International Archives of Occupational and Environmental Health

, Volume 80, Issue 8, pp 701–710

Air pollution, socioeconomic position, and emergency hospital visits for asthma in Seoul, Korea

Authors

  • Sun-Young Kim
    • Department of Environment and Occupational Health Sciences, School of Public Health and Community MedicineUniversity of Washington
  • Marie S. O’Neill
    • Departments of Epidemiology and Environmental Health Sciences, School of Public HealthUniversity of Michigan
  • Jong-Tae Lee
    • Department of Health Management, Graduate SchoolHanyang University
  • Youngtae Cho
    • Department of Epidemiology and Biostatistics, Graduate School of Public Health and The Institute of Health and EnvironmentSeoul National University
  • Jaiyong Kim
    • Primary Health Care Team, Research DepartmentKorea Health Insurance Review Agency
    • Department of Epidemiology and Biostatistics, Graduate School of Public Health and The Institute of Health and EnvironmentSeoul National University
Original Article

DOI: 10.1007/s00420-007-0182-3

Cite this article as:
Kim, S., O’Neill, M.S., Lee, J. et al. Int Arch Occup Environ Health (2007) 80: 701. doi:10.1007/s00420-007-0182-3

Abstract

Objective

Some epidemiological literature has observed that air pollution effects on health differed across regional or individual socioeconomic position. This study evaluated whether regional and individual socioeconomic position, as indicated by health insurance premiums, modified the effect of air pollution on hospital visits for asthma.

Methods

Effects of ambient air pollutants (particulate matter, carbon monoxide, sulfur dioxide, nitrogen dioxide, and ozone) on 92,535 emergency out-patient hospital visits for asthma in Seoul, Korea during 2002 were estimated using case-crossover analysis, adjusting for time trend, weather conditions, and seasonality. Next, interactions between air pollutants and Korean National Health Insurance premium (1) for the individual patient and (2) averaged across the patient’s residence district, were entered, first singly then jointly, in the models.

Results

Relative risks of emergency outpatient hospital visits were all positively and significantly associated with interquartile increases for selected lags for all air pollutants. In the regression model with interaction terms for both individual premium and regional-average premium, associations with all five-air pollutants ranged from 1.03 to 1.09 times higher among the lowest premium districts compared to the highest premium districts. Of all the pollutants, nitrogen dioxide showed the strongest associations in lower premium districts compared to the higher premium districts. Individual socioeconomic position did not modify the associations in either the single or joint interaction models.

Conclusion

In Seoul, community but not individual socioeconomic conditions modified risk of asthma hospital visits on high air pollution days.

Keywords

Air pollutionAsthmaEffect modifierResidence characteristicsSocioeconomic factors

Abbreviations

CI

Confidence interval

CO

Carbon monoxide

KNHI

Korean National Health Insurance

NO2

Nitrogen dioxide

PM10

Particulate matter ≤10 μm in aerodynamic diameter

REM

Relative effect modification

RR

Relative risk

SEP

Socioeconomic position

SO2

Sulfur dioxide

Introduction

Numerous epidemiological studies have shown links between air pollution exposure and adverse health outcomes (Brunekreef and Holgate 2002). Other studies have explored whether these associations vary by socioeconomic position (SEP) as measured at the individual level (e.g., educational attainment) (Hoek et al. 2002), or at the area level (e.g., percent residents with low income or education, Martins et al. 2004); percent housing with electricity or running water (O’Neill et al. 2004); or with deprivation indices (Gouveia and Fletcher 2000). Although several studies have evaluated how socioeconomic factors modify the effect of air pollution on daily mortality (Hoek et al. 2002; Martins et al. 2004; O’Neill et al. 2004; Gouveia and Fletcher 2000; Samet et al. 2000; Zanobetti and Schwartz 2000; Jerrett et al. 2004; Villeneuve et al. 2003), only a few have explored SEP, air pollution, and morbidity. In Ontario, Canada, risk of asthma hospitalization associated with gaseous air pollutants (sulfur dioxide and nitrogen dioxide) was greater among individuals living in lower income areas (Lin et al. 2004). In New York City, individuals in racial minority groups had higher relative and attributable risks of ozone-associated respiratory hospital admissions, which were explained mainly by socioeconomic differences and health insurance status (Gwynn and Thurston 2001). Socioeconomic circumstances may affect exposure to air pollution, due to housing characteristics, proximity to roadways or industrial pollution sources, and/or job occupation. People of lower SEP may be more vulnerable to air pollution if health care access or affordability results in inadequate medication or treatment for conditions such as asthma.

Although an individual’s SEP can affect susceptibility and exposure, health responses to air pollution may also depend on socioeconomic characteristics of a community. Pollution concentrations or composition, housing characteristics, and/or health care accessibility and quality can vary in different communities. Socioeconomic variables measured at both the individual and group-level can therefore be used to examine hypotheses about how environmental exposures may interact with other determinants of health, and evaluate the relative importance of socioeconomic conditions at the community and individual level.

The previously cited studies on asthma and air pollution used individual or regional SEP indicators, but not both simultaneously. This study examined whether SEP, as marked by health insurance premium, measured at both the individual and regional levels, modified the association of air pollutants with emergency asthmatic outpatient hospital visits in Seoul, Korea, and whether the effect modification was altered when considering both SEP levels.

Methods

Air pollution data for particulate matter ≤10 microns in aerodynamic diameter (PM10), carbon monoxide (CO), sulfur dioxide (SO2) nitrogen dioxide (NO2), and ozone in 2002 was obtained from outdoor monitoring stations located in each of 25 administrative districts in Seoul, Korea. Measurements were available from only 1 monitoring site in 23 of the districts; the remaining 2 districts each had 2 monitoring sites and the average concentrations from the 2 stations were calculated to estimate exposure in those 2 districts. Figure 1 shows the 25 administrative districts called “Gu”. The average population of each district is about 400,000 persons and the average area is 24 km2.
https://static-content.springer.com/image/art%3A10.1007%2Fs00420-007-0182-3/MediaObjects/420_2007_182_Fig1_HTML.gif
Fig. 1

Twenty-five administrative districts (Gu), Seoul, Korea

We deleted the top 1% of values of daily levels of PM10 as they were extremely distributed (outliers). The effect of ozone was analyzed only for the summer season because the relation of ozone and health differed by season in Korea and temperature and ozone were highly correlated during summer (Kim et al. 2004). We used the daily maximum of 8 h means for CO, daily maximum for ozone, and daily means for the other pollutants. Meteorological variables included daily mean temperature, relative humidity, and air pressure measured at the meteorological monitoring center in Seoul during 2002.

Emergency outpatient hospital visits for asthma were extracted from Korean National Health Insurance (KNHI) records. The KNHI program has covered almost the entire Korean population since 1989 (Yu 1992) and all hospital visit records were compiled for 2002. We selected emergency visits for asthma attacks. We defined these as being emergency room visits hospital visits on holidays, weekends or at night, among outpatients with diagnosed asthma (International Classification of Disease, tenth revision, codes “J45” and “J46”). Among these selected visits, the subjects for this analysis were restricted to people with utilization of inpatient service for asthma more than once or outpatient medical service more than three times from 1999 to 2001. This restriction was made because of prior evidence that respiratory diseases showing bronchial symptoms may be misdiagnosed as asthma in KNHI records (Park et al. 2003).

Exposure was assigned as the air pollution concentration measured at the monitor(s) in the administrative district corresponding to the visited hospital. Since these were emergency visits, the assumption was that people would attend the hospital nearest to their residence.

Socioeconomic position was represented by premium contributed to the KNHI program. This single insurer program covers about 97% of the Korean population; the remaining 3% of the population are indigent or belong to low-income brackets and are covered by a medical benefit program administered locally (NHIC 2005). Premiums are paid according to two categories of the insured, the employed and the self-employed. Premiums of the employed are calculated on the basis of income while those of the self-employed are calculated based on the value of properties and motor vehicles, and participatory rate of economic activity, other than income. This socioeconomic information is furnished by state and local governments and other public organizations (Ministry of Health and Welfare 2005). We categorized patients into quintiles for their individual premiums based on premium statistics for all Koreans reported in December 2001, consistent with previous research (Kim et al. 2002). For dependents, including children and elderly family members, the premium of their corresponding insured was assigned. Quintile 1 indicates people who pay or are covered by the highest 20% of premium distribution, and Quintile 5 represents the lowest 20%. The 25 administrative districts in Seoul were also classified into 5 groups comprised of 5 districts each, according to the average premium of the self-employed in districts, and were classified as Region 1 to 5. We could not use premium of the employed to calculate regional SEP because of inaccuracies in residential address records of the employed insured.

To evaluate the association of air pollution and asthma hospital visits by SEP, we adopted the case-crossover design using conditional logistic regression. The case-crossover approach is a variant of the case-control design in which only cases are sampled. For this study, the case corresponds to an asthma hospital visit and the unit of analysis is the day on which it occurred. One or more control days are selected at fixed intervals before or after the day each case arose, and case-day exposure is compared with exposure on the control days when the event did not occur (Lee and Schwartz 1999; Basteson and Schwartz 2004; Maclure and Mittleman 2000). In this design, the same subject sampled as a case provides control information, analogous to pair matching in the case-control study. Therefore, control for individual characteristics that might confound the air pollution-asthma association is unnecessary.

We selected control days as the same weekday 1 week (1:2 pair-matching) or 1 and 2 weeks (1:4 pair-matching) before and after the day when the emergency visit for asthma occurred. This bidirectional control selection approach has been recommended to control confounding by seasonal patterns and time-trends in air pollutant concentrations (Lee et al. 2000). The pair-matching approach resulting in the higher effects of air pollution on asthma was chosen for each model. The 1:2 design was used for the four pollutants other than ozone. In the initial model, we investigated the relation of air pollution and asthma visits, adjusting for time trend, weather conditions, and seasonality. To control the seasonality, we modeled sine and cosine curves with six different periods from 365/1 to 365/6 and finally included significant terms selected by the stepwise method.

We evaluated exposures on the same day as the event day and four previous days, singly and in moving averages. We evaluated effect modification using the lag structures for each of the five pollutants that yielded the strongest air pollution and asthma associations. Lag 0 corresponds to the day the asthma attack occurred, lag 1 the previous day, and so forth. We report results for the following exposures: PM10 moving average of lags 0, 1, 2; CO moving average of lags 1, 2, 3; SO2 lag 3; NO2 moving average of lags 2, 3, 4; and for ozone, lag 2. One recent suggestion in the case-crossover study was to display the relevant exposure term, which is the absolute value of the difference between the ambient concentration on event day and control days (Kunzli and Schindler 2005). To allow readers to assess statistical power, we presented this relevant exposure term as well as daily concentration of air pollutants evaluated.

To evaluate whether the association of air pollution and asthma differed by SEP, we adopted three interaction models. The first interaction model added a term for quintile of individual insurance premium multiplied by air pollution concentration to the initial model. The second interaction model included the regional category for average insurance premium in districts multiplied by air pollution to the model. The third model included interaction terms for both the regional and individual level indicators. To evaluate the degree of effect modification, we calculated “relative effect modification” (Basteson and Schwartz 2004). Relative effect modification (REM) is defined as the ratio of the relative risk for a given level in individual or regional level SEP to the relative risk for the corresponding reference category, which for this study was the highest level for both regional and individual SEP. We reported relative risk for the initial model and relative effect modification for the single and joint interaction models by interquartile increase of each of the pollutants.

Results

Table 1 shows summary statistics for daily concentrations of the five air pollutants, weather variables, and emergency asthmatic outpatient hospital visits in Seoul, Korea, 2002. Daily mean concentrations of the five air pollutants in Seoul were 67.6 μg/m3, 0.9 ppm, 4.7 ppb, 36.0 ppb, and 39.9 ppb for PM10, CO, SO2, NO2, and ozone, respectively. Most daily concentration levels for the five air pollutants remained under the Korean national ambient air quality standard. The medians of the relevant exposure characteristics for all air pollutants were smaller than the interquartile range of daily concentration. In relative terms, evaluated by dividing the absolute difference by the concentration on event days for each pollutant, the medians for the differences in concentrations between the event days and the control days for PM10, CO, SO2, NO2, and ozone were less than 50% of the event day concentrations. The average number of daily outpatient hospital visits for asthma in Seoul during 2002 was 253.5. Demographic characteristics of patients visiting the hospital with asthma attack on an emergency basis in 2002 are observed in Table 2. The total number of asthmatic outpatient hospital visits was 92,535. The top quintile of premium distribution based on all the Koreans contained 22.0% of the emergency asthmatic outpatient visits, while the bottom quintile included 12.4%. According to the five categories based on regional premiums for 25 districts, the amounts of the premium were significantly different (Table 3). Average concentrations of air pollution were similar across the 25 districts, but the rates of asthma hospital visits were higher in the lower regional SEP districts.
Table 1

Daily air pollution, weather conditions, and emergency outpatient hospital visits for asthma in Seoul, Korea, 2002

 

N

Minimum

Median

Maximum

IQR

Mean

SD

Pollutanta

 Daily concentration

  PM10 (μg/m3)

365

4.9

61.9

302.0

47.4

67.6

39.0

  CO (0.1 ppm)

365

0.8

7.6

44.0

5.6

8.6

4.6

  SO2 (ppb)

365

0.9

4.1

21.2

3.3

4.7

2.7

  NO2 (ppb)

365

2.3

34.3

108.0

20.1

36.0

14.7

  Ozone (ppb)b

92

1.0

37.0

158.5

38.5

39.9

25.3

 Relevant exposure termc

  PM10 (μg/m3)

87,421

0.0

21.6

143.1

26.8

26.0

19.7

  CO (0.1 ppm)

89,356

0.0

2.0

30.4

2.8

2.8

2.8

  SO2 (ppb)

88,248

0.0

1.4

14.8

2.2

2.0

2.0

  NO2 (ppb)

88,929

0.0

8.3

60.4

10.7

10.1

7.8

  Ozone (ppb)

18,419

0.0

17.8

114.0

25.0

21.6

18.0

Weather

 Temperature (°C)

365

−9.7

14.0

30.3

16.8

12.9

9.6

  Relative humidity (%)

365

27.3

62.7

95.4

20.8

62.1

14.2

  Air pressure (hPascal)

365

990.7

1015.6

1035.9

11.7

1015.6

7.9

 Asthma hospital visits

  Hospital visits per day

365

19.0

231.0

804.0

120.0

253.5

114.8

  The rates of hospital visits d

365

1.9

22.6

78.6

11.7

24.8

11.2

aPM10 particulate matter, CO carbon monoxide, SO2 sulfur dioxide, NO2 nitrogen dioxide

b01/06/2002–31/08/2002 (summer)

cAbsolute value of difference between concentration on event day and mean of concentrations on control days

dHospital visits divided by population size and multiplied by 1,000,000

Table 2

Demographic characteristics for emergency outpatient hospital visits for asthma in Seoul, Korea, 2002

Variable

Populationa

Eventb

Rate

N

(%)

N

(%)

Gender

 Male

5,125,353

50.1

49,399

53.4

0.0096

 Female

5,109,963

49.9

43,136

46.6

0.0084

Age in years

 0–9

1,214,113

7.7

66,126

71.5

0.0545

 10–19

2,695,463

17.0

3,462

3.7

0.0013

 20–29

1,987,576

12.5

812

0.9

0.0004

 30–39

3,849,604

24.3

2,237

2.4

0.0006

 40–49

3,422,269

21.6

3,127

3.4

0.0009

 50–59

1,061,675

6.7

4,422

4.8

0.0042

 60–69

1,279,040

8.1

6,556

7.1

0.0051

 70–

348,764

2.2

5,793

6.3

0.0166

Premium

 Medical benefitc

178,394

1.7

1,844

2.0

0.0103

 Quintile 1d

2,498,000

24.4

20,369

22.0

0.0082

 Quintile 2

1,973,076

19.3

22,542

24.4

0.0114

 Quintile 3

1,870,392

18.3

19,432

21.0

0.0104

 Quintile 4

1,942,693

19.0

16,879

18.2

0.0087

 Quintile 5

1,756,524

17.2

11,469

12.4

0.0065

aMean population of December 2001 and December 2002 in Seoul for gender and age, and population on December 2001 for premium category of the Korean National Health Insurance records

bEmergency outpatient hospital visits for asthma in Seoul during 2002, with medical records of utilization of inpatient service for asthma more than once or outpatient service more than three times from 1999 to 2001

cPeople in low income households (under 3% of total population) provided with health care services funded by tax revenue sources as public assistance

dThe highest 20% of premium distribution

Table 3

Mean daily air pollution concentration, socioeconomic characteristics, and rates of daily emergency outpatient hospital visits for asthma in each district, Seoul, Korea, 2002

Regional level SEPa

District (Gu)

Populationb (person)

Premiumc ($)d

Asthma

Pollutantg

Visite

Ratef

PM10

CO

SO2

NO2

Ozoneh

Region 1

Gangnam-gu

537,889

23.9

4.4

4.1

78.9

8.9

5.4

39.7

42.6

Seocho-gu

391,553

23.3

4.7

6.0

60.5

6.7

5.1

37.0

45.6

Songpa-gu

651,260

17.2

7.5

5.8

63.7

6.4

4.7

30.9

44.0

Yongsan-gu

237,843

16.9

4.8

10.1

67.1

6.5

4.5

39.5

36.1

Jongno-gui

182,728

16.6

5.3

14.4

     

Meanj

2,001,273

19.6

25.2

6.3

66.1

7.1

4.9

36.7

42.0

Region 2

Jung-gu

140,575

16.5

5.1

18.3

87.1

11.9

4.7

30.4

45.6

Yeoungdeungpo-gu

407,807

16.2

8.7

10.7

65.3

8.4

7.4

40.2

36.4

Mapo-gu

377,279

16.1

14.7

19.5

69.0

8.9

5.8

40.0

44.7

Gwangjin-gu

387,310

15.9

6.6

8.5

77.7

9.5

4.9

34.3

34.5

Yangcheon-gu

484,432

15.6

6.1

6.3

66.4

9.4

3.5

35.4

42.1

Mean

1,797,402

16.1

39.8

11.1

73.3

9.7

5.2

35.9

40.7

Region 3

Dongjak-gu

404,582

15.6

13.1

16.2

51.5

9.4

4.5

34.5

40.1

Seongdong-gu

341,074

15.6

10.2

15.0

73.4

7.9

3.9

31.1

34.3

Seodaemun-gu

364,421

15.5

10.9

14.9

60.0

8.4

3.6

37.9

34.7

Gangdong-gu

485,356

15.2

8.1

8.4

64.6

8.0

4.0

39.1

44.2

Nowon-gu

644,279

15.0

37.6

29.1

69.3

8.8

4.5

31.8

33.6

Mean

2,239,711

15.4

78.4

17.5

63.7

8.5

4.1

34.9

37.4

Region 4

Dongdaemun-gu

381,820

15.0

7.4

9.7

66.2

11.0

5.4

38.8

39.0

Dobong-gu

371,585

14.8

8.7

11.7

65.5

5.8

5.1

32.6

44.9

Guro-gu

414,347

14.8

9.0

10.9

69.8

7.1

4.4

34.6

36.6

Seongbuk-gu

447,206

14.8

9.1

10.1

44.1

9.5

3.7

35.2

35.1

Gangseo-gu

524,905

14.8

13.9

13.3

66.9

9.4

3.2

36.7

38.2

Mean

2,139,862

14.8

47.4

11.1

65.0

8.5

4.3

35.5

38.8

Region 5

Gwanak-gu

524,986

14.4

19.3

18.4

72.8

9.0

5.7

44.7

42.1

Eunpyeong-gu

468,962

14.3

15.1

16.1

68.4

9.9

4.5

31.3

50.3

Geumcheon-gu

260,945

13.7

4.3

8.2

59.1

8.4

5.1

36.0

38.9

Gangbuk-gu

356,650

13.5

9.6

13.4

78.2

9.2

4.6

33.8

35.2

Jungnang-gu

445,527

13.3

15.6

17.5

68.4

9.2

5.3

43.5

37.7

Mean

2,057,069

13.8

62.9

15.3

69.3

9.2

5.0

37.5

40.8

Mean

 

10,235,316

16.0

253.5

12.4

67.6

8.6

4.7

36.0

30.8

aRegional level socioeconomic position defined by average Korean National Health Insurance premium of self-employed in 25 districts in Seoul, Region 1: five highest premium districts, Region 5: five lowest districts

bMean population size recorded on December 2001 and on December 2002

cAverage premium of self-employed and their dependents in the Korean National Health Insurance in 2002. The premium levels among five regions were significantly different in the Kruskal–Wallis test

dUS $1 was equivalent to 1,273 Korean won in 2002

eAverage number of daily emergency outpatient hospital visits for asthma in 2002

fAverage number of daily emergency outpatient hospital visits divided by population size and multiplied by 1,000,000

gAverage daily concentration of air pollution in 2002

h01/06/2002–31/08/2002 (summer)

iAll values of air pollution for 2002 is missing

jTotal number is shown for population size and asthmatic visit

Air pollution was positively associated with asthma hospital visits in Seoul, Korea during 2002 (Table 4). The relative risk (RR) of asthma hospital visit with a 47.4 μg/m3 increment of PM10 was 1.06 (95% confidence interval, CI = 1.04, 1.08) and the RR of a hospital visit with a 38.5 ppb increment of ozone was 1.11 (95% CI = 1.08, 1.14) in summer. Significant positive RRs were seen consistently across various exposure lags, although the pattern was less consistent for ozone (Fig. 2). Stratified by regional SEP, the significant risk disappeared in Region 1, the highest SEP region, in contrast to the other regions. In addition, the effect of air pollution on asthma tended to increase in the lower SEP regions for PM10 and ozone. The relative risk for PM10 was 1.04 (95% CI = 0.99, 1.10) in Region 1, increasing to 1.05 (95% CI = 1.03, 1.08) and 1.09 (95% CI = 1.06, 1.13) in Regions 3 and 5.
Table 4

Estimated relative risk (RR) and 95% confidence interval (CI) of emergency asthmatic outpatient hospital visits by interquartile increase of five air pollutants in initial models stratified by individual and regional level socioeconomic position in Seoul, Korea, 2002

Pollutant

PM10

CO

SO2

NO2

Ozonea

RRb

95% CIc

RR

95% CI

RR

95% CI

RR

95% CI

RR

95% CI

Individual level SEPd

 Quintile 1

1.06

1.02, 1.09

1.06

1.02, 1.09

1.03

1.01, 1.06

1.06

1.02, 1.10

1.15

1.09, 1.22

 Quintile 2

1.07

1.04, 1.10

1.05

1.02, 1.09

1.04

1.02, 1.07

1.06

1.02, 1.09

1.14

1.08, 1.21

 Quintile 3

1.06

1.03, 1.10

1.05

1.01, 1.08

1.01

0.99, 1.04

1.03

0.99, 1.06

1.09

1.02, 1.15

 Quintile 4

1.03

0.99, 1.07

1.07

1.03, 1.11

1.04

1.01, 1.07

1.06

1.02, 1.10

1.09

1.02, 1.16

 Quintile 5

1.10

1.05, 1.14

1.05

1.00, 1.09

1.01

0.98, 1.05

1.05

1.00, 1.10

1.08

1.00, 1.16

Regional level SEPe

 Region 1

1.04

0.99, 1.10

0.99

0.92, 1.07

0.98

0.94, 1.03

0.96

0.90, 1.02

1.04

0.95, 1.14

 Region 2

1.03

1.00, 1.07

1.06

1.02, 1.11

1.04

1.01, 1.07

1.08

1.04, 1.13

1.11

1.04, 1.19

 Region 3

1.05

1.03, 1.08

1.04

1.02, 1.07

1.02

1.00, 1.04

1.03

1.00, 1.07

1.12

1.07, 1.18

 Region 4

1.06

1.02, 1.10

1.10

1.06, 1.15

1.04

1.02, 1.07

1.06

1.02, 1.10

1.13

1.06, 1.20

 Region 5

1.09

1.06, 1.13

1.06

1.03, 1.09

1.03

1.00, 1.06

1.06

1.02, 1.09

1.13

1.07, 1.19

 Total

1.06

1.04, 1.08

1.06

1.04, 1.07

1.03

1.02, 1.04

1.05

1.03, 1.06

1.11

1.08, 1.14

Initial model: logit[P(asthma)] = air pollution + temperature + relative humidity + air pressure + seasonality

a01/06/2002–31/08/2002 (summer)

bRelative risk

c95% confidence interval

dIndividual level socioeconomic position was indicated by individual premium of employed and self-employed in the Korean National Health Insurance, Quintile 1: the highest 20%, Quintile 5: the lowest 20%

eRegional level socioeconomic position was indicated by average Korean National Health Insurance premium of self-employed in 25 districts in Seoul, Region 1: five highest premium districts, Region 5: five lowest districts

https://static-content.springer.com/image/art%3A10.1007%2Fs00420-007-0182-3/MediaObjects/420_2007_182_Fig2_HTML.gif
Fig. 2

Relative risk (RR) and 95% confidence interval (CI) for the relation of ambient air pollution and emergency outpatient hospital visits for asthma according to various lags, Seoul, Korea, 2002

When interaction terms for both individual and regional level SEP were included (Table 5), relative effect modification values showed that regional level SEP modified the association of SO2 and NO2 with asthma hospital visits, while no significant modifications by individual level SEP were found. Lower levels of regional SEP significantly increased the risks of asthmatic hospital visit for SO2 and NO2. Although not statistically significant, the same trends were found for other air pollutants. This trend was similar across various lags (Fig. 3).
Table 5

Relative effect modification (REM) and 95% confidence interval (CI) of individual and regional level socioeconomic position in the relation of five ambient air pollutants with emergency asthmatic outpatient hospital visits, Seoul, Korea, 2002

Pollutant

PM10

CO

SO2

NO2

Ozonea

REM

95% CIb

REM

95% CI

REM

95% CI

REM

95% CI

REM

95% CI

Individual level (SEP)c

 Quintile 1d

1.00

 

1.00

 

1.00

 

1.00

 

1.00

 

 Quintile 2

1.01

0.97, 1.05

1.00

0.95, 1.04

1.00

0.97, 1.04

0.99

0.95, 1.04

0.98

0.90, 1.05

 Quintile 3

1.00

0.95, 1.04

0.99

0.94, 1.03

0.97

0.94, 1.01

0.96

0.92, 1.01

0.94

0.87, 1.01

 Quintile 4

0.98

0.94, 1.03

1.02

0.97, 1.06

1.00

0.97, 1.04

1.00

0.95, 1.05

0.94

0.87, 1.02

 Quintile 5

1.03

0.98, 1.09

0.99

0.94, 1.04

0.98

0.94, 1.02

0.99

0.93, 1.05

0.92

0.84, 1.00

Regional level (SEP)e

 Region 1f

1.00

 

1.00

 

1.00

 

1.00

 

1.00

 

 Region 2

0.99

0.93, 1.06

1.05

0.97, 1.14

1.05

1.00, 1.11

1.11

1.03, 1.20

1.06

0.96, 1.18

 Region 3

1.01

0.96, 1.07

1.03

0.96, 1.11

1.04

0.99, 1.09

1.07

1.00, 1.15

1.09

0.98, 1.20

 Region 4

1.01

0.95, 1.08

1.08

1.00, 1.16

1.05

1.00, 1.11

1.09

1.02, 1.17

1.06

0.96, 1.18

 Region 5

1.05

0.99, 1.11

1.05

0.97, 1.13

1.05

0.99, 1.10

1.09

1.02, 1.16

1.07

0.97, 1.19

Third interaction model: logit[P(asthma)] = air pollution + air pollution × individual SEP + air pollution × regional SEP + temperature + relative humidity + air pressure + seasonality

REM Relative effect modification is the ratio of the relative risk for a given category in individual or regional level SEP to the relative risk for the corresponding reference category (highest SEP level)

a01/06/2002–31/08/2002 (summer)

b95% confidence interval

cIndividual level socioeconomic position was indicated by individual premium of employed and self-employed in the Korean National Health Insurance Quintile 1: the highest 20%, Quintile 5: the lowest 20%

dReference group is the highest 20% of premium distribution

eRegional level socioeconomic position was indicated by average Korean National Health premium of self-employed in 25 districts in Seoul, Region 1: five highest premium districts, Region 5: five lowest districts

fReference group is five highest premium districts

https://static-content.springer.com/image/art%3A10.1007%2Fs00420-007-0182-3/MediaObjects/420_2007_182_Fig3_HTML.gif
Fig. 3

Relative effect modification (REM) and 95% confidence interval (CI) of regional level socioeconomic position (open circle highest, filled triangle high, rectangle middle, filled circle low, filled square lowest) according to various lags (lag 0 exposure on the same day as the event day, lag1 exposure on the first previous day, lag012 moving average of exposure from the same day through two days prior) in the joint interaction model which includes interaction terms for individual and regional level socioeconomic positions simultaneously with main effect of ambient air pollution on emergency outpatient hospital visits for asthma, Seoul, Korea, 2002

Discussion

Our findings showed that increased levels of air pollution significantly elevated the risk of hospital visits due to asthma among Seoul residents, consistent with previous studies based in various countries (Lipsett et al. 1997; Sunyer et al. 1997). Residence in more deprived regions in Seoul was associated with a stronger adverse effect of air pollution on emergency hospital visits for asthma, while no significant modifying effect was found for individual level SEP. Although regional level SEP modified the air pollution effect significantly only for SO2 and NO2, the same pattern of higher risks in lower SEP districts was also found for the other three air pollutants. These findings are consistent with a previous study that reported residence in deprived areas was associated with adverse effects of NO2 and SO2 on asthma hospitalization among females and males, respectively, in Vancouver, Canada (Lin et al. 2004). However, our study took individual SEP into simultaneous consideration and we were able to explore the gradient and the significance of relative effect modification of individual and regional SEP with a large sample size and case-crossover design.

Asthma prevalence has increased worldwide for the last 20 years (National Heart, Lung, and blood Institute 2004), and a burgeoning number of studies have consistently reported the acute effect of air pollution on asthma attacks in metropolitan areas all over the world including Korea (Lipsett et al. 1997; Sunyer et al. 1997; Lee et al. 2002). Although studies have shown that the adverse effect of air pollution on asthma can be modified by individual (Nauenberg and Basu 1999) or regional level SEP (Lin et al. 2004), there have been very few attempts to explore the effect of both levels of SEP simultaneously. Given the growing interest on the independent roles of regional or neighborhood characteristics, often represented by regional SEP, on individual health (Diez Roux 2001; Macintyre et al. 2002), it is necessary to consider both individual and regional level SEP simultaneously when studying the adverse effect of air pollution on asthma. In Seoul, residential clustering by SEP has taken place since the beginning of the regional autonomous governing system in 1995 (Yoon 1998), making this city an ideal setting for examining how SEP may modify air pollution effects on asthma. Further, asthma attacks are most prevalent among the young and elderly populations (see Table 2), and the daily lives (and hence air pollution exposure) of these populations in Seoul usually take place within the regions they reside, since their mobility across the city is limited.

In general, SEP is known to modify the relationship between air pollution and asthma by creating different levels of exposure and susceptibility (O’Neill et al. 2003). Deprived people are more likely to live and work in polluted areas and to be exposed to air pollution than are their affluent counterparts. It has been reported that individual exposure to PM10 and gaseous pollutants differ by individual income, occupation, and education (Rotko et al. 2000; Rotko et al. 2001; Korc 1996). Further, financially disadvantaged persons are known to utilize medical services less frequently than are needed, even in the case of asthma. Asthma is a chronic condition for which symptoms can be considerably alleviated if cared for regularly and appropriately (Kattan et al. 1997). Active cigarette smoking aggravated asthma symptoms (Thomson et al. 2004), and people residing in deprived areas tend to smoke more than those living in affluent areas (Reijneveld 1998). Individuals with low income have limited access to grocery stores where fresh vegetables and fruits including anti-oxidant vitamins that can help protect against harmful effects of air pollution exposure are available (Morland et al. 2002).

No interaction effect was found for individual SEP and air pollution, whereas the effect of pollution was modified by the residential regions of individuals in Seoul. Although increased exposure is one of the pathways by which lower SEP may modify the air pollution and health association, we do not believe that higher levels of air pollution in less affluent regions created the interaction effect of regional SEP in Seoul. Average concentrations of air pollution were similar across the five SEP regions (see Table 3). Further, while traffic-induced pollution is a main source of air pollution in Seoul (Park et al. 2005) and significantly aggravated lung function (Hong et al. 2005), the amount of traffic is even heavier in the most affluent district (Park et al. 2002). Other mechanisms in Seoul, including access to medical care or access to fresh produce may be generating the observed pattern by regional SEP of the effect of air pollution on asthma hospital visits. Future studies in this field should pay attention to identifying these and other mechanisms, which will be useful for governments to prepare environmental health policies in improving adverse effects of air pollution on asthma in Korea.

Our study has at least the following two limitations. First, the insurance premium was employed as a single proxy for SEP, although SEP could be measured in various ways. This is because the insurance premium is the only available index of individuals’ socioeconomic characteristics in the KNHI data set. However, the KNHI is the only insurance program that universally covers the Korean population, so we could identify individual SEP for all subjects in our study using the KNHI records. A previous study also used the premium as the surrogate of income (Kim et al. 2005). Although the insurance premiums are determined by one’s salary or expected income, and their usefulness has been reported, other indices would be helpful to achieve a better measure of SEP. Second, we employed the administrative boundary for measuring regions in Seoul. Air pollution concentrations measured at a few sites in one administrative area may not represent the exposure for the entire area and every person residing in that area. To reduce the error in future studies, a geographic information system (GIS) could better measure quantitative and/or qualitative aspects of regions regarding air pollution and the monitoring stations (Martins et al. 2004). In terms of error caused by the difference from personal exposure, a recent study showed that outdoor exposure was significantly associated with exhaled nitric oxide in asthmatic children while indoor exposure was not, and concluded that ambient pollution measured in outdoor monitoring stations would be a reasonable surrogate of personal exposure (Koenig et al. 2005). Moreover, misclassification of exposure is not likely to be associated with asthma morbidity and this non-differential misclassification would therefore be expected to result in an underestimate of true effects in this study.

The present study represents one of the very few attempts to simultaneously consider both individual and regional levels of SEP with respect to the relationship between air pollution and asthma hospital visits. Findings from this study should enhance the discourse on how regional socioeconomic and other characteristics may result in increased vulnerability to air pollution exposure, and inform efforts to take comprehensive prevention actions.

Acknowledgments

This study was supported by the grant from the Basic Research Program of Korea Science and Engineering Foundation (no. R01-2002-000-00554-0), the Korean Research Foundation Grant funded by the Korean Government (KRF-2005-214-E00041), and the Robert Wood Johnson Health and Society Scholars Program.

Copyright information

© Springer-Verlag 2007