xvEPA
United States
Environmental Protection
Agency
The U.S. EPA National
Exposure Research
Laboratory's (NERL's)
Workshop on the Analysis of
Children's Measurement Data
Workshop Summary Report
RESEARCH AND DEVELOPMENT
-------
EPA 600/R-06/026
March 2006
The U.S. EPA National Exposure Research
Laboratory's (NERL's) Workshop on the
Analysis of Children's Measurement Data
Workshop Summary Report
By
Nicolle S. Tulve1, Jeff Driver2, Peter E. Egeghy1, Jeff Evans3, Roy C. Fortmann1, John Kissel4, Nancy McMillan5,
Lisa J. Melnyk6, Marsha Morgan1, James Starr1, Daniel M. Stout II1, MarkStrynar1
1US EPA, Office of Research and Development, National Exposure Research Laboratory, Human Exposure and
Atmospheric Sciences Division, Research Triangle Park, NC 27711
2infoscientific.com, Inc., Manassas, VA 20111
3US EPA, Office of Pesticide Programs, Health Effects Division, Washington, DC 20460
4University of Washington, Department of Environmental and Occupational Health Sciences, Seattle, WA 98195
5Battelle, Columbus, OH 43201
6US EPA, Office of Research and Development, National Exposure Research Laboratory, Microbiological and
Chemical Exposure Assessment Research Division, Cincinnati, OH 45268
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC, 27711
-------
Notice
The U.S. Environmental Protection Agency, through its Office of Research and Development, National
Exposure Research Laboratory, sponsored the workshop described in this report. It has been subjected to
the Agency's peer and administrative review and has been approved for publication as an EPA document.
Mention of trade names or commercial products does not constitute endorsement or recommendation for
use. This report was prepared as a general record of the discussions held during the U.S. EPA National
Exposure Research Laboratory's (NERL's) Workshop on the Analysis of Children's Measurement Data
(September 27-28, 2005). This report captures the main points and highlights of the meeting. It is not a
complete record of all details discussed, nor does it embellish, interpret, or enlarge upon matters that were
incomplete or unclear. Statements represent the collective views of the workshop participants. None of the
statements should be used to infer official positions or policy of the EPA.
-------
Executive Summary
The U.S. EPA National Exposure Research Laboratory's (NERL's) Workshop on the Analysis of
Children's Measurement Data was held on September 27 and 28, 2005. A group of recognized experts in
the fields of exposure and health assessments, toxicology, statistics and biostatistics, modeling, analytical
chemistry, and biomarker measurements was assembled to discuss three themes regarding measurements
data from recent children's exposure studies conducted or funded by NERL. To facilitate the two-day
discussions, a summary document containing information on the recent studies of children's exposure was
compiled and distributed to all attendees. The summary included data from laboratory and observational
field studies that had been conducted or funded by NERL over the previous years. The studies took place
in the EPA Research Test House, in child care centers, and in private residences. The field studies were
observational measurement studies. No chemicals or other products were introduced by EPA into the
homes or day care centers, and participants were asked to follow their normal daily routines while the
environmental samples and other types of information were collected. The workshop attendees were
assigned to one of four groups to discuss the themes as presented and any additional ideas that were
identified. The groups were evenly mixed with Program Office, Industry, Academia, and Government
researchers.
Workshop participants were asked to consider the following themes during their discussions:
Theme 1: Major trends in the data across children's exposure studies
a. Identify the major sources and routes of children's exposure to the pesticides based on the data
from these and other children's exposure studies. Does this change by pesticide or class of
pesticide?
b. Identify the major factors that influence children's exposures to the pesticides.
c. Are the observations in the recent NERL studies consistent with results from other studies and
with current assumptions?
Theme 2: Additional hypotheses that should be tested using existing data
a. Provide a list of the major hypotheses that could be tested on existing data.
b. Why are these considered to be important hypotheses?
c. How suitable are the existing data? What is still missing?
Theme 3: Additional statistical analyses that can be performed on these data to better understand
children's exposures to chemicals at homes and daycare centers? Provide a list of statistical analyses
that could be performed on the accrued data. For each proposed analysis, please specify
a. What data gap are you addressing?
b. What key hypothesis are you testing?
c. What result do you expect to find?
d. What is the likelihood of success, given the limitations of the data?
At the conclusion of the meeting, each group produced a summary document that captured their theme
discussions. There were many similar observations recorded for the groups. For example, all groups
recognized the importance of diet as a route of exposure. It was also recognized that adequately capturing
in
-------
children's activities is critically important to understanding and evaluating their exposures. The need for
more statistical evaluations that incorporate the questionnaire data was deemed highly important. Many
additional hypotheses and statistical analyses were proposed during the workshop.
This workshop report lists a summary of the observations related to each theme. These results will be
used to generate a detailed statistical analysis plan that will be implemented using the available data.
IV
-------
Contents
Notice ii
Executive Summary iii
Acknowledgements vi
Chapter 1: Introduction 1
Background 1
References 2
Goal of the Workshop 4
Report Structure 4
Agenda 5
Charge Questions 6
Presentations by Guest Speakers 7
Chapter 2: Summary of the Workgroup Reports 12
ChapterS: Summary of the Open Discussion on Collaboration 16
Appendix A: Individual Workgroup Materials 18
Appendix B: Participant List 40
Appendix C: Speaker Presentations 43
-------
Acknowledgements
The authors would like to thank all the workshop participants for their valuable contributions to this
report.
VI
-------
Chapter 1
Introduction
Background
The U.S. Environmental Protection Agency (U.S. EPA) has pledged to increase its efforts to provide a safe
and healthy environment for children by ensuring that all EPA regulations, standards, policies, and risk
assessments take into account special childhood vulnerabilities to environmental toxicants. Children are
behaviorally and physiologically different from adults. Their interaction with their environment, through
activities such as playing on floors, mouthing of hands and objects, and handling of food, may increase
contact with contaminated surfaces. Proportionally higher breathing rates, relative surface area, and food
intake requirements may increase exposure. Differences in absorption, metabolism, storage, and excretion
may result in higher biologically-effective doses to target tissues. Immature organ systems may be more
susceptible to toxicological challenges. Windows of vulnerability, when specific toxicants may
permanently alter the function of an organ system, are thought to exist at various stages of development.
Children are commonly exposed to a wide variety of chemicals in their daily lives in their homes, schools,
day care centers, and other environments that they occupy. The chemicals to which they are exposed may
originate from outdoor sources, such as ambient air contaminants, and indoor sources such as building
materials and consumer products. One category of consumer products to which children may be exposed is
pesticides that are used to control roaches, rats, termites, ants, and other vermin. Despite widespread
residential and agricultural use of pesticides, little is known about the actual levels of pesticide exposure
among children and the factors that impact children's exposures to pesticides. The Food Quality Protection
Act (FQPA) of 1996 requires EPA to upgrade the risk assessment procedures for setting pesticide residue
tolerances in food by considering the potential susceptibility of infants and children to both aggregate and
cumulative exposures to pesticides. Aggregate exposures are exposures from all sources, routes, and
pathways for individual pesticides. Cumulative exposures include aggregate exposures to multiple
pesticides with the same mode of action for toxicity. Specifically, FQPA requires that risk assessments
must be based on exposure data that are of high quality and high quantity or exposure models using factors
that are based on existing, reliable data.
EPA's Office of Research and Development (ORD) is responsible for conducting research to provide the
scientific foundation for risk assessment and risk management and to meet the mandates of the FQPA. In
2000, ORD released its Strategy for Research on Environmental Risks to Children addressing research
needs and priorities associated with children's exposure to environmental pollutants and providing a
framework for a core program of research in hazard identification, dose-response evaluation, exposure
assessment, and risk management.
The National Exposure Research Laboratory (NERL) is working to achieve three specific objectives of the
ORD Strategy through its Children's Exposure Research Program: (1) develop improved exposure
assessment methods and models for children using existing information; (2) design and conduct research on
age-related differences in exposure, effects, and dose-response relationships to facilitate more accurate risk
assessments for children; and (3) explore opportunities for reducing risks to children. After an exhaustive
review of the volume and quality of the data upon which default assumptions for exposure factors are based
(Cohen Hubal et al., 2000a), a framework for systematically identifying the important sources, routes, and
pathways for children's exposure was developed (Cohen Hubal et al., 2000b).
This framework (see Figure below), based on a conceptual model for aggregate exposure, provides the
foundation for a protocol for measuring aggregate exposures to pesticides (Berry et al., 2001) and for
developing sophisticated stochastic models (Zartarian et al., 2000). Using the framework, four priority
research areas, representing critical data gaps in our understanding of environmental risks to children, have
been identified:
(1) Pesticide use patterns;
(2) Spatial and temporal distribution in residential dwellings;
(3) Dermal absorption and indirect (non-dietary) ingestion (including micro- and macro-activity
-------
approaches); and
(4) Direct ingestion.
Several targeted studies were designed and conducted to address these research needs. These include
laboratory studies, small pilot field studies, and large collaborative field studies. These studies aimed to
(1) evaluate methods and protocols for measuring children's exposure.
(2) collect data on exposure factors to reduce the uncertainty in exposure estimates and risk
assessments.
(3) collect data for use in exposure model development and evaluation.
The studies took place in the EPA Research Test House, child care centers, and private residences. The
field studies were observational measurement studies. No chemicals or other products were introduced by
EPA into the homes or day care centers. Participants were asked to follow their normal daily routines
while the environmental samples and other types of information were collected. Organophosphate and
pyrethroid pesticides and their metabolites were measured in environmental samples (e.g., air, surface dust
residues, diet) representing multiple routes and pathways of exposure and in biological fluids (e.g., urine).
Time/activity and other questionnaire data were also collected. The primary participants in the field studies
were children ranging in age from 8 months to 12 years.
References
Berry, MR, Cohen Hubal, EA, Fortmann, RC, Melnyk, LJ, Sheldon, LS, Stout, DM, Tulve, NS, Whitaker,
DA. 2001. Draft Protocol for Measuring Children's Non-Occupational Exposure to Pesticides by all
Relevant Pathways. EPA/600/R-03/026. Office of Research and Development, Research Triangle Park,
NC.
Cohen Hubal, EA, Sheldon, LS, Burke, JM, McCurdy, TR, Berry, MR, Rigas, ML, Zartarian, VG. 2000a.
Exposure assessment for children: a review of the factors influencing exposure of children, and the data
available to characterize and assess that exposure. Environmental Health Perspectives 108(6):475-486.
Cohen Hubal, EA, Sheldon, LS, Zufall, MJ, Burke, JM, Thomas, K. 2000b. The challenge of assessing
children's residential exposure to pesticides. Journal of Exposure Analysis and Environmental
Epidemiology 10:638-649.
Zartarian, VG, Ozkaynak, H, Burke, JM, Zufall, MJ, Rigas, ML, Furtaw, EJ Jr. 2000. A modeling
framework for estimating children's residential exposure and dose to chlorpyrifos via dermal residue
contact and nondietary ingestion. Environmental Health Perspectives 108(6):505-14.
-------
Source
indoor
residential
outdoor
residential
industrial
agricultural
Release
&
Transfer
Inhalation
Activity
Exposure V^ Patterns
Media
Respiratory
Tract
Inhalation
Exposure
Rate
R.T. Uptake
T Uptake
Rate/
Indoor
air
water
house
dust
food
surfaces
clothes
Modeling framework for children's pesticide exposure from Cohen Hubal et a/. (2000b).
-------
Goal of the Workshop
The U.S. EPA National Exposure Research Laboratory's (NERL's) Workshop on the Analysis of
Children's Measurement Data was held on September 27 and 28, 2005 in Research Triangle Park, NC.
Workshop participants were provided with a summary document that contained the following information:
• Descriptions of recent children's exposure studies conducted or funded by NERL, including
descriptions of the parameters measured and the measurement methods;
• Concentration data and summary statistics for comparisons of the studies;
• Simple comparative analyses; and,
• Highlights of the results of the studies.
The goal of the workshop was to assemble a group of recognized experts in the fields of exposure and
health assessments, toxicology, statistics and biostatistics, modeling, analytical chemistry, and biomarker
measurements to discuss the major trends in the data, additional hypotheses, and additional statistical
analyses that should be undertaken with the data (Chapter 1), using the summary document and an
overview presentation by Dr. Linda Sheldon as the bases for discussions. The results of the two-day
workshop have been captured in this report (Chapter 2, Appendices A and C). An important outcome from
this workshop will be a research plan for additional statistical analyses of the collected children's
measurement data in order to improve our understanding of the relevant factors affecting children's
exposures to pesticides.
Report Structure
The preliminary text contains an Executive Summary. Chapter 1 includes a brief discussion of the
background for the research, a description of the purpose of the workshop, the meeting agenda, the charge
questions, and biographical information on the guest presenters. Chapters 2 and 3 provide summaries of
the charge questions and the discussion on collaborative ideas. Appendix A contains the individual group
discussion materials. Appendix B contains the list of attendees with contact information. Appendix C
contains the overview presentation given by Dr. Linda Sheldon, Acting Director of the Human Exposure
and Atmospheric Sciences Division, and the presentations given by the guest presenters during day one of
the workshop.
-------
Agenda
Day 1: Tuesday September 27,2005
8:00 am - 9:00 am Continental Breakfast (Camillia Room)
9:00 am - 9:30 am Registration
9:30 am -10:00 am Welcome *Breaks on your own
Overview Presentation (Camillia Room)
Presenter: Dr. Linda Sheldon (Acting HEASD Director, EPA)
10:00 am - 12:00 noon Presentations by Guest Speakers (20 min. each)
Speaker 1 - Bob Lordo (Battelle)
Speaker 2 - Richard Fenske (University of Washington)
Speaker 3 - Robin Whyatt (Columbia University)
Speaker 4 - Haluk Ozkaynak (USEPA/ORD)
Speaker 5 - Jeff Evans (USEPA/OPP)
12:00 noon - 1:00 pm Lunch on your own.
1:00 pm - 3:00 pm Theme 1: Major Trends in The Data *Groups (A, B, C, or D) - assigned rooms
3:00 pm - 4:00 pm Theme 2: Additional Hypotheses
4:00 pm - 5:00 pm Report Out for Themes 1 and 2 *A11 Groups - Camillia Room
5:00 pm Adjourn Dinner on your own - suggestions will be provided.
Day 2: Wednesday, September 28,2005
8:00 am - 8:30 am Continental Breakfast (Camillia Room)
8:30 am - 9:00 am Overview of Today's Agenda (Camillia Room)
Presenter: Dr. Linda Sheldon
9:00 am - 11:00 am Theme 3: Additional Statistical Analyses
11:00 am - 12:00 noon Report Out for Theme 3
12:00 noon - 1:00 pm Lunch on your own.
1:00 pm - 2:30 pm Future Collaborations (optional)
2:30 pm - 3:00 pm Closing Remarks
Presenter: Dr. Linda Sheldon
3:00 pm Adjourn
-------
Charge Questions
To facilitate the two-day discussions, charge questions were used to focus the group discussions and are
reproduced here.
Charge Questions for the U.S. EPA NERL's Workshop on the Analysis of Children's Measurement
Data
NERL has conducted and/or funded many children's exposure studies involving pesticides over the past
several years. We have summarized the information from a number of these studies in the provided draft
report entitled "Summary and Comparison of Data Collected in NERL Children's Pesticide Exposure
Studies".
For this workshop, NERL would like to have open discussions with fellow researchers on the data and
results from these studies. In addition, we encourage fellow scientists to share their data and findings from
their children's exposure studies as well.
We have divided the workshop members into four assigned groups A, B, C, or D. The charge for each
group is to provide your input and comments on the following research themes:
Theme 1: Major trends in the data across children's exposure studies
a. Identify the major sources and routes of children's exposure to the pesticides based on the data
from these and other children's exposure studies. Does this change by pesticide or class of
pesticide?
b. Identify the major factors that influence children's exposures to the pesticides.
c. Are the observations in the recent NERL studies consistent with results from other studies and with
current assumptions?
* Two hour time limit
Theme 2: Additional hypotheses that should be tested using existing data
a. Provide a list of the major hypotheses that could be tested on existing data.
b. Why are these considered to be important hypotheses?
c. How suitable are the existing data? What is still missing?
* One hour time limit
Theme 3: Additional statistical analyses that can be performed on these data to better understand
children's exposures to chemicals at homes and daycare centers
Provide a list of statistical analyses that could be performed on the accrued data. For each proposed
analysis, please specify
a. What data gap are you addressing?
b. What key hypothesis are you testing?
c. What result do you expect to find?
d. What is the likelihood of success, given the limitations of the data?
Please prioritize the proposed analyses.
* Two hour time limit
-------
Presentations by Guest Speakers
During Day 1 (September 27, 2005) of the workshop, five guest speakers were asked to discuss the
importance of the data contained in the summary document with respect to their research. Presentation
summaries and speaker biographies are captured below. Speaker presentations can be found in Appendix
C.
Speaker 1: Dr. Robert Lordo, Battelle
Title: Major findings and recommendations based on the outcome of statistical analysis
of children's exposure data from the CTEPP study
Description of Presentation: EPA has recently completed an observational pilot study of Children's Total
Exposure to Persistent Pesticides and Other Persistent Organic Pollutants (CTEPP), in which environmental
monitoring, personal exposure, and biomarker data were collected on selected pesticides and other
pollutants from the homes and day care centers of 257 preschool children, in order to investigate their
aggregate exposures to these pollutants. Statistical modeling techniques were used to characterize how
personal exposure levels may differ among different types of environments (e.g., urban vs. rural, day care
vs. stay-at-home children) and which exposure routes were most dominant for a particular pollutant. In
addition, analysis of chlorpyrifos and chrysene data from the CTEPP study were recently performed using
novel statistical techniques, including structural equations modeling and hierarchical Bayesian modeling, in
order to evaluate the utility of selected exposure biomarkers in understanding potential exposure pathways
to children. Important findings and conclusions from the analyses performed on these data will be
presented, along with questions and issues that remain unresolved and, therefore, could impact the design
of future children's exposure studies.
Dr. Robert Lordo has served as an environmental statistician at Battelle for the past 17 years. His primary
research focus has been in areas of exposure assessment and risk analysis, using statistical modeling
techniques to evaluate the extent to which sensitive subpopulations are exposed to toxic chemicals and to
characterize the pathways of exposure within various environmental media and ultimately to human
biomarkers. For several years, he has provided statistical support to EPA's Lead Program, characterizing
exposures of lead-based paint hazards and their statistical link to elevated blood-lead levels in children. He
recently served as principal statistician on EPA's pilot study of Children's Total Exposure to Persistent
Pesticides and Other Persistent Organic Pollutants (CTEPP).
-------
Speaker 2: Dr. Richard A. Fenske, University of Washington
Title: How can we improve the accuracy of children's measurement data?
Description of Presentation: This presentation will focus on the strengths and limitations of several
approaches to characterizing children's exposure to pesticides: the use of urinary metabolite monitoring in
conjunction with environmental and personal sampling; the potential for saliva monitoring to provide a
better understanding of internal dose; and the value of global positioning system instruments for improving
our understanding of children's macro (time-location) activities. Examples will be drawn from laboratory
and fields studies conducted by University of Washington researchers.
Dr. Richard A. Fenske, PhD, MPH, is a professor in the Industrial Hygiene and Safety Program in the
Department of Environmental and Occupational Health Sciences, School of Public Health and Community
Medicine, University of Washington. He has served as director of the NIOSH-supported Pacific Northwest
Agricultural Safety and Health Center since its establishment in 1996. His research interests include
exposure assessment and intervention studies in the workplace and in communities. His current studies
included reducing children's exposure to pesticides in agricultural communities, using fluorescent tracers
for pesticide safety education, improving risk communication methods, and developing novel exposure
assessment methods. He teaches courses in environmental sampling and analysis, and environmental risk
analysis. He is a member of the U.S. Environmental Protection Agency's Science Advisory Board, and a
member of the National Academy of Sciences/Institute of Medicine Committee to Review the Health
Effects in Vietnam Veterans of Exposure to Herbicides. From 1994-2004 he served on the National
Advisory Panel of the National Cancer Institute's Agricultural Health Study, a prospective epidemiological
study of pesticide applicators and their families.
-------
Speaker 3: Dr. Robin Whyatt, Columbia Center for Children's Environmental Health, Columbia
University
Title: Validating biomarkers of prenatal insecticide exposure. Will this help resolve conflicting
epidemiologic findings?
Description of Presentation: Three recent epidemiologic studies have examined the relationship between
prenatal organophosphate exposures and birth outcomes (see EHP, 112, 2004). Results are not consistent.
Associations reported include decreased birth weight and length (Whyatt et al.); decreased head
circumference with levels of paraoxonase 1 activity as a modifier (Berkowitz et al.) and decreased
gestational age (Eskenazi et al.). Differences in exposure patterns and in the biomarkers used (chlorpyrifos
in cord blood versus chlopyrifos metabolites in maternal prenatal urine) have been identified as factors that
could have contributed to the discrepant findings (Needham, EHP, 113, 2004). Between 2001-04, we
undertook a validation study nested within one of the cohorts to examine relationships between these
biomarkers and measures of external exposure. 85% of the women report using some form of pest control
during pregnancy and 46% report using higher toxicity methods. The organophosphates chlorpyrifos and
diazinon and the carbamate, propoxur were detected in 99.6%-100% of repeat two-week integrated indoor
air samples (n=354 samples) collected from 102 homes during pregnancy. There was little within home
variability in the indoor air concentrations; between home variability accounted for 92% of the variance in
chlorpyrifos levels, 94% in diazinon levels and 87.7% in propoxur levels (p<0.001, mixed model).
Chlorpyrifos levels in repeat indoor air samples were significantly correlated with levels of 3,5,6-trichloro-
2-pyridinol (TCPY) in repeat prenatal maternal spot urine samples (r=0.3, p<0.01) but not with levels of
chlorpyrifos in maternal or umbilical cord blood. TCPY levels in postpartum meconium were significantly
correlated with TCPY levels in maternal prenatal urine samples (r=0.4, pO.OOl) and with chlorpyrifos
levels in maternal and umbilical cord blood collected at delivery (r=0.35-0.44, p=0.001). Chlorpyifos and
TCPY in the environmental and biologic samples declined significantly following the U.S. EPA 2000-2001
regulation action to phase out residential uses. Results suggest that organophosphate metabolites measured
in meconium may provide an alternative dosimeter for examining effects of prenatal exposure on birth
outcomes.
Dr. Robin Whyatt is Associate Professor of Clinical Environmental Health Sciences at the Mailman
School of Public Heath at Columbia University and is Deputy Director of the Columbia Center for
Children's Environmental Health. Her research focus is on the effects of environmental exposures on
women and children, including the developing fetus. This has included molecular epidemiologic research
on prenatal exposures to ambient air pollution and cigarette smoking in Poland and research on the effects
of environmental exposures among African American and Dominican mothers and newborns from New
York City. Dr. Whyatt's particular focus is the effects of prenatal exposure to non-persistent pesticides
(organophosphates, carbamates and pyrethroids) and phthalates among this minority population. She is
principal investigator on a number of federal research grants including studies to validate biomarkers of
prenatal pesticide exposure; to assess exposures to the newer use insecticides and to evaluate effects of
phthalate exposures during pregnancy. She is also collaborating with the Center for Disease Control on the
validation of biomarkers of prenatal exposure to contemporary-use pesticides. Dr. Whyatt has published
widely on research using biologic markers in studies of prenatal exposures and has served on a number of
federal committees. She currently serves on a National Academy of Science Committee on Human
Biomonitoring for Environmental Toxicants and is co-chair of the chemical exposure workgroup for the
National Children's Longitudinal Cohort Study.
-------
Speaker 4: Dr. Haliik Ozkaynak, USEPA's National Exposure Research Laboratory
Title: Modeling Children's Exposure to Pesticides: Issues and Challenges
Description of Presentation: This presentation provides an overview of issues and challenges encountered
in modeling children's exposures to pesticides. Opportunities for reducing model and input uncertainties
are discussed. In particular, areas where field measurement data are especially important for model
evaluation and refinement are identified. Recent pesticide exposure modeling studies indicate that major
uncertainties in model results stem from data gaps for critical factors, including information on pesticide
usage; the pattern and frequency of touching surfaces; hand- to-mouth and object-to mouth activity
patterns; pesticide residue concentrations at different post-application times; residue transfer to the skin
surface upon contact; where people spend their time in relation to where residues exist in the residential
environment; and factors related to intake and uptake into the human body after exposure. As more
exposure measurements data become available and collectively analyzed, uncertainties in the exposure
model results should be reduced. By the same token, as model refinements progress, models will help
define critical areas where further survey and measurement research are needed.
Dr. Haluk Ozkaynak is a Senior Scientist at U.S. EPA's National Exposure Research Laboratory (NERL)
in the Office of Research and Development (ORD). The principal responsibilities of his position includes
developing and applying new exposure analysis and modeling methods for the assessment of population
health risks in order to improve risk assessment and management decisions of EPA, by active collaboration
with scientists from different ORD Laboratories, Centers and various EPA Program Offices. Dr. Ozkaynak
is presently the co-chair of the EPA Workgroup on Probabilistic Risk Analyses. He is also a member of the
World Health Organization's International Programme on Chemical Safety (WHO's IPCS) Uncertainty in
Exposure Assessment and EPA's Integrating Scientific Information Workgroups. Recently, he has served
as the co-chair of the Chemical Exposure Workgroup of an interagency study in the US for the planned
National Children's Study (NCS). Prior to joining EPA in 1998, Dr. Ozkaynak was a Lecturer at the
Department of Environmental Health of Harvard School of Public Health in Boston. His research at
Harvard included directing a multi year environmental epidemiology study in Russia, and participating in
various exposure assessment and community studies, including the National Human Exposure Assessment
Survey (NHEXAS) and the Kanawha Valley Health Studies, both sponsored by the U.S EPA. Dr.
Ozkaynak is a former President of the International Society of Exposure Analysis (ISEA).
10
-------
Speaker 5: Mr. Jeff Evans, Office of Pesticide Programs, Health Effects Division
Title: OPP Regulatory Perspective
Description of Presentation: This presentation will focus on how EPA can best use the data developed by
ORD in the day-to-day regulation of pesticides. What has worked, what has not? What do we still need?
Mr. Jeff Evans is a Senior Exposure Assessor for occupational and consumer uses of pesticides, member
of the Division's Risk Assessment Review Committee, and Chair of the Exposure Science Advisory
Counsel. He is involved in the development of Standard Operating Procedures for Residential Exposure,
Occupational and Residential Exposure Study Guidelines, and Cumulative Exposure Assessments for the
Organophosphate and Carbamate Pesticides.
11
-------
Chapter 2
Summary of the Workgroup Reports
Individual workgroup reports can be found in Appendix A. This section describes the common ideas, by
theme, that were independently developed in each workgroup.
Theme 1: Major trends in the data across children's exposure studies.
a. Identify the major sources and routes of children's exposure to the pesticides based on the data
from these and other children's exposure studies. Does this change by pesticide or class of
pesticide?
• Evidence suggests that diet is one of the major sources of exposure to pesticides.
• Indirect ingestion may be an important route of exposure to pyrethroid pesticides.
• Inhalation may be an important route of exposure to organophosphate pesticides.
• Drinking water was not considered an exposure source for most children.
• In general, beverages other than pure fruit juices can be dismissed as an exposure source.
• Other potentially important sources and routes of exposure that were identified included:
"excess dietary" ingestion (i.e., added contamination to foods from hands and residential
surfaces during eating), prenatal/gestational/lactational exposures, indoor residential pesticide
usage, other indoor environments where young children spend time, outdoor applications on
the lawn, agricultural areas, mosquito control, occupational take-home exposures, pets, and
personal care products.
• While risk assessments must account for all routes of exposure, further analyses should be
undertaken to investigate if chemical class membership or physicochemical properties can be
exploited in determining the dominant route of exposure for individual pesticides.
b. Identify the major factors that influence children's exposure or dose to the pesticides.
• Non-dietary ingestion, especially hand-to-mouth activity.
• Diet.
• Children's activities and behavioral patterns, including hand washing and bathing.
• Food handling behaviors.
• Bioavailability of the pesticide.
• Chemical specific exposure factors such as volatility, water solubility, degradation, and
toxicokinetics.
• Specific product use characteristics, including numbers and types of pesticides applied in the
indoor environment.
• Spatial and temporal variability in the indoor environment.
• Other exposure factors including demographics, specific food consumption patterns,
physiological factors.
c. Are the observations in the recent NERL studies consistent with results from other studies and
with current assumptions?
• In general, the observations presented in the NERL studies are consistent with results from
other studies and with current assumptions.
• All groups noted the importance of gaining a better understanding of the levels of pesticide
metabolites in environmental media and how those levels influence the interpretation of
urinary biomarker data.
12
-------
Theme 2: Additional hypotheses that should be tested using existing data?
The workshop participants were asked to propose additional hypotheses that should be tested using the
existing data. More often than not, the discussions generated a series of important research questions, but
not necessarily research questions that could be tested using NERL's existing data. In theme 2, section a,
we list many of the research questions that were raised during the two-day workshop.
a. Provide a list of the major research questions that could potentially be tested on existing data.
• Are the algorithms correct?
• What are the best inputs to use in the algorithms to assess children's exposures to pesticides?
How can the algorithms be refined?
• Are the toxicokinetics similar for classes of chemicals?
• How do we reduce the uncertainty in the most important exposure factors, such as hand-to-
mouth transfer?
• Is peak exposure or average exposure the most important metric with respect to health
effects?
• Is indirect ingestion a significant route of exposure? How do we accurately estimate indirect
ingestion? What is the magnitude of intra- and inter-personal variability? Are there
discernable characteristics associated with the magnitude of the indirect ingestion?
• Can creatinine levels in children be used as a proxy for activity level?
• How is pesticide fate and transport different now that we are using less volatile pesticides?
What is the relationship between volatile/less volatile pesticides?
• Does changing the resolution of sampling time affect the correlation with urine?
• Do biomarkers correlate better with peak exposures than with average exposures?
• Are there behavioral measurement tools currently used in other fields that we can adapt to
help us gain a better understanding of the activity/behavioral/pesticide exposure relationship?
• Are there alternative approaches that utilize behavioral categories that can be used instead of
the micro-environment approach and can these be related to transfer coefficient estimates? Is
grouping children by macroactivity classification more descriptive as a model input than
using microactivity classifications?
• Would removing the dietary contribution allow a clearer understanding of the remaining
pathways and an enhanced sensitivity analysis?
• How do results from high exposure situations (e.g., post-application scenarios or identified
high-use populations) compare to the overall population's environmental and biological
levels?
• Does input of OPP transfer coefficients (and related assumptions regarding absorption) give
results consistent with observations?
• For studies where dermal transfer coefficients can be calculated, are the values consistent with
the values used by OPP?
• What are the impacts of sampling methods on measured dose calculations? Are different
answers obtained based on the specific methods used?
• Which environmental media contribute significantly to urinary biomarker levels?
• Do one or more environmental measurement parameters correlate with absorbed dose
distributions for biomarkers?
• Are urinary biomarker levels significantly increased due to crack and crevice applications?
• What is the significance of total versus vapor-phase inhalation exposures?
• Do individuals who consume large amounts of leafy vegetables have higher exposures to
pesticides?
• Does accounting for direct intake of the metabolite improve modeled estimates of absorbed
dose?
• Are there spatial distributions in the indoor environment? How do we effectively capture
these distributions?
• Are there regional differences in the magnitude of the residential exposure?
13
-------
• What alternative biomarkers of exposure should be considered? Are urinary biomarkers
really the most appropriate?
b. Why are these considered to be important hypotheses?
Testing these hypotheses will help to determine the following:
• How to better understand and capture children's activities.
• How to apportion the routes of exposure based on relative importance; whether the route
apportionment is related to the physico-chemical characteristics of the chemicals; whether
chemicals can be evaluated as a class.
• How to use biomarkers of exposure to understand and interpret children's exposure.
• How to reduce the default values that are being used in the algorithms, models, and regulatory
risk assessments.
• Whether the algorithms need to be refined, and, if so, how.
c. How suitable are the existing data? What is still missing?
The following were identified as data gaps or issues with the existing data:
• Temporal dimension.
• Data on very young children (under 3 years).
• Missing data elements (e.g., dietary information for young children).
• Measurement methods and method detection limits (MDLs) differ. The treatment of non-
detects has not been consistent across studies.
• Time activity data and pesticide use information limited and of questionable accuracy.
• Scenario specific measurements from food handling establishments.
• A thorough mining of the questionnaire data. Analysis of the questionnaire data to determine
the information that contains predictive ability.
• Unification/standardization/normalization of methodologies and QA/QC procedures.
• A comparison of the studies to national trends or the larger population.
• Aggregated/uniform data sets of larger studies.
• Parent/metabolite relationships in environmental media.
Theme 3: Additional statistical analyses that can be performed on these data to better understand
children's exposures to chemicals at homes and daycare centers?
Provide a list of statistical analyses that could be performed on the accrued data. For each proposed
analysis, please specify
a. What data gap are you addressing?
b. What key hypothesis are you testing?
c. What result do you expect to find?
d. What is the likelihood of success, given the limitations of the data?
The discussions on additional statistical analyses that could be performed on the accrued data were not
as developed as we had hoped. In many instances, additional statistical analyses were discussed, but
the data gap being addressed, the key hypothesis, the result expected, or the likelihood of success were
not discussed in relation to the statistical analysis proposed. The following list of bullets highlights the
various statistical analyses proposed by the workgroups.
• Create a data warehouse where all the data from these studies can be housed for statistical
analyses and model evaluation. Perform more multivariate analyses using all available
environmental measurements and questionnaire data to describe the relationships between the
measurements and the biomarker over time. Pathway analysis should be included in these
14
-------
analyses. Model testing should be included with these analyses. Evaluate the relationship
between the environmental concentrations and the biomarkers. For example, ANOVA,
Bayesian hierarchical analysis, sensitivity analysis.
• Statistically analyze the data set to estimate the parameters that go into the mechanistic
models.
• Evaluate the relationship between the variability in the biomarker data and the dietary data.
• Normalize the data based on age, gender, and body mass.
• Compare the NERL data to other available data.
• Use the principal component analysis as a method for dealing with the interdependence of the
environmental measurements.
• Apply pharmacokinetic data to adjust the dose estimates.
• Examine the currently applied assumptions to investigate the disparity between the
environmental and urinary concentrations.
• Pharmacologically adjust the calculated dose based on the previous days environmental
measurements (when the environmental and biomarker sampling schedules are temporally
coherent).
• Develop exposure estimates based on OPP methods and compare to biomonitoring results.
• Cluster analyses.
• Compare the inventory data collected with the Residential Exposure Joint Venture (REJV)
national survey.
Many research questions and statistical analyses were proposed in themes 2 and 3. The following is
the authors' prioritized list of the research questions and statistical analyses that we believe we can
successfully evaluate with the data.
Research questions:
• Can creatinine levels in children be used as a proxy for activity level?
• Does changing the resolution of sampling time affect the correlation with urine?
• Do biomarkers correlate better with peak exposures than with average exposures?
• Would removing the dietary contribution allow a clearer understanding of the remaining
pathways and an enhanced sensitivity analysis?
• Does input of OPP transfer coefficients (and related assumptions regarding absorption) give
results consistent with observations?
• For studies where dermal transfer coefficients can be calculated, are the values consistent with
the values used by OPP?
• Do one or more environmental measurement parameters correlate with absorbed dose
distributions for biomarkers?
• Do individuals who consume large amounts of leafy vegetables have higher exposures to
pesticides?
Statistical analyses:
• Evaluate the relationship between the environmental concentrations and the biomarkers.
• Evaluate the relationship between the variability in the biomarker data and the dietary data.
• Normalize the data based on age, gender, and body mass.
• Compare the NERL data to other available data.
• Use the principal component analysis as a method for dealing with the interdependence of the
environmental measurements.
• Examine the currently applied assumptions to investigate the disparity between the
environmental and urinary concentrations.
• Develop exposure estimates based on OPP methods and compare to biomonitoring results.
15
-------
Chapter 3
Summary of the Open Discussion on Collaboration
Linda Sheldon opened the session. She indicated that this should be an informal discussion of
ways that exposure researchers could collaborate on data analysis efforts.
Janice Sharp (Valent BioSciences) is currently the chair of the Non-Dietary Exposure Task Force
(NDETF), a group that represents the pyrethrin and pyrethroid pesticide supplier/manufacturer
industry. Janice discussed the research-related work of the Task Force and the potential
availability of data. NDETF has done a lot of research on the pyrethroids. This includes
laboratory testing on pesticide degradation and transfers to surfaces (e.g., hand presses). She was
concerned that a lot of the studies they have done may be considered dosing studies by EPA under
the new rulemaking. Much of the data has been provided to OPP. She may be able to make some
of it publicly available.
The pesticide industry has performed the Residential Exposure Joint Venture (REJV), a survey of
pesticide use in the U.S. They have made the data available to EPA/OPP and EPA/ORD.
Curt Lunchik indicated that industry is very interested in sharing data. There was a discussion of
how the process would work for EPA to obtain industry data. Data can be shared as confidential
business information. It was suggested that HEASD take a lead in trying to move this effort
forward.
Elaine Cohen Hubal suggested that the best way to move collaborations forward to maximize the
utility of data collected by different organizations would be to identify one to three questions of
greatest importance and try to answer these. If there is a hypothesis to be tested, we can then
determine what data are needed, who has the data, and pull the data together from all available
sources for the analyses.
Rich Fenske suggested that the Pesticide Handler Exposure Database (PHED) was an example of
how this has worked in the past. Curt and others indicated that the Agricultural Handler Exposure
Database (AHED), the successor to PHED, is the current example.
A number of researchers discussed potential questions/research areas that could be used to start the
collaborations.
Curt Lunchik, with agreement from a number of others in the group, suggested that the first area
could be analyses of existing biomonitoring data to look at the relationship of pesticide metabolite
levels and measurements of the parent pesticide(s) in environmental media. There are many
groups that have urinary metabolite levels and other measurements. Linda Sheldon suggested that
we needed to start small with a "doable" project. Roy Fortmann suggested that the project should
start with the pyrethroid pesticide metabolites because there is some data and they are the current-
use pesticides. He suggested that the analyses of chlorpyrifos/TCPy relationships may be more
difficult and less relevant. There is an International Life Sciences Institute (ILSI) workgroup
headed by Linda Sheldon that might consider this as a project. The group suggested that Linda's
workgroup pursue this effort.
The second potential area, suggested by Lisa Melnyk, was quantifying behavior/activity. There
was agreement that this is an important area, one in which a number of researchers have worked,
but the approaches to collect information on activities and behavior, as related to children's
exposures, have been highly variable. Researchers in industry, academia, non-profits, and
government have collected data. There was discussion about forming an Activity Workgroup that
would evaluate what data are available and determine how to proceed. Lisa Melnyk was drafted
by acclimation to chair the workgroup.
Carry Croghan suggested that an effort should be made to compile available methods, SOPs, and
questionnaires from children's exposure studies. Roy Fortmann objected, considering this to be
too large of an effort that was not justified unless a clear purpose was identified for this database.
Asa Bradman discussed potential intervention studies that should be considered. These included
(a) a removal study where participants are moved to a pesticide-free home, you control for diet,
and collect urine samples to measure exposure, (b) a deep cleaning home study, (c) a fasting study,
16
-------
and (b) food elimination studies where certain classes of foods are not included in the diet. There
was not extensive discussion on this topic.
• The session was closed, having identified the following areas in which to initiate collaborations
and leads on these efforts:
1. Linda Sheldon's ILSI group to take a look at analyses of existing biomonitoring data to
look at the relationship of pesticide metabolite levels and measurements of the parent
pesticide(s) in environmental media starting with the pyrethroid pesticide metabolites,
and,
2. Lisa Melnyk to start an Activity Workgroup to study how best to quantify and measure
the effect of behavior/activity on children's exposure to pesticides.
17
-------
Appendix A
Individual Workgroup Materials
GROUP A:
In an effort to provide a safe and healthy environment for children, EPA has conducted research to provide
a scientific foundation for risk assessment and risk management to meet mandates of the FQPA. NERL has
designed and conducted several field studies to address these needs. The studies have been summarized
and presented to a panel of exposure assessors and modelers in government, industry, and regulatory fields
during a workshop held September 27 - 28, 2005. Four groups were formed to evaluate the data and
discuss additional research needed to better understand the linkages between the exposure data and risk
associated with an event. The groups were each given 3 themes with some charge questions. A summary
of Group A responses and discussion are given here.
Theme 1: Major trends in the data across children's exposure studies.
a. Identify the major sources and routes of children's exposure to the pesticides based on the data from
these and other children's exposure studies. Does this change by pesticide or class of pesticide?
b. Identify the major factors that influence children's exposures to the pesticides.
c. Are the observations in the recent NERL studies consistent with results from other studies and with
current assumptions?
Discussion
a. Identify the major sources and routes of children's exposure to the pesticides based on the data
from these and other children's exposure studies. Does this change by pesticide or class of pesticide?
All sources and routes were identified as being important depending on the event, i.e., following
application or no application. Diet was recognized as being significant based on the data from CTEPP, but
drinking water was not a concern if it came from a public water supply system. Beverages were
demonstrated to not be an important source, but the group was unwilling to completely dismiss the
collection due to possible contamination in fruit based drinks, e.g., apple juice. An important aspect that
was not completely understood was the contribution of handling of food prior to consumption and the
influence on overall dietary exposure. The small data available did not allow for much discussion on the
influence of surface-to-food and surface-to-hand-to-food to overall exposure and resulting biomarker
measurements. It was shown that low levels on hands did not account for levels on food, but, again, based
on a small number of sample results.
Non-dietary ingestion was recognized as a highly possible source of exposure. Hand-to-mouth activity
measurements were difficult to interpret. Replenishment of contamination could not be assessed with the
data presented. Influences of behavioral aspects on overall exposure were difficult to determine with the
datasets. This source is of great interest to the regulatory community since it directly affects their risk
calculations. Food handling could also fall into this category (changing term to indirect ingestion). More
information is needed in this area to better understand the contributions of non-dietary route to overall
exposure taking into consideration the extreme variability of the class of compound, application time,
activities that influence the measurement or estimation, etc. (ex. 3PBA and TCP in environmental samples
vs. pesticide exposure). Three reasons for the difficulty in determining non-dietary exposure were given
and need to be addressed to adequately understand this route: 1) the difference between extractable vs.
absorbed contaminant on the hand, 2) the difference between bioavailability of contaminant on dust vs.
food, 3) transfer efficiency unknown.
A concern raised for the ingestion route was the unknown bioavailability of the contaminant. An
experiment that was identified was to evaluate the difference in body burden between spiked food vs.
spiked matrix ingestion (dust). Also largely unknown was the transfer efficiency from the hand to the
18
-------
mouth. Many associated exposure factors come into play when trying to determine hand-to-mouth
measurements. In addition to understanding the hand wipe data, methodology issues were identified. What
solvent do you use to estimate this (methanol, saliva simulant)? How do you address the inability to
measure source strength (hand loading)?
Inhalation seemed to be adequately addressed, but the concern was whether or not it is an important route
of exposure for pesticides currently in use. Questions were raised about the adequacy of the measurement,
importance, and whether the exposure changes from particle bound to free contaminant. Importance of the
inhalation route is compound dependent, location specific (indoor vs. outdoor), and dependent on the health
concern. It was recognized that inhalation is well characterized and may not be a major contributor.
Dermal exposure can be measured well for applied levels, perhaps not for real world samples. This is the
major route of exposure after pesticide is applied, although this was not based on the data presented.
Passage through skin to blood may be affected by additives (surfactants, inerts, formulations). Producers
try to minimize dermal adsorption in formulation.
The discussions on routes and sources led to a focus on overall exposure. Low level chronic exposure and
episodic events both contribute to overall exposure. Acute exposure vs. chronic and resulting health effects
were raised as a concern. Does it vary compound to compound? We know more about acute than chronic
health effects. Acute effects are required to be listed for a new pesticide, chronic are not. Chronic
exposure may be a string of high level exposures.
To summarize all the sources and routes with their importance, a table was constructed. All routes need to
be addressed. The relative contribution differs based on the type of study and particular pesticide
addressed.
Study Type
After application
Chronic
Route of Exposure
Dietary
v+
*Non-dietary
v+
Inhalation
V
Possible (class
dependant) ex.
DDVP
Dermal
V
*need non-dietary measurement research to better quantify
V = effected route; V + = greatly affected route
The central concern that pulls all of the exposure data together to address risk is the results of the
biomonitoring. However, the link between exposure measurement and health effect is critical. This is
obtained through a better understanding of bioavailability. Dietary ingestion seems to be well accepted as
complete adsorption. Other routes are not well understood. Being able to measure a contaminant does not
necessarily mean that it's bioavailable. This relationship needs to be better understood, maybe through
PBPK modeling.
b. Identify the major factors that influence children's exposures to the pesticides
The discussion of this question is folded into the responses for question a.
19
-------
c. Are the observations in the recent NERL studies consistent with results from other studies and
with current assumptions?
In general, the trends in the NERL studies for measured absorbed dose were consistent with other
information; however, there are many studies and the data do not seem to be compiled to address absorbed
dose. The information in the summary should be presented in a more useful format so that all the
information can be digested. It was understood that each study had its own objectives and goals which may
make the combining of data difficult, especially with differing measurement methods used.
It was suggested that potential routes of exposure could be eliminated as not being a major contributor
based on planned study results and not speculation. The group questioned the worth of the effort to
measure exposure rather than trying to determine what led to the absorbed dose.
Theme 2: Additional hypotheses that should be tested using existing data?
a. Provide a list of the major hypotheses that could be tested on existing data.
b. Why are these considered to be important hypotheses?
c. How suitable are the existing data? What is still missing?
Discussion
Group A came up with three hypotheses that were worth studying with the existing data:
1. What are the impacts of sampling methods on measured dose calculation? Do you get a different answer
based on what method you use?
For example, Alcohol wipe vs. bologna for surface concentrations.
Which environmental monitoring technique has a better correlation to concurrently monitored absorbed
dose (biomarker/exposure)?
Many techniques to be evaluated (many media).
Why? So we can use the data that we have with assurance of its reliability to improve study guidelines and
to develop consistent measurement protocols.
Data suitable? Yes.
2. Which environmental media contribute significantly to absorbed dose?
Why? So we can focus on the appropriate and most effective media to monitor and to look for mitigation
approaches.
Data suitable? No.
3. Refine risk algorithms.
Data suitable? Yes.
Theme 3: Additional statistical analyses that can be performed on these data to better understand
children's exposures to chemicals at homes and daycare centers?
Provide a list of statistical analyses that could be performed on the accrued data. For each proposed
analysis, please specify
a. What data gap are you addressing?
b. What key hypothesis are you testing?
c. What result do you expect to find?
d. What is the likelihood of success, given the limitations of the data?
Please prioritize the proposed analyses.
20
-------
Discussion
This proved to be the most difficult task for Group A. Some of the discussion centered on the phenomenon
of biomarker concentrations not explained by exposure measurements. It was theorized that a problem
exists in the route/pathway algorithm; either a pathway is missed, the algorithms are off, or some exposure
scenarios are missed. If there were only a 2-fold difference between exposure and biomarker, the producer
of the pesticide would be satisfied. That is rarely the case. It was suggested that the algorithms need some
work. The assumptions may or may not be correct. For example, the current assumption for hand loading
is 100% absorption. This may not be accurate.
Time is one of the biggest factors on which to focus. Too much emphasis is placed on hours when days
would be more appropriate. Hours are too short of a time frame to analyze exposure. One bad assumption
is the linearity of the additive model. Exposure and biomarker output are not linear and treating them as
such results in more questions than answers. It appears that a more systematic way to understand the data
is needed.
After a long discussion about the value of the current data set, the group agreed on the first statistical
analysis.
1) More multivariate analyses using all available environmental measurements to describe the
relationships between the measurements and the biomarker over time. Pathway analysis is part of
this. Other suggested techniques include, ANOVA, Bayesian hierarchical analysis, and sensitivity analysis
on the current data set.
This will help us understand the relationship between the biomarker and the environmental measurements.
For each individual study, correct statistical analyses the first time is critical, and then try to put the studies
together. Pay attention to application events versus non-application. Results should be aggregated
separately for application and non-application studies. Look at logs closely. Meta-analysis across
population-based studies should be evaluated. Meta-analysis across application studies should be evaluated
separately.
Urine analysis was the most common biomarker measurement. The sampling procedure and analysis
varied by study. Understanding the best method for collection and analysis of the biomarker is essential.
Two groups can be defined for analysis; exposed and non-exposed. The ultimate goal is to get an algorithm
to give an accurate estimate of absorbed dose. Three or four exposure measurements that are good
predictors of absorbed dose would be ideal. There was much debate over creatinine correction and whether
to trust this data. Can a first void and spot urine collection be compared?
It was suggested that sensitivity analysis be performed on individual measurements to determine the kind of
distribution that can be assigned. Variability is associated with every measurement, so all have
distributions. This is a huge task to tackle. However, understanding the contribution of each source by
route will lead to better exposure estimates.
2) Statistical analysis of data set to estimate parameters that go into the mechanistic models
A large data set should be used to test SHEDS. The data could be used to validate the model and its
algorithms to test the predictions against the measured exposure; although there was no consensus on this
point.
This suggestion would give us distributions of mechanistic model parameters after previous comments are
incorporated, especially with time and linearity. More focused analysis of the data could be performed.
3) How much variability in biomarker is explained by dietary data?
Since most of the data points show ingestion as the major route of concern, Group A suggested looking
more closely at the diet. The DIYC study was specifically designed to look at exposed vs. not-exposed,
21
-------
and coordinated the timing of the biomarker with the exposure measurements. It was suggested that the
USDA study in which dietary exposure was controlled should be reviewed to determine if the trends are the
same as the data presented.
4) Normalize based on age, gender, body mass?
Time ran out during the discussion of this last recommendation for new statistical analyses. It was
determined that CTEPP may not be the best data to evaluate this idea. It was hoped that this would explain
more of the variability in the biomarker measurements. However, a danger exists in losing the specifics.
22
-------
GROUP B:
Theme 1: Major Trends in the Data
Identify the major sources and routes of children's exposure to the pesticides based on the data from these
and other children's exposure studies. Does this change by pesticide or class of pesticide?
1A Major Sources / Routes
• Diet
o There is compelling evidence that diet is one of the major sources of exposure for
pesticides
• Curl et al. (2003) reported a median total dimethyl phosphate metabolite
concentration approximately six times higher for children with conventional
diets than with organic diets.
• Lu's dietary intervention study (in press) found that the median urinary
concentrations of the specific metabolites for malathion and chlorpyrifos
decreased to the non-detect levels immediately after the introduction of organic
diets and remained non-detectable until the conventional diets were re-
introduced.
• CTEPP estimates (based on multimedia samples) found dietary to be the
dominant route for chlorpyrifos and permethrin exposures.
• CPPAES study found no detectable increase in urinary metabolite concentration
following crack and crevice application, even with air and surface levels
hundreds of times higher than levels measured in CTEPP.
• Consistent lack of correlation between usage questionnaires and measured
metabolites indirectly supports diet hypothesis.
In summary, group consensus suggested that the organic diet study was sufficiently compelling to warrant
additional investigation and validation.
o However, the group was not ready to accept that it is all dietary or that other routes can
be completely de-emphasized.
• Additional intervention studies are needed to validate findings.
• Diet does not explain elevated biomarker levels observed in high-use areas such
as Jacksonville, FL.
• Results of DIYC study suggest that "excess dietary" ingestion is a highly
important factor.
• Studies in which diet is controlled are needed to investigate the other routes.
• Great uncertainty exists in algorithms used to estimate route-specific doses.
• Formal sensitivity analysis of route-specific contributions appears to be lacking,
but remains to be clarified.
• Must account for all relevant pathways for risk assessment.
o Diet needs to be addressed properly in future planning for community-based cumulative
risk research.
In summary, although it was generally agreed that diet is a significant contributor of pesticide exposures,
the group felt that all pathways must be considered. The risk assessors strongly indicated that all pathways,
based on current assessment approaches, must be accounted for and measured inputs provided.
• Prenatal/gestational/lactational exposure as a major route of exposure for fetus/child.
o Extent of breastfeeding is culturally-dependent.
o Evidence of differences in pharmacokinetics during pregnancy.
o The group recommended more research on prenatal exposure.
o Since mother's exposure routes are different than the child's, must not limit
measurements to only those relevant to young children.
o Characterize the mother's exposure across all routes and it may be different than that of
the child.
23
-------
In summary the group agreed that prenatal measures were important for determining the cause of some
health impacts. The association shown between cord blood and both birth outcomes and developmental
milestones merits further research. In addition, the use of cord blood may circumvent difficulties related to
blood draws from children.
• Indoor Residential Pesticide Usage
o Columbia researchers found a correlation between insect infestations and pesticide use.
o Berkley researchers found that pesticide usage is highly dependent upon individual
tolerances for insects.
• They found a 60% cockroach infestation rate, but only a 50% insecticide use
rate.
o Serf-reported pesticide use may not be accurate.
o Dust/indoor air concentration ratio determined by a pesticide' s physical properties.
o Non-persistent pesticides appear to persist in the indoor environment longer than
previously believed.
o Hygiene differences between licensed and non-licensed applicators.
In summary, elucidation of indoor pesticide usage is complex and entwined with human behavior. Research
should continue to understand occupant perceptions of pests, pesticide usage, and safety.
• Lawn and Garden Residential Pesticide Usage
o Anecdotal evidence suggests that garden pesticides are used in quantities exceeding label
recommendations.
o Increasing use of pesticides on lawns for purely aesthetic reasons.
o Track-in of outdoor pesticides has been observed in previous studies.
• Residential Contamination by Agricultural Pesticides
o Many agricultural pesticides have been measured indoors, with persistent pesticides
measured more frequently.
o Classifying residential exposure to non-persistent agricultural compounds is difficult, but
is required for risk assessment purposes.
o Occupational exposures may result in the take-home of pesticide residues and subsequent
exposures for farm families.
o Berkeley researchers have measured elevated levels of the pyrethroids in dust in
agricultural communities.
o Spray drift is not seen as a major source for exposure.
• Occupational Take-Home Exposures
o The Agricultural Health Study has documented exposures to spouses.
o The Washington studies found a correlation between levels of agricultural pesticides in
automobiles and in homes.
o Compliance with hygiene interventions among field workers has been low. For example,
work clothing is often used repeatedly without washing.
In summary, occupational exposures contribute to familial exposures. There are opportunities through
educational efforts (using appropriate, verifiably effective communication techniques) to provide
interventions and reduce potential exposures to the families of the occupationally exposed.
• Pets
o Little is known about the role of pets in children's exposures.
o Few published studies are available.
o Results from the Pet Pilot Study suggest that active pets may serve as an important
vehicle for transport of turf-applied pesticides into dwellings.
In summary, the group felt that pet borne exposures might be important, but a lack of familiarity with the
data prevented the group from developing solid conclusions.
24
-------
• The "Missing Other"
o There are clearly many sources of exposure that remain unidentified or unmeasured.
o Includes pesticide use in institutional and retail settings.
o Individuals are likely to be unknowingly exposed.
o Prophylactic applications
• Performed in retail food handling establishments (e.g., markets, restaurants) for
sanitation reasons.
• Performed in daycare centers/schools for licensing reasons.
In summary, the group felt strongly that other scenario-specific exposures were worth investigating. In
order to understand total exposure it was considered important to quantify the contribution of locations
other than the home.
IB Major Factors
• Measurement Issues
o Are we measuring the right things?
• Are the correct pesticides being measured? Columbia researchers found high
levels of carbamates indoors, but these are not included in the EPA children's
studies because of low reported usage.
• We need to be careful in assuming that all important pesticide exposures are
from non-volatile pyrethroids. The semi-volatiles may not be as dead as
assumed.
• Studies show that pesticides no longer used remain at measurable concentrations
indoors.
In summary, researchers need to be aware of the breadth of pesticides present in indoor environments. The
group would like more pesticides measured in future studies.
o Are we measuring pesticide exposures correctly?
• Are we using the correct items from the "measurement toolbox?" Are the
measures adequate?
• There is a clear lack of standardization of measurement methods. Different
surface measurement methods result in greatly different loading estimates.
In summary, the group strongly felt that there was a lack of standardization among the approaches and
methods employed in human exposure studies. Unanimously it was believed the comparability between
studies would be enhanced by either standardizing approaches or showing the relationship between
different methods.
• Exposure Variability
o Intrinsic variability may not be accounted for, leading to substantial measurement error
and exposure misclassification.
o Temporal variability in environmental concentrations remains largely uncharacterized.
• Believed to be compound-dependent.
• Longer measurements or increased numbers of random measurements may be
necessary to accurately quantify exposures.
o High spatial variability was observed in focused studies (e.g., Test House). Expert
judgment is often used to select surfaces to be sampled; are we capturing the full range of
surface loadings with which children come into contact?
o We may not know where the current resolution (either temporal or spatial) is insufficient.
There is a need for additional laboratory experiments to determine the resolution
required.
o High temporal variability observed in urinary biomarker measurements.
• Spot and 24-h samples are correlated, but samples separated by 3-day intervals
are poorly correlated.
25
-------
• Temporal aspects are multi-factorial (including environmental and
pharmacokinetic).
• Environmental levels do not change as rapidly as biomarkers, pointing to
important effect of activities.
• Higher urinary excretion has been observed in post-partum period.
Need experiments on lipid storage during pregnancy.
Are pesticides being selectively released after birth; is there a higher
level of metabolism?
How do you interpret biomarker measures from pregnant females?
In summary, temporal and spatial variability merits additional considerations. The variability over space
and time may influence biomarker concentrations. In addition, standardized approaches for sampling
surfaces are required to insure that samples are representative of the actual distribution.
• Exposure Metrics
o Exposure peaks may be more important than average levels for certain important
outcomes. Episodic events can also influence exposures and may be more important for
outcomes. Exposure spikes may be critical for vulnerable and susceptible
subpopulations, such as developing fetuses.
o There is a tendency to measure blindly. We should first do back-of-the-envelope
calculations and determine if we could actually capture a measure or determine what that
measure might be (a sensitivity analysis).
In summary, the group felt that peak exposures might be more important for certain outcomes. However,
the evidence for capturing peaks is not well supported particularly when considering generalized outcomes.
Peaks might be difficult to capture and nebulous across multiple media.
• Exposure Classification: Epidemiology vs. Risk Assessment
o Differences exist between epidemiological and risk assessment needs.
o NERL's focus has been on links among exposure, uptake, and dose.
o Risk assessors want distributions.
o Risk assessment may benefit from methods used by Homeland Security.
• Classification of Exposure-related Behaviors
o Exposure variability across individuals is thought to be influenced to a large degree by
individual behaviors.
o What is the theoretical basis for the supposition that higher activity levels lead to greater
exposure?
• Fenske's GPS data of child on bicycle suggests that greater area covered
increases chances of contacting contaminated areas.
• Daycare study found a correlation between qualitative assessment of relative
activity levels and bodysuit loadings.
• Group discussed anecdotal evidence that seemed to suggest that the children
who moved the fastest accumulated the least amount of dermal loading (e.g., "a
rolling stone gathers no moss").
o We have not approached behavior systematically; have not explored other disciplines for
their experiences.
o The activity questionnaire approach may not be the best or get us where we want to go.
Categories of active and passive may not be appropriate.
o Perhaps we cannot capture exposure-related activities mechanistically (hand-to-mouth
counts).
• Instead of distilling down to touches, move to more general factors. Counting is
not practical for capturing behavior, thus some other type of metric is needed to
effectively describe behavior-dose relationships.
o Are there ways to more accurately classify behavior with respect to exposure? The
influence of personality and psychological factors on exposure needs to be explored.
26
-------
• Apply existing instalments like the "Child Behavior Check List" for appropriate
ages.
• Use existing instruments that characterize temperament or personality.
• Other factors like attention span and tactile proclivities may be important.
• Some personality factors may be linked to physical factors; for example,
hyperactive children may have higher respiration rates.
• The field of Environmental Anthropology should be consulted.
• Group discussed how this could be incorporated into scenario specific
distributions required by risk managers for assessments.
The risk assessor present clearly indicated that he had specific needs in order to estimate risk. It should be
noted that the attending risk assessor was somewhat skeptical of several concepts introduced in this section.
The introduction of individual behavior and quantification of behavior are not easily incorporated into the
risk assessment process.
• Reliance on Biomarkers
o Route-specific contributions do not seem to add up to what is excreted.
• Estimated dose exceeds the various routes.
• Are there unaccounted for sources?
• Are algorithms sufficient?
• Are assumptions accurate?
o Creatinine adjustments specific to different age groups are not available.
o There is a current emphasis on interpreting biomarker data.
• For persistent compounds we have a good feel for their meaning, though
perhaps not for kids.
• Captures general population but not the tails.
o Interpretation muddled by uncertainties in the parent-metabolite relationship.
o Why not focus on parent compounds?
• Parent compounds must be measured in blood, but getting blood from kids is
difficult.
• Sufficient volumes are difficult to obtain in order to be able to analyze for
pesticides.
• One idea is to use umbilical cord blood. Parent compound makes it through to
baby, but metabolite does not.
In summary, due to the lack of association between urinary biomarker concentrations and multimedia,
multi-route exposure measures, there was some doubt that the contribution of environmental concentrations
had been fully explored and sources properly identified. We do not know enough about creatinine excretion
in children and fetus/parent relationship to fully integrate exposure to biomarker relationships. The group
considered whether focusing on parent compound might be a better approach than the metabolite.
• Type of Applications
o Potential exposure implications of licensed applicator vs. home applications.
o The data to test this hypothesis is currently available in the large regional studies
presented as part of this workshop.
• Chemical Class
o Dust is a reservoir for pesticides that have been applied indoors or have intruded from
outdoor sources.
o Physical properties are believed to affect transfer from surfaces to skin, and may affect
route-specific contributions.
o Predominant routes of exposure are distinctly affected by the unique properties of an
individual pesticide, particularly vapor pressure.
o Exposure to carbamates has been largely ignored, but there are a few carbamate
insecticides registered for use indoors.
• Transfer efficiency
o Needs to be clarified and is likely an important factor.
27
-------
In summary, the determination of transfer efficiencies is complex and driven by multiple factors (e.g.,
surface concentration, maximal dermal loading, removal, physico-chemical properties of the active
ingredient, etc.). Additional work is required to integrate complex factors in order to derive a
meaningful value. However, it was clear that the risk assessors desired a value as part of the risk
assessment process.
• Age
o In CHAMACOS, an agricultural effect was only seen in kids less than 12 months old.
o Major sources and routes of exposure change with a child's age.
o Diet complexity changes with age.
• Culture
o Influences breast feeding rate and duration. The Berkeley researchers found no
difference in urinary biomarker levels between bottle- and breast-fed babies.
o Determines types of take-home exposures. For example, some cultures remove shoes
prior to entry.
o Diet varies markedly among some cultures.
• Pets
1C. Consistent with Current Assumptions?
• CTEPP nicely demonstrates that residues in food are important.
• The group did not identify inconsistencies with currently held assumptions.
Theme 2. Additional Hypotheses that Should be Tested
2A List of Major Hypotheses
• Does changing the resolution of sampling time affect the correlation with urine?
• Do biomarkers correlate better with peak exposures than with average exposures?
• Would changing the resolution of sampling periods result in improved correlations?
• Would averaging environmental concentrations over longer periods of time improve biomarker
correlations?
• Are there behavioral measurements, for example "temperament", that we can use to test an effect
on the transfer to the child?
• Are there alternative approaches that utilize behavioral categories that can be used instead of the
micro-environment approach, and can these be related to transfer coefficient estimates?
• Would removing the dietary contribution allow a clearer understanding of the remaining pathways
and an enhanced sensitivity analysis?
• How do results from high exposure situations compare to the overall populations' environmental
and biological levels?
• Are dermal transfer factors from the JAX study consistent with values used by OPP?
2C What is still missing?
• Scenario specific measures from food handling establishments.
• A thorough mining of questionnaire data.
• Unification/standardization/normalizations of methodologies and QA/QC procedures.
• A comparison of the studies to national trends or the larger population.
• The treatment of non-detects has not been consistent across studies.
• Aggregated/uniform data sets of larger studies.
• Parent-metabolite relationships in environmental media.
• Is urinary metabolite really the best marker? If not, what are alternative biomarkers (e.g., saliva,
blood, cord blood)?
28
-------
Theme 3: Additional Statistical Analyses
Rough Ideas:
Compare Data to Other Available Data
• Comparisons to other studies (e.g., occupational vs. non-occupational). What is the scale of
exposure compared to occupational exposure (sense of magnitude)?
• Compare studies that applied pesticides to the survey studies.
• How do real measures relate to the model predicted exposures?
Crosswalk: Environmental Concentrations to Biomarkers
• Transfer factor successful in agriculture because of fairly standardized work practices such as time
in field, duration of exposure, limited compounds, repetitive activities, etc. relative to residential
environment.
• There are many different methods employed for determining transferable residues in residential
studies, however, there are fewer and standardized approaches for agricultural assessments.
• Might not be realistic to expect to reconcile exposure and biomarkers measures.
Analysis Methods
• Use principal component analysis as a method for dealing with interdependence of environmental
measurements.
• Apply pharmacokinetic data to adjust dose estimates.
• Perform comparisons of dose with other exposure metrics, such as: amount applied, time in home,
number of family members, eating out vs. home, transportation, weather, un-mined questionnaire
data, all pair-wise comparisons.
• Examine the currently applied assumptions to investigate the disparity between environmental and
urinary concentrations (i.e., breathing rate, soil ingestion, etc.).
• Incorporate existing measurement and variability data into probabilistic models.
Prioritized Additional Statistical Analysis:
Define Expectations
• Define our expectation for correlation.
o Consult lead (Pb) research to determine what the best association that we can hope for is.
o Look at variability in children's blood lead measurements and environmental
measurements and see what is achievable.
• Hypothesis: Data we have collected is as good as it is going to get.
• Expectation: Low correlation coefficient.
• Success: Lowering expectations is likely to be highly successful.
Dietary Background Normalization
• Perform a rank adjustment of biomarker levels using data from organic vs. conventional diet
study, then correlate with environmental measurements.
o Use post-intervention values from Lu's organic diet intervention study.
o Subtract out dietary contribution.
• Hypothesis: Correlation exists between non-dietary routes and biomarkers, but dietary
contribution must first be removed.
• Results: Better correlation between biomarker and environmental measurements.
• Success: Probable.
Adjust for Temporal Lag
• Pharmacologically adjust calculated dose based on previous day's environmental measurements.
o Incorporating chemical !/2-life.
o Use pharmacokinetic derived dose.
o Incorporate temporal lag in biomarker.
29
-------
o Stratify by age.
• Hypothesis: Urine is more related to previous exposures than contemporaneous exposures.
• Expect: Better correlation (or confusion).
• Success: Indeterminate.
Apply OPP SOPs
• Develop exposure estimates based on OPP methods and compare to biomonitoring results.
• Normalize sampling techniques.
• Purpose: Testing algorithms.
• Hypothesis: Current algorithms are very health protective.
• Expect: Health protective estimates.
• Success: High likelihood.
Other Analyses
• Compare dose against other metrics and all available data (questionnaires, weather).
• Take another look at assumptions.
o Breathing rate
o Soil ingestion
• Larger "Taking stock of exposure data" by including other available datasets.
• Employ Principal Component Analysis (PCA) to deal with interdependence of environmental
measurements.
• Compare magnitude of our results to other available results (occupational, etc.).
30
-------
GROUP C:
THEME 1 - Major Trends in the data across children's exposure studies
a. Identify the major sources and routes of children's exposure to the pesticides based on the data from
these and other children's exposure studies. Does this change by pesticide or class of pesticide?
Historically, research regarding potential "personal" exposures in and around homes, including those to
children, focused on airborne pollutants. Increasingly, however, studies have focused on
microenvironments (indoors, outdoors, in-transit), sources (consumer products, combustion appliances,
outdoor sources), media (food, air, water, surface residues, soil, dust), and unique pathways that contribute
to total exposure for a given chemical. Non-dietary incidental ingestion resultant from children's mouthing
behavior, for example, has been the subject of many hypotheses and studies regarding children's exposure
to chemicals such as pesticides. However, pesticide exposure studies, including those that are the subject
of this workshop, demonstrate that direct dietary intake is potentially the most significant source of
exposure to infants and children. Dietary exposure alone, however, does not explain a significant fraction
of total exposure for some subpopulations. For example, in the case of pyrethroid insecticides, non-dietary
ingestion may be an important pathway of exposure to children; whereas, with organophosphates, the
inhalation route appears to contribute more significantly to total or aggregate exposure. Additional
research is needed to better characterize children's total (aggregate) exposure and the underlying
source/pathway attribution. Further, the unique behavioral characteristics and patterns of children (e.g.,
toddlers) require further study to discern the impact to exposures experienced in specific
microenvironments (e.g., indoor residential, outdoor residential, day care centers). Research involving
urinary biomarkers of exposure (e.g., TCPy, 3-PBA) show considerable promise in this regard, but require
concurrent measures of environmental media-specific residues to investigate the array of exposure sources
and their respective attribution to the total personal exposure levels observed via biomonitoring. It is
important to recognize that direct exposure to biomarkers (degradation products of pesticides) can occur in
the environment, suggesting that in some cases, concurrent environmental measurements of the biomarkers
themselves are important. Finally, it is important to recognize that estimation of parent-equivalent
absorbed dose levels from biomarker measurements (e.g., urinary) requires a prior understanding of parent
compound and biomarker toxicokinetics. The major sources and related factors that appear to influence
children's exposure to pesticides are outlined below.
b. Identify the major factors that influence children's exposure to pesticides
Based on the existing studies, which focused on insecticides such as OPs and pyrethroids, the following
observations can be made:
• Key sources of potential exposure to children include
o Food (dietary; note: food-form contribution analyses are critical for children) (highest
source)
o Residential and other environments
• Consumer products (microenvironments, e.g., day care centers, playgrounds,
indoor surfaces, outdoor surfaces, pets, etc.)
• Professional products (see above)
o General area applications (e.g., agricultural areas adjacent to residential or institutional
(such as schools) environments, geographic area mosquito control methods (such as
aerial application, lawn)
o Water (residue source strength and consumption depends upon the specific form,
groundwater or drinking water, tap, bottled, wells, surface; there are important chemical
and site-specific considerations that impact magnitude of potential residues in water;
generally, water-related sources appear to be associated with low exposure potential; it is
important to note that because of low residue concentrations and analytical limitations,
contribution of water-related exposures from beverage consumption is often difficult to
determine)
o Personal products (e.g., head lice shampoos)
31
-------
• Key exposure routes include
o Oral ingestion (direct; diet)
o Indirect ingestion (non-dietary)
o Dermal (very complex exposure route/pathway; requires additional investigation using
controlled studies; chemical-specific and behavioral-related factors are involved;
relationship to incidental ingestion via hand-to-mouth behavior requires concurrent
measurements of hand and environmental surface residue levels)
o Inhalation (source and chemical-specific characteristics must be considered; e.g.,
respirable particles versus vapor-phase)
o Role of dust (requires further investigation, e.g., pathways of exposure such as re-
suspension and resultant particle size distribution)
o Relative significance of each route requires further investigation
• Chemical versus behavioral determinants (intra- and inter-person variability;
source/route contribution variability for individuals)
• Sensitivity analyses
o Route-specific bioavailability is a key component (e.g., often assume 100% oral
absorption of ingested residues; for example, residues associated with dust)
• Key chemical-specific exposure factors include
o Functional group(s)
o Volatility
o Water solubility
o Degradation (e.g., hydrolysis and photolysis)
o Route-specific toxicokinetics
o Chemical-specific product use characteristics (equipment, rates, directions)
• Key general exposure factors include
o Demographics
o Product use patterns (generalized, e.g., pesticide user versus non-user status; types of
products being used; misuse patterns)
o Temporal food consumption for specific populations
• Where are the biggest uncertainties
• Food frequency
• Recipes (model evaluation opportunities)
• Processing factors and degradation products (preparation, cooking)
• Alternative analyses for non-detects; development of more sensitive analytical
methods
• Focused residue monitoring programs
o Temporal children's activity patterns
o Behavioral factors (e.g., surface and body-part-specific contact patterns and frequency;
hand loading and mouthing patterns and frequencies as a function of activity level)
o Physiological factors (e.g., body-part specific surface contact frequency, breathing zone
location, inhalation rate, etc.)
c. Are the observations in the recent NERL studies consistent with results from other studies and with
current assumptions?
• Consistency with results from other studies should be further evaluated with respect to
o Study design similarities/differences (demographics, product use information, if known,
methodologies)
o Review/confirm consistency of how existing studies address non-detect (ND) values
o Must be specific regarding statistical metrics used to determine "consistency"
(descriptive statistics, distributional characteristics)
o Comparison at central tendencies versus specified percentiles
o What to compare? (environmental media, route-specific exposure estimates, absorbed
dose via biomarkers)
o Consideration of outliers/data censoring
32
-------
o Population-based biological monitoring studies appear to be reasonably consistent (e.g.,
EPA-sponsored studies versus CDC data involving the same biomarkers); factors that
may impact comparability can include urinary collection and analysis methods, intake of
the biomarker itself, demographic differences, seasonal differences, geographic
differences, unique dietary practices such as the potential role of imported foods in the
diets of some sub-populations of children
o Differences between scenario/situation-specific short-term measurements and population-
based measurements of "steady-state", background exposures are apparent; further,
results from studies involving specific "user" sub-populations may vary significantly;
additional situational studies with specific user sub-populations recommended
• Consistency with current assumptions and children-specific exposure assessment methods (e.g.,
EPA Office of Pesticide Programs "Standard Operating Procedures" (SOPs) for Residential
Exposure Assessment and related policy documents such as "Policy 12")
o Time to revisit SOPs; coordination with EPA/ORD scientists (e.g., indirect, incidental
ingestion)
o Role of biological monitoring and guidance regarding protocols and interpretation of
biomonitoring data (e.g., spot urine/first void and relationship to 24-hr excretion; review
of historical literature/publications)
THEMES 2 and 3 - Additional hypotheses that could be tested using existing data (Theme 2) and
Additional statistical analyses that can be performed on these data to better understand children's
exposures to chemicals (Theme 3)
~ Provide a list of the major hypotheses that could be tested on the existing data.
• Major hypotheses that could be tested using existing data include the following:
o Do one or more environmental measurement parameters correlate with absorbed dose
distributions for biomarkers?
• Practical, economical measures; assists with contribution analyses; provide
measurement guidelines and protocols
• Consider non-parametric versus parametric statistics (bivariate and multivariate)
• Ongoing SEM and Bayesian analyses - consider including metabolites in SEM
and HBM analyses
o Are absorbed dose distributions significantly increased relative to "background"
following indoor crack and crevice applications?
• Needed to support regulatory safety determinations given high occurrence of the
product types/exposure scenario
• Sample size limitations; lack of pre- and post-urine samples
• Combine CTEPP and JAX?
o Cluster analyses - Environmental Measurements
• CTEPP (e.g., air, soil measurements, etc.) - Do chemicals and/or media cluster
Addresses similarities/differences regarding environmental partitioning
Addresses co-occurrence in environmental media
o Cluster analyses - Questionnaire Data
• Identification of most important factors (e.g., associated with "higher"
measurements for dietary sources)
• Consolidation of questionnaire results to increase statistical power
o What is the significance of total versus vapor-phase inhalation exposures?
o Is indirect ingestion a significant route of exposure (hand-to-mouth, object-to-mouth, dust
ingestion)? What is the magnitude of inter- and intra-individual variability? Are their
discernable characteristics associated with the magnitude of indirect ingestion?
• Sensitivity analyses with multiple "models" including concurrent measurements
of house dust, hand wipe, surface transferability and assumptions, e.g., dust
ingestion rates, hand-to-mouth frequency, surface area of hand
• Comparison to EPA SOP algorithm
33
-------
• Relative importance of exposure route; support evaluation of related factors
(surface area involved, frequency of events)
o Do individuals who consume large amounts of leafy vegetables have higher exposures to
pesticides?
• Look at food diaries
o Does accounting for direct exposure/intake of key metabolites improve modeled
estimates of absorbed dose?
• Model evaluation; relative importance of exposures and bioavailability of
biomarkers
• Multivariate stuff
o Characterize spatial distributions of surface residues in indoor residential
microenvironments (Are there spatial distributions in indoor environments?)
• JAX, Test House data sets (intra-site variability)
• Additional data sets (e.g., NDETF fogger study)
• Significance for dermal exposure estimation
o Are their regional differences in the magnitude of residential exposures and absorbed
dose distributions?
• What level of resolution is needed for regulatory agency decision-making?
• Why are these considered to be important hypotheses?
• How suitable are the existing data? Limitations?
• Limitations of existing data include
o Temporal dimension (time series)
o Age group specificity (e.g., under 3 yrs)
o Missing concurrent data elements (e.g., dietary, time-activity, product use diaries)
o Surface transferable residue measurements vary; no comparative studies to relate (e.g.,
specify as "indoor CA roller equivalents")
o Source attribution for total absorbed dose
o Data quality and accuracy evaluation (e.g., adult "shadow" reporting for children;
verification of actual product used; amount applied; nominal application rate)?
o Assumptions regarding incidental ingestion exposure
o Comparability of dust collection methods
o Measurement methods and method detection limits (MDLs) differ
o Time activity and pesticide use information limited and of questionable accuracy
• Measurements that can be excluded?
o None identified; it is noted that a study providing direct comparison of different methods
of measuring surface transferable residues is required
• Other
o Data Gap: Particle size distribution (role in inhalation exposure, role in incidental
ingestion exposure); relative importance of re-suspension as an exposure pathway;
importance of particle-bound chemicals
o Existing study videography data analyses
o Permethrin, head lice treatment incidence in JAX cohort
o How can these data be used in conjunction with videography data (e.g., do low versus
high surface contact rates correlate with dose magnitude; do low versus high hand-to-
mouth rates correlate with dose magnitude)?
• Key prerequisites for combining data (Data Warehouse) is a data dictionary
o Data dictionary for each study; common data elements that can support analyses of
combined data sets and meta analyses
• Should include selected elements of questionnaires (e.g., demographics)
• Method equivalence (e.g., transferable residue measurements)
• Analytical method comparability (e.g., MDL distributions)
• What data sets "qualify" for inclusion (e.g., TCPy)
34
-------
• Methodological approach would be applicable to analyses with future data sets
(e.g., pyrethroids)
• Descriptive statistics and bivariate and multivariate regression analyses
o e.g., Surface loading versus absorbed dose
• Comparison of inventory data collected versus Residential Exposure Joint Venture (REJV)
national survey
• Model evaluation case study recommended for permethrin
o Address multiple data sets:
• Population-based
- CTEPP, CDC, PEPCOT
• User-subpopulation
- JAX, RTI Star Grant
o Per capita probabilistic modeling
• e.g., comparison to CARES aggregate assessment for permethrin
• Permethrin biological monitoring versus probabilistic simulations for per capita
aggregate modeling (chronic doses represented via background levels observed
in the general population versus predictive estimates of LADD)
• Permethrin scenario-specific biological monitoring versus probabilistic
modeling for selected scenarios (e.g., acute, daily doses following product use
events, e.g., crack and crevice)
• Notes: the children in the JAX study had at least seven times higher levels of 3-
PBA in their urine samples than children in the CTEPP-OH study; participants
in the JAX study (n=9) were selected from homes that reported frequent
pesticide use. Participants in the CTEPP study (n=127) were selected at random.
Theme 3: Additional statistical analyses that can be performed on these data to better understand
children's exposures to chemicals at homes and daycare centers?
~ Provide a list of statistical analyses that could be performed on the accrued data. For each proposed
analysis, please specify: (a) What data gap are you addressing? (b) What key hypothesis are you testing?
(c) What result do you expect to find? (d) What is the likelihood of success, given the limitations of the
data? Please prioritize the proposed analyses.
(Note: Prerequisite to proposed analysis is completion of a data dictionary for each study data set.)
Proposed Analysis 1: Do one or more environmental measurement parameters correlate with absorbed
dose distributions for biomarkers?
• Non-parametric versus parametric statistics
• Bivariate
• Multivariate non-parametric analyses
• Comparison of inventory and use data versus REJV
• Cluster analysis (CC)
• HBM and SEM - include metabolites in SEM and HBM analyses
• Meta-analysis (analysis across the results from individual studies)
• Analysis of combined data sets (data warehouse)
• Model evaluation case studies
o Permethrin biological monitoring versus probabilistic, per capita aggregate modeling
o Permethrin scenario-specific biological monitoring versus probabilistic modeling
o Focus on 3-PBA if resources are limited
o Population-based (CTEPP, CDC, PEPCOT)
o User-subpopulation (JAX, RTI STAR grant)
Proposed Analysis 2: Are absorbed dose distributions significantly increased due to crack and crevice
applications?
35
-------
• Sample size is very small
• May be inadequate data to do this analysis
• JAX 2001 biomonitoring data and questionnaire data for 200 still need to be analyzed
Proposed Analysis 3: Cluster analysis - CTEPP
• Environmental data
• Questionnaire data -identification of most important factors
• Can you do this for different percentiles (e.g., for the highly exposed - 95th percentile versus 50th)
• Food diaries - can you identify food types that are most important?
Proposed Analysis 4: What is the significance of total versus vapor-phase inhalation exposures?
• This is a data gap, not a statistical analysis. Concern is that we may be underestimating exposure
if we don't adequately address PM resuspension.
Proposed Analysis 5: Is indirect ingestion a significant route of exposure?
• Look at the assumptions that are being used; sensitivity analyses of assumptions. What are the
assumptions that are driving it? Are intake assumptions correct? Which concentration
measurement should you use? (house dust, surface wipes, hand wipes, PUF, etc.)
• Need to do analyses of a single data set with multiple models and assumptions to provide
comparisons and uncertainty evaluations
Proposed Analysis 6: Metabolites
• Multivariate
Proposed Analysis 7: Characterize spatial distributions
36
-------
GROUP D:
Theme 1: Major trends in the data across children's exposure studies.
a. Identify the major sources and routes of children's exposure to the pesticides based on the
data from these and other children's exposure studies. Does this change by pesticide or class of
pesticide?
In the past, we thought that children were exposed to pesticides at their homes mainly through inhalation
(air) route of exposure. Today, our research has indicated that children are exposed primarily through
dietary intake (solid food) of pesticides that have been applied on agricultural crops. Other important routes
of exposure appear to be nondietary ingestion for the pyrethroids and inhalation for the organophosphates.
The research issue is that dietary intake still only explains about 60% of the children's potential doses to
pesticides in these environments. More research is needed to understand other sources (e.g., surfaces) and
routes (e.g., nondietary) of children's exposure to pesticides and how their behavior impacts their
exposures.
Research has recently shown that some urinary biomarkers of exposure (e.g., TCP, IMP, and 3-PBA) are
quite measurable in several environmental media at children's homes and day care centers. This questions
the reliability of using these urinary biomarkers of exposure to assess children's exposures to pesticides in
these environments. Published research has shown that sheep orally administered TCP easily absorbed and
excreted this metabolite unchanged in the urine. In addition, industry has conducted research (unpublished)
showing that these degradation products (e.g., TCP) are easily absorbed in the gut and are excreted
unmetabolized in the urine. This shows that degradation products can contribute to the excreted amount of
metabolites in humans. Therefore in future studies, it is essential to measure for the degradation products of
pesticides that are used as urinary biomarkers of exposure.
b. Identify the major factors that influence children's exposures to the pesticides.
Children's behavior and activity patterns most likely impact their exposures to pesticides at homes and day
care centers. These activity patterns include hand-to-mouth and object-to-mouth activity and hygiene
practices (e.g., washing hands, bathing). These activities have not been well-characterized in past studies.
More research is needed to better understand how children's behavior/activity patterns influence their
exposures to pesticides in these environments. We are wasting our time unless we can conduct exposure
studies that link pesticide applications with children's behavior/activity patterns.
c. Are the observations in the recent NERL studies consistent with results from other studies and
with current assumptions?
Generally, the NERL study results are consistent with other/previous studies.
Theme 2: Additional hypotheses that should be tested using existing data?
a. Provide a list of the major hypotheses that could be tested on existing data.
Our group felt that we needed to look at the actual data and descriptive statistics from each study before
we could provide any meaningful hypotheses.
Important research needs/comments:
• We felt that the dietary and inhalation routes of exposure have been well characterized in many of
these studies. However, other routes of exposure (i.e., non-dietary ingestion) have not been well
characterized, particularly the best inputs (e.g., soil, dust, wipes) to use in our algorithms to assess
children's exposures to these pesticides.
• We have little information on the toxicokinetics (absorption, metabolism, and excretion) of these
pesticides in humans. Are the toxicokinetics similar for classes of chemicals or do we have to
assess this by individual pesticides?
37
-------
• There is still too much uncertainty in exposure factors. For example, use of current distributions
for hand-to-mouth transfers seems to lead to very high exposure estimates.
• Do we regulate for high exposures or average exposures?
• There is still a major gap in data on transfer coefficients.
• Intra- and inter-personal variability of children's activities are important to measure.
• Group children by macroactivity levels instead of trying to describe all microactivities.
• Can creatinine levels in children be used as a proxy for activity level?
• We need to focus more on high-end exposures of children to pesticides.
• Anomaly-exposures caused by unexpected activities of children.
• Lack of specific biomarkers is an important factor limiting the use of biomarkers. Will genomics
be useful? Need to have dose-response relationship and time course information (intra- and inter-
person variability).
• We greatly need a research study similar to CHEERS conducted to better understand children's
exposures to pesticides.
• Proximity of pesticide use (temporal and spatial) is very important when assessing children's
exposure.
• Pesticide degradation (including metabolites) in the environment (time decay after application -
indoors and outdoors).
• Reconcile switch from organophosphates to less volatiles; how does this affect fate and
transportfaioavailabirity.
• Laboratory-based studies of pyrethroids versus organophosphates physico-/chemical behavior.
• Effect of vapor pressure on movement/partitioning and uptake.
Important comments:
Data from previous studies (i.e., industry) should be shared and publicly available to the scientific
community. Why fund research that has already been conducted and well characterized (i.e., toxicokinetics
and toxicodynamics of pesticides in humans/animals)? We need to form partnerships or have agreements of
understanding with industry and other similar groups.
Research interests related to the data from these studies:
• Relationship between terminal and absorbed dose in the body.
• Review and comparison of dosimeter data and types. Dosimeters versus hand wipes: how do they
reflect actual loading?
• Laboratory-based evaluations compared with field data.
• Are the Jacksonville results higher than CTEPP because of non-dietary exposure?
• Duplicate diet data compared to standard food diaries.
• SHEDS vs. CARES (or other similar models). Compare estimates using actual data input into
these two models. Are they similar or different?
• Perform meta-analysis of these data - any trends?
• Re-analyze the data and look for associations with an emphasis on regulating the higher end
exposures.
• Stratify existing data by application method (crack and crevice vs. broadcast) and test for
differences.
• Group data by quantiles.
• Compare regulatory estimates vs. data from these studies.
Important comments:
• Can data be combined across studies to better characterize behavior and pesticide applications?
• Can information from other studies (external) be considered and included?
• Can information from studies using different methodologies be compared and included?
• Can studies using different recruitment methods be combined? Are the populations comparable?
• Can differences in dust concentrations, loading etc. be explained by physical chemical properties
(e.g. Henry's law)?
• Need for standardization of methods.
38
-------
• Lack of uniformity across studies.
• Do not over interpret existing data.
• Need better interaction between researchers and regulators to define hypotheses.
d. Why are these considered to be important research questions important?
We are curious, i.e., these areas have the greatest uncertainties in the distribution estimates.
39
-------
Appendix B
Participant List
Participant
Amy Herring
Andrew Geller
Asa Bradman
Beth Doyle
Bob Krieger
Bob Lordo
Carry Croghan
Chris Wible
Curt Lunchick
Dan Stout
Dana Ban-
Dana Vogel
David Miller
Debbie Bennett
Don Whitaker
Elaine Cohen Hubal
ElinUlrich
Frank Selman
Haluk Ozkaynak
Jacqueline Moya
Jane Chuang
Janice Sharp
Jeff Driver
Jeff Evans
Affiliation
UNC
EPA/NHEERL
UC Berkeley
EPA/OW
UC Riverside
Battelle
EPA/NERL
Scotts
Bayer CropScience
EPA/NERL
CDC
EPA/OPP
EPA/OPP
UC Davis
EPA/NERL
EPA/COMPTOX
EPA/NERL
Dow Agro Sciences
EPA/NERL
EPA/NCEA
Battelle
Valent Biosciences
Infoscientific.com
EPA/OPP
Contact Information
Email: amy herring(@,unc.edu
Phone: 919-843-6368
Email: seller. andrew@epa. gov
Phone: 919-541-4208
Email: abradman(@,socrates.berkeley.edu
Phone:510-643-3023
Email: doyle.elizabeth(@,epa.gov
Phone: 202-566-0056
Email: bob .krie ger@ucr.edu
Phone: 951-827-3724
Email: lordor(@,battelle.org
Phone:614-424-4516
Email: croghan. carry(@,epa. gov
Phone:919-541-3184
Email: chris.wible(@,scottsco.com
Phone: 937-644-7012
Email:
curt.lunchick(S>bavercropscience.com
Phone: 919-549-2986
Email: stout. dan(@,epa. gov
Phone: 919-541-5767
Email: dlbl(@,cdc.gov
Phone: 770-488-7886
Email: vo gel . dana@,epa. gov
Phone: 703-305-0874
Email: miller.davidj(@,epa. gov
Phone: 703-305-5352
Email: dhbennett@,ucdavis.edu
Phone: 530-754-8282
Email: whitaker.donald(@,epa. gov
Phone: 919-541-1571
Email: hubal.elaine(@,epa. gov
Phone: 919-541-4077
Email: ulrich.elin@,epa. gov
Phone: 919-541-3717
Email: fbselman(@,dow.com
Phone:317-337-5108
Email: ozkaynak.haluktgiepa.gov
Phone: 919-541-5172
Email: moya.j acqueline(g),epa. gov
Phone: 202-564-3245
Email: chuangj @,battelle. org
Phone: 614-424-5222
Email: Janice.sharp(@valent.com
Phone: 847-968-4844
Email: ddriver448(@,aol.com
Phone: 703-361-8449
Email: evans.ieff(@,epa.sov
40
-------
Participant
Jeff Morgan
Jim Olson
Jim Raymer
Jim Stan-
Jim Xue
John Kissel
John Ross
Linda Birnbaum
Linda Sheldon
Lisa Melnyk
Lynn Delpire
Maria Morandi
Mark Strynar
Marsha Morgan
MaryKay O'Rourke
Nancy McMillan
Nicolle Tulve
Paul Jones
Paul Lioy
Peter Egeghy
Philip Villanueva
Richard Fenske
Robin Whyatt
Ross Highsmith
Roy Fortmann
Affiliation
EPA/NERL
SC Johnson
RTI
EPA/NERL
EPA/NERL
U of Washington
Infoscientific.com
EPA/NHEERL
EPA/NERL
EPA/NERL
EPA/OPPT
U Texas at Houston
EPA/NERL
EPA/NERL
U of Arizona
Battelle
EPA/NERL
EPA/NERL
EOHSI
EPA/NERL
EPA/OPP
U of Washington
Columbia U
EPA/NERL
EPA/NERL
Contact Information
Phone: 703-305-7877
Email: morgan.j effrev(@,epa. gov
Phone:513-569-7738
Email: jjolson(@,scj.com
Phone: 262-260-2768
Email: jraymerfg.rti.org
Phone: 919-541-5924
Email: starr. j ames(@,epa. gov
Phone: 919-541-4608
Email: xue .j ianpingtgiepa. gov
Phone: 919-541-7962
Email: jkissel(@,u. washington.edu
Phone:206-543-5111
Email: john(@,infoscientific.com
Phone: 916-486-6100
Email: birnbaum. lindatgiepa. gov
Phone: 919-541-2655
Email: sheldon. lindatgiepa. gov
Phone: 919-541-2205
Email: melnyk.lisa(@,epa. gov
Phone:513-569-7494
Email: delpire.lynn(@,epa. gov
Phone:202-564-8531
Email: maria.t.morandi(@,uth. tmc.edu
Phone: 713-500-9288
Email: strynar.mark(g),epa.gov
Phone: 919-541-3706
Email: morgan. marshafg.epa. gov
Phone: 919-541-2598
Email: mkor(g),email.arizona.edu
Phone: 520-626-6835
Email: mcmillann(g),battelle.org
Phone: 614-424-7942
Email: tulve.nicolle(@,epa. gov
Phone: 919-541-1077
Email: jones.paul-atgiepa.gov
Phone:919-541-1516
Email: prioy(@,eohsi.rutgers.edu
Phone: 732-445-0150
Email: egeghy.peter(@.epa.gov
Phone: 919-541-4103
Email: villanueva.philip(@,epa. gov
Phone: 703-308-8665
Email: rfenske@.u. Washington. edu
Phone: 206-543-0916
Email: rmw5 @,columbia. edu
Phone: 646-459-9609
Email: highsmith.ross(@,epa. gov
Phone: 919-541-7828
Email: fortmann.royfg.epa.gov
Phone: 919-541-1021
41
-------
Participant
Ryan Williams
Sarah Winfield
Susan Schober
Tom McCurdy
Zaida Figueroa
Affiliation
Clorox
EPA/OPP
CDC
EPA/NERL
EPA/OPP
Contact Information
Email: rvan.williams(@,clorox.com
Phone: 925-425-4338
Email: winfield. sarah(@,epa. gov
Phone: 703-305-7016
Email: sschober(@,cdc.gov
Phone: 301-458-4484
Email: mccurdv .thomas(@,epa. gov
Phone: 919-541-0782
Email: figueroa.zaidatg.epa. gov
Phone: 703-308-0015
42
-------
vvEPA
United States
Environmental Protection
Agency
Office of Research
and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
EPA 600/R-06/026
March 2006
www.epa.gov
Please make all necessary changes on the below label,
detach or copy, and return to the address in the upper left-
hand corner.
If you do not wish to receive these reports CHECK HERE I I:
detach, or copy this cover, and return to the address in the
upper left-hand corner.
PRESORTED
STANDARD
POSTAGE & FEES
PAID
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
------- |