Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Ambient Particulate Pope
1. Research | Article
Ambient Particulate Air Pollution, Heart Rate Variability, and Blood Markers
of Inflammation in a Panel of Elderly Subjects
C. Arden Pope III,1 Matthew L. Hansen,1,2 Russell W. Long,3,4 Karen R. Nielsen,5 Norman L. Eatough,3
William E. Wilson,6 and Delbert J. Eatough 3
1Department of Economics, Brigham Young University, Provo, Utah, USA; 2Case Western Reserve University School of Medicine,
Cleveland, Ohio, USA; 3Department of Chemistry and Biochemistry, Brigham Young University, Provo, Utah, USA; 4Human Exposure
and Atmospheric Sciences, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina, USA; 5Department of
Radiation Oncology, University of Utah School of Medicine, Salt Lake City, Utah, USA; 6Office of Environmental Information, U.S.
Environmental Protection Agency, Research Triangle Park, North Carolina, USA
north and south along the western front of the
Wasatch Mountains. During winter low-level
Epidemiologic studies report associations between particulate air pollution and cardiopulmonary
temperature inversion episodes, PM concen-
morbidity and mortality. Although the underlying pathophysiologic mechanisms remain unclear, it
trations become elevated as local emissions are
has been hypothesized that altered autonomic function and pulmonary/systemic inflammation may
trapped in a stagnant air mass near the valley
play a role. In this study we explored the effects of air pollution on autonomic function measured by
floor. Thus, PM pollution levels in the winter
changes in heart rate variability (HRV) and blood markers of inflammation in a panel of 88 elderly
are generally higher and have much greater
subjects from three communities along the Wasatch Front in Utah. Subjects participated in multiple
variability than during other seasons. In
sessions of 24-hr ambulatory electrocardiographic monitoring and blood tests. Regression analysis
was used to evaluate associations between fine particulate matter [aerodynamic diameter ≤ 2.5 µm Hawthorne, data were collected during the
winter of 1999/2000 and the summer of
(PM2.5)] and HRV, C-reactive protein (CRP), blood cell counts, and whole blood viscosity. A
100-µg/m3 increase in PM2.5 was associated with approximately a 35 (SE = 8)-msec decline in stan- 2000. In Bountiful and Lindon, data were
collected during the winter of 2000/2001.
dard deviation of all normal R-R intervals (SDNN, a measure of overall HRV); a 42 (SE = 11)-msec
Panel participants. Panels of elderly resi-
decline in square root of the mean of the squared differences between adjacent normal R-R intervals
dents of the three communities were recruited
(r-MSSD, an estimate of short-term components of HRV); and a 0.81 (SE = 0.17)-mg/dL increase in
to participate in 24-hr ambulatory electro-
CRP. The PM2.5–HRV associations were reasonably consistent and statistically robust, but the CRP
cardiographic (ECG) monitoring and blood
association dropped to 0.19 (SE = 0.10) after excluding the most influential subject. PM2.5 was not
tests. Potential participants were initially
significantly associated with white or red blood cell counts, platelets, or whole-blood viscosity. Most
recruited by directly contacting persons living
short-term variability in temporal deviations of HRV and CRP was not explained by PM2.5; how-
in the neighborhoods adjacent to the moni-
ever, the small statistically significant associations that were observed suggest that exposure to PM2.5
toring sites and asking for neighborhood
may be one of multiple factors that influence HRV and CRP. Key words: cardiovascular disease,
referrals. Information about the study was
C-reactive protein, ECG monitoring, heart rate variability, inflammation, particulate air pollution,
given, and for those who indicated a willing-
PM 2.5 . Environ Health Perspect 112:339–345 (2004). doi:10.1289/ehp.6588 available via
ness to participate, an eligibility questionnaire
http://dx.doi.org/ [Online 12 November 2003]
was completed. A total of 89 persons were ini-
tially enrolled in the study. Six persons in
Evidence is accumulating that particulate study, we hypothesized that altered autonomic Hawthorne who participated in the winter
matter (PM) air pollution is associated with function and pulmonary inflammation play a panel also participated in the subsequent
increased cardiopulmonary morbidity and role in the pathogenesis of cardiopulmonary summer panel and, for purposes of this analy-
mortality [Committee of the Environmental disease related to fine particle air pollution sis, were treated as separate subjects, which
[aerodynamic diameter ≤ 2.5 µm (PM2.5)]. A
and Occupational Health Assembly of the resulted in 95 subjects. Three subjects with-
American Thoracic Society (CEOHA-ATS) primary objective of this study was to further drew from the study before data collection
1996; Pope and Dockery 1999; Pope et al. examine PM’s influence on cardiac autonomic began, and four more withdrew with only a
2002]. Although the underlying physiologic function, as measured by HRV and blood single day of data collection, finally resulting in
mechanisms for these effects are still being markers of inflammation, by studying a panel 88 subjects with 250 total observations (~2.84
explored, it has been postulated that PM’s of elderly persons who may be more suscepti- observations per subject). All subjects were
influence could involve altered autonomic ble or exhibit greater response to PM. nonsmokers living in homes with no smokers.
function and inflammatory responses indi-
Materials and Methods
cated by various blood markers (Glantz 2002; Address correspondence to C.A. Pope III, 142 FOB,
Godleski et al. 2000; Seaton et al. 1995; Study areas and periods. This study was con- Brigham Young University, Provo, UT 84602 USA.
Stone and Godleski 1999; Utell et al. 2002). ducted in three Utah communities: a) a com- Telephone: (801) 422-2157. Fax: (801) 422-0194.
Epidemiologic studies have observed ele- munity in Salt Lake City, Utah, referred to as E-mail: cap3@email.byu.edu
The U.S. Environmental Protection Agency
vated PM exposures to be associated with spe- Hawthorne, located 2.5 mi (4 km) southeast
(U.S. EPA) through its Office of Research and
cific physiologic end points, including reduced of the city’s urban center; b) Lindon, located
Development partially funded and collaborated in the
lung function (Hoek et al. 1998), increased in the Provo/Orem metropolitan area approx- research described here under IMPACT Cooperative
blood plasma viscosity (Peters et al. 1997), imately 40 mi (65 km) south of Salt Lake Agreement CR827364 and STAR grant R82799301
reduced heart rate variability (HRV; Gold City; and c) Bountiful, a Salt Lake City sub- with Brigham Young University. However, the views
et al. 2000; Liao et al. 1999; Pope et al. 1999), urb, located approximately 12 mi (20 km) expressed in this article are those of the authors and
do not necessarily reflect the views or policies of the
and markers of inflammation (Ghio and north of the urban center. Sources of PM in
U.S. EPA. Mention of trade names or commercial
Devlin 2001; Peters et al. 2001; Salvi et al. the communities included substantial traffic
products does not constitute U.S. EPA endorsement
1999, 2000; Schwartz 2001; Tan et al. 2000). and urban-related sources, an integrated steel or recommendation for use.
It has also been suggested that certain groups mill, and local oil refineries. All three commu- The authors declare they have no competing
may be more at risk for the effects of PM expo- nities are located along the Wasatch Front, a financial interests.
sure, including the elderly (Pope 2000). In this relatively densely populated area running Received 11 July 2003; accepted 12 November 2003.
339
Environmental Health Perspectives • VOLUME 112 | NUMBER 3 | March 2004
2. |
Article Pope et al.
Subjects were retired persons between 54 and (Perma Pure LLC, Toms River, NJ) dryer, and of Utah Division of Air Quality Hawthorne,
89 years of age, and 57% were female. All TEOM monitor technology. This sampler has Bountiful, and Lindon monitoring stations.
subjects lived in private homes or were resi- been described in detail elsewhere (Eatough Because of the inability to predict pollution
dents of a retirement home without special air et al. 1999, 2001). The second method used episodes more than a few days in advance and
filtration systems and had no serious medical the Particle Concentrator-Brigham Young because of the complexity and experimental
conditions that would preclude their partici- University Organic Sampling System (PC- nature of some of the pollution monitoring, var-
pation. Medical conditions that precluded BOSS) to measure fine particulate mass, crustal ious scheduling and equipment failures meant
participation included diabetes, renal failure, and trace elements, sulfate, carbonaceous mate- that some pollution data were missing during
Parkinson’s disease, mental illness, chronic rial (elemental and organic), nitrate, semi- blocks of time when health data were being col-
alcohol abuse, treatment with oxygen therapy, volatile organic compounds, and semivolatile lected. The most consistently reliable data were
abnormal heart rhythm, pacemaker use, nitrate. The configuration and operation of the the PM2.5 FRM data. However, even these data
implanted defibrillator use, heart transplant, PC-BOSS as used in this study have been previ- had some days missing. These missing data
or heart failure within the previous 6 months. ously described in detail (Lewtas et al. 2001). were estimated and filled in by extrapolation of
Research protocols and consent forms were These monitors were co-located with the State the available data from the neighboring days
approved by the institutional review board
for human subjects at Brigham Young A 20
University. Before entering the study, all par- 60
No. of observations
ticipants read and signed consent forms and 15
PM2.5 (µg/m3)
40
completed a questionnaire pertaining to back-
ground information, medical history, and 10
20
prescription medications. Subjects received
$100 for participating in the study. 5
0
The specific days of health data collection
for each panel along with PM2.5 concentra- 0
tions are presented in Figure 1. Multiple obser- Nov 29 Dec 19 Jan 8 Jan 28 Feb 17 Mar 8 Mar 28
Hawthorne winter (1999/2000)
vations on the subjects were collected over the
study periods. The days that health end point 20
B
60
data were collected were not random because
No. of observations
of the effort to collect health data for each sub-
PM2.5 (µg/m3)
15
40
ject at least once during periods of relatively
high pollution and at least once during periods 10
20
of relatively low pollution.
Pollution and weather data. Daily data 5
0
for temperature and relative humidity were
obtained for the Salt Lake City, Utah, 0
Jun 11 Jul 1 Jul 21 Aug 10 Aug 30 Sep 19
International Airport monitoring station
Hawthorne summer (2000)
from the National Climatic Data Center
(www.ncdc.noaa.gov). In addition, the C 20
60
National Weather Service computes an air
No. of observations
stagnation index for the Wasatch Front. This 15
PM2.5 (µg/m3)
40
index, also called the clearing index, is a profile
of the atmosphere that incorporates tempera- 10
20
ture, moisture, and winds into an index that
measures the vertical and horizontal motion of 5
0
particles in the air (Jackman and Chapman
1977). The clearing index ranges from 0 to 0
Nov 17 Dec 17 Jan 16 Feb 15 Mar 17 Apr 16 May 16 Jun 15
1,050, where the values indicate the relative air
Bountiful (2000/2001)
stagnation. As climatic conditions lead to
more stagnant air, the clearing index falls, and 20
D
60
as conditions lead to less stagnant air, the
No. of observations
PM2.5 (µg/m3)
clearing index rises, usually quite rapidly. 15
40
Daily 24-hr monitoring of airborne
PM2.5 was conducted by the State of Utah 10
20
Division of Air Quality according to the U.S.
Environmental Protection Agency’s (U.S. EPA) 5
0
federal reference method (FRM; U.S. EPA
1997). Nonvolatile, PM2.5 mass concentrations 0
Nov 17 Dec 17 Jan 16 Feb 15 Mar 17 Apr 16 May 16 Jun 15
were determined using tapered element Lindon (2000/2001)
oscillating microbalance (TEOM) monitors
Figure 1. PM2.5 concentrations and number of observations of health end points for (A) Hawthorne winter,
(Patashnick and Rupprecht 1991). Total PM2.5
(B) Hawthorne summer, (C) Bountiful, and (D) Lindon. Black dots represent PM2.5 data collected by the State
mass, including semivolatile mass (SVM), was of Utah Division of Air Quality according to the U.S. EPA’s FRM. Green dots represent missing data that
measured using two sampling methods. The were estimated using neighboring days and the clearing index. Bars represent the number of observations
first was a real-time total ambient mass sampler of health end points collected on given days. Black, red, green, yellow, and blue bars represent the first,
(RAMS), based on diffusion denuder, Nafion second, third, fourth, and fifth time period of health data collection, respectively.
340 112 | NUMBER 3 | March 2004 • Environmental Health Perspectives
VOLUME
3. |
Article Particulate air pollution, HRV, and inflammation
if there were no large changes in the clearing technician in their homes. The hookup of a c) r-MSSD, the square root of the mean of
index, or by projection of data consistent with modified V5 and VF bipolar lead placement the squared differences between adjacent
observed changes in the clearing index. As were used, and skin preparation, electrode NN intervals. SDNN is an estimate of overall
reported elsewhere (Eatough et al. 2003), the placement, and related protocols were similar to HRV, SDANN reflects variability due to
RAMS and PC-BOSS monitors provided those described elsewhere (Marquette Medical cycles longer than 5 min and is an estimate of
nearly equivalent estimates of the PM2.5 mass, Systems 1996). ECGs were recorded digitally long-term components of HRV, and r-MSSD
including SVM. For this analysis, the averages (sampling rate, 256 Hz/channel) on removable reflects high-frequency variations and is an esti-
of the nonmissing values from 24-hr concen- flash cards using a lightweight, two-channel, mate of short-term components of HRV. A
trations calculated from RAMS and PC-BOSS ambulatory ECG monitor (Trillium3000; detailed discussion of the physiologic interpre-
were used as estimates of total fine particulate Forest Medical, East Syracuse, NY). The ECG tation of these measures is presented elsewhere
mass. Figure 1 illustrates the PM2.5 concentra- digital recordings were processed, and mean (Task Force 1996).
tions using FRM-filled data and the time of heart rate and various measures of HRV were Blood collection and analysis. Immediately
health data collection for each panel. Figure 2 calculated for each 24-hr session using PC- after each 24-hr ECG monitoring period,
presents PM 2.5 measurements from FRM- based software (Trillium3000 PC Companion blood was drawn using standardized proce-
filled, TEOM, and RAMS/PC-BOSS moni- Software for MS Windows; Forest Medical). dures for venipuncture, collection, storage,
tors plotted with the clearing index. Three measures of HRV were calculated and and shipment. Blood cell counts and differen-
Ambulatory ECG monitoring. Repeated used in this analysis: a) SDNN, the standard tial white cell counts were determined using
24-hr ambulatory ECG monitoring was con- deviation of all normal R-R (or NN) intervals an automated cell counter with flow differen-
ducted on the subjects during periods of during the 24-hr period; b) SDANN, the stan- tial (Advia 120 hematology system; Bayer
both low and high air pollution. Participants dard deviation of the average NN intervals in Corporation, Leverkusen, Germany); C-reac-
were hooked up to the monitors by a trained all 5-min segments of the 24-hr period; and tive protein (CRP) was measured using neph-
elometry (IMMAGE Immunochemistry
Systems, CRP Kit 447280; Beckman Coulter
5,000
A
Instruments, Brea, CA); whole blood viscosity
60
4,000
Clearing index
was measured using a cone-plate viscometer
PM2.5 (µg/m3)
40
(model DZ-2+; Brookfield Engineering
3,000
Laboratories Inc., Middleboro, MA). All of
20
2,000
the blood tests were conducted at ARUP
0
Laboratories in Salt Lake City.
1,000
Analytic methods. As illustrated in Figure 1,
–20
0
we attempted to collect health data during
Nov 29 Dec 19 Jan 8 Jan 28 Feb 17 Mar 8 Mar 28
times of both high and low PM2.5 concentra-
Hawthorne winter (1999/2000)
5,000
tions. Associations for each of the heart rate,
B
HRV, and blood measures were evaluated sta-
60
Clearing index
4,000
tistically using various regression models. To
PM2.5 (µg/m3)
40
3,000
account for subject-specific differences two
approaches were used: a) Deviations from each
20 2,000
individual subject’s mean were calculated and
0 1,000
used as the dependent variables in the regres-
sion models, and b) subject-specific fixed effects
–20 0
Jun 11 Jul 1 Jul 21 Aug 10 Aug 30 Sep 19 (Greene 2000) were included in the regression
Hawthorne summer (2000) models. Three approaches were used to control
5,000
for weather variables: a) Both linear and qua-
C
60
Clearing index
dratic terms for daily average temperature and
4,000
PM2.5 (µg/m3)
relative humidity were included in the models,
40 3,000
b) additive cubic smoothing splines with 3
20
degrees of freedom (df) for average temperature
2,000
and relative humidity were included in the
0 1,000
models, and c) cubic smoothing splines with a
functional two-way interaction between average
–20 0
Nov 17 Dec 17 Jan 16 Feb 15 Mar 17 Apr 16 May 16 Jun 15
temperature and relative humidity using 6 df
Bountiful (2000/2001)
were included in the models. Mean heart rate
5,000
D
was included in some of the models for SDNN,
60
Clearing index
4,000
SDANN, and r-MSSD. Although most of the
PM2.5 (µg/m3)
40
models included PM2.5 as a linear term, models
3,000
that included cubic smoothing splines with 3 df
20 2,000
for PM2.5 were also estimated. The statistical
analysis was conducted using SAS statistical
0 1,000
software (SAS Institute 2001) estimating the
–20 0
fully parametric regression models using PROC
Nov 17 Dec 17 Jan 16 Feb 15 Mar 17 Apr 16 May 16 Jun 15
REG and estimating the regression models that
Lindon (2000/2001)
included cubic smoothing splines using the
Figure 2. PM2.5 measurements for (A) Hawthorne winter, (B) Hawthorne summer, (C) Bountiful, and (D)
“spline” and “spline2” functions in the model
Lindon; from FRM-filled (thick black lines), RAMS/PC-BOSS (red), and TEOM (blue) monitors plotted with
statement of PROC GAM.
the clearing index (thin black lines).
341
Environmental Health Perspectives • VOLUME 112 | NUMBER 3 | March 2004
4. |
Article Pope et al.
We conducted residual analysis of the As Table 1 and Figure 1 show, during the However, the associations between HRV and
estimated regression models by labeling resid- study period PM 2.5 pollution levels rarely these weather variables were nonlinear and
uals and partial residuals by subject number exceeded the U.S. National Ambient Air likely interactive. For example, model V
Quality 24-hr standard of 65 µg/m3 (U.S. EPA
and plotting them over pollution levels. We included a cubic smoothing spline that allowed
then evaluated the sensitivity of the results to 1996, 1997). The mean levels of the HRV and for both nonlinearity and two-way interactions
influential subjects by estimating the regres- blood measures were not remarkable. with temperature and relative humidity. The
sion models with the influential subjects Table 2 presents the regression coefficients chi-squared test for this smoothing spline indi-
excluded. Also, we estimated models for dif- for five alternative regression models. Elevated cated a statistically significant fit (p < 0.05) for
ferent lagged days of exposure and for PM2.5 concentrations of PM2.5 were consistently asso- SDNN and r-MSSD. Nevertheless, negative
concentrations measured by the alternative ciated with declines in all three measures of associations between PM2.5 and HRV were
monitoring approaches. HRV, SDNN, SDANN, and r-MSSD. As clearly observed even when accounting for
expected, there were substantial cross-subject subject-specific differences and controlling
Results differences in these HRV measures. There were for temperature and relative humidity using
Summary statistics for the pollution, weather, also statistically significant associations between various modeling approaches.
and health variables are presented in Table 1. HRV and temperature and relative humidity. Elevated concentrations of PM2.5 were
associated with increases in CRP and mono-
Table 1. Summary statistics of key variables used in analysis.
cytes. As with the HRV measures, these asso-
Reference rangea ciations were not highly sensitive to the
Variable Unit No. Mean ± SD Min–Max
various modeling approaches presented in
µg/m3
PM2.5 (FRM-Filled) — 250 23.7 ± 20.2 1.7–74.0
Table 2. PM2.5 was not significantly associ-
µg/m3
PM2.5 (not filled) — 182 25.8 ± 21.2 1.7–74.0
µg/m3
PM2.5 (TEOM) — 223 18.9 ± 13.4 2.2–61.5 ated with whole-blood viscosity, other white
µg/m3
PM2.5 (RAMS/PC-BOSS) — 236 26.5 ± 18.8 5.6–72.4 blood cell counts, red blood cells, or platelets.
Average temperature °F — 250 43.3 ± 20.8 19.0–87.0
Figure 3 presents partial residuals (after
Average relative humidity % — 250 67.5 ± 18.9 24.5–92.0
controlling differences in individual means and
Mean heart rate bpm — 247 72.1 ± 9.8 50–104
for temperature and relative humidity) from
SDNN msec — 246 131.4 ± 43.1 45–317
model I for SDNN and CRP. For conve-
SDANN msec — 246 109.0 ± 38.1 37–376
r-MSSD msec — 246 69.7 ± 59.5 14–323 nience, the residuals in Figure 3 are labeled
CRP mg/dL 0.0–0.8 244 0.50 ± 0.60 0.10–7.8 by subject number and plotted over PM2.5.
Whole blood viscosity cP 3.6–6.0 248 5.15 ± 0.76 0.80–8.3
Similar partial residual plots are obtainable
White blood cells k/µL 3.2–10.6 250 6.99 ± 1.59 3.60–14.11
using the other four modeling approaches. Full
Granulocytes k/µL 1.3–7.0 237 4.26 ± 1.20 2.10–11.10
residual plots for all of the end points were also
Lymphocytes k/µL 0.8–3.1 237 2.14 ± 0.72 0.50–4.60
created and analyzed. As was observed in the
Monocytes k/µL 0.1–0.5 237 0.41 ± 0.16 0.10–1.30
Basophils k/µL 0.0–0.1 237 0.05 ± 0.05 0.00–0.20 residual plots and as shown in Figure 3, some
Eosinophils k/µL 0.0–0.4 237 0.19 ± 0.11 0.00–0.70 subjects provided highly influential observa-
Red blood cells M/µL 4.69–6.07 250 4.75 ± 0.43 3.32–6.16
tions. For CRP, one subject (subject 15) was
Platelets k/µL 177–406 250 249.3 ± 58.7 101.0–457.0
clearly highly influential. A single influential
Abbreviations; bpm, beats per minute; cP, centipoise; k/µL, thousands per microliter; Max, maximum; Min, minimum; subject was also observed for mean heart rate
M/µL, millions per microliter.
and monocytes (subjects 77 and 5, respectively;
aInterpretive normal reference ranges for blood parameters assessed by ARUP Laboratories.
Table 2. PM2.5 regression coefficients [×100 (SEs)] for various regression models using all subjects and FRM-filled PM2.5 data.
Model parameters I II III IV V
Control for Dependent Subject-specific Subject-specific Subject-specific Subject-specific
cross-subject variables are fixed effects fixed effects fixed effects fixed effects
differences deviations from
individual means
Control for Quadratic Quadratic Additive spline Additive spline Interactive spline
functionsa for functionsa for
other factors smooths for smooths for smooths for temp,
temp, RHb temp, RHb plus
temp, RH temp, RH RH; partial
control for HRc control for HRd
Mean heart rate –3.83 (2.5) –5.23 (3.47) –5.14 (2.42)** — –4.49 (1.73)**
–34.36 (12.13)# –41.71 (11.67)# –34.94 (8.32)#
SDNN –31.45 (12.27)** –37.76 (17.01)**
SDANN –16.80 (12.55) –19.92 (17.46) –17.42 (12.43) –22.93 (12.20)* –18.98 (8.67)**
–45.82 (15.42)# –51.15 (15.27)# –42.25 (10.90)#
r-MSSD –39.31 (15.64)** –50.59 (21.60)**
0.78 (0.25)# 0.95 (0.35)# 0.96 (0.25)# 0.81 (0.18)#
CRP —
Whole blood viscosity 0.10 (0.30) 0.12 (0.42) –0.00 (0.30) — 0.07 (0.21)
White blood cells 0.15 (0.55) 0.25 (0.75) 0.21 (0.53) — –0.07 (0.38)
Granulocytes 0.21 (0.53) 0.38 (0.72) 0.33 (0.51) — 0.02 (0.37)
Lymphocytes –0.06 (0.20) –0.09 (0.28) –0.06 (0.20) — –0.07 (0.14)
0.12 (0.04)#
Monocytes 0.13 (0.06)** 0.15 (0.08)** 0.14 (0.05)** —
Basophils –0.02 (0.02) –0.02 (0.03) –0.01 (0.02) — –0.01 (0.01)
Eosinophils –0.03 (0.03) –0.03 (0.04) –0.02 (0.03) — –0.01 (0.02)
Red blood cells 0.04 (0.09) 0.03 (0.13) 0.00 (0.09) — 0.03 (0.06)
Platelets 0.58 (13.33) –1.18 (18.49) 2.29 (13.14) — 0.31 (9.34)
Abbreviations; RH, relative humidity; temp, temperature.
aIncludes linear and quadratic terms for average temperature and relative humidity. bIncludes cubic smoothing splines with 3 df for average temperature and relative humidity. cIncludes
24-hr mean heart rate. dIncludes a cubic smoothing spline with a functional two-way interaction between average temperature and relative humidity using 6 df. Models for SDNN,
SDANN, and r-MSSD also includes mean heart rate. *p ≤ 0.10; **p ≤ 0.05; #p ≤ 0.01.
342 112 | NUMBER 3 | March 2004 • Environmental Health Perspectives
VOLUME
5. |
Article Particulate air pollution, HRV, and inflammation
residual plots not shown). For the three meas- decreasing for longer exposure lag times. For inflammation, as measured by various blood
ures of HRV (only SDNN shown in Figure 3), CRP, the concurrent-day exposures were also markers of inflammation, would be associated
two subjects (89 and 66) appeared to be most most strongly associated with CRP, but a with ambient concentrations of PM 2.5 .
influential, with two other subjects (36 and 77) relatively large effect was also observed for the Reasonably consistent and statistically robust
successively less so. For the other end points, 3 days’ previous exposure. negative associations between PM2.5 and meas-
no influential subjects were clearly observed. As shown in Figures 1 and 2, for all meas- ures of HRV were observed. PM2.5 was signifi-
Table 3 presents the PM2.5 regression coeffi- ures of PM2.5 during prolonged periods with a cantly associated with CRP but not with
cients using model V excluding various num- very low and stable clearing index (indicating changes in white blood cell counts, red blood
bers of influential subjects. For the measures of very stagnant air conditions), PM2.5 concentra- cells, platelets, or whole-blood viscosity. The
HRV, excluding influential subjects produced tions tended to rise and then drop rapidly with PM2.5 associations with CRP are interesting
somewhat smaller estimated PM2.5 effects, but the eventual and inevitable rapid rise in the and suggestive, but given that they depend
the effects remained negative and generally sta- clearing index. Using model V, measures of largely on a highly influential subject, they are
tistically significant. For CRP, excluding the HRV and CRP were also regressed on other not compelling.
single highly influential subject produced a estimates of concentrations of PM2.5, including Previous studies have reported HRV asso-
substantially smaller estimated PM2.5 effect FRM without filling in missing values and ciations with ambient PM air pollution expo-
that was marginally significant (p = 0.06). For the TEOM and RAMS/PC-BOSS measures sure (Gold et al. 2000; Liao et al. 1999; Pope
mean heart rate and monocytes, excluding the (results not shown). For the FRM (not filled) et al. 1999), occupational PM exposure
influential subjects resulted in reduced and and for TEOM, PM2.5 effect estimates were (Magari et al. 2001), and PM exposure from
statistically insignificant associations with PM2.5. nearly identical to those reported in Table 2. secondhand cigarette smoking in an airport
As illustrated for SDNN and CRP in For the PM2.5 concentrations measured using smoking lounge (Pope et al. 2001). A previ-
Figure 3, associations with PM2.5 were nearly RAMS/PC-BOSS that included SVM, the ous study that used 24-hr ambulatory ECG
linear. Goodness-of-fit tests, based on models results were similar for CRP, but statistically monitoring was also conducted along the
that included cubic smoothing splines with significant negative associations were not Wasatch Front in Utah (Pope et al. 1999), but
3 df for PM2.5, indicated that PM2.5 associa- observed for measures of HRV. However, it had only seven participants with a total of
tions with r-MSSD but not SDNN, SDANN, there were 10 days with abnormally high levels 39 observations and had air pollution monitor-
ing only for PM ≤ 10 µm in aerodynamic
of SVM (> 15 µg/m3), caused by abnormally
or CRP were significantly different from linear
(p < 0.05). However, after influential observa- elevated concentrations of ammonium nitrate. diameter (PM10). This study estimated that a
100-µg/m3 increase in PM10 was associated
tions were removed, no associations between When these 10 days were deleted from the
PM 2.5 and SDNN, SDANN, r-MSSD, or analysis, statistically significant negative associ- with an 18-msec decrease in 24-hr SDNN.
CRP were significantly different from linear ations, similar to those presented in Table 2, Assuming a PM2.5:PM10 ratio of 0.60, this
(p > 0.25). were observed for PM 2.5 from the RAMS/ would suggest a 30-msec decline in 24-hr
SDNN associated with a 100-µg/m3 increase in
Table 4 presents regression coefficients PC-BOSS monitors.
with model V using different lagged days of PM2.5—an estimate remarkably close to that
Discussion
exposure. The strongest associations between observed in this study. However, unlike in the
PM2.5 and HRV were with concurrent-day In this study we hypothesized that altered present analysis, PM pollution was associated
exposures with effect estimates generally cardiac autonomic function, as indicated by with increases in 24-hr r-MSSD. Another study
changes in measures of HRV, and pulmonary conducted in Boston, Massachusetts, estimated
100 A
SDNN, partial residuals
Table 3. PM2.5 (FRM-filled) regression coefficients [×100 (SEs)] using model Va, excluding various numbers
of influential subjects.
50
Excluding two Excluding three Excluding four
0
Excluding most most influential most influential most influential
influential subjectb subjectsb subjectsb subjectsb
–50
Mean heart rate –1.66 (1.55) — — —
–23.70 (7.13)# –31.01 (6.75)# –26.39 (6.84)#
SDNN —
–100
–23.09 (6.31)# –19.94 (6.41)#
SDANN — –14.37 (6.96)**
–25.63 (8.95)# –34.96 (8.45)# –21.77 (7.96)#
r-MSSD —
0 10 20 30 40 50 60 70
CRP 0.19 (0.10)* — — —
Monocytes 0.02 (0.03) — — —
B
4
aThis model included subject-specific fixed effects and cubic smoothing splines with a functional two-way interaction
CRP, partial residuals
between average temperature and relative humidity using 6 df. Models for SDNN, SDANN, and r-MSSD also included
mean heart rate. bThese models excluded highly influential subjects as determined by residual analysis. Only one subject
2
was excluded for mean heart rate (subject 77), CRP (15), and monocytes (5). For the HRV measures, SDNN, SDANN, and
r-MSSD, initially two subjects were excluded (66 and 89), then three (36, 66, 89), and then four (36, 66, 89, 77). *p ≤ 0.10;
0
**p ≤ 0.05; #p ≤ 0.01.
Table 4. PM2.5 (FRM-filled) regression coefficients [×100 (SEs)] using model Va using different lagged days
–2
of pollution exposure.
–4 Concurrent day Previous day Two days previous Three days previous
0 10 20 30 40 50 60 70
–34.94 (8.32)#
SDNN –19.37 (9.21)** –17.03 (9.62)* –10.22 (10.12)
PM2.5 (µg/m3) SDANN –18.98 (8.67)** –10.22 (9.51) –6.80 (9.94) –1.52 (10.43)
–42.25 (10.90)#
r-MSSD –26.27 (12.02)** –27.66 (12.54)** –20.77 (13.20)
Figure 3. Partial residuals (after controlling for
0.81 (0.18)# 0.64 (0.21)#
CRP 0.33 (0.20)* 0.38 (0.20)*
subject-specific differences, temperature, and rela-
tive humidity) from model I for (A) SDNN and (B) aThis model included subject-specific fixed effects and cubic smoothing splines with a functional two-way interaction
CRP plotted over PM2.5. Residuals are labeled by between average temperature and relative humidity using 6 df. Models for SDNN, SDANN, and r-MSSD also included
mean heart rate. *p ≤ 0.10; **p ≤ 0.05; #p ≤ 0.01.
subject number.
343
Environmental Health Perspectives • VOLUME 112 | NUMBER 3 | March 2004
6. |
Article Pope et al.
the effect of average PM2.5 levels over the pre- approximately 28,000 American women, the meeting a measurement challenge. Atmos Environ
37:1277–1292.
vious 4 hr on SDNN and r-MSSD during a relative risk of a first cardiovascular event
Eatough DJ, Obeidi F, Pang Y, Ding Y, Eatough NL, Wilson WE.
25-min period of alternating rest and exercise increased with higher quintiles of CRP. The 1999. Integrated and real-time diffusion denuder samplers
in a panel of 21 participants 53–87 years of age CRP range for the fourth quintile reported by for PM 2.5 based on BOSS, PC and TEOM technology.
with a total of 163 monitoring sessions (Gold Ridker et al. (2002) was 2.09–4.19 mg/L. The Atmos Environ 33:2835–2844.
Ghio AJ, Devlin RB. 2001. Inflammatory lung injury after
et al. 2000). A 100-µg/m3 increase in exposure median CRP in our present study (0.4 mg/dL bronchial instillation of air pollution particles. Am J Respir
to PM2.5 was associated with an approximately or 4 mg/L), therefore, would fall within the Crit Care Med 164:704–708.
24-msec decline in both SDNN and r-MSSD. upper range of this fourth quintile. Also, the Glantz SA. 2002. Air pollution as a cause of heart disease: time
for action. J Am Coll Cardiol 39:943–945.
The physiologic importance of these estimated increase in CRP associated with a Godleski JJ, Verrier RL, Koutrakis P, Catalano P, Coull B,
100-µg/m3 increase in PM2.5 reported in this
observed changes in HRV is not fully under- Reinisch U, et al. 2000. Mechanisms of morbidity and mor-
stood, yet there is growing recognition of the study, even excluding the highly influential tality from exposure to ambient air particles. Res Rep
Health Inst 91:5–88.
role of autonomic dysfunction in cardiovascu- observation (0.19 mg/dL or 1.9 mgL), was Gold DR, Litonjua A, Schwartz J, Lovett E, Larson A, Nearing B,
lar mortality, and HRV measures provide approximately equal to or greater than the et al. 2000. Ambient pollution and heart rate variability.
quantitative, well-defined indicators of cardiac range of the CRP quintiles reported by Ridker Circulation 101:1267–1273.
Greene WH. 2000. Econometric Analysis. 4th ed. Upper Saddle
autonomic function (Task Force 1996). For et al. (2002). However, it is uncertain how
River, NJ:Prentice-Hall.
example, decreases in HRV are strong predic- these acute changes in CRP effect or reflect Hoek G, Dockery DW, Pope CA III, Neas L, Roemer W,
tors of mortality (Kennedy 1997; La Rovere potential changes in cardiovascular risk. Brunekreef B. 1998. Association between PM10 and decre-
ments in peak expiratory flow rates in children: reanalysis
et al. 2003), and autonomic nervous system– The analysis using the RAMS/PC-BOSS
of data from 5 panel studies. Eur Respir J 11:1307–1311.
activated changes in HRV may increase the data observed that the PM–HRV associations Jackman DN, Chapman WT. 1977. Some meteorological
likelihood of sudden cardiac death (Task Force were much smaller on days with high SVM, aspects of air pollution in Utah with emphasis on the Salt
1996). In addition, decreased 24-hr SDNN especially if the SVM was rich in nitrate. Lake Valley. Technical Memorandum NWS WR-120. Salt
Lake City, UT:National Oceanic and Atmospheric
has been associated with all-cause mortality These results are not well understood, but Administration, National Weather Service, Western
and has been found to be a significant predic- they imply that nitrate may have a smaller Region.
tor of death due to progressive heart failure in impact than do other fine particles or that Kennedy HL. 1997. Beta blockade, ventricular arrhythmias, and
sudden cardiac death. Am J Cardiol 80:29J–34J.
patients with chronic heart failure (Nolan et al. nitrate exposure is not well accounted for La Rovere MT, Pinna GD, Maestri R, Mortara A, Capomolla S,
1998). In a panel of 433 participants with a by central-site monitoring (Patterson and Febo O, et al. 2003. Short-term heart rate variability
mean age of 62 years, a 41.2-msec decrease in Eatough 2000). The health associations for strongly predicts sudden cardiac death in chronic heart
failure patients. Circulation 107:565–570.
SDNN from the mean SDNN of study par- PM2.5 measured by TEOM (which includes Lewtas J, Booth D, Pang Y, Reimer S, Eatough DJ, Gunde L.
ticipants (113.4) was associated with a relative no SVM), PM2.5 measured by FRM (which 2001. Comparison of sampling methods for semi-volatile
risk of 1.62 for all-cause mortality and a rela- includes some SVM), and PM2.5 measured by organic carbon (SVOC) associated with PM2.5. Aerosol Sci
Technol 34:9–22.
tive risk of 2.45 for progressive heart failure. RAMS/PC-BOSS (which is designed to meas-
Liao D, Creason J, Shy C, Williams R, Watts R, Zweidinger R.
In our present analysis, similar declines in ure total PM including SVM) after deleting 1999. Daily variation of particulate air pollution and poor
SDNN were associated with an increase in days with abnormally high levels of ammo- cardiac autonomic control in the elderly. Environ Health
exposure of approximately 100 µg/m3 PM2.5, Perspect 107:521–525.
nium nitrate were all similar. These results
Magari SR, Hauser R, Schwartz J, Williams PL, Smith TJ,
but it is unknown how these acute declines in suggest that organic semivolatiles may have Christiani DC. 2001. Association of heart rate variability
HRV may reflect increased risk for adverse effects similar to those of the nonvolatile PM with occupational and environmental exposure to particu-
cardiovascular events. and point toward the need for further particle late air pollution. Circulation 104:986–991.
Marquette Medical Systems. 1996. EKG Lead Configurations:
The suggested associations between PM2.5 characterization in health studies of PM pollu- Series 8500 Ambulatory Tape Recorders Operator’s
and CRP are intriguing. The MONICA- tion. Additional study with other pollution Manual. Milwaukee, WI:Marquette Medical Systems.
Augsburg study also reported evidence of measures including gaseous pollutants may be Nolan J, Batin PD, Andrews R, Lindsay SJ, Brooksby P, Mullen M,
et al. 1998. Prospective study of heart rate variability and
increased CRP associated with PM air pollu- important. mortality in chronic heart failure: results of the United
tion (Peters et al. 2001). A 3-fold increase in Although in this study we observed statis- Kingdom heart failure evaluation and assessment of risk trial
the odds of having CRP > 5.7 mg/L was asso- tical associations between PM2.5 and HRV (UK-heart). Circulation 98:1510–1516.
Patashnick H, Rupprecht EG. 1991. Continuous PM10 measure-
ciated with a marked pollution episode. Using and CRP, most of the relevant variability in ments using the tapered element oscillating microbalance.
multivariate regression analysis, the study also the temporal deviations of these physiologic J Air Waste Manag Assoc 41:1079–1083.
found an increase in CRP of 0.88 mg/L asso- end points was not explained by PM2.5. The Patterson E, Eatough DJ. 2000. Indoor/outdoor relationships for
ciated with an increase of 26 µg/m3 in the ambient PM2.5 and associated pollutants: epidemiological
fact that small but statistically significant asso-
implications in Lindon, Utah. J Air Waste Manage Assoc
previous 5-day average concentration of total ciations were observed suggests that PM2.5 50:103–110.
suspended particles, which converts to a CRP may be one of multiple factors that influence Peters A, Doring A, Wichmann HE, Koenig W. 1997. Increased
increase of 0.34 mg/dL for a 100-µg/m 3 plasma viscosity during an air pollution episode: a link to
HRV and CRP. These results suggest that
mortality? Lancet 349:1582–1587.
increase in the 5-day average total suspended altered cardiac autonomic function and pul- Peters A, Frohlich M, Doring A, Immervoll T, Wichmann H-E,
particles. This estimated change is roughly monary/systemic inflammation may play a Hutchinson WL, et al. 2001. Particulate air pollution is
comparable with changes observed in the role in pathogenesis relating to the harmful associated with an acute phase response in men; results
from the MONICA-Augsburg Study. Eur Heart J
present analysis. Seaton et al. (1999) also effects of PM air pollution on human health. 22:1198–1204.
reported a positive association between PM10 Pope CA III. 2000. Epidemiology of fine particulate air pollution
REFERENCES
and CRP. A 147% (95% confidence interval, and human health: biologic mechanisms and who’s at
risk? Environ Health Perspect 108(suppl 4):713–723.
20–477) increase in CRP was associated with Pope CA III, Burnett RT, Thun MJ, Calle EE, Krewski D, Ito K,
a 100-µg/m3 increase in PM10. Committee of the Environmental and Occupational Health
et al. 2002. Lung cancer, cardiopulmonary mortality, and
Assembly of the American Thoracic Society. 1996. Health
Recently, much attention has been placed long-term exposure to fine particulate air pollution. JAMA
effects of outdoor air pollution. Am J Respir Crit Care Med
287:1132–1141.
on CRP as an indicator of acute phase response 153:3–50.
Pope CA III, Dockery DW. 1999. Epidemiology of particle
Eatough DJ, Eatough NL, Obeidi F, Pang Y, Modey WK, Long
and as a risk factor for cardiovascular events. effects. In: Air Pollution and Health (Holgate ST, Samet
RW. 2001. Continuous determination of PM2.5 mass, includ-
Remarkably small changes in CRP have been JM, Koren HS, Maynard RL, eds). London:Academic Press,
ing semi-volatile species. Aerosol Sci Technol 34:1–8.
671–705.
associated with changes in cardiac risk. Eatough DJ, Long RW, Modey WK, Eatough NL. 2003. Semi-
Pope CA III, Eatough DJ, Gold DR, Pang Y, Nielsen KR, Nath P,
In a study by Ridker et al. (2002) involving volatile secondary organic aerosol in urban atmospheres:
344 112 | NUMBER 3 | March 2004 • Environmental Health Perspectives
VOLUME
7. |
Article Particulate air pollution, HRV, and inflammation
et al. 2001. Acute exposure to environmental tobacco and GRO-α production in healthy human airways. Am J pollution caused by forest fires. Am J Respir Crit Care Med
smoke and heart rate variability. Environ Health Perspect Respir Crit Care Med 161:550–557. 161(4 pt 1):1213–1217.
109:711–716. SAS Institute. 2001. The SAS System for Windows, Release Task Force of the European Society of Cardiology and the North
Pope CA III, Verrier RL, Lovett EG, Larson AC, Raizenne ME, 8.02. Cary, NC:SAS Institute Inc. American Society of Pacing and Electrocardiography. 1996.
Kanner RE, et al. 1999. Heart rate variability associated with Schwartz J. 2001. Air pollution and blood markers of cardiovas- Heart rate variability: standards of measurement, physiologi-
particulate air pollution. Am Heart J 138(5 pt 1):890–899. cular risk. Environ Health Perspect 109(suppl 3):405–409. cal interpretation, and clinical use. Circulation 93:1043–1065.
Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. 2002. Seaton A, MacNee W, Donaldson K, Godden D. 1995. U.S. Environmental Protection Agency. 1996. National ambient
Comparison of C-reactive protein and low-density lipopro- Particulate pollution and acute health effects. Lancet air quality standards for particulate matter. Fed Reg
tein cholesterol levels in the prediction of first cardiovas- 345:176–178. 61(241):65638.
cular events. N Engl J Med 347:1557–1565. Seaton A, Soutar A, Crawford V, Elton R, McNerlan S, Cherrie J, ———. 1997. Revised requirements for the designation of
Salvi S, Blomberg A, Rudell B, Kelly F, Sandstrom T, Holgate ST, et al. 1999. Particulate air pollution and the blood. Thorax reference and equivalent methods for PM2.5 and ambient
et al. 1999. Acute inflammatory responses in the airways 54:1027–1032. air quality surveillance for particulate matter; final rule.
and peripheral blood after short-term exposure to diesel Stone PH, Godleski JJ. 1999. First steps toward understanding Fed Reg 62(138):5725–5726.
exhaust in healthy human volunteers. Am J Respir Crit Care the pathophysiologic link between air pollution and car- Utell MJ, Frampton MW, Zareba W, Devlin RB, Cascio WE.
Med 159:702–709. diac mortality. Am Heart J 138(5 pt 1):804–807. 2002. Cardiovascular effects associated with air pollution:
Salvi SS, Nordenhall C, Blomberg A, Rudell B, Pourazar J, Kelly FJ, Tan WC, Qiu D, Liam BL, Ng TP, Lee SH, van Eeden SF, et al. potential mechanisms and methods of testing. Inhal
et al. 2000. Acute exposure to diesel exhaust increases IL-8 2000. The human bone marrow response to acute air Toxicol 14:1231–1247.
345
Environmental Health Perspectives • VOLUME 112 | NUMBER 3 | March 2004