Modeling and analysis of personal exposures to VOC mixtures using copulas.

摘要: Environmental exposures typically involve mixtures of pollutants, which must be understood to evaluate cumulative risks, that is, the likelihood adverse health effects arising from two or more chemicals. This study uses several powerful techniques characterize dependency structures mixture components in personal exposure measurements volatile organic compounds (VOCs) with aims advancing understanding environmental mixtures, improving ability model a statistically valid manner, and demonstrating broadly applicable techniques. We first describe characteristics introduce terms, including fraction represents component's share total concentration mixture. Next, using VOC data collected Relationship Indoor Outdoor Personal Air (RIOPA) study, are identified positive matrix factorization (PMF) by toxicological mode action. Dependency examined fractions modeled copulas, address dependencies multiple variables across entire distribution. Five candidate copulas (Gaussian, t, Gumbel, Clayton, Frank) evaluated, performance fitted models was evaluated simulation fractions. Cumulative cancer risks calculated for results multivariate lognormal compared observed data. Results obtained RIOPA dataset showed four representing gasoline vapor, vehicle exhaust, chlorinated solvents disinfection by-products, cleaning products odorants. Often, single compound dominated mixture, however, were generally heterogeneous composition changed concentration. Three action, VOCs associated hematopoietic, liver renal tumors. Estimated lifetime exceeded 10(-3) about 10% participants. Factors affecting high included city, participant ethnicity, house air exchange rates. The Gumbel (two mixtures) t (four types emphasize tail dependencies. Significantly, reproduced both risk predictions degree accuracy, performed better than distributions. Copulas may method choice particularly highest extreme events, cases poorly fit distributions represent greatest risks.