Exposure to air pollution is an established risk factor for asthma, reduced lung function as well as inflammatory and oxidative processes which are in turn linked with obesity and diabetes. Risk for these adverse outcomes begins early in childhood and well-documented racial/ethnic differences and social disparities render minority children especially vulnerable, yet these populations are underrepresented in the literature. Analysis for truly representative effect estimates should consider a wide array of genetic, social and environmental factors, biologic pathways from exposure to disease, with consideration for causal mediators and interactions. Mediation and interaction analysis can also aid in determining the transportability of estimated effects from one population to another. However, estimation and interpretability of target parameters in mediation analysis is complicated by issues such as mediator-outcome interactions, non-linearities, and exposure-induced confounding, which cannot be addressed using traditional regression approaches. My long-term career goals are to assess optimal interventions on environmental health risk factors that best reduce overall risk in populations of interest. I will utilize advanced epidemiologic methods maximinzing internal validity and efficiency of estimation of target. I will expand on existing methods for estimation of effects in the presence of time varying exposures and covariates as well as exposure-induced mediator outcome confounding (estimation of controlled direct effects and the randomized intervention analogues for the natural direct and indirect effects) which cannot be addressed with traditional regression approaches. The proposed methodology will be suitable for assessment of potential interventions on continuous exposures, which will be especially beneficial in the area of environmental epidemiology.
In Aims 1 & 2 of this proposal I will use these proposed methods to assess direct, indirect and total effects of air pollution exposures on risk of asthma, overall lung function and metabolic syndrome, within a counterfactual framework. I am well suited to perform the proposed research based on 1) my past experience in environmental health and advanced methods and counterfactual approaches 2) the exceprional interdisciplinary mentoring team I have assembled and 3) the unique research opportunity offered by the datasets in the proposal, comprised largely of minority children. This proposed study will enable me to quantify mediated effects of air pollution exposures in especially vulnerable populations. I will be advised by a world-class team of mentors to expand my expertise in integrating advanced epidemiologic methods with causal inference applications (e.g., machine learning and efficient estimators of causal inference parameters) in environmental epidemiological studies; epigenetic and exposomic factors as potential modifiers or mediators of effect; and health disparities and social factors associated with environmental exposures. The proposed research and training will enable me to establish an independent career as a leader in intervention assessment and causal inference in environmental epidemiology.
There are multiple pathophysiologic pathways through which air pollution exposures exert their effects on health, while racial/ethnic differences exist in terms of prevalence of health outcomes and genetic and social factors linked to levels of exposure and that potentially modify exposure effects. Mediation and interaction analyses are important to help as answer questions about how exposures exert their effects and who may be particularly susceptible, but limitations exist in estimation and interpretability. I propose mediation analysis in a counterfactual framework using data adaptive estimators to address complex epidemiologic issues and quantify direct, indirect and total effects of air pollution health effects on asthma, overall lung function and metabolic syndrome parameters in especially vulnerable minority children populations.