Air pollution and age-dependent changes in emotional behavior across early adolescence in the U.S

Abstract

Recent studies have linked air pollution to increased risk for behavioral problems during development, albeit with inconsistent findings. Additional longitudinal studies are needed that consider how emotional behaviors may be affected when exposure coincides with the transition to adolescence - a vulnerable time for developing mental health difficulties. This study investigates if annual average PM_2.5 and NO₂ exposure at ages 9-10 years moderates age-related changes in internalizing and externalizing behaviors over a 2-year follow-up period in a large, nationwide U.S. sample of participants from the Adolescent Brain Cognitive Development (ABCD) Study®. Air pollution exposure was estimated based on the residential address of each participant using an ensemble-based modeling approach. Caregivers answered questions from the Child Behavior Checklist (CBCL) at the baseline, 1-year follow-up, and 2-year follow-up visits, for a total of 3 waves of data; from the CBCL we obtained scores on internalizing and externalizing problems plus 5 syndrome scales (anxious/depressed, withdrawn/depressed, rule-breaking behavior, aggressive behavior, and attention problems). Zero-inflated negative binomial models were used to examine both the main effect of age as well as the interaction of age with each pollutant on behavior while adjusting for various socioeconomic and demographic characteristics. Against our hypothesis, there was no evidence that greater air pollution exposure was related to more behavioral problems with age over time.

Keywords: Adolescence; Air pollution; Externalizing; Internalizing; Neurodevelopment.

Figure 1

Results for internalizing behavior. A) Displays the estimated probability of being in the absolute zero category as compared to the non-zero category (i.e., any value for CBCL scores). B) Displays the estimated CBCL score for subjects whose scores were in the non-zero category. Numerous results are presented which include: 1) Age only which displays the main effects of age excluding air pollution with all other variables held constant from 9 to 12 years-of-age; 2) PM_2.5-by-age interaction which displays differences in 9 and 12 years-of-age for the WHO recommended PM_2.5 levels - 5 μg/m³ (light blue) - versus the EPA’s - 12 μg/m³ (dark blue); 3) NO₂-by-age interaction which displays differences in 9 and 12 years-of-age for the WHO recommended NO₂ levels - 5.33 ppb (light purple) - versus the 90^th percentile NO₂ level in our sample - 26.1 ppb (dark purple) (The EPA level is 53 ppb which is outside our sample range). All graphs display percent change with age. All covariates held constant at the largest N category (sex = “male”, race/ethnicity = ‘White’, caregiver education = ‘Post Graduate Degree’, caregiver employment = “Employed”, and household income = “≥$100K”), and mean for neighborhood safety ($\overline{x}=3.88$); for interaction models, NO₂ is set to the WHO standard (5.33 ppb) for the PM_2.5-by-age models and PM_2.5 is set to the WHO standard (5 μg/m³) for the Age-only and NO₂-by-age models; p-value_FDR = p-value for predictor graphed once FDR corrected for multiple comparisons; N.S. = not significant.

Figure 2

Results for externalizing behavior. A) Displays the estimated probability of being in the absolute zero category as compared to the non-zero category (i.e., any value for CBCL scores). B) Displays the estimated CBCL score for subjects whose scores were in the nonzero category. Numerous results are presented which include: 1) Age only which displays the main effects of age excluding air pollution with all other variables held constant from 9 to 12 years-of-age; 2) PM_2.5 main effect which displays the percent change from the WHO recommended PM_2.5 levels - 5 μg/m³ - versus the EPA’s - 12 μg/m³ with age held constant at 9 years-of-age; 3) NO₂-by-age interaction which displays differences in 9 and 12 years-of-age for the WHO recommended NO₂ levels - 5.33 ppb (light purple) - versus the 90^th percentile NO₂ level in our sample - 26.1 ppb (dark purple) (The EPA level is 53 ppb which is outside our sample range). All graphs display percent change with age. All covariates held constant at the largest N category (sex = “male”, race/ethnicity = ‘White’, caregiver education = ‘Post Graduate Degree’, caregiver employment = “Employed”, and household income = “≥$100K”), and mean for neighborhood safety ( = 3.88); for interaction models, NO₂ is set to the WHO standard (5.33 ppb) for the PM_2.5-by-age models and PM_2.5 is set to the WHO standard (5 μg/m³) for the Age-only and NO₂-by-age models; p-value_FDR = p-value for predictor graphed after FDR corrected for multiple comparisons; N.S. = not significant.