Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Ambient air pollution and lung function in children
(USC Thesis Other)
Ambient air pollution and lung function in children
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
AMBIENT AIR POLLUTION AND LUNG FUNCTION IN CHILDREN
by
Robert Urman
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(EPIDEMIOLOGY)
August 2015
Copyright 2015 Robert Urman
Acknowledgments
I would like to express my sincerest gratitude to Dr. Rob McConnell for his invaluable
mentorship and for the countless hours he spent reading over my dissertation and
providing me constructive feedback. I would also like to thank my remaining committee
members (Drs. Jim Gauderman, Frank Gilliland, Scott Fruin, and John Wilson) for their
invaluable guidance as I progressed through my graduate studies.
I would also like to acknowledge the contributions of several colleagues of mine. I would
like to thank Dr. Rima Habre for sharing her wealth of knowledge on exposure
assessment. I would like to thank Drs. Talat Islam and Towhid Salam for the professional
advice they have given me. I would like to thank Hita Vora and Feifei Liu for their
statistical and programming assistance. I would like to thank Ed Rappaport for his
dedication in maintaining all of the data sets related to the Children’s Health Study.
Finally, I would also like to thank my family and friends for their continued support and
encouragement.
ii
Table of Contents
Acknowledgments............................................................................................................... ii
List of Tables ...................................................................................................................... v
List of Figures ................................................................................................................... vii
CHAPTER 1: Introduction and Background ...................................................................... 1
1.1 OVERVIEW.............................................................................................................. 1
1.2 BRIEF REVIEW OF REGIONAL POLLUTION EFFECTS ON LUNG
FUNCTION ..................................................................................................................... 3
1.3 NEAR-ROADWAY AIR POLLUTION .................................................................. 6
1.4 PARTICULATE MATTER AND COMPOSITION ................................................ 9
1.5 POSSIBLE BIOLOGICAL MECHANISMS INVOLVED IN AIR POLLUTION
EFFECTS ON THE LUNGS ........................................................................................ 11
1.6 PROPOSED DISSERTATION ............................................................................... 12
1.7 REFERENCES ........................................................................................................ 16
CHAPTER 2: Associations of Children’s Lung Function with Ambient Air Pollution:
Joint Effects of Regional and Near-roadway Pollutants ................................................... 20
2.1 ABSTRACT ............................................................................................................ 20
2.2 INTRODUCTION ................................................................................................... 21
2.3 MATERIALS AND METHODS ............................................................................ 21
2.4 RESULTS ................................................................................................................ 24
2.5 DISCUSSION ......................................................................................................... 28
2.6 TABLES AND FIGURES ...................................................................................... 33
2.7 SUPPLEMENTARY MATERIAL ......................................................................... 43
2.8 REFERENCES ........................................................................................................ 55
CHAPTER 3: Determinants of the Spatial Distributions of Elemental Carbon and
Particulate Matter in Eight Southern Californian Communities ....................................... 58
3.1 ABSTRACT ............................................................................................................ 58
3.2 INTRODUCTION ................................................................................................... 59
3.3 MATERIALS AND METHODS ............................................................................ 61
3.4 RESULTS ................................................................................................................ 67
3.5 DISCUSSION ......................................................................................................... 71
3.6 TABLES AND FIGURES ...................................................................................... 79
3.7 SUPPLEMENTARY MATERIAL ......................................................................... 91
3.8 REFERENCES ...................................................................................................... 106
iii
CHAPTER 4: Associations between lung function and ambient transition metals
exposure among children in the Southern California Children’s Health Study .............. 109
4.1 ABSTRACT .......................................................................................................... 109
4.2 INTRODUCTION ................................................................................................. 110
4.3 MATERIALS AND METHODS .......................................................................... 111
4.4 RESULTS .............................................................................................................. 115
4.5 DISCUSSION ....................................................................................................... 118
4.6 TABLES AND FIGURES .................................................................................... 124
4.7 SUPPLEMENTARY MATERIAL ....................................................................... 131
4.8 REFERENCES ...................................................................................................... 134
CHAPTER 5: Summary and suggestions for future research ......................................... 137
5.1 SUMMARY AND DISCUSSION ........................................................................ 138
5.2 SUGGESTIONS FOR FUTURE RESEARCH .................................................... 144
5.3 REFERENCES ...................................................................................................... 148
iv
List of Tables
Table 2.1. Characteristics of 1,811 CHS participants with lung function testing. ........ 33
Table 2.2. Effects of measures of near-roadway air pollution on lung function level. . 34
Table 2.3. Sensitivity analysis for lung function effects of near-roadway residential
NO
x
. ............................................................................................................................... 35
Table 2.4. Effect of averaged regional pollutants on lung function level. .................... 36
Table 2.5. Joint analysis of regional air pollution and near-roadway NO
x
on lung
function. ......................................................................................................................... 37
Supplemental Table 2.1. Characteristics of participants and non-participants in lung
function testing. ............................................................................................................. 50
Supplemental Table 2.2. Distribution of residential distances to freeways and other
major roads for CHS participants in 8 study communities. .......................................... 52
Supplemental Table 2.3. Correlation of regional pollutants from central sites. ............ 53
Supplemental Table 2.4. Correlation of near-roadway predicted exposures. ............... 54
Table 3.1. Eight-week geometric mean concentration of measured pollutants (in µg/m
3
)
and coefficient of variation (CV in %) in each community. ......................................... 79
Table 3.3. Pairwise correlations between deviated (community-centered) pollutants
levels and potential predictors. ...................................................................................... 81
Table 3.4. Prediction models across all eight communities
a
. ........................................ 82
Table 3.5. Leave-one-out cross-validated (LOOCV) R
2
for prediction models in Table
3.4 applied to each community. .................................................................................... 83
Table 3.6. Community specific EC
2.5
models. .............................................................. 84
Table 3.7. Community specific EC
0.2
models (reported betas in each column followed
by R
2
for each community). .......................................................................................... 85
Table 3.8. Leave-one-out cross-validated (LOOCV) R
2
of various hierarchical
combined models. .......................................................................................................... 86
Supplemental Table 3.1: Pearson correlation
a
of eight-week averaged levels of
measured pollutants. ...................................................................................................... 97
Supplemental Table 3.2. Pairwise correlation of EC
2.5
a
with traffic and other land-use
predictors (by community). ........................................................................................... 98
Supplemental Table 3.3. Pairwise correlation of EC
0.2
a
with traffic and other land-use
predictors (by community). ......................................................................................... 100
Supplemental Table 3.4. EC prediction models for five communities with high R
2
in
the 8-community model. ............................................................................................. 102
Supplemental Table 3.5. The relative variability of eight-week mean concentrations of
measured pollutants in each community expressed as range/mean (in %). ................ 103
Table 4.1. Description of study population. ................................................................ 124
v
Table 4.2. Descriptive statistics of measured metals (in ng/m
3
) and PM mass (in
μg/ m
3
). ......................................................................................................................... 125
Table 4.3. Associations of metals and PM with FEV
1
. ............................................... 126
Table 4.4. Associations of metals and PM with FVC. ................................................ 127
Table 4.5. Sensitivity analysis for water-soluble fine Ni. ........................................... 128
Table 4.6. Associations of metals with FEV
1
adjusted for mass................................. 129
Table 4.7. Associations of metals with FVC adjusted for mass. ................................. 130
vi
List of Figures
Figure 2.1. Distribution of predicted local (A) NO, (B) NO
2
, and (C) NO
x
within each
of the eight study communities based on a spatial land-use regression model. ............ 38
Figure 2.2. Associations of local NO
x
with (A) FEV
1
and (B) FVC within each study
community. .................................................................................................................... 39
Figure 2.3. Adjusted average FEV
1
versus 2002-2007 community-average pollutant
levels. ............................................................................................................................. 40
Figure 2.4. Adjusted average FVC versus 2002-2007 community-average pollutant
levels. ............................................................................................................................. 41
Figure 2.5. Joint effect of regional PM
2.5
and NRAP on FEV
1
. .................................... 42
Figure 3.1. Map of CHS communities. ......................................................................... 87
Figure 3.2. Distribution of eight-week averaged concentrations of EC and PM in 2.5
and 0.2 µm size fractions. ............................................................................................. 88
Figure 3.3. Distribution of selected predictor variables by community. ....................... 89
Supplemental Figure 3.1. Scatter plot of community specific estimates of total
CALINE4 from EC
2.5
model as a function of average 8 week NO
x
concentrations from
fixed monitoring stations. ............................................................................................ 104
Supplemental Figure 3.2. Scatter plot of community specific estimates of total
CALINE4 from EC
0.2
model as a function of average distance to shoreline. ............. 105
Supplemental Figure 4.1. Distribution of total metal concentrations (in ng/m
3
). ....... 131
Supplemental Figure 4.2. Distribution of water-soluble metal concentrations (in
ng/m
3
). ......................................................................................................................... 132
Supplemental Figure 4.3. Distribution of water-insoluble metal concentrations (in
ng/m
3
). ......................................................................................................................... 133
vii
CHAPTER 1: Introduction and Background
1.1 OVERVIEW
The effect of air pollution on human health has long been a concern. One of the
more exemplary cases in recent history of the dangers of air pollution was the London
Smog of 1952. During this five day event that took place from December 5
th
until
December 9
th
, a thick layer of fog, comprising mostly sulfurous smoke, blanketed the
Greater London region. Recent studies have attributed about 12,000 deaths to this event
and the couple of months that followed (1). Since this event, numerous published
epidemiological studies have provided a wealth of information that would indicate that
air pollution, even at concentrations that are considered relatively low by many of today’s
standards, can have a wide array of health impacts. Extensive epidemiological studies of
air pollution have shown associations with a number of outcomes including respiratory
and cardiovascular diseases, cognitive function, stroke, cancer, and death (2, 3). The aim
of this dissertation is to explore the negative impact of probable causal pollutants or
pollutant mixtures on respiratory health, with the focus mainly on the impacts of local or
near-roadway air pollution and that of elemental composition of particulate matter on
lung function in children and adolescence. In urbanized locations, such as the Southern
California communities that constitute the Children’s Health Study (CHS) and from
which the population at focus in this dissertation is derived, motorized vehicles contribute
a significantly large amount of the air pollution at both the local and regional scales (2).
However, other non-traffic sources also emit a sizeable amount of pollutants into the air
in Southern California (4-6).
1
Many studies have been conducted that have studied the impacts of regional air
pollution in children. Regional air pollution is assessed at fixed-site monitoring stations
that are often times situated away from heavy pollution sources in order to capture
background air pollution levels. High levels of regional air pollution has been linked with
increased acute respiratory illness, increased hospital admissions due to asthma
exacerbation, increased respiratory symptoms such as cough and phlegm, and decreases
in lung function (7). The testing of one’s lung function is done to determine the health of
the lungs or to see how it responds to some stimuli. Lung function testing, or pulmonary
function testing, is also done to help diagnose asthma in children and chronic obstructive
pulmonary disease in adults. Lung function testing can include a number of parameters
that measure different aspects of the lungs. Forced vital capacity (FVC) is the total
volume of air that is forcibly expired and is used a measure of the size of the lungs.
Forced expiratory volume in 1 second (FEV
1
) is the total amount of air that is forcibly
expired within the first second and is used to measure obstruction in the airways. Other
measures of lung function include maximum midexpiratory flow (MMEF) which is the
average expiratory flow over the middle half of the FVC (also referred to as FEF
25-75%
),
and peak expiratory flow (PEF), which is the maximum flow during expiration.
Lung function is an important outcome because studies have shown that low lung
function in children is associated with increased development of asthma (8).
Additionally, low lung function in children may have long lasting impacts as some
studies have shown that in adulthood, low lung function is predictive of a number of
diseases including coronary artery disease, chronic obstructive pulmonary disease and
mortality (9-13). Thus, understanding how air pollution affects lung function in children
2
may be critical for preventing subsequent diseases later in life. In addition, children are
more likely to be susceptible to the effects of air pollution compared to adults for a
number of reasons including that their lungs are still in development, their lungs have a
higher surface area to volume ratio compared to adults, and they are more likely to
participate in outdoor activities which lead to higher exposure to air pollutants due to
increased ventilation (14).
1.2 BRIEF REVIEW OF REGIONAL POLLUTION EFFECTS ON LUNG FUNCTION
While not entirely consistent, studies have shown detrimental effects of regional
air pollution and lung function in children. Most studies have been cross-sectional
designs. For example, as part of the Second National Health and Nutrition Examination
Survey (NHANES II), Schwartz (15) reported significant negative associations between
FVC, FEV
1
, and peak expiratory flow and annual concentrations nitrogen dioxide (NO
2
),
ozone (O
3
), and total suspended particles among children and young adults between 6 and
24 years of age. Raizenne et al. (16) examined the association between lung function of
children between 8 and 12 years of age and particle strong acidity, sulfate particles,
particulate matter with aerodynamic diameter less than 2.1 µm and 10 µm (PM
2.1
and
PM
10
) in 22 communities in the United States and Canada. This study found negative
associations with a number of lung function measures, with the strongest associations
being with particle strong acidity. Another study found that among University of
California, Berkeley students who were lifelong residents of either Los Angeles or the
San Francisco Bay areas, those who had been exposed to high levels of lifetime ambient
levels of O
3
had decreased levels of FEF
75%
and FEF
25-75%
(markers of small airway
3
function) (17). Similar associations were also observed with PM
10
and NO
2
. However, a
cross-sectional study from the Six Cities Study of Air Pollution and Health showed no
associations between various measures of lung function and air pollution among children
between 10 and 12 years of age (18). Despite the null findings with lung function, there
were associations between air pollution and respiratory symptoms among children with a
history of wheeze or asthma. In a series of longitudinal studies evaluating lung function
growth among Austrian children, medium-term summer time O
3
and PM
10
were initially
associated with lower lung function, but over a longer averaging period of 3.5 years, the
effect of O
3
was no longer present (19-21).
Findings from the CHS
Cross-sectional and longitudinal epidemiological studies from the CHS have also
shown the harmful effects of regional air pollution on lung development. The CHS is one
of the few prospective studies of regional air pollution and children’s lung function
growth. Participants were selected from communities in the Southern California region in
order to capture the large between community contrasts between regional NO
2
, O
3
,
ambient PM
10
, and strong acid (22). Beginning in 1993, over 11,000 children across five
separate cohorts have been recruited though public schools.
The first study from the CHS was a cross-sectional analysis that examined the
relationship between a number of collected pollutants from fixed-site monitoring stations
and attained lung function level in children (22). This study, which looked at children
from different grades levels (fourth grade, seventh grade, and tenth grade), showed
inverse associations of O
3
, PM
10,
PM
2.5
, NO
2
, and acid vapor with measures of lung
4
function (FVC, FEV
1
, MMEF, and PEF).The reported associations were stronger among
girls than boys and among girls who indicated spending more time outdoors than indoors.
In a follow-up study among this same group of children, lung function growth was
examined. The results from this second study showed that exposure to higher
concentrations of PM
10
, PM
2.5
, NO
2
, and inorganic acid vapor were associated with lower
four-year lung function growth only among the fourth grade cohort (23). An analysis of
the fourth grade cohort with follow-up through high school graduation at age 18 showed
that higher concentrations of regional NO
2
, acid vapor, PM
10
, PM
2.5
, and elemental
carbon were also associated with lower 8-year growth (24). In a replication study,
analysis of a second cohort of fourth graders found lower four-year lung function growth
with increasing levels of exposure to PM
2.5
, NO
2
, acid vapors, and elemental carbon (14).
While exposure to high levels of air pollution has been demonstrated repeatedly to impair
lung function in this group of children, there is also some evidence from the CHS that
decreasing exposure to air pollution could reverse these effects. Following a small group
of children from the seventh and tenth grade cohorts who had moved away from the
communities in which they were recruited, those children who had moved to
communities with lower PM
10
concentrations showed some improvement in lung
function (25). Overall levels of pollution have decreased over the past 20 years (26).
Additional evidence that air pollution causes lung function detriments is provided by a
recent study which showed that children from a cohort recruited in 2002-2003 (described
in more detail later in this dissertation) had larger lung growth between the ages of 11 and
15 compared to two older cohorts of children who were generally breathing in more
polluted air (27). Together the CHS studies are consistent with a causal relationship
5
between a package of correlated regional pollutants and deficits in lung function and lung
function growth over the course of adolescence.
1.3 NEAR-ROADWAY AIR POLLUTION
The study of near-roadway air pollution is an emerging topic in the literature
because of potentially different toxicological effects compared to regional pollutants.
Many prior studies have referred to these near-roadway exposures as “traffic exposures.”
However, motor vehicles or traffic contribute to both local air pollution (on a scale of up
to a few hundreds of meters from roadways) and regional air pollution covering an entire
community fairly uniformly. Regional pollutants collected at fixed-site monitoring
stations are considered aged or secondary pollutants as they are formed as a result of
photochemistry in the atmosphere such as is the case with ozone. Near-roadway air
pollutants emitted directly from vehicles’ tailpipes are often classified as fresh or primary
emissions and are likely to differ in composition (and therefore potentially in health
effects) from regional pollutants. For clarification, the mixture of local “traffic-related
pollutants” will be referred to as near-roadway air pollutants (NRAP) to distinguish the
local scale from traffic’s contribution to regional air pollution. A challenge in linking
near-roadway air pollutants with any health outcome is that motor vehicles emit a large
number of pollutants, which include NO, CO, elemental carbon (black carbon or soot),
organic carbon, and other particulate matter. Motor vehicles can also produce non-
combustible air pollution through brake wear, tire wear and re-suspension of road
particles (28). Many of the particles generated by vehicles can span various size fractions
and which may each have different effects on human health (29).
6
Methods to assess near-roadway air pollution
Due to the complex mixture of air pollutants that vehicles produce, it is not
feasible to directly measure them all. To characterize the impacts of local traffic in
epidemiological studies, a number of methods have been previously been employed.
Some epidemiological studies have used roadway proximity as a surrogate for near-
roadway air pollution. This is a viable option as many of these pollutants show rapid
decay for the first few hundred meters away from a road before leveling off to
background levels (30, 31). Other measures of local traffic have been used including
vehicle counts or density, length of street segments, and self-report of traffic intensity (2).
Another approach is to measure selected individual tracer pollutants, such as NO
x
, at
multiple different locations with respect to roadways to represent the near-roadway air
pollution mixture (2). These pollutants can then be assigned to individuals or can be used
in subsequent steps to model exposures for later assignment to individuals. Three
commonly used modeling techniques are geostatistical interpolation, dispersion
modeling, and land-use regression. In geostatistical interpolation, surfaces of air pollution
are produced by interpolating concentrations at unmeasured locations using data from the
nearest measured locations. Four common geostatistical interpolation techniques have
been identified: 1) spatial averaging, 2) nearest monitor, 3) inverse distance weighting,
and 4) kriging (32). These geostatistical interpolation techniques are commonly used for
predicting regional pollutants, but they can also be used for the local setting. In dispersion
modeling, air pollution dispersion is simulated through mathematical modeling of a
number of input variable which include background concentrations of air pollution,
7
geophysical data (information about the terrain), meteorological data (wind speed, wind
direction, temperature, atmospheric stability), and on-road motor-vehicle emissions (2).
Land-use regression (LUR) modeling has recently become popular due to its ease of use
and its ability to better explain these small-scale variations compared to kriging (33, 34).
LUR works by regressing measured pollutant against a number of geographic
information systems (GIS) aided predictor variables including land-use categories, traffic
characteristics, and the physical environment. LUR has repeatedly been used to explain
variations of NO
2
and particulate matter (PM) across parts of Europe and North America
(35).
Review of near-roadway effects on lung function
Most studies of near-roadway effects on lung function have been cross-sectional.
In an early study of fourth grade children living in Munich, Germany, researchers found
that an increase in traffic volume on main roads through school districts was associated
with peak expiratory flow (36). Another early study that investigated the impacts of near
roadway traffic exposure and lung function in children found strong negative associations
of truck density on lung function among children living within 300 meters of motorways
(37). Measured concentrations of black smoke and NO
2
in school attended by these
children were also significantly associated with lower measures of lung function. Oftedal
et al. (38) used the EPISODE dispersion model to estimate exposure to NO
2
, PM
10
, and
PM
2.5
for different time scales among 9- and 10-year old children who had spent their
entire lifetime in Oslo, Norway. Early and lifetime pollutants were significantly
associated with lower PEF, FEF
25%
, and FEF
50%,
with more pronounced effects seen
8
among girls. In an Italian study, LUR modeled NO
2
was inversely associated with
FEV
1
/FVC, FEF
25-75
, and PEF (39). However in this same study, self-reported measure of
traffic intensity and distance to busy roads had mixed associations as these exposures
were positively associated with some measurements of lung function and negatively
associated with others.
In a study of German children between the ages of 9 and 11, no associations were
found between lung function and local traffic (assessed via traffic counts and modeled
concentrations of soot, benzene, and NO
2
), but associations were found between local
traffic and a number of respiratory outcomes including asthma, wheeze, and cough (40).
Consistent findings from another German study of children (ages 5-7 and 9-11) found
associations of near-roadway air pollution (SO
2
, NO
2
, CO, benzene, and O
3
assessed via
extensive monitoring of 1 x 1 km grid) with cough and bronchitis but not measures of
lung function (41).
In the CHS, a longitudinal analysis of the effects of near-roadway air pollution on
8-year growth was performed on both fourth grade cohorts. Children living within 500
meters of a freeway showed significantly stunted FEV
1
and MMEF growth as well as
possible effects on FVC growth when compared to children living farther than 1500
meters from a freeway (42). In addition, higher levels of model-based estimates of
freeway pollution (derived from a dispersion model) were marginally associated with
lower lung function growth.
9
1.4 PARTICULATE MATTER AND COMPOSITION
Particulate matter (PM) is a heterogeneous mixture of different chemical
components in various size fractions and has been linked to various health outcomes (29).
The composition of PM (organic and inorganic compounds) can vary greatly depending
on source. In urban areas, PM is generated mainly from motor vehicles either through
direct motor vehicle emissions or through brake and tire wear and dust resuspension.
Examples of other sources of man-made PM include fossil fuel burning in power stations
and factories, industrial processes, construction, and burning of fuels used for cooking
and heating. These particles that are released directly into the atmosphere are often
referred to as primary particles. Secondary particles, such as sulfates and nitrates, belong
to another group of particles that are formed from chemical reactions in the atmosphere.
PM is often characterized by its size. Ultrafine particles consist of PM with an
aerodynamic diameter of 0.1 µm or less (PM
0.1
) and are generally the smallest sized
particles measured. These particles are potentially the most toxic as they can penetrate
deep into the alveolar region of the lung and potentially enter the bloodstream. In the
atmosphere, these particles are able to coagulate and condense into larger particles known
as fine PM and are characterized as having aerodynamic diameters of between 0.1 and
2.5 µm. PM
2.5
includes the total of the fine fraction and ultrafine fractions. The last
fraction of PM often studied are the coarse particulates which are defined as particles
having aerodynamic diameters of between 2.5 and 10 µm. Together, PM
2.5
and coarse
PM make up PM
10
. The ultrafine particles in PM
10
, contain very little of the total mass
but contribute the most to particle number.
10
Transition metals
Metals have often been implicated as the mostly likely toxic component in PM
(43). Transition metals are of special interest because of their ability to produce reactive
oxidative species (ROS) through Fenton reactions or by interacting with cellular proteins
(44). Transition metals come from a number of different sources and the toxicity of PM
may vary by its metal composition. For example, copper (Cu), iron (Fe), and zinc (Zn)
are all produced from brake and tire wear while nickel (Ni) and vanadium (V) are
markers of ship emissions (45). A number of studies have shown relationships between
transition metals and respiratory morbidities such as airway inflammation, asthma
exacerbation, and wheeze in children (46-48), but few studies have shown associations
with lung function.
1.5 POSSIBLE BIOLOGICAL MECHANISMS INVOLVED IN AIR POLLUTION
EFFECTS ON THE LUNGS
A likely mechanism by which air pollution affects the lungs is through the
oxidative stress and inflammatory pathways (49). Oxidative stress occurs when a buildup
of oxidants overwhelms antioxidant defenses. As a result of this imbalance between
oxidants and antioxidants, free radicals (atoms with an unpaired electron) react with
nearby lipids, proteins, and nucleic acids leading to cellular damage. A number of
exogenous (air pollution, radiation, carcinogens, trauma) and endogenous sources
(mitochondrial electron transport during cellular respiration) have the potential to induce
oxidative stress. The lungs are a highly vulnerable target of oxidative damage because
gas exchange between ambient air and the body occurs in the lungs and the lung’s large
11
surface area allows for many possible interactions with pollutants (50). A consequence of
air pollution induced oxidative stress in the lungs is an influx of activated inflammatory
cells. These inflammatory cells can further enhance oxidative stress in the lungs as
inflammatory cells have the potential to generate and release free radicals. The free
radicals then attack the lung causing tissue damage.
Different pollutants in ambient air can induce oxidative stress through varying
mechanisms. Ozone interacts with substrates that are present in the lung lining fluid and
initiates a cascade of secondary, free radical derived, ozonation products (49). NO
2
also
interacts with substrates in the lung lining fluid leading to the formation of oxidized
species that are capable of lung damage. Particulate matter can contain transition metals
and organic components such as polycyclic aromatic hydrocarbons that directly or
indirectly induce oxidative stress. Certain transition metals can undergo redox cycling
and form damaging hydroxyl radicals through Fenton or Fenton-like mechanisms (29,
51). Human exposure chamber studies have shown that short term exposure to diesel
exhaust elicits an acute inflammatory effect in healthy subjects (52). Experimental studies
have shown that exposing human bronchial or alveolar cells to transition metals resulted
in increased markers of inflammation (53, 54).
1.6 PROPOSED DISSERTATION
While studies have shown effects of regional and near-roadway air pollution on
health, it is unclear whether these effects are independent of one another and what the
magnitude of the joint effect of regional and near-roadway air pollution is on lung
function. A number of studies have used modeled NO
2
as a surrogate for the near-
12
roadway air pollution mixture, but it is unclear whether NO
2
is most responsible for
observed effects of near-roadway air pollution on lung function or whether other
pollutants cause these effects. What are these pollutants that may inhibit lung function in
children and how well are we able to model them? Composition and size of PM are likely
important determinants of its toxicity. Transition metals have been suggested as the
component of PM that is most likely to be harmful. How does PM and its constituents
across different size fractions affect children’s lung function? To answer these questions,
a fifth cohort of children from kindergarten and first grade were recruited into the CHS in
the 2002-2003 school year from several Southern California communities. During the
sixth year of the study (2007-2008), children from 8 of these communities had their lung
function tested at age 11. A supplemental group of children was recruited that year to
increase the power to detect associations of lung function with near-roadway pollution
and regional pollution. Near-roadway pollution was assessed as part of a campaign in
2004-2005 to measure the intra-community variability of nitrogen monoxide (NO), NO
2
and oxides of nitrogen (NO
x
) and then subsequently develop LUR models for predicting
the small scale variability of these pollutants for use in health studies (55). Over 900
samplers were deployed across 12 communities in Southern California. Ogawa samplers
were used to measure levels of NO
2
and NO
x
outside the homes of a subset of active
subjects in the current CHS cohort, at the schools from which participating CHS subjects
were selected, and at community central monitoring stations (NO was calculated by
taking the difference of the two measurements). Measurements were taken during a 2-
week period in the winter and another 2-week period in the summer. All locations within
each season were switched on at the same time to avoid any temporal ambiguity in the
13
measurements. The average of the two 2-week seasonal concentrations was taken prior to
modeling in order to get a single annual estimate at each location. Using a land-use
regression modeling framework and accounting for spatial autocorrelation, a model
containing CALINE-modeled pollution (a dispersion-modeled estimate of pollution),
distance to freeway and other major roads, non-freeway traffic within a 300m buffer,
population within a 300m buffer, and elevation explained 63% of the variability of NO
and 71% of the variability of both NO
2
and NO
x
. In a previous study, these predicted
near-roadway exposures were found to be associated with lower lung function with
stronger effects seen in children whose parents reported high levels of stress (56). In
Chapter 2, I will investigate whether the effects of these predicted, near-roadway
concentrations of NO, NO
2
, and NO
x
at the residence of CHS subjects are independent of
the effects of regional air pollution in order to see how spatial scales may be important in
understanding air pollution’s impact on lung function.
While NO
2
and NO
x
have repeatedly been used in the development of spatial
models, there have been fewer studies that have explored the spatial variability of other
constituents of near-roadway pollution (35). A second campaign of intense collection of
various pollutants to study intra-community variability in these CHS communities took
place in 2008-2009 in which potentially more biologically relevant EC, OC and elements
were measured. In Chapter 3, I will describe the development of land-use regression
models for two pollutants (elemental carbon and particulate matter) in two different size
fractions (2.5µm and 0.2µm) in a subset of communities from the CHS. Models that can
predict a substantial proportion of the intra-community variability of particular pollutants
can be used for predicting exposures among subjects within the CHS. These pollutants
14
could potentially provide stronger evidence for impacts of near-roadway air pollution on
any particular health endpoint.
Transition metals were measured at multiple locations only in the quasi-ultrafine
size fraction. The intra-community spatial variation was not predictable based on traffic
and other characteristics available. However, as part of this second exposure campaign,
concentrations of various elements in multiple size fractions were collected at the schools
of CHS participants, who generally lived in neighborhoods around the schools. This
wealth of information has allowed for possible testing of the impact of transition metals
across several size fractions on lung development in children. In Chapter 4, I will explore
the relationship between transition metals and lung development among CHS
participants.
In Chapter 5, I will summarize the findings and provide my synthesis of the
findings in Chapters 2-4. I will conclude by providing some thoughts about future
research directions.
15
1.7 REFERENCES
1. Davis DL, Bell ML, Fletcher T. A look back at the London smog of 1952 and the
half century since. Environmental health perspectives 2002;110(12):A734-5.
2. HEI. Traffic-related air pollution: a critical review of the literature on emissions,
exposure, and health effects. HEI Special Report: Health Effects Institute, 2010.
3. Guxens M, Sunyer J. A review of epidemiological studies on neuropsychological
effects of air pollution. Swiss Med Wkly 2012;141:w13322.
4. Schauer JJ, Rogge WF, Hildemann LM, et al. Source apportionment of airborne
particulate matter using organic compounds as tracers. Atmospheric Environment
1996;30(22):3837-55.
5. Minguillón MC, Arhami M, Schauer JJ, et al. Seasonal and spatial variations of
sources of fine and quasi-ultrafine particulate matter in neighborhoods near the
Los Angeles–Long Beach harbor. Atmospheric Environment 2008;42(32):7317-
28.
6. Mysliwiec MJ, Kleeman MJ. Source apportionment of secondary airborne
particulate matter in a polluted atmosphere. Environmental science & technology
2002;36(24):5376-84.
7. Bates DV. The effects of air pollution on children. Environmental health
perspectives 1995;103 Suppl 6:49-53.
8. Islam T, Gauderman WJ, Berhane K, et al. Relationship between air pollution,
lung function and asthma in adolescents. Thorax 2007;62(11):957-63.
9. Schroeder EB, Welch VL, Couper D, et al. Lung function and incident coronary
heart disease: the Atherosclerosis Risk in Communities Study. Am J Epidemiol
2003;158(12):1171-81.
10. Kannel WB, Hubert H, Lew EA. Vital capacity as a predictor of cardiovascular
disease: the Framingham study. Am Heart J 1983;105(2):311-5.
11. Knuiman MW, James AL, Divitini ML, et al. Lung function, respiratory
symptoms, and mortality: results from the Busselton Health Study. Ann Epidemiol
1999;9(5):297-306.
12. Schunemann HJ, Dorn J, Grant BJ, et al. Pulmonary function is a long-term
predictor of mortality in the general population: 29-year follow-up of the Buffalo
Health Study. Chest 2000;118(3):656-64.
13. de Marco R, Accordini S, Marcon A, et al. Risk factors for chronic obstructive
pulmonary disease in a European cohort of young adults. Am J Respir Crit Care
Med 2011;183(7):891-7.
14. Gauderman WJ, Gilliland GF, Vora H, et al. Association between air pollution
and lung function growth in southern California children: results from a second
cohort. Am J Respir Crit Care Med 2002;166(1):76-84.
15. Schwartz J. Lung function and chronic exposure to air pollution: a cross-sectional
analysis of NHANES II. Environ Res 1989;50(2):309-21.
16. Raizenne M, Neas LM, Damokosh AI, et al. Health effects of acid aerosols on
North American children: pulmonary function. Environmental health perspectives
1996;104(5):506-14.
17. Tager IB, Balmes J, Lurmann F, et al. Chronic exposure to ambient ozone and
lung function in young adults. Epidemiology 2005;16(6):751-9.
16
18. Dockery DW, Speizer FE, Stram DO, et al. Effects of inhalable particles on
respiratory health of children. The American review of respiratory disease
1989;139(3):587-94.
19. Kopp MV, Bohnet W, Frischer T, et al. Effects of ambient ozone on lung function
in children over a two-summer period. Eur Respir J 2000;16(5):893-900.
20. Ihorst G, Frischer T, Horak F, et al. Long- and medium-term ozone effects on
lung growth including a broad spectrum of exposure. Eur Respir J
2004;23(2):292-9.
21. Horak F, Jr., Studnicka M, Gartner C, et al. Particulate matter and lung function
growth in children: a 3-yr follow-up study in Austrian schoolchildren. Eur Respir
J 2002;19(5):838-45.
22. Peters JM, Avol E, Gauderman WJ, et al. A study of twelve Southern California
communities with differing levels and types of air pollution. II. Effects on
pulmonary function. Am J Respir Crit Care Med 1999;159(3):768-75.
23. Gauderman WJ, McConnell R, Gilliland F, et al. Association between air
pollution and lung function growth in southern California children. Am J Respir
Crit Care Med 2000;162(4 Pt 1):1383-90.
24. Gauderman WJ, Avol E, Gilliland F, et al. The effect of air pollution on lung
development from 10 to 18 years of age. N Engl J Med 2004;351(11):1057-67.
25. Avol EL, Gauderman WJ, Tan SM, et al. Respiratory effects of relocating to areas
of differing air pollution levels. Am J Respir Crit Care Med 2001;164(11):2067-
72.
26. Lurmann F, Avol E, Gilliland F. Emissions reduction policies and recent trends in
Southern California’s ambient air quality. Journal of the Air & Waste
Management Association 2014;65(3):324-35.
27. Gauderman WJ, Urman R, Avol E, et al. Association of improved air quality with
lung development in children. N Engl J Med 2015;372(10):905-13.
28. Thorpe A, Harrison RM. Sources and properties of non-exhaust particulate matter
from road traffic: a review. The Science of the total environment 2008;400(1-
3):270-82.
29. Kelly FJ, Fussell JC. Size, source and chemical composition as determinants of
toxicity attributable to ambient particulate matter. Atmospheric Environment
2012;60(0):504-26.
30. Zhu YF, Hinds WC, Kim S, et al. Concentration and size distribution of ultrafine
particles near a major highway. Journal of the Air & Waste Management
Association 2002;52:1032-42.
31. Roorda-Knape MC, Janssen NAH, De Hartog JJ, et al. Air pollution from traffic
in city districts near major motorways. Atmospheric Environment
1998;32(11):1921-30.
32. Wong DW, Yuan L, Perlin SA. Comparison of spatial interpolation methods for
the estimation of air quality data. Journal of exposure analysis and environmental
epidemiology 2004;14(5):404-15.
33. Briggs DJ, Collins S, Elliott P, et al. Mapping urban air pollution using GIS: a
regression-based approach. Int J Geogr Inf Sci 1997;11(7):699-718.
17
34. Lebret E, Briggs D, van Reeuwijk H, et al. Small area variations in ambient NO2
concentrations in four European areas. Atmospheric Environment 2000;34(2):177-
85.
35. Hoek G, Beelen R, de Hoogh K, et al. A review of land-use regression models to
assess spatial variation of outdoor air pollution. Atmospheric Environment
2008;42(33):7561-78.
36. Wjst M, Reitmeir P, Dold S, et al. Road traffic and adverse effects on respiratory
health in children. BMJ 1993;307(6904):596-600.
37. Brunekreef B, Janssen NA, de Hartog J, et al. Air pollution from truck traffic and
lung function in children living near motorways. Epidemiology 1997;8(3):298-
303.
38. Oftedal B, Brunekreef B, Nystad W, et al. Residential outdoor air pollution and
lung function in schoolchildren. Epidemiology 2008;19(1):129-37.
39. Rosenlund M, Forastiere F, Porta D, et al. Traffic-related air pollution in relation
to respiratory symptoms, allergic sensitisation and lung function in
schoolchildren. Thorax 2009;64(7):573-80.
40. Nicolai T, Carr D, Weiland SK, et al. Urban traffic and pollutant exposure related
to respiratory outcomes and atopy in a large sample of children. Eur Respir J
2003;21(6):956-63.
41. Hirsch T, Weiland SK, von Mutius E, et al. Inner city air pollution and respiratory
health and atopy in children. Eur Respir J 1999;14(3):669-77.
42. Gauderman WJ, Vora H, McConnell R, et al. Effect of exposure to traffic on lung
development from 10 to 18 years of age: a cohort study. Lancet
2007;369(9561):571-7.
43. Chen LC, Lippmann M. Effects of metals within ambient air particulate matter
(PM) on human health. Inhalation toxicology 2009;21(1):1-31.
44. Schwarze PE, Ovrevik J, Lag M, et al. Particulate matter properties and health
effects: consistency of epidemiological and toxicological studies. Human &
experimental toxicology 2006;25(10):559-79.
45. Hasheminassab S, Daher N, Saffari A, et al. Spatial and temporal variability of
sources of ambient fine particulate matter (PM2.5) in California. Atmos Chem
Phys 2014;14(22):12085-97.
46. Rosa MJ, Perzanowski MS, Divjan A, et al. Association of recent exposure to
ambient metals on fractional exhaled nitric oxide in 9-11 year old inner-city
children. Nitric oxide : biology and chemistry / official journal of the Nitric Oxide
Society 2014;40:60-6.
47. Hirshon JM, Shardell M, Alles S, et al. Elevated ambient air zinc increases
pediatric asthma morbidity. Environmental health perspectives 2008;116(6):826-
31.
48. Patel MM, Hoepner L, Garfinkel R, et al. Ambient metals, elemental carbon, and
wheeze and cough in New York City children through 24 months of age. Am J
Respir Crit Care Med 2009;180(11):1107-13.
49. Kelly FJ. Oxidative stress: its role in air pollution and adverse health effects.
Occupational and environmental medicine 2003;60(8):612-6.
50. Bargagli E, Olivieri C, Bennett D, et al. Oxidative stress in the pathogenesis of
diffuse lung diseases: a review. Respiratory medicine 2009;103(9):1245-56.
18
51. Stohs SJ, Bagchi D. Oxidative mechanisms in the toxicity of metal ions. Free
radical biology & medicine 1995;18(2):321-36.
52. Xu Y, Barregard L, Nielsen J, et al. Effects of diesel exposure on lung function
and inflammation biomarkers from airway and peripheral blood of healthy
volunteers in a chamber study. Particle and fibre toxicology 2013;10:60.
53. Carter JD, Ghio AJ, Samet JM, et al. Cytokine production by human airway
epithelial cells after exposure to an air pollution particle is metal-dependent.
Toxicology and applied pharmacology 1997;146(2):180-8.
54. McNeilly JD, Heal MR, Beverland IJ, et al. Soluble transition metals cause the
pro-inflammatory effects of welding fumes in vitro. Toxicology and applied
pharmacology 2004;196(1):95-107.
55. Franklin M, Vora H, Avol E, et al. Predictors of intra-community variation in air
quality. J Expo Sci Environ Epidemiol 2012;22(2):135-47.
56. Islam T, Urman R, Gauderman WJ, et al. Parental stress increases the detrimental
effect of traffic exposure on children's lung function. Am J Respir Crit Care Med
2011;184(7):822-7.
19
CHAPTER 2: Associations of Children’s Lung Function with Ambient
Air Pollution: Joint Effects of Regional and Near-roadway Pollutants
Urman R, McConnell R, Islam T, Avol E, Lurmann F, Vora H, Linn W, Rappaport E,
Gilliland F, Gauderman J
2.1 ABSTRACT
Prior studies have reported adverse effects of either regional or near-roadway air
pollution (NRAP) on lung function, but little has been done of the joint effects of these
exposures. This study was conducted to assess these joint effects on childhood lung
function in the Children’s Health Study. Lung function was measured on 1,811 children
from eight Southern Californian communities. NRAP exposure was assessed based on (1)
residential distance to the nearest freeway or major road and (2) estimated near-roadway
contributions to residential nitrogen dioxide (NO
2
), nitric oxide (NO), and total nitrogen
oxides (NO
x
). Exposure to regional ozone (O
3
), NO
2
, particulate matter with aerodynamic
diameter less than 10 µm (PM
10
) and 2.5 µm (PM
2.5
) was measured continuously at
community monitors. A 17.9 ppb (two standard deviation) increase in near-roadway NO
x
was associated with deficits of 1.6% in FVC (P=0.005) and 1.1% in FEV
1
(P=0.048).
Effects were observed in all communities and were similar for NO
2
and NO. Residential
proximity to a freeway was associated with a reduction in FVC. Lung function deficits of
2-3% were associated with regional PM
10
and PM
2.5
(FVC and FEV
1
) and with O
3
(FEV
1
), but not NO
2
, across the range of exposure between communities. These results
indicate that NRAP and regional air pollution have independent adverse effects on
childhood lung function.
20
2.2 INTRODUCTION
Reduced lung function has been associated with subsequent increased risk of
overall mortality, including coronary artery disease and respiratory disease in adults (1)
and with asthma in children (2). Therefore, identifying factors that reduce lung function
but are modifiable could lead to interventions with large public benefits.
Regional air pollutants have been associated with reduced lung function in both
adults and children (3, 4). Studies examining lung function in children exposed to local
residential near-roadway air pollution (NRAP) have not found consistent associations (5-
11), although exposure metrics differed across studies. However, there has been little
investigation of the joint effects of regional and NRAP exposures.
In this study, we assessed the joint effects of NRAP and regional exposures to
ozone (O
3
), particulate matter with aerodynamic diameter of less than 10 µm and 2.5 µm
(PM
10
and PM
2.5
), and nitrogen dioxide (NO
2
) on childhood lung function in the
Children’s Health Study (CHS). We examined associations with both traffic proximity
measures and land-use regression modeled NRAP based on a prior dense air monitoring
study of NO
x
conducted within CHS communities.
2.3 MATERIALS AND METHODS
Study Subjects
The CHS has enrolled over 11,000 children in a series of cohorts investigating the
health effects of air pollution. The current analysis includes a cohort established in 2002-
2003 when participants were 5-7 years of age (12). During the 2007-2008 school year,
21
lung function was measured on 1,811 cohort participants (82% of the active cohort) from
eight communities, as described in detail in the 2.7 Supplementary Material section.
Questionnaires
Questionnaires completed by parents or guardians at study enrollment provided
information on participants’ health, socio-demographic and other exposures, which was
updated yearly. A complete list of covariates is described in the 2.7 Supplementary
Material section.
Lung Function
Trained technicians measured lung function, weight, and height, and collected
information about recent acute respiratory illness. Using pressure-transducer-based
spirometers (Screenstar Spirometers, Morgan Scientific, Haverhill, MA), we identified
the maximal forced expiratory volume during the first second (FEV
1
) and forced vital
capacity (FVC) from a series of seven efforts from each child, as described previously
(13).
Air Pollution Exposure
NRAP exposures at each child’s residence and school were based on estimates of
surrogates, including distance to freeways, highways, and large surface streets. Spatial
land use regression models were developed based on an extensive monitoring campaign
of nitrogen oxides (NO
x
) and nitrogen dioxide (NO
2
) and by subtraction nitrogen oxide
(NO) at over 900 locations in CHS communities, as described previously (14). Key
22
predictors included distance to freeways and major roads, traffic volumes and their
emissions-weighted dispersion estimates, with lesser contributions from population
density and local variation in elevation. The resulting annual average predicted residential
concentrations of near-roadway NO, NO
2
, and NO
x
, incrementally increased above
regional background, was used in analyses, as described below.
The regional level of NO
2
, PM
2.5
, PM
10
, and O
3
was computed as the mean of the
six years of each pollutant measured continuously at a central monitoring location in each
community from cohort recruitment (2002) to the recording of lung function tests
(2007).
Additional details of NRAP and regional pollutant exposure assessment are
provided in the 2.7 Supplementary Material section.
Statistical Methods
We fitted linear regression models (with fixed effects for each study community)
to investigate associations of FVC and FEV
1
with NRAP and a mixed model that
included a random intercept for community to assess associations with regional pollutants
and joint effects with NRAP. Each pulmonary function outcome was log transformed to
satisfy the assumptions of the models. All models were adjusted for demographic and
anthropomorphic characteristics (eg. height) and selected other potential confounders (eg.
spirometry technician). In sensitivity analyses, other potential confounders and effect
modifiers were examined using standard methods described in further detail in the 2.7
Supplementary Material section.
23
The NRAP NO
x
(and NO and NO
2
) predicted residential exposures were deviated
from a community-specific mean. Conceptually, this allowed examination of the effect of
the complex NRAP mixture, for which the nitrogen oxides are only a surrogate, and to
distinguish it from the regional NO
2
effect, which was assessed based on the continuous
measurements at the community monitor so as to be comparable to other regional
pollutant assessments. This procedure was also necessary to make the NRAP NOx
approximately orthogonal (uncorrelated) to cross-community regional exposures in the
mixed models. Health effect estimates were scaled to the range of long-term average
regional pollution across all communities and to two standard deviations in the predicted
NRAP nitrogen oxides.
Based on our final model, we also computed estimated lung function
representative of different combinations of high and low regional and NRAP
environments. Low regional pollution was based on the minimum value of regional PM
2.5
while low NRAP was defined as one standard deviation below the mean value for
deviated NO
x
. Conversely, high regional pollution was based on the maximum value of
regional PM
2.5
and high NRAP was defined as one standard deviation above the mean
value for deviated NO
x
. We expressed the predicted lung function in these different
environments as percentages relative to those in the cleanest environment (low regional
and low NRAP).
2.4 RESULTS
The average age at lung function measurement was 11.2 years (SD=0.6). A
plurality of participants was White (40%) and a majority was of Hispanic ethnicity (57%,
24
Table 2.1). A substantial proportion of participants’ households had incomes less than
$30,000 (27%) or less than a high school education (21%). Eleven percent reported no
health insurance coverage for the child. Most homes had a gas stove (87%) and almost
half of families had at least a dog (36%) and/or a cat (19%). A small proportion of
subjects reported secondhand tobacco smoke exposure (4%) and household mold
problems (11%).
We examined the relationship of lung function participation rate of cohort
members with key sociodemographic and other characteristics. After adjusting for
community (to correspond to the analytic approach for assessment of NRAP effects),
non-participants were more likely to be boys and to have asthma (Supplemental Table
2.1). Otherwise, participants and non-participants were generally similar across a broad
range of demographic, social and housing characteristics.
Overall, 27% of children lived within 500 m of a freeway, while 20%, 15% and
38% lived 500-1,000 m, 1,000-1,500 m, or >1,500 m from a freeway, respectively
(Supplemental Table 2.2). There was 14% of children who lived within 75 m of a major
road (mostly non-freeway), 17% between 75 and 150 m, 28% between 150 and 300 m,
and 40% at least 300 m. The distributions of residential proximity to freeways and major
roads varied substantially from community to community. Predicted residential near-
roadway NO
x
, NO, and NO
2
showed wide variation within most study communities
(Figure 2.1). Correlations among regional pollutant levels ranged from 0.06 (between
PM
10
and NO
2
) to 0.80 (between PM
10
and PM
2.5
; Supplemental Table 2.3). O
3
had
relatively strong positive correlations with PM
2.5
and PM
10
. The correlation between
25
predicted near-roadway NO, NO
2
and NO
x
(within communities) exceeded 0.90
(Supplemental Table 2.4).
The means of FEV
1
and FVC for males were 2,474 ml and 2,902 ml, respectively,
and the corresponding means for females were 2,442 ml and 2,783 ml. Living within 500
m of a freeway was associated with a nearly 2 percent deficit in FVC (–1.96%; 95% CI:
–3.41%, –0.49%; P=0.009) compared to those living at least 1,500 m from a freeway
(Table 2.2). Mean FEV
1
was also lower for children living within 500 m of a freeway but
the association was not statistically significant. Although close proximity to a major road
was negatively associated with each measure of lung function, these associations were
not statistically significant.
Near-roadway residential NO
x
, NO, and NO
2
had statistically significant negative
associations with both FVC and FEV
1
(Table 2.2). For example, a two standard deviation
increase in near-roadway NO
x
exposure (17.9 ppb) was associated with a 1.56% deficit in
FVC (–2.62, –0.49; P=0.005), and a 1.10% deficit in FEV
1
(–2.19, –0.01; P=0.048).
Negative associations between near-roadway NO
x
and lung function were observed
within six of the eight study communities for FEV
1
(Figure 2.2A) and within all eight
study communities for FVC (Figure 2.2B). There was not significant heterogeneity of
near-roadway NO
x
effects across the eight communities for either FEV
1
(P=0.61) or FVC
(P=0.64).
Adjustment for potential confounding variables resulted in only small changes to
the estimated effects of near-roadway residential NO
x
on FEV
1
and FVC (Table 2.3). For
example, across models that included various additional adjustments, the near-roadway
NO
x
-related deficits ranged from –0.96% to –1.12% (main model: –1.10%) for FEV
1
, and
26
from –1.40% to –1.60% (main model: –1.56%) for FVC. In an analysis restricted to
children without asthma, the effect of near-roadway NO
x
was similar to that in the entire
study population (1.19% decline in FEV
1
and 1.51% decline in FVC). The difference in
effects between children with and without asthma was not statistically significant. There
was also no significant heterogeneity in near-roadway NO
x
effects on lung function in
girls compared to boys. Although we have observed associations of lung function with
exposure at schools of participants in this study in conjunction with psychosocial stress
(15), we observed no main effects of exposure in schools in this analysis (results not
shown).
Deficits in FEV
1
of approximately 3% were observed across the range of
community O
3
and PM
2.5
levels (P=0.006 for O
3
and 0.001 for PM
2.5
, Table 2.4 and
Figure 2.3). A greater than 2% deficit was observed across the range of PM
10
exposure.
Deficits in FVC of over 2% were also observed across the range of both PM
2.5
and PM
10
(Table 2.4 and Figure 2.4); however, a single community (Mira Loma) appears to have
driven the association between FVC and PM
10
.
In models assessing the joint effects of regional and NRAP, there was little
change in the strength of the regional pollutant associations with either FVC or FEV
1
,
after adjusting for near-roadway NO
x
(Table 2.5). For FEV
1
, there was little change in the
unadjusted association of near-roadway NO
x
(1.10% deficit in Table 2.2) after adjusting
for regional pollutants effects (1.04% to 1.14% deficits in Table 2.5). For FVC, the
unadjusted association with near-roadway NO
x
(1.56% deficit in Table 2.2) was
somewhat attenuated after adjusting for regional pollutants (1.40% to 1.49% deficits in
Table 2.5), although the associations remained significant. Similar patterns of lung
27
function deficits in two-pollutant models were observed for near-roadway NO and NO
2
(results not shown). The patterns of effects of freeway proximity associations were also
similar in models including a regional pollutant and in models unadjusted for regional
pollution (results not shown). We examined the possibility that background pollutant
exposures might up-regulate pulmonary response to near-roadway pollutants resulting in
larger lung function deficits in communities with high regional pollutants. However, none
of the regional pollutants significantly modified the association between near-roadway
residential NO
x
and each of the lung function endpoints (results not shown).
2.5 DISCUSSION
These results indicate that exposure to near-roadway air pollution adversely
affects childhood lung function. Strengths of the study were the ability to demonstrate
consistent effects of NRAP using both roadway proximity and validated predicted NO
x
markers for the NRAP mixture in communities with differing regional air quality,
roadway networks, and geographical characteristics. The study design offered an unusual
opportunity to demonstrate that associations of lung function with NRAP pollutant
variation were independent of associations also observed with regional air pollution.
NRAP is a complex mixture of particles and reactive gases with oxidant and pro-
inflammatory properties that could plausibly cause the observed lung function deficits
(16, 17). Oxides of nitrogen were selected to develop prediction models for likely near-
roadway variation of the mixture because they are inexpensive to measure with the
spatial density needed to develop valid models. NO
2
also has known oxidant and
immune-modulatory properties and could contribute to the near-roadway lung functions
28
effects (18), although in our analysis it was not possible to distinguish NRAP NO
2
effects
from other components of the mixture. The association of regional PM
2.5
and PM
10
with
both FEV
1
and FVC, and no effect of regional NO
2
, suggests that there were independent
effects of transported or secondary regional particulate matter and of the NRAP mixture
(rather than NO
2
). In addition, previous reports from the CHS (and other studies)
showing associations of NRAP, but not regional pollutants, with prevalent and incident
asthma (12, 19, 20) also are consistent with separate and independent effects of these
diverse pollutant mixtures.
It is also possible that more complex combinations of regional and NRAP account
for the observed associations, as toxicological and experimental studies indicate that
interaction with other pollutants may enhance the effects of particle exposure (21, 22).
Although the study design allowed us to examine the heterogeneity of NRAP health
effects across multiple communities, we found little evidence for interaction between
regional pollutants and NRAP. Rather, the adverse effects were relatively consistent in all
eight study communities, although there was limited precision to each estimate because
of limited community-specific sample size.
We have previously observed associations of regional PM (23) and traffic
proximity (7) with growth of FVC, but accompanied by larger effects in FEV
1
in an older
cohort of CHS participants. Other studies of traffic and lung function in elementary
school and adolescent children have also found larger associations with flow rates than
with FVC (8, 9, 24). However, the current results are consistent with an observed effect
of regional pollutants on FVC in a cross sectional analysis of prior CHS cohorts (13).
Additional follow up of this cohort is ongoing and may help elucidate these relationships.
29
Some previous studies that have looked at associations between residential traffic
related pollution and lung function were performed in multiple geographical regions (5,
7, 8, 10, 11), but many of these studies used only roadway proximity or traffic
count/density metrics rather than validated exposure models. Other studies that have used
land-use regression to estimate the relationship between NRAP and childhood lung
function were performed in relatively limited geographical regions (6, 9). Results have
not been consistent across studies.
These inconsistencies in the strength of association between near-roadway
residential traffic exposure and respiratory health across several prior studies (5-11) may
result in part from the use of different types of NRAP measures, with differing degrees of
uncertainty as proxies for pollution exposure. A strength of this study was the use of
quantitative residential NO
x
exposure assignments derived from a spatial land-use
regression model calibrated to measurements at well characterized locations in study
communities (14). Additionally, the association between lung function and predicted NO
x
was consistent with the inverse relationship between residential distance to a freeway and
lung function, which was also observed in an earlier CHS cohort (7), as concentrations of
NRAP decrease with increasing distance from a freeway (25). Comparable, high quality,
exposure assessment across studies would facilitate qualitative comparisons or pooled
analyses and might lead to more consistent epidemiologic findings.
The adverse associations of lung function with O
3
, PM
2.5
, and PM
10
are consistent
with other studies (3). In earlier CHS cohorts we reported associations of lung function
with PM
2.5
and PM
10
, as well as NO
2
,
but not with O
3
(7, 13). However, O
3
and PM were
30
correlated across communities of the current cohort, and it was therefore not possible to
distinguish effects of each.
This study replicates the general design and general age range of a cross-sectional
report from a previous CHS cohort (19) but expands the scope of that earlier work by
examining both between and within-community pollutant effects. The amount of
between-community regional variation in the present study is less than that found in
previous CHS studies due to our focus on more-urban communities with larger gradients
in NRAP. However, a nearly two-fold difference in the six-year averaged regional
pollution concentrations (Figures 2.3 and 2.4) exists between the highest and lowest
polluted communities, which allowed us to identify between and within-community
effects. We have been collecting additional lung function data and will examine
longitudinal pollutant effects separately.
We considered the possibility that bias explained our results. Participants and
non-participants from the cohort were generally similar across a broad range of
demographic, social and housing characteristics (Supplemental Table 2.1). The only
significant difference was for boys, who were more likely than girls to be non-
participants. However, adjusting for sex and for other characteristics had little impact on
the NRAP effect estimate (Table 2.3). Furthermore, the effect of NOx on lung function in
analyses restricted to girls was generally similar to the effect among all participants.
Although selection bias and residual confounding by other factors cannot be excluded as
an explanation for our results, these analyses provide little reason to believe that this
occurred.
31
There are potentially large public health implications of these findings because
NRAP exposure due to proximity of homes and other locations where children spend
time is common (26, 27) and lung function in childhood tracks into adult life (28-30).
Furthermore, the strong association between exposure and lung function in non-asthmatic
children suggests that traffic-related pollution did not affect only a sensitive subgroup but
rather has a potential impact on all children. Although direct comparison of the
magnitude of effects of regional and near-roadway pollution is difficult, the deficits
associated with near-roadway NOx across a (two-standard deviation) range of within-
community variation encompassing most children in our study communities was only
modestly less than the effects of regional pollutants across the range of community-
average exposure. Compared with a child living in a low NRAP environment in a low
regional PM
2.5
community, the results suggest that a child living in a high NRAP
environment in a community with high PM in Southern California would experience a
greater than 4% decrease in FEV
1
(Figure 2.5) For comparison with another common
exposure, maternal secondhand smoking of 1 pack/day has been shown to be associated
with a 0.4% deficit in childhood level of FEV
1
(31). Prevention of these large pollutant
effects poses a challenge to the current air pollution regulatory framework, which
historically has set standards using risk calculations that consider effects of regional air
quality but not near-roadway traffic-related variation in exposure.
32
2.6 TABLES AND FIGURES
Table 2.1. Characteristics of 1,811 CHS participants with lung function testing.
N (total=1811) %
a
Male 871 48.1
Race
Asian 86 4.8
Black 39 2.2
Don't Know 239 13.2
Mixed 229 12.6
Other 486 26.8
White 732 40.4
Hispanic ethnicity
Don't Know 92 5.1
Hispanic 1028 56.8
Not Hispanic 691 38.2
SES
Household income
<$30,000 402 27.1
$30,000 or more 1084 73.0
Parental education
Did not finish high school 345 20.6
High school diploma or some college 854 51.0
College diploma or greater 477 28.5
Health insurance covers child 1508 89.3
Home characteristics/Potential exposures
Gas stove 1462 86.5
Dog 599 35.8
Cat 312 18.8
Mold past 12 months 172 10.5
Secondhand smoke exposure 67 3.8
In-utero exposure to maternal smoking 99 5.8
Health conditions
Acute respiratory illness 164 9.4
Medical diagnosis of asthma 334 19.5
a
Due to missing values, denominators (n) for each percentage may
differ.
33
Table 2.2. Effects of measures of near-roadway air pollution on
lung function level.
FEV
1
a
FVC
a
%Diff 95% CI %Diff 95% CI
Freeway
>1,500 m Ref Ref
1,000-1,500 m 1.63 (-0.05, 3.34) 0.99 ( -0.65, 2.66)
500-1,000 m -0.50 (-2.05, 1.07) -1.01 (-2.52, 0.53)
<500 m -1.06 (-2.55, 0.45) -1.96 (-3.41, -0.49) **
Trend (P-value) 0.09 0.004
Major Road
>300 m Ref Ref
150-300 m -0.56 (-1.90, 0.79) -0.69 (-2.00, 0.65)
75-150 m -0.50 (-2.04, 1.06) -0.82 (-2.32, 0.72)
<75 m -1.58 (-3.21, 0.09) -1.53 (-3.14, 0.11)
Trend (P-value) 0.09 0.06
Predicted Near-roadway Pollution
b
NO
2
-1.00 (-2.08, 0.09) -1.40 (-2.46, -0.33) *
NO -1.19 (-2.27, -0.09) * -1.68 (-2.74, -0.60) ***
NO
x
-1.10 (-2.19, -0.01) * -1.56 (-2.62, -0.49) ***
a
All models include adjustments for log of height and its squared value,
BMI and BMI
2
, sex, age, sex*age interaction, race, Hispanic ethnicity,
respiratory illness at time of test, field technician, and study community.
b
Near-roadway residential pollutants were scaled to two standard
deviations of their respective community-mean centered distributions (6.4
ppb for NO
2
, 12.3 ppb for NO, and 17.9 ppb for NO
x
).
* P<0.05, ** P<0.01, *** P<0.005
34
Table 2.3. Sensitivity analysis for lung function effects of near-roadway residential NO
x
.
FEV
1
a
FVC
a
% diff (95% CI) % diff (95% CI)
Main model -1.10 (-2.19, -0.01) -1.56 (-2.62, -0.49)
Additional covariates
Main model + family income -1.04 (-2.13, 0.07) -1.51 (-2.58, -0.43)
Main model + parental level of education -0.96 (-2.05, 0.14) -1.44 (-2.51, -0.36)
Main model + diagnosis of asthma by medical doctor -1.06 (-2.14, 0.03) -1.55 (-2.61, -0.47)
Main model + dogs in home -0.97 (-2.06, 0.13) -1.40 (-2.47, -0.33)
Main model + cats in home -1.09 (-2.18, 0.00) -1.55 (-2.61, -0.47)
Main model + exposure to gas stove -1.10 (-2.18, -0.01) -1.56 (-2.62, -0.49)
Main model + in-utero exposure to maternal smoking -1.09 (-2.17, 0.01) -1.60 (-2.66, -0.53)
Main model + exposure to tobacco smoke at home -1.12 (-2.20, -0.02) -1.57 (-2.63, -0.50)
Main model + exposure to mold -1.12 (-2.21, -0.03) -1.57 (-2.64, -0.50)
Main model + insurance coverage -1.10 (-2.18, -0.01) -1.55 (-2.61, -0.48)
Subgroup analysis
Non-asthmatics -1.19 (-2.41, 0.05) -1.51 (-2.72, -0.29)
Asthmatics -0.65 (-3.35, 2.14) -1.20 (-3.91, 1.58)
Boys -0.96 (-2.48, 0.58) -1.13 (-2.60, 0.36)
Girls -1.10 (-2.65, 0.48) -1.81 (-3.34, -0.25)
a
See Table 2.2 for adjustment variables and scaling factor for pollutant effects.
35
Table 2.4. Effect of averaged regional pollutants
on lung function level.
Regional Pollutant % Diff
a
95% CI
FEV
1
O
3
(10am-6pm)
-3.10 (-5.24, -0.91)
**
PM
2.5
-2.94 (-4.65, -1.20)
***
PM
10
-2.19 (-3.98, -0.37)
*
NO
2
-1.19 (-4.14, 1.85)
FVC O
3
(10am-6pm)
-0.31 (-3.11, 2.57)
PM
2.5
-2.25 (-3.94, -0.52)
*
PM
10
-2.05 (-3.54, -0.54)
**
NO
2
-0.79 (-3.52, 2.02)
a
See footnote to Table 2.2 for adjustment variables
(community adjustment not included). Each pollutant was
scaled to the range of the 24-hour average over the study
period from 2002 until 2007 with the exception of O
3
,
which was scaled to the 8-hour average from 10am to 6pm
(22.7 ppb for O
3
10-6, 13.3 µg/m
3
for PM
2.5
, 30.3 µg/m
3
for
PM
10
, 19.4 µg/m
3
for NO
2
).
* P<0.05, ** P<0.01, *** P<0.005
36
Table 2.5. Joint analysis of regional air pollution and near-roadway NO
x
on lung function.
Effect of Regional Pollutant
a
Effect of Near-roadway NO
x
a
Regional Pollutant
% Diff 95% CI
% Diff 95% CI
FEV
1
O
3
(10am-6pm)
-3.24
(-5.32, -1.11) ***
-1.04 (-2.11, 0.05)
PM
2.5
-3.00
(-4.76, -1.21) ***
-1.07 (-2.14, 0.01)
PM
10
-2.24
(-4.04, -0.41) *
-1.14 (-2.22, -0.06) *
NO
2
-1.22
(-4.23, 1.88)
-1.07 (-2.15, 0.02)
FVC O
3
(10am-6pm)
-0.34
(-3.21, 2.63)
-1.47 (-2.53, -0.41) **
PM
2.5
-2.35
(-4.09, -0.57) **
-1.40 (-2.46, -0.34) **
PM
10
-2.17
(-3.68, -0.63) **
-1.49 (-2.54, -0.43) **
NO
2
-0.78
(-3.62, 2.15)
-1.46 (-2.52, -0.39) **
a
See Table 2.2 for adjustment variables (community adjustment not included) and scaling factor
for pollutant effects of near-roadway NO
x
. See footnote to Table 2.4 for scaling factor for
regional pollutants.
* P<0.05, ** P<0.01, *** P<0.005
37
Figure 2.1. Distribution of predicted local (A) NO, (B) NO
2
, and (C) NO
x
within each of
the eight study communities based on a spatial land-use regression model.
38
Figure 2.2. Associations of local NO
x
with (A) FEV
1
and (B) FVC within each study
community.
39
Figure 2.3. Adjusted average FEV
1
versus 2002-2007 community-average pollutant levels.
Average FEV
1
values are referenced to a white, non-hispanic female of average height and BMI
and without a respiratory infection on the day pulmonary function was examined.
40
Figure 2.4. Adjusted average FVC versus 2002-2007 community-average pollutant levels.
Average FVC values are referenced to a white, non-hispanic female of average height and BMI
and without a respiratory infection on the day pulmonary function was examined.
41
Figure 2.5. Joint effect of regional PM
2.5
and NRAP on FEV
1
.
Percentages in different exposure environments are relative to a low regional PM
2.5
and low NRAP environment as described in the Statistical Methods section.
42
2.7 SUPPLEMENTARY MATERIAL
Supplementary Methods
Study Subjects
The current analysis is from a Children’s Health Study (CHS) cohort established
in 2002-2003 in 13 Southern California communities when participants were 5-7 years of
age (12). Eight communities with an original enrollment of 3618 contributed to the
current analysis. (Due to resource limitations, lung function was not measured in five).
These eight communities represent a broad range of regional exposure, including Santa
Barbara, a clean coastal community; Long Beach, a coastal city with high PM and NO
2
levels but low ozone levels; and inland communities with high PM and relatively high
ozone levels. The communities were selected based in addition on the presence of large
gradients in near-roadway exposure. During the 2007-2008 school year lung function was
measured on 1,523 cohort participants (82% of the active cohort). In addition, in schools
with participants with high residential NRAP exposure, a supplementary cohort of 352
children was recruited from classrooms of participants already undergoing follow-up and
parents provided informed consent for lung function testing prior to field testing. Of these
more recently enrolled children, 288 (82%) were available and completed testing. In
total, 1811 children from eight Southern Californian communities (Anaheim, Glendora,
Long Beach, Mira Loma, Riverside, Santa Barbara, San Dimas, and Upland) participated
in lung function testing.
The study was approved by the University of Southern California Institutional
Review Board, and written informed consent was obtained from a parent or guardian of
each participant.
43
Distance-based Exposure Measurements
Participant residence and school addresses were standardized and their locations
were geo-coded to 13 m perpendicular to the side of the adjacent road, using the Tele
Atlas database and software (Tele Atlas, Inc., Boston, CA, www.na.teleatlas.com).
Distance to the nearest major road was estimated using ESRI ArcGIS Version 9.2 (ESRI,
Redlands, CA, www.esri.com). A major road was defined based on functional
classification by the California Department of Transportation as a freeway (with limited
access) or other highway (typically with heavy traffic volume), or a major or minor
arterial thoroughfare. Each direction of travel was represented as a separate roadway, and
the shortest distance was estimated from the residence to the middle of the nearest side of
the freeway or major road. We included in the analysis only children with addresses that
could be geo-coded accurately. Specifically, only residential addresses for which the Tele
Atlas geo-coding software assigned its highest quality match code were included. These
addresses were located on the correct side of the street with their relative position
between cross streets determined by linear interpolation of residence number between the
nearest intersections.
Residential and school address distance to a freeway were categorized as <500 m,
500-999 m, 1,000-1,499 m, and >1,500 m, based on the distribution of the residences of
participants, recent evidence for extended increased concentration of fresh traffic
pollutants on this scale, and results from a previous CHS cohort that have shown
respiratory health associations on this spatial scale (7, 32-35). Distance to the nearest
major road (including freeways) was categorized as <75 m, 75 to 150 m, >150 m to 300
m, and >300 m, based on the markedly increased exposure and risk of asthma within 75
44
m of major roadways in previous studies (including this cohort at study entry), which
decreased to background levels by 150 to 300 m (12, 36-39).
Land-use Regression Modeling of NO, NO
2
, and NO
x
For a detailed description of sample collection and model selection, see Franklin,
2012. (14) Land-use regression (LUR) models of NO, NO
2
, and NO
x
, were developed
based on 942 collected samples across 12 communities in southern California. In 2005-
2006, concentrations of pollutants were collected using Ogawa samplers that were
deployed for 2 weeks in the winter and 2 weeks in the summer. NO, NO
2
, and NO
x
were
chosen as surrogates of the near roadway air pollution because of the existence of
inexpensive monitors that can readily be deployed across multiple locations. A natural
log transformation of these pollutants was performed prior to modeling. Because the
main focus during the development of the models was to understand the within-
community or local distribution of these pollutants, the measured concentrations of these
pollutants and each of the predictors were subtracted (deviated) from the community-
specific town means. The final LUR models included both a freeway and a non-freeway
component of CALINE4-modeled NRAP (a Gaussian line-source dispersion model) (40),
distance to the nearest freeway and its squared value, distance to the nearest non-freeway
major road, non-freeway traffic volume and population within a 300m buffer, and
elevation. To assess the performance of the models, 10-fold cross validation was
performed in which the data was split into 10 parts, and nine of the parts were used for
training the model, while one part was held out for validation purposes. This was
repeated 10 times so that each part had an opportunity to be validated by the rest of the
45
data. The 10-fold cross validation R
2
-values for predicting local variation in traffic
exposures based on these predictors were 63% for NO, 71% for NO
2
, and 71% for NO
x
.
These models were used to predict local annual average concentrations of NO,
NO
2
, and NO
x
at the residence and school of each child in the lung function study.
Predicted concentrations from the LUR model were on a log-deviated scale. To make
these predictions more interpretable, we added back a natural log transformed community
mean (calculated from the original sampling) to each predicted value, which was then
exponentiated to get the predictions on a part per billion scale. These predictions were
then deviated again from the community mean (unlogged) of a given pollutant in order to
examine the local effects of these pollutants.
Regional Exposures
Since the beginning of this study, regional pollutant levels have been measured
continuously at a central monitoring location in the study communities. Measures of
regional pollutants included hourly concentrations of NO
2
(determined from NO
x
-NO
measured by chemiluminescence), 24-hour PM
2.5
concentrations (measured by the
Federal Reference Method (FRM)) or hourly PM
2.5
concentrations (measured by Beta
Attenuation Monitor (BAM)), 24-hour PM
10
concentrations (measured by the FRM) or
hourly PM
10
concentrations (measured by BAM or Tapered Element Oscillation
Microbalance (TEOM)), and hourly O
3
concentrations (measured on ultraviolet
photometers). PM data collected from BAM and TEOM monitors were adjusted, based
on comparison with collocated FRM data, to represent FRM equivalence, while the 8-
hour average concentration of O
3
from 10am to 6pm in each community was used in
46
assessment of O
3
effects. For statistical modeling, the regional level of each pollutant was
computed as the mean of the six years of measurements from cohort recruitment (2002)
to the recording of lung function tests (2007).
Statistical Methods
We fitted multiple linear regression models to investigate associations of FVC
and FEV
1
with each of the above indicators of NRAP exposure. A base model was first
developed using variables that are known predictors of lung function or were suspected to
influence lung function measurements. This base model included age at time of lung
function testing, sex, an interaction term between age and sex, race, an indicator for
Hispanic ethnicity, log of height (measured at time of lung function testing) and its
squared value, body-mass index (BMI) and its squared value, presence of acute
respiratory illness during lung function testing, indicator variables for which field
technician administered the test, and indicator variables for study community. A log
transformation of each pulmonary function measure was used to satisfy the assumptions
of linear regression.
The NRAP exposure values for each child from the spatial land-use regression
model were deviated from their respective community-specific mean. These community-
specific centered NRAP exposure values are constructed by design to be orthogonal
(uncorrelated) to cross-community regional exposures, which allows for the simultaneous
modeling of near-roadway and regional exposures. Each deviated exposure metric was
entered into the base model one at a time to test its association with lung function. Health
effect estimates are reported as the percent change in lung function per increase of two
47
standard deviations in the corresponding exposure. Distance-based exposures were
categorized, and health effect estimates are reported as the percent change in lung
function compared to the reference category.
Effects of regional air pollution on FVC and FEV
1
, either individually or in
combination with NRAP, were assessed with a mixed model that included a random
intercept for each study community. With the exception of indicator variables for study
community, the same set of adjustment variables were included as above. The health
effect estimate for each regional pollutant was scaled to the corresponding range of that
pollutant across study communities. Potential effect modification of NRAP by regional
air pollutants was assessed by formal testing using appropriate interaction terms between
the two pollutants.
For school level analyses, a mixed model that included a random intercept for
each school and a fixed effect for community was used. To model the joint effects of
NRAP at each child’s residence and school attended, the residential assigned exposure
was deviated from the school assigned exposure and both were included in the same
mixed model.
To examine the robustness of findings, additional sensitivity analyses were
performed, including adjustment for potential confounders and stratification by sex and
asthma status. Covariates considered in statistical analyses included race, Hispanic
ethnicity, history of doctor-confirmed diagnosis of asthma, parental income and
education, health insurance coverage, in utero exposure to maternal smoking, secondhand
tobacco smoke (SHS) exposure and presence in the home of pets, mold, or a gas stove.
Variables selected for further adjustments were chosen based on prior literature.
48
Complete covariate information for sex, race, and Hispanic ethnicity was obtained, while
most of the rest of the data had less than 10% missing, although 18% of responses for
income was missing, Missing indicators were included in models for incomplete
covariate information in order to maintain sample size when comparing across models.
Differences between subgroups were examined by testing for effect modification as
described above.
In all analyses, we assumed a two-sided alternative hypothesis and 0.05
significance level. All analyses were performed using Statistical Analysis System (SAS
version 9.2; SAS Institute Inc., Cary, NC).
49
Supplemental Table 2.1. Characteristics of participants and non-participants in lung function testing.
a
Due to missing values, denominators (n) for each percentage may differ.
b
P-value, adjusted for community, comparing the characteristic distribution of non-participating
subjects to subjects with lung function measurements.
Participants Non-participants
N
(total=1811) %
a
N
(total=402) %
a
P-value
b
Male 871 48.1 232 57.7 <0.01
Race
Asian 86 4.8 19 4.7 0.13
Black 39 2.2 16 4.0
Don't Know 239 13.2 71 17.7
Mixed 229 12.6 51 12.7
Other 486 26.8 90 22.4
White 732 40.4 155 38.6
Hispanic ethnicity
Don't Know 92 5.1 34 8.5 0.12
Hispanic 1028 56.8 214 53.2
Not Hispanic 691 38.2 154 38.3
SES
Income
<$30,000 402 27.1 112 34.6 0.10
$30,000 or more 1084 73.0 212 65.4
Parental education
Did not finish high school 345 20.6 79 21.6 0.22
High school diploma or some college 854 51.0 201 55.1
College diploma or greater 477 28.5 85 23.3
Insurance 1508 89.3 321 88.9 0.58
50
Supplemental Table 2.1 Continued. Characteristics of participants and non-participants in lung function testing.
a
Due to missing values, denominators (n) for each percentage may differ.
b
P-value, adjusted for community, comparing the characteristic distribution of non-participating
subjects to subjects with lung function measurements.
c
Respiratory illness information collected only on subjects undergoing PFT.
Participants Non-participants
N
(total=1811) %
a
N
(total=402) %
a
P-value
b
Home characteristics/Potential exposures
Gas stove 1462 86.5 311 86.4 0.53
Dog 599 35.8 127 37.7 0.41
Cat 312 18.8 60 18.0 0.98
Mold past 12 months 172 10.5 36 10.2 0.71
Secondhand smoke exposure 67 3.8 17 4.5 0.64
In-utero exposure to maternal smoking 99 5.8 26 7.2 0.25
Health conditions
Acute respiratory illness
c
164 9.4 - - -
Medical diagnosis of asthma 334 19.5 94 24.3 0.08
51
Supplemental Table 2.2. Distribution of residential distances to freeways and other major roads for CHS participants in 8 study
communities.
Freeway (m)
a
Large road (m)
a
Total <500 500-1,000 1,000-1,500 >1,500
b
<75 75-150 150-300 >300
b
N n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%)
Anaheim 190 115 (62.2) 35 (18.9) 13 ( 7.0) 22 (11.9) 37 (20.0) 44 (23.8) 83 (44.9) 21 (11.4)
Glendora 310 41 (13.3) 23 ( 7.4) 46 (14.9) 199 (64.4) 31 (10.0) 41 (13.3) 77 (24.9) 160 (51.8)
Long Beach 131 16 (12.6) 37 (29.1) 18 (14.2) 56 (44.1) 20 (15.7) 29 (22.8) 49 (38.6) 29 (22.8)
Mira Loma 273 41 (15.4) 39 (14.7) 24 ( 9.0) 162 (60.9) 36 (13.5) 38 (14.3) 55 (20.7) 137 (51.5)
Riverside 167 36 (21.8) 42 (25.5) 23 (13.9) 64 (38.8) 17 (10.3) 26 (15.8) 38 (23.0) 84 (50.9)
San Dimas 249 80 (32.5) 80 (32.5) 62 (25.2) 24 ( 9.8) 26 (10.6) 42 (17.1) 73 (29.7) 105 (42.7)
Santa Barbara 265 110 (44.0) 50 (20.0) 32 (12.8) 58 (23.2) 52 (20.8) 56 (22.4) 72 (28.8) 70 (28.0)
Upland 226 39 (17.3) 46 (20.4) 54 (24.0) 86 (38.2) 35 (15.6) 33 (14.7) 51 (22.7) 106 (47.1)
Total 1811 478 (27.0) 352 (19.9) 272 (15.3) 671 (37.8) 254 (14.3) 309 (17.4) 498 (28.1) 712 (40.2)
a
Do not sum up to total population due to missing values.
b
Reference group for distance-related analyses.
52
Supplemental Table 2.3. Correlation of regional pollutants from central sites.
Pollutant
a
PM
2.5
PM
10
NO
2
O
3
(10am-6pm) 0.66 0.63 0.12
PM
2.5
0.80* 0.60
PM
10
0.06
a
Each pollutant represents the 24-hour average
concentration over the study period from 2002
until 2007 with the exception of O
3,
which
represents the 8-hour average from 10am to
6pm.
* P<0.05
53
Supplemental Table 2.4. Correlation of near-roadway predicted exposures.
Pollutant
NO
2
NO
x
NO 0.92 0.98
NO
2
0.98
54
2.8 REFERENCES
1. Knuiman MW, James AL, Divitini ML, et al. Lung function, respiratory
symptoms, and mortality: results from the Busselton Health Study. Ann Epidemiol
1999;9(5):297-306.
2. Islam T, Gauderman WJ, Berhane K, et al. Relationship between air pollution,
lung function and asthma in adolescents. Thorax 2007;62(11):957-63.
3. Gotschi T, Heinrich J, Sunyer J, et al. Long-term effects of ambient air pollution
on lung function: a review. Epidemiology 2008;19(5):690-701.
4. Downs SH, Schindler C, Liu LJ, et al. Reduced exposure to PM10 and attenuated
age-related decline in lung function. N Engl J Med 2007;357(23):2338-47.
5. Pujades-Rodriguez M, Lewis S, McKeever T, et al. Effect of living close to a
main road on asthma, allergy, lung function and chronic obstructive pulmonary
disease. Occup Environ Med 2009;66(10):679-84.
6. Dales R, Wheeler A, Mahmud M, et al. The influence of living near roadways on
spirometry and exhaled nitric oxide in elementary schoolchildren. Environ Health
Perspect 2008;116(10):1423-7.
7. Gauderman WJ, Vora H, McConnell R, et al. Effect of exposure to traffic on lung
development from 10 to 18 years of age: a cohort study. Lancet
2007;369(9561):571-7.
8. Brunekreef B, Janssen NA, de Hartog J, et al. Air pollution from truck traffic and
lung function in children living near motorways. Epidemiology 1997;8(3):298-
303.
9. Rosenlund M, Forastiere F, Porta D, et al. Traffic-related air pollution in relation
to respiratory symptoms, allergic sensitisation and lung function in
schoolchildren. Thorax 2009;64(7):573-80.
10. Gauvin S, Amro S, Zmirou D, et al. Road traffic, NO2 exposure and respiratory
function among children (VESTA study). International Journal of Vehicle Design
2001;27(1):251-61.
11. Janssen NA, Brunekreef B, van Vliet P, et al. The relationship between air
pollution from heavy traffic and allergic sensitization, bronchial
hyperresponsiveness, and respiratory symptoms in Dutch schoolchildren. Environ
Health Perspect 2003;111(12):1512-8.
12. McConnell R, Berhane K, Yao L, et al. Traffic, susceptibility, and childhood
asthma. Environ Health Perspect 2006;114(5):766-72.
13. Peters JM, Avol E, Gauderman WJ, et al. A study of twelve Southern California
communities with differing levels and types of air pollution. II. Effects on
pulmonary function. Am J Respir Crit Care Med 1999;159(3):768-75.
14. Franklin M, Vora H, Avol E, et al. Predictors of intra-community variation in air
quality. J Expo Sci Environ Epidemiol 2012;22(2):135-47.
15. Islam T, Urman R, Gauderman WJ, et al. Parental stress increases the detrimental
effect of traffic exposure on children's lung function. Am J Respir Crit Care Med
2011;184(7):822-7.
16. Aust AE, Ball JC, Hu AA, et al. Particle characteristics responsible for effects on
human lung epithelial cells. Res Rep Health Eff Inst 2002(110):1-65; discussion 7-
76.
55
17. Li N, Sioutas C, Cho A, et al. Ultrafine particulate pollutants induce oxidative
stress and mitochondrial damage. Environ Health Perspect 2003;111(4):455-60.
18. EPA. Integrated Science Assessment for Oxides of Nitrogen – Health Criteria
(Final Report). U.S. Environmental Protection Agency, 2008.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=194645 (accessed 26 Mar
2013).
19. Peters JM, Avol E, Navidi W, et al. A study of twelve Southern California
communities with differing levels and types of air pollution. I. Prevalence of
respiratory morbidity. Am J Respir Crit Care Med 1999;159(3):760-7.
20. HEI. Traffic-related air pollution: a critical review of the literature on emissions,
exposure, and health effects. HEI Special Report: Health Effects Institute, 2010.
21. Mauderly JL, Samet JM. Is there evidence for synergy among air pollutants in
causing health effects? Environ Health Perspect 2009;117(1):1-6.
22. Huang YC, Rappold AG, Graff DW, et al. Synergistic effects of exposure to
concentrated ambient fine pollution particles and nitrogen dioxide in humans.
Inhal Toxicol 2012;24(12):790-7.
23. Gauderman WJ, Avol E, Gilliland F, et al. The effect of air pollution on lung
development from 10 to 18 years of age. N Engl J Med 2004;351(11):1057-67.
24. Oftedal B, Brunekreef B, Nystad W, et al. Residential outdoor air pollution and
lung function in schoolchildren. Epidemiology 2008;19(1):129-37.
25. Zhu YF, Hinds WC, Kim S, et al. Concentration and size distribution of ultrafine
particles near a major highway. Journal of the Air & Waste Management
Association 2002;52:1032-42.
26. Appatova AS, Ryan PH, LeMasters GK, et al. Proximal exposure of public
schools and students to major roadways: a nationwide US - survey. Journal of
Environmental Planning and Management 2008;51(5):631-46.
27. Perez L, Lurmann F, Wilson J, et al. Near-roadway pollution and childhood
asthma: implications for developing "win-win" compact urban development and
clean vehicle strategies. Environ Health Perspect 2012;120(11):1619-26.
28. Hibbert ME, Hudson IL, Lanigan A, et al. Tracking of lung function in healthy
children and adolescents. Pediatr Pulmonol 1990;8(3):172-7.
29. Sears MR, Greene JM, Willan AR, et al. A longitudinal, population-based, cohort
study of childhood asthma followed to adulthood. N Engl J Med
2003;349(15):1414-22.
30. Morgan WJ, Stern DA, Sherrill DL, et al. Outcome of asthma and wheezing in the
first 6 years of life: follow-up through adolescence. Am J Respir Crit Care Med
2005;172(10):1253-8.
31. Wang X, Wypij D, Gold DR, et al. A longitudinal study of the effects of parental
smoking on pulmonary function in children 6-18 years. Am J Respir Crit Care
Med 1994;149(6):1420-5.
32. Gauderman WJ, Avol E, Lurmann F, et al. Childhood asthma and exposure to
traffic and nitrogen dioxide. Epidemiology 2005;16(6):737-43.
33. Jerrett M, Arain MA, Kanaroglou P, et al. Modeling the intraurban variability of
ambient traffic pollution in Toronto, Canada. J Toxicol Environ Health A
2007;70(3-4):200-12.
56
34. Hu SS, Fruin S, Kozawa K, et al. A wide area of air pollutant impact downwind
of a freeway during pre-sunrise hours. Atmos Environ 2009;43(16):2541-9.
35. Choi W, He ML, Barbesant V, et al. Prevalence of wide area impacts downwind
of freeways under pre-sunrise stable atmospheric conditions. Atmos Environ
2012;62:318-27.
36. Zhu Y, Hinds WC, Kim S, et al. Concentration and size distribution of ultrafine
particles near a major highway. J Air Waste Manag Assoc 2002;52(9):1032-42.
37. Gilbert NL, Goldberg MS, Beckerman B, et al. Assessing spatial variability of
ambient nitrogen dioxide in Montreal, Canada, with a land-use regression model.
J Air Waste Manag Assoc 2005;55(8):1059-63.
38. Venn AJ, Lewis SA, Cooper M, et al. Living near a main road and the risk of
wheezing illness in children. Am J Respir Crit Care Med 2001;164(12):2177-80.
39. van Vliet P, Knape M, de Hartog J, et al. Motor vehicle exhaust and chronic
respiratory symptoms in children living near freeways. Environ Res
1997;74(2):122-32.
40. Benson PE. Caline4 -- A dispersion model for predicting air pollution
concentration near roadways. Sacramento, California: California Department of
Transportation, Office of Transportation Laboratory, 1989:205 pp.
57
CHAPTER 3: Determinants of the Spatial Distributions of Elemental
Carbon and Particulate Matter in Eight Southern Californian
Communities
Urman R, Gauderman J, Fruin S, Lurmann F, Liu F, Hosseini R, Franklin M, Avol E,
Penfold B, Gilliland F, Brunekreef B, McConnell R
3.1 ABSTRACT
Emerging evidence indicates that near-roadway pollution (NRP) in ambient air
has adverse health effects. However, specific components of the NRP mixture
responsible for these effects have not been established. A major limitation for health
studies is the lack of exposure models that estimate NRP components observed in
epidemiological studies over fine spatial scale of tens to hundreds of meters. In this study,
exposure models were developed for fine-scale variation in biologically relevant
elemental carbon (EC). Measurements of particulate matter (PM) and EC less than 2.5
µm in aerodynamic diameter (EC
2.5
) and of PM and EC of nanoscale size less than 0.2
µm were made at up to 29 locations in each of eight Southern California Children’s
Health Study communities. Regression-based prediction models were developed using a
guided forward selection process to identify traffic variables and other pollutant sources,
community physical characteristics and land use as predictors of PM and EC
variation in
each community. A combined eight-community model including only CALINE4 near-
roadway dispersion-estimated vehicular emissions accounting for distance, distance-
weighted traffic volume, and meteorology, explained 51% of the EC
0.2
variability.
Community-specific models identified additional predictors in some communities;
however, in most communities the correlation between predicted concentrations from the
eight-community model and observed concentrations stratified by community were
58
similar to those for the community-specific models. EC
2.5
could be predicted as well as
EC
0.2
. EC
2.5
estimated from CALINE4 and population density explained 53% of the
within-community variation. Exposure prediction was further improved after accounting
for between-community heterogeneity of CALINE4 effects associated with average
distance to Pacific Ocean shoreline (to 61% for EC
0.2
) and for regional NO
x
pollution (to
57% for EC
2.5
). PM fine spatial scale variation was poorly predicted in both size
fractions. In conclusion, models of exposure that include traffic measures such as
CALINE4 can provide useful estimates for EC
0.2
and EC
2.5
on a spatial scale appropriate
for health studies of NRP in selected Southern California communities.
3.2 INTRODUCTION
Emerging evidence suggests that near-roadway air pollution is associated with
chronic respiratory, cardiovascular, and neurological diseases (1, 2). Considerable
uncertainty exists as to the components of the near-roadway pollutant mixture responsible
for chronic health effects. Oxides of nitrogen have been commonly measured to develop
near-roadway prediction models because of the close association between NO
x
and
vehicular emissions and the existence of inexpensive passive NO
x
monitors (2). Although
acute effects of NO
2
have been observed at ambient concentrations, toxicological studies
have identified components of ambient particulate matter as more likely to be responsible
for the chronic effects of near-roadway exposures. Recent epidemiological studies have
reported health associations with estimated exposure to particulate elemental carbon
(EC), employing models based on traffic metrics and other land use (3-5). Particles with
EC may also contain transition metals and organic compounds that cause oxidative stress
59
and inflammation known to be involved in the pathogenesis of asthma and other
respiratory diseases (6, 7). EC, especially smaller particles, carries these toxicologically
relevant particle components deep into the lung. However, there have been few exposure
models estimating components of particulate matter on a fine spatial scale of tens to a
few hundred meters that is relevant for epidemiological studies examining near roadway
effects.
In Southern California, EC is a useful marker for vehicular combustion products,
especially from diesel powered vehicles, which are the primary EC source (8). Smaller
contributions to ambient EC are made by wood smoke (little used in our study
communities), ship emissions, railways, and off-road vehicles (9). For this study, we
measured and modeled Southern California EC concentrations in the fine respirable
fraction less than 2.5 µm in aerodynamic diameter (EC
2.5
) and in a nanoscale size fraction
less than 0.2 µm (EC
0.2
) that we anticipated might better reflect the near-roadway
gradient in ultrafine particles in communities participating in the Children’s Health Study
(CHS), a large prospective study of cardio-respiratory health (10). The study is notable
for the fine spatial scale at which these measurements were made in order to assess small-
scale intra-community variation. Information on traffic, land use and other community
features were used to develop models of within-community exposure, based on
measurements made at selected locations of high and low traffic impacts in each study
community. We also measured and modeled intra-community variation in particulate
matter (PM) mass in the 2.5 and 0.2 µm size fractions (PM
2.5
and PM
0.2
). Additionally,
we assessed both within- and between-community variation of these pollutants.
60
3.3 MATERIALS AND METHODS
Study locations and air sampling
Air pollution samplers for size-resolved PM mass and components were deployed
from November 2008 until December 2009 in up to 29 informatively selected locations
within each of eight Southern California communities (see Figure 3.1) in which CHS
participants are currently being studied. Sample locations were selected from among
participants’ homes based on high or low impacts of freeway, non-freeway, and other
non-traffic sources. All samplers were deployed at the same time in each community for
two consecutive two-week periods during warm and cool times of year, for a total of four
two-week sampling periods per community. Size-resolved PM less than 0.2 (PM
0.2
) and
0.2 to 2.5 µm in aerodynamic diameter (PM
0.2-2.5
) were collected on modified Harvard
cascade impactors (11). PM
2.5
was estimated
by summing the PM
0.2-2.5
and PM
0.2
stage
data. EC was collected from different sampling lines and measured using a thermal-
optical transmittance method. Additional information on the selection of sampling
locations and on air monitoring is available in the 3.7 Supplementary Material section
and in a previous report (12).
Predictors of EC and PM mass
Potential predictors of EC included distance (and inverse distance) to roadways
and other sources, traffic density in distance buffers around sampling locations,
dispersion modeled traffic pollutant exposure, length of road and amount of green space
in buffers around sampling locations, population density and elevation. Predictors were
61
linked to GPS measurements made at the sampling locations using GIS software
(ArcGIS). Details are provided in the 3.7 Supplementary Material section.
Annual average daily traffic (AADT) volumes on roadways and truck percentage
were obtained from the California Department of Transportation (Caltrans) milepost data
for freeways and numbered state highways for 2009 (13) and Dynamap Traffic Count
(Version 10.2) datasets produced by TeleAtlas (Boston, Massachusetts) for other roads.
Roadway classification was based on the Functional Class Code (FCC) as found in the
Dynamap dataset. Density plots were generated within the GIS using a linear decay
function that approximated the decrease in ambient concentrations with increasing
distance away from roadways, i.e., decays to background between 150 and 300 meters
(14).
Estimates of the contributions of local on-road motor vehicle emissions to air
quality were obtained using the CALINE4 Gaussian line-source dispersion model (15).
The CALINE4 dispersion model uses distance to roadways, vehicle counts, vehicle
emission rates, and meteorological conditions as inputs. Although the CALINE4 model
provides estimates of the near-roadway contribution to EC and PM
2.5
(and of multiple
other near-roadway pollutants), these estimates are all highly correlated and should be
regarded as markers for the primary near-roadway mixture. Separate estimates were made
for the contribution of local traffic on freeways and on all other roadways (non-freeway
roads) to concentrations of EC and PM
2.5
. Total CALINE4 was computed as the sum of
the contributions from both freeway and non-freeway roads. Total CALINE4 was highly
correlated with the freeway component of CALINE but not with the non-freeway
component of CALINE. See 3.7 Supplementary Material section for more details.
62
Distances to freeways, active railways, combustion point sources (eg. a port or a
refinery), intermodal transportation facilities (for example, where train to truck transfer of
cargo occurs), and to the nearest Pacific Ocean shoreline were computed by GIS. To
provide another indicator of emissions proximity, roadway lengths within various buffer
distances (50, 100, 150, 200, 250, and 300 m radius) were computed for each FCC road
class and summed together to provide the total length in each buffer (16).
Population density data at the block group level were obtained from US Census
Bureau (2000 data projected to year 2010) via ESRI’s data repository. The population
density within 300m radius buffers of each sampling location was computed as an aerial
extent-weighted average of each block group’s density in the buffer.
Elevation data with ~10m resolution were obtained from the US Geological
Survey (USGS) website (http://seamless.usgs.gov). For each sampling location, we
computed a mean elevation at a neighborhood level based on 10m-grid elevation values
within a 300m buffer.
The Normalized Difference Vegetation Index (NDVI) is an indicator of live green
vegetation derived from satellite remote sensing data (17) and was included as a
predictive variable as a metric of the absence of traffic and other pollution sources.
Additional description of predictor covariates is provided in the 3.7
Supplementary Material section.
Regional pollutants
Additional measures of regional air pollutants were continuously collected at
regulatory agency regional air monitoring stations in each of the study communities.
63
Pollutants of interest from these measurements included PM
2.5
and NO
x
as described
previously (18). Daily measures of these regional pollutants were integrated over the
multi-week study periods and were used in the modeling as modifiers of the association
between predictors and outcome.
Statistical methods
EC and PM samples collected across seasons were averaged to derive a single
eight-week average concentration for each sampling location. Sampling locations that
were not the same across seasons were not included in the primary analyses. At schools
and central sites where duplicate measurements were made, concentrations were
averaged. All eight-week averaged concentrations were natural log transformed to help
satisfy the normality and homoscedasticity assumptions of linear regression and to ensure
model predictions would be positive.
Because the focus of this study was to examine the factors that affected within-
community variation of fine and nanoscale EC (EC
2.5
and EC
0.2
) and particulate mass
(PM
2.5
and PM
0.2
), we implemented a strategy similar to the one used in Franklin et al.
(19) for NO
2
, NO, and NO
x
. To parse out the within-community variation from the cross-
community variation, we subtracted (or “deviated”) the mean concentration of pollutant Y
in community c from the pollutant measurement Y at location i in community c (dY
ci
= Y
ci
-Y
c
). We similarly performed a transformation of each predictor (X) previously listed (i.e.
dX
ci
= X
ci
– X
c
) to estimate a within-community distribution of the predictors. Some
predictor variables (CALINE4 estimates, traffic density) were positively skewed and
were log transformed (prior to deviating) to minimize the potential influence of very high
64
values. We also evaluated the components of within-community and between-community
variation for each pollutant using the VARCOMP procedure in SAS version 9.3 (SAS
Institute Inc., Cary, NC).
We developed both community-specific models and a combined (eight-
community) model for each pollutant. For community-specific analyses, point sources
farther than 10km were excluded from consideration. In the combined model, we
weighted the distance to intermodal facilities by dividing by the mean distance (i.e.
dX
ci
/Xc) , thus giving less weight to communities without intermodal facilities within
10km. We excluded from the combined models combustion point source locations with
NO
x
emission rates that were greater than 50 tons per year as predictors because most
communities had none and distance to a shoreline was excluded because variation at the
within-community scale was not meaningful for most communities that were many
kilometers inland.
We calculated Pearson correlations between each of the deviated pollutants (dY
ci
)
and predictors (dX
ci
) to understand how they varied together within communities.
Supervised forward selection, similar to the one used in the European ESCAPE study
(20), was used to develop combined models as well as community-specific models.
Model selection began with the predictor that produced the highest adjusted R
2
and that
had a beta coefficient in the expected direction. Remaining predictors were then added
one at a time until the addition did not result in at least a 1% improvement in the adjusted
R
2
. The direction of all beta coefficients was checked during each step of this model
selection process. From this group of predictors, those that were not significant at the
0.10 level were dropped one at a time starting with the least significant predictor.
65
Variables that had a variance inflation factor (VIF) greater than 3 were also dropped from
the model.
Leave-one-out cross-validated (LOOCV) and (for the combined models) leave-
one-community-out cross-validated (LOCOCV) R
2
were calculated to assess how well
the models performed across communities and how transferable they might be to other
communities in Southern California. To examine the performance of the combined model
in each community, we took the predicted dY
ci
from this model (using the LOOCV
approach) and calculated the correlation with the observed values by community. The
correlation was then squared to estimate the proportion of variation in each community
that was explained by the combined model.
Finally, we fitted a mixed-effects model to consider the possibility that
community-level variables might modify intra-community prediction models. The
community-level variables considered included measured concentrations of PM
10
and
NO
x
from the regulatory agency regional air stations as well as the community average of
measured EC
2.5
and of selected predictors (population density, which might be an
indicator of additional combustion sources in densely populated areas, and community-
average distance to the shoreline as a proxy for meteorological characteristics that might
affect the models). This mixed model took the form dY
ci
=α + β
1
*dX
ci
+ β
2
*dX
ci
*C +
β
4
*dZ
ci
+ f
c
*dX
ci
+ e
c
+ e
ci
, where Y was the outcome (EC
0.2
or EC
2.5
), X was the predictor
of interest, C was a possible community-level modifier, Z were adjustment covariates
from the combined model, and the e
c
and f
c
were community-level random effects,
assumed to be bivariate normally distributed and independent of the subject-specific
random effect e
ci
. The parameter β
2
and its corresponding level of statistical significance
66
was used to determine whether the intra-community relationship between X and Y varied
by C. All analyses were conducted using SAS version 9.3.
3.4 RESULTS
In this section, the sample size and distribution of each exposure outcome is
described. The distribution of key covariates and their univariate association with PM and
EC by community and size fraction is illustrated. The predictors from a unified model
across all communities and the associated heterogeneity in the cross-validated predictions
from this model in individual communities were examined, and the results from this
approach were compared with a more traditional community-specific modeling approach.
In sensitivity analyses, we examined the influence of regional pollution and other
community characteristics on the heterogeneity of effects of near-roadway traffic metrics.
Because these models will be applied to health outcomes in the CHS, a combined model
of EC
2.5
exposure was developed restricted to communities in which within-community
variability was well-predicted by near-roadway traffic metrics.
Characterization of sampling sites and key predictors
Samples were collected from 228 locations across the eight communities. Of these
locations, 177 remained unchanged across the entire study design and were eligible for
the analysis of eight-week average concentrations across seasons, as described in the 3.7
Supplementary Material section. After eliminating locations with invalid data, mostly due
to equipment failures or power interruptions, we had 148, 152, 130, and 137 locations
with valid eight-week EC
2.5
, EC
0.2
, PM
2.5
, and PM
0.2
data, respectively.
67
The community-specific distribution of the average of the eight-week
measurements is shown in Figure 3.2 and the corresponding geometric means and
coefficients of variation are shown in Table 3.1. The smaller nanoscale fraction of EC
(0.2µm) had a similar pattern of within-community variability to EC
2.5,
based on the
coefficients of variation in Table 3.1. The within-community variation of EC
2.5
and EC
0.2
was about half that of the between-community variation (Table 3.2). In contrast, the
within-community variance of PM
0.2
was greater than its between-community variance.
The between-community variance of PM
2.5
was about ten times as large as its generally
small within-community variance. However, one community (Mira Loma) contributed
most of the between-community variability in PM
2.5
(Figure 3.2). There was strong
correlation between EC
2.5
and EC
0.2
across all locations (0.83; Supplemental Table 3.1).
The community adjusted (within-community) correlation was almost as large (0.76).
Both size fractions of particulate mass were weakly correlated with one another and with
each EC size fraction.
There was substantial variability in the distribution of the predictor variables at
sampling locations in different communities, for example for CALINE4 EC estimates and
population density, which were key explanatory variables in combined models of
exposure (as described below). The CALINE4-modeled freeway concentration varied by
almost 7-fold, considerably more than the CALINE4-modeled concentration from all
other roads. Mean population density varied by approximately 3-fold (See Figure 3.3).
The strongest correlations of measured EC pollutant concentrations in both size
fractions were with traffic metrics (Table 3.3). Correlations with freeway and with the
sum of freeway and non-freeway CALINE4 were approximately 0.7. Weaker correlations
68
were observed with other predictors. Correlations of traffic and other predictor variables
with PM
2.5
and PM
0.2
were much weaker than with EC
2.5
and EC
0.2
with few exceptions
(e.g. NO
x
point sources with PM
0.2
). We also examined the community-specific
correlations of EC
2.5
and EC
0.2
with potential predictor variables and found that there was
considerable heterogeneity across communities for each pollutant (Supplemental Tables
3.2 and 3.3). For EC
2.5
, there were consistently strong and common traffic associations in
five of the eight communities. A different traffic metric, truck count on the nearest
freeway, was strongly correlated with EC
2.5
in Long Beach. However, in Mira Loma and
San Dimas, traffic was poorly correlated with EC
2.5
. EC
0.2
showed strong traffic
associations in all but one community (San Dimas).
Combined eight-community model
CALINE4 was included in the best combined model for EC in each size fraction,
and some form of traffic exposure was included in the best model for every pollutant
studied (Table 3.4). After the traffic metrics, population density had the next largest
effect estimates. Cross validation R
2
for both EC sizes were about 0.5, while the cross
validation R
2
for PM mass were much smaller.
Although the LOOCV R
2
was 51% for EC
2.5
from the combined model, the
performance varied substantially when applied to each community separately. For
example, the R
2
of predicted with measured EC
2.5
in Santa Barbara was 82% but the
model explained none of the intra-community variation in Long Beach (Table 3.5).
Concentrations of EC
2.5
were also poorly predicted in Mira Loma and San Dimas. The
EC
0.2
model predictions explained at least 30% of the measured variation in seven of the
69
eight communities. In contrast, the model for PM
2.5
explained 30% or more of the
measured variation in only two communities and the model for PM
0.2
in no community.
Community-specific models
Models identifying community-specific predictors were fitted for EC
2.5
(Table
3.6). These models explained the variation in some communities, particularly San Dimas
(in which traffic metrics did not contribute to the model), Long Beach and Riverside,
considerably better than the combined models, but the R
2
was still relatively low in San
Dimas. Community-specific models did not substantially improve the R
2
’s for EC
0.2
,
except in San Dimas, in which traffic metrics did not contribute to the model and
community-specific R
2
increased only to 0.21 (Table 3.7). Either freeway or (correlated)
total CALINE was selected in most community-specific models.
Sensitivity analyses
As a post-hoc analysis for EC, we developed models in just those communities in
which the combined model predicted at least 30% of community-specific variability
(Table 3.5). For EC
2.5,
a five-community model (excluding Long Beach, Mira Loma, and
San Dimas) was able to explain 66% of the measured variation (Supplemental Table 3.4),
compared with 51% using data from all communities (from Table 3.4). The five-
community model included total CALINE4, population density, and NDVI as predictors.
Only San Dimas was excluded from the sensitivity model for EC
0.2
, which like the model
for all eight communities included only total CALINE4. The LOOCV R
2
was 53%,
70
compared with 49% in all communities from Table 3.4. The LOCOCV R
2
’s were similar
to those for LOOCV.
In previous analyses examining within-community NO
x
variability, we observed
larger CALINE4 effects in communities with lower average concentrations (19).
Therefore, we investigated whether the heterogeneity in traffic effect estimates in
different communities might be explained by the average of community exposures and by
the average of the continuous regional pollutant measurements made at the central site
monitors during the time of sampling in each community. We focused on the variability
in effects of total CALINE4 as this was a strong predictor in the eight-community models
for both EC
2.5
and EC
0.2
and was selected in a majority of the community-specific
models. In some models that included an interaction between total CALINE4 and these
community-level modifiers, there was substantial improvement in the LOOCV R
2
(Table
3.8). In the EC
2.5
model, we found that the strongest association between total CALINE4
and EC
2.5
were in communities with low levels of regional NO
x
(Supplemental Figure
3.1), while associations between EC
0.2
and total CALINE4 were strongest in communities
nearest to the shoreline (Supplemental Figure 3.2).
3.5 DISCUSSION
Notable features of this analysis included (1) the heterogeneity of the strength of
EC-traffic associations across communities and the potential to partially explain this
variability by community characteristics, a finding with potentially broad implications for
spatial exposure modeling; (2) a comparison of model performance across multiple
communities using two complementary approaches (a combined model and more
71
traditional community-specific models); (3) models that were able to predict EC on a fine
spatial scale; and (4) model development for a novel size fraction (PM
0.2
and EC
0.2
).
Combined prediction models that captured the fine spatial scale of EC across
eight communities were developed. The combined model R
2
’s were substantially
improved by accounting for community characteristics that modified the effects of
CALINE4 (from 51% to 57% for EC
2.5
by accounting for regional NO
x
and 49% to 61%
for EC
0.2
by accounting for average shoreline distance; Table 3.8 and Supplemental
Figures 3.1 and 3.2). It is possible that in the setting of a noisy and more complex high
regional pollution background that a small local traffic effect on EC
2.5
was not identified,
whereas in a community with little transported pollution, the effect of small primary
traffic sources was apparent. This finding is consistent with our previous report with a
larger number of communities in which we measured the within-community variation in
NO, NO
2
and NO
x
(19). In that study, CALINE4 predicted variation better in less
polluted communities outside of the Los Angeles air basin, where regional pollution is
lower, than within the basin. For EC
0.2
, it is not entirely clear why the effect of CALINE4
is stronger closer to the shoreline, but we speculate that it might be due to onshore winds
creating clearer gradients of EC
0.2
concentrations and producing larger contrasts in
communities closer to the shoreline with cleaner background concentrations. These
findings, especially the variability by regional pollution levels, have potentially broad
relevance to modeling of near-roadway pollution and merit further study in other
geographic regions.
In community-specific models, the inclusion of truck counts on the nearest
freeway in Long Beach improved the EC
2.5
R
2
substantially compared to the combined
72
model (0.54 in Table 3.6 and 0 in Table 3.5). There were three freeways in Long Beach
with markedly different truck counts. However, there was little variation in truck counts
within each freeway and there was weak association of EC
2.5
with distance to nearest
freeway (Supplemental Table 3.2). Therefore, truck counts may have reflected
background levels associated with the areas of the city corresponding to the three
freeways rather than a near-roadway effect of truck exhaust. Long Beach is a coastal
community with a major shipping port, refineries, and rail activity. It has complex air
flows due to convergence of westerly and southerly onshore flows during the day. The
CALINE4 estimates of near-roadway traffic impacts may be less accurate than in other
communities because (1) the modeling relied on a single meteorological monitoring site
which did not represent the complex flows and (2) the on-road emission estimates
probably underestimated the heavy truck traffic on arterial corridors due to traffic from
the port. Riverside is another relatively large community with heavy truck traffic en route
from the port to large local warehouse transfer facilities. In this community, truck count
on the nearest freeway and distance to these intermodal transfer facilities improved EC
2.5
prediction. EC
2.5
variability was also explained by predictors other than near-roadway
metrics in San Dimas. Higher elevation predicted lower EC
2.5
(Table 3.6) and, along with
vegetation, EC
0.2
(Table 3.7). While the CALINE4 predictions capture some aspects of
meteorology, the absence of local meteorological measurements may have contributed to
the poor predictability of EC in San Dimas. This community extends into the foothills of
the San Gabriel Mountains, which may have strong and local influence on wind speed
and direction that is not identified by the two closest monitoring stations (Azusa and
Pomona, both far away from the major terrain features). A Mira Loma specific model for
73
EC
2.5
was not reported due to a technical problem with samplers during one particle
collection wave, leaving only 13 locations for analysis. The small sample size might
explain the poor model fit of the combined model in this community.
The leave-one-community-out cross-validated R
2
was approximately 50% for
both EC
2.5
and EC
0.2
, based on the combined models. However, the poor performance of
these models in some communities indicates that further study is warranted to determine
how transferable the models could be to other Southern California communities, and
whether communities to which the combined model would not be transferable could be
identified a priori based on complexity of geographic topology, meteorology and other
pollution sources (eg. San Dimas and Long Beach). A few other studies in Europe and
North America have found that exposure models developed from land use in one city had
reduced R
2
when used to predict measurements in other cities (21-23).
The combined EC
2.5
model predicted poorly in some communities because the
near-roadway exposure metrics that determined variability overall did not explain
variability in some communities. European studies that have examined the variability of
EC
2.5
through land-use regression modeling also found measures of traffic to be
important predictors but to vary between regions (4, 20, 24-27). In the ESCAPE study,
separate models were developed using information on local land use for each of 20 large
European cities, and cross-validated R
2
for the separate models ranged from 40% to 95%
(20). The R
2
’s of 36% to 77% in the community-specific models for EC
2.5
in our
communities (Table 3.6) were somewhat lower but, as in the European studies, were
heterogeneous across communities.
74
Possible reasons for the higher model R
2
in ESCAPE include a wider range of
measured EC
2.5
concentrations to be explained by traffic and other land uses across large
metropolitan regions in Europe, compared with the range in the generally smaller
communities in the CHS as the focus of our modeling was to predict fine spatial scales of
near-roadway mixtures (e.g. 50-150m). Comparing the exposures across the two studies
is not straight forward, because EC
2.5
was assessed by light absorbance of PM
2.5
in
ESCAPE. Although absorbance is highly correlated with measured EC
2.5
, the relationship
between the two measurements of EC can vary depending on location (28). The ratio of
the range to mean of PM
2.5
absorbance was provided in the European study (29), and we
have calculated this index in each CHS study community in order to compare the
variability across studies (Supplemental Table 3.5). In ESCAPE, this measure of
variability ranged from 68% in Gyor (Hungary) to 235% in London/Oxford (United
Kingdom) and about half of the 20 study areas in ESCAPE had values that were greater
than 100%. In contrast, only one among our eight study communities had a value greater
than 100% for EC
2.5
(101% in Anaheim). Levels of residential EC
2.5
in European cities
can be high (30, 31), because unlike Southern California there is a high proportion of
diesel powered passenger vehicles that travel on secondary roads in close proximity to
residences.
We observed stronger correlations between EC
2.5
with freeway sources, compared
with non-freeway sources, of CALINE4-predicted concentrations. A large proportion of
EC is attributable to diesel exhaust from trucks, which are found largely on freeways in
Southern California (8, 32) and elsewhere (33). A Cincinnati study also found diesel
sources, including length of bus routes and truck intensity within 300 meters of
75
monitoring locations, to be strong predictors of EC
2.5
concentrations (34). In a Boston
study, the strongest traffic predictor of EC
2.5
(measured via absorbance) was length of
roadway in a 200m buffer (35).
For EC
2.5
population density contributed to the eight-community model,
suggesting either that there were other anthropogenic sources of EC
2.5
or that population
density provided additional information on traffic emissions. For example, residential
population might be an indicator of “cold starts” that produce more EC after a prolonged
period with the engine off. Other studies have also found population density to predict
near-roadway air pollutants (20, 25, 27). However, in our study this variable added little
to the R
2
(~1%) in models also containing total CALINE4.
To our knowledge, few previous studies have examined the within-community
spatial distribution and predictors of EC in size fractions that are smaller than 2.5µm. We
hypothesized that EC
0.2
(and to a lesser extent PM
0.2
) would be better markers for fresh
near-roadway combustion than EC
2.5
, which might contain a larger proportion of
regionally transported EC. A recent study showed that a larger proportion of EC along
busy roads was found in the smaller 0.25 µm size fraction compared to the 2.5-0.25 µm
size fraction and the concentration of these fractions were highest on a stretch of freeway
containing a large number of diesel trucks (32). Contrary to our hypothesis and these
previous results, in our study the ratio of within- to between-community variance was
similar for both size fractions (Table 3.2), suggesting that the accumulation mode (i.e, 0.1
to 1 µm), which should account for most of the EC
0.2
mass, has a substantial transported
regional component in Southern California. EC
0.2
was highly correlated with EC
2.5
(Supplemental Table 3.1), both within communities (R=0.76) and across all
76
measurements (R=0.83). In addition, only total CALINE4 was a predictor of the within-
community variation in EC
0.2
in the combined model. Although EC
0.2
is likely to
penetrate more deeply into lungs and therefore may be a more biologically relevant
exposure, these results suggest that modeling EC
0.2
exposure may provide little
information for assessing health effects of within-community exposure to primary traffic
source beyond what is provided by EC
2.5
(or by the CALINE4 estimate).
The within-community variability of PM
0.2
was almost as large as for EC (based
on coefficients of variation in Table 3.1), but the cross-validation R
2
for within-
community variability in PM
0.2
(0.12 from Table 3.4) was poor. Although PM
0.2
is
enriched with ultrafine particles less than 0.1 µm in diameter, which have large spatial
gradients downwind from major roadways (14, 36), ultrafine particles have little mass
and most of the PM
0.2
mass is likely to be greater than 0.1 µm in diameter. Determinants
of PM
0.2
variability merit further investigation.
The model for PM
2.5
, which included length of road in a 100m buffer, nearest
freeway truck count, population density, elevation and distance to the nearest point
source of NOx (from Table 3.4), nevertheless poorly predicted the within-community
variation in our study (cross-validated R
2
0.17). This is consistent with the regional
character of PM
2.5
mass and with other studies that have shown little variation with local
traffic predictors (35). Given the relatively smaller within- to between-community ratio in
variance for PM
2.5
(Table 3.2), it is unlikely that predicted exposures from within-
community models would contribute substantially more to understanding health effects
than measurements from a single central site monitoring station. Other studies have
reported better R
2
for PM
2.5
based on traffic and land use, perhaps because they have
77
been conducted across generally larger metropolitan regions with both regional and local
variation in PM
2.5
(20, 25, 37).
In conclusion, a combined model for land use effects on EC on a fine spatial scale
within multiple Southern California communities was generally robust, although there
was marked heterogeneity in effect estimates for the CALINE4 near-roadway traffic
metric that could partially be explained by regional pollutant concentrations and distance
to shoreline. Predictors other than near-roadway traffic metrics substantially improved
model fit in some communities. In addition, traffic prediction models for a novel 0.2 size
fraction we had hypothesized would be a better marker for near-roadway pollution were
not substantially better than for EC
2.5
.
78
3.6 TABLES AND FIGURES
Table 3.1. Eight-week geometric mean concentration of measured pollutants (in µg/m
3
) and coefficient of variation (CV in %) in each
community.
a
Represents the geometric mean and CV using all measurements across all communities.
The geometric CV is a mixture of between and within-community variation.
b
Unweighted average of the reported community specific geometric means and CVs. The
geometric CV represents the within-community variation.
EC
2.5
EC
0.2
PM
2.5
PM
0.2
Geo Geo
Geo Geo
Geo Geo
Geo Geo
Community Mean CV Mean CV Mean CV Mean CV
Anaheim (AN)
1.15 20.5
0.46 22.6
14.97 8.0
2.56 15.7
Glendora (GL)
0.77 21.5
0.26 17.7
13.52 7.4
1.70 27.1
Long Beach (LB)
1.17 10.6
0.53 14.0
14.15 8.9
2.62 13.5
Mira Loma (ML)
0.87 9.6
0.41 8.9
21.98 6.7
2.68 10.7
Riverside (RV)
0.95 21.7
0.35 25.3
15.26 6.2
2.34 19.1
Santa Barbara (SB)
0.55 29.3
0.20 39.8
11.42 11.3
2.05 22.8
San Dimas (SD)
1.12 10.1
0.44 15.5
14.48 6.7
2.45 14.2
Upland (UP) 0.65 18.6 0.34 17.4 11.94 4.3 2.76 9.6
All measurements
a
0.86 33.2
0.36 36.6
14.45 20.7
2.34 24.0
8 community mean
b
0.90 17.7
0.37 20.1
14.71 7.4
2.40 16.6
79
Table 3.2: Components of within- and between-community variance (and ratio of within-
to between-community variance).
Within Between
Variance Variance Ratio
EC
2.5
0.03 0.05 0.53
EC
0.2
0.006 0.01 0.54
PM
2.5
1.23 11.0 0.11
PM
0.2
0.14 0.12 1.14
80
Table 3.3. Pairwise correlations between deviated (community-centered) pollutants levels and potential predictors.
Predictors EC
2.5
a
EC
0.2
a
PM
2.5
a
PM
0.2
a
CALINE4
a
Freeway 0.69 ** 0.65 ** 0.39 ** 0.29 **
Non-freeway 0.35 ** 0.39 ** 0.16
0.16
Total 0.72 ** 0.71 ** 0.41 ** 0.28 **
Distance
Freeway -0.49 ** -0.48 ** -0.23 ** -0.21 *
Large arterial roads -0.04
-0.16
-0.10
-0.01
Traffic density
a
150m buffer 0.49 ** 0.57 ** 0.34 ** 0.14
300m buffer 0.53 ** 0.54 ** 0.30 ** 0.15
Freeway truck count 0.17 * 0.17 * 0.26 ** 0.17
Road buffers (all roads)
50m 0.23 ** 0.26 ** 0.22 * 0.26 **
100m 0.33 ** 0.35 ** 0.28 ** 0.14
150m 0.33 ** 0.37 ** 0.22 * 0.08
200m 0.33 ** 0.37 ** 0.22 * 0.08
250m 0.31 ** 0.35 ** 0.20 * 0.07
300m 0.33 ** 0.34 ** 0.17
0.07
Elevation -0.45 ** -0.47 ** -0.36 ** -0.22 **
Population density (300m buffer) 0.33 ** 0.27 ** 0.30 ** 0.26 **
Normalized difference vegetation index (NDVI) -0.29 ** -0.24 ** -0.19 * -0.05
Distance to railway -0.40 ** -0.44 ** -0.18 * -0.17
Distance to intermodal facility (weighted) -0.21 * -0.26 ** -0.15
-0.14
Distance to point source of NO
x
(10-50 tons/yr) -0.11 -0.03 -0.04 -0.18 *
a
On log scale (and in all following tables).
* P<0.05, ** P<0.01
81
Table 3.4. Prediction models across all eight communities
a
.
Predictors
a
EC
2.5
EC
0.2
PM
2.5
PM
0.2
Logged freeway CALINE4
0.059
Logged total CALINE4 0.255 0.291
Total length of roads in 50m buffer
0.074
Total length of roads in 100m buffer
0.039
Freeway truck count
0.041
Elevation
-0.039
Population density (300 m buffer) 0.050
0.037 0.071
Distance to point source of NO
x
(10-50 tons)
-0.048
Adjusted R
2
0.53 0.51 0.27 0.16
Leave-one-out cross-validated (LOOCV) R
2
0.51 0.49 0.21 0.12
Leave-one-community-out cross-validated (LOCOCV) R
2
0.48 0.47 0.20 0.14
a
Reported betas are scaled to two standard deviations of deviated predictors across all eight
communities as follows: 1.6 units for logged freeway CALINE4, 1.1 units for logged total
CALINE4, 116.7 meters for total length of roads in 50m buffer, 340.9 meters for total length
of roads in 100m buffer, 6250 trucks for freeway truck count, 84.5 meters for elevation, 1574
individuals/km
2
for population density, and 3210 meters for distance to point source of NO
x
(10-50 tons).
82
Table 3.5. Leave-one-out cross-validated (LOOCV) R
2
for prediction models in Table 3.4
applied to each community.
Towns EC
2.5
EC
0.2
PM
2.5
PM
0.2
Anaheim 0.64 0.91 0.26 0.03
Glendora 0.74 0.42 0.24 0.14
Long Beach 0.00 0.54 0.27 0.14
Mira Loma 0.02 0.37 0.11 0.04
Riverside 0.50 0.50 0.03 0.16
Santa Barbara 0.82 0.80 0.39 0.23
San Dimas 0.08 0.09 0.43 0.12
Upland 0.51 0.54 0.23 0.11
83
Table 3.6. Community specific EC
2.5
models.
Predictors
a
AN GL LB ML
b
RV SB SD UP
Logged freeway CALINE4 0.395
-
Logged total CALINE4
0.324
- 0.304 0.338
0.222
Elevation
-
-0.214
Population density (300m buffer)
-
0.097
Distance to shoreline
-
NDVI
- -0.252
Freeway truck count
0.120
0.204
Distance to intermodal facility
-0.048
Adjusted R
2
0.67 0.73 0.61
0.79 0.82 0.44 0.50
LOOCV R
2
0.58 0.70 0.54 - 0.75 0.77 0.36 0.44
N 18 22 17 13 21 18 16 23
a
Reported betas are scaled to two standard deviations of the deviated predictors across all eight
communities as follows: 1.6 units for logged freeway CALINE4, 1.1 units for logged total CALINE4, 84.5
meters for elevation, 1574 individuals/km
2
for population, 5260 meters for distance to shoreline, 0.1 for
NDVI, 6250 trucks for freeway truck count, and 4300 meters for distance to intermodal facility.
b
A stable model could not be fit for Mira Loma (ML), which was likely attributable to the small number of
samples available for this analysis.
84
Table 3.7. Community specific EC
0.2
models (reported betas in each column followed by R
2
for each community)
a
.
Predictors
a
AN GL LB ML RV SB SD UP
Logged freeway CALINE4
0.351
Logged non-freeway CALINE4
0.164
Logged total CALINE4 0.446 0.170
0.250 0.508
Distance to FCC3
-0.184
-0.141
Logged traffic density (150m radius)
0.142
Elevation
-0.145
-0.202
NDVI
-0.107 -0.113
Adjusted R
2
0.91 0.48 0.56 0.57 0.49 0.82 0.31 0.70
LOOCV R
2
0.86 0.38 0.49 0.52 0.40 0.74 0.21 0.57
N 17 22 19 19 20 17 16 22
a
Reported betas are scaled to two standard deviations of the deviated predictors across all eight communities as
follows: 1.6 units for logged freeway CALINE4, 0.8 units for logged non-freeway CALINE4, 1.1 units for logged total
CALINE4, 393 meters for distance to FCC3, 1.9 units for logged traffic density, 84.5 meters for elevation, and 0.1 for
NDVI .
85
Table 3.8. Leave-one-out cross-validated (LOOCV) R
2
of various hierarchical combined
models
a
.
Community level modifier of
total CALINE4 EC
2.5
EC
0.2
No modifier (from Table 3.4)
0.51 0.49
NO
x
central site
0.57 0.52
PM
2.5
central site
0.54 0.54
Average of community EC
2.5
measurements
0.52 0.50
Averaged population density
0.52 0.54
Averaged shoreline distance 0.54 0.61
a
NO
x
and PM
2.5
measurements came from fixed monitoring
sites, while EC
2.5
, population density, and distance to
shoreline were averaged across pollution measurement sites
by community.
86
Figure 3.1. Map of CHS communities.
87
Figure 3.2. Distribution of eight-week averaged concentrations of EC and PM in 2.5 and 0.2 µm size fractions.
EC
2.5
µg/m
3
EC
0.2
µg/m
3
PM
0.2
µg/m
3
PM
2.5
µg/m
3
88
Figure 3.3. Distribution of selected predictor variables by community.
89
Figure 3.3 Continued. Distribution of selected predictor variables by community.
90
3.7 SUPPLEMENTARY MATERIAL
Supplementary Methods
Study locations and air monitoring
Air pollution samplers for size-resolved particulate mass and components were
deployed across eight Southern California communities (Anaheim, Glendora, Long
Beach, Mira Loma, Riverside, San Dimas, Santa Barbara, and Upland) in which
Children’s Health Study (CHS) participants were under study. These eight communities
have widely differing background pollution profiles, roadway networks, and topography.
Samplers were placed in up to 29 informatively selected locations within each
community to capture size-resolved concentrations and differences in particle mass and
composition in size fractions less than 0.2 (PM
0.2
) and 0.2 to 2.5 µm in aerodynamic
diameter (PM
0.2-2.5
). In each community, sampling locations included up to 24 CHS
participants’ homes, 3-4 elementary schools formerly attended by the CHS cohort, and
the local regional air-monitoring location. Homes were chosen based on a stratified
sampling design to maximize variability of pollutants with respect to potential sources.
Three home locations within each of eight strata were selected based on the following
three criteria: (1) highest one-third versus lowest two-thirds of freeway predicted
pollutants from the CALINE4 dispersion model (details of CALINE4 are found below),
(2) highest one-third versus lowest two-thirds of non-freeway predicted pollutants from
the CALINE4 dispersion model, and (3) positive versus negative predicted residuals from
a kriged surface of residuals from a previous land use regression model for NO
x
(19).
Sampling took place from November 2008 until December 2009 on a rotating
schedule in two communities at a time. All samplers were deployed at the same time
91
within each community for two consecutive two-week periods during warm and cool
times of year in southern California, for a total of four two-week sampling periods per
community. At each school and central site, co-located samplers were deployed to
determine the sampling precision. When measurements were not possible at the homes of
original participants, new home locations were selected in the community based on their
similarity to the criteria used to select the original participating home locations to
represent different traffic environments. Locations that were excluded from these
analyses because data from only a single season were available had measured values, on
average, that were similar to those community and season specific values that were used
in the analyses (data not shown).
PM
0.2-2.5
and PM
0.2
were collected with a modified Harvard cascade impactor that
used polyurethane foam (PUF) and Teflon substrates (11). Mass determinations were
made based on the difference of measurements made before and after sample collection
using a Mettler MX-5 microbalance. EC
2.5
and EC
0.2
were collected on pre-baked quartz
filters from different sampling lines. EC was determined using a thermal-optical
transmittance (TOT) method, based on NIOSH Method 5040 (38). Duplicate samples,
collected at approximately 10% of locations, showed high precision: intra-class
correlation coefficients for EC
2.5
, EC
0.2
, PM
0.2-2.5
, and PM
0.2
were 0.92, 0.94, 0.99, and
0.88, respectively. Some specific measurements of EC and PM concentrations were
invalidated based on laboratory notes, field notes, and/or criteria for maximum plausible
values when data were not consistent with other proximate measurements and had no
obvious near-source emitters of pollutants (e.g. freeway). Additional information about
92
quality control and other features of sample collection and PM and EC measurements has
been reported previously (12).
Traffic data
Annual average daily traffic (AADT) volumes and truck percentage on roadways
were obtained from the California Department of Transportation (Caltrans) milepost data
for freeways and numbered state highways for 2009 (13) and Dynamap Traffic Count
(Version 10.2) datasets produced by TeleAtlas (Boston, Massachusetts) for other roads.
Traffic volumes on small roads were adjusted based on county data for 1990-2010 to
reflect 2009 traffic levels (39). All volumes were assigned to the Dynamap
Transportation (Version 9.2) roadway network. The extrapolated traffic count data
provided 100% coverage on freeways and highways [Functional Class Code (FCC) 1 and
2 roads], 98% coverage on major arterials (FCC3 roads), and 14% coverage on minor
arterials and collectors (FCC4 roads). A geographic weighted regression, using data from
measured local mean traffic volumes on FCC3 and FCC4 roads was used to estimate
traffic volumes on FCC3 and FCC4 roads without traffic counts. Total truck count on
freeways was computed as the product of total volume on FCC1 roads and its
corresponding truck percentage.
CALINE4 dispersion model estimates of local traffic contributions
Estimates of the contributions of local on-road motor vehicle emissions to air
quality were obtained from the CALINE4 Gaussian line-source dispersion model (15).
The model was applied using PM
2.5
emission factors for exhaust, tire wear, and brake
93
wear from the EMFAC2007 model for 2009 in the Southern California Air Basin (40)
and from literature values for the EC fraction of PM
2.5
emissions (41) and re-suspended
road-dust PM
2.5
and EC emissions (42). Average vehicle speed and heavy-duty vehicle
fractions of traffic volumes on specific segments of freeways and numbered highways
were obtained from Caltrans (13, 39). Characteristic diurnal, day-of-week, and month-of-
year traffic volume variations were obtained from Caltrans (39). Hourly surface wind
speed and direction data collected from meteorological stations in or near each
community during the sampling periods were used along with hourly atmospheric
stability estimates and climatological estimates of morning, afternoon, and evening
mixing heights in the modeling. Separate estimates were made for the near-roadway
contribution of freeways and of all other roadways (non-freeways) to concentrations of
EC and PM
2.5
. Separate estimates could also be made for the contribution of local traffic
to concentrations of several pollutants, including carbon monoxide, nitrogen dioxide,
total oxides of nitrogen, elemental and organic carbon and PM
10
and PM
2.5
. These
estimated pollutant exposures should be regarded as indicators of incremental increases
due to primary emissions from local vehicular traffic on top of background ambient
levels. CALINE-estimated total EC, for example, which we used as a predictor of
measured EC in this study, represented only the effect of the incremental contribution of
local traffic to a more homogeneous community background concentration of EC that
included both primary and secondary pollution resulting from long range transport. This
metric was highly correlated with other pollutants estimated by CALINE4.
94
Distance to sources
Using GIS software, distance to the nearest freeway was computed using the
Dynamap Transportation roadway data. Distance to the nearest active railway was
computed using the Federal Railroad Administration location and activity data for 2008.
Combustion point sources locations and emissions were obtained from the California Air
Resources Board Facilities emissions inventory (43) and sources were categorized based
on NO
x
emission rates of 10 to 50 tons per year (TPY), 50 to 150 TPY, 150 to 350 TPY,
and greater than 350 TPY. Large intermodal transportation facilities (truck to and from
railway) were identified from Research and Innovative Technology Administration's
Bureau of Transportation Statistics database. Distance from sampling locations to the
Pacific shoreline, as defined in ESRI’s geophysical database, was computed.
Normalized Difference Vegetation Index
Normalized difference vegetation index (NDVI) is the ratio of the difference
between and sum of the spectral reflectance measurements acquired in the near-infrared
and visible (red) regions, respectively (varies between -1.0 and +1.0) and is used as an
indicator of live green vegetation. The amount of green vegetation is a surrogate for the
absence of pollution sources and enhanced pollutant uptake by dry deposition. Monthly
average NDVI data for the sampling periods with 3 km resolution were acquired
(http://earthobservatory.nasa.gov/Features/MeasuringVegetation/
measuring_vegetation_2.php) and used to assign NDVI for 300m buffers around each
sampling location and sampling time period.
95
Scaling of Predictors
Model predictors were scaled to two standard deviations of the within-community
distribution across all eight communities. This was done in order to compare the effect
size of different predictors relative to the variability observed in our study. Each predictor
was subtracted (deviated) from its community-specific mean such that each community
had mean value 0 and community-specific variance that was the same as for the
undeviated values. For each exposure variable, the standard deviation of the community-
specific deviations was computed and twice this value was used for scaling the
corresponding effect estimate.
96
Supplemental Table 3.1: Pearson correlation
a
of eight-week averaged levels of measured
pollutants.
EC
2.5
b
EC
0.2
b
PM
2.5
b
PM
0.2
b
EC
2.5
0.83 0.54 0.31
EC
0.2
0.76
0.53 0.49
PM
2.5
0.61 0.57
0.39
PM
0.2
0.22 0.20 0.55
a
Unadjusted correlations (mixture of between-
within community correlation) are found in upper
right triangle, and community-adjusted (within-
community) correlations are found on the lower left.
b
On log scale.
97
Supplemental Table 3.2. Pairwise correlation of EC
2.5
a
with traffic and other land-use predictors (by community)
c
.
Predictors AN GL LB ML RV SB SD UP
CALINE4
a
Freeway 0.83 0.82 0.32 0.19 0.67 0.86 0.41 0.66
Non-freeway -0.64 0.40 -0.45 -0.05 0.27 0.63 0.09 0.69
Total 0.82 0.86 0.16 0.14 0.68 0.90 0.35 0.72
Distance
Freeway -0.62 -0.81 0.04 0.13 -0.66
b
-0.83
b
-0.25 -0.18
Large arterial roads 0.51 -0.24 0.18 -0.26 -0.01 -0.49 0.04 0.08
Traffic density
a
150m buffer 0.70 0.55 0.06 0.11 0.37 0.79 -0.23 0.45
300m buffer 0.70 0.60 0.17 0.06 0.47 0.82 -0.13 0.50
Freeway truck count 0.48 0.13 0.80 0.22 -0.09 -0.07 -0.03 0.08
Road buffers (all roads)
50m 0.36 0.19 -0.07 0.34 0.21 0.43 -0.03 0.07
100m 0.60 0.45 0.00 0.27 0.22 0.51 0.14 0.03
150m 0.62 0.38 -0.02 0.22 0.21 0.61 0.03 0.04
200m 0.60 0.55 0.12 0.19 0.12 0.57 -0.08 0.05
250m 0.58 0.56 0.14 0.15 0.04 0.53 -0.12 0.18
300m 0.54 0.61 0.08 0.11 0.13 0.57 0.00 0.20
Elevation 0.18 -0.44 -0.22 -0.19 -0.41 -0.71 -0.69 -0.63
Population density (300m buffer) -0.02 0.39 -0.16 0.19 0.58 0.77 -0.36 0.44
Normalized difference vegetation index (NDVI) -0.50 -0.45 0.50 -0.16 -0.43 -0.39 0.00 -0.18
a
On log scale.
b
In Riverside (RV) and Santa Barbara (SB), distance to the nearest freeway and to the nearest railway were very highly correlated
(railway was adjacent to the freeway).
c
Some correlations not included (indicated by a dash) because the mean distance to the predictor was greater than 10km in those
communities.
See Table 3.1 for community names.
98
Supplemental Table 3.2 Continued. Pairwise correlation of EC
2.5
a
with traffic and other land-use predictors (by community)
c
.
Predictors AN GL LB ML RV SB SD UP
Distance to railway 0.26 -0.24 0.20 0.20 -0.61
b
-0.83
b
-0.05 -0.66
Distance to intermodal facility -0.23 - -0.29 -0.19 -0.46 - 0.56 -0.57
Distance to point source of NO
x
10-50 tons/yr 0.55 -0.29 -0.47 -0.33 -0.21 0.09 0.35 -0.48
Distance to point source of NO
x
50-150 tons/yr - - 0.21 -0.35 - - - -
Distance to point source of NO
x
150-350 tons/yr - - 0.27 - - - - -
Distance to point source of NO
x
>350 tons/yr - - -0.05 - - - - -
Distance to shoreline - - 0.64 - - -0.435 - -
Number of locations contributing to analysis 18 22 17 13 21 18 16 23
a
On log scale.
b
In Riverside (RV) and Santa Barbara (SB), distance to the nearest freeway and to the nearest railway were very highly correlated
(railway was adjacent to the freeway).
c
Some correlations not included (indicated by a dash) because the mean distance to the predictor was greater than 10km in those
communities.
See Table 3.1 for community names.
99
Supplemental Table 3.3. Pairwise correlation of EC
0.2
a
with traffic and other land-use predictors (by community)
c
.
Predictors AN GL LB ML RV SB SD UP
CALINE4
a
Freeway 0.95 0.60 0.76 0.53 0.64 0.83 0.30 0.63
Non-freeway -0.65 0.29 -0.39 0.70 0.44 0.66 0.21 0.77
Total 0.96 0.65 0.74 0.61 0.72 0.90 0.31 0.74
Distance
Freeway -0.61 -0.54 -0.60 -0.44 -0.65
b
-0.87
b
-0.05 0.03
Large arterial roads 0.59 -0.13 -0.11 -0.42 -0.13 -0.59 -0.01 -0.47
Traffic density
a
150m buffer 0.80 0.42 0.27 0.77 0.52 0.83 0.10 0.61
300m buffer 0.70 0.46 0.39 0.76 0.53 0.77 0.22 0.47
Freeway truck count 0.42 0.39 0.39 0.19 -0.60 -0.12 0.41 0.45
Road buffers (all roads)
50m 0.58 0.02 -0.08 -0.01 0.32 0.44 0.07 0.24
100m 0.72 0.14 0.18 0.60 0.34 0.58 -0.31 0.26
150m 0.71 0.00 -0.20 0.62 0.37 0.66 -0.01 0.36
200m 0.69 0.33 -0.17 0.63 0.36 0.54 -0.03 0.26
250m 0.66 0.42 -0.04 0.66 0.24 0.48 0.10 0.29
300m 0.62 0.44 -0.05 0.59 0.28 0.48 0.23 0.13
Elevation 0.20 -0.53 -0.02 0.27 -0.54 -0.82 -0.45 -0.72
Population density (300m buffer) -0.13 0.38 -0.12 0.07 0.61 0.67 -0.18 0.37
Normalized difference vegetation index (NDVI) -0.46 -0.26 -0.03 -0.25 0.02 -0.36 -0.47 0.04
a
On log scale.
b
In Riverside (RV) and Santa Barbara (SB), distance to the nearest freeway and to the nearest railway were very highly correlated
(railway was adjacent to the freeway).
c
Some correlations not included (indicated by a dash) because the mean distance to the predictor was greater than 10km in those
communities.
See Table 1 for community names.
100
Supplemental Table 3.3 Continued. Pairwise correlation of EC
0.2
a
with traffic and other land-use predictors (by community)
c
.
Predictors AN GL LB ML RV SB SD UP
Distance to point source of NO
x
10-50 tons 0.53 -0.38 -0.32 0.28 -0.12 0.27 0.12 -0.58
Distance to point source of NO
x
50-150 tons - - -0.30 0.51 - - - -
Distance to point source of NO
x
150-350 tons - - -0.33 - - - - -
Distance to point source of NO
x
>350 tons - - -0.61 - - - - -
Distance to shoreline - - 0.33 - - -0.54 - -
Number of locations contributing to analysis 17 22 19 19 20 17 16 22
a
On log scale.
b
In Riverside (RV) and Santa Barbara (SB), distance to the nearest freeway and to the nearest railway were very highly correlated
(railway was adjacent to the freeway).
c
Some correlations not included (indicated by a dash) because the mean distance to the predictor was greater than 10km in those
communities.
See Table 1 for community names.
101
Supplemental Table 3.4. EC prediction models for five communities with high R
2
in the
8-community model.
EC
2.5
a
EC
0.2
a
Communities
b
excluded LB,ML,SD SD
Predictors
c
Logged total CALINE4 0.260 0.303
Population density 0.086
NDVI -0.090
Adjusted R
2
0.68 0.54
LOOCV R
2
0.66 0.53
LOCOCV R
2
0.66 0.50
a
On log scale.
b
See Table 3.1 for community names.
c
Reported betas are scaled to two standard deviations of the
deviated predictors across all eight communities as follows: 1.1
units for logged total CALINE4, 1574 individuals/km
2
for
population, and 0.1 for NDVI.
102
Supplemental Table 3.5. The relative variability of eight-week mean concentrations of
measured pollutants in each community expressed as range/mean (in %).
Community
EC
2.5
EC
0.2
PM
2.5
PM
0.2
Anaheim
101 109 28 54
Glendora
92 84 34 84
Long Beach
32 48 31 49
Mira Loma
35 29 24 44
Riverside
76 93 22 71
Santa Barbara
98 118 45 82
San Dimas
33 43 23 47
Upland
77 64 14 40
103
Supplemental Figure 3.1. Scatter plot of community specific estimates of total CALINE4 from EC
2.5
model as a function of average 8
week NO
x
concentrations from fixed monitoring stations.
See Table 3.1 for community names.
104
Supplemental Figure 3.2. Scatter plot of community specific estimates of total CALINE4 from EC
0.2
model as a function of average
distance to shoreline.
See Table 3.1 for community names.
105
3.8 REFERENCES
1. Guxens M, Sunyer J. A review of epidemiological studies on neuropsychological
effects of air pollution. Swiss Med Wkly 2012;141:w13322.
2. HEI. Special Report: Traffic-related air pollution: a critical review of the literature
on emissions, exposure, and health effects. Boston, MA: Health Effects Institute,
2010.
3. Brauer M, Hoek G, Smit HA, et al. Air pollution and development of asthma,
allergy and infections in a birth cohort. Eur Respir J 2007;29(5):879-88.
4. Morgenstern V, Zutavern A, Cyrys J, et al. Respiratory health and individual
estimated exposure to traffic-related air pollutants in a cohort of young children.
Occup Environ Med 2007;64(1):8-16.
5. Ryan PH, Lemasters GK, Biswas P, et al. A comparison of proximity and land use
regression traffic exposure models and wheezing in infants. Environ Health
Perspect 2007;115(2):278-84.
6. Riedl M, Diaz-Sanchez D. Biology of diesel exhaust effects on respiratory
function. J Allergy Clin Immunol 2005;115(2):221-8; quiz 9.
7. Ghio AJ, Smith CB, Madden MC. Diesel exhaust particles and airway
inflammation. Curr Opin Pulm Med 2012;18(2):144-50.
8. Schauer JJ. Evaluation of elemental carbon as a marker for diesel particulate
matter. J Expo Anal Environ Epidemiol 2003;13(6):443-53.
9. EPA. Report to Congress on Black Carbon. United States Environmental
Protection Agency. EPA-450/R-12-001. 2012,
10. McConnell R, Islam T, Shankardass K, et al. Childhood incident asthma and
traffic-related air pollution at home and school. Environ Health Perspect
2010;118(7):1021-6.
11. Lee SJ, Demokritou P, Koutrakis P, et al. Development and evaluation of personal
respirable particulate sampler (PRPS). Atmos Environ 2006;40(2):212-24.
12. Fruin S, Urman R, Lurmann F, et al. Spatial Variation in Particulate Matter
Components over a Large Urban Area. Atmos Environ 2014;83:211-9.
13. CALTRANS. 2009 Annual Average Daily Traffic (AADT) for all vehicles on
California State Highways. Sacramento, CA: California Department of
Transportation, 2010.
14. Zhu YF, Hinds WC, Kim S, et al. Concentration and size distribution of ultrafine
particles near a major highway. Journal of the Air & Waste Management
Association 2002;52:1032-42.
15. Benson P. CALINE4 - A dispersion model for predicting air pollution
concentrations near roadways (Created November 1984, updated June and
November 1989). 1989.
16. Eckel SP, Berhane K, Salam MT, et al. Residential traffic-related pollution
exposures and exhaled nitric oxide in the children's health study. Environ Health
Perspect 2011;119(10):1472-7.
17. Pettorelli N, Vik JO, Mysterud A, et al. Using the satellite-derived NDVI to
assess ecological responses to environmental change. Trends Ecol Evol
2005;20(9):503-10.
106
18. Gauderman WJ, Avol E, Gilliland F, et al. The effect of air pollution on lung
development from 10 to 18 years of age. N Engl J Med 2004;351(11):1057-67.
19. Franklin M, Vora H, Avol E, et al. Predictors of intra-community variation in air
quality. J Expo Sci Environ Epidemiol 2012;22(2):135-47.
20. Eeftens M, Beelen R, de Hoogh K, et al. Development of land use regression
models for PM2.5, PM2.5 absorbance, PM10 and PMcoarse in 20 European study
areas; results of the ESCAPE project. Environmental Science & Technology 2012.
21. Allen RW, Amram O, Wheeler AJ, et al. The transferability of NO and NO2 land
use regression models between cities and pollutants. Atmos Environ
2011;45(2):369-78.
22. Poplawski K, Gould T, Setton E, et al. Intercity transferability of land use
regression models for estimating ambient concentrations of nitrogen dioxide. J
Expo Sci Environ Epidemiol 2009;19(1):107-17.
23. Vienneau D, de Hoogh K, Beelen R, et al. Comparison of land-use regression
models between Great Britain and the Netherlands. Atmos Environ
2010;44(5):688-96.
24. Carr D, von Ehrenstein O, Weiland S, et al. Modeling annual benzene, toluene,
NO2, and soot concentrations on the basis of road traffic characteristics.
Environmental Research 2002;90(2):111-8.
25. Brauer M, Hoek G, van Vliet P, et al. Estimating long-term average particulate air
pollution concentrations: Application of traffic indicators and geographic
information systems. Epidemiology 2003;14(2):228-39.
26. Hochadel M, Heinrich J, Gehring U, et al. Predicting long-term average
concentrations of traffic-related air pollutants using GIS-based information.
Atmos Environ 2006;40(3):542-53.
27. Beelen R, Hoek G, Fischer P, et al. Estimated long-term outdoor air pollution
concentrations in a cohort study. Atmos Environ 2007;41(7):1343-58.
28. Cyrys J, Heinrich J, Hoek G, et al. Comparison between different traffic-related
particle indicators: elemental carbon (EC), PM2.5 mass, and absorbance. J Expo
Anal Environ Epidemiol 2003;13(2):134-43.
29. Eeftens M, Tsai MY, Ampe C, et al. Spatial variation of PM2.5, PM10, PM2.5
absorbance and PMcoarse concentrations between and within 20 European study
areas and the relationship with NO2 - Results of the ESCAPE project. Atmos
Environ 2012;62:303-17.
30. Putaud JP, Raes F, Van Dingenen R, et al. European aerosol phenomenology-2:
chemical characteristics of particulate matter at kerbside, urban, rural and
background sites in Europe. Atmos Environ 2004;38(16):2579-95.
31. Putaud JP, Van Dingenen R, Alastuey A, et al. A European aerosol
phenomenology-3: Physical and chemical characteristics of particulate matter
from 60 rural, urban, and kerbside sites across Europe. Atmos Environ
2010;44(10):1308-20.
32. Kam W, Liacos JW, Schauer JJ, et al. Size-segregated composition of particulate
matter (PM) in major roadways and surface streets. Atmos Environ 2012;55:90-7.
33. Kinney PL, Aggarwal M, Northridge ME, et al. Airborne concentrations of PM2.5
and diesel exhaust particles on Harlem sidewalks: A community-based pilot
study. Environmental Health Perspectives 2000;108(3):213-8.
107
34. Ryan PH, LeMasters GK, Levin L, et al. A land-use regression model for
estimating microenvironmental diesel exposure given multiple addresses from
birth through childhood. Sci Total Environ 2008;404(1):139-47.
35. Clougherty JE, Wright RJ, Baxter LK, et al. Land use regression modeling of
intra-urban residential variability in multiple traffic-related air pollutants. Environ
Health-Glob 2008;7.
36. Beckerman B, Jerrett M, Brook JR, et al. Correlation of nitrogen dioxide with
other traffic pollutants near a major expressway. Atmos Environ 2008;42(2):275-
90.
37. Moore DK, Jerrett M, Mack WJ, et al. A land use regression model for predicting
ambient fine particulate matter across Los Angeles, CA. J Environ Monitor
2007;9(3):246-52.
38. Eller PM, Cassinelli ME. NIOSH Manual of Analytical Methods. 4th Edition (1st
Supplement) ed.: National Institute for Occupational Safety and Health,
Cincinnati, OH; 1996.
39. CALTRANS. 2007 California Motor Vehicle Stock, Travel and Fuel Forecast.
Sacramento, CA: California Department of Transportation, 2008.
40. ARB. EMFAC2007 v2.3 user's guide: calculating emission inventories for
vehicles in California. Sacramento, CA: California Air Resources Board, 2007.
41. Gillies JA, Gertler AW. Comparison and evaluation of chemically speciated
mobile source PM
2.5
profiles. International Journal of Vehicle Design
2001;27:209-16.
42. Fitz DR, Bufalino C. Measurement of PM
10
emission factors from Paved Roads
using on-board particle sensors. Presented at US Environmental Protection
Agency’s 11th Annual Emission Inventory Conference, Emission Inventories--
Partnering for the Future2002.
43. ARB. 2008 California Facilities Emissions Inventory, California Air Resources
Board. Sacramento, CA: California Air Resources Board, 2009.
108
CHAPTER 4: Associations Between Lung Function and Ambient
Transition Metals Exposure Among Children in the Southern
California Children’s Health Study
4.1 ABSTRACT
Numerous studies have reported adverse effects of particulate matter (PM) on
lung function in children, but there has been little investigation of the chronic effects of
PM composition. Transition metals are biologically plausible agents contributing to these
effects. Aerosol size and water-solubility may be important determinants of metal
toxicity. As part of the Children’s Health Study, repeated measurements of forced
expiratory volume in 1 second (FEV
1
) and forced vital capacity (FVC) were assessed on
1911 children between the ages of 11 and 15 in eight Southern Californian communities.
Concentrations of PM and transition metals (copper [Cu], iron [Fe], nickel [Ni],
vanadium [V], and zinc [Zn]), in three size fractions [quasi-ultrafine (<0.2 μm), fine (0.2
to 2.5 μm) and coarse (2.5 to10 μm)] were measured at participants’ elementary schools.
In addition, the water-soluble fraction of each metal was measured. Adjusted linear
regression models were used to assess the relationship between lung function and these
exposures. Water-soluble Ni, V, and Zn in the fine fraction were associated with a 4.0%
(95% confidence interval: 6.0% to 2.0%), 3.6% (5.9% to 1.3%), and 2.7% (5.0% to 0.4%)
deficit in attained FVC level at age 15 across the 5%-95% distribution of each metal.
Associations of similar magnitude were observed for total Ni and V in the fine fraction
and with water-soluble V in the coarse fraction. Ni and V associations in each size
fraction were generally robust to adjustment for total mass. Associations of FEV
1
were
observed with fine PM mass but not with any of the metals. These results suggest that
109
specific water-soluble transition metals found in PM might explain some of the adverse
effects of PM on childhood lung function.
4.2 INTRODUCTION
Lung function is commonly used to assess respiratory health. In the Children’s
Health Study (CHS), we have demonstrated that high ambient concentrations of regional
particulate matter (PM) with an aerodynamic diameter of less than 10 μm (PM
10
) and less
than 2.5 μm (PM
2.5
) were associated with lower lung function and lung function growth
(1, 2). Additionally, high exposure to local traffic, independent of the effects of regional
air pollution, was associated with impaired lung function (3). It is possible that early life
deficits in lung function can have long lasting effects if stunted lung development results
in lower attained lung function in adult life (4). This is important because low lung
function among adults has been linked with cardiovascular disease and mortality (5-7).
PM is a complex mixture of different organic and inorganic compounds (8).
Identifying which of these components of PM induces harmful effects is important from a
biological, public health, and regulatory standpoint. Transition metals are commonly
found in PM and have been implicated as a cause of health effects (9). Transition metals
cause oxidative stress both in vivo and in vitro (10) by production of reactive oxygen
species (ROS) through Fenton-like mechanisms due to their redox cycling properties
(11). Other metals contribute to oxidative stress through other pathways (12, 13). The
solubility of metals is likely a key contributing factor to biological effects as soluble
metals can more easily translocate into cells beyond the site of deposition and are thus
more bioavailable (14).
110
Diesel exhaust, brake and tire wear, ship emissions, and industrial processes are
all important contributors to metal-enriched PM (15). Residential proximity to major
roadways, where metal-enriched PM from vehicular emissions and resuspended road dust
is likely to be high (16), has been associated with reduced lung function and increased
incidence of asthma (17, 18). Some studies have shown acute cardiovascular and
respiratory effects with specific metals found in PM (19, 20), but there has been relatively
little epidemiological investigation of the chronic effects of metals on lung function in
children.
In this study, we investigated associations of selected transition metals (copper
[Cu], iron [Fe], nickel [Ni], vanadium [V], and zinc [Zn]) in coarse (PM
10-2.5
), fine
(PM
2.5-0.2
), and quasi-ultrafine (PM
0.2
) size fractions with lung function growth in
adolescence and attained lung function at age 15 in the CHS. These metals were selected
based on previous human and animal studies which suggested that they were harmful to
the lungs (14). We hypothesized that the adverse effects of these transition metals would
be largest for the smallest size fractions due to its deeper deposition into the lungs and for
water soluble metals.
4.3 MATERIALS AND METHODS
Study design
The CHS includes a series of cohorts in which the relationship between air
pollutants and indicators of cardio-respiratory health has been investigated among
children. The current cohort has been described previously (3, 18). Briefly, children from
kindergarten and first grade classes were recruited in 2002-2003 from public elementary
111
schools across several southern Californian communities selected to represent a large
range of regional pollutant exposures. Questionnaires completed at study entry by parents
or guardians collected information on respiratory health, demographic, social, and indoor
environmental exposures relating to the child. Information collected included sex, race,
Hispanic ethnicity, parental education, insurance coverage, asthma status, exposure to
maternal smoking during pregnancy (in utero) and to second hand smoking (SHS),
presence of pests, dog, cat, carpet, water damage, or mildew in the home, and perceived
stress of the parent or guardian filling out the baseline questionnaire (21). Follow-up
questionnaires were sent out annually (except in years 7 and 9 of follow-up, due to
budgetary constraints) to update specific information pertaining to the child, including
asthma status.
Lung function
Beginning in year 6 of follow-up when children were in either the 5
th
or 6
th
grade
(year 6 of follow-up, average age: 11.2), trained technicians measured maximal forced
expiratory volume in the first second (FEV
1
) and forced vital capacity (FVC) on each
child using pressure transducer-based spirometers (Screenstar Spirometers, Morgan
Scientific, Haverhill, Massachusetts, USA). Lung function testing was conducted again in
years 8 and 10 of follow-up. A total of 1911 children had at least one set of lung function
measurements. All acceptable lung function maneuvers had to meet American Thoracic
Society standards, as previously described (1). On the day of lung function testing,
height, weight, and whether the child currently had a cold or other chest illness were
assessed.
112
Exposures
During year 7 of the study, size-resolved PM samples were collected at CHS
elementary schools using modified Harvard cascade impactors (22) with multiple
collection stages to capture quasi-ultrafine (PM
0.2
), accumulation mode fine (PM
2.5-0.2
),
and coarse (PM
10-2.5
) fractions. Measurements at the schools were made throughout the
year on a rotating schedule in two communities at a time. All schools within each
community were measured at the same time. Sampling at the schools was performed for
two consecutive two-week periods during warm and cool times of the year for a total of
eight weeks. We measured at elementary schools as an indicator of exposures in
neighborhoods around schools where children live. Exposures were assigned to
measurements of lung function during subsequent follow-up. In addition to
measurements at the school, measurements at central site monitors in each community
were made concurrently with the schools. These sites operated for an additional eight
weeks throughout the year for a total sampling period of 16 weeks in order to obtain a
more stable estimate of the yearly average concentration. Additional information on the
exposure assessment has been described elsewhere (23).
Using federal reference gravimetric methods, total mass was determined by
weighing polyurethane foam (for PM
0.2-2.5
and PM
2.5-10
) and Teflon substrates (for PM
0.2
)
using a Mettler MT-5 microbalance in a temperature and relative humidity controlled
room. The total and water-soluble mass of Cu, Fe, Ni, V, and Zn were measured with
inductively-coupled plasma mass-spectroscopy (ICP-MS) at the Wisconsin State
Laboratory of Hygiene. Ambient concentrations were calculated by taking the mass and
113
dividing it by the total volume of air that flowed through the Harvard cascade impactors
(at flow rate of 5L per second).
Statistical methods
Pollutant concentrations at schools were averaged across all waves of
measurements to yield single 8-week estimates. To get better estimates of the long term
average at the schools, these 8-week averages were scaled up or down by multiplying
these values by the ratio of the 16-week average across all measurements at the central
site monitors to the 8 weeks of central site measurements that corresponded to the time at
which measurements were collected at the schools. The water insoluble fraction of each
metal was calculated by taking the difference of the water soluble fraction of a particular
metal from the total concentration. Each pollutant was scaled to the middle 90th
percentile of its own distribution (from the 5
th
to 95
th
percentile) in order to compare
health effect estimates across an environmentally relevant range for each pollutant.
A mixed-effect linear spline model, with knots placed at ages 12 and 14, was used
to model the non-linear trajectory of lung function growth in children. Included in the
spline model were adjustments for sex, race, Hispanic ethnicity, height, height-squared,
body mass index (BMI), BMI-squared, presence of respiratory illness on the day of the
testing, and field technician. Each lung function outcome was log transformed to satisfy
the assumptions of the models. The linear growth model was parameterized to yield
estimates, reported as percent changes, of the effect of a pollutant on 4-year lung function
growth between age 11 and 15 as well as an effect of a pollutant on mean attained lung
function level at age 15, as described previously (24). In addition, the linear spline model
114
included random effects for both level and growth to account for dependency among
participants and clustering of children at both the community and school levels. To test
the robustness of the associations in these models, they were additionally adjusted for
potentially confounding social and home environment exposures. We also investigated
whether the effects of these pollutants differed by gender and asthma status (defined as
ever medically diagnosed during first year of lung function testing). We tested whether
any differences were significant by including an interaction term between the pollutant of
interest and an indicator for gender or asthma status. Lastly, we tried fitting 2-pollutant
models whereby we adjusted the metal concentrations by total PM mass within the same
size fraction in order to assess whether the effect of each metal is independent of the
effect of total PM mass.
4.4 RESULTS
There were a total of 926 males and 975 females across 30 schools with an
average of 2.4 lung function tests performed on each child. A majority of participants
were white and of Hispanic descent (Table 4.1). Approximately twenty percent of the
children indicated that they had a doctor confirmed diagnosis of asthma at the first lung
function testing.
Between the ages of 11 and 15, girls’ average FEV
1
increased by 908 ml (average
attained level at age 15: 3263 ml), while boys saw their average FEV
1
increase by 1568
ml (average attained level at age 15: 3968 ml). Over the same span of age, the average
FVC for girls increased by 1036 ml (average attained level at age 15: 3706 ml) and 1868
ml for boys (average attained level at age 15: 4657 ml).
115
Community specific distributions of total metal concentration and the water
soluble and insoluble fractions of each metal are shown in Table 4.2 and in Supplemental
Figures 4.1, 4.2, and 4.3, respectively. Of the five metals studied, Fe had the highest
average total concentration in each size fraction, while Ni and V had amongst the lowest
concentrations for each fraction. The between-community variability was substantially
larger than the within-community variability for each metal across all size fractions with
the exception of Zn whose between-community variability was between 1.6 and 2.9 times
higher than the within community variability (Table 4.2). Except for V, the correlation of
the coarse water soluble metals were weakly, negatively correlated with total coarse PM
mass. Correlations between metals and total mass within all other fractions ranged from
moderately strong (R=0.28 for quasi-ultrafine Cu) to very strong (R= 0.94 for coarse V).
Across all size fractions, Fe had the lowest water soluble to insoluble ratio compared to
other metals. In the fine and quasi-ultrafine fractions, the majority of the mass for all
metals, except Fe, was in the water-soluble fraction.
The total metal concentration and the water soluble fraction were mostly
negatively associated with attained FEV
1
at age 15 (Table 4.3). There was a marginally,
significant effect of V in the total fine fraction (P=0.08) and water soluble coarse fraction
(P=0.08) and of Ni in the water soluble fine fraction (P=0.05) on attained level at age 15.
There was little evidence of an effect of the water insoluble fraction of each metal on
attained FEV
1
and FEV
1
growth. A 9.9 μg/m
3
increase in total fine PM mass was
associated with approximately a 3% deficit in FEV
1
at age 15 (95% CI: -5.31, -0.61).
Although the associations with lung FEV
1
growth between the ages of 11 and 15 (slope)
were generally negative, none was statistically significant.
116
V in the total coarse fraction was associated with an approximately 3% deficit in
FVC at age 15 (-2.95%, 95% CI: -5.49, -0.34) and with a 4% deficit (-3.86%, 95% CI: -
5.81, -1.93) in the total fine fraction (Table 4.4). Negative associations with FVC attained
level were also observed with Ni in the total and water soluble fine fractions, V in the
water-soluble coarse and fine fractions, Zn in the water-soluble fine fraction, and with
total fine PM mass. In the coarse and fine fractions, estimated water-soluble effects were
larger than insoluble effects for all metals except Fe. In the ultrafine fraction, water-
insoluble Zn was associated with deficits in attained FVC at age 15. There were no
associations with FVC growth.
Adjustment for potential confounders did not result in any marked changes in the
relationship between pollutant and outcome. Focusing on the association between water
soluble fine Ni on FEV
1
and FVC, for example, adjusting for various social and home
environment variables resulted in effect estimates on attained FEV
1
level that ranged
from -2.40% to -2.74% (compared to -2.61% in the base model) and estimates on attained
FVC level that ranged from -3.61% to -4.14% (compared to -4.01% in the base model;
Table 4.5). Stratifying by gender revealed larger effects of water soluble fine Ni on
attained FEV
1
and FVC level at age 15 among boys than among girls. The effect on FVC
level was larger among asthmatics than non-asthmatics; however, differences in effect
estimates between boys and girls and between children with and without asthma were not
statistically significant. We also restricted the analysis to those participants who lived in
the same residence between the first and last set of lung function measurements to reduce
possible exposure misclassification (as exposure measurements were made for only the
initial address and a single year). Among this group of non-movers (N=1181), the
117
associations both with growth and in attained lung function at age 15 were noticeably
stronger compared to the entire population. Other statistically significant associations
(water soluble fine V and Zn with FVC, for example) were also robust to confounding
and showed larger (but not statistically different) deficits in boys and asthmatics (results
not shown).
Because the metal associations may have reflected associations between
correlated total mass and lung function, we fit two pollutant models that co-adjusted each
metal and total mass. After adjustments for total PM mass in the coarse fraction, total
coarse V was associated with FEV
1
(Table 4.6), but the inflated estimate was a result of
collinearity between total coarse V and PM mass (correlation: 0.94 in Table 4.2). In FVC
models adjusted for total mass, associations with water-soluble V in the coarse size
fraction, total V and water-soluble Ni and V in the fine size fraction and water-insoluble
Zn in the quasi-ultrafine size fraction all remained significant for attained level by 15
years of age (Table 4.7). In these two-pollutant models, additional associations with FVC
attained level were observed with water-soluble coarse Fe and Ni and with water-
insoluble Fe in the quasi-ultrafine size fraction.
4.5 DISCUSSION
Deficits in attained FVC at age 15 were associated with transition metals. The
most consistent associations were with water soluble Ni and V in the fine particulate
fraction, although associations were also observed with water soluble Fe, Ni and V in the
coarse size fraction after adjustment for coarse PM mass. Water insoluble Zn in the
quasi-ultrafine fraction was consistently associated with attained FVC. These results
118
suggest that effects of these metals on FVC may be independent of total mass. A deficit
in attained FEV
1
was associated with fine PM mass but there were no statistically
significant associations of FEV
1
with transition metals.
There were no significant associations of PM mass or composition with lung
function growth (with the exception of fine water-soluble Ni with FVC growth among the
subset of participants who had lived at the same residence during the period of study;
Table 4.5), but many of the observed associations were negative. One possible
explanation for an association with attained level and not growth is that the effect of PM
mass and its composition on lung development was cumulative and started before our
first measurements of lung function. If we had been able to measure lung function
starting at an earlier stage in the children’s development, we might have seen a
significant effect of PM mass and its composition on lung function growth.
PM is a heterogeneous mixture of different constituents in various sizes (8, 14).
Some transitional metals (e.g., Fe, Cu, Ni, and V) are capable of undergoing redox
cycling and can generate ROS through Fenton-like mechanisms, while other transitional
metals not undergoing redox cycling (such as Zn) have also shown to induce oxidative
stress in model systems (11-13, 25, 26). When ROS overwhelm antioxidant defenses in
cells, ROS can react with proteins, lipids, and DNA causing tissue damage and promote
the influx of inflammatory cells that are plausibly responsible for lung function deficits
observed in this study (10, 27, 28). Experimental studies provide support to the perceived
harmful effects of the transition metals. For example, exposing rats to short-term
exposure of stainless steel welding fumes, comprised of Fe, Cr, Mn, and Ni, caused lung
119
damage and compromised the rat’s ability to fight bacterial infection compared to a
control group exposed to filtered air (29).
Based on prior knowledge of transition metals and the hypothesized role of
oxidative stress in air pollution-induced lung disease, we would have expected Fe to have
large effects, because the concentrations of total Fe were much larger than those of other
metals in the study communities. However, the effect of total Fe scaled across an
environmentally relevant range was generally weaker than that of both total Ni and total
V. Why Fe might be less toxic is not clear, although one study testing the toxicity of
several metals on a rat lung epithelial cell line showed the order of greatest toxicity to
smallest to be V, Zn, Cu, Ni, and Fe (30). Also of possible relevance to the weaker
associations of Fe is that in all size fractions a much smaller proportion was water-soluble
(albeit still generally more in absolute concentration) than for other metals (Table 4.2).
Water solubility is an indicator of bioavailability and has been implicated in
respiratory toxicity (14). Studies have shown that exposing mice and rats to soluble
transition metals resulted in inflammation in the lungs (31-33). Exposure of human
alveolar epithelial cells to the soluble fraction of three different metal-rich welding fumes
resulted in reduced intracellular glutathione, while exposure to two nickel based fumes
also resulted in an increase in ROS production (34). Exposing human bronchial epithelial
cells to residual oil fly ash (ROFA), a byproduct of heavy oil burning containing water
soluble V, Ni, and Fe, also induced cytokine expression Carter, Ghio, Samet and Devlin
(11). The insoluble fraction of each metal had relatively weaker and less consistent
associations with lung function. However, the water-insoluble fraction was not measured
120
directly, rather calculated by taking the difference of the water-soluble fraction from the
total concentration.
To our knowledge, this is the first study that has examined specific metals across
multiple size fractions with lung function. We hypothesized that metals in the quasi-
ultrafine fraction would have the most impact on children’s lungs because of their ability
to penetrate deeper into the lungs and therefore should be more toxic on a mass basis (35,
36). We did not find consistent associations of quasi-ultrafine metals with lung function.
A couple of possible reason for the lack of significant associations in this size fraction
could be due to missing data (some 2-week samples in this size fraction were lost due to
errors with setup) which would attenuate any true associations or that the smaller amount
of mass in this size fraction, most of which is in accumulation mode, is not enough to
elicit a health response. Our results are consistent with a recent review of effects of
ambient ultrafine PM, which concluded that there are not strong and consistent findings
that would support the notion that ultrafine particles have effects on human health that are
independent of other PM size fractions (37). Although water-soluble metals had
generally stronger associations with lung function, we unexpectedly found associations
on attained FVC with both water-insoluble Fe and Zn in the quasi-ultrafine fraction, but
only after adjustment for mass.
Most epidemiological studies involving the effects of metals exposure on lung
function have focused on occupational exposures in adults (38). There have been few
studies of associations of inhaled metal particles with lung function in children and
results have not been uniformly consistent. In a Chinese study, children living near an e-
waste recycling area had higher blood concentration levels of Ni and Mn compared to a
121
control group of children (39). Boys aged 8-9 years who had higher exposure to these
metals had significantly lower FVC compared to an unexposed group of boys, but no
significant associations with these metals were observed among girls or among boys in
older age groups, perhaps due to limited sample size. In a multi-city European study of
the effects of PM and selected elemental exposures, chosen to reflect different
anthropologic sources, Ni in the PM
10
size fraction were associated with lower FEV
1
in
children, but the authors concluded that total mass was more consistently associated with
lung function deficits (40). Another recent study found that increased exposure to PM
2.5
Cu and Fe were associated with a deficit in FEV
1
(41), results which our study did not
replicate. In a study of the acute effects of metals among young adults, Cd, Zn, and V
were associated with reduced lung function (20).
Studies have also identified effects of Ni, V and Zn, in addition to Fe, on other
respiratory outcomes in children. Ambient Fe, Ni and V were associated with airway
inflammation among 9 to 11 year old children (42). Exposure to ambient levels of PM
2.5
Zn was linked with following day hospitalizations for pediatric asthma exacerbations
(43). Among children less than 2 years of age, higher ambient levels of Ni and V, and not
total PM
2.5
, were associated with an increased probability of wheeze (44).
Strengths of this study include the longitudinal collection of lung function
measurements on a large number of children with detailed metal measurements in three
different size fractions at schools in communities with a wide range of ambient particle
concentration. A limitation is the use of a single estimate of exposure in 2009 to represent
the average neighborhood exposure for the entire study period. PM
10
and PM
2.5
have
steadily decreased over the past decade (45), but the relative ranking of communities in
122
terms of air pollutant concentrations have changed very little. Given the large between
community contrasts in metal exposures (Table 4.2), it is likely that the order of
community averaged metal concentrations across the duration of the study is reflected by
the concentrations measured in 2009. Any potential misclassification of exposure would
likely be non-differential which would result in the reported estimates being
underestimates of the true effect.
Care must also be taken in interpreting these findings as metals are source specific
and other unmeasured compounds that may be spatially correlated with the metals
presented in this study could be responsible for the reported harmful effects. In Southern
California, the likely dominant source of Ni and V is from heavy oil combustion from
ships other combustion sources at the ports of Los Angeles and Long Beach. However,
port-related emissions contribute significant amounts of organic compounds in both the
particle and gaseous phase that could be responsible for the observed effects (46, 47).
Consistent adverse effects of Southern California PM on lung function in children
have been observed across multiple CHS cohorts; moreover, improvements in air quality
have been associated with better lung function in children in the CHS (24). This study
identifies water-soluble transition metals in the fine respirable fraction as components of
PM potentially responsible for these effects. These results have potential relevance for
standards setting based on PM composition.
123
4.6 TABLES AND FIGURES
Table 4.1. Description of study population.
Population characteristics N (%)
a
Male 926 (48.7%)
Race
Asian 87 ( 5.3%)
Black 44 ( 2.7%)
Mixed 241 (14.6%)
Other 501 (30.4%)
White 774 (47.0%)
Hispanic ethnicity
Hispanic 1080 (59.7%)
Non-Hispanic 729 (40.3%)
Education
Less than 12th 366 (20.7%)
High school diploma or some college 899 (50.8%)
College diploma or greater 506 (28.6%)
Insurance coverage 1585 (89.3%)
Exposure to in-utero smoke 104 ( 5.8%)
Environmental tobacco smoke 82 ( 4.4%)
Exposure to pests 1155 (67.8%)
Dog 539 (30.0%)
Cat 307 (17.1%)
Carpet 1608 (89.8%)
Mildew 430 (25.7%)
Water damage 237 (13.4%)
Parental stress
Low 810 (46.7%)
High 924 (53.3%)
Asthma diagnosis 357 (19.7%)
Non-movers (between years 6 and 10) 1181 (62.1%)
a
Due to missing values, denominators (n) for each
percentage may differ.
124
Table 4.2. Descriptive statistics of measured metals (in ng/m
3
) and PM mass (in μg/m
3
).
a
Correlation calculated within same size fraction
Mean (std/5-95%)
Between to
within
community
ratio
Correlation
with PM
mass
a
Mean (std/5-95%)
Between to
within
community
ratio
Correlation
with PM
mass
a
Mean (std/5-95%)
Between to
within
community
ratio
Correlation
with PM
mass
a
Water
soluble to
insoluble
ratio
Coarse
Cu 9.8( 3.09/ 9.9) 4.1 0.38 2.4( 1.28/ 4.2) 17.1 -0.16 7.4( 2.54/ 8.7) 2.8 0.54 0.33
Fe 411.8(135.5/524.1) 12.9 0.91 2.8( 1.48/ 4.3) 20.0 -0.31 409.0(136.0/525.1) 13.0 0.91 0.01
Ni 0.6( 0.16/ 0.6) 12.1 0.73 0.1( 0.05/ 0.2) 12.4 -0.01 0.5( 0.14/ 0.5) 9.7 0.83 0.22
V 1.0( 0.33/ 1.2) 23.5 0.94 0.1( 0.06/ 0.1) 11.0 0.64 0.9( 0.28/ 1.1) 21.6 0.95 0.14
Zn 8.4( 2.85/ 11.3) 2.7 0.55 3.8( 2.27/ 6.1) 3.0 -0.23 4.6( 2.42/ 8.0) 3.1 0.87 0.84
PM Mass 11.8( 2.88/ 11.1) 12.3 . N/A N/A N/A N/A
Fine
Cu 3.5( 1.29/ 4.6) 3.8 0.49 2.0( 0.77/ 2.6) 5.3 0.43 1.5( 0.56/ 1.9) 1.9 0.54 1.39
Fe 85.1(26.45/ 88.6) 3.5 0.74 8.2( 1.96/ 6.8) 3.1 0.59 76.9(25.13/ 82.1) 3.4 0.73 0.11
Ni 0.4( 0.12/ 0.4) 8.6 0.61 0.2( 0.08/ 0.3) 13.8 0.60 0.2( 0.06/ 0.2) 2.0 0.42 1.28
V 0.9( 0.31/ 0.9) 23.2 0.67 0.7( 0.26/ 0.7) 11.2 0.46 0.2( 0.11/ 0.4) 6.2 0.77 3.00
Zn 4.1( 1.68/ 5.4) 1.6 0.59 3.7( 1.98/ 6.9) 3.0 0.77 0.3( 0.89/ 3.1) 16.4 -0.59 12.00
PM Mass 12.2( 2.74/ 9.9) 14.2 . N/A N/A N/A N/A
Ultrafine
Cu 0.3( 0.15/ 0.5) 4.1 0.28 0.2( 0.08/ 0.2) 3.5 0.59 0.1( 0.12/ 0.4) 1.6 -0.03 1.09
Fe 4.7( 2.26/ 7.7) 6.5 0.66 0.8( 0.35/ 1.0) 4.0 0.66 3.9( 2.04/ 6.6) 5.5 0.62 0.19
Ni 0.1( 0.10/ 0.4) 8.3 0.41 0.1( 0.05/ 0.2) 4.1 0.36 0.0( 0.07/ 0.3) 2.7 0.36 1.97
V 0.5( 0.29/ 0.9) 13.5 0.28 0.3( 0.15/ 0.5) 7.9 0.49 0.2( 0.17/ 0.5) 3.6 0.05 1.68
Zn 1.1( 0.67/ 2.2) 2.9 0.67 0.9( 0.44/ 1.4) 1.7 0.62 0.2( 0.39/ 1.2) 2.1 0.44 3.79
PM Mass 1.8( 0.58/ 1.7) 4.0 . N/A N/A N/A N/A
Total Fraction Water Soluble Fraction Water Insoluble Fraction
125
Table 4.3. Associations of metals and PM with FEV
1
a
.
a
Associations of attained lung function at age 15 and growth from 11 to 15 years of age were reported as percent changes along with
95% confidence intervals per unit increase in pollutant which was scaled to the difference between the 5
th
and 95
th
percentiles (see
Table 4.2 for pollutant specific values). All models were adjusted for height, BMI, race, Hispanic ethnicity, sex, respiratory illness,
and field technician.
* P<0.05
Attained age 15 Lung growth Attained age 15 Lung growth Attained age 15 Lung growth
Coarse
Cu 0.15( -2.76, 3.14) -0.47( -2.28, 1.38) -0.65( -3.65, 2.45) -0.00( -1.89, 1.92) 0.64( -2.48, 3.86) -0.56( -2.49, 1.40)
Fe -1.94( -4.86, 1.06) -0.75( -2.86, 1.40) -0.22( -2.89, 2.52) -0.29( -1.94, 1.39) -1.92( -4.84, 1.09) -0.74( -2.85, 1.41)
Ni -1.51( -4.53, 1.61) -1.31( -3.37, 0.80) -1.81( -4.62, 1.08) -0.78( -2.68, 1.16) -0.78( -3.72, 2.25) -1.03( -2.93, 0.91)
V -2.31( -5.05, 0.51) -0.74( -2.78, 1.35) -2.11( -4.38, 0.22) -0.40( -2.13, 1.37) -2.12( -4.89, 0.74) -0.74( -2.77, 1.33)
Zn -0.72( -3.92, 2.59) -1.13( -3.33, 1.12) -0.68( -2.94, 1.62) -0.35( -1.91, 1.23) 0.47( -2.45, 3.47) -0.64( -2.47, 1.22)
PM mass -1.09( -4.17, 2.10) -0.10( -2.23, 2.07)
Fine
Cu 0.97( -2.41, 4.47) -0.40( -2.48, 1.74) 0.46( -2.71, 3.73) -0.74( -2.68, 1.24) 1.28( -1.68, 4.34) 0.13( -1.82, 2.13)
Fe -0.46( -3.24, 2.39) -0.68( -2.51, 1.18) -2.12( -4.75, 0.59) -1.21( -3.17, 0.78) -0.34( -3.04, 2.45) -0.61( -2.39, 1.20)
Ni -2.12( -5.14, 0.99) -0.89( -3.07, 1.33) -2.61( -5.16, 0.00) -1.05( -2.98, 0.91) -0.16( -2.72, 2.48) -0.14( -1.90, 1.66)
V -2.37( -4.92, 0.26) -0.81( -2.74, 1.16) -2.29( -5.01, 0.50) -0.75( -2.71, 1.25) -1.17( -3.69, 1.42) -0.50( -2.24, 1.28)
Zn -0.97( -3.44, 1.55) -0.55( -2.34, 1.27) -1.63( -4.25, 1.05) -0.55( -2.48, 1.41) 1.36( -1.54, 4.34) 0.09( -1.81, 2.03)
PM mass -2.99( -5.31, -0.61) * -0.52( -2.49, 1.50)
Ultrafine
Cu 0.32( -2.42, 3.15) -0.03( -1.81, 1.77) 0.59( -2.25, 3.52) -0.16( -1.99, 1.70) 0.00( -2.40, 2.47) 0.03( -1.63, 1.72)
Fe -1.11( -4.05, 1.92) -0.01( -1.95, 1.96) -0.22( -2.85, 2.49) -0.55( -2.24, 1.16) -1.13( -3.89, 1.71) 0.10( -1.77, 1.99)
Ni -1.08( -4.68, 2.66) -1.71( -4.16, 0.80) -0.30( -3.48, 2.99) -1.02( -3.24, 1.24) -1.32( -4.69, 2.17) -1.74( -4.11, 0.68)
V -1.38( -4.74, 2.10) -1.40( -3.71, 0.97) -1.06( -4.44, 2.45) -1.21( -3.51, 1.13) -1.24( -4.28, 1.90) -1.17( -3.38, 1.09)
Zn -0.73( -3.51, 2.12) -0.43( -2.34, 1.53) -0.80( -3.43, 1.90) -0.51( -2.33, 1.35) -0.28( -2.90, 2.42) -0.11( -2.01, 1.83)
PM mass 0.76( -1.93, 3.53) 0.09( -1.61, 1.82)
Total Fraction Water-Soluble Fraction Water-Insoluble Fraction
126
Table 4.4. Associations of metals and PM with FVC
a
.
a
Associations of attained lung function at age 15 and growth from 11 to 15 years of age were reported as percent changes along with
95% confidence intervals per unit increase in pollutant which was scaled to the difference between the 5
th
and 95
th
percentiles (see
Table 4.2 for pollutant specific values). All models were adjusted for height, BMI, race, Hispanic ethnicity, sex, respiratory illness,
and field technician.
* P<0.05
Attained age 15 Lung growth Attained age 15 Lung growth Attained age 15 Lung growth
Coarse
Cu -0.17( -3.01, 2.75) 0.26( -1.49, 2.05) -0.96( -4.03, 2.22) -0.01( -1.83, 1.83) 0.28( -2.72, 3.38) 0.31( -1.53, 2.18)
Fe -1.91( -4.92, 1.19) -0.37( -2.39, 1.68) -0.40( -3.14, 2.42) -0.14( -1.72, 1.47) -1.88( -4.89, 1.24) -0.36( -2.38, 1.69)
Ni -1.85( -4.75, 1.13) -0.35( -2.33, 1.66) -1.59( -4.54, 1.45) -0.44( -2.26, 1.42) -0.97( -3.85, 1.99) -0.07( -1.91, 1.80)
V -2.95( -5.49, -0.34) * -0.76( -2.65, 1.17) -3.49( -5.21, -1.73) * -0.95( -2.46, 0.59) -2.50( -5.19, 0.25) -0.62( -2.51, 1.32)
Zn -1.37( -4.34, 1.70) -0.42( -2.45, 1.66) -0.66( -2.80, 1.54) -0.30( -1.70, 1.11) -0.34( -3.09, 2.48) 0.01( -1.72, 1.78)
PM mass -2.48( -5.15, 0.26) -0.60( -2.54, 1.38)
Fine
Cu -0.11( -3.25, 3.14) 0.34( -1.64, 2.36) -0.16( -3.23, 3.01) 0.28( -1.60, 2.19) -0.03( -2.69, 2.70) 0.33( -1.45, 2.14)
Fe -1.47( -4.04, 1.17) -0.38( -2.06, 1.33) -0.96( -3.99, 2.16) -0.01( -1.92, 1.95) -1.41( -3.89, 1.14) -0.37( -2.01, 1.28)
Ni -3.80( -6.34, -1.19) * -1.74( -3.64, 0.19) -4.01( -5.96, -2.02) * -1.55( -3.19, 0.13) -0.98( -3.33, 1.43) -0.86( -2.42, 0.72)
V -3.89( -5.81, -1.93) * -1.50( -3.11, 0.14) -3.63( -5.89, -1.32) * -1.38( -3.08, 0.35) -1.78( -4.18, 0.68) -0.73( -2.30, 0.86)
Zn -1.53( -3.76, 0.75) -0.53( -2.12, 1.08) -2.71( -4.96, -0.40) * -0.76( -2.51, 1.02) 1.95( -0.82, 4.79) 0.35( -1.43, 2.17)
PM mass -3.56( -5.64, -1.43) * -0.90( -2.64, 0.88)
Ultrafine
Cu -1.13( -3.65, 1.45) 0.02( -1.61, 1.68) 0.83( -2.03, 3.77) 0.70( -1.00, 2.43) -1.57( -3.64, 0.54) -0.31( -1.77, 1.16)
Fe -2.06( -4.98, 0.95) -0.17( -2.07, 1.76) 0.05( -2.58, 2.75) 0.18( -1.41, 1.79) -2.11( -4.80, 0.66) -0.18( -1.98, 1.65)
Ni -2.31( -5.93, 1.45) -0.44( -2.82, 2.00) -1.32( -4.46, 1.91) -0.34( -2.40, 1.76) -2.02( -5.24, 1.31) -0.26( -2.50, 2.03)
V -2.81( -6.30, 0.82) -0.85( -3.12, 1.48) -1.73( -5.14, 1.80) -0.05( -2.22, 2.17) -2.50( -5.40, 0.49) -1.17( -3.24, 0.94)
Zn -2.16( -4.73, 0.47) -0.35( -2.15, 1.49) -0.77( -3.28, 1.81) -0.08( -1.75, 1.62) -2.41( -4.66, -0.11) * -0.29( -2.02, 1.47)
PM mass 0.15( -2.44, 2.80) 0.62( -0.96, 2.23)
Total Fraction Water-Soluble Fraction Water-Insoluble Fraction
127
Table 4.5. Sensitivity analysis for water-soluble fine Ni
a
.
a
Associations of attained lung function at age 15 and growth from 11 to 15 years of age were reported as percent changes along with
95% confidence intervals per unit increase in pollutant which was scaled to the difference between the 5
th
and 95
th
percentiles (see
Table 4.2 for pollutant specific values). All models were adjusted for height, BMI, race, Hispanic ethnicity, sex, respiratory illness,
and field technician.
* P<0.05
Model Attained age 15 Lung growth Attained age 15 Lung growth
Base model (from Tables 3 and 4) -2.61( -5.16, 0.00) -1.05( -2.98, 0.91) -4.01( -5.96, -2.02) * -1.55( -3.19, 0.13)
Base model + parental education -2.40( -5.08, 0.37) -1.04( -2.99, 0.96) -3.61( -5.65, -1.53) * -1.48( -3.17, 0.24)
Base model + insurance coverage -2.74( -5.32, -0.09) * -1.11( -3.05, 0.86) -4.14( -6.11, -2.12) * -1.62( -3.28, 0.07)
Base model + in utero exposure to tobacco smoke -2.64( -5.19, -0.02) * -1.07( -3.01, 0.90) -3.99( -5.98, -1.96) * -1.59( -3.24, 0.10)
Base model + ETS (time varying) -2.68( -5.22, -0.07) * -1.00( -2.93, 0.96) -4.06( -6.01, -2.06) * -1.53( -3.18, 0.14)
Base model + pests -2.66( -5.21, -0.03) * -1.12( -3.06, 0.86) -3.99( -5.98, -1.96) * -1.67( -3.31, 0.00)
Base model + dog -2.53( -5.18, 0.20) -1.07( -3.01, 0.90) -3.93( -5.93, -1.89) * -1.61( -3.26, 0.07)
Base model + cat -2.65( -5.21, -0.02) * -1.13( -3.06, 0.85) -4.02( -5.98, -2.01) * -1.66( -3.32, 0.02)
Base model + carpet -2.70( -5.20, -0.13) * -1.18( -3.12, 0.81) -3.99( -5.95, -1.99) * -1.62( -3.28, 0.06)
Base model + water damage -2.53( -5.08, 0.10) -1.14( -3.07, 0.83) -3.93( -5.86, -1.97) * -1.61( -3.24, 0.05)
Base model + mildew -2.60( -5.08, -0.05) * -1.11( -3.03, 0.86) -3.94( -5.84, -1.99) * -1.56( -3.20, 0.11)
Base model + parental stress -2.51( -5.06, 0.10) -0.95( -2.89, 1.02) -3.77( -5.81, -1.69) * -1.46( -3.14, 0.24)
Base model + doctor diagnosed asthma -2.65( -5.18, -0.06) * -1.11( -3.04, 0.86) -4.02( -5.99, -2.01) * -1.56( -3.22, 0.12)
Boys only -3.59( -6.71, -0.37) * -0.32( -3.02, 2.46) -5.17( -7.82, -2.45) * -0.89( -3.37, 1.66)
Girls only -1.29( -4.53, 2.07) -1.07( -3.67, 1.59) -2.55( -5.91, 0.93) -1.70( -4.14, 0.80)
Asthmatics only -3.89( -8.91, 1.40) -3.22( -8.59, 2.46) -6.36(-12.28, -0.03) * -5.14(-10.77, 0.84)
Non-asthmatics only -2.42( -6.09, 1.40) -0.75( -2.98, 1.54) -3.83( -6.54, -1.04) * -1.06( -2.97, 0.88)
Non-movers between years 6 and 10 -3.35( -6.28, -0.32) * -1.55( -3.69, 0.64) -5.25( -7.47, -2.97) * -2.01( -3.98, -0.01) *
FEV
1
FVC
128
Table 4.6. Associations of metals with FEV
1
adjusted for PM mass
a
.
a
Associations of attained lung function at age 15 and growth from 11 to 15 years of age were reported as percent changes along with
95% confidence intervals per unit increase in pollutant which was scaled to the difference between the 5
th
and 95
th
percentiles (see
Table 4.2 for pollutant specific values). All models were adjusted for height, BMI, race, Hispanic ethnicity, sex, respiratory illness,
and field technician.
* P<0.05
Attained age 15 Lung growth Attained age 15 Lung growth Attained age 15 Lung growth
Coarse
Cu 0.34( -2.85, 3.63) -0.54( -2.55, 1.51) -1.15( -3.81, 1.58) -0.09( -2.04, 1.89) 1.04( -2.70, 4.91) -0.79( -3.16, 1.65)
Fe -4.19(-11.55, 3.78) -3.73( -8.95, 1.80) -0.64( -3.21, 2.00) -0.42( -2.18, 1.37) -4.13(-11.52, 3.88) -3.69( -8.92, 1.84)
Ni -1.69( -6.48, 3.35) -2.74( -5.95, 0.58) -2.00( -4.50, 0.57) -0.83( -2.79, 1.17) -0.65( -6.28, 5.32) -3.28( -6.94, 0.52)
V -12.01(-20.39, -2.75) * -7.28(-14.10, 0.08) -3.57( -7.23, 0.23) -0.76( -3.33, 1.89) -10.48(-19.35, -0.63) * -7.42(-14.22, -0.07) *
Zn -0.52( -4.34, 3.45) -1.52( -4.22, 1.26) -0.93( -3.13, 1.33) -0.41( -2.04, 1.25) 2.20( -3.12, 7.80) -2.65( -6.50, 1.37)
Fine
Cu 1.38( -1.61, 4.47) -0.41( -2.84, 2.08) 0.62( -2.10, 3.43) -0.75( -2.94, 1.49) 2.11( -0.66, 4.95) 0.22( -2.10, 2.60)
Fe 1.42( -1.85, 4.80) -0.83( -3.59, 2.02) -1.14( -4.21, 2.03) -1.30( -3.84, 1.30) 1.51( -1.61, 4.74) -0.71( -3.37, 2.02)
Ni -0.22( -3.75, 3.43) -0.96( -3.91, 2.07) -1.57( -4.55, 1.50) -1.24( -3.80, 1.39) 1.14( -1.22, 3.56) 0.01( -1.96, 2.02)
V -0.92( -4.24, 2.51) -0.92( -3.73, 1.97) -1.14( -3.80, 1.61) -0.65( -2.96, 1.71) 1.25( -2.13, 4.74) -0.39( -3.23, 2.53)
Zn 0.23( -2.53, 3.06) -0.44( -2.72, 1.90) 0.77( -3.20, 4.91) -0.49( -3.77, 2.91) -0.46( -3.39, 2.55) -0.31( -2.78, 2.22)
Ultrafine
Cu 0.26( -2.74, 3.35) -0.03( -1.91, 1.88) 0.17( -3.24, 3.71) -0.38( -2.73, 2.03) 0.15( -2.36, 2.73) 0.08( -1.61, 1.80)
Fe -2.67( -6.60, 1.43) -0.01( -2.74, 2.80) -1.08( -4.35, 2.30) -1.13( -3.42, 1.22) -2.39( -5.99, 1.34) 0.19( -2.33, 2.78)
Ni -2.30( -6.59, 2.20) -2.32( -5.08, 0.52) -1.33( -4.99, 2.47) -1.49( -3.93, 1.01) -1.87( -5.61, 2.03) -2.04( -4.60, 0.60)
V -2.51( -6.43, 1.58) -1.80( -4.30, 0.76) -2.61( -6.69, 1.64) -1.97( -4.68, 0.83) -1.48( -4.71, 1.85) -1.22( -3.46, 1.07)
Zn -2.06( -5.74, 1.76) -0.85( -3.52, 1.88) -1.66( -4.82, 1.60) -0.88( -3.25, 1.55) -0.78( -3.88, 2.43) -0.16( -2.36, 2.08)
Total Fraction Water-Soluble Fraction Water-Insoluble Fraction
129
Table 4.7. Associations of metals with FVC adjusted for PM mass
a
.
a
Associations of attained lung function at age 15 and growth from 11 to 15 years of age were reported as percent changes along with
95% confidence intervals per unit increase in pollutant which was scaled to the difference between the 5
th
and 95
th
percentiles (see
Table 4.2 for pollutant specific values). All models were adjusted for height, BMI, race, Hispanic ethnicity, sex, respiratory illness,
and field technician.
* P<0.05
Attained age 15 Lung growth Attained age 15 Lung growth Attained age 15 Lung growth
Coarse
Cu -0.01( -2.70, 2.76) 0.28( -1.63, 2.23) -2.02( -4.03, 0.04) -0.36( -2.05, 1.36) 1.05( -2.10, 4.29) 0.47( -1.74, 2.73)
Fe 1.46( -6.09, 9.62) 0.64( -4.56, 6.13) -1.91( -3.67, -0.12) * -0.63( -2.12, 0.88) 1.55( -5.99, 9.71) 0.66( -4.55, 6.15)
Ni -0.21( -4.64, 4.42) 0.13( -3.05, 3.42) -2.17( -4.18, -0.12) * -0.68( -2.36, 1.04) 1.90( -3.26, 7.34) 0.81( -2.81, 4.57)
V -6.15(-14.77, 3.34) -1.78( -8.66, 5.62) -3.78( -6.42, -1.06) * -1.06( -3.38, 1.31) -2.28(-11.57, 7.98) -0.71( -7.67, 6.77)
Zn -0.72( -4.00, 2.68) -0.39( -2.87, 2.15) -1.35( -3.14, 0.48) -0.49( -1.92, 0.97) 3.48( -1.02, 8.18) 0.78( -2.76, 4.44)
Fine
Cu 1.12( -1.61, 3.93) 0.69( -1.51, 2.93) 0.68( -1.81, 3.24) 0.48( -1.51, 2.52) 1.33( -1.14, 3.87) 0.72( -1.33, 2.80)
Fe 1.05( -1.83, 4.02) 0.05( -2.36, 2.53) 0.57( -2.35, 3.57) 0.33( -2.00, 2.71) 1.00( -1.74, 3.83) 0.03( -2.28, 2.40)
Ni -2.40( -5.51, 0.81) -2.13( -4.74, 0.55) -2.96( -5.37, -0.49) * -1.65( -3.80, 0.55) 0.06( -2.08, 2.26) -0.90( -2.67, 0.91)
V -2.82( -5.49, -0.09) * -1.77( -4.12, 0.65) -2.54( -4.69, -0.34) * -1.26( -3.19, 0.71) 0.72( -2.43, 3.96) -0.69( -3.23, 1.92)
Zn -0.29( -2.68, 2.16) -0.44( -2.42, 1.58) -0.55( -3.96, 2.98) -0.68( -3.54, 2.27) 0.13( -2.54, 2.87) -0.05( -2.23, 2.18)
Ultrafine
Cu -1.22( -3.90, 1.53) -0.16( -1.85, 1.57) 0.87( -2.31, 4.15) 0.34( -1.77, 2.49) -1.48( -3.61, 0.70) -0.24( -1.72, 1.26)
Fe -3.59( -7.13, 0.07) -1.03( -3.46, 1.46) -0.06( -3.07, 3.04) -0.46( -2.51, 1.64) -3.49( -6.65, -0.23) * -0.93( -3.16, 1.35)
Ni -3.04( -7.10, 1.21) -1.07( -3.69, 1.63) -1.69( -5.07, 1.80) -0.83( -3.05, 1.45) -2.38( -5.86, 1.24) -0.67( -3.06, 1.78)
V -3.38( -7.16, 0.56) -1.34( -3.76, 1.13) -2.34( -6.14, 1.61) -0.64( -3.15, 1.93) -2.58( -5.58, 0.51) -1.23( -3.32, 0.90)
Zn -3.56( -6.68, -0.33) * -1.27( -3.60, 1.12) -1.06( -3.91, 1.87) -0.72( -2.78, 1.38) -3.16( -5.69, -0.56) * -0.78( -2.71, 1.18)
Total Fraction Water-Soluble Fraction Water-Insoluble Fraction
130
4.7 SUPPLEMENTARY MATERIAL
Supplemental Figure 4.1. Distribution of total metal concentrations (in ng/m
3
).
131
Supplemental Figure 4.2. Distribution of water-soluble metal concentrations (in ng/m
3
).
132
Supplemental Figure 4.3. Distribution of water-insoluble metal concentrations (in ng/m
3
).
133
4.8 REFERENCES
1. Peters JM, Avol E, Gauderman WJ, et al. A study of twelve Southern California
communities with differing levels and types of air pollution. II. Effects on
pulmonary function. Am J Respir Crit Care Med 1999;159(3):768-75.
2. Gauderman WJ, Avol E, Gilliland F, et al. The effect of air pollution on lung
development from 10 to 18 years of age. N Engl J Med 2004;351(11):1057-67.
3. Urman R, McConnell R, Islam T, et al. Associations of children's lung function
with ambient air pollution: joint effects of regional and near-roadway pollutants.
Thorax 2014;69(6):540-7.
4. Ben-Shlomo Y, Kuh D. A life course approach to chronic disease epidemiology:
conceptual models, empirical challenges and interdisciplinary perspectives.
International journal of epidemiology 2002;31(2):285-93.
5. Georgiopoulou VV, Kalogeropoulos AP, Psaty BM, et al. Lung function and risk
for heart failure among older adults: the Health ABC Study. The American
journal of medicine 2011;124(4):334-41.
6. Sin DD, Wu L, Man SF. The relationship between reduced lung function and
cardiovascular mortality: a population-based study and a systematic review of the
literature. Chest 2005;127(6):1952-9.
7. Anderson JO, Thundiyil JG, Stolbach A. Clearing the air: a review of the effects
of particulate matter air pollution on human health. Journal of medical toxicology
: official journal of the American College of Medical Toxicology 2012;8(2):166-
75.
8. Schwarze PE, Ovrevik J, Lag M, et al. Particulate matter properties and health
effects: consistency of epidemiological and toxicological studies. Human &
experimental toxicology 2006;25(10):559-79.
9. Chen LC, Lippmann M. Effects of metals within ambient air particulate matter
(PM) on human health. Inhalation toxicology 2009;21(1):1-31.
10. Merolla L, Richards RJ. In vitro effects of water-soluble metals present in UK
particulate matter. Experimental lung research 2005;31(7):671-83.
11. Carter JD, Ghio AJ, Samet JM, et al. Cytokine production by human airway
epithelial cells after exposure to an air pollution particle is metal-dependent.
Toxicology and applied pharmacology 1997;146(2):180-8.
12. Stohs SJ, Bagchi D. Oxidative mechanisms in the toxicity of metal ions. Free
radical biology & medicine 1995;18(2):321-36.
13. Xia T, Kovochich M, Liong M, et al. Comparison of the mechanism of toxicity of
zinc oxide and cerium oxide nanoparticles based on dissolution and oxidative
stress properties. ACS nano 2008;2(10):2121-34.
14. Kelly FJ, Fussell JC. Size, source and chemical composition as determinants of
toxicity attributable to ambient particulate matter. Atmospheric environment
2012;60(0):504-26.
15. Hasheminassab S, Daher N, Saffari A, et al. Spatial and temporal variability of
sources of ambient fine particulate matter (PM2.5) in California. Atmos Chem
Phys 2014;14(22):12085-97.
16. Lough GC, Schauer JJ, Park JS, et al. Emissions of metals associated with motor
vehicle roadways. Environmental science & technology 2005;39(3):826-36.
134
17. Gauderman WJ, Vora H, McConnell R, et al. Effect of exposure to traffic on lung
development from 10 to 18 years of age: a cohort study. Lancet
2007;369(9561):571-7.
18. McConnell R, Berhane K, Yao L, et al. Traffic, susceptibility, and childhood
asthma. Environ Health Perspect 2006;114(5):766-72.
19. Roemer W, Hoek G, Brunekreef B, et al. PM10 elemental composition and acute
respiratory health effects in European children (PEACE project). Pollution Effects
on Asthmatic Children in Europe. The European respiratory journal
2000;15(3):553-9.
20. Cakmak S, Dales R, Kauri LM, et al. Metal composition of fine particulate air
pollution and acute changes in cardiorespiratory physiology. Environmental
pollution 2014;189:208-14.
21. Islam T, Urman R, Gauderman WJ, et al. Parental stress increases the detrimental
effect of traffic exposure on children's lung function. Am J Respir Crit Care Med
2011;184(7):822-7.
22. Lee SJ, Demokritou P, Koutrakis P, et al. Development and evaluation of personal
respirable particulate sampler (PRPS). Atmospheric environment 2006;40(2):212-
24.
23. Fruin S, Urman R, Lurmann F, et al. Spatial Variation in Particulate Matter
Components over a Large Urban Area. Atmospheric environment 2014;83:211-9.
24. Gauderman WJ, Urman R, Avol E, et al. Association of improved air quality with
lung development in children. N Engl J Med 2015;372(10):905-13.
25. Wilson MR, Foucaud L, Barlow PG, et al. Nanoparticle interactions with zinc and
iron: implications for toxicology and inflammation. Toxicology and applied
pharmacology 2007;225(1):80-9.
26. Morina F, Jovanovic L, Mojovic M, et al. Zinc-induced oxidative stress in
Verbascum thapsus is caused by an accumulation of reactive oxygen species and
quinhydrone in the cell wall. Physiologia plantarum 2010;140(3):209-24.
27. Kelly FJ. Oxidative stress: its role in air pollution and adverse health effects.
Occupational and environmental medicine 2003;60(8):612-6.
28. Rosanna DP, Salvatore C. Reactive oxygen species, inflammation, and lung
diseases. Current pharmaceutical design 2012;18(26):3889-900.
29. Antonini JM, Stone S, Roberts JR, et al. Effect of short-term stainless steel
welding fume inhalation exposure on lung inflammation, injury, and defense
responses in rats. Toxicology and applied pharmacology 2007;223(3):234-45.
30. Riley MR, Boesewetter DE, Kim AM, et al. Effects of metals Cu, Fe, Ni, V, and
Zn on rat lung epithelial cells. Toxicology 2003;190(3):171-84.
31. Costa DL, Dreher KL. Bioavailable transition metals in particulate matter mediate
cardiopulmonary injury in healthy and compromised animal models. Environ
Health Perspect 1997;105 Suppl 5:1053-60.
32. Adamson IY, Prieditis H, Hedgecock C, et al. Zinc is the toxic factor in the lung
response to an atmospheric particulate sample. Toxicology and applied
pharmacology 2000;166(2):111-9.
33. McNeilly JD, Jimenez LA, Clay MF, et al. Soluble transition metals in welding
fumes cause inflammation via activation of NF-kappaB and AP-1. Toxicology
letters 2005;158(2):152-7.
135
34. McNeilly JD, Heal MR, Beverland IJ, et al. Soluble transition metals cause the
pro-inflammatory effects of welding fumes in vitro. Toxicology and applied
pharmacology 2004;196(1):95-107.
35. Delfino RJ, Sioutas C, Malik S. Potential role of ultrafine particles in associations
between airborne particle mass and cardiovascular health. Environ Health
Perspect 2005;113(8):934-46.
36. Oberdorster G. Pulmonary effects of inhaled ultrafine particles. International
archives of occupational and environmental health 2001;74(1):1-8.
37. HEI. Review Panel on Ultrafine Particles. Understanding the Health Effects of
Ambient Ultrafine Particles. HEI Perspectives 3: Health Effects Institute, Boston
MA., 2013.
38. Redlich CA, Tarlo SM. Longitudinal assessment of lung function decline in the
occupational setting. Current opinion in allergy and clinical immunology
2015;15(2):145-9.
39. Zheng G, Xu X, Li B, et al. Association between lung function in school children
and exposure to three transition metals from an e-waste recycling area. Journal of
exposure science & environmental epidemiology 2013;23(1):67-72.
40. Eeftens M, Hoek G, Gruzieva O, et al. Elemental composition of particulate
matter and the association with lung function. Epidemiology 2014;25(5):648-57.
41. Gehring U, Beelen R, Eeftens M, et al. Particulate matter composition and
respiratory health: the PIAMA Birth Cohort study. Epidemiology 2015;26(3):300-
9.
42. Rosa MJ, Perzanowski MS, Divjan A, et al. Association of recent exposure to
ambient metals on fractional exhaled nitric oxide in 9-11 year old inner-city
children. Nitric oxide : biology and chemistry / official journal of the Nitric Oxide
Society 2014;40:60-6.
43. Hirshon JM, Shardell M, Alles S, et al. Elevated ambient air zinc increases
pediatric asthma morbidity. Environ Health Perspect 2008;116(6):826-31.
44. Patel MM, Hoepner L, Garfinkel R, et al. Ambient metals, elemental carbon, and
wheeze and cough in New York City children through 24 months of age. Am J
Respir Crit Care Med 2009;180(11):1107-13.
45. Lurmann F, Avol E, Gilliland F. Emissions reduction policies and recent trends in
Southern California’s ambient air quality. Journal of the Air & Waste
Management Association 2014;65(3):324-35.
46. Minguillón MC, Arhami M, Schauer JJ, et al. Seasonal and spatial variations of
sources of fine and quasi-ultrafine particulate matter in neighborhoods near the
Los Angeles–Long Beach harbor. Atmospheric environment 2008;42(32):7317-
28.
47. Hu S, Polidori A, Arhami M, et al. Redox activity and chemical speciation of size
fractioned PM in the communities of the Los Angeles-Long Beach harbor. Atmos
Chem Phys 2008;8(21):6439-51.
136
CHAPTER 5: Summary and Suggestions for Future Research
The goal of this dissertation was to explore the relationship between the air
pollutant mixture and respiratory morbidity. While the main outcome of focus throughout
this dissertation was lung function, effects of air pollutants have been reported for asthma
and other respiratory symptoms as well as other non-respiratory morbidities (1, 2) and
thus some of the discussion and conclusions made throughout this dissertation should be
applicable to other air pollution studies. Early studies examining the effects of air
pollution on lung function relied on between-community differences in exposure for
making inferences. With this design of using regional air pollutants, it was often difficult
to partition the effects of individual pollutants due to their high correlation with one
another in the Los Angeles Air Basin. One hypothesis was that examining exposures on a
finer spatial scale may allow for the ability to disentangle the effects of individual
pollutants if within-community distributions of pollutants vary considerably. Within
urban communities, traffic represents the dominant source of air pollutants, but other
non-traffic sources, including industrial emissions, biomass burning, and (in coastal
cities) heavy oil burning from ships, may also contribute a significant amount (3-6).
Two important tasks that this dissertation tried to address were 1) to determine
whether there is a better way of modeling and assessing the effects of air pollutants and
2) to identify whether there are particular components of the air pollution mixture that
may be considerably more impactful on lung function than others. In the next section, I
will summarize the findings from Chapters 2 through 4 and comment on the analyses
presented throughout the dissertation. This will be followed by a brief discussion of
137
possible future research that could help further our understanding of the effects of the
pollutant mixture.
5.1 SUMMARY AND DISCUSSION
Chapter 2
The second chapter covered the joint effects of regional and near-roadway air
pollutants on lung function in children in grades 5 and 6. Using the subjects from the
Children’s Health Study as the study population, we demonstrated that near-roadway and
regional air pollution have independent adverse associations with lung function in
children. Approximately 1800 children from 5
th
grade and 6
th
grade classes across eight
Southern Californian communities underwent lung function testing. Near-roadway air
pollution and regional air pollution have been shown previously to adversely affect lung
function in children (7). Predicted near-roadway NO
2
, NO, and NO
x
, derived from a well
validated land-use regression model (8) and regional air pollutants (NO
2
, O
3
, PM
10
and
PM
2.5
) collected from central monitoring sites were assigned to all children. After
adjusting for factors known to be associated with lung function, higher levels of near-
roadway NO
2
, NO, and NO
x
were found to be associated with a reduction in FVC and
FEV
1
. Residential proximity to freeways, a surrogate of high traffic exposure, was also
found to be associated with lower FVC. Deficits in lung function were seen with regional
PM
10
, PM
2.5
and O
3
, but not regional NO
2
.
In a joint model of near-roadway and regional
air pollutant, the effects of each persisted after being adjusted for one another. Regional
air pollution exposure did not modify the effect of the near-roadway air pollutants.
138
Experimental studies have shown that NO
2
can induce lung damage in rats and
cultured cells and can induce an inflammatory response in individuals when exposed to
high occupational levels (9). Earlier CHS studies have shown associations between
regional NO
2
and lung function in children, but these associations may be a result of other
pollutants that are highly correlated with NO
2
(7, 10, 11). In Chapter 2, the associations
of lung function with regional PM
2.5
and PM
10
and lack of an association with regional
NO
2
, but an association with near-roadway NO
2
and NO
x
could be seen as somewhat
perplexing. One possible explanation for the associations with near-roadway NO
2
and
NO
x
but not with regional NO
2
is that the near-roadway NO
2
is simply a marker of the
near-roadway pollution mixture and that other components of this mixture are responsible
for the harmful effects on lung function. PM
2.5
composition has been of interest because
many of its components are known to induce oxidative stress and have been associated
with several health outcomes (12). Additional research is warranted to determine
whether NO
2
is simply a marker for some other causal pollutant or if NO
2
is indeed partly
responsible for the observed effects on lung function.
Chapter 3
The third chapter covered the modeling of elemental carbon (EC) and particulate
matter in two size fractions: 2.5 µm in aerodynamic diameter and in a nanoscale
(ultrafine) size less than 0.2 µm. Studies have shown that near-roadway air pollution is
associated with numerous health outcomes. Oxides of nitrogen have been measured to
develop near-roadway prediction models because of their association with near-roadway
vehicular emissions and the modest measurement cost. However, elemental carbon and
139
other components of particulate matter may be more biologically relevant exposures. Few
studies have developed EC and PM exposure models on a fine spatial scale of tens to
hundreds of meters that are relevant for epidemiological studies. About 240 sites across 8
of the Southern California CHS communities were selected to monitor components of
PM across 8 weeks out of the year. Using a supervised forward selection process, land-
use regression models were developed for EC and PM in each of the size fractions. The
CALINE4 line source dispersion model explained 51% of the within-community
variability of EC
0.2
and, with population density, explained 53% of the within-community
variability of EC
2.5
across all eight communities. Community-specific models identified
additional predictors for EC
2.5
and EC
0.2
. Models for PM
2.5
and PM
0.2
explained a
substantially lower percentage of the total variability.
The variability predicted by the land-use regression models for EC
2.5
and EC
0.2
were lower than the variability predicted by land-use regression models for NO
2
(71%)
and NO
x
(71%) that were eventually used in the analysis presented in Chapter 2 (8). It is
difficult to say why the models for EC performed worse than the earlier presented NO
2
and NO
x
models. One possible reason for the poorer performance is that EC is produced
mainly by diesel-powered trucks, which comprise only a fraction of motor vehicles in
Southern California and are found mostly on freeways and not minor roads. NO
2
and
NO
x
are produced by most vehicles found on both freeways and minor roads. It was not
surprising that the model for PM
2.5
was poor given that prior studies have shown little
spatial variability downwind of roadways (13, 14). The performance of the PM
0.2
model
was on the other hand unexpected given that these same studies also showed strong
distance-decay relationship downwind of roadways for ultrafine particles (13, 14). While
140
PM
0.2
is enriched with PM
0.1
, the majority of the mass is likely to be greater than 0.1 µm
in the accumulation mode and thus might not be expected to have the same spatial pattern
as ultrafine particles. Additional research is needed to determine the spatial variability of
PM
0.2
.
The main tool for assigning home exposures to NO
2
and NO
x
(as surrogate of the
near-roadway pollution mixture) and to model the variability of PM and EC was land-use
regression. Several review articles on land-use regression have been written describing its
uses and its strengths and limitations (15-17). One limitation of land-use regression that I
have highlighted is that several modeled pollutants may be even more highly correlated
than the measured pollutant concentrations themselves. During the collection of PM and
its constituents at selected locations, NO
2
and NO
x
measurements were also taken at a
subset of these locations. A model containing only CALINE4 estimates of near-roadway
pollution, elevation, and NDVI, explained 72% of the within community variability of
NO
x
(data not published), which is similar to the amount of variability predicted from the
earlier NO
x
model briefly described in Chapter 2 and found in more detail in Franklin et
al. (8). The correlation between the observed within-community distributions of EC
2.5
and NO
x
was 0.77. However, the correlation between the predicted within-community
distributions of EC
2.5
and NO
x
was 0.91 despite the fact that the model for EC
2.5
was only
able to predict approximately half of the within-community variability. The high
correlation between the predicted EC
2.5
and NO
x
concentrations was induced because
both models had the traffic marker, CALINE4, explaining most of the predictable
variability. For this reason, the effect of predicted EC on lung function was not
developed as a chapter in this dissertation. This has important implications, as using
141
predicted exposures from land-use regression models in epidemiological studies would
make the ability to disentangle the effects of different PM components difficult if they
have a common, dominant source, hence similar predictor covariates. It may also be
unnecessarily costly to measure multiple tracer pollutants for modeling if they have a
common source. Further research is warranted in other exposure data sets to determine if
this insight is more generally applicable.
Chapter 4
In the fourth chapter, the effect of transition metals on lung function was
investigated. Many studies have shown an adverse effect of total PM on lung function in
children, but few epidemiological studies have examined effects of particle composition.
The same population of children from Chapter 2, now followed longitudinally, was used
to explore the effects of transition metals. Concentrations of metals, including the novel
water soluble fraction, across multiple size fractions were collected as part of the
sampling campaign described in the third chapter. School concentrations of five metals
(Cu, Fe, Ni, V, and Zn) were assigned to children. Water-soluble Ni, V, and Zn in the
fine fraction were associated with deficits on attained lung function at age 15. The
associations of fine water-soluble Ni and V persisted after adjustment for mass. Total fine
PM, but none of the metals, was associated with FEV
1
. Neither total PM nor the metals
were significantly associated with lung growth between 11 and 15 years of age.
Ni and V are both markers of heavy fuel oil combustion in ships. While not
reported in Chapter 4, the correlation between total fine Ni and V at the schools was 0.95,
which explains the similar magnitude of associations of both metals that was reported.
142
Other metals also were highly correlated. For example, in the total fine fraction, Cu, Fe,
and Zn all had a correlation with one another of at least 0.67, potentially indicating that
they were emitted from a common source, for example brake and tire wear. Like the CHS
studies investigating the effects of multiple regional pollutants, the correlation between
metals (and potentially other elements and organic compounds found in PM) limited the
ability to distinguish individual pollutant effects from one another.
In Chapter 4, the focus was on identifying whether certain transition metals are
harmful to children’s lungs. However, for many metals as well as other components of
the air pollution mixture, there could be many potential sources. From a policy
standpoint, an alternative to regulating individual pollutants that are produced by many
sources would be to regulate sources. In addition, policies and regulations can be enacted
to minimize total exposures from potentially harmful sources without having to identify,
at this point in time, those particular chemical compounds that are most likely responsible
for observed health effects. Thus, instead of studying the effects of individual
components of the air pollution mixture, studying the effect of packages of pollutants
originating from potentially hazardous sources (for example vehicle combustion
products, engine wear and brake wear, heavy duty vehicles emitting diesel exhaust
particulate, and power plants) on health endpoints would be an alternative approach to
translating air pollution research into policies that could quickly benefit the general
population.
143
5.2 SUGGESTIONS FOR FUTURE RESEARCH
In Chapter 2, we used NO
x
as a tracer pollutant for traffic, and showed
associations cross-sectionally with lung function. In Chapter 3, we tried modeling EC and
PM as potentially better markers of traffic. In Chapter 4, associations were observed of
water-soluble Ni and V in the fine fraction with attained FVC level. We studied transition
metals because they are potent oxidants and speculated that exposure to these metals
would likely induce oxidative stress and subsequent physiological changes to the lungs.
However, the absence of an Fe effect was puzzling despite its high concentrations. One
reason we saw associations mainly with Ni and V, and not with Fe may have been that
there was something about the source instead of specific metals that was responsible for
the adverse effects observed with lung function. We speculated that the source for Ni and
V was heavy oil combustion from the ships that travel along the coast and unload cargo at
the ports of Los Angeles and Long Beach. Besides traffic and ship emissions, there are
other potential sources of toxic pollutants in the Southern Californian air basin (18). An
in depth analysis of the effects of pollution sources on respiratory outcomes is warranted.
Source Apportionment
One possible future study would be to use data reduction methods or run source
apportionment, for example with receptor model Positive Matrix Factorization (PMF), to
identify sources more formally and to examine their association with respiratory
outcomes. This study might have two main objectives. The first objective could be to run
source apportionment or other data reduction methods on available data (PM composition
and NO
x
) to identify specific sources. One likely signature that we would anticipate
144
seeing is that of a traffic source as we have seen both EC and NO
x
correlate strongly with
traffic predictors in land-use regression models. Potentially it may be possible to identify
different components of a traffic source such as diesel exhaust emissions, tire and break
wear, and resuspended road dust. In addition, we might also identify a signature for a
heavy oil fuel source from the ships entering and leaving the ports of Los Angeles and
Long Beach. Identifying other sources of potentially toxic pollutants within these CHS
communities could lead to new testable hypotheses about whether they are harmful to
children. Source apportionment using PMF has been started by the CHS team.
Currently, we have elemental and organic carbon concentration measurements for
two seasons (4-weeks each) at several homes and schools as described in Chapter 3. We
also have complete 8-week coverage of elemental composition (48 different elements) at
all schools. However, we only have half of the PM analyzed for metal concentrations at
the homes (single season). If we had another season, we can do a whole year source
apportionment which may provide better estimates of the annual levels of these sources.
A small grant will need to be developed to analyze the remaining PM for its elemental
composition.
The second objective could be to use these sources in an epidemiological study to
see if they relate to respiratory outcomes. One possible approach would be to use the
identified source profiles at all of the schools and run an analysis similar to the one in
Chapter 4 whereby all children from that school were assigned the exposures of the
elementary schools they attended. Another approach would be to attempt to model the
sources to predict the source profiles at the children’s residences in the neighborhood of
each school. We could predict the source profiles either through land-use regression
145
modeling or geostatistical interpolation, such as kriging or non-parametric smoothing. A
potential benefit of using geostatistical interpolation is that it might generate predicted
source profiles that are less correlated with one another as might happen if we try
modeling the source profiles with land-use regression, especially if source profiles are
spatially correlated (diesel exhaust and tire wear, for example).
Long Beach Study
In chapter 4, we saw strong associations of Ni and V with attained lung function.
As noted above, the likely source for these metals is heavy oil combustion from the ships
near the ports of Los Angeles and Long Beach. In addition to ship emissions, there are
numerous other sources of pollutants in Long Beach including industrial, utility, and
trucking emissions (19). Another possible future research direction would be to study the
within-community variability of the pollutants emitted from these different sources in the
Long Beach area. However, based on our previous health findings with Ni and V, the
main focus of the study would be to understand the within-community distribution of
ship emissions on a fine scale of kilometers (rather than tens of kilometers) and determine
whether these emissions are related to health outcomes.
To expand on this idea of identifying whether ship emissions could be responsible
for adverse respiratory effects, an epidemiological would compare respiratory
measurements (e.g., lung function, markers of inflammation, asthma status, respiratory
symptoms) from newly recruited children in Long Beach to children downwind and in
neighboring communities, where levels of ship emissions should be lower. By selecting
children in this fashion, it would be possible to maximize the gradient of observed ship
146
emissions. Currently available data show that there is considerable within-community
variability of Ni and V in the quasi-ultrafine fraction in Long Beach that is not present in
other CHS communities (see Supplemental Figure 4.1). Ambient PM can be measured at
several locations and tracers for ships emissions (Ni and V) can be used to help quantify
the fraction of PM that is derived from ships. Additionally, PMF or other data reduction
techniques can again be employed to identify other elements that could possibly spatially
co-vary with Ni and V. After recruiting children in these communities and collecting
information on their respiratory health, spatial models can be employed to assign
residential exposure to ship emissions. Appropriate statistical methods (e.g. linear and
logistic regression) can then be applied to relate exposure to ship emissions to the several
selected markers of respiratory health.
To conclude, both regional air pollutants and near-roadway exposure to NO
x
are
important determinants of lung function. However, it is not apparent whether NO
x
is
harmful or if it is serving as a surrogate for other pollutants. While land-use regression
models were developed to predict EC for use in epidemiologic studies, these models
offered little additional value as the predictions from these models are highly correlated
with predictions from a NO
x
model. Transition metals are potentially toxic for lung
development. Additional research is warranted to explore the relative effects of different
fractions of the air pollution mixture. While identifying the key causal pollutant or
pollutants out of a package of correlated pollutants may be a challenging proposition,
identifying potentially harmful sources of pollutants may lead to continued and directed
efforts to clean the air children and adults breathe and to reduce the risk of lung injury
associated with breathing noxious pollutants.
147
5.3 REFERENCES
1. Ruckerl R, Schneider A, Breitner S, et al. Health effects of particulate air pollution: A
review of epidemiological evidence. Inhalation toxicology 2011;23(10):555-92.
2. Costa S, Ferreira J, Silveira C, et al. Integrating health on air quality assessment--review
report on health risks of two major European outdoor air pollutants: PM and NO(2).
Journal of toxicology and environmental health Part B, Critical reviews 2014;17(6):307-
40.
3. HEI. Traffic-related air pollution: a critical review of the literature on emissions,
exposure, and health effects. HEI Special Report: Health Effects Institute, 2010.
4. Mysliwiec MJ, Kleeman MJ. Source apportionment of secondary airborne particulate
matter in a polluted atmosphere. Environmental science & technology 2002;36(24):5376-
84.
5. Minguillón MC, Arhami M, Schauer JJ, et al. Seasonal and spatial variations of sources
of fine and quasi-ultrafine particulate matter in neighborhoods near the Los Angeles–
Long Beach harbor. Atmospheric environment 2008;42(32):7317-28.
6. Pakbin P, Ning Z, Shafer MM, et al. Seasonal and Spatial Coarse Particle Elemental
Concentrations in the Los Angeles Area. Aerosol Science and Technology
2011;45(8):949-63.
7. Gauderman WJ, Vora H, McConnell R, et al. Effect of exposure to traffic on lung
development from 10 to 18 years of age: a cohort study. Lancet 2007;369(9561):571-7.
8. Franklin M, Vora H, Avol E, et al. Predictors of intra-community variation in air quality.
J Expo Sci Environ Epidemiol 2012;22(2):135-47.
9. Kelly FJ. Oxidative stress: its role in air pollution and adverse health effects.
Occupational and environmental medicine 2003;60(8):612-6.
10. Peters JM, Avol E, Gauderman WJ, et al. A study of twelve Southern California
communities with differing levels and types of air pollution. II. Effects on pulmonary
function. Am J Respir Crit Care Med 1999;159(3):768-75.
11. Gauderman WJ, Avol E, Gilliland F, et al. The effect of air pollution on lung
development from 10 to 18 years of age. N Engl J Med 2004;351(11):1057-67.
12. Kelly FJ, Fussell JC. Size, source and chemical composition as determinants of toxicity
attributable to ambient particulate matter. Atmospheric environment 2012;60(0):504-26.
13. Zhu YF, Hinds WC, Kim S, et al. Concentration and size distribution of ultrafine
particles near a major highway. Journal of the Air & Waste Management Association
2002;52:1032-42.
14. Beckerman B, Jerrett M, Brook JR, et al. Correlation of nitrogen dioxide with other
traffic pollutants near a major expressway. Atmospheric Environment 2008;42(2):275-90.
15. Jerrett M, Arain A, Kanaroglou P, et al. A review and evaluation of intraurban air
pollution exposure models. Journal of exposure analysis and environmental
epidemiology 2005;15(2):185-204.
16. Ryan PH, Lemasters GK, Biswas P, et al. A comparison of proximity and land use
regression traffic exposure models and wheezing in infants. Environmental health
perspectives 2007;115(2):278-84.
17. Hoek G, Beelen R, de Hoogh K, et al. A review of land-use regression models to assess
spatial variation of outdoor air pollution. Atmospheric Environment 2008;42(33):7561-
78.
18. Hasheminassab S, Daher N, Saffari A, et al. Spatial and temporal variability of sources of
ambient fine particulate matter (PM2.5) in California. Atmos Chem Phys
2014;14(22):12085-97.
148
19. Motallebi N, Taylor CA, Jr., Croes BE. Particulate matter in California: part 2--Spatial,
temporal, and compositional patterns of PM2.5, PM10-2.5, and PM10. J Air Waste
Manag Assoc 2003;53(12):1517-30.
149
Abstract (if available)
Abstract
The effect of air pollution on human health has long been a concern. The aim of this dissertation is to explore the negative impact of probable causal pollutants or pollutant mixtures on respiratory health, with the focus mainly on the impacts of local or near‐roadway air pollution and that of elemental composition of particulate matter (PM) on lung function in children and adolescence. In urbanized locations, such as the Southern California communities that constitute the Children’s Health Study (CHS) and from which the population at focus in this dissertation is derived, motorized vehicles contribute a significantly large amount of the air pollution at both the local and regional scales. Prior studies have reported adverse effects of either regional or near‐roadway air pollution on lung function, but little has been done of the joint effects of these exposures. In the first study, analyses were conducted to assess the joint effects of these exposures on childhood lung function in the CHS. Results indicate that near‐roadway and regional air pollution have independent adverse effects on childhood lung function. However, specific components of the near‐roadway pollution mixture responsible for these effects have not been established. A major limitation for health studies is the lack of exposure models that estimate these components observed in epidemiological studies over fine spatial scale of tens to hundreds of meters. In the second study, exposure models were developed for fine‐scale variation in biologically relevant elemental carbon (EC). Models that included traffic measures provided useful estimates for EC₀.₂ and EC₂.₅ on a spatial scale appropriate for health studies of near-roadway pollution in selected Southern California communities. Moreover, numerous studies have reported adverse effects of PM on lung function in children, but there has been little investigation of the chronic effects of PM composition. Transition metals are biologically plausible agents contributing to these effects. Aerosol size and water-solubility may be important determinants of metal toxicity. In the third study, associations between lung development and concentrations of PM and transition metals (copper [Cu], iron [Fe], nickel [Ni], vanadium [V], and zinc [Zn]), in three size fractions [quasi-ultrafine (<0.2 μm), fine (0.2 to 2.5 μm) and coarse (2.5 to 10 μm)] were explored. Results from this study suggest that specific water-soluble transition metals found in PM, namely Ni and V in the fine fraction, might explain some of the adverse effects of PM on childhood lung function.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Long term effects of air pollution on lung function in the European community respiratory health survey
PDF
Evaluation of new methods for estimating exposure to traffic-related pollution and early health effects for large population epidemiological studies
PDF
Disparities in exposure to traffic-related pollution sources by self-identified and ancestral Hispanic descent in participants of the USC Children’s Health Study
PDF
Spatial modeling of non-tailpipe emissions and its association with children's lung function
PDF
Airway inflammation and respiratory health in the Southern California children's health study
PDF
Examining exposure to extreme heat and air pollution and its effects on all-cause, cardiovascular, and respiratory mortality in California: effect modification by the social deprivation index
PDF
Prenatal air pollution exposure, newborn DNA methylation, and childhood respiratory health
PDF
Associations between ambient air pollution and hypertensive disorders of pregnancy
PDF
Association of traffic-related pollution and stress on childhood lung function
PDF
Air pollution, mitochondrial function, and growth in children
PDF
Sensor-based mobile health approaches for personal air pollution and pediatric asthma studies
PDF
Persistence of pollution-induced lung function deficits in early adulthood: evidence from the Children's Health Study
PDF
Air pollution and childhood obesity
PDF
Early life risk factors for childhood asthma
PDF
Effects of stress and the social environment on childhood asthma in the children' s health study
PDF
A cohort study of air-pollution and childhood obesity incidence
PDF
Genetic variation in inducible nitric oxide synthase promoter, residential traffic related air pollution and exhaled nitric oxide in children
PDF
Assessing the impact of air pollution on adverse birth outcomes in a low resource setting
PDF
Early life air pollution exposure, gestational diabetes mellitus, and autism spectrum disorder
PDF
Effects of near-roadway air pollution exposure on obesity, obesity-related behavior, and neurobehavioral deficits during peripuberty
Asset Metadata
Creator
Urman, Robert
(author)
Core Title
Ambient air pollution and lung function in children
School
Keck School of Medicine
Degree
Doctor of Philosophy
Degree Program
Epidemiology
Publication Date
07/10/2015
Defense Date
06/15/2015
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
Air pollution,Children,lung function,OAI-PMH Harvest,Traffic
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
McConnell, Rob (
committee chair
), Fruin, Scott (
committee member
), Gauderman, William James (
committee member
), Gilliland, Frank D. (
committee member
), Wilson, John P. (
committee member
)
Creator Email
rurman@gmail.com,rurman@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-592394
Unique identifier
UC11301882
Identifier
etd-UrmanRober-3589.pdf (filename),usctheses-c3-592394 (legacy record id)
Legacy Identifier
etd-UrmanRober-3589.pdf
Dmrecord
592394
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Urman, Robert
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
lung function