Human biomonitoring for chemical exposures has generated large amounts of data. Analysis of those data presents a challenging problem to epidemiologists and biostatisticians. One prominent characteristic of these environmental data is that exposures are always mixtures of chemicals and the chemicals in a mixture are often moderately or highly correlated. The adverse effect of an individual chemical on any health outcome is usually small due to the low exposure level. However, effects of exposure to chemicals in mixtures can accumulate and act synergistically on health outcomes. The overarching goal of this project is to develop better statistical methods for understanding the detrimental health impacts of exposure to mixtures of chemicals. To accomplish this goal, we propose improvements over the existing genome-wide complex trait analysis approach so that the accumulative effects and the total interaction effects of exposure to chemical mixtures can be estimated with minimal bias. We further propose to estimate the individual chemical effects as the average causal effect through the propensity score adjustment. The estimates will serve as the basis for toxicity assessment of chemicals. Lastly, we propose a flexible network analysis approach to understand the potential causal pathways from exposure to mixtures to health outcomes. The methods will be applied to a number of datasets on which the research team has been working to answer important scientific questions with regards to the associations of persistent organic pollutant exposures with endocrine and cardio-metabolic outcomes, and biological pathways and nutrients relevant to these associations. The datasets also serve as testing formats for developing and using the software package implementing the proposed methods. The software package will be made freely available to environmental research community. The results of this project are expected to substantially improve our ability to understand complex relationships among the many chemical exposures found in human populations and detrimental health outcomes. Our development of innovative methods will potentially facilitate the investigation of biological pathways mediating these relationships and enhance our understanding of nutritional and other factors that may in part ameliorate adverse effects of toxicants.
Populations are exposed to many different chemicals, or mixtures, that may affect health. This project will use existing data from epidemiological studies to develop new statistical approaches to study the role of exposure to multiple chemicals on health outcomes. The results should improve our ability to understand complex relationships among the many chemical exposures found in human populations and detrimental health outcomes.
