0 EDA United States	^ ^2JIJ Office of
v>Cnr\ Environmental Protection Agency	-J? ¦ >? Research and Development
The Children's Total Exposure to Persistent
Pesticides and Other Persistent Organic
Pollutants (CTEPP) Study
Study Design
Nancy K. Wilson, Principal Investigator;
Gary Evans, Robert G. Lewis,
Thomas R. McCurdy, Elaine Cohen-Hubal,
Maurice Berry, Jim Quackenboss, and Carvin Stevens
U.S. Environmental Protection Agency
Office of Research and Development
Human Exposure & Atmospheric Sciences Division
Exposure Measurements & Analysis Branch
Notice: The U.S. Environmental Protection Agency (EPA), through its Office of Research and Development (ORD), partially funded
and collaborated in the research described here. This protocol is part of the Quality Systems Implementation Plan (QSIP)
that was reviewed by the EPA and approved for use in this demonstration/scoping study. Mention of trade names or
commercial products does not constitute endorsement or recommendation by EPA for use.

-------
Design8-r2d2.wpd
February 24, 2000
Update May 22, 2001
The Children's Total Exposure to Persistent Pesticides and Other
Persistent Organic Pollutants (CTEPP) Study
STUDY DESIGN
Nancy K. Wilson, Principal Investigator;
Gary Evans, Robert G. Lewis,
Thomas R. McCurdy, Elaine Cohen-Hubal,
Maurice Berry, Jim Quackenboss, and Carvin Stevens
Abstract
The research study, "Children's Total Exposure to Persistent Pesticides and Other
Persistent Organic Pollutants," (CTEPP) is a pilot-scale project involving about 260 children,
which investigates the possible exposures that young children may have to common
contaminants in their everyday surroundings. These contaminants include several pesticides,
phenols, polychlorinated biphenyls, polycyclic aromatic hydrocarbons, some of which are
suspected of being endocrine disrupters. The targeted compounds are persistent in the indoor and
sometimes the outdoor environments, so that very low levels may exist in the children's
surrounding microenvironments and provide a source of chronic, non-acute exposure. The
primary purposes of the research are to increase our understanding of children's exposures to
persistent pollutants, to gain information on the various activities, environmental media, and
pollutant characteristics that may influence children's exposures, and to generate further
questions and hypotheses for future research.
Young children, especially those of the preschool ages 1-5, are hypothesized to have
greater exposures than do older children or adults to persistent organic pesticides and other
persistent organic pollutants, including some compounds that may have endocrine-disrupting
effects or developmental toxicity. These greater exposures may result from what children eat and
drink, where they spend their time, and what they do there. The impact of the exposures may be
greater on young children because of their smaller body masses, immature body systems, and
rapid physical development. Very young children learn about their environment by exploring not
only the appearance and texture of objects, but also their taste and smell. Thus nondietary
ingestion can play an important role in their exposures.
The Food Quality and Protection Act of 1996 (FQPA) sets new, more stringent standards
for pesticide residues in foods, and provides increased emphasis on health protection for infants
and children. The exposure component of the risk assessment for pesticides is now required to
•	consider the susceptibility of children to increased exposure, and
•	account for aggregate exposures to the pesticides from all sources, including food,
drinking water, and non-occupational applications of the pesticides in homes,
schools, daycare centers, and other microenvironments.

-------
Essentially, the FQPA states that exposure assessments must be conducted for infants and
children and that these exposure assessments must include and be reliable for all sources of
pesticide exposure. However, very little information on children's aggregate exposures is
available at the present time, the methods for obtaining this information need improvement, and
the pathways and media through which such exposures may take place are known uncertainly.
Thus, the CTEPP study has direct practical utility to FQPA. It will provide data on
aggregate chronic, sub-acute pesticide exposures and pathways for approximately 260 children in
several microenvironments, improve the methods for determining their exposures and pathways,
and allow generation of hypotheses for further research . The objectives of CTEPP are thus
twofold: (1) To measure the total exposures at sub-acute levels of a small set of preschool
children in several NC and OH counties to a suite of persistent pesticides and other persistent
organic pollutants that they may encounter in their everyday environments, and (2) To apportion
the exposure pathways and to identify and formulate the important hypotheses to be tested in
future research. Therefore, CTEPP investigates the total exposures to persistent organic
compounds in the environment of a group of pre-elementary school children through the
ingestion, inhalation, and dermal absorption pathways, in several non-occupational settings,
through multiple environmental media. Targeted organic chemical pollutants include polycyclic
aromatic hydrocarbons; chlorinated, carbamate, triazine, pyrethroid, and organophosphate
pesticides; phthalate esters; phenols; and polychlorinated biphenyls. The specific compounds
were selected because they may be carcinogenic, mutagenic, acutely or chronically toxic, or
possibly disruptive to the human endocrine system; and because they are widespread and often
persistent in the indoor or outdoor environment.
Children who stay at home with an adult caregiver and children who attend preschool or
day care are included in the study. Emphasis is on the younger children aged 18 months to
4 years. Exposures of the children and their primary adult caregivers living in the same
household are estimated through the collection and analysis of samples of food, beverages, and
drinking water; indoor and outdoor air; hand wipes; house dust, classroom dust, and play area
soil; and smooth floor and food preparation surface wipes. Urine samples are also collected for
analysis for biomarkers of exposure. Children who are not able to provide one or more spot urine
samples during the day (who are not at least partially toilet-trained) and children who are still
being breast-fed are excluded. Information about the children's activities during the sampling
period is collected via activity diaries and food diaries. Approximately 10% of the children are
videotaped for 3-4 hr periods during the sampling to supplement and validate the activity diaries
and observations. The range of exposures through multiple environmental pathways and media
is estimated. Potential external doses are determined through a combination of micro-
environmental measurements and time-activity diaries; and insofar as is possible, effective doses
are estimated through the analysis of urinary biomarkers. Sample collection in the targeted NC
and OH counties will extend over a two-to-three year period.

-------
Outline
III.	Introduction
IV.	Objectives
A.	Long-Range Objectives
B.	Clients
C.	Hypotheses
D.	Specific Objectives of the Study
V.	Approach
A.	Sample Size
B.	Sampling Plan
C.	Sampling, Sampled Media, and Target Chemicals
D.	Analysis Methods
E.	Supporting Information
F.	Human Subjects Approval
G.	Information Collection Request
H.	Quality Assurance
I.	Statistical Analysis of Data
J. Data Analysis and Model Evaluation and Refinement
K. Initial Calculation of Exposure, Potential Daily Intake and Potential Daily Dose
L. Peer Review
M. Milestones
VI.	Personnel and Implementation
VII.	Funds
VIII.	Selected Outputs
IX.	References

-------
Tables
I. Minimum Sample Size to Detect Various Percent Differences in B2 PAH Potential Dose
II Minimal Sample Sizes (Children/Group) Required to Detect a Difference in Persistent
Organic Pollutant Exposures Between Two Groups of Children Based on a Two-Sample
t-Test Conducted at the 5% Significance Level
III. Numbers of Households in Each Stratification Category
IV Targeted Chemicals
V. Analysis Methods for the Targeted Chemicals
Figures
1.	Daily Potential Dose of B2 PAH from Food, Dust, and Air, ng/kg
2.	Daily Potential Intake (|!g/day) of some Persistent Organic Pollutants through Multiple
Pathways while at Day Care
3.	Overview of Sampling Plan for CTEPP Study
Appendices
Appendix A: Data Analysis Plan

-------
I. Introduction
Young children, especially those of the preschool ages 1-5, are hypothesized to have
greater exposures to persistent organic pesticides and other persistent organic pollutants than are
older children or adults. These greater exposures may result from what children eat and drink,
where they spend their time, and what they do there. The impact of the exposures on children's
health may be greater than that on adults' health, as a result of children's smaller body masses,
immature body systems, and rapid physical development [PI, VI], Very young children learn
about their environment by exploring not only the appearance and texture of objects, but also
their taste and smell. Thus non-dietary ingestion can play an important role in their exposures.
The Food Quality and Protection Act of 1996 (FQPA) sets new, more stringent standards
for pesticide residues in foods, and provides increased emphasis on health protection for infants
and children. The exposure component of the risk assessment for pesticides is now required to
•	consider the susceptibility of children to increased exposure, and
•	account for aggregate exposures to the pesticides from all sources, including food,
drinking water, and non-occupational applications of the pesticides in homes,
schools, daycare centers, and other microenvironments.
Essentially, the FQPA states that exposure assessments must be conducted for infants and
children and that these exposure assessments must include and be reliable for all sources of
pesticide exposure. However, very little information on children's aggregate exposures is
available at the present time, the methods for obtaining this information need improvement, and
the pathways and media through which such exposures may take place are known uncertainly.
Thus, the CTEPP study has direct practical utility to FQPA. It will provide data on
aggregate chronic, sub-acute pesticide exposures and pathways for approximately 260 children in
several microenvironments, improve the methods for determining their exposures and pathways,
and allow generation of hypotheses for further research. It will allow improvement of the
methods used to obtain physical exposure data, and it will facilitate the identification of the
important exposure pathways. The objectives of CTEPP are thus twofold: (1) To measure the
total exposures at sub-acute levels of a small set of preschool children in several NC and OH
counties to a suite of persistent pesticides and other persistent organic pollutants that they may
encounter in their everyday environments, and (2) To apportion the exposure pathways and to
identify and formulate the important hypotheses to be tested in future research. Therefore,
CTEPP investigates the total exposures to persistent organic compounds in the environment of a
group of pre-elementary school children through the ingestion, inhalation, and dermal absorption
pathways, in several non-occupational settings, through multiple environmental media. Targeted
organic chemical pollutants include polycyclic aromatic hydrocarbons; chlorinated, carbamate,
triazine, pyrethroid, and organophosphate pesticides; phthalate esters; phenols; and
polychlorinated biphenyls. The specific compounds were selected because they maybe
carcinogenic, mutagenic, acutely or chronically toxic, or possibly disruptive to the human
endocrine system; and because they are widespread and often persistent in the indoor or outdoor
environment.
Although studies of young children's exposures to various environmental pollutants have
been done in the past, these have been confined largely to studies involving one specific
pollutant, for example lead, one environmental source, for example, environmental tobacco
smoke (ETS), or to studies involving one pathway or route of exposure, for example, inhalation
[01, 02, Rl, W9]. A few small studies have evaluated methods for sampling and analysis to be
used in exposure studies, for example, the Household Infant Pesticide Exposure Study (HIPES)

-------
[LI]. To allow management and reduction of the possible risk from a given pollutant or
pollutant class, however, it is necessary to know the total exposure to individuals from all
environmental media with which they come in contact, through all pathways. This total exposure
estimate can then be used to derive an estimate of the potential intake or applied dose of the
pollutant, which can in turn be used to estimate the potential health impact on the individual. The
pioneering research in the application of such Total Exposure Assessment Methodology (TEAM)
was done by Wallace and coworkers in the late 1980s [W9]. The first major TEAM study, of
volatile organic compound (VOC) exposures, was completed in 1985. A large prospective study
of the exposures of farmers and farm families to agricultural pesticides - the Agricultural Health
Study (AHS) - is ongoing, but its focus is not primarily on children, nor does it include families
of nonfarmers [Al], The Minnesota Study [Ul], which is a component of the National Human
Exposure Assessment Survey (NHEXAS) pilot studies, focuses on children's exposures, but is
limited to households with recent pesticide applications and to children ages 3-12. Additionally,
the Minnesota study measures total exposures in only 50 households, and to only four target
pesticides: chlorpyrifos, diazinon, malathion, and atrazine. It therefore does not include the
more persistent pesticides or other persistent industrial chemicals that have been used in the past,
such as DDT or PCBs.
However, there are many questionnaire and epidemiologically based studies of children's
exposures reported in the literature. Several recent studies have implicated pesticide exposures
and exposures to other xenobiotics, for example potential endocrine disrupters, as possible
causes of children's health problems [B2, Dl, D2, D3, D4, H2, Kl, K2, K3, L2, L3, M3, M4,
07, P2, P3, R2, R4, R6, S7, Tl, W10, Zl, Z2], and several have estimated the exposures of small
numbers of children to specific pesticides [Bl, B6, El, HI, L4, L5, N4, S2, W8].
Over the past four years, Wilson and Chuang, in a series of small methodology studies,
have used the TEAM approach to examine the total exposures of preschool children in low-
income families to polycyclic aromatic hydrocarbons [C7, CIO, C12, W5]. On the basis of their
findings, these investigators extended the research to a small study of total exposures of
preschool children who attend day care centers to an extended list of target persistent chemicals
[Wl, W2, W6, W7]. The results of these studies are provocative in that they suggest that young
children's exposures to some persistent pollutants may be greater than those of adults who
inhabit the same microenvironments, especially when the potential dose, which takes into
account a child's body mass, is considered. Additionally, for these sensitive young persons, both
dietary and nondietary ingestion appear to be significant pathways for exposure to some
compounds. These findings are exemplified in Figures 1 and 2. Figure 1 shows the greater
potential dose of polycyclic aromatic hydrocarbons for children than for adults in the same
household in a nine-home study [W5]. Figure 2 shows the relative importance of exposures
through three major pathways for several classes of persistent chemicals measured in several care
centers [Wl, W2, W6, W7]. Because of the very small size of these studies (nine participants or
fewer) , however, a larger study is needed to confirm these findings with greater confidence.
The survey, recruitment, sampling, and analysis methods, and the associated activity logs
and questionnaires that were developed or refined in the above small methodology studies of
preschool children's exposures; and the findings of those studies are the initial basis for this
expanded pilot study [CI, C2, C3, C8, C9, CI 1, C14, Gl, H3, Ml, Nl, N5, N7, SI, T2, W12],
The CTEPP study is intended to apply the methods already developed as much as is
possible to field measurements of exposures. However, because of the high unit cost of
measuring the total exposures of a large number of children, it is possible to include only about
260 children. For this pilot study, a two-frame sampling plan has been developed, which will

-------
allow random sampling of preschool children who attend day care centers and who stay at home.
This plan, shown schematically in Figure 3, will produce data that are representative of children
in six counties in each of two states, North Carolina and Ohio. However, it is important to note
that the results of this pilot study will apply only to the study population of 260 children, and no
inferences to larger populations can be made.
The field study is an investigation of the total exposures to persistent organic compounds
in the environment of pre-elementary school children through the ingestion, inhalation, and
dermal absorption pathways. Targeted organic chemical classes include polycyclic aromatic
hydrocarbons; chlorinated, carbamate, triazine, pyrethroid, and organophosphate pesticides;
phthalate esters; phenols; and polychlorinated biphenyls. The specific compounds were selected
because they are carcinogenic, mutagenic, acutely or chronically toxic, or potentially disruptive
to the human endocrine system [CI5, CI6]; and because they are persistent and often ubiquitous
in the environment. Although some of the pesticides, for example the organophosphates, are not
generally thought to be persistent, in the absence of sunlight and moisture - in the indoor
environment - they degrade slowly if at all, and hence may pose an exposure risk for many years.
Children who stay at home with an adult caregiver and children who attend preschool or
day care are included in the study. Emphasis is on the younger children aged 18 months to
4 years. Exposures of the children and their primary adult caregivers living in the same
household are estimated through the collection and analysis of samples of food, beverages, and
drinking water; indoor and outdoor air; hand and forearm wipes; house dust, classroom dust, and
play area soil; and smooth floor and food preparation surface wipes. Urine samples are also
collected for analysis for biomarkers of exposure. Children who are not able to provide one or
more spot urine samples during the day (who are not at least partially toilet-trained) and children
who are still being breast-fed are excluded. Information about the children's activities during the
sampling period is collected via activity diaries and food diaries. Approximately 10% of the
children are videotaped for 3-4 hr periods during the sampling to supplement and validate the
activity diaries and observations. The range of exposures through multiple environmental
pathways and media is estimated. Potential external doses are determined through a combination
of microenvironmental measurements and time-activity diaries; and insofar as is possible,
effective doses are estimated through the analysis of urinary biomarkers. Sample collection in
the targeted NC and OH counties will extend over a two-to-three year period.
The expected benefits include a greater understanding of children's total exposures to
persistent pesticides, possible endocrine disrupters, and similar pollutants; improved knowledge
of the environmental pathways that are most important in young children's exposures; and
generation of hypotheses for further research on children's exposures.
CTEPP has direct practical utility in meeting the requirements of the Food Quality and
Protection Act (FQPA). The FQPA establishes a more stringent health-based standard for
pesticide residues on foods and provides increased emphasis on health protection for infants and
children. FQPA allows EPA to establish, modify, leave in effect, or revoke a tolerance (the legal
limit for a pesticide chemical residue in or on a food) only if it is determined to be "safe." Safe is
defined to mean that "there is a reasonable certainty that no harm will result from aggregate
exposures to the pesticide." Risk assessments, including both hazard and exposure assessments,
are required to establish "safe." For FQPA the exposure component of the risk assessment is
required to
• consider the susceptibility of children to increased exposure, and

-------
•	account for aggregate exposures of the pesticide from all sources including food,
drinking water, and non-occupational applications of the pesticides in homes,
schools, daycare centers, and other microenvironments.
As a part of this process, an additional tenfold margin of safety "shall be applied to account. . .
for completeness of the data with respect to exposure." Essentially, the act states that exposure
assessments must be conducted for infants and children and that these exposure assessments
must be reliable for all sources of pesticide exposure.
The critical exposure data needed with regard to FQPA are given in the EPA draft policy
paper, Exposure Data Requirements for Assessing Risks from Pesticide Exposure of Children
(May, 1999). This document defines the components of a complete data set; it also provides
criteria with respect to reliability of the data. The document then describes why the elements of a
complete and reliable data set are currently not available. Critical elements that are missing are
presented in the document. They were also discussed at a meeting of the EPA Science Advisory
Panel on June 25, 1999. Based on both of these sources, the greatest uncertainties in the
assessments are associated with exposures to pesticides from non-occupational applications in
homes, schools, and daycare centers. This is especially true for assessments for very young
children. Critical data gaps are associated with
•	location and activity patterns for children,
•	pesticide use in microenvironments where children spend their time,
•	pesticide distributions in these microenvironments
•	factors for estimating exposure from microenvironmental concentrations (that is,
hand-to-surface and hand-to-mouth contacts, transfer coefficients for various
contacts, time spent in microenvironments, and surface area contacted). For some
scenarios, exposure factors are based on data for fewer than 10 children in a single
age group or from a single set of laboratory experiments.

-------
II. Objectives
A.	Long-Range Objectives
The general objective of this research is to support the mission of the National Exposure
Research Laboratory (NERL) to characterize, predict, and diagnose human exposure [N2],
Within this framework, the CTEPP study has two major objectives:
(1)	To measure the total exposures of a small set of preschool children, approximately
260 children total, in several NC and OH counties, to sub-acute levels of a suite of
persistent pesticides and other persistent organic pollutants that they may
encounter in their everyday environments, and
(2)	To apportion the exposure pathways and identify the important hypotheses to be
tested in future research.
The long-range objectives of this research are thus responsive directly to FQPA
requirements, to EPA goals as expressed in the NERL research strategy [N6], which defines
several long-term goals related to the Strategic Plan for the Office of Research and Development
[03] and those defined in the EPA Children's Risk Strategy [04],
Furthermore, this research supports directly the Government Performance and Results
Act (GPRA) goals for EPA in the following way:
• ORD Science Sub-Objective #2.2: The research will provide improved tools and
data for more quantitative human health risk assessments. It will enhance the
scientific basis for identification, characterization, and assessment of exposures
that pose the greatest health risks. It will provide information on the exposures of
a group of individuals chosen from a susceptible population, preschool children,
to persistent organic pollutants in the environment.
Annual performance goals to which this research is relevant include Goal 194, Provide
Exposure and Effects Methodologies; and Goal 837, Initiate Field Exposure Study of Children to
Two EDCs.
B.	Clients
Clients for the results of this research include the EPA Office of Children's Health
Protection, and the EPA Office of Prevention, Pesticides, and Toxic Substances. Additionally,
the results will complement related research funded by the EPA Office of Research and
Development under its Science to Achieve Results (STAR) grants program, the National Institute
of Environmental Health Sciences, the EPA National Health and Ecological Effects Research
Laboratory (NHEERL), and the Consumer Product Safety Commission (CPSC), the scientific
community, and the public, particularly parents and caregivers of young children. Some of the
specific ongoing projects to which CTEPP is complementary are, for example, at NIEHS, Dr.
Matt Longnecker is looking at persistent organochlorine pesticides and phthalate esters in
mothers' blood and the relationship with children's health outcomes; Dr. Jane Hoppin is also
examining phthalate esters and their potential effects on the health of young children. The
Environmental and Occupational Health Institute at Rutgers University is conducting a nine-
subject study of exposures of children to chlorpyrifos after crack and crevice application,
"Children's Post-Application Pesticide Pilot Study." Current STAR grants include, among
others, "Assessing levels of organophosphate insecticides, which could expose children from
pets treated with flea control insecticides," Dr. Janice Chambers et al.; "Exposure of children to
pesticide in Yuma County, Arizona," Dr. Mary Kay O'Rourke et al.; and "A study of pesticide
exposures in Minnesota children," Dr. Edo Pellizari et al. (Organophosphates only) [Ul].

-------
Most importantly, the enhanced knowledge of children's total exposures and the
improved exposure measurement methodologies afforded by CTEPP will benefit many young
children in addition to the study population.
C.	Hypotheses
The main objectives of the CTEPP study, as discussed above, are: (1) To measure the
total exposures of a small set of preschool children in several NC and OH counties to chronic,
sub-acute levels of a suite of persistent pesticides and other persistent organic pollutants that they
may encounter in their everyday environments, and (2) To apportion the exposure pathways and
identify the important hypotheses to be tested in future research. There are several hypotheses
that can be tested within this group of children using the CTEPP data.
Under main objective (1), total exposure measurement, it is possible to ask the questions:
•	Are the targeted children's exposures at home and at day care/preschool equally
important?
•	Are exposures approximately the same or significantly different for the targeted
children in low-income households compared with the targeted children in
middle/upper-income households?
•	Are exposures approximately the same or significantly different for the targeted
urban and rural children?
Under main objective (2), apportionment of exposure pathways, it is possible to ask:
•	Are the exposure pathways and their relative importance different for the different
chemical classes of persistent pesticides and other persistent organic pollutants?
•	Is the ingestion pathway a major pathway for exposure of the targeted adults and
preschool children living in the same household?
•	Is diet the major contributing factor to the ingestion exposure of this group of
children and
•	In the sample population, are children's exposures to the targeted pollutants
approximately the same or significantly greater than those of the adults living in
the same household?
As mentioned previously, these hypotheses can be tested for the subpopulation of
approximately 260 children, who reside in the selected counties, and who are the focus of the
CTEPP study. The results will not be generalizable to larger populations of children, for
example, they will not be generalizable to "all children in NC or in OH (or in the US, for that
matter)," or to "all children in low-income and middle-income families," or to "all day care
centers," and so on. Neither can they be used to test such hypotheses as "are exposures of NC
children the same as or different from those of OH children?"
D.	Specific Objectives of This Study
The specific objectives of this study are thus as follows:
•	Estimate the total exposures of a group of young, pre-school children to selected
persistent organic pesticides and other persistent organic pollutants via all three
environmental pathways: ingestion, inhalation, and dermal absorption. Include all
environmental media that are likely to provide the opportunity for significant
exposures. Through the use of collected data on activity patterns, environmental
and biological measurements, and appropriate exposure models, estimate the
exposure and potential dose of each of the target chemicals. Compare the

-------
potential doses with actual, biological doses estimated from urinary biomarkers of
exposure.
•	Obtain comparisons, by stratification of the sampling, between children who stay
at home and who attend pre-school or day care, of children in low-income and in
middle/upper-income families, and of children in urban and in rural environments.
Compare actual dose estimates, obtained from urinary biomarkers measurements,
for adults and children in the same households, with potential dose estimates
obtained from exposure measurements. Include children in six counties in each
of two states: North Carolina and Ohio.
•	Apportion the pathways of exposure for the various target chemicals included in
this study, using actual, CTEPP-measured, multimedia concentration data and
observational and questionnaire data. Identify those microenvironmental media
and activities that contribute most to the total exposures.

-------
III. Approach
A. Sample Size
A preliminary estimate of the number of children and households to be sampled was
derived from the results of the PAH studies described earlier [W5, W7, C7, CIO, C12], For
children in 24 homes, the mean potential dose of B2 PAH (PAH that are probable human
carcinogens) was 34 ng/kg-day, with a standard deviation of 24 ng/kg-day. To test whether the
potential dose is the same for children in low- and middle/upper income families, a two-sample t
test was used, at the 5% significance level. The minimum sample sizes necessary to detect
differences in various potential doses of B2 PAH at two statistical powers are shown in Table I.
At 80% power, with a mean difference in the potential dose of 25%, approximately 126
participants are needed in each of the two groups being compared. An improved estimate of the
sample size necessary, based on the results of the day care study [W2, W6, W7, Wl], which
includes most of the target chemicals in this study led to the results shown in Table II, described
in the following paragraphs.
To support the sample size calculations, data on children's exposures to persistent organic
pollutants (POP) from the PAH studies [C7,C10,C12,W5,W7] and the two day care studies [Wl,
W2, W6, W7, C13] were reviewed. The review assessed the distribution and variability of POP
concentrations in house dust, indoor and outdoor air, solid and liquid food, and soil for seven
target POP, namely: B2 PAH, organophosphates, phthalate esters, phenols, diazinon,
chlorpyrifos, and bisphenol-A. In addition, the distribution of hydroxy-PAH in urine was
reviewed. It was found that
(1)	POP concentrations tend to be lognormally distributed.
(2)	Although there are differences in the variability of POP concentrations among the
six media and in urine, the standard deviations of log-transformed (In) POP
concentrations generally range from 0.50 to 2.0.
(3)	Differences in geometric mean POP concentrations between low-income and
higher-income families as well as between daycare centers and homes generally
range from 0 to 500%, between city and rural areas, they range from 0 to 150%,
and between smokers' and non-smokers' homes, they range from 0 to 250% (B2-
PAH only), depending on the compound and medium.
Based on the analysis of the historical and current data, the calculations were performed
with the following assumptions:
(1)	A two-sample t-test was conducted at the 5% significance level on ln-transformed
POP to compare the POP exposures in the following groups of children:
low-income families versus middle/upper income families,
at daycare centers versus at home,
inner city versus rural areas,
smokers' homes versus nonsmoker's homes.
The comparison of POP exposures in smokers' versus non-smokers' homes was
performed on B2 PAH only, because only these two groups have data from the
PAH exposure studies.
(2)	Sample sizes were computed that provide 80 or 90 percent power for detecting a
significant difference between the two groups when the actual percent difference
ranges from 10 to 200 percent. (The power represents the level of confidence
desired to detect a specified difference between the two groups. An experiment

-------
designed to have 90 percent power for detecting a specified difference will be
more sensitive than one designed to have 80 percent.)
(3) Sample sizes were computed assuming that the standard deviation of In-
transformed POP concentrations is either 0.5, 1, 1.5, or 2.
Table II summarizes the estimated sample sizes required to detect specific differences
between any two groups of children, i.e. low- and middle/upper income children, at-center and
at-home children, or inner-city and rural children. For example, to detect a difference between
two groups of children if the standard deviation of ln-transformed POP is 1.0 and the actual
percent difference between the two groups is 100 percent, roughly 34 children are needed in each
group to give 80% power. If the standard deviation of ln-transformed POP is 1.5, the number of
children increases to approximately 75 per group.
For all seven target POP concentrations and all six media (floor dust, indoor and outdoor
air, solid and liquid food, and soil), the median percent difference in POP concentrations between
the two groups of children was 60 percent, and the median standard deviation in ln-transformed
POP concentrations was 1.0. As shown in Table II, if the standard deviation in ln-transformed
POP concentrations is 1.0 and the actual percent difference between the two groups of children is
50 percent, then a sample size of approximately 100 children per group will provide 80 percent
power for detecting a statistical difference in POP exposures between any two groups of children.
Under the same conditions, a sample size of approximately 130 children per group provides 90
percent confidence that a statistical difference will be detected between the two groups. To allow
for missing samples and other data issues, a sample size of 120 to 160 children per group is
recommended.
To test the above hypotheses with maximum power, at least 120 children each from low-
income and middle/upper income families (targeted at 80 % power), who will be between the
ages of 18 mo and 5 yr at the time of sampling, would participate. Those under age 4 will be
preferred. Children must be able to provide spot urine samples during the day, and those who
are still being breast-fed will be excluded. Children who attend daycare centers and children who
stay at home will be included. They will be monitored at their homes and at the centers. Ideally,
the 120+ children in each group (low-and middle/upper income) would be evenly distributed in
each subgroup, for example, inner city versus rural, to provide the same power. U.S. Census
Bureau definitions of urban and rural will be used. Low-income families are defined as those
who meet the qualification criteria for the Women, Infants, and Children program (WIC), i.e.,
household income not exceeding 185% of the federal poverty level.
If equal numbers of children were selected in each stratification, there would be 128
households in each category, as summarized in Table III. The total number of households
would be 256. Overall, there would be 128 households in each category: rural, urban, low-
income, middle/upper income, at home during the day, and at pre-school or day care during the
day. Based on the sample size requirements in Table II and an estimated standard deviation of
1.0, 128 children per group would allow detection of a difference of 50% between groups at 90%)
power, or with a standard deviation of 2.0, detection of a difference of 100%) at 80%> power.
To obtain a probability-based stratified sample, the sample group of low-income subjects
would be smaller than the sample group of middle/upper income subjects; and the rural group
would be smaller than the urban group. Extreme oversampling of the low-income or rural
groups, however, would induce imbalance in sampling rates and weights and inefficient sampling
design. Alternatively, balancing the sizes of the two groups leads to better statistical power for
group comparisons. To increase the representativeness of the study, allow intergroup

-------
comparisons, and meet the study objectives, yet stay within the confines of reasonable
expenditures of resources, a sampling plan has been developed that takes into account the
necessary compromises between the aforementioned conflicting statistical goals.
B.	Sampling Plan
The sampling plan that is presented here is expected to accomplish the following:
•	Acquire the data that are necessary to meet the major objectives of the CTEPP
study regarding total exposure measurement, evaluation and refinement of
models, and apportionment of exposure pathways for young children.
•	Increase the representativeness of CTEPP by recruiting using two sample frames.
•	Maximize the response rates for participation, with a target response rate of 75%.
•	Maximize the amount of useful information that can be obtained, while balancing
the conflicting demands of representativeness and hypothesis testing, and
•	Minimize the total cost of the study, recognizing that the per-child unit costs to
measure total exposures are high.
A target sample of 128 children (and associated caregivers) in each state will be obtained. This
sample will be balanced evenly between the day care and no day care components. The
sampling design will oversample the low-income strata, but sample sizes will still be smaller for
the low-income than for the middle/upper income group overall. Sample sizes will also be
smaller in the rural than in the urban stratum.
Although every attempt will be made to obtain samples that are truly representative of the
populations from which the samples are drawn, there are potential confounders that may limit
this representativeness. For example, the no day care component will be drawn from those
households that have telephones; this will exclude families that have no phone, which means that
the most indigent families, families who have recently moved, migrant workers' families, and
others may not be represented. By inclusion of day care centers that serve primarily low-income
clients, this limitation will be partially ameliorated. A small degree of self-selection into the day
care sample is also possible, if families who receive public assistance are more likely to apply for
day care. Because of the difficulty of obtaining physical and biological samples from children
who are still breast-fed or who are not toilet-trained, these children will also be underrepresented.
Through analysis of the pre-monitoring questionnaire data and related Census data, it should be
possible to obtain rough estimates of the fraction of the targeted population who are
underrepresented; these estimates can then be used to control for the confounders.
Nevertheless, the information that CTEPP will generate will be tremendously useful in
meeting the objectives of the study and in testing the study hypotheses, as discussed earlier.
The proposed sample design provides a compromise in sensitivity between analytical
inferences and population-based inferences. This sampling plan is described in detail in
Appendix A, and is shown schematically in Figure 3.
C.	Sampling, Sampled Media and Targeted Chemicals
As in the small pilot studies, the sampled media will be indoor and outdoor air, house and
classroom dust, play area soil, solid and liquid food, drinking water, hand skin surface, and urine.
Additional sampling in CTEPP includes hard floor surface wipes and wipes of the most-used
food preparation surface, in those households and child care centers where pesticides have been
applied in the previous seven days.
Dislodgeable residues using a polyurethane foam (PUF) roller, food preparation surface
wipes, and hard floor surface wipes will be collected in those households that have had pesticide

-------
applications indoors or outdoors in the 7 days previous to sampling. Indoor and outdoor air
samples will be collected by continuous sampling over 48 hr at 4 L/min, using a URG
(University Research Glassware, Chapel Hill NC) sampler with a 10 |im inlet and a cartridge
containing a quartz fiber filter and XAD-2 resin and PUF [R3] in series. The soil and dust
samples will be obtained at the end of each 48-hr sampling period. Dust samples will be
collected in the room that the child uses most, using an HVS3 vacuum sampler (Cascade Stack
Sampling Systems, Bend OR) [LI, R3] for carpeted areas or a wipe sample for uncarpeted, hard-
surfaced areas. Participants will also be asked to donate a vacuum cleaner bag of floor dust,
collected during the month previous to sampling, which will be used for further methods
development. Soil samples will be collected by scraping up the top 0.5 cm of soil in an 0.095 m2
(1 ft2) area in the middle of the child's play area [LI], Smooth floor surface wipes will be
collected in the area that is pointed out by the teacher or parent as that area where the child is
most likely to spend time. Food preparation surface wipe samples will be collected from the
most-used food preparation surface. Diet samples will be obtained by the duplicate plate method
[N3, Fl], collecting duplicate servings of all foods that the child is served over the two-day
period. In the post-monitoring visit, the caregiver will be asked to describe the food sample
contents and confirm the food diary information. Solid food, liquid food, and drinking water will
be collected separately in glass containers to avoid phthalate ester contamination likely with
some food storage containers. Handwipe samples, with a gauze pad wetted with 50% isopropanol
in water, will be collected prior to participants' washing their hands, just before lunch and just
before supper on each of two days [G1 ]. Urine samples will be approximate 48-hr collections,
collected as spot urine samples accumulated over the two-day sampling period. If the household
has applied pesticides in the preceding 7 days, the spot urine samples will be analyzed separately.
Otherwise, the urine samples will be combined for each 48-hr period. Sampling methods are
described in detail elsewhere [Wl, W2, W6, W7, C10, C13, T2],
Targeted chemicals include representatives of several compound classes: poly cyclic
aromatic hydrocarbons; phthalate esters; organochlorine, organophosphate, carbamate, triazine,
and pyrethroid pesticides; phenols; acid herbicides; and polychlorinated biphenyls. Table IV
summarizes the target compounds and the reasons for their selection.
D.	Analysis Methods
Methods for chemical analysis of the targeted chemicals in the environmental media of
interest were developed previously and are available for use in this research [C10, CI2, CI3, S4,
S5, H3, N5, N7]. These methods are summarized in Table V.
E.	Supporting Information
Information needed for the interpretation of the chemical results will be collected through
survey instruments that were developed and utilized for the small pilot studies [Wl, W2, W6,
W7, C10, C12, C13], and modified as appropriate for this study. Extensive information on
housing characteristics will be obtained, including details on home interior, exterior, location,
GPS coordinates, and surrounding areas. Each subject's age, weight, gender, and race or
ethnicity will be recorded. The questionnaire includes questions regarding pesticide use, smoking
habits, heating, cooking, and cleaning activities, income/education level of the household, and
other information relevant to understanding the chemical measures, such as carpet and hard-
surface floor area, location of residence and school, size of residence etc. Additionally, an
activity log for the child and the adult, a diary of the foods eaten during the sampling period, and
food logs or menus for the two weeks preceding sampling will be collected. Because active

-------
children breathe more, eat more, and move around more than those who are passive in a possibly
polluted environment, the activity logs will include classification of the level of the child's
activity. Approximately 10% of the children will be videotaped for 3-4 hr each during the
sampling period to supplement the activity logs. Subsequent videotape analysis and interpretation
will be done using published methods, refined as necessary for CTEPP [Fl, G2, L6, Z3].
Because the above information may include some sensitive information, or information
that could conceivably be used to identify individual subjects, it is necessary to protect the
privacy of the participants. A Certificate of Confidentiality has been obtained to protect the
information from involuntary release. Additionally, researchers will not disclose any study
information that can be associated with individual participants without the participants' consent,
and study data will be coded into the data base without individual identifying information.
F.	Human Subjects Approval
Collection of urine samples from the children and their caregivers requires human
subjects approval, both from EPA and through the Institutional Review Boards (IRB) of
participating contractors. These approvals have been obtained.
G.	Information Collection Approval
Because this study involves collection of information on more than nine subjects, through
the use of questionnaires, activity logs, and food diaries, approval of the Office of Management
and Budget (OMB) is required. This review process, known as the Information Collection
Request (ICR) under the Paperwork Reduction Act of 1995, will be assisted by the fact that many
of the procedures and survey instruments have been field-tested successfully in the pilot studies.
The process of obtaining ICR approval was begun in February 1999 by publication of a CTEPP
outline in the Federal Register. OMB approval of the ICR was given in March 2000.
II. Quality Assurance
A Quality Systems and Implementation Plan (QSIP), which combines the work plan and
the quality assurance project plan, has been developed for this study. The requirements for the
quality assurance program are stated in EPA Order 5360.1, "Policy and Requirements to
Implement the Quality Assurance Program." This order requires that quality assurance becomes
an integral part of all data collection activities and is totally integrated into the program, to assure
the reliability of environmental measurements and data. Guidance is found in EPA documents
EPA QA/R-2 and EPA QA/R-5.
Included in the QSIP are Project Quality Objectives (PQOs) and, for individual target
analytes, Data Quality Objectives (DQOs). These will be set before field sampling begins.
Preliminary PQOs include the following:
Recruiting
•	Eligibility: The initial recruitment activities will screen the eligibility of the study
participants. Children who stay at home with an adult caregiver and children who
attend preschool or day care are included in the study. Emphasis is on the
younger children aged 18 months to 4 years. Children who are not able to provide
one or more spot urine samples during the day (who are not at least partially
toilet-trained) and children who are still being breast-fed are excluded.
•	Response: With intensive recruiting efforts and guaranteed confidentiality, we
expect to achieve response rates of about 85% in each of these two stages for the
day care sample component. For the random digit-dialing (RDD) sample

-------
component, we anticipate that about 75% of eligible households will participate in
the study. A variety of non-monetary and monetary incentives will be used [K4];
we are seeking advice on non-monetary incentives from the department of
developmental psychology at the North Carolina State University.
Statistical Power
•	Sample Size: Sufficient to detect a 50% difference in POP exposures with a
standard deviation of 1.0 at 90% power
Sampling and Analysis
•	Standard Operating Procedures (SOPs) are included in the QSIP
•	Method Precision: Relative standard deviation < ±25% for urine samples; ±15%
for other samples. These are set specifically for different analytes and different
media.
•	Recoveries: ±80 - 120%
•	Limits of Detection (LODs): These are both medium-specific and compound-
specific. For PAH, for example, the LODs are 0.03 ng/m3 in air, 1 ng/g in soil,
0.02 ng/g in food, and 0.017 ng/mL in urine.
I. Statistical Analysis of Data
All the questions that are posed above are amenable to hypothesis testing. Therefore, the
data will be analyzed and fitted to distributions, which will be tested for normality. If the
distributions are normal, then hypothesis testing will use two-sided t tests (alpha = 0.05). If the
data are not distributed normally, then the non-parametric two-sample K-S test will be used on
the distributions (alpha = 0.05). Because of the small number of samples involved in individual
categories of the stratified design (See Table II), testing will be done on lumped categories, such
as low vs high income.
J. Data Analysis
Interpretations that will be sought in the data analysis include all significant relationships
among the study variables for the population tested, with emphasis on the stratification variables
discussed above. Additionally, interpretive relationships will be sought between the
questionnaire data, such as reported pesticide use or parental occupation, and the chemical and
physical data acquired. Appendix B provides the CTEPP Data Analysis Plan.
K. Initial Calculation of Exposure, Potential Daily Intake, and Potential Daily Dose
The exposure values (ng/day) for inhalation and ingestion (dietary and nondietary) can be
converted to units of maximum potential dose by assuming 100% absorption in the body and
normalizing for body mass. Various factors can be found in the literature to account for physical,
chemical, and/or physiological processes. For maximum estimates, this conversion gives upper
limits on the amount of a pollutant available for delivery to target organs. In subsequent
refinement of the exposure estimates, literature absorption factors for the various targeted
compounds will be used as they become available.
The potential daily dose of a target compound in ng/kg body mass per day can be
estimated using the following equations, which together comprise the most commonly used
Microenvironmental Exposure Model (MEM):

-------
„ _ C,*t, + C„* 
-------
Although exposure factors for children and adults are available in the literature for
inhalation and soil ingestion [E2, B3, S3], and for dermal absorption of some compounds, there
are uncertainties in these factors, which are especially large for the dermal and nondietary
ingestion routes of exposure. Some studies in the past have assumed ventilation rates of 20
m3/day for adults and 15 m3/day for children [06, B3]; more recent values are estimated
according to the subject's age. Estimated ventilation rates will be based on the EPA Exposure
Factors Handbook recommendations [E2], currently 6.8, 8.3, 11.3, and 15.2 m3/day for children
ages 1-2 yr, children ages 3-5 yr, adult females, and adult males, respectively. Dust soil ingestion
rates will be based on the values published in the peer-reviewed literature [E2, S3, LI] , currently
0.06 g/day for children and 0.1 g/day for adults. These factors will be used in the initial
calculations with the MEM. As better estimates become available, they will be incorporated in
the model.
L. Peer Review
Peer review of the study, its implementation, and its outputs is an important part of
assuring the relevance, significance, quality, and scientific merit of this research. Internal peer
review is a part of each step of the study, including the study design, the combined work plan and
quality assurance project plan (QSIP), work assignments/task orders, progress and final reports,
and publications. External peer review by a panel of outside experts in the human exposure field
was obtained for the study design in January 1999, and the peer reviewers' comments are
incorporated in this design. External peer review will be obtained for the resulting publications.
Additionally, an external advisory panel of experts in the exposure field will provide feedback on
the research at appropriate intervals.
M. Milestones
This research is expected to span a period of four to five years, beginning in FY98, and
will occur in several phases. A list of some milestones is given below.
Phase I Milestones - initiated in FY98
•	Design a pilot study of children's total exposure to persistent organic pesticides and
other persistent organic pollutants that considers all three environmental pathways:
ingestion, inhalation, and dermal absorption, and all environmental media that are
likely to provide the opportunity for significant exposures.
•	Obtain peer review of the study design and develop a Quality Systems Implementation
Plan (QSIP), which follows the Quality Integrated Work Plan (QIWP) template that
has been developed for the North American Research Strategy on Tropospheric Ozone.
This template is available on the Internet at the following location:
http://cdiac.esd.ornl.gov/programs/NARSTO/narsto/html. Use of this template ensures
that the implementation plan meets the American National Standards Institute (ANSI)
E-4 standard. The QIWP is equivalent to the NERL Quality Systems Implementation
Plan (QSIP).
•	Initiate human subjects review (Institutional Review board, IRB) and EPA human
subjects approval.
-------
•	Initiate the Information Collection Request (ICR) process with the Office of
Management and Budget (OMB).
•	Select and modify or improve sampling and analytical methods.
•	Develop a communication strategy.
Phase II Milestones - Initiated in FY99
•	Obtain ICR approval.
•	Obtain human subjects approval.
•	Initiate and complete field sampling in six counties in each of two states: North
Carolina and Ohio. (No recruiting or field sampling will be conducted during the U.S.
Census 2000 period, specifically, March 1, 2000 to June 30, 2000.)
Phase III Milestones - Initiated in FY00
•	Analyze samples.
•	Interpret the results in the light of the CTEPP objectives.
•	Publish the results in one or more peer-reviewed scientific journals.
-------
IV. Personnel and Implementation
Insofar as possible, this research is intended to be an in-house activity of the National
Exposure Research Laboratory. The NERL team members who are involved in the study
include:
Dr. Nancy K. Wilson, Principal Investigator
Dr. Wilson is a research chemist with extensive experience in the human exposure
field, including study design, sampling and analysis, and interpretation of results. She has a B.S.
in chemistry from the University of Rochester, and an M.S. and Ph.D. in physical chemistry from
Carnegie-Mellon University. After Dr. Wilson retired Dr. Marsha Morgan becomes the PI and
TOPO for the CTEPP study.
Dr. Marsha Morgan, Staff
Dr. Marsha Morgan was a NERL postdoctoral fellow on the study and became the EPA
PI and TOPO for the CTEPP study in September 2000. She has experience in analytical
laboratory and field work, as well as in toxicology. She has a B.S. in pre-medicine zoology from
Ohio University, a MS in environmental health from East Tennessee State University, and a
Ph.D. in environmental toxicology and animal science from Michigan State University.
Gary Evans, Co-Principal Investigator
Mr. Evans is a chemical engineer with extensive experience in exposure measurements
and modeling. He will be responsible for many aspects of the questionnaires and activity pattern
measurements.
Dr. Robert G. Lewis, Staff
Dr. Lewis is a research chemist with extensive experience in the human exposure field,
especially in the methods development and application areas. He has a significant and extensive
background in the measurement of pesticides exposure in environmental and human media,
including that of young children. Dr. Lewis has a B.S. in chemistry from the University of North
Carolina, Chapel Hill, and a Ph.D. in organic chemistry from the University of Wisconsin,
Madison.
Thomas R. McCurdy, Staff
Mr. McCurdy is a physical scientist with extensive experience in atmospheric
measurement and exposure, with activity patterns, with project management, and with the
statistical handling of research data. He has over 15 years of exposure modeling and assessment
experience.
Dr. Elaine Cohen-Hubal, Staff
Dr. Hubal is a research chemical engineer with a background in mathematical
modeling of environmental and biological systems. She has a B.S. in chemical engineering from
MIT and an M.S. and Ph.D. in chemical engineering from North Carolina State University.
Carvin Stevens, Staff, Quality Assurance Overseer
Mr. Stevens is a chemist with wide experience in a variety of areas, including field
measurements, chemical analysis, and quality assurance.
-------
Dr. Maurice Berry, Consultant
Dr. Berry is the NERL Program Manager for dietary exposure research. He has a
background in dietary exposure measurement, modeling and methods development. He also has
experience in numerous multimedia measurements programs and will serve as an advisor and
consultant to the project.
Dr. James Quackenboss, Consultant
Dr. Quackenboss is a NERL research scientist with an extensive background in
environmental chemistry, mechanisms of exposure, and field measurements of exposure. He
will serve as an adviser and consultant to the project.
Professor Amy Halberstadt, Consultant
Dr. Halberstadt is a professor in the field of developmental psychology at the North
Carolina State University. She has advised us on several aspects of the child videotaping, and
will also advise us on the selection and presentation of non-monetary incentives to day care
centers, parents, and children to increase response rates.
Because of the extensive field work that this research will entail, it was anticipated that
most of the field work and chemical analysis would be done through extramural contracts. A
task order to Battelle Memorial Institute under Contract 68-D99-011 for field and analytical
support was awarded in August 1999.
-------
V. Funding
-------
VI. Selected Outputs
#
APG/APM
Product
Code
ORD
Peer
Review
Date
Type, Title, and Lead NERL Author
1
194
X8

7/98
Draft study design, "Children's Total
Exposure to Persistent Pesticides and Other
Persistent Pollutants," Nancy K. Wilson
[COMPLETE]
2
194
X7
4
1/99
Peer input to draft study design, "Children's
Total Exposure to Persistent Pesticides and
Other Persistent Pollutants," Nancy K.
Wilson and Carvin Stevens [COMPLETE]
3
194
X6
4
8/99
Study design, "Children's Total Exposure to
Persistent Pesticides and Other Persistent
Pollutants," Nancy K. Wilson [COMPLETE]
4
194
X6
4
12/99
QSIP/QIWP, "Children's Total Exposure to
Persistent Pesticides and Other Persistent
Pollutants," Nancy K. Wilson [COMPLETE]
5
194
X9

3/00
Human subjects approval, "Children's Total
Exposure to Persistent Pesticides and Other
Persistent Pollutants," Nancy K. Wilson
[COMPLETE]
6
194
X9

4/00
OMB approval, "Children's Total Exposure
to Persistent Pesticides and Other Persistent
Pollutants," Gary Evans [COMPLETE]
7
837
X8

9/99
Protocol for field exposure study of children
to two endocrine disrupters, Nancy K.
Wilson [COMPLETE. Included in #3
above.]
8
194
C8
4
1/01
Progress Report on Recruitment, To be
determined.
9
194
C2
3
6/01
Journal article, "Results of Field Sampling
and Chemical Analyses," To be determined.
10
194/New
C8
4
5/02
Report, "Statistical Analysis and
Interpretation of Results," To be determined.
11
194
C2
3
12/02
Report/Journal article, "Children's
Exposures to Persistent Organic Pollutants,"
To be determined.
-------
#
APG/APM
Product
Code
ORD
Peer
Review
Date
Type, Title, and Lead NERL Author
12
194
C2
3
12/03
Report/journal article "Children's
Exposures to Persistent Organic Pollutants,"
To be determined.
-------
VII. References
A1 M. C. R. Alavanja, D. P. Sandler, S. B. McMaster, S. H. Zahm, C. J. McDonnell, C. F.
Lynch, M. Pennybacker, N. Rothman, M. Dosemeci, A. E. Bond, and A. Blair, "The
Agricultural Health Study." Environ. Health Perspec., 104, 362-369 (1996).
A large prospective study of the exposures of farmers and farm families to
agricultural pesticides.
B1 M. A. Bradman, M. E. Harnly, W. Draper, S. Seidel, S. Teran, D. Wakeham, and R.
Neutra, "Pesticide Exposures to Children from California's Central Valley— Results of
a Pilot Study." J. Expos. Anal. Environ. Epidem., 7, 217-234 (1997).
Handwipe sampling for OP pesticides on 11 rural children detected 10 of 33
pesticides that were found in their house dust. Dust ingestion scenarios suggest that
diazinon exposures could exceed the reference dose (9xl0e-5 mg/kg/day).
B2 S. Bhatia and J. P. Neglia, "Epidemiology of Childhood Acute Myelogenous
Leukemia." J. Pediatric Hematol. Oncol., 17, 94-100 (1995).
Environmental risk factors include association with parent and family member
exposure to petroleum products and pesticides.
B3 Barclays Official California Code of Regulations (Barclays Law Publishers, South San
Francisco, CA, 1996). "Level of exposure to carcinogens." Vol. 28, §12721
B4 T. J. Buckley, J. M. Waldman, R. Dhara, A. Greenberg, Z. Ouyang, and P. J. Lioy, "An
Assessment of a Urinary Biomarker for Total Human Environmental Exposure to
Benzo[a]pyrene (BaP)." Int. Arch. Occup. Environ. Health, 67, 257-266 (1995).
B6 T. J. Buckley, J. D. Liddle, D. Ashley, V. Burse, D. Paschal, and G. Akland,
"Measurements of Exposures and Biomarkers in the Lower Rio Grande Valley:
Multimedia Results for Pesticides, Metals, PAH, and VOCs." Environ. Int. 23, 705-
732 (1997).
CI J. C. Chuang, "Evaluation and Application of Methods for Estimating Children's
Exposure to Persistent Organic Pollutants in Multiple Media." Progress report to
USEPA for Contract 68-D4-0023, Work Assignment 2-01, September 1997.
C2 J. C. Chuang, M. A. Pollard, Y. Chou, R. G. Menton, and N. K. Wilson, "Evaluation of
Enzyme-Linked Immunosorbent Assay for the Determination of Polycyclic Aromatic
Hydrocarbons in House Dust and Soil." Sci. Total Environ., 224, 189-199 (1998).
C3 J. C. Chuang, P. J. Callahan, R. G. Menton, S. M. Gordon, R. G. Lewis, and N. K.
Wilson, "Monitoring Methods for Polycyclic Aromatic Hydrocarbons and their
Distribution in House Dust and Track-In Soil." Environ. Sci.Technol., 29, 494-500
(1995).
-------
C4 J. C. Chuang, M. R. Kuhlman, and N. K. Wilson, "Indoor Concentrations of Polycyclic
Aromatic Hydrocarbons inNonsmokers' Residences During Winter and Summer
Seasons." Measurement of Toxic and Related Air Pollutants: Proceedings of the 1991
EPA/AWMA International Symposium, Pub. VIP-21, AWMA, Pittsburgh PA, 1991, pp.
1124-1140.
C5 M. D. Cheng, P. D. Hopke, andN. K. Wilson, "Receptor Modeling of Indoor
Polynuclear Aromatic Hydrocarbons Collected in Columbus, Ohio." Proceedings of
the 1991 AWMA National Meeting, AWMA, Pittsburgh PA, 1991.
C6 J. C. Chuang, P. J. Callahan, S. M. Gordon, N. K. Wilson, and R. G. Lewis, "Methods
for Polycyclic Aromatic Hydrocarbons and Tobacco Smoke Markers in House Dust:
Laboratory and Field Evaluation." Measurement of Toxic and Related Air Pollutants:
Proceedings of the 1993 EPA/AWMA International Symposium, Pub. VIP-34, AWMA,
Pittsburgh PA, 1993, pp. 88-93.
C7 J. C. Chuang, N. K. Wilson, and C. Lyu, "Summer Study of Field Methods for
Estimating Exposure of Children in Low-Income Families to Polycyclic Aromatic
Hydrocarbons." Final Program, Annual Meeting of the Society for Risk Analysis and
the International Society of Exposure Analysis, Society for Risk Analysis, McLean VA,
Paper P3.13, p.161.
C8 J. C. Chuang and N. K. Wilson, "Method Validation and Characterization of Phenols
in House Dust and Soil." Presented at the 1997 EPA/AWMA International
Symposium on Measurement of Toxic and Related Air Pollutants, Research Triangle
ParkNC, May 1997, Paper P-l.
C9 J. C. Chuang, M. G. Nishioka, andN. K. Wilson, "Evaluation of Enzyme-Linked
Inmmunosorbent Assays for Polycyclic Aromatic Hydrocarbons, Pentachlorophenol,
and 2,4-D Herbicide in House Dust and Soil." Invited paper presented in the
Immunochemistry Summit VI symposium at the national meeting of the American
Chemical Society, Las Vegas NV, September 1997.
CIO J. C. Chuang, P. J. Callahan, C. W. Lyu, Y.-L. Chou, and R. G. Menton,
"Characterization of Polycyclic Aromatic Hydrocarbons Exposure Among Children of
Low-Income Families from Inner Cities and Rural Areas." Final report on Cooperative
Agreement CR822073, EPA 600/R-98/163a, 163b, and 163c (1998).
CI 1 J. C. Chuang, Y. L. Chou, M. Nishioka, K. Andrews, M. Pollard, and R. Menton,
"Field Evaluation of Screening Techniques for Polycyclic Aromatic Hydrocarbons,
2,4-Diphenoxyacetic Acid, and Pentachlorophenol in Air, House Dust, Soil, and Total
Diet." Final Report, Contract 68-D4-0023, WA 1-04. EPA 600/R-97/109, 1997.
C12 J. C. Chuang, P. J. Callahan, C. W. Lyu, andN. K. Wilson, "Polycyclic Aromatic
Hydrocarbon Exposures of Children in Low-Income Families." J. Expos. Anal.
Environ. Epidem., 2, 85-98 (1999).
-------
C13 J. C. Chuang, C. Lyu, Y-L Chou, P. J. Callahan, M. Nishioka, K. Andrews, M. A.
Pollard, L. Brackney, C. Hines, D. B. Davis, and R. Menton, "Evaluation and
Application of Methods for Estimating Children's Exposure to Persistent Organic
Pollutants in Multiple Media." Report, Contract 68-D4-0023, EPA 600/R-98/164a,
164b, and 164c (1998).
C14 D. E. Camann, H. J. Harding, C. L. Stone, and R. G. Lewis, "Comparison of PM2.5
and Open-Face Inlets for Sampling Aerosolized Pesticides on Filtered Polyurethane
Foam." Measurement of Toxic and Related Air Pollutants: Proceedings of the
EPA/AWMA International Symposium, Air & Waste Management Association,
Pittsburgh, PA, Publication VIP-39, 1994, pp. 838-843, ASTM Standard Practice D
4861.
C15 D. A. Crain, L. J. Guilette, A. A. Rooney, and D. B. Pickford, "Alterations in
Steroidogenesis in Alligators (Alligator Mississippiensis) Exposed Naturally and
Experimentally to Environmental Contaminants." Environ. Health Perspect., 105, 528-
533 (1997).
Studies of the effects of pesticides as EDCs in juvenile alligators showed no
effects of vinclozilin or 2,4-D, but showed an ED effect of atrazine, as measured by
plasma testosterone, aromatase, and gonadal histopathology.
C16 Chemical and Engineering News, June 8, 1998, p. 11.
C17 E. A. Cohen-Hubal, L. S. Sheldon, T. R. McCurdy, M. L. Rigas, J. M. Burke, V. G.
Zartarian, and N. C. G. Freeman, "Children's Exposure Assessment: A Review of
Factors Influencing Children's Exposures and the Data Available to Characterize and
Assess that Exposure." Submitted for publication to Environmental Health
Perspectives, July 1999.
D1 J. R. Davis, R. C. Brownson, R. Garcis, B. J. Bentz, and A. Turner, "Family Pesticide
Use and Childhood Brain Cancer "Arch. Environ. Contam. Toxicol., 24, 87-92 (1993).
Significant positive associations were found with child brain cancer for use of
pesticides in the home — pest strips, termiticides, flea collars, garden diazinon, and
herbicides, comp to non-cancer controls. Associations were significant compared to
other cancers for pest bombs, termiticides, flea collars, insecticides in the garden,
carbaryl in the garden, and herbicides. (Done by survey; no physical measurements
were made.)
D2 J. L. Daniels, A. F. Olshan, and D. A. Savitz, "Pesticides and Childhood Cancers."
Environ. Health Perspect., 105, 1068-1077 (1997).
Epidemiological studies published between 1970 and 1996 were reviewed.
Reported relative risk estimates were modest. Frequent occupational exposure to
pesticides or home pesticide use was related to increased risk of childhood cancer,
especially leukemia and brain cancer. Residence on a farm was associated with
increased risk.
-------
D3 I. R. Danse, R. J. Jaeger, R. Kava, M. Kroger, W. M. London, F. C. Lu, R. P. Maickel,
J. J. McKetta, G. W. Newell, S. Shindell, F. J. Stare, and E. M. Whelan, "Position
Paper of the American Council on Science and Health - Public Health Concerns about
Environmental Polychlorinated Biphenyls (PCBs)." Ecotoxicol. Environ. Safety, 38,
71-84(1997).
A review of the scientific literature on PCBs in the environment and their
potential risk. Studies of workers exposed to high doses of PCBs over long periods of
time have not demonstrated an increased cancer risk. Recent studies implicating
prenatal PCB exposure in neurodevelopmental toxicity have methodological
deficiencies. No conclusive evidence exists at this time that PCB levels in the
population are causing in utero intellectual deterioration. Evidence for estrogenic
effects of environmental PCBs remains weak and circumstantial. Thus more study is
needed.
D4 C. DeRosa, P. Richter, H. Pohl, and D. J. Jones, "Environmental Exposures that Affect
the Endocrine System: Public Health Implications." J. Toxicol. Environ. Health, Part
B, 1,3-16(1998).
An overview of the chemicals that have been implicated as endocrine disrupters.
Approximately one-fourth are the persistent and lipophilic organochlorine compounds,
such as DDT and PCBs.
Environmental Working Group, "Overexposed: Organophosphate Insecticides in
Children's Food." Report, Environmental Working Group, Washington, DC, 1998.
Reports analyses of data from the 80,000 food samples tested by USD A and FDA
and dietary records collected for 4000 children by USDA. Concludes that unsafe levels
of OP exist in baby food and other foods eaten by children, especially apples, peaches,
and grapes. Calls for ban on all home and structural uses of OP pesticides and on
commodities that end up in baby food.
Exposure Factors Handbook, U.S. EPA Office of Health and Environmental
Assessment (1997). Washington, DC.
T. Field and E. Ignatoff, "Videotaping Effects on the Behaviors of Low Income
Mothers and Their Infants During Floor-Play Interactions." J. Appl. Developmental
Psych., 2, 227-235 (1981).
Videotaping initially affects the behavior of mothers interacting with their
children, but repeated sessions or more lengthy exposures reduce the observer effect.
P. W. Geno, D. E. Camann, H. J. Harding, K. Villalobos, and R. G. Lewis, "Handwipe
Sampling and Analysis Procedure for the Measurement of Dermal Contact with
Pesticides." Arch. Environ. Contam. Toxicol., 30, 132-138 (1996).
Cellulose sponges wet with isopropyl alcohol were used to wipe children's hands,
with quantitative removal of chlorpyrifos and pyrethrin. Applicability to 29 other
pesticides is suggested, including acid herbicides.
E2
F1
G1
-------
G2 D. Gross, "Issues Related to Validity of Videotaped Observational Data." Western J.
Nursing, 13, 658-663 (1991).
A review of strategies for obtaining useful videotaped observational data.
HI M. Heil, B. Schiller, A. C. Huggett, and F. Haschke, "Toxicological Aspects of Food
for Infants and Children." Monatsschrift Kinderheilkunde, 144 Suppl. 2, S224-S229
(1996).
The author estimates exposure to pesticide residues in food from food analyses
and the estimated diet of a 4-mo old child.
H2 P. T. C. Harrison, P. Holmes, and C. D. N. Humfrey, "Reproductive Health in Humans
and Wildlife - Are Adverse Trends Associated with Environmental Chemical
Exposure." Sci. Total Environ., 205, 97-106 (1997).
Trends in the incidences of testicular and breast cancer, and concern about
reduced semen quality, cryptorchidism, hypospadias, and polycystic ovaries may be
associated with endocrine disruptors. Suspected compounds include naturally
occurring steroids, phyto- and myco-estrogens, synthetic hormones, organotins,
organochlorine pesticides, polychlorinated biphenyls, dioxins, alkylphenols,
polyethoxylates, phthalates, and bisphenol-A. However, there is no direct causal
evidence in humans.
H3 J. P. Hsu, H. G. Wheeler, Jr., H. J. Schattenberg, III, D. E. Camann, R. G. Lewis, and
A. E. Bond. "Analytical Methods for Determining Nonoccupational Exposures to
Pesticides," J. Chromatogr. Sci. 26, 181-189 (1988).
K1 W. R. Kelce and E. M. Wilson, "Environmental Antiandrogens - Developmental
Effects, Molecular Mechanisms, and Clinical Implications." J. Molec. Medicine., 75,
198-207 (1997).
A review of the mechanism of toxicity and clinical implications of environmental
chemicals that inhibit androgen-mediated sex development. These EDCs include the
fungicide vinclozilin and the pesticide DDT and its metabolite DDE.
K2 R. J. Kavlock, G. P. Daston, C. DeRosa, P. Fennercrisp, L. E. Gray, S. Kaattari, G.
Lucier, M. Luster, M. J. Mac, C. Maczka, R. Miller, J. Moore, R. Rolland, G. Scott, D.
M. Sheehan, T. Sinks, and H. A. Tilson, "Research needs for the risk assessment of
health and environmental effects of endocrine disruptors-a report of the U. S. EPA-
sponsored workshop." Environ. Health Perspect., 104 (Suppl.4), 715-740 (1996).
A review of research needs in endocrine disrupter risk assessment.
K3 W. Karmaus and N. Wolf, "Reduced Birthweight and Length in the Offspring of
Females Exposed to PCDFs, PCP, and Lindane." Environ. Health Perspect., 103,
1120-1125.
The newborn infants of 221 teachers who had been exposed to pentachlorophenol
(PCP) and lindane in wood panels and chlorinated dibenzo-p-dioxin and chlorinated
dibenzofurans in indoor air were compared to the infants of 189 teachers who had not
been exposed. The median difference was 175 g birth weight and 2 cm birth length.
-------
The reductions in birth weight and birth length were significant at the p = 0.04 and p =
0.02 level, respectively.
K4 R. A. Kulka, "The Use of Incentives to Survey 'Hard-to-Reach' Respondents: A Brief
Review of Empirical Research and Current Practice." Proceedings of the Seminar on
New Directions in Statistical Methodology, Council of Professional Associations on
Federal Statistics (COPAFS), Bethesda, MD, May 1994.
LI R. G. Lewis, R. C. Fortmann, and D. E. Camann, "Evaluation of Methods for the
Monitoring of the Potential Exposure of Small Children to Pesticides in the Residential
Environment." Arch. Environ. Contam. Toxicol., 26, 37-46 (1994). HIPES study.
L2 J. K. Leiss and D. A. Savitz, "Home Pesticide Use and Childhood Cancer—A Case-
Control Study." Amer. J. Publ. Health, 85, 249-252 (1995).
For 252 child cases and 232 controls, yard treatment was associated with soft
tissue sarcomas (odds ratio ~4.0). Home pest strip use was associated with leukemia
(OR -1.7-3.0). The findings suggest that the use of home pesticides is associated with
some types of childhood cancer.
L3 M. P. Longnecker, W. J. Rogan, and G. Lucier, "The Human Health Effects of DDT
(Dichlorodiphenyltrichloroethane) and PCBs (Polychlorinated Biphenyls) and an
Overview of Organochlorine in Public Health." Ann. Revw. Public Health, 18, 211-244
(1997).
Many organochlorines are EDCs in experimental assays. High-level exposures to
some OC compounds appear to cause abnormalities of liver function, skin, and the
nervous system. Neonatal hypotonia or hyporeflexia has been associated with PCB
exposure. Epidemiological data is not convincing that OC compounds cause cancer.
However, if animal data is included, a recent risk estimate of 10 (exp-4) per year for
chlorinated dioxins and some PCBs has been suggested.
L4 C. Loewenherz, R. A. Fenske, N. J. Simcox, G. Bellamy, and D. Kalman, "Biological
Monitoring of Organophosphorus Pesticide Exposure Among Children of Agricultural
Workers in Washington State." Environ. Health Perspect., 105, 1344-1353 (1997).
Children up to 6 yr old who lived with pesticide applicators were monitored for
OP pesticide exposure, by measuring the urinary metabolite dimethylthiophosphate
(DMTP). Of 88 children, 47% of applicators' children and 27% of non-applicators'
children had detectable levels of DMTP. Younger children and those who lived less
than 200 feet from an orchard had higher levels.
L5 C. S. Lu and R. A. Fenske, "Air and Surface Chlorpyrifos Residues following
Residential Broadcast and Aerosol Pesticide Applications." Environ. Sci. Technol., 32,
1386-1390(1998).
Ambient air and surface chlorpyrifos residues were measured for seven days
following broadcast (Dursban) and total release aerosol (K-RID) chlorpyrifos
applications for flea control in dormitory rooms. Broadcast applications resulted in 7.5
times more total deposited chlorpyrifos on carpets than aerosol applications;
-------
dislodgeable residues on carpets were 2 times greater. Residues on nontarget surfaces
such as furniture were 140-150 times greater from aerosol applications than from
broadcast applications. However, the estimated total absorbed doses (12-33 |!g/kg)
were near the no observable effect level (NOEL 30 |ig/kg) on the first day, and lower
on following days.
L6 A. Lohaus, "The Effect of Videotaping on Preschool and Primary School Children."
Zeitschrift Pedagogische Psychologie, 1, 131-140(1987).
Videotaping improved the performance of 105 children ages 5-11 on tasks
requiring convergent and divergent thinking. The results were independent of the
children's ages.
Ml L. S. Miller, D. B. Davis, C. S. Preven, J. C. Chuang, J. C. Johnson, J. M. Van Emon,
and N. K. Wilson. "Analysis of Soil and Dust Samples for Polychlorinated Biphenyls
by ELISA." Presented in the Immunochemistry Summit VI symposium at the national
meeting of the American Chemical Society, Las Vegas NV, September 1997.
M3 R. Meinert, P. Kaatsch, U. Kaletsch, F. Krummenauer, A. Meisner, and J. Michaelis,
"Childhood Leukemia and Exposure to Pesticides—Results of a Case-Control Study in
Northern Germany." Europ. J. Cancer, 32A, 1943-1948 (1996).
For 219 cases of childhood leukemia, there was a significant association with
pesticide use in home gardens. Parental agriculture-related exposure was not
significant.
M4 D. Mukerjee et al., "Assessment of risk from Multimedia Exposures of Children to
Environmental Chemicals." 28th Annual Critical Review, J. Air & Waste Manage.
Assoc., 48, 483-501 (1998).
A review of the adverse effects observed in children associated with exposure to
environmental chemicals. Exposure factors and equations for estimating total
exposures and potential doses through the various environmental pathways are
presented. Extensive references.
N1 M. G. Nishioka and K. D. Andrews, "Method Validation and Application for
Semivolatile Organic Compounds in Dust and Soil: Pesticides and PCBs." Final
Report, Contract 68-D4-0023, WA 1-08, Task 3. EPA 600/R-97/141 (1997).
N2 National Academy of Sciences, Committee on Risk Assessment of Hazardous Air
Pollutants, "Science and Judgment in Risk Assessment." National Academy Press,
Washington DC, 1994.
N3 National Academy of Sciences, "Pesticides in the Diets of Infants and Children."
National Academy Press, Washington DC, 1993.
N4 Nonoccupational Pesticide Exposure Study (NOPES), Final Report, Atmospheric
Research and Exposure Assessment Laboratory, U. S. Environmental Protection
Agency, Research Triangle Park NC, 1990. EPA/600/3-90/003.
-------
N5 M. G. Nishioka, H. M. Burkholder, M. C. Brinkman, S. M. Gordon, and R. G. Lewis,
"Measuring Transport of Lawn-Applied Herbicide Acids from Turf to Home-
Correlation of Dislodgeable Turf Residues with Carpet Distribution and Carpet Surface
Residues." Environ. Sci. Technol., 30, 3313-3320 (1996).
A study of the track-in of 2,4-D and dicamba from lawn pesticide applications.
N6 National Exposure Research Laboratory, Research Strategy, 1997 draft.
N7 Development of Analytical Methods for Lawn-Applied Pesticides in House Dust, Report
No. 600/R-97/110, US Environmental Protection Agency, Research Triangle Park, NC,
November 1997.
N8 M. C. Brinkman, J. E. Sawchuk, and S. M. Gordon. "Sample Management and
Reporting of Results Using Database Software." Presented at the 7th Annual Meeting
of the International Society of Exposure Analysis, Research Triangle Park, North
Carolina, November 1997.
01	W. R. Ott, "Human Exposure Assessment: The Birth of a New Science." J. Expos.
Anal. Environ. Epidem., 5, 449-472 (1995).
02	W. R. Ott and J. W. Roberts, "Everyday Exposure to Toxic Pollutants." Scientific
American, February 1998, pp. 86-91.
03	Office of Research and Development, Strategic Plan, U. S. Environmental Protection
Agency, EPA/600/R-96/059, May 1996; updated 1997, EPA/600/R-97/015.
04	Office of Research and Development, U. S. Environmental Protection Agency,
Children's Risk Strategy, draft document, June 1997.
05	M. O'Malley, "Clinical Evaluation of Pesticide Exposure and Poisoning." Lancet, 349,
1161-1166(1997).
Pesticide exposure effects range from skin irritation, acute toxicity to complex
systemic illness as a result of cholinesterase inhibition, e.g. from OP exposure. Possible
links to asthma from exposure to OP pesticide contaminants.
06	ORETG, Occupational and Residential Exposure Test Guidelines (1994, 1997). U.S.
Environmental Protection Agency Office of Prevention, Pesticides, and Toxic
Substances, Washington, DC, Series 875.
07	N. Olea, P. Pazos, and J. Exposito, "Inadvertent Exposure to Xenoestrogens." Eur. J.
Cancer Prevention, 7, Suppl 1, S17-S23 (1998).
Defines endocrine disruptor (EDC) as an exogenous substance that causes
adverse health effects in an intact organism or its progeny, secondary to changes in
endocrine function. The following anthropogenic compounds are identified as EDCs:
o,p-DDT, kepone, methoxychlor, phenolic derivatives, and PCBs. Also toxaphene,
dieldrin, endosulfan, t-butylhydroxyanisole, benzylbutylphthalate, 4-hydroxy alkyl
-------
phenols, and bisphenol-A. However, evidence for ED by these compounds in humans
is meager or nonexistent.
08 W. R. Ott, Environmental Statistics and Data Analysis, CRC Press, Boca Raton, FL,
1995.
PI	F. D. Perera, "Environment and Cancer: Who Are Susceptible?" Science, 278, 1068-
1073 (1977).
Physiological factors associated with the decreased detoxification capability,
DNA repair, and immune function in the elderly and in the very young make these
groups more susceptible to environmental insults. There is a hormonal association with
deregulation of growth and differentiation through receptor binding.
P2 J. M. Pogoda and S. Preston-Martin, "Household Pesticides and Risk of Pediatric Brain
Tumors." Environ. Health Perspect., 105, 1214-1220 (1997).
An investigation of the risk of household pesticide use from pregnancy to
diagnosis in mothers of 224 children with brain tumors and 218 controls. Risk was
significantly elevated for prenatal exposure to flea/tick pesticides (OR 1.7). Sprays and
foggers were the only products significantly related to risk (OR 10.8). Elevated risk
was not observed for termite or lice treatments, pesticides for nuisance pests, or yard
and garden insecticides, fungicides, herbicides, or snail killer.
P3 F. D. Perera, R. M. Wyatt, W. Jedrychowski, V. Rauh, D. Manchester, R. M. Santella,
and R. Ottman, "Recent Developments in Molecular Epidemiology - Study of the
Effects of Environmental Polycyclic Aromatic Hydrocarbons on Birth Outcomes in
Poland." Amer. J. Epidem., 147, 309-314 (1998).
Biomarkers of PAH exposure (DNA adducts) and physical characteristics of
newborns from a heavily industrialized city and a rural town in which coal heating is
predominant were compared. Infants with high levels of DNA adducts had
significantly decreased birth weight, length, and head circumference. Cotinine (a
marker for environmental tobacco smoke exposure) was also significantly associated
with decreased birth weight and length.
R1 J. W. Roberts and P. Dickey, "Exposure of Children to Pollutants in House Dust and
Indoor Air." Rev. Environ. Contam. Toxicol., 143, 59-78 (1995).
R2 D. C. Rice, "Neurotoxicity Produced by Developmental Exposure to PCBs." Mental
Retardation Developmental Disabilities Res., 3, 223-229 (1997).
Prospective studies suggest decreased reflexes, retarded psychomotor
development in early childhood associated with PCB exposure. Decreased IQ and
reading ability were evidenced at age 11. Animal models reveal changes in activity and
cognitive function with developmental exposures to PCBs.
R3
J. W. Roberts, W. T. Budd, M. G. Ruby, A. E. Bond, R. G. Lewis, R. W. Wiener, and
D. E. Camann, "Development and Field Testing of a High Volume Sampler for
-------
Pesticides and Toxics in Dust." J. Expos. Anal. Environ. Epidem. 1, 143-155 (1991),
ASTM Standard Practice D 5438
R4 N. C. Rawlings, S. J. Cook, and D. Waldbillig, "Effects of the Pesticides Carbofuran,
Chlorpyrifos, Dimethoate, Lindane, Triallate, Trifluralin, 2,4-D, and
Pentachlorophenol on the Metabolic Endocrine and Reproductive Endocrine System in
Ewes." J. Toxicol. Environ. Health, 54, 21-36 (1998).
Carbofuran caused a significant increase in serum concentrations of thyroxine,
the major secretory product of the thyroid and a principal regulator of metabolism; all
other pesticides except trifluralin caused a decrease in thyroxine. Serum
concentrations of Cortisol were increased by trifluralin and chlorpyrifos. Insulin
concentrations were increased by dimethoate, lindane, trifluralin, triallate, and
pentachlorophenol. Estradiol concentrations were increased by lindane and trifluralin.
Luteinizing hormone (LH) was decreased by trifluralin, lindane, and dimethoate, but
increased by triallate. Pentachlorophenol and triallate caused increased severity of
oviductal intraepithelial cysts.
R5 R. A. Rudel, S. J. Melly, P. W. Geno, G. Sun, and J. G. Brody, "Identification of
Alkylphenols and Other Estrogenic Phenolic compounds in Wastewater, Septage, and
Groundwater on Cape Cod, Massachusetts." Environ. Sci. Technol., 32, 861-869
(1998).
The potential EDCs nonylphenol, octylphenol, and their ethoxylates, bisphenol-
A, nonylphenol, and phenylphenol were analyzed in wastewater, septic effluent, and
wells. Nonylphenol was detected in all septage samples. Phenylphenol and bisphenol-
A were detected in septage and wastewater. Bisphenol-A and some ethoxylates were
detected in several drinking water wells.
R6 J. R. Reigart et al., "Report of the Children's Health Protection Advisory Committee to
the U. S. Environmental Protection Agency Regarding the Selection of Five
Regulations for Re-Evaluation," Children's Health Advisory Committee, May 28,
1998.
The committee recommends re-evaluation of five regulations in the light of recent
data and the fact that protection of children was not adequately considered in the
original regulations. The five regulations are Mercury, Farm Worker Protection
Standard, Triazine Pesticides, Organophosphates and Carbamates, and Air Quality and
Asthma. Of the triazines, atrazine is identified specifically because of its
carcinogenicity and potential for causing hormonal developmental effects, and because
it has been detected in drinking water throughout the Midwest and other parts of the
nation. Of the organophosphates and carbamates, methyl parathion, dimethoate, and
chlorpyrifos are identified specifically because they represent the bulk of the dietary
risk of neurotoxicity.
SI	D. B. Shealy, M. A. Bonin, J. V. Wooten, D. L. Ashley, L.L. Needham, and A. E.
Bond, "Application of an Improved Method for the Analysis of Pesticides and their
Metabolites in the Urine of Farmer Applicators and their Families." Environ. Int., 22,
661-665 (1996).
-------
52	N. J. Simcox, R. A. Fenske, S. A. Wolz, I. C. Lee, and D. A. Kalman, "Pesticides in
Household Dust and Soil—Exposure Pathways for Children of Agricultural Families."
Environ. Health Perspect., 103, 1126-1134 (1995).
OP pesticides were higher in dust in farm homes than in non-farm homes for 59
residences in WA. Tested pesticides included azinphosmethyl, chlorpyrifos, parathion,
and phosmet. Dust concentrations were greater than soil concentrations.
53	E. J. Stanek, III and E. J. Calabrese, Human and Ecological Risk Assessment, 1, 133
(1995).
54	L. S. Sheldon, J. Keever, J. Beech, J. M. Roberds, and P. Gross, "Manual of Analytical
Methods for Determination of Selected Environmental Contaminants in Composite
Food Samples." Final Report, Contract 68-C2-0103, U.S. Environmental Protection
Agency, Cincinnati, OH.
55	L. S. Sheldon, J. T. Keever, J. M. Roberds, J. B. Beach, and J. N. Morgan, "Methods
for Measuring Base/Neutral and Carbamate Pesticides in Composite Dietary Samples."
J. Expos. Anal. Environ. Epidem., 7, 37-60 (1997).
56	M. D. Shelby, R. R. Newbold, D. B. Tully, K. Chae, and V. L. Davis, "Assessing
Environmental Chemicals for Estrogenicity Using a Combination of in vitro and in
vivo Assays." Environ. Health Perspect., 104, 1296-1300 (1996).
Suspected or known EDCs studied include 17-beta-estradiol, diethylstilbestrol,
tamoxifen, 4-hydroxytamoxifen, methoxychlor, the methoxychlor metabolite 2,2-bis(p-
hydroxyphenyl)-l,l,l-trichloroethane (HPTE), endosulfan, nonylphenol, o,p'-DDT,
and kepone.
57	J. D. Sherman, "Chlorpyrifos (Dursban)-Associated Birth Defects: A Proposed
Syndrome, Report of Four Cases, and Discussion of the Toxicology." Int. J. Occup.
Med. Toxicol., 4, 417-431 (1995).
It is suggested that four cases of unusual birth defects are associated with
maternal exposure to Dursban.
T1 R. D. Thomas, "Age-Specific Carcinogenesis-Environmental Exposure and
Susceptibility." Environ. Health Perspect., 103, 45-48 (1995).
Emphasizes the importance of dietary exposure of children relative to cancer risk.
T2 K. W. Thomas, L. S. Sheldon, E. D. Pellizzari, R. W. Handy, J. M.. Roberds, and M.
R. Berry, "Testing Duplicate Diet Sample Collection Methods for Measuring Personal
Dietary Exposures to Chemical Contaminants." J. Expos. Anal. Environ. Epidem., 1,
17-36 (1997).
U1 U. S. Environmental Protection Agency, Proceedings of the Science to Achieve Results
(STAR) Program Workshop on Children's Exposure to Pesticides, Washington, DC,
April 1998.
-------
U2 U. S. EPA Standard Operation Procedures (SOPs) for Residential Exposure
Assessments, Draft, Contract 68-D4-W6-0030, Work Assignment 3385.102, Versar,
Inc., prepared for the Residential Exposure Assessment Work Group, Office of
Pesticide Programs, Health Effects Division, Washington, DC, July 1997.
VI T. Vial, B. Nicolas, and F. Descotes, "Clinical Immunotoxicity of Pesticides." J.
Toxicol. Environ. Health, 48, 215-229 (1996).
Author suggests potential risk, especially during chronic exposure or to
immunocompromised persons such as those who are old, malnourished or young, but
evidence is scarce.
W1 N. K. Wilson, J. C. Chuang, and C. Lyu, "Multimedia Concentrations of PAH in
Several DayCare Centers." Polycyclic Aromatic Compounds, in press (1999).
W2 N. K. Wilson, J. C. Chuang, and C. Lyu, "Exposures of Nine Children Who Attend
Day Care to Persistent Pesticides and Other Organic Pollutants." J. Expos. Anal.
Environ. Epidem. (1999), to be submitted for publication.
W3 N. K. Wilson, J. C. Chuang, and M. R. Kuhlman, "Sampling Polycyclic Aromatic
Hydrocarbons and Other Semivolatile Organic Compounds in Indoor Air." Indoor Air,
4,513-521 (1991).
W4 N. K. Wilson and J. C. Chuang, "Indoor Levels of PAH and Related Compounds in an
Eight-Home Pilot Study." In M. J. Cooke, K. Loening, and J. Merritt, Eds.,
Polynuclear Aromatic Hydrocarbons: Measurements, Means, and Metabolism,
Battelle Press, Columbus OH, 1991, pp. 1037-1052.
W5 N. K. Wilson, J. C. Chuang, and C. Lyu, "Evaluation of Field Methods for Estimating
Exposure of Children in Low-Income Families to Polycyclic Aromatic Hydrocarbons."
Measurement of Toxic and Related Air Pollutants: Proceedings of the 1996
EPA/AWMA International Symposium, Pub. VIP-64, AWMA, Pittsburgh PA, 1996, pp.
797-802.
W6 N. K. Wilson, J. C. Chuang, and C. Lyu, "Measurements of Persistent Organic
Chemicals in Several Day Care Centers." Presented at the 1997 annual meeting of the
International Society of Exposure Analysis, Research Triangle Park NC, November
1997.
W7 N. K. Wilson, J. C. Chuang, and C. Lyu, "Multimedia Microenvironmental
Concentrations of PAH in Day Care Centers." Presented at the 16th International
Symposium on Polycyclic Aromatic Compounds, Charlotte NC, November 1997.
W8 R. W. Whitmore, F. W. Immerman, D. E. Camaan, A. E. Bond, and R. G. Lewis,
"Nonoccupational Exposure to Pesticides for Residents of Two U. S. Cities." Arch.
Environ. Contam. Toxicol., 26, 47-59 (1994).
-------
W9 L. A. Wallace, "Human Exposure to Environmental Pollutants: A Decade of
Experience." Clin. Exper. Allergy, 25, 4-9 (1995).
W10 B. Weiss, "Pesticides as a Source of Developmental Disabilities." Mental Retardation
Developmental Disabilities Res., 3, 246-257 (1997).
Organochlorine pesticides have been linked with developmental
neurotoxicity; the evidence for organophosphate pesticides is ambiguous.
W11 N. K. Wilson, J. C. Chuang, and C. Lyu, "Persistent Pesticides and Other Organic
Pollutants in Multiple Environmental Media in Nine Day Care Centers." Journal article
to be submitted for publication.
W12 R. S. Whiton, C. Witherspoon, and T. J. Buckley, "A Modified HPLC Method for
Determination of Polycyclic Aromatic Hydrocarbon Metabolites in Human Urine." J.
Chromatog., B 665, 390-394 (1995).
Z1	S. H. Zahm and S. S. Devesa, "Childhood Cancer—Overview of Incidence, Trends, and
Environmental Carcinogens." Environ. Health Perspect., 103, 177-184 (1995).
8000 child cancer occur annually in the US. There is a well-established link with
one EDC (diethylstilbestrol). Some pesticides are possibly EDCs and some data
suggest higher susceptibility in children.
Z2 S. H. Zahm and M. H. Ward, "Pesticides and Childhood Cancer," Environ. Health
Perspect., in press (1998). Presented at the U. S. Environmental Protection Agency
Conference on Avoidable Causes of Childhood Cancer, Arlington VA, September
1997.
Z3 V. G. Zartarian, J. Streicker, A. Rivera, C. S. Cornejo, S. Molina, O. F. Valadez, and J.
O. Leckie, "A Pilot Study to Collect Micro-Activity Data of Two- to Four-Year-Old
Farm Labor Children in Salinas Valley, California." J. Expos. Anal. Environ. Epidern.,
5,21-34(1995).
Methods were developed to videotape activity patterns of children. Four children
in farm labor families were studied. Questionnaires the day after taping and
comparison with the videos tested the hypothesis that recall is inadequate for
specifying children's activity patterns. However, the presence of the observers did
alter the children's behaviors to some extent.
-------
Appendix A
Data Analysis and Modeling
In the CTEPP study, multiple persistent organic pollutants (POP) in multiple chemical
classes, including pesticides, will be measured in multiple environmental media. Separate
chemical analyses will be carried out for each target compound in each environmental medium.
In some cases comparisons across target compounds and compound classes will be carried out,
for example to determine whether or not the relative exposure distribution across pathways is
consistent across compounds.
I. Design
The CTEPP design is a probability-based, stratified sample. Stratification will be
performed at the various sampling stages for the two sample components, the day care sample
and the telephone sample. These two sample components will be selected independently in the
primary strata defined by state (NC and OH). Additional first-stage stratification will be urban
and rural area classification performed at the county level. At subsequent sampling stages, the
sample will be also stratified by income levels: low income versus middle/upper income.
However, the main analysis groups (or domains) need to be distinguished from design
strata. Key study estimates will be computed for groups defined by:
•	Children in urban versus children in rural households
•	Children in low-income versus children in middle/upper income households
•	Children who attend day care versus children who do not
Whenever possible, comparisons will also be made across these groups. For example, we will
compare the exposure of urban and rural children, and similarly, of low- and middle/upper-
income children. However, we will not compare finer cross-classes such as:
•	urban low-income versus urban middle/upper-income, or
•	low-income day care versus low-income home care children, or
•	income subgroups of the day care component.
It should also be qualified that the small study sample sizes will not support definitive
comparisons about different subgroup exposures.
Nevertheless, the multivariate regression modeling approach will adjust for potential
effects of these covariates while estimating the effects of each independent variable. The
independent variables for these models will include each of these dimensions (as dummy two-
level variables).
A target sample of 128 children and associated caregivers in each state will be obtained.
This sample will be balanced evenly between the day care and no day care components. The
sampling design will over-sample the low-income strata, but sample sizes will still be smaller for
the low-income than for the middle/upper-income group overall. Sample sizes will also be
smaller in the rural than in the urban stratum.
The sample design provides a compromise in sensitivity between analytical inferences
and population-based inferences. On one hand, a more representative sample provides more
nearly proportional representation of these groups rather than equal representation. Proportional
representation would be more appropriate if the analytic goals were restricted to population-
Appendix A, Page 35
-------
based inferences. On the other hand, equal group sample sizes maximize the power of group
comparisons, an important element of analytical inferences.
Similar trade-offs arise in the weighting procedures, briefly discussed in Section 4.2, and
in the use of weights in the analyses. For the model-based analyses, weights will not be
necessary, especially when aggregating sample components where survey weights may exhibit
high variability.
2.	Analysis Objectives
2.1	Total exposure estimates
The first major objective of the CTEPP study is to measure the total exposures of the
study subjects to a suite of persistent pesticides and other persistent organic pollutants (POP).
We will characterize children's total exposures to POP and their urinary biomarker
concentrations within the individual strata and compare exposures and biomarker concentrations
across strata. One factor at a time comparisons, averaged across all other factors, will be carried
out. No comparison between the two states will be performed.
These comparisons will include, for example:
•	Urban versus rural
•	Day care versus no day care
•	Low income versus middle/upper income
•	Children versus adults in the same household
2.2	Exposure pathway component analyses
Another major objective of the CTEPP study is to apportion exposure pathways and
identify the important media where exposure can occur. The major component exposure
pathways are inhalation, dietary ingestion, non-dietary ingestion, and dermal absorption. We will
compare the components of exposure across the major pathways, averaging across all the
stratification factors. Examples of such comparisons include:
•	Ingestion - Dietary versus all others. We will determine the proportion of total
exposure attributable to diet (food, beverage, and drinking water).
•	Dietary ingestion versus non-dietary ingestion
•	Comparisons of percentage distribution of exposure by pathway across the target
compounds to be monitored
We will also study how the various exposure pathway components differ across the key
subpopulations. Examples of such analyses include:
•	Percentage of exposure by pathway between low income and middle/upper
income children
•	Percentages of exposure by pathway between
-	Urban and rural	- Day care and home care
-	Children and adults residing in the same household.
The survey is designed to facilitate the realization of these analysis objectives.
3.	Calculation of Daily Potential Exposure
The exposure concentrations in the various media, along with activity patterns, dietary
patterns, and various physiological and body size parameters will be combined using the EPA
microenvironmental exposure model (MEM) [E2, EPA Exposure Factors Handbook, 1996] to
Appendix A, Page 36
-------
derive estimates of daily exposures through the various pathways: inhalation, dietary and
nondietary ingestion, and dermal absorption. The EPA micro environmental exposure model is
displayed below.
The microenvironmental exposure model converts exposure values (ng/day) for
inhalation and ingestion (dietary and nondietary) to units of maximum potential (internal) dose
by assuming 100% absorption in the lung and digestive tract and normalizing for body mass.
Various factors can be found in the literature to account for physical, chemical, and/or
physiological processes. For maximum estimates, this conversion gives upper limits on the
amount of a pollutant available for delivery to target organs. In subsequent refinement of the
exposure estimates, literature absorption factors for the targeted compounds will be used as they
become available.
The MEM estimates potential daily dose of a target compound in ng/kg body mass per
day using the following equations:
D
inh
v_
w
D,
t. * D. + t * P
I	I	0	i
0
M x lOOQ
W
n
D
Cf * Mf * 1000
d
W
Appendix A, Page 37
-------
where
Dinh =	the estimated daily dose through inhalation, ng/kg
Dn =	the estimated daily dose through nondietary exposure, ng/kg
Dd =	the estimated daily dose through dietary exposure, ng/kg
IV	=	the body weight of the subject, kg
Ci =	indoor air concentration, ng/m3
C0 =	outdoor air concentration, ng/m3
tt =	subject's time spent indoors, min
t0 =	subject's time spent outdoors, min
V	=	the estimated subject's daily ventilation rate, 20 or 15 m3
D, =	target compound concentration in the floor dust, |ig/g
P0 =	target compound concentration in the play area soil, |!g/g
M =	subject's estimated daily dust/soil intake, 0.06 or 0.1 g
C) =	target compound concentration in the daily food samples, |!g/kg
Mf =	the daily mass of food intake, kg
For those target compounds that are likely to be absorbed through the skin surface, an
additional increment to the total daily exposure may occur. As a first approximation, this can be
estimated as:
Ddenn=Mw x (Aexp / Aw) x ( Fderm / IV)
where
Ddenn	= the estimated dose through dermal absorption, ng/day
Mw	= the mass of a target compound in the wipe sample, ng
2
Aexp = the exposed skin surface of the subject, m
Aw = the area of the skin from which the wipe sample was taken, m2
Fderm = the fraction of the compound that can be absorbed through the skin.
Although exposure factors for children and adults are available in the literature for
inhalation and soil ingestion, and for dermal absorption of some compounds, there are
uncertainties in these factors, which are especially large for the dermal and nondietary ingestion
routes of exposure. It is commonly assumed that the ventilation rate is 20 m3/day for adults and
15 m3/day for children, and that the dust/soil ingestion rate is 0.06 g/day for adults and 0.1 g/day
for children (EPA, ibid). These factors will be used in the initial model. As refined estimates
become available, they will be incorporated in the model.
The results of applying the potential daily dose model discussed above, and also in the
study design, will be a vector of estimated daily component intake doses (ng/kg) for each subject
via inhalation, dietary ingestion, nondietary ingestion, and dermal absorption. The overall daily
total dose is the sum of these component doses.
All of the target pollutant concentrations in multiple sample media (air, dust, soil, food,
and wipe), activity patterns, food and beverage intake profiles, and physiological parameters
specified in the microenvironmental exposure model will be determined for each child in the
study. Thus the comparisons of the children's total potential doses or component potential doses
across pathways or across strata will be carried out based on the totality of pathways considered
in the microenvironmental exposure model, namely inhalation of indoor and outdoor air, non-
dietary ingestion from dust and soil through hand-to-mouth activity, dietary exposure, and dermal
exposure through contact with the floor or other surfaces.
Appendix A, Page 38
-------
The comparisons of potential dose between children and their caregivers will also be
addressed. In the households where the adult caregivers stay at home with the children (half the
adults sampled, about 128), virtually the same suite of measurements and characteristics will be
determined for the adults as for the children. This includes the target pollutant concentrations
determined in the various media, food and beverage profile diaries, and urine biomarkers. The
determinations for the caregivers are made with the same frequency and at the same times of day
as for the children. The only difference in data collection is that activity profiles for the children
are recorded in activity diaries, whereas those for adults are assessed in post-monitoring
interviews. In the households where the child attends day care and the caregiver works outside
the home during the day, the caregivers' activities, exposure, diet, and urine biomarkers are
measured when the caregivers are at home, but not during the portion of the day when they are at
work. The daily exposure information must thus be interpreted as the portion of exposure that
can be attributed to the residential environment.
Therefore comparisons between exposures to children and their caregivers will be
made separately within the day care and stay-at-home strata. The comparisons can be made
based on the full set of media-specific responses, but for those adults who work outside the
home, their exposures are interpreted as the residential component.
4. Data Analysis
The principal data analyses to satisfy the primary and secondary objectives will be
based on the microenvironmental model-derived total daily intake dose and its component
pathway doses. These analyses will be discussed below. It should be noted that this discussion
does not imply that these will be the only uses of the data. It is very likely that the data will be
used for many additional analysis purposes and to develop, refine, and evaluate additional
mechanistic exposure and dose models. If it is found that the nature of the statistical
comparisons differs from one state to the other, separate comparisons will be done by state. Note
that no state-to-state comparisons will be performed.
	4.1 Preliminary Analysis. Distributional Assumptions and Outlier Testing
The four microenvironmental model-based, component-specific intake (potential) doses
and the corresponding total intake dose for each child will be organized into a multi-factorial
structure, each factor corresponding to a stratification factor at two levels. The models may
include other important variables that will be measured for children and households in the study.
These variables can be selected based on a step-wise regression procedure. Distributional
assumptions will be made and outlier detection procedures will be carried out separately for each
of the estimated component doses and for the total dose.
	— Distributional Assumptions. Test for Normality. Most environmental pollutant
concentrations, including POP concentrations, have been reported in the literature to conform to
lognormal distribution models [Ott, Environmental Statistics and Data Analysis, CRC Press,
1995]. Consequently goodness-of-fit of the derived pathway-specific component doses and the
external total dose to the lognormal distribution will be assessed. For each target compound and
for each response we will calculate the (natural) logarithm of the dose and fit a fully saturated,
homogeneous variance analysis of variance model to this data matrix, modeling all main effects
and all interactions. This corresponds to making a simple mean value correction within each
cell. If no compound is detected for a component concentration, we will set the concentration
equal to half the detection limit and proceed as if it is a measured value. For purposes of this
Appendix A, Page 39
-------
preliminary analysis, all responses will be treated as if they were independent. Studentized
residuals (i.e., residual divided by the standard error of the residual) will be calculated based on
the analysis of variance fit. The studentized residuals will be plotted on a normal probability
scale to assess conformance to a normal distribution model. Goodness-of-fit tests for normality
will be carried out using the Wilk-Shapiro (W) test. When applied to large amounts of data,
goodness-of-fit tests can be very sensitive to minor departures from distributional assumptions.
If the test for normality is not significant (p>0.05), we will accept the normality assumption and
conduct further analyses based on a lognormal distributional model. If the normality test is
significant (p<0.05) we will assess the shape of the normal probability plot to see if the
departures from straight-line behavior are minor or substantial. Sometimes one or more apparent
outliers can result in significant departures from normality based on the goodness-of-fit test.
Thus we will identify all studentized residuals in excess of three in absolute value. If one or
more are isolated values, that is, they are removed from their nearest neighbors, we will treat
them as tentative outliers. By contrast, if the entire probability plot departs from straight-line
behavior, it will be treated as a departure from normal distributional assumptions. In the latter
case we will assess the assumption of constant variance by use of Levene's test on the absolute
residuals. If appropriate we will relax the assumption of homogeneous variance and will refit the
analysis of variance model.
If the departure from normality can be explained by one or a small number of outliers or
by heterogeneous variances then we will adopt a normal distribution model in the logarithmic
transforms of the responses. Otherwise subsequent tests of hypotheses will be carried out based
on a rank transformation of the responses.
	— Outlier Detection. All studentized residuals in excess of three in absolute value will be
considered to be tentative outliers. The concentration values or activity or physiological
parameter values contributing to these extreme values will be reviewed in the basic data records
for correctness. If errors are found, they will either be corrected or the outlier will be deleted. If
no errors are found, the extreme responses will be considered to be natural variation and will be
retained in the data set.
4.2 Survey Weights and Weight Adjustments
Most of the statistical analyses will use weighted survey data. As pointed out in Section 1
of this appendix, however, unweighted data will be used for the model development and
validation parts of the analyses directed at analytical, model-based inferences rather than
population-based inferences.
Survey weights will account for unequal sampling probabilities and reduce potential
biases due to nonresponse. Sampling weights are needed for unbiased estimation under the
sampling design. Nonresponse adjustments will force estimates based on participating children,
for key characteristics, to match those of the entire sample or the entire universe of eligible
children.
The first step in weighting the data is the computation of sampling weights that account
for the varying probabilities of selection for different subgroups of children. Sampling weights
will be computed separately for the two sample components (or frames). In each component, the
probabilities of selection will be unequal due to stratification and disproportionate sampling (i.e.,
oversampling of certain strata). In addition, the day care sample weights may also need to reflect
selection with probabilities proportional to size (PPS), a method that may be used for the first-
stage sample of centers. Sampling weights will be computed as the reciprocal of the probabilities
Appendix A, Page 40
-------
of selection at each sampling stage, and assigned to sample day care centers and children. The
weights assigned to participating children will also be assigned to their parents and households.
These weights will be used in analyses of measurement data collected on a parent- or household-
level.
The next step in weighting consists of non-response adjustments. We plan to use
weighting class (and/or post-stratification) adjustments that make use of class totals known either
on a frame basis, that is, population cell totals; or on a sample basis, that is, for non-participating
as well as for participating households. For example, weighting classes may be based on design
strata, such as counties/states, income groupings and rural versus urban. Non-response
adjustments, designed to reduce the potential bias of non-response, will force the weights for
responding households to sum to known population totals within each cell (post-stratum, or
weighting class).
	4.3 Adjustments for Cluster Sampling
The day care center portion of the sample selection will be based on a two-stage cluster
sample. The sample size estimates are based on the assumption of 14 participating day care
centers, from six counties, in each state. From these 14 day care centers, 64 day care children
within each state will be sampled (an average of 4.6 children per participating center). Each day
care center may be considered a cluster, so the children sampled from the same day care center
would be anticipated to have correlated responses. A substantial portion of the children's 24-
hour daily exposures is obtained from the same day care center. The indoor and outdoor air
concentrations, food concentrations, floor sample concentrations, and play soil concentrations
obtained at the day care center would be the same for all these children. Furthermore, children
attending the same day care center may live near each other and share similar living conditions.
This would lead to correlated target compound concentrations at home.
The correlation in response among children in the same day care center will be estimated
by incorporating day care center as a random effect in the analysis of variance model. This will
lead to two components of variation, a2adc and a2e. The variance component a2adc corresponds to
variation among day care centers. The variance component a2e corresponds to the variance
within day care centers. The correlation between two responses in the same day care center is p
= °2adc /(°2adc + °2J- The correlation reflects the cluster design effects and will be incorporated
into the analyses. Separate values of p will be calculated for each response and each target
compound.
The analysis of variance model will also include the systematic factors (state, urban vs.
rural, low vs. middle/upper income, day care vs. no day care) and possibly other important
covariates that may be selected based on statistical criteria such as a step-wise regression
procedure. Note that while a factor for state will be included in the model, no comparison
between the two states will be performed. If results of the comparisons between urban and rural,
low-income and middle/upper income, and/or day care and non-day-care are found to rely on the
state, these comparisons will be made for each state separately.
4.4 Descriptive Statistics
Mean values and corresponding standard errors will be presented for each stratification
factor level (for example, urban, middle/upper income) and for pairs of factors (for example,
urban, and middle/upper income), averaged over the remaining factors. The sample means will
be weighted averages, using the survey weights as discussed in the section on weight
adjustments. Thus the sample averages will estimate the population averages within each
Appendix A, Page 41
-------
stratum or combination of strata. The standard error estimate will account for the survey weights
assigned to each observation, as well as the correlation among responses for children who attend
the same day care center. Children who do not attend day care centers provide independent
responses. Weighted averages will be based on the logarithms of the responses. Confidence
intervals will be calculated about the logarithmic mean under the assumption that the weighted
averages are approximately normally distributed. The weighted averages and associated
confidence bounds will then be exponentiated to provide inferences in the domain of physical
relevance.
4.5	Comparisons to Satisfy Total Exposure Objective
The first analysis objective is to compare estimated total exposure across one stratum at a
time (except for state), averaging across the remaining strata, for example, total exposure for low
income compared to total exposure for middle/upper income, averaging across urbanicity, day
care attendance status, and state). Comparisons to satisfy the objective will be based on the
doses calculated using the microenvironmental exposure model. A test of hypothesis of equality
of average total doses between the two levels within the stratum will be carried out by comparing
weighted averages and corresponding standard errors, using a two-sample, two-tailed t-test. If
the goodness-of-fit test for normality and the normal probability plot based on the logarithmic
transforms of the responses do not show serious departures from normal distribution
assumptions, the comparison between levels will be based on weighted averages of the
logarithmic responses within each stratum level and associated standard errors. The standard
error calculation will account for weighting and for the correlated responses among children who
attend the same day care center. If serious departures from normal distribution assumptions
occur, the t-test comparisons between average values in the stratum levels will be based on a
weighted mean of ranked responses within each stratum level and associated standard errors.
The weights will be the same as those used with the logarithmic transformation, namely the
survey weights. The estimated correlation among the children who attend the same day care
center will need to be re-estimated, based on the rank transformation data.
Significance levels of the t-tests will be reported. If the t-test is based on the
logarithmically transformed responses, the exponentiated mean difference (i.e. ratio of geometric
means) and associated 95 percent two-sided t-statistic confidence interval bounds will be
reported. If the t-test is based on rank transformed responses, just the exponentiated weighted
logarithmic mean difference (i.e. ratio of geometric means) will be reported.
4.6	Comparisons to Satisfy Pathway Apportionment Objectives
These objectives pertain to comparing the components of total dose by pathway. This
entails comparing the components of dose to one another, averaging across all the strata or
assessing the interaction between components of dose and stratum level. For example,
comparing the ratio of dose attributable to dietary ingestion with that attributable to non-dietary
ingestion, either averaged across all the strata or else comparing the ratio between rural and urban
children.
Such comparisons will be carried out in a similar fashion to the total exposure
comparisons. Let D;, D- denote the i-th and j-th pathway specific component of dose
respectively, or alternatively let D- denote the total dose, summed across pathways. Let R;j =
Dj/Dj, the ratio of D; to Dr If D; and D- are each approximately normally distributed and R,,
ranges from 0 to °° then X;j may be approximately normally distributed. Suppose R;j is bounded
from above by u (e.g. if D- = DT0T then u=l). In that event transform R to X = log [R/(u-R)].
Appendix A, Page 42
-------
Then < X <°° and X may be more nearly normally distributed than is R. We will calculate a
weighted average of the X's (if the X's are approximately normally distributed) or of the signed
ranks of the X's (if the X's depart from normality) across the factor combinations. The weights
will be the survey weights, as discussed in the section on weight adjustments. The standard error
calculation will reflect the survey weights, as well as the correlation among the X's (or the signed
ranks of the X's) for the children who are attending the same day care center. We will also
calculate a two-sided t-statistic based confidence interval based on the weighted average and its
associated standard error. The transformation on the weighted average and on the upper and
lower confidence bounds will be inverted to obtain an estimate of the weighted mean of the R's
and associated confidence bounds in a physically meaningful scale.
For inferences about the two-factor interactions between relative doses in different
exposure pathways and strata we will compare the ratios of external pathway specific doses
among strata. Let RM = Dj/Dj be as defined above. We wish to compare the average value of the
R;j's between two strata. For example we may wish to compare the ratio of dietary dose to total
dose between low income and middle/upper income children. We proceed as above, for the
primary analysis objective, either parametrically or non-parametrically, depending on the
, Ry
approximate normality of X; = log(	:—). We calculate the weighted averages within each
stratum, using survey weights as discussed above, and the corresponding standard errors of the
mean. To compare the ratio of dietary external dose to total external dose between low income
and middle/upper income children we test the hypothesis H(l: |!lo = |imu versus H,: |!lo * |!mu where
|ilo and |imu are the population means of the Xu within each stratum, by two-sample, two-sided t-
tests.
Appendix A, Page 43
-------
NOT DOING THIS SECTION AND BEYOND
IV Uses of CTEPP Data to Evaluate and Refine Curren tly Available Exposure Models
The CTEPP data set will include data from 256 children and their caregivers pertaining to
concentrations of target compounds in indoor and outdoor air, dust, soil, handwipes, food,
beverage, and drinking water; in addition to activity diaries, food diaries, and physiological
information. These data used in combination can provide information to evaluate the
performance of existing exposure models and to extend current exposure models to facilitate
their more common use, and with a markedly reduced input data burden. Two examples of such
applications of the CTEPP data are presented below. The first involves the critical evaluation of
a dietary exposure model. The second shows how the data obtained from the CTEPP study can
be used to extend the EPA microenvironmental exposure model to markedly reduce the burden
of collecting required input data.
5.1	Evaluate the EPA/OPP DEEM Dietaiy Exposure Evaluation Model
Each child's food concentration of each target compound maybe estimated using the
food consumption diaries and the EPA/OPP DEEM model. The predicted dietary residue
concentrations can be directly compared to the measured concentration levels in the food based
on the duplicate plate analysis. It will be determined whether the results are significantly
different. The questionnaire data can be used to determine whether are there environmental or
home factors, such as cooking, washing and other practices that influence the residue levels
found in the food samples. The questionnaire data can also be used to determine if the
differences between observed and predicted vary by socio-economic status (SES), housing or
measured indoor/outdoor environmental conditions. These and other similar questions will be
addressed using the CTEPP measurements or the questionnaire and survey data. Deviations
between the DEEM model and the measured dietary concentration data can be used to modify the
model.
5.2	Extend the EPA Microenvironmental Exposure Model
The EPA microenvironmental exposure model that is discussed in this section requires
extensive monitoring inputs, both indoors and outdoors, personal activities, and detailed food
consumption data. The CTEPP data set can be used to determine relations between indoor air
and dust concentrations and outdoor concentrations which are much simpler and less costly to
obtain, as well as activity patterns in the house. Potential models include:
Ct = p0 + PjPUA7 + p2PUA14 + p3PUA30 + p4C0 + P5FREQ + P6PETS + p7FANS + pgAC +
PgCLEAN + P10COOKING + pnINDSOURCES + ERROR
Ct = p0 + PjPUA7 + p2PUA14 + p3PUA30 + p4C0FREQ + 1 x £ ppQ + ERROR
FREQ
C„= P„ + P,f,C,+ PJfaCd +ERROR
where, (fs, fd = period of outdoor soil or indoor floor/surface dust contact time)
Ch = p0 + P.IDEX + P2IDEdCd + ERROR
Cd= Po + PA + ERROR
Appendix A, Page 44
-------
Cd = P0	X CLEAN + ERROR
FREQ
These models relate indoor concentration values to outdoor concentrations and activity based
variables. Incorporation of these relations into the micro environmental model will simplify the
burden of data collection considerably. An index of dermal exposure (IDE) will be calculated
using the reported level of the child's activity (high, medium or low) in connection with potential
contact with contaminated surfaces based on diary and/or video tape data. IDE score will range at
the minimum from 0 to 3 more likely from 0 to 6 depending on the range and type of activities
noted or observed. Likewise, we will define an incidental ingestion exposure ( HE) score based
on reported hand to mouth and object to mouth activities. A number of these variables will be
used as class (indicator) or interaction variables in the predicted regression models. Pesticide use
(PU) information will be coded by type and application history. Most recent application within a
week will be denoted by the variable PU7. Earlier applications more than a week but less than 2
weeks ago, and more than a month ago, will be denoted by the variables PU14 and PU30,
respectively. These variables will be used in the models. We define:
Q
Indoor air concentration
c0
Outdoor air concentration
Cd
Indoor dust concentration
Cs
Outdoor soil concentration
Ch
Handwipe concentration
FREQ	Frequency of window and door openings
PETS	Presence of pets in home (0, 1)
FANS	Use of ceiling fans (0, 1)
COOKING Cooking source and type (0, 1,2,...)
AC	Use of central air-conditioning and type (0, 1,2,..)
INDSOURCE Potential other non-cooking indoor sources
CLEAN High, low, average amounts of cleaning activities or measures (e.g.,
doormats, shoes removed indoors, etc.)
PUA	PU x Area of pesticide applied
5.3 Evaluate Physical Dermal Exposure and Dose Models
Physical and mechanistic models of dermal exposure models will be developed using the
concentrations obtained in the CTEPP study and transfer coefficients (TC) either derived from
the results of the CTEPP study or presently available either in the literature or in OPP's SOPs.
Therefore, with CTEPP data we have the opportunity of either: 1) developing these physical
models, or 2) evaluating these models against the exposure data generated directly or indirectly
during the study. Some examples of these model applications and evaluations are listed below.
Note that the superscript "p" refers to predicted and "m" refers to measured exposure (E) or dose
(D). ED donates exposure duration and "C" refers to dislodgeable surface or carpet
concentrations. Again, time lags are denoted by subscripts. Since a multiplicative model is used
in the prediction of dermal exposures, log transforms will be used to convert these equations to
an additive regression model form.
EPderm = C x TC x ED (Predicted dermal exposure or potential dose)
Appendix A, Page 45
-------
Ch=po+ p1lnCd+ p2lnTC+ p3ED + ERROR
^ ^air -^food/water ^non-diet ^dermal
^ dermal ^PBPK (-^ ~ ^air ~ ^food/water ~ ^non-diet)
where aPBPK is the coefficient or function based on PK or PPBK modeling relating
exposures to dose and vice-versa.
Dno„.dle, = Cd x IIEd x Kd + Cs x IIES x Ks
where Ks and Kd are the incidental soil or dust ingestion estimates used in the equations
described earlier or obtained from literature.
lnEmdermal = P0 + P.lnC + P2lnTC +P3ED + ERROR
Consequently, we can estimate TC also from:
TC = exp(lnEmdermal - InC - InED)
or
TC = exp(lnCh - InC -InED)
Other pesticide application time lag models may be considered if these models prove to have
acceptable predictive power.
5.4 Compare Exposures and Urinary Biomarker Data
Predicted total exposures and potential dose will be compared statistically to measured
urinary biomarker data in order to evaluate the predictive power of the exposure models
developed. Biological dose will be estimated using a PK-based average absorption and metabolic
conversion rate, as well as in a few cases direct application of a PK model to the estimated
exposure/dose profile [for example, modeling trichloropyridinol (TCP) concentrations in urine
associated with multimedia exposures to chlorpyrifos]. Predicted route and pathway-specific
exposures will be contrasted as well as summed over to estimate total potential exposures by
different study sub-groups. Biomarker measurements will also be correlated with other
behavioral and home and day care potential exposure factors. Use of pesticides in homes and day
care centers, proximity to busy roadways, cooking, hobbies and cleaning activities which could
result in higher indoor and consequently personal exposures will be examined statistically.
Stepwise regression models or CART techniques will be used as exploratory models to examine
the likelihood of various factors that could lead to elevated exposures and absorbed dose.
Children's behavioral characteristics or activities, such as walking barefoot, digging in yard or
playground soil with measurable or elevated pesticide levels, sleeping on the floor, low or high
hand washing frequencies, habitual thumb sucking, etc. will be analyzed as part of this
investigation.
IV Predict Intake Doses With Urinary Biomarker Concentrations
The environmental concentration measurements, activity diaries, and food consumption
diaries are difficult, time-consuming, and expensive to obtain. These data however are needed to
Appendix A, Page 46
-------
obtain estimates of the component pathway external doses and thereby total external dose. In
contrast, urinary biomarker concentrations are relatively quick, easy, and inexpensive to measure.
They can be obtained for very much larger numbers of children and adults than can full suites of
indoor and out environmental measurements. The extent of the ability of the urinary biomarker
information to serve as a surrogate indicator of the external component and total doses can be
assessed based on the CTEPP data.
The correlation between urinary biomarker data and each of the component pathway
doses or total aggregate dose can be assessed by a series of simple regression analyses. Let U;,
Dtot ; denote the urinary biomarker concentration and the total external dose respectively within
the i-th combination of factors. We would like to determine how well U predicts DT0T. We
consider a succession of simple linear regression models.
Dtot;I — Po ^ PiU; +e
D-roT.i — Po; + PiU; +e
Dtot;I — Poi ^ PiiU; +e
to determine how well U can act as a surrogate for DT0T, either within particular combinations of
strata or across strata. These relations might also be expressed in terms of the logarithmic
transformations of DT0T and U. The coefficient of variation and the residual variation about the
model indicate how precisely U can predict DT0T. 95 percent 2-sided prediction intervals can be
constructed to determine upper and lower prediction bounds on DT0T, conditional on observing
U.
Other factors contained in the activity and food diaries can be added to the predictive equations
to determine whether they enhance the predictiveness of the relations. Several such factors
discussed in the paragraph above, such as pesticide use or cleaning agent use, could enhance the
total dose. Component doses can be substituted for DT0T in the above relations to determine how
well U can act as a surrogate for a component dose.
IV Additional Analyses
The above discussion on data summarization and the construction of tests of hypotheses
and point and confidence interval estimates dealt with statistical displays and procedures
designed specifically to address the primary and secondary statistical analysis objectives
specified at the beginning of the section. The data set will be very rich and many additional uses
will be made of the data.
We will utilize the data (questionnaires, diaries, and POP concentrations in multimedia)
generated from the CTEPP study to evaluate and refine U.S. EPA Standard Operation Procedure
(SOPs) for Residential Exposure Assessments (Draft Contract No. 68-W6-0030, Work
Assignment No. 3385.102, prepared by The Residential Exposure Assessment Work Group:
Office of Pesticide Programs, Health Effect Division Versar, Inc. July 18, 1997). For example,
the child activity diary data obtained from the CTEPP study can be used to refine the parameters
used in the SOP 2.3.2 "Postapplication Potential Dose Among Toddlers from Incidental
Nondietary Ingestion of Pesticide Residue on Residential Lawns from Hand-to-mouth Transfer."
We will convert all the collected data from the CTEPP study into EPA TherDBase format. The
data will be accessible easily to EPA OPP to evaluate and refine the SOPs for Residential
Exposure Assessments.
The previously discussed analyses dealt with estimates of total exposure and component
exposure and their comparisons among strata. Additional information obtained from the activity
Appendix A, Page 47
-------
diaries and the food diaries can be used to explain variations in exposure components among
children, either within strata or across strata. Thus, for example, if inhalation exposure or dietary
exposure were found to be major components of total exposure, the identification of factors
associated with high values of these component exposures can lead to specific recommendations
concerning their reduction. For example type, frequency, and amount of pesticide use may be
correlated with indoor or outdoor air concentration, which in turn is correlated with inhalation
exposure. It might also be related to increased concentrations in food preparation surfaces, which
in turn would lead to increases in the ingestion exposure. The food diaries will provide
information about types and amounts of foods eaten. This may be a predictor of target
compound concentrations in daily food intake.
Appendix A, Page 48
-------