vvEPA
United States
Environmental Protection
Agency
Office of Health and
Environmental Assessment
Washington DC 2O460
EPA/600/8-86/020F
December 1986
Research and Development
Second Addendum to
Air Quality Criteria for
Particulate Matter and
Sulfur Oxides (1982):
Assessment of Newly
Available Health
Effects Information
-------
-------
EPA/600/8-86/020F
December 1986
Second Addendum to
Air Quality Criteria for
Particulate Matter and
Sulfur Oxides (1982):
Assessment of Newly Available
Health Effects Information
Environmental Criteria and Assessment Office
Office of Health and 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 Pro-
tection Agency policy and approved for publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
-------
CONTENTS
LIST OF FIGURES •• ......;...>...,... v
LIST OF TABLES • • Y!
ABSTRACT ..-- - . viy
AUTHORS AND CONTRIBUTORS vi 11
REVIEWERS ......;.,........,....v ; xii
OBSERVER xv
1. INTRODUCTION l~l
1.1 PHYSICAL AND CHEMICAL PROPERTIES OF AIRBORNE PARTICULATE
MATTER AND AMBIENT AIR MEASUREMENT METHODS 1-2
1.2 PHYSICAL/CHEMICAL PROPERTIES OF SULFUR OXIDES AND THEIR
TRANSFORMATION PRODUCTS AND AMBIENT MEASUREMENT METHODS .... 1-10
1.3 KEY AREAS ADDRESSED IN EMERGING NEW HEALTH EFFECTS DATA .... 1-15
2. RESPIRATORY TRACT DEPOSITION AND FATE ...., 2-1
2 1 RESPIRATORY TRACT DEPOSITION AND FATE OF INHALED AEROSOLS .. 2-1
2.2 SULFUR DIOXIDE UPTAKE AND FATE 2-14
2.3 POTENTIAL MECHANISMS OF TOXICITY ASSOCIATED WITH INHALED
PARTICLES AND S02 , 2-15
2.4 SUMMARY AND CONCLUSIONS 2-17
3 EPIDEMIOLOGICAL STUDIES OF HEALTH EFFECTS ASSOCIATED WITH
EXPOSURE TO AIRBORNE PARTICLES AND/OR SULFUR OXIDES 3-1
3.1 HUMAN HEALTH EFFECTS ASSOCIATED WITH SHORT-TERM EXPOSURES .. 3-1
3.1.1 Mortality Effects of Short-Term Exposures 3-2
3.1.2 Morbidity Effects of Short-Term Exposures 3-14
3.2 EFFECTS ASSOCIATED WITH LONG-TERM EXPOSURES 3-20
3.2.1 Mortality Effects of Chronic Exposures 3-20
3.2.2 Morbidity Effects of Long-Term Exposures 3-28
3.3 SUMMARY AND CONCLUSIONS 3-50
4. CONTROLLED HUMAN EXPOSURE STUDIES OF SULFUR DIOXIDE HEALTH
EFFECTS • 4-1
4 1 NORMAL SUBJECTS EXPOSED TO SULFUR DIOXIDE .4-8
4 2 CHRONIC OBSTRUCTIVE PULMONARY DISEASE PATIENTS EXPOSED
TO S02 4-11
4.3 FACTORS AFFECTING THE PULMONARY RESPONSE TO S02 EXPOSURE
IN ASTHMATICS j-ll
4.3.1 Dose-Response Relationships 4-11
4.3.2 S02-Induced Versus Nonspecific Airway Reactivity ... 4-24
4.3.3 Oral, Nasal, and Oronasal Ventilation 4-26
4.3.4 Time Course of Response to S02 in Asthmatics 4-29
4.3.5 Exacerbation of the Responses of Asthmatics to
S02 by Cold/Dry Air 4-32
4.3.6 Clinical Relevance 4-37
-------
CONTENTS (continued)
4.4
4.5
4.4.2
4.4.3
MECHANISM(S)
4.4.1 Mode of Action
Breathing Mode and Interaction with Dry Air
Tolerance (Attenuation of Response) to S02 with
Repeated Exposure
SUMMARY AND CONCLUSIONS
EXECUTIVE SUMMARY
5.1 RESPIRATORY TRACT DEPOSITION AND FATE
5.2 SUMMARY OF EPIDEMIOLOGIC FINDINGS ON HEALTH EFFECTS
ASSOCIATED WITH EXPOSURE TO AIRBORNE PARTICLES AND SO
5.3 SUMMARY OF CONTROLLED HUMAN EXPOSURE STUDIES OF
DIOXIDE HEALTH EFFECTS
REFERENCES
APPENDIX
4-38
4-38
4-41
4-43
4-44
5-1
5-1
5-2
5-7
6-1
A-l
IV
-------
LIST OF FIGURES
Figure Pa9e
1 Representative example of typical biomodal mass distri-
bution and chemical composition in an urban aerosol 1-3
2 Regional deposition of monodisperse aerosols by indicated
particle diameter for mouth breathing (alveolar, trachio-
bronchial) and nose breathing (alveolar) , 2-3
3 Estimates of thoracic deposition of particles between 1
and 15 urn by Miller et al. (1986) for normal augmenters
and mouth breathers 2-9
4 Predicted initial dose to the TB region as a function of
body mass 2-11
5 Adjusted frequency of cough for the 27 region-cohorts
from the Six-Cities Study at the second examination
plotted against mean TSP concentration during the
previous year 3-35
6 Adjusted mean percent of predicted FEVx at the first
examination for the 27 region-cohorts from the Six-Cities
Study plotted against mean TSP concentration during the
previous year . 3~36
7 Gradation of physiological responses to S02 4-7
-------
LIST OF TABLES
Table
Page
1 Summary of quantitative conclusions drawn in U.S. EPA (1982a)
from epidemiological studies relating health effects to acute
exposure to ambient air levels of S02 and PM in London 3-3
2 Summary of quantitative conclusions drawn in U.S. EPA (1982a)
from epidemiological studies relating health effects to
chronic exposure to ambient air levels of PM and/or S02 3-29
3 Summary of key quantitative conclusions based on newly avail-
able epidemiological studies or analyses relating health
effects to acute exposure to ambient air levels of S02 and/or
PM 3-51
4 Summary of key quantitative conclusions based on newly avail-
able epidemiological studies relating human health effects to
long-term exposures of S02 and/or PM 3-54
5 Summary of asthmatic subject characteristics from newly
available controlled human exposure studies of effects of
sulfur dioxide on pulmonary function 4-3
6 Summary of normal subject characteristics from newly
available controlled human exposure studies of effects of
sulfur dioxide on pulmonary function 4-6
7 Summary of results from controlled human exposure studies
of pulmonary function effects associated with exposure of
asthmati cs to S02 4-13
8 Clinically significant responses 4-38
vr
-------
ABSTRACT
This Addendum to the earlier 1982 U.S. EPA document, Air Quality Criteria
for Particulate Matter and Sulfur Oxides, evaluates scientific information on
health effects associated with exposure to various concentrations of sulfur
oxides and particulate matter in ambient air. Although the literature through
1986 has been reviewed thoroughly for information relevant to air quality cri-
teria, the present Addendum is not intended as a complete and detailed review
of all literature pertaining to sulfur oxides and particulate matter. Rather,
an attempt has been made to focus on the evaluation of those studies providing
key information by which to delineate quantitative exposure-effect or dose-
response relationships for the subject pollutants.
Although this Addendum is principally concerned with the health effects of
sulfur oxides and particulate matter, other scientific data are presented and
evaluated in order to provide a better understanding of these pollutants in the
environment. To this end, the Addendum also includes discussions of physical
and chemical properties of sulfur oxides and particulate matter; ambient air
monitoring and related analytical techniques; and the respiratory tract deposi-
tion and fate associated with human exposure to the subject pollutants.
VI 1
-------
AUTHORS AND CONTRIBUTORS
The following people served as authors or otherwise contributed to pre-
paration of the present addendum. Names are listed in alphabetical order.
Dr. Lawrence J. Folinsbee
Environmental Monitoring and Services, Inc.
Chapel Hill, NC 27514
Dr. Lester D. Grant, Director
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Timothy R. Gerrity
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Chapel Hill, NC 27514
Dr. Victor Hasselblad
Center for Health Policy Research
Duke University
Durham, NC 27713
Dr. Donald H. Horstman
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Chapel Hill, NC 27514
Dr. Howard Kehrl
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Chapel Hill, NC 27514
Dr. Dennis Kotchmar
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Mr. Frank F. McElroy
Environmental Monitor and Service Labs
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Fred Miller
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
VI 1 1
-------
AUTHORS AND CONTRIBUTORS (continued)
Dr. L. Jack Roger
Environmental Monitoring
Chapel Hill, NC 27514
and Services, Inc.
Dr. Alan Marcus
Dept. of Mathematics
Washington State University
Pullman, WA 99164
Dr. Joel Schwartz
Office of Policy Analysis (PM
U.S. Environmental Protection
Washington, DC 20460
221)
Agency
IX
-------
U.S. Environmental Protection Agency
Science Advisory Board
Clean Air Scientific Advisory Committee
Particulate Matter/Sulfur Oxides Review Committee
The substance of this document was independently peer-reviewed in public
session by the Clean Air Scientific Advisory Committee, Environmental Protec-
tion Agency Science Advisory Board.
Chairman
Dr. Morton Lippman, Professor, Department of Environmental Medicine, NYU
Medical Center, Tuxedo, New York 10987 (914) 351-2396
Director, Science Advisory Board
Dr. Terry F. Yosie, Science Advisory Board, United States Environmental Protec-
tion Agency, Washington, D.C. 20460
Panel Members
Dr. Mary 0. Amdur, Senior Research Scientist, Energy Laboratory, MIT, Room
16-339, Cambridge, Massachusets 02139 (617) 253-3111
Dr. Edward D. Crandall, Professor of Medicine, Cornell University, New York,
New York 10021 (212) 472-5041
Dr. Robert Frank, Professor of Environmental Health Sciences, Johns Hopkins
School of Hygiene and Public Health, 615 N. Wolfe Street, Baltimore,
Maryland 21205 (301) 955-3720
Dr. Warren B. Johnson, Manager, Research Aviation Facility, National Center
for Atmospheric Research, P.O. Box 3000, Boulder, Colorado 80307
(303) 497-1032
Dr. Timothy Larson, Environmental Engineering and Science Program, Department
of Civil Engineering FX-10, University of Washington, Seattle, Washington
98195 (206) 543-6815
Dr. Roger 0. McClellan, Director, Inhalation Toxicology Research Institute,
Lovelace Biomedical and Environmental Research Institute, P.O. Box 5890,
Albuquerque, New Mexico 87185 (505) 844-6835
-------
Dr. Mark J. Utell, Co-Director, Pulmonary Disease Unit, Associate Professor of
Medicine and Toxicology in Radiation Biology and Biophysics, University of
Rochester Medical Center - Box 692, Rochester, New York 14642
(716) 275-4861
Dr. James H. Ware, Associate Professor, Harvard School of Public Health,
Department of Biostatistics, 677 Huntington Avenue, Boston, MA 02115
(617) 732-1056
Dr. Jerry Wesolowski, Air and Industrial Hygiene Lab, California Department of
Health, 2151 Berkeley Way, Berkeley, California 94704 (415) 540-2476
Dr. James L. Whittenberger, Director, University of California Southern
Occupational Health Center, Prof, and Chair, Department of Community and
Environmental Medicine, California College of Medicine, University of
California, Irvine, 19722 MacArthur Blvd., Irvine, California 92717
(714) 856-6269
Executive Secretary
Mr. Robert Flaak, Environmental Scientist, Science Advisory Board (A-101F), U.S.
Environmental Protection Agency, Washington, D.C. 20460
(202) 382-2552
-------
REVIEWERS
_A preliminary draft version of the present addendum was circulated for
review. Written or oral review comments were received from the following
individuals, most of whom participated (along with the above authors and
contributors) in a peer-review workshop held at EPA's Environmental Research
Center in Research Triangle Park, NC on May 22-23, 1986.
Dr. Karim Ahmed
Natural Resources Defense Council
122 E. 42nd Street
New York, NY 10168
Mr. John Bachmann
Ambient Standards Branch (MD-12)
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. David Bates
Department of Medicine
St. Paul's Hospital
University of British Columbia
Vancouver, British Columbia
Canada V6Z 1Y6
*Dr. Robert Bechtel
National Jewish Center for Immunology
& Respiratory Disease
1400 Jackson Street
Denver, CO 80206
Dr. Per Camner
The Karolinska Institute
P.O. Box 60400
S-104 01 Stockholm
Sweden
Mr. Jeff Cohen
Ambient Standards Branch (MD-12)
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Jack Hackney 213/922-7561
Room 51
Environmental Health Service
Rancho Los Amigos Hospital
7601 Imperial Highway
Downey, CA 90242
XI 1
-------
REVIEWERS (continued)
Dr. Carl Hayes
Health Effects Research Laboratory (MD-55)
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Ian Higgins
American Health Foundation
320 E. 43rd Street
New York, New York 10017
Dr. Steve Horvath
Prof. Inst. of Env. Stress
University of California
Santa Barbara, CA 93106
*Dr. Jane Koenig
Dept. of Environmental Health, SC-34
University of Washington
Seattle, WA 98195
Dr. Emmanuel Landau
American Public Health Assoc.
1015 15th Street, N.W.
Washington, DC 20005
Dr. Alan Marcus
Department of Mathematics
Washington State University
Pullman, WA 99164-2930
Dr. Bart Ostro
Research Division
California Air Resources Board (IPA)
1800 15th Street
Sacramento,, CA 95812
Dr. Haluk Ozkaynak
KSG-EEPC
Harvard University
79 JFK Street
Cambridge, MA 02138
*Dr. William Pierson
Northwest Asthma & Allergy Center
4540 Sand Point Way, N.E.
Seattle, WA 98105
xi n
-------
REVIEWERS (continued')
Mr. Larry J. Purdue
Enviromental Monitoring Systems .Laboratory (M.D-77).
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Neil Roth
6115 Executive Blvd.
Rockville, MD 20852
Dr. Neil Schacter
Mt. Sinai Medical Center
24-30 Annenberg
1 Gustave L. Levy Place
New York, NY 10029
*Dr. Dean Sheppard
Cardiovascular Research Institute
University of California
San Francisco, CA 94143
Dr. Frank Speizer
Channing Laboratory
180 Longwood Avenue
Boston, MA 02115
Dr. John Spengler
Department of Environmental Science & Physiology
Harvard School of Public Health
665 Huntington Avenue
Boston, MA 02115
Dr. David L. Swift
Johns Hopkin University
School of Hygiene
615 N. Wolfe Street
Baltimore, MD 21205
^Written reviews only.
xiv
-------
OBSERVER
The following member of the Clean Air Scientific Advisory Committee (CASAC)
of EPA's Science Advisory Board attended the May 22-23, 1986 workshop as an
observer on behalf of CASAC.
Dr. Timothy Larson
Environmental Engineering and Science Program
Dept. of Civil Engineering EX-100
University of Washington
Seattle, WA 98195
xv
-------
-------
CHAPTER 1. INTRODUCTION
The United States Clean Air Act and its 1977 Amendments mandate that the
U.S. Environmental Protection Agency (U.S. EPA) periodically review criteria
for National Ambient Air Quality Standards (NAAQS) and revise such standards as
appropriate. The most recent periodic review of the scientific bases under-
lying the NAAQS for particulate matter (PM) and sulfur oxides (S0x) culminated
in the 1982 publication of the EPA document Air Quality Criteria for Particulate
Matter and Sulfur Oxides (U.S. EPA, 1982a), an associated PM staff paper (U.S.
EPA, 1982b) which examined the implications of the revised criteria for the
review of the PM NAAQS, an addendum to the criteria document addressing further
information on health effects (U.S. EPA, 1982c), and another staff paper re-
lating the revised scientific criteria to the review of the S0x NAAQS (U.S. EPA,
1982d). Based on the criteria document, addendum and staff papers, revised-
24-hr and annual-average standards for PM have been proposed (Federal Register,
1984a) and public comments on the proposed revisions have been received both in
written form and orally at public hearings (Federal Register, 1984b). Consid-
eration of possible revision of the sulfur oxides NAAQS is still under way.
Since preparation of the above criteria document, addendum, and staff
papers (U.S. EPA, 1982a, b, c, d), numerous new scientific studies or analyses
have become available that may have bearing on the development of criteria for
PM or SO and thus may notably impact proposed revisions of those standards now
under consideration by EPA. In December 1985 the Clean Air Scientific Advisory
Committee (CASAC) of EPA's Science Advisory Board met to discuss the PM proposals
and possible implications of the newly available information. CASAC recom-
mended that a second addendum to the 1982 Criteria Document (U.S. EPA, 1982a)r
be prepared to evaluate new studies and their implications for derivation of
health-related criteria for the PM NAAQS. In the process of responding to
CASAC1s recommendations, the Agency also determined that it would be useful to
examine studies that have emerged since 1982 on the health effects of sulfur
oxides.
1-1
-------
Accordingly, the present addendum (1) summarizes key findings from the
1982 EPA criteria document and first addendum (U.S. EPA, 1982a,c) as they
pertain to derivation of health-related criteria, and (2) provides an updated
assessment of newly available information of potential importance for deriva-
tion of health criteria for both the PM and S0x standards, with major emphasis
on evaluation of human health studies published since 1981. Certain background
information of crucial importance for understanding the assessed health effects
findings is also summarized. This includes information on physical and chemi-
cal properties of PM, sulfur oxides, and associated aerosols (including acid
aerosols) and ambient monitoring techniques. However, new studies on associa-
tions between acid aerosols and health effects are being evaluated in a separate
issue paper.
1.1 PHYSICAL AND CHEMICAL PROPERTIES OF AIRBORNE PARTICIPATE MATTER AND
AMBIENT AIR MEASUREMENT METHODS
As noted in the 1982 EPA criteria document (U.S. EPA, 1982a),'airborne
particles exist in many sizes and compositions that vary widely with changing
source contributions and meteorological conditions. However, airborne particle
mass tends to cluster in two principal size groups: coarse particles, general-
ly larger than 2 to 3 micrometers (pm) in diameter; and fine particles, gener-
ally smaller than 2 to 3 pm in diameter. The dividing line between the coarse
and the fine sizes is frequently given as 2.5 urn, but the distinction according
to chemical composition is neither sharp nor fixed; it can depend on the con-
tributing sources, on meteorology, and on the age of the aerosol.
Fine particle volume (or mass) distributions often exhibit two modes.
Particles in the nuclei mode (which includes particles from 0.005 to 0.05 urn in
diameter) form near sources by condensation of vapors produced by high tempera-
ture processes such as fossil-fuel combustion. Accumulation-mode particles
(i.e., those 0.05-2.0 jjm in diameter) form principally by coagulation or growth
through vapor condensation of short-lived particles in the nuclei mode. Typi-
cally, 80 percent or more of the atmospheric sulfate mass occurs in the accu-
mulation-mode. Particles in the accumulation mode normally do not grow into
the coarse mode. Coarse particles include re-entrained surface dust, salt
spray, and particles formed by mechanical processes such as crushing and
grinding.
1-2
-------
Primary particles are directly discharged from manmade or natural sources.
Secondary particles form by atmospheric chemical and physical reactions, and
most of the reactants involved are emitted as gaseous pollutants. In the air,
particle growth and chemical transformation occur through gas-particle and
particle-particle interactions. Gas-particle interactions include condensation
of low-vapor-pressure molecules, such as sulfuric acid (ti^SQ^) and organic
compounds, principally on fine particles. The only particle-particle interac-
tion important in atmospheric processes is coagulation among fine particles.
As shown in Figure 1, fine atmospheric particles mainly include sulfates,
carbonaceous material, ammonium, lead, and nitrate. Coarse particles consist
mainly of oxides of silicon, aluminum, calcium, and iron, as well as calcium
carbonate, sea salt, and material such as tire particles and vegetation-related
particles (e.g., pollen, spores). The distributions of fine and coarse parti-
cles overlap; some chemical species found mainly in one mode may also be found
in the other.
CRUSTAL MATERIAL
(SILICON COMPOUNDS.
IRON, ALUMINUM). SEA
SALT. PLANT PARTICLES
SULFATES. ORGANICS.
AMMONIUM. NITRATES.
CARBON. LEAD, AND
SOME TRACE CONSTITUENTS
.03
1 3 10
Particle Diameter -i
100
300
Figure 1. Representative example of typical bimodal mass distribution (measured
by impactors) and chemical composition in an urban aerosol. Although some
overlap exists note substantial differences in chemical composition of fine versus
coarse modes. Chemical species of each mode are listed in approximate order of
relative mass contribution. Note that the ordinate is linear and not logarithmic.
Source: Modified from Whitby (1975) and National Research Council (1979).
1-3
-------
The carbonaceous component of fine particles contains both elemental
carbon (graphite and soot) and nonvolatile organic carbon (hydrocarbons in
combustion exhaust and secondary organics formed by photochemistry). In many
urban and nonurban areas, these species are the most abundant fine particles,
after sulfates. Secondary organic particles form by oxidation of primary
organics by a cycle that involves ozone and nitrogen oxides. Atmospheric
reactions of nitrogen oxides yield nitric acid vapor (HNOO that may accumulate
as nitrate particles in the fine or coarse modes. Most atmospheric sulfates
and nitrates are water-soluble and tend to absorb moisture. Hygroscopic growth
of sulfate-containing particles markedly affects their size, reactivity, and
other physical properties which influence their biological and physical
effects.
The relative proportions of particles of different chemical composition
and size ranges can vary greatly in ambient air, depending upon emission
sources from which they originate and interactions with meteorological condi-
tions, e.g., relative humidity (RH) and temperature. Particles from;combustion
of fossil fuels or high-temperature processes, e.g., metal smelting, tend to
fall in the fine (<2.5 urn) or small coarse mode (<10 urn MMD) range; those from
crushing or grinding processes, e.g., mining operations, tend to be mainly in
the coarse mode (>2.5 urn), with a substantial fraction in excess of 10 urn.
Another important distinction concerning airborne particles is the broad
characterization that can result from different methods commonly used for rou-
tine monitoring purposes. The most commonly used methods for collection and
«
measurement of airborne particles were described in U.S. EPA (1982a). As noted
there, differences in measurements obtained from various instruments and
methods used to measure PM levels have important implications for derivation of
quantitative dose-response relationships from epidemiologic studies and for
establishing air quality criteria and standards. It is generally not practic-
able to discriminate on the basis of either particle size or chemical composi-
tion when assessing particulate matter data from routine monitoring networks.
Characteristics of the collected samples are dependent on the types of sources
in the vicinity, weather conditions and sampling procedures. Difficulties that
result and limitations of measurements were also discussed in detail in the
1982 EPA criteria document (U.S. EPA, 1982a).
When considering measurements of airborne particles it is essential to
specify the method used and to recognize that results obtained with one method
1-4
-------
and under a given set of conditions are not necessarily applicable to other
situations. For example, attempts have been made to relate findings based on
smoke measurements (that relate mainly to dark-colored characteristics of
particles from incomplete combustion of coal or other hydrocarbon fuels) to
situations involving total suspended particulate matter (TSP) or size-specific
fractions thereof (measured directly in terms of weight). Because the former
(smoke) methods were used in many early epidemiological studies and the latter
are now more often used for monitoring purposes in many countries, conversion
from one type of measurement to the other would be desirable but, for reasons
noted below, there can be no generally applicable conversion factor. Compara-
tive evaluation of the two methods has been undertaken at numerous sites (Ball
and Hume, 1977; Commins and Waller, 1967; Lee et- al. , 1972), but the results
emphasize that they measure different qualities of the particulate matter and
cannot be directly compared with one another (U.S. EPA, 1982a).
Sampling airborne particles is a complex task because of the wide spectrum
of particle sizes and shapes. Separating particles by aerodynamic size pro-
vides a simplification by disregarding variations in particle shape and relying
on particle settling velocity. The aerodynamic diameter of a particle is not a
direct measurement of its size but is the equivalent diameter of a spherical
particle of specific gravity which would settle at the same rate as the mea-
sured particles. Samplers can be designed to collect particles within sharply
defined ranges of aerodynamic diameters or to simulate the deposition pattern
of particles in the human respiratory system, which exhibits a more gradual
transition from acceptance to exclusion of particles. High-volume (hi-vol)
samplers, dichotomous samplers, cascade impactors, and cyclone samplers are the
most common devices with specifically designed collection characteristics.
These samplers rely on inertial impaction techniques for separating particles
by aerodynamic size, filtration techniques for collecting the particles and
gravimetric measurements for determining mass concentrations. Mass concen-
trations can also be estimated using methods that measure an integral property
of particles such as optical reflectance, and empirical relationships between
mass concentrations and the integral measurement can be used to predict mass
concentration, if a valid physical model relating to the measurements exists
and empirical data verify the model predictions.
The hi-vol sampler collects particles on a glass-fiber filter by drawing
air through the filter at a flow rate of -1.5 m /min and is used to measure
1-5
-------
total suspended participate matter (TSP). The hi-vol sampler has outpoints of
~25 urn at a wind speed of 24 kph and 45 pm at 2 kph. Although sampling effec-
tiveness is wind-speed sensitive, no more than a 10 percent day-to-day variabi-
lity occurs for the same ambient concentration for typical conditions. The
hi-vol is one of the most reproducible particle samplers in use, with a typical
coefficient of variation of 3 to 5. One major problem associated with the
glass-fiber filter used on the hi-vol is formation of artifact mass caused by
the presence of acid gases in the air (e.g., artifactual formation of sulfates
3
from S02), which can add 6 to 7 ug/m to a 24-h sample. The hi-vol has been
the sampler most widely used in the U.S. for routine monitoring and has yielded
TSP mass estimates used in many American epidemiological studies.
Hi-vol samplers with size-selective inlets (SSI) have recently been devel-
oped which collect and measure particles <10 urn or <15 pm. Except for the
inlet, these samplers are identical in design and operation to the TSP hi~vol.
Versions are now being used in epidemiologic health effects studies, and
several models are being evaluated for possible routine monitoring use.
The dichotomous sampler is a low-volume gravimetric measurement device
which collects fine (<2.5 pm) and coarse (>2.5 urn to <10 or 15 urn) ambient
®
particle fractions. The sampler uses Teflon filters which minimize artifact
mass formation. The earlier inlets used with this sampler were very wind-speed
dependent, but newer versions are much improved. Because of low sampling flow
rate, the sampler collects submilligram quantities of particles and requires
microbalance analyses, but is capable of reproducibility of +10 percent or
better. The method, however, has only begun to be employed on any major scale
to generate size-selective data on PM mass assessed in relation to health
effects evaluated in epidemiological studies.
Cyclone inlets with cutpoints around 2 pm have long been used to separate
the fine particle fraction, can be used with samplers designed to cover a range
of sampling flow rates and are available in a variety of physical sizes.
Applications of cyclone inlets are found in 10- and 15-pm cutpoint inlets for
both dichotomous and hi-vol samplers. Samplers with cyclone inlets could be
expected to have coefficients of variations similar to those of the dichotomous
or SSI hi-vol samplers and, until recently, have also found only limited use in
epidemiological studies of PM health effects.
Cascade impactors have been used to obtain mass distribution by particle
size. Because care must be exercised to prevent errors (e.g., those due to
1-6
-------
particle bounce between stages), these samplers are normally not used as
routine monitors. A study by Miller and DeKoning (1974) comparing cascade
impactors with hi-vol samplers showed inconsistencies in mass collections by
the impactors.
Samplers that derive mass concentrations by analytical techniques other
than direct weight have been used extensively. One of the earliest was the
British smokeshade (BS) sampler, which measures the reflectance of particles
collected on a filter and uses empirical relationships to estimate mass concen-
trations. These relationships are more sensitive to carbon concentrations than
mass (Bailey and Clayton, 1980) and hence are very difficult to interpret as
either total or size-selective PM mass present in the atmosphere. The BS
method and its standard variations typically collect PM with an =4.5 urn D5Q
cutpoint under field conditions, with some particles ranging from 7 to 9 urn at
times being collected (McFarland et al. , 1982). Thus, even if larger particles
are present in the atmosphere, the BS method collects mainly fine-mode and
small coarse-mode particles. The BS method neither directly measures mass nor
determines chemical composition of collected PM. Rather, it measures light
absorption of particles indicated by reflectance from a stain formed by parti-
cles collected on filter paper. Reflectance of light from the stain depends
both on density of the stain, or amount of PM collected, and optical properties
of collected PM. Smoke particles composed of elemental carbon in incomplete
fossil-fuel combustion products typically make the greatest contribution to
darkness of the stain, especially in urban areas. Thus, the amount of elemen-
tal carbon, but not organic carbon, in the stain tends to be most highly
correlated with BS reflectance readings. Other nonblack, noncarbon particles
also have optical properties which can affect the reflectance readings, but
usually with negligible contribution to optical absorption.
Because the relative proportions of atmospheric carbon and noncarbon PM
can vary greatly from site to site or from one time to another at the same
site, the same absolute BS reflectance reading can be associated with very
different amounts (or mass) of collected particles or even with very different
amounts of carbon. Site-specific calibrations of reflectance readings against
actual mass measurements from collocated gravimetric monitoring devices are
therefore mandatory in order to obtain credible estimates of atmospheric
concentrations of particulate matter based on the BS method. A single calibra-
tion curve relating mass or atmospheric concentration (in ug/m ) of particulate
1-7
-------
matter to BS reflectance readings obtained at a given site may serve as a basis
for crude estimates of the levels of PM (mainly particles <10 (jm) at that site
over time, so long as the chemical composition and relative proportions of
elemental carbon and noncarbon PM do not change. However, the actual mass or
smoke concentration at a given site may differ markedly from values calculated
from a given reflectance reading on either of the two most widely used standard
curves (the British and OECD standard smoke curves). Thus, much care must be
taken in interpreting the meaning of any BS value reported in terms of ug/m3,
and such "nominal" expressions of airborne particle concentrations are not
meaningful unless related to direct determinations of mass by gravimetric
measurements carried out at the same geographical location and close in time to
the BS readings.
The AISI light transmittance method is similar in approach to the BS
technique, collects particles with a D5Q cutpoint S5.0 urn aerodynamic diameter,
uses an air intake similar to that of the BS method, and has been used for
routine monitoring in some American cities. Particles are collected on a
filter-paper tape periodically advanced to allow accumulation of another stain,
opacity of the stain is determined by transmittance of light through the
deposited material and tape, and results are expressed in terms of optical
density or coefficient of haze (CoH) units per 1000 linear feet of air sampled
(rather than mass units). Readings of COH units are more responsive to non-
carbon particles than are BS measurements, but again, the AISI method does not
directly measure mass or determine chemical composition of collected particles.
Attempts to relate COH to ug/m also require site-specific calibration of COH
readings against mass measurements determined by a collocated gravimetric
device, but the accuracy of such mass estimates are subject to question.
Since the hi-vol method collects particles much larger than those collec-
ted by BS or AISI methods, intercomparisons of PM measurements by the BS or
AISI methods to equivalent TSP units, or vice versa, are very limited. For
example, as shown by several studies, no consistent relationship exists between
BS and TSP measurements taken at various sites or at the same site during
various seasons. One exception is the relationship observed between BS and TSP
during severe London air pollution episodes when low wind-speed conditions
caused settling out of larger coarse-mode particles. Because fine-mode particles
3
predominated, TSP and BS levels (in excess of ~500 ug/m ) tended to converge,
as expected if mainly fine-mode particles were present.
1-8
-------
Another technique for determining the mass of PM collected on a filter
without weighing is to quantitatively measure the attenuation by the PM sample
of beta rays from a low-energy radioactive beta source. This method is very
close to a true mass measurement and correlates highly with gravimetric mass
concentration determinations (Lilienfeld, 1979). PM samples may be measured
rapidly in the laboratory, or instruments incorporating both the sampling appa-
ratus and the beta source and detector are available for automatic, on-site mea-
surements. Sample periods typically range from 0.5 to 2 hours, and quasi-con-
tinuous PM monitoring can be achieved in the instrument by configuring the beta
detector to monitor the particle collection area of the filter as the particles
accumulate. Various particle size-selective inlets can be used with a beta
attenuation instrument to effect size-limited PM measurements. The technique
.has generally good precision, but it is subject to errors from detector drift
and absorption of moisture from the atmosphere by the filter material or the
collected PM (Lilienfeld, 1985).
The piezoelectric microbalance is a device that measures PM continuously
by electronically measuring the change in the resonant frequency of a quartz
crystal as PM is deposited on its surface, either by impaction or electrostatic
precipitation. Although very sensitive, this technique is subject to measure-
ment error from imperfect adhesion of particles to the crystal, moisture and
temperature dependence, and sensitivity to certain pollutant gases (Lilienfeld,
1985). In addition, the dynamic range of the technique is limited, and the
crystal must be cleaned frequently.
Another resonant frequency technique is the tapered element oscillating
microbalance (TEOM), which continuously measures the mass of the PM collected
on a filter mounted on the end of a cantilevered element by electronically
measuring the change in the element's resonant frequency of oscillation.
Because the PM is collected on a filter, this method is more rugged than the
piezoelectric microbalance and compares more favorably with other filtration
methods (Lilienfeld, 1985). However, it is less sensitive, suffers from the
same potential interference from absorption of moisture by the filter and col-
lected PM, and requires that the filter be changed periodically.
The integrating nephelometer, an optical instrument which measures light
scattered by suspended particles, can be used to continuously measure ambient
concentrations of PM without collection, but such measurements are only indi-
rectly related to the mass concentration of the particles. Light scattering
1-9
-------
varies with particles size and is maximum in the 0.3 to 0.8 urn (accumulation
mode) size range (Charlson et al.s 1978). Thus, nephelometer measurements are
most useful for fine particles and visibility monitoring and correlate poorly
with broad-size-range PM measurements unless the ambient particle size distri-
bution is dominated by fine particles. Heating the sampled air to reduce its
relative humidity is necessary to minimize the effect of high humidity on
particle size.
Many analytical techniques are available to determine chemical properties
of particles collected on a suitable substrate. Most of the techniques, such
as those for elemental sulfur, have been shown to be more precise than the
analyses for gravimetric mass concentration. Methods are available that pro-
vide reliable analyses for sulfates, nitrates, organic fractions, and elemental
composition (e.g., sulfur, lead, silicon), but not all analyses can be used for
all particle samples because of factors such as incompatible substrates or in-
adequate sample size. Results can be misinterpreted when samples have not been
appropriately segregated by particle size and when artifact mass is formed on
the substrate rather than collected in particulate form, e.g., positive arti-
facts likely in nitrate and sulfate determinations (as noted below).
1.2 PHYSICAL/CHEMICAL PROPERTIES OF SULFUR OXIDES AND THEIR TRANSFORMATION
PRODUCTS AND AMBIENT MEASUREMENT METHODS
The only sulfur oxide that occurs at significant concentrations in the
atmosphere is sulfur dioxide, one of the four known gas-phase sulfur oxides
(sulfur monoxide, sulfur dioxide, sulfur trioxide, and disulfur monoxide). As
discussed in U.S. EPA (1982a), sulfur dioxide is a colorless gas detectable by
taste at levels of 1000 to 3000 ug/m3 (0.35-1.05 ppm). Above 10,000 ug/m3 (3.5
ppm), it has a pungent irritating odor.
As also discussed in U.S. EPA (1982a), S02 is mainly removed from the
atmosphere by gaseous, aqueous, and surface oxidation to form acidic sulfates.
Gas-phase oxidation of S02 by the hydroxyl (OH) radical is well understood; not
so well understood, however, is oxidation of S02 by hydroperoxyl (H02) and
methyl peroxyl (CH..00) radicals. The ready solubility of SO, in water is due
3 £. ?
mainly to formation of bisulfite (HSOg-) and sulfite (S03 -) ions, which are
easily oxidized to form acidic sulfates by reacting with catalytic metal ions
and dissolved oxidants. Sulfur dioxide reacts on the surface of a variety of
1-10
-------
airborne solid particles, such as ferric oxide, lead dioxide, aluminum oxide,
salt, and charcoal.
Sulfur trioxide (SO,), which can be emitted into the air directly or
result from reactions mentioned earlier, is, a highly reactive gas. In the
presence of moisture in the air, it is rapidly hydrated to form sulfuric acid.
In the air, then, it is sulfuric acid in the form of an aerosol that is found
rather than SCL, and it is generally associated with other pollutants in
droplets or solid particles of widely varying sizes. The acid is strongly
hygroscopic, and droplets containing it readily take up further moisture from
the air until they are in equilibrium with their surroundings. If any ammonia
is present, it reacts with sulfuric acid to form various ammonium sulfates,
which continue to exist as an aerosol (in droplet or crystalline form, depend-
ing on the relative humidity).
The sulfuric acid may also react further with other compounds in the air
to produce other sulfates. Some sulfates reach the air directly from combus-
tion or industrial sources, and near oceans, sulfates exist in aerosols gene-
rated from ocean spray. As discussed in U.S. EPA (1982a), sulfate particles
fall mainly in the fine-mode (<2.5 urn) size range. These particles, in the
presence of moisture in air, combine with water to form coarse-mode aerosols
(i.e. , >2.5 urn).
Many sulfur compounds are present in the complex mixture of urban air
pollutants. Some are naturally occurring and some are manmade. Total biogenic
sulfur emissions in the United States have been estimated to be in the range of
5 to 6 million metric tons annually. Additional contributions from coastal and
oceanic sources may also be significant. Anthropogenic (manmade) sources are
estimated to emit about 26 to 27 million metric tons of SO (mostly S0?)
.A. £-
annually in the United States. Most manmade sulfur oxide emissions are from
stationary point sources; over 90 percent of these are S02 and the rest are
sulfates.
Once S0? is emitted into the lower atmosphere, maintenance of a tolerable
environment depends on the ability of wind and turbulence to disperse the
pollutants. Factors affecting the dispersion of S02 from combustion sources
include (1) temperature and efflux velocity of the gases, (2) stack height, (3)
topography and the proximity of other buildings, and (4) meteorology. Some of
the SOp emitted into the air is removed unchanged onto various surfaces,
including soil, water, grass and vegetation. The remaining S02 is transformed
1-11
-------
into sulfuric acid or other sulfates by various processes in the presence of
moisture, and these transformation products are then removed by dry deposition
processes or by precipitation. The relative proportion of SOp and its trans-
formation products resulting from atmospheric processes varies with increasing
distance from emission sources and residence time (age) in the atmosphere.
With long-range transport (over hundreds or thousands of kilometers), extensive
transformation of S02 to sulfates occurs, with dry deposition of acidic sulfates
or their wet depositon in rain or snow contributing to acidic precipitation
processes.
The most commonly used collection and measurement methods for sulfur
oxides were described in the 1982 EPA criteria document (U.S. EPA, 1982a). A
clear understanding of the underlying bases and limitations of particular
methods is essential for adequate interpretation of epidemiological studies
discussed later. If SOp were the only contaminant in air, all measurement
methods for that gas would give comparable results, indicating the true concen-
tration of SOp. In typical urban environments, however, other pollutants are
always present and although sampling procedures can be arranged to minimize
interference from particulate matter by first filtering the air, errors still
arise due to other gases and vapors. Thus, variations in specificity and
accuracy of methods must be taken into account in comparing results from
various studies.
Methods for measurement of SOp include (1) manual methods, which involve
collection of the sample over a specified time period and subsequent analysis
by a variety of analytical techniques, and (2) automated methods, in which
sample collection and analysis are performed continuously and automatically.
In the most commonly used manual methods, the analyses of the collected samples
are based on colorimetric, titrimetric, turbidimetric, gravimetric, x-ray fluo-
rescent, chemiluminescent, and ion exchange chromatographic measurement prin-
ciples.
The most widely used manual method for determination of atmospheric SCL is
the West-Gaeke pararosaniline method. An improved version of this colorimetric
method, adopted in 1971 as the U.S. EPA reference method, can measure ambient
3
SOp at levels as low as 25 pg/m (0.01 ppm) with 30 min to 24 hr sampling time.
The method has acceptable specificity for SOp, if properly implemented; how-
ever, samples collected in tetrachloromercurate(II) can undergo temperature-
dependent decay leading to the underestimation of ambient SOp concentrations.
1-12
-------
A variation of the method uses a buffered formaldehyde solution for sample
collection, reducing the temperature-dependent decay problem. Certain American
epidemiological studies employed the West-Gaeke or other variations of the
pararosaniline method.
A titrimetric (acidometric) method, whereby SO,, is collected in dilute
hydrogen peroxide and the resultant H2$04 is titrated with standard alkali, is
the standard method mainly used in Great Britain and by the Organization for
Economic Cooperation and Development (OECD). The method requires long sampling
times (24 h), is subject to interference from atmospheric acids and bases, and
can be affected by errors due to evaporation of reagent during sampling,
titration errors, and alkaline contamination of glassware. It has been used to
provide aerometric S0? estimates reported in many British and European epidemi-
ological studies.
Some other methods use alkali-impregnated filter papers for collection of
S0? and subsequent analysis as sulfite or sulfate. Most involve extraction
prior to analysis; but nondispersive x-ray fluorescence allows direct measure-
ment of SOp collected on sodium carbonate-impregnated membrane filters. These
methods have not been widely used for routine air monitoring or epidemiological
studies.
Two of the most sensitive methods for measuring S02 are based on chemilu-
minescence and ion exchange chromatography. With the former, S02 is absorbed
in a tetrachloromercurate solution and then oxidized with potassium permanga-
nate; oxidation of the absorbed SO^ is accompanied by chemiluminescence de-
tected by a photomultiplier tube. With the latter, ion exchange chromatography
can be used to determine ambient levels of S02 absorbed into dilute hydrogen
peroxide and oxidized to sulfate, or S02 absorbed into a buffered formaldehyde
reagent. These methods have not yet been widely employed for routine monitor-
ing uses.
Sulfation methods, based on reaction of airborne sulfur compounds with
lead dioxide paste to form lead sulfate, have been used both in the United
States and Europe to estimate ambient SCL concentrations over extended time
periods. However, data obtained by sulfation methods are affected by many
physical and chemical variables and other interferences (such as wind speed,
temperature, and humidity); and they are not specific for S02, since sulfation
rates are als'o affected by other airborne sulfur compounds (e.g., as sulfates).
Thus, although sulfation rates (mg S03/100 cm2/day) have been converted to
1-13
-------
rough estimates of S02 levels (in ppm), these cannot be accepted as accurate
measurements of atmospheric S02 levels. This is notable here because lead
dioxide gauges provided estimates of S02 data used in some pre-1960s British
epidemiological studies and also in some American epidemiologic studies.
Automated methods for measuring ambient S02 levels have been widely used
for air monitoring. Some early continuous S02 analyzers, based on conductivity
and coulometry, were subject to interference by many ambient air substances.
More recent commercially available analyzers using these measurement principles
exhibit improved specificity for S02 through incorporation of sophisticated
chemical and physical scrubbers.
Continuous S02 analyzers that use flame photometric detection (FPD),
fluorescence, or second-derivative spectrometry are now commercially available.
The FPD method involves measurement of the band emission of excited SCL
molecules formed from sulfur species in a hydrogen-rich flame and can exhibit
high sensitivity and fast response, but must be used with selective scrubbers
or coupled with gas chromatographs to achieve high specificity. Fluorescence
analyzers detect characteristic fluorescence of the S02 molecule when irra-
diated by UV light, have acceptable sensitivity and response times, are in-
sensitive to sample flow rate, and require no support gases. However, they can
be affected by interference due to water vapor (quenching effects) and certain
aromatic hydrocarbons and must employ ways to .minimize such effects. Second-
derivative spectrometry can provide highly specific measurement of S02 in the
air, with continuous analyzers based on this principle being insensitive to
sample flow rate and requiring no support gases. U.S. EPA has designated con-
tinuous analyzers based on many of the above principles (conductivity, coulome-
try, flame photometry, fluorescence, and second-derivative spectrometry) as
equivalent methods for measurement of atmospheric S02-
Two main methods have been used to measure total water-soluble sulfates
collected on filters along with other suspended particulate matter. With the
turbidimetric method, samples are collected on sulfate-free glass fiber or
other efficient filters, the sulfate is extracted and precipitated with barium
chloride, and the turbidity of the suspension is measured spectrophotometrically.
Samples are normally collected over 24-h periods by hi-vol sampler. However,
no distinction can be made between sulfates and sulfuric acid present in the
air and collected on the filters; and some material present as acid in the air
may be converted to neutral sulfate on the filter during sampling. With the
1-14
-------
methylthymol blue method, samples are collected as in the turbidimetric method
and the extract is reacted with barium chloride, but the barium remaining in
solution is then reacted with methylthymol blue and the sulfate determined
colorimetrically by measurement of uncomplexed methylthymol blue. This modifi-
cation allows the procedure to be automated, but the same limitations as noted
for the turbidimetric method apply, including lack of distinction between
sulfates and sulfuric acid.
As for sulfuric acid, no fully satisfactory method exists for its measure-
ment in the presence of other pollutants in the air, but some procedures exist
for examining acidic properties of suspended particles or acid aerosols in
general. Almost all of the strong acid content of ambient aerosols consists of
sulfuric acid (H?SO.) and its partial atmospheric neutralization product,
ammonium bisulfate (NH.HSO^); however, ammonium sulfate [(NH^SO^], the final
neutralization product, is only weakly acidic. Nitric acid (HN03) and hydro-
chloric acid (HC1) are other strong acids found in the ambient air (mainly as
vapors or, when incorporated into fog droplets, as constituents of acid
aerosols). Ambient air acidic aerosol concentrations can be expressed in terms
of jjmols H+/m3 or as H2$04 equivalent in ug/m3 (at 98 ug/umol). Unfortunately,
no systematic surveys of average acid aerosol concentrations in United States
airsheds were available at the time the 1982 EPA criteria document (U.S. EPA,
1982a) was prepared, nor is such systematic survey information available for
more current acidic aerosol levels. However, Lioy and Lippmann (1986) have
recently summarized some of the highest levels reported for recent years in
3
North America, including levels in the range of 20 to 30 ug/m H2$04 (1 hr
mean). This is in contrast to the highest level (680 ug/m3 H2S04 1 hr mean)
recorded in the United Kingdom in London in 1962 and even higher levels almost
certainly present during earlier London air pollution episodes.
1.3 KEY AREAS ADDRESSED IN EMERGING NEW HEALTH EFFECTS DATA
Important new health effects information has emerged in three main areas
since preparation of the 1982 EPA criteria document and addendum: (1) new data
which permit more definitive characterization of respiratory tract deposition
patterns for inhaled particles of various size ranges, e.g., fine-mode (<2.5
urn) vs. larger coarse mode particles (>2.5 urn, <10 urn, <15 urn, etc.); (2) new
reanalyses of certain key British epidemiology studies, which used BS methods
1-15
-------
for measuring PM levels, and additional new epidemiologic studies, employing
other non-gravimetric or gravimetric PM measurement methods, that assess health
effects associated with exposures to PM and SO in contemporary urban airsheds
P\
of the 1970s and 1980s; and (3) new controlled human exposure studies which
more precisely define exposure-response relationships for pulmonary function
decrements and respiratory symptoms due to acute S09 exposure.
1-16
-------
CHAPTER 2. RESPIRATORY TRACT DEPOSITION AND FATE
2.1 RESPIRATORY TRACT DEPOSITION AND FATE OF INHALED AEROSOLS
As discussed in U.S. EPA (1982a), the respiratory system is the major
route of human exposure to airborne suspensions of particles (aerosols) and
gases such as S02. In inhalation toxicology, deposition refers to removal from
inspired air of inhaled particles or gases by the respiratory tract and the
initial regional pattern of these deposited materials. Clearance refers to
subsequent translocation (movement of material within the lung or to other
organs), transformation, and removal of deposited substances from the respira-
tory tract. It can also refer to removal of reaction products formed from SO-
or particles. Retention refers to the temporal pattern of uncleared deposited
particulate materials or gases and reaction products. These phenomena are
complicated by interactions that occur among particles, gases such as S02 or
endogenous ammonia, and water vapor in the airways.
Deposition patterns of inhaled aerosols and gases are affected by physical
and chemical properties, e.g., aerosol particulate size distribution, density,
shape, surface area, electrostatic charge, hygroscopicity or deliquescence,
chemical composition, gas diffusivity and solubility, and related reactions.
The geometry of the respiratory airways from nose and mouth to the lung
parenchyma also influences aerosol deposition; important morphological parame-
ters include diameters, lengths, inclinations to vertical, and branching angles
of airway segments. Physiological factors that affect deposition include
breathing patterns, respiratory tract airflow dynamics, and variations of
relative humidity and temperature in the airways. Clearance from the respira-
tory tract depends on many factors, including site of deposition, chemical
composition and properties of deposited particles, reaction products, muco-
ciliary transport in the tracheobronchial tree, macrophage phagocytosis, and
pulmonary lymph and blood flow. An understanding of respiratory tract anatomy
and regional deposition and clearance of particles is essential for interpreta-
tion of the results of health effects studies discussed later.
2-1
-------
The respiratory tract includes the passages of the nose, mouth, nasopharynx,
oropharynx, epiglottis, larynx, trachea, bronchi, bronchioles, and small ducts
and alveoli of the pulmonary acini. In regard to respiratory tract deposition
and clearance of inhaled aerosols, three main regions can be considered: (1)
the extrathoracic (ET) region, which includes the airways extending from the
nares down to the epiglottis and larynx at the entrance to the trachea (the
mouth is included in this region during mouth breathing); (2) the tracheo-
bronchial (TB) region, which includes the primary conducting airways of the
lung from the trachea to the terminal bronchioles (i.e., that portion of the
lower respiratory tract having a ciliated epithelium); and (3) the pulmonary
(P) region, which consists of the parenchymal airspaces of the lung, including
the respiratory bronchioles, alveolar ducts, alveolar sacs, atria, and alveoli
(i.e., the gas-exchange region). The extrathoracic region, as defined above,
corresponds exactly to the nasopharynx, as defined by the International Commis-
sion on Radiological Protection (ICRP) (Task Group on Lung Dynamics, 1966). The
thoracic region corresponds to that portion of the respiratory tract distal to,
and including, the trachea (i.e., TB + P).
As discussed in U.S. EPA (1982a), evaluation of mechanisms by which
inhaled particles ultimately affect human health requires recognition of the
importance of deposition and clearance phenomena in the respiratory tract.
Major regions of the respiratory tract differ markedly in structure, size,
function, and sensitivity or reactivity to deposited particles. They also have
different mechanisms for particle elimination or clearance.
The 1982 EPA criteria document depicted available experimental deposition
data for total and regional deposition in a series of figures (i.e., Figures
11-3 to 11-9 of U.S. EPA, 1982a). Curves for alveolar deposition and estimates
of tracheobronchial deposition, along with an extrapolation of the upper bound
of the TB curve to the point predicted by Miller et al. (1979), are reproduced
here in Figure 2. Added to the figure are the more recent data of Svartengren
(1986), Heyder (1986), and Emmett et al. (1982) for deposition of particles >10
urn in aerodynamic diameter (D_rt-) in healthy adult subjects breathing through a
36 ;,
mouthpiece. '.•-..
In the studies reported by Heyder (1986), mean inspiratory flow rates of
3 -1
250 and 750 cm s were used with a four-second breathing cycle, resulting in
minute ventilations of 7.5 and 22.5 L min , respectively. At the higher flow
rate, TB deposition of 10 urn D__ particles was 0.14; fractional deposition for
ac
2-2
-------
I—I I I I I
RANGE OF ALVEOLAR DEPOSITION,
MOUTH BREATH ING
ESTIMATE OF ALVEOLAR DEPOSITION, NOSE BREATHING
RANGE OF TRACHEOBRONCHIAL DEPOSITION
MOUTH BREATHING
EXTRAPOLATION OF ABOVE TO POINT ( « ) PREDICTED
BY MILLER et al., (1979)
m\
m\
\
-09 EMMETT et al. (1982); 337 cm3 s'1, 6s BREATHING CYCLE
HEYDER (1986); 750 ernes'], 4s BREATHING CYCLE
A A HEYDER (1986) ; 250 cm3 s'1, 4s BREATHING CYCLE
5 |-O* SVARTENGREN (1986)
OPEN SYMBOLS: TRACHEOBRONCHIAL DEPOSITION
SOLID SYMBOLS: ALVEOLAR DEPOSITION
10 121416 20
2.0 3.0 4.0 5.0
0.2 0.3 0.4 0.5
AERODYNAMIC DIAMETER,
PHYSICAL DIAMETER, jum
Figure 2. Regional deposition of monodisperse aerosols by indicated particle diameter for mouth
breathing (alveolar, tracheobronchial) and nose breathing (alveolar). The alveolar band indicates
the range of results found by different investigators using different subjects and flow parameters
for alveolar deposition following mouth breathing. Variability is also expected following nasal
inhalation. The tracheobronchial band indicates intersubject variability in deposition over the
range of sizes as measured by Chan and Lippmann (1980). Deposition is expressed as fraction
of particles entering the mouth (or nose). Also shown is an extrapolation of the upper bound of
the TB curve to the point predicted by Miller et al. (1979). The extrapolation illustrates the likely
shape of the curve in this size range but is uncertain. However, the data of Emmett et al. (1982),
Heyder (1986), and Svartengren (1986) tend to substantiate this extrapolation. In the Svartengren
(1986) studies, subjects took maximally deep inhalations at a flow of 500 cm3 s"'.
2-3
-------
12 pro Dae particles was 0.09. In contrast, the lower flow rate yielded deposi-
tion fractions of 0.17 and 0.12, respectively, for 10 urn and 12 urn D
36
particles. Emmett et al. (1982) observed an average TB deposition of 0.36 in
three subjects who inhaled 10 urn D particles at a mean inspiratory flow rate
3 ~1 i
of 337 cm s with 10 breaths/min (i.e., minute ventilation of 10.1 L min ).
Under these breathing conditions the alveolar region deposition fraction for 10
urn particles averaged 0.06.
The deposition of 11.5, 13.7, and 16.4 urn D particles Was studied by
36
Svartengren (1986) using a different exposure regime. Subjects took four
maximally deep inhalations at a flow of 500 cm s 1 from a glass bulb apparatus
each time particles were sprayed up into the bulb. Exposure times varied from
2 to 5 min. Six subjects were studied at the 11.5 and 13.7 urn sizes, while
five subjects were studied at 16.4 urn D . The average alveolar deposition
aG
fraction was 0.01 at the largest particulate size and 0.04 at the 11.5 and 13.7
|jm sizes. By subtracting alveolar deposition from the measured total lung
deposition, the average TB deposition fractions of the 11.5, 13.7, and 16.4 (jm
Dae Partl~cles were °-27> °-17» and 0.12, respectively. The data of Svartengren
(1986), along with the data of Heyder (1986) and Emmett dt al. (1982), tend to
substantiate the extrapolation of the upper bound of the TB curve in Figure 2
to the point predicted by Miller et al. (1979).
Numerous subject-related and environmental factors can influence deposi-
tion and clearance of aerosols, including inhalation patterns (rate and route),
airway dimensions in relation to pulmonary function measurements, disease
state, particle composition, and the presence- of pollutant gases. Detailed
discussion of effects of such factors on deposition patterns is beyond the
scope of this addendum (for more details, see U.S. EPA, 1982a,b; Lippmann et
al., 1980; Garrard et al., 1981; Svartengren et al., 1986; Lippmann and
Schlesinger, 1984). However, the results of Heyder et al. (1982) on the
biological variability of particulate deposition in controlled and spontaneous
mouth breathing are of interest since this was an important issue raised in the
1982 EPA criteria document. Using both breathing patterns and particulate
sizes ranging from 1 to 7|jm D , they studied total deposition and deposition
d.S
rate in 20 subjects. The variability of deposition rate between subjects
spontaneously breathing the same aerosol is associated with morphological and
physiological factors but is mainly governed by physiological factors (i.e.,
primarily individual flow rate). Heyder et al. (1982) contend that this type
2-4
-------
of variability is the most important when considering health-related issues of
inhaled particulate matter.
Data on respiratory tract deposition can be used to provide an evaluation
of deposition of typically observed ambient participate distributions. The
similarity of experimental deposition data from human subjects breathing
monodisperse aerosols in a laboratory setting to the general population breath-
ing multimodal urban aerosols was examined in studies published after prepara-
tion of the 1982 EPA criteria document (U.S. EPA, 1982a). Hiller et al. (1982)
studied total respiratory tract deposition in five subjects using a mixture of
monodisperse polystyrene latex spheres 0.6, 1, and 2 [jm in size. Their experi-
mental results suggest that the deposition of mixed monodisperse and monodis-
perse single aerosols is similar for fine particles. However, the theoretical
modeling of Diu and Yu (1983) indicate that the regional deposition patterns of
polydisperse aerosols can be quite complex. They assumed a log normal size
distribution and studied total and regional deposition with nasal and mouth
breathing for geometric standard deviations (a ) of 1.0 (monodisperse), 1.5,
2.5, and 3.5. The results of Diu and Yu (1983) are consistent-with the observa-
tion of Morrow (1984) that the mass deposition of mono- and polydisperse aero-
sols differs little if a <2. Typically, a values reported for distribution of
urban and rural aerosols is usually around 2 (see Chapter 5, U.S. EPA, 1982a).
In the theoretical studies of Diu and Yu (1983), larger values of a are pre-
dicted to impart significant complexities in regional deposition patterns due
to competing mechanisms interacting with the sequential filtering effect of the
respiratory tract.
Over half of the total mass of a typical ambient mass distribution would
be deposited in the extrathoracic region, most of this being coarse particles,
during normal nasal breathing (see Chapter 11 of U.S. EPA, 1982a). Clearance
of most of this material to the esophagus would occur within minutes. Some
fraction of the hygroscopic fine mass (e.g., sulfates and nitrates that grow to
2-4 (jm in the respiratory tract) might also be deposited and dissolve in the
extrathoracic region. Smaller fractions of both the hygroscopic and non-
hygroscopic fine particles (mostly <1 pm) would be deposited in the tracheo-
bronchial and alveolar regions, respectively. Clearance of hygroscopic
material by dissolution and reaction would be relatively rapid in both regions.
Clearance of insoluble coarse-mode substances would increase from less than an
2-5
-------
hour for the larger particles deposited in the upper portion of the tracheo-
bronchial region to as much as a day for that deposited more distally.
Insoluble fine and coarse particles deposited in the alveolar region have
clearance half-times varying from weeks to years for the fast phase and slow
phase, respectively.
With mouth-only breathing, the regional deposition pattern changes mark-
edly, with extrathoracic deposition reduced and both tracheobronchial and pul-
monary deposition enhanced. Extrathoracic deposition, although reduced, still
would be dominated by coarse mode aerosols and contain little fine-mode contri-
bution. Endogenous ammonia in human airways may, however, reduce the deposi-
tion of acid aerosols (U.S. EPA, 1982b). Remaining non-hygroscopic fine
particle deposition efficiency would change little over nasal breathing (<20
percent).
In essence, regional deposition of ambient particles in the respiratory
tract does not occur at divisions clearly corresponding to atmospheric aerosol
distributions. Coarse-mode and hygroscopic fine-mode particles are deposited
in all three regions. A fraction (5 to 25 percent) of the remaining fine-mode
particles (e.g., organics and carbon not associated with hygroscopic material)
is deposited in the.,tracheobronchial/alveolar regions. With mouth-only breath-
ing, as illustrated in Figure 2, little particulate mass in excess of 15 pm is
deposited in the thoracic region, and little mass greater than 10 pm is
deposited in the alveolar region.
Oronasal breathing (partly via the mouth and partly nasally) typically
occurs for healthy adults while undergoing moderate to heavy exercise. Swift
and Proctor (1982) computed deposition for oronasal breathing as a function of
particulate size, correcting for deposition in the parallel nasal and oral
airways, and compared these results to those for mouth breathing via tube.
Using minute ventilations of 24.5 and 15 Lmin , their analyses predicted that
total thoracic deposition at all sizes is more or less essentially the same as
for pulmonary deposition noted above for mouth only breathing, i.e., with very
few particles over 10 urn D in size being likely to reach tracheobronchial
ae
regions. Tracheobronchial deposition with oronasal breathing at a higher
minute ventilation (45 Lmin ) has been examined by Miller et al. (1984). .Data
for extrathoracic and tracheobronchial deposition were fit to logistic regres-
sion models yielding significantly improved fits of the deposition data. As
done by Swift and Proctor (1982), a 50/50 split in airflow between the nasal
2-6
-------
and oral pathways was assumed. Simulated oronasal breathing at a minute
ventilation of 45 Lmin resulted in tracheobronchial deposition fractions of
0.21, 0.17, 0.14 and 0.09 for particles of 8, 9, 10, and 12 urn in aerodynamic
diameter, respectively. When the experimental deposition data of Heyder
(1986), separately for nasal and oral breathing, are combined to simulate
oronasal breathing, the results are in agreement with the analyses of Miller et
al. (1984). Bowes and Swift (1986) studied mouth deposition during natural oro-
nasal breathing and found that 58% of 10 urn particles and 78% of 15 urn
particles deposited in the mouth. Nasal deposition efficiencies were, however,
not measured.
More recently, thoracic deposition and its component parts have been
examined by Miller et al. (1986), as a function of particulate size, for
ventilation rates ranging from normal respiration to heavy exercise in individ-
uals who, as per Niinimaa et al. (1981), habitually breathe oronasally (mouth
breathers) and in those who normally employ oronasal breathing when minute
ventilations exceed about 35 Lmin (normal augmenters). Published data from
various laboratories for ET and TB deposition, along with previously unpub-
lished data of Lippmann and co-workers at New York University, were fit to
logistic regression models prior to examining the influences of breathing mode
and activity level on TB, P, and thoracic (TB + P) deposition. For the ET
region, an impaction parameter was used that was a function of aerodynamic
diameter and inspiratory flow rate, and the logistic models provided signifi-
cantly improved fits of the nasal and oral inspiration data compared to the
linear models of Yu et al. (1981) that also used an impaction parameter and
that formed the basis of the Swift and Proctor (1982) analyses. Since TB
deposition is due primarily to inertia! impaction in the. upper airways and to
sedimentation in the lower airways, the logistic analysis for the TB region was
based upon aerodynamic diameter rather than on an impaction parameter. The
proportionality of airflow between the nose and mouth as a function of activity
level was determined from Figure 2 of Niinimaa et al. (1981).
Thoracic deposition results given by Miller et al. (1986) are shown in
Figure 3, along with the thoracic deposition results of Swift and Proctor
(1982). With minute ventilations (V£) of 40 or 60 Lmin'1 (panel A), there is
not much difference between normal augmenters and mouth breathers in thoracic
deposition for D beyond the peak of the deposition curve. For V£ less than
35 Lmin'1, the Miller et al. (1986) analyses result in substantially lower
deposition in normal augmenters compared to mouth breathers. As VE increases,
2-7
-------
thoracic deposition for normal augmenters initially decreases for a given D ,
clG
increases through the oronasal switching point, and then decreases. For mouth
breathers, however, there are minimal changes in thoracic deposition at lower
ventilation rates with monotonic declines in deposition as VV increases beyond
30 Lmin"1.
Swift and Proctor (1982) computed bands of total thoracic deposition as a
fraction of particles entering the mouth and nose during oronasal and oral
breathing, using V£ of approximately 24.5 Lmin" and 15 Lmin"1, respectively.
The shaded area of Panel B (Figure 3) represents a composite of these data
based on the lower band of the low V£ and the upper band of the higher Vp.
While neither Swift and Proctor (1982) nor the U.S. EPA (1982a,b) extended the
bands for TB deposition beyond 8 |jm, some thoracic deposition could be projec-
ted for 10 to 15 pg particles with oronasal breathing. More recent experi-
mental data utilized in Miller et al. (1986) indicate that there is a gradual
decline in thoracic deposition for large particulate sizes and that there can
be significant deposition of particles greater than 10 pm, particularly for
mouth breathers.
It should be noted that the deposition studies cited previously all used
adult subjects, yet many of the epidemiology studies cited in the PM/SO
/\
criteria document (U.S. EPA, 1982a) and in this addendum report effects
observed in children. Anatomical and functional differences between adults and
children are likely to yield complex interactions with the major mechanisms
affecting respiratory tract deposition. In a study of over 1800 Mexican-
American, white, and black children 7 to 20 years of age, Hsu et al. (1979)
found significant differences of lung volume and flow rate among the three
races, and between male and female subjects. Further analyses of these data by
Hsi et al. (1983) demonstrated that using sitting height as a predictor greatly
reduced the racial differences of ventilatory function and allowed the applica-
tion of a single set of prediction equations for children of all three groups.
Other studies are available on normal pulmonary function values (de Swiniarski
et al., 1982), intrasubject variability (Hutchinson et.al., 1981), influence of
physical performance capacity on the growth of lung volumes (Anderson et al.,
1984), and postnatal growth and size of the pulmonary acinus (Osborne et al.,
1983).
To date, experimental deposition data in children's lungs are not avail-
able. Analogous to the development of mathematical models for deposition in
adults, the thrust for age-dependent dosimetry modeling has been from
2-8
-------
0.5
0.4
0.3
a.
UJ
a
o
£0.2
DC
O
I
(-
0.1
1.0
i i i i mil—i i i i iiuj
MOUTH BREATHERS
NORMAL AUGMENTERS —
I I I 1 I I Ml
Itt "I
i ilium
\\
I I I I I Mil
10.0 100 1.0
AERODYNAMIC DIAMETER, //m
10.0
J±
100
Figure 3. Estimates of thoracic deposition of particles between 1 and 15 urn by Miller et al. (1986)
for norrnal augmenters (solid lines) and mouth breathers (broken lines) are shown for minute venti-
lation (VE) exceeding the switch point of 35 L min'1 (A) and for lower VE (B). Normal augmenters
are individuals who normally use oronasal breathing to augment respiratory airflow when VE exceeds
about 35 L min'1, while mouth breather refers to those individuals who habitually breathe oronasally
(Niinimaa et al., 1981). The shaded area (B) is a composite of the.computed bands of thoracic depo-
sition of particles less than 8 jum by Swift and Proctor (1982) for VE of approximately 24.6 and 15
L min'1.
scientists dealing primarily with radiological protection issues (Hofmann et
al., 1979; Hofmann, 1982a,b; Crawford, 1982). More recently, Phalen et al.
(1985) have studied the postnatal enlargement of human tracheobronchial airways
and its implication for the deposition of particles ranging from 0.05 to 10 urn
in size. They made some morphometric measurements in replica lung casts of
people aged 11 days to 21 years. The model predictions for deposition during
inspiration only were computed for three states of physical exertion — low
activity, light exertion, and heavy exertion. Scaling techniques were employed
to make age-dependent adjustments from adult flow rates.
While the predictions of Phalen et al. (1985) indicate that, in general,
increasing age is associated with decreasing particulate deposition efficiency,
high flow rates and large particulate sizes do not exhibit consistent age-
dependent differences. Since VV at a given state of activity is approximately
linearly related to body mass, children will inhale more air per unit body
2-9
-------
mass, resulting in higher TB doses. For resting ventilation, this age-related
dose effect, as a function of particulate size, is illustrated in Figure 4.
Xu and Yu (1986) also computed particle deposition efficiencies as a function
of age utilizing the growth models of Hoffman et al. (1979, 1982a,b). They
take into account the age dependence of head deposition in their calculations.
In contrast to the predictions of Phalen, et al (1985), a peak in TB deposition
efficency at about 2 years is predicted. However, when divided by body weight,
the TB deposition rate would show an age dependence similar to Phalen et al.
(1985).
While children may be at greater risk than adults from exposure to particu-
late matter on the basis of deposition during inspiration, information is needed
on possible age-dependent differences in ET deposition, deposition over the en-
tire breathing cyc\e, mucociliary clearance, and tissue sensitivity, in order
to place this risk into perspective relative to health effects evaluations.
Other deposition characteristics of individuals and atmospheric distribu-
tions (as well as other factors) can cause variations in regional deposition.
The following examples illustrate potentially important variations in exposure/
deposition patterns:
(1) The peak in alveolar deposition efficiency for nasal and mouth-only
breathing (Figure 2) tends to occur at or near the normal minimum in the
bimodal distribution (2 to 4 urn MMAD). However, near emission sources or in
other polluted conditions, substantial increases can occur in the coarse- or
fine-mode contribution to this most efficiently deposited range.
(2) The deposition of both coarse and fine particles in the tracheobron-
chial region can be increased over normal ranges by increased breathing rates
during exercise and by cigarette smoking, in both bronchi tic and asthmatic
subjects, generally reducing alveolar deposition. Since retention of particles
at 24 hr was significantly lower when bronchoconstriction was induced before
inhalation of particles than when bronchoconstriction was induced after inhala-
tion, Svartengren et al. (1984) postulated that bronchoconstriction may serve
as a defense mechanism for the alveolar region. However, enhanced tracheo-
bronchial deposition may not be protective, especially for disease states
(e.g., bronchitis) or other conditions that constrict, inflame, or cause mucous
build-up in airways. Further complicating our understanding of lung clearance
mechanisms in obstructive airways disease is the variety of mucociliary trans-
port patterns that can be observed, including regurgitation, stasis, spiral
motion, and movement toward the opposite bronchus (Isawa et al., 1984).
2-10
-------
1.0
AGE - YEARS
8 10 12
14
16 18
0.001
10
20 30 40
BODY MASS -KG
50
60
Figure 4. Predicted initial dose to the TB region as a
function of body mass. Assumptions include equivalent
upper airway deposition for all ages, inhalation of
particles at 1 mg/m3 concentration in air, and resting
minute ventilation.
Source: From Phalen et al. (1985).
2-11
-------
(3) Regional mass deposition data do not provide insights regarding local-
ized "hot spot" deposition. Significantly higher participate mass to lung
surface ratios can occur in the extrathoracic and tracheobronchial regions as
compared to the alveolar region. Gerrity et al. (1979) computed the average
particle surface concentration of an inhaled 8 urn MMAD aerosols in each genera-
tion of the Weibel lung model (Weibel, 1963) and predicted as much as two
orders of magnitude difference between particulate surface concentration in the
segmental bronchi compared to terminal bronchioles. Local surface concentra-
tions of deposited particles within large airways are probably higher than the
average. Also, respiratory disease states that result in altered breathing
patterns (e.g., increased oral breathing) may lead to increased deposition of
particles in particular respiratory tract regions.
(4) Although the probability of deposition of particles larger than 10 [jm
in the alveolar region is low, small numbers of such particles have been found
in human lungs (U.S. EPA, 1982a,b). Some evidence suggests that those large
insoluble coarse substances that do penetrate may be cleared at a much slower
rate. Animal tests indicate that 15 urn particles instilled in this region
clear much more slowly than smaller particles of the same composition (U.S.
EPA, 1982a,b).
Besides variations in regional deposition patterns found for inhaled
particles and factors affecting typical deposition patterns, regional differ-
ences exist for clearance mechanisms by which inhaled particles penetrating
various levels of the respiratory tract are removed. The effects of inhaled
particulate matter and other noxious agents, e.g., irritative gases, on clear-
ance mechanisms represent one of the major categories of toxic actions exerted
by such air pollutants. Detailed reviews of clearance mechanisms and effects
on them due to inhaled particles and sulfur dioxide (S02) are presented else-
where (U.S. EPA, 1982a,b; Lippmann et al. , 1982; Lippmann and Schlesinger,
1984).
Mucociliary clearance and alveolar clearance mechanisms are of most
concern here. Lung mucociliary clearance is the major defense mechanism by
which inhaled particles deposited in the tracheobronchial airways are removed
from the respiratory tract. Particle-laden mucus is transported by the tips of
cilia which are immersed in an aqueous sol layer. Airway mucus transport rates
decrease distally from the trachea (Asmundsson and Kilburn, 1970; Foster et
al., 1980) with particle residence times of potentially as much as 300 minutes
2-12
-------
in the terminal bronchioles (Lee et al. , 1979). Mucociliary clearance half-
times of the healthy lung can range typically between 30 minutes and several
hours, depending on the initial distribution of particles and mucus transport
rates within each airway. Lung mucociliary clearance can be impaired by
disease states of the lungs (Lippman et al. , 1980). Svartengren et al. (1986)
have observed marked dysfunction of lung mucociliary clearance in patients
with bronchiectasis. Influenza A and respiratory syncitial virus infection
cause a decrease in lung mucociliary clearance (Camner et al., 1973; Levandowski
et al., 1985; Garrard et al. , 1985) and a virtual halt in trachea! mucus trans-
port (Levandoswki et al. , 1985) unless supplemented by cough. Retarded mucus
transport within the lungs can lead to increased residence times of inhaled
particles.
Two general types of alveolar clearance mechanisms are generally recog-
nized: absorptive and non-absorptive. Absorptive mechanisms involve active
and passive transport processes, whereby deposited particles permeate the
alveolar epithelium and penetration of endothelial barriers occurs prior to
uptake into the blood or lymphatic transport. These processes are most effec-
tive in removing highly soluble particles. Phagocytosis of deposited particles
by alveolar macrophages is generally accepted as the chief non-absorptive
clearance process. Some Tow-solubility materials may escape phagocytosis and
accumulate as focal deposits within parenchymal tissues. -In the International
Commission on Radiological Protection (1979) lung model it has been suggested
that as much as 40 percent of particles deposited in alveoli migrate, either
free or phagocytized, to the distal portions of the ciliated airways for subse-
quent removal by mucociliary clearance. Alveolar clearance rates depend in
large part on particle solubility. Several studies of long-term clearance of
highly insoluble particles in the 1- to 4-|jm range (Bailey et al. , 1982; Bohning
et al. , 1982; Philipson et al., 1985) report two phases with half times of ap-
proximately 20 and 300 days, though Philipson et al. (1985) observed slow half-
times of as much as 2500 days. Stahlhofen et al. (1980) measured the long-term
clearance of ferric oxide particles (moderately insoluble) between 1 and 9 urn
MMAD and found single phase clearance half-times of between 70 days for the
smaller particles and 110 days for the larger ones.
Continuous exposures to ambient aerosols result in the simultaneous
deposition and redistribution of particles. The regional dose of particles
inhaled continuously may thus differ significantly from the regional pattern of
2-13
-------
acute aerosol deposition. Brain and Valberg (1974) developed a model of
retention of continuously inhaled particles based on the ICRP (Task Group on
Lung Dynamics, 1966) lung model. Gerrity et al. (1983) further refined it to
the Weibel (1963) lung model, taking into account individual airway mucus trans-
port rates. The Gerrity et al. (1983) model predicts maximum doses to the tra-
chea and respiratory bronchioles for a moderately insoluble 10-|jm aerosol.
Deposition of inhaled sulfate compounds in the respiratory tract is
complex and depends upon breathing patterns and physical properties of the
inhaled particles. Deposition patterns and clearance mechanisms for sulfates
depend upon their particular size ranges (mainly fine particles <2.5 pm) as
discussed above. Of most importance is the fact that deeper penetration of
particles into the respiratory tract occurs during breathing through the mouth
or oronasally than during nasal breathing.
Of particular concern from a health standpoint is the fact that acidic
aerosols exist in ambient air mainly in the size range of 0.3 to 0.6 urn (MMAD),
well within the range of readily inhalable fine-mode particles capable of pene-
trating deeply into tracheobronchial and alveolar regions of the respiratory
tract. Under fog conditions, where acidic components are often incorporated
into water droplets of larger sizes up to 10-15 urn, concern exists in regard to
the potential for health effects being associated with the increased deposition
of acidic fog droplets in the tracheobronchial regions of the respiratory
tract.
2.2 SULFUR DIOXIDE UPTAKE AND FATE
As discussed in U.S. EPA (1982a,c), sulfur dioxide is soluble in water and
readily absorbed upon contact with the moist surfaces of the nose and upper
respiratory passages. It is well established that the gas is almost completely
removed (95 to 99 percent) by nasal absorption under resting conditions in both
man and laboratory animals. A recent study by Schachter and coworkers (1986)
also indicates similar, almost complete, removal of SO,, in nasal passages during
nasal breathing under increased exercise conditions. Schachter et al. (1986)
3
exposed six subjects to 2.62 mg S0?/m (1 ppm) in an environmental chamber to
study nasal absorption of inhaled SOp. A 6 min rest was followed by 4 to 6 min
of exercise at 450 kpm during which subjects breathed only via the nose. A
catheter was placed in the oral cavity and connected to an SO^ analyzer. No
2-14
-------
detectable quantities of S0? could be measured when sampling from the mouth.
In addition, saliva samples were analyzed for dissolved S0?; no dissolved S0?
was detected. These results confirm previous observations that the nose is
extremely efficient in removing SO^.
Other human studies indicate that S02 penetration to the lower respiratory
tract increases with activity and increased ventilation associated with a shift
-1
from nasal to oronasal breathing at a mean VF of 30 L min (Niinimaa et al. ,
1980, 1981; D'Alfonso, 1980). Most studies on the deposition of S00 in animals
4j
and humans have been done at concentrations greater than 2.62 mg/m (1 ppm).
The 95 to 99 percent removal of S0? by the upper respiratory tract has not been
confirmed at levels ordinarily found in ambient air (generally less than 0.1
3
mg/m [0.038 ppm]). It is expected, however, that similar deposition patterns
would be observed at these lower concentrations of S02- Once inhaled, S02 is
absorbed quickly into the mucus layer lining the ET and TB regions, where
reactions can occur which might result in alterations in the viscosity of
mucus. Absorbed SO,, can also be transferred rapidly into the systemic circula-
tion. Less than 15 percent of the total inhaled S02 is likely to be exhaled
immediately, with only small amounts (about 3 percent) being desorbed during
the first 15 minutes after the end of exposure (U.S. EPA, 1982a,b).
2.3 POTENTIAL MECHANISMS OF TOXICITY ASSOCIATED WITH INHALED PARTICLES AND S02
U.S. EPA (1982a) noted that numerous possibilities exist by which a wide
variety of toxic effects may be exerted by inhaled particles once deposited in
the respiratory tract. Certain general types of mechanisms of toxicity can be
identified to apply across a wide range of mixtures of inhaled particles,
either acting alone or in combination with other common gaseous air pollutants,
such as S09, NO , or ozone. These include, for example, possible irritant
L. X
effects that result in decreased air flow due to airway constriction, altered
mucociliary transport and effects on alveolar macrophage activity. Other toxic
effects and underlying mechanisms of action are much more chemical-specific,
and depending on the particular materials involved, may include forms of
systemic toxicity involving non-respiratory system organs and functions. The
main focus of discussion here is on general mechanisms of toxicity rather than
more chemical-specific ones.
2-15
-------
The tracheobronchial portion of the respiratory system is the site of de-
position of a mixture of fine (especially hygroscopically fine) and relatively
small (<10-15 pm) coarse-mode particles. Bronchoconstriction is one common
response to deposition of particles in this region and has been reported in
response to short-term exposure to high levels of various "inert" dusts, as
well as acid and alkaline aerosols of varying particle sizes. Bronchoconstric-
tion produced by acute exposures is likely because of neurologically-mediated
reflexive actions arising from chemical and/or mechanical stimulation of irri-
tant neural receptors in the bronchi. Since particle deposition and epithelial
nerve endings tend to concentrate near airway bifurcations, deposition at such
points may exert an influence on pulmonary mechanical changes due to chemical
or mechanical stimulation of receptors. Reflex coughing and bronchoconstric-
tion due to irritant effects of particles or S02 on tracheobronchial region
receptors may be related to effects observed in various epidemiological
studies, e.g., aggravation of chronic respiratory disease states such as
asthma, bronchitis, and emphysema. Also, as noted earlier, some personsvwith
asthma or other respiratory diseases may have elevated particle deposition
rates in the tracheobronchial region which may contribute to a cascading effect
of further bronchoconstriction and increased particle deposition in that
region.
Referring to the earlier discussion of particle clearance mechanisms,
several more potential mechanisms of toxicity associated with inhalation of
airborne particles can be readily discerned. This includes a plausible sequence
of events by which inhaled particles can contribute to chronic obstructive
pulmonary disease (Albert et al., 1973; Lippmann et al. , 1980). That is, in-
haled particles and noxious gases can stimulate changes in the distribution and
activities of various cell types lining the tracheobronchial airways. Acute
exposures to high levels of airborne particles initially stimulate increased
mucus secretion and mucociliary flow useful in clearing inhaled particles.
However, with continuous or repeated exposures, more marked changes can occur,
e.g., marked and persistent depression in bronchial clearance, increase in
secretory cell number, increase in the thickness of the mucous layer (Lippman
and Schlesinger, 1984). Also, certain particles and gases affect the number of
ciliated cells or their functioning so as to alter (i.e., speed or s.low) muco-
ciliary clearance rates. Mucociliary clearance is affected by fine sulfuric
acid aerosols, high levels of carbon dust, and cigarette smoke.
2-16
-------
Because of the above mucociliary clearance phenomena, airborne particles
may be importantly involved as etiological factors that contribute to various
types of chronic lung diseases, as discussed by U.S. EPA (1982a,b) and Lippmann
et al. (1980). This includes: likely involvement in the pathogenesis of
chronic bronchitis; increasing susceptibility to acute bacterial and viral
infections, especially in populations or groups (e.g., children, the elderly
and cigarette smokers) already predisposed to such infections by other factors;
and likely aggravation of preexisting disease states, e.g., chronic bronchitis
or emphysema, or other respiratory conditions such as bronchial asthma. Also,
some individuals (e.g., those with Kartagener's syndrome) have genetically
inherited defects in ciliated cell function or other disease states, which
result in much reduced mucociliary clearance of inhaled particles and poten-
tially greater vulnerability to toxic effects of such particles.
Particle deposition within the alveolar region of the lungs is mainly
limited to fine and coarse particles of less than 10 |jm D . Several important
clG
characteristics in the alveolar region affect responses to inhaled particles.
Clearance from the alveolar region is much slower than from the tracheobron-
chial region. The alveolar region is the site of oxygen uptake and of various
non-respiratory functions of the lungs that may be affected by pollutant expo-
sures. Many victims of London air pollution episodes were patients suffering
from cardiopulmonary diseases (e.g., emphysema and bronchitis), which normally
reduce the lungs' ability to transfer oxygen to blood. Individuals with
chronic lung disease and nonuniform ventilation distribution will be sensitive
to pollution if only because the delivered dose, to the region that is being
ventilated will be higher than it would be if ventilation were normally distri-
buted. Although this added load (due to pollution exposure) is usually
tolerable in normal individuals, the added stress and chain of events may lead
to fatal or irreversible damage in individuals already compromised with cardio-
pulmonary disease.
2.4 SUMMARY AND CONCLUSIONS
Studies published since preparation of the earlier criteria document (U.S.
EPA, 1982a) and the previous addendum (U.S. EPA, 1982c) support the conclusions
reached at that time and provide clarification of several issues. In light of
previously available data, new literature was reviewed with a focus towards (1)
2-17
-------
the thoracic deposition and clearance of large particles, (2) assessment of
deposition during oronasal breathing, (3) deposition in possibly susceptible
subpopulations, such as children, and (4) information that would relate the
data to refinement or interpretation of ancillary issues, such as inter- and
intrasubject variability in deposition, deposition of monodisperse versus
polydisperse aerosols, etc. Major results for the first three areas are given
below.
The thoracic deposition of particles >10 pm D and their distribution in
36
the TB and P regions was studied by a number of investigators (Svartengren,
1986; Heyder, 1986; Emmett et al., 1982). Depending upon the breathing regimen
used, TB deposition ranged from 0.14 to 0.36 for 10-(jm D particles, while the
36
range for 12-nm D particles was 0.'09 to 0.27. For particles 16.4 urn D , a
3.6 _ 36
maximally deep inhalation pattern resulted in TB deposition of 0.12.
The experimental data cited above were obtained from human exposure
studies in which the subjects inhaled through a mouthpiece. Some of the minute
ventilations employed would more normally occur with oronasal breathing (partly
via the mouth and partly nasally). Various studies (Swift and Proctor, 1982;
Miller et al., 1984, 1986) have simulated deposition during oronasal breathing
by adjusting for parallel nasal and oral deposition as a function of air flow
through the respective compartments. While the magnitude of deposition in
various regions depends heavily upon minute ventilation, there is, in general,
a gradual decline in thoracic deposition for large particle sizes, and there
can be significant deposition of particles greater than 10 pro D , particularly
36
for individuals who habitually breathe through their mouth. Thus, the deposi-
tion experiments wherein subjects inhale through a mouthpiece are relevant to
examining the potential of particles to penetrate to the lower respiratory
tract and pose a potentially increased risk. Increased risk may be due to
increased localized dose of larger particles (Gerrity et al., 1983).
Although experimental data are not currently available for deposition of
particles in the lungs of children, some trends are evident from the modeling
results of Phalen et al. (1985). Phalen and co-workers made morphometric
measurements in replica lung casts of people aged 11 days to 21 years and
modeled deposition during inspiration as a function of activity level. They
found that, in general, increasing age is associated with decreasing particu-
late deposition efficiency. However, very high flow rates and large particu-
late sizes do not exhibit consistent age-dependent differences. Xu and Yu
2-18
-------
(1986) found a small peak in deposition efficiency at 2 years of age followed
by a decline with increasing age. When divided by body weight, though, their
results qualitatively agree with those of Phalen et al. (1985). Since minute
ventilation at a given state of activity is approximately linearly related to
body mass, children receive a higher TB dose of particles than do adults and
would appear to be at a greater•risk, other factors (i.e., mucociliary clear-
ance, particulate losses in the head, tissue sensitivity, etc.) being equal.
2-19
-------
-------
CHAPTER 3. EPIDEMIOLOGICAL STUDIES OF HEALTH EFFECTS ASSOCIATED WITH
EXPOSURE TO AIRBORNE PARTICLES AND/OR SULFUR OXIDES
Extensive published information exists concerning health effects associ-
ated with exposure to airborne particulate matter and sulfur oxides. Detailed
evaluations of much of this extensive literature (including discussions of po-
tential mechanisms of toxicity and findings emerging from animal toxicology ex-
periments, controlled human exposure studies, and epidemiological studies) are
provided in the 1982 EPA criteria document (U.S. EPA, 1982a), as well as
several other critical reviews of the subject (World Health Organization, 1979;
Holland et al. , 1979; Lippmann et al. , 1980; Lippmann and Schlesinger, 1984).
Key health effects findings emerging from the earlier criteria review (U.S.
EPA, 1982a) are summarized below, providing a perspective against which more
recently published studies are then highlighted and evaluated in the present
chapter.
3.1. HUMAN HEALTH EFFECTS ASSOCIATED WITH SHORT-TERM EXPOSURES
As reviewed by U.S. EPA (1982a), much information has been generated by
experimental animal studies and controlled human exposure studies in regard to
health effects associated with short-term (<24 hr.) exposures to airborne
particles and sulfur oxides. Especially important information concerning the
effects of acute sulfur dioxide exposures on pulmonary functions has been
derived from controlled human exposure studies, as later discussed in Chapter 4
of this Addendum. However, other crucial information gained in regard to
effects on human health of short-term exposures to realistic concentrations of
sulfur oxides and/or airborne particles has come from epidemiological studies.
Complicating such studies are the frequent co-occurrence of elevated levels of
sulfur oxides along with airborne particles and difficulties in adequately
controlling or adjusting for the effects of other potentially confounding vari-
ables. Attention is directed here mainly to identification of epidemiological
studies that yield information relevant to the delineation of exposure-effect
or exposure-response relationships.
3-1
-------
3.1.1. Mortality Effects of Short-Term Exposures
As discussed in U.S. EPA (1982a), the most clearly defined effects on
mortality arising from exposure to sulfur oxides and particulate matter have
been sudden increases in the number of deaths occurring, on a day-to-day basis,
during episodes of high pollution. The most notable of these occurred in the
Meuse Valley in 1930, in Donora in 1948, and in London in 1952. Additional
episodes with notable increases in mortality occurred in London during various
Winters from 1948 to 1962. Besides evaluating mortality associated with major
episodes, epidemiology studies also focused on more moderate day-to-day varia-
tions in mortality within large cities in relation to PM and SO pollution.
.A.
The large body of literature concerning such studies carried out in the
United Kingdom, elsewhere in Europe, the United States and Japan was critically
reviewed in detail by U.S. EPA (1982a). As discussed there, various method-
ological problems with most of the studies precluded drawing of quantitative
conclusions regarding exposure-effect or exposure-response relationships of
importance for deriving air quality standards. Among the main problems were
inadequate measurement or control for potentially confounding variables and
inadequate quantitation of exposure to airborne particles, S0? or other associ-
ated pollutants (e.g., sulfates).
Despite such problems, U.S. EPA (1982a) concluded that the then available
studies collectively indicated that mortality was clearly and substantially
increased when airborne particle 24-hr concentrations exceeded 1000 pg/m (as
measured by the BS method) in conjunction with elevations of S0? levels in
3
excess of 1000 pg/m (with the elderly or others with severe preexisting
cardiovascular or respiratory disease mainly being affected). As for evalua-
tion of risks of mortality at lower exposure levels, U.S. EPA (1982a) concluded
that studies conducted in London by Martin and Bradley (1960) and Martin (1964)
yielded useful, credible bases by which to derive conclusions concerning
quantitative exposure-effect relationships. Table 1 summarizes key conclusions
drawn from these and other critical studies of mortality and morbidity effects
associated with short-term (24-hr) exposures to particulate matter and S02, as
stated earlier in the 1982 EPA criteria document (U.S. EPA 1982a).
3-2
-------
TABLE 1.
SUMMARY OF QUANTITATIVE CONCLUSIONS DRAWN IN U.S. EPA (1982a) FROM EPIDEMIOLOGICAL STUDIES RELATING HEALTH
EFFECTS TO ACUTE EXPOSURE TO AMBIENT AIR LEVELS OF S02 AND PM IN LONDON**
Type of Study
lortality
Effects observed
Clear increases in daily total
24-hr average
BS
XLOOO
pollutant level (ug/m3)
S02
XLOOO
Reference
Martin and Bradley
lorbidity
mortality or excess mortality
above a 15-day moving average
among the elderly and persons
with preexisting respiratory or
cardiac disease during the
London winter of 1958-59.
Analogous increases in daily
mortality in London during
1958-59 to 1971-72 winters.
(1960); Martin (1964)
Mazumdar et al. (1981)
Some indications of likely increases
in daily total mortality during the
1958-59 London winter, with greatest
certainty (95% confidence) of
increases occurring at BS and S02
levels above 750 ug/m3.
Analogous indications of increased
mortality during 1958-59 to 1971-72
London winters, again with greatest
certainty at BS and S02 levels above
750 |jg/m3 but indications of small
increases at BS levels <500 ng/m3
and possibly as low as 150-200 |jg/m3.
500-1000
500-1000
Martin and Bradley (1960)
Mazumdar et al. (1981)
Worsening of health status among
a group of chronic bronchitis
patients in London during
winters from 1955 to 1960.
>250-500*
>500-600
Lawther (1958); Lawther
et al. (1970)
No detectable effects in most
bronchitics; but positive
associations between worsening
of health status among a selected
group of highly sensitive chronic
bronchitis patients and London BS
and S02 levels during 1967-68
winter.
<250*
<500
Lawther et al. (1970)
*Note that the 250-500 vg/m3 BS levels stated here may represent somewhat higher PM concentrations than those actually
associated with the observed effects reported by Lawther (1970). This is due to the estimates of PM mass (in |jg/m3 BS)
used by Lawther being based on the O.S.I.R. calibration curve found by Waller (1964) to approximate closely a site-specific
calibration curve developed by Waller in central London in 1956, but yielding somewhat higher mass estimates than another
site-specific calibration developed by Waller a short distance away in 1963. However, the precise relationship between
estimated BS mass values based on the D.S.I.R. curve versus the 1963 Waller curve cannot be clearly determined due to
several factors, including the non-linearity of the two curves and their convergence at low BS reflectance values.
*Source: U.S. EPA (1982a). Subsequent reanalyses of the London mortality data alluded to here have been carried out since
completion of U.S. EPA (1982a) and are described elsewhere in this Chapter. In general, the results of these .more recent
reanalyses demonstrate relatively continuous exposure-response relationships across the entire range of BS levels reported
for London during the winters of 1958-59 to 1971-72, with no clear thresholds evident for significant associations between
daily mortality and BS (but not S02) at levels ranging to below 250-500 ug/m3. The difference in the gravimetric calibra-
tions noted above for 1956 and 1963 and lack of later gravimetric calibration of BS readings, however, limit specification
of precise PM levels (in ug/m3) associated with the relatively small increases in mortality seen at lower BS concentrations.
In addition, new morbidity studies regarding effects of shprt-term PM/SO exposures of a more contemporary nature are "also
discussed elsewhere in this chapter. . ; :
3-3
-------
The studies by Martin and Bradley (1960) and Martin (1964) dealt with a
relatively small body of data on relationships between daily mortality in
Greater London and daily variations in pollution (smoke and sulfur dioxide)
during the winter of 1958-59. Aerometric data from multiple sampling sites
used in their analysis can be considered reasonably representative of outdoor
concentrations in the areas where people lived, although the inclusion of
outer, less-densely populated areas meant that average exposure may have been
underestimated. During the winter of 1958-59, Martin and Bradley (1960)
reported that mortality increased on some days when smoke concentrations
o
increased by more than 100 ug/m over the previous day or when S09 concentra-
o e-
tions increased by 70 ug/m (0.025 ppm). Increases in daily mortality were up
to about 1.2 times expected values assessed from 15-day moving averages. Thick
fog (visibility less than 200 meters) was also associated with increases in
mortality. The relative importance of the three factors (smoke, S02, fog)
could not be clearly determined, but on the basis of other work, the authors
considered that smoke was probably most important. When results were con-
sidered on an absolute basis (Lawther, 1963), it was concluded that increases
in mortality became evident when the 24-hr mean concentrations of smoke and
o o
sulfur dioxide exceeded 750 pg/m and 710 pg/m (-0.25 ppm), respectively.
Studies on day-to-day variations in mortality in London were continued in
successive winters and coupled with the records of emergency hospital admis-
sions. Martin (1964) showed correlations between both the daily mortality and
hospital admission data and concentrations of smoke or $62. There was no
clearly defined level (threshold) above which effects were seen, but fairly
consistent increases in both mortality and hospital admissions occurred when
concentrations of smoke and sulfur dioxide each exceeded a 24-hr mean of about
o
500 ug/m . Based on the above analyses and a reanalysis of the Martin and
Bradley data set by Ware et al. (1981), U.S. EPA (1982a) concluded that notable
increases in mortality among the elderly and chronically ill may have been
q
associated with BS and S02 levels in the range of 500 to 1000 |jg/m . Much less
certainty was attached to suggestions of possible slight increases in mortality
at still lower BS or S02 concentrations, based on the Ware et al. (1981)
reanalyses.
In subsequent years, because of reductions in London BS levels brought
about by implementation of the British Clean Air Act and more gradual S02
reductions, only few occasions occurred when smoke or S02 levels exceeded 500
3-4
-------
. Analyses of daily mortality in London in relation to variations in
smoke and S02 levels during winters from 1958-59 to.1971-72 were reported by
Maz-umdar et al. (1981). These analyses are of special value in attempting to
define lowest levels of exposure to particulate matter and/or S02 associated
with increased mortality, because they include winters when levels of those
o
pollutants never exceeded 500 ug/m . The results obtained for airborne parti-
cles (measured in terms of BS) were analyzed in relation to linear and quadra-
tic models, which Mazumdar et al. (1981) found to provide good fits to the data
examined after relevant potentially confounding variables, e.g., temperature
and humidity, were taken into account statistically. U.S. EPA (1982a) con-
cluded that both of the models suggest small increases in mortality at smoke
3 3
levels below 500 ug/m and, possibly, to as low as 150-250 ug/m .
In a publication newly available since completion of.U.S. EPA (1982a),
Mazumdar et al. (1982) reported further on three types of analyses,of London
mortality during the 1958-59 to 1971-72 winters: (1) year-by-year multiple.
regressions, (2) stratification using nested quartiles of one pollutant within
another, and (3) multiple regression of a subset of high-pollution days. Steps
were taken in each analysis to control for potentially confounding factors.
Mortality and pollution variables were first divided by their winter means
(indexed or percent) to adjust for year-to-year variation. Seasonal trends,
were adjusted for by treating each variable as a deviation (residual) from
15-day moving averages; these residuals were then corrected,for weather factors
by regressing separately indexed mortality, SOp and smoke residuals in tempera-
ture and humidity residuals of the same day, previous day and lag days up to 1
wk; and dummy variables were used to remove day-of-week effects. The corrected
indexed pollution variables were then reconverted to absolute units by
multiplying each value by the corresponding winter mean, but the mortality
values were left in indexed form.
Mazumdar et al. (1982) reported that the year-by-year multiple regressions
yielded generally much smaller coefficients for S09 (14 winter x = 1.17 percent
3 -
mortality increase/mg/m S09; p >0.10) versus those for smoke (14 winter x =
3
25.09 percent/mg/m smoke; p <0.01). Also, the nested quartile analyses using
16 cells (i.e., 4 quartiles of smoke within 4 quartiles of S02 and vice versa),
were reported as only partially successful, in that substantial covariation
remained between the two pollutants in the highest and lowest quartiles.
Visual inspection of other cells, the authors noted, nevertheless suggested a
3-5
-------
much larger smoke than SCL effect. Last, multiple regression analyses, using
the 100 days during the 14 winters when the two pollutants were in their
highest deciles (excluding 5 days during the 1962 episode), were reported as
showing that mortality increases monotonically with smoke for fixed S09 levels
3
but mortality only increased with SO,, levels above 0.7 mg/m for fixed smoke
levels. The authors concluded that their analyses of London data for 14
winters support the conclusion that mortality was significantly and positively
associated with air pollution, but the mortality/pollution association was
almost entirely due to smoke. They also noted possible contributions of SO, at
sufficiently high pollutant levels (i.e., when both SO™ and smoke >0.7 mg/m ).
Results from linear and quadratic models of mortality regressed on smoke alone
led the authors to state a preference for the quadratic model supplemented by a
3
hypothesis that at low smoke levels (<0.3 mg/m ), smoke may serve as a surro-
gate for an unidentified variable (e.g., a highly toxic fraction of particulate
emissions).
More recently, Ostro (1984) reported that new analyses of the same 1958-59
to 1971-72 London winter data indicate some risk of mortality even at smoke
levels below 150 pg/m. Specifically, Ostro (1984) employed a variation of a
standard multiple regression model to test whether the data supported the
o
existence of a "threshold" at BS = 150 pg/m . Observations across the range of
pollutant levels were divided into two segments, those falling below versus
3 3
those above 150 pg/m . Regression analyses for data below 150 pg/m , con-
trolling for important potentially confounding factors (e.g., temperature,
humidity, etc.), indicated a statistically significant pollutant effect on
mortality below the BS = 150 ug/m level. For 11 of 14 winters, the coeffi-
cients for mortality associations with BS values below 150 were statistically
different from zero at p <0.10. Additional analyses focused on the last seven
winters, starting in 1965-66, during which there were no BS values above 500
o
|jg/m . The mortality coefficients were significant at p <0.05 for six years
and at the 0.01 level in four of the years. Ostro (1984) concluded that these
results are suggestive of a strong association of BS with mortality, holding
3
temperature and humidity constant, at levels below 150 pg/m .
The Mazumdar et al. (1982) and Ostro (1984) analyses produced generally
analogous results in relation to reported findings on PM effects: (1) each
found significant positive associations between increased mortality and BS
levels for most of the 14 London winters from 1958-59 to 1970-71, when the data
3-6
-------
were analyzed on a yeat—by-year basis; (2) the coefficients obtained for
mortality associations with lower BS values were generally larger than, values
obtained with higher BS levels, an apparently counterintuitive result; and
(3) no clearly defined threshold for BS-mortality associations could be
identified based on either set of analyses, both of which showed small but
3
significant associations at levels below 500 ug/m BS.
No readily obvious reasons stand out as explaining the reported stronger
correlations between lower BS values and mortality than associations seen at
higher BS levels, although both Mazumdar et al. (1982) and Ostro (1984)
tendered some possibilities (for example, the low levels of smoke in later
years may have contained higher proportions of respirable particles or specific
toxic materials). Still other questions have been raised in regard to these
analyses; for example: (1) whether or not the effects of smoke and S0? can be
credibly separated out, given the very high correlation (generally X).80 or
0.90) between BS and S0? levels in the subject data set; (2) whether unmeasured
variables, such as indoor air pollution levels, might have also covaried with
outdoor BS and S0? concentrations and contributed to observed mortality
effects; or (3) whether other unevaluated longer-term changes in demographic
characteristics of the London population (age, socioeconomic levels, ethnic
mix, etc.) over the 14 winters might not be such as to contribute to spurious
apparent associations between mortality increases and BS or SO^. Also, Roth et
al. (1986) suggested that use of deviations of mortality from 15-day moving
averages may hide the true relationship between pollution and mortality.
Not all of these issues can be definitively resolved at this time. How-
ever, it is unlikely that long-term demographic shifts during the 14 year study
period could account for significant year-by-year associations; nor is it
likely that indoor air exposures would be consistent from year to year, given
variations in yearly climatic conditions coupled with gradual changes in
heating practices (shifts away from open hearth burning of coal in residences)
that occurred during the 14 year study period. In addition, further reanalyses
of the 1958-59 to 1971-72 London mortality data have been carried out in an ef-
fort to address issues of the above types. For example, an unpublished
analysis of the 1958-59 to 1971-72 London winter data set carried out by
Shumway et al. (1983) for the California Air Resources Board (CARB) also
produced results indicative of risk below the 500 ug/m3 level of smoke. These
analyses used a spectral transform multiple regression model and detrending of
3-7
-------
data to correct for temperature and autocorrelation effects. The best model
for predicting cardiovascular, respiratory or overall mortality used lagged
temperature and logs of same day levels of S02 or smoke. Results were reported
to indicate that pollution acts positively and instantaneously, whereas tempera-
ture has both a significant same-day effect and a strong negative effect with a
lag of two days. The largest portion of variance in daily mortality was attri-
buted to cyclical pollution-temperature patterns typified by 7-21 day periods.
Overall, these results suggest that although relatively small increases in PM
air pollution may be associated with increased daily mortality in London, the
effects were likely greater when higher PM concentrations occurred as part of
multi-day cycles than with short duration episodes.
More recent reanalyses, performed by Marcus and Schwartz in cooperation
with CARB, are concisely described in Appendix A to this Addendum. The memo-
randum in Appendix A summarizes their reanalyses as described in an attached
more extensive paper (Schwartz and Marcus, 1986) now being prepared for sub-
mittal for publication. Their reanaTyses indicate that: (1) Clear exposure-
response relationships are evident between the main air pollution variables
(BS, S02 levels) and increases in daily mortality when graphically displayed
either in terms of absolute daily mortality or deviations from 15-day moving
averages of daily mortality; (2) Multiple regression analyses revealed that
either daily mortality or derivations in daily mortality from 15-day moving
averages were positively and significantly correlated with increases in BS or
SCL across the 14 winters, adjusted for time series autocorrelation, tempera-
ture, and humidity; (3) Analyses on a year-by-year basis yielded significant
linear correlations of mortality with BS for 13 of the 14 winters, including
3
later years only having days <250 (jg/m and even for 6 of 11 winters when only
3
days with BS <200 ng/m were included in the analyses; (4) The partial regres-
sion coefficients for BS versus mortality are relatively stable from year to
year, although they tend to increase for later years versus the first 7 years;
(5) The partial regression relationship between mortality and BS is non-linear,
the relationship being convex with somewhat steeper linear slope at lower BS
3 3
levels (<250 pg/m ) than for higher BS levels (>500 ng/m ); (6) S09 is signifi-
3
cantly associated with daily mortality, mainly at high levels (>500 jjg/m ); but
(7) S02 effects appear to be somewhat distinguishable from BS effects, with the
3-8
-------
mortality effects of BS remaining significant and relatively large when SO^ is
included in the regression model whereas inclusion of BS in the model reduces
the S0? coefficients to insignificant values. Overall, these reanalyses fur-
ther substantiate and reinforce major results derived from earlier published
analyses and point more strongly toward PM-mortality associations even at levels
below 150-250 jjg/m . On the other hand, it is difficult to estimate with any
precision what PM levels (in ug/m ) may have been associated with increased mor-
3
tality at lower BS levels (<150 to 250 (jg/m ), given lack of contemporaneous
gravimetric calibration data beyond 1963.
Taking into account the above considerations, the following conclusions
appear to be warranted based on the earlier criteria review (U.S. EPA, 1982a)
and present evaluation of newly available analyses of the London mortality
experience: (1) Markedly increased mortality occurred, mainly among the
elderly and chronically ill, in association with BS and S0? concentrations
3 ^
above 1000 (jg/m , especially during episodes when such pollutant elevations
occurred for several consecutive days; (2) During such episodes coincident high
humidity or fog was also likely important, possibly by providing conditions
leading to formation of hLSO, or other acidic aerosols; (3) Increased risk of
mortality is associated with exposure to BS and SO,, levels in the range of 500
3 3
to 1000 ug/m , for S0~ most clearly at concentrations in excess of ~700 ug/m ;
and (4) Convincing evidence indicates that relatively small but statistically
significant increases in the risk of mortality exist at BS (but not $02) levels
below 500 |jg/m , with no indications of any specific threshold level having
3
been demonstrated at lower concentrations of BS (e.g., at <150 ug/m ).
However, precise quantitative specification of the lower PM levels associated
with mortality is not possible, nor can one rule out potential contributions of
other possible confounding variables at these low PM levels.
In another study of air pollution relationships with mortality reported
since the earlier criteria review (U.S. EPA, 1982a), Mazumdar and Sussman
(1983) evaluated associations between mortality events and daily particulate
matter and S0? levels in Pittsburgh, PA. The analysis, limited to investiga-
tion of same-day events, reported a possible relationship between heart disease
mortality/morbidity and same day particulate levels (measured in terms of COH),
but not same-day S0? levels. The analyses specifically evaluated daily mortal-
ity rates during 1972-1977 for all of Allegheny County, PA in relation to daily
3-9
-------
average COM and SO/, measurements obtained at each of three air monitoring
stations: one at the center of the County within a high pollution section of
Pittsburgh; another situated relatively near the first in a somewhat less
polluted area; and a third in a distinctly cleaner area on the northeast edge
of the County. Corrections for trend and seasonal factors were made by use of
daily deviations from 15-day moving averages for air pollution, temperature and
mortality variables. Multiple regression analyses revealed no statistically
significant associations between mortality for all ages or heart disease
mortality in relation to either S02 or COH when regressed on each variable
alone. When S02 and COH were considered jointly, only the associations between
total or heart disease mortality and COH measurements at the Hazelwood (high
pollution area) station were significant at p <0.05. These results, however,
cannot be accepted as providing meaningful information on mortality-air pollu-
tion associations in the Pittsburgh area in view of: (1) inadequate character-
ization of county-wide air pollution levels against which to compare mortality
rates for the entirety of Allegheny County, the S02 and COH levels at each of
the three monitoring stations used not being highly correlated (mostly r <0.4
to 0.5) with values at the other stations; (2) internal inconsistencies whereby
larger coefficients were obtained for associations of mortality to COH readings
at the cleaner air station on the edge of the County than the intermediate
pollution station near the center of the County; and (3) the use of a large
number of separate mortality regression analyses, from among which only two
were significant at p <0.05.
In addition to the above reanalyses of London mortality data, reanalyses
of mortality data from New York City in relation to air pollution have been
recently reported by Ozkaynak and Spengler (1985). These investigators carried
out time-series analyses on a subset of New York City data included in a prior
analysis by Schimmel (1978) which was critiqued during the earlier criteria
review (U.S. EPA, 1982a). The present reanalyses by Ozkaynak and Spengler
(1985) evaluated 14 years (1963-76) of daily measurements of mortality (the sum
of heart, other circulatory, respiratory, and cancer mortality), COH, SOp, and
temperature. Prior to regression analysis, efforts were made to remove assumed
low-frequency confounding by "filtering" each variable to remove its slow-
moving components. This included not only use of residuals from 15-day moving
averages, but also evaluation of sensitivity of results to other filters.
Initial exploratory analyses estimated regression coefficients for COH and SO
3-10
-------
after all variables were preprocessed with one of several filters (e.g., taking
residuals from 7-, 15-, or 21-day moving averages and other filters that
removed all cycles in the data that fell beyond indicated periods measured in
days). Overall, the regression coefficients for COM ranged from 1.2 to 5.4
daily deaths per unit of COH, most being statistically significant (p <0.05).
Also, a reasonable range of variation in temperature specifications produced
coefficients ranging from 1.3 to 1.8 deaths per COH unit. The risk coeffi-
cients of Schimmel (1978) were near the lower end of the range of coefficients
found by Ozkynak and Spengler (1985). The latter investigators noted then that
they were able to generate a fairly consistent set of estimates by performing a
number of sensitivity analyses. They also correctly note that these initial
estimates were subject to several technical limitations: (1) misclassification
of population exposure can occur in using aerometric data from one fixed
monitoring site; (2) the exposure index, COH, is imperfectly related to respi-
rable particle mass levels; and (3) the range of exploratory models initially
fit may not have been diverse enough. Consequently, an additional reanalysis
was undertaken.
Specifically, more recent reanalysis of the New York City data reported by
Ozkaynak and Spengler (1985) used standard time-series methods to control for
covariates such as temperature and to handle the problem of autocorrelation.
Their previous analysis was also extended by adding records of visibility and
weather from three New York City airports, in order to examine spatial homoge-
neity of daily air pollution in New York City and to use visibility as a
surrogate for aerosol extinction (bext) or for fine particle (FP) pollution as
discussed by Ozkaynak et al. (1985). The most salient feature of the mortality
data found by this reanalysis was a strong seasonal component which confounds
direct regressions involving mortality, air pollution and weather variables. A
simple trigonometric expression was used that removed the temperature cyclic
component and rendered nonseasonal temperature nonsignificant. Another
stationary autoregressive term was also used to exhaust the time-series
structure of the mortality records. Consideration of lagged regressions and
interactions did not improve the model's predictive ability. Time-series
analyses were then performed with a linear model and in a multivariate manner
in which corrections for seasonality and autocorrelation were introduced into
the linear model. Preliminary estimates of excess deaths (e.) or elasticities
for the pollutant variables were thereby calculated, resulting in the following
3-11
-------
findings: (1) the time-series analysis showed SCL levels to be significantly
correlated with mortality (es02 =2.3 percent); (2) COH also contributed
significantly to excess deaths (eCQH = 2.4 percent); (3) B ., a variable used
as a surrogate for FP pollution was also a significant contributor to excess
daily deaths (~1.2 percent); and (4) the total estimated excess deaths attri-
butable to air pollution was ~6.0 percent. The authors concluded that although
these are interim results (they are also analyzing the data one year at a time
and by each quarter), these findings: (1) indicate that during the study period
ambient air pollution of a large urban area was contributing to mortality, (2)
appear to corroborate results from cross-sectional mortality studies, and (3)
indicate that particulate air pollution, even at current levels, could be of
concern for public health. However, the authors again correctly noted limita-
tions of their analyses which preclude full reliance on these preliminary
results for risk assessment purposes: (1) The results reflect aggregate
analyses of 14 years of data and more thorough analyses need to be done to take
into account changing SOp and aerosol composition over the period (preliminary
analyses indicate no differences in pollutant coefficients for 1963 to 1970 and
1971 to 1976); (2) The results are based on aerometric data from one monitoring
station and visibility data from one airport (JFK); and (3) The effects of heat
waves and influenza epidemics during the study period have not been considered
in any detail in these preliminary analyses.
Hatzakis et al. (1986) recently published a study of short-term effects of
air pollution on mortality in Athens, Greece, during 1975-82. Daily concentra-
tions of SCL (acidimetric method) and smoke (standard British Method) measured
by a five-station network in Athens were evaluated in relation to mortality
data abstracted from the Joint Registries of Athens and 18 other contiguous
towns in the Greater Athens area. The authors reported that adjusted daily
mortality (estimated by subtracting the observed mortality value from an
"expected" value, calculated after fitting a sinusoidal curve to the empirical
mortality data) was significantly and positively related to S02 levels (b =
+0.0058, p = 0.05), but not to smoke levels. Separate multiple regression
analyses were done for S02 and smoke, controlling in each case for temperature,
relative humidity, secular, seasonal, monthly and weekly variations in
mortality as well as interactions of the above variables with season.
Evaluation of a possible threshold for the SOp-mortality effect was carried out
by successively deleting from the regression model days with the highest S02
3-12
-------
values. These analyses resulted in the authors suggesting that, if there is an
O
S02 threshold, it must lie slightly below 150 ug/m (mean daily value).
The latter result, as stated by the authors, is not consistent with
results of other studies in which SO- mortality thresholds have been placed
O ^
around the value of 300 ug/m (or, more credibly, around 500 ug/m , as per U.S.
EPA, 1982a). Nor is the failure to find significant associations between
mortality and smoke consistent with other more usual published findings
(although differences in chemical composition of PM in Athens and lack of
calibration of smoke readings against gravimetric measurements make it
difficult to compare smoke levels from Athens versus elsewhere). Other
questions also arise which make it difficult to fully accept the reported
findings, e.g.: (1) how representative are the aerometric data for the entire
Athens metropolitan area from which the mortality data were abstracted,
although the typography of the area, with Athens and adjoining towns situated
in a coastal "bowl" surrounded by mountains, and high correlations (mostly r
>0.50-0.60) between pollutant readings from the five network stations suggest
that the aerometric data may well be quite representative; (2) whether use of
deviations of observed mortality data for 1975-82 from expected values derived
from 1956-58 mortality data as a pre-high pollution baseline period is
statistically sound; (3) whether separate regression analyses for S02 and smoke
alone are sufficient versus analyses with both these pollutants included; and
(4) whether effects of temperature or flu epidemics were adequately compensated
for in the analyses.
In summary, the above newly available reanalyses of New York City data
raise possibilities that, with additional work, further insights may emerge
regarding mortality-air pollution relationships in a large U.S. urban area.
However, the interim results reported thus far do not now permit definitive
determination of their usefulness for defining exposure-effect relationships,
given the above-noted types of caveats and limitations. Similarly, it is pre-
sently difficult to accept the findings of mortality associated with relatively
low levels of SOp pollution in Athens, given questions stated above regarding
representativeness of the monitoring data and the statistical soundness of
using deviations of mortality from an earlier baseline relatively distant in
time. Lastly, the newly reported analyses of mortality-air pollution
relationships in Pittsburgh (Allegheny County, PA) utilized inadequate exposure
characterization and the results contain sufficient internal inconsistencies,
3-13
-------
so that the analyses are not useful for delineating mortality relationships
with either S02 or PM.
3.1.2. Morbidity Effects of Short-Term Exposures
As noted by the World Health Organization (1979), epidemiological studies
can be useful in assessing morbidity effects associated with air pollution in
different communities or in areas where changes in air pollution occurred over
time. In such studies, where respiratory diseases are'followed, it is necessary
to control for age distribution, socioeconomic status, and other possibly con-
founding factors. It is also crucial that adequate characterization of expo-
sure to air pollutants of interest be carried out, if quantitative conclusions
are to be drawn regarding exposure-effect or dose-response relationships.
However, very few of the available epidemiological studies on morbidity effects
associated with short-term exposure to airborne particles allow for such
conclusions, as evaluated by U.S. EPA (1982a).
Those reported by Lawther for London populations (see Table 1) were
identified by U.S. EPA (1982a) as providing credible bases for drawing quanti-
tative-type conclusions about morbidity effects associated with airborne
particles (measured as smoke) and elevated S02 levels. Lawther et al. (1970)
reported on studies carried out from 1954 to 1968 mainly in London, using a
diary technique for self-assessment of day-to-day changes in conditions among
bronchitic patients. A daily illness score was calculated from the diary data
and related to BS and S02 levels and weather variables. Pollution data for
most of the London studies were mean values from the group of sites used in the
mortality/morbidity studies of Martin (1964); those aerometric measurements
likely provide reasonable estimates of average exposure in areas where study
subjects lived or worked. In early years of the studies, when pollution levels
were generally high, well defined peaks in illness score were seen when concen-
trations of either BS or S02 exceeded 1000 ,pg/m3. With later reductions in
pollution, the changes in condition became less frequent and of smaller size.
From the series of studies as a whole, up to 1968, it was concluded that the
minimum pollution levels associated with significant changes in the condition
of the patients was a 24-hr mean BS level of -250 pg/m3 together with a 24-hr
o
mean S02 concentration of ~500 ug/m (0.18 ppm). A later study reported by
Waller (1971) showed that, with much reduced average levels of pollution, there
was an almost complete disappearance of days with smoke levels exceeding 250
3-14
-------
|jg/m3 and S02 levels over 500 ng/m3 (0.18 ppm). As earlier, some correlation
remained between changes in the conditions of the patients and daily concentra-
tions of smoke and S02, but the changes were small at these levels and it was
difficult to discriminate between pollution effects and those of adverse
weather. Thus, as concluded by U.S. EPA (1982a), the observed effects (wors-
ening of health status among chronic bronchi tic patients) were clearly
associated with BS levels of 250 to 500 ug/m3 and, possibly, somewhat lower
levels (<250 M9/m3) for highly sensitive bronchitic patients.*
Since preparation of U.S. EPA (1982a) evaluations summarized in Table 1,
additional studies have appeared concerning morbidity associated with short-
term exposure to airborne particles and/or sulfur oxides. Dockery et al.
(1982), for example, reported on pulmonary function evaluations carried out for
school children in Steubenville, OH as part of the Harvard Six-Cities Study.
Pulmonary function was evaluated immediately before and after air pollution
episodes in 1978, 1979 and 1980, by relating spirometric measurements (appro-
priately corrected for height, etc.) to aerometric data (e.g., TSP and S02
levels) obtained from state air pollution monitors. Data for each individual
child were evaluated. Linear decreases in forced vital capacity (FVC) with
increasing TSP concentrations were found, and slopes were determined for linear
relationships fitting the data for four different observation periods (fall,
1978; fall, 1979; spring, 1980; fall, 1980). The slope of FVC vs. TSP was
calculated for 335 children with three or more observations during any of the
four study periods. Of the 335 children examined, 194 were tested during more
than one study period. On average, estimated FVC was approximately 2 percent
lower following each alert, whereas forced expiratory volume in 0.75 sec
(FEV0 75) did not change during the 1978 study but was decreased by 4 percent
during the 1979 alert. In the spring of 1980, similar declines were seen in
FVC and FEVn 7(- values as were found following the previous alerts, but no
U • / O
significant declines were seen in fall, 1980, when pollutant levels were
distinctly lower than for previous alerts (e.g., TSP levels did not exceed 160
ug/m3 in fall, 1980). The largest declines in lung function were observed one
to two weeks after the episodes. Fifty-nine percent of the children had slopes
*Note: Roth et al. (1986) have recently raised questions regarding how well
the health indicator values used in the Lawther morbidity studies reflect
actual health status and suggest that associations between temperature and
health may be understated in this data set.
3-15
-------
less than zero (i.e., decreasing FVC with increasing TSP). The median slope
Q
for the entire sample was -0.081 mL/ug/m , which is significantly less than
zero (p <0.001) by a Wilcoxan Signed Rank test. The median FVC vs. S00 slope
3 f-
was -0.057 mL/ug/m , also significantly (p <0.01) less than zero, but the
relationship with mean daily temperature was not significantly less than zero.
Similar analyses performed with FEVQ 75 also showed the relationships (slopes)
for S02 and TSP to be significantly less than zero.
Overall, these repeated measurements of lung function showed statistically
significant but physiologically small and apparently reversible declines of FVC
and FEVg j^ levels to be associated with increases of 24-hr mean TSP levels.
On days of testing for pulmonary function effects, the TSP levels ranged from
O Q
11.0 to 272 ug/m and S02 levels ranged from 0.0 to 281 ug/m . However,
maximum TSP levels of 312 or 422 ug/m3 occurring in fall, 1978, 2 to 5 days
prior to spirometric testing may have contributed to the observed declines in
lung function for some children included in data analyses for that period.
o
Similarly, the maximum S02 value of 455 ug/m recorded on days immediately
preceding the spirometric testing during the Fall, 1979 period may have
accounted for observed declines in lung function. The investigators noted that
it was not possible to separate the relative contributions of the two pollu-
tants, nor were any thresholds for the observed pulmonary function decrements
discernable within the above broad range of TSP and S02 levels. Nevertheless,
these results appear to demonstrate that small, reversible changes in. pulmonary
function can occur as the consequence of increased concentrations of TSP and
S02 somewhere in the above ranges. Whether such pulmonary function changes per
se are adverse or can lead to other, irreversible changes or make the lung more
susceptible to later insults remains to be resolved. Evaluations of such
issues may need to take into account an apparent subset of "responders" within
the population of children studied, who showed greater than average declines in
lung function in relation to TSP or S02 levels. For example, the lowest
quart! le of slopes of FVC and FEVQ 75 versus TSP were -0.386 and -0.306
mL/ug/m , respectively.
Results consistent with and supportive of the findings of Dockery et al .
(1982) have emerged from another recently reported study conducted by Dassen et
al. (1986) in the Netherlands. Baseline pulmonary function data were obtained
for a sample of more than 600 children during November, 1984. Then, a subset
of the same children (N = 62) were retested again in January, 1985, during an
air pollution episode when 24-hr mean values for TSP (hi-vol samples), RSP
3-16
-------
(respirable suspended paniculate, D5Q <3.5 by cyclone sampler), and S02
(acidimetric technique) measured via a 6-station network all reached the range
of 200 to 250 ug/m3. Several lung function parameters showed statistically
significant average declines of 3 to 5 percent upon second (episode) testing in
comparison against each child's own earlier baseline values, including
decrements seen on the second day of the episode in both FVC and FEV levels, as
well as in measures reflecting small airway functioning (i.e., maximum
mid-expiratory flow and maximum flow at 50 percent vital capacity). Declines
from their original baseline values for these parameters were still observed 16
days after the episode upon retesting of another subset of the children, but no
differences were found between baseline and retest values for a third subset of
children reevaluated 25 days after the episode. Given the lack of evident
effects at this latter post episode time point, 24-hr mean TSP, RSP, and S02
levels measured in the 100 to 150 ug/m3 range just prior to the last lung
function tests may not be sufficient to cause observable pulmonary function ef-
fects in children. Overall, the Dassen et al. (1986) results are very
analogous to those found by Dockery et al. (1982) in connection with the
Stuebenville episodes. That is, the relative declines in lung function
parameters were similar in magnitude (taking into account corrections made for
lung growth), and the 2 to 3 week time course for decrements persisting after
the episodes were similar.
Mazumdar and Sussman (1983), discussed earlier, not only studied relation-
ships between mortality and measures of PM and SOX pollution in Pittsburgh, PA
during 1972-77, but also included evaluations of morbidity (indexed by
emergency hospital admissions) in relationship to daily COH and S02
concentrations corrected for temperature and seasonal variations. Significant
associations were reported between same-day COH values (which ranged from near
0.0 to 3.5 units) and total morbidity and heart disease morbidity for all ages
(1 to 59 yr) and > 60 yr age groups, but no consistent statistically
significant associations between morbidity categories and same-day S02 levels
(ranging from near 0 to 0.14 ppm) monitored at the same stations. However,
these results cannot be taken as indicative of associations between increased
morbidity and elevated PM or S02 levels in the Pittsburgh area, given limita-
tions identified earlier in relation to the mortality analyses from the same
study, i.e.: (1) inadequate characterization of air pollution concentrations
representative of the entirety of Allegheny County from which the morbidity data
were drawn, and (2) internal inconsistencies in the results, with various
3-17
-------
classes of morbidity variously being more strongly associated with S02 or COH
measured at lower pollution stations than higher pollution stations.
Perry et al. (1983), followed 24 Denver asthmatic subjects,from January
through March, 1979, using twice daily self-obtained measurements of each
subject's peak expiratory flow rates (from Mini-Wright Peak Flow Meters) and
recording use of "as-needed" aerosolized bronchodilators and reports of airway
obstruction symptoms characteristic of asthma. These measures of morbidity
were tested for relationships to air pollutants using a random effects model.
Dichotomous, virtual impactor samplers at two fixed monitoring sites provided
daily measurements (in ug/m3) of inhaled PM (total mass, sulfates,• and
nitrates), for coarse (2.5 to 15 (jm) and fine fractions (<2.5 pm). CO, S02,
03, temperature and barometric pressure were also measured. Of the environmen-
tal variables measured, only fine nitrates were significantly associated with
increased symptom reports and increased bronchodilator usage. During the
course of this study, however, TSP levels were uncharacteristically low. This
limits interpretation of the study in relation to PM effects. Use of aero-
metric data from only two monitoring stations in Denver, with unknown distances
in relation to places of residence for subjects matched to the proximal sta-
tion, also limits the usefulness of the reported findings.
Bates and Sizto (1983, 1986) have also reported results of an ongoing
correlational study relating hospital admissions in southern Ontario to air
pollution levels. Data for 1974, 1976, 1977, and 1978 were discussed in the
1983 paper. The more recent 1985 analyses evaluated data up to 1982 and
showed: (1) no relationship between respiratory admissions and S02 or COHs in
the winter; (2) a complex relationship between asthma admissions and
temperature in the winter; and (3) a consistent relationship between
respiratory admissions (both asthma and nonasthma) in summer and sulfates and
ozone, but not to summer COH levels. However, Bates and Sizto note that the
data analyses are now complicated by long-term trends in respiratory disease
admissions unlikely related to air pollution, but they nevertheless hypothesize
that observed effects may be due to a mixture of oxidant and reducing pollu-
tants which produce intensely irritating gases or aerosols in the summer but
not in the winter. More definitive interpretation of these findings may be
limited until additional results are reported from this continuing long-term
study.
Goldstein and Weinstein (1986) tested for an association between days with
S02 peaks above various levels (0.1, 0.3, and 0.5 ppm hourly readings) and days
3-18
-------
with high numbers of emergency visits for asthmatics at three inner-city
municipal hospitals in New York City during 1969 to 1972. Two areas of the
city were under study and ambient exposure data were derived .from the average
of two local air monitoring stations in these areas. No significant associa-
tions were found using the two-sided Chi-square test. Potentially confounding
factors considered included: day-of-the-week effects, temperature, and trends
in asthma following reduction of air pollution in New York City.
Goldstein and Weinstein (1986) stated that the inferences that can be
drawn from this ecological study are constrained by certain methodological
limitations. For example, they express appropriate concern for the representa-
tiveness of the S0? exposure data derived from roof top measurements. They
also appropriately emphasized that this study does not rule out a relationship
between asthma and ambient levels of S02 since this ecological approach may be
too crude to detect an effect.
Of the newly-reported analyses of short-term PM/SO exposure-morbidity
/s.
relationships discussed above, the Dockery et al. (1982) study provides the
best-substantiated and most readily interpretable results. Those results,
specifically, point toward decrements in lung function occurring in association
with acute, short-term increases in PM and SOp air pollution. The small,
reversible decrements appear to persist for 2-3 wks after episodic exposures to
these pollutants across a wide range, with no clear delineation of threshold
yet being evident. In some study periods effects may have been due to 24-hr
TSP and SC>2 levels ranging up to 422 and 455 pg/m , respectively. Notably lar-
ger decrements in lung function were discernable for a subset of children
(responders) than for others. The precise medical significance of the observed
decrements per sj; or any consequent long-term sequalae remain to be determined.
The nature and magnitude of lung function decrements found by Dockery et al.
(1982), it should be noted, are also consistent with: (1) the recently reported
findings of Dassen et al. (1986) for Dutch children; (2) observations of
Stebbings and Fogleman (1979) of gradual recovery in lung function of children
during seven days following a high PM episode in Pittsburgh, PA (max 1-hr TSP
estimated at 700 ug/m3); (3) and the report of Saric et al. (1981) of 5 percent
average declines in FEV-, Q being associated with high S02 days (89-235 ug/m ).
3-19
-------
3.2 EFFECTS ASSOCIATED WITH LONG-TERM EXPOSURES
3.2.1. Mortality Effects of Chronic Exposures
The World Health Organization (1979) notes that, in countries having reli-
able systems for the collection and analysis of data on deaths, based on cause
and area of residence, death rates for respiratory diseases have commonly been
found to be higher in urban than in rural areas. Many factors, such as differ-
ences in smoking habits, occupation, or social conditions may be involved in
these contrasts; however, in a number of countries, a general association be-
tween death rates from respiratory diseases and air pollution has been apparent
for many decades. Analyses of these data have been of great value as a lead
for epidemiologic studies, but the absence of information concerning other
relevant variables, such as smoking, and the relatively crude nature of indices
of pollution used in many of these studies make them unsuitable for the
quantitative assessment of exposure-effect relationships.
The 1982 U.S. EPA criteria document (1982a) noted that certain large-scale
"macroepidemiological" studies (or "ecologic" studies as termed by some) have
attracted attention on the basis of reported demonstrations of associations be-
tween mortality and various indices of air pollution, e.g., PM or SO levels.
s\
For example, Lave and Seskin (1970) reanalyzed mortality data from England and
Wales, and developed multiple regression equations in terms of pollution and
socioeconomic indices. Their findings of positive correlations between mortal-
ity rates and pollution are of general interest but cannot contribute to the
development of dose-response relationships because of inadequate exposure
indices used in the analyses. The authors also examined similar data for
standard metropolitan statistical areas (SMSAs) in the USA, and in a later
paper (Lave and Seskin, 1972) attempted to assess relative effects of air
pollution, climate, and home heating on mortality rates. Although equations
were obtained relating death rates to measurements of suspended particulate
matter and total sulfates (both by high-volume sampler), it is again doubtful
whether these can be regarded as valid in the absence of more adequate informa-
tion on smoking and because of inadequate characterization of exposure
parameters.
Other studies reported in further publications (Lave and Seskin, 1977;
Chappie and Lave, 1981) extended their earlier analyses. Based on such later
work, analogous positive associations between mortality and air pollution
3-20
-------
variables were reported for the United States. Many criticisms similar to
those indicated above for the earlier Lave and Seskin (1970) study apply here.
Of crucial importance are basic difficulties associated with all of their
analyses in terms of: (1) use of aerometric data without regard to quality
assurance considerations, notably including use of sulfate measurements known
to be of questionable accuracy due to artifact formation during air sampling;
and (2) questions regarding how representative the air pollution data used in
the analyses are as estimates of actual exposures of individuals included in
their study groups. In some instances, for example, data from a single moni-
toring station were apparently used to estimate pollution exposures for study
populations from surrounding large metropolitan areas.
The 1982 U.S. EPA criteria document (1982a) noted that further difficul-
ties in discerning consistent patterns of association between mortality and air
pollution variables are encountered when results of Lave and coworkers are
compared with those obtained by others using analogous macroepidemiological
approaches. For example, Mendelsohn and Orcutt (1979) carried out regression
analyses of associations between 1970 mortality rates (for 404 county groups
throughout the United States) and air pollution exposures retrospectively
estimated on the basis of 1970 and 1974 annual average pollutant data from air
monitoring sites in the same or nearby counties. Their results suggested fairly
consistent (though variable) associations between mortality for some age groups
(increasingly more positive with age) and sulfate levels but much less consis-
tent and sometimes negative associations with TSP or other pollutants, the
combined TSP-SO, pollution-health elasticity obtained by Mendelsohn and Orcutt
(1979) is similar to that obtained in the earlier studies by Lave and
coworkers, all falling in the range of 0.1 to 0.2.
Other results obtained by Thibodeau et al. (1980) in carrying out large
scale cross-sectional analyses of the above type indicate that the regression
coefficients for mortality relationships with air pollution variables are quite
unstable. Also, Lipfert (1980) reported results from an analysis taking into
account a smoking index based on state tax receipts, which he interpreted as
showing sulfates to be least harmful of seven air pollutants (including S02 and
TSP), although no adjustments for urban-rural differences in study population
residences were used. This is in contrast to unpublished analyses of 1970
United States mortality data by Crocker et al. (1979), which found no signifi-
cant relationships between air pollution and total mortality when taking into
3-21
-------
account retrospectively estimated nutritional variables and a smoking index.
Also, results of Gerking and Schultze (1981), using the same data base, indi-
cated a significant positive relationship between TSP and total mortality when
using an OLS model similar.to that of Lave and Seskin (1977) but found nega-
tive, though significant, air pollution coefficients after adding smoking,
nutrition, exposure-to-cold, and medical-care variables to a two-equation
model.
U.S. EPA (1982a) also noted that various criticisms of the above studies
have been advanced by authors of the other respective studies, but it was not
possible to ascertain which findings may be more valid than others. Thus,
although many of the studies qualitatively suggested positive associations
between mortality and chronic exposure to certain air pollutants in the United
States, many key issues remained unresolved concerning reported associations
and whether they are causal or not. Since preparation of the earlier Criteria
Document (U.S. EPA, 1982a) additional ecological analyses have been reported
regarding efforts to assess relationships between mortality and long-term
exposure to particulate matter and other air pollutants.
Chinn et al. (1981), for example, reported an ecological analysis investi-
gating the relationship of mortality to atmospheric smoke and S02 in county and
London boroughs of England and Wales during 1969 to 1973. Weighted multiple
regression analyses showed no significant association between smoke and mortal-
ity from respiratory illness. Annual average BS levels were reported to range
from 15 to 225 [jg/rn3 and S02 levels from 24 to 317 |jg/m3. The lack of signifi-
cant association found should not be taken as an indication of no effect at
these levels because: (1) the BS readings are derived from the use of mass-
reflectance calibration curves with limited or no applicability to the specific
geographic locales included in the study; and (2) ecological studies of this
type are often very insensitive to small effects of pollution.
Lipfert (1984) conducted a series of cross-sectional multiple regression
analyses of 1969 and 1970 mortality rates for up to 112 U.S. SMSA's, using the
same basic data set as Lave and Seskin (1977) for 1969 and taking into account
various demographic, environmental and lifestyle variables (e.g., socioeconomic
status and smoking). Also included in the Lipfert (1984) reanalysis were the
following additional independent variables: diet; drinking water variables;
use of residential heating fuels; migration; and SMSA growth. New dependent
variables included age-specific mortality rates with their accompanying
3-22
-------
sex-specific age variables. Both linear and several nonlinear (e.g., quadratic
or linear splines testing for possible threshold model specifications) were
evaluated. Efforts to replicate the basic analyses of Lave and Seskin (1977)
and to improve upon the fit of models using various specifications led Lipfert
(1984) to conclude that: (1) differences existed between high and low pollu-
tion SMSAs unrelated to the magnitude of the air pollution variables, i.e. that
there appear to be important variables missing from the specification; (2)
correction of errors in the Lave-Seskin data improved the regression fit and
significance of some of the coefficients; but (3) it was not possible to
conclude whether SO, or TSP has a statistically significant effect on total
mortality or whether either response is linear.
Lipfert (1984) then introduced additional variables of the type listed
above into the reanalysis in hopes of improving the specification and to
evaluate possible col linearity with the pollution variables. The fact that
some observations were incomplete for some of the newly added variables neces-
sitated the analysis of certain subsets of the original Lave-Seskin data set.
Overall, for these reanalyses, in which regressions were extended to include
new variables in stepwise fashion (but retaining the 7 Lave-Seskin variables as
the first step in each case), adding new variables significantly improved the
fit, but several of the original Lave-Seskin variables (including SO^) became
non-significant as the result of the additional variables. Further analyses
included regressions for mortality restricted to central city areas versus
SMSA-based regressions, with agreement between coefficients for sulfates being
quite poor (and negative for central city regressions broken down by age groups
<65 or >65 yr). Many of the additional explanatory variables in the above
reanalyses (both for central city and SMSA regressions) were found to be
statistically significant and were then employed in regressions using total
mortality rates adjusted for age, nonwhite population, poverty and cigarette
smoking. Results obtained with use of additional explanatory variables and
varying model specifications were very mixed: (1) Sulfate coefficients were
quite unstable, ranging from near 0.0 to 0.049 (highly significant and corres-
ponding to an elasticity of 6 percent); (2) TSP coefficients were similarly
variable, with similar maximum elasticity; (3) In no case were TSP and sulfate
variables significant in the same regression; and (4) When the full set of
explanatory variables were used with the dummy pollution variables, the coeffi-
cients for the pollution variables became more significant. Lipfert (1984),
3-23
-------
based on these total mortality analyses, concluded that: (1) The Lave-Seskin
specification is inadequate and provides misleading results; (2) Using addi-
tional explanatory variables improves the fit; (3) The existence of thresholds
for the air pollution variables can neither be proved nor disproved; (4)
Although difficult to separate S04 effects from TSP effects, the TSP coeffi-
cients displayed slightly more consistent behavior across all the data sets
considered; and (5) Effects for drinking water, ozone, and (to a lesser extent)
coal and wood heat warrant further investigation.
Results obtained by Lipfert (1984) with further age- and sex-specific
regression analyses for <65 yr old subjects, using all other variables as
defined in the above total mortality regressions, produced similar results as
for the total mortality analyses. That is, as explanatory variables are added,
the pollution variables tend to lose significance and the r2 values are con-
siderably higher than those of Lave and Seskin (1977), even when using the same
specifications. Based on the age- and sex-specific analyses: (1) Sulfate was
never significant for males (except for Lave-Seskin specifications) and only
occasionally significant for females; and (2) TSP was more often significant
for both males and females, especially with threshold specifications. Analo-
gous sex-specific analyses for persons > 65 yr old revealed further interesting
results: (1) The migration variable was the single most important variable and
the age variable was negative; (2) Sulfate was significant only with the
Lave-Seskin specification (both sexes) or with other variables suppressed
(females); and (3) TSP was never significant.
In sum, it is quite evident from the above results that the air pollution
regression results for the U.S. data sets analyzed by Lipfert (1984) are
extremely sensitive to variations in the inclusion/exclusion of specific
observations (for central city versus SMSA's or different subsets of locations)
or additional explanatory variables beyond those used in the earlier Lave and
Seskin (1977) analyses. The results are also highly dependent upon the parti-
cular model specifications used, i.e. air pollution coefficients vary in
strength of association with total or age-/sex-specific mortality depending
upon the form of the specification and the range of explanatory variables
included in the analyses. Lipfert's overall conclusion was that the sulfate
regression coefficients are not to be taken seriously and, since sulfate and
TSP interact with each other in these regressions, caution is warranted for TSP
as well.
3-24
-------
Ozkaynak and Spengler (1985) have also described recent results from
ongoing attempts of a Harvard University group to improve upon some of the
previous analyses of mortality and morbidity effects -of- air pollution in the
United States. Ozkaynak and Spengler (1985) present principal findings, from a
cross-sectional analysis of the 1980 U.S. vital statistics and available air
pollution data bases for sulfates, and fine, inhalable and total suspended
particles. In these analyses, using multiple regression methods, the associa-
tion between various particle measures and 1980 total mortality were estimated
for 98 and 38 SMSA subsets by incorporating recent information on particle size
relationships and a set of socioeconomic variables to control for potential
confounding. Issues of model misspecification and spatial autocorrelation of
the residuals were also investigated. Results from the various regression
analyses indicated the importance of considering particle size, composition,
and source information in modeling of PM-related health effects. In parti-
cular, particle exposure measures related to the respirable and/or toxic
fraction of the aerosols, such as FP (fine particles) and sulfates were the
most consistently and significantly associated with the reported (annual)
cross-sectional mortality rates. On the other hand, particle mass measures
that included coarse particles (e.g., TSP and IP) were often found to be
non-significant predictors of total mortality.
The Ozkaynak and Spengler (1985) results noted above for analysis of 1980
U.S. mortality provide an interesting overall contrast to the findings of
Lipfert (1984) for 1969-70 U.S. mortality data. In particular, whereas Lipfert
found TSP coefficients to be most consistently statistically significant (al-
though varying widely depending upon model specifications, explanatory
variables included, etc.), Ozkaynak and Spengler found particle mass measures
including coarse particles (TSP, IP) often to be non-significant predictors of
total mortality. Also, whereas Lipfert found the sulfate coefficients to be
even more unstable than the TSP associations with mortality (and questioned the
credibility of the sulfate coefficients), Ozkaynak and Spengler found that
particle exposure measures related to the respirable or toxic fraction of the
aerosols (e.g., FP or sulfates) to be most consistently and significantly
associated with annual cross-sectional mortality rates. It might be tempting
to hypothesize that changes in air quality or other factors from the earlier
data sets (for 1969-70) analyzed by Lipfert (1984) to the later data (for 1980)
analyzed by Ozkaynak and Spengler (1985) and Ozkaynak et al. (1986) may at
3-25
-------
least partly explain their contrasting results, but there is at present no
basis by which to determine if this is the case or which set of findings may or
may not most accurately characterize associations between mortality and chronic
PM or SO exposures in the United States.
/^
Selvin et al. (1984) also used regression analyses applied to ecologic
data to study the influence of air quality in the U.S. on mortality. The
analyses used 1968-72 mortality data aggregated by county (3082) or by groups
of counties comprising 410 1970 Census Public Use Sample (PUS) areas (some of
which may be a single heavily populated urban county, e.g. Los Angeles, or
several sparsely populated rural counties grouped together). Total mortality,
rather than cause-specific, rates were calculated for sex-, race-, and
age-specific categories and were then evaluated by regression analyses in rela-
tion to air quality values (for TSP, S02, and N02) extracted from data
collected at 6625 monitoring stations during 1974-76. County level aerometric
estimates were interpolated from average values at individual monitoring
stations, and air pollution estimates for the 410 PUS areas were
population-weighted averages of the county level value. Overall, various
regression analyses (taking into account numerous control variables) for
county-wide or PUS areas in all of the U.S. or broken down into regions (West,
South, etc.) yielded extremely mixed results, with both positive and negative
coefficients being obtained in various analyses for mortality in relation to
TSP, SOp, and NO-. The authors: (1) concluded that their results provided no
persuasive evidence for links between air quality and general mortality levels;
(2) noted that their results were inconsistent with previously published work;
and (3) opined that linear regression analyses applied to nationally collected
ecologic data cannot be usefully employed to infer causal relationships between
air quality and mortality. However, the manner in which the Selvin et al.
(1984) study was conducted provides little basis for assigning any credibility
to the results obtained, especially in view of: (1) use of 1974-76 air quality
data to estimate retrospectively exposures against which to compare 1968-74
mortality data and; (2) use of mortality data aggregated by county or by groups
of counties with highly variable relationships between air monitoring locations
and the population groups from which the mortality data were drawn.
In addition to ecological or macroepidemiological studies of mortality
relationships to chronic air pollution exposures in the U.S., Imai et al.
(1986) have recently published analyses of associations between mortality from
3-26
-------
asthma and chronic bronchitis and air pollution variables in Yokkaichi, Japan.
An industrial city on Ise Bay several hundred miles south of Tokyo, Yokkaichi's
industrial base and harbor facilities were largely destroyed during World War
II. They were later rebuilt to include the establishment in 1957 of a petro-
leum complex that contained the largest oil-fired power plant in Japan, which
burned high-sulfur oil that resulted in large SO,, emissions and consequent high
SO concentrations in immediate residential/commercial areas around the harbor.
s\
This continued until stringent emission controls were put in place and resulted
in dramatic decreases in SO concentrations in the highly polluted area around
s\
the harbor from 1972 to 1973 and thereafter. Mortality rates for the popula-
tion in that high pollution area were compared against analogous rates (for •
bronchial asthma or chronic bronchitis including emphysema, determined from
death certificates issued during 1963-83) for people living in less-polluted
areas of Yokkaichi. Sulfur oxides levels (measured by the lead peroxide
method) averaged across several monitoring sites in the polluted harbor area
ranged from around 1.0 to 2.0 mg/day (annual average) during 1964-72 and then
steadily declined from somewhat less than 1.0 mg/day in 1973 to less than 0.5
mg/day in 1982. This is in contrast to SO levels consistently below 0.3
/S.
mg/day (annual average) at 3 monitoring sites in the low pollution areas of the
city throughout 1967 to 1982. Annual average levels for other pollutants (N02,
TSP, oxidants) monitored in the high pollution area were also consistently low,
i.e. <0.02 ppm (N0?), <0.05 mg/m3 (TSP), and <0.05 ppm (oxidants, daily max
hourly values) from 1974 to 1982. Results obtained indicated significant
differences between chronic bronchitis mortality for persons > 60 yr old in the
high pollution area compared against rates for the same age group from the
low-pollution control area for 1967-70 and extending into 1971-74, somewhat
beyond the point where marked declines became evident in SO levels after
/\
control measures were implemented. Lagged correlations showed large signifi-
cant associations between SO levels and chronic bronchitis mortality occurring
/\
>1 yr later in the high pollution area (the largest correlations were found for
4-5 yr lags). In contrast, bronchial asthma mortality became relatively higher
in the polluted area during the 1967-70 period, and began to decrease
thereafter in more immediate response to the, improvement in air quality.
These findings, overall, are quite interesting in that they relate mortal-
ity changes in populations in circumscribed urban neighborhoods to air pollu-
tion indices obtained from monitoring sites spatially located in close
3-27
-------
proximity to the residences of the population groups for whom mortality rates
were determined. Further, consistently elevated mortality for the elderly in
the high-pollution area (relative to the control area) was evident across many
years while the SO concentrations were high, but then declined following
reductions in the SO levels, thus enhancing the likelihood of a causal rela-
}\
tionship between sulfur-containing air pollution and mortality having been
detected in the study. However, it is not possible to quantitate with any
precision the relative contributions to the observed mortality increases of S0?
versus sulfates or other sulfur agents (e.g., possibly hLSO, aerosols likely
formed in the moist air of the coastal city).
The 1982 EPA document (U.S. EPA, 1982a) also noted that other epidemic-
logical studies have more specifically attempted to relate lung cancer mor-
tality to chronic exposures to sulfur oxides, PM undifferentiated by chemical
composition, or specific PM chemical species. However, the 1982 document
concluded that little or no clear epidemiological evidence advanced to date
substantiates hypothesized links between S02 or other sulfur oxides and cancer;
nor does there now exist credible epidemiological evidence linking increased
cancer rates to elevations in PM as a class, i.e., undifferentiated as to
chemical content.
3.2.2. Morbidity Effects of Long-Term Exposures
Increased incidence of respiratory symptoms, disease states or other pul-
monary function impairments are likely to be among the effects of long-term
exposures to air pollution, since the respiratory system includes tissues that
receive the initial impact when toxic materials are inhaled. Numerous studies
have been conducted in an effort to relate pulmonary function changes to the
presence of PM or sulfur oxides air pollutants in European, Japanese, and Ame-
rican communities. However, few provide more than qualitative evidence
relating respiratory symptoms, disease rates or pulmonary function changes to
airborne particles and/or sulfur oxides. The few studies evaluated earlier by
U.S. EPA (1982a) as providing quantitative evidence for respiratory system
effects due to long-term PM and/or SO exposure are summarized in Table 2.
s\
One series of studies, reported on from the early 1960s to the mid-1970s,
was conducted by Ferris, Anderson, and others (Ferris and Andersen, 1962;
Kenline, 1962; Andersen et al. , 1964; Ferris et a!., 1967, 1971, 1976). The
initial study involved comparison of three areas within a pulp-mill town
3-28
-------
00
LU
i — i
Q
ID
1—
00
•X
J ^
C\J LU
00 —1
CD
* — * i — i
«^
—i
— 1 0
•?" T~
o o
o
o
LU 1—
i— i oo
\— I—
•=c o
1— LU
1 — 1 LJ_
] — LL.
Z LU
3 DT
o-i—
— 1
u_ <=n
O LU
T~
>_
rv CU
•a: z
s: i—
oo —i
LU
rv
,
CM
LU
1
CQ
I—
, — ,
CO
E
•\
cn
~^
s *
1 —
QJ
> i-
OJ QJ
r— -P
+J
-p re
C E
re
-P QJ
=3 -P
r— re
O 3
Q- CJ
«l—
Ol -P
cn s-
re re
i- D.
OJ
re
S-
(—
[
C\l
OJ
u
c:
QJ
5-
QJ
M-
QJ
i"^
OJ
0
00
0-
00
1—
OO
CQ
73
OJ
^>
i-
OJ
CO
o
o
CO
•p
CJ
QJ
4-
LU
>j
-^j
3
] s
CO
4-
O
OJ
Q.
h-
, — ,
I —
CO
CT1
-H
•^_s
,—
n3
+J
01
t —
cr
33
1
cn
r^
C\J
i
r-H
00
i— 1
1
^
1— 1
CD
OO
1
0
00
>: 1
U Q.
C E
oi >, cn c.
3 co c T-
cr =5
Ol >,i— C
S- S- OJ
4- O 73 i- 73
-P Ol 73 C
"O rO <-0 r— • rO
oj i- re T- i —
co T— OJ .d cn
re a. i- u c
QJ CO (J LU
S- O) OJ C
(J S- 73 T- -
(~~ "O
••- S- 73 C i—
QJ C O OJ
>, 3 re T- •<-
c— O -P 4-
QJ r — CO U 4-
-^ E C Ol
•r- 4- 0 =! -C
— 1 O -P 4- CO
r—
re
o
-p
CJ s~^
QJ CO
co re
1 O)
CO S-
co re
0
O ' — '
. / — .
•— CO
re r~~
cn
-p < — i
Oi
f
CO OO
•<— r —
r cr>
5 -^
u_
•X
•X
o
00
i-H
1
73
OJ •
CO L
re
OJ Z
S-
c co cj -
•r- -P QJ C
t— 73 -r-
-P 3 I
C 73 -C t-
QJ fO +J QJ
E •!- CQ
> 0 C
O C -r-
S- C 0
Q- O •!- C
E T- -P 0
"" U •!- -P
-P C L) 3
C 3 O r—
QJ 4— CO i —
i- CO O
re cn re o_
Q. C
a. =i c: s:
- —i -
a; •—*
LL.
•X
^
r— 1
00
rH
1
O
00
.p
C. C.
OJ •!-
s-
re c
CO O- O
-P Q-'i-
u re -p .
OJ f 1 T"
4- O C
4- C 3 Z
OJ 4-
73
4- C CD C
o re c -r-
3 1 —
-i*i " 1 — 1-
CJ CO OJ
re E c: CQ
1 — O "r—
•P C
-P Q- Ol -i-
c E '/i
oj >, re to
S- CO O) -P
re s- i—
Q.73 CJ =!
Q- C OJ 73
S- -P -P
QJ _Q CO C -r-
— -r- •!- • S-
.— Ol — CJ
co re i/i s_
CO U OJ OJ CXJ
re — o~i "o
E U +^ ^_
-^4- ••!- -i- • ' .
CO T- O i- +-
CQ U -P Ol C
— • QJ QJ C T- Ol
Q_ E OJ i — E
S CO •!- tO 5-3
CL. i +J oi re cj
Ol S- OJ O
QJ -P QJ Q- 73
-p -r- -C OJ
re 10 .P >> .c /-N
E r— -P re
•i- 4- Oli— CM
x o c re 4-00
o -r- s- o cn
t- Ol 5- QJ r— 1
CL CO 3 C C — '
CL 3 73 QJ O
re i cn -i— CJ -C
-P O) Ol C -P
re QJ CQ i — T-
E -C CO O
•i— -P C OJ -P
-P -i— O 73
co cn co oi E
OJ C CO -C 3
•i— i — CO CO 73
Ol 73 QJ E •<- C
73 t- > \. < — QJ
3 re QJ O1 JD. 73
S- Cn i— n. 3 73
CJ QJ Q-<;
S- S- 0
>> 3 en co QJ
i — CO M— 1 OJ -C
C Ol r- en •!- -P
O -I- 3 CSJ 73
-P CO V 34-
CO -r— -P O
re 3 cj -p co
cn -i— re £-
73 -I- s- .c co 3 QJ
OJ-CtcO OI-P-P OJ-P
SEE -c cj c a.
oi re •-», Q--P QJ re
•i— Cn l/l CO 4— 4— -CI
>C3_ OQJ4- OCJ
OJ E Cn QJ
oi > c -P cn co -p
jzi -r- -i- re 3 .c -PC
en co -p i— QJ
•P co 3 i — 3 co
co -i — o re re co QJ
3COQJ i — -PQJ CUS-
e -P > re -c s- Q-
U OJ 73 73
co OJ i— OJ 73 Ol OJ
E 4- -P -P OJ -C -C
^4-00 rero> -P-P
cn oj CQ cj .c £-
3- - -r- -P 01 -P 4-
-C 73 73 CO re O
C -P OJ C E -Q JC
•1— 1 -P •!- O O -P-P
TO S- i- X
QJOJO C04—O OJOI
S- JZ D. 73 -P -P-P
QJ OJ O CO O
-C73S- -Cr — QJ :ZOJ
OJ +J OJ +J -C
73 > OJ OJ > 3 -p
QJ S- -C E OJ jQ
-P OJ -P i — -i- • — . C
re co oi s- re T-
•P JD OJ -P OJ.-P CXJ
coo> reoc 0073
•i- i. oo o cn , S-
> T- r- <=C CJ
QJ.CO -PCQJ CL.CO
t — -p-P reo-^ LUOI
•r- 4— •(—•(— 73
CO 3 73 i— -P i—
CQ t— 3 re 00 OJ
73 O) co E -P • S-
-p QJ -i- -r- o ^3 re
re -P 4- -P -p c
-c re 4- re co -p
-P -i— Ol -C Ol 73 . . C
U -C +-> •!— QJ QJ
OJ O OO Ol 73 U E
-P CO QJ 73 S- 3
OCOC -P373 DO
zro-i- os-c oo
* z o re oo 73
* -X
•x *:
•X
3-29
-------
(Berlin, New Hampshire). Kenline (1962) reported average 24-h S02 levels
(estimated from sulfation rates) during a limited summer sampling period
(August-September, 1960) to be only 16 ppb and average 24-h TSP levels for the
o
two-month period to be 183 [jg/m . In the original prevalence study (Ferris
and Anderson, 1962; Anderson et al., 1964), no association was found between
questionnaire-determined symptoms and lung function tests assessed in the
winter and spring of 1961 in the three areas with differing pollution levels,
after standardizing for cigarette smoking. The authors discuss why residence
is a limited indicator for exposure (Anderson et al., 1964). The study was
later extended to compare Berlin, New Hampshire, with the cleaner city bf
Chilliwack, British Columbia in Canada (Anderson and Ferris, 1965). Sulfation
rates (lead candle method) and dustfall rates were higher in Berlin than in
Chilliwack. The prevalence of chronic respiratory disease was greater in
Berlin, but the authors concluded that this difference was due to interactions
between age and smoking habits within the respective populations.
The Berlin, New Hampshire, population was followed up in 1967 and again in
1973 (Ferris et al., 1971, 1976). During the period between 1961 and 1967, all
measured indicators of air pollution fell, e.g., TSP from about 180 ug/m in
1961 to 131 pg/m3 in 1967. In the 1973 follow-up, sulfation rates nearly
doubled from the 1967 level (0.469 to 0.901 mg 50,,/100/cm2 day) while TSP
3
values fell from 131 to 80 p.g/m . Only limited SOp data were available (the
mean of a series of 8-h samples for selected weeks). During the 1961 to 1967
period, standardized respiratory symptom rates decreased and there was an in-
dication that lung function also improved. Between 1967 to 1973, age-sex
standardized respiratory symptom rates and age-sex-height standardized
pulmonary function levels were unchanged. Although some of the testing was
done during the spring versus the summer in the different comparison years,
Ferris and coworkers attempted to rule out likely seasonal effects by retesting
some subjects in both seasons during one year and found no significant differ-
ences in test results. Given that the same set of investigators, using the
same standardized procedures, conducted the symptom surveys and pulmonary
function tests over the entire course of these studies, it is unlikely that the
observed health endpoint improvements in the Berlin study population were due
to variations in measurement procedures, but rather appear to have been associ-
Q
ated with decreases in TSP levels from 180 to 131 ug/m . The relatively small
changes observed and limited aerometric data available, however, argue for
caution in placing much weight on these findings as quantitative ir>H--
3-30
-------
effect or no-effect levels for health changes in adults associated with chronic
exposures to PM measured as TSP.
The earlier criteria review (U.S. EPA, 1982a) also noted a cross-sectional
study conducted by Bouhuys et al. (1978) in two Connecticut towns in which dif-
ferences in respiratory and pulmonary function were examined in 3056 subjects
(adults and children). Hosein, et al. (1977a) reported on aerometric data used
in the study, which were obtained at three sites in Ansonia (urban) and four
sites in Lebanon (rural) near the residences of study subjects. The TSP levels
during the period of the study in Lebanon and Ansonia were 39.5 and 63.1 ug/m
o
and SOy levels were 10.9 and 13.5 |jg/m , respectively. Site-to-site variations
on the same day were frequently significant in Ansonia and also occurred in
Lebanon. During the years 1966-72, annual average TSP levels in Ansonia ranged
3
from 88 to 152 ug/m . No historical data for S02 or TSP in Lebanon were pro-
vided. Size fractionation (Hosein, et al. 1977b) of a limited number of TSP
samples in Ansonia showed 81 percent of the TSP sample to be 9.4 urn or less in
diameter. Binder et al. (1976) obtained for 20 subjects in Ansonia one 24-hour
measure of personal air pollution exposure for particles (<7 urn diameter), S0?,
and NO™. Subjects with smokers in the home were exposed to significantly higher
levels than those without such exposure. Personal exposure and outdoor expo-
sures were also significantly different. The mean personal respiratory particle
3 3
level was 114 ug/m as compared to the outdoor TSP level of 58.4 ug/m .
An extended version of the MRC Questionnaire was administered via a com-
puter data-acquisition terminal (Mitchell et al., 1976) between October 1972
and January 1973 in Lebanon and from mid-April through July 1973 in Ansonia.
For children 7 to 14 yrs, the response rate varied from 91 to 96 percent for
boys and girls. For adults (25 to 64 years) the response rate was 56 percent
in Ansonia and 80 percent in Lebanon. After analysis of non-responder versus
responder differences, the responders were considered to be representative of
the total population, although some significant differences were noted between
responders and non-responders for some symptom reporting and current smoking in
some age groups.
Bouhuys et al. (1978) found no differences between Ansonia and Lebanon for
chronic bronchitis prevalence rates but did note that a history of bronchial
asthma was highly significant for male residents of Lebanon (the cleaner town)
as compared to Ansonia (the higher-pollution area). No differences were
observed between the communities for pulmonary function tests adjusted for sex,
age, height and smoking habits. However, prevalence for three of five symptoms
3-31
-------
(cough, phlegm, and plus one dyspnea) were significantly higher for adult
non-smokers in Ansonia (p <0.001). The mix of positive and negative health
effect results found by this cross-sectional study make it difficult to
interpret. Although few air pollution effects, were observed, the
statistically significantly increased symptom rates raise questions as to
whether some impact on health (due to prior PM exposures, for example) might
have occurred. A follow-up longitudinal . examination could have determined
whether the effects persisted. Also, it may be that the reported effects
related more to historical rather than current pollutant levels or to occupa-
tional exposures which were not examined.
The 1982 Criteria Document (U.S. EPA, 1982a) further indicated that
apparent quantitative relationships between air pollution and lower respiratory
tract illness in children were reported by Lunn et al. (1967), who studied
respiratory illnesses in 5- and 6-year old school children living in four areas
of Sheffield, England. Air pollution levels showed a gradient in 1964 across
the four study areas, the mean 24-hour smoke (BS) concentrations ranging from
o
97 to 301 (jg/m . During 1965, annual BS concentrations of smoke were about 20
percent lower and SOp about 10 percent higher, but the gradient was preserved
for each pollutant. In high-pollution areas, individual 24-h mean BS levels
o
exceeded 500 ug/m 30 to 45 times in 1964 and 0 to 15 times in 1965 for the
o
lowest and highest pollution areas, respectively. SOp exceeded 500 ug/m 11 to
32 times in 1964 and 0 to 23 times in 1965 for the lowest and highest pollution
areas, respectively. Information on respiratory symptoms and illness was
obtained by questionnaires completed by parents, by physical examination, and
by tests of pulmonary function (FEVg j,- and FVC). Socioeconomic factors (SES)
were considered in the analyses, but parental smoking and home-heating systems
were not. Although some differences in SES between areas were noted, gradients
between areas existed even when the groups were divided by social class, number
of children in house, and so on. Positive associations were found between air
pollution concentrations and both upper and lower respiratory illness. Lower
respiratory illness was 33 to 56 percent more frequent in the higher pollution
areas than in the low-pollution area (p <0.005). Also, decrements in lung
function, measured by spirometry tests, were closely associated with respira-
tory disease symptom rates.
Lunn et al. (1970) also reported results for 11-year-old children studied
in 1963-64 that were similar to those found earlier for the younger group1;
Upper and lower respiratory illness occurred more frequently in children
3-32
-------
exposed to annual average 24-h mean smoke (BS) concentrations of 230 to 301
3 3
jjg/m and 24-h mean S0~ levels of 181-275 ug/m than in children exposed to
smoke (BS) at 97 ug/m and S0? at 123 M9/m • This report also provided
additional information obtained in 1968 on 68 percent of the children who were
5 and 6 years old in 1963-64. By 1968, the reported BS levels were only about
one-half those measured in 1964, S0? levels were about 10 to 15 percent below
those of 1964, and the pollution gradient no longer existed; so the combined
three higher pollution areas were compared with the single original
low-pollution area. Lower respiratory illness prevalence measured as "colds
going to chest" was 27.9 percent in the low-pollution area and 33.3 percent in
the combined high-pollution areas, a difference not statistically significant
at p >0.05. Ventilatory function results were similar. Also, the 9-year-old
children had less respiratory illness than the 11-year-old group seen previous-
ly. Because 11-year-old children generally have less respiratory illness than
do 9-year olds, this represented an anomaly that the authors suggested may be
due to improved air quality.
These Lunn et al. (1967, 1970) findings have been widely accepted (World
Health Organization, 1979; Holland et al. , 1979; U.S. EPA, 1982a,b) as valid.
On the basis of the results reported, it appears that increased frequency of
lower respiratory symptoms and decreased lung function in children may occur
3
with long-term exposures to annual BS levels in the range of 230 to 301 ug/m
3
and S0~ levels of 181 to 275 (jg/m . However, these are only very approximate
observed-effect levels because of uncertainties associated with estimating PM
mass based on BS readings. Also, it cannot now be concluded, based on the 1968
follow-up study, that no-effect levels were demonstrated for BS levels in the
o
range of 48 to 169 [jg/m because of: (1) the likely insufficient power of the
study to have detected small changes given the size of the population cohorts
studied, and (2) the lack of site-specific calibration of the BS mass readings
at the time of the later (1968) study. In summary, the Lunn et al. (1967)
study provided the clearest evidence cited in the 1982 EPA criteria document
(U.S. EPA, 1982a) for associations between both significant pulmonary function
decrements and increased respiratory disease illnesses in children and chronic
exposure to specific ambient air levels of PM and SO^.
Since the earlier criteria review (U.S. EPA, 1982a), results of analyses
of data from the ongoing Harvard study of outdoor air pollution and respiratory
health status of children in six cities in the eastern and midwestern United
3-33
-------
States have been reported by Ware et al. (1986). Between 1974 and 1977, ap-
proximately 10,100 white preadolescent children were enrolled in the study
during three successive annual visits to the cities. On the first visit, each
child underwent a spirometric examination and a parent completed a standardized
questionnaire regarding the child's health status and other important
background information. Most of the children (8,380) were seen for a second
evaluation one year later. Measurements of TSP, the sulfate fraction of TSP
(ISO.), and S02 concentrations at study-affiliated outdoor stations were
combined with data from other public and private monitoring -sites to create a
record of TSP, ISO., and S02 levels in each of 9 air pollution regions during a
one-year period preceding each evaluation, and for TSP during each child's
lifetime up to the time of evaluation.
Analyzing data across all six cities, Ware et al. (1986) found that fre-
quency of chronic cough (see Figure 5) was significantly associated (p <0.01)
with the average of 24-hr mean concentrations of all three air pollutants (TSP,
TSO,, S02) during the year preceding the health examination. Rates of bron-
chitis and a composite measure of lower respiratory illness were significantly
(p <0.05) associated with annual average particulate concentrations, as well as
being related to measures of lifetime TSP concentrations. However, within the
individual cities, temporal and spatial variation in air pollutant levels and
symptom or illness rates were not significantly associated. The history of
early childhood respiratory illness for lifetime residents was significantly
associated with average TSP levels during the first two postnatal years within
cities, but not between cities. Furthermore, pulmonary function parameters
(FVC and FEV.J were not associated with pollutant concentrations during the
year immediately preceding the spirometry test (see Figure 6) or, for lifetime
residents, with lifetime average concentrations, although Ferris et al. (1986)
reported a small effect on lower airway function (MMEF) related to FP
concentrations.
Overall, these results appear to suggest that risk may be increased for
bronchitis and some other respiratory disorders in preadolescent children at
moderately elevated TSP, TSO, and S02 concentrations, which do not appear to
be consistently associated with pulmonary function decrements. However, the
lack of consistent significant associations between morbidity endpoints and air
pollution variables within individual cities argues for caution in interpreting
the present results. For example, it might be argued that the non-significant
3-34
-------
CHRONIC COUGH
175
150 —
125 —
o
o 100
T"
cc
Ul
0.
iu 75
50
25
25 50 75 100
MEAN TSP, (//g/m3)
125
150
Figure 5. Adjusted frequency of cough for the 27 region-cohorts from the Six-Cities
Study at the second examination plotted against mean TSP concentration during the
previous year, with between-cities regression equation. LEGEND: P=Portage,
T=Topeka, W-Watertown, C=Carondolet, L=Other St. Louis, R=Steubenville Ridge,
V=Steubenville Valley, K=Kinston, H=Harriman.
Source: Ware et al. (1986).
3-35
-------
FEV., AT SECOND EXAMINATION
1.64
1.62
1.60
OJ
1.58
1.56
1.54
1.52
w
w
w
25 50 75 100 125 150
MEAN JSP,
Figure 6. Adjusted mean percent of predicted FEV., at the first examination for the 27
region-cohorts from the Six-Cities Study plotted against mean TSP concentration
during the previous year, with between-cities regression equation. The slope is not
significantly different from 0. LEGEND: See Figure 5.
Source: Ware et al. (1986).
3-36
-------
associations within cities but significant symptom increases in relation to air
pollutant gradients across the cities may reflect spurious correlations across
the cities. On the other hand, the within city variation in air pollutant
gradients and/or size of study populations within particular cities may not be
sufficiently large to detect associations between the health endpoints and air
pollutant variables included in the analyses. Also, the PM indices employed in
the analyses (e.g., TSP, etc.) may provide a "diluted" measure of exposure to
the most highly toxic PM components (e.g., FP or small coarse-mode particles).
In fact, the reported stronger associations between ISO, levels and other
measures of ambient air FP concentrations are highly suggestive of possible
associations between health effects observed in the Ware et al. (1986) study
and exposure to small particles in contemporary U.S. atmospheres. Available
data (Spengler and Thurston, 1983) from air monitors sampling inhalable parti-
culates (IP; <15 urn) in the same cities included in the Harvard Six Cities
Study analyses discussed here indicate IP mass annually averaged from approxi-
o
mately 20 to 60 |jg/m . This suggests that the observed health effects noted
above may be associated with annual average IP (<15 urn) concentrations below 60
o
ug/m . However, full interpretation of the strength and significance of these
findings is difficult at this point, in light of further follow-up of these
children still being in progress and the expectation that longitudinal analyses
will later be carried out which will relate health data to more extensive
aerometric data (including such data collected in later years).
In another series of studies conducted during the last few years, Ostro
and co-workers evaluated relationships between air pollution indices for 84
standard metropolitan statistical areas (SMSA's) mostly of 100,000 to 600,000
people in size, and indices of acute morbidity effects, using data derived from
the National Center for Health Statistics (NCHS) Health Interview Survey (HIS)
of 50,000 households comprising about 120,000 people (Ostro, 1983; Hausman et
al., 1984; Ostro, 1987).
Ostro (1983) used HIS data to assess the prevalence of illness and
illness-related restrictions in activity in the United States. Data on either
restricted activity days (RADs) or work loss days (WLDs) were aggregated over a
year, and correlated with annual TSP levels, controlling for temperature, wind,
precipitation, population density, smoking, etc. Using the 1976 survey, a
significant relationship between TSP and both outcomes was found, with RAD's
3-37
-------
showing a more significant relationship. The explained variation was much
higher for RADs than for WLDs. This is expected since the decision to take a
day off from work depends on many idiosyncratic factors besides if illness is
present. The average of air pollution monitors for each city was used, rather
than aerometric data aggregated for smaller geographic units in relationship to
individuals residing nearby for whom HIS data were included in the analysis.
The use of city wide TSP data therefore increased possible error in the
exposure variable, but this would more likely bias the results toward zero,
rather than towards finding a significant effect. The Ostro (1983) analysis
was also only for one year of data, and thus Was unable to demonstrate
consistency across years. On the other hand, the use of 84 cities in the Ostro
(1983) analysis reduced the chance that the particular choice of cities
spuriously induced a relationship between air pollution and morbidity due'to
some omitted cofactor. In sum, this first paper suggested a potential
relationship between morbidity and air pollution, which must be viewed with
caution because of the ecological nature of the data, the less than perfect fit
of the annual pollution and acute morbidity variables, and because of the
possibility that results in one year could have occurred by chance.
The Hausman et al. (1984) paper analyzed the same data, but made three
important methodological advances. It used a Poisson specification for the
model, used a fixed effects model that only looks at deviations from the city
mean levels of illness, and used short-term pollution as the exposure variable.
Poisson analysis is appropriate for analyzing low probability events, which is
the case with these morbidity symptoms. The fixed effect model effectively
controls for differences between cities in morbidity levels. This avoids the
potential bias of attributing intercity differences in disease rates to inter-
city differences in pollution. Two-week average TSP levels are used as the
exposure variable. Significant associations between pollution levels and RADs
or WLDs were still found. The magnitude of the within city effects was similar
to the magnitude of the between city effects seen earlier. Again demographic
factors were controlled for on an individual basis, along with climatic
conditions. This analysis considerably strengthens the plausibility of the
association, particularly because of the within city effect. However, it still
only analyzed data for one year, which may be anomalous.
In the most recent analyses reported, Ostro (1987) applied the Hausman et
al. (1984) techniques to analyze HIS results from 1976 to 1981 in relation to
3-38
-------
estimates of fine particle (FP) mass. That is, for adults aged 18 to 65, days
of work loss (WLDs), restricted activity days (RADs) and respiratory-related
restricted activity days (RRADs) measured for a two-week period before the day
of the survey were used as measures of morbidity and analyzed in relation to
estimated concurrent two-week averages of FP or lagged in relation to estimated
2-wk FP averages from 2 to 4 weeks earlier. The FP estimates were produced
from the empirically derived regression equations of Trijonis. These
equations, as used here, incorporated screened airport data and two-week
average TSP readings at population-oriented monitors, using these data taken
from the metropolitan area of residence. Various potentially confounding
factors (such as age, race, education, income, existence of a chronic health
condition, and average two-week minimum temperature) were controlled for in the
analyses. Various morbidity measures (WLDs, RADs, RRADs), for workers only or
for all adults in general, were consistently found to be statistically signifi-
cantly (p <0.01 or <0.05) related to lagged FP estimates (for air quality 2 to
4 weeks prior to the health interview data period), when analyzed for each of
the individual years from 1976 to 1981. However, less consistent associations
were found between the health endpoints and more concurrent FP estimates.
The approach employed by Ostro to estimate PM levels introduces into his
analyses a number of uncertainties, such as those inherent in airport
visibility measurements, FP/visibility relationships, and TSP monitoring
limitations (most notably, use of the Trijonis equations characterizing FP
relationships to visibility in northeastern U.S. areas may not be appropriate
for western U.S. cities); and use of single average pollutant levels to
estimate exposures for an entire city's population. Use of the spatially
averaged indicator over time within a specific area should reduce some of these
uncertainties, but it is unlikely that more than qualitative relationships
between PM levels estimated in this fashion and morbidity effects could be
derived. Additional uncertainties derive from use of the HIS data base, with
the vast majority of data points being "0", representing no incidences of
indicator effects being recalled in the prior two weeks. However, use of the
Hausmann et al. (1984) statistical approach should have adequately dealt with
this problem.
The overall patterns of results obtained from the reported analyses are
interesting but difficult to interpret. They may suggest that acute morbidity
effects are associated with fine-mode particle exposures occurring 2-4 weeks
3-39
-------
earlier, but less so with immediately prior FP exposures. Such a possibility
cannot be ruled out in view of findings reported by other investigators regard-
ing lag structures in data bases relating mortality or morbidity to PM expo-
sures. Nevertheless, these Ostro analyses have found consistent associations
between PM and morbidity measures for adults that are reasonably consistent
between and within contemporary American cities. As such, the results tend to
reinforce the plausibility of the Ware et al. (1986) findings of associations
between morbidity measures in children and PM concentrations found in contem-
poraneous American urban air sheds. However, the Ostro analyses do not allow
for the estimation of quantitative relationships between morbidity effects and
more usual 24-hr or annual average direct gravimetric measures of particulate
matter air pollution (e.g., TSP, PM10, etc.).
In another new American study, by Schenker et al. (1983), respiratory
symptom questionnaires were administered to 5557 adult women in a rural area of
western Pennsylvania. Air pollution data (including SO,, but not PM measure-
ments) were derived from 17 air monitoring sites and stratified in an effort to
define low, medium and high pollution areas. The means of 4-yr (.1975-1978)
3
annual average S0? levels in each stratum were 62, 66, and 99 ug/m , respec-
tively. Risks for respiratory symptoms were assessed by a multiple logistic
model that controlled for several potentially confounding factors (e.g.,
smoking) and used estimated air pollution concentrations at population-weighted
centroids of 36 study districts (i.e., the concentrations were derived from
another model which weighted observed monitoring data for distance from the
district centroid and corrected for terrain effects). The risk of "wheeze most
days or night" in nonsmokers residing in the high- and medium-pollution areas
was 1.58 and 1.26 (p = 0.02), respectively, in relation to the low-pollution
area. For residents living in the same location for >5 yr, these relative
risks were 1.95 and 1.40 (p <0.01), and increased risk of grade 3 dyspnea in
nonsmokers was associated with S02 levels at p <0.11. However, no significant
association was observed between cough or phlegm and air pollution variables.
The results of this study, while suggesting that wheezing may be qualitatively
associated with ambient exposure to S02, are difficult to accept in light of:
(1) the very limited gradient of annual-average S02 levels across which health
effects were reported to have been detected (associations with higher level
exposures versus distinctly lower S02 concentrations would be more credible);
(2) the very rough estimation of S0? exposure concentrations by means of model
3-40
-------
calculations; and (3) the lack of evaluation of possible PM or short-term S02
peak contributions to the evaluated health effects.
Several other recent studies have been reported that evaluated PM and/or
SO effects in populations residing in the southwestern United States. In one,
Chapman et al. (1985) conducted a survey in early 1976 regarding the prevalence
of persistent cough and phlegm (PCP) among 5,623 young adults in four Utah
communities stratified to represent a gradient of sulfur oxides exposures.
Community-specific mean S02 levels had been 11, 18, 36 and 115 ug/m during the
5 years prior to the survey and corresponding mean sulfate levels were 5, 7, 8,
and 14 ug/m3. No gradients of TSP or suspended nitrates were observed across
the communities. Aerometric data were obtained from monitors sited at ground
level. Differences along the sulfur oxides gradient were tested by chi-square
statistics, and data were also analyzed by constructing categorical logistic
regression models that treated PCP as the dependent variable and controlled for
numerous potentially important factors (e.g. smoking, age, SES, etc.). For
nonsmoking mothers, PCP prevalence was 4.2 percent in the high-exposure com-
munity and ~2.0 percent in all other communities. For non-smoking fathers, the
PCP prevalence was 8.0 percent in the high pollution community and 3.0 percent
elsewhere, while the PCP prevalence was less strongly associated with ambient
sulfur oxides exposures for smoking fathers. Overall, intercommunity, preva-
lence differences were significant at p <0.05 for all the above groups except
smoking fathers. The categorical logistic regression model yielded similar
results, providing evidence suggestive of increased cough and phlegm being
o 2
associated with annual average 115 ug/m S02 levels and/or 14 ug/m sulfate
levels. There is much to argue for acceptance of the reported results from
this study, including use of aerometric data from monitors situated in close
proximity to study subjects' homes and nearly equivalent response rates on the
health questionnaire across the communities sampled.
Dodge (1983) studied the respiratory health and lung function of Anglo-
American children (grades 3 to 5) residing in an Arizona smelter community
versus such children residing in another small Arizona community free of
smelter air pollution. Cough prevalence was 25.6 percent in the smelter town
children and 14.3 percent in the non-smelter groups (p <0.05). Baseline
pulmonary function at the outset of the study was equal in the two groups, and
over the four years of the study, lung function growth (measured in terms of
FEV-, after 4 yr. of study minus predicted FEV^) was also equal between the two
3-41
-------
groups. During the study, annual average SCL levels were 55 and 48 ug/m at
company and state monitoring sites, respectively (highest 24-hr SO., levels were
3
611 and 524 ug/m , respectively, at the company and state sites). Annual
average TSP was 28 ug/m in the smelter community. These results suggest that
smelter community children had more cough than the control group children but
no evident differences in lung function. However, it is difficult to ascribe
the reported effects specifically to SOp or TSP (although the very low levels
of the latter are unlikely to account for the effects).
Dodge et al. (1985) more recently reported on a longitudinal study of
children exposed to markedly different concentrations of SCL and moderately
different levels of particulate sulfate (SOp in Southwestern U.S. towns. Four
groups of subjects lived in two areas of one smelter town and in two other
towns, one of which was also a smelter town. In the highest pollution area,
the children were exposed intermittently to high S00 levels (peak 3-hr x
3 -
exceeded 2,500 ug/m or ~1.0 ppm) and moderate particulate SOA levels (x = 10.1
3
ug/m ). When children were grouped by the four observed pollution gradients,
the prevalence of cough (measured by questionnaire) correlated significantly
with pollution levels (trend chi-square = 5.6; p = 0.02). No significant
differences occurred among the groups of subjects over 3 years, and pulmonary
function and lung growth over the study were roughly equal over all groups.
The results tend to suggest that intermittent high level exposures to S0?, in
the presence of moderate particulate sulfate levels, produced evidence of
bronchial irritation (increased cough) but no chronic effect on lung function
or lung function growth. It is difficult to quantitate the SO,, levels specifi-
cally associated with the observed effects, although the intermittent high
level exposures to ~1.0 ppm (3 hr averages) mentioned earlier are likely
implicated. Note that S09 levels for the higher polluted smelter town annually
o
averaged 103 ± 282 (S.D.) ug/m (indicating wide variability in the one hr mean
o
levels) versus 14 ug/m in the lesser polluted town. Other measured air
pollutants, e.g. TSP, differed little between the high and low pollution areas
_ o
(24-hr TSP x = 52 and 58 ug/m , respectively). The observation of increased
cough but lack of lung function changes in children comports well with the
findings of Ware et al. (1986).
Lebowitz et al. (1982) studied 117 families in Tucson, Arizona, selected
from a stratified sample of families in geographical clusters from a represen-
tative community population included in an ongoing epidemiologic study. Both
3-42
-------
asthmatic and non-asthmatic families were evaluated over a two year period,
using daily diaries; and the health data obtained were related to various in-
dices of environmental factors derived from simultaneous micro-indoor and out-
door monitoring in a representative sample of houses for air pollutants,
pollen, fungi, algae and climate. Macromonitoring of air pollutants and pollen
was carried out simultaneously. The data were mainly evaluated in terms of
statistical techniques employing contingency tables and frequency distributions
using SPSS programs. Two-month averages of indoor TSP ranged from 2.1 to 169.6
o
[jg/m . Cyclone measurements of respirable particulate (RSP) ranged from below
o
readable limits up to 28.8 pg/m . CO and NO measurements were also taken, but
/\
no SOp monitoring was reported. Suspended particulate matter and pollen were
reported to be related to symptoms in both asthmatics and non-asthmatics, but
the authors reported that the statistical analyses used were all qualitative
(becase of low sample size) and statistical significance was not computed.
In a recently published Canadian study, Pengelly et al. (1986) reported
results for an ongoing study of associations between particle size and respira-
tory health in children of Hamilton, Ontario. From 1979 to 1982, a cohort of
approximately 3500 elementary school children was studied by determining each
child's health history and respiratory symptoms by means of a questionnaire
administered to their parents. Also, pulmonary function tests were conducted
on the children at school. Particle size and concentrations were determined by
using two networks distributed across the city, one consisting of 7 to 9
Anderson 2000 Cascade impactors and another of 27 hi-vol TSP samplers.
Smoking, use of gas for cooking, SES and other potentially confounding factors
were assessed by parental questionnaire and controlled for in statistical
analyses, i.e., stepwise multiple regression techniques (linear for continuous
dependent variables and logistic for binary dependent variables).
In the present report, Pengelly et al. (1986) focused on two indicators of
respiratory health (cough and bronchitis episodes) and two indicators of
pulmonary function (peak expiratory flow or PF and MEFyr), both adjusted for
body size. Logistic regression analyses found no significant associations
between cough or bronchitis episodes and air pollution indices, correcting for
other factors. Both peak flow and MEFy,- (adjusted for height) were reported to
be significantly associated with the presence of fine particles. However, the
fine fraction (FF) was estimated by adding results for samples collected by the
lower stages of a cascade impactor (nominally reflecting sizes <3.3 pm). Based
3-43
-------
on particle bounce problems associated with this impactor (see discussion in
Chapter 1) and comparison measurements made by the authors in Hamilton between
dichotomous fine (<2.5 pm) and the cascade FF, additional coarse material >3.3
|jm was probably also included in the FF measured by Pengelly et al. (1986).
Overall the FF mass was more than double the dichotomous sampler fine mass.
Also since preparation of the earlier criteria review (U.S. EPA, 1982a),
additional analyses of health effects relationships to PM and SO air pollution
/\
in European cities have emerged. Some of the new European work includes
longitudinal analyses reported by van der Lende et al. (1986) as being conduct-
ed in regard to evaluating relationships between prevalence of respiratory
symptoms and pulmonary function decline and variations in air pollution in two
areas of The Netherlands. That is, health measurements were obtained from
cohorts of approximately 2000 men and women (aged 15 to 64 years), residing in
a highly polluted area (Vlaardingen) or a non-polluted rural area (Vlagtwedde),
with subjects being followed and examined at intervals of three years. Over
the course of the study, air pollution levels (PM measured as British smoke,
SOp, etc.) remained consistently very low in the latter area, whereas pollution
levels declined over time in the former, highly polluted area. Van der Lende
et al. (1986) noted that in a previous publication, they reported both a
significantly higher prevalence of respiratory symptoms in the polluted area
and also a greater decline there in pulmonary function (based on four consec-
utive studies over a 9-year period). In the present update paper (van der
Lende et al., 1986), further findings are provided regarding associations
between respiratory symptoms and pulmonary function decline and air pollution
after six consecutive studies covering a 15-year period. The results, termed
"preliminary" by the authors, provide some indications of more respiratory
symptoms and greater pulmonary function declines in the polluted area than the
control, non-polluted area. However, as currently available, the reported
results do not allow for any quantitative conclusions to be clearly drawn
regarding PM levels associated with observed health effects.
In another study (PAARC, 1982a,b; Lellouch, 1986) relationships between
atmospheric pollution and chronic or recurrent respiratory diseases were evalu-
ated from 1974 to 1976 as part of a French national survey in 28 areas of 7
cities and a newly industrialized region. The following pollutants were mea-
sured: S02 (specific-SP and acidimetric-AF methods); suspended particles
(smoke and modified OECD gravimetric methods); nitrogen oxides (NO and NO^
3-44
-------
measured by modified Griess-Saltzmann method); and sulfates (measured by
colorimetry after reduction). Samples were obtained over 24 hr. periods, but
for the gravimetric measures (48 to 96 h), from 1974-76 except for one summer
month each year and except for the sulfates which were determined only during
the last half of the study and only in one part of the study zones.
Twenty-eight study zones were defined to include 2-4 groups of ~1000 people in
different cities exposed to pollution that differed as much as possible in
quality and quantity (estimated from earlier aerometric data from 1971-72).
Zones included populations situated within 0.5 to 2.3 km (x = 1.3 km) of air
monitoring stations located 2-4 m above ground level in the center of each
zone. National meteorological services supplied climatic data (e.g.,
temperature and humidity) taken at a station best characterizing each city
(usually an airport, sometimes far from the zones investigated), and laboratory
analyses for the air pollutants measured were carried out by laboratories in
each city studied but for sulfates done at a single laboratory. Means for
daily data for the pollutants studied were calculated for 1974-76 (where values
came from data accumulated over several days, it was assumed the pollution was
the same on each day). The extreme mean daily concentrations from various
zones were: 13 and 127 M9/m3 for S09 (AF), 22 and 85 ug/m3 SO- (Sp); 18 and
T 3 "\
152 ug/m (smoke); 45 and 243 ug/m (gravimetric), 7 and 145 ug/m (NO); and 12
to 61 ug/m3 (N02).
As for health evaluations, ventilatory function was measured in both men
and women aged 25 to 59 and children aged 6 to 10 and respiratory symptoms were
ascertained by standardized questionnaire. The results presented by PAARC
(1982a,b) were for ~20,30Q subjects from 20 zones (response rates varied from
70 to >90 percent in the included zones). Analyses of covariance were used for
FEV results and logistical regression for the analysis of symptoms scores,
taking into account control factors such as smoking and socioeconomic status.
It should be noted that efforts were made to standardize the health endpoint
measurements by common training of personnel carrying out testing in various
zones and use of standard protocols.
The results of the study were reported by PAARC (1982b) as follows: (1)
Among both male and female adults, S02 concentrations are significantly
associated with the prevalence of lower respiratory disease (LRD) symptoms; (2)
Among children, S02 is associated with the prevalence of upper respiratory
disease (URD) symptoms; (3) For both adults and children, FEV^ Q varied
3-45
-------
negatively in relation to elevations in S02 levels; and (4) No other pollutants
were associated with ventilatory function or the prevalence of respiratory
symptoms. More specifically, SO,, concentrations were significantly correlated
(r >0.44) with incidence of cough, expectoration, and LRD symptoms in men and
With LRD incidence in women (r = 0.49); and S02 correlated (r = 0.53)
significantly with URD in children. It was noted that, whereas the above
results emerged from analyses including data drawn from across cities, the
gradient of SOp effects on symptom rates was not always evident within the same
city (an analogous situation to findings reported by Ware et al. , 1986, based
on data from six American cities). Similarly, the gradients emerging from
regressions across cities for relationships between S02 and FEV-, 0 measures for
men (r = -0.52), women (r = -0.67) and children (r = -0.70) were not always
evident from data within all individual cities. In contrast to the S0?
results, very mixed correlations (some positive and some negative, but none
significant) were found between symptoms and measures of PM (smoke or
gravimetric) and nitrogen oxides (NO, NO^). Also, PAARC (1982b) reported that
the correlations between FEV., Q and PM or nitrogen oxides measures were
positive (some significantly so for NO or N0?); i.e., they implied improved
lung function as airborne particle or nitrogen oxides levels increased. The
Lellouch (1986) publication, apparently based on final data analyses utilizing
approximately the same number of subjects as noted in the PAARC (1982b) report,
emphasized the findings noted above for S02-related health effects but stated
that no clear correlations were observed for any other pollutants (i.e.
sulfates, particulate matter, or nitrogen oxides).
The results from the PAARC study (PAARC, 1982a,b; Lellouch, 1986) are
interesting but challenging in terms of interpretation. The study appears to
have ensured that aerometric data from the sampling stations used would be rea-
sonably well representative of the surrounding study populations in the various
zones, a definite strong point of the study. Similarly, efforts to standardize
measurements of health endpoints across the different cities is another strong
point. Also, in the case of the S02 .measurements, acceptable analytical techni-
ques were used and periodic intercomparisons made between laboratories, thus
enhancing the credibility of the S0? aerometric data. Much less confidence can
be placed in the data derived for particulate matter, however, in view of the
use of smoke readings and/or gravimetric readings that varied for 48 to 96 h
periods as the basis for generating estimated particle concentrations to compare
3-46
-------
across cities. It is doubtful that any adequate comparison could be made, then,
across cities in terms of relationships between either symptoms or pulmonary
function measures and PM estimates; analyses relating such health endpoints to
PM measures within individual cities (not reported in PAARC, 1982a,b) might be
more credible, but this remains to be evaluated. As for the significant associ-
ations between S02 and health endpoints reported by PAARC (1982a,b), several
factors limit full acceptance of the reported findings, eg: (1) the S02 and PM
indices were only tested in separate regression analyses; (2) the associations
for S0? and lung function changes were significant for only one of the two types
of S0? measurement methods used; and (3) other uncertainties are introduced by
the lack of control for seasonal effects and parental smoking in the analyses of
childrens1 data.
In another European study from the Commission of the European Communities
(Florey et a!., 1983) reported since the 1982 U.S. EPA criteria document was
prepared, various health endpoints in children (6-11 yrs old) were evaluated in
relation to air pollution in 19 geographic areas located in several different
European Community countries. Data were obtained on 22,337 children and
included information on respiratory symptoms obtained by questionnaire and
pulmonary function measurements (peak expiratory flow rate measured by Wright
peak flow meters). Efforts were made to standardize health measurements and
protocols across all study areas. S02 concentrations were determined (using
six different analytical methods) and particulate pollution was measured by
smoke methods in some countries and by unspecified gravimetric methods in a few
other ones. Side by side monitors were set up at 20 sites to help provide a
basis for calibration across sites; these 20 "comparison" monitoring stations
standardly used the British smoke method for PM and acidimetric method for S02-
Significant associations emerged from analyses within some individual
countries, but differed greatly from one country to another. In three
countries, a composition variable called chronic non-specific lung disease
(CNSLD) was highly significantly correlated positively with smoke, but the
magnitude of the effects differed by a factor of about seven. The range of
annual smoke levels was about the same in all three countries, about 15-40
*3
ug/m . In four countries, there were significant associations with S02, but
two of these were negative. In those with positive correlations annual median
S00 levels were 60-160 ug/m3, and for those with negative associations they
3-47
-------
were 20-120 [jg/m , making it likely that the S02 results reflected chance
variations rather than actual pollution effects. However, no significant rela-
tionships between health effects and particulate pollution were found when data
from across countries were pooled. The reported results are difficult to
interpret. The Commission of the European Communities (Florey et al., 1983)
O
report noted that annual average levels of smoke greater than 140 (jg/m in the
o
presence of SO,, at >180 ug/m have been found by other studies to be levels
above which consistent positive associations between health effects and air
pollution are detectable. These levels are higher than any measured in the
present study, and this might explain the lack of consistent effects observed
from city to city or when data were analyzed across all cities. The results of
analyses for data within a given city may warrant further, more detailed
evaluation and may yield useful information on quantitative exposure-effect
relationships. However, given the great difficulty noted by the Commission of
the European Communities (Florey et al. , 1983) report in deriving bases for
comparing air quality measurements for PM and S02 across different cities it is
dubious that useful quantitative conclusions can be drawn from analyses of data
combined across cities. This is especially the case in view of only limited
calibration of smoke readings against gravimetric measurements by collocated
gravimetric devices in the various countries.
Muehling et al. (1985) also studied the relationship between croup and
obstructive bronchitis of German children taken to clinic versus the level of
air pollutants of their residential areas. They show in this retrospective
study that the incidence of these two diseases was greater in the area with
higher SOp and dustfall levels. Several important confounding factors were
examined (i.e., infection incidence, meteorological parameters, social status,
and distance from clinic). Quarterly average values of S0? and dustfall were
provided by the county of Nord Rhein in Westphalia. The authors state that
their results clearly show that the disease frequency depended on whether the
children lived in an area of high or low S0? and dustfall levels, but noted
that it cannot be clearly stated whether or not the measured emissions are the
actual cause of any increased morbidity.
Wojtyniak et al. (1984) studied the symptoms of persistent cough and
phlegm, bronchitis, and reduced ventilatory capacity in Cracow, Poland. This
cross-sectional study used questions based on the MRC questionnaire. An exten-
sive monitoring network of 20 sampling stations covered the entire area of the
3-48
-------
city. The city was divided into two parts: the city center with suspended
3 3
participate levels averaging 180 pg/m and S02 levels of 114 M9/"1 > and tne
remaining areas having suspended particulate levels averaging 109 pg/m and S02
levels averaging 53 pg/m . Multiple logistic regression models were used to
test for the effects of air pollution, age, smoking history, and other factors.
As expected, smoking history was a highly significant determinant, but high
exposure to air pollution did result in 2.3 times (0.05 < p <0.10) the risk of
exacerbated symptoms in men. In women, the prevalence of exacerbated symptoms
was related to indoor air pollution resulting from coal combustion in stoves.
Because only two pollution exposure areas were used, it was impossible to
separate the effects of particulate matter and sulfur dioxide. The study may
also minimize the effect of pollution because of confounding of smoking and
because of the lack of a true "clean" control area.
In summary, of the numerous new studies published on morbidity effects
associated with long-term exposures to PM or SO , only a few provide poten-
s\
tially useful results by which to derive quantitative conclusions concerning
exposure-effect relationships for the subject pollutants. The Ware et al.
(1986) study, for example, provides evidence of respiratory symptoms in
children being associated with particulate matter exposures in contemporary
U.S. cities without evident threshold across a range of TSP levels from ~30 to
T 3
150 pg/m , with more marked effects notable in the 60-150 ug/m range in com-
parison to lower levels. The increase in symptoms appear to occur without con-
comitant decrements in lung function among the same children. The medical sig-
nificance of the observed increases in symptoms unaccompanied by decrements in
lung function remains to be fully evaluated but is of likely health concern.
Caution is warranted, however, in using these findings for risk assessment
purposes in view of the lack of significant associations for the same variables
when assessed from data within individual cities included in the Ware et al.
(1986) study. The findings derived from the series of studies by Ostro (Ostro,
1983; Hausman et al., 1984; Ostro, 1987), qualitatively indicative of morbid-
ity effects in adults being associated with PM exposures with U.S. cities, tend
to support the plausibility that the associations observed by Ware et al, (1986)
reflect actual morbidity effects in children due to contemporaneous PM and/or
S0? exposures in U.S. cities.
Other new American studies provide evidence for: (1) increased respira-
tory symptoms among young adults in association with annual-average S02 levels
of -115 pg/m3 (Chapman et al., 1985); and (2) increased prevalence of cough in
3-49
-------
children (but not lung function changes) being associated with intermittent
exposures to mean peak 3-hr S09 levels of ~1.0 ppm or annual average levels of
3
~103 |jg/m (Dodge et a!., 1985). It is difficult in regard to each of these
two studies, however, to determine if the reported effects are due to repeated
high-level intermittent exposures to SO- or to more chronic low level exposures
to SOp or its transformation products.
Results from one European study (PAARC, 1982a,b) also tend to suggest that
increased lower respiratory disease symptoms and decrements in lung function
in adults (both male and female) may be associated with annual average S0?
levels in the range from about 25 to 130 ug/m3. In addition that study suggests
that upper respiratory disease and lung function decrements in children may also
be associated with annual-average S02 levels across the above range. The S0?-
morbidity effects associations reported by PAARC (1982a,b), however, cannot be
fully accepted in view of several factors discussed earlier, e.g. internal
inconsistencies between results obtained with analyses using different S0? mea-
surement data and lack of control for some important potentially confounding
factors in certain of the analyses yielding significant results.
3.3 SUMMARY AND CONCLUSIONS
As indicated earlier, although key conclusions from the 1982 criteria docu-
ment (U.S. EPA, 1982a) are concisely summarized at the outset of various chapter
subsections, the main focus of this chapter is on the evaluation of epidemiolog-
ical information on the health effects of PM and SO newly available since pre-
s\
paration of the 1982 document. Furthermore, major emphasis has been placed in
this chapter on identification of the newer epidemiological studies or analyses
which provide quantitative information pertinent to delineation of exposure-
effect or exposure-response relationships.
Table 3 summarizes key conclusions drawn from those newer studies or
analyses evaluated in the present chapter as providing the most pertinent and
useful quantitative evidence for mortality or morbidity effects associated
with short-term human exposures to PM or S0?.
Taking into account the first category of studies in Table 3 and various
considerations discussed above in this chapter, the following conclusions
appear to be warranted based on the earlier criteria review (U.S. EPA, 1982a)
and the present evaluation of newly available analyses of the London mortality
3-50
-------
O
— 1
O E
1—
CO C£
o
— 1 ^^
o -z.
O N
—1 O
O CO
"SL LI_
LU O
Q
)— t CO
Q_ ~J
LU LU
1 |
CQ
- I-H
-J CQ
S S
LU -3I
O
O
LU
Q ce
LU ZO
CO CO
•=c o
CQ O_
X
CO LIJ
O LU
CO ZD
ZD CJ
CJ
"Z, O
0 1—
CJ
LU I—
> CJ
TABLE 3. SUMMARY OF KEY QUANTITAT]
ANALYSES RELATING HEALTH EFFf
xp
*G
^-^
Ol
ZL
' '
, —
OJ
QJ
•— •
-P
TO
O
Q.
0)
O)
TO
QJ
TO
t-
CM
U
QJ
t-
QJ
y_
OJ
CJ
LO
Q_
CO
1—
*
CQ
Results Obtained
•o
4->
CO
u—
o
QJ
0.
j_
CM
CO
~J
^
TO
-P
1-
TO
^
N
^
O
r^^
i
0
0
2
\
o
LO
v
o -P to
.C LO TO TO
oi4J r^-
c -i- i ui 3
•i- 3 o oj o
s- o tn r—
u -r- TO
•a c\j ^ QJ in
r-- s- TO
>, 1 -P U
-P i— t TO CZ >>
•i- p^- -i-i —
r- cn -P ^i
TO rH C •— •<-
^-) QJ i— U>
s- o TJ TO in
o -P ••- e o
E > (/I Q.
cn QJ
•O CO M- T3
QJ i in o c
Ul CO -P fTJ
TO LO (J m
QJ cn cu c «
S- rH 4- 0 E
CJ M- •<- ^s.
C 4- QJ -P O)
•r- O TO ZL
CN U
»+- ui O -r~ O •
O t- CO *O O «
aj c uo e
^ 4J ~a •<- v -x.
C C OJ Ol
0 "- -^ T3 Ul ZL
.,- 5 S- £Z r—
*J TO TO OJ O
TO C E > O
U O l
-£
-C
CO
1
1
0
o
LD
V
QJ 1
i — O 1
Q. m TO C "O
C 'r- in i- O QJ
0 -P TO -r- -r- CJ
•o i— Q.-P c:
C ZJ -P U) TO ZJ
Unpublished reanalysis of same 14 year Lo
mortality data using spectral transform m
regression analyses confirming significan
ciations for total, cardiovascular and re
tory mortality, accounting for autocorrel
and temperature. Suggestion of more prono
effects with 7-21 day cycles of exposure.
CD
CO
» en
^
ZJ
u
"
3E_
•o
C
TO
NI
-P
£_
TO
5
.C
u
CO
o
0
ID
A
i .
C -PM
u) O Ul S
Q -P 3 oi<—
ZJ TO O ZL OJ
C T- r— >
•i- U O QJ
-P O S O r—
C Ul O i — 1
O Ul S- V CO
t_j to <4_ -^v CQ
•Q O
C -P CO
O TO TO LO
T3 .C ( V
C -P TO Ol >,
Unpublished reanalysis of same 14 year Lo
mortality data using regression analyses
detrended data for time series autocorrel
tion, humidity, and temperature indicatin
significant associations between mortalit
BS to below 100 ug/m3, but not for S02 at
ug/m3.
CM
CO
en
r-H
TO
J_>
QJ
>>
QJ
CJ
0
a
o
^
o
CO
CM
o
CM
0
CM
CM
]
-— * >• .a aj - c 3
£-§ U_ U Q.CO TO in
ro cy in £
i c -P a; \ o c:
CM ••- CO S- 01 CO —
Evidence for reversible (^2-3 wk) small (
but statistically significant decrements
of school children following episodes in
ville, Ohio when 24-h TSP and S02 levels
tively ranged up to 220-420 and 280-460 u
but not after "sham" episode with TSP = 1
S02 = 190 ug/ma. Larger decrements seen
set of children.
>>
•i—
•o
£_
O
s:
Co"
CO
cn
~~*
_1
TO
j_)
C "
QJ
TO
O
0
CM
1
O
0
CM
0
CM
CD
0
CM
1
si
CO -r-
Evidence for reversible (^-2-3 wk) small (
but statistically significant decrements
O "O
4- C.
TO
TO O to
C 4-»
CZ
Q_ -P TO
monary function. measures (FVC, FEV^ Q MEF
school children in the Netherlands during
after pollution episode when 24-h TSP, RS
S02 levels ranged up to 200-250 ug/m3, bu
effect shortly after day when same pollut
averaged 100-150 ug/m3.
O
OJ
U
•a
- i/>
— 0
QJ -P
> U
QJ TO
i — S—
CO S-
CQ OJ
JZ,
3 P
0 0
"~ 0
-p -p
TO
OJ
^. TD
U1
S- (—
>> TO
• .- +J
^- C
TO QJ
-P -P
S- 0
o a.
E
-o c
OJ 0
TO -P
QJ ZJ
s- .a
(J -r-
c s-
**" "^
-C 0
-P LJ
*3 -t->
-a o
-P O)
TO C
S 2
TO CZ
rt' ui
cn-p
^.f"*
ZJ
C (J
4-
i/l •(—
OJ
> T3
QJ C
•— TO
Q_ CO
impossible to quantify with any precision
:ric calibrations for BS readings beyond 19
ui QJ
•P >
•r- TO
£_
-P O)
TO
r~ »4_
-P 0
-P U
O TO
•Z. r—
«
3-51
-------
experience: (1) Markedly increased mortality occurred, mainly among the
elderly and chronically ill, in association with BS and S00 concentrations
3
above 1000 ug/m , especially during epis'odes when such pollutant elevations
occurred for several consecutive days; (2) During such episodes coincident
high humidity or fog was also likely important, possibly by providing condi-
tions leading to formation of H2S04 or other acidic aerosols; (3) Increased
risk of mortality is associated with exposure to BS and S00 levels in the
3
range of 500 to 1000 ug/m , for S09 most clearly at concentrations in excess
3
of ~700 ug/m ; and (4) Convincing evidence indicates that relatively small but
statistically significant increases in the risk of mortality exist at BS (but
O
not SOg) levels below 500 ug/m , with no indications of any specific threshold
level having been demonstrated at lower concentrations of BS (e.g., at <150
ug/m ). However, precise quantitative specification of the lower PM levels
associated with mortality is not possible, nor can one rule out potential
contributions of other possible confounding variables at these low PM levels.
Besides the above London mortality analyses, additional studies reviewed
in this chapter evaluated relationships between mortality and short-term
PM/SOX exposures in various other geographic locations. For example, newly
available reanalyses of New York City data by Ozkaynak and Spengler (1985)
raise possibilities that, with additional work, further insights may emerge
regarding mortality-air pollution relationships in a large U.S. urban area.
However, the interim results reported thus far do not now permit definitive
determination of their usefulness for defining exposure-effect relationships,
given the above-noted types of caveats and limitations. Similarly, it is pre-
sently difficult to accept the findings reported by Hazakis et al. (1986) of
mortality associated with relatively low levels of S0« pollution in Athens,
given questions stated above regarding representativeness of the monitoring
data and the statistical soundness of using deviations of mortality from an
earlier baseline relatively distant in time. Lastly, newly reported analyses
of mortality-air pollution relationships in Pittsburgh (Allegheny County, PA)
reported by Mazumdar and Sussman (1983) utilized inadequate exposure characteri-
zation and the results contain sufficient internal inconsistencies, so that the
analyses are not useful for delineating mortality relationships with either
S02 or PM.
As for newly-reported analyses of short-term PM/SO exposure-morbidity
}\
relationships discussed in this chapter, the Dockery et al. (1982) study noted
3-52
-------
in Table 3 provides the best-substantiated and most readily interpretable re-
sults. Those results, point toward decrements in lung function occurring in
association with acute, short-term increases in PM and SO,, air pollution. The
small, reversible decrements appear to persist for up to 2-3 wks after episodic
exposures to these pollutants across a wide range, with no clear delineation of
threshold yet being evident. In some study periods effects may have been due
3
to 24-hr TSP and S0? levels ranging up to 422 and 455 ng/m , respectively.
Notably larger decrements in lung function were discernable for a subset of
children (responders) than for others. The precise medical significance of the
observed decrements per s>e or any consequent long-term sequalae remain to be
determined. The nature and magnitude of lung function decrements found by
Dockery et al. (1982) are also consistent with: (1) the recently reported
findings of the Dassen et a]. (1986) study noted in Table 3 for Dutch children;
(2) observations of Stebbings and Fogleman (1979) of gradual recovery in lung
function of children during seven days following a high PM episode in Pitts-
3
burgh, PA (max 1-hr TSP estimated at 700 ug/m ); (3) and the report of Saric
et al. (1981) of 5 percent average declines in FEV-, „ being associated with
high S02 days (89-235 M9/m3).
Table 4 summarizes those newly available epidemiology studies which
appear to provide the most useful quantitative evidence for morbidity effects
associated with long-term (generally annual-average) exposure to PM and/or
S0?. Note that, as was the case for the earlier criteria review (U.S. EPA,
1982a), none of the newly available analyses of relationships between mortality
and chronic PM and/or SO exposures were judged here to yield sufficiently quan-
/\
titative information to be useful for derivation of criteria.
From among the numerous new studies published on morbidity effects associ-
ated with long-term exposures to PM or S0x, only the few listed in Table 4 are
judged here to provide potentially useful results by which to derive,quantita-
tive conclusions concerning exposure-effect relationships for the subject pol-
lutants. The Ware et al. (1986) study provides evidence of respiratory symptoms
in children being associated with particulate matter exposures in contemporary
U.S. cities without evident threshold across a range of TSP levels for ^30 to
O
150 ug/m . The increase in symptoms appears to occur without concomitant decre-
ments in lung function among the same children. The medical significance of the
observed increases in symptoms unaccompanied by decrements in lung function re-
remains to be fully evaluated but is of likely health concern. Caution is
3-53
-------
co
U-l
a
=5
t—
co
o
CO
0 CL
UJ O
O V.
>— Q
a. z
UJ «C
UJ M
—1 O
CO CO
0
> CO
a:
—1 CO
3 O
LU CL,
UJ
Z
o s
a uj
UJ t—
CO 1
a:
•-" I—
>- -J
1— UJ
P
z z
=5 s:
o-=>
n:
UJ CD
Ni^ ^?^
U.P
0 <
—I
>. UJ
^
^p
*g
^^
CO
,
UJ
CD
^
p;
v^
O)
Zl>
Ul
r»
0)
0)
4-J
c
ID
•P
3
^
O
CL
0)
01
ID
t-
01
^£
1
m
c
c
•a:
01
u
0)
i.
01
01
Ol
o
CO
a.
CO
1—
TJ
01
•r—
ID
JD
O
Ul
•^
in
0)
o:
>n
T3
3
P
CO
f-
o
0)
a.
^^
1—
CO ^^
' — * CTi ID
CD rM CO
CO s— ' CA
CT> H
Ci 1-^ '
ID
i — P ID
ID 01
•P CO)
0) ID
E 0)
Q) a. 01
S- ID T3
ID -CO
3 o a
i in M
1 i-H O
1-1 r-H
O
in
M 1 1
1 1 1
O
CD
£ Ul
PS- 01
U O --O 1 £-
>>a)M-oo toJ-oi IE- .r— P
OlIOJi — 01 *+-LOZ-P Ol lOOlUl'i— 3~\C\IS-. — si — C
C'r-oia)3s:uj!-i a. c oa.ca)-c r— 01 a) co o ••-
0)OCID4->-PQ- 1 O) O O. X 10 -r- i*-r— 3.T33EQ.
Ul -C EOlS-tDtl) O-O EIXOJS--PP OO 0)O *^, O
•i-4->m>r-(UEjaO-CO>— sX ID r— 0) -r-C O-rO T3 I— Ol 3 P
•r- C ^^ ID X S- 0) ^-* E '"^ 3POJE3-P O P r— t 0) O i — 0)
JZ'r-UliniOOC'i— *^(N ^Q-CUlS^^EP CU1 UP^J- 3
U>-^0) Q)i— S-i*- O+J Ul O1O OUCO)OOlE-r- 0)0) MX iH-0-0
c o)'.- ID a. •!-•!- ID ZLCO a-iDJCr— O.OE r-i: a)c c
o oi c -P 3 a. -a -PUO) ui •— • 01 us- IDOICMOII>
s-iniD'i— craa)iD s-ox: a) .c ••— ro co 0) >T-OCUI i —
.2 10 JZ U C N i— r— CDIOP PE-P-C COS-P 0)^ICOO) •(— OJ-C -Q
01 U IDt->)0)ID l-i- ID Ol'i- Ul 1 3 C £- P S- O Ol-r-
-Ul . Of— S-3SO3 S-0)3C3i-IO'i- Q.EO)>4-IOCO'r-Ul
JZ'i— CCOJZ ID *OO-^- i — T-uli—i4— OO1OQ.-CU1
OlTJO'-PO)CC-r- Jc T3JZT3 S- O "OS-ID Ei— O-
3 •*— 3 •!— 01 ID •(— > p ^^» o a)Q.a)cod)cu}p 0)4— s-uioiDuio-ui
U S- U CO IDCPT3fl)P-P fb"O!D 'x^ Od) IDC>ID COUlfD
ll-+->3Ca) JZC-i-OlO)'!- S-IDU i-P O U T3 J- S- CL'i- ID U U Q.
o n> in 2 -i-3u u o T-COID UTJi— • — 'iDO)
S- IOU1 -PC-COO CJZU1LOC >, Cr-IO^:--> 4-M
U>-i-OlC'i-Ot03-i- r— Ul -i-OlUli— 13-C+Jr— -i--i-3 £U!Pt-O
•P in 3 £- o U ^^ -P o ul ID >i O ^11 E *i— 0) T- >\ (J c CL 3 0)
ioo)r— •oiniaa)u!'i-t-3 i— u ui o s ••- in r— ID-P o-r- x:
£- £- r— Ul 3--P 3 Ul-P P-P(NUTOU1 P O1-— • C i— 1 P "O • Ol
<*— *i— ID Ul U S-S-C CPi — O uiroo. CC n3P fDul"'-
T3S-OJT r— O 0) Ul ID 0) ID IDC3COJZCJ-Q.U1 10 O 10 -P C£-C^=
0)0) u>>0)inu u a) 73 10301 ^ uso)ps-ooi3
U130) i — >rHt-Ci— •/- -i-PIOOJ+JO U1IO -i-lOS-'l-OT- O-p
fDOUi — -PO) O)0)U /**s4- . ^-U) O1Z3PQ--PO; 4— IDE -PO.PC
0)i— COCi— O -i- c-3'r-.C •!-•!- Ol ID COUO. -r-JZ S-C03CO ID
£- 0) O 10 -P >,T3 -PECO1 CU1CS-0)0)1— Ol COlCO).Ei^|— CP
CJ T3 U) i"~ U ^ +J ID Ul ^v. Ol 3 Ol S- 3 Q) S- S- 4— 04 O) 3 O P '*"*. f~~ O P
CCJ3C)'i-COO'i-S-O Ol-f- O •r-0)O>33Olf-O •i-O'i-CO)OO'>-'i-
>-iioiain>iDuiuiZLUco coup-i-a.Q.2:-PE
ID
c
o
'f O) CO
•P C >> >, .
u ••- -a -a ^
0) O 3 3
CO Ol+J i— PC
1 C CO ID CO S-
Ul O C 0)
Ul 1— 0 r- P
o M- ra -i- ID ui
s- o c P co)
C_> -r- U -r- 3
Ul T3 0) "O JZ
r— 0) 3 CO 3 P
ID U) P 1 P 3
•i- >>••- ui >> T- o in
•Pi — CD U) "O O1CO C
•i- ID C O 3 C 3
CCO S-+J OCO
>-<
-------
warranted, however, in using these findings for risk assessment purposes in view
of the lack of significant associations for the same variables when assessed
from data within individual cities included in the Ware et al.
(1986) study. The findings derived from the series of studies by Ostro (Ostro,
1983; Hausman et al . , 1984; Ostro, 1987), also discussed in the present chapter,
are qualitatively indicative of morbidity effects in adults being associated
with PM exposures over time within U.S. cities and tend to support the plausi-
bility that the associations observed by Ware et al , (1986) reflect actual mor-
bidity effects in children due to contemporaneous U.S. PM and SO^ exposures.
The other new American studies listed in table 4 provide evidence for: (1)
increased respiratory symptoms among young adults in association with annual -
3
average SO levels of vL15 ug/m (Chapman et al . , 1985); and (2) increased pre-
valence of cough in children (but not lung function changes) being associated
with intermittent exposures to mean peak 3-hr S0~ levels of ~1.0 ppm or ahnual-
3
average levels of ^103 (jg/m (Dodge et al . , 1985). It is difficult to deter-
mine if effects observed in these two studies are due to repeated high-level
short-term S0? peak exposures or to more chronic exposure to lower annual -aver-
age levels of S0? or its transformation products.
Results from one European study (PAARC, 1982a,b) also tend to suggest that
increased lower respiratory disease symptoms and decrements in lung function in
adults (both male and female) may be associated with annual average S0? levels
3
increasing across a range from about 25 to 130 pg/m . In addition that study
suggests that upper respiratory disease and lung function decrements in children
may also be associated with annual -average SOp levels across the above range.
However, the S02-morbidity effects associations reported by PAARC (1982a,b)
cannot be fully accepted in view of (1) internal inconsistencies between
findings obtained with S0? exposure estimates based on one type of measurement
method versus those based on another measurement technique, and (2) the lack
of adequate control for potentially important confounding factors in certain
of the analyses yielding significant associations.
3-55
-------
-------
CHAPTER 4. CONTROLLED HUMAN EXPOSURE STUDIES OF SULFUR DIOXIDE
HEALTH EFFECTS
Since the completion of the 1982 EPA criteria document (U.S. EPA, 1982a)
and the first addendum to it (U.S. EPA, 1982c), numerous scientific articles
have been published in the peer-reviewed literature or accepted for publication
in regard to controlled human exposure studies providing important additional
information pertinent to development of criteria for primary (health related)
NAAQS for S0?. This chapter of the present addendum summarizes and evaluates
the newly available studies and discusses their relationship with certain other
key studies and conclusions from Chapter 13 of the 1982 criteria document and
the earlier addendum. Several of the key issues discussed in the previous
addendum have been further investigated. Those discussed here are:
(1) Differences in subject characteristics, medication, and restriction
from medication which may have considerable impact upon the differ-
ences in results reported by different laboratories.
(2) Concentration (S0?)-response relationships in sensitive individuals
under various conditions of exercise activity level or other form of
hyperpnea.
(3) Possible enhancement of SO^-induced bronchoconstriction by cold
and/or dry air and by mouthpiece breathing.
Mechanisms of action of
(asthmatic) individuals.
(4) Mechanisms of action of S02-induced bronchoconstriction in sensitive
The majority of subjects used in the studies summarized in this addendum
were asthmatic. Asthma is a heterogeneous disease classification which in-
cludes a broad range of subjects. The least severe asthmatic may have had
asthma diagnosed by a physician during childhood (by an unknown set of
criteria) and have been mainly symptom-free since childhood and rarely, if
ever, requires medication. On the other end of the spectrum are individuals who
4-1
-------
may be on chronic bronchodilator therapy (theophylline), who may use chromolyn
(disodium chromoglycate) prior to activity, and may also require steroids.
Pulmonary function tests (spirometry and airway resistance) are used to define
the clinical status of an asthmatic at the time the studies are performed.
Since airway obstruction in asthma is variable and often intermittent, and
given that the physiologic status is highly influenced by the quantity and type
of medication being used, tests of lung function cannot be used alone to
determine the severity of the disease at any one time.
In addition to the diversity of clinical status, there was a broad range
of selection criteria used to define asthma in various laboratories and from
study to study. In some of the early studies, a clinical definition of asthma
(i.e., diagnosed by a physician) was the selection criterion. In an effort to
provide more descriptive information about the subjects, other criteria such as
a positive response (i.e, much more reactive than "normal" subjects) to a
pharmacologic stimulus such as methacholine or histamine was used as a criteri-
on for selection. A positive (bronchoconstriction) response to an exercise
test (5 to 10 min at 85 percent of maximum) or to an SCL inhalation challenge
was also used to select subjects. The use of these descriptive criteria is
sometimes useful in comparing results among laboratories.
One further point which relates to severity of asthma is the ability of
the subjects to safely withhold their medication for a particular period of
time. There was considerable variation among laboratories in the duration of
time for which certain types or general classes of medication were restricted.
A number of the characteristics of the subjects who participated in studies
described in this addendum are summarized in Tables 5 and 6 along with other
information on aspects of protocols employed in the studies.
The criteria for adverse health effects of air pollutants have been a mat-
ter of considerable discussion and disparity of opinion. In general terms, the
adversity or "clinical significance" of a response may be discussed in relation
to the magnitude of the functional changes (this must be considered on a test-
specific basis), the duration or persistence of the response (i.e., acute re-
sponses vs. permanent or long-term health effects) the types of symptoms and the
degree of discomfort or distress involved, and also the need for possible the-
rapeutic intervention.
The relationship of symptoms and measured physiological responses to the
health status of asthmatics may not always be readily apparent. Figure 7 may
4-2
-------
1
"3- O O
O O —i
3 :j U
x -^ C -3
CO y?
i -^ *j o
> ••— i/l •~-
J -— .Q +->
: — o
o -u
4-> QJ
ro .c:
CO U-
-
CXL u_
UJ —J
r- ID
t- -i-
O 4-*
ro U
O
>-
cn 4J cn 4-J cn
i— ro
ro -C
S- CZ
O -r-
O O
e s-
O OI
r— O i—
ro s-«f-
^: ca -a
O i —
S- -r-
CQ TJ
•(— ro
dCO
OJ cn
4-3
-------
UJ "O
—I 01
—J 3
O C
a; "c: o o
i—, ra -•»
— O U
VS) 3
t- >» t. =
ty in
U 13
XJ (J
fO CJ
o tn
4J
S- U LA.
-------
_
OJ C O S-
ro OJ fO •
.*-> C -C i
in -.- +->
i- O U
ZJ -
>> s- ^
4-J +-> O
> -r- l/l •!-
OJ •— ^2 •»->
••- o
CO O
•a •*-
-a i
QJ JZ
i— Q-
CO CSJ >> CO CO
E -a .c
O "- -f~
s- o -»->
-
O •!->
•<—) 03
(T5 U
03 i—
a; a; s- cz
U .1-^ > X
^- Q. i— <1J
4-5
-------
a
UJ
i
o
fy^ -^^
H- 0
Z t— <
O r—
0 0
Z
UJ 13
oa
-
I— i 0
- 3D
_j a.
3:
uj z
z o
s: uj
CD a
as i— i
u_ x
o
CO H- 1
O C3
t-H
f— rv*
CO =3
HH Li-
as _i
LLJ =D
H- co
o
' '
UJ u_
r-j UJ
CQ
™~^ 1 i
.—J i-l.
CO O
—1 CO
cn s-
S- 0
CU 4->
i — in
5 :5z
C_3
eu >
c. u_
•r— *^^.
i — O
eu •
in rH
CO ^>
CO UJ
u_
CU
c:
•r— ^
I— CO
CU QS
in co
cO
oa
u_
"y* ^^,
s:
CU
D5
^^
S-
o
f*"
.f_3
3
«^
1
1
|
1
1
ID
en
•
o
i
0
f-.
.
CD
rH
^J-
•
f**.
II
1 X
O
^x^
CO
CM
00
CM
1
en
rH
S~*t
«^f-
00
en
rH
^.^
,
i^—
CO
+J
CU
•r-
T3
CU
CQ
1
1
1
1
1
rH
CO
II
1 X
CO
to
•
p^.
II
1 X
CD
^^
CM
CM
OO
CM
1
en
rH
s*~*.
in
00
cn
rH
s^x
•
r—
CO
.}_)
CU
CU
CU
o
in
c:
•r~
^~
O
u_
I— 1
UJ 1
s: i
•^^* "^^
CO
oo
•X
CO
II 00
1 X CD
CM
LD
•
IT) X
I--
II r^-
1 ^1 ^3"
CD CD
rH \
^^ rH
CD CO
rH CM
/••^
CD
CO 1
1 OO
rH rH
CM •— '
'•""N /^^
«3- co
oo oo
en cn
rH rH
^^ ^^-^
. .
r— t—
CO co
4-^ 4-^
a> cu
OJ >,
i — O
r- CO
3 4J
^*Z CO
1
1
1
1
1
1
1
1
1
1
LO
^^
LT>
rH
,
UD
O^J
/^-^
^f"
oo
cn
rH
v^ ^
•
t—
CO
^_>
CU
S-
a»
r~
(J
CO
U
CO
I—
UJ
s:
~^_
^"
00
II
1 X
1
1
1
CD
"^^
CD
rH
CO
p**.
i
LTJ
,, — s
U3
00
cn
rH
**^S
.
r~"
n3
•4-^
at
•r—
r—
^~
CU
c:
•r™
-a
c
o
cea
,
CO
-4_)
CO
-a
cu
r~~
+3
4-
O
•4_?
CD
tn
-Q
CO
X
4-6
-------
INCAPACI-
TATING
SEVERE
MODERATE
O
2
iii
Z
O
z
b. U
o w
Z
bJ o
9 °-
< OT
tf. Ul
0 a:
INCREASES
»20O%*
INCREASES
MORE THAN
2OO%*
INCREASE
UP TO
2OO%*
INCREASE
LESS THAN
1OO%
Ul
O
O Z
Z <
O
bJ >
O <
< OT
5s
EMERGENCY
TREATMENT
REQUIRED
W
K 2
0^0
IRONCHODILAT
REQUIRED TC
SOLVE SYMPT
ut bJ
K
M
SPONTANEOUS
RECOVERY
<1 HR
PONTANEOUS
RECOVERY
<30 MIN
to
^
Z
z
0 H
2 o
b b. bJ
^ O u.
fH it
D bJ
o
DECREASE
»15%*
DECREASE
>15%*
DECREASE
BUT <15%*
NO
CHANGE
bJ
O
O Z
Z <
O
z K °
~ ^. ^
bJ Ul
02-
z o °
< K ^
I £ &J
0 ft b.
SEVERE
BREATHING
DISTRESS*
V)
* OT
bJ t_ Ul
BVIOUS WHEE:
MARKED CHES
TIGHTNESS
EATHING DISTI
O K
ffl
• _•
SOME WHEEZE
OR CHEST
TIGHTNESS
(0
LD SYMPTOM
NO WHEEZE
OR CHEST
TIGHTNESS
2
K tO
O 2
H O
0 < K
z £ a.
a. 2
w >-
U tf)
a
V)
2
o
a.
2
>•
N
O
9)
O
»
C
0
a
to
2
O
<
O
1
b.
Z
O
5
O
c
O
6
2
3
0)
H
4-7
-------
be used to roughly classify the severity of response using four variables which
are frequently measured in the studies discussed in this chapter, namely: (a)
change in SRaw; (b) duration of effect of SO,.,; (c) changes in spirometry, chiefly
FEV, Q; (d) types of symptoms and relative discomfort. This table is not
intended to provide a quantitative description of what does or does not
constitute an adverse health effect but is primarily intended to demonstrate
that there are an array of responses and to assist the reader in judging the
relative severity of the different responses which are described. There is no
question that the types of response described under INCAPACITATING would be
considered as clinically significant adverse health effects. Most of the
responses identified in this chapter would fall in the moderate and severe
categories or some combination of the responses described under those
categories.
4.1. NORMAL SUBJECTS EXPOSED TO SULFUR DIOXIDE
The pulmonary function effects of S0? in normal healthy adult volunteers
have usually been much less than those seen in S02-exposed subjects with
clinically documented asthma. The newly available information supports this
conclusion in general but also suggests that some mild effects which are of
little if any acute health importance may be observed in normal subjects at
concentrations below 5.0 ppm. The 1982 criteria document (U.S. EPA, 1982a)
presented the conclusion that the probable lowest-observable-effects level in
normal healthy subjects is 5.0 ppm S02 at rest. The first addendum to that
document (U.S. EPA, 1982c) further suggested that normal subjects are about one
order of magnitude (i.e., tenfold) less sensitive to SO,, exposure than asth-
£.
matics.
Bedi et al (1984) studied subjects exposed to 1.0 and 2.0 ppm S02 in an
environmental chamber (22°C, 40 percent RH) for 2h (V£ = 40 L/min for 3 to 30
min exercise periods with intervening 10 min rest). In the initial 9 subjects
tested at both 1.0 ppm and 2.0 ppm S02, these investigators reported a modest
(10.3 percent) but significant increase in SRaw following both exposure
concentrations. Further investigation with a total of 22 subjects at 1.0 ppm
using the same protocol failed to substantiate this finding. Given the trivial
increase in SRaw (well within daily variations), the finding in the initial
group probably occurred by chance. Folinsbee et al. (1985) also reported
exposure of normal subjects to 1.0 ppm S02 in a study in which the effects of
4-8
-------
combined exposure to ozone and S02 were examined. The exposure protocol for
this study was the same as the Bedi et al . (1984) study and included many of
the same subjects. There were no significant changes in forced expiratory
spirometry or airway resistance as a result of 1.0 ppm SO^ exposure reported
for these subjects.
Stacy et al . (1983) exposed subjects to 0.75 ppm S0? alone and in combina-
tion with several particulate pollutants. During the 4-h exposures, subjects
walked on a treadmill on two occasions (VV approximately 55 L/min). There were
no significant effects of this S02 (or SOp plus particulate) exposure on either
forced expiratory spirometry or airway resistance.
Schachter et al . (1984) compared the responses of asthmatics and normals
(4M, 6F) to S0?. Three of the normals were reportedly atopic (i.e., they
probably had some history of allergy). There were no significant effects in
normal subjects at any of the concentrations tested (0.25, 0.50, 0.75, and 1.0
ppm S0?). Measurements were made for 60 min following a 10-min bicycle exercise
period (VV estimated at 35 L/min by measurement at the same workload on another
occasion) in S0?; the S02 level was maintained for the first 30 min post-
exercise. At the higher SO,, concentrations (0.75 and 1.0 ppm) the subjects did
experience upper respiratory symptoms (these included unpleasant taste and odor
and sore throat, symptoms associated with extrathoracic airways).
Koenig and Pierson (1985) in a review of several studies from their
laboratory reported a decline (6 percent) in FEV^^ Q following exposure to 1.0
ppm S00 in 8 healthy normal adolescents. These subjects were exposed via
3
mouthpiece to either 1 ppm S02, 1 mg/m NaCl aerosol, or their combination.
Resting exposure of 30 min was followed by 10 min of exercise (V^ = 39.9
L/min). The apparent decrease in
^ Q
occurred 2 to 3 min following the
exercise period in S02. However, the FEV-j^ Q decrease following saline aerosol
was 4 percent and the absolute post-exposure FEV-j^ Q values were identical
(i.e., 2.89 liters). Furthermore, the authors used repeated pair t-tests in
their analysis without correction for multiple comparisons (e.g., Bonferroni).
These data should be subjected to a more rigorous statistical analysis to
ascertain their significance. Even if these FEV-j^ Q data were statistically
significant, the differences between the air exposure and S02 exposure are so
small that they are of no clinical importance.
3
Exposure to a mixture of S0? (1 ppm) and ammonium sulfate (528 pg/m ) was
studied in 20 normal subjects by Kulle and associates (1984). The subjects
4-9
-------
were young adult nonsmokers (10M, 10F) with normal spirometry and no allergic
or respiratory disease history. Four hour exposures occurred in an environmen-
tal chamber (22°C, 60 percent RH) and included two 15-min exercise periods
(miId-100 watts, VV estimated 40 L/min [4 to 5 times rest]). There were no
significant effects on spirometry or airway resistance after exposure to either
SOg alone, ammonium sulfate alone, or their combination. There was no change
in the response to a methacholine inhalation challenge following any of the
exposures. There were reports of upper respiratory symptoms which were most
prevalent with the combination exposure. This study further supports the
absence of pulmonary function effects of S02 at 1.0 ppm in normal subjects.
Wolff et al. (1984) exposed nine steel workers, two of whom were classi-
fied as asthmatic, to 5 ppm S02 or S02 plus carbon dust for 2.5 h in an
environmental chamber (22°C, 50 percent RH). The exposure included five 4-min
exercise periods (V£ not reported). Mucociliary clearance measurement
exhibited no consistent pattern of change. Histamine reactivity (percent drop
in FEV-, Q at threshold dose) showed a tendency to increase slightly (37
percent; 28 percent excluding asthmatics). There were no notable changes in
pulmonary function among the normal subjects. Symptomatically the subjects
found the SO,, plus carbon dust exposure more unpleasant than S0? alone.
The effects of S0? on ten older men (55 to 73) were studied by Rondinelli
and colleagues (1986). Subjects were exposed via mouthpiece at rest (10
2/min) and exercise (10 min at 31 £/min) first to NaCl droplet aerosol, and
then to either NaCl aerosol plus 0.5 ppm SO™ (n=7), or NaCl aerosol plus 1.0
ppm SOp. FEV, n decreased after exercise in all conditions by 5, 7, and 8
percent respectively. Although these results are suggestive of a small effect
of oral breathing of S0? in older men, the incompletely randomized exposure
sequence and the inappropriate use of repeated paired t-tests in the analysis
raise sufficient questions that the effects cannot be considered conclusive.
In summary, these studies of S0? exposure in normal healthy adults and
adolescents demonstrate minimal, if any, significant pulmonary function effects
of S02 exposure at 0.25 to 2.0 ppm with exposure durations ranging from 10
minutes to four hours including exercise periods, with work outputs sufficient
to increase ventilation to 35 to 55 L/min. The only effect of any consequence
was the increase in upper respiratory symptoms, which was chiefly the result of
the unpleasant taste/odor of sulfur dioxide.
4-10
-------
4.2 CHRONIC OBSTRUCTIVE PULMONARY DISEASE PATIENTS EXPOSED TO S02
In addition to asthmatics, patients with chronic obstructive pulmonary
disease (COPD) have also been exposed to SO™. Linn et al. (1985b) exposed 24
COPD patients (ages 49 to 68) to 0.4 and 0.8 ppm S02. Although there was a
wide range of functional impairment (FEV, Q/FVC ratio ranged from 27 to 70
percent), all patients were able to exercise without supplemental oxygen.
One-hour exposures in an environmental chamber (22.5°C, 86 percent RH) included
two 15-min exercise periods (VV = 18 L/min). In contrast to many previous
studies of mild asthmatics, most of these patients regularly used broncho-
dilators and were permitted their use up to 4 h prior to study.. There were no
effects of SO™ exposure in this subject group and no trends indicative of
change in any of the measured functions (including SRaw, spirometry, and
arterial oxygen saturation). It should be noted that little if any effect
would be anticipated in asthmatics under these exposure conditions. The
authors suggested that these COPD patients may be less reactive to SO™ than
younger asthmatics, although, as the authors discuss, given the low dose rate
of exposure and the marked differences in medication status, this conclusion
may be premature. The ventilations achievable by COPD patients are limited by
the severity of their disease. Further investigations of COPD patients exposed
to SO™ should include groups with less severe disease who are capable of
exercising at moderate intensity (e.g., VE = 30 to 35 £/min) and able to
withhold medication. Only after such investigations have been completed will
sufficient information be available to assess the relative risk of COPD
patients exposed to SO™.
4.3 FACTORS AFFECTING THE PULMONARY RESPONSE TO SO™ EXPOSURE IN ASTHMATICS
4.3.1 Dose-Response Relationships
Important considerations in assessing the response to any inhaled gas or
aerosol include the concentration of the substance in the inspired air, the
rate of exchange of ambient air with the lung (ventilation), and the duration
of exposure. The concentrations to which asthmatics have been exposed in more
recent studies (since 1981) range from 0.10 to 2.0 ppm SO™ although interest
has focused on the range from 0.2 to 1.0 ppm. A broad range of exposure
durations has been utilized ranging from 3 min to 6 h, although the primary
4-11
-------
focus has been on 5 to 10-min exposures which incorporate hyperpnea.
Ventilation rates have ranged from 8 to 10 L/min at rest to 60 to 70 L/min
(exercise or voluntary eucapnic hyperpnea), although most interest has centered
on moderate (VE = 35 to 50 L/min) to heavy (V£ > 50 L/min) exercise levels
which in warm humid environments provoke, at most, only mild to moderate
exercise-induced bronchoconstriction. Results from the recently published
studies are summarized in Table 7.
Schachter et al . (1984) performed a concentration-response study in a
group of 10 normal subjects (see Section 4.1 above) and a group of 10 asthmatic
subjects exposed in an environmental chamber (23°C, 70 percent RH) to 0, 0.25,
0.50, and 1.0 ppm S0?. Subjects rested briefly and then exercised for 10
minutes at 450 kpm (V£ = 35 L/min). In addition, subjects were exposed to 1.0
ppm S02 at rest. A significant decline in FEV-j^ Q followed both the 0.75 (-8.3
percent) and 1.0 (-13 percent) ppm exercise exposures in these asthmatics.
This was accompanied by a significant increase (54 to 68 percent at 1.0 ppm) in
airway resistance (interrupter method). There were also some changes (these
did not occur consistently at all concentrations or time intervals after
exposure) in maximum expiratory flow which mainly occurred at the two highest
concentrations. The recovery was rapid and pulmonary function was within 5
percent of baseline (and no longer significantly different) by 10 min
post-exercise even though S02 exposure continued at rest. As other investi-
gators have reported, there was a considerable range of response among these
subjects, with 3 or 4 subjects demonstrating no appreciable response to S02 at
any concentration while some others showed trends indicative of a dose-response
(SOg-FEV-j^ Q) relationship beginning as early as 0.25 ppm. The responses of
asthmatics seen in this study may appear less severe than those seen by other
investigators at similar SO,, concentrations, although comparisons are difficult
because of the different measurements made; the relatively small changes in Raw
may be partially due to the use of the interrupter method. However, a number
of other factors could account for the discrepancies between this and other
recent studies of asthmatics. First, the subjects were not pre-selected for
the presence of airway hyperreactivity to S02, cold air, exercise, histamine or
methacholine, an approach frequently used by others. Second, the moderate
workload and unencumbered oronasal ventilation probably resulted in a lower
SO,, delivery to the reactive airways than would occur with mouth breathing.
4-12
-------
C LU
LU CZ)
2: LU
ZD 00
31 O
a_
CD >C
LU LU
—I DZ
O 1—
O CZ1
C_J LU
CD »-H
Q> c_J
LU O
OO
r~ - 21
a: cz)
2: i—
ZD 2:
OO ZD
Lu
LU
CO
1 —
in
OJ
u
cr
OJ
t-
o>
4-
OJ
in
C
cu
o
t_J
in
c
o
+J
ro
t.
O)
in
.0
CD
cu
S_ in
ZI Z3
in 4->
o ro
CL40
X OO
LU
O)
!_
ZI CU
in "D
O O
X
LU
4— -X
o in
4J
S- U
CU O)
E JD
ZI ZJ
2: oo
o
4J
ro
s-
ZJ
CD
CZ
0
CJ
^
CO
en
r-H
ro
eu
•a
S-
CL
CL
CU
JZ
00
. cu TD • ro
•a in cu s- oj
CU fd "O •«— >—
i- i- eu c
o u c >> zi
Q. O) S_
cu -a cz -a eu
u o u
. — *- c: c
o cz ro u
C -r- i— CQ -r-
cn-.- i— i 4-
ro s- -u :n •.-
4-i ro c c
ro E cu cu cn
T3 > U -i-
in S- ZJ m
E 4-J OJ "d
O in a. O JZ
4-> O) >, S- 4J
CL cn JZ CL*—
>, zj c: o o)
S-
cu cz -a
in ro
ro cu jz cz
CU CZ 4-> •(—
!_ 'I—
LJ i — in CQ
c cu in i — i
•i- in cu rc
O JD O)
~o c i^» ro
CU -i- -~~s t-
•o 4-> ro
eu m cz a.
c s- E o
\ u
CU S- r—
ro > co o
!_ O - 4-
oo
c &•$ -a
0 CD >, 01
•r- CO JD T3
4J eu
ro >)"eJ cu
r— JD QJ C
cu ry jz Jz -i—
> 1/1 m 4-* ro
o ••—
•r- E
4-> LO
C
eu II
S- LU
CL
:c 4J
o>
OJ
CL r-H
1 JZ DZ
ro zi o
S- 0 CSJ 5^ DZ
O E CSI CD «3T
CO
oo
CL
a.
i— i
CD
CO
cn /— %
^*-N .•— s I — 1 ID ^~"»
JD, ro ^-- co ID
co ^f- en co
CO CC - r-H CT1
en en • <— ^-* r-n
r-H i — 1 ro v-x
cu ro •—
i — i — ro
ro ro t- 4->
CU CU 4->
(ri 4J 4_) QJ
OJ CU JZ i —
u cu s-
c cz ro jz cu
c cz jz 4-J cn
•i— T— LJ CU O
— J — 1 OO CQ Q±
S- ID i—
CZ O CSJ O)
5^ JZ O O CU - >
4— ez> 4-* i — inzzms-i — eu
oo 4-JLom. i — i — -r— ctno rojzr—
•^1 CZ4->CU OJfO rOS_O4-14— L)4->
Cn cz in OJ "*-~ LJ > cu E ro • CL in •>— •<— QJ
i — t- oj oj o m ro 4-* E c cu cn^— o •-—
OQJOU uuocuu -r-c i — oocxcno •— cu
•r- ,— U C C S-'i- .in-^- OJU.— ZJJZ. CDOJS-
injDinro ros-**— 4-Jcro > rooomc-i— QJCU
>)•!— U JZOCU-r— UCU4-3 OJC .OJOininX
JZinE-r- C4-S-C: OJOII- •— -r- CU "O S_-r- >>CU
CXinO4— OJOCn 4— OJ inOJ-JZ4->JZin
O +J *i— CU U «i— 4— CO in 4-J •(— 4-J OJ 4-^ O CU QJ m
OJCLCLC ajinintn 'oj-r— c *r— cus-i — ••— cTi*»-
r— E O1 i— CZ ZJ. JZQJS-OJDOt-OJCJZ
JD jz >,-•- JDOEC oc oj 4->4->eucLro4-)4-j . — ro4-j
ro4-iinm roCLO-r— CCU4— o inxcu--— in JDJZ
CUT- ojm4->ro >oc 4->-i~O)S-s-o)c roo4-i
s- 3 c c s- cu o.4-> incuro rox romos-ro
Z3 ••—•(— Z3S-ES- OJ inO CU>»tn>OO Z3in
inin ro in >>QJ 4->in4->-f- tni>E' — o mEcM
ro O) oj 4-> ro croo>>oJz rooo
cucnms- cuo cz ocno)*i— oooJ4->> — o O)4->oo
Ecrocu EOO zi •r-c-'-ic ajcjz CLJZ LO c E CL
roojo • TDrojDcn 4- E en csi o S E
OJZS-C: O4-DZ4- CTJZZI-i- 4-S_O >,-r- . S_ O >> CL
2ioozt zocco i—totnin LUO4-JinjzoJD 2:mcL
t OJ 4- ^
•r- s- - ro c ?i
4-> o c c cu cnr^.
s_roc4- - coj o s_-f- -om
-OJ4-JT- 3-r-> - -r- o ^J rOCU *i— 3 4-> 4— C 4— f^- O
03O OJ O CrT in . OJ Cn ro— » ino T---NC4- ceO4->
or >>mc oo ••— i — in a^s-ss-OJ &^ro-i— oo 4— • CL
t^jDOJ4-J cs->oju cs—'Oicn CUOOE s-^ COJTD>>
C l-r- S- ZJ •*- O) O) S- -r- -r- r— ZJ C Ul r-H ^^ • -i- 'r- O O in
••— minoja X' — O4-> < — oin-i— -r— cnro c-f—
oinc incucin mi — UlfS--r— incut-c
o> xo -r— eu cu DZ ••— *i— euroo"cu s- •<— me cu'ocu-'—
cn nm_ s_ cns-cc:4-> cnEz:min cuczra ro cncuo.
cE4->o c aj r— ro cm ECZ xro oo> ccLin
ro> cu ro > s- .— 4-> ro oo cuojcci— roE4-Jeu
jz -rom jzoo>roin jzjz.4->o_ Ero o jz-^ccn
o -i — u • UJZE U4-JOQ.U) c * — * aj u cu c:
Qi O •<— E E 3*-1 inO •— i/; E O) -i— . — CTZ* CUEro
4J Lu !- 4— O O 4-1 (1*— C 4-J 3 X>> £- E-MU-i— CZ 4-1OOJJZ
CLU4->-.-4-)4-> COQJ4- c rom ~\CD Ero ccs_u
rocucccLCL rooo 04-* ro-E~o. EOOC-^ rorozi
u oenEE u i- zi o OD> ccum o_ roEw u-Minj-J
•r--u •«—>>>-> -/--ocjzi •,— ^ -^-tn4-> -^CJZCLCD — in ro c
4— c_5 ininin 4— c_>*4— o 4— LUC rou ••— 4-) ^d oo 4— -i— euro
•^ > eu -r— > *^ in -i— LU •>— cu cu cu o — in s u
tZ U- V) >>r— CU C Lu r— 4-> E CSI in I- '•— ) LO^S-CDCZ COJ •—
cn -i— i — -i — c en omo cn o>roujD r^-roeucDcu cnt-4— 4—
•i— -u, — roo -r— -S-CU4-) 'r— -cncuczj or 4-> CD oj **— o •<—
in— S-ros- in — -•-> cn CL in — c S- -i— in jzooror-n^ in- c
>a)ucu>-) r>cczni= 3>roo -u aj4-> >>,cji
o LU x ••— > cz OLUOZI>^ o LU Jz tz 4— CD •«— cz S- -i-J QJ oejC'r-
2ILuCU4JOro 2TLuUOOin 21 Lu U -r— O i— 1 3'r— Cn-:£ JD "ZL \ — rO in
i/l
c -a
•r- T3 O
E O O) ^->. ••—
C C C ~^. LO CD 4-J C S- CZ
E E E *— 'lEEcroa-E
C-— C^- Ci— CIT)-OC043E4-JCU<—
•1— •!— •!— •!— OJ S~^ U) CL 4-> OJ 4J
inco incD in m . in tl 4J c || cu -^ •-— in ZJ Csi
s- 11 s- II s- 11 t- > ••— i D> -T CD eu s- E II
OJ OJ OJ OJ4-)EECD4-»OJ
X LU X LU X LU X - — ' in CL^->\ t— I C X CD u_
LUD> LU> LU> LU T— 1 CU -^ CSJ i — ^ — ' — ' OJ <— 1 >
0
O
LO DZ HZ «3- ' r-H
s-r^ • s- ..s-^ t-DZ- s_r^.
QJDZ'-H QJ&?O05^LO CU HZ i— 1 O) \\ QZ f** CUDZ^H
jDQi J3OLOJ3O: JD JDCt:
EC_> || E LO II CO II Et-J II EO '&< II E 0 11
ro o 3-§ ro cj ro o 5^ ro o cu UD ro o 5^
JZOOLOIC JZO'-^DZ^^DZ JZOOCDDZ JZOO.OODZ JZIDCDDZ
C_3CSJOD
tE Q. o. ex
CL CL CL CU
CU Q.
LO LO tD
CSJ CSJ CSJ CSJ CSJ
CD CD CD CD CD
4-13
-------
—I CO
o o
ce a.
LU co
CC CO
u.
O LO
C£ UJ
o ro
Q.4J
X CO
o to
4_>
t. u
JD ••-»
i-i
t- in o •*-)
o ro s- •— .
o. o o. ro *-
S- U O 1C !
ji — en 4-) u1*
cu-cc-o ••-••-
CU 4-1
cu s- o •*-
_ u c: o
3 TD • C ••- CL
: ,— > c co
T3 s- ro ro c
tn O) °O U LJ
Q-cn" ~~ c: c
E cn ai ca en oiro
>> rj x >-H ••- ••- •
co co OJ Z m co CJ •
O J=
I
. TD O
. "° u
i -^- ro
•r- ro i
T3
o o
o s- u 4-> a) u
•a c E ro OIM-
31^- -a -c ai ro
- uj c: c > c
•> 11 -r- Qj >> QJ
*-> 4j c: -f— CD
*~ TD J= 3:
c ^S QJ 4-J CQ roc
O O 4-> —« Qi •—
•r- CO ^- C 3: CO
J3 C i-H
CQJQ.-N
t_) CO OJ O>
j - — CU CU CU 4-> ••—
-oct! ° ° c-^,t!
co QJ ••-> aj t
ro c •— r--
C o
••- - — E a.--- cu
i- cu ••—
a; ce
So
4.E
— m
t^ !i
m CM •
CM CO —1
II II II
*— 4J cj ce it
ro — s
i. o CM £i? az
CD E CM O
-------
CX
E: oj
1 Zt
ZD tZ
U_ C
O O
CJ
CO * — -
LU
O 0
ZD CO
CO O
LU
or co
ZD CJ
CO I — 1
o i—
LU 3Z
1 —
"Z. CO
O LU
LU CX
— 1 ZD
— 1 CO
O O
ZZ. LU
O
CJ 1C
0 2C
t-lJ
CO | —
ZD CJ
CO CD
LU CO
Q£ CO
LL.
O CO
1 —
>• CJ
cx cu
CX ra
X 4-3
LU CO
OJ
s_
tn a)
o TD
CX 0
x s:
LU
M—
O in
S- CJ
QJ QJ
J3 --5
E -Q
ZJ ZJ
•z. co
c:
o
03
s-
ZJ
cn
u
o
CJ
03
CO
cn
i— i
03
QJ
C
— '
tn
QJ
cn
tz
03
CJ
03
ce:
CO
i~
b
03
in
O
E
O
s_
ce
CO
c:
QJ
m
03
QJ
s~
(J
c
OJ
S-
cx
CL
i—
(J 4-3
•'- cx
C ZJ
•i- O
E S-
Ol
U1
OJ S-
4-3 0
OJ 4—
• i—1 C
•o o
>— 1 4->
f>- QJ
(/) 1 —
QJ (13
Ol E
03
x: jz
CJ 4_>
c:
03 CD
O ~=t
4! LL,
•r- UJ
£Z ^
cn
O UJ
E
in LO
•r— CO
CJ
S- II
QJ
X U
LU >
s_
E CJ
03 0
Jz; ro
CJ CM
CD
i — 1
tz
o •—
00
t— { "$•
E
CX
cx
CD
QJ
s_
QJ
in
CJ
O
;>
. LU
. ai u_
\Z O "T3
X3-J QJ
ftJCX 01
> >> QJ
in S_
c u
r~ C tZ
4-J
cr QJ *»~
a» m o
E 03
QJ QJ in
5_ S- C
CJ U O
QJ £Z "-
^-N S- OJ
3-? m cn
CO 0) CD
II U CO
J
II
1^
CJ
QJ
ZJ
ro
s-
o
cr
QJ
c:
o
QJ
£_
03
oo
CO
cn
QJ
, —
QJ
4-3
QJ
CQ
c:
o
CM'i— "O
O 4J fZ
CO CJ 03
-— S- OJ
•i— in O3
4J C S-
c: o
03 U QJ
4-1 O m
in jz •-—
-O CJ CJ
zj c: s-
in o QJ
S- X
in jo QJ
aj
+-> -a jz:
CJ CJ •«-
•o ~a
£Z (Z ^-3
>-l •!- fO
cn
c: u
03
JZ !Z
+j ra
OJ
- E
CD '«-
CO "^ S-
C (73
cn c:
3= "- 03
03 in QJ
CO /•"*» CJ
c: co c
•r~ CO -f—
CM
QJ QJ
in || in
03 03
QJ CZ QJ
s- ra s-
0 Q) U
£Z E tZ
c:
e
c /— -a
••- OJ
I/) CD 4-3
•r— IJD fO
CJ E
S- U "~
QJ 4-J
LU ^> QJ
QJ 1C
JH CX
ECJ
03 o 3-S
JZ OO r— 1
CJ CM ^J-
O
i— 1
cz
E
LO
E
CX
CX
UT)
O
fO
4->
03 l*-
T3 0
E 4-J
O C
4-> QJ
CX+J
E X
>^ QJ
in
.4-3
O ZI
. -a
1C QJ
t-
- O
•a cx
O QJ
tn
CJ
OJ'
JZl QJ
QJ
C S-
•t- CJ
tz
•*- o
JD t—
03 1
•r- J_>
s- jz:
03 cn
> •!-
cu
fl3 T3
-^- QJ
c: o
03 JZ
4-3 in
JZt OJ
zi c:
CO O
(Z
E
E
cx
O
LT)
CO
**
<+—
0
QJ
>
—
cn
zi
in
in
OJ
CD
03
CJ
r
CO
at
rz
03
CJ
!tl
cz
ai
03
CJ
sc:
o-
(_)
_
JD
oo
CO
cn
03
QJ
, —
QJ
+J
QJ
CO
a>
i 4-3 i jz; QJ -4-> Ji
1 JZ rO -C -i-J S- O3 i->
£Z 1 -*J j-J O O CJ
O£Z 03C01 ES-QJ
"o >•>•>— JT3 n -I-1 jz d. aj
Cj TD ' — - • 01 "— 03 tu ' .M
"O QJ Q)TD O3 O4->-r- i_~C3cO
•i— C to O O S- •— 03 O
03 i. 4— 4-3 jz: a» •*->
COQJS- > S- '<— -C JUC ZI in
O CZ v^* OJ CJ 4J CJ O • —
•r- -^- O3 J-J £Z JZ ••— ro O OJ
•o s- QJ o o cn OT-t-> ji s_ aj
i — i t/i cj CL O •«— JZ E tZ 4-3 Ol
QJ C
?g in cz o
CD S- 03 OJ -r- iz
"~ i. u — E
ift O !5c o ••—
O4~ -t-J — ' oi
. !- O
O •— m' 3= -n
Z! ••— - CM - >~> - O
O — "t3 OJ •*- >- Q; CM X
t/lJZ Xi— tfCCQJm- —
E O -— -r— — *J 0 CD _ JZ
O3QJ-»J-r~CD CZO3i- X "-s Z1
tz c: cz
E E E
tZ CM <3" *-O
> :>
QJ
UJ i— • CM CO
QJ
-^ CJ
in QJ
O3 •«-
E O_^~i
ZC U z: -H
o; 03 c
CJ U, S 11
CD t-O ^-*^-v3I
CO CM i— t CM -* .fOQJ (J •-• *
CJ ^ CO CJ ~~
4-15
-------
G3 "O
y aj
±3 3
Z3 C
CL. *r-
u_ cz
CD O
CJ
UJ
So
r—
CO O
r—
LU
OS CO
13 CJ
i/) t— i
O V~*
uj rn
s^"
ZZJ U-
zco
a uj
oo
ceo_
3= UJ
13
U- Q
UJ
ZDCJ
V) O
QC V)
UU
O CO
H-
>- CJ
§u_
UJ
ID
co ST.
0
• r—
UJ=3
—ILL.
CO
ff
U)
cu
CJ
cu
OJ
a>
Ul
+j
c
cu
E
E
O
CJ
U)
O
4-»
m
>
£.
CU
Ul
J3
O
CU
3 Ul
Ul 3
O 4->
x"-*J
UJ t/)
CO
fc.
3
Ul CO
o "a
s-s
LLJ
<+-
O u)
t. CJ
cu cu
E S
3 3
O
4-1
£_
3
C3
U
c
o
*— X
ro
CO
rH
*-^
ra
cu
ai
c
0)
o
u
QJ T—
CJ 4->
CU tl
I I
a
e c
J= 0
ui DIM—
cu o
ra +->
CJ CJ
•I— CU
C 4—
UJ
x
o t- >•
>• ra -
UJ O
u_ c ui
•r- X
ra
S~ I Lf)
3 ..^ . *-cO
Ul &^ 1
o *sF-*
a. jn«st -
X r-H 0
cu i TJ oo
cu t
cu - ui
u TJ ra "a
cu cu cu cu
•*- u» I- u>
d ro U ro
JZ ¥ "
CM in ic
CM r*-
CU
• . in
cz cz ••-
•r- 4J -r- CJ
E ui E t-
cu a)
O !- CZ> X
ro I-H cu
^
a.
a.
0
c:
ro
CJ
"cr
cn
Ul
S
Facemask:
Ul
ra
cu
CJ
ra
LU
changes.
LU
+J
U)
cu
t-
X
tn
00
1 — •*
ra
4J
cu
cn
c
cu
o
C= -r-
CU
CU U
Ul C
3 ra
ra -t->
U Ul
>!*«/)
ro CU
o ra
ra
Ul C
cu
4-> "a
ro CU
U Ul
•i— rQ
C S-
t— t U
CU LU
U U_
C
ra
Ul C
•r- ra
Ul CJ
*"iii
i — CZ
fO CT
Ul -r—
fO Ul
tr o
•i— £Z.
CU 4->
ul ZJ
ra J3
CU
!-
CJ &5
c CM
OJ
•r-
E to
CU
O S-
ro
E
CL
Ct.
O
^
3 TD 1
in c o
CU ro -C
O) C
I'll
j= ro CU
U CU I-
••- s- o
*=. .a E
- ro • —
Ul S- 4->
u o cz
•i- CU
ro S- CT
E O CU
ui c: o
ro -i- (J
*f CU
1 Ul
ra
O CU
t/J £-
Xp
B-
^-aT
CM 3^
t to
ID
01 i
Ul
CU t"-
S- X
CJ ra
CU E
to
rH
II
<:
cu
. Ul
C •(-
•r- CJ
E £-
CU
O X
CM CU
C
O
CJ
tl
4-J
Ul
O
CJ
"O
cu
Ul
ro
CU
i— S-
ro CJ
ui CU
ro TJ
C
C ~tQ
•t- Z> CM
LU 1
CU U_
Ul tfi
ra - f-
cu ^e x
S-, C3 nj
cj ro E
c ;>
•i— <>-
o -
4-> O
C CU ui
Significa
resistanc
-16% Vmax
in
ra
CU
CJ
ra
in
GO
cn
i— i
^d
cu
t_
cu
O)
0
Qi
I -N ra
cu o -t-> cj
s- LO ui cr ui -r-
| 4-i -r- CU CU <— r—
CU fO 'r-3 Ul U
CU -M O >i-d ra 4->
t- 1 S- 4-J 3 CU ra £-
ro 4-> CQ •!— ui t- rO
S_ > CJ CZ OJ
Ul O -i— H- CZ -i— i —
4J-CZ . 4J O -r- U
CJ U) CU •<— Ul
CU Ul Ul £3 &§ E 4—
M- t--r- CLTJC3 O O
4— CU U CU C\J CZ> 4J
CU 4J S- Ul r-H Cu4->
H- CU 4^ E CJ
4-roXH-3fO>,CU
O CU O O tO ••-?
•o j=i cn .a
4_> CU -O CU ra CZ 3
cz m cu cu ••— -in
cu ro 4J t- -cz •& s
x s- cu cu"-jz:c£ cz
UJ O Q-~O 5 */> LO O
Ul
5 1 4-J
rO •«- «.)
- Oi C u_ — j ra
-.— s- ra -r- u_ jzi -i—
in -i— .cz to 3 4-J
. JZT 4J - in cz
5^ ^5 ui z > o 4->
ro (**> "O ui LU c\j tn
cn ^j- cz cu u_ JD
•*—'•**-' rO i-~ » t- 3
-a - o to
« s- -o . o o **-
o -r- c. cn-i- >•
to ro O "i— s- LL. in .
> i-
ce: 4J t- cn in cj
t/i cz •*- cu 4J cr c
ra ra a. ui ro cz ••—
-a _cz t- -cz ra
CU 4J CU OJ "- O CU -
in ui u) 4— E cn
ra S- ra ••» *J "-
CUCUCUUS-CCLU)
S- Ol S- S- CU fO ZJ
(J£-CJCU4JCJO4-'
craczxM— •!— s-o
^^^_.^. cu raM-(L3 C
to
O CZ
cz ro £- E
CU >— ••
i-H
t- r*-'
CU ZC rH
.a 01
EO II
-CZ to O ZC
c_j CM r-
- cu cz:
ui ul O
L) CD 4->
cui-.ro
•r-3 U U
m cu
•r-'o cn
M- C
•I- i— 1 •.—
.— 3
XJ C CU
C o "a
ro J= cz
> in ra
Ul
E
o
4->
CL
1
"a
pronounce
cu
CD
ro
CL
cn
cz
'•*
o
o
M—
o
cu
3
cz
o
CJ
4-16
-------
ce:
O "O
E: QJ
— J ZJ
ZD C
Q_ •<—
4J
U. C
0 0
u
CO •— '
1 — I CM
czs o
ZD tO
1—
to O
1 —
1 1 1
ce: to
ZD CJ
tO (-H
O 1 —
CL. <^
LU 31
ZD Lu
1C O
CONTROLLED
H EXPOSURE
i—
0 S
LU CU
LU
to >—
r> o
tO CD
UJ tO
u_
o to
1 —
>- CJ
CZ. UJ
<£ U_
g" U_
E: LU
ZD
to *£
0
• h-
[*••• CJ
2:
CQ
X -P
LU to
CJ
in QJ
O T3
p o
X-E:
LU
O in
S- LJ
QJ QJ
JZt 'l~>
§•§
2T to
C
O
.p
£
0
L)
C
O
CJ
CO
cn
i-H
TO
QJ
S-
ro
Q.
CX
QJ
_c
tn
4_> O O
in O •<—
QJ QJ in CO S-
S- LJ -P A CJ
ZJ C CJ *•— ' CL
tn TO QJ
O t- 4- -O S-
CL QJ 4- O QJ
OJ O S- C
4-> QJ CJ O
"O > O i—
OJ OJ ».—
4J U -P 4-> S-
TO Z) U S- O
QJ T3 -r- O 4-
CL C S- .C
Qj .r- 4-> in -P
S- in O
d C TO C
tn TO O
QJ LJ U I- -P
TO CM-C > JZI
U O L) O
•.- CO C '-x
T3 O CM C
C O S- O 'f~
\ — 1 4-3 .O tO S
QJ 4->
OJ in -P tn
in c s- - o s_
s-*. C O QJ JS C -i-
^5 O Q-r— 4-
O tn QJ TO CM S1? £~
i— t QJ S- E cn QJ
it QJ - - ro i— 1 4—
X ZJ C 3^ r— QJ ||
^-s in o LD •<— to QJ
O (J ro 4-J TOf X m
O- QJ tn CJ ^— ' TO
3X m H S- CJ
TOQJ QJ LJ -^ !-
C£ -t X S- C QJ LJ
tO • -P ^— ' ZJ •«— QJ C
C tn tn "y. •(—
C -r- QJ S- O S
•i- E S- CJ CL t9-i E CJ
^- X CU O S-
tn C ro C 4— in
TO -P ••— E "O TO O
QJ tn E in S- . • CL
s_ s_ ^;£'7^£: £
C 4— ro ZJ-POfO-'-
s_ s- s- c
. OJ QJ • QJ II II • *«-
-,— 4- 4- .,- -4-| XI X •«-
to TO «=C m TO — " — ' tn ro
c jz:
0 LJ
•i- TO
4J CJ
TO
• (— O
-P 4-
oj -a -P
> QJ L)
t_ -i- OJ
QJ S- *<— 3
CL TO ^3
>,> ZJ
1C — •* tn
QJ
LJ CTi
QJ
Q. ZC ' — t
f— 4-> CJ \\
TO ZJ O &•?
s- o ro CM in
CD E CM CO tn .^
E Q. E U -P ro 5
CO S--P in TO-i— TOCMi-H
E
C3.
O-
LO
o
CO
cn
I— 1
ro
QJ
r—
QJ
CJ
CQ
cn i c
C O -r- . QJ
•i— i/> m in
. — tn C LJ QJ
o ro o -i- -c
O -i— 4-> -P
L) CM-P TO
0 U E (-
TO*""0 'si 5 £
5 CJ -P in
I- tn in TO E
•^ ro c: tn
ro QJ o cn-.-
s_ (j c cn
-P O O -i- S-
ra c: jz: 4J QJ .
jz: ••— LJ TO c: m
-P C •— >> C
c o •<- (n a
t/1 TO £- -P ••-
CJ (J jZi c tn +•>
+-> QJ -P ro
TO cn~o > in c
LJ C QJ S- QJ -«-
T3 >, ro CL CD E
C t- -r- >, ZJ O
»-H -a u -c: i>o LJ
in
-P 4—
11 (JO
E QJ
S- 1 X T3 •<-> tn C
s- ro '— ' c ja m o
O!5t- TOZJQJ'i— 4—
•r- 5 in c -P o
y~ TO TO CM -P TO
ftt- Q£ O C S- U S-
ce: >,tn to QJ o -i- o
to -c -a -P
C t^l -r- O 10 E U
• r- -O t- -^
in O O in o O ^^ O C
Q) 4- LJ TO •!— "O 4— -r-
in QJ -P X C
TO .5_£-ro-i— ro'O'O
QJCnOLJCtO QJO S
S- ••- . C -r- CJl-i^ O TO
LJ tn S- -i— iQ ctncnoi
c: .,_ E • -^ ro to
.,- 4-> TO . O i- M -P
O CD LJ *i— QJ O O "O
c QJ to t- jz; -p s-
Oitn-i— o>>5 inZJ
•r-E4- 4-s_ --Em
OJ -P "O QJ ' — ^ S- 4-> +J QJ
-O CL'^ C &« -Q O TO CL E
EEOCM".— Q-OJE
>> >, n i— CM o QJ s- >,4-
c
p
-P "O ' — *
TO r- C
•(- .C U -P E
4-J in ro LJ "v..
£Z CJ QJ QJ i — = -
QJ S- ••->
> -C t- JD CD
S~ [— O ZJ ID
OJ = 4— in 1
Q. O
>, O UJ CO
1C -P > ^~'
CO
a) in ^
•i- CJ 0 ~>^ It V
, l-^S^^^rc
r- 4-J C\J !*-* 1 = i —
s_ > c to •»- ro
I -i- -i- O 1 -P
cro-p--— t-ro m4-j
O •-- -C -i- E QJ TO
(J "O "O tn TO -C CT
JZ r— -O TO -P CD QJ
CD i — TO in C
o c QJ o ro
S_4-TO>LJ«— mLJ
.Q O *i— TO CJ *^-
C -P 4- -P 014-
cj tn *i— •(— o o c -i—
jz; *J ~o 4-3 ro c
4-> LJ Q) T3 C ' JZ: O)
QJ CZ TO O C U •>-
TO 4- JD. C -P 5 >n
jz; QJ E ro m c n— QJ
-P "O JZ: QJ O D£ TO S-
CJ LJ -P CT'i— ,UO LJ ZJ
tn > i cn-P -i— -P
4-3 *.— cj in ^ (j c TO
tn -P o tn tn ro • ••— ^
cn-r- .—
roc t^ co QJ CM tz QJ ro
ce: ro i i— i in ro t-
tO QJ TO -P O S-
•— -P T3 QJ- 4J
S- C CL TO SO CL C
U TO E ro &-S E TO
c j= cu c as c co >iJ=
c
••—
E
C i—
tn CD
•i— in
L)
i- II
CJ
X LU
LU >
LO CM •
CM CO* i-H •
II tl \\
S- CJ CJ
cjzco cjo zczcin
E 1 f^* CvJ
.C CD *^NX~N^"^I — t CM CO
CJ CO t-H CM ro ^- 'N— •••—'
CM
C
'E
LO
E
CL
CD
4-17
-------
Of
s: cu
—1 3
a. -r-
u_ c
a o
u
to **-*
UJ
a o
ID to
to o
LU
as to
to »—•
O P—
:< E:
LU zc
P—
2: to
§<:
LU
3= O
CU UJ
LU as
-J tO
oo
z: LU
8=
__l—
as
u, cu
LU
£25
ZZJ CO
to o
LU CO
as to
LU
CD tO
>-<0
as LU
S u-
IE tu
tO 2=
CD
. P—
r-co
2=
UJ =3
CO
h^
in
cu
u
c=
cu
cu
cu
as
in
cz
cu
E
o
c
o
ra
CO
m
o
CO
3 in
in 3
O 4->
CL ro
X 4->
LU I/)
S
in cu
O T3
CL O
x s:
•LU
o in
(- u
cu a>
If "3
z to
cz
o
4-1
ra
s-
3
czt
u
cz
o
CO
ra
m
00
3
ra
4-J
(U
cz
c
cz
o
»*- 1 4-> .
o s- o cz
OJ 3 O
0) UT3 —
o ra cz 4->
CZ X T- U
ra cu •.-
•+-J CU £-
t- cz j= 4J
o ro 4-J tn
CL C
E in c o
.,- OJ*f- (J
o
cu en t- jr
JZ CZ O U
4-> ••- 4-> CZ
>i U O
tn s- 03 t-
CO T3 >*- -Q
4-> 1
ro >> en » ra *i—
t- CM t- ra
CD T3 CU
t-l/l r— >,
••- J= T- I-
ra jz 4-» jz "o
c ••- s cz
$3 CM CO 4-
CZCM -o ••-
+ CU E CO
tn in 3 tn "O
Q) o ra jz ra •*—
O14-J CU CU E
CZ t- J= t- 3
ra ^5 u 4-J u jz
U I •»- 5 •«- CZ
a; u as in as en
to **— to f-H to ro
*!—
cu
in CD
-I- in
u
S- II
X LU
LU >
rH CO
CO
o -
co r- ro
fc. CO -
cu re re co co en co
jzj -as as o o
ECO 11 rH || CO
ra o &3 &3 CM co
-CZ-HCD CD 1C 1C
CM
CM
E
LO
g
CL
CL
tD
CD
(J
•eg-
CO
en
^
ra
cu
cz
cz
1
1 C
cz ro
O JZ
S- JD
O in
~O tn
CJ CJ
- 3
CZ •«-•
1 CZ
C4 O
t/> 4-J
U
t- •(—
o s-
m
•a cz
o o
•r— ••— (J
•o &- o •
CZ CO JZ JZ
>— i CL u m
''r- >,
»
s- i z: ro
*r- - CO "O
ra CD 4-J
CM ra • t-
CZ rH i — CM O
ra
co - - >> in
i — UD rH ra CO
O ro TZ» E
•— 1 >» I T-
s- ra cu 4->
o TD -a 4->
4_ cu i ra cz
m >>i — cu
cz ra r— co
o cu s- - s
•r- s- ra CM 4->
4J U CU CU
U CZ >>JZ1
cu •<— co ra
s- jz -a cu
s- 3i-> i U
o ra >> cz
u as c i — cu
to o s- s-
s_ ra cu
cu -&$ co **—
4-> CO CTJ «—
>
to cu x cu co JD
U CM tn
i— U TZJ
ra 3 in >, CO >>4J
o ra x ra ra
p— CM TD CU T3 t-
E
CL
Q.
*J3
CD
tJ
OO
en
^s
ro
cu
c
cz
4-J -a
CO 3 CZ -
> JD I ro in
•r- S- CZ E 4-
4-J T- " CU 4-> O O
u ro ui cz -»-j 4-J
•r- 4-J - cu CL cu ro
S- 4-J u tn -+-J E >
4-J in cu C in >>••— cu
in •(—«+- o *i— t/i 4^ u
cz O s— -i— in in c
o E co 4-J cz cu ro •
o ••- o • co u tn
O - CJTU CJ ' — • CD"— CU
JZ-DC3J C C4-J 34- S_
• Ui — to 0*1— cz m-(— 3
cz o o ra cz 4-J
oucu tnumcnro
s_ tn cu -r- T- co ••— s-
jD.4— rom 4— en tn cu
O CU CU 4-J -i- CZ CL
m s-jzcrczrO' — E
4-j tn cj 4-J aj ouz ro co
tn4-Jcz E'(—U(_)4-J
cu u -r- s_ cu tn •<—
CD CU CU U 33Z JZ
<-OCUS3JZCZOOUJD
!- £-
•r- C O
ro ra 4-J i-
CD CZ ro-) S- ra
to ra as •>— jz
CO tO t- CL U C
jz i— cu in 3 ra
4.) O "U 4-J CU E CO
T- cu ra s- •—
3 cz in cu tn o
ro ra s- s- co
5c co en cu £- c
ra-J s- 5 o •!-
as o • o o
to s- cz en— i tn cz
co ••- -i- ro
•a 4-> tn E -cz
cu ro • o 4-J
tn co co .— •S.--N4J
ra s-o &$&^ CL ;M
cu cncM CM r- E CD
S- CM CO CM >,LO
cz en4-> — ' cz
•i- -r-
m CM s-5 jz ro ro
CO CO 4-J QJ "O OJ
4-» rH CO -f- i— CZ i~
•
^r- ~n-
• to
ro rH
i
S- CO II ||
co re co o
jzt as o CM zc zc
E LO CM <£ , t-
CU JZ1 -r—
tn cu ra
cu in X3
JZ CZ CU i—
4-> O (J O !-
CL C O CU
cu cu JT, ro
"O S- CZ M— 3
cz cu o
3 CM 5 >>
. O 4-J >! O TTJ
-i-> t/) O 4-> i — ra
ro c ••- cu
jz - -o ro s-
4-J tn >,-*- »—
cz i — E in ro
tn o 4-J 3 ro
4-> -i— CZ JZ JZ 4-J
tn 4-J co cz
OJ ••— S- S- JZ CU
cn*o ra cu u 4-J
CD CZ CL 3 -r- C
3 O CL O JZ O
to u ra i — 3 u
oj c:
cz cu
"O O CO
C r— 3 .
ro ro • — 4-J co
ro cu t-
3CU U JD 3
rcxn -i— 4->
as •/- jz . ro
l/l CLM— i-
rO 4- CO
CZ 1- -r- Q-
QJ . +J
in tn en
ra co -f— m
cu ••— tn *••—
S- 4-> JZ,
U •!— O 4->
cz TD :z
*~ E ro
4-> 3
ro cz co
U JZ O ••-
•*- E 4-> -r- 4->
4- O O 4-> "-
•r- 4-J JD U *O
CZ CL •i--'-~
en e t- CL e
•»— >) O CU 3
to m u— -a jz
.^
C c— 3
•(— 4-J
tn CD m
— in
U 4-J
s- I! o
cu « —
X LU •>-
*=J- CO
re re co in
1 *X Q<
I- II II
E LO CO
-------
CL1!-
1/1 in en
UluiOlO
- ooo
(D QJ C
U — -i-
•i- CL-C
I- CJ QJ O CJ O3
M- -r- -t-J -CJ
QJ QJ CJ C -^ •—
J '— CJ to Ol Q.'—
3 E S- QJ O >-, CJ
Q O
Z3 CO
f—
CO O
1 -•- I O 03 CM fO
coczo: 4-J aj T3 r—i cj
"* 03 LL. H Q-S-Cs—'
QJ QJ -CJ
QJ •!—
S- < 4-J
-r- tn c u
OJ T3 S- I
in QJ T- "O E QJ ?J
-fO to CLQJ O'— O
•— • ••- u .a
I Q) QJ CXI I
••— t_ ro
i a. u
c: cj QJ
in O3 •«— >•
•^ -i— T3 tz; . c
4-) QJ O) i -•*
I U_ C OJ I- S- H-
S_ -^ 4J O QJ
p c • o 3 4- jz: s-
• en i~ i- o -*-1
; QJ O— QJ O3
^C(JE
6-$oiZJ
"-D nsoi
CSJOJ
S_ fD -i- - fO •
(_t QJ <4— *™i QJ C7>
crt--(-^?e-'-
i — i en c oo • _ ' co
QJ QJ O.
01 > (TJ CJ -r-
QJ .^, .^^ ^j 3 O '
t- T3 "O in T3 CT2:
, CJ CZ -r- 3 03 t-
^- en ui wi
E en - c 4J E .
o c c JT -^ u o
+j T— •(— cn M QJ 4-^
tZL"O 3 QJ •-—) CL
>, S CZ (J .C 3 >»
S-i/JXOE'-—
rOfO CD QJ i LO O QJ ft3 CD
cz c cri XQJ - E QJ
- o •—•• •— ft*-1 QJ • o cn
i- n3 E3 t- O".*J C •
o i/i . > o o •— Q. «J en
OJ "
.-
CJ '•—
n:ui> •— «
' OJ "«O 3 Ll_ O, E CJ '•— (& •+!>
>,s-co o LU x >, <= z; .01 ZJ
^2^2r— iEa_ajui-- UIE.Q
Q. 03
X 4-)
LU CO
C£ O3 OJ
01 -r-
^9 O3 CLi-O
t. cn '— 4-j co
t- CJ
QJ OJ
JZl "->
QJ cn
J=. CZ) X CU
CO i-H QJ JD
4-19
-------
C T- OJ 3
o m c oj -u
•r— OJ •*— ^ in
4-1 Cn<— 4->
u c u OJ m
'•— to .a -I—
c i— m cn t.
O ro 4-> CD O
O C (J JD <*-
C (J OJ C CL'O
QJ 01 •«- C C
4-) E? "O O)~O ro
U 4-> ro v> ro ex
•r- O. 3
•o E "a -— LO o
4J a> ui
CD,+J O
>^ aj u
t- 4-> U ro
OJ VI OJ -C
> C i~ (J
O C O 3
E E^ 2
•*-) .Q in o O
ro QJ t- C
J_) QJ -r— (_)
(J 3 LO
Wl 3 CT Ul S-
4-> -O QJ QJ QJ
U I QJ
•i- U >
t- C •«-
O (O 3 U
.c m •»—
u yi 4-
C E in '.-
O O OJ C
E ro ui
t/i >>^Z
a> co (j <—
•r- C
O OJ
C U O QJ QJ U C
•r- i— QJ > C QJ -r-
E QJ !- -i- O > i—
in 4J ;
•r- C t
tn )
— •. S-
$_ E- u ••-
CL CX*'— '•"•
ajoj-JOJOJA
i O S-
4->
£- U
o CM r-.
-------
a;
U_ CZ
CD O
LJ
00 •— '
LU
CD CD
ZD V)
\—
\ —
LU
err oo
ZD CJ
OO I — I
CD t—
CL. - CJ
C£ LU
s-
aj
VI
_Q
0
QJ
. vi zi
0 4-J
CD_ fO
X 4-J
LU OO
QJ
I-
ZJ
in QJ
O T3
CL O
. x s:
M—
o ui
I- LJ
QJ QJ
E JZ?
ZJ Z3
2: OO
c;
o
CD
•
c;
O
CJ
, — .
uo
00
cn
r-H
*"""'
,__!
03
4_3
QJ
j_
QJ
cn
or
O •' —
ja >>
QJ •—
LJ "O 4-J
CZ QJ V)
QJ LJ O
•a zj E
QJ T3
CL C C
E-- 0
>O U
t- 00 •—
0 t-
Respirat
suggests
choconst
QJ
OO I VI
QJ •!-
r— -a a
, — S-
ra QJ QJ
in X
S- c LU
OJ O
4-J CL4->
03 QJ t-
S- T-
a: ZJ
OO Q
QJ
CZ vi E
"~ O +J
QJ -r-
vi t- .cz
O3 QJ +J
U QJ
QJ vi in
"O "— QJ
LJ V>
. S- O3
•i— X '•-
OO QJ CJ
L
OJ ZIZ
E CJ
cj CM r*--
CM
O -i—
CD <-T)
r-H r-.
E
CL
CD,
CD
r-i
QJ QJ
E CJ
4-J OJ
S-
r— gj
4-J r—
•i- O
vi QJ E
>i in S—
03 C QJ
S- CL 1
i— V) 4-J
03 QJ S-
S- O
, — jz:
ro "O in
zn &-
VI -C
03 S-
CJ QJ
r- 4-J
.- ro
U C
QJ
4- >
O QJ
in S-
4-J ZJ
(J U
S- QJ QJ LJ
QJ l/l 4_) l|_ O
JZ 03 VI 4—
CL QJ QJ QJ QJ
r- s- cn LJ
s- LJ ai4-> cz
QJ QJ ZJ ZJ 03
CLQ in JD CJ
cn V ••-
Ci <~>^2s ^
TZJJOO
51? ol cz
CD CUr-
cn QJ
i—i oo | 5
n
ro
T3
sr ru
•^ rora vi
'—.a: .C 03
oo ^j- T3
^ QJ
.,- s- ro
03 U
ceL cz c
oo 03 i — i
JZ
TJ 4-J ..
a> QJ
VI S- VI
to QJ ••—
QJ 4-J LJ
S- nj S-
LJ CLP QJ
C i- X
t-ri cnuj
i-H
^
r-H
I)
4-> QJ
s_
S- U
QJ C
(O
Oj . .
s- a>
cn in
- CJ
cn s-
r- QJ
vi X
LU
i-H -£Z
, — ,
OO
cn
r-H
,— I
ra
J_)
OJ
,
s_
QJ
ic£
VI
CZ
O
CD,
VI
QJ
o
QJ
LJ
ra
QJ
o
in 4^
CZ 4-J
•o ra c
o -cz
i— CJ
S- OJ
QJ E
CL
m
CD
!_
VI
QJ
S-
tn
0
CL
X
QJ
E
£_
QJ
LO
QJ
VI
cz
o
CL
in t-
QJ O
S- 4-
OJ QJ "O
LJ VI O
ZJ •«- «^-
•a LJ
QJ S-
S- QJ
4-> X
1 4-> QJ
4-J
S-
03 O
CL-CZ
CL VI
03
QJ QJ
VI 4-J ^
U CJ 4-
S_ -f-
QJ TD QJ
X CZ r—
QJ r-l JD
CZ QJ
QJ -r- VI
t- in in o3
i- -t-~ to >t QJ
ro in QJ -M S-
>, S_ 4-J -r- CJ
cz •— u m i— o cz
i— ro CZ QJ •(— 4-J •<—
jz: E OITJ i- JD 03
4-J o -i— cz ro ro CJ
4-J vi rO > CZ . ••—
S- CL ZJ 4-J 4-
, OJ 4-J 4-J OJ-'i- C
QJ o QJ ro o CL **~
s_ cz jz: t- 4-J oo
QJ JO QJ C 1
cn 4-J o oo c: e QJ
r- CL U O VI
VI ZJ QJ in CLUO *i—
o "—t vi . in oo LJ
•^ S- JD QJ 4-J QJ S-
JD VI 4-> O CZ X
r-H S- <4~ 4-> O LU
r-H . CD O E LJ >>
*~~ff — ^ CM t~ O QJ QJ 4-J
^ in LJ ••— JJD vi
•M S- m JD !-
CZ> OO O C ••— 3 O •<—
00^-"4--r--OVlCn Ll_
CZ
QJ
4-J
s-
QJ
4-J
1— 4
s-
QJ
E
-a
o
LJ
QJ
OO
:M
r-H
It
X
i
03
o
4-J
c
1—
QJ
VI
o
QJ
X
LU
CJ
o
S3
vi O
QJ ZJ
0 C
"S ~° 4J
4-J ^J CZ
rO CD O
QJ OO LJ
CL
O) O O
S- 4-J 4-J
03
O
CZ
QJ
in
03
QJ
t-
U
CZ
*^'
BLO
rcise
QJ
X
LU
C
E
r-H
U
^
ZC
CZ
ra
o
cz
QJ
VI
QJ
£_
O
CZ
i.
QJ
CL
QJ
E
*J
QJ
comparab
QJ
jz: . m
4-J in o3
f- 4-> QJ
S CJ S-
QJ U
c ••-> c
4-J VI .
ZJ CD ••-
CZ r-H OO
QJ
j-> O • -
cn«=c QJ
•i— *^- VI
. C LJ
3-S QJ&? in
i-H CJ
S-
tl QJ
X
X t-U
Oi^C
i-O 1 —
J
i-H
r^;
i-H
u
ZC
LO T3 XI
O QJ LU
r-H £-
S- in
II ZJ ZJ
LJ O
X CJ ZJ
- — O CZ
C£. ••— O
OO 4-> CJ
CZ I-
• r- QJ
E 4->
CD,
CD
i— 1
oo
cn
03
4-J
OJ
C
03
4-J
in
S-
o
ZC
t QJ O3 VI O3
-a cz vi cz •— 4-> E
vi QJ ra C O fO LJ
QJQJ+JZJO-'-LJaJVl
VI S- S- O* CL 4-> -i— M— 4->
>^Z3O VI LJ CZ 4— LJ
i — mCU OJC-r-QJOJ
rO O QJ . S- ZJ i — ••— 3 E
EZ CL t- •*-» "4— LJ vi JD O.
03 x LD c; 4-J zi CL
QJ4-JQO-I- S- •— in V)
O3 4-J QJ i— 1 >>4— -i— cn QJ CM
cz m E •*— ' 4-J 4-J cn E
•i— T- !- - r— LJ QJ OO V)
4-> H— QJ i — -r- •— 4-J t-
T3EX o3roa,QJ-'-QJ
-aOQJ4-J'«-E LJ CZ
roi- QJ s- jz: y— c c
4— CZ ro4-JorOS-O
4-J 03 QJ ro i/l ••— CJ
LJ4->-CJCnvi QJ4-C-O
QJ 03 o o QJ cn en'*— o cz
. — -a -r- cz. •<— cz cz cz cj QJ
4- S- 4- O 03 Cn 4-J
QJM— QJ CZ — EX:-'— M— X
a: o CL-<- 4-J ro u in o QJ
in
OJ
4-> I/I V)
ZJ CL S- QJ 3:
c '<- o - m o
cz -a CD cn s- o a.
O OQJOOCQJ4-JVCLQJ
CO -r^ 4-J -^-i- X O >
4-J fO CJ LJ QJ t — s~*LT) ro
t- (O i — ex zj CL •rjr-* jC- ,
QJ i— O T3 S- CD •
4-J QJCLQJOQJQJOOOin
U— t- t- CZ S- > > ^^ 4-J
03 QJ "- Q- O -i— CJ V) CJ
Q) 4-J E 4-J Q_ 03 QJ
&^ cz -r- QJ o ns— jz: JD E
OO -O 4-)-r-QiZJ4-J^-NZICL
oo CL vi QJ 4-J oo E '•— ^ m Q.
CM inoj-oro ZJ3=O
QJS-, S-CZO OO M— UO
\\ S-ZJO4-J-1- vi^-xocn
in -P C 4-> cj
n ' O CL4-J CJ m !- QJ C3 /\
•i— X LJ C O3 ••— •<—> CZ CM
"3: 4-JQJQJOQJrajDO3 *«—
rO 03 ••—> CJ S- ZJ •!- 4-> O
yi 4-JZJZJWCZO3 QJO - — -
CZOV1CD-'— QJ4-EJD ^J
i — ..QJM— oo i — ^O n3O
03 CJ -C 5-S LJ 00
4-> CZS-CJQJCD 5^ET3 -*-s
O OOrOJ^CDCLOCLCZCJ)
4-J C_JM— OJ4-Jr-H"-CMCLrDQ_
j.
E
QJ
in CM
LJ
S- II
QJ
X LU
LU >
i — 1
S- r-i
QJ ZC i-H
E CJ II
rO O &^
.a UD o zc
cjwr^i
^-N 0 X CD -r- 03
CM CJ QJ r-H T3 TZI
O
4-J
E
U") O_
CM CL
CD CM
4-21
-------
CU "-
u. c:
o o
is?
t/1 O
—11/1
O O
K- X
2UJ
8=
W1 O
UJ vl
QS m
>.
o *-* cr c s-
(j U O O)
•*- to 3 >, CU
3 > or <^ u LO '
XJ J= JD XJ UJ -i
• ffl C (U !
S- Ul E Ul CT
*-> o -tJ -n u *-> o
fO -r— VI pr— QJ O 'i—
••- 1C U) O
1 >> O
•— ex
3 O Ul .
QJ OJ
"- JZ ••- O
, o cu :
U T3
O S- C *~ •>- O v
-o >> s- —
4J O TD '—
c: ••— cu o
•— •(- O CO
t- a; o c oj
-
CU U O S-
t- c: — o
ro ex s- >> o
s_ . -
••— E S-
• • ,co —'
o • c
oa»o
•— 1 C ••-
0) .,- . 3 r
**- -P C
— O CO
M
O) O O (D •— -^- -r-
mcc ccnuicE •<—
CU CJ CX O) —•-- JD U CX t-
^ •(— j-) CJ CJ ro E
U JZ KJ O. 3 O) CTi—i 13: 1-
•— cn c o cu o •!- •«- .-03-
•^-OCXCS-4->I_C3 —
3x/> •»— S- ••— Zt. Ul CU E TD '
!_ -^ »^-
QJ O *-> 13
(_> T3 t- U -
LO I- CD
JZ Ul U
4_> CU C -
•^ . J- •-- u-> :
3 c: 3 n r-
>, Ul -C . (
cz •— o u CD :
- - cx-~
O O) .C —
t- . C *J
U OJ — U
Ul CU (Z »<
>*-r- QJ X O) O S- CJ S-
3 T3 t- QJ ••—j CJ U Q_ CD,
•a cn e o -*
>-o oo a.
CJ (J -0 —
CD — OJ
•— c: z. CD
ex zu cx^r
,
— .z: o o
CU ••— QJ
••- z; c CD
ex ex CL«=T
JZ LJ QJ U
— O CX
O Ul >,LU
O 3 O II
O CU O
co czi LTI ar
*j c s- ui o
3 •*— 3 ui c: ui
c: •— in QJ -^
"- J3 O U JZ fO
E 3 CL O —»
-------
In a subsequent paper, Witek et al. (1985) described the symptoms experi-
enced by the subjects in the Schachter et al. (1984) study. Both asthmatics
and normal subjects experienced increased respiratory symptoms following S02
exposure. Normal subjects complained chiefly of upper airway (nose and mouth)
symptoms of odor or unpleasant taste; these symptoms were not increased by
exercise. Normals experienced no significant lower respiratory symptoms.
There was an increase in lower respiratory symptoms in asthmatics at 0.75 and
1.0 ppm S02, although the/significance of this trend is not clear (p = 0.09).
Upper airway symptoms tended to be elevated in both asthmatics and normals, but
more consistently in normals. The lower respiratory symptoms increased with
exercise in the asthmatics and were significantly correlated (r = 0.67, p <.05)
with the decrease in FEV-. Q. In contrast, exercise did not affect symptoms in
normals. The authors stated that even the asthmatics' symptoms were generally
mild and required no therapy.
Linn and coworkers (1983b) also evaluated the responses of naturally
breathing asthmatics exposed to SOp in an environmental chamber (23°C, 85
percent RH) while performing 5 min of moderately heavy exercise (V^ = 48
L/min). Twenty-three mild asthmatics (some of whom were hyperreactive to
methacholine and all of whom were reactive to 0.75 ppm S02) were exposed four
times, once each to 0, 0.20, 0.40, and 0.60 ppm. Significant increases in SRaw
occurred after clean air exposure due to exercise-induced bronchoconstriction.
The SRaw increase after 0.20 ppm was not significantly larger than after clean
air, but the SRaw following exposure to the two higher concentrations was
significantly elevated. SRaw demonstrated a significant trend to increase with
increasing S02 concentration but this trend was not linear; the mean increases
in SRaw after 0.2, 0.4 and 0.6 ppm S02> over those seen with clean air, were
0.54, 2.03, and 6.77 cm H20-sec. The response data are suggestive of a
threshold concentration for response to S02- There is a strong possibility of
a concentration threshold for S02 at low concentrations and ventilations since
the scrubbing of S0? by the upper airway mucosal surfaces may be so efficient
that only a relatively small quantity of S02 reaches the reactive portions of
the airways.
Roger et al. (1985) studied 27 mild asthmatics (methacholine sensitive,
not using cromolyn or steroid medication). Exposures were to 0.0, 0.25, 0.50,
and 1.0 ppm S02 in an environmental chamber (26°C, 70 percent RH) utilizing
natural breathing while performing treadmill exercise (V£ = 41 L/min). The
increases in SRaw post-exercise associated with these exposures were 48, 63,
4-23
-------
93, and 191 percent respectively; the increases at the two highest concentra-
tions were significantly greater than with air. The data reported by Roger et
al. (1985) were further analyzed (Horstman et a!., 1986) in order to determine
individual S02-SRaw dose-response relationships. This analysis included
previously unreported data on exposure to 2 ppm in subjects who were non-
responsive to lower concentrations. From interpolation of the dose-response
plots, the concentration of S02 which provoked a 100 percent increase in SRaw
(PCS02) was determined for each subject. All S02 responses were corrected for
the response observed with clean air, i.e., exercise-induced bronchoconstric-
tion. For the most reactive 80 percent of the subjects the PCS02 ranged from
0.28 to 1.38 ppm; it was greater than 1.95 ppm (and therefore basically
indeterminate since the peak exposure level was 2.0 ppm) in the remaining
20 percent of subjects. (This percentage of S02~insensitive asthmatics is in
general agreement with Linn et al., 1984b) The median PCS02 in all subjects
and under these conditions was 0.75 ppm; 25 percent (i.e., 6) of the subjects
had a PCS02 less than 0.50 ppm, the lowest being 0.28 ppm. The dose-response
relationships relate only to the level of exercise used in this study. Dif-
ferent dose-response relationships would be expected for different exercise
levels or different exposure durations.
4.3.2 SO,,-Induced Versus Nonspecific Airway Reactivity
It is well established that most asthmatics are highly reactive to bron-
chial inhalation challenge with histaminergic (histamine) and cholinergic
(acetylcholine, carbachol, methacholine) agents. Clear evidence has also
emerged that asthmatics are substantially more reactive to S02 than are healthy
subjects. The relationship between S02~induced bronchoconstriction and nonspe-
cific airway reactivity has been examined or alluded to in a number of studies
(Horstman et al., 1986; Witek and Schachter, 1985; Sheppard et al. , 1983).
Airway reactivity to methacholine and to histamine are well correlated (r =
0.70) (Chatham et al. , 1982). Methacholine reactivity was more highly corre-
lated with exercise-induced bronchoconstriction and was better able to distin-
guish between normals and asthmatics (Chatham et al., 1982).
Witek and Schachter (1985) reported that the methacholine reactivity of a
group of 8 asthmatics was highly (r = 0.86, p <0.05) correlated with their reac-
tivity to S02. The subjects were a subgroup of 8 of the 10 subjects used in the
Schachter et al. (1984) study (see Schachter et al., 1984, for protocol details).
4-24
-------
The dose of methacholine required to produce a 20 percent drop in the maximal
expiratory flow at 40 percent VC above RV on a partial expiratory maneuver
(MEF40%-P) was determined. From the MEF40%-P vs. S02 response relationship,
the S0? concentration required to produce a 20 percent drop was determined.
The relationship between the methacholine provocative dose and the S02 provoca-
tive concentration was determined by rank correlation. This study suggests
that there is a relationship between methacholine reactivity and severity of
S0?-induced bronchospasm.
On the other hand, Koenig and Pierson (1985) concluded in their recent
review article that the response to a methacholine challenge was not a good
predictor of the degree of S0?-induced bronchoconstriction in asthmatics. They
suggested that a positive response to an exercise challenge was more likely to
predict a positive response to S02. Linn et al. (1983b) present subject data
(their Table 1) for methacholine reactivity, exercise response (SRaw change),
and S0? response (SRaw change), which are sufficient to allow calculation of
correlation coefficients between these three variables. The rank-order cor-
relation coefficient between methacholine reactivity and S02 response was 0.38,
between exercise response and S02 response was 0.46, and between exercise and
methacholine response was 0.47 (these calculations by the authors of the
addendum). The latter two correlation coefficients were significant (p <0.05)
and this observation supports the suggestion of Koenig and Pierson (1985).
Horstman et al. (1986) have compared the methacholine reactivity (interpolated
dose causing a doubling of SRaw) with the S02 response (PCS02; see previous
section). The methacholine and S02 responses were significantly but weakly
correlated (r = 0.31).
The relationship of histamine reactivity to S02-induced bronchoconstric-
tion is less well described. "Tolerance" to S02 exposure reported by Sheppard
et al. (1983) was not accompanied by any decrease in histamine reactivity.
However, this does not necessarily indicate the absence of an overall relation-
ship between histamine reactivity and S02 responsiveness.
One problem in establishing the strength of the relationship between
non-specific airway reactivity and S02 response is the restricted range of the
observations in these studies which deal only with the most reactive segment of
the population, namely asthmatics. Inclusion of data from normal subjects
would undoubtedly result in a higher correlation. Nevertheless it is apparent
that increased S0? responsiveness in asthmatics cannot simply be ascribed to
elevated non-specific airway reactivity.
4-25
-------
4.3.3 Oral. Nasal, and Oronasal Ventilation
For SO,, in particular, but also for many other gases and aerosols, the
inhalation route is an important factor in delivery of the substance to the
lung. Since 1982, a number of studies have been reported which specifically
address this issue. There are important interactions between the inhalation
route, which in many cases is simultaneous oral and nasal breathing (oronasal)
(Proctor et al. , 1981), and the ventilation rate such that the efficiency
of the oral or nasal mucosa in absorbing S02 declines as the air flow increases.
Approximately 80 percent of the adult population breathes nasally at rest, with
some 10 to 20 percent breathing oronasally (Cole, 1982). As noted in the
previous addendum (U.S. EPA, 1982c) the studies of Kirkpatrick et al. (1982)
and Linn et al. (1982b in the earlier Addendum I; 1983a in the present refer-
ence list) indicated the importance of oronasal airway scrubbing of S0? in
mitigating the effects of S02 during nasal or oronasal breathing.
In an effort to further resolve the interaction between exercise ventila-
tion and route of inhalation in asthmatics, Bethel et al. (1983b) studied 9
mild asthmatics breathing humidified air (23°C, 80 percent RH) through either a
mouthpiece or a divided facemask (ventilation could' be measured separately in
nasal and oral chambers). Subjects worked at 250 (V£ = 26 L/min), 500 (VV = 53
L/min), or 750 kpm (V£, 62 L/min) and breathed either clean air or 0.50 ppm
S02 for 5 min. Mouthpiece inhalation of S02 resulted in increased SRaw at
moderate (231 percent) and heavy (306 percent) workloads, but with facemask
breathing, the SRaw only increased at the heavy workload (219 percent work-
load). The oral component of ventilation during mask breathing was estimated to
be approximately 38 L/min at the heavy workload, similar to the oral ventila-
tion of 41 L/min with mouthpiece breathing at the moderate workload; the
similarity of SRaw responses in these two cases is noteworthy. From these
studies it is apparent that oronasal breathing ameliorates some of the effect
of S02 breathing in asthmatics, but this effect becomes less important as the
exercise workload increases and both the overall ventilation rises and the
relative contribution of oral ventilation to total ventilation increases.
Kleinman (1984) has modeled the bronchoconstriction response to S0? in
relation to ventilation, oral/nasal partitioning of ventilation, and differ-
ences in SOy scrubbing capability of the two upper airways. This model
suggests that differences in response to S02 can be quantitatively accounted
for by differences in penetration of S02 to target sites within the lower or
thoracic airways (defined as structures at or just below the laryngopharynx).
4-26
-------
Because of the possible interference with oral breathing during the face-
mask exposures, Bethel et al. (1983a) studied 10 mild asthmatics exposed to
0.50 ppm S02 in an exposure chamber (23°C, 80 percent RH) to determine if
freely breathing subjects would develop bronchoconstriction at this concentra-
tion. Following 5 min exercise at 750 kpm (VE unreported, approximately 50 to
60 L/min), SRaw increased 39 percent in clean air but increased 238 percent in
0.50 ppm S0? similar to that previously observed with facemask breathing. Thus
mild asthmatics performing moderate to heavy exercise exhibited clear evidence
of bronchoconstriction after 5 min exposure to 0.50 ppm S02 while breathing
unencumbered.
In a subsequent study (Bethel et al. , 1985), the effects of 0.25 ppm S02
were studied in 19 mild to moderate asthmatics using a similar protocol (23°C,
36 percent RH with 5 min exercise at 750 kpm). SRaw increased from 6.4 to
11.3 post-exercise in clean air and from 5.7 to 13.3 post-exercise in 0.25
ppm S0?. The slightly greater response following S02 exposure was apparently
significant (p <0.05, Wilcoxon one-tailed sign test). The application of a
signed rank test, preferable in this case, would not confirm this significance.
However, when the workload was increased to 1000 kpm in 9 of the 19 subjects,
the increase in SRaw after clean air exercise was slightly, but not signifi-
cantly, greater than that after exercise with 0.25 ppm S02- The authors
suggested that the threshold concentration of S02 which may cause broncho-
constriction in mild asthmatics under conditions of moderate to heavy exercise
appears close to 0.25 ppm. However, the very small rise in SRaw at only one
work output indicates that the additional effect of 0.25 ppm S02 (over that
produced by exercise) is of minor, if any, clinical significance. Neverthe-
less, it must be stressed that these asthmatics had relatively mild disease.
Koenig et al. (1983b) examined the effects of exposure to 0.5 and 1.0 ppm
S02 combined with a sodium chloride droplet aerosol in nine extrinsic adoles-
cent asthmatics. Judging from their medication requirements, this group of
asthmatics would have to be considered more severe than the adult asthmatics
studied by several other investigators. The exposures were delivered via
mouthpiece (22°C, 75+ percent RH) for 10 min during moderate treadmill exercise
(30 min rest exposures were followed by 10 min exercise). The responses ranged
from a 15 percent decrease in FEV-j^ Q at 0.5 ppm to a 61 percent decrease in
V at 1 0 ppm. The response to 1.0 ppm tended to be greater but this
max75 ^ . . ,
difference between S0£ concentrations did not attain overall statistical
4-27
-------
Nevertheless, the effects of S02 on lung function persisted
FEV
l.O1
Vmax50 and
significance.
longer after the higher concentration exposure.
(partial flow volume curves) were significantly reduced and total respiratory
resistance (forced oscillation) was significantly increased following
mouthpiece breathing of 0.5 or 1.0 ppm S02. Seven of nine subjects were also
exposed to 0.5 ppm S02 plus aerosol delivered via a facemask (ventilation 5 to
6 times rest or 30 to 50 L/min). The pulmonary function changes after
breathing 0.50 ppm S02 plus aerosol via facemask were not significantly dif-
ferent from baseline. However, some of the subjects intentionally breathed
through their nose rather than oronasally; therefore the comparison of the
results of this study with those of Bethel et al. (1983a) would not be
appropriate.
Previous studies (Andersen et al., 1974) cited in the criteria document
have suggested that nasal resistance increases following S02 exposure. Because
this could have an important impact on the route of inhalation and/or the
oronasal ventilation switch point, Koenig and associates (1985) examined the
effects of 0.50 ppm S02 on the work of nasal breathing in a group of moderate
adolescent asthmatics (7/10 were theophylline users). Subjects were exposed
o
to S02 (and H2S04 aerosol — 100 ug/m ) either via mouthpiece or oronasal
facemask (22°C, 75 percent RH). Thirty min resting exposure was followed by 20
min of moderate exercise on a treadmill (VE = 43 L/min). Exposure to S0? via
mouthpiece or facemask resulted in an approximate 30 percent increase in nasal
work of breathing (measured with a divided diving mask containing two pressure
transducers which measured the pressure drop across the nasal passages). Due
to marked inter- and intra-individual variability in these nasal measurements,
only the increase in nasal work of breathing after facemask exposure, measured
at 22 min post-exercise, was found to be statistically significant. Exercise,
per se, is associated with a reduction in nasal resistance which persists for
about 5 to 15 minutes after exposure ceases (Forsyth et al. , 1983). The
effects of S02 on nasal resistance may therefore be offset by the effects
associated with exercise leading perhaps to minimal changes in nasal resistance
immediately post-exercise. No changes in nasal resistance were observed after
clean air or sulfuric acid aerosol exposure. The decreases in FEV., n and
* X. U
Vmax50 were S1~9nificantly greater with mouthpiece than with facemask exposure
to 0.50 ppm S02. The implications of this finding are not clear at present.
If S02 raised nasal resistance during exercise (and this is not presently
4-28
-------
known), the relative proportion of oral ventilation could increase. However,
the reduction in nasal resistance associated with exercise may override or
negate the effect of S02 on nasal resistance. Increased oral breathing of
S0? could result in a greater delivery of SCL to airways below the larynx.
4.3.4 Time Course of Response to S00 in Asthmatics
Early studies of SCL exposure in normal healthy subjects indicated that
the peak response occurred early in exposure and was reduced with continued
exposure. The effect of prolonged or repeated exposure has recently been
addressed in asthmatics.
Sheppard and associates (1983) reported the responses of mild to moderate
asthmatics (n = 8) exposed three consecutive times to 0.5 ppm SO™. The
subjects performed voluntary eucapnic hyperpnea with 0.5 ppm SCL for 3 min at a
ventilation which had previously caused bronchoconstriction (air temperature =
22.6°C, RH = 82 percent). Three subjects failed to reach the target of a 60
percent increase in SRaw above baseline and consequently performed additional
hyperpnea to produce increased SRaw. Twice more, at 30-min intervals, the SO,,
hyperpnea was repeated. SRaw was measured before and after each S02 exposure.
A single bout of S0? hyperpnea was performed on the following day and again one
week later. The first exposure to S02 caused a doubling of SRaw (104 percent
increase). The second and third SQ2 exposures elicited only modest increases
in SRaw. (35 percent, 30 percent respectively). However, 1 day and 7 days
later, the response to S02 was similar (+89, +129 percent) to that on the first
exposure.
In this study, the relationship of SD2 tolerance to histamine-induced
bronchoconstriction was examined in a subgroup of four subjects. A baseline
histamine challenge test was followed 30 min later by two 3-min periods of S02
breathing separated by 30 min (as in the initial part of the study). When the
histamine challenge was repeated after a further 30 min, the histamine dose-
response relationship was unchanged despite the blunted response to S02 inhala-
tion. This study demonstrated that repeated exposure of asthmatics to 0.5 ppm
S0? by mouthpiece at 30-min intervals resulted in a blunted S02 response (toler-
ance) which persisted for at least 30 min but was absent after 24 h and was not
associated with any change in airway reactivity to histamine. The implications
of this study for response mechanisms are discussed in Section 4.4.
4-29
-------
Linn et al. (1984c) also studied the effect of repeated SCL inhalation in
14 mild to moderate asthmatics who were exposed to 0.6 ppm S0? for 6 h on each
of two consecutive days. These were compared with similar clean air exposures.
They performed two 5-min bouts of exercise (V£ = 50 L/min), one immediately
upon entering the exposure chamber (22°C and 85 percent RH) and a second bout 5
h later. SRaw was measured immediately post-exercise and at hourly intervals
between exercise periods. With S02 exposure, SRaw was approximately doubled
following each exercise bout. Small increases in SRaw also occurred following
exercise in clean air. There were no differences in response between early and
late exercise challenges and no significant differences in SRaw response
between exposure days. SGaw, but not SRaw, responses indicated smaller
decreases on the second SO,, exposure day (-0.091 sec~1-cmH00"1) than the first
-I -1
(-0.119 sec -cmH20 ). This difference was of only marginal statistical
significance and not of any clinical importance. The results of this study
indicate that S02~exercise challenges separated by 5 h (between- exercise
periods) produce essentially similar responses and that the responses are not
appreciably different on two consecutive days. The Linn et al. (1984c) and
Sheppard et al. (1983) studies had several methodological differences;
respectively, these were free breathing vs. mouthpiece, exercise vs. eucapnic
hyperpnea, 4.5 h vs. 30 min interexposure interval, 5 min vs. 3 min exposure
duration, and 0.6 ppm vs. 0.5 ppm S02 concentration. Nevertheless, in each
study, an initial SO,, exposure which produced at least a doubling of SRaw was
followed later by a second exposure. With the shorter 30-min interval in the
Sheppard study, the response to S02 was blunted. However, with the longer 5-h
interval in the Linn study, the S02 response was unchanged from the initial
exposure. Evidence from the exercise-induced bronchoconstriction literature
(Edmunds et al., 1978; Stearns et al. , 1981) indicates that the refractory
period following exercise induced bronchoconstriction persists for 2 to 4 h.
The refractory period following SO,,-induced bronchoconstriction lasts at least
30 min but less than 5 h.
Snashall and Baldwin (1982) studied the effect of exposures to 8 ppm SO^
repeated at 4 h and 24 h in 4 normal and 1 asthmatic subjects. Compared to the
initial exposures, S0?-induced bronchoconstriction was reduced 42 percent at 4
h while no difference was observed at 24 h.
In a more comprehensive examination of repeated exercise during continuous
SOp exposure in a large subject population (n=28) exposed to 3 different S02
4-30
-------
levels with repeated exercise, Roger et al. (1985) also observed attenuatfon of
S09-induced bronchoconstriction. The subjects worked at a moderate workload
(VE = 42 L/min) and breathed freely (except for 2 min at the end of exercise
periods 2 and 3). They were not selected for S02 sensitivity, were sensitive to
methacholine challenge, and used no cromolyn or steroids. Each subject was
exposed, on three different days, to three S02 concentrations (0.25, 0.50, and
1.0). During each exposure, the subject exercised three times for 10 min each
separated by 15-min intervals between exercise bouts. SRaw was measured pre-
exposure and following each exercise period. After the first exercise, SRaw
increased significantly over that seen with clean air (48 percent), with
exposure to both 0.5 (+93 percent) and 1.0 ppm S02 (+191 percent). With
subsequent exercise bouts in both 0.5 and 1.0 ppm S02, the SRaw increased only
about half as much (third exercise SRaw increase was 52 percent and 116 percent
in 0.5 and 1.0 ppm, respectively). This attenuation of response was less than
that seen by Sheppard et al. (1983). Nevertheless, there were several
differences between the two studies (exposure duration 3 min vs. 10 min,
inter-exposure interval 30 min vs. 15 min, mouthpiece eucapnic hyperpnea vs.
free breathing exercise, S02-sensitive vs. methacholine-sensitive selection
criterion). The Roger et al. (1985) subjects demonstrated a refractoriness to
both exercise in clean air and to exercise in S02; the latter was of greater
absolute magnitude in terms of less increase in SRaw but the relative reduction
in response from first to last exercise periods was similar for repeated
exercise in either clean air or S02-
A subset of 10 subjects from the Roger et al. (1985) study were further
studied by Kehrl and coworkers (1986). The subjects were selected for moderate
S0? sensitivity (i.e., no subjects nonresponsive to S02 were used and the most
reactive subjects were not studied). In addition to the three 10-min exercise
periods performed previously, these subjects exercised continuously for 30 min
at the same exercise intensity (\/E = 41 L/min) in an environmental chamber
(26°C, 70 percent RH) while exposed to 1.0 ppm S02- The SRaw data for the
original intermittent exercise exposures were similar to those of the original
larger subject group (SRaw: baseline, 5.4; post-exercise-1, 14.7; post-
exercise-2, 12.8; post-exercise-3, 11.1). After 30 min continuous exercise in
1.0 ppm SO-, SRaw significantly increased from 5.2 to 17.3 cm H20-sec. The
SRaw change was not significantly different than that seen after the first 10
minute exercise period of the intermittent exercise exposure. This study
4-31
-------
demonstrated that SO,,-induced bronchoconstriction is elicited by 10-min expo-
sures but a further 20 min of continuous exercise resulted in only a slightly
greater increase in SRaw which did not attain statistical significance.
In order to examine the time course of recovery from S02~induced broncho-
constriction in asthmatics, Hackney et al. (1984) exposed 17 mild to moderate,
nonsmoking, S02~sensitive asthmatics (not using cromolyn or steroid medication)
to 0.75 ppm S02 for 3 h. A secondary objective of this study was to determine
the usefulness of spirometric testing as an adjunct or alternative to
plethysmography under such exposure conditions. The exposure consisted of 3 h
in an environmental chamber with a 10-min exercise period (\L = 45 L/min) at
the beginning of the exposure followed by post-exercise and hourly SRaw mea-
surements. SRaw was approximately quadrupled (+263 percent) after exercise,
returned almost to baseline at one hour (+34 percent, not significant) and was
unquestionably back to baseline after 2 h recovery. In an otherwise identical
exposure sequence which included spirometric testing, the FEV, „ was
significantly reduced (-20 percent) post-exercise. The correlation between the
FEV.^ 0 and SRaw changes was significant (r = 0.60) but accounted for
considerably less than half the variance, indicating that the two measures did
not track each other closely in all subjects. This study demonstrated that
moderate S02/exercise-induced bronchoconstriction will be relieved during rest
(over a 1 to 2 h period) even if a low-level S0? exposure is continued. Second
the authors demonstrated that changes in FEV.. Q are also useful indicators of
S02 exposure in asthmatics, although it is not clear that significant changes
in FEV^ Q would occur with less severe exposure more typical of the ambient
environment.
4.3.5 Exacerbation of the Responses of Asthmatics to S02 by Cold/Dry Air
It has been well established that both cold air and dry air can exacerbate
bronchoconstriction in asthmatics (Deal et al. , 1979a; Strauss et al., 1977).
The precise mechanism(s) for the effect are not universally agreed upon
(Anderson, 1985). Although direct convective cooling of the airway plays a
minor role, the major avenue of heat loss is due to evaporation to humidify the
inspired air. -Evaporation of airway surface liquid may also lead to other
changes discussed in section 4.4. The potential for evaporative cooling by
inhaled air can be most readily appreciated from the determination of the
4-32
-------
absolute humidity of the inspired air. Absolute humidity (AH) expresses the
water content of the air in mg/L (g/m3). The lower the AH., the greater the
potential for evaporative cooling. AH is listed, for each study, in Table 4.
For reference, the AH of saturated air at 37°C (i.e., BTPS) is 44 mg/L.
Therefore, in order to bring inspired air at 0°C, AH = 1 mg/L, to BTPS, the
temperature of each liter of air must be increased to 37°C (0.011 kcal) and 43
mg of water must be evaporated (0.025 kcal) (calculated from the respiratory
heat exchange equation of Deal et al., 1979b).
Sulfur dioxide exposure can occur during the winter months when the
ambient air temperature is low, and consequently the water vapor content is
reduced. Accordingly, Bethel and coworkers (1984) examined the separate and
combined effects of sulfur dioxide and cold dry air in seven asthmatics (mild
to moderate asthma) breathing via mouthpiece. In this study and the following
study by Sheppard and coworkers (1984), a series of bronchoprovocation tests
were used. The methods are as follows:
The subjects breathed a test gas mixture for 3 min, then SRaw was deter-
mined every 30 s for 2 min. This cycle of 3 min exposure and 2 min SRaw
testing was repeated until the desired response was achieved. The
ventilatory bronchoprovocation test consisted of performing voluntary
eucapnic hyperventilation at increasing ventilation levels (20, 30, 40,
50, 60, etc. L/min) while breathing a single test gas mixture. The SO,,
bronchoprovocation test consisted of breathing (eucapnic hyperventilation)
at some fixed ventilation and gas temperature and humidity with succes-
sively doubling levels of sulfur dioxide (e.g., 0, 0.125, 0.25, 0.50, 1.0,
2.0 ppm SOp) used as the stimulus.
Bethel's subjects performed ventilatory bronchoprovocation tests with both 0.50
ppm S02 in warm humid air and with no S0'2 in cold-dry air .(-11°C, dew point
-15°C) until an increase in SRaw was observed in order to determine the venti-
lation which caused "little or no bronchoconstriction" with either stimulus.
At the selected ventilation, subjects breathed on a mouthpiece for 3 min one of
the following mixtures: (1) warm-humid (23°C, dew point = 18.4°C) air, (2) warm
humid air with 0.50 ppm S02, (3)cold dry air, (4) cold dry air with 0.50 ppm
S0?. Modest but nonsignificant increases in SRaw followed each of the first
three conditions [(1) +3 percent, (2) +38 percent, (3) +18 percent]. However,
4-33
-------
the combination of 0.50 ppm S02 and cold dry air caused a striking increase in
SRaw (from 6.94 to 22.35, or a 222 percent increase). In this study, the
combined effect of breathing cold dry air and 0.50 ppm S02 via mouthpiece was
clearly larger than the sum of the individual response to either SO,, or cold
dry air.
Sheppard and coworkers (1984) further explored the interaction of breath-
ing cold dry air and S02 via mouthpiece in a group of 8 mild asthmatics. The
purpose of the study was to determine the relative contributions of decreased
air temperature (-20°C) and reduced water vapor content (0 percent RH). Using
a ventilatory bronchoprovocation test with cold dry air, the highest ventila-
tion which did not cause increased SRaw was determined. The study consisted of
having the subjects perform eucapnic voluntary hyperpnea, at the selected
ventilation, 6 consecutive times for 3 min at a time with 2 min intervals
between efforts. This was done on four separate occasions (different days)
ordered randomly. On one occasion, the subject breathed cold-dry air only;
this did not cause an increase in SRaw. The three other tests consisted of S0?
bronchoprovocation tests at the selected ventilation with successive doubling
of S02 concentrations (starting at 0.125 ppm), one with cold-dry air, one with
warm-dry (22°C, 0 percent RH) air, and one with warm-humid (22°C, 70 percent
RH) air. The S02 concentration required to produce a doubling of baseline SRaw
(PC100) was interpolated from the dose-response curve. The PC100 for cold-dry
air (0.51 ppm) and for warm-dry air (0.60 ppm) were not significantly different
but both were less than the PC100 for warm-humid air (0.87 ppm). The PC100
measured in this study may not be a useful effects index because the response
may be a function of the cumulative effect of all S02 concentrations breathed,
as noted by the authors. In addition, the authors considered the possible
mitigating effect of repeated exposure - tolerance, but the importance of
this is unclear. Further studies were then performed using a ventilatory
bronchoprovocation test while breathing either 0.0, 0.1, or 0.25 ppm S0? in
warm-dry air. From the ventilation-SRaw dose-response plots at each S0?
concentration, the ventilation producing an 80 percent increase in SRaw (PV80)
was determined. The PV80 at 0.0, 0.1, and 0.25 ppm S02 were 54.9, 51.1, and
49.3 L/min, respectively. The differences in PV80 between 0.1 or 0.25 and
clean air (0.0 ppm) reportedly reached significance although it was not clear
how these data were analyzed (presumably repeated measures analysis of
variance). Regardless of whether or not the difference in PV80 between clean
4-34
-------
air and 0.1 and 0.25 ppm S02 was statistically significant, the magnitude of
this difference is small and of no established or obvious clinical importance.
Nevertheless, the first part of this study did confirm that breathing dry air
and cold air potentiates sulfur dioxide-induced bronchoconstriction. This
potentiation could be an additive effect since both cooling (convective and
evaporative) and drying of the airway may act as direct bronchoconstrictive
stimuli, per se (Sheppard et al., 1984). In addition, the drying of the upper
airway also reduces the ability of the oropharynx to scrub S0~ from the inhaled
air and may also cause a concentrating effect of the remaining airway surface
liquid (see Mechanism section).
Concurrent studies by Linn and coworkers (1984a) also were directed at the
possible interaction of inhalation of sulfur dioxide and cold air. They
studied a group of 24 mild to moderate S02-sensitive asthmatics. A preliminary
study to determine the effects of humidity at cold ambient temperatures in-
cluded eight subjects exposed to 0.0, 0.2, 0.4, and 0.6 ppm S02 at 5°C under
two humidity conditions (81 percent and 54 percent). The subjects exercised
for 5 min in an environmental chamber at a workload selected to elicit a
ventilation of approximately 50 L/min (range 37 to 60) and breathed naturally.
SRaw showed a tendency to increase more from pre- to post-exposure with
increased S0? concentration. The post-hoc analyses for changes at each
concentration were not presented, presumably because of the small sample size
and the non-randomized experimental design. No effect of ambient humidity on
response to S0? was seen at the 5°C air temperature. However, the difference
in water vapor content at the low and high humidities was approximately 1.84
mg/L, approximately 1/20 of the difference in water vapor pressure between
ambient and BTPS, and thus the absence of a difference should have been expect-
ed. A second study in this same series compared responses of 24 asthmatic sub-
jects exposed to 0.6 ppm S02 under warm-humid (22°C, 85 percent RH, AH = 16.5)
and cold humid (5°C, 85 percent RH, AH = 3.4) conditions. The same exercise and
natural breathing procedures as above were followed. Breathing 0.0 ppm S02,
subjects had small nonsignificant increases in SRaw under warm (27 percent) and
cold (38 percent) conditions. Exposure to 0.6 ppm S02 under these temperature-
humidity conditions produced significant increases in SRaw in both warm (132
percent) and cold (182 percent) conditions. However, the temperature effect,
unlike in the Sheppard et al. (1984) and Bethel et a.l. (1984) studies, was not
significant although the trend was in the direction of an increased response at
4-35
-------
the lower temperature. The temperature difference between cold and warm air
was larger in the Sheppard et al. and Bethel et al. studies (42°C and 34°C,
respectively) compared to the Linn et al. study (17°C). However the cold-warm
difference in inspired air water content (AH) were similar for the three studies
(14.8, 12.6, 13.1 respectively). Nevertheless, it is apparent that the exacer-
bation of S02~induced bronchoconstriction by cold air, containing small quanti-
ties of water vapor, is minimal in freely breathing asthmatics exposed during
moderately heavy exercise at 5°C air temperature.
In order to determine the possible effects of even colder ambient air
temperatures, Linn et al. (1984b) exposed 24 mild S02-sensitive asthmatics
(including 11 subjects from Linn et al., 1984a) to 0.0, 0.3, and 0.6 ppm S0?
at +21, +7, and -6°C (RH approximately 78 percent). The exposure duration was
5 min. The authors noted that "only 10-20 percent of clinically asthmatic pro-
spective subjects had to be rejected as non-responsive to SO" (10 min exercise
at 40 L/min breathing 0.75 ppm S02). There was a significant effect of decreas-
ing air temperature and of increasing S0? concentration on the post-exercise
SRaw. However, the authors reported that there was no statistically significant
interaction of air temperature and S02 concentration for SRaw although the
interaction was apparently significant for SGaw. The effect of cold air (in
increasing SRaw or decreasing SGaw) was most pronounced with the 0.0 ppm S02
exposures and minimal with 0.6 ppm exposures. The results of this study do not
support the hypothesis that S02 acts synergistically with cold air in freely
breathing, exercising, mild to moderate asthmatics. The authors concluded that
the cold air and S02 effects "acted additively at most." The results for the
7°C and 21°C 0.6 ppm S02 exposures (+207 percent, +150 percent SRaw) were simi-
lar to those seen in their previous study (Linn et al., 1984a) (+182 percent,
+132 percent SRaw), thus demonstrating the reproducibility of these studies.
In order to study the full range of S0?-temperature-humidity interactions,
Linn et al. (1985a) also examined the effects of warm-dry (38°C, 20 percent RH)
and warm-humid (38°C, 85 percent RH) conditions on 22 S02-exposed (0.6 ppm)
asthmatics. The exposure protocol was similar to the two 1984 studies with a 5
min chamber exercise period and ventilation of approximately 50 L/min. The
experimental desfgn was a three-factor (S0?-0.0 and 0.6 ppm; temperature-21 and
38°C; and humidity-20 percent and 80 percent) factorial design with repeated
measures across all factors. In this study, the major differences would be
anticipated to occur between the warm humid (38°C, 85 percent RH) condition and
4-36
-------
the cooler dryer condition (21°C, 20 percent RH). There were significant
effects of temperature, S0? and humidity on the delta-SRaw (pre- to post-
exercise) response and significant -temperature-SOp and humidity-SOp inter-
actions. The largest clean air increase in SRaw (20 percent) occurred with
cool-dry air and the smallest with warm-humid. The largest SO,, induced
increase in SRaw (204 percent) occurred under cool-dry conditions and again the
smallest change (35 percent) occurred under warm-humid conditions. Symptoms
showed a similar pattern of response after S0? exposures with lower symptoms
scores under warm-humid than cool-dry conditions. SRaw responses to 0.6 ppm
SOp under 21°C-humid conditions were similar for all three Linn et al. studies
(1984a, 132 percent; 1984b, 150 percent; 1985a, 157 percent). The response
under warm humid conditions was considerably less. The authors discussed the
possibility that they observed a synergism between SOp exposure and airway
drying/cooling due to reduced temperature or humidity of inspired air.
4.3.6 Clinical Relevance
As discussed in the introduction, there is no obvious or clearcut point
where S0? effects cease to be a mere annoyance and became an adverse health
effect. The "clinical" importance of the various observations cannot be
interpreted in an unequivocal fashion. There were no reports of cases where
subjects required emergency treatment or hospitalization following S02
exposure. Furthermore, there was no evidence reported which indicated that
brief SOp exposure caused either acute or chronic changes in nonspecific
airway reactivity and the majority of subjects recovered spontaneously within
an hour. The responses (SRaw and FEV., n) were no greater than those observed
with exposure to aeroallergens and no delayed effects of SOp were reported.
However, in addition to changes in spirometry, airway resistance, and
symptoms of wheezing and chest discomfort; several "clinically" relevant
observations were documented. These observations are summarized in Table 8.
The "clinical" significance of these responses included the use of medication
following S0? exposure, the modification of activity, or the inability to
complete the SOp exposures. The repeatability of such responses is demon-
strated, by one subject from the Koenig et al. (1985) study, who participated
in two exposure to 0.50 pmm SOp and, in both cases, was unable to complete the
exposure and required a brochodilator to reverse the bronchoconstriction.
4-37
-------
TABLE 8. CLINICALLY SIGNIFICANT RESPONSES
Study Reference
Responses
Bethel et al. (1984)
Koenig et al. (1985)
Linn et al. (1984b)
Linn et al. (1984a)
Linn et al. (1984c)
Roger et al. (1985)
(Horstman et al., 1986)
2/7 subjects required bronchodilater after cold
air + 0.50 ppm S02
2/10 subjects exposed to 0.50 ppm S02 via mouth-
piece could not complete exposure; required
bronchodilator to reverse bronchoconstriction.
One tried again with the same result.
1/24 took isoproterenol after 0.4 ppm S02
3/24 took isoproterenol after 0.6 ppm S02
1/24 took isoproterenol after 0.6 ppm S02
One subject required reduced exercise level
to complete exposure at 0.6 ppm S02
2/28 subjects unable to complete exposure
regimen. One dropped out at 0.5 ppm S02
(he required medication - anecdotal report)
Second subject unable to complete exposure
at 1.0 ppm S02
4.4 MECHANISM(S)
4.4.1 Mode of Action
A single unequivocal definition of asthma is not realistic on the basis of
existing knowledge and the heterogeneity of the disease. The single condition
that is common to all definitions of asthma is the reversibility of slowed
forced expiration presumably due to airway narrowing (smooth muscle
contraction, excess mucous secretion, mucosal edema). Most current definitions
of asthma also include the concept of nonspecific airway hyperreactivity (e.g.,
methacholine, histamine). The present American Thoracic Society definition of
asthma is:
A disease characterized by an increased responsiveness of the airways to
various stimuli and manifested by slowing of forced expiration which
changes in severity either spontaneously or with treatment.
It is noteworthy that the data summarized in this addendum indicate that asth-
matics experience substantial, but transient, bronchoconstriction (slowed
4-38
-------
forced expiration) when exposed to low S02 concentrations (i.e. increased
responsiveness).
Because of its relatively rapid reversibility, SO^-induced bronchocon-
striction in asthmatics is likely the result of decreased airway caliber caused
by contraction of airway smooth muscle. The study of Roger et al. (1985) in-
dicated the largest S0?-induced increases in airway resistance measured by
plethysmography were associated with increases in the low frequency component
of respiratory system impedance measured by the forced random oscillation
(noise) technique. The interpretation of this finding was an elevated peri-
pheral resistance associated with constriction of anatomically peripheral or
small airways. However, narrowing of central upper airway structures such as
the larynx and glottis may accompany increased airway resistance (Cole, 1982)
and it is_possible that some of the increase in airway resistance may be due to
elevated laryngeal or glottal resistance.
Contraction of airway smooth muscle in response to environmental stimuli
can be evoked by intrinsic chemical and/or physical stimuli acting via neural
and/or humoral pathways. SOp may either act directly on smooth muscle or may
cause the release of chemical mediators from tissue, especially the release of
histamine from mast cells. It is beyond the scope of this document to provide
even a brief review of the mechanism of action of all the possible pharmaco-
logic mediators of SCL-induced bronchoconstriction. However, some plausible
candidates include histamine, slow-reacting substance of anaphylaxis, leuko-
trienes, and prostaglandin F?-alpha, all of which are released in the airways
and can cause smooth muscle contraction. • ' .
As reported in the previous addendum (U.S. EPA, 1982c), both activation of
parasympathetically mediated reflexes (Nadel et al. , 1965; Sheppard et al. ,
1980) and mast cell degranulation (Sheppard et al. , 1981) with consequent
release of chemical mediator (most likely histamine) play a significant role
in SOp-induced bronchoconstriction. While the specific mechanism whereby S02
interacts with the airways to induce bronchoconstriction has not been elucidated,
additional studies relevant to the mechanism(s) have appeared since the
previous addendum. These studies assessed the inhibitory effects on SC^-induced
bronchoconstriction of a variety of receptor antagonists (drugs that bind the
receptors but do not stimulate the receptor-induced response). Results from
these studies suggest that mechanisms in addition to reflex bronchoconstriction
and mast cell degranulation may play a significant part in the responses of the
asthmatic airway to SCL.
4-39
-------
Snashall and Baldwin (1982) studied the effects of atropine and cromolyn
on relatively mild bronchoconstriction (Raw increased <100 percent above
baseline) induced by breathing 8 ppm SOp at rest. Both atropine and cromolyn at
least partially blocked SO^-induced bronchoconstriction in all but one of 11
normal subjects. The degree of atropine blockade was inversely related to the
magnitude of the S02~induced response (r = -0.75), i.e., small responses were
completely blocked, while there was little blockade of large responses. For
asthmatics, atropine enhanced S02-induced bronchoconstriction in three of four
subjects tested; minimal blockade was observed in the remaining subject.
Cromolyn blocked the SOp-induced response in three of the four asthmatic
subjects.
Tan et al. (1982) exposed resting normal and atopic subjects to 20 ppm and
asthmatics to 10 ppm S0? to induce bronchoconstriction. Both ipra^ropium
bromide (IB, an anticholinergic agent similar to atropine) and cromolyn par-
tially inhibited the SO^-induced response in all normal and atopic subjects
tested. For asthmatics, IB had little effect on S02-induced bronchoconstriction
in five of nine subjects and afforded only partial blockade in the remaining
four subjects. Cromolyn at least partially inhibited S02~induced bronchocon-
striction in all 18 asthmatics tested. Clemastine (a selective H-, receptor
antagonist without anticholinergic or antiserotinergic activity) effectively
blocked the S02~induced response in five of seven asthmatic subjects tested.
Koenig et al. (1987) studied a group of adolescents with a history of
allergy but without clinically documented extrinsic asthma by the authors'
criteria. All had exercise-induced bronchospasm, defined as a 15 percent or
more reduction in FEV, n after 6 min exercise at 85 percent of V0? . The
aim of this study was to determine whether the beta-2 sympathomimetic drug
albuterol could inhibit SO^-induced bronchoconstriction. Following baseline
lung function tests, the subjects received either placebo or 180 (JQ of albuterol
aerosol, again performed lung function tests, and then exercised for 10 min
(VV = 34 £/min). The subjects were exposed to clean air or 0.75 ppm SOp after
either placebo or albuterol pretreatment. Pretreatment with albuterol resulted
in a 6 to 8 percent increase in FEV, ~ and a 17 to 23 percent decrease in
resistance (forced oscillation method). In clean air, exercise plus placebo
resulted in a 4 percent decrease in FEV, „. In S0?, exercise plus placebo
resulted in a 15 percent decrease in FEV, „ below the preplacebo baseline.
Following pretreatment with albuterol, change in FEV, „ did not differ between
4-40
-------
S0? and clean air exposure and in neither case did FEV, „ drop below the
prealbuterol baseline. This study illustrates that SCL induced broncho-
constriction can be prevented (or at least the response curve can be shifted)
by pretreatment with a beta-2-sympathomimetic drug. Albuterol could act by
inhibiting smooth muscle contraction or by inhibiting mast cell degranulation
or by a combination of the two effects.
Myers et al. (1986a) examined the effect of cromolyn on SOp-induced
bronchoconstriction in a group of 10 asthmatics. They demonstrated a dose-
dependent inhibition of SCL-induced bronchoconstriction by cromolyn given
prior to the S0? exposure. It was also established that the cromolyn did not
reduce nonspecific airway reactivity. In fact, the methacholine reactivity
increased with the higher dose (200 mg) of cromolyn. The mechanism by which
cromolyn exerts this inhibitory effect on S0?-induced bronchoconstriction is
not established but could result from the inhibitory effect of cromolyn on
mast cell degranulation.
In a subsequent study (Myers et al. , 1986b); the effects of chromolyn
plus atropine or S0?-induced bronchoconstriction were studied in 9 subjects, 7
of whom participated in the previous study. It was demonstrated that the
combination of atropine and chromolyn was more effective at inhibiting S0?-
induced bronchoconstriction than either agent above. The SOp dose-response
curve was not reproducible for some subjects who participated in both studies.
4.4.2 Breathing Mode and Interaction With Dry Air
There is no question that the magnitude of S0?-induced bronchoconstriction
is significantly greater with oral than with oronasal or nasal breathing
(Kirkpatrick et al. , 1982). When S02 is inhaled by mouth more S02 penetrates
beyond the pharynx to sites involved in the induction of bronchoconstriction
(Bethel et al. , 1983b; Kleinman, 1984). It is assumed that because of their
geometry and greater relative surface area, the nasal passages are capable of
effectively removing most S02 breathed at rest and a large percentage during
conditions of increased ventilation (exercise, isocapnic hyperpnea). While
there is certainly less relative surface area available for S02 scrubbing in
the oral cavity, other factors may also influence increased bronchoconstriction
associated with mouth breathing of S02, especially at higher ventilation rates.
Increased oral ventilation may result in substantial drying of both upper
(oral and pharyngeal area) and lower (larynx and trachea) airways. The extent
4-41
-------
of airway surface drying will depend upon the ventilation (air flow rate) and
water content of inhaled air. Airway drying could lead to alterations in both
the quantity and properties of surface liquid in the airways. It is not known
whether changes in the volume or water content of the surface liquid lining the
upper airway will result in altered SOp uptake or the penetration of the gas to
sites in the intrathoracic airway more likely involved in the induction of
bronchoconstriction. Another factor which is altered by drying of airway
surface liquid is its osmolarity. Hyperosmolar solutions can induce broncho-
constriction (Anderson, 1985) but it is not known whether changes in mucous
osmolarity may affect the functional response to SO,,.
Two laboratories (Cardiovascular Research Institute, UCSF, and Rancho Los
Amigos Hospital) have performed the bulk of the work on the interaction of SCL
breathing and inhaled air temperature and humidity. Although the results of
the two labs have been qualitatively similar, the mouthpiece breathing studies
(e.g., Bethel et al., 1983b) have typically yielded more pronounced increases in
airway resistance. In S0? exposures using oronasal ventilation, interlabora-
tory differences have been smaller. The use of mouthpiece breathing results in
a more direct airflow path of lower resistance than does unencumbered oronasal
breathing (Proctor et al., 1981; Cole, 1982). Under situations of unencumbered
oronasal breathing, the mouth may act as an effective organ of air modification
(i.e., warming, humidifying, scrubbing of particles and soluble gases). During
mouthpiece breathing, this effectiveness is reduced because of the alteration
in oral airway geometry and the bypassing of some of the oropharyngeal surfaces
involved in air modification. Thus some of the difference between laboratories
may be due to differences in the amount of airway drying and the volume of
nasal ventilation, both of which would favor greater upper airway S0? scrubbing
in studies using oronasal ventilation. Undoubtedly subject selection criteria
and medication also play an important role in the magnitude of response but
such differences between study series are not obvious (see subject table).
Another possibility, noted incidentally by Koenig et al. (1985), is that
subjects may deliberately breathe via the nasal airway, despite the higher
resistance, in order to alleviate both the drying effect due to cold (and/or
dry) air and the effect of S02 which may be associated with the distinctive
odor or taste.
4-42
-------
Cole (1982) notes that approximately 85 percent of adults are preferential
nose breathers who only resort to oral or oronasal breathing under the demand-
ing conditions of exercise, nasal obstruction, or speech. This occurs despite
the fact that upper airway resistance via the nasal airway is about twice that
via a mouthpiece. However, Bethel et al. (1983b) suggest more asthmatics may
breath oronasally and that asthmatics switch from nasal to ore/nasal breathing
at a lower ventilation than normals; this is due to the greater prevalence of
rhinitis in the asthmatic population.
4.4.3 Tolerance (Attenuation of Response) to SO., With Repeated Exposure
Attenuation of S0?-induced bronchoqonstriction with repeated S02 exposure
(with eucapnic hyperpnea) was not associated with a decrease in airway respon-
siveness to. histamine (Sheppard et al. , 1983).. This probably indicates that
the attenuation of response was not related to decreased responsiveness of
airway smooth muscle or decreased responsiveness of vagal reflex pathways.
However it is possible that, within the lung, the regional dosimetry of the
gas, S0?, and the aerosol, histamine, were quite different. If the changes in
response to repeated S0? exposure were due to localized effects, the histamine
aerosol may have been an inappropriate probe. These authors did suggest that
depletion of mediators or a selective inhibition of SO^-sensitive afferents
might be involved in this phenomenon. For equivalent total exercise time,
Kehrl et al. (1986) observed greater S0?-induced bronchoconstriction with
continuous as compared to intermittent exercise during SO,, exposure. Thus
mediator depletion or selective inhibition of afferents, as well as exercise-
induced release of endogenous bronchodilators (epinephrine) are probably not
related to the attenuation of response with repeated exposure (or repeated
intermittent exercise during exposure). These results suggest that alleviation
of S0?-induced bronchoconstriction is related to events that occur during the
post-exposure/post-exercise recovery period.
Attenuation of bronchoconstriction has been reported for exercise (Stearns
et al., 1981) and hyperpnea of cold, dry air (Bar-Yishay et al. 1983, Wilson et
al., 1982) repeated at short time intervals, suggesting that the attenuation of
S0?-induced bronchoconstriction may be secondary to this decline in response.
4-43
-------
4.5 SUMMARY AND CONCLUSIONS
Studies which have been published in the scientific literature since 1982
support many of the conclusions reached in the earlier criteria document (U.S.
EPA, 1982) and the previous addendum (U.S. EPA, 1982c).
The new studies clearly demonstrate that asthmatics are much more sensi-
tive to S02 as a group.. Nevertheless, it is clear that there is a broad range
of sensitivity to SOp among asthmatics exposed under similar conditions. Re-
cent studies also confirm that normal healthy subjects, even with moderate to
heavy exercise, do not experience effects on pulmonary function due to S0?
exposure in the range of 0 to 2 ppm. The minor exception may be the annoyance
of the unpleasant smell or taste associated with SO,,. The suggestion that
asthmatics are about an order of magnitude more sensitive than normals is thus
confirmed.
There is no longer any question that normally breathing asthmatics per-
forming moderate to heavy exercise will experience S0~-induced bronchocon-
striction when breathing S02 for at least 5 min at concentrations less than 1
ppm. Durations beyond 10 min do not appear to cause substantial worsening of
the effect. The lowest concentration at which bronchoconstriction is clearly
worsened by S02 breathing depends on a variety of factors.
Exposure to less than 0.25 ppm has not evoked group mean changes in
responses. Although some individuals may appear to respond to S0?
concentrations less than 0.25 ppm, the frequency of these responses is not
demonstrably greater than with clean air. Thus individual responses cannot be
relied upon for response estimates, even in the most reactive segment of the
population.
In the S02 concentration range from 0.2 to 0.3 ppm, six chamber exposure
studies were performed with asthmatics performing moderate to heavy exercise.
The evidence that 50,,-induced bronchoconstriction occurred at this concentra-
tion with natural breathing under a range of ambient conditions was equivocal.
Only with oral mouthpiece breathing of dry air (an unusual breathing mode under
exceptional ambient conditions) were small effects observed on a test of ques-
tionable quantitative relevance for criteria development purposes. These find-
ings are in accord with the observation that the most reactive subject in the
Horstman et al. (1986) study had a PCS02 (S02 concentration required to double
SRaw) of 0.28 ppm.
4-44
-------
Several observations of significant group mean changes in SRaw have
recently been reported for asthmatics exposed to 0.4 to 0.6 ppm S02. Most if
not all studies, using moderate to heavy exercise levels (>40 to 50 L/min),
found evidence of bronchoconstriction at 0.5 ppm. At a lower exercise rate,
other studies (e.g., Schachter et al., 1984) did not produce clear evidence of
S02~induced bronchoconstriction at 0.5 ppm SO^. Exposures which included
higher ventilations, mouthpiece breathing, and inspired air with a low water
content resulted in the greatest responses. Mean responses ranged from 45
percent (Roger et al. , 1985) to 280 percent (Bethel et al., 1983b) increase in
SRaw. At concentrations in the range of 0.6 to 1.0 ppm, marked increases in
SRaw are observed following exposure. Recovery is generally complete within
approximately 1 h although the recovery period may be longer for subjects with
the most severe responses.
It is now evident that for S02~induced bronchoconstriction to occur in
asthmatics at concentrations less than 0.75 ppm, the exposure must be accom-
panied by hyperpnea. Ventilations in the range of 40 to 60 L/min have been
most appropriate; such ventilations are beyond the usual oronasal ventilatory
switchpoint. There is no longer any question that oral breathing (especially
via mouthpiece) causes exacerbation of S0?-induced bronchoconstriction. New
studies reinforce the concept that the mode of breathing is an important
determinant of the intensity of S0?-induced bronchoconstriction in the follow-
ing order: oral > oronasal > nasal. A second exacerbating factor implicated
in recent reports is the breathing of dry and/or cold air. It is not clearly
established whether the exacerbation of the S0? effects is due to airway
cooling, airway drying, or some other mechanism.
The new studies do not provide sufficient additional information to estab-
lish whether the intensity of the S0?-induced bronchoconstriction depends upon
the severity of the disease. Across a broad clinical range from "normal" to
moderate asthmatic there is clearly a relationship between the presence of
asthma and sensitivity to SOp. Within the asthmatic population, the relation-
ship of S0? sensitivity to the qualitative clinical severity of asthma has not
been studied systematically. Ethical considerations (i.e., continuation of
appropriate medical treatment) prevent the unmedicated exposure of the "severe"
asthmatic because of his dependence upon drugs for control of his asthma. True
determination of sensitivity requires that the interference with SO^ response
caused by such medication be removed. Because of these mutually exclusive
4-45
-------
requirements, it is unlikely that the "true" S02 sensitivity of severe asthmatics
will be determined. Nevertheless, more severe asthmatics should be studied.
Alternative methods to those used with mild asthmatics, not critically depend-
ant on regular medication, will be required. The studies to date have only
addressed the "mild to moderate" asthmatic.
Consecutive SCL exposures (repeated within 30 min or less) result in a
diminished response compared with the initial exposure. It is apparent that
this refractory period lasts at least 30 min but that normal reactivity returns
within 5 h. The mechanisms and time course of this effect are not clearly
established but refractoriness does not appear to be related to an overall
decrease in bronchomotor responsiveness. These observations suggests that the
effects of S0? on airway resistance and spirometry tend to be short lasting
and do not tend to become worse with continued or repeated exposure. Neverthe-
less, the issue of chronic exposure to SOp in asthmatics has not been addressed.
From the review of studies included in this addendum, it is clear that the
magnitude of response (typically bronchoconstriction) induced by any given S02
concentration was variable among individual asthmatics. Exposures to S02
concentrations of 0.25 ppm or less, which did not induce significant group mean
increases in airway resistance also did not cause symptomatic bronchoconstric-
tion in individual asthmatics. On the other hand, exposures to 0.40 ppm S02 or
greater (combined with moderate to heavy exercise) which induced significant
group mean increases in airway resistance,also caused substantial bronchocon-
striction in some individual asthmatics. This bronchoconstriction was often
associated with wheezing and the perception of respiratory distress. In a few
instances it was necessary to discontinue the exposure and provide medication.
The significance of these observations is that some S02-sensitive asthmatics
are at risk of experiencing clinically significant (i.e., symptomatic) broncho-
constriction requiring termination of activity and/or medical intervention when
exposed to S02 concentrations of 0.40 to 0.50 ppm or greater when this exposure
is accompanied by at least moderate activity.
4-46
-------
CHAPTER 5. EXECUTIVE SUMMARY
In general, studies published in the scientific literature since 1981-82
support many of the conclusions reached in the earlier criteria review (U.S.
EPA, 1982a,c). Some of the key findings emerging from the present evaluation
of the newly available information on health effects associated with exposure
to PM and SO are summarized here.
/\
5.1 RESPIRATORY TRACT DEPOSITION AND FATE
Studies published since preparation of the earlier criteria document (U.S.
EPA, 1982a) and the previous addendum (U.S. EPA, 1982c) support the conclusions
reached at that time and provide clarification of several issues. In light of
previously available data, new literature was reviewed with a focus towards (1)
the thoracic deposition and clearance of large particles, (2) assessment of
deposition during oronasal breathing, (3) deposition in possibly susceptible
subpopulations, such as children, and (4) information that would relate the
data to refinement or interpretation of ancillary issues, such as inter- and
intrasubject variability in deposition, deposition of monodisperse versus
polydisperse aerosols, etc.
The thoracic deposition of particles >10 |jm D and their distribution in
36
the TB and P regions has been studied by a number of investigators (Svartengren,
1986; Heyder, 1986; Emmett et al., 1982). Depending upon the breathing regimen
used, TB deposition ranged from 0.14 to 0.36 for 10-|jm D particles, while the
ac
range for 12-|jm Dae particles was 0.09 to 0.27. For particles 16.4 (jm Dae, a
maximally deep inhalation pattern resulted in TB deposition of Or12. While the
magnitude of deposition in various regions depends heavily upon minute ventila-
tion, there is, in general, a gradual decline in thoracic deposition for large
particle sizes, and there can be significant deposition of particles greater
than 10 |jm D , particularly for individuals who habitually breathe through
36
their mouth. Thus, the deposition experiments wherein subjects inhale through
5-1
-------
a mouthpiece are relevant to examining the potential of particles to penetrate
to the lower respiratory tract and pose a potentially increased risk. In-
creased risk may be due to increased localized dose or to the exceedingly long
half-times for clearance of larger particles (Gerrity et al., 1983).
Although experimental data are not currently available for deposition of
particles in the lungs of children, some trends are evident from the modeling
results of Phalen et al. (1985). Phalen and co-workers made morphometric mea-
surements in replica lung casts of people aged 11 days to 21 years and modeled
deposition during inspiration as a function of activity level. They found that,
in general, increasing age is associated with decreasing particulate tracheo-
bronchial deposition efficiency. However, very high flow rates and large parti-
culate sizes do not exhibit consistent age-dependent differences. Since minute
ventilation at a given state of activity is approximately linearly related to
body mass, children receive a higher TB dose of particles than do adults and
would appear to be at a greater risk, other factors (i.e., mucociliary clear-
ance, particulate losses in the head, tissue sensitivity, etc.) being equal.
5.2 SUMMARY OF EPIDEMIOLOGIC FINDINGS ON HEALTH EFFECTS ASSOCIATED WITH
EXPOSURE TO AIRBORNE PARTICLES AND SOV
A
Newly available reanalyses of data relating mortality in London to short-
term (24-h) exposures to PM (measured as smoke) and S0? were evaluated and
their results compared with earlier findings and conclusions discussed in U.S.
EPA (1982a). Varying strengths and weaknesses were evident in relation to the
different individual reanalyses (Mazumdar et al. , 1982; Ostro, 1984; Shumway
et al., 1983; Schwartz and Marcus, 1986) evaluated and certain issues remain un-
resolved. Nevertheless, the following conclusions appear to be warranted based
on the earlier criteria review (U.S. EPA, 1982a) and present evaluation of newly
available analyses of the London mortality experience: (1) markedly increased
mortality occurred, mainly among the elderly and chronically ill, in association
3
with BS and S0? concentrations above 1000 (jg/m , especially during episodes when
such pollutant elevations occurred for several consecutive days; (2) during such
episodes coincident high humidity (fog) was also likely important, possibly in
providing conditions leading to formation of H^SO, or other acidic aerosols; (3)
increased risk of mortality is associated with exposure to BS and S09 levels in
3
the range of 500 to 1000 (jg/m , for S0? most clearly at concentrations in excess
5-2
-------
of ~700 to 750 pg/m ; and (4) convincing evidence indicates that relatfveTy
small but statistically significant increases in the risk of mortality exist
o
at BS (but not S02 levels below 500 ug/m ), with no indications of any specific
threshold level having yet been demonstrated at lower concentrations of BS
o
(e.g., at 150 ug/m ). However, precise quantitative specification of the lower
PM levels associated with mortality is not possible, nor can one rule out poten-
tial contributions of other possible confounding variables at these low PM
levels.
In addition to the reanalyses of London mortality data, reanalyses of mor-
tality data from New York City in relation to air pollution reported by Ozkaynak
and Spengler (1985) were evaluated. Time-series analyses were carried out on
a subset of New York City data included in a prior analysis by Schimmel (1978)
which was critiqued during the earlier criteria review (U.S. EPA, 1982a). The
reanalyses by Ozkaynak and Spengler (1985) evaluated 14 years (1963-76) of daily
measurements of mortality (the sum of heart, other circulatory, respiratory, and
cancer mortality), COM, S0?, and temperature. In summary, the newly available
reanalyses of New York City data raise possibilities that, with additional work,
further insights may emerge regarding mortality-air pollution relationships in
a large U.S. urban area. However, the interim results reported thus far do not
now permit definitive determination of their usefulness for defining exposure-
effect relationships, given the above-noted types of caveats and limitations.
Similarly, it is presently difficult to accept findings reported in another
new study (Hazakis et al. , 1986) of mortality associated with relatively low
levels of SOp pollution in Athens, given questions regarding representativeness
of the monitoring data and the statistical soundness o'f using deviations of mor-
tality from an earlier baseline relatively distant in time. Lastly, a newly re-
ported study (Mazumdar and Sussman, 1983) of mortality-air pollution relation-
ships in Pittsburgh (Allegheny County, PA) was evaluated as having utilized
inadequate exposure characterization and the results contain sufficient inter-
nal inconsistencies, so that the analyses are not useful for delineating mor-
tality relationships with either SO^ or PM.
Of the newly-reported analyses of short-term PM/SOx exposure-morbidity
relationships discussed in this Addendum, the Dockery et al. (1982) study
provides the best-substantiated and most readily interpretable results. Those
results, specifically, point toward decrements in lung function occurring in
5-3
-------
association with acute, short-term increases in PM and S02 air pollution. The
small, reversible decrements appear to persist for 2-3 wks after episodic expo-
sures to these pollutants across a wide range of concentrations, with no clear
delineation of threshold yet being evident. In some study periods effects may
have been due to 24-hr TSP and S02 levels ranging up to 220-420 and 280-460
[jg/m , respectively. Notably larger decrements in lung function were discern-
able for a subset of children (responders) than for others. The precise medical
significance of the observed decrements per s_e or any consequent long-term
sequelae remain to be determined. The nature and magnitude of lung function
decrements found by Dockery et al. (1982), it should be noted, are also consis-
tent with: (1) the recently reported findings of Dassen et al (1986) of pulmo-
nary function decrements of approximately the same magnitude over similar time
periods after episodic exposure of Dutch children to 24-hr TSP and S00 levels
3
in the 200-250 |jg/m range, (2) observations of Stebbings and Fogleman (1979)
of gradual recovery in lung function of children during seven days following a
high PM episode in Pittsburgh, PA (max 1-hr TSP estimated at 700 ng/m3); and (3)
a report by Saric et al. (1981) of 5 percent average declines in FEV-,^ Q being
associated with high S02 days (89-235 pg/m3).
In regard to evaluation of long-term exposure effects, the 1982 U.S. EPA
criteria document (1982a) noted that certain large-scale "macroepidemiological"
(or "ecologic") studies have attracted attention on the basis of reported demon-
strations of associations between mortality .and various indices, of air pollu-
tion, e.g., PM or SO levels. U.S. EPA (1982a) also noted that various criti-
}\
cisms of then-available ecologic studies made it impossible to ascertain which
findings may be more valid than others. Thus, although many of the studies qua-
litatively suggested positive associations between mortality and chronic expo-
sure to certain air pollutants in the United States, many key issues remained
unresolved concerning reported associations and whether they were causal or
not.
Since preparation of the earlier Criteria Document (U.S. EPA, 1982a)
additional ecological analyses have been reported regarding efforts to assess
relationships between mortality and long-term exposure to particulate matter
and other air pollutants. For example, Lipfert (1984) conducted a series of
cross-sectional multiple regression analyses of 1969 and 1970 mortality rates
for up to 112 U.S. SMSA's, using the same basic data set as Lave and Seskin
5-4
-------
(1977) for 1969 and taking into account various demographic, environmental and
lifestyle variables (e.g., socioeconomic status and smoking). Also, the
Lipfert (1984) reanalysis included several additional independent variables:
diet; drinking water variables; use of residential heating fuels; migration;
and SMSA growth. New dependent variables included age-specific mortality rates
with their accompanying sex-specific age variables. Both linear and several
nonlinear (e.g., quadratic or linear splines testing for possible threshold
model specifications) were evaluated.
It became quite evident from the results obtained that the air pollution
regression results for the U.S. data sets analyzed by Lipfert (1984) are
extremely sensitive to variations in the inclusion/exclusion of specific
observations (for central city versus SMSA's or different subsets of locations)
or additional explanatory variables beyond those used in the earlier Lave and
Seskin analyses. The results are also highly dependent upon the particular
model specifications used, i.e. air pollution coefficients vary in strength of
association with total or age-/sex-specific mortality depending upon the form
of the specification and the range of explanatory variables included in the
analyses. Lipfert1s overall conclusion was that the sulfate regression coeffi-
cients are not credible and, since sulfate and TSP interact with each other in
these regressions, caution is warranted for TSP coefficients as well.
Ozkaynak and Spengler (1985) have also newly described results from
ongoing attempts to improve upon previous analyses of mortality and morbidity
effects of air pollution in the United States. Ozkaynak and Spengler (1985)
present principal findings from a cross-sectional analysis of the 1980 U.S.
vital statistics and available air pollution data bases for sulfates, and fine,
inhalable and total suspended particles. In these analyses, using multiple
regression methods, the association between various particle measures and 1980
total mortality were estimated for 98 and 38 SMSA subsets by incorporating
recent information on particle size relationships and a set of socioeconomic
variables to control for potential confounding. Issues of model misspecifica-
tion and spatial autocorrelation of the residuals were also investigated.
The Ozkaynak and Spengler (1985) results for 1980 U.S. mortality provide
an interesting overall contrast to the findings of Lipfert (1984) for 1969-70
U.S. mortality data. Whereas Lipfert found TSP coefficients to be most con-
sistently statistically significant (although varying widely depending upon
model specifications, explanatory variables included, etc.), Ozkaynak and
5-5
-------
Spengler found particle mass measures including coarse particles (TSP, IP)
often to be nonsignificant predictors of total mortality. Also, whereas
Lipfert found the sulfate coefficients to be even more unstable than the TSP
associations with mortality (and questioned the credibility of the sulfate
coefficients), Ozkaynak and Spengler found that particle exposure measures
related to the respirable or toxic fraction of the aerosols (e.g., FP or sul-
fates) to be most consistently and significantly associated with annual cross-
sectional mortality rates. It might be tempting to hypothesize that changes in
air quality or other factors from the earlier data sets (for 1969-70) analyzed
by Lipfert (1984) to the later data (for 1980) analyzed by Ozkaynak and Spengler
(1985) and Ozkaynak et a "I. (1986) may at least partly explain their contrast-
ing results, but there is at present no basis by which to determine if this is
the case or which set of findings may or may not most accurately characterize
associations between mortality and chronic PM or SO exposures in the United
/\
States. Thus conclusions stated in U.S EPA (1982a) concerning ecologic
analyses still largely apply in regard to mortality-PM/SO relationships.
/\
The present Addendum also evaluated a growing body of new literature on
morbidity effects associated with chronic exposures to airborne particles and
sulfur oxides. In summary, of the numerous new studies published on morbidity
effects associated with long-term exposures, to PM or SO , only a few may
.A.
provide potentially useful results by which to derive quantitative conclusions
concerning exposure-effect relationships for the subject pollutants. A study
by Ware et al. (1986), for example, provides evidence of respiratory symptoms
in children being associated with particulate matter exposures in contemporary
U.S. cities without evident threshold across a range of annual-average TSP levels
3 3
of ~30 to 150 ug/m , with most marked effects notable in the 60-150 ug/m range
in comparison to lower TSP levels. The increase in symptoms appears to occur
without concomitant decrements in lung function among the same children. The
medical significance of the observed increases in symptoms unaccompanied by
decrements in lung function remains to be fully evaluated but is of likely
health concern. Caution is warranted, however, in using these findings for risk
assessment purposes in view of the lack of significant associations for the same
variables when assessed from data within individual cities included in the Ware
et al. (1986) study. The findings derived from another series of studies (Ostro,
5-6
-------
1983; Hausman et al. , 1984; Ostro, 1987) are qualitatively indicative of mor-
bidity effects in adults being associated with PM exposures over time within
U.S. cities, and these results tend to support the plausibility that the asso-
ciations found by Ware et al. (1986) reflect actual morbidity effects in
children due to contemporaneous U.S. PM exposures.
Other new American studies provide evidence for: (1) increased respira-
tory symptoms among young adults in association with annual-average S0? levels
o ^
of ~115 pg/m (Chapman et al., 1985); and (2) increased prevalence of cough in
children (but not lung function changes) being associated with intermittent
exposures to mean peak 3-hr S09 levels of ~1.0 ppm or annual average S0? levels
3
of ~103 pg/m (Dodge et al. , 1985). It is difficult to distinguish as to
whether the effects found in these two studies are due to repeated high-level
S0? peak exposures or to chronic exposures to lower concentrations of SCL or
its transformation products.
Results from one European study (PAARC, 1982a,b) also tend to suggest the
likelihood of lower respiratory disease symptoms and decrements in lung function
in adults (both male and female) being associated with anniial average S0« levels
3
ranging without evident threshold from about 25 to 130 pg/m . In addition that
study suggests that upper respiratory disease and lung function decrements in
children may also be associated with annual-average SO™ levels across the above
range. However, the associations between morbidity effects and S02 reported
by PAARC (1982a,b) cannot be fully accepted due to: (1) internal inconsis-
tencies between results obtained with S02 exposure estimates based on one type
of measurement method versus those based on another S02 measurement technique,
and (2) the lack of adequate control for seasonal effects and parental smoking
in certain analyses for childrens' data that yielded significant health
effects associations.
5.3 SUMMARY OF CONTROLLED HUMAN EXPOSURE STUDIES OF SULFUR DIOXIDE HEALTH
EFFECTS
The new studies evaluated in the present addendum (Chapter 4) clearly
demonstrate that asthmatics are much more sensitive to S02 as a group. Never-
theless, it is clear that there is a broad range of sensitivity to S02 among
asthmatics exposed under similar conditions. Recent studies also confirm that
normal healthy subjects, even with moderate to heavy exercise, do not experience
5-7
-------
effects on pulmonary function due to S02 exposure in the range of 0 to 2 ppm.
The minor exception may be the annoyance of the unpleasant smell or taste asso-
ciated with S02. The suggestion that mild "compensated" asthmatics are about
an order of magnitude more sensitive than normals is thus confirmed. There is
not enough information on SO™ response in moderate to severe asthmatics to esti-
mate their sensitivity.
There is no longer any question that normally breathing asthmatics per-
forming moderate to heavy exercise will experience SOp-induced bronchocon-
striction when breathing S02 for at least 5 min at concentrations less than 1
ppm. Durations beyond 10 min do not appear to cause substantial worsening of
the effect. The lowest concentration at which bronchoconstriction is clearly
exacerbated by SCL breathing depends on a variety of factors.
Exposure to less than 0.25 ppm has not evoked group mean changes in
responses. Although some individuals may appear to respond to S0? concentra-
tions less than 0.25 ppm, the frequency of these responses is not demonstrably
greater than with clean air. Thus individual responses cannot be relied upon
for response estimates, even in the most reactive segment of the population.
In the S02 concentration range from 0.2 to 0.3 ppm, six chamber exposure
studies were performed with asthmatics performing moderate to heavy exercise.
The evidence that SO^-induced bronchoconstriction occurred at this concentra-
tion with natural breathing under a range of ambient conditions was equivocal.
Only with oral mouthpiece breathing of dry air (an unusual breathing mode under
exceptional ambient conditions) were small effects observed on a test of ques-
tionable quantitative relevance for criteria development purposes. These find-
ings are in accord with the observation that the most reactive subject in the
Horstman et al. (1986) study had a PCS02 (S02 concentration required to double
SRaw) of 0.28 ppm.
Several observations of significant group mean changes in SRaw have
recently been reported for asthmatics exposed to 0.4 to 0.6 ppm S02. Most if
not all studies, using moderate to heavy exercise levels (>40 to 50 L/min),
found evidence of bronchoconstriction at 0.5 ppm. At a lower exercise rate,
other studies (e.g., Schachter et al. , 1984) did not produce clear evidence of
SOp-induced bronchoconstriction at 0.5 ppm S0?. Exposures which included
higher ventilations, mouthpiece breathing, and inspired air with a low water
content resulted in the greatest responses. Mean responses ranged from 45
percent (Roger et al., 1985) to 280 percent (Bethel et al., 1983b) increase in
5-8
-------
SRaw. At concentrations in the range of 0.6 to 1.0 ppm, marked increases in
SRaw are observed following exposure. Recovery is generally complete within
approximately 1 h although the recovery period may be longer for subjects with
the most severe responses.
It is now evident that for S02~induced bronchoconstriction to occur in
asthmatics at concentrations less than 0.75 ppm, the exposure must be accom-
panied by hyperpnea. Ventilations in the range of 40 to 60 L/min have been
most,appropriate; such ventilations are beyond the usual oronasal ventilatory
switchpoint. There is no longer any question that oral breathing (especially
via mouthpiece) causes exacerbation of S02-induced bronchoconstriction. New
studies reinforce the concept that the mode of breathing is an important
determinant of the intensity of S02-induced bronchoconstriction in the follow-
ing order: oral > oronasal > nasal. A second exacerbating factor strongly
implicated in recent reports is the breathing of dry and/or cold air with S02-
It is not established whether the reduced water content, the reduced tempera-
ture, or both is responsible for this effect.
The new studies do not provide sufficient additional information to estab-
lish whether the intensity of the S02-induced bronchoconstriction depends upon
the severity of the disease. Across a broad clinical range from "normal" to
moderate asthmatic there is clearly a relationship between the presence of
asthma and sensitivity to S09. Within the asthmatic population, the relation-
ship of S0? sensitivity to the qualitative clinical severity of asthma has not
been studied systematically. Ethical considerations (i.e., continuation of
appropriate medical treatment) prevent the unmedicated exposure of the "severe"
asthmatic because of his dependence upon drugs for control of his asthma. True
determination of sensitivity requires that the interference with S02 response
caused by such medication be removed. Because of these mutually exclusive
requirements, it is unlikely that the "true" S02 sensitivity of severe
asthmatics will be determined. Nevertheless, more severe asthmatics should be
studied. Alternative methods to those used with mild asthmatics, not criti-
cally dependant on regular medication, will be required. The studies to date
have only addressed the "mild to moderate" asthmatic.
Consecutive S02 exposures (repeated within 30 min or less) result in a
diminished response compared with the initial exposure. It is apparent that
this refractory period lasts at least 30 min but that normal reactivity returns
within 5 h. The mechanisms and time course of this effect are not clearly
5-9
-------
established but refractoriness does not appear to be related to an overall
decrease in bronchomotor responsiveness.
From the review of studies included in this addendum, it is clear that the
magnitude of response (typically bronchoconstriction) induced by any given S0?
concentration was variable among individual asthmatics. Exposures to S09
concentrations of 0.25 ppm or less, which did not induce significant group mean
increases in airway resistance also did not cause symptomatic bronchoconstric-
tion in individual asthmatics. On the other hand, exposures to 0.40 ppm S0? or
greater (combined with moderate to heavy exercise) which induced significant
group mean increases in airway resistance, also caused substantial broncho-
constriction in some individual asthmatics. This bronchoconstriction was often
associated with wheezing and the perception of respiratory distress. In a few
instances it was necessary to discontinue the exposure and provide medication.
The significance of these observations is that some S0?-sensit.ive asthmatics are
at risk of experiencing clinically significant (i.e., symptomatic) bronchocon-
striction requiring termination of activity and/or medical intervention when
exposed to S02 concentrations of 0.40 to 0.50 ppm or greater when this exposure
is accompanied by at least moderate activity.
5-10
-------
6. REFERENCES
Albert, R. E.; Lippmann, M.; Peterson, H. T., Jr.; Berger, J.; Sanborn, K. ;
Bohning, D. (1973) Bronchial deposition and clearance of aerosols. Arch.
Intern. Med. 131: 115-127.
Amdur, M. 0. (1985) When one plus zero is more than one. Am. Ind. Hyg. Assoc.
J. 46: 467-475.
Andersen, I.; Lundqvist, G. R.; Jensen, P. L.; Proctor, D. F. (1974) Human
response to controlled levels of sulfur dioxide. Arch. Environ. Health 28:
31-39.
Anderson, S. D. (1985) Issues in exercise-induced asthma. J. Allergy Clin.
Immunol. 76: 763-772.
Anderson, D. 0.; Ferris, B. G. , Jr.; Zinkmantel, R. (1964) Levels of air
pollution and respiratory disease in Berlin, New Hampshire. Am. Rev.
Respir. Dis. 90: 877-887.
Anderson, D. 0.; Ferris, B. G. , Jr. (1965) Air pollution levels and chronic
respiratory disease. Arch. Environ. Health 10: 307-311.
Anderson, K. L. ; Rutenfranz, J.; Seliger, V.; Ilmarinen, J.; Berndt, I.;
Kylian, H. ; Ruppel, M. (1984) The growth of lung volumes affected by
physical performance capacity in boys and girls during childhood and
adolescence. Eur. J. Appl. Physiol. 52: 380-384.
Asmundsson, T. ; Kilburn, K. H. (1970) Mucociliary clearance rates at various
levels in dog lungs. Am. Rev. Respir. Dis. 102: 388-397.
Bailey, D. L. R.; Clayton, P. (1980) The measurement of suspended particulate
and carbon concentrations in the atmosphere using standard smoke shade
methods. Stevenage, Hertfordshire, United Kingdom: Warren Spring Laborato-
ry; report no. LR 325 (AP).
Bailey, M. R.; Fry, F. A.; James, A. C. (1982) The long-term clearance kinetics
of insoluble particles from the human lung. Ann. Occup. Hyg. 26: 273-290.
Ball, D. J. ; Hume, R. (1977) The relative importance of vehicular and domestic
emissions of dark smoke in greater London in the mid-1970s, the signifi-
cance of smoke shade measurements, and an explanation of the relationship
of smoke shade to gravimetric measurements of particulate. Atmos. Environ.
11: 1065-1073.
Bar-Yishay, E.; Ben-Dov, I.; Godfrey, S. (1983) Refractory period after
hyperventilation-induced asthma. Am. Rev. Respir. Dis. 127: 572-574.
Bates, D. V.; Sizto, R. (1983) Relationship between air pollutant levels and
hospital admissions in Southern Ontario. Can. J. Public Health 74:
117-122.
6-1
-------
Bates, D. V.; Sizto, R. (1986) A study of hospital admissions and air pollu-
tants in southern Ontario. In: Lee, S. D. ; Schneider, T. ; Grant, L. D.;
Verkerk, P., eds. Aerosols: research, risk assessment, control strategies,
proceedings of the second U.S.-Dutch international symposium; May 1985;
Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 767-777.
Bedi, J. F.; Folinsbee, L. J.; Horvath, S. M. (1984) Pulmonary function effects
of 1.0 and 2.0 ppm sulfur dioxide exposure in active young male nonsmokers.
J. Air Pollut. Control Assoc. 34: 1117-1121.
Bethel, R. A.; Epstein, J.; Sheppard, D.; Nadel, J. A.; Boushey, H. A. (1983a)
Sulfur dioxide-induced bronchoconstriction in freely breathing, exercis-
ing, asthmatic subjects. Am. Rev. Respir. Dis. 128: 987-990.
Bethel, R. A.; Erie, D. J.; Epstein, J.; Sheppard, D.; Nadel, J. A.; Boushey,
H. A. (1983b) Effect of exercise rate and route of inhalation on sulfur-
dioxide-induced bronchoconstriction in asthmatic subjects. Am. Rev.
Respir. Dis. 128: 592-596.
Bethel, R. A.; Sheppard, D.; Epstein, J.; Tarn, E. ; Nadel, J. A.; Boushey, H. A.
(1984) Interaction of sulfur dioxide and dry cold air in causing broncho-
constriction in asthmatic subjects. J. Appl. Physio!. 57: 419-423.
Bethel, R. A.; Sheppard, D.; Geffroy, B.; Tarn, E.; Nadel, J. A.; Boushey, H. A.
(1985) Effect of 0.25 ppm sulfur dioxide on airway resistance in freely
breathing, heavily exercising, asthmatic subjects. Am. Rev. Respir. Dis.
131: 659-661.
Binder, R. E.; Mitchell, C. A.;
the indoor environment in
31: 277-279.
Hosein, H. R.; Bouhuys, A. (1976) Importance of
air pollution exposure. Arch. Environ. Health
Bohning, D. E.; Atkins, H. L.; Cohn, S. H. (1982) Long-term particle clearance
in man: normal and impaired. Ann. Occup. Hyg. 26: 259-271.
Bouhuys, A.; Beck, G. J.; Schoenberg, J. B. (1978) Do present levels of air
pollution outdoors affect respiratory health? Nature (London) 276:
466-471.
Bowes, S. M. (1986) Particle deposition in the mouth during oro-nasal breathing
[dissertation]. Baltimore, MA: John Hopkins University.
Bowes, S. M.; Swift, D. L. (1985) Oral deposition of monodisperse aerosols in
humans during natural oral breathing. Abst. of poster presentation,
inhaled particles VI, Cambridge, England. (Publication status of data
contained in poster presentation is unknown at this time).
Bowes, S. M.; Swift, D. L. (1986) Aerosol deposition in the human oral passage
during natural oronasal breathing. In: Aerosols': formation and reactivity,
proceedings of the 2nd international aerosol conference; Berlin, West
Germany. London, United Kingdom: Pergamon Journals Ltd.; pp. 197-199.
Brain, J. D.; Valberg, P. A. (1974) Models of lung retention based on ICRP task
group report. Arch. Environ. Health 28: 1-11.
6-2
-------
Camner, P.; Jarstrand, C. ; Philipson, K. (1973) Tracheobronchia] clearance in
patients with influenza. Am. Rev. Respir. Dis. 108: 131-135.
Chan, T. L.; Lippmann, M. (1980) Experimental measurements and empirical
modelling of the regional deposition of inhaled particles in humans. Am.
Ind. Hyg. Assoc. J. 41: 399-409.
Chapman, R. S.; Calafiore, D. C.; Hasselblad, V. (1985) Prevalence of persis-
tent cough and phlegm in young adults in relation to long-term ambient
sulfur oxide exposure. Am. Rev. Respir. Dis. 132: 261-267.
Chappie, M.; Lave, L. (1981) The health effects of air pollution: a reanalysis.
J. Urban Econ.: in press.
Charlson, R. J. ; Waggoner, A. P.; Thielke, J. F. (1978) Visibility protection
for class I areas: the technical basis. Washington, DC: Council on Envi-
ronmental Quality; report no. CEQ517714498. Available from: NTIS, Spring-
field, VA; PB-288842.
Chatham, M. ; Bleecker, E. R. ; Smith, P. L. ; Rosenthal, R. R. ; Mason, P.;
Norman, P. S. (1982) A comparison of histamine, methacholine, and exercise
airway reactivity in normal and asthmatics subjects. Am. Rev. Respir. Dis.
126: 235-240.
Chinn, S. ; Florey, C. du V.; Baldwin, I. G.; Gorgol, M. (1981) The relation of
mortality in England and Wales 1969-73 to measurements of air pollution.
J. Epidemic!. Commun. Health 35: 174-179.
Cole, P. (1982) Upper respiratory airflow. In: Proctor, D. F. ; Andersen, I. The
nose: upper airway physiology and the atmospheric environment. New York,
NY: Elsevier Biomedical Press; pp. 163-189.
Commins, B. T.; Waller, R. E. (1967) Observations from a ten-year-study of
pollution at a site in the city of London. Atmos. Environ. 1: 49-68.
Crawford, D. J. (1982) Identifying critical human subpopulations by age groups:
radioactivity and the lung. Phys. Med. Biol. 27: 539-552.
Crocker, T. G. ; Schulze, W. D. ; Ben-David, S. ; Kneese, A. V. (1979) Methods
development for assessing air pollution control benefits: volume I,
experiments in the economics of air pollution epidemiology. Washington,
DC: U. S. Environmental Protection Agency, Office of Health and Ecological
Effects; EPA report no. EPA-600/5-79-001a. Available from: NTIS, Spring-
field, VA; PB-293615.
D'Alfonso, D. A. (1980) The limiting factors of nasal respiration [Ph.D.
thesis]. Santa Barbara, CA: University of California.
Dassen, W. ; Brunekreef, B.; Hoek, G.; Hofschreuder, P.; Staatsen, B.; de Groot,
H. ; Schouten, E. ; Biersteker, K. (1986) Decline in children's pulmonary
function during an air pollution episode. J. Air Pollut. Control Assoc.
36: 1223-1227.
6-3
-------
de Swiniarski, R.; Mataame, M.; Tanche, M. (1982) Plethysmography study and
pulmonary function in well-trained adolescents = Etude plethysmographique
et fpnction pulmonaire chez des adolescents bien entrafnes. Bull. Eur.
Physiopathol. Respir. 18: 39-49.
Deal, E. C., Jr.; McFadden, E. R., Jr.; Ingram, R. H., Jr.; Strauss, R. H. ;
Jaeger, J. J. (1979a) Role of respiratory heat exchange in production of
exercise-induced asthma. J. Appl. Physio!. Respir. Environ. Exercise
Physio!. 46: 467-475.
Deal, E. C., Jr.; McFadden, E. R., Jr.; Ingram, R. H., Jr.; Jaegner, J. J.
(1979b) Hyperpnea and heat flux: initial reaction sequence in exercise-
induced asthma. J. Appl. Physio!.: Respir. Environ. Exercise Physiol. 46:
476-483.
Diu, C. K.; Yu, C. P. (1983) Respiratory tract deposition of polydisperse
aerosols in humans. Am. Ind. Hyg. Assoc. J. 44: 62-65.
Dockery, D. W.; Ware, J. H. ; Ferris, B. G., Jr.; Speizer, F. E.; Cook, N. R.;
Herman, S. M. (1982) Change in pulmonary function in children associated
with air pollution episodes. J. Air Pollut. Control Assoc. 32: 937-942.
Dodge, R. (1983) The respiratory health and lung function of Anglo-American
children in a smelter town. Am. Rev. Respir. Dis. 127: 158-161.
Dodge, R. R.;_ Burrows, B. (1980) The prevalence and incidence of asthma and
asthma-like symptoms in a general population sample. Am. Rev. Respir Dis
122: 567-575.
Dodge, R.; Solomon, P.; Moyers, J.; Hayes, C. (1985) A longitudinal study; of
children exposed to sulfur oxides. Am. J. Epidemic!. 121: 720-736.
Edmunds, A. T.; Tooley, M.; Godfrey, S. (1978) The refractory period after
exercise-induced asthma: Its duration and relation to the severity of
exercise. Am. Rev. Respir. Dis. 117: 247-254.
Emmett, P. C.; Aitken, R. J. ; Hannan, W. J. (1982) Measurements of the total
and regional deposition of inhaled particles in the human respiratory
tract. J. Aerosol Sci. 13: 549-560.
Federal Register. (1984a) Proposed revisions to the national ambient air
quality standards for particulate matter: proposed rule. F. R. (March 20)
49: 10408-10437.
Federal Register. (1984b) Proposed revisions to the national ambient air
quality standards for particulate matter: public hearing announcement. F.
R. (April 2) 49: 13059.
Ferris, B. G.,•Jr.; Anderson, D. 0. (1962) The prevalence of chronic respira-
tory disease in a New Hampshire town. Am. Rev. Respir. Dis. 86: 165-185.
Ferris, B. G., Jr.; Burgess, W. A.; Worcester, J. (1967) Prevalence of chronic
respiratory disease in a pulp mill and a paper mill in the United States.
Br. J. Ind. Med. 24: 26-37.
6-4
-------
Ferris, B. G., Jr.; Higgins, I. T. T.; Higgins, M. W.; Peters, J. M. ; van
Ganse, W. F. ; Goldman, M. D. (1971) Chronic nonspecific respiratory
disease, Berlin, New Hampshire, 1961-1967: a cross-sectional study. Am.
Rev. Respir. Dis. 104: 232-244.
Ferris, B. G. , Jr.; Mahoney, J. R. ; Patterson, R. M.;. First, M. W. (1973) Air
quality, Berlin, New Hampshire, March 1966-December 1967. Am. Rev. Respir.
Dis. 108: 77-84.
Ferris, B. G.. Jr. ; Chen, H.; Pulep, S. ; Murphy, R. L. H., Jr. (1976) Chronic
nonspecific respiratory disease in Berlin, New Hampshire, 1967 to 1973; a
further follow-up study. Am. Rev. Respir. Dis. 113: 475-485.
Ferris, B. G. , Jr.; Ware, J. H. ; Spengler, J. D.; Dockery, D. W.; Speizer, F.
E. (1986) The Harvard six-cities study. In: Lee, S. D.; Schneider, T.;
Grant, L. D.; Verkerk, P. J., eds. Aerosols: research, risk assessment and
control strategies, proceedings of the second U.S.-Dutch international
symposium; May 1985; Williamsburg, VA. Chelsea, MI: Lewis Publishers,
Inc.; pp. 721-730.
Florey, C. du V.; Swan, A. V.; van der Linde, R.; Holland, W. W.; Berlin, A.;
Di Ferrante, E. , eds. (1983) Report on the EC epidemiological survey on
the relationship between air pollution and respiratory health in primary
school children. Brussels, Belgium: Commission of the European Communi-
ties, Environmental Research Programme.
Folinsbee, L. J. ; Bedi, J. F.; Horvath, S. M. (1985) Pulmonary response to
threshold levels of sulfur dioxide (1.0 ppm) and ozone (0.3 ppm). J. Appl.
Physiol. 58: 1783-1787.
Forsyth, R. D.; Cole, P.; Shephard, R. J. (1983) Exercise and nasal patency. J.
Appl. Physiol.: Respir. Environ. Exercise Physiol. 55: 860-865.
Foster, W. M.; Langenback, E.; Bergofsky, E. H. (1980) Measurement of trachea!
and bronchial mucus velocities in man: relation to lung clearance. J.
Appl. Physiol.: Respir. Environ. Excercise Physiol. 48: 965-971.
Garrard, C. S.;
of aerosol
184-193.
Gerrity, T. R.; Schreiner, J, F.; Yeates, D. B. (1981) Analysis
deposition in the healthy human lung. Arch. Environ. Health 36:
Garrard, C. S. ; Levandowski
(1985) The effects of
mucous transport. Arch.
, R. A.; Gerrity, T. R.; Yeates, D. B. ; Klein, E.
acute respiratory virus infection upon trachea!
Environ. Health 40: 322-325.
Gerking, S. ; Schultze, W. (1981) What do we know about benefits of reduced
morality from air pollution control? Am. Econ. Rev. 71: 228-234.
Gerrity, T. R.; Lee, P. S.; Hass, F. J.; Marinelli, A.; Werner, P.; Lourenco,
R. V. (1979) Calculated deposition of inhaled particles in the airway
generations of normal subjects. J. Appl. Physiol.: Respir. Environ.
Exercise Physiol. 47: 867-873.
6-5
-------
Gernty, T. R.; Garrard, C. S.; Yeates, D. B. (1983) A mathematical model of
particle retention in the air-spaces of human lungs. Br. J. Ind. Med. 40:
121-130.
Gillespie, J. R. (1980)
studies on beagles
pollutants.
W. S. , eds
Cincinnati,
Criteria
115-153.
Review of the cardiovascular and pulmonary function
exposed for 68 months to auto exhaust and other air
In: Stara, J. F. ; Dungworth, D. L.; Orthoefer, J. G.; Tyler,
Long-term effects of air pollutants: in canine species.
OH: U. S. Environmental Protection Agency, Environmental
and Assessment
Available from:
Office; EPA report no
NTIS, Springfield, VA;
EPA-600/8-80-014;
PB81-144875.
pp.
Goldstein, I. F.; Weinstein, A. L. (1986) Air pollution and asthma: effects of
exposures to short-term sulfur dioxide peaks. Environ. Res. 40: 332-345.
Hackney, J. D.; Linn, W. S.; Bailey, R. M.; Spier, C. E. ; Valencia, L. M.
(1984) Time course of exercise-induced bronchoconstriction in asthmatics
exposed to sulfur dioxide. Environ. Res. 34: 321-327.
Hatzakis, A.; Katsouyanni, K.; Kalandidi, A.; Day, N. ; Trichopoulos, D. (1986)
Short-term effects of air pollution on mortality in Athens. Int. J.
Epidemic!. 15: 73-81.
Hausman, J. A.; Ostro, B. D.; Wise, D. A. (1984) Air pollution and lost work.
Cambridge, MA: National Bureau of Economic Research; NBER workinq paper
no. 1263.
Hazleton Laboratories America, Incorporated. (1975) Effects of inhaling sulfur
dioxide, sulfuric acid, and fly ash. Palo Alto, CA: Electric Power Re-
search Institute; report no. EPRI-74. Available from: NTIS, Springfield,
VA; PB-241788.
Heyder, J. (1986)
Aerosol Sci.:
Studies of particle
in press.
deposition and clearance in humans. J.
Heyder,"J.; Gebhart, J.; Stahlhofen, W.; Stuck, B. (1982) Biological variabil-
ity of particle deposition in the human respiratory tract during con-
trolled and spontaneous mouth-breathing. In: Walton, W. H.; Critchlow, A.;
Coppock, S. M. , eds. Inhaled particles V: proceedings of an international
symposium organized by the British Occupational Hygiene Society; September
1980; Cardiff, United Kingdom. Ann. Occup. Hyg. 26: 137-147.
Hiller, F. C. ; Mazumder, M. K.; Wilson, J. D.; McLeod, P. C.; Bone, R. C.
(1982) Human respiratory tract deposition using multimodal aerosols. J.
Aerosol Sci. 13: 337-343.
Hofmann, H. (1982a) Dose calculations for the respiratory tract from inhaled
natural radioactive nuclides as a function of age--II: basal cell dose
distributions and associated lung cancer risk. Health Phys. 43: 31-44.
Hofmann, W. (1982b) Mathematical model for the postnatal growth of the human
lung. Respir. Physio!. 49: 115-129.
6-6
-------
Hofmann, VI.; Steinhaeusler, F. ; Pohl, E. (1979) Dose calculations for the
respiratory tract from inhaled natural radioactive nuclides as a function
of age—I: compartmental deposition, retention and resulting dose. Health
Phys. 37: 517-532.
Holland, W. W.; Bennett, A. E. ; Cameron, I. R.; Florey, C. du V.; Leeder, S.
R.; Schilling, R. S. F.; Swan, A. V.; Waller, R. E. (1979) Health effects
of particulate pollution: reappraising the evidence. Am. J. .Epidemiol.
110: 525-659.
Horstman, D.; Roger, L. J.; Kehrl, H.; Hazucha, M. (1986) Airway sensitivity of
asthmatics to sulfur dioxide. Toxicol. Ind. Health 2: 289-298.
Hosein, H. R. ; Mitchell, C. A.; Bouhuys, A. (1977a) Evaluation of outdoor air
quality in rural and urban communities. Arch. Environ. Health 32: 4-13.
Hosein, H. R.; Mitchell, C. A.; Bouhuys, A. (1977b) Daily variations in air
quality. Arch. Environ. Health 32: 14-21.
Hsi, B. P.; Hsu, K. H. K. ; Jenkins, D. E. (1983) Ventilatory functions of
normal children and young adults: Mexican-American, white, and black. III.
Sitting height as a predictor. J. Pediatr. (St. Louis) 102: 860-865.
Hsu, K. H. K. ; Jenkins, D. E. ; Hsi, B. P.; Bourhofer, E.; Thompson, V.;
Tanakawa, N.; Hsieh, G. S. J. (1979) Ventilatory functions of normal
children and young adults - Mexican-American, white, and black. I. Spirom-
etry. J. Pediatr. (St. Louis) 95: 14-23.
Hutchinson, A. A.; Erben, A.; McLennan, L. A.; Landau, L. I.; Phelan, P. D.
(1981) Intrasubject variability of pulmonary function texting in health
children. Thorax 36: 370-377.
Imai, M. ; Yoshida, K. ; Kitabatake, M. (1986) Mortality from asthma and chronic
bronchitis associated with changes in sulfur oxides air pollution. Arch.
Environ. Health 41: 29-35.
International Commission on Radiological Protection. (1979) Limits for intake
of radionuclides by workers. ICRP publication 30. Ann. ICRP. 2:(3/4).
Isawa, T. ; Teshima, T. ; Hirano, T. ; Ebina, A.; Motomiya, M.; Konno, K. (1984)
Lung clearance mechanisms in obstructive airways disease. J. Nucl. Med.
25: 447-454.
Kehrl, H. R. ; Roger, L. J. ; Hazucha, M. J. ; Horstman, D. H. (1986) Differing
response of asthmatics to S02 exposure with continuous and intermittent
exercise. Am. Rev. Respir. Dis.: in press.
Kenline, P. A. (1962) In quest of clean air for Berlin, New Hampshire. Cincin-
nati, OH: U. S. Department of Health, Education, and Welfare, Division of
Air Pollution; technical report no. SEC TR A62-9.
Kirkpatrick, M. B.; Sheppard, D.; Nadel, J. A.; Boushey, H. A. (1982) Effect^of
the oronasal breathing route on sulfur dioxide-induced bronchoconstriction
in exercising asthmatic subjects. Am. Rev. Respir. Dis. 125: 627-631.
6-7
-------
Kitagawa, T. (1984) Cause analysis of the Yokkaichi asthma episode in Japan. J.
Air Pollut. Control Assoc. 34: 743-746.
Kleinman, M. T. (1984) Sulfur dioxide and exercise: relationships between
response and absorption in upper airways. J. Air Pollut. Assoc. 34: 32-37.
Koenig, J. Q.; Pierson, W. E. (1985) Pulmonary effects of inhaled sulfur
dioxide in atopic adolescent subjects: a review. In: Frank, R.; O'Neil, J.
J.; Utell, M. J.; Hackney, J. D.; Van Ryzin, J.; Brubaker, P. E. , eds.
Inhalation toxicology of air pollution: clinical research considerations.
Philadelphia, PA: American Society for Testing and Materials; publication
no. ASTM STP 872; pp. 85-91.
Koenig, J. Q.; Pierson, W. E.; Horike, M.; Frank, R. (1982) Effects of inhaled
sulfur dioxide (S02) on pulmonary function in healthy adolescents: exposure
to S0_2 alone or SC-2 + sodium chloride droplet aerosol during rest and
exercise. Arch. Environ. Health 37: 5-9.
Koenig, J. Q.; Pierson, W. E.;
sulfuric acid on pulmonary
Respir. Dis. 128: 221-225.
Horike, M. (1983a) The effects of inhaled
function in adolescent asthmatics. Am. Rev.
Koenig, J. Q.; Pierson, W. E.; Horike, M.; Frank, R. (1983b) A comparison of
the pulmonary effects of 0.5 ppm versus 1.0 ppm sulfur dioxide plus sodium
chloride droplets in asthmatic adolescents. J. Toxicol. Environ. Health
11: 129-139.
Koenig, J. Q.; Morgan, M. S.; Horike, M.; Pierson, W. E. (1985) The effects of
sulfur oxides on nasal and lung function in adolescents with extrinsic
asthma. J. Allergy Clin. Immunol. 6: 813-818.
Koenig, J. Q.; Marshall, S. G.; Horike, M. ; Shapiro, G. G. ; Furukawa, C. T. ;
Bierman, C. W.; Pierson, W. E. (1987) The effects of albuterol on sulfur
dioxide-induced bronchoconstriction in allergic adolescents. J. Allergy
Clin. Immunol. 79: in press.
Kulle, T. J.; Sauder. L. R. ; Shanty, F.; Kerr, H. D.; Parrel!, B. P.; Miller,
W. R.; Milman, J. H. (1984) Sulfur dioxide and ammonium sulfate effects on
pulmonary function and bronchial reactivity in human subjects. Am. Ind.
Hyg. Assoc. J. 45: 156-161.
Lange Andersen, K.; Rutenfranz, J.; Seliger, V.; Ilmarinen, J.; Berndt, I.;
Kylian, H. ; Ruppel, M. (1984) The growth of lung volumes affected by
physical performance capacity in boys and girls during childhood and
adolescence. Eur. J. Appl. Physio!. Occup. Physiol. 52: 380-384.
Lave, L B.; Seskin, E.
tive effect, with
ment, is considered. Science
P. (1970) Air pollution and human health. The quantita-
an estimate of the dollar benefit of pollution abate-
(Washington, DC) 169: 723-733.
Lave, L. B.; Seskin, E. P. (1972) Air pollution, climate, and home heating:
their effects on U.S. mortality rates. Am. J. Public Health 62: 909-916.
6-8
-------
Lave, L. B. ; Seskin, E. P. (1977) Air pollution and human health. Baltimore,
MD: The Johns Hopkins University Press.
Lawther, P. J. (1958) Climate, air pollution and chronic bronchitis. Proc. R.
Soc. Med. 51: 262-264.
Lawther, P. J. (1963) Compliance with the Clean Air Act: medical aspects. J.
Inst. Fuel 36: 341-344. *
Lawther, P. J.; Waller, R. E.; Henderson, M. (1970) Air pollution and exacerba-
tions of bronchitis. Thorax 25: 525-539.
Lebowitz, M. D. ; O'Rourke, M. K. ; Dodge, R. ; Holberg, C. J. ; Gorman, G. ;
Hoshaw, R. W. ; Pinnas, J. L.; Barbee, R. A.; Sneller, M. R. (1982) The
adverse health effects of biological aerosols, other aerosols, and indoor
microclimate on asthmatics and nonasthmatics. Environ. Int. 8: 375-380.
Lee, R. E.; Caldwell, J. S.; Morgan, G. B. (1972) The evaluation of methods for
measuring suspended particulates in air. Atmos. Environ. 6: 593-622.
Lee, P. S.; Gerrity, T. R. ; Hass, F. J. ; Lourenco, R. V. (1979) A model for
tracheobronchial clearance of inhaled particles in man and a comparison
with data. IEEE Trans Biomed. Eng. BME-26: 624-630.
Lellouch, J. (1986) Pollution atmospherique et affections respiratoires
chroniques [Air pollution and chronic respiratory diseases]. Soz.
Praeventivmed. 31: 8-12.
Levandowski, R. A.; Gerrity, T. R.; Garrard, C.
clearance mechanisms by acute influenza A
106: 428-432.
S. (1985) Modifications of lung
infection. J. Lab. Clin. Med.
Lilienfeld, P. (1979) GCA's model APM-1 ambient particulate monitor - field
applications and results. Presented at: 72nd annual meeting of the Air
Pollution Control Association; June; Cincinnati, OH. Pittsburgh, PA: Air
Pollution Control Association; paper no. 79-32.1.
Lilienfeld, P. (1985) Methods for continuous monitoring of particulates as
applied to the PM10 standard. In: Johnson, T. R.; Penkala, S. J. , eds.
Transactions: quality assurance in air pollution measurements, an APCA
specialty conference; October 1984; Boulder, CO. Pittsburgh, PA: Air
Pollution Control Association; pp. 64-80. (APCA transactions TR-3).
Linn, W. S. ; Bailey, R. M. ; Shamoo, D. A.; Medway, D. A.; Venet, T. G. ;
Wightman, L. H.; Hackney, J. D. (1982) Respiratory responses of young
adult asthmatics to sulfur dioxide exposure under simulated ambient
conditions. Environ. Res 29: 220-232.
Linn, W. S.; Shamoo, D. A.; Spier, C. E.; Valencia, L, M.; Anzar, U. T.; Venet,
T. G.; Hackney, J. D. (1983a) Respiratory effects of 0.75 ppm sulfur
dioxide in exercising asthmatics: influence of upper-respiratory defenses.
Environ. Res. 30: 340-348.
6-9
-------
Linn, W. S.; Venet, T. G.; Shamoo, D. A.; Valencia, L. M.; Anzar, U. T.; Spier,
C. E.; Hackney, J. D. (1983b) Respiratory effects of sulfur dioxide in
heavily exercising asthmatics. Am. Rev. Respir. Dis. 127: 278-283.
Linn, W. S.; Shamoo, D. A.; Venet, T. G.; Bailey, R. M.; Wightman, L. H.;
Hackney, J. D. (1984a) Comparative effects of sulfur dioxide exposures at
5°C and 22°C in exercising asthmatics. Am. Rev. Respir. Dis. 129: 234-239.
Linn, W. S.; Shamoo, D. A.; Vinet, T. G.; Spier, C. E.; Valencia, L. M. ; Anzar,
U. T.; Hackney, J. D. (1984b) Combined effect of sulfur dioxide and cold
in exercising asthmatics. Arch. Environ. Health 39: 339-346.
Linn, W. S.; Avol, E. L.; Shamoo, D. A.; Venet, T. G.; Anderson, K. R.; Whynot,
J. D.; Hackney, J. D. (1984c) Asthmatics' responses to 6-hr sulfur dioxide
exposures on two successive days. Arch. Environ. Health 39: 313-319.
Linn, W. S.; Shamoo, D. A.; Anderson, K. R.; Whynot, J. D.; Avol, E. L.;
Hackney, J. D. (1985a) Effects of heat and humidity on the responses of
exercising asthmatics to sulfur dioxide exposure. Am. Rev. Respir. Dis.
131: 221-225.
Linn, W. S.; Fischer, D. A.; Shamoo, D. A.; Spier, C. E. ; Valencia, L. M. ;
Anzar, U. T.; Hackney, J. D. (1985b) Controlled exposures of volunteers
with chronic obstructive pulmonary disease to sulfur dioxide. Environ.
Res. 37: 445-451.
Lioy, P. J.; Lippmann, M. (1986) Measurement of exposure to acidic sulfur
aerosols. In: Lee, S. D.; Schneider, T. ; Grant, L. D.; Verkerk, P. J.,
eds. Aerosols: research, risk assessment and control strategies, proceed-
ings of the second U.S.-Dutch international symposium; May 1985;
Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 743-752.
Lipfert, F. W. (1980) Sulfur oxides, particulates, and human mortality: synop-
sis of statistical correlations. J. Air Pollut. Control Assoc. 30:
366-371.
Lipfert, F. W. (1984) Air pollution and mortality: specification searches using
SMSA-based data. J. Environ. Econ. Manage. 11: 208-243.
Lippmann, M. (1977) Regional deposition of particles in the human respiratory
tract. In: Lee, D. H. K. ; Falk, H. L. ; Murphy, S. D. ;• Geiger, S. R. , eds.
Handbook of physiology, section 9: reactions to physical agents. Bethesda,
MD: American Physiological Society; pp. 213-232.
Lippmann, M.; Lioy, P. J. (1985) Critical issues in air pollution epidemiology.
EHP Environ. Health Perspect. 62: 243-258.
Lippmann, M.; Schlesinger, R. B. (1984) Interspecies comparisons of particle
deposition and mucociliary clearance in tracheobronchial airways. J.
Toxicol. Environ. Health 13: 441-469.
Lippmann, M.; Yeates, D. B. ; Albert, R. E. (1980) Deposition, retention, and
clearance of inhaled particles. Br. J. Ind. Med. 37: 337-362.
6-10
-------
Lippmann, M.; Schlesinger, R. B. ; Leikauf, G. ; Spektor, D. ; Albert, R. E.
(1982) Effects of sulphuric acid aerosols on respiratory tract airways.
In: Walton, W. H.; Critchlow, A.; Coppock, S. M., eds. Inhaled particles
V: proceedings of an international symposium organized by the British
Occupational Hygiene Society; September 1980; Cardiff, United Kingdom.
Ann. Occup. Hyg. 26: 677-690.
Lunn, J. E. ;
illness
7-16.
Knowelden, J. ; Handyside, A. J. (1967) Patterns of respiratory
in Sheffield infant schoolchildren. Br. J. Prev. Soc. Med. 21:
Lunn, J. E. ; Knowelden, J. ; Roe, J. W. (1970) Patterns of respiratory illness
in Sheffield junior schoolchildren: a follow-up study. Br. J. Prev. Soc.
Med. 24: 223-228.
Martin, A. E. (1964) Mortality and morbidity statistics and air pollution.
Proc. R. Soc. Med. 57: 969-975.
Martin, A. E.; Bradley, W. H. (1960) Mortality, fog and atmospheric pollution:;
an investigation during the winter of 1958-59. Mon. Bull. Minist. Health
Public Health Lab. Serv. G. B. 19: 56-73.
Mazumdar, S.; Sussman, N. (1981) Relationships of air pollution to health:
results from the Pittsburgh study. Presented at: 74th annual meeting of
the Air Pollution Control Association; June; Philadelphia, PA. Pittsburgh,
PA: Air Pollution Control Association; paper no. 81-11.3.
Mazumdar, S. ; Sussman, N. (1983) Relationships of air pollution to health:
results from the Pittsburgh study. Arch. Environ. Health 38: 17-24.
Mazumdar, S. ; Schimmel, H. ; Higgins, I. (1981) Daily mortality, smoke and S02
in London, England 1959 to 1972. In: Frederick, E. R., ed. A specialty
conference on: the proposed SOx and particulate standard; September 1980;
Atlanta, GA. Pittsburgh, PA: Air Pollution Control Association; pp.
219-239.
Mazumdar, S.; Schimmel, H.
ty to air pollution:
Arch. Environ. Health
; Higgins, I. T. T.
an analysis of 14
37: 213-220.
(1982) Relation of daily mortali-
London winters, 1958/59-1971/72.
McFarland, A. R. ; Ortiz, C. A.; Rodes, C. E. (1982) Wind tunnel evaluation of
the British smoke shade sampler. Atmos. Environ. 16: 325-328.
Mendelsohn, R. ; Orcutt, G. (1979) An empirical analysis of air pollution
dose-response curves. J. Environ. Econ. Manage. 6: 85-106.
Miller, K.; DeKoning, H. W. (1974) Particle sizing instrumentation. Presented
at: 67th annual meeting of the Air Pollution Control Association; June;
Denver, CO. Pittsburgh, PA: Air Pollution Control Association; paper no.
74-48.
Miller, F. J. ; Gardner, D. E. ; Graham, J. A.; Lee, R. E., Jr.;,Wilson, W. E.;
Bachmann, J. D. (1979) Size considerations for establishing a standard for
inhalable particles. J. Air Pollut. Control Assoc. 29: 610-615.
6-11
-------
Miller, F. J.; Grady, M. A.; Martonen, T. B. (1984) Coarse mode aerosol behav-
ior in man: theory and experiment. In: Liu, B. Y. H.j.Piu, D. Y. H.;
Fissan, H., eds. Aerosols: science, technology, and industrial applica-
tions of airborne particles. New York, NY: Elsevier Science Publishing
Co. , Inc.; pp. 999-1002.
Miller, F. J.; Martonen, T. B.; Menache, M. G.; Spektor, D. M. ; Lippmann, M.
(1986) Influence of breathing mode and activity level on the regional
deposition of inhaled particles and implications for regulatory standards.
Cambridge, United Kingdom: Inhaled particles VI: accepted for publication.
Mitchell, C. A.; Schilling, R. S. F. ; Bouhuys, A. (1976) Community studies of
lung disease in Connecticut - organization and methods. Am. J. Epidemic!.
103: 212-225.
Morrow, P. E. (1984) Aerosol factors affecting respiratory deposition.
Atemwegs Lungenkr. 10: 231-236.
Muehling, P.; Bory, J.; Haupt, H. (1985) Einfluss der Luftbelastung auf
Atemwegserkrankungen. Untersuchungen bei Saeuglingen und Kleinkindern [The
influence of air pollution on respiratory diseases: studies of babies and
infants]. Staub Reinhalt. Luft 45: 35-38.
Myers, D. J.; Bigby, B. G.; Boushey, H. A. (1986a) The inhibition of sulfur
dioxide-induced bronchoconstriction in asthmatic subjects by cromolyn is
dose dependent. Am. Rev. Respir. Dis. 133: 1150-1153.
Myers, D. J.; Bigby, B. G.; Calvayrac, P.; Sheppard, D.; Boushey, H. A. (1986b)
Interaction of cromolyn and a muscarinic antagonist in inhibiting bronchi-
al reactivity to sulfur dioxide and to eucapnic hyperpnea alone. Am. Rev.
Respir. Dis. 133: 1154-1158.
Nadel, J. A.; Salem, H.; Tamp!in, B.; Tokiwa, Y. (1965) Mechanism of broncho-
constriction during inhalation of sulfur dioxide. J. Appl. Physiol. 20:
164-167.
National Research Council. (1979) Airborne particles. Baltimore, MD: University
Park Press; p. 4.
Niinimaa, V.; Cole, P.; Mintz, S.; Shephard, R. J. (1980) The switching point
from nasal to oronasal breathing. Respir. Physiol. 42: 61-71.
Niinimaa, V.; Cole, P.; Mintz, S.; Shephard, R. J. (1981) Oronasal distribution
of respiratory airflow. Respir. Physiol. 43: 69-75.
Osborne, D. R. S.; Effmann, E. L. ; Hedlund, L. W. (1983) Postnatal growth and
size of the pulmonary acinus and secondary lobule in man. AJR Am. J.
Roentgenol. 140: 449-454.
Ostro, B. D. (1983) The effects of air pollution on work loss and morbidity. J.
Environ. Econ. Manage. 10: 371-382.
6-12
-------
Ostro, B. (1984) A search for a threshold in the relationship of air pollution
to mortality: a reanalysis of data on London winters. EHP Environ. Health
Perspect. 58: 397-399.
Ostro, B. D. (1987) Air pollution and morbidity revisited: a specification
test. J. Environ. Econ. Manage.: in press.
Ozkaynak, H. ; Spengler, J. D. (1985) Analysis of health effects resulting from
population exposures to acid precipitation precursors. In: Conference on
health effects of acid precipitation; November 1984; Research Triangle
Park, NC. EHP Environ. Health Perspect. 63: 45-55.
Ozkaynak, H. ; Schatz, A. D. ; Thurston, G. D. ; Isaacs, R. G. ; Husar, R. B.
(1985) Relationships between aerosol extinction coefficients derived from
airport visual range observations and alternative measures of airborne
particle mass. J. Air Pollut. Control Assoc. 35: 1176-1185.
Ozkaynak, H. ; Spengler, J. D. ; Garsd, A.; Thurston, G. D. (1986) Assessment of
population health risks resulting from exposures to airborne particles.
In: Lee, S. D.; Schneider, T.; Grant, L. D.; Verkerk, P. J., eds. Aero-
sols: research, risk assessment and control strategies, proceedings of the
second U.S.-Dutch international symposium; May 1985; Williamsburg, VA.
Chelsea, MI: Lewis Publishers, Inc.; pp. 1067-1080.
PAARC Cooperative Group. (1982a) Atmospheric pollution and chronic or recurrent
respiratory diseases. I. Methods and material. Bull. Eur. Physiopathol.
Respir. 18: 87-99.
PAARC Cooperative Group. (1982b) Atmospheric pollution and chronic or recurrent
respiratory disease. II. Results and discussion. Bull. Eur. Physiopathol
Respir. 18: 101-116.
Pengelly, L. D. ; Goldsmith, C. H.; Lerigan, A. T.; Furlong, W.; Toplack, S. A.
(1986) The Hamilton study: effect of particle size on respiratory health
in children. In: Lee, S. D. ; Schneider, T.; Grant, L. D. ; Verkerk, P. J.,
eds. Aerosols: research, risk assessment and control strategies, proceed-
ings of the second U.S.-Dutch international symposium; May 1985;
Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 753-766.
Perry, G. B. ; Chai, H. ; Dickey, D. W.; Jones, R. H.; Kinsman, R. A.; Merrill,
C. G.; Spector, S. L.; Weiser, P. C. (1983) Effects of particulate air
pollution on asthmatics. Am. J. Public Health 73: 50-56.
Phalen, R. F.; Oldham, M. J.; Beaucage, C. B.; Crocker, T. T.; Mortensen, J. D.
(1985) Postnatal enlargement of human tracheobronchial airways and impli-
cations for particle deposition. Anat. Rec. 212: 368-380.
Philipson, K. ; Falk, R. ; Camner, P. (1985) Long-term lung clearance in humans
studied with teflon particles labeled with chromium-51. labeled with
chromium-51. Exp. Lung Res. 9: 31-42.
Pierson, W. E.; Covert, D. S. ; Koenig, J. Q. (1984) Air-pollutants,'bronchial
hyperreactivity, and exercise. J., Allergy Clin. Immunol. 73: 717-721.
6-13
-------
Proctor, D. F.; Sheppard, D.; Nadel, J. A.; Boushey, H. A. (1981) Oronasal
breathing and studies of effects of air pollutants on the lungs [letter to
the editor and response]. Am. Rev. Respir. Dis. 123: 242-243.
Roger, L. J.; Kehrl, H. R.; Hazucha, M. ; Horstman, D. H. (1985) Bronchocon-
striction in asthmatics exposed to sulfur dioxide during repeated exer-
cise. J. Appl. Physiol. 59: 784-791.
Rondinelli, R. C. A.; Koenig, J. Q.; Marshall, S. G. (1986) The effects of
sulfur dioxide on pulmonary function in healthy non-smoking male subjects
aged 55 years and older. Am. Ind. Hyg. Assoc. J.: accepted for
publication.
Roth, H.; Wyzga, R.; Hayter, A. (1986) Methods and problems in estimating
health risks from particulates. In: Lee, S. D.; Schneider, T. ; Grant, L.
D.; Verkerk, P. J. , eds. Aerosols: research, risk assessment and control
strategies, proceedings of the second U.S.-Dutch international symposiums;
May 1985; Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp.
1047-1065.
Saric, M.; Fugas, M.; Hrustic,
school-age children. Arch.
0. (1981) Effects of urban air pollution on
Environ. Health 36: 101-108.
Schachter, E. N.; Witek, T. J.; Beck, G. J.; Hosein, H. R.; Colice, G.;
Leaderer, B. P.; Cain, W. (1984) Airway effects of low concentrations of
sulfur dioxide: dose-response characteristics. Arch. Environ. Health 39:
34-42.
Schachter, E. N.; Witek, T. J.; Hosein, H. R. (1986) [Letter to Dr. Lester
D. Grant with a manuscript entitled: Nasal absorption of S02 during rest
and exercise in healthy human volunteers]. New York, NY: Mount Sinai
Medical Center, Department of Medicine and Community Medicine.
Schenker, M. B.; Samet, J. M.; Speizer, F. E.; Gruhl, J.; Batterman, S. (1983)
Health effects of air pollution due to coal combustion in the Chestnut
Ridge region of Pennsylvania: results of cross-sectional analysis in
adults. Arch. Environ. Health 38: 325-330.
Schimmel, H. (1978) Evidence for possible acute health effects of ambient air
pollution from time series analysis: methodological questions and some new
results based on New York City daily mortality, 1963-1976. Bull. N. Y.
Acad. Med. 54: 1052-1108.
Schlesinger, R. B. (1985) Effects of inhaled acids on respiratory tract defense
mechanisms. In: Conference on health effects of acid precipitation;
November 1984; Research Triangle Park, NC. EHP Environ. Health Perspect.
63: 25-38.
Schlesinger, R. B. (1986) The effects of inhaled acid aerosols on lung defens-
es. In: Lee, S. D.; Schneider, t.; Grant, L. D.; Verkerk, P. J., eds.
Aerosols: research, risk assessment and control strategies, proceedings of
the second U.S.-Dutch international symposium on aerosols; May 1985;
Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 617-635.
6-14
-------
Schwartz, J. ; Marcus, A. A. (1986) Statistical reanalyses of data relating
mortality to air pollution during London winters 1958-1972. Washington,
DC: U.S. Environmental Protection Agency, Office of Policy, Planning and
Evaluation.
Selvin, S. ; Merrill, D. ; Wong, L.; Sacks, S. T. (1984) Ecologic regression
analysis and the study of the influence of air quality on mortality. EHP
Environ. Health Perspect. 54: 333-340.
Sheppard, D.; Wong, W. S. ; Uehara, C. F. ; Nadel, J. A.; Boushey, H. A. (1980)
Lower threshold and greater bronchomotor responsiveness of asthmatic
subjects to sulfur dioxide. Am. Rev. Respir. Dis. 122: 873-878.
Sheppard, D. ; Nadel, J. A.; Boushey, H. A. (1981) Inhibition of sulfur dioxide-
induced bronchoconstriction by disodium cromoglycate in asthmatic sub-
jects. Am. Rev. Respir. Dis. 124: 257-259.
Sheppard, D. ; Epstein, J. ; Bethel, R. A.; Nadel, J. A.; Boushey, H. A. (1983)
Tolerance to sulfur dioxide-induced bronchoconstriction in subjects with
asthma. Environ. Res. 30: 412-419.
Sheppard, D.; Eschenbacher, W. L. ; Boushey, H. A.; Bethel, R. A. (1984) Magni-
tude of the interaction between the bronchomotor effects of sulfur dioxide
and those of dry (cold) air. Am. Rev. Respir. Dis. 130: 52-55.
Shumway, R. H.; Tai, R. Y.; Tai, L. P.; Pawitan, Y. (1983) Statistical analysis
of daily London mortality and associated weather and pollution effects.
Sacramento, CA: California Air Resources Board; contract no. Al-154-33.
Snashall, P. D.; Baldwin, C. (1982) Mechanisms of sulphur dioxide induced
bronchoconstriction in normal and asthmatic man. Thorax 37: 118-123.
Spengler, J. D. ; Thurston, G. D. (1983) Mass and elemental composition of fine
and coarse particles in six L). S. cities. J. Air Pollut. Control Assoc.
33: 1162-1171.
Stacy, R. W. ; Seal, E., Jr.; House, D. E. ; Green, J.; Roger, L. J.; Raggio, L.
(1983) A survey of effects of gaseous and aerosol pollutants on pulmonary
function of normal males. Arch. Environ. Health 38: 104-115.
Stahlhofen, W. ; Gebhart, J. ; Heyder, J. (1980) Experimental determination of
the regional deposition of aerosol particles in the human respiratory
tract. Am. Ind. Hyg. Assoc. J. 41: 385-398a.
Stearns, D. R.; McFadden, E. R., Jr.; Breslin, F. J.; Ingram, R. H., Jr. (1981)
Reanalysis of the refractory period in exertional asthma. J. Appl. Physio!,
50: 503-508.
Stebbings, J. H. , Jr.; Fogleman, D. G. (1979) Identifying a susceptible sub-
group: effects of the Pittsburgh air pollution episode upon school-
children. Am. J. Epidemic!. 110: 27-40.
Strauss, R. H. ; McFadden, E. R., Jr.; Ingram, R. H., Jr.; Jaeger, J. J. (1977)
Enhancement of exercise-induced asthma by cold air. New Engl. J. Med. 297:
743-747.
6-15
-------
Svartengren, M. (1986) Lung deposition and clearance of particles in healthy
persons and patients with bronchiectasis. Stockholm, Sweden: Karolinska
Institute.
Svartengren, M. ; Philipson, K. ; Linnman, L. ; Camner, P. (1984) Airway resis-
tance and deposition of particles in the lung. Exp. Lung Res. 7: 257-269.
Svartengren, M.; Hassler, E.; Philipson, K.; Camner, P. (1986) Spirometric data
and penetration of particles to the alveoli. Br. J. Ind. Med. 43: 188-191.
Swift, D. L. ; Proctor, D. F. (1982) Human respiratory deposition of particles
during oronasal breathing. Atmos. Environ. 16: 2279-2282.
Tan, W. C. ; Cripps, E.; Douglas, N.; Sudlow, M. F. (1982) Protective effect of
drugs on bronchoconstriction induced by sulphur dioxide. Thorax 37:
671-676.
Task Group on Lung Dynamics. (1966) Deposition and retention models for inter-
nal dosimetry of the human respiratory tract. Health Phys. 12: 173-207.
Thibodeau, L. A.; Reed, R. B.; Bishop, Y.
pollution and human health: a review
Perspect. 34: 165-183.
M. M. ; Kammerman, L. A. (1980) Air
and reanalysis. EHP Environ. Health
U. S. Environmental Protection Agency. (1982a) Air quality criteria for partic-
ulate matter and sulfur oxides. Research Triangle Park, NC: Environmental
Criteria and Assessment Office; EPA report nos. EPA-600/8-82-029aF-cF. 3v.
Available from: NTIS, Springfield, VA; PB84-156777.
U. S. Environmental Protection Agency. (1982b) Review of the national ambient
air quality standards for particulate matter: assessment of scientific and
technical information, OAQPS staff paper. Research Triangle Park, NC:
Office of Air Quality Planning and Standards; EPA report no.
EPA-450/5-82-001. Available from: NTIS, Springfield, VA; PB82-177874
U. S. Environmental Protection Agency. (1982c) Air quality criteria for particu-
late matter and sulfur oxides: v. 1, addendum. Research Triangle Park, NC:
Environmental Criteria and Assessment Office; pp. A1-A15; EPA report no.
EPA-600/8-82-029aF. Available from NTIS, Springfield, VA; PB84-156785/REB.
U. S. Environmental Protection Agency. (1982d) Review of the national ambient
air quality standards for sulfur oxides: OAQPS staff paper. Research
Triangle Park, NC: Office of Air Quality Planning and Standards; EPA
report no. EPA-450/5-82-007. Available from: NTIS, Springfield, VA;
PB84-102920.
United Kingdom Ministry of Health. (1954) Mortality and morbidity during the
London fog of December 1952. London, United Kingdom: Her Majesty's Sta-
tionary Office. (Reports on public health and medical subjects no. 95).
Utell, M. J. (1985) Effects of inhaled acid aerosols on lung mechanics: an
analysis of human exposure studies. In: Conference on health effects of
acid precipitation; November 1984; Research Triangle Park, NC. EHP
Environ. Health Perspect. 63: 39-44.
6-16
-------
Utell , M. J. ; Morrow, P. E. (1986) Effects of inhaled acid aerosols on human
lung function: studies in normal and asthmatic subjects. In: Lee, S. D. ;
Schneider, T.; Grant, L. D.; Verkerk, P. J., eds. Aerosols: research, risk
assessment and control strategies, proceedings of the second U.S.-Dutch
international symposium; May 1985; Williamsburg, VA. Chelsea, MI: Lewis
Publishers, Inc.; pp. 671-681.
van der Lende, R. ; Schouten, J. P. ; Rijcken, B. ; van der Meulen, A. (1986)
Longitudinal epidemiological studies on effects of air pollution in The
Netherlands. In: Lee, S. D. ; Schneider, T.; Grant, L. D.; Verkerk, P. J. ,
eds. Aerosols: research, risk assessment and control strategies, proceed-
ing of the second U.S.-Dutch international symposium; May 1985;
Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 731-742.
Waller, R. E. (1964) Experiments on the calibration .of smoke filters. J. Air
Pollut. Control Assoc. 14: 323-325.
Waller, R. E. (1971) Air pollution and community health. J. R. Coll. Physicians
(London) 5: 362-368.
Ware, J. H. ; Thibodeau, L. A.; Speizer, F. E.; Colome, S.; Ferris, B. G., Jr.
(1981) Assessment of the health effects of atmospheric sulfur oxides and
particulate matter: evidence from observational studies. EHP Environ.
Health Perspect. 41: 255-276.
Ware, J. H.; Ferris, B. G. , Jr. ; Dockery, D. W.; Spengler, J. D.; Stram, D. 0.;
Speizer, F. E. (1986) Effects of ambient sulfur oxides and suspended
particles on respiratory health of preadolescent children. Am. Rev.
Respir. Dis. 133: 834-842.
Wedding, J. B. ; Carney, T. C. (1983) A quantitative technique for determining
the impact of non-ideal ambient sampler inlets on the collected mass.
Atmos. Environ. 17: 873-882.
Weibel, E. R. (1963) Morphometry of the
Press, Inc., Publishers.
human lung. New York, NY: Academic
Whitby, K. T. (1975) Modeling of atmospheric aerosol particle size distribu-
tions: a progress report on EPA research grant no. R800971, "sampling and
analysis of atmospheric aerosols." Minneapolis, MN: University of Minneso-
ta, Mechanical Engineering Department; p. 111-21; particle technology
laboratory publication no. 253.
Wilson, N. M. ; Barnes, P. J. ; Vickers, H. ; Silverman, M. (1982) Hyperventila-
tion-induced asthma: evidence for two mechanisms. Thorax 37: 657-662.
Witek, T. J. ; Schachter, E. N. (1985) Airway responses to sulfur dioxide
and methacholine in asthmatics. JOM J. Occup. Med. 27: 265-268.
Witek, T. J. ; Schachter, E. N. ; Beck, G. J. ; Cain, W. S. ; Colice, G. ;
Leaderer, B. P. (1985) Respiratory symptoms associated with sulfur
dioxide exposure. Int. Arch. Occup. Environ. Health 55: 179-183.
6-17
-------
Wojtyniak, B.; Krzyzanowski, M.; Jedrychowski, W. (1984) Die Bedeutung
staedtischer Luftverunreinigung fuer chronische [Importance of urban air
pollution in chronic respiratory problems]. Z. Erkr. Atmungsorgane 163:
274-284.
Wolff, R. K.; Obminski, G.; Newhouse, M. T. (1984) Acute exposure of sympto-
matic steelworkers to sulfur dioxide and carbon dust: effects on
mucsciliary transport pulmonary function and bronchial reactivity. Br. J.
Ind. Med. 41: 499-505.
World Health Organization. (1976) Selected methods of measuring air pollutants.
Geneva, Switzerland: World Health Organization. (WHO offset publication
no. 24).
World Health Organization. (1977) Air monitoring programme design for urban and
industrial areas. Geneva Switzerland: World Health Organization. (WHO
offset publication no. 33).
World Health Organization. (1979) Sulfur oxides and suspended particulate
matter. Geneva, Switzerland: World Health Organization. (Environmental
health criteria 8).
Xu, G. B.; Yu, C. P. (1986) Effects of age on deposition of inhaled aerosols in
the human lung. Aerosol Sci. Techno!. 5: 349-357.
Yu, C. P.; Diu, C. K. ; Soong, T. T. (1981) Statistical analysis of aerosol
deposition in nose and mouth. Am. Ind. Hyg. Assoc. J. 42: 726-733.
6-18
-------
APPENDIX
Schwartz and Marcus Statistical Reanalysis of Mortality Data
during 14 London Winters (1958-1972).
A-l
-------
\
%
:/
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
October 14, 1986
MEMORANDUM
SUBJECT: Statistical Reanalyses of Data Relating Mortality to Air Pollution
During 14 London Winters (1958-1972)
FROM:
TO:
Allan Marcus
Ambient Standards Branch
Joel Schwartz
Economic and Regulatory Analysis Division!
Bruce Jordan, Chief
Ambient Standards Branch
Les Grant, Director
Environmental Criteria and Assessment Office
This memo summarizes our continuing analysis of the London mortality
data. These analyses were conducted at your request for the purpose of
delineating further the degree of reliance that can be put on the more
recent published analyses of these data discussed in the criteria document
and staff paper addenda. Our analyses are discussed in more detail in the
attached paper. The recently published studies include three statistical
analyses of the possible relationship between daily air pollution concentrations
and days with excessive numbers of deaths in London during the winters of
1958-1972. (Mazumdar et a!., 1982; Ostro, 1984; Shumway et al., 1983)
We believe that these studies have shown that a relationship exists
between particulate matter as measured by British Smoke or S02 and mortality
in London, and that those relationships continue below a British Smoke
level of 150 ug/m^. However, commenters and others have raised questions
about 1) whether the analyses adequately handled the temporal structure of
the data, both in terms of avoiding confounding due to long term time
trends and seasonal fluctuations, and in terms of avoiding the mis-estimation
of the regression standard errors (and hence significance tests) that
occurs when there is autocorrelation in the regression residuals; 2) whether
the dose response relationship is linear or nonlinear and whether that
relationship is distorted by the techniques used to filter the series, and
3) whether it is British Smoke, S02 or both that are responsible for the
mortality.
A-2
-------
All of the studies have attempted to deal with the autocorrelation in the
data (that is, the number of people who die on day t is correlated with the
number of people who died on day t-1, t-2, etc.). Some used deviations from
15 day moving average to remove these autocorrelations, more recently Ostro
used an autoregressive model. All of the studies used separate regressions
for each year to remove the time trend of falling mortality. None of the
models successfully separated S02 and British Smoke effects. None of the
studies reported any tests to determine whether using separate regressions
for each year adequately dealt with the possibility of a linear time trend
in the data or reported any formal tests to determine whether they had in
fact adequately accounted for the autocorrelations in the data. In addition,
while the studies clearly demonstrated relationships at low levels, and gave
some indication of their magnitude, no detailed exploratory analysis was
presented to determine the potential shape of the dose response curve.
With the assistance of analysts at the California Air Resources Board,
who had obtained the full 14 winter data set from the United Kingdom, we
therefore decided to reanalyze the London mortality data so as to evaluate
the adequacy of the fitted models. We began by grouping and examining the
raw data graphically for all years and each year separately much in the manner
used by Ware et al. (1981) and the 1982 Criteria Document for the 1958-59 winter.
Figure 1 shows the results of raw daily mortality versus smoke for all winters
while Figure 2 shows the same plot for all days with smoke less than 500 ug/m^.
Figure 3 illustrates deviations in daily mortality for a representative
individual winter. In general, these plots show the same kind of continuum
of association between mortality and smoke seen in the earlier analyses,
with no apparent lower limit. Both the curvilinear shape of the dose
response curve and the low level effects are also evident in year by year
plots. We then decided to study the temporal structure of the process,
particularly its autocorrelation. We developed regression models that
control for the effects of autocorrelation. These models were then used
to study the relative usefulness of BS and S02 as predictors of mortality.
A comparison of the key features of these as well as the published regressions
is summarized in Table 1.
These additional analyses have suggested the following conclusions:
(A) Short-term changes in mortality can be very well modeled by an
autoregressive process with two or three terms (i.e., mortality on day t
predicted by a combination of residual mortality on days t-1, t-2, and
possibly t-3). The autoregressive part (AR1-3) alone usually accounted for
about 54% of the variance in each year's daily mortality. When only days
with BS < 200 were considered, the fraction of variance explained by
autoregression increased to about 58%. The AR3 structure of the data was
not completely modeled by either the 15-day average detrending or by the
AR1 model used in an unpublished analysis by Ostro.
(B) When temperature, humidity and one pollutant were considered in an
autoregressive model, the incremental variation in mortality explained was
about 14%. Using British Smoke as the exposure variable, pollution was
significantly related to mortality in 13 out of 14 years. In a random effects
A-3
-------
model that combined the results from all years British Smoke was highly
significant (P < .0001). When smoke levels were restricted to only those
below 500 ug/m3 or even below 200 ug/m3 the overall significance of British
Smoke increased (t = 8.74 for levels below 500 ug/m3 and t = 14.43 for levels
below 200 ug/m3).
(C) Regression models for mortality using BS, S02, temperature and
humidity accounted for substantial additional variance in daily mortality
over and above the autoregressive components. This regression model added
an additional fraction of explained variance, about 14%, to the autoregressive
model. Even when only days with BS < 200 were considered the regression
model explained about 12% of the variance in mortality.
(D) Air pollution variables were usually more significant statistically
than temperature or humidity in explaining mortality.
(E) Due to the multicollinearity of BS and S02, there were no years
when both were significant at the 5% level of significance. However, BS
was always significant statistically for more years than was S02. Using
all data, BS was significant in 6 years and S02 in 2 years out of 14, and
even on days with BS < 500, BS was significant 4 years and S02 was never
significant in 14 years when both variables were used. These results are
shown in Table 2. However, the statistical significance of both variables
was greatly reduced because of the multicollinearity. What is more striking
is that the multiple regression slope for BS was relatively stable whether
or not S02 was used in the model, with a mean slope (weighted by the reciprocal
of the variance) of 0.079 excess deaths per ug/nr BS without SO? in the
model, and 0.061 with S02 included. The estimated slopes for S&2 were
significantly positive only for 1958 and 1962, and were otherwise insignificant
and scattered around zero slope. Even on days with BS < 200 only, the mean
yearly slopes for BS were 0.138 without S02 and 0.135 with S02 included
(both highly significant) (Tables 3,4). Again, the S02 slope was
approximately zero.
Our recent assessments thus confirm that the multiple regression models
for daily mortality and BS do indeed reflect a relationship that cannot be
attributed to time series effects, temperature, S02, or functional
misspecification. The general consistency of the results is shown in
Figure 4. This shows the regression slope of mortality vs. BS for Mazumdar's
linear model (coded M), for Ostro's low-BS linear model (BS < 150, coded
L), and for the CARB/EPA analyses using all days (coded A), days with BS <
500 (coded B), and days with BS < 200 (coded D). There does appear to be
some tendency for higher slopes in later years when both BS and S02 levels
reflect more nearly contemporary conditions. Thus the published analyses
do appear to be relevant in assessing the health effects of particulate
matter.
A-4
-------
REFERENCES
Mazumdar, S.; Schimmel, H.;Higgins, I.T.T. (1983). Letter to the
editor. Arch. Env. HUh. 38: 123-126.
Mazumdar, S.; Schimmel, H.; Higgins, I.T.T. (1982) Relation of daily
mortality to air pollution: an analysis of 14 London winters,
1958/59-1971/72. Arch. Environ. Health 37: 213-220.
Ostro, B. (1984) A search for a threshold in the relationship of air pollution
to mortality: a reanalysis of data on London winters. EHP Environ. Health
Perspect. 58: 397-399.
Ostro, B. (1986). Letter to Dr. L.D. Grant, ECAO, August 15, 1986.
Shumway, R.H.; Tai, R.Y.; Tai, L.P.; Pawitan, Y. (1983) Statistical analysis
of daily London mortality and associated weather and pollution effects.
Sacramento, CA: California Air Resources Board; contract no. Al-154-33.
Shumway, R.H. (1986) Letter to A.H. Marcus, October 2, 1986.
Ware, 0. H.; Thibodeau, L. A.; Speizer, F. E.; Colome, S.; Ferris, B. G., Jr.
(1981) Assessment of the health effects of atmospheric sulfur oxides and
particulate matter: Evidence from observational studies. Environ.
Health Perspect. 41: 255-276.
A-5
-------
J-,.
.3N
in '
^^ r^
~o
§2
toco
j-10
«m
^5 T"
•S M
ȣ
*!
~m e
P
O
•
•
*
•
•
«
*
•g
a
*
• .
•
•
9
^
* . "
•
* •
• •
* . * . .
« • • •
• • • mm
• * * *
• • • • • • •
• • • •
t__ • 1 JL • 1"
o
3
r-
5
§
N
O
O
^*
*"
§
O
*"
§
O)
§
eo
E
a
'"1
ii
&
s
IO
§
«r
o
o
m
o
S
o
o
r*
0
to o 10 o to o
% s a M £ ss
CO *v TO " *^> W
•4-
O
• 10
CM 0)
1*- 3
CTi i—
"7 >
i— i
CTi C
f-H O)
u
15
•c o
•M CM
cn t-
Lf> O
fM^
1 OO •
eo eo to
in eo
en OJ
r- 1 CT1 It-
fl3 O
<«- i-
O O> in
g re^
OJ TJ O)
4-> C i—
c: } O
•r- re
re
^"(ut^
•i- j= a)
T3 CD
4-> (/» 3
t- 4-> 10
O C
E flj re
>, Oi re
E- Q.
o •
•r- re IS)
U- LU CD
A-6
-------
ffk
to
E
^«*
CM
o
^^ IS
fc« 1
^j ST
= 3j
^J V*
& **
Q 0)
lish Smoke (All
Winters 1958-
»i
^3
I Q
^fc _f
,*J
To
i
j^
"3
o
«
••
.
•
*
•
9 ,
,
*
•
•
*
•
'
9
^
• _ '
O
O
* 0
• •
•
. • •
• » .
• •
• a
* . " I
"* . • •
* •
a
*
1 1 I i « "
U)
0
!>
9
*
O
o
CO
s
!*)
o
10
N
O
O
0
to
f—
§
r-
O
O K> Q * O
ID W O ^* IO
m « n ->
P*fc. ^ ^
r^*. ^ ^_
CTl r- O
r— 1 TO O
-<= >1
cn+J i—
3 C t_
o QJ -i-s -
O°t C^ ^^
•—I CM CO
1 C
CO t- O
ID O ''-
CT> <4- 4J
1-1 c/> t^
» H- OQ O)
E o t.
» «0 01 0)
e" OJ l- •(->
•g -^ «
2 C > T3
E T-
a 4J c
c ••- re
•i- ,— ia.
C_ <«
•a t-
o •—
= E r-
o ••-
•O S")-!-'
C i — (O
0 •!-
_i re co
•a ••-
c: •
•i- OJ QTO
>, re^lc -^
-(-> i- to en
•f- O
(O TO *r- O
•!-> +-> O
t- QJ TC) r-H
O ^" .
L- O
^) UO
•p™ ^^ TO t~~
TO CJ Q^ CU
O' CO C >
a> •!— oj
t- »~™ i~™
CSJ QJ > 00
£_ t- OD
OJ 3
£_ 4-> O 5
3 C 0
cr>-i- ai i —
• . *^ i f**
-ON
AIIBQ
A-7
-------
O
M 3
t ->
O
>• K a.
•-0 E
*» O
§ 5!
0 0
CM
cc
0)
O)
OJ "3
+J oj
> c/1
t. "3 OS
o
v-
O (U T3
Ql-r- (O JE
QJ O -*=
10 CD.C 4->
(O {/)
0> I. -M
C Ol >, C
•r- > 4J O)
O ^~ '^
s rti fa o
&^^-> S
"O (/)
i •<->
tn c
t_
o
o
o i- -o 5
L. o.
>«- oj c c
t_ 1- -r-
•r- c o w
4jQ.ro>,
O f > T-
E O O) r—
(O Q ItS
x m
a
ic o
o m
to • (/*)
"O Ch OQ
M- CTi <*-
O ^H O
W) CO
c uo a)
O CTl 3
re It- >
•r- O +•>
01 «- c JE
Q 0) 0) 4->
4-> O
^ ^ "S
0) C
-------
250
a.
o
200 •
175 •
150
125
100
75
50
25
-4-
-t-
58 59 60 61 62 63 64 65 66 67 68 69 70 71
YEAR
Figure 4. Sensitivity of estimated slope (excess deaths per day per 1000
ug/m3 BS) to statistical method. Symbols: M, Mazumdar (1982); L, Ostro
(1984) for BS < 150;-unpublished CARB/EPA analyses: A, all days; B, days
with BS < 500; D, days with BS < 200.
A-9
-------
Table 1. FEATURES OF REGRESSION MODELS FOR LONDON MORTALITY DATA
Study
Mortal ity
BS
SO?
TEMP
Time Series
Mazumdar (1982)* D15
D15
D15
D15
D15
Ostro (1984)* D15
Shumway (1983) DW
OTHER ANALYSES
Ostro (1982)* D15
D15
D15
Shumway (1983) DW
Ostro (1986)* C
D15
D15
CARB/EPA (1986)* C
C
C
D15
D15
D15
*Year by year results.
ARX: Autoregressive model of
C: crude
DX: deviations from moving
L: linear
D15.L
D15.Q
-
-
D15.L & Q
PL150
-
PL75
PL100
PL200
DW, log
L
LT200
LT300
LT200
-
LT200
LT200
_
LT200
order x
average (X=15
D15.L
D15.L
D15.L D15,L
D15.Q D15.L
D15.L & 0 D15.L
L
DW, log DW
L
L
_ 1
DW
L
L
L
L
L L
L L
L
L L
L L
day or weighted)
.
-
-
-
-
--
Freq., Lag
_
-
-
Freq., Lag
AR1
-
-
AR1-3
AR1-3
AR1-3
AR1-3
AR1-3
AR1-3
PLX: piecewise linear with change at x
Q: quadratic
LTX: linear, truncated above
x
A-10
-------
Table 2. STATISTICAL SIGNIFICANCE OF SLOPE ESTIMATES RELATING
MORTALITY TO BRITISH SMOKE BY YEAR*
All days
Days with BS < 500
Year
1958-59
1959-60
1960-61
1961-62
1962-63
1963-64
1964-65
1965-66
1966-67
1967-68
1968-69
1969-70
1970-71
1971-72
w/o SO? w SO? w/o SO? w SO?
+ 0 + 0
+ O 0 0
+ + + o
+ 0 + 0
+ 0 + 0
+ + + +
+
+ 0 + 0
+ o + o
+ + +
+
+ 0 + 0
00 0 0
+ + + 0
Positive, significant at two-tailed 5% level
o: Not significantly different from zero
-: Negative, significant at two-tailed 5% level.
*Autoregressive model with British Smoke, temperature, relative humidity
without S02 or with S02.
A-ll
-------
Table 3. RANDOM EFFECTS MODEL FOR DAILY MORTALITY AND BRITISH SMOKE*
Without Temp.
and Humidity
in Model
With Temp.
and Humidity
in 'Model
All
BS
BS
1 BS
< 500
< 200
B
0
0
0
.0698
.0783
.1225
T
6
9
9
.71
.55
.57
B
0
0
0
.0793
.0857
.1376
T
5
8
14
.83
.74
.64
Table 4. RANDOM EFFECTS MODEL FOR DAILY MORTALITY, BRITISH SMOKE AND S02*
Without Temp, and Humidity
BS $02
With Temp, and Humidity
BS SO?
All BS
BS < 500
BS < 200
B
0.0789
0.1044
0.1352
T
3.12
4.95
3.65
B
-0.004
-0.021
-0.012
T
-0.296
-1.72
-0.616
B
0.0609
0.0839
0.1079
T
2.72
3.73
3.17
B
0.0129
0.0025
0.112
T
1.06
0.172
0.747
*For all years controlling for year for different levels of British Smoke
A t-statistic -1.96 is statistically significant at p 0.05.
A-12
-------
STATISTICAL REANALYSES OF DATA RELATING MORTALITY
TO AIR POLLUTION DURING LONDON WINTERS 1958-1972
Joel Schwartz, Office of Policy, Planning and Evaluation
U.S. Environmental Protection Agency
Washington, DC
Allan H. Marcus, Office of Air Quality Planning and Standards,
U.S. Environmental Protection Agency
Research Triangle Park, NC*
October 10, 1986
*0n -assignment from Washington State University
A-13
-------
INTRODUCTION
This paper discusses our continuing review and analysis of the London
mortality data to assess further the degree of reliance that can be put on the
published studies in criteria development and standard setting. We have
recently reviewed three statistical analyses of the possible relationship
between daily air pollution concentrations and daily deaths in London during
the winters of 1958-1971. General conclusions about these studies (Mazumdar et
al. 1982; Ostro, 1984; and Shumway et a!., 1983) are summarized in the Criteria
Document and staff paper addenda and are discussed more fully in a separate
memorandum.
We believe that these studies have shown that a relationship exists
between mortality in London and particulate matter (measured as British Smoke)
and/or S09, and that those relationships continue below a British Smoke level
3
of 150 jjg/m . However, commentors and others have raised questions about: (1)
whether the analyses adequately handled the temporal structure of the data,
both in terms of avoiding confounding due to long-term time trends and seasonal
fluctuations, and in terms of avoiding the misestimation of the regression
standard errors (and hence significance tests) that occurs when there is
autocorrelation in the regression residuals; (2) whether the dose-response
relationship is linear or nonlinear and whether that relationship is distorted
by the techniques used to filter the series, and (3) whether it is British
Smoke, SOp or both that are responsible for the mortality.
All of the studies have attempted to deal with the autocorrelation in the
data (that is, the number of people who die on day t is correlated with the
number of people who died on day t-1, t-2, etc.). Some used deviations from
15-day moving averages to remove these autocorrelations; more recently Ostro
used an autoregressive model. All of the studies used separate regressions for
each year to remove the time trend of falling mortality. None of the models
successfully separated S02 and British Smoke effects. None of the studies
reported any tests to determine whether using separate regressions for each
year adequately dealt with the possibility of a linear time trend in the data
or reported any formal tests to determine whether they had in fact adequately
accounted for the autocorrelations in the data. In addition, while the studies
A-14
-------
clearly demonstrated relationships at low levels, and gave some indication of
their magnitude, no detailed exploratory analysis was presented to determine
the potential shape of the dose-response curve.
To address these issues, we joined with analysts at J:he California Air
Resources Board and reanalyzed the London Mortality data to evaluate the
adequacy of the fitted models. Following our examination of the underlying
data for all years and each year (discussed below), we first studied the
temporal structure of the process, particularly its autocorrelation. We
computed autocorrelation functions for the dependent and independent variables,
ran regressions that accounted for that autocorrelation, diagnosed their
residuals to assure that all of the autocorrelation had been controlled for and
re-estimated where necessary, examined plots for indications of a relationship
and clues to the shape that it might have, and examined the question of whether
we could separate the effects of the two variables (i.e. BS vs S02).
AUTOCORRELATION FUNCTIONS AND AUTOREGRESSIVE MODELS
First, we examined the autocorrelation functions for mortality, deviations
from 15-day moving average mortality, and British Smoke separately for each
year 1958-1971. This allowed us to determine whether the data was stationary
in each year, to diagnose the nature of the autocorrelation using formal
Box-Jenkins techniques, and to determine if the autocorrelations followed the
forms used in the examples of Roth et al. (1986) to illustrate the potential
for distortion of the dose-response relationship.
Our analyses indicated that in each year, the autocorrelations fell off
continuously with increasing lag, showing that the data were stationary. The
autocorrelations for both mortality and British Smoke were positive and clearly
autoregressive in nature (that is, with a gradual and continuous falling off of
the correlation between the value at time t and the value at time t-n as n
increased). The autocorrelation function for daily mortality was almost always
generated by the first two autoregressive parameters, although for a few years
one or three parameters appeared significant. The autocorrelation function for
deviations from the moving average of mortality showed greatly reduced levels
of autocorrelation. However, even after subtraction of a moving average,
enough autocorrelation remained to require an additional autoregressive parame-
ter in order to achieve stationary residuals in most years. The moving average
A-15 .
-------
term did induce some short-term cyclical fluctuations in the autocorrelation
function that were not previously there; this is consistent with Shumway's
finding that a simple 15-day moving average is a filter with short-term
oscillations.
In sum, we concluded that the previous analyses were correct in assuming
that using separate regressions for each year would achieve stationarity, and
correctly assumed that deviations from a moving average, or the use of auto-
regressive terms would reduce the autocorrelation in the model. However,
neither the use of deviations from moving average nor the use of a single
autoregressive term seems to completely eliminate the problem, and higher
autoregressive parameters are necessary to assure that there is no bias in
estimating the significance of the parameters.
At this point it is worthwhile to discuss the problems that autocorrela-
tion in the data can cause. There are cyclical patterns of increase and
decrease in both the dependent and independent variables. If those patterns
are caused by some omitted factor, including them can induce a false correla-
tion in the data or reduce a true correlation in the data. This requires the
omitted factor to be correlated with both mortality and air pollution. The
sign of the respective correlations of the omitted factor will determine the
direction of the bias. Statistically, if mortality on day t is correlated with
mortality on day t-1, etc. then the residuals of the regression of mortality
and some independent variables may also be correlated. This violates the
classical regression assumption that the errors are uncorrelated, and means
that the results of an ordinary least squares regression are not reliable.
It is only autocorrelation in the residuals, and hot autocorrelation in
the dependent or independent variables, that matters for this problem. The
issue of autocorrelation in the series inducing or masking correlations between
them is the additional problem of omitted variable bias discussed above. Yet
another problem with misspecifying the temporal structure of the model is the
misestimation of the regression parameter standard errors, and hence the
distortion of significance tests of the parameters.
We performed regressions using an autoregressive model with up to four
autoregressive parameters, which remove autocorrelation from the residuals.
The residuals of these models were analyzed by standard ARIMA techniques, the
regression models were respecified, and repeated, and once again tested to
ensure no autocorrelation was present in the residuals.
A-16
-------
These regressions were done separately for each year, for dependent
variables of daily mortality, and deviation from moving average of mortality.
In addition, they were performed in each case with and without controlling for
temperature and humidity. Both British Smoke and S02 were analyzed as indepen-
dent variables separately. This gave 112 separate final models (2 dependent
variables x 2 pollutants x with or without temperature and humidity x 14
years). In addition, having assured stationarity by performing separate
regressions for each year, the overall significance of the results for the full
data set was then assessed using a random effects model to incorporate a
between year and within year variance in estimating the overall effect. This
is described in more detail in the appendix. The results are summarized below.
For daily mortality, British Smoke was significant for 13 out of 14 years,
with or without temperature and humidity in the model. The coefficients were
generally similar to, but slightly higher than, those reported by Ostro in his
regressions using this outcome. In most cases two autoregressive parameters
rather than the one used by Ostro were necessary to completely account for the
autocorrelation of the residuals. The net effect of removing the remaining
autocorrelation that Ostro left in his model was to make British Smoke signifi-
cant for 2 more years than Ostro found, and to increase the t statistic in all
but 3 of the 14 years. This indicates that the temporal patterns in mortality
tend to mask rather than enhance the relationship with British Smoke. The
coefficients of British Smoke and their t-statistics, for models with and
without temperature and humidity, are shown in Table 1. Table 1 also includes
the coefficients and t -statistics for the Ostro regressions (which included
temperature and humidity).
Note that for both daily mortality and deviations from daily mortality,
the regression coefficients tend to increase in the later years, when pollution
levels were lower. This is consistent with the results reported by Ostro, who
3
found a higher regression coefficient below 150 than above 150 ug/m in his
spline regressions. The random effects model coefficient for British Smoke was
0.0698 without the weather terms and 0.0793 with them, with t-statistics of
6.71 and 5.83 respectively (p <0.0001). A signed rank test performed to assess
the overall significance of British Smoke across all 14 years was highly
significant (p = 0.0011), both with and without the weather factors.
For deviations from daily mortality, which was more often analyzed in
earlier analyses, an autoregressive parameter was necessary in most of the
A-17
-------
TABLE 1. DAILY MORTALITY AND BRITISH SMOKE
CONTROLLING FOR AUTOCORRELATION
Without Temperature
and Humidity
With Temperature
and Humidity
Ostro Results
Year
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
Beta
.0487
.0259
.0541
.0849
.0781
.0544
.0726
.0708
.1010
.1046
.1474
.1031
.0505
.1345
t-Statistic
7.07
2.88
3.74
4.98
5.81
3.50
3.69
2.45
3.16
2.24
3.27
2.25
1.19
2.26
Beta
.0511
.0247
.0549
.0951
.0750
.0669
.0723
.0792
.1017
.1411
.1495
.1346
.0479
.1631
t-Statistic '
7.06
2.45
3.74
5.05
5.57
4.00
3.49
2.72
2.98
2.94
3.27
2.60
1.12
2.75
Beta
.062
.028
.062
.093
.063
.065
.065
.072
.106
.227
.170
.094
.066
.061
t-Statistic
5.89
2.34
2.91
4.69
4.08
3.07
2.48
2.40
3.16
3.98
3.34
1.77
1.20
0.79
Mean Coefficient
.0808 .0898
Signed Rank Test for Overall Significance
S = 52.5 P = 0.0011 S = 52.5 P = 0.0011
Random Effects Model for All Years
B = .0698 t = 6.71 B = .0789 t = 5.83
.0881
S = 52.5 P =.0011
B = .0787 t = 5.55
models. However, after its inclusion, British Smoke was significant for all 14
years, with or without the inclusion of temperature and humidity. Six of these
years had no days with British Smoke above 500. Coefficients in our random
effects model were .0662 (t = 6.49) and .0747 (t = 5.62) without and with
temperature and humidity. Note that once autocorrelation is fully accounted
for, as in these models, the regression coefficients for using either daily
mortality or deviations from daily mortality are quite similar, as one would
expect. We conclude that British Smoke is highly significant in this data,
after fully accounting for the autocorrelation in the data, whether daily
mortality or deviations from daily mortality are used as outcomes. The coeffi-
cients of British Smoke and its t-statistic for each year for deviations in
mortality are shown in Table 2.
A-18
-------
TABLE 2. DEVIATIONS FROM DAILY MORTALITY AND BRITISH SMOKE
CONTROLLING FOR AUTOCORRELATION
Without Temperature and Humidity
With Temperature and Humidity
Year
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
Beta
.0479
.0412
.0491
.0483
.0626
.0362
.0531
.0914
.0884
.1298
.1180
.1013
.0982
.1294
t
7.39
2.77
3.43
3.47
5.15
2.56
2.85
3.64
3.20
2.62
2.92
2.29
3.02
2.51
Beta
.0490
.0178
.0529
.0538
.0709
.0527
.0568
.0899
.1055
.1451
.1468
.1130
.0929
. 1442
t
6.89
1.61
3.43
3.46
5.68
3.58
2.81
3.33
3.58
2.79
3.79
2.44
2.77
2.66
Random Effects Model for All Years
B = .0662 t = 6.49 B = .0747 t = 5.62
A similar, but not quite as strong pattern is obtained when S02 is used as
the pollutant. For daily mortality, S02 is significant for 10 out of 14 years
when temperature and humidity are not in the model, and for 11 out of 14 years
when they are included. For deviations from daily mortality, S^ is signifi-
cant for 12 out of the 14 years when temperature and humidity are not included,
and for 11 out of 14 when they are included. The coefficients for S02 and
their t - statistics are shown in Table 3. The random effects model gave
weighted coefficients of .0371 without weather terms and .0543 with them (p <
0.0001). Again a signed rank test showed a highly significant relationship
(p = 0.0011) across the years.
DIAGNOSTIC PLOTS AND FUNCTIONAL FORM
The next issue we addressed was the shape of any dose-response relation-
ship, with particular attention to low levels. To examine this, we first
plotted the data in various ways. Because pollution accounts for at most a few
percent of the mortality in London, and a similar share of its variation, to
A-19
-------
TABLE 3. DAILY MORTALITY AND S02 CONTROLLING FOR AUTOCORRELATION
Without
Year
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
Temperature
Beta
.0644
.0321
.0340
.0714
.0549
.0304
.0388
.0496
.0544
.0296
.0462
.0568
.0281
.0302
and Humidity
t-Statistic
7.25
2.73
2.30
4.71
7.58
2.40
2.84
2.50
2.85
1.20
1.90
2.57
1.41
1.04
With Temperature
Beta
.0652
.0325
.0360
.0871
.0541
.0356
.0387
.0820
.0622
.0543
.0588
.0662
.0286
.0542
and Humidity
t-Statistic
7.35
2.42
2.26
5.24
6.84
2.94
2.65
4.00
2.90
1.93
2.33
2.96
1.37
1.83
Signed Rank Test for Overall Significance
S = 52.5 P = 0.0011
Random Effects Model for all Years
B = .0371 t = 11.24
S = 52.5 P = 0.0011
B = .0543 t = 20.88
detect any curvature in the relationship it is necessary to reduce the varia-
tion somewhat by grouping the data in a fashion analogous to that used by Ware
et al. (1981) and the 1982 Criteria Document for the 1958-59 London winter
data. We examined plots for both outcomes for each year, and for all years
combined. We summarized the data in the plots by sorting the observations in
order of increasing pollutant, and taking the means of groups.
First, we ran some descriptive statistics on the frequencies of British
Smoke at different levels by year. These statistics show that even the early
years are dominated by lower levels of British Smoke. In total, only 85 out of
the 1540 days in the 14 winters had British Smoke levels above 500 pg/m3. In
12 out of the 14 years (all years except 1958 and 1959) over 90 percent of the
2
days were below 500 ug/m . When we did a cut at a lower level of British Smoke
we found that 73 percent of the days were below 200 ug/m , including more than
90 percent of the days from 1965 onward, and the majority of the days in 1961
and later.
A-20
-------
Figure 1 presents the a scatter plot of daily mortality versus British
Smoke, where each point represents the mean of 20 consecutive observations in
increasing order of British Smoke. It clearly shows a relationship starting at
the lowest observed levels, on the order of 20 ug/m , and also clearly indi-
cates that the slope of the relationship decreases at the highest levels. This
is consistent with the results of the published papers. For example, Mazumdar
found higher regression coefficients in the later years, when pollution was
low, than in the early years, when the average pollution level was much higher.
3 3
Ostro also reported a higher coefficient below 150 pg/m than above 150 ug/m
in his spline analysis of the data. The shape of the curve suggests that the
log transform used by Shumway should give a better fit than a linear regression.
A log transform always has the problem of an infinite slope at the low end.
3
However, since the lowest observed values of smoke were about 20 ug/m , this is
unlikely to have been a problem in fitting the regression, although it suggests
that the regression should not be extrapolated to values below these.
The same curvilinear shape occurs in the plot of deviations from 15 day
moving average mortality against British Smoke (Figure 2). Figure 3 depicts
the relationship for British Smoke when only days with smoke levels below 500
ug/m are included and clearly shows a relationship continuing to the lowest
levels. The curvilinear slope is not a function of some change that occurs
over time, since it occurs within individual years, as shown in Figure 4, which
plots daily mortality versus British Smoke in 1963. The continuation of the
relationship to low levels is shown for daily mortality, in Figure 5, and for
deviation from daily mortality in Figure 6. Note that these plots also provide
a second answer to the issue raised by Roth. The plots of the dose-response
relationship using deviations from moving average mortality or using mortality
as outcome measures both have a similar shape.
The curvilinear shape of the relationship may be an artifact of the
autocorrelation of the exposure variable. Since very high pollution days
generally follow high pollution days, the population of responders may have
been depleted so that it cannot respond proportionately to the very high levels
on the following day. In addition the very highest days in 1958 and 1959 (when
most of the days over 500 ug/m occurred) were accompanied by extremely low
visibility, and both logic and anecdotal evidence suggests that avertive
behavior occurred on those days. However, BS was a larger fraction of
particulate mass at higher levels of BS than at lower levels. Such a changing
A-21
-------
§
m
to
a
§
II
gin
coco
"5 CD
I.
.
To
a>
I
a
£
'2-s
a g
•g-2
OS
(A
§1
E
U.CO
10
rs
'ON
'A»J|OJJOW A|I9Q
A-22
-------
«••»
jg
1
(0
CD
Is
9 *
Ji a^
CO)
M *"
*s **
5 if)
con
52
SB
iations in Daily Mortal
London Winte
S
a
'
,
0
—
—
-
•
9
a
»
9
• ^
•
. •
•
•
• • ""
• •
• •
* * » • •
*••••*• ""
• • O * €
•I ' 8. J ."II-*
• • • *
I ii ii iiii i> ' —
o
1
o
o
*•
o
o
*~
o
o
o
§
0)
o
o
00
,E
O O)
^5 «i
•g
OT
o -g
« 1
03
o
S
o
o
0
CO
o
o
o
o
o
e
c
c
I-
11
CM'0
o? i
T5
? S.
1_ £
O Tn
M ^
Ii
for the wi
Mote the h
moke for 20 adjacent va
**.
— ii
s
O <0
|i
.2*-o
CO C
•0 «
•8S
c "ra
.2 t
*• o
•IE
sl
*3
1!
minOtOOIOOlOO ID O 1O
§ « CO « N t- t- .^.^.
Aea-S I UJQJJ) Ajt|»wow AHBQ ui
A-23
-------
V)
CO
o
°csl
*t%
II
8
tf CO
S-in
e>o>
jer-
|&
Is
si
•f^f *^^
.t-1
a
. •
at
British S
o
CO
•ON
in
S
/-,
s
•a
t.s
(0 >.
2|
Si
El
» o
€ «
ch point r
ote the re
51
c4
8!
J*
*>T3
is
o o
COCO
ii
li
3 "5
il
fl
2 c
is
££
A-24
-------
w o g o S §
"> CM o E; Jg SI 2
CO CO CO CM CM CM CM
o
I
tqieap -ou |B;OJ '
A1IVQ
A-25
-------
e>
^2
t
f £
«r
i *
ily Mortality -
London Winl
e
Q
-
IU
Z
1
1
« £
o
Is _
o **
IU •
e •
u. ;
M .
• (0 "
O ID
CD .
f ^
o «< __
IU
O
X
Z
«"
(0
Z
1 1 1 1 1 1 1 1 1 1 1 !
1
IO
r-
o
in
T-
IO
01
1
§
LJ
100
RITISH SMOK
CD
IO
fs
O
ID
IO
CM
O
i S88sBIIE8B58
«
a>
•B
1
UJ
en
oo
If
•§'
I-
b..
£:
g>
•g
•
18
OQ ffil
>£
= OJ I
ID *; I
Q.E|
f qjeep -ou |B»oj 'AIPVIUOW A1IVQ
A-26
-------
e
o
tn
£
.2
™ O>
n c
2|
•«E «^
•9
0
•-B
1
o
flD
o
<
IU
X
• z
-
.
1 II 1 1 1 1
O IO O ID O IO O.IOC
oKlOdiOfxIONC
^^^
^~
s
IU
E
ff
O
g
u.
O
Z S
§N
i *
o a
B > ^^
? ^7
—
•
N
ID
ii i i
S ID O ID O
» N IB* K 0
O
o c
CM g
E
|
1
10 0
t E
1
CD
I
•c
c
a
g)
o E
IO e
*
i
D.
E
ID £ -
N 0«
? ctE
1 -s g
C (o *t
^» "g
* eo*"
s s§
§ i- Jl
«- w S-S-
x ?|
g -si
t »- •"-
c *•!
a c £
"ite
R l=^
•N
0) ai
JC S
o o
iz.
-C 81
•5.x
*: o
g ml
3»
L. ^*
0) "Z
>OD
.^•s
2w
o|
B EI
«M e.£
o«
7 C
co-t
'> d
o a.
Qo
CD a
£ o
0 »•=
O) » /*
" .— w
U. CD
(B6BJGAV BUIAOW ABQ-SI UIDJJ) AinVlHOW AllVd Nl SNOI1VIA3Q
A-27
-------
relationship would transform a linear relationship between participates and
mortality into a concave relationship with British Smoke, such as is observed.
However, no definitive conclusions as to what caused this curvilinear relation-
ship can yet be made. While examination of the plots suggests that a log or
fractional power transformation would fit the data better, that is unnecessary
to establish the significance of the correlation. At lower, levels, a linear
approximation appears to fit almost as well, and we have continued our use of
linear models in our subsequent analyses.
LOW LEVEL EFFECTS
Our plots indicated that "hockeystick" regressions would not be appropri-
ate for examining the relationship at lower levels, and that if anything, a
higher slope was expected. To examine this quantitatively, we reran all of our
regression models using only days when British Smoke was less than 500 pg/m .
Since the exclusion reduces the sample size and automatically reduces signifi-
cance levels, we were more interested in what happened to the slopes than in
what happens to the p values.
For daily mortality we found that British Smoke was significant for 12 out
of the 14 years, with or without temperature and humidity in the model. SOp
was significant for 10 out of the 14 years without including temperature and
humidity, but for 11 of the years when they were included. For deviations from
daily mortality, British Smoke was significant for 11 out of 14 years without
including temperature and humidity and for 12 out of 14 years when they were
included. S02 was significant for 9 out of the 14 years without temperature
and humidity corrections, and for 11 out of the 14 years when they were includ-
o
ed. However, excluding days over 500 pg/m reduced the sample size by 40
percent for 1958, giving little statistical power, so that year should probably
be excluded from consideration at these levels (which would leave British Smoke
significant in 12 out of 13 years for daily mortality and 11 out of 13 for
deviations from daily mortality).
Table 4 shows the coefficients of British Smoke and their t - statistics.
o
These show that even with a cutoff of 500 |jg/m , the coefficients tend to
increase in later years when the pollution levels were lower. This is consis-
tent with Figure 1, which shows a higher slope at lower levels, when only
3 3
looking at data below 500 ug/m . The exclusion of the days over 500 ug/m
A-28
-------
TABLE 4. DAILY MORTALITY AND BRITISH SMOKE (SMOKE < 500 ug/m3 ONLY)
CONTROLLING FOR AUTOCORRELATION
Without
Year
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
Temperature
Beta
.0774
.0399
.0628
. 1080
.1366
.0476
.0680
.0708
.1010
.1046
.1474
.1032
.0505
.1345
and Humidity
t-Statistic
2.18
1.56
2.70
4.62
4.23
2.50
3.08
2.45
3.16
2.24
3.27
2.24
1.19
2.26
With Temperature
Beta
.0836
.0375
.0604
.1192
.1264
.0550
.0667
.0792
.1017
.1411
.1495
. 1346
.0479
.1631
and Humidity
t- Statistic
3.47
1.37
2.39
4.96
3.94
2.91
2.82
2.72
2.98
2.94
3.27
2.60
1.12
2.75
Random Effects Model for All Years
B = .0783 t = 9.55
B = .0857 t = 8.74
increases the coefficients in the early years from the values they had when all
days were included, which again indicates that the higher slope at lower levels
occurs within each year as well as between them and, therefore, is not simply
due to some other time trend. Also, note that, even in the years when British
Smoke was insignificant, its coefficient was always within the range of the
years when it was significant, indicating that the lack of significance was due
more to a higher variance in that year, and that a consistent pattern of effect
was being seen.
When the random effects model was used, the mean coefficient for British
Smoke was .0783 (t = 9.55, p < 0.0001) without weather terms and .0857 (t =
8.74 p < 0.0001) with those terms. Note that despite the reduction in sample
size, the overall model shows that British Smoke is more significant when
restricted to days with pollution less than 500 pg/m than when the higher days
are included. The smaller sample size of the individual winters masks this
strength in the individual statistics.
To investigate the low level effects further, the model was rerun using
Q
only those days when British Smoke was less than 200 ug/m . While the reduced
sample size within each year meant that British Smoke was only significant for
A-29
-------
6 out of the 11 years with sufficient data to run the analyses (8 out of the 11
years if one-tailed tests are used), the coefficients were stable, suggesting
that using the full data set (which has enough days so that the p-value is not
dominated by small sample size) would give a different result. In fact, the
coefficient of British Smoke in the random effects model is 0.1225 (t = 9.57,
p < 0.0001) without weather terms and 0.1376 (t = 14.64 p < 0.0001) with
temperature and humidity. Thus the overall relationship between British Smoke
and mortality is stronger if the data are restricted to only those days when
smoke was less than 200 ug/m . These results are shown in Table 5.
TABLE 5. DAILY MORTALITY AND BRITISH SMOKE (SMOKE < 200 ug/m3 ONLY)
CONTROLLING FOR AUTOCORRELATION
With
Year
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
Temperature
Beta
.1198
.1939
.0436
.1229
.1428
.1436
.1382
.1721
.0786
.0624
.0826
and Humidity
t-Statistic
1.76
2.88
0.81
2.28
2.68
3.08
2.33
3.75
1.28
1.21
1.18
Without
Beta
.1207
.1509
.0798
.1242
.1476
.1638
.1861
.1705
.1062
.0455
.1192
Temperature and Humidity
t-Statistic
1.60
2.23
1.51
2.14
2.80
3.28
3.08
3.71
1.67
0.88
1.70
Random Effects Model for All Years
B = .1225 t = 9.57
B = .1376 t = 14.64
To further investigate the relationship between British Smoke and
mortality we looked at the years 1965-1972, for smoke levels <200 pg/m3. This
restricts us to a range of pollution similar to that in the United States
today, and a period in London when the sources of the particulate matter were
also likely closer to United States sources.
We used a nonlinear regression incorporating all seven years together,
with dummy variables for each year, temperature, humidity, and autoregressive
parameters. British Smoke was significantly related to mortality (p = 0.138,
t = 7.67) with a coefficient that was identical to the one found in our random
effects model of the year by year coefficients.
A-30
-------
SEPARATING BRITISH SMOKE FROM S02
The high degree of collinearity between British Smoke and S02 makes it
difficult to distinguish between them. Nevertheless we felt that it was
important to try, in order to see what could be learned. Reviewing the data
above, where only one pollutant was used in the regressions, note that British
Smoke was consistently significant more often than S02, both in the regressions
using temperature and humidity and in those omitting the weather variables, and
for both daily mortality and deviations from daily mortality. To examine this
further we repeated the above regression for all the ranges of pollution, for
inclusion or exclusion of weather variables, and for all years, using both
British Smoke and S02 in the model. We looked at which pollutant achieved
significance and at the stability of the regression coefficients.
For all British Smoke levels, when both pollutants were in the model,
British Smoke was a significant predictor of daily mortality for 6 out of the
14 years with or without the weather factors in the model. S02 was significant
for 3 of the years excluding temperature and humidity, and for 2 of the years
if they were included. We then only considered those years when the correla-
tion between the S02 coefficient and the Smoke coefficient was less than 0.9.
These years have less collinearity and therefore, allow a better chance to
distinguish between effects associated with the different variables. This
criteria was met for 6 years for the daily mortality regressions that excluded
weather. British Smoke was significant for 4 of those years, but S02 was not
significant for any of them. The criterion was met for 7 years when the
weather factors were included; and British Smoke was again significant for 4 of
the years, but S02 was significant for only 1 year.
When we restricted our models to those days with British Smoke less than
500 ug/m (Table 6), British Smoke was significant for 7 years and S02 for none
of the years when temperature and humidity were excluded. When they were in-
cluded, British Smoke was significant for 5 years and, again, S02 was always
nonsignificant. Without temperature and humidity, there were 10 years when the
correlation of the regression coefficients were less than 0.9, and British
Smoke was significant for 5 of them. There were 9 such years when temperature
and humidity were included in the model, and British Smoke was significant for
3 of them.
When we looked at the stability of the coefficients, an even stronger
story emerged. Using the random effects model, British Smoke was highly
A-31
-------
TABLE 6. DAILY MORTALITY, BRITISH SMOKE AND S02 ONLY DAYS
WITH BRITISH SMOKE < 500 |jg/m3 CONTROLLING FOR AUTOCORRELATION
WITHOUT TEMP AND HUMIDITY
Year
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
BS
.0646
.0876
.0652
.1691
.0950
.1173
• .1357
.0299
.1071
.2085
.2147
-.0462
-.0030
.2699
t
1.28
2.08
1.44
2.83
1.50
2.93
2.34
.478
1.36
2.36
2.84
-.375
-.036
2.40
S02
.0260
-.0638
-.0028
-.0482
.0345
-.0622
-.0467
. 0315
-.0040
-.0628
-.0436
.0773
.0294
-.0769
t
.367
-1.43
-.061
-1.07
.761
-1.95
-1.26
.741
-.086
-1,37
-1.11
1.30
.745
-1.40
Random Effects Model for all Years
BS = .1044 t = 4.95 S02 = -.0212 t = -1.76
significant (B = 0.0789, T = 3.12). Using daily mortality as the outcome, for
British Smoke at all levels, the mean BS coefficient with S02 in the model was
0.089, compared to a mean of 0.081 when only British Smoke was in the model.
The impact of adding SOp to the model was only to change the variation about
that mean, with the coefficient of variation increasing from 43 percent to 105
percent. A sign rank test of the overall significance of British Smoke over
the full 14 year period was still significant (p = 0.0011). By contrast, the
mean value of the S02 coefficient became negative in the joint models, and its
coefficient of variation increased to 750 percent. The sign rank test for an
overall effect was highly insignificant (p = 0.66).
When temperature and humidity were included in the models, the mean value
of the British Smoke coefficient was 0.0737, with a coefficient of variation of
119 percent and signed rank test p-value of 0.0144. This compares to a mean of
0.090 and COV of 47 percent without S02 in the model. For S02, the mean
coefficient was 0.0111 with a coefficient of variation of 440 percent and a
signed rank test p-value of 0.47. This compared to a mean of 0.054, COV of 34
percent, and overall significance level of p = 0.0011 without British Smoke in
the model. In the random effects model, British Smoke was again significant
(B = 0.0609, t = 2.72) and S02 nonsignificant.
A-32
-------
When both pollutants were included in models for only those days when
•3
British Smoke was less than 200 M9/m (Table 7), and analyzed in the random
effects model, the coefficient of British Smoke was 0.1352 (t = 3.65 p = .0019)
without temperature and humidity terms, and 0.1079 (t = 3.17 p = 0.0045) with
those terms. S02 was highly nonsignificant in both cases. This compares to
the equivalent coefficients in models excluding S02 of 0.1225 and 0.1376
respectively, as noted previously. Again, the coefficient of British Smoke is
little changed by the addition of S02 to the model.
TABLE 7. DAILY MORTALITY. BRITISH SMOKE AND S02 FOR ONLY DAYS
WITH SMOKE < 200 \ig/m& CONTROLLING FOR AUTOCORRELATION
WITHOUT TEMP AND HUMIDITY
Year
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
BS
.0485
.1006
.1215
.1274
.1850
.1683
. 2517
.2472
-.1879
.0225
.2070
t
.417
.883
1.52
1.30
1.97
1.64
2.54
3.27
-1.19
.247
1.65
S02
.0588
.0583
-.0723
-.0034
-.0277
-.0135
-.0687
-.0483
.1278
.0214
-.0648
t
.754
1.03
-1.31
-.056
-.538
-.271
-1.40
-1.26
1.81
.534
-1.18
Random Effects Model for All Years
B = .1352 t = 3.65 B = -.0117 t = -.616
The stability of the British Smoke coefficient to the addition of S02 to
the model, the fact that it remains significant in about half of the individual
years and in the analysis of all years in contrast to the instability of S02,
and the general nonsignificance of S02 both in individual years and overall,
suggests different conclusions about the two variables. The multiple regres-
sion with both factors included looks for significance just for that portion of
British Smoke and S02 that vary independently of each other. For British
Smoke, we find that the coefficient, so restricted, is the same as when all of
its variance is considered (in the regressions with only one pollutant).
Moreover, this effect was statistically significant over the whole data set.
S0?, in contrast, had a different mean coefficient when only its variation that
was independent of smoke was considered, and that coefficient was not signifi-
cantly different than zero. This suggests that the significance of S02 in
A-33
-------
separate regressions may only be due to its collinearity with British Smoke,
whereas British Smoke is significant, with the same magnitude effect, whether
or not its covariance with S02 is included. Because of the high degree of
collinearity, this cannot be used to exclude the possibility of an independent
S02 effect; however, we feel that it is good grounds for concluding that
British Smoke is significantly correlated with mortality independent of S02.
The year by year analysis serves an additional purpose besides
controlling for possible linear trends in the data. It also serves as a
partial control for omitted variable bias. Any epidemiological study always
faces the issue of the omitted confounding variable. Given that British smoke
is significant in 13 out of 14 years taken individually means any omitted
factor accidentally linking particulate matter and mortality cannot be a
happenstance but must be a long-term systematic factor. It is of course
possible to imagine such factors, such as weather, which was controlled for,
but perhaps imperfectly. However what makes this relationship so impressive
is that it was so stable across a period of 14 years when the nature of air
pollution in London was changing drastically.
In the 1950's and early 1960's particulate matter was dominated by the
open hearth burning of coal, which was banned by the Clean Air Act of 1963.
The growth in diesel bus and truck traffic in London combined with the fall in
open combustion substantially changed the source and weather-sensitive nature
of the particulates. For example in the first 4 years (1958-1961) the average
correlation coefficient of British Smoke with temperature was -.300 and with
relative humidity was +.325. In the last 4 years of our data they had fallen
to -.188 and +.084 respectively. Given that when we restrict our regressions
to the linear end of the dose-response curve (BS <200 ug/m3) the coefficients
are stable from the beginning years to the end ones, while the sources of
particulate matter and their relationship to weather change significantly, such
omitted variable bias seems unlikely.
REFERENCES
Mazumdar, S.; Schimmel, H.; Higgins, I. (1981) Daily mortality, smoke and S02
in London, England 1959 to 1972. In: A specialty conference on the proposed
SO and particulate standard; September 1980; Atlanta, GA. Pittsburgh, PA:
Air Pollution Control Association; pp. 219-239.
Ostro, B. (1984) A search for a threshold in the relationship of air pollution
to mortality: a reanalysis of data on London winters. EHP Environ. Health
Perspect. 58: 397-399.
A-34
-------
Roth, H. D.; Wyzga, R. E.; Hayter, A. J. (1986) Methods and problems in esti-
mating health risks from particulates in aerosols (Lee, S. 0.; Schneider,
T; Grant, L. D.; Verkerk, P. J. eds). Lewis Publishers, Chelsea, MI, pp.
837-957.
Shumway, R. H.; Tai, R. Y.; Tai, L. P.; Pawitan, Y. (1983) Statistical analysis
of daily London mortality and associated weather and pollution effects.
Sacramento, CA: California Air Resources Board; contract no. Al-154-33.
A-35
-------
APPENDIX
RANDOM EFFECTS MODEL FOR ESTIMATING AN OVERALL RELATIONSHIP BETWEEN POLLUTION
AND MORTALITY FOR ALL YEARS
We assume a model where;
b.j is estimated separately for each year, with variance V.
and residual e.
the random effects model is that
b.j = B + e^ +er where er is a random variance
component with variance Vr
then we can estimate Vr by
and
where
Vr = [Z (bi - bavg)2]/(k -1) - Z
B = Z w-t^./Z w.
i + VR)
and
se of B = [Z W..]-0.5]
*U.S. GOVERNMENT PRINTING OFFICE : 1987-748-121/40694
A-36
-------
-------
-------
-------
ro
C
o
o
OD
i
00
cn
o
ro
CD
~n
3- o O
HI
CO o £0
3 -B
° =
a o
o, X
ft?
i-S S
to 3
-i CO
fl) cn
i I
D •<
55-
5§
CD CO
-, 2.
CO
0)
g
D
O
'
O
I
CJl
ro
en
oo
------- |