\documentclass[10pt]{article}
\usepackage{fullpage}
\usepackage{setspace}
\usepackage{parskip}
\usepackage{titlesec}
\usepackage[section]{placeins}
\usepackage{xcolor}
\usepackage{breakcites}
\usepackage{lineno}
\usepackage{hyphenat}
\PassOptionsToPackage{hyphens}{url}
\usepackage[colorlinks = true,
linkcolor = blue,
urlcolor = blue,
citecolor = blue,
anchorcolor = blue]{hyperref}
\usepackage{etoolbox}
\makeatletter
\patchcmd\@combinedblfloats{\box\@outputbox}{\unvbox\@outputbox}{}{%
\errmessage{\noexpand\@combinedblfloats could not be patched}%
}%
\makeatother
\usepackage{natbib}
\renewenvironment{abstract}
{{\bfseries\noindent{\abstractname}\par\nobreak}\footnotesize}
{\bigskip}
\titlespacing{\section}{0pt}{*3}{*1}
\titlespacing{\subsection}{0pt}{*2}{*0.5}
\titlespacing{\subsubsection}{0pt}{*1.5}{0pt}
\usepackage{authblk}
\usepackage{graphicx}
\usepackage[space]{grffile}
\usepackage{latexsym}
\usepackage{textcomp}
\usepackage{longtable}
\usepackage{tabulary}
\usepackage{booktabs,array,multirow}
\usepackage{amsfonts,amsmath,amssymb}
\providecommand\citet{\cite}
\providecommand\citep{\cite}
\providecommand\citealt{\cite}
% You can conditionalize code for latexml or normal latex using this.
\newif\iflatexml\latexmlfalse
\providecommand{\tightlist}{\setlength{\itemsep}{0pt}\setlength{\parskip}{0pt}}%
\AtBeginDocument{\DeclareGraphicsExtensions{.pdf,.PDF,.eps,.EPS,.png,.PNG,.tif,.TIF,.jpg,.JPG,.jpeg,.JPEG}}
\usepackage[utf8]{inputenc}
\usepackage[greek,ngerman,english]{babel}
\usepackage{float}
\begin{document}
\title{Effect of air-pollutants and environmental factors on COPD
exacerbations: time series analysis stratified by age and concentration
of pollutants}
\author[1]{Jong Seung Kim}%
\author[1]{Hyu Seok Shin}%
\author[1]{Eun Jung Lee}%
\author[1]{Min Gul Kim}%
\author[1]{Sang Woo Yeom}%
\author[1]{Sam Hyun Kwon}%
\author[2]{Min Hee Lee }%
\author[3]{Doo Hwan Kim}%
\author[1]{Sang Jae Noh}%
\affil[1]{Chonbuk National University Hospital}%
\affil[2]{Presbyterian Medical Center}%
\affil[3]{National Health Insurance Service}%
\vspace{-1em}
\date{\today}
\begingroup
\let\center\flushleft
\let\endcenter\endflushleft
\maketitle
\endgroup
\selectlanguage{english}
\begin{abstract}
Background: Chronic obstructive pulmonary disease (COPD) exacerbation
(CE) is characterized by the rapid deterioration of respiratory symptoms
caused by air pollution, but there have been no comprehensive studies
with regard to the age-stratified effect of air pollution. We
investigate the short-term effects of air pollution and environmental
factors on CE. Methods: By merging the individualized environmental data
and Korea's National Health Insurance cohort, we conducted Quasi-Poisson
analysis to evaluate the effects of air pollutants and environmental
factors on CE. Results: From January 2007 to December 2012, a total of
15110 CE events occurred, which showed seasonal dynamics in CE events,
air pollutants (particulate matter less than 10 \selectlanguage{greek}µ\selectlanguage{english}m (PM10), NO2, SO2) and
environmental factors (wind speed (WS), solar sunshine (SS)). The only
factor affecting CE was PM10, and this occurred on lag days 4, 5, and 6
and had a peak hazard ratio (HR) of 1.0404 on lag day 5. PM10 also had a
lag-cumulative effect on CE risk from lag day 6 to lag day 10. In
subgroup analysis on age and level of parameters, each factor had a
different significant effect on CE on different lag days. Conclusions:
PM10 uniquely affects CE at lag-specific day 5 (from lag day 4 to lag
day 6). PM10 also had a lag-cumulative effect on CE risk which showed a
pattern proportional to the concentration of PM10. Considering the
subgroup analysis, we need to devise different strategies for air
pollutants and age for patients with COPD exacerbation.%
\end{abstract}%
\sloppy
\textbf{Abbreviations}
COPD, Chronic Obstructive Pulmonary Disease; CE, COPD exacerbation; HR,
hazard ratio; PM10, Particulate matter less than 10 \selectlanguage{greek}µ\selectlanguage{english}m;
SO\textsubscript{2}, Sulfur dioxide; NO\textsubscript{2}, Nitrogen
dioxide; O\textsubscript{3}, ozone; WS, Wind speed; SS, Solar sunshine;
KNHIS-NSC, Korea National Health Insurance Service-National Sample
Cohort; DLNM, Distributed Lag Linear and Non-Linear Model; QAIC, quasi
Akaike information criterion; PACF, partial autocorrelation function;
DOW, Day of Week; RR, Risk ratio; CO, Carbon monoxide; DTR, diurnal
temperature range; HM, humidity.
\textbf{Introduction}
Chronic obstructive pulmonary disease (COPD) is a disease in which lung
function decreases faster than other organ functions due to abnormal
inflammatory reactions in the lungs. It is a heterogeneous disease,
characterized by airflow limitation caused by a mixed form of
parenchymal destruction (emphysema) and small airway disease
(obstructive bronchiolitis). COPD is currently the fourth leading cause
of death and is expected to be the third leading cause by 2020 (GOLD
CRITERIA).
Symptoms of COPD include cough, sputum, and exertional dyspnea. The
cause may be smoking or air pollution. For example, it has been reported
that air pollution such as fine dust less than 2.5 \selectlanguage{greek}µ\selectlanguage{english}m/10 \selectlanguage{greek}µ\selectlanguage{english}m (PM2.5/10)
and NO\textsubscript{2} affects COPD prevalence.\textsuperscript{1}There
have also been reports of significant adverse effects on lung maturation
and development in children exposed to high levels of
NO\textsubscript{2} and particulate matter less than 2.5 \selectlanguage{greek}µ\selectlanguage{english}m
(PM2.5).\textsuperscript{2} However, there have been no studies on the
delayed effect of short-term and high peak exposure of air pollution on
COPD.
COPD exacerbation (CE) is a disease that requires additional treatment
due to the rapid deterioration of respiratory symptoms caused by COPD,
and is a serious disease which may lead to death.\textsuperscript{3}Some
of the factors that are known to be closely related to CE include a
history of two or more frequent CEs and an increase in blood eosinophil
levels.\textsuperscript{4} However, there have been no studies showing
the relationship between air pollution and CE.
Therefore, in this paper, we investigated the effects of air pollution,
including fine dust, and environmental factors on CE using 6 years of
cohort data.
\textbf{Methods}
\textbf{Population of this study}
This study passed the scrutiny of the Institutional Review Board of
Jeonbuk National University Hospital (IRB number 2020-08-012). This time
series analysis was conducted using the anonymized Korea National Health
Insurance Service-National Sample Cohort (KNHIS-NSC) database, and data
in the period from January 1, 2007 to December 31, 2012 were analyzed.
This database includes day of hospital visit, the patient's age, sex,
residential area, economic level, diagnosis, medical history, and
disease code. The diagnostic codes from KNHIS-NSC are based on the
International Classification of Diseases, Tenth Revision (ICD-10). In
2007, the population in Korea was 49,194,085, of which 1 million cohorts
of randomly stratified samples were used by the KNHIS-NSC databases.
The inclusion criteria of the CE population from this database were: 1)
patients assigned code J441 in the ICD-10 classification at the time of
the hospital visit; 2) patients taking oral steroids; and 3) patients
with chest computed tomography scans.\textsuperscript{5,6} Individuals
were divided into four age groups: preschool age (0 to 4 years old),
school age (5 to 19 years old), adult (20 to 59 years old), and elderly
([?]60 years old).
\textbf{Air pollution and environmental factors}
Particulate matter less than 10 \selectlanguage{greek}µ\selectlanguage{english}m (PM10), sulfur dioxide
(SO\textsubscript{2}), nitrogen dioxide (NO\textsubscript{2}), ozone
(O\textsubscript{3}), and carbon monoxide (CO) are factors affecting
COPD exacerbation.\textsuperscript{7} These air pollutants are known to
have a direct harmful effect or a lag day effect on COPD exacerbation.
Wind speed (WS), solar sunshine (SS), humidity (HM), diurnal temperature
range (DTR) are also important daily ecological factors. In this study,
the lag effect affecting COPD exacerbation was considered using
SO\textsubscript{2}, NO\textsubscript{2}, O\textsubscript{3}, PM10, WS,
SS, HM, CO, and DTR as the main variables.
\textbf{Individualization}
Korea's administrative districts are divided into eight provinces, 77
cities, and 240 local districts. In this study, the place of residence
of patients with CE was individualized using the KNHIS-NSC database, and
the environmental factors in these individualized addresses were used as
input variables. It was thought that these individualized data would
reflect the environmental setting of each region much better than the
national average for each environmental indicator.
\textbf{Data source of Air pollutants}
As of 2010, there were 239 air quality monitoring stations for the
entire Korea according to the city measurement network. Since air
quality measurements from these stations were too coarse to represent
geographic variations of air pollutants across the study area, we
applied the Kriging method, a widely-used statistical approach for
spatial interpolation, in order to obtain refined estimates of spatial
distributions of air quality. The estimated results were then summarized
to the spatial resolution of the local district and were used as an
input for the dlnm modeling of this study.
\textbf{Model construction}
In general, since there is no clear criterion for configuring the
optimal DLNM model\textsuperscript{8}, various approaches can be taken
to construct this. Many studies use Akaike's information criterion to
select the model with the best performance.\textsuperscript{9}Therefore,
we also considered a quasi Akaike information criterion (QAIC) and a
partial autocorrelation function (PACF) to choose the optimal
construction for the DLNM model. The quasi Akaike information criterion
(QAIC)\textsuperscript{10}, the quasi-likelihood adjustments of Akaike's
information criterion (AIC)\textsuperscript{11}, provides important
information on the explanatory power of quasi Poisson models that is
used for overdispersed count data. The partial autocorrelation function
(PACF) evaluates the level of partial autocorrelation in model
residuals. We compared the mean of the absolute values of PACF (mPACF)
for the first 100 days of the models.
Table 1 shows the values of QAIC and mPACF for various models we
compared. Each model was selected by forward selection using a greedy
approach. The M5 model was selected because its QAIC was the lowest
(3289.498) while its PACF value was the second lowest (0.02849), which
was greater than the lowest by only 0.00002.
\textbf{Statistical analysis}
The Distributed Lag Linear and Non-Linear Model (DLNM) was used to view
the direct and delayed effect of air pollution and environmental factors
on CE. Considering the distribution of number of daily CE events, the
Quasi-Poisson model is a more suitable model than the Poisson model.
Quasi-Poisson is usually used when the variance is greater than the
average, and this was the reason why we used this model. It was used in
combination with a DLNM to investigate the lagged and non-linear
effects. For all variables, the maximum lag day \emph{L} was considered
up to 10 days.
The multivariate DLNM model for the air pollutants is:
\begin{equation}
\text{log\ E}\left(Y_{t}\right)=s\left({PM10}_{t};\eta_{\text{PM}10}\right)+s\left({NO2}_{t};\eta_{NO2}\right)+s\left({SO2}_{t};\eta_{SO2}\right)+s\left(\text{WS}_{t};\eta_{\text{WS}}\right)+s\left(\text{SS}_{t};\eta_{\text{SS}}\right)+ns\left(time,\left(4\times 6+4\right)\text{df}\right)+DOW+\sum_{k=1}^{6}{\gamma_{k}\mu_{\text{tk}}}\backslash n\nonumber \\
\end{equation}
\textbf{Results}
\textbf{Baseline data}
From January 2007 to December 2012, a total of 15110 CE events occurred
over the 6 years. CE increased from year to year: 2018 in 2007, 2232 in
2008, 2443 in 2009, 2385 in 2010, 2854 in 2011, and 3178 in 2012. By age
group, there were 97 boys and 41 girls of preschool age, 59 boys and 106
girls of school age, 1643 men and 1076 women aged 20--59, and 7778 men
and 4310 women over 60 years of age(Figure 1A).
\textbf{Seasonal dynamics of COPD exacerbation and outdoor environmental
factors}
We noted seasonal changes in CE that differed by age group (Figure 1B).
COPD exacerbation was predominantly in winter and less frequent in
summer, and this tendency was particularly severe in those aged 60 or
older.
There were definite seasonal variations in NO\textsubscript{2}, PM10,
SO\textsubscript{2}, WS and SS (Figure 1C). Contaminants such as
NO\textsubscript{2}, PM10, and SO\textsubscript{2} mainly showed a
cyclic seasonal pattern with highs in winter and lows in summer.
Environmental factors such as WS and SS also showed cyclic patterns with
highs in late winter and early spring and lows in the autumn. Table 2
shows the median values for the air pollutants and environmental
factors. We used these median values when calculating the risk ratio
(RR).
\textbf{Estimated lag day effect and lag-cumulative effect of outdoor
environmental factors on COPD exacerbation}
Of the three air pollutants and two external environmental factors, the
only factor affecting COPD exacerbation was PM10 (Figure 2A-E). In
particular, PM10 affected COPD exacerbation on lag days 4, 5, and 6 and
had a hazard ratio (HR) of 1.0404 (95\% CI {[}1.0089, 1.0729{]}) at lag
day 5 (Table 3). PM10 also had a cumulative effect from lag day 6 to lag
day 10 (Figure 2A). The remaining factors NO\textsubscript{2},
SO\textsubscript{2}, SS, and WS did not influence COPD exacerbation
(Figure 2B-E).
\textbf{Subgroup analysis -- calculated RR stratified by concentration
of air pollutants and patient age}
As a subgroup analysis, we calculated the RR values according to the
concentration of air pollutants (Table 3). From this, the concentrations
of air pollutants NO\textsubscript{2} and SO\textsubscript{2} were not
correlated with COPD exacerbation risk, nor were the environmental
factors WS and SS. Only PM10 showed an increase in HR with
concentration: 1.0253 at 30 \selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}, 1.0704 at 80
\selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}, and 1.1370 at 150 \selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}(Table
3, Figure 3A, B).
In addition, we examined the RR values for COPD exacerbation according
to patient age by time series (Table 4). Individuals aged under 20 years
(preschool, school age) did not have significant RR values for any of
the air pollutants or environmental factors. In adults (20--59 years),
PM10 at lag day 3 (which was significant from lag day 3 to 4) had an RR
value of 1.0372 and NO\textsubscript{2} at lag day 0 had a significant
RR value of 1.4613. For the environmental factors, WS and SS had
significant RR values at lag day 0 (1.1320, 0.9444, respectively).
This pattern was the same in the elderly over 60 years of age. In the
case of air pollutants, PM10 at lag day 3 (which was significant from
lag day 6) had an RR value of 1.0372, and NO\textsubscript{2} at lag day
6 had a significant RR value of 1.0774. SO\textsubscript{2} did not show
significant RR values in any of the age groups. For the environmental
factors, WS and SS in the elderly group had significant RR values at
lag day 0 (1.1320 and 0.9444, respectively).
\textbf{Discussion}
Various air pollutants such as PM10, NO\textsubscript{2}, and
SO\textsubscript{2} pose a great danger to public health, in particular
to the respiratory system of individuals.\textsuperscript{12} It is
known that, for every increase of PM10 by 10 \selectlanguage{greek}μ\selectlanguage{english}g/m\textsuperscript{3},
mortality from all causes, cardiopulmonary diseases and lung cancer
rises by 4\%, 6\% and 8\%, respectively.\textsuperscript{13} Carbon
monoxide (CO) and nitrogen dioxide (NO\textsubscript{2}) from diesel
engines are known to raise the risk of lung cancer from 15\% to
40\%.\textsuperscript{14} These environmental pollutants are known to
have a significant adverse effect on public health including mortality
rates. As a result of increased residence times in the air and higher
concentrations of pollutants, mortality rates are significantly affected
and public health generally deteriorates.\textsuperscript{15} These air
pollutants are closely affected by environmental factors such as wind
speed and sunlight.
In our study, air pollutants such as PM10, NO\textsubscript{2}, and
SO\textsubscript{2} showed seasonal variations. In Korea, the use of
coal or oil increases during cold weather, mainly in winter, and
atmospheric pollution increases leading to high air pollution indicators
(PM10, SO\textsubscript{2}, NO\textsubscript{2}) in winter (Figure 1C).
In our study, the same sinusoidal pattern appeared in the frequency of
COPD exacerbations (Figure 1B). During the entire 6-year observation
period, 15110 individuals suffered COPD exacerbation, of which 17.9\%
were adults in their 20s, and 80\% were elderly individuals older than
60 years of age (Figure 1A). When the distribution of the entire patient
population was drawn in a time series, the number of individuals
affected tended to be particularly high in winter, which is consistent
with the pattern of major air pollution indicators (Figure 1B).
So which of the three air pollutants, at what concentrations and for how
long, were individuals affected by COPD exacerbation? We utilized ``DLNM
packages'' using R to analyze these factors. In conclusion, of the
various air pollutants, only PM10 was significantly related to COPD
exacerbation (Figure 2A), and it was found to affect lag-specific days 4
to 6. In particular, on lag-specific day 5, the risk of COPD
exacerbation reached peak values (Figure 2A, Table 3). For a PM10
concentration of 30 \selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}, which is the upper limit of
``acceptable'' fine dust levels, the risk of COPD exacerbation increased
by 2.5\%. At the median concentration of normal PM10 (46.98
\selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}), it increased by 4.0\%. When it was 80
\selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}, which is the lower limit of ``unacceptable''
fine dust levels, the risk of COPD exacerbation increased by 7.0\%. In
the case of PM10 over 150 \selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}, which is an extremely
unacceptable level of PM10, CE risk increased by 13.7\% (Table 3, Figure
3A, B). In addition, for the lag-cumulative effect of PM10, we
calculated the risk of COPD exacerbation, and it was found that it
continued to affect CE risk from lag day 6 to lag day 10 after exposure
to PM10 (Figure 2B).
We analyzed the entire patient group in four age subgroups. (Table 4)
For preschool age (less than 5 years old) and school age (less than 20
years old) individuals, the number of COPD exacerbations was small, so
the confidence interval of the RR value was large and so did not produce
meaningful results.
In the group of adults 20 years or older, PM10 level significantly
increased the risk of COPD exacerbation on lag days 3 and 4. At lag day
0, NO\textsubscript{2} and WS levels significantly increased the risk of
COPD, and SS level significantly decreased the risk of COPD
exacerbation.
Similar results were found in older people over the age of 60 with PM10
significantly increasing CE risk at lag days 3, 4, 5, and 6, and
NO\textsubscript{2} significantly increasing CE risk at lag day 6, WS
increasing CE risk at lag day 0, and SS lowering CE risk at lag day 0.
Through this, it can be considered that CE risk was not high in adults
over 20 years of age on a clear day without wind. On the other hand, on
windy days when it was not sunny, CE risk increased in adults over 20
years of age. Unlike air pollutants, all these climate factors were
found to affect CE risk only on the lag-specific day, not the lag days
(after lag day 1).
The limitations of this paper are: 1) The study only covers data since
2007 when fine dust (PM10) measurement started in Korea. 2) Detailed
indicators, such as PM2.5, were not tested at the time and were not
included in the study.
Nevertheless, this paper has the following advantages. 1) This is a
large-scale cohort study of patients in Korea. 2) Modeling considers
each of the air pollutants and environmental factors in combination with
linear, non-linear and lagged effect. 3) The study attempts to analyze
each age group and the concentrations of pollution indicators.
\textbf{Conclusion}
Among air pollutants, PM10 uniquely affects CE at lag-specific day 5
(from lag day 4 to lag day 6). For the lag-cumulative effect of PM10, CE
risk showed a pattern proportional to the concentration of PM10. When we
look at how other air pollutants and environmental factors affect each
age group differently, we need to devise different strategies for air
pollutants and age of COPD patients.
\textbf{References}
1. Liu S, Zhou Y, Liu S, Chen X, Zou W, Zhao D, et al. Association
between exposure to ambient particulate matter and chronic obstructive
pulmonary disease: results from a cross-sectional study in China. Thorax
2017;72(9):788--95.
2. Gauderman WJ, Avol E, Gilliland F, Vora H, Thomas D, Berhane K, et
al. The effect of air pollution on lung development from 10 to 18 years
of age. N Engl J Med 2004;351(11):1057--67.
3. Hurst JR, Wedzicha JA. What is (and what is not) a COPD exacerbation:
thoughts from the new GOLD guidelines. Thorax 2007;62(3):198-9.
4. Hurst JR, Vestbo J, Anzueto A, Locantore N, M\selectlanguage{ngerman}üllerova H, Tal-Singer
R, et al. Susceptibility to exacerbation in chronic obstructive
pulmonary disease. N Engl J Med 2010;363(12):1128--38.
5. Jinjuvadia C, Jinjuvadia R, Mandapakala C, Durairajan N,
Liangpunsakul S, Soubani AO. Trends in outcomes, financial burden, and
mortality for acute exacerbation of chronic obstructive pulmonary
disease (COPD) in the United States from 2002 to 2010. COPD 2017
Feb;14(1):72--9.
6. Gedebjerg A, Szépligeti SK, Wackerhausen LH, Horváth-Puhó E, Dahl R,
Hansen JG, et al. Prediction of mortality in patients with chronic
obstructive pulmonary disease with the new Global Initiative for Chronic
Obstructive Lung Disease 2017 classification: a cohort study. Lancet
Respir Med 2018 Mar;6(3):204--12.
7. Ko FW, Chan KP, Hui DS, Goddard JR, Shaw JG, Reid DW, et al. Acute
exacerbation of COPD. Respirology 2016 Oct;21(7):1152--65.
8. Gasparrini A. Distributed Lag Linear and Non-Linear Models in R: The
Package dlnm. J Stat Softw 2011;43(8):1--20.
9. Talagala T. Distributed lag nonlinear modelling approach to identify
relationship between climatic factors and dengue incidence in Colombo
District, Sri Lanka. Epidemiol Biostat Public Health 2015 Dec 21;12(4).
10. Lebreton JD, Burnham KP, Clobert J, Anderson DR. Modeling survival
and testing biological hypotheses using marked animals: a unified
approach with case studies. Ecol Monogr 1992;62(1):67-118.
11. Anderson D, Burnham K. Model selection and multi-model inference.
2nd ed. NY: Springer-Verlag; 2004; 63(2020):10.
12. The Lancet. Editorial. UK air pollution and public health. Lancet
2017 May 13;389(10082):1860.
13. Pope CA 3rd, Burnett RT, Thun MJ, Calle EE, Krewski D, Ito K, et al.
Lung cancer, cardiopulmonary mortality, and long-term exposure to fine
particulate air pollution. JAMA 2002 Mar 6;287(9):1132--41.
14. Benbrahim-Tallaa L, Baan RA, Grosse Y, Lauby-Secretan B, El
Ghissassi F, Bouvard V, et al., International Agency for Research on
Cancer Monograph Working Group. Carcinogenicity of diesel-engine and
gasoline-engine exhausts and some nitroarenes. Lancet Oncol 2012
Jul;13(7):663--4. doi: 10.1016/s1470-2045(12)70280-2.
15. Huang J, Pan X, Guo X, Li G. Impacts of air pollution wave on years
of life lost: A crucial way to communicate the health risks of air
pollution to the public. Environ Int 2018 Apr;113:42--9.
Hyu Seok Shin, PhD\textsuperscript{1,4*}, Eun Jung Lee, MD,
PhD\textsuperscript{1,2*}, Min Gul Kim MD, PhD\textsuperscript{3*}, Jong
Seung Kim, MD, PhD\textsuperscript{1,2,4}, Sang Woo Yeom,
MS\textsuperscript{4}, Sam Hyun Kwon, MD, PhD\textsuperscript{1,2}, Min
Hee Lee MD\textsuperscript{5}, Doo Hwan Kim, MD\textsuperscript{6}, Sang
Jae Noh, MD, PhD\textsuperscript{7}
\textsuperscript{1}Department of Otolaryngology-Head and Neck Surgery,
College of Medicine, Jeonbuk National University, Jeonju, Republic of
Korea
\textsuperscript{2}Research Institute of Clinical Medicine of Jeonbuk
National University--Biomedical Research Institute of Jeonbuk National
University Hospital, Jeonju, Republic of Korea
\textsuperscript{3}Department of Pharmacology, Jeonbuk National
University, Jeonju, Republic of Korea
\textsuperscript{4}Department of Medical Informatics, Jeonbuk National
University, Jeonju, Republic of Korea
\textsuperscript{5}Department of Internal Medicine, Division of Allergy
and Pulmonology, Presbyterian Medical Center, Jeonju, Republic of Korea
\textsuperscript{6}Director of Big-Data Center, National Health
Insurance Service (NHIS), Wonju, Republic of Korea
\textsuperscript{7}Department of Forensic medicine, College of Medicine,
Jeonbuk National University, Jeonju, Republic of Korea
* These authors contributed equally.
Corresponding author: Jong Seung Kim, MD, PhD, Department of Medical
Informatics and Department of Otorhinolaryngology, College of Medicine,
Jeonbuk National University, 20, Geonji-ro, Deokjin-Gu, Jeonju-si,
Jeonbuk, 561-712, Republic of Korea
Tel: +82 632501980/Fax: +82 632501986
Email address: \emph{kjsjdk@gmail.com}
Conflict of Interest: none
FINANCIAL DISCLOSURE: none
\textbf{Author contributions}
EJL, JSK, DHK, MHL contributed to the study design, protocol and study
materials. HSS, JSK, SWY collected study data. HSS, SWY, SHK, SJN
designed the statistical plan and data analysis. HSS, SWY performed the
statistical analysis. EJL, MGK, JSK, HSS, wrote the first draft of the
manuscript. All authors contributed to interpretation of the data and
revision of the manuscript.
\textbf{Running title} : Effect of outdoor air pollutants on the risk of
COPD exacerbations
\textbf{Keywords} : Chronic Obstructive Pulmonary Disease exacerbation,
air pollutant, environmental factors
\textbf{Sources of funding} : None
\textbf{Text word count:} 2602 words
\textbf{Number of tables and/or figures:} 4/3
\textbf{Figure legends}
Figure 1. A. CE counts by age and sex. B. Seasonal variations according
to age group. C. Seasonal variations of air pollutants and environmental
factors according to time series.
Figure 2. Lag-specific effect and lag-cumulative effect of air
pollutants. A. PM10. B. NO\textsubscript{2}. C. SO\textsubscript{2}. D.
SS. E. WS.
Lag day 0: current day; Lag day 1: 1 day previously, to Lag day 10: 10
days previously.
Figure 3. A. Heat-map of PM10 stratified by concentration and lag day
effect. B. 3D image of PM10 stratified by concentration and lag day
effect.
Lag day 0: current day; Lag day 1: 1 day previously, to Lag day 10: 10
days previously.
Table 1. DLNM model selection process through quasi Akaike information
criterion (QAIC) and a partial autocorrelation function (PACF)
(Abbreviations: M1, model 1; M2, model 2; M3, model 3; M4, model 4; M5,
model 5; M6, model 6; M7, model 7; M8, model 8; M9, model 9; PM10,
Particulate matter less than 10 \selectlanguage{greek}µ\selectlanguage{english}m; NO\textsubscript{2}, Nitrogen
dioxide; SO\textsubscript{2}, Sulfur dioxide; WS, Wind speed; SS, Solar
sunshine; O\textsubscript{3}, ozone; HM, humidity; CO, Carbon monoxide;
DTR; diurnal temperature range; QAIC, quasi Akaike information
criterion; PACF, partial autocorrelation function; DOW, Day of Week; RR,
Risk ratio\selectlanguage{english}
\begin{longtable}[]{@{}llll@{}}
\toprule
& Model & QAIC & mPACF\tabularnewline
\midrule
\endhead
M1 & PM10 & 3305.782 & 0.02962\tabularnewline
M2 & PM10+NO\textsubscript{2} & 3303.790 & 0.02862\tabularnewline
M3 & PM10+NO\textsubscript{2}+SO\textsubscript{2} & 3292.022 &
0.02899\tabularnewline
M4 & PM10+NO\textsubscript{2}+SO\textsubscript{2}+SS & 3294.994 &
0.02847\tabularnewline
\textbf{M5} &
\textbf{PM10+NO\textsubscript{2}+SO\textsubscript{2}+SS+WS} &
\textbf{3289.498} & \textbf{0.02849}\tabularnewline
M7 &
PM10+NO\textsubscript{2}+SO\textsubscript{2}+SS+WS+O\textsubscript{3} &
3327.636 & 0.02919\tabularnewline
M8 &
PM10+NO\textsubscript{2}+SO\textsubscript{2}+SS+WS+O\textsubscript{3}+HM
& 3318.287 & 0.02931\tabularnewline
M9 &
PM10+NO\textsubscript{2}+SO\textsubscript{2}+SS+WS+O\textsubscript{3}+HM+CO
& 3324.240 & 0.02953\tabularnewline
M10 &
PM10+NO\textsubscript{2}+SO\textsubscript{2}+SS+WS+O\textsubscript{3}+HM+CO+DTR
& 3348.394 & 0.03135\tabularnewline
\bottomrule
\end{longtable}
Table 2. Summary of outdoor air pollutants and environmental factors\selectlanguage{english}
\begin{longtable}[]{@{}ll@{}}
\toprule
\textbf{Factor} & \textbf{Daily median {[}range{]}}\tabularnewline
\midrule
\endhead
\textbf{Outdoor air pollutant (data from Ministry of Environment)}
&\tabularnewline
PM10 (\selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}) & 46.9794 {[}8.2000,
626.1082{]}\tabularnewline
NO\textsubscript{2} (ppb) & 23.245 {[}5.33000,
71.17143{]}\tabularnewline
SO\textsubscript{2} (ppb) & 4.80625 {[}1.60000
17.32143{]}\tabularnewline
O\textsubscript{3} (ppb) & 34.27667 {[}4.55, 99.04{]}\tabularnewline
CO (ppb/24hr) & 480.05 {[}239.8, 1357{]}\tabularnewline
\textbf{Weather condition (data from Korea Meteorological
Administration)} &\tabularnewline
WS (m/s) & 2.04125 {[}1.011688 7.186957{]}\tabularnewline
SS (h) & 5.917177 {[}0.00, 12.26{]}\tabularnewline
HM (\%) & 68.7424 {[}30.4688, 91.9677{]}\tabularnewline
DTR ([?]C) & 9.6687 {[}2.8012, 18.4892{]}\tabularnewline
\bottomrule
\end{longtable}
PM10, Particulate matter less than 10 \selectlanguage{greek}µ\selectlanguage{english}m; SO2, Sulfur dioxide; NO2,
Nitrogen dioxide; O3, ozone; WS, Wind speed; SS, Solar sunshine; CO,
Carbon monoxide; HM, humidity; DTR, diurnal temperature range.
Table 3. HR of CE according to representative value of air pollutants
and environmental factors\selectlanguage{english}
\begin{longtable}[]{@{}llllllllll@{}}
\toprule
\begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{Factor}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{Lag day}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{Median}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{HR} \textbf{Mean {[}LL, UL{]}}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{Concentration}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{HR}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{Concentration}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{HR}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{Concentration}\strut
\end{minipage} & \begin{minipage}[b]{0.10\columnwidth}\raggedright\strut
\textbf{HR}\strut
\end{minipage}\tabularnewline
\midrule
\endhead
\begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\textbf{Outdoor air pollutant}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage}\tabularnewline
\begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
PM10 (\selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3})\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
5\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
46.98\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\textbf{1.0404 {[}1.0089, 1.0729{]}*}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
30\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\textbf{1.0253 {[}1.0056, 1.0454{]}*}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
80\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\textbf{1.0704 {[}1.0153, 1.1285{]}*}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
150\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\textbf{1.1370 {[}1.0292, 1.2562{]}*}\strut
\end{minipage}\tabularnewline
\begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
NO\textsubscript{2} (ppb)\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
0\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
23.25\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1.0554 {[}0.9140, 1.2187{]}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
30\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1.0737 {[}0.8883, 1.2979{]}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
50\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1.1278 {[}0.8186, 1.5537{]}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage}\tabularnewline
\begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
SO\textsubscript{2} (ppb)\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
4.81\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1.0198 {[}0.9473, 1.0979{]}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
10\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1.0452 {[}0.8854, 1.2339{]}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
20\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1.0979 {[}0.7734, 1.5586{]}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage}\tabularnewline
\begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\textbf{Weather condition}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage}\tabularnewline
\begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
WS (m/s)\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
2.04\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1.0181 {[}0.9779, 1.0598{]}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage}\tabularnewline
\begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
SS (h)\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
7\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
5.92\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
1.0213 {[}0.9934, 1.0500{]}\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage} & \begin{minipage}[t]{0.07\columnwidth}\raggedright\strut
\strut
\end{minipage}\tabularnewline
\bottomrule
\end{longtable}
Abbreviations: CE, chronic obstructive pulmonary disease exacerbation;
HR, hazard ratio; {[}LL, UL{]}, 95\% confidence interval {[}lower limit,
upper limit{]}.
\textbf{*: statistically significant.}
Lag day 0: current day; Lag day 1: 1 day previously to Lag day 10: 10
days previously.
Table 4. HR of CE according to stratified age and lag days\selectlanguage{english}
\begin{longtable}[]{@{}llllllllll@{}}
\toprule
\textbf{Factor} & \textbf{Concentration (median)} & \textbf{Preschool
age} & \textbf{Preschool age} & \textbf{School age} & \textbf{School
age} & \textbf{Adult} & \textbf{Adult} & \textbf{Elderly} &
\textbf{Elderly}\tabularnewline
\midrule
\endhead
& & \textbf{Lag day} & \textbf{HR {[}LL, UL{]}} & \textbf{Lag day} &
\textbf{HR {[}LL, UL{]}} & \textbf{Lag day} & \textbf{HR {[}LL, UL{]}} &
\textbf{Lag day} & \textbf{HR {[}LL, UL{]}}\tabularnewline
\textbf{Outdoor air pollutant} & & & & & & & & &\tabularnewline
PM10 (\selectlanguage{greek}µ\selectlanguage{english}g/m\textsuperscript{3}) & 46.98 & 5 & 1.0296 {[}0.0951, 11.142{]}
& 0 & 0.9797 {[}0.3999, 2.3998{]} & \textbf{3 (3,4)} & \textbf{1.0372
{[}1.0058, 1.0696{]}*} & \textbf{3 (3,4,5,6)} & \textbf{1.0372
{[}1.0155, 1.0593{]}*}\tabularnewline
NO\textsubscript{2} (ppb) & 23.25 & 5 & 1.4613 {[}0.1080, 19.769{]} & 0
& 1.4613 {[}0.6989, 3.0556 & \textbf{0} & \textbf{1.4613 {[}1.3305,
1.6050{]}*} & \textbf{6} & \textbf{1.0774 {[}1.0162,
1.1423{]}*}\tabularnewline
SO\textsubscript{2} (ppb) & 4.81 & 1 & 1.0884 {[}0.0177, 66.736{]} & 0 &
1.0884 {[}0.0300, 39.447{]} & 1 & 1.0884 {[}0.7157, 1.6552{]} & 0 &
1.0884 {[}0.7446, 1.5909{]}\tabularnewline
WS (m/s) & 2.04 & 1 & 1.0215 {[}0.1542, 6.7653{]} & 1 & 1.0215
{[}0.6505, 1.6040{]} & \textbf{0} & \textbf{1.1320 {[}1.0736,
1.1936{]}*} & \textbf{0} & \textbf{1.1320 {[}1.0850,
1.1811{]}*}\tabularnewline
SS (h) & 5.92 & 4 & 1.0148 {[}0.3907, 2.6359{]} & 1 & 0.9748 {[}0.5410,
1.7562{]} & \textbf{0} & \textbf{0.9444 {[}0.9079, 0.9824{]}*} &
\textbf{0} & \textbf{0.9444 {[}0.9154, 0.9744{]}*}\tabularnewline
\bottomrule
\end{longtable}
Abbreviations: CE, chronic obstructive pulmonary disease exacerbation;
HR, hazard ratio; {[}LL, UL{]}, 95\% confidence interval {[}lower limit,
upper limit{]}.
\textbf{*: statistically significant.}
Lag day 0: current day; Lag day 1: 1 day previously, to Lag day 10: 10
days previously.\selectlanguage{english}
\begin{figure}[H]
\begin{center}
\includegraphics[width=0.70\columnwidth]{figures/Figure-1/Figure-1}
\end{center}
\end{figure}\selectlanguage{english}
\begin{figure}[H]
\begin{center}
\includegraphics[width=0.70\columnwidth]{figures/FIGURE-2-V5/FIGURE-2-V5}
\end{center}
\end{figure}\selectlanguage{english}
\begin{figure}[H]
\begin{center}
\includegraphics[width=0.70\columnwidth]{figures/Figure-3/Figure-3}
\end{center}
\end{figure}
\selectlanguage{english}
\FloatBarrier
\end{document}