February 11, 2008

January 17, 2008 HSRB Work Group Teleconference 

on AHETF and AEATF Protocols

Chair’s Summary

Participants

Celia B. Fisher, HSRB Chair

Jan Chambers, Member

Richard Fenske, Member

Dallas Johnson, Member

KyungMann Kim, Member

Michael Lebowitz, Member

Lois Lehman McKeeman, Member

Paul Lewis, HSRB DFO

William Jordan, EPA OPP

Work Group Goals

The HSRB Workgroup [Workgroup] was convened by teleconference to discuss
HSRB concerns about the sampling designs proposed by the Agricultural
Handler Exposure Task Force (AHETF) and the Antimicrobial Exposure
Assessment Task Force (AEATF) [TFs] for their respective pesticide
handler exposure monitoring programs. The Workgroup reviewed materials
provided by the EPA Office of Pesticide Programs (OPP) in preparation
for the April 2008 HSRB meeting. OPP anticipates presenting at least two
AEATF protocols for HSRB review at the April meeting.  The Summary of
EPA/OPP Teleconferences with AHETF submitted by William Jordan was the
primary focus of the conference call. The Workgroup members appreciated
the detailed information and summary provided by OPP. It was clear from
the summary and materials that OPP has carefully considered and in most
instances supported all of the HSRB’s earlier recommendations. 

 

The goal of the Workgroup teleconference was to (1) seek clarification
on some elements of the OPP summary document and (2) identify a set of
criteria for assessing the sampling design that could be fairly and
consistently applied to protocols from the AHETF and AEATF that will
come before the Board at a later HSRB meeting.

Points for Clarification

The Workgroup wanted to learn how OPP intended to use the data from the
AHETF and AEATF studies, and, thus, sought clarification for the Board
on a number of issues related to the design and collection of the data.

Point #5 of the EPA Summary stated that “EPA has determined that
generation of handler exposure data using a purposive diversity sampling
(PDS) design is acceptable” based on (a) the time and resources
already expended on this design, (b) the delay and extra costs
associated with developing a random sampling approach; and (c) the
ability of these data to meet OPP’s needs. “

OPP has informed the Task Forces that OPP expects their submissions to
include scenario-specific justification of their proposed sampling
strategies.  For example, in the case of research on agricultural
handlers’ exposure, the justification should contain full
documentation of:

 

The methods and rationale for selection of locale, study site, crop,
equipment, workers, etc., 

Relevant agricultural statistics and production figures, chemical
sales/use data, a description of the equipment to be used and a
rationale for considering this equipment to be representative for the
scenario.  

Identification of all professional contacts who contributed information
to the design process, with a description of their position, expertise,
and experience; 

Incorporation of random elements to be considered in each
scenario-specific design, and an explanation as to how they will be
implemented whenever feasible.  

Cost estimates for all alternatives considered, documenting the basis
and rationale for all estimates, including estimated costs of rejected
alternatives.  

To the extent that the materials provided by the Task Forces do not
contain adequate information, OPP expects to ask clarifying questions to
the TFs. 

There was general agreement among Workgroup members and OPP that (a) the
current PHED data set has many limitations and that it is advisable to
obtain a better database on handler exposure for use in future risk
assessments in almost all cases; and (b) a randomized sampling strategy
would support some statistical analyses of data that are precluded by
the use of a PDS design.  The remainder of the discussion was aimed at
ascertaining the extent to which data generated by PDS might produce
useful data. Below are the clarification questions posed to William
Jordan and a summary of his responses.

INFORMATION SESSION

Did the Task Forces provide detailed information on why randomized
sampling would be excessively expensive? 

No. OPP does not have a specific estimate of the likely increased cost
of a randomized sampling design for any scenarios.  The consultant to
EPA confirmed that a randomized sampling design would be expected to be
more expensive, probably significantly so, and primarily because of
higher costs to identify candidates and recruit subjects.  The
consultant also acknowledged the difficulty in identifying potential
participants in the studies and that ultimately, whether in a purposive
or random sampling scheme, these participants may be the same.  For most
scenarios too little is known about the target population even to
estimate with confidence the cost of a randomized sample design.

Were there other reasons, in addition to those listed in the Summary
that OPP decided to accept the PDS design?

In addition to considerations of the value of the data and the cost and
feasibility of different sampling approaches, OPP also considered how a
decision to insist on having the TFs use random sampling strategies
would affect the timing of data generation.  Assuming that the TFs
agreed to use a probability based sampling strategy, OPP estimates that
redesigning the AHETF and AEATF exposure monitoring programs to
incorporate a randomized sampling design would probably delay data
submission by another year or longer.  The TFs, however, have informed
OPP that the cost of employing random sampling strategies would
significantly increase the testing costs, to a point that they would no
longer be able to afford to conduct the research.  They indicated they
would consider ending the program or significantly reducing the number
of monitoring units per scenario. If the TFs refused to develop the data
voluntarily, OPP would have to issue Data Call-In notices to compel the
pesticide industry to generate handler exposure data.  Preparing and
issuing DCIs would probably take OPP at least several years, with no
assurance that a regulatory requirement for the data would be approved
by the Government.  Data generation and submission would take several
more years.  Given the acknowledged limitations of PHED, OPP thinks
these likely delays and the uncertainty of the eventual outcome would
not be in the best interests of currently exposed pesticide handlers.

How does OPP intend to use the AHETF and AEATF data? 

OPP plans to (a) generate conservative exposure estimates based on the
use of the high-end of the distribution and other information and (b)
determine whether exposure is proportional to the amount of active
ingredient handled in each scenario monitored.  More specifically, OPP
plans to use the data for each scenario in at least these ways:

To generate estimates of mid-range and high-end exposure from the
distribution of the data.

To determine how the amount of active ingredient handled relates to
exposure.

Determine if the proportionality assumption is not supported by data, to
look for other variables that might have significant influences on
exposure in order to develop hypotheses for examination and control in
future studies.

Can an estimate of uncertainty for high-end values be determined with
the PDS design?

No

Does the OPP’s decision to accept PDS mean that sponsors are not
expected to use randomized designs when possible?

No. Sponsors are expected to incorporate random elements into the design
whenever that is feasible—as for example, when lists are available to
identify the universe of relevant sites and when obtaining or generating
such a list is practical and economically feasible.

Are there sampling requirements for PDS protocols?

Yes.  Each scenario-specific design document must specify the sampling
frame. OPP will evaluate the frame for representativeness and bias.

WORK GROUP DISCUSSION AND RECOMMENDATIONS TO THE FULL HSRB

The Work Group discussed the potential implications of the written
materials and information session on protocol submissions, OPP
presentations and HSRB review at the April Meeting. The Work Group
concluded that it would recommend to the HSRB that:

Random sampling designs are preferred.

When random sampling is not possible, a PDS protocol must nonetheless
have a well-developed sampling frame based on knowledge of the range of
ingredient concentrations and distribution of methods used in the field.

Each protocol should be individually assessed for the feasibility of
random assignment. When random sampling is not possible each protocol
should be individually assessed for the adequacy of the PDS sampling
frame. 

With respect to the format of protocols submitted to OPP/HSRB, the
protocols should include:

A detailed description of the methods and rationale for data collection
(e.g., neck wipes)

If random sampling is not used, a detailed description of efforts made
to incorporate random elements in each scenario-specific design and why
it was not feasible (in terms of availability of information, costs, and
time) to obtain a random sample.

For both random and PDS designs, a detailed description, rationale and
justification for the scenario, selection of clusters, and what will be
done within each cluster and why. 

For all protocols, a detailed explanation of how data will be analyzed
and interpreted by AHETF & AEATF. 

For all protocols a detailed explanation of how the data is anticipated
to be analyzed by EPA and how it will be useful for EPA risk
assessments.

With respect to the format of OPP presentations to the Board:

OPP should develop a written glossary of terms (e.g., cluster, scenario)
for HSRB and public reference. This glossary should be distributed but
not summarized during OPP presentations.

For each protocol OPP should provide a brief (1 page if possible)
abstract in terms appropriate for a lay audience describing the nature
and purpose of the study and how EPA intends to use the data.

OPP’s oral presentation should not focus on details. The Work Group
believes that such detailed presentations distract from focusing
attention on those aspects of the protocol for which OPP is eliciting
Board feedback. 

OPP’s oral presentation on the science should not be a summary of the
protocol, but a focused discussion of OPP’s evaluation of why they
think the study has sufficient scientific validity; the presentation
should include questions regarding scientific validity that OPP wishes
the Board to address. 

OPP’s oral presentation should also include a description of how the
Agency plans to analyze and use the data. 

Similarly, OPP’s oral presentation should not focus on the details
regarding the protection of the human subjects as such details are
described in the written materials. Rather, a brief oral presentation
should identify those aspects of the design that OPP believes raise
human subjects concerns.

With respect to the participation of members of the AHETF and AEATF at
HSRB meetings:

Since the HSRB makes its recommendations to EPA and not directly to
sponsors, it is the responsibility of OPP and not the sponsor to present
the protocol to HSRB, along with EPA’s critique and conclusions.

Sponsors have the opportunity to express their perspectives and clarify
information during public statement periods.

During Board discussion of protocols, sponsors should be available for
additional clarifications that may be needed.

In addition, if sponsors believe that a specific point has not been
adequately addressed they should have the opportunity to alert OPP to
their concerns during the time allotted to the protocol; OPP in
consultation with the Chair and DFO may recommend to the Board that the
sponsor be heard on this issue.

With respect to criteria for “expedited review” of protocols in the
future:

The Work Group agrees with OPP that an expedited review process is
desirable for protocols similar to those that have already been
evaluated by OPP and the Board.  

The Work Group believes such a process can be developed only after the
HSRB has had experience evaluating exemplary protocols. 

 PAGE   

 PAGE   1 

