Air pollution and emergency department visits for chest pain and weakness in Edmonton, Canada.

IJOMEH 2010;23(1) 15
O R I G I N A L P A P E R S
International Journal of Occupational Medicine and Environmental Health 2010;23(1):15 – 19
DOI 10.2478/V10001-010-0001-z
AIR POLLUTION AND EMERGENCY DEPARTMENT
VISITS FOR CHEST PAIN AND WEAKNESS
IN EDMONTON, CANADA
MIECZYSŁAW SZYSZKOWICZ1 and BRIAN ROWE2
1 Air Health Effects Research Section, Health Canada,
Ottawa, Ontario, Canada
2 Department of Emergency Medicine, University of Alberta,
Edmonton, Alberta, Canada
Abstract
Objectives: Chest pain or weakness can be first signal of health problems. Many studies demonstrate that these condi-
tions can be related to air pollution. This work uses time-series data to investigate the association. Material and Methods:
This is a study of 68 714 emergency department (ED) visits for chest pain (ICD-9: 786) and of 66 092 ED visits for weak-
ness (ICD-9: 780). The hierarchical method was applied to analyse the associations between daily counts of ED visits
for chest pain and weakness (separately) and the levels of the air pollutants and meteorological variables. The counts
of visits for all patients, males and females were analysed separately by whole period (I–XII), warm (IV–IX) and cold
(X–III). Results: The results are presented in the form of the excess risks associated with an increase in the interquartile
range (IQR) for the pollutant. Chest pain: 2.4% (95% CI: 1.0–3.9) for CO, females, I–XII; 3.8% (95% CI: 0.0–7.8)
for NO2, males, IV–IX; 4.5% (95% CI: 0.9–8.3) for O3 (1-day lagged), males, IV–IX; 2.8% (95% CI: 0.5–5.2), for PM10,
males, X–III; 2.0% (95% CI: 0.0–4.0), for SO2, females, X–III; 2.1% (95% CI: 0.2–4.0) for PM2.5, all, X–III. Weak-
ness: 2.1% (95% CI: 0.4–3.7) for CO (2-day lagged), males, X–III; 3.4% (95% CI: 1.0–5.9) for NO2 (2-day lagged), males,
X–III; 2.4% (95% CI: 0.9–3.9) for SO2, females, I–XII; 4.6% (95% CI: 1.0–8.2) for O3 (1-day lagged), females, IV–IX.
Conclusions: Obtained findings provide support for the hypothesis that ED visits for chest pain and weakness are associated
with exposure to ambient air pollution.
Key words:
Emergency department visit, Chest pain, Weakness, Air pollution, Temperature, Relative humidity
Received: April 12, 2007. Accepted: September 3, 2007.
Address reprint request to M. Szyszkowicz, Air Health Effects Research Section, Environmental Health Science and Research Bureau, Health Canada, 269 Laurier
Avenue, Room 3-030, Ottawa, ON, K1A 0K9, Canada (e-mail: mietek_szyszkowicz@hc-sc.gc.ca).
INTRODUCTION
Chest pain is a common presentation to the emergency de-
partment (ED). Non-specific chest pain or weakness can
be early signals of cardio-pulmonary disease. This study
is based on 10 years daily summarized counts of ED visits
for chest pain and weakness. The study takes in account
ambient air pollution exposures and meteorological fac-
tors. ED data were linked to concentrations of ambient air
pollutants and weather variables. We constructed models
for different air pollutants: gases (SO2, NO2, CO and O3)
and particulate matters (PM2.5 and PM10) — particles with
median aerodiameter of 2.5 and 10.0 μm or less, respec-
tively. The goal is to verify a hypothesis that environmen-
tal exposures (exposure to air pollution and/or to mete-
orological conditions) may be associated with ED visits for
chest pain or weakness. Chest pain or weakness can be
early signals of cardio-pulmonary symptoms.
There are no studies on the associations between ED vis-
its for chest pain or weakness and ambient air pollution.
Here we use a time-series methodology to assess associa-
tions between air pollution and chest pain and weakness
sufficiently serious to warrant an ED visit. A large num-
ber of time series studies of air pollution already exist, but

O R I G I N A L P A P E R S M. SZYSZKOWICZ ET AL.
IJOMEH 2010;23(1)16
To relate short-term effects of air-pollution and weather
factors to the number of daily ED visits for chest pain and
weakness we applied a generalized linear mixed models
(GLMM) methodology [3]. We first defined clusters based
on the following triplet (year, month, day of week). Our
clusters may have 4 or 5 observations (days). The clusters
have a hierarchical structure: days are nested in days of
the week, which are nested in months, and months are
nested in years. The days of the same day of the week, in
the same month and in the same year belong to the same
cluster. We applied a Poisson model to clustered counts. In
our models we assumed fixed slope and random intercept
on the constructed clusters. The method was already pro-
posed to analyse air pollution impact on health outcomes
[4]. Among existing software which realizes this method
we have chosen the glmmPQL function from the R statis-
tical package [5].
Relative risks of chest pain and weakness ED visits at-
tributable to the single pollutant and weather factors us-
ing current day exposure level, 1-day and 2-day lagged
shared exposure levels were estimated for an increase in
value of current day interquartile range (IQR). Results
are expressed as excess risk (%RR = (RR-1)×100%,
where RR is relative risk): percentage changes in daily
visits associated with the pollutant and weather factors.
The 95% confidence intervals (95% CI) were also cal-
culated.
RESULTS
The results are presented in four tables and two figures.
Table 1 contains the number of ED visits for chest pain
and weakness by age and sex. Chest pain: Of the 68 714 to-
tal visits in the study, 51.1% (n = 35 134) occurred among
males. Between 1992 and 2002, the percentage of the
number of ED visits for chest pain, by month, ranged
from 7.6% in June to 9.2% in March. Percentage of total
visits by days of the week changed from 13.8% on Fridays
to 15.6% on Mondays. Weakness: Of the 66 092 total visits
in the study, 48.7% (n = 32 180) occurred among males.
Between 1992 and 2002, the percentage of the number
of ED visits for weakness, by month, ranged from 7.7%
these mostly assess mortality, morbidity, hospital admis-
sions and emergency department admissions for respira-
tory and cardiovascular health outcomes. In this work we
fit in a trend to assess other endpoints than the traditional
cardio-pulmonary health outcomes. As we have already
mentione there is no such study: ambient air pollution
and ED visits for chest pain or weakness.
MATERIALS AND METHODS
Data on ED visits were supplied by Capital Health for
all five Edmonton area hospitals and covered the period
between April 1, 1992 and March 31, 2002. ED visits for
chest pain were identified based on a discharge diagnosis
of chest pain using the International Classification for Dis-
eases 9th revision (ICD-9), rubric 786 [1]. The visits were
date-tagged at the day of arrival to the ED. In total, the
analysis is based on 68 714 ED visits for chest pains over
a span of 3652 days. This represented approximately 2.3%
of all (n = 2 946 714) recorded and diagnosed ED visits to
these hospitals over the study period.
From the same data base we extracted all cases related
to ED visits for weakness based on a discharge diagnosis
of weakness using the ICD-9 codes, rubric 780 [1]. In this
case, we had 66 092 ED visits for weakness, which is ap-
proximately 2.2% of all recorded and diagnosed ED visits.
Environment Canada supplied hourly data for selected
weather variables; this included: relative humidity, tem-
perature (dry bulb) and atmospheric pressure (sea level).
For our study we created a daily average of 24 measure-
ments. After preliminary analysis, in our models we
used only temperature and relative humidity as weather
factors.
The hourly data on ambient air pollutants concentrations
were obtained from a few monitoring stations in Edmon-
ton. These data were also supplied by Environment Cana-
da as part of the National Air Pollution Surveillance Net-
work [2]. We defined the daily concentration level for any
specific monitor station as an average of 24 measurements
taken at hourly intervals. The daily shared exposures of
the population were expressed as mean values among
stations.

AIR POLLUTION AND CHEST PAIN IN EDMONTON, CANADA O R I G I N A L P A P E R S
IJOMEH 2010;23(1) 17
obtained for temperature and relative humidity. The both
parameters are often used in other studies.
Table 3 and 4 represent the estimated excess relative risks
(%RR) and their 95% confidence intervals (95% CI). The
percentage changes in daily ED admissions are in relation
to an increase in the interquartile range (IQR) of each
pollutant. The calculated risks were adjusted for the ef-
fects of temperature and relative humidity. In the tables
only positive statistically significant results are shown. The
tables also present detailed specifications for gender and
period. Chest pain: (Table 3) elevated risks observed be-
tween levels of air pollution and chest pain for CO, NO2,
and O3 (1-day lag) were 2.4% (95% CI: 1.0–3.9), 3.8%
in April (7.9% in September) to 8.9% in July. The percent-
age of total visits by days of the week changed from 13.6%
on Tuesdays to 15.2% on Sundays.
Table 2 contains a summary on ambient air pollutants and
weather components. Ambient air pollutant concentra-
tions and meteorological parameters were used to show
environmental characteristics in Edmonton in the time
period of the study. In the table the number of days for
which the values were available was shown. The models
constructed by the hierarchical method are based on sin-
gle pollutant and two weather factors: temperature and
relative humidity. The models were evaluated for differ-
ent configurations of weather parameters. The best fit was
Table 1. Frequency of emergency department visits for chest pain (ICD-9 = 786) and weakness (ICD-9 = 780) by age group
and gender. Edmonton (April 01, 1992 – March 31, 2002)
Age of
respondents
(years)
Chest Weakness
n %
Female
n
Male
n
n %
Female
n
Male
n
0–20 7 400 10.8 3 678 3 722 21 685 32.8 10 772 10 913
20–30 8 780 12.8 4 234 4 546 7 040 10.7 3 839 3 201
30–40 11 150 16.2 5 014 6 136 7 746 11.7 3 860 3 886
40–50 12 119 17.6 5 471 6 648 6 669 10.1 3 196 3 473
50–60 10 007 14.6 4 837 5 170 5 144 7.8 2 412 2 732
60–70 8 116 11.8 3 962 4 154 5 203 7.9 2 420 2 783
70–80 7 093 10.3 3 778 3 315 6 460 9.8 3 460 3 000
80–105 4 049 5.9 2 606 1 443 6 145 9.3 3 953 2 192
Total 68 714 100.0 33 580 35 134 66 092 100.0 33 912 32 180
Table 2. Number of days, mean, standard deviation (SD),
median, interquartile range (IQR, the 75th–25th percentile
values) of daily average concentrations of the ambient
air pollutants and meteorological factors, Edmonton
(April 01, 1992 – March 31, 2002)
Variable (unit) Days Mean SD Median IQR
CO (ppm) 3 652 0.7 0.4 0.6 0.4
NO2 (ppb) 3 652 21.9 9.4 19.7 12.8
SO2(ppb) 3 616 2.6 1.8 2.2 2.3
O3 (ppb) 3 652 18.6 9.3 17.8 14.0
PM10 (μg/m
3) 2 813 22.6 13.1 19.4 15.0
PM2.5 (μg/m
3) 1 444 8.5 6.2 7.2 6.2
Temperature (°C) 3 652 3.9 11.9 5.4 17.9
Relative humidity (%) 3 652 66.0 13.6 66.1 18.5
Fig. 1. Chest pain: The percentage changes in the relative risk
(%RR) by the pollutants (lagged by none, 1 and 2 days), sex
and season.

O R I G I N A L P A P E R S M. SZYSZKOWICZ ET AL.
IJOMEH 2010;23(1)18
(95% CI: 0.0–7.8) and 4.5% (95% CI: 0.9–8.3), respec-
tively. Particulate matter exposures were associated
with an increased risk of chest pain with %RR equal
to 2.8% (95% CI: 0.5–5.2), and 2.1% (95% CI: 0.2–4.0)
for PM10 and PM2.5, respectively. Weakness: (Table 4)
elevated risks were estimated between levels of ambi-
ent air pollution and ED visits for weakness as 3.4%
(95% CI: 1.0–5.9), 2.4% (95% CI: 0.9–3.9), and 4.6%
(95% CI: 1.0–8.2), for NO2 (2-day lag), SO2, and O3
(1-day lag), respectively. Figure 1 and 2 shows the esti-
mated %RRs for the considered combination of the air
pollutants (lagged), sex and season.
DISCUSSION AND CONCLUSIONS
In this study, the short-term effects of air pollutants, after
adjusting for temperature and relative humidity, on daily
ED visits for chest pain and weakness in Edmonton were
found to be positive and statistically significant. Chest pain
or weakness is related to exposure to air pollution. Clus-
tering and overlapping symbols on the figures support the
Fig. 2. Weakness: The percentage changes in the relative risk
(%RR) by the pollutants (lagged by none, 1 and 2 days), sex
and season.
Table 3. The excess risks (%RR) and 95% confidence intervals
(95% CI) for ED visits for chest pain by pollutants, period
and gender
Pollutant Period, patients % RR 95% CI
CO I–XII, all 1.9 0.9–3.0
CO X–III, all 1.3 0.2–2.4
CO I–XII, female 2.4 1.0–3.9
CO X–III, female 2.0 0.4–3.6
CO I–XII, male 1.8 0.4–3.1
CO_1 I–XII, female 1.7 0.3–3.1
CO_1 X–III, female 1.8 0.2–3.4
CO_2 I–XII, female 1.5 0.1–2.9
NO2 I–XII, all 2.6 1.3–4.0
NO2 X–III, all 1.7 0.1–3.4
NO2 IV–IX, all 2.9 0.0–5.8
NO2 I–XII, female 3.3 1.4–5.2
NO2 X–III, female 2.8 0.5–5.1
NO2 I–XII, male 2.4 0.6–4.3
NO2 IV–IX, male 3.8 0.0–7.8
NO2_1 X–III, female 2.4 0.1–4.7
NO2_1 I–XII, female 1.9 0.1–3.8
SO2 I–XII, all 1.1 0.0–2.2
SO2 X–III, female 2.0 0.0–4.0
O3_1 IV–IX, all 3.9 1.2–6.7
O3_1 IV–IX, male 4.5 0.9–8.3
O3_2 IV–IX, all 2.7 0.0–5.5
PM10 I–XII, all 1.7 0.5–2.8
PM10 X–III, all 2.1 0.5–3.8
PM10 I–XII, male 2.4 0.9–4.0
PM10 X–III, male 2.8 0.5–5.2
PM2.5 X–III, all 2.1 0.2–4.0
Table 4. The excess risks (%RR) and 95% confidence intervals
(95% CI) for ED visits for weakness by pollutants, period
and gender
Pollutant Period, patients %RR 95% CI
CO_2 X–III, all 1.9 0.7–3.1
CO_2 I–XII, all 1.5 0.4–2.5
CO_2 X–III, male 2.1 0.4–3.7
CO_2 I–XII, male 1.6 0.2–3.0
CO_2 X–III, female 1.7 0.1–3.2
NO2_2 X–III, male 3.4 1.0–5.9
NO2_2 X–III, all 2.7 1.0–4.5
NO2_2 I–XII, all 2.1 0.7–3.6
NO2_2 I–XII, male 2.4 0.5–4.4
NO2_2 I–XII, female 2.1 0.2–4.0
NO2_2 X–III, female 2.3 0.1–4.6
SO2 I–XII, female 2.4 0.9–3.9
SO2 X–III, female 2.4 0.5–4.5
SO2 I–XII, all 1.6 0.4–2.7
SO2 X–III, all 1.6 0.1–3.1
SO2_2 X–III, male 2.2 0.1–4.4
SO2_2 X–III, all 1.5 0.0–3.1
O3_1 IV–IX, female 4.6 1.0–8.2

AIR POLLUTION AND CHEST PAIN IN EDMONTON, CANADA O R I G I N A L P A P E R S
IJOMEH 2010;23(1) 19
ACKNOWLEDGEMENT
The authors acknowledge Environment Canada for providing
the air pollution data from the National Air Pollution Surveil-
lance (NAPS) network that was analyzed in this study. The au-
thors appreciate the efforts of Ms. Chris Houston from Infor-
mation Services, Capital Health and Virginia Willis, from the
Emergency Medicine Research Group (EMeRG), for securing
these data. The first author appreciates the efforts of Health
Canada for securing these data and for funding data acqui-
sition.
REFERENCES
1. World Health Organization. The International Classification
of Diseases, 9th Revision. Geneva: WHO; 1997.
2. Environment Canada. National Air Pollution Surveillance
Network (NAPS) website. Available from: http://www.etc-cte.
ec.gc.
3. Molenberghs G, Verbeke G. Models for discrete longitudinal
data. New York: Springer; 1997.
4. Szyszkowicz M. Use of generalized linear mixed models to ex-
amine the association between air pollution and health out-
comes. Int J Occup Med Environ Health 2006;19:224–7.
DOI 10.2478/v10001-006-0032-7.
5. R. 2.6.1. The R Foundation for Statistical Computing. Available
from: http://www.r-project.org/.
6. Uva AS. Aircraft cabin air quality: exposure to ozone. Act Med
Port 2002;15:143–51.
7. Ekosse G, de Jager L, van den Heever DJ. The occurrence of
chest pains and frequent coughing among residents living within
the Selebi Phikwe Ni-Cu mine area, Botswana. Afr J Health
Sci 2005;12:37–48.
8. Jaakkola MS, Yang L, Ieromnimon A, Jaakola JJ. Office work,
SBS and respiratory and sick building syndrome symptoms. Oc-
cup Environ Med 2007;64:178–84.
findings. For example the association between exposure
to nitrogen dioxide and ED visits for chest pain is empha-
sised by the cluster of symbols (%RR) on a positive side
of the coordinate system. Our results suggest it is possible
that vehicular traffic, which is one of the main producer
of NO2, contributes to an increased incidence of ED visits
for chest pain.
It was shown that exposure to ozone present in an aircraft
cabin can be an irritant of the respiratory system result-
ing in chest pain [6]. The residents living in sites closest
to a mine and smelter/concentrator plant in Botswana
reported a higher incidence of chest pain and frequent
coughing, compared to those living in another part of the
study area [7]. In-office exposure to carbonless copy pa-
per, paper dust, and fumes from photocopiers and printers
can be associated with chest pain and fatigue [8].
The limitations of this study are typical of this type of
research. They include the adequacy of the used model
and impact of measurements error in the exposure to pol-
lutants and weather factors and outcome variables. The
outcome variables are identified on the basis of the ICD9
codes. Each ED chart was coded by an experienced medi-
cal record nosologist using triage information, nursing
notes, ED records and consultation notes. We may assume
that chest pain or weakness was rather properly diagnosed.
Fixed-site monitors provide daily pollution exposures of
ambient air pollution and are applied to represent shared
population exposure. Edmonton is a large city geographi-
cally and people have different exposures. Like most other
time series and case-crossover studies, our risk estimates
may be biased by assigning regional measures of air pollu-
tion from fixed site-monitoring station to patients. On the
other hand the results are based on well accepted study
methodology among researches and with a long positive
history. The proposition to use the GLMM technique in
such a type of study is a relatively new concept.
Overall, this study provides support for the hypothesis that
ED visits for chest pain and weakness are associated with
exposure to ambient air pollution.