Content uploaded by Jaymin Kwon
Author content
All content in this area was uploaded by Jaymin Kwon on Mar 30, 2016
Content may be subject to copyright.
Effects of air pollution on asthma hospitalization rates
in different age groups in metropolitan cities of Korea
Minjeong Park &Sheng Luo &Jaymin Kwon &
Thomas H. Stock &George Delclos &Ho Kim &
Hong Yun-Chul
Received: 31 May 2012 /Accepted: 25 February 2013 / Published online: 9 March 2013
#Springer Science+Business Media Dordrecht 2013
Abstract Many studies have shown associations between air
pollution and asthma admissions in Korea, but have not reported
whether these effects differ by age classification. The purpose of
this study was to determine whether air pollution effects on
asthmatic hospital admissions are different by three age groups
(years): children (less than 15), adults (15–64; reference group),
and the elderly (over 65). Daily time-series data from seven
metropolitan cities in South Korea were analyzed in two stages.
In the first stage, relative asthma morbidity rates associated with
air pollution were estimated for each city and age group, using
semiparametric log-linear regression. In the second stage, esti-
mates from all seven cities were combined by age group using
Bayesian hierarchical modeling. The effects of exposure to
particulate matter <10 μm in aerodynamic diameter (PM
10
),
carbon monoxide (CO), and nitrogen dioxide (NO
2
) varied
significantly by age groups. Using adults as the referent, the
relative rate (RR) of asthma admissions with 10 μg/m
3
increase
of PM10 is 1.5 % (95 % CI 0.1–2.8 %) lower for children and
1.3 % (95 % CI 0.7–1.9 %) higher for the elderly; RR with
1 ppm increase of CO is 1.9 % (95 % CI 0.3–3.8 %) lower for
children; RR with 1 ppb increase of NO
2
(1 ppb) is 0.5 % (95 %
CI 0.3–0.7 %) higher for the elderly. No significant age group
difference in relative rate was found for ozone or sulfur dioxide.
Keywords Asthma .Air pollution .Age group .Log-linear
regression .Hierarchical model
Introduction
Numerous studies reported relationships between air
pollution and mortality or morbidity over many decades.
Especially, air pollution has been considered as a possible
contributor to the prevalence of asthma (Baibergenova
et al. 2005; Berktas and Bircan 2003; Dab et al. 1996;
Galan et al. 2003; Halonen et al. 2008;Hoetal.2007).
Air Pollution and Health European Approach (APHEA)
conducted pivotal studies about effects of air pollution
on health in ten European countries, finding evidence of
adverse air pollution effects on respiratory diseases in-
cluding asthma in urban areas using an epidemiologic
time-series approach (Katsouyanni et al. 1995,1997;
Vigotti 1999). One of the largest studies in the USA,
The National Morbidity, Mortality, and Air Pollution
Study (NMMAPS) reported the morbidity of asthma
associated with air pollution (Samet et al. 2000). Also,
in Korea, several studies on acute asthma caused by air
pollution have been conducted in order to address the
high mortality rate (Ho et al. 2007;Kimetal.2006,
2007a,b), since mortality from respiratory diseases,
including asthma, increased 100 % in the decade from
1992 to 2002 (Cho et al. 2006).
Asthma is separately diagnosed by age, usually two
groups, age <15 year as child-onset and age >15 as adult-
onset asthma. Hence, asthma researches associated with air
pollution have been investigated separately by age group,
mostly focused on childhood asthma (Dockery et al. 1989;
Gergen et al. 1988; Kim et al. 2005; Lee et al. 2002,2006;
Lin et al. 2002; Mortimer et al. 2002; Norris et al. 1999;
Schildcrout et al. 2006; Son et al. 2006).
However, no study has been conducted in Korea to assess
differences of the air pollutant effect on asthma prevalence
by age group and whether these are significant differences.
M. Park (*):S. Luo :T. H. Stock :G. Delclos
Houston, USA
e-mail: min.park@uth.tmc.edu
J. Kwon
Fresno, USA
H. Kim :H. Yun-Chul
Seoul, Korea
Air Qual Atmos Health (2013) 6:543–551
DOI 10.1007/s11869-013-0195-x
Therefore, it is necessary to evaluate whether air pollution
effects associated with acute asthma prevalence differ
according to age classification.
We tested the null hypothesis that air pollution effects are
statistically different on three age groups: (1) ages less than
15 years (childhood asthma), (2) ages 15–64 years, and (3)
ages over 65 (elderly) with the two-stage model. In the first
stage, we used time-series analysis based on Poisson regres-
sion to estimate age-specific effects for each city (Dominici
et al. 2002; Peng et al. 2006; Schwartz et al. 1996). And,
Bayesian hierarchical modeling was used in the second state
to combine the estimates across seven cities by age group in
Korea (Dominici et al. 2000).
Materials and methods
Hospital administration data
The Korean National Health Insurance Corporation registered
asthma-related hospital admissions data between June 1,
1999, and December 31, 2003. The data have emergency
room visits including night time or holiday visits prompted
by asthma (classified as “J45”and “J46,”by ICD-10). Two or
more visits by the same person in the same day were counted
as one episode. Daily counts for asthma were aggregated in
each of the seven metropolitan cities (Fig. 1) for the three age
groups between June 1999 and December 2003.
Meteorological data
Hourly data on temperature (in degree Celsius), humidity (in
percent), sea level atmospheric pressure (in hectopascal),
and wind speed (in meter per second), and 3-h dew point
temperature (in Celsius) data were obtained from the Korea
Meteorological Offices in the seven cities. Three-hour at-
mospheric Pasquill stability classifications were obtained
from the US National Oceanic and Atmospheric
Administration. Atmospheric stability was designated on a
scale of A (extremely unstable) to G (extremely stable), and
data on this variable were also obtained from each city's
observatory. This variable was dichotomized by substituting
a value of zero for A–C status levels, and 1 for D–G status
levels (Kwon et al. 2006).
Environmental monitoring data
Hourly air pollutant monitoring data were collected at
each of the following monitoring stations (the following
numbers in parentheses represent the number of monitor-
ing stations in each city): Seoul (27), Busan (9), Incheon
(10), Daegu (6), Gwangju (4), Daejeon (3), and Ulsan
(11). Reported air quality concentration data were collect-
ed for PM
10
(in microgram per cubic meter), CO (in part
per million), O
3
(in part per billion), NO
2
(in part per
billion), and SO
2
(in part per billion). Five percent
trimmed daily averages were taken from hourly measured
air pollutant concentrations for each monitoring station
(i.e., an average after trimming 5 % of the data points
from the top and bottom of the distributions). The aver-
age concentration was then calculated for each city using
the daily averaged concentrations from each monitoring
station within the city.
First stage analysis
Data were analyzed in two stages. The first stage was involved
in estimating air pollution effect by age group for each city by
using a log-linear generalized additive model based on
Poisson regression with smoothing splines. This method is
commonly used in time-series studies related to mortality or
hospital counts (Schwartz 1994) such as NMMAPS data
(Dominici et al. 2002). The “mgcv”package was used for
estimating parameters in the first stage by generalized cross
validation or unbiased risk estimation (Wood 2009), which is
different from others in model fitting and variance estimation.
In this paper, long-term seasonality, day of week effect, and
meteorological data were considered as potential confounders.
The day of week effect was included due to the properties of
hospital admission data. In the previous studies, temperature,
dew point temperature, average temperature, and average dew
point temperature for the three previous days had been con-
sidered as major confounders of evaluating air pollution effect
on asthma admissions (Dominici et al. 2000; Samet et al.
2000). Hence, these six variables were always included as
confounders in models, setting as a baseline model. In
Fig. 1 Map of seven metropolitan cities in South Korea
544 Air Qual Atmos Health (2013) 6:543–551
addition to the baseline model, other meteorological variables
which are humidity, sea level atmospheric pressure, wind
speed, and atmospheric Pasquill stability were also added only
if each factor showed significant contribution to the model by
backwards variable selection with 5 % significance level
(Wood 2009).
Yc
atμc
at
jPoisson μcat
ðÞ;t¼1;...;1675;c¼1;...;7;a¼1;2;3
Log μcat
ðÞ¼bcair pollutantðÞ
t
cþbage<15
cair pollutantðÞ
t
cIage<15 þbage65
cair pollutantðÞ
t
cIage65
þgcDOWtday of weekðÞþS1ctime;8=yearðÞ
þS2
ctemp0;3ðÞþS3
ctemp13;3ðÞ
þS4
cdewpoint0;4
þS5
cdewpoint13;4
þSi
cothers;λi
ðÞ
þintercept for age group a b0c;b0;age<15cIage<15;b0 age65 cIage65
þseparate smoothing functions of time for age group
where separate smoothing functions of time for age group
is the age-specific seasonality smoothing function with 4
degrees of freedom per year. The S
c
(x,λ) is smoothing
function with variable xand smoothing parameter λ. The
natural splines model was applied for smoothing functions
in this paper. The smoothing parameter for long-term
seasonality was 8/year in terms of best fit in the data after
checking through 16 parameters/year for seasonality. A
smoothing parameter for other meteorological confounders
except the baselines was selected by minimizing GVC. In
addition to daily mean on the same day as asthma admis-
sions, the average mean from the previous 3 days for each
meteorological variable was also considered. For air pol-
lution, the largest log-relative rates (RR)s were selected
from the last 6 days after rates were estimated for each
day separately. To specify this approach, let Y
c
at
and μ
c
at
be daily sum and expected sum per 10,000 asthma admis-
sions for each age group a (1 =age< 15, 2 =15≤age < 65,
Table 1 Population size, hospi-
tal admissions, and rates (per
10,000) for asthma in each city
by age group from June 1999 to
December 2003
City Age group Population size Admissions Admissions per 10,000
Seoul <15 1,814,097 44,432 245
15–64 7,284,748 22,547 31
≥65 533,053 19,136 359
Busan <15 682,292 25,630 376
15–64 2,655,396 10,133 38
≥65 223,275 10,799 484
Incheon <15 572,823 12,024 210
15–64 1,713,564 5,460 32
≥65 135,455 5,247 387
Daegu <15 512,759 17,092 333
15–64 1,758,418 5,532 31
≥65 146,140 5,509 377
Gwangju <15 308,722 14,296 463
15–64 930,174 4,385 47
≥65 74,714 3,980 533
Daejeon <15 302,798 6,766 223
15–64 945,561 3,593 38
≥65 74,089 3,576 483
Ulsan <15 252,810 5,257 208
15–64 692,876 2,119 31
≥65 40,729 2,530 621
Air Qual Atmos Health (2013) 6:543–551 545
3= age≥65). (Air pollutant)
t
c
is the concentration level of one
of the air pollutants at day tin city c. The indicator function Iis
defined as follows: I
A
(x)=1 if x∈A,I
A
(x)=0 if x∉A.For
example, I
age<15
=1, if the subject is in age< 15 and 0, other-
wise. The Y
c
at
is daily sum per 10,000 asthmatic hospital
admissions adjusted by population size for each city. Hence,
the estimate β
c
is the log-RR per 10,000 asthma admissions
associated with one unit change of air pollutant in the city cfor
adults in age between 15 and 64 (reference group; no subscript
for age group). Additional changes of log-RRs for each of the
children and the elderly from log-RR for adults are β
age< 15
c
and β
age≥65
c
in city c, respectively. If at least one of these
additional changes are not equal to zero, we conclude that air
pollution effect on asthma admission differs by age group.
Each βwas estimated separately by city and age group.
Second stage analysis
The null hypothesis in the second stage is H
0
,β
age< 15
=β
age≥65
=0. The estimates of additional changes of log-
relative rates from two age groups, age< 15 and age ≥65,
c
bacwere assumed to be normally distributed with the true
additional changes for two age groups (age<15, age ≥65)
from adults (ages 15–64) for each city, bacand its
variances σ2
bc
a. These estimated changes for each city
were combined assuming normal for prior distribution
with mean baand variance σ2
ba.Weassumednoprior
information for variance σ2
ba. This was applied to all five
air pollutants in this paper. For this study, WinB UG S,
software for Bayesian analysis using Markov Chain
Monte Carlo, was used to estimate combined additional
changes of log-relative rates for two age groups (age< 15,
age≥65) comparing to the log-relative rate for adults
(ages 15–64), with very large variance implicating no
prior information given for posterior estimation.
c
bacbacN
jbac;σ2
bc
a
bacba~Njba;σ2
ba
;c¼1;...;7;a¼1;3
Fig. 2 Time-series plot for
daily number of hospital
admissions for asthma in each
city by age group (<15, 15–64,
≥65), from June 1999 to
December 2003
546 Air Qual Atmos Health (2013) 6:543–551
Results
Table 1shows population size, hospital admission numbers
for asthma, and asthma admission rates (per 10,000: adjust-
ed by population size) for each city by age group.
Admission rates for adults were much less than for the other
age groups, and the elderly tended to have the highest rates
among the three. Time-series plots for daily asthma admis-
sion rates are presented for each city by age group in Fig. 2.
The dots are actual asthma admission rates, and the curves
in the plot were fitted by generalized linear model using
natural splines function with 8/year as smoothing parameter.
The same format was also applied to Fig. 3. The graphs
showed three peaks of admission rates in 1999, 2000, and
2002. Usually, asthma admission rates tended to increase in
the fall and winter. However, the peak time and pattern are
not identical in age groups. In children, the first peak oc-
curred around November 1999, whereas it occurred between
the end of December 1999 and January 2000 for adults and
elderly people. The adults and elderly groups did not have a
second peak. The third peak occurred at the same time
period in all age groups. Children's admission rates showed
a lot more variability than other age groups.
Statistical summaries of each of the five air pollutant
concentrationsforeachcityareshowninTable2.The
capital of South Korea, Seoul, had the highest average
concentrations of PM
10
and NO
2
, but the lowest levels of
ozone. Figure 3shows the time-series plot for daily averages
of air pollution concentrations for each city. PM
10
tended to
increase in spring for all seven cities. Daejeon showed a
much higher level of CO than any other city between 1999
and 2001. CO showed a huge drop in Gwangju around
February 2003. CO also had a pattern of increasing in winter
and decreasing in summer. O
3
had a reverse pattern which
had two peaks per year; one was around June and another
one was in September from 2001 to 2003. NO
2
and SO
2
did
not show any particular seasonality, but tended to be higher
during the winter and show a minimum around September.
Table 3shows estimated changes of log-relative rates, its
standard errors, relative rates, and its 95 % confidence in-
tervals on asthmatic hospital admission rates (per 10,000)
across seven cities in South Korea associated with one unit
change of air pollution for each age group. RRs of asthma
admission rates (per 10,000) with 10 μg/m
3
increase of
PM
10
significantly increased 0.8 % (95 % CI 0.2–0.16 %)
for adults (15–64 years) and 2.1 % (95 % CI 1.3–2.9 %) for
Fig. 3 Time-series plot for
10 % daily trimmed means of
air pollutant concentrations for
each city, from June 1999 to
December 2003
Air Qual Atmos Health (2013) 6:543–551 547
the elderly (≥65 years). RRs with 1 ppm increase of CO also
significantly increased 1.9 % (95 % CI 0.5–3.3 %) for adults
and 2.2 % (95 % CI 1.2–3.3 %) for the elderly. And RRs
with 1 ppb increase of NO
2
and SO
2
increased 0.6 % (95 %
CI 0.2–0.9 %) and 1.9 % (95 % CI 0.7–3.2 %) for the
elderly. Table 4shows estimated additional changes of log-
relative rates, its standard errors, relative rates and its 95 %
confidence intervals on asthmatic hospital admission rates
(per 10,000) across all seven cities associated with one unit
change of air pollution regarding adults (15–64 years) as a
reference group. The estimates in Table 4are additional
changes of log-RRs of each pollutant for children
(<15 years) and the elderly (≥65 years), referring to log-
RRs for adults (ages 15–64). RRs with 10 μg/m
3
increase of
PM
10
significantly differ by age group; RR for children is
0.15 % lower and RR for the elderly is 0.13 % higher than
Table 2 Distribution of daily
average (for ozone, average for
9–17 h) air pollution concentra-
tions for each city from June
1999 to December 2003
†
For ozone, average for 09–17 h
Mean Standard deviation Median 25 percentile 75 percentile Range [min, max]
PM
10
(μg/m
3
) Korean ambient air quality standard for 24-h averaging time; 100 μg/m
3
Seoul 68.81 38.94 62.88 42.46 84.24 [10.39, 422.72]
Busan 59.85 27.86 52.95 41.31 70.72 [22.15, 321.06]
Incheon 53.45 29.48 47.00 33.04 65.92 [11.51, 270.38]
Daegu 61.66 27.34 56.21 43.02 74.51 [9.63, 287.62]
Gwangju 50.17 26.17 43.81 33.66 59.58 [10.60, 237.18]
Daejeon 48.06 25.12 42.27 30.96 58.40 [10.09, 265.57]
Ulsan 49.09 23.89 43.08 34.02 58.43 [16.26, 286.20]
CO (ppm) Korean ambient air quality standard for 24-h averaging time; 9 ppm
Seoul 0.783 0.343 0.700 0.536 0.934 [0.253, 4.847]
Busan 0.728 0.257 0.679 0.545 0.848 [0.178, 1.823]
Incheon 0.682 0.247 0.625 0.517 0.776 [0.303, 2.309]
Daegu 0.755 0.324 0.685 0.539 0.898 [0.131, 2.554]
Gwangju 0.624 0.302 0.570 0.424 0.760 [0.100, 2.107]
Daejeon 0.969 0.511 0.829 0.617 1.179 [0.113, 4.570]
Ulsan 0.665 0.227 0.630 0.500 0.786 [0.225, 1.583]
O
3
(ppb)
†
Korean ambient air quality standard for 8-h averaging time; 0.06 ppm
Seoul 0.021 0.013 0.013 0.008 0.019 [0.002, 0.048]
Busan 0.029 0.011 0.023 0.017 0.028 [0.004, 0.058]
Incheon 0.024 0.012 0.017 0.012 0.024 [0.002, 0.050]
Daegu 0.026 0.013 0.018 0.012 0.024 [0.003, 0.060]
Gwangju 0.022 0.011 0.016 0.010 0.022 [0.000, 0.066]
Daejeon 0.026 0.013 0.017 0.011 0.025 [0.002, 0.071]
Ulsan 0.028 0.010 0.021 0.016 0.027 [0.004, 0.052]
NO
2
(ppb) Korean ambient air quality standard for 24-h averaging time; 0.06 ppm
Seoul 0.036 0.012 0.035 0.027 0.045 [0.011, 0.075]
Busan 0.026 0.009 0.024 0.019 0.032 [0.006, 0.059]
Incheon 0.027 0.010 0.026 0.019 0.033 [0.007, 0.071]
Daegu 0.027 0.010 0.025 0.020 0.033 [0.007, 0.068]
Gwangju 0.022 0.009 0.021 0.015 0.027 [0.005, 0.059]
Daejeon 0.022 0.011 0.019 0.014 0.027 [0.004, 0.069]
Ulsan 0.020 0.007 0.020 0.015 0.024 [0.005, 0.047]
SO
2
(ppb) Korean ambient air quality standard for 24-h averaging time; 0.05 ppm
Seoul 0.005 0.002 0.005 0.003 0.006 [0.002, 0.016]
Busan 0.008 0.004 0.008 0.005 0.010 [0.001, 0.030]
Incheon 0.007 0.002 0.006 0.005 0.008 [0.002, 0.017]
Daegu 0.007 0.004 0.006 0.004 0.009 [0.002, 0.024]
Gwangju 0.005 0.002 0.004 0.003 0.006 [0.001, 0.016]
Daejeon 0.006 0.003 0.005 0.003 0.007 [0.002, 0.020]
Ulsan 0.008 0.003 0.007 0.006 0.009 [0.003, 0.021]
548 Air Qual Atmos Health (2013) 6:543–551
RR for adults. It shows significantly lower RR (1.9 %) with
CO for children and higher RR (0.5 %) with NO
2
for the
elderly comparing to RR for adults. Based on the hypothesis
test, we concluded that air pollution effects for three air
pollutants (PM
10
, CO, NO
2
) on asthma hospitalization rates
are significantly different in at least two of three groups.
There appeared to be no significant difference among age
groups for O
3
and SO
2
.
Discussion
The study was to investigate whether air pollution effects on
acute asthma differ by age classification. In Table 3, esti-
mated changes of log-relative rates and confidence intervals
across seven cities have been shown for three age groups,
and none of the pollutants have shown a protective effect.
Nonetheless, Table 4shows significantly negative effects in
PM
10
and CO for children since these are additional changes
compared to the effects for adults which mean that child-
hood asthma was less sensitively affected by these two
pollutants than adult-onset asthma. In the same manner,
positive additional changes of log-relative rates in PM
10
and NO
2
for elderly persons mean that the elderly was at
significantly higher risk than adults on these two pollutants.
Based on the results, we concluded that the effects of PM
10
,
CO, and NO
2
significantly differ by age group, and the
elderly was the most vulnerable group to the effect of
PM
10
and NO
2
. In light of previous research which reported
a significant association between air pollutants and child-
hood asthma hospitalizations in Seoul, Korea (Lee et al.
2002), the elderly might be at even higher risk, although it
is not comparable in direct since it had different study
periods. Based on the results of this study, elderly persons
were affected by PM
10
and NO
2
significantly more than
children or adults. On the other hand, other studies found
Table 3 Estimated changes of
log-relative rates, its standard
errors, relative rates and its 95 %
confidence intervals on asthmat-
ic hospital admission rates (per
10,000) across seven cities in
South Korea associated with one
unit (PM
10
=10 μg/m
3
,CO=
1 ppm, O
3
=1 ppb, NO
2
=1 ppb,
SO
2
=1 ppb) change of air pol-
lution for each age group, from
June 1999 to December 2003
a
Statistically significant
b
For ozone, average for 9–17 h
Air pollutant Age group (years) Estimate Standard error RR (95 % CI)
PM
10
(μg/m
3
) <15 −0.007 0.006 0.9930 (0.9814, 1.0048)
15–64 0.008 0.004 1.0080 (1.0002, 1.0160)
a
≥65 0.021 0.004 1.0212 (1.0132, 1.0293)
a
CO (ppm) <15 −0.0007 0.0048 0.9993 (0.99, 1.0087)
15–64 0.0187 0.007 1.0189 (1.0049, 1.033)
a
≥65 0.0219 0.0053 1.0221 (1.0115, 1.0329)
a
O
3
(ppb)
b
<15 −0.0011 0.0016 0.9989 (0.9957, 1.0021)
15–64 −0.0004 0.0007 0.9996 (0.9981, 1.001)
≥65 −0.0001 0.0013 0.9999 (0.9974, 1.0023)
NO
2
(ppb) <15 0.0015 0.0016 1.0015 (0.9984, 1.0046)
15–64 0.0006 0.001 1.0006 (0.9986, 1.0026)
≥65 0.0056 0.0019 1.0056 (1.0019, 1.0093)
a
SO
2
(ppb) <15 −0.0125 0.0095 0.9876 (0.9694, 1.0061)
15–64 0.0083 0.0052 1.0084 (0.9981, 1.0188)
≥65 0.019 0.0062 1.0192 (1.007, 1.0316)
a
Table 4 Estimated additional
changes of log-relative rates, its
standard errors, relative rates and
its 95 % confidence intervals on
asthmatic hospital admissions (per
10,000) across seven cities in
South Korea associated with one
unit (PM
10
=10 μg/m
3
,CO=
1 ppm, O
3
=1 ppb, NO
2
=1 ppb,
SO
2
=1 ppb) change of air pollu-
tion compared to adults (15–
64 years) as a reference group,
from June 1999 to December 2003
a
Statistically significant
b
For ozone, average for 9–17 h
Air pollutant Age group (years) Estimate Standard Error RR (95 % CI)
PM
10
(μg/m
3
) <15 −0.015 0.007 0.9851 (0.9717, 0.9987)
a
≥65 0.013 0.003 1.0131 (1.0071, 1.0191)
a
CO (ppm) <15 −0.0195 0.0098 0.9807 (0.9621, 0.9997)
a
≥65 0.0031 0.0032 1.0031 (0.9969, 1.0094)
O
3
(ppb)
b
<15 −0.0007 0.0016 0.9993 (0.9962, 1.0024)
≥65 0.0003 0.0009 1.0003 (0.9985, 1.0021)
NO
2
(ppb) <15 0.0009 0.0017 1.0009 (0.9976, 1.0042)
≥65 0.0049 0.0011 1.0049 (1.0028, 1.0071)
a
SO
2
(ppb) <15 −0.0208 0.0113 0.9794 (0.9581, 1.0013)
≥65 0.0106 0.0056 1.0107 (0.9997, 1.0218)
Air Qual Atmos Health (2013) 6:543–551 549
no age difference of associations between air pollution and
asthma or pulmonary diseases as follows. In the APHEA 2
report in 2001 (Atkinson et al. 2001), investigators estimat-
ed PM
10
/total suspended particulate (TSP) effects on asth-
matic hospital admission for eight cities separately for each
age group (0–14 years, 15–64 years, 65+ years). They com-
bined city-specific effects by age group counting on hetero-
geneity across the cities. And, these combined estimates for
each age group were close to each other and did not show
the difference among age groups in PM
10
/TSP. This is a
different conclusion from our finding in this paper. There
can be many reasons for this discrepancy; two possibilities
are specified as follows. One is the different measurement
on particles. Of eight cities, two cities (Milan, Rome)
recorded median daily TSP levels and one city (Paris) mea-
sured PM
13
. In contrast, we measured PM
10
for all seven
cities. Even though the correlations among PM
10
,PM
13
, and
TSP are relatively high, this may contribute some of the
discrepancy since the results were combined across cities.
Another possibility is different population for the elderly.
The APHEA 2 report in 2001 included population in hos-
pital admissions for COPD in addition to asthma, yet we
included only asthma admissions due to limited access to
information. Since it is difficult to distinguish between
COPD and asthma in the elderly population due to other
factors such as smoking (Bellia et al. 2003; Tinkelman et al.
2006), a different population setting may provoke substan-
tial differences in conclusion. This may be an important
limitation of this study. Nonetheless, the use of hospital
admission and discharge data has been found to be more
reliable and lessprone to misclassification when differentiating
COPD and asthma in elderly persons as compared to self-
reported data since the asthma data used are hospital admission
discharge data (Radeos et al. 2009). We also recognize that
exploring better ways to avoid this misclassification is needed.
However, for children (<15 years), the results in Table 3
showed consistency with previous researches that there is no
significant increase of relative rates on childhood asthma ad-
missions associated with air pollutant levels (Anderson et al.
2010,2012).
For hypothesis testing, we combined seven estimates
from each city by using a Bayesian approach assuming
homogeneity across seven cities since these seven cities
are designated as metropolitan areas in Korea based on
criteria such as population size. However, possible hetero-
geneity, caused by different locations or contributors of the
air pollution effects on asthma hospital admissions, may be
an important limitation of the study. For example, socioeco-
nomic status (Lee et al. 2006), one of the important factors
in childhood asthma, could be different in the populations of
these cities. Polluting sources could be also different among
seven cities, for example, the largest auto-making industry
in Ulsan.
We considered only meteorological factors, which have
been shown to be associated with respiratory diseases (Abe
et al. 2009; Dominici et al. 2006; Gosai et al. 2009; Wong et
al. 1987), as confounders in this paper. However, other than
meteorological factors, childhood asthma may be more
influenced by some other factors such as genetic predispo-
sition which needs to be considered in further research.
Multiple pollutant models also can be considered in further
studies since some air pollutants are related to each other.
Bias may be addressed in this study due to the characteris-
tics of hospital admission data which counted only emer-
gency room visit patients who may not represent all
asthmatic patients. We suggest further study considering
possible limitations we address here in addition to our major
suggestion based on the study results which is to be aware of
different air pollution effects by age classification and to pay
more attention to asthma in the elderly differing from child-
hood asthma.
References
Abe T, Tokuda Y, Ohde S, Ishimatsu S, Nakamura T, Birrer RB
(2009) The relationship of short-term air pollution and weather
to ED visits for asthma in Japan. Am J Emerg Med 27(2):153–
159
Anderson HR, Butland BK, van Donkelaar A, Brauer M, Strachan DP,
Clayton T et al (2012) Satellite-based estimates of ambient air
pollution and global variations in childhood asthma prevalence.
Environ Heal Perspect 120(9):1333–1339
Anderson HR, Ruggles R, Pandey KD, Kapetanakis V, Brunekreef B,
Lai CK et al (2010) Ambient particulate pollution and the world-
wide prevalence of asthma, rhinoconjunctivitis and eczema in
children: phase one of the international study of asthma and
allergies in childhood (ISAAC). Occup Environ Med 67(5):293–
300
Atkinson RW, Anderson HR, Sunyer J, Ayres J, Baccini M, Vonk JM et
al (2001) Acute effects of particulate air pollution on respiratory
admissions: results from APHEA 2 project. Air pollution and
health: a European approach. Am J Respir Crit Care Med
164(10 Pt 1):1860–1866
Baibergenova A, Thabane L, Akhtar-Danesh N, Levine M, Gafni A,
Moineddin R et al (2005) Effect of gender, age, and severity of
asthma attack on patterns of emergency department visits due to
asthma by month and day of the week. Eur J Epidemiol 20(11):947–
956
Bellia V, Battaglia S, Catalano F, Scichilone N, Incalzi RA, Imperiale
C et al (2003) Aging and disability affect misdiagnosis of COPD
in elderly asthmatics: the SARA study. Chest 123(4):1066–1072
Berktas BM, Bircan A (2003) Effects of atmospheric sulphur dioxide
and particulate matter concentrations on emergency room admis-
sions due to asthma in Ankara. Tuberkuloz Ve Toraks 51(3):231–
238
Cho SH, Park HW, Rosenberg DM (2006) The current status of asthma
in Korea. J Korean Med Sci 21(2):181–187
Dab W, Medina S, Quenel P, Le Moullec Y, Le Tertre A, Thelot B et al
(1996) Short term respiratory health effects of ambient air pollution:
results of the APHEA project in Paris. J Epidemiol Community
Health 50(Suppl 1):s42–6
550 Air Qual Atmos Health (2013) 6:543–551
Dockery DW, Speizer FE, Stram DO, Ware JH, Spengler JD, Ferris BG
Jr (1989) Effects of inhalable particles on respiratory health of
children. Am Rev Respir Dis 139(3):587–594
Dominici F, McDermott A, Zeger SL, Samet JM (2002) On the use of
generalized additive models in time-series studies of air pollution
and health. Am J Epidemiol 156(3):193–203
Dominici F, Peng RD, Bell ML, Pham L, McDermott A, Zeger SL et al
(2006) Fine particulate air pollution and hospital admission for
cardiovascular and respiratory diseases. JAMA: J Am Med Assoc
295(10):1127–1134
Dominici F, Samet JM, Zeger SL (2000) Combining evidence on air
pollution and daily mortality from the 20 largest US cities: a
hierarchical modelling strategy. J R Stat Soc 163:263–302
Galan I, Tobias A, Banegas JR, Aranguez E (2003) Short-term effects
of air pollution on daily asthma emergency room admissions. Eur
Res J: Offic J Eur Soc Clin Respir Physiol 22(5):802–808
Gergen PJ, Mullally DI, Evans R 3rd (1988) National survey of
prevalence of asthma among children in the United States, 1976
to 1980. Pediatrics 81(1):1–7
Gosai A, Salinger J, Dirks K (2009) Climate and respiratory disease in
Auckland, New Zealand. Aust N Z J Public Health 33(6):521–526
Halonen JI, Lanki T, Yli-Tuomi T, Kulmala M, Tiittanen P, Pekkanen J
(2008) Urban air pollution, and asthma and COPD hospital emer-
gency room visits. Thorax 63(7):635–641
Ho WC, Hartley WR, Myers L, Lin MH, Lin YS, Lien CH et al (2007)
Air pollution, weather, and associated risk factors related to asth-
ma prevalence and attack rate. Environ Res 104(3):402–409
Katsouyanni K, Touloumi G, Spix C, Schwartz J, Balducci F, Medina S et
al (1997) Short-term effects of ambient sulphur dioxide and particu-
late matter on mortality in 12 European cities: results fromtime series
data from the APHEA project. Air pollution and health: a European
approach. BMJ (Clinical Research Ed) 314(7095):1658–1663
Katsouyanni K, Zmirou D, Spix C, Sunyer J, Schouten JP, Ponka A et
al (1995) Short-term effects of air pollution on health: a European
approach using epidemiological time-series data. The APHEA
project: background, objectives, design. Eur Respir J: Offic J
Eur Soc Clin Respir Physiol 8(6):1030–1038
Kim DH, Kim YS, Park JS, Kwon HJ, Lee KY, Lee SR et al (2007a)
The effects of on-site measured ozone concentration on pulmo-
nary function and symptoms of asthmatics. J Korean Neurol
Assoc 22(1):30–36
Kim JH, Kim JK, Son BK, Oh JE, Lim DH, Lee KH et al (2005)
Effects of air pollutants on childhood asthma. Yonsei Med J
46(2):239–244
Kim SY, Kim H, Kim J (2006) Effects of air pollution on asthma in
Seoul: comparisons across subject characteristics. J Prev Med Pub
Health= Yebang Uihakhoe Chi 39(4):309–316
Kim SY, O'Neill MS, Lee JT, Cho Y, Kim J, Kim H (2007b) Air
pollution, socioeconomic position, and emergency hospital visits
for asthma in Seoul, Korea. Int Arch Occup Environ Heal
80(8):701–710
Kwon J, Weisel CP, Turpin BJ, Zhang J, Korn LR, Morandi MT et al
(2006) Source proximity and outdoor-residential VOC concentra-
tions: results from the RIOPA study. Environ Sci Technol
40(13):4074–4082
Lee JT, Kim H, Song H, Hong YC, Cho YS, Shin SY et al (2002) Air
pollution and asthma among children in Seoul, Korea. Epidemiology
13(4):481–484
Lee JT, Son JY, Kim H, Kim SY (2006) Effect of air pollution on
asthma-related hospital admissions for children by socioeconomic
status associated with area of residence. Arch Environ Occup
Heal 61(3):123–130
Lin M, Chen Y, Burnett RT, Villeneuve PJ, Krewski D (2002) The
influence of ambient coarse particulate matter on asthma hospi-
talization in children: case-crossover and time-series analyses.
Environ Heal Perspect 110(6):575–581
Mortimer KM, Neas LM, Dockery DW, Redline S, Tager IB (2002)
The effect of air pollution on inner-city children with asthma.
Eur Respir J: Offic J Eur Soc Clin Respir Physiol 19(4):699–
705
Norris G, YoungPong SN, Koenig JQ, Larson TV, Sheppard L, Stout
JW (1999) An association between fine particles and asthma
emergency department visits for children in Seattle. Environ
Heal Perspect 107(6):489–493
Peng RD, Dominici F, Louis TA (2006) Model choice in time series
studies of air pollution and mortality. J R Stat Soc: Series A
169(Part 2):179–203
Radeos MS, Cydulka RK, Rowe BH, Barr RG, Clark S, Camargo CA
Jr (2009) Validation of self-reported chronic obstructive pulmo-
nary disease among patients in the ED. Am J Emerg Med
27(2):191–196
Samet JM, Zeger SL, Dominici F, Curriero F, Coursac I, Dockery DW
et al (2000) The national morbidity, mortality, and air pollution
study. part II: Morbidity and mortality from air pollution in the
United States. Research Report (Health Effects Institute) 94(Pt
2):5–70, Discussion, 71–9
Schildcrout JS, Sheppard L, Lumley T, Slaughter JC, Koenig JQ,
Shapiro GG (2006) Ambient air pollution and asthma exacerba-
tions in children: an eight-city analysis. Am J Epidemiol
164(6):505–517
Schwartz J (1994) Nonparametric smoothing in the analysis of air
pollution and respiratory illness. Can J Stat 22(4):471–487
Schwartz J, Spix C, Touloumi G, Bacharova L, Barumamdzadeh T, le
Tertre A et al (1996) Methodological issues in studies of air
pollution and daily counts of deaths or hospital admissions. J
Epidemiol Community Health 50(Suppl 1):S3–11
Son JY, Kim H, Lee JT, Kim SY (2006) Relationship between the
exposure to ozone in Seoul and the childhood asthma-related
hospital admissions according to the socioeconomic status. J
Prev Med Public Health=Yebang Uihakhoe Chi 39(1):81–86
Tinkelman DG, Price DB, Nordyke RJ, Halbert RJ (2006)
Misdiagnosis of COPD and asthma in primary care patients
40 years of age and over. J Asthma: Official J Assoc Care
Asthma 43(1):75–80
Vigotti MA (1999) Short-term effects of exposure to urban air pollu-
tion on human health in Europe. the APHEA projects (air pollu-
tion and health: a European approach. [Effetti a breve termine
prodotti dalla esposizione ad inquinamento atmosferico urbano
sulla salute umana in Europa. I Progetti APHEA (Air Pollution
and Health: a European Approach)]. Epidemiologia e Prevenzione
23(4):408–415
Wong KW, Davies DP, Lau EM (1987) Asthma and climatic condi-
tions: experiences from Hong Kong. Br Med J (Clin Res Ed)
294(6564):119
Wood S (2009) Mgcv: GAMs with GCV/AIC/REML smoothness
estimation and GAMMs by PQL
Air Qual Atmos Health (2013) 6:543–551 551