Page 1
Ambient Metals, Elemental Carbon, and Wheeze and Cough in New York City Children
through Age 24 Months
Molini M. Patel1,2
, Lori Hoepner2,3
, Robin Garfinkel2,3
, Steven Chillrud2,4
, Andria Reyes2,3
,
James W. Quinn5, Frederica Perera
2,3, Rachel L. Miller
1,2,3
1Division of Pulmonary, Allergy, and Critical Care Medicine, Department of Medicine
College of Physicians and Surgeons, Columbia University
2Columbia Center for Children’s Environmental Health, Columbia University
3Department of Environmental Health Sciences, Mailman School of Public Health, Columbia
University
4Lamont-Doherty Earth Observatory, Columbia University
5The Institute for Social and Economic Research and Policy, Columbia University
Correspondence and request for reprints should be addressed to Rachel L. Miller M.D.,
PH8E, Columbia University College of Physicians and Surgeons, 630 W. 168th
St, New York,
NY 10032. Phone: 212-305-7759, fax: 212-305-2277, Email: [email protected] .
This manuscript has an Online Data Supplement, which is accessible in this issue’s table of
content online www.atsjournals.org
Sources of funding: Funding for the study is provided by the National Institute of
Environmental Health Sciences (grants R01 ES013163, P50ES015905, P01 ES009600, P30 ES
009089, and R01 ES008977), U.S. Environmental Protection Agency (grants R827027, RD-
832141), Irving General Clinical Research Center (grant RR00645), Educational Foundation of
Page 1 of 47 AJRCCM Articles in Press. Published on September 10, 2009 as doi:10.1164/rccm.200901-0122OC
Copyright (C) 2009 by the American Thoracic Society.
Page 2
1
America, Gladys & Roland Harriman Foundation, The New York Community Trust, and
Trustees of the Blanchette Hooker Rockefeller Fund.
Running Head: Ambient Metals and Asthma in Children
Descriptor Number: 101 – Asthma in children or 57 – Asthma: epidemiology
Word Count: 4319
AT A GLANCE COMMENTARY
Scientific Knowledge on the Subject
Associations between PM2.5 and asthma development and acute asthma exacerbations are well-
documented. However, health effects of exposure to specific airborne components from traffic
and heating oil combustion, including metals and elemental carbon, have not been fully
characterized.
What This Study Adds to the Field
This paper presents new evidence that implicates exposures to ambient nickel, vanadium, and
elemental carbon as possible risk factors for respiratory symptoms in a young inner city cohort.
The report provides evidence that exposures to PM2.5-associated metals and elemental carbon
from sources such as heating oil combustion and traffic may be important health-relevant PM2.5
fractions associated with asthma morbidity in urban children as young as age 2 years.
Page 2 of 47
Page 3
2
Abstract
Rationale: The effects of exposure to specific components of ambient fine particulate matter
(PM2.5), including metals and elemental carbon (EC), have not been fully characterized in young
children.
Objectives: To compare temporal associations among PM2.5; individual metal constituents of
ambient PM2.5, including nickel (Ni), vanadium (V), and zinc (Zn); and EC and longitudinal
reports of respiratory symptoms through age 24 months.
Methods: Study participants were selected from the Columbia Center for Children’s
Environmental Health (CCCEH) birth cohort recruited in New York City between 1998 and
2006. Respiratory symptom data were collected by questionnaire every 3 months, through age 24
months. Ambient pollutant data were obtained from state-operated stationary monitoring sites
located within the study area. For each subject, 3-month average inverse-distance weighted
concentrations of Ni, V, Zn, EC, and PM2.5 were calculated for each symptom reporting period
based on the questionnaire date and the preceding 3 months. Associations between pollutants and
symptoms were characterized using generalized additive mixed effects models, adjusting for sex,
ethnicity, environmental tobacco smoke exposure, and calendar time.
Measurements and Main Results: Increases in ambient Ni and V concentrations were associated
significantly with increased probability of wheeze. Increases in EC also were associated
significantly with cough during “cold/flu season”. Total PM2.5 was not associated with either
wheeze or cough.
Conclusions: These results suggest that exposure to ambient metals and elemental carbon from
heating oil and/or traffic at levels characteristic of urban environments may be associated with
respiratory symptoms among very young children.
Page 3 of 47
Page 4
3
247 words
Keywords: traffic, heating oil combustion, metals, asthma
Page 4 of 47
Page 5
4
Introduction
Epidemiologic evidence links increases in ambient levels of fine particulate matter
(PM2.5) to asthma exacerbations, lung function decrements, and greater utilization of medical
services for asthma [1-3]. Because of geographic and seasonal differences in PM2.5 composition
and PM2.5-associated health effects [4-7] current mass-based standards for ambient PM2.5 may
not adequately target specific components that are causally associated with adverse health
effects. Diesel exhaust particles (DEP) are a significant driver of local urban PM2.5 levels and are
a dominant source of atmospheric elemental carbon (EC) [8]. Traffic is an important source of
ambient metals from tailpipe emissions, brake and tire abrasion, and resuspended roadway dust
[9, 10]. In New York City (NYC), residual oil combustion for heating contributes to ambient
nickel (Ni) and vanadium (V) concentrations that exceed levels in most other US cities [6, 11].
Given the large contributions of traffic and heating oil combustion to urban ambient PM2.5 levels,
there is a need to characterize the contributions of specific components such as metals and EC to
adverse health effects.
Studies have demonstrated that communities with higher EC concentrations have higher
prevalence of asthma and chronic respiratory symptoms [12, 13]. More recently, proximity to
major roadways has been associated with chronic respiratory symptoms, asthma, and allergic
sensitization [14, 15]. One key longitudinal study in Southern California observed that long-term
exposure to EC, PM2.5, nitrogen dioxide, and acid vapors, derived primarily from motor vehicle
emissions, were associated with deficits in lung function growth between ages 10-18 [16].
Relatively fewer studies have examined respiratory health effects associated with ambient
metals exposures. In a national-scale study, PM2.5-associated risks of respiratory and
cardiovascular hospital admissions were higher in communities with higher levels of PM2.5-
Page 5 of 47
Page 6
5
related Ni, V, and EC [5]. Recently, increases in ambient zinc (Zn) were associated with
increases in asthma emergency department visits and hospital admissions among children living
in Baltimore [17]. Mechanistic support is provided by observations of greater release of
proinflammatory cytokines from airway cells in response to metals exposure [18, 19]. Studies are
needed that elucidate the potential health effects of ambient metal exposures in young children
living in urban communities with high asthma morbidity [20] and examine the differential health
effects associated with exposures to ambient metals, EC, and PM2.5.
We hypothesized that exposure to ambient metals and EC would be associated with
wheeze and cough among young urban children. In a longitudinal design, associations between
local measurements of ambient metals, EC, and PM2.5 and concurrent respiratory symptoms
among children through age 24 months were characterized. The findings provide evidence of a
link between the disproportionately high burden of ambient metals and diesel emission sources
and disproportionately high asthma morbidity among young residents of NYC and possibly other
cities. Some results have previously been reported in the form of abstracts [21, 22].
METHODS
Study cohort data
Detailed methods are provided in the online supplement. Children living in Northern
Manhattan and the South Bronx were enrolled between 1998 and 2006 into a prospective birth
cohort study conducted by the Columbia Center for Children’s Environmental Health (CCCEH)
[23-25]. Briefly, 725 fully enrolled pregnant women, recruited from prenatal clinics associated
with New York Presbyterian Medical Center or Harlem Hospital, were followed throughout
Page 6 of 47
Page 7
6
pregnancy and provided maternal and/or cord blood at delivery. Informed consent was obtained
in accordance with the Columbia University Institutional Review Board. Data on subject
characteristics, residence, environmental tobacco smoke exposure (ETS), and respiratory
symptoms were collected by questionnaires administered to mothers in person or by telephone
every 3 months, between child ages 3 and 24 months.
Stationary site monitoring
Twenty-four hour average ambient concentrations of PM2.5 and PM2.5 fractions of Ni, V,
Zn, and EC were measured every third day by the New York State Department of Environmental
Conservation between 1999 and 2007. Datasets were downloaded (http://www.dec.ny.gov/) for
two sites in the Bronx that were located in the study area: New York Botanical Gardens (NYBG)
and Intermediate School 52 (IS52). Data were aggregated by week and site, as described [26].
Statistical analysis
Associations between metals, EC, and PM2.5 and presence of wheeze and cough were
analyzed using generalized additive mixed effects models (GAMM) using the mgcv library in R
version 2.9.0 (R Foundation for Statistical Computing, Vienna, Austria). Nitrogen dioxide (NO2)
was evaluated as a gaseous indicator of traffic emissions. Single pollutant models were
constructed in which each pollutant was analyzed as a parametric continuous variable. For each
subject, 3-month moving average concentrations of Ni, V, Zn, EC, and PM2.5 were calculated for
each symptom reporting period based on the follow-up questionnaire date and the preceding 3
months. Exposures were assigned to subjects by calculating inverse-distance weighted
concentrations using pollutant measurements from IS52 and NYBG. Address data were collected
Page 7 of 47
Page 8
7
only at prenatal, 6, 12, and 24 month questionnaires, and addresses at interim time points were
assigned using data on moves since last questionnaire and previous addresses. A first order
autoregressive correlation structure was specified to account for correlation among the repeated
observations collected over a 2-year period from each subject. Other covariates included
parametric terms for sex, ethnicity, postnatal ETS, and a nonparametric smoothed term for
calendar time using natural cubic splines (4.7 degrees of freedom per calendar year).
The robustness of results was evaluated using the following methods: models that
included gaseous and particulate copollutants related to traffic, models that excluded the highest
5% of pollutant concentrations, and analyses stratified by season. For stratified analyses, season
was defined as a dichotomous variable: “cold/flu season” (September 1 to March 31) or “non-
cold/flu season” (April 1 to August 31).
In descriptive summaries of symptom prevalence and pollutant levels, season was defined
by calendar year as follows: winter = December 21-March 20, spring = March 21-June 20,
summer = June 21-September 20, and fall = September 21-December 20. Except for GAMM,
statistical tests and modeling were performed using SAS 9.1.3 (Cary, NC, release 2005), and
results with p<0.05 were considered statistically significant.
RESULTS
Cohort characteristics
Among 687 subjects who reached their 2nd
birthday by October 31, 2007, 653 (90% of
the fully enrolled) provided any follow-up data, and thus were included in this study. Seventy
five percent of participants completed at least 5 of the 8 follow-up questionnaires, and 20%
Page 8 of 47
Page 9
8
completed all 8 questionnaires. Thirty-four subjects (5%) were lost to follow-up between ages 3
and 24 months. Characteristics of the study population are shown in Table 1. Approximately
64% of mothers identified themselves as Dominican, and 36% identified themselves as African
Americans. A majority of mothers had at least a twelfth-grade education, and greater than 90%
reported receiving Medicaid at enrollment. At age 24 months, 30% of children were told by a
doctor that they have or may have asthma. There were no significant differences in any of the
displayed demographics between the full cohort at age 24 months, and the subgroup that had
completed one or more follow-up questionnaires.
Spatial and temporal trends in ambient metals, EC, and PM2.5
Across all time points, 72-78% of subjects lived closer to IS52, and 22-28% of subjects
lived closer to NYBG. The range of mean distance was 3.9-4.2 km for IS52 and 5.8-5.9 km for
NYBG. Between 2000 and 2007, mean concentrations of Ni, EC, and NO2 varied significantly
between IS52 and NYBG (Tables E1-E4 of online supplement). Mean concentrations of
pollutants also varied significantly by season (Table E5 of online supplement). Concentrations of
metals in fall and winter often were double the levels in spring and summer, whereas EC
concentrations were higher in winter and fall by approximately 27%. PM2.5 concentrations were
significantly higher in winter and summer by 24%. NO2 concentrations were significantly higher
in winter and spring, however, by less than 10%.
Prevalence of wheeze and cough
Forty seven percent of subjects reported wheeze during at least 1 follow-up period
through age 24 months, whereas 89% reported cough. The overall prevalence of wheeze and
Page 9 of 47
Page 10
9
cough did not change over the study period between 1998 and 2007, but did display consistent
seasonal patterns, with maxima in winter and fall months and minima in spring and summer
months (Figure 1). The proportions of subjects reporting wheeze in the fall and winter were
similar and averaged approximately 20% and 19%, respectively. The proportions of subjects
reporting wheeze in the spring and summer were similar and averaged approximately 14% each.
The proportion of subjects reporting cough was highest in the fall at 56.2%. The proportion of
subjects reporting cough in the winter, summer, and spring and summer were 53.3%, 40.3%, and
37.0%, respectively.
Association between ambient metals, EC, PM2.5 and wheeze and cough
Significant positive associations were observed between metals and wheeze but not
cough. Among all pollutants evaluated, the largest effect estimates were observed in association
with Ni exposure. In models that adjusted for sex, ethnicity, postnatal ETS exposure, and
calendar time, an increase in interquartile range (IQR) concentration of ambient Ni (0.014 µg/m3)
was associated significantly with 28% increased probability of wheeze (p = 0.0006) (Table 2).
These findings were robust to the inclusion of the copollutants EC, NO2, copper (Cu), and iron
(Fe), with an 11% decrease in the magnitude of effect.
Vanadium and wheeze were not significantly associated in the singlepollutant model
(Table 2). An IQR (0.003 µg/m3) increase in 3-month average concentrations of V was
associated with a 10% increased probability of wheeze (p = 0.13). However, after adjusting for
EC, NO2, Cu, and Fe, there was suggestion of association between V and wheeze (β = 0.14 per
IQR increase in V, p = 0.08). Zinc was not associated with wheeze in either single or
multipollutant models.
Page 10 of 47
Page 11
10
Elemental carbon was not significantly associated with wheeze or cough in either single
or multipollutant models that included NO2 and Ni (Table 2). Additionally, total PM2.5 was not
associated significantly with wheeze or cough in single pollutant models. PM2.5 was negatively
associated with wheeze in a multipollutant model that included NO2 and Ni (β = -0.13 per IQR
increase in PM2.5, p = 0.03). Adjustment for Ni but not NO2 resulted in the change of the PM2.5
effect estimate from positive and nonsignificant to negative and significant. Ni was strongly and
positively associated with wheeze and was more positively correlated with PM2.5 in cold/flu
season (r = 0.41) than in non-cold/flu season (r = -0.21), which may explain the apparent
protective effects of PM2.5 on wheeze. An IQR (0.004 ppb) increase in NO2 was significantly
associated with 26% increased probability of wheeze (p = 0.002) (Table 2). However, in the
multipollutant model that included EC and Ni, the effect estimate decreased to 0.13 per IQR
increase in NO2, and the association became nonsignificant (p = 0.27). The association between
NO2 and cough was not significant in the single pollutant model, but there was suggestion of an
association in a model that adjusted for EC and Ni (β = 0.14 per IQR increase in NO2, p = 0.08).
Effects of cold/flu season
To examine differences in effects by season, multipollutant analyses were performed after
stratifying by cold/flu season. Despite the 50% smaller sample sizes in these models, significant
relationships were observed between several pollutants and symptoms, mostly during the cold/flu
season (Table 3). For example, Ni and V remained significantly associated with wheeze in the
model that included only observations during cold/flu season, and the effect estimates were
larger than those estimated in the all-season models. EC was significantly associated with cough
in analyses restricted to observations during cold/flu season, and the association between NO2
Page 11 of 47
Page 12
11
and cough was borderline significant (p = 0.05). In analyses restricted to non-cold/flu season
(April 1-August 31), NO2 was significantly associated with wheeze (Table 3). Significant
negative associations also were observed during cold/flu season between Zn and cough and
PM2.5 and wheeze in multipollutant models. For PM2.5 and wheeze, adjustment for Ni resulted in
a negative effect estimate for PM2.5. For Zn and cough, adjustment for Fe produced a negative
effect estimate for Zn. Fe was significantly associated with cough in the multipollutant model
and also was more highly correlated with Zn during cold/flu season (r = 0.29) than during non-
cold/flu season (r = 0.52), which may provide explanation for the apparent protective effect of
Zn on cough in cold/flu season.
Sensitivity and exploratory analyses
To examine whether the observed findings were driven by extreme pollutant
measurements, relationships with symptoms were examined after excluding the highest 5%
pollutant concentrations. Extreme measurements were clustered by season and year. Timing of
peak measurements also varied among pollutants. For example, peak concentrations of both Ni
and V were measured between December 2000 and February 2001. High concentrations of EC
and Zn were measured between January and February 2006. Peak NO2 concentrations were
predominately clustered between February and May 2000. After excluding peak Ni
measurements, associations with wheeze remained significant in both single and multipollutant
models (Table E6 of online supplement). After excluding extreme V concentrations, associations
with wheeze were no longer significant in the multipollutant model. There was suggestion of
negative association between Zn and cough, and the association between PM2.5 and wheeze was
significantly negative only in multipollutant models. Similar to the full model, NO2 was
Page 12 of 47
Page 13
12
significantly associated with wheeze after excluding the highest 5% concentrations, however, the
association was not robust to the inclusion of the copollutants Ni and EC.
Exploratory cross-sectional analyses were conducted to examine the effects of prenatal
pollutant exposures, implicated in asthma pathogenesis [24, 27], on wheeze and cough at later
ages (data not shown). Ambient metals, EC, and PM2.5 concentrations from the 3 months prior to
birth were not associated with symptoms at age 9 months, and exposures between 3 and 6
months before birth were not associated with symptoms at age 12 months. Models that included
prenatal ETS as a covariate did not produce results that differed from models that included
postnatal ETS as a covariate (Table E7 of online supplement).
DISCUSSION
Our objective was to characterize the differential relationships between exposure to
ambient PM2.5 and its specific components, including metals and EC, and respiratory symptoms
in a cohort of very young children living in high-density NYC neighborhoods. We found that Ni
and V were associated significantly with wheeze in this cohort during the first 24 months of life,
after adjusting for sex, ethnicity, ETS, seasonal trends, and copollutants. Additionally, EC was
associated significantly with cough only during cold/flu season. This study provides new
evidence using an individual-level longitudinal study design that specific components of PM2.5
related to residual oil combustion and/or traffic are associated with adverse respiratory health
effects in children during the first two years of life. PM2.5, a heterogeneous mix of particles of
various chemical constituents from multiple sources, was not associated significantly with
wheeze or cough. This latter result suggests that mass-based standards for total PM2.5 may not
adequately protect against adverse health effects from exposures to the individual toxic metals
Page 13 of 47
Page 14
13
and EC components which are believed to represent approximately only 4% and 3% of the mass,
respectively [28].
Children participating in this study reside in NYC communities with very high pediatric
asthma prevalence and hospitalization rates [20] and that contain major trucking thoroughfares,
bus depots, and waste transfer stations that emit multiple air pollutants [29]. Traffic emissions,
particularly from diesel vehicles, are a dominant source of EC in the atmosphere. Traffic also
contributes to ambient metals from direct tailpipe emissions, brake and tire abrasion, and
resuspension of roadway dust [9, 10]. Residual oil fuel, which is the major source of ambient Ni
and V in NYC, continues to be used for space heating in older residential and commercial
buildings that are common in the study area [11]. Concentrations of EC, Ni, V, and Zn are higher
at the Bronx monitoring sites in our study area, compared to an average of 87 US counties [7],
and Ni concentrations at the Bronx sites are higher than those at other NYC monitoring sites
[11]. Hence, these results suggest that metals and EC from heating oil combustion and diesel
traffic and may be important ambient pollutants that contribute to asthma-related symptoms in
these communities.
The largest effect size and most consistent associations were observed between Ni and
wheeze. The effects of Ni on wheeze were robust to the inclusion of indicators of traffic
emissions such as EC and NO2. Although NO2 was significantly associated with wheeze in a
single pollutant model, associations became nonsignificant when Ni and EC were included in the
model. Therefore, residual oil combustion, an important non-traffic source of ambient Ni in the
study area, could be responsible for many asthma-related symptoms among young residents of
these communities. Recent studies support a role for Ni in increasing risk of asthma -related
outcomes. In a national scale study, county- and season-specific PM2.5 risk estimates for
Page 14 of 47
Page 15
14
respiratory and cardiovascular admissions were higher in counties and seasons with a PM2.5-Ni
fraction in the 75th
compared to the 25th
percentile [5]. Additionally, in reanalyses of the National
Mortality and Morbidity Air Pollution Study data, PM10-mortality risk estimates were higher for
communities with higher long-term averages of ambient Ni and V [4, 6], and this effect
modification was driven by strong associations observed in NYC [4]. Although these previous
studies included adult populations, their findings support even further the premise that Ni and V
may be important airborne pollutants that contribute to adverse respiratory health effects in
NYC.
In analyses stratified by cold/flu season, larger effect estimates for Ni and V on wheeze
and significant effects of EC on cough were observed in models containing observations from
only cold/flu season (September 1-March 31) (Table 3). Concentrations of metals and EC are
higher in the winter due to emissions from heating sources such as roof-top furnaces and due to
lower mixing height in the atmosphere, resulting in diminished dispersion of emitted pollutants
[7, 11]. Respiratory symptoms and asthma exacerbations show peaks in the fall and winter, as
well, and are related to viral infections [30]. In models that excluded the highest 5% of pollutant
concentrations, V and EC were no longer associated with wheeze and cough, respectively,
suggesting that extreme concentrations occurring primarily during winter may be highly
influential in terms of their effects on respiratory symptoms. Nickel remained significantly
associated with wheeze after removing the highest 5% measurements. In a study of human
airway cells, coexposure to human rhinovirus and nitrogen dioxide (NO2) or ozone (O3)
stimulated greater production of the pro-inflammatory cytokine IL-8 than did exposure to
rhinovirus or either pollutant alone [31]. Therefore, significant associations between Ni and V
and wheeze and EC and cough during cold/flu season may occur as a consequence of synergistic
Page 15 of 47
Page 16
15
effects on airway inflammation induced by exposures to viral infections and airborne Ni. NO2
was significantly associated with wheeze during the non-cold/flu season, after adjusting for Ni
and EC. NO2 concentrations did not display strong seasonal variation (Table E5 of online
supplement). Hence, the effects of NO2 on wheeze may have been masked by the larger effects
of Ni and/or viral infections exposures during the cold/flu season and became apparent in the
absence of exposures to high Ni concentrations and/or viral infections during non-cold/flu
season. Unexpected significant negative associations were observed between PM2.5 and wheeze
and Zn and cough that were driven by effects in cold/flu season. Because these apparent
protective effects were observed only in multipollutant models, they are likely explained by
inclusion of copollutants such as Ni and Fe that were found to have strong positive effects on
symptoms and by higher correlations between pollutants observed in cold/flu season.
Ambient levels of Ni, V, or EC may be serving as surrogates of pollutant mixtures or
other individual components from residual oil combustion and/or traffic that are causally
associated with respiratory symptoms. Many PM2.5 species evaluated in our models displayed
high correlation between sites (Table E1 and E4 of online supplement) and also were highly
correlated with other trace elements within sites (Tables E2 and E3 of online supplement),
making it difficult to distinguish among the effects of pollutants from common sources. For
example, due to high correlation among Ni and V, we did not include them in the same model to
evaluate as potential confounders. Copper (Cu) and iron (Fe) were moderately correlated with
Ni, V, and Zn and have been associated with increased mortality [32, 33] and stimulation of
airway inflammation [34] in the literature. In the current study, however, neither Cu nor Fe was
associated with wheeze or cough in single pollutant models (data not shown), and neither altered
the observed associations between Ni or V and wheeze. EC and NO2, both indicators of traffic
Page 16 of 47
Page 17
16
tailpipe emissions, were moderately correlated at NYBG but not significantly correlated at IS52.
EC was significantly associated with cough during the cold/flu season after adjusting for NO2
and Ni, and NO2 was significantly associated with wheeze during the non-cold/flu season after
adjusting for Ni and EC. Thus, although traffic emissions appear to contribute to respiratory
morbidity, it is difficult to distinguish between the effects of particulate and gaseous pollutants.
We acknowledge several limitations to this study. The study population was comprised of
only Dominican and African American children living in Northern Manhattan and the South
Bronx, and populations that differ in ethnic composition also may differ with respect to the
relative strength of association between particular outcomes and exposures. Furthermore, this
cohort may differ from the overall population in several other factors including asthma
prevalence, the distribution of traffic- and oil combustion-related pollutants, genetic
polymorphisms, and cultural differences that may influence symptom reporting and behaviors
relevant to dose of environmental exposures.
To characterize associations between EC, metals, and PM2.5 and respiratory symptoms,
exposures were assigned using data from two monitoring sites located in the study area: IS52 and
NYBG. Previously, personal exposures of NYC adolescents to PM2.5-associated Ni, Zn, and BC
were observed to display the greatest spatial variability, whereas exposures to V showed the least
spatial variability [35]. In the current study, significant differences were observed in mean
concentrations of Ni and EC between sites, and EC was weakly correlated between sites. Small-
scale differences in EC, occurring mostly in winter and spring periods, have been attributed to
local stack emissions of EC that cause random spikes in ambient concentrations (Dr. Oliver
Rattigan, NYSDEC, personal communication). Using data from existing monitoring stations to
represent individual exposures to pollutants with high spatial variability may not represent true
Page 17 of 47
Page 18
17
exposure as accurately as personal or residential measurements. To incorporate spatial
heterogeneity in ambient concentrations of PM2.5 components, exposure estimates were assigned
to subjects using inverse-distance weighted pollution measurements from the two stationary
monitoring sites. Furthermore, in these longitudinal analyses, relationships between pollutants
and symptoms were examined within subjects over time, and previous studies have shown that
central site measurements of PM2.5, EC, and metals are correlated temporally with personal
exposures within subjects. Thus, in the current analyses, central site measurements may provide
reasonable estimates of exposure [36]. Given the significant spatial differences observed in
ambient Ni and EC concentrations, exposure misclassification may be higher for these pollutants.
However, such measurement error is likely to be random and would tend to underestimate the
effects of Ni and EC on respiratory symptoms.
Symptom data covered a three month period and were compared with concurrent 3-
month averages of metals, EC, and PM2.5. Much of the previous evidence regarding effects of
particulate matter or its components pertains to acute (daily) exposures to metals or EC in time-
series analyses or to long-term exposures (yearly or multi-year) in cohort studies. For example,
deficits in lung function growth have been observed in children in association with community-
level pollution exposures between ages 10 and 18 years [16]. In a recent population-level time
series study of children ages 0-17 living in Baltimore, Maryland, high ambient Zn levels,
measured at a central monitoring site, were associated with increases in asthma emergency
department visits and hospitalizations on the following day [17]. From its onset, CCCEH chose
the 3-month time interval as the shortest duration in which structured, high quality questionnaires
could be administered to hundreds of women as part of the parent cohort design. The intent was
to capture recent chronic (i.e. subacute) exposures and related symptoms. For the purpose of this
Page 18 of 47
Page 19
18
specific study, our hypothesis testing was based on determining the effects of subacute
environmental exposures on respiratory symptoms, in part to ascertain a signal that goes beyond
those related to hourly or daily changes in activities. Given the longitudinal design of our study,
collecting data about symptoms on shorter lags, for example, the previous 7 days, may have
improved our characterization of the associations between metals and EC exposures and
respiratory symptoms. However, our findings provide evidence that subacute (i.e. 3 month)
exposures of very young urban children may be associated with increased probability of
respiratory symptoms in addition to acute and long-term exposures to metals and BC or EC that
are documented in the literature.
In conclusion, the associations between increases in ambient concentrations of Ni, V, and
EC, but not total PM2.5, and increased probability of respiratory symptoms, after adjusting for
copollutants, suggest that specific PM2.5 components related to residual oil combustion and/or
traffic may be health-relevant PM2.5 fractions associated with increased respiratory morbidity in
children through age 24 months. While it has been previously demonstrated that exposures to
traffic-related air pollution early in life may be important risk factors for later development of
asthma [24, 37], the current results improve our understanding of the potential deleterious
consequences of exposure to specific metals for children in inner cities. Given that metal and EC
components of ambient PM2.5 are only indirectly regulated as part of the PM2.5 mass-based
standard, improved regulatory action directed at specific sources such as traffic and residential
boilers or at ambient concentrations of individual components such as EC and metals, may be
needed to help protect young children living in urban areas.
Page 19 of 47
Page 20
19
Acknowledgements: The authors acknowledge the contribution of Oliver Rattigan, David
Wheeler, Paul Sierzenga, Dirk Felton, and Patrick Lavin from the Bureau of Air Quality
Surveillance of the New York State Department of Environmental Conservation. The authors
thank Dr. Shuang Wang for her statistical consultation and also thank the women and children
participating in the study.
Page 20 of 47
Page 21
20
References
1. Mar TF, Larson TV, Stier RA, Claiborn C, Koenig JQ, An analysis of the association
between respiratory symptoms in subjects with asthma and daily air pollution in Spokane,
Washington. Inhal Toxicol 2004;16:809 - 815.
2. Delfino RJ, Staimer N, Tjoa T, Gillen D, Kleinman MT, Sioutas C, Cooper D, Personal
and ambient air pollution exposures and lung function decrements in children with
asthma. Environ Health Perspect 2008;116:550-558.
3. Halonen JI, Lanki T, Yli-Tuomi T, Kulmala M, Tiittanen P, Pekkanen J. Urban air
pollution, and asthma and COPD hospital emergency room visits. Thorax 2008;63:635-
641.
4. Dominici F, Peng RD, Ebisu K, Zeger SL, Samet JM, Bell ML. Does the effect of PM10
on mortality depend on PM nickel and vanadium content? A reanalysis of the NMMAPS
data. Environ Health Perspect 2007;115:1701-1703.
5. Bell ML, Ebisu K, Peng RD, Samet JM, Dominici F. Hospital admissions and chemical
composition of fine particle air pollution. Am J Respir Crit Care Med 2009;179:1115-
1120.
6. Lippmann M, Ito K, Hwang JS, Maciejczyk P, Chen LC. Cardiovascular effects of nickel
in ambient air. Environ Health Perspect 2006;114:1662-1669.
7. Bell ML, Dominici F, Ebisu K, Zeger SL, Samet JM, Spatial and temporal variation in
PM(2.5) chemical composition in the United States for health effects studies. Environ
Health Perspect 2007;115:989-995.
8. Schauer JJ, Evaluation of elemental carbon as a marker for diesel particulate matter. J
Expo Anal Environ Epidemiol 2003;13:443-453.
Page 21 of 47
Page 22
21
9. Lough GC, Schauer JJ, Park JS, Shafer MM, Deminter JT, Weinstein JP. Emissions of
metals associated with motor vehicle roadways. Environ Sci Technol 2005;9:826-836.
10. Salma I and Maenhaut W, Changes in elemental composition and mass of atmospheric
aerosol pollution between 1996 and 2002 in a Central European city. Environ Pollut
2006;143:479-488.
11. Peltier RE, Hsu S-I, Lall R, Lippmann M, Residual oil combustion: a major source of
airborne nickel in New York City. J Expos Sci Environ Epidemiol 2008 Oct 8. [Epub ahead
of print].
12. McConnell R, Berhane K, Yao L, Jerrett M, Lurmann F, Gilliland F, Künzli N,
Gauderman J, Avol E, Thomas D, Peters J. Traffic, susceptibility, and childhood asthma.
Environ Health Perspect 2006;114:766-772.
13. Kim JJ, Smorodinsky S, Lipsett M, Singer BC, Hodgson AT, Ostro B. Traffic-related air
pollution near busy roads: The East Bay children's respiratory health study. Am J Respir
Crit Care Med 2004;170:520-526.
14. Ryan PH, Lemasters GK, Biswas P, Levin L, Lindsey M, Bernstein DI, Lockey J,
Villareal M, K. KHG, Grinshpun SA. A comparison of proximity and land use regression
traffic exposure models and wheezing in infants. Environ Health Perspect 2007;115:278-
284.
15. Morgenstern V, Zutavern A, Cyrys J, Brockow I, Koletzko S, Kramer U, Behrendt H,
Herbarth O, von Berg A, Bauer CP, Wichmann HE, Heinrich J, for the Gini Study Group
and the LISA Study Group. Atopic diseases, allergic sensitization, and exposure to
traffic-related air pollution in children. Am J Respir Crit Care Med 2008;177:1331-1337.
16. Gauderman WJ, Avol E, Gilliland F, Vora H, Thomas D, Berhane K, McConnell R,
Kuenzli N, Lurmann F, Rappaport E, Margolis H, Bates D, Peters J. The effect of air
Page 22 of 47
Page 23
22
pollution on lung development from 10 to 18 years of age. N Engl J Med 2004;351:
1057-1067.
17. Hirshon JM, Shardell M, Alles S, Powell JL, Squibb K, Ondov J, Blaisdell CJ. Elevated
ambient air zinc increases pediatric asthma morbidity. Environ Health Perspect 2008;
116:826-831.
18. Schaumann F, Borm PJA, Herbrich A, Knoch J, Pitz M, Schins RPF, Luettig B, Hohlfeld
JM, Heinrich J, Krug N. Metal-rich ambient particles (Particulate Matter2.5) cause
airway inflammation in healthy subjects. Am J Respir Crit Care Med 2004;170:898-903.
19. Salnikow K, Li X, Lippmann M. Effect of nickel and iron co-exposure on human lung
cells. Toxicol Appl Pharmacol 2004;196:258-265.
20. Statewide Planning and Research Cooperative System. 2007 New York State Asthma
surveillance summary report. New York State Department of Health. New York, NY.
21. Patel MM, Chillrud SN, Hoepner L, Reyes A, Garfinkel R, Whyatt RM, Kinney PL,
Perera FP, Miller RL. Effects of ambient elemental carbon and metals on wheeze and
cough during early childhood. Epidemiology 2008;19:S159.
22. Patel MM, Hoepner L, Garfinkel R, Chillrud SN, Reyes A, Perera FP, Miller RL,
Ambient metals and elemental carbon in fine particulate matter predict wheeze and cough
in very young urban children. J Allergy Clin Immunol 2009;123:S172.
23. Perera FP, Illman SM, Kinney PL, Whyatt RM, Kelvin EA, Shepard P, Evans D,
Fullilove M, Ford J, Miller RL, Meyer IH, Rauh VA. The challenge of preventing
environmentally-related disease in young children: community-based research in New
York City. Environ Health Perspect 2002;110:197-204.
Page 23 of 47
Page 24
23
24. Miller RL, Garfinkel R, Horton M, Camann D, Perera FP, Whyatt RM, Kinney PL.
Polycyclic aromatic hydrocarbons, environmental tobacco smoke, and respiratory
symptoms in an inner-city birth cohort. Chest 2004;126:1071-1078.
25. Donohue KM, Al-alem U, Perzanowski MS, Chew GL, Johnson A, Divjan A, Kelvin EA,
Hoepner LA, Perera FP, Miller RL, Anti-cockroach and anti-mouse IgE are associated
with early wheeze and atopy in an inner-city birth cohort. Journal of Allergy and Clinical
Immunology, 2008. 122(5): p. 914-920.
26. Narvaez RF, Hoepner L, Chillrud SN, Yan B, Garfinkel R, Whyatt RM, Camann D,
Perera FP, Kinney PL, Miller RL. Spatial and temporal trends of polycyclic aromatic
hydrocarbons and other traffic-related airborne pollutants in New York City. Environ Sci
Technol 2008;42:7330-7335.
27. Mortimer K, Neugebauer R, Lurmann F, Alcorn S, Balmes J, Tager I. Air pollution and
pulmonary function in asthmatic children: Effects of prenatal and lifetime exposures.
Epidemiology 2008;19:550-557.
28. Wilson R, Spengler J. Emissions, dispersion, and concentration of particles. In: Wilson R,
Spengler J, editors. Particles in Our Air: Concentrations and Health Effects, Cambridge,
MA: Harvard University Press; 1996. p. 41-62.
29. Tonne CC, Whyatt RM, Camann DE, Perera FP, Kinney PL. Predictors of personal
polycyclic aromatic hydrocarbon exposures among pregnant minority women in New
York City. Environ Health Perspect 2004;112:754-759.
30. Lin RY, Pitt TJ, Lou WWY, Qilong Y. Asthma hospitalization patterns in young children
relating to admission age, infection presence, sex, and race. Ann Allergy Asthma Immunol
2007;98:139-145.
Page 24 of 47
Page 25
24
31. Spannhake WE, Reddy SPM, Jacoby DB, Yu XY, Saatian B, Tan J. Synergism between
rhinovirus infection and oxidant pollutant exposure enhances airway epithelial cell
cytokine production. Environ Health Perspect 2002;110:665-670.
32. Burnett RT, Brook J, Dann T, Delocla C, Philips O, Cakmak S, Vincent R, Goldberg MS,
Krewski D. Association between particulate- and gas-phase components of urban air
pollution and daily mortality in eight canadian cities. Inhal Toxicol 2000;12:15-39.
33. Ostro B, Feng W-Y, Broadwin R, Green S, Lipsett M. The effects of components of fine
particulate air pollution on mortality in California: Results from CALFINE. Environ
Health Perspect 2007;115:13-19.
34. Gavett SH, Haykal-Coates N, Copeland LB, Heinrich J, Gilmour IM. Metal composition
of ambient PM2.5 influences severity of allergic airways disease in mice. Environ Health
Perspect 2003;111:1471-1477.
35. Kinney PL, Chillrud SN, Sax S, Ross JM, Pederson DC, Johnson D, Aggarwal M,
Spengler JD. Toxic exposure assessment: A Columbia-Harvard (TEACH) study (The
New York City report). Mickey Leland National Urban Air Toxics Research Center
Research Report No. 3. 2005.
36. Janssen, NAH, Lanki T, Hoek G, Vallius M, de Hartog JJ, Van Grieken R, Pekkanen J,
Brunekreef B. Associations between ambient, personal, and indoor exposure to fine
particulate matter constituents in Dutch and Finnish panels of cardiovascular patients.
Occup Environ Med 2005;62:868-877.
37. Salam, MT, Gauderman WJ, McConnell R, Lin PC, Gilliland FD. Transforming growth
factor- 1 C-509T polymorphism, oxidant stress, and early-onset childhood asthma. Am J
Respir Crit Care Med 2007;176:1192-1199.
Page 25 of 47
Page 26
25
Figure Legends
Figure 1. Seasonal trends in wheeze and cough. Prevalence calculated as proportion of subjects
reporting presence of wheeze or cough each season. Wheeze prevalence was higher in the fall
(September 21-December 20) and winter (December 21-March 20) (p<0.0001), compared with
spring (March 21-June 20). Similar proportions of subjects reported wheeze in the spring and
summer (June 21-September 20). Prevalence of cough was higher in fall, winter, and summer
(p<0.0001 for all 3 seasons), compared with spring. Su = summer, F = fall, W = winter, Sp =
spring.
Page 26 of 47
Page 27
26
Tables Table 1. Selected cohort characteristics
Cohort at
age 24
months
(n = 687)
Children with
follow- up
data
(n = 653)
Male 48% 49% Child’s sex
Female 52% 51%
Dominican 65% 64% Mother’s
Ethnicity African American 35% 36%
Mother with at least 12th grade education 64% 64%
Child age 6 months 13% 13%
Child age 12 months 12% 12%
Child age 24 months 11% 11%
Maternal history
of smoking
Any time child age 0-24 months 11% 12%
Child age 6 months 25% 25%
Child age 12 months 23% 23%
Child age 24 months 19% 19%
Smoker in
household
Any time child age 0-24 months 25% 27%
Maternal history of asthma 23% 22%
Mother receiving Medicaid at enrollment 91% 90%
Child with asthma/possible asthma* 30% 30%
*Doctor says child has or might have asthma at time of 24 month questionnaire
Page 27 of 47
Page 28
27
Table 2. Effect estimates of presence of wheeze or cough associated with 3-month average
ambient pollutant concentrations (β-coefficient*, p)
Pollutant
(IQR)
Symptom
Single pollutant model Multipollutant model
nll 636 (3085) 636 (3075)
Wheeze 0.28 (0.0006) 0.25 (0.0006)
Ni†
(0.014 µg/m3)
Cough -0.05 (0.51) -0.14 (0.10)
n 636 (3085) 636 (3075)
Wheeze 0.10 (0.13) 0.14 (0.08)
V†
(0.003 µg/m3)
Cough 0.04 (0.49) -0.04 (0.59)
n 636 (3085) 636 (3075)
Wheeze 0.04 (0.66) 0.01 (0.94)
Zn†
(0.018 µg/m3)
Cough 0.03 (0.75) -0.17 (0.15)
n 636 (3075) 636 (3075)
Wheeze 0.04 (0.43) 0.02 (0.66)
EC‡
(0.29 µg/m3)
Cough 0.04 (0.34) 0.05 (0.25)
n 638 (3131) 636 (3075)
Wheeze -0.0009(0.89) -0.13 (0.03)
PM2.5‡
(2.1 µg/m3)
Cough -0.03 (0.62) -0.06 (0.36)
n 650 (3553) 636 (3075)
Wheeze 0.26 (0.002) 0.13 (0.27)
NO2§
(0.004 ppm)
Cough 0.05 (0.44) 0.14 (0.08)
Page 28 of 47
Page 29
28
* Beta coefficient estimates change in probability of outcome per interquartile range (IQR)
increase in pollutant concentration adjusted for sex, ethnicity, smoking by mother or other
smoker in the home, calendar week (df = 4.72).
†Copollutants include EC, NO2, copper, and iron.
‡Copollutants include NO2 and Ni.
§Copollutants include EC and Ni.
llTotal number of subjects included in model (number of subjects*number of observations per
subject).
Values in boldface are statistically significant (p<0.05).
Page 29 of 47
Page 30
29
Table 3. Effects estimates of presence of wheeze or cough associated with 3-month average
ambient pollutant concentrations stratified by season (β-coefficient*, p)
Pollutant (IQR) Symptom Cold/Flu Season†
n††
= 580 (1661)
Non-Cold/Flu Season‡
n = 606 (1414)
IQR 0.012 µg/m3 0.009 µg/m
3
Wheeze 0.31 (0.003) 0.46 (0.07)
Ni
Cough -0.14 (0.10) -0.20 (0.30)
IQR 0.0033 µg/m3 0.0029 µg/m
3
Wheeze 0.31 (0.0003) 0.17 (0.39)
V
Cough -0.15 (0.13) 0.12 (0.46)
IQR 0.011 µg/m3 0.007 µg/m
3
Wheeze -0.13 (0.46) 0.34 (0.25)
Zn
Cough -0.31 (0.04) 0.20 (0.44)
IQR 0.319 µg/m3 0.232 µg/m
3
Wheeze 0.07 (0.32) -0.02 (0.80)
EC
Cough 0.11 (0.04) -0.001 (0.99)
IQR 2.1 µg/m3 1.9 µg/m
3
Wheeze -0.30 (0.008) 0.02 (0.85)
PM2.5
Cough -0.06 (0.36) -0.13 (0.18)
IQR 0.0040 ppm 0.0038 ppm
Wheeze -0.08 (0.47) 0.38 (0.02)
NO2
Cough 0.22 (0.05) 0.12 (0.33)
Page 30 of 47
Page 31
30
* Beta coefficient estimates change in probability of outcome per increase in interquartile range
(IQR) of pollutant concentration, adjusted for sex, ethnicity, smoking by mother or other smoker
in the home, and calendar week, and copollutants as described in Table 2.
†Includes observations between September 1 and March 31.
‡Includes observations between April 1 and August 31.
††Total number of observations included in model (number of subjects*number of observations
per subject).
Values in boldface are statistically significant (p<0.05).
Page 31 of 47
Page 32
Figures Figure 1.
0
10
20
30
40
50
60
70
Su F W Sp Su F W Sp Su F W Sp Su F W Sp Su F W Sp Su F W Sp Su F W Sp Su F W Sp Su F W Sp Su
Season
Pro
po
rtio
n o
f su
bje
cts
(%)
Wheeze
Cough
2007 1998
Page 32 of 47
Page 33
0
Ambient Metals, Elemental Carbon, and Wheeze and Cough in New York City Children
through Age 24 Months
Molini M. Patel, Lori Hoepner, Robin Garfinkel, Steven Chillrud, Andria Reyes, James W.
Quinn, Frederica Perera, Rachel L. Miller
ONLINE DATA SUPPLEMENT
Page 33 of 47
Page 34
1
METHODS
Study cohort recruitment
Pregnant women who identified themselves as Dominican or African-American and who
were residents of Washington Heights, Central Harlem, or the South Bronx were screened and
recruited from prenatal clinics associated with New York Presbyterian
Medical Center or Harlem
Hospital between 1998 and 2006 as part of the Columbia Center for Children’s Environmental
Health (CCCEH) cohort. Eligibility was limited to nonsmoking women aged 18 to 35 years who
had obtained their first prenatal visit by the twentieth week of pregnancy. Additional exclusion
criteria included pre-existing conditions such as diabetes, hypertension, and known HIV as well
as self-reported illegal drug use. Of the 2835 women screened, 1917 were eligible, and 841
expressed interest in the study and completed prenatal questionnaires. Seven hundred twenty five
women were followed throughout pregnancy and provided maternal and/or cord blood sample at
delivery and thus, were considered fully enrolled. Informed consent was obtained in accordance
with the Columbia University Institutional Review Board.
Questionnaire data
Data on subject characteristics, residence, environmental tobacco smoke (ETS) exposure
and respiratory symptoms were collected by questionnaires administered to the mothers by
telephone or in person prenatally and every 3 months postnatally, between 3 and 24 months of
age for a maximum of 8 follow-up questionnaires. Data were collected only about the previous
3-month period, regardless of the actual time elapsed since the last follow-up questionnaire.
Thirty four subjects (5%) were lost to follow-up between birth and age 24 months and were
Page 34 of 47
Page 35
2
excluded from analyses. There were no significant differences in subject characteristics between
study participants and those lost to follow-up.
Stationary site monitoring for metals, elemental carbon, and PM2.5
Twenty-four hour average ambient concentrations of PM2.5 and concentrations of EC, Ni,
V, and Zn contained within the PM2.5 fraction of particles were measured every third day by the
New York State Department of Environmental Conservation (NYSDEC) between 1999 and 2007
as part of the US Environmental Protection Agency’s Speciation Trends Network (STN).
Publicly available datasets were downloaded (http://www.dec.ny.gov/) from the two sites
in the Bronx which were located in the study area: New York Botanical Gardens (NYBG) and
Intermediate School 52 (IS52). At NYBG, data on PM2.5 were available between January 1999
and December 2005. EC and metals data were available between April 2000 and December
2005. At IS52, PM2.5 data were available between January 1999 and May 2007, and EC and
metals data were available between December 2000 and May 2007. Because daily data were not
available for both sites, and data were not always available on the same day between sites,
pollutant measurements were aggregated by week and site, as described [25]. Analysis was
conducted by the Research Triangle Institute (Research Triangle Park, NC), and protocols were
standardized across the years queried. Flagged observations indicating deviations from protocol
(most commonly, shipping temperature of samples outside of specifications), equipment
malfunction, and atypical incidents in the local environment (e.g., sandblasting, unusual traffic
congestion) were excluded from calculations of 1-week average concentrations.
Statistical analysis
Page 35 of 47
Page 36
3
Distributions of continuous variables were examined, and data were log transformed
when necessary to reduce variance and fulfill the distribution requirements of the statistical test
being used. To assess potential selection bias with respect to subjects included in these analyses,
Student’s t-test (continuous variables) and Χ2 (discrete variables) were used to compare
demographic characteristics of the fully enrolled CCCEH cohort at age 24 months with those of
the subgroup that had completed follow-up questionnaires.
Data from NYSDEC’s web sites were integrated into CCCEH's database, which is
maintained in Scientific Information Retrieval 2002. Initially, descriptive statistics were
compared using the complete DEC datasets versus the verified, non-flagged DEC dataset. Some
significant differences were identified, therefore, analyses continued with only verified data
points. Because of non-normal distribution of metals, EC, and PM2.5 concentrations, data are
presented as geometric means (GM) and standard deviation (GSD). Descriptive statistics were
performed on 1-week average pollutant concentrations. Mean pollutant levels between sites were
compared using non-parametric Mann Whitney U-test. Correlations among pollutants within
sites and between the same pollutant measured at the 2 sites were examined using Spearman
correlations. For quality control, PM2.5 duplicate samples from the IS52 site were verified as
highly correlated (r = 0.810, p < 0.001). Seasonal differences in pollutant concentrations were
analyzed using Kruskal Wallis test. In descriptive summaries of symptom prevalence and
pollutant levels, season was defined by calendar year as follows: winter = December 21-March
20, spring = March 21-June 20, summer = June 21-September 20, and fall = September 21-
December 20.
Generalized linear models were used to examine whether the prevalence of wheeze or
cough varied by season. Prevalence was calculated as the proportion of subjects reporting
Page 36 of 47
Page 37
4
presence of symptoms each season. Associations between metals, EC, and PM2.5 and presence of
wheeze and cough were analyzed using generalized additive mixed effects models (GAMM)
using the mgcv library in R version 2.9.0 (R Foundation for Statistical Computing, Vienna,
Austria). A first order autoregressive correlation structure was specified to account for
correlation among the repeated observations collected over a 2-year period from each subject
since it is expected that correlations between observations diminish over time. Other covariates
included parametric terms for sex, ethnicity, postnatal ETS, and a nonparametric smoothed term
for calendar time using natural cubic splines (4.7 degrees of freedom per calendar year). Data on
ETS were available only on questionnaires administered prenatally, and at 6, 12, and 24 months
of age. In models, exposure to ETS was analyzed as a dichotomous variable and was defined as
smoking by mother or presence of smoker in the home at any time prenatally to 24 months of
age.
For each subject, 3-month moving average concentrations of Ni, V, Zn, EC, and PM2.5
were calculated for each symptom reporting period based on the follow-up questionnaire date
and the preceding 3 months. As a comparison to traffic-associated particles, nitrogen dioxide
(NO2) was evaluated as a gaseous indicator of traffic emissions. Single pollutant models were
constructed in which each pollutant was analyzed as a parametric continuous variable.
Exposures were assigned to subjects by calculating inverse-distance weighted concentrations
using pollutant measurements from IS52 and NYBG. Address data were collected only at
prenatal, 6, 12, and 24 month questionnaires, and addresses during interim time points were
assigned using data on moved since last questionnaire and previous addresses.
The robustness of results was evaluated using the following methods: models that
included gaseous and particulate copollutants related to traffic, models that excluded the highest
Page 37 of 47
Page 38
5
5% of pollutant concentrations, and analyses stratified by season. Season was defined as a
dichotomous variable: “cold/flu season” (September 1 to March 31) or “non-cold/flu season”
(April 1 to August 31). Separate models included observations only from cold/flu season and
only from non-cold/flu season.
Except for GAMM, statistical tests and modeling were performed using SAS 9.1.3 (Cary,
NC, release 2005), and results with p<0.05 were considered statistically significant.
Page 38 of 47
Page 39
6
TABLE E1. SUMMARY STATISTICS FOR 1-WEEK AVERAGE POLLUTANT
CONCENTRATIONS*, 1999-2007
Pollutant Site Geometric
mean (GSD) Min Median Max
Spearman’s
correlation
(r)
IS52 (n=210) 0.016 (0.015) 0.003 0.017 0.120 Ni
†
NYBG (n=190) 0.021 (0.014) 0.003 0.024 0.100
0.75‡
IS52 (n=210) 0.006 (0.005) 0.0003 0.006 0.027 V
NYBG (n=190) 0.006 (0.004) 0.0002 0.007 0.024 0.73
‡
IS52 (n=210) 0.032 (0.022) 0.002 0.032 0.192 Zn
NYBG (n=190) 0.031 (0.019) 0.002 0.032 0.115 0.73
‡
IS52 (n=205) 1.1 (0.7) 0.0002 1.1 4.0 EC
†
NYBG (n=189) 1.3 (0.6) 0.40 1.3 4.9 0.38
‡
IS52 (n=393) 13.0 (5.1) 3.4 13.4 37.5 PM2.5
NYBG (n=413) 12.3 (6.3) 3.2 12.3 38.4 0.77
‡
IS52 (n=360) 0.029 (0.006) 0.017 0.029 0.053 NO2
NYBG (n=416) 0.027 (0.005) 0.019 0.027 0.045 0.79
‡
* All concentrations are expressed as µg/m3, except for NO2, which is in parts per
million.
p<0.05 for difference in mean concentrations between sites, Mann-Whitney U-test.
‡p<0.0001 for Spearman’s correlation.
Page 39 of 47
Page 40
8
TABLE E2. SPEARMAN’S CORRELATION COEFFICIENTS FOR 3-MONTH MOVING AVERAGES OF POLLUTANTS
WITHIN THE IS52 SITE*
IS52
NO2
IS52
EC
IS52
PM2.5
IS52
Br
IS52
Ca
IS52
Cl
IS52
Cu
IS52
Fe
IS52
Mn
IS52
Ni
IS52
Pb
IS52
V
IS52 EC -0.32
IS52 PM2.5 0.38 -0.15
IS52 Br 0.22 0.32 -0.02
IS52 Ca -0.22 0.56 -0.40 0.52
IS52 Cl 0.32 0.25 -0.05 0.63 0.37
IS52 Cu -0.21 0.51 -0.09 0.43 0.74 0.22
IS52 Fe -0.11 0.56 0.15 0.53 0.42 0.23 0.52
IS52 Mn 0.32 0.61 -0.31 0.05 0.84 0.27 0.71 0.51
IS52 Ni 0.47 0.07 0.11 0.67 0.19 0.76 0.09 0.31 0.06
IS52 Pb 0.03 0.33 -0.13 0.64 0.49 0.49 0.47 0.43 0.43 0.59
IS52 V 0.27 0.36 0.09 0.81 0.57 0.67 0.56 0.58 0.41 0.76 0.64
IS52 Zn 0.14 0.37 -0.17 0.75 0.63 0.71 0.45 0.48 0.52 0.78 0.68 0.82
*p>0.05 for shaded correlation coefficients
Page 40 of 47
Page 41
9
TABLE E3. SPEARMAN’S CORRELATION COEFFICIENTS FOR 3-MONTH MOVING AVERAGES OF POLLUTANTS
WITHIN THE NYBG SITE*
NYBG
NO2
NYBG
EC
NYBG
PM2.5
NYBG
Br
NYBG
Ca
NYBG
Cl
NYBG
Cu
NYBG
Fe
NYBG
Mn
NYBG
Ni
NYBG
Pb
NYBG
V
NYBG EC 0.49
NYBG PM2.5 0.19 0.02
NYBG Br 0.35 0.47 0.24
NYBG Ca 0.29 0.52 -0.14 0.49
NYBG Cl -0.03 0.11 -0.26 0.24 0.17
NYBG Cu 0.03 0.32 0.16 0.47 0.39 0.06
NYBG Fe 0.04 0.71 0.14 0.41 0.58 -0.25 0.42
NYBG Mn -0.07 0.51 0.04 0.50 0.55 -0.05 0.65 0.69
NYBG Ni 0.59 0.69 -0.04 0.35 0.23 0.28 -0.06 -0.06 -0.03
NYBG Pb 0.58 0.53 0.16 0.33 0.11 0.03 0.07 0.24 0.04 0.62
NYBG V 0.30 0.68 0.02 0.69 0.44 0.40 0.32 0.25 0.32 0.62 0.48
NYBG Zn 0.33 0.82 -0.28 0.61 0.61 0.41 0.23 0.21 0.36 0.64 0.40 0.78
*p>0.05 for shaded correlation coefficients.
Page 41 of 47
Page 42
10
TABLE E4. SPEARMAN’S CORRELATION COEFFICIENTS FOR 3-MONTH MOVING AVERAGES OF METALS, EC,
AND PM2.5 BETWEEN THE IS52 and NYBG SITES*
IS52
EC
IS52
PM2.5
IS52
Br
IS52
Ca
IS52
Cl
IS52
Cu
IS52
Fe
IS52
Mn
IS52
Ni
IS52
Pb
IS52
V
IS52
Zn
IS52
NO2
NYBG EC 0.22 -0.09 0.67 0.44 0.76 0.43 0.30 0.40 0.80 0.64 0.69 0.73 0.14
NYBG PM2.5 0.19 0.77 0.12 -0.10 0.00 0.22 0.36 0.02 0.07 0.03 0.21 0.00 0.21
NYBG Br 0.47 0.07 0.71 0.46 0.47 0.54 0.56 0.58 0.46 0.54 0.55 0.55 -0.07
NYBG Ca 0.52 -0.25 0.57 0.90 0.31 0.63 0.50 0.71 0.32 0.47 0.55 0.66 -0.07
NYBG Cl 0.11 -0.23 0.17 0.14 0.66 0.03 0.02 0.13 0.41 0.03 0.21 0.40 0.20
NYBG Cu 0.32 0.16 0.18 0.25 -0.04 0.54 0.36 0.43 -0.07 0.20 0.15 0.10 -0.15
NYBG Fe 0.71 -0.06 0.24 0.55 -0.04 0.50 0.68 0.70 -0.04 0.28 0.28 0.28 -0.38
NYBG Mn 0.51 -0.11 0.18 0.49 -0.06 0.50 0.48 0.70 -0.13 0.22 0.14 0.20 -0.32
NYBG Ni 0.05 0.02 0.46 0.16 0.76 0.03 0.14 -0.04 0.84 0.64 0.59 0.53 0.28
NYBG Pb 0.24 0.12 0.33 0.12 0.45 0.14 0.38 0.12 0.64 0.75 0.49 0.47 0.09
NYBG V 0.24 -0.03 0.69 0.44 0.69 0.48 0.50 0.41 0.74 0.64 0.85 0.76 0.02
NYBG Zn 0.19 -0.33 0.64 0.57 0.72 0.39 0.30 0.45 0.75 0.61 0.67 0.81 0.04
NYBG NO2 0.09 0.29 0.56 0.13 0.46 0.15 0.33 0.06 0.72 0.61 0.58 0.52 0.56
*p>0.05 for shaded correlation coefficients
Page 42 of 47
Page 43
11
TABLE E5. SEASONAL VARIATIONS IN POLLUTANT CONCENTRATIONS*
Ni† V† Zn† EC† PM2.5† NO2†
IS52 NYBG IS52 NYBG IS52 NYBG IS52 NYBG IS52 NYBG IS52 NYBG
GM (GSD) 0.024
(0.014)
0.033
(0.011)
0.007
(0.006)
0.009
(0.004)
0.037
(0.020)
0.043
(0.017)
1.0
(0.76)
1.6
(0.68)
15.2
(5.2)
14.0
(5.4)
0.031
(0.005)
0.029
(0.004)
Min 0.003 0.010 0.001 0.002 0.002 0.021 0.28 0.70 8.6 5.9 0.021 0.022
Median 0.025 0.034 0.008 0.009 0.039 0.043 1.1 1.6 14.6 14.1 0.031 0.029
Winter
Max 0.075 0.067 0.026 0.020 0.111 0.115 3.7 4.9 37.5 34.1 0.046 0.045
GM (GSD) 0.011
(0.008)
0.014
(0.014)
0.004
(0.005)
0.003
(0.003)
0.025
(0.016)
0.019
(0.014)
0.96
(0.68)
0.96
(0.46)
11.7
(4.2)
10.7
(5.7)
0.029
(0.006)
0.027
(0.005)
Min 0.003 0.004 0.001 0.001 0.009 0.002 0.06 0.37 4.7 4.0 0.016 0.015
Median 0.011 0.015 0.005 0.004 0.023 0.021 0.98 0.92 11.6 10.7 0.030 0.027
Spring
Max 0.036 0.100 0.014 0.012 0.090 0.072 3.5 2.5 25.8 30.7 0.043 0.043
GM (GSD) 0.007
(0.005)
0.011
(0.003)
0.005
(0.003)
0.005
(0.002)
0.024
(0.009)
0.020
(0.007)
1.3
(0.57)
1.2
(0.21)
14.9
(5.3)
13.9
(7.1)
0.027
(0.006)
0.024
(0.005)
Min 0.003 0.006 0.001 0.002 0.010 0.011 0.52 0.93 5.8 4.1 0.017 0.015
Median 0.007 0.012 0.006 0.005 0.025 0.020 1.3 1.2 14.8 14.3 0.027 0.025
Summer
Max 0.021 0.015 0.013 0.010 0.044 0.035 2.1 1.6 30.9 38.4 0.053 0.034
GM (GSD) 0.016
(0.019)
0.023
(0.013)
0.005
(0.006)
0.006
(0.005)
0.037
(0.029)
0.038
(0.020)
1.2
(0.78)
1.3
(0.58)
11.8
(4.7)
10.9
(6.1)
0.028
(0.006)
0.027
(0.005)
Min 0.003 0.003 0.001 0.001 0.007 0.007 0.44 0.45 3.4 3.2 0.010 0.011
Median 0.017 0.025 0.005 0.006 0.039 0.041 1.1 1.3 9.4 10.6 0.029 0.027
Fall
Max 0.120 0.058 0.027 0.024 0.192 0.109 4.0 3.2 30.8 35.2 0.045 0.044
All concentrations are expressed as µg/m3, except for NO2, which is in parts per million.
† p<0.05, for differences in means across seasons, evaluated by Kruskal-Wallis Test, except for IS52 EC.
Page 43 of 47
Page 44
11
Table E6. Effect estimates of presence of wheeze or cough associated with 3-
month average ambient pollutant concentrations, excluding highest 5% pollutant
concentrations (β-coefficient*, p)
Pollutant (IQR) Symptom Single pollutant model Multipollutant model
nll 632 (2933) 632 (2923)
Wheeze 0.39 (<0.0001) 0.31 (0.0004)
Ni†
(0.014 µg/m3)
Cough 0.03 (0.74) -0.08 (0.51)
n 633 (2924) 632 (2914)
Wheeze 0.05 (0.52) 0.07 (0.49)
V†
(0.003 µg/m3)
Cough 0.07 (0.27) -0.03 (0.74)
n 636 (2919) 636 (2909)
Wheeze 0.12 (0.31) 0.06 (0.67)
Zn†
(0.018 µg/m3)
Cough 0.008 (0.94) -0.25 (0.08)
n 636 (2914) 636 (2910)
Wheeze 0.10 (0.10) 0.09 (0.15)
EC‡
(0.27 µg/m3)
Cough 0.05 (0.28) 0.06 (0.21)
n 635 (2982) 633 (2934)
Wheeze -0.03 (0.65) -0.18 (0.003)
PM2.5‡
(1.9 µg/m3)
Cough -0.06 (0.25) -0.09 (0.14)
n 648 (3382) 633 (2924)
Wheeze 0.21 (0.02) 0.12 (0.23)
NO2§
(0.0038 ppm)
Cough 0.07 (0.31) 0.12 (0.16)
Page 44 of 47
Page 45
12
* Beta coefficient estimates change in probability of outcome per interquartile range
(IQR) increase in pollutant concentration adjusted for sex, ethnicity, smoking by mother
or other smoker in the home, calendar week.
†Copollutants include EC, NO2, copper, and iron.
‡Copollutants include NO2 and Ni.
§Copollutants include EC and Ni.
llTotal number of subjects included in model (number of subjects*number of
observations per subject).
Values in boldface are statistically significant (p<0.05).
Page 45 of 47
Page 46
13
TABLE E7. Associations of wheeze or cough with 3-month average ambient
pollutant concentrations, adjusting for postnatal versus prenatal ETS exposure
(β-coefficient*, p-value)
Pollutant
(IQR)
Symptom Controlling for postnatal
ETS exposure
nll = 636 (3075)
Controlling for prenatal
ETS exposure
nll = 630 (3047)
Wheeze 0.25 (0.0006) 0.25 (0.005) Ni
(0.014 µg/m3) Cough -0.14 (0.10) -0.15 (0.07)
Wheeze 0.14 (0.08) 0.16 (0.05) V
(0.003 µg/m3) Cough -0.04 (0.59) -0.05 (0.50)
Wheeze 0.01 (0.94) 0.03 (0.83) Zn
(0.018 µg/m3) Cough -0.17 (0.15) -0.18 (0.14)
Wheeze 0.02 (0.66) 0.02 (0.71) EC
(0.29 µg/m3) Cough 0.05 (0.25) 0.06 (0.19)
Wheeze -0.13 (0.03) -0.14 (0.02) PM2.5
(2.1 µg/m3) Cough -0.06 (0.36) -0.06 (0.35)
Wheeze 0.13 (0.27) 0.12 (0.19) NO2
(0.004 ppm) Cough 0.14 (0.08) 0.14 (0.08)
ETS = environmental tobacco smoke
* Beta coefficient estimates change in probability of outcome per interquartile range
(IQR) increase in pollutant concentration adjusted for sex, ethnicity, ETS (either prenatal
or postnatal, calendar week (df = 4.72), and copollutants as described in Table 2.
Page 46 of 47
Page 47
14
llTotal number of subjects included in model (number of subjects*number of
observations per subject).
Values in boldface are statistically significant (p<0.05).
Page 47 of 47