STA TUS OF
CUMULATIVE RISK
ASSESSMENT FOR
ORGANOPHOSPHA TE
PESTICIDES
U.S. EPA Office of Pesticide Programs
January 15, 2002
Revised Draft
-------
I. Introduction
A. General
This document summarizes the basic principles that underlie OPP's
approach to cumulative risk assessment. It also summarizes and explains the
information in the "Preliminary Organophosphorus Cumulative Risk Assessment,"
which was released on December 3, 2001. Other subjects presented here are
discussed more fully in the documents:
~ "A Common Mechanism of Action: The Organophosphate Pesticides,"
dated November 2, 1998;
~ "Guidance for Identifying Pesticide Chemicals and Other Substances that
Have a Common Mechanism of Toxicity," dated February 5, 1999;
~ "General Principles for Performing Aggregate Exposure and Risk
Assessments," dated November 28, 2001;
~ "Proposed Guidance on Cumulative Risk Assessment of Pesticide
Chemicals That Have a Common Mechanism of Toxicity," dated June 22,
2000;
~ "Endpoint Selection and Determination of Relative Potency in Cumulative
Hazard and Dose-Response Assessment: A Pilot Study of
Organophosphorus Pesticide Chemicals," dated September 5, 2000;
~ "Cumulative Risk: A Case Study of the Estimation of Risk From 24
Organophosphate Pesticides," dated November 9, 2000; and
~ "Preliminary Cumulative Hazard and Dose Response Assessment for
Organophosphorus Pesticides: Determination of Relative Potency and
Points of Departure for Cholinesterase Inhibition," dated July 31, 2001.
This guide is designed to assist the reader by identifying and explaining
the key features of the preliminary organophosphorus (OP) cumulative risk
assessment. The goal is to help stakeholders better understand the assessment
and the potential issues involved in the assessment and, ultimately, provide
input on the conduct and conclusions of the assessments. Because the
assessment is preliminary, some elements may change before release of the
revised assessment. Changes are possible as a result of the public comment
period on the preliminary risk assessment; review by the FIFRA Scientific
Advisory Panel scheduled for February 2002; as well as continuing work by the
Agency.
The preliminary cumulative risk assessment for the OPs was placed in the
public docket on December 3, 2001. It is available on the internet at
www.epa.gov/pesticides/cumulative. The other documents noted above are
Revised Draft
-------
posted on the Internet atwww.epa.gov/scipoly/sap/ orwww.epa.gov/trac/science.
A public comment period on the preliminary risk assessment was announced
following the opening of the docket. This comment period will close March 8,
2002.
B. Common Mechanism Group/Cumulative Assessment
Group
OPP has determined that it is appropriate to treat the organophosphates
(OPs) as sharing a common mechanism of toxicity: the inhibition of
cholinesterase activity. A preliminary cumulative assessment was conducted to
evaluate the combined risk from food, water, and residential/non-occupational
exposure resulting from all relevant uses of OPs.
All of the OPs, which have been determined to cause a common toxic
effect by the same major biochemical event, that is, inhibition of acetyl
cholinesterase form the "Common Mechanism Group" or CMG for the OPs. The
40 pesticides in the CMG include the 39 OPs that are currently registered or
have tolerances for import purposes plus a new pesticide fosthiazate.
Fosthiazate was examined in the hazard assessment to determine its relative
potency. It may be considered for registration in the future. Fosthiazate is a
potential alternative to methyl bromide. The 40 members of the CMG are listed
in Section II, "Common Mechanism Group/Technical Registrants."
However, not all of these chemicals contribute meaningfully to the OP
cumulative risk, for a variety of reasons. Therefore, some chemicals are not
included in the assessment. The chemicals that are included in the quantification
of cumulative risk are referred to as the "Cumulative Assessment Group" or CAG.
The Cumulative Assessment Group for the OPs includes 31 pesticides. Section
III, "Cumulative Assessment Group," describes the decisions leading to formation
of this group.
C. Relationship Between Individual Chemical and Cumulative
Assessments
To fully understand the goals and methods of the cumulative OP
assessment it is necessary to understand the relationship of the single chemical
OP risk assessments to the multi-chemical cumulative OP risk assessment.
Revised Draft
3
-------
Comparison of Individual and Cumulative Risk Assessment
Individual Assessment
Cumulative Assessment
Focus is on specific chemical
Goal-determine "safe" level for
most sensitive endpoint
Considers all effects and
exposures
Emphasis on the effect shared
by members of the common
mechanism group
Considers relative potency of
chemicals in the group
Must look at the likelihood of co-
occurrence of exposures
In general, the individual chemical risk assessments should be done first.
The aggregate assessments for the individual chemicals provide information
needed to define the parameters of the cumulative exposure assessment. They
permit evaluation of the strengths and weaknesses of the available data. This
information is important for directing the process for deciding whether a particular
pesticide source and/or pathway combination should be included in the
cumulative assessment. In any case, it is necessary that both the individual and
cumulative assessments be done, since they consider the risks of the chemicals
in different ways.
As noted above, the cumulative risk assessment considers only the
common mechanism effect. The effect identified as "common" may or may not
be the effect that was used as the basis for establishing an individual chemical's
endpoint. The common toxic effect may be produced at, above, or below doses
that produce other toxicological effects that are not associated with the common
mechanism of toxicity. For example, an OP may have an affect that is not
associated with cholinesterase inhibition that may occur at a different dose level
than the cholinesterase inhibition. In addition, because the emphasis is on the
common effect, the endpoint selected for the cumulative assessment may be
generally the same as in the individual assessment, for example the inhibition of
cholinesterase, while the specific measure(s) used, for example plasma, red
blood cell or brain, or specific test animal may be different for the two
assessments.
Exposures are only relevant for a cumulative assessment if they have the
potential to result in a cumulative risk. For example, for the OPs, potential for
concurrent or overlapping exposure exists because the effects on cholinesterase
may overlap given the effect can persist over several days to weeks depending
on the magnitude of exposure. This is in contrast to, for example, most chronic
and cancer endpoints for which the effect occurs after long-term exposure. In
Revised Draft
4
-------
that case, concurrent or overlapping exposures are not necessary for evaluation
of a common mechanism effect.
To analyze the potential for concurrent exposures, the exposure
assessments for the OP cumulative risk assessment must address:
~ Critical window for the common mechanism effect (i.e., time from
exposure to the pesticide and expression of the common
mechanism effect until the effect is reversed and the individual has
returned to a pre-exposed condition),
~ Regional patterns in usage, which result in exposures to multiple
chemicals that can be expected to occur only in a defined spatial or
geographic area; and
~ Temporal issues, for example, whether the pesticides are applied
during the same season or time period, so that multiple exposures
are possible, and the temporal relationship between exposures in
food, water, and the home.
The critical window of expression for the common toxic effect and exposure
duration, pattern, and frequency, therefore, become paramount in determining
where there is an opportunity for an individual to be exposed to two or more
pesticides concurrently. In addition, to maintain the appropriate relationship
among the components of the assessment (food, water, and residential), it is
necessary to maintain the appropriate demographic element of the assessment,
so for example, a two-year-old's dietary exposure would not be combined with a
homeowner applicator's exposure from treating his lawn. Finally, because the
assessment combines many data sets into a single assessment, reducing the
likelihood of compounding conservative assumptions and over-estimation bias
becomes very important in constructing the cumulative risk assessment.
Developing a modeling tool that permits the assessment of co-occurrence
is a necessary aspect of the development of cumulative methods. The model
must be able to integrate exposure through food, water, and residential/non-
occupational pathways to reflect both the probability of exposure by any given
pathway and the timing of exposures through different pathways. Therefore, the
model should reflect the exposure of discrete individuals/population members in
which routes of exposure are linked and the estimated exposures reflect the
individual's location, and other demographic characteristics of the
individual/population member such as age and weight; the time of year; the
individual's anticipated patterns of pesticide use (for residential exposure); and
the individual's history of exposure. For example,
~ if an individual's house was treated for termites today, that
Revised Draft
5
-------
exposure could continue for a period of time for that individual, but
would not be randomly spread through a population
~ similarly, for drinking water, the source of an individual's drinking
water today is likely to be the same source tomorrow, and that
spatial and temporal linkage must be preserved.
The following chart illustrates how potential exposure to an
individual/population member should consider and link temporal, spatial, and
demographic components for the specific individual/population member.
Illustration of Exposure Linkages for an Individual in the Population
Example(s) of
Individual
Characteristics
Dimension
Correlation for an Individual in the Population
~Season of the Year
Temporal
~ Drinking water exposure and residential pesticide
application pattern correlate with season of year
~ Location of home
(Urban or rural area,
region of country)
Spatial
~Drinking water estimates correlate with region of country
~Residential pesticide usage likely for region of country
~Gender
~ Person's Age
Demographic
~ Reproductive status consistent with age and gender
~Age correlates with consumption pattern, activity pattern,
inhalation rate
~ Personal preferences, behaviors, and characteristics
consistent with data on home pesticide usage and type of
home
Individual Example: An individual who is part of a population of concern is a 1-year old female, in
New England, during the winter, in a rural location without municipal water, whose food and water
consumption is that reported for her in the CSFII. She encounters potential residential pesticide use
consistent with a rural, New England location in the winter. She does not apply home pesticides, but
may come in contact with pesticides by crawling on the floor. Body weight, height, surface area,
inhalation and other biological factors are consistent with her other demographic characteristics, as
recorded in the CSFII.
Revised Draft
6
-------
The following chart summarizes the differences in the major exposure
components of the risk assessments for the individual and cumulative
assessments for food, water, and residential exposures and the resulting
differences in the outputs of the assessments.
Differences in Individual OP Chemical And
Cumulative OP Exposure Assessments
Exposure Pathways
Element to
Compare
Individual
Assessments
Cumulative
Assessment
Food
Type of
Assessment:
Probabilistic
Probabilistic
Input:
If an individual eats a particular food
item, his probability of exposure to
an individual chemical's residue is
determined only by the probability of
the residue being present on the
food. In the individual
assessments, estimates are made
for all food items, and all the
estimates are independently made,
because it can be assumed that the
probability of a single chemical
being on any given item (say
carrots) is unrelated to the
probability of it being on any other
item (say green beans) or to the
probability of other chemicals being
present on these items.
An individual's probability of
exposure to multiple chemical
residues depends not on the
additive probabilities of the single
chemical being present on a given
food item, but on the probability of
their co-occurrence on a single food
item and across the multiple food
items that the individual consumes.
These probabilities, unlike with a
single chemical, cannot be
assumed to be independent of each
other. Thus, for example, if a given
field were treated with one OP for a
particular pest, it would not be likely
that it would also be treated with the
other 15 OPs registered on that
crop for that pest. Reliance on
monitoring data and use of
composite samples allows the
assessment to capture co-
occurrence of OPs on food.
Output:
Distribution of exposures for
population of concern on a national
scale.
Distribution of exposures for
population of concern on a national
scale; however, these distributions
will also be presented as regional
distributions when integrated with
the regional assessments being
done for water and residential
exposures.
Revised Draft
7
-------
Exposure Pathways
Element to
Compare
Individual
Assessments
Cumulative
Assessment
Water
Type of
Assessment:
Deterministic
Probabilistic
Input:
Generally uses a screening level
conservative assessment, which
uses a point estimate from a
reasonable high-end exposure
scenario, which is generally
selected to represent all use areas
for a given crop. The point estimate
typically does not take into account
seasonal variations in exposure
concentrations. Thus, variations in
exposure overtime are not
considered in the screening
estimates. Such variation may be
considered in more refined
assessments, if sufficient
information is available to do so,
(e.g., water monitoring with frequent
sample intervals). Point estimates
are also used for water
consumption values.
Uses a distribution of daily pesticide
concentrations over multiple years
rather than a single point estimate,
and uses a regional approach
based on geographic location, crops
grown and agricultural practices as
opposed to having one scenario
represent all crops. Since
determining the probability of co-
occurrence or exposure to multiple
pesticides at the same time is
important to calculating total
exposure for cumulative risk
assessment, the timing of pesticide
use, the place where the pesticide
is used and the probability that it will
occur in the drinking water in one or
more regions are all accounted for
in order to develop reasonable
estimates of exposures to
pesticides in drinking water. Water
consumption values are taken from
the CSFII
Output:
Point estimate is compared to the
residue level that could be in water
and still be "safe", given the
amount of residues estimated to be
in food. This residue level is termed
the Drinking Water Level of
Comparison (DWLOC).
Distribution of exposures for
populations of concern. These
distributions are presented as
regional/location-specific estimates
designed to represent the region of
concern. They are combined with
exposure estimates from food,
using food and water consumption
data from the CSFII as the
common, linking factor.
Analysis of
Modeling
Results:
When model estimates exceed the
DWLOC, use all available
refinements. Obtain all available
monitoring data and compare to
modeled values.
Model estimates refined as
extensively as possible and
compared to available monitoring
data.
Revised Draft
8
-------
Exposure Pathways
Element to
Compare
Individual
Assessments
Cumulative
Assessment
*c3
Type of
Assessment:
Deterministic
Probabilistic
(L)
(L)
PtH
Input:
Individual exposure scenarios are
developed to represent reasonable
high-end exposures from
application (home-owner
applicators) and post-application
exposures. The scenarios are
generally taken to represent all
areas of the country. Timing of
exposure is not generally
considered (except for the duration
of exposure, for example, short-
term, intermediate-term, or long-
term).
Individual exposures are estimated
along with the probability of co-
occurrence with other exposures, all
of which are presented, not in the
context of the individual, but as
probability distributions for the
population of interest. To estimate
co-occurrence the temporal and
spatial aspects of residential use,
together with the probability of use
at any given time period are
incorporated in the assessments.
For example, termite applications
would only be considered in certain
areas of the country and lawn
exposures would only occur at
certain times of the year for most
areas of the country. To establish
these relationships, assessments
are done for separate regions and
for specific time periods.
Output:
Risk estimates for individuals for
representative scenarios, e.g.,
toddlers on a treated lawn, or
combined applicator and post-
application exposures for adults
who treat their own lawn. These
risk estimates are evaluated to
determine if the use is "safe" for the
individual/population member
exposed.
Distribution of exposures for
populations of concern, rather than
for a specific individual/population
member subject to the exposure.
These distributions are presented
as regional/location-specific
estimates designed to represent the
region of concern and are combined
with food and region-specific water
exposure estimates.
In summary, it is important to see these two different assessments
(individual chemical and cumulative) as distinct, in the questions they address,
the methods they use, and the regulatory outcome that may be appropriate.
Revised Draft
9
-------
|l^^Commo^Mechanis^^^up^^chnica^egistremts
The following table lists the 40 OPs that are currently in the common mechanism
group. This list includes the 39 OPs that are currently registered or have tolerances for
import purposes, and also includes a new chemical, fosthiazate, which was examined in
the hazard assessment to determine its relative potency. It may be considered for
registration in the future. Fosthiazate is a potential methyl bromide alternative. The
table also shows the registrant(s) primarily responsible for the data on the chemicals.
Chemical
Registrant(s)
Acephate
Valent
Azinphos methyl
Bayer; Mahkteshim-Agan
Bensulide
Gowan
Cadusafos
FMC
Chlorpyrifos
Dow
Chlorpyrifos methyl
Dow
Chlorethoxyfos
AMVAC
Coumaphos
Bayer
Diazinon
Syngenta; Mahkteshim-Agan
Dichlorvos
AMVAC
Dicrotophos
AMVAC
Dimethoate
Cheminova
Disulfoton
Bayer
Ethion
Cheminova
Ethoprop
Aventis
Ethyl Parathion
Cheminova
Fenamiphos
Bayer
Fenitrothion
Sumitomo
Fenthion
Bayer
Fosthiazate
ISK Biosciences
Malathion
Cheminova; Bayer
Methidathion
Gowan
Revised Draft
10
-------
Chemical
Registrant(s)
Methamidophos
Bayer
Methyl Parathion
Cheminova; Griffin; CerexAgri
Mevinphos
AMVAC
Naled
AMVAC
Oxydemeton Methyl (ODM)
Gowan
Phorate
BASF; Aceto
Phosalone
Aventis
Phosmet
Gowan
Phostebupirim
Bayer
Pirimiphos methyl
Agriliance
Profenofos
Syngenta
Propetamphos
Wellmark
Sulfotepp
Plant Products; Fuller
Temephos
Clark Mosquito Control
Terbufos
BASF
Tetrachlorvinphos
Boehringer Ingelheim Vetmedica; Hartz Mountain Corporation
Tribufos
Bayer
Trichlorfon
Bayer
Revised Draft
11
-------
^T7^^^3umLHativ^^ssessmen^Group
Not all of the chemicals in the common mechanism group contribute meaningfully
to the OP cumulative risk for a variety of reasons. Therefore, some chemicals are not
included in the assessment. In addition, some chemicals and some chemical/use
combinations are not evaluated quantitatively. The following summarizes which OP
chemicals the Agency has excluded from the CAG, and discusses several for which
only qualitative assessments were performed.
A. Excluded Chemicals
Ethion, ethyl parathion, sulfotepp, cadusafos, fenitrothion, temephos,
propetamphos, and coumaphos were not included in the cumulative assessment
group, for the reasons discussed below.
Ethion, ethyl parathion, and sulfotepp are not included in the
cumulative assessment group because these chemicals are being phased out
according to specific legal agreements with the registrants. These legal actions
call for a near term removal of the uses. In addition, the result of these actions in
practice is often an accelerated move away from the chemical. As a result, if the
Agency chose to include the chemicals in an assessment, it would be difficult to
estimate the continuing exposure contribution. Finally, the Agency believes,
given that these actions have already taken place, there could be an
inappropriate regulatory effect if other chemicals or uses were considered for
removal from the market now, as the result of considering these phased out uses
in the assessment. It should be noted that phased out uses of certain other
chemicals will also be excluded from the assessment.
Cadusafos, fenitrothion, temephos, and propetamphos are not
included in the cumulative assessment group because it was determined in each
of their individual assessments that there were negligible, if any, exposures.
~ Cadusafos is used exclusively on imported bananas. No detectable food
residues are expected from this use.
~ Fenitrothion has a tolerance for imported wheat gluten from Australia and
is used in the U.S. only in containerized bait stations in child resistant
packaging. Monitoring data show negligible residues for wheat gluten,
and exposure resulting from the containerized bait stations in child
resistant packaging is expected to be insignificant also.
~ Temephos is used only as a mosquito larvicide. Applications are limited
to brackish water areas where exposure to both bystanders and drinking
water is expected to be negligible.
~ Propetamphos is used only as a crack and crevice treatment. It is not
Revised Draft
12
-------
allowed to be used in structures children or the elderly occupy, such as or
including homes, schools, day-cares, hospitals, and nursing homes with
the exception of areas of food service within those structures, when food
is covered or removed prior to treatment. As the result of these
restrictions, exposure is expected to be negligible.
Coumaphos is used for direct application to livestock and to swine
bedding. The Agency anticipates that there is not likely to be appreciable
transfer to meat and milk as the result of these uses.
B. Chemicals to Be Examined Qualitatively
Three chemicals-chlorethoxyfos, phostebupirim, and profenofos-
have no detectable residues in PDP monitoring data and are each used on a
single crop However, a screening analysis for water was conducted to assess
whether their contribution to water exposure is also negligible. Tetrachlorvinphos
has only pet and livestock uses. The pet uses are not included in this
assessment due to lack of exposure data suitable for probabilistic assessment
methods. The individual chemical assessment shows risks of concern for this
use. Any possible residues resulting from the livestock use are expected to be
covered by the conservative residue estimate for meat commodities that is being
used in the assessment. Fostiazate, the new chemical which may be considered
for registration in the future, is included but has only a hazard assessment.
C. Current Status of Each Chemical
The following table summarizes the current status of the OPs in regard to
their inclusion in the cumulative assessment. For included pesticides it indicates which
assessments have been done (F = Food, W = Water, R = Residential) It also indicates
the 11 pesticides for which residential uses are registered.
Organophosphates: Current Status
Chemical/Uses
Included:
F=Food;
W= Water;
R=Residential
Included:
Screening
Assessment
Only
Excluded
Residential
Use
Registered
Acephate
/ (F,W,R)
/
Azinphos methyl
/ (F,W)
Bensulide
/ (W,R)
/
Cadusafos
/
Revised Draft
13
-------
Chemical/Uses
Included:
F=Food;
W= Water;
R=Residential
Included:
Screening
Assessment
Only
Excluded
ResiRdential
Use
Registered
Chlorethoxyfos
/(W)
Chlorpyrifos
/ (F,W)
/
(Not included
due to low risk
of remaining
uses)
Chlorpyrifos methyl
/(F)
Coumaphos
/
Diazinon
/(F,W)
Dichlorvos
/ (F, R)
Included in Water
only as a
degradate of
Naled
/
Dicrotophos
/(W)
Dimethoate
/ (F,W)
Disulfoton
/ (F,W,R)
/
Ethion
/
Ethoprop
/ (F,W)
Ethyl parathion
/
Fenamiphos
/ (F,W,R)
/
Fenitrothion
/
Fenthion
/(R)
/
Fosthiazate*
/ (Hazard Only)
Malathion
/ (F,W,R)
/
Methidathion
/ (F,W)
Methamidophos
/(F,W)
Methyl parathion
/ (F,W)
Mevinphos
/(F)
Naled
/ (F,W,R)
/
Revised Draft
14
-------
Chemical/Uses
Included:
F=Food;
W= Water;
R=Residential
Included:
Screening
Assessment
Only
Excluded
Residential
Use
Registered
Oxydemeton methyl (ODM)
/ (F,W)
Phorate
/(F,W)
Phosalone
/(F)
Phosmet
/ (F,W)
Phostebupirin
/ (W)
Pirimiphos methyl
/(F)
Profenophos
/ (W)
Propetamphos
/
Sulfotepp
/
Temephos
/
Tetrachlorvinphos
/(F,R)
/
(No quantitative
assessment
due to lack of
data-screening
level
assessment
indicates risks
of concern)
Terbufos
/(F,W)
Tribufos
/(F,W)
Trichlorfon
/(R)
/
*A new chemical being examined to determine if it might be considered for registration in the future-it is a
potential methyl bromide alternative.
Revised Draft
15
-------
IV. Endpoint Selection
A. Uncertainty Factors
1. Individual Chemical Uncertainty Factors
Chemical-specific uncertainty factors are applied, if necessary, to
the individual chemicals in the CAG, in considering the relative toxic
potency of each chemical member in the group. However, no uncertainty
factors are carried over from the individual assessments. Chemical-
specific adjustments are based on issues with the toxicity data for an
individual chemical, for example, to account for use of a LOAEL rather
than a NOAEL or use of sub-chronic data in the absence of chronic data.
These adjustments allow each chemical's database to express a uniform
effect level, allowing them to provide equivalent measures of toxicity to the
extent possible. No chemical specific uncertainty factors were used in the
preliminary OP cumulative risk assessment.
2. Group Uncertainty Factor
The group uncertainty factor for the CAG is applied after estimating
the toxicity of the group. The group uncertainty factor covers areas of
scientific uncertainty that pertain to the group as a whole rather than to an
individual chemical's database. This includes, for example, differences
between species (inter-species) and among individuals within a species
(intra-species). In addition, EPA analyzes any overall database
uncertainty. This includes any issues concerning the quality and
completeness of the database as it relates to the common toxic effect for
the group as a whole. The preliminary assessment does not specify a
group uncertainty factor because that decision has not yet been made.
3. FQPA Safety Factor Determination
The Agency is preparing a science policy paper containing
proposed guidance on the relationship between the FQPA Safety Factor
and cumulative risk assessment. This document will further the policy
development process to address questions surrounding how the FQPA
Safety Factor relates to cumulative risk assessments. The Agency
anticipates issuing this paper in mid-February 2002, shortly after the
revised generic guidance document on the FQPA Safety Factor is
released. Following this process, EPA will consider the specific case of
the OP cumulative assessment. Therefore, the preliminary OP cumulative
Revised Draft
16
-------
risk assessment has not considered the FQPA safety factor.
C. Endpoint Selection & Relative Potency of Chemicals
1. Calculating Relative Potency Factors
Once an endpoint has been selected and before an exposure
assessment can be done, the chemicals must be ranked according to their
ability to produce the toxic effect of concern. In a cumulative risk
assessment the toxic effect of concern is the effect which is common to all
members of the group. The common endpoint for the OPs has been
determined to be the inhibition of cholinesterase activity. The ability to
produce this effect is quantified by a "potency" value. The method to
estimate the relative potency of the OPs in producing the toxic effect of
concern has been termed the "relative potency factor" method. This
method includes the following elements:
~ Determine the potency of each chemical.
~ Select an index chemical.
~ Express each chemical's potency in terms of the index chemical.
~ Select the endpoints for the index chemical
The result of the method is the determination of a relative potency
factor or RPF for each chemical. The table below shows the RPFs that
have been developed. Only those chemicals with residential/non-
occupational exposures have RPFs for the dermal and inhalation routes of
exposure. The sections that follow describe how EPA is calculating RPFs
for the dermal, inhalation, and oral routes of exposure. An example
calculation is provided for each route.
Relative Potency Factors
Chemical
Oral
Dermal
Inhalation
Acephate
0.13
0.0025
0.208
Azinphos-methyl
0.092
Bensulide
0.003
0.0015
Chlorpyrifos
0.10
Chlorpyrifos-methyl
0.012
Diazinon
0.024
Dichlorvos
0.037
0.677
Dicrotophos
1.95
Revised Draft
17
-------
Chemical
Oral
Dermal
Inhalation
Dimethoate
0.33
Disulfoton
1.23
0.47
6.596
Ethoprop
0.049
Fenamiphos
0.039
1.5
0.315
Fenthion
0.35
0.015
Fosthiazate
0.16
Malathion
0.0003
0.015
0.003
Methidathion
0.37
Methamidophos (Index Chemical)
1.00
1.00
1.00
Methyl Parathion
0.058
Mevinphos
1.36
Naled
0.083
0.075
0.820
Oxydemeton Methyl (ODM)
0.90
Phorate
0.39
Phosalone
0.024
Phosmet
0.020
Pirimiphos methyl
0.029
Terbufos
0.84
Tetrachlorvinphos
0.0008
0.00075
Tribuphos
0.045
Trichlorfon
0.014
0.0075
0.087
Note: Three pesticides included in the preliminary assessment, phostebupirim, profenofos, and
chlorethoxyfos did not have quantified RPFs. These chemicals have no detectible residues in PDP
monitoring data. They were included in a screening assessment for drinking water using RPFs of 25.
This screening assessment demonstrated that their contribution to drinking water risk is very low.
Chemical specific RPFs are being developed for these three chemicals.
The relative potency factor for each chemical is expressed in relationship
to an index chemical. The relative potency of the index chemical is, by
definition, one. The index chemical's measure of potency is divided by each
chemical's measure of potency to produce its relative potency, as illustrated
below.
Revised Draft
18
-------
Index Chemical Measure of Potency
Index Chemical RPF = —-—— ;— = 1
Index Chemical Measure of Potency
Index Chemical Measure of Potency
Chemical A RPF = —— :— — = 0.5
Chemical A Measure of Potency
Index Chemical Measure of Potency
Chemical B RPF = — ——— — -= 2.0
Chemical B Measure of Potency
In this example chemical A is half as potent as the index chemical in
producing the effect of concern, while chemical B is twice as potent as the index
chemical in producing the effect.
Use of Relative Potency Factors to Express All Residues As Residues of the
Index Chemical
After calculating the relative potencies of all of the chemicals in the CAG,
for each exposure route that is being assessed (i.e., oral, dermal and inhalation),
the residues of each chemical are multiplied by that chemical's relative potency
factor for each exposure of interest (e.g., food residues). Where exposure to
these residues can co-occur to the same population member, the resulting
values are added together to get the total, cumulative exposure in terms of
residues of the index chemical, as illustrated below.
Residue Index Chemical x 1.0
Residue Chemical A x 0.5
+ Residue Chemical B x 2.0
Total Residues (expressed as
residues of the index chemical)
2. Background on Different Types of Endpoints Used in Risk
Assessment
The following provides some very basic background on different types of
endpoints that can be used for risk assessment (and as measures of relative
potency). The terms and definitions presented here will be used in the following
discussion.
Revised Draft
19
-------
In individual chemical risk assessment OPP most often uses a No
Observed Adverse Effect Level ( NOAEL) as the endpoint. This is defined as the
highest dose level that does not produce a significant increase in an adverse
response. Significance usually refers to both statistical and biological
considerations. Significance may depend on a number of factors including the
number of dose levels tested, the number of animals per dose, and the
background incidence of the adverse response in the control (non-exposed)
group.
The NOAEL approach has several limitations, some of which are
particularly important for cumulative risk assessment where one of the goals is to
determine relative potency. These limitations include:
~ the NOAEL by definition must be one of the experimental doses
tested
~ once the NOAEL is identified, information in the rest of the dose-
response curve is ignored
~ experiments that use fewer animals can result in NOAELs at higher
dose levels thus rewarding testing procedures that produce less
certain rather than more certain NOAELs
~ the NOAEL approach does not identify actual (i.e., significant)
responses and the NOAELs will vary based solely on the dose
levels tested-resulting in NOAELs that may represent widely
varying levels of risk; therefore, the NOAELs do not represent a
common level or "common footing" on which to compare different
chemical's potency
An alternative approach that addresses some of these limitations is
the benchmark dose (BMD) method. Using this method all of the data
points being considered are plotted to produce a dose response curve.
Using various statistical techniques, this curve is then used to calculate a
specified response level (the benchmark response). The benchmark
response is usually specified as a 1 to 10% response (compared to the
control). For example, the BMD10 is the benchmark dose associated with
a 10% response compared to the untreated control. The Agency
benchmark dose guidance recommends use of the lower bound
confidence limit for a dose at the specified response level as an alternative
endpoint to the NOAEL. Because of very narrow confidence limits, in the
preliminary cumulative assessment the estimated benchmark dose itself is
used, rather than the lower confidence interval. The following graphs
illustrate these concepts.
Revised Draft
20
-------
NOAEL/LOAEL
CD A
tr
"Observable
Range of Data'
A
NOAEL
x LOAEL
A A
0
Dose
Benchmark Dos
CD
(/)
c
O
QJ
fj)
CD
en
"Observable
Range of Data
"10% Level of
Change"
0
Point of Departure
~ Dose
Benchmark Dose (BMD10)"
Revised Draft
-------
It should be noted that it is actually possible for the lower bound
confidence limit (BMDL) on the BMD to be at a dose level below the
NOAEL. This is because the confidence limits on the BMD take into
account the quality of the data being used to estimate the response and
reflect the variability in response. If there is a large amount of uncertainty
and/or variability resulting in relatively large confidence intervals on the
BMD, then the lower bound confidence interval may be below the NOAEL.
In the preliminary cumulative assessment of the OPs, it was not
possible to use dose-response modeling to estimate relative potencies for
the dermal and inhalation routes of exposure. For these routes a
"Comparative Effect Level" (CEL) was used. A CEL is simply a defined
response for the common mechanism which can be used for comparison.
In the preliminary cumulative risk assessment the comparative effect level
is defined as the dose causing no greater than15% cholinesterase
inhibition compared to the control. CELs are dose levels from a study and
in the preliminary cumulative risk assessment for the OPs the CEL was in
many cases, in fact, the same dose identified as the NOAEL. The NOAEL
chosen for a study reflects a weight-of-evidence decision from different
types of toxic effects while the CEL is simply a defined response for the
common mechanism effect used for comparison purposes.
Another term used in the cumulative assessment that is not used in
the individual OP assessments is "Point of Departure". The Point of
Departure (POD) is defined as the point in the dose-response curve at
which a change in response can be reliably said to be due to dosing with
the chemical, but is still within background variability. In the individual
assessments it is equivalent to the "endpoint" used to calculate risk. In the
cumulative assessment the POD is the level of response used to
represent the toxicity of the index chemical, i.e., the "endpoint" for the
index chemical. It is used to calculate the cumulative risk.
As will be discussed in detail below, for the oral route of exposure
benchmark doses are used for the measures of relative potency and the
point of departure of the index chemical. For the dermal and inhalation
routes CELs are used for the measures of relative potency. The points of
departure for the dermal and inhalation routes of exposure are the
respective benchmark doses for the dermal and inhalation routes for the
index chemical.
3. Implementing the Relative Potency Factor Method
The method itself, as illustrated above, is straightforward; however
the details of its implementation in any given case are more complex. In
order to implement the method, four critical pieces are necessary.
Revised Draft
22
-------
~ Selection of a specific common endpoint and duration of
exposure on which to compare potencies (e.g., for the OP's
common mechanism the endpoint is cholinesterase
inhibition-cholinesterase data are available for plasma,
brain, or red blood cell (RBC) cholinesterase inhibition in
male or female rats, rabbits, dogs, or mice and studies using
many different time frames are available);
~ Estimation of the measures of potency (e.g., BMD10's, CELs)
and calculation of the relative potencies;
~ Selection of an index chemical; and
~ Selection of the specific level of response (e.g., BMD10
NOAEL) to represent the toxicity of the index chemical. This
is the Point of Departure (POD).
The index chemical is selected based on which chemical in the
CAG has the best data base for all routes of exposure (oral, dermal,
inhalation) and has the best-characterized dose-response curve for the
toxic effect. This allows a more reliable analysis of all the potential data
available on the relative potencies of the other chemicals. The selection
of the index chemical does not affect the individual chemical potency
values used to calculate the relative potencies. The importance of the
index chemical selection lies in the determination of the endpoint used in
risk estimation (the Point of Departure mentioned above). It is desirable to
have high confidence in the selected endpoints. Therefore, again, it is
desirable that the index chemical have the best and most complete toxicity
data base for the common endpoint.
In the OP preliminary assessment the selection of the index
chemical has no effect on the estimated risks for the oral route of
exposure, i.e., the estimated risks from the oral route of exposure would
be the same regardless of which chemical was the index chemical. This is
because the measures of potency and the Point of Departure use the
same measure, the BMD10. [This was not the case in the previous
analysis where the measures of potency were the slope scaling factors
(m) while the Point of Departure was the BMD10.] For the dermal and
inhalation routes, where CELs are the measures of potency while the
Point of Departure is the BMD10 of the index chemical, the selection of the
index chemical may affect the estimated risks.
Selection of a specific common endpoint, duration of exposure, and
the method to compare potencies is based on a detailed analysis of the
toxicity database. In presentations to the Scientific Advisory Panel (SAP),
the Agency has discussed several approaches that could be used. The
Revised Draft
23
-------
Agency first proposed a method to the SAP in a pilot analysis in
September 2000. In this analysis a single "representative" study was
chosen for each chemical and route of exposure (oral, dermal, inhalation).
Dose response was modeled with a linear equation using a probit model.
In response to this analysis the SAP provided the following
recommendations:
~ There would be much greater confidence in the measure of
relative potency if it were derived from several, relatively
consistent studies as opposed to a single study, without
benefit of confirmation by other studies.
~ Reevaluate the selection of the probit model for determining
the relative potencies. They specifically suggested
considering Michaelis-Menton kinetics or an exponential
model as the potential alternative methods.
The SAP comments were addressed in a second analysis
completed in July 2001. In response to this second analysis, the panel
recommended several refinements. A detailed discussion of those
recommendations and how the Agency has addressed them in the
preliminary risk assessment is provided below in Section 4-"Oral Relative
Potency Factors". Another SAP meeting to review this work is scheduled
for February 2002. The following table summarizes the key aspects of
each analysis.
Summary of Three Hazard Analyses
Oral Route of Exposure
Pilot
(September 2000)
July 2001
Current
(December 2001)
Type of Model
Linear Equation (probit
model)
Nonlinear Equation
(exponential model)
Nonlinear Equation
(exponential model
expanded to include
possibility of flat area in
low dose region)
Studies Used
One Representative
Study
All Studies
All Studies
Study Duration
> 21 Days
> 21 Days
> 21 Days
Proposed
Compartment/Sex for
RPFs
None Proposed
RBC/Male
Brain/Female
Potency Measure
bmd50
Slope Scaling Factor
(m)
BMD10
Revised Draft
24
-------
Data Used in Model
CHEI Percent Inhibition
Mean CHEI, Standard
Deviation, Sample Size
Mean CHEI, Standard
Deviation, Sample Size
Analysis of Variability
No Measure
Weighting of Variance;
Confidence Intervals
Calculated
Weighting of Variance;
Confidence Intervals
Calculated
Analysis of Model
Uncertainty
No Measure
Goodness-of-Fit Test
Statistical comparison
of fit between basic
and expanded model
Determination of the Relative Potency Factors for the Dermal and
Inhalation Routes of Exposure
The Agency used two different methods to estimate potency, one
for the oral, and another for the dermal and inhalation routes of exposure.
This was necessary because there are different amounts of data available
for the different routes. Determination of the relative potency for the
dermal and inhalation routes of exposure will be discussed first, because
these are the simplest cases. This will be followed by discussion of the
oral route; the selection of the index chemical, which is the same chemical
for all routes of exposure; and the Points of Departure for the index
chemical.
The dermal and inhalation routes of exposure are only applicable to
residential exposures. Therefore, RPFs were only determined for those
chemicals which have residential uses.
4. Dermal Relative Potency Factors
Relative potencies for the dermal route of exposure were
determined using Comparative Effect Levels (CELs) observed in dermal
toxicity studies. The CEL is the dose causing up to15% cholinesterase
inhibition. This is in contrast to the relative potency factors for the oral
route, which, as will be discussed shortly, were determined through
modeling. Even though the dermal data were not suitable for modeling,
the dermal studies were used in the relative potency analysis and
endpoint evaluation. They were chosen (as opposed to, for example,
using oral data as a surrogate) because of the importance of using the
same route of exposure, in this case dermal, for both the toxicity and
exposure estimates. There are only a limited number of dermal studies for
OPs with high quality dose-response data. Therefore, it was determined
that the database of dermal toxicology studies, when considered across all
of the chemicals, was not appropriate for dose-response modeling.
As noted above, determinations of relative potencies based on
Revised Draft
25
-------
tests using the same sex and species are preferred. As will be explained
in detail in the section below on "Oral Relative Potency Factors," the
Agency is using the data on inhibition of brain cholinesterase activity in
female rats as the measure of relative potency for the preliminary OP
cumulative risk assessment. Therefore, CELs for female brain
cholinesterase inhibition from dermal toxicity studies were used to
determine the dermal relative potency measures.
In the case of dermal exposure, tests on the same species were not
always available. Four chemicals were tested by the dermal route in rats.
Only rabbit studies were available for four OPs. Thus, both rat and rabbit
data were used.
One chemical, dichlorvos, had no dermal exposure study of any
kind. OPP waived the requirement for a dermal toxicity study due to the
volatility of the chemical, which makes it very difficult to conduct such a
study. Residential/non-occupational dermal exposure was not assessed
for dichlorvos in the preliminary cumulative risk assessment of the OPs.
This is because there is a limited potential for significant exposure via the
dermal route. DDVP's high volatility limits its residence time on skin
surfaces thus making the dermal (and subsequent oral) routes of
exposure unlikely.
Based on the above considerations, the following CELs were
chosen as the measures of potency for the dermal route of exposure.
Measures of Potency for the Dermal Route of Exposure:
CELs for Female Brain Cholinesterase
Activity from Dermal Toxicity Studies
Chemical
Species
CEL(mg/kg/day)
Acephate
rat
300*
Bensulide
rat
500*
Dichlorvos
Dermal exposure study waived due to volatility of compound.
Disulfoton
rabbit
1.6
Fenamiphos
rabbit
0.5
Fenthion
rabbit
50
Malathion
rabbit
50
Methamidophos
rat
0.75
Naled
rat
10
Revised Draft
26
-------
Tetrachlorvinphos
rat
1000*
Trichlorfon
rabbit
100
* Highest dose tested.
The following examples illustrate how these CELs are used to
calculate the relative potency factors. Using the measure of potency for
the index chemical, 0.75 mg/kg/day, as explained above, the relative
potency factors are calculated as:
Index Chemical Measure of Potency 0.75
Index Chemical RPF = ——— ;——— — = —— = 1
Index Chemical Measure oi Potency 0.75
Index Chemi cal Measure of Potency 0.75
Acephate RPF = — — = —— = 0.0025
Acephate Measure of Potency 300
Index Chemical Measure of Potency 0.75
Bensulide RPF = — = = 0.0015
Bensulide Measure of Potency 500
The remaining relative potencies can be calculated in a similar manner.
All of the RPFs for the dermal route of exposure are listed in the table,
"Relative Potency Factors," at the beginning of this section.
5. Inhalation Relative Potency Factors
Relative potencies for the inhalation route of exposure were
determined using Comparative Effect Levels (CELs) from inhalation
toxicity studies. The CELs are defined as the dose causing up to a 15%
brain cholinesterase inhibition (compared to the control). This is in
contrast to the relative potency factors for the oral route which, as will be
discussed shortly, were determined through dose-response modeling. As
described in the case of dermal exposure, the inhalation studies were
chosen because of the importance of using the same route of exposure, in
this case inhalation, for both the toxicity evaluation and the exposure
estimate. As in the case of the dermal toxicity database, the number of
available inhalation toxicity studies with quality dose-response data was
limited. Therefore, it was determined that the database of inhalation
toxicology studies, when considered across all of the chemicals, was not
appropriate for dose-response modeling.
As noted above, determination of relative potencies based on tests
using the same sex and species is preferred. As will be explained in detail
in the section below on "Oral Relative Potency Factors", the Agency is
Revised Draft
27
-------
using the data on inhibition of brain cholinesterase activity in female rats
as the measure of relative potency for cumulative risk assessment.
Therefore, CELs for female brain cholinesterase inhibition from inhalation
toxicity studies were used to determine the inhalation relative potency
measures.
All of the inhalation studies were performed with the same species
(rat); however four different strains of rats were used. The exposure
conditions varied among the chemicals tested. Four used whole-body
exposure while three used nose only exposures. The index chemical,
methamidophos, used head/nose exposure. The studies were sub-
chronic (21 to 90 days), with the exception of dichlorvos, which had only a
chronic inhalation study.
No inhalation toxicity studies were available for three chemicals,
tetrachlorvinphos, fenthion, and bensulide. No inhalation risk assessment
was necessary for two of these chemicals, tetrachlorvinphos and fenthion,
in the OP cumulative assessment. A quantitative risk assessment for
tetrachlorvinphos was not included in the OP cumulative assessment due
to lack of exposure data suitable for use in a probabilistic assessment.
Tetrachlorvinphos's only remaining uses are pet uses. The individual
chemical screening level assessment indicates risks of concern.
Inhalation risks were not estimated for public health mosquitocide uses.
This is the only remaining use of fenthion. Bensulide's inhalation RPF was
estimated using the oral data for bensulide.
Based on the above considerations, the following CELs were chosen
as the measures of potency for the inhalation route of exposure.
Measures of Potency for the Inhalation Route of Exposure:
CELs for Female Brain Cholinesterase Activity
from Inhalation Toxicity Studies
Method (species
Female
Chemical
tested was the
CEL
rat in all cases)
(mg/kg/day)
Acephate
nose only
1.492*
Bensulide
No inhalation toxicity study available.
Dichlorvos
whole body
0.458
Disulfoton
nose only
0.047
Fenthion
No inhalation toxicity study available.
Fenamiphos
nose only
0.984*
Revised Draft
28
-------
Malathion
whole body
121
Methamidophos
head/nose
0.310
Naled
whole body
0.378
Tetrachlorvinphos
No inhalation toxicity study available.
Trichlorfon
whole body
3.574
* Highest dose tested.
These CELs are used to calculate the relative potency factors in
exactly the same way as in the case of the dermal RPFs. Using the
measure of relative potency for the index chemical, 0.310 mg/kg/day, as
explained above, the relative potency factors are calculated as:
Index Chemi cal Measure of Potency 0.310
Index Chemical RPF = — ;— = = 1
Index Chemi cal Measure of Potency 0.310
Index Chemical Measure of Potency 0.310
Acephate RPF = = = 0.208
Acephate Measure of Potency 1.492
Dichlorvos RPF = Index Chemical Measure of Potency = 0.310 = 0.677
Dichlorvos Measure of Potency 0.458
The remaining relative potencies can be calculated in a similar manner.
All of the RPFs for the inhalation route of exposure are listed in the table,
"Relative Potency Factors," at the beginning of this section.
6. Oral Relative Potency Factors
a. Model Used to Estimate RPFs for the Oral Route of
Exposure
In the case of the oral route of exposure, numerous oral studies
with comparable methodologies are available and suitable for dose-
response analysis. Therefore, it was possible to determine relative
potency factors for the oral route of exposure using a model developed in
response to SAP comments. In response to the pilot analysis presented
in September 2000, the Agency developed the following exponential
equation to model the dose-response curves and estimate oral relative
potencies.
Revised Draft
29
-------
y = B + (A - B) x e"m*dose
where:
y=cholinesterase activity
B=the y-asymptote (value of maximum cholinesterase
inhibition)
A=background cholinesterase activity
m=slope scale factor (the measure of potency in the
July assessment)
Dose=dose of the OP, in mg/kg/day
While the equation itself may appear rather daunting, the idea is
fairly simple. All of the relevant data points are assembled and the
equation employs a mathematical exercise that attempts to find a curve
that comes the closest to the most data points (simultaneously) as
possible. Statistical methods are then available to assess if this curve,
and the measures derived from it (e.g. m, the BMD10) are really good
representations of these data points. A graph of this exponential function
is provided below.
100
0.25
Dose (mg/kg/day)
Revised Draft
30
-------
This model did provide a good representation of the data. Out of a total of 1312
data sets available for modeling, the above exponential function was a good
representation of the dose-response for 1306 data sets. A data set in this case
consisted of the cholinesterase measurements at a specific time point from a specific
study for a specific compartment (plasma/RBC/brain) and sex combination (e.g.,
male/plasma).
The dose-response analysis was performed using a computer
program developed for this purpose by the Agency's Office of Research
and Development's National Health and Environmental Effects Laboratory
(NHEERL). This program, OPCumulativeRisk (OPCumRisk), is publicly
available on the internet at
www.epa.qov/scipolv/sap/index.htm#september.
The SAP was very supportive of this approach when it was
presented in September 2001. However, some additional analyses and
revisions were recommended. The key recommendations that have been
incorporated into the current analysis include:
1) reevaluation of the procedure for estimating the horizontal-asymptote,
i.e., the "B" term in the above equation (the Panel suggested that the
decision rules used to estimate "B" could be improved to result in more
consistent values for the horizontal asymptote)
2) determination of the appropriate measure for relative potency (some
members of the Panel suggested that a Benchmark Dose, e.g., BMD10,
was a more appropriate measure of potency than m, the slope scaling
factor in the above equation)
3) a formal analysis of residuals (the residuals are the measures of how
far each of the actual measured data points is from the estimated dose-
response curve-the Panel suggested that the results of this analysis
would help the Agency remove some bias in the potency estimates-see
#4 below)
4) revision of the statistical procedures for weighting the various
cholinesterase data points and calculating confidence intervals (the Panel
was concerned that the residual plots presented at the September 2001
meeting appeared to indicate the residuals were larger in the low dose
area of the curve than in the high dose area, the Panel suggested revising
the weighting procedure would improve this bias-the weights determine
the relative importance of different data points in the analysis; the Panel
also suggested a particular statistical technique called "bootstrapping" for
calculating the confidence intervals)
Revised Draft
31
-------
5) consideration of repeated measures of cholinesterase inhibition in the
animal studies (in toxicity studies, measures of blood cholinesterase such
as RBC and plasma can be obtained from a single laboratory animal
multiple times, therefore, blood cholinesterase is often a "repeated
measure"-in contrast, brain cholinesterase can only be measured from a
laboratory animal one time and cannot be a "repeated measure"; statistical
procedures need to incorporate information about the variability within a
study-in the case of repeated measures, where animals are potentially
measured more than once, it is necessary to track each individual animal
in the statistical procedures-the Panel recommended the Agency
consider how repeated measures impacted potency estimates)
Finally, there was considerable discussion at the technical briefing (August
2001) and the SAP meeting in September about the potential for a flat
region in the low dose portion of the dose-response curve. (The concern
was that the model only allowed for exponential decline. Some argued
that initially—at low doses-the curve would not decline exponentially for
some chemicals but would have a flat area where cholinesterase activity
was not declining as quickly.)
The current analysis addresses these comments by incorporating
the following changes. The analysis uses all of the data being considered
all at once, together in a joint analysis rather than a tiered approach
working up from single measures in individual studies. This joint analysis
enabled the Agency to address two of the above recommendations.
Recommendation (1) is addressed because there is only a single estimate
of the horizontal asymptote for each sex and chemical, rather than
multiple ones for each study. The joint analysis, because it considers all
of the data at once, is based on more dose levels. As a result it was
possible to look in more detail at the shape of the low-dose area of the
dose-response curve.
A new equation was developed to include the possibility of a flatter
area in the lose-dose region of the dose-response curve. This new
equation is shown below.
idose=0.5[(Dose - S - D) + /(Dose - S - D)2 + 4 X Dose X S]
where:
idose is the scaled internal dose
Dose is the administered dose level
S controls the shape of the curve in the low-dose region
D controls the horizontal width of the low dose part of the
curve that is declining less rapidly than the rest of the curve
Revised Draft
32
-------
This equation can be added to the first equation, which we will now
call the "basic" equation, by replacing Dose in the first equation with idose.
When this is done the result is what we will refer to as the expanded
equation. The expanded equation becomes equivalent to the basic
equation as S gets larger and D approaches 0. A graph illustrating these
relationships is shown on the following page.
Revised Draft
33
-------
- 10
- 8
- 6
- 4
- 2
- 0
CD
CO
O
Q
ro
c
CD
-t—•
C
"O
ffl
CO
o
CO
0
8
10
Dose
The graph shows the effect of increasing values of S. On the left
hand side of the graph are various exponential functions (the solid lines)
with different values of S. Looking at these graphs from right to left, it can
be seen that as S gets larger, the flat portion in the low dose area
disappears, until when S is very large the curve is the same as the basic
exponential equation. On the right hand side of the graph, the dashed
lines represent the relationship between estimated internal dose and
administered dose that is expressed in the expanded equation, with D=2
and increasing values of S going from right to left. This part of the graph
demonstrates that as S gets larger Administered Dose and Internal Dose
become equivalent.
The expanded equation estimated a dose-response curve that fit
the data better than the basic curve for eight of the chemicals. These
chemicals, whose dose-response curves were modeled using the
expanded equation, are azinphos-methyl, bensulide, disulfoton, malathion,
Revised Draft
34
-------
methyl-parathion, phorate, phosmet, and terbufos. The remaining
chemicals dose-response curves were estimated using the basic equation.
As noted above (recommendation 2) there was considerable
discussion at the SAP about whether m, the slope scaling factor in the
basic equation, or a Benchmark Dose (e.g., BMD10) was the appropriate
measure of potency. In the current analysis the BMD10 has been used as
the measure of potency. This was necessary because the shapes of the
dose-response curves for the basic and expanded models differ in the low
dose region. Therefore, the slope-scaling factor (m) is no longer an
appropriate measure of potency across all of the chemicals. In addition,
the value of the slope-scaling factor is dependent on the value of the
horizontal asymptote. The current analysis clearly shows that the values
of the horizontal asymptotes for the different OP chemicals are not similar
to each other. Thus, the slope scaling factor is not an appropriate
measure for determining relative potency.
Another statistical concern noted above- (recommendation 4)
revision of the statistical procedures for weighting the cholinesterase data
and calculating confidence intervals was partially addressed by use of a
new estimation procedure in the joint analysis. This procedure, a
nonlinear mixed effects model, was performed using the nlme package of
R (an open source statistical programming language; http://cran.r-
proiect.org). The R programs used in the current analysis are contained in
Appendix B of the preliminary risk assessment.
In the July and in the current analysis data points were weighted to
give those data points that were more reliable more influence on the
estimated dose-response curve. In the July analysis the weights were
based on the square of the estimated mean. In the current analysis the
weights are proportional to the mean. The confidence intervals have not
been recalculated using the method suggested by the SAP
(bootstrapping), however; the calculation of the confidence intervals has
been revised. Bootstrapping is a very time and resource-intensive
procedure. Although it may be the preferred approach for calculating
confidence intervals, the Agency has not conducted any bootstrapping
procedures. The current method for calculating confidence intervals is
adequate and satisfactory for this assessment. To address another
statistical concern noted above-(recommendation 3), an analysis of
residuals was done which indicates that the models generally capture the
trend of the mean of the data, and the weighting function used in the
current analysis is generally superior to that used in the July analysis.
Revised Draft
35
-------
b. Selection of Species/Compartment/Sex and Duration of Exposure
for Comparison of Potencies
A central principle of the relative potency factor method is that
relative potency should be determined using a uniform basis of
comparison. This requires using to the extent possible a common
response derived from a comparable measurement methodology, species,
and sex for all the exposure routes of interest. Although many different
methods are available for measuring cholinesterase activity, for this
assessment they are all assumed to be comparable if the study was found
to be acceptable. Studies are available for various species (e.g., dog,
mouse, rat, and rabbit), however; toxicology studies in the rat provided, by
far, the most extensive cholinesterase activity data for all routes (dermal,
inhalation and oral) and in the three compartments (plasma, red blood cell,
and brain) in both sexes. Therefore, only rat studies were used in
determining relative potencies, except in the case of five chemicals for the
dermal route, for which no rat study was available.
The Agency decided to use only those data that reflect steady-
state conditions for cholinesterase inhibition to estimate relative potencies.
Steady-state as used here is the point where continued dosing at the
same level results in no further increase in cholinesterase inhibition. This
was done because the steady state values produce relative potency
factors that are reproducible and reflect less uncertainty due to the rapidly
changing, time-sensitive differences in measures of cholinesterase that
are observed prior to achieving a steady-state. Steady-state for each OP
was determined qualitatively. The analysis showed that most chemicals
appeared to reach steady-state by 21 to 28 days of exposure in both
sexes and all three compartments (plasma, RBC, and brain). The
available data sets for each chemical-sex-compartment included a range
of exposure durations from 21 days to greater than 700 days.
In addition, monitoring data show that people generally have had
some level of OP exposure, making it unlikely that any individual would
encounter exposure to OP pesticides without having a previous exposure
from other sources. Therefore, the Agency does not consider the use of
toxic endpoints based on single-day exposures to be reflective of the
actual human exposure situation. Furthermore, the effects of OP
exposure can persist for several days to weeks depending on the
magnitude of exposure, making the exposed individual potentially more
vulnerable to subsequent exposures during that period. These
considerations together with the very stable and reproducible levels of
cholinesterase inhibition in studies of 21 days or longer resulted in the use
of only those cholinesterase measures based on study duration of 21 days
or longer in the development of the RPFs.
Revised Draft
36
-------
In its September 2000 review the SAP recommended against
combining data (at least initially) across compartments, i.e., plasma, red
blood cell (RBC), and brain, or for males and females. This led the
Agency to analyze six separate compartment/sex combinations for each
chemical, i.e., male/plasma; male/RBC; male/brain; female/plasma;
female/RBC; and female/brain. These were analyzed in order to
determine an appropriate compartment/sex on which to compare
potencies of the chemicals. Overall there is a good agreement between
potency values calculated for males and females. Therefore, the selection
of either males or females would make little difference in the RPFs. Males
were chosen for use in the comparison of potencies for the July analysis.
For the current analysis which is based on brain cholinesterase inhibition,
females were chosen instead of males. This was because, for the brain
compartment, female rats were more sensitive than male rats for five OPs
(diazinon, dichlorvos, pirimiphos-methyl, tetrachlorvinphos, and
trichlorfon).
For most of the chemicals, the relative potencies were similar when
calculated using plasma, RBC, and brain measurements. In the July
analysis RBC cholinesterase inhibition was chosen for comparison of
potencies. After considering the comments from the September 2001
SAP meeting in addition to the comments from the public, the Agency has
decided to use brain cholinesterase data for quantifying cumulative risk for
the OPs. This decision was based on:
~ Compared to relative potency estimates using RBC, the estimates
of relative potency based on brain cholinesterase inhibition have
smaller confidence intervals and therefore, will result in less
uncertainty in cumulative risk estimates. (Confidence intervals give
the range of values within which the BMD10 is expected to actually
fall. Thus, if the BMD10 is estimated to be 0.08 mg/kg/day and the
confidence limits are from 0.0001 to 10.0 mg/kg/day, they are said
to be wide, and this would not be a very good estimate. If the
confidence limits are from 0.05 to 0.1, they are said to be narrow
and the estimate is better.)
~ Brain data represent a direct measure of the common mechanism
of toxicity as opposed to RBC or plasma which are surrogate
measures for the brain and peripheral nervous system. As noted
above, the toxic potencies and points of departure estimated using
brain cholinesterase inhibition are generally similar to the estimates
using RBC data for the OPs.
~ The SAP recommended that the Agency address the issue of
repeated measures. This issue, which concerns repeated
cholinesterase measures derived from a single animal, only
pertains to the plasma and RBC data because blood data can be
Revised Draft
37
-------
collected several times from a single animal, while brain data can
only be collected once. Therefore, using the brain data, repeated
measures are not an issue.
Although the Agency has emphasized that the relative potencies
and points of departure were similar when calculated using plasma, RBC,
and brain measurements, extensive analysis was done to characterize
the likely differences in the risk estimates resulting from using female brain
data compared to male RBC data. Extensive analysis was also done to
evaluate the effects of the changes in the statistical methods. These
analyses showed:
~ The revised statistical methods, while providing refined statistical
estimates, had very little impact on the BMD10 values that were
estimated, with the exception of those chemicals modeled with the
expanded equation.
~ 21 OPs have very similar oral RPFs based on female brain
compared to male RBC data-the difference was less than 3X.
These slight differences are likely due to experimental
variability/errors rather than real differences in sensitivity between
the RBC and brain measures.
~ Oral RPFs are lower (i.e., less potent) using brain data for diazinon,
malathion, fenamiphos, and tribufos. Malathion is the least potent
of all the OPs and this difference is unlikely to impact the
cumulative risk estimates. Diazinon's residential uses are being
phased out as well as many of its agricultural uses. Tribufos does
not have residential uses and is only used as a defoliant on cotton.
The only residential exposure to fenamiphos is on golf courses. It
has few detections in PDP. Because of limited exposure potential,
using either RPFs based on RBC or brain for diazinon, tribufos, and
fenamiphos would have little impact on the total cumulative risk
estimates.
~ Oral RPFs are higher (i.e, more potent) using brain data for
mevinphos, methidathion, acephate, and naled. Dietary exposure
to mevinphos is very low because it is only used on imported
bananas. Methidathion does not have many detects in PDP.
Because of limited exposure potential, using either RPFs based on
RBC or brain for mevinphos and methidathion would have little
impact on the total cumulative risk estimates. Dietary risks for
acephate and naled could be underestimated using male RBC
relative potency factors because both pesticides are used on
numerous commodities.
Revised Draft
38
-------
In summary, the oral relative potency values are based on
cholinesterase activity data derived from female rat brain data, taken from
studies that lasted 21 days or longer. This choice was made after an
extensive analysis of all available oral data and multiple reviews by the
SAP. As we have seen, such extensive databases are not available for
the dermal and inhalation routes of exposure. Therefore, because of the
extensive oral database, which makes a detailed comparison between
compartments for males and females possible, this same selection of
female rat brain data was also used in the case of the dermal and
inhalation routes. The only exception is that, when rat data were not
available for the dermal route of exposure, rabbit data were used. In
addition, as we will see shortly, the same selection was made for
determining the points of departure for risk assessment.
After determining that female brain measures in the rat are the
most appropriate for comparison of relative potencies and after selecting
the BMD10 as the appropriate measure of potency, RPFs can be
calculated. This is done using the BMD10 from the relevant (basic or
expanded) exponential equation. These BMD10's for each chemical are
listed in the table below.
ORAL BMD10s
(chemicals marked with *were modeled using t
he expanded equation)
Chemical
BMD10 (mg/kg/day)
Acephate
0.63
Azinphos methyl*
0.90
Bensulide*
32.85
Chlorpyrifos
0.83
Chlorpyrifos-methyl
7.51
Diazinon
3.43
Dichlorvos
2.25
Dicrotophos
0.04
Dimethoate
0.25
Disulfoton*
0.07
Ethoprop
1.70
Fenamiphos
2.11
Fenthion
0.24
Fosthiazate
0.50
Revised Draft
39
-------
Chemical
BMD10 (mg/kg/day)
Malathion*
326.37
Methidathion
0.22
Methamidophos (Index Chemical)
0.08
Methyl Parathion*
1.41
Mevinphos
0.06
Naled
1.00
Oxydemeton Methyl (ODM)
0.09
Phorate*
0.21
Phosalone
3.38
Phosmet*
4.13
Pirimiphos methyl
2.88
Terbufos*
0.10
Tetrachlorvinphos
101.92
Tribuphos
1.81
Trichlorfon
6.03
The following illustrates how these BMD10's are used to calculate
the relative potency factors. Using the measure of potency for the index
chemical, 0.08 mg/kg/day, the relative potency factors are calculated as:
Index Chemical RPF = Index Chemical Measure of Potency = 0.08 = 1
Index Chemical Measure of Potency 0.08
Acephate RPF = Index Chemical Measure of Potency = 0.08 =0.13
Acephate Measure of Potency 0.63
Bensulide RPF = Index Chemical Measure of Potency = 0.08 = 0.003
Bensulide Measure of Potency 32.85
Where the Measures of Potency for all of the
chemicals are the BMD10's estimated using the
relevant exponential equation calculated using that
chemical's female brain data, from studies 21 days or
longer.
The oral relative potencies for the remaining chemicals can be calculated
Revised Draft
40
-------
in a similar manner. All of the RPFs, for the oral route of exposure are
listed in the table, "Relative Potency Factors," at the beginning of this
section.
7. Selection of the Index Chemical and the Points of
Departure for Risk Assessment
The index chemical is selected based on which chemical in the
cumulative assessment group has the best data base for all routes of
exposure (oral, dermal, inhalation) and the best-characterized dose-
response curve for the toxic effect. It is important that it acts
toxicologically as purely as possible by the common mechanism defining
the group, that is, it has no other mechanisms of appreciable toxicity; and
that quantitative data for assessing potency be available for as many
routes of exposure, genders, species, and strains of animals as possible.
This allows a more reliable analysis of all the potential data available on
the relative potencies of the other chemicals.
Methamidophos was chosen to be the index chemical for the
preliminary OP cumulative assessment. The oral database contains
studies that characterize the entire dose-response range from very low
doses to high doses. Within the oral route of exposure, potency values
for methamidophos were consistent between adult male and female rats
and among the three compartments (plasma, RBC, and brain). Quality
dose-response data were also available for the dermal and inhalation
routes of exposure. Available data from the literature support the
conclusion that methamidophos acts "toxicologically as purely as
possible." It is a direct-acting anti-cholinesterase OP that appears to
selectively inhibit cholinesterase, the target enzyme.
The selection of the index chemical does not affect the potency
values used to calculate the relative potencies for the individual chemicals,
since these are based solely on the individual chemical's data, nor does it
affect the relative potencies of the chemicals, which is simply an indexing
exercise. The importance of the index chemical selection lies in the
determination of the dose level that will be used in risk estimation. This
dose level is called the Point of Departure or POD. It can be an observed
NOAEL from a single study, as was the case in the individual OP risk
assessments or it can be a Benchmark Dose based on a modeled
estimate.
In the OP preliminary assessment the selection of the index
chemical has no effect on the estimated risks for the oral route of
exposure, i.e., the estimated risks from the oral route of exposure would
be the same regardless of which chemical was the index chemical. This is
because the measures of potency and the Point of Departure use the
same measure, the BMD10. [This was not the case in the previous
Revised Draft
41
-------
analysis where the measures of potency were the slope scaling factors
(m) while the Point of Departure was the BMD10.] For the dermal and
inhalation routes, where CELs are the measures of potency while the
Point of Departure is the BMD10 of the index chemical, the selection of the
index chemical may affect the estimated risks.
The oral, dermal, and inhalation PODs for the cumulative
assessment are based on benchmark dose modeling of the rat female
brain data for studies of 21 days or longer for methamidophos. The
benchmark dose where cholinesterase activity is reduced by 10%
compared to background activity (BMD10) is the effect level selected. OPP
has traditionally used 10% cholinesterase inhibition for plasma and RBC
as the decision-point for selecting an effect level when cholinesterase
inhibition is the effect of interest. These PODs are listed in the following
chart. They are the endpoints the Agency used in the preliminary OP
cumulative risk assessment. The lower bound confidence limit (BMDL) on
these PODs is also listed. The narrow confidence intervals demonstrate
the high quality data available for methamidophos.
Points of Departure (from the Index Chemical Methamidophos):
Female Rat Brain Cholinesterase Activity from
Toxicity Studies 21-Days or Longer
Route of Exposure
BMD10 (mg/kg/day)
BMDL (mg/kg/day)
Oral
0.08
0.07
Dermal
2.12
1.77
Inhalation
0.39
0.21
8. Summary and Example Risk Calculation
Three elements are required for endpoint selection in the case of
cumulative assessments:
~ Selection of an index chemical,
~ Calculation of relative potency factors, and
~ Selection of points of departure.
These elements perform exactly the same function as the elements
in an individual chemical assessment. The following summary relates the
elements used in the cumulative assessment back to the basic risk
assessment equation that is used in all risk assessments:
Risk = Hazard x Exposure
Revised Draft
42
-------
The exposure part of the equation is obtained by summing all of the
relevant residues to which a person is exposed, for the relevant time
period. In individual chemical assessments these residues can simply be
added together, because they are all residues of the same chemical. In a
cumulative assessment these residues must first be put on a common
scale before they can be added. This is done by multiplying each of the
potentially multiple residues of each chemical by a number that represents
that chemicals relative potency, as shown below.
Relative
Residues Expressed as
Residues
Potencv Factor
Residues of the Index Chemical
1 mg Chemical A
X
0.5
= 0.5 mg
2 mg Chemical A
X
0.5
= 1.0 mg
0 mg Chemical B
X
2.0
= 0.0 mg
3 mg Chemical B
X
2.0
= 6.0 mg
2 mg Chemical Index
X
1.0
= 2.0 mg
4 mg Chemical Index
X
1.0
= 4.0 mq
13.5 mg/day of the index
chemical
Once all of the residues have been converted by this process, the
"Exposure" side of the equation is exactly the way it is for an individual
chemical-it is as if all of the residues are residues of the index chemical.
Just as in the case of an individual chemical assessment the
"Hazard" part of the equation is obtained by selecting the endpoints that
will be used for risk assessment. Since all of the residues are now
expressed in terms of the index chemical, the endpoints for use in risk
assessment are selected for the index chemical and compared to the
residues, to obtain the estimate of risk.
For example, to perform a dermal risk assessment using a margin
of exposure (MOE) approach, the methamidophos point of departure for
dermal risk assessment, 2.12 mg/kg/day, and the above exposure
estimate, 13.5 mg/day of methamidophos (converted to mg/kg/day by
dividing by body weight = 13.5- 62 kg = 0.22 mg/kg/day), would be used
to calculate the following MOE.
MOE = Hazard or MOE = Point of Departure = 2.12 mq/kq/dav = 9.6
Exposure Exposure 0.22 mg/kg/day
The "new and complicated" part of the OP cumulative risk
assessment is determining (and keeping track of) what measures are
being used for relative potency, and what points of departure for the index
chemical are being used in risk assessment. The measures of potency
were selected to provide the best measures of relative potency. The
Revised Draft
43
-------
points of departure were selected to provide the best measures of the
index chemical's toxicity for use in risk assessment. All of the measures
of potency for each route (dermal, inhalation, and oral) are listed in the
tables above, as are the points of departure for methamidophos. The
following table provides a summary of what measures were used in each
case.
Route of Exposure
Measure of Potency
Point of Departure
Dermal
CELs (from a dermal study
for each chemical using
female rat brain data and a
study 21 days or longer)
BMD10 (modeled from
Methamidophos Dose-
Response Curve based
on female rat brain data
from one methamidophos
dermal study)
Inhalation
CELs (from an inhalation
study for each chemical
using female rat brain data
and a study 21 days or
longer)
BMD10 (modeled from
Methamidophos Dose-
Response Curve based
on female rat brain data
from one methamidophos
inhalation study)
Oral
BMD10 (modeled using all
acceptable oral studies for
each chemical using female
rat brain data from studies
21 days or longer)
BMD10 (modeled from
Methamidophos Dose-
Response Curve based
on female rat brain data
from three
methamidophos oral
studies 21 days or
longer)
Revised Draft
44
-------
V. Cumulative Exposure Model And Interpretation Of
Model Outputs
A. Background
Developing a modeling tool that permits the assessment of co-occurrence
is a necessary aspect of the development of cumulative methods. The model
must be able to integrate exposure through food, water, and residential/non-
occupational pathways in a manner that reflects both the probability of exposure
by any given pathway and the timing of exposures through different pathways.
This means the model should reflect the exposure of discrete
individuals/population members in which routes of exposure are linked.
The estimated exposures should reflect the individual's location and other
demographic characteristics of the individual such as age and weight; the time of
year; the individual's anticipated patterns of pesticide use (for residential
exposure); and the individual's history of exposure. For example, if an
individual's house was treated for termites today, that exposure could continue
for a period of time for that individual, but would not be randomly spread through
a population. Similarly, for drinking water, the source of an
individual's/population member's drinking water today is likely to be the same
source tomorrow, and the spatial and temporal linkage must be preserved. As a
result, the building blocks for the cumulative risk assessment are specifically
defined individuals/population members for whom the spatial, temporal, and
demographic aspects of their exposures are linked. The outputs included in the
preliminary OP risk assessment are:
~ Cumulative risk from OPs in food
~ Cumulative risk from OPs in drinking water
~ Cumulative risk from OPs in residential/non-occupational settings
~ Cumulative risk from OPs across multiple pathways (food, water, and
residential/ non-occupational)
~ All of the above assessments contain some elements that are dealt with
qualitatively
The following section describes the attributes of the software model,
Calendex™, in some detail. This is the model that was used in the preliminary
OP cumulative risk assessment. In addition, the attributes and current status of
other models that allow assessment of cumulative risks will be briefly reviewed.
Calendex™ is a proprietary software package licensed from Novigen Sciences,
Inc. The Calendex™ model and its dietary component, DEEM™, have been the
Revised Draft
45
-------
subject of review at two SAP meetings. [The following papers were presented at
those meetings: "Dietary Exposure Evaluation Model (DEEM™) and DEEM™
and Max LIP (Maximum Likelihood Imputation Procedure) Pesticide Residue
Decompositing Procedure and Software," dated February 29, 2000 and
"Calendex™; Calendar-Based Dietary & Non-Dietary Aggregate and Cumulative
Exposure Software System", dated September 27, 2000],
B. Calendex™
Calendex™ contains demographic and food consumption data for a
sample of individuals/population members that is representative of the U.S.
population. This is the CSFII (USDA's Continuing Survey of Food Intakes by
Individuals) for 1994-1996 together with the 1998 Supplemental Children's
Survey. The demographic variables (e.g., age, sex, weight) for each
individual/population member in the survey can be used as part of the basis for
selecting potential non-food exposures for the individuals as well as to link these
non-food exposures to the food exposure for these individuals. For each
scenario that is developed, routes (e.g., oral and dermal) can be linked if
exposures are dependent on each other. If the exposures are linked, then the
model assumes that the exposures occur at the same time. For example, the
inhalation and dermal exposures that may result from application of a lawn
pesticide should occur on the same day. Calendex™ uses the calendar day as
the unit of time for calculating exposure. If exposure estimates for more than one
day are required, these are built by adding together sequential daily exposures
for an individual and averaging them over the number of days to provide an
average daily exposure over the desired time frame. If single-day exposures are
considered, the output of the analysis is a distribution of daily exposures.
Calendex™ calculates daily food exposure using the DEEM™ dietary
exposure model. This is the same model OPP currently uses for individual
chemicals. In the cumulative analysis, however, time is an important
consideration and it is necessary to estimate food exposure for every single day
of the year so that the daily food exposures can be combined with daily drinking
water and residential exposures. It is assumed in this analysis, that the
consumption data in the CSFII is reflective of food choices across the year and
around the country. No attempt has been made to estimate seasonal or regional
differences in food exposures. Drinking water concentrations are, however,
related in space and time. The pesticide concentration in drinking water at a
particular site on any given day is correlated with the concentration on a
subsequent day. The model must preserve this time-series relationship. A
similar relationship exists for residential exposures in which concentrations
present on Day 1 are related to concentrations on Day 2.
Calendex™ uses the following steps to estimate food and water
exposures in the case of single day exposures. Starting with January 1st,
~ It calculates exposure from food for individual #1 using one of his two diets
in the CSFII and randomly assigns a residue value from PDP for each
Revised Draft
46
-------
food included; after multiplying each amount of food consumed by its
selected residue value the total exposure for food for this individual is
calculated by summing the exposure through each food
~ For water, it selects a random year from the multiple years of daily
concentrations that are available, and calculates the water exposure on
January 1. These daily concentrations were estimated with
PRZM/EXAMS-IR, which provides a distribution of daily concentrations
over a wide range of years. Calendex™ calculates the exposure from
water for individual #1 by multiplying the concentration in water by the
water consumption reported in the CSFII by individual #1
~ It then sums the total exposure for food and water for individual #1 on that
day.
~ This process is repeated multiple, e.g., ten times, for each consumption
record (for the relevant age group) in the CSFII to develop a distribution of
exposures for January 1.
~ This process is then repeated for each calendar day of the year. In this
way a distribution of single-day exposures is generated for the entire year.
Finally, the whole process is repeated for each region. The output is a
distribution of exposures for the population subgroups of concern for each region.
Calendex™ uses the following analogous steps to calculate
residential/non-occupational exposures. Starting with January 1st,
~ It uses the probability that the individual would be using a pesticide for a
particular purpose to determine if the individual might have a chance of
being exposed to various pesticides that might be used that day.
~ If the answer to that question is "yes" it determines the specific dates that
are possible for those exposures to make a probability decision on
whether the individual is actually exposed on that day, i.e., a probability-
based decision is made to determine whether or not that scenario will
actually be encountered by the individual on that date.
~ If a scenario is assigned a "yes" answer, then the appropriate values
defining the exposure are selected from the many distributions of input
parameters for residential exposure scenarios.
~ The exposures for the appropriate dermal, oral and inhalation exposures
are calculated for all selected residential scenarios. In doing these
calculations the model is able to use information on the frequency and
amount of chemical used and the degradation of the chemical over time.
The estimates of the amount of residues available to be contacted, how
easily they dislodge (i.e., come off) when contacted, and how often
contact is made are provided as inputs into the model. The model also
evaluates for each day whether an individual applied a pesticide on a
previous day-and, if so, estimates exposure as a result of that previous
application (appropriately considering any degradation that may have
subsequently occurred).
~ All the exposures are converted to route-specific MOEs, i.e., separate
MOEs for oral, dermal, and inhalation, and added together with the food
Revised Draft
47
-------
and water exposure for that individual for that day to estimate the total
exposure for that individual on January 1.
~ As noted above, this process is repeated multiple, e. g., ten times, for
each consumption record (for the relevant age group) in the CSFII to
develop a distribution of exposures for January 1.
~ This process is then repeated for each calendar day of the year. In this
way a distribution of single-day exposures is generated for the entire year.
Finally, the whole process is then repeated for each region. The output is
a distribution of exposures for the population subgroups of concern for each
region.
A more detailed description of how Calendex™ operates, including
numerous examples, can be found in the Appendices to this document.
The Agency has worked extensively with the components of Calendex™
and has developed the capability to track the exposure input data that
correspond to individual daily risk estimates. This allows analysis of specific
pesticide residue inputs, including the specific pesticide and commodity for food
and water exposure or specific use of the pesticide for residential exposure. As
such, Calendex™ will permit the Agency to identify and analyze sources of
exposure in order to identify further refinements or mitigation strategies.
C. Interpretation of Model Outputs
As discussed above, the model outputs are a series of daily exposure
distributions, one for each day of the year. Each daily distribution represents the
result of repeatedly estimating the possible exposures for each individual in the
relevant population for that day of the year. For any given percentile of exposure
of interest, e.g., the 90th, these daily distributions can be shown as a time series
of MOEs across the entire year. Taking January 1st as an example, this is done
by selecting from the January 1st daily distribution the exposure level that
corresponds to the 90th percentile of exposure and calculating the corresponding
MOE. This MOE is placed on January 1st of the yearly graph for the 90th
percentile. This is illustrated for January 1st in the following graphs.
Revised Draft
48
-------
. s* _ ¦ .| x. Graph of Daily
January Distribution MOEs at 90th Percentile for One Year
Exposure (/^g/kg/day)
In a similar manner, time series could be developed for any other
percentile of exposure by selecting the exposure for that percentile from each
daily distribution. These 365 MOEs would be placed on another graph for the
year that would represent the new percentile of exposure.
In the case of the food assessment, interpretation of the output distribution
is the same as for the individual chemical assessments. Because of the wide
distribution of both fresh and processed foods, food is assumed to be a
"national" commodity with little seasonal or regional variation. Any differences
resulting from time of the year or region of the country are reflected in the PDP
sampling of food residues and the sampling scheme used in the CSFII to reflect
consumption. Therefore, the results for food alone really represent single
Margins of Exposure (MOE) corresponding to a given percentile of exposure
which are considered representative of any day of the year in any region of the
country. Nevertheless, the food output distributions were calculated and arrayed
for each day across the year as described above, so that they can be
incorporated with the water and residential output for each region. However, as
can be seen in the calculated food output distributions, there is very little
variability in the results from day to day and the same distribution is used for
every region.
In the case of the water and residential assessments, there is both a
spatial and temporal component that can be seen very clearly in the output
distributions. The spatial component is reflected in the 12 separate regional
assessments that were done. The temporal component is reflected in the
significant variation that can be seen in the water and residential output
distributions over the course of a year. A typical graph showing the food, water,
and residential output distributions in a specific region is shown below. It should
Revised Draft
49
-------
Cumulative Assessment Children Ages 1 -2 in the Southern S eaboard (Region 6) Single D ay
Analysis
JiJia n Days
RJ Pi
c-j cn cm cn c-j c--i cn c-j co co co co co co
10000
S 1000000
Total MOE
Inhalation MOE
Derrns/ MOE
Oral (Non-ci etary) MOE
Food MOE
PRZM-EXAMS Weter Exposure (433+435) MOE
be noted that the residential output distributions are really three separate
distributions representing dermal, inhalation, and non-dietary oral MOEs.
The analysis and interpretation of the temporal component of the risk
assessment involves two related aspects: first the exposure duration used to
calculate the MOEs and second the length of the exposure period that is relevant
when viewing the above output distributions across the year. The nature of the
toxic response, i.e., brain cholinesterase inhibition is also relevant to determine
an appropriate time frame over which to consider exposure to OPs
Cholinesterase inhibition is not immediately reversible, with effects persisting for
days to weeks depending on the magnitude of exposure. Because of this and
because there is a continued background exposure to OPs from food, a period of
multiple days might be considered an appropriate window over which to evaluate
the pattern of exposures and resulting MOEs for the OP cumulative assessment.
In the preliminary OP cumulative assessment, single day exposures have
been used to calculate the MOEs. in this case, analysis and interpretation of the
output distributions relies heavily on examination of these distributions to discern
changing patterns of exposure. When viewed in this way, there are periods of
higher exposure (i.e., periods with low MOEs) and periods with lower exposure
(i.e., periods with higher MOEs). Changes in the pattern of exposure can be
interpreted by examining different pathways of exposure (food, water, residential)
and different routes of exposure (oral, dermal, and inhalation) separately to
determine the factors causing any increased exposure estimates. Given the
Revised Draft
50
-------
hazard framework defined for the OPs, elevated exposures that continue over
multiple days would likely be of more concern using this method of analysis (i.e.,
MOE's calculated using a single day exposure), than if the elevated exposure
was for only a very short period, such as a few days. Using this mode of analysis
raises issues concerning the appropriate interpretation of examining the elevated
exposure of different individuals over multiple days. This population based
approach to risk estimation is discussed in more detail below.
As noted above Calendex™ can also calculate exposure over multiple
sequential days (e.g., 7, 14, 21) for each individual. In this mode of analysis
Calendex™ calculates single consecutive daily exposures for each individual as
described above, for the selected number of days, and then averages these
together. The resulting exposure distributions can be arrayed across the year in
the same way as the single day distributions. However, their interpretation would
likely be different. In this case elevated exposures over short time periods would
likely be of more concern than in the case of the single day analyses. Using this
mode of analysis raises issues concerning how appropriate the available data are
for conducting such a longitudinal (multiple consecutive day) analysis for an
individual.
It should be noted that the single day analysis does not depend on any
knowledge of the day-to-day exposure patterns of any particular individual, since
each day is modeled separately for each individual. This type of assessment,
therefore, highlights between-individual patterns of exposure (population risk)
rather than within- individual patterns of exposure (individual risk). Using this
approach the focus is on a snapshot of potential population risk from a variety of
sources. The likelihood of a sustained elevation in an individual's exposure is
anticipated to be lower than an elevated population exposure at any given
percentile. The rationale behind this conclusion is provided in the following
example. Few individuals are likely to repeat residential applications for every
day of the pest season. However, on a population basis, the upper percentiles of
exposure will reflect the phenomenon of a large number of individuals
encountering an increase in exposure due to performing these tasks. It is this
increase in population risk that may be a concern. As a result, this approach to
calculating the exposure to the population is considered to be health protective.
OPP will continue to pursue a series of further analyses to evaluate
alternative strategies for combining the data and selecting appropriate time
frames to consider. This will also be the subject of several questions presented
to the SAP in February 2002.
D. Other Models
The Agency is aware of three other models that have been developed or
are under development to conduct multi-pathway assessments and that can be
adapted to incorporate inputs for data from multiple chemicals. Two of these
have been presented to the SAP as aggregate risk assessment models:
Revised Draft
51
-------
LifeLine™, a model developed by Hampshire Research Institute, currently
licensed by the LifeLine Group; and Rex™, a product of Infosciences. Neither of
these packages appeared to provide the scope with regard to the number of
pathways, routes, or sources of pesticides required in the current OP cumulative
assessment. CARES, a product of ACPA, is still under development. It is
expected to be presented to the SAP in April/May 2002.
LifeLine™ is a multi-pathway model that can be adapted to evaluate
multiple chemicals. It focuses on identifying the periods during an individual's life
where pesticide exposures are likely to occur, and identifying the source of those
exposures. LifeLine™ produces a longitudinal estimate of possible exposures,
focusing on looking across many years of a person's life. It draws upon a subset
of natality records from the U.S. Census to develop the demographic
characteristics of the population under evaluation. Consumption data from the
CSFII are matched to the other information available using the demographic,
regional, and seasonal information from the two surveys. Residential exposure is
estimated by linking data from a group of surveys to develop scenario
characteristics that are anticipated to occur due to the use patterns of the group
of chemicals under evaluation.
CARES is intended to perform cumulative and aggregate assessments,
focusing on a population-based, cross-sectional analysis of a hypothetical year of
exposure. CARES is anticipated to generate a series of exposure estimates
moving across the calendar year, similar to that described for Calendex™. The
demographic characteristics of the population being assessed will be drawn from
a subset of the U.S. Census. CARES is intended to provide the user with a
flexible, easily used tool to develop total and pathway-specific estimates of
exposure, and to facilitate identification of the sources of exposure.
Revised Draft
52
-------
VI. Dietary Risk Assessment
A. Dietary Risk From Food
1. Residue and Consumption Data Used in the Food
Assessment
Dietary exposure from food is calculated considering what is eaten by
individuals in one day and residue values for each food. The food exposure
assessment is extensively refined using probabilistic Monte Carlo analyses.
Information on the amount of residues that may be on foods was obtained mainly
from the USDA's Pesticide Data Program (PDP) supplemented with information
from the Food and Drug Administration (FDA) Surveillance Monitoring Program
and Total Diet Study. PDP data were used for the commodities that have been
directly monitored as part of the program and were also used to estimate
residues on commodities where this data can be reliably used as a surrogate
(e.g., measured data for broccoli was used to estimate cauliflower residues).
Commodities directly monitored by PDP accounted for approximately 86% of the
diet of children 3-5 years old. Commodities for which surrogate PDP data were
used accounted for an additional 1.3 % of the diet of children 3-5 years old.
Consumption data were taken from the USDA Continuing Survey of Food
Intake by Individuals (CSFII 1994-96), and the 1998 Supplemental Children's
Survey. The CSFII records one-day food and nutrient intake data and is
considered to be representative of the U.S. population. The CSFII 1994-1996
contains survey data on 20,607 participants interviewed over two discontinuous
days. The supplemental children's survey includes an additional 5,459 children,
birth through 9 years old.
The Agency limited the food assessment to use of mainly PDP monitoring
data for several reasons. The PDP program is designed to provide the best
available data for risk assessments. PDP collects samples of selected food
commodities throughout the year on a nationwide basis. It focuses on foods
consumed by children and on foods typically available throughout the year.
Foods are washed and inedible portions removed before analysis. These
samples are analyzed for numerous pesticide residues and, therefore, capture
co-occurrence of different pesticide residues on a particular sample. The
distribution of residues that results from this program reflects a range of pesticide
use patterns. It also takes into account the percentage of the crop nationwide to
which each pesticide is typically applied (known as percent crop treated). Data
collected between 1994 and 2000 were used in the assessment. The PDP data
were adjusted to remove chemicals or uses that have been cancelled or are
being phased out.
Other available monitoring data are collected for different purposes than
Revised Draft
53
-------
those of the PDP program and are not necessarily designed to reflect the overall
consumption by the U.S. population. However, some FDA monitoring data were
used to supplement the PDP data. The FDA surveillance data measure residues
on commodities generally sampled closer to the point of production than for PDP.
The program has extensive data on eggs and fish, two commodities not sampled
in PDP. These data were used in the current assessment to support the
judgement that OP residues are negligible on eggs and fish. The FDA total diet
study is excellent for assessing the occurrence of pesticides in foods that have
actually been prepared for consumption; however, the number of samples
analyzed is very small. These data have been used in the current assessment to
estimate residues in meats other than poultry. The data show that only limited
residues of OP pesticides have been found on a few meats at low levels. This
information was used to develop a conservative residue estimate for meat
commodities. The maximum residue found for each type of meat in the 26
market baskets collected between 1991 and 1999 was used in the assessment.
Commodities for which FDA data were used in the assessment accounted for 6.3
% of the diet of children 3-5 years old.
The last case in which supplemental information was used in the
assessment is highly refined sugars and syrups. PDP includes high fructose
corn syrup and has found no pesticide residues. However, no other sugar or
syrup sources are included in PDP. The FDA total diet survey has analyzed
refined sugar and maple sugar and found no OPs in 26 market baskets. This
limited residue data together with the knowledge of the highly refined nature of
sugars and syrups is the basis for assuming negligible residues of OPs on
sugars and syrups. In the current assessment residues were assumed to be
zero for these foods derived from sugarcane, sugar beet, and maple. These
foods account for about 3% of the consumption of children 3-5 years old.
The following table summarizes the above discussion on the sources of
residue information used in the assessment and the percentage of children's
diets covered by each source.
PROPORTION OF THE DIET OF CHILDREN (3-5 years old)
COVERED IN THE CUMULATIVE FOOD ASSESSMENT
Source of Residue Estimate
Percent of Per Capita Consumption
PDP
85.7
Translation of PDP (PDP data used as
surrogate for other commodities)
1.3
FDA Data (eggs, fish, meat other than
poultry)
6.3
Assumed Negligible (sugars and syrup
sources)
3.1
Not Covered in Current Assessment
3.6
Revised Draft
54
-------
OP Market Basket Data
A task force of pesticide producers has provide the Agency with an OP
pesticide market basket survey. The results of this survey, conducted in 1998,
were submitted to the Agency in 2001. The final report is still under review but
the data are being examined to determine what they show concerning cumulative
exposure for OPs on food. Samples were taken from grocery stores and single-
serving size samples were analyzed by methods with very low limits of detection.
The foods collected, all of which are also covered by PDP, were apples, broccoli,
cherries, cucumbers, green beans, grapes, peaches, sweet corn, lettuce,
oranges, potatoes, strawberries, and tomatoes. Preliminary examination of the
data indicate that cumulative exposure estimates using these data are in general
agreement with a similar assessment using PDP data. These data will continue
to be examined.
2. Processing Factors Used
Residues of organophosphates may be either concentrated or reduced by
the activities of drying (e.g., prunes), processing (e.g., juice), washing, peeling
and cooking. The Agency uses processing factors to account for these situations
in the risk assessment. EPA has utilized, to the extent possible, the processing
studies that have been submitted to the Agency in support of the registration and
reregistration activities for the individual OP pesticides. In cases where no
acceptable data were available, the assessment relies on assumptions regarding
processing factors. The preliminary assessment lists the processing factors that
were used for each chemical/commodity (see Section 6 "Data" below).
3. Pesticides Included in the Food Assessment
After exclusion of data on pesticides that have been cancelled or do not
have food uses, there are residues for 22 OPs in the PDP data. The following 22
OPs have, therefore, been included in the food assessment.
acephate
azinphos methyl
chlorpyrifos
chlorpyrifos-m ethyl
disulfoton
diazinon
dichlorvos
dimethoate
ethoprop
fenamiphos
malathion
methidathion
methamidophos
mevinphos
methyl-parathion
oxydemeton-methyl
phorate
phosalone
phosmet
pirimiphos-methyl
terbufos
tribufos
The following chemicals, which have not been sampled in PDP, are not
expected to contribute to food risk for the reasons described below.
~ Naled degrades rapidly to dichlorvos and is analyzed and included in PDP
Revised Draft
55
-------
as dichlorvos
~ Bensulide is expected to have negligible residues based on field trial data
~ Cadusafos is used only on imported bananas and field trial data indicate
residues will not occur in the edible portion of the banana
~ Chlorethoxyfos and phostebupirim are used only in soil applications to
corn, therefore, significant residues in edible portions of the corn would not
be expected
~ Dicrotophos and profenofos are only used on cotton; cottonseed oil is the
only food commodity derived from cotton and any residues are expected
to be low due to the extent of refining and blending of the oil
~ The only food related use of trichlorfon and tetrachlorvinphos are livestock
uses-as a pour-on treatment of beef cattle and for livestock and livestock
premises respectively; any potential residues are expected to be covered
by the conservative residue estimate for meat commodities that is being
used in the assessment
4. Elements of the Cumulative Analysis Which May Differ From
Individual Chemical Assessments
a. Use of Composite Samples/Estimating Residues on a Sample-by-
Sample Basis
Only the residue data from composite samples were used in the
preliminary OP cumulative assessment. A single composite sample may
contain several individual servings of some foods (e.g., five pounds of
apples). For this assessment, it was assumed that residues found on the
composite samples adequately reflected the residues that would be on
any given single-serving contained in the sample.
In addition, all of the different chemical residues found on a sample
were summed to generate a single cumulative residue for each sample.
By summing on a sample-by-sample basis, the potential for capturing any
co-occurrence on the same commodity is enhanced. A majority of PDP
samples contained no detectable residues of any OP. For those that
contained detectable residues, a single OP was most common, but many
multi-residue samples were found. The maximum number of OPs on a
single PDP sample was five (this occurred on only 5 samples during the
period 1994-1999). For food forms (e.g., grains) that are highly blended
before consumption, the residue value used was the average of all the
cumulative residues for that food form.
b. Use of Zero as the Residue Estimate When No Detectable
Residues are Found
It has been the usual practice in Agency assessments on individual
pesticides to assume, for samples which showed non-detectable residues,
Revised Draft
56
-------
that residues are present at % the limit of detection (LOD) of the analytical
method for that part of the crop that is treated. For the untreated part of
the crop, residues are assumed to be zero. This procedure becomes
problematic for a cumulative assessment. It is not enough to simply
estimate the percent crop treated for each of the pesticides in the
cumulative assessment; it is also important to consider the potential for
co-occurrence of multiple residues on the same crop. In the current
assessment all OP residues reported as non-detectable are assumed to
be zero.
In a complex analysis such as this assessment, in which there are
abundant samples with detectable residues, the assumption of zero for
non-detects would not be expected to greatly impact the exposure
estimates at the highest percentiles of exposure. This assumption was
tested and found to be the case in an earlier stage of the assessment as
reported in the case study presented to the SAP in December 2000.
Cumulative food exposure assessments were conducted using two
extreme default assumptions: all non-detects = 0 and all non-detects = %
LOD for the chemical with the highest percent crop treated for a given
food.
c. Use of Measured PDP Data on Related Commodities that Were Not
Measured in PDP (Translation of Data)
In chemical specific dietary exposure assessments the Agency
routinely translates residue data from one food commodity to related ones
if the pesticide use patterns are similar on those commodities. For
example, data on cantaloupes is often used as surrogate data for
watermelons and other melons. For a cumulative assessment, in which a
grower has a choice of several chemicals from the cumulative assessment
group, these translations become more difficult to make. In the current
assessment, translations of the residue data were made exactly as they
are in the individual assessments except that a residue was not included if
the chemical was not registered on the crop that the data were being
translated to. This allowed maximum use of the PDP data. The
uncertainty introduced by this method is not expected to have a major
impact on the assessment because the foods for which translated data
were used comprise a relatively small portion of the per capita
consumption of children. An analysis of critical commodities contributing
to the higher percentiles of exposure in this assessment is currently under
way. If any translated foods appear in this analysis then the sources of
data for those specific contributors will be examined even more closely for
their validity as surrogate residue estimates.
d. Over-Tolerance and Other "Violative" Residues
Residue values that exceed the tolerance on the commodity and
residue values for commodities with no registered use for the associated
Revised Draft
57
-------
pesticide have been excluded. These "violative" residues are rare in both
PDP and FDA monitoring.
5. Risk Estimates for Food Alone
Separate assessments were conducted for children 1-2 years old, children
3-5, adults 20-49 , and adults 50+. The most highly exposed subgroup, that is
the group with highest estimated risks, are children 1-2 years old. The risk
estimates for this sub-group are presented below.
Estimated Percentile of per capita Days Falling Below Calculated
Exposure (mg/
kg/day) with Margin Of
Exposure for Children 1-2
Percentile
Exposure
Margin of Exposure (MOE)
90.00
0.000100
800
95.00
0.000176
454
97.50
0.000285
280
99.0
0.000499
160
99.50
0.000735
108
99.75
0.001045
76
99.90
0.001541
51
It is assumed in this assessment that food distribution and storage
systems in the U.S. result in essentially a national distribution of the major foods
in the U.S. diet that is constant throughout the year. Thus, there is no regional or
temporal component for the food only assessment. For some of the seasonal
changes in availability of certain foods, PDP has designed its sampling program
to concentrate on these time frames so that the residue data reflect the foods,
including imports, available to the consumer. This same national food estimate
is, therefore, combined with the specific regional water and residential
assessments to calculate each regional assessment.
6. Data
All of the data used in the preliminary OP cumulative risk assessment are
available in the public docket and on the internet at
www.epa.gov/pesticides/cumulative. The following summarizes the major data
tables related to the food assessment that are included in the preliminary risk
assessment and where they may be found.
~ Table III.C.1 in Section III (Appendices), beginning on page III.C.1 Page 1,
Revised Draft
58
-------
contains the source of the residue values used for every food in the
assessment (e.g., PDP, FDA total diet, etc.)
~ Table III.C.5 in Section III (Appendices), beginning on page III.C.5 Page 1,
contains all of the Processing Factors used in the assessment
~ Other Tables in Section C of the Appendices summarize the residues
found in PDP and FDA monitoring; show the co-occurrence of OPs in PDP
samples; provide the translation scheme used to apply PDP data to other
crops; and include a series of tables showing how the data were input into
the DEEM™ dietary software
B. Dietary Risk From Water
1. Introduction
Drinking water exposure to pesticides can occur through surface water
and ground water contamination. Potential for exposure to pesticides in drinking
water varies for different parts of the country and in different times of the year.
Contributing factors to these differences include time of pesticide application and
weather conditions shortly after application. These differences are also
influenced by the inherent local and regional differences in soils, crops, and site
vulnerabilities.
To make the water assessments reflect geographic variations as
realistically as possible, OPP used USDA Economic Research Service maps to
divide the continental United States into12 regions. These regions are grouped
according to similarity in crops. They take into account the geographic and
climatic differences that lead to different agronomic practices, pest pressures,
pesticide application methods and rates, and factors that affect pesticide
transport to water. Water was assessed for watersheds that are potentially
vulnerable to OP contamination within each of these regions. This regional
approach allows the assessments to account for effects on drinking water that
are driven by the different characteristics of these regions.
Scenarios for developing estimates of pesticides in drinking water within
each region were chosen based on organophosphate use, watershed
vulnerability (which accounts for such factors as rainfall frequency and intensity,
slope of the land, and soil type which affect pesticide runoff), and source of
drinking water (surface water or ground water). Information on the use of
different pesticides within the same region, the timing of use, and the fate and
transport properties of the pesticides was used to identify pesticides that are
likely to co-occur.
Factoring drinking water exposure into the framework for food exposures
means developing a person-by-person approach to estimating drinking water
exposure over time, which preserves the individual's demographic characteristics
and associates only those exposures that are appropriate for such an individual,
as described above in Section V. "Cumulative Exposure Models and
Revised Draft
59
-------
Interpretation of Model Outputs." The probabilistic cumulative risk assessment
for the organophosphates necessitates that drinking water exposures be based
on daily concentrations of pesticides in the drinking water sources. When longer
term exposure estimates are used, multiple sequential daily exposures would be
averaged to obtain the relevant exposure estimate. To estimate risk the
assessment used modeled distributions of daily concentrations of pesticides in a
probabilistic analysis.
The differences in the individual chemical and cumulative approaches for
the determination of pesticide concentrations in drinking water are summarized in
the following table.
Aggregate Screening vs Cumulative Assessments
Aggregate Screening Assessment
for A Single Pesticide
Cumulative Assessment for
Multiple Pesticides
point estimate (single value), 99.9th
percentile concentration
distribution of all daily concentrations
(13,000+ days)
national estimate (single site represents
entire US)
regional estimate (multiple sites,
regional differences)
national Percent Crop Area (PCA)
regional Cumulative Adjustment Factor
(CAF), reflecting variation in crop
intensity
maximum label rates & frequency,
minimum interval between applications
typical rates, frequencies, intervals
comparison of point estimate to DWLOC
value
probabilistic assessment of water
exposures
one compound at a time
multiple compounds considering co-
occurrence
2. Available Monitoring Data
EPA's three main sources of monitoring data for organophosphates
in water are:
(1) USGS ambient water samples, which include 9 currently
registered OPs,
(2) USGS-EPA reservoir monitoring project, which includes 27 OP
parent compounds and 19 OP transformation products, and
Revised Draft
60
-------
(3) ground and surface water monitoring information collected by
states or submitted by registrants.
The Agency is committed to using all available monitoring data as
extensively as possible. Monitoring data were used extensively in the
individual assessments, and the Agency has relied on these assessments
in developing its approach to the cumulative assessment. In addition to
guiding the Agency in focusing its regional assessments, monitoring data
were used for comparison to the modeling distributions for the cumulative
assessment.
However, two main considerations made it difficult to base the
cumulative assessment solely on monitoring data. First, the monitoring
databases are not robust enough to assess even a single chemical over
time in various regions of the country. Sampling is too infrequent to
assess daily concentrations. The lack of monitoring data for some
compounds makes it difficult to use the available data to assess the co-
occurrence of multiple chemicals over time across the country. The
available monitoring data was, however, used where possible to help
assess co-occurrence. Secondly, mitigation developed as the result of the
risk management for individual OP chemicals has resulted in use
deletions, lower application rates, and reduced numbers of applications.
The available monitoring data do not reflect these changes.
In summary, although the quantitative assessment was based on
modeled distributions used in a probabilistic assessment, water monitoring
data were used throughout the assessment in three main ways.
~ Groundwater monitoring data were used to assess the
vulnerability of groundwater to organophosphates.
~ Any available monitoring was used as background
information for scenario selection. The primary criterion for
scenario selection was actual use information, but available
monitoring data were also considered.
~ Monitoring data were used to evaluate modeling results at
every level of the assessment process.
Monitoring data confirm that OPs do occur in surface water drinking
water sources. The frequency of detections is generally low, except for
chlorpyrifos, diazinon, and, in some instances, malathion. While the
residential uses of chlorpyrifos and diazinon, which contributed to many of
those detections, have been cancelled, the individual chemical risk
mitigation for malathion is in progress. The magnitude of detections
generally ranges from sub-parts per billion to a few parts per billion.
Significantly greater frequencies of detection occur in the limited number
of targeted monitoring studies. In general, surface water sources are
Revised Draft
61
-------
more likely to be vulnerable to OP contamination than are ground water
sources.
The USGS-EPA Reservoir Monitoring Project although only in the
pilot stage, included more OPs than any previous study. Therefore, it is
particularly useful for considering the possibility of exposure to multiple
OPs. Of 314 intake samples, 137 (44%) had one or more detectable OPs.
Of the 137 with detectable OPs, 16 (12%) had more than one OP
detected. Of 67 outfall samples, 17 (25%) had one or more OPs detected.
Of these 17, 2 (12%) had more than one OP detected. Of the 12
reservoirs included in the study, no more than 3 or 4 were located in areas
with substantial OP use. A comparison of weather data during the
sampling period with long-term trends indicates that the first year of this
1.5 year study had drier than normal rainfall. Thus, results of this study
are not reflective of particularly vulnerable sites or weather conditions for
this OP assessment.
Of 218 finished water samples collected in the study, 24 (11 %) had
one detectable OP. None of the finished samples had more than one OP
detected. It is important to note that available evidence suggests that
water treatment may convert the parent OP compounds into compounds
that are also of toxicological concern. Not all of these transformation
products were included in the monitoring study. Thus, EPA cannot draw
definitive conclusions regarding the co-occurrence of parent OP and toxic
transformation products based solely on the results of this study.
Model estimates were compared to available monitoring data. This
comparison indicates that the assessment is by no means worst case or
unrealistic. In each region, levels of one or more OP pesticides detected
in monitoring studies are greater than that estimated by the cumulative
water assessment. In some cases, the model estimates are lower by an
order of magnitude or more. However, in that same region, estimates of
other OP pesticides are similar to or greater than detections found in
monitoring studies. Because the cumulative assessment focuses on the
cumulative impact from multiple OP pesticides, it does not necessarily
focus on the conditions that lead to the highest concentration of one
particular OP.
Although monitoring for OPs in treated drinking water is very
limited, the weight of evidence from available studies is that chlorination
may transform the OPs to oxons, sulfoxides, and sulfones which are of
toxicological concern. A few studies indicate that the oxon transformation
product will be stable in chlorinated water for at least 24 to 48 hours after
treatment.
3. Regional Approach
As shown on the map on the following page, the 48 contiguous
Revised Draft
62
-------
states were divided into 12 regions. These twelve regions were recently
developed by the USDA Economic Research Service to depict geographic
specialization in the production of U.S. farm commodities. The regions
represent areas with similar types of farms and similar physiographic, soil,
and climatic traits. By design, there are many similarities within each
region such as crops grown, application timing (use season), and
application rates. There are also many similarities in key environmental
factors affecting runoff, such as precipitation, irrigation practices, soil
types, and average slopes of the land. These regions provide a
framework for identifying one or more locations that represents an area of
the greatest concern for drinking water exposure within each region.
Considered together, this set of locations represents drinking water
exposure throughout the U.S. for the cumulative OP assessment.
Revised Draft
63
-------
Fruitful,Rim TslW
Northern GreaTPIains
Morthern.Gre'scent
Basin and Range
Heartland
EastirnJUp lands
FruitfuliRimfSW
Prairie'.Gateway
SouthentSeaboard
Mississippi Portal
Fruitful Rim^SE
Fruitful Rim TX
Source: USDA ERS
Revised Draft
64
-------
Each region in the assessment is represented by a geographic area
within the region that has the highest apparent potential for cumulative
exposure to OPs in drinking water. Each of these locations has a
relatively high usage of multiple OPs (in relation to other parts of the
region) which coincides with surface and/or ground water sources of
drinking water which are vulnerable to potential contamination by these
OPs. Since the purpose of the assessment is to identify the impact from
multiple OPs occurring in water in the same area, the area(s) selected for
the assessment do not necessarily represent the highest exposure of a
single chemical, but rather the highest multiple OP exposure within the
region. Locations within each region were selected using the following
steps:
~ Identify the high OP usage areas and high agricultural intensities
with each region
~ In each high usage area within the region determine the types and
locations of drinking water sources
~ Assess the vulnerability of drinking water sources within the high
usage area within the region; OPP adapted vulnerability schemes
proposed by Kellog and others at USDA for this purpose
~ Compare locations of surface drinking water intakes overlain on
runoff vulnerability maps to determine where potentially vulnerable
surface water sources of drinking water coincided with high use
areas; for groundwater, compare OP use areas with a leaching
vulnerability map
Details of this process of selecting a location to represent each
region are provided in each regional assessment. One region, Region
7-the Fruitful Rim, Ca-which covers the central and coastal valleys of
California, southern California, and south-central Arizona had two different
locations selected. The remaining 11 regions were represented by a
single location. Region 8-the Basin and Range-which covers Nevada,
Utah, most of Oregon, and part of California, Washington, Idaho,
Wyoming, Colorado, New Mexico, Arizona and Montana was represented
by the same scenario developed for Region 3, the Northern Great Plains.
Region 11 —the Fruitful Rim, Texas-which covers much of the eastern
coast of Texas, was represented by the same scenario developed for
Region 4, the Prairie Gateway.
The Northwest fruitful Rim provides an illustration of the location
selection process within a region. Three high OP-use areas occur in the
Northwest Fruitful Rim: Yakima County and eastern Washington are the
highest OP use area (predominantly on orchards) and have the highest
percent crop area. The Snake River Valley in southeast Idaho is the
second highest use area (predominantly on potatoes and sugar beets).
The Willamette Valley, Oregon, is the third high-use area with a mix of OP
uses. There are predominantly ground-water sources of drinking water in
Idaho and eastern Washington, with vulnerability to leaching potentially
Revised Draft
65
-------
higher in eastern Washington. A few surface-water intakes occur in the
Yakima County area. The Willamette Valley has more surface water
intakes and is more vulnerable to runoff. Available monitoring from
NAWQA study units in Willamette Valley, Snake River Basin, and Pugett
Sound suggest that the Willamette Valley will be more vulnerable to OP
contamination with a higher potential for co-occurrence of multiple
pesticides.
The surface water assessment for the Northwest Fruitful Rim was
therefore based on the Willamette Valley in Oregon. Potential impacts of
OP pesticides on ground water resources in eastern Washington and
southeast Idaho were qualitatively analyzed, relying largely on ground-
water monitoring available through the USGS NAWQA program and state
monitoring programs.
The following table shows, for each region, the location selected for
the surface water assessment and the crops and pesticides used in the
estimates.
Locations, Crops, and Pesticides
Inc
uded in the
Regional Water Assessments
REGION
LOCATION
CROPS
PESTICIDES
1) Heartland
Central Illinois
Corn
Terbufos
Chlorethoxyphos
Chorpyrifos
Phostebupirim
2) Northern Crescent
South Central
Apple
Chlorpyrifos
Pennsylvania
Pear
Dimethoate
Peach
Azinphos-methyl
Corn
Diazinon
Alfalfa
Terbufos
Pumpkin
Methyl parathion
Cantaloupe
Phosmet
Corn
Methidathion
Phostebupirim
3) Northern Great Plains
Red River Valley
Corn
Chlorpyrifos
(Minnesota, North
Wheat
Dimethoate
Dakota)
Sugar Beet
Azinphos-methyl
Potato
Phorate
Terbufos
4) Prairie Gateway
Central Hills,
Cotton
Acephate
Texas
Corn
Azinphos-methyl
Alfalfa
Chlorpyrifos
Wheat
Dicrotophos
Potato
Malathion
Sorghum
Methyl parathion
Phorate
Phostebupirim
Terbufos
Tribufos
Dimethoate
Revised Draft
66
-------
5) Eastern Uplands
Western North
Apple
Azinphos-methyl
Carolina
Alfalfa
Chlorpyrifos
Corn
Dimethoate
Methyl Parathion
Phosmet
Terbufos
6) Southern Seaboard
East Coastal
Cotton
Acephate
Plain, North
Tobacco
Chlorpyrifos
Carolina
Corn
Dimethoate
Peanuts
Disulfoton
Ethoprop
Fenamiphos
Phorate
Terbufos
Tribufos
7a) Fruitful Rim, CA
North Central
Corn
Nectarine
Acephate
Methidathion
Valley, California
Alfalfa
Asparagus
Azinphos-methyl
Methyl-
Almond
Melons
Chlorpyrifos
pa rathion
(Walnut)
Cucumber
Diazinon
Naled
Pear
Pumpkin
Dimethoate
ODM
Peach
Squash
Disulfoton
Phorate
Apricot
Tomato
Fenamiphos
Phosmet
Grape
Dry Beans
Fonofos
Plum
Sugar Beet
Malathion
Apples
Cantaloupe
Methamidophos
(Pear)
Broccoli
7b)Fruitful Rim, CA
South Central
Cotton
Cantaloupe
Acephate
Malathion
Valley, California
Citrus
Apricot
Azinphos-methyl
Methyl-pa rathion
Alfalfa
Nectarine
Bensulide
Naled
Almond
Plum
Chlorpyrifos
ODM
(Walnut)
Grape
DDVP
Phorate
Apple
Sugar Beet
Diazinon
Phosmet
(Pear)
Lettuce
Dimethoate
Profenofos
Orange
Prune
Disulfoton
Tribufos
Melon
Broccoli
Fenamiphos
Peach
Methamidophos
Methidathion
8) Basin and Range
Region 3 location used.
9) Mississippi Portal
Southern
Cotton
Acephate
Methyl-pa rathion
Louisiana
Corn
Chlorpyrifos
Phorate
Soybean
Dicrotophs
ProfenofosteTer
Dimethoate
bufos
Disulfoton
Tribufos
Malathion
Methamidophos
Phostebupirim
10) Fruitful Rim, NW
Willamette Valley,
Apple
Cauliflower
Acephate
Oregon
Pear
Blackberries
Azinphos-methyl
Cherry
Blueberries
Bensulide
Pea
Raspberries
Chlorpyrifos
Broccoli
Cucumber
Diazinon
Cabbage
Sweet Corn
Dimethoate
Nursery
Mint
Siculfoton
(Trees &
Hazelnut
Ethoprop
Shrubs)
Sweet and Tart
Malathion
Squash
Cherries
Methidathion
Onion
Christmas
Methyl-pa rathion
Snap
Trees
Naled
Beans
ODM
Phosmet
Revised Draft
67
-------
11) Fruitful Rim, TX
Region 4 location used.
12) Fruitful Rim, FL
South Central,
Tomato
Acephate
Florida
Pepper
Chlorpyrifos
Cucumber
Diazinon
Watermelon
Ethoprop
Sweet Corn
Methamidophos
Lettuce
Phorate
Citrus
Corn
Sugarcane
4. Assessment Tools
The limitations of the available monitoring data for estimating daily
concentrations of OP pesticides in multiple regions, together with
recommendations from the SAP, resulted in the evaluation of modeling
tools that would allow production of time-linked regional assessments
which are as realistic as possible.
Surface Water
After consideration of available predictive tools, EPA used the
PRZM/EXAMS-IR model which has been modified by using scenarios and
inputs that are specifically designed for performing drinking water
assessments. The model simulates runoff into an index drinking water
reservoir (IR), which is based on Shipman City Lake in Shipman, Illinois.
The PRZM component of the model is designed to predict the
pesticide concentration dissolved in runoff waters and carried on
sediments from the field where it was applied to an adjoining edge-of-field
surface water body. The model can simulate specific site, pesticide, and
management properties including soil properties (organic matter, water
holding capacity, bulk density), site characteristics (slope and surface
roughness of the land, field geometry), pesticide application parameters
(application rate, frequency, spray drift, application depth, application
efficiency, application methods), agricultural management practices
(tillage practices, irrigation), and pesticide environmental fate and
transport properties (soil half-life, soil:water partitioning coefficients, foliar
degradation and dissipation, and volatilization). OPP selected a
combination of these different properties to represent a location-specific
scenario for each particular pesticide-crop regime in each region.
The EXAMS component of the model is used to simulate
environmental fate and transport processes of pesticides in surface water
(after they have reached the edge-of-field water body). The model
simulates abiotic and biotic degradation, sediment:water partitioning, and
volatilization. The actual inputs used in the model for each
region/location/crop/pesticide combination are provided in the Appendices
of the risk assessment (see Section 6 "Data" below).
Revised Draft
68
-------
Changes have been made to model input parameters to produce
outputs that focus on co-occurrence of pesticides, the central concern of
the cumulative risk assessment. For example, a regional as opposed to a
national Percent Crop Area (PCA) is being used in the model to account
for the amount of land on which crops are grown in the different localities
where the drinking water is being assessed. In addition, all available crop-
pesticide use information was utilized. Instead of using the maximum
label rates, maximum numbers of applications, and minimal time intervals
between applications, EPA used typical rates, frequencies and intervals,
which again makes the model outputs more realistic and likely closer to
the actual agronomic practices of the growers. The use of the model
output has also been changed. Instead of generating one conservative
high end exposure number, the Agency used all 13,000-plus daily
concentration values to produce distributions that were used in a
probabilistic risk assessment for the different regions.
The Agency believes that this approach is the best method
currently available to estimate daily residue concentrations in drinking
water:
~ it allows for the assessment of multiple chemicals
~ the data generated span a time frame of over 30 years, which
captures the variability due to changing weather conditions
~ distributions can be generated in different locations across the
entire country thus capturing regional variability
~ since daily distributions are generated, it is possible to maintain the
time dependency that is needed for this type of risk assessment
~ because the entire distribution was used in the Calendex™ model
the differences in exposures on different days are taken into
account.
Each of these aspects of the assessment is discussed in more
detail below.
Use of the Full Distribution of PRZM/EXAMS-IR Output
The following graph illustrates the full output of the PRZM/EXAMS-
IR model for a given region when 36 years of weather data are available.
It also shows the concentration point estimate that would be selected
when the screening level assessment is used.
Revised Draft
69
-------
Daily Distribu
~.~ID
0.000 ^
— PesticideA
— Screen
l-Jan 31-Dec 31-Dec 31-Dec 31-Dec 30-Dec 30-Dec 30-Dec 30-Dec 23-Dec 23-Dec
Time (spans multiple years)
Drinking water residues are estimated from this output for a specific
day in this specific region, by selecting one of the 36 estimates available
for that day. This is in contrast to the individual chemical assessments
where a single point estimate corresponding to approximately the 99.9th
percentile concentration is used in the initial assessment. Use of the full
distribution in the cumulative assessment allows the probabilistic risk
assessment to take account of the day-to-day variations in expected
pesticide concentrations across the year in a specific location.
Because the application rates, frequencies, and timing of
applications are held constant, the PRZM/EXAMS-IR estimates over
multiple years evaluate the impact of the variability in precipitation on the
amount of pesticide that reaches surface water. Because weather data
spanning 24 to 36 years is available for many locations across the
country, PRZM/EXAMS-IR can account for OP runoff from a wide range of
weather patterns not otherwise possible with monitoring studies that span
relatively few years.
Cumulative Adjustment Factors for Cropped Area and OP Crop Use
The estimation of separate regional risk assessments allows use of
a Cumulative Adjustment Factor (CAF) based on the total reported
number of acres which receive OP applications in a particular region. In
the individual chemical assessments, the adjustment factor most often
used was the Percent Cropped Area (PCA) of 87% which represents the
highest percentage of agriculture (cropped area) in any large watershed in
Revised Draft
70
-------
the U.S. Using a regional approach, regions with less intense cropping
will have lower estimated concentrations based on a regional CAF
compared to the national PCA.
The following example illustrates how CAFs are calculated and
used in the cumulative assessment. The CAF is calculated as:
CAFop/crop = (Acres Planted AIIOPcrops/Total Acres inWatershed) X
(Acres Treated OP.Crop/Acres Planted A„opCrops)
The CAF for each OP chemical/crop combination is used to adjust the
initial output of PRZM/EXAMS-IR for that chemical/crop combination. This
adjustment allows the PRZM/EXAMS-IR output to reflect only the
contribution of the number of acres estimated to be treated with that
pesticide.
EXAMPLE:
~ The total area for a location (watershed) is 800,000 acres.
~ Agricultural crops treated with OPs account for 320,000 acres.
~ Two crops, corn and alfalfa, are treated with OP Pesticides.
~ 60,000 acres of corn are treated with Pesticide A and 40, 000 with
Pesticide B.
~ 16, 000 acres of alfalfa are treated with pesticide A and 10,000 with
Pesticide B.
The four CAFs for this region are calculated below.
CAF Corn.OP(A)= (320,000/800,000) X (60,000/320,000) = 0.075
CAF Corn.OP(B)= (320,000/800,000) X (40,000/320,000) = 0.05
CAF A|fa|fa-op(A)= (320,000/800,000) X (16,000/320,000) = 0.02
CAF A|fa|fa.OP(B) = (320,000/800,000) X (10,000/320,000) = 0.0125
Each daily residue estimate from PRZM/EXAMS-IR for pesticide A
application to corn in the region would be multiplied by 0.075. Similarly
each daily residue estimate from pesticide B application to corn would be
multiplied by 0.05. And the corresponding adjustments would be made for
the alfalfa estimates. In this manner, since the use statistics come from
actual reported data in the region, competing and compatible uses of the
various OPs applied in that region are taken into account.
It should be noted that the percent cropped area part of the regional
CAFs (Acres Planted AIIOPCrops/Total Acres inWatershed) is based on data from
Revised Draft
71
-------
a large area. The size of the hydrologic units used (average > 1000
square miles) generally span multiple counties and may contain several
watersheds that supply drinking water intakes. These regional PCAs
represent the aggregation of crop areas from county-level NASS data and
assume that the cropping area is uniformly distributed. (For the source of
these data see Section 6 "Data" below.) However, in reality, cropping
intensity is variable and smaller watersheds, including those capable of
supporting drinking water supplies, may have a much higher percentage
of crop land than the rest of the large basin.
An example is Zollner Creek in the Willamette River Valley. This
watershed had the highest concentrations and frequencies of detection of
OPs among all of the NAWQA monitoring sites in the Willamette Valley.
This stream drained a watershed that was 99% agriculture, much greater
than the regional PCA of 60% used in this assessment. The regional
assessment areas generally coincided with the area with the highest PCA.
However, in some regions, such as the Northwest Fruitful Rim and the
Eastern Uplands, the regional assessment focused in a lower-intensity
agricultural area which was other wise more vulnerable because of OP
usage and/or the nature of the drinking water source.
The percent acres treated part of the CAF (Acres Treated op_
Crop/Acres Planted ANOp crops) 's derived from state-level data (or NASS
reporting districts) and assumes uniform use practices across the state. In
reality an uneven distribution of percent acres treated would be expected
in response to differing pest pressures. Thus, this assumption will
underestimate areas where pest pressures may dictate a higher
percentage of acres treated in a given year; similarly, it will overestimate
areas where low pest pressures will require fewer acre treatments.
Pesticide Application Information Used in the Assessment
Typical usage was estimated by dividing the pounds reported as
applied in a given area by the acres treated in that area. Estimates for
these statistics were generally taken from the most recent USDA pesticide
use surveys to reflect current practices. This derivation of the "typical"
number assumes that all applications were made at this typical or average
rate and that frequencies of applications were constant year to year. The
assessment considered only yearly variations in weather, and not
variations in application rates. This contrasts with the individual chemical
assessments where the range of rates considered always includes the
maximum label rates.
With the exception of Region 7 (California), application dates were
determined based on pesticide application windows established for each
of the OP pesticide/crop combinations in each region. This window
represents an approximate beginning and ending date for the use of the
pesticide on a particular crop. In many cases USDA handbooks also
Revised Draft
72
-------
provide "most active" periods during the planting and/or harvesting
windows. The mid-point of the most-active period was selected as the
application date for a pesticide applied at the "planting" stage of crop
production and, for example, for the defoliant tribufos used as a harvest
aid for corn. When "most active" periods were not available, the single
application date for a pesticide is the beginning of the crop stage window.
Multiple applications, such as OP cover sprays for tree fruits, were placed
at the beginning and equidistant within the application window. A most
likely, or predominant, application method was designated for each
pesticide-either air or ground application.
Sensitivity analyses may be conducted on each of these use
parameters: the percent of crop treated, the typical application rate, and
number and corresponding dates of applications. A preliminary evaluation
of selecting the midpoint date of the most active application period as the
application date of the pesticide on a particular crop has been done for the
Heartland region. This evaluation found that variations based on date
selection may result in differences of approximately two to three times in
cumulative concentration estimates. In the case of the Heartland, the
highest concentrations were estimated when applications were assumed
to be made at the end of the most active application period rather than at
the midpoint, which was used in the probabilistic exposure assessment.
Section 6 "Data" describes the source of the data used to make
these decisions and where the complete information can be found in the
risk assessment.
Ground Water
Ground water estimated concentrations were not included
quantitatively in the risk estimates. However, in areas of the United States
that receive their drinking water from ground water, monitoring data from
vulnerable ground water sources were examined. In each region, it was
determined that the surface water estimates would be protective of
groundwater, i.e., the surface water estimates would be expected to be
higher than any groundwater estimates. In those cases the surface water
estimates were considered to cover groundwater.
The concentrations of OPs in ground water were not significant in
most regions due to the fate parameters (chemical properties) of the
organophosphate class of compounds. This class is not very persistent or
mobile in the environment. Persistence and mobility are necessary for
pesticides to move through soils and contaminate ground water. The
available data generally do not provide evidence that parent OP pesticides
will co-occur in groundwater. Few data are available to determine if OP
transformation products might co-occur in groundwater.
An example of an exception to these general conclusions is
Revised Draft
73
-------
Fenamiphos. Fenamiphos and its degradates, fenamiphos sulfoxide and
fenamiphos sulfone, have been detected at high levels in ground-water
studies conducted in Florida, and to a lesser extent in California. Such
detections led to the phase-out of fenamiphos use on citrus in the Central
Ridge area of Florida. The individual chemical risk mitigation for
fenamiphos has not been completed. As noted above, each regional
assessment discusses the available information on OPs in groundwater.
5. Results
The preliminary OP cumulative risk assessment indicates that
drinking water is not a major contributor to the total cumulative risk. For all
regions the results of the assessment indicate that the contribution to the
OP cumulative risk from drinking water is generally at least one order of
magnitude lower than the contribution from OPs in food at percentiles of
exposure above the 95th, for all population subgroups evaluated. As the
percentile of exposure increases, the difference between the food and
water contributions increase.
6. Data
All of the data used in the preliminary OP cumulative risk assessment are
available in the public docket and on the internet at
www.epa.gov/pesticides/cumulative. The following summarizes the major
data sources used, the tables related to the water assessment that are
included in the preliminary risk assessment, and where they may be
found.
Water Monitoring Data
~ Section III.E. 1 (Appendices), beginning on page III.E.1 Page 1,
contains the USGS NAWQA program data on the occurrence of
OPs in ambient water
~ Section III.E.2 (Appendices), beginning on page III.E.2 Page 1,
contains data from state monitoring programs
~ Section III.E.3 (Appendices), beginning on page III.E.3 Page 1,
contains data from the USGS-EPA Pilot reservoir monitoring
program
Inputs to PRZM/EXAMS-IR
~ Section III.E.5 (Appendices), beginning on page III.E.5 Page 1,
contains, for each OP in the water assessment, the chemical
specific inputs that were used and their source
~ Section III.E.6 (Appendices), beginning on page III.E.6 Page 1,
contains, for each OP/Crop combination in each region the
scenario specific inputs: application method, incorporation depth,
Revised Draft
74
-------
application rate, application efficiency, spray drift, application dates,
and application frequencies and intervals.
~ Section III.E.7 (Appendices), beginning on Page III.E.7 Page 1,
contains background information on the remaining inputs to PRZM
which are used for each crop scenario
Water Treatment Data
~ Section II.E.4 (Appendices), beginning on page III.E.4 Page 1,
contains the available information of the effects of drinking water
treatment on OPs
Other Information
~ Each regional assessment contains other use information including
the sources of that information as well as information on locations
of surface water intakes of drinking water in the region.
Revised Draft
75
-------
VII. Residential (& Other Non-
occupational) Risk Assessment
The residential component of the preliminary OP cumulative risk assessment is
the most sophisticated analysis of its type that OPP has ever conducted. It is the first
application of distributional analysis to residential exposure assessments. It also factors
in the seasonal and regional aspects of pesticide use.
Potential for exposure to pesticides from residential and other non-occupational
uses differs in different parts of the country and at different times of the year.
Contributing factors to these differences include amount and time of pesticide
application. OPP has used the calendar based model Calendex™ to address the
temporal aspects of the residential use of pesticides in 12 distinct geographic regions
throughout the U.S. These regions are the same regions used in the water
assessment. Although based on major crop growing areas, these regions also present
an opportunity to consider the unique climate patterns and pest patterns that influence
residential pesticide use and expected exposure. Calendex™ allows delineation of the
critical timing aspects of seasonal use of OPs that result in exposure to pesticides and
enables the identification of potential co-occurrences from multiple sources.
Exposures to pesticides can occur through dermal, inhalation, and non-dietary
ingestion routes as a result of homeowner (i.e., "do-it-yourself) and commercial
applications in residential and public areas. Exposure can result from mixing, loading,
and applying the pesticide, and/or reentering a treated site. Residential exposure to
organophosphates in outdoor settings may result from applications to lawns,
ornamentals, and "backyard" orchards and vegetable gardens. Indoor
organophosphate exposures may result from crack and crevice treatments, use of pest
strips, and from pet products (e.g., impregnated collars, dips, powders). EPA also
considered post-application exposures in indoor/outdoor public areas such as parks,
recreational areas, golf courses, schools or office buildings. Furthermore, the risk
assessment includes residential bystander exposures from public health uses of
organophosphates (e.g., mosquito and blackfly abatement). Certain residential uses
that are assumed to result in negligible exposure (e.g., ant/roach bait stations in child
resistant packaging or post-application exposure to treated fire ant mounds) were not
included in the assessment. That was the case in the individual chemical assessments
as well. The following chart delineates the current residential use picture for the
organophosphates:
Revised Draft
76
-------
Residential Uses for the Organophosphates
Chemical
Indoor
Residential
Uses
Outdoor
Residential
Uses
Golf Course
and Public
Area Uses
Pet Uses
Public
Health
Uses
acephate
N/A
Ornamentals,
residential turf, sod
farms
Golf course turf
N/A
N/A
bensulide
N/A
Residential turf
Golf course turf
N/A
N/A
chlorpyrifos
N/A
N/A
Golf course and
sod farm turf
N/A
Mosquito
adulticide
dichlorvos (DDVP)
Resin pest strips,
crack and crevice
(professional
applicators only)
Residential turf and
ornamental plants
(professional
applicators only)
N/A
Flea collars,
sponge, spray and
dip (applied by
veterinarians only)
N/A
disulfoton
Potted plant
treatments
Flower gardens,
roses, ornamentals,
shrubs, small trees.
N/A
N/A
N/A
fenamiphos
N/A
N/A
Golf course turf
N/A
N/A
fenthion
N/A
N/A
N/A
N/A
Mosquito
adulticide
malathion
N/A
Residential turf,
ornamentals,
garden, fruit trees.
Golf course turf,
pick-your-own
strawberries/orchar
ds, turf in public
areas
N/A
Mosquito
adulticide
naled
N/A
N/A
N/A
N/A
Mosquito
adulticide,
black fly
control
tetrachlorvinphos
N/A
N/A
N/A
Dips, powders,
sprays, and flea
collars.
N/A
trichlorfon
N/A
Residential turf and
ornamentals
Golf course turf,
turf in public areas
N/A
N/A
Seventeen OPs had registered uses when the Food Quality Protection Act
(FQPA) was passed in 1996. Seven of these have been excluded from the
cumulative residential assessment since all residential uses with any significant
exposure or risk have been eliminated. These pesticides are: chlorpyrifos,
diazinon, dimethoate, ethoprop, fenitrothion, phosmet, and protetamphos. Six of
the remaining 10 OPs have completed individual risk mitigation and the
cumulative assessment reflects the most up-to-date residential use picture for
these chemicals: acephate, bensulide, disulfoton, fenthion, naled, and
Revised Draft
77
-------
trichlorfon. Four OPs are still in the risk mitigation process, and any future risk
mitigation actions will be incorporated into the revised cumulative assessment:
dichlorvos (DDVP), fenamiphos, malathion, and tetrachlorvinphos.
Two OPs, tetrachlorvinphos and DDVP are currently registered for use on
pets. EPA did not have sufficient data on exposure for these uses to include
them in a calendar-based probabilistic assessment. The screening level
assessments for these uses indicate risks of concern. As noted above, the
individual chemical risk mitigation for these chemicals is in progress.
Other OP uses were not included because they resulted in negligible
exposures or because their single chemical assessments showed very low risk.
These low exposure uses include ant and roach baits, paint additives, post-
application residential exposure from sod farm applications, and applications to
fire ant mounds. Chlorpyrifos use for mosquitoes was not included because very
low exposures and risk were estimated in the single chemical, screening level
assessment.
A. Spatial and Temporal Aspects of the Residential
Assessment
Information relating to both the temporal and spatial aspects of exposure
is reflected in the residential portion of the cumulative risk assessment. The
assessment matches exposure scenarios with uses representative of a particular
region. The residential risk assessment scenarios are based on application
timing, duration of use, and frequency of application for each chemical in each
region. For example, if you live in Buffalo, New York, and it's January, you will
not be exposed to pesticides by mowing your lawn.
Chemical use patterns greatly affect potential exposure scenarios. By
evaluating a pesticide's geographic and temporal pattern of use, a profile for
each chemical can be developed to establish the potential routes, durations, and
frequencies of exposure. Also, the evaluation of chemical use profiles allows for
the identification of exposure scenarios that may overlap, co-occur, or vary
among chemicals. These possible exposures will then be associated to
individuals in the assessment, again preserving linkages to demographic
characteristics of the individuals as well as appropriate linkages in uses. In some
cases, products may serve essentially the same purpose, such that the use of
one will almost certainly preclude the use of the other, that is, they are
competitors.
The chart on the following page provides a visual example of the results of
the likelihood and frequency assumptions for the assessment within one example
region. It displays the various residential applications and their timing (including
repeated applications) over the course of a year, for one region/site. Each
regional assessment contains a chart like this for that region's uses.
Revised Draft
78
-------
January | February | March | April | May [ June | July | August | September | October | November | December
Chanical
O chinch
Cherni cal C: cr ack &crevice
Cherni cal U: fleas
Chemical U: fleas
Chemical T: grubs
Chemical T:crack & crevice
Chemical P: cyuhs
Chemical P: crack & crevice
Chemical Et: fleas Chemical B: fleas
Chemica
hshruhs
Chemi cd'Is hruhs
Chemi cal V: -s hrti*s
ChemicahV: shrubs
These likelihood and frequency assumptions for residential scenarios were
used to superimpose a pattern of relevant residential exposures that could
reasonably be expected to occur throughout the year for a given
individual/population member in the region. Any individual's exposure is based
on probabilistic methods that account for the percentage of the population likely
to be using the product in the first place, as well as preserve relevant time,
space, and demographic characteristics associated with the individual and his
probability of exposure. A detailed discussion of these methods is contained in
Appendix II of this document.
Five residential scenarios were used in the assessment. They represent
the critical OP uses that have the potential for significant exposure or risk when
considered in a cumulative assessment. These are:
~ Lawn care
~ Home vegetable gardens/ornamentals/orchards
~ Golf courses
~ Wide area public health sprays
~ Indoor (crack and crevice sprays and impregnated pest strips)
The following table shows, for each region, the residential scenarios that were
assessed and the pesticides used in the estimates.
Revised Draft
79
-------
Residential Scenarios and Pesticides
Included in the Regional Residential Assessments
REGION
SCENARIO
PESTICIDES
1) Heartland
Lawn
DDVP, Malathion, Trichlorfon
Golf Course
Bensulide, Trichlorfon
Gardens
Malathion, Acephate, Disulfoton
Indoor
DDVP
2) Northern Crescent
Lawn
Malathion, Trichlorfon
Golf Course
Bensulide, Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
Public Health
Malathion, Naled
3) Northern Great Plains
Lawn
DDVP, Trichlorfon
Golf Course
Bensulide, Malathion
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
4) Prairie Gateway
Lawn
Bensulide, DDVP, Malathion, Trichlorfon
Golf Course
Acephate, Bensulide, Fenamiphos, Malathion,
Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
Public Health
Malathion
5) Eastern Uplands
Lawn
DDVP, Malathion, Trichlorfon
Golf Course
Acephate, Bensulide, Fenamiphos, Malathion,
Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
6) Southern Seaboard
Lawn
Malathion, Trichlorfon
Golf Course
Acephate, Bensulide, Fenamiphos, Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
Public Health
Malathion
7a) Fruitful Rim, CA
Lawn
Malathion, Trichlorfon
Golf Course
Bensulide, Fenamiphos, Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
Revised Draft
80
-------
7b)Fruitful Rim, CA
Lawn
Malathion, Trichlorfon
Golf Course
Bensulide, fenamiphos, Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
8) Basin and Range
Lawn
Malathion, trichlorfon
Golf Course
Bensulide, Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
9) Mississippi Portal
Lawn
Malathion, Trichlorfon
Golf Course
Acephate, Trichlorfon, Malathion
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
Public Health
Malathion, Fenthion
10) Fruitful Rim, NW
Lawn
Malathion, Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
11) Fruitful Rim, TX
Lawn
Malathion, Trichlorfon
Golf Course
Acephate, Bensulide, Fenamiphos, Malathion,
Trichlorfon
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
Public Health
Malathion
12) Fruitful Rim, FL
Lawn
Malathion, Trichlorfon
Golf Course
Acephate, Bensulide, Fenamiphos
Gardens
Acephate, Disulfoton, Malathion
Indoor
DDVP
Public Health
Malathion, Naled
B. Hazard
The estimated exposures to each pesticide will be converted to index
chemical equivalents using route-specific relative potency factors for oral,
dermal, and inhalation exposures, as described above in Section IV. "Endpoint
Selection." Exposures will be compared to route-specific BMD10 values of the
index chemical to develop route specific MOEs. Oral, dermal, and inhalation
MOEs were combined by taking the inverse of the MOE for each route, adding
these together, and then taking the inverse of that sum to get the total MOE for
the oral, dermal, and inhalation routes of exposure.
Revised Draft
81
-------
C. Types of Data Used in the Assessment
Three types of data are used in the residential assessment: pesticide use;
pesticide residue dissipation; and exposure contact and exposure factors.
Pesticide use data are utilized to determine the percent of households using a
pesticide, whether the applicator is a professional or not, the timing of the
pesticide treatments, and frequency and duration of exposure. Use data are also
important in identifying geographic regions where the pesticide may be applied.
This type of information is needed together with chemical residue fate, residue
contact data and exposure factors to predict the potential for co-occurrence of
exposure events in cumulative assessments.
In this assessment, use data are specific to the region under evaluation.
Pesticide residue dissipation data address the fate of the pesticides once applied
and much of this data is region specific also. Exposure contact data describes
how often humans are expected to come into contact with the chemical or its
residues. Human exposure factors, such as breathing rates, body weight and
surface areas used in this assessment come from the Agency Exposure Factors
Handbook. Other exposure factors such as the size of the area being assessed
(e.g., the lawn) and time spent in the area are also important in assessing risk.
The data used in the assessment are discussed below.
1. Use Data
The majority of use-related information in the cumulative risk assessment
was obtained from the sources described below:
~ National Garden Survey (1996 -1997) tracks the percent of households
employing lawn care applicators and was used to estimate the number of
households that may use a given pesticide; it also contains information on
variables such as vegetable garden size
~ National Home and Garden Pesticide Use Survey (1989-1990) delineates
percent of households using pesticides based on a large national survey.
These values consider users and non-users as well as homes having
lawns and those that do not.
~ National Home and Garden Pesticide Use Survey (1991) provided
information on indoor use of DDVP
~ Survey data were also used to estimate frequency of applications, the type
of application equipment used, and the type of clothing worn while making
applications
~ Doanes GolfTrak (1998-1999) was used to identify the percent of golf
courses treated with OP pesticides; a 1992 survey conducted by the
Center for Golf Course Management was used to establish the percent of
individuals playing golf
~ The Occupational and Residential Exposure Task Force (ORETF)
provided estimates of time spent in the garden performing post-application
activities as well as information on the frequency of applications
Revised Draft
82
-------
~ Regional Cooperative Extension Services recommendations were used to
determine the timing of pesticide application windows, especially for turf
uses, but also for timing aspects of various gardening activities.
~ For Public Health Uses estimates of use and timing of use are based on
information provided by representatives of Florida Mosquito Abatement
Districts, Florida A&M; and Health Canada (black fly). Where specific
timing information was not available for regions having public health spray
uses, a spray schedule of once every two weeks was assumed for the
summer season.
~ Non-Occupational Pesticide Exposure Survey (NOPES)-provided
information on pest strips
2. Exposure Data
The major generic exposure factors used for each exposure scenario
included in the assessment are shown below. In addition to this information,
each regional assessment contains a chart showing, for each specific use
scenario considered in that region, the specific input data that were used for:
application method, amount applied, number and frequency of applications, the
period of time over which it may be used, % applied by a professional, % applied
by the homeowner, % of households in the region using the chemical, and the
active exposure period (how long residues are available for contact after
application). In addition, each regional assessment contains information on the
region specific residue dissipation data sources that were used.
The reasoning behind the selection of the following exposure factors and
the specific data source for each of them is contained in the risk assessment in
section I.D. pages 1-19. It should be noted that all of the data were obtained
from actual measured data of some kind-e.g., registrant submitted chemical
specific data, ORETF data, literature studies, etc-and do not rely on default
assumptions.
Two types of input distributions were used in the residential assessment.
A uniform distribution is one in which each value within the range specified has
an equal probability of being selected. Therefore, it does not reflect what the
actual shape of the distribution may be. A log-normal distribution approximates
the expected shape of the data distribution, with low values having a higher
probability of selection because there are more low values in a log-normal
distribution.
a. Lawn Scenario (DDVP, Bensulide, Malathion, Trichlorfon)
Application:
Unit Exposure for Granular Applications:
Dermal: 0.02-7.6 mg/lb ai handled (uniform distribution)
Inhalation: 0.00019-0.0096 mg/lb ai handled (uniform distribution)
Unit Exposure for Hose-end Sprayer Applications:
Revised Draft
83
-------
Dermal: 0.017-49 mg/lb ai handled (uniform distribution)
Inhalation: 0.007-0.089 mg/lb ai handled (uniform distribution)
Each of these distributions reflects a range of clothing from short-pants and short
sleeved shirts to long pants and long sleeved shirts. ORETF data showed that
55% of those who treat their lawns wear short sleeved shirts and 38% wear short
pants when applying liquid formulations while70% wore short sleeved shirts and
32% wore short pants while applying granulars. The distributions for the hose-
end sprayers also reflect the range derived from study data that included a ready
to use hose-end sprayer and a sprayer that required pouring pesticide into the
hose-end device.
Lawn Size:
500-15,000 ft2 (Uniform Distribution)
Post-Application
Dermal:
Adult transfer coefficient: 1,930-13,200 cm2/hr (uniform distribution)
Child transfer coefficient: 700-16,000 cm2/hr (uniform distribution)
Oral (for hand-to-mouth activity if children):
# of mouthing events: 0-26 (uniform distribution)
Surface area of hand associated with each mouthing event: 0-20 cm2/event
(uniform distribution)
Adjustment for greater residue transfer on wet hands: 1.5-3X (uniform
distribution)
Removal efficiency of residues on hands by saliva: 10-50% (uniform distribution)
The last two adjustments are applied to the residue data to account for the
expected greater residues that are picked up on wet hands and the expected
greater efficiency of removal of those residues in the mouth by saliva.
Time Spent on Lawn:
Adult: 0-2 hours (cumulative distribution)
Child: 0-3.5 hours (cumulative distribution)
b. Vegetable Gardens/Orchards/Ornamentals (acephate, disultofon,
malathion)
Application:
Unit Exposure for Hand Pump Sprayer:
Dermal: 7.99-354.4 mg/lb ai handled (uniform distribution)
Inhalation: 0.002-0.0142 mg/lb ai handled (uniform distribution)
Unit Exposure for Hand Garden Duster:
Dermal: 7.99-1375.4 mg/lb ai handled (uniform distribution)
Inhalation: 0.0044-8.29 mg/lb ai handled (uniform distribution)
Unit Exposure for Ornamental Granular Incorporated Treatment:
Revised Draft
84
-------
Dermal: 0.0034-0.356 mg/lb ai handled (uniform distribution)
Inhalation: 0.00001 mg/lb ai handled (point value)
All of the above distributions reflect a range of clothing from short-pants and
short sleeved shirts to long pants and long sleeved shirts, with the exception of
the granular treatment which is a point estimate based on the Limit of
Quantitation from the study.
Ornamental Bed Size:
500-2,000 ft2 (Uniform Distribution)
Vegetable Garden/Orchard Size:
135-8,000 ft2 (Log-normal Distribution)
Post-Application
Dermal:
Transfer coefficient: 100-5,000 cm2/hr (uniform distribution)
Time Spent in Garden:
0.083-1 hour (Uniform Distribution)
c. Golf Course (acephate, bensulide, fenamiphos, malathion, trichlorfon)
Post-Application
Dermal:
Transfer coefficient: 200-760 cm2/hr (uniform distribution)
Time Spent Golfing:
4 hours (point estimate for all chemicals except bensulide)
2-4 hours (uniform distribution-used for bensulide because its use is restricted to
tees and greens)
d. Public Health (malathion, naled, fenthion)
Post-Application
% of Application Deposited on Lawns:
3.8 - 30% (uniform distribution)
This distribution combines ground and aerial applications for which data show a
deposition range for ground from 3.8 to approximately 5% and for air values that
range from approximately 15-30%.
Estimates of lawn residues were based on the chemical specific transfer
efficiency of malathion (up to 2.2%) and naled (up to 1.5%). The Malathion
estimate was used for fenthion since the two chemicals have very similar
formulations, vapor pressures and molecular weights.
Exposures to residues on lawns were estimated using the same dermal transfer
coefficients, hand to mouth variables, and duration of time spent on the lawn as
shown above for lawn uses.
Revised Draft
85
-------
e. Indoor
The only indoor use chemical is DDVP. The only relevant route of exposure for
DDVP is inhalation due to its volatility. Exposure while handling the impregnated
pest strips is considered minimal and was not included.
Application:
Unit Exposure for Crack and Crevice Applications:
Inhalation: 0.72-2.499 mg/lb ai handled (uniform distribution)
Post-Application
Inhalation:
Post-application airborne concentration from crack and crevice treatments:
0.0754-0.548 mg/m3 (uniform distribution)
Post-application airborne concentration from pest strips: 0.11-0.005 mg/m3
(samples taken at 1, 7, 14, 28, 56,and 91 day intervals-uniform distribution for
each sample period)
Breathing Rate Multiplier:
1 for at rest; 2 for moderate activity (uniform distribution)
Breathing rates were taken from the exposure factors handbook and the
multipliers reflect the fact that people were assumed to be at rest half of the time
and engaged in moderate activity the other half of the time.
Time Spent in Home:
0-24 hours (Cumulative Distribution)
E. Individual Versus Cumulative Assessment
In general, the individual chemical assessments are designed to reflect
reasonable high-end risks to an individual/population member represented in
each exposure scenario (e.g., adults applying pesticides to a lawn with push-type
spreader, children playing on treated lawns). The cumulative risk analysis
focuses not on risk to the individual, but population risk (see discussion in
section V. "Cumulative Exposure Model and Interpretation of Model Outputs").
To estimate these risks, It combines many data sets into a single assessment.
As a result it is important to reduce the likelihood of compounding conservative
assumptions and over-estimation bias. Therefore, the assessment is not based
on high-end risk estimates but on estimates of potential exposure that
appropriately bound the risks while realistically capturing possible multiple
exposures.
F. Results
The results of the residential portion of the cumulative risk assessment are
relatively straight-forward to interpret. Inhalation exposures to DDVP from No-
Pest strips and crack and crevice treatments are the major contributors to indoor
residential exposures. This determination is simple to make because these are
Revised Draft
86
-------
the only remaining indoor uses for OPs. Some of the regional assessments from
the southern regions also indicate hand-to-mouth activities by children in
conjunction with lawn scenarios are a contributor to exposure. In examining
these potential risks after the release of the preliminary assessment the Agency
found an error in the computer input. Correction of this error resulted in
estimated risks that do not appear to be significant for hand-to-mouth activities by
children in conjunction with lawn scenarios.
Revised Draft
87
-------
VIII. Cumulative Risk and Risk Characterization
Risk characterization is the interpretation phase of the assessment
process. Appropriate interpretation of results is especially important for an assessment
as complex as the OP cumulative assessment. EPA has combined many types of data,
derived from a variety of sources, to produce detailed estimates of risk from multiple
OPs in food, drinking water, or use in residential areas. The outputs of the assessment
should be evaluated in a variety of ways. The risk characterization identifies potential
biases in input parameters, the direction of the bias, and the uncertainly surrounding the
inputs and the exposure model with regard to their potential impact on the results of the
assessment. An entire section of the preliminary OP cumulative risk assessment is
devoted to risk characterization. It can be found in Section I part G.
In summary, the results of the OP cumulative assessment indicate that the
contribution to OP cumulative risk from drinking water is generally at least 10 times
lower (one order of magnitude) than the contribution from OPs in food at percentiles of
exposure above the 95th for all population subgroups evaluated. As the percentile of
exposure increases, the difference between the food and water contributions increases.
This pattern is consistent for all regions. Those regions with the lowest total MOEs
(highest risk estimates) at the upper percentiles in the exposure distribution generally
reflect the contribution of the inhalation route of exposure from residential indoor uses of
DDVP. The exposures occur from the No Pest Strips and crack and crevice treatments.
This observation is consistent for all regions evaluated. The same pattern of risk from
each pathway is observed for all regions. At these higher percentiles of population
exposure, residential uses are a major source of risk-specifically, inhalation exposure
by all age groups. These patterns occur in all sub-groups, although estimated risks
appear to be higher for children than for adults regardless of the population percentile
considered. EPA believes that the results of the assessment provide a highly refined,
health protective estimate of the cumulative risk to the U.S. public from the use of OPs.
^^^^ccupationa^mc^cologicg^
Cumulative occupational and ecological risk assessments are not required by
FQPA and have not been conducted. Occupational and ecological risks were
addressed in the individual risk assessments for the OPs.
Revised Draft
88
-------
X. Summary of Future Work
The preliminary OP cumulative risk assessment provides a detailed picture of
possible exposure to oprganophosphorus pesticides. Details in the assessment provide
the basis to evaluate the effects of the methods and assumptions on the results of the
assessment. This evaluation process is particularly important for a cumulative OP
assessment because it reflects additional data compared to previous, single-chemical
assessments. It uses distributions of data in place of point estimates to the extent
possible, and introduces new data sources, particularly in the residential portion of the
assessment. EPA has used the OP cumulative risk assessment as a vehicle to
introduce a number of advances in its risk assessment methodology. These changes
are most evident in the hazard, drinking water and residential components, as well as in
the methods used to combine pathways to develop a total risk profile for all of the OPs
together. Therefore, EPA plans to conduct additional analyses of the data before
reaching final conclusions. At this point in the planning process, EPA in cooperation
with USDA has developed a set of follow-up analyses that will be conducted to assist in
interpreting the results of the preliminary analysis, and to prepare an OP cumulative risk
assessment for making regulatory decisions. Some examples of planned analyses are:
~ Conduct a series of sensitivity analyses for input parameters to determine
those most likely to impact the outcome of the assessment
~ Conduct detailed analysis of food exposure to identify major contributors
to risk, identifying specific food-pesticide combinations
~ Evaluate the tails of the food distribution to determine whether isolated
data points reflecting unusual consumption patterns or residue levels are
inappropriately affecting the results of the assessment
~ Evaluate the effect of assumptions about residue concentrations in baby
foods in the assessment.
~ Verify residential use patterns for OPs
~ Define the data that are needed to better characterize the toxicity of OP
degradates and treatment by-products in water systems. Evaluate and
summarize existing data
Revised Draft
89
-------
List of Abbreviations
a.i.
AGDCI
BMD
BMR
CAF
CAG
CEL
CL
CMG
CNS
CWS
CSF
CFR
CSFII
DCI
DEEM
DFR
DWLOC
EC
ED10
EEC
EP
EPA
FDA
FIFRA
FFDCA
FQPA Food
G
GIS
GLC
GLN
GM
GOF
GRAS
HDT
HED
ILSI
idose
IR
Active Ingredient
Agricultural Data Call-In
Benchmark Dose
Benchmark Response
Cumulative Adjustment Factor
Cumulative Assessment Group (of chemicals)
Comparative Effect Level
Confidence Limit
Common Mechanism Group (of chemicals)
Central Nervous System
Community Water Systems
Confidential Statement of Formula
Code of Federal Regulations
Continuing Surveys for Food Intake by Individuals (from USDA)
Data Call-In
Dietary Exposure Evaluation Model
Dislodgeable Foliar Residue
Drinking Water Level of Comparison.
Emulsifiable Concentrate Formulation
Effective Dose: central estimate on a dose associated with a 10%
response adjusted for background.
Estimated Environmental Concentration-The estimated pesticide
concentration in an environment, such as a terrestrial ecosystem.
End-Use Product
U.S. Environmental Protection Agency
Food and Drug Administration
Federal Insecticide, Fungicide, and Rodenticide Act
Federal Food, Drug, and Cosmetic Act
Quality Protection Act
Granular Formulation
Geographical Information System
Gas Liquid Chromatography
Guideline Number
Geometric Mean
Model Goodness-of-Fit
Generally Recognized as Safe as Designated by FDA
Highest Dose Tested
Health Effects Division
International Life Sciences Institute
Scaled internal dose
Index Reservoir
Revised Draft
-------
LCO
Lawn Care Operator
LED10
Lower Limit on an Effective Dose (95% lower confidence limit on a dose
associated with 10% response adjusted for background)
LEL
Lowest Effect Level
LOC
Level of Concern
LOD
Limit of Detection
LOAEL
Lowest Observed Adverse Effect Level
MCLG
Maximum Contaminant Level Goal-used by the Agency to regulate
contaminants in drinking water under the Safe Drinking Water Act
mg/kg/day
Milligram Per Kilogram Per Day
mg/L
Milligrams Per Liter
MOE
Margin of Exposure
MP
Manufacturing-Use Product
MPI
Maximum Permissible Intake
MRID
Master Record Identification (number)-EPA's system of recording and
tracking studies submitted.
N/A
Not Applicable
NAS
National Academy of Sciences
NAWQA
USGS National Water Quality Assessment
NHEERL
National Health and Environmental Effects Laboratory
nlme
Non-linear mixed effects model
NOEC
No Observable Effect Concentration
NOAEL
No Observed Adverse Effect Level
NPDES
National Pollutant Discharge Elimination System
NR
Not Required
NRC
National Research Council
OP
Organophosphate
OPCumRisk
Organophosphate Cumulative Risk (computer program)
OPP
Office of Pesticide Programs
OPPTS
Office of Prevention, Pesticides and Toxic Substances
ORETF
Occupational and Residential Exposure Task Force
PAD I
Provisional Acceptable Daily Intake
PAG
Pesticide Assessment Guideline
PBPK
Physiologically Based Pharmacokinetics
PAM
Pesticide Analytical Method
PCA
Percent Crop Area
PCO
Pest Control Operator
PDP
Pesticide Data Program (USDA)
PHED
Pesticide Handler's Exposure Database
POD
Point of Departure
PPb
Parts Per Billion
PPm
Parts Per Million
PRN
Pesticide Registration Notice
PRZM /
Pesticide Root Zone Model/EXposure Analysis Model System-Coupled
EXAMS
models used to estimate pesticide concentrations in surface water.
RAC
Raw Agriculture Commodity
RBC
Red Blood Cell
RED
Reregistration Eligibility Decision
Revised Draft
91
-------
RfD
Reference Dose
RPF
Relative Potency Factor
RUP
Restricted Use Pesticide
SAP
FIFRA Scientific Advisory Panel
SCI-GROW
Tier 1 Ground Water Computer Model
SF
Safety Factor
SLN
Special Local Need (Registrations Under Section 24© of FIFRA)
SOP
Standard Operating Procedures
TC
Toxic Concentration-The concentration at which a substance produces
toxic effect.
TD
Toxic Dose-The dose at which a substance produces a toxic effect.
TEP
Typical End-Use Product
TGAI
Technical Grade Active Ingredient
TLC
Thin Layer Chromatography
UF
Uncertainty Factor
Mg/g
Micrograms Per Gram
Mg/L
Micrograms Per Liter
USDA
United States Department of Agriculture
USGS
United States Geological Survey
UV
Ultraviolet
WARP
Water Analysis Regression Program
WHO
World Health Organization
WP
Wettable Powder
Revised Draft
-------
Glossary of Terms
Absorbed Dose: The amount of a substance penetrating across the absorption
barriers (the exchange barriers) of an organism, via either physical or biological
processes. Synonymous with internal dose.
Additivity: When the "effect" of a combination of chemicals is estimated by the sum of
the exposure levels or the effects of the individual chemicals.
Aggregate Dose: The amount of a single substance available for interaction with
metabolic processes or biologically significant receptors from multiple routes of
exposure.
Aggregate Exposure: The amount of a chemical available at the biological exchange
boundaries (e.g., respiratory tract, gastrointestinal tract, skin) for all routes of exposure.
Aggregate Exposure Assessment: A process for developing an estimate of the
extent of a defined population to a given chemical by all relevant routes and from all
relevant sources.
Aggregate Risk: The risk associated with all pathways & routes of exposure to a
single chemical.
Analog(s): Analog is a generic term used to describe substances that are chemically
closely related. Structural analogs are substances that have similar or nearly identical
molecular structures. Structural analogs may or may not have similar or identical
biological processes.
Antagonism: The ability of a substance to prevent or interfere with another substance
interacting with its biological targets, thereby reducing or preventing its toxicity.
Benchmark Dose (BMDL): A statistical lower confidence limit on the dose producing a
predetermined level of change in adverse response compared with background
response. The BMD is derived by fitting a mathematical model to the dose-response
data. A BMD10 is a benchmark dose with 10% change in adverse response compared
with background response.
Benchmark Response(BMR): A designated level or percent of response relative to
the control level of response used in calculating a BMD.
Biomonitoring: Measurement of a pesticide or its metabolites in body fluids of
exposed persons, and conversion to an equivalent absorbed dose of the pesticide
based on a knowledge of its human metabolism and pharmacokinetics.
Revised Draft
93
-------
Common Mechanism of Toxicity: Common mechanism of toxicity pertains to two or
more pesticide chemicals or other substances that cause a common toxic effect(s) by
the same, or essentially the same, sequence of major biochemical events (i.e.,
interpreted as mode of action). Hence, the underlying basis of the toxicity is the same,
or essentially the same, for each chemical.
Common Mechanism Group (CMG): A group of pesticides determined to cause
adverse effects by a common mechanism of toxicity. The CMG is defined using the
previously released "Guidance for Identifying Pesticide Chemicals and Other
Substances that Have a Common Mechanism of Toxicity" (February 5, 1999). Not all
members of a CMG will necessarily be incorporated in the cumulative risk assessment.
Common Toxic Effect: A pesticide and another substance that are known to cause
the same toxic effect in or at the same anatomical or physiological site or locus (e.g.,
the same organ or tissue) are said to cause a common toxic effect. Thus, a toxic effect
observed in studies involving animals or humans exposed to a pesticide chemical is
considered common with a toxic effect caused by another chemical if there is
concordance with both site and nature of the effect.
Comparative Effect Level (CEL): A dose by which potency of chemicals may be
compared; e.g. the dose causing a maximum of 15% cholinesterase inhibition.
Concurrent Exposure: The potential human exposure by all relevant pathways &
routes that allows one chemical to add to the exposure of another chemical such that
the total risk of a group of common mechanism chemicals is an estimate of the sum of
the exposures to the individual chemicals. The accumulation of the common toxic effect
may or may not depend on simultaneous or overlapping exposures depending on the
duration and recovery time for the toxic effect.
Cumulative Adjustment Factor (CAF): Accounts for the percent of land in a given
location that is planted to crops and treated with a given OP.
Cumulative Assessment Group (CAG): A subset of the CMG. The CAG is that group
of pesticides selected for inclusion in the cumulative risk assessment. The chemicals in
the CAG are judged to have a hazard and exposure potential that could result in the
expression of a cumulative risk.
Cumulative Dose: The amount of multiple (two or more) substances which share a
common mechanism of toxicity available for interaction with biological targets from
multiple routes of exposure.
Cumulative Exposure Assessment: A process for developing an estimate of the
extent to which a defined population is exposed to two or more chemicals which share
a common mechanism of toxicity by all relevant routes and from all relevant sources.
Revised Draft
94
-------
Cumulative Toxicity or Toxic Effect: A cumulative toxic effect(s) is the net change in
magnitude of a common toxic effect(s) resulting from exposure to two or more
substances that cause the common toxic effect(s) from a common mechanism, relative
to the magnitude of the common toxic effect(s) caused by exposure to any of the
substances individually.
Cumulative Risk: For the purpose of implementation of FFDCA as amended by
FQPA, cumulative risk is the likelihood for the cumulation of a common toxic effect
resulting from all pathways and routes of exposure to substances sharing a common
mechanism of toxicity.
Dependent (events): The probability of one event occurring is affected by whether or
not another event has or has not occurred.
Deterministic: This approach to risk assessment uses point estimates, for example,
single maximum values or average values, to represent input variables in an exposure
model. This can be compared to a probabilistic approach which considers the full range
of potential exposures incurred by members of a population.
Dislodgeable Residues: The portion of a pesticide (which may or may not include its
metabolites) that is available for transfer from a pesticide treated surface.
Dose: The amount of substance available for interaction with metabolic processes or
biologically-significant receptors after crossing the outer boundary of an organism.
Dose Rate: Dose per unit time (e.g., mg/day). Also called dosage. Dose rates are
often expressed on a per-unit-body-weight basis (mg/kg/day). Dose rates may also be
expressed as an average over a time period (i.e., lifetime).
Dose Additivity: The Agency's assumption when evaluating the joint risk of chemicals
that are toxicologically similar and act at the same target site. In other words, it is
assumed that each chemical behaves as a concentration or dilution of every other
chemical in the CAG (or chemical mixture). The response of the combination is the
response expected from the equivalent dose of an index chemical. The equivalent dose
is the sum of component doses scaled by their potency relative to the index chemical.
Effective Dose (ED): The effective dose is a measured or estimated dose level
associated with some designated level or percent of response relative to the control or
baseline level of response. For example, the ED10 is a dose associated with a 10%
response. The effective does is essentially the same as a benchmark dose (BMD). It is
determined by using a curve-fitting procedure that is applied to the dose-response data
for a chemical.
Exposure: Contact of a substance with the outer boundary of an organism. Exposure
is quantified as the concentration of the agent in the medium in contact integrated over
the time duration of that contact.
Revised Draft
95
-------
Exposure Assessment: The qualitative or quantitative determination or estimation of
the magnitude, frequency, duration, and rate of exposure of an individual or population
to a chemical.
Exposure Scenario: A combination of facts, assumptions, and inferences that define a
discrete situation or activity where potential exposures may occur.
Independent (events): The probability of one event occurring is not affected by
whether or not another event has or has not occurred.
Index Chemical: The chemical selected as the basis for standardization of toxicity of
components in a CAG (or a mixture). The index chemical should have a clearly defined
dose-response relationship.
LED10 : The lower confidence limit on an effective dose, that is, in this case the 95%
lower confidence limit on a dose associated with 10% response adjusted for
background.
Level of Comparison: A drinking water level of comparison is a theoretical upper limit
on a pesticide's concentration in drinking water in light of total aggregate exposure to a
pesticide in food, drinking water, and through residential uses.
Lowest Observed Adverse Effect Level (LOAEL): The lowest dose in a toxicity study
resulting in adverse health effects.
Margin of Exposure: The point of departure divided by a human environmental
exposure(s) of interest, actual or hypothetical.
Mechanism of Toxicity: Mechanism of toxicity is defined as the major steps leading to
an adverse health effect following interaction of a substance with biological sites. All
steps leading to an effect do not need to be specifically understood. Rather, it is the
identification of the crucial events following chemical interaction that are required in
being able to describe a mechanism of toxicity.
Monte Carlo Analysis: One of several mathematical techniques for performing
probabilistic assessments. The method relies on the computational powers of modern
computers to simulate the range and frequency of all possible outcomes of a process
based on repeatedly sampling from the inputs provided by the user. These inputs are
combined according to the model that is specified by the user.
No Observed Adverse Effect Level (NOAEL): The highest dose in a toxicity study
which does not result in adverse health effects.
Pathway of Exposure: The physical course a pesticide takes from the source to the
organism exposed (e.g., through food or drinking water consumption or residential
pesticide uses).
Revised Draft
96
-------
Point of Departure (POD): Point on the dose-response curve where each chemical's
response is close to or within the background level of response, in other words, the
dose at which effects from a pesticide are first distinguishable. Depending on the kind
of data available and the purpose of the analysis, there are differing procedures for
estimating the point of departure.
Reference Dose (RfD): NOAEL/UF.
Relative Potency Factor (RPF): The ratio of the toxic potency of a given chemical to
that of an index chemical in the CAG. Relative potency factors are used to convert
exposures of all chemicals in the CAG into their exposure equivalents of the index
chemical.
Relative Potency Factor (RPF) Method: The RPF approach expresses the potency of
each chemical in a CAG in relation to the potency of another member in the group
which has been selected as the index chemical. A relative potency factor is calculated
for each chemical for each route of exposure (e.g., oral, dermal, inhalation). For
example, if compound A is determined to be one-tenth as toxic as the index compound
the RPF for compound A is 0.1. Using this approach, for each route of exposure for
each chemical, exposure is expressed as exposure equivalents of the index chemical.
The exposure equivalents are calculated by multiplying the residues and the RPF for
each route. These exposure equivalents are summed to obtain an estimate of total
exposure by route in terms of the index chemical.
Route of Exposure: The way a chemical enters an organism after contact, e.g.,
ingestion, inhalation, or dermal absorption. Note that all three routes of exposure can
occur within an exposure pathway. A pathway is not route specific.
Site of Toxic Action: The physiological site(s) where a substance interacts with its
biological target(s) leading to a toxic effect(s).
Steady State Inhibition: The time point at which continued dosing at the same level
results in no further increase in cholinesterase inhibition.
Structure-Activity Relationships: Substances that contain or are bioactivated to the
same toxophore may cause a common toxic effect by a common mechanism. The
relative toxic efficacy and potency among the substances in their ability to cause the
toxic effect may vary substantially. Differences in potency or efficacy are directly related
to the specific or incremental structural differences between the substances and the
influence these differences have on the ability of the toxophore to reach and interact
with its biomolecular site of action, and on the intrinsic abilities of the substances to
cause the effect. The ability of two or more structurally-related substances to cause a
common toxic effect and the influence that their structural differences have on toxic
efficacy and potency are referred to as structure-activity relationships.
Surrogate Data: Substitute data or measurements on one substance (or population)
used to estimate analogous or corresponding values for another substance (or
population).
Revised Draft
97
-------
Toxic Action: The interaction with biological targets that leads to a toxic effect.
Toxic Effect: An effect known (or reasonably expected) to occur in humans that results
from exposure to a chemical substance and that will or can reasonably be expected to
endanger or adversely affect quality of life.
Toxic Endpoint: A quantitative expression of a toxic effect occurring at a given level of
exposure. For example, acute lethality is a toxic effect, an LD50 value (median lethal
dose) is the toxic endpoint that pertains to the effect.
Toxic Potency: The magnitude of the toxic effect that results from a given exposure.
Relative potency refers to comparisons of individual potencies of chemicals in causing a
common toxic effect at the same magnitude (e.g, LD50, ED50) by a common mechanism.
Transfer Coefficient: Residue transfer rate to humans during the completion of
specific activities (e.g., cm2 per hour), calculated using concurrently collected
environmental residue data.
Uncertainty: Lack of knowledge about specific factors, parameters, or models.
Uncertainty Factor: Uncertainty factors applied to account inter- and intra-species
differences in relation to toxic effects, and uncertainties associated with the data.
Unit Exposure: The amount of a pesticide residue's to which individuals are exposed,
normalized by the amount of active ingredient used.
Variability: Differences attributed to true heterogeneity or diversity in a population or
exposure parameter.
Weight-of-the-Evidence: Weight-of-the-evidence refers to a qualitative scientific
evaluation of a chemical substance for a specific purpose. A weight of evidence
evaluation involves a detailed analyses of several or more data elements, such as data
from different toxicity tests, pharmacokinetic data, and chemistry data followed by a
conclusion in which a hypotheses is developed, or selected from previous hypotheses.
Revised Draft
98
------- |