NEHA April 2023 Journal of Environmental Health

ADVANCEMENT OF THE SCIENCE

We ran separate models for each pollut- ant variable of interest (PM concentrations, NO 2 , and O 3 ) with various exposure peri- ods (24-hr, 48-hr, 72-hr, and 96-hr means). Meteorological variables, such as temperature and relative humidity, were averaged over the same periods. We included exposure windows from 24-hr up to 96-hr averages of pollution before the physical activity measurements, as an e‚ect of air pollutants on physical activity might require more exposure time to manifest a change in time spent in physical activity. We controlled for temperature and relative humid- ity because 96-hr means of temperature and relative humidity showed the strongest asso- ciations with the measured outcomes. Also, we considered a model using the maximum 8-hr average concentration of ozone dur- ing each exposure interval, as the 8-hr mean aligns with the safe exposure limit for human health established by agencies such as the U.S. Environmental Protection Agency. E‚ect esti- mates for each measurement are presented as the percent change in time spent performing physical activity per increase in pollutant con- centrations. We considered a p -value of <.05 as statistically significant. Results The air pollutant concentrations we measured had a considerable range and are listed in Table 1. We examined 24-hr, 48-hr, 72-hr, and 96-hr means measured at the school; these values were also compared to the 96-hr mean concentrations from the CAMS. Concentra- tions at the CAMS monitoring site were lower and standard deviations were higher compared with the school measurements. The participants were 8.3 ± 1.5 years of age with a body mass index (BMI) of 17.9 ± 5.0 kg/m 2 (Table 2). The BMI-for-age percen- tile was 49.8 ± 41.2%. The physical activity levels for MVPA, light, and sedentary activ- ity were 63.4 ± 8.2%, 10.1 ± 1.7%, and 26.5 ± 7.9% of the time, respectively. A pairwise t -test indicated the three activity levels were significantly di‚erent from each other (all p <.001 with Bonferroni adjustment). The participant-specific factors including medication information are characterized in Table 3. We compared percent time spent in MVPA and sedentary activities by their factor levels using the Kruskal–Wallis test to exam- ine if the mean proportions between factor levels were statistically di‚erent. The test

TABLE 1

Air Pollutants Measured at the School and by Continuous Ambient Monitoring Stations (CAMS)

Air Pollutant PM 2.5 (µg/m

24-hr Mean (School)

48-hr Mean (School)

72-hr Mean (School)

96-hr Mean (School)

96-hr Mean (CAMS)

3 )

Mean

12.5

11.7

11.5

12.2

10.2

SD

3.7

2.4

1.9

2.8

5.3 9.8 5.2

Median

13.2

11.1

11.4

11.3

IQR

4.9

4.1

3.1

4.1

Maximum Minimum

18.9

15.7

14.3

17.6

18.7

6.3

9.0

8.6

8.6

3.4

PM 10 (µg/m

3 )

Mean

45.3 17.4 40.3 24.6 74.1 24.5

43.1 12.5 38.5 19.1 62.3 25.9

42.6

44.9

36.9 12.4 38.7 16.8 51.6 13.8

SD

8.7

9.1

Median

40.3 11.9 57.0 31.4

45.8

IQR

9.6

Maximum Minimum

60.1 28.5

NO 2 (ppb) Mean

17.6

18.2

18.4

18.9

17.9

SD

6.1

3.3

3.1

3.7

5.1

Median

19.2

18.6

18.5

19.0

16.3

IQR

7.8

4.8

2.8

5.0

5.2

Maximum Minimum

26.2

22.2 12.2

22.7 12.2

23.6 11.6

27.1 13.0

7.2

O 3 (ppb)

Mean

21.4 10.5 19.6 18.1 38.9

20.4

21.8

20.4

19.9

SD

6.7

7.3

5.5

5.1

Median

18.9 11.7 31.1 12.5

19.4 12.3 34.5 13.9

18.3

18.9

IQR

8.6

7.5

Maximum Minimum

29.7 15.6

28.4 14.8

9.2

Note. IQR = interquartile range; NO 2 = nitrogen dioxide; O 3 = ozone.

results showed significantly di‚erent propor- tions for some factors (gender, BMI category, a father with asthma, siblings with asthma, having eczema, health insurance status, smoking status) and medications (leukot- riene blockers, long-acting bronchodilators and inhaled corticosteroids, and nasal corti- costeroids) with both MVPA and sedentary activities (see bolded p -values in Table 3). For example, type of insurance (i.e., Med- icaid versus private) was a significant factor

( p = .003): participants with Medicaid spent more time in MVPA (66.5%) than did those with private insurance (61.2%). Conversely, participants with Medicaid spent less time in sedentary activities (23.9%) than did those with private insurance (27.9%, p = .04).

Models Predicting Physical Activity Data

Table 4 presents e‚ect estimates using GEE models, 95% confidence intervals (CIs), and

18

$3091*  • 91'*6 

Powered by