Incorporating Time Spent at Schools into Estimates of Traffic-Related Air Pollution Exposure

摘要

Background: Children spend a substantial amount of time at school, yet most epidemiologic studies investigating the effects of traffic related air pollution (TRAP) characterize exposure based solely on the participants' primary home. The objective of this analysis was to examine whether estimated TRAP exposures at school and estimates that integrate home and school are significantly different than exposures based on the home address alone. Methods: Complete address histories, including school locations, were collected for participants in the Cincinnati Childhood Allergy and Air Pollution Study, a longitudinal birth cohort. TRAP exposure was estimated at ages 4 and 7 for each participant's home and school using a validated land use regression model. Time-weighted average TRAP exposure was derived based on the proportion of hours spent at each location. Estimates of TRAP were log-transformed, and paired t-tests were used to evaluate the differences between exposures at home and school, as well as estimates at the home and time-weighted averages. Results: A smaller proportion of children reported attending school at age 4 (152, 22%) compared to age 7 (556, 91%), and children spent a correspondingly lower percentage of time at home per week (median: 93% vs 76% at ages 4 and 7, respectively). Estimated TRAP was, on average 6% higher at schools compared to homes at age 4 (p = 0.02), but there was no significant difference at age 7. Time-weighted estimates were significantly different from estimates based solely on the home at ages 4 (p = 0.03) and 7 (p < 0.01). However, time-weighted estimates were only 1% greater on average than exposures derived at the home at both ages. Conclusion: Incorporating school locations into time-weighted estimates of TRAP exposure results in small increases in estimated exposure, likely due to the greater proportion of time spent at the home. Future studies will investigate how geographic or socioeconomic characteristics influence these results.