EPA/600/R-99/086
May 2000
ASSESSMENT OF HEALTH EFFECTS OF
PESTICIDE EXPOSURE IN YOUNG CHILDREN
National Health and Environmental Effects Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
EPA/600/R-99/086
May 2000
ASSESSMENT OF HEALTH EFFECTS OF PESTICIDE EXPOSURE
IN YOUNG CHILDREN
Proceedings of a Workshop held in El Paso, Texas in December 1997
Edited by
David Otto, Rebecca Calderon, Pauline Mendola and Elizabeth Hilborn
Epidemiology and Biomarkers Branch
Human Studies Division
National Health and Environmental Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
National Health and Environmental Effects Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
DISCLAIMER
This document has been reviewed in accordance with the U. S. Environmental Protection Agency' s
peer and administrative review policies and approved for publication. The views expressed are
those of the participating scientists and should not be construed as representing any Agency
position. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use
11
-------
TABLE OF CONTENTS
Page
Preface iv
Executive Summary v
Introduction 1
Presentations
Issues in Pediatric Epidemiology — Robert Bornschein, Ph.D 3
Assessing Neurobehavioral Effects of Environmental Toxicants on Children: Options
and Issues — David C. Bellinger, Ph.D 5
Do Pesticides and Other Environmental Exposures Play an Important Role in the
Development of Diseases of Immune Dysregulation along the United States
Mexican Border? — Anthony A. Horner, M.D 17
Evaluation of Developmental Neurocognitive and Neurobehavioral Changes Associated
with Pesticide Exposure: Recommendations for the U.S. Environmental
Protection Agency Workshop on the Assessment of Health Effects of
Pesticide Exposure in Infants and Young Children — Antolin M. Llorente, Ph.D 22
Pesticides and Children: Pulmonary Outcome Measures — Maria D. Martinez, M.D 33
Pesticides and Childhood Cancer - An Overview — Jonathan Buckley, M.D 37
Health Effects of Pesticides - D.J. Ecobichon, Ph.D 45
Increased Sensitivity to Pesticides in Young Children: Possible Mechanisms-
Stephanie Padilla, Ph.D 57
Design of Children's Pesticide Exposure Survey—Jim Quackenboss, Ph.D 59
Pesticide Usage Along the U.S.-Mexican Border—Gerry Akland 79
Pesticide Use and Assessment along the Arizona Border—Mary Kay O'Rourke, Ph.D 85
Issues in Studying Populations along the U.S.-Mexico Border —
James VanDersIice, Ph.D 102
Some Observations on Studies of Pesticides and Children on the U.S.-Mexico
Border - Rob McConnell, M.D 113
Resources for Pediatric Research in the Border Region - James Ellis, M.D 116
Workgroup Reports
Introduction 118
Neurobehavioral
120
Developmental 127
Immunology and Pulmonary 141
Cancer 151
Day 3: Summary of Group Discussion 156
Appendices
Agenda 159
List of Attendees/Contributors 161
Workshop Discussion Groups 165
-------
PREFACE
The mission of the U.S. Environmental Protection Agency (EPA) is to protect human health
and safeguard the natural environment. An area of special concern is the U.S.-Mexican border that
stretches nearly 2,000 miles from the Pacific Ocean to the Gulf of Mexico. Mexico and the U.S.
have been working together since the 1983 La Paz Agreement to solve the environmental and health
problems in the border region. An Environmental Health Workgroup (EHW) was established in
1996 under the auspices of the Border XXI program specifically to address these concerns. The
EHW is co-chaired by the EPA Office of Research and Development, the U.S. Department of Health
and Human Services, and the Mexican Secretariat of Health.
Much of the border region is devoted to the cultivation of fruits and vegetables. As a
consequence, the potential health risks of chronic, low-level pesticide exposure in young children
were identified as a high binational priority concern of the EHW. Of special interest are the risks
to very young children from persistent, multipathway multipesticide exposures. Recognizing the
challenges in studying these risks in young children, a number of projects have been initiated to
provide basic tools, methodologies and data to guide the design and conduct of future epidemiologic
studies by the EHW and other interested organizations.
To assist in meeting this priority objective, the National Health and Environmental Effects
Laboratory of the EPA sponsored a workshop on December 4-6, 1997, in El Paso, Texas, that
focused on the identification of health effects associated with exposure to pesticides and how those
effects might be measured in very young children. Thirty participants were divided into five
working groups to review research methods in neurotoxicity, developmental toxicity,
carcinogenicity, immunological effects and respiratory effects. Participants included leading
scientists in these five fields, local health care providers, state public health officials and researchers
from EPA and the Centers for Disease Control and Prevention. This report contains reviews of
state-of-the-art methods currently available in each of these areas for field studies of young children,
health effects associated with pesticide exposure in children, and recommendations from each of
these working groups for research on the health effects of pesticide exposure in young children.
Recommendations will provide guidelines for future EHW projects addressing the assessment of
exposure to pesticides and potential health risks to children living in the border region.
Harold Zenick, PhD
Associate Director for Health,
National Health & Environmental Effects Research Laboratory,
U.S. Environmental Protection Agency
and
Co-Chair, Environmental Health Workgroup
IV
-------
EXECUTIVE SUMMARY
Background
Pesticides are a broad group of chemicals used to kill insects, fungus and undesirable plant species;
they are by design, biologically active compounds. Millions of pounds are applied in agricultural,
industrial, institutional, commercial, and residential settings within the United States (US) each year.
Pesticide exposure is ubiquitous via contamination of food, soil, air, and water, yet the health effects of
chronic, low-dose exposure are unknown. There is growing concern that current tolerance levels may not
be sufficient to protect the health of children.
Children are at greater risk than adults for increased exposure to pesticides. Hand-to-mouth activity
increases the risk of pesticide exposure by the oral route. Children have a proportionally greater surface
area and tend to have more dermal contact with soil and indoor floor surfaces than adults. Children
consume more food and beverage as a portion of their body weight than adults. Foods such as fruits and
fruit juices may form a large part of a young child's diet and represent a major source of pesticide exposure.
Children may be particularly vulnerable to the effects of pesticides due to rapidly maturing organ and
nervous systems. Children's metabolism of pesticides is not well understood. Differences in
biotransformation and elimination may result in children experiencing a greater toxic effect than adults.
Children along the US-Mexican border may be at increased risk for pesticide exposure due to the
prevalence of year-round agriculture in this region. Research efforts to determine if there are measurable
health effects associated with chronic low dose exposure to pesticides will initially be focused in the border
region as part of the Border XXI Program.
During December 7-9 1997, the Environmental Health Workgroup sponsored a workshop on the
assessment of health effects of pesticide exposure in young children to discuss the current state of the
science, and to identify priorities for future research.
Workshop Structure and Goals
Experts from a variety of fields participated in the workshop. The three-day meeting was structured
to include discussion of pesticide exposure measurement, potential health effects in the pediatric
population, and current research efforts in the US-Mexican border area. Small workgroups were organized
to address five health endpoint domains: cancer, neurobehavioral, respiratory, immunologic, and
developmental effects. After day one, the respiratory and immunology groups combined to discuss areas
of subject overlap. Workgroup members were selected to represent multiple areas of expertise:
epidemiology, exposure assessment, clinical medicine, and those with experience conducting research
along the US-Mexican Border.
Day One
The first day focused on issues related to the study of children. Dr. Bornschein gave the keynote
address: 'Issues in Pediatric Epidemiology' based on his work with lead-exposed children. Dr. David
Bellinger discussed potential methods for the neurobehavioral assessment of children at different stages
of development. Dr. Anthony Homer's presentation focused on published studies of immunotoxic effects
of pesticides in human and animal models. Dr. Antolin Llorente reviewed the neurotoxicant effects of
pesticides in adults and the need for well-designed studies of children exposed during periods of rapid
neurologic development. Dr. Maria Martinez discussed techniques available to measure pulmonary
function in children of various ages. Dr. Jonathan Buckley presented information about the study of
children diagnosed with cancer and highlighted the difficulties associated with the evaluation of pesticide
exposure assessment in these children.
-------
Workgroup Reports: Day One - Health Effects
The focus of the workgroups on Day One was to identify lists of health endpoints that were relevant
and measurable in young children and infants. There is very little documentation of health effects related
to low-dose pesticide exposure in young children, so workgroup members extrapolated from their
knowledge of occupational exposure studies, acute toxicity reports, and animal studies.
Cancer: In the United States, cancer is the second leading cause of death for children between the ages of
one and 14 years. Although overall cancer rates have generally been declining, the rate of childhood cancer
has increased in North America. Specific cancer diagnoses such as: acute lymphoid leukemia, tumors of
the central nervous system and bone would be of particular interest in the study of environmental
exposures.
Developmental: Exposures that may occur during the prenatal period or infancy may have the greatest
impact on the developing child. The following developmental health endpoints are listed in priority order
as defined by the workgroup: birth defects, stillbirths, and spontaneous abortion; mental, motor, adaptive
development; growth; language; birth weight related to gestational age; social development; infant
mortality; puberty, age at menarche and development of secondary sex characteristics.
Neurobehavioral: The workgroup identified assessment tools appropriate for evaluation of children in
various age groups. The Bayley Scales of Infant Development, Wechsler Preschool and Primary Scales
of Intelligence-Revised, the Wide Range Assessment of Visual Motor Abilities, Wide Range Assessment
of Memory and Learning, Peabody Developmental Motor Scales, Visual acuity, Wechsler Intelligence
Scales for Children, 3rd ed., visual contrast sensitivity, and Neurobehavioral Evaluation System all received
priority rankings.
Immunology: The workgroup identified immunologically-associated health endpoints of interest. These
included: asthma (reactive airway disease); allergy; primary immunodeficiency; contact dermatitis; lupus
erythromatosus; inflammatory bowel disease; infectious diseases; and adverse reproductive outcomes.
Respiratory: The respiratory workgroup discussed both the utility of validated disease endpoints and self-
reported symptomatology in assessing overall pulmonary health. • The workgroup discussed four
respiratory diseases: upper respiratory infection, acute bronchitis, asthma, and interstitial lung disease.
Day Two - Research issues
The keynote address, 'Health Effects of Pesticides' was delivered by Dr. Donald Ecobichon. He
discussed the health effects of acute pesticide poisonings in adult agricultural workers. He emphasized that
the research challenge will be to develop methods to measure subtle psychologic, behavioral, and
neurologic deficits in children exposed to lower doses of toxic mixtures of the 'inert' and active ingredients
in pesticide formulations. Dr. Stephanie Padilla reported that her laboratory is investigating the effects of
sub-lethal doses of chlorpyrifos on young rats. Young, postnatal rats are more sensitive to organophosphate
pesticides than adults. Differences in levels of detoxification enzymes may account for some of this
observed effect.
Dr. Jim Quackenboss discussed the design of a Children's Pesticide Exposure Survey. He
discussed that one of the major difficulties of research in the field of health effects of pesticides on children
has been the difficulty in selecting 'high' exposure individuals from the general population. Mr. Gary
Robertson reported the results of a survey of pesticide use near the US-Mexico border. Methods used to
evaluate pesticide usage were different in each state; some states collect actual usage data, in others, usage
was estimated from agricultural crop records and acreage under cultivation. Dr. Mary Kay O'Rourke
vi
-------
discussed her current exposure assessment projects. These include: the National Human Exposure
Assessment Survey, a survey of residents along the US-Mexican border in Arizona, a Children's Pesticide
Survey in Yuma, Arizona, and multiple projects requested by communities along the border, studying
health effects such as asthma, diabetes mellitus and lupus erythromatosus. Dr. Jim VanDerslice raised
issues related to studying populations along the U.S.-Mexican Border. He stressed that although the Border
Region is referred to as a single entity, it is actually a very diverse collection of communities along a 2000
mile long corridor. Dr. Rob McConnell discussed cultural considerations in the conduct of epidemiologic
studies along the U.S.-Mexican Border. Dr. James Ellis highlighted potential resources for pediatric
research in the Border area. He emphasized the need to build trust between researchers and community
members before research is initiated.
Workgroup Reports: Day Two- Development of Strawman Study Proposals
The second day of the workshop was designed to integrate selected health endpoints identified
during the Day One workgroup sessions into a collection of potential study designs for implementation
along the border region.
Cancer: The workgroup outlined and discussed several possible types of studies including:
1) Use of existing cancer data bases.
2) An ecological study could compare pesticide usage in border and non-border regions and determine if
there is a difference in cancer patterns in these areas.
3) A case-control study: cancer cases could be obtained from clinics and hospitals.
4) A prospective cohort study: exposure would be measured with the use of a biomarker and incident cases
of cancer recorded.
5) A case control design could identify children with leukemia and determine if they have a higher level
of V(D)J recombinase mediated chromosomal rearrangements.
Developmental: Proposed studies were classified as analytic, descriptive or capacity building.
Analytic Studies
1) A prospective prenatal cohort. The study hypothesis would be: pesticide exposure is related to delayed
and/or altered development and long term developmental problems.
2) A poisoned children case study. The study hypothesis would be: there are persistent neurobehavioral and
neurodevelopmental sequelae of acute pesticide exposure.
3)A prospective closed-cohart study of symptomatic children. The study hypothesis would be: there are no
developmental differences between symptomatic children with detectable urinary metabolites of
organophosphate pesticides and symptomatic children without detectable urinary metabolites.
Descriptive Studies
1) A cross-sectional study of any correlation between levels of pesticides, anticholinesterase, and related
enzymes in maternal and infant biologic samples.
2) A descriptive cross-sectional study using a Geographic Information System approach of infant health
status. The main hypothesis is that infant mortality and birth weight are not different in areas with high
agricultural pesticide use compared to geographic areas with lower agricultural pesticide use.
Capacity Building
1) Pesticide Dose: a summary of pesticide dose information in young children is needed. 2)Adaptation of
neurodevelopmental tests to populations within the border region.
vn
-------
Neurobehavioral: The Neurobehavioral workgroup discussed three study designs--(l) a retrospective
cohort design, (2) a cross-sectional study, and (3) a longitudinal cohort study. The basic hypothesis
addressed by these studies is that exposure to pesticides produces neurotoxic effects in children.
1) Retrospective Acute, High-exposure Cohort Study. A retrospective cohort study of a group of children
with clearly defined, high-level exposure will be selected for an initial study to determine whether or not
pesticide exposure produces neurotoxic effects in young children.
2) Cross-sectional Chronic, Low-exposure Study. An exposure questionnaire will be administered to
parents of children aged 1.5-2.5 years to select three groups—high, middle and low exposure deciles (10%).
The Bayley Test is recommended for neurobehavioral assessment of children. Exposure measures should
include house dust and urine samples for biological measures.
3) Longitudinal Cohort Study. 100 children living in a high-risk area could be selected. The Bayley Test
would be administered at 3-month intervals for one-two years. Urine samples should be obtained at each
testing for measurement of OP levels, metabolites and a-esterases.
Immunology / Respiratory: The group agreed on some study designs to evaluate the association between
pesticide exposure and immunological and pulmonary health effects.
1) Cross-sectional study, questionnaire derived exposure combined with self-reported health endpoints.
Exposure assessment supplemented by GIS and some environmental sampling. Hypothesis: the prevalence
of asthma and other diseases will be higher in individuals with increased pesticide exposure.
2) A cross-sectional study based on exposure status. The study hypothesis: pesticide exposure increases
the incidence of and/ or exacerbates pre-existing asthma.
3) A methacholine challenge test to objectively assess airway reactivity would be administered to a group
of healthy children. A case - control study would follow with case status assigned to those with airway
hyper reactivity. The study hypothesis is that pesticide exposure contributes to airway hyper reactivity.
4) Cross sectional study of children < 1 years of age as a pilot for a longitudinal study of a birth cohort.
The study hypothesis: pesticide exposure affects the development of the immune system in infants resulting
in altered antibody response to vaccine administration and increased incidence of infectious disease.
Day 3- Group discussion
The following seven research priorities were assembled based on reports from workgroups and
individual participants' comments during the day's discussion.
1) The development of efficient methods for screening children for exposure status
2) Questionnaire development and validation in Border communities
3) Targeted environmental sampling to increase efficiency
4) Validation of biochemical measures of exposure
5) The need to establish normal ranges for health endpoints
6) The development of sophisticated modeling techniques to more accurately predict the health effects of
exposure to multiple pesticides by multiple exposure routes
7) Studies must have adequate power to detect subtle pesticide-associated health effects
Workshop members stressed that public health officials and health care providers from the border
community are requesting better exposure measurements. They would like to know the extent of
environmental pesticide contamination, would like to know the potential health effects associated with
pesticide exposure, and if current levels of pesticide exposure are causing health problems.
vm
-------
ACKNOWLEDGMENTS
Many people have contributed to the creation of this document. We would particularly like to
express our appreciation to Marcia Gardner (SRA Technologies) who assembled the initial draft
and Jennifer Hawks, Vickie Worrell, Shanika and Tasha Rogers, Shalaunda Johns and Sully
Jaffer who assisted in word processing.
IX
-------
INTRODUCTION
Over the last 30 years, the border region between the U.S. and Mexico has experienced a dramatic
increase in population and industrialization. This growth has exceeded the existing infrastructure capabilities
of the region, leading to inadequate sewage treatment, insufficient drinking water supplies, and dramatic
impact on habitat and biodiversity along the border. In order to address environmental problems associated
with growth, the U.S. and Mexican governments signed the Agreement for the Protection and Improvement
of the Environment in the Border Area (La Paz Agreement) in 1983. The La Paz Agreement defined the
border region as the area lying 100 kilometres north and south of the U.S.-Mexican border. The Integrated
Environmental Plan for the Mexican-U.S. Border Area (IBEP) released in 1992 extended the scope of the
La Paz Agreement to include environmental health and natural resource issues. Passage of the North
American Free Trade Agreement (NAFTA) in 1993 extended U.S.-Mexican activities along the border,
including creation of the Commission for Environmental Cooperation. The Border XXI Program, operating
under these mandates, is a comprehensive program designed to achieve a clean environment, protect public
health and natural resources, and encourage sustainable development.
Much of the border region is devoted to agriculture and aerial pesticide spraying is widespread. A
major concern is the risk from repeated, often year-round pesticide exposure. The problem is complicated
by exposure to multiple pesticides from different sources (residential as well as agricultural) and multiple
pathways (food, water, air), the cumulative impact of which is unknown. Of particular concern are young
children (from birth to age five) whose developmental vulnerability puts them uniquely at risk. The
Environmental Health Workgroup, part of the Border XXI Program, has identified this concern as a high
priority research issue and initiated an extensive project—Pesticide Exposure and Health Effects in Young
Children along the U.S.-Mexican Border—to assess the nature and extent of this problem.
Phase I of this project was a survey of pesticide usage along the border. Planning of health surveys
and a subsequent epidemiological health effects study are also in progress and will be refined on the basis
of results obtained in earlier phases of the project. There is a paucity of data on the health effects of pesticide
exposure in humans in general and in young children in particular. Most available health effects data are
from case reports of accidental or intentional acute poisonings. Little evidence is available on chronic, low-
level pesticide exposure in populations living in agricultural areas. Health effect studies of young children
are further complicated by language and behavioral limitations. The advice of experts from a variety of
disciplines including psychometrics, developmental psychology, immunology, pulmonology and oncology
concerning the health endpoints likely to be most sensitive to pesticide exposure in humans and which tests
can realistically be administered to young children (five years and younger) will be crucial for planning
purposes.
An important objective of this workshop was to review and evaluate appropriate endpoints for use
in health effect studies of young children exposed to pesticides. Other objectives included: (1) examination
of the existing infrastructure to support such studies in the border region; (2) identification of possible
populations for study; and (3) recommendations for possible study designs. Participants in the workshop
were assigned to workgroups corresponding to the disciplines considered relevant for pesticide research in
children—i.e., psychometrics, developmental psychology, immunology, pulmonology and cancer. The
proceedings include papers by invited speakers on health endpoints appropriate for use in studies of young
children, issues specific to pediatric research in the border region, and recommendations of the individual
break-out groups from the workshop.
-------
PRESENTATIONS
-------
KEYNOTE ADDRESS: Issues in Pediatric Epidemiology*
Robert Bornschein
Department of Environmental Health
University of Cincinnati
The keynote address focused on important issues in determining the potential relation
between an environmental exposure and an adverse health effect in young children. Dr.
Bornschein drew from his experience investigating health effects associated with pediatric lead
exposure. His presentation included a historical perspective on improvements in exposure
measurement and the determination of relevant health endpoints.
The first topic addressed was environmental exposure assessment in young children.
Standardization of data collection, preparation and analyses was stressed. Sampling should be
uniform and based on the child's behavior to achieve the best estimates of dose. For example,
environmental sampling should be targeted in areas (e.g., rooms, yard, etc.) where the child
frequently spends time. Internal and external quality control including use of standard reference
materials, bench-top controls, external reference laboratories and proficiency programs are
necessary to optimize exposure assessment. Limiting variability and laboratory error is critical in
evaluating low level environmental exposures. National databases such as the National Health
and Nutrition Examination Survey can provide general population exposure prevalence estimates
for selected compounds. Screening data may also be available from county health departments
and through the Centers for Disease Control and Prevention, particularly from the Morbidity and
Mortality Weekly Report.
Evaluation of measurements that serve as markers of exposure and effect was discussed.
There is a continuum from external dose to the development of exposure-related disease.
External dose — Internal dose -- Biologically effective dose — Biologic response — Disease
It is a big step from external dose to biologic response. Many assumptions are made
based on data derived from adjacent steps in the continuum, but it is important not to extrapolate
too far. For example, lead in housedust correlates poorly with a child's blood lead level, but
hand wipes from the child will be more strongly related to both blood lead and housedust than
they are to each other. So, in this case, without knowing the concentration of the intermediate
step (hand related exposure) there does not appear to be a relation between housedust lead and
blood lead in children. This is an important lesson as we begin investigating the pathways of
pesticide exposure in young children.
Exposure models are needed to estimate body burdens in children and pregnant women.
These models should address multiple pathways of exposure and multiple exposure sources.
Empirical models are the most desirable such as the Integrated Exposure Uptake Biokinetic
model for lead exposure and physiologically-based pharmacokinetic exposure models. It is
important to realize that typically you can explain very little of the variance associated with body
burden of an environmental exposure. For example, an exhaustive model for blood lead in
children will typically explain less than 20% of the variance in levels. Analytic variance is
-------
always higher than the explained variance. Unexplained variance may be attributed to poor
measurement of diet, air, hobbies or job exposures, prior accumulation of a body burden due to
earlier exposures, hematocrit, and other measurement error. Additionally, the relation between
the source of the marker (e.g., urine) and the tissue where the compound is deposited or
concentrated (e.g., bone, fat) should be known. The clinical sample results should be interpreted
based on the physiology and kinetics of the substance in the body. Temporal considerations are
also needed. Microenvironments change frequently over time and recent measurements may not
be reflective of long term levels. Kinetics can also change over time in young children. Calcium
in bone, for example, changes over three times in the first year of life with a much slower
calcium exchange rate later in childhood.
Moving from the measurement of exposure and dose to childhood adverse health
outcomes, Dr. Bomschein talked about the difficulty in measuring child development. A simple
"one exposure leads to one outcome" model may work when there is a large high-dose exposure
and the associated outcome is severe. This is typically not the case for many environmental
exposures. In a multiple main effects model, each factor is considered an independent
contributor to risk (e.g., socio-economic status, parental education, etc.) for poor child
development. An interactive model is probably closer to reality, where factors are related to each
other as well as having an impact on the outcome under study. For example, socio-economic
status is related to maternal IQ, child rearing practices, nutrition, etc.., all of which can interact
with each other as well as potentially exert an independent effect on child development.
A variety of study designs are appropriate for investigating the potential health effects in
young children associated with an environmental exposure. Dr. Bomschein discussed using
cross-sectional and longitudinal studies of child development. Data analyses require particular
attention to the treatment of missing data, particularly in the case of exposure measurements
below the analytic limit of detection (LOD). Exposure data are not typically normally distributed
and decisions about how to handle values below the LOD (e.g., make them zero) can have a
major influence on the analyses. Structural equation models can be very useful because they are
designed to model complex interactions and a variety of pathways in the analysis.
This is a summary of the presentation by Dr. Robert Bornschin
-------
Assessing Neurobehavioral Effects of Environmental Toxicants on Children:
Options and Issues
David C. Bellinger
Children's Hospital (Boston)
Harvard Medical School
Because many environmental toxicants interfere with the development of the central
nervous system, a comprehensive assessment of the health effects of such substances in the
pediatric population require valid and reliable methods of evaluating neurobehavioral function.
Considerable effort has been invested in developing batteries of methods for assessing such
endpoints in adults, resulting in options such as the Neurobehavioral Core Test Battery (Anger et
al, 1993), the Adult Environmental Neurobehavioral Test Battery developed by the ATSDR
(Amler et al., 1995), and the computer-administered Neurobehavioral Evaluation System (Letz,
1991). In contrast, only one battery has been developed specifically for use in studying the
neurobehavioral impact of community-level exposures on children, the Pediatric Environmental
Neurobehavioral Test Battery (PENTB) (Amler et al., 1996). As a result, the assessment
batteries used to investigate neurotoxicant exposures in children have tended to be study-specific,
even in the case of the set of prospective lead studies, which were characterized by considerable
interaction among the major research groups (Bornschein and Rabinowitz, 1985; Grant, Smith,
and Sors, 1989).
The evaluation of neurotoxicant effects in children involves some challenges less
germane to adult assessment, due largely to the rapid pace of development over the first few
years in the nature and breadth of the response modalities which can be exploited as "windows"
onto the status of children's skills in different functional domains. Development is, by
definition, a "moving target," requiring the discrimination of changes in behavior that reflect a
toxicant effect from changes that reflect normal development. The assessment of children, in
contrast to adults, thus requires consideration of the added dimension of time, in that exposure to
a toxicant may affect the rate at which a skill emerges as well as the form which that skill takes.
As a result, the question, "What is the appropriate age at which to conduct assessments" requires
careful weighing of competing considerations. Carrying out assessments within the first year of
life has important advantages. Because children's neurobehavior is affected by many factors,
isolating outcome variation uniquely attributable to a specific exposure from the variation
attributable to other, often correlated (confounding) factors can pose a formidable challenge
(Bellinger et al., 1989). Because some of these factors are not strongly associated with children's
neurobehavioral status prior to approximately 18 months of age (Golden and Birns, 1982), their
impact can be minimized by assessing children in infancy. In addition, the shorter the interval
between exposure and assessment, the fewer the opportunities for the occurrence of intervening
medical and social events that might change outcome in ways that are independent of the
exposure. On the other hand, the ways in which an infant can "tell" us what he or she currently
"knows" are limited, impoverishing the empirical bases on which to draw inferences about the
likelihood of adverse impact from neurotoxicant exposure. In addition, prior knowledge of the
site or mechanism of a toxicant's effects on the neural substrate may lead to the expectation that
its primary impact will be on a domain of function that cannot easily be assessed in infants, such
as "executive" functions (planning and organization), abstract reasoning, or reading. Thus,
although a neurotoxicant may have greatest impact on brain systems undergoing the most rapid
-------
change at the time of exposure, the status of such systems may not yet be clearly reflected in
observable behavior. If such systems represent the endpoint domain of greatest interest, it would
be necessary to delay follow-up assessments until such time that the response modalities
available to a child are sufficiently differentiated that a clear and reliable picture of a child's
strengths and weaknesses in these domains can be obtained. Thus, a decision regarding the age
at which assessments should be conducted in any particular study involves significant trade-offs.
The specific choices one makes among the large number of assessment instruments
available depend on several considerations. One critical factor is the primary goal of the
assessment. Is the goal, for instance, to understand the neuropsychological mechanism(s) of
neurotoxicant effects, or to estimate the public health or "real world" impact of neurotoxicant
exposure? If it is the former, one might select instruments that focus in detail on cognitive
"building blocks" (i.e., vigilance, visual-spatial skills, working memory), the component low-
level information processing skills that underlie higher-order functions. If the goal is the latter,
one might prefer instruments that assess practical skills such as reading, mathematical reasoning,
and oral expression. The results of such tests would suggest what "real world" skills may have
been affected, but provide relatively little insight as to why. Logistical factors also bear on the
selection of instruments, such as the length of time available for assessment of each child, and
the level of expertise of study personnel.
To some extent, a discussion of which measures are most suitable for use in
neurotoxicant studies can be pursued without specifying a particular neurotoxicant.
Neurobehavior represents the highest level of neurological organization, thus depending on the
integrity of many more basic processes. As a set of final common pathways, it would be
expected to reflect the impact of any neurotoxicant exposure, should the level of exposure exceed
the threshold of effect. Although two neurotoxicants may impair different combinations of
lower-level neurobehavioral processes, their overall impact might be expressed in the same
apical behaviors, particularly in infancy when the response options are so limited. It should not
be surprising that certain apical tests, which average performance over many functional domains,
have proven to be sensitive to compounds that vary widely in their chemical properties and
biological actions (e.g., heavy metals and persistent organic pollutants). Thus, apical
neurobehavioral measures appear to be quite sensitive to neurotoxicant exposures but not
specific.
The following sections discuss selected approaches to assessing the neurobehavioral
function of children of different ages, as well as general assessment issues germane to studies of
neurotoxicant exposures.
Infants (0-3 Years)
The two major approaches to the developmental assessment of infants involve sensory-
motor and information-processing tests. A third approach, based on informant reports, will be
discussed briefly. A fourth option, involving assessments of newborn behaviors using
instruments such as the Neonatal Behavioral Assessment Scale (Brazelton, 1984) will not be
discussed due to the difficulties in conducting and interpreting such assessments in the context of
an epidemiological study of neurotoxicant exposures (Dietrich and Bellinger, 1994).
-------
Sensory-motor (SM) Tests
Currently, SM tests are viewed as the "gold standard" for infant assessment. These tests
issue from what is referred to as the psychometric tradition. The selection of items is based on a
purely statistical criterion, i.e., the ability to discriminate among children of a certain age, rather
than on a set of principles or hypotheses about child development. Such tests simply provide for
a broad sampling of the behaviors children display at different ages. As a result, they permit
inferences about whether a particular child's development is progressing at an age-appropriate
pace and in an age-appropriate form. Because children's abilities change rapidly as they age, the
items administered to children of different ages differ considerably. For example, items
administered to a 3 month old child typically involve orienting and tracking, object
manipulation, and awareness of novelty. At age 14 months old, simple visual-spatial skills,
language comprehension, and elementary problem-solving skills are assessed. At 2 1/2 years,
more complex expressive and receptive language skills and abstract concepts are assessed. The
item sets administered to children at different ages essentially constitute qualitatively different
tests, which should give pause to one seeking to interpret multiple scores over time as true
repeated measures that define a developmental function. To some extent this is true of all
childhood assessments, but the differences are greatest during the years of infancy.
Although many SM-based tests are available, the one most commonly used is the Bayley
Scales of Infant Development (second edition) (BSID) (Bayley, 1993), which covers the age
range of 1 to 42 months. Its primary components are the Mental Development Index (MDI),
which assesses cognitive skills (e.g., memory, habituation, problem solving, early number
concepts, classification, language, social skills); and the Psychomotor Development Index (PDI),
which assesses gross motor planning, balance, ability to imitate postures, visual-motor
integration, and fine motor skills. In addition to global MDI and PDI scores, "facet" scores in the
following domains are also obtained: cognitive, language, personal-social, motor. The BSID
were standardized on a sample of 1700 U.S. children in which the distribution by race/ethnicity
mirrored that of the U.S. population.
One disadvantage of all SM tests warrants comment. Although they satisfy most criteria
by which the adequacy of psychological tests is evaluated (e.g., test-retest reliability, concurrent
validity), scores typically have low predictive validity, even over periods as brief as a few years.
For example, the median correlation between 1 year BSID scores and IQ at age 5 to 7 years in
low-risk children is approximately 0.1, although it is often considerably higher among infants at
medical risk (Kopp and McCall, 1982; Rose and Feldman, 1991). Test scores obtained after age
2 are usually more strongly related to later IQ scores, however, presumably because the skills
assessed more closely resemble those assessed by IQ tests. That young infants' scores on SM
tests have such low predictive validity is not necessarily a flaw in such tests but most likely
reflects the fact that developmental trajectories are dynamic and that the speed with which a child
achieves a criterion level of skill within one domain is not necessarily related to how quickly a
criterion level will be achieved in a different domain or even a later criterion within the same
domain. For example, is it reasonable to expect that the age at which a child masters visually-
directed reaching for an object will predict how well that child reads at age 7? Developmental
delay does not necessarily reflect developmental deviance and the two may differ in prognostic
significance. Scores on SM tests can be viewed as analogous to birth weight (McCall, 1979).
Although it does not predict school-age weight, neonatologists find it a useful indicator of a
newborn's current health. Similarly, scores on SM tests provide a valid measure of an infant's
-------
current developmental status. Nevertheless, a score on an SM test should not be considered an
estimate of "infant IQ."
Information-processing (IP Tests)
The poor predictive validity of SM tests has been attributed to dissimilarities in the
domains assessed by these tests and by IQ tests. It has been proposed that prediction from infant
behaviors would be improved if it were possible to evaluate close analogues of the skills later
called upon by IQ tests. Infant abilities considered likely candidates include elements of
information processing such as perception, discrimination, storage, retrieval, and classification.
Because performance on habituation and expectancy tasks involves such skills, several new
assessment procedures based on them have been developed. The best known among these is the
visual recognition memory (or novelty preference) procedure called the Fagan Test of Infant
Intelligence (FTII) (Fagan et al., 1986). Other methods are based on nonvisual information
processing (e.g., cross-modal transfer of information, tactual recognition memory; Rose et
al.,1992).
The FTII rests on the principle that an infant will usually prefer to look at something
novel rather than something familiar. It consists of 10 trials in which the infant is presented with
such a choice. Relying on corneal reflections, an observer records how the infant distributes
looking time to a novel and a familiar picture (all photographs effaces). In the last half of the
first year, infants allocate an average of 60% of time to a novel picture. In an attempt to combine
the best features of SM and IP tests, the most recent revision of the BSID includes several items
that assess visual recognition memory in 1 to 3 month olds.
To a limited extent, research bears out the hope that IP tests are more predictive of later
IQ than are SM tests (Fagan and Detterman, 1992). A meta-analysis of 31 studies calculated a
weighted correlation of 0.36 between novelty preference and preschool IQ (McCall and Carriger,
1993). Scores obtained between the ages of 2 and 8 months appear to be more predictive than
scores obtained later and, as with SM tests, preschool IQ scores of children at medical risk can be
predicted more accurately than IQ scores of low-risk children. The predictive validity of IP tests
does not exceed that of SM tests in samples of at-risk infants, however. Moreover, despite its
statistical significance, the prediction afforded by scores on IP tests only accounts for 15% of the
variance in preschool age IQ, a level of accuracy that is no higher than the level achieved by
relying solely on parental education and socioeconomic status.
An important limitation of the FTII is the narrow age window within which it is normed
(up to 12 months). Furthermore, whereas SM tests provide a broad characterization of an
infant's developmental status, IP tests provide information about a highly restricted set of
behaviors.
Informant-based Methods
Many parent-completed questionnaires on infant and child development are available,
developed in large-part to help clinicians identify infants who may require additional evaluations.
They vary in format and in the breadth of their coverage. One measure of general development
is the Child Development Inventory (Ireton, 1992), a 270-item questionnaire applicable to
children 15 months to 6 years. It assesses 8 domains: social, self-help, gross motor, fine motor,
expressive language, language comprehension, letters, and numbers. The Vineland Adaptive
Behavior Scales (Sparrow et al., 1984), administered as a semi-structured interview involving
8
-------
297 items, provides information on child behaviors in four domains: communication, daily living
skills, socialization, and motor skills.
Because of the lack of standardization in parents' experience, observation, and reporting
skills, such measures are generally not satisfactory for the purpose of evaluating subtle
developmental consequences of toxicant exposure. These factors may be less problematic for
questionnaires that focus intensively on salient domains such as language development and
behavior problems. If the items are sufficiently specific, parents can be extremely accurate
reporters. A set of two instruments, the MacArthur Communicative Development Inventories,
take advantage of this and can be used with children 8 to 30 months of age (Fenson et al., 1993).
The Vocabulary Production scale, for instance, asks a parent to indicate which words, from a list
of 680, a child currently uses. The Syntactic Complexity scale asks the parent to indicate which
of two alternative constructions (e.g., "I no do it," "I can't do it") sounds most like the way their
child speaks "right now." A widely used questionnaire for assessing behavior problems in 2-to-3
year old children is the Child Behavior Checklist (Achenbach et al., 1987), which asks a parent
to indicate the frequency (not true, somewhat or sometimes true, very true or often true) with
which a child engages in 99 specific problem behaviors.
Children (Ages 4 and Up)
For children ages 4 and above, the number of instruments for assessing neurobehavioral
status is enormous, precluding detailed discussion of individual tests. The key issue is not
finding the "correct" or "best" tests. As Bernstein noted, "...there is no single battery for
evaluating the potential impact of toxic agents on the developing child. I cannot recommend any
specific tests in this endeavor; many are appropriate. Overall strategy, a principled theoretical
framework, and adequately specified behavioral domains are what counts, not tests" (1994).
Among the available choices of general intelligence tests rooted in the psychometric
tradition are the Wechsler Preschool and Primary Scale of Intelligence-Revised (Wechsler, 1989;
ages 3-7 years 3 months), the McCarthy Scales of Children's Abilities (McCarthy, 1972; ages 2
years 7 months-8 years 7 months), the Stanford-Binet, 4th Edition (ages 2 years-adult), the
Differential Ability Scales (Elliot, 1990; 2 years 6 months-18 years), and the Kaufman-
Assessment Battery for Children (Kaufman and Kaufman, 1983; ages 2 years 6 months-12 years
6 months). The mixes of skills assessed by these instruments overlap considerably, although
direct comparison studies reveal that a child's scores on the various tests may vary by several
points. This is especially true in clinically-defined samples who present with difficulties
performing particular types of tasks, which may be represented more or less prominently on a
given test. Several of these tests have proven sensitive to various low-level neurotoxicant
exposures, including lead and PCBs.
Several "off the shelf test batteries are also available, although they have not been
widely used in studies of neurotoxicant exposures. Two examples are the Reitan-Indiana
Neuropsychological Test Battery (Boll, 1981) and the Luria-Nebraska Children's Battery
(Golden, 1981). A recently published test battery, the NEPSY, provides for a developmental
neuropsychological assessment of children ages 3 to 12 based on Luna's model. It includes 27
subtests that assess five functional domains: attention/executive functions, language,
sensorimotor abilities, visuospatial abilities, and memory/learning. It does not produce a
summary IQ-like score. Few published data are available on the NEPSY, although it is the
primary endpoint in an NIEHS clinical trial evaluating whether administration of the oral
-------
chelating drug succimer improves the long-term neurobehavioral outcome of lead poisoned
children.
As mentioned earlier, the Pediatric Environmental Neurobehavioral Battery (PENTB)
was developed specifically to evaluate the behavioral impact of community-level toxicant
exposures on children. Its primary goal is to provide a rapid, relatively inexpensive, and
reasonably comprehensive evaluation of children from 1 to 16 years of age. Among the factors
motivating the selection of tests were the need to limit an evaluation to one hour and to use tests
whose administration does not require professional supervision. Because of the expertise
required for infant assessment, the battery for children ages 1 to 3 involves only informant-based
methods. The 10 performance-based and four-informant-based instruments cover the affective,
cognitive, motor, and sensory domains. Affect is assessed using the Vineland Adaptive Behavior
Scales and the Personality Inventory for Children. Assessments of cognitive skills include the
Kaufman-Brief Intelligence Test, story memory (and delayed recall) from the Wide Range
Assessment of Memory and Learning, a divided attention test, the Developmental Test of Visual-
Motor Integration, and a verbal cancellation task. Motor skills are assessed using a finger
tapping task, the Purdue Pegboard, and the divided attention test. The tests of sensory status are
visual acuity, visual contrast sensitivity, and vibration threshold. Field studies demonstrate that
this battery of tests is acceptable to parents and children and can be administered reliably. Its
ability to detect subclinical impact of neurotoxicants is not yet known, although the battery
includes several measures shown in prior studies to be sensitive to such exposures. A useful next
step in research on the PENTB would be to determine whether the performance of children with
identified developmental or learning problems differs from that of normal controls. If the
PENTB does not discriminate such groups, its utility in detecting subtle neurotoxicant effects
would be drawn into question.
A major issue in assembling an assessment battery is whether to include a test of general
intelligence. On the one hand, such tests are familiar and the construct they measure is valued by
policy makers. In addition, because IQ is an endpoint that is easily monetized, it is suitable for
use in cost-benefit analyses. On the other hand, some investigators have advocated the strong
position that "...the general use of summary scores [e.g., IQ...] is both inappropriate and
unscientific" (White et al., 1994, p.513) in part because important exposure-related differences in
performance may be obscured when summary scores are employed as the outcome index. This is
because such scores, in effect, average an individual's performance over the multiple domains
assessed by an apical test. For instance, if exposure to a neurotoxicant impairs only visual-
spatial functioning, an exposure-related difference in the full-scale IQ scores might not be
apparent because only some of the subtests that contribute to full-scale IQ rely heavily on visual-
spatial skills. Similar arguments are often made by investigators who study primarily animal
models (Rice, 1993). Although this claim has intuitive appeal (and must necessarily be correct to
some extent), it has generally not been supported by the results of human epidemiological studies
of neurotoxicants. In lead studies, for instance, the most consistent finding across studies is an
inverse association between an exposure biomarker and full-scale IQ, but only limited
consistency in the associations between lead and scores on tests that focus on specific
neuropsychological domains (National Research Council, 1993). Furthermore, even within
specific studies, stronger associations have usually been found on IQ tests than on domain-
specific tests. In the Boston prospective study, for instance, whereas blood lead at age 2 and IQ
at age 10 were inversely related (Bellinger et al., 1992), only a chance number of significant
associations was found between blood lead and scores on tests such as the California Verbal
10
-------
Learning Test-Children, the Wisconsin Card Sorting Test, the Developmental Test of Visual-
Motor Integration, story recall, finger tapping, and grooved pegboard (Stiles and Bellinger,
1993).
The stronger associations found on apical tests than on focused tests could have a
toxicologic or methodologic explanation (or both). Whether one type of test is more sensitive
than the other may depend on the neuropsychological mechanism of toxicity. The performance
decrements associated with increased lead exposure could reflect the joint impact of several
independent impairments or impairment of a small set of key functions that underlie many
diverse cognitive abilities. If the latter, the result might be slight, nonsignificant decrements in
performance across diverse neuropsychological domains. If the decrements are summed in the
form of an apical score, such as IQ, a significant exposure-related decrement might be apparent.
Moreover, the specific form in which lead's impact on neurobehavior is expressed might differ
depending on a variety of host (e.g., age, sex, socioeconomic status) and contextual
characteristics (e.g., dosing regimen), what in the animal literature is referred to as the
"experimental system." Applied to human studies, this perspective suggests that exposure to a
particular neurotoxicant may not produce the same "behavioral signature" under all scenarios
(Bellinger, 1995a). If so, exposure-related effects will again be most apparent on tests that
average performance over many domains, reducing the impact of cohort-specificity in the
expression of toxicity.
Another reason why apical tests more reliably reveal exposure-related differences in
performance may be the greater strength of their psychometric properties as compared to
domain-specific tests. Unlike IQ tests, domain-specific tests were designed more to identify
individuals with clinically-significant deficits in certain types of skills than to discriminate
between levels of performance within the normal range. This difference in sensitivity will be
especially important when the effects to be characterized are subtle. Domain-specific tests may
be most useful for more highly exposed cohorts in which the neurobehavioral effects are
substantial and thus detectable using tests that are less sensitive than apical tests (e.g., Bellinger
etal., 1994).
Computer-administered Tests
Recently several tests included in the NES2 battery for adults (Letz, 1991) have been
adapted for use with children (Winneke et al., 1994; Otto et al, 1996; Dahl et al., 1996). These
computer-administered (or computer-assisted) tests are finger tapping, continuous performance
test, and hand-eye coordination, which assess aspects of motor response speed, sustained visual
attention and response latency, and motor coordination (tracking), respectively. Each of these
tests has a long history of use in clinical child neuropsychology. Their cost efficiency and the
standardization of administration, data capture, and scoring make these procedures attractive
options for use in epidemiologic field studies. In their present form, these tests may be most
appropriate for children 8 years and older as substantial numbers of the younger children in the
initial studies were unable to complete all tasks. In addition, visual contrast sensitivity may be
an important covariate of children's (and adults') performance on these three tasks (Hudnell et
al., 1996a), accounting for as much as (or more) of the outcome variance as does low-level
neurotoxicant exposures. It appears important, therefore, that visual contrast sensitivity be
measured in studies using NES2 tasks (and perhaps conventional, individually-administered tests
as well). First, it would reduce error variance in endpoint scores and thus increase the power of
hypothesis tests involving neurotoxicant exposure. Second, alteration in visual contrast
11
-------
sensitivity function may itself be a result of neurotoxicant exposure, representing a mechanism
for the exposure-related decrements seen in other functional domains (Hudnell et al., 1996b).
Informant-based methods. As with infants, many parent- or teacher-completed instruments are
available for identifying stable socio-emotional/behavioral characteristics that are not readily
assessed during a standard neurobehavioral evaluation. These include the Child Behavior
Checklist/4-18 (Achenbach, 199la), the Personality Inventory for Children (Wirt et al., 1990),
the Connors' Teacher and Parent Rating Scales (Connors, 1990), and Parent and Teacher Ratings
Scales of the Behavior Assessment System for Children (BASC) (Reynolds and Kamphaus,
1992). The CBCL/4-18 (Achenbach, 1991b; Achenbach, 1991c) and BASC families of
instruments also include teacher and self-report versions, providing the option to collect data
from multiple informants on a given child.
These methods should be considered as adjuncts to, not replacements for, individually-
administered assessments of neurobehavioral function.
Additional Issues
Identifying a "Behavioral Signature"
Several issues should be considered when attempting to infer the "behavioral signature"
of a neurotoxicant on the basis of differences in scores on a battery of tests. First, an individual's
score on a neuropsychological test does not reflect the integrity of a single functional domain or
region of the brain (Krasnegor et al., 1994). Successful performance on any test depends on
multiple skills. For instance, a low score on a test of design copying (e.g., the Developmental
Test of Visual-Motor Integration) may be due to problems with graphomotor control, visual
perception, planning and organization, motivation, or behavioral modulation (e.g., impulsivity)
(Bellinger, 1995b). Second, drawing inferences about the relative sensitivity of
neuropsychological domains to a neurotoxicant is problematic unless the tests used to assess
different domains are equivalent in their discriminating power, specifically their true-score
variance (Chapman and Chapman, 1978). If they are not, differential performance across
domains may not be due to differential ability across domains. Deficits may appear to lie in a
particular domain simply because the tools used to assess that domain are technically superior to
the tools used to assess other domains. One way to address this issue is to employ an instrument
that was designed to assess several related aspects of function. One example of such an
instrument is the Wide Range Assessment of Memory and Learning (Sheslow and Adams, 1990)
for children ages 5 to 17, which includes 9 subtests that assess different types of verbal memory
(story, sentence, number/letter strings), different types of nonverbal memory (picture, design,
sequence of actions), and different types of learning (verbal, visual, sound-symbol). Another
example is the Wide Range Assessment of Visual-Motor Abilities (Adams and Sheslow, 1995)
for children ages 3 to 17, which assesses visual-motor integration (design copying), visual-spatial
matching, and fine motor function (pegboard). Because the components of such tests are normed
on the same population, their reliability coefficients are directly comparable, permitting
inferences about the relative magnitudes of toxicant effects on the different aspects of function
assessed. The benefits of direct comparisons made possible by such co-norming also extend to
sets of instruments. For instance the Wechsler Individual Achievement Test (The Psychological
Corporation, 1992), which assesses academic skills such as reading, mathematics, and spelling,
was co-normed with the Wechsler intelligence tests (i.e., WPPSI-R, WISC-III, WAIS-R).
Because the most frequently employed criterion of a "learning disability" is a significant ability-
12
-------
achievement discrepancy, the paired use of the WIAT and the age-appropriate IQ test provides
solid psychometric grounds for identifying such discrepancies.
Quality Control
Although the need to implement a protocol for maintaining the quality of analytical
measurements of biomarkers is universally acknowledged (e.g., Coggon, 1995), implementing
analogous procedures for maintaining the quality of neurobehavioral measurements is equally
important although less frequently explicitly acknowledged. While this applies to all
neurobehavioral measurements, it is especially true with respect to infants. Developmental
testing of an infant occurs within the context of an interpersonal interaction that is loosely
scripted by prescribed administration procedures (e.g., number of demonstrations, number of
trials permitted). However, the examiner is also given latitude to determine "on the fly" the best
way in which to engage the infant in the evaluation. Unlike school-age children, infants have no
"test-taking set." They do not experience any social pressure whatsoever to cooperate with an
assessor, attending only to a task that engages their interest. If a child fails to perform a target
behavior, it may be because he or she has not yet developed the underlying skills or simply
because of a lack of motivation to perform it (or both). One of the assessor's major tasks is
establish a setting that increases the likelihood that a target response will be elicited, should it be
in the infant's repertoire. This requires considerable training and practice. Clear procedures
must be in place from the outset of a study for thorough training of testers and for periodic
evaluation of tester performance. These procedures should also include review of a tester's
proficiency in scoring responses and in deriving summary and standard scores from the raw data.
Measurement of Covariates
As final common pathways, neurobehavioral endpoints are multi-determined and
responsive to a variety of biological and sociological factors. In most studies, approximately
50% of the variation in IQ is "explained" by such factors. In ascertaining whether an exposure is
associated with neurobehavior, it is critical that information about these biological/sociological
factors be taken into account. This will serve two purposes. First, it will reduce the outcome
error variance, providing for a statistically more powerful test of the hypothesis that the
environmental exposure is associated with the endpoint. This is important because a low-level
environmental exposure is likely to explain a relatively small percentage of outcome variance. In
most lead studies it was less than 5%. Second, it provides the possibility of controlling for
confounding, which is present when one of the biological/sociological determinants of the
endpoint is also correlated with the exposure. In studies on low-level lead exposure, the factors
usually identified as critical confounders included family socioeconomic status, parental IQ, and
the quality of the home environment (as assessed by the Home Observation for Measurement of
the Environment; Bradley, 1994). The likelihood that confounding is a major issue, and the
specific variables that need to be considered as confounders, may be toxicant-specific, depending
in particular on the key exposure pathways.
Conclusion
Deciding which tests to use to assess the impact of a toxicant on children's neurobehavior
requires consideration of several factors, followed by efforts to reconcile incompatibilities in
their implications. Among the most important factors are the following:
(1) the goal of the assessment, e.g., to guide public policy or to clarify basic mechanisms
oftoxicity,
13
-------
(2) the design of the study, e.g., cross-sectional versus prospective,
(3) the children's ages at the time of assessment, e.g., infant versus school-age,
(4) the range of children's ages at the time of assessment, e.g., all less than 3 years of age
versus a broader span, such as 1 to 15 years,
(5) the length of time available for assessment, e.g., one hour versus three,
(6) the level of training of personnel available to conduct the assessments, e.g.,
bachelor's degree versus clinical neuropsychologists.
Whatever choices one makes in addressing one of these issues may well constrain the
range of options available for addressing others, requiring trade-offs in the breadth and quality of
the data that can be obtained.
References
Achenbach, T., Edelbrock, C., and Howell, C. 1987. Empirically based assessment of the
behavioral/emotional problems of 2- and 3-year old children. Journal of Abnormal Child
Psychology 15:629-50.
Achenbach, T. 1991a. Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT:
University of Vermont.
Achenbach, T. 1991b. Manual for the Teacher's Report Form and 1991 Profile. Burlington, VT:
University of Vermont.
Achenbach, T. 1991c. Manual for the Youth Self-Report and 1991 Profile. Burlington, VT: University of
Vermont.
Adams, W., and Sheslow, D. 1995. Wide range assessment of visual motor abilities. Wilmington, DE:
Wide Range, Inc.
Amler, R., Anger, W.K., and Sizemore, O. (Eds.) 1995. Adult environmental neurobehavioral battery.
Atlanta: Agency for Toxic Substances and Disease Registry.
Amler, R., and Gibertini, M. (Eds.) 1996. Pediatric environmental neurobehavioral battery. Atlanta:
Agency for Toxic Substances and Disease Registry.
Anger, W.K., Cassitto, M.G., and Liang, Y-X. 1993. Comparison of performance from three continents
on the WHO-recommended neurobehavioral core test battery. Environmental Research 62:125-47.
Bayley, N. 1993. Bayley Scales of infant development. 2nd. ed. San Antonio: The Psychological
Corporation.
Bellinger, D. 1995a. Interpreting the literature on lead and child development: The neglected role of the
"experimental system." Neurotoxicology and Teratology 17:201-12.
Bellinger, D. 1995b. Neuropsychologic function in children exposed to environmental lead.
Epidemiology 6:101-3.
Bellinger, D., Leviton, A., Waternaux, C., Needleman, H., and Rabinowitz, M. 1987. Longitudinal
analyses of pre- and postnatal lead exposure and early cognitive development. New England
Journal of Medicine 316:1037-43.
Bellinger, D., Leviton, A., and Waternaux, C. 1989. Lead, IQ, and social class. International Journal of
Epidemiology 18:180-5.
Bellinger, D., Stiles, K., and Needleman, H. 1992. Low-level lead exposure, intelligence, and academic
achievement: A long-term follow-up study. Pediatrics 90:855-61.
14
-------
Bellinger, D., Hu, H., Titlebaum, L., and Needleman, H. 1994. Attentional correlates of dentin and bone
lead levels in adolescents. Archives of Environmental Health 49:98-105.
Bernstein, J. 1994. Assessment of developmental toxicity: Neuropsychological batteries. Environmental
Health Perspectives 102:141-4.
Boll, T. 1981. The Halstead-Reitan neuropsychological battery. In: Handbook of clinical
neuropsychology (Filskov, S., and Boll, T., eds.). New York: Wiley, 577-607.
Bornschein, R., and Rabinowitz, M. 1985. Environmental Research 38: (entire issue).
Bradley, R. 1994. The HOME inventory: Review and reflections. In: Advances in child development and
behavior, Vol. 25, 241.
Brazelton, T. 1984. Neonatal behavioral assessment scale. Clinics in developmental medicine No.88.
Philadelphia; J.B. Lippincott.
Chapman, L., and Chapman, J. 1978. The measurement of differential deficit. Journal of Psychiatric
Research 14:303-11.
Coggan, D. 1995. Assessment of exposure to environmental pollutants. Occupational and Environmental
Medicine 52:562-4.
Connors, K. 1990. Connors' rating scales manual. North Tonawanda, NY: Multi-health Systems, Inc.
Dahl, R., White, R., Weihe, P., et al. 1996. Feasibility and validity of three computer-assisted
neurobehavioral tests in 7-year-old children. Neurotoxicology and Teratology 18:413-9.
Dietrich, K., and Bellinger, D. 1994. Assessment of neurobehavioral development in studies of the
effects of fetal exposures to environmental agents. In: Prenatal exposure to toxicants:
Developmental consequences (Needleman, H., and Bellinger, D., eds.). Baltimore: The Johns
Hopkins Press, 57-85.
Elliott, C. 1990. Differential ability scales. San Antonio: The Psychological Corporation.
Pagan, J., Singer, L., Montie, J., and Shepherd, P. 1986. Selective screening device for the early detection
of normal or delayed cognitive development in infants at risk for later mental retardation.
Pediatrics 78:1021-6.
Pagan, J., and Detterman, D. 1992. The Pagan Test of infant intelligence: A technical summary. Journal
of Applied Developmental Psychology 13:173-93.
Fenson, L., Dale, P., Reznick, S., Thai, D., Bates, E., Hartung, J., Pethick, S., and Reilly, J. 1993.
MacArthur communicative development inventories. User's guide and technical manual. San
Diego: Singular Publishing Group, Inc.
Golden, C. 1981. The Luria-Nebraska children's battery: Theory and formulation. In:
Neuropsychological assessment of the school-age child (Hynd, G., and Obrzut, eds.). New York:
Grune and Stratton, 227-302.
Golden, M., and Birns, B. 1983. Social class and infant intelligence. In: Origins of intelligence: Infancy
and early childhood, 2nd. ed. (Lewis, M., ed.). New York: Plenum Press, 347-98.
Grant, L., Smith, M., and Sors, A. (eds.). 1989. Lead exposure and child development: An international
perspective. Boston: Kluwer Academic Publishers.
Hudnell, H., Otto, D., and House, D. 1996a. The influence of vision on computerized neurobehavioral
test scores: A proposal for improving test protocols. Neurotoxicology and Teratology 18:391-400.
Hudnell, H., Skalik, I., Otto, D., House, D., Subrt, P., and Sram, R. 1996b. Visual contrast sensitivity
deficits in Bohemian children. NeuroToxicology 17:615-28.
Ireton, H. 1992. Child development inventory. Minneapolis: Behavior Science Systems, Inc.
Kaufman, A., and Kaufman, N. 1983. Kaufman-assessment battery for children. Administration and
scoring manual. Circle Pines, MN: American Guidance Service.
Kopp, C., and McCall, R. 1982. Predicting later mental performance for normal, at risk, and handicapped
infants. In: Life-span development and behavior, Vol.4 (Baltes, P., and Brim, O., eds.). New York:
Academic Press, 33-61.
Krasnegor, N., Otto, D., Bernstein, J., Burke, R., Chappell, W., Eckerman, D., Needleman, H., Oakley,
G., Rogan, W., Terracciano, G., and Hutchinson, L. 1994. Neurobehavioral test strategies for
environmental exposures in pediatric populations. Neurotoxicology and Teratology 16:499-509.
Letz, R. 1991. NES2 User's manual (version 4.4). Winchester, MA: Neurobehavioral Systems, Inc.
15
-------
McCall, R. 1979. The development of intellectual functioning in infancy and the prediction of later IQ.
In: Handbook of infant development (Osofsky, J., ed.). New York: John Wiley & Sons, 707-41.
McCall, R., and Camger, M. 1993. A meta-analysis of infant habituation and recognition memory
performance as predictors of later IQ. Child Development 64:57-79.
McCarthy, D. 1972. McCarthy scales of children's abilities. New York: The Psychological Corporation.
National Research Council. 1993. Measuring lead exposure in infants, children, and other sensitive
populations. Washington, DC: National Academy Press.
Otto D., Skalik I, House D., Hudnell H.K. 1996. Neurobehavioral Evaluation System (NES):
Comparative Performance of 2nd-, 4th-and 8th-grade Czech children. Neurotoxicol. Teratol.
18:421-428.
Psychological Corporation. 1992. Wechsler Individual Achievement Test. San Antonio: The
Psychological Corporation.
Reynolds, C., and Kamphaus, R. 1992. Behavior assessment system for children manual. Circle Pines,
MN: American Guidance Service, Inc.
Rice, D. 1993. Lead-induced changes in learning: Evidence from behavioral mechanisms from
experimental animal studies. Neurotoxicology 14:167-78.
Rose, S., Feldman, J., Wallace, I., and McCarton, C. 1991. Information processing at 1 year. Relation to
birth status and developmental outcome during the first 5 years. Developmental Psychology
27:723-37.
Rose, S., Feldman, I., and Wallace, I. 1992. Infant information processing in relation to six-year
cognitive outcomes. Child Development 63:1126-41.
Sheslow, D., and Adams, W. 1990. Wide range assessment of memory and learning. Wilmington, DE:
Wide Range, Inc.
Sparrow, S., Balla, D., and Cicchetti, D. 1984. Vineland Adaptive Behavior Scales Interview Edition
Survey Form Manual. Circle Pines, MN: American Guidance Service.
Stiles, K., and Bellinger, D. 1993. Neuropsychological correlates of low-level lead exposure in school-
age children: A prospective study. Neurotoxicology and Teratology 15:27-35.
Wechsler, D. 1989. Wechsler Preschool and Primary Scale of Intelligence-Revised. San Antonio: The
Psychological Corporation.
White, R., Gerr, F., Cohen, R., Green, R., Lezak, M., Lybarger, J., Mack, J., Silbergeld, E., Valciukas, J.,
Chappell, W., and Hutchinson, L. 1994. Criteria for progressive modification of neurobehavioral
function. Neurotoxicology and Teratology 16:511-24.
Wirt, R., Lachar, D., Klinedinst, J., et al. 1991. Multidimensional description of child personality: A
manual for the personality inventory for children, 1990 edition. Los Angeles: Western
Psychological Services.
16
-------
Do Pesticides and Other Environmental Exposures
Play an Important Role in the Development of Diseases of
Immune Dysregulation along the United States Mexican Border?
Anthony A. Horner, M.D.
Assistant Clinical Professor of Pediatrics
University of California San Diego Medical School
Introduction
The immune system has evolved over millions of years to protect self from non-self.
Immune competence relies on the rapid activation and replication of clones of antigen specific T
and B lymphocytes, and other mononuclear and polymononuclear white blood cells which
recognize and respond to infectious agents, foreign proteins, and other small molecular weight
molecules in an effort to protect the host. Many immune responses are protective, some are
clinically irrelevant, and some are deleterious to the individual. Environmental exposures can
cause perturbation of the immune system in a number of ways. Agents can be immunotoxic, act
as adjuvants, induce hypersensitivities to themselves, or induce autoimmunity, all of which have
unique clinical consequences. Although not covered here, immunocytes, due to their rapid
turnover, are also particularly susceptible to malignant transformation. In order to assess
whether the U.S. Mexican border is safe from an immunological point of view, it is important to
consider these various mechanisms by which immune status can be altered.
To assess the impact of environmental exposures on immunological health in individuals
living in target areas, a host of complementary investigations should be utilized. Preliminary
studies would include identification of which potentially immuno-modulating materials are
present in the environment and a survey of the prevalence and severity of diseases with a
potential immunologic etiology. It will be important to compare health information with
laboratory data such as: (1) quantitative laboratory studies of antibody including IgE, and cellular
constituents of the immune system, (2) qualitative assessment of vaccine specific antibody liters
and T-cell proliferation responses, (3) hypersensitivity testing for relevant agents, either
serologically or by skin tests, (4) markers of auto-immunity, and (5) in the case of immunologic
diseases of the respiratory tract, pulmonary function testing and chest x-rays. A philosophical
approach to the assessment of immunological health and its clinical consequences in individuals
living on the U.S. Mexican border will be outlined.
Immunological Principles
Immunotoxicity
Cytotoxic compounds generally affect rapidly dividing and metabolically active cells
selectively. Malignant cells, hair follicles, bowel mucosa, and bone marrow derived red and
white blood cells are particularly sensitive to cytotoxic agents because they fall into this
category. Immunotoxin is a term which is occasionally used to describe agents which stimulate
the immune system. I will reserve the term immunotoxin to describe agents which cause a
deterioration in immune parameters. Agents which promote immune responses are called
adjuvants in most of the immunology literature. I will use the term immuno-adjuvant to describe
17
-------
those agents which promote immune responses. Therefore, by my narrow definition, an
immunotoxin would be expected to lead to a fall in the numbers and health of immunocytes over
time leading to poor immunity and susceptibility to infectious disease. Malignant and auto-
immune disease can also be seen as a consequence of immunodeficiency.
The immunotoxic potential of pesticides and other manmade products is generally
assessed in animal models before their use is approved (Ladies et al., 1994; Vohr, 1995; Madsen
et al., 1996). One major limitation of animal models, however, is that they may not accurately
reflect the consequences of exposure in humans. For example, the pesticide pentachlorophenol
(PCP) has been shown to have a number of immune-inhibitory effects in animals. However, a
recently published study demonstrated that factory workers, some exposed to high levels of PCP
(up to 1442mg/l in plasma) for greater then 10 years had essentially no differences in
immunologic parameters nor health when compared to age matched controls (Colosio et al.,
1993). In the case of the insecticides deltametrin and a-cypermetrin, published animal studies
have shown immunostimulatory, immuno-inhibitory and neither effect (Masden et al., 1996).
The route of exposure has further immunologic consequences. For example in the case of
carbaryl, inhalation leads to a fall in antibody titers while dermal and oral exposure has no effect
(Ladies et al., 1994). These cited examples reflect the limitations of experimental animal models
in the identification of immunotoxins and other immuno-modulators of clinical relevance.
Agents with high immuno-modulatory potential should be easy to identify, with consistent
results across species, and identification as a health risk in the laboratory before their use is
approved. Therefore, those agents that make it to market and into common use are likely to have
relatively weak immuno-modulatory potential. Epidemiological studies play an important role in
monitoring the ongoing health consequences of the environment people live in. Given the
limitations of animal studies, population based research needs to be conducted to identify
immunotoxins and other immune modifying agents in the environment which might not be
identified in animal studies.
I m m u n o-ad j u van ts
Some compounds promote immune reactivity to other compounds without necessarily
inducing an antigen specific immune response themselves. This phenomenon has been termed
an adjuvant effect in immunology. In the laboratory, adjuvants are used to induce allergy, auto-
immunity, and other immunologically reactive states. Allergic immune responses (Th2) are seen
in experimental animal models with the co-injection of antigen with alum, pertussis toxin, or
cholera toxin (Snider et al., 1994; Oettgen et al., 1994). Immune responses more characteristic of
viral or mycobacterial infections (Thl responses) are seen with the co-injection of Freund's
adjuvant (Dvorak and Dvorak, 1974), and more recently with immuno-stimulatory sequence
DNA(ISS) (Raz et al., 1996). Although the adjuvant effect has been well described in animal
models, less is known about the possible adjuvant effect of environmental exposures on the
immune status and health of people. However, recently, Diaz-Sanchez et al. (1997) have
demonstrated that diesel exhaust promotes IgE synthesis in in vitro systems, and increases
ragweed specific IgE in nasal secretions of ragweed allergic individuals exposed to diesel
exhaust and allergen. This type of phenomenon may well occur with exposure to other manmade
and organic compounds whose adjuvant properties have yet to be identified. Given that allergic
disease is more common in developed then in primitive populations, there is concern that
exposure to diesel exhaust and other forms of manmade pollution may be responsible for this
excess disease burden.
18
-------
Antigen/allergen Specific Hypersensitivities
Exposure to pollens, molds, chemicals, and other agents in the air can lead to hyper-
sensitivity, and clinical disease. Since the majority of these exposures is via the respiratory tract,
skin, and gastro-intestinal tract, clinical disease is likely to manifest itself in one of these organ
systems. These immunologically mediated hypersensitivities can take many clinical forms
including: (1) classic IgE mediated allergic diseases such as asthma and allergic rhinitis, (2)
hypersensitivity pneumonitis and contact dermatitis which are thought to be due to cell mediated
hypersensitivities, and (3) diseases due to multiple immune mechanisms such as allergic
broncho-pulmonary aspergillosis (ABPA). Hypersensitivity mediated illness has been well
described in humans exposed to high levels of pesticides and other industrial products and
byproducts in the workplace. Examples include isocyanate induced asthma and specific IgE
production, and IgG mediated hypersensitivity pneumonitis (Baur et al., 1994), and chromate
induced asthma and contact dermatitis (Bright et al., 1997; Fisher, 1983). Rural environments
are also rich in unique organic allergens of animal and vegetable origin, which can provoke
asthma and other classic allergic diseases (Berstein and Berstein, 1993). In addition, more
complex immunologic diseases such as ABPA and hypersensitivity pneumonitis have been well
described in agricultural workers exposed to high concentrations of aspergillosis and other molds
in green houses and other settings (Yoshida et al., 1993). Clearly, a comprehensive assessment
of the prevalence and severity of common allergic diseases, and the spectrum of allergen
sensitivities will be an important part of the entire immunologic profile for the people living in
the U.S. Mexican border region.
Auto-immunity
The role of the environment in the development of autoimmune disease has not been well
studied. However, based on our current immunologic understanding, there is reason to believe
that auto-immune disease can be induced by the environment. We know that some medications
can induce a lupus-like disease and hemolytic anemia, with serological evidence of auto-
immunity (Hess, 1995). In this vein, it has been published that: (1) breast augmentation using
silicone implants increases the risk of scleroderma (Hess, 1995), (2) some cases of systemic
lupus erythematosus might be caused by industrial pollution (Koeger et al., 1997), and (3) that
Raynaud's phenomenon, sclerodermatous skin changes and acroosteolysis can be induced by
exposure to vinyl chloride. Certainly, a potential exists that some of the micro-environments
within the U.S. Mexican Border region promote the development of auto-immune phenomenon.
Assessment of Immune Status and Clinical Consequences at the U.S. Mexican Border
Clearly, pesticides and many other environmental exposures can affect the immune and
many other organ systems in animals and humans. Immunologic pathology can obviously lead
to clinically evident disease or may be so minor as to not be of concern. In order to carry out an
investigation of the clinical impact of environmental exposures on the development and
perpetuation of disease along the U.S. Mexican border two important issues need to be addressed
initially, as they will aid in the development of an enlightened and therefore more productive and
focused secondary investigation. The first issue to be addressed is, to what extent and to what
agents are the inhabitants of the U.S. Mexican border region exposed? The second is, what are
the active health problems of the people living in the various communities of the border region?
Is there an excess disease burden? And what potential influence could relevant exposures play in
the development of the diseases identified?
19
-------
Given our present state of knowledge, we can make some educated guesses about some of
the immunologically mediated health problems we might identify in the area under study. The
environment clearly plays an important role in the pathogenesis of respiratory disease. In the
case of asthma, there is a large body of information which suggests that poor communities with
poor air quality are at particular risk for experiencing an excess disease burden. Therefore,
asthma is likely to have a high prevalence along the U.S. Mexican border, and other
immunologically mediated pulmonary diseases may also be common. Survey data on the
incidence and severity of respiratory illness should be sought in the target communities, and this
should be collected in conjunction with air sampling studies to identify potential agents of
disease. Spirometric and CXR evaluation of the population should be pursued along with allergy
testing to provide more objective measures of disease and hypersensitivity. Finally, pulmonary
hypersensitivity testing, utilizing environmental agents identified in the ambient air should be
used to establish a cause and effect relationship between exposure and respiratory disease.
A link between asthma and other pulmonary and cutaneous hypersensitivity mediated
diseases and the environment has been clearly established. However, the role of the environment
in the development of other immunologically mediated conditions including those of the gastro-
intestinal tract is much less well understood. Survey information of individuals, their health care
providers, and hospitals will be important, to establish whether the immune status and health of
the population living along the U.S. Mexican border is in fact compromised compared to control
populations. The general imrmmological health of the population should be assessed by focusing
on the frequency and severity of community acquired viral and bacterial disease, as well as more
atypical and invasive infections. In addition, information regarding auto-immune manifestations
should be sought. Screening immunologic evaluations should include complete blood counts.
T-cell and B-cell immunity should be assessed both quantitatively and qualitatively to get an
accurate assessment of function. Quantitative T-cell and B-cell studies would include T/B cell
subsets and total immunoglobulin and IgG subclass levels. Qualitative assessment of T-cell and
B-cell function should include delayed type hypersensitivity skin testing to agents such as
Candida antigen which does not require previous vaccination or disease for positivity, and antigen
specific antibody liters which might require vaccination for reliability. Hypersensitivity testing
using skin or serum should be conducted to identify potential agents of disease. Provocation
testing may also be appropriate to see if exposure induces symptoms of disease. In addition,
physical examinations and hematologic markers of auto-immunity may be needed to objectively
assess for auto-immune disease should it be identified as a potential health problem in the
population under study.
Conclusion
Not much is known about the health risks for people living along the U.S. Mexican
border. However, this is a region of rapid change. With the opportunities and consequences of
NAFTA yet to be realized, the EPA has a unique opportunity to identify potential environmental
health hazards and to incorporate their findings into the policies of NAFTA that will govern the
border region in the years to come. However, study of the border and the health of the people
living along it is complicated by the simple fact that the U.S.. Mexican border is not
homogeneous. Agricultural, industrial, urban, and rural areas with undoubtedly unique
environmental conditions exist along the border. However, one consistent theme along the
border is duplicity. There exist two governments establishing two sets of environmental
20
-------
standards for two sets of industries, and two populations. In addition, there is great heterogeneity
in the ethnic, socio-economic, and cultural backgrounds of the people in this area. These issues
will be important considerations in developing a plan to study the health consequences of living
along the U.S. Mexican border.
References
Baur, X., Marek, W., Ammon, I, et al. 1994. Respiratory and other hazards of isocyanates. Int. Arch.
Occup. Environ. Health 66:141-52.
Berstein, D.I., and Berstein, I.L. 1993. Occupational asthma in allergy: Principles and practice. St. Louis:
Mosby, 1369-93.
Bright, P., Burge, P.S., O'Hickey, S.P., et al. 1997. Occupational asthma due to chrome and nickel
electroplating. Thorax 52:28-32.
Colosio, M.M., Maroni, M., Barcellini, W., et al. 1993. Toxicological and immune findings in workers
thexposed to pentachlorophenol. Arch. Environ. Health. 48:81-8.
Diaz-Sanchez, D., Tsien, A., Feming, J., and Saxon, A. 1997. Combination diesel exhaust particulate and
ragweed allergen challenge markedly enhances human in vitro nasal ragweed-specific IgE and
skews cytokine production to a T helper cell 2-type pattern. J. Immunol. 158:2406-13.
Dodson,V., and Dinman, B.D. 1971. Occupational acroosteolysis III. A clinical study. Arch. Env. Health
22:83-91.
Dvorak, A.M., and Dvorak, H.F. 1974. Structure of Fruend's complete and incomplete adjuvant:
Relation of adjuvanticity to structure. Immunology 27:99.
Fisher, A.A. 1983. The chromates: Prime causes of industrial allergies in contact dermititis. Cutis 32:24.
Hess, E.V. 1995. Role of drugs and environmental agents in lupus syndromes. Curr. Opin. Rheumatol.
34:597-59.
Kardestuncer, T., and Frumkin, H. 1997. Systemic lupus erythematosis in relation to environmental
pollution: An investigation in of an African-American community in Northern Georga. Arch, of
Env. Health 52:85-90.
Koeger, A.C., Nguyen, J.M., and Fleurette, F. 1997. Epidemiology of scleroderma among woman:
Assessment of the risk from exposure to silicone and silica. J. Rheum. 24:1853-5.
Kramer, M.N., Kurup, V.P., and Fink, J.N. 1989. Allergic bronchopulmonary aspirgillosis from a
contaminated dump site. Am. Rev. Respir. Dis. 140:1086-8.
Ladies, G.S., Smith, C., Heaps, K., and Loveless, S. 1994. Evaluation of the humeral immune response of
CD rats following a 2 week exposure to the pesticide carbarl by oral, dermal, or inhalation routes.
J. Tox. Environ. Health. 42:143-56.
Madsen, C., Claesson, M.H., and Ropke, C. 1996. Immunotocxicity of the pyrethroid insecticides
daltametrin and a-cypermetrin. Toxic. 107:219-27.
Markowitz, S.S., and McDonald, C.J. 1992. Occupational acroosteolysis. Arch. Dermatol. 106:219-23.
Oettgen, H.C., Martain, T.R., Wynshaw-Boris, et al. 1994. Active anaphylaxis in IgE deficient mice.
Nature 370:367-70.
Raz, E., Tighe, H., Sato, Y., et al. 1996. Preferential induction of Thl response by intradermal gene
vaccination. Proc. Nat. Acad. Sci. USA 93:4733-7.
Snider, D.P., Marshall, J.S., Perdue, M.H., et al. 1994. Production of IgE antibody and allergic
sensitization of intestinal and peripheral tissues after oral immunization with protein antigen plus
pertussis toxin. J. Immunol. 153:647-57.
Vohr, H.W. 1995. Experiences with an advanced screening proceedure for the identification of chemicals
with an immunotoxic potential in routine toxicology. Toxic. 104:149-58.
Yoshida, K., Ueda, A., Yamasaki, H., et al. 1993. Hypersensitivity pneumonitis resulting from
Aspergillus fumigatus in a greenhouse. Arch, of Env. Health 48:260-2.
21
-------
Evaluation of Developmental Neurocognitive and Neurobehavioral Changes
Associated with Pesticide Exposure: Recommendations for the
U.S. Environmental Protection Agency Workshop on the Assessment of
Health Effects of Pesticide Exposure in Infants and Young Children
Antolin M. Llorente
Department of Pediatrics
Baylor College of Medicine
and
Texas Children's Hospital
Houston, Texas
Abstract
This manuscript provides recommendations for the assessment of chronic low-level pesticide
exposure-related cognitive changes in infants and young children. In addition to a review of critical
issues associated with the detection of neurocognitive and neurobehavioral alterations, a list of tests
and procedures is presented from which a test battery could be devised to optimally assess the effects
of neurotoxins in this population. The domains assessed by these instruments include attention and
concentration, overall developmental or intellectual functioning, language, learning and memory
(verbal and visual), motor skills, neurobehavioral functioning (adaptation, behavior, etc.), and visual
processing. The use of repeated observations, in conjunction with a set of instruments encompassing
a broad assessment scope with enough sensitivity, should be capable of detecting subtle
neurocognitive changes subsequent to neurotoxic exposure.
Introduction
Although the neurodevelopmental effects of certain neurotoxicants on the Central Nervous
System (CNS) have been well documented, including the impact of lead (Pb) (Beattie et al., 1975;
David et al., 1982; Needleman, 1993; Needleman et al., 1990; Rutter, 1980; Yule et al., 1981),
polychlorinated biphenyls (PCB's) (Jacobson, Jacobson, and Humphrey, 1990; Rice, 1997; Rogan
and Gladden, 1991, 1992), and acute pesticide poisonings (Kaplan et al., 1993; Zwiener and
Ginsburg, 1988), the effects of chronic low-level pesticide exposure on neurocognitive and
neurobehavioral outcomes in infancy and childhood have received limited attention. This is
unfortunate as these agents have been shown to cause significant neuropsychological (NP)
impairments in adults (Ecobichon and Joy, 1994; Hartman, 1995), through acute poisonings (Dean
et al., 1984; Muldoon and Hodgson, 1992; Ratner, Oren, and Vigder, 1983) or chronic exposure to
these substances (Metcalf and Holmes, 1969; Rosenstock et al., 1990). The fact that the effects of
chronic low-level pesticide exposure on the developing brain have received increased attention
recently is also timely as the use and production of these compounds for commercial and domestic
purposes have spiraled in recent years (Ecobichon and Joy, 1994; Lang, 1993).
Therefore, the study of developmental cognitive and behavioral effects due to chronic low-
level exposure to pesticides warrant special consideration. The study of the CNS sequelae of these
compounds in individuals along the U.S.-Mexico border, a primary focus of this workshop, also
merits resources and time allocation as a substantial number of infants and young children are at risk
22
-------
in this frontier due to the extensive use of agricultural pesticides in this region. However, prior to
embarking on a review and tabulation of assessment procedures useful in measuring developmental
neurocognitive and neurobehavioral changes due to chronic low-level pesticide exposure through
multiple pathways, critical developmental, methodological, strategic, and theoretical issues
associated with the evaluation of these variables are addressed.
Review of Selected Child Development and Maturation Variables
A brief review of developmental issues merits consideration since they are intrinsically
related to the assessment of cognitive changes subsequent to neurotoxicant exposure. Brain weight
is a marker capable of elucidating the rapid developmental changes occurring in the CNS of infants
and young children. During the first year of life, an infant's brain more than doubles its weight
(Lemire et al., 1975). In addition to substantial increases in brain weight, there are considerable
maturational changes taking place in the developing brain. Schade and Ford (1965) showed that a
fourfold increase in the number of branching points from dendrites in layer III of the middle frontal
gyrus occurred during the first six months post-birth. These investigators also reported substantial
increments in the total length of these dendrites in the same brain region (Schade and Ford, 1965).
Accompanying these surges in dendrite development, extensive increments in synaptic growth,
dendrite pruning, and myelination occur postnatally partly responsible for the increase in the
complexity of the developing CNS (Huttenlocker, 1984, 1990; Kolb and Fantie, 1997).
Although other developmental markers are worthy of consideration (e.g., glial cell
development, head circumference), the indices presented above underscore the staggering and rapid
maturational changes that a developmental neurocognitive and neurobehavioral assessment must be
able to gauge when evaluating the effects of chronic low-level exposure to pesticides in the
developing CNS. The assessment of neurocognitive and neurobehavioral changes must also be
capable of disentangling the effects of pesticides and similar neurotoxins from these maturational
changes. A thorough understanding of these developmental issues is additionally important since
domains in critical periods of development, or those undergoing the greatest amount of change, tend
to be the ones at greatest risk (Scott, 1962) to the effects of neurotoxins.
Neurodevelopmental and Neurophysiological Factors Associated with Increased Neurotoxic
Susceptibility
It is critical to review factors that make the developing CNS of infants and young children
highly susceptible to neuro toxicants. According to Hartman (1995), the developing brain is
especially vulnerable to the effects of these substances. The added vulnerability is partially the result
of the substrate composition of the CNS (e.g., 50% of the total dry weight of the brain is lipide),
making it a preferred site of accumulation for neurotoxins as these substances tend to be lipophilic
(Hartman, 1995). The CNS of the young child may also be more susceptible to neurotoxins due to
the disruptive effects of these substances during critical and active periods of brain development,
quite capable of interrupting or hampering cell division or other developmental processes. Hartman
(1995) also notes that the young brain is especially sensitive to hypoxia, a frequent mechanism of
neurotoxic action. Finally, the effects of neurotoxic compounds, including pesticides, may be more
pronounced in the rapidly maturing brain secondary to the underdeveloped status of the blood-brain
barrier of infants and young children (Chusid, 1982; Claudio, 1992).
23
-------
Theoretical, Methodological, and Strategic Issues in the Assessment of Cognitive Change
Secondary to Neurotoxicant Exposure
As with any area of scientific inquiry, it is critical that the intensity, chronicity, frequency,
and routes of exposure be clearly defined. Although exposure to pesticides, particularly low-level
chronic exposure, may be difficult to measure and quantify, it is imperative that its measurement and
definition be as specific and well defined as possible, thus allowing for later comparison with other
investigations assessing the effects of these substances. In addition to well-defined exposure, the
definition of neurodevelopmental or neurobehavioral impairment should be clearly demarcated with
impairment cut-off scores clearly stated for the same reasons. Albeit studies with adults suggest that
pesticides may be capable of infringing upon NP performance (Ecobichon and Joy, 1994),
investigators should not assume that the effects of exposure to pesticides in infants and youths cause
the same type or magnitude of sequelae as those observed in adults. Researchers should also
recognize that their conceptualizations of exposure simply represent hypotheses requiring flexibility
and the ability to evolve as new information emerges modifying data-based findings of the
neurodevelopmental effects of low-level contamination through multiple pathways.
Although cross-sectional designs may be initially required as an exploratory method to
elucidate the effects of various types of exposures or other parameters, as these designs or methods
may aid understanding this phenomenon, an argument is presented for the use of longitudinal or
repeated measure designs to assess subtle alterations in neurocognition and neurobehavior associated
with chronic low-level pesticide exposure. Longitudinal designs are best qualified to assess the
additive and chronic impact of exposure to neurotoxins. Longitudinal designs are superior relative
to other designs in assessing rates of changes in development (Achenbach, 1978) coupled with
superimposed alterations in cognitive functioning associated with neurotoxins. In addition, they
allow experimental subjects to serve as their own controls, permitting an idiographic rather than a
nomothetic approach to the evaluation of cognitive change. The latter is an important issue to be
addressed as a large proportion of NP measures necessary to assess alterations in cognitive
functioning may not have available normative data for ethnic minority populations living along the
U.S.-Mexico border. Finally, longitudinal relative to cross-sectional designs are superior in
detecting delayed-onset impairments that may occur as a result of chronic low-level pesticide
exposure.
Validity and reliability are critical issues worthy of significant consideration. It should be
noted that the issue of validity not only relates to the populations under study, but also addresses the
applications under investigation (pesticides and similar neurotoxins) (see Franzen, 1989). In other
words, although neurodevelopmental procedures may have been shown capable of assessing delays
as a result of prematurity, developmental delays, or similar syndromes, the majority of instruments
used to assess changes in cognition in infants and young children have not been validated for the
specific application at hand (assessment of pesticide-related effects).
Aside from issues related to sample size capable of infringing upon an investigation
evaluating the effects of pesticides on neurodevelopment, other issues associated with subject
selection strategies should be strongly considered when evaluating NP functioning. Subject
selection biases may occur as a result of idiosyncratic characteristics specific to the desired study and
border populations. Study populations selected as a mere result of experimental convenience may
also reflect biases associated with these populations. For example, the selection of participants
referred from clinics near agricultural areas with significant pesticide exposure may exhibit greater
24
-------
symptomatic levels relative to levels exhibited by infants and young children selected from non-
referred agricultural regions with elevated pesticide exposure. Lack of multi-site participant
selection may also fail to capture differences in border populations or exposures because certain
areas within the U.S.-Mexico border may vary substantially in their predominant industrial versus
agricultural utilization and their subsequent environmental conditions.
Although exclusion criteria play a major role in any investigation as some subjects must be
excluded from a study because their participation could adversely influence an investigation (e.g.,
history of child abuse or head trauma), some variables which may meet criterium for exclusion may
interact with neurotoxins. In other words, studies investigating the effects of neurotoxicants
including pesticides should not discard specific variables simply as a matter of convenience since
many of these variables may interact with the toxins under investigation (e.g., maternal alcohol or
drug use). A large number of exclusionary variables may also render a study ecologically invalid
as it becomes unrepresentative of the population under investigation. A posture making use of
formal experimental designs rather than quasi-experimental designs with multiple control groups
(e.g., infants and children with high and low exposures, infants and children exposed to other
neurotoxicants, unexposed children) is critical to the generalization of the findings from such an
investigation. Although this may seem a moot point, a number of investigations with adults
assessing the toxic impact of solvents on neurocognition employed designs without the use of
controls leaving their findings to much speculation (see Juntunen et al., 1980).
A balance in examination scope versus length is another issue which should be given special
consideration, particularly in investigations employing infants as participants. With regard to
breadth, the evaluation of neurotoxicant-related cognitive changes should have a broad range of
domains in its assessment aims. It should further comprise breadth in the assessment of participants,
namely the infant and the caretaker as respondent for the infant's functioning. This assessment
approach has been shown to be significantly helpful in the assessment of infants and young children
for obvious reasons (c.f., Edelbrock et al., 1985). An evaluation that includes the child and the
caretaker may be productive in the assessment of toxicants should early impairments associated with
these substances become evident in the behavioral repertoire of youths noticeable to caretakers prior
to the emergence of subtle alterations in cognition detectable during NP assessments. The length
of a developmental neurocognitive assessment should also be regarded as critical when assessing
this population as individual variables such as frustration, motivation, and stamina may play key
assessment roles. Finally, the length and scope of a test battery have time allocation implications
associated with economic factors that must be taken into account.
Whenever longitudinal assessment or similar designs employing repeated measures to
investigate cognitive changes require sequential examination or repeated exposure of test materials,
the issue of practice effect associated with multiple test administration should be given due weight
(Sattler, 1988). This is especially important when assessing the effects of factors whose variance
may be small (e.g., pesticides) relative to the variance due to the developmental changes taking place
or the effects associated with the repeated presentation of the procedures used in an investigation.
hi this regard, it should also be noted that the sensitivity of certain psychological instruments to
detect neurotoxic-related cognitive changes depends on the novelty of the test or procedure as well
as the frequency of administration. Therefore, repeated NP task administrations to the same subject
diminishes its novelty and consequently its sensitivity to detect changes in NP functioning. One
approach to reduce practice effects is to use measures resistant to the gains associated with repeated
25
-------
administration. Another approach is to use alternative equivalent measures of the same NP
procedures when available. Alternatively, an assessment stance which employs a monitoring battery
nestled within a comprehensive battery (comprising the monitoring battery), whereby the procedures
comprising the comprehensive battery are repeated with less frequency (due to their diminished
reticent to test-retest effects) relative to the procedures in the monitoring battery, may be of aid in
dealing with the issue of frequency of test exposure.
Assessment Considerations with Ethnic Minorities and Populations Along the U.S.-Mexico
Border
Since the majority of individuals will come from the frontier between the U.S. and Mexico,
or from regions within the U.S. with predominant ethnic minority representation, it is critical to
address issues associated with the cognitive evaluation of these persons. A large portion of
immigrants from Mexico and its border with the U.S. tend to be from low educational,
socioeconomic, and under-served backgrounds (Fortes and Rumbaut, 1990; U.S. Bureau of the
Census, 1984). This is an important factor to consider in NP evaluation as it directly impacts
performance on most developmental neurocognitive procedures used to detect neurocognitive and
neurobehavioral alterations (Adams et al., 1982; Ardila, Roselli, and Rosas, 1989; Laosa, 1984;
Perez-Arce, 1984).
Language and its impact on neuropsychological performance (Gordon, 1980; Paradis, 1978)
and the limited availability of suitable instruments in Spanish from which valid inferences can be
drawn are two factors deserving careful consideration since a large proportion of individuals living
along the U.S.-Mexico frontier are monolingual (Spanish) while others are bilingual
(English/Spanish). The use of interpreters, as it relates to the assessment of cognitive functions,
should be avoided as there is precedent in the NP literature to indicate that such an assessment
posture is capable of biasing results (c.f., LaCalle, 1987).
The availability of normative data for these populations is also scarce (Ardila, Roselli, and
Puente, 1994) precluding the appropriate use of a nomothetic approach to the interpretation of test
results. This limitation is a strong argument supporting the use of experimental designs employing
longitudinal or repeated measures allowing subjects to serve as their own controls as noted earlier.
Issues associated with the standardized administration of developmental neuropsychological
procedures, examiner competence, and migrational variables associated with these populations (c.f.,
Llorente, 1997; Llorente et al., in-press; U.S. Immigration and Naturalization Service, 1991; Warner,
1992), play a major role in these types of investigations. These influences, left unaccounted, are
likely to bias the assessment of neurodevelopmental variables.
Other Potential Confounding Factors Associated with Cognitive Assessment
Other potential confounds capable of infringing upon the assessment of cognitive changes
as a result of exposure to neurotoxins or pesticides in youths deserve special merit. Maternal drug
use during gestation and nursing periods demand proper attention as they have direct bearing on
child development outcome (Coles et al., 1992; Jacobson et al., 1994). Issues associated with the
treatment of childhood illnesses through the use of multiple health care delivery systems readily
available near both borders should also be addressed (Warner, 1992). Nutritional variables and
family history of medical and psychiatric illnesses (c.f., Brockman and Ricciuti, 1971; Cravioto and
DeLicardie, 1966) are potential confounds worthy of exploration as they may also account for a
26
-------
portion of the variance in neurodevelopmental performance which otherwise may be inadvertently
attributed to neurotoxins.
Tests and Procedures to Evaluate Developmental Neurocognitive and Neurobehavioral
Alterations Subsequent to Neurotoxicity
Certain health and research organizations and individual investigators have made
recommendations with regard to the assessment of neurotoxicity in humans (see Hartman [1995] for
a comprehensive review). The World Health Organization (WHO) Neurobehavioral Core Test
Battery was developed for adolescents and adults for this purpose. This battery assesses several
domains including motor skills (steadiness and fine motor coordination), attention and concentration
(response speed), perceptual speed, visual processing, learning and memory (visual and auditory),
and affect (Baker and Letz, 1986). With regard to children, Winneke and Collet (1985), as part of
an effort spearheaded by the WHO, also recommended assessing a broad range of domains when
evaluating cognitive changes associated with exposure to neurotoxins including overall intellect,
attention and concentration (e.g., reaction time), visual processing and visuo-motor abilities, and
motor skills. Similarly, a relatively recent attempt at developing a comprehensive battery of tests
and procedures for youths was conducted by the U.S. Government, Agency for Toxic Substances
and Disease Registry (ATSDR). The outcome of that effort was the Pediatric Environmental
Neurobehavioral Test Battery (PENTB) (ATSDR, 1997). This battery of tests and procedures
assesses a broad range of abilities and skills through the use of instruments administered to the child
and the caretaker. Unfortunately, the assessment of infants as part of this effort was left to the sole
use of rating scales to be completed by the caretaker failing to take advantage of individually
administered procedures available for infants and young children useful in detecting changes in
neurocognition. Nevertheless, these efforts represent great strides in the assessment of
neurobehavioral effects of neurotoxins in infancy and childhood. Table 1 below shows a list of
recommended developmental neurocognitive and neurobehavioral tests and procedures capable of
detecting cognitive changes subsequent to chronic low-level exposure to pesticides and other
neurotoxicants through multiple pathways from infancy to late childhood. The domains assessed
by these instruments include overall developmental or intellectual functioning, attention and
concentration, language, learning and memory (verbal and visual), motor skills, neurobehavioral
functioning including adaptation, and visual processing. The overall aim of such a list of
instruments is the development of a battery of tests with enough sensitivity, which in conjunction
with the use of repeated observations, will be capable of detecting subtle neurocognitive changes
subsequent to pesticide exposure.
27
-------
References
Achenbach, T.M. 1978. Research in developmental psychology': Concepts, strategies, methods. New York:
Free Press.
Adams, R.L., Boake, C., and Grain, C. 1982. Bias in neuropsychological test classification related to
education, age and ethnicity. Journal of Consulting and Clinical Psychology 50:143-5.
Agency for Toxic Substances and Disease Registry. 1997. Pediatric Environmental Neurobehavioral Test
Battery. Washington, DC: U.S. Department of Health and Human Services.
Ardila, A., Rosselli, M., and Puente, A.E. 1994. Neuropsychological assessment of the Spanish speaker. New
York: Plenum Press.
Ardila, A., Roselli, M., and Rosas, P. 1989. Neuropsychological assessment of illiterates: Visuo-spatial and
memory abilities. Brain and Cognition 11:147-66.
Baker, E.L., and Letz, R. 1986. Neurobehavioral testing in monitoring hazardous workplace exposures.
Journal of Occupational Medicine 28:987-90.
Brockman, L., and Ricciuti, H. 1971. Severe protein calorie malnutrition and cognitive development in
infancy and early childhood. Developmental Psychology 4:312.
Beattie, A.D., Moore, M.R., Goldberg, A., Finlayson, M.J.W., Mackie, E.M., Graham, J.F., McLaren, D.A.,
Murdock, R.M., and Stewart, G.T. 1975. Role of chronic low-level lead exposure in the etiology of
mental retardation. Lancet 1:589.
Chusid, J.G. 1982. Correlative neuroanatomy andfunctional neurology, ISthed. California: Lange Medical
Publications.
Claudio, L. 1992. An analysis of the U.S. Environmental Protection Agency neurotoxicity testing guidelines.
Regulatory Toxicology and Pharmacology 16:202-12.
Coles, C.D., Platzman, K.A., Smith, I., James, M.E., andFalek, A. 1992. Effects of cocaine and alcohol use
on neonatal growth and neurobehavioral status. Neurotoxicology and Teratology 14:23-33.
Cravioto, J., and DeLicardie, E. 1966. Nutrition, growth, and neurointegrative development: An
experimental and ecologic study. Pediatrics 38:319.
David, O.J., Grad, G., McGann, B., and Kolton, A. 1982. Mental retardation and "nontoxic" lead levels.
American Journal of Psychiatry 139:806-9.
Dean, A., Pugh, J., Embrey, K., Cain, J., Lane, L., Brackin, B., and Thompson, F.E. 1984. Organophosphate
insecticide poisoning among siblings-Mississippi. Morbidity and Mortality Weekly Report 33:592-4.
Ecobichon, D., and Joy, R.M. 1994. Pesticides and neurological diseases, 2nd ed. Florida: CRC Press.
Edelbrock, C., Costello, A.J., Dulcan, M.K., Kales, R., and Conover, N.C. 1985. Age differences in the
reliability of the psychiatric interview of the child. Child Development 56:265-75.
Franzen, M.D. 1989. Reliability and validity in neuropsychological assessment. New York: Plenum Press.
Gordon, H.W. 1980. Cerebral organization in bilinguals. Brain and Language 9:255-68.
Hartman, D.E. 1995. Neuropsychological toxicology: Identification and assessment of human neurotoxic
syndromes, 2nd ed. New York: Plenum Press.
Huttenlocker, P.R. 1984. Synapse elimination and plasticity in developing human cerebral cortex. American
Journal of Mental Deficiency 88,488-96.
Huttenlocker, P.R. 1990. Morphometric study of human cerebral cortex development. Neuropsychologia
28:517-27.
Jacobson, J.L., Jacobson, S.W., and Humphrey, H.E.B. 1990. Effects of exposure to PCB's and related
compounds on growth and activity in children. Neurotoxicology and Teratology 12:319-26.
Jacobson, J.L., Jacobson, S.W., Sokol, R.J., Martier, R.J., Martier, S.S., Ager, J.W., and Kaplan-Estrin, M.G.
1994. Teratogenic effects of alcohol on infant development. Acoholism: Clinical and Experimental
Research 17:174-83.
Juntunen, J., Hernberg, S., Eistola, P., and Hupli, V. 1980. Exposure to industrial solvents and brain atrophy:
A retrospective study of pneumoencephalographic findings among 37 patients. European
Neuroradiology 19:3 66-75.
28
-------
Kaplan, J.G., Kessler, J., Rosenberg, N., Pack, D., and Schaumburg, H.H. 1993. Sensory neuropathy
associated with Dursban (chlorpyrifos) exposure. Neurology 43:2193-6.
Kolb, B., and Fantie, B. 1997. Development of the child's brain and behavior. In: Handbook of clinical child
neuropsychology (Reynolds, C.R., and Fletcher-Janzen E. (eds.). New York: Plenum Press, 17-41.
LaCalle, J.J. 1987. Forensic psychological evaluations through an interpreter: Legal and ethical issues.
American Journal of Forensic Psychology 5:29-43.
Lang, L. 1993. Are pesticides a problems? Environmental Health Perspectives 101:578-83.
Laosa, L.M. 1984. Ethnic, socioeconomic and home language influences upon early performance on
measures of abilities. Journal of Educational Psychology 76(6): 1178-98.
Lemire, R.J., Loeser, J.D., Leech, R.W., and Alvord, E.G. 1975. Normal and abnormal development of the
human nervous system. New York: Harper and Row.
Lezak, M.D. 1995. Neuropsychological assessment, 3rd ed. New York: Oxford University Press.
Llorente, A.M., Ponton, M.O., Taussig, I.M., and Satz, P. In-press. Patterns of American immigration and
their influence on the acquisition of neuropsychological norms for Hispanics. Archives of Clinical
Neuropsychology.
Llorente, A.M. 1997. Neuropsychologic assessment of Hispanic populations: The influence of immigration
on assessment. Paper presented at the 105th Annual Convention of the American Psychological
Association, Chicago, IL.
Metcalf, D.R., and Holmes, J.H. 1969. EEG, psychological and neurological alterations in humans with
organophosphate exposure. Annals of the New York Academy of Science 160:357.
Muldoon, S.R., and Hodgson, M.J. 1992. Risk factors for nonoccupational organophosphate pesticide
poisoning. Journal of Occupational Medicine 34:38-41.
Needleman, H.L. 1993. The current status of childhood low-level lead toxicity. Neuro Toxicology 14:161-6.
Needleman, H.L., Schell, A., Bellinger, D., Leviton, A., and Allred, E. 1990. The long-term effects of
exposure to low doses of lead in childhood. The New England Journal of Medicine 322:83-8.
Paradis, M. (ed.) 1978. Aspects of bilingualism. Columbia, S.C.: Hornbeam Press.
Perez-Arce, E. 1984. The role of culture and SES on cognition. The Clinical Neuropsychologist 8:350
(Abstract).
Fortes, A., and Rumbaut, R.G. 1990. Immigrant America: A portrait. Los Angeles: University of California
Press.
Ratner, D., Oren, B., and Vigder, K. 1983. Chronic dietary anticholinesterase poisoning. Israeli Journal of
Medical Science 19:810-4.
Rice, D.C. 1997. Neurotoxicity produced by developmental exposure to PCBs. Mental Retardation and
Developmental Disabilities 3:223-9.
Rogan, W.J., and Gladden, B.C. 1992. Neurotoxicology of PCB's and related compounds. Neuro Toxicology
13:27-36.
Rogan, W.J., and Gladden, B.C. 1991. PCB's DDE and child development at 18 and 24 months. Annals of
Epidemiology 1:407-13.
Rosenstock, L., Daniell, W., Bamhart, S., Schwartz, D., andDemers, P.A. 1990. Chronic neuropsychological
sequelae of occupational exposure to organophosphate insecticides. American Journal of Industrial
Medicine 18:321-5.
Rutter, M. 1980. Raised lead levels and impaired cognitive-behavioral functioning: A review of the
evidence. Developmental Medicine and Child Neurology 22(Suppl. 1):126.
Sattler, J.M. 1988. Assessment of children, 3rd ed. San Diego, CA: Jerome M. Sattler, Publisher.
Scott, J.P. 1962. Critical periods in behavioral development. Science 138:949.
Schade, J.P., and Ford, D.H. 1965. Basic neurology. Amsterdam: Elsevier.
Spreen, O., and Strauss, E. 1991. A compendium of neuropsychologic tests: Administration, norms, and
commentary. New York: Oxford University Press.
U.S. Bureau of the Census. 1984. 1980 census: Socioeconomic characteristics of U.S. foreign-born
population detailed in Census Bureau tabulations. Washington, DC: U.S. Department of Commerce.
U.S. Immigration and Naturalization Service. 1991. 1990 statistical yearbook of the Immigration and
Naturalization Service. Washington, DC: U.S. Government Printing Office.
29
-------
Warner. D.C. 1992. Health care on the U.S.-Mexico border. In: Health policy and the Hispanic (Furino, A.,
ed.). Colorado: Westview Press.
Winneke, G., and Collet, W. 1985. Components of test batteries for the detection of neuropsychological
dysfunction in children. Neurobehavioral methods in occupational and environmental health
(Document 3). Copenhagen: World Health Organization, 44-8.
Yule, W., Lansdown, R., Millar, I.E., and Urbanowicz, M.A. 1981. The relationship between blood lead
concentrations, intelligence and attainment in a school population: A pilot study. Developmental
Medicine and Child Neurology, 23:567-76.
Zwiener, R.J., and Ginsburg, C.M. 1988. Organophosphate and carbamate poisoning in infants and children.
Pediatrics 81:121-6.
30
-------
TABLE 1
I.
II.
Recommended Developmental Neuropsychological and Neurobehavioral
Tests and Procedures by Age and Domain1
Developmental Tests and Procedures (Ages 4 months-42 months)2
Instrument Recommended Age Range3
Bayley Scales of Infant Development
Child Development Inventory
Minnesota Infant Development Inventory
Vineland Adaptive Behavior Scales
> 4 months-42 months
> 15 months
4-15 months
> 4 months
Procedures and Tests for Pre-school and School-age Children (Ages 48 months and older)4
Domain/Instrument Recommended Age Range3
Intellectual
Woodcock-Johnson Tests of Cognitive Abilities
(WJ-R; Subtests 1-14)
Attenti on/Concentration
Children's Color Trails 1 and 2
Numbers Reversed (WJ-R, Subtest 9)
Reaction Time (e.g., Gordon Diagnostic)
Trailmaking Test A and B (Children's Version)
Language
Verbal Fluency
Clinical Evaluation of Language Fundamentals
(CELF-Third Edition, expressive and receptive
tests only)
Learning and Memory
WJ-R Memory Scales (Subtests 1, 2, 8, 9, 15, 16)
Stanford-Binet-4th ed. (Bead Memory)
Rey-Osterrieth Complex Figure
Motor
Bruininsks-Oseretsky Motor Proficiency Tests
Grip Strength Test (Child Version)
> 4 years
> 6 years
> 3 years
> 6 years
> 9 years
> 6 years
> 6 years
> 3 years
> 3 years
> 6 years
> 5.5 years
> 6 years
31
-------
Table 1. Recommended Developmental Neuropsychological and Neurobehavioral Tests
and Procedures by Age and Domain' (continued)
Grooved Pegboard Test (Child Administration) > 5 years
Peabody Motor Scales > 1 year-6 years
Visual Processing
Developmental Test of Visual-Motor Integration > 3 years
Stanford-Binet-4th ed. (Pattern Analysis) > 3 years
Rey-Osterrieth Complex Figure (Copy) > 6 years
Adaptive/Behavioral/Emotional
(Administered or Completed by the Caretaker)
Child Behavior Checklist > 4 years
Conners' Rating Scales > 3 years
Vineland Adaptive Behavior Scales > 4 months
Notes:
The test selection shown above attempted to take into consideration as much as possible the use of tests and procedures
that have been adapted (not translated) in Spanish so they can be used with populations along the U.S.-Mexico
border.
2The battery of tests for infants should not surpass an hour's time in its administration.
3 Age-range shown does not necessarily represent the lower age-bound of the test as published by the test developer.
4 A battery from this list of tests and procedures should be selected to be administered to children at specific age levels.
The battery selected should not surpass more than four hours in administration time (including test breaks) for
children 48 months of age or older.
5The reader is referred to Lezak (1995), Sattler (1988), and Spreen and Strauss (1991) for references for the tests and
procedures presented above.
32
-------
Pesticides and Children: Pulmonary Outcome Measures
Maria D. Martinez, M.D.
Arizona Respiratory Science Center
Department of Pediatrics
University of Arizona
Air pollution and pesticides are two important health concerns for communities along the
U.S.-Mexico border. Abundant literature exists regarding the association of air pollution and
pesticides with pulmonary complications, but not much is known about the potential effects of
chronic pesticide exposure on lung function. To begin with air pollution, Ware et al. (1986) found
that higher particulate levels were associated with increases in cough frequency, bronchitis, and a
composite measure of lower respiratory illness. In addition, a review of several studies of ambient
air pollution and respiratory disease concluded that pollution, while not a causal factor in its
development, aggravates preexisting respiratory disease (Abramson and Voigt, 1991). Air pollution
has been associated with bronchoconstriction both in healthy individuals as well as in asthmatics.
Turning to pesticides, there is extensive literature available regarding acute pesticide poisoning, but,
little has been published describing lung function abnormalities and chronic pesticide exposure.
Both bronchorrhea and bronchoconstriction have been observed in acute pesticide poisonings, while,
interstitial lung disease has been described in association with paraquat exposure (Reigart, 1995).
Thus, the literature reveals little about chronic pesticide exposure and its potential effects on
lung function. In planning a study to identify pulmonary sequelae of such exposure, there are several
outcome measures which might be used. These include clinical data, questionnaires, and physiologic
measures.
The first outcome measure to consider is clinical data. Advantages of utilizing clinical data
include such features as economy and availability, as well as a contemporaneous record. Being a
contemporaneous record, recall bias is nearly eliminated since the data were obtained at the time of
the visit. Health care utilization patterns, on the other hand, may bias the results. Subjects of lower
socioeconomic status, though they may actually have worse lung disease, may not frequent a
physician on a regular basis. This population may wait till their acute symptoms are severe before
going to an emergency department or urgent care facility, or may not go at all. In this setting,
continuity of care and, therefore, records are less likely.
Questionnaires are another relevant aspect of studying lung function abnormalities in
different populations. Before getting into questionnaires I will briefly explain the Tucson Children's
Respiratory Study (CRS) since I will be using it as an example in the remainder of this presentation.
This study is an ongoing, longitudinal study of respiratory health in 1,246 children which utilizes
a combination of physician reporting, questionnaires, home visits, and physiologic studies to
examine potential risk factors and their association with the development of respiratory disease
(Taussig et al., 1989). Questionnaires provide more detailed and in-depth information about the
targeted population than clinical charts do since questionnaires can be geared to the specific study
group. In addition, they can be self-administered as well as coded and stored in a computerized data
base. When administering a questionnaire, both linguistic and cultural compatibility with the
33
-------
targeted population must be considered. If not self-administered, trained personnel are required. In
contrast to what we observed with the clinical data, recall bias can be introduced with questionnaires.
Other challenges to be considered when preparing a questionnaire include labeling of
symptoms, translation, and cultural differences. Regarding labeling of symptoms, one must know
what terms subjects in the study area may use to define the outcome variables of interest. After
knowing the local idioms and local definitions of respiratory outcome variables, as well as of their
symptoms, one can translate the questionnaire. In doing so, one can aim to get a better
understanding of what is happening in the study populations.
Some examples of questionnaires used to assess respiratory outcomes include: the CRS
questionnaire, the American Thoracic Society/NIH Division of Lung Diseases (ATS/DLD)
questionnaire, The International Study of Asthma and Allergies in Childhood (ISAAC), and the
American Thoracic Society questionnaire. The CRS questionnaire has been useful in categorizing
children with respect to distinct asthma phenotypes which has resulted in the ability to correlate
physiologic measures with clinical disease expression. In older children one can use the ISAAC
which consists of a video tape with five different vignettes. The video tape is shown in conjunction
with the administration of a questionnaire to which the parents respond. No words are spoken in the
tape: rather images and audio of different symptom complexes are provided. Presently there is an
international effort by ATS and NIH to create a new standardized questionnaire that contains
different modules (i.e., a module for respiratory core questions for children and another for adults).
In addition to clinical data and questionnaires, physiologic outcome measures can be
entertained. Pulmonary physiologic measures can correlate exposure with disease, are relatively
simple to perform, are noninvasive, and are reproducible. Challenges include the requirement for
special equipment and trained personnel resulting in increased cost and subject cooperation.
The pulmonary physiologic parameters that can be measured depend on the child's age. In
young children, four to five years old, one can obtain partial expiratory flow volume curves (PEFV).
An example of a PEFV curve can be seen in Figure 1. These curves are reproducible, correlate with
disease, and allow tracking over time. Of the original 1246 children in the CRS, PEFV curves were
obtained in approximately 600 (males: n=279, females: n=298). Seventy-five percent of these 600
also completed a cold dry air challenge. The mean age was 6.04 years with 40% being younger than
six years of age. These curves allow for measurement of maximal flow at functional residual
capacity, a good measure of airway function. In assessing the effectiveness of this measure, the
Tucson group found that intersubject variability was 29.5%. While this may appear to be high,
variability was actually lower than that found in the adult population using comparable measures.
The intrasubject variability was only 7.9%.
One can standardize these measures by indexing V'max FRC by lung volume (V'max
FRC/FRC). When V'max FRC/FRC is plotted against the four different groups of wheezing
obtained from the questionnaire, one can see how this measure can assess lung function at a given
point in time. The four different wheezing groups consist of: (1) children who had not wheezed by
age six, (2) transient early wheezers who wheezed early in life, but not at six years of age, (3) late
wheezers who did not wheeze early in life, but did wheeze by six years old, and (4) persistent
wheezers who wheezed both in early life as well as at six years of age (Martinez et al., 1995).
34
-------
It is important to keep in mind factors other than disease or exposure, such as height, which
may be important in interpreting physiologic measures. The taller the subject, the larger the lung
volume (FRC). PEFV curves can be used to assess lung function at a point in time, as well as for
long-term follow up. In addition, PEFV curves can also be used in conjunction with cold dry air
(CDA) and methacholine challenges to assess bronchial reactivity.
When assessing lung function in older children one can use standard spirometry, since greater
subject cooperation is obtainable. These two measures (spirometry and PEFV) are comparable in
utility. However, spirometry is a more complete assessment of pulmonary function. Another
measure that can be used in both older and a younger child is peak expiratory flow rate (PEFR).
Though there is wide intersubject variability, the intrasubject reproducibility is good. This measure
is very good for trending. If a restrictive abnormality of lung function is found, one can further test
the older child by doing a diffusing capacity of the lung for carbon monoxide (DLCO). The DLCO
is a measure of gas exchange at the level of the alveolar membrane. Though it is a good indicator
of interstitial lung disease, it requires extensive subject cooperation and additional equipment.
In summary, it is clear that pollutants aggravate lung disease in children, but little is known
about chronic effects of pesticide exposure on the lung health of children. We have also seen that
while clinical data is inexpensive and reliable, it provides only non-directed information and may
not be available for all potential subjects. Questionnaires provide more in-depth information, but
require consideration of cultural and linguistic features of the population. Lastly, physiologic
outcome measures provide correlates to clinical/questionnaire data, correlates with disease, and
allow longitudinal tracking, but also add complexity and expense.
Figure 1
FRC
FLOW
(ml/sec)
FRC
VOLUME
35
-------
References
Abramson, M., and Voigt, T. 1991. Ambient air pollution and respiratory disease. Med J Aust
154:543-53.
Martinez, F.D., Wright, A.L., Taussig, L.M., Holberg, C.J., Halonen, M., Morgan, W.J. 1995.
Asthma and wheezing in the first six years of life. The Group Health Medical Associates. N
EnglJMed 332(3): 133-8.
Reigart, J.R. 1995. Pesticides and children. Pediatric Annals 24:663-8.
Taussig, L.M., Wright, A.L., Morgan, W.J., Harrison, H.R., and Ray, C.G. 1989. The Tucson
Children's Respiratory Study. I. Design and implementation of a prospective study of acute
and chronic respiratory illness in children. Am JEpi 129(6):1219-31.
Ware, J.H., Ferris, E.G., Dockery, D.W., et. al. 1986. Effects of ambient sulfur oxides and
suspended particles on respiratory health of preadolescent children. Am Rev Respir Dis
133:834-42.
36
-------
Pesticides and Childhood Cancer - An Overview
Jonathan Buckley
Department of Preventive Medicine
University of Southern California
Background
The epidemiological literature contains numerous studies that implicate pesticides as a risk
factor for increased cancer risk, in both adults and children. However much of that literature reports
modest increases in risk, does not establish that these associations are causal, and commonly does
not implicate specific agents. Appreciation of the limitations of the epidemiological approach, and
of the fact that many reported exposure-disease associations that have enjoyed brief notoriety have
faded under closer scrutiny, suggests that we need to be cautious about the link between pesticides
and cancer. Perhaps the most persuasive argument in support of the association is its specificity: in
study after study, where increases in cancer risk have been reported the malignancy arises from the
lymphoid or hematopoietic system. This is apparent in both case-control studies in adults, where
lymphomas are the malignancies most frequently linked to pesticide exposure, and in children where
studies of acute lymphoblastic leukemia (ALL), acute myoblastic leukemia (AML), and lymphoma
have all yielded positive results.
Methodological Considerations
In children, a cohort approach is infeasible because of the rarity of the endpoint—each year,
approximately 3 new cases of the most common cancer (ALL) occur per 100,000 children under the
age of 15. Occasionally, cancer clusters come under investigation, and some of these are in locales
where above average pesticide exposure naturally leads to suspicion of these chemicals, but the study
of clusters is generally a frustrating and unrewarding business and rarely yields useful insights into
the basis for the cluster. Ecological studies, showing spacial (and/or temporal) patterns of cancers
that correlate with documented patterns of pesticide use can provide indirect evidence for an
association, but such studies will be limited in the U.S. until a national population-based registry for
pediatric cancers is available.
In practice, case-control designs, with cases necessarily drawn from a very large population,
have provided most of the data on cancer risk associated with pesticide use.
Published Data on Pesticides and Cancer
Zahm and Blair (1992) recently summarized the data concerning pesticides and non-
Hodgkins lymphoma (NHL). They compiled reports from 21 cohort studies of farmers and found
11 with odds ratios (OR) greater than 1,3 significantly so, with OR ranging from 0.6 to 2.6. In 19
case-control and cross-sectional studies, which might be expected to provide better data, 12 gave OR
greater than 1 with 8 significant. Zahm and Blair conclude that these data are equivocal, possibly
due to the fact that exposure is inferred from a broad occupational category "farming."
Studies based on more specific exposure data have generally shown higher risk estimates.
Hardell et al. (1981) reported a 6-fold risk of lymphoma for persons with exposure to phenoxyacetic
37
-------
acid herbicides (a class including 2,4-D and 2,4,5-T) or chlorophenols. Hoar et al. (1986) found an
OR of 2.2 for NHL in farmers who used phenoxy herbicides, with OR over 7 for those reporting
more than 20 days use of 2,4-D per year. Risk was also increased in those not reporting use of
protective equipment. In a similar study in Nebraska the risk was 3.3 for farmers handling 2,4-D for
more than 20 day per year. LaVecchia et al. (1989) report a significant trend with duration of
exposure to herbicides and Persson et al. (1989) report an OR of 4.9 for NHL for occupations
exposed to phenoxy acids. Pearce et al. (1987) reported a significant OR of 3.7 for orchard workers.
Ollson and Brandt (1981) found a greatly elevated risk of cutaneous NHL (OR=10.0) with exposure
to phenoxy acids, but a much smaller OR (1.3) overall. Other pesticides have been studied less
extensively, but significant positive results have been reported for atrazine, chlorophenols, and
fungicides in general.
Similar associations have been reported for leukemia and multiple myeloma in adults (Viel
and Richardson, 1991; Burmeister, 1990; Brown et al., 1990; Cantor and Booze, 1989) and in
children (Buckley et al., 1989; Buckley et al., 1994; Lowengart et al., 1987; Savitz et al., 1995).
Childhood Cancer - Results from the Children's Cancer Group
The Children's Cancer Group (CCG) is a cooperative clinical trials group that includes over
100 hospitals in the U.S. and Canada. It has had an active epidemiological research program for
many years, and has conducted case-control studies on most of the major cancers in children
(Robison et al., 1995). Although the focus of these studies has varied, most have included questions
about exposure of the child (and parents) to pesticides.
The first generation of studies tended to cover a wide range of exposures and associations
of possible relevance, and consequently most topics received relatively superficial attention - in the
case of pesticide exposure, the questionnaire typically asked about pesticides as a single entity; in
some studies we asked about home, garden, and occupational exposures separately. Significant
associations were found for both acute myeloblastic leukemia (AML - Study CCG-E05, Table 1,
Buckley et al., 1989) and NHL (Study CCG-E08, Table 2).
As interesting as these results were, the lack of detail concerning the types of pesticides, the
pests being treated, periods of use, protective measures employed, and who was exposed, made
interpretation difficult. In the next generation of studies (CCG-E14 for AML, CCG-E15 for ALL,
and a proposed study of NHL) we attempted to obtain this detail, through a much more
comprehensive interview and, for CCG-E15, through in-home measurements. Analysis of these data
is underway.
Methods
As outlined earlier, ecologic studies and investigation of clusters have a role to play, but it
is likely that case-control studies will be continue to be the method of choice. However, the
investigator hoping to confirm and better delineate a link between pesticide exposure and childhood
cancer risk faces some formidable obstacles. These may include issues relating to the diagnosis, to
case and control selection, to control for sociodemographic confounders—as for any case-control
study—but the overriding problem is EXPOSURE ASSESSMENT.
38
-------
Table 1: Odds Ratios (OR) for Childhood AML Associated with Pesticide Exposures. (The
Exposure
Household pesticides (child)
Occupational pesticides (parent)
All ages
Age 5 and under
Monocytic/myelomonocytic
Category
None
1,000 days
None
1-1, 000 days
> 1,000 days
None
1-1, 000 days
> 1,000 days
Cases/Controls
128/148
46/33
13/9
8/3
164/182
16/15
24/7
75/89
8/5
12/1
49/59
7/2
6/1
OR
1.00
1.77
2.02
3.48
1.00
1.50
4.26
1.00
1.91
12.79
1.00
8.88
16.17
Table 2 : Odds Ratios (OR.) for Childhood NHL Associated with Pesticide Exposures.
exposed is given in parenthesis, p-values are tests of trend.)
Exposure
Household insecticides (mother)
Garden sprays (mother)
Exterminate around home (mother)
Herbicides or pesticides (child)
Occupational pesticides (parent)
Category
Never
l/mo
No
Yes
No
Yes
No
Yes
Cases/Controls
185/199
46/51
18/10
6/1
237/253
9/6
14/8
238/257
31/12
215/244
50/23
248/256
21/13
OR
1.00
0.97
1.94
6.45
1.00
1.60
1.87
1.00
2.78
1.00
2.47
1.00
1.67
p-value
0.04
0.001
0.001
0.002
(The person
p-value
0.06
0.13
0.004
0.0009
0.21
39
-------
Exposure Assessment
The simplest, cheapest, and most direct method is to ask the parents about exposures. While
most parents can provide information about general patterns of pesticide use, this approach in
unlikely to yield useful detail regarding the chemicals used, and generally gives only a crude
estimate of the amount the child will have been exposed to over his/her lifetime. For older children,
the questionnaire must elicit details of use going back many years. In addition there will always be
concerns that differential response rates of case and control parents will introduce a systematic bias.
Further complicating the issue is uncertainty about the critical period of exposure—pre-conception,
in utero, or postnatal?
Similarly, determining parental occupational exposure by questionnaire is problematic.
Again, the specific chemicals may be unknown and the quantities impossible to estimate reliably,
but in addition the pathway from exposure to effect in the child is not known. If it is via a genotoxic
effect to the testes/ovaries, pre-conception exposure is what counts; if the effect is due to exposure
of the child, the model must include a means of transferring the chemical from the workplace to the
home. Quantitating exposure, based on questionnaire data, under such a model is clearly impossible.
In CCG studies we have tried a number of approaches to determining pesticide exposures for
children in our studies. One is to concentrate on the products used (e.g., insect sprays); an alternative
is to concentrate on the pests (e.g., termites, weeds). Another approach that relies less on memory,
but may give a limited snapshot of pesticide use, is to take an inventory of products in the home at
the time of interview.
An alternative to the questionnaire-based approach is to sample from the child, parent, or the
home environment. While this seems to be an appealing way of avoiding the limitations of a
questionnaire-based approach, it is not without its own difficulties.
Environmental/Sampling
Direct measurement of residues of pesticides inside or around the home(s) that the child has
lived in is certainly possible under some circumstances. I will focus on housedust sampling, since
this is the approach used in a recent CCG study, but similar considerations will apply to other
sampling methods (e.g., air, water, or soil sampling).
Advantages to this approach are obvious there: such assays provide a quantitative and
objective measure of specific pesticide residues in proximity to the child.
While this is very appealing, the limitations and difficulties need also to be carefully
considered:
(1) Proper sampling requires specialized sampling equipment (essentially a purpose-built
vacuum cleaner), a clearly defined sampling protocol that covers such things as the area(s)
to be sampled, the vacuuming procedure, and cleaning procedures between samples. Clearly,
it also requires home access, which could be problematic for many reasons, especially for
national studies.
40
-------
(2) The specificity of the assays is a double-edged sword, since there is no way of knowing
whether the important residue(s) have been included in the list of substances to measure.
(3) Assays are expensive, which can limit the number of subjects being tested, and/or the
number of samples per household.
(4) There can be technical problems with the assays due in part to the fact that the sample
(dust) contains many organic contaminants. In the CCG study, there was substantial batch-
to-batch variation for many analytes.
(5) This approach measures only one source of exposure. Thus housedust sampling ignores
the contributions from water and food.
(6) The effective (internal) dose to the child is not known.
(7) The measurement is of the current concentrations of the analytes. This reflects not only
patterns of recent use, but also stability of the residue in housedust—which will vary greatly
for different residues—and patterns of past use, including use by previous occupants. Since
we know little about the most important period of exposure for these children, the extent to
which the measurement captures relevant data is not clear.
It is useful to compare housedust measurements with questionnaire data. Questionnaire data
can potentially be targeted to specific time periods, such as when the child was in utero, or can cover
an interval of many years and may include both the current and previous homes. However, the
quality of the data is questionable. Since a recent CCG case-control study of childhood ALL
included both questionnaire and housedust measurements, it is possible to determine the extent to
which these approaches agree, at least for some types of pesticides. Presented in Tables 3a and 3b
are preliminary results from this study. These tables show odds ratios for any versus no self-reported
use of pesticides to treat selected pests, based on questionnaire data alone, classifying individuals
according to the amount of pesticide (grouped according to the common use(s) of the pesticide)
found in the housedust of the child's home. In general, it can be seen that the self-report is consistent
with measurements made on the dust in that parents reporting that they treated fleas or ticks, say,
tended to fall into the upper tertile with respect to measured quantities of products used to treat fleas
and ticks. The associations are not particularly strong, but are remarkable none the less. There are
a host of reasons why associations might be expected to be weak and difficult to detect. (1) Products
classified under a particular use may have many other uses. (2) Products classified under a particular
use represent only a portion of all products available for that purpose and thus may not include the
product used by the parent. (3) The measurement reflects recent use plus a (variable) contribution
from past use in the home. The sample is from a single site, that may not be ideal for any given type
of pesticide. In contrast the reported use is either during a specific time period (the
pregnancy)—commonly many years previously—or averaged over most of the child's life. This use
may actually have been in a previous home. Usage could have occurred in any room, or outdoors.
(4) Misclassification is likely in the self-reported use due to difficulties in recalling events long in
the past. (5) Technical difficulties with the assays will create misclassifications in the measured
values.
41
-------
Biological Sampling
Biological sampling may circumvent a number of the problems associated with
environmental sampling (e.g., internal dose, exposure from multiple sources) but may not solve
others (e.g., choice of relevant analytes, cost). Drawing samples directly from the case (or control)
may be undertaken for essentially two reasons:
(1) to measure pesticide load in the body—for example the concentration of DDT residues
in fat.
(2) to measure possible biological effects of pesticide exposure. Such effects could be
sufficiently specific as to implicate a single compound (e.g., adduct formation) or may be
quite non-specific and only be tied to pesticide exposure through ancillary data, such as self-
reported pesticide use. One biological marker of interest is the frequency of translocations
mediated by VDJ-recombinase. This enzyme system is responsible for splicing of genomic
DNA in maturing lymphocytes to produce diversity in antibody and T-cell receptor proteins.
It also appears to act aberrantly, at low frequency in normal lymphocytes to produce
translocations at a variety of sites (Cortopassi and Amheim, 1992; Fuscoe et al., 1992), and
some of these translocations are important in the pathogenesis of leukemia (Gu et al., 1992)
and lymphoma (Kirsch and Lista, 1996). Completing this model is a study which shows that
agricultural workers had an elevated frequency of aberrant VDJ recombination during the
summer months, when they were most exposed to pesticides (Lipkowitz et al., 1992). CCG
currently has a proposal under review to obtain questionnaire data on pesticide use from
approximately 600 children with lymphoma and an equal number of matched controls, and
to examine the frequency of aberrant VDJ recombinase translocations in the cases.
Conclusions
While there is ample reason to suspect that exposure to some pesticides may increase
the risk of lymphoma and leukemia in children, confirming the link in a convincing fashion
and determining what the agent(s) and mechanism(s) are, is a formidable challenge.
Questionnaires can be improved to obtain the most reliable information possible, but it is
unlikely that questionnaire data alone will suffice. Quantitative assessment of exposure
and/or biological markers associated with pesticide exposure, combined with interview data,
may be necessary to make substantial progress.
42
-------
Table 3a: Comparison of Pests Treated (By Questionnaire) and Detection of Pesticide
Residues in Housedust. The period (for self-report) is during the index
pregnancy only, (p-values are for trend.)
Pesticide Group Pests Treated (According to Questionnaire)
Fleas/ticks Termites Weeds
Flea/tick products
Termite control
Herbicides
Tertile
1
2
3
1
2
3
1
2
3
1.00
1.62
2.60
p=0.009
1.00
1.06
1.61
p=0.06
1.00
1.43
0.63
p=0.93
-
1.00
1.83
3.02
p=0.23
1.00
1.81
0.87
p=0.74
1.00
1.67
1.04
p=0.39
1.00
1.38
1.05
p=0.62
1.00
2.27
1.47
p=0.01
Table 3b: Comparison of Pests Treated (By Questionnaire) and Detection of Pesticide
Residues in Housedust. The period (for self-report) is from the index pregnancy
to a year prior to diagnosis, (p-values are for trend.)
Pesticide Group Pests Treated (According to Questionnaire)
Fleas/ticks Termites Weeds
Flea/tick products 1 1.00 1.00 1.00
0.63
0.78
p=0.23
Termite control 1 1.00 1.00 1.00
0.83
0.84
p=0.44
Herbicides Below 1.00 1.00 1.00
1.58
p=0.08
Tertile
1
2
3
1
2
3
Median
Below
Above
1.00
4.32
5.16
p<0.0001
1.00
1.69
3.61
pO.OOOl
1.00
1.00
p=0.99
1.00
0.88
1.32
p=0.59
1.00
0.61
2.06
p=0.07
1.00
0.85
p=0.93
43
-------
References
Brown, L.M., Blair, A., Gibson, R., et al. 1990. Pesticide exposure and other agricultural risk factors for
leukemia among men in Iowa and Minnesota. Cancer Res 50:6585-91.
Buckley, J.D., Versteeg, C.M., Ruccione, K., et al. 1994. Epidemiological characteristics of childhood acute
lymphocytic leukemia. Analysis by immunophenotype. Leukemia 8:856-64.
Buckley, J.D., Robison, L.L., Swotinsky, R., et al. 1989. Occupational exposures of parents of children with
acute nonlymphocytic leukemia. Cancer Res 49:4030-7.
Burmeister, L.F. 1990. Cancer in Iowa farmers: Recent results. Am JIndust Med 18:295-301.
Cantor, K.P., and Booze, C.F. 1991. Mortality among aerial pesticide flight instructors. Arch Environ Med
46:110-6.
Cortopassi, G.A., and Arnheim, N. 1992. Using the polymerase chain reaction to estimate mutation
frequencies and rates in human cells. Mutation Res 277:239-49.
Fuscoe, J.C., Zimmerman, L.J., Harrington-Brock, et al. 1992. V(D)J recombinase-mediated deletion of the
hprt gene in T-lymphocytes from adult humans. Mutation Res 283:13-20.
Gu, Y., Cimino, G., Alder, H., et al. 1992. The t(4;ll)(q21;q23) chromosomal translocations in acute
leukemias involve the VDJ recombinase. Proc Natl Acad Sci 89:10464-8.
Hardell, L., Eriksson, M., Lenner, P., and Lungren E. 1981. Malignant lymphoma and exposure to chemicals,
especially organic solvents, chlorophenols and phenoxy acids: A case-control study. Br J Cancer
43:169-76.
Hoar, S.K., Blair, A., Holmes, F.F., et al. 1986. Agricultural herbicide use and risk of lymphoma and soft-
tissue sarcoma. JAMA 256:1141-7.
Kirsch, I.R., and Lista, F- 1996. Transrearrangements as biomarkers for risk of lymphoid malignancy.
Cancer Surveys 28:311-27.
LaVecchia, C., Negri, E., D'Avanzo, B., and Franceschi. 1989. Occupation and lymphoid neoplasms. Br J
Cancer 60:385-8.
Lipkowitz, S., Garry, V.F., and Kirsch, I.R. 1992. Interlocus V-J recombination measures genomic instability
in agricultural workers at risk for lymphoid malignancies. Proc Natl Acad Sci 89:5301-5.
Lipkowitz, S., Stem, M.-H., and Kirsch, I.R. 1990. Hybrid T cell receptor genes formed by interlocus
recombination in normal and ataxia-telangiectasia lymphocytes. JExp Med 172:409-18.
Lowengart, R.A., Peters, J.M., Cicioni, C., Buckley, J., et al. 1987. Childhood leukemia and parent's
occupational and home exposures. J7VC7 79:39-46.
Ollson, H., and Brandt. L. 1981. Non-Hodgkin's lymphoma of the skin and occupational exposure to
herbicides. Lancet 2:579.
Pearce, N.E., Sheppard, R.A., Smith, A.H., andTeague, C.A. 1987. Non-Hodgkin's lymphoma and farming:
An expanded case-control study. Int J Cancer 39:155-61.
Persson, B., Dahlander, A., Fredriksson, M., et al. 1987. Soft tissue sarcoma and non-Hodgkin's lymphoma
in relation to phenoxyherbicide and chlorinated phenol exposure in western Washington. JNCI78:
899-910.
Robison, L.L., Buckley, J.D., and Bunin, G. 1995. Assessment of environmental and genetic factors in the
etiology of childhood cancers: The Children's Cancer Group epidemiology program. Environ
Health Perspect 6:111-6.
Viel, J.-F., and Richardson, S.T. 1991. Adult leukemia and farm practices: An alternative approach for
assessing geographical pesticide exposure. Soc Sci Med 32:1067-73.
Zahm, S.H., and Blair, A. 1992. Pesticides and non-Hodgkin's lymphoma. Cancer Res 52:5485s-8s.
44
-------
Health Effects of Pesticides
D.J. Ecobichon, Ph.D.
Queen's University
Department of Pharmacology and Toxicology
Kingston, Ontario, Canada
While the efficacy of pesticides for the protection/preservation of health, food, and fiber
cannot be disputed, there are costs for these benefits, one being an estimated, annual, 3 million cases
of acute, life-threatening intoxications worldwide, perhaps as many or more unreported cases and
some 220,000 deaths (Hayes, 1992). There is a long history of pesticide poisonings among workers
in agriculture, in residential treatment and in public health (Hayes, 1992). However, none of the
numbers quoted take into account persistent and/or delayed effects arising from either acute, high-
level or from prolonged, low-level exposures, including adverse reproductive outcomes (fertility,
abortions, teratology), neurological and neurobehavioral development. Such endpoints of toxicity
are poorly studied or go unreported except as "interesting" anecdotal cases in the literature. The
incidence of acute poisonings in emerging nations is some 13-fold higher than in industrialized,
agricultural nations. One can only suspect that there might be an association of similar magnitude
for other endpoints of toxicity.
I am not going to address pesticide-induced, acute toxicity since this aspect has been well
described by many authors, myself included (Ecobichon, 1996; Ecobichon, 1995). My interest lies
with the immediate and later-developing neurotoxicity as a consequence of acute and prolonged
exposure to pesticides. By way of example, look at the scenarios-presented in Figure 1. The
afflicted individuals all showed similar neurological sequelae upon presentation and during later
progression of their conditions. Identification of the causative agent would have required extensive
chemical analysis. Neurological assessment would not have been conclusive. To simplify the cases,
the first was caused by manzidan (Maneb-zineb, a fungicidal mixture), the second by diazinon. an
insecticide, and the third by the herbicide, 2.4-D. Similar signs/symptoms can be elicited by
different chemicals (Ecobichon and Joy, 1994).
Exposure may involve concentrated products or diluted, prepared-for-use formulations, either
by dermal and/or inhalation: (1) during preparation and application; (2) as bystanders in sprayed
areas (fields, greenhouses, homes); or (3) via environmentally deposited agents in water, food, or
soil. As is shown in Figure 2, one is not only dealing with the active ingredients but also with
"inerts," both petrochemical-based or completely synthetic compounds, many having the potential
to elicit neurotoxicity themselves, without even contemplating possible interactions with the toxicity
elicited by the pesticides. Perhaps, the best example is that of the herbicide glyphosate (N-
phosphonomethylglycine) where the toxicity (gastrointestinal, respiratory, cardiovascular, CNS) can
be attributed largely to the surfactant poly-oxyethyleneamine (POEA) used in formulations to
promote wetting of plant surfaces and rapid penetration (Sawada et al, 1988; Adam et al., 1997).
My experience in Central America and southeast Asia with agricultural pesticide usage is
that, once the crop is planted, the husband usually migrates to an urban area in search of work,
leaving the care of the crop, including pesticide applications, to his wife and the older, pre- and post-
45
-------
pubertal children. When spraying occurs, everyone goes to the fields or paddies, the younger
children tending the toddlers on site, frequently on ground areas where chemicals are stored, spilled
from concentrates during dilution or dumped at the day's end when tanks are rinsed. With little in
the way of protective clothing, those spraying (mothers, sons, daughters) receive extensive dermal
exposure. The younger family members, staying at the storage/mixing/loading site, will receive oral
as well as dermal exposure from the contaminated soil.
Do such agricultural practices elicit adverse health effects? In southeast Asia, there is an
exceptionally high incidence of birth defects in agricultural (rice growing) regions. With rice, there
is extensive use of herbicides, insecticides, and fungicides from preplanting through to harvesting.
A colleague in Thailand just shakes his head when questioned about the obvious association. He has
been unable to amass a suitable data base through regional hospitals. In the same agricultural region,
there were an estimated 8,268 pesticide poisonings/100,000 workers reported for 1983 (Boon-Long
etal., 1986).
Returning to neurotoxicity, early data from Washington state reveals just how much material
(DDT, parathion) was being applied to apple orchards (Table 1) (Batchelor, 1953). In 1951 -52, DDT
was rapidly becoming the panacea for all insect problems while parathion was making its debut as
an alternative insecticide. Not surprisingly, acute toxicity was observed! One of my colleagues
claims that if it had not been for parathion, it might have taken much longer to leam about this class
of insecticides. The paper of Grob et al. in 1950 provided conclusive evidence of the methanism(s)
of organophosphorus ester (parathion) toxicity (Grob et al., 1950). The paper of Summerford et al.
in 1953 and the landmark paper of Batchelor and Walker in 1954 demonstrated both the toxicity and
the estimated occupational exposure to parathion (Summerford et al., 1953; Batchelor and Walker,
1954).
By the mid-1950s and early 1960s, there was ample evidence in the literature of the persistent
neurological effects following organophosphorus ester poisoning in agricultural workers. However,
much of this data was ignored or was treated as anecdotal or irrelevant. In Canada, Davignon and
her colleagues, in a study of apple growers, demonstrated higher (and persistent) incidence of
neurological manifestations and anomalies among insecticide handlers which correlated well with
the number of years of exposure to both organochlorine and organophosphorus insecticides
(Davignon et al., 1965). The 1983 study by Wharton and Obrinsky was, in my opinion, a watershed
publication for those interested in long-term effects, the study demonstrating the persistence of
blurred vision, muscle weakness, headaches, and nausea for up to four months after poisoning
(Whorton and Obrinsky, 1983). It would not be surprising to learn that, even in the present decade,
there are studies (and highly sophisticated ones) still being conducted on orchard workers in
Washington state (Rosenstock et al., 1990).
46
-------
TABLE 1
Ground Application of DDT and Parathion to Fruit Orchards
in North Central Washington State in 1951 and 1952
Formulations Material Applied (Ib/acre)
Liquid
(suspension
or emulsion)
Dusts
H
22.5
25.0
12.0
DDT
L M
1.0 15.1
1.8 9.4
10.7 10.3
Parathion
H
5.0
9.8
1.6
L
0.3
0.2
1.2
M
2.3 *
1.6**
1.4**
* Hand-spray equipmen
** Portable, power-driven (truck, tractor drawn)
My personal involvement in the persistent neurological problems of pesticide poisonings began
in 1976 with a case of acute intoxication of an adult, female technician by the organophosphorus ester,
fenetromion (Ecobichon et al., 1997). The patient developed an interesting set of neurological and
psychiatric sequelae approximately three weeks after the subsidence of the characteristic acute signs and
symptoms, these persisting to some degree for almost a year after the event (Table 2). These
signs/symptoms were common in the nerve gas literature but not for insecticides except in "anecdotal"
published papers. Intrigued, a colleague and I reviewed the literature pertaining to pesticides and
neurological diseases, these efforts culminating in a book of the same title, in 1982 and a revised, second
edition in 1994 (Ecobichon and Joy, 1982; 1994). By 1994, there was a lot more relevant material in the
literature, as people began to recognize subtle neurological changes. To make a long story much shorter,
in addition to the neuromuscular effects (fasciculations, tremors, persistent muscular weakness), there are
a large number of neurobehavioral changes associated with pesticide exposure (Table 3). How does one
begin to test those parameters?
TABLE 2
PERSISTENT SYMPTOMS ASSOCIATED WITH
ACUTE FENITROTHION POISONING*
Anxiety
Frequent headaches
Nausea
Inability to concentrate
Trembling, muscle spasms
Muscle cramps, especially in legs
Generalized muscle weakness
Fatigue and lethargy
Mental depression
* From Ecobichon et al. Can. Med. Assoc. J. U6, 377-380 (1977)
47
-------
Over the years, a variety of test batteries have been developed, the significant characteristic
being the frustrating mix-and-match approach as investigators select some tests and reject others in
attempts to fine-tune measurements in the hope of improving the results in a meaningful manner. This
has been only partially successful. One early test battery is that developed by Feldman et al. (Feldman et
al., 1980) (Table 4). The WHO-UNDP core test battery has proven useful but, as can be seen in Table 5,
it is specific only for anti-cholinesterase insecticides (Maroni, 1986). WHO has also described a
behavioral test battery (Table 6). The most recent assessment system has been that of Stephens and his
colleagues, developed and tested in the "sheep dip" problem in the U.K., involving diazinon,
propetamphos, or a mixture of diazinon and chlorfenvinphos (Stephens et al., 1995; Stephens et al.,
1996) (Table 7). While we have not achieved "nirvana" in this field, the fine-tuning is producing more
sophisticated endpoints of toxicity, as Stephens" results show.
TABLE 3
NEUROBEHAVIORAL SEQUELAE RELATED TO
PESTICIDE POISONINGS
FEATURES
EFFECTS
Cognitive
Disturbances
Expressive Language
Deficits
Psychopathological
Sequelae
Reduced vigilance and
alertness
Attention deficits
Slowed information processing
Psychomotor retardation
Impaired memory functions
Reduced comprehension
Speech difficulties
Speech slurring, reduced
enunciation
Difficulties in
saying what is intended
formulating thoughts
repetition
Depression
Restlessness, insomnia
Excessive dreaming
Emotional lability
Weeping spells
Schizophrenic reactions
Irritability
Phobias
Outbursts of temper (rage)
Belligerent behavior
Obsessive-compulsive behavior
The current legal matters in the southern U.S.A. associated with the indoor use of chlorpyrifos
and effects on children bear examination. Given the now-familiar scenario that children are not "little
48
-------
adults", body size/surface area differences and the ease of penetration of the skin, it is not surprising that
toxicity has been observed with this rather potent organophosphorus ester. As is shown in Table 8, the
number of children and adults "poisoned" by homeowner use of one agent, chlorpyrifos, is impressive to
say the least (Blondell, 1997). To my knowledge, no one has addressed the possibilities of delayed
and/or persistent effects. In a recent paper studying acute poisonings by two rather potent carbamate
insecticides (methomyl, aldicarb), it would appear that children showed predominantly CNS effects
(depression, seizures, hypotonia), while adults showed mostly PNS symptoms (miosis, fasciculations,
bradycardia, bronchorrhea) (Lifshitz et al., 1997) (Table 9). These results suggest differences in
susceptibility as well as in target sites.
TABLE 4
Common Tests Used to Detect Behavioral
Effects of Neurotoxins*
TEST TYPE
TEST FUNCTION
Memory
Overall
Intelligence
Sustained Attention
Dexterity and Eye-
Hand Coordination
Reaction Time
Wechsler Memory Scale
Wechsler Adult Intelligence
Scale (WAIS)
Continuous Performance Test
Bourdon-Wiersma Vigilance
Neisser Letter Search
Santa Ana Dexterity Test
Flanagan Coordination Test
Michigan Eye-Hand Coordination
Test
Finger-tapping Test
Simple reaction time test
Choice reaction time test
Psychomotor Function
Mira Test
Digit-symbol Substitution
Task
Personality (Mood)
Eysenck personality Inventory
Rorschach Test
Feeling-tone Checklist
*Feldman et al. Am. J. Indus. Med. 1,211-227
(1980).
How will we test subtle, neurotoxicological, psychological, and/or behavioral
changes/adaptations in children? Most of the test batteries in use are focused on adults, are computer-
driven, and have complex instructions and paradigms developed to isolate and identify specific
endpoints/deficits. Reading some of the instructions for conducting these tests is a challenge in itself.
49
-------
We have gone so far in the sophistication of these tests, from what used to be done with paper and pencil
to keyboards and VDTs, that these have become impractical as simple tests. There are test batteries, e.g.,
the acquired cognitive test (ACT) by Stollery, which have seen use in heavy metal poisoning and solvent
exposure (Stollery, 1996a; Stollery, 1996b; Stollery, 1996c). There are neurological assessment tests for
children exposed to heavy metals. Can these be simplified and adapted for children's exposure to
pesticides, co-solvents, emulsifiers, and surfactants? It would be a challenge for pediatric neurologists.
While many young children, aged 7 years or older are computer literate, how can younger children (aged
5 or less) be tested?
TABLE 5
WHO-UNDP Core Methods for Health Assessment
Following Pesticide Exposure*
Assessment Data Collected
Exposure/Absorption Type of agent, amount
Metabolites in urine
Plasma and erythrocytic
cholinesterase activities
Health Effects Background (health and
Occupational)
Symptoms/signs (based on
VBC 82/1 and WHO/NIOSH
questionnaire
Neurological examination
(semiquantitative)
Neurobehavioral examination
Nerve conduction (sural, ulnar
and peroneal nerves).
Neuromuscular junction function
Behavioral tests (WHO battery)
*Maronietal. Toxicol. Letters 33, 115-123 (1986)
The question that comes to mind is: why are we observing persistent neurological and
neurobehavioral effects in adults and children as a consequence of exposure to pesticides? Are there any
testable hypotheses? The anticholinesterase-induced neuropathic sequelae are poorly understood, and
the search for other mechanisms of action continues (Figure 3). Firstly, there is considerable evidence of
a direct interaction between organophosphorus esters and secondary targets such as muscarinic (mAChR)
and nicotinic (nAChR) acetylcholine receptors, the results showing a competitive block (antagonism) at
mAChR and an induced desensitization of nAChR following a partial agonistic effect (Eldefrawi et al.,
1992). Evidence suggests that organophosphates may cause a hyperpolarization of nAChR with an
inhibition of ACh binding (Bartels and Nachmansohn, 1969). Some of this work goes back to the
extensive studies by Dettbam where injections of small doses of paraoxon into the regions caused
necrotic damage that now appears NOT to be artifact (Wecker and Dettbarn, 1977). At micromolar
levels in the circulation, certainly attainable in poisonings, organophosphorus esters (or oxon
metabolites) may directly induce toxicity at mAChR and nAChR whereas, at nanomolar levels, toxicity
is due to the inhibition of nervous tissue AChE and the effects of ACh (Bakry, 1988). However, in many
50
-------
studies, the agents tested (nerve gases, DFP, echothiophate) are not really representative of the
insecticide molecule beyond sharing the same chemical classification.
TABLE 6
The WHO Behavioral Test Batterv*
Reaction Times
Santa Ana Test
Wechsler Digit-
Symbol Test
Profile of Mood
States
Aiming Test
Digit Span Test
Benton Visual
Retention Test
Digit-Symbol Test
Helsinki Subjective
Symptom Ques-
tionnaire
Test Objectives/Purposes
Auditory and visual
Timed perceptual-motor
coordination
Measuring perceptual and
motor speed
Measuring mood and affective
states
Measuring hand steadiness
Immediate auditory memory
Visual memory
Measuring perceptual organi-
zation, motor dexterity,
attention, speed of per-
formance
Investigates psychological,
neuro-vegetative, gastrointestinal and
neurological symptoms
*Maroni et al Toxicol. Letters 33, 115-123 (1986)
Secondly, there is also considerable experimental evidence of a loss of acetylcholine (ACh)
receptors during severe poisonings by organophosphorus ester insecticides AND prevention of
their recovery (or re-synthesis) by localized damage and necrosis. This aspect was reviewed in
a number of chapters in Chambers and Levi (1992). As is shown in Figure 4, the
superabundance of ACh at nerve endings during the acute phase of the poisoning can have two
effects: (1) stimulation, with a subsequent depolarizing blockade which is reversible; or (2)
overstimulation, with a desensitization (reduced binding and affinity) of the receptors and a
down-regulation of the receptor population as a built-in safety measure, with possible damage
(necrosis?) to the postsynaptic membrane and an inability to synthesize new receptors. This
has been shown to occur with both mAChR and nAChR in the peripheral and central nervous
systems.
51
-------
TABLE 7
Test Batteries Used in the "Sheep Dip'
Studies in the United Kingdom*
Questionnaires Demographic, lifestyles
pesticide exposure history
general health
Cognitive Tests Automated Cognitive Test
short-term memory
visual spatial memory
Sustained Attention-reaction time
Information Processing
symbol-digit test
Syntactic Reasoning
truth-falsehood of statements
Long-Term Memory Test
Serial Word Learning Test
*Stephens et al - Lancet 345, 1135-1139 (1995)
- Neurotox. Teratol. 18, 449-453 (1996)
Can either of these hypotheses, admittedly focussed on the organophosphorus esters,
explain the persistent effects seen following anticholinesterase-type insecticide poisoning? In
my opinion they can and are testable! Some years ago, carbaryl, administered subchronically
to swine, was found to produce CNS edema and fragmentation of myelin tracts along with
neuromuscular lesions showing necrosis (ischemic myodegeneration, acute hyaline and
vacuolar degeneration, dystrophic calcification) (Smalley et al., 1969). More recently, carbaryl
has been associated with persistent neuromuscular weakness and CNS effects in at least one
individual. This and other incidents known to me have involved exceptionally high-level,
exposure (Ecobichon and Joy, 1994). Is the active ingredient always the culprit? Could it be
something else in the formulation (organic co-solvents, emulsifiers, surfactants, wetting or
stabilizing agents) that causes the neurotoxicity? This takes me back to Figure 1 and scenarios
#1 and #3 where the neurotoxicities were due to the fungicide manzidan and 2,4-D,
respectively. There were other "ingredients" in those formulations, exposure to solvents and/or
emulsifiers being excessive. Can these neurotoxicological test batteries differentiate between
solvent-induced neuropathic changes and a pesticide-related toxicity? This is a definite
challenge for future research!
52
-------
TABLES
Adult and Child Chlorpyrifos-Related Cases Seen or Referred
to Health Care Facilities (HCF) and Hospitalized-U.S.
Poison Control Centers, 1985-1992.
Age Group Health Care Facility Hospitals
Homeowner PCO Homeowner PCO
Non-occupat.
adults
877
393
70
34
Children 828 150 84 23
0-5 yrs
Homeowner - 44 products used in home
PCO - 10 products used by pest control operators
TABLE 9
Symptoms Observed in Children and Adults
With Methomyl and Aldicarb Intoxication*
Children
Symptoms
Coma/Stupor
Hypotonia
Seizures
Miosis
Fasciculations
Bradycardia
Diarrhea
Salivation
Bronchorrhea
N
36
36
3
20
2
6
12
0
0
(%)
100
100
8
55
5.5
16
33
Adults
N
0
0
0
22
20
8
0
2
4
(%)
91.7
83.3
33.3
8.3
16.6
*Lifshitz et al ClinToxicol. 35,25-27(1997)
53
-------
POISONING SCENARIOS
#1 MANZIDANP (Maneb, zineb) FUNGICIDE
Ethylene bis-
dithiocarbamates
Dizziness, tiredness, muscle weakness, headache, nausea, slurred spech, disorientation, tonic and
cloinic convulsions, loss of consciousness
#2 DIAZINON organophosphorus INSECTICIDE
ester
Flu-like signs (headache, nausea, vomiting), physical weakness, slurred speech, muscle spasms,
disorientation. memory loss, mental confusion
#3 2,4,-D chlorophenoxy HERBICIDE
acetic acid
Nausea, vomiting, headache, dizziness, fatigue, muscle pain, myotonia
Fig. 1 Three poisoning scenarios involving pesticides, the signs and symptoms being similar, although
caused by different pesticides - a fungicide, an insecticide and an herbicide.
PESTICIDE EXPOSURES
CONCENTRATES
SOLUBLE
EMULSIFIABLE
WETTABLE POWDERS
DUSTS
DILUTIONS
"INERTS"
DUSTS
LIQUIDS
AEROSOLS
Cold fog
Smoke generated
ACTIVE
INGREDIENTS
CO-solvents
Emulsifiers
Diluents
Surfactants
Fig. 2 Effects resulting from exposure to pesticides should be considered in the light of not only the
formulation active ingredients but also the "inert" ingredients, agents required in the final end
use product to maintain stability during spraying.
54
-------
OPs
(CARBA1TES ?)
SITES
9
AChE (n.t.) — ACh-related
effects
ACh-related
partial agonist
other effects
•7
in the CNS
Ineuromuscular
function
Fig. 3 A schematic diagram of sites of action of organophosphonis and possibly of carbamate ester insecticides in
neuronal tissue. At nanomolar (nM) concentrations, inhibition of nervous tissue acetylcholinesterase
(chE) would occur, with acetylcholine-related signs/symptoms. At micromolar (uM) levels, frequently seen in
poisonings, direct action at muscarinic receptors (MAChR) and/or nicotinic receptors (nAChR) may result in a
competitive blockade, hyper-polarization and/or desensitization of the receptors.
ACh
STIMULATION
A
OVER-STIMULATION
r«v«r«ltilt
DEPOLARIZING
BLOCKADE
DESENSITIZATION
DOWN-REGULATION
OF RECEPTORS
DAMAGE
NECROSIS
Fig. 4 A diagramatic representation of acetylcholine-induced effects of exposure to anticholinesterase-type
insecticides. Low levels of agent might produce the classical events of stimulation with a reversible depolarizing
blockade. High levels of accetylcholine (Ach) might elicit over-stimulation followed by desensitization and down-
regulation of receptor numbers, with necrotic damage to postsynaptic and neuromuscular membranes and no
resynthesis of the normal receptor population.
55
-------
References
Adam. A., et al. 1997. The oral and intratracheal toxicities of Roundup and its components in rats. Vet. Human
Toxicol. 39:147-51.
Bakry, N'.M., et al. 1988. Direct actions of organophosphate anti-cholinesterases on nicotinic and muscarinic
acetylcholine receptors. J. Biochem. Toxicol. 3:235-59.
Bartels, E., and Nachmansohn, D. 1969. Organophosphate inhibitors of acetylcholine-receptor and -esterase tested
on the electroplax. Arch. Biochem. Biophys. 133:1-10.
Batchelor, G.S. 1953. Survey of insecticide spray practices used in the fruit orchards of north central Washington.
AMA Arch. Indus Hyg. Occup. Med. 7:399-401.
Batchelor. G.S., and Walker, K.C. 1954. Health hazards involved in use of parathion in fruit orchards of north
central Washington. AMA Arch. Indus. Hyg. Occup. Med. 10:522-9.
Blondell, J. 1997. Review of chlorpyrifos poisoning data. Memorandum. Washington, DC: U.S. EPA.
Boon-Long, J., et al. 1986. Toxicological problems in Thailand. In: Environmental toxicity and carcinogenesis
(Ruchirawat, M., and Shank, R.C., eds). Test and Journal Corp. Bangkok, 283-93.
Chambers, I.E., and Levi, P.E. 1992. Organophosphates. Chemistry, fate and effects. San Diego: Academic Press.
Davignon, L., et al. 1965. A study of the chronic effects of insecticides in man. Can. Med. Assoc. J. 92:597-602.
Ecobichon, D.J. 1995. Pesticides. In: Principles of pharmacology. Basic concepts and clinical applications
(Munson, P.L, ed). New York: Chapman Hall, Chapter 107, 1563-79.
Ecobichon, D.J. 1996. Toxic effects of pesticides. In: Casarett and Doull's toxicology. The basic science of
poisons. Fifth Edition (Klaassen, C.D., ed.). New York: McGraw Hill, Chapter 22, 643-89.
Ecobichon, D.J., et al. 1977. Acute fenitrothion poisoning. Can. Med. Assoc. J. 116:377-9.
Ecobichon, D.J., and Joy, R.M. 1982. Pesticides and neurological diseases. Boca Raton, FL: CRC Press, Inc.
Ecobichon, D.J., and Joy, R.M. 1994. Pesticides and neurological diseases. Second edition. Boca Raton, FL:
CRC Press, Inc.
Eldefrawi, A.T., et al. 1992. Direct actions of organophosphorus anticholinesterases on muscarinic receptors. In:
Organophosphates. Chemistry, fate and effects (Chambers, J.E., and Levi, P.E., eds.). San Diego:
Academic Press, 257-70.
Feldman, R.G., et al. 1980. Neuropsychological effects of industrial toxins: A review. Am. J. Indus. Med. 1:211-
27.
Grob, D., et al. 1950. The toxic effects in man of the anti-cholinesterase insecticide parathion (p-nitrophenyl
diethyl thionophosphate). Bull Johns Hopkins Hasp. 87:106-29.
Hayes, Jr., W.J. 1982. Pesticides studied in man. Baltimore: Williams and Wilkins.
Lifshitz, M., et al. 1997. Carbamate poisoning in early childhood and in adults. Clin. Toxicol. 35:25-7.
Maroni, M., et al. 1986. The WHO-UNDP epidemiological study on the health effects of exposure to
organophosphorus pesticides. Toxicol. Letters 33:115-23.
Rosenstock, L., et al. 1990. Chronic neuropsychological sequelae of occupational exposure to organophosphate
insecticides. Am. J. Indus. Med. 18:321-5.
Sawada, Y., et al. 1988. Probable toxicity of surface-active agent in commercial herbicide containing glyphosate.
Lancet 1:299.
Smalley, H.E., et al. 1969. The effects of chronic carbaryl administration on the neuromuscular system of swine.
Toxicol. Appl Pharmacol. 14:409-19.
Stephens, R., et al. 1995. Neuropsychological effects of long-term exposure to Organophosphates in sheep dip.
Lancet 345:1135-9.
Stephens, R., et al. 1996. Organophosphates: The relationship between chronic and acute exposure effects.
Neurotox.Teratol. 18:449-53.
56
-------
Increased Sensitivity to Pesticides in the Young: Possible Explanations
S. Padilla1, S.R. Mortensen2, S.M. Chanda3, and V.C. Moser1
'Neurotoxicology Division, U.S. EPA, Research Triangle Park, NC
2American Cyanamid Co., Princeton, NJ
3NIEHS, ResearchTriangle Park, NC
Our laboratory is especially interested in determining if children may be more sensitive
than adults to the effects of anticholinesterase pesticides. Our investigations have centered
around characterizing the mechanism(s) for this postulated age-related sensitivity in a standard
laboratory animal, Long-Evans rats.. We have chosen the most commonly used
organophosphorus pesticide, chlorpyrifos [Dursban® or Lorsban®, O,O-diethyl 0-(3,5,6-
trichloro-2-pyridyl) phosphorothionate], as our first pesticide to investigate in depth.
It has been known for over 30 years that young postnatal animals may be more
sensitive than mature animals to the lethal effects of organophosphorus pesticides, but little
work has been done comparing the biochemical and behavioral endpoints in young and adult
animals at less-than-lethal dosages, or to explore the mechanisms for this differential
sensitivity. Acutely administered chlorpyrifos is approximately 5 times more toxic to young
rats (postnatal day 17, PND17) as compared to adult rats, measured by changes in clinical signs
and motor activity (1,4). The present group of experiments was designed to determine the basis
for this increased sensitivity in the young. Our research plan was to look at two general factors:
(1) whether the target enzyme, cholinesterase, was more sensitive to chlorpyrifos inhibition in
the young and (2) whether the young were less able to detoxify the pesticide and its
metabolites.
Our research to date has shown that the brain cholinesterase in very young rats (i.e.,
PND4) and adults is equally sensitive to inhibition by various organophosphorus or carbamate
pesticides (2,3). These data would not support target enzyme sensitivity as an explanation for
age-related sensitivity to the acute effects of pesticides.
In other experiments which were designed to assess some of the toxicokinetic factors of
chlorpyrifos toxicity, we delineated the developmental profiles of two detoxification enzymes:
A-esterase and carboxylesterase activity. A-esterases detoxify by hydrolyzing the active
metabolite of chlorpyrifos, chlorpyrifos oxon, and carboxylesterases detoxify by binding up
and deactivating chlorpyrifos oxon. We found that young animals are severely deficient in
these enzymes (2-4). As the animals mature, their A-esterase and carboxylesterase activities
generally increase and their sensitivity to acute chlorpyrifos toxicity concurrently decreases.
The above data indicate that one explanation for increased sensitivity of the young to
chlorpyrifos toxicity may be their inability to detoxify chlorpyrifos as efficiently and rapidly
as adults. Given this explanation, what evidence is there that children may also be deficient in
these detoxification enzymes? A search of the literature revealed that a few studies (5,6) using
human serum/plasma indicate that young humans (i.e. below 2 to 8 years of age) are also
57
-------
deficient in these detoxification enzymes—a fact which should cause some concern when
considering the potential effects of anticholinesterase pesticides in children.
References
Moser,V.C, Padilla, S., Hunter, D., Marshall, R.S., McDaniel, K.L., and P. M. Phillips. (1998) Age-
and gender-related differences in the time-course of behavioral and biochemical effects
produced by oral chlorpyrifos in rats. Fundam. Appl. Toxicol. 149:107-119.
Mortensen, S.R., Chanda, S.M., Hooper, M.J. and S. Padilla. (1996) Maturational differences in
chlorpyrifos-oxonase activity may contribute to age-related sensitivity to chlorpyrifos. J.
Biochem. Toxicol. 11:279-287.
Mortensen, S.R., Hooper, M.J., and S. Padilla. (1998) Rat brain acetylcholinesterase activity:
Developmental profiles and maturational sensitivity to carbamate and organophosphorus
inhibitors. Toxicol. 125:13-19.
Moser, V.C., Chanda, S.M., Mortensen, S.R., and S. Padilla. Age- and gender-related differences in
sensitivity to chlorpyrifos in the rat reflect developmental profiles of esterase activities.
Toxicol. Sci., in press.
Augustinsson, K. -B. and Barr, M. (1963) Age variation in plasma arylesterase activity in children.
Clin. Chem. Acta. 8:569-573.
Ecobichon, D.J. and Stephens, D.S. (1973) Perinatal development of human blood esterases. Clin.
Pharmacol. Ther. 14:41-47.
58
-------
Screening Design for a Children's Pesticide Exposure Study
J.J. Quackenboss1, R.W. Whitmore2, P. Shubat3, C. Stroebel3, A. Kukowski3,
A. Clayton2, H.S. Zelon2, N.C.G. Freeman4 and E.D. Pellizzari2
U.S. E.P. A., National Exposure Research Laboratory, Human Exposure and Atmospheric
Sciences Division, Human Exposure Research Branch, P.O. Box 93478, Las Vegas, NV
89193-3478
Research Triangle Institute, P.O. Box 12194, Research Triangle Park, NC 27709-2194
3Minnesota Department of Health, P.O. Box 64975, St. Paul, Minnesota 55164-0975
4Environmental and Occupational Sciences Institute, P.O. Box 1179, Piscataway, NJ
08855-1179
Introduction
The Minnesota Children's Pesticide Exposure Study (MNCEPS) was conducted as a
demonstration/scoping project for a NHEXAS (National Human Exposure Assessment Survey)
"Phase-Ill" study, "to determine the causes of exposure for high risk groups, including those at
the high end of the exposure distribution and those who are more biologically susceptible (Sexton
et al., 1995a; 1995b)." This report describes the survey design and questionnaires, which were
developed to identify and screen households and individuals from the "high end" of the exposure
distribution for selected pesticides, and from a "susceptible" population sub-group (i.e., children).
Children have been identified as a susceptible population sub-group in terms of the potential for
exposure to environmental contaminants and the likelihood of adverse response to these exposures
(NRC, 1993). Behavioral patterns and diets may result in greater exposures to contaminants in the
environments where children live and play, and in the foods they consume. Small body size, and
developing organ systems and immune systems may put children at greater risk from these
exposures.
Households were selected based on the usage of pesticides reported in questionnaires, and
on the presence of products with target compounds identified through a product inventory.
Questionnaires are frequently used to obtain exposure-related information, and to classify
population groups based on their likelihood of exposures. These classification groups may be used
to stratify and select individuals for more detailed monitoring, as described in an accompanying
manuscript (Quackenboss et al., 1999), or for follow-up on health status or disease outcomes (for
epidemiological studies). In both cases, there is an interest in the ability of the questionnaire items
to distinguish more highly "exposed" individuals. Comparisons between the questionnaires used
in each screening phase, and with the measurements obtained in the follow-up exposure
monitoring phase, may provide important quantitative evidence regarding the utility of
questionnaire and household inventory data for this purpose.
Urban and suburban application of pesticides to lawns, golf courses, roadsides, public
buildings, homes, and apartment buildings may contribute to human population exposures, even
though they are a smaller proportion of total pesticide use than are agricultural applications. The
59
-------
1992 "National Home and Garden Pesticide Use Survey" collected data from 2078 randomly-
selected households in 29 states (Whitmore et al., 1993). This survey collected information on the
frequency and types of pesticide use in and around homes, both by household members and
professional applicators. Approximately 35% of households reported treating the primary living
area with insecticides at least once per year; 10% reported frequent insecticide use, on average,
more than once per month. In addition, nearly 20% of households reported using a pest control
service to treat homes for fleas, roaches, or ants during the past year. The Minnesota Department
of Health identified urban (structural) and suburban (turf) applications as being of concern due to
the numbers of complaints received (by the Minnesota Department of Agriculture), especially for
multi-unit and renter-occupied dwellings.
The follow-up exposure monitoring component of the study, described by Quackenboss
et al. (1999), evaluated the feasibility of making multi-media exposure measurements for a sample
of children.
Methods
The screening survey design is described below, including a description of the sample
population and questionnaires used. The hypothesis addressed here is that a screening design and
questionnaires are useful in identifying (stratifying) households and individuals (i.e., classification
groups) with higher exposures. The survey design assigned a higher probability of selection to
children considered "more likely" to be exposed to pesticides in and around their homes, based on
screening questionnaire data and a household inventory of pesticide products. The first phase of
the survey was to "identify" households with age-eligible children and with more frequent
pesticide use. The second phase was to "screen" these households for more detailed and specific
information about pesticide use and characteristics (in-home screening questionnaire and product
inventory). The third phase was to conduct "follow-up" monitoring on those more likely to be
"exposed," and a sample of other households.
Due to limitations in the staff available for both in-home screening and the follow-up
monitoring phases, samples of between 125 and 150 telephone numbers were selected for each of
17 "team-weeks" in Phase 1. The larger number of telephone numbers was selected for non-urban
areas, in order to provide a sufficient number of households on private wells. These were
contacted to identify and select about 18-20 households for MDH in-home screening
(questionnaire and product inventory), which was conducted over a 10-to-12 day period (Phase 2).
These results were then used to select six monitoring participants for each field-team and week
(Phase 3).
60
-------
Phase 1: IDENTIFY (telephone interview) households
• age-eligible children • frequent pesticide usage
(1,033) (668)
Phase 2: SCREEN (in-home by MDH)
• Questionnaire • Inventory of pesticide
(general pesticide products (specific
usage patterns) active ingredients)
n= 1,388
Increasing
Level of
Detail
n=102
PhaseS: FOLLOW-UP
EXPOSURE MONITORING
• exposure, biomarker, and environmental
concentration measurements
• time/activity diary, diet diary, activity patterns
• baseline and follow-up questionnaires
Decreasing
Sample
Sizes
Figure 1. Summary of Three-Phases in Survey Design collecting minimum level of detail on larger
number of subjects (Phase 1) and intensive monitoring (Phase 3) with sub-sample of subjects.
Phase 1: Identification of Households
A sample of 2,303 telephone numbers was selected from a list obtained from Genysis
Systems, Inc. (Fort Washington, PA) that was predicted to have age-eligible children (ages 3-12)
in the cities of Minneapolis and St. Paul, or in Rice or Goodhue counties in Minnesota. Eligibility
for participation was limited to children between the ages of 3 and 12 (inclusive) to maximize the
potential for collecting the first morning void urine samples, while still covering a portion of the
age ranges identified for children ("younger" ages 1-6; "older" 7-12) and pesticide exposures
(NRC, 1993). As a partial adjustment for raider-representation in the sample list, which is based
on households with listed telephone numbers, telephone numbers from lower socioeconomic (inner
city) Census Tracts in Minneapolis and St. Paul were sampled in proportion to population size in
the 1990 Census.
Telephone Screening. Given the frequency of households with children between the ages of 3 and
12, which was estimated using 1990 Census data to be less than 20%, a telephone-based interview
was conducted by RTI to identify potential study participants. The call was proceeded by a letter
which notified household residents of the call, and indicated that the study was being conducted
in cooperation with the Minnesota Department of Health. The telephone survey was conducted
using a Computer-Assisted Telephone Interview (CATI) approach. An adult member of the
household was interviewed to confirm eligibility (within city limits of Minneapolis or St. Paul; or
in Rice or Goodhue counties), and to obtain a roster of household members who were full-time
residents (i.e., those who lived in the residence year round). General information about the
frequency of pesticide use for indoor and outdoor insect control was also requested.
61
-------
Questions were based on the NOPES (Whitmore et al., 1994) and the Home and Garden
survey (Whitmore et al., 1993) questionnaires. These were adapted to identify specific types of
pest problems likely to be treated in Minnesota homes, and included applications by the
respondent, other household members, and pest control services. The following four "yes/no"
questions were used for over-sampling the frequent pesticide users.
(Q4) During the summer months, that is from June to September, do you or any member of your
household apply pesticides inside your home or apartment more than once to control insect
pests, such as fleas, ants, roaches, or silverfish?
(Q5) During the summer months, that is from June to September, do you or any member of your
household apply pesticides in the yard outside your home or apartment more than once to
control insect pests, such as fleas, ants, mites, aphids, or webworms?
(Q6) During the summer months, that is from June to September, do you use a pest control
service to treat the yard outside your home or apartment to control insects such as ants?
(Q7) During the past year, have you used a pest control service to treat the inside of your home
or apartment more than once for insect pests, such as fleas, ants, roaches, or silverfish?
A.11 households that responded "Yes" to either Q4 or Q7 were selected as "frequent" insecticide
users for Phase 2 (in-home screening); 50% of the remaining households were randomly sampled
for screening. Those with only one age-eligible child were sub-sampled (50%) to reduce the
survey design effect which would result from the unequal probability of selection relative to multi-
child households.
Selection of the Sample Frame. Children were over-sampled in this module because they are a
potentially sensitive sub-population for exposures to pesticides. In order to select a sample of
children for this study, four alternative sampling frames were considered:
1) an area household sampling frame (like the NHEXAS Region V field study);
2) a random-digit-dialing (RDD) telephone number frame;
3) all telephone numbers listed in the current telephone directories serving the target areas;
and
4) all households with telephone numbers listed in the current telephone directories
serving the target areas that are predicted by the vendor to contain children in the
target age range.
Costs for the first two approaches, area household frame and RDD, were increased by the large
proportion of households without age-eligible children and of non-residential phone numbers. The
third approach might provide better coverage of the target population, but at the cost of contacting
a large proportion of households without age-eligible children. The minimum survey cost, which
provided adequate representativeness to support the objectives for this study (i.e., to evaluate
feasibility and to make comparisons between exposures, environmental concentrations, and
questionnaires), was associated with option 4.
Thus, the sampling frame selected for the Children's Pesticide Study was the list of
households in Minneapolis and St. Paul (urban and sub-urban areas), or in Goodhue and Rice
counties (rural areas), which are predicted by Genesys Sampling Systems, Inc. to have age-eligible
children. These households are essentially the households listed in the current telephone
directories that are predicted to contain children aged 3 to 12 based on birth records and other
publicly-available data used in marketing research. We recognized, however, that the list was not
62
-------
complete. It excluded households not in the current telephone directories; 75% coverage was
estimated for Minneapolis and St. Paul. More importantly, it includes only that portion of
households with children in the target age range for whom marketing information is available on
the ages of household members. Since selection of households from a commercial list might bias
the income level upward, proportionately more urban households were selected from areas
identified as "inner city neighborhoods" based on census characteristics. Limited inferences
regarding the central tendency in exposures for the population covered by these listings are
possible, although these are limited by the sample size and unequal sampling weights (effective
sample size).
Phase 2: Household Screening
The second phase was for MDH staff to "screen" the households to identify those with
recent (or routine) pesticide applications. During this visit, a household screening questionnaire
was completed, which included
a roster of all household members, including age, gender, race, ethnicity, whether employed
outside the home, educational level, smoking status, and occupational exposures;
demographic information and housing characteristics, including household income, home
ownership, type of home, ground cover on the area around the house, and whether the property was
used as a farm;
insecticide usage inside and on the exterior/foundation of building, including who and where
applied, the frequency by user, and when used;
regular lawn or yard treatments (who, type, frequency, and last use); and
activity information for the selected child.
Non-leading, scripted probes were used to ensure that questions were answered as completely as
possible.
Pesticide use questions. These questions were based on the NHEXAS questionnaires, with some
items added or modified to meet specific objectives for the MDH and University of Minnesota
components of the study, and to identify specific insect pests and treatments that were likely to be
encountered in Minnesota. Some of the pesticide usage questions are listed below.
(S18) In the past 6 months, were any chemicals for the control of fleas, roaches, ants or other
insects used inside this (house/apartment)?
{If YES, the interviewer continued with SI9}
(S 19) What room(s) in your home were treated?
{the rooms were listed, with a Yes/No response recorded for each}
(S20) Which areas within the room(s) were treated?
{the areas were listed, with a Yes/No response recorded for each; if "Other" was selected,
the respondent was asked to specify a location}
Questions were used to record the frequency of applications inside this (house/apartment) during
the past six months by 1) a household resident, 2) a professional exterminator, or 3) someone else.
The last month that insect control products were applied indoors was recorded, together with
information about how the product was prepared for application, and who and where the product
was mixed. A similar series of questions was asked about "chemicals for the control of fleas,
roaches, ants or other insects" that were used during the past six months "on the exterior or
63
-------
foundation of this (house/apartment)." Regular treatments to the lawn or yard outside the
respondent's house or apartment were identified, together with information about who applied, the
type of application (wet, dry), the number of treatments for weed control and insect control, and
the month of last treatment.
Product Inventory. Respondents were asked "Do you have any pesticide products used to control
for insects or weeds in or around your home?" They were also shown a card listing different types
of pesticide products in order to obtain a complete inventory of all pesticide products (Figure 2).
Disinfectants were excluded from the inventory. Agricultural pesticides were also excluded unless
they had been used in or around the home. Interviewers recorded the name and EPA registration
numbers of each pesticide product. In a few instances, an EPA registration number could not be
located on the product. Interviewers noted the presence of specific, commonly used pesticide
ingredients as indicated in the list of active ingredients for each product. These ingredients were
those which had been selected for analysis in the subsequent sampling phase. Neither duplicate
products within a household nor product volume was recorded. As each product was inventoried,
interviewers asked whether that product had been used during the past year. Interviewers prompted
respondents to identify all areas inside and outside the home where pesticides might be stored.
After all products had been inventoried, respondents were asked to identify the last product used.
In order to limit the time required to complete the interview, detailed usage information was not
obtained for each product.
The EPA registration numbers were entered into a database to create an electronic list of
products by household. Based on the EPA registration number products which contained any of
the target pesticides (chlorpyrifos, diazinon, malathion, atrazine, and 2,4-D) were identified.
Products determined not to be pesticides were eliminated. Active ingredients of inventoried
products were identified using PEST-BANK software and an internet site maintained by the
California Department of Pesticide Regulation (DPR) of the California Environmental Protection
Agency, in affiliation with the EPA, Office of Pesticide Programs (CalEPA, 1997). For several
products that were not listed in the database, some information about active ingredients was
derived from the product name. Additional details regarding the inventory and analyses of these
data are described by Adgate et al. (1999).
Screening Score. A "screening score" was assigned and used to select a sample of about 10-to-12
households per monitoring team and week. This score assigned a higher ranking to households
having and using a target pesticide, than for other insecticides, or to having products containing
target or non-target pesticides stored in the home but not used in the past year (Figure 3). Another
factor included was occupational contact with pesticides by one or more family members. For
each team-week, about 11 homes were selected to complete the baseline study questionnaire and
for recruitment into the Phase 3 exposure monitoring. The top five scores were selected "with
certainty"and a simple random sample of six of the remaining households was selected from those
screened by MDH in that week. Six of those completing the baseline questionnaire, and who
agreed to participate in the monitoring phase, were assigned to a field-team for monitoring during
a specified one-week period (team-week).
64
-------
Types of Controls*
Insect sprays
Baits and traps
Insect repellents, lotions
Pet collars
Shampoos
Bombs and room foggers
Fly and insect strips
Mouse and rat bait
Lawn chemicals
Weed and other plant sprays
Flower and shrub insect or mold control
Vegetable garden insect or mold control
Slug controls
*Any product used to control pests, such as
weeds, insects, or rodents.
Types of Pests
Insects
Ants, spiders, mosquitoes, ticks, chiggers, fleas,
cockroaches, bees, hornets, wasps, moths, lice, flies,
soil-dwelling insects (nematodes), plant-chewing insects,
plant-sucking insects (aphids).
Microorganisms
Mildew, mold, wood decay or rot, plant diseases
Plants
Algae or moss, brush, grass-like weeds, broad-leaf weeds
Animals
Slugs and snails, mice and rats, birds, bats, otiier mammals
Figure 2. Show card used for MDH Product Inventory.
Target pesticide used in past year [I]
Indoor insect treatment in past 6 month [Q]
Outdoor insect treatment in past 6 months [Q]
Routine insecticide exposure at work [R]
Target pesticide stored (not used) [I]
Non-target pesticide used in past year [I]
Non-target pesticide, not used [I]
Score Assigned
Highest
I
I
I
I
Lowest
Figure 3. Screening score assigned based on Inventory [I], in-home screening
Questionniare [Q], or Roster [R].
65
-------
Phase 3: Follow-up Exposure Measurements
The third phase was to "follow-up" on those expected to have higher exposures (based on
their screening data) by monitoring of environmental, exposure, and biological media
concentrations for specific pesticide compounds for 102 households and children. Environmental
and personal exposure sample collection methods were selected to measure the extent of an
individual's exposure to specific pesticide compounds. Personal sample collection methods (air,
diet, dermal) were used to assess the total (or aggregate) exposure to selected pesticides and
polyaromatic hydrocarbons (PAHs). Selection of pesticides was based on information about both
the ranges of likely population exposures and on the hazard of the chemicals. A set of "primary"
pesticides were used for development of the survey design, to define quality assurance (QA) goals
for selection of sampling and analytical methods, and were collected for most sample types and
media (Table 1). Secondary pesticides were reported from the analyses of selected samples.
Samples of vacuum dust were collected as part of special study for the MDH.
Environmental sample collection methods were selected to provide information about the
source of the chemicals and the exposures and the relative importance of the media and location
to total exposure, dose, and risk. Some of the samples were collected by the participants (dietary,
urine), while the remainder were collected or setup by field technicians (personal, indoor, and
outdoor air monitors; tap water, surface wipe and press, dermal rinse, soil). The types of samples
collected included:
Air — Indoor, Outdoor (10% of urban homes), Personal (6-day-integrated);
Surface Wipes and Press ~ at two indoor locations (main play area; inside family room);
Water Sample ~ Tap during the week (10% of urban homes);
Foods and beverages— Duplicate diet, 4 day composite;
Activity Observations (video-tape) — four hours (approximately 20 homes);
Dermal Rinse for adhesion — one day;
Urine ~ first morning void on days 3,5, and 7; and
Baseline and follow-up questionnaires, and activity diaries.
This combination of environmental, exposure, human activity, and biological
measurements was selected to evaluate the relative contribution of multiple media, pathways, and
routes to "aggregate" (or total) exposure and to body burden measurements. This includes an
evaluation of the importance of understanding detailed activity patterns (e.g., surface contact and
mouthing activities) in children in interpreting the relationship between environmental media
concentrations and exposure. More detailed descriptions of the methods used are provided in
Quackenboss et al. (1999).
Sample Size. The target study sample sizes were as follows: 72 for Minneapolis and St. Paul (36
in each city, and in each of the urban and suburban domains within those cities combined), and 30
in the rural area (Goodhue and Rice counties). The survey design effect due to unequal
probabilities of selection reduces the effective sample size for population inferences relative to a
simple random sample of population members. For the MNCPES, the survey design effect was
expected to be approximately 1.31 for most statistics. The factors contributing to this design effect
were: oversampling for MDH monitoring of households that appear to be regular pesticide users
based on the telephone screening data (deff, = 1.11); selection of one age-eligible child at random
66
-------
from the age-eligible children within sample households (deff2 = 1.09; reduced from about 1.30
selecting only half the households with one age-eligible child); and oversampling children who
appear to be most likely to have been exposed based on the MDH screener and inventory of
household pesticide products (deffj = 1.08).
Therefore, the effective sample sizes in these areas were expected to be the following: 55 in
Minneapolis and St. Paul; 27 in each city and in each of the urban and suburban domains; and 23
in the rural area. The standard errors of population proportions (e.g., proportion of children age
3 to 12 in the target area with measurable exposure to chlorpyrifos) are adequate to support
accurate estimates when the true proportion is larger than about 10 percent for Minneapolis and
St. Paul or greater than about 20 percent for the other, smaller domains. Several objectives for
the MNCPES involved establishing relationships or correlations between study observations (e.g.,
pesticide exposures and concentrations of pesticide metabolites in the urine). These sample sizes
provide sufficient power to detect correlations (i.e., significantly different from zero) of about 0.50
or greater for the urban, suburban, and rural analysis domains for which the effective sample size
will be about 23 to 27, as discussed above. For the larger sample available for the entire
Minneapolis and St. Paul area, correlations of about 0.40 or greater will be significantly different
from zero. Thus, these sample sizes were considered adequate because inferences are limited to
making comparisons (i.e., among environmental, exposure, and biomarker concentrations),
evaluating exposure assessment models, and summarizing central tendency in children's pesticide
exposures.
Table 1. Identification of target pesticides for the Minnesota Children's Pesticide
exposure Study.
Emphasis
Primary
Secondary
In Vacuum Dust only
(Special study for MDH)
Pesticides
Atrazine
Chlorpyrifos
Alachlor
Dichlorvos
Dieldrin
/raws-Chlordane
czs-Clordane
4,4'-DDE
4,4'-DDD
4,4'-DDT
2,4-D
Dinoseb
MCPP
MCPA
Diazinon
Malathion
Endosulfan 1
Heptachlor
Metolachlor
cz'5-Permethrin
Jra«s-Permethrin
Pyrethrums
Simazine
Methyl Parathion
Silvex
2,4,5-T
67
-------
RESULTS
The Children's Pesticide exposure Study was conducted during the summer of 1997. The
"identification" phase started in May, 1997; the "follow-up" field monitoring was completed by
the end of September. This report summarizes the results of the Phase 1 telephone "Identification"
survey, and of the Phase 2 in-home "Screening" using the household questionnaire and pesticide
product inventory. Results from the Phase 3 sampling of the Minnesota Children's Pesticide Study
are not yet complete and are not included. These data provide a picture of pesticide storage and
use patterns in the surveyed homes.
Survey
Identification phase. Telephone interviews were completed for 1,388 households (67% of those
determined as eligible). Of these, 1,030 had at least one age-eligible child and were asked to
complete the pesticide related questions. The overall proportion of those identified as "frequent
users," based on "yes" responses to questions Q4 and Q7, was 27.8%. This rate was slightly
higher for inner city Census Tracts (33.3% and 31.8% in Minneapolis and St. Paul, respectively)
than in the other urban areas (28.8% and 27.8%). The rate was higher in Rice county (30.3%) than
in Goodhue county (17.6%), the two rural areas. Households responding to the telephone survey
were more likely to use pesticides inside the home than in the yard, and were far more likely to
apply pesticides themselves than to hire a professional applicator. In order to improve the
likelihood of obtaining measurable insecticide concentrations in the monitoring phase, 602
households were selected based on those reporting frequent pesticide use and a 50% sample of the
others. Those households with only one age-eligible child (32.8%) were sub-sampled to reduce
the variance inflation (survey design effect) that results from randomly selecting one child from
all age-eligible children in a household, resulting in a sample of477 homes. Of these, 348 homes
(73%) agreed to set an appointment for a "screening" visit by MDH staff.
Household Screening phase. The MDH was able to complete in-home screening in 294 homes
(88% of 335 attempted) within the time required (10 days) to use this data to select households into
the follow-up (monitoring) portion of the study; an additional 14 homes were competed after this
time. Of these 308 households, 225 (73%) were in the urban area (Minneapolis and St. Paul) and
83 (27%) were located in Goodhue and Rice Counties. Although most homes in Goodhue and
Rice Counties were in rural areas, some were within population centers, such as Red Wing and
Faribault, Minnesota. Thirty-four percent of households in Goodhue and Rice Counties indicated
that their homes were located on "working farms."
The "screening scores," based on results of the telephone and household
questionnaires/inventory, were used to select 183 households for follow-up. Baseline (NHEXAS)
questionnaires were completed for 174 (96%) of these cases; 159 (91%) were available for
monitoring during their selected sample-week. Of these, 109 households were selected to set
monitoring appointments, and 102 households (94%) participated in follow-up visits ("core"
components).
Follow-up Monitoring phase. Participation rates in the optional components of the monitoring
phase were very good. Of the 102 participants, 75 participated in the personal air monitoring and
68
-------
89 provided urine samples. In addition, 88 children provided hair samples and 61 provided blood
samples (archived). The overall response rate, calculated as the product of responses to each item
identified above, was 37% (Quackenboss et al, 1999), which is comparable with other human
exposure surveys (Callahan et al., 1995). The largest contributions to losses were in the
"identification" phase, with 47% response rate. One factor may have been the use of a telephone
call as the initial contact, since telephone surveys traditionally have lower response rates than
personal contact in a door-to-door survey, which was not feasible to apply in this study. Another
factor which may have contributed to this rate was the limited time available (10 days) to complete
the telephone interviews for a "cohort" of households, so that the results could be transferred to
MDH for them to begin the "screening" phase. Close coordination of between these efforts was
needed to ensure that an adequate number of households was available at the same time that the
monitoring teams were ready to begin the next cohort.
Once the "identification" phase was completed, participation rates in the household
"screening" and "follow-up" monitoring phases Were excellent, with response rates of 88% and
82.1%, respectively. In part, this might be due to the active participation and support of the state
health department (MDH). An additional factor may have been the high level of interest among
participating families in a study that addressed an issue of concern to parents—the potential
exposure of their children to pesticides in foods and from other sources.
Household Screening Questionnaire
Frequency and Locations of Applications. Sixty-nine percent of all homes indicated some
pesticide use during the six months preceding the survey. Pesticide products were used inside the
house more frequently than they were used outside: 52% of all respondents indicated that they had
used pesticide products for insect control inside their homes within the past 6 months: 21% had
insect control treatments to the foundation or exterior of the house; and 38% had regular treatments
to the lawn or yard (Table 2). Of the 115 households with regular lawn treatments, 60 (52%)
reported one treatment and 35 (30%) reported two or more treatments for weed control in the past
six months. In contrast, only 8 households (7%) reported lawn or yard treatments with
insecticides. No difference was identified between urban and rural households with respect to lawn
treatments. Nearly a quarter of all households used pesticides in flower or vegetable gardens or
on fruit trees.
Within the home, the kitchen was the room most likely to have been treated:
approximately 80% of households that had used pesticides inside the home, and nearly 42% of
all homes, had applied pesticides in the kitchen. The floor and baseboard were the most common
sites of application within rooms: nearly two-thirds of households that had used pesticides inside
the home, and just over one-third of all households had applied pesticides to the floor. Of
households applying pesticides inside the home, 11 % had applied pesticides to cupboards in which
food was stored and/or to cupboards in which dishes, pots, and pans were stored, and 24% had
applied the products to counter tops. Since this latter response was not an alternative offered by
the interviewer, but had to be elicited in response to a general query about "other" areas, the actual
percentage may have been even higher. The combination of greater frequency of use in the
kitchen, and self-reported use on counter-tops (associated with the kitchen as the room treated)
may be a significant finding in terms of the relationship between patterns of insecticide use and
69
-------
human exposures, especially in terms of the potential for contamination of foods during storage
and preparation in the home and the contribution of this to dietary exposures.
Professional applications. Few households had employed professional pesticide applicators for
pest problems associated with their house itself. Within the past 6 months, professional
exterminators had applied pesticides inside less than 2% of all homes and on the exterior or
foundation of less than 3% of all households. Employment of a professional to apply pesticides
to the lawn or yard was somewhat more common. Over 7% of all households, and just over 24%
of households that had treated the lawn and yard with pesticides, indicated that application had
been performed by a professional.
Table 2. Rooms and Surface Areas Treated for Insect Control in the Past Six Months for all
households and for those reporting indoor insecticide use ("users" = 52.3% of total).
Rooms Treated
1 . Living Room
2. Family Room
3. Dining Room
4. Kitchen
5. Bathroom(s)
6. Bedroom(s)
7. Basement
8. Other
Room(s)
# (%) %
total users
41 (13.3) 25.5
24 ( 7.8) 14.9
30 (9.7) 18.6
129 (41.9) 80.1
42 (13.6) 26.1
34 (11.0) 21.1
52 (16.9) 32.3
30 (9.7) 18.6
Don't Know
2 ( 0.6)
Surface Areas Treated
1 . Floors
2. Baseboards
3. Lower half walls
4. Upper half walls
5. Ceilings
6. Cupboards with
dishes, pots, pans
7. Cupboards with food
8. Storage cabinets
9. Closets
10. Window sills
1 1 . Other Areas
# (%) %
total users
109 (35.4) 67.7
58 (18.8) 36.0
16 ( 5.2) 9.9
14 (4.5) 8.7
13 (4.2) 8.1
18 (5.8) 11.2
18 (5.8) 11.2
24 (7.8) 14.9
15 (4.9) 9.3
34 (11.0) 21.1
66 (21.4) 41.0
[total=308 respondents] [users=161 households; 52.3%]
Pets. Because of the potential for frequent, close contact between pets and children, use of
pesticide products on pets (e.g., to control fleas and ticks) is of particular concern to those
investigating children's exposure to pesticides. Over 71% of households had pets such as dogs,
cats, gerbils, hamsters, rabbits, guinea pigs, birds, or horses. Respondents in nearly one quarter
of these homes (or approximately 17% of all homes) gave an affirmative response to the question
70
-------
"Are any chemicals or collars used on any of these pets to control fleas or ticks?" Rural
households were more likely than urban households both to have pets (80% compared to 68%),
and to use pesticide products on pets (24% compared to 14%).
Inventory of Pesticide Products
Pesticide products were found in all but 9 households (97.1%). A total of 2,058 pesticides
products were inventoried. Respondents indicated that 1,083 products, or slightly over half of
those inventoried, had been used within the past year. Two hundred seventy-eight households
(90.3%) reported having used pesticide products at least once during the past year. The mean
number of products inventoried per household was 6.7(±5.4); the mean number of products used
was 3.5 (±2.8). No more than 26 products were inventoried in any other household. The
maximum number of inventoried products used in any household during the past year was 16.
While, in the screening questionnaire, 69% of households reported using pesticides within the past
six months. 90% of households reported use of specific inventoried pesticide products within the
past year. The 21 % difference may reflect, in part, the timing of the survey. For those households
surveyed early in the process, the preceding six months would have been approximately mid-
November through mid-May; the six month period applicable to households surveyed later in the
process would have encompassed more of the summer season in Minnesota.
Active ingredients. Nearly 170 active ingredients were identified in inventoried products. DEET
(diethyl-meta-toluamide and other isomers), an ingredient of many insect repellents, was the most
common ingredient both found (64% of households) and used (53% of households). Products
containing piperonyl butoxide, a synergist, were inventoried in 49% of homes and used in 30%
of homes surveyed. Pyrethrins, which are derived from plants, were found and used in 48% and
29% of homes surveyed, respectively. Of homes surveyed, 30% had products containing
permethrin, a synthetic pyrethroid compound, and 21 % had used a product containing permethrin
within the past year. Chlorpyrifos, an organophosphate insecticide, was found in 27% of
households and used by 18% of households. The last of the most frequently used pesticide
ingredients listed above, MCPP, DMA salt, is a chlorophenoxy herbicide. MCPP, DMA salt was
found in 35% of surveyed homes and used in 18% of surveyed homes.
Types of pesticides. Approximately 90% of all pesticide products found and used were
insecticides, insect repellents, or herbicides. Insecticides alone accounted for nearly 50% of all
products found or used. Together, insecticides and insect repellents accounted for more than 75%
of product use. Use of pesticides to control for non-insect pests (e.g., weeds or rodents) was far
less common. The graph below shows the breakdown of major types of products used (Figure 4).
Types of products included in the category "Other" are acaricides, growth regulators,
molluscicides, and vertebrate repellents. Products that could not be associated with any type of
pest are also included in this category.
71
-------
OTHER (1.48%)
FUNGICIDE (1.75%>
RODENTICIDE(3.42%
HERBICIDE (15.24%>
-INSECTICIDE (49.86%)
^;i'V^5^~fV'-5f:ft-4--;gJ >0
INSECT REPELLENT (28.25%)-
Figure 4. Types of Pesticide Products Used in the Past Year.
Comparisons
Between Questionnaires. Comparisons were first made between the telephone- ("identification")
and household- ("screening") based classification groups for pesticide usage: "frequent users,"
defined from the telephone survey as those responding "yes" to either Q4 or Q7, and "others." The
proportions of households reporting insecticide usage on the household screening questionnaires
(MDH) are shown in Table 3 for two groups, as defined by their telephone survey responses. More
than 75% of the 139 "frequent users," who indicated pesticide use during the summer months
(June-September) on the telephone survey, also reported using products for insect control indoors
within the past six months on the household screener (MDH). This compares with less than 32%
of the 162 "others." The corresponding rates for using insecticides in the kitchen were 61% and
25%, for the "frequent users" and "others." Although this classification was only based on indoor
use, reported pesticide usage by the "frequent" group was also slightly greater for exterior insect
control (28% vs. 16%), regular lawn or yard treatments (44% vs. 32%), and for application of
chemicals to control weeds or insects in a flower, vegetable, or fruit garden (27% vs. 19%). The
identification of specific insecticides (active ingredients) in the household product inventory is
compared with the telephone-based classification in Table 4. There was an increase in chlorpyrifos
use among the "frequent users" households (20% vs. 12%), although the proportion of households
having chlorpyrifos-containing products was similar (30% vs. 24%).
The use of pesticides for indoor insect control during the past six months, reported in the
household questionnaire, was compared with the total number of products found and reported to
be used during the past year in the product inventory. Distributions for the number of products
found and used in each home were skewed to the right and there were several outliers. Thus, non-
parametric analyses (Wilcoxon rank-sum test and Median test) were used to compare these
classification groups (Table 5). The number of products present was only slightly higher for those
reporting indoor insecticide use. This difference was greater, and consistently significant using
both tests, when use of these products during the past year was considered. There was
considerable overlap between these groups, which might be reduced when considering specific
types of pesticides and their likely uses. Applications to indoors, exterior/foundation, lawn/yard,
and garden were combined to form a general index of pesticide use for comparison with the
72
-------
product inventory. The number of products found and used were both higher for the combined
pesticide use group (Table 6).
With other studies. Two national surveys of household pesticide storage have been conducted (US
EPA, 1980; RTI, 1992). In addition, a number of regional studies have included pesticide
inventories and there have been local surveys that reflect on pesticide use in the Minneapolis and
St. Paul areas (Kamble, 1982; US EPA, 1990; Davis, 1992; Bradman, 1997). Although some
results of this study can be compared to the results of the earlier studies, caution should be
exercised in drawing any conclusions, given the different geographical areas, pests, and climates;
the fact that some of these studies are now nearly 20 years old; and the differences between the
underlying purposes and resultant methodologies of the studies.
The mean number of pesticide products inventoried in this study (6.7) is considerably higher
than means estimated for the 1976-1977 National Household Pesticide Usage Study (1.7) and
found in the National Home and Garden Pesticide Usage Study (3.8), conducted in 1988-1989
(USEPA, 1980; RTI, 1992). However, in neither of these earlier studies were frequent pesticide
users over-sampled as they were in the current study. Part of the study population in the Non-
occupational Pesticide Exposure Study was chosen to represent higher pesticide use; in this study,
the mean number of pesticides products for frequent pesticide users was 5.3 (USEPA, 1990).
However, even within this group of frequent pesticide users, the maximum number of products in
any home was 23, substantially lower than the maximum number of 45 found in this study.
Notwithstanding the differences in sample selection, with one exception, all national and regional
studies, including this study, reported that approximately 90% of participating households used
pesticide products.
Generally, a comparison of the most commonly found active ingredients in these studies shows
increasing use of pyrethins and pyrethroids. This is consistent with a trend toward using pesticides
that are considered less hazardous and reflects the fact that pyrethroids entered the market only as
recently as 1980. Other noticeable differences can be explained by the banning of certain popular
ingredients.
DISCUSSION
The simple set of yes/no questions used for the telephone "screening" phase were useful to
identify or classify households by general patterns of pesticide use, when compared to the in-home
screening questions and the product inventory, given differences in time frame and level of detail.
The household screening phase indicated that health department interviewers could
successfully inventory and identify pesticide products which were stored in the home. Nearly all
households in the survey (97%) had pesticide products. Approximately 90% of households
reported having used pesticides during the year preceding the survey. Although households
determined to be "frequent pesticide users" were over-sampled, this percentage is similar to that
found in other studies. The mean number of pesticide products inventoried per household was
6.7. This is higher than the number of products reported in other studies. This may reflect
geographic variations or temporal shifts in pesticide usage patterns, or may be an artifact of
different survey objectives and strategies. Approximately 50% of all pesticide products
inventoried and used were insecticides. Insect repellents comprised approximately 23% of
73
-------
products inventoried and 28% of products used. Herbicides constituted the third most common
type of pesticide product, accounting for approximately 18% of products inventoried and 15% of
products used. Insect repellents were more likely to be used than either insecticides or herbicides.
Just over half of respondents reported that insecticide products had been used inside their
homes within the past six months. The room in which insecticides were most often applied was
the kitchen. In nearly 42% of the households in which the kitchen had been treated, at least one
of the areas treated was the counter tops. Overall, floors and baseboards were the most frequently
treated surface areas within rooms. The most frequently reported "other" surface area treated was
the counter-top (associated with kitchen as the room treated).
Demographic analysis indicates that selection of the sampling frame, which was based on
listed telephone numbers and commercial marketing data, may introduce some bias into the results
of this survey. Despite efforts to increase the proportion of lower income participants by over-
sampling inner city areas, the median household income of the study population (approximately
$56,250), was high relative to a two year moving average of the median income in Minnesota
($40,022). Therefore, results may not accurately reflect pesticide practices in lower income
groups. In addition, the homes in this study were almost exclusively (95%) single family
unattached homes. Therefore, no conclusions can be made regarding pesticide storage or use in
other housing types, such as apartments, town homes, or mobile homes. A large proportion of
these homes were owner-occupied (94%). Both the high income level and the high proportion of
single family homes in this study may have resulted from the use of a commercial telephone list.
Higher income families may have been more likely to have engaged in activities that would result
in their inclusion on the initial telephone list. Restriction of selection to a phone list also excluded
families without telephones.
Another aspect of selection bias is self-selection. Selection was a multi-step process and
attrition occurred at each step. The response rate to the telephone survey was low, due in part to
calls that were not answered, or were answered only by a machine. No information regarding these
households is available, precluding any meaningful evaluation of the potential non-response bias.
Overall, the completion rate for the telephone survey was low, less than 70%. Response rates for
the in-home interviews and inventories were better; interviews and inventories were completed for
308 of 335 (92%) households visited. Higher response rates may have been due to having the local
health department (MDH) contact, or to elimination of "nonresponders" at an earlier stage.
Several issues relating to the sampling design affect the extent to which conclusions derived
from this survey can be generalized beyond the survey population. Although the sample was
population based, certain households were assigned a higher probability of selection in order to
ensure that particular groups, e.g. frequent pesticide users, were represented in the final sample.
In order to generalize the survey results beyond the survey population, data must be weighted (by
the inverse of a household's probability of selection) to offset this unequal probability of selection.
Weighted results will be available in later publications based on data from this study.
74
-------
Table 3. Comparison of Telephone "Identification" and MDH Household "Screening'
Questionnaires for Reported Pesticide Use*
Household Screening
Location Treated
(in Past 6 Months)
Inside (insect control)
-> in the Kitchen
Exterior (insect control)
Regular Lawn or Yard
Garden (insect control)
Number4" in each "Group"
Total
(%)
51.8
41.5
21.6
37.5
22.6
301
Telephone Identification
User Group
"Indoor Users"
in Summer
(%)
75.5
61.2
28.1
43.9
26.2
139
"Others"
(%)
31.5
24.7
16.1
32.1
19.1
162
* % of households from each Telephone Identication "User Group" who also reported using
pesticides in the MDH Household Screening Questionnaire
+ total=301 (missing 7 cases for telephone or questionnaire records)
Table 4. Comparison of Telephone "Identification" and MDH Household "Screening" for
Insecticides Present and Used In Past Year*
Household "Screening"
Insecticide on
Product
Inventory
Chlorpyrifos
Diazinon
Malathion
Product
Present
Used
Present
Used
Present
Used
Number in each "Group"+
Total
(%)
26.9
16.0
20.9
12.3
9.0
4.0
301
Telephone "Identification"
User Group
"Indoor Users"
in the Summer
(%)
30.2
20.1
21.6
12.2
10.8
5.0
162
"Others"
(%)
24.1
12.4
20.4
12.4
7.4
3.1
139
*% of households from each Telephone Identication "Group" with target pesticides identified on
MDH Screening Product Inventory and their use in the past year
+ total=301 (missing 7 cases for telephone or inventory records)
75
-------
Table 5. Comparison of the product inventory with the household screening questionnaire:
number of pesticides found and used in each home, by reported indoor insecticide use.
Product Inventory
# Products Found
Mean
Median
25th - 75th percentiles
# Products Used
in past year
Mean
Median
25th - 75th percentiles
Number of Homes
Indoor Insecticides
Used in Past 6 months
No
6.2
5
3-8
3.0
2
1-4
145
Yes
7.2
5
4- 10
4.0
3
2-5
161
Wilcoxon
rank-sum
Median test
p<.05
n.s.
p<.01
p<.01
Table 6. Comparison of the product inventory with the household screening questionnaire:
number of pesticides found and used in each home, by combined pesticide use
(indoors, exterior/foundation, lawn/yard, or garden)
Product Inventory
# Products Found
Mean
Median
25th - 75th percentiles
# Products Used
in past year
Mean
Median
25th - 75th percentiles
Number of Homes
Combined
Pesticide Use
No
5.2
3
2-7
2.2
1
1-3
89
Yes
7.3
6
4-10
4.0
3
2-5
219
Wilcoxon
rank-sum
Median test
p<.01
p<.01
p<.01
p<.01
The need to restrict the interview and inventory to a reasonable amount of time limited
questions regarding use of specific pesticides. For example, the only product specific information
collected was whether a product was present and whether it was used during the past year. No
76
-------
information was collected as to where each product was used, how frequently it was used, and
what quantity was used.
These data provide a rare, inside view of household pesticide storage and use practices in a
sample of Minnesota homes. From a public health perspective, some of the data are reassuring:
the most common active ingredients are considered to be among those pesticide ingredients less
hazardous to health. While the percentage of households that used pesticides was similar to results
in other studies, the numbers of products stored and used were much higher. This leads to
questions about the rate of application. Does the larger average number of products reflect a
higher rate of application or simply more varied applications? Common use in the kitchen leads
to concerns about the safety of such applications, which occur near food and food preparation
areas. This study appears to have under-represented lower income households and renters. Do
lower income households and/or renters have unique pesticide practices? This study did not
address other questions directly related to the potential for exposure. For example, the
questionnaire and inventory did not attempt to identify how products were used, patterns of use
(e.g., following label directions), or accessability of storage locations to children.
The Minnesota Children's Pesticide Study will make a significant contribution toward
quantifying (1) concentrations of pesticides in environmental media in and around the home and
(2) children's exposure levels for multiple pathways. The survey and monitoring results will assist
in a determining whether household pesticide practices are associated with body burden.
Ultimately, these studies will be important to evaluating policies affecting residential pesticide use
and children's exposures to pesticides.
REFERENCES
Adgate, J.L., Kukowski, A., Stroebel, C., Shubat, P.J., Morrell, S., Quackenboss, J.J., Whitmore, R.W.,
and Sexton, K. (1999). "Household Pesticide Storage and Use Patterns in Minnesota." Submitted
to J. Exp. Anal. Environ. Epidem.
Bradman M.A., Hamly M.E., Draper W., Seidel S., Teran S., Wakeman D., and Neutra, R. (1998).
Pesticide Exposure to Children from California's Central Valley: Results of a Pilot Study. Journal
of Exposure Analysis and Environmental Epidemiology. 7:217-234.
Callahan, M.A., Clickner, R., Whitmore, R.W., Kalton, G., Sexton, K. (1995). "Overview of important
design issues for a national human exposure assessment survey." J. Exp. Anal.Environ. Epidem.
5(3): 257-282.
California Environmental Protection Agency (CalEPA, 1997). Department of Pesticide Regulation.
"USEPA/OPP Pesticide Related Database Queries."
(AccessedOctober, 1997-March 1998).
Kamble ST., Gold R.E., and Parkhurst A.M. (1982). "Nebraska Residential Pesticide Use Survey (1979
and 1980)." Agricultural Experimental Station, University of Nebraska, Lincoln, Nebraska.
National Research Council (NRC), Committee on Pesticides in the Diets of Infants and Children (1993).
Pesticides in the Diets of Infants and Children. National Academy Press, Washington, DC.
PEST- BANK. SilverPlatter, (1997). The information in PEST-BANK is developed by the Center for
Environmental and Regulatory Information System from data supplied by the U.S. Environmental
Protection Agency and state pesticide regulatory agencies.
77
-------
Quackenboss. J.J, Pellizzan, E.D., Shubat, P., Whitmore, R.W., Adgate, J.L., Thomas, K.W., Freeman,
N.C.G.. Stroebel, C., Lioy, P.J., Clayton, C.A., and Sexton, K. (1999). "Design Strategy for a
Multrpathway Pesticide Exposure Study in Children." Submitted to J. Exp. Anal, and Environ.
Epidem.
Research Triangle Institute (RTI, 1992). National Home and Garden Pesticide Use Survey. Prepared for
U.S. Environmental Protection Agency. Report No. RTI/5100/17-03F.
Sexton, K., Kleffman, D.E., and Callahan, M.A. (1995a). "An Introduction to the National Human
Exposure Assessment Survey (NHEXAS) and Related Phase I Field Studies." J. Exp. Anal, and
Environ. Epidem., 5(3): 229-232.
Sexton, K., Callahan, M.A., Bryan, E.F., Saint, C.G., and Wood, W.P. (1995b). "Informed decisions about
protecting and promoting public health: rationale for a national human exposure assessment
survey." J. Exp. Anal. Environ. Epidem. 5(3): 233-256.
United States Environmental Protection Agency (USEPA, 1980). National Household Pesticide Usage
Study, 1976-1977. Final Report. EPA 540/9-80-002. Office of Pesticide Programs, Washington
D.C.
United States Environmental Protection Agency (USEPA, 1990). Nonoccupational Pesticide Exposure
Study, (NOPES), Final Report. EPA/6-3-9/3. Atmospheric Research and Exposure Assessment
Laboratory, Office of Research and Development, Research Triangle Park, N.C.
Whitmore, R.W., Kelly, I.E., Reading, P.L., Brandt, E., and Harris, T. (1993). "National Home and
Garden Pesticide Use Survey." In K.D. Racke and A.R. Lesslie, eds., Pesticides in Urban
Environments: Fate and Significance, ACS Symposium Series 522, American Chemical Society,
Washington, DC, pp. 18-36.
Whitmore, R.W., Immerman, F.W., Camann, D.E., Bond, A.E., Lewis, R.G., and Schaum, J.L. (1994).
"Non-occupational exposures to pesticides for residents of two U.S. cities." Arch. Environ.
Contam. Toxicol. 26: 47-59.
78
-------
Pesticide Usage Along the U.S.-Mexican Border*
Gerry Akland1 and Brian Schumacher2
'Research Triangle Institute
Research Triangle Park, NC 27708-2195
2U.S. Environmental Protection Agency
National Exposure Research Laboratory
Las Vegas, NV 98193-3478
Introduction
The passage of the North American Free Trade Agreement (NAFTA) and the accompanying
environmental side agreements commit the U.S. government and the U.S. Environmental
Protection Agency (EPA) to insure a safe environment as industrialization, trade, and population
growth occur along the U.S./Mexico border. Even before the passage of NAFTA, many
communities along the border were beset by infrastructure deficiencies, e.g., lack of public
drinking water, sewage system, and garbage disposal. These infrastructure deficiencies are a direct
result of the rapid growth along the border, especially over the past 15 years. During this period,
the population of the border region has doubled to more than six million people. Economic growth
has been accompanied by increased potential for water and air quality degradation. Residents of
these communities have strong concerns about their possible exposures to environmental
contaminants, including those which may be coming from across the border (transboundary
pollution), or from local sources, including traffic, refuse burning, and extensive pesticide use
throughout the agricultural areas.
Understanding and evaluating the nature of pesticide exposure to thepediatric population (i.e.,
multiple pathway/multiple pesticide exposure) and potential health implications are of national
interest and certainly relevant to border residents. As a result of discussions with the state agencies
and border communities, this project was identified as one of the priority areas for the
Environmental Health Workgroup of the Border XXI program. In a general way, this project is
the first step of a multi-phase program to assess the effects (if any) of multiple pathway, multiple
pesticide exposures on children's health.
*This paper is abridged from a report entitled "Pesticide Exposure and Health Effects in Young Children. Part
I: Pesticide Data" prepared by RTI for the U.S. Environmental Protection Agency. The complete report
(#EPA/600/R-99/008) is available from the National Technical Information Service, 5285 Port Royal Rd, Springfield,
VA 22161. It has been peer reviewed by the EPA and approved for publication. Mention of trade names does not
constitute endorsement or recommendation for use.
79
-------
The goal of this first phase of the program was to inventory and gather the data available in
each of the four border states, namely, Arizona, California, New Mexico, and Texas. Data were
obtained related to existing agricultural and pesticide usage practices in counties within 100
kilometers of the U.S./Mexico border. These data are summarized and examined to provide
descriptive analyses of simple comparisons (e.g., counties of highest pesticide usage, pesticides
used most heavily in these counties, and major pesticide usage by crop) among counties within
each state. Implementation of the project began with a meeting with officials in the State Health
Department, State Department of Agriculture, and other interested parties in each of the four states:
Arizona, California, New Mexico, and Texas. During the meetings, the project was introduced and
the need for pesticide and health information was discussed together with identifying the
appropriate state and local contacts who could provide the local, county, and state data. The type
of pesticide information of interest was discussed, including the common names and chemical
formulae of pesticides; when and where they were typically applied; and the frequency and typical
rates of application. Based on discussions with the state officials in California and Arizona, all of
the border counties within 100 km of the U.S./Mexico border were to be included in the study area.
These counties included Imperial and San Diego counties in California and Cochise, Pima, Santa
Cruz, and Yuma counties in Arizona. Only two counties in New Mexico (i.e., Dona Ana and Luna
counties) and four counties in Texas (i.e., Cameron, El Paso, Hildago, and Webb counties) were
to be included in this phase of the project because state officials felt that they would adequately
represent both the major agricultural areas within the state and the majority (-90%) of the
population living along the border.
The availability of pesticide information data varied by state depending upon the existence of
state-specific regulations which permitted collection of the information either from sales or actual
usage reports submitted by the licensed distributor or pesticide applicator. In Arizona and
California, state regulations require all commercial applicators to file reports indicating pesticide
usage by location, crop, year, and quantity applied in pounds of active ingredient. These records
are collected and maintained in computerized databases. Therefore, data presented in this report
for Arizona and California are from actual pesticide application records.
In contrast, Texas and New Mexico do not have state regulations which require pesticide data
submission to the state authorities, so actual pesticide usage data are not available. Thus, pesticide
usage in New Mexico and Texas was estimated using crop-specific pesticide usage information
obtained from the Arizona data and adjusting these estimates by the actual crop acreages grown
in New Mexico and Texas. Average pesticide usage rates, in pounds of active ingredient per acre
by crop category (e.g., small grains, vegetables, orchards, etc.), in Cochise, Pima, and Yuma
counties, Arizona, were calculated to provide an estimate of typical pesticide usage rates in the
U.S./Mexico border region. The number of acres, by crop category, in each border county for the
study area in New Mexico and Texas were obtained from both states Departments of Agriculture
for the years 1992-1995, where available. After discussions with local county agricultural
extension agents in New Mexico and Texas to determine similarities in agricultural practices and
pesticide usage between their state and Arizona, the average pesticide usage rates for the Dona Ana
and Luna counties in New Mexico and for El Paso and Webb counties in Texas were derived from
the average rates of pesticide usage in Cochise and Pima counties, Arizona. Yuma county,
Arizona, pesticide application rates were used to characterize pesticide usage for Hidalgo and
Cameron counties, Texas. To obtain the final estimated pesticide usage figures presented in this
report, the appropriate county' s average usage rate of active ingredients per acre by crop categories
was multiplied by the number of acres in that county.
80
-------
For each agriculturally active county within the four states bordering Mexico, a summary of
the acres of cropland, crops grown, the most heavily-used pesticides (in pounds of active
ingredient applied per year), and pesticide use on major crop types are presented for the study
period 1992-1995 in Arizona, New Mexico, and Texas while California data are presented for the
years 1991 through 1994.
The most agriculturally active counties (as indicated by the most acres harvested) are Imperial
County, California; Hidalgo and Cameron Counties, Texas; and Yuma County, Arizona,
respectively, for calendar year 1992. These 4 counties comprised 78.8% of the total harvested
acres associated with all 12 agriculturally active counties examined in this report. The most
commonly grown crops in these four counties are cotton, orchards (predominantly citrus), and
vegetables. Other major crops grown throughout the border region include hay and small grains.
In general, across all counties in the four border states, insecticides were the most frequently
applied pesticides and accounted for greater than 50% of all the pesticides used. Herbicides,
accounting for about 30% to 40% of all pesticides used, were the second most frequently applied
pesticides. The use of fungicides ranked third (about 15% or less of the total pesticides used) in
the U.S./Mexico border region and were generally associated with orchard and vegetable crops.
The least used (<5% of total applied pesticides) of the major classes of pesticides were the soil
fumigants (except in California) and defoliants.
The heaviest usage of pesticides on a per acre basis was associated with growing orchard
crops. Vegetable and cotton crops required lesser yet still substantial quantities of pesticides to
be applied when compared to "other crops" (hay, grain, etc.) which required the least amount of
pesticides. As a result of this general relationship, the counties where the greatest total pounds of
active pesticide ingredient were applied are Imperial and San Diego Counties, California; Hidalgo
County, Texas; Yuma County, Arizona; and Cameron County, Texas, in order of decreasing
pesticide usage rates. Selected tables from the complete report are presented below including the
most heavily used pesticides in Imperial County, California (Table 1), Yuma County, Arizona
(Table 2), and the most frequently sold pesticides in Cameron, Hildago, and Willacy Counties,
Texas (Table 3).
81
-------
Table 1. Most Heavily-Used Pesticides in Imperial County, California: Pounds of Active
Ingredient Aoolied bv Year.
Pesticide
Metam Sodium
Sulfur
Trifluralin
Malathion
EPIC
Methomyl
Dacthal
Chlorpyrifos
Dimethoate
Endosulfan
Linuron
Maneb
Methyl Bromide
TOTAL
Primary Use
Fungicide
Fungicide
Herbicide
Insecticide
Herbicide
Insecticide
Herbicide
Insecticide
Insecticide
Insecticide
Herbicide
Fungicide
Fumigant
1991
983422
2489597
303811
257352
259598
138544
220471
141170
144548
143117
X
X
X
5081630
(77.7)**
Pounds of A.I.
1992
1755879
3520508
132552
182039
186045
76394
340050
97062
X*
X
78985
X
137626
6507140
(85.7)
Applied by year
1993
1115498
2121163
149661
169643
143652
100638
157516
94151
X
X
159428
X
X
4211350
(68.2)
1994
1906805
1724271
217380
214289
157259
135844
129842
107145
79083
X
X
473622
199586
5345126
(81.3est.)
* X = less than lowest number shown in column.
** () = % of the sum of tabled entries in column to overall total pesticides used [14].
82
-------
Table 2. Yuma County: Most Heavily-Used Pesticides by Year in Pounds of Active Ingredient.
Year
Pesticide
Guthion
Lorsban
Orthene
Sulfur
Diazinon
Dimethoate
Lannate
Kerb
Phosdrin
Thiodan
Balan
Dacthal
Permethrin
Malathion
Treflan
Alliette
Carzol
Diuron
DiSyston
Ridomil
Maneb
Prowl
Prometryn
Prefar
EPTC
TOTAL
Primary Use
Insecticide
Insecticide
Insecticide
Fungicide
Insecticide
Insecticide
Insecticide
Herbicide
Insecticide
Insecticide
Insecticide
Herbicide
Insecticide
Insecticide
Herbicide
Fungicide
Insecticide
Herbicide
Insecticide
Fungicide
Fungicide
Herbicide
Herbicide
Herbicide
Herbicide
Insecticide
Herbicide
Fungicide
1992
111194.0
43808.2
23493.8
142158.4
10864
18644.3
108559.1
20990.5
27196.0
29215.0
19494.6
24120.8
14649.2
10341.6
25775.7
<
<
<
12442.8
<
<
<
<
<
<
542,948
310408
(57.2%)**
90381.6
(16.6%)
142158.4
(26.2%)
1993
<*
49228.5
23447.1
60029
12375
34891.1
110768.1
<
33200.0
59220.0
24010.8
29779.5
16171.9
32029.0
15585.6
64264.0
16663.5
27856.5
<
20721.7
76382.6
<
<
<
<
706624
387954
(54.9%)
97232.4
(13.8%)
221397.3
(31.3%)
1994
<
32553.5
36595.5
113073
<
12484.4
20056.5
<
30796.0
29771.0
13761.0
<
<
42808.0
20390.9
<
<
<
<
<
10400.8
10128.8
12232.0
<
<
385,052
205064.9
(53.2%)
56512.7
(14.7%)
123473.8
(32.1%)
1995
<
54753.0
44919.6
95522.0
19497.5
45931.9
139597.6
<
17560.0
54337.5
33856.2
22847.2
24551.8
37929.6
17163.2
93392.0
11929.6
29154.5
<
<
26230.1
<
<
25940.0
15388.0
810,501
451008.1
(55.6%)
144349.1
(17.8%)
215144.1
(26.5%_)_
* < = less than 500 pounds active ingredient applied.
**() = % of total used within pesticide type.
83
-------
Table 3. Pesticide Sales Data (1992) of the 20 Most Sold Pesticide Products from Cameron, Hidalgo, and
Willacy Counties, Texas*.
Formulation (active ingredient)
Temik 1 5G (aldicarb)
Lorsban 15G (chlorpyrifos)
Ridomil/Bravo 81 W SP (metalaxyl & chlorothalonil)
Furadan 15G (carbofuran)
Dacthal 75% WP (DCPA)
Terrachlor Super X (PCNB)
Iron Sulphate (Iron Sulphate)
Javelin WG (Bacillus thuriglensis var. kurstaki)
Bravo 90 DF (chlorothalonil)
Orthene 90S (acephate)
Atrazine 4L (atrazine)
Guthion 2L (azinphos-methyl)
Roundup (glyphosate)
Karmex DF (diuron)
Methyl Parathion (MP)
Treflan TR-10 (trifluralin)
DiSyston 15G (disulfoton)
Dropp 50 WP (thidiazuron)
Solicam (norflurazon)
Ridomil MZ 58 (mancozeb & metalaxyl)
Quantity Sold
(in Ibs or Ibs/gal)
184842
124890
116366
112950
95280
50000
47870
47689
48809
43280
40665
30975
28980
28104
26806
23900
22560
22218
22100
21756
Total
#A.I.*
27726.3
18733.5
94256.5
16942.5
71460.
5000.
47870.
**
43928.1
38952.
162660.
61950.
101430.
22483.2
107224.
2390.
3384.
11109.
17370.6
12618.5
819,618.2
- data compiled by Texas Department of Agriculture.
** - only chemically-based pesticides included in total pounds of A.I.
84
-------
Pesticide Use and Assessment along the U.S.-Mexico Border
Mary Kay O'Rourke, Ph.D.
Environmental and Occupational Health Unit
The University of Arizona, Tucson, AZ
Introduction
This paper addresses three topics related to pesticide use and assessment along the U.S.-
Mexico border. They are: (1) a brief synopsis of the projects underway at the University of
Arizona and the practical problems/issues encountered, (2) a summary of questionnaire
information gleaned about potential exposure of children to pesticides, and (3) a synopsis of
preliminary environmental pesticide data collected from homes in the NHEXAS Survey. The
information provided in my lecture was delivered to stimulate workshop exercises.
The presentation utilizes information collected by the Arizona NHEXAS Consortium under
the direction of Michael D. Lebowitz. The paper is written from the perspective of project
design, field collection and data assembly. None of the laboratory or modeling issues are
addressed here. Further, the presentation is based on analyses of incomplete data sets. Final
numbers will differ from those contained in this report, although the data probably represents the
dominant trend.
Current Projects Evaluating Pesticide Exposure
The University of Arizona is collaborating with Bartelle Memorial Institute and Illinois
Institute of Technology in two surveys: NHEXAS (The National Human Exposure Assessment
Survey) and a special "Border" Survey (Total Human Exposure in Arizona: A comparison of
border communities with the rest of the state). A third study has been undertaken by the
University of Arizona and the Western Arizona Health Education Center (WAHEC) to evaluate
the exposure of children to pesticides in Yuma County. Lay health workers from the community
(promotores) serve as interviewers, field technicians and health educators.
NHEXAS: The objectives of this field study are to determine the distributions of total human
exposures to multi-media pollutants in the classes of metals, pesticides and volatile organic
compounds
(VOCs) by studying a proportionate-based sample of the total population (with a nested design
for the different stages of sampling). Specific aims are: 1) document the occurrence, distribution
and determinants of total exposures in the general population; 2) characterize the 90th percentiles
Although the research described in this article has been funded wholly or in part by the United States
Environmental Protection Agency through Cooperative Agreement CR821560 and CR824719 to Michael D.
Lebowitz and CR825169 to Mary Kay O 'Rourke, it has not been subjected to Agency review and therefore does
not necessarily reflect the views of the Agency, and no official endorsement should be inferred.
85
-------
of total exposures to each of the pollutants; 3) evaluate the different media, personal-time and
activity factors that contribute to current total exposure; 4) evaluate biomarkers in blood and urine
for the target pollutants; 5) to perform evaluations of relationships between exposure reports,
environmental measurements, and biomarkers of the target pollutants; and 6) to predict total
exposures for individuals and populations including Hispanics. The proportional based
population sampling of households within blocks occurs in stage 1.
The target is 1200 such households, interviewed utilizing the NHEXAS questionnaires. In
stage 2, additional questionnaires will be completed and environmental sampling will take place
in about 450 households, representatively selected from the respondents. Environmental sampling
will include: metals in dust, soil, outside air and some tap water; pesticides in dust, soil, and some
tap water, and total VOCs in air. In stage 3, a representative selected subset of households will be
reevaluated for metals, pesticides and VOCs using methods with greater resolution and reliability.
Subjects in the households will be asked to complete questionnaires and provide biological
samples.
Total exposure models to pollutants sampled during the study (VOCs, metals, pesticides) will
be developed. These models will be associated with multi-media contact. Probabilistic exposure
models have been developed for the NHEXAS AZ study population and projected upward to
assess risk in the state.
Border Survey: There are concerns among border communities that exposures are high
relative to other parts of the country. These communities believe they encounter elevated exposure
related to their proximity with Mexico. Associated with increased exposure, is a community-wide
fear of increased health effects. Currently, there are no data available to validate this perception
of elevated exposure among the border communities. However, EPA has funded a project to
characterize the total exposure of residents in the state of Arizona (NHEXAS AZ). In NHEXAS
AZ, multiple media (air, soil, house dust, skin, food and beverages, water, blood and urine) will
be evaluated to determine contributions to exposure through various pathways (inhalation,
absorption, ingestion). We have been funded to conduct a special, complementary exposure study
along the Arizona-Mexico border, so for the first time, border exposures can be compared with
those from an adjacent non-border area (NHEXAS AZ).
Like NHEXAS AZ the "Border Study" will determine the distribution function of exposure
to selected metals, pesticides and volatile organic compounds (VOCs). Most of the target
contaminants will be the same for the two studies. However, we have added selected pesticides
(organochlorines) and polycyclic aromatic hydrocarbons (PAHs) to the analyte list since we
expect to find greater concentrations of these along the border.
Along the border, 300 households will be contacted with a minimum enrollment of 225 homes.
Preliminary work will be performed in all 225 homes. Detailed, intensive sampling will be
performed in a 100 home subset (of the 225 homes). This will provide enough homes to test
differences in the geometric means of contaminant concentrations between the Border and State.
These homes will be selected using the same population based probability research design as
NHEXAS AZ. All census divisions have already been randomized for the NHEXAS study. We
will select the next blocks in sequence. Thus the populations of the two studies will be
independent and non-overlapping. An exhaustive quality assurance plan (QSIP) complete with
standard operating procedures (SOPs) was developed for every aspect of the NHEXAS study, and
will be employed in the Border Study. (Additional SOPs will be added for PAHs.)
86
-------
In the Arizona-Mexico border study, exposure information will be gathered directly from
subjects, from environments frequented by subjects (primarily subject home environments) and
from public records. Questionnaires will be employed to characterize the study population,
evaluate common practices believed to contribute to exposures and evaluate potential bias in the
study due to non-participation. Blood and urine samples will be collected directly from subjects
and concentrations of target pollutants will be measured. Additional concentrations of target
pollutants will be measured from the air, dust, soil and water of home environments. Duplicate
diets (regardless of food and beverage source) will be collected. Public records containing usable
information on target pollutants (soil, air, and water) will be used where available. Exposure
assessment models will be generated using direct and surrogate measures varying in the intensity
of detail.
Total exposure models to pollutants sampled during the study (VOCs, metals, pesticides and
PAHs) will be developed. These models will be associated with multi-media contact.
Probabilistic exposure models developed for NHEXAS AZ will be applied to this proposed border
population study. These models will be fine-tuned to reflect the differences between the two study
populations as needed. The precision and accuracy of the previously developed models will be
tested with the independent data obtained from the border population. The objectives of these
models are to estimate the multi-media pollutant exposures to the subject and determine the
sources of inter-individual variability.
Children's Pesticide Survey: Yuma County is responsible for growing much of the nation's
fresh fruit and vegetable supply during the winter months. These crops are tended by seasonal and
migrant laborers who frequently live near the edge of fields with their families. Many of the
pesticides used are pyrethroids, some are dinitroanilines. Further, organophosphates (OPs) like
diazinon and chlorpyrifos are used in fields against pests, and in homes to combat termites and
roaches. As a result, children living in these homes are at great risk for routine exposure to
pesticides. Health effects in response to OP exposure have been observed. For instance, OPs
inhibit acetylcholinesterase causing the accumulation of acetylcholine. In turn, this affects the
central nervous system (sympathetic and parasympathetic) and elicits symptoms of sweating,
diarrhea and others. Poisoning is usually associated with occupational exposure associated with
agriculture. Mortality rates are high among the poisoned and are usually caused by respiratory
insufficiency. Some investigators report cardiac complications associated with OP poisoning.
The association with agriculture is particularly worrisome for the citizens of the Yuma area.
Many of them work in the orchards and vegetable fields and are concerned with associated health
risks. Investigators have documented different cholinesterase levels between farm workers (30.28
U/g hemoglobin) and others living in the community (32.3 U/g hemoglobin). Very low levels of
cholinesterase were found among farm workers who actually report being sprayed while in the
fields. Spraying of OPs is also a common practice for flower growers and OP poisoning has been
reported among some florists.
Since many of the fields in the Yuma area are sprayed from the air, residents with no
occupational exposure are also concerned. Housing adjacent to fields can be contaminated. In
other cases, people intentionally spray the inside of their homes to reduce pests. Misapplication
can result in heavy exposures and toxic responses particularly for children.
Ingestion is a major path for exposure to some OPs. Low income farm workers are frequently
given produce and fruit from the fields. They may not have a suitable running water supply to
87
-------
appropriately clean the produce. In some cases they may lack the knowledge of how to handle this
additional risk.
Toxic pesticide exposures are a community concern, but cancer is a wide-spread community
fear. Numerous studies have been conducted and suggest an association. One Canadian study
examined a population based tumor registry's records and found an association between lung
cancer tumors and farming in Saskatchewan; further brain cancer associations were reported by
researchers in Rome.
Although there are currently no data available to validate the perception of elevated exposure
in Yuma County, community-based organizations are actively educating the local population to
reduce exposure. One group, WAHEC, has been a pioneer in using lay-educators (promotores)
to educate farm-workers about exposure in the fields and possible secondary exposures to their
families.
We propose a study of 300 children living in the agricultural areas of Yuma County, formerly
served by the Valley Health Clinic, and recruited primarily through Head Start Programs. Most
participants will be low income Hispanics. The study will be undertaken with local participation
from WAHEC (Western Arizona Area Health Education Center) and employ promotores as
interviewers and field technicians. Screening samples (urine and house dust) will be collected
from parents at the Head Start Program or by apromotore (a) at the home. In 100-150 homes,
screening collections will be made. 30 homes will be selected for multi-media sampling to
evaluate OP concentrations in homes. All 300 children will be evaluated for urinary metabolites
indicative of pesticide exposure. We expect to sample households of the upper 20% for
pesticides. We will sample air, dust and other media as funding permits. We will seek separate
funding to evaluate other media (food and beverage). To model "total" exposure, we will
supplement these databases with regional information garnered while sampling for the NHEXAS
project.
We expect to find that children from low SES households have greater exposure than those of
the rest of the state as determined by the NHEXAS evaluation. We expect that children living in
homes where parents have knowledge of pesticide exposure mitigation will have lower exposures
than homes where little is known about pesticides. Even so, since more pesticides are used in the
Yuma area, we expect to find greater pesticide exposure in Yuma than elsewhere along the U.S.-
Mexico border of Arizona.
Community Requested Projects: Researchers write proposals and enter communities to
evaluate issues they consider to be priorities. Frequently, the communities have concerns not
addressed by these studies. Our group has become involved in several projects at the request of
the communities. For instance, in both Nogales and Douglas, Arizona, the communities are
interested in the potential for elevated asthma prevalence among children. We have undertaken
a study in each city in conjunction with the Arizona Department of Health Services. The Hispanic
community in Douglas is also interested in whether they are at elevated risk for diabetes. Dr.
Lebowitz is working with the Rural Health Office on this study. CDC and ADHS are pursuing an
evaluation of Lupus in Nogales at the request of residents, while Rural Health and the University
of Arizona Epidemiology Unit re-examine the original questionnaire results for validity. In
Sommerton, Gadsden and San Luis, substance abuse is an issue.
Description of Selected Field Implementation Issues
-------
Selection of Target Analytes: Literally thousands of pesticides are in use in the country.
Many are used for more than one insect and in more than one environment. Some are used in
agriculture and the same pesticide may be used safely for residential treatment. A good example
is chlorpyrifos. It is used on selected crops as an insecticide and used to treat foundations of
homes as a termiticide. The pesticide is not used inside the house, but under the slab. In some
cases, misapplication of agricultural pesticides inside residences can prove fatal (i.e., methyl
paraquat). Selection of pesticides for our studies was based on: (1) common usage as determined
by sales records and information from the state agricultural inspector, (2) potential exposure for
many people throughout the state, (3) established collection and analysis protocols for all media
and biomarkers, and (4) reasonable sample stability under storage and shipment conditions to
insure sample integrity. A great study design can be developed but if few people are exposed, or
sample collection, analysis and stability are questionable, then little can be done to understand the
nature and extent of exposure or associated health effect.
The Researcher's Commitment to the Community: Fortunately for the well-being of science,
most researchers believe in the proj ects they propose. They believe their proposed proj ect is in the
best interest of the community. Sometimes communities see these research projects as a benefit
for the community, but they often believe that only the researcher or the institution will benefit
from the project. Once a researcher is involved with a community, the commitment cannot be
taken lightly. Carpet-bag research is rarely welcomed by a community. Researchers should enter
the relationship with a long-term community commitment in mind, the commitment must be there
after the project ends.
Initial Contact: Prior to entering a community it is useful to release general project
information to the newspaper, radio and television. Notify local agencies that concerned residents
can call, including the public relations department of the police or sheriffs office and Border
Patrol. Many people will still miss the announcement, so mail to people living on the specific
blocks to be sampled is essential. Attempts to contact potential study participants should be made
during the day and evening and on weekend and weekdays. Special calling cards can be left
announcing the intent to contact and inviting the participant to call the office. When going door
to door, identity badges, a project vehicle and a project handout should be used. Provide business
cards and phone numbers to all contacted. Employ mixed gender, mixed age, mixed race and
mixed ethnicity field teams appropriate to the region being sampled. In the southwest each field
team should have at least one bilingual speaker at all times.
Culturally Appropriate Staff and Materials: Border communities present an interesting
challenge when performing survey work. Major issues include gender, language, culture and
degree of acculturation. Some issues are common: women are unlikely to answer the door to a
man if there is no man in the house. Many people speak Spanish; this does not mean they can
read Spanish as well or better than English. Many people speak Spanish and read English. There
may be a great diversity of language and reading skills within a single home crossing several
languages. Approach the home with a mixed field team and be prepared for any outcome. We
have had some cases where Anglos refused to participate because of efforts to include those who
do not speak or read English. Be prepared for illiteracy among all groups. Some study
participants cannot read or write in any language.
Field Truthing: From 1991 -1995, US Census data indicate that Arizona was the second fastest
growing state in the Union. As a result all selected block groups had to be evaluated for residential
89
-------
density prior to sampling. Numerous census block groups that were reported empty in 1990 had
residential development in 1995.
Logistics of Large-scale Environmental and Biological Sample Collection: Pesticides are
semi-volatile organic compounds that are particularly sensitive to volatilization by heat or
degradation by ultraviolet light. Availability of cold storage either in a refrigerator, freezer or ice
chest is an essential component of sample collection, transport and storage. This includes shipping
samples for analysis on "ice" by an overnight carrier. The need for cold storage and transport
influences the size of the field vehicle, number of samples collected and the distance and timing
of field trips.
Complexity of Methods: When working in a field environment use the simplest method
possible to accomplish the job. Select sturdy, lightweight, durable equipment that provides
consistent, reliable, precise and accurate measurements. Complex systems are more likely to
break, require extensive cleaning between operations or be inappropriately operated. Some
problems we have experienced: large vacuum systems that are not easily transported and require
cleaning of multiple parts, programmable pumps with digital displays that do not function at high
temperature and mechanical pump systems that fail to hold flow rate.
Students as Professionals: The purpose of research in a university setting is to provide learning
experiences for students. Advantages of student labor include a low cost, partially trained work
force that is reasonably intelligent. Disadvantages include limited flexibility of scheduling field
work, maturity level and work force turnover. We are constantly training new students to provide
the simplest skills.
Quality Assurance Issues: With a highly mobile work force it is essential to provide consistent
training, cross checks to unravel problems, and adequate supervision including an independent
quality assurance officer. An overall quality assurance plan should be implemented, all procedures
and changes should be documented and all documents should be available to project staff at all
times. The more complex the project, the more important the plans and procedures become.
Research Projects and Clinical Practice: Many procedures and treatments are available to the
physician treating a patient for a given disease, syndrome or symptom. Physicians routinely
prescribe medications and laboratory evaluations viewed as nearly risk-free. However,
procedures used on the general population for research purposes, are judged more critically. Low
risk procedures used routinely in the clinic may not be allowed in the field where there is no health
risk to the patient. Expectations of field implementation must be realistic relative to what the
researcher can get approved by an institutional review board.
Selected Preliminary Results from the NHEXAS Arizona Project
Several surveys are described above, but only the NHEXAS Survey is advanced to the point
where any results are currently available. These are all PRELIMINARY results. Not all results
are currently available.
Table 1 is derived from data in the NHEXAS Descriptive Questionnaire. The descriptive
questionnaire briefly assesses the demographic characteristics of a each enrolled household. Each
household has one subject designated as the primary respondent. This person is assigned an
90
-------
individual respondeni number (IRN) of 01. The IRN 01 is the index case for the study. These
percentages apply only to the recruited population.
Tables 2 and 3 examine pesticide use reported inside and outside the homes by responses to
questions on the NHEXAS Baseline Questionnaire. Table 2 examines pesticide use throughout
the state for all IRN Ols, pesticide use inside homes of those under the age of 18. Unfortunately,
only 4 children with IRNs of 01 were under the age of 6 so the percentages are not meaningful.
In general, there appears to be lower indoor use of pesticide among the border counties. The
reduction may reflect the higher elevation of Santa Cruz and Cochise Counties. The resulting
decrease in temperatures may result in fewer indoor pests, fewer lawn treatments or lower incomes.
Table 4 reports specific pesticides used in homes in the non-border counties of the State and
in the Border counties. Professional application of pesticides twice as great in the Border Counties
over the rest of Arizona and knowledge of specific pesticides applied is low; 3 times as many
unknown pesticides are applied in the Border region. Some of this lack of information flows from
professional application, but professional application alone cannot explain the difference. I will
speculate that proximity to pesticides available in Mexico may account for part of this unknown
component. Residents are aware of their exposure to pyrethroids, chlorpyrifos and diazinon.
Additional information about specific use of pesticide can be found in the NHEXAS Time-
Activity Questionnaire, and the Follow-up Questionnaire. IRN Ols applied pesticide 2% of the
days where information was sought (2560 person days) and 1% mixed the pesticide. During the
week we sampled 11.7% reported using or being near applied pesticide (315 person weeks); of
these .6% report using pesticides for more than .5 hours and appear to be professional applicators.
1.5% report using protective equipment and 9.2 % washed hands following application.
Environmental Data
With the possible exception of house dust, food and beverage, few samples contained
measurable amounts of pesticide. Tables 5 and 6 are provided ONLY as illustrations of potential
exposures. They illustrate the population exposure at the 50th and the 90th percentile of
distribution for each medium. Several assumptions are inherent in these numbers. First, the
example assumes that an individual has the 50th percentile (or 90th) of exposure for all media; a
very unlikely scenario. I have assumed a sedentary rate of inhalation at 101/min over the course
of the day with 85% of time spent indoors and 15% of time spent outdoors. I assume 1 kg of food
was consumed and 4 liters of beverage, 3 liters of which were water. Ingestion appears as the most
important pathway of exposure in terms of volume of pesticide for chlorpyrifos (Table 5), but not
so for diazinon (Table 6).
Summary
The NHEXAS approach will provide a wealth of information to determine the extent of public
exposure to target analytes and to evaluate the mechanisms of that exposure. Projects like this will
point the way toward exposure mitigation methods once pathways are identified. These projects
are also incredibly expensive to perform. The question becomes what proxy data can be used to
determine the high end of exposure and what focused study types can be applied?
There are a great many pesticides in use in our environment and we lack the methods to
evaluate many of them, Questionnaires indicate that about half the homes in Arizona use
91
-------
pesticides and 11.7% of the primary respondents report applying or working near applied
pesticide each week. Preliminary data suggest that homes with children appear to use less
pesticide, but this trend needs additional evaluation.
One approach is to employ screening techniques like urinary biomarkers as an indication of
exposure and then evaluate the high end of the exposure for multiple media. In NHEXAS, we
found few elevated levels within each medium but the biomarker for chlorpyrifos was found by
CDC in every urine sample submitted.
A second, or perhaps joint approach, is to evaluate a proxy media like house dust. Following
the meeting I performed a simple chi square evaluation of the relationship between house dust and
the urinary metabolite of chlorpyrifos. There appeared to be no significant relationship between
the concentration of chlorpyrifos in the house dust and the concentration of the biomarker in urine.
We are currently examining the value of the questionnaires in predicting urinary biomarkers.
We are in the process of applying the screening approach within the Children's Pesticide Project.
92
-------
Table 1. Baseline Questionnaire Arizona Subject Characteristics of State-wide IRN Ols by
Age Class. N represents the number of people in each category.
All IRN = 01
N=1015
Gender
Male
Female
Age
< 6 years
>6&<18years
>18&< 45 years
> 45 & < 65 years
> 65 years
Race
White
African American
Native American
Asian/Pacific Islander
Other
Ethnicity
Hispanic
Non-Hispanic
School Completed
None
Primary/Middle School
Some High School
High School Grad
Some College
College Grad
Post-Graduate
Smoker
Yes
No
Smoke Indoors
Yes
No
40.5%
59.5%
3.3%
13.4%
39.7%
27.7%
15.9%
93.6%
2.5%
2.5%
0.5%
1.5%
35.4%
63.7%
3.6%
20.2%
8.9%
20.6%
28.5%
10.0%
7.6%
18.2%
81.2%
12.2%
5.9%
IRN <18 years
N=151
53.3%
47.7%
13.9%
86.1%
n/a
n/a
n/a
88.1%
3.3%
5.3%
3.3%
38.4%
60.9%
19.9%
63.6%
13.9%
2.0%
0.7%
-
-
0.7%
99.3%
0.7%
n/a
IRN < 6 vears
N = 21
42.9%
57.1%
100%
n/a
n/a
n/a
n/a
81.0%
9.5%
4.8%
4.8%
28.6%
71.4%
95.2%
4.8%
-
-
-
-
-
0
100%
n/a
n/a
93
-------
Table 2. Percent Homes Using Pesticides as reported in the Baseline Questionnaire. Results
are reported in percentages for the age class of the IRN 01. N represents the number of homes
sampled in each area.
Baseline QX
Stage 2 & 3
All IRN 01
N = 359
Pesticides at Work/School
Work Pesticide= Raid
Work Pesticide= Repellent
Work Pesticide= Chlorpyrifos
Work Pesticide= Malathion
Work Pesticide= Diazinon
Work Pesticide= Carbaryl
Work Pesticide=Other
termat/pesticide
Work Pesticide= Atrazine
Work Pesticide= Other
Herbicide
Work Pesticide= Fungicide
Work Pesticide= Unknown
Pesticide
Work Pesticide=
Other — specified
Use Pesticides Indoors @ Home
Living Room
Family Room
Dining Room
Kitchen
Bathroom
Bedroom
Treatment Areas
Floors
Baseboards
Lower Half of Wall
Upper Half of Wall
Ceilings
Cupboards with dishes
Cupboards with Food
Storage Cabinets
Closets
Other
6.4%
0.6%
1.7%
0
0
0.6%
0.3%
0.6%
0
0.6%
0.3%
4.5%
1.2%
52.4%
37.0%
24.2%
32.6%
45.7%
42.3%
36.8%
22.0%
39.3%
7.0%
2.5%
2.2%
4.7%
4.2%
9.2%
12.0%
10.0%
Baseline QX
Stage 2 & 3
IRNOKlSyrs
N = 53 "
7.5%
0
0
0
0
0
0
0
0
0
0
0
0
41.5
22.6%
17.0%
24.5%
34.0%
28.3%
26.4%
13.2%
26.4%
5.7%
1.9%
0
5.7%
5.7%
7.5%
3.8%
9.4%
Baseline QX
Stage 2 & 3
IRN OK 6 yrs
N = 4
0
0
0
0
0
0
0
0
0
0
0
0
0
50.0%
25.0%
0.0%
25.0%
25.0%
25.0%
25.0%
50.0%
25.0%
0
0
0
0
0
0
0
0
94
-------
Table 2 (cont.). Percent Homes Using Pesticides as reported in the Baseline Questionnaire. Results
are reported in percentages for the age class of the IRN 01. N represents the number of homes
sampled in each area.
Baseline QX
Stage 2 & 3
All IRN 01
N = 359
Personal Indoor Application
during the last 6 months
Professional Indoor Application
during the last 6 months
Other Resident applied Indoors
during the last 6 months.
Pesticide State
Needs Dilution
Diluted/Mixed by
Respondent
Diluted/Mixed by
Professional
Diluted/Mixed by Another
Applied directly
Don't Know
Use Pesticides Outdoors @
Home
Personal Outdoor Application
during the last 6 months
Professional Outdoor Applic.
During the last 6 months
Other Resident applied Indoors
during the last 6 months.
22.0%
(1-24 treatments)
25.9%
(1-12 treatments)
7.2%
(1-12 treatments)
15.6%
2.2%
12.3%
1.7%
25.3%
11.4%
59.6%
16.7%
(1-24 treatments)
28.1%
(1-24 treatments)
16.7%
(1-12 treatments)
Baseline QX
Stage 2 & 3
IRNOKlSvrs
N = 53 "
3.8%
( 1 treatment)
15.1%
(1-6 treatments)
20.8%
(1-12 treatments)
13.2%
0
11.3%
1.9%
22.6%
5.8%
58.5%
18.9%
(1-6 treatments)
15.1%
(1-6 treatments)
35.8%
(1-12 treatments)
Baseline QX
Stage 2 & 3
IRN OK 6 yrs
N=4
0
25.0%
(1 treatment)
25.0%
(2 treatments)
0
n/a
n/a
n/a
25.0%
25.0%
50.0%
0
0
50%
(1-3 treatments)
95
-------
Table 2 (cont.). Percent Homes Using Pesticides as reported in the Baseline Questionnaire. Results are
reported in percentages for the age class of the IRN 01. N represents the number of homes sampled in each
area.
Baseline QX
Stage 2 & 3
All IRN 01
N = 359
Pesticide State
Needs Dilution
Diluted/Mixed by Respondent
Diluted/Mixed by Professional
Diluted/Mixed by Another
Applied directly
Don't Know
Lawn Treatment
Lawn treatment with insect control
Mothball Use
Pets
Pets treated for Fleas/Ticks
General Health
Good
Fair
Poor
Reside on a Farm/Ranch
Family Income Level
< $9,999
$10,000 to $19,999
$20,000 to $29,999
$30,000 to $39,999
$40,000 to $49,999
$50,000 to $74,999
$75,000 to $99,999
>$ 100,000
Don't Know
Refused to divulge
23.4%
8.4%
12.8%
2.2%
23.7%
12.5%
27.3%
5.0%
5.6%
68.0%
21.2%
75.5%
20.1%
3.6%
3.3%
7.5%
8.9%
13.1%
19.5%
13.9%
20.3%
5.0%
3.1%
1.9%
4.7%
Baseline QX
Stage 2 & 3
IRN OKI 8 yrs
N = 53
20.8%
3.8%
11.3%
5.7%
32.1%
5.7%
35.8%
0
1.9%
81.1%
24.5%
90.6%
9.4%
5.7%
5.7%
9.4%
13.2%
18.9%
7.5%
32.1%
1.9%
5.7%
1.9%
3.8%
Baseline QX
Stage 2 & 3
IRN OK 6
yrs
N = 4
0
n/a
n/a
n/a
50.0%
0
n/a
0
50.0%
25.0%
75.0%
25.0%
0
25.0%
25.0%
25.0%
25.0%
96
-------
Table 3. Percent Homes Using Pesticides as reported in the Baseline Questionnaire for the Non-
Border counties of Arizona and for the Border Counties. Results are reported in percentages for
the region lived in by the IRN 01. N represents the number of homes sampled in each area.
Baseline QX
Stage 2 & 3
All IRN 01 in AZ
Non Border Co.
N = 329
Pesticides at Work/School
Work Pesticide=Raid
Work Pesticide=Repellent
Work Pesticide= Chlorpyrifos
Work Pesticide= Malathion
Work Pesticide= Diazinon
Work Pesticide= Carbaryl
Work Pesticide=Other
termat/pesticide
Work Pesticide= Atrazine
Work Pesticide= Other
Herbicide
Work Pesticide= Fungicide
Work Pesticide= Unknown
Pesticide
Work Pesticide=
Other — specified
Use Pesticides Indoors @
Home
Living Room
Family Room
Dining Room
Kitchen
Bathroom
Bedroom
Treatment Areas
Floors
Baseboards
Lower Half of Wall
Upper Half of Wall
Ceilings
Cupboards with dishes
Cupboards with Food
Storage Cabinets
Closets
Other
6.4%
0.6%
1.8%
0
0
0.6%
0.3%
0.6%
0
0.6%
0.3%
4.3%
1.2%
52.3%
36.0%
24.2%
32.5%
45.6%
42.2%
36.2%
22.8%
39.5%
7.0%
2.7%
2.4%
4.9%
4.6%
9.1%
11.6%
8.2%
Baseline QX
Stage 2 & 3
IRN 01
Border Co.
N = 30
6.7%
0
0
0
0
0
0
0
0
0
0
6.7%
0
41.5
40.0%
26.7%
33.3%
46.7%
43.3%
43.3%
13.3%
36.7%
6.7%
0
0
3.3%
0
10.0%
16.7%
30.0%
97
-------
Table 3 (con't). Percent Homes Using Pesticides as reported in the Baseline Questionnaire for the
Non-Border counties of Arizona and for the Border Counties. Results are reported in percentages
for the region lived in by the IRN 01. N represents the number of homes sampled in each area.
Baseline QX
Stage 2 & 3
All IRN 01 in AZ
Non Border Co.
N = 329
Personal Indoor Application
during the last 6 months
Professional Indoor Application
during the last 6 months
Other Resident applied Indoors
during the last 6 months.
Pesticide State
Needs Dilution
Diluted/Mixed by Respondent
Diluted/Mixed by Professional
Diluted/Mixed by Another
Applied directly
Don't Know
Use Pesticides Outdoors @
Home
Personal Outdoor Application
during the last 6 months
Professional Outdoor Applic.
During the last 6 months
Other Resident applied Indoors
during the last 6 months
22.2%
(1-24 treatments)
25.8%
(1-12 treatments)
13.7%
(1-12 treatments)
15.8%
2.4%
12.2%
1.8%
25.2%
11.2%
59.0%
25.5%
(1-24 treatments)
28.2%
(1-24 treatments)
16.4%
(1-12 treatments)
Baseline QX
Stage 2 & 3
IRN 01
Border Co.
N = 30
20.0%
(1-24 treatments)
26.7%
(1-6 treatments)
10.0%
(1-12 treatments)
13.3%
0
13.3%
0
26.7%
13.3%
58.5%
33.3%
(1-24 treatments)
26.6%
(1-6 treatments)
20.0%
(1-3 treatments)
98
-------
Table 3 (con't). Percent Homes Using Pesticides as reported in the Baseline Questionnaire for the
Non-Border counties of Arizona and for the Border Counties. Results are reported in percentages
for the region lived in by the IRN 01. N represents the number of homes sampled in each area.
Baseline QX
Stage 2 & 3
All IRN 01 in AZ
Non Border Co.
N = 329
Pesticide State
Needs Dilution
Diluted/Mixed by Respondent
Diluted/Mixed by Professional
Diluted/Mixed by Another
Applied directly
Don't Know
Lawn Treatment
Lawn treatment with insect control
Mothball Use
Pets
Pets treated for Fleas/Ticks
General Health
Good
Fair
Poor
Reside on a Farm/Ranch
Family Income Level
< $9,999
$10,000 to $19,999
$20,000 to $29,999
$30,000 to $39,999
$40,000 to $49,999
$50,000 to $74,999
$75,000 to $99,999
>$ 100,000
Don't Know
Refused to divulge
23.7%
8.4%
12.8%
2.2%
22.8%
12.5%
28.9%
5.2%
4.6%
69.3%
21.6%
75.4%
20.1%
3.6%
3.6%
7.0%
8.5%
12.5%
19.5%
14.9%
19.8%
5.5%
3.3%
1.8%
2.1%
Baseline QX
Stage 2 & 3
IRN 01
Border Co.
N = 30
20.0%
6.7%
13.3%
0
33.3%
13.5%
10.0%
3.3%
16.7%
53.3%
16.7%
76.4%
20.0%
3.3%
0
13.3%
13.3%
20.0%
20.0%
3.3%
26.7%
0
0
3.3%
0%
99
-------
Table 4. Types of pesticide applied inside and outside NHEXAS homes in the Non-Border Counties
of Arizona and the Border Counties. Results are reported in percentages for the region lived in by
the IRN 01. N represents the number of homes sampled in each area.
Households from
Non-Border Counties
in Arizona
N = 329
Indoor
4.5%
3.9%
1.3%
1.3%
0.1%
0.1%
3.0%
1.0%
0.1%
7.3%
Outdoor
4.5%
2.4%
1.2%
3.3%
1.2%
0.4%
1.2%
. 1.2%
0
1.6%
Specific Pesticides
Reported
Used
Professional
Pyrethroids
Chlorpyrifos
Diazinon
Other OPs
Carbamates
Hydromethylon
Herbicides
Other
Unknown
Households from
Border Counties in
Arizona
N = 329
Indoor
10.0%
13.1%
0
3.3%
0
0
0
0
3.3%
20.0%
Outdoor
10.0%
3.3%
3.3%
10.2%
0
0
12.2%
0
0
26.9%
Table 5. An idealized view of exposure to chlorpyrifos through multiple media and along several
pathways
Media
50* Percentile
90th Percentile
Floor Dust
Sill Wipe
Dermal
Foundation Soil
Yard Soil
Food/Beverage
Air Indoors
Air Outdoors
0.1 ug/g
0.1 ug/m 2
0.0 ug/sample
0.0 ug/g
0.0 ug/g
0.0 ug/kg
0.0 ug/m3
0.0 ug/m3
0.2 ug/g
0.3 ug/m2
0.2 ug/sample
0.2 ug/g
<0.1 ug/g
2.0 ug/kg
0.1 ug/m3
<0.1 ug/m
3
100
-------
Table 6. An idealized view of exposure to diazinon through multiple media and along several
pathways
Media
Floor Dust
Sill Wipe
Dermal
Foundation Soil
Yard Soil
Food/Beverage
Air Indoors
Air Outdoors
50th Percentile
0.0 ug/g
<0.1 ug/m2
0.0 (J.g/sample
0.0 ug/g
0.0 ug/g
0.0 ug/kg
0.0 ng/m3
0.0 ng/m3
90'h Percentile
0.2 ug/g
0.3 ug/nr
0.9 ug/sample
<0.1 ug/g
<0.1 ug/g
0.0 ug/kg
4.2 ng/m3
27.4 ng/m3
101
-------
Issues in Studying Populations along the U.S. - Mexico Border
James VanDerslice, Ph.D.; Theresa L. Byrd, Dr.P.H.; Kathleen O'Rourke, Ph.D.
University of Texas
School of Public Health
El Paso, Texas
Introduction
The purpose of this paper is to discuss some of the unique conditions found along the U.S.-
Mexico border and their implications for the design and conduct of epidemiologic studies. While the
views and experiences presented are primarily those of the authors, approximately 15 other
researchers with extensive experience along the U.S.-Mexico border were contacted during the
development of this paper for their input.
General Considerations
The U.S.-Mexico Border: Large, Diverse and Dynamic
Over the last five years, the U.S.-Mexico border has received much attention, primarily due
to issues of illegal immigration and the development of the North American Free Trade Agreement
(NAFTA). While the use of the term "U.S.-Mexico border" can give an impression of a well-defined
homogenous area, the area is, in fact, quite large, diverse, and very dynamic. The La Paz Agreement,
signed by the U.S. and Mexican governments in 1983, defines the border region as that encompassed
by a 100-km buffer on each side of the political boundary between the United States and Mexico.
This region is approximately 125,000 square miles, about the same area as the states of Maine, New
Hampshire, Vermont, Massachusetts, Connecticut, Rhode Island, New York, New Jersey, and
Delaware combined. This area stretches along 2,000 miles of border and transects several ecological
zones.
The main urban areas developed from small towns along traditional north-south trade routes.
There are 19 major population centers (>50,000 population) within the 100-km buffer, 12 of which
are in San Diego County, California. Of the remaining seven population centers, four are in Texas
(El Paso, Brownsville, McAllen, and Laredo), two are in Arizona (Yuma and Tucson), and one is in
New Mexico (Las Cruces) (1990 U.S. Census figures). Of the 25 counties that are on the border, only
three have populations greater than 500,000. Ten counties have a population greater than 90,000
people, and the 15 remaining all have populations less than 15,000.
There is significant variation in the demographic characteristics of the U.S. population living
in the border region. In 9 of the 25 border counties, more than two-thirds (66%) of households are
predominantly Spanish speaking, with a majority being Spanish speaking in an additional 13 counties.
Only three counties are predominantly (>66%) English speaking.
Low educational attainment is common in border communities. For example, in 22 of the 25
border counties, over 25% of the adult population (> 25 years of age) do not have a high school
diploma; in seven counties, more than half of the adult population have not graduated from high
school. While the area is predominantly Hispanic, the proportion of Hispanics ranges from less than
10 to over 80 %, with the highest proportions found in the lower Rio Grande Valley.
102
-------
People Cross the Border, Existing Data Systems Don't
For many decades, the pairs of "sister cities" along the border acted much like single social
and economic units. For example, until four years ago residents of Cuidad Juarez would freely cross
the Rio Grande into El Paso for work, returning to their homes and families each night. Citizens from
one side would (and still do) shop and eat in the other. Even with the increased efforts to prevent
illegal crossings into the U.S., the border can be likened to a "semi-permeable membrane," with the
"permeability" varying from day to day and location to location.
This in itself presents unique difficulties for designing and conducting epidemiological
studies. What is the "population at risk"? Does it include Mexican nationals who cross into the U.S.
temporarily, or even for extended periods of time? If health outcomes occurring to such individuals
are counted in a study, how does one account for the denominator from which these health outcomes
have been generated? Complicating the situation is the fact that, while the individuals do cross,
information systems generally do not. Thus information about these individuals is often difficult, if
not impossible, to access.
Immigration Status Is Always an Issue
Many families along the border have members living in both countries. However, because
of the economic opportunities on the U.S. side, and the consequences of not having the appropriate
immigration status to remain or work in the U.S., many residents are not willing to provide accurate
information to researchers, government agencies, and health care providers. This leads to serious
selection and observation bias as an individual's decisions regarding whether to seek health care,
where to seek health care, whether to participate in clinic- or community-based epidemiological
studies, as well as the accuracy of information provided to health care practitioners or researchers,
will certainly be influenced by a person's immigration status. In addition, immigration status may
well be correlated with other potential risk factors (e.g., income, education level, years lived in the
U.S.).
Structural Barriers to Binational Research
There are several structural barriers to conducting binational research. Few sources of funding
exist for conducting research in both countries. When such funding is obtained, transferring money,
equipment, supplies, biological samples, and even the researchers themselves, can be complicated and
time consuming.
There are a number of structural differences which make binational studies challenging. There
are significant differences in institutions, especially in health care systems, making it difficult to
recruit comparable populations and generate comparable estimates of disease occurrence. There are
problems in identifying appropriate review boards for research on human subjects in Mexico.
Universities are not comfortable having such studies reviewed solely by institutional review boards
(IRBs) in the U.S.. It can take an enormous amount of time to develop an appropriate review board
and get a protocol reviewed and approved.
There are many well-trained and dedicated scientists in the border cities of Mexico who are
quite eager to conduct research on public health problems. Many of them have very limited access
to computers or adequate laboratories. There are often differences in the underlying expectations, and
103
-------
in the sense of what is timely and adequate progress on a project. This can be a source of frustration
and conflict.
Factors Affecting Study Design
Arising from the situations described above are a number of specific factors that can directly
affect study design.
Prospective and Cross-sectional Cohorts
Identifying Cohorts.
There are several problems which arise when trying to identify sampling frames for selecting
representative cohorts. The first issue is the definition of the target population. If the target
population is defined on the basis of location of residence, then there is the potential that a small, yet
a significant proportion of the population in some areas is not U.S. citizens or permanent residents.
In such situations, many of the civil databases commonly used as sampling frames, such as voter
registration and driver's licenses, would not be representative of the target population. Furthermore,
many newer immigrants have a fear of dealing with government officials for any reason (e.g.,
registering to vote) even though they have the appropriate documentation. Thus, if the target
population is limited to individuals who are found in such civil databases, then newer immigrants and
Mexican nationals living in the study area will not be included in the study.
Using telephone listings may result in similar biases. For example, according to 1990 Census
figures, there were four census tracts in El Paso County where the proportion of households lacking
phones was greater than 30%, and these were in the lower income inner-city and peri-urban areas of
the county. Community-based surveys have found the proportion of households lacking phones to
range from 20% to 41% in rural West Texas.
Recruiting cohorts through clinics is often problematic. On several occasions we have
encountered clinic administrators who wanted clear evidence that the research was going to directly
benefit the clinic or the clinic population before agreeing to participate. Some administrators feel that
researchers have used clinic resources for recruiting and data collection, but have failed to involve
the clinic in the analysis or follow-up. In some instances, we have been told that the clinic never even
received a copy of the study results. In addition, clinics rarely have any form of IRB in place, and
yet feel the desire to have some form of oversight in addition to university IRBs.
It is our experience that once the clinic administration agrees to participate, there is a very
high level of cooperation. Many institutions, however, (including hospitals) do not have sophisticated
information systems and it can take significant time and effort to work with information system
managers to locate and extract the desired data.
A more fundamental issue is the population served by any given clinic. Many border residents
have limited access to health care. A recent Behavioral Risk Factor Survey conducted by the Paso
del Norte Health Foundation in El Paso, Texas, found that a third of the adults contacted by phone
did not have any kind of health care coverage (Paso del Norte Health Foundation, unpublished data,
1997). Sixteen percent reported that they had not seen a doctor when they needed to, and just under
half of these cited cost or lack of insurance as the main reason. Three percent said that they didn't see
a doctor because of "fear."
104
-------
Many residents use a variety of providers, so that the records from a single provider may not
contain all information about a given patient's complaints, diseases, or treatments. For example, there
has been a rapid increase in the number of providers serving areas south of El Paso where many
colonias are located. This has led to some degree of competition and the use of multiple providers.
The situation is further complicated by the use of health care providers in Mexico. Estimates
of the number of people using health care providers in Mexico vary significantly. A recent household
survey in Presidio found that 11 % of the respondents used medical care in neighboring Ojinaga, 65%
sought dental services, and 61% purchased medicines (Serrano, personal communication, 1997). A
survey in Brownsville, Texas, found that 41% of the respondents went to Mexico for medicines and
health care (Zveleta, 1985). In another survey in San Diego County just over 40% said they had used
medical services in Mexico (Nichols, 1991). Cost and the quality of treatment are the primary
reasons cited in most studies for the use of health care services in Mexico (see Warner and Reed,
1993 for further information on this issue).
Of equal importance is the use of U.S. health care by Mexican nationals. A 1987 study of
2,954 randomly-selected Tijuana residents found that 2.5% had used the U.S. health care system in
the past 6 months, and that half of these were U.S. citizens or legal residents (Geundelman and Jasis,
1990). Such cross-border utilization of health care makes it difficult to use hospital or clinic records
to gather consistent information for a sample of individuals representative of a population living in
a specific geographical region, such as on the U.S. side of the border.
Cross border utilization also impacts the identification of birth cohorts. In No gales, almost
11% of U.S. residents receive some sort of prenatal care and 1% deliver in Mexico, while 4% of
women from Nogales, Sonora, receive prenatal care in the U.S., and 8% deliver in the U.S. (Homedes
et al., 1992). Preliminary findings from a recent survey of women giving birth at a public hospital
in El Paso found that 5% said they lived in neighboring Cuidad Juarez (Byrd, personal
communication, 1997).
In some areas, a high proportion of pregnant women do not receive prenatal care until late in
their pregnancy. The El Paso study found that only 57% of the women had their first prenatal visit
in the first trimester, while 11% had their first prenatal visit in the third trimester or never received
prenatal care. NCHS estimated that in 1992 between 5.3% and 9.9% of pregnant women in the four
border states have late or no prenatal care (MRSA, 1997). As a result, generating birth cohorts
through prenatal care may miss up to 11% of the women, and these women are quite likely to be
systematically different than the rest of the population. WIC clinics have the potential of being a good
means for recruiting birth cohorts as eligibility for WIC is based solely on proof of place of residence,
not immigration status.
Public schools are another potential source of information. As with WIC clinics, proof of
place of residence is required to enroll children in public school in Texas; however, school officials
usually do not request information regarding immigration status. Such populations are relatively
stable. Further, in many smaller areas the school is an important focus of the community and school
officials are respected community leaders. If school officials recognize the importance of a study and
are involved in the planning stages, their active support and participation can be of immense value
(Redlinger, O'Rourke, and VanDerslice, 1997).
Farmworker organizations may provide a basis for identifying cohorts. Local organizations
are keenly interested in pesticides as an issue and many groups are very well organized. Church-
105
-------
based organizations are also important in many communities and might be used to generate a cohort
of individuals. It is our experience that church leaders are highly respected and trusted.
Selecting cohorts based on a specific geographical region can be problematic. There are few
accurate maps available for the border regions, particularly in peri-urban areas where colonias and
sub-divisions have been developed on former agricultural lands. Aerial photos are generally available
and can easily be used to generate a sampling frame of houses in a community. However, it can be
quite difficult to contact residents through household visits. A recent household survey in Presidio,
Texas, found almost no one at home during the first weekend of data collection (Serrano, personal
communication, 1997). In a survey of households in four colonias near El Paso, interviewers
approached every home during daytime hours, but were only able to make contact with an adult
resident in 53% of the attempts (VanDerslice, Byrd, and Mroz, 1995). While scheduling household
visits for the evenings and weekends will be necessary, it does not guarantee high rates of contact.
In a study of randomly-selected households in Sunland Park, New Mexico, 20 of the 296 selected
households could not be contacted after five or more attempts, including visits on the weekend, in the
evening, and in the early morning (VanDerslice and Shapiro, 1996).
Follow-up: Long term or short-term follow-up can be quite difficult due to the dynamic
nature of the population, particularly in out-lying areas. In Sunland Park, New Mexico, 29% of the
residents had lived there for less than 5 years (VanDerslice and Shapiro, 1996), while in Presidio,
Texas, 35% of the residents had lived in the town and 48% had lived in the same house for 5 years
or less (Serrano, personal communication, 1997). A survey of women delivering in the public
hospital in El Paso found that 20% had changed their residence during pregnancy (Byrd, personal
communication, 1997).
Addresses and phone numbers provided by residents are frequently incorrect. For example,
in a study of Hepatitis A seroprevalence among school children in San Elizario, Texas, experienced
promatoras from the community were unable to locate 20% of the children's families using the phone
number and address provided to the school (Redlinger, O'Rourke, and VanDerslice, 1997). In many
cases families will use the address of a relative, particularly for billing information. Obviously, loss-
to-follow-up in these situations will not be random, but will occur more frequently among families
with unskilled workers, as well as among newer immigrants and immigrants lacking documentation.
Follow-up may be more efficient if the school system or health care system is used as the
means of continued contact. However, there seems to be significant loss to follow-up afterbirth, even
for 2-month post-partum check-ups (Shapiro, personal communication, 1997). WIC clinics may be
a useful mechanism for maintaining contact, and it appears that mothers who enroll in WIC prenatally
are more apt to continue to use WIC after childbirth (O'Rourke, personal communication, 1997).
Case-control Studies
Case Ascertainment: In general, there is limited availability of computerized health
outcomes in the four border states. The statewide registries which do exist are relatively new. A
notable exception is New Mexico which has had a tumor registry for over 20 years. Unfortunately,
many other centralized health data sets are poorly organized or suffer from inadequate reporting.
Datasets from hospitals on the border vary widely in quality. Most computerized records exist for
billing purposes, and as such, the degree of medical information is limited. Few of these databases
have ever been used for research, and as a result, it can take significant time and effort to get approval
to access such data, and work with the information specialists to generate the correct dataset in the
106
-------
desired format.
Selection of Controls: While many of the problems described above regarding the
identification of cohorts also apply to the selection of population-based controls, the more
fundamental issue is how to actually sample from the same population that generated the cases.
Based on tumor registry or hospital records it may be impossible to tell which cases were from
Mexico. Further, Mexican nationals coming to the U.S. for health care would not be proportionately
represented among those presenting to the hospital for the types of conditions used to generate
controls (e.g., minor trauma) as compared to those presenting for more serious conditions defining
the cases (e.g., cancer). Thus hospital controls as a group may not represent the population from
which the cases were generated.
Unique Study Factors
There are a variety of unique study factors in border areas which may complicate the design
and analysis of epidemiologic studies of environmental risk factors. Studies which rely on place of
residence as a proxy for exposure may be confounded by the high degree of mobility along the border.
A household survey in Presidio, Texas, found that 43% of the respondents had lived in Ojinaga,
Mexico, before moving to Presidio (Serrano, personal communication, 1997). A survey of women
delivering at the public hospital in El Paso, Texas, found that 76% had previously lived in Mexico
(44% from neighboring Ciudad Juarez, 15% from other parts of Chihuahua, and 17% from other parts
of Mexico). It is possible that living in or working in Mexico, particularly in agricultural areas, may
have led to exposure to pesticides not licensed for use in the U.S.
Border populations also have access to many products and Pharmaceuticals not readily
available in other parts of the U.S. A recent study found several highly potent arsenic based
rodenticides available from hardware stores in Ciudad Juarez (Lugo, personal communication, 1997).
Border residents frequently cross the border to purchase Pharmaceuticals without a prescription. Lead
based paints, and a variety of foods may be obtained in border towns and easily brought into the U.S.
Other important factors include the use of pesticide containers or industrial 55-gallon drums to store
drinking water, and unique dietary patterns.
Dealing with People
Recruiting
Research Not Seen as Important: We have encountered many residents with negative
attitudes regarding research, particularly during community-based household surveys. Comments
we have heard include: "we have been studied to death and nothing has changed," "we are used and
nothing is left behind," and "all studies, no action." This appears to be particularly true in colonias,
where in a convenience sample of 269 households, 16% remember being interviewed in their home
(VanDerslice et al., 1995).
In contrast to water and sewerage which are almost universally viewed as important issues to
residents of colonias, pesticides are not seen as much of a threat. In a community survey conducted
just south of El Paso, "pesticides in food" ranked 15th out of 20 named environmental and social
problems potentially posing a "risk to myself and my family" with only 38% of the respondents rating
this risk as "high" (Byrd, VanDerslice, and Petersen, 1997). In a subsequent phone survey covering
all of El Paso County, only 13% felt that pesticide use was an "extremely serious" or "very serious"
107
-------
problem in their community (VanDerslice et al., unpublished data, 1997). A similar study in Arizona
and Sonora found that 33% of Arizona residents and 39% of Sonoran residents felt that they were "at
risk a lot" from pesticides and herbicides in foods (Udall Foundation, 1996). As one resident of
Soccorro, Texas, responded when asked if she was concerned about planes applying pesticides to
fields near her home: "How can I smell the pesticides over the smell of sewage from the (failed)
septic tank next door?"
In contrast, farmworker groups both in the U.S. and Mexico that we have contacted are very
concerned with this issue, and a local group organized and hosted a three-day conference last year to
promote interaction between their organization and researchers interested in initiating studies.
Contacting Potential Participants: Being "introduced" to the community before starting
recruiting is extremely important for maximizing the level of cooperation. Ideally, such an
introduction should be made through local, well-respected community groups (e.g., churches, schools,
civic groups, local clinics). The introduction adds a degree of legitimacy and sends the message that
the work is worthwhile.
Even an introduction through the mail, phone, or via the mass media can improve legitimacy
and remove some of element of suspicion and surprise when a interviewer knocks on the door.
Researchers with the University of Arizona found that an introductory letter stating approximately
when they would be visiting the household was quite effective (M. O'Rourke, personal
communication, 1997).
A number of surveys have collected data on the types of mass media used by border residents
and the results vary substantially. While television appears to be the most commonly used media,
results on the proportion who relied on television for environmental or health information ranged
from 10% in Sunland Park, New Mexico (VanDerslice and Shapiro, 1996), 64% for El Paso as a
whole (Byrd, VanDerslice, and Petersen, 1997), and 35% in Arizona and 81% in Sonora (Udall
Foundation, 1996). Radio appears to be the second most important media, with 27% of El Paso
residents, 7% of Arizona residents and 57% of Sonoran residents using radio to get information.
Interviewers and Interviewers: The perceived identify of the interviewers is crucial for gaining
some level of immediate trust. In some communities there is widespread distrust of strangers, and
in particular, a fear that any unknown person is an INS agent trying to locate illegal aliens. In a recent
pilot project, two professors were observing drivers leaving a school parking lot in an effort to
estimate the level of seat belt use. Many drivers would not stop and the school received several
complaints about the "government agents." Local universities are well known and generally well-
respected. Using university identification badges and university hats and shirts provides a means of
instantaneous recognition.
Female interviewers seem to be trusted more than male interviewers, and pairs of interviewers
(one male and one female) can provide security as well as a level of trust. Many communities have
informal health educators (promatoras), and these workers have been effective interviewers. During
training the difference between being an educator and data collector needs to be discussed explicitly,
and consistency with data
collection protocols between interviewers needs to be checked in the field. Some promatoras have
now been working intermittently as interviewers for various surveys over three years. Being from
the community or from nearby communities, they are familiar with study areas, and are very good
at approaching and convincing residents to participate in a study. While using Hispanic interviewers
108
-------
helps recruiting in Hispanic communities, there have been instances where such interviewers were
treated with hostility in predominantly white communities (M. O'Rourke, personal communication
1997).
Incentives for Participation: Cash or product incentives are often used as a means of increasing
participation. In our experience, such incentives seem to have only a marginal effect on participation
rates, but may be justified more on grounds of being a symbol of compensation for a person's time.
Researchers with the Arizona National Human Exposure Study (NEXUS) report that potential study
subjects seemed to either immediately see the value of the study and chose to participate, or did not
see the study as having value and refused to participate. Offering an incentive did not appear to
change a person's mind (M. O'Rourke, personal communication, 1997).
Community incentives (e.g., health fairs, free walk-in consultations, talks at community
groups) seem to be as, if not more, effective at increasing participation. Such measures give the
message that the researcher wants to see action that addresses the community's health problems and
is willing to make an investment to help bring about such changes. While there are always concerns
that such community activities might affect the study results, such potential effects can be minimized
by the sequencing of activities and the types of community incentives used.
Participation Rates from Selected Studies: Participation rates have been estimated for a number
of studies conducted along the U.S.-Mexico border. Raw participation rates measure the overall
effectiveness of locating and contacting the specific persons or households chosen to be in a study
as well as the participants decision whether to participate. Raw participation rates varied from just
under 50% to just over 90% (Table 1). Adjusted rates, which measure only the person's decision to
participate, ranged between 70 and 90 percent.
Data Collection
Instrument Development: The most obvious issue regarding instrument development for
the border region is language. Depending on location, between 20 and 90% of border residents speak
Spanish as their preferred language. However, there is substantial local variation in word usage,
particularly for words describing foods and products.
Of particular difficulty is gathering comparable data from different ethnic groups using different
languages. Standard instrument development techniques include translation and back-translation.
Few investigators, however, have assessed whether different cultures or regional groups assign
different meanings to the same word. For example, a local study found that 50% of monolingual
Spanish speakers either reversed or equated the meaning of the words "probable" and "possible" (L.
Cohn, unpublished data, 1997). Further, a similar study found that 42% of English-speaking
adolescents had the same type of misunderstanding (Cohn et al., 1995).
In bilingual populations, language use varies over time, and there are clear differences in the
use of Spanish between young adults and older adults. In addition, there are differences between
recent and less-recent immigrants, as well as for immigrants from different parts of Mexico. Such
variability in language use makes it difficult to develop valid survey instruments.
109
-------
Table 1: Participation Rates in Selected Surveys along the U.S.-Mexico Border
Location
San Elizario, TX
Santa Cruz, AZ
Webb County, TX
El Paso County, TX
El Paso County, TX
AZ-Sonora border
AZ-Sonora border
Brownsville, TX
Zavaleta, 1986
TX border
Sunland Park, NM
El Paso, TX
El Paso, TX
"raw rate = # completed
badj. rate = # completed
Study Design Participation Rates (%)
School-based, Entire School
76.6 (raw3)
Community, Cluster, Convenience
83.7 (adjusted11)
Colonias, Random Sample
69.7 (raw)
97.2 (adjusted)
Colonias, Random Sample
70.0 (raw)
93.3 (adjusted)
Colonias, All Households
49.2 (raw)
92.4 (adjusted)
Mail Survey to All Dentists
74.0 (raw)
Mail Survey to All Dentists
70.0 (raw)
Community
87.3 (raw)
Cluster Sample
83.8 (adjusted)
Community, Random Sample
66.9 (raw)
72.5 (adjusted)
WIC Clinic, Eligible Clients
^90.0 (adjusted)
Community, Convenience
90.0 (adjusted)
interviews / total number of units selected.
interviews / number of units contacted.
Citation
Redlinger et al, 1997
Clark etal., 1994
Rogers et al., 1994
Rogers etal., 1994
VanDerslice et al., 1995
Homedes et al., 1994
Homedes et al., 1994
Dutton, unpublished
VanDerslice et al., 1996
O'Rourke, unpublished
Byrd et al., 1997
Using language-based to instruments to measure psychological development in children and
changes in development over time presents unique challenges and difficulties in bilingual populations
such as in the border region. Children often use both languages to communicate, and the relative use
of one language over the other changes as they get older and are encouraged to develop the use of
English in school. As such, subtle differences in language-based ratings over time may reflect
changes in language use rather than changes in development. This is an area needing much new
research.
Common data collection techniques may also be used differently by different social or ethnic
groups. Some researchers have observed that Hispanics are less apt to select extreme values (e.g.,
strongly agree, strongly disagree) on Likert scales (L. Conn, personal communication, 1997).
Personnel in local emergency rooms report that when they ask patients to rate their pain on a scale
of 1 to 10, many Hispanic patients are unable to translate the intensity of pain into a numeric score
(Z. Green, personal communication, 1997). These examples point to the need for more in-depth
110
-------
instrument development and validation when working with populations along the U.S.-Mexico
border.
Finally, instruments to be used along the border, or in any specific region, need to be more
than just accurate translations of the original; they need to reflect the reality of the region, from the
types and names of foods being consumed, to beliefs and about specific health events. Key
informants and focus groups are crucial for finding out about these aspects of "local reality."
Interviewing
Many of the factors and problems discussed above regarding recruiting in the border region
apply equally well to interviewing. One unique issue is the use of computer-assisted interviewing.
It is our experience that the presence of a portable computer distracts the respondent, while others feel
that the use of computers has had minimal effect on the interview process.
Collecting Biological Samples
While collecting biological samples, particularly using invasive techniques, is always more
problematic than asking questions, with close cooperation from community groups or clinic staff it
can be quite successful. In the Hepatitis A seroprevalence study of children 3 to 7 years of age, we
were able to obtain blood samples from 561 of the 682 children registered in the school, an 85%
participation rate. In a recent study of folate levels in women attending a WIC clinic, approximately
90% provided a blood sample (O'Rourke, unpublished data, 1997). In the Texas Department of
Health border survey, preliminary figures indicate that while only 16% of the families contacted
refused to be in the study, 40% refused to let a blood sample be taken from their child, in spite of a
$40 cash incentive. Native Americans appear to be extremely reluctant to provide blood samples (M.
O'Rourke, personal communication, 1997).
Conclusions
A truly successful large research project will require concerted effort in preparation and in
follow-through. It should be designed as part of a larger commitment to working with a community
to help them improve health conditions over a sustained period of time. Interventions developed as
a result of research must be culturally appropriate and acceptable to the community. They should be
involved not only in the research, but in the development of interventions. This will require listening
to the community and responding to their "felt needs," as well as developing working relationships
with agencies serving the community. All groups in the community, including industries, health care
agencies, and community leaders should be involved in the process of assessment and planning. This
inclusion of all stakeholders will increase the likelihood that programs will be adopted and
institutionalized. Without this intersectoral commitment, programs cannot be sustained.
Acknowledgments
We wish to thank the many individuals who offered ideas, experiences, and suggestions in
the development of this paper, including: Dan Green, Mimi Roddy, Dr. Craig Ham's, and Dr. Hardy
Loe of the University of Texas, Health Science Center - Houston, School of Public Health; Beatriz
Vera of PSR; Dr. Tom Redlinger and Amy Liebman of the University of Texas at El Paso; Dr. Mary
Kay O'Rourke of the University of Arizona; Dr. Cynthia Lopez of the University of New Mexico;
111
-------
Rebecca Hart of the National Center for Environmental Health, Centers for Disease Control and
Prevention; Dr. Tim Wilkosky of RTI; Dr. RJ Dutton of TDK; Dr. Andres Lugo of the West Texas
Poison Control Center; Kitty Richards from the New Mexico Department of Health; and Dr. Paul
English of the California Department of Health Services.
References
Clark, L.C., et al. 1994. The Santa Cruz County community health survey. Arizona Department of Health.
(Unpublished)
Cohn, L.D., Schydlowe M., Foley J., and Copeland, R.L. 1995. Adolescents' misinterpretation of health risk
probability expressions. Pediatrics 95(5):713-6.
Guendelman, S.,and Jasis-Silberg, M. 1992. Electronics and garment maquiladoras in Tijuana: The health of
working women. Border Health 8(3): 1-55.
Homedes, N.,et al. 1994. Utilization of health services along the Arizona-Sonora border: The providers'
perspective. Salud Publica de Mexico 36(6):633-45.
Paso del Norte Health Foundation. 1997. El Paso health report. Paso del Norte Health Foundation, El Paso,
Texas. (Unpublished)
Redlinger, T., et al. 1997. Seroepidemiology of Hepatitis A among school children in a U.S.-Mexico border
community. American Journal of Public Health 87(10): 1715-7.
Rogers, G.O. 1994. Las Colonias del Alto Rio Bravo: Baseline conditions in Webb and El Paso Counties.
Center for Housing and Urban Development, Texas A&M University, College Station, Texas.
(Unpublished)
Selwyn, B.J., et al. 1992. The primary health care review approach to binational community based health care
evaluation and action along the U.S.-Mexico border. Border Health 8(3):56-66.
Udall Foundation. 1996. United States/Mexico border environmental health survey. The Morris K. Udall
Foundation. (Unpublished)
VanDerslice, J., et al. 1995. Survey of health and environmental conditions in selected colonias of El Paso
County, Texas. Texas Department of Health, Office of Border Health. (Unpublished)
VanDerslice, J., and Shapiro, C. 1996. Environmental health assessment of Sunland Park, New Mexico. New
Mexico Department of Health, Border Health Office. (Unpublished)
Warner, D.C., and Reed, K. 1993. Health care along the border. U.S.-Mexican Studies Program Policy
Studies Program, Policy Report No. 4. Lyndon B. Johnson School of Public Affairs, University of
Texas at Austin, Austin, Texas.
Zavelata, T. 1986. Health needs assessment survey: A U.S.-Mexico border community case study, 1984-85.
South Texas Institute of Latin and Mexican American Research. Texas Southmost College,
Brownsville, Texas.
112
-------
Some Observations on Studies of Pesticides and Children on the U.S.-Mexico Border
Rob McConnell, M.D.
University of Southern California School of Medicine
Los Angeles, California
Cross Cultural Considerations
Although there is an abundant literature on the challenges of cross cultural studies, these
challenges are relatively new for environmental epidemiology. Three examples from the
environmental health literature are presented.
The British Medical Research Council Respiratory Questionnaire is available in many
different languages, and it is widely used for evaluation of environmental pulmonary effects.
1 However, the standard practice of translation and back translation is not sufficient to guarantee
comparable responses across cultures. Data from Central American banana plantation workers
demonstrated high prevalence rates of dyspnea among unexposed, healthy workers (McConnell,
in press). South African miners interviewed in their mother tongue have much higher prevalence
of dyspnea than when interviewed in another language with which they are also fluent (Becklake,
1987). These culturally based differences in reporting may be relevant to the interpretation of
respiratory questionnaire data obtained from parents of differing ethnic backgrounds (or from
different ethnic sub-populations) about their pesticide exposed children.
'The application of the WHO Neurobehavioral Core Test Battery, which was designed to
be applicable cross culturally for evaluation of neurotoxic exposures, has shown unexpected
dramatic differences in neuroperformance in different parts of the world (Anger, 1993). These
differences were not explained by exposure to neurotoxins and were particularly marked for a
group of Central American farm workers, a population from which there have been large
migrations to the United States in recent years. These findings would be relevant to the
evaluation of neurobehavioral effects of pesticides, at least among older children from some
border sub-populations.
Finally and perhaps most important for the proposed border studies, the common practice
of grouping all Hispanic populations in epidemiologic studies may mask important differences
between sub-populations. The high incidence of a birth defect with clear environmental
determinants, neural tube defects, among offspring of Mexican women provides an example of
the complexity of these differences between different Hispanic populations (Shaw, 1997).
Although children of U.S.-bom women of Mexican descent have a risk of neural tube defect
similar to non-Hispanic white women, children (born in the U.S.) of Mexican bom mothers have a
risk 2.4 times that of non-Hispanic white women. A review of rates of neural tube defect within
Mexico has shown that there are marked differences between different parts of the country,
differences which may be related to the wide variety of ethnic sub-populations within Mexico or
to environmental factors.
These examples demonstrate some of the difficulties in identifying appropriate
populations for cross sectional study of potentially pesticide related respiratory, neurobehavioral,
and developmental effects among border children. One strategy for controlling for potential
confounders to such associations, many of which are likely to be unidentified and unmeasured,
113
-------
would be to restrict the study population to a group as homogeneous as possible with respect to
cultural and ethnic factors, for example children of U.S. born parents of Northern Mexican
descent. Alternatively, prospective studies of children, using each child as his or her comparison
for evaluation of the outcome of interest, would help to control for confounders associated with
different ethnic subgroups.
Identifying Heavily Exposed Populations
Previously poisoned patients, if accessible, may provide attractive high-dose population
groups for study of some outcomes of pesticide exposure, especially neurobehavioral effects.
Adults previously poisoned with cholinesterase inhibitors have demonstrated clearly and
consistently the neurobehavioral sequela of heavy exposure (Savage, 1988; Rosenstock, 1991;
Steenland, 1994). Because the results of studies of heavily occupationally exposed, but not
poisoned, groups have been less conclusive, it would make sense first to evaluate children
previously poisoned with pesticides for neurobehavioral (and perhaps developmental or
immunologic) sequelae. If such a study were to identify associations between previous pesticide
poisoning and measurements of specific neurobehavioral or other outcomes, these would be the
specific outcome measurements which might be further developed as instruments for evaluating
the effect of pesticides among children heavily exposed environmentally, but not poisoned, with
cholinesterase inhibitors.
The difficulty with this proposal is in identifying and finding previously poisoned children
for study. The California pesticide poisoning reporting system, an international model for
surveillance of occupational poisoning for many years, receives reports from physicians, who are
required by law to report any illness they suspect of being caused by pesticides. Nevertheless,
this surveillance system leaves most poisonings unreported, and the system is biased toward
capturing occupational exposures, although surveillance reports from poison control centers,
where poisoned children are often treated, are now being included routinely. Active surveillance
of pesticide poisoning, in Fresno County in California (Maizlish, 1994) and internationally
(McConnell, 1994) suggests that it would be worth exploring the feasibility of identifying a large
enough cohort of children previously poisoned with cholinesterase inhibitors from several border
counties to evaluate chronic neurobehavioral sequelae.
California also has a pesticide use data base which contains geographically referenceable
records of all agricultural applications in the state by date. The design for a potentially relevant
pilot study linking this little known resource to a data base of school absences will be presented.
114
-------
References
Anger, K.W., Cassitto, M.G., Liang, Y., Amador, R., Hooisma, J., Chrislip, D.W., et al. 1993. Comparison
of performance from three continents on the WHO-recommended neurobehavioral core test
battery. Environ Res 62:125-47.
Becklake, M.R., Freeman, S., Goldsmith, C., Hessel, P.A., Mkhwelo, R., Mokoetle, K., et. al. 1987.
Respiratory questionnaires in occupational studies: Their use in multilingual workforces on the
Witwatersrand. Ml J Epi 16:606-11.
McConnell, R., and Hruska, A. 1993. An epidemic of carbofuran and methamidophos poisoning in maize
cultivation in Nicaragua. Am J Public Health 83:1559-62.
Castro, N., McConnell, R., Anderson, K., Pacheco, F., and Hogsted, C. In press. Respiratory symptoms,
spirometry, and chronic occupational paraquat exposure. Scandin J Work Env Health.
Rosenstock, L., Keifer, M., Daniell, W., McConnell, R., Claypoole, K., et al. 1991. Chronic central
nervous system effects of acute organophosphate pesticide intoxication. Lancet 338:223-7.
Savage, E.P., Keefe, T.J., Mounce, L.M., Heaton, R.K., et al. 1988. Chronic neurologic sequelae of acute
organophosphate pesticide poisoning. Arch Environ Health 43:38-45.
Shaw, G.M., Velie, E.M., and Wasserman, C.R. 1997. Risk for neural tube defect-affected pregnancies
among women of Mexican descent and white women in California. Am J Public Health 87:1467-
71.
Maizlish, N., Rudolph, L., and Dervin, K. 1995. The surveillance of work-related pesticide illness: An
application of the sentinel event notification system for occupational risks (SENSOR). Am J
Public Health 85:806-11.
Steenland, K., Jenkins, B., Ames, R.G., O'Malley, M., Chrislip, D., and Russo, J. 1994. Chronic
neurological sequelae to organophosphate pesticide poisoning. Am J Public Health 84(5):731-6.
115
-------
Resources for Pediatric Research in the Border Region
James Ellis, M.D.
El Centro, CA
For the past 30 years, I have been a private pediatrician in California's Imperial Valley. The
Imperial Valley is located on the Mexican border about 100 miles east of San Diego. The land is desert,
but because it is irrigated by the Colorado River, it has a year-round growing season. The population of
the Imperial Valley is approximately 140,000 on the United States side of the border and close to one
million on the Mexican side. Both areas are extremely active agriculturally. My experience in the
Imperial Valley leads me to share with the U.S. Environmental Protection Agency (EPA) in order to assist
their efforts to identify and measure health end points associated with pesticide usage and its effect on
young children.
Since the health organization resources for measuring these end points will vary from area to area,
I will not attempt to catalogue these resources. I will instead offer guidelines to EPA on how to best
access resources and also foster community cooperation, understanding, and acceptance of
environmentally beneficial projects whether the project is short lived or extends over a period of years.
As a private pediatrician, I have experience in all the areas that I will address.
Cooperation is essential between EPA and the community. In conducting studies, both EPA and
the community have needs that must be met to successfully complete a project. EPA needs:
1) access to community resources,
2) access to information, and
3) access to the study population.
The community needs:
1) trust in the outside entity and
2) trust in the benefit to be derived from the project.
It is imperative that both sets of objectives be reached for their mutual benefit. The key to
meeting these objectives is in building relationships between the community and the entity (in this case
EPA).
Outsiders in any area can be met with reserve. Along the U.S./Mexican border, outsiders,
especially government representatives, are viewed as a possible threat by some segments of the
population. Therefore, knowledge of and involvement in any studies must involve the entire community,
including health care providers, in order to assure the best outcomes.
Accessing the community can be difficult. A crucial step is developing a relationship with key
physicians in the community. The physician can join as a member of the team and assist in making
rounds to other key entities with information and recruiting participants as necessary. The County
Medical Society can provide information and contact with local physicians. Working with the society and
referred physicians can build relationships that will extend into and involve the community. The most
intimate knowledge of the health community with access to unpublished data can be obtained from the
health department, clinics, private practitioners, and local hospitals. In areas where centralized structures
do not exist, the creation of an advisory committee comprised of health care components within the
community might prove productive.
116
-------
WORKGROUP REPORTS
-------
Introduction to Workgroup Reports
The workshop was structured to bring together experts from a variety of fields and attempt
to draw on their collective expertise to identify priority pediatric research questions related to
pesticide exposure. Workgroups were organized around five health endpoint domains: neurology,
developmental effects, pulmonary, immunology and cancer. After day one, the pulmonary and
immunology groups combined to discuss areas of mutual interest.
Workshop participants were assigned to the multidisciplinary workgroups on the basis of
their areas of expertise. Each group had a least one epidemiologist, exposure assessment expert,
physician and representative from the border area (e.g., health department). EPA facilitators
conducted the workgroup sessions and were responsible for producing the workgroup reports.
Day One - Health Effects
The purpose of the day one breakout group was to generate a comprehensive list of child
health endpoints under the health domain of interest that could be studied in young children and
infants. The groups were also charged with assembling sufficient information to rank the priority
of each endpoint for Phase II (pilot) study. The discussions were fueled by the health effects
speakers presentations given earlier in the day.
The groups were encouraged to list as many health endpoints as possible which could be
measured in infants and young children. Although certainly important, facilitators tried to avoid
an overemphasis on "pesticide-related" endpoints at this point. The literature is sparse and the
group leaders encouraged people to think about measurement in a relatively broad way. Some
groups organized part of the discussion around a validated test which was applicable to multiple
endpoints. A facilitators guide listed some relevant points for discussion including the gender and
age groups affected; the population prevalence (or availability of population norms); the
persistence of the endpoint (acute, chronic, etc.); whether measurement tools were available and if
special training was need to measure the endpoint; and the probability that the endpoint could be a
result of a pesticide exposure.
Each group rated the priority of the health endpoints they generated on a five-point scale
with one as the highest priority and five the lowest. All factors were considered important in the
ranking (biologic plausibility, feasibility, etc.). For example, an endpoint could be rated as a high
priority because it is easy to measure in children even if there is no strong evidence to link it
directly to pesticide exposure (e.g., growth).
A plenary group discussion was held to discuss the highest rated endpoints from each
health domain. The group considered whether the constellation of endpoints discussed for Phase
II study would adequately assess the likely health effects for children with respect to variations in
age; gender; representativeness of the potential sample (citizenship, school participation, access
to health care providers); and comprehensiveness across organ systems or endpoint domains.
Day Two - Development of Strawman Study Proposals
The purpose of the day two breakout group was to generate a collection of potential
studies for Phase II implementation. Study hypotheses were generated based on the day one
118
-------
discussion of health endpoints, and the speakers presentations on pesticide exposures and border
populations.
A collection of potential Phase II (pilot) studies were generated under each health endpoint
domain, rather than designing a "cadillac" study. Generally, the discussion was organized around
the study design rather than the endpoint (in some cases multiple endpoints could be addressed in
the same study design), but it was also appropriate to generate a series of similar designs which
require different measurement strategies (e.g., a series of cross-sectional studies). Groups were
not asked to develop a study design for every endpoint generated in the previous days' discussion,
but to try to be comprehensive and at least get some ideas about the highest priority endpoints.
The facilitators were asked to elicit information and encourage discussion about the main
hypotheses or aim of the study; the target population; the appropriate study group; exposure and
outcome assessment issues; strengths and weakness of the design; and the timeframe needed for
the study.
Groups discussed the relative merits and feasibility of the studies generated and rated the
priority of the proposed studies on a five-point scale. One is the highest priority and five the
lowest. Ranking was based on the information gathered in the earlier discussion. Groups were
instructed to be mindful of the time frame for Phase II studies. If a priority study was a longer
term (larger scale, etc.) endeavor, could a pilot test be done at this phase?
Reports on the strawman proposals were made to the entire group on the morning of day
three in a plenary session.
119
-------
Neurobehavioral Workgroup Report
Workgroup Members: Dave Otto (facilitator), Stephanie Padilla (rapporteur), David
Bellinger, Rebecca Gomez, Sue McMaster, Rob McConnell, Rossanne Philen, Gary
Robertson
Introduction
Pesticides are widely used in the border region for agricultural purposes and for residential
pest control. Unlike other chemicals encountered in the environment, pesticides are designed to
kill and the mechanisms of action are well known. Organophosphorus pesticides are the most
widely used class of pesticides along the border (add ref). OP pesticides produce acute
cholinergic effects and, in some cases, delayed neuropathy. While the acute effects of high level
pesticide exposure have been documented extensively, the effects of chronic, low-level exposure
in human populations—a situation typical of the border and other agricultural areas—are poorly
understood.
Day 1: Recommended Neurobehavioral Tests
The initial question to be examined in the workshop was what measures are available in
selected disciplines for studying health effects in young children (aged 1-5 years), without
considering the specific features of pesticide exposure. This workgroup was charged with
discussing neurobehavioral and psychometric measures. Sensory measures were also considered.
Selection of appropriate tests to assess the neurobehavioral effects of chemical exposure in
young children is a challenging task for several reasons including: (1) the range of cognitive and
sensorimotor behaviors in young children is limited; (2) the sensorimotor and cognitive abilities
of young children change very rapidly over time, necessitating different measures at different ages
(see Table 1); and (3) the predictive validity of available sensorimotor tests before age three is
very poor (Bellinger, this volume). The logistic constraints of field testing also impose
limitations on which tests can be done. The neurobehavioral group reviewed tests of sensory,
motor and cognitive functions appropriate for use in testing infants and children of different ages.
Test characteristics that were considered included general psychometric qualities (test-retest
reliability, concurrent validity, standardization of instrument, and sensitivity to neurobehavioral
impairment); applicability for use in field testing (cost, simplicity and time of test administration;
training needed to administer and score the test); and age range of children for which the test is
appropriate. The group also considered if the test has been adapted for use with Hispanic
populations. Results of this discussion are summarized in Table 1. The only tests recommended
for use in children under age 3 are the Bayley Scales of Infant Development (Bayley, 1993). The
Bayley includes two primary scales, the Mental Development Index (MDI) of cognitive skills and
the Psychomotor Development Index (PDI) of motor skills. The sensitivity of these scales to
environmental insult has been demonstrated repeatedly in pediatric lead studies (Dietrich &
Bellinger, 1994). The strengths and weaknesses of the Bayley and other tests available for
assessing neurobehavioral function in young children are discussed in detail elsewhere in this
volume (see chapters by Bellinger and Llorente).
120
-------
TABLE la. SUMMARY OF RECOMMENDED NEUROBEHAVIORAL TESTS (Part A)
END- POINT
Reliability
Validity
Standardized
Simplicity
Admin Time
Cost
Training
Age Range
Sensitivity
Spanish
Translation
Confounders
Priority
BAYLEY-II
(2nd ed.)
High
Concurrent, not
Predictive
•
•
30-40 mill
$500-$600
Moderate
l-42mos
Pb/PCB
Yes
Iron deficiency;
not affected by SES
<2 yrs of age
1
WPPS1-R
High
Concurrent
and Predictive
•
•
1 hr
$500/kit
$1.50/test
Moderate
3-7 yrs
•
?
SES
1
WRAML
(Memory & Learning)
High
•
•
•
40-45 min
$240/kit
simple
5-17 yrs
?
9
1
WRAVMA (Visun Motor
Abilities)
High
•
•
•
25-30 min
$250
simple
5-17 yrs
7
Neutral
1
PEABODY
?
?
•
•
45 mill
Supplies
simple
3-7 yrs
V
1
Abbreviation Key for Tables la and Ib
Bayley-II Bayley Scales of Infant Development, 2nd ed. (Bayley, 1993)
NES Neurobehavioral Evaluation System (Letz, 1994)
Peabody Peabody Developmental Motor Scales (Folio & Sewell, 1993)
Vineland Vineland Adaptive Behavior Scales (Sparrow et al, 1984)
WISC-IH Wechsler Intelligence Scales for Children, 3rd ed. (Wechsler, 1991)
WPPSI-R Wechsler Preschool and Primary Scales of Intelligence-Revised (Wechsler, 1989)
WRAML Wide Range Assessment of Memory and Learning (Sheslow & Adams, 1990)
WRAVMA Wide Range Assessment of Visual Motor Abilities (Adams & Sheslow, 1995)
121
-------
TABLE Ib. SUMMARY OF RECOMMENDED NEUROBEHAVIORAL TESTS (Fart B)
END-POINT II VINELANO
II
Reliability
Validity
Standardized
Simplicity
Admin Time
Cost
Training
Age Range
Sensitivity
Spanish Translation
Confounding Factors
Priority
•
•
•
•
30 min
$50
Simple
3 yrs - teens
7
culture-
neutral
3
VISUAL ACUITY
•
•
•
•
3 min
Simple
4+
—
•
1
VISUAL CONTRAST
SENSITIVITY
•
•
•
•
3 min
$500
Simple
7+
•
•
1
TACTILE
SENSITIVITY
•
•
not for kids
~
1 5 min
$2,000
Simple
7+
•
•
1
wisc-m
•
•
•
•
1 hr*
$5004600
Moderate
6-16
•
sns
i
NFS
•
•
~
•
$5001
Simple
5/7+
•
•
1
* Shorter screening test available
122
-------
For children older than three, a wide range of tests are available. Tests recommended by
this group include the Wechsler Preschool and Primary Scales of Intelligence-Revised
(Wechsler, 1989) for children aged 3-7, the Wechsler Intelligence Scales for Children, 3rd ed.
(Wechsler, 1991) for children aged 6-16, the Wide Range Assessment of Memory and Learning
(heslow & Adams, 1990), the Wide Range Assessment of Visual Motor Abilities (Adams &
Sheslow, 1995), the Peabody Developmental Motor Scales (Folio & Sewell, 1993), and the
Vineland Adaptive Behavior Scales (Sparrow et al, 1984). These tests were selected because
they cover most of the major functional domains (except sensory) that one would want to assess
when the targets of a toxicant are unknown and the objective is to be comprehensive. Moreover
the psychometric characteristics of these tests are very robust. Both Wide Range tests have the
advantage of providing for the assessment of multiple factors within a broad domain
(memory/learning or visual-motor function). A child's performance on different wide range
factors can be compared directly because all parts of these tests were normed on the same group.
The Vineland Scales, administered to parents in interview form, provide additional information
on four aspects of child behavior—communication, daily living skills, socialization and motor
skills.
Traditional psychometric tests, including some of those mentioned above, have been
applied extensively in studies of children exposed to lead, methylmercury and to a lesser extent,
polychlorinated biphenyls (see Bellinger, this volume). These tests, in some cases, are time
consuming, require highly trained personnel to administer and interpret, and often yield only
global measures of cognitive function. During the past decade many traditional tests have been
adapted for computer administration and are available in rapid, reliable and easy-to-administer
form. The Neurobehavioral Evaluation System (NES) is the most widely used computerized
system developed specifically for human neurotoxicity testing (Letz, 1994). NES was designed
for testing workers with a minimum education level of 5th grade, although many of the tests can
be performed by school children (Otto et al, 1996) and even preschool children (Winneke et al,
1994). A pictorial version of the continuous performance test was recently developed
specifically for use with preschool children (Dahl et al, 1996). NES tests provide a very useful
and efficient alternative to traditional paper-and-pencil tests for children aged six and older.
Tests of sensory function, particularly vision, have also proven to be sensitive indicators
for the early detection of neurotoxicity (e.g., Mergler, 1995; Hudnell et al, 1996) and are
integral components of several human neurotoxicity testing batteries (Anger et al, 1994;
ATSDR, 1995). Three sensory tests are recommended for use with children-visual acuity,
visual contrast sensitivity, and tactile sensitivity. Snellen charts for testing visual acuity are
available in a variety of forms including pictures suitable for use with young children. Visual
acuity is a measure of the ability to focus images on the retina. Contrast sensitivity is a measure
of spatial vision-the ability to discriminate line gratings with varying spatial frequencies.
Preferential looking methods are available to measure visual acuity and contrast sensitivity in
infants and young children (Adams et al, 1992; Adams & Courage, 1993). The Functional
Acuity Test, F.A.C.T. 101 (Stereo Optical Co, Chicago, IL), is recommended for measuring
contrast sensitivity in children aged seven and above. Tactile sensitivity is a measure of
somatosensory function assessed by administering vibratory stimuli to the fingers or toes and
determining the intensity threshold for detecting the vibratory stimulus. Several tactile testing
devises are available commercially (Amler and Gibertini, 1996). Tactile sensitivity can be
measured in children six and older.
123
-------
Day 2: Proposed Neurobehavioral Study Designs
The Neurobehavioral Workgroup discussed three study designs--(l) a retrospective cohort
design, (2) a cross-sectional study, and (3) a longitudinal cohort study. The basic hypothesis
addressed by these studies is that exposure to pesticides produces neurotoxic effects in children.
A brief description and rationale for these study designs follow.
(1) Retrospective Acute, High-exposure Cohort Study. While one of the long term
objectives of the pesticide studies in children is to determine if exposure to low levels of
pesticides has measurable adverse health effects, one obvious starting place to examine the issue
of pesticide health effects in children is in those who are exposed to high levels and who may
have suffered acute symptoms from the exposure. Thus, we propose a retrospective cohort
study of a fairly small group of children with clearly defined, high-level exposure for one initial
study to determine unequivocally whether or not pesticide exposure at acutely toxic levels
produces neurotoxic effects in young children. The hypothesis would be that exposure to high
levels of pesticides, enough to produce acute toxic effects in children, also will produce
measurable adverse neurologic effects. In this study common confounders to be controlled for
include socioeconomic status, ethnicity, birth weight, gestational age, parental IQ, diet, and
exposure to other potential neurotoxicants such as the heavy metals arsenic, mercury, and lead.
The study would be designed as a retrospective cohort study. Thirty children poisoned
during the period 1990-1995 while aged 0-4 years would be selected from the registries of
Poison Control Centers. One or two age (+/- 6 months) and gender matched unexposed controls
would be selected per case. Exposure would be determined from hospital records. Age
appropriate tests would be administered: these would include psychometric tests such as the
Bayley-II, WPPSI-R, WISC III, and sensory tests to assess vision, vibration sense, and balance.
Results of psychometric tests would be compared to national norms as well as to age and gender
matched controls. A positive result of this study would be defined as the finding that children
who were poisoned by pesticides during early childhood have measurable adverse neurologic
effects on psychometric and sensory neurologic testing. A positive result would lead to cross-
sectional or longitudinal studies of children with chronic, low-level pesticide exposure. If the
retrospective cohort study yielded negative results, additional neurobehavioral testing of
pesticide-exposed children may still be needed, since chronic low level pesticide exposure may
lead to different neurological effects than a one time high level exposure. However, further
discussion of the value of such studies would be warranted.
(2) Cross-sectional Chronic, Low-exposure Study. If results of the retrospective cohort
study suggest that acute, high level exposure to pesticides produces neurotoxicity in children,
then a further study of exposure in children would be advisable. In this study the hypothesis
would be that differences can be detected in the neurobehavioral measures among children in
three exposure level groups. Three groups- high, middle, and low deciles (10%) of exposure--
would be selected based on responses to an exposure and pesticide usage questionnaire which
would be administered to mothers. Children in this study would be aged 1.5-2.5 years, and only
mothers with children in this age range at the beginning of the study would be eligible for
participation. Approximately 100 child participants would be recruited. Possible sources for
participants are Women Infants and Children's (WIC) Clinics or state health department well-
baby clinics. An alternative approach would be to identify groups of children more likely to be
exposed to high levels of pesticides. Possibilities might include children of pesticide
124
-------
exterminators, children living in apartment buildings sprayed on a regular schedule with
chlorpyrifos, or children of farm workers-particularly those whose parents take them to the
fields with them. In selecting an exposed group, careful consideration must also be given to an
appropriate and readily available non-exposed control group. For instance, finding controls for
children of exterminators—e.g., children of appliance repairmen—or children living in
chlorpyrifos-treated apartment buildings—e.g., children living in apartments not treated for
cockroaches—might be easier than finding comparable non-exposed matches for farm children.
The Bayley-II Test (2nd-ed.) is recommended for neurobehavioral assessment of children
in the age range of 1.5 to 2.5 years, and would be administered to child participants. Exposure
measures would include house dust as well as urine samples for biological measures of agents
such as specific organophosphates (OP), OP metabolites, a-esterases, and an OP screen;
carbamates could also be measured. Since the consistency of OP urinary measures over time is
not well documented, as a preliminary pilot study a subset of 30 children would be selected to
participate in repeated urine sampling at time intervals such as weekly or every two weeks. This
would give data regarding the consistency of the urinary pesticide measures over time in the
three different exposure groups. A positive result of this study would be defined as the finding
that children who were chronically exposed to pesticides at levels less than those considered
acutely toxic, during early childhood, have measurable adverse neurologic effects on
psychometric neurologic testing. This study would also provide information to determine how
well questionnaire derived exposure information correlates with biological measures of
exposure. Common confounders that would be controlled for in the neurobehavioral analyses
include socioeconomic status, ethnicity, birth weight, gestational age, parental IQ, diet, and
heavy metals such as arsenic, mercury and lead.
(3) Longitudinal Cohort Study. If neurobehavioral function is shown to be impaired by
low-level chronic OP exposure in the cross-sectional study, a logical next question is whether
neurobehavioral function changes over time as a result of chronic exposure. We propose to
select 100 children, aged 1.5 to 2.5 years, living in a high-risk area, such as an agricultural area,
from WIC or well-children clinics, or Head Start or day-care centers. The Bayley Test would be
administered to these children at 3-month intervals for one to two years. Urine samples would
also be obtained at each testing for measurement of OP levels, metabolites and a-esterases. This
would provide a perspective over time of changes in both neurological function and urine
pesticide levels. If it were necessary to economize on analysis costs, urine samples could be
frozen and then possibly aggregated over time. Frozen urine samples could also be analyzed
retrospectively, such as when an abnormal Bayley score was obtained. However, the most
desirable study design would be to analyze each urine specimen obtained over the entire time
period. A positive result of this study would be defined as the finding that children who were
chronically exposed to pesticides at levels less than those considered acutely toxic, during early
childhood, have measurable adverse neurologic effects on psychometric neurologic testing, and
that these effects can be measured longitudinally over time. Common confounders that would
be controlled for in the neurobehavioral analyses include socioeconomic status, ethnicity, birth
weight, gestational age, parental IQ, diet, and heavy metals such as arsenic, mercury and lead.
125
-------
References
Adams R.J., Courage M.L.. Contrast sensitivity in 24- and 36-month-olds as assessed with contrast
sensitivity card procedure. Optom. Vis. ScL, 1993, 70:97-101.
Adams R.J., Mercer M.E., Courage M.L., van-Hof-van Duin J. A new technique to measure contrast
sensitivity in human infants. Optom. Vis. Sci., 1992, 69:440-446.
Adams W., Sheslow D. Wide Range Assessment of Visual Motor Abilities. Wide Range Inc.,
Wilmington, DE, 1995.
Amler R, Gibertini M. (Eds) Pediatric Environmental Neurobehavioral Test Battery. Agency for Toxic
Substances and Disease Registry, Atlanta, 1996.
Anger W.K., Letz R.E., Chrislip D.W. et al. Neurobehavioral test methods for environmental health
studies of adults. Neurotoxicol. Teratol, 1994, 16:489-497.
Bayley N. Bavlev Scales of Infant Development (2nd ed.). The Psychological Corp., San Antonio,
1993.
Bellinger D.C. Assessing neurobehavioral effects of environmental toxicants on children: Options and
issues. Proceedings of Workshop on the Assessment of Health Effects of Pesticide Exposure on
Young Children held in El Paso, TX, Dec. 1997.
Dahl R., White R.F., Weihe P. et al. Feasibility and validity of three computer-assisted neurobehavioral
tests in 7-year old children. Neurotoxicol. Teratol. 1996, 18:413-419.
Deitrich K., Bellinger D.C. Assessment of neurobehavioral development in studies of the effects of
fetal exposures to environmental agents. In: Prenatal Exposure to Toxicants: Developmental
Consequences. H. Needleman, D. Bellinger (eds.) The Johns Hopkins Press, Baltimore, 1994,
pp.57-85.
Folio M., Sewell R. Peabodv Developmental Motor Scales and Activity Cards Manual. DLM Teaching
Resources, Allen, TX, 1993.
Hudnell H.K., Skalik I., Otto D., House D., Subrt P., Sram R. Visual contrast sensitivity deficits in
Bohemian children. Neuro.Toxicol. 1996, 17:615-628.
Letz R. NES2 User's Manual (version 4.6). Neurobehavioral Systems, Atlanta, GA, 1994.
Llorente A.M. Evaluation of developmental neurocognitive and neurobehavioral changes associated
with pesticide exposure: Recommendations for the U.S. Environmental Protection Agency.
Proceedings of Workshop on the Assessment of Health Effects of Pesticide Exposure on Young
Children held in El Paso, TX, Dec. 1997.
Mergler D. Behavioral neurophysiology: Quantitative measures of sensory toxicity. In:
Neurotoxicology: Approaches and Methods. Chang L.W., Slikker W. (eds.). Academic Press,
San Diego, 1995,pp.727-736.
Otto D., Skalik I, House D., Hudnell H.K. Neurobehavioral Evaluation System (NES): Comparative
performance of 2nd-, 4th-and 8th-grade Czech children. Neurotoxicol. Teratol. 1996, 18:421-
428.
Sheslow D., Adams W. Wide Range Assessment of Memory and Learning. Wide Range, Inc.,
Wilmington, DE, 1990.
Sparrow S., Balla D., Cicchetti D. Vineland Adaptive Behavior Scales Interview Edition Survey Form
Manual. American Guidance Service, Circle Pines, MN, 1984.
Wechsler D. Wechsler Preschool and Primary Scale of Intelligence-Revised. Psychological Corp., San
Antonio, 1989.
Wechsler D. Wechsler Intelligence Scales for Children (3rd ed) Manual. Psychological Corp., San
Antonio, 1991.
Winneke G., Altmann L., Kramer U. et al. Neurobehavioral and neurophsyiological observations in six
year old children with low lead levels in East and West Germany. Neuro.Toxicol. 1994, 15:705-
714.
126
-------
Developmental Workgroup Report
Workgroup Members: Pauline Mendola (facilitator), Asa Bradman, R.J. Button, Steve Hern,
Antolin Llorente, Andres Lugo, Anne Sweeney.
Day 1 - Health Endpoints
Introduction and Synopsis of Discussion
The purpose of the first breakout group discussion was to generate a list of health
endpoints in pediatric populations which could be classified as "developmental". Our task was
to try and enumerate potential health effects without giving much attention to exposure or study
design issues at this stage. This proved difficult and there was considerable productive
discussion of the need to begin with a prenatal, longitudinally-followed cohort in order to
adequately assess exposure and pediatric developmental health while controlling for a variety of
confounding factors. Even with a prenatal cohort, health effects occurring earlier in the
reproductive spectrum (e.g., fertility, fecundability) could not be ruled out.
There was substantial concern that it would be difficult to attribute any effect of pesticide
exposure to child developmental measurements without understanding more about the potential
mechanisms of action and controlling for factors that are known to influence child development
(e.g., gestational age, maternal education, social class, etc.). The timing of exposure, particularly
during gestation, may influence developmental outcomes as well as the dose. In this case, the
issues are particularly complex because of the limited information on pesticide-related health
risks for children. Exposure could be directly and independently related to a developmental
health risk, but it seemed likely that the relation between exposure and child development would
be more complex. Exposure could be related to social class, education and income and could
also be related to other risk factors such as congenital anomalies, birthweight, nutrition and
general health status. The probability of biologic relation between pesticide exposure and
developmental effects was discussed in general as were issues regarding latency (time between
exposure and health effect). Prenatal exposure was thought to be important for many
developmental endpoints, either through a direct mechanism or as a potential initiator or
confounder in the presence of a later postnatal exposure.
In the absence of a clear understanding of the likely pathway and mechanism for pesticide
exposure to influence child development, the group proceeded to discuss potential health
endpoints for study that could help define the relation or provide background information on the
health status of border children to aid in the interpretation of future study findings.
The workgroup identified nine health endpoint groups, which are outlined in table 1
below. A more detailed discussion of each endpoint follows the table. Endpoints are presented
in order of the priority ranking given by the group. Rankings were based on a five-point scale
with "1" as the highest and "5" as the lowest priority. Priority scores were based on relevance
of the endpoint for a study of pesticide-related effects and need for baseline health status
information on the population.
127
-------
Birth Defects, Stillbirths, Spontaneous Abortion (priority ranking: 1)
Gender/Age. This endpoint addresses males and females from recognition of pregnancy
through the newborn period (primarily). Some birth defects may not be diagnosed until the
second year of life.
Population prevalence. The population prevalence of major birth defects among live born
children is about 3-4/100 live births. Neural tube defects (NTDs) are a particular concern in the
border area. The general prevalence of NTD in the United States is about 1-1.5/1000 live births.
In the border area, the prevalence may be higher (1.4-2.8/1000 live births). One to two percent
of births are stillbirths and 12-15 percent of recognized pregnancies end in spontaneous
abortion. Anomalous embryos and fetuses are at a higher risk of loss, particularly early in
gestation. This, as well as the prevalence of prenatal diagnostic procedures and elective
terminations will effect the birth prevalence of major birth defects, reinforcing the importance of
prenatal assessments to reduce survivor bias.
Measurement. Potential sources of data discussed were: birth defects registries, birth
certificates and fetal loss registries; hospital discharge data; and clinic and OB/GYN medical
records.
Modification/pilot testing needed. There was interest in learning about the availability of
genetic information, both to potentially classify aborted fetuses based on karyotype and to
potentially develop genetic markers of susceptibility, exposure and/or effect. The need for
standardization of methods across the border area was expressed for all studies using abstraction
of registry or medical record data.
Mental, Motor, Adaptation (priority ranking: 1)
Gender/Age. Appropriate for both males and females. Different instruments are available
to assess these endpoints from newboms through age 18. The group focused discussion on
children from 4 months to 42 months.
Population norms. Norms are available for all tests. Children would be assessed for age-
appropriate development.
Measurement. The Bayley scales (0 to 42 months) have been adapted for Spanish language
populations and should be used here. Other inventory measures such as the Minnesota Infant
Development Inventory (4 to 15 months), Child Development Inventory (s 15 months), Child
Behavior Checklist (two forms: ages 2-3 years and ages 4-18 years) should also be included in
the assessment. The idea is to get a relatively comprehensive picture of child development using
both infant/child examinations and parental reports. Indices will include fine and gross motor
skills, an index of mental function, and adaptation. Other possible tests for young children
include the Peabody Motor Scale (1 year to 6 years old), and the Vineland Adaptive Behavior
Scales (0 to 18 months). The Woodcock-Johnson test of cognitive abilities can be conducted on
children three years of age or older.
Examiner needs. Testing must be done by trained personnel. Minimally, a masters level
psychologist with special training or certification in the test protocols is needed.
128
-------
Modification/pilot testing needs. The Bayley scale and Woodcock-Johnson test have been
adapted for Spanish speaking populations. Other measures will need adaptation and pilot testing.
Gender/Age: Males and females of all ages could be addressed under this endpoint. Main focus
is on young children (under 42 months).
Population prevalence: This was hard to determine. There is a consensus that acute
pesticide poisonings in children are grossly underreported. It was estimated that greater than 20
cases per year would be reported to poison control centers; national average was estimated at
1/2500 children. Emergency room visits were discussed as another potential source of cases and
also as a datasource to estimate prevalence.
Measurement. The idea behind this endpoint is to establish some notion of the natural
history of acute pesticide exposure. Children with documented acute high exposures can be
carefully examined for subsequent health effects. The developmental status of children would be
characterized using age appropriate measures (see Mental, Motor, Adaptation above). Children
would be followed over time to observe persistence of developmental effects and potential
sequelae of acute poisoning.
Examiner needs. Testing must be done by trained personnel. Minimally, a masters level
psychologist with special training or certification in the test protocols is needed.
Modification/pilot testing needs. Case finding through poison centers and hospitals must
be piloted. The Bayley scale and Woodcock-Johnson test have been adapted for Spanish
speaking populations. Other measures will need adaptation and pilot testing.
Growth (priority ranking: 1.5)
Gender/Age: This endpoint can be measured in males and females of all ages. The most
interest is in birth measurements and longitudinal measures from birth through age five.
Population norms. Gender-specific norms are available. There was some discussion of the
appropriateness of the available growth curves for border populations. Children can be
compared to curves but there is a benefit to following children over time so that they can serve as
their own controls as to rate of growth, etc.
Measurement. Birth measurements, particularly head circumference, were of interest. All
birth measurements (weight, length, etc.) should be interpreted relative to gender and gestational
age. These measures might be found in medical records and birth certificates but there was
concern about error in measurement (particularly for gestational age, head circumference, and
length). Height (length) and weight could be measured prospectively over time or could be
ascertained from pediatric medical records for children with regular care at the same clinic/site
over time.
129
-------
TABLE 1: ENDPOINTS RECOMMENDED FOR DEVELOPMENTAL ASSESSMENT
Health Endpoint
Birth defect, stillbirths, spontaneous
abortion
Mental, motor, adaptation
Acute poisoning sequelae
Growth
Language
Birthweight, gestational age
Social development
Infant mortality
Puberty
Hearing
Priority
i
i
1.5
1.5
1.5
2
4
5
5
Not rated
G ender
M&F
M&F
M&F
M&F
M&F
M&F
M&F
M&F
M&F
M&F
Age
Gestation to birth
Birth to 42 mo.
Focus < 42 mo.
Birth to 5 yrs.
Birth to 30 mo.;
36 mo. to preschool
Birth
Birth to 3 yrs.
Birth to 1 yr.
8 to 18 yrs.
Birth to 18 yrs.
Prevalence/Norms
Available
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Measurement Tools
Available
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Pilot
needed
N
Y except
for Bayley
Y except
for Bayley
N
Y
except for
ci-:i,r
N
Y except
for Bayley
N
Y
Y
Acute Poisoning Developmental Sequelae (priority ranking: 1.5)
130
-------
Examiner needs. Standardization of measurement techniques is needed. Substantial error
in measurement could obscure the potential relation between exposure and outcome. Most of
these measurements are typically done by nursing staff, hospital house staff, or pediatricians. If
measurements were made prospectively, someone with training to follow a measurement
protocol (no special educational requirements) would be needed.
Language (priority ranking: 1.5)
Gender/Age: Males and females could be evaluated. Our focus was on children from
birth through 30 months and 36 months through preschool.
Population norms. Norms are available for this test. Children can be compared to age-
appropriate norms.
Measurement. MacArthur Scales can be used to assess language development in children
from 0 to 30 months. The Comprehensive Evaluation of Language Fundamentals can be used
with children 36 months or older.
Examiner needs. Testing must be done by trained personnel. Minimally, a masters level
psychologist with special training or certification in the test protocols is needed.
Modification/pilot testing needs. The Comprehensive Evaluation of Language
Fundamentals has been adapted for Spanish speaking populations. The MacArthur Scale needs
adaptation and pilot testing.
Birthweight, Gestational Age (priority ranking: 2)
Gender/Age. Appropriate for all newboms, male and female.
Population norms. Available for general population in gender-specific form. Less well
established for preterm infants and applicability of general norms to border population is
unknown.
Measurement. Birthweight could be measured using birth certificate data or medical
records. There was concern about the validity of gestational age reported on the birth certificate.
In the absence of research protocols for recording gestational age, the clinical or pediatric
estimate of gestational age is assumed to be more accurate in the medical record.
Examiner needs. Standardization of measurement techniques is needed. Ideally, a
research protocol would be in place to aid in the assessment of gestational age.
Social Development (priority ranking: 4)
Gender/Age. Males and females could be evaluated. Our focus was on infants and
toddlers (0-3 years old).
Population norms. Norms are available and children can be compared to age-appropriate
norms.
131
-------
Measurement. Bayley Scales of Infant Development, Child Development Inventory and
the Vineland Adaptive Behavior Scales.
Examiner needs. Testing must be done by trained personnel. Minimally, a masters level
psychologist with special training (certification?) in the test protocols is needed.
Modification/pilot testing needs. Instruments need adaptation and pilot testing for
Spanish speaking populations.
Infant Mortality, Neonatal and Postneonatal (priority ranking: 5)
Gender/Age. Appropriate endpoint for male and female livebom infants. Neonatal
mortality occurs within the first 28 days of life and postneonatal mortality occurs between day 29
and one year.
Population prevalence. There is substantial variation in infant mortality by race and
ethnicity. Infant mortality was approximately 8.4/1000 live births in the United States in 1994.
The rate for Mexican Americans is similar to Caucasians (6.5 and 6.6/1000, respectively). There
is also a suggestion in the literature that Mexican-born mothers have better birth outcomes than
US-born mothers (may or may not impact on this question?).
Measurement. It was thought that all border states had linked birth-death files. Concern
was expressed about error in recording the underlying cause of death for infants. The reliability
and similarity of matching procedures across states should be assessed to ensure that the data are
of comparable quality.
Puberty, Age at Menarche, Secondary Sex Characteristics (priority ranking: 5)
Gender/Age. Males and females, aged 8 to 18 could be assessed. There was some
interest in looking at younger children for signs of precocious puberty, but it was felt this
endpoint would have to be studied on a more focused basis rather than on a population level
since the incidence would likely be low.
Population norms. Age-specific norms are available for pubertal development for both
males and females (Tanner scales).
Measurement. Mailed questionnaires have been developed for pubertal self-assessments
based on the Tanner scales. Physical examinations could also be conducted by physicians.
Modification/pilot testing needs. The acceptability and adaptability of existing measures
to the border population would need pilot testing and potential adaptation.
Hearing (no ranking)
This endpoint was raised during the plenary discussion after the first workgroup session.
The workgroup saw some relevance for including hearing as a sensory endpoint, but in the
absence of data to suggest that hearing might serve as a target endpoint for pesticide exposure,
there was more interest in general developmental motor and cognitive outcomes.
132
-------
Day 2 - Study Designs
Introduction and Synopsis of Discussion
The purpose of the workgroup session on day two was to generate a collection of ideas
for Phase II studies that addressed the developmental health domain. The summary sheets from
the health endpoint session on day one were available for reference. The group began with a
discussion of "priority" populations including: poisoned children (who could also serve as
sentinels for areas of high pesticide exposure and who are more likely to experience and thus
identify health endpoints with the strongest relation to exposure); farm families (farm workers,
people who live in agricultural areas); children in homes/schools/daycare with regular structural
(or regular indoor) use of pesticides; siblings of poisoned children; children in colonias; clinic
populations with nonspecific symptomatology consistent with pesticide exposure; and clinic
populations with identified developmental problems. Building on this discussion and the
previous dialogue on developmental health endpoints, a series of potential study designs were
assembled.
For the purpose of this report, the eight study designs are described in the order that they
were presented at the strawman report on day three. There are three analytic studies, two
descriptive studies and three "capacity building" studies. The actual day two discussion was
much more dynamic, with ideas going back and forth between priority populations, endpoint and
exposure measurement, and other design issues. As the products of a dynamic process, all of the
study designs described here have some relation to the overall program of research proposed.
Analytic Studies - Prospective Prenatal Cohort
Study design: Prospective prenatal cohort.
Level of inquiry: Formal test of hypothesis.
Main hypothesis: Pesticide exposure is related to delayed and/or altered development and
long term developmental problems.
Secondary hypothesis: Prenatal exposure to pesticides is more relevant than postnatal.
Target (referent) population: Exposed children.
Study group: Enroll women prenatally from a "priority population" such as an
occupational group (farm workers) or from a clinic in an agricultural area; compare these women
and their children to women and children from a group/clinic assumed to have low exposure.
Exposure measurement: Prenatally: questionnaire exposure/residence history and current
practices; maternal blood and urine for exposure assessment and cholinesterase levels; house
dust. Perinatally: maternal blood and urine, cord blood for exposure and cholinesterase levels;
house dust. Collect breast milk samples from lactating women. Every six months for 2-3 years,
characterize the environment, house dust, pesticide questionnaires, biologic samples from child.
133
-------
Table 2: Outline of Proposed Studies
Proposed Study
Prospective Prenatal
Cohort
Poisoning Case
Followup
Symptomatic Children
Biologic Sample
Correlations
G1S Infant Health Status
Compile Pesticide Dose
Information
Adaptation of
Neurodevelopmental
Tests
Border Physician
Training
Level
Analytic
Analytic
Analytic
Descriptive
Descriptive
Capacity Building
Capacity Building
Capacity Building
Design
Prospective cohort
Case Series
Nested Case-Control
Cross-sectional
Cross-sectional
N/A
N/A
N/A
Target Population
Exposed children
Children with acute exposure
Children with chronic exposure
General population of mothers
and infants
Border area infants
Young children and juvenile
animal research
Spanish speaking children and
families
Health care providers
Time Frame
4-5 yrs. to conduct;
2 yr. pilot
2-3 yrs.
2-2.5 yrs.
lyr.
3-6 mo.
< 3 mo.
1 yr.
lyr.
134
-------
Outcome measurement: Start with the prenatal maternal blood and urine measures to further
identify (+) and (-) mother-infant pairs. Follow all (+) mothers and selected (-) mothers (this
assumes that + mothers will be less common). Assess infant development using a standard battery
of tests (see discussion under health endpoints). Neonatal tests should be conducted to assess
alertness, habituation, reflexes, etc.. Developmental testing should occur at the following intervals,
months 4, 9,12,15, 18, 24, 30, 36, 42. Tests at 12 and 15 months are anticipated to be particularly
informative. House dust and other exposure measures should be tied to developmental testing
whenever possible. It would also be possible, within this design, to assess a number of other
reproductive outcomes including spontaneous abortion, preterm delivery, and birth weight.
Overall strengths: Longitudinal follow-up beginning as close to conception as possible
with repeated exposure and developmental measures could address the main research question
definitively.
Overall weaknesses: More information is needed to be able to design this study efficiently,
such as: how to identify truly exposed versus unexposed (or high versus low) women and children
for study; how to maximize the relevance of repeated exposure assessments in a variety of media;
which developmental tests will be more sensitive to the effects of pesticides, etc.
Time frame: Approximately 4-5 years to conduct. Needs extensive piloting prior to
beginning the study (2 years?).
Analytic Studies - Poisoning Case Series
Study design: Case series.
Level of inquiry: Formal test of hypothesis.
Main hypothesis: There are long-term neurobehavioral and neurodevelopmental sequelae of
acute pesticide exposure that persist over time.
Secondary aim: Characterize the natural history of acute poisoning and determine
neurobehavioral and neurodevelopmental presentation of poisoned children. Children could also be
followed for alterations in sexual maturation which may be associated with exposure.
Target (referent) population: Age-specific general population on whom the norms for
developmental tests are based.
Study group: Acutely poisoned children identified through poison control centers and/or
hospital emergency room visits/admissions. Enroll children reported to be poisoned who are aged 0
to 42 months. Interest was expressed in including Mexican children who are reported to US health
care facilities.
Exposure assessment: Based on poisoning reports (medical records, etc.). There is a
potential to add followup pesticide exposure characterization to ensure that observed effects are
sequelae of the poisoning incident and not due to ongoing exposure.
Outcome assessment: Conduct a baseline assessment within two weeks of reported
135
-------
poisoning episode. The sequence of developmental followup testing will be based on age at
enrollment and use a standard battery of tests (see discussion under health endpoints). Depending
on age at enrollment, testing should occur at the following intervals, months 4, 9, 12, 15, 18, 24, 30
36, 42. Tests at 12 and 15 months are anticipated to be particularly informative.
Overall strengths: Description of the natural history and potential long-term neurobehaviora
and neurodevelopmental sequelae of acute poisoning is needed. Longitudinal followup can assess
effects over time. This study may help further refine the developmental endpoints which are most
sensitive to pesticide exposure and help to define the testing time frames and protocols which best
discriminate the effects of pesticides on child development. Studying poisoned children as sentinels
may provide information on geographic areas and/or behaviors of high risk children. Modeled after
occupational studies like those of Zweiner.
Overall weaknesses: Need to pilot case ascertainment process. Comparison to general
population norms may be inappropriate. Dependent on our ability to develop a "fast response" team
or individual psychologist who can travel to testing sites. The assumption is that information from
acute poisoning provides information on potential effects of chronic low dose exposure - may or
may not be true.
Time frame: Collect cases within the first year and follow all children at least one year or
until age 42 months. Total time will be less than two years.
Analytic Studies - Prospective Nested Case-Control Study of Symptomatic Children
Study design: Nested case-control within a cohort of symptomatic children seen at a health
care facility (or multiple sites?). Symptoms of pesticide poisoning are strikingly similar to the flu.
Level of inquiry: Formal test of hypothesis.
Main hypothesis: There are no developmental differences between symptomatic children
with (+) pesticide urine screens and symptomatic children with (-) pesticide urine screens.
Secondary aim: Characterize the prevalence of nonspecific illnesses treated in a clinical
setting which are likely due to pesticide exposure.
Target population: Children with chronic low dose exposure to pesticides. We assume
children with chronic exposure will be more likely to present with clinical symptoms than children
who are reported to poison control centers. No data are currently available to confirm this
assumption.
Study group: Symptomatic children who present in a clinical setting for treatment. "Case
definition" will be developed based on the flu-like symptoms associated with pesticide exposure.
All children aged 2-3 1/2 years who fit the definition will attempt to provide a urine sample for
pesticide screening and possibly a finger-stick blood sample for cholinesterase testing. Children
with (+) screens will be compared to selected children with (-) screens.
Exposure measurement: Urine at the time of presentation will be screened for pesticides.
Turnaround time for the initial screen will be a few weeks. Finger stick blood samples may be
136
-------
available for measurement of cholinesterase and related enzymes. Subsequent to enrollment into
case-control study (+/- screen), house dust and urine samples will be collected at a baseline
approximately four-six weeks after clinic visit. Quarterly house dust and urine samples will be
collected after the baseline visit. Possible "complete characterization" of selected (+) children to
determine likely pathways of exposure including: air, duplicate diet and other environmental
measures.
Outcome measurement: Establish a case definition based on symptomatology. Assess case
(+screen) and control (- screen) children within six weeks of index clinic visit. For children under
42 months, use the standard tests described above. For children over 42 months, more complex
testing could be done. Developmental testing should be timed with environmental and biological
sample collection. Medical records can be reviewed to examine previous growth information and
look for differences between cases and controls. Maternal education, socioeconomic status and
access/utilization of health care are important confounders.
Overall strengths: Identifies children for study on the basis of "high risk" presentation.
Allows for longitudinal follow-up of children with symptoms related to pesticide exposure who
were not identified as "poisoned". As a companion to the poisoning case-series, this study can
identify factors which may predict whether or not the pesticide exposure experienced by a child will
come to the attention of health care providers, parents, caregivers.
Overall weaknesses: Dependent on clinic staff to identify cases, provide informed consent
and collect urine and blood samples. Should we provide an incentive? Need a quick turnaround
time from laboratory for urine analysis. Need a "fast response" team/individual similar to
poisoning project. Older "young" children studied, don't have information on prior exposures.
Time frame: Training of physicians, medical staff should be done first (see below) - three to
six months. After that, collect cases for six-nine months and follow cases and controls for at least
one year. Total time - approximately 2 -2 1/2 years.
Descriptive Studies - Correlation Between Maternal and Infant Biologic Samples
Study Design: Cross-sectional.
Level of Inquiry: Feasibility pilot.
Primary aim: Describe the correlation and variability between pesticide levels (as well as
cholinesterase and related enzymes) in maternal and infant biologic tissue samples in the general
population.
Target (referent) population: General population of mothers and infants.
Study group: NTD control group in the Texas Department of Health NTD study.
Approximately 140 general population control mother-infant pairs have provided multiple biologic
specimens. Another potential study group is the population-based controls from a University at
Texas - Houston, School of Public Health which has prenatal as well as postnatal samples from
approximately 120 mother-infant pairs.
137
-------
Exposure measurement: A variety of tissues are potentially available including maternal,
fetal and infant. Some samples are banked and some have already been analyzed (possibly
including pesticide data). Potential to collect house dust samples at the enrollment of new control
subjects.
Outcome measurement: No health effects measured.
Overall strengths: Fills an important data gap. Will help us determine the appropriate
biologic specimens to collect to adequately assess exposure. Data can be used to refine a biologic
sampling strategy.
Time frame: Depending on what has already been done, could have results in about one
year.
Descriptive Studies - CIS Studies of Infant Health Status
Study design: Cross-sectional.
Level of inquiry: Primarily descriptive but potential to test formal hypothesis.
Main hypothesis: Infant mortality and birthweight are not different in areas with high
agricultural pesticide use compared to geographic areas with lower agricultural pesticide use.
Target (referent) population: Infants bom in the border area.
Study population: Infants with live birth certificates filed in the border area (for the past
year? Multiple years?) for studies of birthweight. Infants who died in the first year of life and have
a death certificate linked to their birth certificate (most recent year available, usually a lag of 2-3
years; multiple years?)
Exposure measurement: Geographic mapping of pesticide usage including pounds per acre
of active ingredients. Attempts will be made to map the best available pesticide usage data in the
smallest area.
Outcome measurement: Vital registry data will be used to locate the place of residence at
birth. Birthweight and mortality will be assessed in high pesticide use areas compared to low
pesticide use areas.
Overall strengths: May help to identify high risk areas and provides baseline data on the
health status of the border population. It is a concern that national statistics may not apply as a
good reference population for the border. This exercise will provide data to address that concern.
All border states have developed compatible GIS systems and this project makes good use of that
capability.
Overall weaknesses: Pesticide use data is non-specific exposure data. No information on
individual matemal/patemal/infant exposure is available.
Time frame: Depending on the availability of data, approximately three-six months.
138
-------
Capacity Building - Pesticide Dose
Our assumption is that a summary of pesticide dose information in young children (and
juvenile animals) is needed. The summary should include LOAL, NOAL, RfD, and LD50 for a
variety of pesticides. This information would be very helpful in determining which pesticides to
measure in children (assuming exposure to multiple compounds) and provide some information
about susceptibility. The focus should be on pesticides known to be used or present in the border
area.
Time frame: Less than 3 months.
Capacity Building - Adaptation of Neurodevelopmental Tests
Our assumption is that many of the currently available neurobehavioral and
neurodevelopmental tests are inappropriate for use in the border population. Back translation is not
sufficient and tests should be adapted for use with Spanish speaking populations. These efforts will
increase the validity of studies of developmental endpoints in this population.
Time frame: approximately one year?
Capacity Building - Training of Border Physicians
Our assumption is that children suffering from the effects of pesticide exposure are not
identified by health care workers or their parents/caregivers as having been exposed to pesticides.
Training should be provided to physicians and health care providers in the border area. This
program has relevance nation-wide as well. It is possible to stagger implementation, to use pre and
post-testing to evaluate the effectiveness of the training and provide CME credits for training. This
exercise is an important precursor to our proposed nested case-control study based on a cohort of
sick children.
Exposure Assessment Issues Raised
Although the focus of the meeting was on the potential health effects which could be studied
in young children in relation to pesticide exposure, the measurement of exposure was a recurrent
theme. The following issues were raised by members of the developmental workgroup.
The difficulties of exposure assessment were a big focus of concern in various discussions.
Investigation of methods and standardization of exposure measurement was seen as a high priority.
Since the organophosphate pesticides were of particular interest, yet have a very short biologic half-
life, it is difficult to classify exposure with one measurement. Serial measurements and modeling
efforts to describe the relation between exposure and environmental levels of pesticides were
stressed. Environmental samples are likely to be more persistent, particularly indoors, and may
give an indication of usual or long-term exposure.
Housedust was perceived as a particularly useful media for exposure assessment. Many
more analytes can be measured in environmental media than can be measured in biologic samples.
Until the methods for biologic fluids are developed, thoughtful assessment of the environmental
samples can provide otherwise unavailable data on commonly used pesticides.
139
-------
The focus on acetylcholinesterase inhibitors, particularly common organophosphate
compounds was also the topic of discussion. While restricting our focus to one class of compound
or one mechanism of action may help to refine the study question, it is critically important that we
remember there are others which are operating also. Other mechanisms of action and other
compounds may have independent effects on the health outcomes we study and which may interact
with the exposures we choose to measure. Aggregate exposures and multiple pathways are
important issues.
Most of the biologic assays which were discussed are conducted on urine samples. The
difficulties of collecting urine from small diapered children and infants seem daunting but the
pediatricians think it is feasible to try. First morning voids are most problematic and may mean less
for young children who urinate during the night.
There is a need for more qualified laboratories that can analyze research samples (biologic
and environmental). Contract labs may not have the equipment and quality control procedures to
achieve a limit of detection appropriate for a non-occupational study. The few research level
laboratories available may not be able to keep up with the demands. Perhaps there should be a
regional laboratory program to fund (or otherwise support) facilities whose purpose is to support
epidemiologic exposure studies.
140
-------
Immunology and Pulmonary Report
Immunology Workgroup Members: Betsy Hilborn (facilitator), Susanne Becker (rappateur),
Lina Balluz, Donald Echobichon, David Camann, Anthony Homer.
Pulmonary Workgroup Members: Rebecca Calderon (facilitator), Jerry Blondell
(rappateur), Bob Bornschein, James Ellis, Maria Martinez, Mary Kay O'Rourke, Enrique
Paz, Jim VanDerslice.
On day one of the workshop, the immunology and pulmonary workgroups initially met as two
separate groups. Toward the end of the session that day, the two groups met together to discuss relevant
health endpoints identified by both workgroups. On the second day, the two groups met together to discuss
both the immunology and pulmonary study designs. For the purposes of this report, each group has a
separate report for day one (health endpoints) even though there may be some redundancy. For day two
there is a single report from the group that covers both immunology and pulmonary study designs.
Day 1: Health Endpoint, Pulmonary
The pulmonary workgroup discussed both the utility of validated disease endpoints and self-
reported symptomatology in assessing overall pulmonary health. Also, the group discussed physiologic
measurements for use in either a clinical or an epidemiologic setting in order to accurately classify
individuals' pulmonary health. The workgroup discussed four pulmonary disease classifications along with
an extensive list of symptoms and physiologic measurements. Signs and symptoms of high interest were
either suggestive or pathognomonic for each of the diseases. The issue of whether symptoms or disease
could be ascertained by a questionnaire or had to be diagnosed in a clinical setting was considered in
evaluating each disease. The four diseases and their measurement by symptomatology are summarized in
Table 1. Physiologic measurements that could be used are summarized in Table 2. For each of the
endpoints additional covariates were identified, whether the illness was acute, chronic or persistent and
special considerations related to diagnosis and presentation of the disease by age group. Finally the group
considered whether it was biologically plausible that exposure to pesticides could be associated with the
health endpoint of concern.
Table 1. Pulmonary Disease and Symptoms
Signs and
Symptoms
cough
wheezing
bronchorrhea
shortness of breath
cyanosis
congestion
malaise
fever
Upper Respiratory
Infections
X
X
X
X
X
X
Asthma
X-dry
X
X
X- rarely
Acute
Bronchitis
X- productive
X
X
X
X
X
X
Interstitial Lung
Disease
X
X- exertion
141
-------
Table 2. Pulmonary' Disease and Physiologic Measures or Diagnostic Tests
Physiology
Measurement
PEFV
PEFR
Spirometry
D2CO
Pulse oximetry
Swabs- microbial
Upper Respiratory
Infections
X
Acute Bronchitis
X
Asthma
X
X
X
Interstitial Lung
Disease
X
X
X
X
Upper Respiratory Infections (URIs). Many of the symptoms could be ascertained through
the use of a questionnaire. The incidence of these infections is highest in infants and declines to about
the same incidence as adults by six years of age. The majority of these infections are acute and
transient. Other important risk factors to consider in epidemiologic studies include: prematurity,
overcrowding, day care, seasonality, and nutritional status. The majority of the workgroup felt there
was a low probability that pesticide exposure would be related to an increase incidence of URIs.
Acute Bronchitis. Many of the same symptoms associated with URIs are also characteristic of
acute bronchitis. Similar to URIs many of these symptoms could be ascertained through
questionnaire. Unlike URIs the incidence of this disease increases as children become older.
Important risk factors to consider include: prematurity, overcrowding, day care, seasonality, and
nutritional status. The issue of biological plausibility split the workgroup between suspected
plausibility and a frank unknown.
Asthma. Many of the symptoms of asthma could be ascertained by a questionnaire but
asthma should be diagnosed by a physician. The condition of asthma is generally intermittent after an
acute onset. Asthma is a rare diagnosis in children under the age of three. After three years of age, the
incidence increases, peaking at about 10-12 years of age and then it begins to decline slowly in young
adults, ages 18-21. Important risk factors to consider in studies of asthma include: smoking (active
and passive), viral respiratory infections, genetic disposition, air pollution, breast feeding, allergens,
low socioeconomic status, and parental occupations. It was the overwhelming consensus of the
workgroup that there was a biologically plausible relationship between pesticide exposures and
asthma. The group had a small discussion about the difference between exacerbation and induction of
asthma. The workgroup felt that pesticides could be involved in causing an asthma condition as well
as exacerbating an existing condition.
Interstitial Lung Disease. Similar to asthma, signs and symptoms could be obtained by
questionnaire, but cases must be confirmed through a physician diagnosis. Physiologic measurements
(PEFV and spirometry) could be used in identifying possible cases or measuring severity of the
condition. This disease is rare in infants and children, since it is often the result of a long term
chronic exposure. Important risk factors to consider include: dust, air pollutants, parental occupation
142
-------
and other underlying disease. The majority felt there was a low probability that pesticide exposure
was related to interstitial lung disease.
Day 1: Health Endpoints. Immunology
In order of priority, the four diseases and methods for clinical and laboratory diagnosis are summarized in
Table 3.
Table 3. Priority Immunologic Health Endpoints and Physiologic Measures or Diagnostic Tests
Physiology
Measurement
Pulmonary function
tests
Total and/or specific
IgE
Skin testing
Provocation testing
Total Ig levels
Antibody titer response
to antigenic stimulation
Delayed type
hypersensitivity testing
T-cell subset measures
B-cell subset measures
Lymphocyte
proliferation assay
Patch testing
Physical exam
Medical records
Questionnaire
Asthma
X
X
X
X
X
X
X
Allergy
X
X
X
X
X
Immunodeficiency
X
X
X
X
X
X
X
X
X
Contact Dermatitis
X
X
X
X
1. Asthma (reactive airway disease) This health endpoint was of major interest to
researchers and clinicians who work along the border. The incidence of asthma is on the rise
with an estimated 30% of children experiencing at least one episode of wheezing during their
first 3 years of life. All ages and both sexes are potentially affected. Some children 'outgrow'
their asthma symptoms. There is some evidence that low socioeconomic status is associated with
higher rates of childhood asthma. Approximately 5% of the adult population was estimated to be
diagnosed with asthma.
Specific measures of disease were discussed. Pulmonary function testing was considered the
best clinical measure. This test has some limitations of use in the youngest age groups. Some
other clinical measures discussed were: measurement of total IgE, skin testing for allergies,
143
-------
allergen-specific IgE, and provocation testing. These tests would all require skilled clinical
and/or laboratory personnel to perform. The usefulness of data derived from questionnaires and
medical records was also discussed. Questionnaire administration would require some training.
2. Allergic diseases (allergic rhinitis, eczema, allergic broncho-pulmonary aspergillosis)
Allergies manifest themselves in multiple organ systems; the workgroup focused on the
respiratory and dermatologic systems as among the most commonly affected. Eczema is most
common in infants, but all ages are potentially affected. The sexes were considered to be equally
affected. There is some variability in the duration of allergic disease. Discussion focused on
allergic children and the fact that they may cease to have symptoms of allergy during mid-
childhood, but that these symptoms may reappear during later years.
Allergen skin testing is a valuable tool in the diagnosis of allergy. There was some
controversy in the group about the use sensitivity/usefulness of skin testing in children under 5
years of age. Other measures that would require clinically trained personnel include:
measurement of total IgE, physical exam and interpretation of findings. Questionnaires and
medical records may also be useful.
3. Immunodeficiency There was much interest in the possibility of pesticide exposure and the
development of primary or secondary immunodeficiency in an exposed child. Males are more
likely than females to experience primary immunodeficiency due to X-linked disease. It was
unknown if there was any modifying effect on the association between pesticide exposure and
immunodeficiency due to socioeconomic status or ethnicity.
Tests to assess immunologic competency include: total Ig levels, with a result less than 2
standard deviations below the mean was considered abnormal. Antibody titer response to
antigenic stimulation, T-cell and B-cell subset measures, a complete blood count with
differential, delayed type hypersensitivity testing, and lymphocyte proliferation studies were all
considered to be potentially useful measures of immune structure and function. These tests
would require a phlebotomist and trained laboratory personnel to perform and interpret the
assays. An accurate infectious disease history obtained during physical exam or questionnaire
interview may provide valuable information about functional immune status.
4. Contact dermatitis Contact dermatitis was considered of interest because it is common and
easy to measure in all ages, including infants. There is no known difference in prevalence by
age, sex, ethnicity or socioeconomic status. It is known to be associated with exposure to
phenoxy herbicides, fungicides such as maneb, mancozeb, and some 'inert' petrochemical
ingredients of pesticide formulations.
Contact dermatitis manifests as a rash that may be difficult to distinguish from eczema.
A trained clinician would be required to make the diagnosis. Patch testing is the method of
choice to confirm the diagnosis of contact dermatitis.
5. Autoimmune diseases- There was concern voiced by a clinician that works along the border,
that there is an increased number of lupus erythematosus cases being diagnosed in children living
along the border. The overall prevalence in the U.S. pediatric population is 'very low'. Females
are affected more frequently than males. African-Americans are affected more frequently than
Caucasians. The disease is usually diagnosed during adolescence, so this would not be a useful
144
-------
health endpoint in very young children. The disease is chronic, but symptoms may remiss, then
recur.
Diagnosis is made by documenting the presence of a number of signs and symptoms.
Clinical samples such as blood and urine may be examined for: anti-nuclear antibody, an
elevated erythrocyte sedimentation rate, an abnormal complete blood count with differential, and
a complete urinalysis with examination of sediment. A clinician is needed to perform a physical
exam, a phlebotomist to draw blood samples, and a laboratorian to analyze the clinical assays.
6. Inflammatory bowel disease (inflammatory colitis) Inflammatory bowel disease (IBD) was
discussed as a health endpoint believed to be related to a disorder of the immunological system.
There was discussion about the possibility of a genetic component in the development of IBD
because of familial and racial predilection in some disorders (e.g.: Crohns disease within some
Jewish families). It was unknown if there was a difference in prevalence based on gender or
socioeconomic status. IBD is a chronic disease of long duration; no one knew of any association
between pesticide exposure and the development of IBD.
Diagnosis of IBD is made by a trained clinician with history, physical exam, and
colonoscopy. A nonspecific test for blood in the stool (guaiac assay) may be may also be
performed by anyone with minimal training.
7. Infectious diseases During group discussions, a hypothesis was made that an individual with
immunocompromise would be more susceptible to and experience more frequent infectious
diseases. It was unknown if there would be any difference in rates in males and females, but it
was felt that children under the age of 2 would be more susceptible to infection. There was also
some discussion that children from families with low socioeconomic status may experience more
infections due to the potential lack of safe food, water, proper sanitation, and adequate nutrition.
Infectious diseases may be diagnosed by a variety of methods depending on the infectious
agent. In most cases a trained clinician would be required to perform a physical exam, and there
would be a need for clinical sample collection and analysis. Laboratory capabilities in
microbiology, microscopy, immuno-cellular techniques, and possibly molecular methods may be
needed. Questionnaires may also be used.
8. Adverse reproductive outcomes- The group also briefly discussed that imrmmopathology in
an adult female may contribute to adverse reproductive outcomes, ranging from fetal death to
prematurity. There are known associations of maternal race and some maternal behaviors with
adverse reproductive outcomes. We did not know of any association with exposure to
organophosphates.
This health endpoint could be extracted from medical records or by questionnaire so
minimal training would be required to assess reproductive history.
Day Two - Development of Strawman Study Proposals
The purpose of the day two breakout group was to generate a collection of potential
studies for Phase II implementation. Study hypotheses were generated based on the day one
discussion of health endpoints, and the speakers presentations on pesticide exposures and border
populations.
145
-------
A collection of potential Phase II (pilot) studies were generated under each health
endpoint domain, rather than designing a "Cadillac" study. Generally, the discussion was
organized around the study design rather than the endpoint (in some cases multiple endpoints
could be addressed in the same study design), but it was also appropriate to generate a series of
similar designs which require different measurement strategies (e.g., a series of cross-sectional
studies). Groups were not asked to develop a study design for every endpoint generated in the
previous days' discussion, but to try to be comprehensive and at least get some ideas about the
highest priority endpoints. The facilitators were asked to elicit information and encourage
discussion about the main hypotheses or aim of the study; the target population; the appropriate
study group; exposure and outcome assessment issues; strengths and weakness of the design; and
the time frame needed for the study.
Groups discussed the relative merits and feasibility of the studies generated and
prioritized the proposed studies. Ranking was based on the information gathered in the earlier
discussion. Groups were instructed to be mindful of the time frame for Phase II studies. If a
priority study was a longer term (larger scale, etc.) endeavor, could a pilot test be done at this
phase? Reports on the strawman proposals were made to the entire group on the morning of day
three in a plenary session.
Day 2: Proposed Immunology and Pulmonologv Studies
On the second day, the Immunology and Pulmonary groups were merged to discuss study
protocol development since the two immune-related diseases of most concern, asthma and
allergy were identified as the #1 and #2 health endpoint of concern by both groups. There was
some discussion about the diagnosis of asthma in young children, so the diagnosis "reactive
airway disease" was adopted to describe asthma-like disease in infants and young children.
The first issue to be discussed was of an ethical nature; individuals may be motivated to
participate in community studies by the promise of information about the health and exposure
status of the community as a whole. Communication of study results may evolve into a sensitive
issue. If results are released in an inappropriate manner, it may lead to concern and distress
beyond what is warranted. However, part way through the study, the data may indeed indicate an
association between exposure and health problems. Should the community be informed of
preliminary results, or should all the data be collected and analyzed before community
involvement?
Epidemiological studies require a long period of data collection, then analysis and
interpretation. The long lagtime between participation in a survey and feedback from
investigators may frustrate community members anticipating an immediate interpretation of
study results. Some participants felt that subjects should be kept informed on an ongoing basis to
avoid the appearance of neglect and disinterest. Another approach would be to communicate a
proposed time-line during the recruitment phase of the study, so that participants will have a
realistic expectation of when to expect an analysis of the study data. Any preventive methods to
reduce exposure discovered during the course of investigation should be communicated.
A second issue of concern was the use of adequate interview tools. Spanish should be the
native language of the interviewers. All health questionaires used should be available in Spanish
and proven to be appropriate for the health concerns of interest.
146
-------
Some study designs were proposed to evaluate the association between pesticide
exposure and immunological and pulmonary health effects. It was felt that a cross sectional
survey was needed in target communities along the border. This would serve as a pilot study of
data and sample collection, and form a basis for future studies with more specific objectives.
The approach was the refinement of existing questionnaires for evaluation of pesticide exposure
such as the National Exposure Health Assessment Survey, and the use of questions from pre-
existing questionnaires for airway disease and other health effects such as those developed by the
American Thoracic Society and the National Health Interview Survey. All of these are available
in Spanish and are proven appropriate.
Study Questions: Is there excess pulmonary or infectious disease in this community? What is
the distribution of pesticide exposure? Is exposure associated with excess pulmonary or
infectious disease in this community?
Study Design: Cross Sectional Study, questionnaire derived exposure combined with self-
reported health endpoints. Exposure assessment supplemented by GIS and some environmental
sampling.
Study population: The goal is to recruit children, less than 11 years of age. Survey
questionnaires will be administered to parents of the study population. Recruitment of study
population will occur through the Women Infant Children Program (WIC), clinics, day care
centers, and schools. It was suggested that highly exposed children could be recruited through
the Poison Control center. Benefits to study population; financial incentives could be offered in
exchange for study participation.
Questionnaires: The pesticide exposure survey should yield at least the following information:
occupation of parents, address, type of residence, presence of livestock or pets, proximity of the
home to an agricultural area, if yes, type of agriculture, use of pesticides in home, on property, on
pets or children, frequency of use, identity of applicator, home pesticide inventory, known
pesticide exposures, sources of food and drinking water.
Parents will be interviewed concerning children's history of and presence of asthma,
allergies, eczema, lupus erythematosus, respiratory tract infections, other infections, and
gastrointestinal problems.
Survey information may be combined with agricultural pesticide use data obtained from
the pesticide control boards and health departments. Currently the most comprehensive pesticide
usage data is available in California and Arizona.
Strengths of approach: The strengths of this approach are: 1) standard questionnaires are
available, 2) some questionnaires have Spanish versions that may be more useful in
predominantly Hispanic communities, 3) surveys can be an economically efficient way to collect
information, 4) large samples can be collected in multiple communities and 5) data can be used
to generate hypotheses.
147
-------
The study designs are summarized in Table 4: Proposed Immunologic/Pulmonary Epidemiologic Studies.
Table 4. Proposed Immunologic/Pulmonary Epidemiologic Studies
Proposed Study
Pilot study of
immunologic status and
development of infants
exposed to pesticide
Longitudinal study of a
birth cohort
Survey of border
familie
Case-control study
Case-control study of
children with hyper
reactive airways.
Level/ Design
Cross sections
Longitudinal study
Cross sectional study
Case-control
study(cases=exposed )
Case-control study
using methacholine
challenge
Target Population
Border children <1 year old
Border families
Border families
Border children <1 1 years of
age
Border children < 6 years
old.
Issues
Feasibility of studying infants and collecting
clinical samples
Loss to follow-up, resource intensiv
May study large number of subjects cost
effectively, but health endpoints not
validated. Potential for bias in exposure and
outcome measurement, May be useful to
develop hypotheses about health endpoints of
interest. Use of environmental samples to
assess exposure will increase cost of study.
Difficult to assess chronic exposure levels
resulting in mis-classification.
Risks to children with preexisting airway
disease associated with exposure to
methacholine. Exposure assessment costly.
Time Frame
2 years
5 years, minimum
2 years
2 years
2 years
Hypothesis 1: The prevalence of asthma and other diseases will be higher in individuals with increased pesticide exposure.
148
-------
Weaknesses of approach: The weaknesses of this approach are: 1) it is a survey, and only
one point in time is assessed, 2) cross sectional studies have limited ability to detect rare health
events or exposures and 3) conducive to mis-classification, and information bias.
Hypothesis 2: Pesticide exposure increases the incidence of and/ or exacerbates pre-existing
asthma.
Case-control study: The next study design to be discussed was a case-control study. A case
control study based on health outcome (asthmatics as cases), was proposed. Households would
be surveyed for pesticide usage and exposure, and the homes monitored for pesticide residues.
Doubt was raised that this design was appropriate since the development of asthma is known to
have numerous associations with the presence of other allergens such as house dust mites, cats,
cockroaches, and Alternaria spp. The group discussed designating cases and controls based on
exposure status.
Study design: Case-control study based on exposure. Those persons with highest pesticide
exposure would be designated as cases.
Study population: Children < 11 years of age.
The approach would be to select cases and controls based on exposure, and then evaluate
these groups for prevalence of asthma. High and low pesticide exposure households with
children < 11 years of age would be identified, for detailed pesticide sampling and monitoring.
Asthma prevalence in the households would then be assessed by questionnaires, and medical
records.
Strength of study: May evaluate specific health outcome hypothesized to be associated with a
rare exposure.
Weaknesses of study: The weaknesses are: (1) difficult and expensive to assess exposure
levels to assign case or control designation; 2) mis-classification of a problem, as pesticide
exposure is ubiquitous and current measures of exposure may have no correlation with past
exposure levels.
The group discussed another study related to the association between asthma and pesticide
exposure.
Hypothesis 3: Pesticide exposure contributes to airway hyper reactivity.
Study design: Case-control study based on response to a methacholine challenge test.
Study population: Children > 6 years of age.
A methacholine challenge test to objectively assess airway reactivity would be
administered to a group of healthy children. Children would be assessed for pesticide exposure.
Exposure history (possibly urinary metabolites, household samples) of children with airway
hyper-reactivity would be compared to those of children with no evidence of hyper-reactivity.
149
-------
Strengths of study: i) Specific, objective outcome measure, 2) strong biologic plausibility.
Weaknesses of study: 1) Difficult and expensive to administer, patient cooperation and trained
pulmonologist required; 2) retrospective exposure assessment by questionnaire imprecise,
urinary and household measures expensive and current measures of exposure may have no
correlation with past exposure levels.
A pilot of a longitudinal study was proposed to study immune system development in
infants.
Hypothesis 4: Pesticide exposure affects the development of the immune system in infants and
young children resulting in altered antibody response to vaccine administration and increased
incidence or severity of infectious disease.
Study design: Cross sectional study as a pilot for a longitudinal study of a birth cohort.
Study population: Children < 1 years of age.
Initially, a pilot study would be performed on approximately 30 subjects in each of four
groups; maternal/neonatal, four month, six month, and twelve month old groups. Subjects would
be recruited from a community health clinic. In the neonatal group, women near delivery would
be interviewed to assess pesticide exposure, samples of blood and urine would be collected. At
delivery, cord blood and placental samples would be collected to assess immune system function
and pesticide levels. Infants at the age of four, six, and 12 months would be assessed, mothers
would be interviewed and urine samples would be obtained at each assessment. Blood, urine,
hand wipes and household environmental samples would be collected at six and twelve months.
Questionnaires about health status and pesticide exposure can be given to the mother when the
child is seen in the clinic. Blood samples would be analyzed for indicators of immune function
such as antibody response to vaccination, T-cell immunity, leukocyte marker analysis, and T-cell
proliferative response to superantigen. The full-scale study would evaluate a cohort of children in
a longitudinal study design, would follow children longer than one year, and would incorporate
developmental assessments.
Strengths of study: 1) May evaluate specific immune system parameters in children at an age
when immune function is rapidly evolving; 2) may evaluate response to antigenic challenge in
the form of naturally occurring infection and planned vaccinations; 3) short-term exposure
history for an infant, with few sources of food and water and 4) crawling/mouthing infants have
greater potential for exposure.
Weaknesses of study: 1) Invasive assessment (blood draw) of health outcome, may be poorly
tolerated by the infant or parent; 2) longitudinal study requires geographically stable population
and 3) ethics of exposure results/interpretation/reporting need to be explored.
150
-------
CANCER WORKGROUP
Workgroup Members: Martha Moore (facilitator), Jim Quackenboss (rappateur),
Jonathan Buckley, Luis Escobedo, Debra Gilliss, Luis Ortega, David Camann, and
Judy Henry.
Introduction
In the United States, cancer is the second leading cause of death for children between
the ages of one and 14 years. Although overall cancer rates have generally been declining, the
rate of childhood cancer (specifically acute lymphoid leukemia, tumors of the central nervous
system and bone) has increased in North America. The cause of this increase is unclear, but
the possibility that it might relate to environmental exposure is real and should be investigated.
There is some evidence that parental occupation is associated with an increased risk for
childhood cancers. In addition to paints, solvents, radiation and hydrocarbons, parental
exposure to agricultural chemicals has been linked to cancer in children. A number of
epidemiological studies involving adults have shown an association between pesticide exposure
and cancers (see Buckley, this volume). It seems reasonable to postulate that pesticide
exposure would be of particular concern to children who are growing rapidly. Cancer is a
disease requiring cellular expansion as a part of its etiology. Thus, the normal growth
processes of fetal and child development provide the ideal setting for pesticides, that can
modify genetic material and processes, to cause the initiation and/or promotion of tumor
development.
Possible Strategies to Determine if Border Children Have a Higher Risk for Cancer
The Cancer Workgroup was tasked with discussing possible strategies to investigate
whether children, who live along the U.S.-Mexican border, might be at increased risk for
cancer. In approaching this issue, the group formulated the following questions: Does pesticide
"exposure" cause (or contribute) to an increased risk of cancer in children who live along the
border? Is there a difference in the pattern of childhood cancer and (or) pesticide usage
between border regions and non-border regions?
The workgroup outlined and discussed several possible types of studies including: (1)
using existing data bases; (2) performing an ecological study that would geographically
compare pesticide usage and cancer incidence; (3) performing a case-control study that would
identify cases and then determine if the cancers were associated with pesticide exposure; (4)
conducting a prospective cohort study that might link exposure to a biomarker and then to the
cancer; and (5) conducting a study that could link cancer-relevant biomarkers with pesticide
exposure. A series of questions were raised that could be discussed for each approach. These
included: (1) the size and characteristics of the population, (2) the existing data and current
ongoing studies, (3) existing infrastructure-groups and organizations already in place and
working in relevant areas, (4) appropriate strategies for exposure assessment, (5) confounders,
(6) study strengths, (7) study weaknesses, (8) time frame and cost, (9) probability of useful
outcome and (10) what questions might be answered.
151
-------
Use of Existing Databases
There are currently a number of cancer registries that can provide cases and incidence
rates for various cancers. Some state registries may already have information by county. There
are also two national cooperative clinical trial groups that are dealing with children's cancer—
the Children's Cancer Group (CCG) and the Pediatric Oncology Group (POG). Both groups
provide good resource information and possibilities for collaborative study. Using this
approach it would be relatively easy to obtain the relative proportion of cancer types and the
number of cancer cases that occur within any given geographic area. It is, however, difficult to
get reliable rates due to the small population sizes.
On the exposure side, there are or soon will be CIS maps and data that link agricultural
usage, crops and pests. There is also a very active and potentially helpful agricultural
extension service. These agents are very familiar with activities within their geographic
regions. Pesticide suppliers can also provide information. It is, however, hard to get actual
usage information from most border states. California is the one exception; it has a very good
reporting system. The real issue, however, is not pesticide use, but exposure levels attained by
individuals living along the border. This information is really not available from any existing
database.
Ecological Study
It would be possible to look at border and non-border regions and determine if there is a
difference in cancer patterns and/or pesticide usage. This might provide a relatively crude
ecological assessment. If both pesticide usage and cancer incidence is greater in the border
areas, that would provide some additional hypotheses for further testing. The biggest danger
from this type of approach would be the possibility of a cancer cluster. Upon further analysis
of this approach it was determined that this approach would give little, if any, definitive
information. If the study showed a positive association, it might support the need for additional
study. If the study showed no association, it would not rule out the need for a further study.
There was little enthusiasm for pursuing this approach.
Case-Control Study
Given that cancer is a very rare disease, the group discussed whether it would be
possible to do a case control study. The group estimated that there would be about 150 new
cancer cases per year in children aged 0-15 living in states along the border. Under the best
case scenario, it might be possible to find and enroll about 120 of these cases. Looking just at
leukemia incidence, it was estimated that there would be only about 35 cases per year. Cases
could be obtained from clinics and hospitals. However, with this small number of cases, the
group felt that it was unlikely that the study would have enough power to detect an association
with environmental factors.
The limited power of the study was further exacerbated when the exposure assessment
component was discussed. Three basic approaches to exposure assessment were discussed: (1)
use of questionnaires such as are used by the CCG, (2) environmental samples from the home,
particularly house dust, and (3) overall usage of agricultural pesticides. Past and current
152
-------
studies to assess exposure were briefly discussed. A Brownsville study extensively
evaluated 9 households over two seasons. NHEXAS is obtaining data in Arizona from 175
households. In addition, 100 households are being evaluated in Minnesota. None of these
studies have really answered the question as to how many and which sampling strategies should
be used. Given that the science of exposure assessment has yet to answer some of the basic
design questions, it seems prudent to wait on conducting a case control cancer study until such
issues can be adequately addressed.
At this juncture further issues that might make the study problematic were identified. It
was noted that the majority of cancer cases would likely come from urban rather than rural
areas. This would occur because the highest populations live in urban areas, particularly San
Diego. This fact further limits the possibility that this study design could evaluate the
association between agricultural pesticides and cancer. The number of people available for
sample would fall below the number required to make the analysis. Furthermore this design
requires that cancer cases be identified and then evaluated as to their possible pesticide
exposures. The time between case ascertainment and exposure evaluation means that
population mobility and lack of telephones would limit the number of cancer cases that could
actually be enrolled and evaluated. It was also noted that there would be confounders to this
study including socioeconomic status, infections and other potentially carcinogenic
environmental exposures.
As to what questions a case control study might address, it was decided that one might
see a weak association between cancer outcome and one or more possible measures of pesticide
exposure. A negative study would provide no information due to the low power of the design.
Thus, the utility of such a case control study in relation to the resources required would be very
low.
The group discussed current projects that might provide insight for the border
population. The two national children's cancer networks (CCG and POG) could provide useful
information and identify possible cases. For instance in the CCG study, there are
approximately 2000 cases of acute lymphoblastic leukemia (ALL) and 600 of acute
myeloblastic leukemia (AML). Some of these cases would be in border states. Participants in
the CCG study have completed extensive pesticide usage questionnaires. Thus the workgroup
recommended waiting for the results of the CCG and POG studies.
Following the El Paso Meeting, Dr. Debra Gillis obtained some of the cancer statistics
from the California Cancer Registry. This information is provided to further emphasize the
difficulty of obtaining enough border County cases to conduct a case control study for
childhood cancer. From 1988 to 1994, there were 47 childhood cancer cases in Imperial
County, California. Of these, 15 were leukemias (32%), 10 were central nervous system
cancers (21%) and 22 were other cancers (46%). In San Diego County, during this same time
period, there were 583 childhood cancers of which 204 were leukemias (35%), 114 were central
nervous system cancers (20%) and 265 were other cancers (45%). Thus, the pattern of the main
types of childhood cancers is similar between the two counties and is also similar to the rest of
the state. It is important to note that this is only 7 cases per year in Imperial County and 83
cases per year in San Diego County. It would therefore take a number of years to accrue
adequate numbers of cases for analysis.
153
-------
Biomarker Approaches
The low incidence of cancer makes it very difficult and expensive to assess whether
pesticide exposure contributes to its etiology. Perhaps a better strategy is to utilize cancer-
relevant biomarkers. The risk assessment paradigm relating exposure to disease is outlined
below:
Usage -»
Agric.
Resid.
Environ. -*
Cone.
Dust
Surface
Food
Air
Water
Exposure
"Contact"
Air-Inh.
Ingest.
Dermal
:
Activity
-4 Internal
Dose
Blood
Urine
-* Target ->
Dose
DNA
Adducts
Biol. -»
V(D)J
gene
Health
Effect & Disease
Diagnosis
mutation
The workgroup discussed the possible utility of studying the relationship between
pesticide exposure, biomarkers of exposure and biomarkers of biological effect. In the past,
researchers have used DNA adducts as a biomarker of exposure and various genetic markers
such as gene mutation. Unfortunately DNA adduct analysis requires the development of
specific techniques for each pesticide. It is difficult to use this approach in a more generic way.
Thus the workgroup recommended that the target dose biomarker not be used and that blood or
urine levels be directly associated with the biological effect biomarker (see below).
The genetic markers that have been used, although measuring the types of genetic
damage seen in the etiology of tumors, are not directly involved in the development of tumors.
More recently, new molecular approaches have been developed to quantitate genetic alterations
known to be involved in cancers. Some of these chromosome rearrangements are mediated by
V(D)J recombinase. The workgroup recommended that this marker be utilized for the
biological effect marker.
Env. -»
Cone.
"Dust"
Exposure •* Int. -*
Dose
Urine
Target
Dose
•4 Biol. -» Health/
Effect Disease
>VfDM >Dicf»acp
Process
The workgroup considered possible prospective and case control studies that might
incorporate the use of biomarkers. For the prospective study, highly exposed and unexposed
individuals would be identified and compared as to their biomarker frequency. By focusing
first on highly exposed individuals it should be possible to determine the utility of the
approach. Such a strategy is already being followed by one panel member, Dr. Moore.
Ideally, one would want to conduct a prospective study that could determine whether
the biomarker predicted the development of cancer. Unfortunately, the number of samples that
154
-------
would need to be taken and achieved, the time required for tumors to develop and be
ascertained, and thus the cost of the study, make such an approach prohibitively expensive.
The case control design would identify children with leukemia and ask if they have a
higher level of V(D)J recombinase-mediated chromosomal rearrangements. The cases would
then be evaluated as to their exposure status. Dr. Buckley, one of the panel members, is
currently pursuing such an approach.
Both designs have issues with confounders. In particular, there would be a number of
mutagens in addition to possible pesticides that might cause an increase in V(D)J recombinase-
mediated rearrangements. It is also possible that V(D)J recombinase activity levels might be
under genetic control and that some individuals might be inherently more susceptible to such
rearrangements. The normal distribution of V(D)J recombinase activity has not been defined.
The workgroup identified several strengths of this biomarker approach. It seems likely
that these types of studies would provide useful information. These genetic markers are non-
specific for exposure and therefore might be used to provide insight into cumulative biological
effects and thus cumulative risk and disease. Furthermore, the biomarker approach can be
applied to a relatively small (compared to cancer assessment) number of cases.
The weakness of this specific biomarker approach is obvious. It is tied to the V(D) J
"model" for cancer etiology. In addition, all of the previously described issues concerning
exposure assessment apply equally to the biomarker studies. However, a careful coupling of
biomarker analysis with exposure analysis might lead to useful conclusions as to the
appropriate exposure assessment approach.
Conclusion
Reflecting upon the problems associated with the above approaches, the workgroup
focused on the scientific issues underlying them. In all cases, the questions associated with
exposure assessment compromised the conclusions that might be drawn from the study.
Although databases exist, it is extremely difficult to access even the most fundamental pesticide
usage information by geographic region. As already noted, the usage information provides little
insight as to the level of exposure that either individuals or groups of individuals might receive.
The workgroup, four of whom were representative of the border states, concluded and strongly
recommended that the issues associated with proper exposure assessment be solved prior to
conducting an analysis of health outcome. This would apply both to cancer and non-cancer.
Under the best of circumstances, exposure assessment is difficult. In addition, there are some
issues that are "unique" to the border. There are a number of pesticides that are legal for use in
Mexico that are not allowed in the U.S. Pesticides from Mexico readily find their way across
the border, often without proper labeling as to content. Thus there are products that are used in
border communities that are not used in other parts of the U.S. There is essentially no
information on these pesticides and their usage levels.
155
-------
The workgroup concluded that the weakest part of the analysis rests with the exposure
assessment. Therefore the group strongly recommended that resources be focused first on
improving approaches to exposure assessment. Also, other efforts are already underway
investigating childhood cancers, developing databases and evaluating approaches using
biological biomarkers. The workgroup recommended waiting for these researchers to complete
their work. Once the exposure assessment can be more adequately conducted, and the
information from the cancer studies is available, it should be possible to revisit and make
recommendations concerning studies to investigate the association of childhood cancers and
exposure to pesticides.
Day 3: Group Discussion
The focus was on themes common among workgroups: efficient methods to screen for
pesticide exposure, questionnaire development and validation, environmental sampling, and
validation of biochemical measures of exposure. There is a need for information about the
range of pesticide exposures in this pediatric population. Many outstanding questions remain:
how well correlated are current methods of environmental exposure measurement and actual
absorbed dose? What is the clinical relevance of metabolite or cholinesterase levels detected in
biological samples? What is the population distribution of exposure endpoints for each agent of
interest? What are the determining factors that result in a portion of the population being
categorized in either 'tail' of the exposure distribution? What are the resources available for
performing studies in the border region? What are the difficulties that will be encountered in
attempting to study this specific population? What, if any, is the value of studying acutely
poisoned individuals, and can this information be extrapolated to persons with chronic, low-
dose exposures?
Individual workgroups presented proposed studies for group discussion. The
neurobehavioral workgroup raised the question of what the next step would be if a cross-
sectional study failed to show any health effects associated with high levels of pesticide
exposure. The group suggested initiating the investigation into health effects by studying those
children with a history of acute pesticide poisoning. If those children display altered
neurolobehavioral function in some area, then at least a health endpoint of interest has been
identified, one that can be examined in children with lower dose or chronic exposure. The
group raised the issue of there being a scientific need for long term follow-up of those children
who have experienced an episode of poisoning; it is unknown if neurobehavioral changes occur
and how they may change over time. A cohort of children with repeated neurobehavioral
assessment during a longitudinal study was proposed. Participants suggested a cohort could be
derived from clinics or a Women Infants and Children program group.
The combined immunology/pulmonology workgroup presented three proposed study
designs. A cross-sectional study of border populations would provide essential information
about demographic, occupational, geographic charactaristics, and pesticide use. Techniques
such as household pesticide inventory, environmental and biologic sample collection for
pesticide exposure could be pilot tested. One goal would be to validate questionnaire-derived
exposure data through the use of pesticide inventories and environmental and clinical sample
collection. Based on the hypothesis that chronic pesticide exposure might increase airway
reactivity, a cohort study was proposed. A group of children with quantified pesticide exposure
156
-------
would be recruited. Children would undergo a methacholine challenge test and pulmonary
function testing results from the high and low exposure groups would be compared. Another
study was proposed to study immune development in infants. Initially, a pilot study would be
performed: women near delivery would be interviewed to assess pesticide exposure. Samples of
blood and urine would be collected. At delivery cord blood samples would be collected to
assess pesticide levels. Infants at the age of four, six, and 12 months would be assessed, urine
samples would be obtained at each assessment. Blood, urine and household environmental
samples would be collected at six and twelve months. Blood samples would be analyzed for
indicators of immune function such as antibody response to vaccination, leukocyte marker
analysis, and T-cell immunity. Health histories would be collected at each assessment to
evaluate the frequency of infection. The full-scale study would evaluate a cohort of children in
a longitudinal design, would follow children longer than one year, and would incorporate
developmental assessments.
The developmental workgroup proposed three study designs. Studies would be
composed of a prenatal cohort with a prospective study of infant development, a case series of
acutely poisoned children, and a prospective study of toddlers that would incorporate
developmental assessment with questionnaire and environmental sampling to assess exposure.
Some of the study questions discussed included: what is the relationship between maternal and
fetal blood levels of pesticides, associations between geographic information system (GIS)
exposure assessment and infant birth weight and mortality. The possible use of serum
cholinesterase levels as a measure of exposure was discussed.
The cancer endpoint workgroup raised the question: is there a difference in patterns of
pesticide usage between border and non-border regions? One approach would be to examine
cancer registries to determine the relative proportion of specific cancer diagnoses in different
geographic regions. One drawback of utilizing registries is the problem of under reporting and
the difficulty of assessing the true rates of illness. Another approach would be to conduct an
ecologic study to determine the association between rates of reported illness and GIS data on
agricultural pesticide usage and residential proximity to agricultural areas. One risk associated
with this approach would be the possibility of discovering an apparent 'border-related cancer
cluster' which may create difficult community communication issues. Another approach would
be to perform a case-control study using reported leukemia cases over a 2 year period.
Approximately 70 cases would be expected during this time period. Exposure would be
assessed by questionnaire, diet and pesticide usage assessment, household sampling, and
occupation. Problems with this approach include: recall bias, confounding exposure factors,
possible weak association between leukemia and pesticide exposure, high cost, and possible
lack of telephones to facilitate contact. Two potential approaches for pilot study include: a
retrospective cohort study among those diagnosed with cancer, and a prospective cohort study
among those persons categorized as having experienced 'high exposure levels'. A possible
source of confounding would be other environmental mutagens.
A comment from a member of the assembled group: the border community is requesting
better exposure measurements. Local physicians would like to know the potential health effects
associated with pesticide exposure, and to know if current levels of pesticide exposure are
causing health problems.
157
-------
A group discussion was held after the workgroup presentations. The following section is
a record of the comments raised in the course of discussion.
1. Participants raised the issue that questionnaire validation was very important for these
studies, that it is a difficult process to assess the health effects of mixtures of toxins.
2. Pathway analysis may be useful, but would be difficult to perform with multiple agents.
Additionally, pesticides are semi-volatile; they may be found in multiple media.
3. The likelihood of questionnaire data correlating with exposure is low. We must focus on
environmental and biochemical measures of exposure to increase precision of exposure
assessment.
4. This is state-of-the art work. We may never achieve a linear correlation. More
sophisticated modelling techniques may be needed.
5. In our work with measuring exposure to chlorpyriphos, we had 100% of subjects with
measurable pesticide metabolites. How do we start to separate out the proportional
contribution by media in assessing the source(s) of exposure?
6. We can't afford to analyze a large number of samples for the presence of pesticides
when most of them will have no detectable levels.
7. We need to establish 'normal' ranges for health endpoints; how else can we interpret
data? We need to consider both exposed and affected persons in the development of
ranges. Yes, the 'tails' of ranges are rich in information.
8. It's essential to have a control group.
9. We need a lot of subjects (a high 'N'), not just a control group. We want to measure the
whole range of 'normal' endpoints.
10. We know so little about the health effects of environmental contaminants here. The
population exposure levels are unknown, the health status of this population is unknown.
11. It may be worthwhile to gather theses, dissertations and other literature about border
populations that may contain both exposure and health information. There is not much
on Medline.
12. We can target small grant recipients, universities and academic departments in the area.
There is ongoing survey work along the border.
13. What is the value of acute poisonings? Answer: the clinical evaluation of poisoning
cases. We may need to look outside the border for these. There are databases of
poisonings, therefore one method may be to respond to an outbreak of poisonings , and
then to study the group longitudinally. For instance, there is currently a follow-up study
of methylparathion poisoning of children in the South.
158
-------
AGENDA
WORKSHOP ON THE ASSESSMENT OF HEALTH EFFECTS
OF PESTICIDE EXPOSURE IN YOUNG CHILDREN
DAY 1 Sunday. December 7.1997
8:00 am REGISTRATION
8:30 am Welcome, Introductions, Workshop Objectives
Dr. Hal Zenick, U.S. EPA Co-chair, Environmental Health Workgroup
8:45 am Keynote Address: Issues in Pediatric Epidemiology - Dr. Robert
Bornschein
9:30 am Review of Methods Available for Assessing Toxicity in Young Children -
NEUROLOGY - Dr. David Bellinger
10:15 am IMMUNOLOGY - Dr. Anthony Homer
11:00 am COFFEE BREAK
11:45 am DEVELOPMENTAL - Dr. Antolin Llorente
12:30 pm LUNCH
1:30 am PULMONARY - Dr. Maria Martinez
1:45 pm CANCER - Dr. Jonathan Buckley
2:30 pm Charge to Workgroups - Dr. Rebecca Calderon
Workgroup Breakout (Neurobehavioral, Immunology, Developmental,
Pulmonary, Cancer)
4:30 pm Plenary Session: Workgroup Reports
6:30 pm Adjourn
Dav 2 Monday. December 8.1997
8:30 am Keynote: Health Effects of Pesticides - Dr. Donald Ecobichon
9:15 am Increased Sensitivity to Pesticides in Young Children: Possible Mechanisms
-Dr. Stephanie Padilla
10:00 am Design of Children's Pesticide Exposure Survey - Dr. Jim Quackenboss
159
-------
10:30 am BREAK
10:45 am Pesticide Use near the U.S.-Mexico Border - Dr. Gary Robertson
11:15 am Pesticide Use and Assessment along the Arizona Border - Dr. Mary Kay
O'Rourke
12:00 pm LUNCH
1:00 pm Issues in Studying Border Populations - Dr. Jim VanDerslice
1:45 pm Cultural Considerations in Conduct of Epidemiolgic Studies - Dr. Robert
McConnell
2:15pm Dr. James Ellis
3:00 pm Charge to Breakout Groups - Dr. Rebecca Calderon
Breakout Groups for Development of Strawman Study Designs
(Neurobehavioral, Developmental, Immunology-Pulmonary, Cancer)
6:00 pm Adjourn
DAY 3 Tuesday. December 9.1997
8:30 am Reports on Strawman Proposals
NEUROBEHAVIORAL
9:00 am IMMUNOLOGY-PULMONARY
9:30 am DEVELOPMENTAL
10:00 am CANCER
10:30 am Closing Remarks and Discussion
11:00 am Adjourn
160
-------
List of Attendees/Contributors
Name
Gerry Akland
Lina Balluz
Susanne Becker
David Bellinger
Jerome Blondell
Robert Bornschein
Asa Brad man
Jonathan Buckley
Theresa L. Byrd
Rebecca Calderon
David Camann
A. Clayton
R. J. Dutton
Address
Analytical and Chemical Studies
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, N.C. 27709
Centers for Disease Control and Prevention
NCEH/EHHE/HSB
4770 Buford Hwy
F-46
Atlanta, GA 30341
U.S. EPA
Human Studies Division
MD-58D
Research Triangle Park, NC 27711
Neuro Epidemiology Unit
Children's Hospital
Mail Stop CA-503
300 Longwood Avenue
Boston, MA 02115
U.S. EPA
Health Effects Division (7509C)
401 M Street, SW
Washington, DC 20460
Department of Environmental Health
University of Cincinnati
3223 Eden Avenue
Cincinnati, OH 45267
Department of Environmental Health
Sciences, School of Public Health
7th Floor, University Hall MC-7360
University of California at Berkeley
Berkeley, CA 94720
Norris Cancer Center
MS-44
1441 East Lake Avenue
Los Angeles, CA 90033
University of Texas
School of Public Health at El Paso
1100 North Stanton, Suite 110
El Paso, TX 79902
U.S. EPA
Human Studies Division
MD-58C
Research Triangle Park, NC 27711
Southwest Research Institute
Post Office Drawer 28510
San Antonio, TX 78228-0510
Research Triangle Institute
Post Office Box 12194
Research Trianlge Park, NC 27709-2194
Director, Office of Border Health
Texas Department of Health
1100 W. 49th Street
Austin, TX 78756
Telephone/E-mail
919/217-2594
919/217-2591 fax
akland@rti.org
770/488-7353
770/488-7335 fax
Iib7@cdc.gov
919/966-0676
919/966-6271 fax
Becker.susanne@epa.gov
617/355-6565
617/734-6527 fax
703/305-5336
703/305-5147 fax
Blondell.jerry@epa.gov
513/558-0996
513/558-4838 fax
510/528-8319
510/642-5815 fax
abradman@socrates.berkeley.edu
213/764-0431
213/764-0143 fax
jbuckley@hsc.usc.edu
915/747-8504
915/747-8512 fax
tbyrd@utep.edu
919/966-0617
919/966-0655 fax
Calderon.rebecca@epa.gov
210/522-3649 fax
dcamann@swri.edu
919/541-6392
919/541-5966fax
clayton@rti.org
512/458-7675
512/45 8-7262 fax
rjdutton@comm.tdh.state.tx.us
161
-------
Donald J. Ecobichon
James A. Ellis
Luis Escobcdo
N.C.G. Freeman
Debra Gilliss
Rebecca Gomez
Judith Henry
Stephen Hern
Elizabeth Hilborn
Anthony Homer
A. Kuukowski
Antolin Llorent
Andres M. Lugo
Maria Martinez
Rob McConnell
Suzanne McMaster
Queen's University
Dept. of Pharmacology and Toxicology
Kingston, Ontario
Canada
1205 Aurora Drive
El Centre, CA 92243
Border Health Office
P.O. Box 30001
Department B3BHO
New Mexico State University
Las Cruces, NM 80003
Environmental and Occupational
Sciences Institute
Post Office Box 1179
Piscataway, NJ 08855-1179
California Department of
Health Services
5900 Hollis Street, Suite E
Emeryville, CA 94608
Centro de Salud Familiar La Fe
700 S. Ochoa Street
El Paso, TX 79901
Centro de Salud Familiar La Fe
700 South Ochoa Street
El Paso, TX 79901
U.S. EPA
National Exposure Research Laboratory
P.O. Box 93478
Las Vegas, NV 89193-3478
U.S. EPA
Human Studies Division
MD-58-A
Research Triangle Park, NC 27711
Dept. of Preventive Medicine
Univ. of Southern California
Los Angeles, CA 90033
Minnesota Department of Health
Post Office Box 64975
St. Paul, Minnesota 55164-0975
Baylor College of Medicine
6621 Fannin Street, Suite 530
Houston, TX 77030
West Texas Poison Center
Thomason Hospital
4815 Alameda Avenue
El Paso, TX 79905
University of Arizona
College of Medicine
Respiratory Sciences Center
1501 N. Campbell Avenue
Tucson, AZ 85724
University of Southern California
School of Medicine
1540 Alcazar Street, CHP 236
Los Angeles, CA 90033
U.S. EPA
613/359-5510 phone & fax
760/352-7216
760/352-1028 fax
505/528-5156
505/528-6045 fax
732/445-0151
732/445-0116
nfreemanfijorchid. rutgers.edu
510/450-3818
510/450-3773 fax
dgillis@hw2.cahwnet.gov
915/545-7036
512/458-7222
512/45 8-7776 fax
jhenry@epi.tdh.tx.state.us
702/798-2691
702/798-2261 fax
Hem.stephen@epa.gov
919/966-0658
919/966-7584 fax
hilborn.betsy@epa.gov
619/543-6222 pager
619/29-3758 fax
Aahom@aol.com
651/215-0854
651/215-0975fax
kukowal@mdh-envh.health.state.mn.us
713/668-0494
713/770-3399 fax
llorente@bcm.tmc.edu
915/534-3800
915/534-3809 fax
520/626-7780
520/626-6970 fax
213/342-1593
213/342-3272 fax
rmcconne@hsc.usc.edu
919/541-3844
162
-------
Pauline Mendola
Martha Moore
Kathleen O'Rourke
Mary Kay O'Rourke
Luis Ortega
David Otto
Stephanie Padilla
Enrique Paz
National Health and Environmental Effects
Research Laboratory
MD-51A
Research Triangle Park, NC 27711
U.S. EPA
Human Studies Division
MD-58-A
Research Triangle Park, NC 27711
U.S. EPA
Environmental Carcinogenesis Division
MD-68
Research Triangle Park, NC 27711
University of Texas
School of Public Health at El Paso
1100 North Stanton, Suite 110
El Paso, TX 79902
EOH, University of Arizona
1435 N. Fremont Street
Tucson, AZ 85719
Arizona Border Health Office
Arizona Department of Health Services
3815 N. Black Canyon Hwy
Phoeniz, AZ85015
U.S. EPA
Human Studies Division
MD-58-B
Research Triangle Park, NC 27711
U.S. EPA
Neurotoxicology Division
MD-74-B
Research Triangle Park, NC 27711
Pan American Health Organization
6006 North Mesa, Suite 600
El Paso, TX 79912
mcmaster.suzanne@epa.gov
919/966-6953
919/966-75 84 fax
mendola.pauline@epa.gov
919/541-3933
moore.martha@epa.gov
915/747-8503
915/747-8512 fax
kathleen@utep.edu
520/626-6835
520/882-5014 fax
maryk@hrp.arizona.edu
602/230-5808
602/230-5959 fax
lortega@hs.state.az.us
919/966-6226
919/966-6367 fax
otto.david@epa.gov
919/541-3956
919/541-4849 fax
padilla. stephanie@epa.gov
915/581-6645
915/833-4768 fax
Rossanne Philen
E.D. Pellizzari
James Quackenboss
Gary Robertson
Brian Schumacher
P. Shubat
Centers for Disease Control
NCEH/EHHE/HSB
4770 Buford Hwy, F-46
Atlanta, GA 30341
and Prevention
Research Triangle Institute
Post Office Box 12194
Research Triangle Park , NC 27709-2194
U.S. EPA (HERB)
National Exposure Research Laboratory
P.O. Box 93478
Las Vegas, NV 89193-3478
U.S. EPA
National Exposure Research Laboratory
P.O. Box 93478
Las Vegas, NV 89193-3478
U.S. EPA
National Exposure Research Laboratory
P.O. Box 93478
Las Vegas, NV 89193-3478
Minnesota Department of Health
Post Office Box 64975
St. Paul, Minnesota 55164-0975
770/488-7299
770/488-7335 fax
rhp2@cdc.gov
919/541/6579
919/541/7208fax
edp@rti.org
702/798-2442
702/798-2261 fax
702/798-2691
702/798-2261 fax
robertson.gary@epa.gov
702/798-2242
207/798-2107
Schumacher.brian@epa.gov
651/215-0927
651/215-0975
pamela.shubat@health .state.mn.us
163
-------
C. Stroebel
Anne Sweenev
James VanDerslice
Minnesota Department of Health
Post Office Box 64975
St. Paul, Minnesota 55164-0975
University of Texas
School of Public Health
RASW1040
P.O. Box 20186
Houston, TX 77225
Office of Environmental Health
Assessment Services
Washington Department of Health
Post Office Box 47846
Olympia, WA 98504-7846
651/215-0919
651/215-0975fax
stroebec@mdh-envh.health.state.mn.us
713/500-9471
713/500-9442 fax
asweeney@utsph.sph.uth.tmc.edu
360/236-3183
360/236-2251 fax
javl303@doh.wa.gov
R.W. Whtimore
H.S. Zelon
Hal Zenick
Research Triangle Institute
Post Office Box 12194
Research Triangle Park, NC 27709-2194
Research Triangle Institute
Post Office Box 12194
Research Triangle Park, NC 27709-2194
U.S. EPA
National Health and Environmental
Effects Laboratory
MD-87
Research Triangle Park, NC 27711
919/541-5809
919/541-5985fax
rww@rti.org
919/541-5888
919/541-7198fax
hsz@irt.org
919/541-2883
919/541-4201 fax
zenick.hal@epa.gov
164
-------
Workshop Discussion Groups
Neurobehavioral
DAVID BELLINGER
REBECCA GOMEZ
DAVID OTTO
SUZANNE McMASTER
STEPHANIE PADILLA
ROSEANNE PHILEN
GARY ROBERTSON
Developmental
ASA BRADMAN
STEVE HERN
ANTOLIN LLORENTE
ANDRES LUGO
ROBERT McCONNELL
PAULINE MENDOLA
ANNE SWEENEY
R.J. DUTTON
HALE VANDERMER
Immunology
LINA BALLUZ
SUSANNE BECKER
DONALD ECOBICHON
DAVID CAMANN
BETSY HILBORN
ANTHONY HORNER
Pulmonary
JERRY BLONDELL
BOB BORNSCHEIN
REBECCA CALDERON
JAMES ELLIS
MARIA MARTINEZ
MARY KAY O'ROURKE
ENRIQUE PAZ
JIM VanDERSLICE
Cancer
JONATHAN BUCKLEY
LUIS ESCOBEDO
DEBRA GILLISS
JUDY HENRY
MARTHA MOORE
LUIS ORTEGA
JIM QUACKENBOSS
165
------- |