United States
Environmental Protection
Agency
Exposure of High Risk
Subpopulations to Particles:
Final Report (ARM 21)
-------
EPA/600/R-03/145
December 2003
Exposure of High Risk
Subpopulations to Particles
Final Report (APM-21)
by
Lance A. Wallace, Ronald W. Williams
Jack Suggs, Linda Sheldon, Roy Zweidinger, Anne W. Rea, Alan Vette, Kelly W. Leovic,
Gary Norris, Matt Landis, Carvin D. Stevens, Teri Conner, Carry Croghan
National Exposure Research Laboratory
U.S. EPA, RTF, NC
Research Triangle Park, NC 27711
Charles Rodes, Philip A. Lawless, Jonathan Thornburg
Research Triangle Institute, RTP, NC
Lee-Jane S. Liu, Ryan Allen , David Kalman, Joel Kaufman, Jane Koenig, Timothy Larson,
Thomas Lumley, Lianne Sheppard
University of Washington, Seattle, WA 98101
Kathleen Brown, Jeremy Samat, Helen Suh, Amanda Wheeler, Petros Koutrakis
Harvard University, Cambridge, MA 02138
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
/T~y Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free.
-------
Disclaimer
The U.S. Environmental Protection Agency through its Office of Research and Development
partially funded and collaborated in the research described here under contract numbers 68-D2-
0134 (QST Environmental), 68-D2-0187 (SRA Technologies, Inc), 68-D-99-012, 68-D5-0040
(Research Triangle Institute) and cooperative agreement numbers CR-827159 (Harvard School
of Public Health), CR-827177 (University of Washington), CR-827164 (New York University),
CR-820076 (University of North Carolina-Chapel Hill), and CR-828186-01-0 (Shaw
University). It has been subjected to the Agency's peer and administrative review, and it has been
approved for publication as an EPA document. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
-------
Abstract
This final report describes results of studies EPA conducted on the exposure of high-risk
subpopulations to particles. The overall goal of these studies was to relate personal exposure to
outdoor concentrations of particles, particularly fine particles less than 2.5 micrometers in
diameter (PMa.s). The studies were carried out by EPA's National Exposure Research
Laboratory (NERL), and by two university consortia (Harvard University School of Public
Health and the University of Washington Department of Environmental Studies). All studies
included repeated measurements over 10-28 days of personal, indoor, and outdoor air PM2 5,
including indoor and outdoor air PMio and associated co-pollutants. Nearly 2500 personal PIVb.s
filters and a much larger number of indoor and outdoor PlV^.s and PMio filters were collected
from more than 200 participants in 5 cities. All participants filled out identical household
questionnaires and time-activity diaries providing information on where they spent their time and
what particle sources were active. Participants were chosen from several high-risk
subpopulations: adults with chronic obstructive pulmonary disease (COPD), coronary heart
disease (CHD), and hypertension, and children with asthma. Healthy cohorts were also included
as controls. Some participants were in retirement homes but most lived in their own homes or
apartments. All subjects were chosen on a non-probabilistic basis, and therefore all conclusions
apply only to the subjects themselves; they must not be extrapolated to larger groups of people.
Results indicated little difference in exposure between disease cohorts and healthy persons in the
same geographic area. Across the cities, mean PM2.5 exposures ranged between 9.3 and 23
u.g/m3, compared to mean indoor concentrations of 7.4-20 ng/m3, and mean outdoor
concentrations of 9.0-22 |ug/m3. Seasonal variations were important in some cities, unimportant
in others. Median longitudinal correlations of personal PM2.5 exposure with outdoor
concentrations ranged from 0.1 to 0.65. However, all cohorts had some persons with high
correlations with outdoor air and others with low correlations, indicating the importance of
activities and household characteristics that either change over time or are basically repeatable
from day to day. Different cities had different estimates of the mean infiltration factor, a
measure of the influence of outdoor air particles on indoor concentrations. For private
residences, calculated values of the infiltration factor for PM2 5 ranged from 0.40 to 0.53 during
the heating season and from 0.45 to 0.79 during the non-heating seasons.
111
-------
Contents
Disclaimer ii
Abstract iii
Figures v
Tables vii
Acknowledgement ix
Introduction 1
Study Design 2
Quality Assurance 7
Results 16
Discussion 54
Conclusions 57
Future Work 59
References 60
APPENDIX
Journal Articles Published or in Preparation from the Panel Studies A-l
IV
-------
Figures
1 Collocated personal (APMTW) and indoor (AHMTW) monitors with
EPA's Federal Reference Method (FRM) monitor at the
Research Triangle Park central site 8
2 Wintertime comparison of the SKC Personal Exposure Monitor (PEM)
with the Harvard Impactor (HI) at the Atlanta site 9
3 Springtime comparison of the Harvard Personal Exposure Monitor (HPEM)
with the Harvard Impactor (HI) at the Atlanta central site 10
4 The Harvard PEM (HPEM2.5) vs. the Harvard Impactor (HI2.5) when
collocated at the Boston central site 11
5 Multi-Pollutant Sampler developed by Harvard School of
Public Health 12
6 Collocated PEMs and His at the Los Angeles central site 13
7 Comparison of Harvard Impactor (HI), Harvard personal environmental
Monitor (HPEM), and federal reference method (FRM) measurements
forPM25: Seattle (central sites) 15
8a Cumulative PM2.5 distributions: RTP Summer 2000 (N = 206-224) 17
8b Cumulative PM2.5 distributions: RTP Fall 2000 (N = 204-210) 17
8c Cumulative PM25 distributions: RTP Winter 2001 (N = 182) 18
8d Cumulative PM25 distributions: RTP Spring 2001 (N = 132-147) 18
8e Cumulative PM25 distributions: RTP All Seasons (N = 735-763) 19
9 Comparison of PMio and PM2 5 combined over all seasons: RTP 20
10 Daily average temperatures (°F) during field study: RTP 21
11 Boxplots of the PMio and PM2 5 concentrations (|ig/m3): RTP 22
12 Boxplots of the PM2 5 infiltration factor by season: RTP 23
-------
1 3 Boxplots of longitudinal correlation coefficients (Pearson) between
personal exposure and ambient concentration of PM2.s: RTF ........................................... 31
14 PMi 5 concentrations by location and season: Atlanta - Boxplots
indicate 5th, 25th, 50th (median), 75th and 95th percentiles .................................................. 34
1 5 Atlanta: The four types of PM? 5 measurements plotted on lognormal
probability coordinates. Plotted points are the 1st, 5*, 10th, 25th, 50th,
75*, 90th, 95th and 99th percentiles (N = 258-282) [[[ 35
16 EC concentrations by location and season: Atlanta [[[ 36
17 SO42" concentrations by location and season: Atlanta [[[ 37
18 Longitudinal correlations between personal exposures and ambient
concentrations of PM2s: Atlanta [[[ 39
19 Boston: Cumulative PM25 distributions of personal, indoor, residential
outdoor and central site concentrations (N = 266-301) [[[ 41
20 Longitudinal correlations between personal exposures and ambient
concentrations of PMas: Boston [[[ 42
21 Los Angeles: Cumulative PM2 5 distributions of personal, indoor, and
residential outdoor concentrations (N = 179-189) [[[ 44
22 Longitudinal correlations between personal exposures and ambient
concentrations of PM2 5: Los Angeles [[[ 45
-------
Tables
1 Study Design: Research Triangle Park 3
2 Study Design: Atlanta, Boston, and Los Angeles 4
3 Study Design: Seattle (Subpopulations) 5
4 Study Design: Seattle (Measurement Methods) 6
5 PM Mass Concentrations (ug/m3): RTP 16
6 Air exchange rates by season: RTP 22
7 Maximum Likelihood Estimates of Outdoor Source Contribution to Indoor
and Personal PM2.5: RTP 24
8 Least squares estimate of indoor filtration and personal exposure factors: RTP 25
9 Nonlinear Model Estimates of the Deposition Rate k, Penetration Coefficient P, and
Indoor Source Terms: RTP 28
10 Comparison of Linear and Nonlinear Model Estimates of Parameters in the Mass
Balance Equation: RTP 30
11 Descriptive Statistics for Outdoor, Indoor, and Personal Pollutant Concentrations
in Atlanta: Particles (in ug/m3) 33
12 Descriptive Statistics for Outdoor, Indoor, and Personal Pollutant
Concentrations in Atlanta: Gases (in ppb) 38
13 Descriptive Statistics for Outdoor, Indoor, and Personal Pollutant
Concentrations in Boston: Particles (in ng/m3) 40
14 Descriptive Statistics for Outdoor, Indoor, and Personal Pollutant
Concentrations in Los Angeles: Particles (in ng/m3) 43
15 Individual-Specific Spearman Correlation Coefficients: PM2.5 in
Los Angeles 46
16 Summary of PM Concentrations Between Oct 1999 and May 2001
by Health Group: Seattle 47
vn
-------
17 Correlations Between Personal, Indoor, Outdoor, and Central Site
Monitors for PM2 5 and PMio: Seattle 48
18 Summary of Estimated Particle Penetration (P), Air Exchange Rate (a),
Particle Decay Rate (k), and the Ambient PM Infiltration Efficiency (Fmf) Using
the Recursive and Nonlinear Regression Models: Seattle 50
19 Percentage of Time Spent in Microenvironments by Health Group: Seattle 53
20 Arithmetic Mean PM2 5 Concentrations (u.g/m3)—All Studies 54
21 Arithmetic Mean PMio Concentrations (ng/m3)—All Studies 55
22 Variation of PM2 5 Infiltration Factor by Season—All Studies 56
23 Longitudinal Correlations Between Personal and Outdoor Air for Participants—All
Studies 56
vm
-------
Acknowledgements
The authors would like to thank Daniel Vallero, Laura Kildosher, John Creason, Debra Walsh,
Richard Kwok, Jose Sune, Janet Burke, Haluk Ozkaynak, Antonio Leathers, and Ellen Streib of
the U.S. Environmental Protection Agency (US EPA) for their administrative and technical
support throughout the scope of these studies. The authors also acknowledge the contributions
of Mike Hermann, Elizabeth Rodes, Randy Newman, Jeff Nichols, and Donald Whitaker
(Research Triangle Institute) for their contribution in support of the Baltimore, Fresno, and
Research Triangle Park panel studies. The authors thank Dr. Ademola Ejire (Shaw University,
Raleigh, NC) for his contributions regarding community outreach and Dr. William E. Sanders,
Jr. and Margaret Herbst (University of North Carolina-Chapel Hill, NC) for their efforts in
recruiting cardiac defibrillator participants in the Research Triangle Park studies. Researchers
Gaun Lau and Barbara Turpin (Rutgers University-Environmental and Occupational Health and
Safety Institute) and Barry Ryan and Czerne Reid (Emory University) are acknowledged for
their contribution to the panel studies performed by the Harvard School of Public Health.
Grateful acknowledgement is made to the hundreds of participants and supportive family
members who made these studies possible.
IX
-------
INTRODUCTION
In the 1990s, many studies around the world indicated that daily mortality increased with
increasing outdoor air pollution. Suspicion mainly fell on particles, although associated
co-pollutants (sulfates, SO2, NO2, CO, and ozone), which are often highly correlated with
particles, continue to be considered possible contributors. The findings led the US EPA
to propose a new National Ambient Air Quality Standard (NAAQS) for fine particles
(PM2 5). The United States Congress appropriated additional money to study the
problem, and the National Academy of Sciences (NAS) was asked to outline a research
program. The NAS report (NRC-NAS, 1998) envisioned a 13-year research "portfolio,"
the first three years of which would concentrate on measuring actual personal exposure,
particularly for high-risk subpopulations, to fine particles and associated co-pollutants.
Even before the NAS report was published, the National Exposure Research Laboratory
(NERL) at Research Triangle Park, NC had planned such a study and had already
published a Request for Applications. The NAS report and Congressional funding
increased the funds available for the study and made it possible to fund a series of studies
instead of the one originally planned. Three of these were to be carried out by University
consortia under co-operative agreements and the remainder by EPA. The studies began in
1999, and fieldwork has been completed in all but one of the studies.
Prior to these studies, which concentrated on persons living in their own homes and
apartments, the NERL carried out two studies in retirement homes in Baltimore, MD and
Fresno, CA. Last year, a report on the progress of the studies was published (US EPA
2002). That report provided the study designs and proposed measurement methods for all
of the studies. Because the measurement methods were fully described in last year's
report, these descriptions will not be repeated in this report. (An exception is the
description of a new multipollutant monitor developed by the Harvard University School
of Public Health. This description is found below in the Quality Assurance section.) Last
year's report also included summaries of PM25 and PMio mass concentrations for the
Baltimore and Fresno studies, and therefore these results will not be repeated in this
report. (Again one exception is a table in the Discussion section comparing results from
all cities.) This final report summarizes work on particles from studies in five cities:
Research Triangle Park, NC; Atlanta, GA; Boston, MA; Los Angeles, CA; and Seattle,
WA.
-------
STUDY DESIGN
All studies had a basic study design involving personal, indoor, and outdoor monitoring
of persons in certain high-risk subpopulations. All studies fielded after 1999 used the
same questionnaire, which was approved by the Office of Management and Budget
(OMB) in July of 1999. Within the basic design, each group was free to emphasize
different aspects, so that some variation in pollutant monitoring existed among the groups
and the various field studies. The major goal of all studies was to relate personal
exposure to outdoor concentrations, a crucial need to support regulation of outdoor
particles.
Because the epidemiology studies are time series studies, the relation of personal
exposure to outdoor concentrations must be calculated over a number of days for each
subject. Earlier studies had indicated that the minimum number of needed days would be
greater than 3 and less than 7. Therefore all studies followed participants for a minimum
of 7 days. All studies calculated longitudinal correlations of personal exposure to
outdoor concentrations. Several hypotheses were developed that could be tested in the
course of the studies. One hypothesis was that high-risk subpopulations would engage in
fewer dust-generating activities and would therefore have lower indoor air concentrations
and lower exposures than healthy cohorts. A second related hypothesis was that their
personal exposures would have stronger correlations with outdoor air than healthy
cohorts, due to fewer particle sources indoors. A brief description of the study design for
each city follows.
Research Triangle Park, NC
Following two earlier studies of retirement home residents in Baltimore and Fresno, fully
described in the first report in this series (US EPA 2002), the Research Triangle Institute
International, a not-for-profit research institute, was contracted to collect environmental
samples based upon a NERL study design (Ronald Williams, Project Officer). NERL
scientists analyzed the resulting field and laboratory data. This group studied two cohorts
in the Research Triangle Park, NC area: an African-American cohort with controlled
hypertension living in a low-moderate socioeconomic status (SES) neighborhood, and
persons with implanted cardiac defibrillators. Participants were non-smoking, 50+ years
of age, and living in their own homes. The participants were monitored for 7 consecutive
days during four consecutive calendar seasons. Besides personal particle monitors to
measure mass, participants also carried a real-time light-scattering device, the MIE pDR
(personal DataRAM®) to estimate short-term (1 minute average) particle exposures. Air
exchange rates in each home were also measured. The study design is shown in Table 1.
-------
Table 1. Study Design: Research Triangle Park
Study city
Panel description
Number of
participants
Seasons
(days/season)
PM 2.5 mass
PM 10 mass
PM nephelometer
PM number count
and distribution
EC-OC
NO2
03
CO
Elements (SO4)
Air Exchange
Health Measures
SE Raleigh
Low SES neighborhoods,
minorities with
controlled hypertension
27
4
(7)
P, I, 0, A
1,0, A
P,I
I, O select homes
P, I, 0, A
I, A
P,A
1,0, A
P, I, 0, A
PFT
PEF, FEV, pulse, 02 sat.
Chapel Hill
Cardiac Defibrillators
8
4
(7)
P, I, 0, A
1,0, A
P,I
I, O select homes
P, I, 0, A
I, A
P,A
1,0, A
P, I, 0, A
PFT
PEF, FEV, pulse,
O2 sat.
Monitor Types
—
~
PEM, HI, Dichot, FRM,
TEOM,
PEM, Dichot
MIE pDR
SMPS-APS
PEM
Ogawa badge
Ogawa badge, TECO
Draeger, TECO
PEM, HI, Dichot, FRM
PFT tubes
AirWatch, Nellcor N-20
P = personal, I = indoor residential, O = Outdoor residential, A= ambient, EC-OC = elemental and organic
carbon, PFT = perfluorotracer method, PEF = Peak expiratory flow, FEV = Forced expiratory volume, O2
sat= blood oxygen saturation. PEM = personal exposure monitor, HI = Harvard impactor, Dichot =
dichotomous sampler, FRM = Federal Reference Method, pDR = personal Data Ram, SMPS-APS =
Scanning Mobility Particle Sizer - Aerodynamic Particle Sizer. MIE, Ogawa, Airwatch, Draeger, TECO,
TEOM, Nellcor = manufacturers or trade names.
-------
Atlanta, Boston, and Los Angeles
The Harvard School of Public Health (Petros Koutrakis, Principle Investigator) studied
subpopulations in Atlanta, Boston, and Los Angeles. In each city, personal, indoor and
outdoor multi-pollutant samples were collected over seven-day periods in two seasons.
The high-risk subpopulations monitored included those with chronic obstructive
pulmonary disease (COPD) and those with heart disease, including myocardial
infarctions (MI). A healthy cohort was also monitored. The pollutants monitored
included personal, indoor, and outdoor PM2 5, sulfate, ozone, SC>2, NC>2, elemental carbon
and organic carbon (EC/OC). Personal PMio was also monitored in Los Angeles only.
The measurement methods included a multipollutant personal sampler developed at
Harvard and capable of sampling PM2.s, PMio, SO2 and sulfate (in Eastern cities), NO2
and nitrate (in Western cities), ozone, and EC/OC. Indoor and outdoor particles were
measured using Harvard Impactors (His) for both PM2 5 and PMio. The basic study
design is provided in Table 2.
Table 2. Study Design—Atlanta, Boston, Los Angeles
Atlanta
Boston
Los Angeles
TOTAL
Fall 1999
Spring 2000
Winter 1999-
2000
Summer 2000
Winter 1999-
2000
Summer 2000
COHORT
Healthy
N
--
--
8
6
~»
—
14
Days
~
~
7
7
—
—
COPD
N
15
13
4
5
15
14e
66
Days
T
7b
T
1
7d
7
MIf
N
9
9
11
10
—
—
39
Days
7
7
7
7
—
—
Target
Person-
days
168
154
159
147
103
98
829
f Boston MI cohort includes individuals with conjunctive heart failure, history of by-pass
surgery, and medication-treated angina
a One subject monitored for four days and another for six days.
b Three subjects monitored for only six days each.
c One subject monitored for 5 days.
d One subject monitored for 5 days.
e One subject was excluded, as the person was admitted to the hospital and spent little
time at home during the sampling period.
-------
Seattle
This city was studied by the University of Washington Dept. of Environmental and
Occupational Studies (L-J. Sally Liu, Principle Investigator). Four different cohorts were
studied: persons with chronic obstructive pulmonary disease (COPD), coronary heart
disease (CHD), children with asthma, and healthy elderly persons (Table 3). Participants
were monitored during two "seasons"—the heating season (October through February)
and the non-heating season. The pollutants monitored are summarized in Table 4.
Table 3. Study Design: Seattle (Subpopulations)
Year
1999
2000
2001
Starting
Date
Oct26
Nov8
Nov29
Jan 10
Feb7
Feb21
Mar 6
Mar 27
Apr 10
May 1
May 15
Jul 10
JulSl
Sep25
Octl6
Nov 6
Nov27
Dec 25
Jan 8
Jan 22
Feb5
Feb26
Mar 29
Apr 16
Apr 30
May 14
Asthmatics
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
2
4
5
3
2
4
5
3
4
1
33
CHD
0
0
0
0
1
0
1
0
0
0
0
1
1
2
3
6
4
1
1
3
3
4
3
2
1
3
40
COPD
5
5
5
3
2
3
3
4
3
5
4
4
1
4
5
-
-
-
-
-
-
-
-
-
-
-
56
Healthy
3
4
3
6
3
3
3
4
2
3
0
2
2
-
-
-
-
-
-
-
-
-
-
-
-
-
38
Total
8
9
8
9
6
6
7
8
5
8
4
7
4
6
8
6
6
5
6
6
5
8
8
5
5
4
167
Number of subjects by cohort and session. A total of 108 subjects were monitored, about
50% were monitored twice.
-------
Table 4. Study Design: Seattle (Measurement Methods)
Measurement
PM,0
PM25
Pump
PM,
WS/SVOC
EC/OC
Biomarker
CO
NO2/SO2
ACH
Continuous RH
Continuous T
Compliance
Time/activity
PEF/FEV,C
Pulse rate / O2
saturation
HRV and BP
Personal
-
HPEM (41pm)
BGI
pDR
PUF
HPEM
Urine sample
Breath sample
Ogawa
-
-
-
Motor on/off
Diary form
Airwatch
Pulse Oximeter
Holter
Indoor
HI (10 1pm)
HI (10 1pm)
Medo
Neph or pDR
PUF
HI& IOGAPS
-
Langan/bag
Ogawa
TelAir/PFT
Onset logger
Onset logger
-
-
-
-
Outdoor
HI (10 1pm)
HI (10 1pm)
Cast
Neph
PUF
HI & IOGAPS
-
-
Ogawa
TelAir
-
-
-
-
-
-
Central Site
HI (10 1pm) x 2
HI (10 1pm) x 2
Gast
Neph
PUF
HI & IOGAPS
-
-
Ogawa
-
-
-
-
-
-
-
HI = Harvard Impactor; HPEM = Harvard Personal Exposure Monitor; pDR = Thermo-
MIE personal DataRam; Neph = Radiance Reflectance nephelometer; WS = wood
smoke; SVOC = semi-volatile organic compounds; PUF = polyurethane foam; IOGAPS
= indoor-outdoor gaseous air pollution sampler; Langan = Langan Instruments, Inc. T15
CO Measurer; BGI, Medo, Gast, TelAir and Ogawa = manufacturers; PFT =
perfluorotracer method; PEF/FEVi = peak expiratory flow/forced expiratory flow (1
second); Airwatch = spirometer; HRV = heart rate variability; BP = blood pressure.
-------
QUALITY ASSURANCE
Since the personal, indoor, and residential outdoor monitors employed in these studies are
not EPA reference instruments, side-by-side comparisons with the EPA Federal
Reference Method (FRM) or with instruments that have previously been compared with
the FRM, such as the Harvard Impactor (HI) were carried out. The following text and
figures illustrate that the precision and accuracy of these monitors compared well with
EPA reference methods.
Research Triangle Park, NC
A major effort was made to compare all monitors used in this study and a journal article
focusing on monitor performance was produced (Williams et al., 2000b). In general,
precision was very good at levels of about 5%, and agreement with EPA reference
methods was also good. The personal and indoor monitors (HI) were collocated with the
EPA Federal Reference Method (FRM) monitors at the central site. The personal
monitor showed a slightly positive bias with an intercept of 4 |ig/m3, a slope of 0.95, and
an R2 of 86% (Figure 1). The indoor monitor agreed better with a negligible intercept, a
slope of 1.04 and an R2 of 97%. The regression of the personal monitor vs. the HI had an
intercept of 2.7 jag/m3, a slope of 0.95, and an R2 of 88%.
-------
50 -
E
o> 40 -
< 30 H
20 -
10
APMTW = 4.1352 + 0 9538*AFRTW
R2 = .86
AHMTW = 0 4824 + 1 0416*AFRTW
R2= 97
APMTW
AHMTW
10
20
30
40
50
PM^j (APRTW) iig/m3
Figure 1. Collocated personal (APMTW) and indoor (AHMTW) monitors
with EPA's Federal Reference Method (FRM) monitor at the Research
Triangle Park central site. Regression lines are compared to the one-to-one
line.
-------
Atlanta
During the fall of 1999, the SKC PEM fine particle personal monitor was run side by side
at the Atlanta central site against the HI, which has previously shown that it agrees well
with the EPA FRM. Agreement was poor, with an intercept of 6 fig/m3 and a low slope
of 0.78 (Figure 2).
Fall 1999 PM2.5 Collocation-Atlanta
50
in
cvi
40 -
30 -
Ill
Q.
O 20 -
10 -
SKC PEM = (0.78*HI) + 5.98
(r2D0.91)
1:1 Line
10
20 30
HI PM2.5
40
50
Figure 2: Wintertime comparison of the SKC Personal Exposure Monitor (PEM)
with the Harvard Impactor (HI) at the Atlanta central site.
-------
During the spring of 2000, the SKC PEM was replaced by the Harvard PEM, and
agreement with the HI at the central site was considerably improved, with a negligible
intercept, a slope of 1.06 and an R2 value of 84% (Figure 3).
Spring 2000 PM2.5 Collocation-Atlanta
40
30 -
in
LU
OL
I
20 -
10 -
HPEM= (1.06* HI)-0.28
(r 2D0.84)
1:1 Line
10
20
HI PM2.5
30
40
Figure 3. Springtime comparison of the Harvard Personal Exposure Monitor
(HPEM) with.the Harvard Impactor (HI) at the Atlanta central site.
10
-------
Boston
The Harvard PEM (HPEM) fine particle personal monitor was run side by side at the
Boston central site against the Harvard Impactor (HI). Agreement was good, with a slope
of 1.13 and an R2 value of 95% (Figure 4).
30 -
ro 20 -
in
eg
LJJ
a.
I
10 -
HPEM=1.13*H 1-1.77
(R2=0.95)
• Greased HPEM (Central site)
1:1
10
20
30
HI
2.5
Figure 4. The Harvard PEM (HPEM2.s) vs. the Harvard Impactor (HI2.5) when
collocated at the Boston central site.
11
-------
Los Angeles
Following the Atlanta and Boston studies referenced above, the Harvard University
School of Public Health developed an integrated multi-pollutant monitor to measure the
simultaneous paniculate and gaseous exposures for 22 COPD patients in Los Angeles.
This sampler is essentially several individual samplers that have been joined together to
form a simple, compact, and relatively lightweight personal monitor (Figure 5). The
entire monitor (plus pump and battery pack) weighs approximately six pounds.
Participants were able to wear the monitor successfully throughout the monitoring period.
Participants were allowed to remove the monitor and place it nearby when they would be
stationary for long periods of time, such as when they were sleeping or reading. For
participants using oxygen, monitors were strapped to their oxygen tanks to ease the
sampling burden.
For indoor, outdoor home, and central site monitoring, the monitors were placed on a
tripod, with the inlets approximately one meter above the ground. Both indoor and
outdoor monitors were placed away from any objects (e.g., trees, houses, vents) to
minimize interference with pollutant measurements. Outdoor monitors were placed
under a rain cap to protect the samplers from precipitation. For personal monitoring, the
monitor was attached by Velcro to the shoulder strap of a padded backpack at breathing
level. If the participant was mobility-restricted or otherwise hampered, the samplers were
attached to fixed objects near the participant's body, with the inlet protruding into the
breathing zone.
PM2 5 and PM10 PEMs
Nitrate Mini-Sampler
EC/OC Mini-Sampler
O3, SO2/NO2 Samplers
Figure 5. Multi-Pollutant Sampler developed by Harvard University School of
Public Health.
The multi-pollutant sampler measured PM,0 and PM2 5 concentrations using PEMs, small
inertial impactors designed specifically for personal and micro-environmental monitoring
12
-------
(Marple et al, 1987; Thomas et al, 1993; Chang et al, 1999; Demokritou et al., 2001a).
Impactor plates in all samplers were greased to minimize particle bounce (Demokritou et
al., 200la; Demokritou et al., 200Ib). In both seasons, indoor and outdoor PMio and
PM2.5 measurements were made using Harvard PEMs operated at flow rates of 4 LPM.
In the winter, personal PMio and PM2 5 concentrations were measured using PEMs
manufactured by SKC. Since these samplers were designed to operate at flow rates of 4
LPM, the SKC PEMs were modified to allow their use at flow rates of 1.6 and 2 LPM for
PMio and PM2 5 sampling, respectively (Rojas-Bracho et al., 2000). Because the cut-
point of the impactors are a function of flow rate, the number of nozzle holes was
reduced from ten to four for the PMio PEM and to five for the PM2.5 PEM to maintain the
same size cut-offs as originally designed. In the summer, personal PMio and PM2.5
concentrations were measured using PEMs designed by Harvard to operate at 1.8 LPM,
since these samplers were lighter, could be used without modification, and would be
comparable to the Harvard PEM samplers used to sample indoors and outdoors.
Both the SKC and Harvard PEMs used Teflon filters as the particle collection media and
included drain disk rings to prevent metal contamination for future ICP-MS analysis.
The PM2.5 and PMio PEMs were attached to either side of the monitor using a 10 cm long
elutriator (Figure 5). Nitrate and EC/OC mini-samplers were attached to the front of the
elutriator using clips. The passive O? and SO2/NO2 badges were placed in the side of the
elutriator, with their face exposed to the sample air stream to allow for constant sampler
collection rates.
Collocated PEMs and His at the Los Angeles central site showed reasonable agreement
with a slope of 0.95 and an R2 value of 90% (Figure 6). However, the intercept of 3
(j.g/m3 was a bit higher than could be hoped for. During the second season of the Los
Angeles study, PMio and PM2.5 personal samples were collected simultaneously. The
PMio PEM was collocated at the L.A. central site with an HIio, with reasonable
agreement (R2 = 0.92, slope = 1.22).
PM
90
80
70
60
50
40 -
30 -
20
10
0
10
y=1.22x
R2 = 0.92
0 10 20 30 40 50 60 70 80 90
HI
50
40
30
20
10
0
PM
-2.5
= 0.95x + 2.99
R2 = 0.90
10
20 30
HI
40
50
Figure 6. Collocated PEMs and His at the Los Angeles central site: winter
and winter and summer PM2.s Measurements
13
-------
Seattle
The limit of detection (LOD) for the 24-h integrated HI was 1 ng/m3 and for the 24-h
integrated HPEV^.s was 6.2 (ig/m3 for the first 4 sessions and was reduced to 4.5 ng/m3
afterwards. The improvement was achieved by replacing the oiled porous impaction
plate with vacuum grease to reduce contamination from silicon oil and by adding a drain
disc downstream of the Teflon filter. All duplicates were highly correlated with each
other, with a Pearson's r of 0.96 or higher. The mean difference between the duplicates
was not significantly different from zero. The precision, calculated as the standard
deviation of duplicate differences divided by V2, was 1.2 ug/m3 for HI and 2.2 jug/m3 for
HPEM2.5.
The accuracy of the PMa.s measurements was calculated by comparing with the
collocated FRM2 5 measurements at the central site. The investigators also collocated
HIa 5 and HPEM2 5 whenever possible: 77 pairs at the stationary ambient monitoring sites
and 17 pairs at subjects' homes (Figure 7). The Pearson's r between samplers was 0.93 or
greater. There was a positive bias, 7.7 ug/m3 (pO.OOl), for HPEN^.s with an oiled
impaction plate. For HPEM2 5 with a greased impaction plate, the bias was negligible
(0.4 jig/m3, p=0.08). All HI and HPEM measurements were corrected for average blank
values. The HPEM2 5 measurements with oiled impaction plates during the first four
monitoring sessions (N=269 out of 1347 personal filters) were removed from analysis
due to the oil contamination problem.
14
-------
CM
I
LU
Q.
X
32
30
28
26
24
22
20
18
16
14
12
10 -
8 -
6 -
4 -
2 -
0
HPEM=1.64+0.88*FRM
(R2=0.87)
HI=0.09+0.97*FRM
(R2=0.97)
A
•
FRM
FRM
vs
vs
HPEM
HI2.5
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
FRM2 5 (|jg/m3)
Figure 7. Comparison of Harvard impactor (HI), Harvard personal environmental
monitor (HPEM), and federal reference method (FRM) measurements for PM2.s:
Seattle (central sites).
15
-------
RESULTS
Research Triangle Park, NC
No significant differences in mean cohort exposures were observed between either of the
two cohorts, their residences, locality or for the four seasons. Therefore all results were
combined (Table 5). Mean personal PM2.s exposures (23.0 |ig/m3) were approximately 4
|ag/m3 greater than the comparable indoor and outdoor PM2.s concentrations. No
differences were observed between the mean indoor, outdoor residence, and central site
PM2 5 concentrations. Mean outdoor PMio (-30-31 |ag/m3) was slightly higher than the
indoor PMio concentrations (28 |ig/m3), and 8-12 fig/m3 higher than the corresponding
PM2 5 concentrations.
Table 5. PM Mass Concentrations (jig/m3): RTF
Variable
"Personal PM2 5
blndoor PM2.5
bOutdoor PM2 5
bCentral site PM2.5
alndoor PMio
aOutdoor PMio
aCentral site PMio
°Indoor PMio-25
°Outdoor PMio-2.5
cCentral site PMi0.2.5
N
712
761
761
746
761
761
752
761
761
210
Geo Mean
19.2
15.3
17.5
17.3
23.2
27.5
27.9
6.3
8.5
8.6
Mean
23.0
19.1
19.3
19.2
27.7
30.4
31.4
8.6
11.1
10.0
Min
3.4
2.3
5.0
5.0
4.4
7.9
4.8
-2.8
-2.8
2.6
Max
142.3
119.4
51.6
49.5
155.7
105.1
105.0
116.6
82.2
32.1
RSD
70.1
80.1
43.7
44.9
70.6
46.4
51.5
111.8
86.9
62.3
"Measured using PEMs, bmeasured using HI samplers, "measured by difference in PEM PM,0 monitor and
collocated HI PM2 s mass concentrations. Negative values associated with the minimum PM,0.2 5
measurements are suspected to be the result of methodological differences between the two monitoring
methods relative to semi-volatile retention and particle cut-off parameters.
The cumulative distributions for the four types of PM2 5 samples (personal, indoor,
residential outdoor, and central site) are displayed by season and for all seasons combined
in Figure 8. The plotted points are the 1st, 5th, 10th, 25th, 50th, 75th, 90th, 95th, and 99th
percentiles.
16
-------
100
I 10
• Personal
Indoor
Outdoor Residential
- Outdoor Central Site
-2
-1
0
Z-score
Fig. 8a. Cumulative PM2.s distributions: RTF Summer 2000 (N = 206-224)
100
• Personal
- Indoor
- Outdoor Residential
• Outdoor Central Site
-3
-2
Fig. 8b. Cumulative PM2.5 distributions: RTF Fall 2000 (N = 204-210).
17
-------
100
I
10
-*- Personal
— Indoor
-*- Outdoor Residential
—— Outdoor Central Site
-3
-2
-1 0
Z-score
Fig. 8c. Cumulative PM2.s distributions: RTF Winter 2001 (N = 182).
100
Personal
Indoor
Outdoor Residential
Outdoor Central Site
Fig. 8d. Cumulative PM2.S distributions: RTF Spring 2001 (N = 132-147).
18
-------
1UU
I io
-3
-*-Personal
-*- Indoor
-*- Outdoor Residential
^~ Outdoor Central Site
-2
-1
0
Z-score
Fig. 8e. Cumulative PM2.S distributions: RTF AH Seasons (N = 735-763).
19
-------
The cumulative distributions for the three types of PM]0 samples (indoor, residential
outdoor, central site) are compared to their counterpart PMa.s samples in Figure 9.
1000
100-
0)
10-
•Indoor PM10
Outdoor Residential PM10
Outdoor Central Site PM10
-Indoor PM2.5
Outdoor Residential PM2.5
Outdoor Central Site PM2.5
0
Z-score
Fig. 9. Comparison of PM10 and PM2.5 combined over all seasons: RTF.
20
-------
Outdoor temperatures are important influences on behavior patterns that can influence air
exchange rates and infiltration factors. Figure 10 displays the daily temperature
variations over the course of the RTF study.
Daily Temperature (°F) at Ambient Site
80 -
160-
40 -
20
' ' ' *~T ' ' ' T"1 ' ' T1 ' ' """r ' ' ' 'i ' ' ' T ' ' ' i ' ' ' r1 ' ' rrt ' ' ""i ' ' ' r~" '
Jul Aug Sep Oct Ncv Dec Jan Feb Mar Apr May
2000-2001
Figure 10. Daily average temperatures (°F) during field study: RTF.
21
-------
Air exchange rates were surprisingly high during the winter and low during the summer,
the reverse of what has been observed in other studies in temperate zone cities (Table 6).
The unusually mild winter, with few daily average temperatures below freezing, may
have contributed to the high and highly variable air exchange rates that season.
Table 6. Air exchange rates by season: RTF
Season
Summer 2000
Fall 2000
Winter 2001
Spring 2001
N
34
33
28
34
Mean (h^)
0.48
0.63
1.07
0.70
SD (h'1)
0.51
0.36
0.81
0.40
Particle concentrations showed little seasonal variation (Figure 11). Indoor
concentrations, although their mean values were lower than outdoor for both size
fractions, had a larger number of high values in every season.
140 -
60 -
season Summer 2000
season Fall 2000
season: Winter 2001
- 140
- 60
season Sprtnq 2001
Method
Figure 11. Boxplots of the PM™ and PM2.5 concentrations (ug/m3): RTF.
A=Ambient (central site); I = indoor; O = Outdoor (backyard); P = Personal.
22
-------
Infiltration factors were calculated for each home (Figure 12). Again no significant
seasonal difference was noted. The number of unphysical values (<0 or >1) occurring in
every season suggests difficulties in the mass balance model employed.
2-
-1 -
-3
Fall 2000 Spring 2001 Summer 2000 Winter 2001
Season
Figure 12. Boxplots of the PM2.5 infiltration factor by season: RTF.
Several calculations using linear regression were performed to estimate the contribution
of outdoor particles to indoor concentrations and personal exposures (Table 7). All
approaches agreed in finding an infiltration factor Finf ranging between 0.40 and 0.45.
The personal exposure attenuation factor Fpex ranged between 0.54 and 0.58 in all
calculations except one, in which it was 0.47. The average PM2 5 indoor air concentration
due to indoor sources C;g (indoor-generated) ranged between 9 and 10 ug/m3, while the
average personal exposure due to indoor sources and personal activities Ena (non-
ambient) ranged between 12 and 15 ug/m3. Since the mean indoor PlVb.s concentration
was 19 ug/m , the contribution of indoor sources (9-10 ug/m3) was close to half the total.
The contribution of indoor sources and personal activities to the mean personal exposure
of 23 ug/m3 was slightly more than half the total.
23
-------
Table 7. Maximum Likelihood Estimates of Indoor and Outdoor Source
Contributions to Indoor and Personal PIVh.s: RTF
Mixed Model Specifications Indoor a(± SE)
1 . Single fixed intercept
Single fixed slope
2. Multiple fixed intercepts
Single fixed slope
3. Random intercepts
Single fixed slope
4. Single fixed intercept
Multiple fixed slopes
5. Multiple fixed intercepts
Multiple fixed slopes
6. Single fixed intercept
Random slopes
7. Random intercepts
Random slopes
Clg= 10.34(±1.32)
Finf=0.40(±0.06)
9.02(±1.24)
0.43(± 0.06)
9.18(±1.59)
0.45(±0.06)
9.32(±1.16)
0.45(±0.06)
10.19(±1.64)
0.42(± 0.06)
9.73(± 1.22)
0.42(± 0.08)
9.18(± 1.68)
0.45(± 0.06)
Personalb(± SE}
Ena= 11.82(±1.52)
FPex= 0.58(±0.07)
13.59(±1.62)
0.54(± 0.07)
12.64(± 1.78)
0.56(± 0.07)
12.47(±1.35)
0.59(±0.08)
15.16(±2.05)
0.47(±0.07)
12.27(± 1.40)
0.56(±0 .08)
12.59(±1.64)
0.56(±0.07)
" Subject specific indoor model: Indoor PM2 5 = C,g + Finf Outdoor PM2 5 + Residual,
where Indoor PM2.5 = daily indoor HI measurements (ug/m3), and Outdoor PM2.5 =
outdoor backyard HI measurements (ug/m3).
b Subject specific personal model: Personal PM2 5 = Ena + Fpex* Outdoor PM2 5 +
Residual, where Personal PM2 5 = personal Marple sampler measurements (ug/m3).
The house-by-house estimates of the infiltration factor Fmf, indoor-generated
concentration Clg, and personal exposure attenuation factor Fpex are provided in Table 8
24
-------
Table 8. Least squares estimate of indoor filtration and personal exposure factors
Subject
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
26
27
28
29
31
32
33
34
35
36
37
38
Mean
SD
CiE
8.7
8.5
7.0
20.0
4.2
9.3
-3.2
23.9
10.2
2.7
35.3
9.3
1.8
6.1
17.9
2.5
-0.8
6.0
6.8
2.2
23.5
4.9
36.7
36.0
8.5
30.9
4.8
-0.8
4.0
10.1
6.0
7.3
17.6
6.4
6.0
0.9
10.6
10.7
Finf
0.38
0.33
0.6
0.87
0.74
0.43
0.99
-0.22
0.24
0.42
-0.26
0.22
0.51
0.31
0.02
0.47
0.81
0.51
0.64
0.45
0.45
0.97
-0.55
0.17
0.53
0.21
0.36
1.62
0.69
0.2
0.34
0.43
0.22
0.11
0.32
0.53
0.42
0.38
R2
0.14
0.07
0.24
0.06
0.28
0.19
0.59
0.15
0.09
0.74
0.02
0.13
0.46
0.11
0
0.78
0.29
0.33
0.59
0.89
0.21
0.15
0.04
0
0.29
0
0.28
0.6
0.78
0.04
0.12
0.15
0.02
0.25
0.28
0.89
0.28
0.27
Eig
8.3
4.7
21.3
19.6
8.0
7.6
-5.4
29.2
16.2
8.2
44.0
10.2
19.1
6.7
15.6
9.4
10.8
39.4
6.0
6.3
37.1
13.0
21.5
25.3
23.0
14.8
6.1
17.1
4.7
11.4
8.1
8.1
10.2
13.0
19.4
4.7
14.5
10.5
Fpex
0.92
0.65
0.62
0.85
1.1
0.74
1.27
-0.22
0.18
0.29
-0.37
0.24
-0.42
0.34
0.27
0.51
0.62
-0.36
0.69
0.51
0.28
0.78
0.2
0.71
0.47
0.47
0.63
0.08
0.78
0.3
0.47
0.62
0.79
0.21
0
0.78
0.44
0.40
R2
0.24
0.22
0.05
0.07
0.11
0.07
0.43
0.07
0.03
0.16
0.02
0.17
0.06
0.15
0.02
0.22
0.06
0.02
0.51
0.41
0.07
0.1
0.02
0.06
0.14
0.1
0.25
0.01
0.62
0.04
0.13
0.12
0.2
0.1
0
0.53
0.16
0.16
The first model is defined by indoor PM2 5 = C,s + Fkf (outdoor PMj.5), where Q = concentration of indoor generated PM2.S
(regression intercept), and Fto/ = infiltration of ambient PM2 s (regression slope). Ci, and E* have units of fig m°. R2 =
coefficient of determination (square of correlation between indoor and outdoor PMj.5). The second model is defined by
personal PMu = Eit + Fpa, (outdoor PM2.5), where Eit =, exposure to indoor generated PMis (regression intercept), and Fpo =
ambient PM2 5 contribution to personal exposure (regression slope). R2 = coefficient of determination (square of correlation
between personal and outdoor PM2 5)
25
-------
A nonlinear mass-balance model was applied to the 24-h average gravimetric data
collected by the indoor and outdoor Harvard impactors. The model is (Ozkaynak et al.,
1996):
Cm = PaCoul/(a+k) +fCOOkSc0oi/[(a+k)VJ + folherSother/[(a+k)V]
where
Cm = indoor concentration (jJ.g/m )
Cou, = outdoor concentration (u.g/m3)
P = penetration coefficient (dimensionless)
a = air change rate (h"1)
k = deposition rate (h"1)
fcook = fraction of time cooking (dimensionless)
father = fraction of time not cooking (dimensionless)
V = volume of house (m3)
Scook = source strength (mass flux) from cooking (|ag h"1)
Soiher = source strength (mass flux) from all other indoor sources (jag h"1)
Measured quantities include Cin, Cout, a, and V, while the time spent cooking was
obtained from the time-activity diaries. The four unknowns (P, k, SCOOk, S0ther) were
obtained from the SAS NLIN procedure. P was constrained to be between 0 and 1; k,
Scook, S0,her were constrained to be >0.
Thirty-six homes of 38 measured had sufficient data (up to 28 days) to carry out the
calculations. The results are shown in Table 9. In this table, CCOOk, the average
concentration produced by cooking during the time of cooking, is obtained from the 2nd
term on the right-hand side of the equation above (the ratio fCOokSCook/[(a+k)V], and C0ther,
the average concentration due to all other indoor sources, is obtained from the third term
in the equation above, the ratio f0,herS0iher/[(a+k)V].
Table 9 shows considerable scatter. For example, six values of A: are at the lower limit of
0, and 14 values of P are at the upper limit of 1. This indicates a high degree of error,
either measurement error, errors in entering the cooking times on the time-activity diary,
or errors in the assumption of the model. However, many of the individual estimates for
the homes appear to be in reasonable agreement with what is known or guessed to be the
case. Few previous studies have been able to arrive at individual estimates for the
penetration coefficient and deposition rates in individual homes.
When all data are combined, the overall nonlinear estimate for the penetration coefficient
P is 0.99. This is likely to be an underestimate, since P was constrained from above by 1;
values greater than 1 would have increased the average value. In fact, the regression was
run again with P unconstrained and this resulted in a value of P averaged across the 36
homes of 1.17. Even so, this value of P is in good agreement with the value of P obtained
from the University of Washington approach (0.97) and the PTEAM result (1). However,
26
-------
if the average of the estimated penetration coefficient is calculated for the homes
individually, the estimate of P drops to 0.76. This is an underestimate, since 14 values of
P are equal to 1 and some would have been greater than 1 if unconstrained.
The estimates of the concentration due to cooking (while cooking) range from zero to
16.7 (ig/m3, with an average of 6.94 u.g/m3. However, since the fraction of time cooking
was only 4%, the average concentration over 24 hours due to cooking was only 0.3
u.g/m3. The estimates of the concentration due to other indoor activities range from zero
to 16.5 ng/m3, with an average of 5.56 ng/m3, somewhat less than the PTEAM estimate.
This leads to an estimate of the fraction of indoor air concentrations due to outdoor air of
0.81. (The PTEAM estimate was 0.71). Finally, the estimates for the deposition rate k
vary from 0 to 1, with an average rate of 0.25 h"1. This average is less than the estimated
PTEAM average of 0.39 h"1. Also, it is likely to be an overestimate, since A: was
constrained from below by zero. However, some theoretical estimates of deposition rates
support a lower value (Lai and Nazaroff, 2001).
27
-------
Table 9. Nonlinear Model Estimates of the Deposition Rate k, Penetration Coefficient P, and Indoor Source Terms
Subject Days
F,nf
C out
C cook C other residual C PRED C in
'cook
'other
K)
oo
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
26
27
28
31
21
21
22
20
20
26
25
6
21
21
9
27
7
26
25
28
18
19
21
20
24
14
13
27
17
26
24
14
0.36
0.56
0.81
0.78
0.67
0.39
0.88
0.17
0.23
0.45
0.63
0.08
0.44
0.36
0.57
0.42
0.39
0.65
0.69
0.43
0.69
1.00
0.68
0.79
0.48
1.00
0.40
0.66
17.67
22.88
24.90
22.77
19.93
19.15
23.96
27.58
19.78
17.30
20.92
18.66
13.35
16.59
15.88
16.44
18.19
20.01
20.41
17.98
21.72
19.45
21.01
18.13
26.23
19.15
18.83
25.60
0.00
0.62
0.19
0.07
0.14
0.00
0.05
0.01
0.48
0.18
0.18
0.46
0.28
0.18
0.18
0.18
0.00
0.34
0.47
0.05
0.21
0.00
0.34
0.11
0.80
0.00
0.18
0.30
0.36
1.00
1.00
1.00
0.88
0.39
0.92
0.18
0.46
0.61
0.80
0.17
1.00
0.55
0.77
0.52
0.39
1.00
0.98
0.50
1.00
1.00
1.00
1.00
1.00
1.00
0.56
0.78
0.84
0.91
0.97
0.27
0.59
0.56
1.31
0.25
0.54
0.59
1.71
0.45
0.23
0.36
0.53
0.89
0.20
1.00
1.23
0.36
0.59
0.39
0.79
0.45
0.86
1.41
0.52
2.18
6.85
0.00
10.67
7.10
7.81
11.36
0.00
7.16
10.52
8.39
0.00
11.18
1.03
10.63
5.10
4.03
7.61
11.57
10.56
1.17
10.87
11.12
5.38
11.16
16.67
7.83
8.73
0.00
6.85
5.30
0.68
7.10
3.84
5.38
0.00
3.00
10.52
1.37
12.40
10.45
1.22
2.55
5.10
3.02
0.33
2.52
9.31
0.88
10.87
0.19
5.38
11.16
16.48
7.83
2.52
10.42
0.81
0.51
0.47
2.54
0.04
-0.03
-0.38
0.60
0.19
0.02
2.14
-0.01
0.00
0.02
1.23
-0.05
-0.11
0.38
0.09
-0.02
3.00
1.66
6.08
3.62
0.63
7.11
0.16
0.13
15.55
16.27
21.04
39.30
19.84
17.75
21.10
17.15
15.49
10.11
27.75
13.51
8.53
11.27
16.37
10.26
11.89
15.97
19.77
10.32
30.29
21.99
18.94
35.45
22.94
27.22
11.63
21.39
16.35
16.78
21.50
41.83
19.88
17.72
20.73
17.75
15.68
10.13
29.89
13.50
8.53
11.28
17.60
10.21
11.78
16.35
19.86
10.30
33.29
23.65
25.02
39.06
23.57
34.32
11.79
21.52
0.05
0.05
0.08
0.07
0.02
0.04
0.03
0.04
0.01
0.04
0.02
0.02
0.00
0.03
0.04
0.06
0.06
0.05
0.07
0.04
0.02
0.07
0.06
0.06
0.04
0.01
0.02
0.04
0.95
0.95
0.92
0.93
0.98
0.96
0.97
0.96
0.99
0.96
0.98
0.98
1.00
0.97
0.96
0.94
0.94
0.95
0.93
0.96
0.98
0.93
0.94
0.94
0.96
0.99
0.98
0.96
-------
Table 9. Continued
Subject Days Flnl Cjout k P a C_cook Cjother residual C_PRED CJn feook f0o,er
32
33
34
35
36
37
26
20
12
24
5
19
0.26
0.41
0.52
0.47
0.16
0.31
16.48
17.53
25.88
17.14
26.59
18.18
0.72
0.30
0.37
0.53
1.00
0.41
0.95
0.70
1.00
1.00
0.79
0.62
0.28
0.54
0.41
0.50
0.26
0.46
0.00
10.42
8.01
13.59
1.02
4.94
9.21
0.94
3.20
13.59
6.57
5.21
0.02
0.14
0.07
0.87
0.02
0.04
13.33
9.82
18.15
21.49
9.52
11.88
13.35
9.97
18.22
22.35
9.54
11.91
0.01
0.09
0.07
0.10
0.00
0.05
0.99
0.91
0.93
0.90
1.00
0.95
38 14 0.53 12.73 0.04 0.61 0.39 3.40 0.31 -0.01 7.69 7.68 0.00 1.00
Mean 0.54 19.91 0.26 0.76 0.68 6.94 5.56 0.90 18.04 18.94 0.04 0.96
SD 0.24 3.66 0.25 0.26 0.44 4.50 4.37 1.68 7.62 8.68 0.03 0.03
to
-------
The nonlinear model provided estimates of the infiltration factor Finf, the deposition rate k, the
penetration coefficient P and the fraction of indoor air particle concentrations produced by
outdoor air infiltration (frac out). Linear regression provided estimates of Finf and frac out.
Parameters P and k were estimated with an iterative procedure by first allowing values of P and k
each to increment between 0.1 and 1.0 by 0.1. All possible combinations ofP and k were then
combined with each 24-h measurement of a to produce a calculated infiltration factor Fcaic-
Calculated infiltration factors were then matched with weekly estimates of F,n/obtained from the
mixed-model where multiple slopes and intercepts were estimated for each subject by season and
assuming an autoregressive variance-covariance structure. Values of P, k and a were retained for
calculated values Fcaic that differed from estimates of F,«/by no more than ±0.01.
The nonlinear model was run for each house separately with P unbounded and also with P
bounded from above at 1. The model was also run by combining all data (N = 708 days). The
linear model, as mentioned above, was run for each house separately and also for all data
combined. The results (Table 10) indicate that the two models disagree. In particular, the linear
model predicts a much smaller effect of outdoor air (44-46% of the total indoor concentration)
than the nonlinear model (63-81%). Similar disagreements between models have been observed
by collaborating researchers at the University of Washington and Harvard School of Public
Health on data from their respective studies. At present, the reasons for the discrepancies are
unknown.
Table 10. Comparison of Linear and Nonlinear Model Estimates of Parameters in the
Mass Balance Equation: RTF
Model Site N Finf P k frac out
Nonlinear; P<1 RTP 36 0.54 0.76 0.26 0.63
NL; P unbounded RTP 36 0.73 1.17 0.36 0.68
NL, overall RTP 708 0.73 0.99 0.18 0.81
Linear; by house RTP 36 0.44 0.72 0.42 0.44
Linear; overall RTP 761 0.40 0.75 0.51 0.46
For the RTP cohorts, regression and correlation analyses indicated that, although there were
significant associations between ambient concentrations of PM2 5 and the gaseous co-pollutants,
personal PM2 5 exposures were only significantly related to ambient O3. Personal exposures and
indoor concentrations of the pollutants were not associated with their corresponding ambient
concentrations, except for
X-ray fluorescence (XRF) studies of the personal, indoor, and outdoor PM2 5 samples collected
on Marple PEMs indicated that sulfates made up the bulk of the mass (35-46%). Measurements
on co-located indoor and outdoor monitors with quartz filters followed by thermal quantitation of
elemental and organic (EC/OC) indicated that total carbon made up the bulk of the remainder of
the outdoor mass (30-34%). A small amount of crustal material, nitrate, and sodium chloride
30
-------
accounted for the remainder of the mass. However, a positive artifact was observed for the
indoor samples and data uncertainty for the OC component has to be considered.
Finally, the ultimate goal of these studies was to determine the longitudinal correlation between
personal exposures and central site measurements. Because no particular seasonal component
was noticed for the 37 subjects, their four 7-day sampling periods were combined into a single
28-day sequence and Pearson correlation coefficients between exposure and ambient
concentration were determined for each subject. Because the cohorts showed some differences,
they are kept separate in the summary boxplot of correlations (Figure 13). Median correlation
coefficients are quite low at 0.45 and 0.3, suggesting that for these persons, only 10-20% of the
variance in personal exposures could be explained by ambient concentrations. For some persons,
correlations were quite high, but for others they were not only very low, but even negative.
Longitudinal correlation (Pearson's r) between personal and central site PM,5 for each
subject by cohort.
1.0-
;§ 0.5
n
"3
e
U
•e
3
!
o.o -
-0.5 -
-1.0
ICD Low
Cohort
Figure 13: Boxplots of longitudinal correlation coefficients (Pearson) between personal
exposure and ambient concentration of PM2.5: RTF, NC.
31
-------
Atlanta
Summary statistics for personal, indoor, and outdoor PM2 5, elemental carbon (EC), and
paniculate sulfate (SCU ") stratified by season and sample type are presented in Table 11. For all
pollutant measures, outdoor concentrations tended to be higher than corresponding indoor and
personal levels, with differences generally most pronounced for SC>42~, reflecting the fact that its
major sources are located outdoors. Indoor and outdoor concentrations were higher in the spring
as compared to fall for each of the measured pollutants. In contrast, spring and fall personal
exposures tended to be comparable, with no consistent pattern observed across the three
measured pollutants (Figure 14).
32
-------
Table 11. Descriptive Statistics for Outdoor, Indoor, and Personal Pollutant
Concentrations in Atlanta: Particles (in
Pollutant/
Season/Sample
Fall PM2.5
Outdoor
Indoor
Personal
Spring.PM2.s
Outdoor
Indoor
Personal
Fall EC
Outdoor
Indoor
Personal
Spring EC
Outdoor
Indoor
Personal
Fall SO42"
Outdoor
Indoor
Personal
Spring SO42"
Outdoor
Indoor
Personal
N
141
143
141
118
127
142
141
152
146
134
139
138
102
112
131
137
144
135
Mean + Std. Dev.
14.7 ±7.7
14.3 ± 10.8
16.6 ±9.8
22.0 ±10.8
19.9 ±14.2
15.0 ±7.5
1.5 ±1.0
1.0 ±0.7
1.5 ±0.8
1.8 ±0.8
1.7 ±0.7
1.7 ±0.6
4.0 ± 2.2
2.5 ±1.4
2.7 ±1.6
5.4 ±2.9
3.4 ±1.9
2.8 ± 1.6
Median
13.1
12.2
14.3
20.8
16.0
14.2
1.2
0.9
1.3
1.7
1.5
1.7
3.9
2.2
2.5
4.8
3.0
2.5
Max
46.7
89.3
77.2
53.8
99.1
45.8
5.9
6.5
4.6
5.2
5.1
4.2
10.5
8.7
9.3
14.1
10.7
9.3
33
-------
Distribution of PM2.5 measurements by location and season
Atlanta fall and spring
50
40 -
1 30
ID
ON
20 -
10 -
-T- 95%
5%
• • • *
N=141 N=141 N=143 N=126 N=141 N=117 N=30 N=56
fall spr
Personal
fall spr
Indoor
fall spr
Outdoor
fall spr
Central site
Sample Location and Season
rth
Figure 14. PM2.s concentrations by location and season-Atlanta. Boxplots indicate 5* ,
25* , 50* (median), 75th and 95th percentiles by horizontal lines, with the mean indicated by
the dashed line and outliers indicated by dots.
34
-------
The cumulative PM2.5 distributions over all seasons are compared in a lognormal probability
graph in Figure 15. On such graphs, lognormal distributions appear as a straight line. Personal,
indoor, and outdoor residential measurements are very similar from the 1st to the 95th percentiles.
The outdoor central site records higher concentrations than the outdoor residential sites for the
lower half of the distribution, suggesting that it is impacted by sources more strongly.
100
I
10 -
«- Personal
» Indoor
*- Outdoor Residential
— Outdoor Central Site
-3
-2
Figure 15. Atlanta. The four types of PMz.s measurements plotted on lognormal
probability coordinates. Plotted points are the 1st, 5th, 10th, 25th, 50th, 75 ,90th, 95th, and
99th percentiles. N = 258-282.
35
-------
Concentrations of elemental carbon (EC), an indicator of diesel traffic, were increased in the
spring compared to the fall for all types of measurements (personal, indoor, outdoor) (Figure 16).
Sulfate concentrations, on the other hand, were elevated in the spring for the indoor and outdoor
measurements, but were nearly unaffected by season for the personal measurements (Figure 17).
3 -
E
^)
I2
c
a)
o
c
o
O
1 -
O
O
Fall Spring
Personal
Fall Spring
Indoor
Fall Spring
Outdoor
Figure 16. EC Concentrations by location and season: Atlanta.
36
-------
\4" Concentrations by location and season: Atlanta
37
-------
Summary statistics for personal, indoor, and outdoor ozone (Ch), nitrogen dioxide (NO:), and
sulfur dioxide (SO2) stratified by season and sample type are presented in Table 12. During both
seasons, outdoor concentrations for each of the gases tended to be higher than their respective
indoor concentrations and personal exposures. Outdoor gaseous pollutant concentrations were
comparable during both sampling seasons.
Table 12. Descriptive Statistics for Outdoor, Indoor, and Personal Pollutant
Concentrations in Atlanta: Gases (in ppb).
Pollutant
Season/Sample Type N Mean + Std. Dev Median Max Min
Fall O3
Outdoor
Indoor
Personal
Spring O3
Outdoor
Indoor
Personal
Fall NO,
Outdoor
Indoor
Personal
Spring NO?
Outdoor
Indoor
Personal
Fall SO2
Outdoor
Indoor
Personal
Spring SO:
Outdoor
Indoor
Personal
150
151
146
146
145
144
143
142
138
143
145
143
143
142
138
143
145
143
20.5 ± 11.2
2.7 ±5.3
3.6 ±5.0
26.7 ± 10.4
1.3 ±2.4
2.1 ±2.8
13.9 ±8.3
9.2 ± 12.6
9.5 ±7.8
13.5 ±9.6
11.6± 15.8
12.1 ± 15.2
3.2 ±3.6
1.4 ±6.2
-0.4 ±4.5
4.8 ±7.0
-1.3 ±9.8
-1.2 ±3.8
18.5
1.1
2.1
25.6
0.7
1.3
11.0
6.1
7.3
11.0
6.4
7.7
2.6
0.4
-0.6
3.7
-1.0
-0.5
57.7
48.0
36.8
60.9
13.6
12.8
40.8
123.3
49.7
55.0
85.8
128.2
19.0
54.6
24.4
48.0
105.2
9.8
-2.4
-2.3
-2.4
8.1
-2.9
-3.1
-0.3
1.4
1.6
-0.6
0.2
0.6
-5.3
-5.6
-5.8
-8.5
-8.6
-10.1
38
-------
Longitudinal correlations (Spearman) between personal exposures and outdoor concentrations of
PM2.5 were calculated for the Atlanta participants (Figure 18). Median correlations were about
0.65 for the COPD cohort and about 0.55 for those with myocardial infarctions.
Longitudinal correlation (Spearman's r) between personal and outdoor PM2.5
Atlanta
g
03
1.0 -
0.8 -
0.6 -
0.4 -
ro 0.2 -
-0.2-
-0.4 -
-0.6
N=24
N=17
COPD Ml
Health Status
Figure 18. Longitudinal correlations between personal exposures and ambient
concentrations of PM2.s: Atlanta
Boston
Summary" statistics for personal, indoor, and outdoor PM: 5, EC and sulfate stratified by season
and sample location (personal, indoor, and outdoor) are presented in Table 13. For all pollutant
measures, concentrations tended to be similar among the microenvironments. This result is not
surprising for sulfate and EC, which have few indoor sources; however, previous studies have
shown higher personal exposures than indoor or outdoor concentrations. While median
exposures were higher than outdoor, the concentrations differed by less than 1 ug/m3.
39
-------
; 13. Descriptive Statistics for Outdoor, Indoor, and Personal Pollutant
entrations in Boston: Particles (in ng/m3).
Pollutant
Season/Sample Type
Winter PM2 5
Outdoor
Indoor
Personal
Second Personal
Summer PM2 5
Outdoor
Indoor
Personal
Second Personal
Summer sulfate
Outdoor
Indoor
Personal
Second Personal
Winter EC
Outdoor
Indoor
Personal
Second Personal
Summer EC
Outdoor
Indoor
Personal
Second Personal
N
75
72
55
47
44
86
63
31
97
105
103
39
94
102
94
54
95
100
101
38
Mean + Std. Dev.
11.4 ±7.7
10.4 ±9.4
17.6 ±26.9
12.5 ±17.3
12.8 ±7.7
12.7 ±8.7
9.4 ±5.1
12.0 ±8.5
4.0 ±2.7
3.1 ±2.2
3.1±2.1
2.9 ±2.4
2.3 ± 1.4
2.0± 1.5
1.5 ± 1.6
1.5± 1.7
1.5 ±0.7
1.5 ±0.6
1.6 ±0.6
1.5 ±0.6
Median
9.9
7.8
10.6
9.9
11.9
10.5
8.3
11.0
3.3
2.6
2.6
1.8
1.9
1.5
1.1
1.2
1.4
1.4
1.5
1.3
Max
46.3
48.9
155.9
122.8
39.8
49.8
26.1
37.0
11.6
10.6
8.9
10.9
6.7
10.8
11.6
11.5
4.7
3.3
4.9
3.8
40
-------
The cumulative PM2.s distributions observed in Boston are plotted on lognormal probability
coordinates in Figure 19. Between the 5th and 95th percentiles, all four types of measurements
appear to be very similar. The personal measurements show the greatest variation, with both the
lowest and highest concentrations of any of the four types of measurements. The similarity of
the outdoor residential concentrations to the outdoor central site suggests both the spatial
homogeneity of the fine particles across the city and the representativeness of the central site in
estimating residential outdoor concentrations.
1000
100 -
•ft, 10 -
0.1
-*~ Personal
-•- Indoor
--*- Outdoor Residential
— Outdoor Central Site
-2
-1
0
Z-score
Figure 19. Boston. Cumulative PM2j distributions of personal, indoor, residential outdoor
and central-site concentrations. N = 266-301
41
-------
Longitudinal correlation coefficients between personal and ambient concentrations were
calculated for each subject in the three cohorts: cardiovascular heart disease, spouses of patients,
and persons with COPD (Figure 20). As at RTF and Atlanta, median coefficients were relatively
low (range of 0.55-0.65). Mean coefficients (dashed line) were even lower due to some strongly
negative coefficients.
PM2.5 - Personal vs Ambient
nal correlation
O 0 0 O 0 ->•
o M *. en co o
"§ -0.2 -
g* -0.4-
0
-0.6 -
-0.8 -
-1.0 -
t T t
j
••
*
_i_
*
•
(N=22) (N=14) (N=8)
CHD Spouse COPD
Cohort
Figure 20. Longitudinal correlations between personal exposures and ambient
concentrations of PlV^.s: Boston.
42
-------
Los Angeles
Summary statistics for personal, indoor, and outdoor PM2 5, elemental carbon (EC), and
paniculate sulfate (864 ) stratified by season and sample type are presented in Table 14.
Table 14. Descriptive Statistics for Outdoor, Indoor, and Personal Pollutant
Concentrations in Los Angeles: Particles (in
Pollutant
Season/Sample Type
Winter PM2 5
Outdoor
Indoor
Personal
Summer PIVb.s
Outdoor
Indoor
Personal
Winter NO3"
Outdoor
Indoor
Personal
Summer NOs"
Outdoor
Indoor
Personal
Winter EC
Outdoor
Indoor
Personal
Summer EC
Outdoor
Indoor
Personal
N
92
92
87
96
97
92
92
94
98
95
96
97
94
90
91
95
95
85
Mean + Std. Dev.
13.5 ±8.5
16.9 ± 11.7
19.6 ±14.5
19.3 ± 9.0
18.1±11.1
25.1 ±20.8
3.1 ±2.6
1.1 ±1.0
1.2± 1.1
2.8 ±1.5
1.7 ±0.8
1.6 ±0.9
1.9±1.1
1.6 ±0.9
1.9±1.0
0.1 ±0.7
0.2 ±0.7
0.3 ±0.8
Median
11.2
12.8
14.4
17.4
17.0
18.8
2.2
0.9
0.8
2.5
1.5
1.4
1.7
1.4
1.7
0.0
0.2
0.2
Max
56.5
49.5
63.5
53.5
94.8
137.8
11.8
4.7
6.4
7.1
4.2
5.0
5.5
5.2
4.9
2.7
2.1
3.3
43
-------
Cumulative PM2.s distributions for Los Angeles are plotted on lognormal probability coordinates
in Figure 21. The most striking aspect of this plot, unlike those for Atlanta and Boston, is the
evidence of a "personal cloud", with personal exposures consistently higher than either indoor or
outdoor concentrations from the 5th to the 99th percentiles.
1000
100 -
I
10 -
-2
-1
0
Z-score
Figure 21. Los Angeles. Cumulative PM2.s distributions of personal, indoor, and
residential outdoor concentrations. N = 179-189
44
-------
Longitudinal correlations (Spearman) between personal exposure and ambient concentrations of
PM2.5 for the summer and winter seasons were calculated for the respondents in Los Angeles
(Figure 22). The median personal-outdoor correlation was quite low in summer at 0.29 and only
slightly higher in winter at 0.49. Surprisingly, personal-indoor correlations were very little
higher. Mean values were again lower than the medians due to several negative relationships.
i.u -
n a -
0.6 -
0.4 -
f 0.2 -
S
o.o -
-0.2 -
-0.4 -
-0.6 -
n a ._
O
O
0
n
— — . -
0
Q
0
° 9 ~?~
— — - '::::i:u::u::::::::_
1 i ' '
i
±
V
1 0
0
Sum Win
Sum Win
Sum Win
Personal vs. outdoor Personal vs. Indoor
Outdoor vs. Indoor
Figure 22. Longitudinal correlations between personal exposures and indoor and outdoor
concentrations of PlVh.j: Los Angeles.
45
-------
Table 15 provides a more detailed look at the individual longitudinal correlations between
personal and outdoor exposure to PM2 5. The table shows that only 4 of the 28 week-long
monitoring efforts resulted in significant relations between personal exposures and outdoor
concentrations.
Table 15. Individual-Specific Spearman Correlation Coefficients: PM2.S in Los Angeles
SEASON
Summer
Winter
Indoor vs. Outdoor
Personal vs. Indoor
Personal vs. Outdoor
SUBJECT
n R p-value n r p-value n r p-value
LPD-16
LPD-17
LPD-18
LPD-19
LPD-21
LPD-22
LPD-23
LPD-24
LPD-25
LPD-26
LPD-27
LPD-28
LPD-29
LPD-30
LPD-01A
LPD-02
LPD-03
LPD-04
LPD-05
LPD-06
LPD-07
LPD-08
LPD-09
LPD-10
LPD-11
LPD-12
LPD-13
LPD-14
LPD-15
7
7
7
6
6
7
7
7
7
7
7
6
7
7
1
7
6
7
5
3
6
6
6
5
4
6
7
7
7
0.61
0.61
0.89
0.14
0.71
0.82
-0.29
0.43
0.89
0.61
0.79
0.54
0.43
0.82
—
0.07
0.49
0.57
0.20
--
0.89
0.37
0.94
0.70
-0.40
0.49
0.61
0.93
0.57
0.15
0.15
0.01
0.79
0.11
0.02
0.53
0.34
0.01
0.15
0.04
0.27
0.34
0.02
—
0.88
0.33
0.18
0.75
—
0.02
0.47
0.00
0.19
0.60
0.33
0.15
0.00
0.18
5
7
6
7
6
6
7
6
6
7
7
7
7
7
2
7
3
6
5
3
6
6
6
4
6
5
7
7
7
0.90
0.32
0.94
0.29
0.14
0.14
0.32
0.31
-0.03
-0.32
0.43
0.54
0.32
-0.18
—
0.61
0.50
0.60
-0.10
~
-0.26
0.31
-0.26
-0.60
0.43
0.60
0.93
0.89
0.61
0.04
0.48
0.00
0.53
0.79
0.79
0.48
0.54
0.96
0.48
0.34
0.22
0.48
0.70
..
0.15
0.67
0.21
0.87
—
0.62
0.54
0.62
0.40
0.40
0.28
0.00
0.01
0.15
5
7
6
6
7
6
7
6
6
7
7
6
7
7
2
7
4
6
7
2
7
6
5
6
4
5
7
7
7
0.70
-0.07
0.89
0.54
0.39
0.43
0.11
0.77
0.14
0.29
0.00
0.20
0.32
-0.43
—
-0.21
0.80
0.49
0.43
—
0.36
0.94
0.10
0.49
-0.40
0.00
0.50
0.82
0.79
0.19
0.88
0.02
0.27
0.38
0.40
0.82
0.07
0.79
0.53
1.00
0.70
0.48
0.34
—
0.64
0.20
0.33
0.34
—
0.43
0.00
0.87
0.33
0.60
1.00
0.25
0.02
0.04
46
-------
A mixed-effects multiple regression model was run to determine the factors affecting indoor
PM2.5 concentrations. The model estimate for the infiltration factor was 0.42 in winter and 0.70
in summer, consistent with increased air exchange rates in summer.
Seattle
PM mass results for all four cohorts are presented in Table 16. Mean personal PM2 5
concentrations ranged from 9.3-10.8 }ig/m3 for adults but were significantly higher (13.3 jig/m3)
for asthmatic children. Indoor means were slightly lower at 7.4-9.5 jo.g/m3, and outdoor means
slightly higher at 9.0-12.6 ug/m3. Considering the difference between personal and indoor
concentrations as an estimator for the personal cloud, mean values were only about 1-2 fig/m3 for
the three adult cohorts but 4 n/m3 for the asthmatic children.
Table 16. Summary of PM Concentrations Between Oct 1999 and May 2001 by Health
Group: Seattle
Location
Personal
Indoor
Outdoor
Central Site
Pollutant
(Hg/m3)
PM25
PM25
PM10
PM2.5
PM10
PM2.5
PM10
Group
COPD
Healthy
Asthmatic
CHD
COPD
Healthy
Asthmatic
CHD
COPD
Healthy
Asthmatic
CHD
COPD
Healthy
Asthmatic
CHD
COPD
Healthy
Asthmatic
CHD
All
All
N
307
183
263
325
443
193
276
329
437
206
274
324
437
194
272
323
435
200
269
324
222
221
Mean
10.5
9.3
13.3
10.8
8.5
7.4
9.2
9.5
14.1
12.6
19.4
16.2
9.2
9.0
11.3
12.6
14.3
14.5
16.4
18.0
10.1
17.3
Std
Dev
7.2
8.4
8.2
8.4
5.1
4.8
6.0
6.8
6.6
7.8
11.1
11.3
5.1
4.6
6.4
7.9
6.8
7.0
7.4
9.0
5.7
9.1
GM
8.6
7.7
11.1
8.8
7.3
6.1
7.9
8.0
12.7
10.6
16.8
13.6
8.0
7.9
9.8
10.6
12.8
13.0
14.7
16.1
8.6
14.9
GSD
1.9
1.8
1.9
1.9
1.7
1.9
1.7
1.8
1.6
1.9
1.7
1.8
1.7
1.7
1.7
1.8
1.6
1.6
1.6
1.6
1.8
1.8
Min
0.8
0.8
1.0
1.4
1.0
0.4
2.2
1.6
2.5
0.6
2.2
0.6
-0.2
0.7
2.8
1.3
2.9
2.9
1.2
3.3
1.0
0.4
Max
45.6
96.2
49.4
66.6
49.9
38.0
36.3
65.3
40.1
62.2
107.7
110.6
28.9
24.5
40.4
41.5
41.4
54.9
47.3
54.3
29.5
49.9
47
-------
Correlations between the paniculate samples are presented for both size fractions in Table 17.
Correlations were high for the residential outdoor and central site PM2.5 measurements, and
slightly lower for the PMio measurements. Personal exposures correlated fairly well with indoor
concentrations (0.57) but not with outdoor levels (0.34). The complete PM2 5 distributions,
summed across all cohorts, are compared in Figure 23. The indoor distribution was substantially
lower than the personal or outdoor distributions.
Table 17. Correlations Between Personal, Indoor, Outdoor, and Central Site Monitors for
PM2.5 and PMi0: Seattle
n l nx * C0n"
Personal PM2 5
, , _., con-
Indoor PM2 5 N
Outdoor PM2 5 °°rr
Central PM2 5 °°rr
I j r,\/i corr
Indoor PMio
„ ^ , „,, corr
Outdoor PMio
Central PM,0 C°rr
Personal
PM25
1
1078
0.57
996
0.34
1010
0.29
974
0.43
1008
0.31
1003
0.25
965
Indoor
PM25
1
1500
0.49
1426
0.40
1293
0.75
1455
0.46
1428
0.34
1288
Outdoor Central
PM2 5 PM2 5
1
1498
0.83
1298
0.33
1441
0.89
1447
0.67
1290
1
1408
0.30
1304
0.79
1294
0.87
1368
Indoor Outdoor Central
PM10 PM10 PM10
1
1515
0.31
1448
0.31
1298
1
1497
0.73
1288
1
1398
Note: All p-values < 0.0001.
48
-------
100
r>
10
-*-Personal
-•- Indoor
-*- Outdoor Residential
-<- Outdoor Central Site
-2
0
Z-score
Figure 23. Cumulative distribution functions for personal, indoor, outdoor, and central site
PM2.s: Seattle. N = 1078-1500.
49
-------
At equilibrium, as mentioned above, the infiltration factor F,nf= Pa/(a+k), where P is the
penetration factor (dimensionless), a is the air exchange rate (h'1) and k is the deposition rate
(h"1). The magnitudes of these parameters were estimated using the recursive mass balance
model based on the MIE pDR measurements (Allen et al., in press). An algorithm was developed
to identify all indoor peaks. These peaks were removed, leaving only the concentrations due to
penetration of outdoor particles. The 10-minute average measured concentrations from the
indoor and outdoor pDRs were combined to form one-hour concentrations. Using a one-hour
time step, the recursive mass balance model was run to provide separate estimates of P, a, k, and
Fm/for each home. Summary statistics from this effort are provided in Table 18 and Boxplots
showing the range of values for each parameter across the 65 homes are shown in Figure 24.
The penetration factor P was close to 1, as was also the case in the earlier PTEAM Study
(Pellizzari et al., 1992; Ozkaynak et al., 1996a.b; Clayton et al., 1993). The deposition constant k
was 0.15 (± 0.19) h"1, smaller than the value of 0.39 h"1 observed in PTEAM.
Table 18. Summary of Estimated Particle Penetration (P), Air Exchange Rate (a), Particle
Decay Rate (A"), and the Ambient PM Infiltration Efficiency (F,n/) Using the Recursive and
Nonlinear Regression Models: Seattle
Variable
N
Mean
Median
Std Dev
P
a
k
F,nf
65
65
65
65
0.97
0.59
0.15
0.78
1.00
0.73
0.07
0.81
0.08
0.26
0.19
0.19
50
-------
1.2
1.0-
0.8-
0.6-
0.4-
0.2-
0.0 J
T
0
I
Penetration Air exchange Particle decay particle infiltration
rate (p) rate (a) rate (k) efficiency (Finf)
Figure 24. Distribution of P, a, k, and F,-n/ among residences: Seattle.
51
-------
Longitudinal correlations between personal and outdoor PM2.s were calculated for each subject
(Figure 25). Correlations covered a wide range from negative to nearly 1, as has been observed
in other studies, with the median correlation between 0.3 and 0.4 for all four cohorts. The
hypothesis that the sick persons would have higher longitudinal correlations with outdoor air
than healthy persons was not confirmed.
(a)
1.0 -
0.8 -
o 0.6 -
Jo
b 0.4-I
o
o
"(5 0.2 -
o
-1 -0.2 -
-0.4 -
-0.6 -
(N=24)
(N=28)
(N=24)
(N=17)
Asthma CHD COPD
Health Status
Healthy
Figure 25. Longitudinal correlation between personal and central site PM2.s for each
subject by health status (N>6 for each subject): Seattle
Elderly high-risk subpopulations may spend more of their time in the home and less in transit
than their healthy counterparts. This hypothesis was tested in all studies by administering a
questionnaire asking for time spent in several microenvironments. In the Seattle study, the
hypothesis was confirmed (Table 19), although the differences were relatively small.
52
-------
Table 19. Percentage of Time Spent in Microenvironments by Health Group: Seattle.
Group
Asthmatic
Kids (N=33)
CHD,
Adults
f\l—1Q\
(.IN 38J
COPD,
Adults
/TvT «\
(JN— JO)
Healthy,
Adult
f\l 1O\
^1N jy)
Microenvironment
Home
Yard
In Transit
Work
Outdoors
Indoors away from home
Cooking, self
Cooking, others
Home
Yard
In Transit
Work
Outdoors
Indoors away from home
Cooking, self
Cooking, others
Home
Yard
In Transit
Work
Outdoors
Indoors away from home
Cooking, self
Cooking, others
Home
Yard
In Transit
Work
Outdoors
Indoors away from home
Cooking, self
Cooking, others
Percentage of time spent in each
microenvironment
Mean
66.4
1.7
4.4
1.1
4.7
21.0
0.1
0.7
85.5
1.0
3.6
0.3
0.9
6.9
1.7
0.2
87.6
0.8
3.2
0.1
1.0
6.1
1.0
0.2
82.7
1.2
4.0
1.0
1.7
8.0
1.0
0.3
Std Dev
5.7
2.6
1.7
3.5
3.5
6.4
0.1
0.5
7.8
1.4
2.3
1.7
1.2
5.1
1.6
0.3
6.9
1.0
1.9
0.6
1.9
4.6
1.3
0.6
8.3
1.6
2.5
2.8
1.8
5.3
1.0
0.6
Min
55.5
0.0
1.3
0.0
0.1
4.5
0.0
0.0
65.0
0.0
0.1
0.0
0.0
0.1
0.0
0.0
71.4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
66.8
0.0
0.5
0.0
0.0
0.1
0.0
0.0
Max
80.0
8.2
8.2
16.5
17.5
33.2
0.5
1.9
96.5
6.0
9.2
10.6
4.8
20.9
5.8
1.5
100.0
4.3
7.3
3.1
11.6
21.3
5.6
2.7
99.2
6.7
9.3
12.4
7.9
19.4
4.4
2.7
53
-------
DISCUSSION
The mean concentrations of PM2.5 are compared across all studies in Table 20. (Results from two
earlier studies of retirement homes by NERL are included for comparison.) PM2 5 exposures
were lower in Seattle than in the other cities. Seasonal differences were not strong in Boston or
Raleigh, but were higher in spring than in fall in Atlanta, higher in summer than in winter in Los
Angeles, and (not shown) higher in the fall-winter heating season in Seattle.
Indoor PlV^.s concentrations were often quite similar to outdoor concentrations in all cities. The
only exceptions were the two retirement homes in Baltimore and Fresno, where recirculation and
filtering of outdoor air was provided on a constant basis by the HVAC systems. Personal
exposures were often but not always higher than indoor concentrations, but only by a few (1-4)
u.g/m3.
Table 20. Arithmetic Mean PM2.s Concentrations (|ig/m3) — All Studies.
Site
Raleigh/Chapel Hill
Atlanta-Fall
Atlanta—Spring
Boston— Winter
Boston— Summer
Los Angeles— Winter
Los Angeles— Summer
Seattle-COPD
Seattle-healthy
Seattle-CHD
Seattle — asthmatic children
Baltimore retirement home
Fresno retirement home
N (Personal)
712
141
142
55/47a
63/31
87
92
307
183
325
263
325
120
Personal
23.0
16.6
15.0
17.6/12.53
9.4/12.0
19.6
25.1
10.5
9.3
10.8
13.3
13.0
13.3
Indoor
19.1
14.3
19.8
10.4
12.7
16.9
18.1
8.5
7.4
9.5
9.2
10.0
9.7
Outdoor
19.3
14.7
22.0
11.4
12.8
13.5
19.3
9.2
9.0
12.6
11.3
22.0
20.5
Second personal sample from same household.
54
-------
Over 2100 indoor and outdoor PMio samples were collected in three cities (Table 21). Once
again Seattle had the lowest outdoor concentrations. Indoor and outdoor PMio concentrations
were quite comparable in all areas, again with the exception of the two retirement homes, which
were able to reduce the outdoor levels penetrating indoors by substantial amounts.
Table 21. Arithmetic Mean PMio Concentrations (|4,g/m3)—All Studies.
Site
Raleigh/Chapel Hill
Seattle-COPD
Seattle-healthy
Seattle-CHD
Seattle — asthmatic children
Baltimore retirement home
Fresno retirement home
N (Indoor)
761
437
206
324
274
28
24
Indoor
23.2
14.1
12.6
16.2
19.4
11.0
15.1
Outdoor
27.2
14.3
14.5
18.0
16.5
30.0
28.2
A fundamental goal of all these studies was to relate personal exposure to outdoor
concentrations. For different cohorts, median longitudinal correlation coefficients (either
Pearson or Spearman) ranged between 0.10 and 0.65. These values suggest that for the median
person in each of the high-risk sensitive subpopulations, outdoor air explained as little as 1% up
to a maximum of about 40% of the variation in personal exposure. All studies included some
persons with very high correlations with ambient air and others with very low or even negative
correlations. These correlations are lower than those found for healthy adults and children
(Ebelt et al., 2000; Janssen 1998; Janssen et al., 1998; Williams et al., 2000a).
An important parameter governing the effect of outdoor air particles on indoor concentrations is
the infiltration factor:
Finf = Pa/(a+k)
The infiltration factor is expected to be lower when the house is closed and higher when
windows are open. Therefore Finf was calculated separately for the heating and non-heating
seasons in each city. During the heating season, the infiltration factor varied over a tight range
of 0.40 to 0.53. In the non-heating seasons, the range was much wider, from 0.40 to 0.79 (Table
22). Every city except for RTF showed an increase in Fjnf between the heating and non-heating
seasons. The increase was quite large in Los Angeles, Boston, and Seattle, and relatively small
in Atlanta.
55
-------
Table 22. Variation of PM2.s Infiltration Factor by Season—AH Studies
City
ITP: linear model
.os Angeles
Boston
Boston sulfate
Atlanta
Atlanta sulfate
Seattle
N
29
15
14
N/A
24
24
55
Heating
0.46
0.42
0.40
N/A
0.43
0.40
0.53
SE
0.05
0.08
0.13
N/A
0.10
0.04
0.16
N
25
15
15
15
22
22
55
Non-heating
0.40
0.70
0.67
0.75
0.49
0.45
0.79
SE
0.04
0.11
0.10
0.03
0.14
0.04
0.18
A summary of the longitudinal correlations between personal and outdoor air for participants in
all the panel studies is provided in Table 23. The three studies all succeeded in collecting valid
personal and outdoor data for at least four consecutive days on about 100 separate occasions. All
studies agreed with each other, and also with previous studies, in having roughly half the
participants with correlations below 0.5. These values indicate that for the median person in
each of the studies, outdoor air explained between 18 and 25% of the variance in personal
exposure. The studies also agreed in showing less than one fourth of the participants (14-21%)
with significant personal-outdoor correlations.
Table 23. Longitudinal Correlations Between Personal and Outdoor Air for Participants—
All Studies
Organization
NERL-RTI
Univ. Wash.
Harvard
Cities
RTF
Seattle
Boston-Atlanta-L .A .
N
persons1
112
98
105
Nwith
p<0.05
16
16
22
Fraction with
p<0.05
0.14
L 0.16
0.21
Nwith
r>0.5
48
49
49
Fraction with
r>0.5
0.43
0.50
0.47
Number with at least four consecutive days of personal and residential outdoor measurements
within one season. Persons monitored in two or more seasons are counted separately for each
season.
56
-------
CONCLUSIONS
A strong effort has been made to gather information about the actual exposures of those persons
at high risk from pollution. Over 200 persons were monitored in five cities and over 2500
personal, indoor, and outdoor PMz.s samples were collected, with another 2000 indoor and
outdoor PMio samples collected. Associated co-pollutants and elements were also sampled,
many with personal as well as indoor and outdoor measurements. Thousands of questionnaires
were administered, creating a rich database on activity patterns, time budgets, and particle-
generating activities. These data are a valuable resource for current and future analyses, and it is
expected that a large number of journal articles will be produced in the next few years
investigating different aspects of these studies.
The main goal of the studies, to document how the exposure of high-risk subpopulations to fine
particles is related to ambient concentrations, both cross-sectionally and longitudinally, has been
amply fulfilled. All major identified high-risk subpopulations were included in the study efforts.
Good geographic distribution across the country was incorporated into the study designs, and
highly precise measurements of both exposure and outdoor concentrations performed. The
second goal of the studies, to calculate the contribution of ambient concentrations to total
exposure, was more difficult. While estimates have been achieved in some of the studies,
additional work in this area will depend on elemental data still being analyzed and on which of
several statistical approaches might be adopted. However, the basic data that will ultimately lead
to considerable advances in fully completing this goal have been collected and will form a major
focus of future journal articles bearing on this question.
The organizations performing the work gained valuable experience in field studies involving
extensive personal and indoor monitoring. As a result, new improved personal monitors were
developed and the Harvard Multipollutant Personal Sampler, developed as part of this study, is
now available commercially.
Hundreds of volunteers took part in the studies and in most cases successfully completed them,
indicating that the burden of carrying the monitoring equipment, making room for the indoor
monitors, keeping activity diaries, and filling out questionnaires each day had been successfully
alleviated by good planning and study design.
Two hypotheses were advanced at the beginning of these studies. One was that high-risk
subpopulations would engage in fewer dust-generating activities and would therefore have lower
indoor air concentrations and lower personal exposures than healthy cohorts. This hypothesis
was not confirmed. In Seattle, for instance, the healthy cohort had the lowest personal exposures
of the four cohorts studied. In Boston, the spouses of the high-risk persons had lower
concentrations in one season and higher concentrations in the other, but with no overall
difference.
57
-------
A second related hypothesis was that the personal exposures of the high-risk cohorts would have
stronger correlations with outdoor air than healthy cohorts, due to fewer particle sources indoors.
Again the hypothesis was not confirmed, with the healthy cohort in Seattle actually showing
higher mean correlations with outdoor air than the three high-risk cohorts.
The following conclusions may be ventured. However, it should be carefully noted that all
subjects were chosen on a non-probabilistic basis, and therefore all conclusions apply only to
the subjects themselves; they must not be extrapolated to larger groups of people.
• Personal PMz.s monitors were fully evaluated in side-by-side comparisons with EPA
reference or equivalent instruments at all sites and generally agreed well.
• Mean personal PN^.s exposures for the different cohorts ranged from 9 to 25 ng/m3. The
Seattle cohorts were at the low end and the Los Angeles and Raleigh/Chapel Hill cohorts
at the high end.
• Mean indoor and outdoor PMa.5 concentrations were similar for all cohorts living in
private homes in all cities, and ranged from 7-20 yg/m3 indoors and from 9-22 jig/mS
outdoors.
• Personal clouds on the order of 1-4 |ig/m3 were observed for PM2.5 for most cohorts.
• Mean indoor and personal PM2.5 concentrations were much lower than outdoor
concentrations for cohorts living in retirement homes, due possibly to extensive filtering
and recirculation of outdoor air by the HVAC systems.
• Personal PM2.s exposures were similar for the healthy and sick cohorts. For these
subjects, there was no indication that the sensitive persons were reducing their exposure.
• Longitudinal correlations of personal PM2.5 exposure with outdoor air concentrations
were low for some individuals in each cohort and high for others, suggesting the
importance of individual activity patterns and household characteristics in affecting the
personal-outdoor relationship. In all three studies, fewer than half of these correlation
coefficients exceeded 0.5, and fewer than one fourth were significant at the p<0.05 level.
• Calculations of the infiltration factor, which determines the contribution of outdoor air to
indoor concentrations, are difficult and have resulted in divergent estimates between
studies and even between different applications of the mass-balance model (linear vs.
nonlinear) in the same study. It is not clear if this is fully explained by differences among
the cities in climate, house construction practices, use of air conditioning, or dependence
on the assumptions of the mass balance model. This is a problem that will require further
analyses of the collected data.
58
-------
FUTURE WORK
Although much has been learned, there remain some important questions requiring further
research, development, and field studies. Presently our estimates of the infiltration factor, the
penetration coefficient, and the PMb.5 deposition rate appear to be conflicting. The assumptions
underlying the use of the mass balance model need to be carefully examined.
A continuing challenge is to understand the organic chemical (OC) loading on particles, which
provides a substantial portion of the total mass. Because of both positive and negative artifacts
on quartz fiber filters, the standard thermo-optical method of determining OC is usually unable
to arrive at a trustworthy estimate of the actual OC mass on the particles in their normal state in
the atmosphere. This problem is being further studied at EPA-NERL.
Because of increasing interest in air toxics, future studies may combine studies of particles with
studies of polar and non-polar volatile organic compounds (VOCs) and semi-volatile organic
compounds (SVOCs). EPA-NERL is presently planning studies that will add these air toxics to
their continuing PM studies. The importance of ultrafine particles in contributing to mortality
and morbidity continues to be of interest, but a lack of personal monitors capable of measuring
ultrafine particles has limited our knowledge of personal exposure to ultrafines, and data is
sparse on indoor concentrations. Additional work in the area of low-burden, low-cost personal
exposure monitors and the adaptation of survey instruments (activity diaries and questionnaires)
to better characterize potential personal exposures to PM of ambient sources in the general
population is needed.
59
-------
References
Allen, R., Larson, T., Wallace, L.A., Sheppard, L., and Liu, L.-J. S. Investigation of indoor and
outdoor contributions to total indoor particulate matter exposure. Environmental Science &
Technology, in press.
Chang, L.T., Samat, J., Wolfson, J.M., Rojas-Bracho, L., Suh, H.H., Koutrakis, P. (1999).
Development of a personal multi-pollutant exposure sampler for particulate matter and criteria
gases. Pollution Atmospherique. Numero Special 40e Anniversaire de L'APPA 31-39.
Clayton, C.A., Perritt, R.L., Pellizzari, E.D., Thomas, K.W., Whitmore, R.W., Wallace, L.A.,
Ozkaynak, H., Spengler, J.D. (1993). Particle Total Exposure Assessment Methodology
(PTEAM) study: distributions of aerosol and elemental concentrations in personal, indoor, and
outdoor air samples in a southern California community. J. Expos. Anal. Environ. Epidem. 3:
227-250.
Demokritou, P., Kavouras, I., Ferguson, S., Koutrakis, P. (2001a). Development and laboratory
performance evaluation of a personal multi-pollutant sampler for simultaneous measurements of
particulate and gaseous pollutants. Aerosol Sci. Technol. 35: 741-752.
Demokritou, P., Kavouras, I.G., Harrison, D., Koutrakis, P. (200 Ib). Development and
evaluation of an impactor for a PM2.s speciation sampler. J. Air & Waste Management
Association 51:514-523.
Ebelt, ST., Petkau, A.J., Vedal, S., Fisher, T.V., Brauer, M. (2000). Exposure of chronic
obstructive pulmonary disease patients to particulate matter: relationship between personal and
ambient air concentrations. J.Air& Waste Management Association 50: 1081-1094.
Janssen, N. (1998). Personal exposure to airborne particles: validity of outdoor concentrations
as a measure of exposure in time-series studies. Thesis, Department of Environmental Sciences,
University of Wageningen, Utrecht, Netherlands.
Janssen, N.A.H., Hoek, G., Brunekreef, B., Harssema, H., Mensink, I., Zuldhof, A. (1998).
Personal sampling in adults: relation among personal, indoor, and outdoor air concentrations.
Am. J. Epidem. 147: 537-547.
Marple, V.A., Rubow, K.L., Turner, W., Spengler, J.D. (1987). Low flow rate sharp cut
impactors for indoor air sampling: design and calibration. J. Air Pollut. Control Assoc. 37:
1303-7.
National Research Council-National Academy of Science (1998). Research priorities for
airborne particulate matter I: immediate priorities and a long-range research portfolio.
National Academy Press, Washington, DC.
60
-------
Ozkaynak, H., Xue, J., Weker, R., Butler, D., Koutrakis, P., Spengler, J. (1996a). The Particle
TEAM (PTEAM) study: analysis of the data. Volume III, Final Report. EPA Contract #68-02-
4544. Research Triangle Park, NC: US Environmental Protection Agency.
Ozkaynak, H., Xue, J., Spengler, J.D., Wallace, L.A., Pellizzari, E.D., Jenkins, P. (1996b).
Personal exposure to airborne particles and metals: results from the Particle TEAM Study in
Riverside, CA. J. Exposure Anal. Environ. Epidem. 6: 57-78.
Pellizzari, E.D., Thomas, K.W., Clayton, C.A., Whitmore, R.W., Shores, R.C., Zelon, H.S.,
Perritt, R.L. (1992). Particle Total Exposure Assessment Methodology (PTEAM): Riverside,
California pilot study. Volume I, Final Report. EPA Contract # 68-02-4544, Research Triangle
Park, NC: US Environmental Protection Agency.
Rojas-Bracho, L., Suh, H.H., Koutrakis, P. (2000). Relationships among personal, indoor, and
outdoor fine and coarse particle concentrations for individuals with COPD. J. Expos. Anal.
Environ. Epidem. 10: 294-306.
Thomas, K.W., Pellizzari, E.D., Clayton, C.A., Whitaker, D.A., Shores, R.C., Spengler, J.D.,
Ozkaynak, H., Wallace, L.A. (1993). Particle Total Exposure Assessment Methodology
(PTEAM) 1990 study: method performance and data quality for personal, indoor, and outdoor
monitoring. J. Exposure Anal. Environ. Epidem. 3: 203-226.
US EPA (2002). Preliminary particulate matter mass concentrations associated with
longitudinal panel studies: assessing human exposures of high-risk subpopulations to
paniculate matter. National Exposure Research Laboratory, Office of Research and
Development, EPA/600/R-01/086. Washington, D.C.: U.S. Environmental Protection Agency.
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: 4193-4204.
Williams, R., Suggs, J., Zweidinger, R., Evans, G., Creason, J., Kwok, R., Rodes C., Lawless, P.,
Sheldon, L. (2000b). Comparison of PM2.5 and PMio monitors. J. Exposure Anal. Environ.
Epidem. 10: 497-505.
61
-------
APPENDIX
Journal Articles Published or in Preparation
from the Panel Studies
-------
Journal Articles Published or in Preparation from the Panel Studies
Research Triangle Park. NC
Conner, T., Norris, G., Landis, M., and Williams, R. (2001). Individual particle analysis of
indoor, outdoor, and personal samples from the 1998 Baltimore retirement home study.
Atmospheric Environment, 35:3935-3946.
Landis, M.S., Norris, GA., Williams, R.W., and Weinstein, J.P. (2001) Personal exposures to
PM2.5 mass and trace elements in Baltimore, Maryland. Atmospheric Environment 35: 6511-
6524.
Rea, A., Zufall, M., Williams, R., Reed, C., and Sheldon, L. (2001). The influence of human
activity patterns on personal PM exposure: a comparative analysis of filter-based and continuous
particle measurements. Journal of Air and Waste Management Association, 51:1271-1279.
Williams, R., Suggs, J., Rea A., Leovic, K., Vette, A., Sheldon, L., Rodes C., and Thornburg J.
The Research Triangle Park particulate matter panel study: modeling ambient source
contribution to personal and residential PM mass concentrations. Atmospheric Environment, in
press.
Williams, R., Suggs, J., Rea, A., Leovic, K., Vette, A., Croghan, C., Sheldon, L., Rodes, R.,
Thornburg, J., Ejire, A., Herbst, M., and Sanders, W. The Research Triangle Park particulate
matter panel study: PM mass concentration relationships. Atmospheric Environment, in press.
Rodes C, Lawless P., Thornburg J., Williams, R., Evans G., Zweidinger R., Norris, G., McDow
S. The potential influence efface velocity on the loss of volatile species collected on Teflon
filters. Submitted to Journal of Air and Waste Management Association.
Thornburg J., Rodes, C.E., Williams, R. Relationship between HVAC system operation, air
exchange rate, and indoor-outdoor particulate matter ratios. Submitted to Atmospheric
Environment.
Lawless, P., Rea, A., Williams, R. Personal monitoring compliance observed in the NERL
Research Triangle Park particulate matter panel study. Planned journal submission.
Rea A., Croghan C., Thornburg J., Rodes, E., Williams R. PM concentrations associated with
personal activities based on real-time personal nephelometry data from the NERL RTP PM panel
study. Submitted to Atmospheric Environment.
A-l
-------
Boston
Brown et al. Assessing exposures to paniculate and gaseous pollutant for senior adults,
individuals with COPD and MI patients in Boston, MA. Planned journal submission.
Sarnat et al. Examining the impact of ambient fine paniculate matter sources on personal
exposures: the effect of cohort, city and season. Planned journal submission.
Atlanta
Chang et al. The relationship between outdoor, indoor, and personal exposures to PM2 5 and Its
components for two sensitive cohorts. Planned journal submission.
Reid et al. Factors affecting the relationship between indoor and outdoor concentrations for
ozone, NO2, and SO2. Planned journal submission.
Wheeler et al. associations between cardiovascular health and particulate exposures for two
sensitive cohorts. Planned journal submission.
Los Angeles
Chang et al. Characterization of PM2.5, EC and NOs" Exposures for the Metropolitan Los
Angeles Area. Planned journal submission.
Lau et al. Potential for confounding by gaseous pollutants: results from Los Angeles. Planned
journal submission.
Seattle
Liu, L.-J. S., Slaughter, C., Larson, T. (2002). Comparison of light scattering devices and
impactors for particulate measurements in indoor, outdoor, and personal environments.
Environmental Science & Technology, 36, 2977-2986.
Goswami, E., Larson, T., Lumley, T., Liu, L.-J. S. (2002). Spatial characteristics of fine
particulate matter: identifying representative monitoring locations in Seattle. Journal of Air &
Waste Management Association 52: 324-333.
Allen, R., Box, M., Larson, T., Liu, L.-J. S. (2001). A cost-effective weighing chamber for
particulate matter filters. Journal of Air & Waste Management Association 51: 1659-1653.
A-2
-------
Seattle (ConU
Pang, Y., Gundel, L. A., Larson, T., Finn, D., Liu, L.-J. S., Claiborn, C (2002). Development and
evaluation of a novel personal particulate organic and mass sampler (PPOMS). Environmental
Science & Technology 36:5205-5210.
Liu, L.-J. S., Box, M., Kalman, D., Kaufman, J., Koenig, J., Larson, T., Sheppard, L., Slaughter,
C., Lewtas, J., Wallace, L.A. (2003). Exposure assessment of particulate matter for susceptible
populations in Seattle, WA. Environmental Health Perspectives, 111 (7): 909-918.
Allen, R., Larson, T., Wallace, L.A., Sheppard, L., and Liu, L.-J. S. Investigation of indoor and
outdoor contributions to total indoor particulate matter exposure. Environmental Science &
Technology, in press.
Koenig, JQ; Jansen, K; Mar, TF; Lumley, T; Kaufman, J; Sullivan, J; Liu, L-J S; Shapiro, GG;
Larson, TV. Measurement of offline exhaled nitric oxide in an air pollution health effect study.
Environmental Health Perspectives, in press.
Mar, T.F., Koenig, J. Q., Jansen, K., Sullivan, J., Kaufman, J., Trenga, C. A.,Siahpush, H., Liu,
L-J. S. Neas, L. An analysis of the association between air pollution and blood pressure, heart
rate and pulse oximetry in elderly subjects. Submitted to Epidemiology.
Lianne Sheppard, Chris Slaughter, Jon Schildcrout, L.-J. Sally Liu, Thomas Lumley. Exposure
measurement error in epidemiologic studies of air pollution. Submitted to J. Exposure Analysis
& Environmental Epidemiology.
Wu, CF Wu; C.F., Delfino; R.J., Floro; J.N., Samimi; B.S., Quintana P.J.E.; Kleinman, M.T.;
Liu, L.-J. S. Evaluation of personal nephelometers in indoor, outdoor and personal
environments. Submitted to J. Exposure Analysis & Environmental Epidemiology.
Allen, Ryan; Wallace, Lance; Liu, L.-J. Sally. Estimating hourly personal exposures to indoor-
and outdoor-generated particles among sensitive populations in Seattle. Submitted to J. Air &
Waste Management Association.
Larson, Timothy; Gould, Timothy; Simpson, Chris; Claiborn, Candis; Lewtas, Joellen; and Liu,
L.-J. Sally. Source apportionment of indoor, outdoor and personal PM2.5Jn Seattle, WA using
positive matrix factorization. Submitted to J. Air & Waste Management Association.
Liu et al., Outdoor contribution to personal PM2.5 (planned submission in Fall 2003).
Claiborn et al. Indoor OC artifact (planned submission in Fall 2003).
Larson et al. Source apportionment for indoor and outdoor PM exposures (planned submission in
Spring 2003).
A-3
-------
Seattle (Cont.)
Johns et al. Comparisons of breath, personal, indoor, and ambient CO (planned submission in
Fall 2003).
Claiborne et al. Characterization of indoor and outdoor EC/OC in Seattle (planned submission in
Fall 2003).
Wu et al. Use of real-time CO2 monitors and nephelometer measurements to estimate and
characterize air exchange rates (planned submission in Winter 2004).
Larson et al. Source apportionment for personal PM exposures, using elements, WS markers,
and PAHs (planned submission in Winter 2004)
Allen et al. Sensitivity analysis of recursive model for estimating Finf (planned submission in
Winter 2004).
Simpson et al. Validation of using PAHs as markers for gasoline exposure using personal air
PAHs and urine PAHs data (planned submission in Fall 2003).
Simpson et al. Methoxyphenol methods paper (planned submission in Summer 2003).
Simpson et al. Evaluation of usefulness of woodsmoke tracers in outdoor filters (planned
submission in Fall 2003).
Simpson et al. Indoor-outdoor relationships for woodsmoke tracers, as compared with other
tracer methods (planned submission in Winter 2004).
Larson et al. Using woodsmoke tracers including levoglucosan in source apportionment
(planned submission in Winter 2004).
Simpson et al. Urine biomarker - Seasonal, weekday/ weekend variations, attenuation factors
(planned submission in Winter 2004).
A-4
------- |