Proposed Final Draft v 1 Dated September 26, 2007 Do Not Cite or Quote

September 26, 2007

EPA-HSRB-07-03

George Gray, Ph.D.

Science Advisor

Office of the Science Advisor

1200 Pennsylvania Avenue, NW

Washington, DC 20460 

Subject:  June 27-29, 2007 EPA Human Studies Review Board Meeting Report

Dear Dr. Gray:

	The United States Environmental Protection Agency (EPA or Agency)
requested the Human Studies Review Board (HSRB) to review scientific and
ethical issues addressing: (1) Carroll-Loye Picaridin Mosquito
Repellency Protocol LNX-001; (2) ICR Picaridin Mosquito Repellency
Protocol ICR 1 A 044; (3) Acrolein; (4) 4-Amino Pyridine and; (5)
Antimicrobial Exposure Assessment Task Force (AEATF) and Agricultural
Handler Exposure Task Force (AHETF) Research Programs.  The enclosed
HSRB report addresses the Board’s response to EPA charge questions at
its June 27-29, 2007 meeting.  

A summary of the Board’s conclusions is provided below.  Specific
Board discussion addressing potential source of underestimation bias,
QA/QC controls, design of scenario-level sampling scenarios and
within-worker variability concerning the AEATF and AHEFT research
programs, in addition to the topics listed above, is provided in the
body of this report.

Carroll-Loye Picaridin Mosquito Repellency Protocol LNX-001

The protocol LNX-001 to study the efficacy of a cream formulation and a
pump spray formulation of picaridin for repelling mosquitoes is
sufficiently sound, from a scientific perspective, to be used to assess
the repellent efficacy of these formulations against mosquitoes.

The Board concurred with the assessment of the Agency that the protocol
LNX-001 submitted for review by the Board, if revised as suggested in
EPA’s review, would meet the applicable requirements of 40 CFR 26,
subparts K and L.  In addition, with the submission of the amended
protocol, the Board believed that the protocol meets the applicable
requirements of 40 CFR 26, subparts K and L.

ICR Picaridin Mosquito Repellency Protocol ICR 1A 044

If amended and consistent with the Board’s concerns and
recommendations, the protocol ICR 1A 044 studying the efficacy of two
aerosols formulations of picaridin for repelling mosquitoes would be
sufficiently sound, from a scientific perspective, to be used to assess
the repellent efficacy of these formulations against mosquitoes.

The Board concurred with the assessment of the Agency that the protocol
ICR 1A 044 submitted for review by the Board, if revised as suggested in
both EPA’s review and by the Board, would meet the applicable
requirements of 40 CFR 26, subparts K and L.

Acrolein

The Board concluded that the Weber-Tschopp et. al. study contains
information sufficient for assessing human risk resulting from potential
acute inhalation exposure. In addition, the study was sufficiently
sound, from a scientific perspective, to be used to estimate a safe
level of acute inhalation exposure to acrolein.

There was not clear and convincing evidence that the conduct of the
Weber-Tschopp et al. study was fundamentally unethical.  In addition,
despite the lack of adequate information to assess the affirmative, most
of the HSRB agreed that there was not there clear and convincing
evidence that the conduct of the study was significantly deficient
relative to the ethical standards prevailing at the time the research
was conducted.

4-Amino Pyridine 

The Board concluded that the three clinical studies,  Grijalva et al.
2003, Segal et al. 1999, and Van Diemen et al. 1993, were sufficiently
sound, from a scientific perspective, to be used to derive a point of
departure for estimating risk to humans from exposure to 4-AP.  Thus
considering the three studies, an estimate of the LOAEL of 0.07
mg/kg/day was determined.  The Board was reluctant to endorse the use of
a 5 mg/day (0.07 mg/kg-day) LOAEL, given the multiplicity of side
effects seen among patients receiving this dose, and the steep
dose-response curve of 4-AP.  Thus, the Board cautioned that this
conclusion comes with a degree of uncertainty and advised the Agency to
take such uncertainty into account when using the published information
to arrive at a point of departure for 4-AP. 

The Board concurred with the initial assessment of the Agency that for
each of these three clinical studies (Segal et al., 1999; Grijalva et
al., 2003; and Van Diemen et al., 1993), there was no clear and
convincing evidence that the conduct of the study was fundamentally
unethical, or that the conduct of the study was significantly deficient
relative to the ethical standards prevailing at the time the research
was conducted.

AEATF and AHETF Research Programs

The AHETF governing documents have provided the HSRB with a detailed and
thoughtful analysis of expected benefits and risks associated with the
conduct of human exposure monitoring. The Board concurred that study
participants would receive little if any direct benefit from
participation in these studies although formalizing the nature of safety
feedback provided to participants might increase the potential benefit.
The Board also agreed that growers and landowners will receive a modest
benefit through the use of the test substances at no cost, and that such
a benefit was not unreasonable. The Board recommended that the
particular arrangements for providing the test substance be outlined in
the individual protocols. Finally, the Board concurred with the AHETF
that the database developed from these studies will improve the quality
of risk assessments, and that this should be considered a valuable
societal benefit, provided that the data collected are accurate, or at
the least do not underestimate real-world exposures.

The Board recommended that AHETF pay more careful attention to the issue
of safety when asking participants to operate equipment with which they
do not normally work. In particular, AHETF should be more explicit about
the level of competency expected of workers when operating such
equipment. The Board commended the AHETF for developing clear guidelines
for stopping work based on a heat index. However, the Board concluded
that the approach described in the governing documents was not fully
protective of workers, and recommended that additional attention be
given to this matter, including consideration of a lower heat index
threshold for stopping work.

Addressing Potential Sources of Underestimation Bias

The HSRB was comfortable with not including concurrent biomonitoring in
the protocols.  In fact, the Board recommended that the use of
additional monitoring units was more appropriate than the inclusion of
biomonitoring in these programs.  

Design of Scenario-Level Sampling

The Board acknowledged the great complexity of study design development,
given the many variables associated with exposure and the practical
constraints that arise in the conduct of human exposure studies. The
Board commended the Agency and the task forces for their efforts in
developing a detailed discussion of sampling strategies. The Board
remained concerned that the number of variables is large, and that the
relative importance of these variables has not yet been defined
adequately. The Board recommended that the Agency develop a process
whereby the critical variables associated with exposure are ranked,
accompanied by an appropriate rationale and justification.

The major limitation of non-random sampling is that it provides no means
for estimating the error associated with any estimate based on the
sample. The exposure distributions based on this type of sample might or
might not be anywhere close to the true exposure distributions and there
is not way to tell if the results are representative or not. If the
estimated exposure distributions will be used for regulatory purposes it
is particularly important to base those estimates on samples that at
least approximate a random sample and that permit obtaining data-driven
estimates of uncertainty around quantities of interest. Error estimates
and other estimates relevant to determination of quantities in the AEATF
and AHETF reports are based on strong and un-testable assumptions. 

Statistical Justification of Number of Cluster 

As the sample size justification for the AEATF II program is based on
the ICC estimate from the AHETF program, it is recommended that the
AEAFF II program update the proposed sample size based on an analysis of
whether the ICC estimates agree with that from the AHETF program.

Subject Recruitment and Enrollment Issues

The Board agreed that the Governing Documents and associated Standard
Operating Procedures do include comprehensive and appropriate
protections for human subjects.  

The Board agreed that the handling of language differences is an area
requiring further refinement and is appropriate to protections for human
subjects.  Related issues include mechanisms to ensure understanding and
voluntariness in the consent process. 

Sincerely,

William S. Brimijoin, Ph.D. Vice Chair

EPA Human Studies Review BoardNOTICE

This report has been written as part of the activities of the EPA Human
Studies Review Board, a Federal advisory committee providing advice,
information and recommendations on issues related to scientific and
ethical aspects of human subjects research.  This report has not been
reviewed for approval by the Agency and, hence, the contents of this
report do not necessarily represent the view and policies of the
Environmental Protection Agency, nor of other agencies in the Executive
Branch of the Federal government, nor does the mention of trade names or
commercial products constitute a recommendation for use.  Further
information about the EPA Human Studies Review Board can be obtained
from its website at http://www.epa.gov/osa/hsrb/.  Interested persons
are invited to contact Paul Lewis, Designated Federal Officer, via
e-mail at lewis.paul@epa.gov.

	In preparing this document, the Board carefully considered all
information provided and presented by the Agency presenters, as well as
information presented by public commenters.  This document addresses the
information provided and presented within the structure of the charge by
the Agency.

United States Environmental Protection Agency Human Studies Review
Board Members

Chair

Celia B. Fisher, Ph.D., Marie Ward Doty Professor of Psychology,
Director, Center for Ethics Education, Fordham University, Department of
Psychology, Bronx, NY *

Vice Chair

William S. Brimijoin, Ph.D., Chair and Professor, Molecular Pharmacology
and Experimental Therapeutics, Mayo Foundation, Rochester, MN  

Members 

Alicia Carriquiry, Ph.D., Professor, Department of Statistics, Iowa
State University

Snedecor Hall, Ames, IA 

Gary L. Chadwick, PharmD, MPH, CIP, Associate Provost, Director, Office
for Human Subjects Protection, University of Rochester, Rochester, NY 

Janice Chambers, Ph.D., D.A.B.T., William L. Giles Distinguished
Professor, Director, Center for Environmental Health Sciences, College
of Veterinary Medicine, Mississippi State University, Mississippi State,
MS 

Richard Fenske, Ph.D., MPH, Professor, Department of Environmental and
Occupational Health Sciences, University of Washington, Seattle WA  

Susan S. Fish, PharmD, MPH, Professor, Biostatistics & Epidemiology,
Boston University School of Public Health, Co-Director, MA in Clinical
Investigation, Boston University School of Medicine, Boston, MA 

Suzanne C. Fitzpatrick, Ph.D., DABT, Senior Science Policy Analyst,
Office of the Commissioner, Office of Science and Health Coordination,
U.S. Food and Drug Administration, Rockville, MD 

Dallas E. Johnson, Ph.D., Professor Emeritus, Department of Statistics,
Kansas State University, 

1812 Denholm Drive, Manhattan, KS **

Kannan Krishnan, Ph.D., Professor, Département de santé
environnementale et santé au travail, Faculté de medicine, Université
de Montréal, Montréal, Canada *

KyungMann Kim, Ph.D., CCRP, Professor & Associate Chair, Department of
Biostatistics & Medical Informatics, School of Medicine and Public
Health, University of Wisconsin-Madison, Madison, WI  

Michael D. Lebowitz, Ph.D., FCCP, Professor of Public Health & Medicine.
University of Arizona, Tucson, AZ 

Lois D. Lehman-Mckeeman, Ph.D., Distinguished Research Fellow, Discovery
Toxicology, Bristol-Myers Squibb Company, Princeton, NJ  

Jerry A. Menikoff, M.D., Office of the Director, National Institutes of
Health, Bethesda, MD

Sean Philpott, PhD., MS Bioethics, Policy and Ethics Director, Global
Campaign for

Microbicides, Program for Appropriate Technology in Health, Washington
D.C.

Richard Sharp, PhD., Director of Bioethics Research, Department of
Bioethics, Cleveland Clinic,

Cleveland, OH

Human Studies Review Board Staff

Paul I. Lewis, Ph.D., Designated Federal Officer, United States
Environmental Protection Agency, Washington, DC 

* Not in attendance at June 27-29, 2007 Public Meeting

** Become HSRB member August 31, 2007  TABLE OF CONTENTS

  TOC \o "1-3" \h \z \u    HYPERLINK \l "_Toc175991099"  United States
Environmental Protection Agency Human Studies Review Board Members	 
PAGEREF _Toc175991099 \h  6  

  HYPERLINK \l "_Toc175991100"  INTRODUCTION	  PAGEREF _Toc175991100 \h 
9  

  HYPERLINK \l "_Toc175991101"  REVIEW PROCESS	  PAGEREF _Toc175991101
\h  14  

  HYPERLINK \l "_Toc175991102"  CHARGE TO THE BOARD AND BOARD RESPONSE	 
PAGEREF _Toc175991102 \h  15  

  HYPERLINK \l "_Toc175991103"  Proposed Carroll-Loye Picaridin Insect
Repellent Efficacy Study LNX-001	  PAGEREF _Toc175991103 \h  15  

  HYPERLINK \l "_Toc175991104"  Proposed ICR Picaridin Insect Repellent
Efficacy Study	  PAGEREF _Toc175991104 \h  21  

  HYPERLINK \l "_Toc175991105"  Completed Inhalation Study with Acrolein
  PAGEREF _Toc175991105 \h  29  

  HYPERLINK \l "_Toc175991106"  Completed Studies on the Therapeutic and
non-Therapeutic Effects of Administration of 4-aminopyridine	  PAGEREF
_Toc175991106 \h  37  

  HYPERLINK \l "_Toc175991107"  Design of Research on the Levels of
Exposure Received by Pesticide Handlers	  PAGEREF _Toc175991107 \h  42  

  HYPERLINK \l "_Toc175991113"  REFERENCES	  PAGEREF _Toc175991113 \h 
64  

 

INTRODUCTION

From June 27-29, 2007, the United States Environmental Protection
Agency’s (EPA or Agency) Human Studies Review Board (HSRB) met to
address scientific and ethical issues concerning:

   

A.  Proposed Carroll-Loye Picaridin Insect Repellent Efficacy Study
LNX-001 

EPA requires data from efficacy studies using appropriate insect species
to support claims of greater efficacy than have previously been
approved.  

EPA’s regulation, 40 CFR §26.1125, requires the sponsor or
investigator to submit to EPA, before conducting a study involving
intentional exposure of human subjects, materials describing the
proposed human research in order to allow EPA to conduct scientific and
ethics reviews.  In addition, EPA’s regulation, 40 CFR §26.1601,
requires EPA to seek HSRB review of the research proposal.  

In previous meetings the HSRB has reviewed and commented favorably on
several proposed insect repellent efficacy protocols to be conducted by
Carroll-Loye Biological Research, submitted by Dr. Scott Carroll.  Dr.
Carroll has submitted a proposal for new research to evaluate the
efficacy of two conditionally registered repellent products containing
the active ingredient picaridin.  The research protocol, identified as
LNX-001, describes a field study of the efficacy of the test
formulations against mosquitoes.  The proposal bears many similarities
to the protocols EMD-004, SCI-001, and WPC-001 that the HSRB has
previously reviewed.  

EPA has reviewed Dr. Carroll’s protocol and has concluded that, with
some required refinements, it appears likely to generate scientifically
sound, useful information and to meet the applicable provisions of the
EPA regulations in 40 CFR part 26, subparts K and L.  The sponsor wishes
to submit the data to EPA later this year to satisfy the requirement to
provide efficacy data imposed when it received a conditional
registration for picaridin.  In the interest of providing a thorough and
timely response to the proposal, and because EPA finds the protocol
generally meets applicable scientific and ethical standards, EPA is
presented this protocol for review at the Board’s June 2007 meeting.

B.  Proposed ICR Picaridin Insect Repellent Efficacy Study

EPA requires data from efficacy studies with human subjects to support
claims of efficacy of a new pesticide product intended to repel insects
that transmit human diseases.  

EPA’s regulation, 40 CFR §26.1125, requires the sponsor or
investigator to submit to EPA, before conducting a study involving
intentional exposure of human subjects, materials describing the
proposed human research in order to allow EPA to conduct scientific and
ethics reviews.  In addition, EPA’s regulation, 40 CFR §26.1601,
requires EPA to seek HSRB review of the research proposal.  

Dr. Niketas Spero has submitted a proposal for new research to evaluate
the efficacy of two new formulations of a skin-applied repellent product
containing picaridin, to be conducted by Insect Control & Research, Inc.
(ICR).  The research protocol, identified by Protocol ID
G0590307001A044, describes a field study of the efficacy of the test
formulations against mosquitoes. 

EPA has reviewed ICR’s protocol and has concluded that, with a number
of required revisions, it appears likely to generate scientifically
sound, useful information and to meet the applicable provisions of the
EPA regulations in 40 CFR part 26, subparts K and L.  The sponsor wishes
to submit the data to EPA later this year in support of an application
to register one or more new picaridin products.  In the interest of
providing a thorough and timely response to the proposal, and because
EPA finds the protocol can meet applicable scientific and ethical
standards, EPA is presented this protocol for review at the Board’s
June 2007 meeting.

C.  Completed Inhalation Study with Acrolein

In its reregistration program EPA reexamines the safety of previously
registered pesticides.  The Agency is currently reviewing pesticides
containing the active ingredient acrolein.  Acrolein is registered for
use as a biocide in agricultural and industrial water supply systems. 
It is also formed as a byproduct in various industrial processes and is
a component of cigarette smoke.

In a review of the published scientific literature, EPA identified a
study published in German in 1977 in which human subjects were exposed
to acrolein for various durations and at varying concentrations in an
inhalation chamber.  Researchers collected data on subjective irritation
sensations and on eye-blink and respiratory rates. The Agency intends to
use the results of this study in its hazard assessment to derive a
“point of departure” (POD) for assessing acute toxicity resulting
from acute exposure to this chemical.  

The Agency’s regulation, 40 CFR §26.1602, requires EPA to seek HSRB
review of an EPA decision to rely on the results of any study if the
research was “initiated before April 7, 2006, and the research was
conducted for the purpose of identifying or measuring a toxic effect.”
 EPA has reviewed the study, applying the standards in 40 CFR
§§26.1703 and 26.1704.  Those provisions state:

§26.1703  Prohibition of reliance on research involving intentional
exposure of human subjects who are pregnant women (and therefore their
fetuses), nursing women, or children.

Except as provided in §26.1706, in actions within the scope of
§26.1701 EPA shall not rely on data from any research involving
intentional exposure of any human subject who is a pregnant woman (and
therefore her fetus), a nursing woman, or a child.

§26.1704  Prohibition on reliance on unethical research with
non-pregnant, non-nursing adults conducted before April 7, 2006

Except as provided in §26.1706, in actions within the scope of
§26.1701, EPA shall not rely on data from any research initiated before
April 7, 2006, if there is clear and convincing evidence that the
conduct of the research was fundamentally unethical (e.g., the research
was intended to seriously harm participants or failed to obtain informed
consent), or was significantly deficient relative to the ethical
standards prevailing at the time the research was conducted.  This
prohibition is in addition to the prohibition in §26.1703.

The Agency’s reviews concluded that the data were scientifically sound
and that there was no clear and convincing evidence that the conduct of
the research was fundamentally unethical or significantly deficient
relative to the ethical standards prevailing at the time the research
was conducted.   Nor was there evidence to show that the subjects
included nursing or pregnant women or children.

D.  Completed Studies on the Therapeutic and non-Therapeutic Effects of
Administration of 4-aminopyridine

In its reregistration program EPA reexamines the safety of previously
registered pesticides.  The Agency is currently reviewing pesticides
containing the active ingredient 4-aminopyridine (4-AP).  4-AP is
registered by EPA as a bird repellent under the name Avitrol.  It has
also been investigated as a drug to treat various neurological diseases,
and was recently approved for the treatment of chronic functional motor
and sensory deficits resulting from Guillain-Barré syndrome.

In a review of the published scientific literature EPA identified three
studies in which human subjects were exposed to 4-AP to evaluate whether
it alleviated neurological symptom in patients with either spinal cord
injury or multiple sclerosis.  These clinical trials also report on the
non-therapeutic effects of 4-AP.  The Agency intends to use the results
of these studies to derive a point of departure for assessing the risks
to humans resulting from all potential durations of exposure–acute,
short term, intermediate or subchronic, and chronic exposure.

The Agency’s regulation, 40 CFR §26.1602, requires EPA to seek HSRB
review of an EPA decision to rely on the results of any study if the
research was “initiated before April 7, 2006, and the research was
conducted for the purpose of identifying or measuring a toxic effect.”
 EPA has concluded that the three studies with 4-AP are subject to HSRB
review under 40 CFR §26.1602.  The Agency reviewed the studies,
applying the standards in 40 CFR §§26.1703 and 26.1704.  Those
provisions state:

§26.1703  Prohibition of reliance on research involving intentional
exposure of human subjects who are pregnant women (and therefore their
fetuses), nursing women, or children.

Except as provided in §26.1706, in actions within the scope of
§26.1701 EPA shall not rely on data from any research involving
intentional exposure of any human subject who is a pregnant woman (and
therefore her fetus), a nursing woman, or a child.

§26.1704  Prohibition on reliance on unethical research with
non-pregnant, non-nursing adults conducted before April 7, 2006

Except as provided in §26.1706, in actions within the scope of
§26.1701, EPA shall not rely on data from any research initiated before
April 7, 2006, if there is clear and convincing evidence that the
conduct of the research was fundamentally unethical (e.g., the research
was intended to seriously harm participants or failed to obtain informed
consent), or was significantly deficient relative to the ethical
standards prevailing at the time the research was conducted.  This
prohibition is in addition to the prohibition in §26.1703.

The Agency’s reviews concluded that the data were scientifically sound
and that there was no clear and convincing evidence that the conduct of
any of the research was fundamentally unethical or was significantly
deficient relative to the ethical standards prevailing at the time the
research was conducted.   None of the studies included as subjects
nursing or pregnant women or children.

	

E.  Background Materials Relating to the Design of Research on the
Levels of Exposure Received by Pesticide Handlers

Under FIFRA, EPA requires that all pesticide products must be
“registered” before they may be sold or distributed in commerce. 
The applicant for registration has the burden of demonstrating that its
pesticide will not cause “unreasonable adverse effects on the
environment.”  Among other potential risks, EPA requires applicants to
provide information that allows EPA to assess the potential for adverse
effects on people who mix, load, or apply a pesticide (referred to as
pesticide “handlers.”)  Accurately characterizing handlers’
potential exposure is essential to EPA’s risk assessment and
regulatory decision-making.

EPA currently relies on a collection of exposure studies mostly
contained in the Pesticide Handlers Exposure Database (PHED) to develop
estimates of handlers’ potential exposure.  When dealing with
pesticide that have low volatility, EPA assumes that, if field data are
corrected for chemical-specific losses under field conditions, the
amount of exposure a handler receives is independent of the chemical
composition of the pesticide he is using, and that his exposure depends
on the amount of active ingredient handled, as well as the particular
activity, the particular type of pesticide formulation and the
particular type of equipment used.  The Agency uses the PHED data to
develop estimates of “unit exposures” – expressed as an amount of
exposure per amount of active ingredient handled – for specific
scenarios.  (A scenario is defined by the activity, formulation, and
equipment, e.g. applying a liquid formulation by using airblast
equipment in an open cab.)  Using this information, EPA estimates
handlers’ potential exposures for each use of a pesticide and compares
those levels with toxicity data.  If the comparisons show potential
exposure is acceptably low, EPA concludes there is no risk to handlers. 
If, however, the comparisons show that in some scenarios a handler may
receive unacceptably high exposure, EPA takes actions to mitigate the
risk.  The range of possible actions to reduce handlers’ exposure
includes requiring the use of personal protective equipment, reduced
application rates, changes in formulation, use of specific types of
application equipment or engineering controls, or prohibition of the use
pattern.

 The data currently used to estimate handlers’ potential exposure has
a number of limitations.  The Agency believes that data from new handler
exposure studies would provide a much sounder basis for estimating
potential exposure.  In particular, new data should provide a basis for
characterizing the distribution of unit exposures across the population
of handlers performing activities in each scenario.  Two industry groups
have arisen to undertake the research necessary to develop new databases
– the Agricultural Handlers Exposure Task Force (AHETF) and the
Antimicrobials Exposure Assessment Task Force II (AEATF).  The AHETF is
focusing on studies that relate to the use of pesticides in agriculture,
and the AEATF will characterize exposures received by people while
handling antimicrobial pesticides, e.g., disinfectants, materials
preservatives, etc. 

Both Task Forces would like to initiate research soon – the AEATF
during the winter of 2007–2008, and the AHETF during the pesticide use
season in 2008.  At least some Task Force studies would involve
intentional exposure of a human subject.  EPA’s regulation, 40 CFR
§26.1125, requires the sponsor or investigator to submit to EPA, before
conducting  research involving intentional exposure of a human subject,
materials describing the proposed human study in order to allow EPA to
conduct scientific and ethics reviews.  In addition, EPA’s regulation,
40 CFR §26.1601, requires EPA to seek HSRB review of the research
proposal.  

The HSRB has considered the prospect of new handler research at two
previous meetings.  In June 2006 the Board reviewed five proposed
protocols developed by the AHETF.  The Board raised questions and made
numerous comments on both scientific and ethical aspects of the
proposals. 

Over the past year EPA and the Task Forces have worked hard to address
the issues identified by the HSRB.  In response to scientific concerns
raised by the HSRB, EPA analyzed the existing handler exposure database
and relevant scientific literature, and presented its analysis to the
FIFRA Scientific Advisory Panel (SAP) in January 2007.  The Agency asked
the SAP to comment on, among other topics, the “limitations [of
existing data] and on EPA’s conclusion that additional data could
improve significantly EPA’s ability to estimate worker exposure.” 
The SAP report was released April 2, 2007, and is available at:  
HYPERLINK
"http://www.epa.gov/scipoly/sap/meetings/2007/january/january2007finalme
etingminutes.pdf" 
http://www.epa.gov/scipoly/sap/meetings/2007/january/january2007finalmee
tingminutes.pdf  .  At its April meeting the HSRB received a copy of the
SAP report and a presentation by two members of the Panel that prepared
the report.

In addition, for the April 2007 HSRB meeting EPA prepared a draft
document identifying the major elements of the recruitment and
enrollment processes that should be considered by investigators as they
prepare protocols for handler exposure research.  The document discussed
broad principles which should be considered in the course of research
design.  In the future, through a participatory process involving
investigators, workers, and other stakeholders EPA intends to add to the
document specific best practices, and to identify publicly available
resources that contain additional discussion, information, and guidance
relevant to the implementation of general ethical principles in
occupational exposure research. The draft document is available at:   
HYPERLINK
"http://www.epa.gov/osa/hsrb/files/meeting-materials/apr-18-20-2007-publ
ic-meeting/DraftFrameworkForDevelopingBest-Practices0315007.pdf" 
http://www.epa.gov/osa/hsrb/files/meeting-materials/apr-18-20-2007-publi
c-meeting/DraftFrameworkForDevelopingBest-Practices0315007.pdf  

Both the AHETF and AEATF have prepared extensive materials explaining
and justifying their proposed research, and have revised these materials
in response to EPA comments.  These materials provided to the HSRB for
discussion at its June 2007 meeting, generally explain the scope of the
proposed research programs and describe the general framework for
conducting the research.  In addition, each Task Force has provided
Standard Operating Procedures which will guide the conduct of the
studies.   These materials provide essential background information to
support the Board’s evaluations of Task Force protocols and related
materials at subsequent meetings.  Because EPA regards the proposed
studies as “research involving intentional exposure of human
subjects,” EPA regulations require the Agency and the Board to review
these proposals before the investigators initiate the studies.  

  This report transmits the HSRB’s comments and recommendations from
its June 27-29, 2007 meeting.        

REVIEW PROCESS

From June 27-29, 2007, the Board had a public face-to-face meeting in
Arlington, Virginia.  Advance notice of the meeting was published in the
Federal Register “Human Studies Review Board: Notice of Public Meeting
(72 Federal Register 108, 31323).  At the public meeting, following
welcoming remarks from Agency officials the Board then heard
presentations from the Agency on the following topics: 

A research proposal from Carroll-Loye Biological Research to evaluate
the efficacy of two conditionally registered products containing
picaridin in repelling mosquitoes in the field.  

A research proposal from Insect Control & Research, Inc. to evaluate the
efficacy of two unregistered products containing picaridin in repelling
mosquitoes in the field.

A completed study measuring the effects on human subjects of acute
inhalation exposure to acrolein. 

Three completed studies of the efficacy and side effects of
4-aminopyridine used as a therapeutic agent.

Extensive background materials concerning research to quantify the level
of exposure received by people who mix, load, and apply pesticides. 
These materials were prepared by the Agricultural Handlers Exposure Task
Force and by the Antimicrobial Exposure Assessment Task Force. 

 The following oral comments were presented at the meeting: 

 

Scott Carroll, Ph.D. representing Carroll-Loye Biological Research and
Ghona Sangha, Ph.D. representing LANXESS Corporation addressing the
proposed Carroll-Loye picaridin insect repellent efficacy study LNX-001;


Mr. Niketas Spero and Robin Todd, Ph.D. representing ICR, Inc.
addressing the ICR picaridin mosquito repellency protocol;  

Richard Collier, Ph.D. of Landis International representing the
Agricultural Handlers Task Force, John Ross, Ph.D. of
infoscientific.com, representing the Agricultural Handlers Task Force, 
Ray McAllister of CropLife America, Inc., and Larry Holden, Ph.D. of
Sielken and Associates Consulting, Inc.  addressing the design of
research on the levels of exposure received by pesticide handlers.

	For their deliberations, the Board considered the materials presented
at the meeting, written public comments and Agency background documents
(e.g. pesticide human study, Agency data evaluation record (DER) of the
pesticide human study, weight of evidence review, ethics review,
pesticide human study protocols and Agency evaluation of the protocol). 

CHARGE TO THE BOARD AND BOARD RESPONSE

Proposed Carroll-Loye Picaridin Insect Repellent Efficacy Study LNX-001

Charge to the Board

If the proposed research described in Protocol LNX-001 from Carroll-Loye
Biological Research is revised as suggested in EPA’s review, does the
research appear likely to generate scientifically reliable data, useful
for assessing the efficacy of the test substances for repelling
mosquitoes? 

Board Response

. The active ingredient will be formulated into a 20% pump spray and
into a 20% cream. All experiments will be conducted using Good
Laboratory Practices.  A dosimetry experiment with 10 individuals will
be performed to determine the amount of product that would be utilized
by people using the product as directed.  

The experiment will be a field study.  Two locations in California could
be used, either in the Central Valley or in southern California. A
mixture of Culex and Aedes species will be present at these sites.

Legs and/or arms will be tested.  There will be two experienced persons
serving as negative controls (i.e., without any repellent product) to
confirm mosquito biting pressure. Experimental subjects, in pairs, will
monitor landings with intent to bite (LIBe’s) during a one minute
interval each 15 minutes, until the First Confirmed LIBe (FCLIBe) can be
determined. Stopping rules will be employed. The Complete Protection
Time (CPT) will be determined, expressed as mean and standard deviation
plus 95% confidence interval, if data are normally distributed, and
methods are described to assess normality.

With respect to the science criteria established earlier by the HSRB,
the following assessments are made: 

General HSRB Scientific Criteria

The scientific question was stated (i.e., to test the efficacy of
picaridin formulated as either a pump spray or a cream in repelling
mosquitoes).

Existing data were not adequate to answer the question of efficacy of
these new formulations.

Because existing data were not adequate to answer the question of
efficacy, new studies involving human subjects are necessary.

The potential benefits of the study are clear, i.e., that an effective
repellent would be available that would have either greater efficacy
and/or fewer drawbacks than what was currently approved.

It is likely that the benefits would be realized because repellent
efficacy will be determined in carefully designed field experiments.

The risks are minimal because the active ingredient is of very low
toxicity, the other formulation ingredients are of very low toxicity,
the mosquitoes will be aspirated before they have an opportunity to
bite, and the regions selected will not have evidence of West Nile
Virus.

The most likely relevant risk would be irritation from mosquito bites,
but participants will be instructed to remove mosquitoes before they are
bitten, or the possibility of infection with West Nile Virus, but the
regions selected will have no evidence of the virus. Serology tests will
be performed on captured mosquitoes.

Study Design Criteria

The purpose of the study is clearly defined (i.e., efficacy testing).

There are specific objectives (i.e., to determine the Complete
Protection Time that picaridin in two formulations displays as a
mosquito repellent).

There was a formally stated hypothesis; however, it is broad and
untestable.  This does not detract from the value of the study because a
hypothesis is not really necessary for an efficacy study such as this.

The sample size will be 10 individuals per product along with 2
experienced individuals to confirm mosquito biting pressure. A dosimetry
experiment prior to the field experiment will quantify the amount of
repellent being used. 

There is a plan allocating individuals to treatments.

It is anticipated that the findings from this study can be generalized
beyond the study sample.

Participation Criteria

There is justification for the selection of the target population (i.e.,
selection primarily or completely from the existing Volunteer Data Base,
comprised of individuals previously participating in similar studies or
interested in doing so, who routinely are active outdoors, and are
routinely exposed to mosquitoes).

The participants will be representative of some of the population of
concern.  However, there are others in the population unlike these
participants who are likely to use these products, but it would either
be unethical to test them or would be less appropriate to test them. 
The participating population, while not completely representative, is
considered appropriate and reasonable.

The inclusion/exclusion criteria are appropriate.

The sample will not be a vulnerable group.

Measurement Criteria

The measurements will be accurate and reliable as defined. The endpoint
will be the First Confirmed Landing with Intent to Bite (FCLIBe). While
this was viewed as an appropriate endpoint by many of the Board members,
there was some concern that the confirming LIBe criterion was not a
sufficiently protective/conservative, and that the first unconfirmed
LIBE should be the endpoint.

The measurements will be appropriate to the question being asked.

Quality assurance will be a part of the experimental plan.

Statistical Analysis Criteria

The data were designed to be analyzed to calculate Complete Protection
Time with a range of variability. There was concern that specific
criteria from the standpoint of statistics for the selection of 10
subjects were not provided or available.  There was also concern about
the handling of censored data.

Measures of uncertainty were addressed.

Laboratory and Field Conditions

Laboratory experiments are not proposed, except for the dosimetry

Field experiments will be appropriate.

The study will include a stop rule plan, medical management plan, and a
safety monitor.

EPA’s science analysis identified a deficiency in the lack of a stated
hypothesis, and one was subsequently added. However, the objective of
this study, i.e, length of time of efficacy in repelling mosquitoes in
the field, is clear and the lack of a formally stated hypothesis, or a
vague and broad hypothesis, does not detract from the scientific value
of the study. EPA also identified a deficiency in a lack of an
explanation for the negative control in the dosimetry experiment;
however, it is a necessary control to determine whether any factors
besides the formulated product (e.g., sweat) might alter the weight of
the dosimeters. EPA also noted that the method of measuring the
treatment area was not described.  The Board concurred with this
deficiency and urged that it be addressed. There were also two
statistical deficiencies identified, and the Board urged greater
consideration of statistical issues with respect to determination of
sample size and analysis of the data.

HSRB Consensus and Rationale

The protocol LNX-001 to study the efficacy of a cream formulation and a
pump spray formulation of picaridin for repelling mosquitoes is
sufficiently sound, from a scientific perspective, to be used to assess
the repellent efficacy of these formulations against mosquitoes.

Charge to the Board

b. If the proposed research described in Protocol LNX-001 from
Carroll-Loye Biological Research is revised as suggested in EPA’s
review, does the research appear to meet the applicable requirements of
40 CFR part 26, subparts K and L?  

Board Response

Background on Study

The proposed study would evaluate the efficacy of two different
skin-applied formulations of an already registered and marketed insect
repellent, Icaridin (registered by the Agency as Picaridin). Icaridin is
also known under the registered trade name BayrepelTM and marketed under
the brand name Autan.

The research is to be conducted by Carroll-Loye Biological Research, a
private laboratory in Davis, CA. The sponsor of this study is LANXESS,
Inc. of Pittsburg, PA. The submitted documents assert that the study
will be conducted in accordance with the ethical and regulatory
standards of 40 CFR 26, Subparts K and L, as well as the requirements of
FIFRA §12(a)(2)(P), the U.S. EPA’s Good Laboratory Practice (GLP)
Standards described at 40 CFR 160, and the California State EPA
Department of Pesticide Regulation study monitoring (California Code of
Regulations Title 3, Section 6710). Finally, the protocol was reviewed
and approved by an independent human subjects review committee,
Independent Investigational Review Board (IIRB), Inc., of Plantation, FL
prior to submission to the Agency.

The revised research protocol submitted consists of two interdependent
studies: 1) a dosimetry study, performed under controlled laboratory
conditions, designed to determine the amount of an insect-repelling
compound, known as KBR 3023 (picaridin; Icaridin), that normal subjects
would typically apply when provided with one of two compound
formulations (lotion or pump-spray); and 2) an efficacy study, performed
at field sites in Central California and/or Southern California,
designed to measure the effectiveness of 20% KBR 3023 (Picaridin;
Icaridin), as a mosquito repellent. Dosimetry will be determined either
by passive dosimetry using self-adhesive roll-gauze (spray and aerosol
formulations) or by direct measurement of compound application (lotion
formulation). The efficacy of 20% KBR 3023 (picaridin; Icaridin) as a
mosquito repellent will be determined by measuring the ability of the
two formulations to prevent mosquito landings (defined as “Lite with
Intent to Bite”; LIBe) under field conditions. Mosquitoes will be
aspirated mechanically after landing but prior to biting; prior to
initiation of the efficacy study, all volunteers will be trained, using
laboratory-raised, pathogen-free mosquitoes in a controlled laboratory
setting, both to recognize a mosquito landing with the intent to bite
(LIBe) and to remove such mosquitoes with an aspirator. The strengths
and weaknesses of each study design are described above.

 volunteers, a number that appears to be adequately justified (Carroll
2007a; Carroll 2007b).

Ethics and Regulatory Compliance Review

The Board concurred with the factual observations of the strengths and
weaknesses of the study, as detailed in the EPA’s initial Science and
Ethics Review, dated May 24, 2007 (Carley and Sweeney 2007a). With the
provision of an amended protocol on June 14, 2007 (Carroll 2007b), the
proposed research described in Protocol LNX-001 comports with the
applicable ethical and regulatory requirements of 40 CFR 26, subparts K
and L.

 For example, the original and amended protocols were reviewed and
approved by IIRB. Documentation previously provided to the EPA by IIRB
indicates that it reviewed this study pursuant to the standards of the
Common Rule (45 CFR 46, Subpart A) and determined it to be in compliance
with that Rule. 

With respect to study design, the risks to participants are minimal and
justified by the likely societal benefits, including data on the
efficacy of 20% KBR 3023 (picaridin; Icaridin) as a mosquito repellent.
The nature and likelihood of any side effects or adverse events are
described clearly in the informed consent documents (with separate
documents outlining the risks for treated volunteers and experienced,
untreated controls). The risks to study participants are three-fold: 1)
allergic reaction to test materials themselves; 2) exposure to biting
arthropods; and 3) possible exposure to arthropod-borne diseases. 

Reasonable attempts have been taken to minimize any potential harm, and
plans for the medical management of any side effects or adverse events
have been developed. Although 20% KBR 3023 (picaridin; Icaridin) is not
currently used as an insect repellent in the United States, for example,
repellent formulations containing 20% KBR 3023 are commercially
available in Europe and Australia, and have been used for years with
little evidence of toxic effects. Laboratory analyses, as summarized by
Dr. Ghoma Sangha of LANXESS at the public meeting of the HSRB, also
suggest that participants enrolled in this study are unlikely to be at
increased risk of experiencing adverse side effects upon exposure to the
test materials. 

Reactions to mosquito bites are usually mild and easily treated with
over-the-counter steroidal creams. Excluding subjects who have a history
of such severe skin reactions will minimize the risk of a subject
experiencing a severe physical reaction to a mosquito bite. In addition,
the study protocol is designed specifically to minimize the likelihood
that a mosquito will bite, through the use of clear stopping rules,
limited exposure periods, pre-bite aspiration and joint observation. 

To minimize the risk that study subjects will be exposed to illnesses
such as West Nile Virus, field tests of repellent efficacy will be
conducted only in areas where known vector-borne diseases have not been
detected by county and state health or vector/mosquito control agencies
for at least one month. Finally, mosquitoes collected while attempting
to bite control and treated subjects during the field tests will be
subjected to multiplex RT-PCR assays for several known arthropod-borne
diseases—including West Nile Virus, Western Equine Encephalitis Virus,
and St. Louis Encephalitis Virus—with clear plans to contact study
participants and alert them if a transmissible pathogen is detected.

In accordance with the provisions in the EPA’s final human studies
rule (40 CFR §§ 26.1701-1704), minors and pregnant women are
explicitly excluded from participation, the latter being confirmed by
requiring all female volunteers to undergo a self-administered
over-the-counter pregnancy test on the day of the study. The use of
so-called “alternate” subjects ensures that the results of
over-the-counter pregnancy tests would be kept private; that study
participants may be designated as alternate subjects and automatically
excluded from participation allows for potentially pregnant volunteers
to withdraw without compromising their confidentiality.

Finally, the study protocol also included several mechanisms designed to
minimize coercive subject recruitment and enrollment. For instance,
although the study is to be conducted by Carroll-Loye Biological
Research, a private research laboratory in Davis, California, the
Principal Investigator of the study and Co-Owner of the research
laboratory, Dr. Scott P. Carroll, also is an adjunct faculty member of
the Department of Entomology at the University of California, Davis. The
majority of research participants will be recruited from the
University’s student population, including from Dr. Carroll’s own
department, but the protocol specifically excludes any student or
employee of the Study Director and includes a substantial waiting period
between recruitment and study enrollment and an interview by Dr. Carroll
designed to minimize coercive subject recruitment and enrollment. In
addition, compensation for study participation is not so high as to
unduly influence enrollment.  It is important to note, however, that the
planned use of a convenience sample of study participants may limit the
broad applicability of the study results to the general population; this
fact is noted by the study investigators in the protocol.

HSRB Consensus and Rationale

The Board concurred with the assessment of the Agency that the protocol
LNX-001 submitted for review by the Board, if revised as suggested in
EPA’s review, meets the applicable requirements of 40 CFR 26, subparts
K and L.  In addition, with the submission of the amended protocol, the
Board believed that the protocol meets the applicable requirements of 40
CFR 26, subparts K and L.

Proposed ICR Picaridin Insect Repellent Efficacy Study

Charge to the Board

If the proposed research described in ICR’s proposed picaridin
protocol is revised as suggested in EPA’s review, does the research
appear likely to generate scientifically reliable data, useful for
assessing the efficacy of the test substances for repelling mosquitoes? 

Board Response

This protocol was submitted with Confidential Business Information (CBI)
redacted.  The name of the sponsor was withheld as was the concentration
of the active ingredient, picaridin. A workgroup of the HSRB had decided
prior to the HSRB review that the information being withheld was not
critical to the scientific review of the protocol.

The protocol was clearly written and described a field study to be
conducted on two picaridin aerosol formulations at two locations. The
locations, selected to achieve different species composition of
mosquitoes, would be: Site 1: Savannah-Ogeechee Canal Museum and Nature
Center, Savannah, GA, and Site 2: Pine Island, Lee County Mosquito
Abatement District, FL. The testing would last up to 14 hours on any
particular test day.

There is an adequate pool of test subjects from individuals previously
tested for repellent efficacy from which the specific subject pool will
be selected. Fourteen individuals will be selected for the conduct of
the protocol at each site, 10 test subjects, 2 alternates, and 2
untreated controls.  The treated subjects will have one test formulation
on one arm and the other on the other arm.  A standard amount, as
indicated in EPA test guidelines, will be applied to the subjects’
arms; no dosimetry study will be conducted to determine likely consumer
use rate. The treated skin will be exposed to potential mosquito bites
for 5 minutes out of each 30 minutes. During the remaining 25 minutes of
each 30 minute interval, the subjects will be in a screened area that is
impervious to mosquitoes. The treated area of skin will not be covered
during these 25 minute intervals to avoid any rubbing off of the
product. The endpoint will be time to First Confirmed Bite (FCB),
monitoring bites from only those mosquitoes which have all 6 legs on the
skin. Two negative controls will be selected randomly; these individuals
are not anticipated to receive bites. No positive control will be
conducted. The ICR staff will record the data.

There was concern raised by the Board because of the lack of a dosimetry
study. The Board understands that a standardized amount of the repellent
is the guideline. However, the Board considered a dosimetry experiment
as a valuable addition to the protocol and recommended that it be
included so that the most likely amount of product applied by the
consumer will be tested.

The Board recommended that a sample of mosquitoes be taken from those
biting the participants so that subsequent serology for pathogenic
viruses can be performed; this information can be provided to the
participants to assist them in medical management if there is any
evidence of diseases present in the mosquitoes.

The Board also had concerns about the use of the First Confirmed Bite as
the endpoint.  The Board understands that the FCB is at present the
guideline endpoint, and appreciates that it may be the more rigorous
criterion for efficacy.  However, the Board has been impressed with the
added safety to participants of the Landing with Intent to Bite (LIBe)
endpoint on some of the previously reviewed protocols, even though this
endpoint is not that stated in the current guidelines. There is
insufficient information to judge how reflective the LIBe is of the
bite. The Board is also cognizant of the possibility of a lack of
consistency in the labels if new criteria are introduced, and the
resultant confusion and/or unfairness to consumers who wish to compare
products.  However, balancing these issues (consistency of endpoints,
safety to test subjects, and accuracy of the results) needs to be a
priority item for EPA; thus the Board strongly urged EPA to confirm or
update its guidelines with the best possible balance of safety,
consistency with past studies, and accuracy of the results (including
statistical validity).

With respect to the science criteria established earlier by the HSRB,
the following assessments are made: 

General HSRB Scientific Criteria

The scientific question was stated (i.e., to test the efficacy of
picaridin formulated as one of two aerosols in repelling mosquitoes).

Existing data were not adequate to answer the question of efficacy of
these new formulations.

Because existing data were not adequate to answer the question of
efficacy, new studies involving human subjects are necessary.

The potential benefits of the study are clear, i.e., that an effective
repellent would be available that would have either greater efficacy
and/or fewer drawbacks than what was currently approved.

It is likely that the benefits would be realized because repellent
efficacy will be determined in carefully designed field experiments.

The risks are low because the active ingredient is of very low toxicity,
and the regions selected will not have evidence of West Nile Virus
during the previous week, although this one week interval may not be
sufficient protection. Because the protocol specifies that the test
subjects must be bitten, there is the possibility of infection from
virus-bearing mosquitoes in the repellent-treated subjects.

The most likely relevant risk would be irritation from mosquito bites,
or the possibility of infection with West Nile Virus or other
mosquito-borne diseases, but the regions selected will have no evidence
of the West Nile Virus during the week prior to the test. A more
intensive medical management plan should be described.

Study Design Criteria

The purpose of the study is clearly defined (i.e., efficacy testing).

There are specific objectives (i.e., to determine the protection time
that picaridin in two formulations displays as a mosquito repellent).

There was no formally stated hypothesis; however, this does not detract
from the value of the study.

The sample size will be 10 individuals (plus selection of 2 alternates)
along with 2 experienced individuals to confirm mosquito biting
pressure. A dosimetry experiment prior to the field experiment will not
be performed, but the standardized amount in the guidelines will be
used.  The Board recommended that a dosimetry test be performed to
quantify the amount of repellent a consumer would apply, and that this
be used in the field tests. 

There is a plan allocating individuals to treatments.

It is anticipated that the findings from this study can be generalized
beyond the study sample.

Participation Criteria:

There is justification for the selection of the target population (i.e.,
selection primarily or completely from the existing Volunteer Data Base,
comprised of individuals previously participating in similar studies).

The participants will be representative of some of the population of
concern; however, there are others in the population unlike these
participants who are likely to use these products, but it would either
be unethical to test them or would be less appropriate to test them. 
Even though the participant group is not totally representative, the
participating population is considered appropriate and reasonable.

The inclusion/exclusion criteria are appropriate, except that
individuals over the age of 55 should not be participants.

The sample will not be a vulnerable group.

Measurement Criteria

The measurements will be accurate and reliable as defined. The endpoint
will be the First Confirmed Bite (FCB), which is the endpoint in the
testing guidelines. While this would be consistent with the tests that
have been conducted in the past, there was some concern that the bite
would place the participants at greater risk than necessary because of
the possibility of contracting a mosquito-borne disease.

The measurements will be appropriate to the question being asked, with
the possible exception of the exclusion of bites from mosquitoes which
do not have all 6 legs on the skin.

Statistical Analysis Criteria

The data were designed to be analyzed to calculate mean time to FCB,
plus standard deviation and 95% confidence interval. There was concern
that specific criteria from the standpoint of statistics for the
selection of 10 subjects were not provided or available.  

Information on determining the normality of data was not included.

The originally-suggested exclusion of “outliers” should be abandoned
and the data from all participants should be included.

Measures of uncertainty were addressed, but need to be considered in
relationship to whether the data are normally distributed or not.

Laboratory and Field Conditions

Laboratory experiments are not being proposed.

Field experiments will be appropriate.

EPA’s science analysis identified a deficiency in the lack of a stated
hypothesis, and there was a promise to add a hypothesis about a test of
a specified length of repellent efficacy; however, the objective of this
study, i.e., determination of the length of efficacy in repelling
mosquitoes in the field, is clear and the lack of a formally stated
hypothesis does not detract from the scientific value of the study. In
reality, for this type of study, it does not matter whether there is a
hypothesis of a specific time or not because the study’s data will
indicate what the length of efficacy is for that particular product,
which is what the EPA needs for its regulatory purposes. EPA also
identified a deficiency in a lack of information about determination of
the normality of the acquired data, and the handling on non-normally
distributed data (should this be the case), and the Board concurred that
such information is needed. The Board concurred with EPA regarding the
necessity for addressing the identified deficiencies in statistical
approaches and Good Laboratory Practices.

HSRB Consensus and Rationale

If amended and consistent with the Board’s concerns and
recommendations, the protocol ICR 1A 044 studying the efficacy of two
aerosols formulations of picaridin for repelling mosquitoes would be
sufficiently sound, from a scientific perspective, to be used to assess
the repellent efficacy of these formulations against mosquitoes.

Charge to the Board

If the proposed research described in ICR’s proposed picaridin
protocol is revised as suggested in EPA’s review, does the research
appear to meet the applicable requirements of 40 CFR part 26, subparts K
and L?  

Board Response

Background on the Study

The proposed study (Spero 2007) would evaluate the efficacy of two
different skin-applied formulations of picaridin-based insect
repellents. Picaridin is already registered and marketed as an insect
repellent in the United States, under the registered trade name
BayrepelTM and marketed under the brand name Autan. The goal of this
study is to evaluate the efficacy of two formulations under field
conditions.

The research is to be conducted by ICR, Inc., a commercial organization
based in Catonsville, Maryland; ICR provides testing and regulatory
consulting services for companies developing and marketing pesticides
and insecticides in the United States and Canada. The study is managed
by toXcel, LLC of Gainesville, Virginia. The sponsor of this study is
unknown; because of claims of CBI, the documents for this study were
provided to the HSRB in redacted form, with neither the sponsor’s
identity nor the exact composition of the two test formulations
provided. The submitted documents assert that the study will be
conducted in accordance with the ethical and regulatory standards of 40
CFR 26, Subparts K and L, as well as the requirements of FIFRA
§12(a)(2)(P), and the U.S. EPA’s Good Laboratory Practice (GLP)
Standards described at 40 CFR 160. Finally, the protocol was reviewed
and approved by an independent human subjects review committee, Essex
Investigational Review Board (EIRB), Inc., of Lebanon, NJ prior to
submission to the Agency.

Efficacy of the two picaridin-based formulations will be evaluated under
field conditions by using healthy volunteers. The study will be
performed at two field sites in Georgia (Savannah-Ogeechee Canal Museum
and Nature Center) and Florida (Pine Island, Lee County). The
effectiveness of the two compounds as mosquito repellents will be
determined by measuring the ability of each formulation to prevent
mosquito bites under field conditions. The strengths and weaknesses of
the study design are described above.

 in the protocol as either two confirmed bites in a single five-minute
exposure period, or one bite in each of two consecutive exposure
periods. Treated study participants will work in pairs, observing
mosquito landings and alerting attendant ICR staff of potential bites;
ICR staff will determine whether or not these are confirmed bites.
Probes (i.e., "bites" where the mosquito punctures the skin but does not
collect blood) and bites from mosquitoes that do not fully alight (i.e.,
all six legs on the surface of the exposed skin) will not be considered
as confirmed bites. Once breakdown has occurred for a particular
repellent formulation, no further exposure of the subject's treated skin
will occur. 

 Landing mosquitoes will be brushed away by attendant ICR staff, and a
small number will be collected for later laboratory identification.

Ethics and Regulatory Compliance Review

The Board concurred with the factual observations of the strengths and
weaknesses of the study, as detailed in the EPA’s Science and Ethics
Review, dated May 24, 2007 (Carley and Sweeney 2007b). If the
recommended changes included therein are incorporated, the proposed
research described in Protocol ICR 1A 044 is likely to meet with the
ethical and regulatory requirements of 40 CFR 26, subparts K and L.  As
submitted to the Agency, however, the current protocol (Spero 2007)
fails to meet the applicable requirements of the Agency’s Final Human
Studies Rule40 CFR 26, subparts K and L.

Subpart K of the Agency’s Final Human Studies Rule requires that the
investigator submit to the EPA all information that pertains to the IRB
review of proposed research (40 CFR 26.1115a) as well as additional
information specified in 40 CFR 26.1125, if not already included in the
IRB documentation.  The information requested under 40 CFR 26.1125
includes a discussion of the potential risks to human subjects, the
measures proposed to minimize these risks, expected benefits if any and
to whom, alternative means to obtain comparable information, and the
balance of risk and benefits of the research.  In addition, subject
information sheets and approved written informed consent documents
should be provided, along with any information about recruitment and the
presentation of this subject information.  Finally, the investigator
should provide copies of all correspondence with the IRB, including
official notification of IRB review and approval.

, minor deficiencies not withstanding.

With respect to study design, the risks to participants (if properly
minimized) are justified by the likely societal benefits, including data
on the efficacy of these new picaridin-based formulations of mosquito
repellents. The risks to study participants are three-fold: 1) allergic
reaction to test materials themselves; 2) exposure to biting arthropods;
and 3) possible exposure to arthropod-borne diseases. Although plans for
the medical management of any side effects or adverse events have been
developed, it is not unreasonable to expect greater efforts to be taken
to minimize any potential harm. Allergic reactions to the test materials
themselves are unlikely. Picaridin is commercially available and has
been used at higher doses as a repellent with little evidence of toxic
effects, so the subjects enrolled in this study are unlikely to be at
increased risk of experiencing adverse side effects upon exposure to the
test materials. The inert ingredients are also widely used in cosmetic
and personal care products, and have previously been reviewed and
approved for use in other pesticide products registered by the Agency
under FIFRA. It is disturbing to note, however, that picaridin is listed
in the informed consent documents as a seemingly innocuous Toxicity
Category IV compound. This may mislead participants into believing that
picaridin is less toxic than it actually is, given that the EPA lists
picaridin as a Toxicity Category III compound for acute oral, dermal and
eye exposure. In toXcel’s response to the Agency’s science and
ethics review of the protocol, however, Dr. Micah Reynolds indicates
that this misleading information has been corrected in the informed
consent documents (Reynolds 2007).

 for efficacy testing; OPPTS 810.3300 and 810.3700 specify use of first
bite or first confirmed bite for determining protection time. However,
many Board members argued—from both a scientific and an ethical
perspective—that the Agency should give considerable weight to
revising these guidelines to specify use of LIBe for determining
protection time in future studies submitted to the EPA for registration.

The protocol states that to minimize the risk that study subjects will
be exposed to arthropod-borne illnesses such as West Nile Virus, field
tests of repellent efficacy will be conducted only when “cases” of
known vector-borne diseases have not been detected by local
vector/mosquito control agencies for at least one week prior to
initiation of the field trials. However, disease prevalence and
incidence can change rapidly. Conducting field trials in areas that have
been certified as “disease-free” only for a single week may be
inadequate to minimize the potential risk of vector-borne disease for
study participants, and the Board recommended that the protocol be
amended such that trials only be conducted in areas in which known
arthropod-borne viruses have not been detected in weekly testing for at
least one month prior to initiation of the field studies.  As
researchers also will be collecting mosquitoes for species
identification, to further minimize the risk of arthropod-borne disease
these mosquitoes should be subjected to serologic or DNA-based assays
for known arboviruses, with clear and workable plans to contact study
participants and alert them if a transmissible pathogen is detected.
Given that many of these diseases can have long incubation periods in
infected individuals (e.g., 2-14 days for West Nile Virus), and that
many of the symptoms of arthropod-borne illnesses can be diffuse or
subacute, simply following up with study participants via phone two
weeks after the completion of the field studies is insufficient.

Study investigators also should clarify in both the protocol and the
informed consent document whether or not “cases” refers to actual
reports of human disease, or the detection of pathogens in the local
mosquito population by using sentinel flocks or other laboratory based
methods. In the site-specific informed consent documents, the term
“case” is used to describe both actual cases of human disease and
detection of known arboviruses in mosquito pools in adjoining paragraphs
detailing the risk to study participants of vector-borne disease. This
may be confusing to study participants, and interfere with their
accurate assessment of the risks. The Board recommended that reference
to rates of West Nile Virus and other arthropod-borne diseases (human
“cases”) be deleted from the informed consent documents, as should
the suggestion that the species known to transmit these viruses most
frequently are uncommon at the sites where the field trials will be
conducted. These statements may lead study participants to underestimate
the risk of exposure to these agents. Although these illnesses were rare
in Georgia and Florida in 2006, for example, rates of vector-borne
diseases like West Nile Virus were substantially higher in previous
years and it is difficult for even the most seasoned arbovirologist to
predict what the likely incidence will be in the field test areas in
2007 and 2008.

Finally, as effects of arthropod-borne diseases are particularly severe
in the elderly or those with compromised immune systems, it may be
prudent to: 1) exclude participants greater than 55 years of age (rather
than the current protocol’s current 65 years and older), and 2)
explicitly describe the risk for those with immune disorders—many with
asymptomatic HIV-disease, for example, may consider themselves healthy
enough for study participation yet may be at increased risk of illness.
In written comments provided to the Agency (Reynolds 2007), ICR and
toXcel apparently agree with the recommended change to the protocol’s
age-based inclusion and exclusion criteria.

In accordance with the newly promulgated provisions in the EPA’s final
human studies rule (40 CFR §§ 26.1701-1704), minors and pregnant women
are explicitly excluded from participation, the latter being confirmed
by requiring all female volunteers to undergo a self-administered
over-the-counter pregnancy test on the day of the field study. In order
to protect the confidentiality of these results, however, it may be
prudent to have study participants conduct the pregnancy tests just
prior to travel to the sites in Georgia and Florida, with an
investigator-confirmed test performed before exposure at the field site,
so that pregnant participants can self-exclude themselves from study
participation prior to travel; exclusion of female participants once
they have arrived at the field sites may otherwise be difficult to
explain while ensuring that the results of over-the-counter pregnancy
tests are kept private. Additional procedures to ensure confidentiality
are also recommended. As noted in the Agency’s science and ethics
review (Carley and Sweeney 2007b), identification of study participants
by first name and last initial was inappropriate and it is rather
surprising that the EIRB did not request the use of unique identifiers
to further protect participant identity. In written comments provided to
the Agency (Reynolds 2007), however, ICR and toXcel apparently agree
with this assessment and have modified to protocol accordingly to
require the use of unique numerical identifiers.

A number of additional concerns should also be raised. First, a more
detailed explanation of study recruitment is needed, particularly a
description of plans to minimize coercive subject recruitment and
enrollment. Although compensation for study participation is not so high
as to unduly influence enrollment, it may be appropriate to exclude all
employees and contractors of ICR, toXcel and the sponsor (as well as
family members), not just full-time employees of ICR, in order to
minimize coercion; in written comments provided to the Agency (Reynolds
2007), ICR and toXcel apparently agree with this assessment and have
modified the protocol. Enrollment and informed consent procedures should
also be described in greater detail; informed consent is a process, not
just a discrete moment in time. Although the informed consent document
will be described and discussed with potential subjects—in person or
via telephone—prior to initiation of the field studies, discussion of
the risks and benefits of study participation should be ongoing. It is
insufficient to simply state that, on the evening prior to the field
trials, “[investigators] will review with [study participants] the
specifics of the study as described in the ICD” (e.g., Spero 2007,
53).   A detailed explanation of study procedures for risk and benefit
is required. Furthermore, the additional risks to untreated control
subjects (chosen by lottery rather than via separate enrollment of more
experienced study participants) should be clearly listed in both the
protocol and the informed consent document.  However, it should be noted
that mosquitoes are supposed to be aspirated from untreated controls
prior to biting

Finally, there was considerable debate about whether or not the
recruitment of research subjects from distant sites, with transport to
field sites in Georgia and Florida, might be coercive with respect to
enabling study withdrawal; volunteers may be less willing to withdraw
from study participation if withdrawal involves considerable
inconvenience and delay in returning home or if they believe that they
have some reciprocal obligation to the researchers for the travel and
lodging. In addition, given the amount of time involved and the
additional risks associated with study-related travel and out-of-state
housing (though not directly related to the study intervention),
researchers and study sponsors should more clearly justify the
recruitment and transport of experienced study participants from across
the United States rather than recruiting and enrolling volunteers from
the local populations in Savannah, Georgia and Lee County, Florida. At
least one member of the Board, for example, believed that a more
ethically-appropriate study design would involve recruitment of local
research participants and specialized training sessions for those
volunteers.

HSRB Consensus and Rationale

	The Board concurred with the assessment of the Agency that the protocol
ICR 1A 044 submitted for review by the Board, if revised as suggested in
both EPA’s review and by the Board, would meet the applicable
requirements of 40 CFR 26, subparts K and L.

Completed Inhalation Study with Acrolein

Charge to the Board

The Agency has concluded that the Weber-Tschopp et. al (1977) study
contains information sufficient for assessing human risk resulting from
potential acute inhalation exposure.  Please comment on whether the
study is sufficiently sound, from a scientific perspective, to be used
to estimate a safe level of acute inhalation exposure to acrolein.  

Board Response

The Board began by highlighting the toxicological evaluation of acrolein
as prepared by the Agency for Toxic Substances and Disease Registry
(ATSDR, 2005) and the EPA Integrated Risk Information System (IRIS
2003).  The information below is reproduced from these reviews. Acrolein
is toxic by inhalation, oral, and dermal exposures (toxicity category I
for all routes). It is a potent irritant to the mucous membranes. As
such, its toxicity is exerted at the point of contact with tissues.
Signs and symptoms resulting from inhalation exposure to airborne
acrolein may include irritation of the nose, throat and lungs, pulmonary
edema, lung hemorrhage, and death. The nasal tissues appear to be the
most sensitive target of inhalation exposure, with onset of noticeable
irritation occurring in seconds (0.3 ppm). Higher airborne
concentrations of acrolein (2– 5 ppm) result in increasingly severe
manifestations of irritation over the entire respiratory tract.  Oral
acrolein exposure may result in gastrointestinal discomfort, vomiting,
and stomach ulceration and/or hemorrhage. The stomach epithelium appears
to be the most sensitive target for oral exposure (0.75 mg/kg). Higher
concentrations of ingested acrolein have primarily resulted in
increasingly severe irritation effects in the stomach (2 mg/kg and
higher). Exposure to acrolein vapors or liquids may cause stinging of
the eyes, lacrimation, and reddening, ulceration, or necrosis of the
skin (10% acrolein solution). The eye appears to be the most sensitive
target for exposure (0.3 ppm). Histological changes in respiratory and
gastrointestinal epithelium have been observed from both inhalation and
oral exposures, respectively. Changes in body and organ weights,
hematology, and serum biochemistry, as well as developmental effects
have been observed.  Some of these effects are believed to be secondary
effects of gastrointestinal and/or respiratory tract irritation (i.e.,
loss of appetite and weight loss due to gastrointestinal irritation).
Inhaled acrolein is retained primarily in the upper respiratory tract
(Egle, 1972) because of its high solubility and reactivity. Draminski et
al. (1983) identified a low level of acrolein derived conjugates in the
urine of rats following oral dosing. Orally administered acrolein is
excreted (as metabolites) in the urine, feces and as carbon dioxide. The
main pathway of metabolism for acrolein is the addition of GSH to the
activated double bond followed by conversion to mercapturic acid. A
second pathway is that of epoxidation of the double bond followed by
attack on the epoxide by glutathione. A third pathway is addition of
water to acrolein to form 3-hydroxypropionaldehyde, which can be further
metabolized and ultimately incorporated into normal metabolic pathways
(Parent et al., 1998). Exposure of the general population occurs
primarily through atmospheric contact (HSDB, 2003). 

or levels 2- to 20-fold times (Environment Canada, 2000). Levels between
2.3 and 275 μg/m3 have been reported in smoky indoor environments such
as bars and restaurants (IARC, 1995). In residences where wood stoves
were used, concentrations from 0.7-6.0 μg /m3 have been reported (IARC,
1995). IARC (1995) noted that the acrolein concentrations in the smoke
from various cigarettes ranged from 3-220 μg/cigarette. Levels as high
as 463-684 μg/cigarette were reported (Kuwata et al., 1979). Jones et
al. (1999) reported concentrations of acrolein in mainstream smoke
ranging from 10 – 140 μg per cigarette, and estimated concentrations
in side stream smoke in the range of 100 – 1700 μg per cigarette
(IRIS 2003)” 

It is fairly well known that the annoyance effect of environmental
tobacco smoke (ETS) is likely due to acrolein, as indicated in the EPA
document on ETS (1992, reprinted by NIH in 1992-93); Weber (ibid.)
indicated it was more pronounced for nasal irritation than eye
irritation, though probably not the primary cause of irritation from
ETS.

While the Weber-Tschopp et al. study is the focus of the review, the EPA
IRIS review (2003) reports another human exposure study by Sim & Pattle
(1957) in which 12 volunteers were exposed in a chamber to 0.8 and 0.12
ppm acrolein for 10 and 5 minutes, respectively; the volunteers reported
it was extremely irritating to all exposed mucosal surfaces.  No chronic
studies of human exposure to acrolein have been reported. NIOSH has
recommended that the concentration in workroom air be limited to 0.1 ppm
averaged over an 8-hour shift. The ATSDR toxicology profile stated that
“acrolein exposure levels were very comparable for the appearance of
cellular changes in nasal epithelium of animals and onset of nasal
irritation in humans” (Weber-Tschopp et al., infra vide), implying
that acute nasal effects are similar.

The Weber-Tschopp et al. study provides the most comprehensive
description of acute effects in humans.  For this research, healthy male
and female college student volunteers were exposed to acrolein in a 30
m3 chamber at an 0.1 hourly air exchange rate in 3 trials: 

(1) A continuous exposure at constantly increasing acrolein
concentrations, 

(2) Discontinuous short exposures to successively increasing
concentrations, and 

(3) Constant concentration for one hour. 

Acrolein was injected with a micro liter syringe, vaporized and blown
into the test chamber via a carrier gas stream. Acrolein concentration
in the test chamber was quantitatively determined and results were
reproducible [sd = 0.023 ppm = 3.8%]. 

In the first experiment, 31 male and 22 female students in groups of
three participated. One trial with acrolein and one control trial under
identical conditions but without acrolein were performed with each
subject. Students were exposed to increasing acrolein concentration from
0 to 0.6 ppm in the first 35 minutes and to a constant 0.6 ppm
concentration in the last 5 minutes. The subjects had to fill out a
questionnaire every 5 minutes. The questions were: Is air quality good?
Acceptable or bad? And do you have a desire to leave the chamber? After
that, two subjects in each group were immediately compared for eye
blinking frequency. With the third subject the breathing frequency
during the entire exposure was measured. Eye irritation was
significantly higher (p<0.01) than controls at 0.09 ppm and above. Nasal
irritation was significantly higher (p<0.01) than controls beginning at
0.26 ppm. Throat irritation was experienced at 0.43 ppm and above. Eye
blinking rate was experienced at 0.26 ppm and above (p<0.01).
Respiration rate decreased by 25% (p<0.01) at 0.6 ppm concentration. 

In the discontinuous short exposure experiment there were 42 students
(17 males and 25 females). The subjects in groups of 4 were each exposed
5 times for 1 ½ minutes to variously high acrolein concentrations (0,
0.15, 0.3, 0.45, and 0.6 ppm). After a minute of exposure, they were
given the questionnaire form to fill. Between each exposure they were
allowed to recuperate in a clean room for 8 minutes. The same controls
from the first experiment were used. Eye and nasal irritation was
significantly higher (p<0.05) than controls beginning at 0.3 ppm and
0.06 ppm, respectively. Throat irritation was not evident. 

In the constant one hour exposure duration, 46 students in groups of
threes (21 males and 25 females) were exposed to 0.3 ppm acrolein
concentration for 60 minutes. Measurements of eye blinking frequency,
breathing frequency and subjective symptoms of irritation were taken at
the beginning of exposure and during exposure. Measurement of control
values were obtained in the subjects at the beginning of exposure. Eye,
nose and throat irritation increased significantly (p < 0.01), reached a
plateau after 20-30 minutes of exposure, while eye blinking frequency
plateaued after 10 minutes. Respiratory rate decreased 20% after 40
minutes exposure (p<0.01) in 16 subjects. The severity of the annoyance
significantly increased almost immediately after acrolein was
introduced. Eye, nose and throat irritation and eye blink frequency
increased with increasing exposure duration. After 40 minutes, the
subjective irritation reached a constant intensity while eye blink
frequency after 10 minutes reached a definite rate. Throat irritation,
which was insignificant in the other exposures, reached significance
after only 10 minutes at this long exposure. There was a significant
individual correlation (p between <0.05 and <0.01) between eye blink
frequency and the subjective eye irritation. Every person with a sharp
increase in eye blink frequency also had a sharp increase of eye
irritation. 

The volunteers were asked about the air quality during the exposure if
it was good, bad or for the desire to leave the chamber and the degree
of irritation to the eyes, nose and throat. The effects to continuous
exposure as well as discontinuous exposure increased with acrolein
concentration. Some indication of adaptation to the annoyance, not
irritating effects of acrolein was suggested by the study investigators
as intermittent exposures were more annoying. The eyes were more
sensitive than the nose to the irritating effects of acrolein. 

In the continuous exposure the irritation was significantly greater both
in the eyes and nose than in the discontinuous short exposures which the
investigators attribute to an increase in the sensitivity of both organs
as a function of increasing exposure time. Throat irritation in both
experiments was not as sensitive a criterion: in continuous exposure it
increased significantly through 0.43 ppm, in discontinuous exposure it
showed no change. The eye blink frequency of 34 subjects in the
continuous trial was a function of the acrolein concentration. It
increased from 0.17 ppm to 0.26 ppm (p<0.01) and it doubled at about 0.3
ppm. The breathing frequency of 19 subjects in the continuous exposure
trial decreased slightly with increasing acrolein concentration. This
decrease was statistically significant at 0.6 ppm (p<0.05). At this
concentration the decrease in breathing frequency reached 4 breaths per
minute - a decrease corresponding to about 25%.  An increase in
irregular breathing frequency in 11/19 subjects compared to controls was
observed, very soon after the addition of acrolein but mostly in the
second half or last third of the exposure time. Nearly half of the
subjects displayed more or less pronounced tendency to lengthen the
expiration cycle or more rarely the inspiration cycle holding the breath
toward the end of the acrolein exposure. 

Based on the results of this investigation, it was concluded that the
threshold for the effects measured are: Eye irritation 0.09 ppm; Nasal
irritation 0.15 ppm; Eye blink frequency 0.26 ppm; Breathing frequency
0.30 ppm; Throat irritation 0.30 ppm.   The investigators concluded also
that the threshold value for irritation is at least at the lower end of
the spectrum, i.e., around 0.1 ppm.

Based on the nose and throat irritation and a decrease in respiratory
rate in humans exposed to acrolein, ATSDR derived an acute-duration
inhalation MRL of 0.003 ppm calculated from the LOAEL of 0.3 ppm from
the Weber-Tschopp et al. 1977 study. They also derived an intermediate
duration inhalation MRL of 0.04 ppb from their extrapolation based on
animal nasal epithelial metaplasia. EPA also has an RfC for acrolein. 
EPA concluded in its WOE that “the study demonstrated that subjective
eye irritation was the most sensitive indicator for the acute acrolein
exposure in humans with a threshold effect of 0.09 ppm (0.2 mg/m3).
Protection of the eyes from the irritating effects of acrolein will
protect against other respiratory effects of nasal and throat irritation
and breathing effects which occurred at slightly higher thresholds.” 

Critique of Weber-Tschopp et al. Study

Strengths

This was a very well designed and conducted study in a lab with known
scientific ability, QA capabilities, and institutional ethical review.
Acrolein monitoring, chemical analysis, and low variability in chamber
concentrations are indicative of the study’s strength. A large number
of healthy subjects were used (though the total number of subjects
involved is unknown), including controls (though the controls may not
have been very “blind”). The measurements of the subjects were
performed well with appropriate methods, the most important being the
objective measurements. The three studies looked at intermittent and
continuous exposures over a sufficient range of concentrations and of
times for the acute effects to be manifested. LOAELs could be determined
from these data.  The study appears to meet the HSRB scientific
criteria.  

Weaknesses

No positive controls were used, and negative controls may have been
biased by not being completely “blind”. There is no justification
for the sample size, so there may have been a false negative rate in
some comparisons where statistically significant differences were not
found. The periods between studies are unknown. The results may not be
generalizable to younger or older groups.

HSRB Consensus and Rationale 

Charge to the Board

Please comment on the following:

1) Is there clear and convincing evidence that the conduct of the study
was fundamentally unethical?

2) Is there clear and convincing evidence that the conduct of the study
was significantly deficient relative to the ethical standards prevailing
at the time the research was conducted?

Board Response

Introduction 

This is a report of third party research involving human subjects that
was 

conducted prior to April 6, 2006 and was not conducted with the
intention of submission to EPA for the purpose of product
registrationunder the pesticide laws. Rather, this study was conducted
in order determine the relative contribution of acrolein to the
irritating effects of cigarette smoke. The study was conducted at the
Institute for Hygiene and Occupational Physiology, Swiss Federal
Engineering College, in Zurich Switzerland. Financial support for the
study was provided by the Association Suisse des Fabricants de
Cigarettes, located in Fribourg, Switzerland. 

 to EPA by a registrant, 40 CFR §26.1303, which requires documentation
of ethical conduct of studies submitted after April 6, 2006, does not
apply. In addition, because the study was conducted prior to that date,
it is not subject to those documentation requirements.  40 CFR
§26.1602(b)(2) requires HSRB review, 40 CFR §26.1703 forbids EPA
reliance on research involving intentional exposure of pregnant or
nursing women or children, and 40 CFR §26.1704 forbids EPA reliance on
pre-rule research if there is “clear and convincing evidence” that
its conduct was fundamentally unethical or significantly deficient
relative to standards prevailing when it was conducted.  When evidence
concerning subject age and reproductive status is both absent and
unobtainable, EPA’s policy is that §26.1703 does not prohibit
reliance on a study.  

This report describes three sub-studies: 

 A) Continuous exposure to acrolein concentrations increasing over 40
minutes from zero to 0.60 parts per million (ppm) 

B) 90-second exposures separated by 8-minute recovery periods to
concentrations increasing from 0.15 to 0.60 ppm

C) Continuous exposure over 60 minutes to constant concentration of 0.30
ppm.  

All sub-studies were conducted in a 30-cubic meter (m3) chamber using
healthy research subjects.  Subjective measures of annoyance were
obtained from the responses of the research volunteers to questions
about air quality (“good”, “acceptable”, or “poor”), a wish
to leave the room (“no”, “don’t know”, or “yes”), and
perceived eye, nose, and throat irritation (1=not at all; 2=a little;
3=medium; 4=strong).  Objective measures of response were recorded only
for tests A and C and included measurement of eye blink rate for two out
of three subjects and respiratory rate and depth for the remaining one
out of three subjects.  

There were 53, 42, 46 subjects in sub-studies A, B and C respectively.
It is not known whether this represents a maximum of 141 individual
subjects, a minimum of 56 individual subjects ([based on 25 women, which
is the largest number of women in any of the three sub-studies, and 31
men, which is the largest number of men in any of the three
sub-studies], or some number in between. This range of somewhere between
56 and 141 individual research subjects in the three sub-studies assumes
that no subject participated more than once in a given substudy. The
only information provided about the subjects is the gender distribution,
and that they were “healthy college students.” 

There is no indication in the results or discussion sections of the
publication that any subjects withdrew from participation in any of the
three sub-studies. 

The presentation by EPA described the compound acrolein as “highly
toxic,” although clarification from HSRB members familiar with
acrolein indicated that at the exposure levels of this study, acrolein
was not considered “highly toxic.” Airborne concentrations higher
than those used in this study (between 2 and 5 ppm) result in increasing
irritation over the entire respiratory tract.  Relative to this, the
doses used in this study were not considered highly toxic.  Acrolein is
highly irritating, has an odor threshold, and manifests toxicity at the
point of contact, rather than leading to systemic toxicity. 

	Critique of Study 

The prevailing standard is assumed to be the 1975 version of Declaration
of Helsinki (DoH), because the study was published in 1977. However, the
previous version of the DoH (1964) may have been in effect if the study
was conducted prior to 1975. In addition, because this study was not
performed by medical doctors, the DoH may not have applied, regardless
of when the study was conducted.  

Because this study was published in 1977, the report is missing much
information on which to base a thorough assessment of the ethical
conduct of the study. The number of individual subjects used could range
from 56 to 141 and how many may have participated in more than one test
is unknown.  There also is little information about the subjects
themselves: whether any were students or employees of the investigators;
how they were recruited; and what they were told about risks, their
freedom to withdraw, or the informed consent process.  The description
of expressing a “wish” to leave the chamber is problematic because
it does not necessarily indicate that a subject would leave if given the
opportunity; a subject may have “wished” to leave but might not
actually leave.  The Agency provided a more direct translation of the
original German language publication.  This offered the Board a more
exact understanding of the actual question being asked of the research
volunteers, which was whether they would have preferred to leave the
room, rather than did they want/need to leave the room.  There is no
information concerning compensation, possible undue influence, or the
applicable version of the DoH, or whether, in fact, the DoH would have
been applicable to research.  There was no justification for the sample
size, which raised questions as to whether more subjects than necessary
were placed at risk.  Another issue concerns whether the study should
have been stopped when indications of a wish to leave (e.g., 72 % of
subjects expressed a wish to leave the chamber at 20 minutes but did
not withdraw from the study) became evident. The subjects were tested in
groups of three and it is unclear whether the investigators analyzed the
data as it was obtained to determine irritation levels or if the
analysis was not performed until all subjects had been tested.  Thus, it
is unknown whether the study should have been stopped sooner, as
accumulating evidence suggested irritation was occurring.  

40 CFR §26.1703 forbids EPA from using research involving intentional
exposure of pregnant or nursing women or children; the research subjects
were described as college students, and thus were likely to have been at
least 18 years old.  Approximately half the subjects were female, but
the report does not indicate their reproductive or nursing status.

The Board was in agreement that there was a great deal of information
lacking in the published report of this study. In addition, several
members of the HSRB expressed discomfort in having to assess the ethics
of the research in this situation. Many members agreed with the
recommendation to EPA that as they decide whether or not to use this
information, they should carefully consider whether use of this
information would lead to more protective standards and consider whether
the information from the animal studies would suffice for its risk
assessment work.

Although a majority of members found no clear and convincing evidence
that the research was significantly ethically deficient or was so
deficient as to place subjects at risk or seriously impair the informed
consent process, members felt that they were limited by the standard of
“clear and convincing evidence”, and were uncomfortable with
assessing the conduct of this study in the absence of information
necessary to make that assessment. 

 and the potential societal benefit at the time of study conduct was not
felt by one Board member to be reasonable in relation to the risks to
subjects. 

HSRB Consensus and Rationale

There was not clear and convincing evidence that the conduct of the
Weber-Tschopp et al. study was fundamentally unethical.  In addition,
despite the lack of adequate information to assess the affirmative, most
of the HSRB agreed that there was not there clear and convincing
evidence that the conduct of the study was significantly deficient
relative to the ethical standards prevailing at the time the research
was conducted.

Completed Studies on the Therapeutic and non-Therapeutic Effects of
Administration of 4-aminopyridine

Charge to the Board

The Agency’s weight-of-evidence (WOE) document for 4-aminopyridine
describes the study design and results of three clinical trials
(Grijalva et al. 2003, Segal et al. 1999, and Van Diemen et al. 1993). 
The WOE document also discusses the Agency’s conclusion that these
studies provide sufficient information to establish a point of departure
for the assessment of the risk to humans resulting from all potential
durations of exposure to 4-AP.  Please comment on whether the studies
are sufficiently sound, from a scientific perspective, to be used to
derive a point of departure for estimating risk to humans from exposure
to 4-AP.

Board Response

To inform this question, EPA provided HSRB with a number of background
documents, including an extensive review of the health and environmental
effects of 4-AP conducted by the Agency in January 1989, as well as the
WOE document and electronic copies of the 3 published studies on which
the proposed estimates of risk are to be based.

Salient background information includes the following.  4-AP exerts its
major biological actions by blocking fast acting potassium channels. 
This effect enhances or prolongs action potentials in muscle and nerve,
and also increases transmitter release at neuronal synapses in the
periphery and brain.  4-AP is acutely toxic in animals and humans. 
Toxicity at low doses is primarily mild, including tremors, sweating,
and salivation.  The dose-effect curve is very steep.  Higher doses
cause muscular incoordination, seizures, and death.  There is little
evidence of metabolism, most of the drug appears unchanged in urine. 
There is little selectivity between humans and other mammals, or birds. 
The compound is tightly bound to soil particles and persists in the
environment for a year or longer under many circumstances.

4-AP has been used or considered for therapy of botulism, overdose of
non-depolarizing muscle relaxant drugs in surgical anesthesia, for
spinal cord injury, and for certain demyelinating disorders such as
multiple sclerosis and Guillain-Barré syndrome (FDA approval for the
latter use was granted by FDA under the orphan drug program in December
2006).

Present EPA concern is focused on the use of 4-AP as a “bird
repellant” (eg., Avitrol).  This terminology appears somewhat
misleading as the chemical does not drive birds away by adverse sensory
stimuli such as odor but by poisoning, which elicits distress calls that
warn the flock to avoid a baited area.  The database on animal studies
is quite extensive.  In fact the 1989 EPA document presents tabular
information on LD50 values for a wide variety of vertebrates and
invertebrates and numerous species of birds.  The WOE document, however,
states that “there are no reliable animal toxicity studies to derive
an appropriate point for departure for assessing human health risk”. 
The limitation appears to be that the animal studies have chiefly
focused on LD50 values—ideal for comparing potency across species and
genera but insufficient to establish a point of departure such as NOAEL,
LOAEL, or BMD10.  Hence the perceived need to rely on human toxicity
studies.  It should be noted, however, that the 1972 study by Mistov and
Uzunov (summarized by EPA, 1989) did reveal dose-related histopathology
in white rats treated for 1 or 6 months with 1 to 5 mg/kg 4-AP.  Thus,
the LOAEL in rats appears to lie at 5 mg/kg or lower.  It should also be
noted that tabular data in EPA 1989 show rats to be one of the less
sensitive species with regard to 4-AP toxicity.

EPA’s WOE document concluded that 4-AP has short residency time in the
body, and therefore “one can conclude that a single PoD value is
sufficient for risk assessments of 4-AP for different potential exposure
scenarios (short-, intermediate- or long-term exposures).” The EPA
analysis also noted that the minimal daily oral dose of 4-AP producing
side effects ranged from 5 to 30 mg/day. EPA reviewers also concluded
that 4-AP has a very steep dose-response relationship. In conclusion,
the Agency proposed a 5 mg/day (0.08 mg/kg-day) as a LOAEL, and a point
of departure for risk assessments.

  

While the studies were conducted primarily to evaluate efficacy, they
all included a discussion of safety. The investigators frequently
indicated that subjects were able to “tolerate” doses, which is
quite understandable, given the severity of the disease/injury in these
patients. In many cases it was not clear which subjects (at which doses)
suffered side effects. The lowest oral dose tested in these studies was
5 mg/day. The Agency translated this dose to 0.08 mg/kg-day.

A critique of the three studies are is provided below.  As clinical
studies, the three reference studies have major weaknesses of design or
outcome. 

Segal et al. 

Segal et al. (1999) created an active control group of 5 spinal cord
injury patients, each of whom received doses of 6 mg/day, as well as 16
patients in a high dose group. It was not clear from the article whether
any of the active control patients experienced side effects. The article
simply states, “Nervousness, giddiness or dizziness, and
gastrointestinal upset manifesting as mild abdominal cramping or nausea
were the most frequent side effects.” The frequency of these side
effects was not provided. The article also states, “All side effects
were transient, self-limited, or disappeared with changes in dosage or
the timing of drug ingestion to coincide with meals or snacks.” In
summary, it was not clear from this study the extent to which patients
who received a 6 mg/day, the lowest dose in the study, experienced side
effects.

Grijalva et. al.

Grijalva et al. (2003) completed a study with 21 spinal cord injury
patients. The lowest dose in the study was 5 mg/day. The authors
reported that 56 probable adverse reactions were registered over the
26-week study. Adverse effects (dry mouth, dizziness, and gastritis)
began with 4-AP at 5 or 10 mg/day. Fourteen patients receiving 4-AP
treatment had 26 probable adverse reactions. The authors report the
frequency of these side effects, but did not specify which side effects
occurred at which dose level.

Van Diemen et al.

Van Diemen et al. (1993) conducted a randomized, double-blind,
placebo-controlled cross-over trial with 70 multiple sclerosis patients.
In the intravenous phase of the study, parathesias occurred at a minimal
dose of 1 mg. In the oral phase of the study (69 patients), 54 of 69
(78%) patients experienced at least one side effect. Table 2 in the
article indicated that 75 side effects were observed in those patients
who received the minimal daily dose of 5 mg, so presumably some of the
patients reported multiple side effects. At 5 mg/day, these side effects
included paresthesias/dysesthesias (15), dizziness/light-headedness
(36), gait instability (11), nausea/vomiting (9), and
restlessness/anxiety (4).

Conclusion of Studies

 Overall the therapeutic effects as noted from each of the three studies
were of minor degree or of marginal statistical significance,
particularly as concerns improved function after spinal cord injury. 
Powerful placebo effects were noted, further weakening confidence in
treatment-related improvements.  These weaknesses do not in themselves
impair the potential usefulness of these studies in defining toxic
endpoints.  Unfortunately, clinical studies are typically not designed
in a manner that allows one to estimate toxic endpoints with confidence,
but rather to establish efficacy at doses that are not “overly
toxic” in relation to a therapeutic benefit. 

As for toxic signs and symptoms, the results of the Grijalva study are
difficult to interpret because adverse reactions were reported by 56% of
the treated patients and, apparently, by a still higher proportion of
the placebo controls.  The Segal study (1999) is weakened, by the
absence of a placebo control.  For this reason it is uncertain which if
any adverse effect is truly treatment-related.  Actually, Segal et al
did not specifically report adverse effects but merely stated that
“nervousness, giddiness or dizziness, and GI upset like mild abdominal
cramping or nausea were the most frequent side effects.  It therefore
seems reasonable to state that neither the Grijalva study nor the Segal
study by itself is sufficiently sound for the purpose of deriving a
point of departure for estimating risk to humans from exposure to 4-AP. 
On the other hand, the Van Diemen study (1993) was comparatively
rigorous and provided a wealth of detail on the occurrence and intensity
of treatment-related side effects.  Unfortunately, this study did not
employ doses low enough to establish a NOAEL but it does seems to
indicate that a total daily oral dosage of 5 mg was associated with
definite but mild discomfort unaccompanied by changes in blood chemistry
or EEG.  

HSRB Conclusion and Rationale

The Board concluded that the studies that the three clinical studies, 
Grijalva et al. 2003, Segal et al. 1999, and Van Diemen et al. 1993,
were sufficiently sound, from a scientific perspective, to be used to
derive a point of departure for estimating risk to humans from exposure
to 4-AP.  Thus considering the three studies, an estimate of the LOAEL
of 0.07 mg/kg/day was determined.  The Board was reluctant to endorse
the use of a 5 mg/day (0.07 mg/kg/day) LOAEL, given the multiplicity of
side effects seen among patients receiving this dose, and the steep
dose-response curve of 4-AP.  Thus, the Board cautioned that this
conclusion comes with a degree of uncertainty and advised the Agency to
take such uncertainty into account when using the published information
to arrive at a point of departure for 4-AP. 

 

Charge to the Board

Please comment on the following:

1) Is there clear and convincing evidence that the conduct of any of 
the three clinical studies (Segal et al., 1999; Grijalva et al., 2003;
Van Diemen et al., 1993) any of the clinical studies was fundamentally
unethical?

2) Is there clear and convincing evidence that the conduct of any of the
clinical studies was significantly deficient relative to the ethical
standards prevailing at the time the research was conducted?

Board Response

Three studies are being evaluated. The EPA is seeking to use these
studies to derive a point of departure for estimating risks to humans
from exposure to 4-aminopyridine (4-AP). Each of them is a completed and
published study, and the information about the studies is derived solely
from the published articles.

The earliest of the three studies, by Van Diemen and colleagues, was
conducted at the Free University Hospital in Amsterdam in the early
1990s, for the primary purpose of determining the efficacy and safety of
4-AP in treating persons with multiple sclerosis. This was a cross-over
study, in which subjects were randomized between 4-AP and placebo. The
initial portion of the study involved the use of intravenous 4-AP, while
a second phase used oral 4-AP. It is stated in the article that the
study was approved by the “ethical committee” of the hospital where
it was conducted, and that the informed consent of all subjects was
obtained. No specific ethical standard is mentioned in the article,
although presumably this study would have been governed by the 1989
version of the Declaration of Helsinki.

The second of the three studies, by Segal and colleagues, was conducted
at the VA Medical Center in Long Beach California in the late 1990s, for
the primary purpose of determining the efficacy and safety of 4-AP in
treating persons with chronic traumatic spinal cord injury. One group of
subjects was randomized between low and high doses of oral 4-AP, while
another group of subjects (which had previously been exposed to this
compound) received only the high dose. It is stated in the article that
the study was “institution-approved”, and that the written informed
consent of all subjects was obtained. No specific ethical standard is
mentioned in the article, although because this study was conducted in a
VA hospital which also holds a Federal-Wide Assurance from the federal
Office for Human Research Protections, presumably this study was
governed by the Common Rule (45 CFR 46 Subpart A). 

The third of the studies, by Grijalva and colleagues, was conducted at
the Specialties Hospital, Centro Médico Nacional Siglo XXI, in Mexico
City, in 1999 and 2000, for the primary purpose of determining the
efficacy and safety of 4-AP in subjects with long-term spinal cord
injury. This was a cross-over study in which subjects were randomized
between oral 4-AP and placebo. It is stated in the article that the
study was approved by the local research committee of the hospital where
it was conducted, and by the National Research Council of the Instituto
Mexicano del Seguro Social, and that all subjects were fully informed
about the study and signed an “informed consent letter.” No specific
ethical standard is mentioned in the article, but the Institute Mexicano
del Seguro Social has an IRB listed with OHRP, and holds a Federal-Wide
Assurance, and thus presumably this study was governed by the Common
Rule (45 CFR 46 Subpart A). 

	Critique of Studies

Because each of these three studies was completed prior to the effective
date of the EPA’s final rule, “Protections for Subjects in Human
Research” (April 7, 2006), the Board is required to evaluate these
studies under a rule that allows their results to be used by the EPA
unless there is clear and convincing evidence that either a study was
fundamentally unethical, or that it was significantly deficient relative
to the ethical standards at the time the research was conducted. The
consequence of that review standard is that in the absence of
information about particular aspects of a study, all uncertainties must
be resolved in favor of the study having been properly conducted. Given
that circumstance, the Board believed it would be appropriate, in the
future where such studies that pre-date the effective date of the final
rule are being reviewed, for the EPA (as it has reported to the Board
with other similar studies previously) to attempt to collect additional
information about the studies, such as the records of IRB review
including any consent forms.

Given that such additional information was not available to the Board
with regard to these three studies, the Board can only rely on the
information presented in the published reports regarding these studies.
Those reports indicate that in each case the study was reviewed by an
IRB or an equivalent type of body. In addition, in each instance, it was
stated that informed consent of the subjects was obtained. With regard
to that aspect of the studies, there is no evidence before the Board
suggesting that the consent obtained was not appropriate under the
then-applicable standards.

iven the pre-existing information that was known about the possible
risks of using 4-AP in each instance, there was no evidence before the
Board suggesting that there was not an appropriate relationship between
risks and benefits under the then-applicable ethical standards.

In none of the published reports for these three studies is there any
evidence suggesting that any subjects under age 18 were enrolled, or
that any pregnant or nursing women were enrolled. Given that
circumstance, it appears that the provisions of 40 CFR § 26.1703 have
been complied with.

HSRB Consensus and Rationale

The Board concurred with the initial assessment of the Agency that for
each of these three clinical studies (Segal et al., 1999; Grijalva et
al., 2003; and Van Diemen et al., 1993)., there is was no clear and
convincing evidence that the conduct of the study was fundamentally
unethical, or that the conduct of the study was significantly deficient
relative to the ethical standards prevailing at the time the research
was conducted.

Design of Research on the Levels of Exposure Received by Pesticide
Handlers

Risks and Benefits of Handler Research

Charge to the Board

Will the Task Forces’ Governing Documents considered in conjunction
with the additional study- and scenario-specific information specified
above provide an adequate basis for assessing whether the risks of
conducting a particular study are justified by the expected benefits of
the proposed research?  If not, what additional information should be
provided for an IRB, EPA, and the HSRB?

	

Board Response

The Agency has provided the HSRB with a document entitled, “AHETF
Human Research Monitoring Program”. This report serves as the
“governing document” for the pesticide handler studies sponsored by
the Agricultural Handlers Exposure Task Force. Section 4 of the report
discusses study benefits, section 5 discusses risks to subjects, and
section 6 provides a benefit-risk comparison.

Both documents indicated that no direct benefits will accrue to study
participants, and that the risks to participants must be justified by
societal benefits. The primary societal benefit cited by the authors was
the ability for EPA and other regulatory agencies to use new handler
exposure to data to improve the quality of worker risk assessments. The
AHETF authors pointed out that growers or landowners who participate in
these studies would benefit through use of the test substance at no
cost.

The documents also indicated that the database would be a benefit to the
sponsors of the studies. However, a benefit to pesticide manufacturers
is not considered in the benefit-risk analyses mandated by FIFRA.

The document indicated that risks to subjects in these studies were
greater than minimal. The authors identified six types of risks:
heat-related illness, exposure to surrogate chemicals, scripting of
field activities, psychological, exposure to detergents used in
sampling, and injuries. Each of these risks was discussed in detail, and
plans to minimize risks were included in the report.

The authors concluded that the risks to study participants are
outweighed by the benefit to society in the form of high quality
exposure data for use in evaluating pesticide safety.

	Critique

The Board’s discussion focused primarily on the Agricultural Handler
Exposure Task Force (AHETF) document. The document states that there
will be no direct benefits to participants. However, it is arguable that
there could be a direct benefit of feedback on work and safety
performance. Study participants will be provided with their own results.
Knowledge of one’s ranking among workers could have educational or
motivational value for a worker. There is also the potential for direct
feedback related to safe and unsafe practices. For example, an
individual using a ground boom sprayer with a blocked nozzle might
remove the nozzle and attempt to dislodge the blockage by blowing
through it. This is clearly poor practice, but it does happen.  If the
study supervisor observed such behavior then it would be sensible for
him/her to provide advice to that individual on safe work practices. A
second example would be the identification of more widespread poor
working practices.  In the United Kingdom, studies in seed treatment
facilities documented that, contrary to good practice, some
operativesors were using compressed air to clean residues of seed
treatment products from application equipment.  This obviously created
an airborne hazard.  The fact that this practice was commonplace
indicated a need to communicate the hazard to the whole industry sector.
 In this particular case, the industry task force involved with the
studies produced a poster illustrating the findings that was distributed
to seed treatment facilities. The dissemination of this type of feedback
as part of stewardship could be seen as a benefit to pesticide handlers
in general. How such information will be communicated to participants
and the pesticide handler community in general should be described in
the protocol.

The discussion of benefits to growers, landowners, or commercial
applicators focused on the provision of free pesticide product for use
in the study. The Board agreed with the AHETF’s conclusion that the
magnitude of this benefit is not likely to result in coercion of
employees to volunteer to participate in a study. However, the timing
and conditions of how this free offer is made should be explained
explicitly as a part of each protocol’s discussion of recruitment.
This would permit the HSRB to evaluate the specifics of the arrangement,
and would permit the AHETF to assure some uniformity in how such offers
are being made in different studies (clusters) or/and scenarios.

The nature of risks is discussed thoroughly in the governing documents,
but the Board concluded that several issues warranted further clarity.
The first issue is related to the scripted nature of the studies, and
the possibility that workers may be asked to use equipment that they
would not normally use. Lack of familiarity with equipment could
increase the risk of injury; for example, a mechanical injury through
collapsing hydraulic systems or an electrocution through folded or
folding booms coming into close proximity or contact with overhead power
cables. The AHETF document indicates that workers need to be
“familiar” with the type of equipment to hbe used. The Board was
concerned that familiarity with equipment was not sufficiently described
and might be inadequate. In the United Kingdom, for example, the
requirement would be that the workers should be “competent”. The
Board recommended that future AHETF protocols identify equipment use
requirements for workers and the steps that will be taken to ensure that
workers can operate the equipment unsupervised in a safe manner.

The second issue is related to potential heat stress. The AHETF document
outlines a strategy that includes encouraging participants to drink
water or sports drinks throughout the monitoring period. It is
considered bad practice in occupational hygiene to combine working with
hazardous substances and drinking (or eating and smoking). The document
provides no advice to minimize potential inadvertent exposure during
drinking. In fact the document states that hand washes are not
necessary. The document’s only guidance is for researchers to remind
workers just prior to participation about general ways to minimize
exposure to chemicals, such as washing their hands before eating and
before removing clothing. The Board recommended that the AHETF develop a
consistent policy regarding drinking water or sports drinks and personal
hygiene.

The Board commended the AHETF for developing clear stopping rules to
minimize the risk of heat-related health concerns based on the National
Oceanic and Atmospheric Administration's (NOAA) National Weather Service
heat index. In particular, the AHETF has proposed hourly measurement of
heat and humidity when the ambient air temperature exceeds 70oF, with
increasing vigilance for signs of heat exhaustion and sunstroke on the
part of study investigators as the heat index increases. A study would
be halted when the heat index exceeds 130oF, as severe heat-related
illness is likely with prolonged physical exertion under such
conditions. AHETF researchers also recognize that direct exposure to the
sun can contribute to heat-related illness, and have proposed adjusting
the heat index accordingly for agricultural handlers working in the
direct sun. The Board was very supportive of these efforts, but
concluded that further protection of study participants was warranted.

As calculated, the National Weather Service heat index assumes that the
person in question is 5' 7" tall, 147 pounds, Caucasian, clothed in long
trousers and a short-sleeved shirt, walking at a speed of 3.1 mph in the
shade in a breeze of 6 mph, and not dripping with sweat. It is unclear
whether or not these characteristics and conditions apply to the
agricultural handlers likely to be enrolled in the proposed studies. The
whole body dosimeter used for the proposed research, for example, is
described in study documents as "long underwear". The Board recommended
that the heat index threshold be adjusted to account for the increased
amount of clothing that volunteers will wear during study participation.
The Board further recommended that AHETF document the expected levels of
physical exertion in its protocols, and consider whether a further
adjustment to the heat index threshold would be appropriate.

Finally, the Board was concerned with the use of the 130oF heat index
value as a threshold. According to the NOAA website,
"heatstroke/sunstroke [is] highly likely with continued exposure" under
such conditions (http://www.crh.noaa.gov/arx/heatindex.php). It is
important to note, however, that the website also states "sunstroke,
heat cramps or heat exhaustion [are] likely, and heat stroke [is]
possible with prolonged exposure and/or physical activity" with a heat
index of 105-129oF. The Board recommended that the AHETF revisit the
issue of heat stress, and develop a stopping point that will minimize
risk for the study participants.

The AHETF governing documents have provided the HSRB with a detailed and
thoughtful analysis of expected benefits and risks associated with the
conduct of human exposure monitoring. The Board concurred that study
participants would receive little if any direct benefit from
participation in these studies, although formalizing the nature of
safety feedback provided to participants might increase the potential
benefit. The Board also agreed that growers and landowners will receive
a modest benefit through the use of the test substances at no cost, and
that such a benefit was not unreasonable. The Board recommended that the
particular arrangements for providing the test substance be outlined in
the individual protocols. Finally, the Board concurred with the AHETF
judgment that the database developed from these studies will improve the
quality of risk assessments, and that this should be considered a
valuable societal benefit, provided that the data collected are
accurate, or at the least do not underestimate real-world exposures.

The Board recommended that AHETF pay more careful attention to the issue
of safety when asking participants to operate equipment with which they
do not normally work. In particular, AHETF should be more explicit about
the level of competency expected of workers when operating such
equipment. The Board commended AHETF for developing clear guidelines for
stopping work based on a heat index. However, the Board concluded that
the approach described in the governing documents was not fully
protective of workers, and recommended that additional attention be
given to this matter, including consideration of a lower heat index
threshold for stopping work.

	Addressing Potential Sources of Underestimation Bias

	Charge to the Board

1) Has EPA appropriately characterized the limitations on the scientific
usefulness of a handler database that does not include data
characterizing the efficiency of residue removal procedures?  If not,
what limitations have been overlooked?

Board Response

Introduction

The Agency presented its concerns regarding potential underestimation of
dermal exposure. It relied on the work of the January 2007 EPA
Scientific Advisory Panel (SAP) report, as well as a review of the
scientific literature. The SAP observed that whole body passive
dosimetry and biological monitoring comparisons did not seem to indicate
a systematic bias. The SAP concluded that bias may exist, but the extent
of potential bias between dermal exposure and biological monitoring
could not be detected because of the statistical uncertainty inherent in
the exposure and biomonitoring data.  The SAP also noted that passive
dosimetry can generate data that can be used to develop predictive
estimates of exposure for a number of different scenarios and
activities. The SAP suggested that biological monitoring could be a
useful check on passive dosimetry, but declined to suggest that require
biological monitoring be included in a protocol. EPA agreed with the
overall SAP conclusions and described some disadvantages to biological
monitoring including additional cost, logistical considerations (e.g.,
number of days required for metabolites to clear), and a lack of
acceptable biomonitoring methods for many of the surrogate compounds
proposed for the AHETF and AEATF studies.

The SAP gave particular attention to potential underestimation bias
resulting from the use of hand wash and skin rinse techniques. The
scientific literature indicates that hand wash/rinse performance can be
influenced by the chemical properties of the pesticide, such as
solubility, octanol/water partition coefficient, or formulation type;
residence time on the skin before hand rinsing is performed; type of
solvent used to rinse the hands (e.g., alcohol, soap and water);
concentration of the chemical on the skin (microgram/cm2); duration of
the exposure monitoring period; and nature of the residue (whether
exposed to pesticide concentrates, dilute sprays, or field residues).
Hand rinse removal efficiency values from several studies involving
human subjects ranged from approximately 70 to 90 percent; however, in
the case of chlorpyrifos the efficiency was approximately 20 to 40
percent).  An unpublished AEATF hand rinse efficiency study reported up
to 90 percent efficiency for didecyl dimethyl ammonium, but further
details were not provided.

 The Agency has concluded that exposure to the head/face and neck is
expected to be very low for the majority of exposure scenarios planned
by the AHETF.  The Agency pointed to an exception in the case of the
open-cab airblast application AHETF studies.

The SAP was equivocal about the need to correct the results from hand
washing for its efficiency at recovering pesticides from skin.  The SAP
would accept a rinse validation study if it could decrease the
uncertainty in exposure estimates at a reasonable cost, and be done
within approved human studies guidelines. 

The AHETF argued that no correction was needed for any potential method
bias because of reasonable congruence in exposure estimates between
studies based on biological monitoring and those using passive
dosimetry.  AEATF argued that no correction was needed in studies where
individuals will not be wearing gloves and that the hand correction
factor was reasonable.  

The EPA concluded that substantial underestimation by whole body
garments was unlikely, and that the most relevant methods to be
corrected for potential underestimations were the handwash and the
face/neck wipe. For the proposed AHETF studies, the Agency stated that
the contribution of hand exposure was expected to be minimal because all
subjects will be wearing chemical resistant gloves (CRG) during all
operations. The AEATF intends to collect data based on individuals not
wearing gloves (consumer products), but for most scenarios the Agency
again expects exposure to head, face, and neck to be low. The Agency
proposed two options to the task forces: biological monitoring could be
included as a check for potential breakthrough or other losses when
using surrogate chemicals that have well-established methods; cotton
gloves beneath the CRG and hat patches when measuring head, face, and
neck exposures could be used in scenarios for which exposures to these
body regions might be relatively high.

The Agency concluded that conditions should be established for
correcting hand rinse and face/neck wipe exposure values. The EPA
proposed a set of conditions for consideration by both task forces;
namely, if measured exposures from hands, face, and neck contribute less
than 20 percent of total exposure, no action is required; if measured
exposure contribution represents between 20 and 60 percent of total
exposure, an automatic 50 percent adjustment can be made or a validation
study can be submitted; if measured exposure contribution is greater
than 60 percent, a validation study is required.  Because validation
studies involve intentional exposure of human subjects, review of such
studies by the HSRB would be required.  

Critique

or absorbed into the skin. Visual techniques have been most useful in
qualitative evaluations of exposure and worker education.

The use of the term “passive dosimetry” to describe the dermal
sampling methods proposed by AHETF and AEATF can lead to some confusion.
First, the approach relies on both interception (whole body garments)
and removal (handwash, face/neck wipe) techniques. It is difficult to
reconcile the very active procedures required to remove chemicals from
the skin with the notion of a “passive” measurement method. Second,
the term “dosimetry” is a misnomer, as it is exposure rather than
dose that is being measured.

The Board expressed some reservation regarding the SAP and Agency
conclusion that passive dosimetry does not underestimate dermal
exposure. The conclusion was based primarily on a recent article by Ross
et al. (2007). In that article it is stated, in regard to the methods
proposed by the AHETF and AEATF, that “the passive dosimetry methods
used have never been validated.” The article then analyzed 14
concurrent or consecutive passive dosimetry-biomonitoring studies, and
reported that this analysis produced “generally similar” estimates
of absorbed dose from these two methods. However, when these datasets
were examined in detail, this conclusion was not well-supported. For
some chemicals the passive dosimetry estimates under-predicted the
biomonitoring estimates, while in others the opposite was true (see
Table 3 of the article, with reference to chlorpyrifos and atrazine,
respectively). Dose estimates based on passive dosimetry measurements
were dependent on laboratory studies of dermal absorption. Dose
estimates from biomonitoring studies were based on laboratory studies of
urinary metabolite excretion. In each case, the laboratory data are
characterized by very high variability and uncertainty. Given the
variability of these data, the Board was not persuaded that this
analysis provided a validation of passive dosimetry measurements.

that exposure to the hands, face, and neck was likely to be minimal for
most scenarios. No scientific evidence was presented to support this
conclusion. Many studies of pesticide handlers have demonstrated that
hand exposure can be a major contributor to total dermal exposure. The
heretofore cited OECD guidance document states, “Monitoring of hand
exposure may be the most important measurement in a dermal exposure
study. The contribution of the hands to total exposure has been well
documented by many investigators, using a variety of methods.”

The Board concurred with the Agency’s concerns regarding potential
under-estimation of exposure by the handwash and face/neck wipe methods.
No justification has been given to support the validity of AHETF’s
proposal to use “AOT” (or of the AEATF’s proposal to use either
propanol or “AOT”) in the hand wash or/and face/neck wipe sampling
methods to assess dermal exposure.  A great deal of information is
available suggesting that wash data can significantly underestimate
exposure and wipe data can be worse than wash data.  The Board
recommended that the task forces either generate data supporting the
efficiency of removing their surrogate pesticides from skin by washing
and by wiping, or accept the automatic adjustments being proposed by the
Agency; in fact, the adjustment for wipes could even be increased beyond
that being proposed for washing. This recommendation is consistent with
the advice provided in the 1997 OECD guidance document: “The best that
can be achieved for a hand wash or hand rinse method is a laboratory
validation of the efficiency of recovery of material from the hands of
human volunteers.”

Existing data clearly indicate that adsorption (more than absorption) of
certain pesticides can occur within a matter of minutes after the
exposure has occurred.  For example, data presented in Fenske and Lu
(1994) show that several handwashings recovered less than 50% of
chlorpyrifos from the skin immediately after exposure, and recovered
only about 20% from the hands one hour after exposure.  The handwashing
efficiency data summarized in Table 1 of a review by Brouwer et al.
(2000) range from 23 to 96%, an even wider range than the values
summarized in the Agency’s presentation.  These data indicate that the
results from washing can range from negligibly biased to a four-fold
underestimation of the true exposure, although the frequent washings
indicated within the preliminary protocols may limit the bias to about
two-fold.  

In contrast to hand washing, much of the hand wipe data presented (such
as the 10% mean recovery with a CV of 33% for azinphos-methyl from
Fenske et al. (1999) and the many chemicals with circa 50% mean recovery
with a similar CV in Table 2 of the review by Brouwer et al. (2000))
indicate that head/neck wipes may be both more biased and more variable
than hand washes. The above data support the Agency’s proposal that,
lacking a validation study of the wash method, a 50% adjustment
(multiply the results by 2x) should be applied.  The above data support
an even larger adjustment (of circa 3x) for unvalidated wipe data.  In
both cases, a validation study for recovery efficiency of wash or wipe
samples can be tested in vitro and would not require human exposure
testing. Some Board members also recommended that the Agency explore the
use of modeling to adjust hand exposure, and offered as an example an
algorithm based on some of the literature cited by the Agency in the SAP
documentation.

Charge to the Board

2) Has EPA identified the relevant scientific and practical
considerations affecting the choice to include biomonitoring, and has
EPA appropriately characterized the limitations on the scientific
usefulness of the resulting data if no biomonitoring is conducted?  If
not, what other considerations should bear on a decision to conduct
biomonitoring in addition to WBD?

Board Response

The HSRB discussed the perspective that, although inclusion of
biomonitoring data in the conduct of the AEATF and AHETF research
programs was intellectually satisfying, this was not an easy task. 
Furthermore, the difficulties in assessing the utility of biomonitoring
in the conduct of the Task Force’s programs were evidenced by the 
response previously provided by the EPA’s Scientific Advisory Panel. 
After lengthy discussion, the HSRB recommended that biomonitoring did
not need to be included in the AEATF and AHETF programs.  The Board felt
that the EPA had not fully characterized the scientific and practical
issues and considerations relative to the use of biomonitoring.  If
anything, the HSRB added to the reasons to not use biomonitoring, with
the following points emphasized:

The goal of the Task Force programs is to describe exposure from a
particular use scenario.  Biomonitoring provides data that is
chemical-specific rather than scenario-specific.  Hence, whole body
dosimetry is the most appropriate measure by which to provide an
estimate of exposure under a specific use condition.   

Given the list of surrogate chemicals provided by the Task Forces, it is
not clear whether biomonitoring is technically feasible and would
provide reliable data.  Technical feasibility requires that there are
well-established analytical methods in place along with knowledge of the
metabolism and kinetics of a given compound in order to accurately
assess internal dosimetry.  Such studies would also increase the
complexity of the design and execution, as they would require pre- and
post-exposure sample collection, with the post-exposure time period
determined by the kinetic properties of the compound.

Biomonitoring is important to understanding, determining and estimating
risk.  The Task Force programs are focused on establishing exposure so
as to assess risk, and any risk assessment would be performed in a
chemical-specific manner.

Additional points raised during Board discussion included observations
that requiring concurrent biomonitoring would impose additional burdens
on participants and would severely restrict study participants to those
with no recent prior or immediately subsequent exposure to the chemical,
a restriction that has the potential to seriously bias the results. 
Inclusion of biomonitoring would also restrict the range of surrogate
chemicals that have sufficiently sensitive metabolites to be useful at
the low levels expected in these studies.  Furthermore, the variability
implicit in back-calculating any detectable biomonitoring data to dermal
dose (necessary for use in the data base) is likely to add as much
uncertainty as clarity to the conclusions.  .  

HSRB Consensus and Rationale

The HSRB was comfortable with not including concurrent biomonitoring in
the protocols.  In fact, the Board recommended that the use of
additional monitoring units was more appropriate than the inclusion of
biomonitoring in these programs.  

QA and QC Controls

Charge to the Board

Do the Task Forces’ Standard Operating Procedures appear adequate to
ensure that the data resulting from the proposed research will be of
high quality?  If not, what other Quality Assurance or Quality Control
procedures need to be addressed?

Board Response

The HSRB noted that the volume of Standard Operating Procedures (SOPs)
provided by the Task Forces represents a significant effort to develop
the infrastructure required to develop the procedures that govern this
work. To this point, it was noted that several of the governing
documents provided to the Board contained information that was relevant
to the SOPs, and should be added as appropriate.  Overall, the SOPs
outlining the overall administration, report generation and quality
assurance (QA) oversight seem reasonably complete.   The HSRB reviewers
noted two major areas that should be expanded and/or revised for
additional clarity, namely the SOPs that focused on data quality and
sample integrity and compliance.  Specific recommendations were as
follows:

The SOPs need to define what represents a “good sample.”  What
general guidance will be provided to define sample quality?  How long
after completion of work is a sample collected?  What tolerances are
allowed in targeted airflow or environmental conditions?  What
conditions determine whether a sample is to be “weathered” and how
will “weathering” be performed?

On-site spiking of samples is intended to be used for analytical
standardization and reliability.  No details were provided on how such
samples were generated, and how they were handled to simulate actual
exposure conditions.

The roles of the study director and the principal investigator are
unclear and should be expanded.

There was a recognized need for training in the execution of these
studies along with information detailing how compliance to the protocol
would be established.

The SOPs should provide for the means by which incidents relating to
lack of compliance or possible negligent conduct can be reported.

Design of scenario-level sampling strategies

Charge to the Board

With regard to the AHETF and AEATF plans to conduct their proposed
handler research using purposive diversity sampling strategies: 

1) Has EPA identified the relevant scientific and practical
considerations affecting the choice of a strategy for sample selection? 
If not, what other considerations should bear on the choice?

Board Response to the Charge

.  For example, the National Health and Nutrition Examination Survey
(NHANES) has used complex probability sampling to sample a
representative U.S. population.  Complications Considerations associated
with complex probability sampling for the Task Force exposure monitoring
programs include likelyits high cost, the absence of a sampling frame,
and the likelihood of significantissues related to selection bias. 
Selection bias is a particular issue for these protocols because the
studies will only use volunteers, increasing chances of bias.

In light of these issues, the Agency and the task forces have considered
two alternative sampling strategies: purposive representative sampling
and purposive diversity sampling (PDS). Purposive representative
sampling captures a small sample of handler-days that is a
“miniature” of the target population, with respect to important
factors concerning the range and extent of exposure, while PDS captures
a small sample of handler-days that are diverse with respect to factors
related to the range and extent of exposure.  The task forces have
proposed PDS as the strategy more likely to reflect a broad range of
heterogeneous conditions. PDS can be diversified on the amount of active
ingredient handled, the individual (MU), location and time, and other
factors (such as equipment type, crops, rates, and micro-location). 
Site selection will emphasize more common conditions and the task forces
will be required to provide a rationale and/or justification for
selection of sites or site conditions based on diversity criteria.

The Agency has indicatedTask Forces’ statistical consultants have
argued that PDS permits a non-random sample to perform at least as well
as a small, same-sized probability sample.  It provides greater
assurance of obtaining a sample that reflects a broad range of
conditions, and makes it less likely that high end or low end exposure
conditions would be missed.  Augmenting scenario data with new clusters
in the future would be straightforward, and conditions of interest would
be easier to target. Nonetheless, the Agencyit was acknowledged that PDS
is not a probability-based sample and can only be used to establish a
surrogate distribution of exposures. A surrogate distribution cannot be
equated to the actual distribution in a target population using pure
statistical sampling theory; however, PDS can capture major aspects of
an actual distribution. The Task Forces assert that the Rresults using
this type of sample are not expected to be substantially different from
those derived using a small, same-sized cluster random sample. 
Therefore the Task Forces argue PDS should also is be considered to be
adequate for practical regulatory purposes.

The EPA’s Scientific Advisory Panel (SAP) has expressed concern with
the proposed purposive nature of sample selection because PDS assumes
underlying random selection can be used to estimate sample sizes.  In
Appendix C of its January 2007 report, the SAP provided a discussion of
potential for bias and an alternative stratified approach.  The SAP
expressed concern that use of a non-probability sample would essentially
preclude consideration of appropriate weighting to estimate
distributional parameters including means, standard deviations, upper
percentiles, etc.  Thus, the SAP recommended an informal approach for
identifying top factors and for assigning probability weights to
approximate frequencies. In response to the SAP concerns, the task
forces have outlined the constraints regarding available data and
resources. The Agency plans to evaluate the data and documentation that
will be submitted by the task forces to support their approach. 
However, given the unique aspects of this monitoring program and its
relatively small size, the Agency continues to believe that PDS is
adequately representative of the target population and can be used to
develop exposure assessments of occupational handler populations.

	Critique

The Board commended the Agency and the task forces for the work they
have conducted over the past year to develop a viable sampling strategy
for pesticide handler and consumer exposures. However, the Board had a
number of questions regarding the sampling strategy, and in particular
the selection of purposive diversity sampling as the foundation for
these studies. The more practical aspects of the sampling strategy are
addressed in this section of the Board’s report, while certain
scientific aspects are addressed below in response to specific charge
questions put forth by the Agency.

, accompanied by an appropriate rationale and justification for the
ranking. This ranking would then inform the study design in terms of how
sampling sites and individual participants are selected.

HSRB Consensus and Rationale

whereby the critical variables associated with exposure are ranked,
accompanied by an appropriate rationale and justification.

Charge to the Board

2) Does the HSRB agree with EPA that the Task Forces should provide
scenario-specific information about the availability of data to identify
significant variables (other than AaiH) potentially influencing exposure
and about the feasibility of developing a sampling strategy to address
those variables quantitatively?   If not, what additional information is
needed?

Board Response to the Charge

To be scientifically assessed, and for the Board to provide its
scientific advice, scenario-specific information needs to be provided.
This information would necessarily include significant variables that
would potentially influence exposure and its effect, in addition to
AaiH, and a feasible sampling strategy, that would be essential to meet
the scientific criteria to provide reliable and useful data. Such would
reaffirm the EPA Scientific Advisory Panel’s  recommendation that all
major factors of importance be included in each scenario-specific study
to be conducted. The information could be provided briefly for the Board
to consider, not necessarily to the extent provided by the AEATF
example.  It could be presented as one would a study design for the
scenario-specific study to be evaluated (with appropriate references)
and would contain in such design not only the specifics of the
population (including its size) to be studied but also the list of
variables and how they were to be collected and analyzed.  It could also
briefly respond to the scientific criteria proposed previously by the
Board.  

Critical to the Board’s evaluation, and to the representativeness and
usefulness of the exposure data collected and provided, would be the
information as to the relationship between the scenario-specific
exposure assessment and the representative exposure in such scenarios in
the target population.

In terms of the relevant variables, one could group them into those that
are essential or less so, and whether they would significantly affect
the exposure (or exposure scenario and data collection) or might be
important to have. Description of the target population and the subjects
selected are essential, including inclusion and exclusion criteria.
Description of the primary measuring instruments and how they are to be
used is essential.  Because environmental conditions are significant
determinants of exposure, essential environmental variables (e.g., site
description, temperature and humidity) would have to be measured, and
others (e.g., wind speed, microclimatic conditions [including, presence
of significant factors that affect airflow]) might or might not be
necessary. The subjects’ type of external clothing could be considered
essential, and the subjects’ work history with the type of application
and pesticide would be important, whereas minor differences between the
subjects would not; gender and age would be necessary. It might be
important to record, if possible, the subjects’ health status and any
physical deformities that might influence how they handle the pesticide
and perform the application required.  It would be worth noting by
observation and recording significant features of the handling and
application.  Of course, prior experience and documented studies of this
nature will illustrate what was and should have been critically
measured.

Known variability in similar data previously collected, including inter-
and inter-subject variability in exposures, would even help design the
study and determine the type and size of the population to be studied.
Finally, and repetitiously, how one would analyze these variables is
important to know.

Charge to the Board

3) Has EPA appropriately characterized the limitations on the scientific
usefulness of the resulting data attributable to the choice of the
sampling strategy?  If not, what has EPA overlooked?

Board Response

 argument in favor of purposive sampling. The reports are very well
written and the issue of sampling has clearly received much  of
attention. The task forces should be commended for a thorough and rather
sophisticated analysis. Many of their points are well taken. The
limitations of the proposed sampling strategy have not been fully
discussed.

No one knows what factors affect exposure. It appears that not only are
there a myriad factors but that their importance is not so clear. The
only way to protect against potentially significant biases introduced by
unknown or unaccounted for factors is through randomization.

Purposive sampling has a definite role to play in qualitative research
and in fact is most popular common in applied social sciences
applications. The goal in qualitative studies is often to explore and
describe a universe. In quantitative studies, in contrast, the goal is
typically to obtain a point estimate of some quantity and also a measure
of the uncertainty around that estimate. Because purposive sampling
includes no random mechanism, the machinery of probability theory is not
available to the researcher and thus an estimate of the uncertainty
around point estimates cannot be computed. As a consequence, results
from a purposive sample cannot be generalized beyond the sample except
on faith. An example of faith is the discussion in the reports that
state that the estimated exposure distributions obtained from the
purposive samples will approximate the true exposure distributions at
least in the major attributes. 

It is true that purposive sampling or its close relative, judgment
sampling, is sometimes justifiable when the sample is very small.
However, this is true if in addition, the universe is also small and its
characteristics are known to the investigator. One argument given
against the use of a probability sample in the report is that the
sampling frame (the universe) is unknown and difficult to characterize.
Unfortunately, this fact also diminishes considerably the advantages of
purposive sampling even in small samples (see, e.g., Jessen, 1978,
Statistical Survey Techniques one of the standard reference survey
sampling books).

A basic question arises: “ is how small does the sample have to be
before non-random selection is a better option?”.  The answer is
“very small.”  It has been argued (again, see Jessen and others)
that with sample sizes as small as about 8 or 10, the advantages of the
non-random sampling mechanism vanish. 

 and not 5 as is argued in the Governing Documents. If all MUs collected
from the five clusters will be combined to estimate the distribution of
exposures, then the sample size we must focus on is that of the combined
(across clusters) sample which here is 25 or so. By the time we have
reached that number there is no real excuse for not carrying out at
least a semi-random selection procedure at the handler-day level.

. However, a single sampler (even over repeated sampling exercises) can
introduce very large biases that are very difficult (or impossible) to
quantify or anticipate. 

Reliability (defined as the closeness of each observation to its own
average over repeated trials) is typically higher in purposive samples
than in random samples of similar size. Accuracy (also known as
validity, a combination of bias and sampling error and a measure of
closeness of an estimator to the targeted value) however is difficult to
forecast and can be quite low. Because non-random samples provide no
means for computing accuracy, the value of such samples is rather
questionable.

. Suppose that the five clusters are selected purposively. Given a
cluster and relative to the actual cost of collecting the dosimetry
data, it should be possible for example to first randomly select firms
and within firms, randomly select handler-days. Not everyone will agree
to participate and not everyone might be handling the desired products
on the selected dates, but all that will require is a few additional
visits or phone calls to the selected firms. 

Purposively select locations (e.g., counties or group of counties).

Within location, list farming operations. These data are available such
as through  NASS, extension agents, etc.

Roughly stratify producers by crop and by size.

 Observe them next time they apply the chemical of interest.

HSRB Consensus and Rationale

 Error estimates and other estimates relevant to determination
quantities in the AEATF and AHETF reports are based on strong and
un-testable assumptions. 

Statistical justification for number of clusters and monitoring units

	Charge to the Board

What additional information, if any, would the HSRB need to assess the
adequacy of the justification for the number of clusters and number of
MUs in specific AHETF and AEATF study proposals?

	Board Response to the Charge

The primary objective of the AHETF and AEATF II human exposure
monitoring programs is to collect sufficient data for each handler
scenario to characterize the distribution of the exposure level, both
dermal and inhalation.  In other words, both programs are interested in
knowing the statistical distribution of the exposure level within an
acceptable bound (K) for their relative accuracy.  The fold relative
accuracy (fRA) measures how far the sample estimate is from the true
parameter in a relative sense.  The sample size estimation for both
programs is based on the same justification of a 3-fold accuracy (K=3),
i.e. fRA is less than or equal to 3, for the parameter of interest.  

, and as such the sampling of the clusters and the MUs within clusters
is not random.  Therefore it requires a set of assumptions for the
“surrogate sampling model”:

Observed exposures are viewed as arising (at least approximately) from a
random sample of clusters and then from a random sample of MUs within
each cluster.

The sampling distribution of normalized exposures within and between
clusters is, at least approximately, lognormal.

The second assumption is reasonable, and even if it is violated, its
impact should be minimal.  However, the first assumption is problematic.
 Depending on how the clusters and the MUs within clusters are selected,
bias can be introduced in a way that cannot be corrected. 

Therefore the sample size justification based on Monte Carlo simulations
has its limits.

Under these assumptions, the sample size is estimated based on the
nested variance component model for the normalized exposure level.  In
order to determine the relative accuracy of the estimates of the
parameters associated with the statistical distribution for the exposure
level, one needs as design parameters reasonable estimates for the
geometric standard deviation (GSD) of the exposure level and the
“intra-cluster” correlation coefficient, i.e. intraclass correlation
coefficient (ICC) due to cluster sampling.

Nested lognormal variance component assumptions were used in a
surrogate-sampling model to determine the sample sizes necessary to
achieve a 3-fold relative accuracy of distributional parameter
estimates. Reasonable values for the GSD and the ICC of exposure
normalized by the amount of ai handled were obtained from an analysis of
existing data.

The Board provides specific comments to each task force proposal as
noted below

AEATF II: A GSD of 2.86 was derived from four dermal exposure monitoring
studies (three by the Chemical Manufacturers Association in wipe, mop
and aerosol-hands setting and one by the Pesticide Handlers Exposure
Database in aerosol), a coefficient of variation 1.42 was derived from
the log-scale standard deviation under the assumption of a lognormal
distribution, and an ICC between 0 (independence among observations
within a cluster) and 0.3 (a moderate dependence) was assumed.
Simulation analyses indicate that Nc=3 clusters with Nm=6 MUs per
cluster will achieve the desired benchmark goal and is more
cost-effective than other feasible configurations.

The recommended sample size of five clusters with five MUs per cluster
for the AHETF program and three clusters with six MUs per cluster for
the AEATF II program is considered a ‘default’ or ‘standard’
configuration only.  It strictly applies only to scenarios without
existing data and when the default variability is GSD=4 or 2.86 and
ICC=0.3, respectively, for the AHETF and AEATF II program, and benchmark
accuracy (K=3) is considered reasonable.  In other cases, the simulation
techniques can be used to develop optimal sampling plans for each
scenario it addresses.

The AHETF and AEATF II human exposure monitoring programs have done an
outstanding job of considering the effects of both the numbers of
clusters and the numbers of MUs within each cluster.

Given the assumptions made regarding the surrogate sampling, the
Governing Documents from the AHETF and the AEATF II human exposure
monitoring programs provide a very thorough justification for the sample
size in terms of the number of clusters and the number of MUs per
cluster in specific AHETF and AEATF II study proposals.  The sample size
justification includes determination of feasible values of Nc and Nm,
optimal configuration of Nc and Nm based on relative cost of sampling
cluster vs MUs, sensitivity of the relative accuracy bound to the GSD
and ICC, and the impact of unequal number of MUs per cluster which may
be expected when MUs drop out.  It also recognizes its limitations and
suggests scenario-specific simulation studies to estimate adequate
sample size.

The AEATF II program should give strong consideration to increasing the
number of monitoring units to 25 with no more than 5 MUs per cluster,
which results in at least five clusters per scenario.

HSRB Consensus and Rationale

As the sample size justification for the AEATF II program is based on
the ICC estimate from the AHETF program, it is recommended that the
AEATF II program update the proposed sample size based on their study in
the future to verify whether the ICC estimates agree with that from the
AHETF program.

	Within-Worker variability

	

	Charge to the Board

Has EPA appropriately characterized the limitations on the scientific
usefulness of a database that does not include repeated measures?  If
not, what limitations has EPA overlooked?

	Board Response

A database that does not include repeated measures will have limited
usefulness in the context of some analysis goals but not in the context
of others.

If the objective of the study is to obtain an estimate of the mean
exposure of workers in a given scenario, then a single observation per
person will suffice. This is true even when the quantity of interest is
a median, a geometric mean or in general, any distributional attribute
associated with the center of the distribution.

For regulatory purposes, the EPA needs an estimate of the exposure
distribution for a given scenario whose variability reflects the
between-person variance in exposure. We refer to this as the
distribution of usual exposures under a given scenario. Ideally, this
usual exposure distribution would be estimated from the mean handler
exposures computed from observing each handler during a large number of
randomly selected days. This is clearly an impractical approach.
Alternatively, it can also be estimated from a database that includes at
least one replicate observation on at least a randomly selected
sub-sample by fitting the appropriate random effects model (see, e.g.,
Nusser, Carriquiry, Dodd and Fuller, Journal of the American Statistical
Association, 1996; Carriquiry, Public Health Nutrition, 1999;
Carriquiry, Journal of Nutrition,2003). This approach has been
recommended by both the US National Academy of Sciences (2002) and by
the World Health Organization (2006) for estimating the distributions of
usual exposures to components in food and drinking water.

The estimated distribution that is based on a single measurement,
however, has a variance that includes the between-person variance and
the within-person (or day-to-day) variance. The latter is problematic
and inflates the overall variance of the exposure distribution by an
amount that can be quite significant. As a consequence, upper-tail
quantiles of the exposure distribution tend to be overestimated. We
refer to this estimate as the one-day estimate of the distribution. 

From a regulatory stand-point, using the one-day distribution in lieu of
the usual exposure distribution tends to be conservative. That is,
estimated percentiles such as the 95th percentile will typically be
larger than what they would be if estimated using the distribution with
the correct variance. In this sense, the one-day distribution is
protective.  If the sample of MUs in each scenario were large enough to
allow estimation of the one-day distribution, then using the one-day
distribution as a proxy for the usual exposure distribution would be
acceptable.

The reports by the AHETF and AEATF argue strongly against collecting
replicate observations. The two main arguments are the following:

Cost: given the already rather small sample sizes per scenario,
collecting replicate observations would imply increasing the number of
measurements since reducing the number of MUs is not reasonable.

The distribution of mean exposures can be analytically derived given
information about the one-day distribution of exposures.

  

The second argument, however, is not convincing. First, it is not at all
a given that the log-normal is the correct probability model for
exposures under all possible scenarios. Even if the log-normal model was
the appropriate model, the report confuses the distribution of
within-person means with the distribution of the mean of a log-normal.
Those two distributions are in general not the same unless we are
willing to believe that the within-worker exposure distribution (i.e.,
the distribution of exposures in a worker observed during a very large
number of days) is also log-normal.  Thus, without imposing the
log-normal model at the worker level as well, the Board was not sure it
is possible to analytically derive the distribution of the conditional
expectation of daily exposure given handler (which is the distribution
of interest).

The task forces propose “borrowing” an estimate of the intra-class
correlation in order the “correct” the one-day distribution so that
it will better approximate the usual exposure distribution. In
principle, using an external estimate of the within-person variance (or
of the ICC) might be reasonable, but the actual estimate needs to be
carefully selected and very well justified. Is it reasonable to expect,
for example, that the within-person variance in exposure will be the
same across scenarios? 

The limitations introduced by the lack of replication and the assumed
model are compounded by the very small sample sizes. Without imposing
the log-normal model on the data it would not even be possible to obtain
a point estimate of the 95th percentile (at least a reliable one) let
alone get an estimate of its standard error.  In addition to allowing
for estimation of the two relevant variance components, replicate
observations would increase the sample size even after accounting for
correlation between observations collected from the same person.

e

	Subject recruitment and enrollment issues

	Charge to the Board

1) Does the Board agree that the Governing Documents and associated SOPs
of the AHETF and AEATF research programs include comprehensive and
appropriate protections for human subjects of the research?  If not,
what has been overlooked?

Board Response

The Board was very impressed with the quality of the work done, and the
amount of effort that has clearly gone into producing these documents.
The Board had only a few suggestions:

The documents might provide greater clarity regarding input from
organizations that represent the subjects (such as farm labor
organizations or advocates). For example, at page 94 of 468 of the AHETF
materials, it indicated that such organizations “might” be invited
to provide input with regard to a particular study. It could be helpful
to provide some guidance regarding those situations in which the
participation of such organizations would be especially important. In
addition, such organizations could also provide useful advice with
regard to the Governing Documents and associated SOPs themselves, and
not merely limited to particular protocols.

The sample consent form that was provided would benefit from increased
attention to the reading level. In a variety of places (e.g., in
discussing how compounds will be applied to “vertical” and
“horizontal” surfaces), the language could be rewritten at a lower
reading level (e.g., by including examples referring to floors, walls,
and blinds), particularly given the likely educational background of
many of the subjects.

With regard to pregnancy testing of subjects, the AEATF and AHETF
documents appear to take different approaches. The former group tests
only subjects under age 50, while the latter group tests all female
subjects. It would seem appropriate to use the same standard in both
types of studies, or else to provide a justification for the difference.

Also on the issue of pregnancy testing, the documents mention that if a
pregnancy test is positive, the records relating to that person would be
discarded. It would be better to use a word such as “shredded,” to
make it clearer that the documents will be destroyed, not merely put
aside or thrown out where they could be accessed by others.  .

). By taking this measure, it would decrease the likelihood that an
employer could determine which employees chose not to participate,
because the employer could not be sure that some of them were merely
unable to participate due to the limit on participation.

	HSRB Consensus and Rationale

The Board agreed that the Governing Documents and associated SOPs do
include comprehensive and appropriate protections for human subjects.  

Charge to the Board

2) In singling out the handling of language differences as an area
requiring further refinement, has EPA overlooked other areas in need of
revision?  If so, what?

	Board Response

The Board reaffirmed the importance of making all paper and electronic
documents distributed to study volunteers available in both English and
Spanish. The Board also expressed strong support for the need for both
Spanish- and English-speaking study staff to be available at study sites
to answer any questions that research volunteers may have regarding the
study or their participation in the study. Spanish- and English-speaking
staff should also be available during other study interactions with
research volunteers, including data-collection encounters and via
telephone lines that are made available to subjects who may have
questions about their participation in the study or rights as a research
volunteer. The Board expressed concerns about reliance on translators of
convenience, such as co-workers and others who may lack sufficient
familiarity with the study.

The Board expressed support for the Agency's proposal to have impartial
third-party witnesses observe the consent process when a research
subject is unable to read relevant study documents. As specific studies
are proposed, however, it will be important for investigators to
describe the procedures to be employed in recruiting these witnesses. It
would be inappropriate, for example, to ask translators to serve as
witnesses (as suggested in the materials reviewed by the Board), because
one of the main purposes of employing a witness is to ensure that the
communication of study-related materials is adequate (and the translator
would be conflicted with regard to that assessment). If feasible, the
Agency may wish to consider using impartial "consent monitors" or
"research subject advocates" as witnesses, as is increasingly done in
certain clinical studies.

The Board also returned to an issue raised at the last Board meeting. At
that earlier meeting the Board suggested that in many areas of the U.S.,
agricultural workers often speak neither English nor Spanish but instead
speak another language, such as one of several indigenous languages of
Northern Mexico. The Agency may wish to consider this possibility (and
its implications for sample bias and just distribution of research
benefits and risks) in deciding whether to restrict eligibility to
English- and Spanish-speaking subjects.

HSRB Consensus and Rationale

The Board agreed that the handling of language differences is an area
requiring further refinement and is appropriate to protections for human
subjects.  Related issues include mechanisms to ensure understanding and
voluntariness in the consent process. 

REFERENCES

Brouwer DH, Boeniger MF, Van Hemmen J. 2000. Hand Wash and Manual Skin
Wipes.  Ann. Occup. Hyg. 44(7):501-510.

Carley, J.M., and K. Sweeney. (2007a) Science and Ethics Review of
Protocol for Human Study of Mosquito Repellent Performance. Ref: Carroll
2007a. Unpublished memorandum, dated May 24, 2007.

Carroll, S. (2007a) Efficacy Test of KBR 2023 (Picaridin;
Icaridin)-Based Personal Insect Repellents (20% Cream and 20% Spray)
with Mosquitoes under Field Conditions: Efficacy Test Protocol LNX-001.
Unpublished document prepared by Carroll-Loye Biological Research, dated
April 10, 2007.

Carroll, S. (2007b) Amendments to Protocol LNX-001. Unpublished document
prepared by Carroll-Loye Biological Research, dated June 14, 2007

Carley, J.M., and K. Sweeney. (2007b) Science and Ethics Review of
Protocol for Human Study of Mosquito Repellent Performance. Ref: Spero
2007. Unpublished memorandum, dated May 24, 2007.

  SEQ CHAPTER \h \r 1 Fenske RA, Lu C (1994). Determination of handwash
removal efficiency: incomplete removal of the pesticide chlorpyrifos
from skin by standard handwash techniques.  AM. Ind. Hyg. Assoc. J.
55(5):425-432.

Fenske, RA Simcox, NJ, Camp, JE, Hines, CJ (1999). Comparison of three
methods for assessment of hand exposure to azinphos-methyl (Guthion)
during apple thinning. Appl. Occup. Env. Hyg. 14:618-623.

OECD. 1997. Guidance Document for the Conduct of Studies of Occupational
Exposure to Pesticides During Agricultural Application. OECD
Environmental Health and Safety Publications, Series on Testing and
Assessment, No. 9. Environment Directorate. Organization for Economic
Cooperation and Development. Paris. [Accessed at (  HYPERLINK
"http://www.oecd.org/ehs/"  http://www.oecd.org/ehs/ ]

Reynolds, M. (2007) Electronic Submission to the Docket Re: Submission
of Written Comments for Inclusion in the Upcoming Meeting of the Human
Studies Review Board, June 27-29, 2007; EPA Docket ID:
EPA-HQ-ORD-2007-0403. Unpublished letter submitted to Dr. Paul Lewis,
Designated Federal Officer, Human Studies Review Board. Prepared by
toXcel, LLC, dated June 18, 2007.

Ross J, Chester G, Driver J, Lunchick C, Holden L, Rosenheck L, Barnekow
D. 2007. Comparative evaluation of absorbed dose estimates derived from
passive dosimetry measurements to those derived from biological
monitoring: Validation of exposure monitoring methodologies. J Expo Sci
Environ Epidemiol. Jun 27; [Epub ahead of print]

Spero, N. (2007) Protocol for Conducting Insect Repellent Field Efficacy
Testing on Mosquitoes Including Supporting Materials Satisfying 40 CFR
§26.1125 for Test Materials 1003715-019 & 1003715-020. Unpublished
document prepared by Insect Control & Research, Inc., under Protocol ID
G0590307001A044, dated April 11, 2007.

Weber-Tschopp, A.; Fischer, T.; Gierer, R.; and Grandjean, E. (1977)
Experimentally Induced Irritating Effects of Acrolein on Man
(unpublished English translation of “Experimentelle Reizwirkungen von
Akrolein auf den Menschen”) International Archives of Occupational and
Environmental Health 1977 (40): 117-130. 

  PAGE  1  of   NUMPAGES  65 

Neither this summary nor the Board’s full report addresses the risks
and benefits of the research proposed by AEATF-II.  OPP found the AEATF
presentation on this to be adequate.  

If all EPA has is purposive diversity sampling data then it is unclear
how EPA could estimate the bias attributable to that design. EPA
believes that it could get an estimate only by comparing the TFs’ data
with results from other research using a random sampling strategy. 
Currently there are no such data.

It is also registered for use within the U.S.  The use history is longer
outside the U.S.

If the same subjects participate in the dosimetry phase and as treated
subjects in both field tests, and if the same untreated
“experienced” subjects participated in both field tests, the minimum
number of subjects would be 20 treated (of which 10 participated in
dosimetry) + 2 untreated = 22.

If entirely different subjects participated in the dosimetry phase and
in each day of field testing, and if different untreated controls were
used on each day of field testing, the maximum number of subjects would
be 10 + 20 + 20 + 2 + 2 = 54.

Was the intention here to say the protocol meets the completeness
standards of 26.1125, or that it meets all the requirements of subparts
K and L?  See p. 21, lines 6-10 where the overall conclusion is stated
as the Board’s consensus.

This definition is consistent with EPA’s guideline standard: a bite
confirmed by another (confirming) bite in the same or the next
subsequent exposure period.  This standard is confused here with another
idea from the ICR protocol—that subjects cannot be relied on to know
when they’ve been bitten, but only ICR technicians can identify a bite
unambiguously.  Suggest deleting the word “confirmed” from the first
highlighted passage.  Suggest for the second passage “ICR staff will
determine whether to count an event as a bite.”  In the third instance
suggest “will not be counted as bites.”

The minimum is 1 and the maximum is 10.

Significance of quotation marks unclear.  If this is a quotation, what
is the source?  If the words are simply being used in their dictionary
meanings, quotation marks aren’t needed.

Unclear which “regulatory requirements of review and documentation”
are being referred to.  Is the intent to say the submission complies
with 40 CFR 26.1125?  Or are other requirements involved as well?

The EPA guidelines do not mention landings as an endpoint, but EPA has
accepted studies with landings as an endpoint.  The guidelines are
silent with respect to bites from insects with fewer than six legs on
the treated area, and describe probes as equivalent to bites.

Is the point here that the results cannot be generalized beyond the
population tested?  If so, in what sense is the study “sufficient for
assessing human risk resulting from acute inhalation exposure to
acrolein?” OPP suggests deleting “for the population tested.”

The applicability of this passage of the rule is not limited to
submitters who are also registrants.

All subsequent discussion makes it clear that one Board member is
involved.

Relevant to what?  The decision to conduct the study cannot be undone,
whether it was a good decision or not; is the point here that post hoc
benefit is never relevant to EPA’s decision whether to rely on a
completed study?

Suggest deleting sentence, which adds nothing to the discussion while
introducing irrelevant variations on important language from the
regulations that should frame the Board’s decision.  The rules address
“research involving intentional exposure of human subjects [to a
substance]”, never mentioning “deliberate exposure,” “to a toxic
substance” or “for non-medical purposes.”  

The deleted passage is an aside, irrelevant to a discussion of these
three studies.

This acknowledgement of receipt is the only mention of the AEATF-II
program in this passage.  All examples are from AHETF; all
recommendations are specific to AHETF.  

These cited section numbers seem to have fallen by the wayside.

The consensus does not address the proposal from AEATF-II, which OPP
found generally adequate.

The fact wipe sampling is not mentioned in the draft 875 GLs does not
mean that EPA would not accept its use.  To our knowledge, no regulatory
body here or abroad rejects studies including face/neck wipes.

Page 29 of the OECD document states: “Exposure to the head, neck and
face should be determined by measuring deposition on a cotton hat or
cap, on a hood as part of the whole body dosimeter, or on a head patch
as for the patch method.  Exposure to the face and neck could also be
determined through the use of wipe samples.”  

OPP finds the phrasing confusing.  We think that the material present on
skin is either adsorbed or absorbed, and hand rinse and face wipe
methods are designed to remove adsorbed materials. OPP recognizes that
these methods may not completely remove adsorbed material.

OPP recommends deletion of this paragraph.  Had the HSRB asked OPP to
explain the basis of its view, OPP would have noted that data in PHED,
as well as other studies, show that when gloves are worn, hand exposure
is minimized.  Likewise data in PHED indicate that face and neck
exposure is generally very low in most scenarios.  

Even if skin contributions are not always relatively small compared to
exposure on other parts of the body, OPP explained that we intend to
evaluate the data produced by the TFs for each scenario to determine the
relative contribution from skin residues vs. WBD residues. If these data
show a high proportion of exposure comes from skin residues, OPP plans
to require either an automatic adjustment or residue recovery efficiency
studies. 

The Board’s consensus answer to the charge question, and its
rationale, would be helpful.

The Board’s consensus answer to the charge question, and its
rationale, would be helpful.

Does this range of handler-days include both the AHETF and AEATF-II
programs?

We do not think this is necessarily a correct statement of EPA’s
understanding.  More specificity would be helpful.  Does this mean for
“occupational exposure monitoring projects?”  If so, the Agency does
not believe complex probability sampling is typical.  If not, what types
of projects are intended?

The primary purpose of the Task Forces’ proposed research is
collection of improved measurements of the range of individual handler
exposure-days for the various scenarios.  Generation of data to assess
the relationship of AaiH to exposure is a secondary goal.

OPP believes that a number of considerations bear on the decision about
which factors to study.  In OPP’s view it is critically important to
test the validity of the assumption that serves as the basis for our
approach to data requirements and risk management decisions – that
exposure is proportional to AaiH.  While the study of other factors may
be of interest, OPP thinks that they are less important to an effective
regulatory program.  

This recommendation seems to reflect an incomplete understanding of the
regulatory requirements regarding applicator certification.  Many
applicators who apply restricted use pesticides are not themselves
certified, but make applications under the supervision of a certified
applicator.  The owner is often the certified one (so they can buy
product) and they may hire others to make the application. Perhaps the
recommendation might better suggest including applicators with differing
levels of training.

This discussion is specific to the AHETF program; similar comments
concerning the AEATF-II program would be helpful.

Wouldn’t it make better sense for EPA to require the TFs to identify
and rank all critical variables, and to fully document their rationale
and justification for the ranking?  Or is the intention that EPA define
the process to be used?

See the previous comment. 

The Board’s consensus answer to the charge question, and its
rationale, would be helpful.

The rest of this discussion suggests that the Board wasn’t persuaded
by the argument.  

OPP thinks the effective sample size of a 5 X 5 sampling design would be
somewhere between 5 and 25.  As described in Snijders and Bosker (p.23),
based on an ICC of 0.3 (for sampling locations) the effective sample
size is approximately 11 for this design.  We question the usefulness of
random sampling for such a small sample size. 

How big is “conceptually large”?

More explanation would be helpful.  Is the Board not convinced that a
“quasi-random selection of MUs” would cost more than a PDS?  If not,
why?

What does the Board think is the threshold of significance here, and
what is the basis for the assertion that once stratified the costs
increases are more modest?

How would the Board suggest applying this approach to the research of
the AEATF-II?

How would the Board recommend dealing with a low rate of response by
randomly selected producers and/or handlers? How would the Board propose
dealing with a limited pool of applicators from which to select randomly
(most farmers do not have more trained handlers than they need to
perform day-to-day operations)?

This suggests that the Board thinks its recommended approach probably
would be enough more expensive that it would require reduction in the
scope of the program.  Is that the intent?  

Why does the Board think it is particularly important that exposure
estimates used for regulatory purposes be based on samples that at least
approximate a random sample?  This is not discussed.

We understand that the selection of MUs within a cluster is proposed to
be random, not purposive.  The TFs propose purposive selection of only
the sites; once a location is selected, they will randomly draw subjects
from among the workers who volunteer, assuming there are more than the
required minimum number of volunteers.

This statement does not seem to agree with the results of the TF
simulations.  The number of MUs per sampling location (i.e. cluster)
must be approximately equal to satisfy the K-factor benchmark criteria. 
If  cluster sizes vary widely, then additional locations would be
selected to collect the target total sample size.

This section of the report seems inconsistent with the Board’s
recommendation during the meeting.  During the public meeting the Board
agreed that given the expected lack of correlation of exposures within a
location, three sampling locations would likely be sufficient for
meeting the benchmark criteria for the K-factor.  Also an ICC of 0.3
based on the outdoor agricultural handler data serves as an upper bound
estimate of the correlation likely to be observed in the indoor
antimicrobial locations.  Finally, if the K-factor benchmark criterion
is not met (after a post hoc analysis of the measurements collected from
3 locations), the TF will collect MUs from additional locations.

We understand that the cost for collecting a given number of
measurements at multiple locations is substantially higher than
collecting the same number of measurements at one or a few locations. 
Costs increase if you have to rent more buildings instead of studying
more subjects per building.

Note: a similar statement is included in the summary on page 4 of this
report.

The Agency agrees that the within-person variance cannot be estimated
when there is only one observation per person.  The measurements
collected will reflect the total variance—the combined impact of
within-person variance and between-person variance.  We also agree it
will be impossible to disentangle these two sources of variation when
only one measurement is collected from each individual.  As we explained
in the meeting, however, we think the “impact of ignoring the
within-person variance” is modest and acceptable, since the Agency is
interested in the distribution of single-day exposures, not the
distribution of usual exposures.

D

G

´



 h 

 h 

|

}

-





F

F

 hqj

hyN

„@

„Ð^„@

 hqj

  hqj

hyN

üõüõüîçüçÕÄç¼çîüîüîüîüîüõµ®ü®ü®ü®ü®ü
®ü®ü®§ü§ü§ü§ü§ü§îçüçüçüçÕÄç¼çî h¥

 h¥

hyN

hyN

j™

hyN

  h 

 h 

hyN

  h 

 h 

 h 

H* h 

 h 

 h 

  h 

  h 

=

>

Y

  h 

 h 

 h 

 h 

h²+»& h 

 h 

  h 

  h 

  h 

 h 

o

p

-o

p

N਀&䘋

N਀&䘋

䠀$摧娓-N਀&䘋

摧娓-N਀&䘋

䠀$摧娓-N਀&䘋

 h 

 h 

Ž

‘

”

B

à

á

 h 

 h 

Ž

à

intended to use the Task Forces’ data to characterize the distribution
of usual exposures (i.e. the distribution of long-term average
exposures), the Board’s conclusions would be valid.

If, having considered OPP’s comments above, the Board decides to
retain this recommendation, more explanation of this assertion would be
helpful.  What are the characteristics of an efficient design?  What are
they assumed to add to the cost?  What gains in information would
result, and how are they valued?

The Board’s consensus answer to the charge question, and its
rationale, would be helpful.

More explanation of the basis for assuming vulnerability of many of the
prospective subjects would be helpful.  

While it’s possible that a janitorial service might have as many as 50
employees meeting basic eligibility criteria for an AEATF-II study, a
more typical number would be much lower, especially in agriculture.  A
farmer might have half a dozen experienced handlers; an aerial
application service might employ a similar number of pilots.

