Estimating error in using residential outdoor PM2.5 concentrations as proxies for personal exposures: a meta-analysis.

Abstract

Studies examining the health effects of particulate matter <or= 2.5 microm in aerodynamic diameter (PM2.5) commonly use ambient PM2.5 concentrations measured at distal monitoring sites as proxies for personal exposure and assume spatial homogeneity of ambient PM2.5. An alternative proxy-the residential outdoor PM2.5 concentration measured adjacent to participant homes-has few advantages under this assumption.
We systematically reviewed the correlation between residential outdoor PM2.5 and personal PM2.5 (-rj) as a means of comparing the magnitude and sources of measurement error associated with their use as exposure surrogates.
We searched seven electronic reference databases for studies of the within-participant residential outdoor-personal PM2.5 correlation.
The search identified 567 candidate studies, nine of which were abstracted in duplicate, that were published between 1996 and 2008. They represented 329 nonsmoking participants 6-93 years of age in eight U.S. cities, among whom -rj was estimated (median, 0.53; range, 0.25-0.79) based on a median of seven residential outdoor-personal PM2.5 pairs per participant. We found modest evidence of publication bias (symmetric funnel plot; pBegg = 0.4; pEgger = 0.2); however, we identified evidence of heterogeneity (Cochran's Q-test p = 0.05). Of the 20 characteristics examined, earlier study midpoints, eastern longitudes, older mean age, higher outdoor temperatures, and lower personal-residential outdoor PM2.5 differences were associated with increased within-participant residential outdoor-personal PM2.5 correlations.
These findings were similar to those from a contemporaneous meta-analysis that examined ambient-personal PM2.5 correlations (rj = median, 0.54; range, 0.09-0.83). Collectively, the meta-analyses suggest that residential outdoor-personal and ambient-personal PM2.5 correlations merit greater consideration when evaluating the potential for bias in studies of PM2.5-mediated health effects.

Full text

Page 1

Environmental Health Perspectives • volume 118 | number 5 | May 2010
673
Research
Numerous epidemiologic and toxicologic
studies have linked particulate matter (PM)
air pollution with adverse health outcomes,
including mortality (Burnett et al. 2000;
Dominici et al. 2003; Katsouyanni et al.
2003), hospital admissions (Burnett et al.
1995; Linn et al. 2000; Oftedal et al. 2003),
and subclinical disease (Diez Roux et al. 2008;
Liao et al. 2009; Whitsel et al. 2009). A com-
mon feature of such studies is their reliance on
ambient PM concentrations measured at distal
monitoring sites as proxies for personal expo-
sure to PM of ambient origin. The reliance is
consistent with regulatory policies developed
under the Clean Air Act (1970) which have
been informed by studies of the correlation
between personal exposures to PM originating
outdoors and residential outdoor PM con-
centrations (Wallace 2000). However, ambi-
ent PM may not adequately represent total
PM exposure, because human activity pattern
surveys suggest that, on average, individuals
spend > 85% of their time inside (Klepeis et al.
2001), where they are exposed to numerous
sources of indoor PM, the physicochemical
properties and toxicities of which often differ
from those of ambient PM (Monn and Becker
1999; Wainman et al. 2000).
Available exposure studies, although small
in number, have suggested that several fac-
tors may influence the relationship between
ambient and total PM exposure, including
home ventilation, indoor PM sources, and
time–activity patterns (Rodes et al. 2001;
Sarnat et al. 2006; Williams et al. 2003b).
Because these factors are not well quantified
(Janssen et al. 1998), we previously reviewed
the literature that examined the within-
participant ambient-personal PM2.5 correla-
tion to determine the magnitude and sources
of measurement error inherent in using ambi-
ent PM2.5 as a surrogate for personal exposure
(Avery et al. 2010). We found that character-
istics of participants, studies, and the environ-
ments in which they were conducted affect
the accuracy of ambient PM2.5 as a proxy for
personal exposure and that the potential for
exposure misclassification may be substantial.
Although the residential outdoor PM2.5
concentration measured adjacent to participant
homes may be equally prone to misclassifica-
tion under the assumption of spatial homo-
geneity, use of this measure as an alternative
proxy for personal exposure may have some
advantages if this assumption is not uniformly
applicable. Studies of spatial variability in
ambient PM2.5 concentrations among 27 U.S.
urban areas (Pinto et al. 2004) suggest that
this may be the case. The fact that PM2.5 var-
ies at the microenvironmental level as a func-
tion of, for example, topography, proximity
to PM2.5 point sources, adjacency to major
traffic arterials, and prevailing winds [U.S.
Environmental Protection Agency (EPA)
2009; Zhu et al. 2002] also is consistent with
this suggestion. Nonetheless, how spatial vari-
ability and outdoor microenvironments affect
the use of ambient PM2.5 concentrations as
a proxy for personal PM2.5 exposure remains
unclear. Thus, we performed a meta-analysis
using the literature that examined the within-
participant residential outdoor-personal PM2.5
correlation and contrasted these findings with
those from the review of the within- participant
ambient-personal PM2.5 correlation (Avery
et al. 2010). Findings from the two meta-
analyses will facilitate the quantification of bias
that resulted from the use of surrogates for
personal PM2.5 exposure in studies that relied
on outdoor PM2.5 measurements.
Methods
Systematic review strategy. We devised
a search strategy to identify studies of the
Address correspondence to C. Avery, Department of
Epidemiology, University of North Carolina–Chapel
Hill, Bank of America Center, 137 E. Franklin St.,
Suite 306, Chapel Hill, NC 27514 USA. Telephone:
(919) 966-8491. Fax: (919) 966-9800. E-mail:
christy_avery@unc.edu
We acknowledge C. Croghan (U.S. Environmental
Protection Agency) for providing the additional data
analyses used in this article.
This research was supported by grant R01-
ES012238 and P30-ES10126 from the National
Institute of Environmental Health Sciences and by
grant T32-HL007055 from the National Heart,
Lung, and Blood Institute.
This work has been reviewed by the U.S.
Environmental Protection Agency and approved for
publication but may not necessarily reflect official
agency policy.
The authors declare they have no competing
financial interests.
Received 01 July 2009; accepted 14 January 2010.
Estimating Error in Using Residential Outdoor PM2.5 Concentrations as
Proxies for Personal Exposures: A Meta-analysis
Christy L. Avery,1 Katherine T. Mills,1 Ronald Williams,2 Kathleen A. McGraw,3 Charles Poole,1 Richard L. Smith,4
and Eric A. Whitsel1,5
1Department of Epidemiology, University of North Carolina–Chapel Hill, Chapel Hill, North Carolina, USA; 2U.S. Environmental
Protection Agency, National Exposure Research Laboratory, Research Triangle Park, North Carolina, USA; 3Health Sciences Library,
4Department of Statistics and Operations Research, and 5Department of Medicine, University of North Carolina–Chapel Hill, Chapel Hill,
North Carolina, USA
Background: Studies examining the health effects of particulate matter ≤ 2.5 µm in aerodynamic
diameter (PM2.5) commonly use ambient PM2.5 concentrations measured at distal monitoring sites
as proxies for personal exposure and assume spatial homogeneity of ambient PM2.5. An alternative
proxy—the residential outdoor PM2.5 concentration measured adjacent to participant homes—has
few advantages under this assumption.
oBjectives: We systematically reviewed the correlation between residential outdoor PM2.5 and
personal PM2.5 (–rj) as a means of comparing the magnitude and sources of measurement error asso-
ciated with their use as exposure surrogates.
Methods: We searched seven electronic reference databases for studies of the within-participant
residential outdoor-personal PM2.5 correlation.
results: The search identified 567 candidate studies, nine of which were abstracted in duplicate,
that were published between 1996 and 2008. They represented 329 nonsmoking participants 6–93
years of age in eight U.S. cities, among whom –rj was estimated (median, 0.53; range, 0.25–0.79)
based on a median of seven residential outdoor-personal PM2.5 pairs per participant. We found
modest evidence of publication bias (symmetric funnel plot; pBegg = 0.4; pEgger = 0.2); however, we
identified evidence of heterogeneity (Cochran’s Q-test p = 0.05). Of the 20 characteristics examined,
earlier study midpoints, eastern longitudes, older mean age, higher outdoor temperatures, and lower
personal-residential outdoor PM2.5 differences were associated with increased within-participant
residential outdoor-personal PM2.5 correlations.
conclusions: These findings were similar to those from a contemporaneous meta-analysis that
examined ambient-personal PM2.5 correlations (‒rj = median, 0.54; range, 0.09–0.83). Collectively,
the meta-analyses suggest that residential outdoor-personal and ambient-personal PM2.5 cor-
relations merit greater consideration when evaluating the potential for bias in studies of PM2.5-
mediated health effects.
key words: air pollution, measurement error, meta-analysis, PM2.5. Environ Health Perspect
118:673–678 (2010). doi:10.1289/ehp.0901158 [Online 14 January 2010]

Page 2

Avery et al.
674
volume 118 | number 5 | May 2010 • Environmental Health Perspectives
within-participant residential outdoor-personal
PM2.5 correlation. No limitations on document
type, language, or publication date were used.
On 12 November 2007, we conducted searches
in PubMed (http://www.ncbi.nlm.nih.gov/
pubmed; 1950 to 12 November 2007), Web of
Science (http://thomsonreuters.com/ products_
services/science/science_ products/a-z/web_of_
science; 1955 to 12 November 2007), BIOSIS
Previews (http://www. thomsonscientific.com/
cgi-bin/jrnlst/jloptions.cgi?PC=BP; 1969 to
12 November 2007), CSA Environmental
Sciences and Pollution Management
(http://www.csa.com/factsheets/envclust-
set-c.php; 1967 to 12 November 2007),
TOXLINE (http://toxnet.nlm.nih.gov/;
1965 to 12 November 2007), and Proquest
Dissertations and Theses (http://www.
proquest.com/en-US/catalogs/databases/
detail/pqdt.shtml; 1861 to 12 November
2007). We searched EMBASE (http://www.
embase.com/; 1974 to 12 November 2007),
on 14 December 2007.
The following strategy was used to search
PubMed: (PM 2.5 OR PM2.5 OR PM25
OR PM 25 OR fine particle) AND (ambient
OR outdoor OR outdoors OR outside OR
exterior OR external OR background OR
fixed site*) AND (individual OR personal)
AND (correlat* OR associat* OR relat* OR
compar* OR pearson OR spearman). The
same four sets of key words were adapted for
input into Web of Science, BIOSIS Previews,
CSA Environmental Sciences and Pollution
Management, TOXLINE, and EMBASE.
The Dissertations and Theses search required
only the first three sets of key words to create
a result set small enough for review.
We downloaded citations to an electronic
reference manager (EndNote X1; Thomson
Reuters, New York, NY), de-duplicated, and
supplemented with secondary references cited
in articles identified in the primary search.
The citations were independently reviewed
with respect to three inclusion criteria: mea-
surement of residential outdoor PM2.5, mea-
surement of personal PM2.5, and estimation
of the within-participant residential outdoor-
personal PM2.5 correlation. Study, participant,
and environment characteristics were extracted
from all articles meeting the inclusion crite-
ria. The study characteristics were journal of
publication, publication date, setting, study
dates, sample size, duration of study, timing
(consecutive, nonconsecutive), lower limit of
PM2.5 detection, number (minimum, mean)
of paired PM2.5 measures, and correlation
metric (Pearson, Spearman). Participant char-
acteristics included age (mean, minimum,
maximum), percent female, and the presence
of comorbidities (pulmonary, cardiovascular,
multiple, neither). Environmental characteris-
tics included the mean, median, and standard
deviation of PM2.5 concentrations (residen-
tial outdoor, personal), the within-participant
residential outdoor-personal PM2.5 correla-
tion coefficients and corresponding number
of paired measurements, season, distance to
monitor, monitor type, air exchange rate,
percentage of time using air conditioning,
and percentage of time with windows open.
Discrepant exclusions and extractions were
adjudicated by consensus. Supplemental
data were requested from authors by elec-
tronic mail as needed. City-specific longi-
tudes and latitudes were obtained from the
GEOnet Names Server (National Geospatial-
Intelligence Agency 2009). Meteorologic data
were obtained from the National Climatic
Data Center (2009).
Table 1. Characteristics of nine U.S. studies examining the within-participant residential outdoor- personal
PM2.5 correlation.
Study dates (month/day/year)
Setting
Start CityState
Wallace 1996Azusa CA03/06/1989
Rojas-Bracho et al. 2000BostonMA 02/05/1996
Williams et al. 2000a, 2000b TowsonMD07/26/1998
Rodes et al. 2001
1FresnoCA02/01/1999
2 FresnoCA 04/19/1999
Suh et al. 2003
1Los AngelesCA06/12/2000
2Los AngelesCA 02/11/2000
Liu et al. 2003
1SeattleWA10/26/1999
2SeattleWA10/26/1999
3SeattleWA 02/07/2000
4 SeattleWA 11/27/2000
Reid 2003
1 AtlantaGA09/21/1999
2AtlantaGA04/01/2000
Williams et al. 2003aRaleighNC06/09/2000
Brown et al. 2008
1Boston MA 11/15/1999
2 BostonMA 06/06/2000
All nine studies totaled
(1996–2008), 16 substudies
Abbreviations: C, consecutive; N, nonconsecutive; P, Pearson product-moment correlation coefficient; r, within-par-
ticipant residential outdoor-personal PM2.5 correlation estimation method; S, Spearman’s rank correlation coefficient.
Summary statistics are reported as counts, range, proportion, or median. “Pairs” indicates average number of outdoor-
personal paired measurements for estimation of within-participant correlations. Williams et al. 2000a and 2000b refer to
the same study.
Study/substudyEnd
Duration
(months)
0.2
11.7
0.9
PM2.5 measures
TimingPairs
N
C
C
r
P
P
P
03/13/1989
02/02/1997
08/23/1998
7
13
16
02/28/1999
05/16/1999
0.9
0.9
C
N
8
7
P
P
07/24/2000
03/22/2000
1.4
1.3
C
C
6
6
S
S
08/10/2000
10/26/2000
05/24/2001
02/24/2001
9.3
11.8
15.2
2.9
C
C
C
C
7
7
7
7
P
P
P
P
11/23/1999
05/13/2000
05/21/2001
2.0
1.4
11.2
C
C
N
6
6
S
S
P20
01/29/2000
07/25/2000
2.4
1.6
1.9
C
C
6
5
7
S
S
861989 – 200170% C63% P
Table 2. Characteristics of participants in nine studies that examined the within-participant residential
outdoor-personal PM2.5 correlation.
Participant Age
Study SubstudynMeanMinimum
Wallace 19961034.1
Rojas-Bracho et al. 200017—b
Williams et al. 2000a,
2000b
Rodes et al. 200115 85
21485
Suh et al. 20031 1468.1
2 1370
Liu et al. 200313076.3
2 4877.3
33376.6
4229
Reid 20031 2364
222 63
Williams et al. 2003a3670
Brown et al. 20081 12—c
2 11—c
All nine studies totaled
1996–2008
Abbreviations: N, no disease; P, chronic pulmonary disease; C, chronic cardiovascular disease.
aSummary statistics reported as counts, range, proportion, or median; bRequested but not provided as of 18 November
2009. cNot collected. Williams et al. 2000a and 2000b refer to the same study.
Maximum
52
—b
93
Percent female
30
—b
81
Comorbiditya
N
P
N, C, P
11
—b
7219 81
55
55
55
60
66
65
57
—b
—b
84
84
88
89
86
13
88
84
85
—c
—c
93
68
68
87
93
61
55
35
24
33
50
74
20
27
55%
N
N
P
P
N
P
C
P6
33
33
55
40
40
C, P
C, P
C
C, P
C, P
25% N16329706

Page 3

Outdoor and personal PM2.5
Environmental Health Perspectives • volume 118 | number 5 | May 2010
675
Statistical analysis. Summary correla-
tion and variance estimates for the jth study
were estimated from the personal ambient
PM2.5 correlations measured for each of the
ith participants. Each within-participant
correlation coefficient (ri) was converted to
its variance-stabilizing Fisher’s z-transform:
Zri = (1 ÷ 2)loge[(1 + ri) ÷ (1 – ri)] (Fisher
1925). Estimates of the within- participant
variance [vi = 1 ÷ (ni – 3)] and between-
participant variance (τj2 = [Qj – (kj – 1)] ÷ c)
for the jth study were estimated from the
number of paired personal-residential out-
door PM2.5 measurements for each partici-
pant (ni), the number of participants per
study (kj), the weighted sum of squared
errors [Qj = Σk
and a constant (c) = Σki=1(ni – 3) –
[Σk
formed effect size for the jth study is given
by -Zj = Σk
ipant-specific weights [wi = ([1 ÷ (ni – 3)]
+ τj2)–1], study-specific standard errors
[Sj = (1 ÷ Σk
weights [Wj = (1 ÷ sj)2]. Negative τ2 estimates
were set to 0 (Field 2001).
We assessed publication bias, which
is present when study results influence the
chance or timing of publication (Begg and
Berlin 1989), using a “funnel plot” of Wj
versus -Zj. In the absence of publication bias,
plots usually resemble a symmetrical funnel,
with the more precise estimates forming the
spout and the less precise estimates forming
the cone. We also evaluated the adjusted rank
correlation (Begg and Mazumdar 1994) and
regression asymmetry tests (Egger et al. 1997)
as well as a nonparametric “trim-and-fill”
method that imputes hypothetically missing
results due to publication bias (Duval and
i=1(ni – 3)(Zri – Zri)2],
i=1(ni – 3)2 ÷ Σk
i=1(ni – 3)]). The trans-
i=1wiZri ÷ Σk
i=1wi with partic-
i=1wi)1/2], and study-specific
Tweedie 2000). Low p-values associated with
the former tests (pBegg, pEgger) give evidence of
asymmetry.
Interstudy heterogeneity was evaluated
using a plot of -Zj ÷ Sj versus 1 ÷ Sj (Galbraith
1988) and with Cochran’s Q-test (Cochran
1954). The plot and test are related in that
the position of the jth study along the vertical
axis illustrates its contribution to Q-test statis-
tic. In the absence of appreciable evidence of
heterogeneity, all studies fall within the 95%
confidence interval (CI) and pCochran > 0.1.
We first assessed variation in the strength
and precision of -Zj across levels of the study,
environment, and participant characteristics
with a summary random-effects estimate of
-
Z within each study, environment, and par-
ticipant category (Berkey et al. 1995). We also
constructed a series of univariable random-
effects meta-regression models to relate each
study, environment, and participant character-
istic to differences in -Zj. Lastly, a multivariable
random-effects meta-regression model and a
backward elimination strategy were used to
evaluate 8 study, participant, and environ-
ment characteristics routinely available in epi-
demiologic studies of PM2.5 health effects:
latitude, longitude, mean age, percent female,
relative humidity, sea level pressure, mean
temperature, and mean residential outdoor
PM2.5 (measured in this setting or spatially
interpolated in other studies). Interval-scale
characteristics were analyzed before and after
dichotomization at their medians unless
noted otherwise. We used STATA (version 9;
StataCorp LP, College Station, TX) to per-
form all the analyses. To facilitate interpre-
tation, summary estimates (i.e., -
back-transformed to their original metric ‒r
after data analysis.
Z ) were
Results
The systematic review identified 567 candi-
date studies for screening. Of these studies,
nine (2%) met the criteria for critical appraisal
and were abstracted (Brown et al. 2008; Liu
et al. 2003; Reid 2003; Rodes et al. 2001;
Rojas-Bracho et al. 2000; Suh et al. 2003;
Wallace 1996; Williams et al. 2000a, 2000b,
2003a). Abstracted studies were published
between 1996 and 2008 (Table 1), were set
in eight cities in six U.S. states, and were con-
ducted between 1989 and 2001. The median
study duration was 1.9 months (range, 0.2–
15.2 months), a period in which 70% of the
studies collected PM2.5 data over consecutive
days. During data collection, the investiga-
tors recorded a median of seven (range, 5–20)
pairs of residential outdoor and personal
PM2.5 concentrations per participant, on
which the within-participant Pearson (63%)
and Spearman (37%) correlation coefficients
were based (Table 1).
The studies represented 329 nonsmok-
ing participants 6–93 (median, 70) years
old, 55% of whom were female and 25% of
whom did not report chronic pulmonary or
cardiovascular disease (Table 2). On aver-
age, residential outdoor PM2.5 concentra-
tions (range, 8.6–42.6 µg/m3) were lower
than personal PM2.5 concentrations (range,
9.3–70.0 µg/m3), with a median residential
outdoor-personal PM2.5 difference of –1.55
µg/m3 (range, –27.4 to 9.0 µg/m3; Table 3).
The estimated ‒rj (median, 0.53; range,
0.25–0.79) and its standard deviation varied
widely (Figure 1), the latter reflecting variabil-
ity in sample weights (median, 53.6; range,
9.4–548.1). Temperature (range, 2.0–24.0°C)
and relative humidity (range, 27.3–78.9%)
were also variable.
Table 3. Environmental characteristics for nine studies that examined the within-participant correlation between residential outdoor and personal PM2.5.
Residential outdoor
PM2.5 (µg/m3)PM2.5 (µg/m3)
Study SubstudyMean ± SDMean ± SD
Wallace 199642.6 ± NR70 ± NR
Rojas-Bracho et al. 2000 14.2 ± 11.221.6 ± 13.6
Williams et al. 2000a,
2000b
Rodes et al. 20011 20.5 ± 13.413.1 ± 5.9
2 10.1 ± 3.211.1 ± 2.8
Suh et al. 20031 19.3 ± 9.0 25.1 ± 20.8
2 13.5 ± 8.5 19.6 ± 14.5
Liu et al. 20031 9.0 ± 4.69.3 ± 8.4
2 9.2 ± 5.1 10.5 ± 7.2
3 12.6 ± 7.9 10.8 ± 8.4
4 11.3 ± 6.413.3 ± 8.2
Reid 20031 14.5 ± 7.3 16.3 ± 8.4
222.7 ± 10.615.0 ± 7.5
Williams et al. 2003a19.3 ± 8.4323.0 ± 16.1
Brown et al. 20081 8.6 ± 5.212.0 ± 6.0
212.5 ± 7.610.0 ± 6.2
All nine studies totaled
1996–2008
Abbreviations: DP, dew point; NR, not reported; –rj, mean within-participant residential outdoor PM2.5-personal PM2.5 correlation coefficient; RH, relative humidity; SD, standard devia-
tion; SLP, sea level pressure; T, temperature.
Personal
rMeteorologic data, mean over study dates
DP (°C)
52.0
45.4
64.0
–rj
SD
0.16
0.11
0.08
T (°C)
11.7
13.2
24.0
SLP (kPa)
101.81
101.56
101.85
RH (%)
27.3
68.0
68.3
0.41
0.64
0.7922.0 ± 12.013.0 ± 3.2
0.58
0.65
0.32
0.59
0.47
0.51
0.55
0.41
0.76
0.48
0.35
0.25
0.75
0.53
0.18
0.20
0.14
0.16
0.10
0.09
0.13
0.11
0.18
0.12
0.04
0.22
0.35
0.14
9.6
17.5
21.1
13.7
9.9
10.8
10.0
6.9
15.7
17.2
17.2
2.0
20.4
13.4
41.8
41.2
60.3
46.8
43.6
44.8
42.8
37.8
49.7
49.8
51.9
22.7
58.6
46.1
102.27
101.42
101.34
101.70
101.78
101.78
101.82
101.90
102.01
101.64
101.92
101.67
101.43
101.78
75.2
43.9
71.3
69.7
78.9
77.8
76.0
77.1
68.3
62.0
67.4
59.0
70.3
69.016 13.9 ± 7.9 13.2 ± 8.2

Page 4

Avery et al.
676
volume 118 | number 5 | May 2010 • Environmental Health Perspectives
Figure 2, a funnel plot of -
tle evidence of asymmetry. This was consis-
tent with pBegg = 0.4, pEgger = 0.2, although
the “trim-and-fill” analysis imputed seven
hypothetically missing studies. Figure 3, a
Galbraith plot in which three observations
fell outside the 95% CIs, provides evidence
of heterogeneity. This evidence was consistent
with pCochran = 0.05.
Several study, participant, and envi-
ronmental characteristics were suggestively
associated with moderate increases in the
within- participant residential outdoor-personal
PM2.5 correlation coefficient in univariate meta-
regression models (Figure 4), including earlier
study midpoints, eastern longitudes, older mean
age, lower personal-residential outdoor PM2.5
differences (and ratios), and higher mean tem-
peratures (Figure 5). For example, every 5°C
increase in mean temperature was associated
with a 0.10 95% CI, (–0.02, 0.21) unit differ-
ence in ‒r . The direct association between mean
temperature and ‒rj also was apparent when eval-
uating mean temperature dichotomized at the
median: In studies with a mean temperature
≥ 13.43°C, ‒r was 0.59 (range, 0.40–0.74), and
in those with a mean temperature < 13.43°C, ‒r
was 0.50 (range, 0.44–0.56).
When evaluating multivariable meta-
regression models, only higher mean ages
and eastern longitudes were associated with
an increased within-participant residential
outdoor-personal PM2.5 correlation coeffi-
cient (p < 0.05).
Discussion
Epidemiologic studies of the health effects
of PM2.5 typically estimate PM2.5 expo-
sures using daily mean concentrations either
obtained from a single ambient PM2.5 moni-
toring site or averaged across several sites
Zj, shows lit-(U.S. EPA 1996). Although rapid dispersion
and secondary formation of atmospheric
PM2.5 via chemical reactions of such gases as
sulfur dioxide, nitrogen oxides, and ammonia
ensure some geographic uniformity of the
monitored concentrations, primary sources of
anthropogenic PM2.5, including traffic, con-
struction, and industry (Samet and Krewski
2007), can increase the spatial variability of
PM2.5. Additional factors that influence the
relationship between ambient PM2.5 concen-
trations and PM2.5 exposures include home
ventilation, indoor activities associated with
generation or resuspension of PM2.5 like
cooking or cleaning, and time–activity pat-
terns (Liu et al. 2003; Williams et al. 2000b).
Thus, estimates of PM2.5 exposure based on
ambient PM2.5 concentrations are associated
with an acknowledged degree of uncertainty
(Janssen et al. 1998).
To further characterize this uncer-
tainty, in the present study we extended a
prior meta-analysis of the within-participant
ambient-personal PM2.5 correlation (Avery
et al. 2010) by examining the within-
participant residential outdoor-personal
PM2.5 correlation using analogous meta-
analytic methods. In both cases, the examina-
tion generated little evidence for publication
bias of Fisher’s z-transformed ‒rj but strong
evidence of heterogeneity. Several study,
participant, and environment characteristics
were associated with an increased ‒rj, including
earlier study midpoints, eastern longitudes,
lower personal-residential outdoor PM2.5
differences (and ratios), higher mean ages,
and higher mean temperatures. Moreover,
the direct association between eastern longi-
tudes and increased ‒rj was consistent with the
prior meta-analysis of the within-participant
ambient-personal PM2.5 correlation.
The direct association between eastern
longitudes and increased ‒rj may reflect sev-
eral regional factors, including higher urban
PM2.5 concentrations (Rom and Markowitz
2006) or a greater influence of secondary
PM2.5 sources in eastern locales (Pinto et al.
2004). The inverse associations between the
residential outdoor-personal PM2.5 difference
(or ratio) and mean temperature with ‒rj may
also suggest lower microenvironmental varia-
tion in PM2.5 or an increased contribution of
residential outdoor to personal PM2.5 expo-
sure, through either time–activity patterns or
increased air exchange. We were unable to
fully evaluate the influence of these factors
given the limited number of published stud-
ies and their inconsistent reporting of other
geographic, household, and personal factors
potentially responsible for the above associa-
tions. However, higher mean ages and eastern
longitudes were associated with increased ‒rj
in the multivariable prediction model that
included study, participant, and environment
characteristics routinely available in epidemio-
logic studies of PM2.5 health effects.
Figure 1. Forest plot for 16 estimates of –rj (95% CIs) from nine studies of the within-participant residential
outdoor-personal PM2.5 correlation.
Wallace et al. 1996
Rojas-Bracho et al. 2000
Williams et al. 2000a, b
Rodes et al. 2000, 1
Rodes et al. 2000, 2
Suh et al. 2003, 1
Suh et al. 2003, 2
Liu et al. 2003, 1
Liu et al. 2003, 2
Liu et al. 2003, 3
Liu et al. 2003, 4
Williams et al. 2003b
Reid et al. 2003, 1
Reid et al. 2003, 2
Brown et al. 2008, 1
Brown et al. 2008, 2
–0.2500.25
r–
j (95% CI)
0.500.75 1.00
Study
Figure 2. Funnel plot for 16 estimates of the
within-participant residential outdoor-personal
PM2.5 correlation.
600
500
400
300
200
100
0
–0.5
Wj
Z– –
j
01.0 1.5
Reported
Imputed
0.5
Figure 3. Galbraith plot with 95% CIs for 16 esti-
mates of the within-participant residential outdoor-
personal PM2.5 correlation.
1 ÷ Sj
Z––
j ÷ Sj
15
10
5
0
0510 1520 25

Page 5

Outdoor and personal PM2.5
Environmental Health Perspectives • volume 118 | number 5 | May 2010
677
Although the meta-analyses of the
ambient-personal and residential outdoor-
personal PM2.5 correlations summarized a
wide range of published correlation coef-
ficients, both of them estimated a median
‒rj of 0.5, which suggests that attempting to
account for spatial variability and outdoor
microenvironments does not appreciably
affect the use of outdoor PM2.5 concentra-
tions as proxies for personal PM2.5 exposure
in the settings examined by the source studies.
Nonetheless, these simple measures of central
tendency have potentially important implica-
tions for studies using PM2.5 concentrations
measured at distal or proximal monitoring
sites. For example, an ‒r of 0.5 implies that, on
average, only ‒r2 or one-fourth of the variation
in personal PM2.5 is explained by ambient
or residential outdoor PM2.5 concentrations.
Under a simple measurement error model, it
also implies that the variances of ambient or
residential outdoor PM2.5 concentrations are
1/‒r2, or four times as large as the variance
of the true, but often unmeasured, personal
PM2.5 exposure. Moreover, ‒r values of 0.5 in
diseased and nondiseased subpopulations (i.e.,
nondifferential exposure measurement error)
imply that a) sample sizes needed to detect
between-group differences in mean ambient
or residential outdoor PM2.5 concentrations
are 1/r2, or 4-fold as large as those needed to
detect the same differences in personal PM2.5
exposures, and b) effect estimates expressed
as microgram per cubic meter increases in
ambient or residential outdoor PM2.5 con-
centrations are equal to those associated with
the same microgram per cubic meter increases
in personal PM2.5 exposure, albeit attenu-
ated toward the null by the power r2 or 0.25.
The latter form of attenuation is capable of
obscuring weak to modest health effects of
PM2.5 (White et al. 2003), yet it cannot be
adequately controlled by methods commonly
used to account for confounding (Greenland
and Robins 1985).
Given the above considerations, it is
tempting to assume that all health effect esti-
mates based on ambient or residential outdoor
PM2.5 concentrations would be consider-
ably larger if they were instead based on per-
sonal PM2.5 exposures, but to do so would
yield more biased estimates if the original
PM2.5–disease associations were spurious due
to chance or confounding (Armstrong 1998).
This justifies the application of the present
findings to the PM2.5–disease associations that
are the most precise and least biased accord-
ing to criteria used to judge epidemiologic
evidence (Hill 1965; Poole 2001; U.S. EPA
2009). Furthermore, factors associated with ‒r,
such as mean age and eastern longitudes, may
differ among participants and the studies in
which they are enrolled. It is therefore difficult
to predict the degree to which PM2.5 health
effects estimates may be biased by exposure
measurement error. Nonetheless, the above
examples clearly illustrate that the impact of ‒r
on the interpretation of findings from studies
of PM2.5 health effects may be substantial.
Although in the present study we
attempted to quantify the error associated with
using residential outdoor and ambient PM2.5
concentrations as proxies for total personal
exposure, the approach adopted here has sev-
eral limitations. First, residential outdoor and
ambient PM2.5 concentrations are likely to be
poor proxies for exposure to nonambient PM
because PM originating indoors has different
Figure 4. Unadjusted summary correlations (95% CIs) and differences (95% CIs) by study, participant, and
environment characteristics for nine studies examining the within-participant residential outdoor-personal
PM2.5 correlation. Summary correlations represent stratum-specific estimates of –r. Increases in –r per unit
change of study, participant, and environment characteristics are provided by –r difference estimates. SLP,
sea level pressure.
1.00.50 –0.5–1.0 1.0 0.50 –0.5–1.0
Year


Study midpoint

Measurement type

Latitude


Longitude


Correlation coefficient

Mean number of
paired measures

Participant characteristics
Comorbidity



Mean age, restricted
to adults

Percent female


Environment characteristics
Mean outdoor PM2.5


Mean personal PM2.5


Personal-outdoor PM2.5


Personal/outdoor PM2.5


Relative humidity


Dew point


SLP (kPa)


Mean temperature


Per 1-year increase
10/1996–03/2003
6/2003–7/2008
03/1989–03/2000
03/2000–03/2001
Consecutive
Nonconsecutive
Per 5° increase
≥ 38.07°
< 38.07°
Per 10° increase
≥ –117.9
< –117.9
Spearman
Pearson
Per 1-pair increase
≥ 7
< 7

0.59 (0.42, 0.72)
0.49 (0.4, 0.58)
0.62 (0.48, 0.73)
0.45 (0.37, 0.52)
0.56 (0.46, 0.65)
0.43 (0.25, 0.58)

0.57 (0.43, 0.69)
0.50 (0.38, 0.61)

0.61 (0.39, 0.76)
0.48 (0.42, 0.54)
0.53 (0.36, 0.67)
0.55 (0.42, 0.66)

0.55 (0.42, 0.66)
0.53 (0.36, 0.67)

0.65 (0.44, 0.79)
0.43 (0.21, 0.61)
0.49 (0.39, 0.59)
0.50 (0.39, 0.60)

0.57 (0.41, 0.70)
0.50 (0.34, 0.64)

0.53 (0.38, 0.66)
0.54 (0.38, 0.66)


0.57 (0.38, 0.71)
0.50 (0.43, 0.56)

0.50 (0.39, 0.59)
0.59 (0.43, 0.72)

0.49 (0.40, 0.57)
0.64 (0.46, 0.77)

0.49 (0.40, 0.57)
0.64 (0.46, 0.77)

0.48 (0.41, 0.55)
0.57 (0.4, 0.71)

0.59 (0.41, 0.73)
0.49 (0.42, 0.55)

0.56 (0.39, 0.70)
0.52 (0.43, 0.60)

0.59 (0.40, 0.74)
0.50 (0.44, 0.56)
–0.02 (–0.07, 0.04)
0.13 (–0.12, 0.37)
0
0.25 (0.03, 0.44)
0
0.15 (–0.17, 0.44)
0
–0.01 (–0.13, 0.11)
0.09 (–0.16, 0.34)
0
0.04 (–0.02, 0.09)
0.46 (–0.09, 0.39)
0
–0.02 (–0.30, 0.26)
0
0.01 (–0.02, 0.04)
0.02 (–0.26, 0.30)
0

0.22 (–0.12, 0.51)
–0.11 (–0.46, 0.27)
–0.04 (–0.35, 0.28)
0
0.07 (–0.01, 0.14)
0.09 (–0.23, 0.40)
0
0.01 (–0.05, 0.07)
–0.01 (–0.30, 0.28)
0

–0.01 (–0.18, 0.15)
0.07 (–0.19, 0.32)
0
–0.02 (–0.07, 0.03)
–0.14 (–0.37, 0.12)
0
–0.06 (–0.14, 0.02)
–0.22 (–0.45, 0.03)
0
–0.17 (–0.35, 0.03)
–0.22 (–0.45, 0.03)
0
0.01 (–0.10, 0.11)
–0.11 (–0.35, 0.15)
0
0.12 (–0.14, 0.37)
0.13 (–0.12, 0.37)
0
0.05 (–0.08, 0.11)
0.05 (–0.22, 0.30)
0
0.10 (–0.02, 0.21)
0.12 (–0.14, 0.37)
0
Combined
Cardiovascular
Pulmonary
Healthy
Per 10-year increase
≥ 69 years
< 69 years
Per 10% increase
≥ 50%
< 50%
Per 10-µg/m3 increase
≥ 13.85 µg/m3
< 13.85 µg/m3
Per 5-µg/m3 increase
≥ 13.3 µg/m3
< 13.3 µg/m3
Per 5-µg/m3 increase
≥ 0 µg/m3
< 0 µg/m3
Per 0.5-µg/m3 increase
≥ 1
< 1
Per 10% increase
≥ 69.72%
< 69.72%
Per 10°C increase
≥ 44°C
< 44°C
Per 0.15 unit increase
≥ 101.8
< 101.8
Per 5°C increase
≥ 13.43°C
< 13.43°C
r– differenceSummary r–
Study characteristics
Figure 5. Plot for 16 estimates of the within- participant
residential outdoor-personal PM2.5 correlation
(95% CI) versus mean outdoor temperature, including
the univariate random-effects meta- regression line.
Z– –
j (95% CI)
2.0
1.5
1.0
0.5
0.0
–0.5
05 1015
Mean temperature (°C)
20 25

Page 6

Avery et al.
678
volume 118 | number 5 | May 2010 • Environmental Health Perspectives
compositions and biological properties (Long
et al. 2001). Although the relative toxicity of
outdoor and indoor PM remains under inves-
tigation, a panel study of 16 chronic obstruc-
tive pulmonary disease patients in Vancouver,
British Columbia, reported that only the
PM originating outdoors was associated with
adverse cardiopulmonary effects (Ebelt et al.
2005). Moreover, in the present study we did
not evaluate the correlation between concen-
trations of PM originating almost exclusively
outdoors (e.g., sulfate or elemental carbon)
and personal PM2.5 exposure, despite reports
that their associations with ambient PM2.5 are
particularly strong (Ebelt et al. 2000; Sarnat
et al. 2006). Further work examining the rela-
tive contributions of PM2.5 constituents to
PM-mediated health effects is clearly needed.
In summary, the results presented here
and in the previous meta-analysis of the
within-participant ambient-personal PM2.5
correlation (Avery et al. 2010) suggest that
greater scrutiny of the effects of exposure
measurement error is warranted. Further
inquiry should involve quantifying the impact
of using ambient or residential outdoor PM2.5
concentrations as proxies for personal PM2.5
exposure, as well as the development of meth-
odologies to apply such findings. A compre-
hensive understanding of the degree to which
these proxies influence PM2.5–disease associa-
tions is especially important in air pollution
epidemiology because the health effects of
PM2.5 exposure may be subtle. Such subclini-
cal effects are particularly difficult to detect in
the presence of measurement error because
sensitivity of detection varies inversely with
the degree of misclassification (Rom and
Markowitz 2006).
RefeRences
Armstrong BG. 1998. Effect of measurement error on epide-
miological studies of environmental and occupational
exposures. Occup Environ Med 55(10):651–656.
Avery CL, Mills KT, Williams R, McGraw K, Poole C, Smith RL,
et al. 2010. Estimating error in using ambient PM2.5 concen-
trations as proxies for personal exposures. Epidemiology
21(2):215–223.
Begg CB, Berlin JA. 1989. Publication bias and dissemination of
clinical research. J Natl Cancer Inst 81(2):107–115.
Begg CB, Mazumdar M. 1994. Operating characteristics of
a rank correlation test for publication bias. Biometrics
50(4):1088–1101.
Berkey CS, Hoaglin DC, Mosteller F, Colditz GA. 1995. A ran-
dom-effects regression model for meta-analysis. Stat Med
14(4):395–411.
Brown KW, Sarnat JA, Suh H, Coull BA, Spengler JD,
Koutrakis P. 2008. Ambient site, home outdoor and home
indoor particulate concentrations as proxies of personal
exposures. J Environ Monit 10(9):1041–1051.
Burnett RT, Brook J, Dann T, Delocla C, Philips O, Cakmak S,
et al. 2000. Association between particulate- and gas-
phase components of urban air pollution and daily mortal-
ity in eight Canadian cities. Inhal Toxicol 12(suppl 4):15–39.
Burnett RT, Dales R, Krewski D, Vincent R, Dann T, Brook JR.
1995. Associations between ambient particulate sulfate
and admissions to Ontario hospitals for cardiac and respi-
ratory diseases. Am J Epidemiol 142(1):15–22.
Clean Air Act. 1970. 42USC7401.
Cochran WG. 1954. The combination of estimates from different
experiments. Biometrics 10(1):101–129.
Diez Roux AV, Auchincloss AH, Franklin TG, Raghunathan T,
Barr RG, Kaufman J, et al. 2008. Long-term exposure to
ambient particulate matter and prevalence of sub-
clinical atherosclerosis in the Multi-Ethnic Study of
Atherosclerosis. Am J Epidemiol 167(6):667–675.
Dominici F, McDermott A, Daniels M, Zeger SL, Samet J.
2003. Mortality among residents of 90 cities. In: Revised
Analyses of Time-Series Studies of Air Pollution and
Health; Revised Analyses of the National Morbidity,
Mortality, and Air Pollution Study (NMMAPS), Part II.
Boston:Health Effects Institute, 9–24.
Duval S, Tweedie R. 2000. Trim and fill: a simple funnel-plot-
based method of testing and adjusting for publication bias
in meta-analysis. Biometrics 56(2):455–463.
Ebelt ST, Petkau AJ, Vedal S, Fisher TV, Brauer M. 2000.
Exposure of chronic obstructive pulmonary disease
patients to particulate matter: relationships between per-
sonal and ambient air concentrations. J Air Waste Manag
Assoc 50(7):1081–1094.
Ebelt ST, Wilson WE, Brauer M. 2005. Exposure to ambient and
nonambient components of particulate matter: a compari-
son of health effects. Epidemiology 16(3):396–405.
Egger M, Davey Smith G, Schneider M, Minder C. 1997. Bias in
meta-analysis detected by a simple, graphical test. BMJ
315(7109):629–634.
Field AP. 2001. Meta-analysis of correlation coefficients: a
Monte Carlo comparison of fixed- and random-effects
methods. Psychol Methods 6(2):161–180.
Fisher RA. 1925. Statistical Methods for Research Workers.
Edinburgh:Oliver & Boyd.
Galbraith RF. 1988. A note on graphical presentation of esti-
mated odds ratios from several clinical trials. Stat Med
7(8):889–894.
Greenland S, Robins JM. 1985. Confounding and misclassifica-
tion. Am J Epidemiol 122(3):495–506.
Hill AB. 1965. The environment and disease: association or
causation? Proc R Soc Med 58:295–300.
Janssen NA, Hoek G, Brunekreef B, Harssema H, Mensink I,
Zuidhof A. 1998. Personal sampling of particles in adults:
relation among personal, indoor, and outdoor air concen-
trations. Am J Epidemiol 147(6):537–547.
Katsouyanni K, Touloumi G, Samoli E, Petasakis Y, Analitis A, Le
Tertre A, et al. 2003. Sensitivity analysis of various models
of short-term effects of ambient particles on total mortal-
ity in 29 cities in APHEA2. In: Revised Analysis of Time‑
series Studies of Air Pollution and Health. Special report.
Boston:Health Effects Institute, pp. 157–164.
Klepeis NE, Nelson WC, Ott WR, Robinson JP, Tsang AM,
Switzer P, et al. 2001. The National Human Activity Pattern
Survey (NHAPS): a resource for assessing exposure to
environmental pollutants. J Expo Anal Environ Epidemiol
11(3):231–252.
Liao D, Whitsel EA, Duan Y, Lin HM, Quibrera PM, Smith R,
et al. 2009. Ambient particulate air pollution and ectopy—
the environmental epidemiology of arrhythmogenesis in
Women’s Health Initiative Study, 1999–2004. J Toxicol
Environ Health A 72(1):30–38.
Linn WS, Szlachcic Y, Gong H Jr., Kinney PL, Berhane KT. 2000.
Air pollution and daily hospital admissions in metropolitan
Los Angeles. Environ Health Perspect 108:427–434.
Liu LJ, Box M, Kalman D, Kaufman J, Koenig J, Larson T, et al.
2003. Exposure assessment of particulate matter for sus-
ceptible populations in Seattle. Environ Health Perspect
111:909–918.
Long CM, Suh HH, Kobzik L, Catalano PJ, Ning YY, Koutrakis P.
2001. A pilot investigation of the relative toxicity of indoor
and outdoor fine particles: in vitro effects of endotoxin
and other particulate properties. Environ Health Perspect
109:1019–1026.
Monn C, Becker S. 1999. Cytotoxicity and induction of proin-
flammatory cytokines from human monocytes exposed to
fine (PM2.5) and coarse particles (PM10–2.5) in outdoor and
indoor air. Toxicol Appl Pharmacol 155(3):245–252.
National Climatic Data Center. 2009. Weather/Climate Events.
Available: http://www.ncdc.noaa.gov/oa/climateresearch.
html [accessed 18 November 2009].
National Geospatial-Intelligence Agency. 2009. About GEOnet
Names Server. Available: http://geonames.nga.mil/ggma-
viewer/MainFrameSet.asp [accessed 18 November 2009].
Oftedal B, Nafstad P, Magnus P, Bjorkly S, Skrondal A. 2003.
Traffic related air pollution and acute hospital admission
for respiratory diseases in Drammen, Norway 1995–2000.
Eur J Epidemiol 18(7):671–675.
Pinto JP, Lefohn AS, Shadwick DS. 2004. Spatial variability of
PM2.5 in urban areas in the United States. J Air Waste
Manag Assoc 54(4):440–449.
Poole C. 2001. Low p-values or narrow confidence intervals:
which are more durable? Epidemiology 12(3):291–294.
Reid CM. 2003. Assessment of Exposure to Selected Criteria
Pollutants for Two Sensitive Population Cohorts in Atlanta,
Georgia. [Dissertation]. Atlanta, GA:Emory University.
Rodes CE, Lawless PA, Evans GF, Sheldon LS, Williams RW,
Vette AF, et al. 2001. The relationships between personal
PM exposures for elderly populations and indoor and out-
door concentrations for three retirement center scenarios.
J Expo Anal Environ Epidemiol 11(2):103–115.
Rojas-Bracho L, Suh HH, Koutrakis P. 2000. Relationships
among personal, indoor, and outdoor fine and coarse
particle concentrations for individuals with COPD. J Expo
Anal Environ Epidemiol 10(3):294–306.
Rom WN, Markowitz SB. 2006. Environmental and Occupational
Medicine. 4th ed. Philadelphia:Lippincott Williams &
Wilkins.
Samet J, Krewski D. 2007. Health effects associated with expo-
sure to ambient air pollution. J Toxicol Environ Health A
70(3–4):227–242.
Sarnat SE, Coull BA, Schwartz J, Gold DR, Suh HH. 2006.
Factors affecting the association between ambient con-
centrations and personal exposures to particles and
gases. Environ Health Perspect 114:649–654.
Suh H, Koutrakis P, Chang L. 2003. Characterization of the
Composition of Personal, Indoor, and Outdoor Particulate
Exposures. Contract 98-330. Boston:Harvard School of
Public Health, Environmental Science and Engineering
Program.
U.S. Environmental Protection Agency. 1996. Air Quality
Criteria for Particulate Matter. EPA/600/P-95/001aF-cF.3v.
Research Triangle Park, NC:U.S. EPA.
U.S. Environmental Protection Agency. 2009. Integrated
Science Assessment for Particulate Matter (Second
External Review Draft). EPA/600/R-08/139-B. Washington,
DC:U.S. EPA.
Wainman T, Zhang J, Weschler CJ, Lioy PJ. 2000. Ozone and
limonene in indoor air: a source of submicron particle
exposure. Environ Health Perspect 108:1139–1145.
Wallace L. 1996. Indoor particles: a review. J Air Waste Manag
Assoc 46(2):98–126.
Wallace L. 2000. Correlations of personal exposure to particles
with outdoor air measurements: a review of recent stud-
ies. Aerosol Sci Technol 32(1):15–25.
White E, Armstrong BK, Saracci R. 2003. Principles of Exposure
Measurement in Epidemiology. Oxford:Oxford University
Press.
Whitsel EA, Quibrera PM, Christ SL, Liao D, Prineas RJ,
Anderson GL, et al. 2009. Heart rate variability, ambient
particulate matter air pollution, and glucose homeostasis:
the environmental epidemiology of arrhythmogenesis in the
women’s health initiative. Am J Epidemiol 169(6):693–703.
Williams R, Creason J, Zweidinger R, Watts R, Sheldon L,
Shy C. 2000a. Indoor, outdoor, and personal exposure
monitoring of particulate air pollution: the Baltimore
elderly epidemiology-exposure pilot study. Atmos Environ
34(24):4193–4204.
Williams R, Suggs J, Creason J, Rodes C, Lawless P, Kwok R,
et al. 2000b. The 1998 Baltimore Particulate Matter
Epidemiology-Exposure Study: part 2. Personal exposure
assessment associated with an elderly study population. J
Expo Anal Environ Epidemiol 10(6 pt 1):533–543.
Williams R, Suggs J, Rea A, Leovic K, Vette A, Croghan C, et al.
2003a. The Research Triangle Park Particulate Matter
Panel Study: PM mass concentration relationships. Atmos
Environ 37(38):5349–5363.
Williams R, Suggs J, Rea A, Sheldon L, Rodes C, Thornburg J.
2003b. The Research Triangle Park Particulate Matter
Panel Study: modeling ambient source contribution to
personal and residential PM mass concentrations. Atmos
Environ 37:5365–5378.
Zhu YF, Hinds WC, Kim S, Sioutas C. 2002. Concentration and
size distribution of ultrafine particles near a major high-
way. J Air Waste Manag Assoc 52(9):1032–1042.