Environmental Statistics
For four decades, we have provided support to government and industry clients regarding statistical analyses to address their environmental concerns. Our environmental epidemiologists and statisticians work with health and social scientists, economists, toxicologists, biologists, ecologists, engineers and environmental scientists to provide advice on environmental issues such as species abundance, pollutant plumes, and human exposure and risk assessment.
Our work results in developing methodologies, technical cooperation, data collection, and implementation of environmental, spatial, and spatio-temporal statistics. Since 1980, we have led in designing and implementing studies of human exposures to toxicants, conducting more than 30 exposure field studies concerning multimedia, multipathway, and multichemical components.
Focus Areas
- Environmental sampling design
- Geostatistics, spatial, and spatio-temporal models
- Exposure studies
- Toxicology studies
- Risk assessment
Methodologies/Techniques
- Standard probability-based sample designs including stratified, cluster, systematic, multistage, and double sampling or two-phased sampling
- Environmental specific sample designs such as generalized random tessellation designs, capture recapture designs, ranked set sampling, line transects, variable circular plot sampling, area sampling, grid sampling, adaptive sampling, and composite sampling
- Experimental designs for causality studies, for controlling spurious causality and to evaluate the effects of combinations of treatments
- Kriging, nearest neighbor, spatial smoothing techniques
- Hierarchical/multilevel models to account for uncertainty and latent (spatial/non-spatial) effects
- Bayesian models to account for a priori scientific knowledge
- Markov Chain Monte Carlo techniques for parameter estimation
- Parametric (mixed models involving repeated measures and/or cluster correlated data, as well as generalized mixed models) and non-parametric models
- Exploratory data analysis
- Simulation (Monte Carlo techniques), jackknife, and bootstrap
- Sensitivity analysis
Capabilities
- Propose probability- and non-probability-based surveys as a basis for field studies involving direct exposure measurements, monitoring network design, and any environmental and ecological variable
- Apply spatial and spatio-temporal techniques for monitoring network assessment and modeling environmental processes
- Propose models that facilitate incorporating data from multiple sources (surveys, GIS layers, monitoring networks, etc.)
- Provide statistical consulting services for a variety of animal and plant toxicology studies conducted under the Federal Insecticide, Fungicide, and Rodenticide Act's good laboratory practice (GDP) guidelines
- Analyze safety assessment studies, including neurobehavioral, developmental toxicity and teratology, carcinogenicity, endocrine toxicity, avian toxicity, and acute studies for estimating LD50s (lethal dose 50%) and related parameters
- Apply parametric, non-parametric, and Bayesian methods for analyzing exposure data and performing risk assessments
- Use simulation techniques (e.g., as Monte Carlo methods, jackknife, and bootstrap) to assess uncertainties in current risk assessment practices, exposure studies and emission factors
Applications
- Through the years, we have developed sampling designs for collecting data on water quality, air quality, and human exposure to contaminants, including the Total Exposure Assessment Methodology studies and the national Human Exposure Assessment Survey field test in EPA Region 5.
- We have applied experimental design techniques to evaluate the effects of four management practices on the reproduction and growth of a threatened orchid species and to assess optional modes of data collection, as well as to identify and assess the magnitude and characteristics of measurement error and to examine how these vary temporally.
- We developed designs for estimating dose-response relationships and have investigated properties of calibration designs (spacing and replication of standard concentrations) aimed at optimal estimation of detection limits.
- We participated in studies to determine association between outcomes and exposures. For example, the association between adverse birth outcomes and exposure to drinking water arsenic, and the effects of environmental exposures to PCBs and other hormonally active agents on the ability of couples to conceive a child.
- Working with risk experts, we have applied modeling techniques to help EPA evaluate regulatory versus non-regulatory options for fossil fuel waste management.
- We have conducted a wide variety of toxicology studies for government and commercial clients, including identifying and characterizing hazards to human health and the environment and supporting pharmaceutical development.
Projects
- EPA/OAQPS/SPPD Pulp and Paper Multi-Pollutant Sector Strategies Project. We developed a geostatistical model of the impact of sulfur dioxide (SO2) emissions from individual pulp and paper mills on ambient PM2.5 levels and visibility (regional haze). The model allowed the EPA to identify those pulp and paper mills that have the most significant contributions to PM2.5 levels in or near PM2.5 nonattainment areas and Class I areas (national parks).
- EMAP Southeast Isolated Wetlands Assessment. Working with GIS and wetland scientists, we will develop a population frame using geodatabases, and select a probability sample of sites that will be used to assess population frame accuracy, and to estimate the number, extent and value of isolated wetlands in three states of North and South Carolina.
- Technical Support for Evaluating PBPK/PD Models for Interpreting Biomarkers of Exposure to Assess Cumulative Risk. We conducted exploratory research to evaluate the potential techniques for interpreting biomarkers of exposure using PBPK/PD approaches. We used statistical analysis to evaluate the utility of selected biomarkers of exposure in understanding potential pathways of exposure and dose.
- National Human Exposure Assessment Survey (NHEXAS) Phase I Field Study. We developed the design for NHEXAS, the objective of which was to evaluate comprehensive human exposure to multiple chemicals on a community and regional scale. The information gained from NHEXAS help individuals, communities, states, the EPA, and other organizations understand the greatest health risks from various chemicals and decide whether steps to reduce those risks are needed.
- Endocrine Disruptor Screening and Testing Standardization and Validation Technical Support Services (EDSP). We applied mixed models to validate results obtained from several labs performing assays to detect endocrine-active chemicals by measuring body and organ weight changes, organ microscopic changes, and nine circulating hormones according to a standard protocol, and in compliance with regulatory agency guidelines and GLP and laboratory SOPs.
- Developmental Toxicity Testing and Research for the National Toxicology Program of NIEHS. Our statisticians developed, documented, and implemented optimal methods of analysis for a variety of developmental toxicology studies. We developed methods of analysis for dose-interaction studies involving multiple compounds and cluster-correlated data, with continuous and binary outcomes. We wrote the statistical analysis plans for three dose-interaction studies with incomplete factorial designs, and reviewed all project protocols and final reports for accuracy of interpretation of statistical methods and results.
- The Non-Occupational Pesticide Exposure Study. For the EPA, we studied human, non-occupational exposures to 32 pesticides and their degradation products in two U.S. cities during three seasons. Environmental monitoring consisted of 24-hour indoor, personal, and outdoor air samples.
Contacts
- Breda Munoz, statistics
- Tim Wilcosky, epidemiology
- Gayle Bieler, toxicology