UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, DC 20460
OFFICE OF CHEMICAL SAFETY AND
POLLUTION PREVENTION
\
MEMORANDUM December 16, 2021
SUBJECT: Science Review of the AEATF II Immersion/Dip/Soak (IDS) Human Exposure
Monitoring Study (AEATF II Project ID AEA12; MRID 51588901).
PC Code(s): Not Applicable (NA)
DP Barcode(s)/No(s): NA
Petition No(s).: NA
Regulatory Action: Human Health
Risk Assess Type: Surrogate Handler Exposure Data
CaseNo(s).: NA
MRIDNo(s).: 51588901
40CFR: None
Risk Assessment and Science Branch (RAB1)
OPP/Antimicrobials Division (7510P)
Statistician
ICF (EPA Contractor)
Thru: Timothy Dole, CIH "Twi&y C-
Risk Assessment and Science Support Branch (RAB2)
OPP/Antimicrobials Division (7510P)
TO: Melissa Panger, Ph.D., Acting Branch Chief
Risk Assessment and Science Support Branch (RAB1)
OPP/Antimicrobials Division (7510P)
This memorandum presents the revised EPA/Office of Pesticide Program (OPP) Antimicrobials
Division (AD) science review of the human exposure immersion/dip/soak (IDS) study submitted
by the Antimicrobial Exposure Assessment Task Force II (AEATF II). The original EPA
Science Review, dated September 20, 2021, was reviewed by the Human Studies Review Board
(HSRB) and three revisions have been incorporated based on that review: (1) correct a typo in
Table 1 for the dermal unit exposure for the Sink scenario; (2) clarify the concentrations in the
inset table on page 17 of this review; and (3) clarify for the Sink scenario that the monitoring
times are sampling-based, not based on work efficiency. The dermal and inhalation exposure
data as represented in this review are acceptable and, subject to the considerations described
below, are recommended for use for pesticide handler exposure assessments.
Page 1 of 41
-------
EXECUTIVE SUMMARY
This document represents the USEPA, Office of Pesticides Program, Antimicrobials Division
(AD) review of the Antimicrobial Exposure Assessment Task Force II (AEATF II)
immersion/dip/soak (IDS) study. The AEATF II designed the study to develop unit exposures for
people who immerse, dip, and/or soak (restaurant-type) equipment and utensils, or use a
rag/sponge in a bucket to wipe surfaces, with a treatment solution of an antimicrobial product.
The results of the study are reported herein. The protocol for this completed study was
previously reviewed by the EPA and the Human Studies Review Board (HSRB) for ethical and
scientific design. Both EPA and HSRB approved the protocol and provided recommendations
for some modifications (discussed within this memo). This memo contains the scientific review,
recommended unit exposures, and study limitations to be considered by users. The ethics review
is contained in a separate memo. Both reviews are to be presented to the HSRB at the planned
October 21, 2021 meeting.
The study investigators monitored inhalation and dermal exposures to a total of 54 different test
subjects. The 54 subjects were separated into three distinct "sub" scenarios within this
overarching study called immersion/dip/soak (IDS). The three "sub" scenarios for which the
distinct inhalation and dermal exposures have been developed are (1) Bucket and rag/sponge, (2)
3-compartment sink, and (3) Clean-out-of-place (COP). Each of the three "sub" scenarios
comprised 18 of the subjects, and none of the subjects participated in more than one of the
scenarios. Both the C14 analog of alkyl dimethyl benzyl ammonium chloride (ADBAC)
(dermal) and didecyl dimethyl ammonium chloride (DDAC) (inhalation) were used as the active
ingredients (a.i.) as the surrogate test compound by all test subjects. ADBAC inhalation
exposures were also measured but were not used in this science review because of issues with
background contamination in the controls, instead DDAC was used for inhalation sampling.
These two Quats were selected as surrogate compounds because of their stability, low vapor
pressure, low detection limits, and registered uses. All test subjects were recruited from the food
service, janitorial, hotel, etc. industries. All cleaning (sanitizing) activities were performed
indoors. The term "cleaning" is used colloquially within this review as the test substance as well
as the tasks being performed are meant to be used to "sanitize", which is a specific pesticidal
claim to control a specific microorganism. Each subject was randomly assigned within each of
the three scenarios to perform the specific cleaning (i.e., sanitizing) tasks for a given
concentration of a.i. and time. Subjects were instructed to work as they normally would. EPA
confirms that the data are considered the best available data for assessing handler exposures from
antimicrobial treatment solutions for immersing, dipping, and soaking equipment and utensils, as
well as cleaning surfaces with a bucket and rag/sponge. The reader is referred to Section 3.0 for
a discussion on the data limitations and use of the data as a surrogate for other a.i.s.
EPA intends to use these AEATF II immersion/dip/soak unit exposures for hard surface
sanitizing and disinfectant uses in food service areas as well as for wiping when product labels
allow for bucket and rag/sponge. These scenarios do not cover the pouring of an antimicrobial
product into the containers to make up the treatment solutions. Those mixer/loader scenarios are
monitored in separate AEATF II studies (the mixer/loader portion was conducted separately
because many different formulations can be used such as liquids, powders, flakes, metering
systems, etc).
Page 2 of 41
-------
Select summary statistics for the "unit exposures" (i.e., exposures normalized to the
concentration of a.i. in the treatment solution (ppm a.i.) and to the duration of exposure (hours))
are presented in Table 1 for the dermal and inhalation routes of exposure. Each test subject wore
both inner and outer whole-body dosimeters (WBD) that were sectioned and analyzed separately
for each body part (e.g., lower leg, upper leg, forearm wipe or lower arm, upper arm, etc.).
Table 1. Unit Exposures (UE) for the AEATF II Immersion/Dip/Soak (IDS) Scenarios.
Exposure Route
Clothing and/or Inhalation3
(Normalization Units)
AEATF IIb c(n= 18)
Arithmetic
Meand
95th Percentile6
Bucket Rag & Sponge
Dermal
Long pants/short-sleeves, no gloves
(mg/(ppm a.i. x hours))
0.096
0.211
Inhalation
Dose
(mg/(ppm a.i. x hours))f
1.98E-6
5.61E-6
8-hr TWA
((mg/m3)/(ppm a.i. x hours))8
2.47E-7
7.01E-7
3-Compartment Sink
Dermal
Long pants/short-sleeves, no gloves
(mg/(ppm a.i. x hours))
0.00371
0.0072
Inhalation
Dose
(mg/(ppm a.i. x hours))f
3.88E-6
1.48E-5
8-hr TWA
((mg/m3)/(ppm a.i. x hours))8
4.85E-7
1.85E-6
Clean-Out-of-Place (COP)
Dermal
Long pants/long-sleeves, gloves
(mg/(ppm a.i. x hours))
0.000734
0.00258
Inhalation
Dose
(mg/(ppm a.i. x hours))f
5.73E-5
2.10E-4
8-hr TWA
((mg/m3)/(ppm a.i. x hours))8
7.16E-6
2.63E-5
a Unit Exposures (UEs) reported in Table 1 have been converted to represent "hours" rather than the
"minutes" which are the duration units reported throughout this review and Appendix A.
bDermal and inhalation UEs are corrected for laboratory recoveries (field recoveries >100%).
c Statistics are estimated using a lognormal simple random sampling model. Non detected (ND) values are
estimated using substitution by 'A the limit of quantification (LOQ). Details are described in Appendix A.
d Arithmetic Mean (AM) = geometric mean (GM) * exp{0.5*(lnGSD)2}
e 95th percentile = GM * geometric standard deviation GSD1645
f Inhalation (mg/(ppm a.i. x hours) = air cone ((mg/m3) / (ppm ai * duration)) * breathing rate (1 m3/hour) *
exposure duration (hours/day)
8 8-Hour Time Weighted Average (TWA) ((mg/m3)/(ppm a.i. x hours)) = air cone (mg/m3) / (ppm a.i. x
hours) * study exposure duration (hours/day) / 8 (hours)
Page 3 of 41
-------
The following important points with respect to these data are noted:
• The unit exposures reported in Table 1 have had their units converted from the review
herein and Appendix A from "ppm x minutes" to "ppm x hours" for ease of use by
exposure assessors as worker exposure durations are reported in "hours" rather than
"minutes". The unit exposure conversion was calculated by multiplying the unit
exposures reported herein by "60" to convert the minutes to hours.
• The dermal unit exposures recommended in Table 1 are based on the short-sleeved shirt,
long pants, no gloves for the bucket and sink scenarios and the long pants, long-sleeved
shirt, gloves for the COP scenario. The hand exposure for all three scenarios represents
nearly 100% of the exposure (i.e., the other body parts round-out for the bucket and sink
scenarios and nearly round-out for the COP scenario).
• Estimates of the dermal unit exposures for the geometric mean (GM), arithmetic mean
(AM), and 95th percentile (P95) were shown to be accurate within 3-fold with 95%
confidence for all but the P95 for the COP scenario for the empirical simple random
sampling model (see Table 7 below). The inhalation unit exposures meet the 3-fold
relative accuracy objective for all but the P95 for all three of the IDS scenarios for the
empirical simple random sampling model. At this time, no additional monitoring for the
three IDS scenarios is required.
• The statistical analysis (Section 2.4) provides evidence consistent with log-log-linearity
with a slope of 1111 between dermal exposure and the treatment solution concentration
and exposure duration for the bucket and sink scenarios, but not the COP scenario. An
ideal result of the log-log-linearity test is an estimated slope between 0 and 1 with a
confidence interval that includes 1 but not zero indicating that independence between
exposure and ppm x duration (a slope of zero) is rejected and that log-log-linearity with a
slope of 1 is not rejected. The results reported in Section 2.4, Table 8 of this analysis
indicate the following:
o For the bucket scenario, the confidence intervals for the slope exclude 0 and
include 1 for both dermal and the inhalation 8-hr TWA. Thus, the "unit exposure"
approach for both the dermal and inhalation for the 8-hr TWA is a reasonable
approximation.
o For the sink scenario, the confidence intervals for the slope exclude 0 and include
1 for dermal. Thus, the "unit exposure" approach for the dermal is a reasonable
approximation. However, for inhalation 8-hr TWA exposure, the slope is negative
and the confidence intervals include 0 but not 1, thus the assumption of
independence was supported and the assumption of log-log-linearity with slope 1
was rejected. The results for inhalation exposure seem to be counterintuitive.
[1] The statistical analysis of log-log-linearity tests whether the slope of log exposure against log a.i., or for this IDS
study, the log ppm a.i. x duration, is 1, which supports the use of the data in the "unit exposure" formats. We now
refer to these analyses as the log-log-linearity analyses. In the Governing Documents and in previous reviews of
the AEATF II studies we have referred to these analyses as a "proportionality" analysis, but this has caused some
confusion because the statistical models do not assume that the exposure is directly proportional to the amount of a.i.
handled (AaiH), or in the IDS study, ppm x duration, but instead assume that the logarithm of the exposure is linear
in the logarithm of a.i. (or in the IDS study, ppm a.i. x duration) with a slope of 1, which is a related finding but a
very different model, as explained in more detail in Appendix A. We have therefore changed the terminology from
"proportionality" to "log-log-linearity with a slope of 1."
Page 4 of 41
-------
o For the COP scenario for both the dermal and inhalation 8-hr TWA exposure, the
confidence intervals include 0 but not 1, thus the assumption of independence was
supported and the assumption of log-log-linearity with slope 1 was rejected. This
suggests that the exposure does not depend on the normalizing factor.
• A secondary study objective for EPA is to meet 80% power for detecting log-log-
linearity with a slope of 1. This objective is met if the widths of the confidence intervals
for the slopes are < 1.4. This secondary objective was met for all scenarios; therefore, the
statistical (post-hoc) power is greater than 80%.
• The statistical analyses reported here and in Appendix A are for exposure normalized by
the product of ppm and duration, which is a reasonable way to account for the effects of
both concentration and duration on exposure. In the Supplement to Appendix A, we
explored and report results where exposure is either normalized by ppm alone, which
does not account for the effects of duration, or where exposure is not normalized at all.
To assess the risks resulting from cleaning restaurant equipment/utensils and cleaning with a
bucket and rag/sponge, EPA will combine appropriate unit exposure (UE) values with chemical-
specific inputs (e.g., maximum labeled application rates, dermal absorption, toxicological
endpoints of concern) and default inputs (e.g., hours worked) in the standard pesticide handler
exposure algorithm: Potential exposure = UE (mg/ppm ai/hour) x absorption (%) if applicable x
maximum label rate (ppm a.i. weight) x hours worked conducting the task.
1.0 Background
The AEATF II is developing a database representing inhalation and dermal exposure during
many antimicrobial handler scenarios. A scenario is defined as a pesticide handling task based
on activity (e.g., application or mixing/loading) and equipment type (e.g., paint brush/roller,
airless paint sprayer, ready-to-use wipes, bucket and rag/sponge, trigger pump sprayer, mopping,
pressure treatment of wood, etc.). The AEATF II is monitoring residues on both inner and outer
dosimeters, which will allow the EPA to estimate exposures to various clothing configurations
(e.g., long pants, long-sleeved shirt or long pants, short-sleeved shirt or short pants, short-sleeved
shirt). Hand exposure as well as inhalation exposures are also being monitored. Prior to
conducting intentional exposure studies in humans, the protocols are reviewed by the HSRB.
The HSRB reviewed this IDS exposure study protocol on October 23, 2018.
1.1 Immersion/Dip/Soak (IDS) Scenario Defined
The three IDS "sub" scenarios in this study are defined as subjects, recruited from the food
service and janitorial industries, cleaning restaurant equipment/utensils and surfaces using
techniques as they normally would do. The study conditions were simulated/designed to mimic
actual work conditions and the subject's own routines that cause them to interact/be exposed to
the a.i. were based on their own experiences.
Each of the three "sub" scenarios are described in detail as follows (the mixing/loading of the
concentrate was not performed by the subjects, the mixing/loading exposure data are available in
prior AEATF II studies):
Page 5 of 41
-------
(1) Bucket and rag/sponge: Subjects from the janitorial industry dipped a rag or sponge
into a diluted sanitizing solution in a bucket, wrung out the rag/sponge, and wiped
vertical and horizontal surfaces.
(2) 3-compartment sink: Subjects from the food service industry manually washed,
rinsed, and sanitized cookware/bakeware that do not fit in dishwashers. This scenario
is typical in restaurants/bars/schools/etc. "The first sink is used to wash the items, the
second sink is to rinse the items, and the third sink is to sanitize the items as
described in the Food and Drug Administration (FDA) 2017 Food Code. The
sanitization step requires that articles are placed into a sanitizing solution following
the time and temperature requirements of the sanitizer being used. After the articles
have been dipped and/or soaked in the sanitizer, they are removed from the sink and
place on a clean dry surface to air dry. Although the focus of this study is on exposure
during the immersion of equipment and/or utensils into an antimicrobial solution, in
the case of the food service industry, workers must follow a strict three-step process
to clean, rinse, and sanitize. Since this is a sequential activity, the entire process of
using a three-compartment sink was monitored in the study. " (AEATF 2021, page
21)
(3) Clean-Out-of-Place (COP): Subjects from the food processing industry used a
stainless-steel COP tank to clean and sanitize industrial equipment parts. "Equipment
that [is] cleaned using COP include removable articles such as fittings, clamps,
product handling utensils, tank vents, pump rotors, impellers, blades, knives, casings,
and hoses. ... Once the equipment has been disassembled, manual dry cleaning may
take place to remove debris from the equipment parts followed by placement into the
COP tank. ... After the cleaning step, the tank is drained... Once the tank is drained...
the parts may be rinsed by spraying them with a hose, and then the tank is refilled,
and a sanitizing solution is added, and the jets are turned back on. After the specified
circulation time, the sanitizing solution is drained, and the cleaned and sanitized
articles are removed and placed on racks to air dry.
The entire cleaning and the sanitization processes in COP tanks were monitored in
this study. It was not necessary to use dirty equipment parts because ...all the parts
are placed into a single tank, and they remain in the same tank throughout the
cleaning, rinsing, and sanitizing steps. Once the parts are in the tank, there is no
manual contact with the parts until they are removedfrom the tank after the
sanitizing step. ...Most of the time workers conduct one cycle of equipment cleaning
in a COP tank during a work shift, with the majority of their time spent on taking
apart equipment for cleaning, running CIP cleaning systems, and manual
cleaning/scrubbing of equipment.
The monitored activity for this use scenario included placing various pieces of
equipment into an empty COP tank, adding water and turning on the circulation to
simulate the cleaning cycle, draining the wash water, rinsing of the items in the tank
by spraying with a hose, filling the tank with water for the sanitizing step, allowing
Page 6 of 41
-------
the items to soak in the circulating sanitizing solution for at least 15 minutes,
draining the tank, andfinally removing the items and placing them on racks to air
dry. ... once the water has reached the correct level and the detergent, acid, or
sanitizer has been added, the worker turns off the water supply, turns on the jets, and
walks away to do other tasks such as disassembling equipment, cleaning floors, and
running clean-in-place systems in certain equipment. This was simulated in the study
by having the subjects move away from the COP tank for a minimum of 30 minutes
during the simulated cleaning cycle and 15 minutes during the sanitizing cycle. Once
the 15 minutes was up following the sanitizing cycle, the subject drained the COP
tank and removed the sanitized articles from the tank, placing them on a wire rack to
air dry. " (AEATF 2021, pages 22-23)
Subjects wore whole body dosimeters (WBD) underneath short-sleeved shirts, and long pants,
and no gloves plus one personal air sampler in the bucket & rag/sponge and 3-compartment sink
scenarios. The subjects participating in the COP scenario wore similar clothing configurations
except they wore long-sleeved shirts and gloves. The conditions under which the study
participants handle the pesticide as they are monitored are referred to as the scenario. Both inner
and outer dosimeters were worn by the monitored study participants, and both inner and outer
dosimeters were analyzed for residues.
1.2 Study Objective
The AEATF II's study objective is to monitor inhalation and dermal exposures to be used as
inputs in exposure algorithms to predict future exposures to persons sanitizing/disinfecting
surfaces and equipment by IDS cleaning methods. Dermal and inhalation exposure monitoring
was conducted while study participants sanitized various surfaces and equipment by various
methods (i.e., the three exposure scenarios discussed above). These exposures will be used in
pesticide exposure assessments as "unit exposures".
"Unit exposure" (UE) is defined as the expected external chemical exposure an individual may
receive (i.e., "to-the-skin" or "in the breathing zone") per weight-unit of chemical handled and is
the default data format used in pesticide handler exposure assessments. Unit exposures are
typically expressed as the amount of active ingredient (a.i.) handled by participants in scenario-
specific exposure studies (e.g., mg a.i. exposure/lb a.i. handled). In these IDS scenarios, the
logical normalization variable, based on the job function/task of dipping one's hand in a solution
over time, is the treatment concentration times the duration of exposure (e.g., ppm a.i. x hour).
EPA uses these UEs genetically to estimate exposure for other chemicals having the same or
different application rates.
Criteria for determining when a scenario is considered complete and operative have been
developed (SAP 2007). As outlined in the AEATF II Governing Document (ACC 2011), the
criteria are briefly summarized as follows:
• The AEATF II's objective for this study design is to be 95% confident that key statistics
of normalized exposure are accurate within 3-fold. Specifically, the upper and lower
95% confidence limits should be no more than 3-fold (K=3) higher or lower than the
Page 7 of 41
-------
estimates for each of the geometric mean, arithmetic mean, and 95th percentile unit
exposures. To meet this objective, AEATF II proposed an experimental design with 18
monitoring events (MEs) for "professional/commercial" employed subjects
cleaning/sanitizing hard surfaces and restaurant-type of equipment.
A secondary objective for EPA is to meet 80% power for detecting log-log-linearity with a slope
of 1. This objective is approximately met if the widths of the confidence intervals for the slope
based on the lognormal model are <1.4.
1.3 Protocol Modifications, Amendments, and Deviations
1.3.1 Protocol Modifications Based on EPA and HSRB Reviews
EPA and the HSRB provided science-based changes to the IDS protocol during the review (EPA
2018 and HSRB 2019). The review comments and AEATF II responses are summarized in
Table 2a for the EPA comments and Table 2b for the HSRB comments.
Table 2a. EPA Science R
.eview and AEATF II Responses.
EPA Issue Raised
AEATF Response
EPA Comments
1. Increase the range of
quat concentrations
used in the bucket &
rag/sponge and the 3-
compartment sink
scenarios in order to
increase the statistical
power; two options
were proposed. If
AEATF chooses
Option 1, this is
above the rate for no-
rinsing of food
contact surfaces, so
the articles used in
the study must be
rinsed prior to them
being used for food
contact.
Agreed to incorporate
Option 1 (increase levels to
1000, 600, and 100 ppm
quat). Agreed to rinse all
the test articles
(bakeware/cookware and
equipment parts) when all
the monitoring is done.
These will be items AEATF
has purchased and/or
borrowed for the study and
will not be used for food
contact until after the
monitoring is complete.
These changes will be
added to the protocol.
According to AEATF (2021) page 38 of the study report,
for the sink scenario, "After the subject completed the
work activity and was escorted out of the kitchen, a
researcher rinsed all the sanitized cookware to
remove quat residues. The rinsed items were placed
on clean metal wire racks and allowed to air dry.
Soiled cookware that had not been cleaned by the
subject was washed and rinsed by the researcher
and allowed to air dry on the wire racks. The sinks
and areas around the sinks were rinsed with water
and wiped dry to remove any quat residues. The
sponge/scrub pad(s) used by the test subject was
thrown away." Similar rinsing and wiping were
performed by the researchers after the bucket
scenario (see AEATF 2021, page 42) and COP
scenario (see AEATF 2021, page 46).
2. Make every effort to
record the volume of
dilute test solution
used by each ME
Agreed, but noted that in
some cases these will be
estimates; this is already
noted in the protocol.
The following volumes were estimated:
Bucket and Rag/Sponge: 4.75 gal of water prepared for
multiple MEs (AEATF 2021, Table 17)
Sink: 9.8 to 22.4 gallons (AEATF 2021, Table 8)
COP: 50 to 201 gal water per tank, half the MEs used two
tanks per monitoring period (AEATF 2021, Table 26)
Page 8 of 41
-------
Table 2a. EPA Science R
.eview and AEATF II Responses.
EPA Issue Raised
AEATF Response
EPA Comments
3. Allowing some
subjects to participate
in multiple scenarios
will result in a
statistical correlation
which is not desirable.
Either have all or
almost all of the
subjects participate in
both the bucket & rag
and 3-compartment
sink scenarios or have
no one participate in
more than one
scenario.
Agreed to change the
protocol to specify that no
one can participate in more
than one scenario.
There were no repeat measurements of test subjects (i.e.,
each ME was a different individual).
4. Various editorial
changes to the
protocol
These will be incorporated
No further comments.
Table 2b. HSRB Review and AEATF II Responses.
HSRB Recommendation
AEATF Responses
EPA Comments
1. Measure the water temperature
The water temperature of the
sanitizing solution used by each ME
will be measured; additionally, the
temperature of the wash and rinse
water in the sinks will be measured as
well as the wash and rinse water used
in the COP tank scenario. This will be
added to the protocol.
Water temperatures are reported in
the study report.
2. Add willingness to conduct the
work (for bucket & rag and 3-
comparetment sink) without
wearing gloves as part of the
inclusion criteria
This will be added to the inclusion
criteria in the protocol, ICF, and
recruiting materials
No further comments in the science
review.
3. Make sure that the test sites
have a range of tables and
chairs and surfaces for the
subjects to wipe during the
bucket & rag scenario
This is the plan; this detail will be
added to the protocol
Appendix C of the study report
shows pictures of the short and
long rectangle tables, round tables,
and various types of chairs.
4. Remove the statement that
having Spanish speakers adds
diversity to the study
This statement was not found in the
protocol
No further comments in the science
review.
Page 9 of 41
-------
Table 2b. HSRB Review and AEATF II Responses.
HSRB Recommendation
AEATF Responses
EPA Comments
5. Measure the height of the
subjects or the height of the
sampler to the sink; measure
the size of hands of subjects
The protocol already states that the
height of the subjects will be
recorded. Because subjects will be
moving around while they work, it
would not be practical to measure the
height of the air-sampler above the
sink; however, the height of the sinks
and COP tanks will be recorded.
Work habits including whether a
subject gets unusually close to the
sink or tank will be documented.
Measuring hand size is not practical
nor would it provide useful
information regarding exposure
potential and will not be done.
The water depths in the sinks were
measured (6 to 9 inches) as well as
the depths of the wash and rinse
sinks (AEATF 2021, Table 7 page
99). The subject's heights were
also measured. Detailed
observational notes are not
included in the study report for all
MEs, rather a few observational
notes are added such as "nothing in
the observation notes to
explain... high-end residues'". For
ME 16 bucket scenario it is noted
that the subject's face was 1 to 2
inches from surfaces he was wiping
and air sampling tube touched
surface. Hand exposure was
monitored using hand washes,
although relative hand size among
subjects might have given some
insight into the magnitude of
exposure among MEs, it would not
have been definitive to change how
the unit exposures are being
estimated or used in risk
assessments.
Page 10 of 41
-------
Table 2b. HSRB Review and AEATF II Responses.
HSRB Recommendation
AEATF Responses
EPA Comments
6. Address what will be done if
the subject wants to
drink/eat/take a break with
respect to sample collection
and making sure that residues
are not lost; what will be done
if the subject wants to wipe
their face with their hand or
forearm? How will this affect
exposure?
As requested by EPA, the AEATF
will clarify that the subjects will be
allowed to handle and hold a beverage
container. It will be made clear that if
the subject needs to stop to eat during
the monitoring event, a hand wash and
face/neck wipe will be conducted
first; however due to the relatively
short monitoring periods it is unlikely
this will occur. Subjects will be
allowed to take breaks if they want to
(this will be added to the [Informed
Consent Form] ICFs). A chair covered
with plastic will be provided as well
as water and Gatorade/sports drink.
Wiping sweat from one's forehead or
scratching/rubbing one's face or other
body parts are normal activities that
occur while working and therefore
such activities will not be prohibited;
however, this type of activity will be
noted in the observation records so
that any unusually high residues might
be explained.
For the bucket scenario, "Most
subjects worked continuously
for the monitoring period; a few
took breaks at their discretion."
MEs 4, 8, and 12 took breaks.
For the sink scenario, "Ten of
the 18 subjects worked
continuously for the monitoring
period, while the other eight
took one or more breaks at their
discretion. As kitchen workers
are normally allowed to drink
from containers with lids,
subjects were allowed to drink
the provided bottles of Gatorade
or water while they worked
and/or during their breaks, and
most did." (AEATF 2021, page
66)
For the COP scenario, "Subjects
were allowed to drink the
provided bottles of Gatorade or
water while they sat at the table
during the wash and sanitizing
cycles." (AEATF 2021, page
72)
No observations of noteworthy
activities during breaks were
reported.
1.3.2 Protocol Amendments
AEATF (2021) (page 86) lists 4 protocol amendments. The amendments included (1) increased
the compensation to the subjects in 2 of the 3 scenarios (approved by institutional review board
(IRB)); (2) made changes to heat index cutoff for stop work to be consistent with SOP AEATF
II-l IB. 1 (approved by IRB); (3) increased the compensation to the 3rd (and final) scenario
(approved by IRB); and (4) "corrected the reporting procedures for sending protocol deviations
to the IRB to harmonize with the IRB requirements" (AEATF 2021, page 86) (approved by IRB).
Page 11 of 41
-------
1.3.3 Protocol, Method, and SOP Deviations
Two laboratory, 17 protocol, and four standard operating procedure (SOP) deviations were noted
in the study (study report pages 86 and 88). Examples of the reported deviations include
different size buckets for the wipe scenarios were used (i.e., 6 and 10 qt buckets instead of 3 and
6 qt buckets); direction of airflow in relationship to subjects wiping was not recorded because
subjects were constantly changing directions; airflow measurements for the COP scenario were
not taken since there was no HVAC system and vents to outside were shut; anti foam agent was
needed for COP; LOQ for face/neck wipes and forearms was increased because of recovery
issues and background interferences in blank samples during method development; inhalation
samples were analyzed for both C14-ADBAC (as planned) and DDAC because of background
levels found in blank OSHA Versatile Sampler (OVS) tubes; the 3-compartment sink sites ended
up with one less subject at one site and one extra subject at another; etc. For a detailed
description of each of the deviations, the reader is referred to the study report. EPA accepts the
study author's conclusion that these deviations did not adversely affect the outcome of the study.
Although switching the active ingredient appears to be a major deviation, the active ingredient
included (i.e., DDAC) was already part of the pesticide formulation being monitored and was
assessed during the protocol review by EPA and the HSRB (and has existing analytical methods
as it has been used by the HSRB in previous studies).
1.4 Material & Methods
The following is a summary of the key field aspects of the study.
• Study Location: The IDS study was conducted indoors at three sites in Orlando, FL for
the bucket & rag/sponge and 3-compartment sink scenarios. Each of the two scenarios
were monitored in the kitchens/banquet halls of two churches and an Elks Lodge to
increase the variability in the sink sizes and variety of the room layouts. For the COP
scenario, the demonstration room in a COP tank manufacturing facility in Madison, WI
was used for the monitoring. Test site schematics and photos of the site/rooms are in
Appendix C starting on page 330 of the study report.
• Substance Tested: The product applied and monitored in the study was the Oasis® 146
Multi-Quat Sanitizer, EPA Reg. No. 1677-198. The two test substances in Oasis® 146,
used in the study, were alkyl (C14, 50%; C12, 40%; C16, 10%) dimethyl benzyl
ammonium chloride (known as ADBAC) and didecyl dimethyl ammonium chloride
(known as DDAC). "The specific quaternary ammonium analog that was analyzed in all
matrices is the C14-ADBAC (CAS Number 139-08-2) which makes up 50% of total
ADBAC in Oasis 146. In addition, because high levels of background CI4- ADBAC were
found in new air-sampling tubes during the analysis of the study samples, DDAC
residues, along with C14-ADBAC residues, were quantified in the OVS air-sampling
tubes generated during the field phase. " (AEATF 2021, page 24)
• Test System: The study was designed to monitor exposures to subjects cleaning (i.e.,
sanitizing) equipment and surfaces within the three IDS scenarios while varying
concentrations of the a.i.(s) and volumes of treatment solutions applied. Thus, the total
amount of active ingredient handled (AaiH) was also varied. The test systems for each of
the three IDS scenarios were setup as follows:
Page 12 of 41
-------
o Bucket & rag/sponge: Subjects wiped horizontal and vertical surfaces (e.g.,
countertops, refrigerators, tables, etc) in the kitchens and banquet halls at the 3
sites discussed above. Specifics of the sites and equipment are as follows:
¦ Site 1 - Kitchen -16 ft x -27 ft; banquet hall 64 ft x -61 f with round and
rectangular tables and plastic chairs to wipe
¦ Site 2 - Kitchen 15 ft x 24 feet; banquet hall 69 ft x -71 f with rectangular
tables and plastic chairs to wipe
¦ Site 3 - Kitchen -69 ft x -20 ft; banquet hall 50 ft x 93 ft; banquet hall
with a serving buffet/salad bar and rectangular and round tables and
plastic chairs to wipe
¦ Subjects were also given their choices of cleaning implements to use from
the following list of buckets and rag/sponges:
• 6-quart red sanitizing Kleen-Pail (WebstaurantStore)
• 10-quart red sanitizing Kleen-Pail (WebstaurantStore)
• 3-gallon plastic blue pail with a spout (Home Depot)
• 11-quart grey bucket made by Design (Target)
• 16 x 19-inch 100% cotton bar towels (white and gold stripe)
• 14 x 18-inch microfiber bar cleaning towels (white)
• 11.5 x 24-inch Chix food service wipers (pink)
• 3M C31 jumbo sponge (yellow)
• 8.25 x 4.25-inch extra-large sponge (yellow)
• Premiere pads, large cellulose commercial cleaning sponge (yellow)
o 3-compartment sink: Subjects washed (i.e., sanitized) kitchen ware at the same
three sites used and described above for the bucket & rag/sponge scenario. Each
of the three sinks had capacities of 22.4, 20.8, and 34.3 gallons at Sites 1, 2, and
3, respectively. Subjects filled each sink to the level they would normally fill it to.
"A variety of used and new cookware and bakeware was purchasedfor this study
from a commercial kitchen supply store. A total of 128 items such as mixing
bowls, cookie sheets, round cake pans, cupcake pans, frying pans, spatulas,
ladles, gravy boats, loafpans, pizza rounds, pots and pans were purchased
(page 35) each day of monitoring and prior to each ME, items were soiled by
smearing hot oatmeal over the surfaces with gloved hands to mimic soiled
cookware that would be washed in a restaurant or other food service
establishment. This was done by researchers prior to the start of the ME by
boiling water, adding oatmeal (Quaker Oats, Quick 1 Minute), and letting the
mixture thicken for about 3 to 4 minutes before applying it to the items(AEATF
2021, page 35)
Subjects were provided the following sponges/scouring pads to choose from:
• 3M 74CC Scotch Brite medium duty scrub sponge (yellow and green,
6.125 x 3.625 inches)
• Scotch Brite non-scratch scrub sponge (blue, 4.4 x 2.5 inches)
• Scotch Brite #96 general purpose scouring pad (green, cut in half to 6 x
4.5 inches)
Page 13 of 41
-------
o COP: Subjects were provided various food processing-type equipment to clean
(i.e., sanitize) in stainless steel COP tanks that were setup in the demonstration
room discussed at the site location above. The room was 26 ft x 44 ft and included
floor drains to drain the tanks. The HVAC system was not operating and the
doors were closed, thus there was no air flow during the monitoring. The COP
tank volumes were 95, 185, and 275 gallons and each tank included jets. The
tanks were roughly 2 ft deep, roughly 2 ft wide, and their lengths varied from
roughly 4, 6, and 10 ft and roughly 4 ft off the ground. (AEATF 2021, pages 42
and 43)
Figures 1-4 below illustrate the test systems, including equipment types and cleaning
(sanitizing procedures).
Figure 1. Selection of bucket & rag/sponges.
ngure 2. Wringing out a rag/sponge.
Page 14 of 41
-------
Figure 3. Researcher adding sanitizing product to 3-compartment sink
Figure 4. Subject sanitizing kitchen equipment/utensils.
Page 15 of 41
-------
Figure 5. Subject removing sanitized equipment/parts from COP tank.
• Sample Size: The study consisted of 18 monitoring events (ME) in each of the three IDS
scenarios. Each ME is a different subject (i.e., different person/individual, no repeat
subjects between IDS scenarios). The 54 MEs in this study generated a total of-1,700
samples of individual dosimeters and QA/QC samples.
• Duration: All the sampling times were time-based, not work-efficiency based. For the
bucket & rag/sponge scenario, half of the subjects were monitored for 20 minutes and half
for 60 minutes (AEATF 2021, Table 15 page 107). The duration of the monitoring for the
3-compartment sink scenario was evenly split between 1 hour and 2 hours (AEATF 2021,
Table 6 page 98). The COP monitoring duration was also evenly split between 1 and 2
hours (since the COP tank was either run once or run twice) but the times were more varied
(i.e., 66 to 75 minutes and 126 to 153 minutes as reported in AEATF (2021) Table 24, page
116). A summary of the individual ME start/stop sampling times is also reported in the
study report alongside the durations.
• Concentration of Active Ingredient (ppm): Typically, the amount of active ingredient
(AaiH) is used to normalize the inhalation and dermal exposures to calculate unit exposure
Page 16 of 41
-------
normalized by the pounds of a.i. handled. The protocol for this IDS study had planned to
normalize the exposures by the concentration (ppm ai) because workers are exposed while
immersing their hands into buckets, sinks, and tanks rather than using a definite amount of
treatment solution. The concentrations of the two a.i.s used for the three IDS scenarios are
reported below.
MEs
Average Measured
Average Measured
IDS Scenario
(Carboy ID for
C14-ADBAC
DDAC
Bucket to pre-mix
Concentration
Concentration
treatment solution)
(ppm)
(ppm)
Bucket &
MEs 1-6
88.7
92.3
Rag/Sponge
(Carboys 1,2,7)
MEs 7-12
171
185
(Carboys 3,6,9)
MEs 13-18
331
420
(Carboys 4,5,11)
3-Compartment
MEs 1-6
20
22.7
Sink
MEs 7-12
124
146
MEs 13-18
198
231
COP Tank
MEs 1-6
19.4
26.7
MEs 7-12
123
143
MEs 13-18
192
228
Notes: For ME4 in the Bucket scenario, t
le treatment solution came from two different
Carboys, the average of the two concentrations was used/reported in Table 4a below. For
the Bucket & Rag scenario the averages in the above table are across the three Carboy
averages and so do not exactly match the averages from Table 4a across the six MEs. For
the COP scenario the averages in the above table are across replicate averages and so do
not exactly match the averages from Table 4c across the six MEs.
• Surface Area Wiped and Equipment Cleaned/Handled: For the bucket scenario, the
subjects wiped surfaces "...such ... as countertops, backsplashes, refrigerators, ice
makers, stoves, tables, and chairs. Each surface to be wiped in the bucket and rag/sponge
scenario was identified and measured beforehand. " (AEATF 2021, page 39). The surfaces
wiped for the MEs in the 20 minute duration group ranged from 107 to 846 ft2. In the 60
minute duration group, the surface area wiped ranged from 569 to 1,635 ft2 (AEATF 2021,
page 107). For the sink scenario, "...a total of 128 items such as mixing bowls, cookie
sheets, round cake pans, cupcake pans, frying pans, spatulas, ladles, gravy boats, loaf
pans, pizza rounds, pots and pans were purchased. " (AEATF 2021, page 35). The MEs in
this scenario were grouped in either the 1- or 2-hour monitoring timeframes. The 9 MEs in
the 1-hour group cleaned 31, 40, 40, 46, 47, 58, 60, 66, and 77 items while the 9 MEs in
the 2-hour group cleaned "not reported", 32, 38, 78, 114, 115, 122, 128, and 218 items.
"Because subjects cleaned at different rates and to ensure that subjects met a minimum
time periodfor this work activity, the ME ended when the target monitoring duration was
reached, rather than basing it on cleaning a set number of items. In two cases, subjects
cleaned at a faster than expected rate and were told to reclean their items so that they
workedfor the target time. The actual number of items cleaned was recordedfor each ME
Page 17 of 41
-------
to give an idea of the work efficiency. " (AEATF 2021, page 35). For the COP scenario, the
food processing-type of equipment cleaned (equipment was placed in a choice of 5
different size wire baskets by the subjects for cleaning in the COP tanks) included: 4
valves, 14 miscellaneous fittings, 22 miscellaneous clamps, 3 end caps (2 inches diameter),
28 gaskets (rubber, 2-3 inches diameter), 5 pipes (14-24 inches long), 1 long pipe (3 feet
long), and 3 hoses (2 to 4 feet long).
• Cleaning (Sanitizing) Procedures . For all three of the IDS scenarios, researchers added
the a.i. to the treatment solution, not the subjects (mixer/loader exposure was monitored
in previous AEATF II studies).
For the bucket scenario, the subjects were " ...shown the selection of buckets, rags, and
sponges and asked to pick the one(s) he/she wanted to use. It was explained to the subject
that he/she could use more than one rag or sponge and that if the water in the bucket got
too dirty, they could get a fresh bucket of sanitizing solution. ... Most subjects worked
continuously for the monitoring period; a few took breaks at their discretion. Most
subjects placed their bucket either on the floor and bent over to immerse the rag/sponge
into the bucket and wring it out or they placed the bucket on a table or chair where they
could access it without bending over. They would pick up and move the bucket as they
movedfrom table to table in the banquet hall and to various places in the kitchen. ...If
the subject finished wiping all the designated surfaces in both rooms, but had not met the
target activity duration (either 20 or 60 minutes), the subject was brought back to the
first room and told to re-wipe the same surfaces. ... The last activities done by the
subjects was to wring out the rag/sponge and pour the used solution down the sink
drain. " (AEATF 2021, page 41-42).
For the sink scenario, the subjects were " ...shown the sink, drying racks, and soiled
cookware and then shown the selection of sponges/souring pads and asked to pick the
one(s) he/she wanted to use. It was explained to the subject that he/she needed to fill each
sink with water to a typical depth (at a minimum half full) and add detergent to the first
sink as they normally would. ... Subjects filled the sinks with water to the desired depths,
adding as much detergent as they wanted to the wash sink, then stepped aside so that the
observer could measure the depth and temperature of each sink. ... Workers placed the
dirty items into the wash sink and then moved them to the rinse sink, and then to the
sanitizing sink after which the items were placed on a drying rack or surface to air dry.
This process was repeated until the target work time was reached. Splashing onto the
subjects, especially in the torso area, was noted with most subjects. The last activity done
by the subjects was to drain the sinks by removing the stoppers by inserting their hand
into each sink to pull out the plug; subjects were allowed to do this in any order they
wanted. " (AEATF 2021, pages 37-38).
For the COP scenario, the subjects placed " ...the various pieces of equipment into the
empty COP tank, placing the smaller parts into one or more wire baskets. Next the
subject turned on the tank water valve andfilled the tank for the simulated wash cycle.
Once the water reached the desired level, the subject turned off the water and turned on
the circulation which operated the jet agitation. At this point, the subject removed his
Page 18 of 41
-------
safety glasses and gloves, placed them on the wire rack, and sat at a table placed on the
far side of the test room. After allowing the water to circulate for approximately 30
minutes for the simulated wash cycle, the subject put his safety gear back on and drained
the tank by flipping a lever on the bottom of the tank which released water out of the
bottom of the tank onto the floor and down the floor drain. Once the water had drained,
the subject conducted a simulated rinse cycle by spraying the items in the tank with water
from a hose. Next the subject filled the tank with water for the sanitization step. When the
water level reached the desired height, the subject turned off the water valve and stepped
aside so that a researcher could add the defoamer and Oasis 146 to the tank. Once this
was complete, the subject turned on the circulation for a few minutes to allow the
solution to mix thoroughly before turning it off and stepping back again so that
researchers could collect the aliquots. Once this was done, the subject turned the
circulation back on, checked to see if everything was working correctly, and then
removed his/her PPE, and sat down at the table again. The subject was required to put
his/her PPE back on and to walk to the tank to check it one time during the
approximately 15-minute sanitizing cycle. Once the time was up, the subject put the PPE
back on and drained the sanitizing solution from the tank by opening the valve at the
bottom of the tank which released the solution onto the floor and down the drain. Once
the tank had been drained, the subject picked up the parts from the tank and placed them
on the wire racks to air dry. This included removing all the small parts from the wire
baskets. Since the parts were wet, it was not uncommon to see diluted sanitizing solution
drip from the parts as they were being moved. Once all the parts had been placed on the
rack, the work activity was completed. For subjects doing one cleaning/sanitizing cycle,
at this point the monitoring event was complete; for those who had to conduct two
cleaning/sanitizing cycles, the entire process was repeated with a second set of clean, dry
equipment parts. Subjects doing a second cycle rinsed the COP tank with a hose after
removing the first set of parts or they added water to the empty tank using the valve at the
bottom on the tank and then drained it. The method used to rinse the tank between cycles
was up to the subject(AEATF 2021, pages 45-46).
• Environmental Conditions: Environmental conditions (humidity and indoor
temperatures) are reported for each individual ME on page 78 of the study report. Indoor
temperatures for the bucket scenario ranged from 65.4 to 79.6 F and the humidity indoors
ranged from 24.1 to 77.8% (AEATF 2021, page 42); for the sink scenario indoor
temperatures ranged from 64.1 to 85 ° F and the humidity indoors ranged from 30.9 to
80.2% (AEATF 2021, page 38); and for the COP scenario indoor temperatures ranged
from 69.7 to 76.8 F and the humidity indoors ranged from 28.9 to 59.7%. The sites for
the bucket and sink scenarios had air flow measurements taken which showed in the
kitchens 12.7, 11.7, and 15.2 air changes per hour (ACH), at sites 1, 2, and 3 respectively.
For the bucket scenario, the banquet halls had air flows of 3.6, 7.8, and 10.4 ACH, at sites
1, 2, and 3, respectively. The air flow direction relative to the subjects at the sinks
indicated no significant sustained air flow based on anemometers and visual observations.
The air flow direction relative to the subjects for the bucket scenario was not meaningful
as the subjects moved around the room as they wiped/cleaned. For the COP scenario,
there was no air flow as there was no vents/open doors.
Page 19 of 41
-------
2.0 Results
2.1 QA/QC
Controls. The non-fortified method validation control samples (blanks) indicated contamination
in the OVS tubes. The study report's Table 30 (AEATF 2021, page 122) shows for CI 4-
ADBAC that all 8 of the OVS tube controls had detectable residues ranging from 10.3 to 47.4
ng/tube. Because of the background contamination, and the need for low LOQ for the IDS
scenario that would potentially lead to low inhalation exposures, the AEATF II decided to
include OVS sampling of DDAC residues (and only the inhalation results for DDAC are reported
herein). For DDAC, 6 of the 8 controls for the OVS tubes also had detectable residues, but at
much lower levels, ranging from 0.08 to 3.4 ng/tube. Finally, the outer dosimeters also showed
background interference with C14-ADBAC and the AEATF II chose to increase the LOQ for
this matrix from 3 to 10 |ig/sample.
The results of the non-fortified field and laboratory control samples (blanks) were as follows: All
the C14-ADBAC and DDAC field control matrix samples were less than the limit of
quantification (LOQ) (AEATF 2021, page 75); the C14-ADBAC laboratory control matrix
samples were also all less than the LOQ (AEATF 2021, page 74). One OVS tube DDAC non-
fortified laboratory control sample had detectable residues of DDAC (12.8 ng compared to the
LOQ of 12 ng).
The C14-ADBAC LOQs (LODs) for the various matrices were: air sampling OVS tubes 100
(30) ng/sample, neck/face wipe 0.25 (0.075) [j,g/sample, forearm wipe 1 (0.3) [j,g/sample, outer
WBD sections 10 (3) [j,g/section, inner WBD sections 3 (0.9) [j,g/section, and handwash 1 (0.3)
[j,g/sample. The DDAC OVS LOQ (LOD) was 12 (3.6) ng/tube (page 58).
Method Validation. In this IDS study, the AEATF II noted that since the analytical methods for
C14-ADBAC in the same sampling matrices have been previously developed and used in many
other AEATF II exposure studies, that the method validation was conducted under non-GLP
(good laboratory practices) (AEATF 2021, pages 73-74). Each of the sampling matrices were
fortified using triplicate samples at a low-, mid- and high-levels (as opposed to the typical 7
samples per fortification level) (AEATF 2021, page 58). The results of the method validation for
C14-ADBAC ranged from 83.0±2.4% (mean ± standard deviation) for the low-level fortification
of the inner dosimeters to 105±6.7% for the high-level fortification of the outer dosimeters
(AEATF 2021, Table 29 page 121).
As discussed above, the OVS tubes showed background levels of C14-ADBAC that would
potentially interfere with the expected low inhalation exposures in the IDS scenarios, and
therefore, the AEATF II also included the monitoring of DDAC specifically for the OVS tubes.
The results of the method validation for DDAC in the OVS tubes at the fortifications of 10, 500,
and 4,000 ng/tube were 73.7%, 91.4%, and 88.4%, respectively (AEATF 2021, Table 30 page
122).
Laboratory Recoveries. The concurrent laboratory recovery samples for C14-ADBAC and
DDAC were fortified at levels expected in the exposure study. Typically, the field recoveries are
Page 20 of 41
-------
used to correct the actual field monitoring samples, however, the field recoveries in this study
were all greater than 100% and therefore the laboratory recoveries were used to correct the field
samples (if the laboratory recoveries <100%). "The number of replicates per fortification level
within a matrix rangedfrom 1 to 20. The overall average recoveries across fortification levels
for each matrix rangedfrom 90.8% (face/neck wipes) to 105% (OVS tubes, C14-ADBAC) ... the
concurrent fortification recoveries within each analytical set were averaged; if that average was
less than 100%, the subject samples (andfieldfortified samples if included in the analytical set)
within that set were adjusted by that average recovery. " (AEATF 2021, page 75) The results of
the laboratory recoveries are provided in AEATF (2021) Tables 31 to 37, pages 123-129.
Field Recoveries. The field recovery samples were transported, stored, and analyzed with the
corresponding field (dosimeter) samples. Results of the field recoveries for C14-ADBAC are
summarized in AEATF (2021) on pages 75-76, Tables 38-41. Since the field recoveries were
>100%), they were not used to correct the actual field exposure samples. The field recoveries for
C14-ADBAC averaged 105±7.66% (n=18) for the hand wash, 103±4.81% (n=18) for the
face/neck wipes, 101±9.56%> (n=18) for the inner dosimeters, and 116±14.7%> (n=9) for the OVS
tubes.
" Unlike the solutions used to fortify other matrices, the field fortification solutions prepared by
the analytical lab to fortify the OVS tubes were made using C14-ADBAC reference standard, not
Oasis 146. For this reason, the fieldfortified OVS tubes could not be used to determine DDAC
recovery. However, DDAC has been used as the test substance in several AEATF II studies
including AEAO2 (Spray and Wipe Study, MRID No. 48375601), AEA03 (Mop Study, MRID No.
48210201), and A EA 05 (Pour Liquid Study, MRID No. 48917401). The stability of DDAC in
OVS tubes under field andfrozen conditions is well documented. " EPA notes the average field
recoveries in the AEATF IF s liquid pour study for all matrices for both C14 ADBAC and DDAC
were roughly 90 to 100%. "To support the longer time that the OVS tubes were in frozen
storage during this study, a new DDAC freezer storage stability study in OVS tubes was
conducted... Tubes were spiked at 123 ngDDAC/tube and 4,930 ng DDAC/tube with Oasis 146
and analyzed on day 0 and at 8 months after fortification. The study showed no loss of DDAC
through 8 months in frozen storage (recoveries ranging from 94.3 to 117%)..." (AEATF 2021,
page 76-77).
2.2 Calculating Unit Exposures
Dermal Unit Exposure. As discussed in the protocol, dermal exposure was measured using
100%) cotton inner and outer whole-body dosimeters (WBD). The inner WBDs were worn
underneath normal work clothing (i.e., long-sleeved shirt and long pants for the COP scenario
and short-sleeved shirt and long pants for the other two IDS scenarios as the subject's arms were
dipped into the buckets and sinks). The normal work clothing worn over the inner WBDs were
also analyzed and reported as outer dosimeters. Dermal exposures monitoring techniques also
included hand washes, face/neck wipes, and forearm wipes for the bucket and sink scenarios
(since those techniques included dipping subject's arms into the treatment solution). The inner
and outer WBDs were sectioned and analyzed by body part (i.e., upper and lower arms, front and
rear torso, and upper and lower legs). Samples were adjusted, as appropriate, according to
recovery results from laboratory fortification samples (i.e., all field recovery results were >100%
Page 21 of 41
-------
so laboratory recovery results were used to correct the actual exposure/field samples where
laboratory recoveries were <100%).
Hand wash removal efficiency studies were previously conducted by the AEATF II and
reviewed by EPA and used in other AEATF II scenarios (e.g., USEPA 2010, USEPA 2012).
These same two studies were used in this study to correct hand and face/neck samples. "A study
to measure the removal efficiency of DDAC (CAS 7173-51-5) from skin using a washing
technique showed an average recovery of 90.3% DDAC (Boatwright 2007). A similar study was
conducted with the structurally related compound alkyl dimethyl benzyl ammonium as the
saccharinate salt (ADBAS; CAS 39387-42-3) using a wipe technique rather than a wash
technique resulting in an average recovery of 89% (Boatwright, 2008). Based on these studies, a
90% dermal removal efficiency correction factor is used to adjust the C14-ADBAC hand-wash
residues while a 89% correction factor is used to adjust the face/neck wipe residues. " (AEATF
2021, pages 62-63)
One final adjustment factor was used for the face/neck samples to correct for the area of the face
covered by the safety glasses. A correction factor of 1.1 (as per AEATF SOP 9K.0, Section
2.2.1) was used to correct the face/neck residue values (AEATF 2021, page 62). Dermal samples
were not adjusted for background levels of C14-ADBAC.
The analyses of residues on the dosimeters worn by each individual subject allow for the
estimation of exposure for various clothing configurations from long- to short-sleeved shirts
(long-sleeved shirts not available for bucket and sink scenarios because the subjects lower arms
were monitored by forearm wipes rather than the WBD) and long- to short-pants. The results of
these various clothing configurations are available in Appendix A. For brevity and usefulness
(i.e., the majority of the dermal exposure is to the hands while the other body parts round-out
when values are reported to 3-significant figures) only the following clothing configurations are
reported in the main body of this review:
(1) "Long-Short Dermal" = long pants, short-sleeved shirt, and no gloves for bucket &
rag/sponge and 3-compartment sink scenarios; and
(2) "Long-Long Dermal" = long pants, long-sleeved shirt, and gloves for the COP
scenario.
Total dermal exposure is calculated by summing exposure across all body parts for each
individual monitored. The following WBD sections are summed to calculate the clothing
configuration of long pants, short-sleeved shirts (Long-Short) plus face/neck wash, forearm
wipes, and hand wash for the bucket and sink scenarios:
• inner upper arms,
• inner front and inner rear torso, and
• inner lower and inner upper legs.
The following WBD sections are summed to calculate the clothing configuration of long pants,
long-sleeved shirts (Long-Long) plus face/neck wash and hand wash for the COP scenario:
• inner lower arms,
Page 22 of 41
-------
• inner upper arms,
• inner front and inner rear torso, and
• inner lower and inner upper legs.
Dermal unit exposures (mg a.i.) are normalized by the product of the concentration of the
treatment solution (ppm a.i.) and the duration of exposure (hour) to yield units of mg/(ppm * hr).
The dermal unit exposure (mg/(ppm*hr)) is calculated by dividing the summed total exposure by
the measured ppm a.i. concentration*measured exposure duration.
Inhalation Exposure. Inhalation exposure was measured using a single personal air sampling
pump. The inhalation sampling consisted of " ...a low-volume, SKC personal air-sampling pump
was attached to the subject's belt or waistband. This was connected to an OSHA Versatile
Sampler (OVS) air-sampling tube containing a glass filter andXAD-2 sorbent (SKC catalog
number 226-30-16). The OVS tube is designed to capture both particulates or aerosols and
vapor to provide total inhalable residue. The tube was attached to the subject's collar in the
subjects' breathing zone... The tube intake was positioned downward to simulate the nasal
passage of the subject. The airflow of each pump was calibrated to a target airflow of
approximately 2.0 liters per minute prior to use and documented. " (AEATF 2021, page 48).
Background levels of DDAC in the OVS tubes were low and thus the background levels were
not used to adjust any of the samples (AEATF 2021, page 74). Based on the high background
levels for C14-ADBAC in the OVS sampling tubes described above, only the DDAC sampling
for inhalation exposures are reported herein.
Inhalation unit exposures for the DDAC OVS sampling tubes (measuring total inhaled residues)
are provided using the two following methods:
(1) Air concentration expressed as an 8-hour time weighted average (TWA) and normalized
by ppm and duration (i.e., (mg/m3)/(ppm x minutes)) is calculated as the air concentration
((mg/m3) / (ppm a.i. x minutes)) * sampling duration (hours/day) / 8 (hours / day).
(2) Inhalation exposure (mg/ppm a.i. x minutes) or dose is calculated as the air concentration
((mg/m3) / (ppm a.i. x minutes) * breathing rate (1 m3/hour) * sampling duration
(hours/day).
2.3 Dermal and Inhalation Exposure Results
Results. A summary of the individual and mean dermal and inhalation results from the three
IDS scenarios are presented in Table 4. Both empirical means and the results of the lognormal
simple random sample means are provided for comparison; the latter being the recommended
values summarized in Table 1. The various clothing configurations for the three IDS scenarios
are provided. Also shown for comparison to the total dermal exposure are the dermal results for
the hand exposures only. These tables report the results for each individual subject along with
empirical and lognormal simple random sampling method statistical summaries.
Appendix A to this memo provides statistical models to estimate the unit exposure summary
statistics, including:
Page 23 of 41
-------
• Empirical simple random sampling model; and
• Lognormal simple random sampling model.
The results of the lognormal simple random sampling model have been selected to best represent
the summary statistics for the unit exposures (for summary results of recommended unit
exposures see Table 1 above). The estimates using substitution of half the LOQ for non-detected
values below the LOQ or below the LOD are recommended. For a detailed discussion of the
lognormal simple random sampling model calculations and results the reader is referred to
Appendix A, which includes quantile plots to compare normal and log-normal distributions for
the unit exposures.
Appendix A also provides various alternative statistical models for estimating the exposure from
the ppm x duration instead of simply using the unit exposures multiplied by the ppm x duration.
The main model is a linear regression model for log exposure against the log of the ppm x
duration. Also included is the HSRB-recommended quadratic regression model regressing log
exposure against log (ppm x duration) and log (ppm x duration) squared. Quantile and
regression plots are used to evaluate the linear regression model. Additional models considered
in Appendix A are log-log-logistic, three-parameter logistic, and gamma regression models
recommended by the HSRB. Of these alternative regression models, the best-fitting models for
most exposure routes are either the linear or gamma models, based on the AIC statistical
criterion. Since the gamma model's AIC scores were not very different from the linear models
and the linear models are much easier to implement, the linear models were selected.
Impact of Non-detects. All the hand sampling results for the three IDS scenarios had detectable
residues. All the forearm sampling results for the bucket and sink scenarios had detectable
residues. The outer lower arm for the WBD sampling had detectable residues for 16 of the 18
MEs for the COP scenario. The neck/face was detectable for most of the MEs (44 of 54 MEs).
The rest of the individual WBD for inner sectioned body parts were mostly below the limit of
quantification (LOQ), while most of the outer dosimeters had detectable residues. Most of the
OVS tubes had detectable residues where only 6, 4, and 0 of the samples were below the LOQ
for the bucket, sink, and COP, respectively. Both dermal and inhalation exposure results were
estimated using various methods of handling non-detects, including V2 the LOQ, substituting of
the non-detects with the midpoint of lowest and highest value, maximum value, minimum value,
and the maximum likelihood method for censored data. Because the dermal exposures are
dominated by the hand exposures, the non-detects had no impact on the dermal unit exposures
for the bucket and sink scenarios and a very minimal impact on the COP scenario. Most
inhalation exposures had detectable residues and the handling of the non-detects had only slight
impact on the results, except in some cases using the maximum or minimum substitution method.
The alternative estimates for handling non-detects (i.e., substituting the maximum and minimum
LOQ values and censored data maximum likelihood (MLE)) are provided in Appendix A (bucket
scenario Table AB10 page 14; sink scenario Table AS10 page 56; COP scenario Table AC10
page 92).
Page 24 of 41
-------
T:ihle4;i. Bucket & U;i«/Spon«e: Summnrv of Dennnl iiml Inhiihilion I nil Kxposnre Kslimsiles.
Monitoring Kvent (Ml.)
ADBAC
Cone
(ppin)
l)l)AC
Cone
(ppm)
Diimlion
(Mill)
Dermsil I nil exposure
(ni«/(ppm ADIJAC - mill))
Inhiihilion S 11 onr
i\\ A I nil Kxposnre
((in«/in')/(ppin l)l)AC
* min))
Long-Short
1 hinds
1
87.28
92.16
22
0.00538
0.00537
3.11E-09
2
87.28
92.16
60
0.00124
0.00123
1.17E-09
3
87.28
92.16
20
0.00116
0.00114
7.61E-09
4
88.45
92.49
60
0.00180
0.00179
4.86E-09
5
89.23
92.03
21
0.00248
0.00248
3.21E-09
6
89.23
92.03
61
0.00050
0.00049
3.35E-09
7
175.06
187.32
21
0.00143
0.00143
1.64E-09
8
175.06
187.32
61
0.00281
0.00281
3.46E-09
9
173.40
187.06
22
0.00156
0.00153
3.24E-08
10
175.06
187.32
60
0.00101
0.00101
2.54E-09
11
164.11
181.61
20
0.00120
0.00120
1.71E-09
12
164.11
181.61
61
0.00124
0.00124
1.78E-09
13
331.05
435.46
20
0.00221
0.00220
1.92E-09
14
331.05
435.46
61
0.00076
0.00076
9.25E-10
15
340.97
436.55
20
0.00134
0.00133
4.57E-09
16
340.97
436.55
61
0.00131
0.00129
5.18E-09
17
320.94
388.95
20
0.00050
0.00049
8.00E-10
18
320.94
388.95
61
0.00095
0.00094
2.23E-09
Empirical Mean
196.75
232.62
40.67
0.00161
0.00160
4.58E-09
Empirical SD
103.87
142.69
20.59
0.00113
0.00113
7.15E-09
Lognormal Simple
Random Sample Mean
199.45
236.41
41.29
0.00160
0.00159
4.12E-09
Lognormal Simple
Random Sample SD
120.12
167.67
24.81
0.00101
0.00102
4.29E-09
Page 25 of 41
-------
Let X; be the ith AaiH or unit exposure value and let Y, = ln(Xi).
18
Empirical Mean = X = ^ X; /18
1=1
Empirical SD = Sx = - xf HI . Suppose X is lognormally distributed, so that Y = ln(X) is normally distributed with a
population mean |i and a population variance g2
Lognormal Simple Random Sample Mean = Estimated population mean of X = Estimate of exp([j, + '/2 g2) = exp( Y + '/2 SY2) where
18 I 18 .
Y = £Y;/18and SY = JZ(Yi "Y) 117 ¦
Lognormal Simple Random Sample SD = Estimated population standard deviation of X = Estimate of
exp(^i + !/2 g2) ^/exp(a2)-l = exp( Y + V2 SY2)^/exp(sY2)-l.
Page 26 of 41
-------
Tsihlc 4h. 3-('omp;irlmen( Sink: Siiiiiinsirv of Dermsil sind 1 nliiihilion I nil Kxposnre Ksliniiiles.
Monitoring Kvenl (Ml.)
ADIJAC
Cone
(ppm)
l)l)AC
Cone
(ppm)
Dnrsilion
(mill)
Dermsil I nil exposure
(ni
-------
Let X; be the ith AaiH or unit exposure value and let Y, = ln(Xi).
18
Empirical Mean = X = ^ X; /18
1=1
Empirical SD = Sx = - xf HI . Suppose X is lognormally distributed, so that Y = ln(X) is normally distributed with a
population mean |i and a population variance g2
Lognormal Simple Random Sample Mean = Estimated population mean of X = Estimate of exp([j, + '/2 g2) = exp( Y + '/2 SY2) where
18 I 18 .
Y = £Y;/18and SY = JZ(Yi "Y) 117 ¦
Lognormal Simple Random Sample SD = Estimated population standard deviation of X = Estimate of
exp(^i + !/2 g2) ^/exp(a2)-l = exp( Y + V2 SY2)^/exp(sY2)-l.
Page 28 of 41
-------
Tsihle 4c. Clesin-oiil-of Place (COP): Suininnrv of Derm:i 1 nml 1 nlinhilion I nil Kxposnre Kslimntes.
Monitoring Kvenl (Ml!)
ADIJAC
Cone
(ppm)
l)l)AC
Cone
(ppm)
Diimlion
(Mill)
Dermsil I nil exposure
(mg/(ppm ADIJAC - mill))
Inhiihilion S lionr
TW A I nil Lxposnre
((nig/nr')/(ppni l)l)AC
mill))
Long-Long
lln lids
(gloves)
1
18.13
23.55
71
5.58E-05
5.22E-05
2.05E-07
2
19.76
26.95
133
2.92E-05
2.58E-05
5.98E-08
3
19.31
26.40
67
1.28E-05
9.23E-06
7.82E-08
4
20.17
27.85
139
1.14E-05
9.24E-06
6.94E-08
5
19.65
30.40
72
2.36E-05
1.01E-05
9.67E-07
6
19.15
25.35
151
2.64E-05
2.19E-05
6.54E-07
7
119.69
145.76
66
5.31E-06
4.03E-06
1.39E-08
8
121.28
144.12
136
1.46E-06
6.16E-07
1.83E-08
9
127.49
135.61
69
1.14E-05
9.73E-06
3.89E-08
10
115.78
136.22
126
4.90E-06
2.21E-06
1.92E-08
11
129.49
148.26
75
4.22E-06
6.18E-07
7.00E-08
12
126.56
147.84
152
6.37E-06
5.08E-06
1.07E-07
13
191.43
229.78
68
1.57E-06
8.71E-07
2.02E-08
14
191.14
228.45
131
8.83E-07
5.52E-07
1.63E-08
15
190.96
228.05
73
1.53E-06
1.51E-06
8.88E-09
16
191.05
227.30
139
2.10E-06
1.80E-06
1.06E-08
17
193.18
230.64
73
4.88E-06
2.90E-06
5.14E-08
18
194.93
228.24
153
2.06E-06
2.98E-07
4.10E-08
Empirical Mean
111.62
132.82
105.22
1.14E-05
8.82E-06
1.36E-07
Empirical SD
73.14
85.24
36.48
1.42E-05
1.31E-05
2.56E-07
Lognormal Simple
Random Sample Mean
130.16
150.07
105.69
1.22E-05
1.05E-05
1.19E-07
Lognormal Simple
Random Sample SD
177.06
181.86
38.99
2.22E-05
2.99E-05
2.58E-07
Page 29 of 41
-------
Let X; be the ith AaiH or unit exposure value and let Y, = ln(Xi).
18
Empirical Mean = X = ^ X; /18
1=1
Empirical SD = Sx = - xf HI . Suppose X is lognormally distributed, so that Y = ln(X) is normally distributed with a
population mean |i and a population variance g2
Lognormal Simple Random Sample Mean = Estimated population mean of X = Estimate of exp([j, + '/2 g2) = exp( Y + '/2 SY2) where
18 I 18 .
Y = £Y;/18and SY = JZ(Yi "Y) 117 ¦
Lognormal Simple Random Sample SD = Estimated population standard deviation of X = Estimate of
exp(^i + !/2 g2) ^/exp(a2)-l = exp( Y + V2 SY2)^/exp(sY2)-l.
Page 30 of 41
-------
2.4 Evaluation of Scenario Benchmark Objective
Benchmark Objective. The data from the study has been analyzed to see if the IDS scenario
meets the AEATF II objective of a relative 3-fold accuracy (i.e., K = 3). These analyses used the
SAS code originally developed by the Agricultural Handler Exposure Task Force (AHETF) and
independently confirmed by the Health Effects Division (HED) (and now modified by the
Antimicrobial Division (AD)). Appendix A (starting page 15) provides the detailed benchmark
analysis which is summarized as follows:
Benchmark Objective: fold Relative Accuracy (fRA)
The benchmark objective for AEATF II scenarios is for select statistics - the geometric mean
(GM), the arithmetic mean (AM), and the 95th percentile (P95) - to be accurate within 3-fold
with 95% confidence (i.e., "fold relative accuracy" also expressed as "K-factor"). EPA has
analyzed the data using various statistical techniques to evaluate this benchmark. First, to
characterize the unit exposures (also referred to as "normalized exposure"), normal and
lognormal quantile plots of dermal and inhalation UEs are provided in Appendix A (bucket
scenario Figures AB1 to AB14 for empirical quantile plots and Figures AB15 to AB21 for
quantile plots for residuals starting on pages 19 and 30; sink scenario Figures AS1 to AS14 for
empirical quantile plots and Figures AS 15 to AS28 for quantile plots for residuals starting on
pages 60 and 69; and COP scenario Figures AC1 to AC14 for empirical quantile plots and
Figures AC 15 to AC28 for quantile plots for residuals starting on pages 96 and 105) to illustrate
that the lognormal distribution is a better fit than the normal distribution for the normalized
exposure (albeit in some cases the difference between the normal and log-normal fit is small).
Overall, these plots support the assumed lognormal distributions for the normalized exposure.
Note: all logarithms defined in this review are natural logarithms.
Next, EPA calculated estimates of the GM, AM and P95 based on two different calculation
methods:
• Empirical estimates; and
• Assuming a lognormal distribution and a simple random sample (SRS).
The 95% confidence limits for each of these estimates were obtained by generating 10,000
parametric bootstrap samples from the fitted lognormal distribution. Then, the fRA for each was
determined as the 95th percentile of the maximum of the two ratios of the sample statistic to the
parameter, after the parameter is replaced by its estimated value. The results of the long pants,
short sleeved shirts, no gloves (Long-Short) for the Bucket and Sink scenarios and the long
pants, long sleeved shirts, gloves (Long-Long) for the COP scenario, as well as the inhalation
exposures for the OVS 8-hr TWA are presented below in Table 5 for the bucket scenario
(Appendix A pages 16 to 18); Table 6 for the sink scenario (Appendix A pages 57 to 60); and
Table 7 for the COP scenario (Appendix A pages 94 to 96). Appendix A also presents fRA
values calculated using a non-parametric bootstrap approach, with generally similar results. The
results indicate that for the dermal unit exposures under consideration, the IDS study meets the
3-fold relative accuracy objective for all but the 95% confidence limit for the COP scenario for
the empirical simple random sampling model (i.e., P95s in Table 7 below). The inhalation unit
exposures meet the 3-fold relative accuracy objective for all but the 95% confidence limit for all
three of the IDS scenarios for the empirical simple random sampling model (i.e., P95s in Tables
5, 6, and 7).
Page 31 of 41
-------
T:ihk*5: Results of Primim Bench 111;
rk An;il\sis lor (lie Bucket K;iii/Snon;ic Scenario.
Dcrniiil l.xnosnrc (l.finii Shorn
liihiiliiliun r.\|)(isui
v (S-lir TWA)
Sliiiislic
I nil l-lxposiirc l-lsliniiilc
(ni/ (|)|)in \ mill))
CI
I'KA
GMS
0.0014
0.001 to
0.0018
1.3
2.85E-9
1.93E-9 to
4.28E-9
1.5
GSDs
1.788
1.47 to 2.18
1.2
2.358
1.772 to
3.155
1.3
GMS = geometric mean assuming SRS = "exp(average of 18 ln(UE)) values"
GSDs = geometric standard deviation assuming SRS = "exp(standard deviation of 18 ln(UE)) values"
AMS
0.0016
0.0012 to
0.0021
1.3
4.578E-9
2.55E-9 to
6.583E-9
1.7
AMu
0.0016
0.0012 to
0.0022
1.3
4.117E-9
2.617E-9 to
6.707E-9
1.6
AMS = average of 18 unit exposures
AMu = arithmetic mean based on GMS =
GMs*exp{0.5*(ln(GSDs)2}
P95s
0.0054
0.0023 to
0.0077
2.2
3.235E-8
6.233E-9 to
3.691E-8
4.7
P95u
0.0035
0.0023 to
0.0053
1.5
1.168E-8
6.278E-9 to
2.158E-8
1.9
P95s= 95th
P95u= 95th
percentile (i.e., estimated as the maximum unit exposure from the 18 unit exposures)
percentile based on GMS = GMS * GSDs 1645
T;ihlc(>: Results of Primim Benchm;
rk Aiiiihsis lor (lie 3-C oni n;i rl men I Sink Sccnnrio.
Dci iiuil l.xnosiirc (l.onti Sliorl)
Inhiiliilion r.xposnrc (X-ln
TWA)
Sliiiislic
I nil l-lxposiirc r.s(ini;i(c
(nili/|)|)in \ mills)
<)5" n ( I
I'KA
I nil l-lxposiirc l-lsliniiilo
((niii/ni ")/ (|)|)in \ mill))
')5"n CI
I'KA
GMS
0.00055
U.UUU45 to
0.00069
1.2
2.709E-9
1.383E-y to
5.469E-9
2.0
GSDs
1.60
1.37 to 1.88
1.2
4.388
2.683 to
7.249
1.6
GMS = geometric mean assuming SRS = "exp(average of 18 ln(UE)) values"
GSDs = geometric standard deviation assuming SRS = "exp(standard deviation of 18 ln(UE)) values"
AMS
0.00061
0.00049 to
0.00078
1.3
6.882E-9
2.836E-9 to
2.097E-8
2.7
AMu
0.00062
0.00049 to
0.00078
1.3
8.086E-9
3.149E-9 to
2.401E-8
2.7
AMS = average of 18 unit exposures
AMu = arithmetic mean based on GMS =
GMs*exp{0.5*(ln(GSDs)2}
P95s
0.00144
0.00085 to
0.00226
1.7
2.877E-8
1.044E-8 to
2.241E-7
5.8
P95u
0.00120
0.00085 to
0.00168
1.4
3.085E-8
1.057E-8 to
8.886E-8
2.9
P95s= 95th
P95u= 95th
percentile (i.e., estimated as the maximum unit exposure from the 18 unit exposures)
percentile based on GMS = GMS * GSDs 1645
Page 32 of 41
-------
liihlc"7: Results olPriniiin Bench in;
rk An;il\sis lor (lie (k'nn-nul-ol'-nkicc (COP) Scenario.
Doi'iiiiil l.xnosiMY (l.nnti Shorn
liihiiliilion Mxposuro (X-ln
TWA)
Sliiiislic
I nil llxposiiiv I'.sliiiiiilo
(m^/ppm \ iniiis)
')5V'ii CI
IRA
I nil l'l\poMiiv 11 stimuli'
((inii/iii V (ppm \ miii))
CI
IRA
GMS
5.91E-6
3.4E-6 to
1.05E-5
1.8
5.01E-8
2.75E-8 to
9.37E-8
1.9
GSDs
3.342
2.24 to 5.03
1.5
3.737
2.41 to 5.85
1.6
GMS = geometric mean assuming SRS = "exp(average of 18 ln(UE)) values"
GSDs = geometric standard deviation assuming SRS = "exp(standard deviation of 18 ln(UE)) values"
AMS
1.14E-5
5.68E-6 to
2.54E-5
2.1
1.361E-7
4.97E-8 to
2.71E-7
2.5
AMu
1.22E-5
6.06E-6 to
2.70E-5
2.1
1.194E-7
5.37E-8 to
2.94E-7
2.3
AMS = average of 18 unit exposures
AMu = arithmetic mean based on GMS =
GMs*exp{0.5*(ln(GSDs)2}
P95s
5.58E-5
1.78E-5 to
2.17E-4
3.5
9.672E-7
1.67E-7 to
2.57E-6
5.2
P95u
4.30E-5
1.80E-5 to
1.02E-4
2.4
4.379E-7
1.69E-7 to
1.124E-6
2.6
P95s= 95th
P95u= 95th
percentile (i.e., estimated as the maximum unit exposure from the 18 unit exposures)
percentile based on GMS = GMS * GSDs 1645
Presumption of Log-log-linearity With Slope 1. EPA evaluated the presumption that the mean
exposure (more precisely, the expected value of the exposure) is a multiple of the concentration
of the treatment solution x the exposure duration (ppm x minutes). In the Governing Document
and in statistical reviews of some previous AEATF II studies, this presumption has been referred
to as "proportionality", but we are now referring to this analysis as a "log-log-linearity" analysis
to clarify that the statistical models do not assume that the exposure is directly proportional to
either the amount of active ingredient handled or in the case of the IDS scenarios the ppm x
minutes. If the log-log-linear model has a slope of 1, then the arithmetic mean exposure will be a
multiple of the ppm x minutes. The statistical test compares the slope of 1 with a slope of 0,
where 0 corresponds to complete independence between exposure and the ppm x minutes.
To evaluate the relationship for this scenario EPA performed regression analysis of
log(exposure) against log(ppm x minutes) to determine if the slope of this log-log-linear model is
not significantly different than 1 - providing support for a "proportional" (an abbreviation for "
log-log-linear with slope 1") relationship - or if the slope is not significantly different than 0 -
providing support for an independent relationship. If the slope is positive, not zero and not 1,
then the arithmetic mean exposure tends to increase with the ppm x minutes but not
proportionally, so that, for example, doubling the ppm x minutes will not tend to double the
exposure. If the slope confidence interval excludes both 1 and 0 but the slope is positive, then
the statistical evidence rejects both proportionality and independence and shows that the
exposure tends to increase with the ppm x minutes but not proportionally. Note: the slope for
the dermal (or inhalation) exposure measures the change in log mg dermal (or inhalation)
exposure for each unit change in log ppm x minutes. A slope of 1 implies that the log of the unit
exposure (mg/(ppm x minutes)) is equal to a constant plus a random error, so that the unit
exposure has the same mean for any ppm x minutes, and thus the mg dermal (or inhalation)
exposure is proportional to the ppm x minutes.
Page 33 of 41
-------
The resulting regression slopes and confidence intervals for the clothing scenarios and 8-hr TWA
inhalation exposures to be used by EPA in our assessments are summarized in Table 8. A more
detailed discussion and table of the slopes along with the other clothing scenarios is presented in
Appendix A (for the Bucket scenario pages 27-29 and Table AB18, for the Sink scenario pages
68-69 and Table AS 18, for the COP scenario pages 104-105 and Table AC 18).
For the bucket scenario, the confidence intervals for the slope exclude 0 and include 1 for both
dermal and the inhalation 8-hr TWA. Thus, the assumption of independence was rejected and the
assumption of log-log-linearity with slope 1 was supported (more precisely, did not reject
proportionality (a slope of one)). Therefore, the "unit exposure" approach for both the dermal
and inhalation for the 8-hr TWA is a reasonable approximation.
For the sink scenario, the confidence intervals for the slope exclude 0 and include 1 for dermal.
Thus, the assumption of independence was rejected and the assumption of log-log-linearity with
slope 1 was supported (more precisely, did not reject proportionality (a slope of one)). Therefore,
the "unit exposure" approach for the dermal is a reasonable approximation. However, for
inhalation 8-hr TWA exposure the slope is negative and the confidence intervals include 0 but
not 1, thus the assumption of independence was supported and the assumption of log-log-
linearity with slope 1 was rejected. The results for inhalation exposure seem to be
counterintuitive.
For the COP scenario, for both the dermal and inhalation 8-hr TWA exposure the confidence
intervals include 0 but not 1, thus the assumption of independence was supported and the
assumption of log-log-linearity with slope 1 was rejected. This suggests that the exposure does
not depend on the normalizing factor.
A secondary objective for EPA is for meeting 80% power for detecting log-log-linearity with a
slope of 1. This objective is met if the widths of the confidence intervals for the slopes are at
most 1.4. This secondary objective was met for all scenarios and so the statistical (post-hoc)
power is greater than 80%.
Tsihle S. 95 Percent Confidence Inlervsils lor (lie Slope of l.og Kxposure (m«) versus Log
ppiii x minutes lor Dernisil sind Inhnlnlion Kxposures.
Scen:irio/Clolhin»
Slope
Confidence
Inlen ill
Con I'idciicc
Inlen ill Width
Appendix A
Bucket
0.711
0.348- 1.074
0.726
Table AB18
(Long-Short, no gloves)
Long Short
Sub mid value
Sink
0.923
0.694- 1.152
0.458
Table AS 18
(Long-Short, no gloves)
Long Short
Sub mid value
COP
0.038
-0.257 to 0.334
0.591
Table AC 18
(Long-Long, gloves)
Long
Sub mid value
Bucket
0.712
0.193 - 1.231
1.038
Table AB18
Inhalation
TWA
(8-hr TWA)
Sub mid value
Sink
-0.174
-0.541 to 0.194
0.735
Table AS 18
Inhalation
TWA
(8-hr TWA)
Sub mid value
Page 34 of 41
-------
COP
0.102
-0.394 to 0.598
0.992
Table AC 18
Inhalation
TWA
(8-hr TWA)
Sub mid value
Figures 7 to 9 show the data and corresponding fitted regression models for the dermal exposure
routes. The data points marked with the symbols "A" are the shorter durations with lower C14
ADBAC concentrations (i.e., 20-minutes for the bucket scenario and 1-hour for both the sink and
COP scenarios) and "B" are the longer durations and higher C14 ADBAC concentrations (i.e., 1-
hour for the bucket scenario and 2-hours for both the sink and COP scenarios). Appendix A also
presents probability plots of the residuals from these fitted regression models (figures for specific
quantile plots and page numbers in Appendix A are referenced above); these probability plots
show that this simple log-log-linear regression model fits reasonably well except for the
inhalation exposure for the COP scenario (Appendix A pages 107-108). Appendix A also
includes the fitted regression models for the inhalation exposure routes (Appendix A pages 34 to
41 for the bucket scenario, pages 73 to 79 for the sink scenario, and pages 109 to 115 for the
COP scenario).
Regression Plot For Long Short Dermal Exposure
Normalized by ug/ml ADBAC * mins
Scenario Bucket
log ug/ml ADBAC * mins
H log Long Short Dermal Predicted Mean
A: Duration 20 mins, B: Duration 60 mins
Figure 7. Bucket Scenario: Regression plot for Long Short Dermal (mg/(ppm x mins))
Page 35 of 41
-------
Regression Plot For Long Short Dermal Exposure
Normalized by ug/ml ADBAC * mills
Scenario Sink
3 -
tS
Q
c5
O
1 -
0 -
log ug/ml ADBAC * mins
| M log Long Short PermaT"
Predicted Mean |
A: Duration 1 hour, B: Duration 2 hours
Figure 8. Sink Scenario: Regression plot for Long Short Dermal (mg/(ppm x mins))
Regression Plot For Long Dermal Exposure
Normalized by ug/ml ADBAC * mins
Scenario COP
-2.0 ¦
-2.5 ¦
-3.0 -
-3.5 ¦
-4.0 -
8 9
log ug/ml ADBAC * mins
| M log Long Dermal
Predicted Mean |
A: Duration 1 hour, B: Duration 2 hours
Figure 9. COP Scenario: Regression plot for Long Long Dermal (mg/(ppm x mins))
Page 36 of 41
-------
3.0 Discussion of Data Generalizations and Limitations
The regulatory need for a generic data base of pesticide handlers for antimicrobial pesticide
products has been discussed previously (SAP 2007). The study design for the three IDS
scenarios incorporated random diversity selection where feasible. Such a study design requires a
discussion of how the data can be generalized and the limitations of the results. The following
items are provided to potential users of these data to characterize the results of this sampling
effort:
(1) The study purposively selected sites in Orlando, FL, and Madison, WI, as the study
locations. This selection criterion, rather than a random selection of sites across the country,
limits to some degree the statistical generalizations of the data. Thus, we cannot determine
whether these results provide unbiased estimates of exposure distributions from using
sanitizers in locations other than Orlando, FL, and Madison, WI, and it is not possible to use
these data to estimate the potential bias or geographic variability. To generalize these results
to the whole country requires an assumption that the exposure distribution for these scenarios
is independent of the geographic location. The statistical limitations of the purposive site
selection are deemed acceptable by the Joint Regulatory Committee (JRC). It is reasonable
to assume that the cleaning routines to wipe hard surfaces and clean (sanitize) restaurant-type
equipment in sinks and tanks in Orlando and Madison are not substantially different than
cleaning the same types of surfaces and equipment throughout the country. Given a limited
set of resources for the overall AEATF II monitoring program, the assumption that
cleaning/sanitizing does not vary geographically was sufficiently reasonable to forgo the
random site selection (of all buildings throughout the country) in favor of spending the
limited resources to monitor additional distinctly different scenarios (e.g., pressure treatment
of wood, trigger pump spray & wipe, painting, hand held spray wands, etc).
(2) The data generated in this study are acceptable to use as surrogate for assessing other
chemicals considered to have low volatility (i.e., vapor pressures less than -1E-4 mmHg @
20°C). This "rule-of-thumb" for the vapor pressure threshold is reviewed by EPA on a case-
by-case basis, particularly for those antimicrobial pesticides with vapor pressures that are
near to this threshold. For example, for those chemicals with vapor pressures of-1E-4
mmHg, EPA reviews the available inhalation toxicity data to see if the toxicity studies were
performed as a gas or with an aerosol.
(3) The small sample size by itself does not create statistical limitations since the confidence
intervals for the summary statistics based on the primary statistical model were reasonably
narrow (meeting better than the 3-fold relative accuracy goal, except in a few instances as
discuss above).
More important is the fact that the original sets of subject participants, locations, and dates
from which the subjects, and sampling dates were chosen were limited and hence might not
be representative of all experienced restaurant/food processing workers (i.e., dishwashers,
banquet servers, hotel/housekeeping, busboys, janitors, caterers, bartenders,
creamery/dairy/food processing plant workers) living in Florida and Wisconsin (e.g., those
that had experience sanitizing surfaces with bucket & rag/sponge, 3-compartment sink
setups, or COP tanks but did not volunteer), buildings (e.g., churches, Elk Lodge, and
manufacturer of COP tanks were selected for this study), and time periods (e.g., summer
versus winter, day versus night, etc.). In other words, the most significant limitation is that
Page 37 of 41
-------
these data were not derived from a fully stratified random sample of MEs even though the
statistical analyses made that assumption. At a minimum this increases the uncertainty of
the estimates (so the calculated confidence intervals are too narrow) and there may also be
some bias (e.g., study participants not in the volunteer pool might be more or less prone to
exposure than the selected group).
(4) In this study/scenario/review we evaluated the presumption of "proportionality" that the
mean exposure is a positive multiple of the concentration of a.i. and duration of exposure
(i.e., the mean exposure is proportional to the ppm a.i. times hours and the exposure tends to
increase with increasing ppm and duration of exposure). Proportionality is evaluated by
testing if the log-log-linear model has a slope of 1. The analyses of log-log-linearity then
shows that dermal and inhalation exposure tends to increase proportionally with ppm a.i.
times hours exposed. Although this proportionality holds true for nearly all of the exposure
scenarios developed throughout the AEATF II's monitoring program, a few scenarios have
not, including the COP scenario within this study. For the long-long dermal exposure in the
COP scenario, the slope was 0.038 and the confidence interval was from -0.257 to 0.334,
which does not include a slope of 1. Several theories may help explain this lack of exposure
trending proportionally with increasing treatment solution concentration and exposure
duration, such as the fact that the subjects wore long-sleeve shirts and additional personal
protective equipment (PPE) in the form of gloves, which may be more relevant to this
task/job. The clothing scenario of long pants and short-sleeved shirts (with gloves, because
all subjects were monitored wearing gloves) showed higher forearm exposures and resulted
in a slope of 0.416 and a confidence interval from -0.066 to 0.898. Although the long pants
short-sleeved shirt clothing scenario fit the unit exposure modeling approach better, EPA did
not choose to use this estimate because workers operating COP tanks typically wear the
long-sleeved shirts and gloves. Another theory is that the duration the equipment soaked in
the tanks was irrelevant to exposure and the normalization factor of concentration is more
relevant. The Supplement to Appendix A provides alternative normalization evaluations (i.e.,
by ppm, page 74, or by "1" to represent unnormalized, page 144). Normalizing by the
treatment solution concentration only for the long-long clothing scenario did not improve the
outcome (when normalized by ppm the slope is -0.019 and the confidence interval is from
-0.333 to 0.295 (Supplement to Appendix A, Table BC18, page 91)). Another explanation is
that exposure is inherently highly variable and the sample size was not large enough to
accurately model the trend in exposure to the normalization variables. EPA did consider
using the data for the COP scenario to estimate the exposure using the fitted log-log-linear
model with the estimated intercept and slope rather than using unit exposures that correspond
to a slope of one. However, EPA has decided to continue using the unit exposure approach
corresponding to a slope of one for the COP scenario as the conservative estimate of
exposure and to be consistent with the surrogate unit exposure approach developed through
the SAP (2007) and the AEATF Governing Document (ACC 2011).
(5) The subjects monitored in this study were professional workers employed in the
restaurant/hotel/food processing plant occupations for a duration from 1 to 30 years for the
bucket scenario, 1 to 40 years for the sink scenario, and 4 months to 30 years for the COP
scenario. The rationales for selecting professionals instead of consumers as test subjects
were discussed in the protocol review (e.g., the 3-compartment sink and COP tank are less
common for consumers). The use of occupational workers as test subjects is representative
of the use pattern based on the equipment (e.g., 3-compartment sink and COP); but
somewhat less known for the bucket & rag/sponge scenario. There is a potential for the unit
exposures for the bucket & rag/sponge to be different than if consumers were selected due to
Page 38 of 41
-------
the subject's greater experience with the task. EPA's regulatory approval process for
sanitizers in the past has been based on the trigger pump spray and wipe; now EPA has the
availability to compare the trigger pump spray and wipe to the results of the bucket and
rag/sponge scenario which will allow for better risk characterization. The duration of
exposure by the worker compared to the consumer will tend to drive the higher daily
exposure towards the worker's but it is unknown if the differences in the unit exposure (and
if the consumer even has a higher unit exposure) outweighs the duration of exposure.
(6) The planned use of C14 ADBAC for the inhalation OVS tubes was interrupted by
background residues of C14 ADBAC. Although the researchers could have increased the
LOQ to above background levels, the anticipated inhalation exposures for the IDS study
were for low air concentrations and a low LOQ would be needed. Therefore, the AEATF II
researchers switched to using DDAC as the surrogate compound for the OVS monitoring.
The switch from C14 ADBAC to DDAC for the OVS tubes because of the background
levels was a sound choice. Unfortunately, the OVS tubes had been fortified with the C14
ADBAC reference standard instead of the formulated product Oasis 146, which contains
both DDAC and C14 ADBAC. This resulted in no OVS field recovery samples for the
inhalation monitors. Although the lack of field recoveries for the OVS tubes results in an
uncertainty in the inhalation monitoring, DDAC has been previously shown to be stable in
OVS tubes in the field and during transport in several other AEATF II monitoring studies.
Furthermore, a new storage stability study was conducted specifically for this study to
account for any losses during sample storage. EPA believes the inhalation results for DDAC
are sufficiently sound to be used in risk assessments but the high background contamination
in the ADBAC OVS tubes were such that they are unusable.
4.0 Conclusions
EPA has reviewed the AEATF II IDS study and concludes that the AEATF II made the
appropriate changes to the protocol proposed by the EPA and HSRB and has properly executed
the study. The protocol deviations that occurred and were properly reported have not adversely
impacted the reliability of these data. The EPA recommends that the inhalation and dermal UEs
generated in this IDS study be used provided the data are used within the boundaries set forth in
this review. The following is a summary of our conclusions:
• The AEATF II data for inhalation and dermal exposures represent reliable data for
assessing sanitizing of hard surfaces and restaurant-type equipment with bucket &
rag/sponge, 3-compartment sinks, and COP tanks. The AEATF II unit exposures
summarized in Table 1 are recommended to be used for regulatory purposes.
• Estimates of the GM, AM, and P95 were shown to be accurate within 3-fold with 95%
confidence (except in only a few instances). At this time, no additional monitoring for
the IDS scenarios is required.
Page 39 of 41
-------
5.0 References
AEATF 2018. Immersion/Dip/Soak Study Design Document: A study for Measurement of
Potential Dermal and Inhalation Exposure During Antimicrobial Applications Involving
Immersion, Dip, and Soak. Antimicrobial Exposure Assessment Task Force II (AEATF II).
Sponsor Study Identification: AEA12. August 2, 2018.
AEATF 2021. A study for Measurement of Potential Dermal and Inhalation Exposure During
Antimicrobial Applications Involving Immersion, Dip, and Soak. Antimicrobial Exposure
Assessment Task Force II (AEATF II). Sponsor Study Identification: AEA12. May 5, 2021.
MRID 51588901
ACC. 2011. American Chemistry Council, Antimicrobial Exposure Assessment Task Force II
(AEATF II) Governing Document for a Multi-year Antimicrobial Chemical Exposure
Monitoring Program. Interim Draft Document. Version 3. July 8,2011.
FDA. 2017. Food and Drug Administration (FDA) 2017 Food Code.
https://www.fda.gov/food/fda-food-code/food-code-2017.
HSRB Report. 2019. EPA-HSRB-19-01, Subject: October 23, 2018, EPA Human Studies
Review Board Meeting Report, from Liza Dawson, PhD, Chair EPA Human Studies Review
Board to Jennifer Orme-Zavaleta, Ph.D., EPA Science Advisor, Office of the Science Advisor,
1200 Pennsylvania Avenue, NW, Washington, DC 20460.
https://www.epa.gov/sites/default/files/2Q19-
02/documents/hhrb final report science aea i _ t• t<)tocol.pdf
SAP. 2007. Memorandum: Transmittal of Meeting Minutes of the FIFRA Scientific Advisory
Panel Meeting Held January 9 - 12, 2007 on the Review of Worker Exposure Assessment
Methods. U.S. Environmental Protection Agency.
USEPA. 2010. Science Review of AEATF II Mop Human Exposure Monitoring. MRID
Numbers 48210201, 48231201, and 48231901. Memorandum from Timothy Leighton (USEPA)
to Nader Elkassabany, PhD, Branch Chief (USEPA), dated October 4, 2010.
USEPA. 2012. Science Review of AEATF II Aerosol Human Exposure Monitoring.
Memorandum from Timothy Leighton (USEPA) to Nader Elkassabany, PhD, Branch Chief
(USEPA), dated January 3, 2012.
USEPA. 2018. Science and Ethics Review of AEATF II Immersion/Dip/Soak Scenario Design
and Protocol for Exposure Monitoring. Memorandum from Timothy Leighton (USEPA) to
Laura Parsons (USEPA), dated September 27, 2018.
Page 40 of 41
-------
Appendix A
Statistical Review of the AEATF II Immersion/Dip/Soak (IDS) Study
And
Supplement to Appendix A
(To be included as a separate electronic file)
Page 41 of 41
------- |