This proposal focuses on the development of observational study designs and statistical analysis methods to capitalize on group formation in the study of health effects. We focus on estimation of the effect of exposures on health outcomes with particular emphasis environmental exposures. There are two specific aims. The first is to develop and evaluate statistical methods for grouping in model predictors.
This aim i s has two components. One is motivated by occupational epidemiologic studies where individual exposures cannot be accurately measured. The methods development focuses on a regression calibration approach to exposure estimation that combines individual and group information. The goal is to minimize bias and maximize the precision of the exposure effect (or relative risk) parameter estimate in an exposure-response model. The role of the grouping process in the estimation procedure will be evaluated. The common practice of substituting ambient pollution exposure measures for individual exposures in air pollution studies motivates the second component. We will evaluate the role of using proxy group exposure measures in place of individual exposure measures in models of the health effects of air pollution where the goal is estimation of exposure effect parameters.
The second aim i s to research methods for the design and analysis of grouped data structures as motivated by multipopulation studies. The first subaim concentrates on aggregate data studies based on disease rates and risk factor survey data. Properties of the aggregate data study relative risk parameter will be evaluated given varying confounding factor, covariate of the nuisance parameters, to more flexibly accommodate varying data structures, and to accommodate spatial dependence. in the second subaim, study of the properties of individual studies with grouped data components will be motivated by multipopulation cohort studies designed to study of health effects of air pollution. We will address the role of between group information in such studies and our ability to differentiate between chronic and acute effects. Methods used in this proposal include clear problem definition, asymptotic statistical theory, and Monte Carlo simulation. three datasets are described to motivate and apply the proposed methods, other datasets will be utilized as they become available over the funding period. The methods are relevant to a wide range of environmental health applications, including occupational studies, air pollution research, and the assessment of health risks due to other environmental exposures exhibiting between- group variation and exposure measurement error.
| Sheppard, Lianne (2005) Acute air pollution effects: consequences of exposure distribution and measurements. J Toxicol Environ Health A 68:1127-35 |
| Sheppard, Lianne; Slaughter, James C; Schildcrout, Jonathan et al. (2005) Exposure and measurement contributions to estimates of acute air pollution effects. J Expo Anal Environ Epidemiol 15:366-76 |
| Hoppin, Jane A; Ulmer, Ross; London, Stephanie J (2004) Phthalate exposure and pulmonary function. Environ Health Perspect 112:571-4 |
| Sheppard, Lianne (2003) Insights on bias and information in group-level studies. Biostatistics 4:265-78 |
| Guthrie, Katherine A; Sheppard, Lianne; Wakefield, Jon (2002) A hierarchical aggregate data model with spatially correlated disease rates. Biometrics 58:898-905 |
| Akhmedkhanov, Arslan; Revich, Boris; Adibi, Jennifer J et al. (2002) Characterization of dioxin exposure in residents of Chapaevsk, Russia. J Expo Anal Environ Epidemiol 12:409-17 |
| Sheppard, L; Levy, D; Checkoway, H (2001) Correcting for the effects of location and atmospheric conditions on air pollution exposures in a case-crossover study. J Expo Anal Environ Epidemiol 11:86-96 |
