cicerone, R. J. (1988) How has the atmospheric concentration of CO changed? In: Rowland,  F. S.; Isaksen, I. S. A., eds. The
        changing atmosphere: report of the Dahlem workshop on the changing atmosphere; November 1987; Berlin, Federal
        Republic of Germany. Chichester, United Kingdom: John Wiley & Sons, Ltd.; pp. 49-61. (Physical, chemical, and earth
        sciences research report no. 7).
Cohen, J.; Iwamiya, R. (1998) Analyses of diurnal patterns in  hourly  and eight-hourly average  ambient carbon monoxide
        concentrations. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and
        Standards; contract no.  EPA 68-D7-0066.
Collins, W. J.; Stevenson, D.  S.; Johnson, C. E.; Derwent, R. G. (1997) Tropospheric ozone in a global-scale three-dimensional
        Lagrangian model and its response to NO^ emission controls.  J.  Atmos. Chem. 26: 223-274.
Colome, S. D.; Wilson, A. L.; Tian, Y. (1994) California residential indoor air quality study. Volume 2. Carbon monoxide and air
        exchange rate: an univariate and multivariate analysis. Chicago, IL: Gas Research Institute; report no. GRI-93/0224.3.
Colucci, J. M.; Begeman, C. R. (1969) Carbon monoxide in Detroit, New York, and Los Angeles  air. Environ. Sci. Technol.
        3:41-47.
Comrie, A. C.; Diem, J. E. (1999) Climatology and forecast modeling of ambient carbon monoxide in Phoenix, Arizona. Atmos.
        Environ. 33: 5023-5036.
                                                      3-56

-------
Conrad, R. (1996) Soil microorganisms as controllers of atmospheric trace gases (H2, CO, CH4, OCS, N2O and NO). Microbiol.
        Rev. 60: 609-640.
Conrad, R.; Seiler, W. (1985) Influence of temperature, moisture, and organic carbon on the flux of H2 and CO between soil and
        atmosphere: field studies in subtropical regions. J. Geophys. Res. [Atmos.] 90: 5699-5709.
Cortese, A. D.; Spengler, J. D. (1976) Ability of fixed monitoring stations to represent personal carbon monoxide exposure. J. Air
        Pollut. Control Assoc. 26:  1144-1150.
Cote, W. A.; Wade, W. A., Ill; Yocom, J. E. (1974) A study of indoor air quality. Washington, DC: U.S. Environmental Protection
        Agency, Office of Research and Development; report no. EPA-650/4-74-042.
Croes, B. E.; Effa, R. C.; Pasek, R. J. (1996) A comparison of the California emission inventory with 1995 ambient concentration
        data in Los Angeles. Presented at: 6th CRC on-road vehicle emissions workshop; March; San Diego, CA. Sacramento,
        CA: California Air Resources Board.
Davidson, C. I; Borrazzo, J. E.; Hendrickson, C. T. (1987) Pollutant emission factors for gas stoves: a literature survey. Research
        Triangle Park, NC: U. S. Environmental Protection Agency, Air and Energy Engineering Research Laboratory; report no.
        EPA-600/9-87-005.
Dignon, J.; Ando, T.; Brenninkmeijer, C.; Conny, J.; Granier, C.; Khalil, A.; King, G.; King, S.; Law, K.; Levine, J. S.;
        Rockmann, T.; Stedman, D.; Yung, Y.; Zafiriou, O. (1998) Carbon monoxide sources and sinks: working group summary
        report. In: Khalil, M. A. K.; Pinto, J. P.; Shearer, M. J., eds. Atmospheric carbon monoxide and its environmental effects
        [proceedings of an international conference]; December 1997; Portland, Oregon. Washington, DC: U.S. Environmental
        Protection Agency; report no. EPA/600/R-98/047; pp. 359-374.
Fan, C.; Zhang, J.; Cheung, J. P. (1997) Air pollutants emitted from household combustion sources. In: Measurement of toxic and
        related air pollutants. Volume 2: proceedings of an A& WMA/EPA specialty conference; April-May; Research Triangle
        Park, NC. Pittsburgh, PA: Air & Waste Management Association; pp. 837-846. (A&WMA publication VIP-74).
Federal Trade Commission.  (1994)  Tar, nicotine, and carbon monoxide of the smoke of 933 varieties  of domestic cigarettes.
        Washington, DC: Federal Trade Commission.
Federal Trade Commission.  (1997) Federal Trade Commission report to Congress for 1995, pursuant to the Federal Cigarette
        Labeling  and  Advertising   Act.   Washington,  DC:   Federal  trade   Commission.   Available:
        www.ftc.gov/os/1997/9708/cigrep97.htm [2000, February 14].
Federal Trade Commission. (1999) Cigar sales and advertising and promotional expenditures for calendar years 1996 and 1997.
        Washington,  DC: Federal Trade Commission. Available: www.ftc.gov/os/1999/9907/cigarreportl999.htm [2000,
        February 14].
Flachsbart, P. G.; Ott, W. R. (1984) Field surveys of carbon monoxide in commercial settings using personal exposure monitors.
        Washington, DC: U.S. Environmental Protection Agency, Office of Monitoring Systems and Quality Assurance; EPA
        report no. EPA-600/4-84-019.
Flachsbart, P. G.; Mack, G. A.; Howes, J. E.; Rodes, C. E. (1987) Carbon monoxide exposures of Washington commuters. JAPCA
        37: 135-142.
Fortmann, R.  C.;  Borrazzo, J.  E.; Davidson,  C.  I. (1984) Characterization  of parameters influencing indoor  pollutant
        concentrations. In: Berglund, B.; Lindvall, T.; Sundell, J., eds. Indoor air: proceedings of the 3rd international conference
        on indoor air quality and climate, v. 4, chemical characterization and personal exposure; August; Stockholm, Sweden.
        Stockholm, Sweden:  Swedish Council for Building Research; pp. 259-264.
Fuglestvedt, J.  S.; Jonson, J. E.; Isaksen, I. S. A. (1994) Effects of reductions in stratospheric ozone on tropospheric chemistry
        through changes in photolysis rates. Tellus Ser. B 46: 172-192.
Fujita, E. M.; Croes, B. E.; Bennett, C. L.; Lawson, D. R.; Lurmann, F. W.; Main, H. H. (1992) Comparison of emission inventory
        and ambient concentration ratios of CO, NMOG, and NOX in California's south coast air basin. J. Air Waste Manage.
        Assoc. 42: 264-276.
Gertler, A. W.; Sagebiel, J. C.; Dippel, W. A.; O'Connor,  C. M. (1999) The impact of California phase 2 reformulated gasoline
        on real-world vehicle emissions. J. Air Waste Manage. Assoc. 49: 1339-1346.
Grosjean, D.; Fung, K. (1984) Hydrocarbons and carbonyls in Los Angeles air. J. Air Pollut. Control Assoc.  34: 537-543.
Haan, D.; Martinerie, P.; Raynaud, D. (1996) Ice core data of atmospheric carbon monoxide over Antarctica and Greenland during
        the last 200 years. Geophys. Res. Lett. 23: 2235-2238.
Hallock-Waters, K. A.; Doddridge, B. G.; Dickerson, R.  R.; Spitzer, S.; Ray, J. D. (1999)  Carbon  monoxide in the U.S.
        mid-Atlantic troposphere: evidence for a decreasing trend. Geophys. Res. Lett. 26: 2861-2864.
Hameed, S.; Pinto, J.  P.;  Stewart, R. W. (1979) Sensitivity  of the  predicted CO-OH-CH4 perturbation to tropospheric NO^
        concentrations. J. Geophys. Res. C:  Oceans Atmos. 84: 763-768.
Hard, T. M.; Mehrabzadeh, A. A.; Chan, C. Y.; O'Brien, R.  J. (1992) PAGE measurements of tropospheric HO with measurements
        and model of interferences. J. Geophys. Res. [Atmos.] 97: 9795-9817.
Harke, H.-P.; Peters, H. (1974) Zum Problem des Passivrauchens. III. Uber den EinfluB desRauchens auf die CO-Konzentration
        im Kraftfahrzeug bei Fahrten im Stadtgebiet [The problem of passive smoking. III. The influence of smoking on the CO
        concentration in  driving automobiles]. Int. Arch. Arbeitsmed.  33: 221-229.
                                                      3-57

-------
Hartwell, T. D.; Clayton, C. A.; Ritchie, R. M.; Whitmore, R. W.; Zelon, H. S.; Jones, S. M.; Whitehurst, D. A. (1984) Study of
        carbon monoxide exposure of residents of Washington, DC and Denver, Colorado. Research Triangle Park, NC: U.S.
        Environmental Protection Agency, Office of Research and Development, Environmental Monitoring Systems Laboratory;
        report no. EPA-600/4-84-031.
Haste, T. L.; Chinkin, L. R.; Kumar, N.; Lurmann, F. W.; Hurwitt, S. B. (1998) Use of ambient data to evaluate a regional emission
        inventory for the San Joaquin Valley in California. In: Chow, J.; Koutrakis,  P., eds. PM25: a fine particle standard.
        Volume I: proceedings of an international specialty conference; January; Long Beach, CA. Pittsburgh, PA: Air & Waste
        Management Association; pp. 451-462. (A&WMA publication VIP-81).
Hedrick, R. L.; Krug, E. K. (1995) Conventional Research House measurements of emissions from unvented gas space heaters.
        Chicago, IL: Gas Research Institute; report no. GRI-95/0205.
Himmel, R. L.; DeWerth, D. W. (1974) Evaluation of the pollutant emissions from gas-fired ranges. Project EP-1-23: analysis of
        flue products from gas-fired appliances. Cleveland, OH: American Gas Association Laboratories; report no. 1492.
Hofmann, D. J.; Peterson, J. T.; Rosson, R. M., eds. (1996) Climate monitoring and diagnostics laboratory no. 23. Summary report
         1994-95.  Boulder,  CO: U.S.  Department  of  Commerce,  National  Oceanic and Atmospheric Administration,
        Environmental Research Laboratories.
Holcomb, L.  C. (1993) Indoor air quality and environmental tobacco smoke: concentration and exposure. Environ. Int. 19: 9-40.
Humphreys, M. P.; Knight, C. V.; Pinnix, J. C. (1986) Residential wood combustion impacts on  indoor carbon monoxide and
        suspended particulates In: Proceedings of the 1986 EPA/APCA symposium on measurement of toxic air pollutants; April;
        Raleigh, NC. Research Triangle Park, NC: U. S. Environmental Protection Agency, Environmental Monitoring Systems
        Laboratory; pp. 736-747; report no. EPA-600/9-86-013.
Ingalls, M. N.; Smith, L. R.; Kirksey, R. E. (1989) Measurement of on-road vehicle emission factors in the California South Coast
        Air  Basin:  volume I,  regulated  emissions. Atlanta,  GA: Coordinating Research Council,  Inc.;  report no.
        CRC-APRAC-AP-4-SCAQS-l.
Inman, R. E.;  Ingersoll, R.  B.; Levy, E. A. (1971) Soil:  a natural sink for carbon  monoxide. Science (Washington, DC)
         172: 1229-1231.
Ipps, D.  T.; Popejoy, D. (1998) An updated verification of the on-road vehicle emission inventory for the Los Angeles basin using
        ambient air quality data. Presented at: 8th CRC on-road vehicle emissions workshop; April; San Diego, CA. Sacramento,
        CA: California Air Resources Board.
Jaasma,  D. R.; Stern, C. H.; Honeycutt, D. J.; Fugler, D. (1995) Effect of glass doors on carbon monoxide spillage from masonry
        fireplaces. Presented at: 88th annual meeting and exhibition of the Air & Waste  Management Association; June;
        San Antonio, TX. Pittsburgh, PA: Air & Waste Management Association; paper no. 95-RP137.01.
Jacob, D. J.; Wofsy, S. C. (1990) Budgets of reactive nitrogen, hydrocarbons, and ozone over the Amazon forest during the wet
        season. J. Geophys. Res. [Atmos.] 95: 16,737-16,754.
Jeffries, H. E. (1995) Photochemical air pollution. In: Singh, H. B., ed. Composition, chemistry, and climate of the atmosphere.
        New York, NY: Van Nostrand Reinhold; p. 337.
Johnson, T. (1984) A study of personal exposure to carbon monoxide in Denver, Colorado. Research Triangle Park, NC: U.S.
        Environmental Protection Agency, Environmental Monitoring Systems Laboratory; report no. EPA-600/4-84-014.
Johnson, D.; Billick, I.; Moschandreas, D.; Relwani, S. (1984) Emission rates from unvented gas appliances.  In: Berglund, B.;
        Lindvall, T.; Sundell, J., eds. Indoor air: proceedings of the 3rd international conference on indoor air quality and climate,
        v. 4, chemical characterization and personal exposure; August; Stockholm,  Sweden. Stockholm, Sweden:  Swedish
        Council for Building Research; pp. 367-373.
Johnson, T.; Capel, J.; Paul, R.; Wijnberg, L. (1992) Estimation of carbon monoxide exposures and associated carboxyhemoglobin
        levels in Denver residents  using a probabilistic version of NEM. Research Triangle Park, NC:  U.S. Environmental
        Protection Agency, Office of Air Quality Planning and Standards; contract no. 68-DO-0062.
Kanakidou, M.; Crutzen, P. J. (1993) Scale problems in global tropospheric chemistry modeling: comparison of results obtained
        with  a three-dimensional  model,  adopting longitudinally uniform  and varying  emissions  of NOX  and NMHC.
        Chemosphere 26: 787-801.
Kanakidou, M.; Singh, H. B.; Valentin, K. M.; Crutzen, P.  J. (1991) A two-dimensional study of ethane and propane oxidation
        in the troposphere. J. Geophys. Res. [Atmos.] 96: 15,395-15,413.
Kern, R. A.; Knoedler, J. A.; Trester, R.  J.  (1990) Automotive carbon monoxide emissions in a closed garage.  Presented at:
        International congress and exposition; February-March; Detroit, MI. Warrendale, PA: Society of Automotive Engineers;
        pp.  33-38. (SAE technical paper series no. 900499).
Khalil, M. A. K.; Rasmussen, R. A. (1988a) Carbon monoxide in the earth's atmosphere: indications of a global increase. Nature
        (London) 332: 242-245.
Khalil, M. A. K.; Rasmussen, R. A. (1988b) Carbon monoxide in an urban environment: application of a receptor model for source
        apportionment. JAPCA38: 901-906.
Khalil, M. A.  K.; Rasmussen, R.  A. (1989)  Urban carbon monoxide:  contributions  of  automobiles  and wood burning.
        Chemosphere  19: 1383-1386.
                                                      3-58

-------
Khalil, M. A. K.; Rasmussen, R. A. (1994) Global decrease in atmospheric carbon monoxide concentration. Nature (London)
        370:639-641.
Khalil, M. A. K.; Rasmussen, R. A. (1999) Non-automotive sources of carbon monoxide in urban areas. In: Khalil, M. A. K.; Pinto,
        J. P.; Shearer, M. J., eds. Carbon monoxide [special issue of papers from an international conference; December 1997;
        Portland, OR]. Chemosphere Global Change Sci. 1:111-114.
King, G. M. (1999) Characteristics and significance of atmospheric carbon monoxide consumption by soils. In: Khalil, M. A. K.;
        Pinto, J. P.; Shearer, M. J., eds. Carbon monoxide [special issue of papers from an international conference; December
        1997; Portland, OR]. Chemosphere Global Change Sci.  1: 53-63.
Klepeis, N. E.; Ott, W. R.; Switzer, P. (1995) Modeling the time series of respirable suspended particles and carbon monoxide
        from multiple smokers: validation in two public  smoking lounges. Presented at: 88th annual meeting and exhibition of
        the Air &  Waste Management Association; June; San Antonio, TX. Pittsburgh, PA: Air & Waste  Management
        Association; paper no. 95-MPH.03.
Klepeis, N. E.; Ott, W. R.; Switzer, P. (1996) A multiple-smoker model for predicting indoor air quality in public lounges. Environ.
        Sci. Technol. 30: 2813-2820.
Klepeis, N. E.; Ott, W. R.; Repace, J. L. (1999) The effect of cigar smoking on indoor levels of carbon monoxide and particles.
        J. Exposure Anal. Environ. Epidemiol. 9: 622-635.
Koontz, M. D.; Nagda, N. L. (1987) Survey of factors affecting NO2 concentrations. In: Seifert, B.; Esdorn, H.; Fischer, M.; Rilden,
        H.; Wegner, J., eds. Indoor air '87: proceedings of the 4th international conference on indoor air quality and climate, v.
        1, volatile organic compounds, combustion gases, particles and fibres, microbiological agents; August; Berlin, Federal
        Republic of Germany. Berlin, Federal Republic of Germany: Institute for Water, Soil and Air Hygiene; pp. 430-434.
Koontz, M. D.; Nagda, N. L. (1988) A topical report on a field  monitoring study of homes with unvented gas space heaters.
        Volume III—methodology and results. Chicago, IL: Gas Research Institute; report no. GRI-87/0044.2.
Koontz, M. D.; Rector, H. E. (1995) Estimation of distributions for residential air exchange rates: final report. Washington, DC:
        U.S. Environmental Protection Agency,  Office of Pollution Prevention and Toxics; contract nos. 68-D9-0166 and
        68-D3-0013.
Koontz, M. D.; Mehegan, L. L.; Nagda, N. L. (1992) Distribution and use of cooking appliances that can affect indoor air quality:
        topical report (November 1985—December 1991). Chicago, IL: Gas Research Institute; report no. GRI-93/0013.
Korc, M. E.; Roberts, P. T.; Chinkin, L. R.; Main,  H. H. (1993) Comparison of emission inventory and ambient concentration
        ratios of NMOC, NOX, and CO in the Lake Michigan air quality region. Santa Rosa, CA: Sonoma Technology, Inc.; report
        no. STI-90218-1357-FR
Krol, M.; Van Leeuwen, P. J.; Lelieveld, J. (1998) Global  OH trend inferred from methylchloroform measurements. J. Geophys.
        Res. [Atmos.] 103: 10,697-10,711.
Lagus Applied Technology, Inc. (1995) Air change rates in non-residential buildings in California. Sacramento, CA: California
        Energy Commission; contract no. 400-91-034; July.
Law, K. S. (1999) Theoretical studies of carbon monoxide distributions, budgets and trends. In: Khalil, M. A. K.; Pinto, J. P.;
        Shearer, M. J., eds. Carbon monoxide [special issue of papers from an international conference; December 1997;
        Portland, OR]. Chemosphere Global Change Sci. 1:19-31.
Lawson, D. R.; Groblicki, P. J.; Stedman, D. H.; Bishop, G. A.; Guenther, P. L. (1990) Emissions from in-use motor vehicles in
        Los Angeles: a pilot study of remote sensing and the inspection and maintainence program. J. Air Waste Manage. Assoc.
        40: 1096-1105.
Leaderer, B.  P. (1982) Air pollutant emissions from kerosene space heaters.  Science (Washington, DC) 218: 1113-1115.
Leaderer, B. P.; Stolwijk, J. A. J.; Zagraniski, R. T.;  Qing-Shan, M. (1984) A field study of indoor air contaminant levels
        associated  with unvented combustion sources. Presented  at: 77th annual  meeting  of the Air Pollution Control
        Association; June; San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; paper no. 84-33.3.
Lobert, J. M.; Scharffe, D. H.; Hao, W.-M.; Kuhlbusch,  T. A.; Seuwen, R.; Warneck, P.; Crutzen, P. J. (1991) Experimental
        evaluation of biomass burning emissions: nitrogen and carbon containing compounds. In:  Levine, J. S., ed. Global
        biomass burning: atmospheric, climatic, and biospheric implications. Cambridge, MA: MIT Press; pp. 289-304.
Lofroth, G.; Burton, R. M.; Forehand, L.; Hammond, S. K.; Seila, R. L.; Zweidinger, R. B.; Lewtas, J. (1989) Characterization of
        environmental tobacco smoke. Environ. Sci. Technol. 23: 610-614.
Logan, J. A.; Prather, M. J.; Wofsy, S. C.; McElroy, M. B.  (1981) Tropospheric chemistry: a global perspective. J. Geophys. Res.
        C: Oceans Atmos. 86: 7210-7254.
Magliano, K. L.; Lurmann, F. W.; Watson, J. G. (1993) Reconciliation of emissions inventory and ambient data in the San Joaquin
        Valley and the San Francisco Bay area. Presented at: 86th annual meeting & exhibition of the Air & Waste Management
        Association; June; Denver, CO. Pittsburgh, PA:  Air & Waste Management Association; paper no. 93-WP-100.8.
Mahieu, E,; Zander, R.; Delbouille, L.; Demoulin, P.; Roland, G.; Servais, C. (1997) Observed trends in total vertical column
        abundances of atomspheric gases from IR solar spectra recorded at the Jungfraujoch. J. Atmos. Chem. 28: 227-243.
Marr, L. C.; Morrison, G. C.; Nazaroff,  W. W.; Harley, R. A. (1998) Reducing the risk of accidental death due to vehicle-related
        carbon monoxide poisoning. J. Air Waste Manage. Assoc. 48:  899-906.
                                                      3-59

-------
McCarthy, S. M.; Yarmac,R. F.; Yocom, J. E. (1987) Indoor nitrogen dioxide exposure: the contribution from unvented gas space
         heaters. In:  Seifert, B.; Esdorn, H.; Fischer, M.; Rilden, H.; Wegner, J., eds. Indoor air '87: proceedings of the 4th
         international conference on indoor air quality and climate, v. 1, volatile organic compounds, combustion gases, particles
         and fibres, microbiological agents; August; Berlin, Federal Republic of Germany. Berlin, Federal Republic of Germany:
         Institute for Water, Soil and Air Hygiene; pp. 478-482.
Menkedick, J.; Niemuth, N.; Hartford, P.; Landstrom, D. K. (1993) Metered ranges, cooktops and ovens in the northern Illinois
         gas residential load study data base: topical report. Chicago, IL: Gas Research Institute; report no. GRI-93/0204; July.
Miller, S. L.; Hannigan, M. (1999) Indoor carbon monoxide levels from the use of unvented natural gas fireplaces in Boulder, CO.
         Presented at: The first National Science Foundation international conference on indoor air health; May; Denver, CO.
Moschandreas, D. J.; Zabransky, J., Jr. (1982) Spatial variation of carbon monoxide and oxides of nitrogen concentrations inside
         residences. Environ. Int. 8: 177-183.
Moschandreas, D. J.; Relwani, S. M.; O'Neill, H. J.; Cole, J.  T.; Elkins, R. H.; Macriss, R. A. (1985) Characterization of emission
         rates from indoor combustion sources. Chicago, IL: Gas Research Institute; report no. GRI 85/0075.
Mount, G. H.; Williams, E. J. (1997) An overview of the tropospheric OH photochemistry experiment, Fritz Peak/Idaho Hill,
         Colorado, fall 1993. J. Geophys.  Res. [Atmos.] 102: 6171-6186.
Moxley,  J. M.; Cape, J. N.  (1997) Depletion  of carbon monoxide  from the nocturnal boundary layer.  Atmos. Environ.
         31: 1147-1155.
Mueller Associates. (1985) Indoor air quality environmental information handbook: combustion sources. Washington, DC: U.S.
         Department of Energy, Office of Environmental Analysis; report no. DOE/EV/10450-1.
Mumford, J. L.; Lewtas, I; Burton, R. M.; Svendsgaard, D. B.; Houk, V. S.; Williams, R. W.; Walsh, D. B.; Chuang, J. C. (1990)
         Unvented kerosene heater emissions in mobile homes: studies on indoor air particles, semivolatile organics, carbon
         monoxide, and mutagenicity. In: Indoor air '90: precedings of the 5th international conference on indoor air quality and
         climate, volume 2, characteristics of indoor air; July-August; Toronto,  ON, Canada. Ottawa, ON, Canada: International
         Conference  on Indoor Air Quality and Climate, Inc.; pp. 257-262.
Mumford, J. L.; Williams, R. W.; Walsh, D. B.; Burton, R. M.;  Svendsgaard, D.  J.; Chuang, J. C.; Houk, V. S.; Lewtas, J. (1991)
         Indoor air pollutants from unvented kerosene heater emissions in mobile homes: studies on particles, semivolatile
         organics, carbon monoxide, and mutagenicity. Environ. Sci. Technol.  25: 1732-1738.
Murray, D. M.; Burmaster, D. E. (1995) Residential air exchange rates in the United States: empirical and estimated parametric
         distributions by season and climatic region. Risk Anal. 15: 459-465.
Nagda, N. L.; Koontz, M. D.; Fortmann, R. C.; Billick, I. H. (1987) Detailed evaluation of range-exhaust fans. In: Seifert, B.;
         Esdorn, H.; Fischer, M.; Rilden, H.; Wegner, J., eds. Indoor air '87: proceedings of the 4th international conference on
         indoor air quality and climate. Volume 3:  developing countries, guaranteeing adequate indoor air quality, control
         measures, ventilation effectiveness, thermal climate and comfort, policy and strategies; August; Berlin, Federal Republic
         of Germany. Berlin, Federal Republic of Germany: Institute for Water, Soil and Air Hygiene; pp. 324-328.
Nagda, N. L.; Koontz, M. D.; Billick, I. H.; Leslie, N. P.; Behrens, D. W. (1996) Causes and consequences of backdrafting of
         vented gas appliances. J. Air Waste Manage. Assoc. 46: 838-846.
National Aeronautics and Space Administration, Panel for Data Evaluation. (1997) Chemical kinetics and photochemical data for
         use in stratospheric modeling: evaluation number 12. Pasadena, CA: National Aeronautics and Space Administration,
         Jet Propulsion Laboratory; JPL publication 97-4.
National  Oceanic and Atmospheric  Administration. (1999a) Global distribution of atmospheric carbon monoxide.  Boulder,  CO:
         Climate Monitoring & Diagnostics Laboratory, Carbon Cycle Group. Available: ww w .cmdl.noaa.gov/ccgg/figures/corug.gif
         [2000, February 14].
National  Oceanic and Atmospheric Administration. (1999b) CO trends and growth rates. Boulder, CO: Climate Monitoring &
         Diagnostics Laboratory, Carbon Cycle Group. Available: www.cmdl.noaa.gov/ccgg/figures/cotrend_global.gif [2000,
         February 14].
National Research Council. (1986) Environmental tobacco smoke: measuring exposures and assessing health effects. Washington,
         DC: National Academy Press.
Novelli, P. C.; Masarie, K. A.; Tans, P. P.; Lang, P. M.  (1994) Recent changes in atmospheric carbon monoxide. Science
         (Washington, DC) 263:  1587-1590.
Novelli, P. C.; Connors, V. S.; Reichle, H. G., Jr.; Anderson, B. E.; Brenninkmeijer, C. A. M.; Brunke, E. G.; Doddridge, B. G.;
         Kirchhoff, V. W. J. H.; Lam, K. S.; Masarie, K. A.; Matsuo, T.; Fairish, D. D.; Scheel, H. E.; Steele, L.  P. (1998)
         An internally consistent set of globally distributed atmospheric carbon monoxide mixing ratios developed using results
         from an intercomparison of measurements. J. Geophys. Res. [Atmos.] 103:  19,285-19,293.
Ott, W.; Langan, L.; Switzer, P. (1992) A time series model for cigarette smoking activity patterns: model validation for carbon
         monoxide and respirable  particles in a chamber and an automobile. In:  Measuring, understanding, and predicting
         exposures in the 21 st century: proceedings of the conference; November 1991; Atlanta, GA. J. Exposure Anal. Environ.
         Epidemiol. 2(suppl. 2): 175-200.
                                                       3-60

-------
Ott, W.; Switzer, P.; Willits, N. (1994) Carbon monoxide exposures inside an automobile traveling on an urban arterial highway.
        J. Air Waste Manage. Assoc. 44: 1010-1018.
Ott, W. R.; Klepeis, N. E.; Switzer, P.  (1995) Modeling environmental tobacco smoke in the home using transfer functions.
        Presented at: 88th annual meeting and exhibition of the Air & Waste Management Association; June; San Antonio, TX.
        Pittsburgh, PA: Air & Waste Management Association; paper no. 95-WP84B.03.
Pacyna, J. M.; Graedel, T. E.  (1995) Atmospheric emissions inventories: status and prospects. Annu. Rev. Energy Environ.
        20: 265-300.
Park, J.-H.; Spengler, J. D.; Yoon, D.-W.; Dumyahn, T.; Lee, K.; Ozkaynak, H. (1998) Measurement of air exchange rate of
        stationary vehicles and estimation of in-vehicle exposure. J. Exposure Anal. Environ. Epidemiol. 8: 65-78.
Pierce, T. E.; Dudek, M. P. (1996) Biogenic emission estimates for 1995. In: The emission inventory: key to planning, permits,
        compliance, and reporting; proceedings of a specialty conference; September; New Orleans, LA. Pittsburgh, PA: Air &
        Waste Management Association; pp. 309-326. (AWMA publication VIP-65).
Pierson, W. R.; Gertler, A. W.; Bradow, R. L. (1990) Comparison of the SCAQS tunnel study with other on-road vehicle emission
        data. J. Air Waste Manage. Assoc. 40: 1495-1504.
Pierson, W. R; Gertler, A. W.; Robinson, N. F.; Sagebiel, J. C.; Zielinska, B.; Bishop, G. A.; Stedman, D. H.; Zweidinger, R. B.;
        Ray, W. D. (1996) Real-world automotive emissions—summary of studies in the Fort McHenry and Tuscarora mountain
        tunnels. In: Parrish, D.; Trainer, M.; Rao. S. T.; Solomon, P. A., eds. A&WMA international specialty conference on
        regional  photochemical measurements and modeling, part 2; November 1993; San Diego, CA. Atmos. Environ.
        30: 2233-2256.
Pinto, J. P.; Khalil, M. A. K. (1991) The stability of tropospheric OH during ice ages, inter-glacial epochs and modern times. Tellus
        43B: 347-352.
Pollack, A. K.; Bunker, A. M.; Fieber, J. K.; Heiken, J. G.; Cohen, J. P.; Shepard, S. B.; Schleyer, C. H.; Yarwood, G. (1998)
        Revision of light-duty vehicle emission inventories using real-world measurements-Auto/Oil Program, phase II. J. Air
        Waste Manage. Assoc. 48: 291-305.
Poppe, D.; Zimmermann, J.; Bauer, R.; Brauers, T.; Brilning, D.; Callies, J.; Dorn, H. -P.; Hofzumahaus, A.; Johnen, F. -J.; Khedim,
        A.; Koch, H.; Koppmann, R.; London, H.; Milller, K. -P.; Neuroth, R.; Plass-Dulmer, C.; Platt, U.; Rohrer, F.; Roth, E. -P.;
        Rudolph, J.; Schmidt, U.; Wallasch, M.; Ehhalt, D. H. (1994) Comparison of measured OH concentrations with model
        calculations. J. Geophys. Res.  [Atmos.] 99: 16,633-16,642.
Prinn, R.; Cunnold, D.;  Simmonds, P.; Alyea, F.; Boldi, R.; Crawford, A.; Fraser, P.; Gutzler, D.; Hartley, D.; Rosen, R.;
        Rasmussen, R. (1992) Global average concentration and trend for hydroxyl radicals deduced from ALE/GAGE
        trichloroethane (methyl chloroform) data  for 1978-1990. J. Geophys. Res. [Atmos.] 97: 2445-2461.
Prinn, R. G.; Weiss, R. F.; Miller, B. R.; Huang, I; Alyea, F. N.; Cunnold, D. M.; Fraser, P. I; Hartley, D. E.; Simmonds, P. G.
        (1995) Atmospheric trends and lifetime of CH3CC13 and global OH concentrations. Science (Washington, DC) 269:
         187-192.
Research Triangle Institute. (1990) An investigation of infiltration and indoor air quality. Albany, NY: New York State Energy
        Research and Development Authority; report no. NYERDA 90-11.
Rickert, W. S.; Robinson, J. C.; Collishaw, N. (1984) Yields of tar, nicotine, and carbon monoxide in the sidestream smoke from
         15 brands of Canadian cigarettes. Am. J. Public Health 74: 228-231.
Rinsland, C. P.; Levine, J. S. (1985) Free tropospheric carbon monoxide concentrations in 1950 and 1951 deduced from infrared
        total column amount measurements. Nature (London) 318: 250-254.
Ritter, J. A.; Barrick, J. D. W.; Watson, C. E.; Sachse, G. W.; Gregory, G. L.; Anderson, B.  E.; Woerner, M. A.; Collins, J. E., Jr.
        (1994) Airborne boundary layer flux measurements of trace species over Canadian  boreal forest and northern wetland
        regions. J. Geophys. Res. [Atmos.] 99:  1671-1685.
Robinson, E.; Bell, G. B., Jr. (1956) Low-level temperature structure under Alaskan ice fog conditions. Bull. Am. Meteorol. Soc.
        37: 506-513.
Rodes, C.; Sheldon, L.; Whitaker, D.; Clayton, A.; Fitzgerald, K.; Flanagan, J.; DiGenova, F.; Hering,  S.; Frazier, C. (1998)
        Measuring concentrations of selected air pollutants inside California vehicles [final report]. Sacramento, CA: California
        Environmental Protection Agency, Air Resources Board; contract no. 95-339.
Rogers, J. D. (1990) Ultraviolet  absorption cross sections and atmospheric photodissociation rate constants of formaldehyde.
        J. Phys. Chem. 94:4011-4015.
Ryan, J. V.; McCrillis, R. C. (1994) Analysis of emissions from residential natural gas furnaces. Presented at: 87th annual meeting
        and exhibition of the Air & Waste Management  Association; June; Cincinnati, OH.  Pittsburgh, PA: Air & Waste
        Management Association; paper no. 94-WA75A.04.
Seila, R. L.; Lonneman, W. A.;Meeks, S. A.  (1989) Determination of C2 to C12 ambient air hydrocarbons in 39 U.S. cities from
         1984 through 1986. Research Triangle Park, NC: U.S. Environmental Protection Agency, Atmospheric Research and
        Exposure Assessment Laboratory; report no. EPA/600/3-89/058.
                                                      3-61

-------
Seller, W.; Conrad, R. (1987) Contribution of tropical ecosystems to the global budget of trace gases especially CH4, H2, CO and
        N2O. In: Dickinson, R. E., ed. The geophysiology of Amazonia: vegetation and climate interactions. New York, NY: John
        Wiley; pp. 133-162. (Glantz, H. H., ed. Wiley series in climate and biosphere).
Shadwick, D.; Glen, G.; Lakkadi, Y.; Lansari, A.; del Valle-Torres, M. (1997) Analysis of carbon monoxide for the Denver,
        Colorado MSA. Research Triangle  Park,  NC: U.S.  Environmental Protection Agency,  Office of Research and
        Development, National Exposure Research Laboratory; contract no. 68-D5-0049; December.
Shadwick, D.; Glen, G.; King, J.; Chen, X. (1998a) Analysis of carbon monoxide for the Los Angeles, California CMSA. Research
        Triangle Park, NC: U.S. Environmental Protection Agency, Office of Research and Development, National Exposure
        Research Laboratory; contract no. 68-D5-0049; June.
Shadwick, D.; Glen, G.; King, J.; Chen, X. (1998b) Analysis of carbon monoxide for the New York, New York CMSA. Research
        Triangle Park, NC: U.S. Environmental Protection Agency, Office of Research and Development, National Exposure
        Research Laboratory; contract no. 68-D5-0049; June.
Shadwick, D.; Glen, G.; King, J.; Chen, X. (1998c) Analysis of carbon monoxide for the Phoenix,  Arizona  MSA. Research
        Triangle Park, NC: U.S. Environmental Protection Agency, Office of Research and Development, National Exposure
        Research Laboratory; contract no. 68-D5-0049; June.
Slack, H. H.; Heumann, M. A. (1997) Use of unvented residential heating appliances—United States, 1988-1994. Morb. Mortal.
        Wkly. Rep. 46: 1221-1224.
Spicer, C. W.; Billick, I. H. (1996) Persistence and fate of natural gas appliance emissions. In: Proceedings of the  1995
        international gas research conference; November 1995; Cannes, France. Rockville, MD: Government Institutes, Inc.; pp.
        1752-1761.
Stedman, D. H.; Bishop, G. A.; Aldrete, P.; Slott, R. S. (1997) On-road evaluation of an automobile emission test program.
        Environ. Sci. Technol. 31: 927-931.
Stedman, D. H.; Bishop, G. A.; Slott, R. S. (1998)  Repair avoidance and evaluating inspection and maintenance programs.
        Environ. Sci. Technol. 32: 1544-1545.
Stephens, R. D.; Cadle, S. H. (1991) Remote sensing measurements of carbon monoxide emissions from on-road vehicles. J. Air
        Waste Manage. Assoc.  41: 39-46.
Sterling, T. D.; Sterling, E. (1979) Carbon monoxide levels in kitchens and homes with gas cookers. J. Air Pollut. Control Assoc.
        29:238-241.
Sze, N. D. (1977) Anthropogenic CO emissions: implications for the atmospheric CO-OH-CH4 cycle. Science (Washington, DC)
        195: 673-675.
Tegart, W. J. McG.; Sheldon, G. W., eds. (1993) Climate change 1992: the supplementary report to the IPCC impacts assessment.
        Canberra, Australia: Australian Government Publishing Service.
Thompson, A. M.; Cicerone, R. J. (1986) Possible perturbations to atmospheric CO, CH4, and OH. J. Geophys. Res. [Atmos.]
        91: 10,853-10,864.
Thompson, A. M.; Chappellaz, J. A.; Fung, I. Y.; Kucsera, T.  L.  (1993) The atmospheric CH4 increase since the last glacial
        maximum. (2). Interactions with oxidants. Tellus 45B: 242-257.
Thrasher, W. H.; DeWerth, D. W. (1979) Evaluation of the pollutant emissions from  gas-fired room heaters. Cleveland, OH:
        American Gas Association, Cleveland Laboratories; research report no. 1515.
Tiegs, P.; Bighouse, R. (1994) Spillage of combustion byproducts from woodstoves operated in negative pressure environments.
        Presented at: 87th annual meeting and exhibition of the Air  & Waste Management Association; June; Cincinnati, OH.
        Pittsburgh, PA: Air & Waste Management Association; paper no. 94-FA150.04.
Tobacco Information and Prevention Source. (1996)  1995 national household survey on drug abuse, tobacco related statistics,
        SAMHSA,   August   1996.  Atlanta,  GA:  U.S. Centers  for  Disease  Control  and  Prevention.  Available:
        www.cdc.gov/tobacco/samhsa.htm [2000, February 14].
Traynor, G. W.; Anthon, D. W.; Hollowell, C. D. (1982) Technique for determining pollutant emissions from a gas-fired range.
        Atmos. Environ.  16: 2979-2987.
Traynor, G. W.; Allen, J. R.; Apte,M. G.; Girman, J. R.; Hollo well, C. D. (1983) Pollutant emissions from portable kerosene-fired
        space heaters. Environ. Sci. Technol. 17: 369-371.
Traynor, G. W.; Apte, M. G.; Carruthers, A. R.; Dillworth, J. F.; Grimsrud, D. T. (1984) Pollutant emission rates from unvented
        infrared  and convective gas-fired space heaters.  Berkeley, CA:  U.S. Department  of Energy, Lawrence Berkeley
        Laboratory; report no. LBL-18258.
Traynor, G. W.; Girman, J.  R.; Apte, M. G.; Dillworth, J. F.; White, P. D. (1985) Indoor air pollution due to emissions from
        unvented gas-fired space heaters. J. Air Pollut. Control Assoc. 35:  231-237.
Traynor, G. W.; Apte, M. G.; Carruthers, A.  R.; Dillworth, J. F.; Grimsrud, D. T.; Gundel, L. A. (1987) Indoor air pollution due
        to emissions from wood-burning stoves. Environ. Sci. Technol. 21: 691-697.
                                                      3-62

-------
Traynor, G. W.; Aceti, J. C.; Apte, M. G.; Smith, B. V.; Green, L. L.; Smith-Reiser, A.; Novak, K. M.; Moses, D. O. (1989)
         Macromodel for assessing residential concentrations of combustion-generated pollutants: model development and
         preliminary predictions for CO, NO2, and respirable suspended particles. Berkeley, CA: U.S. Department of Energy,
         Lawrence Berkeley Laboratory; report no. LBL-25211.
Turk, B. H.; Grimsrud, D. T.; Brown, J. T.;  Geisling-Sobotka, K. L.; Harrison, J.; Prill, R.  J. (1989) Commercial  building
         ventilation rates and particle concentrations. ASHRAE Trans. 95(part 1): 422-433.
Turner, S.; Cyr, L.; Gross, A. J. (1992) The measurement of environmental tobacco smoke in 585 office environments. Environ.
         Int. 18: 19-28.
U.S. Census Bureau. (1998) American housing survey: detailed tables from the 1993 AHS-N data chart. How do you fuel your
         home? - Table 2-5. Washington, DC: U.S. Department of Commerce.
U.S. Census Bureau. (1999) USA Statistics in brief—law, education, communications, transportation, housing. Washington, DC:
         U.S. Department of Commerce. Available: www.census.gov/statab/www/part2.html [2000, February 14].
U.S. Environmental Protection Agency. (1981) U.S. Environmental Protection Agency Intra-Agency Task Force on Air Quality
         Indicators. Research Triangle Park, NC:  Office of Air Quality Planning and Standards; report no. EPA-450/4-81-015.
U.S. Environmental Protection Agency. (1990) National air quality and emissions trends report, 1988. Research Triangle Park,
         NC: Office of Air Quality Planning and  Standards; report no. EPA/450/4-90/002.
U. S. Environmental Protection Agency. (199 la) User's guide to MOBILE4.1 (mobile source emission factor model). Ann Arbor,
         MI: Office of Mobile Sources; report no. EPA-AA-TEB-91-01.
U.S. Environmental Protection Agency. (1991b) Air quality criteria for carbon monoxide. Research Triangle Park, NC: Office of
         Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA/600/8-90/045F.
U.S. Environmental Protection Agency. (1994) User's guide to MOBILES (Mobile Source Emission Factor Model). Ann Arbor,
         MI:  Office  of Air  and  Radiation, Office  of Mobile Sources; report no.  EPA-AA-TEB-94-01. Available:
         www.epa.gov/oms/models/mobile5/mob5ug.pdf [2000, March 3].
U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research Triangle Park, NC: National
         Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
U.S. Environmental Protection Agency. (1998) National air quality and emissions trends report, 1997. Research Triangle Park,
         NC: Officeof Air Quality Planning and Standards; report no. EPA454/R-98-016. Available: www.epa.gov/oar/aqtrnd97/
         [2000, February 14].
Wallace, L. A. (1983) Carbon monoxide in air and breath of employees in an underground  office. J. Air Pollut. Control Assoc.
         33:678-682.
Williams, R.; Walsh, D.; White, J.; Jackson, M.; Mumford, J.  (1992) Effect on carbon monoxide levels in mobile homes using
         unvented kerosene heaters for residential heating. Indoor Environ. 1: 272-278.
Wilson, A. L.; Colome,  S. D.; Tian, Y. (1993) California residential indoor air quality study.  Volume 1:  methodology and
         descriptive statistics. Appendices. Chicago, IL: Gas Research Institute; report no. GRI-93/0224.2.
Wilson, A. L.; Colome,  S. D.; Tian, Y.; Becker,  E. W.; Baker, P. E.; Behrens, D. W.;  Billick, I.  H.; Garrison, C. A. (1996)
         California residential air exchange rates  and residence volumes.  J. Exposure Anal. Environ. Epidemiol. 6:311-326.
Yung, Y. L.; Shia, C.; Herman, R. L. (1999) Is the biomass burning source of CO decreasing? In: Khalil, M. A. K.; Pinto, J. P.;
         Shearer,  M. J.,  eds. Carbon monoxide  [special issue of papers from an international conference; December 1997;
         Portland, OR]. Chemosphere Global  Change Sci. 1: 83-90.
Yurganov, L. N; Grechko, E. L; Dzhola,  A. V. (1999) Zvenigorod carbon monoxide total column  time  series: 27 yr of
         measurements. In: Khalil, M. A. K.; Pinto, J. P.; Shearer, M. J., eds. Carbon monoxide [special issue of papers from an
         international conference; December 1997; Portland, OR]. Chemosphere Global Change Sci. 1: 127-136.
Zawacki, T. S.; Cole, J. T.; Huang, V.  M. S.; Banasiuk, H.; Macriss, R.  A. (1984) Efficiency and emissions  improvement of
         gas-fired space  heaters. Task 2. Unvented space heater emission reduction [final  report]. Chicago, IL: Gas Research
         Institute; report no. GRI-84/0021.
Zhang, Y.; Stedman, D. H.; Bishop, G. A.; Beaton, S. P.; Guenther, P. L.  (1996) On-road evaluation of inspection/maintenance
         effectiveness. Environ. Sci. Technol. 30: 1445-1450.
Ziskind, R. A.; Rogozen, M. B.; Carlin, T.; Drago, R. (1981) Carbon monoxide intrusion into  sustained-use vehicles. Environ. Int.
         5: 109-123.
                                                      3-63

-------
                                    APPENDIX 3A

                 Spatial Correlation  Coefficients
                          for Carbon Monoxide
       This appendix contains information about the spatial variability of carbon monoxide (CO) that was
not presented in Section 3.4.3. Spatial correlation coefficients have been calculated in ambient monitoring
studies to provide insight into the geographical distribution of sources of particulate matter (Suh et al,
1995) or into pollutants' spatio-temporal exposure characterization errors (Ito et al., 1998; Peters et al.,
1999).
       Two types of spatial correlations, the Pearson correlation coefficient and the Kendall tau statistic,
were used to evaluate the relationship of daily CO maximum 8-h averages among the different monitoring
sites in the  five geographically diverse urban areas of Denver, CO; Los Angeles, CA; New York,  NY;
Phoenix, AZ; and Fairbanks, AK  (Shadwick et al., 1997,  1998a,b,c).  Both types of nonparametric
correlation coefficients are intended to measure the "strength of relationship"; however, each does so in
different ways. The Pearson correlation coefficient stresses numerical levels, and the Kendall tau statistic
emphasizes differences between ranked sets of values.  The Kendall tau correlation coefficients are
presented in this section because they allow for a more simplistic interpretation of the resulting statistic than
do Pearson coefficients (Hollander and Wolfe, 1973). For example, in any two time series, a given pair of
(bivariate) observations can exhibit a simultaneous increase (or decrease) in ranks (i.e., a concordant pair)
or a change of ranks in opposing directions (i.e., a discordant pair). Kendall tau estimates the probability
that two pairs of observations are concordant. As the sample statistic approaches +1, a high probability of
concordance between two randomly selected pairs is implied, and as it approaches -1, a low probability
of concordance (high probability of discordance) is implied.
       Table  3A-1 shows  the Kendall tau correlation coefficients for the daily maximum 8-h  CO
concentration at the sites in the Denver Metropolitan Statistical Area (MSA).  The correlation is strongest
among the downtown sites, which also are located the closest together. A number of factors contribute to
the generally low values that are seen. These factors include the influence of local sources, differences in
meteorological conditions across the MSA, nonoverlapping averaging periods, and measurement error. The
correlation coefficients for the site pairs generally increased from the 1986/87 to 1995/96 winter seasons
(Table 3A-1). Littleton, CO, provides a good example of the increase in correlation with several of the
other CO sites in the Denver area. Littleton (1990 population of approximately 33,700 people) is a city
located  approximately 20 km south of downtown Denver.  Its correlation coefficient has shown large
increases, along with the Broadway and Julian Denver CO sites, as well as with the  Arvada and Greeley,
CO, sites.  These increased correlations were for the weekday data and were all of at least one order of
magnitude (Table 3A-1).  One order of magnitude increases also were seen for the weekend data between
Littleton and the  Arvada  and Greeley sites.  Arvada and Greeley are approximately 25 km and 95 km,
respectively, from Littleton. Therefore, it appears that the increase in correlation coefficients is related to
the population spreading out from downtown Denver, resulting in the surrounding suburbs becoming more
similar relative to traffic patterns and, hence, to the distribution of the ambient CO.
                                            3A-1

-------
     Table 3A-1. Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average
              Carbon Monoxide Data in the Denver Metropolitan Statistical Area

                                    1986/87 Winter Season

                                    Boulder-    Denver-   Denver-   Denver-
                 Welby   Littleton    Marine    Broadway   Albion    Julian   Arvada   Greeley
Welby

Littleton

Boulder-Marine

Denver-
Broadway

Denver-Albion

Denver-Julian

Arvada

Greeley
                                         WEEKEND
Welby

Littleton

Boulder-Marine

Denver-
Broadway

Denver-Albion

Denver-Julian

Arvada

Greeley
                                    1995/96 Winter Season
                                    Boulder-    Denver-   Denver-   Denver-
                  Welby   Littleton    Marine    Broadway   Albion    Julian   Arvada   Greeley
                                         WEEKEND
                                            3A-2

-------
       Table 3 A-2 shows the correlations for the daily maximum 8-h average CO concentration between
sites for the Los Angeles Consolidated Metropolitan Statistical Area (CMSA). The upper table shows the
correlations for the 1986/87 winter season, whereas the table at the bottom shows the correlations for the
1995/96 winter season. With some exceptions, the between-site correlations generally increased for each
site-pair in 1995/96 compared with those in 1986/87. This was analogous to the Denver site correlations.
The city-pair that had the highest correlation in the 1986/87 analysis remained unchanged for the 1995/96
analysis in 8 out of the 11 city-pairs in the weekday data and in 7 out of the  11 city-pairs in the weekend
data. This is indicative of the Los Angeles CMSA having several distinct air sheds.  The monitoring sites
that "reside" within a particular air shed likely will exhibit similar  characteristics in ambient CO. For
example, the Burbank, Hawthorne, Lynwood, and Reseda, CA, monitors show a history of violations of the
daily maximum 8-h average over the  last 5 years  of monitoring data. These sites are located in heavily
urbanized areas of Los Angeles County, with numerous major freeways nearby.  The spatial correlations
among these four sites are weak to moderate (range: 0.3 to 0.7 for the 1995/96 weekday data). Peripheral
to the area containing the urbanized monitors are the monitoring sites in La Habra, Long Beach, Anaheim,
and Riverside, CA.  These sites had violations before 1991, but have not had any violations of the daily
maximum 8-h average since then. The last group of sites, consisting of Asuza, Barstow, El Toro, and West
Los Angeles, CA did not have any violations over the 10-year period.
       A notable exception to  the general increase in the between-site correlations  was seen in the
correlations between Barstow and each of the other sites for the weekday data, where the correlations were
uniformly lower in 1995/96 compared to the correlations in 1986/87.  However, this site is outside the Los
Angeles Basin (lying across the San Bernardino Mountains from Los Angeles), and hence there is probably
very little relationship between ambient CO in Barstow and the other Los Angeles CO sites.
       Table 3A-3 shows the Kendall tau correlation coefficients calculated for the daily maximum 8-h
average CO concentration for the New York, NY, CMSA monitoring sites for the earliest year and the most
recent  year in these analyses.   The  correlation coefficients exhibit a wide range (from a correlation
coefficient of zero [i.e., no correlation at all] to a correlation coefficient of 0.8) of values indicating
heterogeneity in the 8-h maximum CO concentrations among the monitoring sites. Although the correlation
coefficients generally rose from the 1986/87 winter season to the 1995/96 winter season, the increase was
not as pronounced as those in either Denver or Los Angeles, and there were numerous exceptions where
the correlations decreased during the same time period.
       Care should be taken in attempting to draw any conclusions based on correlation coefficients
involving any two sites. For example, the site that has the highest correlation coefficient (0.622) for the
weekday data with the Flatbush Avenue  site (the site in Brooklyn, NY, that  had some of the highest
measured CO values in the New York  CMSA analysis) for the  1986/87 winter season was located in
Morristown, NJ, a suburban, mostly residential community with seemingly little in common with Brooklyn.
Interestingly, Morristown also had the highest correlation (0.615) with Brooklyn for the weekend data.
Then, for weekdays in the 1995/96  winter season, Brooklyn had  the highest correlation (0.60) with
Elizabeth, NJ, and, for weekends, Brooklyn again had the highest correlation (0.54) with the Elizabeth site.
The population density in downtown Brooklyn is more like that of Elizabeth than Morristown; however,
they are still quite far apart.  Apparently, characteristics shared by the New Jersey sites with the Flatbush
Avenue site, other than proximity, are influencing the correlations between them.
       Table 3 A-4 contains the Kendall tau correlation coefficients for the Phoenix MSA monitoring sites
for the earliest year and the  most recent year in these analyses. For the 10 pairs of sites for the weekday
data, six pairs had correlations that rose between the 1986/87 winter season and the 1995/96 winter season,
and four pairs remained about the same.  For the 10 pairs of sites for the weekend data, only three pairs of
correlations increased, five pairs decreased, and two pairs remained about the same. No discernable pattern
is evident in the matrix of correlations in Table 3 A-4. Indications are that the Phoenix air shed is more well
mixed than are the others investigated here.

                                              3A-3

-------
 Table 3A-2. Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average Carbon Monoxide Data in the Los Angeles
	Consolidated Metropolitan Statistical Area	
                                                  1986/87 Winter Season
                     West                                Long                           El
            Azusa    LA     Burbank   Reseda  Lynwood   Beach   Hawthorne  Anaheim    Toro
 La
Habra   Riverside  Barstow
                                                      WEEKDAY

                                                0.23      0.28       0.27

                                                0.48      0.49       0.51

                                                0.52      0.55

-------
>
en
            Table 3A-2 (cont'd). Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average Carbon Monoxide Data in the
           	Los Angeles Consolidated Metropolitan Statistical Area	

                                                            1995/96 Winter Season
                     Azusa
West
 LA
Burbank   Reseda   Lynwood
Long
Beach
Hawthorne   Anaheim
 El
Toro
 La
Habra
Riverside   Barstow

-------
(35
          Table 3A-3.  Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average Carbon Monoxide Data in the New York City
                                                       Consolidated Metropolitan Statistical Area
                                                                   1986/87 Winter Season
Bridgeport

Stamford

Fort Lee

Hackensack

Newark

Jersey City

Perth Amboy

Freehold

Morristown

Elizabeth

Flatbush

7th Avenue

Manhattan
                                                                        Jersey    Perth
                      Bridgeport  Stamford   Fort Lee   Hackensack   Newark   City    Amboy   Freehold   Morristown
                                                                                                                         7th
                                                                                                    Elizabeth   Flatbush   Avenue   Manhattan
                                                                        WEEKDAY
                                                                        WEEKEND

-------
>
-^1
             Table 3A-3 (cont'd). Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average Carbon Monoxide Data in the

                                             New York City Consolidated Metropolitan Statistical Area

                                                                 1995/96 Winter Season


                                                                    Jersey    Perth                                            7th
                     Bridgeport   Stamford   Fort Lee   Hackensack   Newark    City   Amboy   Freehold   Morristown   Elizabeth   Flatbush  Avenue  Manhattan


                                                                     WEEKDAY
                                                                     WEEKEND

-------
      Table 3A-4. Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average
     	Carbon Monoxide Data in the Phoenix Metropolitan Statistical Area	
                                     1986/87 Winter Season
                   South Phoenix   West Phoenix   East Butler Drive
Central
Phoenix
 South Phoenix

 West Phoenix

 East Butler Drive

 Central Phoenix

 North Miller Road
North Miller
   Road
                                     1995/96 Winter Season
South Phoenix
West Phoenix
East Butler Drive
Central
Phoenix
North Miller
Road
 South Phoenix

 West Phoenix

 East Butler Drive

 Central Phoenix

 North Miller Road
       There were large variations in the Kendall tau correlation coefficients calculated for the daily
maximum 8-h average CO concentrations between individual sites in a given MSA during the winters of
1986/87 and 1995/96, suggesting a high degree of heterogeneity in the daily maximum 8-h average ambient
CO levels in the MS As that were characterized, likely indicating that daily maximum 8-h average CO levels
observed at particular monitoring sites may not be related to CO levels occurring some distance away from
the monitoring site.  Further analyses will determine whether the correlation coefficients will increase
significantly with an increase in averaging time to 24 h.
       Table 3A-5 shows Kendall spatial tau correlation coefficients for the daily maximum 8-h CO
concentration at  the three  sites in the Fairbanks, AK, MSA.  As can be seen from the table, there are
uniformly high correlation coefficients among each of the site pairs with much smaller variation among
them than among those given in Tables 3A-1 through 3A-4. There is very little change in values over the
10-year period of interest.  In addition, for each winter  season in  Table  3A-5, there is no substantial
difference between the weekday correlations and those for the weekend. No other discernable pattern or
trend is revealed by the correlation analysis.
                                             3A-8

-------
       Table 3A-5. Kendall Tau Spatial Correlations for the Daily Maximum 8-Hour Average
      	Carbon Monoxide Data in the Fairbanks Metropolitan Statistical Area	

                                         1986/87 Winter Season
                            Federal Building
State Building
Hunter Elementary
                                              WEEKDAY
 Federal Building

 State Building

 Hunter Elementary
                                               WEEKEND
                                          1995/96 Winter Season
                             Federal Building
 State Building
 Hunter Elementary
                                               WEEKDAY
 Federal Building

 State Building

 Hunter Elementary
                                               WEEKEND
        Based on the above results, caution should be exercised in using the daily maximum 8-h average
data to characterize the exposure of the general population within a given MSA (e.g., for studies relating
health outcomes to ambient CO levels or in health studies in which CO may be viewed as a confounding
variable).
References
Hollander, M.; Wolfe, D. A. (1973) Nonparametric statistical methods. New York: John Wiley & Sons.
Ito, K.; Thurston, G. D.; Nadas, A.; Lippmann, M. (1998) Multiple pollutants' spatio-temporal exposure characterization errors.
        In: Measurement of toxic and related air pollutants: proceedings of a specialty conference, volume I; September; Gary,
        NC;pp. 53-63. (A&WMA publication VIP-85).
Peters, A.; Kotesovec, F.; Skorkovsky, J.; Brynda, J.; Heinrich, J. (1999) Akute Auswirkung der Schwebstaubkonzentrationen in
        der AuBenluft auf die Mortalitat - Vergleichsstudie Nordost-Bayern / Nordbohmen. AbschluBbericht [Acute effects of
        suspended particle concentrations in the atmosphere on mortality - a study comparing northeast Bavaria and north
        Bohemia.  Final report]. Bavaria, Federal Republic of Germany: Institut fur Epidemiologie; report no. GSF-EP S 1/99.
Shadwick, D.; Glen, G.; Lakkadi, Y.;  Lansari, A.; del Valle-Torres, M. (1997) Analysis of carbon monoxide for the Denver,
        Colorado  MSA. Research Triangle Park, NC:  U.S.  Environmental Protection Agency, Office  of Research and
        Development, National Exposure Research Laboratory; contract no. 68-D5-0049; December.
Shadwick, D.; Glen, G.; King, J.; Chen, X. (1998a) Analysis of carbon monoxide for the Los Angeles, California CMS A. Research
        Triangle Park, NC: U.S. Environmental Protection Agency, Office of Research and Development, National Exposure
        Research Laboratory; contract no. 68-D5-0049; June.
Shadwick, D.; Glen, G.; King, J.; Chen, X. (1998b) Analysis of carbon monoxide for the New York, New York CMSA. Research
        Triangle Park, NC: U.S. Environmental Protection Agency, Office of Research and Development, National Exposure
        Research Laboratory; contract no. 68-D5-0049; June.
                                                  3A-9

-------
Shadwick, D.; Glen, G.; King, I; Chen, X. (1998c) Analysis of carbon monoxide for the Phoenix, Arizona MSA. Research
        Triangle Park, NC: U.S. Environmental Protection Agency, Office of Research and Development, National Exposure
        Research Laboratory; contract no. 68-D5-0049; June.
Suh, H. H.; Allen, G. A.; Koutrakis, P.; Burton, R. M. (1995) Spatial variation in acidic sulfate and ammonia concentrations within
        metropolitan Philadelphia. J. Air Waste Manage. Assoc. 45: 442-452.
                                                     3A-10

-------
                                    CHAPTER 4
       Population  Exposure to Carbon Monoxide
4.1  Introduction
       National Ambient Air Quality Standards (NAAQS) have been set to protect public health and
welfare. The NAAQS for carbon monoxide (CO), which are not to be exceeded more than once per year,
are 9 ppm for an 8-h average and 35 ppm for a 1-h average. These standards include a margin of safety to
protect the population from adverse effects of CO exposure. Accordingly, this chapter reviews studies of
population exposure to CO concentrations from different sources and explains why CO exposure studies
are necessary and how they are done. It also discusses how population exposures are estimated, describes
typical levels and durations of CO exposure in various microenvironments, and examines how CO
exposures have changed over time in the United States.
       Because Americans spend substantial amounts of time indoors, it is important to determine the total
population exposure to CO  from both indoor and outdoor  CO  sources.  In this chapter, "outdoor"
concentrations are those measured in air that immediately surrounds an indoor microenvironment. Because
of air exchange, outdoor CO concentrations have a direct effect on CO concentrations measured indoors.
"Ambient" concentrations are those measured at fixed-site, air quality monitoring stations that are used to
determine compliance with the NAAQS.
       As discussed in Chapter 3, one cannot assume that "outdoor" and "ambient" concentrations are
always the same because sometimes outdoor sources (e.g., highway emissions) are downwind of an ambient
monitoring station." However, one generally can assume that indoor and "outdoor" CO concentrations are
approximately the same, except for situations when CO  is emitted by indoor sources  (e.g., unvented
combustion appliances inside a home)  or when CO emissions from an immediate outdoor source directly
contaminate indoor microenvironments (e.g., when a vehicle's undiluted exhaust infiltrates the passenger
cabin of that vehicle or a following vehicle).
       After inhalation, CO binds with hemoglobin (Hb) in the blood to form carboxyhemoglobin (COHb).
Besides endogenous CO production developed within the body from Hb catabolism, everyone is exposed
to a global background level of CO in the ambient air on the order of 0.1 ppm (see Section 3.2).  These
combined sources constitute a reference or baseline exposure  as reflected in an endogenous COHb level
on the order of 0.5% that varies individually, based on physiological differences. These differences reflect
variation in basal metabolisms and other metabolic sources, as discussed in detail in Sections 5.3 and 5.4.
This chapter discusses the exposure of nonsmokers to CO.  Smokers are excluded, because they represent
a source of CO, and because of their higher baseline levels of COHb and adaptive response to elevated
COHb.
       The study of population exposure is multidisciplinary, and the definition of personal exposure has
evolved over time (Ott, 1982;Duan, 1982;Lioy, 1990; U.S. Environmental Protection Agency, 1992a;Last
et al, 1995; Zartarian et al., 1997).  A recent definition offered by Zartarian et al. (1997) states that
exposure is the contact between an agent and a target at a specified contact boundary, defined as a surface
in space containing at least one exposure point (a point at which contact occurs). Using this definition, this
chapter assumes that an inhaled CO molecule (the agent) reaches a human (the target) at the lining of the
lung (the contact boundary) where CO exchange takes place between air and blood. In actual field studies,


                                           4-1

-------
the air in the immediate vicinity of the target often is assumed to be well mixed, such that a measured CO
concentration in the air can be assumed to represent a nonsmoking person's  actual  exposure from CO
inhalation.
       This chapter is concerned only with CO exposures that occur at concentrations capable of increasing
COHb levels above  a  reference  baseline level.  Besides exposure to CO  concentrations  above the
background level, human COHb levels can be elevated because of metabolic degradation of many drugs,
solvents (e.g., methylene chloride), and other compounds to CO (for details see Section 5.3).  Because the
endogenous production of CO from drugs and solvents may continue for several hours, it can prolong any
cardiovascular stress from COHb.  Moreover, the maximum COHb level from endogenous CO production
can last up to twice as long as comparable COHb levels caused by comparable exposures to exogenous CO
(Wilcosky and Simonsen, 1991; Agency for Toxic Substances and Disease Registry, 1993).  Hence, the
literature on exposure to methylene chloride also is discussed in this chapter.
       Descriptive studies of exposure typically report average and peak concentrations to which people
are exposed, the temporal aspects of exposure (i.e.,  averaging times), where exposures occurred (i.e.,
outdoor and indoor microenvironments), and the sources of microenvironmental exposures (e.g., motor
vehicles, combustion appliances). Explanatory studies of microenvironmental concentrations when sources
are operating try to identify factors that affect or contribute to exposure, because that information may
enable mitigation of high-level exposures.  Unlike epidemiologic or clinical studies of health effects,
exposure studies rarely identify the health outcomes (e.g., headache, dizziness, nausea, etc.) associated with
measured exposures.
       For completeness, this chapter also briefly discusses high-level nonambient CO exposures that can
lead to CO poisoning and death (e.g., accidental exposure to undiluted motor vehicle exhaust). Although
such effects are not health outcomes presently used in setting the NAAQS for CO, they may be affected by
reduced emissions from sources regulated by the U. S. Environmental Protection Agency (EPA) to meet the
CO NAAQS.   Chapter 6 discusses  relevant  health effects  resulting from exposure to ambient  CO
concentrations.
       This chapter  is  organized as follows:  the  first section summarizes the state of knowledge on
population exposure to CO as of 1991, when EPA published the previous CO air quality criteria document.
Next follows a discussion of more recently published studies of population exposure to all sources of CO
(except the active inhalation of tobacco smoke), which delineates typical and peak levels of exposure as
people engage in daily  activities, including those related to an occupation. Factors affecting trends in
population exposure are then described, including factors such as public policies affecting motor vehicle
emissions, travel behavior, and societal changes in human  activity patterns, particularly those related to
motor vehicles (which are amajor source of CO emissions, as shown in Chapter 3).  The conclusion section
summarizes findings of this assessment and discusses their implications for CO exposure models, such as
the probabilistic NAAQS Exposure Model (NEM) used for evaluating CO exposures under different air
quality indicators for the NAAQS.  The conclusion also examines the extent to  which  CO exposures have
changed since the previous criteria document.
4.2  Brief Summary of Population Exposure Studies Prior to 1991
       This section briefly reviews key population exposure studies that were completed by 1990.
It identifies populations sensitive to CO exposure and discusses studies of population exposure based on
fixed-site and personal monitors, as well as relevant population exposure models.  This section does not
discuss many pre-1990 exposure studies for two reasons: (1) the pre-1990 studies are reviewed in the
previous criteria document, and the primary purpose of the present criteria document is to focus on more
recent studies; and (2) the results of older studies no longer may  be indicative of current population
exposures, given the major reductions in CO emissions per motor vehicle, social  changes  affecting

                                             4-2

-------
commuting patterns and growth in vehicle miles of travel, and other factors. Factors affecting changes in
population exposure are discussed in Section 4.4.

4.2.1  Sensitive Populations
       The NAAQS are intended to protect the general public, including probable high-risk groups of the
general population. These groups differ from one air pollutant to another. In the case of CO, these groups
include the elderly; pregnant women; fetuses; young infants; and those suffering from anemia or certain
other blood, cardiovascular, or respiratory diseases. People thought to be at greatest risk from exposure to
ambient CO levels are those with ischemic heart disease who have stable exercise-induced angina pectoris
(cardiac chest pain). Individuals with this disease represented about 3% of the U.S. population in 1994.
Studies show that earlier time to onset of cardiac chest pain occurred in these people while they exercised
during exposures to CO concentrations that produced levels of COHb in the bloodstream in the range of
2 to 3% (U.S. Environmental Protection Agency, 1991). The National Health and Nutrition Examination
Survey (NHANES) II study reported that 6.4% of the U S. population who never smoked had COHb levels
above 2.1%, based on a national random sample of people (n = 3,141) ranging in age from 12 to 74 years
(Radford and Drizd,  1982).  The NHANES H study was done in the late  1970s, when ambient CO
concentrations were much higher (see Figure 3-4).

4.2.2  Estimates of Population Exposure Based on  Fixed-Site Monitors
       In the United States, NAAQS attainment is based on ambient air quality measurements recorded
at a nationwide network of fixed-site monitors.  Based on this network, EPA's Office of Air Quality
Planning and Standards estimated that 9.1 million people lived in three counties where ambient CO levels
exceeded the NAAQS in 1997 and, thus, were at increased risk of exposure to CO levels above the NAAQS
(U.S. Environmental Protection Agency, 1998).  The estimate was  made by combining census data on
county populations with data on violations of the CO NAAQS recorded by stationary monitors. Previous
studies have shown why such estimates should not be interpreted as assessments of population exposure
to CO, for the two reasons discussed below:
(1)  Ambient CO concentrations are not spatially homogeneous within the area monitored.  For example,
    Ott and Eliassen (1973) reported average CO levels ranging from 5.2 to 14.2 ppm for sidewalks along
    congested streets of downtown San Jose, CA. Corresponding CO averages at fixed-site monitors were
    only 2.4 to 6.2 ppm.  A decade later,  Ott and Flachsbart (1982) found a narrower gap between
    simultaneous CO measurements from fixed-site and personal exposure monitors deployed at indoor
    and outdoor commercial settings in five California cities.
(2)  In the absence of indoor CO sources and immediate outdoor sources (e.g., idling motor vehicles),
    indoor CO concentrations tend  to equal outdoor concentrations over 24 h.  In  buildings with
    mechanical ventilation systems, the timing and scheduling of outdoor "make-up" air into the building
    affects ratios of indoor-outdoor concentrations both in the short and long term (Yocom, 1982).  For
    example, when make-up air was introduced into an air-conditioned building during morning rush hours
    (when  outdoor CO levels were high), indoor CO concentrations exceeded outdoor levels for the
    remainder of the day (Yocom et al, 1971). In the presence of incompletely vented indoor combustion
    sources such as gas stoves used for supplementary space heating, kerosene heaters, etc., indoor CO
    concentrations often exceed the outdoor levels (U.S. Environmental Protection Agency, 1991). In a
    1985 Texas study of a low-socioeconomic population, CO concentrations were greater than or equal
    to  9 ppm in 12% of surveyed homes.  Residential CO concentrations were high where unvented gas
    space heaters were used as the primary heat source (Koontz and Nagda, 1988).
These facts take on added significance given that many Americans spend most of their time indoors (Szalai,
1972; Chapin, 1974; Meyer, 1983; Johnson, 1989; Schwab et al., 1990).  Hence, studies of actual personal
                                             4-3

-------
exposure to CO are preferred over crude estimates of population exposure to ambient CO in determining
what risk CO poses to public health from a total exposure perspective (Sexton and Ryan, 1988).

4.2.3  Surveys of Population Exposure Using Personal Monitors
       With the development of personal exposure monitors (PEMs) in the 1970s, researchers began to
measure  either the total human  exposure of a population or the exposures  of subpopulations in
microenvironments that posed higher risks of CO exposure, such as inside a motor vehicle moving slowly
in congested traffic.  In theory, a microenvironment exists if the CO concentration at a particular location
and time is sufficiently homogeneous yet significantly different from the concentrations at other locations
(Duan, 1982).
       Human exposure studies of target populations typically use either a direct or an indirect approach.
In the direct approach,  PEMs are distributed either to a representative  or "convenience" (nonrandom)
sample of a human population.  Population exposure parameters cannot be estimated from a convenience
sample, because it does not represent the population from which it was drawn. Using PEMs, people record
exposures to selected air pollutants as they engage in their regular daily activities. In the indirect approach,
trained technicians use PEMs  to  measure pollutant concentrations in  specific  microenvironments or
populations. This information then must be combined with additional data on human activity patterns to
estimate the time spent in those microenvironments (Duan, 1982; Sexton and Ryan, 1988).  For further
discussion of these topics, see  Section 8.2 of the previous CO criteria  document (U.S. Environmental
Protection Agency, 1991) and Mage (1991).
       Sexton and Ryan (1988) discuss types of personal monitors and research methods used by the direct
and indirect approaches. Although small passive monitors may be placed near a person's oral/nasal cavity
where exposure actually occurs, larger monitors must be carried by a person or placed nearby.  Using data
from PEMs, one can construct exposure-time profiles for a particular activity,  such as commuting, or the
integrated exposure between two points in time. From this information, one  can determine the average
concentration to which a person has been exposed for a given time period. Based on the superposition
principle, one also can determine a net microenvironmental concentration by subtracting the outdoor
concentration, as measured by an appropriate fixed-site monitor, from a microenvironmental concentration
measured by a personal monitor (Duan, 1982).  Because  ambient CO  concentrations  are not spatially
homogeneous at any given moment, the net microenvironmental concentration can be either positive or
negative in value.  A negative  net value can occur, for example, in  homes with no CO sources during
morning periods when ambient CO concentrations from rising traffic emissions on highways have not yet
diffused into residential areas. A negative net value simply indicates that the indoor microenvironment has
a lower positive CO concentration than the outdoor environment at a given moment.
       In an early pilot study in Los Angeles, using the direct approach, subjects recorded their exposures
and corresponding activities in diaries (Ziskind et al., 1982).  Because this was cumbersome and potentially
distorted the activity, later studies used data loggers to store concentrations electronically, as done by major
studies of the urban populations of Denver, CO, and Washington, DC (Akland  et al., 1985).  In these
studies, subjects still used  diaries to record pertinent information about their activities in specified
microenvironments while monitoring personal exposures.  Data then were transferred electronically from
data loggers and manually from diaries to computer files for analysis.
       The direct approach, which uses the total exposure assessment methodology, provides a frequency
distribution of air pollutant concentrations for a sample of people selected randomly from either a general
or specific population (defined by demographic, occupational, and health  risk factors) for a particular time
period of interest (e.g.,  a day).  Studies using the  direct  approach enable researchers to assess  what
percentage of a large population is exposed to pollutant concentrations in excess of ambient air quality
standards (Akland et al., 1985).  Studies using the indirect approach may focus on situations that bring large
numbers  of people in contact with high concentrations in specific microenvironments.  For example,

                                              4-4

-------
FlachsbartandBrown(1989) determined what percentage of employees were exposed to CO concentrations
in excess of national and state ambient air quality standards at a large shopping center attached to a parking
garage in Honolulu, HI.
      Direct studies of general populations are rare because of their expense and the logistical problems
of monitor distribution. Two examples for CO were those done in Denver and Washington, DC, during the
winter of 1982 and 1983 (Akland  et  al, 1985).   In both studies, the target population included
noninstitutionalized, nonsmoking residents, 18 to 70 years of age, who lived in the city's metropolitan area,
an estimated 1.2 million adults in Washington and 500,000 in Denver. In both cities, the composite network
of fixed-site monitors overestimated the  8-h exposures of people with low-level personal exposures and
underestimated the 8-h exposures of people with high-level personal exposures.  With respect to the
underestimates, over 10% of the daily maximum 8-h  personal exposures in Denver exceeded the NAAQS
of 9 ppm, and about 4% did so in Washington. The end-expired breath CO levels were in excess of 10 ppm,
which is roughly equivalent to 2% COHb in about 12.5% of the Denver participants and about 10% of the
Washington participants (after correcting the measured breath concentrations for the influence of room air
CO concentrations). Simultaneous CO measurements at fixed-site monitors exceeded 9 ppm only 3% of
the time in Denver and never exceeded 9 ppm in Washington (Akland et al., 1985).
      The Denver and Washington studies identified certain  activities associated with higher CO
exposures. The two highest average CO concentrations occurred when subjects were (1) inside a parking
garage and (2) traveling by car. Those who commuted 6 h or more per week had higher average exposures
than those who commuted fewer hours per week.  Table 4-1 shows that higher mean CO concentrations
occurred for travel by motor vehicle (motorcycle, bus, car, and truck) than that by walking or bicycle, and
that high indoor concentrations (above the 8-h NAAQS of 9 ppm) occurred in public garages, service
stations, or motor vehicle repair facilities.   Denver had higher  average CO concentrations than  did
Washington for all microenvironments because of Denver's higher altitude and colder winter climate (Ott
etal., 1992a).
4.3  Population Exposure Models
       Many studies developed computer models to predict exposure  in both general and special
populations (U.S. Environmental Protection Agency, 1991). These models are important because it is
impossible and logistically impractical to measure the hourly and daily exposure of every person in a
population on a real-time basis.   Models of human exposure are  empirically derived  mathematical
relationships, theoretical algorithms, or hybrids of these two. To support policy decisions related to the
setting of ambient and emission standards, EPA supported development of two general population exposure
models: (1) the NEM and (2) the Simulation of Human Activity and Pollutant Exposure (SHAPE) model.
These models assume that an individual's total CO exposure over a specified time interval can be estimated
as the sum of the average concentration within a microenvironment, multiplied by the amount of time spent
in that microenvironment (Duan, 1982).
       The SHAPE model used a stochastic approach to simulate the exposure of an individual over a 24-h
period (Ott, 1984).  The model replicates a person's daily activity pattern by sampling from probability
distributions  representing the chance of entry,  time  of entry,  and  time  spent in  22  different
microenvironments.  Transition probabilities determine a person's movement from one microenvironment
to another. The model assumes that microenvironmental concentrations  reflect the contribution of an
ambient concentration and a component representing CO sources within each microenvironment. Because
SHAPE relies on field surveys of representative populations, the data requirements of the model are fairly
extensive.
       The SHAPE model can estimate the frequency distribution of maximum standardized exposures
to CO for an urban population and the cumulative frequency distribution of maximum exposures for both

                                             4-5

-------
       Table 4-1.  Carbon Monoxide Concentrations in Selected Microenvironments of
                                 Denver, CO, 1982 and 1983
                   (listed in descending order of mean CO concentration)
Microenvironment
In-Transit
Motorcycle
Bus
Car
Truck
Walking
Bicycling
Outdoor
Public garages
Residential garages or carports
Service stations or vehicle repair facilities
Parking lots
Other locations
School grounds
Residential grounds
Sports arenas, amphitheaters
Parks, golf courses
Indoor
Public garages
Service stations or vehicle repair facilities
Other locations
Other repair shops
Shopping malls
Residential garages
Restaurants
Offices
Auditoriums, sports arenas, concert halls
Stores
Health care facilities
Other public buildings
Manufacturing facilities
Homes
Schools
Churches
Number of
Observations3

22
76
3,632
405
619
9

29
22
12
61
126
16
74
29
21

116
125
427
55
58
66
524
2,287
100
734
351
115
42
21,543
426
179
Mean
(ppm)

9.79
8.52
8.10
7.03
3.88
1.34

8.20
7.53
3.68
3.45
3.17
1.99
1.36
0.97
0.69

13.46
9.17
7.40
5.64
4.90
4.35
3.71
3.59
3.37
3.23
2.22
2.15
2.04
2.04
1.64
1.56
Standard Error
(ppm)

1.74
0.81
0.16
0.49
0.27
1.20

0.99
1.90
1.10
0.54
0.49
0.85
0.26
0.52
0.24

1.68
0.83
0.87
1.03
0.85
0.87
0.19
0.002
0.48
0.21
0.23
0.30
0.39
0.02
0.13
0.25
aAn observation was recorded whenever a person changed a microenvironment and on every hour; thus, each
 observation had an averaging time of 60 min or less.

Sources: Johnson (1984) and Akland et al. (1985), as reported in U.S. Environmental Protection Agency (1991).
                                             4-6

-------
1-h and 8-h periods, thereby allowing estimates of the proportion of the population that is exposed to CO
concentrations above the NAAQS. An evaluation of SHAPE by Ott et al. (1988), using survey data from
the aforementioned Denver study, showed that the observed and predicted arithmetic means of the 1 -h and
8-h maximum average CO exposures were in close agreement; however, SHAPE overpredicted low-level
exposures and underpredicted high-level exposures.
       Unlike SHAPE, which uses diary data from each person in a population, the NEM model aggregates
people into cohorts. The NEM model has evolved over time from deterministic to probabilistic versions.
As described elsewhere (Johnson and Paul,  1983; Paul and Johnson,  1985), the deterministic version of
NEM simulates movements of selected groups (cohorts) of an urban population through a set of exposure
districts or neighborhoods and through different microenvironments. Cohorts are identified by district of
residence and, if applicable, district of employment, as well as by age-occupation group and activity pattern
subgroup. The NEM uses empirical adjustment factors for indoor and in-transit microenvironments, and
accumulates exposure over 1 year. Although the deterministic NEM was able to estimate central tendencies
in total exposure accurately, it did less well estimating the associated uncertainty caused by variation in time
spent in various microenvironments (Quackenboss et  al.,  1986) or variation in microenvironmental
concentrations (Akland et al., 1985). Paul et al. (1988) discussed advancements in the deterministic version
of NEM.
       In recent years,  EPA  developed  the
probabilistic  NEM  for CO  (pNEM/CO);  see
Johnson et al. (1992) for a description of the
assumptions and algorithms of pNEM/CO, as
those details are beyond the  scope  of this
chapter.   Figure 4-1  shows the  conceptual
overview of the logic and  data  flow of the
pNEM/CO model. It shows how any alternative
CO NAAQS can be evaluated by establishing
the distributions of personal exposures to CO
when that alternative CO  standard is met.
       McCurdy (1995) examined the history of
both the NEM and pNEM models and the role
they have played  in reviews  of criteria  air
pollutants  such as CO.    The EPA used
pNEM/CO, rather than the SHAPE model, in its
previous  review of the  CO NAAQS (U.S.
Environmental Protection Agency, 1992b).  At
the request of the Clean Air Scientific Advisory
Committee,   as  part of that  review, EPA
performed  a  limited  evaluation  of  the
predictions of pNEM/CO against observed data for subjects  of the Denver study (Johnson et al., 1992).
That evaluation concluded that there was generally good agreement between the distributions of observed
and predicted 1 -h daily maximum exposures, but that the model tended to underpredict the highest 8-h daily
maximum exposures (i.e., >12 ppm) and overpredict the lowest 8-h daily maximum exposures (i.e.,
<5  ppm).  Extending the earlier evaluation  of the pNEM/CO model, Law  et al. (1997) performed
20 simulated runs,  the average values of which were used for evaluation purposes.  Based on  this
evaluation, Law et al. (1997) reported the predicted and observed population exposure cumulative
frequency distributions (CFD), with and without gas stove use.  Figures 4-2 and 4-3 show that, regardless
of gas stove use, pNEM/CO overpredicted the CFD at low exposures and underpredicted the CFD at high
exposures for 8-h daily maximum exposures.  The proportion of the Denver population exposed at or above
                Seasonal
               Considerations
               (Temperature)
Human Activity
 and Exertion
  Patterns
 Emission Rates and
 Use Patterns for
 Indoor Sources
(e.g., gas appliances,
 passive smoking)
    Exposure Algorithms
     Ambient
     Fixed-Site
   Concentrations
        Distribution of People and
        Occurrences of Exposures
        Linked with Breathing Rate


Air Quality
Specification
Figure 4-1. Conceptual overview of pNEM. Model inputs (e.g.,
activity patterns, ambient monitoring data, air exchange rates)
are in round-cornered boxes, and model calculations are
shown in rectangles.
Source: Johnson et al. (1999).
                                              4-7

-------
                          Observed Exposure (n = 591)
                          Simulated Exposure
                                         99.99 99 999
               Cumulative Probability (percent)
 Figure  4-2.    Observed  versus simulated  8-h  daily
 maximum exposure for persons residing in homes with
 gas stoves in Denver, CO.
 Source:  Lawet al. (1997).	
                         Observed Exposure (n = 188)
                         Simulated Exposure
                                                                                      99.9  99.99 9999!
              Cumulative Probability (percent)
Figure  4-3.   Observed  versus simulated 8-h daily
maximum exposure for persons residing in homes without
gas stoves in Denver.
Source: Lawet al. (1997).
9 ppm (the 8-h NAAQS) was 12.7% (observed) versus 9.9% (highest predicted) for those with gas stoves
and 13.3% (observed) versus 8.8% (highest predicted) for those without gas stoves.  Similar results were
reported for the  1-h daily maximum exposures, with and without gas stove use.  The proportion of the
Denver population exposed at or above 35 ppm (the 1-h NAAQS) was 3.2% (observed) versus  1.2%
(highest predicted) for those with gas stoves and 2.1% (observed) versus 1.1% (highest predicted) for those
without gas stoves. Relatively close agreement between simulated and observed PEM data occurred for
CO concentrations near the average exposure, within the range of 6 to 13 ppm for the 1-h case and within
5.5 to 7.0 ppm for the 8-h case.
       Law et al. (1997) proposed four factors that may explain why pNEM/CO underpredicted the
observed CFD of people with high-level exposures (>90th percentile).
(1) The pNEM/CO modeled only two indoor CO sources (passive smoking and gas stoves) and omitted
    other home combustion sources (e.g., attached garages, fireplaces, kerosene heaters, woodstoves, etc.)
    that may have impacted  observed levels using personal monitors.
(2) Although people with high-level exposures were observed for two consecutive  days, the model
    randomly sampled each  day separately, thereby diluting the chances of sampling that person on both
    days.
(3) The model used activity pattern data from three cities (Cincinnati, OH; Denver; and Washington) to
    predict exposures  of Denver residents.   In reality, activity patterns of Cincinnati or Washington
    residents may not reflect those residing in Denver.
(4) The model used a constant vehicular air-exchange rate and a constant secondary-smoke CO value that
    are known to vary in reality.
       The EPA's Office of Air Quality Planning and Standards is in the process of revising the pNEM/CO
model.  Given the Law et al. (1997) study and EPA's  mandate under the Clean Air Act (CAA) to protect
public health, research is needed to improve the abilities of both SHAPE and pNEM to predict high-end
exposures.  One possible improvement is to consider addressing autocorrelation of inputs for time and
concentration, particularly inputs of possibly nonindependent microenvironments. For example, the
                                               4-8

-------
commuting time and CO exposure of someone going from home to work may not be independent of their
commuting time and exposure from work to home.  Also, both SHAPE and pNEM may need to consider
CO emission sources, microenvironments, and activity patterns excluded from previous versions of the
models to improve characterization of high-end exposures.  Model refinements should recognize that high-
end exposures may come from either regulated or unregulated CO emissions under the CAA.
4.4  Survey of Recent Exposure Studies of Nonsmokers
       This section focuses primarily on population exposure studies published or accepted for publication
in the scientific, peer-reviewed literature since the previous CO criteria document was published in 1991,
and includes some earlier studies that were omitted.  It compares results from both old and new studies
wherever possible.  However, the scope is limited to the  exposure of nonsmokers.   Studies involving
smokers were excluded because it is impossible to determine whether CO concentrations inhaled by
smokers originated from smoking or other sources (e.g., motor vehicles, gas appliances, etc.)." As such,
smokers have higher baseline levels of COHb than nonsmokers and an adaptive response to elevated COHb.
       The section is subdivided into three parts. The first part discusses nonoccupational exposure studies
because the NAAQS are intended to protect the general public. The second part reviews occupational
studies because urban population exposure models (e.g., pNEM/CO) must account for that portion of total
exposure that occurs in occupational settings.  The third part describes recent activity pattern studies
because they focus on the important temporal component of exposure.

4.4.1  Nonoccupational Exposures
       This subsection focuses on nonoccupational CO exposures that occur because of a variety of human
activities that require contact with sources of CO emissions, such as motor vehicles and fuel-burning tools
and appliances.  Section 3.5 discusses studies where CO has been measured by area monitors in  indoor
microenvironments because such measurements constitute an indirect exposure estimate. In addition, this
section discusses studies of breath CO in populations and studies of exposure to methyl ene chloride because
it can indirectly increase COHb levels.  Also discussed are studies of the passive exposure of nonsmokers
to CO concentrations in cigarette and cigar smoke because the pNEM/CO model accounts for this type of
exposure.

4.4.1.1  Exposure to Carbon Monoxide from Motor Vehicles
       This subsection presents nonoccupational studies of exposure to CO concentrations from motor
vehicles. Because these studies date to the mid-1960s, many of them were reviewed in the previous CO
criteria document and, therefore, are not reviewed again here. Subsection 4.4.2.2 examines the effects of
progressively tighter CO emission standards on passenger cabin exposure; this subsection, on the other
hand, focuses on other factors that affect the CO exposure of motorists and bicyclists.
       Several studies done prior to 1991 reported passenger exposure to engine or tailpipe emissions of
CO (Amiro, 1969; Clements,  1978; Ziskind et al., 1981).  More recently, Hampson and Norkool  (1992)
reported that 20 children were treated for accidental CO poisoning after riding in the back of eight closed
pickup trucks in Seattle, WA, between 1986 and 1991.  Forty-eight children riding with them in the backs
of these same pickup trucks did not require treatment.  Seventeen of the 20 children rode under  a rigid
closed canopy attached to the bed of the truck,  and the other three rode beneath  a tarpulin.  Six pickup
trucks had known exhaust system leaks; three had rear-end tailpipes, and three had side-mounted tailpipes.
For the 20 children, average COHb levels measured in an  emergency room were 18.2% (mean)  ±2.4%
(standard error), and levels ranged from 1.6 to 37.0% COHb.
       Studies done both before and after 1991 continue to show that fixed-site monitors underestimate
in-vehicle CO exposures. Flachsbart (1995) reported that 14 of 16 in-vehicle exposure studies performed

                                             4-9

-------
in the United States between 1965 and 1992 simultaneously measured both ambient and passenger cabin
concentrations.  Regardless of the study, Table 4-2 shows that the mean CO concentrations inside vehicles
always exceeded the mean ambient CO concentrations measured at fixed-site monitors. The ratio between
a study's mean in-vehicle CO concentration and its mean ambient CO concentration fell between 2 and
5 for most studies, regardless of when the study was done, but exceeded 5 for two studies done during the
early  1980s. Of the more recent studies, Chan et al. (1991) found that median CO concentrations were
11 ppm inside test vehicles driven on hypothetical routes in Raleigh, NC, during August and September
1988, but median ambient concentrations were only 2.8 ppm at fixed-site monitors. Fixed-site samples
were collected about 30 to 100 m from the midpoint of each route.
       Flachsbart (1995) proposed that the results of the 16 studies could be explained partly by different
study approaches (direct versus indirect) and by other aspects of study design, including choice of city,
season of the year, the surveyed road's functional type and location, travel mode, and vehicular ventilation.
For example, by pairing three direct studies (studies 5,11 and 13 in Table 4-2) with three indirect studies
(studies 4, 12 and 14, respectively, in Table 4-2) that were done at the same time, Flachsbart showed that
the mean in-vehicle exposure measured by the direct approach was always lower than that measured by the
indirect approach. Although direct studies sampled real populations engaged in a variety of trips in all types
of traffic, most  indirect studies focused on hypothetical commuters with higher exposures in rush hour
traffic. In another example, a comparison of studies 8 through 10 in Table 4-2 shows the effect of roadway
type.  Study 9 had a sizeable component of residential driving, which may explain why the mean in-vehicle
CO exposure of 7.7 ppm for study 9 was lower than the mean exposures for studies 8 and 10.  When study 9
data were disaggregated by roadway location, mean CO concentrations were 10 ppm for major commuting
routes and 5.5 ppm for drives in residential areas. The mean concentration of 10 ppm for major commuting
routes in study 9 is similar to the mean CO concentrations reported for arterial highways by other studies
(i.e., 9.8 ppm for study 8 and 10.6 ppm for study 10), which also were done during the early 1980s.
       Like earlier studies, recent ones also have looked at effects of different routes and travel modes on
CO exposure.  Chan et al. (1991) reported significantly different in-vehicle exposures to CO for
standardized drives on three routes  that varied in traffic volume and speed. The median in-vehicle CO
concentration was 13 ppm for 30 samples in the downtown area of Raleigh, which had heavy traffic
volumes, slow speeds, and frequent stops. The next highest concentrations (median =11 ppm, n = 34)
occurred on an  interstate beltway that had moderate traffic volumes and high speeds, and the lowest
concentrations (median = 4 ppm, n = 6) occurred on rural highways with low traffic volumes and moderate
speeds.  Similarly, Dor et al. (1995) reported CO exposures of 12 ppm for 19 trips lasting an average of
82 min on a route through central Paris, France, which was 2 to 3 ppm higher than the mean exposure for
30 trips split between two suburban  routes. In terms of travel modes, both Joumard (1991) and Dor et al.
(1995) found differences in CO exposures for public and private modes of travel in French cities and towns,
confirming findings made earlier in  the United States by Flachsbart et al. (1987).
       Ott et al. (1994) developed statistical models of passenger cabin  exposure to CO concentrations
from highway emissions, based on  88 trips taken during a 13.5-mo period in 1980 and 1981. All trips
occurred in one vehicle with windows set in a "standard position" as  it traveled an arterial highway
(El Camino Real) in the  San Francisco Bay area of California.  The models are noteworthy because they
examined the explanatory power of nine variables. The best model predicted the average CO exposure per
trip as a function of just two variables: (1) traffic conditions, as measured by the proportion of travel time
stopped, and (2) a seasonal trend term expressed as a cosine function of the day of the year on which the
trip was taken (adjusted multiple correlation coefficient  [R2] = 0.67).  A model that included ambient CO
concentrations from a fixed-site monitor slightly improved the explanatory power of the model (adjusted
R2 = 0.71).
       Flachsbart (1999a) developed a set of statistical models of passenger exposure to CO concentrations
inside a motor vehicle as it traveled a coastal artery (the Kalanianaole Flighway) in Honolulu, HI. All trips

                                             4-10

-------
                Table 4-2. Summary of Studies of In-Vehicle Exposure and Ambient
                            Carbon Monoxide Concentrations, 1965 to 1992
Study
Sites
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Study
Period
=4965
=4966
April 1966 - June 1967
-Summer 1973
October 1974 - February 1975
July - August 1978
October 1979
January 1980 -February 1981
January - March 1981
November 1981 -May 1982
November 1982 - February 1983
January - March 1983
May 1987 -March 1988
August - September 1988
January 1991 -March 1992
November 1991 -December 1992
Mean In-Vehicle
CO Concentration
(ppm)
37.0
31.8b
25.4C
17.5
13.4
13.8
13.1
9.8
7.7d
10.6
6.5f
11.7
8.6
11.3
4.6
3.0
Variation of In-Vehicle
CO Concentration
(ppm)a
23-58
7-77
18-36
15-20
±5.5
±7.2
10.9-15.3
±5.8
±3.6 -±7.7
5-30
0.14-0.32g
±2.2 -±9.0
±4.95
±5.1
±2.1
±2.9
Typical Ambient
CO (ppm)
20-30
14.3
No data
6.0
6.0
3.5
3.4
<1.5
2.5-8.4e
1.1
3.2-6.6 e
2.3
3.7
2.9
<1.0
No data
Study Sites:
1. Los Angeles, CA
2. Chicago, IL; Cincinnati, OH; Denver, CO;
   St. Louis, MO; Washington, DC
3. 14 cities
4. Los Angeles, CA
5. Boston, MA
6. Washington, DC
7. Los Angeles, CA
8. Menlo Park, Palo Alto, and Los Altos, CA
 9.  Denver, CO; Los Angeles, CA; Phoenix, AZ;
    Stamford, CT
10.  Honolulu, HI
11.  Denver, CO; Washington, DC
12.  Washington, DC
13.  Los Angeles, CA
14.  Raleigh, NC
15.  Menlo Park, Palo Alto, and Los Altos, CA
16.  New Jersey Turnpike and Route 18, NJ	
Note: Numbers shown below in parentheses are mean in-vehicle CO concentrations in ppm.
"Range or one standard deviation except as noted.
"•Chicago (37), Cincinnati (21), Denver (40), St. Louis (36), Washington (25).
'Atlanta (29), Baltimore (21), Chicago (24), Cincinnati (23), Denver (29), Detroit (25), Houston (23), Los Angeles (36), Louisville (20),
 Minneapolis-St. Paul (28), New York (34), Phoenix (29), St. Louis (18), Washington (19).
dDenver (10.3), Los Angeles (8.5), Phoenix (6.7), Stamford (5.2), commuting and residential driving microenvironments weighted by sample
 size.
'Range across all cities studied.
"Denver (8.0), Washington (5.0).
8Range in standard error for all cities studied.

Source:  Flachsbart (1995).
                                                     4-11

-------
occurred during morning periods over a 6-mo period in 1981 and 1982. The 6.2-km study site was divided
into three links. The models predicted the average CO concentration inside the vehicle's passenger cabin
on the third link as a function of several variables. Based on data for 80 trips, the three most powerful
models  (adjusted R2 = 0.69) were nonlinear combinations of several  variables:   the  average CO
concentration inside the cabin for the second link; wind speed and direction; and either travel time, vehicle
speed, or the estimated motor vehicle CO emission factor for the third link. Several nonlinear models were
based on data for 62 trips for which nonzero, ambient CO concentrations were available. For this smaller
database, the most practical models (adjusted R2 = 0.67) combined three variables: (1) the ambient CO
concentration; (2) the second-link travel time; and (3) either the travel time, vehicle speed, or CO emission
factor for the third link.  The models showed that cabin exposure was strongly affected by travel time and
average vehicle speed, both of which were affected by the time that the test vehicle entered each link of the
highway. This implied that stochastic simulations of exposure (e.g., the SHAPE model by Ott [1984] and
Ott et al. [1988]) should not assume that trip times and commuter exposures are independent of trip-starting
times.
       The most recent U.S. study of in-vehicle CO exposure reported results for hypothetical commutes
on standardized routes in Los Angeles and Sacramento, CA, during early fall, 1997 (Rodes et al., 1998).
Continuous CO concentrations were measured over 2-h periods at different times of day for 29 trips by two
test vehicles (one vehicle following the other).  The CO levels were measured both inside and outside (at
the base of the windshield) of the vehicles, and outdoors  along surveyed routes and at  nearby fixed
monitoring stations. Since the minimum quantification limit (MQL) of the portable monitor was 2 ppm,
all data below the MQL were treated as zero concentrations.  The research design employed a balanced
factorial design to determine the range of CO concentrations encountered in each city under different
scenarios.  The scenarios represented combinations of different test vehicle types and ventilation settings,
roadway types, level of roadway congestion, and time of day. The study is unique in that the lead test
vehicle, equipped with a video camera, followed  vehicles with detectable  emissions (by eye or nose)
whenever possible.  Although the lead vehicle frequently trailed city buses and heavy duty  diesel trucks,
it also targeted visibly gross-emitting, gasoline-powered vehicles.
       Because the California study design purposely emphasized  scenarios  likely to result  in high
in-vehicle emissions, the study results cannot easily be compared to results of the 16 studies in Table 4-2.
Mean CO concentrations inside both test vehicles were reported for each scenario based on only two to four
commutes in each city.  These concentrations ranged from less than MQL to 2.6 ppm, based on a total of
13 trips in Sacramento, and from 3.0 to  6.0 ppm, based on a total of 16 trips in Los Angeles.  As in similar
studies, the means of ambient CO concentrations measured at fixed-site monitors fell below the means of
in-vehicle CO levels in both  cities, and typically were less than the MQL of the portable monitor. Because
ambient CO levels vary from city to city, the study computed net microenvironmental concentrations of test
vehicles by subtracting ambient from interior CO  levels to  estimate CO exposure resulting solely from
roadway emissions.   They found average net microenvironmental concentrations to be higher in
Los Angeles (4.6 to 4.9 ppm) than in Sacramento (2.1 to 3.1 ppm) during rush hour trips on freeways; but,
there was little difference in  average net microenvironmental concentrations between freeway and arterial
trips during rush hour trips in both cities. Based on preliminary analysis of five trips, the study reported that
CO concentrations could reach short-term peaks, ranging from =15 to 70 ppm,  when the vehicle trailed
gasoline-powered delivery trucks and older sedans.
       The CO exposure of cycling as a travel mode has been studied and compared to the exposure of
motorists. In England, Bevan et al. (1991) reported that the mean CO exposure of cyclists in Southhampton
was 10.5 ppm, based on 16 runs over two 6-mi routes that took an average of 3 5 min to complete. Note that
the CO exposures of European cyclists may not be comparable to cyclists' exposures in the United States
because installation of catalytic converters on new cars in Europe occurred in 1988, about 13 years after
their introduction  in the United States (Faiz et al., 1996).  In The Netherlands, Van Wijnen et al. (1995)

                                             4-12

-------
compared exposures of volunteers serving as both car drivers and cyclists on several routes in Amsterdam
during winter and spring. For a given route, the mean personal 1-h CO concentrations were always higher
for car drivers than for cyclists regardless of when sampling occurred during the year. However, a volunteer
breathed 2.3 times more air per minute on  average as a cyclist than as a car driver.  When  adjusted for
variation in breathing rate, the range in median 1-h averaged uptakes of CO of cyclists (2.4 to 3.2 mg)
approached that of car drivers (2.7 to 3.4 mg).
       Studies have quantified the effect of traffic  volume  and speed  on in-vehicle CO exposure.
Flachsbart et al. (1987) reported that in-vehicle CO exposures fell by 35% when  test vehicle speeds
increased from 10  to 60 mph on eight commuter routes in Washington.  In a similar study of typical
commuter routes in central Riyadh, Saudi Arabia, Koushki et al. (1992) found that in-vehicle CO exposures
fell by 36% when vehicle speeds increased from 14 to 55 km/h (9 to 34 mph). They also found that mean
in-vehicle CO concentrations increased by 71.5% when traffic volumes increased from 1,000 to 5,000
vehicles per hour. Mean CO levels ranged from 30 to 40 ppm, averaged over trips of 25 to 43 min during
peak hours, and ranged from 10 to 25 ppm for trips of 15 to 20 min during off-peak hours.
       The effects of diurnal and seasonal variation on  in-vehicle CO exposure were not discussed
completely in the previous CO criteria document.  Studies of diurnal effects on in-vehicle exposure during
peak travel periods have been inconclusive because they did not control for covariation in traffic volumes
and speeds, ambient CO concentrations, or meteorological conditions (e.g., temperatures, wind speeds)
during different periods of the day. In Los Angeles, Haagen-Smit (1966) found evidence that CO exposures
during afternoon commutes were greater than those during morning commutes. Similar results were found
later by Cortese and Spengler (1976) in Boston, MA, by Wallace (1979) in Washington, and by Dor et al.
(1995) in Paris. However, contrary evidence was reported by Holland (1983) for four U.S. cities and by
Joumard (1991) in France.  Recently, Rodes et al. (1998) compared commuter exposures for morning and
evening rush hour periods for 12 freeway trips in two California cities. Ambient CO levels were subtracted
from in-vehicle CO levels to estimate the vehicle's net microenvironmental concentration. In Los Angeles,
net microenvironmental concentrations during evening commutes were  about 25% lower than morning
values because of higher wind speeds. In Sacramento, such net concentrations during evening trips were
slightly higher than morning values because of higher traffic congestion levels. In another recent study,
Aim et al. (1999) reported that the geometric mean CO concentration of 11  morning trips (3.1 ppm)
exceeded that of 12 afternoon trips (2.0 ppm) on a standard route in Kuopio, Finland, but attributed this
difference to weather and traffic variables.
       Seasonal variations in ambient temperatures, wind conditions, and traffic volumes affect passenger
cabin exposure to CO, as shown in studies by Shikiya et al. (1989) in southern California, Ott  et al. (1994)
in northern California, Dor et al. (1995) in France, and Flachsbart (1999a) in Hawaii. Ott et al. (1994) and
Dor et al. (1995), who both measured exposures for an entire year, reported that exposures were generally
higher in the fall and winter than in the spring and summer. Such results usually are attributed to colder
temperatures in temperate climates, which increase CO emissions per vehicle mile during winter months.
In Hawaii,  where temperatures are never  cold enough to have a substantial  effect on motor  vehicle
emissions, Flachsbart (1999a) found that traffic flows and wind speeds had reinforcing effects on passenger
cabin exposures to CO concentrations on a coastal highway in Honolulu.  During late fall, exposures were
low because traffic flows were light and wind speeds were high. During winter and spring, exposures were
relatively higher because traffic flows were greater and winds calmer than during the fall.

4.4.1.2   Exposure to Carbon Monoxide in Recreational Vehicles
       Two studies examined personal exposure to CO in the exhaust of recreational vehicles. In the first
study, Simeone (1991) collected CO concentrations in the passenger areas of large power boats with
side-mounted exhausts during routine cruises offshore of Annapolis, MD, and Boston. In Boston Harbor,
CO concentrations averaged 56 ppm during  a 60-min cruise and 28 ppm after a 30-min  cruise.  For


                                             4-13

-------
Chesapeake Bay cruises near Annapolis, average stabilized CO concentrations at the helm ranged from
93 to 170 ppm over 20- to 30-min periods and 272 ppm over 30 min on the rear deck near the transom of
the boat. In both studies, exhaust gas was affected significantly by airflow about the boat under certain head
winds.  At head wind speeds of 10 to 30 knots, turbulent mixing occurred in closer proximity to the rear of
the boats, enabling exhaust gases to migrate freely into each boat.
       In the second study, Snook (1996) studied the CO exposure of a snowmobiler while traveling in the
wake of a lead snowmobiler on a 2- to 3-mi straight trail over level terrain in Grand Teton National Park,
WY. The CO exposure of the following snowmobiler was measured under stable atmospheric conditions
in Tedlar™ bags. The distance between the two snowmobiles ranged from 25 to  125 ft, and speeds ranged
from 10 to 40 mph. The follower's  maximum average centerline exposure was 23.1 ppm, which occurred
at 10 mph and 25 ft behind the lead  snowmobile.  Although Snook (1996) reported no averaging times for
exposures,  one  can  estimate that  these times ranged from 3 to 18 min from the data given on the
snowmobiler's travel distance and vehicle speed. In general, the centerline CO  concentrations decreased
with increasing distance between snowmobiles and increased with greater speeds,  but only for distances
greater than 25 ft  between snowmobiles.  At 15 ft off centerline, average concentrations fell sharply to
levels of 0 to 7.5 ppm. When the  snowmobiler drove alone (self-exposure), the average concentration
minus the background concentration was 1.3 to 3.0 ppm. Background concentrations ranged from 0.2 to
0.5 ppm.
       Snowmobile tourism has become a booming business across the nation including several national
parks. For example, over 87,000 tourists traveled by snowmobile in Yellowstone National Park (Wilkinson,
1995) during the winter of 1993 and 1994. Under steady-state conditions, a snowmobile may emit from
10 to 20 g/mi of CO, while a modern U.S. automobile equipped with a catalytic converter emits far less
(0.01 to 0.04 g/mi) at speeds of 10 to 40 mph.   There are no federal laws regulating the exhaust from
snowmobile engines, and states are preempted from implementing snowmobile  emission standards.  The
typical snowmobile utilizes a two-stroke engine, because it is less expensive than a four-stroke engine and
provides a high power: weight ratio. However, a two-stroke engine produces relatively high emissions of
CO (Snook and Davis, 1997).

4.4.1.3  Residential Exposure to Carbon Monoxide
       Residential sources of CO concentrations include motor vehicle operation inside an attached garage
and the use of unventilated or poorly ventilated kerosene space heaters, gas appliances, and charcoal grills
and hibachis in the living area of the home. Studies of exposures to nonfatal concentrations are discussed
first, followed by studies of unintentional deaths  caused by high indoor concentrations.
       According to the Barbecue Industry Association, 44 million American households owned a charcoal
grill in 1989, and an estimated 600 million charcoal-barbecuing events take place  annually (Hampson et al,
1994).  An early study showed that the air stream from charcoal grills contains 20  to 2,000 ppm of CO, with
75% of grills emitting 200 ppm and above (Yates, 1967).  Gasman et al.  (1990) reported COHb levels
ranging from 6.9 to 17.4% in a family of four people in northern California who had been exposed to smoke
from cooking indoors on a barbecue grill, which was found by fire fighters in the middle of the living room.
       Mumford et al. (1991) and Williams et al. (1992) assessed CO exposure to emissions from unvented
portable kerosene heaters in eight small mobile homes with no gas appliances and low air exchange rates.
Each home was monitored for an average of 6.5 h per day for 3 days per week for 4 weeks. For 2 weeks,
the heater was on, and, for 2 weeks,  it was off. When the heater was turned on, it was in use for an average
of 4.5 h. When the heater was in use, study participants (all nonsmokers) spent most of their time in the
family room or kitchen. Sampling took place in the living area about 1.5 to 3 m from the heater.  The mean
8-h CO concentrations were 7.4 ppm (1 -h peak =11.5 ppm) when the heater was  on and 1.4 ppm (1 -h peak
= 1.5 ppm) when it was off.  Peaks usually were observed at the end of the combustion period. The ambient
CO level measured 0.5 h prior to heater use ranged from 0 to 8 ppm.  When the heater was on, three of the


                                            4-14

-------
eight homes had 8-h average CO levels that exceeded the NAAQS, and one home routinely had levels of
30 to 50 ppm.
       Wilson et al. (1993a,b) and Colome et al. (1994) reported CO exposures for a random sample of
California homes that used gas appliances during a 48-h period from December 1991 to April  1992. For
periods of 48 h,  the median CO concentration was 1.2 ppm (indoors) and 0.8 ppm (outdoors), and the
median of the maximum 8-h average CO concentration was 2.0 ppm (indoors) and 1.4 ppm (outdoors). Of
surveyed homes, 13 of 286 homes (4.5%) had indoor CO concentrations above the NAAQS of 9 ppm for
8 h, and 8 of 282 homes (2.8%) had outdoor CO concentrations above this standard.  Although most of the
exceedances occurred in the Los Angeles basin, these percentages  could be low because the  basin was
underrepresented in the statewide sample. The study did not translate these percentages into statewide
estimates.
       Figures 4-4 and 4-5 show log-probability plots of the 1-h and 8-h maximum indoor minus outdoor
CO concentrations, respectively, for a common sample of 277 homes. These figures show that 17 homes
(6.1%) had 1-h maximum concentrations indoors that were at least 5 ppm higher than outdoor levels, and
that 10 homes (3.6%) had 8-h maximum CO concentrations indoors that were at least 5 ppm higher than
outdoor levels.  They suggest that a small percentage of California homes would still have indoor CO
problems even if outdoor CO levels  at these homes complied with federal ambient standards.  Using
univariate regression analysis, outdoor CO concentrations explained approximately 55% of the variation
found in indoor CO concentrations.
                    20 30 40 50 60 70
                      Percentage
 Figure 4-4.  Log-probability plot of the maximum 1-h
 indoor minus outdoor CO concentrations based on a
 random sample of 277 homes that used gas appliances in
 California, 1991 and 1992.  (Only those indoor minus
 outdoor values greater than or equal to 0.1 ppm are
 shown.)
 Source: Wilson et al. (1993a).
                                                 E
                                                 Q.
                                                 B
                                                 8 10t
                                                 .c
                                                 OD
                                                 E
                                                 E
                                                             12  5  10 20 30 40 50 60 70
                                                                                       9.6 99.9 99 99
                     Percentage
Figure 4-5.  Log-probability plot of the maximum 8-h
indoor minus outdoor CO concentrations based  on a
random sample of 277 homes that used gas appliances in
California, 1991  and 1992. (Only those indoor minus
outdoor values greater than or equal to 0.1 ppm are
shown.)
Source: Wilson et al. (1993a).	
Higher net indoor CO levels (indoor minus outdoor CO concentrations) were traced definitively to space
heating with gas ranges and gas-fired wall furnaces, use of gas ranges with continuous gas pilot lights, small
home volumes, and cigarette smoke; however, several other factors also may have contributed to the higher
CO levels: malfunctioning gas furnaces, automobile exhausts leaking into homes from attached garages
and carports, improper use of gas appliances (e.g., gas fireplaces), and improper  installation of gas
appliances (e.g., forced air unit ducts).
       Unintentional deaths caused by CO poisonings have been studied in California, New Mexico, and
Washington.  Two California studies collected data for the 1979 to 1988 period. In the first, Liu et al.
                                             4-15

-------
(1993) reported that 13.3% of 444 deaths were caused by improper use of charcoal grills and hibachies, of
which 54% occurred inside motor vehicles (e.g., vans, campers) and 46% in residential structures (e.g.,
homes, apartments, shacks, tents). Relative to their share of the state population, higher death rates
occurred among Asians, blacks, males, and people aged 20 to 39. In the second study, Girman et al. (1998)
identified specific factors that contributed to unintentional deaths caused by CO from several combustion
sources (e.g., charcoal grills and hibachis, other heating and cooking appliances, motor vehicles, small
engines, camping equipment). There was a strong association between alcohol use and CO poisoning from
motor vehicles. Typically, motorists under the influence of alcohol would pull into their garages, leave the
engine running while listening to cassette tapes, and then fall asleep. Faulty heating equipment used during
winter months was implicated in about 50% of all unintentional deaths in studies by both Girman et al.
(1998) in California and Yoon et al. (1998)  in New Mexico.
       Based on data for  10  counties in Washington, Hampson et  al. (1994) reported  features  of
unintentional CO poisoning cases that occurred between  1982  and 1993.  Most cases occurred when
electrical power was interrupted during fall and winter months, because of either regional storms or unpaid
utility bills.   Of 509 patients treated with  hyperbaric oxygen, 79 (16%) were  exposed when charcoal
briquets were burned for heating or cooking in 32 separate incidents.  Non-English speaking Hispanic
whites and Asians were disproportionately represented among the cases. The COHb levels averaged 21.6%
and ranged from 3.0 to 45.8%.
       The National Center for Health Statistics and the U.S. Consumer Product Safety Commission
(CPSC) estimated that 212 deaths in 1992 were caused by fuel-burning appliances used in the home.  Of
these deaths,  13 involved use of gasoline-powered appliances (National Center for Health  Statistics and
U.S. Consumer Product Safety Commission, 1992).  The CPSC also estimated that 3,900 accidental CO
injuries occurred in 1994, of which about 400 were associated with the use of gasoline-powered engines
or tools (National Institute for Occupational Safety and Health, 1996). In response to the problem, several
federal government agencies issued a joint alert concerning exposure  to CO emitted  by  these sources
(National Institute for Occupational Safety and Health, 1996).  These sources involved use of pressure
washers, air compressors, concrete-cutting saws, electric generators, floor buffers, power trowels, water
pumps, and welding equipment. Unintentional CO poisonings frequently happened indoors even when
people took precautions to ventilate the building.

4.4.1.4  Exposure to Carbon Monoxide at Commercial  Facilities
       Although motorists typically turn off their engines during refueling, people still may be exposed to
CO concentrations from idling cars and other cars entering and leaving the fueling area of the station.
Wilson et al. (1991) randomly sampled 100 self-service filling stations and, for comparison,  took
convenience  samples at 10 parking garages and 10 nearby office buildings in Los Angeles,  Orange,
Riverside, and San Bernardino counties of Southern California.  They took 5-min samples of 13 motor
vehicle air pollutants, including  CO, in  each microenvironment and  in  the  ambient  environment.
Microenvironmental and ambient concentrations were measured on the same day but not simultaneously.
The highest median CO concentration occurred in parking garages (11.0 ppm), followed by service stations
(4.3 ppm), and office buildings (4.0 ppm).  The median ambient CO concentration was  2.0 ppm.
       Ice skating, motocross, and tractor pulls are sporting events in which significant quantities of CO
may be emitted in short periods of time by machines in poorly ventilated indoor arenas. The  CO is emitted
by several sources, including ice resurfacing machines and ice edgers during skating events; gas-powered
radiant heaters used to heat viewing stands; and motor vehicles atmotocross, monster-truck, and tractor-pull
competitions. These competitions usually involve many motor vehicles with no  emission controls.
       Several studies of CO exposure in commercial facilities were not cited in  the previous CO criteria
document. First, Kwok (1981) reported episodes of CO poisoning among skaters inside four arenas in
Ontario, Canada. Mean CO levels ranged from 4 to  81 ppm for periods of about 80 min. The CO levels


                                             4-16

-------
in the spectator areas ranged from 90 to 100% of levels on the ice rinks. The ice resurfacing machines
lacked catalytic emission controls. Second, both Sorensen (1986) and Miller et al. (1989) reported CO
concentrations greater than 100 ppm in rinks from the use of gasoline-powered resurfacing machines. High
concentrations were attributed to poorly maintained machines and insufficient ventilation in one rink.
Third,  based on  data collected  in the Quebec city area, Levesque et al.  (1990)  developed  a linear
relationship between CO exposure and the CO concentration in exhaled breath (see Section 4.3.1.5 for
discussion of CO exposure and breath CO relationships) but could not eliminate other factors affecting the
relationship. In a later study, Levesque et al. (1991) measured the alveolar CO of 14 male adult nonsmokers
who played ice hockey, but who were not exposed in occupational settings. Rink CO concentrations ranged
from 0 to 76.2 ppm.  The study again found a linear relationship between exposure and absorbed CO such
that, for each 10  ppm of CO in the indoor air, the players absorbed enough CO to raise alveolar CO by
4.1 ppm or about 0.76% COHb.
       In the United States, surveys of CO exposure were done at ice arenas in Vermont, Massachusetts,
Wisconsin, and Washington.  For a rink in Massachusetts, Lee et al. (1993) showed that excessive CO
concentrations can occur even with well-maintained  equipment and  fewer  resurfacing  operations if
ventilation is inadequate. Average CO levels were less than 20 ppm over 14 h, with no significant source
of outdoor  CO.  Ventilation systems could not disperse pollutants emitted and trapped by temperature
inversions and low  air circulation at ice level.  In another  study, Lee et al.  (1994) reported that CO
concentrations measured inside six enclosed rinks in the Boston area during a 2-h hockey game ranged from
4 to  117 ppm, whereas outdoor levels were about 2 to 3 ppm, and the alveolar CO of hockey players
increased by an average of 0.53 ppm per 1 ppm CO exposure over 2 h.  Fifteen years earlier, Spengler et al.
(1978) found CO levels ranging from 23 to 100 ppm in eight enclosed rinks in the Boston area, which
suggests that CO exposure levels in ice arenas have not improved.
       In a letter, Paulozzi et al. (1991) reported that  25  people  exposed to CO  during a Vermont
high-school ice hockey game had mean COHb levels of 8.9%, but did not report whether any of them were
smokers. Although Paulozzi et al. (1991) was unable to measure CO concentrations at the game, Smith
et al. (1992) reported CO levels of 150 ppm (no averaging time was given) at an indoor ice-hockey rink in
Wisconsin. To document the extent of the problem in Vermont, Paulozzi et al. (1993) measured CO during
eight high-school games in the state, and reported that average CO levels for the entire game ranged from
<5 to 101 ppm, with a mean of 3 5 ppm.  Hampson (1996) reported a maximum CO level of 3 54 ppm inside
an ice arena in Seattle in March 1996. Based on data for 17 persons, whose tobacco use was not reported,
the  average COHb level was 8.6%  (range 3.3 to  13.9%).  The source of CO was a malfunction in a
20-year-old ice resurfacing machine. Hampson also reported that CO may have diffused into an adjacent
bingo hall through an open door.  In view of these studies, the State of Minnesota declared in Regulation
No. 4635 that CO measurements taken 20 min after ice  resurfacing must be less than 30 ppm.
       Studies also  have been done in sports  arenas that allow motor vehicles. Boudreau et al. (1994)
reported  CO levels for three indoor sporting events (i.e., monster-truck competitions, tractor pulls) in
Cincinnati. The CO measurements were taken before and during each event at different elevations in the
public seating area of each arena with most readings obtained at the midpoint elevation where most people
were seated. Average CO concentrations over 1 to 2 h ranged from 13 to 23 ppm before the event to 79 to
140 ppm during the event. Measured CO levels were lower at higher seating levels. The ventilation system
was operated maximally, and ground-level entrances were completely open.
       High CO levels also have been found at motor vehicle competitions in Canada.  In a study not cited
in the previous CO criteria document, Luckurst and Solkoski (1990)  recorded CO concentrations at two
tractor-pull events in Winnipeg, Manitoba. The mean instantaneous  concentration at 25 locations in the
arena ranged from 68 ppm at the start of the first event to 262 ppm by the end. At the second event, the
range was 78 to 436  ppm. Levesque et al. (1997) reported CO levels  at an indoor motocross competition
held in a skating rink in the Quebec city region.  The May 1994 event lasted from roughly 8 p.m. to

                                            4-17

-------
midnight.  Average CO concentrations determined at five stations located at different points in the arena
ranged from 19.1 to 38.0 ppm, with the higher levels measured during the second half of the show. High
CO concentrations forced a health official to interrupt the event seven times to  help clear the air.
Covariance analysis showed that CO levels were related to the initial CO concentration, the event duration,
engine size, and especially the number of motorcycles on the track.

4.4.1.5  Studies of Breath Carbon Monoxide in Populations:  The Effects of Exposure to
        Carbon Monoxide
       The concentration of CO in the end-tidal breath of a nonsmoker, after a standardized breathholding
maneuver, can be related to an exponentially time-weighted average of the previous CO exposures (U.S.
Environmental Protection Agency, 1991). As described in Chapter 2, the breath concentration of CO also
can be related to percentage of blood COHb by constructing a calibration curve from simultaneous blood
and end-tidal breath sampling.  These  end-tidal breath CO measurements can demonstrate whether
nonsmoking subjects recently have exceeded a protective level of 2.1% COHb (U.S. Environmental
Protection Agency, 1992b).  Such measurements are more informative than an 8-h personal CO exposure
measurement because a nominal 8-h average of 9 ppm  can attain different COHb concentrations.  For
example, a subj ect starting with 0.5% COHb will reach a higher COHb level after a 4-h exposure to 3 ppm
CO followed by a 4-h exposure to 15 ppm CO than the reverse combination (i.e., exposed to 4-h at 15 ppm
CO followed by 4-h at 3 ppm CO).
       The EPA  reviewed the  pre-1990 literature reports of breath CO measurements in various
populations (U.S.  Environmental Protection Agency, 1991; Section 8.5.2.2).  These data and the more
recent data on breath CO found  in the following part of this section often are collected with different
breathhold-time, often are uncorrected for the CO content of the inhaled air (Smith, 1977; Wallace, 1983),
and also may be subj ect to a positive hydrogen-interference if the breath CO is analyzed electrochemically
(Vreman et al, 1993) (see Section 2.6.2 of this document). Consequently, this should be taken as a caveat
by the reader that a portion of the variance among the results of different studies may be related to different
breath collection methods and different breath CO measurement techniques.
       Lando et al. (1991) collected breath samples of 4,647 workers using MiniCO breath kits (Model
1000, Catalyst Research Corporation, Owings Mills, MD).  The latter part of a breath was collected in a
balloon following a 15-s breathhold, but the method of analysis was not described. Although the authors
cite Jarvis et al. (1980) forthis method, Jarvis et al. (1980) used the Jones et al. (1958) method that requires
a 20-s breathhold.  Furthermore, these data are uncorrected for the amount of CO in the maximal inhalation
prior to the breathhold step (Smith, 1977; Wallace, 1983).  Consequently, these data are not compatible with
other studies using 20-s for the breathhold time and corrected data.  Mean CO levels (Table 4-3) ranged
from 4.2 (±1.66 standard deviation [SD]) ppm for never-smokers to 33.3 (±11.22 SD) ppm for heavy
smokers (25 cigarettes/day or more). Based on cutoffs of 3  and 6 ppm above ambient, a larger number of
ex-smokers (1.7to 3.3%) than never-smokers (0.4 to 1.9%) appeared to be falsely reporting their smoking
status.
       Chung et al. (1994) employed the Lee and Yanagisawa (1992, 1995) sampler to measure personal
exposure to CO of 15 Korean housewives using charcoal briquettes for cooking.  The COHb levels also
were  measured using a CO-Oximeter (CO-Ox). Although the personal sampler had somewhat high
imprecision based on four duplicate samples (average of 2.1 ppm difference), the investigators were able
to  document a higher level of both exposure to CO and blood COHb when the charcoal briquettes were
used.  Levels of COHb were generally high, even without use of the briquettes,  leading the experimenters
to  hypothesize that the high prevalence  of smoking (all 15 subjects had smokers in their homes) had
elevated the level above the levels found in the U.S.  among nonsmokers.
       Seufert and Kiser (1996) measured CO levels in the end-tidal breath after a 10-s breathhold of
126 crew members of a nuclear submarine just before and just after a62-h submerged period.  The CO level


                                            4-18

-------
            Table 4-3.  Mean (M) Breath Carbon Monoxide Levels and Sample Sizes
                         Across Smoking Categories and Job Types3
Smoking Category
Never-smokers


Quitters


Occasional smokers


Light smokers
(1 to 15 cigarettes per day)

Moderate smokers
(16 to 24 cigarettes per day)

Heavy smokers
(25 cigarettes per day or more)


M
SD
n
M
SD
n
M
SD
n
M
SD
n
M
SD
n
M
SD
n
Total
4.2
1.66
2,328
4.6
3.10
1,148
7.6
6.12
178
14.3
8.40
238
24.7
10.47
351
33.3
11.22
273

Blue Collar
4.5
1.89
294
5.1
5.03
217
7.6
3.98
22
15.6
8.94
48
24.6
11.72
97
32.6
9.61
95
Job Type
Clerical
4.1
1.69
958
4.4
2.19
427
8.0
7.05
90
14.0
8.70
131
25.4
10.06
180
34.1
12.73
117

White Collar
4.1
1.55
1,076
4.5
2.63
504
7.1
5.34
66
13.79
7.21
59
23.4
9.67
74
33.0
10.50
61
aSample size (n) refers to those with CO measurements; CO measurements were taken on 97.2% of those interviewed.
The CO levels are in parts per million. Data are for end-tidal breath collected after a 15 -s breathhold, without the required
correction for the CO in the inhaled air (Smith, 1977; Wallace, 1983)

Source: Lando et al. (1991).
in the submarine (called "ambient" by the authors) increased from 2.6 ppm to 9.2 ppm in the fan room and
in two other spaces.  The authors state that the increase was caused primarily by cigarette smoke from the
40 smokers aboard because auxiliary diesel engines were not used during the submersion period.  The
nonsmokers'  breath CO  increased from 9  to  21  ppm.  Although the authors  did  not  comment on
the considerable difference between the nonsmokers' breath CO of 21 ppm and the measured "ambient"
concentration of only  9.2 ppm, it may  have been because of higher  smoking rates in  nonmonitored
duty sections than in monitored sections, the absence of a correction for higher CO concentrations  in
air inhaled for the  10-s  breathhold than in the end-tidal breath  CO, and  the end-tidal breath CO
and "ambient" CO measurements being made with two different instrument systems.  Operation of a
revitalization system that removed CO also may have contributed to a lower monitored "ambient" CO than
the CO nonsmokers were exposed to in the nonmonitored duty sections.
       Zayasu et al. (1997) present the first study showing that asthmatics untreated by corticosteroids have
higher 20-s breathhold end-tidal breath CO than either healthy controls or treated asthmatics, as determined
by subtracting the background level from the observed reading (Figure 4-6).  This is not the required
correction for CO in the inhaled air reported by Smith (1977) and Wallace (1983), so these data are not
consistent with those studies where this correction was made.  They attribute the higher levels to lung

                                             4-19

-------
   15-1
E
Q.
Q.

O
O
T3
LU
   10-
    5-
            Control
Untreated
Asthmatic
 Treated
Asthmatic
                                                     inflammation, leading to a possible increase in
                                                     heme oxygenase, which creates endogenous
                                                     CO. (For more information on endogenous CO
                                                     production, see  Section 5.3.)
                                                            Shenoi et al. (1998) tested 470 youths
                                                     (aged 5 to 20 years) in hospital admissions for
                                                     CO  in breath, using  the  electrochemical
                                                     Vitalograph Breath CO monitor (Vitalograph,
                                                     Inc., Lenexa, KS). The results, showing that
                                                     1.9% (9 of 470)  had end-tidal breath CO levels
                                                     greater than 9  ppm after a  20-s breathhold,
                                                     were confirmed by CO-Ox testing of blood.
                                                     Five of the nine  patients with the higher breath
                                                     CO were believed to be cigarette smokers, one
                                                     may have been exposed to fumes from a faulty
                                                     furnace, and three were believed to be exposed
                                                     to environmental  tobacco smoke  or traffic
                                                     exhaust.   No corrections were  made for the
                                                     parts per million of CO in the air inhaled for
                                                     the breathhold.

                                                     4.4.1.6 Nonoccupational Exposure to
                                                            Methylene Chloride
                                                            Nonoccupational   exposure    to
                                                     halogenated  hydrocarbons  like  methylene
                                                     chloride (see Sections  5.3 and  6.8.3), which
can be metabolized to CO in the body, potentially occurs when the chemicals are found in  contaminated
ambient air and groundwater used as drinking water and in consumer products that contain the chemical
as a solvent, flame-retardant additive, or propellant. Exposure to methylene chloride in the  home, for
example, primarily occurs through use of paint and varnish removers. Exposure also may occur through
use of aerosol propellants such as those found in hair sprays, antiperspirants, air fresheners,  and  spray
paints. The Agency for Toxic Substances and Disease Registry (1993) reported that some aerosol products
may contain up to 50% methylene chloride. However, the current extent of methylene chloride in aerosol
products apparently has not been studied recently; nor are typical population exposures  to methylene
chloride from consumer products known.
       Ambient exposure may occur near production and use facilities or near hazardous waste sites that
store methylene chloride. Ambient concentrations of methylene chloride near organic solvent cleaning and
paint  and varnish removal operations  range  from 7.1 to 14.3 ppb, averaged over  1 year  (Systems
Applications, Inc., 1983), and ambient levels at other locations were  reported by the U.S. Environmental
Protection Agency (1985). Although methylene chloride readily disperses when released into the air, it may
remain in groundwater for years and be ingested in drinking water or inhaled when it volatilizes during
showering and laundering (Agency for Toxic Substances and Disease Registry, 1993).
       Exposure to about 500 ppm of methylene chloride for several hours can elevate COHb levels to
15%.  Increases in COHb levels can be detected in the blood of nonsmokers about 30 min after exposure
to methylene chloride. Stewart et al. (1972) demonstrated that elevated COHb levels were proportional to
a series of controlled exposures to methylene chloride. In a controlled experiment, Stewart and Hake
(1976) observed postexposure levels of COHb ranging from 5 to 10% after 3 h of use of a liquid-gel paint
remover  containing 80% methylene chloride and 20% methanol by weight. Concurrent exposure to
Figure 4-6.  Excess CO concentrations in the exhaled air of
nonsmoking control subjects (n = 30), untreated asthmatics
(n = 30), and treated asthmatics (n = 30).  The values shown
were determined by subtracting the background level from the
observed  reading.    "Untreated"  means  no  inhaled
corticosteroids,  "Treated"  refers  to   regularly  inhaled
corticosteroids, and the horizontal bar indicates the mean
value.
Source: Zayasu et al. (1997).	
                                            4-20

-------
methylene chloride and methanol prolongs the period of elevated COHb in the body (Stewart and Hake,
1976; Buie et al, 1986; Wilcosky and Simonsen, 1991). Peterson (1978) reported COHb levels of up to
about 10% saturation after inhalation of methyl ene chloride concentrations ranging from 50 to 500 ppm
over 5 days for 7.5 h per day.

4.4.1.7 Exposures to Carbon Monoxide from Passive Smoking
       Given that the pNEM/CO exposure model accounts for the passive exposure of nonsmokers to CO
concentrations from smoking, this section briefly reviews two studies of this type.  In April 1992, Ott et al.
(1992b) took continuous readings of CO concentrations inside a passenger car for an hour-long trip through
a residential neighborhood of the San Francisco Bay Area. Measurements were taken in both the front and
back seats of the vehicle as a passenger smoked cigarettes. The neighborhood had low ambient CO levels,
because it had little traffic and few stop signs. During the trip, the air conditioning system was operated
in the recirculation mode. Concentrations in the front and rear seats were similar indicating that CO
concentrations were well mixed throughout the passenger compartment. CO concentrations reached a peak
of 20 ppm after the third cigarette.  Using the breath measurement technique of Jones et al. (1958),  the
breath CO level of the driver (a nonsmoker) increased from 2 ppm before the trip to 9.2 ppm at the end of
the trip. Based on a mathematical model, the predicted mean CO exposures for three air exchange rates
were 0.8  ppm for windows  open (20 mph); 7.2 ppm for windows closed (20 mph); and 63.3 ppm  for
windows closed (0 mph).
       Klepeis et al. (1999) reported CO concentrations for two social events that featured cigar smoking.
At one event, about 50 people were exposed to CO concentrations that averaged 5.8 ppm (range 5 to
11 ppm) for nearly a 2 h period. After subtracting the ambient CO level  (1.5 ppm), the event's net CO
concentration (4.5 ppm) was comparable to the CO exposure from freeway traffic during the drive to  the
event.  At the second event inside  a restaurant  in San Francisco, the CO concentration averaged over
200 min reached 10 ppm  (net 9 ppm after subtracting the ambient CO  level of 1 ppm). About  75% of
40 persons at the event were smoking cigars at any instant of time.

4.4.2  Occupational  Exposures
       This subsection discusses occupational exposures to CO and methylene chloride.

4.4.2.1  Exposures to Carbon Monoxide in the Workplace
       A survey by the National Institute for Occupational Safety and Health found that 3.5  million
workers in the private sector potentially are exposed to CO primarily from motor exhaust.  This number of
persons potentially exposed to CO in the work environment is greater than that for any other physical or
chemical agent (Pedersen and Sieber, 1988). In 1992, there were 900 work-related CO poisonings resulting
in death or illness in private industry as reported by the U. S. Bureau of Labor Statistics (as cited in National
Institute for Occupational Safety and Health [1996]).  Three risk factors  affect industrial occupational
exposure:  (1) the work environment is located in a densely populated area  that has high background (i.e.,
ambient)  CO concentrations; (2) the work environment produces CO as a product or by-product of an
industrial process, or the  work environment tends to accumulate CO concentrations that may result in
occupational exposures; and (3) the work environment involves exposure to methylene chloride, which is
metabolized to CO in the body. Proximity to fuel combustion of all types elevates CO exposure for certain
occupations:  airport employees; auto mechanics; small gasoline-powered tool operators (e.g., users of
chainsaws); charcoal meat  grillers; construction workers; crane deck operators; firefighters;  forklift
operators; parking garage or gasoline station attendants; policemen; taxi, bus, and truck drivers; toll booth
and roadside workers; and warehouse workers (U.S. Environmental Protection Agency, 1991).
       Studies of firefighters are discussed briefly below because these studies were not discussed in  the
previous CO criteria document. Other occupational studies of CO exposure are summarized in Table 4-4,


                                             4-21

-------
                                 Table 4-4. Studies of Occupational Exposures and Dosages3
Occupational Category
Airport workers
Bus drivers
Chainsaw/gasoline tool
operators
Charcoal meat grillers
Firefighters
Forklift operators and
workers in facilities
with forklifts
Garage mechanics
Manufacturing jobs
Traffic/roadway workers
CO Concentration (ppm)
and Averaging Period
5.0-13.6 (0.25 h)
5-300 (0.1-1.7 h) (interior of
vehicle)
5. 8-12.5 TWA (0.5-1.0 h)
NA
>200 (<2 min)
16.2-24.3 TWA (8 h)
NA(3.0-5.6h)
NA
14-20 (0.5-3. 0 + h)
0-105 ppm (inside mask) (0-1.7 h)
14.4 TWA (3. 5 h)
1.2-24.2 (9 h)
250-300 (5 h)
NA (4.4 h)
370-386 (NA)
25-47 TWA (8-12 h)
3-34 (8 h)
42.6%>35(lh)
0-83 TWA (4 h)
2-7 (8 h)
1-4.3 (8 h)
5-42 (2 s)
Measured or Estimated
Percent COHb
NA
NA
NA
9.2-75.6 in 5 farmers
NA
>4 in 10 NS
5.7-7.0 in 56 NS
2.45 in 207 NS
3-7 in 9 NS
NA
NA
NA
5-22 for 4 NS
4.2-28.7 for 7 NS
21.1 ±0.7
6.3-13. 3 for 4 NS
>3.25in5%ofNS
>5in45%ofNS
>3.5in71.4%ofNS
NA
NA
<5
State/Country
Massachusetts, U.S.
U.S.
France
U.S.
U.S.
Germany
Bahrain
Maryland, U.S.
Australia
U.S.
California, U.S.
California, U.S.
North Carolina, U.S.
North Carolina, U.S.
North Carolina, U.S.
Colorado, U.S.
California, U.S.
Ontario, Canada
Seven European
countries
Four states, U.S.
Denmark
Massachusetts, U.S.
References
Bellin and Spengler (1980)
McCammon et al. (1981)
Limasset et al. (1993)
Kahleretal. (1993)
National Institute for Occupational
Safety and Health (1996)
Btingeretal. (1997)
Madanietal. (1992)
Radford and Levine (1976)
Brotherhood etal. (1990)
Jankovic et al. (1991)
Materna etal. (1992, 1993)
National Institute for Occupational
Safety and Health (1994)
Baucometal. (1987)
Fawcett etal. (1992)
Ely etal. (1995)
McCammon et al. (1996)
Apte (1997); Apte et al. (1999)
Gourdeau etal. (1995)
Gardiner etal. (1992)
Boeniger(1995)
Raaschou-Nielsen et al. (1995)
Kamei and Yanagisawa (1997)
aNA = not available, NS = nonsmokers, TWA = time-weighed average.




Source: Adapted from Apte (1997) and updated.

-------
which shows CO concentrations for each study (typical values or ranges), averaging periods, and the
measured or estimated percent COHb levels for nonsmokers, if reported.
Lees (1995) reviewed studies of firefighter exposures to combustion products, including CO. During severe
fires, firefighters were exposed to CO concentrations in excess of 500 ppm in approximately 29% of 1,329
min sampled by Burgess et al. (1977) and in 48% of measurements taken by Barnard and Weber (1979).
Gold et al. (1978) reported a geometric mean concentration of 110 ppm for a log-normal distribution of 65
samples with average duration of less than 10 min. The short-term exposure limit (STEL), designed to
prevent acute effects of CO exposure, is 400 ppm averaged over 15 min. In three studies, the STEL was
exceeded in 15 to 33% of measurements (Treitman et al., 1980; Brandt-Rauf et al., 1988; Jankovic et al.,
1991).  Inside a self-contained breathing apparatus, CO measurements ranged from 1 to 105 ppm in six
samples (Jankovic et al., 1991).
       Firefighters are exposed to lower CO levels when they suppress bushfires, wildland fires, and forest
fires. For bushfires, Brotherhood et al. (1990) estimated that Australian firefighters were exposed to CO
levels averaging 17 ppm, based on COHb measurements taken afterwards. In a study of wildland fires in
California, Materna et al. (1992) reported an average CO level of 14.4 ppm over a 3.5-h period (range 1.4 to
38 ppm) during fireline mop-up and a prescribed burn. Concentrations were higher during evening hours,
when inversions occurred, and could range up to 300 ppm near gasoline-powered  pumping engines.
Materna  et al. (1993) found comparable results using different methods. For forest fires, the National
Institute for Occupational Safety and Health (1994) reported an average CO concentration of 11.5 ppm for
a 9-h period.

4.4.2.2   Exposures to Methylene Chloride in the Workplace
       Certain occupations expose workers to organic solvents such as methylene chloride. The solvent
is widely used as a degreaser, paint remover, aerosol propellant, and blowing agent for polyurethane foams.
It is used as an extractant for foods and spices, a grain fumigant, and a low-pressure refrigerant.  It also is
used in the manufacturing of synthetic fibers, photographic film, polycarbonate plastics, pharmaceuticals,
printed circuit boards, and inks. More than one million workers have significant potential for exposure to
methylene chloride (Agency for Toxic Substances and Disease Registry, 1993).  Moreover, the highest
levels of exposure to methylene chloride often occur in the workplace. To protect worker health, the 8-h
TWA threshold limit value for methylene chloride was set at 50 ppm by the American Conference of
Governmental Industrial Hygienists (1996). Exposure at this concentration leads to COHb levels of about
1.9% in experimental subjects. Exposure to 500 ppm for several hours may elevate COHb levels as high
as 15%.  An 8-h exposure to about 500 mg/m3 (3.5 mg/m3 = 1 ppm) of methylene chloride vapor is
equivalent to an 8-h exposure to 35 ppm of CO (U.S. Environmental Protection Agency, 1985).
       Methylene chloride stored in tissue may continue to metabolize to CO after several hours of acute
exposure. In such cases, COHb levels will continue to rise and peak as high as 25% about 5 to 6 h after
exposure (Agency for Toxic Substances and Disease Registry, 1993). Shusterman et al. (1990) reported
an apparent linear elevation of COHb as a function of hours worked by a furniture refinisher who used paint
stripper containing methylene chloride.  Ghittori  et al.  (1993) reported a significant linear correlation
(correlation coefficient [r] = 0.87) between methylene chloride concentration in air and CO in alveolar air
of nonsmoking and sedentary factory workers in Italy. Exposure to 600 mg/m3 of methylene chloride for
7.5 h was associated with a COHb level of 6.8% in  eight volunteers. Exposure to methylene chloride also
can be fatal.  Leikin et al. (1990) reported fatalities of two people who were exposed to unknown
concentrations of methylene chloride while they removed paint in an  enclosed space.  Death was caused
not by CO poisoning, but by solvent-induced narcosis. Before they died, their COHb levels continued to
rise following cessation of exposure,  despite treatment by high levels  of oxygen.
                                             4-23

-------
4.4.3  Activity Pattern Studies
       In assessing population exposure, studies of human activity patterns over a fixed time period (e.g.,
24 h) are necessary to determine how many people potentially are exposed to sources of an air pollutant,
and how long people spend in activities that involve use of these sources. Accordingly, this section reviews
studies of human activity patterns that pertain to population exposure to CO and methylene chloride.
Studies reviewed in the previous CO criteria document reported that many Americans spent most of their
time indoors at home, school, or work,  etc. (Szalai, 1972; Chapin, 1974; Meyer,  1983; Johnson, 1989;
Schwab et al., 1990).  Although more recent activity pattern studies largely confirm this finding, their
sampling and questionnaire designs provide new insights.  This section reviews, in chronological order,
newer studies that include two surveys of activity patterns in California, a similar survey of preadolescents
in six states, a comparative study between California and the nation, a study in southwest England, a Boston
study, and a recent survey at the national level.  Most of the available  data from activity pattern studies have
been combined into one comprehensive database called the Consolidated Human Activity Database
(CHAD) and containing over 20,000 person-days of 24-h activity (Glen et al.,  1997). The information in
CHAD will be accessible for constructing population cohorts of people with diverse characteristics that are
useful for analysis and modeling (McCurdy, 2000).

4.4.3.1  Activity  Patterns of California Residents
       The California Air Resources Board  (CARB) conducted two surveys to determine the activity
patterns of California residents. In each case, projectable probability samples were drawn from English-
speaking households who had telephones.  The first surveyed 1,762 adults and adolescents over 11 years
of age from fall 1987 through summer 1988 (Wiley et al., 1991a;  Jenkins et al., 1992), and the second
surveyed 1,200 children under age 11 years from April 1989 through February 1990 (Wiley et al., 1991b;
Phillips et al., 1991). Using telephone interviews, both surveys asked participants to complete a 24-h diary
for the preceding day.  People ages 9 years and over responded directly to the interview, and the primary
adult careprovider responded for young children. The diaries enabled estimates of time spent in various
activities and locations, and determinations of whether respondents used or were near sources of pollutants,
including consumer products, combustion appliances, and motor vehicles.
       Similar to previous studies, the results showed that all age groups spent most of their time indoors.
Adults and adolescents in California spent, on average, 87% of their time indoors (62% at home and 25%
elsewhere), and only 6% of their time outdoors. They also spent 7% of their time in transit mostly in a car,
van, or pickup truck. Compared to adults and adolescents, children spent a similar amount of time indoors
(86%), but more time at home (76%) and outdoors (10%), and less  time indoors elsewhere (10%)  and in
transit (4%). About 46% of nonsmoking adults and adolescents reported being near others' tobacco smoke
at some time during the day.
       Table 4-5 summarizes results of the two California surveys for various microenvironments pertinent
to CO exposure. For each microenvironment, the table shows the mean and range in time spent per day by
both the entire sample and by those who actually did  an activity in  the microenvironment (i.e., "doers").
The results show the disparity in mean time spent by the population  and by doers of an activity, which has
implications for calculating population exposure in risk assessment.  Table 4-6 gives the percentage of each
sampled population who  reported use of or proximity to potential sources of either CO or methylene
chloride on a given day.  The study did not measure CO or methylene chloride concentrations from these
sources in microenvironments. Also, the surveys did not indicate  whether respondents lived in a home
where combustion appliances were vented.

4.4.3.2  Activity  Patterns of Children in Six States
       In 1990  and 1991, Silvers et al. (1994) surveyed the activities of preadolescent children (ages  5 to
12 years) from a projectable probability sample of 1,000 households in six states.   These states included


                                             4-24

-------
                        Table 4-5.  Time Spent in Different Microenvironments by Californians, 1987 to 1990
                                                     (minutes per day; weighted)3
Adults/ Adolescents
Microenvironment
Motor Vehicle
Inside a garage
Inside an auto repair shop,
parking garage, or gasoline station
Inside a vehicle:
Car
Van or pickup truck
Bus
Potential Gas Appliance
Kitchen
Utility or laundry room
Basement
Industrial plant or factory
Restaurant
Bar/nightclub
Outdoor Transit
Walking
Bicycle or skates
Motorcycle or scooter
Bus, train, or ride stop
In stroller or carried
Other
Population
Mean

9

11

73
18
4

74
3
<1
35
28
8

10
1
1
NA

1
Percent
Doers

9.4

11.6

74.2
17.5
3.4

75.3
5.0
0.5
8.9
34.5
4.6

26.4
3.1
1.9
NA

1.3
Doer
Mean

97

91

99
102
114

98
53
79
393
81
174

38
41
62
NA

38
Doer
Range

1-845

1-685

1-585
2-785
5-1,320

1-930
3-380
5-180
4-750
1-885
5-825

1-360
5-160
5-430
NA

5-270
Population
Mean

2

<1

43
13
3

47
<1
<1
<1
6
NA

6
1

<1
<1
<1
Children
Percent
Doers

4.1

1

67.1
10.4
7.4

70.0
0.6
<0.1
0.2
12.7
NA

24.7
4.4

2.6
2.0
0.4
Doer
Mean

40

11

65
129
39

66
34
75
34
49
NA

24
22

15
40
72
Doer
Range

2-300

3-47

1-630
1-985
1-134

1-320
5-180
75-75
15-45
3-255
NA

1-195
2-155

3-40
1-195
10-110
NA = not available.
To generalize the results of the survey sample to the entire state population, the data were weighted to correct for nonuniform probabilities of including
 certain individuals in the sample.

Source:  Adapted from Jenkins et al. (1992); Phillips et al. (1991).

-------
           Table 4-6.  Percentage of Californians Who Use or Who Are in Proximity to
           Potential Sources of Either Carbon Monoxide or Methylene Chloride on a
	Given Day, 1987 to 1990 (weighted)3	

                                                   Adults/Adolescents              Children
 Potential Pollutant Source                                 1987 to 88                 1989 to 90

 Consumer Products'"
      Personal care aerosols                                   40                        36
      Scented room fresheners                                 31°                        37
      Solvents                                              12                         3
      Oil-based paints                                         5                         2

 Activities/Places
      Went to a gasoline station, parking garage, or                 26                        11
       auto repair shop
      Pumped gasoline                                       15                         1
      Have attached garage0                                   62                        63
      Had vehicle in attached garage0                            37                        36
      Took a hot showerb                                     77                        26

 Near Combustion Appliances
      Had gas heat on0                                       26                        24
      Had gas oven/range on                                  35                        29

 Nonsmokers Near Environmental Tobacco Smoke at
 Any Time During the Day
      Adults (18 years or older)                                 43
      Youths (12 to 17 years)                                  64
      Adults and youths (12 years and older)                      46
      Youths (0 to 11 years)	38

 aTo generalize the results of the survey sample to the entire state population, the data were weighted to correct for
  nonuniform probabilities of including certain individuals in the sample.
 bPotential methylene chloride exposure.
 °Data presented for adult respondents (age 18 years or older) only.

 Source: Adapted from Jenkins et al. (1992); Phillips et al. (1991).
three on the East Coast (New Jersey, New York, and Pennsylvania)  and three on the West Coast
(California, Oregon, and Washington).  Comparisons between this study known as the Children's Activity
Survey (CAS) and the CARB children's study are possible because both were done over an entire year
at about the same time, and both used a retrospective time diary for a 24-h day. Both studies reported very
similar results in terms of the mean hours per day spent by preadolescent children for locations designated
"indoors" (21.5 h for CARB versus 21.7 h for CAS) and "at home" (18.0 h for CARB versus 17.8 h for
CAS). For each study, these results varied by ±1 h for different seasons of the year.  There was variation
in specific activities (e.g., the CAS study reported that preadolescents spent less time per day "riding in
a vehicle" in California [0.52 h] than they did in the  five other states [0.82 h], when the five were
combined as a group). The CAS study did not report time spent near other CO sources.

4.4.3.3   A Comparative Study Between California and the Nation
       Robinson and Thomas (1991) compared results of activity pattern studies, one conducted by
CARB in California in 1987 and 1988 (Wiley et al., 199la; Jenkins et al., 1992), and the other done at the
national  level in 1985 (Cutler,  1990; Cornish  et al., 1991).   Although the two surveys used different
                                              4-26

-------
methods of gathering and coding data, the data were receded to enable comparisons.  The comparison
showed that Californians averaged more time at work and in commuting to work than was the case
nationally, but averaged less time doing housework and caring for children. California men also spent
more time traveling. The national study appeared to show greater time spent at home and in the yard, but,
these results might be explained by differences in location codes between the two studies, rather than by
actual differences in participant activity patterns. For example, the national study did not ask participants
to identify whether they worked indoors or outdoors.  Because the national study was not designed for
exposure assessment, the authors proposed that the CARB study become a model for a future national
study oriented to exposure assessment.  Such a study is discussed in Section 4.3.3.6.

4.4.3.4  An English Study
       Farrow et al. (1997) studied time spent inside the home from a sample of 170 households in Avon,
England, from November 1990 through June 1993. A pregnant woman lived in each household at the start
of the study. Households completed a weekly diary for 1 year that covered roughly the last 6 mo of the
woman's pregnancy and the first 6 mo of the new infant's life. The results indicated that the average
amount of time spent inside the home per day varied by family member as follows:  mothers,  18.4 h
(76.7%); fathers, 14.7 h (61.3%); and infants, 19.3 h (80.4%).  Infants spent more time at home during
winter than summer. Although fathers spent more time at home on weekends, mothers and infants spent
less time.  The applicability of the study results for U.S. households was not determined, and it is hard to
judge without comparative information about the two countries. However, the study in England indicates
that exposure may be a  function of a parent's gender or household role, supporting a similar conclusion
based on a nationwide study of U.S. activity patterns (see Section 4.3.3.6).

4.4.3.5  A Boston Study of Household Activities, Life Cycle, and Role Allocation
       Using activity diary data from 150 households that participated in a 1991 Boston survey, Vadarevu
and Stopher (1996) tested several hypotheses about household travel. One study hypothesis was that there
are significant differences in mean time allocations of activities among different "life-cycle groups" based
on age,  working status, and household size.  They tested the theory that life-cycle stage affects which
activities fall into mandatory,  flexible,  and optional categories; how much time can be allocated to
different activities; and  which household member does each activity. They found that time allocated by
households to specific activities varied according to whether the household consisted of a single working
adult, multiple adults, a young family, an older family, or a nonworking adult.  However, they found no
significant differences among the life-cycle groups or between any life-cycle group and the population
mean in terms of the total time spent in mandatory activities (work, work-related, school, and certain
at-home activities), which required on average 21 h per day.  The amount of time spent in all flexible,
optional, and travel activities was about 3 h per day.

4.4.3.6  The National  Human Activity Pattern Survey
       The  EPA's National Human Activity Pattern Survey (NHAPS) collected 24-h diary data of
activities and locations provided by 9,386 respondents interviewed nationwide in the United States
between October 1992  and  September 1994 (Klepeis et al., 1996).  To enable projections to a larger
population, the sample was weighted by the 1990 U.S. Census data to account for disproportionate
sampling of certain population groups defined by age and gender.  Results were analyzed across a dozen
subgroups:  gender, age, race, Hispanic, education, employment, census region, day-of-week, season,
asthma, angina, and bronchitis/emphysema. The weighted results showed that, on average, 86.9% of a
person's day was spent  indoors (68.7% at residential locations), 7.2% of the day was spent in or near
vehicles, and 5.9% of the day was spent in outdoor locations.
                                            4-27

-------
       The study also  reported unweighted descriptive statistics and percentiles for both the full
population and various subpopulations (i.e., people who actually did certain activities or who spent time
in certain microenvironments) (Tsang and Klepeis, 1996). Of all respondents, 38.3% reported having a
gas range or oven at home, and 23.7% said that the range/oven had a burning pilot light. In terms of motor
vehicle use, 10% of 6,560 people (7.0% of the total sample) spent more than 175 min per day inside a car,
and 10% of 1,172 people (1.2% of the total sample) spent more than 180 min inside a truck or van. Of
those who were  inside a car and knew they had angina (n = 154 respondents), 10% of them spent more
than 162 min per day inside a car.  The  survey also asked  about sources of household pollutants.
Of  4,723  respondents,  10.5% were exposed to solvents,  10.4% to  open flames,  and 8.4% to
"gasoline-diesel" powered equipment; 6.3% of these respondents were in a garage or indoor parking lot;
and 5.7% reported that someone smoked cigarettes at home. Only  1.8% of 4,663 respondents reported
having a kerosene space heater at home.
4.5  Major Factors Affecting Population Exposure
       This section discusses major factors that have and may continue to affect population exposure to
CO.  These factors include public policies affecting urban transportation planning and air quality, motor
vehicle emissions, and social and technological changes affecting human activity patterns.

4.5.1  Federal Policies Affecting Transportation and Air Quality in Urban Areas
       In the United  States, the national effort to improve air quality can be traced to the CAA
amendments of 1970, 1977, and 1990.  As discussed in Chapter 3, the effect of these CAA amendments
on ambient CO concentrations has been substantial. Moreover, emissions from on-road vehicles have
declined since 1970, even as other socioeconomic indicators of growth have increased. Between 1970
and 1995, nationwide emissions of CO from on-road vehicles fell 33.4% (U.S. Environmental Protection
Agency, 1996), despite compound annual growth rates of 1.0% in the nation's population and 3.2% in
vehicle miles of travel (VMT) during the same period (U.S. Department of Transportation, 1996). The
faster growth rate of VMT can be attributed to many factors that have decentralized housing and jobs
within urban regions since World War n.
       Since the mid-1960s, major construction projects intended to expand highway capacities have
been opposed in some metropolitan areas. Opponents claimed that these proj ects promoted urban sprawl
and induced motor vehicle travel that raised regional air pollutant emissions. To address these concerns,
the 1990 CAA amendments state that transportation actions (plans, programs, and proj ects) cannot create
new NAAQS violations, increase the frequency  or severity of existing NAAQS violations, nor delay
attainment of the NAAQS (U.S. Code, 1990).  Pursuant thereto, the EPA promulgated its Transportation
Conformity Rule. Complementary provisions of the 1991 Intermodal Surface Transportation Efficiency
Act  offered financial incentives  under the Congestion Management and  Air  Quality (CMAQ)
improvement program. Under CMAQ, metropolitan planning organizations were offered federal funds
to improve air quality by implementing transportation control measures (TCMs). Examples of TCMs
include programs to promote car and van pooling, flextime, special lanes for high occupancy vehicles, and
parking restrictions.
       Austin et al. (1994) examined how TCMs have changed travel activity, including number of trips,
vehicle miles of travel, vehicle speed, travel time, and the extent to which commuters have shifted travel
from peak to off-peak periods. Using an emission factors model (i.e., MOBILES [for a description of
MOBILES, see U.S. Environmental Protection Agency,  1999a]), the study inferred how much TCMs
would change average speeds of motor vehicles and CO emissions therefrom. The direct effect of TCMs
on commuter exposure to CO has  received only limited study.  Flachsbart (1989)  found that priority
(with-flow and contra-flow) lanes were effective in reducing exposure  to CO concentrations in motor

                                            4-28

-------
vehicle exhaust on a coastal artery in Honolulu.  Compared to commuter CO exposure in adjacent,
congested lanes, exposure in priority lanes was about 18% less for those in carpools, 28% less for those
in high-occupancy vehicles (e.g., vanpools), and 61% less for those in express buses.  These differences
occurred possibly because commuters in priority lanes traveled faster than those in the congested lanes.
Faster vehicles created more air turbulence, which may have helped to disperse pollutants surrounding
vehicles in priority lanes. Furthermore, these differences existed even though the priority lanes were often
downwind of the congested lanes. Although higher speeds were related to lower exposures in priority
lanes, differences in exposure also could have been caused by differences in vehicle type and ventilation,
both of which were not controlled.
       More recently, Rodes et al. (1998) compared the  CO concentrations of two test vehicles driven
on standardized routes that included a freeway carpool lane in Los Angeles. One vehicle used the carpool
lane and the other the unrestricted lanes. Each vehicle repeatedly drove its route for both the morning and
evening rush hour periods of one day, and CO measurements were taken continuously both inside and
outside of each vehicle. Because the vehicles had different air exchange rates, comparisons of external
CO levels (measured at the base of the windshield) are appropriate.  Based on these measurements, the
average CO concentration of the vehicle in the noncarpool lane (5.6 ppm) was twice as high as that of the
vehicle in the carpool lane (2.8 ppm).  For a hypothetical 48-km commute, exposure in the noncarpool
lane (measured in parts per million-minutes) was estimated to be 187% greater than exposure in the
carpool lane. The study suggests that carpoolers may have lower total CO exposure for their entire
commute, because they are exposed to  lower CO concentrations and spend less time commuting in heavy
traffic.  However, the study did not account for the extra time  to collect nonfamily members of a carpool.
       Models to estimate the direct effects of TCMs on commuter CO exposure are not apparent in the
literature.  However, Flachsbart (1999a) developed a series of statistical models to predict passenger cabin
exposure to CO based on trip variables for a 6.2-km Honolulu artery divided into three links. Based on
data for 80 trips, the most practical models of third-link exposure (adjusted R2 = 0.69) combined three
variables:  (1) the ambient CO concentration; (2) the second-link travel time; and (3) either the travel time,
vehicle speed, or CO emission factor for the third link. The models showed that the vehicle's travel time
and average speed and the CO emission factor for a given  link of the roadway had equal ability to predict
passenger cabin exposure to CO on the third link because of mathematical relationships among these three
predictor variables.

4.5.2  Federal and State Policies Affecting Temporal Trends in Exposure
       Studies show significant decreasing trends in population exposure to CO concentrations from
motor vehicle emissions based on different indicators. One indicator is unintentional death rates from CO
poisoning, and another is based on direct measurements of passenger cabin exposure to CO concentrations
from traffic emissions.  Table 4-7 summarizes data on these indicators from several U.S. studies and
shows  the federal and California tailpipe CO emission standards by model year for comparison.
In Table 4-7, the net mean CO concentration value represents the microenvironmental component of total
exposure.  This value  equals  the mean in-vehicle CO  concentration minus the mean ambient CO
concentration, as recorded simultaneously at a fixed-site monitor.

4.5.2.1 Effects of Motor Vehicle Emission Standards on Unintentional  Death Rates
       Based on death certificate reports compiled by the National Center for Health Statistics, Cobb and
Etzel (1991) reported statistics on the annual rate of unintentional deaths from CO poisoning in the United
States.  As shown in Table 4-7,  the annual death rate per 100,000 population declined from 0.67 in 1979
to 0.39 in 1988.  Motor vehicle exhaust gas accounted for 6,552 deaths or 56.7% of the total 11,547
unintentional deaths occurring  during the 10-year period. The highest death rates per 100,000 persons
occurred among males, blacks,  the elderly, and residents  of northern states.  Monthly variation in death


                                            4-29

-------
      Table 4-7.  Motor Vehicle Carbon Monoxide Emission Standards, Typical In-Vehicle
          Carbon Monoxide Exposures, and Unintentional Carbon Monoxide-Related
                               Death Rates in the United States
New Passenger Car
CO Emission Standard3
Year
Pre -control
=4965
1966
1968
1970
1972
1973
1974
1974-75
1975
1976
1977
1978
1979
1980
1981
1981
1981
1981
1981-82
1982
1982-83
1982-83
1983
1984
1985
1986
1987
1987-88
1988
1989
1990
1991-92
1992

1995
Federal
(g/mi)
84.0
84.0
84.0
51.0
34.0
28.0
28.0
28.0
15.0
15.0
15.0
15.0
15.0
15.0
7.0
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4

3.4
California
(g/mi)
84.0
84.0
51.0
51.0
34.0
34.0
34.0
34.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0

3.4
Net Mean
In-Vehicle CO
Concentration15
(ppm)

12.0
17.5



11.5

7.4



10.3
9.7
8.3
5.2
4.3
2.9
2.9
9.5

1.4
1.8
9.4




4.9
8.4


* 3.6
<3.0


CO Exposure Study
Location

Los Angeles, CA
Five U.S. cities



Los Angeles, CA

Boston, MA



Washington, DC
Los Angeles, CA
Santa Clara Co., CA
Denver, CO
Los Angeles, CA
Phoenix, AZ
Stamford, CT
Honolulu, HI

Denver, CO
Washington, DC
Washington, DC




Los Angeles, CA
Raleigh, NC


Santa Clara Co., CA
New Jersey suburbs of
New York City, NY

U.S. Unintentional
CO-Related Annual
Death Rate per
100,000 Population













0.67
0.55
0.58
0.58
0.58
0.58

0.56


0.53
0.49
0.49
0.44
0.39

0.39






 ""Standards apply at a useful life of 5 years/50,000 miles.
 bMean in-vehicle CO concentration minus mean ambient CO concentration.

 Source: Johnson (1988); Cobb and Etzel (1991); Flachsbart (1995); and Faiz et al. (1996).
rates indicated a seasonal pattern, with January fatalities routinely about two to five times higher than in
July.
                                             4-30

-------
    » Exposure-direct  n Ambient-direct
                         Year
                         * Exposure-indirect
                                    • Ambient-indirect
     1 Los Angeles CA
     2 Chicago, IL; Cincinnati OH: Denver, CO,
      St Louis MO; Washington, DC
     3 14 cities
     A Los Angeles. CA
     5 Boston, MA
     6 Washington, DC
     7 Los Angeles CA
     8 Menlo Park, Palo Alto, and Los Altos. CA
     9 Denver, CO: Los Angeles, CA; Phoenix, AZ;
      Stamford, CT
                            10 Hoi
                                  i, HI
11 a Denver, CO
11 b Washington. DC
12 Washington, DC
13a Los Angeles, CA (summer)
13b Los Angeles, CA (winter)
14 Raleigh, NC
15 Menlo Park Palo Alto, and Los Altos, i
16 New Jersey Turnpike and Route 18, N
       Although mortality is not a health effect used in setting the NAAQS for CO, the Cobb and Etzel
(1991) study  still has  value in its revelations about cofactors of personal exposure  to  high  CO
levels.   Moreover,  the study  speculated  that  declining  death rates  could be attributed in part
to automaker compliance with the motor vehicle CO emission standards of the CAA. The investigators
argued that tighter CO emission standards may enable cars to emit exhaust into an enclosed space for a
longer period of time before CO builds up to toxic levels.

4.5.2.2   Effects  of Motor Vehicle Emission Standards on Passenger Cabin Exposure
       Based on  a review of 16 U.S. studies
that  occurred between  1965  and  1992,
Flachsbart  (1995)  reported  a  long-term,
downward trend in commuter exposure levels
(Table 4-2).  Evidence of this downward
trend appears in Figure 4-7, which shows the
ambient   (lower  line)  and   mean    CO
concentrations inside vehicles (top line) for
these studies.  These lines  do not imply that
CO  concentrations  can be inferred  from
points  on  the lines  themselves, or  that
relationships  exist   between   results   for
different  cities.   Studies  reported typical
(mean or median)  CO concentrations  for
trips, most of which lasted an hour or less.
Mean CO concentrations fell from 37 ppm in
1965, as reported by  Haagen-Smit (1966) for
a study in Los Angeles to 3 ppm in 1992 for
a study by Lawryk et al. (1995) in the New
Jersey suburbs of New York City. If  one
assumes that these results are representative
of typical commuter CO exposures in other
large cities during the same time periods, then exposures fell approximately 90% over this 27-year period.
This reduction implies that CO exposure levels reported in the past for a particular place and time in the
United States may not be indicative of current exposures.
       In the United States, the effect of progressively tighter CO emission standards on in-vehicle CO
exposures over time is readily apparent in Table 4-7.  Prior to 1968, each new passenger car emitted
84g/miof CO, but by the 1981 model year and thereafter, each new car sold outside of California emitted
only 3.4 g/mi of CO, a reduction of 96%  (Johnson, 1988).  This reduction in certified CO emissions for
new passenger cars is roughly the same magnitude as the 90% reduction in commuter exposure reported
above for the  same period. Further analysis reveals that net mean exposure data and the  applicable
emission standard data in Table 4-7 are highly correlated (r = 0.74, p < 0.0005 for a one-tailed test of the
hypothesis). In this analysis, the applicable emission standard (federal or California) was determined by
the location of the exposure study.  Because the exposure studies did not adhere to a standard protocol,
Flachsbart (1995) recommended that future in-vehicle CO exposure studies should use standard protocols
to facilitate  comparisons and to document the effect on exposure of future measures taken under motor
vehicle emission control programs.
       Two of the 16 studies did follow a standard protocol. Ott et al. (1994) measured in-vehicle CO
concentrations on 88 standardized trips over a 1-year period in  1980 and 1981 on a suburban highway
nearSan Jose.  They reported a mean CO concentration of 9.8 ppm for trips of 35 to 45 min. In 1991 and
Figure 4-7.   Trends in  ambient  CO concentrations  and
in-vehicle CO exposures, 1965 to 1992. (The upper and lower
lines are provided to make a  clear distinction  between
exposure and ambient CO data reported for each city; these
lines do not imply that results for cities are related.)
Source:  Flachsbart (1995).	
 4-31

-------
1992, Ott et al. (1993) resurveyed this highway using a methodology similar to their previous study to
determine in-vehicle exposure trends. They reported that the mean in-vehicle CO concentration had
dropped to 4.6 ppm or 47% of the mean value estimated 11 years earlier. They attributed the exposure
reduction to replacement of older vehicles with newer ones that have lower CO emission factors.  This
reduction is particularly significant, as daily traffic volumes on this highway grew by 19.1% during the
intervening period, according to estimates by Yu et al. (1996).
       For this highway, Yu et al. (1996) developed a mathematical model known as the STREET model
to predict trends in CO emissions and exposures.  Based on fleet turnover and no changes in the  1990
California motor vehicle CO emission standards, the model predicted that the median CO concentrations
would drop from 3.9 ppm in 1991 and 1992 to 1.6 to 1.8 ppm in 2002 and 2003. At the 99% percentile,
the model predicted that the CO concentrations would drop from 10 ppm  in 1991 and 1992 to 4.0 to
4.6 ppm in 2002 and 2003.  This prediction was based on an additional expected reduction of up to 60%
in tailpipe emissions of CO, primarily because of continued replacement of older cars with newer,
low-emission vehicles. However, these predictions could be too low because the study did not anticipate
the phenomenal growth in sport utility vehicle (SUV) use in California during the 1990s. The certified
CO emissions of SUVs exceed that of standard passenger cars.
       Similar studies of commuter CO exposure were done by Flachsbart et al. (1987) in the United
States, Koushki et al. (1992) in Saudi  Arabia, Fernandez-Bremauntz and Ashmore (1995a,b) in Mexico,
and Dor et al. (1995) in France. These studies used similar methods of data collection and analysis, with
one exception: smoking was allowed for some trips in the Saudi study, but was not allowed in the other
studies.  Table 4-8 shows typical values of the net mean CO concentration by travel mode for three of the
studies.  The net mean CO concentration for the Saudi study could not be determined.  The net CO
concentrations for each travel mode in Mexico City were much higher than for comparable modes in both
Washington and Paris, where net CO  concentrations were similar.  The similarity between the U.S. and
French studies occurred even though catalytic converters existed on 62% of American cars in 1982 (U.S.
Department of Commerce,  1983) but  were not yet common on French cars in 1992 (Dor et al., 1995).
   Table 4-8.  Typical Net Mean Carbon Monoxide Concentration Ranges by Travel Mode for
                                 Cities in Three Countries3-"
Washington, DC, USA
(1983)

Travel Mode
Automobile
Diesel bus
Rail transit
Net Mean CO
Concentrations
(ppm)
7-12
2-6
0-3
Averaging
Times
(min)
34-69
82-115
27-48
Mexico City, Mexico
(1991)
Net Mean CO
Concentrations
(ppm)
37-47
14-27
9-13
Averaging
Times
(min)
35-63
40-99
39-59
Paris, France
(1991 to 92)
Net Mean CO
Concentrations
(ppm)
7-10
2-3
1
Averaging
Times
(min)
82-106
NA
NA
 a"Typical" means do not include outlier values that can be attributed to unusual circumstances.
 bNet mean CO concentration = mean in-vehicle CO concentration minus mean ambient CO concentration.
 NA = not available.

 Source: Adapted from Flachsbart (1999b).
                                            4-32

-------
       The reasons for the similarity in results between the U.S. and French studies are not readily
apparent. However, passenger cabin exposure levels in North and Central America can be explained
partly by comparing the history of automotive emission standards in the United States and Mexico. The
United States initiated nationwide emission standards on new passenger cars in  1968  and adopted
progressively tighter controls throughout the 1970s (Johnson, 1988). By the 1975 model year, catalytic
converters became standard equipment on new passenger cars. Mexico adopted a tailpipe CO emission
standard of 47.0 g/mi for the 1975 model year, and, by the 1993 model year, Mexico finally reached parity
with the 1981 U.S. standard of 3.4 g/mi (Faiz et al, 1996).

4.5.3  Social Changes Affecting Human Activity Patterns
       Between  1965 and 1985, the Americans' Use of Time Project at the University  of Maryland
reported that the average time spent in travel for leisure trips increased from 2.7  to 3.1 h per week
(Cornish et al., 1991).  In contrast, there is evidence that average commuting times between home and
work have remained stable. The decennial census collected travel time data for the first time in 1980.
By 1990, the census showed that the nation's average commuting time of 21.7 min in 1980 increased only
40 s to 22.4 min in 1990. Although  the number of workers who commuted 45 min or more increased
from 10.9 million in 1980 to 13.9 million in 1990, the mean travel time of this commuter cohort actually
decreased slightly from 59.6 min in  1980 to  58.5 min in 1990. One reason for this is that more people
were taking their morning commute from home to work during the "shoulder hours"  from 6 to 7 a.m. or
from 8 to 9 a.m. than during the "peak hour"  from 7 to 8 a.m. In 1990, the shoulder hours accounted for
about 37% of worker trip starts, whereas the "peak hour"  accounted for only 32% of all trip starts
(Pisarski, 1992).
       Typically, average  commuting times in large metropolitan  regions are greater than those
nationwide.  In the Washington metropolitan region, for example, the average daily commuting time for
all modes of travel between home and work was 62 min per day in  1957 (Bello, 1958).  This value
increased to 69 min per day by 1968  and fell  slightly to 68.3 min in 1987 and 1988 when the region was
resurveyed.  The stability in daily commuting times between the 1968 and the 1987 and 1988 surveys was
achieved by  an increase  in travel speed. This increase in travel speed offset greater travel distances
between home and work that occurred during that 20-year period. For those who commuted  alone by
automobile in Washington, the average euclidean, round-trip travel distance from home to work increased
from 13.8 miles in 1968 to 16.1 miles in 1987 and 1988. However, the average trip speeds of solo
commuters  in  Washington also increased  (by 10.7% for  home-to-work trips and by 20.3% for
work-to-home trips) to offset increased commuting distance (Levinson and Kumar, 1994). This increase
in trip speeds is significant because passenger cabin exposure to CO concentrations  has been shown in
a separate study to be inversely related to travel speed in the Washington area (Flachsbart et al., 1987).
       In another study of the Washington area based on the same data and time period, Levinson and
Kumar (1995) observed an 85% overall increase in the number of jobs and a decline in average household
size from 3.34 to 2.67 people.   During the 20-year period from 1968 to 1988, vehicle registrations
increased 118%,  but road capacity increased only  13%. The average number of autos per household
increased from 1.6to2.0. However, the most important change was a higher percentage of women in the
work  force,  which forced readjustments and  reallocations of time spent in household activities.
Specifically, workers had more per capita income but spent less time at home and engaged in more travel
for nonwork trips during peak travel periods.  Compared to 1968, working men spent  20 min less time at
home in 1988, and working women spent about 40 min less.  Commuters made multiple stops (i.e., trip
chaining) on their way home from work (e.g., visiting health clubs, picking up children at day-care,
shopping, eating at restaurants).  In 1968, such errands and activities usually were done after the primary
worker returned home with the household car. By 1988, these trips often were made in separate vehicles
by each household member on their way home from work. By 1988, average time spent daily in travel

                                            4-33

-------
per person in the Washington area had increased by 14 min for workers and by 11 min for nonworkers
over 1968.   Levinson and Kumar (1995) said that these results do not support the hypothesis "that
individuals spend a fixed amount of time per day (just over 1 h) in transportation, and make all budget
allocation adjustment on non-travel times."  Instead, the investigators suggested that  some urban
households have been spending more time in travel and less time at home, and have been buying more
household services outside the home.
       On the other hand, Levinson and Kumar (1995) anticipate that some people will spend more time
at home in the future.  They noted:  "Several factors suggest that work at home, telecommuting, and
teleshopping may be on the verge of wide-spread adoption. The technology is coming into place with the
long-awaited advent of videophones, and of the 'information superhighway', that is, broad-band two-way
communications facilitated by the recent consolidations in the telecommunications and entertainment
industries."  The percentage of people working at  home increased from 2.3% in 1980 to 3% in 1990
(Pisarski, 1992). Currently, an estimated 52 million Americans are self-employed to some extent, working
either in home offices for themselves or for companies as telecommuters.  In 1975,  only 2.5 million
Americans worked at home. In 1994, there were fewer than 4,000 telecommuters working for the federal
government and Webster (1998) reported that the U.S. General Services Administration expected to see
60,000 such  workers by the end of 1998. This employment shift could have beneficial implications for
reduction in population CO exposure.
4.6  Conclusions
       This chapter has reviewed studies of population exposure to carbon monoxide, including some
key studies from the previous CO criteria document and studies that have been published in the
peer-reviewed scientific literature since 1991. This section draws several conclusions from this review,
and identifies both the extent to which CO exposures have changed since the previous criteria document
and some current gaps in knowledge about population exposure to  CO.  The previous CO criteria
document concluded that, on an individual basis, personal exposure is poorly correlated with ambient CO
concentrations as measured at fixed-site monitors, because of personal mobility relative to the monitor's
fixed location, and the spatial and temporal variability of CO concentrations (U.S. Environmental
Protection Agency, 1991). Like earlier studies, more recent ones indicate that the extent and magnitude
of observed personal CO exposures may be greater than those predicted from fixed-site monitors used to
determine compliance with the NAAQS (Wilson etal, 1993a,b; Colomeetal, 1994). Yet, when ambient
CO levels are either high or low on a given day, fixed-site monitors still reflect the corresponding high
or low aggregates of personal exposures on those days.  Otherwise, the stations do not adequately
represent the CO exposures of community residents while they are exposed to motor vehicle exhaust
during commuting, to occupational and residential sources of unvented fuel combustion, or to tobacco
smoke.  The mean COHb level of people exposed to CO from these sources will be greater than their
mean COHb level predicted solely from exposure to CO of ambient origin.
       Implementation of motor  vehicle emission  standards, catalytic converters, motor vehicle
inspection and maintenance programs, and cleaner burning fuels during the past three decades has reduced
the CO exposures of urban commuters (Flachsbart,  1995,  1999b).   This conclusion has important
implications, because it suggests that modeled estimates of current commuter exposure, based on data
inputs from pre-1990 exposure studies, may be too high.  Moreover, the Yu et al. (1996) study  indicates
that the average CO concentrations in passenger cabins of motor vehicles are expected to drop further in
the near future.  However, those projections could be too low because the study did not anticipate or
account for the phenomenal growth in SUV use during the 1990s. The certified CO emissions of SUVs
presently exceed those of standard passenger cars, and EPA did not propose tighter emission standards
for SUVs until 1999. As compared to the federal (Tier 1)3.4 g/mi emission standards shown in Table 4-7,

                                            4-34

-------
the corresponding emission standards for SUVs (3,750 to 5,750 Ibs) are 4.4 g/mi (Federal Register, 2000;
U.S. Environmental Protection Agency,  1999b).  Hence, new studies using standard protocols of
in-vehicle CO exposure would appear to be needed, not only to assess current commuter exposure, but
also to enable comparisons with past studies and projections of future exposure.
       Likewise, there are uncertainties over the extent to which population exposure to CO has changed
in other ways since the previous criteria document. First, there are no published trend  studies of CO
exposure in other important microenvironments (e.g., indoor parking garages, pedestrian sidewalks on
commercial streets, or home environments affected by greater use of microwaves for cooking in lieu of
gas ranges).  Second, the net  effect of various travel  behavior trends on  commuter CO exposure is
uncertain. Trends noted in this chapter include disproportionately high growth rates in vehicle miles of
travel, growth in travel during shoulder hours of peak-traffic periods, and growing use of personal
computers for telecommuting and teleshopping from home in lieu of trips by motor vehicles. These trends
and their implications for exposure suggest that the results of earlier personal CO monitoring studies, such
as those of Akland et al. (1985) summarized in Table 4-1, are probably no longer indicative of present CO
levels and population exposures. These types of personal exposure studies  would need to be redone to
determine current CO  exposure levels of similar urban populations.
       The previous CO  criteria document reported that people are exposed to elevated CO levels in
certain indoor microenvironments (e.g., unventilated parking garages, motor vehicles with leaky exhaust
systems, small homes  with unvented gas stoves and space heaters).  More  recent studies in California
homes indicate that elevated CO concentrations (>9 ppm) still exist and can be caused by several factors,
such as attached garages  and carports; ranges with continuous gas pilot lights; and improper use and
installation of gas appliances, especially in small homes (Wilson et al., 1993a,b; Colome et al., 1994).
Also, recent studies have found elevated CO concentrations (>9 ppm) when people ride certain types of
recreational vehicles (i.e., snowmobiles, powerboats), gather indoors to barbecue food (sometimes to cope
with electrical power outages), and watch  sporting events held at indoor arenas.  High-level exposures
(>25 ppm) may occur inside arenas when they are used for ice skating or motocross, monster-truck, and
tractor pull competitions.  Vehicles used in these competitions often lack any type of emission controls.
In some cases, ventilation alone has not lowered CO sufficiently to safe levels (<9 ppm) at these events.
Moreover, recent studies  report that high-level CO exposures can occur when people use unregulated
gasoline-powered appliances, engines, and tools (e.g., chainsaws), even under ventilated  conditions.
       The previous CO criteria document reported that many Americans spend most of their time
indoors.  This finding still appears to be true, according to more recent studies of activity patterns in
California and nationwide (e.g., the NHAPS study). These studies indicate  that Americans now spend,
on average, between 87 and 89% of their day indoors and about 7% of their time  in or  near vehicles.
However, activity patterns have shifted since the pioneering studies of the early 1970s. Recent study of
travel behavior in Washington indicates that some people are spending relatively more time in travel and
less time at home, compared with the past, because of growth in the service sector of the nation's
economy.  This growth has enabled a growing number of people to buy services outside the home that
once were provided by household members.  Buyers often visit one or more retailers (i.e., trip chaining)
as part of the daily commute to and from  work (Levinson and Kumar, 1995), which extends their trip
times.
       In analyzing activity patterns, the California and national studies both analyzed activities by
conventional  social categories (e.g., gender, age, race, etc.).  In a travel behavior study, Vadarevu and
Stopher (1996) used a different social construct that revealed significant differences in activity patterns
among different life cycle groups defined by age, working status, and household size. Hence, although
their study did not focus on CO exposure assessment per se, it still has useful lessons for the design and
analysis of activity pattern studies.  Their  study also revealed that people who are part of a household
make continuous tradeoffs in their activity  patterns and household role allocations (in terms of who does

                                            4-35

-------
what and when) in response to ongoing social and technological changes.  However, the effects of these
activity and role adjustments by householders on personal CO exposure still needs to be documented by
empirical study.
        In light of the above, population exposure models (e.g., pNEM/CO, SHAPE) may need to sample
from distributions that more accurately represent current microenvironmental CO concentrations and time
budgets and add certain high-exposure level microenvironments (e.g., tobacco smoke exposure while in
a vehicle, sporting events involving motor vehicles at indoor arenas) to their current list.  In the future,
simulation  models  of exposure should consider that trip  times  and commuter exposures  are  not
independent of trip-starting times, and that the distribution of CO exposure is not homogeneous for all
types of commuters. As evidence of the former, Flachsbart (1999a) showed that a commuter's travel time
and CO exposure inside a passenger car for a trip from home to work was  related to trip departure time.
Not surprisingly, travel during off-peak hours (i.e., shoulder periods) to avoid congested traffic resulted
in both shorter travel time and less CO exposure.  As evidence of the latter, carpoolers can reduce their
CO exposure if they use high-occupancy vehicle lanes on highways,  as shown by both Flachsbart (1989)
in Hawaii and by Rodes et  al. (1998) in Southern California.
References
Agency for Toxic Substances and Disease Registry. (1993) Methylene chloride toxicity. Am. Fam. Phy. 47: 1159-1166.
Akland, G. G.; Hartwell, T. D.; Johnson, T. R.; Whitmore, R. W. (1985) Measuring human exposure to carbon monoxide in
      Washington, B.C., and Denver, Colorado, during the winter of 1982-1983. Environ. Sci. Technol. 19:911-918.
Aim, S.; Jantunen, M. I; Vartiainen, M. (1999) Urban commuter exposure to particle matter and carbon monoxide inside an
      automobile. J. Exposure Anal. Environ. Epidemiol. 9: 237-244.
American Conference of Governmental Industrial Hygienists.  (1996) Threshold limit values for chemical substances and
      physical agents: biological exposure indices. Cincinnati, OH: American Conference of Governmental and Industrial
      Hygienists.
Amiro, A. G. (1969) Carbon monoxide presents public health problem. Presented at: 33rd annual educational conference,
      National Association of Sanitarians; June; Houston, TX. J. Environ. Health 32: 83-88.
Apte, M. G. (1997) A population-based exposure assessment methodology for carbon monoxide: development of a carbon
      monoxide passive sampler and occupational dosimeter [dissertation]. Berkeley, CA:  U.S. Department of  Energy,
      Lawrence Berkeley National Laboratory, Environmental Energy Technologies Division; LBNL-40838.
Apte, M. G.; Cox, D. D.; Hammond, S. K.; Gundel, L. A. (1999) A new carbon monoxide occupational dosimeter: results from
      a worker exposure assessment survey. J. Exposure Anal. Environ. Epidemiol. 9: 546-559.
Austin, B. S.; Heiken, J. G.; Shepard, S. B.; Duvall, L. L. (1994) Methodologies for estimating emission and travel activity
      effects of TCMs. Washington, DC: U.S.  Environmental Protection Agency, Office of Mobile Sources; report no.
      EPA-420-R-94-002.
Barnard, R. J.; Weber, J. S.  (1979) Carbon monoxide: a hazard to firefighters. Arch. Environ. Health 34: 255-257.
Baucom, C. D.; Freeman, J. I; MacCormack, J. M. (1987) Carbon monoxide poisoning in a garment-manufacturing plant -
      North Carolina. Morb. Mortal. Wkly. Rep. 36: 543-545.
Bellin, P.; Spengler, J. D. (1980) Indoor and outdoor carbon monoxide measurements at an airport. J. Air Pollut. Control Assoc.
      30: 392-394.
Bello, F. (1958) The city and the car. In: The exploding metropolis. Garden City, NY: Doubleday Anchor Books; pp. 32-61.
Bevan, M. A. J.; Proctor, C. J.; Baker-Rogers, J.; Warren, N. D. (1991) Exposure to carbon monoxide, respirable suspended
      particulates, and volatile organic compounds while commuting by bicycle. Environ. Sci.  Technol. 25: 788-791.
Boeniger, M. F. (1995) Description of facilities and employee exposure to carbon monoxide at four different bridge and tunnel
      toll authorities in the United States. Cincinnati, OH: National Institute for Occupational Safety and Health, Division of
      Surveillance, Field Studies, Hazard Evaluations; report no. IWS-45-02.
Boudreau,  D. R.; Spadafora, M.  P.; Wolf, L.  R.; Siegel, E. (1994) Carbon monoxide levels during indoor sporting
      events—Cincinnati, 1992-1993 [reprint from MMWR 43: 21-23]. JAMA J. Am. Med. Assoc. 271: 419.
Brandt-Rauf, P. W.; Fallon, L. F., Jr.; Tarantini, T.; Idema, C.; Andrews, L. (1988) Health hazards of fire fighters: exposure
      assessment. Br. J. Ind. Med. 45: 605-612.
Brotherhood, J. R.; Budd, G. M.; Jeffery, S. E.; Hendrie, A. L.; Beasley, F. A.; Costin, B.  P.; Wu, Z. (1990) Fire fighters'
      exposure to carbon monoxide during Australian bushfires. Am. Ind. Hyg. Assoc. J. 51: 234-240.
Buie, S. E.; Pratt, D. S.; May, J. J. (1986) Diffuse pulmonary injury following paint remover exposure. Am. J. Med. 81: 702-704.
                                                  4-36

-------
Hunger, I; Bombosch, F.; Mesecke, U.; Hallier, E. (1997) Monitoring and analysis of occupational exposure to chain saw
      exhausts. Am. Ind. Hyg. Assoc.  J. 58: 747-751.
Burgess, W. A.; Sidor, R.; Lynch, J.  J.; Buchanan, P.; Clougherty, E. (1977) Minimum protection factors for respiratory
      protective devices for firefighters. Am. Ind. Hyg. Assoc. J. 38: 18-23.
Chan, C.-C.; Ozkaynak, H.; Spengler, J. D.; Sheldon, L. (1991) Driver exposure to volatile organic compounds, CO, ozone and
      NO2 under different driving conditions. Environ. Sci. Technol. 25: 964-972.
Chapin, F. S., Jr. (1974) Human activity patterns in the city: things people do in time and in space. New York, NY: John Wiley
      & Sons.
Chung, Y.; Park, S. E.; Lee, K.; Yanagisawa, Y.; Spengler, J. D. (1994) Determinations of personal carbon monoxide exposure
      and blood carboxyhemoglobin levels in Korea. Yonsei Med. J. 35: 420-428.
Clements, J.  S. (1978) School bus carbon monoxide intrusion. Washington, DC: U.S. Department of Transportation, National
      Highway Traffic Safety Administration; report no. DOT-HS-803 705.
Cobb, N; Etzel, R. A. (1991) Unintentional carbon monoxide-related deaths in the United States, 1979 through 1988. JAMA
      J. Am. Med. Assoc. 266: 659-663.
Colome, S. D.; Wilson, A. L.; Tian, Y.  (1994) California residential indoor air quality study. Volume 2. Carbon monoxide and
      air exchange rate: an univariate and multivariate analysis. Chicago, IL:  Gas Research Institute; report no. GRI-93/0224.3.
Cornish, E.; Willard, T.; Fields, D. (1991) How Americans use time: an interview with sociologist John P. Robinson. Futurist
      Sept./Oct: 23-27.
Cortese, A. D.; Spengler, J. D. (1976)  Ability of fixed monitoring stations to represent personal carbon monoxide exposure.
      J. AirPollut. Control Assoc. 26: 1144-1150.
Cutler, B. (1990) Where does the free time go? Am. Demograph. (November): 36-39.
Dor, F.; Le Moullec, Y.; Festy, B.  (1995) Exposure of city residents to carbon monoxide  and monocyclic aromatic hydrocarbons
      during commuting trips in the Paris metropolitan area. J. Air Waste Manage. Assoc. 45: 103-110.
Duan, N. (1982) Models for human exposure to air pollution. Environ. Int. 8: 305-309.
Ely, E. W.; Moorehead, B.; Haponik, E. F. (1995) Warehouse workers' headache: emergency evaluation and management of
      30 patients with carbon monoxide poisoning. Am. J. Med. 98: 145-155.
Faiz, A.; Weaver, C. S.; Walsh, M. P.; Gautam, S. P.; Chan, L.-M. (1996) Emission standards and regulations. In: Air pollution
      from motor vehicles: standards and technologies for controlling emissions. Washington, DC: World Bank; pp. 1-24.
Farrow, A.; Taylor, H.; Golding, J. (1997) Time spent in the home by different family members. Environ. Technol.  18: 605-613.
Fawcett, T. A.; Moon, R. E.; Fracica, P. J.; Mebane, G. Y.; Theil, D. R.; Piantadosi, C. A. (1992) Warehouse workers' headache:
      carbon monoxide poisoning from propane-fueled forklifts. J. Occup.  Med. 34: 12-15.
Federal Register. (2000) Control of air pollution from new motor vehicles: tier 2 motor vehicle emissions standards and gasoline
      sulfur control requirements; final rule. F. R. 65 (February  10): 6697-6870.
Fernandez-Bremauntz, A.  A.;  Ashmore, M.  R.  (1995a)  Exposure of commuters  to  carbon monoxide  in Mexico
      City—I. Measurement of in-vehicle concentrations. Atmos. Environ.  29: 525-532.
Fernandez-Bremauntz, A.  A.; Ashmore, M.  R. (1995b) Exposure of commuters  to carbon monoxide in Mexico City
      II. Comparison of in-vehicle and fixed-site concentrations. J. Exposure Anal. Environ. Epidemiol. 5:  497-510.
Flachsbart, P. G. (1989) Effectiveness of priority  lanes in reducing travel time and carbon monoxide exposure. Inst. Transport.
      Eng. J. 59: 41-45.
Flachsbart, P. G. (1995) Long-term trends in United States highway emissions, ambient concentrations, and in-vehicle exposure
      to carbon monoxide in traffic. J. Exposure Anal. Environ. Epidemiol. 5: 473-495.
Flachsbart, P. G. (1999a) Models of exposure to carbon monoxide inside a vehicle on a Honolulu highway. J. Exposure Anal.
      Environ. Epidemiol.  9: 245-260.
Flachsbart, P. G. (1999b) Human exposure to carbon monoxide from mobile  sources. In: Khalil, M. A. K.; Pinto, J. P.; Shearer,
      M. J.,eds. Carbon monoxide [special issue of papers from an international conference; December 1997; Portland, OR].
      Chemosphere Global Sci.  Change 1: 301-329.
Flachsbart, P. G.; Brown, D. E. (1989) Employee exposure to motor vehicle exhaust at a Honolulu  shopping center. J. Architect.
      Plan. Res. 6: 19-33.
Flachsbart, P. G.;  Mack, G.  A.; Howes, J. E.; Rodes, C. E. (1987) Carbon  monoxide exposures of Washington commuters.
      JAPCA37: 135-142.
Gardiner, K.; Trethowan, W. N.; Harrington, J. M.; Calvert, I. A.; Glass,  D. C.  (1992) Occupational exposure to carbon
      monoxide and sulphur dioxide during the manufacture of carbon black. Ann. Occup. Hyg. 36: 363-372.
Gasman, J. D.; Varon, J.; Gardner, J.  P. (1990)  Revenge of the barbecue grill: carbon monoxide poisoning. West. J. Med.
      153:656-657.
Ghittori, S.; Marraccini, P.; Franco, G.; Imbriani, M. (1993) Methylene chloride exposure in industrial workers. Am. Ind. Hyg.
      Assoc. J. 54:27-31.
Girman, J. R.; Chang, Y. -L.; Hayward, S. B.; Liu, K. -S. (1998) Causes of unintentional deaths from carbon monoxide poisonings
      in California. West. J. Med. 168: 158-165.
                                                     4-37

-------
Glen, G.; Lakkadi, Y.; Tippett, J. A.; del Valle-Torres, M. (1997) Development of NERL/CHAD: the National Exposure
      Research Laboratory consolidated human activity database. Research Triangle Park, NC: U. S. Environmental Protection
      Agency, Office of Research and Development; contract no. 68-D5-0049.
Gold, A.; Burgess, W. A.; Clougherty, E. V. (1978) Exposure of firefighters to toxic air contaminants. Am. Ind. Hyg. Assoc. J.
      39: 534-539.
Gourdeau, P.; Parent, M.; Soulard, A. (1995) Exposition a 1'oxyde de carbone dans les garages d'automobiles: evaluation chez
      les mecaniciens [Exposure of carbon monoxide in car repair shops:  evaluation of mechanics]. Can. J. Public Health
      86:414-417.
Haagen-Smit, A.  J. (1966) Carbon monoxide levels in city driving. Arch. Environ. Health 12: 548-551.
Hampson, N. B. (1996) Carbon monoxide poisoning at an indoor ice arena and bingo hall—Seattle, 1996 [reprint of MMWR
      1996; 45: 265-267]. JAMA J. Am. Med. Assoc. 275:  1468-1469.
Hampson, N. B.; Norkool, D. M. (1992) Carbon monoxide poisoning in children riding in the back of pickup trucks. JAMA
      J. Am. Med. Assoc. 267: 538-540.
Hampson, N. B.; Kramer, C. C.; Dunford, R. G.; Norkool, D. M. (1994) Carbon monoxide poisoning from indoor burning of
      charcoal briquets. JAMA J. Am. Med. Assoc. 271: 52-53.
Holland, D. M. (1983) Carbon monoxide levels in microenvironment types of four U.S. cities. Environ. Int. 9: 369-377.
Jankovic, J.; Jones, W.; Burkhart, J.; Noonan, G. (1991) Environmental study of firefighters.  Ann. Occup. Hyg. 35: 581-602.
Jarvis, M. J.; Russell, M. A. H.; Saloojee, Y. (1980) Expired air carbon monoxide: a simple breath test of tobacco smoke intake.
      Br. Med. J. 281:484-485.
Jenkins,?. L.; Phillips, T. J.; Mulberg, E. J.;Hui, S. P. (1992) Activity patterns of Californians: use of and proximity to indoor
      pollutant sources. Atmos. Environ. Part A 26: 2141-2148.
Johnson, T. (1984) A study of personal exposure to carbon monoxide in Denver, Colorado. Research Triangle Park, NC: U.S.
      Environmental Protection Agency, Environmental Monitoring Systems Laboratory; report no. EPA-600/4-84-014.
Johnson, J. H. (1988) Automotive emissions. In: Watson, A.  Y.; Bates, R. R.; Kennedy, D., eds. Air pollution, the automobile,
      and public health. Washington, DC: National Academy Press; pp. 39-75.
Johnson, T. (1989) Human activity patterns in Cincinnati, Ohio [final report]. Palo Alto, CA: Electric Power Research Institute;
      project no. RP940-06; report no. EN-6204.
Johnson, T.; Paul, R. A. (1983) The NAAQS Exposure Model (NEM) applied to carbon monoxide. Research Triangle Park, NC:
      U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards; report no. EPA-450/5-83-003.
Johnson, T.;  Capel, J.; Paul,  R.;  Wijnberg, L. (1992)  Estimation  of carbon monoxide  exposures  and associated
      carboxyhemoglobin levels in Denver residents using a probabilistic version of NEM. Research Triangle Park, NC: U.S.
      Environmental Protection Agency, Office of Air Quality Planning and Standards; contract no. 68-DO-0062.
Johnson, T.; Mihlan, G.; LaPointe, J.; Fletcher, K.; Capel, J.  (1999) Estimation of carbon monoxide exposures and associated
      carboxyhemoglobin levels in Denver residents using pNEM/CO (version 2.0) [draft]. Research Triangle Park, NC: U. S.
      Environmental Protection Agency, Office of Air Quality Planning and Standards; March 15.
Jones, R. H.; Ellicott,  M. F.; Cadigan, J. B.; Gaensler, E.  A. (1958) The relationship between alveolar and blood carbon
      monoxide concentrations during breathholding. J. Lab. Clin. Med. 51: 553-564.
Joumard, R. (1991) Pollution de 1'air dans les transports [Air pollution from transportation]. Pollut. Atmos. 33: 20-24.
Kahler, M.; Kuhse, W.; Wintermeyer, L. A. (1993) Unintentional carbon monoxide poisoning from indoor use  of pressure
      washers—Iowa, January 1992-January 1993. Morb. Mortal.  Wkly. Rep. 42: 777-779, 785.
Kamei, M.; Yanagisawa, Y. (1997) Estimation of CO exposure of road construction workers in tunnel. Ind. Health 35: 119-125.
Klepeis,N. E.; Tsang, A. M.;Behar, J. V. (1996) Analysis of the national human activity pattern survey (NHAPS) respondents
      from a standpoint of exposure assessment. Washington, DC: U. S. Environmental Protection Agency, Office of Research
      and Development; report no. EPA/600/R-96/074.
Klepeis, N. E.; Ott, W. R.; Repace, J. L. (1999) The effect of cigar smoking on indoor levels of carbon monoxide and particles.
      J. Exposure Anal. Environ. Epidemiol. 9: 622-635.
Koontz, M. D.; Nagda, N. L. (1988) A topical report on a field monitoring study of homes with unvented gas space heaters.
      Volume III—methodology and results. Chicago, IL: Gas Research Institute; report no.  GRI-87/0044.2.
Koushki, P. A.; Al-Dhowalia, K. H.; Niaizi, S. A. (1992) Vehicle occupant exposure to carbon monoxide. J. Air Waste Manage.
      Assoc. 42: 1603-1608.
Kwok, P. W. (1981) Reduction of carbon monoxide in indoor skating arenas. Environ. Health Rev. 25: 60-63.
Lando, H. A.; McGovern, P. G.; Kelder, S. H.; Jeffery, R. W.;  Forster, J. L. (1991) Use of carbon monoxide breath validation
      in assessing exposure to cigarette smoke in a worksite population. Health Psychol. 10: 296-301.
Last, J. M.; Abramson, J. H.; Friedman, G. D.; Porta, M.; Spasoff,  R. A.; Thuriaux, M. (1995) A dictionary of epidemiology.
      3rd ed. New York, NY:  Oxford University Press; pp.  5, 60.
Law, P. L.; Lioy, P. J.; Zelenka,  M. P.; Huber, A. H.; McCurdy, T. R. (1997) Evaluation of a probabilistic exposure model
      applied to carbon monoxide (pNEM/CO) using Denver personal exposure monitoring data. J. Air Waste Manage. Assoc.
      47: 491-500.
                                                     4-38

-------
Lawryk, N. J.; Lioy, P. J.; Weisel, C. P. (1995) Exposure to volatile organic compounds in the passenger compartment of
      automobiles during periods of normal and malfunctioning operation. J. Exposure Anal. Environ. Epidemiol. 5:511-531.
Lee, K.; Yanagisawa, Y.  (1992) Development of a sampler for carbon monoxide in expired air. Presented at: 85th annual
      meeting and exhibition of the Air & Waste Management Association; June; Kansas City, MO. Pittsburgh, PA: Air &
      Waste Management Association; paper no. 92-145.03.
Lee, K.; Yanagisawa, Y. (1995) Sampler for measurement of alveolar carbon monoxide. Environ. Sci. Technol. 29: 104-107.
Lee, K.; Yanagisawa, Y.; Spengler, J. D. (1993) Carbon monoxide and nitrogen dioxide levels in an indoor ice skating rink with
      mitigation methods. Air Waste 43: 769-771.
Lee, K.; Yanagisawa, Y.;  Spengler, J. D.; Nakai, S. (1994) Carbon monoxide and nitrogen dioxide exposures in indoor ice
      skating rinks. J. Sports Sci. 12: 279-283.
Lees, P. S. J. (1995) Combustion products and other firefighter exposures. Occup. Med.: State of the Art Rev.  10: 691-706.
Leikin, J. B.; Kaufman, D.; Lipscomb, J. W.; Burda, A. M.; Hryhorczuk, D. O. (1990) Methylene  chloride: report of five
      exposures and two deaths. Am. J. Emerg. Med. 8: 534-537.
Levesque, B.; Dewailly, E.; Lavoie, R.; Prud'Homme, D.; Allaire, S. (1990) Carbon monoxide in indoor ice skating rinks:
      evaluation of absorption by adult hockey players. Am. J. Public Health 80: 594-598.
Levesque, B.; Lavoie, R.; Dewailly, E.; Prud'Homme, D.; Allaire, S. (1991) An experiment to evaluate carbon monoxide
      absorption by hockey players in ice skating rinks. Vet. Hum. Toxicol. 33: 5-8.
Levesque, B.; Allaire, S.; Prud'homme, H.; Rhainds, M.; Lebel,  G.; Bellemarre, D.; Dupuis,  K. (1997) Indoor motocross
      competitions: air quality evaluation. Am. Ind. Hyg. Assoc. J. 58: 286-290.
Levinson, D. M.; Kumar, A. (1994) The rational locator: why  travel times  have remained  stable. J.  Am. Plan. Assoc.
      60:319-332.
Levinson, D.; Kumar, A. (1995) Activity, travel, and the allocation of time. J. Am. Plan. Assoc.  61: 458-470.
Limasset, J.-C.; Diebold,  F.; Hubert, G. (1993) Exposition des conducteurs de bus urbains aux polluants de la circulation
      automobile [Assessment of bus drivers' exposure to the pollutants of urban traffic]. Sci. Total Environ. 134:  39-49.
Lioy, P. J. (1990) Assessing total human exposure to contaminants: a multidisciplinary approach.  Environ. Sci. Technol.
      24: 938-945.
Liu, K.-S.; Girman, J. R.; Hayward, S. B.; Shusterman, D.; Chang, Y.-L. (1993) Unintentional carbon monoxide deaths in
      California from charcoal grills and hibachis. J. Exposure Anal. Environ. Epidemiol. 3(suppl. 1): 143-151.
Luckhurst, D. G.; Solkoski, G. (1990) Carbon monoxide levels in indoor "tractor-pull" events—Manitoba. Can. Dis. Wkly. Rep.
      16-17:79-81.
Madani, I. M.; Khalfan, S.; Khalfan, H.; Jidah, J.; Aladin, M. N. (1992) Occupational exposure to carbon monoxide during
      charcoal meat grilling. Sci. Total Environ. 114: 141-147.
Mage, D. T. (1991) A comparison of the direct and indirect methods of human exposure. In: Gledhill, B. L.; Mauro, F., eds. New
      horizons in biological dosimetry: proceedings of the international symposium on trends in biological  dosimetry; October
      1990; Lerici, Italy. New York, NY: Wiley-Liss, Inc.; pp. 443-454. (Progress in clinical and biological research: v. 372).
Materna, B. L.; Jones, J. R.; Sutton, P. M.; Rothman, N.; Harrison, R. J. (1992) Occupational exposures in California wildland
      fire fighting. Am. Ind. Hyg. Assoc. J. 53: 69-76.
Materna, B. L.; Koshland, C. P.; Harrison, R. J. (1993) Carbon monoxide exposure in wildland  firefighting: a comparison of
      monitoring methods. Appl. Occup. Environ. Hyg. 8: 479-487.
McCammon, C. S., Jr.; Halperin, W. F.;Lemen, R. A. (1981) Carbon monoxide exposure from aircraft fueling vehicles. Arch.
      Environ. Health 36: 136-138.
McCammon,  J. B.;  McKenzie, L. E.; Heinzman, M. (1996) Carbon monoxide poisoning related to  the  indoor use of
      propane-fueled forklifts in Colorado workplaces. Appl. Occup. Environ. Hyg. 11: 192-198.
McCurdy, T. (1995) Estimating human exposure to selected motor vehicle pollutants using the NEM series of models: lessons
      to be learned. J. Exposure Anal. Environ. Epidemiol.  5: 533-550.
McCurdy, T. (2000) Conceptual basis for multi-route intake dose modeling using an energy expenditure approach. J. Expos.
      Anal. Environ. Epidemiol. 10: 86-97.
Meyer, B. (1983) Indoor air quality. Reading, MA: Addison-Wesley Publishing Company, Inc.
Miller, R. K.; Ryan, M. C.; Bilowus, P. (1989) Carbon monoxide poisoning in indoor ice skating  arenas. Va. Med. 116: 74-76.
Mumford, J. L.; Williams,  R. W.; Walsh, D. B.; Burton, R. M.; Svendsgaard, D. I; Chuang, J. C.; Houk, V. S.; Lewtas, J. (1991)
      Indoor air pollutants from unvented kerosene heater emissions in mobile homes: studies on particles, semivolatile
      organics, carbon monoxide, and mutagenicity. Environ. Sci. Technol. 25: 1732-1738.
National Center for Health Statistics and U.S. Consumer Product Safety Commission. (1992) Death certificate file. Atlanta, GA:
      U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention.
National Institute for Occupational Safety and Health. (1994) Health hazard evaluation report HETA 90-0365-2415, U.S.
      Department of the Interior, National Park Service, Yosemite National Park, California. Cincinnati, OH: U. S. Department
      of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention.
                                                     4-39

-------
National Institute  for  Occupational Safety and  Health.  (1996)  Preventing carbon monoxide  poisoning  from  small
      gasoline-powered engines and tools [NIOSH alert]. Cincinnati, OH: U.S. Department of Health and Human Services,
      Public Health Service; report no. DHHS (NIOSH) publication no. 96-118.
Ott, W. R. (1982) Concepts of human exposure to air pollution. Environ. Int. 7: 179-196.
Ott, W. R. (1984) Exposure estimates based on computer generated activity patterns. J. Toxicol. Clin. Toxicol. 21: 97-128.
Ott, W.; Eliassen, R. (1973) A survey technique for determining the representativeness of urban air monitoring stations with
      respect to carbon monoxide. J. AirPollut. Control Assoc. 23: 685-690.
Ott, W.; Flachsbart, P. (1982) Measurement of carbon monoxide concentrations in indoor and outdoor locations using personal
      exposure monitors. Environ. Int. 8: 295-304.
Ott, W.; Thomas, J.; Mage, D.; Wallace, L. (1988) Validation of the simulation of human activity and pollutant exposure
      (SHAPE) model using paired days from the Denver, CO, carbon monoxide field study. Atmos. Environ. 22:2101-2113.
Ott, W. R.; Mage, D. T.; Thomas, J. (1992a) Comparison of microenvironmental CO concentrations  in two cities for human
      exposure modeling. J. Exposure Anal. Environ. Epidemiol. 2: 249-267.
Ott, W.; Langan, L.; Switzer, P. (1992b) A time series model for cigarette smoking activity patterns: model validation for carbon
      monoxide and respirable particles in a chamber and an automobile.  In: Measuring, understanding, and  predicting
      exposures in the 21 st century: proceedings of the conference; November 1991; Atlanta, GA. J. Exposure Anal. Environ.
      Epidemiol. 2(suppl. 2): 175-200.
Ott, W.; Switzer, P.; Willits, N; Hildemann, L.; Yu, L.  (1993) Trends of in-vehicle CO exposures  on a California arterial
      highway over one decade. Research Triangle Park, NC: U.S. Environmental Protection Agency, Atmospheric Research
      and Exposure Assessment Laboratory; report no. EPA/600/A-93/220.
Ott, W.; Switzer, P.; Willits, N. (1994) Carbon monoxide exposures inside an automobile traveling on an urban arterial highway.
      J. Air Waste Manage. Assoc. 44: 1010-1018.
Paul, R. A.; Johnson, T. (1985) TheNAAQS exposure model (NEM) applied to carbon monoxide: addendum. Research Triangle
      Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards; report no. EPA
      450/5-85-004.
Paul, R. A.; Johnson, T.; McCurdy, T. (1988) Advancements in estimating urban population exposure. Presented at: 81 st annual
      meeting of the Air Pollution Control Association; June; Dallas, TX. Pittsburgh, PA: Air Pollution Control Association;
      paper no. 88-127.1.
Paulozzi, L. J.; Satink, F.; Spengler, R. F. (1991) A carbon monoxide mass poisoning in an ice  arena in Vermont [letter]. Am.
      J. Public Health 81: 222.
Paulozzi, L. J.; Spengler, R. F.; Vogt, R. L.; Carney, J. K. (1993) A survey of carbon monoxide  and nitrogen dioxide in indoor
      ice arenas in Vermont. J. Environ. Health 56(5): 23-25.
Pedersen, D. H.; Sieber, W. K. (1988) National occupational exposure survey. Volume III: analysis of management interview
      responses. Cincinnati, OH: U.S. Department of Health and Human Services, National Institute for Occupational Safety
      and Health; DHHS (NIOSH) publication no.  89-103.
Peterson, J. E. (1978) Modeling the uptake, metabolism and excretion of dichloromethane by man. Am. Ind. Hyg. Assoc. J.
      39:41-47.
Phillips, T. J.; Jenkins, P. L.; Mulberg, E. J. (1991) Children in California: activity patterns and presence  of pollutant sources.
      In: Health risk and risk communication: papers  from the 84th annual meeting & exhibition of the Air & Waste
      Management Association; June; Vancouver, BC, Canada. Pittsburgh, PA: Air & Waste Management Association; paper
      no. 91-172.5.
Pisarski, A. E. (1992) New perspectives in commuting: based on early data from the 1990 decennial census and the 1990
      Nationwide Personal Transportation Study. Washington, DC: U.S. Department of Transportation, Federal Highway
      Administration; report no. FHWA-PL/92-026.
Quackenboss, J.  J.;  Spengler, J. D.; Kanarek, M. S.; Letz, R.; Duffy, C. P. (1986) Personal  exposure to nitrogen dioxide:
      relationship to indoor/outdoor air quality and activity patterns. Environ. Sci. Technol. 20: 775-783.
Raaschou-Nielsen, O.; Nielsen, M. L.; Gehl, J. (1995) Traffic-related air pollution: exposure and health effects in Copenhagen
      street cleaners and cemetery workers. Arch. Environ. Health 50: 207-213.
Radford, E. P.; Drizd, T. A. (1982) Blood carbon monoxide  levels in persons 3-74 years of age: United States, 1976-80.
      Hyattsville, MD: U.S.  Department of Health and Human Services, Public Health  Service, National Center  for Health
      Statistics; publication no. (PHS) 82-1250. (Advance data from vital and health statistics: no. 76).
Radford, E. P.; Levine, M. S. (1976) Occupational  exposures to carbon monoxide in Baltimore firefighters. J. Occup. Med.
      18: 628-632.
Robinson, J. P.; Thomas, J. (1991) Time spent in activities, locations, andmicroenvironments: a California-national comparison.
      Las Vegas, NV: U.S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory;  report no.
      EPA/600/4-91/006.
                                                     4-40

-------
Rodes, C.; Sheldon, L.; Whitaker, D.; Clayton, A.; Fitzgerald, K.; Flanagan, J.; DiGenova, F.; Hering, S.; Frazier, C. (1998)
      Measuring concentrations of selected air pollutants inside California vehicles [final report].  Sacramento, CA: California
      Environmental Protection Agency, Air Resources Board; contract no. 95-339.
Schwab, M.; Colome, S. D.; Spengler, J. D.; Ryan, P. B.; Billick, I. H. (1990) Activity patterns applied to pollutant exposure
      assessment: data from a personal monitoring study in Los Angeles. Toxicol. Ind. Health 6: 517-532.
Seufert, K. T.; Kiser, W. R. (1996) End-expiratory carbon monoxide levels as an estimate of passive smoking exposure aboard
      a nuclear-powered submarine.  South. Med. J. 89:  1181-1183.
Sexton, K.; Ryan, P. B. (1988) Assessment of human exposure to air pollution: methods, measurements, and models. In: Watson,
      A. Y.; Bates, R. R.; Kennedy, D., eds. Air pollution, the automobile, and public health.  Washington, DC: National
      Academy Press; pp. 207-238.
Shenoi, R.;  Stewart, G.; Rosenberg, N. (1998) Screening for carbon monoxide in children. Pediatr. Emerg. Care 14: 399-402.
Shikiya, D.  C.; Liu, C. S.; Kahn, M. I.; Juarros, J.; Barcikowski, W. (1989) In-vehicle air toxics characterization study in the
      South Coast Air Basin [draft]. El Monte, CA: South Coast Air Quality Management District, Office of Planning and
      Rules; May.
Shusterman, D.; Quinlan,  P.; Lowengart,  R.; Cone, J. (1990)  Methylene  chloride intoxication in a furniture refinisher:
      a comparison of exposure estimates utilizing workplace air sampling and blood carboxyhemoglobin measurements.
      J. Occup. Med. 32: 451-454.
Silvers, A.; Florence, B. T.; Rourke, D. L.; Lorimor, R. J. (1994) How children spend their time: a sample survey for use in
      exposure and risk assessments. Risk Anal. 14: 931-944.
Simeone, L. F. (1991) The intrusion of engine exhaust into the passenger areas of recreational powerboats. Washington, DC:
      U.S.  Coast Guard, Office of Navigation Safety and Waterway Services; report no. DOT-VNTSC-CG-91-1.
Smith, N. J.  (1977)  End-expired  air technic  for determining  occupational carbon monoxide  exposure. J.  Occup.  Med.
      19: 766-769.
Smith, W.; Anderson, T.; Anderson, H. A.; Remington, P. L. (1992) Nitrogen dioxide and carbon monoxide intoxication in an
      indoor ice arena - Wisconsin, 1992. Morb. Mortal. Wkly. Rep. 41: 383-385.
Snook, L. M. (1996) An investigation of driver exposure to carbon monoxide while traveling in the wake of a snowmobile
      [dissertation]. Knoxville, TN: University of Tennessee. Available from: University Microfilms International, Ann Arbor
      MI; publication no.  AAG9709060.
Snook, L. M.; Davis,  W. T. (1997) An investigation of driver exposure to carbon monoxide while traveling in the wake of a
      snowmobile. Presented at: 90th annual meeting & exhibition of the Air & Waste Management Association; Toronto, ON,
      Canada. Pittsburgh, PA: Air & Waste Management Association; paper no. 97-RP143.02.
Sorensen, A. J. (1986) The importance of monitoring carbon monoxide levels in indoor ice skating rinks. J. Am. Coll. Health
      34: 185-186.
Spengler, J. D.; Stone, K. R.; Lilley, F. W. (1978) High carbon monoxide levels measured in enclosed skating rinks. J. Air Pollut.
      Control Assoc. 28: 776-779.
Stewart, R.  D.; Hake, C. L. (1976) Paint-remover hazard. JAMA J. Am. Med. Assoc. 235: 398-401.
Stewart, R. D.; Fisher, T. N; Hosko, M. J.; Peterson, J. E.; Baretta, E. D.; Dodd, H. C. (1972) Experimental human exposure to
      methylene chloride. Arch. Environ. Health 25: 342-348.
Systems Applications, Inc. (1983) Human exposure to atmospheric concentrations of selected chemicals. Volume II. Research
      Triangle Park, NC: U. S. Environmental Protection Agency, Office of Air Quality Planning and Standards; EPA contract
      no. 68-02-3066.
Szalai, A., ed. (1972)  The use of time: daily activities of urban and suburban populations in twelve countries. The Hague, The
      Netherlands: Mouton & Co.
Treitman, R. D.; Burgess,  W. A.; Gold, A. (1980) Air contaminants encountered by firefighters. Am.  Ind. Hyg. Assoc. J.
      41:796-802.
Tsang, A. M.; Klepeis, N. E. (1996) Descriptive statistics tables from a detailed analysis of the national human activity pattern
      survey (NHAPS) data. Las Vegas, NV: U.S. Environmental Protection Agency, Office of Research and Development;
      report no. EPA/600/R-96/148.
U.S. Code.  (1990) Clean Air Act, as amended by PL 101-549, November 15, 1990. U. S. C. 42: §§7401-7671q.
U.S. Department of Commerce. (1983) Statistical abstract of the United States: 1982-83. National data book and guide to
      sources. 103rd ed. Washington, DC: Bureau of the Census; p. 618.
U.S. Department of Transportation. (1996) Transportation and air quality: selected facts and figures. Washington, DC: Federal
      Highway Administration; FHWA/PD/96-006; p. 6.
U.S. Environmental Protection Agency.  (1985) Health assessment document for dichloromethane (methylene chloride).  Final
      report. Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and
      Assessment Office; report no. EPA-600/8-82/004F.
U.S. Environmental Protection Agency. (1991) Air quality criteria for carbon monoxide. Research Triangle Park, NC: Office
      of Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA/600/8-90/045F.
                                                     4-41

-------
U.S. Environmental Protection Agency. (1992a) Guidelines for exposure assessment. Washington, DC: Risk Assessment Forum;
      report no. EPA/600/Z-92/001.
U. S. Environmental Protection Agency. (1992b) Review of the national ambient air quality standards for carbon monoxide: 1992
      reassessment of scientific and technical information.  OAQPS  staff paper. Research Triangle Park, NC: Office of Air
      Quality Planning and Standards; report no. EPA-452/R-92-004.
U.S. Environmental Protection Agency. (1996) National air pollutant emission trends, 1900-1995. Research Triangle Park, NC:
      Office of Air Quality Planning and Standards; report no. EPA 454/R-96-007.
U.S. Environmental Protection Agency. (1998) National air quality and emissions trends report, 1997. Research Triangle Park,
      NC: Office of Air Quality Planning and Standards; report no. EPA 454/R-98-016. Available: www.epa.gov/oar/aqtmd97/
      [1999, November 24].
U.S. Environmental Protection Agency. (1999a) AP-42: compilation of air pollutant emission factors. Volume II: mobile sources
      (AP-42), pending 5th edition. Ann Arbor, MI: Office of Mobile Sources. Available: www.epa.gov/oms/ap42.htm [2000,
      February, 24].
U. S. Environmental Protection Agency. (1999b) Federal and California light-duty exhaust emission standards. Washington, DC:
      Office of Transporation and Air Quality; report no. EPA420-B-98-001. Available: www.epa.gov/oms/stds-ld.htm [2000,
      February 4].
Vadarevu, R. V.; Stopher, P. R. (1996) Household activities, life cycle, and role allocation. Trans. Res. Rec. 1556: 77-85.
Van Wijnen, J. H.; Verhoeff, A. P.; Jans, H. W. A.; van Bruggen, M. (1995) The exposure of cyclists, car drivers and pedestrians
      to traffic-related air pollutants. Int. Arch. Occup. Environ. Health 67: 187-193.
Vreman, H.  J.; Mahoney, J. J.; Stevenson, D. K. (1993)  Electrochemical measurement of carbon monoxide in breath:
      interference by hydrogen. Atmos. Environ. Part A 27:  2193-2198.
Wallace, L.  (1979)  Use of personal monitor to measure commuter exposure to carbon monoxide in vehicle passenger
      compartments. Presented at:  72nd annual meeting of the Air Pollution Control Association; June; Cincinnati, OH.
      Pittsburgh, PA: Air Pollution Control Association; paper no. 79-59.2.
Wallace, L. A. (1983) Carbon monoxide in air and breath of employees in an underground  office. J. Air Pollut.  Control Assoc.
      33:678-682.
Webster, H. (1998) Would you like to work at home? Read.  Dig. 152: 128-134.
Wilcosky, T. C.; Simonsen, N R. (1991) Solvent exposure and cardiovascular disease. Am. J. Ind. Med. 19: 569-586.
Wiley, J. A.; Robinson, J. P.; Piazza, T.; Garrett,  K.; Cirksena, K.; Cheng, Y.-T.; Martin G.  (1991a) Activity patterns of
      California residents. Final report.  Sacramento, CA: California Air Resources Board; report no. ARB/R93/487.
Wiley, J. A; Robinson, J. P.; Cheng, Y.-T.; Piazza, T.; Stork,  L.; Pladsen, K. (199 Ib) Study of children's activity patterns: final
      report. Sacramento, CA: California Air Resources Board; report no. ARB-R-93/489.
Wilkinson, T. (1995) Snowed under. Natl. Parks 69: 32-37.
Williams, R.; Walsh, D.; White, J.; Jackson, M.; Mumford, J. (1992) Effect on carbon monoxide levels in mobile homes using
      unvented kerosene heaters for residential heating. Indoor Environ. 1: 272-278.
Wilson, A. L.; Colome, S. D.;  Tian, Y. (1991) Air toxics microenvironment exposure and monitoring study.  Final report.
      El Monte, CA: South Coast Air Quality Management  District; February.
Wilson, A. L.; Colome, S.  D.; Tian, Y. (1993a) California residential indoor air  quality study. Volume 1: methodology and
      descriptive statistics. Chicago, IL: Gas Research Institute; report no. GRI-93/0224.1.
Wilson, A. L.; Colome, S. D.; Tian, Y. (1993b) California residential indoor air  quality study. Volume 1: methodology and
      descriptive statistics. Appendices. Chicago, IL: Gas Research Institute; report no. GRI-93/0224.2.
Yates, M. W. (1967) A preliminary study of carbon monoxide gas in the home. J.  Environ. Health 29: 413-420.
Yocom, J. E. (1982) Indoor-outdoor air quality relationships: a critical review. J. Air Pollut. Control Assoc. 32:  500-520.
Yocom, J. E.; Clink, W. L.; Cote, W. A. (1971) Indoor/outdoor air  quality relationships.  J. Air Pollut.  Control Assoc.
      21:251-259.
Yoon, S. S.; Macdonald, S. C.; Fairish, R. G. (1998) Deaths  from unintentional carbon monoxide poisoning and potential for
      prevention with carbon monoxide detectors. JAMA J.  Am. Med. Assoc. 279: 685-687.
Yu, L. E.; Hildemann, L. M.; Ott, W. R. (1996) A mathematical model for predicting trends in carbon monoxide emissions and
      exposures on urban arterial highways. J. Air Waste Manage. Assoc. 46: 430-440.
Zartarian, V. G.; Ott, W. R.; Duan, N. (1997) A quantitative definition of exposure and related concepts.  J. Exposure Anal.
      Environ. Epidemiol. 7: 411-437.
Zayasu, K.; Sekizawa, K.; Okinaga, S.; Yamaya, M.; Ohrui, T.; Sasaki, H. (1997) Increased carbon monoxide in exhaled air of
      asthmatic patients. Am. J. Respir. Grit. Care Med. 156: 1140-1143.
Ziskind, R. A.; Rogozen, M. B.; Carlin, T.; Drago, R. (1981)  Carbon monoxide intrusion into sustained-use vehicles. Environ.
      Int. 5: 109-123.
Ziskind, R. A.; Fite, K.; Mage, D. T.  (1982) Pilot field study:  carbon monoxide exposure monitoring in the general population.
      Environ. Int. 8: 283-293.
                                                      4-42

-------
                                    CHAPTER 5

  Pharmacokinetics and Mechanisms of Action  of
                             Carbon Monoxide
5.1  Introduction
       Basic research on the physiology, pharmacokinetics, and toxicology of carbon monoxide (CO) that
ended in the late seventies was followed by studies focused primarily on the cardiopulmonary effects of CO
as an ambient air pollutant. Although research in this area continues, more recent studies have refocused
on the mechanisms of action and pathophysiological effects of CO at a cellular level and on its role as a
cytotoxic agent and neural messenger. In this chapter, the sections discussing basic pharmacokinetics draw
heavily from Chapter 9 of the previous CO criteria document (U.S. Environmental Protection Agency,
1991). However, all sections were revised and consolidated, many were expanded, and several new sections
were added. In particular, sections on tissue production and metabolism of CO and intracellular effects of
CO have been revised extensively and expanded.  The new section on conditions affecting uptake and
elimination of CO discusses the influence of physical activity, altitude, physical characteristics, and health
status on carboxyhemoglobin (COHb) formation. Also, new sections on the  mechanisms of CO and a
review of the developing concepts have been added.
       Although the focus of this document is on the effects of ambient and near ambient levels of CO
leading to low COHb levels (< 5%), this chapter discusses, where appropriate, findings of a selected number
of human studies carried out at moderate COHb levels (<20%). Also discussed are observations from a
limited number of relevant animal studies at even higher COHb levels. The purpose for the inclusion of
such observations from human studies at higher CO concentrations, and animal studies in general, is to
facilitate the understanding of CO kinetics, related pathophysiologic processes, and mechanisms of
cytotoxicity.  Despite much higher CO uptake  and  elimination rates in animal species than in humans,
primarily because of substantially higher ventilation rates, the laboratory animal data still fill, although only
partially, the knowledge gaps for which no human data are available in these areas of research. Over the
range of CO concentrations that are most relevant experimentally to typical environmental CO exposures
(e.g., 50 to 500 ppm), the rate of both CO uptake and elimination in mammals is inversely proportional to
body mass (i.e.,the smallerthe animal, thefasterthe rate [Klimischetal., 1975;Tyumaetal., 1981]). Over
this same range of CO concentrations, the most widely used predictive model of COHb formation, the
Coburn-Forster-Kane (CFK) equation, accurately predicts the resulting COHb levels not only in human
subjects, but also in laboratory rats and mice (Tyuma et al, 1981; Benignus and Annau, 1994; Kimmel
et al., 1999).  Thus, despite many well identified interspecies differences in the toxicokinetics of CO, the
basic mechanisms of CO toxicity between laboratory animals and humans are similar and,  in many respects,
close to identical.  Although a more detailed discussion of interspecies differences as they relate to humans
may aid in interpretation of data and elucidation of mechanisms, it is not essential for understanding the
material presented in this chapter and is well beyond the scope of this document (see Chapter 1). Despite
interspecies differences, especially  in the uptake  and elimination kinetics of CO,  extrapolation of
observations from animals to man as applied in this chapter, even with its many assumptions, may be useful
                                           5-1

-------
in identifying potential pathophysiologic and histotoxic processes associated with ambient or near ambient
CO exposure.
5.2  Absorption, Distribution, and Pulmonary Elimination
5.2.1  Pulmonary Uptake
       Although CO is not one of the respiratory gases, the similarity of physico-chemical properties of
CO and oxygen (O2) permits an extension of the findings of studies on the kinetics of transport of O2 to
those of CO. The rate of formation and elimination of COHb, its concentration in blood, and its catabolism
is controlled by numerous physical  factors and physiological mechanisms.  The relative contribution of
these mechanisms to the overall COHb kinetics will depend on the environmental conditions, the physical
activity of an individual, and many other physiological processes, some of which are complex and still
poorly understood (see Section 5.4 for details).  All of the pulmonary uptake occurs at the respiratory
bronchioles and alveolar ducts and sacs. The rate of CO uptake depends on the rate of COHb formation.
At the low concentration of CO in inhaled air, the rate of uptake and the rate of COHb formation could, for
all practical purposes, be considered to be qualitatively the same.

5.2.1.1  Mass Transfer of Carbon Monoxide
       The mass transport of CO between the
airway opening (mouth and nose)  and the red
blood   cell  (RBC)   hemoglobin  (Hb)  is
predominantly controlled by physical processes.
The  CO transfer to the  Hb-binding  sites is
accomplished  in  two  sequential  steps:
(1) transfer of CO in a gas phase, between the
airway opening and the alveoli, and (2) transfer
in  a "liquid"  phase,  across  the air-blood
interface, including the RBC.  In the gas phase,
the key mechanisms of transport are convective
flow, by the mechanical action of the respiratory
system, and diffusion in the acinar  zone of the
lung (Engel et al.,  1973). Subsequent molecular
diffusion of  CO  across  the  alveolo-capillary
membrane  along the  CO pressure  gradient,
plasma, and RBC is the virtual mechanism of the
liquid phase.  The principal transport pathways
and body stores of CO are shown in Figure 5-1
(Coburn, 1967).
            Carbon Monoxide in the Ambient Air
  \ Endogenous j
  j  production
  i   ofCO
  !  Metabolism
  \  of CO to CO, i
                 Extravascular compartment
Figure 5-1.  Diagrammatic presentation of CO uptake and
elimination pathways and CO body stores.
Source: Adapted from Coburn (1967).
5.2.1.2  Effects of Dead Space and Ventilation/Perfusion Ratio
       The effectiveness of alveolar gas exchange depends on effective gas mixing and matching of
ventilation and perfusion. During normal tidal breathing, the inhaled gas is not distributed uniformly across
the tracheobronchial tree.   With increased inspiratory flow,  as  during exercise, intrapulmonary gas
distribution becomes more uniform, but gas concentration inhomogeneity still will persist. Considering that
almost 90% of gas is contained within the acinar zone of the lung, any increase in gas inhomogeneity in this
terminal region will have about the same negative effect as an additional increase in the alveolar dead space
or a decrease in the alveolo-capillary diffusion capacity (Engel et al.,  1973).
                                              5-2

-------
       The inefficiency of gas mixing and a consequent decrease in the effectiveness of alveolar gas
exchange is aggravated by ventilation/perfusion (VA/Q) mismatch. Because of the gravity dependence of
ventilation and even more of perfusion in an upright posture, regional VA/Q ratios will range from 0.6 (at
the base of the lung) to 3.0  (at the apex), the overall value being 0.85.  As a result, the VA/Q ratio is the
principal variable controlling gas exchange, and any inequalities not only will impair transfer of gases to
the blood but also will interfere with unloading of gases from the blood into the alveolar air. In humans,
a change in posture to recumbent or exercise will increase the uniformity of VA/Q ratios and promote more
efficient gas exchange, whereas increased resting lung volume, increased airway resistance, decreased lung
compliance, and, generally, any lung abnormality will aggravate VA/Q ratio inequality.
       The simplest indicator of the VA/Q ratio inequalities is the volume of physiological dead space (VD),
which comprises both the anatomical and alveolar dead space.  The alveolar dead space results from
reduced perfusion of alveoli, relative to their ventilation (Singleton et al,  1972). Both right-to-left and
physiological  shunts under normal  conditions contribute little to VA/Q inequality (West, 1990a).
An increase in tidal volume or respiratory frequency, or both, will increase moderately to substantially the
VD in healthy subjects and in individuals with lung function impairment, respectively (Lifshay et al., 1971).

5.2.1.3  Lung Diffusion of Carbon Monoxide
       The next step in the transfer of gases across the alveolar air-Hb barrier is accomplished by gas
diffusion, which is an entirely passive process. To reach the Hb-binding sites, CO and other gas molecules
have to diffuse across the alveolo-capillary membrane, through the plasma, across the RBC membrane, and,
finally, into the RBC stroma before reaction between CO  and Hb can take place. The molecular transfer
across the membrane and the blood phase  is governed by  general physico-chemical laws, particularly by
Pick's first law of diffusion (West, 1990b).  The exchange and equilibration of gases between the two
compartments (air and blood) is very rapid. The dominant driving force is a partial pressure differential of
CO across this membrane; for example, inhalation of a bolus of air containing levels of CO above blood
baseline rapidly increases blood COHb. The rapidity of CO binding to Hb keeps  a low partial pressure of
CO within the RBC, thus maintaining a high pressure differential between air and blood and consequent
diffusion of CO into blood. Subsequent inhalation of CO-free air reverses the gradient (higher CO pressure
on the blood side than alveolar air), and CO is released into alveolar air. The air-blood gradient for CO
pressure is usually much higher than the blood-air gradient; therefore, CO uptake will be a proportionately
faster process than CO  elimination.  The rate of CO release also will be affected by back pressure from
endogenous production of CO.
       Diurnal variations in  CO diffusion capacity of  the lung (DLCO) related to variations in Hb
concentration have been reported in normal, healthy subjects (Frey et al., 1987). Others found the changes
to be related  also to physiological factors such as  oxyhemoglobin (O2Hb), COHb, partial pressure of
alveolar carbon  dioxide (CO2), ventilatory pattern, O2 consumption,  blood flow, functional residual
capacity, etc. (Forster, 1987). Diffusion capacity seems to be relatively independent of lung volume within
the mid-range of vital capacity. However,  at extreme volumes, the differences in diffusion rates could be
significant; at total lung capacity, diffusion is higher, whereas, at residual volume, it is lower than the
average (McClean et al., 1981). In a supine  position at rest, DLCO  has been shown to be significantly
higher than that at rest in a sitting position (McClean et al.,  1981). Carbon monoxide diffusion capacity
increases with exercise, and, at maximum work rates, diffusion will be maximal regardless of body position.
This increase is attained not only by increases in both the diffusing capacity of the alveolar-capillary
membrane  and the pulmonary  capillary blood flow (Stokes et  al., 1981)  but  also by   increased
"V^/Q uniformity (Harf et al., 1978).  Under pathologic conditions, where several components of the air-
blood interface may be affected severely, and the "V^/Q ratio inequality also may increase (as in emphysema,
and fibrosis, or edema), both the uptake and elimination of CO will be affected (Barie et al., 1994).
                                              5-3

-------
5.2.2  Tissue Uptake
5.2.2.1  The Lung
       Although the lung in its function as a transport system for gases is exposed continuously to CO, very
little CO actually diffuses into the lung tissue itself (as dissolved CO), except for the alveolar region where
it diffuses across the lung tissue and into blood. The epithelium of the conductive zone (nasopharynx and
large airways) presents a significant barrier to diffusion of CO.  Therefore, diffusion and gas uptake by the
tissue, even at high CO concentration, will be slow; most of this small amount of CO will be dissolved in
the mucosa of the airways.  Diffusion into the submucosal layers and interstitium will depend on the
concentration and duration of CO exposure and on the relative surface area. Experimental exposures of the
oronasal cavity in monkeys to very high concentrations of CO (>400 ppm) for a very short period of time
(5 s) increased the blood COHb level to <3.5%.  Comparative exposures of the whole lung, however,
elevated COHb to almost 60% (Schoenfisch et al., 1980). Thus, diffusion of CO across the airway mucosa
will contribute little, if at all, to overall COHb concentration.

5.2.2.2  The Blood
       The rate of CO binding with Hb is about 20% slower, and the rate of dissociation from Hb is an
order of magnitude slower than are these rates for O2. However, the CO chemical affinity (represented by
the Haldane coefficient, M) for Hb is about 218 (210 to 250) times greater than that of O2 (Roughton, 1970;
Rodkey et al., 1969). Under steady-state conditions (gas exchange between blood and atmosphere remain
constant), one part of CO and 218 parts of O2 would form equal parts of O2Hb and COHb, which would be
achieved by breathing air containing 21% oxygen and 650 ppm CO. Moreover, the ratio of COHb to O2Hb
is proportional to the ratio of their respective partial pressures, PCO and PO2. The relationship between the
affinity  constant M and PO2 and PCO, first expressed by Haldane (1897-1898), has the following form:
                           COHb/O2Hb = M x  (PCO/PO2).
                                            (5-1)
At equilibrium, when Hb is maximally saturated by O2 and CO at their respective gas tensions, the M value
for all practical purposes is independent of pH, CO2, temperature, and 2,3-diphosphoglycerate (Wyman
et al., 1982; Gr0nlund and Garby, 1984).
       Under dynamic conditions, competitive
binding of O2 and CO to Hb is complex; simply
said, the greater the number of heme molecules
bound to CO, the greater is the affinity of free
hemes for O2.  However, CO  not only occupies
O2-binding sites, molecule for molecule, thus
reducing the amount of  available  O2, but also
alters the characteristic  relationship between
O2Hb and  PO2,  which, in  normal blood,  is
S-shaped.   Figure  5-2  illustrates  the basic
mechanisms of CO toxicity operating at any CO
concentration. The a and a' points represent the
arterial values of PO2.  The v represents the
venous  PO2  of healthy   individuals  after
extraction of 5 vol % of O2.  With increasing
concentration  of  COHb  in  blood,  the
dissociation curve is shifted gradually to the left,
and  its  shape  is transformed  into  a near
rectangular hyperbola. Because the shift occurs
                     PCX (mm Hg)
Figure 5-2.  Oxyhemoglobin dissociation curve of normal
human blood, of blood containing 50% COHb, and of blood
with only 50%  Hb because of anemia.  See the text for
additional details.
Source: U.S. Environmental Protection Agency (1991).
                                              5-4

-------
over a critical saturation range for release of O2 to tissues, a reduction in O2Hb by CO binding will have
more severe effects on the release of O2 than the equivalent reduction in Hb caused by anemia. Thus, in
an acute anemia patient (50% of Hb) at avenous PO2 of 26 torr (v'), 5 vol % of O2 (50% desaturation) was
extracted from blood, an amount sufficient to sustain tissue metabolism. In contrast, in a person poisoned
with CO (50% COHb), the venous PO2 will have to drop to 16 torr (v2; severe hypoxia) to release the same,
5 vol % O2. Any higher demand on O2 under these conditions (e.g., by exercise) might result in brain
oxygen depletion and loss of consciousness of the CO-poisoned individual.
       Because so many cardiopulmonary factors determine COHb formation, the association between
COHb concentration in blood and duration of exposure is not linear but S-shaped.  With progression of
exposure, the initial slower COHb formation gradually accelerates, but, as COHb approaches equilibrium,
the build-up slows down again.  The S-shape form becomes more pronounced with higher CO levels or with
exercise (Benignus et al, 1994; Tikuisis et al., 1992).
       As  Figure 5-1 shows,  CO not only is exchanged between alveolar air and blood but also is
distributed by blood to other tissues. Studies on dogs (Coburn, 1967; Luomanmaki and Cob urn,  1969)
found that, over the range of 2 to 35% COHb, an average of 77% of total body CO remains in the vascular
compartment. The rest of CO diffused to extravascular tissues, primarily skeletal muscle where it is bound
to myoglobin (Mb). Compared to dogs, the extravascular CO stores in men are smaller and account for 10
to 15% of total body CO, and less than 1% of the body CO stores appears to be physically dissolved in body
fluids (Coburn, 1970a). Similar to animals, no shift between blood and extravascular compartments in men
was found at low (<4%) COHb.

5.2.2.3   Heart and Skeletal Muscle
       Myoglobin, as a respiratory hemoprotein of muscular tissue, will undergo a reversible reaction with
CO in a  manner similar to  O2.  Greater affinity of O2 for Mb than Hb  (hyperbolic versus S-shaped
dissociation curve) is, in this instance, physiologically beneficial because a small drop in tissue PO2 will
release a large amount of O2 from oxymyoglobin. The main function of Mb is thought to be a temporary
store of O2 and to act  as a diffusion facilitator between Hb and the tissues (Peters et al., 1994).
       Myoglobin has a CO affinity constant approximately eight-times lower than Hb (M = 20 to 40
versus 218, respectively) (Haab and Durand-Arczynska, 1991; Coburn and Mayers, 1971). AswithHb,the
combination velocity constant between CO  and Mb is only  slightly  lower than that for O2, but  the
dissociation velocity constant is much lower than that for O2. The combination of greater affinity (Mb is
90% saturated at PO2 of 20 mmHg) and lower dissociation velocity constant for CO favors retention of CO
in the muscular tissue. Thus, a considerable amount of CO potentially can be stored in the skeletal muscle
(Luomanmaki and Coburn, 1969).  The binding of CO to Mb (carboxymyoglobin [COMb]) in heart  and
skeletal muscle in vivo has been demonstrated at levels of COHb below 2% in heart and 1% in skeletal
muscle (Coburn, 1973; Coburn and Mayers, 1971).  At rest, the COMb/COHb ratio (0.4 to 1.2) does not
increase with  an increase in COHb up to 50% saturation and appears to be independent of the duration of
exposure (Sokal et al., 1984). During exercise, the relative rate of CO binding increases more for Mb than
for Hb, and CO will diffuse from blood to skeletal muscle (Werner and Lindahl, 1980); consequently, the
COMb/COHb will increase for both skeletal and cardiac muscles (Sokal et al., 1986). A similar shift in CO
has been observed under hypoxic conditions because a fall in myocyte intracellular PO2 below a critical
level will increase the relative affinity of Mb to CO (Coburn and Mayers, 1971). Consequent reduction in
O2 storage capacity of Mb may have a profound effect on the supply of O2 to the tissue. Apart from Hb  and
Mb, other hemoproteins will  react with CO; however, the exact role of such compounds on O2-CO kinetics
still needs to be ascertained.  For more discussion on this topic, see Section 5.6.1.
                                             5-5

-------
5.2.2.4  The Brain and Other Tissues
       The concentration of CO in brain tissue has been found to be about 30- to 50-times lower than that
in blood.  During  the elimination of CO from the brain, the  above ratio of concentrations was still
maintained (Sokal et al., 1984). However, the energy requirement of brain tissue is very high and varies
greatly with local metabolism. Because oxygen demand also is coupled to local functional activity, which
at times may be very high, and because the brain's oxygen storage is minimal, any degree of hypoxia if
uncompensated will have a detrimental effect on brain function.  The primary effects of low ambient
concentrations of CO on other organs (e.g., liver, kidney) is viahypoxic mechanisms (see Section 6.6).

5.2.3  Pulmonary and Tissue Elimination
       An extensive amount of data available on the rate of CO uptake and the formation of COHb contrast
sharply with the limited information available on the dynamics of CO washout from body stores and blood.
Although  almost all of the studies investigating  CO elimination  pattern and processes were done at
moderate COHb levels (<20%), the physiologic mechanisms involved in CO elimination kinetics also are
effected at lower blood COHb, including levels resulting from ambient exposures (<5%). The elimination
rate of CO from an equilibrium state will follow a monotonically decreasing, second-order (logarithmic or
exponential) function (Pace et al., 1950). The rate, however,  may not be constant when the steady-state
conditions have not yet been reached.  Particularly after very short and high CO exposures, it is possible
that COHb decline  could be biphasic, and it can be approximated best by a double-exponential function;
the initial rate of decline or "distribution" might be considerably faster than the later "elimination" phase
(Wagner et al., 1975).  The reported divergence of the COHb  decline rate in blood and in exhaled air
suggests that the CO elimination rates from extravascular pools are slower than those reported for blood
(Landaw,  1973).   Although the absolute elimination rates  are associated  positively  with  the initial
concentration of COHb, the relative elimination rates appear to be independent of the initial concentration
of COHb (Wagner et al., 1975).
       The same factors that govern CO uptake will affect CO  elimination. This suggests that the CFK
model (see Section 5.5.1) may be suitable to predict CO elimination as  well. Surprisingly, few  studies
tested this application. When breathing  air, the  CFK model  predicted very well the COHb decline.
However,  at a higher partial pressure  of O2 in  humidified inspired air (P:O2)  or  under hyperbaric
O2 conditions, the  key CFK equation parameters, particularly the DLCO value, must be adjusted for
hyperoxic conditions so that CFK will predict more accurately the elimination of CO (Tikuisis, 1996;
Tikuisis et al., 1992; Tyuma et al.,  1981).  The half-time of CO disappearance from blood under normal
recovery (air) showed a considerable between-individual variance. For COHb concentrations of 2 to 10%,
the half-time ranged from 3  to 5 h (Landaw, 1973); others reported the range to be 2 to 6.5 h for slightly
higher initial concentrations of COHb (Peterson and Stewart, 1970).  The CO elimination half-time in
nonsmokers  is considerably longer in men (4.5 h) than in women (3.2 h). During sleep, the elimination rate
slowed in both sexes, but, in men, it became almost twice as slow (8.0 h) as during waking hours. Although
no ventilation variables were obtained during the study, the day-to-night differences have been attributed
to lower ventilation rates at sleep. The authors speculate that the sex differences in elimination half-time
are related to the skeletal muscle mass and intrinsically to the amount of Mb (Deller et al., 1992). The half-
time elimination rate appears to be independent of the CO exposure source (e.g., fire, CO  intoxication).
Normobaric  O2 administered to fire victims and CO-poisoned individuals resulted in about the same CO
elimination half-time, 91 and 87 min, respectively  (Levasseur et al., 1996).
       Increased inhaled concentrations of O2 accelerated elimination of CO; by breathing 100%  O2, the
half-time was shortened by almost  75% (Peterson and Stewart, 1970). The average half-life of COHb in
individuals with very low COHb level (1.16%) breathing hyperbaric O2 was 26 min, compared with 71 min
when breathing normobaric O2 (Jay and McKindly, 1997).  The elevation of PO2 to 3 atm reduced the
half-time to about 20 min, which is approximately a 14-fold decrease over that seen when breathing room


                                             5-6

-------
air (Britten and Myers, 1985; Landaw, 1973). Although the washout of CO can be somewhat accelerated
by an admixture of 5% CO2 in O2, hyperbaric O2 treatment is more effective in facilitating displacement
of CO.  Therefore, hyperbaric oxygen is used as a treatment of choice in CO poisoning. A mathematical
model of COHb elimination that takes into account P:O2 has been developed but not yet validated (Singh
et al., 1991; Selvakumar et al, 1993).
5.3  Tissue Production and Metabolism of Carbon Monoxide
       In the process of natural degradation of RBC Hb to bile pigments, a carbon atom is separated from
the porphyrin nucleus and, subsequently, is catabolized by heme oxygenase (HO) into CO.  The major site
of heme breakdown and, therefore, the major production organ of endogenous CO is the liver (Berk et al.,
1976).  The spleen and the erythropoietic system are other important catabolic generators of CO.  Because
the amount of porphyrin breakdown is stoichiometrically related to the amount of endogenously formed
CO, the blood level of COHb or the concentration of CO in the alveolar air have been used with mixed
success as quantitative  indices of the rate of heme catabolism (Landaw et al., 1970; Solanki et al., 1988).
Diurnal variations in endogenous CO production are significant, reaching a maximum around noon and a
minimum around midnight (Levitt et al., 1994; Mercke et al., 1975a).  Week-to-week variations of CO
production are greater than day-to-day or within-day variations for both males and females (Lynch and
Moede, 1972; Mercke et  al., 1975b).
       Any disturbance leading to accelerated destruction of RBCs and accelerated breakdown of other
hemoproteins would lead to increased production of CO. Hematomas, intravascular hemolysis of RBCs,
blood transfusion, and ineffective erythropoiesis all will elevate COHb concentration in blood.  In females,
COHb  levels fluctuate with the menstrual  cycle;  the mean rate of CO production in the premenstrual,
progesterone phase is almost doubled (Delivoria-Papadopoulos et al., 1974; Mercke  and Lundh, 1976).
Neonates and pregnant women also showed a significant increase in endogenous CO production related to
increased breakdown of RBCs.  Degradation of RBCs under pathologic conditions such as anemia
(hemolytic, sideroblastic, and sickle cell),  thalassemia, Gilbert's syndrome with hemolysis, and other
hematological diseases also will accelerate CO production (Berk et al.,  1974; Solanki et al.,  1988).
In patients  with hemolytic anemia, the  CO production rate  was 2- to 8-times higher, and blood COHb
concentration was 2- to 3 -times higher than in healthy individuals (Coburn et al., 1966). Anemias also may
develop under many pathophysiologic conditions characterized by chronic inflammation, such as malignant
tumors or chronic infections (Cavallin-Stahl et al., 1976) (see also Section 5.4.3).
       Not all  endogenous CO comes from RBC degradation.   Other hemoproteins, such as  Mb,
cytochromes, peroxidases, and catalase, contribute approximately 20 to 25% to the total amount of CO
(Berk et al., 1976).  Approximately 0.4 mL/h of CO is formed by Hb catabolism, and about 0.1 mL/h
originates from non-Hb sources (Coburn etal, 1963,1964). This will result in a blood COHb concentration
between 0.4 and 0.7% (Coburn et al., 1965).
       A large  variety of drugs will affect endogenous CO production.  Generally, any drug that will
increase bilirubin production, primarily  from the catabolism of Hb, will promote endogenous production
of CO.  Nicotinic acid (Lundh et al., 1975), allyl-containing compounds (acetamids and barbiturates)
(Mercke et al., 1975c), diphenylhydantoin (Coburn, 1970b), progesterone (Delivoria-Papadopoulos et al.,
1974),  and contraceptives (Mercke  et al., 1975b) all will elevate tissue bilirubin and, subsequently, CO
production.
       Another mechanism that will increase CO production is a stimulation of HO and  subsequent
degradation of cytochrome P-450-dependent, mixed-function oxidases. Several types of compounds, such
as a carbon disulfide and sulfur-containing chemicals (parathion and phenylthiourea), will act on different
moieties of the P-450 system leading to an increase in endogenous CO (Landaw et al., 1970). Other sources
of CO involving HO activity include auto-oxidation of phenols, photooxidation of organic compounds, and

                                             5-7

-------
lipid peroxidation of cell membrane lipids (Rodgers et al, 1994).  The P-450 system also is involved in
oxidative dehalogenation of dihalomethanes, widely used solvents in homes and industry (Kim and Kim,
1996).  Metabolic degradation of these solvents and other xenobiotics results in the formation of CO that
can lead to very high (>10%) COHb levels (Manno et al., 1992; Pankow, 1996).
       Ascent to high elevations will increase the endogenous level of COHb in both humans and animals
(McGrath, 1992; McGrath et al., 1993). The baseline  COHb level has been shown to be positively
dependent on altitude (McGrath, 1992).  Assuming the same endogenous production of CO at altitude as
at sea level, the increase in COHb most likely is consequent to a decrease in PO2 (McGrath et al.,  1993).
Whether other variables, such as an accelerated metabolism or a greater pool of Hb, transient shifts in body
stores, or a change in the elimination rate of CO are contributing factors, remains to be explored. Animal
studies suggest that the elevated basal COHb production is not a transient phenomenon but persists through
a long-term adaptation period (McGrath, 1992).
       In recent years, new discoveries in molecular biology identified the CO molecule as being involved
in many physiological responses, such as smooth muscle relaxation, inhibition of platelet aggregation, and
as a neural messenger in the brain (for details, see Sections 5.6 and 5.7). Most recently, several studies
reported yet another function of CO, that of a possible marker of inflammation in individuals with upper
respiratory tract infection (Yamaya et al., 1998) and bronchiectasis (Horvath et al., 1998a) and in asthmatics
(Zayasu et al., 1997; Horvath et al., 1998b).  In the Zayasu et al. (1997) study, the investigators found that
exhaled concentrations of CO in asthmatics taking  corticosteroids  were about the same as in healthy
individuals (1.7 and 1.5 ppm, respectively), whereas, in asthmatics who did not use corticosteroids, the
average CO concentration was 5.7 ppm.  The authors speculate that one of the anti-inflammatory  effects
of corticosteroids is the  down-regulation of HO.  Whether asthmatics have an increased COHb level was
not measured in this study or reported in other studies. Patients with chronic inflammatory lung disease,
such as bronchiectasis may produce a substantial amount of CO (e.g., 11.8 ppm). As with asthma, induction
of heme oxygenase appears to be the primary mechanism involved in the production of CO (Horvath et al.,
1998a,b).  Critical illness also seems to be associated with elevated production of CO (Meyer et al.,  1998).
When compared with controls, ill patients (not characterized) had higher COHb in both arterial and central
venous blood not attributable to an elevated inspired concentration of O2 used to treat patients. Moreover,
the higher COHb in arterial blood than in central venous blood measured in both ill and control individuals
has lead  the  authors to speculate that a positive arterio-venous COHb difference results from  the
up-regulation of the inducible isoform of heme oxygenase (HO-1) in the lung and subsequent production
of CO (see Section 5.6.4).
5.4 Conditions Affecting Carbon Monoxide Uptake and Elimination
5.4.1  Environment and Activity
       During exercise, increased demand for O2 requires adjustment of the cardiopulmonary system, so
that an increased demand for O2 is met with an adequate supply of O2. Depending on the intensity of
exercise, the physiologic changes may range from minimal, involving primarily the respiratory system, to
substantial, involving extensively the respiratory, cardiovascular, and other organ systems, inducing local
as well as systemic changes.  Exercise  will improve the VA/Q ratio in the lung, increase the respiratory
exchange ratio (RER), increase cardiac output, increase DLCO, mobilize RBC reserves from the spleen, and
induce other compensatory changes. Heavy exercise will cause a decrease in plasma volume, leading to
hemoconcentration and a subsequent decrease in blood volume. Of the many mechanisms operating during
exercise, the two most important physiologic variables are (1) the alveolar ventilation (VA) and (2) cardiac
output.  Although some physiologic changes during exercise may impair CO loading into blood (e.g.,
relative decrease in DLCO during severe exercise), the majority of the changes will facilitate CO transport.
Thus, by increasing gas exchange efficiency, exercise also will promote CO uptake. Consequently, the rates

                                             5-8

-------
of CO uptake and COHb formation will be proportional to the intensity of exercise.  During a transition
period from rest to exercise while exposed to CO (500 ppm/10 min), the diffusing capacity and CO uptake
were reported to rise faster than O2 consumption for each exercise intensity (Kinker et al, 1992).
       Apart from physiological factors, the concentration of CO, as well as the rate of change of CO
concentration in an individual's immediate environment, can have a significant impact on COHb.  For
example, at intersections with idling and accelerating cars, pedestrians will be exposed for a short period
of time to higher CO concentrations than those present at other places on the same street.  Around home,
an individual working with a chain saw, lawnmower,  or other gasoline-powered tools will be exposed
transiently to higher, and occasionally to much higher (e.g., breathing near the exhaust of a chain saw),
concentrations of CO (up to 400 ppm) (B linger et al., 1997). In indoor environments, exposure to elevated
CO from unventilated gas appliances or from environmental tobacco smoke may increase transiently the
COHb level of apreviously unexposed individual.  Occupationally, there are many instances and conditions
under which workers may be exposed briefly to moderate-to-high levels of CO from operating equipment
or other sources. Despite the shortness of each exposure episode, such transients may result in a relatively
high COHb concentration.  As an example, exposure for 5 min or less of a resting individual to 7,600 ppm
CO in inhaled air will result in almost 20% COHb (Benignus et al., 1994).  On repeated brief exposures to
high CO, the COHb will increase further until  the concentrations in inhaled CO  and  in blood reach
equilibrium. Once the distribution of CO within body stores is complete, the COHb will remain constant,
unless the ambient CO concentration changes (either up or down) again.  As is the CO uptake, so is the
elimination of CO from blood governed by the gas concentration gradient between blood and alveolar air.
However, the elimination of CO from blood is a much slower process (see Section 5.2.3) and, therefore,
will take  many hours of breathing clean air before the baseline COHb value is reached.
       Recently, a unique source of CO exposure was identified. It has been found repeatedly that the use
of volatile anesthetics (fluranes) in closed-circuit anesthetic machines, when CO2 absorbent (soda lime) is
dry, can result in a significant production of CO caused by a degradation of the anesthetic and subsequent
exposure of a patient to CO (up to 7% COHb) (Woehlck et al., 1997a,b).

5.4.2 Altitude
       Altitude may have a significant influence on the COHb kinetics (U.S. Environmental Protection
Agency,  1978). These  changes are consequent to compensatory and adaptive physiologic mechanisms.
At sea level, at a body temperature of 37 °C, barometric pressure (PB)  of 760 torr, and air (gas) saturated
with water vapor (BTPS conditions) the P:O2 is 149 torr. At an altitude of 3,000 m (9,840 ft; PB = 526 torr),
the P:O2 is only 100 torr, resulting in an acute hypoxic hypoxia. Direct measurements of blood gases on
over 1,000 nonacclimatized individuals at this altitude found the partial pressure of O2 in alveolar air to be
only 61 torr (Boothby  et al., 1954).  The hypoxic drive will trigger a complement of physiological
compensatory mechanisms (to maintain O2 transport and supply), the extent of which will depend on
elevation, exercise intensity, and the length  of a stay at the altitude.  During the first several days, the
pulmonary  ventilation  at  a given  O2  uptake (work  level) will increase progressively  until a new
quasi-steady-state level is achieved (Bender et al., 1987; Burki, 1984).  The DLCO will not change
substantially at elevations below 2,200 m but was reported to  increase above  that  altitude, and the
spirometric lung function will be reduced as well (Ge et al., 1997). The maximal aerobic capacity and total
work performance will decrease, and the RER will increase (Horvath et al., 1988). Redistribution of blood
from skin to organs and within organs from blood into extravascular compartments, as well as an increase
in cardiac output, will promote CO loading (Luomanmaki and Cob urn, 1969).  Because of a decrease in
plasma volume (hemoconcentration), the Hb concentration will be higher than at sea level (Messmer, 1982).
The blood electrolytes and acid-base equilibrium will be readjusted, facilitating transport of O2. Thus,  for
the same  CO concentration as at sea level, these compensatory changes will favor CO uptake and COHb
formation (Burki, 1984). By the same token, the adaptive changes will affect not only CO uptake but CO

                                              5-9

-------
elimination as well.  Carboxyhemoglobin levels at altitude have been shown to be increased in both
laboratory animals and humans (McGrath, 1992; McGrath et al, 1993).  Breathing CO (9 ppm) at rest at
altitude has produced higher COHb levels than those at sea level (McGrath et al., 1993). Surprisingly,
exercise in a CO atmosphere (50 to 150 ppm) at altitude appeared either to suppress COHb formation or
to shift the CO storage, or both.  The measured COHb levels were lower than those found under similar
conditions of exercise and exposure at sea level (Horvath et al., 1988).
       The short-term acclimatization (within a week or two) will stabilize the compensatory changes.
During a prolonged stay at high altitude (over a few months), most of the early adaptive changes gradually
will revert to the sea level values, and long-term adaptive changes, such as an increase in tissue capillarity
and Mb content in the skeletal muscle, begin to take place. Smokers appear to tolerate short-term hypoxic
hypoxia caused by high altitude (7,620 m [25,000 ft]) much better than nonsmokers, who experience more
severe subjective symptoms and a greater decline in task performance  (Yoneda and Watanabe, 1997).
Perhaps smokers, because of chronic hypoxemia (because of chronically elevated COHb), develop partial
tolerance to hypoxic hypoxia. Although the mechanisms of COHb formation in hypoxic hypoxia and CO
hypoxia are different, the resultant decrease in O2 saturation and activation of compensatory mechanisms
(e.g.,  an  increased  cerebral  blood flow)  appear to be at  least  additive    (McGrath,  1988).
Psychophysiological studies, in particular, seem to support the possibility of physiological equivalency of
hypoxic effects, whether induced by altitude at equlibrium or ambient CO concentration. However, it must
be remembered that, although some of the mechanisms of action of hypoxic hypoxia and CO hypoxia are
the same, CO elicits other toxic effects not necessarily related to O2 transport mechanisms (Ludbrook et al.,
1992; Zhu and Weiss, 1994). Recently, Kleinman et al.  (1998) demonstrated that the effects of CO and
simulated altitude were not synergistic but additive.

5.4.3  Physical Characteristics
       Physical characteristics (e.g., sex, age, race, pregnancy) are not known to directly modify the basic
mechanisms of CO uptake and COHb formation and elimination.  However,  the baseline values of many
cardiopulmonary variables  that may influence COHb kinetics are known to change with physical
characteristics.
       The CO uptake and elimination rates either at rest or with exercise decrease with age. During the
growing years (2 to 16 years of age), the COHb elimination half-time increases rapidly with age in both
sexes and is relatively shorter for boys  than for girls.  Beyond  teenage years, the half-time for CO
elimination continues to grow longer but at a lower rate.  In contrast to the adolescent period, the COHb
half-life during the adult years was found to be persistently shorter (-6%) in females than that in males
(Joumard et al., 1981).  Furthermore, it has been well established that the DLCO decreases with age
(Guenard and Marthan, 1996).  The rate of DLCO decline is lower in middle-aged women than it is in men;
however, at older ages, the rates evened and are about the same for both sexes (Neas and Schwartz, 1996).
The decrease in DLCO, combined with an increase in VA /Q mismatch, which increases with age, means that
it will take longer to both load and eliminate CO from blood.
       In pregnancy, increased requirement for iron and hemodilution may lead to iron deficiency and
anemia (for further details see Sections 6.4 and 7.7.1).  Pregnant women who smoked showed a more
pronounced shift of the O2 dissociation curve to the left (-5% COHb) than one would expect from the
same COHb concentration in nonpregnant women. Thus, increased O2 affinity, combined with decreased
O2-carrying capacity of blood of CO-exposed women, may promote fetal hypoxia (Grote et al., 1994).
Animal studies found that protein deficiency in pregnant mice had no modulating effect on maternal COHb
but resulted in a greater concentration of placental COHb (Singh et al., 1992, 1993; Singh and Moore-
Cheatum, 1993).
       Young women were found to be more resistant to altitude hypoxia than were men, but the
physiological factors for this difference remain unexplored  (Horvath et al., 1988). Carboxyhemoglobin

                                            5-10

-------
levels, although elevated at altitude, were found to be about the same for both males and females (McGrath
etal., 1993).
       Whether the dynamics of COHb formation and elimination or the absolute COHb levels for the
same exposure conditions are different in any way between races have not been studied. Blacks have lower
diffusion capacity than whites (Neas and Schwartz, 1996), which transiently will slow CO loading and
unloading. It also is well documented that the black population has a higher incidence of sickle cell anemia,
which may be a risk factor for CO hypoxia (see Section 5.4.4 below).

5.4.4  Health Status
       An individual with any pathophysiologic condition that reduces the blood O2 content will be at a
greater risk from CO exposure because additional reduction in the O2-carrying capacity of blood resulting
from COHb formation will increase hypoxemia. Depending on the severity of initial hypoxia, exposure to
CO may lower the O2 content to the point where O2 delivery to the tissues becomes insufficient.
       One group of disorders that encompasses a range of etiologically varied diseases characterized by
a reduction in total blood Hb and subsequent insufficiency to meet O2 demands are the anemias. Anemia
is a result of either impaired formation  of RBCs or increased loss or  destruction of RBCs.  The former
category includes disorders of altered O2 affinity, methemoglobinemias,  and diseases with functionally
abnormal and unstable Hb. By far, the most prevalent disorder in this group is a single-point mutation of
Hb, causing sickle cell diseases, the most typical of which is a sickle cell anemia. The O2-carrying capacity
of individuals afflicted with sickle cell anemia is reduced not only because of a smaller amount of Hb, but
also the O2 dissociation curve is shifted to the right, reducing the O2 affinity as well.  Initial compensation
involves primarily the cardiovascular system. The cardiac output will increase as both heart  rate and stroke
volume increase.
       The  opposite condition of anemia is  polycythemia, an  increased number of RBCs in  blood.
Although in polycythemia the total amount of Hb generally is elevated, under certain conditions the arterial
O2 saturation may be decreased, leading to a higher risk of additional hypoxia when exposed to CO (Foster
etal., 1978; Stork etal., 1988).
       A distinctive characteristic of chronic obstructive pulmonary disease (COPD) is increased VD and
VA/Q inequality (Marthan et al., 1985).  Subsequently, impaired gas mixing because of poorly ventilated
lung zones will result in decreased arterial O2 saturation and hypoxemia.  These pathophysiologic conditions
will slow both CO uptake and elimination. Any COHb formation will further lower the O2 content of blood
and increase hypoxemia.  Because COPD patients very often operate at the limit of their O2 transport
capability, exposure to CO may severely compromise tissue oxygenation.
       Because O2  extraction by the myocardium is high, a greater O2 demand by the myocardium  of
healthy individuals  is met by an increased coronary blood flow.  Patients with coronary  artery disease
(CAD) have a limited ability to increase coronary blood flow in response to increased O2 demand  during
physical activity. If this compensatory mechanism is further compromised by decreased O2 saturation from
CO inhalation, the physical activity of patients with CAD may be restricted severely consequent to more
rapid development of myocardial ischemia.
       Individuals  with  congestive heart failure,  right-to-left  shunt in congenital heart disease,  or
cerebrovascular disease also may be at a greater risk from CO exposure because of already compromised
O2 delivery.
                                             5-11

-------
5.5  Modeling Carboxyhemoglobin Formation
5.5.1  The Coburn-Forster-Kane and Other Regression Models
5.5.1.1  Empirical Regression Models
       The most direct approach to establishing a prediction equation for COHb is to regress observed
COHb values against the concentration and duration of exogenous CO exposure. Inclusion of other
predictors such as initial COHb level and VA generally will improve the precision of the predictions. Most
of the CO regression models are purely empirical and have no physiological basis. Their applicability
therefore is limited to more or less exact conditions that were used to collect the data on which they are
based.
       Peterson and Stewart (1970) developed a regression equation for estimating  percent COHb
following a 15-min to 8-h exposure of resting nonsmokers to moderate levels of CO (25 to 523 ppm):

                 Log % COHb = 0.858 Log CO + 0.630 Log t - 0.00094 t' - 2.295,             (5-2)

where CO refers to the concentration of CO in inhaled ambient air in parts per million, t is the exposure
duration in minutes,  and t' is the postexposure time in minutes (set to 0 until the end of exposure).  Data
from a subsequent study were used to derive a new empirical formula for much higher concentrations of
CO (1,000 to 35,600 ppm) but much shorter exposure times (45 s to 10 min) (Stewart et al., 1973).  These
equations still are used occasionally  in field conditions to quickly estimate COHb concentration.
       To predict changes in COHb as a function of ambient CO concentration in an urban setting, Ott and
Mage (1978) developed a linear differential equation where only ambient CO concentration varied with
time. All other parameters were empirically derived constants.  With this simple model, they were able to
show that the presence of CO spikes in data averaged over hourly intervals may lead to underestimating the
COHb concentration by as  much as 21% of the true value.  Consequently, they recommended that
monitored CO be averaged over 10 to 15 min periods.  Based on a similar approach, other empirical models
have been developed but not validated (Chung, 1988; Forbes etal, 1945). Comparison of predicted COHb
values  by these two  models  revealed a  progressive divergence of the estimated COHb curves between
models as exposure (100 ppm) progressed, with absolute differences approaching almost 7% COHb.  Such
wide variations in predicted COHb best demonstrate the inaccuracy of these types of models when applied
outside of a narrowly defined range and make their utility in practical applications questionable (Tikuisis,
1996).
       Several more sophisticated mathematical models have been developed to predict COHb as a
function of exposure time (Singh etal., 1991; Sharanetal, 1990) or altitude (Selvakumar etal., 1992). The
physiological variables  used by Peterson and Stewart (1970) were employed to test these models and
compare the results to the CFK predictions.  The agreement between predicted COHb values by these
models and the CFK model was very good; however, these theoretical models have not been validated by
experimental studies.

5.5.1.2  Linear and Nonlinear Coburn-Forster-Kane Differential Equations
       In 1965, Coburn, Forster, and Kane developed a differential equation to describe the maj or physical
and physiological variables that determine the concentration of COHb in blood for the examination of the
endogenous production of CO. The equation, referred to as the CFK model, either in its original form or
adapted to special conditions is still much in use today for the prediction of COHb consequent to inhalation
of CO. Equation 5-3 represents the linear CFK model that assumes O2Hb is constant:
                                            5-12

-------
           d[COHb]t
               dt
=  Vco-
[COHb]0Pc02
[O2Hb]M
1
1 1
DLCOT<,,
+
PjCO
1 1
DLCO + vy
^ V A ,
(5-3)
where Vb is blood volume in milliliters; [COHb]t is the COHb concentration at time t in milliliters CO per
milliliter blood, standard temperature and pressure, dry (STPD); Vco is the endogenous CO production rate
in milliliters per minute, STPD; [COHb]0 is the COHb concentration at time zero in milliliters CO per
milliliter blood, STPD; [O2Hb] is the oxyhemoglobin concentration in milliliters O2 per milliliter blood,
STPD; PcO2 is the average partial pressure of O2 in lung capillaries in millimeters of mercury; VA is the
alveolar ventilation in milliliters per minute, STPD; DLCO is the lung diffusing capacity of CO in milliliters
per minute per millimeter of mercury, STPD; and P:CO is  the CO partial pressure in inhaled air  in
millimeters of mercury. The  model also assumes an instant equilibration of gases in the lung, COHb
concentration  between venous and  arterial  blood, and COHb concentrations between  blood  and
extravascular tissues. Because O2 and CO combine with Hb from the same pool, higher COHb values do
affect the amount  of Hb available for bonding  with O2.  Such  interdependence can be modeled by
substituting (1.38 Hb -  [COHb]) for [O2Hb], where Hb refers to the number of grams of Hb per milliliter
of blood (Tikuisis et al, 1987a). The CFK differential equation (Equation 5-3) then becomes nonlinear:
                    d[COHb]t   Vco     1
                        dt
                                             PjCO-
                              [COHb]0P502
                                 [O2Hb]M
(5-4)
where P is (1/DLCO) + (PB - 47)/VA, and PB is the barometric pressure in millimeters of mercury. The
nonlinear CFK model is more accurate physiologically but has no explicit solution. Therefore, interactive
or numerical integration methods must be used to solve the equation (Muller and Barton,  1987; Johnson
et al., 1992). One of the requirements of the method is that the volumes of all gases be adjusted to the same
conditions (e.g., STPD) (Muller and Barton, 1987; Tikuisis et al., 1987a,b).
       In general, the linear CFK equation is a better approximation to the nonlinear equation during the
uptake of CO than during the elimination of CO. As long as the linear CFK equation is used to predict
COHb levels at or below 6% COHb, the solution to the nonlinear CFK model will deviate no more than
±0.5% COHb (Smith, 1990). Over the years, it has been empirically determined that minute ventilation
and the DLCO have the greatest influence on the CO uptake and elimination. The relative importance of
other physiologic variables will vary with exposure conditions and health status.  A comprehensive
evaluation of fractional sensitivities of physiologic variables for both the linear and nonlinear CFK
equations shows that each  variable will exert its maximal influence at different times of exposure
(McCartney, 1990). The analysis found that only the fractional concentration of CO in inhaled air, in parts
per million (F:CO), and Vco will not affect the rate at which equilibrium is reached. Figure 5-3 illustrates
the temporal changes in fractional  sensitivities of the principal determinants of CO uptake for the linear
form of the CFK equation; THb is the total blood concentration of Hb.  The fractional sensitivity of unity
means that, for example, a 5% error in the selected variable induces a 5% error in predicted COHb value
by the nonlinear model.

5.5.1.3  Confirmation Studies of Coburn-Forster-Kane Models
       Since the publication of the original  paper (Coburn et al.,  1965), several investigators have
tested the fit of both the linear and nonlinear CFK model to experimental data using different CO exposure
                                             5-13

-------
    0.75-
                                           - 0.75
>   0.25-
    ^0.50-
                      Log Time
Figure 5-3. Plot of fractional sensitivities of selected variables
versus time of exposure (see text for details).
Source:  Modified from McCartney (1990).	
                                                    profiles, a variety of experimental conditions,
                                                    and different  approaches  to evaluating the
                                                    parameters of the model.  In all of these studies,
                                                    almost all of the physiologic coefficients either
                                                    were assumed or  estimated  based on  each
                                                    individual's physical characteristics; the COHb
                                                    values  were   measured  directly  and  also
                                                    calculated for each individual.
                                                           Stewart et  al.  (1970)  tested the  CFK
                                                    linear differential equation on 18 resting, healthy
                                                    subjects exposed to 25 different CO exposure
                                                    profiles for periods of 0.5 to 24 h and to CO
                                                    concentrations  ranging from 1 to 1,000 ppm.
                                                    In a later  study, they tested the nonlinear CFK
                                                    equation  on 22  subjects at various levels of
                                                    exercise while being exposed to up to 200 ppm
                                                    CO for up  to  5.25 h  (Peterson and  Stewart,
                                                    1975).    From  the  obtained  values,  they
concluded that, either at rest or with exercise, the agreement between the predicted and measured COHb
values was good (correlation coefficient [r] > 0.74).
       The first study to test both the linear and nonlinear CFK models for CO uptake and elimination in
pedestrians and car passengers exposed to ambient CO levels in a city was conducted by Joumard et al.
(1981). The two cohorts exposed for 2 h to street and traffic concentrations of CO, respectively, comprised
73 nonsmokers (18 to 60 years of age). Blood COHb samples were taken only at the beginning and the end
of each journey, where the COHb value reached 2.7%, on average.  Both equations performed well in
estimating accurately COHb levels, although the value for males was underestimated slightly.
       The predictive strength of the CFK model under variable CO concentrations was tested by Hauck
and Neuberger (1984).  A series of experiments was performed on four subjects exposed to a total of
10 different CO  exposure profiles at several exercise levels,  so that each exposure was a unique
combination of CO concentration and exercise pattern. The ventilation and COHb values (measured and
calculated) were obtained at 1 -min intervals. The agreement between measured and predicted COHb under
these varied conditions was very good; the mean  difference was  only  7.4% of the nominal (maximal
predicted) value.
       A series of studies has tested the accuracy of the CFK equation under transient exposure conditions
that  would violate several assumptions of the CFK model, specifically the assumption of a single,
well-mixed vascular compartment.  These studies were designed to simulate everyday conditions  (e.g.,
environmental, occupational, military) that may involve frequent but brief (75 s to 5 min) exposures to high
(667 to 7,500 ppm) CO concentrations at rest and with exercise. Moreover, the experiments were designed
to test the accuracy of the CFK equation under transient exposure conditions during the CO uptake and early
elimination phases from arterial and venous blood.  Attempts were  made to measure directly some of the
key physiologic parameters of the CFK equation for each subject (Tikuisis et al., 1987a,b; Benignus et al.,
1994). The studies have shown that during and immediately following exposure, the arterial COHb was
considerably higher (1.5 to  6.1%), and the venous COHb was considerably lower (0.8 to 6%) than the
predicted COHb.  The observed COHb differences between arterial and venous blood ranged from 2.3 to
12.1% COHb among individuals (Benignus et al., 1994).  The overprediction of venous COHb increased
during exercise (-10% of the  true value).  Provided that  the total CO dose (concentration x time) is the
same and within the time constant for the CO uptake and elimination, the COHb value was found to be the
same, regardless of the pattern of exposure.  Because VA affects both the equilibrium and the rate at which
                                            5-14

-------
it is achieved, inconsistencies in the estimates or conversion of gas volumes (ATPS and BTPS to STPD)
will affect the predicted values.  The interindividual and intraindividual disparities between measured and
predicted COHb values were attributed primarily to delays in mixing of arterial and venous blood and
differences in cardiac output; but, other factors, such as lung wash-in, also contribute to this phenomenon.
Modification of the CFK equation by adj usting for regional differences in blood flow produced a model that
predicted with much greater accuracy both the arterial (<0.7% COHb difference) and venous (<1% COHb
difference) COHb during transient uptake and elimination of CO from blood (Smith et al, 1994).
       Although the CO concentrations used in these studies are several orders of magnitude higher than
the usual CO concentrations found in ambient air,  under certain conditions (see Section 5.4.1), people can
be exposed briefly (<10 min) to  such (or even higher) levels of CO in their immediate environment.
Because the physiologic mechanisms (but not the kinetics) of COHb formation are independent of CO
concentration, high COHb transients, particularly in at-risk individuals, could be of clinical importance.
Even briefly, higher arterial COHb may lead to functional impairment of the hypoxia-sensitive heart and
brain (see Sections 5.2.2.3  and 5.2.2.4). In these situations, the predicted instantaneous arterial COHb level
will be substantially underestimated.

5.5.1.4  Application of Coburn-Forster-Kane Models
       To obviate measurements of CFK equation parameters, many of which are complex techniques,
attempts were made to simplify the CFK equation, because  it may be difficult or even impossible  to
measure directly some of these parameters, particularly during physical activity. In one study, by relating
physiological parameters to the O2 uptake by the body, which was in turn  related to an activity level, a
simplified linear form of the CFK model was developed (Bernard and Duker, 1981). The model was used
subsequently to draw simple nomograms of predictive relationships between pairs of variables, but the
accuracy of the nomograms was not tested experimentally.
       The  need for more accurate COHb prediction under more complex physiologic or exposure
conditions requires either modification or expansion of the CFK  model.  Benignus (1995) combined a
physiological model of respiratory gas exchange, MACPUF (Ingram et al., 1987), with the CFK model.
The new model  allows for continuous output and input of 60 cardiopulmonary variables, including F:CO.
The usefulness of the model is particularly in its  ability to continuously update COHb  concentration  in
response to dynamically changing physiologic parameters. The model also allows COHb prediction under
conditions that otherwise would be very difficult to duplicate in the laboratory.
       A fundamental modification of the CFK model was made by Hill et  al. (1977) to study the effects
of CO inspired by the mother on the level of fetal COHb.  The Hill equation combines the CFK equation
(for maternal COHb) with a term denoting COHb transfer from a placenta into the fetus.  Comparative
evaluation of predicted and measured fetal  COHb concentrations under time-varying  and steady-state
conditions in both women and animals showed acceptable agreement only under steady-state conditions
(Hill et al., 1977; Longo and Hill, 1977).
       As mentioned in  Section 5.5.1.3, Smith  et al. (1994)  expanded the CFK model to  allow for
prediction of arterial and  venous COHb during transient CO uptake and early elimination phases.  The
model incorporated regional differences in blood flow, particularly in the forearm, because the forearm is
used most frequently  for blood sampling.   This more elaborate model performed extremely well  in
predicting blood COHb. Although the model was validated on a small number of subjects using the same
experimental setting, the validation was not performed under more demanding conditions of physical
activity and varying CO concentrations.
       To accurately predict COHb in individuals exposed to dihalomethanes, which are a source  of
endogenous CO (see Section 5.3), the CFK model was extended to account for the CO production caused
by oxidation of a parent chemical (Andersen et al., 1991).  The model developed and validated on rats
employed a variety of exposure scenarios to dichloromethane. It subsequently was tested on six volunteers


                                             5-15

-------
exposed to dichloromethane, and, after adjustment of a few parameters, the COHb level was predicted
remarkably well.  After further validation, this model has potential use in predicting accurately COHb
caused by exogenous  and endogenous CO originating from different sources (e.g., Hb  degradation,
metabolism of dihalomethanes, inhaled CO).
       Reexpression of the solution of the CFK model from percent COHb to parts per million of CO
allows the examination of a variety of CO concentration profiles, while keeping a simple preselected target
COHb as a constant. Application of the transformed model to urban hourly averaged CO concentrations
that just attained alternative 1-h and 8-h CO National Ambient Air Quality Standards (NAAQS) showed
that, depending on the air quality pattern used, between 0.01 to 10% of the population may exceed a target
2.1% COHb level in blood without ambient CO concentrations ever exceeding the standard. By including
transients, the models predicted COHb more accurately,  particularly when built into the 8-h running
averages (Venkatram and Louch, 1979; Biller and Richmond,  1982, 1992).  Actually, the ambient CO
concentrations could be averaged over any time period less than or equal to the half-life of COHb (Saltzman
and Fox, 1986).
5.6  Intracellular Effects of Carbon Monoxide
5.6.1  Introduction
       Traditional concepts for CO pathophysiology have been based on the high affinity of CO for
deoxyhemoglobin and consequent reduction of O2 delivery.  This mechanism is relevant for high CO
concentrations, but it is less likely to be relevant to the concentrations of CO currently found in the ambient
environment. This section will summarize recently published information on biochemical mechanisms that
is not related to an impairment of oxygen delivery from elevations in COHb.  Some of the studies outlined
in this section were done with cells in culture and others with laboratory rats. To be relevant to human
exposures from environmental contamination, it is important to note what concentrations of CO are likely
to occur in vivo.  Lung parenchyma represents a special situation where cells may be exposed to ambient
CO without the reduction in concentration associated with Hb-bound CO. Elsewhere in the body, only a
fraction of COHb will dissociate to elevate extravascular CO concentrations.  This elevation is in the range
of approximately 2 to 10 nmol when the COHb concentration is from 0.8 to 3.8% (Coburn, 1970a; Gothert
et al, 1970).  The COHb values  near steady-state conditions in laboratory rats are close to values for
humans (Kimmell et al., 1999). This strengthens the potential for human relevance in recent animal studies
that show that newly identified biochemical mechanisms do have adverse physiological effects.  However,
caution still is warranted because  direct evidence for the occurrence of these mechanisms in humans has
not been shown.

5.6.2  Inhibition of Hemoprotein Function
       Carbon monoxide can inhibit a number of hemoproteins found in cells, such as Mb, cytochrome
c oxidase, cytochrome P-450, dopamine P hydroxylase, and tryptophan oxygenase (Coburn and Forman,
1987).  Inhibition of these enzymes could have adverse effects on cell function.
       Carbon monoxide acts as  a competitive inhibitor, hence biological effects depend on the partial
pressures of both CO and O2. The cellular hemoprotein with the highest relative affinity for CO over that
for O2 is Mb. Carbon monoxide will inhibit Mb-facilitated oxygen diffusion, but physiological compromise
is seen only with high concentrations of COMb. Wittenberg and Wittenberg (1993) found that high-energy
phosphate production was inhibited in isolated cardiac myocytes, maintained at a physiologically relevant
oxygen concentration, when COMb exceeded 40%.  The authors estimated that formation of sufficient
COMb to impair oxidative phosphorylation in vivo would require a COHb level of 20 to 40%.
       Coefficients for binding CO versus O2 among cytochrome P-450-like proteins vary between 0.1 and
approximately 12, and there have been recent discussions suggesting that CO-mediated inhibition of these

                                            5-16

-------
proteins could cause smooth muscle relaxation in vivo (Coburn and Forman, 1987; Wang et al, 1997a;
Wang, 1998).  The issue relates to inhibition of cytochrome P-450-dependent synthesis of several potent
vasoconstricting agents  (Wang, 1998).   Vasodilation has been shown via this mechanism with high
concentrations of CO (ca. 90,000 ppm) (Coceani et al., 1988).  It is unclear, however, whether this could
arise under physiological conditions and CO concentrations produced endogenously.  The competition
between CO and O2 for cytochrome c oxidase was well outlined in the previous review (U. S. Environmental
Protection Agency, 1991), but some additional information has been published since then.  Based on its
Warburg partition coefficient of between 5 and 15, CO binding is favored only in situations where oxygen
tension is extremely low (Coburn and Forman, 1987). Carbon monoxide binding to cytochrome c oxidase
in vivo will occur when COHb is high (ca. 50%), a level that causes both systemic hypotension as well as
impaired oxygen delivery (Brown and Piantadosi, 1992).  Mitochondrial dysfunction, possibly linked to
cytochrome inhibition, has been shown to inhibit energy production, and it also may be related to enhanced
free radical production (Piantadosi etal, 1995,1997a). There has been no new information published since
the last air quality criteria document that pertains to the  effects of CO on dopamine P hydroxylase or
tryptophan oxygenase.

5.6.3  Free Radical Production
       Laboratory animal studies indicate that nitrogen-  and oxygen-based free radicals are generated
in vivo during CO exposures.  Exposure  to CO at concentrations of 20 ppm or more for 1  h will cause
platelets to become a source of the nitric oxide free radical  ('NO) in the systemic circulation of rats (Thorn
etal., 1994; Thorn and Ischiropoulos, 1997). Studies with cultured bovine pulmonary endothelial cells have
demonstrated that exposures to CO at concentrations as low as 20 ppm  cause cells to release 'NO, and the
exposure will cause death by a 'NO process that is manifested 18 to 24  h after the CO exposure (Thorn et
al., 1997; Thorn and Ischiropoulos, 1997).  The mechanism is based  on elevations in steady-state 'NO
concentration and production  of peroxynitrite (Thorn et al., 1994, 1997). Peroxynitrite is a relatively
long-lived, strong oxidant that  is produced by the near diffusion-limited reaction between 'NO and
superoxide radical (Huie and Padmaja, 1993).
       The mechanism by which CO concentrations of 11 nmol or more cause an elevation of steady-state
'NO concentrations appears to be based on altered intracellular "routing" of 'NO in endothelial cells and
platelets. It is well established that the association and dissociation rate constants of'NO with hemoproteins
exceed the rate constants for O2 or CO (Gibson et al., 1986). However, Moore and Gibson (1976) found
that when  CO was incubated with nitrosyl ('NO)-Mb or 'NO-Hb, CO slowly displaced the 'NO.  Carbon
monoxide replacement occurred even when there was excess 'NO-heme protein, and replacement rates were
enhanced by increasing the CO concentration or by carrying out the reaction in the presence of agents such
as thiols, which will react with the liberated 'NO. These conditions, including the presence of thiols, exist
in cells exposed to environmentally relevant concentrations of CO. Exposures to up to 1,070 nmol CO do
not alter the rate of production of'NO by platelets and endothelial cells, yet liberation of'NO was enhanced
by CO (Thorn and Ischiropoulos, 1997; Thorn et al., 1994; Thorn et al., 1997).
       Carbon monoxide will increase the concentration of'NO available to react with in vivo targets in
both lung and brain, based on electron paramagnetic resonance studies with rats exposed to 50 ppm CO or
more (Ischiropoulos et al., 1996;  Thorn et al., 1999a). The concentrations of nitric oxide synthase isoforms
in lung were not altered because of CO, and the mechanism  for elevation in 'NO was thought to be the same
as that found in isolated cells (Thorn et al., 1994,1997). Exposure to 50 to 100 ppm CO also  will increase
hydrogen peroxide (H2O2) production in lungs of rats (Thorn et al., 1999a). The phenomenon depended on
'NO production, as it was inhibited in rats pretreated with N°nitro-L-arginine methyl ester, a nitric oxide
synthase inhibitor.  Production of 'NO-derived oxidants also is increased in CO-exposed rats, based on
measurements of nitrotyrosine, a major product of the reaction of peroxynitrite with proteins (Ischiropoulos
et al., 1996; Thorn et al., 1998, 1999a,b).

                                             5-17

-------
       The mechanism for enhanced H2O2 production in lungs of CO-exposed rats is not clear.  It is
possible that 'NO or peroxynitrite may perturb mitochondrial function.  Peroxynitrite inhibits electron
transport at complexes I through IE, and 'NO targets cytochrome oxidase (Cassina and Radi, 1996;
Lizasoain et al., 1996; Poderoso  et al,  1996).  It is important to note, however, that alterations in
mitochondrial function and an increase of cellular H2O2 were not found in studies where cultured bovine
endothelial cells were exposed to similar CO concentrations (Thorn et al., 1997). An alternative possible
mechanism to mitochondrial dysfunction is that  exposure to CO may inhibit antioxidant  defenses.
Mechanisms linked to  elevations in 'NO could be responsible for inhibiting one or more enzymes. Nitric
oxide-derived oxidants can inhibit manganese superoxide dismutase  and glyceraldehyde-3-phosphate
dehydrogenase and deplete cellular stores of reduced glutathione (Ischiropoulos et al., 1992; Luperchio
etal., 1996).
       Exposure to  high CO concentrations (2,500 to 10,000 ppm)  cause mitochondria in brain cells to
generate hydroxyl-like radicals (Piantadosi et al., 1995,1997a). An additional source of partially reduced
O2 species found in animals exposed to CO is xanthine oxidase. Conversion of xanthine dehydrogenase,
the enzyme normally involved with uric acid metabolism, to xanthine oxidase, the radical-producing form
of the enzyme, occurred in the brains of rats exposed to approximately 3,000 ppm CO (Thorn, 1992).
Lower CO concentrations did not trigger this change.  Therefore, xanthine oxidase is unlikely to be a free
radical source following exposures to CO at concentrations found  in  ambient air.  Moreover, enzyme
conversion was not a primary effect of CO; rather, it occurred only following sequestration and activation
of circulating leukocytes (Thorn, 1993).

5.6.4 Stimulation of Guanylate Cyclase
       In recent years, CO has been demonstrated to play a physiological role in vasomotor control and
neuronal  signal transduction (Morita et al., 1995; Ingi et al., 1996).  Carbon monoxide is  produced
endogenously by oxidation of organic molecules, but the predominant source is from the degradation of
heme (Rodgers et al., 1994). The rate-limiting enzyme for heme metabolism is heme oxygenase (HO),
which converts heme to biliverdin, free iron, and CO. Three isoforms of HO have been characterized. The
HO-1 is an inducible enzyme found in vascular endothelial cells, smooth muscle cells, bronchoalveolar
epithelium, and pulmonary macrophages.  The HO-1 is induced by its substrate, heme, as well as 'NO,
H2O2, several cytokines, and lipopolysaccharide (Arias-Diaz et al., 1995; Durante et al., 1997; Morita et al.,
1995; Motterlini et al., 1996). The HO-2 is a constitutive enzyme  found in certain neurons within the
central nervous system, testicular cells, and vascular smooth muscle  cells (Marks, 1994). Little is known
about HO-3, which recently was identified in homogenates from a number of organs (McCoubrey et al.,
1997).
       A main physiological role for CO is thought to be regulation of soluble guanylate cyclase activity.
Both CO and 'NO can activate guanylate cyclase, although activation by CO is approximately 3 0-fold lower
(Stone and Marietta, 1994).  In neuronal cells possessing both heme oxygenase and nitric oxide synthase,
regulation of cyclic  guanosine monophosphate (cGMP) synthesis is mediated in a reciprocal fashion by
producing either CO or 'NO (Ingi et al.,  1996; Maines et al.,  1993). A compensatory interrelationship
between nitric oxide synthase and heme oxygenase also has been found in endothelial cells and activated
macrophages, although its functional  significance is unknown (Kurata et al.,  1996; Seki et al., 1997).
In macrophages, cGMP synthesis promotes chemotaxis, and cGMP-mediated synthesis and secretion of
tumor necrosis factor a has been linked to both CO and 'NO (Arias-Diaz et al., 1995; Belenky et al., 1993).
Carbon monoxide causes smooth muscle relaxation by stimulating soluble guanylate cyclase (Utz and
Ullrich, 1991; Wang et al., 1997b). Smooth muscle relaxation also may occur because of activation of
calcium dependent potassium channels, although this effect may be linked to guanylate cyclase activity
(Trischmann et al., 1991; Wang  et al., 1997a).  Carbon monoxide-mediated smooth muscle relaxation is
involved  with control of microvascular hepatic portal blood flow (Goda et al., 1998; Pannen and Bauer,

                                             5-18

-------
1998) and suppressing contractions in the gravid uterus (Acevedo and Ahmed, 1998). It also may play a
role in gastrointestinal motility (Farrugia et al., 1998).
5.7  Mechanisms of Carbon Monoxide Toxicity
5.7.1 Alterations in Blood Flow
       Carbon monoxide from environmental pollution may exert similar effects in vivo to those of
endogenously produced CO, because the nanomolar tissue concentrations resulting from inhalation of CO
are comparable or  greater  than concentrations produced by cells possessing heme oxygenase.  Liver
parenchyma has been estimated to generate approximately 0.45 nmol CO/gram liver/min (Goda et al.,
1998).  Carbon monoxide synthesis by smooth muscle cells is approximately 8 pmol/mg protein/min for
human aorta and 23 to 37 pmol/mg protein/min for rat aorta (Cook et al., 1995; Grundemar et al., 1995).
Carbon monoxide production by unstimulated pulmonary macrophages is 3.6 pmol/mg protein/min, and,
after stimulation with lipopolysaccharide, it increases to about 5.1 pmol/mg protein/min (Arias-Diaz et al.,
1995).  The rate of synthesis of CO varies widely for nerve cells.  Cerebellar granule  cells generate
approximately 3 fmol/mg protein/min, olfactory nerve cells produce 4.7 pmol/mg protein/min, and rat
cerebellar homogenates can generate as much as 56.6 pmol/mg protein/min (Ingi and Ronnett, 1995; Ingi
et al., 1996; Mames, 1988; Nathanson et al., 1995).
       Vasodilation is a well-established effect caused by exposure to environmental CO.  At high CO
concentrations, on the order of 500 to 2,000 ppm, the mechanism is related to impairment of O2 delivery
(Kanten et al., 1983; MacMillan, 1975).  However, a portion of the observed increases in cerebral blood
flow are independent of perturbations in O2 supply (Koehler et al., 1982).  In a setting where cellular
oxidative metabolism was not impaired by CO, elevations in cerebral blood flow appeared to be mediated
by 'NO (Meilin et al., 1996). Whether the mechanism was the same as that outlined in the section above,
which causes oxi dative stress, remains to be determined.
       It is unclear whether disturbances in vascular tone by environmental CO is a generalized, systemic
response, and the impact  of variables such as the duration of exposure have not been adequately
investigated. Although cerebral vasodilation mediated by 'NO was reported with exposures to 1,000 ppm
CO, that level of exposure did not alter pulmonary vasoconstriction in an isolated-perfused rat lung model
(Cantrell and Tucker, 1996). Exposure to 150,000 ppm CO caused no changes in pulmonary artery pressure
in isolated blood-perfused lungs, although CO did inhibit hypoxic pulmonary vasoconstriction (Tamayo et
al., 1997). Humans exposed to CO for sufficient time to achieve COHb levels of approximately 8% were
not found to have alterations in forearm blood flow, blood pressure, or heart  rate (Hausberg and Somers,
1997).
       Animals exposed to high CO concentrations (e.g., 3,000 to 10,000 ppm) have diminished organ
blood flow, which contributes to CO-mediated tissue injury (Brown and Piantadosi, 1992; Ginsberg and
Meyers, 1974; Okeda et al., 1981; Song et al., 1983; Thorn, 1990).  The mechanism is based on CO-
mediated hypoxic stress and cardiac dysfunction; therefore, these effects do not arise at CO concentrations
relevant to ambient air quality.

5.7.2 Mitochondria! Dysfunction and Altered Production of High-Energy Intermediates
       When exposed to 10,000 ppm CO, rats exhibit impaired high-energy phosphate synthesis and
production of hydroxyl free radicals because of mitochondrial dysfunction (Brown and Piantadosi, 1992;
Piantadosi et al., 1995).  Exposure to 2,500 ppm CO also will cause hydroxyl radicals to be produced,
apparently by mitochondria, because of a process that could not be related to hypoxic stress (Piantadosi
et al., 1997a).  Evidence for mitochondrial dysfunction  has not been observed in vivo at lower CO
concentrations.  However, under conditions of high metabolic demand, exposure to even 1,000 ppm CO
                                            5-19

-------
in the absence of an overt hypoxic stress will result in impaired energy production in brain (Meilin et al,
1996).
       Carbon  monoxide binding to mitochondrial cytochromes of respiring cells in vitro has been
documented only when either the CO concentration was extraordinarily high, or O2 tension was extremely
low, such that the CO/O2 ratio exceeded 12:1 (Coburn and Forman, 1987). Following CO exposure and
removal to fresh air, CO diffuses out from cells, and mitochondrial function is restored. This process is
enhanced by inspiration of hyperbaric oxygen (Brown and Piantadosi, 1992). Studies in mice indicate that
high CO concentrations inhibit synthesis of high-energy phosphates during exposure to 5,000 ppm CO for
15  min and these changes do not  persist  following removal to  fresh air  (Matsuoka et al., 1993).
In summary, mitochondrial dysfunction and impaired high-energy phosphate synthesis have been shown
by  several independent laboratories to  occur  during exposures to high CO concentrations.   Current
information suggests that this alteration does not occur at CO concentrations relevant to ambient air quality,
and that changes in energy production are not persistent for long periods of time following CO exposure.

5.7.3  Vascular Insults Associated with Exposure to Carbon Monoxide
       There are two primary variables that impact on CO toxicity.  One is the concentration of CO, the
other is the duration of exposure. Traditionally, these two variables have been viewed as merely alternative
ways of elevating COHb concentration in the body.  The concentration of CO breathed dictates the duration
of exposure required to achieve a particular blood level of COHb or tissue level of CO. This view is
predicated on the notion that  CO  pathophysiology is determined by its binding to one or  another
hemoprotein and to inhibition of oxygen delivery or oxidative metabolism.
       There is a substantial body of literature to suggest that, at least with regard to vascular effects, the
duration of exposure has a greater impact on the magnitude of CO pathophysiology than what is predicted
based on the concentration of CO that is inspired.  For example, the lungs are the first site for potential
action of environmental CO.  Results from  investigations have been conflicting regarding  the risk for
pulmonary injury from CO.  Because of the lack of consensus and also the absence of a recognized
biochemical mechanism, low concentrations of CO have been viewed as posing little risk to lung physiology
(U.S. Environmental Protection Agency, 1991, 1992).  When  animals have been exposed  to high CO
concentrations sufficient to raise COHb levels to 56  to 90%, exposures have lasted for only minutes because
of the hypoxic stress.  In these studies, evidence of increased capillary permeability was inconsistent (Fein
etal., 1980;NidenandSchulz, 1965; Penney etal,  1988), and no other alterations in lung physiology were
detected (Fisher et al., 1969; Robinson et al., 1985; Shimazu et al., 1990; Sugi et al., 1990).  In contrast,
when human  beings or experimental animals were exposed to CO for many hours at a time, capillary
leakage of macromolecules from the lungs and systemic vasculature has been documented, but the presence
of hypoxic stress was questioned (Kjeldsen et  al., 1972; Maurer, 1941; Parving et al., 1972; Siggaard-
Andersen et al., 1968).
       In light of the physiological role for CO in vasomotor control, protracted exposures may be prone
to disturb vascular homeostasis, giving rise to pathophysiological responses.  Monkeys exposed to 250 ppm
CO for 2 weeks  exhibited coronary artery damage consisting of subendothelial edema, fatty streaking, and
lipid-loaded cells (Thomsen, 1974).  This study and others (Armitage et al., 1976; Davies et al.,  1976;
Turner et al.,  1979; Webster et al., 1968) have suggested a link between atherosclerosis and chronic CO
exposure. However, other studies have failed to find evidence for an association (Hugod et al., 1978; Penn
etal., 1992).
       Carbon monoxide may cause vascular insults. Leakage of albumin and leukocyte sequestration have
been shown following exposures of rats to  50 ppm or more for 1 h, and the process is mediated by
'NO-derived oxi dants (Ischiropoulos etal., 1996; Thorn, 1993; Thorn etal., 1998,1999a,b). Brain oxidative
stress associated with this mechanism has been shown with rats exposed to 1,000 to 3,000 ppm CO for 1 h
(Ischiropoulos et al., 1996; Thorn, 1993). However, it is unclear whether the flux of'NO, resulting from

                                             5-20

-------
exposures to lower CO concentrations contribute to oxidative or nitrosative stress in vivo.  Important
differences in the patterns of leakage from pulmonary and systemic vascular beds suggest that they may be
caused by different mechanisms. For example, systemic vascular leakage was present for several hours after
CO exposure,  and the leakage resolved within 18 h,  whereas pulmonary vascular leakage was not
measurable until 18 h after CO exposure, and it resolved by 48 h.  Both pulmonary and systemic vascular
leakage occurred after hour-long exposures to CO, but not when exposures lasted for only 30 min, and
vascular changes were not different whether rats were exposed to just 50 ppm or as much as 1,000 ppm CO.
These  are  recent observations,  and further  investigations  are required before their  relevance to
environmental  CO contamination can be assessed adequately.  Moreover, it should be emphasized that the
vascular leakage  observed in  lungs and systemic microvasculature following exposures to  CO at
concentrations as low as 50 ppm may have no pathophysiological impact if regional lymphatics can sustain
a higher flow so that edema does not occur (Thorn et al, 1998, 1999a,b).
5.8  Other Effects of Carbon Monoxide
       Among the most concerning pathophysiological effects of CO is its propensity for causing brain
damage. There has been considerable effort focused on potential mechanisms for this process. With regard
to ambient air standards, however, it is important to note that recent studies were done with high CO
concentrations. Carbon monoxide poisoning is not a "pure" pathological process, as injuries may be
precipitated by a combination of cardiovascular effects linked to hypoperfusion or frank ischemia, COHb-
mediated hypoxic stress, and intracellular effects, including free radical production and oxidative stress.
For example, CO poisoning causes elevations  of glutamate and dopamine in experimental models and
human fatalities (Arranz et al., 1997; Ishimaru et al., 1991, 1992; Nabeshima et al., 1990, 1991; Newby et
al., 1978; Piantadosi et al., 1997b). These elevations occur because of the CO-associated cardiovascular
compromise and, possibly, other direct CO-mediated effects.  Based on the effects of agents that block the
N-methyl-D-aspartate (NMD A) receptor, elevations of glutamate in experimental CO poisoning have been
linked to a delayed type (but not an acute type) of amnesia, to loss of CA1 neurons in the hippocampus of
mice, and to loss of glutamate-dependent cells in the inner ear of rats (Ishimaru et al., 1991, 1992; Liu and
Fechter, 1995; Nabeshima et al., 1990,1991). Antioxidants can protect against CO-mediated cytotoxicity
of glutamate-dependent nerve cells (Fechter et al., 1997).  Mechanisms of glutamate neurotoxicity include
excessive  calcium influx, free-radical-mediated injury that may include calcium-calmodulin-dependent
activation of cytosolic NO synthase,  and lipid peroxidation.  Moderate stimulation by excitatory amino
acids may cause  mitochondrial dysfunction with impaired synthesis of adenosine triphosphate  and
production of reactive O2 species (Beal, 1992). Cell death can be through necrosis or programmed cell
death, depending on the intensity of the stimul us (Gwagetal., 1995). There also may be a synergistic injury
with other forms of oxidative stress because reactive O2 species can intensify excitotoxicity (Bridges et al.,
1991; Pellegrini-Giampietro et al., 1990). Glutamate also can injure cells in the central nervous system that
do not have NMDA receptors by competing for cysteine uptake, which inhibits synthesis of glutathione
(Lipton et al., 1997; Murphy et al., 1989; Oka et al., 1993).
5.9  Summary
       The most prominent pathophysiological effect of CO is hypoxemia caused by avid binding of CO
to Hb. Formation of COHb reduces O2-carrying capacity of blood and impairs release of O2 from O2FIb to
tissues. Failure of vasodilation to compensate causes tissue hypoxia. In addition to tissue hypoxia, ultimate
diffusion of CO to cells may affect adversely their function. The brain and heart tissues are particularly
sensitive to CO-induced hypoxia and cytotoxicity. The rate of COHb formation and elimination depends
on many physical and physiological factors.  The same factors that govern CO uptake determine CO

                                             5-21

-------
elimination as well.  The flow of CO between blood and either alveolar air or the tissues, and vice versa,
is  governed by the CO concentration gradient between these  compartments and  becomes the rate-
determining step in the mass transfer  of CO for COHb formation and elimination.  Because of a small
blood-to-air CO gradient and tight binding of CO to Hb, the elimination half-time is  quite long, varying
from 2.0 to 6.5 h.  Apart from the CO concentration in ambient air, the principal determinants of CO uptake
are minute ventilation and lung diffusion capacity.  Thus, any physiological conditions that affect these
variables (e.g., exercise, age) also will affect the kinetics of COHb. Both the physical and physiological
variables have been integrated into many empirical and mathematical models of COHb formation and
elimination under static and dynamic  conditions of ambient CO concentration and physiologic function.
The nonlinear CFK equation is the most widely used predictive model of COHb formation, and it still is
considered the best all-around model for COHb prediction. Altitude may have a significant influence on
COHb kinetics. The effects of hypoxic hypoxia (altitude) and CO-induced hypoxia appear to be additive.
Adaptation to altitude will moderate COHb formation. In addition to exogenous sources of CO, the gas also
is produced endogenously through catabolism of Hb, metabolic processes of drugs,  and degradation of
inhaled solvents  and other xenobiotics.   This last source may lead to very high (up to 50%) COHb
concentrations. Many disorders, particularly anemias of any etiology, will predispose affected individuals
to CO hypoxia. Furthermore, patients  with a variety of cardiopulmonary diseases (e.g., COPD, CAD) and
chronic inflammatory diseases may be at increased risk because of elevated  COHb.
        Apart from impaired O2 delivery to the tissues because of COHb  formation, recent studies of CO
pathophysiology  suggest cytotoxic effects  independent of O2.  New investigations have expanded the
understanding of CO in two areas. First, there is a growing recognition  of the role that CO may play in
normal neurophysiology and in microvascular vasomotor control.  The impact of CO from ambient air on
these processes has not been investigated adequately; hence, there is insufficient information available to
influence decisions on ambient air quality standards.  The second area of investigation of CO is related to
its propensity for causing free-radical-mediated changes in tissues. Mechanisms for these changes have
been  linked both to  mitochondria and to a CO-mediated disturbance  of intracellular "buffering" of
endogenously generated free radicals  (e.g., 'NO).  The role these mechanisms play in pathophysiology
currently is being investigated. Where dose-response studies are available, the concentrations of CO that
cause adverse effects in animals exceed current NAAQS.
References
Acevedo, C. H.; Ahmed, A. (1998) Hemeoxygenase-1 inhibits human myometrial contractility via carbon monoxide and is
      upregulated by progesterone during pregnancy. J. Clin. Invest. 101: 949-955.
Andersen, M. E.; Clewell, H. M., Ill; Gargas, M. L.; MacNaughton, M. G.; Reitz, R. H.; Nolan, R. J.; McKenna, M. J. (1991)
      Physiologically based pharmacokinetic modeling with dichloromethane, its metabolite, carbon monoxide, and blood
      carboxyhemoglobin in rats and humans. Toxicol. Appl. Pharmacol. 108: 14-27.
Arias-Diaz, J.; Vara, E.; Garcia, C.; Villa, N.; Balibrea, J. L. (1995) Evidence for a cyclic guanosine monophosphate-dependent,
      carbon monoxide-mediated,  signaling system in the regulation of TNF-ct production by human pulmonary macrophages.
      Arch. Surg. (Chicago) 130: 1287-1293.
Armitage, A. K.; Davies, R.  F.; Turner, D. M. (1976) The effects of carbon monoxide on the development of atherosclerosis in
      the White Carneau pigeon. Atherosclerosis 23: 333-344.
Arranz, B.; Blennow, K.; Eriksson,  A.; Mansson, J.-E.; Marcusson, J. (1997) Serotonergic, noradrenergic, and dopaminergic
      measures in suicide brains. Biol. Psychiatry 41: 1000-1009.
Barie, P. S.; Wu, W.; Hariri, R. J.; Halebian, P. H.; Shires, G. T. (1994) Alterations of pulmonary gas exchange after superimposed
      carbon monoxide poisoning  in acute lung injury. Surgery (St. Louis) 115: 678-686.
Beal, M. F. (1992) Does impairment of energy metabolism result in excitotoxic neuronal death in neurodegenerative illnesses?
      Ann. Neurol. 31: 119-130.
Belenky, S. N.; Robbins, R. A.; Rubinstein, I. (1993) Nitric oxide synthase inhibitors attenuate human monocyte chemotaxis in
      vitro. J. Leukocyte Biol. 53:  498-503.
                                                5-22

-------
Bender, P. R.; Weil, J. V.; Reeves, J. T.; Moore, L. G. (1987) Breathing pattern in hypoxic exposures of varying duration. J. Appl.
       Physiol. 62: 640-645.
Benignus, V. A. (1995) A model to predict carboxyhemoglobin and pulmonary parameters after exposure to O2, CO2, and CO.
       Aviat. Space Environ. Med. 66: 369-374.
Benignus, V. A.; Annau, Z. (1994) Carboxyhemoglobin formation due to carbon monoxide exposure in rats. Toxicol. Appl.
       Pharmacol. 128: 151-157.
Benignus, V. A.; Hazucha, M. J.; Smith, M. V.; Bromberg, P. A. (1994) Prediction of carboxyhemoglobin formation due to
       transient exposure to carbon monoxide. J. Appl. Physiol. 76: 1739-1745.
Berk, P. D.; Rodkey, F. L.; Blaschke, T. F.; Collison, H. A.; Waggoner, J. G. (1974) Comparison of plasma bilirubin turnover and
       carbon monoxide production in man. J. Lab. Clin. Med. 83: 29-37.
Berk, P. D.; Blaschke, T.  F.; Scharschmidt, B. F.; Waggoner, J.  G.; Berlin, N.  I. (1976) A new approach to quantitation of the
       various sources of bilirubin in man. J. Lab. Clin. Med. 87: 767-780.
Bernard, T. E.; Duker, J. (1981) Modeling carbon monoxide uptake during work. Am. Ind. Hyg. Assoc. J. 42: 361-364.
Biller,  W. F.; Richmond, H. M. (1982) Sensitivity analysis on Coburn model predictions of COHb levels associated with
       alternative CO standards. Research Triangle Park,  NC: U.S. Environmental Protection Agency, Office of Air Quality
       Planning and Standards; EPA contract no. 68-02-3600.
Biller, W. F.; Richmond, H. M. (1992) COHb module for a probabilistic CONEM. In: Johnson, T.; Capel, J.; Paul, R.; Wijnberg,
       L.  Estimation of carbon monoxide exposures and  associated carboxyhemoglobin levels in Denver residents using a
       probabilistic  version of NEM. Research Triangle Park, NC: U. S. Environmental Protection Agency, Office of Air Quality
       Planning and Standards; contract no. 68-DO-0062; appendix C.
Boothby, W. M.; Lovelace, W. R., II; Benson, O. O., Jr.; Strehler, A. F. (1954) Volume and partial pressures of respiratory gases
       at altitude. In: Boothby, W. M., ed. Respiratory physiology in aviation. Randolph Field, TX: U.S. Air Force School of
       Aviation Medicine; pp. 39-49.
Bridges, R. J.; Koh,  J.-Y.; Hatalski, C. G.; Cotman, C. W. (1991) Increased excitotoxic vulnerability of cortical cultures with
       reduced levels of glutathione. Eur. J. Pharmacol. 192: 199-200.
Britten, J. S.; Myers, R. A. M. (1985) Effects of hyperbaric treatment on carbon monoxide elimination  in humans. Undersea
       Biomed. Res. 12:431-438.
Brown, S. D.; Piantadosi, C. A. (1992) Recovery of energy metabolism in rat brain after carbon monoxide hypoxia. J. Clin. Invest.
       89: 666-672.
Bilnger, J.; Bombosch, F.; Mesecke, U.; Hallier, E. (1997) Monitoring and analysis of occupational exposure  to chain saw
       exhausts. Am. Ind. Hyg. Assoc. J. 58: 747-751.
Burki, N. K. (1984) Effects of acute exposure to high altitude on ventilatory drive and respiratory pattern. J. Appl. Physiol. Respir.
       Environ. Exercise Physiol. 56: 1027-1031.
Cantrell, J. M.; Tucker, A. (1996) Low-dose carbon monoxide does not reduce vasoconstriction in isolated rat lungs. Exp. Lung
       Res. 22: 21-32.
Cassina, A.; Radi, R. (1996) Differential inhibitory action of nitric oxide and peroxynitrate on mitochondrial electron transport.
       Arch. BiochemBiophys. 328: 309-316.
Cavallin-Stahl, E.; Mercke, C.; Lundh, B. (1976) Carbon monoxide production in patients with breast carcinoma. Br. J. Haematol.
       32: 177-182.
Chung, S. J. (1988)  Formulas predicting carboxyhemoglobin resulting from carbon monoxide exposure. Vet. Hum. Toxicol.
       30: 528-532.
Coburn, R. F. (1967) Endogenous carbon monoxide production and body CO stores. Acta Med. Scand. Suppl. 472: 269-282.
Coburn, R. F. (1970a) The carbon monoxide body stores. In: Coburn, R. F., ed. Biological effects of carbon monoxide. Ann. N. Y.
       Acad. Sci.  174: 11-22.
Coburn, R. F. (1970b) Enhancement by phenobarbital and diphenylhydantoin of carbon monoxide production in normal man.
       N.Engl.J. Med. 283:512-515.
Coburn, R. F. (1973) Mean myoglobin oxygen tension in skeletal and cardiac muscle. In: Bicher, H. I.;Bruley, D. F., eds. Oxygen
       transport to tissue: instrumentation, methods, and physiology. New York, NY: Plenum Press; pp. 571 -577. (Advances in
       experimental medicine and biology: v. 37A-B).
Coburn, R. F.;Forman, H. J. (1987)  Carbon monoxide toxicity. In: Fishman, A. P.;Farhi, L. E.;Tenney, S. M.;Geiger, S. R.,eds.
       Handbook of physiology: a critical, comprehensive presentation of physiological knowledge and concepts.  Section 3: the
       respiratory system. Volume IV. Gas exchange. Bethesda, MD: American Physiological Society; pp. 439-456.
Coburn, R. F.; Mayers, L. B. (1971) Myoglobin O2 tension determined from measurements of carboxymyoglobin in skeletal
       muscle. Am.  J. Physiol. 220: 66-74.
Coburn, R. F.; Blakemore, W.  S.;  Forster, R.  E. (1963) Endogenous carbon monoxide production  in  man. J. Clin. Invest.
       42: 1172-1178.
Coburn, R. F.; Williams, W. J.; Forster, R. E. (1964) Effect of erythrocyte destruction on carbon monoxide production in man.
       J. Clin. Invest. 43: 1098-1103.
                                                      5-23

-------
Coburn, R. F.;  Forster, R. E.; Kane, P. B. (1965) Considerations of the physiological variables that determine the blood
       carboxyhemoglobin concentration in man. J. Clin. Invest. 44: 1899-1910.
Coburn, R. F.; Williams, W. J.;Kahn, S. B. (1966) Endogenous carbon monoxide production in patients with hemolytic anemia.
       J. Clin. Invest. 45: 460-468.
Coceani, F.; Breen, C. A.; Lees, J. G.; Falck, J. R.; Olley, P. M. (1988) Further evidence implicating a cytochrome P-450-mediated
       reaction in the contractile tension of the lamb ductus arteriosus. Circ. Res.  62: 471-477.
Cook, M. N.; Nakatsu, K.; Marks, G. S.; McLaughlin, B. E.; Vreman, H. J.; Stevenson, D. K.; Brien, J. F. (1995) Heme oxygenase
       activity  in the adult rat aorta and  liver as measured by carbon monoxide formation. Can. J. Physiol. Pharmacol.
       73:515-518.
Davies, R. F.; Topping, D. L.; Turner, D. M. (1976)  The  effect of intermittent carbon monoxide exposure on experimental
       atherosclerosis in the rabbit. Atherosclerosis 24: 527-536.
Delivoria-Papadopoulos, M.; Coburn, R. F.; Forster, R. E.  (1974) Cyclic variation of rate of carbon monoxide production in
       normal women. J.  Appl. Physiol. 36: 49-51.
Deller, A.; Stenz, R.; Forstner, K.; Konrad, F. (1992) Die elimination von kohlenmonoxydhamoglobin - geschlechtsspezifische
       und zirkadiane einflusse [The elimination of carboxyhemoglobin COHB sex-specific and circadian influences in healthy
       volunteers]. Infusionstherapie 19: 121-126.
Durante, W.; Kroll, M. H.; Christodoulides, N.; Peyton, K. J.; Schafer, A. I. (1997) Nitric oxide induces heme oxygenase-1 gene
       expression and carbon monoxide production in vascular smooth muscle cells. Circ. Res. 80: 557-564.
Engel, L. A.; Wood, L. D. H.; Utz, G.; Macklem, P. T.  (1973) Gas mixing during inspiration. J. Appl. Physiol. 35: 18-24.
Farrugia, G.; Miller, S. M.; Rich, A.; Liu, X.; Maines, M. D.;  Rae, J. L.; Szurszewski, J. H. (1998) Distribution of heme oxygenase
       and effects of exogenous carbon monoxide in canine jejunum. Am. J. Physiol. 274: G350-G358.
Fechter, L. D.; Liu, Y.; Pearce, T. A. (1997) Cochlear protection from carbon monoxide exposure by free radical blockers in the
       guinea pig. Toxicol. Appl. Pharmacol. 142: 47-55.
Fein, A.; Grossman, R. F.;  Jones, J. G.; Hoeffel, J.; McKay, D. (1980) Carbon monoxide effect on alveolar epithelial permeability.
       Chest 78: 726-731.
Fisher, A.  B.; Hyde, R. W.; Baue, A. E.; Reif, J. S.; Kelly, D. F. (1969) Effect of carbon monoxide on function and structure of
       the lung. J. Appl. Physiol. 26: 4-12.
Forbes, W. H.;  Sargent, F.; Roughton, F. J. W.  (1945) The rate of carbon monoxide uptake by normal men. Am. J. Physiol.
       143: 594-608.
Forster, R. E. (1987) Diffusion of gases across the alveolar  membrane. In: Fishman, A. P.; Farhi, L. E.; Tenney, S. M.; Geiger,
       S.  R., eds. Handbook of physiology: a critical, comprehensive presentation of physiological knowledge and concepts.
       Section 3: the respiratory system. Volume IV. Gas exchange. Bethesda, MD: American Physiological Society; pp. 71-88.
Foster, L.  J.; Corrigan, K.; Goldman, A. L. (1978) Effectiveness of oxygen therapy in hypoxic polycythemic  smokers. Chest
       73: 572-576.
Frey, T. M.; Crapo, R. O.; Jensen, R. L.; Elliott, C. G.  (1987) Diurnal variation of the diffusing capacity of the lung: is it real?
       Am. Rev. Respir. Dis. 136: 1381-1384.
Ge, R.-L.; Matsuzawa, Y.; Takeoka, M.; Kubo, K.; Sekiguchi, M.; Kobayashi, T.  (1997) Low pulmonary diffusing capacity in
       subjects with acute mountain sickness. Chest 111: 58-64.
Gibson, Q. H.;  Olson, J.  S.; McKinnie, R. E.; Rohlfs, R.  J. (1986) A kinetic description of ligand binding to sperm whale
       myoglobin. J. Biol. Chem. 261:  10,228-10,236.
Ginsberg, M. D.; Myers, R. E. (1974) Experimental carbon monoxide encephalopathy in the primate. I. Physiologic and metabolic
       aspects.  Arch. Neurol. 30: 202-208.
Goda, N.; Suzuki, K.; Naito, M.; Takeoka, S.; Tsuchida, E.; Ishimura, Y.; Tamatani, T.; Suematsu, M. (1998) Distribution of heme
       oxygenase isoforms in rat liver: topographic basis for carbon monoxide-mediated micro vascular relaxation. J. Clin. Invest.
       101:604-612.
Gothert, M.; Lutz, F.; Malorny, G. (1970) Carbon monoxide partial pressure in tissue of different animals. Environ. Res.
       3: 303-309.
Granlund, J.; Garby, L. (1984) Numerical values of the classical Haldane coefficient. J. Appl. Physiol. 57: 850-859.
Grote, J.; Ball, P.; Oltmanns, K.; Stolp, W. (1994) The effect of increased blood carbon monoxide levels on the hemoglobin
       oxygen  affinity  during pregnancy.  In:  Vaupel, P.; Zander R.;  Bruley, D. F., eds.  Oxygen transport to tissue XV.
       Proceedings of the twentieth annual meeting of the International Society on Oxygen Transport to Tissue; August 1992;
       Mainz, Germany. New York, NY: Plenum Press; pp.  145-150. (Advances in experimental medicine and biology: v. 345).
Grundemar, L.;  Johansson, M.-B.; Ekelund, M.; Hogestatt, E. D. (1995) Haem oxygenase activity in blood vessel homogenates
       as measured by carbon monoxide production. Acta Physiol. Scand. 153: 203-204.
Guenard, H.; Marthan, R.  (1996) Pulmonary gas exchange in elderly subjects. Eur. Respir. J.  9: 2573-2577.
Gwag, B. J.;Lobner, D.;Koh, J. Y.; Wie, M. B.;Choi, D. W.  (1995) Blockade of glutamate receptors unmasks neuronal apoptosis
       after oxygen-glucose deprivation in vitro. Neuroscience 68: 615-619.
                                                      5-24

-------
Haab, P. E.; Durand-Arczynska, W. Y. (1991) Carbon monoxide effects on oxygen transport. In: Crystal, R. G.; West, J. B.;
       Barnes, P. J.; Cherniack, N. S.; Weibel, E. R., eds. The lung: scientific applications, v. 2. New York, NY: Raven Press;
       pp. 1267-1275.
Haldane, J. (1897-1898) Some improved methods of gas analysis. J. Physiol. (London) 22: 465-480.
Harf, A.; Pratt, T.; Hughes, J. M. B. (1978) Regional distribution of VA/Q in man at rest and with exercise measured with
       krypton-Sim. J. Appl. Physiol.: Respir. Environ. Exercise Physiol. 44: 115-123.
Hauck, H.; Neuberger, M. (1984) Carbon monoxide uptake and the resulting carboxyhemoglobin in man. Eur. J. Appl. Physiol.
       53: 186-190.
Hausberg, M.; Somers, V. K.  (1997) Neural  circulatory responses  to carbon monoxide in healthy humans.  Hypertension
       29: 1114-1118.
Hill, E. P.; Hill, J. R.; Power, G. G.; Longo, L. D. (1977) Carbon monoxide exchanges between the human fetus and mother:
       a mathematical model. Am. J. Physiol. 232: H311-H323.
Horvath, S. M.; Agnew, J. W.; Wagner, J. A.; Bedi, J. F. (1988) Maximal aerobic capacity at several ambient concentrations of
       carbon monoxide at several altitudes. Cambridge, MA: Health Effects Institute; research report no. 21.
Horvath, I; Loukides, S.; Wodehouse, T.; Kharitonov, S. A.; Cole, P. J.; Barnes, P. J. (1998a) Increased levels of exhaled carbon
       monoxide in bronchiectasis: a new marker of oxidative stress.  Thorax 53: 867-870.
Horvath, I; Donnelly, L. E.; Kiss, A.; Paredi,  P.; Kharitonov, S. A.; Barnes, P. J. (1998b) Raised levels of exhaled  carbon
       monoxide are associated with an increased expression of heme oxygenase-1 in airway macrophages in asthma: a new
       marker of oxidative stress. Thorax 53: 668-672.
Hugod, C.; Hawkins, L. H.; Kjeldsen, K.; Thomsen, H. K.; Astrup, P. (1978) Effect of carbon monoxide exposure on aortic and
       coronary intimal morphology in the rabbit. Atherosclerosis 30: 333-342.
Huie, R. E.; Padmaja, S. (1993) The reaction of NO with superoxide. Free Radical Res. Commun. 18: 195-199.
Ingi, T.; Ronnett, G. V. (1995) Direct demonstration of a physiological role  for carbon monoxide in olfactory receptor neurons.
       J. Neurosci. 15:8214-8222.
Ingi, T.; Cheng, J.; Ronnett, G. V. (1996) Carbon monoxide: an endogenous modulator of the nitric oxide-cyclic GMP signaling
       system. Neuron 16: 835-842.
Ingram, D.; Dickinson, C.  J.; Ahmed, K. (1987) MACPUF software package. London, United Kingdom: Centre for  Health
       Informatics and Multiprofessional Education (CHIME), University College London, Medical School and Health Sciences
       Centre, McMaster University. Available: www.chime.ucl.ac.uk/Models/macpuf.htm.
Ischiropoulos, H.; Zhu, L.; Chen, J.; Tsai, M.; Martin, J.  C.; Smith, C. D.; Beckman, J. S. (1992) Peroxynitrate-mediated tyrosine
       nitration catalyzed by superoxide dismutase. Arch. Biochem. Biophys. 298: 431-437.
Ischiropoulos, H.; Beers, M. F.; Ohnishi, S. T.; Fisher, D.; Garner,  S. E.; Thorn, S.  R. (1996) Nitric oxide production and
       perivascular tyrosine nitration in brain after carbon monoxide poisoning in the rat. J. Clin. Invest. 97: 2260-2267.
Ishimaru, H.; Nabeshima, T.; Katoh, A.; Suzuki, H.; Fukuta, T.; Kameyama, T. (1991) Effects of successive carbon monoxide
       exposures on delayed neuronal death in mice under the maintenance of normal body temperature. Biochem. Biophys. Res.
       Commun. 179: 836-840.
Ishimaru, H.; Katoh, A.; Suzuki, H.; Fukuta, T.; Kameyama, T.; Nabeshima, T. (1992) Effects of N-methyl-D-aspartate receptor
       antagonists on carbon monoxide-induced brain damage in mice. J. Pharmacol. Exp. Ther. 261: 349-352.
Jay, G. D.;McKindley, D. S. (1997) Alterations in pharmacokinetics of carboxyhemoglobin produced by oxygen under pressure.
       Undersea Hyperbaric Med. 24: 165-173.
Johnson, T.; Capel, J.; Paul, R.; Wijnberg, L. (1992) Estimation of carbon monoxide exposures and associated carboxyhemoglobin
       levels in Denver residents using a probabilistic version  of NEM.  Research Triangle Park, NC: U.S. Environmental
       Protection Agency, Office of Air Quality Planning and Standards; contract no. 68-DO-0062.
Joumard, R.; Chiron, M.; Vidon, R.; Maurin, M.; Rouzioux, J.-M. (1981) Mathematical models of the uptake of carbon monoxide
       on hemoglobin at low carbon monoxide levels. Environ. Health Perspect. 41: 277-289.
Kanten, W. E.; Penney, D. G.; Francisco, K.; Thill, J. E. (1983) Hemodynamic responses to acute carboxyhemoglobinemia in the
       rat. Am. J. Physiol. 244: H320-H327.
Kim, S. K.; Kim, Y. C. (1996) Effect of a single administration of benzene, toluene or m-xylene on carboxyhaemoglobin elevation
       and metabolism of dichloromethane in rats. J. Appl.  Toxicol.  16: 437-444.
Kimmel, E. C.; Carpenter, R. L.; Reboulet, J. E.; Still, K. R. (1999) A physiological model for predicting carboxyhemoglobin
       formation from exposure to carbon monoxide in rats. J.  Appl. Physiol. 86:  1977-1983.
Kinker, J. R.; Haffor, A.-S.; Stephan, M.; Clanton, T. L. (1992) Kinetics of CO uptake and diffusing capacity in transition from
       rest to steady-state exercise. J. Appl. Physiol. 72: 1764-1772.
Kjeldsen, K.; Astrup, P.; Wanstrup, J. (1972) Ultrastructural intimal changes in the rabbit aorta after a moderate carbon monoxide
       exposure. Atherosclerosis 16:  67-82.
Kleinman, M. T.; Leaf, D. A.; Kelly, E.; Caiozzo, V.; Osann, K.; OTSIiell, T. (1998) Urban angina in the mountains: effects of
       carbon monoxide and mild hypoxemia on subjects with chronic stable angina. Arch. Environ. Health 53: 388-397.
                                                      5-25

-------
Klimisch, H. -J.; Chevalier, H. -J.; Harke, H. -P.; Dontenwill, W. (1975) Uptake of carbon monoxide in blood of miniature pigs and
       other mammals. Toxicology 3: 301-310.
Koehler, R. C.; Jones, M. D., Jr.; Traystman, R. J. (1982) Cerebral circulatory response to carbon monoxide andhypoxic hypoxia
       in the lamb. Am. J. Physiol. 243: H27-H32.
Kurata, S.; Matsumoto, M.; Yamashita, U. (1996) Concomitant transcriptional activation of nitric oxide  synthase and heme
       oxygenase genes during nitric oxide-mediated macrophage cytostasis. J. Biochem. 120: 49-52.
Landaw, S. A. (1973) The effects of cigarette smoking on total body burden and excretion rates of carbon monoxide. J. Occup.
       Med. 15:231-235.
Landaw, S. A.; Callahan, E. W., Jr.; Schmid, R. (1970) Catabolism of heme in vivo: comparison of the simultaneous production
       of bilirubin and carbon monoxide. J. Clin. Invest. 49: 914-925.
Levasseur, L.; Galliot-Guilley, M.; Richter, F.; Scherrmann, J. M.; Baud, F. J. (1996) Effects of mode of inhalation of carbon
       monoxide and of normobaric oxygen administration on carbon monoxide elimination from the blood. Hum. Exp. Toxicol.
       15:  898-903.
Levitt, M. D.; Ellis, C.; Levitt, D. G. (1994) Diurnal rhythm of heme turnover assessed by breath carbon monoxide concentration
       measurements. J. Lab. Clin. Med.  124: 427-431.
Lifshay, A.; Fast, C. W.; Glazier, J. B. (1971) Effects of changes in respiratory pattern on physiological dead space. J.  Appl.
       Physiol. 31:478-483.
Lipton, S. A.; Kim, W.-K.; Choi, Y.-B.; Kumar, S.; D'Emiha, D.  M.; Rayudu, P.  V.; Arnelle, D. R.; Stamler, J. S. (1997)
       Neurotoxicity associated with dual actions of homocysteine at the jV-methyl-D-aspartate receptor. Proc. Natl. Acad. Sci.
       U. S. A. 94: 5923-5928.
Liu,  Y.; Fechter, L. D. (1995) MK-801 protects against carbon monoxide-induced hearing loss. Toxicol. Appl. Pharmacol.
       132: 196-202.
Lizasoain, I; Moro, M. A.; Knowles, R. G.; Darley-Usmar, V.; Moncada, S. (1996) Nitric oxide and peroxynitrate exert distinct
       effects on mitochondrial respiration which are differentially blocked by glutathione or glucose. Biochem. J. 314: 877-880.
Longo, L. D.; Hill, E. P. (1977) Carbon monoxide uptake and elimination in fetal and maternal sheep. Am. J. Physiol.
       232: H324-H330.
Ludbrook, G. L.; Helps, S. C.; Gorman, D. F.; Reilly, P. L.; North, J. B.; Grant, C. (1992) The relative effects of hypoxic hypoxia
       and carbon monoxide on brain function in rabbits. Toxicology 75: 71-80.
Lundh, B.; Cavallin-Stahl, E.; Mercke, C. (1975) Nicotinic acid and the endogenous production of carbon monoxide. Acta Med.
       Scand. 197: 173-176.
Luomanmaki, K.; Coburn, R. F. (1969) Effects of metabolism and distribution of carbon monoxide on blood and body stores.
       Am. J. Physiol. 217:  354-363.
Luperchio, S.; Tamir, S.; Tannenbaum, S. R. (1996) NO-induced oxidative stress and glutathione metabolism in rodent and human
       cells. Free Radical Biol. Med. 21:  513-519.
Lynch, S. R.; Moede, A. L. (1972) Variation in the rate of endogenous carbon monoxide production in normal human beings.
       J. Lab. Clin. Med. 79: 85-95.
MacMillan, V. (1975) Regional cerebral blood flow of the rat in acute carbon monoxide intoxication. Can. J. Physiol. Pharmacol.
       53:644-650.
Maines, M. D. (1988) Heme oxygenase:  function, multiplicity, regulatory mechanisms, and clinical applications. FASEB J.
       2: 2557-2568.
Maines, M. D.; Mark, J. A.; Ewing, J. F. (1993) Heme oxygenase, a likely regulator of cGMP production in the brain: induction
       in vivo of HO-1 compensates for depression in NO synthase activity. Mol. Cell. Neurosci. 4: 398-405.
Manno, M.; Rugge, M.; Cocheo, V. (1992) Double fatal inhalation of dichloromethane. Hum. Exp. Toxicol. 11: 540-545.
Marks, G. S. (1994) Heme oxygenase: the physiological role of one of its metabolites, carbon monoxide and interactions with zinc
       protoporphyrin, cobalt protoporphyrin and other metalloporphyrins.  Cell. Mol. Biol. (Paris) 40: 863-870.
Marthan, R.; Castaing, Y.; Manier, G.; Guenard, H. (1985)  Gas exchange alterations in patients with chronic obstructive lung
       disease. Chest 87: 470-475.
Matsuoka, M.; Igisu, H.; Tanaka, I.; Hori, H.; Koga, M. (1993) Brain energy metabolites in mice after an acute exposure to carbon
       monoxide. Res. Commun. Chem. Pathol. Pharmacol. 81: 15-20.
Maurer, F. W. (1941) The effects of carbon monoxide anoxemia on the flow and composition of cervical lymph. Am. J. Physiol.
       133: 170-179.
McCartney, M. L. (1990) Sensitivity analysis applied to Coburn-Forster-Kane models of carboxyhemoglobin formation. Am. Ind.
       Hyg. Assoc. J. 51: 169-177.
McClean, P. A.; Duguid, N. J.; Griffin, P. M.; Newth, C. J. L.; Zamel, N. (1981) Changes in exhaled pulmonary diffusing capacity
       at rest and exercise in individuals with impaired positional diffusion. Clin. Respir. Physiol. 17: 179-186.
McCoubrey, W. K.,  Jr.; Huang, T. J.; Maines, M. D. (1997) Isolation and characterization of a cDNA from the rat brain that
       encodes hemoprotein heme oxygenase-3. Eur. J. Biochem. 247: 725-732.
McGrath, J. J. (1988) Carbon monoxide studies at high altitude. Neurosci. Biobehav. Rev. 12: 311-314.
                                                      5-26

-------
McGrath, J. J. (1992) Effects of altitude on endogenous carboxyhemoglobin levels. J. Toxicol. Environ. Health 35: 127-133.
McGrath, J. J.; Schreck, R. M.; Lee, P. S. (1993) Carboxyhemoglobin levels in humans: effects of altitude. Inhalation Toxicol.
       5: 241-249.
Meilin, S.; Rogatsky, G. G.; Thorn, S. R.; Zarchin, N.; Guggenheimer-Furman, E.; Mayevsky, A. (1996) Effects of carbon
       monoxide on the brain  may be mediated by nitric acid. J. Appl. Physiol. 81: 1078-1083.
Mercke, C.; Lundh, B. (1976) Erythrocyte filterability and heme catabolism during the menstrual cycle. Ann. Intern.  Med.
       85: 322-324.
Mercke, C.; Cavallin-Stahl, E.; Lundh, B. (1975a) Diurnal variation in endogenous production of carbon monoxide. Acta Med.
       Scand. 198: 161-164.
Mercke, C.; Cavallin-Stahl, E.; Lundh, B. (1975b) Carbon monoxide production and reticulocyte count in normal women: effect
       of contraceptive drugs and smoking. Acta Med. Scand. 198: 155-160.
Mercke, C.; Cavallin-Stahl, E.; Lundh, B. (1975c) Heme catabolism during short-term treatment with phenobarbital, diazepam
       and oxazepam. Acta Med. Scand. 198: 149-154.
Messmer, K. (1982) Oxygen transport capacity. In: Brendel, W.; Zink, R. A., eds. High altitude physiology and medicine.
       New York, NY: Springer-Verlag; pp. 117-122.
Meyer,  J.;  Prien,  T.; Van Aken, H.;  Bone,  H.-G.;  Waurick,  R.; Theilmeier,  G.; Booke,  M.  (1998)  Arterio-venous
       carboxyhemoglobin difference suggests carbonmonoxide production by human lungs. Biochem. Biophys. Res. Commun.
       244: 230-232.
Moore, E. G.; Gibson, Q.  H. (1976) Cooperativity  in  the dissociation of  nitric oxide from hemoglobin.  J. Biol. Chem.
       251:2788-2794.
Morita, T.; Perrella, M. A.; Lee, M.-E.; Kourembanas,  S. (1995) Smooth muscle cell-derived carbon monoxide is a regulator of
       vascular cGMP. Proc. Natl.  Acad. Sci. U. S. A. 92: 1475-1479.
Motterlini, R.; Foresti, R.; Intaglietta, M.;  Winslow, R.  M. (1996) NO-mediated activation of heme oxygenase: endogenous
       cytoprotection against oxidative stress to endothelium. Am. J. Physiol. 270: H107-H114.
Muller, K. E.; Barton, C. N. (1987) A nonlinear version of the Coburn, Forster and Kane model of blood carboxyhemoglobin.
       Atmos. Environ. 21: 1963-1967.
Murphy, T. H.; Miyamoto, M.; Sastre, A.; Schnaar, R. L.;  Coyle, J. T. (1989) Glutamate toxicity in a neuronal cell line involves
       inhibition of cystine transport leading to oxidative stress. Neuron 2: 1547-1558.
Nabeshima, T.; Yoshida, S.; Morinaka, H.; Kameyama, T.; Thurkauf, A.; Rice, K. C.; Jacobson, A. E.; Monn, J. A.; Cho, A. K.
       (1990)MK-801 ameliorates delayed amnesia, but potentiates acute amnesia induced by CO. Neurosci. Lett.  108: 321-327.
Nabeshima, T.; Katoh, A.; Ishimaru, H.;  Yoneda, Y.; Ogita, K.; Murase, K.; Ohtsuka, H.; Inari, K.; Fukuta, T.; Kameyama, T.
       (1991) Carbon monoxide-induced delayed amnesia, delayed neuronal death and change in acetylcholine concentration in
       mice. J. Pharmacol. Exp.  Ther. 256: 378-384.
Nathanson, J. A.; Scavone, C.; Scanlon, C.; McKee, M. (1995) The cellular Na+pump as a site of action for carbonmonoxide and
       glutamate: a mechanism for long-term modulation of cellular activity. Neuron 14: 781-794.
Neas, L. M.; Schwartz, J. (1996) The determinants of pulmonary diffusing capacity in a national sample of U.S. adults. Am. J.
       Respir. Grit. Care Med. 153: 656-664.
Newby, M. B.; Roberts, R. J.; Bhatnagar, R. K. (1978) Carbon monoxide- and hypoxia-induced effects on catecholamines in the
       mature and developing rat brain. J. Pharmacol.  Exp. Ther. 206: 61-68.
Niden, A. H.; Schulz, H. (1965) The ultrastructural effects of carbon monoxide inhalation on the rat lung. Virchows Arch. Pathol.
       Anat. Physiol. 339: 283-292.
Oka, A.; Belliveau, M. J.; Rosenberg, P. A.; Volpe, J. J. (1993) Vulnerability of oligodendroglia to glutamate: pharmacology,
       mechanisms, and prevention. J. Neurosci. 13: 1441-1453.
Okeda, R.; Funata, N.; Takano, T.; Miyazaki, Y.; Higashino, F.; Yokoyama, K.; Manabe, M. (1981) The pathogenesis of carbon
       monoxide encephalopathy in the acute phase—physiological and morphological correlation. ActaNeuropathol. 54: 1-10.
Ott, W. R.; Mage, D.  T. (1978) Interpreting urban carbon monoxide concentrations by means of a computerized blood COHb
       model. J. AirPollut.  Control Assoc. 28: 911-916.
Pace, N.; Strajman, E.; Walker, E.  L. (1950) Acceleration of carbon  monoxide elimination in man by high pressure  oxygen.
       Science (Washington, DC)  111: 652-654.
Pankow, D. (1996) Carbon monoxide formation due to metabolism of xenobiotics. In: Penney, D. G., ed. Carbon monoxide. Boca
       Raton, FL: CRC Press; pp. 25-43.
Pannen, B. H.; Bauer, M. (1998) Differential regulation of hepatic arterial and portal venous vascular resistance by nitric oxide
       and carbon monoxide in rats. Life Sci. 62: 2025-2033.
Parving, H.-H.;  Ohlsson, K.; Buchardt Hansen, H. J.; Rorth, M. (1972) Effect  of carbon monoxide exposure on capillary
       permeability to albumin and a2-macroglobulin. Scand. J. Clin. Lab. Invest. 29: 381-388.
Pellegrini-Giampietro, D. E.; Cherici, G.; Alesiani, M.; Carla, V.; Moroni, F. (1990) Excitatory amino acid release and free radical
       formation may cooperate in the genesis of ischemia-induced neuronal damage. J. Neurosci.  10: 1035-1941.
                                                      5-27

-------
Perm, A.; Currie, J.; Snyder, C. (1992) Inhalation of carbon monoxide does not accelerate arteriosclerosis in cockerels. Eur. J.
       Pharmacol. 228: 155-164.
Penney, D. G.; Davidson, S. B.; Gargulinski, R. B.; Caldwell-Ayre, T. M. (1988) Heart and lung hypertrophy, changes in blood
       volume, hematocrit and plasma renin activity in rats chronically exposed to increasing carbon monoxide concentrations.
       J. Appl. Toxicol. 8: 171-178.
Peters, T.; Jilrgens, K. D.; Gunther-Jilrgens, G.; Gros, G. (1994) Determination of myoglobin-diffusivity in intact skeletal muscle
       fibers: an improved microscope-photometrical approach. In: Vaupel, P.; Zander R.; Bruley, D. F., eds. Oxygen transport
       to tissue XV. Proceedings of the twentieth annual meeting of the International Society on Oxygen Transport to Tissue;
       August 1992; Mainz, Germany. New York, NY: Plenum Press; pp. 677-683. (Advances in experimental medicine and
       biology: v. 345).
Peterson, J. E.; Stewart, R. D.  (1970) Absorption and elimination of carbon monoxide by inactive young men. Arch. Environ.
       Health 21: 165-171.
Peterson, J. E.; Stewart, R. D. (1975) Predicting the carboxyhemoglobin levels resulting from carbon monoxide exposures. J. Appl.
       Physiol. 39: 633-638.
Piantadosi, C. A.; Tatro, L.; Zhang, J. (1995) Hydroxyl radical production in the brain after CO hypoxia in rats. Free Radical Biol.
       Med. 18:  603-609.
Piantadosi, C. A.; Zhang, J.; Demchenko, I. T. (1997a) Production of hydroxyl radical in the hippocampus after CO hypoxia or
       hypoxic hypoxia in the rat. Free Radical Biol. Med. 22: 725-732.
Piantadosi, C. A.; Zhang, J.; Levin, E. D.; Folz, R. J.; Schmechel, D. E. (1997b) Apoptosis and delayed neuronal damage after
       carbon monoxide poisoning in the rat. Exp. Neurol.  147: 103-114.
Poderoso, J.  J.; Cerreras, M. C.; Lisdero, C.; Riobo, N;  Schopfer, F.; Boveris, A. (1996) Nitric oxide inhibits electron transfer
       and increases superoxide radical production in rat heart mitochondria and submitochondrial particles.  Arch. Biochem.
       Biophys.  328: 85-92.
Robinson, N. B.; Barie, P. S.; Halebian, P. H.; Shires, G. T. (1985) Distribution of ventilation and perfusion following acute
       carbon monoxide poisoning. In: 41st annual forum on fundamental surgical problems held at the 71st annual clinical
       congress of the American College of Surgeons; October; Chicago,  IL. Surg. Forum 36: 115-118.
Rodgers, P. A.; Vreman, H. J.;  Dennery, P. A.; Stevenson, D. K. (1994) Sources of carbon monoxide (CO) in biological systems
       and applications of CO detection technologies. Semin. Perinatol. 18: 2-10.
Rodkey, F. L.; O'Neal, J.  D.; Collison, H. A.  (1969) Oxygen and carbon monoxide equilibria of human adult hemoglobin at
       atmospheric and elevated pressure. Blood 33: 57-65.
Roughton, F. J. W. (1970) The equilibrium of carbon monoxide with human hemoglobin in whole blood. In: Coburn, R. F., ed.
       Biological effects of carbon monoxide. Ann. N. Y. Acad. Sci. 174: 177-188.
Saltzman, B. E.; Fox, S. H. (1986) Biological significance of fluctuating concentrations of carbon monoxide.  Environ.  Sci.
       Technol. 20: 916-923.
Schoenfisch, W.  H.; Hoop, K. A.; Struelens, B. S. (1980) Carbon monoxide absorption through the oral and nasal mucosae of
       cynomolgus monkeys.  Arch. Environ. Health 35: 152-154.
Seki, T.; Naruse, M.; Naruse, K.; Yoshimoto, T.; Tanabe, A.; Imaki, T.; Hagiwara, H.; Hirose, S.; Demura, H. (1997) Interrelation
       between nitric oxide synthase and heme oxygenase in rat endothelial cells. Eur. J. Pharmacol. 331: 87-91.
Selvakumar, S.; Sharan,M.; Singh, M. P. (1992) Mathematical model for the exchange of gases in the lungs with special reference
       to carbon monoxide. Med. Biol. Eng. Comput. 30: 525-532.
Selvakumar, S.; Sharan, M.; Singh, M. P. (1993) A mathematical model  for the elimination of carbon monoxide in humans.
       J. Theor. Biol. 162: 321-336.
Sharan, M.; Selvakumar, S.; Singh, M. P. (1990) Mathematical model for the computation of alveolar partial pressure of carbon
       monoxide. Int. J. Biomed. Comput. 26:  135-147.
Shimazu, T.; Ikeuchi, H.; Hubbard, G. B.; Langlinais, P. C.; Mason, A. D., Jr.; Pruitt, B. A., Jr. (1990)  Smoke inhalation injury
       and the effect of carbon monoxide in the sheep model. J. Trauma 30: 170-175.
Siggaard-Andersen, J.; Bonde Petersen, F.; Hansen, T. I; Mellemgaard, K. (1968) Plasma volume and vascular permeability
       during hypoxia and carbon monoxide exposure. Scand. J. Clin. Lab. Invest.  Suppl. 103: 39-48.
Singh, J.; Moore-Cheatum, L. (1993) Gestational protein deficiency enhances fetotoxicity of carbon monoxide. In: Keen, C. L.;
       Bendich,  A.; Willhite, C. C., eds. Maternal nutrition and pregnancy outcome. Ann. N. Y. Acad. Sci. 678:  366-368.
Singh, M. P.; Sharan, M.;  Selvakumar, S.  (1991) A mathematical model for the computation of carboxyhaemoglobin in human
       blood as a function of exposure time. Philos. Trans.  R. Soc. Lond. B 334: 135-147.
Singh, J.; Smith, C. B.;  Moore-Cheatum, L.  (1992) Additivity of protein deficiency and carbon  monoxide  on  placental
       carboxyhemoglobin in  mice. Am. J.  Obstet. Gynecol. 167: 843-846.
Singh, J.; Aggison, L., Jr.; Moore-Cheatum, L. (1993) Teratogenicity and developmental toxicity  of carbon monoxide in
       protein-deficient mice.  Teratology 48: 149-159.
Singleton, G. J.; Olsen, C. R.; Smith, R. L. (1972) Correction for mechanical dead space in the calculation of physiological dead
       space. J. Clin. Invest. 51: 2768-2772.
                                                      5-28

-------
Smith, M. V. (1990) Comparing solutions to the linear and nonlinear CFK equations for predicting COHb formation. Math. Biosci.
       99:251-263.
Smith, M. V.; Hazucha, M. J.; Benignus, V. A.; Bromberg, P. A. (1994) Effect of regional circulation patterns on observed HbCO
       levels. J. Appl. Physiol. 77: 1659-1665.
Sokal, J. A.;Majka, J.; Palus, J. (1984) The content of carbon monoxide in the tissues of rats intoxicated with carbon monoxide
       in various conditions of acute exposure. Arch. Toxicol. 56: 106-108.
Sokal, J.; Majka, J.; Palus, J. (1986) Effect of work load on the content of carboxymyoglobin in the heart and skeletal muscles
       of rats exposed to carbon monoxide. J. Hyg. Epidemiol. Microbiol. Immunol. 30: 57-62.
Solanki, D. L.;  McCurdy, P.  R.; Cuttitta, F. F.;  Schechter, G.  P.  (1988) Hemolysis in sickle cell disease as measured by
       endogenous carbon monoxide production: a preliminary report. Am. J. Clin. Pathol. 89: 221-225.
Song, S.-Y.; Okeda, R.; Funata, N.; Higashino, F. (1983) An experimental study of the pathogenesis of the selective lesion of the
       globus pallidus in acute carbon monoxide poisoning in cats: with special  reference to the chronologic change in the
       cerebral local blood flow. Acta Neuropathol. 61: 232-238.
Stewart, R. D.; Peterson, J.  E.; Baretta, E. D.; Bachand, R. T.; Hosko, M. J.; Herrmann, A. A. (1970) Experimental human
       exposure to carbon monoxide.  Arch. Environ. Health 21: 154-164.
Stewart, R. D.; Peterson, J. E.; Fisher, T. N.; Hosko, M. J.; Baretta, E. D.; Dodd, H. C.; Herrmann, A. A. (1973) Experimental
       human exposure to high concentrations of carbon monoxide. Arch. Environ. Health 26: 1-7.
Stokes, D. L.; Maclntyre, N. R.; Nadel, J. A. (1981) Nonlinear increases in diffusing capacity during exercise by seated and supine
       subjects. J. Appl. Physiol.: Respir. Environ. Exercise Physiol. 51: 858-863.
Stone, J. R.; Marietta, M. A. (1994) Soluble guanylate cyclase from bovine lung: activation with nitric oxide and carbon monoxide
       and spectral characterization of the ferrous and ferric states. Biochemistry 33: 5636-5640.
Stork, R. L.; Bredle, D. L.; Chapler, C. K.; Cain, S. M. (1988) Regional hemodynamic responses to hypoxia in polycythemic dogs.
       J. Appl. Physiol. 65:  2069-2074.
Sugi, K.; Theissen, J. L.; Traber, L. D.; Herndon, D. N.; Traber, D. L. (1990) Impact of carbon monoxide on cardiopulmonary
       dysfunction after smoke inhalation injury. Circ. Res. 66: 69-75.
Tamayo,  L.; Lopez-Lopez, J. R.; Castaneda, J.;  Gonzalez, C.  (1997) Carbon monoxide  inhibits hypoxic  pulmonary
       vasoconstriction in rats by a cGMP-independent mechanism. Pfluegers Arch. 434: 698-704.
Thorn, S. R. (1990) Carbon monoxide-mediated brain lipid peroxidation in the rat.  J. Appl. Physiol. 68: 997-1003.
Thorn, S. R. (1992) Dehydrogenase conversion to oxidase and lipid peroxidation in brain after carbon monoxide poisoning.
       J. Appl. Physiol. 73:  1584-1589.
Thorn, S. R. (1993) Leukocytes in carbon monoxide-mediated brain oxidative injury. Toxicol. Appl. Pharmacol. 123: 234-247.
Thorn, S. R.; Ischiropoulos, H. (1997) Mechanism of oxidative stress from low levels of carbon monoxide. Cambridge, MA:
       Health Effects Institute; research report no. 80.
Thorn, S. R.; Ohnishi, S. T.; Ischiropoulos, H. (1994) Nitric oxide released by platelets inhibits neutrophil B2 integrin function
       following acute carbon monoxide poisoning. Toxicol. Appl. Pharmacol. 128: 105-110.
Thorn, S. R.; Xu, Y. A.; Ischiropoulos, H. (1997) Vascular endothelial cells generate peroxynitrite in response to carbon monoxide
       exposure. Chem. Res. Toxicol. 10: 1023-1031.
Thorn, S. R.; Garner, S.; Fisher, D.; Ischiropoulos, H. (1998) Vascular nitrosative stress from CO exposure. In: Program and
       abstracts [of the] Undersea andHyperbaric Medical Society annual scientific meeting: pre- and post-courses; May; Seattle,
       WA. Undersea Hyperbaric Med. 25(suppl.): 47.
Thorn, S. R.; Ohnishi, S. T.; Fisher, D.; Xu, Y. A.;  Ischiropoulos, H. (1999a) Pulmonary vascular stress from carbon monoxide.
       Toxicol. Appl. Pharmacol. 154: 12-19.
Thorn, S. R.; Fisher, D.; Xu, Y. A.; Garner, S.; Ischiropoulos, H. (1999b) Role of nitric oxide-derived oxidants in vascular injury
       from carbon monoxide in the rat. Am. J. Physiol. 276: H984-H992.
Thomsen, H. K. (1974) Carbon monoxide-induced atherosclerosis in primates: an  electron-microscopic study on the coronary
       arteries of Macaco irus monkeys. Atherosclerosis 20: 233-240.
Tikuisis, P.  (1996) Modeling the  uptake and elimination of carbon  monoxide.  In:  Penney, D.  G., ed. Carbon monoxide.
       Boca Raton, FL: CRC Press; pp.  45-67.
Tikuisis, P.; Buick, F.; Kane, D. M. (1987a) Percent carboxyhemoglobin in resting humans exposed repeatedly to 1,500 and
       7,500 ppmCO. J. Appl. Physiol. 63: 820-827.
Tikuisis, P.; Madill, H. D.; Gill, B.  I; Lewis, W. F.; Cox, K. M.; Kane, D. M. (1987b) A critical analysis of the use of the CFK
       equation in predicting COHb formation. Am. Ind. Hyg. Assoc. J. 48: 208-213.
Tikuisis, P.; Kane, D. M.; McLellan, T. M.; Buick, F.; Fairburn, S. M. (1992) Rate of formation of carboxyhemoglobin  in
       exercising humans exposed to carbon monoxide. J. Appl. Physiol. 72: 1311-1319.
Trischmann, U.; Klockner, U.; Isenberg, G.; Utz, J.; Ullrich, V. (1991) Carbon monoxide inhibits depolarization-induced Ca rise
       and increases cyclic GMP in visceral smooth muscle cells. Biochem. Pharmacol. 41: 237-241.
Turner, D. M.; Lee, P. N.; Roe, F. J. C.; Gough, K. J. (1979) Atherogenesis in the White Carneau pigeon: further studies of the
       role of carbon monoxide and dietary cholesterol. Atherosclerosis 34: 407-417.
                                                      5-29

-------
Tyuma, I.; Ueda, Y.; Imaizumi, K.; Kosaka, H. (1981) Prediction of the carbonmonoxyhemoglobin levels during and after carbon
       monoxide exposures in various animal species. Jpn. J. Physiol. 31: 131-143.
U.S. Environmental Protection Agency. (1978) Altitude as a factor in air pollution. Research Triangle Park, NC: Office of
       Research and Development, Environmental Criteria and Assessment Office; report no. EPA 600/9-78-015.
U.S. Environmental Protection Agency. (1991) Air quality criteria for carbon monoxide. Research Triangle Park, NC: Office of
       Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA/600/8-90/045F.
U.S. Environmental Protection Agency. (1992) Review of the national ambient air quality standards for carbon monoxide: 1992
       reassessment of scientific and technical information. OAQPS staff paper. Research Triangle Park, NC: Office of Air
       Quality Planning and Standards; report no. EPA-452/R-92-004.
Utz, J.; Ullrich, V. (1991) Carbon monoxide relaxes ileal smooth muscle through activation of guanylate cyclase. Biochem.
       Pharmacol. 41: 1195-1201.
Venkatram, A.; Louch, R. (1979) Evaluation of CO air quality criteria using a COHb model. Atmos. Environ. 13: 869-872.
Wagner, J. A.; Horvath, S. M.; Dahms, T. E. (1975) Carbon monoxide elimination. Respir. Physiol. 23: 41-47.
Wang, R. (1998) Resurgence of carbon monoxide:  an endogenous gaseous vasorelaxing factor. Can. J. Physiol. Pharmacol.
       76: 1-15.
Wang, R.; Wu, L.; Wang, Z. (1997a) The direct effect of carbon monoxide on KCa channels in vascular smooth muscle cells.
       Pfluegers Arch. 434: 285-291.
Wang, R.; Wang, Z.; Wu, L. (1997b) Carbon monoxide-induced vasorelaxation and the underlying mechanisms. Br. J. Pharmacol.
       121:927-934.
Webster, W. S.; Clarkson, T. B.;Lofland, H. B. (1968) Carbon monoxide-aggravated atherosclerosis in the squirrel monkey. Exp.
       Mol. Pathol. 13: 36-50.
Werner, B.; Lindahl, J. (1980) Endogenous carbon monoxide production after bicycle exercise in healthy subj ects and in patients
       with hereditary spherocytosis. Scand. J. Clin. Lab. Invest. 40: 319-324.
West, J. B. (1990a) Ventilation/blood flow and gas exchange.  5th ed. Oxford, United Kingdom: Blackwell Scientific Publications.
West, J. B. (1990b) Respiratory physiology—the essentials.  4th ed. Baltimore, MD: Williams & Wilkins.
Wittenberg, B. A.; Wittenberg, J. B. (1993) Effects of carbon monoxide on isolated heart muscle cells. Cambridge, MA: Health
       Effects Institute; research report no. 62.
Woehlck, H. J.; Dunning, M. B., Ill; Kulier, A. H.; Sasse,  F. J.; Nithipataikom, K.; Henry, D. W. (1997a) The  response of
       anesthetic agent monitors to trifluoromethane warns of the presence of carbon monoxide from anesthetic  breakdown.
       J. Clm. Momt. 13: 149-155.
Woehlck, H. J.; Dunning, M., Ill; Connolly, L. A. (1997b) Reduction in the incidence of carbon monoxide exposures in humans
       undergoing general anesthesia. Anesthesiology 87: 228-234.
Wyman, J.; Bishop, G.;  Richey, B.; Spokane, R.; Gill, S. (1982) Examination of Haldane's first law for the partition of CO and
       O2 to hemoglobin A0. Biopolymers 21: 1735-1747.
Yamaya, M.; Sekizawa, K.; Ishizuka, S.; Monma,  M.; Mizuta, K.; Sasaki, H. (1998) Increased carbon monoxide in exhaled air
       of subjects with upper respiratory tract infections. Am. J. Respir. Crit. Care Med. 158: 311-314.
Yoneda, I; Watanabe, Y.  (1997) Comparisons of altitude tolerance and hypoxia symptoms between nonsmokers and habitual
       smokers. Aviat.  Space Environ. Med. 68: 807-811.
Zayasu, K.; Sekizawa, K.; Okinaga, S.; Yamaya, M.; Ohrui, T.; Sasaki, H. (1997) Increased carbon monoxide in exhaled air of
       asthmatic patients. Am. J. Respir. Crit. Care Med.  156: 1140-1143.
Zhu, N.; Weiss, H. R. (1994) Effect of hypoxic and carbon monoxide-induced hypoxia on regional myocardial segment work and
       O2 consumption. Res. Exp. Med. 194: 97-107.
                                                      5-30

-------
                                     CHAPTER 6

                  Health Effects of Exposure to
                     Ambient Carbon  Monoxide
6.1  Introduction
       This chapter assesses the current understanding of health effects that may occur in individuals
breathing carbon monoxide (CO) in ambient air.  Concerns about the potential health effects of exposure
to CO have been addressed in extensive controlled-exposure studies and more limited population-exposure
studies. Under varied experimental protocols, considerable information has been obtained on the toxicity
of CO, its direct effects on blood and tissues, and the manifestations of these effects in the form of changes
in organ function. Experimentally derived knowledge of CO health effects has been augmented in recent
years by findings from a small, but growing, body of community epidemiology observational studies of
human populations exposed to CO in ambient air. Additional discussion of studies reporting CO-associated
health effects can be found in the previous  document, Air Quality Criteria for  Carbon Monoxide
(U.S. Environmental Protection Agency, 1991), and in a number of excellent reviews (Kleinman, 1992;
Bascom et al., 1996; Penney, 1996a).
       Although evidence from laboratory animal  studies indicates that CO can adversely affect the
cardiovascular and nervous systems of both mature animals and developing offspring, the concentrations
of CO often used during experimental exposure and consequent levels of carboxyhemoglobin (COHb)
saturation are much higher than typical of ambient human exposures.  The laboratory animal studies,
therefore, must be interpreted with caution. However, they can be useful for exploring the properties and
possible mechanisms of an effect much more thoroughly and extensively than is possible in humans. An
effort is made in this chapter to compare CO health  effect levels found in laboratory animal and human
controlled-exposure studies. Epidemiologic studies of ambient CO exposure effects on human populations
also are assessed in this chapter, and their reported are results considered in light of experimental study
findings.
       The next  section of this  chapter (Section 6.2) emphasizes CO effects  in humans, especially
cardiovascular effects of CO. The section begins with a discussion of epidemiologic studies (Section 6.2.1)
because of their potential importance in assessing community health effects of ambient CO exposure.
Section 6.2.1 emphasizes studies of ambient CO and heart disease exacerbation because short-term ambient
CO concentrations have been associated more  frequently with such exacerbation than with other health-
related endpoints. For purposes of continuity in the epidemiologic discussion, Section 6.2.1 also addresses
studies that have evaluated ambient CO in relation to health indices other than heart disease exacerbation,
including studies of daily mortality, incidence of lowbirth weight, and daily frequency of respiratory illness.
       The remainder of Section 6.2 (Section 6.2.2) summarizes controlled-exposure studies of CO effects
on maximal exercise performance and in subjects with reproducible exercise-induced angina.  In 1991,
these studies formed the major scientific basis for U. S. Environmental Protection Agency (EPA) review of
the levels and adequacy of the National Ambient Air Quality Standards (NAAQS) for CO. Although the
scientific data have changed little since 1991,  controlled-exposure studies continue to provide the most
quantitative evidence on low-level CO effects in humans.
                                            6-1

-------
       Next in importance to cardiovascular effects, but of questionable impact for the young and healthy
population, are studies of neurobehavioral effects, which had earlier provided the scientific basis for the first
CO NAAQS. Subsequent assessments of the neurobehavioral literature, however, have raised questions
about both the methods and results of the early studies. Articles published since the last assessment in 1991
have been mostly reviews that attempt to explain the equivocal results found at low COHb levels and to
provide a physiological basis for behavioral effects.  The section on neurobehavioral effects (Section 6.3)
illustrates the difficulty in studying an organ system with exquisite compensatory responses to a reduced
oxygen supply (hypoxia).
       The rest of the chapter summarizes current knowledge predominantly from experimental laboratory
studies on developmental toxicity (Section 6.4); acute pulmonary effects (Section 6.5); other systemic
effects of CO (Section 6.6); physiologic responses to CO exposure (Section 6.7); and combined exposure
of CO  with other pollutants, drugs, and environmental factors (Section 6.8).  Little new information has
been published on these areas of CO toxicity, and their summaries remain essentially the same as published
in the previous criteria document (U.S. Environmental Protection Agency, 1991). Significant new studies
have been  added to the summaries, but none of the newly published studies draw into  question the
conclusions drawn from the previous 1991 assessment of these topics. Finally, a summary section (Section
6.9) provides a concise review of the key human health effects most clearly  associated  with exposure to
ambient CO.
       It should be noted that a review of the health effects literature on CO since the last assessment was
published in 1991 finds many new published articles on CO poisoning, possibly reflecting increased media
attention to this topic. Many of these articles, however, reported effects at CO levels far higher than in
ambient air. Severe effects from acute exposure to high levels of CO are not directly germane to problems
associated with exposure to current ambient levels of CO  and, thus, are not discussed in detail in this
chapter. They are, however, mentioned briefly in the summary of this and the following chapter to present
a snapshot of the full range of CO toxicity and to provide public health information about potential effects
of accidental exposure to CO, particularly those exposures occurring indoors.
6.2  Cardiovascular Effects
       Cardiovascular disease (CVD) is the leading cause of death in the United States (American Heart
Association, 1997; U.S. Centers for Disease Control and Prevent on, 1997). An estimated 5 8 million people
in the  United States  (-20% of the  population) have one  or  more types of CVD (American Heart
Association, 1997). For the major diseases within the category of total CVD, about 50 million Americans
have high blood pressure, 14 million have ischemic (coronary) heart disease, 5 million have heart failure,
4 million have cerebrovascular  disease (stroke), and 2 million have rheumatic fever or heart disease.
Because  the numbers of affected people are so high, even relatively small  percentage  increases in
cardiovascular mortality or morbidity in the population could have a large impact on both public health and
health care costs.
       Carbon monoxide is notable among air pollutants because it is especially harmful in individuals
with impaired cardiovascular systems.  Persons with a normal cardiovascular system can tolerate substantial
concentrations of CO, if they vasodilate  in response to the  hypoxemia produced by CO.  In contrast,
individuals unable to  vasodilate in response to CO exposure may  show evidence of ischemia at  low
concentrations of COHb.  For this reason, experiments on healthy animals are unlikely to show effects at
low CO  concentrations  of exposure.   In the following  discussion, the effects of CO  on potentially
susceptible population groups are explored through epidemiologic and controlled laboratory studies  (for
further discussion of subpopulations at risk, see Chapter 7).
                                              6-2

-------
6.2.1  Epidemiologic Studies of Cardiovascular and Other Disorders
6.2.1.1   Introduction
       In recent years, many epidemiologic studies have shown associations of short-term ambient air
pollution exposure with mortality, exacerbation of preexisting  illness, and physiologic changes.  The
epidemiologic database regarding short-term ambient air pollution exposure is growing rapidly, and its
interpretation is changing over time.  As recently as the mid-1990s, many  epidemiologic studies had
reported associations of mortality and exacerbation of preexisting disease with ambient levels of parti culate
matter (PM), but relatively few had investigated or reported such  associations with gaseous pollutants,
including CO.  Since then, however, investigators have given more thorough consideration to PM and
gaseous pollutants and have frequently observed positive, statistically significant associations of harmful
effects with both. Thus, although associations of PM with harmful effects continue to be observed, the roles
of gaseous pollutants appear to be stronger than in previous evaluations.
       Another important recent change relates to the biological plausibility of observed epidemiologic
associations of ambient air pollution with harmful health outcomes. A few years ago, it was frequently
argued that associations of low-level ambient air pollution with harmful health outcomes were biologically
implausible. Now, with a considerable amount of new experimental evidence in hand, and after much new
and sharpened thought on this issue, it is possible to hypothesize ways by which ambient exposure to one
or multiple air pollutants could plausibly be involved in complex chains of biological events leading to
harmful health effects in the most sensitive population groups.
       In epidemiologic studies of ambient air pollution, small positive estimates of air pollutant health
effects have been observed quite frequently, often being statistically significant at a = 0.05.1 If ambient air
pollution actually promotes or produces harmful health effects, relatively small effect estimates would
generally be expected on biological and epidemiologic grounds. Also, the magnitudes and  significance
levels of observed air pollution-related effects estimates have varied somewhat from place to  place.  This,
too,  would be expected if the observed epidemiologic  associations  reflect actual effects, because
populations differ in characteristics that could affect susceptibility  to air pollution health effects.  These
characteristics include demographic and socioeconomic factors, underlying health status, indoor-outdoor
activities, diet, medical care systems and access to them, and exposure to risk factors other than ambient
air pollution, such as extreme weather conditions. Thus, although it has been argued that epidemiologic
studies are trustworthy only if they show relatively large effects estimates (e.g., large relative risks), if
observed small effects estimates for ambient air pollution are sufficiently consistent and coherent across
epidemiologic studies and comport well  with plausibility considerations based on other findings, then the
small effects estimates gain more credibility as likely reflecting underlying air pollution-health effects
quantitative relationships. It should also be borne in mind that, in any large population exposed to ambient
air pollution, even a small relative risk for a prevalent health disorder could calculate to a substantial public
health burden attributable to air pollution exposure.
       At  the same  time, important biological, epidemiologic, and statistical uncertainties remain in the
current  epidemiologic  database  for  ambient  CO   and other   air pollutants.    Biologically  and
epidemiologically, it has not been confirmed that the  magnitudes of observed statistical health effects
         lrThe Greek letter alpha (a) is the probability of type I error that the investigator is willing to accept in judging
 whether a finding is statistically significant. Type I error (alpha error) is the error of rejecting a null hypothesis that is
 actually true.  An alpha level of a =  0.05 is often chosen in judging statistical significance. The p value is the
 probability of obtaining a test statistic as large or larger than the one actually observed, if the null  hypothesis is true.
 (Examples of test statistics include the z-statistic, t-statistic, F-statistic, and chi-square statistic.) Evidence against the
 null hypothesis increases as the p value decreases. When an alpha level of a = 0.05 is chosen, the finding is judged to
 be statistically significant if p < 0.05.  In this chapter, the term "statistically significant" assumes an alpha level of
 a = 0.05 (p < 0.05), unless stated otherwise.

                                                6-3

-------
estimates for ambient CO are quantitatively commensurate with actual underlying population susceptibility
to CO exposure. Also, it has not been confirmed that the observed spatial variation in air pollution effects
estimates reflects actual cross-population differences in susceptibility.
       The ambient atmosphere contains numerous air pollutants, and there is increasing realization that
health  effects associated statistically with any single pollutant actually may be  mediated by multiple
components of the complex ambient mix.  Specific attribution of effects to any single pollutant therefore
may convey an overly  simplistic notion of biological reality.   Carbon monoxide is one  of many air
pollutants generated by combustion sources, including mobile sources. These pollutants include CO, PM,
and nitrogen oxides, which have been considered in epidemiologic studies to date.  These pollutants also
include numerous volatile or semivolatile organic compounds, which have not yet been considered
systematically in relation to the noncancer health outcomes usually associated with  exposure to criteria air
pollutants. In available epidemiologic studies, harmful health outcomes frequently are associated with
multiple combustion-related or mobile-source-related air pollutants.  Many investigators have raised the
possibility that CO may be a surrogate or marker for a larger subset of the overall ambient air pollution mix,
and some investigators have argued that ambient-level CO may be a biologically passive surrogate and that
statistical associations of ambient CO with health effects may actually reflect effects of pollutants other than
CO.  However, most investigators have reserved judgment on this issue, and several have emphasized the
need for further research on CO.
       The health effects of long-term exposure to CO and other air pollutants have received little attention
in epidemiologic studies and are  not well understood as yet.  Health effects of long-term exposure at
present-day ambient pollutant levels, which are generally lower than past levels, are especially uncertain.
Also, it is not known whether long-term exposure to ambient CO plays a role in the induction (incidence)
of new cases of illness. Further research on long-term health  effects of exposure to  ambient CO and other
pollutants is needed.
       Important statistical issues need to be considered in critically assessing the available epidemiologic
database for CO and other air pollutants. Many of these are especially pertinent to daily time series studies,
which are the majority of the epidemiologic studies available for ambient CO and which form a large part
of the epidemiologic database for other criteria air pollutants as well.  Statistical uncertainties, coupled with
existing  biological and epidemiologic uncertainties,  can pose difficulties in judging the quantitative
accuracy of pollutant effect estimates themselves and, perhaps in some cases, their qualitative validity.
       Key statistical issues that need to be considered include the following.
(1)  Sensitivity of effects estimates to different choices of statistical models/model specifications. Effects
    estimates for CO and other air pollutants can vary, depending on different choices of statistical models
    or specifications for important model parameters. For example, as the following sections show, effects
    estimates for CO or other pollutants at times differ with different choices of metric for the same
    pollutant, with different choices of modeled covariates (independent variables), with different lag or
    moving-average structures for air pollutants and covariates, with different choices of time spans in
    nonparametric smoothing procedures, and with different choices of adjustment for autocorrelation and
    overdispersion in the data.  Also, in parametric models, different choices of functional forms of
    modeled  concentration-response  relationships can lead to  different  interpretations of results.
    Furthermore, effects estimates for CO and other pollutants in single-pollutant models frequently differ
    from those derived from in multi-pollutant models. It is  increasingly clear that ambient air pollution
    health effects may arise from exposure to multiple pollutants, and that single-pollutant models do not
    necessarily describe adequately the effects of ambient CO and other  pollutants.  Recent increased
    attention to multi-pollutant models is, therefore, highly appropriate.  Even so,  time series of ambient
    CO levels often are highly correlated with time series levels of other pollutants, such as PM. Thus,
    effects estimates for different pollutants remain subject to confounding in multi-pollutant  models.
    Scientific consensus as to optimal modeling strategies for time series air pollution studies has not yet

                                               6-4

-------
    been achieved. Application of nonparametric techniques, which generate exposure-response surfaces
    for more than one pollutant at a time, may well prove useful in future analyses. Nevertheless, however,
    much progress has been made in evaluating issues related to uncertainties associated with model
    selection and alternative specifications for important model parameters, as illustrated by U.S. EPA
    (1996) evaluations of time-series analyses of ambient PM exposure effects.
       As mentioned above, small health effects estimates generally have been observed for ambient air
    pollutants, and small effects would be expected on biological and epidemiologic grounds. At the same
    time, because the effects estimates are small, they can be sensitive to different model specifications.
    These can calculate to substantial differences in estimated numbers of cases of illness or mortality
    attributable to ambient air pollution exposure.  On balance, there remains uncertainty as to the proper
    choice of effects estimates to employ in estimating risks of exposure to ambient CO and other air
    pollutants in the  human population.
(2)  Measurement error in ambient CO metrics. In this document, Appendix 3-A and Sections 4.2.2 and
    4.2.3 indicate that there is substantial spatial variability in ambient CO concentrations, and that fixed-
    site CO measurements may not adequately index widely  varying actual  population exposures to
    ambient CO.  Current evidence also suggests that ambient CO is more spatially heterogeneous than
    other criteria pollutants  assessed in epidemiologic studies to  date (e.g., PM25, PM10, ozone [O3]).
    In many instances, misclassification of exposure leads to down ward bias in statistical effects estimates.
    Thus, effects estimates for ambient CO may be biased downward in available epidemiologic studies,
    and downward bias may be stronger for estimates of CO effects than for estimates of other pollutants'
    effects.  However, this has not been confirmed. Further research is needed to quantify the degree to
    which fixed-site  measurements of ambient CO and other pollutants overestimate or underestimate
    actual population exposures  to these pollutants.  This research will require characterization of
    relationships between fixed-site CO measurements and personal CO exposures over time. It will also
    require accurate  apportionment of total CO exposure into CO exposures of ambient and nonambient
    origin. Further research  also is needed to characterize influences of measurement error on estimates
    of air pollution effects in statistical models for CO and other ambient pollutants.
(3)  Potential confounding of air pollution and weather effects. Meteorologic events  and ambient air
    pollutant concentrations  may be highly correlated on short time scales, even when longer time trends
    have been filtered.  It is essential to model joint effects of air pollution and weather with great care.
    Such joint modeling has been conducted only rarely in time series studies of ambient air pollution.  To
    date, simple additive or proportional assumptions generally have been made in modeling health effects
    of air pollutants and weather.  These assumptions are not necessarily fully adequate, largely because
    health effects estimates for air pollutants are small and subject to large proportional differences with
    different model  specifications.  One example of recent efforts to model better possible complex
    relationships involving combinations of meteorological factors (e.g., temperature, barometric pressure,
    etc.) and to assess their impacts on air pollutant health effects estimates was the modeling of "synoptic
    weather pattern"  effects in conjunction with ambient PM exposures (as described in U.S. EPA 1996).
    Application of nonparametric techniques to jointly model CO and meteorologic effects also may prove
    useful in future CO analyses.
(4)  Insufficient reporting of statistical uncertainty. In available studies, statistical uncertainty generally
    has been assessed rather superficially, without formal consideration of the model tuning performed by
    the investigators. For example, lag times and averaging times for air pollutant metrics are sometimes
    selected to maximize statistical effects estimates for pollutants. This may, at times, lead to inflated
    reported effects  estimates and,  perhaps,  unduly narrow confidence intervals for these estimates.
    In future studies, uncertainty arising from model tuning should be assessed more carefully. In this
    effort, resampling or simulation procedures, which would recreate the entire model estimation process,
                                              6-5

-------
    should be considered as a means  for evaluating the accuracy and robustness of reported effects
    estimates.
(5) Health effects averaged over extended time periods. None of the available time  series studies of
    ambient CO are capable of assessing the incremental effect of pollutants over extended time periods.
    For example, current models cannot confidently predict whether reduction in pollution will decrease
    monthly rates of hospital admissions or mortality, even if they imply a reduction of admissions on days
    with low pollution. This public-health-related issue cannot be addressed by daily time series analysis,
    using only admission or mortality counts. In future studies, investigators also could consider time-
    averaged health effects over,  say, 1 or 3  mo, in  relation to pollution exposure metrics  for the
    corresponding periods. Consideration of extended time-averaged health  effects would tend to allow
    for detection of more chronic impacts beyond any short-term "harvesting" that might be observed in
    daily analyses. (In time series studies of air pollution, harvesting is a  short-term elevation in the
    frequency of a health outcome during or just after a short period of elevated ambient air pollution,
    followed quickly by a short-term reduction in frequency of the same outcome below baseline
    frequency, then by a return to baseline. It has been argued that presence of harvesting would suggest
    that elevated air pollution exposure hastens occurrence of the health outcome by only a short time, but
    brings about little or no net increase in occurrence of the outcome.  It also has been argued that absence
    of harvesting would suggest that, without the elevated air pollution exposure, the outcome might have
    been delayed for a long time or might not have occurred at all.)
       To date, short-term fluctuations in ambient CO have been examined in time series studies of daily
exacerbations of preexisting cardiovascular disease, mortality, and changes in respiratory illness frequency.
Associations have been observed most frequently for short-term ambient CO fluctuations and exacerbation
of heart disease (a subcategory of cardiovascular disease), as usually indexed by daily hospital admissions.
These associations generally have been stronger during  cold weather than during warm weather.  These
associations are biologically plausible to some degree.  In heart disease  patients, the coronary arteries
usually are narrowed and unable to dilate normally. Increased hypoxic stress resulting from small increases
in COHb saturation could conceivably lead to clinically apparent worsening  of their illness.   Cold
temperatures also exacerbate hypoxic effects (e.  g., as occurs at high altitude).
       If observed associations of ambient CO with heart disease exacerbation prove to be real and specific
to CO, they would be of genuine public health concern.  In the United States in  1996, there were about
4,239,000 hospital discharges with heart disease as the first-listed diagnosis (Graves and Owings, 1998).
Among these, about 2,262,000 (53.4%) were for ischemic heart disease (IHD), 825,000 (19.5%) for
myocardial infarction (MI) or heart attack (a subcategory of IHD), 870,000 (20.5%) for congestive heart
failure (CHF), and 618,000 (14.6%) for cardiac dysrhythmias. Also, there were 733,361 deaths caused by
heart disease (Peters et al, 1998). Even a small percentage reduction in admissions or deaths caused by
heart disease would result in a large number of avoided cases.
       Even so, fluctuations in ambient CO levels would be expected to produce only very small changes
in COHb saturation and only small changes in tissue oxygenation.  Thus, the observed associations of
ambient CO with heart disease exacerbation remain difficult to rationalize pathophysiologically. Also, such
exacerbation has been associated not only with ambient CO, but also with other combustion-related ambient
pollutants such as nitrogen dioxide (NO2) and PM.  Thus, the extent to which such exacerbation truly is
attributable to ambient CO exposure is not yet clear.
       Studies of short-term ambient CO levels and daily mortality have yielded mixed results. Observed
associations of ambient CO with mortality are of potential public health concern because associations of
ambient CO with heart disease exacerbation have been observed frequently, and because heart disease is
the leading cause of death in the United States. For example, in the United States in 1996, 733,361 (31.7%)
of all 2,314,690 deaths were caused by heart disease (Peters  et  al.,  1998).  Again, however, no firm
pathophysiologic basis for reported ambient CO-mortality associations is apparent, and the degree of

                                              6-6

-------
correspondence between the reported  statistical effects estimates in populations and real effects in
individuals remains to be more firmly and plausibly delineated.
       Some investigators have observed associations of short-term fluctuations in ambient CO with daily
frequency of respiratory illness. In most cases, exacerbation of preexisting respiratory illness has been
assessed, although some cases of acute respiratory infection constitute occurrence of new illness, especially
in young people. The biological plausibility of these associations is tenuous, because there are as yet no
demonstrated mechanisms by which ambient-level CO  exposure could produce or promote harmful
respiratory effects. Also, in  epidemiologic studies to date, associations of ambient CO with respiratory
disease frequency have been observed less consistently than with heart disease exacerbation. Nevertheless,
the observed associations of ambient CO with respiratory disease frequency are worthy of discussion
because they suggest that ambient CO exposure may not be specifically linked just with heart disease.
       Two available U. S. studies suggest that subchronic ambient CO exposure averaged over about 3 mo.
may be associated with increased incidence of low birth weight.  These studies raise concern, because fetal
hemoglobin (Hb) binds CO somewhat more strongly than does adult Hb, and because increased COHb
saturation would be expected to impair tissue molecular oxygen (O2) delivery more in the fetus than in the
child or the adult (Longo, 1976).  At the same time, these studies are not conclusive, and they are subject
to potential confounding by unmeasured factors, such as maternal smoking, that are known to  influence
birth weight.  Also, outdoor CO levels may be correlated with indoor levels of CO and other pollutants,
which could be higher than  outdoor levels, and which were not measured in these  studies.  Common
socioeconomic factors could be associated with both ambient CO levels and such potential confounding
variables.
       Overall, then, recent epidemiologic studies have tended to increase concern regarding potential
harmful effects of present-day ambient CO exposure, especially with respect to heart disease exacerbation,
possibly with respect to mortality and low birth weight, and even, conceivably, with respect to increased
frequency of respiratory illness.  However, more research on the health  effects of long- and short-term
ambient CO exposure is strongly  warranted to more clearly elucidate quantitative health risks ascribable
to ambient CO specifically  among those pollutants typically found in  urban air mixes impacted by
combustion processes. This research should assess effects not only of changes in COHb levels and tissue
oxygenation, but also of CO dissolved in the blood and of CO in tissues  other than blood, and it should
assess effects of CO alone and as  a component of the complex ambient air pollution mix.
       Individual epidemiologic  studies that have considered ambient CO are summarized and assessed
below. Time series studies of short-term ambient CO are discussed first, in descending order of consistency
of findings.  Health-related outcomes are discussed in the following order:  daily exacerbations of heart
disease, daily mortality counts, and daily frequency of respiratory illness.   Studies of subchronic ambient
CO and reduction of birth weight are discussed after the time series studies.

6.2.1.2   Ambient Carbon Monoxide and Exacerbation of Heart Disease
       Recent epidemiologic studies in the United States, Canada, and Europe suggest that short-term
variations in ambient CO levels are associated with daily hospital admissions for heart disease. In several
studies of such admissions, effects of lagged ambient air pollutant levels have been examined, in addition
to effects of air pollutant levels on the same day as the admissions (0-day lag). When averaging times for
ambient pollution metrics have been 24 h or shorter (e.g., 24-h average CO or daily maximum hourly CO),
modeled effects of CO generally have been strongest with a 0-day lag. When averaging times have been
longer than 24 h, CO effects generally have been strongest when the last day of the averaging period is
lagged 0 days.
       As mentioned above, observed associations of ambient  CO with heart disease exacerbation have
some biological plausibility and are of potential public health concern. However, these associations should
be interpreted cautiously. The average  daily 1-h maximum CO concentrations measured by stationary


                                              6-7

-------
monitors in the epidemiologic studies generally have been low (<5 ppm). Any increase over endogenous
COHb levels produced by a 1-h exposure to <10 ppm exogenous CO, for example, would be difficult even
to measure accurately or, possibly, even to detect reliably. Even 8 h of exposure to 10 ppm CO with light
to moderate exercise (ventilation about 20 L/min) would be expected to produce only a 1.0 to 2.0% increase
in COHb saturation over the baseline level of about 0.5% COHb. Pathophysiologically, it remains difficult
to reconcile such small expected ambient CO-induced changes in COHb saturation with reported increased
overt exacerbation of heart disease in the community setting.
       Also, epidemiologic  studies of ambient CO have relied heavily on pollutant measurements at
stationary outdoor monitors.  Carbon monoxide levels at these monitors may not be correlated well with
personal CO exposures and doses, especially in compromised persons such as cardiac patients, who spend
much of their time indoors.   Additional  information on personal CO exposures and other individual
characteristics, such  as active and passive smoking, would be highly desirable in future epidemiologic
studies of ambient CO.  Furthermore, as discussed in Chapter 3, in most U.S. metropolitan areas, there is
considerable spatial  variation in simultaneous CO measurements made at different monitoring sites.
In most epidemiologic studies to date, exposure metrics have consisted of CO measurements  averaged
across sites.  The effects of such multi-site averaging of CO levels on statistical health effects estimates are
not yet well understood.
       The  diagnostic category  "heart  disease" is smaller and  more specific  than the  category
"cardiovascular disease," which comprises heart disease and other disorders such as cerebrovascular disease
(including stroke), hypertension (high blood pressure), and diseases of the blood vessels.  To date, short-
term variation in ambient CO  levels has been more strongly associated with heart disease exacerbation than
exacerbation of other cardiovascular diseases. At the same time, heart disease itself comprises several
diagnostic subcategories,  such as HtD (including MI, coronary artery atherosclerosis, and angina), heart
failure (including CHF), and disturbances of cardiac rhythm (dysrhythmias, arrhythmias). In this document,
the terms "ischemic heart  disease" (or 'THD"), "coronary artery disease" (or "CAD"), and "coronary heart
disease" may be  considered interchangeable.   The available epidemiologic database is not entirely
consistent regarding the  specific heart disease subcategories with which ambient CO  levels  are most
strongly associated.
       Morris et al. (1995) conducted a time-series analysis of ambient levels of gaseous air pollutants
(CO, NO2, sulfur dioxide [SO2], and O3), in relation to Medicare hospital admissions for CHF in seven U. S.
cities (Chicago, IL; Detroit, MI; Houston, TX;  Los Angeles, CA; Milwaukee, WI; New York City, NY; and
Philadelphia, PA) during the  4-year period, 1986 to 1989.  (Medicare covers only persons >65 years old.)
The average daily maximum  1-h CO levels (mean ± standard deviation [SD]) ranged from 1.8 (±1.0) ppm
in Milwaukee to 5.6 (±1.7) ppm in New  York City.  The relative risk of admissions associated with a
10-ppm increase in ambient CO ranged from 1.10 in New York City to 1.37 in Los Angeles.  All seven
cities showed similar patterns of increasing admissions with increasing ambient CO concentrations.
In almost all analyses, CO effects were stronger on the day of admission (0-day lag) than on previous days.
In single-pollutant models, the effect of CO was statistically significant in all cities but Houston.  In multi-
pollutant models, the CO effect was significant in all cities but New York and Milwaukee.  In the transition
from single-pollutant to multi-pollutant models, effects of CO were  more stable and retained statistical
significance more frequently than effects of the other pollutants. Figure 2 from Morris et al. (1995), which
shows nonparametrically smoothed exposure-response curves for ambient CO levels and CHF admission
rates in each city, is reproduced in Figure 6-1.  The  authors estimated that, each year, approximately
3,250 hospital admissions for CHF could be  attributed to the observed association with short-term
elevations in ambient CO levels. (Note that the Y-axis in Figure 6-1 does not begin at zero. The exposure-
response curves in this figure would appear somewhat less steep if the Y-axis began at zero.)
       Schwartz and Morris (1995) examined air pollution and hospital admissions for heart disease (CHF,
IHD, and cardiac dysrhythmias) in people  aged 65 years and older in the Detroit metropolitan area from

                                              6-8

-------
                                                     1986 through  1989.   Air quality data were
                                                     available for CO,  PM10, O3, and SO2.  For
                                                     gaseous pollutants and PM10, exposure metrics
                                                     were the daily maximum 1 -h concentration and
                                                     the 24-h  average concentration, respectively.
                                                     The  average  of daily maximum  1-h CO
                                                     concentrations over the whole study period was
                                                     2.4 ppm.  The overall  mean of 24-h average
                                                     PM10 concentrations was 48.0 //g/m3. Data were
                                                     analyzed with Poisson auto-regressive models,
                                                     with independent variables for temperature, dew
                                                     point,  month,  and linear and  quadratic  time
                                                     trends.  For each pollutant, the "interquartile
                                                     range"  (IQR), i.e., the difference between the
                                                     25th and 75th percentiles of the distribution of
                                                     ambient concentrations during the study period,
                                                     was calculated.  Relative risks for the health
                                                     outcomes assessed were reported per increment
                                                     in  ambient  concentration  equal   to   each
                                                     pollutant's IQR range.
       Daily admissions for IHD were associated with IQR increases of 1.28 ppm for CO (relative risk
[RR] = 1.010; 95% confidence interval [CI] = 1.001, 1.018), 32  //g/m3 for PM10 (RR= 1.018; 95% CI =
1.005,1.032), 18 ppb for SO2 (RR= 1.014; 95% CI= 1.003,1.026). However, both the CO and SO2 effects
lost statistical significance after controlling for PM10, whereas the PM10 effect remained significant after
controlling for the other pollutants.  Daily admissions for congestive heart failure were associated
independently and significantly with IQR increases in both CO (RR = 1.022; 95% CI = 1.010, 1.034) and
PM10 (RR = 1.024; 95% CI = 1.004, 1.044).  These results are summarized in Table 6-1.  Effects of CO
were stronger on the day of admission (0-day lag) than  on previous days. No pollutant was found to be a
significant risk factor for dysrhythmia admissions.
                  Carbon Monoxide (ppm)

Figure 6-1.  Nonparametric smoothing of the association
between ambient levels of CO and hospital admissions for
CHF among elderly people after adjustment for temperature,
month, day of week,  and year, 1986 through 1989.
Source: Morris et al. (1995).	
    Table 6-1.  Modeled Relative Risks of Interquartile Range Increases in Ambient Pollutant
         Concentrations for Daily Heart Disease Admissions in Persons > 65 Years Old,
   	Detroit, Ml, 1986 to 1989	
                                  Single-Pollutant Models
                                                                    Two-Pollutant Models
                                                                 (including PM10 and CO or SO2)a
Disease Category
Congestive Heart Failure
(Table 6)
Ischemic Heart Disease
(Table 4)
Dysrhythmias
CO
1.022b
1.010b
SO2
1.002
1.014b
03 PM10
1.022 1.032b
1.010 1.018b
CO
1.022b
1.006
PM10
1.024b
1.016b
1.015b
SO2

1.009
No Significant Pollutant Effects
 Tor CHF, no model with PM10 and SO2 was reported.
 bStatistically significant at a = 0.05.
 Source: Modified from Schwartz and Morris (1995).
                                              6-9

-------
       Ambient CO could be a surrogate for general combustion-related or mobile-source air pollution.
In some locations, CO is highly correlated with PM during the winter months. The ambient PM10 level was
associated with heart disease exacerbation in Schwartz and Morris (1995), but was not assessed in Morris
et al. (1995).  Also, ambient PM concentrations were associated with hospital admissions for both heart
failure and IHD in Ontario, Canada (Burnett et al., 1995).
       Pantazopoulou et al.  (1995) investigated cardiac and respiratory disease exacerbation in the area
of Athens, Greece, in relation to short-term ambient air pollution levels during 1988. The health outcomes
analyzed were daily outpatient emergency department visits, and daily hospital admissions for cardiac and
respiratory causes. Pollutant metrics were daily maximum 8-h moving average CO, daily maximum hourly
NO2, and 24-h average black smoke (BS), each averaged over multiple monitoring stations in the Athens
area.   Separate analyses were conducted for "winter"  (January 1 to March 21 and September 22 to
December 31) and "summer" (March 22 to September 21). Mean levels of CO for all available monitoring
stations were 4.5 mg/m3 (3.9 ppm) in winter and 3.4 mg/m3 (3.0 ppm) in summer. These levels were quite
high in comparison to current U.S. levels.   Data were analyzed with multiple linear regression, with
adjustment for meteorological and chronological variables.  Only single-pollutant models were reported.
The report did not mention lagged analysis, so effects were presumably reported for pollutant levels on the
same day as the visit or admission.
       Pollutant effects were reported as the modeled increment in the number of visits or admissions from
the 5th to 95th percentile of the pollutant's concentration distribution during 1988.  Winter and summer
findings from Pantazopoulou et al. (1995) are summarized in Table 6-2.  No pollutant had a statistically
significant effect on any health outcome in summer.  In contrast, all three pollutants had positive,
statistically significant effects  on both cardiac and respiratory "emergency admissions"  (unscheduled
hospital admissions) in winter. Pollution effects were stronger for admissions than for outpatient visits, and
accounted for a higher proportion of respiratory admissions than cardiac admissions.
       Separate analyses were conducted for CO effects when ambient CO levels were averaged across
two, three, or five monitoring stations.  Mean CO levels were 8.2, 6.5, and 5.2 mg/m3 (7.2, 5.7, and
4.5  ppm), respectively.  Interestingly, estimated 5th- to 95th-percentile increments in winter cardiac
admissions were nearly identical in these different analyses (11.0,11.4, and 11.2 admissions, respectively).
Corresponding CO-related increments in winter respiratory admissions were also similar (9.9, 12.1, and
11.3 admissions, respectively).  Effects of BS varied somewhat more than those of CO when averaged
across different numbers of monitoring stations. These observations suggest that there may be less spatial
variation in ambient CO in Athens than is typical for many U.S. locations, or that spatial heterogeneity in
ambient CO levels may not always greatly distort CO-related health effects estimates, even when CO levels
are averaged  across multiple monitoring sites. The observations also underscore that further research is
required on the effects of different choices of ambient pollutant metrics in time series studies.
       Schwartz (1997) examined  relationships  of short-term  ambient  air  pollution  levels with
cardiovascular hospital admissions in people at least 65 years old in Tucson, AZ, from 1988 through 1990.
The analyzed range of diagnoses (International Classification of Diseases, Version 9 [ICD-9] codes 390 to
429) included heart disease, hypertension, rheumatic fever, and pulmonary circulatory disorders (U.S.
Department of Health and Human Services, 1998). It did not include cerebrovascular disease or peripheral
blood vessel  diseases.  Thus, heart disease constituted the large majority, but not all, of the analyzed
disorders.  The author assessed effects of CO, SO2, O3, NO2, and PM10, as measured at a single monitoring
station. Poisson regression models included air pollution metrics, temperature, humidity, and day of week.
Nonparametric smoothing was used to adjust for long-term temporal patterns. Exposure-to-admission lags
of 0, 1, and 2 days were apparently  assessed in different models, and effects estimates of pollutant levels
on the same day as admission (0-day lag) were apparently reported.  During the study, the medians of
maximum hourly CO concentrations and of 24-h average PM10 concentrations were 3.03 ppm and 3 9 yUg/m3,
respectively.  The correlation between PM10  and SO2 was lower than in eastern U.S. cities.


                                              6-10

-------
  Table 6-2. Modeled Effects of 5th- to 95th-Percentile Increments in Ambient Air Pollutant Concentrations on Daily Numbers of
           Cardiac and Respiratory Hospital Admissions in Single-Pollutant Models, by Season, Athens, Greece, 1998
Winter (January 1 -March 21 and September
Disease
Category
Cardiac


Respiratory


Pollutant
CO
BS
NO2
CO
BS
NO2
5th- to 95th-
Percentile Increment
(Mg/m3)a
8,200 (7.2 ppm)
235
76
8,200 (7.2 ppm)
235
76
Increase in
Number of
Admissions15
11.0
11.8
11.2
9.9
9.2
10.4
22 -December 31)
95% CI
3.4, 18.5
4.7, 18.9
3.3, 19.2
2.5, 17.4
2.0, 16.3
2.7, 18.2
Summer (March 22-September 21)
5th- to 95th-
Percentile
Increment (,wg/m3)a
5,600 (4.9 ppm)
134
108
5,600 (4.9 ppm)
134
108
Increase in
Number of
Admissions15
1.4
3.0
-0.06
2.2
3.2
3.9
95% CI
-5.5,8.4
-2.9,8.9
-6.6,6.5
-5.3, 11.5
-4.0, 10.3
-5.9, 10.2
a5th- to 95th-percentile increments in ,wg/m3 as measured at two sites in central Athens.
bThe average daily numbers of total cardiac admissions were 73.8 in winter and 63.9 in summer. The average daily numbers of total respiratory admissions
were 41.8 in winter and 41.4 in summer.

Source: Modified from Pantazopoulou et al. (1995).

-------
       In  Schwartz  (1997),  relative  risk  estimates  of CO  and PM10  for admissions  were
of similar magnitude, independent, additive, and statistically significant at a = 0.05. In a model assessing
both pollutants simultaneously, the estimated percentage increases in admissions across the IQRs of CO
andPM10 levels were 2.33 and 2.37%, respectively.  The IQRs of ambient CO andPM10 were 1.66 ppm and
23 yUg/m3, respectively. Effects estimates for both pollutants appeared to be quite stable across seasons and
did not appear to be confounded with the meteorologic parameters assessed.  There were no appreciable
associations of admissions with ambient levels of SO2, O3, or NO2.
       Burnett et al. (1997a) examined temporal relationships between short-term ambient air pollution
levels and hospitalizations for CHF in the elderly (persons >65 years of age) in the 10  largest Canadian
cities, during the 11 -year period 1981 to 1991.  The average daily number of CHF admissions was 3 9 in the
134 catchment hospitals. Time series random-effects models, adjusted for long-term time trends, seasonal
and subseasonal temporal variation, and day-of-week effects, were used to explore the relationship between
hospitalizations and the ambient air pollutants CO, NO2, SO2, O3, and coefficient of haze  (COH, an optical
index of ambient PM concentration).  Fixed-effects models also were employed. After stratifying by month
of the year and adj usting for temperature, dew point, and other pollutants, the log of the daily 1 -h maximum
CO concentration on the day of admission (0-day lag) was associated most strongly with hospitalization for
CHF. Over the study period, the relative risk across all cities was 1.065  (95% CI = 1.028, 1.104) for an
increase from 1 to 3 ppm CO (the overall IQR of the ambient CO concentration distribution). Associations
of other pollutants with admissions also were observed in single-pollutant models. However, risk estimates
for these other pollutants were  more sensitive to  simultaneous adjustment for multiple pollutants  and
weather variables than were the estimates for CO.  The authors noted that CO could be acting as a marker
for pollution from transportation sources in general, and that independent  effects of non-CO pollutants
could not be ruled out.
       A summary of the Burnett et al. (1997a) study results is presented for each city in Table  6-3, which
is slightly modified from Table 5  of the published report.  The authors employed both random-effects
models and fixed-effects models to estimate RRs of ambient CO (from 1 to 3 ppm) for CHF admissions in
each city. These RRs varied substantially across cities, were generally somewhat larger in single-pollutant
models than in two-pollutant models, and were slightly larger in fixed-effects models than in random-effects
models. The across-cities average RR ranged from 1.058 (two-pollutant random-effects models) to 1.086
(single-pollutant fixed-effects models). Modeled city-specific relative risk was not correlated statistically
significantly with city-specific study mean CO concentration or with city population (as  calculated by the
authors of the present document).
       Burnett et al. (1997b) investigated summertime ambient air pollution in relation to unscheduled
hospital admissions for cardiac and respiratory diseases in Toronto, Ontario, Canada, in the summers of
1992, 1993, and 1994 (total 388 days).  Hourly measurements of CO, O3, SO2, NO2, and COH were
available from multiple monitoring stations.  Daily measurements of fine and coarse PM (8:00 a.m. to
8:00 a.m.) were available from a dichotomous sampler at a downtown site chosen to be representative of
the Toronto area. Measurements of COH also were available.  Levels of PM10 were calculated as the sum
of daily fine and coarse PM mass. Ambient CO levels were low; the mean and 95th percentile of the daily
1-h maximum CO concentration were 1.8  and 3.2 ppm, respectively.
       In data analysis,  pollutant concentrations were lagged 0 to 4 days  before admission in separate
models. Additional pollutant metrics were computed as multi-day average  ambient concentrations, with
the  last day of the averaging period lagged 0 to 2 days before admission.  The number of days in the
averaging period, and the last day of the period, varied from pollutant to pollutant.  In single-pollutant
models, there were positive, usually significant, effects of all pollutants on both cardiac and respiratory
admissions.  Many  two-pollutant models, each with one particulate metric and one gaseous pollutant
metric,were constructed.  In these,  there was little evidence of a CO effect on cardiac or respiratory
admissions. Effects of CO remained slightly positive, but were not statistically significant. Effects of PM

                                             6-12

-------
     Table 6-3. Canadian City-Specific Relative Risks of a Change in Daily Maximum 1-Hour
   Carbon Monoxide Levels from 1 ppm to 3 ppm for Congestive Heart Failure in the Elderly,
      Based on Random Effects Models (Random) and a Fixed Effect Analysis of Each City
                     Separately (Fixed), for Selected Model Specifications
                                                           Model Specification
                                                    CO
CO, NO2, Temperature,
     Dew Point
City
Montreal
Ottawa
Toronto
Hamilton
London
Windsor
Winnipeg
Edmonton
Calgary
Vancouver
Average
Population3
(x 105)
24.6
8.0
31.4
4.2
3.3
3.2
5.9
7.2
8.1
14.0
10.99
cob
(ppm)
2.0
2.5
2.7
2.0
1.7
2.0
1.6
2.7
3.3
2.5
2.30
Fixed
1.05
1.23
1.19
1.04
1.00
1.01
1.05
1.06
1.01
1.22
1.086
Random
1.05
1.16
1.16
1.05
1.02
1.02
1.05
1.06
1.04
1.17
1.078
Fixed Random
1.04
1.18
1.17
1.05
0.99
1.01
0.99
1.02
0.99
1.22
.04
.12
.12
.05
.01
.02
.01
.04
.02
.15
1.066 1.058
 "Based on 1986 census.
 b Average of daily maximum 1-h ambient concentrations of carbon monoxide.

 Source: Modified from Burnett et al. (1997a).
metrics adjusted for CO were similar to PM effects in single-pollutant models. The gaseous pollutant least
sensitive to adjustment for PM metrics was O3, and the gaseous pollutant that most attenuated PM effects
was NO2.  The authors hypothesized that the absence of a CO effect may have reflected the fact that
summertime ambient CO levels were low and emphasized the potential importance of the overall ambient
air pollution mix. They recommended that "all available air pollution measures be considered in assessing
the effects of any single pollutant on health."
       Poloniecki et al. (1997) investigated cardiovascular hospital admissions in relation to ambient air
pollution concentrations 1 day before admission in London, UK, from April 1, 1987, to March 31, 1994.
Pollutant metrics were 24-h mean CO, BS, SO2, and NO2 and hourly mean O3  between 9:00 a.m. and
5:00 p.m.   Measurements of all gaseous pollutants were taken from a single site in central London.
Measurements of BS were taken from one central site and four suburban sites. Median and 90th-percentile
CO concentrations were 0.9 and 1.8 ppm,  respectively.  Corresponding BS concentrations were 12 and
22 //g/m3.  Health outcomes considered were admissions for all  cardiovascular diseases  and  for the
following  seven diagnostic subgroups:  (1) MI, (2) other IHD, (3)  heart failure, (4) angina, (5)  cardiac
dysrhythmia, (6) cerebrovascular disease, and (7) other circulatory diseases.
       Analytical models were adjusted for day of week, holidays, an influenza epidemic in 1989, and
several temporal cycles  ranging from 20 days to the  whole study period.  Pollutant concentration,
temperature, and humidity one day before admission (1 -day lag) also were entered into the models.  Single-
pollutant Poisson models were constructed for each health outcome. In these models, CO was positively
and statistically significantly associated with admissions for all cardiovascular diseases, MI,  and other
                                             6-13

-------
circulatory diseases.  Black smoke was significantly associated with admissions for all cardiovascular
diseases, MI, and angina. Admissions for all cardiovascular diseases and MI were significantly associated
with all pollutants except O3. The only diagnostic subgroup signficantly associated with more than one
pollutant  was  MI.   Admissions for heart  failure, other IHD, and cerebrovascular disease were not
significantly associated with any pollutants.
       Additional single-pollutant and two-pollutant Poisson models were constructed for MI during the
warm  season (April through September) and the cool season (October through March).   Ambient air
pollution  levels were not reported  separately by season.  In the warm season, there were no  significant
associations of air pollution with MI admissions (p > 0.25). In the cool season, CO was positively and
significantly associated with MI in the single-pollutant model and the two-pollutant model with O3.  The
same was true for all other pollutants except O3, and SO2 was significant in all two-pollutant models.  The
p values and regression coefficients from Table 4 of Poloniecki et al. (1997), which show pollutant effects
in single-pollutant and two-pollutant models in the cool season, are presented in Table 6-4.
  Table 6-4. Acute Myocardial Infarction:  One- and Two-Pollutant Models with Cool Season3-",
	London, UK, April 1, 1987, to March 31, 1994	

                                                         Second Pollutant

 Test Pollutant                   O3            NO2              SO2               CO                BS

 03(ppb)

    p Value                    0.22           0.72              0.91               0.93               0.95
    Coefficient                -0.0013         0.0004             0.0001            -0.0001              0.0001

 NO2 (ppb)

    p Value                    0.0004°         0.002V            0.15               0.84               0.23
    Coefficient                 0.0022         0.0013             0.0008             0.0002              0.0008

 S02 (ppb)

    p Value                    0.0005°         0.03C              0.0004"            0.02C               0.03C
    Coefficient                 0.0025         0.0015             0.0021             0.0020              0.0015

 CO (ppb)

    p Value                    O.OOP          0.15              0.39               0.02e               0.38
    Coefficient                 0.0324         0.0205             0.0083             0.0227              0.0100

 Black smoke (,ug/m3)

    p Value                    0.0006°         0.23              0.10               0.13               0.002'
    Coefficient	0.0033	0.0014	0.0015	0.0019	0.0024

 "Diagonal elements (bold italics) are single-pollutant models; off-diagonal elements are test pollutant modeled with a second pollutant
 bCool season is October to March.  Coefficient = Poisson regression coefficient—for example, percent admissions per unit of pollutant = [exp
 (coeff)-!] x 100.
 °p<0.05.

 Source: Modified from Poloniecki et al. (1997). The original table shows results in both cool and warm seasons.
        This study suggests that, during the cool season in London, short-term elevations in ambient air
pollution levels are related to cardiac hospital admissions, especially MI admissions, 1 day later.  Unlike
North American investigators, Poloniecki et al. (1997) reported no association of ambient CO with heart
failure.   Even  so, there is some consistency between  the  European and North American studies.
                                                 6-14

-------
Specifically, these studies showpositive, statistically significant relationships of ambient CO concentration
and other pollutant concentrations with hospital admissions for heart disease, and these relationships tend
to be sronger for heart disease than other types of cardiovascular disease such as stroke.
       Poloniecki et al. (1997) present a balanced discussion of consistencies and inconsistencies in the
available epidemiologic database regarding ambient air pollution.  They emphasize the potential for
confounding of air pollution  and  meteorologic effects.   They  state,  "With so much  potential for
confounding, and with a good many conflicting results to be found in the scientific literature, we chose to
place the statistical emphasis on the direction and consistency of the relations (p values) rather than on
estimating the size of effects from the models which produce the largest effects, as has usually been done."
They also state that their results make a case for further study of CO.
       In Poloniecki  et al. (1997), a statistically significant association of ambient CO with angina
admissions might well have been expected, because ambient CO was associated significantly with MI
admissions, and because CO exposure reduces the time to angina onset in experimental studies of exercising
cardiac patients (see Section 6.2.2).  Failure to observe an association of ambient CO with angina could be
interpreted as a caveat on the observed association of ambient CO with MI. It is also conceivable that the
1-day lag employed in data analysis was not an optimum lag time for detection of any association of
ambient CO exposure and angina that might have existed.
       Morris and Naumova (1998) investigated joint effects of short-term ambient CO concentration and
ambient temperature on daily hospital admissions for CHF in people >65 years of age, in  Cook County
(Chicago), IL, from 1986 through 1989. Data were analyzed with general linear models (GLM) and general
additive models (GAM). The pollutant metrics assessed were daily maximum hourly levels of CO, NO2,
SO2, and O3, as well as 24-h average PM10. For each day of the study, gaseous pollutant measurements were
averaged across Cook County's eight monitoring stations, six of which were in downtown Chicago.  The
PM10 levels were measured at only one station on 80% of study days. The 25th and 75th percentiles of daily
maximum hourly CO, over the whole  study period, were  1.81  and 3.05 ppm, respectively.   The
corresponding  percentiles of 24-h average PM10 were 28 and 51 yUg/m3, respectively.  In addition to the
pollutant metrics, models included variables for daily maximum hourly temperature, day of week, month
of year, and year of study.
       In single-pollutant GLMs, the level of each pollutant except O3 was associated positively and
statistically significantly with CHF admissions on the same day. In a GLM that included all pollutants, only
CO was  associated significantly with admissions.  In this model, the RR for CHF admission at the
75th percentile of maximum hourly CO concentration was 1.08  (95% CI = 1.03, 1.12), as compared to
RR= 1 at a hypothetical, modeled CO concentration of zero. Associations of CO and other pollutants with
admissions were strongest on the day of admission (0-day lag) and weakened rapidly with successively
longer lag times.
       The authors conducted a detailed analysis to assess effects of ambient CO at different temperatures.
These effects were analyzed in three ways:  (1) inclusion of a CO-temperature interaction term in GLM;
(2)  simultaneous  inclusion of a CO term and a temperature  term in GAM, generating an additive
CO-temperature effects surface; and (3) analysis with stratification on daily maximum ambient temperature
(<40°, 40 to 75°, and >75 °F).  Effects of CO on CHF admissions consistently were associated inversely
with temperature (stronger effects at lower temperatures). For example, in a multi-pollutant GLM, RRs for
CHF admissions at the 75th percentile of maximum hourly CO concentration were  1.09 (95% CI = 1.01,
1.18), 1.07(95%CI=1.01, 1.13), and 1.01 (95%CI = 0.92, 1.11) when maximum temperature was <40,
40 to 75 °F, and >75 °F, respectively. Figure 3 in Morris and Naumova (1998) shows temperature-specific
exposure-response curves for percentiles of ambient CO distributions and relative risk of CHF admission
in single-pollutant and multi-pollutant models.  This figure is reproduced in Figure 6-2.  The authors
hypothesized that CHF patients may be unusually susceptible to  CO effects, and that cardiovascular and
other stresses imposed by cold weather may heighten this susceptibility.

                                             6-15

-------
•S 1.10-
     0   20  40  60  80   100
       CO Exposure Percentile
 20  40   60  80  100
CO Exposure Percentile
Figure 6-2. The relative risk associated with the exposure
percentiles of CO at specific temperature strata, based on
results of the  (A) multi-pollutant and (B) single-pollutant
generalized linear models of hospital admissions for heart
failure among the elderly in Chicago, IL, 1986 to 1989. (Note
that curves are  plotted by  percentiles of ambient  CO
concentrations, not measured concentrations.  Temperature-
specific percentiles of CO concentrations were not reported.)
Source: Morris and Naumova (1998).
                                                            Morris and Naumova (1998) is one of
                                                    the few time series studies that has modeled
                                                    temperature-specific  effects  estimates  for air
                                                    pollutant metrics. It is an unusually strong time
                                                    series study, and its findings regarding ambient
                                                    CO effects on CHF admissions are therefore of
                                                    unusual interest and concern.  However,  even
                                                    these findings should be interpreted cautiously.
                                                    For example, as the authors point out, fixed-site
                                                    CO measurements do not give exact estimates of
                                                    individual  subjects'  total   CO  exposures.
                                                    Discrepancies  between estimated  and actual
                                                    total  CO  exposures may be  greatest in  cold
                                                    weather, when indoor CO levels may be higher
                                                    than in warm weather, and when CHF patients
                                                    may spend  more time indoors.  Also, a given
                                                    day-to-day difference in ambient CO level could
                                                    reflect a greater day-to-day difference in  total
                                                    CO   exposure  in  cold  weather  than  in
                                                    warm weather.  Thus, the observed stronger
                                                    associations of ambient CO with CHF in colder
                                                    weather might reflect an underlying effect of
                                                    in creased total CO dose in colder weather, rather
                                                    than a specific effect of ambient CO exposure.
Finally, the observed associations could reflect a true effect of ambient CO fluctuations, superimposed on
higher baseline CO doses from nonambient sources during colder weather.
       As mentioned above, Burnett et al. (1997b) did not observe an  association of ambient CO  with
cardiac admissions.  However, their results do not necessarily contradict those of Morris and Naumova
(1998), who observed a strong association of CO with CHF admissions at cold temperatures but little or no
association at warm ones.  Pantazopoulou  et al.  (1995) and Poloniecki et  al. (1997) also observed
associations of CO and other air pollutants with cardiac admissions during cooler weather but not during
warmer weather. Thus, these four studies generate the hypothesis that ambient CO effects on heart disease
exacerbation are stronger during cooler weather than during warmer weather, a hypothesis that should be
tested further in future studies.
       Yang et al. (1998) studied cardiovascular hospital  admissions in relation to short-term CO levels
in Reno-Sparks, NV, during 6 years (1989 to  1994).  The study area is about 4,400 ft above sea level. At
this elevation, the effect of COHb on tissue O2 delivery might well differ from that at sea level. Exposure
to a given concentration of ambient CO might be expected to reduce tissue O2 delivery more at this
elevation than at sea level.  The range of ICD-9 codes studies (390 to 459.99) included heart disease and
other cardiovascular disorders such as stroke and hypertension.  The modeled daily CO metric was the
highest of the 1-h average CO concentrations occurring between midnight and noon.  The mean of this
metric over the study period was 3.09 ppm. Data were analyzed with weighted least-squares regression and
autoregressive integrated moving average (ARIMA) regression for time series data. Models were adjusted
with dummy variables for day of week and month of year. Wind speed and the previous day's minimum
temperature also were considered.  Carbon monoxide effects estimates were reported for the ambient CO
level on the day of admission (0-day lag). Pollutants other than CO were not considered.
       The  authors  stated,  "All hospital  admissions for CV [cardiovascular disease] and IHD were
significantly associated with CO concentrations...." Tabular displays of results were limited to all CV and
                                             6-16

-------
IHD admissions. No results were reported for other diagnostic subcategories. Modeled CO effects on
admissions were larger in males than females and did not differ significantly by age. The significance levels
of CO effects are somewhat difficult for the reader to discern. Table 5 of Yang et al. (1998) suggests that
regression results were highly significant (p < 0.006).  However, the text states, "According to the ARDVIA
models, CV andlHD hospital admissions increased 1.19% (95% CI = 0.99, 1.39%) and 2.83% (95% CI =
2.07, 3.60%), respectively, for each 1 ppm increase in the 1-h maximum CO level...." This suggests much
clearer significance of the CO effect on IHD admissions than for the effect of CO on total cardiovascular
admissions.   Overall,  this report suggests  an  association of  short-term ambient  CO levels  with
cardiovascular admissions, especially IHD admissions, at an elevation of 4,400 ft.  The report does not
enable inference as to effects of pollutants other than CO.
       Schwartz (1999) evaluated effects of short-term ambient CO and PM10 exposure on daily hospital
admissions for cardiovascular disease (ICD-9 codes 390 to 429) in persons aged at least 65 years, in eight
U.S. counties over 3 years, 1988 to 1990.  The analyzed health outcomes were the same as in Schwartz
(1997) (i.e., ICD-9 codes 390 to 429). The eight study locations were Chicago, IL; Colorado Springs, CO;
Minneapolis and St. Paul MN; New Haven, CT; and Seattle, Spokane, and Tacoma, WA. Previous findings
from Tucson, AZ, also were summarized (Schwartz, 1997).  Aero metric data were taken from population-
oriented CO monitors (neighborhood scale), rather than from compliance monitors, which may be located
away from population centers. Daily maximum 1 -h CO levels and 24-h average PM10 levels were analyzed,
apparently only on the same day as admissions (0-day lag). During the study period, 50th-percentile CO
and PM10 metrics ranged from 2.0 to 4.7 ppm and 23 to 37 yUg/m3, respectively, across the study locations.
The IQRs of the ambient CO and PM10 metrics, across all eight locations during the study period, were
about 1.75 ppm and 25 yUg/m3, respectively. Pollutants other than  CO and PM10 were not analyzed.
       Poisson regression was employed with locally weighted regression scatter plot smoothing (LOESS)
to adjust for effects of temperature, dew point, and seasonal patterns. Effects estimates were reported as
percentage increases in admissions over approximate IQRs of ambient pollutant concentrations. In all study
locations, both  CO and PM10 levels were associated positively with daily cardiac admissions.  Carbon
monoxide effects were statistically significant in seven of nine locations, andPM10 effects were significant
in six of nine locations.  Relative effects of CO and PM10 differed widely among locations. Overall, the
interquartile percentage increase in admissions for CO was 2.79% (95% CI = 1.89 to 3.68), and that for
PM10 was 2.48% (95% CI = 1.81  to 3.14). Effects  estimates for CO and PM10 were not related to the
location-specific correlation between these two pollutants' concentrations, suggesting their effects on
cardiac admissions were at least partially independent. Table 3 of Schwartz (1999), which shows modeled
ambient CO and PM10 effects in all eight study locations and  in Tucson, is reproduced  with slight
modification in Table 6-5.  In two-pollutant models, effects estimates for CO and PM10 were somewhat
smaller than in single-pollutant models. Pollution  effects  estimates were not reported for diagnostic
subcategories (e.g., IHD and heart failure).
       This study's findings leave a general impression of similar, but partially independent statistical
effects of short-term ambient CO and PM10 on total cardiac admissions in the elderly.  At the same time,
it is not clear why CO exerted  a substantially larger statistical effect than PM10 on admissions in
Minneapolis, but no CO effect was evident in the adjacent location of St. Paul. It is also not clear why the
LOESS seasonal smoothing window was 34% longer in St. Paul than in Minneapolis. Similarly, CO had
a larger effect than PM10 on admissions in Seattle, whereas the reverse was true in nearby Tacoma.
Furthermore, the seasonal smoothing window in Seattle was 34% longer than in Tacoma.
       Schwartz (1999) points out that, although CO is a known "cardiovascular toxin," cardiac effects in
experimental studies of humans have been observed only at higher than ambient CO levels.  He states that
one would expect to observe CO effects at lower levels in epidemiologic studies because epidemiology
examines the entire population, which comprises a broader range of disease states and exposure to potential
effects modifiers than do experimental subject groups. He also raises the possibility that ambient CO may

                                             6-17

-------
      Table 6-5. Modeled Percentage Increases in Hospital Admissions for Heart Disease,
         Associated with Interquartile Range Increases in Ambient PM10 (25 ^g/m^and
           Carbon Monoxide (1.75 ppm) in Eight Locations, Across These Locations,
                               and in Tucson, AZ, 1988 to 1990
City
Chicago, IL
Colorado Springs, CO
Minneapolis, MN
New Heaven, CT
St. Paul, MN
Seattle, WA
Spokane, WA
Tacoma, WA
Across the Locations
PM10
2.31
2.76
2.03
2.87
4.19
1.77
3.28
2.63
2.48
95% CL
1.31,3.33
-3.2,9.09
-1.87,6.09
1.04,4.73
1.44,7.00
-0.07,3.64
0.43,6.21
0.47,2.63
1.81,3.14
CO
2.84
0.51
4.09
3.04
0.74
4.22
2.71
1.84
2.79
95% CL
1.59,4.10
-2.41,3.51
1.59,6.65
1.18,4.93
-1.84,3.39
2.44,6.02
0.69,4.78
0.24,3.46
1.89,3.68
Window*
76
180
143
172
191
161
145
120
NA
  Above

 Tucson, AZ	2.99	0.55,5.50	2.94	0.54,5.71	83_

 aSize of window (in days) for LOESS smoothing of time to control season.

 Source: Modified from Schwartz (1999).
be a marker for automotive pollution, and states, "One potential constituent of automotive [pollution] that
may have acute cardiovascular toxicity is the volatile and semivolatile organic aerosols.... Hence, CO may
be serving as a proxy for these compounds. Nevertheless, associations between cardiovascular hospital
admissions and CO have been seen in more  than 20 North American cities,...suggesting that a better
understanding of this pollutant should have a high priority."
       Burnett  et al.  (1999)  assessed  daily unscheduled  hospital admissions  for  eight types of
cardiovascular and respiratory disease exacerbation in Toronto, Canada, in relation to short-term ambient
air pollution exposure over 15 years, 1980 to 1994. Daily average concentrations of CO, NO2, SO2, and O3
were obtained from four monitoring stations. Daily average concentrations of fine PM (PM2 5), coarse PM
(PM10_2 5), and PM10 were estimated from total  suspended particles (TSP) and sulfate levels measured at a
single central station. Daily average COH levels were measured at the same station. The cardiovascular
disease types were HTD, heart failure, cardiac dysrhythmias, cerebrovascular disease (including stroke), and
peripheral circulatory disorders. The respiratory disease types were asthma, obstructive lung disease, and
respiratory infection.
       Data were analyzed with general additive models, with accommodation for over- or underdispersion
from Poisson variation.  The data were prefiltered with a 31 -day LOESS smoother. LOESS smoothers also
were employed to adjust for effects of temperature and relative humidity. Models were adjusted for day-of-
week effects.  Exploratory models were run to select climate variables for final models. The report did not
present seasonal air pollution effects estimates. Pollution metrics were 1-, 2-, or 3-day averages, with the
final day of the averaging period lagged 0 to 2 days before admission. All possible single-pollutant models
were run for each disease type. Pollutant metrics  were constructed to give the log of relative risk per unit


                                             6-18

-------
change in pollutant concentration. Pollutant effects were reported as the percentage increase in total daily
hospital admissions, from zero concentration to the mean pollutant concentration during the study period.
       In the single-pollutant models (Table 3 in Burnett et al., 1999), pollutant effects estimates were
weak or negative for cerebrovascular disease and peripheral circulatory disease, so these two disease types
were not considered in subsequent multi-pollutant analyses.  This left only three heart disease categories
and three respiratory disease categories for further consideration. In single-pollutant models, the CO effect
in each of these six categories was strongest for a 1- or 2-day averaging time, ending on the same day as
the admission (0-day lag).
       The authors constructed multi-pollutant models in  an effort to ascertain effects of the overall
ambient air pollution mix and to estimate health benefits that would follow air pollution reduction. They
selected pollutant metrics from the single-pollutant models for inclusion in the multi-pollutant models.  For
each pollutant, they selected the averaging period and lag that yielded the largest ratio of log-relative  risk
to standard error (largest T-ratio or t-statistic). They constructed two  sets of multi-pollutant models.  In the
first set (Table 4 in Burnett et al., 1999), four models were run for each of the six cardiac and respiratory
disease categories. In one of these four models, all PM metrics were excluded. In each of the other three
models, the investigators forced inclusion of only one PM metric (PM25, PM10_25, or PM10). After these
constraints were imposed, gaseous pollutants were  entered into each of the 24 models using stepwise
regression. Specific gaseous pollution metrics in final models were selected using the Akaike Information
Criterion.
       In this first set of multi-pollutant models, inclusion of gaseous pollutants led to  substantially smaller
PM effects estimates than observed in  single-pollutant  models. Also, modeled percentage increases in
admissions (PIAs) with a PM metric in the model were only moderately larger than corresponding increases
with no PM metric in the model. The maximum diagnosis-specific percentage increase in the PIA with a
PM metric in the model,  over the PIA with no PM metric,  ranged from  7.4% for HTD to  28.1% for
respiratory infection. These increases were larger for respiratory diseases (average = 21.0%) than for heart
diseases (average = 8.5%).
       In the second set of multi-pollutant models (Table 5 in Burnett et al., 1999), all gaseous and PM
metrics competed equally for inclusion, and no metric was constrained to be included or excluded.  In these
models, percentage increases in diagnosis-specific hospital admissions, at the mean concentrations of
included pollutants, ranged from 9.29% (fflD) to 14.45% (respiratory infection). Effects of CO and NO2,
which are generated by mobile sources, largely were confounded.  Carbon monoxide, but not NO2,  was
included in models for asthma, obstructive lung disease, and  dysrhythmias, and NO2, but not CO,  was
included for HTD. Both CO and NO2 were included as independent predictors for heart failure. Taken
together, effects estimates for CO and NO2 were  larger and more frequently statistically significant for heart
diseases than for respiratory diseases.
       Table 6-6 summarizes modeled effects of CO in single-pollutant models and modeled effects of CO
and other pollutants in multi-pollutant models as taken from Tables 3 and 5 of Burnett et al. (1999).  No
PM metric was included in multi-pollutant models for IHD and heart failure (which accounted for 73.9%
of all U. S. hospital admissions for heart disease in 1996).  When a PM metric was included, the contribution
of the modeled PM effect to the overall pollution-associated increase in admissions ranged from 19.3%
(dysthythmias) to 42.1% (respiratory infection).  The  average contribution of PM metrics to the  air-
pollution-associated percentage increase in heart disease admissions was only 6.4% ([0% + 0% + 19.3%]
/ 3). If weighted by the number of admissions for each of the three analyzed heart disease categories in the
United States in 1996 (see Section 6.2.1.1), this contribution would  calculate to only 3.2%.  The average
contribution of PM metrics to percentage increases in respiratory disease admissions was 34.2%. Fine PM
was included in models for dysrhythmias and respiratory infection, and coarse PM was included for asthma
and obstructive lung disease. No more than one  PM metric was included for any disease, and PM10 was not
included for any disease. Ozone was associated much more strongly with respiratory disease admissions

                                              6-19

-------
Table 6-6. Modeled Percentage Increases in Hospitalizations at Mean
Pollutant Concentrations in Single- and Multi-Pollutant Models,
Toronto, Canada, 1980 to 1994
Single-Pollutant Models (Table 3)
Disease
Category CO NO2 SO2 O3 PM10 PM25
(1.2 ppm)' (25.2 ppb)' (5.4 ppb)' (19.5 ppb)' (30 //g/m3)' (18,ug/m3)'
Heart Disease
Dysrhythmias 8.99(3.60)b 5.33(1.73) 0.80(1.43) 3.51(1.71) 5.00(3.03) 4.33(2.91)
Heart Failure 8.33(5.71) 9.48(6.33) 1.93(3.85) 1.42(1.42) 5.75(3.51) 4.70(3.20)
Ischemic Heart 7.31(6.46) 9.73(8.40) 2.32(6.13) 0.61(0.99) 4.97(5.55) 5.73(6.08)
Disease
Respiratory Disease
Asthma 5.35(3.92) 3.33(2.37) 1.01(1.76) 6.32(4.63) 5.27(3.39) 4.60(3.22)
Obstructive 2.93(1.48) 2.21(1.07) 0.03(0.05) 7.29(4.23) 4.11(2.44) 3.42(1.89)
Lung Disease
0)
^ Respiratory 5.00(4.25) 6.89(5.53) 2.40(5.04) 4.42(4.29) 8.35(5.96) 7.64(6.09)
O Infection
"Overall study means of pollutant metrics are in parentheses.
Multi-Pollutant Models (Table 5)
All
PM10.25 CO NO2 SO2 O3 PM Pollutants
(12 //g/m3)'

2.47(1.88) 7.00(2.50) — c — 3.34(1.63) 2.47d(1.49) 12.81
3.77(2.79) 4.09(2.08) 6.89(3.44) — — — 10.98
1.81(3.02) — 8.34(6.10) 0.95(2.07) — — 9.29


5.25(4.20) 4.00(3.86) — — 4.99(3.48) 4.00e(3.04) 12.99
6.07(3.26) 3.00(1.52) — — 6.08(2.74) 3.86e(1.90) 12.94


4.44(4.00) — 4.44(3.31) — 3.93(3.80) 6.08d(4.46) 14.45


bT-statistics for pollutant effects estimates are in parentheses, and are approximately equal to Z-statistics. Two-tailed Z-statistics of 1 .96 or greater, and 2.58 or greater, denote statistical significance at a = 0.05
and a = 0.01, respectively.
°A dash denotes that the indicated pollutant was not included in the multi-pollutant model.
dThe PM metric included was PM2 5 (fine PM).
'The PM metric included was PM25.10 (coarse PM).
Source: Modified from Burnett et al. (1999).






-------
than with heart disease admissions.  Single-pollutant model effects of SO2 could be explained largely by
inclusion of other pollutants.
       In Burnett et al. (1999), PM metrics were estimated, not measured. Thus, the findings are subject
to quantitative uncertainty regarding specific pollutant effects and regarding gaseous pollution effects
relative to PM effects.  Modeled effects of PM may have been biased downward, but this is not certain.
Also, the extent to which findings in Toronto can  be generalized to other locations is  not  clear.
Nevertheless, this report presents an unusually comprehensive effort to examine multi-pollutant effects
across a variety of health outcomes. This study, like many other recent studies, emphasizes the importance
of considering both gaseous and particulate pollutants in data analyses. It underscores the importance of
gaseous pollutants as contributors to health effects  attributable to the overall ambient air pollution mix,
especially in relation to heart disease exacerbation. Also, unless both gaseous pollutants and PM are taken
into account, modeled effects estimates for single pollutants are likely to be inaccurate.  Furthermore, using
effects estimates derived from multi-pollutant models may compensate, at least partly, for confounding of
effects among individual pollutants in estimating potential public health risks.
       Time series studies of ambient CO and daily exacerbations of heart disease are summarized in
Table 6-7.

6.2.1.3   Ambient Carbon Monoxide and Daily Mortality Counts
       Epidemiologic  studies of the  relationship  between CO exposure and daily mortality are not
conclusive. Early studies in Southern California (Goldsmith andLandaw, 1968; Cohen et al., 1969; Hexter
and Goldsmith, 1971) suggested an association between atmospheric levels of CO and increased mortality
from cardiovascular disease, but potential confounders were not controlled thoroughly. In contrast, Kuller
et al. (1975) observed no association between ambient CO levels and cardiovascular disease or sudden
death in Baltimore,  MD.
       Kinney and Ozkaynak (1991) investigated effects of short-term variation in ambient air pollution
levels on daily nonaccidental, nonviolent deaths in Los Angeles County over 10 years, 1970 through 1979.
The mean daily death count was 152; of these,  87  (57%) and 8 (5%) resulted from cardiovascular and
respiratory causes, respectively. Pollution metrics were "daily maximum" CO, daily maximum hourly total
oxidants  (O3 in 1979),  24-h average SO2,  NO2,  and KM (a particulate metric similar to British smoke,
related to elemental carbon), and visual extinction coefficient (Bext, related to fine particulate). In multiple
regression models,  adjusted for meteorology and temporal patterns, there were statistically significant
associations of total and cardiovascular mortality  with temperature, oxidants (lagged 1  day) and the
automotive pollutants CO, NO2, and KM (each lagged 0 days). Levels of the latter three pollutants were
too highly intercorrelated to enable confident assessment of their separate effects.  Respiratory mortality
was associated with temperature but not with any pollution metric, although the power to test for pollutant
effects was limited by small numbers of daily respiratory deaths.
       More recent time series studies in North and South America and in Europe also have yielded mixed
results in relating day-to-day variations in CO levels with daily mortality. For example, no relationship was
found between CO and daily mortality in Los Angeles, Chicago, or Philadelphia (Ito et al., 1995; Kinney
et al., 1995; Ito and Thurston, 1996; Kelsall et al., 1997), after adjusting for ambient particles, time trends,
and weather.  Verhoeff et al. (1996) found no relationship between 24-h average CO concentrations and
daily mortality in Amsterdam, with or without adjustment for PM10 and other pollutants.
       Three other studies (Touloumi et  al., 1994; Salinas and Vega, 1995; Wietlisbach  et al.,  1996)
showed small, statistically significant relationships between CO and daily mortality.  However, effects of
other pollutants (e.g., TSP, SO2, NO2, BS)  and of meteorologic variables (e.g., temperature, relative
humidity) were also significant.   Further research  will be  needed to determine whether low-level CO
exposure actually is increasing mortality (particularly in the elderly population), whether CO is a surrogate
                                              6-21

-------
             Table 6-7. Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Frequency of Heart Disease Exacerbation
         Reference
                 Health Outcomes,
                 Study Locations,
                  Period of Study
                         Ambient Air Pollutants
                         Assessed and Pollutant
                            Concentrations
                                    Findings:
                              Single-Pollutant Models
                                    Findings:
                             Multiple-Pollutant Models
                               Comments
         Morris et al.
         (1995)
               Daily Medicare
               hospital admissions
               for CHF, seven U.S.
               cities,  1986-1989
         Schwartz and  Daily hospital
0)
ro
ro
Morris
(1995)
admissions for CHF,
IHD, and cardiac
dysrhythmias in
persons >65 years old,
Detroit, MI,
1986-1989
         Pantazo-
         poulou et al.
         (1995)
               Daily emergency
               department visits and
               hospital admissions
               for cardiac and
               respiratory causes,
               Athens, Greece, 1988
                       CO, NO2, SO2, O3.
                       The average of daily
                       maximum 1-h CO ranged
                       from 1.8 ppmin
                       Milwaukee, WI, to 5.6 ppm
                       in New York, NY.
CO, PM10, O3, SO2.
The average of daily
maximum 1-h CO =
2.4 ppm.  The average of
24-h average PM10 =
48.0 ,wg/m3.
                      Daily maximum 8-h moving
                      average CO, daily
                      maximum hourly NO2, 24-h
                      average BS. Mean
                      CO = 4.5 mg/m3 in winter
                      and 3.4 mg/m3 in summer.
RR for CHF admissions, per
10-ppm increase in CO,
ranged from 1.10 in
New York to 1.3 7 in
Los Angeles, CA. The CO
effect was statistically
significant in all cities but
Houston, TX. CO effects
were usually strongest with a
0-day lag.


RRs for CHF admissions,
per IQR increase in pollutant
level, were 1.022 for CO  and
1.024 for PM10.  RRs for
IHD admissions were 1.010
for CO and 1.018 for PM10.
Ambient air pollution was
not associated with
dysrhythmia admissions.
                           In winter, all three pollutants
                           were statistically
                           significantly associated with
                           cardiac and respiratory
                           hospital admissions. Air
                           pollution effects were
                           stronger for admissions than
                           for emergency department
                           visits. Ambient air pollution
                           was not associated with
                           admissions or visits in
                           summer.
                                                        The effect of CO was
                                                        more stable and retained
                                                        statistical significance
                                                        more often than other
                                                        pollutants' effects.
Effects of CO and PM10 on
CHF admissions were
independent and
statistically significant
(RR= 1.022 for CO and
1.024 for PM10). Effects
ofPM10onIHD
admissions retained
statistical significance, but
effects of CO and SO2 did
not.

None reported.
PM effects were not
assessed in this
study. The authors
estimated that about
3,250 CHF
admissions (about
0.4% of all CHF
admissions) could be
attributed to short-
term increases in
ambient CO.

CO effects on
admissions were
strongest with a
0-day lag.
                                                       Ambient CO levels
                                                       were high in
                                                       comparison to
                                                       current U.S. levels.
                                                       The report did not
                                                       mention lagged
                                                       analysis, so effects
                                                       presumably were
                                                       reported for a 0-day
                                                       lag.

-------
         Table 6-7 (cont'd). Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Frequency of Heart Disease Exacerbation
         Reference
  Health Outcomes,
   Study Locations,
   Period of Study
  Ambient Air Pollutants
  Assessed and Pollutant
      Concentrations
         Findings:
  Single-Pollutant Models
        Findings:
Multiple-Pollutant Models
     Comments
         Schwartz
         (1997)
         Burnett et al.
         (1997a)
0)
ro
         Burnett et al.
         (1997b)
Daily cardiovascular
hospital admissions in
persons >65 years old,
Tucson, AZ,
1988-1990
Daily hospitalizations
for CHF in persons
> 65 years old,
10 Canadian cities,
1981-1991
Daily hospital
admissions for cardiac
and respiratory
diseases, Toronto,
Canada, summers of
1992-1994
CO, PM10, SO2, NO2, O3.
Medians of daily maximum
hourly CO and 24-h average
PM10 concentrations =
3.03 ppm and 39 ^g/m3,
respectively. The IQRs for
CO and PM10 were about
1.7 ppm and 23 ^g/m3,
respectively.

CO, SO2, NO2, O3, COH.
Mean daily 1-h maximum
CO = 2.32 ppm.  Mean
daily 8-h maximum
CO= 1.59 ppm.
CO, fine PM (PM2 5),
coarse PM (PM10.2 5), PM10
(calculated), COH, O3, SO2,
NO2.  Mean and 95th
percentile of daily 1-h
maximum CO =1.8 ppm
and 3.2 ppm, respectively.
Effects of CO and PM10 on
admissions were of similar
magnitude and statistically
significant.  Admissions
were not appreciably
associated with SO2, O3, or
NO2.
The effect of CO on
hospitalizations was stronger
than effects of other
pollutants. The effect of CO
was strongest with a 0-day
lag.
All pollutants had positive
effects on cardiac and
respiratory admissions.
Effects were usually
statistically significant.
In a model that included
both CO and PM10,
percentage increases in
admissions, per IQR
increase in pollutant level,
were 2.33% for CO and
2.37% for PM10.
The effect of CO was less
sensitive to simultaneous
adjustment for other
pollutants and
meteorologic variables
than were effects of
non-CO pollutants.


CO had little effect on
admissions in two-
pollutant models that
included one PM metric
and one gaseous pollutant
metric. PM effects were
relatively insensitive to
adjustment for gaseous
pollutants.
Admissions for heart
disease constituted
most, but not all, of
the admissions
considered in this
analysis.
Size-specific PM
metrics were not
assessed. The authors
noted that CO could
be acting as a
surrogate for
transportation-related
air pollutants.

Ambient CO levels
were low. This
partially may explain
the lack of association
of CO with
admissions in this
study.

-------
         Table 6-7 (cont'd). Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Frequency of Heart Disease Exacerbation
         Reference
                 Health Outcomes,
                  Study Locations,
                  Period of Study
                         Ambient Air Pollutants
                         Assessed and Pollutant
                             Concentrations
                                     Findings:
                              Single-Pollutant Models
                                     Findings:
                             Multiple-Pollutant Models
                                Comments
         Poloniecki
         etal. (1997)
0)
ro
Morris and
Naumova
(1998)
Daily cardiovascular
hospital admissions,
London, England,
April 1, 1987-
MarchSl, 1994.
Admissions were
divided into seven
diagnostic subgroups.
Air pollution effects
on admissions for MI
were analyzed
separately in warm
and cool seasons.

Daily hospital
admissions for CHF in
persons >65 years old,
Chicago, IL,
1986-1989
                                      24-hmeanCO,BS, SO2,
                                      NO2. Hourly mean O3
                                      between 9:00 a.m. and
                                      5:00 p.m. Median and 90th
                                      percentile CO levels = 0.9
                                      and 1.8 ppm, respectively.
                                      Median and 90th percentile
                                      BS levels = 12 and
                                      22 Mg/m3, respectively.
Daily maximum hourly CO,
NO2, SO2, O3, and 24-h
average PM10.  Mean and
75th percentile CO = 2.51
and 3.05 ppm, respectively.
Mean and 75th percentile
PM10 = 41and51,wg/m3,
respectively.
                            Over all seasons, CO was
                            statistically significantly
                            associated with admissions
                            for all cardiovascular
                            diseases and MI. BSwas
                            significantly associated with
                            admissions for all
                            cardiovascular diseases, MI,
                            and angina. No  pollutant
                            was associated with MI
                            admissions in the warm
                            season.
In single-pollutant GLMs,
all pollutants but O3 were
significantly associated with
CHF admissions. In GLMs
stratified on ambient
temperature, CO effects were
strongest in the lowest
temperature range (<40 °F)
and weakest in the highest
temperature range (>75 °F).
CO effects were also
strongest at lower
temperatures in a generalized
additive model that generated
a joint CO-temperature
effects surface.
                             Two-pollutant models
                             were constructed. In the
                             cool season, CO and BS
                             were significantly
                             associated with MI
                             admissions in models that
                             included O3.  In the model
                             that included CO and BS,
                             neither pollutant had a
                             significant effect on MI
                             admissions.
In a GLM that included all
pollutants, only CO was
significantly associated
with admissions.  In
temperature-stratified
GLMs, CO effects were
somewhat smaller than in
single-pollutant GLMs, but
clearly still were
associated inversely with
ambient temperature.
                           Pollutant effects were
                           analyzed with a 1-day
                           lag. This report
                           presents a balanced
                           discussion of
                           strengths and
                           limitations of time
                           series studies.
                           Emphasizes potential
                           for confounding of air
                           pollution and
                           meteorologic effects.
Data were analyzed
with unusual attention
to joint effects of
ambient air pollution
and temperature.
Pollutant effects on
admissions were
strongest with a 0-day
lag, and decreased
rapidly with
successively longer
lag times.

-------
           Table 6-7 (cont'd). Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Frequency of Heart Disease Exacerbation
         Reference
Health Outcomes, Study
  Locations, Period of
         Study
  Ambient Air Pollutants
  Assessed and Pollutant
      Concentrations
          Findings:
   Single-Pollutant Models
          Findings:
   Multiple-Pollutant Models
        Comments
         Yang et al.
         (1998)
         Schwartz
         (1999)
0)
ro
en
         Burnett et al.
         (1999)
Daily cardiovascular
hospital admissions,
Reno,NV, 1989-1994


Daily cardiovascular
hospital admissions in
persons >65 years old,
eight U.S. counties,
1988-1990 (Results
from Tucson, AZ, also
were summarized.)
Daily hospital
admissions for five
categories of
cardiovascular disease
and three categories of
respiratory disease,
Toronto, Canada, 1980-
1994
CO only. Mean
CO = 3.09ppm.
CO,PM10.  The median of
daily maximum hourly CO
ranged from 2.0 ppm
(Chicago, IL) to 4.7 ppm
(Spokane, WA).  The
median 24-h average PM10
ranged from 23 Mg/m3
(Colorado Springs, CO) to
37 ,ug/m3 (New Haven, CT;
Spokane, WA; and Tacoma,
WA). The overall IQRs for
CO and PM10 were about
1.75 ppm and 25 Mg/m3,
respectively.

Measured daily average CO,
NO2, SO2, O3, COH.
Estimated daily average
PM10, fine PM, coarse PM.
Overall means of pollutant
metrics:  CO =1.2 ppm;
PM10 = 30 Mg/m3; fine
PM =18 Mg/m3; coarse
PM=12Mg/m3.
The effect of ambient CO
(0-day lag) on IHD admissions
and cardiovascular admissions
was statistically significant.

Effects of CO on admissions
were statistically significant in
seven of nine locations. PM10
effects were significant in six
of nine locations.  CO and
PM10 effects on admissions
appeared to be at least partially
independent.
All possible single-pollutant
models were constructed. CO
effects on admissions were
strongest with a 1-day or 2-day
averaging time, ending on the
same day as admission (0-day
lag). Pollutant effects were
weak or negative for
cerebrovascular disease and
peripheral circulatory disease.
None reported.
In two-pollutant models, effects
estimates for CO and PM10
were slightly smaller than in
single-pollutant models.
In two-pollutant models, PM
effects were substantially
smaller than in single-pollutant
models. In multi-pollutant
models, CO and NO2 effects
were usually confounded, but
CO and NO2 contributed
independently to admissions for
heart failure. PM metrics
accounted for 6.4% and 34.2%
of the overall ambient air
pollution effect on admissions
for cardiovascular and
respiratory diseases,
respectively.	
CO effects on admissions
were analyzed only with a
0-day lag.


Pollutant effects on
admissions were apparently
analyzed only with a 0-day
lag. Interpretation of
findings is complicated by
observations of different
relative effects of CO and
PM10, and different lengths
of LOESS smoothing
windows, in adjacent and
nearby locations.
This report employs a
thorough, logical strategy
for sequential use of single-
pollutant and multi-
pollutant models to estimate
air pollution-related health
risks. The authors
emphasized the importance
of considering the overall
ambient air pollution mix in
data analysis.

-------
marker for some other mobile-source or combustion-related pollutant, or whether CO is a surrogate for the
overall combustion-related or automotive pollution mix.
       Touloumi et al. (1994) investigated air pollution and daily all-cause mortality in Athens from 1984
through 1988. Daily mean pollution indicators for SO2, BS, and CO were averaged over all available
monitoring stations.  Autoregressive models  were used, with  log-transformed daily  mortality as the
dependent variable and with adjustment for temperature, relative humidity, year, season, day of week, and
serial correlations in mortality.  Separate models for log(SO2), log(BS), and log(CO) yielded statistically
significant effects estimates (p < 0.001). Air pollution measurements lagged by 1 day were associated most
strongly with daily mortality. In a multi-pollutant model, SO2 and BS were independent predictors of
mortality, although to a lesser extent than temperature and relative humidity.  Addition of an independent
variable for CO concentration did not improve this model's ability to predict daily mortality, suggesting that
CO may be a surrogate marker for other mobile-source or combustion-related pollutants.
       In one of the Air Pollution and Health—A European  Approach (APHEA) studies  in Athens,
Touloumi et al. (1996) observed a distinct positive association of ambient CO levels with daily mortality.
Ambient CO concentrations were compiled from three fixed outdoor monitoring stations over 5 years, 1987
to 1991. Median, mean, and maximum 8-h CO levels were 6.1 mg/m3 (5.3 ppm), 6.6 mg/m3 (5.8 ppm), and
24.9 mg/m3 (21.7 ppm), respectively. The relative risk for daily mortality of a 10 mg/m3 (9 ppm) increase
in the daily maximum 8-h ambient CO concentration was 1.10 (95% CI = 1.05,  1.15). This finding may
be attributable to  yet unknown health effects of ambient levels of CO,  to  the  presence of highly
compromised susceptible groups in the population, or, again,  to CO  acting as a surrogate for  other
combustion-generated air pollutants. Also, ambient CO concentrations were high during this study, and
this may have predisposed toward observing a positive statistical effect of CO on mortality.
       Salinas and Vega  (1995) examined  the  effect of urban air pollution on daily mortality in
Metropolitan Santiago,  Chile, from 1988 through 1991.  Measurements of maximum 8-h average CO;
maximum hourly O3; daily mean SO2, PM10, and PM2 5; and meteorologic variables were obtained from five
monitoring stations. Maximum ambient PM10 concentrations ranged from 339 to 500 /^g/m3 across the
monitoring stations in 1989.  Ambient concentrations of CO and fine PM were not reported. Total daily
mortality (excluding death by injury and poisoning) was regressed on daily maximum 8-h average CO, 24-h
average fine PM, humidity, and temperature, using Poisson regression. When all days were included in the
model, effects of CO, humidity, and temperature were statistically significant, but the fine PM effect was
not.  When the model included only days with fine PM levels under 150 //g/m3, the fine PM effect also
became significant. The authors constructed maps showing standardized mortality ratios for total and
respiratory mortality in different zones of the Santiago area. They reasoned from these maps, together with
the regression analyses described above, that ambient air pollution exposure might well be a more important
risk factor than socioeconomic status for total mortality and respiratory mortality in adults. However, no
analytical models included environmental and  socioeconomic variables simultaneously. Cardiovascular
mortality was not assessed.
       Wietlisbach et al.  (1996)  assessed associations between  daily mortality and  air pollution in
metropolitan Zurich, Basel, and Geneva, Switzerland, from 1984 through 1989. Daily counts were obtained
for total mortality, mortality in people 65 years  of age or older, and respiratory and cardiovascular disease
mortality. Daily measurements of weather variables and CO, TSP, SO2, NO2, and O3 were obtained in each
city (TSP and O3 measurements were not available in Geneva).  For each pollutant except O3, the 3-day
moving average ambient concentration, ending on the day of death (0-day lag), was used in the analysis.
A 1-day average was used for O3.  The study mean daily average CO ranged from 0.95 mg/m3 (0.83 ppm)
in Basel to 1.93 mg/m3 (1.68 ppm) in Geneva. The study mean daily average TSP was 46.2 /^g/m3 in Zurich
and 45.2 /^g/m3 in Basel.  Single-pollutant Poisson models were used to regress daily  death counts on
pollutant levels in each city, controlling for time trends, seasonal factors, and weather variables.
                                             6-26

-------
       In the Wietlisbach et al. (1996) study, air pollution-mortality associations in single-pollutant models
were generally somewhat stronger in the elderly than in the population as a whole and were somewhat
stronger for respiratory mortality than cardiovascular mortality.  Associations of mortality with CO were
somewhat weaker than with TSP, NO2, and SO2. In Zurich, TSP and NO2 were associated statistically
significantly with total daily mortality; TSP, NO2, and CO were associated significantly with mortality in
the elderly; all five pollutants were associated significantly with respiratory mortality (the association with
O3 was negative); and no pollutants were associated significantly with cardiovascular mortality. In Basel,
all pollutants but O3 were associated significantly with total mortality; all pollutants  were associated
signficantly with mortality in the elderly; TSP and NO2 were associated significantly  with respiratory
mortality; and all pollutants but O3 were associated significantly with cardiovascular mortality. In Geneva,
SO2 and NO2 were associated  significantly with total mortality; SO2, NO2, and  CO  were associated
significantly with mortality in the elderly and with respiratory mortality; and TSP and SO2 were associated
significantly with cardiovascular mortality.  When all pollutants  were  modeled simultaneously, the
regression coefficients were unstable and not statistically significant.
       In two recent studies, Burnett and colleagues investigated associations of CO and other pollutants
with daily nonaccidental mortality in Canada.   In one study (Burnett et al., 1998a), the  investigators
assessed the roles of 24-h average concentrations of ambient CO, other gaseous pollutants, sulfates, TSP,
COH, estimated PM2 5, and PM10 and meteorology in Toronto from 1980 through 1994. The overall  study
mean of 24-h average ambient CO concentrations was 1.18 ppm. Overall means of estimated PM10 and fine
PM concentrations were 30.2 and 18.0  yUg/m3, respectively. The time series was adjusted for long-term
trends and temporal cycles. Effects of several different exposure-to-mortality lags were explored, and the
final choice of lags was based on the Akaike Information Criterion.  A 2-day moving average was selected
as the optimum metric for CO, but not for all pollutants.  Final models  included same-day dew  point
temperature.   In single-pollutant models, ambient  levels of all pollutants except O3  were associated
positively and statistically significantly with daily mortality, and this association was strongest for CO.
Two-pollutant models also were constructed, each including CO and one of the other pollutants. In these
models, the magnitudes of relative risks for CO differed little from that in the single-pollutant model for
CO. In contrast, the relative risks for other pollutants generally decreased appreciably. Also, the relative
risks for CO remained statistically significant in all two-pollutant models. Although the relative risk of CO
was highest for deaths from cardiac causes, there was also a clear positive association of CO with deaths
from other causes.
       Burnett et al. (1998b) also examined associations of ambient levels of gaseous pollutants (CO, NO2,
O3, and SO2) with daily nonaccidental mortality in 11 Canadian cities from 1980 through 1991. In single-
pollutant models, relative risks of CO for mortality were more consistent across cities than were relative
risks of the other pollutants.  However, in multi-pollutant models, CO-associated relative risks decreased
substantially, and NO2 and SO2 appeared to explain much of the CO effect on mortality. The estimated
percentage increase in mortality frequency attributable to combined exposure to all four pollutants differed
widely among cities, ranging from 3.6% in Edmonton and Windsor to 11.0% in Quebec.  The authors
reasoned that reductions in gaseous pollutant levels might be more effective than reductions in PM levels
in reducing mortality. It is not possible to interpret this study quantitatively, because metrics of PM and
PM constituents were not included in the analyses. At the same time, these results underscore the need for
measurement and statistical treatment of abroad range of pollutants and for further systematic assessment
and comparison of the public health importance of exposure to ambient CO, other ambient gaseous
pollutants, and PM.
       There have been few studies of ambient CO and mortality in children. Sal diva et al.  (1994, 1995)
observed no association between CO and daily mortality among children or the elderly in Sao Paulo, Brazil,
after adjusting for nitrogen oxides and PM10, although Pereira et al. (1998) did report a relationship of
ambient CO concentration with intrauterine mortality. Interestingly, in the latter study, COHb levels in cord

                                              6-27

-------
blood were correlated with short-term ambient CO levels, although intrauterine mortality was associated
somewhat less strongly with CO than with other pollutants.  At the same time, Pereira et al. (1998) is
difficult to interpret because the investigators assessed fetal loss occurring only after 28 weeks of gestation,
whereas the large majority of spontaneous abortions occur before that time.
       Time series studies of ambient CO and daily mortality counts are summarized in Table 6-8.

6.2.1.4  Ambient Carbon Monoxide and Frequency of Respiratory Illness
       Short-term variation in ambient CO levels has been reported in several studies to be associated with
daily variation in indices of respiratory illness frequency. In most cases, these indices reflect exacerbation
of preexisting respiratory illness. Significant positive associations of ambient CO with respiratory illness
frequency have been observed less often than with heart disease exacerbation. Also, there is as yet no
demonstrated biological mechanism by which CO at ambient exposure levels could plausibly promote
respiratory illness exacerbation or new respiratory illness. Therefore, observed associations of ambient CO
with such exacerbation should be interpreted very cautiously and, by no means, should be considered
confirmed. At the same time, it is appropriate to discuss these associations because they indicate that short-
term ambient CO exposure may not be specifically linked epidemiologically only with heart disease. The
correct interpretation of this apparent lack of specificity is not yet known.  On one hand, it could suggest
that short-term ambient CO exposure effects are not confined to the cardiovascular system.  On the other
hand, it could also be taken as a caveat in regard to the observed associations of ambient CO with heart
disease exacerbation because, although the pathophysiologic connection of CO with respiratory disease is
more tenuous than with heart disease, statistical associations of CO with respiratory disease frequency have
nevertheless been reported.
       Sunyer  et al. (1991) investigated  daily emergency  department visits for chronic obstructive
pulmonary disease in Barcelona, Spain, in relation to short-term ambient air pollution levels during 1985
and 1986.  Emergency department visits for asthma  were  excluded.  Pollutants considered were daily
average and daily maximum hourly SO2, daily average BS, and daily maximum hourly CO, NO2, and O3.
Ambient levels of CO  were quite low during the study:  yearly mean of daily maximum hourly levels
= 5.4 mg/m3 (4.7 ppm), and 98th percentile =14.9 mg/m3 (13.0 ppm). In single-pollutant regression models
adjusted for meteorology, season, and day of week, positive, statistically significant effects estimates were
observed most consistently for SO2 lagged 0 or 1  day, but not 2 days.  Effects were significant for both daily
average and maximum hourly SO2 levels.  Effects of BS  and CO were also positive and statistically
significant (p < 0.01).  Effects  of NO2 and O3 were not significant. Air pollution effects on visits were
weaker in fall than in other seasons. Effects of SO2 remained positive and significant even when days with
daily average SO2 above 72 //g/m3 (181 days, 24.8% of all 730 study days) were excluded from analysis.
       Gordian et al. (1996) examined relationships between short-term ambient air pollution levels and
daily outpatient visits for asthma, bronchitis, and upper respiratory illness (e.g., sore throat, sinusitis,
earache, rhinitis) in Anchorage, AK, from May 1992 to March 1994. Numbers of visits were derived from
medical insurance claims by state and municipal employees and their dependents. Measurements of CO
were made only during winter months.  An increase of 10 //g/m3 in PM10 was associated with a 3 to 6%
increase in visits for asthma and a 1 to 3% increase  in visits for upper respiratory illness.  Winter CO
concentrations were associated with increased numbers of visits for bronchitis and upper respiratory illness,
but not  for asthma.  At the same time, these  CO concentrations were tightly correlated with overall
automobile exhaust emissions, including NO2, fine particles, and VOCs such as benzene.  Thus, visits for
respiratory illness could not be linked specifically to ambient  CO exposure.
       Yang et al. (1997) investigated asthma emergency room visits in Reno, NV, in relation to short-term
ambient air pollution levels during 3 years, 1992 to 1994.  Analytical methods were similar to those
used in Yang et al. (1998).  Briefly, there was a positive, statistically significant association of visits with
                                              6-28

-------
                    Table 6-8.  Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Mortality Counts
Reference
   Health Outcomes,
   Study Locations,
   Period of Study
    Ambient Air Pollutants
    Assessed and Pollutant
       Concentrations
         Findings:
  Single-Pollutant Models
         Findings:
  Multiple-Pollutant Models
        Comments
Kinney and
Ozkaynak
(1991)
Ito et al.
(1995)
Ito and
Thurston
(1996)
Kelsall et al.
(1997)
Daily nonaccidental,
nonviolent deaths from
all causes,
cardiovascular causes,
and respiratory causes,
Los Angeles County,
CA,  1970-1979
Daily nonaccidental
deaths in Los Angeles
County, CA, and Cook
County (Chicago), IL,
1985-1990

Daily nonaccidental
deaths in population
subgroups (e.g. race-
and sex-specific
deaths), Cook County
(Chicago), IL,
1985-1990

Daily mortality
(excluding external
causes) in persons <65,
65-74, and >75 years
old, Philadelphia, PA,
1974-1988
"Daily maximum" CO, daily
maximum hourly total
oxidants, 24-h average SO2,
NO2, KM, and visual
extinction coefficient.  Mean
annual levels of daily
maximum CO decreased over
time (11.2 ppm in 1970,
6.1ppminl979).

Daily maximum 1-h CO.
PM10, O3. Ambient CO
concentrations were not
reported.


CO, PM10, O3, SO2.  Mean and
90th percentile ambient CO
levels = 2.05 and 3.25 ppm,
respectively. Mean and 90th
percentile PM10 levels = 40.7
and 65.0 Mg/m3, respectively.


CO, TSP, SO2, NO2, O3. The
mean and 75th percentile of
daily CO levels = 1.74 and
2.05 ppm, respectively. Mean
and 75th percentile TSP = 67.3
and 82.0 Mg/m3, respectively.
None reported.
An association of mortality
with CO (2-day lag) was
observed in Los Angeles,
but not in Cook County.
Total mortality was
associated most consistently
with PM10 and O3. There
was no consistent
association of CO with
mortality.
CO was maximally
predictive of mortality with
a 3-day lag. TSP, SO2,
lagged CO, and O3 had
positive effects on
mortality.
In two-pollutant models, total
mortality and cardiovascular
mortality were associated
statistically significantly with
total oxidants (1-day lag) and
each of the following
pollutants:  CO, NO2, and
KM.
None reported for CO.
None reported for CO.
Lagged CO was not
associated statistically
significantly with mortality.
Levels of CO, NO2, and
KM were highly correlated.
Numbers of respiratory
deaths were too small to
give powerful tests of
ambient pollutant effects
on respiratory mortality.
Size-specific PM metrics
were not available.

The authors concluded that
mortality risk estimates for
PM10 depend on the choice
of monitoring sites to
include in analysis.

Air pollution effects on
mortality were especially
pronounced in black
women.
The effect of TSP on
mortality increased with
increasing age.

-------
                         Table 6-8 (cont'd).  Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Mortality Counts
        Reference
Health Outcomes, Study
  Locations, Period of
         Study
                                               Ambient Air Pollutants Assessed
                                                 and Pollutant Concentrations
                                         Findings:
                                  Single-Pollutant Models
                                       Findings:
                               Multiple-Pollutant Models
                                       Comments
        Verhoeff et al.  Total daily deaths in
        (1996)         Amsterdam, The
                       Netherlands, 1986-1992
        Touloumi      Daily all-cause
        et al. (1994)    mortality, Athens,
                       Greece, 1984-1988
         Salinas and
         Vega (1995)
Daily deaths, Santiago,
Chile, 1988-1991
0)
CO
o
        Burnett et al.
        (1998b)
Daily nonaccidental
mortality, 11 Canadian
cities, 1980-1991
        Wietlisbach
        etal. (1996)
Daily deaths from all
causes, respiratory
causes, and
cardiovascular causes,
Zurich, Basel, and
Geneva, Switzerland,
1984-1989
CO, PM10, BS, SO2, O3. Mean
and 75th percentile of daily
average CO = 0.97 and
1.21 mg/m3, respectively.  Mean
and 75th percentile of PM10 = 39
and 45 Mg/m3, respectively.

Daily average CO, BS, SO2.
Mean annual ambient CO levels
ranged from 4.63 mg/m3 (1988)
to 6.92 mg/m3 (1984).


CO, fine PM. Ambient
concentrations were not
reported.
CO, NO2, O3, SO2. Overall
mean of daily average ambient
CO =1.0 ppm, ranging from
0.4 ppm in London, Ontario, to
1.5 ppm in Toronto.
Daily CO, TSP, SO2, NO2, O3.
Mean ambient CO levels in
Zurich, Basel, and Geneva were
1.25, 0.95, and 1.93 mg/m3
(1.09, 0.83, and 1.68 ppm),
respectively.  Mean TSP levels
in Zurich and Basel were 46.2
and 45.2 Mg/m3, respectively.
Modeled effects of CO were
not statistically significant
and did not change
appreciably with lags of 0, 1,
and 2 days.


CO, BS, and SO2 were
associated with mortality.
Mortality-pollutant
associations were strongest
with a 1 -day lag.

None reported.
Modeled effects estimates for
ambient CO were more
consistent across cities than
were estimates for other
pollutants.
Associations of mortality
with CO were statistically
significant, but somewhat
smaller than with TSP, SO2,
andNO2. CO effects
appeared strongest with a
3-day lag.
                                                                                     Effects of CO were not
                                                                                     statistically significant in
                                                                                     models that also included BS
                                                                                     or PM10.  Effects of BS were
                                                                                     somewhat stronger than effects
                                                                                     ofPM10.

                                                                                     Addition of CO to a model
                                                                                     with BS and SO2 metrics did
                                                                                     not improve the model's ability
                                                                                     to predict mortality.
When all days were included,
CO was statistically
significantly associated with
daily deaths, but fine PM was
not. When analysis was
restricted to days with fine PM
<150 Mg/m3, both pollutants
were associated significantly
with deaths.

CO effects estimates were
substantially smaller than in
single-pollutant models.
Inclusion of SO2 and NO2
explained much of the
CO-associated effect on daily
mortality.

When all pollutants were
modeled simultaneously, their
estimated effects were unstable
and were not statistically
significant.
                              Modeled effects of
                              particulate metrics were
                              somewhat stronger than
                              effects of gaseous pollutant
                              metrics.
Analyses did not include PM
metrics, so quantitative
effects of gaseous pollutants
cannot be inferred with
confidence.
TSP measurements were not
available in Geneva.

-------
                        Table 6-8 (cont'd). Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Mortality Counts
        Reference
                Health Outcomes,
                 Study Locations,
                 Period of Study
                           Ambient Air Pollutants
                           Assessed and Pollutant
                               Concentrations
                                       Findings:
                                Single-Pollutant Models
                                    Findings:
                             Multiple-Pollutant Models
                                    Comments
        Saldiva et al.
        (1995)
        Saldiva et al.
        (1994)
0)
CO
Pereira et al.
(1998)
              Daily deaths in persons
              >65 years old, Sao
              Paulo, Brazil, May
              1990-April1991
              Daily deaths from
              respiratory causes in
              children <5 years old,
              Sao Paulo, Brazil,
              May 1990-April 1991
Daily intrauterine
mortality occurring
after 28 weeks of
gestation, Sao Paulo,
Brazil, 1991-1992
                       CO, PM10, NOX, SO2, O3.
                       Mean daily CO and
                       PM10 = 6.18 ppm and
                       82.4 Mg/m3, respectively.
                       CO, PM10, NOX, S02, 03.
                       Mean daily CO and
                       PM10 = 6.18 ppm and
                       82.4 Atg/nij, respectively.
CO, PM10, NO2, SO2, O3.
Mean and maximum ambient
CO levels = 5.7 and 12.5 ppm,
respectively. Mean and
maximum PM10 levels = 65.0
and 192.8 ^g/m3, respectively.
                              Modeled effects of CO and
                              SO2 were statistically
                              significant at a = 0.05.
                              Effects of PM10 and NOX
                              were significant at a = 0.01.

                              None reported (see
                              Comments).
The modeled effects of CO
and SO2 were statistically
significant at a = 0.10.
The effect of NO2 was
significant at a = 0.01.
The modeled PM10 effect was
statistically significant at
a = 0.01. Effects of other
pollutants were not
significant.

In a model with all pollutants,
the effects estimate for CO
was small, negative, and not
statistically significant (see
Comments). There was a
strong association of
childhood respiratory
mortality with NOX.

In a model with all pollutants,
the effect of NO2 was
significant at a = 0.01, and
effects of other pollutants
were not significant. The
overall air pollution effect
was best explained by a
composite air pollution index
that incorporated ambient
levels of NO2, CO, and SO2.
                                                        It is not entirely clear
                                                        whether the reported
                                                        pollutant effects estimates
                                                        are from single-pollutant or
                                                        multi-pollutant models.
The authors' use of a
composite air pollution
index is of interest. Even
so, the reported results are
difficult to interpret
quantitatively, because the
most spontaneous
abortions occur before
28 weeks of gestation. In a
second study, ambient CO
levels, adjusted for passive
smoking and birth weight,
were associated
significantly with cord
blood CO levels at
a = 0.01.

-------
the ambient O3 level lagged 0 days, but not 1 or 2 days. Visits were not associated with ambient CO or
PM10 levels.
       Two reports investigating asthma exacerbation in relation to short-term ambient levels of CO and
other air pollutants, published in 1999 by University of Washington investigators (Sheppard et al, 1999;
Norris et al., 1999), highlight difficulties inherent in efforts to specify single-pollutant effects in populations
exposed to complex ambient air pollution mixes. Sheppard et al. (1999) considered admissions for asthma
to 23 greater Seattle hospitals from 1987 through 1994 in persons aged <65 years. Fifty-four percent of
these admissions occurred in persons less than 20 years old. Pollutants considered were daily average CO,
PM10, PM25  (fine PM), PM10_25 (coarse PM), SO2, and daily maximum 8-h average O3. The CO monitors
were located in street canyons, not residential areas. The PM10 was measured with both the EPA reference
method and with light-scattering nephelometry.   Fine PM levels were estimated  largely  from PM10
measurements. Coarse PM levels were calculated as the difference between PM10 and fine PM levels. The
PM measurements from residential sites were given higher weight (80%) than those from an industrial site
(20%). During the study period, 50th-percentile ambient levels of average CO, PM10, fine PM, and coarse
PM were quite moderate (1.7 ppm and 27,13, and 14/^g/m3, respectively). Data were  analyzed to treat all
measured pollutants evenhandedly.  Semiparametric Poisson regression models were used and included
dummy variables for day of week.  Multiple lag times were considered; for each pollutant, the lag time
showing the strongest statistical  association with admissions was selected.  Single- and two-pollutant
models were constructed, and pollutant effects estimates  were reported over IQRs of ambient pollutant
concentrations.
       In single-pollutant models, CO lagged 3 days and O3 lagged 2 days were associated most strongly
with asthma admissions.  Associations of admissions with PM10, fine PM, and coarse PM, each lagged
1 day, were also positive and statistically significant.  The association of admissions with SO2 (lagged
0 days) was positive but not significant. In season-specific, two-pollutant models, generally similar effects
were observed for CO lagged 3 days and fine PM lagged 1 day. Over all seasons, effects of both pollutants
were positive  and significant.  Both pollutants were positively, and at least marginally significantly,
associated with admissions in fall, winter, and spring.  Both pollutants were negatively associated with
admissions in summer. Sheppard et al. (1999) stated, "In striving for a balanced approach to all measured
pollutants..., we observed unexpected associations for CO that dominated the PM effects. Nevertheless,...
there is no evidence for an effect [of CO] on the underlying physiology of asthma. CO may be an important
environmental indicator of incomplete combustion, particularly from mobile sources."
       Norris et al. (1999) investigated short-term ambient air pollution and emergency department visits
for asthma in Seattle children aged <18 years,  from September 1995 through December 1996. Analyses
were conducted for the entire study population, for urban children (in whom the reported overall annual
hospitalization rate was more than 600/100,000), and for suburban children (in whom the reported overall
rate was less than 100/100,000). Pollutants considered were daily average CO, PM10, and PMl (as measured
by nephelometer); hourly average SO2; and daily maximum hourly average and daily average NO2. During
the study period, mean ambient levels of CO, PM10, and fine PM were  1.6 ppm, 21.7 yUg/m3, and 12 /^g/m3
(estimated),  respectively. There were too few O3 measurements to include in analytical models. Relative
risks were reported over IQRs of pollutant concentrations.  Lag times of 0 through 4 days were  considered.
Models were  adjusted with dummy variables for day  of week, smoothing splines for time trends
(approximately a 2-mo moving average), ambient temperature, and dew point temperature.
       Pollutant effects on asthma visits usually were reported with a 1 -day lag. In single-pollutant models,
effects of PM]  and PM10 were consistently positive and at least marginally statistically significant. Effects
of CO were  also consistently positive, but not significant in the urban children. Effects of daily average
NO2, lagged 2 days, were positive and marginally significant.  Mixed results were observed  for other
pollution metrics. In multi-pollutant models, effects of PMl and PM10 remained positive and  statistically
significant, but SO2 and NO2 effects did not retain significance. Multi-pollutant models did not include CO

                                             6-32

-------
because CO was assumed a priori to be a surrogate for stagnant conditions.  Norris et al. (1999) stated,
"In summary, this study found a small but significant association between air pollution and increased ED
[emergency department] visits for asthma....  PM and CO concentrations...were associated with increased
childhood ED visits for asthma and represent the daily variation in incomplete combustion products...."
       The exclusion of CO from multi-pollutant models by Norris et al. (1999) appears dubious, because
both they and Sheppard et al. (1999) observed statistically significant associations of both PM and CO with
asthma exacerbation in single-pollutant models.  The assumption that CO is a surrogate for stagnant
conditions may well be valid, but no rationale was presented for why the same assumption could not be
made for the other combustion-related air pollutants that were included in the multi-pollutant models. Also,
if ambient CO is merely a biologically inert fellow-traveler with ambient PM, it would be difficult to
understand why, in Sheppard et al. (1999), the CO-asthma association was strongest when lagged 3 days,
whereas the PM-asthma association was strongest when lagged only 1 day. Furthermore, although it is true
that there is no known biological mechanism by which CO at ambient levels could exacerbate asthma, the
mechanistic linkage of combustion-related, nonbiological PM with asthma exacerbation remains to be more
clearly elucidated as well.  Finally, the authors' assumption that children under 18 years old are more
"susceptible" than the general population is somewhat questionable, because short-term elevations in
ambient air pollution levels have been associated most strongly with unequivocally harmful health effects
in the elderly, and because asthma death rates and  case-fatality rates are higher in adults than in children.
       Time series studies of ambient CO and daily  frequency of respiratory illness are summarized in
Table 6-9.

6.2.1.5  Ambient Carbon Monoxide and Low Birth Weight
       Low birth weight (typically defined as birth weight <2,500 g) is associated with infant mortality and
childhood morbidity and may predict increased risk of morbidity into adulthood (Joseph and Kramer, 1996;
Institute of Medicine, 1985). Although low  birth weight is probably not a direct cause of these harmful
outcomes, it is a useful marker for developmental disturbances that are more directly responsible (Weinberg
andWilcox, 1998).
       Alderman et al. (1987) conducted a case-control study of birth weight in relation to ambient CO
concentration in Denver, CO, from 1975 through 1983. The CO metric was the time-weighted geometric
mean ambient CO level as measured in the mother's neighborhood during the last 3 mo of gestation. The
large majority of mothers lived within 2 mi of their neighborhood monitoring  sites.  Median and 95th
percentile CO concentrations ranged among monitoring sites from 0.5 to 3.6 ppm and 0.8 to 4.8 ppm,
respectively.  Air pollutants other than CO  were  not considered  in analysis. Individual-level data on
maternal age, race, education, marital status,  parity, and prior pregnancy history were available from birth
certificates, but data on mother's personal CO exposure and smoking were not  available. Ambient CO
exposures were divided into quintiles and analyzed with Mantel-Haenszel methods, adjusting for mother's
race and education. No association of ambient CO level with frequency of birth weight <2,500 g was
observed in these analyses.  After consultation with exposure assessment experts, the investigators divided
subjects into two groups for whom the monitoring  data were considered to reflect true CO exposure more
and less accurately.  Separate analyses of these two groups were conducted. Interestingly, there was more
suggestion of a positive, monotonic CO effect in  the former group (p value for chi-square test of trend
equaled  0.07 [marginally significant]) than in  the latter group  (p  = 0.56  [not significant]).
       Ritz and Yu (1999) assessed low birth weight in southern California, in relation to ambient CO
levels, from 1989 through 1993. The health outcome  analyzed was incidence of birth weight <2,500 g in
singleton babies born at term (37 to 44 weeks of gestation), treated as a dichotomous variable.  Birth
weights <1,000 g and >5,500 g were excluded. The main exposure variable was the average ambient CO
level from 6:00 to 9:00 a.m. during the third trimester of pregnancy, as measured at the South Coast Air
Quality Management District (SCAQMD) monitoring station nearest the mother's residence.  This metric


                                             6-33

-------
                  Table 6-9. Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Frequency of Respiratory Illness
         Reference
  Health Outcomes,
   Study Locations,
   Period of Study
   Ambient Air Pollutants
   Assessed and Pollutant
      Concentrations
         Findings:
  Single-Pollutant Models
        Findings:
Multiple-Pollutant Models
      Comments
         Sunyer et al.
         (1991)
         Gordian et al.
         (1996)
0)
CO
         Yang et al.
         (1997)
         Sheppard
         etal. (1999)
Daily emergency
department visits for
chronic obstructive
pulmonary disease
(excluding asthma),
Barcelona, Spain,
1985 and 1986

Insurance claims for
daily outpatient visits
for asthma, bronchitis,
and upper respiratory
illness (URI),
Anchorage, AK, May
1992-March 1994
Daily emergency
room visits for
asthma, Reno, NV,
1992-1994
Daily hospital
admissions for asthma
in persons aged
<65 years, Seattle,
WA, 1987-1994
Daily maximum hourly CO,
NO2, O3, SO2. Daily average
SO2,BS. Mean
CO = 5.4mg/m3. 98th
percentile CO =14.9 mg/m3.
MeanBS = 72.9,wg/m3.
PM10, CO (fall/winter only).
MeanPM10 = 45.5 ^g/m3.
Mean CO = 2.5 ppm.
CO, PM10, O3.  Mean
CO = 4.6 ppm. MeanPM10
(gravimetric) = 38.0 ^g/m3.
Daily average CO, PM10, fine
PM, coarse PM, SO2. Daily
maximum 8-h average O3.
SO2 exerted the strongest
pollutant effects.  BS and
CO also had statistically
significant positive effects.
PM10 was associated with
increased numbers of visits
for asthma and URI.
Fall/winter CO was
associated with increased
numbers of visits for
bronchitis and URI.
Asthma visits were
associated with daily
maximum O3 lagged 0 days,
but not 1 or 2 days. Visits
were not associated with CO.

The strongest associations
were observed for CO lagged
3 days and O3 lagged 2 days.
Associations also were
observed for PM metrics
lagged 1 day.
None reported.
Models that included PM10
and CO simultaneously
produced essentially the
same results as single-
pollutant models. This
suggests that CO and PM10
effects were at least partly
independent.

None reported.
In two-pollutant models,
similar effects were
observed for CO lagged 3
days and fine PM lagged 1
day. Over all seasons,
effects of CO and fine PM
were positive and
statistically significant
Effects of CO could not
be separated confidently
from effects of other
automotive pollutants,
including benzene.
CO and fine PM exerted
similar statistical
effects, with different
lags. The authors noted
that CO may be an
indicator of incomplete
combustion, especially
from mobile sources.

-------
               Table 6-9 (cont'd). Summary of Time Series Studies of Ambient Carbon Monoxide and Daily Frequency of Respiratory Illness
        Reference
   Health Outcomes,
   Study Locations,
   Period of Study
    Ambient Air Pollutants
    Assessed and Pollutant
        Concentrations
         Findings:
   Single-Pollutant Models
        Findings:
Multiple-Pollutant Models
      Comments
        Norris et al.
        (1999)
Daily emergency
department visits for
asthma in persons aged
<18 years, Seattle, WA,
September 1995-
December 1996
Daily average CO, PM10
Hourly average SO2. Daily
average and daily maximum
hourly average NO2. Daily
average fine PM (estimated).
Daily average CO =1.6 ppm.
Daily average
PM10 = 21.7,wg/m3. Daily
average PM, = 0.4 m'1.
Effects of PMj and PM10
(1-day lag) were positive and
at least marginally statistically
significant.  Effects of CO
(1-day lag) were positive, not
always significant. Effects of
NO2 (2-day lag) were
marginally significant.
CO was not included in
multi-pollutant models.
In these models, effects of
PMj and PM10 retained
statistical significance, but
effects of gaseous
pollutants did not.
This study and
Sheppard et al. (1999),
show generally similar
statistical effects of PM
metrics and CO on daily
frequency of asthma
exacerbations.
0)
CO
en

-------
was chosen after consultation with SCAQMD aerometric experts. Births were included only if the mother's
residential zip code was entirely or largely within 2 mi of one of 18 SCAQMD stations. The large majority
of study mothers lived within 2 mi of the nearest station.  Thus, this study, like that of Alderman et al.
(1987), addressed the fact that ambient CO levels often exhibit considerable spatial variability. The overall
average third-trimester ambient CO level was 2.6 ppm; average CO levels ranged from 1.44 to 3.72 ppm
across the monitoring stations.  Data were analyzed with logistic regression, adjusting for gestational age,
parity, time since previous birth, infant's gender, and mother's age, educational level, and ethnic group.
Ecological variables for commuting habits, constructed from census data for the respective zip codes, also
were included.
       A total of 125,573 term births (92.1% of 136,376 eligible births) was included in analysis.  Birth
weight was <2,500 g in 2,813 (2.2%) of these.  The analysis predicted a 22% increase in incidence of low
birth weight among babies born to mothers with average ambient CO exposure above the 95th percentile
of 5.5 ppm (odds ratio 1.22 [95% CI 1.03, 1.44]). This rose to 33% when first births were excluded and to
54% for mothers <20 years old.  Average ambient CO levels during the first and second trimesters and
during all trimesters combined were not associated with low birth weight incidence.
       Measurements of O3, NO2, and PM10 were made at six of the 18 monitoring stations. Multi-pollutant
models were constructed for subj ects living near these six stations. In these models, ambient third-trimester
ambient CO levels were categorized into 0 to 50th percentile (reference category), 50 to 95th percentile (2.2
to 5.5 ppm),  and >95th percentile (>5.5 ppm). The authors stated, "The effects of CO appeared more
pronounced after adjustment for concurrent exposures to NO2, PM10, and ozone," although specific effects
estimates for the non-CO pollutants were not reported. In the multi-pollutant models, incidence of low birth
weight consistently increased with increasing ambient CO level.  Effects estimates for the highest CO
exposure category achieved statistical significance for births after the first birth and for births to mothers
<20 years old, but not for all births.
       Ritz and Yu (1999) pro vide justification for their choices of study population, exposure period (third
trimester), pollution averaging time (6:00 to 9:00 a.m.), and allowable distance of subjects' residences from
monitoring stations.  They identify important, relevant unmeasured factors such as maternal  smoking,
nutrition,  prepregnancy weight,  adverse pregnancy  experience,  and  occupational history.   They
acknowledge that confounding because of these factors is possible. They also argue that their study design
and analytical approach render serious confounding unlikely.
       Taken together, the Alderman et al. (1987) and Ritz and Yu (1999) studies tend to raise concern as
to whether contemporary ambient CO exposure is a risk factor for low birth weight. The findings have
some biological plausibility because the CO binding affinity of fetal Hb is somewhat greater than that of
adult Hb, and, at a given level of CO exposure, tissue O2 delivery is reduced more in the fetus than in the
child or the adult, in whom fetal Hb has been replaced by adult Hb  (Longo, 1976). The observation by
Pereira et al. (1998) of an association of ambient CO concentration with cord blood COHb level reinforces
this concern. Both Alderman et al. (1987) and Ritz and Yu (1999) recommend further research with
individual-level measurements of CO exposure and relevant covariates.
       At the same time, the Alderman et al. (1987) and Ritz and Yu (1999) studies are  not conclusive
because neither study controlled for relevant covariates at the individual level.  Lack of control for lead
exposure is noteworthy in this regard because lead, like  CO, is associated  with mobile sources.
Furthermore, lead is associated with a variety of harmful reproductive outcomes, including low birth weight
(U.S. Environmental Protection Agency, 1986, 1990).  Potential confounding by lead exposure may be of
particular concern for the Alderman et al. (1987) study, which assessed births occurring at least a decade
earlier than those assessed in the Ritz and Yu (1999) study.
       Studies  of ambient CO and low birth weight are summarized in Table 6-10.
                                             6-36

-------
                              Table 6-10. Summary of Epidemiologic Studies of Ambient Carbon Monoxide and Low Birth Weight
         Reference
 Health Outcomes,
 Study Locations,
  Period of Study
    Ambient Air Pollutants
    Assessed and Pollutant
        Concentrations
           Findings:
    Single-Pollutant Models
         Findings:
 Multiple-Pollutant Models
      Comments
         Alderman et al.   Case-control study of
         (1987)          birth weight, Denver,
                         CO, 1975-1983
         Rite and Yu
         (1999)
05
CO
Large ecologic study
of incidence of birth
weight <2,500 g,
southern California,
1989-1993
                     Lime-weighted geometric mean
                     ambient CO level during the last
                     3 mo of gestation (third trimester
                     of pregnancy). Lhe median CO
                     ranged from 0.5 to 3.6 ppm across
                     monitoring sites.
Mean ambient CO level from
6:00-9:00 a.m. during the third
trimester of pregnancy.  Overall
average ambient CO = 2.6 ppm
(range across 18 monitoring
stations = 1.44-3.72 ppm).  PM10,
O3, and NO2 measurements were
available from six of the
18 monitoring stations.
No association of ambient CO
with birth weight <2,500 g was
observed in the study population as
a whole. A marginally statistically
significant association was observed
in subjects for whom ambient CO
exposure estimates were considered
to be relatively accurate.

Analysis predicted a 22% increase
in low birth weight incidence in
mothers with average ambient CO
exposure above the 95th percentile
of 5.5 ppm. Lhe CO effect was
stronger for births after the first
birth, and in mothers <20 years old,
than in the study population as a
whole.
                                                                  None reported.
Effects of CO appeared more
pronounced than in single-
pollutant models.  Low birth
weight incidence was
monotonically related to
estimated ambient CO
exposure. Effects of CO in the
highest exposure category were
statistically significant for
births after the first birth, and
in mothers <20 years old.
                             Data on mother's
                             smoking, personal CO
                             exposure, and other
                             relevant characteristics
                             were not available.
Data on mother's
smoking, personal CO
exposure, and other
relevant characteristics
were not available.
Despite this, the authors
argued thoughtfully that
observed CO effects were
unlikely to be seriously
confounded with
unmeasured factors.

-------
6.2.2  Controlled Laboratory Studies
       The most extensive human experimental studies on the cardiovascular effects of CO have been those
conducted in predominantly young, healthy, nonsmoking subjects during exercise. Previous assessments
of those effects (U.S. Environmental Protection Agency, 1979,1984,1991;Horvath, 1981;Shephard, 1983,
1984) have identified what appears to be a linear relationship between the level of COHb in the blood and
decrements in human exercise performance, measured as maximal oxygen uptake. Short-term maximal
exercise performance significantly decreases at COHb levels ranging from 5 to 20% (Pirnay et al, 1971;
Vogel and Gleser, 1972; Ekblom and Huot,  1972; Weiser et al., 1978; Stewart et al., 1978; Klein et al.,
1980; Koike and Wasserman, 1992). One study (Horvath et al., 1975) observed a marginal decrease in
maximal exercise performance at a COHb level as low as  4.3% COHb.  Short-term maximal exercise
duration also has been shown to be significantly reduced at COHb levels ranging from 2.3 to 20% (Ekblom
and Huot, 1972; Drinkwater et al., 1974; Raven et al.,  1974a,b; Horvath et al., 1975; Weiser et al., 1978;
Koike and Wasserman, 1992).  The observed decreases in maximal exercise  performance and duration,
however, are so small that they are only of concern primarily for competing athletes, rather than for healthy
people conducting everyday activities at less than maximal exercise levels.  In fact, no significant effects
on oxygen uptake or on exercise ventilation and heart rate were reported during submaximal exercise at
COHb saturations as high as 15 to 20%  (see Section  10.3.2 in U.S. Environmental Protection Agency,
1991), especially at work rates below the metabolic acidosis threshold (Koike et al., 1991).
       Of greater concern at more typical ambient CO levels are certain cardiovascular effects during
exercise that are likely to occur in a smaller, but sizeable,  segment of the general population having a
deficiency of blood supply  (ischemia) to the heart muscle.   This group of patients with CAD and
reproducible exercise-induced angina  (chest pain) is regarded  as  the most sensitive risk group for
CO-exposure effects. Several important studies (Anderson et al., 1973; Sheps et al., 1987; Adams et al.,
1988; Kleinman et  al.,  1989; Allred et al.,
1989a,b, 1991) have provided the cardiovascular
database  for CO in  CAD patients.   In these
studies, discussed in detail  in  the previous
document   (see  Section   10.3.2  in   U.S.
Environmental   Protection   Agency,   1991),
significant ischemia was measured subjectively
by  the  time  of exercise   required  for  the
development of angina (time of onset of angina)
and  objectively  by  the  time required  to
demonstrate a 1-mm change in the ST segment
of the electrocardiogram. Adverse effects were
found with postexposure COHb levels as low as
3 to 6% when compared on the basis of optical
measurements (Figure  6-3).  This  represents
incremental increases of 1.5 to 4.4% COHb from
preexposure baseline levels.  Effects on silent
ischemia  episodes  (no  chest pain),  which
represent  the  majority  of episodes in these
patients, have not been studied.
       Only   one  new  study  has become
available  since  publication  of   the  1991
document.  As part of an investigation of CO
exposure  at   high  altitude,  17  men  with
documented CAD and stable angina performed
30-
25-

20-
15-

10-
5-
0-

-5-

I I I



I
.

Anderson et al.
(1973) |

-

I
.
Kleinman
Kleinman
etal.
(1989)
etal. T
(1998) 1
III I
T Allred et al.
^ (1989a,b,
1991) i


- Allred etal. T
(1989a,b, •
1991) I ,
1 Sheps etal.
(1987)

. Adams et al. -
1 (1988)

~
                      4                6
             Percent COHb by Optical Methods
Figure 6-3.  The effect of CO exposure on time to onset of
angina. For comparison across studies, data are presented
as mean percent differences between air- and CO-exposure
days for individual subjects calculated from each study. Bars
indicate calculated standard errors of the mean. The COHb
levels  were measured at the end of exposure;  however,
because of protocol differences among studies and lack of
precision in optical measurements of COHb, comparisons
must be interpreted with caution.
Source: Modified from U.S.  Environmental Protection
Agency (1991), Allred et al. (1989b, 1991).	
                                             6-38

-------
exercise stress tests after random 2-h exposures to either clean air or 100 ppm CO at sea level (Kleinman
et al., 1998; Leaf and Kleinman, 1996a; Kleinman and Leaf, 1991).  The methods used were similar to
those previously reported by Kleinman  et  al.  (1989).   Group  mean  COHb  levels  measured
by CO-Oximetry were 0.6 ± 0.3 (SD)% and 3.9 ± 0.5 (SD)% for clean air and CO exposures, respectively.
Repeated measures analysis of variance for a subgroup (n = 13) with angina on all test days demonstrated
a statistically significant (p < 0.05) decrease of 9.1 ± 0.6% in the time to onset of angina (from 5.94 to
5.40 min) during exercise after exposure to CO. The results are in good agreement with those observed in
the previously reported studies (see Figure 6-3). There was no statistically significant effect on ST segment
change, on the duration of angina, or on hemodynamic factors such as blood pressure and heart rate.
       Despite  clearly demonstrable effects of low-level CO exposure in patients with IHD, the adverse
health consequences of these types of effects are very  difficult to predict in the at-risk population of
individuals with heart disease. There is a wide distribution of professional judgments  on the clinical
significance of  small performance decrements occurring with the levels of exertion and CO  exposure
defined in the studies noted above. The decrements in performance that have been described at the lowest
levels (<3% COHb) are in the range of reproducibility of the test and may not be alarming to some
physicians. On the other hand, the consistency of the responses in time to onset of angina across the studies
and the dose-response relationship described by Allred et al.  (1989a,b, 1991) between COHb and time to
ST segment changes strengthen the argument in the minds of other physicians that, although small, the
effects could limit the activity of these individuals and affect their  quality of life. In addition, it has been
argued by Bassan (1990) that 58%  of cardiologists believe recurrent episodes of exertional angina are
associated with  a substantial risk of precipitating an MI (heart attack), a fatal arrhythmia (abnormal heart
rhythm), or slight but cumulative myocardial damage.
       Exposures to low levels of CO resulting in 5 to 20% COHb do not produce significant changes in
cardiac rhythm  or  conduction during rest or exercise in healthy  humans (Davies and Smith,  1980;
Kizakevich et al., 1994). Effects of CO on resting and exercise-induced ventricular arrhythmia in patients
with CAD are dependent on their clinical status.  Hinderliter et al. (1989) reported no effects of 4 and 6%
COHb in patients with IHD who did not have chronic arrhythmia (ectopy) during baseline  monitoring. In
more severely compromised individuals with higher levels of baseline ectopy, exposures to CO that produce
6% COHb have been shown to significantly increase the number and complexity of arrhythmias (Sheps
etal, 1990), but this does not occur at lower COHb levels (Sheps etal., 1990,1991; Chaitmanetal, 1992;
Dahmsetal, 1993). This finding, combined with the epidemiologic evidence of CO-related morbidity and
mortality noted  above and the morbidity and mortality studies of workers who are routinely exposed to
combustion products (e.g., Stern et al., 1981, 1988; Edling and Axelson, 1984; Sardinas et al.,  1986;
Michaels and Zoloth, 1991; Koskela, 1994; Melius, 1995; Strom et al., 1995), suggests that CO  exposure
may provide an  increased risk of hospitalization or death in patients with more severe heart disease.
       There also is  evidence from  experimental studies with  laboratory animals  that  CO  can
affect adversely  the cardiovascular system. The lowest-observed-effect level (LOEL) varies, depending on
the exposure regime used and species tested (see Table 6-11). Results from animal  studies (reviewed in
U.S. Environmental Protection Agency, 1979, 1991; Turino, 1981; McGrath, 1982; Penney, 1988, 1996a)
suggest that inhaled CO can cause disturbances in cardiac rhythm and conduction in healthy and cardiac-
impaired animals that are consistent with the human data. Results from animal studies (U. S.  Environmental
Protection Agency, 1991) also indicate that inhaled CO can increase Hb concentration and hematocrit ratio,
probably representing compensation for the reduction in oxygen transport caused by CO.  At high CO
concentrations,  excessive increases in Hb and hematocrit may impose an additional workload on the heart
and compromise blood flow to the tissues.
       There is conflicting  evidence  suggesting that  CO exposure may  enhance  development of
atherosclerosis in laboratory animals, but most studies show no measurable effect when the animals are
fed normal diets without added cholesterol,  even at high (=20%) COHb saturations (U.S.  Environmental

                                             6-39

-------
  Table 6-11. Estimated Lowest-Observed-Effect Levels for Cardiovascular Effects of Exposure of
  	Laboratory Animals to Carbon Monoxide	
                                  LOEL
 Health Effect Category
CO (ppm)    COHb (%)     Duration
Species
Reference
Cardiovascular effects
Cardiac rhythm
Cardiomegaly
Hemodynamics
Hematology
Atherosclerosis and
thrombosis

50
200
150
100
250


2.6
15.8
7.5
9.3
20.0


6 weeksa
30 daysa
30 min
46 daysa
10 weeksb


Dog
Rat
Rat
Rat
Rabbit


Preziosi et al. (1970)
Penney etal. (1974)
Kantenetal. (1983)
Penney etal. (1974)
Daviesetal. (1976)

 aContinuous daily exposure.
 Intermittent daily exposure, 4 h/day.
Protection Agency, 1979, 1991; Penn et al., 1992; Penn, 1993; Mennear, 1993; Smith and Steichen, 1993;
Strom et al., 1995).  Similarly, the possibility that CO promotes significant changes in lipid metabolism that
may accelerate atherosclerosis is suggested in only a few laboratory animal studies (see Table 10-7 in U. S.
Environmental Protection Agency, 1991) but not in humans (Leaf and Kleinman, 1996b); however, any
such effect must be subtle at most.  More recent in vitro studies utilizing cell culture techniques have
explored the hypothesis that CO  causes cellular  oxidative stress and leads to injuries of the vascular
endothelium that may precipitate atherosclerosis  (Thorn and Ischiropoulos, 1997; Thorn et al., 1997).
Unfortunately, the ability of environmentally relevant CO concentrations to  mediate this activity  in the
intact organism has not been  evaluated.  Finally, CO probably inhibits rather than promotes platelet
aggregation (U.S. Environmental Protection Agency, 1991; Min et al., 1992), lending support to forensic
observations that thrombosis is not a prominent feature of CO-mediated injury.  In general, there are few
data to indicate that an atherogenic effect of exposure is likely to occur in human populations at frequently
encountered levels of ambient CO.
6.3  Central Nervous System and Behavioral Effects
6.3.1  Brain Oxygen Metabolism
6.3.1.1  Whole Brain
       It has been documented amply in the literature that, as COHb is formed, vasodilation in the brain
(and increased blood supply) occurs sufficiently to keep the supply of oxygen (O2) to the brain constant
(Helfaer and Traystman, 1996; U.S. Environmental Protection Agency,  1991). The increased blood flow
adequately compensates not only for the oxygen supply decrease caused by reduced arterial O2 content
(CaO2), but it is also sufficient to compensate for the increased difficulty of extraction of O2 because of the
shifted oxyhemoglobin dissociation curve.  This compensatory vasodilation appears to be effective from
low levels to very high levels of COHb (at least up to 60%) and is similar in the fetus, neonate, and healthy
adult.
       Despite the compensatory regulation of O2 supply to the brain, it appears that O2 consumption,
measured as the cerebral metabolic rate for O2 (CMRO2), is reduced as COHb rises. The reason for this is
unclear, but the fact  is well documented (Doblar et al.,  1977; Jones  and Traystman, 1984; U.S.
Environmental Protection Agency, 1991; Langston et al., 1996). The amount of reduction in CMRO2 as
a function of COHb can be seen by combining the information of Doblar et al. (1977) from goats and
Langston et al. (1996) from sheep into one graph (see Figure 6-4).  Although information from Jones and
                                            6-40

-------
                       COHb (%)
Figure 6-4.  The relationship between COHb and CMRO2 for
goats and sheep.  Means from Doblar et al. (1977) were
taken from  their Tables 1 and 3, and CMRO2 values were
transformed to percent of baseline.  Figures 1 and  3 of
Langston et al. (1996) were converted digitally (Summasketch
III graphical to digital conversion) and also were transformed
to percent of baseline. Data from both sources were merged
into the same database and a logit function was fitted to the
data using PROC NUN (SAS Institute Inc., 1990). The solid
line is the best fit, dashed lines are the 95% confidence limits,
and the points plotted are means from the published studies.
                                                    Traystman (1984)  and associated studies was
                                                    expressed as a function of CaO2, not COHb, and
                                                    was difficult to incorporate into Figure 6-4 and
                                                    the associated analysis, their data corroborate
                                                    those of the other workers.
                                                            From Figure 6-4, it may be seen that the
                                                    CMRO2 does not decrease to 90% of baseline
                                                    until -27% COHb (95% confidence limits were
                                                    -21 to 32% COHb). The  data from sheep and
                                                    goats agreed with the results of Paulson et  al.
                                                    (1973), who reported that  the mean human
                                                    CMRO2 did not decrease significantly, even for
                                                    COHb up to 20%. Because Paulson et al. (1973)
                                                    did not report the value of their means, it was
                                                    not possible  to  include their results as data
                                                    points in Figure 6-4.

                                                    6.3.1.2 Subregions of the Brain
                                                            There are  a number of reports of the
                                                    blood-flow response to COHb of subregions of
                                                    the  brain (U.S.  Environmental   Protection
                                                    Agency,   1991).    The   results  generally
                                                    demonstrate that some areas of the brain have
                                                    less baseline blood flow than others, and that the
COHb- compensatory increase in blood flow is not the same for all areas. Generally, however, the percent
increases over baseline are nearly the same for all areas except the neurohypophysis (Hanley et al., 1986).
It is important to note that the latter area serves homoeostatic and not ongoing behavioral functions.  Thus,
it would appear that the subregions of the brain have compensatory increased blood  flow in the presence
of COHb that is similar to the whole brain.  To be sure, all possible regions of the  brain have not been
tested, but no evidence to indicate otherwise has been found.
       Work by Sinha et al. (1991), measuring regional capillary perfusion and blood flow in the presence
of COHb elevation, indicates that the problem of compensation for COHb-reduced CaO2 is not as simple
as indicated above.  Blood flow was measured using radiolabeled dye and capillary morphology was
measured by fluorescence microscopy. With these methods, there appeared to be an increase in the number
of perfused capillaries and in the amount of blood flow as COHb increased. Thus, not only may more blood
be delivered, but increased capillary perfusion would decrease the diffusion distance to the tissue.
       Presumably the compensatory mechanisms in subareas of the brain would work in a manner similar
to that of the whole brain and, thus, would show similar decreases in CMRO2  as COHb increases.
No corroborative studies, however, have been reported in the literature.
       Better and more detailed documentation of regional CMRO2in humans, as well as in other species,
seems appropriate but does not have high priority because not much evidence exists to suggest that the
results would differ from whole-brain results.  It appears that what is needed is not more descriptive work,
but an effort should be made to  understand the mechanism by which COHb elevation reduces CMRO2.
Furthermore, information is needed about brain conditions  under which brain compensatory mechanisms
might be impaired (e.g., injury,  inflammation, ailments  associated  with aging and co-exposure to other
pollutants). If such information were available, specific theoretical (biologically based) predictions could
be made, and behavioral experiments designed to test them.
                                            6-41

-------
6.3.2  Behavioral Effects of Carbon Monoxide
       The effects of CO on behavior, especially the ability to perform certain time discrimination tasks,
provided the scientific basis for the first CO NAAQS in 1971 (see Section 1.2).  As further research data
became available, however, the results on human behavior at lowlevels of CO exposure (<5% COHb) were
called into question and subsequently dismissed as the basis for the standard (U. S. Environmental Protection
Agency, 1979). After reviewing available studies, the previous criteria document (U.S. Environmental
Protection Agency, 1991) concluded that effects on behavior were demonstrated unambiguously in both
humans and laboratory animals at COHb elevations above 20%.  Below this  level, the results were less
consistent. The document also concluded, however, that it seems unwise to ignore the historical evidence
in favor of effects on human performance at COHb levels between 5 and 20% (e.g., Horvath et al, 1971;
Fodor and Winneke, 1972; Putz et al., 1976, 1979; Putz, 1979; Benignus et al., 1987). Even if behavioral
effects are small or occasional, they may be important to the performance of critical tasks.
       Behavioral experiments on the effects of elevated  COHb frequently have been marred  by
methodological problems.  In particular, experiments employing single-blind designs were shown to be
2.5 times as likely to find significant results as similar studies employing double-blind methods (Benignus,
1993,1996). This problem was noted previously, and reports of findings of behavioral effects of CO were
summarized with respect to whether a double-blind procedure had been followed (see Table 10-25 of U.S.
Environmental Protection Agency, 1991). From this summary, it was concluded that, at most, there was
credible evidence for effects on only three (somewhat  artificially defined)  categories of  behavior:
(1) tracking, (2) vigilance, and (3) continuous performance. Even within these categories, considering only
double-blind studies, it was noted that less than 50% of all studies found significant effects. Furthermore,
most of the double-blind experiments reporting significant results were not replicable.
       Benignus (1994) performed extensive meta-analyses of the CO-behavioral literature. Only double-
blind human CO studies were included  (see Section  10.4.2 of U.S. Environmental Protection Agency,
1991).  In this report, two dose-effect curves were estimated from the literature by converting all behavioral
endpoints  to percent of baseline.  A dose-effect curve for COHb and  behavior was estimated from  rat
experiments and corrected for the effects of hypothermia. The COHb for rats was estimated from exposure
conditions by use of a rat-specific version of the Coburn-Forster-Kane Equation (Benignus and Annau,
1994).  Another dose-effect curve was estimated from the human literature on hypoxic hypoxia, which was
converted to equivalent COHb via equal arterial oxygen contents and corrected for effects of hypocapnia.
These two curves virtually overlapped each other. Human data points from CO behavior experiments then
were plotted onto the curve fitted to the rat CO data (no curve was fitted to human data because of the small
effect sizes and small COHb levels).  The conclusion from this meta-analysis was that human behavioral
impairments of 10% (ED-10) should not be  expected until COHb exceeds  20%.
       Data for the rat studies from Benignus (1994) were refitted for present purposes using the same
dose-effect model (a logit)  as for the CMRO2 data above (originally a different function was used); the
results are plotted in Figure  6-5.  With the logit function, it is estimated that a 10% decrement should be
produced in rats by -25% COHb (95% confidence limits of -20 to 30%). Data from all available double-
blind human studies also were converted to percent of baseline and plotted (Figure 6-6), along with the logit
curve fitted to the rat data (from Figure 6-5).
       The human data plotted in Figure 6-6 were mostly not statistically significant (thin solid lines) and
seem, as a group, not to have a dose-related trend of decrements; thus, it could be argued reasonably that
effects in humans cannot be shown to differ from those in rats. Some of the human data, however, at low
levels of COHb (4 to 10%) do appear belowbaseline and were declared statistically significant (thin dashed
lines) by the authors of the original reports.
       The low-COHb significant results plotted in Figure 6-6 invariably were reported in studies in which
only a few levels of COHb were evaluated. The two studies in which more and higher COHb levels were
tested invariably did not find statistically significant effects, even at much higher levels. Furthermore,  for

                                            6-42

-------
           0     10    20    30     40    50
                       COHb (%)
Figure  6-5.   The relationship between  COHb and
behavior   effects in rats.   Points  plotted  are  the
hypothermia-corrected means from four studies of effects
of COHb in rats in which COHb was estimated by a rat-
specific version of the Coburn-Forster-Kane Equation
(see Benignus, 1994). The solid line  is a best-fit logit
curve to  the  rat  data.   The dashed lines are  95%
confidence limits.   The data points are means  from
various studies coded by letter as follows: A = Ator
(1982), B =Atoretal. (1976), C = Smith etal. (1976), and
D = Merigan and Mclntire (1976).	
Figure  6-6.   The relationship between  COHb  and
behavior effects in humans compared to rats.  Points
plotted are means from various human studies.  The
heavy dashed line is the line fitted to rat data as shown in
Figure 6-5. The points for the human data are connected
by thin lines to depict the dose-effect curve found in each
study.  Solid thin lines are from studies in which no
significant effect was found; dashed lines are from studies
in which a significant effect was reported. The data points
are  coded such  that alphabetic symbols are from
nonsignificant studies and other symbols from significant
studies.   Investigators are coded as follows:  X =
Benignus et al. (1990), C = Benignus et al. (1977), K =
Groll-Knapp et al.  (1982), U = Harbin et al. (1988), O =
Otto et al. (1979), W = Ramsey (1973), P = Roche et al.
(1981), E = Stewart et al. (1973), V = Stewart et al.
(1970), & = Weir etal. (1973), Z = Wright etal. (1973), + =
Benignus et al. (1987), $ = Putz et al. (1976), * = Putz
etal. (1979).
every study reporting low-COHb level impairments,
other studies failed to replicate the findings, or
highly similar studies failed to find effects.
        In  summary,   no   reliable   evidence
demonstrating decrements in neural or behavioral
function in healthy young adult humans has been
reported for  COHb levels below 20%, and even
these studies are untested by replication.   The
low-COHb behavioral effects that have sometimes
been reported cannot be taken at face value because
they are not reliably repeatable, and they do not fit
into wider range, dose-effect patterns reported in
other studies. It is more reasonable to conclude that
no statistically detectable behavioral impairments
occur  until  COHb exceeds 20  to 30%.   The
conclusion, based on behavioral evidence alone, is
bolstered by  the findings that whole-brain CMRO2
is not reduced by a similar amount until COHb rises
to 21 to 32%. Because a dose-effect curve has been
fitted, any level of effect may be considered (e.g.,
ED-5, ED-20).  The interpolation of a curve to an
ED-5 point would imply that the COHb levels for
such an effect size would be  15 to  26%. Such an
interpolation is  more speculative than an ED-10,
however,  because the experimental  verification
would  be  difficult, requiring large numbers of
subjects and careful  control of error variance.
Additionally, as interpolation  approaches  small
effect   sizes, the  error  possibility because  of
statistical  model   selection  (threshold  versus
continuous) increases dramatically.
        Behavioral work should be encouraged to
determine whether reliable decrements in behavior
truly  are  associated with low levels  of COHb.
However,  any new experiments should involve
several  CO  exposure  levels, including one high
enough to produce changes.  In addition, inclusion
of some other procedure or  a reference dose of
some other active substance would serve to verify
the sensitivity of the behavior under study, thereby
facilitating  interpretation  of  any  negative  data
collected at the chosen COHb levels. Studies that
do not satisfy these specifications most likely would
be unfruitful and only further confuse the present
understanding  of the  published  literature.    In
addition, other experiments should be designed to
contribute  to the understanding of how CMRO2
relates to COHb elevation and behavioral changes.
                                              6-43

-------
       Behaviors implicated by the research findings involve detection of infrequent events (vigilance),
hand-eye coordination (compensatory tracking), and other forms of continuous performance (U.S.
Environmental Protection Agency,  1991).  Because of the unreliability of the findings, however, it is
questionable whether these behaviors should be cited as effects. Until better evidence of reliable behavioral
effects are published, preferably in  studies that demonstrate dose-related changes, conclusions must be
formed from the extant set of experimental data.
       Because COHb elevates brain blood flow, it has the possibility of altering the delivery of other
toxicants to the brain or altering the biotransformation or elimination of toxicants (e.g., Doi and Tanaka,
1984; Kim and Carlson, 198 3; Roth and Rubin, 1976a,b). In combination with exercise or hypoxic hypoxia,
the interactions would become even more  complex.   Disease and ailments associated with aging
concomitant with all of the above also  could be important.  Interactions such as these are understood from
physiological theory and could be given quantitative estimates through the use of physiological simulation
using whole-body physiological models that are currently under development.
6.4  Developmental Toxicity
       An issue directly relevant to the protection of public health is the potential effect of CO on growth
and function of the developing fetus, infant, and child. Results obtained from new research on this specific
outcome of CO exposure (e.g., Carratu et al, 1993, 1996; Di Giovanni et al, 1993; De Salvia et al., 1995;
De Luca et al., 1996) have not changed the conclusions presented in Section 10.5 of the previous criteria
document (U.S. Environmental Protection Agency, 1991).  From all of the laboratory animal studies, it is
clear that severe, acute CO poisoning can be fetotoxic, although specification of maternal and fetal COHb
levels is difficult because such exposures rarely involve the achievement of steady-state COHb levels or
permit careful and rapid determination of COHb levels.  Available data (reviewed in U.S. Environmental
Protection Agency, 1991; Annau and Fechter, 1994; Carratu et al., 1995; Penney, 1996b) provide strong
evidence that maternal CO exposures of 150 to 200 ppm, leading to approximately 15 to 25% COHb,
produce reductions in  birth weight, cardiomegaly, delays in behavioral development, and disruption of
cognitive function in laboratory animals of several species. Isolated experiments (Prigge and Hochrainer,
1977; Abbatiello and Mohrmann, 1979; Singh,  1986) suggest that some of these effects may be present at
concentrations as low as 60 to 65 ppm (approximately 6 to 11% COHb) maintained throughout gestation.
Studies relating human CO exposure from ambient sources or cigarette  smoking to reduced birth weight
(e.g., Martin and Bracken, 1986; Rubin et al., 1986; Alderman et al., 1987; Wouters et al., 1987; Brooke
et al., 1989; Spitzer et al., 1990; Wen et al., 1990; Peacock et al., 1991a;  Zaren et al., 1996; Jedrychowski
and Flak, 1996; Seeker-Walker et al., 1997) are of concern because of the risk for developmental disorders
(Olds et al., 1994a,b; Olds, 1997); however, many of these studies have not considered all sources of CO
exposure, other pollutants (Wang et al., 1997), or other risk factors during gestation (Peacock et al., 1991b;
Luke, 1994;Robkm, 1997).
       Results from laboratory animal studies suggest that exposure to lower levels of CO, leading to < 10%
COHb, should not have much of an effect on the developing fetus until possibly later in gestation when the
embryo  is much larger and more dependent on transport of oxygen by red blood cells  (Robkin, 1997).
In addition, results from a multicenter, prospective study (Koren et al., 1991) of fetal outcome following
mild to moderate accidental CO poisoning in pregnancy suggest that hypoxemia associated with measured
COHb saturations of up to 18% (or even higher estimated levels) does not impair the growth potential of
the fetus when pregnancy continues normally. Therefore, it is unlikely that ambient levels of CO typically
encountered by pregnant women would cause increased fetal risk. It is necessary, however, to consider the
combined effects of CO with the other common risk factors that may cause adverse fetal outcome (e.g.,
tobacco use, lead exposure, alcohol consumption, genetic background, maternal general health, obstetric
history).

                                             6-44

-------
       One of the more important determinants of the course and outcome of pregnancy that was not
previously discussed is maternal-fetal nutrition (Luke,  1994). Laboratory animal studies conducted to
determine the combined effect of gestational CO exposure and nutritional deficiency suggest that CO has
a greater effect on the fetus in protein-deficient mice (Singh and Moore-Cheatum, 1993; Singh etal., 1993).
Reductions in the rate of pregnancy, lower fetal weights, and increased fetal malformations were reported
at CO concentrations as low as 65 ppm maintained between 6 and 23 h per day during the first trimester
of pregnancy (Gestational Days 8 through 18).  Previous evidence of the fetotoxic and teratogenic effects
of CO in laboratory animals (U.S. Environmental Protection Agency, 1991) came largely from high levels
of exposure (i.e., in the range  of 500 ppm for rodents).
       There are studies (e.g., Schoendorf and Kiely,  1992; Scragg et al, 1993; Mitchell et al., 1993;
Klonoff-Cohen et al., 1995; Blair et al., 1996; Hutter and Blair, 1996; MacDorman et al., 1997) linking
maternal  cigarette smoking with sudden  infant death syndrome (SIDS), but the role of CO is uncertain,
especially in relation to other known risk factors for SIDS, such as developmental abnormalities (Schwartz
et al., 1998), prone sleeping (Kahn et al., 1993; Franco et al., 1996), overheating (Douglas et al., 1996), and
soft bedding (Ponsonby et al.,  1993; Kemp et al., 1998).  Data from human populations (Hoppenbrouwers
et al., 1981) suggesting a link between ambient CO exposures and SIDS are weak, but further study should
be encouraged. Children may experience neurological symptoms such as dizziness or fainting after an acute
episode of CO poisoning (>15% COHb),  or, in some cases, neurological impairment may develop days to
weeks after very high exposures (Crocker and Walker, 1985). Human data from these cases of accidental
high CO exposures are difficult to use in identifying a LOEL for CO because of the small number of cases
reviewed and problems  in documenting exposure levels. However, such data, if systematically gathered
and reported, could be useful in identifying possible ages of special sensitivity to CO and co-factors or other
risk factors that might identify sensitive subpopulations.
6.5  Acute Pulmonary Effects
       It is unlikely that CO has any direct effects on lung tissue, except for extremely high concentrations
that can cause cell damage and edema (Niden and Schulz, 1965; Fein et al., 1980; Burns et al., 1986). No
new information has been published in the literature to change this conclusion drawn from Section 10.2 of
the previous criteria document (U.S. Environmental Protection Agency, 1991).  Experimental studies on
the effects of CO exposures producing COHb saturations up to 56% failed to find any consistent effects on
pulmonary cells and tissue or on the vasculature of the lung (Fisher et al., 1969; Weissbecker et al., 1969;
Hugod, 1980; Chen et al., 1982; Shimazu et al., 1990).  Human studies on the pulmonary function effects
of CO are complicated by the lack of adequate exposure information, the small number of subjects studied,
and the short exposures explored. Decrements in lung function have been observed with increasing severity
of CO poisoning (Kolarzyk, 1994a,b, 1995). For example, occupational or accidental exposure to the
products of combustion and pyrolysis, particularly indoors, may lead to acute decrements in lung function
if COHb levels are greater than 17% (Sheppard et al., 1986) butnot at saturations less than 2% (Cooper and
Alberti, 1984; Hagberg et al., 1985; Evans et al., 1988). It is difficult, however, to separate the potential
effects of CO from the  effects of other respiratory irritants in smoke and exhaust. Community population
studies on CO in ambient air have generally not found strong relationships with pulmonary function,
symptomatology, and disease (Lutz, 1983; Robertson and Lebowitz, 1984; Lebowitz et al., 1987).
6.6  Other Systemic Effects of Carbon Monoxide
       Laboratory animal studies (reviewed in Section 10.6 of U.S. Environmental Protection Agency,
1991) suggest that enzyme metabolism and the P-450-mediated metabolism of xenobiotic compounds may
be affected by CO exposure (e.g., Montgomery and Rubin, 1971; Pankow et al., 1974; Roth and Rubin,

                                            6-45

-------
1976a,b,c). Most of the authors of these studies have concluded, however, that effects on metabolism at
low COHb levels (< 15%) are attributable entirely to tissue hypoxia produced by increased levels of COHb
because the effects are no greater than those produced by comparable levels  of hypoxia produced by
insufficient oxygen delivery.  No new studies have been published at CO levels relevant to ambient
exposures. At higher levels of exposure, where COHb concentrations exceed  15 to 20%, there may be
direct inhibitory  effects of CO on the activity of mixed-function oxidases,  but more basic research is
needed. The decreases in xenobiotic metabolism shown with CO exposure may be important to individuals
receiving drug treatment.
       Inhalation of high levels of CO, leading to COHb concentrations greater than 10 to 15%, have been
reported to cause a number of other systemic effects in laboratory animals and effects in humans suffering
from acute CO poisoning.  Tissues of highly active oxygen metabolism, such as heart, brain, liver, kidney,
and muscle, may be particularly sensitive to CO poisoning.  The impairment of function in the heart and
brain caused by CO exposure is well known and has been described above. Other systemic effects of CO
poisoning are not as well known and are therefore less certain.  There are reports of effects on liver
(Katsumata et al, 1980), kidney (Kuska et al, 1980), bone (Zebro et al, 1983), and immune capacity in
the lung and spleen (Snella and Rylander, 1979).  It generally is agreed that these effects are caused by the
severe tissue damage occurring during acute CO poisoning resulting from one or more of the following:
ischemia resulting from the formation of COHb, inhibition of oxygen release from oxyhemoglobin,
inhibition of cellular cytochrome function (e.g., cytochrome oxidases), and metabolic acidosis.
6.7  Physiologic Responses to Carbon Monoxide Exposure
       The only evidence for short- or long-term compensation to increased COHb levels in the blood is
indirect. Experimental animal data (reviewed in Section 10.7 of U.S. Environmental Protection Agency,
1991) indicate that incremental increases in COHb produce physiological responses that tend to offset the
deleterious effects of CO exposure on oxygen delivery to the tissues. Experimental human data (presented
in a report by Kizakevich et al., 1994) indicate that compensatory cardiovascular responses to submaximal
upper- and lower-body exercise (e.g., increased heart rate, cardiac contractility, cardiac output) occur after
CO exposures.  These  changes were highly  significant for  exposures attaining 20% COHb.  Other
compensatory responses are increased coronary blood flow, cerebral blood flow, Hb (through increased
hemopoiesis), and oxygen consumption in muscle.
       Short-term compensatory responses in blood flow or oxygen consumption may not be complete or
may even be absent in certain persons. For example, from the laboratory animal  studies, it is known that
coronary blood flow is increased with COHb, and, from human clinical studies, it is known  that subjects
with IHD respond to the lowest levels of COHb (6% or less).  The implication is that, in some cases of
cardiac impairment, the short-term compensatory mechanism is impaired.
       From neurobehavorial studies (see  Section 6.3.2 of the present document), it  is apparent that
decrements resulting from CO exposure have not been consistent in all subjects, even in the same studies,
and have not  demonstrated a dose-response relationship with increasing COHb  levels.   The implication
from these data suggests there may be some threshold or time lag in a compensatory mechanism such as
increased blood flow. Without direct physiological evidence in either laboratory  animals or  humans, this
concept can only be hypothesized.
       The mechanism by which long-term adaptation may occur, if it can be demonstrated in humans, is
assumed to be increased Hb concentration via an increase in hemopoiesis. This alteration in Hb production
has been demonstrated repeatedly in laboratory animal studies, but no recent studies have been conducted
that indicate  the occurrence of some  adaptational benefit.  Even if the  Hb increase is a  signature of
adaptation, it has not been demonstrated at low ambient concentrations of CO.
                                            6-46

-------
6.8  Combined Exposure of Carbon Monoxide with Other Pollutants, Drugs, and
     Environmental Factors
6.8.1  High-Altitude Effects
       Although there are many studies comparing and contrasting the effects of inhaling CO with those
produced by short-term, high-altitude exposure, there are relatively few reports on the combined effects of
inhaling CO at high altitudes. There are data (reviewed in Section 11.1 of U.S. Environmental Protection
Agency, 1991) to support the possibility that the effects of these two hypoxic factor episodes are at least
additive.  Most of these early data were obtained at CO concentrations too high to have much meaning for
regulating the amount of CO in ambient air.  More recent studies by Kleinman et al. (1998) evaluated the
combined effects of lower levels of CO at high altitude.  In general, the results confirm the additivity of
hypoxic effects at a simulated altitude of 2.1 km  and CO exposures resulting in 4% COHb.
       There are even fewer studies of the long-term effects of CO at high altitude.  These studies,
identified in Table 11-2 of the previous criteria document (U.S. Environmental Protection Agency, 1991),
indicate few changes at CO concentrations below 100 ppm and altitudes below 4,572 m (15,000 ft). The
fetus may be particularly sensitive to the effects of CO at altitude (Longo, 1977), as is especially true with
the high levels of CO associated with maternal smoking (Moore et al., 1982).
       The potential effects  on  human  health of inhaling CO at  high altitudes  are complex (see
Section 5.4.1) Whenever CO binds to Hb, it reduces the amount of Hb available to carry oxygen. People
visiting high altitudes (where the partial pressure of oxygen in the atmosphere is lower) will experience
reduced levels of oxygen in the blood (hypoxemia) because of a relative hypoventilation that occurs,
particularly during sleep. Carbon monoxide, by binding to Hb, intensifies the hypoxemia existing at high
altitudes by further reducing transport of oxygen to the tissues. In addition, COHb saturations are higher
at altitude than at sea level because, in part, of changes in elimination of endogenous CO and of more rapid
uptake of exogenous CO (McGrath, 1992; McGrath et al.,  1991,  1993). However, within hours of arrival
at high altitude, certain physiological adj ustments begin to take place (Grover et al., 1986), and, over several
days,   these  mechanisms  will  operate  to  lessen  the  initial  impact  of  atmospheric  hypoxia.
Hemoconcentration occurs, and the increased  Hb  concentration offsets  the decreased blood oxygen
saturation and restores oxygen concentrations to  former levels.  Consequently, the simple additive model
of COHb and altitude hypoxemia may be valid only during early altitude  exposure.  The new visitor to
higher altitudes, especially the elderly and those with CAD (Kleinman et al., 1998; Leaf and Kleinman,
1996a), may be at greater risk from the added effects of ambient CO than the adapted resident. The period
of increased risk probably is prolonged in the elderly because adaptation to high altitude proceeds more
slowly with increasing age (Dill et al., 1985).

6.8.2  Interaction with  Drugs
       There remains little direct information on the possible enhancement of CO toxicity by concomitant
drug use or abuse; however, there are some data suggesting cause for concern. There is some evidence that
interactions of drug effects with CO exposure can occur in both directions, that is, CO toxicity may be
enhanced by drug use, and the toxic or other effects of drugs may be altered by CO exposure. Nearly all
published data available on CO combinations with drugs concern psychoactive drugs (Montgomery and
Rubin, 1971,1973; McMillan and Miller, 1974; Medical College of Wisconsin, 1974;Pankowetal, 1974;
Rockwell and Weir, 1975; Roth and Rubin, 1976a,b,c; Mitchell etal, 1978; Topping etal, 1981;Kimand
Carlson, 1983; Engen, 1986; Knisely et al., 1987, 1989). Descriptions of these studies were provided in
Section 11.2 of the previous criteria document (U.S. Environmental Protection Agency,  1991).  The
following summary, excerpted from the last review, still applies because nothing significant has appeared
in the recently published literature.
       The use and abuse of psychoactive drugs  and alcohol are widespread. Because of the effect  of CO
on brain function, interactions between CO and psychoactive drugs could be anticipated. However, very

                                            6-47

-------
little systematic research has addressed this question. In addition, very little of the research that has been
done has utilized models for expected effects from treatment combinations. Thus, often it is not possible
to assess whether the combined effects of drugs and CO exposure are additive or differ from additivity. It
is important to recognize that even additive effects of combinations  can  be  of clinical significance,
especially when the individual is unaware of the combined hazard. The greatest evidence for a potentially
important interaction of CO comes from studies with alcohol in both laboratory animals and humans, where
at least additive effects  have  been obtained (Mitchell et al.,  1978; Knisely et al., 1987, 1989).  The
significance of these effects is augmented by the probable high incidence of combined alcohol use and CO
exposure in the population.
       Besides interaction with psychoactive drugs, there is growing concern that prescribed medications,
especially nitric oxide blockers and calcium channel blockers, could interact with CO. There are no known
published data available, however, on CO combinations with these drugs.

6.8.3  Interaction with Other Air Pollutants and Environmental Factors
       Much of the data concerning the combined effects of CO and other pollutants found in ambient air
are based on laboratory animal experiments that were discussed in Section  11.3 of the previous criteria
document (U.S. Environmental Protection Agency, 1991).  More recent studies published since then have
confirmed the conclusions made  at that time and are included here  for completeness.  Only a few
controlled-exposure studies of humans are available, and the results were discussed in more detail in the
previous document. These early studies in healthy human subjects (Drinkwater et al., 1974; Raven et al.,
1974a,b; Gliner et al., 1975; Hackney et al., 1975a,b; DeLucia et al., 1983) on relevant concentrations of
common air pollutants such as CO, NO2, O3,  and peroxyacetylnitrate failed to show any interaction from
combined exposure. The more recent epidemiology studies (e.g., Morris et al., 1995; Schwartz andMorris,
1995; Schwartz,  1997, 1999; Burnett et al.,  1997a,b; Morris and Naumova, 1998; Burnett et al., 1999)
suggest an association between hospital  admissions for cardiovascular disease and ambient exposure to
multiple pollutants, including CO and PM.  In animal studies, no interaction  was observed following
combined exposure of CO and common air pollutants such as NO2 and SO2 (Busey, 1972; Murray et al.,
1978; Hugod, 1979). However, an additive effect on learning behavior was observed following combined
exposure of high levels (>100 ppm) of CO and NO (Groll-Knapp et al., 1988), and a synergistic dose effect
(increased COHb) was observed after combined exposure to CO and O3 (Murphy, 1964).
       Toxicological interactions of combustion products, primarily CO, carbon dioxide (CO2), NO2, and
hydrogen cyanide (HCN), at levels typically produced by indoor and outdoor fires, have shown a synergistic
effect on mortality following CO plus CO2 exposure (Rodkey and Collison, 1979; Levin et al.,  1987a) and
CO plus NO2 exposure (Levin, 1996) and an additive effect with HCN (Levin  et al.,  1987b). Additive
effects on mortality also were observed when CO, HCN, and low oxygen were combined; adding CO2 to
this combination was synergistic (Levin et al., 1988).
       Finally, laboratory animal studies (Young etal, 1987; Yang etal, 1988;Fechteretal., 1988,1997;
Fechter, 1995; Gary et al., 1997) suggest that combinations of environmental factors such as heat stress and
noise may be important determinants of health effects occurring in combination with CO exposure.  Of the
effects described, one potentially most relevant to typical human exposures is a greater decrement in the
exercise performance seen when heat stress is combined with 50 ppm CO (Drinkwater et al., 1974; Raven
et al., 1974a,b; Glmer et al., 1975).
       All of the studies discussed above involve interactions with exogenous exposure to CO. There are
endogenous sources of CO (e.g., heme degradation, peroxidative degradation of unsaturated  fatty acids,
xenobiotic metabolism) that also can lead to increased COHb saturation. These are discussed in more detail
in Section 5.3 of this document.  Possibly one of the greatest concerns regarding potential risk in the
population comes from inhalation exposure  to the halogenated hydrocarbons widely used as solvents,
especially  the   dihalomethanes   (e.g.,  methylene  chloride [dichloromethane],   dibromomethane,

                                             6-48

-------
diiodomethane, and bromochloromethane). There is some indication from the older literature (Fodor and
Roscovanu, 1976) that oral exposure to trihalogenated methane derivatives also will lead to increased
COHb. Other volatile solvents (e.g., carbon tetrachloride, chloroform, methanol) were tested in laboratory
animals, but none produced increased levels of COHb (Pankow, 1996).
       Methylene chloride provides the greatest potential  exposure to the population because it has been
used widely as a paint remover, degreaser,  and aerosol propellant (Wilcosky and Simonsen, 1991). When
inhaled, it will undergo metabolic breakdown by cytochrome P-450 in liver to form CO, chloride, and CO2.
Increased levels of CO from metabolic breakdown of exogenous chemicals will increase COHb measured
in the blood and add to the increased COHb levels resulting from exogenous CO exposure (DiVincenzo and
Kaplan, 1981a,b; Kurppa et al,  1981). The metabolism to CO  can be saturated,  leading to a slower
elimination of COHb than after CO exposure (Pankow, 1996). In addition, any co-exposures to other
chemicals or drugs that affect cytochrome P-450 also will  affect COHb saturation (Kim and Kim, 1996;
Wirkner et al., 1997).

6.8.4  Tobacco Smoke
       Although tobacco smoke is another source of CO for smokers as well as nonsmokers, it is also a
source of other chemicals (e.g.,  nicotine, NO2, HCN, polyaromatic  hydrocarbons  [PAHs], aldehydes,
ketones) that could interact with environmental CO. Available data suggest that some of these components
can affect the cardiovascular system.  For example, nicotine clearly aggravates the decrease in oxygen
capacity induced by CO through an increase in the oxygen demand of the heart (Khosla et al., 1994;
Benowitz, 1997), and PAHs have been implicated in atherosclerosis (Glantz and Parmley, 1991). Little is
known, however, about the relative importance of CO compared with the other components of tobacco
smoke.
       The association between active smoking and CVD is fully established (Surgeon General of the
United States, 1983). Passive smoking exposes an individual to all components in the cigarette smoke, but
the CO component dominates heavily because  only  1% or less of the nicotine is  absorbed from
environmental tobacco smoke (ETS), compared with 100% in an active smoker (Wall et al., 1988; Jarvis,
1987). Therefore, passive smoking will be closer to pure  CO exposure than active smoking, even if the
resultant levels of COHb are low (about 1 to 2%) (Jarvis, 1987).  The relationship between passive smoking
and increased risk of CVD is controversial. Early studies on this relationship were reviewed in the 1986
report of the Surgeon General of the United States (1986)  and by the National Research Council (1986).
Since that time, the epidemiological evidence  linking passive smoking exposure to heart disease has
expanded rapidly. The available literature on the relationship between passive exposure to ETS in the home
and the risk of cardiovascular-associated morbidity or mortality in the nonsmoking spouse of a smoker
consists of numerous published reports (e.g., Glantz and Parmley, 1991; Steenland, 1992; Wells, 1994;Kritz
et al., 1995; LeVois and Layard, 1995; Steenland et al., 1996; Kawachi et al., 1997; Howard et al., 1998;
He et al., 1999).  The data suggest that  nonsmokers exposed to ETS  had a relative risk of CVD of
approximately 1.3 (95% CI of 1.2 to 1.4). The association of CVD with prolonged exposure to ETS could
be caused by any number of biochemical mechanisms, including greater platelet aggregation, endothelial
cell damage, reduced oxygen supply, greater oxygen demand, and the direct effects of CO (Kalmaz et al.,
1993; Zhu and Parmley, 1995; Weiss, 1996; Werner and Pearson, 1998). Unfortunately, given the size of
this association (25 to 30%) compared to active smoking (=75%) and the inherent problems with the
studies, it is still not known, with accuracy,  how much, or even whether, exposure to ETS increases the risk
ofCVD(Bailar, 1999).
                                            6-49

-------
6.9 Summary
       The effects of exposure to low CO concentrations, such as the levels found in ambient air, are far
more subtle and considerably less threatening than those occurring in frank poisoning from high CO
concentrations.  Because the COHb level of the blood  is the best indicator of potential health risk,
symptoms of exposures to excessive ambient air levels of CO are described here in terms of associated
COHb levels.   The LOEL,  however,  depends on the method used for analysis  of COHb.   Gas
chromatography (GC) is the method of choice for measuring COHb, particularly at saturation levels <5%,
because of the large variability and potential high bias of the optical methods such as CO-Ox.
       The key human health effects most clearly demonstrated by controlled exposure studies to be
associated with the low COHb levels (< 5%) that are expected to occur from exposure to ambient CO are
summarized in Table 6-12. As shown in that table, maximal exercise duration and performance in healthy
individuals have been shown to be reduced at COHb levels of >2.3% and >4.3% (GC), respectively. The
decrements in performance at these levels are small and likely to affect only competing athletes rather than
people engaged in everyday activities. In fact, no effects were observed during submaximal exercise in
healthy individuals at COHb levels as high as 15 to 20%.
 Table 6-12.  Key Health Effects of Carbon Monoxide Demonstrated by Controlled-Exposure Studies

                        Health Effectsa'b                  Tested Population0           References
Target
Organ
 Lungs    Reduced maximal exercise duration with 1-h peak
          CO exposures resulting in >2.3% COHb (GC)
 Heart    Reduced time to ST segment change of the ECG
          (earlier onset of myocardial ischemia) with peak
          CO exposures resulting in >2.4% COHb (GC)

 Heart    Reduced exercise duration because of increased
          chest pain (angina) with peak CO exposures
          resulting in >3% COHb (CO-Ox)
 Heart    Increased number and complexity of arrhythmias
          (abnormal heart rhythm) with peak CO exposures
          resulting in >6% COHb (CO-Ox)
 Brain    Central nervous system effects, such as decrements
          in hand-eye coordination (driving or tracking) and
          in attention or vigilance (detection of infrequent
          events), with 1-h peak CO exposures (~5 to 20%
	COHb)	
                                                    Healthy individuals
                                                     Individuals with
                                                     coronary artery
                                                     disease (CAD)

                                                     Individuals with
                                                     CAD
                                                     Individuals with
                                                     CAD and high
                                                     baseline ectopy
                                                     (chronic arrhythmia)

                                                     Healthy individuals
Drinkwater et al. (1974)
Raven etal. (1974b)
Horvathetal. (1975)

Allredetal. (1989a,b; 1991)
Anderson etal. (1973)
Sheps etal. (1987)
Adams et al. (1988)
Kleinman et al. (1989, 1998)
Allredetal. (1989a,b; 1991)

Sheps etal. (1990)
Horvathetal. (1971)
Fodor and Winneke (1972)
Putz etal. (1976, 1979)
Benignus etal. (1987)
 aThe EPA has set significant harm levels of 50 ppm (8-h average), 75 ppm (4-h average), and 125 ppm (1-h average).
  Exposure under these conditions could result in COHb levels of 5 to 10% and cause significant health effects in
  sensitive individuals.
 bMeasured blood COHb level after CO exposure.
 Tetuses, infants, pregnant women, elderly people, and people with anemia or with a history of cardiac or respiratory
  disease may be particularly sensitive to CO.
                                               6-50

-------
       Adverse effects have been observed in individuals with CAD at 3 to 6% COHb by optical methods
of measurement. At these levels, individuals with reproducible exercise-induced angina (chest pain) are
likely to experience a reduced capacity to exercise because of decreased time to onset of angina. The
indicators of myocardial ischemia during exercise, which are detectable by electrocardiographic (ECG)
changes (ST depression) and associated angina, were statistically significant in one study at >2.4% COHb
(GC) and  showed a dose-response relationship with increasing COHb.  An increase in the number and
complexity of exercise-related arrhythmias also has been observed at >6% COHb (CO-Ox) in some people
with CAD and high levels of baseline ectopy (a chronic arrhythmia) that may present an increased risk of
sudden death.
       Central nervous system effects, including reductions in hand-eye coordination (driving or tracking)
and in attention or vigilance, have been reported at peak COHb levels of 5% and higher, but later work
indicates that significant behavioral impairments in healthy individuals should not be expected until COHb
levels exceed 20%. It must be emphasized, however, that even a 5% COHb level is associated with 1 -h CO
concentrations of 100 ppm  or higher.  Thus, at typical ambient air levels of CO, no observable central
nervous system effects would be expected to occur in the healthy population.
       In addition to the controlled-exposure study findings highlighted in Table 6-12 for key human health
effects most clearly demonstrated to be associated with COHb levels expected to occur from ambient or
near-ambient level CO exposures, certain newly emerging community epidemiology study results warrant
note here,  as well. In time series epidemiologic studies, investigators have reported statistically significant
associations of short-term ambient CO concentrations with daily frequencies of heart disease exacerbation
in the elderly.  Such associations have been observed most consistently during cold weather.  Also, some
investigators have observed associations of short-term ambient CO levels with daily mortality counts and
with daily frequencies of respiratory illness. Results from experimental laboratory studies of high level
exposures to CO, and occupational  reports of increased morbidity and mortality in workers routinely
exposed to combustion products, raise the possibility of community-level associations of ambient CO with
harmful health outcomes.  However, the available epidemiologic database must be considered inconclusive
as to whether the reported statistical associations reflect actual and specific health effects of ambient CO
exposure.  As discussed above, short-term fluctuations in ambient CO at current U. S. levels would produce
only very  small changes in  COHb levels. The rather low (typically <5.0 ppm) average daily maximum
ambient CO levels evaluated in the epidemiologic analyses would be projected to increase COHB levels
by barely  detectable amounts with 1-h exposures and even  8-h exposures to 10 ppm CO with light to
moderate exercise would produce only a 1.0% to 2.0% increase in COHb over the baseline level of about
0.5%. On pathophysiologic grounds, it remains difficult to reconcile such small expected changes in COHB
concentrations with statistically detectable exacerbation of preexisting illness or with increased mortality
at the community level.
       Also, the epidemiologic database itself exhibits some degree of internal inconsistency. Specifically,
the available studies suggest a considerably stronger association of ambient CO with cardiac morbidity than
with respiratory morbidity.  In view of this, a stronger association of ambient CO with cardiovascular
mortality than with respiratory mortality might well be expected. In fact, however, the available database
shows no such preferential association with cardiovascular mortality. Indeed, associations of ambient CO
with respiratory mortality have been observed at least as frequently as with cardiovascular mortality.
Further research is needed on short- and long-term exposure to ambient CO and other combustion-related
air pollutants, and on the relative influence of exposure to pollutants from nonambient sources. In future
epidemiologic studies of CO and other  air pollutants, cause-specific relationships  of pollution with
morbidity and mortality should be characterized more thoroughly, and the consistency of findings on
morbidity with findings on mortality should be critically assessed.
       The current ambient air quality standards for CO (9 ppm for 8 h and 35 ppm for 1 h) are intended
to keep COHb levels below 2.1% to protect the most sensitive members of the general population (i.e.,

                                             6-51

-------
individuals with CAD). Individuals in motor vehicles are at the greatest risk from ambient CO exposure,
followed by pedestrians, bicyclists, and joggers in the proximity of roadways and the rest of the general
urban population exposed to vehicle exhaust.  Several hours of exposure to peak ambient CO concentrations
found occasionally at downtown urban sites during periods of heavy traffic would be required to produce
COHb levels of concern in the most sensitive nonsmokers. Carbon monoxide levels occurring outside the
downtown urban locations are expected to be lower and are probably more representative of levels found
in residential areas where most people live. Significant health effects from ambient CO exposure are not
likely under these latter exposure conditions.  Active cigarette smoking increases the risk for developing
cardiovascular and pulmonary disease, and passive smoking also can elevate COHb levels in nonsmokers
under conditions of poor ventilation, increasing risks for nonsmoking co-workers and family members.
Carbon monoxide poisoning from indoor exposures to higher than  ambient CO levels occurs frequently,
has more severe consequences, and often is overlooked.
References
Abbatiello, E. R.; Mohrmann, K. (1979) Effects on the offspring of chronic low exposure carbon monoxide during mice
     pregnancy. Clin. Toxicol. 14: 401-406.
Adams, K. F.; Koch, G.; Chatterjee, B.; Goldstein, G. M; O'Neil, J. I; Bromberg, P. A.; Sheps, D. S.; McAllister, S.;Price,
     C. J.; Bissette,  J. (1988) Acute elevation of blood carboxyhemoglobin to 6% impairs exercise performance and
     aggravates symptoms in patients with ischemic heart disease. J. Am. Coll. Cardiol. 12: 900-909.
Alderman, B. W.; Baron, A. E.; Savitz, D. A. (1987) Maternal exposure to neighborhood carbon monoxide and risk of low
     infant birth weight. Public Health Rep. 102: 410-414.
Allred, E. N.; Bleecker, E. R.; Chaitman, B. R.; Dahms, T. E.; Gottlieb, S. O.; Hackney, J. D.; Pagano, M.; Selvester, R. H.;
     Walden, S. M.;  Warren, J. (1989a) Short-term effects of carbon monoxide exposure on the exercise performance of
     subjects with coronary artery disease. N. Engl.  J. Med. 321: 1426-1432.
Allred, E. N.; Bleecker, E. R.; Chaitman, B. R.; Dahms, T. E.; Gottlieb, S. O.; Hackney, J. D.; Hayes, D.; Pagano, M.;
     Selvester, R. H.; Walden, S. M.; Warren, J. (1989b) Acute effects of carbon monoxide exposure on individuals with
     coronary artery disease. Cambridge, MA: Health Effects Institute; research report no. 25.
Allred, E. N.; Bleecker, E. R.; Chaitman, B. R.; Dahms, T. E.; Gottlieb, S. O.; Hackney, J. D.; Pagano, M.; Selvester, R. H.;
     Walden, S. M.; Warren, J. (1991) Effects of carbon monoxide on myocardial ischemia. Environ. Health Perspect.
     91:89-132.
American Heart Association. (1997) 1998 heart and stroke statistical update. Dallas, TX: American Heart Association.
Anderson, E. W.; Andelman, R. J.; Strauch, J. M.;  Fortuin, N. J.; Knelson,  J. H. (1973) Effect of low-level carbon
     monoxide exposure on onset and duration of angina pectoris: a study in ten patients with ischemic heart disease. Ann.
     Intern. Med. 79: 46-50.
Annau, Z.; Fechter, L. D. (1994) The effects of prenatal exposure to carbon monoxide. In: Needleman, H. L.; Bellinger, D.,
     eds. Prenatal exposure to toxicants: developmental consequences. Baltimore, MD: Johns Hopkins University Press;
     pp. 249-267. (Annau, Z., ed. The Johns Hopkins series in environmental toxicology).
Ator, N.A.(1982) Modulation of the behavioral effects of carbon monoxide by reinforcement contingencies. Neurobehav.
     Toxicol. Teratol. 4: 51-61.
Ator, N. A.; Merigan, W. H., Jr.; Mclntire, R. W. (1976) The effects of brief exposures to carbon monoxide on temporally
     differentiated responding. Environ. Res. 12: 81-91.
Bailar, J. C., III. (1999) Passive smoking, coronary heart disease, and meta-analysis [letter]. N. Engl. J. Med. 340: 958-959.
Bascom, R.; Bromberg, P. A.; Costa, D. L.; Devlin, R.; Dockery, D. W.; Frampton, M. W.; Lambert, W.; Samet, J. M.;
     Speizer, F. E.; Utell, M. (1996) Health effects of outdoor air pollution [part 2]. Am. J. Respir. Crit. Care Med.
     153:477-498.
Bassan, M. M. (1990) Sudden cardiac death [letter to the editor]. N. Engl. J. Med. 322: 272.
Benignus, V. A. (1993) Importance of experimenter-blind procedure inneurotoxicology. Neurotoxicol. Teratol. 15:45-49.
Benignus, V. A. (1994) Behavioral effects of carbon monoxide: meta analyses and extrapolations. J. Appl. Physiol.
     76: 1310-1316.
Benignus, V. A.  (1996) Behavioral effects of carbon monoxide exposure: results and mechanisms. In: Penney, D. G., ed.
     Carbon monoxide. Boca Raton, FL: CRC Press; pp. 211-238.
                                                 6-52

-------
Benignus, V. A.; Annau, Z. (1994) Carboxy hemoglobin formation due to carbon monoxide exposure in rats. Toxicol. Appl.
      Pharmacol. 128: 151-157.
Benignus, V. A.; Otto, D. A.; Prah, J. D.; Benignus, G. (1977) Lack of effects of carbon monoxide on human vigilance.
      Percept. Mot. Skills 45:  1007-1014.
Benignus, V. A.; Muller, K. E.; Barton, C. N.; Prah, J. D. (1987) Effect of low level carbon monoxide on compensatory
      tracking and event monitoring. Neurotoxicol. Teratol. 9: 227-234.
Benignus, V. A.; Muller, K. E.;  Malott, C. M. (1990) Dose-effects functions for carboxyhemoglobin and behavior.
      Neurotoxicol. Teratol. 12: 111-118.
Benowitz, N. L. (1997) The role of nicotine in smoking-related cardiovascular disease. Preventive Med. 26: 412-417.
Blair, P. S.; Fleming, P. J.; Bensley, D.; Smith, L; Bacon, C.; Taylor, E.; Berry, J.; Golding, J.; Tripp, J. (1996) Smoking
      and the sudden infant death syndrome: results from 1993-5 case-control study for confidential inquiry into stillbirths
      and deaths in infancy. Br. Med. J. 313: 195-198.
Brooke, O. G.; Anderson, H. R.; Bland, J. M.; Peacock, J. L.; Stewart, C. M. (1989) Effects on birth weight of smoking,
      alcohol, caffeine, socioeconimic factors, and psychosocial stress. Br. Med. J. 298: 795-801.
Burnett, R. T.; Dales, R.; Krewski, D.; Vincent, R.; Dann, T.; Brook, J. R. (1995) Associations between ambient paniculate
      sulfate and admissions to Ontario hospitals for cardiac and respiratory diseases. Am. J. Epidemiol. 142: 15-22.
Burnett, R. T.; Dales, R. E.; Brook, J. R.; Raizenne, M. E.;  Krewski, D. (1997a) Association between ambient carbon
      monoxide levels and hospitalizations for congestive heart failure in the elderly in 10 Canadian cities. Epidemiology
      8: 162-167.
Burnett, R. T.; Cakmak, S.; Brook, J. R.; Krewski, D. (1997b) The role of paniculate size and chemistry in the association
      between summertime ambient air pollution and hospitalization for cardiorespiratory  diseases. Environ. Health
      Perspect.  105: 614-620.
Burnett, R. T.; Cakmak, S.; Raizenne, M. E.; Stieb, D.; Vincent, R.; Krewski, D.; Brook, J. R.; Philips, O.; Ozkaynak, H.
      (1998a) The association between ambient carbon monoxide levels and daily mortality in Toronto, Canada. J. Air
      Waste Manage. Assoc. 48: 689-700.
Burnett, R. T.; Cakmak, S.; Brook, J. R. (1998b) The effect of the urban ambient air pollution mix on daily mortality rates
      in 11 Canadian cities. Can. J. Public Health 89: 152-156.
Burnett, R. T.; Smith-Doiron, M.; Stieb, D.; Cakmak, S.; Brook, J. R. (1999) Effects of paniculate and gaseous air pollution
      on cardiorespiratory hospitalizations. Arch. Environ. Health 54: 130-139.
Burns, T. R.; Greenberg, S. D.; Cartwright, J.; Jachimczyk,  J. A.  (1986) Smoke inhalation: an ultrastructural study of
      reaction to injury in the human alveolar wall. Environ.  Res. 41: 447-457.
Busey, W. M. (1972) Chronic exposure of albino rats to certain airborne pollutants. New York, NY: Coordinating Research
      Council, Inc.; report no. CRC-APRAC-CAPM-6-68-6.
Carratu, M. R.; Renna, G.; Giustino, A.; De Salvia, M. A.; Cuomo, V. (1993) Changes in peripheral nervous system activity
      produced in rats by prenatal exposure to carbon monoxide. Arch. Toxicol. 67: 297-301.
Carratu, M. R.;  Cagiano, R.; De Salvia, M. A.; Trabace,  L.;  Cuomo, V. (1995) Developmental neurotoxicity of carbon
      monoxide. In: Degen, G. H.; Seiler,  J. P.; Bentley, P., eds. Toxicology in transition: proceedings of the  1994
      EUROTOX congress; August 1994; Basel, Switzerland. Arch. Toxicol. Suppl. 17: 295-301.
Carratu, M. R.; Cagiano,R.; Cortese, L; Renna, G.; SiroBrigiani, G.; Trabace, L.; Cuomo, V. (1996)Myelinationdisorders
      produced by prenatal exposure to moderate carbon monoxide  concentrations in  rat  offspring. Fundam.  Clin.
      Pharmacol. 10:  172.
Gary, R.; Clarke, S.; Delic, J. (1997) Effects of combined exposure to noise and toxic substances—critical review of the
      literature. Ann. Occup. Hyg. 41: 455-465.
Chaitman, B. R.; Dahms, T. E.; Byers,  S.; Carroll, L. W.; Younis, L. T.; Wiens, R. D. (1992) Carbon monoxide exposure
      of subjects with documented cardiac arrhythmias. Cambridge, MA:  Health Effects Institute; research report no. 52.
Chen, S.; Weller, M. A.; Penney, D. G. (1982) A study of free lung cells  from young rats chronically exposed to carbon
      monoxide from birth. Scanning Electron Microsc. 2: 859-867.
Cohen, S. L; Deane, M.; Goldsmith, J. R. (1969) Carbon monoxide and survival frommyocardial infarction. Arch. Environ.
      Health 19: 510-517.
Cooper, K. R.; Alberti, R. R. (1984) Effect of kerosene heater emissions on indoor air quality and pulmonary function.
      Am. Rev. Respir. Dis. 129:  629-631.
Crocker, P. J.; Walker, J. S. (1985) Pediatric carbon monoxide toxicity. J. Emerg. Med. 3: 443-448.
Dahms, T. E.; Younis, L. T.; Wiens, R. D.; Zarnegar, S.; Byers, S. L.; Chaitman, B. R. (1993) Effects of carbon monoxide
      exposure  in patients with documented cardiac arrhythmias. J. Am. Coll. Cardiol. 21: 442-450.
                                                   6-53

-------
Davies, D. M; Smith, D. J. (1980) Electrocardiographic changes in healthy men during continuous low-level carbon
      monoxide exposure. Environ. Res. 21: 197-206.
Davies, R. F.; Topping, D. L.; Turner, D. M. (1976) The effect of intermittent carbon monoxide exposure on experimental
      atherosclerosis in the rabbit. Atherosclerosis 24: 527-536.
De Luca, A.; Pierno, S.; Tricarico, D.; Carratu, M. R.; Cagiano, R.; Cuomo, V.; Camerino, D. C. (1996) Developmental
      changes of membrane electrical properties of rat skeletal muscle fibers produced by prenatal exposure to carbon
      monoxide. Environ. Toxicol. Pharmacol. 2:  213-221.
De Salvia, M. A.; Cagiano, R.; Carratu, M. R.; Di Giovanni, V.; Trabace, L.; Cuomo, V. (1995) Irreversible impairment
      of  active  avoidance behavior in  rats prenatally  exposed to  mild  concentrations  of carbon  monoxide.
      Psychopharmacology (Berlin) 122: 66-71.
DeLucia, A. J.; Whitaker, J. H.; Bryant, L. R. (1983) Effects of combined exposure to ozone and carbon monoxide (CO)
      in humans. In: Lee, S. D.; Mustafa, M. G.; Mehlman, M. A., eds. International symposium on the biomedical effects
      of ozone and related photochemical oxidants; March; Pinehurst, NC. Princeton, NJ: Princeton Scientific Publishers,
      Inc.; pp. 145-159. (Advances in modern environmental toxicology: v. 5).
Di Giovanni, V.; Cagiano, R.; De Salvia, M. A.; Giustino, A.; Lacomba, C.; Renna, G.; Cuomo, V. (1993) Neurobehavioral
      changes produced in rats by prenatal exposure to carbon monoxide. Brain Res. 616: 126-131.
DiVincenzo, G. D.; Kaplan, C. J. (1981a) Effect of exercise or smoking on the uptake, metabolism, and excretion of
      methylene chloride vapor. Toxicol. Appl. Pharmacol. 59: 141-148.
DiVincenzo, G. D.; Kaplan, C. J. (1981b) Uptake, metabolism, and elimination of methylene chloride vapor by humans.
      Toxicol. Appl. Pharmacol. 59: 130-140.
Dill, D. B.; Alexander, W. C.; Myhre, L.  G.; Whinnery, J. E.; Tucker,  D.  M.  (1985) Aerobic capacity of D.B. Dill,
      1928-1984. Fed. Proc. 44:  1013.
Doblar, D. D.; Santiago, T. V.; Edelman, N. H. (1977) Correlation between ventilatory and cerebrovascular responses to
      inhalation of CO. J. Appl. Physiol.: Respir. Environ. Exercise Physiol. 43: 455-462.
Doi, R.; Tanaka, H. (1984) Effects of carbon monoxide exposure on the distribution of methylmercury in mice. J. Toxicol.
      Sci. 9: 11-22.
Douglas,  A. S.; Allan, T. M.; Helms, P. J. (1996)  Seasonality and the sudden infant death syndrome during 1987-9 and
      1991-3 in Australia and Britain. Br. Med. J. 312: 1381-1383.
Drinkwater, B. L.; Raven, P. B.; Horvath, S. M.; Gliner, J. A.;  Ruhling, R. O.; Bolduan, N. W.; Taguchi, S. (1974)
      Air pollution, exercise, and heat stress. Arch. Environ. Health 28: 177-181.
Edling, C.; Axelson, O. (1984) Risk factors of coronary heart disease among personnel inabus company. Int. Arch. Occup.
      Environ. Health 54: 181-183.
Ekblom, B.; Huot, R. (1972) Response to submaximal and maximal exercise at different levels of carboxyhemoglobin. Acta
      Physiol. Scand. 86: 474-482.
Engen, T. (1986) The combined effect of carbon monoxide and alcohol on odor sensitivity. Environ. Int. 12: 207-210.
Evans, R. G.; Webb, K.; Homan, S.; Ayres, S. M. (1988) Cross-sectional and longitudinal changes in pulmonary function
      associated with automobile pollution among bridge and tunnel officers. Am. J. Ind. Med. 14: 25-36.
Fechter, L. D. (1995) Combined effects of noise and chemicals. Occup. Med. State of the Art Rev. 10: 609-621.
Fechter, L. D.; Young, J. S.; Carlisle, L. (1988) Potentiation of noise induced threshold shifts and hair cell loss by carbon
      monoxide. Hear. Res. 34: 39-47.
Fechter, L. D.; Liu, Y.; Pearce, T. A. (1997) Cochlear protection from carbon monoxide exposure by free radical blockers
      in the guinea pig. Toxicol.  Appl. Pharmacol. 142: 47-55.
Fein, A.;  Grossman, R. F.; Jones, J. G.; Hoeffel,  J.; McKay, D.  (1980) Carbon monoxide effect on alveolar epithelial
      permeability. Chest 78: 726-731.
Fisher, A. B.; Hyde, R. W.; Baue, A.  E.; Reif, J. S.; Kelly, D. F. (1969) Effect of carbon monoxide on function and
      structure of the lung. J. Appl. Physiol. 26: 4-12.
Fodor, G. G.; Roscovanu, A. (1976) Erhoehter Blut-CO-Gehalt bei Mensch und Tier  durch inkorperierte halogenierte
      Kohlenwasserstoffe [Increased blood CO content in humans and animals caused by incorporated halogenated
      hydrocarbons]. Zentralbl. Bakteriol. Parasitenkd. Infektionskrankh. Hyg. Abt. 1: Orig. Reihe B 162: 34-40.
Fodor, G. G.; Winneke, G. (1972) Effect of low CO concentrations on resistance to monotony and on psychomotor
      capacity. Staub Reinhalt. Luft 32: 46-54.
Franco, P.; Groswasser, J.; Sottiaux,  M.; Broadfield, E.; Kahn, A. (1996) Decreased cardiac  responses to auditory
      stimulation during prone sleep. Pediatrics 97: 174-178.
Glantz, S. A.; Parmley, W. W. (1991) Passive smoking and heart disease: epidemiology, physiology, and biochemistry.
      Circulation 83: 1-12.


                                                   6-54

-------
Gliner, J. A.; Raven, P. B.; Horvath, S. M; Drinkwater, B. L.; Sutton, J. C. (1975) Man's physiologic response to long-term
      work during thermal and pollutant stress. J. Appl. Physiol. 39: 628-632.
Goldsmith, J. R.; Landaw, S. A. (1968) Carbon monoxide and human health. Science (Washington, DC) 162: 1352-1359.
Gordian, M. E.; Ozkaynak, H.; Xue, J.; Morris, S. S.; Spengler, J. D. (1996) Paniculate air pollution and respiratory disease
      in Anchorage, Alaska. Environ. Health Perspect. 104: 290-297.
Graves, E. J.; Owings, M. F. (1998) 1996 summary: National Hospital Discharge Survey. Hyattsville, MD: National Center
      for Health Statistics; DHHS publication no.  (PHS) 98-1250. (Advance data from vital and health statistics; no. 301).
      Available: www.cdc.gov/nchs/data/ad301.pdf [2000, June 8].
Groll-Knapp, E.; Haider, M.; Jenkner, H.; Liebich, H.; Neuberger, M.; Trimmel, M. (1982) Moderate  carbon monoxide
      exposure during sleep:  neuro- and psychophysiological effects in young and elderly people. Neurobehav. Toxicol.
      Teratol. 4: 709-716.
Groll-Knapp, E.; Haider, M.; Kienzl, K.; Handler, A.; Trimmel, M. (1988) Changes in discrimination learning and brain
      activity (ERP's) due to  combined exposure to NO and CO in rats. Toxicology 49: 441-447.
Grover, R. F.; Weil, J. V.; Reeves, J. T.  (1986) Cardiovascular adaptation to exercise at high altitude. Exercise Sport Sci.
      Rev. 14: 269-302.
Hackney, J. D.; Linn, W.  S.;  Mohler, J. G.; Pedersen, E. E.; Breisacher, P.; Russo, A. (1975a) Experimental studies on
      human health effects of air pollutants: II. four-hour exposure to ozone alone and in combination with other pollutant
      gases. Arch. Environ. Health 30: 379-384.
Hackney, J.  D.; Linn, W. S.; Law, D. C.; Karuza,  S. K.; Greenberg,  H.; Buckley, R. D.; Pedersen, E. E.  (1975b)
      Experimental studies on human  health effects of air pollutants: III. two-hour exposure to ozone  alone and in
      combination with other pollutant gases. Arch. Environ. Health 30: 385-390.
Hagberg,  M.;  Kolmodin-Hedman, B.; Lindahl, R.; Nilsson, C.-A.;  Norstrom,  A.  (1985) Irritative  complaints,
      carboxyhemoglobin increase and minor ventilatory function changes due to exposure to chain-saw exhaust. Eur. J.
      Respir. Dis. 66: 240-247.
Hanley, D. F.; Wilson, D. A.; Traystman, R. J. (1986) Effect of hypoxia and hypercapnia on neurohypophyseal blood flow.
      Am. J. Physiol. 250: H7-H15.
Harbin, T. J.; Benignus, V. A.; Muller, K. E.; Barton, C. N. (1988) The effects of low-level carbon monoxide exposure
      upon evoked cortical potentials in young and elderly men. Neurotoxicol. Teratol. 10: 93-100.
He, J.; Vupputuri,  S.; Allen, K.; Prerost, M. R.; Hughes, J.; Whelton, P. K. (1999) Passive smoking and the risk of coronary
      heart disease—a meta-analysis of epidemiologic studies. N. Engl. J. Med. 340: 920-926.
Helfaer, M.  A.; Traystman,  R. J. (1996) Cerebrovascular effects of carbon monoxide.  In: Penney, D.  G., ed. Carbon
      monoxide. Boca Raton, FL: CRC Press; pp. 69-86.
Hexter, A. C.; Goldsmith, J. R. (1971) Carbon monoxide: association of community air pollution with mortality. Science
      (Washington, DC) 172: 265-267.
Hinderliter, A. L.;  Adams, K.  F., Jr.; Price, C. J.; Herbst, M. C.; Koch, G.; Sheps, D. S.  (1989) Effects of low-level carbon
      monoxide exposure on resting and exercise-induced ventricular arrhythmias in patients with coronary artery disease
      and no baseline ectopy. Arch. Environ. Health 44: 89-93.
Hoppenbrouwers,  T.; Calub, M.; Arakawa, K.; Hodgman, J. E. (1981) Seasonal relationship of sudden infant death
      syndrome and environmental pollutants. Am. J. Epidemiol. 113: 623-635.
Horvath, S. M. (1981) Impact of air quality in exercise performance. Exercise Sport Sci. Rev. 9: 265-296.
Horvath, S. M.; Dahms, T. E.; O'Hanlon, J. F. (1971) Carbon monoxide and human vigilance: a deleterious effect of present
      urban concentrations. Arch. Environ. Health 23: 343-347.
Horvath, S.  M.; Raven, P.  B.; Dahms, T. E.; Gray, D.  J.  (1975) Maximal aerobic capacity at different levels of
      carboxyhemoglobin. J.  Appl. Physiol. 38: 300-303.
Howard, G.; Wagenknecht, L. E.; Burke, G. L.; Diez-Roux, A.; Evans, G. W.; McGovern, P.; Nieto, J.;  Tell, G. S. (1998)
      Cigarette smoking and progression of atherosclerosis: the atherosclerosis risk in communities (ARIC) study. JAMA
      J. Am. Med. Assoc. 279: 119-124.
Hugod, C. (1979)  Effect of exposure to 0.5 ppm hydrogen cyanide singly or combined with 200 ppm  carbon monoxide
      and/or 5 ppm nitric oxide on coronary arteries, aorta, pulmonary artery, and lungs in the rabbit.  Int. Arch. Occup.
      Environ. Health 44: 13-23.
Hugod, C. (1980)  The effect of carbon monoxide exposure on morphology of lungs  and pulmonary arteries in rabbits:
      a light- and  electron-microscopic study. Arch. Toxicol. 43: 273-281.
Hutter, C. D. D.; Blair, M. E.  (1996) Carbon monoxide—does fetal exposure cause sudden infant death syndrome? Med.
      Hypotheses  46: 1-4.
Institute of Medicine. (1985)  Preventing low birthweight. Washington, DC: National Academy Press.


                                                   6-55

-------
Ito, K.; Thurston, G. D. (1996) Daily PM10/mortality associations: an investigation of at-risk subpopulations. J. Exposure
      Anal. Environ. Epidemiol. 6: 79-95.
Ito, K.; Kinney, P.; Thurston, G. D. (1995) Variations in PM-10 concentrations within two metropolitan areas and their
      implications for health effects analyses. In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the colloquium on
      paniculate air pollution and human mortality and morbidity, part II; January 1994; Irvine, CA. Inhalation Toxicol.
      7: 735-745.
Jarvis, M. J. (1987) Uptake of environmental tobacco smoke. In: O'Neill, I. K.; Brunnemann, K. D.; Dodet, B.; Hoffmann,
      D., eds. Environmental carcinogens: methods of analysis and exposure measurement, v. 9 - passive smoking. Lyon,
      France: International Agency for Research on Cancer; pp. 43-58. (IARC scientific publications no. 81).
Jedrychowski, W.; Flak, E.  (1996) Confronting the prenatal effects of active and passive tobacco smoking on the birth
      weight of children. Cent. Eur. J. Public Health 4: 201-205.
Jones, M. D.,  Jr.; Traystman, R. J. (1984) Cerebral oxygenation  of the fetus, newborn, and adult. Semin. Perinatol.
      8: 205-216.
Joseph, K. S.; Kramer, M. S. (1996) Review of the evidence on fetal and early childhood antecedents of adult chronic
      disease. Epidemiol. Rev. 18: 158-174.
Kahn, A.; Groswasser, J.; Sottiaux, M.; Rebuffat, E.; Franco, P.; Dramaix, M. (1993) Prone or supine body position and
      sleep characteristics in infants. Pediatrics 91: 1112-1115.
Kalmaz, E. E.; Saunders, P. J.; Kalmaz, G. D. (1993) Effects of carbon monoxide on arterial endothelial permeability and
      its possible relation to arterial disease. J. Occup. Med. Toxicol. 2: 31-39.
Kanten, W. E.; Penney, D. G.; Francisco, K.; Thill, J. E. (1983) Hemodynamic responses to acute carboxyhemoglobinemia
      in the rat. Am. J. Physiol. 244: H320-H327.
Katsumata, Y.; Aoki, M.; Oya, M.; Yada, S.; Suzuki, O. (1980) Liver damage in rats  during acute carbon monoxide
      poisoning. Forensic Sci. Int.  16: 119-123.
Kawachi, I.; Colditz, G. A.; Speizer,  F. E.; Manson, J. E.; Stampfer, M. J.; Willett,  W. C.; Hennekens,  C. H. (1997)
      A prospective study of passive smoking and coronary heart disease. Circulation  95: 2374-2379.
Kelsall, J. E.; Samet, J. M.; Zeger, S.  L.; Xu, J. (1997) Air pollution and mortality in Philadelphia,  1974-1988. Am. J.
      Epidemiol. 146:  750-762.
Kemp, J. S.; Livne,  M.; White, D. K.; Arfken, C. L. (1998) Softness and potential to cause  rebreathing: differences in
      bedding used by infants at high and low risk for sudden infant death syndrome. J. Pediatr. 132: 234-239.
Khosla, S.; Laddu, A.; Ehrenpreis, S.; Somberg, J. C. (1994) Cardiovascular effects of nicotine: relation to deleterious
      effects of cigarette smoking. Am. Heart J. 127: 1669-1672.
Kim, Y. C.; Carlson, G. P. (1983) Effect of carbon monoxide inhalation exposure in mice on drug metabolism in vivo.
      Toxicol. Lett.  19: 7-13.
Kim, S.K.;Kim, Y.  C. (1996) Effect of a single administration of benzene, toluene or /w-xylene on carboxyhaemoglobin
      elevation and metabolism of dichloromethane in rats. J. Appl. Toxicol. 16: 437-444.
Kinney, P. L.; Ozkaynak, H. (1991) Associations of daily mortality and air pollution in Los Angeles County. Environ. Res.
      54: 99-120.
Kinney, P. L.; Ito, K.; Thurston, G. D. (1995) A sensitivity analysis of mortality/PM10 associations in Los Angeles.
      In: Phalen, R.  F.; Bates, D. V., eds. Proceedings of the colloquium on paniculate air pollution and human mortality
      and morbidity; January 1994; Irvine, CA. Inhalation Toxicol. 7: 59-69.
Kizakevich, P. N.; Hazucha, M.; VanHoose, L.; McCartney, M. L.; Bolick, K.; Jochem, W. J.; Clayton, C. A.; Alexander,
      B. V.; Boehleche, B.; Hackney, A. J.; Sheps, D. S.; Clapp, L. (1994) Noninvasive ambulatory assessment of cardiac
      function and myocardial ischemia  in healthy subjects exposed to carbon monoxide. Frederick, MD:  U.S. Army
      Medical Research and Materiel Command; contract no. DAMD17-91-C-1007.
Klein, J. P.; Forster, H. V.; Stewart, R. D.; Wu, A. (1980) Hemoglobin affinity for oxygen during short-term exhaustive
      exercise. J. Appl. Physiol.: Respir. Environ. Exercise Physiol. 48: 236-242.
Kleinman, M.  T. (1992) Health effects of carbon monoxide. In: Lippmann, M.,  ed.  Environmental toxicants: human
      exposures and their health effects. New York, NY: Van Nostrand Reinhold; pp.  98-118.
Kleinman, M. T.; Leaf, D. (1991) Effects of exposure to low-level carbon monoxide at  high altitude in sensitive subjects.
      Sacramento, CA: California Air Resources Board.
Kleinman, M. T.; Davidson, D. M.; Vandagriff, R. B.; Caiozzo, V J.; Whittenberger,  J. L. (1989) Effects of short-term
      exposure to carbon monoxide in subjects with coronary artery disease. Arch. Environ. Health 44: 361-369.
Kleinman, M. T.;Leaf, D. A.; Kelly, E.; Caiozzo, V;Osann, K.;O'Niell, T. (1998) Urban angina in the mountains: effects
      of carbon monoxide and mild hypoxemia on subjects with chronic stable angina. Arch. Environ. Health 53: 388-397.
                                                    6-56

-------
Klonoff-Cohen, H. S.; Edelstein, S. L.; Lefkowitz, E. S.; Srinivasan, I. P.; Kaegi, D.; Chang, J. C.; Wiley, K. J. (1995) The
      effect of passive smoking and tobacco exposure throughbreast milk on sudden infant death syndrome. JAMA J. Am.
      Med. Assoc. 273:795-798.
Knisely, J. S.; Rees, D. C.; Salay, J. M; Balster, R. L.; Breen, T. J. (1987) Effects of intraperitoneal carbon monoxide on
      fixed-ratio and screen-test performance in the mouse. Neurotoxicol. Teratol. 9: 221-225.
Knisely, J. S.; Rees, D. C.; Balster, R. L. (1989) Effects of carbon monoxide in combination with behaviorally active drugs
      on fixed-ratio performance in the mouse. Neurotoxicol. Teratol. 11: 447-452.
Koike, A.; Wasserman, K. (1992) Effect of acute reduction in oxygen transport on parameters of aerobic function during
      exercise. Ann. Acad. Med. Singapore 21: 14-22.
Koike, A.; Wasserman, K.; Armon, Y.; Weiler-Ravell, D. (1991) The work-rate-dependent effect of carbon monoxide on
      ventilatory control during exercise. Respir. Physiol. 85: 169-183.
Kolarzyk, E. (1994a) The effect of acute carbon monoxide poisoning on the respiratory system efficiency. I. Values of
      spirometric parameters in different degrees of poisoning. Int. J. Occup. Med. Environ. Health 7: 225-235.
Kolarzyk, E. (1994b) The effect of acute carbon monoxide poisoning on the respiratory system efficiency. II. Types of
      ventilatory disorder and dynamics of changes according to the severity  of carbon monoxide poisoning. Int. J. Occup.
      Med. Environ. Health 7: 237-243.
Kolarzyk, E. (1995) Regulation of breathing in cases of acute carbon monoxide poisoning. Int. J. Occup. Med. Environ.
      Health 8:  89-101.
Koren, G.;  Sharav, T.; Pastuszak, A.; Garrettson, L. K.; Hill, K.;  Samson, L; Rorem, M.; King, A.; Dolgin, J. E. (1991)
      A multicenter, prospective study of fetal outcome following accidental carbon monoxide poisoning in pregnancy.
      Reprod. Toxicol. 5: 397-403.
Koskela, R.-S. (1994) Cardiovascular diseases among foundry workers exposed to carbon monoxide. Scand. J. Work
      Environ. Health 20: 286-293.
Kritz, H.; Schmid, P.; Sinzinger, H. (1995)  Passive smoking and cardiovascular risk. Arch. Intern. Med. 155: 1942-1948.
Kuller, L. H.; Radford, E. P.; Swift, D.; Perper, J. A.; Fisher, R. (1975) Carbon monoxide and heart attacks. Arch. Environ.
      Health 30: 477-482.
Kurppa, K.; Kivisto, H.; Vainio, H. (1981) Dichloromethane and carbon monoxide inhalation: carboxyhemoglobin addition,
      and drug metabolizing enzymes in rat. Int. Arch. Occup. Environ. Health 49: 83-87.
Kuska, J.; Kokot, F.; Wnuk, R. (1980) Acute renal failure after exposure to carbon monoxide. Mater. Med. Pol. (Engl. Ed.)
      12: 236-238.
Langston, P.; Gorman, D.; Runciman, W.; Upton, R. (1996) The effect of carbon monoxide on oxygen metabolism in the
      brains of awake sheep. Toxicology 114: 223-232.
LeVois, M. E.; Layard,  M. W. (1995) Publication bias in the environmental tobacco smoke/coronary heart  disease
      epidemiologic literature. Regul. Toxicol. Pharmacol. 21:  184-191.
Leaf, D. A.; Kleinman, M. T. (1996a) Urban ectopy in the mountains: carbon monoxide exposure at high altitude. Arch.
      Environ. Health 51: 283-290.
Leaf, D. A.; Kleinman, M. T. (1996b) Acute exposure to carbon monoxide does not affect plasma lipids, lipoproteins, and
      apolipoproteins. Angiology 47: 337-341.
Lebowitz, M. D.; Collins, L.; Holberg, C. J. (1987) Time series analyses of respiratory responses to indoor and outdoor
      environmental phenomena. Environ.  Res. 43: 332-341.
Levin, B. C. (1996) New research avenues in toxicology: 7-gas N-gas model, toxicant suppressants, and genetic toxicology.
      In: Chaturvedi, A. K.;  Sanders, D. C.;Higgins, E. A., eds. Proceedings of the international colloquium on advances
      in combustion toxicology; April 1995; Oklahoma City, OK. Toxicology 115: 89-106.
Levin, B. C.; Paabo, M.; Gurman, J. L.; Harris, S. E.; Braun, E. (1987a) Toxicological interactions between carbon
      monoxide and carbon dioxide. Presented at:  16th conference of toxicology, Air Force Medical Research Laboratory;
      October 1986; Dayton, OH. Toxicology 47:  135-164.
Levin, B.C.; Paabo, M.; Gurman, J. L.; Harris, S. E. (1987b) Effects of exposure to single or multiple combinations of the
      predominant toxic gases and low oxygen atmospheres produced in fires. Fundam. Appl. Toxicol. 9: 236-250.
Levin, B.C.; Paabo, M.; Gurman, J. L.; Clark, H. M.; Yoklavich, M. F. (1988) Further studies of the lexicological effects
      of different time exposures to the individual and combined fire gases—carbon monoxide, hydrogen cyanide, carbon
      dioxide and reduced  oxygen. In: Polyurethanes 88: proceedings of the SPI  31st annual technical/marketing
      conference; October; Philadelphia, PA. Society of the Plastics Industry, Inc.; pp. 249-252.
Longo, L.  D. (1976) Carbon monoxide:  effects on oxygenation of the  fetus in utero. Science (Washington, DC)
      194: 523-525.
                                                   6-57

-------
Longo, L. D. (1977) The biological effects of carbon monoxide on the pregnant woman, fetus, and newborn infant. Am. J.
      Obstet. Gynecol. 129: 69-103.
Luke, B. (1994) Maternal-fetal nutrition. Clin. Obstet. Gynecol. 37: 93-109.
Lutz, L. J. (1983) Health effects of air pollution measured by outpatient visits. J. Fam. Pract. 16: 307-313.
MacDorman, M. F.; Cnattingius, S.;Hoffman,H. J.; Kramer, M. S.;Haglund,B. (1997) Sudden infant death syndrome and
      smoking in the United States and Sweden. Am. J. Epidemiol. 146: 249-257.
Martin, T. R.; Bracken, M. B.  (1986) Association of low birth weight with passive smoke exposure in pregnancy. Am. J.
      Epidemiol. 124: 633-642.
May, D. S.; Kelly, J. J.; Mendlein, J. M.; Garbe, P. L. (1991) Surveillance of major causes of hospitalization among the
      elderly, 1988. Morb. Mortal. Wkly. Rep. 40: 7-21.
McGrath, J. J. (1982) Physiological effects of carbon monoxide. In: McGrath, J. J.; Barnes, C. D., eds. Air pollution—
      physiological effects. New York, NY: Academic Press; pp. 147-181.
McGrath,  J. J.  (1992)  Effects of altitude on endogenous carboxyhemoglobin levels. J. Toxicol.  Environ.  Health
      35: 127-133.
McGrath, J. J.; Lee, P.; Schreck, R. M. (1991) Effects of altitude on expired breath carbon monoxide levels. Presented at:
      84th annual meeting and exhibition of the Air & Waste Management Association; June; Vancouver, BC, Canada.
      Pittsburgh, PA: Air & Waste Management Association; paper no. 91-138.2.
McGrath, J. J.; Schreck, R. M.; Lee, P. S. (1993) Carboxyhemoglobin levels in humans: effects of altitude. Inhalation
      Toxicol. 5: 241-249.
McMillan, D. E.; Miller, A. T., Jr. (1974) Interactions between carbon monoxide and J-amphetamine or pentobarbital on
      schedule-controlled behavior. Environ. Res. 8: 53-63.
Medical College of Wisconsin. (1974) Exposure of humans to carbon monoxide combined with ingestion of ethyl alcohol
      and the comparison  of human performance when exposed  for varying periods of time  to carbon monoxide.
      Milwaukee, WI: Medical College of Wisconsin; report no. MCOW-ENVM-CO-74-2.
Melius, J. M. (1995) Cardiovascular disease among firefighters. Occup. Med.: State of the Art Rev. 10: 821-827.
Mennear, J. H. (1993) Carbon monoxide and cardiovascular disease: an analysis of the weight of evidence. Regul. Toxicol.
      Pharmacol. 17: 77-84.
Merigan, W. H.; Mclntire, R. W. (1976) Effects of carbon monoxide on responding under a progressive ratio schedule in
      rats. Physiol. Behav.  16: 407-412.
Michaels, D.; Zoloth, S. R.  (1991) Mortality among urban bus drivers.  Int. J. Epidemiol. 20:  399-404.
Min, H. Y.; Wei, Z.; Quan,  Q. M.; Rong, Z. (1992) Correlation between hemorheological parameters and viscoelasticity
      of thrombus in rats after exposure to carbon monoxide. Clin. Hemorheol. 12: 593-598.
Mitchell, D.  S.;  Packham, S. C.; Fitzgerald, W. E. (1978) Effects of ethanol and carbon monoxide on two measures of
      behavioral incapacitation of rats. Proc. West. Pharmacol. Soc. 21: 427-431.
Mitchell, E. A.; Ford, R. P. K.; Stewart, A. W.; Taylor, B. J.; Becroft, D. M. O.; Thompson, J. M. D.; Scragg, R.; Hassall,
      I. B.; Barry, D. M. J.; Allen, E. M.;  Roberts, A. P. (1993) Smoking and the sudden death syndrome. Pediatrics
      91: 893-896.
Montgomery, M. R.; Rubin, R. J. (1971) The effect of carbon monoxide inhalation on in vivo drug metabolism in the rat.
      J. Pharmacol. Exp. Ther. 179: 465-473.
Montgomery, M. R.; Rubin,  R. J. (1973) Oxygenation during inhibition of drug metabolism by carbon monoxide orhypoxic
      hypoxia. J. Appl. Physiol. 35: 505-509.
Moore, L.  G.; Rounds, S. S.; Jahnigen, D.;  Grover, R. F.; Reeves, J. T. (1982) Infant birth weight is related to maternal
      arterial oxygenation at high altitude. J. Appl. Physiol.: Respir. Environ. Exercise Physiol. 52: 695-699.
Morris, R. D.; Naumova, E.  N. (1998) Carbon monoxide and hospital admissions for congestive heart failure: evidence of
      an increased effect at low temperatures. Environ. Health Perspect. 106: 649-653.
Morris, R. D.; Naumova, E. N.; Munasinghe, R. L. (1995) Ambient air pollution and hospitalization for congestive heart
      failure among elderly people in seven large US cities. Am. J. Public Health 85: 1361-1365.
Murphy, S. D. (1964) A review of effects onanimals of exposure to auto exhaust and some of its components. J. AirPollut.
      Control Assoc. 14: 303-308.
Murray, F. J.; Schwetz, B. A.; Crawford, A.  A.; Henck, J. W.; Staples, R. E. (1978) Teratogenic potential of sulfur dioxide
      and carbon monoxide in mice and rabbits. In: Mahlum, D. D.; Sikov, M. R.; Hackett, P. L.; Andrew, F. D., eds.
      Developmental toxicology of energy-related pollutants: proceedings of the seventeenth annual Hanford biology
      symposium; October 1977; Richland, WA. Oak Ridge, TN: U.S. Department of Energy, Technical Information
      Center; pp. 469-478.
                                                   6-58

-------
National Research Council. (1986) Environmental tobacco smoke: measuring exposures and assessing health effects.
      Washington, DC: National Academy Press.
Niden, A. H.; Schulz, H. (1965) The ultrastructural effects of carbon monoxide inhalation on the rat lung. Virchows Arch.
      Pathol. Anat. Physiol. 339: 283-292.
Norris, G.; Young-Pong, S. N.; Koenig, J. Q.; Larson, T. V.; Sheppard, L.; Stout, J. W. (1999) An association between fine
      particles and asthma emergency department visits for children in Seattle. Environ. Health Perspect. 107: 489-493.
Olds, D. (1997) Tobacco exposure and impaired development: a review of the evidence. Ment. Retard. Dev. Disabil. Res.
      Rev. 3:257-269.
Olds, D. L.; Henderson, C. R.; Tatelbaum, R. (1994a) Intellectual impairment in children of women who smoke cigarettes
      during pregnancy. Pediatrics 93: 221-227.
Olds, D. L.; Henderson, C. R.; Tatelbaum, R. (1994b) Prevention of intellectual impairment in children of women who
      smoke cigarettes during pregnancy. Pediatrics 93: 228-233.
Otto, D. A.; Benignus, V. A.; Prah, J. D. (1979) Carbon monoxide and human time discrimination: failure to replicate
      Beard-Wertheim experiments. Aviat. Space Environ. Med. 50: 40-43.
Pankow, D. (1996) Carbon monoxide formation due to metabolism of xenobiotics. In: Penney, D.G.,ed. Carbon monoxide.
      Boca Raton, FL: CRC Press; pp. 25-43.
Pankow, D.; Ponsold, W.; Fritz, H. (1974) Combined effects of carbon monoxide and ethanol on the activities of leucine
      aminopeptidase and glutamic-pyruvic transaminase in the plasma of rats. Arch. Toxicol. 32: 331-340.
Pantazopoulou, A.; Katsouyanni, K.; Kourea-Kremastinou, J.; Trichopoulos, D. (1995) Short-term effects of air pollution
      on hospital emergency outpatient visits and admissions in the greater Athens, Greece area. Environ. Res. 69: 31-36.
Paulson, O. B.; Parving, H.-H.; Olesen, J.; Skinhoj, E.  (1973) Influence of carbon monoxide and of hemodilution on
      cerebral blood flow and blood gases in man. J. Appl. Physiol. 35: 111-116.
Peacock, J. L.; Bland, J. M; Anderson, H. R.; Brooke, O. G. (1991a) Cigarette smoking and birthweight: type of cigarette
      smoked and a possible threshold effect. Int. J. Epidemiol. 20: 405-412.
Peacock, J. L.; Bland, J. M.; Anderson, H. R. (1991b) Effects on birthweight of alcohol and caffeine consumption in
      smoking women. J. Epidemiol. Commun. Health 45: 159-163.
Perm, A. (1993) Determination of the atherogenic potential of inhaled carbon monoxide. Cambridge, MA: Health Effects
      Institute; research report number 57.
Perm, A.; Currie, J.; Snyder, C. (1992) Inhalation of carbon monoxide does not accelerate arteriosclerosis in cockerels.
      Eur. J. Pharmacol. 228: 155-164.
Penney, D. G. (1988) A review: hemodynamic response to carbon monoxide. Environ. Health Perspect 77: 121-130.
Penney, D. G., ed. (1996a) Carbon monoxide. Boca Raton, FL: CRC Press.
Penney, D. G. (1996b) Effects of carbon monoxide exposure on developing animals and humans. In: Penney, D. G., ed.
      Carbon monoxide. Boca Raton, FL: CRC Press; pp. 109-144.
Penney, D.; Benjamin, M.; Dunham,  E. (1974) Effect of carbon monoxide on cardiac weight as compared with altitude
      effects. J. Appl. Physiol. 37: 80-84.
Pereira, L. A. A.; Loomis, D.; Conceicao, G. M. S.; Braga, A. L. F.; Areas, R. M.; Kishi, H. S.; Singer, J. M.; Bohm, G. M.;
      Saldiva, P. H. N. (1998) Association between air pollution and intrauterine mortality in Sao  Paulo, Brazil. Environ.
      Health Perspect. 106: 325-329.
Peters, K. D.; Kochanek, K. D.; Murphy, S. L.; (1998) Deaths: final data for 1996. Atlanta, GA: U.S. Centers for Disease
      Control and Prevention, National Center for Health Statistics. (National vital  statistics  reports: v. 47,  no. 9).
      Available: www.cdc.gov/nchswww/releases/98facts/98sheets/finmort.htm [2000, June 8].
Pirnay, F.; Dujardin J.;  Deroanne, R.; Petit, J. M. (1971) Muscular exercise during intoxication by carbon monoxide.
      J. Appl. Physiol. 31: 573-575.
Poloniecki, J. D.; Atkinson, R. W.; Ponce de Leon, A.; Anderson, H. R. (1997) Daily time series for cardiovascular hospital
      admissions and previous day's air pollution in London, UK. Occup. Environ. Med. 54: 535-540.
Ponsonby, A.-L.; Dwyer, T.; Gibbons, L. E.; Cochrane, J. A.; Wang, Y.-G. (1993) Factors potentiating the risk of sudden
      infant death syndrome associated with the prone position. N. Engl. J. Med. 329: 377-382.
Preziosi, T. J.; Lindenberg, R.; Levy,D.; Christenson, M. (1970) An experimental investigation in animals of the functional
      and morphologic effects of single and repeated exposures to high and low concentrations of carbon monoxide. Ann.
      N. Y. Acad. Sci. 174: 369-384.
Prigge, E.; Hochrainer, D. (1977) Effects of carbon monoxide inhalation on erythropoiesis and cardiac hypertrophy in fetal
      rats. Toxicol. Appl. Pharmacol. 42: 225-228.
Putz, V. R. (1979) The effects of carbon monoxide on dual-task performance. Hum. Factors 21: 13-24.
                                                   6-59

-------
Putz, V. R.; Johnson, B. L.; Setzer, J. V. (1976) Effects of CO on vigilance performance: effects of low level carbon
     monoxide on divided attention, pitch discrimination, and the auditory evoked potential.  Cincinnati, OH: U.S.
     Department of Health, Education, and Welfare, National Institute for Occupational Safety and Health; report no.
     NIOSH-77-124.
Putz, V. R.; Johnson, B. L.; Setzer, J. V. (1979) A comparative study of the effects of carbon monoxide and methylene
     chloride on human performance. J. Environ. Pathol. Toxicol. 2: 97-112.
Ramsey, J. M. (1973) Effects of single exposures of carbon monoxide on sensory andpsychomotorresponse. Am. Ind. Hyg.
     Assoc. J. 34: 212-216.
Raven, P. B.; Drinkwater, B. L.; Horvath, S. M.; Ruhling, R. O.; Gliner, J. A.; Sutton, J. C.; Bolduan, N. W. (1974a) Age,
     smoking habits, heat stress, and their interactive effects with carbon monoxide and peroxyacetylnitrate on man's
     aerobic power. Int. J. Biometeorol. 18: 222-232.
Raven, P. B.;  Drinkwater, B. L.; Ruhling, R. O.; Bolduan, N.; Taguchi, S.; Gliner, J.; Horvath, S. M. (1974b) Effect of
     carbon monoxide and peroxyacetyl nitrate on man's maximal aerobic capacity. J. Appl. Physiol. 36: 288-293.
Ritz, B.; Yu, F. (1999) The effect of ambient carbon monoxide on low birth weight among children born in southern
     California between 1989 and 1993. Environ. Health Perspect. 107: 17-25.
Robertson, G.; Lebowitz, M. D. (1984) Analysis of relationships between symptoms and environmental factors over time.
     Environ. Res. 33: 130-143.
Robkin, M. A. (1997) Carbon monoxide and the embryo. Int. J. Dev. Biol. 41: 283-289.
Roche, S.; Horvath, S.; Gliner, J.; Wagner, J.; Borgia, J. (1981) Sustained visual attention and carbon monoxide:
     elimination of adaptation effects. Hum. Factors 23: 175-184.
Rockwell, T. J.; Weir, F. W. (1975) The interactive effects of carbon monoxide and alcohol on driving skills. Columbus,
     OH: The Ohio State University Research Foundation; CRC-APRAC project CAPM-9-69.
Rodkey, F. L.; Collison, H. A. (1979) Effects of oxygen and carbon dioxide on carbon monoxide toxicity. J. Combust.
     Toxicol. 6: 208-212.
Roth, R. A., Jr.; Rubin, R. J. (1976a) Role of blood flow in carbon monoxide- and hypoxic hypoxia-induced alterations in
     hexobarbital metabolism in rats. Drug Metab. Dispos. 4: 460-467.
Roth, R. A.,  Jr.; Rubin, R.  J. (1976b) Comparison of the effect of carbon  monoxide and  of  hypoxic hypoxia.
     II. Hexobarbital metabolism in the isolated, perfused rat liver. J. Pharmacol. Exp. Ther. 199: 61-66.
Roth, R. A., Jr.; Rubin, R. J. (1976c) Comparison of the effect of carbon monoxide and of hypoxic hypoxia. I. In vivo
     metabolism, distribution and action of hexobarbital. J. Pharmacol. Exp. Ther. 199:  53-60.
Rubin, D. H.; Leventhal, J. M.; Krasilnikoff, P. A.; Weile, B.; Berget, A. (1986) Effect of passive smoking on birth-weight.
     Lancet (8504): 415-417.
SAS Institute  Inc. (1990) The NLIN procedure. In: SAS/STAT® user's guide, version 6, fourth edition. Volume 2. Cary,
     NC: SAS Institute Inc.; pp. 1135-1193.
Saldiva, P. H. N.; Lichtenfels, A. J. F. C.; Paiva, P. S. O.; Barone, I. A.; Martins, M. A.; Massad, E.; Pereira, J. C. R.;
     Xavier, V. P.; Singer, J. M.; Bohm, G. M. (1994) Association between air pollution and mortality due to respiratory
     diseases in children in Sao Paulo, Brazil: a preliminary report. Environ. Res. 65: 218-225.
Saldiva, P. H. N.; Pope, C. A., Ill; Schwartz, J.; Dockery, D. W.; Lichtenfels, A. J.; Salge, J. M.; Barone, I.; Bohm, G. M.
     (1995) Air pollution and mortality in elderly people: a time-series study in Sao Paulo, Brazil. Arch. Environ. Health
     50: 159-163.
Salinas, M.; Vega, J. (1995) The effect of outdoor air pollution on mortality risk: an ecological study from Santiago, Chile.
     World Health Stat. Q. 48: 118-125.
Sardinas, A.; Wang Miller, J.; Hansen, H. (1986) Ischemic heart disease mortality of fireman and policemen. Am. J. Public
     Health 76:  1140-1141.
Schoendorf, K. C.; Kiely, J. L. (1992) Relationship of sudden infant death syndrome to maternal smoking during and after
     pregnancy. Pediatrics 90: 905-908.
Schwartz, J. (1997) Air pollution and hospital admissions for cardiovascular disease in Tucson. Epidemiology 8: 371-377.
Schwartz, J. (1999) Air pollution and hospital admissions for heart disease ineightU.S. counties. Epidemiology 10:17-22.
Schwartz,  J.; Morris, R. (1995) Air pollution and hospital admissions for cardiovascular disease in Detroit, Michigan.
     Am. J. Epidemiol. 142: 23-35.
Schwartz,  P. J.; Stramba-Badiale,  M.;  Segantini, A.; Austoni, P.; Bosi, G.; Giorgetti, R.;  Grancini, F.; Marni, E. D.;
     Perticone, F.; Rosti, D.;  Salice, P. (1998) Prolongation of the QT interval and the sudden infant death syndrome.
     N. Engl. J. Med. 338: 1709-1761.
Scragg, R.; Mitchell, E. A.; Taylor, B.  J.; Stewart, A. W.; Ford, R. P. K.; Thompson, J. M. D.; Allen, E. M.; Becroft,
     D. M. O. (1993) Bed sharing, smoking, and alcohol in the sudden infant death syndrome. Br. Med J. 307:1312-1318.


                                                   6-60

-------
Seeker-Walker, R. H.; Vacek, P. M; Flynn, B. S.; Mead, P. B. (1997) Smoking in pregnancy, exhaled carbon monoxide,
     and birth weight. Obstet. Gynecol. 89: 648-653.
Shephard, R. J. (1983) Carbon monoxide: the silent killer. Springfield, IL: Charles C. Thomas. (American lecture series
     no. 1059).
Shephard, R. J. (1984) Athletic performance and urban air pollution. Can. Med. Assoc. J. 131: 105-109.
Sheppard, D.; Distefano, S.; Morse, L.; Becker, C. (1986) Acute effects of routine firefighting on lung function. Am. J. Ind.
     Med. 9: 333-340.
Sheppard, L.; Levy, D.; Norris, G.; Larson, T. V.; Koenig, J. Q. (1999)  Effects of ambient air pollution on nonelderly
     asthma hospital admissions in Seattle, Washington, 1987-1994. Epidemiology 10: 23-30.
Sheps, D. S.; Adams, K. F., Jr.; Bromberg, P. A.; Goldstein, G. M.; O'Neil, J. J.; Horstman, D.; Koch, G. (1987) Lack of
     effect of low levels of carboxyhemoglobin on cardiovascular function in patients with ischemic heart disease. Arch.
     Environ. Health 42: 108-116.
Sheps, D. S.; Herbst, M. C.; Hinderliter, A. L.; Adams, K. F.; Ekelund, L. G.; O'Neil, J. J.; Goldstein, G. M.; Bromberg,
     P. A.; Dalton, J. L.; Ballenger, M. N.; Davis, S. M.; Koch, G. G.  (1990) Production of arrhythmias by elevated
     carboxyhemoglobin in patients with coronary artery disease. Ann. Intern. Med. 113: 343-351.
Sheps, D. S.; Herbst, M. C.; Hinderliter, A. L.; Adams, K. F.; Ekelund, L. G.; O'Neil, J. J.; Goldstein, G. M.; Bromberg,
     P. A.; Ballenger, M.; Davis, S. M.; Koch,  G. (1991) Effects of 4 percent and 6 percent carboxyhemoglobin on
     arrhythmia production in patients with coronary artery disease. Cambridge, MA: Health Effects Institute; research
     report no. 41.
Shimazu, T.; Ikeuchi, H.; Hubbard, G. B.; Langlinais, P. C.; Mason, A. D., Jr.; Pruitt, B. A., Jr. (1990) Smoke inhalation
     injury and the effect of carbon monoxide in the sheep model. J. Trauma 30: 170-175.
Singh, J. (1986) Early behavioral alterations in mice following prenatal carbon monoxide exposure. Neurotoxicology
     7: 475-481.
Singh, J.; Moore-Cheatum, L. (1993) Gestational protein deficiency enhances fetotoxicity of carbon monoxide. In: Keen,
     C. L.; Bendich, A.; Willhite, C. C., eds. Maternal nutrition and pregnancy outcome.  Ann. N. Y. Acad. Sci.
     678: 366-368.
Singh, J.; Aggison, L., Jr.; Moore-Cheatum, L. (1993) Teratogenicity and developmental toxicity of carbon monoxide in
     protein-deficient mice. Teratology 48: 149-159.
Sinha, A. K.; Klein, J.; Schultze, P.; Weiss, J.; Weiss, H. R. (1991) Cerebral regional capillary perfusion and blood flow
     after carbon monoxide exposure. J.  Appl. Physiol. 71: 1196-1200.
Smith, C. J.; Steichen, T. J. (1993) The  atherogenic potential of carbon monoxide. Atherosclerosis  (Shannon, Irel.)
     99: 137-149.
Smith, M. D.; Merigan, W. H.; Mclntire, R. W.  (1976) Effects of  carbon monoxide on fixed-consecutive-number
     performance in rats. Pharmacol. Biochem. Behav. 5: 257-262.
Snella, M.-C.;  Rylander, R. (1979) Alteration in local and systemic immune capacity after exposure to bursts  of CO.
     Environ. Res. 20: 74-79.
Spitzer, W. O.; Lawrence, V.; Dales, R.; Hill, G.; Archer, M. C.; Clark, P.; Abenhaim, L.; Hardy, J.; Sampalis, J.; Pinfold,
     S. P.; Morgan, P. P. (1990) Links between passive smoking and disease: a best-evidence synthesis. Clin. Invest. Med.
     13: 17-42.
Steenland, K. (1992) Passive smoking and the risk of heart disease. JAMA J. Am. Med. Assoc. 267: 94-99.
Steenland, K.; Thun, M.; Lally, C.; Heath, C., Jr. (1996) Environmental tobacco smoke and coronary heart disease in the
     American Cancer Society CPS-II cohort. Circulation 94: 622-628.
Stern, F. B.; Lemen, R. A.; Curtis, R. A. (1981) Exposure of motor vehicle examiners to carbon monoxide: a historical
     prospective mortality study. Arch. Environ. Health  36: 59-66.
Stern, F. B.; Halperin, W. E.; Hornung, R. W.; Ringenburg, V. L.; McCammon, C. S. (1988) Heart disease mortality among
     bridge and  tunnel officers exposed to carbon monoxide. Am. J. Epidemiol. 128: 1276-1288.
Stewart, R. D.;  Peterson, J. E.; Baretta, E. D.; Bachand, R.  T.; Hosko, M. J.; Herrmann, A. A. (1970) Experimental human
     exposure to carbon monoxide. Arch. Environ. Health 21: 154-164.
Stewart, R. D.; Newton, P. E.; Hosko, M. J.; Peterson, J. E. (1973) Effect of carbon monoxide on time perception. Arch.
     Environ. Health 27: 155-160.
Stewart, R. D.; Newton, P. E.; Kaufman, J.; Forster, H. V.; Klein, J. P.; Keelen, M. H., Jr.; Stewart, D. J.; Wu,  A.; Hake,
     C. L. (1978) The effect of a rapid 4% carboxyhemoglobin saturation increase on maximal treadmill exercise.
     New York, NY: Coordinating Research Council, Inc.; report no.  CRC-APRAC-CAPM-22-75.
Strom, J.; Alfredsson, L.; Malmfors, T. (1995) Carbon monoxide: causation and aggravation of cardiovascular diseases—
     a review of the epidemiologic and toxicologic literature. Indoor Environ. 4: 322-333.


                                                    6-61

-------
Sunyer, J.; Anto, J. M.; Murillo, C.; Saez, M. (1991) Effects of urban air pollution on emergency room admissions for
      chronic obstructive pulmonary disease. Am. J. Epidemiol. 134: 277-286.
Surgeon General of the United States. (1983) The health consequences of smoking: cardiovascular disease - a report of the
      Surgeon General. Rockville, MD: U.S. Department of Health and Human Services, Public Health Service, Office
      on Smoking and Health; publication no. DHHS(PHS) 84-50204.
Surgeon General of the United States. (1986) The health consequences of involuntary smoking: a report of the Surgeon
      General. Rockville, MD: U.S. Department of Health andHuman Services, Office on Smoking and Health; publication
      no. DHHS (CDC) 87-8398.
Thorn, S. R.; Ischiropoulos,H. (1997) Mechanism of oxidative stress fromlow levels of carbon monoxide. Cambridge, MA:
      Health Effects Institute; research report no. 80.
Thorn, S. R.; Xu, Y. A.; Ischiropoulos, H. (1997) Vascular endothelial cells generate peroxynitrite in response to carbon
      monoxide exposure. Chem. Res. Toxicol. 10: 1023-1031.
Topping, D. L.; Fishlock, R. C.; Trimble, R. P.; Storer, G. B.; Snoswell, A. M. (1981) Carboxyhaemoglobin inhibits the
      metabolism of ethanol by perfused rat liver. Biochem. Int. 3: 157-163.
Touloumi, G.; Pocock, S. J.; Katsouyanni, K.; Trichopoulos, D. (1994) Short-term effects of air pollution on daily mortality
      in Athens: a time-series analysis. Int. J. Epidemiol. 23: 957-967'.
Touloumi, G.;  Samoli, E.; Katsouyanni, K. (1996) Daily mortality and "winter type" air pollution in Athens, Greece —
      a time series analysis within the APHEA project. In: StLeger, S., ed. The APHEA project. Short term effects of air
      pollution on health: a European approach using epidemiological time series data. J. Epidemiol. Commun. Health
      50(suppl. 1): S47-S51.
Turino, G. M. (1981) Effect of carbon monoxide on the cardiorespiratory system. Carbon monoxide toxicity: physiology
      and biochemistry. Circulation 63: 253A-259A.
U.S. Centers for Disease Control and Prevention. (1997) Mortality patterns—preliminary data, United States, 1996. Morb.
      Mortal. Wkly. Rep. 46: 941-944.
U.S. Department of Health and Human Services. (1998) International classification of diseases, ninth revision (ICD-9).
      Hyattsville, MD: U.S. Centers for Disease Control and Prevention, National Center for Health Statistics.
U.S. Environmental Protection Agency. (1979) Air quality  criteria for carbon monoxide. Research Triangle Park,  NC:
      Office of Health and Environmental  Assessment, Environmental Criteria and Assessment Office; report no.
      EPA-600/8-79-022.
U.S. Environmental Protection  Agency.  (1984) Revised evaluation of health effects associated with carbon monoxide
      exposure: an addendum to the 1979 EPA air quality criteria document for carbon monoxide. Research Triangle Park,
      NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no.
      EPA-600/8-83-033F.
U.S. Environmental Protection Agency. (1986) Air quality criteria for lead. Research Triangle Park, NC: Office of Health
      and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA-600/8-83/028aF-dF.
      4v.
U.S. Environmental Protection Agency. (1990) Air quality criteria for lead: supplement to the 1986 addendum. Research
      Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office;
      report no. EPA/600/8-89/049F.
U.S. Environmental Protection Agency. (1991) Air quality  criteria for carbon monoxide. Research Triangle Park,  NC:
      Office of Health and Environmental  Assessment, Environmental Criteria and Assessment Office; report no.
      EPA/600/8-90/045F.
U.S. Environmental Protection Agency. (1996) Air quality  criteria for paniculate matter. Research Triangle Park,  NC:
      National  Center for  Environmental  Assessment-RTF  Office;  report  nos.  EPA/600/P-95/001aF-cF.   3v.
      Available from: NTIS, Springfield, VA; PB96-168224.
Verhoeff, A. P.; Hoek, G.; Schwartz, J.; Van Wijnen, J. H.  (1996) Air pollution and daily mortality in Amsterdam.
      Epidemiology 7: 225-230.
Vogel, J. A.; Gleser, M. A. (1972) Effect of carbon monoxide on oxygen transport during exercise. J. Appl. Physiol.
      32: 234-239.
Wall, M. A.; Johnson, J.; Jacob, P.; Benowitz, N. L. (1988) Cotinine in the serum, saliva, and urine of nonsmokers, passive
      smokers, and active smokers. Am. J. Public Health 78: 699-701.
Wang, X.; Ding, H.; Ryan, L.; Xu, X. (1997) Association between air pollution and low birth weight: a community-based
      study. Environ. Health Perspect. 105: 514-520.
Weinberg, C. R.; Wilcox, A. J. (1998) Reproductive epidemiology. In: Rothman, K. J.; Greenland, S., eds. Modern
      epidemiology. 2nd ed. Philadelphia, PA: Lippincott-Raven Publishers; pp. 585-608.


                                                   6-62

-------
Weir, F. W.; Rockwell, T. H.; Mehta, M. M.; Attwood, D. A.; Johnson, D. F.; Herrin, G. D.; Anglen, D. M; Safford, R. R.
     (1973) An investigation of the effects of carbon monoxide on humans in the driving task: final report. Columbus,
     OH:  The Ohio  State University Research Foundation; contract no. 68-02-0329 and CRC-APRAC project
     CAPM-9-69.
Weiser, P. C.; Morrill, C. G.; Dickey, D. W.; Kurt, T. L.; Cropp, G. J. A. (1978) Effects of low-level carbon monoxide
     exposure on the adaptation of healthy young men to aerobic work at an altitude of 1,610 meters. In: Folinsbee, L. J.;
     Wagner, J. A.; Borgia, J. F.; Drinkwater, B. L.; Gliner, J. A.; Bedi, J. F., eds. Environmental stress: individual human
     adaptations. New York, NY: Academic Press, Inc.; pp. 101-110.
Weiss, S. T. (1996) Cardiovascular effects of environmental tobacco smoke. Circulation 94: 599.
Weissbecker, L.; Carpenter, R. D.; Luchsinger, P.  C.; Osdene, T. S. (1969) In vitro alveolar macrophage viability: effect
     of gases. Arch. Environ. Health 18: 756-759.
Wells, A. J. (1994) Passive smoking as a cause of heart disease. J. Am. Coll. Cardiol. 24: 546-554.
Wen, S. W.; Goldenberg, R. L.; Cutter, G. R.;  Hoffman, H. J.; Cliver, S. P.; Davis, R. O.; DuBard, M. B. (1990) Smoking,
     maternal age, fetal growth, and gestational age at delivery. Am. J. Obstet. Gynecol. 162: 53-58.
Werner, R. M.; Pearson, T.  A. (1998) What's so passive about smoking? Secondhand smoke as a cause of atherosclerotic
     disease [editorial]. JAMA J. Am.  Med. Assoc. 279: 157-158.
Wietlisbach, V.; Pope, C. A., Ill; Ackermann-Liebrich, U. (1996) Air pollution and daily mortality in three Swiss urban
     areas. Soz. Praeventivmed. 41: 107-115.
Wilcosky, T. C.; Simonsen, N. R. (1991) Solvent exposure and cardiovascular disease. Am. J. Ind. Med. 19: 569-586.
Wirkner, K.; Damme, B.; Poelchen, W.; Pankow, D. (1997) Effect of long-term ethanol pretreatment on the metabolism
     of dichloromethane to carbon monoxide in rats. Toxicol. Appl. Pharmcol.  143: 83-88.
Wouters, E. J. M.; de Jong, P. A.; Cornelissen, P.  J. H.; Kurver, P. H. J.; van Oel, W. C.; van Woensel, C. L. M. (1987)
     Smoking and low birth weight: absence of influence by carbon monoxide? Eur. J. Obstet. Gynecol. Reprod. Biol.
     25:35-41.
Wright, G.; Randell, P.; Shephard, R. J. (1973) Carbon monoxide and driving skills. Arch. Environ. Health 27:  349-354.
Yang, L.; Zhang, W.; He, H.; Zhang, G. (1988) Experimental studies on combined effects of high temperature and carbon
     monoxide. J. Tongji Med. Univ. 8: 60-65.
Yang, W.; Jennison, B. L.; Omaye, S.  T. (1997) Air pollution and asthma emergency room visits in Reno, Nevada.
     Inhalation Toxicol. 9: 15-29.
Yang, W.; Jennison, B. L.; Omaye, S.  T. (1998)  Cardiovascular disease hospitalization and ambient levels of carbon
     monoxide. J. Toxicol. Environ. Health Part A 55: 185-196.
Young,  J. S.;  Upchurch,   M.  B.; Kaufman,  M.  J.; Fechter, L. D. (1987)  Carbon monoxide  exposure potentiates
     high-frequency auditory threshold shifts induced by noise. Hear. Res.  26: 37-43.
Zaren, B.; Lindmark, G.; Gebre-Medhin, M. (1996) Maternal smoking andbody composition of the newborn. Acta Paediatr.
     85:213-219.
Zebro,  T.; Wright, E. A.;  Littleton, R.  J.; Prentice, A. I. D. (1983)  Bone changes in mice after prolonged continuous
     exposure to a high concentration of carbon monoxide. Exp. Pathol. 24: 51-67.
Zhu, B.-Q.; Parmley, W. W. (1995) Hemodynamic and vascular effects of active and passive smoking.  Am. Heart J.
      130: 1270-1275.
                                                   6-63

-------
                                     CHAPTER 7
          Integrative Summary and  Conclusions
7.1  Introduction
       Carbon monoxide (CO) is a colorless, tasteless, odorless, and nonirritating gas that is a product of
incomplete combustion of carbon-containing fuels. It also is produced within living organisms by the
natural degradation of hemoproteins (e.g., hemoglobin [Hb],  myoglobin [Mb], cytochromes) or as  a
by-product of xenobiotic metabolism, especially the breakdown of inhaled organic solvents containing
halomethanes (e.g., methylene bromide, iodide, or chloride). With external exposure to additional CO,
subtle health effects can begin to occur, and exposure to very high levels can result in death.
       The health significance of CO in the air largely results from CO being absorbed readily from the
lungs into the bloodstream,  there  forming  a  slowly  reversible complex with  Hb, known  as
carboxyhemoglobin (COHb). The presence of significant levels of COHb in the blood causes hypoxia (i.e.,
reduced availability of oxygen to body tissues). The blood COHb level, therefore, represents a useful
physiological marker to predict the potential health effects of CO exposure. The amount of COHb formed
is dependent on the CO concentration and duration of exposure, exercise (which increases the amount of
air removed and replaced per unit of time for gas exchange), the pulmonary diffusing capacity for CO,
ambient pressure, health status, and the specific metabolism of the exposed individual. The formation of
COHb is a reversible process, but, because of the high affinity of CO for Hb, the elimination half-time is
quite long, varying from 2 to 6.5 h depending on the initial COHb levels. This may lead to accumulation
of COHb, especially if exposure is to varying concentrations of CO over  extended periods of time.
Fortunately, mechanisms exist in normal, healthy individuals to compensate for the reduction in tissue
oxygen caused by increasing levels of COHb.  Cardiac output increases and blood vessels dilate to carry
more blood so that the tissue can  extract adequate amounts of oxygen from the blood. There are several
medical disorders, however, that can make an individual more susceptible to the potential adverse effects
of low levels of CO, especially during exercise. Occlusive vascular disease (e.g., coronary heart disease,
cerebrovascular disease) limits blood flow to the tissues, obstructive lung disease (e.g., bronchitis,
emphysema, asthma) causes gas-exchange abnormalities that limit the amount of oxygen that diffuses into
the blood, and anemia reduces the oxygen-carrying capacity of the blood.  Under any of these conditions,
exposure to CO could reduce further the amount of oxygen available to affected body tissues. A reduction
in oxygen delivery  caused by elevated COHb levels, combined with impaired air or blood flow to the
diseased tissues, will further reduce organ system function and limit exercise capacity.
       The existing National Ambient Air Quality Standards (NAAQS) for CO of 9 ppm for 8 h and
35 ppm for 1 h (Federal Register,  1994) have been established to reduce the risk of adverse health effects
in the population groups most sensitive to the presence of CO in the ambient air.  The term "ambient air"
is interpreted to mean outdoor air measured at ground level where people live and breathe.  A great majority
of people,  however, spend most of their time indoors. A realistic assessment of the health effects from
exposure to ambient CO, therefore, must be set in the context  of total exposure, a major component of
which is indoor exposure.
       This chapter provides a summary of the key factors discussed in Chapters 2 through 6 of the present
document  that determine what risk ambient CO  poses to public health.  An effort  also is made to


                                            7-1

-------
qualitatively delineate key factors that contribute to anticipated health risks from ambient CO in special
subpopulations that form a significant proportion of the population at large.  Risk factors such as age,
gender, and pregnancy are discussed, as well as preexisting heart, lung, vascular, and hematologic diseases.
Subpopulations at risk because of exposure to ambient CO alone, or CO combined with other environmental
factors, are identified. This information will be used by the U. S. Environmental Protection Agency' s Office
of Air Quality Planning and Standards for development of the staff paper and associated assessments that
will help to evaluate the adequacy of the existing CO NAAQS.
7.2  Environmental Sources
       Carbon monoxide is produced by both natural and anthropogenic processes. About half of the
atmospheric CO is released at the earth's surface from fossil fuel and biomass burning, and the rest is
produced as the result of photochemical reactions in the atmosphere.  About two-thirds of the CO in the
atmosphere arises from human activities; natural processes account  for the remaining one-third.  The
background concentration of CO in the troposphere influences the abundance of hydroxyl radicals (OH),
thus affecting the global cycles of many natural and anthropogenic trace gases, such as methane, that are
removed from the atmosphere by reacting with OH. During the 1980s, CO concentrations in remote marine
areas increased at approximately 1% per year. More recent reports, however, show that CO concentrations
in these locations declined rapidly between 1988 and 1993.  Since 1993, the downward trend in CO has
slowed or leveled off, depending on the measurement laboratory, and it is not clear whether CO will
continue to decline or will increase.
7.3  Environmental Concentrations
       The annual average CO concentration is about 0.13 ppm at monitoring sites located in the marine
boundary layer of the Pacific Ocean in the mid-latitudes of the Northern Hemisphere. These sites are
remote from local pollutant sources, and the values obtained at these sites are thought to represent global
background values for CO.  Because of seasonal variations in the emissions and chemical loss of CO
through reaction with OH radicals, mean global background CO levels vary between about 0.09 ppm in
summer and about 0.16 ppm in winter.  Annual 24-h average  CO concentrations  obtained at U.S.
monitoring sites in rural areas away from metropolitan areas are typically about 0.20 ppm, compared with
an annual 24-h average of 1.2 ppm across all monitoring sites in the Aerometric Information Retrieval
System network in 1996.
       In the United States, ambient air 8-h average CO concentrations monitored at fixed-site stations in
metropolitan areas are generally below 9 ppm and have decreased significantly since 1990 when the last
CO criteria document was completed (U.S.  Environmental Protection Agency, 1991). In the latestyear of
record, 1997, annual mean CO concentrations were all less than 9 ppm. However, in spite of the vehicle
emission reductions responsible for the decrease in ambient CO, high short-term peak CO concentrations
still can occur in certain outdoor locations  and situations associated with  motor vehicles and other
combustion engine sources, for example, riding behind high emitters or in a vehicle with a defective exhaust
system and using lawnmowers, weeders, tillers, or other garden equipment. Also, air  quality data from
fixed-site monitoring stations underestimate the short-term peak CO levels in heavy traffic environments.
       Indoor and in-transit concentrations of CO can be significantly different from the typically low
ambient CO concentrations.  The CO levels in homes without combustion sources are usually lower than
5 ppm. The highest residential concentrations of CO that have been reported are associated with vehicle
startup and idling in attached garages and the use of unvented gas or kerosene space heaters where peak
concentrations of CO as high or higher than 50 ppm have been reported. Carbon monoxide concentrations
also have exceeded 9 ppm for 8 h in several  homes with gas stoves and, in one case, 35 ppm for  1 h;

                                             7-2

-------
however, these higher CO concentrations were in homes with older gas ranges that had pilot lights that burn
continuously. Newer or remodeled homes have gas ranges with electronic pilot lights. Also, the availability
of other cooking appliances (e.g., microwaves, heating plates) has decreased the use of gas ranges in meal
preparation.
       Average CO concentrations as high as 10 to 12 ppm have been reported in human exposure studies
for  in-vehicle compartments of moving automobiles.  Carbon monoxide concentrations will depend,
however, on the season and traffic pattern in a particular locale, and the findings of more recent studies
suggest that results from pre-1990 studies in major cities across the United States are no longer applicable.
For example, commuter exposure to motor vehicle exhaust fell from a historically high value of 3 7 ppm CO
in Los Angeles, CA, during 1965 to a low value of 3 ppm CO on the New Jersey Turnpike in 1992. For
San Francisco, CA, using the same data collection protocol, typical commuter exposures fell about 50%
in the 11 -year period from 1980 to 1991, despite a 19% increase in average daily traffic. Carbon monoxide
levels in other indoor environments affected by engine exhaust (e.g., parking  garages, tunnels) follow
similar trends but tend to be higher than in other indoor environments.
       Because indoor and outdoor air quality differ substantially, and because people spend much of their
time indoors, ambient air quality measurements alone do not provide accurate  estimates of personal  or
population exposure to CO from ambient and nonambient sources.  Whereas the  ambient monitoring data
reflect exposure to ambient sources of CO only, the measurement of CO from personal monitors reflects
more accurately the actual total human population exposure to CO.
7.4 Carboxyhemoglobin Levels in the
       Carbon monoxide diffuses rapidly across
the alveolar and capillary membranes and more
slowly across the placental  membrane.    At
equilibrium, approximately 95% of the absorbed
CO binds with Hb to form COHb that, when
elevated  above  the  endogenous level,  is  a
specific  biomarker of  CO  exposure.   The
remaining  5% is distributed  extravascularly.
During continuous exposure to a fixed ambient
concentration of CO, the COHb concentration
increases rapidly at the onset of exposure, starts
to level off after 3 h, and approaches a steady
state after 6 to 8 h of exposure.  Therefore, an
8-h COHb value should be closely representative
of any longer continuous exposures. In real-life
situations, prediction of individual COHb levels
is difficult because of large spatial and temporal
variations in both indoor and outdoor levels of
CO  and  temporal  variations  of  alveolar
ventilation rates.  Because COHb measurements
    not  readily available  in  the  exposed
                                               Population
are
population,  mathematical models have been
developed to predict COHb levels from known
CO exposures under a variety of circumstances
(see Figure 7-1).
                                                14
                                                13-
                                                12-
                                                11-
                                                10-
                                              4-J
                                              £  9H
                                              a 7H
                                              J3
                                              S  6H
                                              O  5-]
                                                 4
                                                 3-
                                                 2-
                                                 1-
      ______ g ^ 20 |_/mjn

      	8h, 10 L/min
      	1 h, 20 L/min
      	1 h, 10 L/min
                                                          20
40       60
  CO, ppm
                                     80
100
Figure 7-1. Predicted COHb levels resulting from 1- and 8-h
exposures to CO at rest (minute ventilation rate of 10 L/min)
and with light exercise (20 L/min) are based on the Coburn-
Forster-Kane  equation,  using  the  following  assumed
parameters for nonsmoking adults:  altitude = 0 ft, initial
COHb level = 0.5% Haldane coefficient = 218, blood volume
= 5.5 L,  Hb level  = 15 g/100 mL,  lung diffusivity =
30 mL/torr/min, and  endogenous rate of CO production =
0.007 m L/min.
                                              7-3

-------
                                                   Table 7-1. Predicted Carbon Monoxide
                                                        Exposures in the Population
                                                                           Predicted COHb
                                                                             Response3*
                                               Exposure Conditions0
1 h, Light
Exercise
8 h, Light
Exercise
                                               Nonsmoking adults exposed
                                               to 25 to 50 ppm

                                               Workplace or home with
                                               faulty combustion appliances
                                               at -100 ppm	
 2 to 3%
 4 to 7%
 4 to 5%    12 to 13%
                                               a See Figure 7-1 for assumed parameters of the Coburn-
                                                Forster-Kane equation (Coburn et al, 1965).
                                               b Light exercise at 20 L/min.
                                               0 Exposures are steady state.
       Evaluation  of  human  CO   exposure
situations indicates that occupational exposures in
some workplaces, or exposures in homes  with
faulty or unvented combustion sources, can exceed
100 ppm CO, leading to COHb levels of 4 to 5%
with 1 -h exposure and 10% or more with continued
exposure for 8 h or longer (see Table 7-1). Such
high exposure levels are encountered rarely by the
general public under  ambient conditions.  More
frequently, short-term exposures to less than 25 to
50 ppm CO occur in the general population, and, at
the low exercise levels usually engaged in under
such  circumstances,  resulting  COHb  levels
typically  remain  below  2 to   3%  among
nonsmokers. Those levels can be compared to the
physiological baseline for nonsmokers, which is
estimated to be in the range of 0.3 to 0.7% COHb.
Unfortunately, no new data have become available
on the distribution of COHb  levels in the U.S.  population since large-scale nationwide surveys (e.g.,
National Health and Nutrition Examination Survey n [Radford and Drizd, 1982]) and human exposure field
studies (e.g., Denver, CO, and Washington, DC [Aklandetal, 1985]) were conducted in the late 1970s and
early 1980s.
       The major source of total exposure to CO for smokers comes from active tobacco smoking.
Baseline COHb concentrations in smokers average 4%, with a usual range of 3 to 8% for one- to two-
pack-per-day smokers, reflecting absorption of CO from inhaled smoke. Carboxyhemoglobin levels as high
as 15% have been reported for chain smokers.   Exposure to tobacco smoke not  only increases COHb
concentrations in smokers, but, under some circumstances, it also can affect nonsmokers.  In some of the
studies cited in this document, neither the smoking habits of the subjects, nor their passive exposure to
tobacco smoke, have  been taken into account. In addition, as the result of their higher baseline COHb
levels, smokers actually may be exhaling more CO into the air than they are inhaling from the ambient
environment when they are not smoking. Smokers may even show an adaptive response to the elevated
COHb levels, as evidenced by increased red blood  cell volumes or reduced plasma volumes.  As a
consequence, it is not clear if incremental increases in COHb caused by typical ambient exposures actually
would raise the chronically elevated COHb levels resulting from smoking.
7.5  Mechanisms of Carbon Monoxide Activity
       A clear mechanism of action underlying the effects of low-level CO exposure is the decreased
oxygen-carrying capacity of blood and subsequent interference with oxygen release at the tissue level that
is caused by the binding of CO with Hb, producing COHb. The resulting impaired delivery of oxygen can
interfere with cellular respiration and cause tissue hypoxia. The critical tissues (e.g., brain, heart) of healthy
subjects have intrinsic physiologic mechanisms  (e.g., increased blood flow and oxygen extraction) to
compensate  for CO-induced hypoxia.   In compromised subjects, or  as  CO levels increase, these
compensatory mechanisms may be overwhelmed, and tissue hypoxia, combined with impaired tissue
perfusion and systemic hypotension induced by hypoxia, may cause identifiable health effects.
       Carbon monoxide will bind to intracellular  hemoproteins  such  as Mb, cytochrome oxidase,
mixed-function oxidases (e.g., cytochrome P-450), tryptophan  oxygenase, and dopamine hydroxylase.
Hemoprotein binding to CO would be favored under conditions of low intracellular partial pressure of
                                            7-4

-------
oxygen (PO2), particularly in brain and myocardial tissue when intracellular PO2 decreases with increasing
COHb levels. The hemoprotein most likely to be inhibited functionally at relevant levels of COHb is Mb,
found predominantly in heart and skeletal muscle. The physiological significance of CO uptake by Mb is
uncertain,  but  sufficient concentrations  of carboxymyoglobin  potentially  could  limit  maximal
oxygen uptake of exercising muscle.  There is also some evidence that binding of CO to intracellular
hemoproteins may secondarily precipitate oxidative stress. The health risks associated with this mechanism
have not been clearly established.
7.6  Health Effects of Carbon Monoxide
       This document deals primarily with the relatively low concentrations of CO that may induce effects
in humans at or near the lower margin of detection by current technology. Yet, the health effects associated
with exposure to this pollutant range from the more subtle cardiovascular and neurobehavioral effects at
low-ambient concentrations, as identified in the preceding chapter, to unconsciousness and death following
acute exposure to high concentrations. The morbidity and mortality resulting from the latter exposures are
described in several recent reports (Jain,  1990; Penney, 1996; Ernst and Zibrak, 1998).
       The health effects from exposure to low CO concentrations, such as the levels found in ambient air,
are considerably less threatening than those occurring in frank poisoning from high CO concentrations.
Effects of exposure to excessive ambient air levels of CO are summarized here in terms of COHb levels;
however, the  lowest-observed-effect level depends on the method used for  analysis of COHb.   Gas
chromatography (GC) is the method of choice for measuring COHb at saturation levels <5%, because of
the large variability and potentially high bias of optical methods such as CO-Oximetry (CO-Ox). Health
effects are possible in sensitive nonsmoking individuals exposed to ambient CO if peak concentrations are
high enough, or of sufficient  duration,  to  raise the  COHb saturation to critical levels  above their
physiological  baseline of 0.3 to 0.7% (GC).  At 2.3% COHb (GC) or higher, some (predominantly young
and healthy) individuals may experience decreases in maximal exercise duration.  At 2.4% COHb (GC) or
higher, patients with coronary artery disease (CAD) experience reduced exercise time before the onset of
acute myocardial ischemia, which is detectable either by symptoms (angina) or by electrocardiographic
changes (ST  segment  depression).  At 5% COHb (CO-Ox) or higher,  some healthy individuals may
experience impaired psychomotor performance; however, there is too much variability in response across
studies that have tested the same concentrations of CO, and research conducted since the last criteria
document review (U.S. Environmental Protection Agency, 1991) indicates that significant behavioral
impairments in healthy individuals should not be expected until COHb levels exceed 20% (CO-Ox).  At
6% COHb (CO-Ox) or  higher, some people with CAD  and high levels  of baseline ectopy  (chronic
arrhythmia) may experience an increase in the number and complexity of exercise-related arrhythmias.
       Some recent epidemiologic studies have reported findings suggestive of ambient CO levels being
associated with increased exacerbation of heart disease in the population. However, these findings must
be considered to be inconclusive at this time because of questions concerning (a) internal inconsistencies
and overall coherence of the epidemiologic results, (b) how well the community average ambient CO levels
used in the studies (typically derived from a few fixed-site monitors) index either spatially widely variable
ambient CO  levels or personal exposures often augmented by indoor-generated CO sources, (c)  the
extremely small changes (from virtually undetectable up to ca 1.0%) in COHb over baseline levels proj ected
(see Figure 7-1) to occur with the low average ambient CO concentrations (most all < 5 ppm; daily max 1 -h
values) reported in the studies, and (d) the pathophysiologic implausibility of any harmful effects being
exerted at such levels.  Putting the ambient CO levels into perspective, exposures to cigarette smoke or to
combustion exhaust gases from small engines and recreational vehicles typically raise COHb to levels much
higher than levels resulting from  mean  ambient CO exposures, and, for most people, exposures to
indoor sources of CO often exceed controllable outdoor exposures.  The possibility has been posed that the

                                             7-5

-------
average ambient CO levels used as exposure indices in the epidemiology studies may be surrogates for
ambient air mixes impacted by combustion sources and/or other constituent toxic components of such
mixes. More research will be needed to clarify better CO's role.

7.7  Subpopulations Potentially at Risk from Exposure to Ambient Carbon
     Monoxide
     As can be seen from the preceding section, CO-related health effects are most likely to occur in
individuals who are physiologically stressed, either by exercise or by medical conditions that can make
them more susceptible to low levels of CO.
     Most of the known quantifiable concentration-response relationships regarding the human health
effects of CO come from two carefully defined population groups: (1) healthy, predominantly male, young
adults and (2) patients with diagnosed CAD.   On the basis  of the effects described, patients with
reproducible exercise-induced angina appear to be best established as a sensitive group within the general
population that is at increased risk of experiencing the health effects (i.e., decreased exercise duration
because of exacerbation of cardiovascular symptoms) of concern at ambient or near-ambient CO-exposure
concentrations that result in COHb levels as low as 2.4% (GC). Healthy individuals also experience
decreased  exercise duration at similar levels of CO exposure, but only during short-term maximal exercise.
Decrements in exercise duration in the healthy population, therefore, primarily would be a concern for
athletes, rather than for people performing everyday activities.
       It can be hypothesized, however, from both clinical and theoretical work and from experimental
research in laboratory animals, that certain other groups in the population are at potential risk to exposure
from CO.  Probable risk groups that have not been studied adequately, but that could be expected to be
susceptible to CO because of gender differences, aging, or preexisting disease or because of the use of
medications or alterations in their environment include fetuses and young infants; pregnant women; the
elderly, especially those with compromised cardiovascular function; individuals with partially obstructed
coronary arteries but not yet manifesting overt symptomatology of CAD; those with heart failure; people
with peripheral vascular or cerebrovascular disease; individuals with hematologic diseases (e.g., anemia)
that affect  oxygen-carrying capacity or transport in the blood; individuals with genetically unusual forms
of hemoglobin associated with reduced oxygen-carrying capacity; those with chronic obstructive pulmonary
disease; people using medicinal or recreational drugs with central nervous system depressant properties;
individuals exposed to other chemical substances (e.g., methylene chloride) that increase endogenous
formation  of CO; and individuals who have not adapted to high altitude and are exposed to a combination
of high altitude and CO.  Little empirical evidence is available by which to specify health effects associated
with ambient or near-ambient CO exposures in these probable risk groups.

7.7.1  Age, Gender, and Pregnancy as Risk Factors
       The fetus and newborn infant are theoretically susceptible to CO exposure for several reasons. Fetal
circulation is likely to have a higher COHb level than the maternal circulation because of differences in
uptake and elimination of CO from fetal Hb. Because the fetus normally has a lower oxygen tension in the
blood than does the mother, a drop in fetal oxygen tension resulting from the presence of COHb could have
potentially serious effects. The newborn infant, with a comparatively high rate of oxygen consumption and
lower oxygen-transport capacity for Hb than those of most adults, also would be potentially susceptible to
the hypoxic effects of increased COHb. Data from laboratory animal studies on the developmental toxicity
of CO suggest that prolonged exposure to high levels (>60 ppm) of CO during gestation may produce a
reduction  in birth weight, cardiomegaly, and delayed behavioral development.  Limited epidemiologic
findings suggest some association of subchronic ambient CO exposure with low birth weight, but the data
are not conclusive.  Additional studies are needed to determine if chronic exposure to CO, particularly at
                                             7-6

-------
low, near-ambient levels, can compromise the already marginal conditions existing in the fetus and newborn
infant. The effects of CO on maternal-fetal relationships are not well understood.
       In addition to fetuses and newborn infants, pregnant women also represent a susceptible group
because  pregnancy  is associated with  increased  alveolar  ventilation  and  an  increased  rate  of
oxygen consumption that serves to increase the rate of CO uptake from inspired air.  Perhaps a more
important  factor is  that pregnant women experience an  expanded  blood volume  associated with
hemodilution and thus are anemic because of the disproportionate increase in plasma volume compared
with erythrocyte volume. This group may be at increased risk and, therefore, should be studied to evaluate
the effects of ambient CO exposure and elevated COHb levels.
       Changes in metabolism with age may make the aging population particularly susceptible to the
effects of CO. Maximal oxygen uptake declines with age.  Thus, many healthy individuals at 75 years of
age are on the borderline with respect to being able to meet daily metabolic requirements  for ordinary
activities. It is quite possible, therefore, that even low levels of CO exposure might be enough to critically
impair oxygen delivery to the tissues in this aging population and limit daily metabolic requirements. The
rate of decline varies widely among individuals because of the many confounding factors such as heredity,
changes in body mass and composition, and level of fitness.

7.7.2  Preexisting Disease as a Risk Factor
7.7.2.1   Subjects with Heart Disease
       As discussed in Chapter 6, cardiovascular disease comprises many types of medical disorders,
including heart disease, cerebrovascular disease (e.g., stroke), hypertension (high blood pressure), and
peripheral  vascular diseases.  Heart  disease, in turn,  comprises several types of disorders, including
ischemic heart disease (i.e., coronary heart disease [CHD], CAD, myocardial  infarction, and angina),
congestive heart failure, and disturbances of cardiac rhythm (dysrhythmias and arryhthmias). Heart disease
patients often have markedly reduced circulatory capacity and reduced ability to compensate for increased
circulatory demands during exercise and other stress. Therefore, they are especially susceptible to harmful
effects from the reduction in oxygen-carrying capacity of the blood.  Exogenous CO exposure causes such
reduction and, thus, could have serious consequences in heart disease patients.
       Coronary heart disease  remains  the
major  cause   of death   and  disability  in
industrialized societies. In the United States,
CHD  is the single largest killer of both males
and females, causing 481,000 deaths  in 1995
(American Heart Association, 1997), two-thirds
of all deaths from heart disease (U. S. Centers for
Disease  Control and  Prevention,  1997)  and
about half of all deaths from cardiovascular
disease (see Figure  7-2).   Almost 14 million
Americans  have  a  history of  CHD,  with
increased prevalence in both males and females
at increasing ages (see Figure 7-3). Individuals
with CHD 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).   Also, patients who
                50% Coronary Heart Disease

                            22% Other
                                1% Rheumatic Fever/
                                   Rheumatic Heart Disease
                                1% Congenital Heart Defects
                                2% Atherosclerosis
                                4% Congestive Heart Failure

                               4% High Blood Pressure
                        16% Stroke
Figure  7-2.   Percentage  breakdown  of deaths  from
cardiovascular diseases in the United States (1996 mortality
statistics).
Source: American Heart Association (1997); National Center
for Health Statistics (1995).	
                                               7-7

-------
I
Q.
•5
    80
    60
40
    20
                                             79
59


13
6 - 6 4
rim nr-i I
2|
-17
0
45
45





56





70
72





70


       18-19 20-29 30-39 40-49 50-59 60-69 70-79  >80
               Males
                              Females
Figure 7-3. Estimated prevalence of cardiovascular disease
by age and sex for the United States, 1988 to 1991.
Source: American Heart Association (1997); Collins (1997);
Adams and Marano (1995).	
                                                    experience angina  typically  have  additional
                                                    ischemic  episodes  that  are  asymptomatic.
                                                    Unfortunately,   some  individuals  in  the
                                                    population  have  CAD  but   are   totally
                                                    asymptomatic and, therefore, do not know they
                                                    are potentially at risk. It has been estimated that
                                                    5% of middle-aged men show signs of ischemia
                                                    during exercise stress  testing;  a  significant
                                                    number of these men  will have angiographic
                                                    evidence  of  CAD.    Persons   with  both
                                                    asymptomatic and symptomatic CAD  have a
                                                    limited coronary flow reserve and, therefore,
                                                    should  be  sensitive  to   a   decrease   in
                                                    oxygen-carrying  capacity  induced by  CO
                                                    exposure.
                                                           Heart failure is a  major and growing
                                                    public  health  problem.   Almost  5  million
                                                    Americans have heart failure, and about 400,000
                                                    new  cases occur  each year (American Heart
Association, 1997).  Because the prevalence of heart failure increases with age,  prolongation of life
expectancy in the general population would be expected to increase the magnitude of the problem over the
next few decades.  The etiology of heart failure is diverse, but the most common secondary conditions
observed in hospitalized patients are CHD, hypertension, chronic obstructive pulmonary disease, diabetes,
and cardiomyopathy (Croft et al, 1997).  The exacerbation of some of these secondary conditions by CO
are not well known;  however, any heart failure patients with CAD, for example,  might be especially
sensitive to CO exposure.

7.7.2.2   Subjects with Other Vascular Diseases
       Vascular disease, including cerebrovascular disease, is present in both males and females and is
more prevalent above  65  years of age because of the increasing likelihood of adverse  effects from
atherosclerosis or thickening of the artery walls. Atherosclerosis is a leading cause  of deaths from heart
attack and stroke (American Heart Association, 1997). In fact, when considered separately from other
cardiovascular diseases, stroke ranks as the third leading cause of death behind heart disease and cancer
(U.S. Centers for Disease Control and Prevention, 1997). Vascular  diseases are associated with a limited
blood flow capacity and, therefore, patients with these diseases should be sensitive to CO exposure. It is
not clear, however, how low levels of exposure to CO will affect these individuals. For example, only one
study (reviewed in the previous criteria document [U. S. Environmental Protection Agency, 1991]) has been
reported on patients with peripheral vascular disease. Ten men with diagnosed intermittent claudication
(lameness) experienced a significant decrease in time to onset of leg pain when exercising on a bicycle
ergometer after breathing 50 ppm CO for 2 h (2.8% COHb). Further research is needed,  therefore, to
determine more precisely the sensitivity to CO of individuals with vascular disease.

7.7.2.3   Subjects with Anemia and  Other Hematologic Disorders
       Clinically diagnosed low values of Hb, characterized as anemia, are a relatively prevalent condition
throughout the world.  If  the anemia is mild to moderate, an inactive  person is often  asymptomatic.
However, because of the limitation in the oxygen-carrying capacity resulting from the low Hb values, an
anemic person should be more sensitive to low-level CO exposure than would be a person with normal Hb
                                             7-8

-------
levels.  Anemia is more prevalent in pregnant women and in the elderly, two already potentially high-risk
groups. An anemic person also will be more sensitive to the combination of CO exposure and high altitude.
       Individuals with hemolytic anemia often have higher baseline levels of COHb because the rate of
endogenous CO production from heme catabolism is increased. One of the many causes of anemia is the
presence of abnormal Hb in the blood. For example, in sickle-cell disease, the average lifespan of red blood
cells with abnormal Hb S is 12 days compared to an average of 120 days in healthy individuals with normal
Hb. As a result, baseline COHb levels can be as high as 4%.  In subjects with Hb Zurich, where affinity
for CO is 65 times that of normal Hb, COHb levels range from 4 to 7%. Presumably, exogenous exposure
to CO, in conjunction with higher endogenous CO levels, could result in critical levels of COHb. However,
it is not known how ambient or near-ambient levels of CO would affect individuals with these disorders.

7.7.2.4  Subjects with Obstructive Lung Disease
       Chronic obstructive pulmonary disease (COPD) is a prevalent disease especially among smokers,
and a large number (>50%) of these individuals have limitations in their exercise performance demonstrated
by a decrease in oxygen  saturation during mild to moderate exercise. As a consequence, individuals with
hypoxia resulting  from  COPD such as bronchitis and  emphysema may be susceptible to CO during
submaximal  exercise typical of normal daily activity.  In spite of their symptoms, many of them (=30%)
continue to smoke and may have baseline COHb levels of 4 to 8%.  The COPD patients with hypoxia are
also more likely to have CO cause a progression of the disease resulting in severe pulmonary insufficiency,
pulmonary hypertension, and right heart failure.
       Hospital admissions for asthma have increased considerably in the  past few years, particularly
among individuals less than 18 years of age.  Because asthmatics also can experience exercise-induced
airflow limitation, it is likely that they also would experience hypoxia during attacks and be susceptible to
CO. It is not known, however, how exposure to CO actually would affect these individuals. Epidemiologic
observations on the relationship between short-term ambient CO levels and the frequency of respiratory
disease cannot yet be interpreted with confidence.

7.7.3  Subpopulations at Risk from Combined Exposure  to Carbon  Monoxide and Other
       Chemical Substances
7.7.3.1  Interactions with Drugs
       There is an almost complete absence of data on the possible toxic consequences of combined CO
exposure and drug use. Because of the diverse classes of both cardiovascular and psychoactive drugs, and
the many other classes of drugs that have not been examined at all, it must be concluded that this is an area
of concern that is difficult to address meaningfully at the present time.

7.7.3.2  Interactions with Other Chemical Substances in the Environment
       Besides direct exposure to ambient CO, there are other chemical substances in the environment that
can lead to increased COHb saturation when inhaled. Halogenated hydrocarbons used as organic solvents
undergo metabolic breakdown by cytochrome P-450 to form CO and inorganic halide. Possibly the greatest
concern regarding potential risk in the population comes from exposure to one of these halogenated
hydrocarbons (methyl ene chloride) and some of its derivatives that could result in potentially harmful levels
of COHb in individuals  at risk.

7.7.4  Subpopulations Exposed to Carbon  Monoxide at High Altitudes
       For patients with CAD, restricted coronary blood flow limits oxygen delivery to the myocardium.
Carbon monoxide also has the potential for compromising oxygen transport to the heart. For this reason,
such patients have been identified as the subpopulation most sensitive to the effects of CO. A reduction
in PO2 in the atmosphere, as at high altitude, also has the potential for compromising oxygen transport.


                                             7-9

-------
Therefore, patients with coronary artery disease who visit higher elevations may be unusually sensitive to
the added effects of atmospheric CO.
       It is important to distinguish between the long-term resident at high altitude and the newly arrived
visitor from low altitude. Specifically, the visitor will be more hypoxemic than the fully adapted resident.
The combination of high altitude with CO will pose the greatest risk to persons newly arrived at high
altitude who have underlying cardiopulmonary  disease, particularly because  they  are usually older
individuals.
       It is  known that low birth weights occur both in infants born at altitudes above 6,000 ft and in
infants born near sea level, whose mothers had elevated COHb levels because of cigarette smoking. It also
has been shown that COHb levels in smokers at high altitude are higher than those in smokers at sea level.
Although it  is possible that the combination of hypoxic hypoxia and hypoxia  resulting from  ambient
exposure to CO could reduce birth weight  further at high altitude and conceivably  modify future
development, available data are not adequate to confirm this hypothesis.
7.8  Conclusions
       Ambient CO concentrations measured at central, fixed-site monitors in metropolitan areas of the
United States have decreased significantly since the late 1980s, when air quality was reviewed in the
previous criteria document (U.S. Environmental Protection Agency, 1991).  The decline in ambient CO
follows approximately the decline in motor vehicle emissions. Exposure to tobacco smoke, to CO indoors
from unvented or inadequately vented combustion sources, and to CO from uncontrolled outdoor sources
(e.g, small combustion engines) may represent a significant portion of an individual's total CO exposure.
Unfortunately, there is not a good estimate of CO exposure distribution for the current population.
       Health assessment information provided in the present document does not warrant changing the
conclusions of the previous document. The principal cause of CO-induced effects at low levels of exposure
still is thought to be increased COHb formation and the consequent reduction of oxygen delivery to the
body's organs and tissues. The air quality criteria used to support the existing CO NAAQS were primarily
those data obtained from experimental studies of nonsmoking  coronary  artery disease patients during
exercise. These studies identified adverse effects with CO exposures that lead to COHb levels of 2.4%
(GC) or higher. Young, healthy individuals appear to be at little or no health risk because of ambient CO
exposure. In these individuals, the only observed effect of CO exposures resulting in <5% COHb has been
reduction of maximal exercise.  No effects of CO exposures in this range have been observed in healthy
individuals performing submaximal exercise at levels typical of normal human activities.  Greater concern,
therefore, has focused on subpopulations in which biological and pathophysiologic considerations would
suggest increased susceptibility to low-level CO exposure. Indeed, recent epidemiologic studies that have
become available since publication of the previous document are stimulating increased scientific interest
regarding ambient CO exposures as a potential risk factor for exacerbation of heart disease, mortality, and
low birth weight. Results of these studies argue for further research on the health effects of ambient CO
exposure.  This  research  should  address CO alone and CO as a component of the overall ambient air
pollution mixture. Nevertheless, the epidemiologic studies remain subject to considerable biological and
statistical uncertainty, and the available epidemiologic database does not provide convincing evidence that
further selective reduction of ambient CO  levels would substantially benefit public health.
                                             7-10

-------
References
Adams, P. F.; Marano, M. A. (1995) Current estimates from the National Health Interview Survey, 1994. Hyattsville, MD: U.S.
      Department of Health and Human Services, Public Health Service, National Center for Health Statistics; publication no.
      6-1521. (Vital and health statistics: v.  10, no.  193). Available: www.cdc.gov/nchs/products/pubs/pubd/series/srlO/
      199-190/selO_193.htm [2000, February 28].
Akland, G. G.; Hartwell, T. D.; Johnson, T. R.; Whitmore, R. W. (1985) Measuring human exposure to carbon monoxide in
      Washington, B.C., and Denver, Colorado, during the winter of 1982-1983. Environ. Sci. Technol.  19:911-918.
American Heart Association. (1997)  1998 heart and stroke statistical update. Dallas, TX: American Heart Association.
Coburn, R. F.; Forster, R.  E.; Kane, P. B. (1965)  Considerations of the physiological variables  that determine the blood
      carboxyhemoglobin concentration in man. J. Clin. Invest. 44: 1899-1910.
Collins, J. G. (1997) Prevalence of selected chronic conditions: United States, 1990-1992. Hyattsville, MD: U.S. Department of
      Health and Human Services, Public Health Service, National Center for Health Statistics; publication no. 97-1522. (Vital
      and health statistics: v. 10, no. 194). Available: www.cdc.gov/nchs/data/srlO_194.pdf [2000, February 28].
Croft, J. B.; Giles, W. H.; Pollard, R. A.;  Casper, M. L.; Anda, R. F.; Livengood, J. R. (1997) National trends in the initial
      hospitalization for heart failure. J. Am. Geriatr. Soc.  45: 270-275.
Ernst, A; Zibrak, J. D. (1998) Carbon monoxide poisoning. N. Engl. J.  Med. 339: 1603-1608.
Federal Register. (1994)  National ambient air quality standards  for  carbon monoxide—final  decision.  F. R.  (August 1)
      59: 38,906-38,917.
Jain, K. K. (1990) Carbon monoxide poisoning. St. Louis, MO: Warren H.  Green, Inc.
National Center for Health Statistics. (1995) Health, United States, 1994.  Hyattsville, MD: U. S. Department of Health and Human
      Services, Public Health Service; publication no. 95-1232. Available: www.cdc.gov/nchs/data/hus_94.pdf [2000, February
      28].
Penney, D. G., ed. (1996) Carbon monoxide. Boca Raton, FL: CRC Press.
Radford,  E. P.; Drizd, T. A. (1982) Blood carbon  monoxide levels in persons 3-74 years of age: United States, 1976-80.
      Hyattsville, MD: U.S. Department of Health  and Human Services, Public Health Service, National  Center for Health
      Statistics; publication no. (PHS) 82-1250. (Advance  data from vital and health statistics: no. 76).
U. S. Centers for Disease Control and Prevention. (1997) Mortality patterns—preliminary data, United States, 1996. Morb. Mortal.
      Wkly. Rep. 46: 941-944.
U. S. Environmental Protection Agency. (1991) Air quality criteria for carbon monoxide. Research Triangle Park, NC: Office of
      Health and Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA/600/8-90/045F.
                                                       7-11

-------
                               APPENDIX A
               Abbreviations and Acronyms
AER



AIRS



ATPS



a



BS



BTPS



cGMP



C



CAA



CAD



CaO2



CARB



CAS



CASAC



CFD



CFK



CH3



CH4



CHAD
CH3Br
CH3CC13



CH3CHO
CH3C1
Air exchange rate



Aerometric Information Retrieval System



Ambient temperature and pressure, saturated with water vapor



Alpha, the level of acceptable Type 1 error



Black smoke



Body temperature and pressure, saturated with water vapor at 37 °C



Cyclic guanosine monophosphate



Carbon



Clean Air Act



Coronary artery disease



Arterial oxygen content



California Air Resources Board



Children's Activity Survey



Clean Air Scientific Advisory Committee



Cumulative frequency distribution



Coburn-Forster-Kane



Methyl radical



Methane



Consolidated Human Activity Database



Methyl bromide



Methyl chloroform



Acetaldehyde



Methyl chloride
                                      A-1

-------
CH3CO



CHD



CHF



CH2O



CH3O2
CH3OOH
CI



CMAQ



CMRO2



CMSA



CO



C02



COH



COHb



COMb



CO-Ox



COPD



CPSC



CRM



CTM



CVD



DLCO



ECG



ED



ED



EMFAC7C



EPA



ETS



F:CO
Acetyl radical



Coronary heart disease



Congestive heart failure



Formaldehyde



Methyl peroxy radical



Methyl hydroperoxide



Confidence interval



Congestion Management and Air Quality



Cerebral metabolic rate for oxygen



Consolidated metropolitan statistical area



Carbon monoxide



Carbon dioxide



Coefficient of haze



Carboxyhemoglobin



Carboxymyoglobin



CO-Oximetry or CO-Oximeter



Chronic obstructive pulmonary disease



Consumer Product Safety Commission



Certified Reference Material



Chemical Transport Model



Cardiovascular disease



Diffusing capacity for carbon monoxide



Electrocardiogram



Effective  dose for a specific decrement in function



Emergency department



Emissions Factor 7C



U.S. Environmental Protection Agency



Environmental tobacco smoke



Fractional concentration of carbon monoxide in inhaled air
                                          A-2

-------
FDA                 Food and Drug Administration



FID                  Flame ionization detection or detector



FTP                  Federal testing procedures



GAM                General additive model



GC                  Gas chromatography or gas chromatograph



GFC                 Gas filter correlation



GLM                General linear model



hu                   Photon



H                   Atomic hydrogen



H2                   Molecular hydrogen



Hb                  Hemoglobin



HCN                Hydrogen cyanide



HCO                Formyl radical



HO                  Heme oxygenase



HOI                 Isoform of heme oxygenase



HO2                  Hydroperoxy radical



H2O2                Hydrogen peroxide



HOCO               Carboxyl radical



ICD                  International Classification of Diseases



IHD                  Ischemic heart disease



IQR                  Interquartile range



IR                   Infrared



LOEL                Lowest-observed-effect level



LOESS               Locally weighted regression scatter plot smoothing



M                   Haldane coefficient



Mb                  Myoglobin



MI                  Myocardial infarction



ML                  Mercury liberation



MQL                Minimum quantification limit






                                           A-3

-------
MSA

n

N

N2

NAAQS

NAMS

NASA

NDIR

NEM

NHANES

NHAPS

Ni(CO)4

NIST

NMDA

NMHC

NMi

NMOC

NO

•NO

NO2

NOX

NOAA/CMDL


NTRM


O

02

03

OH

O9Hb
Metropolitan Statistical Area

Number

North

Molecular nitrogen

National Ambient Air Quality Standards

National Air Monitoring Station

National Aeronautics and Space Administration

Nondispersive infrared

National Ambient Air Quality Standards Exposure Model

National Health and Nutrition Examination Survey

National Human Activity Pattern Survey

Nickel tetracarbonyl

National Institute of Standards and Technology

N-methyl-D-aspartate

Non-methane hydrocarbon

Nederland Meetinstitut (i.e., Dutch Bureau of Standards)

Non-methane organic compounds

Nitric oxide

Nitric oxide free radical

Nitrogen dioxide

Nitrogen oxides

National Oceanic and Atmospheric Administration Climate Monitoring
  Diagnostics Laboratory

National Institute of Standards and Technology Traceable
  Reference Material

Atomic oxygen

Molecular oxygen

Ozone

Hydroxyl radical

Oxyhemoglobin
                                          A-4

-------
p
"atm
pNEM
PB
PAH
PAN
PCO
PEM
PIA
PM
PM25
PM
   10
P02
PRM
Q
r
R2
RBC
RER
RR
S
SCAQMD
SD
SHAPE
SI
SIDS
Probability
Pressure in atmospheres
Probabilistic National Ambient Air Quality Standards Exposure Model
Barometric pressure
Polyaromatic hydrocarbon
Peroxyacetyl nitrate
Partial pressure of carbon monoxide
Personal exposure monitor
Percentage increases in hospital admissions
Parti cul ate matter
Particulate matter with an aerodynamic diameter < 1
Particulate matter with an aerodynamic diameter <2.5
Parti culate matter with an aerodynamic diameter <10
Average partial pressure of oxygen in lung capillaries in millimeters of mercury
Partial pressure of oxygen in humidified inspired air
Partial pressure of oxygen
Primary Reference Material
Perfusion
Linear regression correlation coefficient
Multiple correlation coefficient
Red blood cell
Respiratory exchange ratio
Relative risk
South
South Coast Air Quality Management District
Standard deviation
Simulation of Human Activity and Pollutant Exposure
International System of Units
Sudden infant death syndrome
                                            A-5

-------
SLAMS



SO2



SRM



ST



STEL



STPD



SUV



TCM



TDL



TDLS



Tg



THb



TSP



TWA



URI



UV



VA



Vb



Vco



VD



VMT
State and Local Air Monitoring Station



Sulfur dioxide



Standard Reference Materials



Segment of the electrocardiogram



Short-term exposure limit



Standard temperature and pressure, dry



Sport utility vehicle



Transportation Control Measure



Tunable diode laser



Tunable diode laser spectroscopy



Teragram



Total blood concentration of hemoglobin



Total suspended particulate



Time-weighted average



Upper respiratory illness



Ultraviolet



Alveolar ventilation



Blood volume



Endogenous carbon monoxide production rate



Volume of physiological  dead space



Vehicle miles of travel
                      A-6

-------