  UNITED STATES ENVIRONMENTAL PROTECTION AGENCY

WASHINGTON, D.C.  20460

                                                                        
                                                                        
                                            

                                                                        
                                                                        
                                                              OFFICE OF 
   

 PREVENTION, PESTICIDES

AND TOXIC SUBSTANCES

MEMORANDUM

DATE:  	November 1st, 2006

SUBJECT:	Aldicarb: Acute Dietary Exposure Assessment to Support the
Reregistration Eligibility Decision  

PC Code: 098301		DP Number: D299889

FROM:	Steven M. Nako, Statistician

Chemistry and Exposure Branch

OPP\Health Effects Division (7509P)

Jianping Xue, Research Physical Scientist

Exposure Modeling Research Branch 

ORD\NERL\Human Exposure and Atmospheric Sciences Division (E205-D2) 

THROUGH:	David J. Miller, Chief

Chemistry and Exposure Branch 

Health Effects Division (7509P)

TO:		Felecia Fort, Aldicarb Risk Assessor

		Reregistration Branch 1

OPP\Health Effects Division (7509P)

The Special Review and Reregistration Division (SRRD) requested that HED
 assess dietary risks to aldicarb to support its Reregistration
Eligibility Decision.  The acute adverse effect of cholinesterase
inhibition tends to reverse itself within hours following exposure to
aldicarb.  The available toxicological data indicates that aldicarb has
an estimated half-life for RBC cholinesterase inhibition of two hours
based on data from rats and human subjects.  Since the food diaries used
by Dietary Exposure Evaluation Model-Food Consumption Intake Database
(DEEM-FCID Version 2.03) are based on total daily intake, the estimated
risks produced by this software will overestimate risks to the extent
that foods and drinking water are consumed throughout the day, rather
than during only one event.  To provide a better approximation of the
potential exposure leading to peak RBC ChE inhibition, potential
exposure from food and/or water to aldicarb was computed incrementally
throughout the day.  This computation was made by incorporating
information on the time of day and amounts consumed during each occasion
from the USDA CSFII food diaries.  The potential for accumulation of
toxicity was accounted for by computing the degree to which exposures
could be discounted between exposure occasions, assuming a two-hour
half-life.  

This document is divided into five sections.  Section I introduces
issues associated with the kinetics of aldicarb and recovery of AChE
inhibition.  Section II discusses the dietary inputs used in the
modeling exercises and covers the anticipated residues in food and
predicted drinking water concentrations used in the assessment.  Section
III highlights the approach and method for calculating exposures on an
eating/drinking occasion basis using the DEEM and SHEDS models.  
Section IV summarizes some exploratory analyses of drinking water
consumption patterns.  Bayer CropScience sponsored a Drinking Water
Consumption Survey (DWCS), collecting 7 day diaries from over 4,000
participants.  These data were used to conduct alternative dietary
exposure analyses using the SHEDS model, in which the DWCS diaries were
used to empirically allocate direct drinking water consumption
throughout the day. Finally, Section V provides a brief summary of our
conclusions and provides characterization of the risk.

I.  Toxicological Background and Recovery Half-Life

Aldicarb is a member of the N-methyl carbamate (NMC) pesticides common
mechanism group. Like other NMCs, aldicarb inhibits acetylcholinesterase
(AChE) by carbamylation of the serine hydroxyl group located in the
active site of the enzyme.  NMC toxicity is characterized by maximal
inhibition of cholinesterase which occurs rapidly followed by recovery
typically occurring within hours.   A key consideration in risk
assessment is appropriate matching of the duration of exposure with the
duration of the toxic effect.  Typically, HED’s food and water
exposure assessments sum exposures over a 24 hour period.  This 24 hour
total is typically used in acute dietary risk assessment.  In the case
of the aldicarb, because of the rapid nature of aldicarb toxicity and
recovery, it may be appropriate to consider durations of exposure less
than 24 hours.  Conceptually, a physiologically-based pharmacokinetic
model and/or biologically-based dose-response model would be available
to account for the dynamic nature of exposure, absorption, toxicity,
recovery, and elimination of aldicarb in animals and humans.  However,
such as model does not exist at this time.  In the interim, HED has
developed an analysis using information about external exposure, timing
of exposure within a day, and half-life of ChE inhibition from rats and
humans to estimate risk to aldicarb at durations less than 24 hours. 
Specifically, HED has evaluated individual eating and drinking occasions
and used the ChE half-life information to estimate the residual effects
from aldicarb from previous exposures within the day.  

Table 1 below provides information on the recovery of ChE inhibition in
rats and human subjects.  For both species, the recovery half-life for
RBC ChE inhibition is approximately two hours.  At high doses in rat,
the half-life is up to approximately 6 hours in females.  The estimates
of half-life at the lower doses are most relevant for risk assessment
and are thus the focus here.  As can be seen in the table, the estimated
recovery half life of for aldicarb-inhibited AChE in the human is
estimated to be on the order of 2 hours using RBC AChE activity   This 2
hour recovery half-life is what is used in this refined dietary exposure
assessment which incorporates information on eating/drinking occasions. 
There is some uncertainty associated with the use of the two hour
recovery half-life.  As discussed in detail below, infants and children
are the focus of the current analysis.  Although there are dose-response
ChE data in juvenile animals exposed to aldicarb, there are no such data
to characterize ChE recovery in the young.  As such, the Agency has
assumed that the half-life to recovery in the young is similar to that
seen in adults.  The Agency is requiring such data in young animals to
confirm this assumption.

 

Table 1.  Recovery half-life information for ChE inhibition following
oral exposure to aldicarb in rats and human subjects

Chemical	Brain	RBC

	Recovery Half-Life Estimate (hrs)	Upper & Lower Confident Intervals
(hrs)	Recovery Half-Life Estimate (hrs)	Upper & Lower Confident
Intervals (hrs)

Rat	1.52	1.16-1.99	F (-inf, 0.1)  1.10 

(0.1.0.3)  2.91

(0.3,0.5) 3.39

(0.5, Inf) 5.90

M (-inf,0.1) 1.91

(0.1,0.3) 1.20

(0.3,0.5) 1.62

(0.5, Inf) 1.50	F  0.50-2.40

1.96-4.33

2.35-4.90

3.52-9.91

M  1.31-2.79

0.87-1.64

1.19-2.21

0.80-2.82

Human	N/A	2.07	1.74-2.46



II.	Dietary Inputs: Anticipated Residues

a.  Anticipated Residues-Food 

Table 2 presents the dietary inputs that were used in both the
DEEM-based eating occasion and SHEDS simulations.  These anticipated
residues are based on the most updated food residues, processing
factors, percent crop treated estimates, and predicted drinking water
concentrations.  These data are presented and described in detail in the
Aldicarb Dietary Risk Assessment memo, Fort (2006).  Following Fort
(2006), both food and drinking water concentrations model inputs are
expressed in aldicarb sulfone equivalents.  The results from the
probabilistic risk assessment models (DEEM and SHEDS) were then
converted into aldicarb (parent) equivalents (by multiplying 0.86), and
these adjusted exposures are used to calculate risk, based on the acute
population adjusted dose (aPAD), which is expressed in aldicarb parent
equivalents.  



Table 2. Anticipated Residues Used in Eating Occasion Analyses

Commodity	Source/Notes	Total Samples	Est. Max PCT	Total Residues	Range
(ppm)

Grapefruit	Carbamate MBS  (NB, Fresh)	213	25%	53	0.00147-0.02906

Grapefruit	Carbamate MBS  (PB, Proc.)	162	33%	53	0.00147-0.02906

Lemon	Carbamate MBS	1778	3%	53	0.00147-0.02906

Lime	Carbamate MBS	762	7%	53	0.00147-0.02906

Orange	Carbamate MBS (NB, Fresh)	399	13%	52	0.00147-0.02906

Orange	Carbamate MBS  (PB, Proc.)	399	23%	92	0.00147-0.02906

Pecan	Field Trial	275	8%	22	0.005-0.27

Potato	PDP (NB, Fresh) 	3200	5%	160	0.00758-0.40232

Potato	PDP (PB, Proc.)	1425	24%	342	0.00758-0.17292

Sweet Potato	PDP (NB, Fresh) 	432	37%	160	0.00758-0.40232

Sweet Potato	PDP (PB, Proc.)	1755	37%	650	0.00801-0.11825



b. Predicted Drinking Water Concentrations 

Table 3 presents the drinking water inputs were used in the
eating/drinking occasion analyses.  Fort (2006) provides further
description of these scenarios.  

Table 3. Modeled Drinking Water Scenarios (aldicarb equiv.)

Filename	Notes

Aldicarb_GACoastalGW_300.csv	GA 300ft setback

Aldicarb_GACoastalGW_500.csv	GA 500ft setback

Aldicarb_GACoastalGW_1000.csv	GA 1000ft setback

Aldicarb_GW_FLCit30.csv	FL 1000ft setback

Aldicarb_NCCoastalGW_300.csv	NC 300ft setback

Predicted Drinking Water Concentrations

Pctile	DW Concentration (ppb)

	GA 300ft	GA 500ft	GA 1000ft	FL 1000ft	NC 300ft

10%	2.4	1.4	0.3	1.1	0.2

25%	2.8	1.6	0.4	1.4	0.6

50%	3.2	1.9	0.5	1.9	0.7

75%	4.2	2.4	0.6	2.1	0.9

90%	4.8	2.8	0.7	2.5	1.0

80%	4.4	2.5	0.6	2.2	0.9

90%	4.8	2.8	0.7	2.5	1.0

95%	5.2	3.0	0.7	2.6	1.1

97.5%	5.5	3.1	0.8	2.7	1.1

99%	6.0	3.5	0.8	2.8	1.3

100%	6.5	3.7	0.9	3.0	1.3



III.	Method for Estimating Exposure Based Risks on Eating/Drinking 
Occasions

Baseline Analysis

The Agency began its analysis by conducting a baseline analysis using
its standard modeling practices and modeling software.   The DEEM-FCID
model has been used extensively by the Agency to conduct probabilistic
dietary risk assessments.  The overall concept has been reviewed by a
FIFRA Science Advisory Panel in 2000.  A general overview of the DEEM
model is provided in each dietary risk assessment, and is not reproduced
here. As noted in the DEEM reports, DEEM simulates dietary exposure by
randomly drawing a residue for each commodity-food form and multiplies
that by the total amount  of each commodity consumed during the day. 
These commodity-specific exposures are then summed to produce a total
daily exposure which is converted to a ug/kg bw-day basis.  Because
total food + water consumption amounts are considered on a total daily
basis -- and not on an eating/drinking occasion basis -- DEEM-FCID
cannot account for the recovery of AChE.  Specifically, DEEM sums all
exposures during the day and does not account for the fact that exposure
events occurring early in the day should be partially discounted when
they are summed with exposure events that occur later in the day.   

The Agency also conducted a similar baseline analysis using a version of
SHEDS (termed SHEDS-NMC)  that utilizes the two-day CSFII respondents as
its reference population, fixes the number of ‘iterations’ that each
diary is used in a simulation to the same frequency, and utilizes the
corresponding USDA CSFII sampling weights to calculate per capita
exposures and risks. This version of SHEDS also restricts the method for
drawing anticipated food residues in the Monte Carlo simulations to the
standard approach used by the other models (DEEM-FCID, Calendex,
Lifeline and CARES). In this way, SHEDS-NMC is structured to best
approximate the assumptions, data, and algorithms used in the standard
models currently used by OPP in its risk assessments. 

Table 4 compares the use of the CSFII data by DEEM-FCID and SHEDS-NMC
and highlights the similarities and differences.   Table 5 presents the
baseline figures from DEEM and SHEDS of total daily exposure (%aPAD) at
the per capita 99.9th percentile:  estimated exposures and associated
risks are similar between the two models and differ for all scenarios
and population subgroups by less than 10%.  As indicated before, these
estimates a based upon total daily exposure and do not consider the
effects of a finite recovery half-life.



Table 4 Comparison of SHEDS-NMC and DEEM-FCID

Variable	DEEM-FCID (2.2)	SHEDS-NMC

#Diaries Used (RefPop)

Food Only

Food+Water	(CSFII 2-Day)

41,214

40,476	(CSFII 2-Day)

         41,214

         41,214

Model Weights

(Per capita 99.9th)	CSFII 2-Day	CSFII 2-Day

Frequency

Used in MC simulations	User Specified	User Specified

Data Available

For Eating Occasion	Top 5%,

Max=40K records	All Simulated Records



b. DEEM-Based Eating Occasion Analyses

To the extent that the individual may have consumed those foods and
drinking water throughout the day, the timing and amounts of those
exposures on each of those eating occasions is not provided by the
DEEM-FCID model.  However, since some of this information is available
in the USDA CSFII food diaries, OPP used that data -- together with the
DEEM simulated outputs -- to obtain a DEEM-based estimate of dietary
exposure by eating/drinking occasion. OPP used the DEEM-FCID model which
estimates exposure on a total daily -- and not eating/drinking occasion
-- basis to refine its risk calculations to incorporate the effects of
reversibility of AChE over a two hour half life.  Specifically,
information on the per occasion timing and amounts from the CSFII survey
was combined with CEC output  from the DEEM-FCID model  in order to
produce estimates of risk which incorporate the 2 hour recovery
half-time for AChEI.    The specific steps and mechanics of these
calculations are highlighted in the Appendix. But briefly, the DEEM
program was run with the aldicarb food and drinking water residues, and
the (food commodity-based)  diaries  associated with the top 5% of
exposure values identified by the CEC were saved.   These files were
then merged with information from the CSFII food diaries (which contain
the foods consumed and the associated time of day) and the DEEM recipe
(aka 100 gram) files to produce a file which contained the times
associated with each eating occasion for the sample of the top 5% of
consumers identified in DEEM’s CEC.  This information was then used,
as described in more detail below, to account for AChEI recovery by
discounting early exposures using the 2 hour AChEI recovery half-life.  


Figures 1 through 3 depict these steps with actual DEEM output.  Figure
1 depicts outputs from the three different DEEM-FCID reports: (i)
Summary Table, (ii) Plot File, and (iii) Critical Exposure Commodity
(CEC) Analyses.  The summary table, depicted in Figure 1a, displays the
estimated exposure and risks (%aPAD) at the per capita 95th, 99th, and
99.9th percentiles.  This report also specifies the percent of all food
diaries that are ‘users’.  A food diary is considered a ‘user’
if one or more of the foods for which anticipated residues have been
assigned, including drinking water, was consumed.  In this example,
89.38% of all infant-diaries are ‘users’; that is, approximately 90%
of these infants consume any of these foods and/or drinking water
(Section IV provides further description on Drinking Water Consumption
patterns).  The Plot File presents the total number of diaries
(N=2,940), the total projected person-days (N=7,548,892), and the
projected person-days in each ‘exposure bin’ for all ‘simulated
users’   based on the number of iterations specified in the Monte
Carlo simulation (200 iterations).  The data in this plot file was used
to construct the projected per capita estimates for the entire
subpopulation, as depicted in Figure 1b.   

DEEM’s ‘CEC’ report provides a summary of exposure at the upper
percentile.  The first half of the CEC report provides shares of total
exposure by commodity; in this case, indirect water, food form=130
accounts for 63.52% of total exposures between the 95th and 100th
percentile.  This indirect drinking water is primarily infant formula,
with food form=130 (cooking status=uncooked, form=dried, cooking
method=not specified) referring to the powder component.  Other forms of
both direct and indirect drinking water, as well as foods, constitute
the remaining shares of total exposure at this upper percentile.  In
this case, the top 5% of simulated exposure diaries are saved in this
output file.  

The second part of the CEC report provides the foods consumed and
residues drawn for all simulated diaries at 95th through 100th
percentile.  Figure 1c presents a few selected simulated diaries; the
total number of diaries in this top 5 percentile is determined by the
total number of diaries in the subpopulation (N=2,940), the total number
of iterations (200 iterations), and the sampling weights for the
simulated diaries that tend to fall in this upper percentile.  In this
example, DEEM output 29,138 person-days (records), from this simulation.
 The DEEM CEC report has the following limitations: (i) a maximum of
40,000 records is output, (ii) the lower interval for which CEC focuses
upon is the 95th percentile (any range between 95th and 100th
percentile), (iii) foods contributing less than 1% of the simulated
daily exposure are not saved in the simulated output (lower half). It is
important to ensure that the number of actual (and printed) records do
not exceed the 40,000 limit, and that the two day CSFII sampling weights
are used to obtain an accurate DEEM-based estimate from this CEC output.
 If too many iterations are specified in the DEEM simulation, then DEEM
may print out more than this maximum limit, however, these records may
not provide a comprehensive, random set of the top 5 percent.  Agency
risk assessors typically specify 1,000 iterations when conducting
probabilistic risk assessments using DEEM, since the model is extremely
efficient and quick in conducting the Monte Carlo simulations.  However,
due to the limitations listed above, fewer iterations were specified
here to obtain a complete set of records for the top 5 percent. While
the number of iterations conducted here were significantly lower than
the typical 1000 iterations used in OPP risk assessments in order to
avoid exceeding the 40,000 record limit, the estimates were very similar
to baseline results found with 1000 iterations.

 

Table 5. Comparison of DEEM and SHEDS Baseline Risks

At the per capita  99.9th Percentile

Subpopulation	DEEM Total Daily Exposure/Risks 

(expressed as % aPAD)

	Food Only	Food + Drinking Water 





GA-GW 300 ft	GA-GW 500 ft	GA-GW 1000 ft	FL-SW 1000 ft

USPop	35%	119%	74%	39%	68%

All Infants	42%	285%	168%	53%	154%

Children 1-2 yrs	77%	145%	98%	80%	92%

Children 3-5 yrs	64%	135%	93%	66%	88%

Children 6-12 yrs	50%	87%	60%	49%	57%

Youth 13-19 yrs	30%	91%	58%	33%	52%

Adults 20-49 yrs	30%	94%	58%	32%	53%

Adults 50+ yrs	35%	72%	49%	35%	46%

Females 13-49 yrs	30%	92%	57%	32%	52%

Subpopulation	SHEDS Total Daily Exposure/Risks

	Food Only	Food + Drinking Water 





GA-GW 300 ft	GA-GW 500 ft	GA-GW 1000 ft	FL-SW 1000 ft

USPop	37%	116%	73%	39%	67%

All Infants	41%	278%	160%	53%	148%

Children 1-2 yrs	82%	144%	99%	84%	94%

Children 3-5 yrs	60%	131%	91%	62%	84%

Children 6-12 yrs	45%	84%	58%	47%	55%

Youth 13-19 yrs	32%	88%	58%	33%	52%

Adults 20-49 yrs	31%	91%	57%	32%	52%

Adults 50+ yrs	33%	70%	48%	35%	45%

Females 13-49 yrs	31%	89%	57%	32%	52%

Subpopulation	Ratio DEEM/SHEDS

	Food Only	GA-GW 300 ft	GA-GW 500 ft	GA-GW 1000 ft	FL-SW 1000 ft

USPop	0.96	1.03	1.02	1.00	1.02

All Infants	1.01	1.02	1.05	1.00	1.04

Children 1-2 yrs	0.94	1.01	0.99	0.95	0.98

Children 3-5 yrs	1.07	1.03	1.03	1.07	1.05

Children 6-12 yrs	1.09	1.04	1.03	1.03	1.04

Youth 13-19 yrs	0.92	1.02	0.99	0.99	1.00

Adults 20-49 yrs	0.98	1.03	1.01	1.00	1.02

Adults 50+ yrs	1.04	1.03	1.03	1.01	1.02

Females 13-49 yrs	0.97	1.02	1.00	0.97	1.00



The individual demographic information is provided (CSFII
Household-Person-Day identification) so that one can go back to the USDA
CSFII food diaries to link other information that is not used by the
DEEM model.  For the eating occasion analyses, information on the amount
and timing of all eating occasions was retrieved from the CSFII diaries,
and then merged with the output from the CEC report.  This process is
depicted in Figure 2.  Figure 2 illustrates for one individual how
eating occasion estimates were computed from the DEEM CEC output.  Data
on the timing and amounts of foods and indirect drinking water consumed
throughout the day are extracted from the CSFII food diaries, and merged
into the respective DEEM CEC diaries to obtain eating occasion
estimates.  The maximum cumulative exposure was computed for each eating
occasion for each simulated person-day diary in the DEEM CEC report (Top
5 percentile) Assumptions are required with respect to the timing and
amounts of direct (as opposed to indirect which are supplied by the
CSFII) drinking water consumption throughout the day since direct water
consumption data is not available in CSFII.  The Agency elected to
evenly allocate each individual’s reported total amount of direct
drinking water over the day to 6 fixed events spaced evenly throughout a
15 hour period: 360 minutes after midnight or 6 am, 9 am, 12 noon, 3 pm,
6 pm, and 9 pm (note that indirect drinking water per occasion timing
and amounts were reported in the CSFII diaries and no assumption
concerning these parameters were necessary). 

As depicted in Figure 2, the total daily exposure for this particular
simulated diary is 0.00201 mg/kg/day, or 266% of the aPAD, while the
maximum cumulative exposure with a two hour recovery half-life is
0.000475 mg/kg, or 73% of the aPAD under the eating/drinking occasion
approach.  This analysis is based on equally allocating the total amount
of direct drinking water over six fixed events occurring every 3 hours
from 6 am to 9 pm (timing and amount of indirect drinking water is
available through the CSFII diaries) Sensitivity analyses for two
alternate options for allocating direct drinking water consumption
throughout the day are presented in Section III of this document.  

The maximum cumulative exposure was computed for each eating occasion
for each simulated person-day diary in the DEEM CEC report (Top 5
percentile).  Figure 3a illustrates the total daily exposure values for
these top 5 percent of simulated diaries, together with the paired
eating occasion values.  Re-sorting the eating occasion values permits
the calculation of the 99.9th percentile for the DEEM-based eating
occasion analyses, as depicted in Figure 3b; the two distributions are
overlapped and plotted over the per capita percentiles.



Figure 1. Example of DEEM Outputs

Figure 1a. DEEM Summary Table (AC7)

DEEM-FCID ACUTE Analysis for ALDICARB                           
(1994-98 data)

Residue file: Water GA 300 final.R98                  Adjustment factor
#2 used.

Daily totals for food and foodform consumption used.

MC iterations = 200      MC list in residue file     MC seed = 10

========================================================================
=======

Summary calculations (per capita):

                    95th Percentile      99th Percentile      99.9th
Percentile

                   Exposure   % aRfD    Exposure   % aRfD    Exposure  
% aRfD 

                  ---------- --------  ---------- --------  ----------
--------

All infants:        0.000852   131.14    0.001318   202.79    0.002171  
333.99 

Aldicarb Equiv./1   0.000733   112.78    0.001133   174.39    0.001867  
287.23 

       Percent of Person-Days that are User-Days =  89.38%

1 - Values converted to Aldicarb (parent) equivalents by multiplying
(0.86).

Figure 1b. DEEM Plot File (PLT) – Plot generated in Excel based on
DEEM bins

Total person days (weighted & unweighted) =,      7548892,         2940

Total user days   (weighted & unweighted) =,      6747448,         2642

Bin totals based on 200 iterations. 



Figure 1c. DEEM Contribution Exp (CEC)

 Demographic data for each record, Exposure contribution data by food
(Selected Records):

PID,  HH-Indiv, Day,Sex,  Age,    Bw-kg,  Nf,  Nx,   Tot Expos, 
Samplwt,

      Food,     FF,  Amt(g),   Residue , Adj#1, Adj#2,   Contributn, 
Percnt

19984 ,46309-02 , 2  ,M   ,10M  ,  9.99 , 2  ,   1 ,  0.0101373 ,  1844
,

      1033660  ,211,  246.1 ,  0.402325 , 1.00 , 1.00 ,  0.0099013 ,  
97.67

      86020000 ,240,  368.3 ,  0.003900 , 1.00 , 1.00 ,  0.0001436 ,   
1.42

18832 ,28892-02 , 2  ,F   ,0M   ,  3.18 , 2  ,   2 ,  0.0020118 ,  2242
,

      86010000 ,110,  295.7 ,  0.005900 , 1.00 , 1.00 ,  0.0005484 ,  
27.26

      86020000 ,130,  789.2 ,  0.005900 , 1.00 , 1.00 ,  0.0014633 ,  
72.74





Figure 2.  Example Illustrating Method to Compute Eating Occasion
Exposure from DEEM CEC Output

Data on the timing and amounts of foods and indirect drinking water
throughout the day are taken from the CSFII food diaries, and merged
into the respective DEEM CEC diaries to obtain eating occasion
estimates.  Assumptions are required regarding the timing and amounts of
direct drinking water consumption since that information is not
available in CSFII.  One option, depicted here, is to equally allocate
the total amount over six fixed events: 360 minutes after midnight or 6
am, 9 am, 12 noon, 3 pm, 6 pm, and 9 pm.  For this particular
simulation, total exposure (aldicarb sulfone equivalents) is 0.00201
mg/kg/day, or 266% of the aPAD (=(0.00201x.86)/.00065).  Under the
eating occasion approach, the maximum cumulative exposure with a two
hour half-life is 0.000475 mg/kg, or 73% of the aPAD.  

Source	Amt (mL/Day)	Residue (mg/L)	Exposure (mg/day)	Exposure
(mg/kg/day)	Share of Exposure

With a two hour half-life:     

    0.71

	2.71E-04	3.62E-04

Indirect DW	-	107	0.0059	1:00 PM (780)	0.00020

0.00020	780	60	0.71	2.56E-04	4.55E-04

Direct DW	4	49	0.0059	3:00 PM (900)	0.00009

0.00009	900	120	0.50	2.27E-04	3.19E-04

Indirect DW	-	134	0.0059	6:00 PM (1080)	0.00025

0.00034	1080	180	0.35	1.13E-04	4.53E-04

Direct DW	5	49	0.0059	6:00 PM (1080)	0.00009

0.00032	1260	180	0.35	1.60E-04	4.75E-04

Indirect DW	-	121	0.0059	9:00 PM (1260)	0.00022

 	 	 	 	 	4.75E-04

Direct DW	6	49	0.0059	9:00 PM (1260)	0.00009

( 73% aPAD)







 The results of this analysis are shown in Table 6 for Food Only and
also for Food + Drinking Water for the three Georgia  groundwater and
one Florida surface water scenarios at the per capita 99.9th percentile
for the eating/drinking occasion analysis using a two hour half-life for
aldicarb.  These results are presented as % of the acute Population
Adjusted Dose (aPAD).  Four drinking water concentration scenarios were
modeled for aldicarb: 3 ground water scenarios for aldicarb use on
peanuts/cotton in Georgia with an assumption of 300 ft, 500 ft and 1000
ft well set-backs, and one ground water scenario for aldicarb use on
Florida citrus with a 1000 ft set-back. 

From the table below, it can be seen that incorporating eating occasion
analysis and the 2 hr. recovery half life for aldicarb into the Food
Only analysis does not significantly change the risk estimates when
compared to baseline levels (for which a total daily consumption basis
– and not eating occasion - was used).  From this, it is apparent that
the modifying the analysis such that information on eating (i.e. food)
occasions and aldicarb half life is incorporated results in only minor
reductions in estimated risk: generally on the order of several percent
at most for all age groups.  However, risk estimates for which food and
drinking water are jointly considered and incorporated  (i.e, Food +
Drinking Water in the table below) are reduced considerably (by a factor
of 2 or more in some cases) compared to baseline and  is  not
unexpected: infants receive much of their exposures from indirect
drinking water in the form of water used to prepare infant formula.  The
only scenario for which the aPAD exceeds 100% is the Food + Water (GA-GW
300 ft) scenario for infants.  For all other scenarios and
subpopulations, the estimated risk at the 99.9th percentile  is less
than 100% of the aPAD for Food + Water when information on the timing of
eating/drinking occasions is considered and incorporated into the
analysis.         

Table 6.  Estimated Risks at the Per Capita 99.9th Percentile (2 hr
half-life)

Subpopulation	DEEM-Based Eating /Drinking Occasion Analysis

(Risk Expressed as %aPAD at 99.9th Percentile)

	Food Only	Food + Drinking Water



GA-GW 300 ft	GA-GW 500 ft	GA-GW 1000 ft	FL-SW 1000 ft

US Pop	34%	58%	44%	36%	42%

All Infants	41%	147%	88%	43%	80%

Children 1-2 yrs	72%	95%	80%	76%	78%

Children 3-5 yrs	60%	77%	64%	59%	62%

Children 6-12 yrs	46%	48%	45%	42%	44%

Youth 13-19 yrs	28%	46%	33%	28%	30%

Adults 20-49 yrs	29%	54%	38%	30%	36%

Adults 50+ yrs	34%	47%	37%	34%	37%

Females 13-49 yrs	29%	50%	35%	28%	34%



b  Stochastic Human Exposure and Dose Simulation (SHEDS) Model  

	The above analysis was also conducted with SHEDS –NMC, a modified
version of ORD’s Stochastic Human Exposure and Dose Simulation Model. 
As indicated before, SHEDS-NMC is designed to approximate the
assumptions, data, and algorithms used in the standard models currently
used by OPP in its risk assessments. The SHEDS-NMC exposure estimates
for each scenario and subpopulation of interest are shown below in Table
7. The corresponding DEEM results (from Table 6) are indicated
parenthetically for comparison.  As can be seen, the results are
generally within several percent of one another, with no difference
greater than 10%.   

   

Table 7.  Estimated Risks at the Per Capita 99.9th Percentile (2 hr
half-life) – SHEDS vs. DEEM Estimates

Subpopulation	SHEDS-NMC Eating/Drinking Occasion Analysis

(DEEM Eating/Drinking Occasion Analysis)

	Food Only	

Food + Drinking Water



GA-GW 300 ft	GA-GW 500 ft	GA-GW 1000 ft	FL-SW 1000 ft

USPop	35%

(34%)	55%

(58%)	42%

(44%)	36%

(36%)	41%

(42%)

All Infants	41%

(41%	139%

(147%)	85%

(88%)	42%

(43%)	77%

(80%)

Children 1-2 yrs	77%

(72%)	91%

(95%)	80%

(80%)	78%

(76%)	79%

(78%)

Children 3-5 yrs	57%

(60%)	71%

(77%)	61%

(64%)	57%

(59%)	60%

(62%)

Children 6-12 yrs	43%

(46%)	46%

(48%)	44%

(45%)	43%

(42%)	44%

(44%)

Youth 13-19 yrs	31%

(28%)	44%

(46%)	34%

(33%)	31%

(28%)	33%

(30%)

Adults 20-49 yrs	30%

(29%)	52%

(54%)	37%

(38%)	30%

(30%)	36%

(36%)

Adults 50+ yrs	32%

(34%)	45%

(47%)	36%

(37%)	33%

(34%)	35%

(37%)

Females 13-49 yrs	30%

(29%)	50%

(50%)	37%

(35%)	30%

(28%)	36%

(34%)



IV.	Drinking Water Consumption Patterns

a. CSFII Data  

As noted in Section II above, the relatively high contributions from
drinking water in some scenarios is due to high amounts of consumption
among infants and toddlers.  Drinking water intake differs between these
two subpopulations:  infants receive much of their exposures from
indirect drinking water, generally via formula intake, while toddlers
and older children (as well as adults) receive much of their drinking
water exposures through consumption of direct drinking water as well as
indirect drinking water.   Figures 4a and 4b plot drinking water
consumption from the CSFII/FCID data base, by age group, in mL/day and
mL/kg bwt/day, respectively.  Similarly, Figures 4c and 4d plot drinking
water consumption for infants <1, by month, in mL/day and mL/kg bwt/day,
respectively.  As Figure 4b depicts, infants tend to have higher overall
drinking water consumption rates (mL/kg bwt/day) than children, who in
turn, tend to have higher consumption rates than adults. 

As with any survey, there are a number of drinking water consumption
amounts which are “high end” and might be considered outliers.  For
example, one teenager (HHID-SPNUM-DAY: 22749-3-1) reported consuming
over 20 Liters/day (direct), and several adults reported consuming more
than 10 Liters/day, some via beverages (indirect).   For neither of
these groups, however, was the aPAD exceeded at the 99.9th percentile. 
In addition, there are a few significant consumers among infants and
toddlers (1 to 2 yr olds).  Figure 5 for example highlights two newborn
infants that weighed less than 4 kg, and consumed nearly 2 liters of
water (primarily through formula).  A preliminary inspection of the food
diaries reported for these two individuals indicate that a set amount of
formula was reportedly prepared and consumed by the infants on multiple
occasions throughout the day.  The first infant diary (28892-2-1) was
for a  newborn (0 month old) weighing 3.2 kg, indicated that a total of
8 oz of formula (6 ounces consumed directly + 2 oz used to prepare 0.25
cup of dry rice cereal) was prepared and consumed at 8:00 am, 9:30, 11,
1:30, 4:30, 6:00, 10 and 11:30 pm; an additional 4 oz of formula alone
was prepared/consumed at 1:00 am.  The second infant-dairy (26837-3-2)
was a one month old that weighed 3.6 kg, and consumed 8 oz of formula
nine times during the day at 4:00 am, 6, 8, 10, 12, 2, 6, 8 and 10 pm.
These two consumption amounts are depicted in Figure 5 as the two points
in the upper right hand corner of the plot that deviate from the near
linear pattern established by the majority of the remaining reported
consumptions. These two values are, respectively, 52% and 41% higher on
a ml/kg bw basis than the next (third) highest reported consumption
value.  In order to verify that these consumption diaries were not in
and of themselves responsible for the 300 ft set-back GA groundwater
scenario exceeding the aPAD at the 99.7 percentile of exposure and to
evaluate the sensitivity of the results for the infant subpopulation to
these and other high-end consumption amounts by infants, the Agency
conducted a number of sensitivity analyses using both SHEDS and the
DEEM-based approach.  Under one sensitivity analysis, , the reported
amounts consumed were reduced by 50 percent and in the other analysis 
these high-end consumption amounts were dropped altogether.  The 99.9th
percentile exposures did not change considerably in either analysis,
even when the reported amounts consumed was reduced for an expanded set
(top five instead of top two) of diaries.  







Figure 5. Box-Cox Plot of Infant Direct Water Consumption (ml/kg bw
basis)

b  Bayer Drinking Water Consumption Survey (Direct Water Only, no
infants included)

 

As noted above for direct drinking water, the USDA CSFII collected
information on only the total amount consumed during the survey date; it
did not collect information on the per-occasion amounts and timing of
drinking water events throughout the day.  For newborn infants, indirect
drinking water (via formula) is their primary source of water
consumption, and information on timing and amount consumed is available
in the CSFII.  But the primary source of water intake for many toddlers,
older children and adults is often direct drinking water and the CSFII
diaries do not contain information on the timing of the consumption
events or the per-occasion amounts.  To address this deficiency, Bayer
CropScience sponsored a study in 2004 on direct drinking water
consumption entitled “Drinking Water Consumption Survey” , and
submitted their report and the raw data to the Agency.  The objective of
this study was to obtain a distribution of water intake for a 24-hour
time period that was nationally representative sample of the US
population.  Participants recorded their direct drinking water
consumption (time of day and amount consumed) over a one-week (7 day)
period during summer 2000 (August), and winter 2001 (March).  A total of
4,198 individuals from 2,154 households participated in the survey
providing a total of 27,282 person-day diaries, i.e., 93% of the total
of all participants returned diaries for all 7 days.   

According to the report (Barraj, L.M. et.al. (2004), Exponent®, Inc.;
National Product Database (NPD) Group), one of the potential uses of
these data is to refine a probabilistic exposure assessment: 

“It may be possible, using the information collected by the DWCS to
“allocate” the total daily water consumption amount reported in the
CSFII into various drinking occasions. Specifically, if each subject in
the CSFII survey was randomly matched to subjects in the DWCS, based on
survey season, region, age, gender, and total amount of drinking water
consumed per day, then the total amount reported by that CSFII
participant can be allocated to the same number of drinking occasions as
those reported by the matching DWCS participant.  Similarly, the
proportion of the total daily water consumption allocated to each of
these drinking occasions can be assumed to be similar to that reported
by the matching DWCS participant. This approach would then allow a less
than 24-hour assessment of both food and drinking water (aggregate
assessment) for a pesticide.” (Bayer 2005, p.17) 

Figures 6a and 6b were taken from DWCS report (Barraj, L.M. et.al,
2004).  Figure 6a depicts the total number of occasions that survey
respondents reported consuming (direct) drinking water throughout the
day.  The data support the expectation that drinking water is consumed
throughout the day, and that a drinking occasion analyses may be useful
in refining a dietary risk assessment for aldicarb.  Figure 6b indicates
that individuals consume drinking water at all times during the day. 
While this chart may lend support to the modeling assumption used for
direct drinking water (6 equally fixed times), it is not directly
applicable since this distribution applies to the entire population, and
not to any particular individual.  

The Bayer DWCS data provide an alternative approach to allocating direct
drinking water consumption, rather than the six equally fixed occasions
scenario used above.   In order to investigate the sensitivity of the
DEEM- and SHEDS- estimated exposures using the 6 fixed direct drinking
water occasions, the Agency used the Bayer DWCS information for various
age-gender-season populations(Infants <1 year old were not included in
the Bayer survey so investigation of the sensitivity of these results to
alternate direct water consumption timing scenarios was not possible;
however, the majority of the high-end exposures for infants were derived
from indirect water consumption (primarily infant formula) for which
intra-day timing and consumption amounts WERE available).    Table 8
provides the total number of drinking water diaries in the DWCS by
gender, age and season.  











The steps used to incorporate these DWCS data into an alternate direct
drinking water occasion-based analysis are as follows: 

Generate cohort by gender, age, season (36 bins in Table 10)

Calculate percentage of direct DW by each E.O.

Merge total DW from CSFII with Bayer DW data

Use Total DW from CSFII and percentage of DW from Bayer DW data to
calculate DW amount for each O.E. (occ_time from Bayer data)

This was done by using SHEDS and SAS statistical software.  Table 9
compares the results of this alternative allocation of direct drinking
water consumption to those derived from the six equally fixed-based
approach.  As the two estimates indicate, the risks at the per capita
99.9th percentile appears to be relatively robust with respect to the
allocation of direct drinking water consumption over the day, with the
risks estimated using the Bayer DWCS drinking occasion  data slightly
higher than those estimated using the 6 fixed interval assumption used
in the previous analysis.  In all cases, however, estimated risks are
less than 100% of the aPAD for the 500+ foot set-back scenarios.  

Table 9. SHEDS Estimated EO Risk at Per Capita 

99.9th Percentile

2 hr Half-Life, Direct DW Consumption: Baseline=6 fixed events

Subpopulation	Risk (%aPAD) at the Per Capita 99.9th%

	Food Only	

Food + Drinking Water



GA-GW 300 ft	GA-GW 500 ft	GA-GW 1000 ft	FL-SW 1000 ft

US Population	35%	55%	42%	36%	41%

Infants	41%	139%	85%	42%	77%

1-2 yrs	77%	91%	80%	78%	79%

3-5 yrs	57%	71%	61%	57%	60%

6-12 yrs	43%	46%	44%	43%	44%

13-19 yrs	31%	44%	34%	31%	33%

20-49 yrs	30%	52%	37%	30%	36%

50Plus 	32%	45%	36%	33%	35%

Females 13-49 yrs	30%	50%	37%	30%	36%

2 hr Half-Life, Direct DW Consumption: Bayer DWCS

Subpopulation	Risk (%aPAD) at the Per Capita 99.9th%

	Food Only	

Food + Drinking Water



GA-GW 300 ft	GA-GW 500 ft	GA-GW 1000 ft	FL-SW 1000 ft

US Population	35%	66%	46%	36%	44%

Infants	NA	NA	NA	NA	NA

1-2 yrs	77%	100%	81%	78%	80%

3-5 yrs	57%	101%	72%	58%	71%

6-12 yrs	43%	62%	46%	44%	45%

13-19 yrs	31%	53%	37%	31%	36%

20-49 yrs	30%	64%	43%	30%	41%

50Plus 	32%	46%	36%	33%	35%

Females 13-49 yrs	30%	55%	39%	30%	37%

N/A:  not available.  The Bayer DWCS did not survey infants



V.	Risk Characterization & Summary

This memorandum summarizes OPP’s dietary exposure modeling for
aldicarb eating/drinking occasions using both the DEEM-FCID and the
SHEDS-NMC models.  

Like DEEM-FCID, SHEDS-NMC was designed to utilize the CSFII two day
diaries as its primary reference population; this leads to the similar
estimated exposures and risks  between the two models

The DEEM outputs (along with data from the USDA CSFII food diaries and
the Recipe files) can be used  to compute  Eating/Drinking
Occasion-based exposure estimates which incorporate the AChE recovery
half-life associated with aldicarb ; while this approach has a few
limitations relative to SHEDS-NMC, it produces reasonably similar
results. 

The estimated risks under an eating/drinking occasions approach which
incorporates decay rates are  significantly lower than the total daily
approach to the extent that exposures -- and in particular drinking
water exposures -- occur throughout the day rather than during one
instantaneous event.  

Several infant food/water consumption diaries appear to be distinctly
higher than would be expected and might be considered outliers; EPA’s
preliminary sensitivity analyses, however, indicates that the estimated
risks at the per capita 99.9th percentile are relatively insensitive to
use of these diaries:  specifically, discarding these diaries, or
reducing the amounts consumed by fifty percent, has only small effects
upon the estimated risks. 

The CSFII did not collect information on the timing or per-occasion
amounts for  direct drinking water intake (only the timing and
per-occasion amounts of indirect water intake was provided).  For direct
drinking water, then, it was necessary to allocate the total amount as
reported by the CSFII respondents based on simple assumptions (e.g.,
direct drinking would be allocated equally over 6 equally fixed times),
or by use of survey data.

EPA empirically utilized the Bayer sponsored DWCS data to produce an
alternative method for allocating drinking water intake throughout the
day for the non-infant subpopulations.  The corresponding exposures and
risks at the per capita 99.9th percentile were not significantly altered
from the simple assumption that was initially developed.

APPENDIX

The steps which were followed are highlighted below: 

1.  Run DEEM with the aldicarb food and drinking water residues,
specifying DEEM output files for Summary Report (*.AC7), Plot File
(*.PLT) and CEC (*.CSV) .  For the CEC report, specify the range from
95% to 100%, and the minimum of 1% of total daily exposure for
contributions from food forms, and save it as a CSV file.  

2.  Specify the following iterations for each subpopulation: (US Pop=10
iterations, Infants=200 iterations, 1to2 yrs old=100 iterations, 3to5=75
iterations, 6to12=100 iterations, 13to19=100 iterations, 20to49=75
iterations, 50+=75 iterations, Females 13to49=100 iterations.  If too
many iterations are specified, then the top 5% of all simulated records
will exceed the 40,000 record limit in the CEC module.  

3.  Import the CEC output containing the top 5 percent of simulated
diaries into a (SAS) data base; specifically:   open the output file
(*.CSV)  using MS Notepad, cut the first section of text before the top
5 percent of diaries, add 1 comma separated row with the variable names
for each of the fields, save and close the file, and import it into the
data base.  Note:  Each record is a unique simulated person-day: many
diaries will appear multiple times in this top 5% due to different draws
of food and or drinking water residues.  Even a unique simulated
person-day may appear multiple times in the Monte Carlo simulation.  To
save space, DEEM prints these ‘unique’ records only once; that is
why the “Number of Printed Records” is generally smaller than the
“Number of Records Represented”.

4.  Separately, from the CSFII food diary table (rt30) extract the data
on the time of day (OCC_TIME), and amounts consumed (FOODAMT) for each
food code for each food consumed (FOODCODE, MODCODE) for each unique
diary: Household (HHID), Person (SPNUM), and Day of Intake (DAYCODE).

5.  Merge results from Step 4 with the FCID Recipe table (Recipe_FCID),
by foods (FOODCODE, MOD) , and calculate the amounts of raw agricultural
commodities (FCID_CODE) consumed on each eating occasion
(=FOODAMTxConsumption/100). 

6.  Merge simulated residues from the DEEM CEC output with the timing
and corresponding amounts for each eating occasion from the CSFII/FCID
table (Step 5), by raw agricultural commodity (FOOD=FCID_CODE).  

7.  For each unique simulated person-day in the top 5%, calculate the
time between eating occasions (in minutes), and the discount rate based
on a 2 hour half-life.   For example, if the first eating occasion
occurred at  6:00 am (or 360 minutes after midnight), and the next
eating occasion occurred at 8:00 am (or 480 minutes after midnight),
then the time difference is 120 minutes and the discount rate between
these two eating occasions is: (0.5)^(120/120)=0.5.   Alternatively, if
the second eating occasion occurred at 10:00 am instead of 8:00, then
with the four hours (or 240 minutes between eating occasions), the
discount rate is (0.5)^(240/120)=0.25.   

8.  Calculate the Total Discounted Exposure on Each Eating/Drinking
Occasion by summing the exposure realized on the current eating occasion
and the Total Discounted Exposure on the Previous Eating Occasion
multiplied by the discount rate. 

9.  For that unique simulated person day, calculate the Maximum “Total
Discounted Exposure on Eating Occasion” or ‘Eating Occasion
Estimates’ realized during the day, and output that value into a
table. 

10.  Sort the ‘Eating Occasion Estimates’ by descending order and
compute the contribution of this unique person-day: the overall weight
is the product of the CSFII 2-Day Sampling weight, and the number of
occurrences that permutation appeared in the simulation
(WeightPrj=SAMPLWT*NX).  The cumulative percentile can be calculated by
adding the contributions from that unique simulated person-day
(WeightPrj) with the values from all person-days with higher exposure
values.

11.  Calculate the Total Simulated Projected Population by multiplying
the Projected Population (Sum of SAMPLWT printed in PLT file)  by the
number of iterations (printed in the AC7 output).  Calculate Per Capita
Percentiles by subtracting the cumulative percentiles from this Total
Simulated Projected Population.  These per capita percentile values are
used for the corresponding Eating/Drinking Occasion Exposure Estimates. 
The 99.9th percentile is used as the DEEM-Based Eating Occasion value
for that subpopulation.

12.  Check Computations: the sum of these simulated diary weights
(WeightPrj) for the entire top 5% should approximately equal to 5% of
the total simulated projected population; this total is proportional to
the number of iterations specified in the DEEM simulation.  Check the
estimates at the various calculated per capita percentiles (99.95%,
99.9%, 99.85%, etc.) with the estimates in the DEEM Plot file (per
capita estimates are in the last column).  Note: The lowest exposure
value in this CEC output is often quite different than the actual DEEM
estimate at the per capita 95th; this is expected since DEEM saves these
simulated records ‘on-the-fly’, and not after all simulations are
completed.  You can increase (slightly) the number of iterations
specified in Step 2.  However, if too many iterations are specified, and
the total number of printed records reaches the maximum 40,000 limit,
then these calculations may be inaccurate since those outputted records
may not include the entire top 5% of all records, nor will they not be a
random sample of the top 5 percent of All simulated person-days.  

  For further details, see
http://www.epa.gov/scipoly/sap/meetings/2000/index.htm.

  This is a version of SHEDS that was developed for use with the
N-methyl carbamate insecticides.

  For infants, the estimated risks at the per capita 99.9th percentile
exceeds the level of concern under the Georgia 300 ft scenario (147% of
the aPAD).  The estimated risks for the Georgia-GW 300 foot setback
distance equals 100% of the aPAD at approximately the per- capita 99.7
percentile of the infant subpopulation.  

  The figure represents a Box-Cox transformation of the infant direct
water consumption values (expressed on  a ml/kg basis).  The Box-Cox
transformation is a transformation which can be used to identify
statistical outliers and highlight values which do not reflect typical
patterns.  In this case, the Box-Cox procedure indicated that a square
root transformation of direct drinking water consumption would be 
appropriate to convert the drinking water consumption amounts  to a near
normal distribution which would plot as a straight line on a normal
probability plot.  Figure 5 is a  normal probability plot of the square
root of reported direct water consumption (ml/kg bw).  

Aldicarb                                            Dietary Exposure and
Risk Assessment                             DP Number: 299889

PC Code: 098301

________________________________________________________________________
______

 PAGE   

Page   PAGE  2  of   NUMPAGES  24 

Aldicarb                                            Dietary Exposure and
Risk Assessment                             DP Number: 299889

PC Code: 098301

________________________________________________________________________
_____________________

Page   PAGE  1  of   NUMPAGES  24 

