Assessment of Hypothetical Nitrogen Dioxide and Particulate Matter Concentration Interventions on Asthma Incidence in Southern Californian Children: A Causal Inference Approach

Abstract

Numerous observational studies have linked air pollution with incident childhood asthma. We implemented g-computation, a method based in the counterfactual framework, to estimate the effect on asthma of hypothetical air pollution interventions in a population of Southern California children. We used data from the three cohorts in the Southern California Children’s Health Study (recruited in 1993, 1996, and 2002) with prospectively identified incident asthma cases and continuously monitored nitrogen dioxide (NO2) and particulate matter ≤2.5 µm (PM2.5). Multilevel Poisson regression models were developed to estimate the relation between asthma incidence and baseline year annual average pollution concentration, adjusted for potential confounders. For both NO2 and PM2.5 we estimated the effect on asthma of 1) remaining at 1993 concentrations—the highest of all three time points—and 2) several hypothetical air quality standards—dynamic interventions where air pollution were reduced to a standard only in communities exceeding the hypothetical standard. All comparisons are to the natural course. The three cohorts included 4,140 children with no history of asthma at baseline and an average of 5.9 years of follow-up (mean baseline age: 9.5 years; 53% female; 55% White; 41% Hispanic). Remaining at 1993 NO2 levels was estimated to increase asthma incidence by 19.4% (95% CI: 9-32%), whereas implementing hypothetical standards of 30, 20, and 10 ppb were estimated to reduce asthma incidence by 7.2%, 19.5%, and 39.0% respectively. For PM2.5, there was an increase of 9.7% (0.8-20%) estimated for remaining at 1993 levels and reductions of 10.8%, 14.7%, and 17.5% for implementing hypothetical standards of 15, 12, and 10 μg/m3, respectively. Had NO2 and PM2.5 not improved over the study period we expect there would have been more childhood asthma cases; however, had air pollution levels been lower than observed there would have been even lower asthma incidence.