Journal of Affective Disorders 157 (2014) 45–51

Contents lists available at ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

Research report

Air pollution as a risk factor for depressive episode in patients with cardiovascular disease, diabetes mellitus, or asthma Jaelim Cho a, Yoon Jung Choi b, Mina Suh c, Jungwoo Sohn a, Hyunsoo Kim a, Seong-Kyung Cho c, Kyoung Hwa Ha a, Changsoo Kim a,n, Dong Chun Shin a,d a

Department of Preventive Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea Research and Development Center, Health Insurance Review and Assessment Service, Seoul, Republic of Korea c National Cancer Center, Koyang, Republic of Korea d Institute for Environmental Research, Yonsei University College of Medicine, Seoul, Republic of Korea b

art ic l e i nf o

a b s t r a c t

Article history: Received 12 October 2013 Received in revised form 1 January 2014 Accepted 2 January 2014 Available online 10 January 2014

Background: There is currently insufficient evidence to confirm the effect of ambient air pollution on mental disorders, especially among susceptible populations. This study investigated the short-term effect of ambient air pollution on the risk of depressive episode and the effect modification across disease subpopulations. Methods: Subjects who visited the emergency department (ED) for depressive episode from 2005 to 2009 (n ¼4985) in Seoul, Republic of Korea were identified from medical claims data. We conducted a time-stratified case-crossover study using conditional logistic regression. Subgroup analyses were conducted after the subjects were stratified by underlying disease (cardiovascular disease, diabetes mellitus, chronic obstructive pulmonary disease, asthma, and depressive disorder). The risk was expressed as an odds ratio (OR) per 1 standard deviation of each air pollutant. Results: SO2, PM10, NO2, and CO were positively associated with ED visits for depressive episode. The maximum risk was observed in the distributed lag 0–3 model for PM10 (OR, 1.120; 95% confidence interval, 1.067–1.176). PM10, NO2, and CO significantly increased the risks of ED visits for depressive episode in subjects with either underlying cardiovascular disease, diabetes mellitus, asthma, or depressive disorder. Limitations: Our data may include a misclassification bias due to the validity of a diagnosis determined from medical services utilization data. Conclusions: SO2, PM10, NO2, and CO significantly increased the risk of ED visits for depressive episode, especially among individuals with pre-existing cardiovascular disease, diabetes mellitus, or asthma. & 2014 Elsevier B.V. All rights reserved.

Keywords: Air pollution Depressive episode Cardiovascular disease Diabetes mellitus Asthma

1. Introduction Unipolar depressive disorder, which was the third leading cause of disease burden in 2004, is projected to become the primary cause of disease burden worldwide by 2030, according to the Global Burden of Disease report (World Health Organization, 2008). Individuals who suffer from mental disorders, including depressive disorder, are at an increased risk for suicide, which is one of the leading causes of death (World Health Organization, 2005). Therefore, depressive disorder has become a major public health challenge. Ambient air pollution significantly increases the risk of acute myocardial infarction, arrhythmia, stroke, and type 2 diabetes mellitus (DM) (Sun et al., 2010). Recent epidemiological studies have also suggested an association between ambient air pollution and mental disorders such as depression and suicide. These

n

Corresponding author. Tel.: þ 82 2 2228 1880; fax: þ 82 2 392 8133. E-mail address: [email protected] (C. Kim).

0165-0327/$ - see front matter & 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jad.2014.01.002

studies have reported a short-term effect of PM10, SO2, NO2, and CO on emergency department (ED) visits for depressive disorder (Szyszkowicz et al., 2009) and a long-term effect on depressive symptoms in the elderly (Lim et al., 2012). Consequently, air pollutants, especially particulate matter, may increase the risk of suicide attempt (Szyszkowicz et al., 2010) or suicide (Kim et al., 2010). Depressive symptoms might be associated with inflammatory processes (Raison et al., 2006), which can be induced by air pollutants (Block and Calderon-Garciduenas, 2009). Air pollutants might also cause vascular depression by damaging endothelial vasculature in the brain (Steffens et al., 2003). However, there is currently insufficient evidence to confirm a relationship between air pollution and mental disorders. Depressive disorder is frequently comorbid with other diseases such as cardiovascular disease (CVD), DM, and asthma (Marshall, 2004; Musselman et al., 1998; Ridker et al., 1997; Stuart and Baune, 2012). Several experimental studies have suggested psycho-endocrine–immune interactions through hypothalamic– pituitary–adrenal (HPA) axis dysregulation and inflammatory

46

J. Cho et al. / Journal of Affective Disorders 157 (2014) 45–51

processes (Reichlin, 1993) to explain the bidirectional association of depression with DM (Stuart and Baune, 2012) or allergic asthma (Marshall, 2004). CVD is also related to depression through HPA axis dysregulation (Musselman et al., 1998) and systemic inflammation (Ridker et al., 1997). Air pollutants may trigger depressive symptoms via inflammatory processes in the brain (Block and Calderon-Garciduenas, 2009), especially among patients with CVD, DM, or asthma who may also have an undiagnosed depressive disorder. Moreover, the association between elevated inflammatory cytokines and the risk of suicide in major depression has been reported (Lindqvist et al., 2009; Nassberger and Traskman-Bendz, 1993). However, most studies on the association between air pollution and mental disorders used secondary data that did not include information about past medical history (Szyszkowicz, 2007; Szyszkowicz et al., 2009). Although one previous study has used a panel data design, researchers did not explore the effect of medication or underlying disease, except CVD, on the association between air pollution and mental disorders (Lim et al., 2012). Therefore, in this study, we investigated the short-term effect of ambient air pollution on ED visits for depressive episode and the interactive effect of ambient air pollution on the risk of depressive episode across disease subpopulations using large representative samples from Seoul, Republic of Korea.

2. Methods 2.1. Study subjects Subjects who were enrolled in the study visited the ED for depressive episode from January 1, 2005 to December 31, 2009 in Seoul, Republic of Korea. We identified the subjects from medical claims data which were reported to the Health Insurance Review and Assessment Service (HIRA), which is part of the National Health Insurance program of the Republic of Korea. The information identified from the HIRA data included age, sex, diagnosis code (International Classification of Diseases, 10th revision), and the date of the visit (Cho et al., 2013; Kim et al., 2010). The pre-existing illness of each subject was identified from prior medical services utilization history. However, due to the limitations of these data, we defined pre-existing illness as a disease that had been reported on at least 3 outpatient visits or Z1 hospitalization for the disease during the three years prior to the ED visit. All of the cases were classified into 5 groups: CVD (hypertensive disease [I10–I15], ischemic heart disease [I20–I25], and stroke [I60–I69]), DM (E10–E14), chronic obstructive pulmonary disease (COPD) (J40–J44), asthma (J45–J46), and depressive episode (F32). 2.2. Air pollutants and meteorological variables The air pollutants that were considered in this study included sulfur dioxide (SO2), particulate matter with an aerodynamic diameter less than 10 μm (PM10), ozone (O3), nitrogen dioxide (NO2), and carbon monoxide (CO). Data were available from 27 recording stations in Seoul from January 2005 to March 2009, 26 stations in April 2009, and 25 stations from May 2009 to the end of the study period. The average daily concentration from these stations was obtained from the Ministry of Environment. The methods of measurement used for each air pollutant were the beta-ray absorption method (PM10), the pulse ultraviolet (UV) fluorescence method (SO2), the chemiluminescent method (NO2), the UV photometric method (O3), and the non-dispersive infrared method (CO). Meteorological data, including temperature, relative humidity, sunlight hours, and air pressure, were obtained from the National Meteorological Office. The period of air pollutant and

meteorological data included in the analysis matched the period of ED visits. 2.3. Statistical analysis To investigate the short-term effect of ambient air pollution on daily ED visits for depressive episode, a time-stratified casecrossover design was used. The case period was defined as the date of each ED visit and the control periods were defined as the same days of the week as the case period within the same month, which allowed for 3 or 4 control days per case day. Unmeasured confounders such as area of residence and socioeconomic status among subjects were controlled by the study design because cases served as their own controls (Janes et al., 2005; Kim et al., 2010). Cox proportional hazard models were used to estimate the risk of ED visits for depressive disorder due to each air pollutant in both single and distributed lag models. Single lag models were used to explore the association of ED visits with air pollutant levels on the current day, previous day, 2 days, or 3 days prior to the day of a visit (described as lag 0, lag 1, lag 2, and lag 3). However, since ED visits would not only depend on the same day (lag 0) or previous day0 s (lag 1) effect of air pollutants, we estimated the cumulative effect of air pollution with distributed lag models, which included all air pollutant concentrations from the day of a visit to subsequent specified days in the same statistical model (described as lag0–1, lag0–2, or lag0–3) (Schwartz, 2000). The models included standardized air pollutant concentrations, national holidays, sunlight hours, and the natural cubic spline of the following variables: temperature (df ¼30), relative humidity (df¼ 15), and air pressure df ¼15) (Kim et al., 2010). We conducted the same analyses using 1-day to 3-day single lags (lag1, lag2, and lag3) as well as distributed lags (lag0–1, lag0–2, and lag0–3). Subgroup analyses were performed by season (spring, summer, fall, and winter), age ( o19, 19–39, 40–64, and Z65 years), sex, and pre-existing illness (CVD, DM, COPD, asthma, and depressive disorder) using the distributed lag model that resulted in the highest risk estimate (lag0–3 for SO2 and PM10 and lag0–2 for the other pollutants). Because air pollutants share common sources and are highly correlated, we also used two-pollutant models that included two pollutants in the same model (PM10 with SO2, PM10 with NO2, PM10 with CO, SO2 with NO2, SO2 with CO, and NO2 with CO) to explore independent effects of air pollutants on the risk of ED visits for depressive episode. Risk was expressed as an odds ratio (OR) and 95% confidence interval (95% CI) per 1 standard deviation of each air pollutant. We examined statistically significant differences between risk estimates across strata of potential effect modifiers (i.e., the difference between each underlying disease and none of the five diseases) (Altman and Bland, 2003). SAS version 9.3 (SAS institute, Cary, NC) was used to conduct the statistical analyses. All analyses were performed using the HIRA computer systems. The identity of each subject was blinded by the use of a randomly generated identification number.

3. Results A total of 4985 cases visited the ED for depressive episode during the study period (Table 1). The average age of the study subjects was 44 7 17, 73.3% were female, and 31.7% had a prior history of depressive disorder. In the same-day exposure model (lag0), the adjusted OR of PM10 was 1.065 (95% CI, 1.032–1.098) per 1 standard deviation (36.70 μg/m3) (Fig. 1). All air pollutants were positively associated with ED visits for depressive episode in the lag0 model with the exception of O3. The associations between air pollutants and the

J. Cho et al. / Journal of Affective Disorders 157 (2014) 45–51

risk of depressive disorder remained significant in the lag1 and lag2 models. The adjusted ORs of PM10 and SO2 were 1.120 (1.067–1.176) and 1.103 (1.043–1.166), respectively, in the distributed lag0–3 model. The highest risk of O3 (OR, 1.059; 95% CI, 0.995–1.127), NO2 (1.082; 1.033–1.133), and CO (1.077; 1.026– 1.130) was observed in the distributed lag0–2 model. After stratification of the subjects by underlying disease, PM10, NO2, and CO were positively associated with ED visits for depressive episode in each disease strata with the exception of COPD (Table 2); SO2 exposure was significantly associated with the risk of ED visits for depressive episode in subjects with either underlying CVD, asthma, or depressive disorder. Among the subjects with a past history of asthma, the increased risk was appeared to be statistically significant for PM10 (OR, 1.305; 95% CI, 1.045–1.628). After excluding asthma patients with a past history of depressive disorder, all air pollutants with the exception of O3 were positively

Table 1 Characteristics of emergency department (ED) visits for depressive episode in Seoul, Republic of Korea from 2005 to 2009. N (%)

Variables Total number of visits Age (years)

47

associated with the risk of ED visits for depressive episode. In subjects who did not have any of the five diseases, SO2 and O3 were significantly associated with the risk of ED visits for depressive episode. However, in subjects with any of the five diseases, all air pollutants with the exception of O3 were significantly associated with the risk of ED visits for depressive episode. No synergistic effect modification was associated with either SO2, PM10, NO2, or CO for any group with an underlying disease. The results of the subgroup analyses by age, sex, and season are shown in Table 3. Subjects aged 19–39 years showed an increased risk associated with PM10 and CO, and in subjects aged 40–64 years, the risk estimate was statistically significant for SO2 and PM10. In both males and females, increased risks were observed in all air pollutants with the exception of O3. The risk of ED visits for depressive disorder was significantly increased during the spring (SO2, PM10, and NO2) and fall (SO2, PM10, NO2, and CO). In the two-pollutant model, PM10 and SO2 were independently associated with a risk of ED visits for depressive disorder after adjustments for other air pollutants, whilst the risk estimates were not statistically significant for NO2 and CO (Table 4).

4. Discussion

o 19 19–39 40–64 Z 65

4985 260 1965 2036 724

(100.0) (5.1) (38.8) (40.2) (14.3)

Sex

Male Female

1275 (25.2) 3710 (73.3)

Underlying disease

CVDa Without depressive disorderb DM (E10-E14) Without depressive disorderb COPD (J40-J44) Without depressive disorderb Asthma (J45-J46) Without depressive disorderb Depressive disorder (F32) Any of the five diseases

861 493 450 259 548 294 311 196 1579 2486

(17.3) (14.5) (9.0) (7.6) (11.0) (8.6) (6.2) (5.8) (31.7) (49.9)

Season

Spring Summer Fall Winter

1357 1296 1260 1072

(27.2) (26.0) (25.3) (21.5)

CVD: Cardiovascular disease, DM: Diabetes mellitus, COPD: Chronic obstructive pulmonary disease. a CVD refers to hypertensive disease (I10–I15), ischemic heart disease (I20–I25), and stroke (I60-I69). b Subjects with a past history of depressive disorder prior to emergency department visits were excluded.

In the present study, we found that ambient air pollutants such as PM10, CO, NO2, and SO2 were significantly associated with ED visits for depressive episode. The maximum risk was observed in the distributed lag 0–2 (NO2 and CO) and lag 0–3 (PM10 and SO2) models. Our results were consistent with previous studies, which reported a positive correlation between ED visits for depressive episode and same-day exposure to PM10, CO, and NO2 (Szyszkowicz, 2007; Szyszkowicz et al., 2009). Our results were also consistent with a study that reported a positive association between depressive symptoms and air pollutant exposure in a cumulative lag model (PM10 over 0–2 days, NO2 over 0–7 days, and O3 over 0–2 days) among an elderly population (Lim et al., 2012). Given that approximately 90% of individuals who commit suicide have pre-existing psychiatric disorders, especially depressive disorders (Arsenault-Lapierre et al., 2004; Henriksson et al., 1993), air pollutants may play an important role in the aggravation of depressive symptoms and consequently may increase the risk of suicidal behavior or suicide. Recent epidemiologic studies have reported a possible association between PM10 and suicidal behavior or suicide (Kim et al., 2010; Szyszkowicz et al., 2010). Meanwhile, the risk of ED visits for depressive episode was not significant for O3 exposure. O3 is a secondary air pollutant generated from primary air pollutants such as NO2 via photochemical reaction

Fig. 1. Adjusted ORsa and 95% CI of ED visits for depressive episode using single lag (lag0, lag1, lag2, and lag3) and distributed lag (lag0–1, lag0–2, and lag0–3) models. a Adjusted for national holidays, sunlight hours, and natural cubic spline variables (mean temperature [df ¼30], relative humidity [df ¼ 15], and air pressure [df ¼ 15]); odds ratios per 1 standard deviation of each air pollutant (SO2, 2.33 ppb; PM10, 36.70 μg/m3; O3, 10.04 ppb; NO2, 12.04 ppb; CO, 0.24 ppm).

48

J. Cho et al. / Journal of Affective Disorders 157 (2014) 45–51

Table 2 Adjusted ORsa and 95% CI of ED visits for depressive episode by underlying disease using distributed lag models (lag0–3 for SO2 and PM10, lag0–2 for the other pollutants). SO2 CVDb Without depressive disorderc DM Without depressive disorderc COPD Without depressive disorderc Asthma Without depressive disorderc Depressive disorder Any of the five diseases None of the five diseases

1.098 1.203 1.130 1.124 0.961 0.979 1.211 1.360 1.126 1.156 1.103

PM10 (0.960–1.257) (1.010–1.432) (0.934–1.367) (0.878–1.438) (0.805–1.147) (0.771–1.243) (0.964–1.521) (1.035–1.788) (1.020–1.243) (1.073–1.245) (1.019–1.193)

1.146 1.186 1.289 1.361 1.003 1.070 1.305 1.408 1.155 1.168 1.023

O3 (1.016–1.291) (1.009–1.394) (1.092–1.523) (1.083–1.710) (0.851–1.181) (0.850–1.347) (1.045–1.628) (1.075–1.844) (1.058–1.262) (1.094–1.246) (0.950–1.103)

NO2

0.985 0.962 0.902 0.855 0.968 1.141 1.046 0.954 1.004 0.859 1.248

(0.843–1.150) (0.784–1.182) (0.725–1.122) (0.650–1.126) (0.799–1.171) (0.881–1.479) (0.811–1.350) (0.693–1.313) (0.896–1.125) (0.789–0.936) (1.145–1.361)

1.089 1.189 1.245 1.239 1.027 1.048 1.068 1.330 1.098 1.131 1.053

CO (0.973–1.219) (1.026–1.377) (1.059–1.463) (1.006–1.526) (0.887–1.190) (0.863–1.273) (0.883–1.292) (1.056–1.676) (1.011–1.192) (1.062–1.206) (0.987–1.124)

1.130 1.196 1.244 1.214 1.036 1.083 1.107 1.292 1.135 1.171 0.985

(1.006–1.269) (1.030–1.390) (1.054–1.470) (0.986–1.493) (0.892–1.203) (0.890–1.316) (0.914–1.340) (1.014–1.646) (1.043–1.236) (1.095–1.252) (0.921–1.053)

CVD: Cardiovascular disease, DM: Diabetes mellitus, COPD: Chronic obstructive pulmonary disease. a Adjusted for national holidays, sunlight hours, and natural cubic spline variables (mean temperature [df ¼ 30], relative humidity [df ¼ 15], and air pressure [df ¼ 15]); odds ratios per 1 standard deviation of each air pollutant (SO2, 2.33 ppb; PM10, 36.70 μg/m3; O3, 10.04 ppb; NO2, 12.04 ppb; CO, 0.24 ppm). b CVD refers to hypertensive disease (I10–I15), ischemic heart disease (I20–I25), and stroke (I60–I69). c Subjects with a past history of depressive disorder prior to emergency department visits were excluded.

Table 3 Adjusted ORsa and 95% CI of ED visits for depressive episode by age, sex, and season using distributed lag models (lag0–3 for SO2 and PM10, lag0–2 for the other pollutants). PM10

SO2

O3

NO2

CO

Age (years)

o 19 19–39 40–64 Z 65

1.250 1.076 1.101 1.151

Sex

Male Female

1.131 (1.015–1.260) 1.097 (1.027–1.171)

1.138 (1.034–1.251) 1.112 (1.051–1.177)

1.000 (0.883–1.134) 1.064 (0.990–1.144)

1.128 (1.028–1.239) 1.076 (1.019–1.137)

1.152 (1.049–1.267) 1.066 (1.007–1.129)

Season

Spring Summer Fall Winter

1.238 0.786 1.207 1.049

1.207 0.772 1.165 0.904

1.071 0.834 1.544 1.043

1.197 1.023 1.077 1.035

1.088 1.000 1.119 1.028

(0.989–1.580) (0.985–1.176) (1.008–1.203) (0.993–1.334)

(1.098–1.396) (0.578–1.070) (1.028–1.418) (0.940–1.171)

1.113 1.124 1.106 1.122

(0.883–1.403) (1.041–1.214) (1.022–1.197) (0.985–1.279)

(1.107–1.317) (0.639–0.933) (1.016–1.335) (0.771–1.059)

0.894 1.087 1.044 0.983

(0.669–1.194) (0.983–1.201) (0.946–1.152) (0.832–1.162)

(0.933–1.229) (0.725–0.959) (1.244–1.917) (0.804–1.354)

1.293 1.067 1.077 1.149

(1.048–1.596) (0.989–1.150) (1.001–1.158) (1.016–1.300)

(1.066–1.343) (0.865–1.209) (0.975–1.191) (0.926–1.157)

1.207 1.081 1.061 1.156

(0.973–1.496) (1.001–1.167) (0.982–1.146) (1.020–1.310)

(0.913–1.297) (0.806–1.241) (1.011–1.238) (0.938–1.127)

a Adjusted for national holidays, sunlight hours, and natural cubic spline variables (mean temperature [df ¼30], relative humidity [df ¼ 15], and air pressure [df ¼ 15]); odds ratios per 1 standard deviation of each air pollutant (SO2, 2.33 ppb; PM10, 36.70 μg/m3; O3, 10.04 ppb; NO2, 12.04 ppb; CO, 0.24 ppm).

Table 4 Adjusted ORsa and 95% CI of ED visits for depressive episode using two-pollutant models that include interaction terms between pollutants.

(Haagen-Smit et al., 1952), so it is possible that the primary air pollutants might trigger depressive symptoms before O3 concentration starts to increase.

OR (95% CI)

4.1. Differences by underlying diseases PM10 and SO2 (lag0–3)

PM10 SO2 PM10  SO2

1.080 (1.016–1.147) 1.103 (1.007–1.209) 0.954 (0.902–1.008)

PM10 and NO2 (lag0–3)

PM10 NO2 PM10  NO2

1.109 (1.048–1.174) 1.024 (0.959–1.093) 0.986 (0.940–1.035)

PM10 and CO (lag0–3)

PM10 CO PM10  CO

1.120 (1.055–1.189) 1.010 (0.938–1.087) 0.989 (0.939–1.042)

SO2 and NO2 (lag0–3)

SO2 NO2 SO2  NO2

1.156 (1.045–1.279) 0.996 (0.921–1.078) 0.964 (0.926–1.004)

SO2 and CO (lag0–3)

SO2 CO SO2  CO

1.148 (1.047–1.258) 1.020 (0.936–1.110) 0.962 (0.927–0.999)

NO2 and CO (lag0–2)

NO2 CO NO2  CO

1.060 (0.960–1.170) 1.059 (0.944–1.189) 0.962 (0.926–1.000)

a Adjusted for national holidays, sunlight hours, and natural cubic spline variables (mean temperature [df ¼30], relative humidity [df ¼15], and air pressure [df ¼ 15]); odds ratios per 1 standard deviation of each air pollutant (SO2, 2.33 ppb; PM10, 36.70 μg/m3; O3, 10.04 ppb; NO2, 12.04 ppb; CO, 0.24 ppm).

In the present study, subjects with a pre-existing depressive disorder were at an increased risk of ED visits due to PM10 exposure. PM10 was also significantly associated with ED visits for depressive episode in subjects with pre-existing CVD, DM, or asthma, which is consistent with a previous study on ambient particulate matter that reported a significantly increased risk of suicide among subjects with pre-existing CVD (Kim et al., 2010). In subjects with pre-existing DM or asthma, PM10 was significantly associated with an increased risk of ED visits for depressive episode even after the exclusion of cases with pre-existing depressive disorder. We did not observe any significant risk in subjects with underlying COPD even after subjects with an underlying depressive disorder were excluded. Recently, PM10 has been demonstrated to be a strong inflammatory agent (Calderon-Garciduenas et al., 2008). Thus, one possible hypothesis is that the difference in significance and the intensity of association across pre-existing illnesses may arise from the interactive effects of air pollution on psycho-endocrine–immune connections via an inflammatory process. Depressive disorder, DM, and asthma belong to the psycho-endocrine–immune network system;

J. Cho et al. / Journal of Affective Disorders 157 (2014) 45–51

CVD and COPD do not. Abnormalities in psycho-endocrine–immune connections are based on genetics as well as fetal or early childhood events such as infection or stress (Patterson, 2011). CVD and COPD are also related to systemic inflammation (Fabbri and Rabe, 2007; Ridker et al., 1997), but genetic susceptibility seems to be the more significant factor in the relationship between air pollution and depressive disorder. Therefore, individuals with any illnesses related to the psycho-endocrine–immune system may be at a higher risk for other illnesses related to this system due to air pollution exposure. The exact relationship and interaction of asthma and COPD with air pollutant exposure and ED visits for depressive episode may be difficult to elucidate because we were unable to identify the severity of these diseases in each case. The disease severity of asthma and COPD are related to the occurrence of depression through corticosteroid use and a reduced quality of life (Opolski and Wilson, 2005; Wilson, 2006). A study also suggested that the type and amount of medications such as steroids may play a role in the association between asthma and suicide (Kuo et al., 2010). Therefore, future studies to examine the relationship between air pollution and depressive disorder among asthma or COPD patients should incorporate the disease severity based on corticosteroid use. In subjects who did not have any of the five underlying diseases that we studied, the risks associated with air pollutants were either not statistically significant or lower than in those subjects with CVD, DM, COPD, asthma, or depressive disorder, except for SO2 and O3. The significance of the effect of SO2 on subjects without any of the five diseases might be associated with the algorithm that was used to identify the underlying disease; therefore, misclassified subjects might be present in our study (Supplementary Tables 2 and 3). The risks for O3 were also significantly higher in subjects without an underlying disease regardless of the algorithm that was used. A high risk due to O3 may be related to the diurnal variation of O3 and a difference in the pattern of personal behaviors between patients and healthy individuals (Kim et al., 2007), but the reason for this result remains unclear. Overall, the risks associated with air pollutants appeared to increase after subjects with pre-existing depressive disorder were excluded among the disease groups. However, these disease groups may have an undiagnosed or untreated depressive disorder, which may confound the results observed for CVD, DM, and asthma patients. Further investigations are necessary to confirm our findings. 4.2. Age, sex, and seasonal differences In the present study, SO2, PM10, and NO2 were positively associated with ED visits for depressive episode in 40–64 year olds and NO2 and CO were positively associated with ED visits for depressive episode in the 65 years or older group. In a cohort study, PM10, NO2, and O3 were significantly associated with increased depressive symptoms in the elderly despite differences in study design, study region, and lag structures (Lim et al., 2012). In the present study, the number of ED visits for depressive episode was high in female subjects, but the risk estimates were relatively high in males compared with females. Mood is more likely to be affected by estrogen fluctuation in females due to differences in the serotonin systems and HPA axis between males and females (Pitychoutis and Papadopoulou-Daifoti, 2010). This observation suggests that a hormonal effect may supersede an impact of air pollution on depressive disorder in females. In the present study, air pollutants were also significantly associated with ED visits for depressive episode during the spring and fall, which is consistent with a previous study that reported a relationship between particulate matter and suicide (Kim et al., 2010).

49

We also found that the risk associated with air pollutants was the highest for O3 in the fall, but not in the spring. The reason for this result is unclear because O3 concentration was lower in the fall than in the spring and summer. The highest risk may be related to the diurnal variation of O3 concentrations, which are higher during the daytime when outdoor activities are frequent (Kim et al., 2007). Especially, PM10, NO2, and SO2 concentrations were also lower in the fall, so solar radiation may be more intense compared to the other seasons. During the day, sunlight generates O3 from air pollutants such as oxides of nitrogen (NOx), volatile organic compounds (VOCs), and hydrocarbon via photochemical reactions (Haagen-Smit et al., 1952). Further measurements of NOx, VOCs, and hydrocarbon are necessary to examine this inference. Although our results suggest both sex and seasonal differences associated with the risk of ED visits for depressive episode, our results may be related to differences in personal exposure to air pollutants (Sarnat et al., 2005; Zeka et al., 2006). Due to the ecological nature of our study, we could not assess personal factors related to air pollutant exposure. Therefore, further studies are necessary to address seasonal variation and sex differences. 4.3. Two-pollutant models The concentrations of ambient air pollutants are highly correlated because they share common sources. Therefore, the effects estimated from single-pollutant models are more likely to be confounded by other pollutants. We further explored the association between ED visits for depressive episode and air pollutants using two-pollutant models. In the two-pollutant models, PM10 and SO2 were independently associated with ED visits for depressive episode after adjustments for other air pollutants. Our findings suggest that each air pollutant is at least independently associated with an increased risk of ED visits for depressive episode. 4.4. Possible biological mechanisms Several studies have suggested a possible biological mechanism for the effect of air pollutants on mental health that involves neuroinflammation (Block and Calderon-Garciduenas, 2009). PM10 may cause dopamine neuron damage followed by changes in neurotransmitter levels through an inflammatory process (Sirivelu et al., 2006; Veronesi et al., 2005). An animal experiment reported that O3 exposure may cause oxidative stress to brain structures, including the frontal cortex and striatum (DoradoMartinez et al., 2001). Oxidative damage caused by SO2 has also been observed in the rat brain (Meng et al., 2003). Oxidative stress and lipid peroxidation in rat brains exposed to CO have also been reported, but these experiments involved high concentrations of CO (Thom, 1990; Zhang and Piantadosi, 1992). These mechanisms could manifest in a sudden occurrence of severe depressive symptoms that lead to suicidal behavior. However, because of the limited biological evidence for the effect of air pollution on the brain, further studies are necessary to delineate the causal relationships between air pollution and mental disorders. 4.5. Strengths and limitations In the present study, we used a large and representative dataset. Because outpatient data included scheduled visits in addition to the exacerbation of symptoms, we restricted our analysis to ED cases that are acute and relatively severe. ED data also reflects where symptoms urgently are initiated or exacerbated, and so can cover a more limited area compared to other types of medical care. Therefore, using ED data and air pollution data of a city might be sufficiently suitable for identifying the short-term effect of air pollutants. Also, we used the average levels of air pollutants in Seoul, which might lead to

50

J. Cho et al. / Journal of Affective Disorders 157 (2014) 45–51

underestimation of the risk, but we observed positively significant associations. However, there are several limitations to consider. First, our results are not generalizable and should be cautiously applied to other regions because there may be potential confounding factors specific to Seoul that may influence ED visits. Furthermore, air pollutant concentrations are relatively high in Seoul (Supplementary Table 1). To explore the risk threshold associated with the exposure to each air pollutant, we used a Cox proportional hazard model that included air pollutant concentration as a penalized spline (df¼3) using cumulative lag models (moving average of lag0 to lag3 for SO2 and PM10, lag0 to lag2 for NO2 and CO) (Supplementary Fig. 1). The risk estimates showed linear trends in the concentration of each air pollutant. We did not find a threshold effect in any air pollutants, although our models could not account for potential threshold effects in lower concentrations. However, a study in Canada, which has lower concentrations of air pollutants than Seoul (e.g., mean PM10 ¼ 19.4 μg/m3 vs. 10th percentile¼ 22.0 μg/m3, respectively), also showed a positive association between ambient air pollution and ED visits for depressive disorder (Szyszkowicz et al., 2009). Second, our data may include a misclassification bias due to the validity of a diagnosis determined from medical services utilization data. To rule out medical illness is essential in depressive patients who present in an emergency setting (Lagomasino et al., 1999); therefore, patients may be diagnosed with a depressive episode when they do not present with any other abnormalities in test results. However, patients with depressive symptoms may not meet the criteria for depressive episode. In the present study, a focus on the symptomatology was more important than defining the presence of a depressive episode because the aim of our study was to examine whether air pollution had a short-term effect on mental health that could trigger severe depressive symptoms leading to suicidal behavior. Therefore, even if ED visits for depressive episode were misclassified, the misclassification bias is likely to be non-differential and underestimate the true risk. Meanwhile, ED cases for depressive episodes might have various psychiatric symptoms other than depression such as anxiety, suicidal ideation and psychosis, and it is possible that air pollution might trigger those conditions as well. In this study, because the health insurance claim data that we used contained only diagnosis codes, we could not ascertain accompanying psychiatric symptoms. However, diagnosis with depressive episode implies that depression might be of main interest in treatment, and we found that about 55% of ED cases of depressive episode were hospitalized for the same diagnosis. Additional clinical information on symptoms and prescribed medication might be helpful to improve diagnostic validity, and further studies on other psychitric conditions would be needed. Furthermore, we used an algorithm to improve the accuracy in identifying underlying diseases. Misclassification bias seems to exist because risk estimates vary by algorithm in sensitivity analyses (Supplementary Tables 2 and 3). Analyses that use algorithms that include at least 1 or 2 outpatient visits show that the risks are underestimated. However, the associations remained significant in subjects with underlying CVD, DM, asthma, and depressive disorder. Finally, due to observational nature of the study design, we could only demonstrate an association but not a causal relationship between air pollutants and ED visits for depressive episode. Also, there is a possibility that air pollutants might influence behavior related to ED visits rather than progression of depression. However, our results are consistent with previous epidemiological studies and several lines of biological evidence support the relationship suggested by our study. 4.6. Implications for managing mental health problems The present study suggests that SO2, PM10, NO2, and CO are positively associated with ED visits for depressive episode. Furthermore, the short-term exposure of SO2, PM10, NO2, and CO may significantly

increase the risk among subjects with pre-existing diseases such as CVD, DM, or asthma. Individuals with pre-existing CVD, DM, or asthma may also have undiagnosed or untreated depressive disorder. Therefore, the short-term exposure to air pollutants may trigger depressive symptom among these individuals. To prevent the exacerbation of depressive disorder and suicide due to air pollution, target strategies such as screening programs may be helpful among these disease subpopulations. Role of funding source This research was supported by the Yonsei University College of Medicine, Seoul, Republic of Korea (Grant number 6-2011-0116).

Conflict of interest The authors declare no conflict of interests.

Acknowledgments The authors would like to thank the Ministry of Environment, National Meteorological Office, and Health Insurance Review and Assessment Service for providing the data.

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jad.2014.01.002. References Altman, D.G., Bland, J.M., 2003. Interaction revisited: the difference between two estimates. BMJ 326 (7382), 219. Arsenault-Lapierre, G., Kim, C., Turecki, G., 2004. Psychiatric diagnoses in 3275 suicides: a meta-analysis. BMC Psychiatry 4, 37. Block, M.L., Calderon-Garciduenas, L., 2009. Air pollution: mechanisms of neuroinflammation and CNS disease. Trends Neurosci. 32 (9), 506–516. Calderon-Garciduenas, L., Solt, A.C., Henriquez-Roldan, C., Torres-Jardon, R., Nuse, B., Herritt, L., Villarreal-Calderon, R., Osnaya, N., Stone, I., Garcia, R., Brooks, D. M., Gonzalez-Maciel, A., Reynoso-Robles, R., Delgado-Chavez, R., Reed, W., 2008. Long-term air pollution exposure is associated with neuroinflammation, an altered innate immune response, disruption of the blood–brain barrier, ultrafine particulate deposition, and accumulation of amyloid beta-42 and alpha-synuclein in children and young adults. Toxicol. Pathol. 36 (2), 289–310. Cho, J., Kang, D.R., Moon, K.T., Suh, M., Ha, K.H., Kim, C., Suh, I., Shin, D.C., Jung, S.H., 2013. Age and gender differences in medical care utilization prior to suicide. J. Affect. Disord. 146 (2), 181–188. Dorado-Martinez, C., Paredes-Carbajal, C., Mascher, D., Borgonio-Perez, G., RivasArancibia, S., 2001. Effects of different ozone doses on memory, motor activity and lipid peroxidation levels, in rats. Int. J. Neurosci. 108 (3–4), 149–161. Fabbri, L.M., Rabe, K.F., 2007. From COPD to chronic systemic inflammatory syndrome? Lancet 370 (9589), 797–799. Haagen-Smit, A.J., Darley, E.F., Zaitlin, M., Hull, H., Noble, W., 1952. Investigation on injury to plants from air pollution in the Los Angeles Area. Plant Physiol. 27 (1), 18–34. Henriksson, M.M., Aro, H.M., Marttunen, M.J., Heikkinen, M.E., Isometsa, E.T., Kuoppasalmi, K.I., Lonnqvist, J.K., 1993. Mental disorders and comorbidity in suicide. Am. J. Psychiatry 150 (6), 935–940. Janes, H., Sheppard, L., Lumley, T., 2005. Case-crossover analyses of air pollution exposure data: referent selection strategies and their implications for bias. Epidemiology 16 (6), 717–726. Kim, C., Jung, S.H., Kang, D.R., Kim, H.C., Moon, K.T., Hur, N.W., Shin, D.C., Suh, I., 2010. Ambient particulate matter as a risk factor for suicide. Am. J. Psychiatry 167 (9), 1100–1107. Kim, S.-W., Yoon, S.-C., Won, J.-G., Choi, S.-C., 2007. Ground-based remote sensing measurements of aerosol and ozone in an urban area: a case study of mixing height evolution and its effect on ground-level ozone concentrations. Atmos. Environ. 41 (33), 7069–7081. Kuo, C.J., Chen, V.C., Lee, W.C., Chen, W.J., Ferri, C.P., Stewart, R., Lai, T.J., Chen, C.C., Wang, T.N., Ko, Y.C., 2010. Asthma and suicide mortality in young people: a 12year follow-up study. Am. J. Psychiatry 167 (9), 1092–1099. Lagomasino, I., Daly, R., Stoudemire, A., 1999. Medical assessment of patients presenting with psychiatric symptoms in the emergency setting. Psychiatr. Clin. North Am. 22 (4), 819–850 (viii–ix). Lim, Y.-H., Kim, H., Kim, J.H., Bae, S., Park, H.Y., Hong, Y.-C., 2012. Air pollution and symptoms of depression in elderly adults. Environ. Health Perspect. 120 (7), 1023–1028. Lindqvist, D., Janelidze, S., Hagell, P., Erhardt, S., Samuelsson, M., Minthon, L., Hansson, O., Bjorkqvist, M., Traskman-Bendz, L., Brundin, L., 2009. Interleukin-6

J. Cho et al. / Journal of Affective Disorders 157 (2014) 45–51 is elevated in the cerebrospinal fluid of suicide attempters and related to symptom severity. Biol. Psychiatry 66 (3), 287–292. Marshall, G.D., 2004. Neuroendocrine mechanisms of immune dysregulation: applications to allergy and asthma. Ann. Allergy Asthma Immunol. 93 (2 Suppl. 1), S11–17. Meng, Z., Qin, G., Zhang, B., Geng, H., Bai, Q., Bai, W., Liu, C., 2003. Oxidative damage of sulfur dioxide inhalation on lungs and hearts of mice. Environ. Res. 93 (3), 285–292. Musselman, D.L., Evans, D.L., Nemeroff, C.B., 1998. The relationship of depression to cardiovascular disease: epidemiology, biology, and treatment. Arch. Gen. Psychiatry 55 (7), 580–592. Nassberger, L., Traskman-Bendz, L., 1993. Increased soluble interleukin-2 receptor concentrations in suicide attempters. Acta Psychiatr Scand. 88 (1), 48–52. Opolski, M., Wilson, I., 2005. Asthma and depression: a pragmatic review of the literature and recommendations for future research. Clin. Pract. Epidemiol. Ment. Health 1, 18. Patterson, P.H., 2011. Infectious Behavior: Brain–Immune Connections in Autism, Schizophrenia, and Depression. The MIT Press, Cambridge, Massachusetts Pitychoutis, P.M., Papadopoulou-Daifoti, Z., 2010. Of depression and immunity: does sex matter? Int. J. Neuropsychopharmacol. 13 (5), 675–689. Raison, C.L., Capuron, L., Miller, A.H., 2006. Cytokines sing the blues: inflammation and the pathogenesis of depression. Trends Immunol. 27 (1), 24–31. Reichlin, S., 1993. Neuroendocrine–immune interactions. N. Engl. J. Med. 329 (17), 1246–1253. Ridker, P.M., Cushman, M., Stampfer, M.J., Tracy, R.P., Hennekens, C.H., 1997. Inflammation, aspirin, and the risk of cardiovascular disease in apparently healthy men. N. Engl. J. Med. 336 (14), 973–979. Sarnat, J.A., Brown, K.W., Schwartz, J., Coull, B.A., Koutrakis, P., 2005. Ambient gas concentrations and personal particulate matter exposures: implications for studying the health effects of particles. Epidemiology 16 (3), 385–395. Schwartz, J., 2000. The distributed lag between air pollution and daily deaths. Epidemiology 11 (3), 320–326. Sirivelu, M.P., MohanKumar, S.M., Wagner, J.G., Harkema, J.R., MohanKumar, P.S., 2006. Activation of the stress axis and neurochemical alterations in specific

51

brain areas by concentrated ambient particle exposure with concomitant allergic airway disease. Environ. Health Perspect. 114 (6), 870–874. Steffens, D.C., Taylor, W.D., Krishnan, K.R., 2003. Progression of subcortical ischemic disease from vascular depression to vascular dementia. Am. J. Psychiatry 160 (10), 1751–1756. Stuart, M.J., Baune, B.T., 2012. Depression and type 2 diabetes: inflammatory mechanisms of a psychoneuroendocrine co-morbidity. Neurosci. Biobehav. Rev. 36 (1), 658–676. Sun, Q., Hong, X., Wold, L.E., 2010. Cardiovascular effects of ambient particulate air pollution exposure. Circulation 121 (25), 2755–2765. Szyszkowicz, Jeff, Eric, B.W., Rowe, G., Ian, C., 2010. Air pollution and emergency department visits for suicide attempts in Vancouver, Canada. Environ. Health Insights 15 (4), 79–86. Szyszkowicz, M., 2007. Air pollution and emergency department visits for depression in Edmonton, Canada. Int. J. Occup. Med. Environ. Health 20 (3), 241–245. Szyszkowicz, M., Rowe, B.H., Colman, I., 2009. Air pollution and daily emergency department visits for depression. Int. J. Occup. Med. Environ. Health 22 (4), 355–362. Thom, S.R., 1990. Carbon monoxide-mediated brain lipid peroxidation in the rat. J. Appl. Physiol. 68 (3), 997–1003. Veronesi, B., Makwana, O., Pooler, M., Chen, L.C., 2005. Effects of subchronic exposures to concentrated ambient particles. VII. Degeneration of dopaminergic neurons in Apo E  / mice. Inhal. Toxicol. 17 (4–5), 235–241. Wilson, I., 2006. Depression in the patient with COPD. Int. J. Chron. Obstruct. Pulmon. Dis. 1 (1), 61–64. World Health Organization, 2005. Suicide Huge but Preventable Public Health Problem. World Health Organization, 2008. The Global Burden of Disease: 2004 Update. WHO Press, Geneva, Switzerland Zeka, A., Zanobetti, A., Schwartz, J., 2006. Individual-level modifiers of the effects of particulate matter on daily mortality. Am. J. Epidemiol. 163 (9), 849–859. Zhang, J., Piantadosi, C.A., 1992. Mitochondrial oxidative stress after carbon monoxide hypoxia in the rat brain. J. Clin. Invest. 90 (4), 1193–1199.