Combining Environmental Information. Ii: Environmental Epidemiology and Toxicology

Abstract

SUMMARYAn increasingly important concern in epidemiological and toxicological studies of environmental exposures is the need to combine information from diverse sources that relate to a common endpoint. This is clearly a statistical activity, but statistical techniques for data combination are still only developmental. Herein, we illustrate some current applications of combining information in environmental epidemiology and toxicology, with emphasis on the burgeoning use of meta-analyses for environmental settings. Our goal is to inform readers about modern statistical techniques useful for combining environmental information, with emphasis on more recently developed approaches.lit^ WORDS

From Quantal Counts to Mechanisms and Systems: The Past, Present, and Future of Biometrics in Environmental Toxicology

“…In the latter case, e.g., progress in using Monte Carlo methods such as Gibbs sampling (Robert and Casella, 1999) has opened up broad applications of environmental study. A promising area of future development using hierarchical modeling involves more efficient and practical methods for combining historical or other environmental information in meta-analyses of toxicological data sets (cf., Piegorsch and Cox, 1996;Givens, Smith, and Tweedie, 1997;Ibrahim, Ryan, and Chen, 1998). In these and many other application areas, advances in 21st century biometry will enhance the science of environmental toxicology, and the findings should prove exciting.…”

Benchmark Dose Estimation Incorporating Multiple Data Sources

“…Other work estimating the BMD using mixed models has been described in the environmental literature. Piegorsch and Cox ( 5 ) reviewed generalized linear models with random effects in the estimation of the effective dose in ordinal and continuous regression. Simmons et al .…”

“…Other work estimating the BMD using mixed models has been described in the environmental literature. Piegorsch and Cox (5) reviewed generalized linear models with random effects in the estimation of the effective dose in ordinal and continuous regression. Simmons et al (6) looked at modeling mutagenic potencies from different Salmonella strains, and Coull et al (7) conducted a meta-analysis for methyl-mercury using a Bayesian hierarchical model to combine BMD point estimates from multiple studies.…”

Environmental justice and statistical summaries of differences in exposure distributions

“…values exposure, disease, and demographics over the same geographic area. Environmental health studies often require the linking of such data from many sources and of Ž many types Cox and Piegorsch, 1996;Piegorsch and Cox, . 1996 .…”