
SAP Review of Proposed Guidelines for Efficacy Testing of Topically Applied Pesticides Used Against Certain Ectoparasitic Pests on Pets (June 11-14, 2019)

FIFRA SAP Peer Review 
EPA Response

GENERAL METHODS

Charge Question 1. 
1) The proposed guideline describes test methods for evaluating the efficacy of a variety of pesticidal pet products to control fleas, ticks, mosquitoes, and biting flies on pets. Please discuss whether, given the objectives and the types of products being evaluated, the test methods described in the following sections are appropriate to evaluate the efficacy based on the pesticide labeling claims made, e.g. relating to repellency, mortality, and residual control:
    i. Fleas (section (j))
    ii. Ticks (section (k))a
    iii. Mosquitoes and biting flies (section (l))
    iv. Insecticidal shampoo products (section (m))
    v. Environmental conditions (section (n))

Panel Recommendations:
1
The Panel concluded that the general approaches used for testing the efficacy of products for ticks are appropriate, though there are concerns about the number of dogs and cats that need to be tested. The Panel recommended that the proposed guidelines should be revised to provide specific recommendations about which tick species should be used for testing on cats and dogs. This may lead to a reduction in the numbers of animals used for testing. Utilization of tick species that reliably feed on these animals can facilitate a) typical tick attachment and feeding behavior during trials, b) reduction in the use of vertebrate subjects by preventing wasted trials, and c) results that reflect real-world exposure scenarios for each animal. The Panel suggested the protocols should recommend using 2 to 3 species for testing from: Rhipicephalus sanguineus, Dermacentor variabilis/D. andersoni/D. occidentalis, Amblyomma americanum, A. maculatum, and Ixodes scapularis for dogs (Koch and Sauer, 1984), and among D. variabilis, A. americanum, and I. scapularis/Ixodes pacificus for cats (Coles and Dryden, 2014).  PG18-19
The Agency acknowledges the SAP comment; however, the required pest species for testing will not be included in the guideline as they will be included in a product performance rule that will publish separately.
2
The SAP expressed concerns about the general approaches proposed for testing the efficacy of products for biting flies. It may not be practical to form and maintain stable colonies for some fly species (e.g., horse fly tests are typically conducted with wild populations and, it is hard to get a uniform age and gender split for black flies). The Panel recommended that EPA clearly specify when laboratory colonies of single species should be used and under which conditions using wild insects or a mixture of species is acceptable.  PG 19
The guideline was updated to reflect that if the fly species desired for testing do not have lab colonies, they may be obtained from wild populations. 
3
The Panel noted that one issue that needs more consideration across all pests is the request to regenerate colonies with wild species every three years. This may introduce pathogens and could substantially change the genetics of the colony, making it difficult to compare a study to historic data and across studies. The Panel also questioned whether a three-year time interval is appropriate and recommended the use of a well-defined strain would be best.  PG 19
The outcrossing recommendations listed in the tick and mosquito sections was deleted. 
4
However, if the Agency decided to provide more guidance on outbreeding, the Panel suggested that the frequency should be based on the life cycle of each pest type and the length of time needed for the colony to stabilize after new genetic material has been added. A frequency based on the number of generations required to replenish a colony rather than a specific number of years may be more appropriate. In the updated European Medicines Agency document (EMEA, 2016), a longer time frame of 6 years is suggested.  PG 19
Based on the response to SAP comment #3, the outbreeding recommendation will no longer be included. 
5
While the Panel recognized the importance of fleas, ticks, and mosquitoes used in studies to be pathogen free, it may not be practical for all species. In the proposed guideline, EPA needs to provide a clear plan on which pathogens should be eliminated and how they should be removed, as well as a verification or certification process to demonstrate that a colony is considered pathogen free. This may be particularly problematic if wild individuals are introduced into a colony as they may introduce new pathogens to the colony.  PG 19
The specification of "pathogen-free" has been revised to "free of vector-borne pathogens" to be more specific.
6
The Panel also noted several issues that need clarification relative to the welfare of the pet animals. The Panel recognized that it may be necessary to sedate dogs and cats for short time periods when they are infested with particularly aggressive biting insects, such as stable flies, but more specifics are needed regarding how that can be accomplished. This is discussed in more detail in the response to Charge Question 18.  PG 19
The guideline currently already states to sedate dogs and cats for mosquito and biting fly tests. Based on comments from the public that some sedations do not diminish pain, the Agency added a statement to include the use of pain relievers.
7
The Panel also noted several issues that need clarification relative to the welfare of the pet animals. Allowing the pet animal to groom is important; however, since grooming can affect the outcome of pest counts, methods to document similar grooming behaviors in control and treatment groups should be provided.  PG 19
The guideline has been revised to include the use of cones/E-collars, however specifying only during the periods between infestation and comb out counts, so that most of the study days the animals can groom. 
8
The Panel also noted several issues that need clarification relative to the welfare of the pet animals. Details of animal welfare considerations from the perspective of housing individual or multiple animals for socialization, particularly for longer studies, should be provided as these can affect infestations and the outcome of pest counts.  PG 19
EPA did not revise the guideline based on this comment because this topic is outlined adequately under section "v. Housing of test animals." 
9
The Panel also identified some specific, detailed recommendations regarding clarification of terms  --  Several terms needed clearer definitions:
The guideline's definition for moribund insects reflects incapacitated arthropods that could be considered either alive or dead. The Panel suggested that moribund insects be held longer (24-48 hours) under observation to determine mortality vs. recovery. When doing so, the conditions under which they are being held should be appropriate for recovery (e.g. relative humidity and temperature).  PG 20
The counts in the guideline are allowed up to 48 hours (72 hours for the first time point). Therefore, to gain a kills or controls claim, the pests should be dead by 48 hours after exposure. No changes to the guideline are necessary.
10
The Panel also identified some specific, detailed recommendations regarding clarification of terms  --  Several terms needed clearer definitions:
The Panel noted that a clear definition of tick engorgement is needed, as there are different stages for engorgement. The time frame of 48 hours is insufficient for full engorgement of ticks to occur and engorgement to repletion is not the appropriate endpoint to use. Clarification is also needed as to which tick species is being used for testing. This is discussed in more detail in response to Charge Question 14.  PG 20
Based on information and comments provided by the SAP and public, the Agency removed the term "engorged" in the guideline and the endpoint of "engorgement" was changed to "live ticks on treated animal" vs. "live ticks on control animal" by the specified time points. This is also discussed in response to question 14. 

With respect to adding the tick species required for testing, this question was responded to in comment #1.
11
The Panel also identified some specific, detailed recommendations regarding clarification of terms  --  Several terms needed clearer definitions:
The Panel also noted that a clear definition of what is considered to be an adequately infested vertebrate animal is needed; including, the number of pests that should be added and whether retention rates should be based on the group mean count or counts per individual vertebrate animal (i.e., dog or cat). There needs to be clarification as to which species of fleas should be considered. Some designations are too broad.  PG 20
The guideline was clarified to include that retention rates of individual animals will be used to qualify an animal into the study, however, during the study, retention rates are based on group means. 

In regard to the "clarification as to which species of fleas should be considered" the guideline already states that Ctenocephalides felis (cat fleas) are the species to be used. 
12
The Panel also identified some specific, detailed recommendations regarding clarification of terms  --  Several terms needed clearer definitions:
The Panel further noted that a clear definition of what is considered to be an untreated or negative control should be provided. A potential definition of an untreated control group would be those vertebrate animals which receive no treatment that will affect the number or status of pests with which they have been infested. This definition of untreated control would not fit for pets that have been shampooed (or water washed) with non-insecticide shampoo since, at least the process of shampooing will affect the number of pests attached. Perhaps that could be referred to as a vehicle control.  PG 20
The definition of a negative control was included in section (d)(1)viii.b of the guideline. However, it was also added to the definition section, with the revision of deleting "except for studies testing shampoo products."

13
The Panel also identified some specific, detailed recommendations regarding clarification of terms  --  Several terms needed clearer definitions:
The Panel also recommended that simple definitions for the terms repel, knock down, and kill, should be provided. The Panel suggested that, where appropriate, the European Medicines Agency definitions (EMEA, 2016) be used.  PG 20
Definitions have been added or clarified as suggested. 
14
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The Panel concluded that the counting of dead fleas is problematic, and it is better to count live fleas and utilize the difference between control and treatment groups. This is discussed in more detail in the response to Charge Question 22.  PG 20
Currently, for general mortality tests for fleas, the endpoint is counting live fleas on the treated animals vs. the control animals.  

Per SAP recommendations for "speed of kills claims," the guideline has been revised and will no longer recommend counting dead fleas. The guideline will be revised to align with the EMA guideline, which allows testing for claims in less than 48 hours using the same methods already provided at the same thresholds. 
15
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The guidelines should clarify whether the group means or counts on individual animals meet the minimum retention rate (adequacy of infestation).  PG 20
The guideline has been updated to clarify that individual pest counts should be used to qualify each animal to the study. However, once the study has begun, the group means are used to identify the minimum retention rates in the control groups. 
16
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The Panel recommended that the level of variation in the ratio of male to female pests used for animal infestations that is acceptable be presented. For some species of fleas, the Panel noted, it may not be possible to obtain an exact 50:50 split. A +- 10% variation seems appropriate.  PG 20
Since determining the sex of fleas is difficult and unnecessary for testing, the sex ratio recommendation has been deleted. This revision is consistent with the European guideline. 
17
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The guidelines should clarify the timing of when infestations should occur, how to classify the feeding stage of ticks that are removed, and at what time point removal should take place. This may vary based on the goals of an individual study. This is discussed in more detail in the response to Charge Questions 14 and 15.  PG 21
The timing of the infestations and pest removals are all outlined in the guideline per the pest or endpoint desired. 

The guideline was revised by deleting the "engorged" tick endpoint per SAP and public input and changing the end point to "live ticks on treated" vs. "live ticks on control." 
18
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The guidelines should state that the arthropods should be starved or unfed before infestation, though mosquitoes cannot be starved for too long (maybe < 24 hours). Mosquitoes could be provided sugar water to keep them alive until they are used in test trials.  PG 21
Based on input from the SAP and citations, the guideline was revised to:

"Mosquitoes should not be blood fed prior to testing but can be provided sugar water up until trials. However, 4-24 hours of sugar water starvation is recommended (Franc and Bouhsira 2009, Tiawsirisup et al. 2007)." 

"Biting flies should be starved with justification for the amount of time chosen."

19
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The Panel agreed that combing is the best way to determine flea counts. Thumb counts were not considered to be acceptable. Also, comb counts may not be the best way to fully count and remove ticks, since forceps are typically used during the tick removal process.  PG 21
Based on SAP and public comments, hand counting options for fleas was removed from the guideline since they are considered unreliable. Hand counting options will remain for ticks, which move differently than fleas and are therefore easier to hand count.

The guideline was revised to allow comb counts for fleas at earlier time points.  The fleas removed by comb counts should be counted and then replaced onto the animal with the stipulation that the technician avoid the treated area (so not to mechanically spread the solution). Controls should be handled in the same manner. 
20
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
To identify and quantify the amount of blood taken up by fleas and ticks, the Panel proposed using qPCR as a tool (Fourie et al., 2014).  PG 21
The current end point for mortality testing for both fleas and ticks and the revised endpoint for repellency for ticks is "live pests on treated animal" vs. "live pests on control animals." Therefore, the previously stated endpoint of blood feeding or engorgement for ticks was removed as the SAP pointed out that it was impractical, and ticks would not have been engorged (and possibly not even attached) by 48 hours. There is no blood feeding endpoint for fleas. 
21
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
For tick mortality testing, the placement of ticks "over the entire vertebrate animal" (as worded in the guideline) suggests that they may attach over the entire surface of the animal. Regardless of placement location, ticks will select preferred attachment sites on the vertebrate. Ticks tend to congregate in groups on the vertebrate in locations where they are harder to remove by grooming behaviors of the vertebrate (e.g., ears, head, back, between toes, and axilla for R. sanguineus) (Dantas-Torres and Otranto, 2011). The guidelines should instead suggest that ticks are placed on animals away from treatment sites and allowed to freely disperse for attachment.  PG 21
The guideline was revised to "ticks should be placed on the host animals' bodies away from treatment sites." 
22
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The guidelines do not clearly define when the endpoint of the trials and removal of attached live/dead and unattached ticks should occur. Although the timeframes for checking are mentioned in the proposed guidelines, tick removal is not specified. The Panel recommended stating that tick checks and removal should occur at 48 hours post-infestation or post-treatment.  PG 21
The guideline was revised to clarify that at the final comb count for an infestation period, all ticks should be combed out and removed. 
23
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
Regarding use of laboratory colonies of ticks for trials, the Panel agreed with the statement that "similarly aged" ticks should be used for infestations, and perhaps the importance of this should be more clearly emphasized in the guidelines. The age of ticks used in studies could dramatically affect mortality/repellency trial outcomes. Cohorts of adult ticks used in trials should be the same age post-molt to the adult life stage, within a range of time in which vigor is evident, to avoid age-related deterioration effects on trial outcomes. Physiological age of ticks can affect attachment, feeding, and susceptibility of ticks to pesticides (Heller-Haupt and Varma, 1982; Rupes et al., 1972, 1977; Uspensky and Ioffe-Uspensky, 2006).  PG 21
Based on the SAP comment, the guideline has been revised to state that, "Ticks should be of similar age post-molt to the adult life stage."
24
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The effects of the pesticidal active agents on pet animals can vary, e.g., small breed dogs may be more adversely affected than larger dogs. The Panel expressed concern that the weight/dose ranges may be too wide for some products (i.e., smaller animals get more active ingredient per pound). Attention should be paid to label claims to protect the animals, though the Panel recognized that the lowest labeled dose across all weight classes is being considered. This is discussed in more detail in response to Charge Question 8.  PG 21
The dosing used in the studies are based on mg/kg of the animal for efficacy. Companion animal safety studies are also required for pesticide registration but are not relevant for this efficacy guideline. No changes to the guideline are necessary. 
25
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The guidelines proposed: "Animals should not be treated with any substance (e.g., pesticidal shampoo) that could affect the results of the study for at least 3 months prior to initiating the study." The Panel suggested that the following wording from the EMEA 14 July 2016 guidelines be used instead: "It should be ensured that included animals have not been treated with an ectoparasitic substance within a timeframe that might impact the study outcome." The time period after a test animal is treated should reflect the products' actual residual effectiveness on an animal rather than using a standard 3-month period. The Panel also suggested that the same restrictions should be made for other animals that have been housed with the test animals.  PG 22
The guideline has been revised to: "Animals should not be treated with any substance (e.g., pesticidal shampoo) that could affect the results of the study for at least 3 months prior to initiating the study, or for 30 days after the longest claimed period for products with efficacy claims lasting longer than 3 months. One exception is for products that are known to have very short acting periods and no residual efficacy, such as pills that kill fleas on pets within 24 hours. Products such as these may be given at least 1 week prior to the initiation of testing to be sure they have been cleared completely from the animals' bodies."



26
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --  Procedural Concerns
The Panel noted that it is important to consider the vertebrate animal's welfare when exposing a dog or cat to multiple pests. This is discussed in more detail in the response to Charge Question 7.  PG 22
No specific recommendation was provided by the SAP here; the overall issue will be responded to under Charge Question 7.
27
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures  --    
Procedural Concerns
The Panel recommended more detail be provided in the guidelines on the use of sedatives, i.e., the type of sedative (provide a suggested list of approved sedatives), dose by animal type-weight, frequency that it can be repeated on a single animal, and whether the pest will also be sedated. This is discussed in more detail in the response to Charge Question 18.  PG 22
Based on SAP and public comments, the following statement has been added to the guideline: "For infestations using painful or aggressively biting mosquitoes or biting flies, the animals should be administered pain medications or a drug that has proven analgesic and sedative properties." 

28
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures (including new protocols and emerging techniques)  --    
Concerns with New Protocols and Emerging Techniques
The Panel recognized that the goal of these guidelines is to provide appropriate protocols that are deemed acceptable for regulatory purposes based on the published literature and on previous studies. However, due to the extended time period that guidelines are often in place before being modified, a statement on how a new protocol and/or technique may be implemented beyond saying that it will be reviewed by the Agency would be useful. In particular, some specifics to judge if new approach methodologies (NAMs) can be used to satisfy regulatory requirements and how they can be submitted by organizations or manufacturers would be helpful. The Panel suggested that the Agency encourage product researchers to run non-vertebrate (in vitro) tests along with the approved traditional animal testing methods (in vivo) to determine if the same outcomes are obtained, to develop a database of acceptable in vitro assays. For example, if product testing on pests exposed to treated animal hair in vitro demonstrates similar results to in vivo tests in repellency assays, these may be considered as useful surrogates to reduce vertebrate animal testing for certain product applications (Stanneck et al., 2012a).  PG 22
EPA has included a statement in the guideline indicating that protocols may be submitted for new approach methodologies (NAMs), and that protocols proposing new approach methods, especially those methods which do not rely on utilizing vertebrate animals for testing, should be submitted to EPA for review before the study begins.
29
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures (including new protocols and emerging techniques)  --    
Concerns with New Protocols and Emerging Techniques
Furthermore, as the EPA finalizes the proposed guidelines, the Panel recommended that the different types of tests be defined more clearly. Specifically, the Panel suggested the term non-vertebrate animal testing (NAT, NVT, or NVAT) for tests that do not use the host be substituted for in vitro as this term has different meanings in different disciplines (see also the response to Charge Question #10).  PG 22
The term "in vivo" has been revised to specify testing "utilizing vertebrate animals" in the guideline. 
30
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures (including new protocols and emerging techniques)  --    
Concerns with New Protocols and Emerging Techniques
The database should include an understanding of the biological basis for selected in vitro tests and the conditions under which they are conducted, so that future users would have the same premise to follow. The issue of non-vertebrate testing is discussed in more detail in response to Charge Question 2.  PG 22
EPA is not including a database at this time but will consider this as in-vitro methods are developed. 
31
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures (including new protocols and emerging techniques)  --    
Concerns with New Protocols and Emerging Techniques
Some Panel members suggested that field tests should be considered in judging the efficacy of applied pesticides. These have strengths in that they are potentially more representative of real-world conditions as they could include a greater diversity of pet animals and pests and may have reduced requirements for housing of animals. However, they may require more animals and veterinarians on site. The approaches described in EMEA 2016 guidelines could be used as a starting script if field testing is acceptable to the EPA.  PG 22
EPA acknowledges that field tests have strengths but will not include field testing for these products in the guideline as they are conducted with client-owned pets and not ideal for testing as previously submitted field studies have not yielded useful data. Issues include inadequate or inconsistent pest pressure, lack of appropriate controls, and inaccurate pest counting by owners. However, protocols for field studies may be submitted if desired. 
32
The Panel also identified some specific, detailed recommendations regarding clarification of .... procedures (including new protocols and emerging techniques)  --    
Concerns with New Protocols and Emerging Techniques
Some Panel members suggested that an unbalanced study design be considered when statistically appropriate: i.e., fewer animals in the control group compared to the number of animals in the treatment group, in order to reduce the overall number of animals tested. This could be appropriate when previous data have shown that the variance in the number of pests that remain on the control animals is small.  PG 22-23
EPA added a sentence in the protocol development section to indicate that novel methods to reduce animal numbers may be considered on a case by case basis. Reducing the number of control animals is a scenario that could be considered; however, at this time EPA does not have evidence to support a recommendation for a study design with a reduced number of control animals.

Public Recommendations (docket ID # in parentheses):
33
"Recommend adding definitions for "negative control treatment" and "placebo control"." (0008)
The definition of a negative control was included in section (d)(1)viii.b of the guideline. However, it will also be added to the definition section, with the revision of deleting "except for studies testing shampoo products."
A definition of placebo control is no longer necessary as the recommendation to use a placebo will be removed from the guideline. 
34
"Challenge/infestation/exposure (c)(1)  -  suggest adding definition to include "or close to" (the most important being the successful infestation in the control group) as long as it is representative of what happens in nature. For example, most ticks will infest dogs by climbing up from the ground or grass. If ticks are applied on the dog, there is a risk to place the pest where the product was applied." (0008)
For mortality testing, the guideline has been revised to indicate that fleas and ticks should be placed on the host animals' bodies away from treatment sites.

For repellency testing, the guideline has been revised to "...placed inside a cage but not on the animal" to test if the tick moves onto the pet. 
35
"Comb out (c)(2) statement on removal should include live, moribund and dead" (0008)
The guideline has been revised adding "moribund" to this statement. 
36
"Control/residual (c)(3) states that treated animals should be tested at least 24 hours after application. Timing should match the proposed registrant's desired label claim (i.e., if the target label claim is efficacy after 8 hrs., testing at that time interval should be acceptable). Suggest removing time restriction." (0008)
The guideline is defining here what a residual product is (one that works longer than 24 hours). Time points shorter than that would be for "kills" claims, but not "controls" claims. No change to the guideline is necessary. 
37
"Hand count (c)(5) The statement "Live pests are not removed during a hand count" is true for ticks, not for fleas. For example. with studies on fleas, the parasite is counted and then replaced on the animal. Recommend removing or adding "generally not removed."" (0008)
The guideline was revised to remove the hand counting option for fleas and therefore this comment will not be addressed. 
38
"Knocked down (c)(7) Unclear how this definition differs from moribund. Recommend adding additional detail to clearly differentiate between both terms.... Moribund (c)(9) Unclear how this definition differs from Knocked down. Recommend adding additional detail to clearly differentiate between both terms." (0008)
The guideline was revised to define the term knockdown in terms of "the action of rendering an arthropod incapable of coordinated movement or unable to right itself following exposure to a pesticide product." All other uses have been deleted. 
39

"Lowest labeled dose (c)(8) The definition appears to be overly prescriptive for guidelines. There is concern that there will be limitations upon the final packaging (pipette vs. syringe) and that the study will not evaluate the full product. What about our actual claims, will it be revised (lost) if we do have min dose studies for the relevant parasites? Recommend revising to address these concerns." (0008)
The definition for the lowest labeled dose is to determine and test the most conservative dosing on the product label and apply that dose to all animals in treatment group. It will have no bearing on the final packaging or application method (pipette vs. syringe) or the formulation. Claims will be based on efficacy achieved with the lowest labeled dose, which is how efficacy testing is conducted for all product types. No change to the guideline is necessary. 
40
"Should other companion animals such as rabbits and ferrets be considered under this guideline as well?" (0026)
Other companion animals such as rabbits and ferrets are rarely tested for efficacy claims. If a company would like to add them to a label, they may discuss with the Agency or submit a protocol. No changes to the guideline are necessary. 
41
"Specific values for efficacy were not provided (e.g. >=90% corrected treatment mortality). We would like for efficacy thresholds to be included in the document" (0008)
Specific thresholds for efficacy will be included in the upcoming product performance rule. 
42
"i Test materials and treatments :: Ambient temperature and humidity are not defined. Is there an acceptable temperature range?" (0008)
The comment seems to be in reference to the following statement in the guideline: "Test materials should be stored at ambient temperature and humidity for at least one day before use." Since the study conductor will have the information for and be responsible for how to handle the product, it is not necessary to add that into the guideline. 
43
"Housing of test animals. :: Group housing of animals of the same treatment groups during periods of non-infestation should not be a requirement. Cross-contamination could occur for spot-on or collar applications. Further, collars could be removed and/or consumed by other animals when group housed in a laboratory situation. In addition to effects on treatments, compatibility issues could develop between the animals within the housing group (e.g., fighting, antisocial behavior) presenting a concern for animal welfare. Group housing also allows bias into the study design. Standard environmental enrichment programs provide consideration to animal welfare aspects related to individual housing." (0008)
The guideline indicates that animals "may" be housed together and is therefore at the discretion of the study director. No changes are necessary for the guideline. 
44
"v. Selection and allocation of animals :: Body weight could also be considered for allocation (before pest attractiveness), especially if using a prescribed label dose." (0008)
Since the dosing for testing of products (with the exception of pet collars) is on a mg/kg basis and exact for each dog's weight, there is no need to consider body weights for allocation. Pest attractiveness is the most relevant factor. No change to the guideline is necessary.
45
"vii. Test organisms. If pests are to be certified as free of disease-causing pathogens, it should be outlined as to by whom and what certification is acceptable. Without these specifications, the requirement should be omitted. Certification of pathogen free colonies is a difficult unit to measure (e.g., who does it, how often, when, what does the certification look like, how long is the certification valid). Is EPA aware of any instance where established colonies of arthropods have been diagnosed with a pathogen? Certification is an unnecessary hardship with little apparent benefit. Recommend removing requirement." (0008)
The recommendation for pests to be "certified disease-free" was removed from the guideline. 
46
"viii. Representative sampling. c. Positive control  -  Recommend changing "should not be used" to "is not recommended". There may be cases where a positive control is necessary" (0008)
A positive control is not warranted for any of the tests, however the guideline will be revised to read "a positive control is not recommended."
47
Re: disease free: "Does this refer only to zoonotic pathogens? How should strains be "certified?" Other regulatory agencies, such as FDA Center for Veterinary Medicine, do not require certification. Such a requirement may be excessive and impractical in terms of cost for some CROs. Additionally, AHI recommends wild-caught ticks be allowed to be used in studies. It will not be possible to certify wild-caught ticks... as wild-caught ticks more closely represent current circulating parasites... Also, some parasites such as Ixodes scapularis has a life cycle of 2 years. How would one get them to be parasite free?" (0026)
Based on numerous SAP and public comments, the specification of "pathogen-free" has been revised to "free of vector-borne pathogens" to be more specific.
48
"AHI recommends that the decision to use untreated controls vs placebo controls should be flexible and made based on the type of product, where the study is conducted, study design, etc. Ideally the guideline would allow for either." (0026)
Variation of the study designs described in the guideline may be warranted based on the reasons presented by the commenter; it is recommended that before such an experiment commences, the Agency be contacted to discuss the protocol deviations or that a  protocol be submitted for approval. No changes to the guideline are necessary. 
49
"Does EPA want to review all efficacy protocols or only those associated with novel testing methods and when testing new species?" (0026)
The guidelines are being published to outline established testing methods for certain typical tests and claims. If a company would like to run different types of tests or use novel testing methods, they have the option to submit a protocol to gain written approval or comments, if desired, but submitting protocols prior to testing are not typically required. 
50
"Amendments typically cover any change to an activity or task that has not started yet, rather than the initiation of the study itself. (As an example, tick counts on Day 14 could be amended between the start of the study and before Day 14. Documenting such changes by amendment is important as it indicates that the change was planned prior to occurrence rather than as a deviation, which implies that the change was not planned but occurred without being planned" (0026)
All amendments and deviations to the study are to be documented and included in the final report. 
51
"Test animals. A general requirement that animals should not be treated with any substance that could affect the results of the study for at least 3 months prior to initiating the study is excessive. Standard hygiene and husbandry practices following a study include bathing and cleaning the environment where the previously treated animals were housed. The "quarantine" period should be changed to 30 days to allow any animal that might be still be contaminated to be identified during the prequalification challenge. Prequalification will show whether an animal still has remnants of a previous treatment. It is the responsibility of the registrant, with due knowledge of their active and formulation, to require a longer wash-out period if required." (0008)

 See response to comment #25 above. 
52
Re: Section (g)): "Is this level of detail needed in the final report? It can be documented in the raw data but details of how the parasites are reared and maintained would be cumbersome in a final report." (0026)

Yes, these basic items should be included in the final report as they are important information to assess the validity of the test and are not considered cumbersome as they are already included in many studies submitted already. No changes to the guideline are necessary. 

53
"Why are animal body weights required post study completion? Change in body weight is not typically a study variable for an ectoparasite efficacy study." (0026)

It is standard to include some basic health evaluations in efficacy studies such as body weight pre- and post-study. No change to guideline is necessary.
54
Re: Section (g)): "Does this imply that the study protocol and amendments and deviations need to be appended to the final study report? As other agencies do not require this, it will be cumbersome to do this. It would be more flexible and effective to include these documents in the submission as stand-alone documents instead of incorporated into the final report." (0026)

Yes, all amendments and deviations should be appended in the final study report and should not be in a standalone document. This would ensure that all parts of the study would be easily associated with the study and trackable. This is not considered cumbersome and is typically currently done. 
55
Re: field tests: "Is this section referring to, for example, using a European field trial to support a US-EPA registration? Does the EPA have statutory authority to approve a field trial in companion animals? If yes, is the review of the protocol going to be under a PRIA action?" (0026)
It is unclear what specific statement in the guideline this comment is referring to as there are no field test sections in the guideline. The only discussion about field testing is: 

(2) laboratory tesing "...Field tests are not a preferred method to test efficacy of pet products but on occasion can be used to supplement laboratory data or to support specific claims; field studies should have protocols submitted to the agency for approval prior to commencing the experiment."

If field tests are submitted, they should be submitted in a similar manner of all other studies for review. 
56
Re: not treating with any substance: "This statement may be overly restrictive. The prohibition is likely not necessary provided that the pre-treatment infestation used for allocation reveals adequate retention rate(s). For example, this statement would preclude the use of animals that had been treated with nitenpyram to clear fleas; this product has a label claim for only 24 hours and would be cleared from animal's system within 1 week. On the other hand, as part of inclusion criteria, animals should not have been treated with any substance that could affect the results for at least 3 months (90 Days). This is important as we now have long acting products in the market." (0026)
 See response to comment #25.
57
"Application method and exposure to fleas. The proposed 50:50 ratio of male to female fleas is unrealistic, as this would be extremely time consuming and subject the fleas to additional handling stress. The normal sex ratio of a colony should be used." (0008)
The proposed 50:50 ratio of male to female fleas was deleted from the guideline, allowing the normal sex ratio of a colony to be used instead. This is consistent with the European guideline.
58
"In addition, the proposal to have them "evenly distributed over the host animal" is unrealistic. The fleas should be infested onto the animal in a manner to avoid direct exposure to the treatment site...  As in the flea section, the proposal to have them "evenly distributed over the host animal" is unrealistic. Ticks should be infested onto the animal in a manner to avoid direct exposure to the treatment site." (0008)
Neither the flea nor tick sections were worded this way. However, the guideline has been revised to indicate that ticks should be placed on the host animals' bodies away from treatment sites.
59
"Also, 60% of fleas emerging is high. Consider changing to 50% or less. This will be corrected for by using Abbot's." (0008)
The 60% emergence will be maintained since >60% emergence is typically observed in submitted studies (most are even >70%). Any lower than this would indicate another issue with the study. 
 
60
"Hand counts are not appropriate for fleas and would result in inaccurate data due to the mobility of adult fleas through the haircoat and the fact that they are not going to be attached like ticks." (0008)
The option of hand counting fleas was deleted from the guideline based on SAP and public comments. 
61
"IGRs There is no need to separate these into two groups, even when both endpoints are being measured. Larvae that eclose can be counted, recorded and transferred to larval rearing media from which adult emergence can then be counted and recorded. In some cases, 50 or even 25 eggs can be difficult to obtain because some IGR's can inhibit egg production in exposed female fleas" (0008)
The guideline is already worded to allow either way, depending on if both endpoints are to be tested and how the test will be set up. The guideline reads: "Collected eggs can be divided into two groups to determine both larval hatch and adult emergence rates or can be combined into a single group if only one of the endpoints is being measured. At least 50 eggs total per animal should be collected so that both the larval hatch group and the adult emergence group each have a minimum of 25 eggs per animal in both the treated and control group at each time point for statistical comparisons between treated and control animals." 

62
"In both the flea and tick sections I dont see the scientific need of change in preforming dead tray counts. None of the calculations presented use this collected data... 2nd you need to more carefully consider the need for collecting the dead flea and tick data. Specifically the dead pan counts. Because you can't guarantee collection of all the dead this data will be  inaccurate. Also currently I don't see the dead data is actually not used in any of the calculations. So why do the extra unnecessary work." (0011, 0032)
Dead pest counts and tray counts have been removed from the guideline for all test methods. See response to comment #14 on the changes to "speed of kill" claim testing, which previously used dead pest numbers in calculations. 
63
"In section j(4)(i) I was surprised to once again see reference to hand counts for short time points. My understanding from the past was that hand counts were deemed to inaccurate for fleas. In my lab we have been successfully using comb and replace count for several years." (0011)
Hand counts for fleas have been removed based on SAP and public comments. 
64
"In section j(4)(ii) for IGR hatch rates counting at 96 hours after collections is appropriate, however counting again at day 7 could be problematic. The hatched larvae will cannibalize each other within that time without a food source provided. Adding a food source or moving them on to rearing media greatly complicates the counting process. Again, on the adult emergence count doing two counts (35&42) is unnecessary. Procedure in my lab during the day 35 count includes dissecting any intact pupae and any normal fully formed fleas are counted as emerged." (0011)
We have re-reviewed the addition of a second egg count. Based on the literature, it seems in most situations that even if egg hatch was delayed by the IGR exposure, they would not mature, therefore the second count would not be necessary, and it was deleted.
65
"Treating with only the IGR component of the formulation is an interesting concept and this option is appreciated. However, could this be included as an option rather than a requirement? Other regulatory agencies may not accept this which could lead to repeating studies for different geographies." (0026)
The guideline has been revised to clarify that in evaluating efficacy in relation to some IGR claims, it may be possible to use the  end-use product in a study; however, EPA expects obtaining sufficient numbers of viable eggs to assess larval emergence and support a claim like prevents emergence of biting adults is likely to be difficult in studies conducted on a formulation containing an effective adulticide component and in these cases, the IGR only component should be tested.  
66
"Why is computer-generated randomization mandatory? Other manual methods would work equally well." (0026)
The "computer-generated" part of the randomization statement has been deleted.
67
"Fleas are not typically sexed prior to infestation onto animals. Sexing fleas requires anesthetizing the fleas and examining them under a microscope, which can damage the fleas and decrease their viability. As both sexes infest and feed on the Guidance published by the European Medicines Agency (EMEA) is 50% retention rate for fleas on dog and cats. AHI requests EPA explain the rationale for the proposed 40/60 split." (0026)
The guideline previously indicated that a 50:50 ratio of male to female fleas be used (not a 60:40), however the guideline has been revised to delete that recommendation based on SAP and public comments. 
68
"Since some a.i. (active ingredients) take time to spread, registrants can now measure efficacy at 72 h for earlier time points and 48 h for later time points. These are different standards at different time points. AHI recommends EPA consider offering greater flexibility for when the last test count should be recorded to account for the variability in a.i." (0026)
If a product is meant to last <30 days and it takes longer than 48 hours to kill/control/repel a pest (after the initial 3 days allowed to spread and start working), then efficacy of the product is not considered adequate for protecting animals from these pests. No change to the guideline is necessary.
69
Re: ticks. "Test species. Outcrossing laboratory colonies with field populations every 6 years is not appropriate. The idea behind having an established laboratory colony is to have a "reference" colony to which things can be compared back to over long periods of time. The outcrossing of field populations has the potential to introduce pathogens which would deem the colony unacceptable for use. Additionally, it poses the risk of introducing an arthropod pathogen that could destroy the entire colony. If claims to control resistant populations are sought, separate tests should be conducted on arthropods with known resistance to the active ingredient (or mode of action) to support such claims." (0008)
Outcrossing recommendations were deleted from the guideline based on SAP and public comments. 
70
"Test species. Although current efficacy guidance 3 different species for general tick claims, the protocol defines no more than testing 2 species at same time on same animal. HCPA recommends reconciling this discrepancy and applying appropriate scientific foundation." (0008)
The guideline recommends testing two species at a time to help decrease the total number of studies run, and therefore reducing the number of animals used in testing, since previously each tick was typically run as a separate study. All 3 representative species however should be tested to gain a general tick claim. Therefore, two total studies should be run to test all 3 representative species. However, it is common for registrants to test additional tick species and suggesting them all run 2 at a time will cut down on the total number of tests. No change to the guideline is necessary. 
71
"Recommend including additional detail to ensure consistent counting... Do they want us to count dead fleas on the tray/cage and the animal compared to live fleas on the control animal? What is the threshold for starts killing claim (90%)?" (0008) 
These sections were revised based on public comments; see response to comments #14 & 62 for more information.

The threshold for efficacy is 90% and will be listed in the upcoming product performance rule.  
72
"Table 2 change in infestation day to Day -1 vs. current Day -2" (0008)

Based on SAP recommendations, the guideline has revised the first tick infestation day from -1 to day -2 in Table 2 (in the tick section). The comment has this in reverse. 
73
Re: ticks. "Repellency testing. Cage substrate reference is confusing. Is a light-colored substrate or carpet the only material that can be used? If so, there should be some definition on what type of carpet, as pile height and fiber density, etc., can be quite variable and may result in substantially different outcomes. It is recommended that carpet be eliminated from use as a cage substrate." (0008)
The guideline was updated to read: 
"For each infestation/re-infestation, line individual pet carriers with a light-colored material, such as a light colored felt sheet. For each tick challenge, animals should be placed in the cage overnight or during a 12-hour dark period and 50 (+- 5) unfed and similarly aged adult ticks of each species should be placed inside the cage away from the animal."
74
"Mortality. For some treatments, i.e. collars, 72 hours after the initial infestation my not be long enough for the treatment to release and disperse across the surface of the body, which would not result in over 90% reduction. More importantly, some active ingredients are known to be slower acting against ticks. Knowing this, 72- hour pest counts should be an accepted endpoint for each re-infestation challenge." (0008)
If a product that is meant to last <30 days takes longer than 48 hours to kill/control/repel a pest (after the initial 3 days allowed to spread and start working), then efficacy of the product is not considered adequate for  protecting animals from these pests. For products lasting over 30 days, longer initial efficacy starting points may be justified. 

For all product types, it is recommended that efficacy be determined for each re-infestation at 48 hours to demonstrate that the product provides continuous adequate protection of the animal. 
75
"Tick removal by forceps should be added to comb counts as an accepted method for final tick counts, or "other appropriate means". (0008)
The guideline never prescribed a way to physically remove the ticks, only how to do the final counts, and therefore forceps could be used. No change to the guideline is necessary.  
76
"(4)(i) Adult mortality counts can be made 72 hrs. after initial and 48 hrs. after residual (currently using 24 and 48 hrs.) Change could mean less efficacious products or existing products could reformulation of existing products to meet the reduced standard." (0008)
The standards in the guideline are based on what is consistently seen in studies and the standards that have been used for several years. If products work quicker than this, they may indicate so on the label.
77
"In both the flea and tick sections I dont see the scientific need of change in preforming dead tray counts. None of the calculations presented use this collected data. Most specifically the second paragraph of section K(6) is completely contradictive of itself stating to use the dead status ticks in the calculation but the last sentence defines survival as the no acaricidal effect which is the Live categories. Then the calculation presented uses survival or live only. Even in a tray collection system not all dead ticks are going to be recovered. This number cannot be accurately used to calculate the efficacy. Finally, moving the animals to special collection cages and isolating them for these infestation collection periods could be considered an unnecessary stressor simply for a meaningless visual confirmation of yes there are a few dead parasites" (0011)
See response to comments #14 & 62. 

In response to the comment about moving the animals to special collection cages and isolating them during infestations, this is to prevent pests from moving from one animal to another between the time of infestation to comb out/final counts. 
78
"In section
n k(3)(ii) for short time tick repellency, it is more scientifically accurate to real world application if ticks are not placed directly on the animal. Procedure should include infestation of the chamber floor and allow the ticks to naturally find and infest the animal as is procedure in the Dumont et al 2015 paper." (0011)
Based on SAP and public comments, the tick repellency section was revised to have ticks not applied directly to the animal, but rather in the same environment (i.e., in a pet carrier) and left overnight or in a 12-hour dark cycle. Repellency will then be assessed by whether the ticks are found on the host or not.
79
"Request that the EPA justify the need for three tick species and which species are required. As multiple tick species carry a variety of infectious agents, it seems more rationale to allow Sponsors the flexibility to choose the species that they feel most necessary to control." (0026)
The three recommended tick species for testing were chosen based on the potential for carrying the most common diseases and for host specificity (e.g., ticks that infest cats are different than ticks that infest dogs). The species will be published in the upcoming Product Performance Rule.
80
"AHI agrees with the infestation methods suggested in the document. However, we note that some sites are moving away from the use of an infestation chamber because of animal welfare reasons." (0026)
The guideline does not mention the use of infestation chambers, rather that the animals be individually caged during the infestation period which can represent many options.  Section 2 of the mosquito and biting fly section states "...animals should be individually caged and exposed to 25  -  100 adults... Cages should be sized to comfortably house the sedated animal; typical cages are approximately 24 inches x 24 inches x 24 inches, but larger animals may need larger cages (Meyer et al. 2003, Ross et al. 1997)." No change to the guideline is necessary. 
81
"why are the pre-treatment infestations restricted to occurring by Day -5? We suggest that the latest time point for pre-treatment infestations should be decided by the Sponsor to allow flexibility in scheduling studies." (0026)
The guideline was revised to add the word "approximately" to allow more flexibility. 
82
Re: using sticky tape to prevent escape: "This is a task that is up to the study conductor to determine  -  and should not be mandated by the guidelines." (0026)
The guideline has been revised to "may be used."
83
Re: timing of infestation: "The initial infestation (listed as Day -1) should be allowed to be on Day -1 or Day -2. Day -2 is required by the EMA guidance. Allowing that in EPA studies will have no impact on the study outcomes but will allow studies to support registration in both geographies" (0026)
Based on SAP and public comments, and based on biology of the pests, the first day of infestation for fleas should be day -1 and for ticks should be day -2. The guideline was revised for ticks.
84
"The required use of heavy-gauge plastic pet carriers is too specific for a guideline. The guideline should allow flexibility in the enclosures used for tick exposures" (0026)
The Agency agrees with this comment and the guideline has been revised.
85
"In addition, the time period of 1 hour is not sufficient to allow proper evaluations for repellency or blood feeding. Circadian rhythms are important to mosquito feeding biology and, depending on time of day, the prime feeding time may be missed. The minimum of 60% of insects taking a blood meal is extremely high. This is highly dependent upon species and host preference. Some species or genera that are required to be tested do not prefer to feed on some species of companion animals and, in some cases, on mammals in general. This minimum percentage should be eliminated, with efficacy demonstrated where the average number of mosquitoes taking a blood meal in the control group is statistically significantly different that the average number taking a blood meal in the treated group." (0008)
If the time of day is important to get valuable results from a test, then it would be up to the study director and laboratory to test during optimal times. The guidelines do not prescribe all times best for testing.  

The minimum of 60% blood feeding is based on historical data submitted to the Agency as well as statistical confidence needed to make claims. If arthropod behavior is not adequate to ensure the desired effect is real, then the label claim would not be substantiated. 

EPA does not use statistical significance between the control and treatment groups as threshold for achieving efficacy.  Instead EPA determines a product to be efficacious when it achieves a minimum of 90% mortality to the treated pest.

No changes to the guideline are necessary. 

86
"Exposure to mosquitoes. Should "containing 150 (+- 10)" be containing 150 (+- 15)? If the proposal is for 50 (+- 5) and 100 (+- 10), this should translate to 150 (+- 15)" (0008)
The guideline will be revised to reflect this comment. 
87
"The proposal that "if more than 10% of insects in a treated group blood-fed prior to death, mortality should not be considered a viable endpoint" is unreasonable. Dead should be dead. This puts EPA regulated products at a disadvantage when compared to some FDA regulated products that are systemic and require feeding to affect an arthropod population. As FDA-regulated, these products would support a claim, but the EPA product would not, even though it results in the death of the mosquito. It is not clear the intent of excluding insects that have died which ultimately skews the data set. Some active ingredients are slower in speed of kill than others, but if the insects die within the test period it should be counted as dead." (0009)
Section (m)(4) i. of the guideline now reads: 

Mortality. Mortality may be considered a viable endpoint for a study if it occurs prior to the insect taking a blood meal. Mosquitoes and biting flies that take a blood meal are not repelled and can transmit agents that may cause disease; therefore, if more than 10% of mosquitoes blood fed, mortality should not be considered a relevant endpoint.

The EPA does not regulate systemic products, and thus, the intent of EPA registered products is contact toxicity and/or repellency, unlike for systemic products which require the mosquito to bite the animal for an effect to be seen. Thus, the purpose of topically applied, non-systemic products, is to repel and/or kill the mosquito.  Because claims preventing disease transmission are not made on EPA registered products, bites from mosquitoes which could transmit pathogens would not be providing the implied protection for the treated animal.  

88
"For mosquitos in section l, I am worried that the 60% blood feeding rate is to lofty a goal to target. In my lab I typically see about 40% blood feeding on controls. However, I dont sedate the animals on my current protocols. This non sedation could possibly account for the reduction that I see. I was not able to access all the scientific references listed in this section to verify that other researchers are meeting the new recommend amounts. Is this value supported by known science or is there a more appropriate value that should be used." (0012)
The minimum of 60% blood feeding is based on historical data submitted to the Agency as well as statistical confidence needed to make claims. If arthropod behavior is not adequate to ensure the desired effect is real, then the label claim would not be substantiated. Lowering the feeding to 50% would decrease confidence in the statistics, which would result in increasing the number of animals necessary for the tests. 

No changes to the guideline are necessary. 
89
Re: % bloodfeeding for inclusion: "AHI suggests that 50% would be appropriate because it would ensure adequate infestation but also reduce animal usage. Requiring 60% may lead to more animals being included in the studies to ensure an adequate number of dogs have the adequate infestation level." (0026)
See response to comment #88 above. 

No changes to the guideline are necessary. 

90
"Request that the EPA justify the need for three mosquito species and which species are required. As multiple mosquito species have been shown to transit heartworms, it seems more rationale to allow Sponsors the flexibility to choose the species that they feel most necessary to control." (0026)
The representative species have been chosen to represent the different genera that transmit diseases of public health importance. These will be included in the upcoming Product Performance Rule.


Charge Question 2
New approach methodologies (NAMs) is a broadly descriptive reference to any non-animal technology, methodology, approach, or combination thereof that can be used in this case for efficacy purposes. EPA seeks to reduce the number of vertebrate animals needed for testing, where possible, while ensuring confidence in test results.
a. For each of the following sections from the guideline, please discuss potential alternative methods, including ways to reduce or replace, the use of animals in testing:
    i   Fleas (section (j))
    ii  Ticks (section (k))
    iii Mosquitoes and biting flies (section (l))
    iv Insecticidal shampoo products (section (m))
    v. Environmental conditions (section (n))
b. Considering the importance of accurate animal behavior (e.g., grooming) during these tests and confidence in the results of these tests for protecting public health, for each of the following sections of the guideline please discuss methods for refining, e.g., reduce the pain and suffering, the use of animals.
    i. Fleas (section (j))
    ii. Ticks (section (k))
    iii. Mosquitoes and biting flies (section (l))
    iv. Insecticidal shampoo products (section (m))
    v. Environmental conditions (section (n))

Panel Recommendations:
91
2a. Most Panel members suggested that the Agency should pursue an area that holds great promise for the development of NAMs for chemical safety testing, namely, predictive mathematical and computer-based models. These models can be used to predict biological effects by incorporating data from chemical descriptors and do not involve laboratory experiments. However, due to variability in model performance, the reliability of in silico models should be confirmed for each application. Including biological information can improve predictive modeling. One source of such data is in vitro bioactivity from quantitative high throughput screening (HTS) assays, which simultaneously produces concentration-response profiles for thousands of test agents in a single experiment. For example, an important goal of the NRC report "Toxicity Testing in the 21st Century: A Vision and a Strategy" (2007) involved moving away from whole animal testing towards in vitro methods.  PG 23-24
EPA added language to the guideline encouraging the submission of protocols for NAMs; many computer-based models can be considered NAMs.
92
2a. One Panel member completely disagreed with the Tox21 approach. This panelist stated that while in silico models might have predictive abilities for some systems (e.g., identifying new drugs), that is completely different from product efficacy testing, which is what the EPA testing protocols are about. This panelist also noted that FDA does not allow in silico models as the justification for a new drug use (even if the company discovered it using such a pipeline). FDA would still require product testing and EPA should require the same.  PG 24
EPA added language to the guideline encouraging the submission of protocols for NAMs; many computer-based models can be considered NAMs.
93
2a. The Panel recognized that the goal of the guidelines is to provide concrete protocols that are deemed acceptable for regulatory purposes based on the scientific literature and previous studies. Therefore, many on the Panel agreed that NAMs required for the efficacy testing in the proposed guidelines do not currently exist and, therefore, are not currently recommended.  PG 24
EPA is not incorporating any specific guidance for NAM testing into the guideline, however, to prepare for future opportunities for developing NAMs and reduce animal testing a statement was added to encourage submission of NAM protocols.
94
2a. However, the Panel recognized that guidelines are anticipated to have a long shelf-life, e.g., the current set of guidelines that these will replace are 20+ years old. Due to the extended time period that guidelines are often in place before being modified, it is likely that NAMs that sufficiently replicate the animal tests proposed in these guidelines will be developed within the anticipated lifespan of this document. The Panel noted that section (e) of the proposed guidelines allows for the introduction of novel testing methods, indicating that they should be submitted to the Agency for review prior to testing. While the purpose for including this allowance was to recognize that novel products may arise that require different methods, it seems reasonable that this could also pertain to the introduction of NAMs. The Panel recommended adding specific language to this section that makes it clear that the development of NAMs is encouraged and that the Agency will accept these for review.  PG 24-25
EPA added language to the guideline encouraging the submission of protocols for NAMs.  
95
2a. In addition, the Panel suggested adding language that specifies how a new protocol and technique may be implemented beyond saying that it will be reviewed by the Agency. In particular, the Panel recommended adding some specifics as to how NAMS will be deemed appropriate replacements for the proposed animal tests. One suggestion is to encourage product researchers to run non-vertebrate (in vitro) tests along with the approved traditional animal testing methods (in vivo) to determine if the same outcomes are obtained. For example, if product testing on pests exposed to treated animal hair in vitro demonstrates similar results to in vivo tests in repellency assays, these may be considered as a useful surrogate to reduce vertebrate animal testing for certain product applications (Stanneck et al., 2012a).  PG 25
See response to comment #28.
96
2a. At least one Panel member viewed the tests in the proposed guidelines much like Phase II and Phase III clinical trials that the FDA requires for human and veterinary drugs. In this regard, the panelist agreed that NAMs are not currently recommended, and concluded that, as for drugs regulated by the FDA, to determine whether products work, testing of these products must be done on the patients themselves (i.e., using cats and dogs). The panelist noted that the numbers of test animals suggested by the Agency are far fewer than those for FDA clinical trials.  PG 25
See response to comment #28.
97
2a. The panelist also approved of the Agency's use of refinement rather than replacement for animal testing, as the Agency has conducted power analyses to help decision-making on number of animals needed and is recommending that several test species be tested simultaneously on the same vertebrate animal, where possible.  PG 25
EPA appreciates the comment and will continue efforts to refine animal testing to use fewer animals when possible.
98
2a. Several Panel members applauded the Agency on their use of Monte Carlo simulations in the accompanying document titled, "Sample Size for Pet Product Studies". One Panel member noted that in using such simulations, as detailed in Portier and Kaplan (1989), they were able to enhance the robustness of experiments with limited sample size. Thomas et al. (1996), with limitation of a theoretical N=1 in their experiments, applied Monte Carlo Simulations to model probability distribution functions of physiologically based pharmacokinetic (PBPK) parameters to estimate means, medians, standard deviations, first and third quartiles, skewness, and kurtosis and obtained age- and chronic dosing related pharmacokinetic differences in mice in parallel with a two-year chronic toxicology study.  PG 25
These panel members recommended the Agency continue to apply computational NAMs that include Monte Carlo simulations [underline added by EnDyna for emphasis] for justifying a reduction in the required number of vertebrate animals to six per treatment group.  PG 26
They also suggested adding text to the guideline document that details these approaches and encourages manufacturers to apply computational technologies to reduce animal usage.  PG 26
EPA added language to the guideline encouraging the submission of protocols for NAMs, many computer-based models can be considered NAMs.
99
2a. Several Panel members applauded the Agency on their use of Monte Carlo simulations in the accompanying document titled, "Sample Size for Pet Product Studies". One Panel member noted that in using such simulations, as detailed in Portier and Kaplan (1989), they were able to enhance the robustness of experiments with limited sample size. Thomas et al. (1996), with limitation of a theoretical N=1 in their experiments, applied Monte Carlo Simulations to model probability distribution functions of physiologically based pharmacokinetic (PBPK) parameters to estimate means, medians, standard deviations, first and third quartiles, skewness, and kurtosis and obtained age- and chronic dosing related pharmacokinetic differences in mice in parallel with a two-year chronic toxicology study.  PG 25
They also recommended exploring Markov Chain Monte Carlo Simulations, [underline added by EnDyna for emphasis] which consider covariance and other uncertainties, and coupling simulations with Bayesian approaches [underline added by EnDyna for emphasis] when new scientific data are available.  PG 26
They also suggested adding text to the guideline document that details these approaches and encourages manufacturers to apply computational technologies to reduce animal usage.  PG 26
See response to comment #98. 
100
2a. A few Panelists held a contrasting view regarding the readiness of NAMs. They did not think that the testing on animals is necessary or justified to evaluate the insecticidal activity of a product in all cases. They indicated that while new approaches were not sufficiently developed, there are existing alternative methods to evaluate spray on insecticides against mosquitoes and biting flies, and probably for fleas and ticks as well. They made the point that there are a multitude of insecticide resistance studies done on fleas, ticks, mosquitoes and biting flies that all use nonvertebrate testing methods. They concluded that the methods proposed in the guidelines, although realistic, are a very expensive residual bioassay. They recommended to the Agency that an alternative is to require that the proposed insecticide has activity in a residual assay that is on par with the insecticides that are already known to be highly effective. There are abundant technologies for conducting residual insecticide bioassays on substrates such as paper, glass, plastic, etc. (Ferrero et al., 2006; Deletre et al., 2016; Borges et al., 2019; Navarro et al., 2013; Saytal et al., 2019). While these technologies may not be appropriate to verify claims of persistence of product performance on a pet, they are more than appropriate for claims of "kills."  PG 26
EPA does accept non-animal testing for spray-on type products that only make "kills" claims and do not make residual claims. However, the residual assays conducted on paper, etc., do not sufficiently replicate residual activity on an animal so until it is demonstrated these assays could be used in place of animal testing, EPA does not intend to accept them for residual efficacy claims.
101
2a.  .....one Panel member indicated that there probably are non-vertebrate tests that could replace the use of cats and dogs in some of these studies. The panelist suggested that repellency of a material can be effectively evaluated without using animals. The panelist's recommendation is to use laboratory choice and/or no-choice assays performed and compared against known standards that make claims of efficacy quite certain. The panelist further suggested that tests with cats and dogs could be replaced with tests that employ a model animal, such as a rat or mouse, which would cut the number of vertebrate animals needed for testing in half. The panelist concluded there is no reason to test both cats and dogs and at a minimum most tests should only be conducted in either dogs or cats, especially if the arthropod pest is found on both animals.  PG 26
The Agency does not currently have any data to provide evidence that rodents would be able to represent cats or dogs in these tests nor is there evidence that cats and dogs would be able to represent each other. No changes to the guideline are necessary.

102
2b. Fleas, ticks, mosquitoes and biting flies commonly affect dogs and cats. The Panel concluded that one should consider the role of animal behavior in testing. For example, could it change due to the test product and/or the parasite? The scratching seen after a flea infestation is more likely due to a provoked inflammatory reaction than the flea itself, while ticks may go unnoticed in some cases.  PG 26
The prequalification stage of the animals is intended to help address and rule out any issues with an animal's behavior. Any non-standard test should address any potential behavioral issues in the proposed protocol.     
103
2b. The Panel proposed that the test arthropods be in the environment of cats and dogs for only a short period of time.  PG 27
The guideline outlines the timing between infestation of arthropods and removal/counts. No changes to the guideline are necessary. 
104
2b. While NAMs may be appropriate as screening tests for repellency, for when loss of efficacy occurs, for initial kill, and possibly for rearing pests, the Panel expressed concern regarding how to incorporate pet behavior, such as grooming, that can affect the arthropod counts into the NAMs test. The Agency should provide some guidance on how NAMs protocols and the results from NAMs could be assessed so that they could eventually be used for testing products. For these reasons, the Panel's recommendation that NAMs be assessed via direct comparison to the animal tests they are attempting to replace becomes critical.  PG 27
See response to comment #28. 

Public Recommendations (docket ID # in parentheses):
105
The following article was provided to the Agency for consideration: "Itching for change: Embracing modern flea and tick product development. Please contact the EPA Docket Center, Public Reading Room to view this document." (0009)
EPA has read and considered the article. The only current actionable recommendation for this guideline is that test parasites be fed on artificial membrane feeding systems.  As the guideline is currently written, arthropod colonies are not required to be fed on animals, thus in vitro feeding is acceptable to the Agency.
106
2a. "Similarly, before being included in the guidelines, the ability to extrapolate efficacy data from dogs to cats, or vice-versa, should be analyzed. The data would show whether a product that is effective in one species can predict efficacy in the other species, and, therefore, whether testing on one species could be waived. Waiving testing for one species would be in line with EMA guidance, which permits efficacy data from tests on dogs to be extrapolated to cats when scientifically justified and one dose confirmation study in cats for each parasite species on the label is conducted (CVMP, 2016)." (0019)
EPA does not have specific methods or modeling to extrapolate efficacy data from one species to another and thus has not provided methods for this in the guideline. 
107
2a. "The guidelines do not sufficiently address the use of field studies to reduce and replace testing on animals in laboratories. In the updated draft guidelines, field studies are described as not "preferred" by the agency and only used to supplement laboratory data or for specific claims on occasion (OCSPP 810.3300(i)(ii)). However, justification for largely excluding the use of field studies is not provided. Due to ethical considerations, diligence must be given to incorporating the use of field studies to replace animal use in laboratories for testing ectoparasiticides. In addition, robust clinical studies are more indicative of efficacy than the artificial conditions created in laboratory tests. Unlike laboratory tests, field studies can assess efficacy against natural infestations under a wide range of conditions, from differences in physical and behavioral characteristics of the animals to environmental differences (Marchiondo, 2007). The One Health Company (www.theonehealthcompany.com/), for example, designs and conducts multicentric clinical efficacy trials and has the expertise to assist with the coordination of efficacy trials across veterinary clinics in order to run multi-arm studies using real-time analytics to compare naturally non-treated controls, positive controls, and a new treatment." (0019)
Field studies may have inadequate or inconsistent pest pressure, lack of appropriate controls, and inaccurate pest counting. They are often conducted using endpoints which do not support claims commonly made by pesticide companies which fall under EPA's regulatory authority (e.g., severity of flea allergy dermatitis).  In addition, field studies are also not conducted in ways that allow different claimed effects (e.g., repellency vs. mortality) to be differentiated. Thus, although field studies can sometimes provide useful information, they do not typically directly support the types of efficacy claims made on pesticidal products and EPA recommends laboratory studies be conducted instead of field studies. However, protocols for field studies may be submitted if desired.

108
2a. "While we believe that demonstrating effectiveness in each target parasite is important. Nevertheless, in selected cases and with appropriate justification, the use of foreign data on a particular parasite could minimize the need for testing in dogs and cats." (0026)

Studies are not required to be conducted within the United States; studies conducted outside the U.S. are commonly submitted. The EPA may accept studies conducted in another country, so long as the data support the product use directions and claims. 


109
"HCPA is supportive of new approach methodologies (NAMs) that reduce the number of vertebrate animals needed for testing, where possible, while ensuring confidence in test results" (0008)
See response to comment #28.
110
"We do not agree with the statement that, "at this time no reliable non-animal alternatives are available to avoid the use of animals for efficacy testing of fleas, ticks, mosquitoes and biting flies." In our view, that conclusion is premature since the agency has not sufficiently considered available methods and how they might be adapted. For example, some of the established skin models and skin sensitization systems already used in toxicity testing may provide potential for reducing or replacing animal use in this case. Further, EPA has a unique opportunity to advance the field by engaging with test method developers to determine how the requirements of those making regulatory decisions can be met. It may be that adaptations can be made without too much effort, thereby preventing significant animal suffering. "We urge the agancy and the panel to consider the following: :: In consultation with experts in the field of alternative methods, performing a thorough and formal review of animal replacement technologies that have relevance or potential usefulness, including computational methods; :: Field testing or clinical studies using animals who could benefit from treatment; :: Information cited and discussed in Groff, K., & Bishop, P. (2017). Itching for change: Embracing modern flea and tick product development. Regulatory Toxicology and Pharmacology (6 July online) (http://dx.doi.org/10.1016/j.yrtph.2017.07.002)" (0010)
We appreciate the comment and citation. For discussion of the relevance of the citation please see response to comment #105. For discussion on NAMs, please see response to comment #28 and #91.
111
"Test methods to replace animal use in efficacy testing of ectoparasites on animal companions have been available for decades and have been submitted to regulatory agencies. For example, data submission packages to the EPA and the European Medicines Agency (EMA) Committee for Medicinal Products for Veterinary Use (CVMP) have included results from in vitro efficacy tests in which fleas and ticks were directly exposed to products in a petri dish (EPA, 2010; CVMP, 2006) and from in vitro studies on adulticidal activity of a product on fleas using an artificial membrane system (CVMP, 2011). Artificial membrane systems, which allow parasites to naturally attach and feed on blood or media through a membrane, are described in the literature, and they can be used to evaluate the efficacy of both oral and topical ectoparasiticides that are added to the blood in the in vitro system or applied to the membranes (Krober and Guerin, 2007a, b, c; Kuhnert et al., 1995; Andrade et al., 2014; Kuhnert, 1996; de Moura et al., 1997; Tajeri and Razmi, 2011; Bohme et al., 2014;Wade and Georgi, 1988; Li et al., 2015; Banks et al., 2000; Williams et al., 2014; Meola et al., 2000).  Artificial membrane systems offer advantages over in vivo methods, including greater control and standardization, such as the ability to control components of the blood meal, quantify dose effects of products or pathogen concentration in the blood, eliminate variability caused by animal grooming, and avoid complications from host-arthropod-pathogen interactions (Musyoki et al., 2004; Waladde et al., 1996; Macaluso et al., 2001; Krober and Guerin, 2007b; Kuhnert, 1996; Bonnet and Liu, 2012; Wade and Georgi, 1988). Further, variability between animals can confound the interpretation of results from studies in which fleas or ticks are fed using in vivo methods, an issue in vitro methods can overcome by using blood or media from the same source to evaluate controls and test compounds (Kuhnert, 1996). Researchers have successfully demonstrated the efficacy of parasite control products using artificial membrane systems (for review, see Groff and Bishop, 2017). For example, Li et al. (2015), Krober and Guerin (2007b), and Kuhnert et al. (1995) used silicone membrane systems to confirm the efficacy of commercial ectoparasiticides against Amblyomma americanum adults, Ixodes ricinus adults, and Amblyomma hebraeum nymphs; and Williams et al. (2014) used a parafilm membrane system to demonstrate that a commercial flea product controlled flea reproduction. The SAP should advise the EPA to convene an expert working group or workshop to develop a plan to incorporate non-animal methods in OCSPP 810.3300. The goal of these discussions would be to devise an approach to evaluate the efficacy of ectoparasite products using a combination of in silico models, in vitro tests, and field studies. Further, we appreciate that the updated guidelines note that the EPA is open to reviewing protocols proposing novel testing methods (OCSPP 810.3300(e)), and language included in the guidelines should encourage companies to propose methods that reduce and replace animal testing. Expedited review of such protocols will facilitate the adoption of non-animal test methods. To ensure timely review, we propose the addition of a practicable timeline to OCSPP 810.3300(e)." (0019)
EPA is aware of many of these types of tests (e.g., membrane feeding tests), and these tests have strengths especially in confirming that an active ingredient has a specific effect and controlling for dose. However, there are insufficient data to show that these tests adequately evaluate residual efficacy of products applied to animal skin. Residual efficacy can be affected by a myriad of factors including degradation, movement across skin, and exposure to elements. As mentioned previously, EPA is open to protocols and studies assessing the methods simultaneously. Review timeframes for submitted protocols can be found under the appropriate categories for protocol submission in PRIA.
112
"The potential expansion of the use of in-vitro techniques to demonstrate contact activity of ectoparasiticides against fleas and ticks could be considered as means to decrease the use of dogs and cats in research. Several well developed and validated methods exist that are currently used by many companies in early development stages. These include impregnated filter paper tests for ticks or coated glass surfaces for fleas and ticks." (0026)
See response to comment #28. 
113
"Animals are used to rear ectoparasites as well as assess product efficacy. These animals are immobilized and areas of skin are shaved to allow parasites to feed undisturbed. During efficacy testing dogs and cats are literally infested with fleas and ticks. Use of animals in these ways leads to obvious adverse effects, including stress, anemia, unrelieved itching, and constant pain and discomfort. NAMs that replace animals in these situations already exist. Substantial research has been carried out over the last twenty years on use of artificial membrane systems, which essentially act as skin and allow fleas or ticks to attach and feed on blood or synthetic media through the membrane. These systems have been used with varying degrees of success to rear and test products on fleas and ticks. They have also been employed in studies of the transmission of disease by these vectors. Recent work has demonstrated the utility of these systems in feeding and maintaining mosquitoes as well. However, further research and development is needed to optimize and standardize artificial membrane systems if they are to be used for regulatory purposes. We recommend that government should make funding available to support work on non-animal methods of rearing fleas and ticks and testing ectoparasiticides. Additionally, the ectoparasite product industry should play a role by dedicating a portion of profits to development of these systems. Researchers, regulators, and industry need to collaborate on identification of specific research needs, as well as optimization and standardization of protocols using in vitro systems for efficacy testing and rearing of different species and life stages. This could be accomplished through a series of workshops, which the Humane Society would be happy to help organize. We look forward to further dialog with EPA on this subject. Thank you." (0027)

As the guideline was initially drafted, EPA does not require arthropod colonies to be fed on animals for use in testing, nor is EPA making this recommendation for the final guideline. EPA will consider ways (e.g., workshops, etc.)  to promote the development of NAMs for efficacy testing.
114
"During the process of bringing a new ectoparasiticide to market, animals are used in three areas: toxicity testing, rearing of ectoparasites (includes breeding and feeding) for later use in efficacy testing, and efficacy testing itself. A recent paper published in Regulatory Toxicology and Pharmacology, Itching for change: Embracing modern flea and tick product development (Groff and Bishop 2017), gives a comprehensive overview of both current and alternative systems of feeding/rearing and efficacy testing. The paper's main findings are summarized here. To rear large numbers of parasites for later use in efficacy testing, animals are often immobilized with areas of skin shaved to allow fleas and ticks to feed undisturbed. For efficacy testing, dogs and cats are repeatedly infested with 100+ ticks or fleas for up to a month or more. Use of animals in these ways leads to obvious adverse effects, including stress, anemia, unrelieved itching, and constant pain and discomfort. Alternative feeding systems for rearing and efficacy testing have been developed and used with varying degrees of success. Artificial membranes which essentially act as skin, allow fleas or ticks to attach and feed on blood (typically obtained from a slaughterhouse) or synthetic media through the membrane. The membranes can be modified to accommodate different species and life stages, and they permit researchers to directly observe attachment to the membrane, feeding, reproductive output, and mortality. The membranes may be made from silicone, parafilm, or animal-derived skins. The cat flea Ctenocephalides felis (C. felis), the most common flea species infesting cats and dogs, has been successfully reared using in vitro parafilm membrane systems, with survival rates equal to fleas fed on cats, although reproductive rates were lower. For efficacy assessment of oral and topical products that act through the blood, these in vitro systems provide greater control and standardization, such as the ability to accurately quantify doses, eliminate variability caused by animals grooming, and control components of the blood meal. Variability among test animals, which can confound the interpretation of results, can be overcome by using in vitro methods. In addition, they offer cost savings when compared to the use of animals in test systems. Researchers have successfully demonstrated the efficacy of tick control products using silicone membranes with comparable data from paired in vitro/in vivo studies (e.g., Krober et al. 2007a, 2007b; Fourie et al. 2013; Li et al. 2015). In vitro parafilm membrane systems also have been used successfully for determining the efficacy of products to control C. felis. Other studies have demonstrated the viability of the parafilm membrane system by using it to study the transmission of viruses and bacteria by fleas, including the length of infectivity and mode of transmission (e.g., transmission from blood to fleas, from fleas to blood, through flea excrement, and from fleas to their progeny) (e.g., Vobis et al. 2003; Bouhsira et al. 2013, Kernif et al 2014). The same disease transmission capabilities and routes of infection were captured as in the in vivo studies. A recent study by Tajeri et al. (2016) of Theileria lestoquardi infection by ticks found that while in this case an in vitro system using silicone membrane was largely unsuccessful, high rates of attachment and feeding were observed using an in vitro mouse skin system. Alternative systems are not without challenges, however. Multiple variables affect the success of membrane feeding, including the choice of host species blood used in the in vitro system, preservation of the blood, the addition of supplements to the blood, and attachment stimuli. Hard ticks have complex feeding requirements, depending on a range of olfactory, mechanical, and host-contact stimuli conditions. In addition, the use of anticoagulants, antibiotics, and antifungals in the blood may affect success of tick feeding. Also, meals have to be long-lasting, as hard ticks feed for up to two weeks at a time depending on life stage. To overcome some of these obstacles, researchers have experimented with ways to improve feeding and rearing success. For instance, Krull et al. (2017) showed that the ixodid tick, Dermacentor reticulatus, experienced increases in both the engorgement mass and fecundity when fed at an increased CO2 level. Similarly, membrane systems have been improved to increase successful use by fleas. One researcher demonstrated the importance of providing hair as a substrate for survival and reproduction of the cat flea (Pullen and Meola 1995), while another found that human blood can be used successfully as a blood meal and has a relatively long shelf life (Kernif et al. 2015). While further studies are needed to identify a complete set of optimal conditions for fleas, current parafilm membrane systems are suitable for rearing fleas for ectoparasiticide testing and, with regulatory approval, for efficacy testing. An alternative system has been recently reported for use with mosquitoes. Luo (2014) describes a novel multiple membrane blood‐feeding system for mosquitoes that can be used for the study and routine maintenance of Aedes aegypti L. and Aedes albopictus Skuse, which require a meal of vertebrate blood to produce eggs. This blood‐feeding system uses cattle collagen sausage‐ casing membrane to facilitate feeding and adds adenine nucleotides to the blood to stimulate the feeding response. The efficiency of this blood‐feeding system was compared to a live mice blood source. The Ae. aegypti fed on pig whole blood had 89.7% (without ATP) and 90.7% (with ATP) blood‐feeding rates, which were not significantly different from the mice‐fed ones (98.0%). The Ae. albopictus fed on pig whole blood (with ATP) had a success rate of 84.4%, which was significantly different (better) from the mice‐fed mosquitoes (51.1%). The feeding rates did not differ between sausage‐casing membrane and Parafilm‐M(R). The survival rate, fecundity, pupation, and pupal emergence rates of Aedes females fed on pig whole blood were not significantly different from the mice‐fed ones. While a significant amount of research has been conducted in this area, further work is needed in development and standardization of artificial membrane systems to rear ectoparasites and for use in regulatory efficacy testing. In addition, to eventually transition to completely non-animal systems, development of appropriate synthetic feeding media that could replace the use of animal blood is needed. Groff and Bishop (2016) lay out a series of steps that can be taken for this to occur as summarized below: * Governments should make additional funding available for research on non-animal methods of rearing fleas and ticks and testing ectoparasiticides. * A portion of the profits from the ectoparasite product industry should be dedicated to research and development of these systems. A series of workshops should be convened to bring together researchers, regulators, and industry to collaborate on identification of specific research needs, as well as optimization and standardization of protocols using in vitro systems for efficacy testing and rearing of different species and life stages. * Studies comparing results of in vivo and in vitro systems should be conducted to increase acceptance by regulators. * Once accepted protocols are developed, regulatory agencies must adapt their guidelines to allow and encourage the use of these alternative systems. * To encourage acceptance and uptake of these methods, governments should provide incentives for companies that use in vitro systems to rear parasites and test products whenever possible. The Groff and Bishop paper is included with these comments for the SAP's perusal. HSUS and HSLF feel strongly that this area of testing is ripe for change and that systems currently exist that could be used successfully to replace some or all of the animals used for these purposes. Significant progress could be made by convening a series of workshops that that focuses on optimization and standardization of these approaches for regulatory use." (0028)
See response to comments #28 and #113. 
115
"The draft guidelines recommend a heavy ectoparasite burden. Animals experience obvious physical and emotional adverse effects from infestations with 100 fleas or 50 ticks, which is well in excess of infestations typically present on companion animals. As a reference, a study for a flea control product notes that a cat, who was artificially infested with 50 fleas throughout the test, died from anemia due to severe flea infestation. Animals in field studies are considered infested and included in studies when they have just three to ten ectoparasites. We recommend further discussion around the increase in parasite burden in the draft guidelines." (0031)
EPA understands the concern with parasite load. Parasite loads as high as those recommended would be on the high end of parasite loads that might be observed on companion animals. However, reducing parasite load typically results in a higher number of test animals because arthropod behavior can be difficult to account for.
116
"The recommendation to conduct testing for all species of ticks as well as fleas, if included on the proposed label, results in the use of a large number of dogs and cats (OCSPP 810.3300(d)(1)(vii), OCSPP 810.3300(i)(2), (OCSPP 810.3300(k)(1)). A comparison of efficacy data for different species of ticks and fleas must be conducted before requiring such a large number of animals to be used. This   evaluation would determine whether one species of tick or flea is the most difficult to control, and therefore, if a product controls that species, testing on other species can be avoided. This analysis would consider how the modes of action of different active ingredients may affect the results. The EPA OPP has undertaken similar retrospective analyses in order to inform testing guidelines (e.g., Hilton et.al., 2019)." (0019)

EPA does not have information on which species is the most conservative and thus is not including that in this guideline.  However, EPA is open to an analysis and comparison of susceptibility of the different species for future testing.
117
"We support the exclusion of a positive control (OCSPP 810.3300(d)(1)(viii)(b)) and the clear language that the EPA uses in the draft guideline. The use of animals in negative control groups should be reconsidered in the draft guideline as these tests are particularly painful to the animals used (OCSPP 810.3300(d)(1)(i)). Instead, for example, the efficacy of a product can be determined by the percent reduction in number of fleas and ticks on animals from pre- to post-dosing without the use of negative control groups, as is common in field studies (Karadzovska et al., 2017; Cavalleri et al., 2017a, b; Rohdich et al., 2014; Meadows et al., 2017; Navarro et al., 2016). The SAP should advise the EPA to conduct an evaluation of the use of control groups. Kramer and Font (2017) note that "[c]ontrol groups...offer a relatively untapped potential to reduce sample size without sacrificing statistical power" and conclude that "the number of current control subjects can be reduced by more than half by including historical controls in the analyses. In other experimental scenarios, control groups may be unnecessary." They describe how the results of prior experiments can help to determine the need for control groups and write that "[w]hen similar experiments are performed repeatedly in the same laboratory, using the same standard research and husbandry protocols and the same animals...there will often be scope for the incorporation of historical controls"" (0019)
At this time, the EPA does not have any information on validated methodologies to  generally support this across study types, however the Agency is open to considering a reduced number of animals in the control group on a case-by-case basis with justification provided prior to study commencement. 
118
"The draft guidelines state that end-use formulations should be tested (OCSPP 810.3300(d)(1)(ii)), but do not specify testing requirements for substantially similar formulations. The EPA and FDA have guidelines to minimize animal testing using existing information and bridging to similar products (e.g., EPA, 1998; FDA, 2001)." (0019)
When formulations are considered substantially similar, EPA allows bridging of data between the two products, similar to what is done with toxicity data.
119
"Dogs and cats are social animals whose suffering is exacerbated when isolated. At a minimum, the guidelines should require enrichment for when animals are housed alone. Further, the draft guidelines stipulate that animals "should" be housed individually during infestations, but only that they "may" be housed together during other time periods (OCSPP 810.3300(d)(1)(v)). Considering the social needs of dogs and cats, the guidelines should stipulate that when not infested with parasites, animals "should" be socially housed." (0019)
EPA has incorporated both suggested revisions. 
120
"(2) Laboratory testing What about "ex vivo" studies such as the ones using dog's hair sampled from treated or untreated animals at different time point. This is much more ethical especially when many different species must be tested and when biting is painful (for example, stomoxys flies). Recommend that" (0008; note that the sentence just ends)
See response to comment #28.


Charge Question 3
Should protocols for testing ear mites and Sarcoptic mange mites be included in the guideline?
a. Why or why not?
b. If so, please provide appropriate detailed test methods for evaluating the efficacy against these pests for on-animal treatments.

Panel Recommendations:
121
3a. The Panel recommended that protocols for testing for ear mites (Otodectes cynotis) and Sarcoptic mange mites (Sarcoptes scabiei) should be included in the proposed guidelines. These are two of the most common mite pests found on dogs and cats, and ectoparasiticides such as fipronil and selamectin are used to treat fleas, ticks and mites. Ear mites and Sarcoptic mange mites can be important pests of companion animals and mange mites can be transferred to humans. Therefore, standardization of testing procedures for mites is essential for the development of multi-use products.  PG 27
Specific mite testing guidance will not be added to the guideline at this time. Currently, EPA has received very few studies for mites over the last 15 years and therefore these studies are not common enough, nor have a study design that produces reliably consistent data to establish standardized testing. Registrants seeking mite claims are encouraged to submit protocols for review. The Agency will use the information provided here by the SAP to guide discussions that may arise about methods for mites. 

For soft ear ticks, currently data do not need to be submitted to support labeling claims; however, registrants are required to have data on file supporting all pests if they are listed on the label. 
122
3a. No rationale is provided by the Agency for the exclusion of Demodex mites (Demodex spp.) that are also one of the most common mite pests on dogs and cats. Generalized demodicosis is a severe condition that is difficult to control with currently approved therapies, and new treatments are always being sought. Tests with all three of these mite species and the effects of various topical preparations are in the literature, and some products currently have some or all of them included on their labels (Stanneck et al., 2012b).  PG 27

Overall Summary
The Panel recommended that in addition to ear mites (Otodectes cynotis) and Sarcoptic mange mites (Sarcoptes scabiei) that Demodex mites (Demodex spp.) should be included in the proposed guidelines (Panahi et al., 2015).  PG 30
See response to comment #121.
123
3a. A Panel member also suggested that testing protocols for the soft tick, Otobius megnini (spinose ear tick), should also be considered as it can be a pest of dogs and occasionally cats in the Western U.S. Testing protocols for this tick would differ from hard ticks as soft tick life cycles differ substantially from hard ticks. O. megnini is an ear canal parasite. Only larvae and nymphs are parasites. Adults are free living and lay eggs in the environment, so only subadult ticks would be tested. A similar testing approach as suggested for Otodectes cynotis (ear mites) could be employed for trials using subadult O. megnini.  PG 27-28
See response to comment #121.
124
3b. Protocols for testing the efficacy and/or safety of topically applied ectoparasiticides against all three species of mites are in the published literature. Some examples are listed below.  PG 28

Overall Summary
Examples of possible test methods for evaluating efficacy against each of these 3 species for on-animal treatments are summarized from the literature. Test methods include both clinical trials and laboratory studies primarily in pets with natural infestation, although an experimental infestation study is cited for ear mites. In addition, several methodological issues that the Panel felt were germane to the evaluation of dependent variables and infestations are presented.  PG 30
See response to comment #121.
125
3b. Protocols for testing the efficacy and/or safety of topically applied ectoparasiticides against all three species of mites are in the published literature. Some examples are listed below.  PG 28

Demodectic Mange:
1.  Becskei et al. (2018) describe a safety and efficacy, randomized, single-blind, multi-center clinical study, in which monthly oral doses of sarolaner was evaluated in comparison with weekly topical applications of imidacloprid plus moxidectin for the treatment of generalized demodicosis in client-owned dogs.  Efficacy evaluation involved the number of live Demodex mites relative to pretreatment in five deep skin scrapings from each dog, and the proportion of dogs with no live mites on days 0, 30 and 60, and, if applicable, on days 90, 120, 150 and 180, in both treatment groups.

2.  Fourie et al. (2019) describe a laboratory study that compared the efficacy of two topical spot-on medications, fluralaner or a combination of imidacloprid and moxidectin, against naturally acquired generalized demodicosis in dogs. Client-owned dogs were transferred to the study site and individually housed indoors.  On Day 0, dogs in one group were treated once with fluralaner spot-on solution. Dogs in the other group were treated with the imidacloprid/moxidectin spot-on solution on 3 occasions (Days 0, 28 and 56) or weekly in severe cases. Mites were counted in skin scrapings and demodectic lesions were evaluated on each dog before treatment, and at 28-day intervals over the 12-week period. Deep skin scrapings were made from the same 5 sites on each dog at each examination.

Summary: Both clinical and laboratory-based studies have been used.  Both protocols use a positive control; however, Becskei et al. (2018) used geometric means, while Fourie et al. (2019) used arithmetic means. Housing conditions differed markedly, which could impact socialization issues and contact with other pets with the same infestation. Natural infestations were used in both studies.  PG 28
See response to comment #121.
126
3b. Protocols for testing the efficacy and/or safety of topically applied ectoparasiticides against all three species of mites are in the published literature. Some examples are listed below.  PG 28

 Sarcoptic Mange [Sarcoptes scabiei]:
1.  Stanneck et al. (2012a) describe a laboratory study with naturally infested dogs.  Ten (10) mixed breeds were used in a within subject design. On Days −2, 29, 60 and 90, skin scrapings (+/− 4 cm²) were taken from five places on the dog's body likely to be infested with mites, and the number of mites in these scrapings were counted. The clinical signs and extent of lesions on each dog were assessed on the days on which scrapings were made. Success rate was defined as a dog that complied with all of the following conditions: no live mites, a complete resolution of the presence of papules and skin crusts and a > 90% improvement in body areas with hair loss by Day 90 after the collars had been fitted.

2. Fourie et al. (2019) describes a clinical trial in which 16 client-owned dogs with naturally acquired generalized demodicosis were randomly allocated to 1 of 2 study groups: on Day 0, dogs in 1 group were treated once with fluralaner spot-on solution. Dogs in the other group were treated with the imidacloprid/moxidectin spot-on solution on 3 occasions (Days 0, 28 and 56) or weekly in severe cases. Mites were counted in skin scrapings and demodectic lesions were evaluated on each dog before treatment, and at 28-day intervals over the 12-week period. Deep skin scrapings were made from the same 5 sites on each dog at each examination.

Summary: Both a laboratory-based study and a clinical trial have been used; efficacy evaluations have been made using a between group (positive control) and within group design. Skin scrapings are used in both studies.  Natural infestations were used in both studies.  PG 28-29
See response to comment #121.
127
3b. Protocols for testing the efficacy and/or safety of topically applied ectoparasiticides against all three species of mites are in the published literature. Some examples are listed below.  PG 28

Ear Mite (Otodectes cynotis)
1.  In a laboratory study, Six et al. (2016) used 32 dogs with a previously induced Otodectes cynotis infestation (100 mites in each ear), in a randomized, complete block design to evaluate product efficacy. Total ear mite counts, range and percent efficacy relative to placebo control (not defined) were compared to dogs treated with either a single oral dose or two monthly doses of sarolaner. Counts of mites were made otoscopically on Day -4 and Day 30 and 60.

2.  In a second laboratory study, Taenzler et al. (2018) employed an experimental infestation of O. cynotis in dogs, by harvesting mites by lavage from donor animals and transferring approximately 50 to 100 mites, depending on the intensity of infestation in donor animals, into each ear of the recipient animal. Animals used in the study were experimentally infested within one month prior to study start. To determine efficacy, an otoscopic examination of both ears from each animal was performed prior to treatment and at 14 and 28 days after treatment to determine the number of live mites. However, 28 days after treatment, animals were sedated, and both ears were flushed to determine the number of live mites (adults, larvae, nymphs).
Summary: These 2 laboratory-based studies described in this section evaluated the efficacy of chewable systemically acting drugs rather than topically applied acaricidal products.  Both natural and experimentally induced (Six et al. 2016; Taenzler et al. 2018) infestations were used to determine product efficacy. Number of mites used for initial infestation varied from 50-100 between the studies. Primary assessment in both was the total number of live mite counts after treatment.  PG 29
See response to comment #121.
128
3b. Protocols for testing the efficacy and/or safety of topically applied ectoparasiticides against all three species of mites are in the published literature. Some examples are listed below.  PG 28

The methods used to determine the number of mites vary: Six et al. (2016) employed otoscopic counts at all time points. However, Taenzler et al. (2018) at the final evaluation (Day 28) flushed both ears of sedated dogs to determine the number of live mites, i.e., adults, larvae, nymphs.

Different Techniques to Count Ear Mites:

Bosco et al. (2019) states: "Assessment of O. cynotis infestation. Prior to enrollment, O. cynotis infestation was confirmed by direct or otoscopic examination of the external ear canal of both ears. On Days 28, 56, and 84 post-treatment, cats were sedated with dexmedetomidine hydrochloride (40 μg/kg IM), the ear ducts were filled with a solution of 0.9% sodium chloride (3 to 5 ml per ear canal) and the ears were massaged externally to displace the contents."

Machadoa et al. (2018) states: "The diagnosis of otoacariasis was confirmed by bilateral videootoscopy, and by evaluating the presence of mites."  PG 29
See response to comment #121.
129
3b. Additional Methodological Issues:
Several other issues were raised by the Panel:
a. One involved the necessity of using 5 deep-skin scrapings in dogs with either type of mange to evaluate the number of mites. Becskei et al. (2016) describe the procedure:
      "To count mites, deep skin scrapings were taken from at least four separate sites on each dog. If no mites were detected in the first four scrapings, additional scrapings were made until live mites were found or the maximum of ten scrapings was reached. Selected scraping sites were those that had the most severe or most likely evidence of current mite infestation. Scrapings were conducted to an approximately constant depth (to capillary bleeding) over an area of approximately 2.5 cm[2]."  PG 30
See response to comment #121.
130
3b. Additional Methodological Issues:
Several other issues were raised by the Panel:
A Panel member suggested that evaluation of mite mortality could be done by dipping (Croft et al., 1982), drenching (Pap et al., 1997) or residual exposure (Pasay et al., 2008).  PG 30
See response to comment #121.
131
3b. Additional Methodological Issues:
Several other issues were raised by the Panel:
b. Given that these mites tend to be localized and that long-term residual activity is unlikely to be needed, non-vertebrate testing would be a logical alternative for scabies.
Panel members indicated that it is possible to conduct scabies tests without vertebrate subjects using methods from the literature (Pasay et al., 2008; Pap et al., 1997) and recommended EPA state that non-vertebrate studies would be considered as an acceptable alternative.  PG 30

Overall Summary
With respect to methodology, the Panel recommended that ...... EPA state that non-vertebrate studies would be considered as an acceptable alternative for scabies tests.  PG 31
See response to comment #121.
132
3b. Additional Methodological Issues:
Several other issues were raised by the Panel:
The Panel also recommended that EPA consider that efficacy studies involving client-owned pets be conducted at clinics with volunteer patients. Cited above are several papers that outline procedures for testing product efficacy in three mite species, and the Panel recommended that EPA provide some latitude in the designs to be submitted.  PG 30

Overall Summary
With respect to methodology, the Panel recommended that EPA consider that efficacy studies involving client-owned pets be conducted at clinics with volunteer patients, .....  PG 31
See response to comment #121.

Public Recommendations (docket ID # in parentheses):
133
"Additionally, models of artificial infection of pets with mites exist and are validated. These have been previously used in support of product registration with other regulatory agencies, such as FDA and EMA. Whilst these rely on the use of animals like cats and dogs to demonstrate efficacy, recent developments in genotyping can help demonstrate similarities of various mite species. This may allow for using studies across geographies and therefore reducing the overall number of studies." (0035)
See response to comment #121.
134
"Do the guidelines incorporate the responsibilities outlined in Memorandum of Understanding Between the Environmental Protection Agency and The United States Department of Health, Education and Welfare Food and Drug Administration which incorporates ear mites (225-73-80102)?" (0008)

EPA has registered products with claims to control ear mites, however, this guideline is not intended to clarify the responsibilities outlined in the MOU. In addition, we have not included mites in the guideline due to reasons discussed in comment #121. 


Charge Question 4
For each section listed below, comment on the proposed sample sizes for vertebrate test animals (e.g., dogs, cats). EPA recommends these numbers based on the power vs. sample size analysis provided to ensure adequate power using the minimum number of vertebrate animals. Are these numbers of animals practical?
a. Fleas (section (j))
b. Ticks (section (k))
c. Mosquitoes and biting flies (section (l))

Panel Recommendations:
135
The 1998 OPPTS 810.3300 efficacy testing guidelines required at least 6 animals per group, while preferring 10 animals per group. The "Sample Size for Pet Product Studies" document states that historically the EPA has recommended 6 animals per group, but often received studies with 2-12 animals per study group (see pages 2-3 of the document "Sample Size for Pet Product Studies"). The Panel recommended that the EPA provide more historical data describing the number of animals per study group that was received and describe how often each study design (by group size) was deemed to be acceptable; such information could help to inform these new guidelines.

Aside from considering the historical data in more detail, the Panel agreed that the recommended group sizes in the proposed guidelines are in line with the EPA documentation of power vs. sample size that suggests statistical support for 6-14 vertebrate animals per group. The Panel also found that the sample sizes suggested by the EPA are generally consistent with the stated EPA historical practice requiring 6 animals per group (preferring 10 animals per group) and accepting studies with up to 12 animals per group.  PG 31
Thank you for the agreement on sample sizes. The sample sizes provided were based on power analyses using previously submitted studies which fit into the proposed testing methods and reflect our efficacy standards. 
136
The Panel viewed the number of vertebrate animals for the fleas and mosquitoes and biting flies as practical. The Panel had some concern about the number of animals recommended, particularly for the tick studies.

The Panel discussed a variety of logistical issues in the context of tick lab studies by having 14 treated and 14 control dogs. In order to have 28 dogs that are eligible to be enrolled in the study with adequate infestation levels (25% of infested tick species), there will likely be a need to start with 32 to 34 dogs. This would result in increased study costs in terms of the number of animals and per diems and goes against the 3R's (i.e., Replacement, Reduction, and Refinement). It would also result in more ticks needed at the study site for each infestation. Other issues noted by the Panel include: more dogs to comb that often results in technician fatigue; more data to record for each dog if multiple tick classifications need to be recorded (dead attached, dead free, live attached, live free; moribund, etc); an inability of contract research labs to schedule multiple studies on the same day/week; possibly reduced revenue for contract labs which are small businesses; contract labs needing to hire more technicians; and the inability to accommodate all study dogs into a single animal room (which would introduce another source of variability if more than one room is required).

The Panel noted that there are several design modifications that could potentially lessen these concerns. Given that these data are 'binomial', the variance depends on the true rate. As such, the group with the more extreme (farther away from 1/2) rate should use fewer animals. The animals could still be blocked (matched), but instead of using matched pairs, there would be matched sets (e.g., 2 controls + 1 treated or vice versa). Such matching is common in human clinical trials and could simply be implemented here (it is not more complex than the current blocking approach). In addition, the modeling does not seem to include the blocking/matching effect. Inclusion of such terms could help further increase the power by reducing the relevant variance for testing. Another possible design would be a cross-over design (see e.g., Section 5.4 of Chow and Liu, 2008) which, if it could be implemented, would decrease the sample sizes by 50% (but would require each animal to participate in both the control and treatment arms).

In addition, the Panel noted that the parameters of the power vs. sample size analysis have a large influence on the sample size calculations. For instance, the number of animals (dogs or cats) could be reduced by requiring a lower observed efficacy, a greater true efficacy accompanying a 90% observed efficacy, a greater proportion of pests retained, or less precision.

The Panel recommended that power simulations be explored to determine whether design modifications, like those proposed above, could be used to decrease the number of vertebrate animals in tick studies (and potentially in flea, mosquito and biting fly studies). The Panel also recommended that the impact of the values of each of the parameters and criteria, on the power and sample size in the simulations should be assessed carefully.  PG 32
The Agency appreciates the feedback about sample size. The tick and stable fly models have been re-run with to explore lowering the number of host animals or pests used. For tick testing, the number of animals (cat or dog) used per treatment group was lowered from 14 to 11. 

Currently, most studies run a positive control group and the groups average 8 animals per group regardless of pest. Therefore, tick studies are on average run with 8 animals x 3 groups (= 24 dogs) and have never indicated issues with counting space in labs or technician fatigue. In this guideline we are specifically stating not to run positive controls, as well as encouraging the utilization of multiple pests where possible, to lower the total number of tests required for registration, which would in turn lower the total number of animals to be tested to gain basic registration.

Public Recommendations (docket ID # in parentheses):
137
"HCPA believes that 6-10 animals should be sufficient and is concerned with the practicality of additional animals. There also appears to be an inconsistent approach applied between different invertebrates." (0008)
The minimum number of animals indicated by EPA is based on a power analysis conducted using the study design and control retentions of pests on the different animals (e.g., dog vs. cat) and for each different pest (e.g., biting flies vs. fleas). These sample sizes will ensure that data submitted are adequate to show that if a product reaches the 90% efficacy threshold, it is effective. 
138
"AHI supports the principles described in the WAAVP guidelines, where a minimum sample size of 6 animals per group is acceptable. A requirement to demonstrate statistical difference in the primary outcome of efficacy  -  such as number of live fleas or ticks  -  is appropriate." (0026)
See response to comment #137. 
139
Re: flea test: "What is the scientific justification of requiring seven animals in studies using cats and six animals for studies using dogs? Power would be calculated by the registrant. This is usually less than six animals per group. The conventional six animals per group is an arbitrary number... What is the justification for requiring 8 animals per group for biting fly studies but only six per group for mosquito studies? Six per group is traditionally required. If Sponsors wish to include more to reduce risk of not having adequate infestations, this should be their option, rather than a requirement." (0026)
See response to comment #137.
140
"The statistical modeling that the EPA has conducted serves as an underpinning to many of the recommended changes in requirements for the guideline. It is this methodology that has led to increased sample sizes (up to 14 animals per group required in tick studies). But the statistical modeling is based on assumptions of pest retention that are questionable. Testimony has been provided by Dr. Everett, for instance, that the assumption/requirement of 60% blood feeding of mosquitoes in controls is not seen in his laboratory (as a CRO). If the assumption is lowered to more reasonable values, the size of the treatment group must increase to maintain the same level of statistical power. The assumptions made by the EPA are at the high end of our experience in the conduct of these studies, and are not representative values." (0033)
See response to comment #137.
141
"Selection and allocation of test animals. Why are 14 animals the minimum number to be used for tick studies when only 6 or 7 are the minimum for flea studies? This is a significant difference that is not justified. Recommend a consistent, science-based and statistically sound value." (0008)
See response to comment #137.
142
"In section k(2) the new minimum of 14 animals seems way overkill to me. For ticks on cats it is a little more understandable as those studies can be very difficult with ticks not really liking cats and cats possibly being groomers. We have run studies with 12-15 cats to ensure that we get the minimum of 6 that meet the 25% hold rate. Now if you insist on a minimum of 14 per group meet the minimum hold rate will we have to run 25-30 per group. Again, raising my question on group average or individual animal requirement. For ticks on dogs the idea of 14 animals being needed is absolutely ridiculous. Ticks like dogs and dogs are typically not good at grooming ticks off. My lab typically runs 8 dogs per group and havent had a tick infestation study failure in many years. I would suggest repeating your powers calculations using real data provided from past submitted studies." (0011)
The power calculations were run using data from past submitted studies and have been rerun for the tick models to account for some of the factors in the guideline and changes based on SAP comments. The number of animals recommended for tests for ticks have been reduced from 14 to 11. 




Charge Question 5
For each section listed below, comment on the proposed numbers of different arthropod pest species. EPA recommends these numbers based on review of studies available in the literature studies submitted to EPA, and analysis of power vs. sample size. Are there ways to reduce the number of pests used per test?
a. Fleas (section (j))
b. Ticks (section (k))
c. Mosquitoes and biting flies (section (l))

Panel Recommendations: 
143
Based on Tables A3.1 and A3.2 in Appendix III of the EPA supplemental document titled "Sample Size for Pet Product Studies," the required blood-fed/retention proportions (referred to as "base values" hereafter) play an important role in generating the number of pest species per vertebrate animal presented in the proposed guideline (last column of the table above); these base values are likely determined by the biology of pest-host interactions. For example, pests that are "aggressive" feeders on the vertebrate host tested may have a much higher host retention proportion than pest species with lower host feeding success. Accordingly, one way of reducing the number of pests per animal for testing is to "modify" these base values based on the "observed" (i.e., historical) information on pest-host interactions. Despite the importance of these base values in terms of sample size and power determinations, the proposed guidelines provided minimal information on their sources and biological rationales. Hence, the Panel recommended that the Agency should provide support of these base values or reexamine these base values for potential modifications; the potential sources of historical information of host-pest interaction are open literature, e.g., Taenzler et al. (2016); studies submitted by the pesticide registrants (since the inception of the 1998 testing guideline); and professional judgment from experts of these ectoparasites.

In particular, the Panel recommended that the EPA provide a clear justification for the arthropod "% retained" for inclusion of a vertebrate animal in the study. Using a higher "% retained" would decrease the number of vertebrate animals needed.  PG 33
The Agency agrees that the pest-host interactions played a role in the numbers of animals generated. The models were based on real data submitted to the Agency from studies that followed the methods chosen for the guideline and incorporated the retention rates, which were also chosen based on data typically submitted. The % retained was also generated based on historical numbers in submitted tests. 
144
The Panel further recommended that the EPA should carefully consider whether there are interactions between arthropod species when making sample size recommendations for scenarios in which multiple species are permitted to be tested on the same animal. Input should be sought from subject matter experts (biologists) on these possible interactions. For mosquitoes, two experts (personal communication with Dr. Dan Kline, USDA, and Dr. Bill Reisen, UC-Davis) anticipate no problems in testing multiple species simultaneously (see response to Charge Question #18).

The statistical models in the power vs. sample size calculations did not consider interactions between arthropod species (the species were treated in the models as not interacting with each other). Significant interactions among different pest species could result in reduced retention of each pest species (from that assumed in the power analysis) and potentially could also impact the effectiveness of the treatment (e.g., effective when single species but not for the same species when with other species [i.e., the multi-species scenario]). Thus, the Panel observed that interactions could have important implications on power (e.g., reduced power with reduced retention) and could bias the estimated efficacy for a single species when assessed from a multi-species study.  PG 33-34
The suggestion to use multiple pests for testing was refined based on recommendations by the SAP, the public and peer reviewed literature to mitigate negative interactions between the pests which may affect (e.g., reduce) retention. 

Based on this comment, the mosquito section was not changed and will continue to allow up to 3 mosquito species tested at a time. 

The flea and tick sections were revised to indicate that a flea infestation may be combined with an infestation of 1 species of tick, however it is not encouraged since grooming or scratching due to flea bites may detach ticks.

As for mixing tick species, the recommendations will be revised to state that Ambylomma maculatum (Gulf Coast tick) should be tested alone due to the effect of their bites. Also ticks with similar appearance or markings (before or after feeding) should not be tested together, since it may lead to counting errors.
145
In addition, for the larger numbers of pests per vertebrate animal, the Panel expressed concern about the logistics of counting (and subsequent errors).  PG 34
The pest numbers presented in the guideline are not larger than what is typically used in tests currently, so the Agency does not believe counting errors are a concern. 
146
4a. Below, the Panel provided Specific guidance on each arthropod.
The Panel had considerable discussion about fleas regarding the number of live fleas recovered in controls and flea retention. The literature is clear that visual thumb counting is vastly inferior to combing techniques. However, Dryden et al. (1994) provides a possible method by which the actual number of fleas could be estimated from thumb counts. This would be relevant to the earliest time studies in the guidelines. Dryden et al. (1994) also provides a study comparing counting procedures. Visual counting (six regions of a dog) was compared with a 10-minute count using a flea comb. After the counts the animals were sprayed with an alcohol based pyrethrin spray and combed for 10 minutes and the number of fleas counted. These animals were retained for 24 hours to determine if any additional fleas were counted. Area visual counting accounted for 21.5-28.0% of the fleas placed on the dog and combing counts provided 76.0-85.6 % of the fleas placed on dogs. After applying the pyrethrin sprays, 82.5-86.4% of the fleas were accounted for in the study at 24 hours. In a cross over study, Heckenberg et al. (1994) found that thumb counting found means of 8.8% and 7.7% on dogs infested with 50 and 100 fleas, respectively. Comb- counting provided 67.6% and 75.4 % of the fleas placed on the dogs. In another study, combing for 5, 10, and 15 minutes provided at least 81.5, 90.5, and 85.1% recovery of fleas placed on beagles 1 hour earlier (Zackson et al., 1995).  PG 34

For fleas, the Panel recommended:
1) Only short-haired animals like beagles should be used in the tests. Issues regarding long- haired vertebrate animals and environmental issues like sunlight or rain could be addressed in field studies.
2) Comb counting should be used based on the technique from (Dryden et al. 2016).
3) If beagles are used in the studies, retention estimates needed for the models might be obtained from the study by Dryden et al. (1994).
4) The methodology used by Dryden et al. (1994) should be repeated with short haired cats. Instead of the alcohol-based pyrethrin spray, nitenpyram could be used on the cats to remove all fleas. The use of an Elizabethan collar should also be considered as it can prevent cat grooming behaviors so its value for increasing flea retention can be determined.
5) The Charge Question regarding the issue of counting fleas for "the earliest time (e.g., 8 hours, 12 hours, etc.) needs to be revisited when EPA revises the guidelines. All on-animal counts prior to the final count may be conducted using hand counts. Values can be estimated for live fleas on beagles using the Dryden approach (Dryden et al., 1994).
6) A similar study based on the approach of Dryden et al. (1994) with short-haired cats should also be conducted.  PG 34-35
Based on SAP and public comments, hand counting options were removed from the guideline for fleas since they are considered unreliable. They will remain for ticks, which move differently than fleas and are therefore easier to hand count. The guideline was also revised to allow comb counts for fleas at those earlier time points, during which the fleas are removed then re-infested minutes later, with the stipulation that the comber avoids the treated area (so not to mechanically spread the solution) and treats the controls in the same manner with combing out and re-infesting.  
147
4b. Below, the Panel provided Specific guidance on each arthropod.
The use of 50 ticks per test was viewed as reasonable since decreasing the number of ticks per test would result in requiring more than the currently recommended 14 vertebrates per group (which was a concern  -  see response to Charge #4).

For ticks, the Panel recommended 50 ticks per test.  PG 34
Thank you for the concurrence on the number of ticks used per test. No change to the guideline is necessary. 
148
4c. Below, the Panel provided Specific guidance on each arthropod.
For mosquitoes, the rationale for using 50 per species when testing three species simultaneously, and 100 for individual species tests, is unclear. Unless there is a justification for the larger number for individual species tests, the Panel recommended 50 for the individual species tests, as well.

For mosquitoes, the Panel recommended 50 for the individual species tests.  PG 34
The intent of this text in the guideline was to indicate that if 100 mosquitoes or flies will be used in a study, then fewer animals (only 6) should be tested for each group.  However, if 50 mosquitoes (or specified flies) are used, then 8 cats or dogs per group should be tested. This section was revised for clarity.
149
4c. Below, the Panel provided Specific guidance on each arthropod.
For biting flies, there was considerable concern about using 50 flies per test (in terms of animal safety and welfare). Twenty-five per test was viewed as a more humane number of pests.

For biting flies, the Panel recommended 25 biting flies per test; however, the impact on power and sample size would need to be assessed.  PG 34
Based on numerous concerns from the SAP and public on using 50 painful or aggressive types of biting flies per test, the model was re-run for stable flies using 25 flies per test since they are the most painful/aggressive. Based on that model, the guideline has been revised to indicate that for studies conducted using 50 stable flies per animal8 animals should be in each group; for studies conducted using 25 stable flies per animal 12 animals should be in each group. In addition, the exposure time has been reduced from 1 hour to 30 minutes for all biting flies to reflect the study used in the model. Currently the Agency is only aware of one study conducted using stable flies on dogs and does not expect many future submissions but understands that any product that would qualify for claims against such aggressive/biting flies would be valuable.
150
The Panel also provided the following, general recommendations:
The Agency should provide further documentation of the validity of the parameter values in the simulations based on historical information regarding host-pest interaction from the literature or studies submitted to the EPA in the past.  PG 35
The models using 100 fleas were based on Beugnet et al., 2011.  The Agency relied on historical data to inform the 40% for cats and 60% for dogs control retention of fleas. The efficacy threshold of 90% by 48 hours was based on the Agency's 90% standard to support product registration.

The models using 50 ticks and 25% retention on control animals were based on Taenzler et al., 2016. The efficacy threshold of 90% by 48 hours was based on the Agency's 90% standard to support product registration. 

The models using 100 mosquitoes per cage and 60% blood-feeding rates for control mosquitoes were based on Meyer et al. 2003, Hayasaki et al. 2009, and Franc et al. 2012. Although 50 mosquitoes per species (150 total mosquitoes per cage) was not a parameter found in published studies, the Agency conducted these simulations to attempt to reduce the number of animals used in a single test. The efficacy standards after 1 hour of exposure was based on the Agency's 90% standard to support product registration.

For sand flies, the models were based on Halbig et al. 2000 for 100 flies and Liénard et.al. 2013 for the 60% control blood-feeding rate. The efficacy standards after 1 hour of exposure was based on the Agency's 90% standard to support product registration.

The models using 50 stable flies were based on Fourie et al. 2006.The efficacy standards after 30 minutes of exposure was based on the Agency's 90% standard to support product registration.
151
The Panel also provided the following, general recommendations:
Simulations should be conducted using lower numbers of pests, particularly for biting flies and fleas, and improved estimates of retention rates, to assess the implication on power and sample size (i.e., the number of animals needed).  PG 35
The original simulations were run using the best estimates of retention rates based on available public literature; therefore, they were not re-run with different retention rates. The simulation for stable flies has been re-run using 25 stable flies. The guideline was revised to include options of using either 50 stable flies per animal and 8 animals per test or 25 stable flies per animal and 12 animals per test.  
152
The Panel also provided the following, general recommendations:
For multiple species studies, when making an assumption of no interaction, the lack of interaction should be carefully documented from the literature. Otherwise, the power vs. sample size simulations and their validity would be suspect in the presence of interactions.  PG 35
See response to comment #144. 

Public Recommendations (docket ID # in parentheses):
153
"The number of parasites proposed within the draft (i.e., 100 fleas and 50 ticks) is considered appropriate. For consideration, reduction in number of pests would lessen resources required for the study; however, reduction would also lead to smaller recoveries on control animals thereby affecting efficacy evaluation" (0008)

The Agency acknowledges that the number of parasites is appropriate. 
154
Re: biting fly #: "What is the rationale for only using 50(+-5) biting flies per animal? Could a Sponsor justify using more than this number?" (0026)
See response for comment #148, however this number is the minimum recommended number; more could be used if desired.


Charge Question 6
Comment on the timing of exposing vertebrate test animals to fleas and ticks.
a. Are the number of exposures for products with different durations of efficacy adequate for determining efficacy? Why or why not?
b. If the number of exposures to pests can be decreased, please indicate specifically which exposures can be skipped. If efficacy should be evaluated at more time points, indicate when exposures should occur and for what product types (e.g., spot-ons, collars, and residual shampoos), and discuss the value provided by the additional time points.

Panel Recommendations: 
155
The Panel recommended that the number of exposures should adequately reflect the actual claims of the products. For example, if a product claims efficacy for 60 days, there should be data for at least 60 days. The exact number of exposures necessary will depend on the types of products, the claims, and how the Agency will evaluate the submitted data. If it is necessary to provide >= 90% reduction of fleas and ticks at the "longest labeled duration of efficacy" claim, then there is little need for the day 1 exposures. If the claim is "provides control (assume >90%) throughout a period of 60 days," then some periodic exposures are appropriate. In addition, for products with claims of effectiveness over several months, acquired immunity to tick infestations may occur, leading to host-induced tick mortality (European Medicines Agency, 2016; Wada et al., 2010). Therefore, infestation frequencies should be reduced to the minimum necessary to test the duration of product efficacy.

However, it is unlikely that removing some exposure times would actually change the methodology or the numbers of vertebrate animals being treated because the investigators will still design the study to generate data necessary to show efficacy and residual performance. Exposure tests will be repeated until the treatment fails to reach the necessary control level. The main factor dictating the numbers of vertebrate animals to be tested are the findings from assessment provided in the EPA document entitled, "Sample Size for Pet Product Studies."  PG 36
The first exposure period for mortality claims was revised to day -1 for fleas and day -2 for ticks. The exposure before the product is applied is important to determine if the product kills current infestations. Since the power analysis includes the 4 infestations periods in the first month, and that information is useful for interpreting the data, they will all remain. After 30 days, the infestations move to monthly to limit the testing frequency. 

For the mosquito and biting fly section, one of the infestations in the first month was already removed for products lasting >1 month and all weekly challenges were removed for products lasting over 1 month. 
156
6a. For products with efficacy claims of > 7 days, there is no need to expose treated animals at 1 or 2 days to fleas and ticks, respectively. Again, the testing protocol should reflect the specific claims. For research purposes, testing for efficacy occurs at multiple time points up to and past the maximum claim. Laboratory bioassays without animals could be conducted to test for breakdown of the product. For tick mortality trials, the Panel suggested changing the infestation timing in Table 2 to be Day -2 for ticks (not Day -1), to give adequate time for attachment, as outlined in European Medicines Agency (2016).  PG 36
The purpose of the initial pest exposure/infestation (which is on day -1 or -2 for mortality claims) and "up to 24 hours post application" for repellency claims) is to determine if the product begins working in an adequate time that provided adequate protection.  


 
157
6b. The Panel recommended that, depending on the specific product efficacy claims, at a minimum 14- and 21-day exposures could be removed for products with >30-day claims. The 1 or 2-day exposures could also be removed from >7-day claims depending on the EPA's goals for having 1-day data. The >4-week infestation timing is excessive; weekly repeated infestations of animals with fleas or ticks for the >4-week endpoints are unnecessary. Re-infestation of hosts every 2-4 weeks for longer duration product trials is common practice (Stanneck et al., 2012a, Wengenmayer et al., 2014; Jones et al., 2015; European Medicines Agency, 2016). For the >4- week group, infestations on Days -2, 28 and then monthly should provide the same level of information as weekly infestations. Timing of application for tick repellency testing should be similar to mortality trials at the >4-week endpoint.  PG 36
See responses to comments #s 155 & 156. 

Public Recommendations (docket ID # in parentheses):

158
"Table 1 [and Table 2 and Table 3]. For Claimed Duration of Efficacy of >4 weeks, the timing of infestation is excessive and puts undo stress on the test animals. The timing should be amended to Days -1, 14 & 28 or 30, then monthly up to the longest labeled duration of efficacy" (0008)

See response to comment #155.
159
"Consideration should be given to number and timing of exposures for spot-ons and collars with potential omission of mid-month infestation count(s)" (0008)
See response to comment #155.
160
"We propose that the duration of testing to demonstrate adequate efficacy should be dependent on the proposed claims of the product. On other words, long-lasting products would be required to demonstrate its sufficient duration. This will then allow for differentiation among available products and provide users the information for better product selection. The number of exposures should be customized to the profile of the proposed product. In other words, firms should be allowed the flexibility to customize the exposure intervals to fit their proposed product. Consultation with the Agency, in advance of initiating a development program, could provide alignment between the Agency and the firms towards agreeable study designs." (0026)
The proposed schedule is currently broken down for products with varying durations. The purpose of the guideline is to standardize the data and use the power calculations, which were based on the time points given, to ensure adequate statistical data sets. For more information, see responses to comment #s 155 & 156.


Charge Question 7
In the interest of reducing the use of vertebrate animals in testing, we have suggested in the guideline to allow testing of up to two species simultaneously for fleas and ticks (e.g., fleas and 1 tick species, 2 tick species). Are there any potential known interactions between multiple tick species or ticks and fleas if on a single vertebrate animal at the same time which would require individual pest species be tested separately? If there are interactions, please describe the interactions and provide references.

Panel Recommendations:
161
The Panel observed that testing multiple species on the same host could decrease the number of vertebrate animals to be used. While flea and tick co-infestations are commonly encountered on companion animals, this would make testing 1 flea + 1 tick species simultaneously acceptable and potentially even desirable. It is, however, unknown if increased grooming by cats in response to the presence of fleas might influence normal tick attachment. While retention is not expected to
be altered by co-infestation with two different tick types, some species, such as Rhipicephalus sanguineus and Ixodes scapularis, may be very similar in body size and markings after being attached to a dog for 48-72 hours. This may make the efforts to separately identify and count these two tick species too slow and cumbersome.  Dermacentor variabilis and Ambylomma maculatum (Gulf Coast tick) also share similar coloration. Additionally, A. maculatum is an aggressive species that causes painful, inflamed attachment sites. Thus, from an animal welfare standpoint, adding a second co-infested tick species with A. maculatum exposure would probably be excessively stressful to the test animal.

Regardless of the previous comments, when combining 1 flea + 1 tick or 2 tick species, would it be practically possible to adequately "tease out" (i.e., count) the effects of each of the two species from the same animal? And is it known if one species could potentially affect the survival rate of the other, thus influencing the experimental outcomes? If one of the two species used during a trial doesn't meet the test criteria, the guidelines should address what should be done with these data for the other insect species.  PG 37
Given SAP and public comments that normal tick attachment could be affected by increased grooming caused by itching from fleas in combined flea and tick tests, the encouragement of testing with fleas and ticks combined in a single test was removed. However, a study combining the two pests may still be acceptable if it produced good results.

From an animal welfare aspect, the guideline was revised to explain that Gulf Coast ticks should be tested alone, due to their potentially painful or inflammatory bites. 

Studies utilizing 2 tick species at a time are still encouraged in the guideline but will indicate that the ticks should be different in size and markings (before and after feeding) to ensure that they can be distinguished during counting.
162
When using more than one pest, the Panel recommended that the numbers of each species should be modified. Depending on the frequency of feedings and arthropod numbers used, the amount of blood consumed by a combination of ticks and fleas could be substantial, especially on control animals. Fewer ticks and fleas would need to be used on animals where multiple species are combined. There is also a potentially increased risk of inflammation at feeding sites that should be monitored. While 50 ticks total would likely be well-tolerated, doubling this to 100 ticks total to test 2 species may be excessive and infested animals should be monitored closely for any adverse reactions. In contrast, placement of 150 adult female mosquitoes (3 species x 50 mosquitoes each) will not likely lead to a risk of either undesirable mosquito-mosquito interactions or undue stress on the test animal.  PG 37
We acknowledge that this would be on the high side of a parasite burden, however animals are monitored for health during studies and it is not mandatory to run 2 tick tests at a time. Decreasing the number of pests would increase the number of animals used for testing.  
163
Finally, the proposed guidelines specify using pathogen-free ticks; however, the Panel noted that a more specific definition of this is needed. There is some concern that combining 2 tick species could facilitate dermal or sub-dermal pathogen exchange that might adversely affect test results.  PG 37
Based on numerous SAP and public comments, the specification of "pathogen-free" has been revised to "free of vector-borne pathogens" to be more specific. 
164
The Panel recommended allowing simultaneous testing of 1 flea + 1 tick species only, except when the tick is Amblyomma maculatum. The Panel has also provided the EPA with further information about tick species that have been used to co-infest animals without interactions that might hinder accurate data collection (see response to Charge Question # 15).  PG 37
See response to comment #161.

Public Recommendations (docket ID # in parentheses): n/a


Charge Question 8
 For pet collars, one of the typical methods of application requires cutting the collar to size and therefore the collar application rates are often dependent on neck size and not animal's weight as for other pet products like spot-ons that have dosing labeled based on weight ranges. Because neck size may not correlate with body weight or surface area, there is uncertainty that testing with the formulated collar sized for the neck may not represent the most conservative situations encountered in the field (e.g., dog breeds with small necks but relatively big bodies).
 :: Please discuss how application rate should be addressed during testing of collars. Is cutting the formulated collar to neck size sufficient for testing, even though the typical breeds used in testing may not represent the most conservative ratio of neck size to body weight/surface area? Why or why not? If not, how should this issue be practically addressed during testing?

Panel Recommendations:
165
The Panel found it unclear that the neck size to body weight/surface area or cutting off excess pieces of the collar after it is placed on the pet is having a negative impact on efficacy of the pests claimed on the label for each respective collar. There is limited peer reviewed scientific literature that discusses the efficacy of collars in controlling fleas or ticks.

The availability of collars meant for small versus large dogs (defined by weight of dog) would appear to be an effort to address this issue. Additionally, the active ingredients (a.i.) listed by % weight of a.i. on commercial collars is the same in collars for small and large dogs.

The registrant for the collar likely conducted initial efficacy testing to show that the collar delivers residual efficacy (flea and/or tick) for the desired length of time listed on the product label with any excess collar trimmed off prior to laboratory testing in both small and large dogs. Therefore, the excess collar that is cut off in pivotal studies mimics early testing and the desired length of residual flea/tick efficacy is confirmed, as well as, in clinical field studies conducted globally using dogs of different sizes and breeds (see Stanneck et al., 2012b).

Excerpt from Stanneck et al., 2012b:  Seresto(R) (Bayer Animal Health), a new collar for dogs and cats, provides long term broad spectrum parasiticidal activity by combining the insecticidal properties of imidacloprid with the acaricidal properties of flumethrin. The collar matrix system ensures that both active ingredients are slowly and continuously released from the collar towards the animal thereby avoiding peak concentrations and ensuring that acaricidal/insecticidal concentrations of both active ingredients are present in the cat's or dog's hair coat during the entire efficacy period. The active ingredients spread from the site of direct contact over the entire skin surface of the treated animal.

The Panel recommended that the EPA request that the registrant defend why they are instruct users to cut off excess collar material not needed to make contact with the neck region of the treated pet regardless of neck and body size (surface area). Additionally, the registrant should clearly indicate to the EPA reviewers how the a.i. is being released from the collar matrix, including distribution and/or release kinetics of the a.i (or multiple active ingredients).

There are two primary reasons for cutting off the excess formulated collar to the neck size. The first is that only the collar matrix that touches the skin/hair coat is going to release a.i. so any excess collar not touching the animal will not release and provide a.i to the animal so it's not needed. Removing the excess reduces the potential for exposing the pet, the pet owner, and the pet's surroundings to unnecessary amounts of a.i. The second reason is for vertebrate animal safety. Animal safety studies for the end-use collar formulation are tested with the excess collar cut off so that the animal is not exposed to excess amounts of the a.i. and to minimize potential accidental entanglement. Since some collars will break and detach as a safety feature, this may also lead to a potential health concern if the vertebrate animal ingests it.  PG 38
Since the neck size to body weight ratio are correlated for most dog breeds (except for a few outlier breeds), the current method of determining application rates of collars for efficacy testing (measuring and cutting off the excess collar) is likely the most accurate and practical method. In addition, for most active ingredients, the amount of active ingredient released from a collar is dependent on the collar material/plasticizer as well as the amount of rubbing along the animals' necks.


Public Recommendations (docket ID # in parentheses):
166
"HCPA believes that the testing currently required for registration is adequate for evaluation of product safety. Toxicity testing allows in depth consideration of potential target organs and adverse effects as related to the active(s). We acknowledge that the CAS population (e.g., Beagles, purpose-bred hounds, domestic shorthair cats) is not an ideal representation of the breeds to be exposed in the field; however, the evaluation of 5x exposures in the CAS study as designed to reveal sensitivities within the species provides a solid dataset for evaluation. Further, by their nature, collars are designed to slowly release actives to the fur and haircoat via contact with the animal over the course of the treatment period, in comparison to a spot-on product for which the entire dose is applied at one time. Finally, appropriate post-marketing surveillance ensures attention is maintained for potential safety issues" (0008)
The Agency agrees with this comment. Please see response to comment #165 for full response. 
167
"The problem here is that collar application rates cannot be calculated directly from the length of the collar applied. Since the collar represents an impregnated device with slow-release characteristics, active ingredient is only slowly released to the animal, and the application rate is a function of the collar material itself and any solvents that may be incorporated as well. In addition, a correction, if it were to be applied, would depend on a study of the variance in neck size and animal weight. That data does not exist to our knowledge, so no correction can be applied." (0033)
The Agency agrees with this comment. Please see response to comment #165 for full response.


Charge Question 9
Historically the Agency has requested studies for fleas, ticks, and mosquitoes on pets be conducted with 6 to 8 animals per treatment, although, the Agency has received studies with sample sizes ranging from 2 to 12 animals per treatment. Based on the statistical simulations provided in Appendix 1, EPA is proposing a minimum number of vertebrate animal subjects for each test type (i.e., pest) in the guideline. This number varies according to test type, ranging from 6 to 14, and is intended to achieve 80% statistical power (confidence) with a precision of 4 or 5%, depending on the specific methods employed for each pest. Please discuss and provide comment on:
a. The appropriateness of statistical methods to analyze the data and calculate percent efficacy and the associated 95% confidence interval.
b. The appropriateness of statistical methods and simulations EPA has developed to estimate the power of the proposed design, and specifically to achieve an adequate estimate of precision around the estimated mortality/repellency in the treated group for each section below:
    i. Fleas (section (j))
    ii. Ticks (section (k))
    iii. Mosquitoes and biting flies (section (l))
c. The assumptions used to inform the simulations (e.g., minimum blood-feeding rate of mosquitoes on control test animals, number of pests per animal) for each section. Specifically, please discuss any information that can be used to update or clarify these assumptions, especially where doing so would increase power with a smaller sample size (i.e., using fewer animals). Where possible, indicate which assumptions can be changed for each pest species or type of test animal and provide value estimates.
    i. Fleas (section (j))
    ii. Ticks (section (k))
    iii. Mosquitoes and biting flies (section (l))

Panel Recommendations:
168
Overall the Panel considered the proposed GLMM method to be consistent with commonly accepted statistical principles for point estimation (i.e., of efficacy); however, two Panel members showed a preference for conventional data analytical methods (e.g., geometric and arithmetic means) (Pfister and Armstrong, 2016) for characterizing the product efficacy, and the public comment by Dr. Kenneth Portier (see docket # EPA-OPP-2019-0161-0034) also suggested that the implementation and interpretation of the results of the proposed GLMM method may pose challenges to practitioners without advanced statistical training. Accordingly, given the importance of using valid methods (and their complexity), the Panel recommended that the Agency create generic or example "software," a step-by-step guidance document, and fully worked-out examples for implementing the GLMM method.  PG 40
EPA will add an example using synthetic data to show step-by-step data analysis for pest product efficacy studies to the document "Sample Size for Pet Product Studies". The updated version of this document will be provided later.
169
Proposed GLMM Method for Efficacy Determination
The Panel recommended expanding the justification of why day was included as a random effect in the GLMM scenarios. Also, the day random effects were assumed to be equally correlated in the proposed GLMM framework. The Panel recommended using a temporal correlation structure to model the correlation between different days, such as a decaying correlation as the difference between days increases (e.g., an autoregressive structure).  PG 40
EPA will expand the justification of why day was included as a random effect in the GLMM scenarios in the supplemental document.
170
Proposed GLMM Method for Efficacy Determination
The use of GLMM with a random animal intercept is essential to account for the non-independence of Bernoulli trials (i.e., the binary result on each arthropod) on an animal, and standard GLMs that do not account for correlations between counts within each animal should be avoided. Also, the random animal effects account for different animal characteristics (animals have different length, weight, grooming patterns, etc.).  PG 40
GLMM with random animal effect will be used to analyze the non-independent data (each animal has multiple measurements).  In the simulations to calculate the sample sizes, GLM was used to analyze a dataset on each day separately ONLY after the GLMM with random animal effect failed to converge.  However, the non-convergence issue with GLMM has been resolved (please see the attached memo of Dr. Cohen from ICF), and GLMM with random animal effect will be used to analyze data from studies where each animal has multiple measurements.
171
Proposed GLMM Method for Efficacy Determination
Other Panel suggestions for improving the proposed method included computing a population rate (as opposed to a rate conditional on the animal random effect) after fitting the GLMMs (in particular binominal GLMM with a logit link), ......  PG 40
GLMM with logit link that computes a population rate (as opposed to a rate conditional on the animal random effect) substantially improves the convergence of the model; however, the GLMM that computes a population rate is not able to account for the overdispersion in the data.  Therefore, a GLMM model that computes a population rate will be used only after a GLMM model that computes a rate conditional on animal random effect fails to converge.  
Some quick simulations showed that the power results were similar between the approaches (GLMM with conditional random animal effect model vs. GLMM for population rate); therefore, the simulations using GLMMs for population rate would not be conducted to re-calculate the sample size (please see the memo from Dr. J. Cohen from ICF).
172
Proposed GLMM Method for Efficacy Determination
Other Panel suggestions for improving the proposed method included...... the adoption of a Bayesian approach with priors to address issues such as uncertainty in "binomial" sample sizes and measurement error in counting (likely more of an issue of undercounting),....  PG 40
EPA may evaluate and consider the Bayesian approach later when there are more available data to validate the assumptions such as prior distribution proposed in the approach.
173
Proposed GLMM Method for Efficacy Determination
Other Panel suggestions for improving the proposed method included...... and the application of data submitted by the registrants to validate the proposed GLMM.  PG 40
See response of comment #172.
174
Proposed GLMM Method for Efficacy Determination
In terms of efficacy estimation from the GLMM, the Panel realized that the approach to deriving a point estimate for efficacy and effectiveness is based on a metric/statistic that has been used for many years where the difference between treatment rate and control rate is normalized by the control rate. However, the uncertainty in these estimates was computed using an asymptotic (large sample) standard error (based on the delta method) which may be problematic for small sample sizes. Accordingly, the Panel expressed concerns about the accuracy of this uncertainty calculation and the corresponding coverage of 95% interval estimates. Hence, the Panel recommended that the Agency consider alternative approaches to obtain improved characterization of the uncertainty and more accurate confidence intervals including bootstrapping or using an 'exact' fully Bayesian approach.  PG 40-41
EPA recommends using a bootstrapping approach to obtain the 95% CIs of the estimates when analyzing the data of each actual study.  However, the bootstrapping approach is not practical for simulations to calculate sample sizes of Pet Product Efficacy studies due to the large number of datasets in each of the simulations.  Furthermore, based on the results of small number of simulations (Dr. J. Cohen's ICF memo), the power resulting from using bootstrapping to obtain the 95% CIs of the estimates was slightly greater or similar to the power that resulted from using asymptotic approach to estimate 95% CIs of the estimates. Therefore, the sample sizes estimated from simulation using bootstrapping techniques would be similar to the sample sizes obtained from the simulation using asymptotic (large sample size) standard error.
175
Proposed GLMM Method for Efficacy Determination
One of the major concerns of the Panel regarding the proposed GLMM was the widespread lack of model convergence. The lack of convergence represents a major limitation for using the GLMM reliably for power versus sample size calculations. Further, an analyst outside the Agency may also have difficulty with GLMM model convergence when performing their own analysis of the data. To partially compensate for the problem of model convergence, the more proper binomial model (conditional on random effects) was replaced by a Poisson model (also conditional on random effects) in the simulations and a log pseudo-likelihood function was maximized (due to the true marginal likelihood not being able to be computed in closed form). However, the Panel noted that while the lack of convergence was not solved using this approach as the GLMM models also failed to converge in many cases, the convergence issue could be addressed by using more accurate estimation methods including the Laplace approximation and quadrature (both available in SAS) and "exact" maximization of true marginal likelihood using Monte Carlo methods (available in R). Furthermore, the choice between binomial and Poisson modeling approaches should be based on the appropriateness of the model framework for the type of data, and not on minimizing the lack of model convergence. In this case, the binomial modeling approach is more appropriate than the Poisson modeling approach. Accordingly, the Panel recommended the use of binomial modeling approach with a logit link function instead of the Poisson approach.  PG 41
EPA has revised both the guideline and the supplemental document to reflect this comment. The GLMMs for binomial distribution would be used to analyze the data of Pet Product Efficacy studies. The sample sizes in the guideline were updated with the results from simulations using GLMMs for binomial distribution.
176
Proposed GLMM Method for Efficacy Determination
Other Panel suggestions for resolving the convergence issue included analyzing the data with only the random animal effect.....  PG 41
The non-convergence issue has been resolved for most of the datasets (almost 100%), using one of the Panel's recommendations (please see response to comment #171). Removing day effect from the model to have a simpler variance-covariance matrix will be considered after the model that computes population rate fails.
177
Proposed GLMM Method for Efficacy Determination
Other Panel suggestions for resolving the convergence issue included...... and fitting the model separately by each of the "true efficacy of treatment" levels (0.95, 0.925, 0.90, and 0.85) instead of including day as an additional random effect in the model (fewer random effects should lead to fewer convergence problems).  PG 41
See response of comment #176.
178
Proposed GLMM Method for Efficacy Determination
To facilitate the implementation of the aforementioned suggestions, the Panel recommended consulting the reference by Diggle et al. (2002) and the following R packages (function performed in parentheses): MCMCglmm (fit Bayesian GLMM), glmm (fit GLMM using MC likelihood), bootstrap (bootstrapping), vglm (fit betabinominal), and Rjags (fit Bayesian GLMM with random effects distributions besides normal and complexities including measurement error etc.).  PG 41
EPA recommends using a bootstrapping approach to obtain the 95% CI of the estimates when analyzing the data of each actual study (see response of comment #174).
EPA has not been able to evaluate the recommended methods available in R packages; however, EPA has added language for submissions using any statistical methods that are appropriate for the data.
179
Also, the Panel agreed that power analysis using GLMM via simulation allows the determination of sample size for the efficacy testing of pesticide products; however, additional efforts remain to clarify the methodology and assumptions employed (please see below).  PG 40

Power Analysis Using GLMMs
The Agency performed power versus sample size analysis based on selected parameter settings and criteria for sample size determination, including (a) a true efficacy of treatment equal to 92.5%, (b) an observed efficacy of 90.0%, (c) a precision of 4%-5%, (d) 80% power, and (e) the blood-fed proportion of arthropod ectoparasites on each animal. The Panel noted that the Agency's analyses could be strengthened by including a detailed discussion of the rationale for the chosen settings. For example, animal group sizes could be reduced by requiring a greater true efficacy, less observed efficacy, or less precision. Also, increasing the number of arthropod ectoparasites per host animal will allow smaller sample sizes to be used. Guidance from subject matter experts on the proportion of arthropod ectoparasites retained would be very valuable.  PG 40-41
EPA has added more detailed discussion of the rationale for the chosen settings in the supplemental document.
180
Power Analysis Using GLMMs
One particularly challenging issue surrounds the declaration of "true efficacy" in the power calculations. In the simulations performed by the Agency, the true efficacy was set to 92.5%. The rationale for this true efficacy seems to be that setting the true efficacy to 90% with 5% precision and 80% power would result in required samples that are too large to be practical. Since setting a true efficacy equal to 92.5% in the power versus sample size calculation moves the Agency away from the previous standard of 90.0% efficacy, the Panel recommended discussing this constraint in detail.  PG 42
See response to comment #179.
181
Power Analysis Using GLMMs
Also, in standard (more common) power calculations, the statistical power might refer to an 80% chance of rejecting the null hypothesis of no treatment effect in a study when there is a true non-null effect. The Agency employed a different approach of considering accuracy in the estimation of efficacy with a pre-specified level of precision (5%); the Panel observed that this approach is a lesser known, but still an acceptable form of a power calculation.  PG 42
More discussion of the rationale for the approach of pre-specified precision power vs. typical power of testing a non-null effect will be added to the document "Sample Size for Pet Product Studies."
182
Power Analysis Using GLMMs
The Panel also realized that based on different parameter values in simulations (e.g., parameters change for different study designs in Appendix I of the "Sample Size for Pet Product Studies" document), the results of power analyses are highly variable with the power often larger than needed (which then results in a larger sample size than needed). The Panel recommended consulting subject matter experts to find other biologically plausible parameter settings that could lead to smaller sample sizes. For example, increasing the number of arthropod ectoparasites per animal would increase the study's power and potentially lead to smaller required sample sizes. However, as discussed in response to Charge Question # 5, it appears to be neither practical nor humane to further increase the number of arthropod ectoparasites per animal beyond the level already specified in the Agency's proposed guidelines (please refer to Charge Question #5 for exploring the options in reducing the number of animals used).  PG 42
The required number of animals has decreased, for example, from 14 animals down to 11 animals in studies of testing topically applied products against ticks, due to the assumption of generating the random effects from normal distributions rather than Weibull distributions. However, the option to reduce the required numbers of animals by increasing the number of pests per animals is neither practical nor feasible due to humane reasons.
183
Power Analysis Using GLMMs
Also, the Panel noted that alternative designs and (or) models could be explored to potentially decrease the number of animals needed or to address the problem of lack of GLMM model convergence. For example, the Rao-Scott approach (Rao and Scott, 1992) can be used to model correlated binary data exhibiting over dispersion and is computationally simpler than GLMMs; computer code for implementing the Rao-Scott approach are available in SAS and R (e.g., the raoscott() function in the aod package).  PG 42
Rao-Scott cannot be used to analyze the data of Pet Product Efficacy studies. There are two important reasons that Rao-Scott approach is not applicable for the data of pet product efficacy studies. The first one is that Rao-Scott approach is suitable for testing significance of the treatment effect but not for estimating the efficacy of the treatment.  Another reason is that Rao-Scott approach assumes the treatment effect does not vary from day to day, which is not true in the pet product studies where the treatment efficacy decreases over time.
184
Power Analysis Using GLMMs
As mentioned previously, the GLMM proposed by the Agency has both animal and day as random effects. GLMM modeling often assumes that the random effects are normally distributed, and SAS only allows random effects to be normally distributed. However, in the "Sample Size for Pet Product Studies" document, the power versus sample size calculations evaluated cases in which the random effects were simulated using normal distributions or Weibull distributions. The Weibull distribution is flexible and can assume many different shapes depending on the shape and scale parameters. However, with small sample sizes, or small changes to the Weibull parameters, the shape of Weibull distribution can change substantially. This difference in shape could potentially alter the sample size calculations. Tables A1.4-2 and A1.4-3 show that the GLMM based results can differ considerably when assuming the random effects follow a normal distribution or Weibull distribution. Accordingly, the Panel recommended using only the normal random effect assumption for any final sample size determination based on SAS PROC GLIMMIX and to use other software (e.g., Rjags) for non-normally distributed random effects.  PG 42
The guideline has been revised to reflect this comment. The sample sizes resulting from simulation in which random effects are generated from normal distributions will be used for the studies.
185
Appropriateness of the Assumptions Employed
Based on Tables A3.1 and A3.2 in Appendix III of the "Sample Size for Pet Product Studies" document, the blood-fed/retention proportions of 0.25-0.6 and the numbers of arthropod ectoparasites per animal of 50-100 were used by the Agency for performing the simulation exercises of sample size and power determinations. However, the Agency provided no information on their sources or biological rationales. Nevertheless, the blood-fed/retention proportions appear to be consistent with the literature values of >25% attachment rate needed for establishing an adequate tick infestation and >50% attachment rate needed for establishing an adequate flea infestation (Taenzler et al., 2016). Similarly, the numbers of arthropod ectoparasites per animal employed in the simulations are consistent with the efficacy studies involving ticks (50) (Taenzler et al., 2016), fleas (100) (Beugnet et al., 2011), and mosquitoes (100) (Meyer et al., 2003). Given the importance of the blood-fed/retention proportions and the numbers of arthropod ectoparasites per animal, the Panel recommended that the Agency verify these values based on information obtained from the open scientific literature, registrant submitted studies (since the inception of 1998 testing guideline), and experts of these ectoparasites.  PG 43
See response to comment #150. 
186
Appropriateness of the Assumptions Employed
As stated above, if an ectoparasite retention proportion parameter setting that can be used is higher than the setting assumed in the current models, smaller sample sizes may be possible. Also, the "actual" retention proportion can help in a case-by-case basis for "flexibly" deciding the number of animals needed for the product efficacy testing. For example, the tick Rhipicephalus sanguineus has a high expected retention proportion on dogs, in comparison to Ixodes scapularis, with a typically low retention proportion on cats, suggesting that the number of dogs used for the product efficacy testing with R. sanguineus can be reduced. The data for ectoparasite retention proportions on individual animals may be limited in the published literature. However, the testing laboratories may already have historical ectoparasite retention data collected and available to the Agency upon request. Average retention rates can also be informative for use in reducing the number of animals used, and some sources do cite individual animal data that may be useful for the Agency's consideration (Dryden et al., 2008; Halos et al., 2014; Dumont et al., 2015a, 2015b; Baker et al., 2016; Six et al., 2016; Vatta et al., 2019).  PG 43
Companies may use their own historic data to determine a sufficient sample size of adequate power to address the study objectives for specific pests. This clarification has been added to the guideline.
187
Editorial Comments
1. The use of "true efficacy" in the EPA document titled, "Sample Size for Pet Product Studies" is difficult to follow. For example, on page 7 (line 5 under the heading) the true efficacy is for sample size calculations was listed as "...true efficacy >= 90% with a precision of >= 5%..." But it appears that the sample sizes are based on true efficacies of >= 92.5% with a required efficacy of >= 90%, and a precision of 4 or 5%.  PG 43
The revision to reflect this comment will be made to the document "Sample Size for Pet Product Studies."
188
Editorial Comments
2. Is there a discrepancy between true blood fed proportion in control group of 0.45 in the table on page 7 and the listing of 0.40 under the "input parameters for the simulations" on page 8 of the document "Sample Size for Pet Studies"?  PG 43
The value 0.40 on page 8 was a typo.  It has been corrected and updated to 0.45.
189
Editorial Comments
3. The numbers under "Details simulations" on page 37 that reads in part "P*(1-TE) and equal 0.0475, 0.045, 0.04, and 0.3." It appears that "0.3" is an error and should be "0.03."  PG 43
The value 0.3 on page 37 was a typo.  It has been corrected and updated to 0.03
190
One Panel member suggested that, as an addition to the power analysis, the Agency could address the sample size issue in terms of number of animals per test via in silico approaches such as conducting "experiments" on computers using Monte Carlo simulation or Markov Chain Monte Carlo simulation techniques integrated with biologically-based modeling (Portier and Kaplan, 1989; Thomas et al., 1996; Bernillon and Bois, 2000; EPA, 2010). Considering that the original publication of the 810.3300 Guidelines was in 1998, this Panel member considered that the wealth of information possessed by the Agency could be used to strengthen the probability distributions of the parameters needed for conducting the suggested computer simulations, and therefore, may substantially reduce the number of animals needed.  PG 40 
A section about future use of NAMs has been added to the guideline. Although EPA did not pursue this for the guideline because of time constraints, using historical data submitted to the Agency in the suggested manner is an area the Agency may explore in the future.
191
The effort by the statisticians of the OPP to clarify the details of the math modeling and computer simulations in their work on SAS program of simulations (docket # EPA-HQ-OPP-2019-0161- 0003) on "Sample Size for Pet Product Studies," were very helpful and appreciated. Given the fact that the "1000 sets of data" were created by Monte Carlo Simulations, the EPA are already "doing experiments on computers." Therefore, the Panel further encouraged EPA colleagues to develop in silico approaches for the goal of reducing animal usage.  PG 43-44
EPA added language to the guideline encouraging the submission of protocols for NAMs, many computer-based models can be considered NAMs.

Public Recommendations (docket ID # in parentheses):
192
"As related to sample or group size, EPA claims to be committed to reducing animal testing in some cases while proposing more animals than currently used in others. Could the EPA please explain?" (0008)
EPA is committed to reducing animal testing while ensuring scientifically sound and efficacious products are registered for the market for public use. The numbers to be used are based on sufficient power analysis to address the study objectives which should decrease the number of studies that are currently deemed unacceptable and need to be rerun.  
193
"The "Sample size for pet product studies" document was simulation based on alpha values, false positive rates, and confidence intervals, yet lacked biological evidence or data sets to justify these values. :: The power analysis presented is based on a true random sample, yet every study will have stratification of treatment groups that will not be random. :: The data generated for a typical pet product will not be normally distributed, yet this is an assumption for most statistical models. Historically, there has been no requirement for statistical modeling, because there has been no need when utilizing the cut-off values of >=90% corrected treatment mortality. The Agency reported that typical studies received were between 2 and 12 animals per treatment. The reported range seems to be in agreement with the simulated data. :: Often a product meets the cutoff criteria and "works", or it does not. Manipulating data into a statistical model is not relevant when evaluating if products satisfy the cutoff criteria (>=90% corrected treatment mortality)... We believe that for most products being developed statistical models should not be required because meeting the efficacy requirements (>=90% corrected treatment mortality) at a given time point is sufficient to support that a given product is effective. Data generated in observance of these cut-off values would lack variation that would be informative in statistical processes. Utilizing the previously established minimum of 6 animals per treatment group (6 treated, 6 untreated), and the threshold efficacy requirements should give an adequate evaluation for product efficacy while reducing the number of vertebrate animals needed for testing." (0008)
The Agency disagrees that the statistical analysis recommended in the guideline document do not provide useful information.  The purpose of the power versus sample size analysis was to determine if the historical sample sizes were correct or if the sample size was inadequate. For testing against most pests, the historical number of 6 animals per treatment group turned out to be similar to the numbers generated using the statistical models; however, when evaluating efficacy against ticks on dogs the historical number of 6 animals per group is clearly not adequate to provide confidence that a result of 90% efficacy reflects the true efficacy of the product. Basing a sample size on historical precedent is not adequate and can lead to problems with underpowered studies, as would be the case if the Agency recommended 6 animals per treatment group for tick studies and falsely concluding that the product reaches the 90% efficacy threshold when it does not reach the 90% threshold. Registering pesticide products based on underpowered study designs which result in a false conclusion of 90% efficacy when the product is not 90% effective and could have negative implications for public health. Moreover, historically registrants have calculated 90% efficacy using several different methods which were typically calculated incorrectly and often did not provide a measurement of variance.  The recommended statistical model will not only calculate 90% efficacy appropriately but will also provide 95% confidence limits to allow the agency to assess the variability in the data. Because EPA is recommending the model for calculation, using the model will improve the confidence in and consistency of the studies submitted to EPA and promote a level playing field for all registrants.

194
"The Guidelines, as currently presented, do not make clear to the reader exactly which data need to be collected, but allow the registrant choices depending on the product's control outcome. For example, in section (j)  -  specific guidance for control of flea infestations, six conditions must be specified that establish when and what data are collected.
The number of animals (6 cats, 7 dogs) to be tested per treated and control groups, 
The number of fleas (100+-10) to which each animal will be exposed,
The timing and number of days (1, 2, or 6+) of exposure, 
The duration of exposure (24-48h), and,
When counts are recorded (up to 72 h post treatment, but "last count <=48h post infestation).
When re-infestations are applied, and outcome counts made for studies of long-term control.
Section(j)(3)i discusses the data collected for efficacy assessment. The guidance recommends comparing counts of "live fleas on the treated animals to live fleas on the control animals" or, in the case where killing ability is being assessed, comparing counts of "dead fleas on the animal and in trays/cages between treated and control animals".  This recommendation drives the choice of statistical distribution assumed in the final data analysis. If the product under test is fully efficacious and measurement is not too long after treatment, one would expect the live count to be very low which supports the assumption of Poisson distributed counts.
The form of the Poisson distribution applicable for a count can be affected by any number of factors that can and should be controlled for in the study design. It is critical that condition 2 above - similar numbers of fleas applied per animal, holds for all animals. Too much variation in exposure counts leads to expected counts with different Poisson distributional forms.  This could lead to the need to account for extra heterogeneity in the analysis (impacting the estimates of product efficiency).  Also important is that during the selection phase of the study, animals with wildly varying retention fractions are not used.  The expected retention fraction also changes the effective number of fleas to which a animal is exposed. If animals used in the study vary wildly in number exposed or in retention fractions, blocking on retention fraction or exposure number should be used in allocating animals to treatments and a block effect accounted for in the final statistical analysis.  While this will reduce the effective sample size (number of animals by number of treatments) it should also produce more precise comparisons of average counts (Poisson or Negative Binomial assumption, or success rates for Binomial assumption) between control and treated animals. Instead of blocking, the disparity in exposure levels may be accounted for in the final statistical analysis model, but this can make the analysis potentially more difficult to accomplish successfully from unanticipated computational issues leading to lack of convergence in model fitting.
Extra variability may also be introduced to counts made following re-infestation to assess claims of sustained product efficacy.  If all fleas are not removed (or killed) at the time of re-infestation, the effective number of fleas to which each animal is exposed will also be different. This is especially important in comparisons involving negative controls where little impact (e.g. mortality) is expected. For long-duration studies without flea removal from controls before re-infestation, the effective number of fleas on control animals would be expected to go up compared to an almost constant number of fleas on treated animals. The present guidance draft is silent on this issue. 
Section (j)(5) discusses data analysis. This section would benefit from much more detail. The guidance should provide clearer direction on when a Binomial, Poisson or Negative Binomial distributional assumptions should be used.  
Poisson distributed counts would be most appropriate for counts of successes (live or dead) from a large number of Bernoulli trials (e.g. when 100 fleas are tested per animal as recommended). The decision to include extra variation (e.g. assume a Negative Binomial distribution) in the analysis model could be made after consideration of whether there is wide variability in the effective number of fleas applied to each test animal. In the situation where all fleas on control animals survive and all fleas on treated animals die, each animal is a Bernouilli response.  With a fully effective product and tight study design this outcome in not completely infeasible. While the data are very convincing of product efficacy, statistical analysis would be difficult. The document should provide some guidance on what should be done in this situation. 
In general, any analysis using the binomial distribution requires data consisting of a count of number of pest on test per animal (e.g. total number of fleas applied) and a count of successes. This requires a clear description of what is a success (e.g. for efficacy study a success is on-animal and live, not moribund, for a killing study, moribund or dead, on or off animal).  Of most importance is the establishment of the total number tested. The uncertainty in the exact number (specified 100+-10 fleas) is less important than that animals have near similarity in the "effective number on test". In addition, if in the original acceptability assessment, a test animal is estimated to only retain 80% of fleas applied, one might assume that only 80% of fleas applied to the animal are potentially exposed to the treatment (an a priori assumption in a Bayesian formulation).  As mentioned above, complicating the analysis is the expectation that potentially 100% of fleas applied to control animal might be successes and 100% of fleas applied to treated animal might be failures. 
The guidance suggests the use of a Generalized Linear Model (GLM) for comparing treatment to control (more precisely, for estimating the statistical significance of the difference between treatment and control count "averages" or success "rates") in one-infestation studies.  These models improved on the logit (or probit) "regression" models which might have been used in the past and to which many biologists and risk assessors have been exposed.  The guidance document should provide more justification for why use of a GLM is preferred over a logistic model (e.g. relies on assumptions with greater face validity and results in better overall estimates and statistical tests).
Multiple infestation studies require a more complex model depending on the inference desired. The guidance recommends use of a Generalized Linear Mixed Model (GLMM) for this kind of study. With a GLMM, both model fit and interpretation become more difficult, often requiring the help of a statistician with more advanced training. A GLMM provides potentially a better way of estimating the statistical significance of differences in treatment of estimated control average counts or success rates, both at individual sampling time points, and overall.  The guidance is not clear on which of these two types of inference are preferred, or whether both should be provided. The improvements provided by use of a GLMM (over, say use of a GLM at each test time point) comes with increased complexity of model, the need for more assumptions that may not be easily checked, and the potential that the final model will difficulty computationally "converging" (e.g. fitting).  Biologists and risk assessors will be unfamiliar with these models and have difficulty with interpretation of results. The guidance should provide more details and in particular should provide worked examples where model formulation, fitting and parameter interpretation issues are illustrated and discussed. While extensive examples are not needed, I am unable to find well worked examples in the current literature that are accessible to and understandable by the Master's level biologists most likely to be conducting these studies. Finally, statistical packages that accomplish the computations needed for fitting a GLMM often have the capability to compute and display the parameter estimates and associated (approximate or asymptotic) standard errors in the original count metric (e.g. average count or proportion and not a logit or log odds) that are needed for tables in reports submitted for product review. The document should help the analyst by pointing to these options in the functions used in the more common statistical packages (e.g. SAS, SPSS, and R). Demonstrating the same examples using SAS, SPSS and R would help as well.
Section (j)(5) suggests using Abbott's formula to compute estimates of product effectiveness at each measured time point. There are multiple issues with this formula that need to be discussed in the document but are not discussed in this draft. 
First, if a Poisson-based analysis is used, the statistics produced are essentially averages, not rates. Abbott's formula is presented as a ratio of rates, so it is not clear how the output from the GLM or GLMM is to be applied in the efficacy estimate. It may be that the expectation is to use "raw" rates in Abbott's formula, but this then bypasses the benefits of using formal statistical analysis models for the estimation (and, as I discuss in the next paragraph, leads to further issues with computing a standard error estimate for the efficacy estimate) . As the sample size simulation study demonstrates there are often computational convergence issues with using a Binomial distribution-based analysis which would be expected to produce better rate estimates, and which would provide standard error estimates directly for these rates. What does the analyst do when faced with computational convergence errors?
The guidance document also recommends computation of the 95% confidence interval for the efficacy estimate computed using Abbott's formula. This is not straightforward since Abbott's formula is essentially the ratio of two estimates for which direct estimates of associated estimate uncertainty cannot be easily derived. While not mentioned in the guidance but pointed out in the second paragraph of the footnote on page 6 of the sample size simulation document, statistical software packages will not directly compute Abbott's formula nor do they provide standard error estimates and confidence intervals. As also pointed out in the footnote, estimates for standard error of the efficacy estimate can be approximated using the delta method which is based on a Taylor series expansion of Abbott's formula. But the delta method itself is not easy to apply. Typically only a first order Taylor's series approximation is used. This can be a very rough (that is, not very good) approximation and depends on estimated partial derivatives of the GLM or GLMM likelihood function in the region around the final estimates. Clearly estimated partial derivatives will not be available if the model does not converge.  If it does converge, the analyst needs to find, output and use these estimates in the formula of the delta method. Not every statistician/analyst will be familiar with the process of extracting these values or even with the delta method. This is another area where fully worked-out examples in SAS, SPSS and R should be provided in the guidance document or at least in an appendix to that document.
Section (k)(1) Test Species recommends that when testing for tick mortality and repellency a "laboratory study should be conducted in vivo with adults of three representative species of ticks." Further it specifies that more species of ticks may be used but "no more than two species should be tested on the same animal simultaneously."  Finally, section (k)(2) recommends no fewer than 14 animal subjects per treatment group be used and that animals be allocated to control or treatment groups using stratification based on attractiveness. [Note this issue does not come up with mosquito species since each animal is tested to equal number from the three major species.] 
The allocation of animals based on attractiveness represents restrictions on randomization which essentially creates "blocks" in the study design.  The blocking of experimental subjects will need to be accounted for in the final analysis model, something not mentioned in the data analysis section.
The restriction that no more than two tick species be used on any one animal brings up the issue of "balance" in the design.  To illustrate, assume 14 animals are selected from the study and three tick species are used. Allocation may look like the following.

Animal
Tick Species 
(number of Ticks)
Block/Strata

A
B
C

T-1
50
50
 
1
T-2
50

50
1
T-3
 
50
50
1
T-4
50
50

2
T-5
50

50
2
T-6
 
50
50
2
T-7
50
50

3
T-8
50

50
3
T-9
 
50
50
3
T-10
50
50

4
T-11
50

50
4
T-12
 
50
50
4
T-13
50
50

5
T-14
50

50
5
T-15
 
50
50
5

Note that with only 14 animals (T-1 to T-14) used, more of species A individuals (500) are tested than for species B and C (450) which can lead to a slight biasing of results to the response of tick species A to the treatment.  Without adding a 15[th] test animal the design is unbalanced. In addition, if blocks of say 6 animals per block (3 assigned to treatment and 3 assigned to control) are used, the last block would be incomplete leading to an incomplete block design and further issues with accommodating block effects. 
These issues of blocking, balance and incompleteness of blocks is not addressed in the simulations used to estimate the number of animals required for even the most basic of tick study.  Note that blocking can lead to more precise estimates and hence could result in the need for fewer animals (say 12 per treatment and control versus 14). Balance ensures that no one species has advantage over another in its impact on the final estimates.
Imbalance and incompleteness of blocks are not particularly difficult conditions to accommodate into GLM and GLMM analysis models, and hence would not be expected to keep the analyst from obtaining estimates. But imbalance and incompleteness can be easily avoided by better design (and in this case, the use of one additional test animal for both treatment and control), and the resulting average and rate estimates and hence the efficacy estimate will be more precise and potentially less biased.
Note that balance and completeness become more difficult the more tick species that need to be accommodated. With 4 tick species, 6 animals are needed to handle the 6 different unique pairs of species.  In this case, sample sizes of 6,12 or 18 animals per treatment would be used to ensure balance and complete blocks.
If blocking is used, it is not clear that each block requires the same number of control animals as there are treated animals for the design to be optimal. Savings in total number of needed animals might be possible if this situation is considered. Increased precision through the use of blocking is only possible if the factor used to stratify animals is related to the response (in this case the expected count or rate). If the relationship is weak, blocking will only marginally improve estimates and reducing the number of animals in each block receiving the control might lead to reduced precision in the comparison of treatments to control. This really needs to be researched (simulated) before this option were allowed.
Nowhere in the document is the issue of who does the counts discussed, although this may be considered part of standard GLP.  In most of these studies, counting is clearly time consuming and likely to be tedious. The temptation is to assign multiple individuals to this task to ensure the task is done within a reasonable time frame and without the biases caused by fatigue. But the use of multiple individual counters creates another factor that potentially can impact the data and the precision of the final estimates. Identifying counters and taking intra-counter variability into account in the final analysis can lead to more precise estimates of averages and/or rates and better estimates of efficacy.
It is admirable that the SAS code is provided along with the study report. This allows reviewers to rerun the simulations, assuming they have access to SAS and its macro facility (which I do not have).  But each SAS code document comes with a very low number of internal comments. Some of the variables (primarily bounds on key simulation conditions) used in each macro are defined in the comment block at the beginning of the macro. The flow of the macro is not well described making it somewhat difficult to follow exactly that is happening or to determine whether there might be issues with the simulations. As far as I could tell the simulation of the data follows closely the description provided in the Sample Size for Pet Product Studies document.
From the available code I was able to better understand the simulation by rewriting one of the SAS programs (Sample size for Pet Product Studies (normal).sas) as an R program (see appendix). In the time allowed for this project I was not fully able to replicate the Proc Glimmix analysis to the respective glm() or glmer() analysis in  R. This exercise reinforced in me the need for EPA to help analysts who may not have access to SAS to perform the needed analysis in other statistical packages.
I did note that three of the four simulation rely on Proc Glimmix with the poisson distribution assumption and the log link as the analysis scenario/model used to determine the significance of treatment effects. This means that the simulations do not provide sample size estimates for situations where the binomial or negative binomial distribution would be expected to be used as the basis of the data analysis. I am uncertain whether the recommended sample sizes based as they are on the poisson are adequate for these other situations. This should be explored before the final recommendations are published." (0034)
The guideline has been updated to more clearly distinguish what data should be collected and has been revised to only collect live fleas on treated vs. control animals. 

The parts of this comment about statistics differs in opinion than the SAP panel and the EPA agrees wth the SAP panel comments. 

195
"Analyzing data. If corrected mortality is required to meet 90% in the treatment group, the rest of this statistical analysis is irrelevant and should be omitted." (0008)
See response to comment #193.


Charge Question 10
Please provide comments on the overall clarity, accuracy, and completeness of the draft pet product guidelines. Please provide any additional comments that highlight areas of the draft guidelines that may need to be clarified and note any critical topics that are missing. Please include references to published literature that could help improve the completeness and clarity of the draft guidelines.

Panel Recommendations:
196
Given the importance of the blood-fed/retention proportions and the numbers of arthropod ectoparasites per animal, the Panel recommended that the Agency verify these values based on information obtained from the open scientific literature, registrant submitted studies (since the inception of 1998 testing guideline), and experts of these ectoparasites.  PG 44
The references used for blood-feeding/retention and the numbers of arthropods per animal were derived from peer reviewed literature and studies submitted to the Agency. They represent what is seen consistently in submitted studies that use methods like those described in the guideline. For mosquitoes and biting flies, values were obtained from Tiawirisup et al. 2007, Franc et al. 2012, Lienard et al. 2013, and Franc et al. 2015 which show mosquito bloodfeeding in the controls to be consistently at or above 60%. For fleas on dogs, retention rates of at least 60% of fleas has been shown in Lienard et al. 2013.
197
1. There is a mix of scientific names (genus species) and common names. The Panel recommended that the scientific name be used exclusively or at least at the first use.  PG 44
The guideline has been revised accordingly.
198
2. The term "Biting fly species" is not entirely accurate, at least for the list of flies presented. For example, face flies are not biting flies. Thus, either they should be removed from the list or the term modified. If the term is modified, to include face flies, then house flies should also be included. The Panel recommended that the language be changed to "mosquitoes and stable flies" and that some language about "evaluation of other biting flies would be considered on a case-by-basis (contact EPA for guidance)."  PG 44
The guideline has been revised to delete all flies except for stable flies and sand flies and indicate that all other flies would be considered case by case. 
199
3. The need for "outcrossing" is unclear and what is meant by this is not sufficiently documented. 
:: Outcrossing is commonly done when the investigator wants to retain a "field relevant trait", such as a parasitoids ability to hunt for food.
:: Outcrossing is avoided for toxicology studies because responses to insecticides do not generally change and having a standard reference strain is desirable and makes reproducibility feasible.
:: Thus, for the types of assays in the proposed guidelines, The Panel suggested that outcrossing is unnecessary and even perhaps detrimental. Using a standard strain is more desirable. Outcrossing changes the strain in unpredictable ways, making reproducibility problematic.
:: The Panel suggested that outcrossing be removed from the proposed guidelines.
:: If outcrossing is to be retained in the protocol, then more details are needed as to what strain should be used for outcrossing, what methods should be employed (e.g., reciprocal crosses?), how many individuals are to be used, etc. It seems that this level of insect husbandry is not needed for these protocols.  PG 44-45
The guideline was revised to remove the outcrossing recommendation and to allow the use of wild species if lab colonies are not adequate for a pest. 
200
4. The design of the repellency tests are worth reconsidering. The current method is a no-choice assay where the pests are applied directly onto the animal, whereas, the purpose of the test is to determine if a product prevents pests from associating with an animal. The Panel noted that alternative protocols could include adjoining cages where the pests can move from one animal to another (choice) or releasing the pests into the cage and determining if they are repelled (willing to get on the animal or not) (no-choice test). The Panel recommended that repellency assays be done independently of mortality assays.  PG 45
The repellency methods have been revised to include releasing the pests in a cage instead of onto the animal directly. 
201
5. The procedure for the assessment of blood feeding is a bit vague. Most blood feeding arthropods can take a full blood meal or some partial bit of a blood meal. The current assay ("crushed on a light background") suggests that even a tiny blood meal is counted. Is that desired or are people being asked to score the relative size of the blood meal taken (and if so, how)? The Panel recommended that blood feeding be determined by PCR so that the blood meal can be quantitated, and the species identified.  PG 45
The guideline has been revised to include the determination of blood feeding for mosquitoes only. For topically applied products, blood feeding from mosquitoes is considered as a failure. 
202
6. For shampoos, the Panel recommended that the formulation blank be used as the control because of the variability in shampoo products. Registrants will almost certainly have the formulation blank available to them.  PG 45
After consideration of SAP and public comments, the use of a non-pesticidal (vehicle) control to compare against for shampoo products has been changed to a negative control. See response to comment #237 for more information.  
203
7. Pests that have fallen off the animal can be hard to find. It is unlikely that 100% of the ticks applied can be recovered. The Panel noted that this presents some statistical challenges to dealing with the data. If 50 fleas are released and only 35 fleas are recovered, how will the statistical analysis be carried out?  PG 45
The guideline has been revised to count only live fleas or ticks on treated vs. control animals, therefore, the statistics should be run using live fleas or ticks. 
204
8. The Panel recommended that the title of the guideline be changed from "control" to "efficacy".  PG 45
The name of the guideline has been changed to:
Test Guidelines OCSPP 810.3300: The Efficacy of Topically Applied Pet Products Against Certain Invertebrate Pests

205
9. The Panel recommended that the suitability of a vertebrate animal to be used for the flea or tick tests be determined individually, but that the final control and treatment groups be compared as averages.  PG 45
The guideline has been revised to reflect this comment.
206
10. Although the guidelines address efficacy of new products, it appears that only pre- and posttreatment body weights are required. The same ingredients that kill ectoparasites can also cause a variety of neurological and physiological reactions in dogs/cats; scratching of bites can lead to inflammation and infections. Anemia has been reported in control dogs. The Panel recommended daily inspection of animals for health problems be conducted.  PG 45
Inspection of the health of the animals must comply with GLP regulations at 40 CFR part 160.9 and is not covered specifically in the guideline as it is not part of the efficacy methods. 

207
11. Page 7 of the proposed guidelines: the last sentence of d(3) Analyzing data, second to last sentence - "Software for analysis using GLMs or GLMMs is available in many widely sold statistical analysis packages." In this sentence, the Agency appears to be endorsing the use of commercially available statistical software packages; however, there are open source software packages (e.g., R) available to perform the same task. The Panel recommended that the guidelines clarify when specific software is required and when specific software is only a suggestion.  PG 45
The reference to "widely sold statistical analysis packages" have been deleted from the guideline. 
208
12. Page 9 of the proposed guidelines: first sentence of j(2) Selection and allocation of test animals  -  "The minimum number of qualified animals per group should be determined to achieve sufficient power in addressing the study objectives....." The Agency already employed the power analysis technique to determine the minimum number of animals required. Perhaps, the purpose of the aforementioned statement is to address the situation when the number of animals employed is different from (i.e., less than) the recommended value? The Panel recommended that this concept needs to be clarified. The same comment also applies to k(2) (page 12) and l(2) (page 16).  PG 45-46
To increase clarity, this statement was deleted form the individual pest sections and has been moved to the "Development of protocols for efficacy studies" section.
209
13. Page 11 of the proposed guidelines: first sentence of j(5) Data Analysis: "If the study has only one measurement (one infestation) per animal, GLMs for binomial distribution data (with log link function), Poisson distribution data, or negative binomial distribution data should be used to estimate the survival or hatching rate of each group." The Panel noted that this sentence is in contrast to the recommendation specified in section (d)(3) that techniques other than GLM or GLMM may also be used. Also, additional guidance on employing different distribution functions should be provided, e.g., betabinomial distribution for the "binomial" data exhibiting over dispersion etc. The same comment also applies to k(6) (page 15) and l(6) (page 18).  PG 46
The guideline has been revised to reflect this comment.
336  -  out of order
14. The Panel noted that the term "mode-of-action" as it appears on pages 14, 15, and 18 of the proposed guidelines needs to be defined in section (c). The "definition" of mode of action on page 15 (e.g., volatile repellent, toxicant) is not correct.  PG 46
The term "mode of action" has been changed to "effects on the arthropod pest."  
210
15. Page 4 of the proposed guidelines: (c) Definitions The definitions of "Knock down" and `Moribund" are very similar and difficult to distinguish. The Panel recommended they be clarified.  PG 46
The definitions have been clarified and the recommendation to count "knocked down" pests has been deleted.
211
16. Page 6 of the proposed guidelines: d (1), v. Housing of Animals: the Panel recommended that the permissive "may" be replaced with "should" with respect to group housing of dogs and cats at times other than during infestation.  PG 46
The guidelines already indicated that prolonged isolation should be avoided and used the word "may" to allow them to be housed together at other points. The recommendation by the Panel will not be incorporated into the guideline to avoid changing the way labs currently house animals or how they may handle incompatible animals. 
212
17. The Panel recommended that for analyzing and presenting data, the guidelines need to indicate that the arithmetic means must be used.  PG 46
This suggestion has been incorporated into the guideline.
213
18. The Panel noted that in the tick efficacy section, EMA tick efficacy wording should be used with in-house revisions.  PG 46
There are no specific examples given here with respect to which wording is in question, however the European Medicines Agency's guideline is cited and referenced. 
214
19. The Panel recommended inserting the phrase "adequately infested" when discussing minimum percent tick/flea retention thresholds.  PG 46
The suggestion has been incorporated to clarify retention thresholds. 
215
20. The Panel recommended that the guidelines define the different types of tests discussed. Specifically, the Panel suggested the term non-vertebrate animal testing (NAT, NVT, or NVAT) for tests that do not use the host. The Panel recommended that the terms in vivo, in vitro and in situ be avoided because they have different meanings in different scientific circles (e.g. some would claim that a NAT was an in vitro test, while others would claim it is not). Thus, the guidelines should avoid terms that have non-equivalent synonyms in different scientific disciplines.  PG 46
The term "in vivo" has been changed to "on animal" to provide more clarity.
216
21. The Panel recommended that the guidelines should state how many laboratory efficacy studies are required for each parasite that is included on the final approved product label. For example, FDA approved drugs require two separate pivotal dose confirmation efficacy studies for each specific parasite. Additionally, these two studies must be conducted at two different independent locations using different parasite isolates.  PG 46-47
The number of studies required depends on the specific claims desired. Adding this information would be outside the scope of the guidelines.
217
22. One Panel member indicated a desire to add a section on "The Use of Computational Technology" to encourage manufacturers to utilize such applications of computational technologies to reduce animal usage.  PG 47
See response to comment #98. 
218
23. Page 5: (d) (1) i. of the proposed guidelines: The guidelines should define in practice what the term "Negative control" should encompass.  PG 47
The definition has been added to the guideline. 
219
24. There are some formatting issues (for example, Tables break across pages; citations that need to be separated by a hard paragraph).  PG 47
Formatting errors have been corrected.
220
25. It is not clear if the words `must', `should' and `may' are always used appropriately in the guidelines. The Panel recommended that the use of these words be examined carefully.  PG 47
The Agency has confirmed that the proper terminology has been used throughout the guideline.
221
26. Section (k)(5)I of the proposed guidelines: The Panel suggested that the Agency consider reducing the number of categories of ticks examined from six to four. Efficacy should be based on the number of attached/free live ticks in the control group versus number of attached/free live ticks in the a.i. treated group.  PG 47
The Agency has made this revision. 
222
27. Section (k)(5)i top of page 15 of the proposed guidelines, The paragraph starting with "in situations:"
:: "In situations demonstrating the product has an acaricidal effect only after the tick blood fed (engorged)", is this a realistic outcome for a topical application?
:: The statement, "Dead ticks combined with live ticks to calculate mortality?" is confusing.  PG 47
The endpoints and counting recommendations have been revised to only include live counts on the animals. See response to comment #20 for more information. 
223
28. Section (k)(3)ii provides detailed information for repellency testing against ticks. The Panel recommended the equivalent information be provided for fleas if registrants also have repellency claims for fleas.  PG 47
The sentence stating that repellency should not be considered for fleas was deleted and language was added to encourage the submission of a protocol for flea repellency testing. 
224
29. Table 1: The heading of column #2 should be changed to "Timing of infestations and reinfestations."  PG 47
This change is not necessary as each reinfestation is an infestation. 
225
30. Section (l)(1)ii: The Panel recommended that the species that should be tested needs to be clarified. This would alleviate the need to contact EPA when registrants want to do these tests. Clarity is needed to understand if each species must be tested if it goes on the label. Also, does the Agency want only some species (or 3 of five species) tested to give a "biting flies" label?  PG 47
The guidelines indicate how to test a product's efficacy. The specific species necessary to gain a general registration will be indicated in a subsequent Product Performance Rule. 
226
31. Section (l)(2): The Panel recommended that the phrase, "caged and exposed to 50-100 adults" be changed to "caged and exposed to 50-100 female adults" since blood meals are the end point for this bioassay.  PG 47
The guideline has been revised to include only female mosquitoes for testing.
227
32. Section (m)(3)ii: Delete "for testing mosquitoes" from the end of the last sentence. This is already stated.  PG 47
The guideline has been revised to incorporate the suggested revision.
228
33. Page 6: viii. c: The Panel noted that in some field studies, it might be necessary to use a positive control for animal welfare concerns; therefore, the statement that "a positive control should not be used" should be modified.  PG 48
The guideline does not provide any tests where a positive control should be used. If a study would be conducted that would need a positive control, a protocol should be submitted with the use of the positive control justified. No change to the guideline is necessary. 

Public Recommendations (docket ID # in parentheses):
229
"We also note throughout the document usage of passive language in which registrants "should" do this and recommend more assertive and descriptive language to minimize ambiguity" (0008)
Since this document is a guideline and not a rule, passive language is used where appropriate, which it is most of the time. No changes to the guideline are necessary.
230
"In all sections we need better clarification on qualifying control sample numbers. So, for example in the fleas on dogs setup at least 60% retention is required, is this an average of the group total or individual animal basis. In other words, if I have 6 animals on study and 5 animals hold 90 fleas and the 6th holds 59 fleas for a count day. Average calculates to 84.8 but this study could be invalidated because of the 1 low count animal. This is an item which could be better explained and if true averages were used instead it could reduce the overall number of animals on study. Because of this possible invalidation due to random biological variability, studies are now run with 8 or 10 animals to ensure the minimum of 6 meet the requirement." (0011)
Based on SAP and public comments, these areas have been revised in the guideline to clarify the requirements. 
231
"I would ask the committee to carefully consider my two major concerns. First acceptable control number percentages in all section. I highly recommend these numbers need to be based on real numbers from submitted studies and scientific literature. Have to consider biological variances. Numbers good for 1 species might not be good for all. If it is set to high and applied to all species, studies will fail due to biology not because we are doing bad science. And in My opinion it needs to be defined as average of the control group not individual animal. If it is an average it will lower the number of animals used in trials as we can use the actual minimum and not need extra buffer animals." (0032)
The references used for blood-feeding/retention and the numbers of arthropods per animal were derived from open literature and studies submitted to the Agency. Maintaining higher retention rates also allows for better statistics and in turn lower numbers of animals needed per group.

The guideline has been revised to clarify that for pre-qualification of each animal, the retention is based on each individual animal and during the study is based on the control group mean retention. 

As noted in response to comment #196, mosquito and biting fly blood-feeding rates are based on studies in the peer reviewed literature.
232
"what seems to be proposed is a "Cadillac" version of test methodology versus a "Ford" version of test methodology. The EPA is not disputing that the products they have approved based on existing methodology are effective and valuable products for pest control in the marketplace. Yet it seems to be their opinion that many more details need to be specified, sample sizes need to increase, and statistical methodology should be required to determine if a product is effective. We submit that current methodology as practiced by researchers in the field is still valid, and represents a "Ford" version of studies without additional bells and whistles. This opinion (without the images) has also been expressed by a minority on the Panel, including some who have experience in the conduct of these studies. We urge the panel to resist the temptation to add requirements in order to feel like valid studies are being specified by the guideline" (0033)
The guideline has been drafted in consideration of study designs submitted to the Agency, while also aligning with the European Medicines Agency guideline where possible. The guideline is intended to describe acceptable methodologies to prevent study failures, thereby decreasing the number of studies performed. Other methodologies may also be acceptable. 
233
"Generally, the draft guideline overstates pest retention averages and understates sample sizes. In addition, assumptions/requirements for pest retention do not account for the variability encountered in actual experiments. While dogs may be qualified in a study to retain more than 25% of ticks (for instance), there is no guarantee that the retention rate will be the same throughout the study. (It is often less.) It is unacceptable for a study to be rejected when control retention at a sampling point is less than 25%, even though animals were properly qualified for the study." (0033)
See response to comment #231. 
234
"AHI encourages EPA to utilize a risk-based approach in determining study requirements. In this way, companies could consult with the Agency, in advance of initiating a development program, to reach alignment on agreeable study designs. Additionally, AHI proposes that the species of parasites to be testing and the duration of testing to demonstrate adequate efficacy against the selected parasites be dependent on the proposed product claims" (0035)
A protocol submission or discussion with the Agency is encouraged when novel approaches are proposed. Data requirements are based on the particular product and proposed claims. 

The species of parasites required are based on representative species that were chosen based on biological factors as well as disease transmission and will be provided in the Product Performance Rule separately, or registrants may contact the Agency. 
235
"Why are the WAAVP Guidelines (World Association for the Advancement of Veterinary Parasitology (W.A.A.V.P.) second edition: Guidelines for evaluating the efficacy of parasiticides for the treatment, prevention and control of flea and tick infestations on dogs and cats) not considered as a source document for the guidelines? They are used as a guidance document by FDA and other agencies and considering them would help make study designs more consistent across products and geographies, which could lead to fewer studies and less animals used." (0026)
The WAAVP Guidelines were referenced but due to a typo it was attached to another citation. It is still being referenced in the final version. 
236
"FDA and EMA require GCP studies. Can the acceptance of GCP studies for EPA be considered? This would allow for a single study to be used in multiple geographies, which would lead to fewer animal studies." (0026)
Good Clinical Research Practice or GCP studies are often submitted to the Agency and are widely accepted in place of GLP. 

The guideline states in (a)(4)i. "...However, with the exception that studies will comply with 40 CFR 160.90 to ensure animal welfare, studies that do not otherwise comply with GLP standards may be considered if, in the agency's judgment, the design and conduct of the study provide results that are scientifically reliable..."


METHODS FOR TESTING AGAINST FLEAS (SECTION (J))

Charge Question 11
The proposed studies currently are not blinded but untreated controls are used.
a. Should the study be blinded and/or utilize an inert control (i.e., formulation minus active ingredient)?
b. Why or why not?

Panel Recommendations:
237
The Panel concluded that the answer to this Charge Question depends on whether it is essential to identify all of the active ingredients within the product or formulation and to determine their contribution as it relates to the label claims. If it is necessary to identify the effects of formulations, carriers, diluents, ingredients, adjuvants, etc. on performance, then it will be necessary to conduct studies with the blank formulations. If only determining the performance of a formulation at some designated end point (e.g., >= 90% kill) is the primary objective, then conducting studies with "so-called" inert controls is probably not necessary, nor is conducting a blinded study. 

The Panel found that this question seems most relevant to products such as powders, shampoos, and dips in which carriers, adjuvants, emulsifiers, and other ingredients may in themselves have some lethal effects against fleas. For example, the product Avon(R) Skin So Soft is widely reported to have repellent activity against cat fleas (Fehrer and Halliwell, 1987). Contact with diluted household detergents and cleaning products is lethal to German cockroaches (Szumlas, 2002; Baldwin and Koehler, 2007). Components of soaps such as fatty acids (Sims et al., 2014) and linear alcohol ethoxylates (Sims and Appel, 2007) are toxic to German cockroaches. Aliphatic alcohols and mineral oil have been shown to be toxic to German cockroaches and combinations are synergistic (Sims and O'Brien, 2011). The Panel found that it is also likely that some of these potential ingredients may synergize conventional insecticides included in the formulated products and that these combinations will also be toxic to fleas.  PG 48
The guideline has been revised to remove the recommendation to use a non-pesticidal (vehicle) control for shampoos. See responses to comment #s 310, 311 & 317 below for more information.
238
To isolate and identify each of the compounds that may contribute to the toxicity of these formulations will require a considerable amount of testing. However, the Panel noted that it should be possible to conduct these studies without the use of vertebrate animals. Topical applications of serial dilutions of individual compounds and combinations of compounds in various formulations applied to adult fleas will provide data regarding their contact toxicity allowing for the determination of synergism, additivity or antagonism. Similarly, adult fleas confined to surfaces treated with serial dilutions of individual compounds and combinations of compounds in various formulations should permit the determination of residual activity and allow for the determination of synergism, additivity or antagonism.  PG 48
See response to comment #237.

Public Recommendations (docket ID # in parentheses):
239
"The types of studies being conducted should not require blinding as there is a due process to qualify the test animals prior to the start of the study (including determination of their attractiveness to the pest of interest). To incorporate a blind study design will negate the qualification process. Additionally, including an inert control is not necessary as the outcome of the study should almost always simply answer, `does this test formulation reach 90% efficacy' as compared to an untreated control. At the stage of in vivo experimentations, the question of the efficacy of the active ingredient itself should already be answered. Further, including an inert control group together with an untreated control results in an increase in test animals." (0008)
Thank you for your comment in response to the blinding of the study; we did not add blinding to the guideline. In response to inert controls, see response to comment #237.


Charge Question 12
Repellency is typically determined by whether or not an insect takes a blood meal. Based on flea feeding behavior, repellency is not considered relevant for flea control and therefore is not included in the guideline.
a. Please discuss whether repelling fleas is a viable endpoint. Why or why not?
b. If repellency can be considered a viable endpoint, please provide detailed methods to test for repellency against fleas.
c. Can repellency be determined by assessing the number of live fleas on treated vs. untreated animals? If so, can repellency based on counts of live fleas on treated vs. untreated animals be differentiated from mortality, which is also currently measured by counts of live fleas on treated vs. untreated animals? If the two endpoints can be differentiated please describe the methods, including specifically how and why the counts of live fleas on treated vs untreated animals differ for repellency and mortality, and provide literature references.

Panel Recommendations:
240
The Panel found that the statement, "Repellency is typically determined by whether or not an insect takes a blood meal," is not supported by the scientific literature. Repellents cause oriented movements away from the source. In relation to feeding, a better term might be a feeding inhibitor (Deletre et al., 2016). Even the European guidelines define repellent effect as "a product with a repellent effect will cause the parasite to avoid contact with a treated animal completely and/or to leave a host" (EMEA/CVMP/EWP/005/2000-Rev. 3). Flea repellents have never gained wide acceptance as a strategy to control fleas or protect vertebrate animals. Repellents have been better defined for tick control. Marchiondo et al. (2013) write, "Two types of repellency are defined: sensu stricto for repellency characterized by an irritant effect, causing the tick to move away from the treated animal or leading it to fall off soon after contact with the treated hair coat within ∼6 - 8 h and sensu lato for all other tick repellency (or expellency) up to ∼24 h. The first, repellency sensu stricto, may be attributed to the vapor phase of a compound or irritant effect through direct contact (for example, oil of Citronella but also some synthetic pyrethroids), while the second repellency (or expellency) causes inhibition of attachment or detachment of already attached ticks, (for example, some synthetic pyrethroids or amitraz)."

Pfister and Armstrong (2016) write, "Repellency refers to the action of a product that causes ectoparasites to avoid or leave the dog, or to fail to feed on the dog, i.e. the ability of the compound to prevent the parasite from attaching or migrating onto the dog. A similar concept is the "anti-feeding" effect, which refers to the ability of the compound to stop the parasite from taking a meal from the dog." Herein lies the problem. The Panel recommends that the authors of the guidelines not confuse/conflate repellency or avoidance and anti-feeding properties. The Panel recommended that the Agency provide a concise definition of repellency similar to that found in the European guidelines.  PG 49
See response to comment #296.
241
a. The central premise regarding this Charge Question is based upon the following statement in the proposed guidelines section (j). "Based on flea feeding behavior, repellency should not be considered relevant for flea control (Pfister and Armstrong, 2016)." The Pfister and Armstrong (2016) paper is a review of two types of insecticides and this statement may be true for the systemic fluralaner and cutaneous delivered permethrin, but that does not mean it is true for all compounds.

Flea repellents have been reported in the literature. Chemicals were tested for repellency (prevention of feeding) with the flea Xenopsylla cheopis (Bar-Zeev and Gothilf, 1972). Unfed fleas were allowed to feed on treated skin surfaces of guinea pigs. The fleas were examined for blood feeding. Of the 538 compounds tested, four compounds were as effective as DEET. The repellency of essential oils was evaluated in a filter paper strip choice test with adult cat flea, Ctenocephalides felis. Thymol and trans-cinnamaldehyde were as repellent as 15% DEET (Su et al., 2013).  PG 50
This statement and citation have been removed from the guideline. 
242
b and c. 
The Panel found that if there is a claim that a compound repels fleas, then repellency must be considered as the viable endpoint. The number of live fleas off host compared with the number of live fleas off host in control should provide a relative estimate of the repellency. If the compound also exhibits some contact toxicity, then both live and dead fleas may be present off the host.

Cats or dogs could be treated with the test compound. The treatments would be permitted to dry and spread in the pelage of the animal (12-24 hours). The treated animals would be held within cages constructed with Plexiglass to ensure that fleas don't escape and are collected. Aliquots of fleas (25-50) would be placed on the fur or placed in the cage and allowed access to the treated and control animals. At 1 hour, the contents of the cage would be vacuumed, and the live and dead fleas trapped.  PG 50
See response to comment #223.
243
Repellency is not a widespread label claim of many topical products applied to cats and dogs. The Panel noted that EPA has some latitude to accept non-vertebrate studies to demonstrate repellency or request protocols for determining repellency studies on vertebrates.  PG 50
EPA has included a statement in the guideline indicating that protocols may be submitted for new approach methodologies (NAMs), and that if a protocol is proposing test methods which do not use vertebrate animals, they should also propose to run a traditional (on animal) test concurrently to provide evidence that the new methods are useful surrogates for testing that utilizes vertebrate animals.

Public Recommendations (docket ID # in parentheses):
244
"Recent literature indicates that flea repellency can be assessed by counting fleas falling off the animal in kinetic after infestation (Varloud et al., 2015; Halos et al., 2016). https://parasitesandvectors.biomedcentral.com/articles/10.1186/s13071-016-1345-4  and https://link.springer.com/article/10.1007%2Fs00436-015-4470-7  Flea repellency can be assessed by counting fleas falling off the animal in kinetic after infestation (Varloud et al., 2015; Halos et al., 2016) https://parasitesandvectors.biomedcentral.com/articles/10.1186/s13071-016-1345-4  https://link.springer.com/article/10.1007%2Fs00436-015-4470-7" (0008)

The referenced literature was considered, and a statement was added to the guideline to encourage the submission of a protocol for flea repellency tests. 
245
"Yes, repelling fleas is a viable endpoint. Companies need to be given a chance to develop products and formulations that give the end user a variety of options in order to control flea populations. If the flea is never given the opportunity to take a blood meal on the host animal, there is less opportunity for transport of fleas from outdoor environments indoors. This would allow a reduction in cross contamination among multiple pet households and flea infestations on humans. b. This would have to be a defined endpoint in the protocol before trial initiation. Methods to assess repellency can be very similar to mortality assessments. Simply the presence or lack of presence of the flea is a very simple start. This method can be very similar to what is already outlined in section (5) Data collection and recording for ticks. c. Yes, it can be. The endpoint must be defined in the protocol before the initiation of the study. As the methods for evaluation of repellency versus mortality endpoints are similar, they must be tested by two separate trials (i.e., one trial to determine mortality effects and one trial to determine repellency effects). d. Flea repellency is relevant and given the oral market which does not provide repellency this is very important measure for consumers" (0008)
The guideline will be revised as indicated above in the response to comment #242. 


METHODS FOR TESTING AGAINST TICKS (SECTION (K))

Charge Question 13
The guideline states that tick infestations should consist of a 50:50 ratio of male to female ticks for all species. The European guideline contains a clause that for Ixodes spp. infestations the sex ratio should consist of approximately 10% males: 90% females since males do not readily attach (Marchiondo et al., 2013). However, the scientific literature for Ixodes spp. appears to be split between using a 50:50 ratio or a 90:10 ratio. Please discuss the merits of each ratio:
a. Is one of the ratios better for testing Ixodes spp., or are they both appropriate? If one ratio is better than the other, please explain why.
b. Is one of the ratios better for testing Rhipicephalus spp., or are they both appropriate? If one ratio is better than the other, please explain why.
c. Is one of the ratios better for testing Amblyomma spp., or are they both appropriate? If one ratio is better than the other, please explain why.

Panel Recommendations:
246
a. The Panel noted that it appears that for testing Ixodes spp., both ratios 50:50 and 90:10 (females:males) would be appropriate. In general, placement of a somewhat balanced number of both females and male ticks on hosts is important because tick mating while on the host promotes feeding success by female ticks to engorgement (Akov, 1982; Wang et al., 1998; Donohue et al., 2009). The role of mating on female feeding during the very early feeding stages is unclear. Insufficient numbers of males that may survive during trials could possibly confound testing results by causing females to poorly feed due to lack of mating opportunities. So, a 50:50 ratio is probably best for all species tested to ensure normal mating behaviors. 

The European guideline (European Medicines Agency, 2016) contains a clause that for Ixodes spp., there appears to be split between using a 50:50 ratio or a 90:10 ratio annotated below in the following references: Vatta et al. (2018), Becskei et al. (2017), EMEA (2016), Dumont et al. (2015), Fourie et al. (2015), Kuzner et al. (2013).

A reason some researchers prefer to use more female than male Ixodes spp. in testing is because male Ixodes spp. do not necessarily require a blood meal (attachment) to mate, which can skew the attachment rate statistics on untreated hosts, particularly when testing systemics.

Studies have shown that attachment rates of I. ricinus can be increased by using a higher number of ticks ( 60) and a lower number of males compared to females (∼25:75) as male I. ricinus attach only briefly or not at all (Kiszewski et al., 2001; Kuzner et al., 2013). However, Marchiondo et al. (2013) noted the presence of male I. ricinus is sufficient to stimulate female tick attachments and to achieve greater attachment rates.

"The metastriate ticks, including Dermacentor, Amblyomma, and Rhipicephalus, invariably attain sexual maturity and mate solely on their hosts. The more primitive prostriate Ixodes spp. ticks; however, may copulate both in the absence of hosts and while the female engorges. These expanded opportunities for insemination complicate the mating systems of the Ixodes ricinus complex of species. In these ticks, autogenous spermatogenesis must precede host contact..."
(Kiszewski et al., 2001).

In the U. S., Ixodes pacificus and I. scapularis mate on and off the host and adult males rarely attach to the host, taking little blood if they do. In general, the blood loss to a dog host is less than that taken by females and males of the other genera (Koch and Sauer, 1984). Placing 50% female and 50% male ticks (N = 50 ticks) challenges the host with 25 potentially attaching ticks, whereas 90% female to 10% male could lead up to 45 females attaching to the host. Especially if repeated infestations occur on the animal, or if more than one species of tick is used in a trial, the 50:50 ratio would cause less stress to control animals in particular. A 50:50 ratio would ensure adequate mating opportunities and normal feeding behaviors for females while being consistent with most previous studies for comparison (McCall et al., 2011).   PG 51
To allow flexibility, the tick section of the guideline has been revised to read:

Infestations should consist of a 50:50 ratio of female to male ticks (Dumont et al. 2015, Fourie et al. 2015,  Stanneck et al. 2012, Young et al. 2003, ), however other ratios may be justified (e.g., 90:10 female to male ticks ratio for Ixodes spp.)
247
b. The Panel noted that a 50:50 ratio is best for testing Rhipicephalus spp. As indicated above, with the metastriate ticks, such as Rhipicephalus spp., the males must attach and feed before they mate so there would be no statistical advantage to skewing the ratio of males to females away from 50:50.

Thus, for Dermacentor, Amblyomma, and Rhipicephalus, mating occurs only on the host, and both females and males feed. Males consume substantially less blood than females, but unlike Ixodes spp., they do take some blood. Using either ratio of ticks could lead to up to 50 ticks attaching. Placement of 50:50 would likely result in fewer adverse effects to the host during trials than 90:10, although not much difference would be seen if ticks are removed at 48 hours from the host.  PG 51-52
No changes to the guideline are necessary.
248
c. The Panel noted that a 50:50 ratio is best for testing Amblyomma spp. As indicated above, with the metastriate ticks, such as Amblyomma spp., the males must attach and feed before they mate.  PG 52
No changes to the guideline are necessary.
249
Although not included in the Charge Question, Dermacentor results should also be considered, as this is a common tick found on U.S. pets, in particular, D. variabilis on dogs.  PG 52

REPEATED FROM b:  Thus, for Dermacentor, Amblyomma, and Rhipicephalus, mating occurs only on the host, and both females and males feed. Males consume substantially less blood than females, but unlike Ixodes spp., they do take some blood. Using either ratio of ticks could lead to up to 50 ticks attaching. Placement of 50:50 would likely result in fewer adverse effects to the host during trials than 90:10, although not much difference would be seen if ticks are removed at 48 hours from the host.  PG 52
No changes to the guideline are necessary.

Public Recommendations (docket ID # in parentheses):
250
"For all three species, the ratio should be kept at 50:50. The point of efficacy trials is the eradication of the pest population in that given environment. Aside from a few parthenogenic species in the US, both males and females are needed for sexual reproduction. A decrease in both sexes could allow for an overall reduction in the population of the pest. A special caveat could be created in order to allow a full 100% ratio for species in which only females are present" (0008)
See response to comment #246.
251
"It is true that male Ixodes spp. ticks do not need to feed to reproduce, and sometimes Ixodes spp. ticks mate off host. For mortality and repellency studies, the more females used for infestations the more likely you will have adequate infestations on the control animals. That being said, you are also increasing the attachment burden on the treated animals which may result in more attached ticks in the treated group at counts." (0026)
See response to comment #246.
252
Rhipicephalus: "Males will move from host to host to find a female, feeding as they go. They also need to take a blood meal before they can mate. Therefore, a ratio of 50:50 is most appropriate." (0026)
No changes to the guideline are necessary.
253
Amblyomma: "Literature sources state, the presence of males increases the feeding behavior of the female ticks, an equal sex ratio of 50:50 is appropriate for this tick species. We recognize that the scientific literature is not fully aligned on this matter, and propose that a study should be deemed acceptable if the number of ticks in the untreated group of animals is sufficient to demonstrate that the host is able to harbor ticks, regardless of the sex ratio of the parasites used for the challenge. We believe that if a dog or cat can retain approximately 25% of the ticks used for challenge, this challenge should be considered valid for the purpose of demonstrating efficacy in a treated group" (0026)
See response to comment #246.
254
Ixodes: "This ratio is not appropriate for Ixodes scapularis ticks where ratio should be 40:10, female to male. In addition, as both sexes of all common species except Ixodes feed on the host, there is no advantage to handling the ticks to determine their sex. Infesting with 50 ticks, regardless of sex ratio, will produce a robust challenge." (0026)
See response to comment #246.


Charge Question 14
The guideline states that ticks that feed to engorgement can transmit pathogens that may cause disease; therefore, mortality after blood-feeding to the point of engorgement should not be considered a viable endpoint.
a. Is the determination of engorgement a feasible data collection point? Why or why not?
b. If so, please provide detailed methods with references for determining engorgement.
c. Can blood feeding be determined prior to engorgement and would this be a viable endpoint? If so, please describe methods for determining blood feeding prior to engorgement and provide any relevant references.

Panel Recommendations:
255
a. The Panel concluded that, no, the determination of engorgement is not a feasible data collection point. The Panel recommended that the term "engorgement" needs to be revised in the guidelines as the EMEA and WAAVP guidelines are not being used correctly. Adult tick counts are generally performed 48 hours after each infestation time point, but initiation of blood feeding during this period will not be readily apparent as very little blood is consumed within this time frame. It is impossible to accurately visualize any increase in size (plumpness) over this short period. This is even difficult if individual ticks are weighed (milligrams). All instances of "engorgement" in the guidelines should instead be changed to "partially fed." If feeding to engorgement and drop off is to be mentioned, then the term "repletion" should be used. Throughout the guidelines, for clarity, it should be specified that all ticks found on the host (attached and unattached) are removed at 48 hours post-infestation, not just counted. This is in addition to off-host ticks that are also counted and collected.

The efficacy of systemically-acting acaricides depends on tick attachment and ingestion of the compound. Transmission of bacterial/protozoan pathogens by hard (ixodid) ticks typically starts to take place between 36-72 hours after attachment, before substantial blood feeding occurs in adults. If the kill occurs prior to the 48 hours tick removal time point, this delay after attachment can allow a systemic acaricide to take effect prior to transmission of many pathogens (Uspensky and Ioffe-Uspensky, 2006; Wengengmayer et al., 2014). However, viruses such as Powassan fever virus could be transmitted within minutes to hours after attachment (Richards at al., 2017).  PG 52-53
The use of engorgement as an endpoint has been removed from the guideline and replaced with counting live ticks on the treated animal vs. the control animal for mortality testing and counting all ticks on the treated animal (live or dead) vs. the control for repellency testing, which is how these studies have been historically conducted and is consistent with the European guideline.  
256
b and c. 
The Panel found that engorgement status does not appear to be relevant to these revised guidelines. Researchers are assessing survival of partially fed ticks after 48 hours of exposure (contact) with a topically applied acaricide. As long as female ticks fail to survive to repletion and oviposition, the lifecycle is interrupted. As noted by the Panel in response to question 14a., all instances of "engorgement" in the guidelines should instead be changed to "partially fed."  PG 53
The guideline has been revised to incorporate this comment. 
257
b and c. 
If blood feeding status and duration of attachment are desirable factors for protocol guidance, effective methods have been developed to determine these parameters. Blood consumption during the early feeding phase can be verified by PCR, as well as, identification of the host blood meal source (Che et al., 2015). However, depending on the sensitivity of the assay, it is also possible to detect remnant blood from the host on which the adult tick previously fed during the nymphal stage. The duration of tick attachment and phase of blood feeding could also be scored morphologically using tick engorgement indices based on comparisons of scutal and full body dimensions = "scutal index" (Yeh et al., 1995; Falco et al., 2018). The approximate timing (in days) for initiation of blood feeding by adults feeding on mammals has been established for Ixodes pacificus/scapularis, D. variabilis, A. americanum, and R. sanguineus. Prior to and during the early blood feeding phase, when other fluids are primarily being taken up by the feeding tick, the tick cuticle does not look substantially different in color and no inflation of the body size is evident. Physically, the tick cuticle begins to change color and the tick becomes inflated during the initiation of the rapid blood consumption phase. On the day this occurs, the change in appearance may be quite rapid over the course of several hours as the "big sip" commences (body typically turns a dark gray color as it inflates).

As an example, an adult Ixodes ricinus female takes ~7-8 days to feed to repletion. The slow feeding period occurs between ~1-5 days after attachment, during which the female ingests ~1/3 of the total blood meal. Between 24-72 hours very little blood is ingested. About 2/3 of host blood is consumed during the rapid engorgement phase ("big sip"), which occurs during the last 24-48 hours before the engorged tick drops off the host (~6-8 days post-attachment) (Figure 1, Franta et al., 2010; Sojka, 2015).  PG 53
The Agency appreciates the background information. However, based on previous SAP comments, the determination of tick feeding status has been removed from the guideline because the timing of the count/removal of ticks would occur before engorgement takes place. 

Public Recommendations (docket ID # in parentheses):
258
"Define what constitutes an "engorged" tick. Partially engorged, fully engorged?... "Blood-feeding to point of engorgement" is not adequately described; is it assess by scutum enlargement, increase in weight, etc.? Recommend including additional detail on "point of engorgement" to better assess criteria." (0008)
Engorgement as an endpoint for ticks has been removed from the guideline. 
259
"A fact that I also disagree with in the WAAVP and EMA guidelines is discussion of engorged ticks. The study designs are setup with maximum removal time of 72h and most often within 48h. Ticks will not show engorgement within this amount of time. I have even performed squish test in my lab and could not tell the difference in body fluid results between an unfed female and a 72hr attached female R.s. tick. At most 4 categories should be counted attached and unattached in each of Live & Dead..."All references to engorged ticks should be removed. Engorgment takes 4 to 7 days and we are removing them at most 3 days after infest. You can not determine engorgement in that short time." (0032)
Engorgement as an endpoint for ticks has been removed from the guideline. 
260
Re: determination of engorgement: "No, it is a subjective determination. Males do not feed to engorgement, so if you are measuring engorgement you will only have engorged females. Additionally, pathogens are transmitted well before full engorgement occurs (usually within 48 hours of attachment). In addition, ticks will not engorge in the time that they are on animals in these studies (typically only 48h) so it is not possible to accurately assess engorgement in this study design." (0026)
Engorgement as an endpoint for ticks has been removed from the guideline. 
261
"Yes, blood feeding can be determined without engorgement. Squish the tick between two clear pieces of plastic and see if blood comes out. This is a similar method for what is used to determine the fed status of some biting flies. However, presence of blood does not equal engorgement for ticks. This is unlikely to be a useful endpoint for tick studies." (0026)
Engorgement as an endpoint for ticks has been removed from the guideline. 


Charge Question 15
The endpoints for repellency and mortality contain numerous different categories which define how ticks should be recorded for each endpoint (e.g., repellency).
a. Are there any additional categories which might be needed for the different endpoints (e.g., repellency)? Is the separate presentation of the endpoints clear?
b. Can both repellency and mortality be assessed in a single study, or does each endpoint need to be assessed separately? Please explain.

Panel Recommendations:
262
a. The Panel did not think there is a need for more categories; rather, the Panel recommended a reduction in the number of categories. While the multiple categories offer some potential increased resolution into the responses of the ticks to product testing, the number of categories would be unrealistic in practice.  PG 54
The Agency agrees with this comment and the number of categories has been reduced.
263
Efficacy Endpoints
The proposed guidelines state: "Mortality should be determined by comparing dead, unengorged ticks on the animals and in trays/cages to the control group. Ticks that feed to engorgement can transmit agents that may cause disease; therefore, mortality after blood-feeding to the point of engorgement should not be considered a viable endpoint."

The Panel suggested revising this to indicate that mortality at or before 48 hours would be a reasonable endpoint for mortality to prevent pathogen transmission, if that is the goal.  PG 55
The guideline has been revised to incorporate this comment. Since tick counts for mortality testing are to be conducted by 48 hours, which is before "blood-feeding to the point of engorgement" would happen (or possibly any blood-feeding for that matter), the endpoint for the tick mortality testing has been changed to "live ticks on the treated animal vs. live ticks on the control animal." 
264
Efficacy Endpoints
The proposed guidelines state: "For testing repellency against ticks, a repellent effect is a reaction by a tick to avoid contact with a treated animal. Since repelled ticks will move away from the treated animal and possibly be unrecovered, repellency should be determined indirectly by counting the number of live or dead ticks (attached or unattached) on the animal plus live or dead ticks in the cage that are engorged compared to the control animals. Moribund ticks should be considered alive. Other methods of evaluating repellency against ticks (e.g., hot foot, inhibition of attachment, disruption of attachment; Halos et al. 2012) may also be considered but should be justified."

The Panel noted that the time frame is important here. If following EMEA guidelines (European Medicines Agency, 2016), the ticks should not attach to the host within a 24- hour time frame. Therefore, host and cage surroundings should be checked for ticks and numbers on and off the host need to be compared to control animals. Off-host live, or dead ticks could both be counted. Antifeedant could be considered as an additional endpoint. The number of live and unattached ticks would be a measurement of this.  PG 55
The methods for tick repellency have been revised to have the ticks placed near the animal instead of directly on the animals and will compare the number of ticks (dead or alive) on the animal 24 hour later to determine repellency.  
265
Data collection and recording for ticks
For mortality trials, the mode of action of the product would dictate when the data is collected. Systemic products must be ingested by ticks for mortality to occur and could take several days to act based on the prolonged feeding period of ixodid ticks. The Panel noted that a time point of 48- 72 hours should be reasonable to capture tick mortality with systemic acaricides. For some acaricides that act much sooner, earlier measurements should be made. Comb counts should be specified to include tick removal, unless mortality measurements are being made for retained ticks over longer time points in studies.  PG 55
For mortality trials, the 48 to72 hour counts will remain, with the option to count sooner. The guideline has also been revised to clarify tick removal as suggested. 
266
Data collection and recording for ticks
For the tick mortality effect categories, as the Panel has noted previously, all instances of "engorged" should be removed and replaced with "partially fed." The Panel observed that it seems like there are too many categories here to be truly useful. Reducing the number of categories to evaluate would simplify study interpretations and perhaps reduce the sample sizes of ticks and animals needing to be used. The simpler categories from the European Medicines Agency (EMA, 2016) would probably suffice for most study goals. Also, it seems questionable for moribund ticks to be counted as alive; it is probably more accurate to count them as dead because it is unlikely that they will adequately recover to the point of being capable of re-attaching to a host. Holding moribund ticks off the host for at least 24 or up to 48 hours, under optimal temperature and humidity conditions for survival would allow the study to differentiate live from dead ticks unambiguously. For the tick repellency categories, the Panel agreed the data should typically be collected within 24 hours post-infestation, though sometimes ticks may take longer to attach to the host; up to 48 hours may be appropriate to evaluate repellency in some host-tick systems. Also, there are too many categories for tick scoring here. Within 24 hours, there is no blood acquisition by ticks, and so these categories are not accurate. Blood feeding as a factor for repellency should be removed when using short time frames.  PG 55-56
The number of categories in both mortality and repellency tick testing has been reduced. Since the final counts for the mortality endpoint are already at 48 hours, "moribund" ticks will still be counted as "live." 

Since the repellency endpoint will now be whether the tick is on or off a host animal at 24 hours, the blood feeding category will be removed. 
267
Data collection and recording for ticks
In mortality testing studies, how can host immune response to ticks be differentiated from the effects of the product tested? The Panel noted that it may be important to provide guidance for individual variations that occur in host rejection of ticks due to an immune response vs. product induced mortality determinations during trials as an endpoint (European Medicines Agency, 2016; Wada et al. 2010). This may not be an issue for R. sanguineus, but it could be the case for other tick species tested, such as I. scapularis (Szabo and Bechara, 1999; Gebbia et al., 1995).  It is possible to have large numbers of ticks attach to a host, over the 25% threshold, and rejection by the host may occur during feeding trials that is due to host immune response, not due to the product's effects. Monitoring of host inflammation around feeding ticks would provide insight regarding product-induced mortality vs. immune-induced mortality. Animals with noted inflammation around bite sites should be removed from the study to prevent confounding of results. This could be of particular importance as animals are infested multiple times over the course of several months and may develop acquired immunity to tick feeding during later trials.  PG 56
While this comment is appreciated, it is outside the scope of the guideline to differentiate host immune responses. 
268
Data collection and recording for ticks
b. Optimally, mortality and repellency should be determined in different studies with different designs, because they are different outcomes and have the strong potential to bias the outcome. The current methods muddle the distinction between mortality and repellency. For example, if 50 ticks are put on an animal, and at 24 hours investigators find 10 dead ticks on the animal, 0 live ticks on the animal, and 40 moribund ticks in the cage, this could be scored as a repellent, when equally likely this was a slow acting acaricide without repellent activity. While some materials are both toxic and can act as a repellent, other materials may do one or the other. Thus, the Panel suggested that the design for repellency assays needs reconsideration, as discussed in more detail in the response to Charge Question #10.  PG 56
These guideline sections have been revised. Based on the revised methods, assessing repellency would likely not be able to be combined with mortality. 
269
Data collection and recording for ticks
b. The Panel also recommended that for a repellency study, ticks should not be placed directly onto an animal. Instead, ticks should be placed in the local environment with the animal to allow ticks to naturally attach to the host; if there is a repellent effect, fewer ticks will be found on the animal (Dumont et al., 2015b).  PG 56
The guideline has been revised to incorporate this comment.

Public Recommendations (docket ID # in parentheses):
270
"Adult mortality advises live fleas vs. dead ticks to determine mortality. Recommend assessing by counting the live fleas only (not moribund) and comparing treated vs. untreated groups. Additionally, this should be for all "kills" claims not just "starts killing". (0008) 
See response to comment #14. 
271
"Calculation of Repellency differs EMA guidelines and the proposed formula includes dead ticks in the control & treatment retention rate Efficacy = (Attached ticks on control - Attached ticks on treated)/(attached ticks on control) x 100" (0008) 
The guidelines have been revised to be consistent with the EMA guidelines. 
272
"Tick efficacy should be based on live ticks on dogs, similar to flea efficacy. With that approach, there is no need for collecting ticks that fall off the animals as this is inherently variable with ticks escaping and/or being lost in feces, urine, vomitus, etc... Can the EPA justify why tick efficacy would be based on dead ticks? Tick efficacy is routinely calculated based on a comparison of live ticks on the treated dogs compared to the control dogs (similar to flea efficacy as described in this draft guidance). Quantifying dead ticks is highly variable and inaccurate. Measuring live ticks is much more accurate and appropriate for evaluating tick efficacy." (0026)
See response to comment #14.
273
"Yes, repelling ticks is a viable endpoint. Companies need to be given a chance to develop products and formulations that give the end user a variety of options in order to control tick populations. b. This would have to be a defined endpoint in the protocol before trial initiation. Methods to assess repellency can be very similar to mortality assessments. c. Yes, it can be. The initial study question must be defined in the protocol before the initiation of the study. As the methods and endpoints are similar ratings for repellency versus mortality then these need to be tested by two separate trials i.e., one trial to determine mortality effects and one trial to determine repellency effects." (0008)
See response to comment #268.
274
"The second edition of the WAAVP guidelines (published in 2013) updated the tick categories to only four categories: live free, live attached, dead free, dead attached. Engorgement status was removed from consideration as most ticks will not be engorged during the time that they are on dogs in laboratory studies and as it can be difficult to assess when ticks are not fully engorged. Suggest that the EPA guidelines should be aligned with this to ensure all data collected is meaningful and adds value (in regard to engorgement status) and to allow studies to be used in multiple geographies." (0026)
We agree with this comment and these sections have been revised to align with the EMA guidelines.
275
"Based on the current endpoints. It seems that the mortality endpoint is dead ticks compared to the control animals, where repellency is looking at live and dead ticks, so they can be assessed in the same study. If counts are done at 24-72 hours post infest or treatment, there shouldn't be engorged ticks in kennels." (0026)
See response to comment #268.




METHODS FOR TESTING AGAINST MOSQUITOS OR BITING FLIES (SECTION (L))

Charge Question 16
Comment on the adequacy of outcrossing lab-reared colonies of mosquitoes every 3 years.
a. Is three years an adequate timeframe to retain genetic diversity in lab-reared populations? Why or why not?
b. If not, how often should lab-populations be outcrossed with wild-type mosquitoes? Please provide references if relevant.

Panel Recommendations:
276
What is the goal of periodic outcrossing of mosquito (as well as other ectoparasite) colonies? The Panel observed that retaining genetic diversity is irrelevant for toxicology testing (unless for example, examining frequency of resistance alleles) and questionable for behavioral studies. Most insect species including mosquitoes do not die off with inbreeding. Mosquito colonies at the USDA laboratory in Gainesville, FL have never been outcrossed and have been used successfully for a variety of tests for years. If the goal is to maintain some part of a large population with relevant proportions of resistance alleles, then the Panel found that this protocol is not detailed enough.  A recent paper by Ross et al. (2019) concludes that inbreeding depression occurs in this species due to low population size. Colonized mosquito populations suffer inbreeding depression and adaption to laboratory conditions. Laboratory environments are inherently artificial, and colonized mosquito populations experience an entirely different set of selective pressures compared to natural populations, which could lead to declines in reproduction (Bryant and Reed.,1999). Ross et al. (2017) suggested that while life history traits of Ae. aegypti do not change consistently with laboratory maintenance, traits where selective pressures are absent in the laboratory, such as flight ability, feeding behavior, and thermal tolerance might still be compromised.

Ross et al. (2019) concluded that variation in fitness and performance was dependent on colony size, but that most individual traits were unaffected, and patterns of adaptation were not consistent across populations. Performance of laboratory populations of Ae. aegypti maintained in a larger population increases. Rates of laboratory adaptation can be slowed or minimized by using more natural rearing environments such as maintaining large populations (400 vs. 100 in larger cages at a variable temperature and environments that are more complex.

The Panel agreed that many traits are lost in colonized insects shortly after they are selected for colony conditions. However, genetic conditions tend to stabilize. Adding new (uncharacterized) genetic material periodically would change the genetics of the colony so they are different than they were when the colony was used for previous studies. This could produce different results in certain types of tests.  PG 57
The outcrossing recommendation has been removed. 
277
Rather than periodic outcrossing, the Panel suggested testing a standard strain that is well-studied and has a history in the scientific literature and comparing the counts of live mosquitoes on treated versus untreated animals. Believing that there will be widespread differences between strains assumes facts not in evidence. Furthermore, every time a strain is outcrossed the strain changes and could result in non-reproducible results, a major drawback to any manipulations of the strain used for testing. In addition, outcrossing with uncharacterized field-collected individuals could introduce pathogens, as well as, resistance to ectoparasiticidal compounds.  PG 58
The outcrossing recommendation has been removed.
278
If the Agency decides to include protocols regarding outbreeding in the updated guidelines, the Panel cautioned that it would be more appropriate to suggest outbreeding after every so many generations than to specify a time period (i.e., every 3 years). Several articles indicated that the simplest way to maintain the fitness of colonized mosquito populations, especially Ae. aegypti, is to cross laboratory colonies to an outbred population. Yeap et al. (2011) suggest a 2-year period for out-crossing, but a rationale was not readily discernable from the article.

With an estimated generation time of approximately 30 days for Ae. aegypti, (Sowilem et al.,
2013), a 3-year outcrossing interval could result in >36 inbreed generations.  However, Ross et al. (2019) established replicate populations of Ae. aegypti mosquitoes and maintained them in the laboratory for twelve generations at different census sizes [<100 individuals or 400 individuals]. At F5, life history traits were compared between replicated large populations and inbred lines. After two generations of brother-sister mating, the inbred lines had reduced fitness relative to the large populations, with substantial costs to larval survival and development time. Although the study design involved a comparison among replicates established from eggs from an ancestral population, the data suggest adaptive changes to laboratory conditions in the F5 generation. Therefore, if outbreeding was important, it should be done more frequently than on a 3-year basis.

Alternatively, Ross et al. (2017) describe a simple protocol for maintaining Ae. aegypti mosquitoes in the laboratory, which minimizes laboratory adaptation and implement outcrossing to increase the relevance of experiments to field mosquitoes. They suggest outcrossing be repeated for at least three consecutive generations to produce colonies that are at least 87.5% similar, genetically, to the field population.  PG 58
The outcrossing recommendation has been removed.

Public Recommendations (docket ID # in parentheses):
279
"Outcrossing laboratory colonies with field populations every 3 years is not appropriate. The idea behind having an established laboratory colony is to have a "reference" colony to which things can be compared back to over long periods of time. The outcrossing of field populations has the potential to introduce pathogens which would deem the colony unacceptable for use. Additionally, it poses the risk of introducing an insect pathogen that could destroy the entire colony. If claims to control resistant populations are sought, separate tests should be conducted on arthropods with known resistance to the active ingredient (or mode of action) to support such claims. If the intent is to introduce "alternate" genes into the colony, these alternates will be diluted in a short period of time due to the generational efficiency of most mosquito species." (0008)
The outcrossing recommendation has been removed.


Charge Question 17
In the interest of reducing vertebrate animal testing, we propose simultaneous testing of up to three mosquito species on a single vertebrate animal.
a. Is simultaneous testing of three mosquito species feasible? Why or why not?
b. If simultaneous testing of three mosquito species on a single vertebrate animal is feasible, then please comment if the proposed methods are adequate, and if there are alternative methods please provide detailed methods and references.
c. If testing three species at once is not feasible, provide additional detailed methods and associated references to limit the number of vertebrate test animals necessary to complete efficacy testing against the three required mosquito species and maintain adequate statistical power.

Panel Recommendations:
280
After consulting mosquito experts (e.g., Drs. Dan Kline and Bill Reisen, personal communication), the Panel determined that simultaneous testing with three mosquito species should not affect the test results. The mosquitoes do not compete for feeding sites and do not otherwise interact in ways to alter results expected with a single species. Proposed methods look reasonable, but it would be nice to reduce the number of recommended vertebrate animals used.  PG 59
No changes to the guideline are necessary.
281
After an extensive discussion of this Charge Question, the Panel recommended the following: If three mosquito species are tested simultaneously, 50 mosquitoes of each species should be used. .......  PG 60
No changes to the guideline are necessary.
282
After an extensive discussion of this Charge Question, the Panel recommended the following: ......... For flies known to exhibit aggressive behavior and (or) inflict a painful bite (e.g., stable flies), efficacy testing conducted with 50-100 flies may not be recommended because of the adverse effects on the vertebrate animals involved. Under natural conditions, stable flies rarely complete a blood meal on the first try because they are dislodged by the host because the bite is painful. The fly merely flies around the host and tries again. This continues until the fly is replete. When the fly is undisturbed, as it would be under laboratory testing conditions, the time to repletion averages 147 seconds, as documented Schofield and Torr (2002).  PG 60
The models were re-run for stable flies using 25 flies and determined if 25 flies per animal is desired then 12 dogs per study should be used. The infestation time has also been revised to only for 30 minutes.



Public Recommendations (docket ID # in parentheses):
283
"For a general mosquito claim, product must be effective against 3 species of mosquitoes, some products may repel, and others efficacy is measured by landing, protocol discusses testing multiple species with single test. We do not believe this would be possible." (0008)
The mosquito section allows the testing of multiple species at the same time; however, the study does not have to. No changes to the guideline are necessary. 


Charge Question 18
The proposed methods to test mosquitoes and biting flies indicate vertebrate test animals should be sedated prior to exposure to invertebrate pests.
a. In the interest of minimizing stress of sedation on vertebrate test animals please discuss whether sedation is necessary. If sedation is not necessary, please provide detailed methods and references and discuss the strengths and weaknesses of the methods for conducting efficacy testing.
b. If sedation is necessary, please provide methods to minimize the number of times vertebrate animals should be sedated.

Panel Recommendations:
284
Because the efficacy of a product is being evaluated, a sedated animal allows this evaluation to be conducted without the effects of host defensive behaviors. An un-sedated animal confined in a 2 ft3 cage with 50-100 flies or mosquitoes feeding on it would be in constant motion. Pain associated with the bites of mosquitoes and stable flies would interfere with an accurate evaluation of product efficacy and the pain would be difficult for a test vertebrate (e.g., a dog) to endure for 1 hour. The Panel suggested reducing the number of fly exposure trials requiring animal sedation to every 2 weeks (Days 7, 28, etc.), when possible.  PG 60
It is important to balance realistic testing with humane treatment to the host animals to produce reliable results. The proposed guideline already reduced infestations for mosquitoes and flies compared with fleas. Currently, for a typical 1 month product, there are three infestations: day 3, day 14 and day 28. This will provide enough endpoints to gain reliable data with the reduction of 1 infestation point. No changes to the guideline are necessary. 
285
If starved insects are used, this should cause more rapid feeding on the test vertebrate, and sedation time and exposure to the insects might thus be reduced. Young (3 to 5 days old) stable flies are routinely starved for 4 hours before a test where feeding is involved. Stable flies will remain alive for 24 hours or more if given water on cotton ball or pads (Jones et al., 1992). Stable flies will live several days if provided with a 10% sucrose solution or Gatorade (powder is easy to keep and store). Mosquitoes are not usually starved but maintained on water or 10% sucrose.  PG 60-61
See response to comment #18.
286
If one of the criteria for a successful test is the presence of blood in the insects' gut, the Panel observed that small amounts of blood would be sufficient. Insects do not need to feed to repletion. Stable flies can fill up in an average of 147 seconds (Schofield and Torr, 2002). Some mosquitoes can fill up in about <= 3 minutes, if undisturbed.  PG 61
No changes were proposed in this comment. 

Public Recommendations (docket ID # in parentheses):
287
"Selection and allocation of test animals. Methods other than sedation of animals should be considered. Sedation can be hard on the animals, especially if there are multiple data points over long periods such as a collar evaluation." (0008)
Sedation of animals is only suggested when testing against mosquitoes as they are difficult to count on moving animals and against painful/aggressive biting flies. These pest claims are rarely proposed. Sedation should not be used for testing against other pests. No changes to the guideline are necessary. 
288
"The effects of sedatives can differ -- some pain medications have sedative properties, but not all sedatives have pain management properties. In addition, not all sedatives are anxiolytics, meaning that some sedatives prevent movement but mental anxiety is still present. If the guidelines recommend to sedate animals, they should also provide that the animals are administered pain medications or a drug that has proven analgesic and sedative properties." (0019)
The following sentence has been added:

"For infestations using painful or aggressively biting mosquitoes or biting flies, the animals should be administered pain medications or a drug that has proven analgesic and sedative properties."
289
"Sedation is considered necessary in this model to prevent animals from interfering with the feeding of the mosquitoes or flies." (0026)
 This feedback has been incorporated into the guideline. See response to comment #288. 


Charge Question 19
Comment on the timing of exposing vertebrate test animals to mosquitoes and biting flies.
a. Are the number of exposures for products with different durations of efficacy adequate for determining efficacy? Why or why not?
b. If the number of exposures to pests can be decreased, please indicate specifically which exposures can be skipped. If efficacy should be evaluated at more time points, indicate when exposures should occur and for what product types (e.g., spot-ons, collars, and residual shampoos), and discuss the value provided by the additional time points in relation to the additional stress on test animals from additional sedations.

Panel Recommendations:
290
a. The proposed guidelines variably specify that each test vertebrate animal should be exposed to 50-150 adult, pathogen-free mosquitoes or biting flies for one or more time points. As has been previously mentioned, the Panel questioned whether adequate numbers of all insect species listed in the proposed guidelines are readily available for this.  PG 61
Language has been added to indicate that flies may be obtained from wild populations if they are not readily available in lab colonies. In addition, the specification of "pathogen-free" has been revised to "free of vector-borne pathogens" to be more specific.
291
a. Furthermore, the Panel strongly recommended that the requirement for exposing vertebrate test animals to 50 stable flies be rewritten due to the extreme pain and distress caused by exposure to this aggressive biting species. Even decreasing the number of stable flies from 50 to 25 may still be excessive for an individual animal to tolerate for the proposed one-hour exposure period, let alone multiple exposures. The reduction in power and corresponding need for an increased number of vertebrate test animals caused by decreasing the pest number to 25 stable flies, and potentially even lower, should be further considered through modeling. The Panel was interested in seeing just how much decreasing the number of stable flies would increase the number of necessary test animals. Similarly, the number of control animals should be related to the expected degree of response. Since the control mortality would be expected to be very low in the case of stable flies, the Panel suggested that perhaps fewer control vertebrate animals could be used.  PG 61
The stable fly model was re-run using 25 stable flies vs. 50 and the guideline has been revised to add that if 25 flies are tested vs. 50 then 12 dogs per groups should be used. The guideline has also been revised to state that the exposure time for flies should be 30 minutes rather than 1 hour. The guideline already had reduced amounts of exposures (only days 3, 14 & 28.) Finally, language with sedative and pain medication suggestions to help alleviate pain and stress to the animals has been added. 

292
a. As previously raised by the Panel in discussing other charge questions, better definitions of "adequate infestation" of test animals, preferably specific to both pest and test animal species, would be another strategy for reducing overall sample sizes. The Panel also discussed specifying appropriate expected levels of blood feeding on control animals. The figure of 75% currently appears in the "Sample Size for Pet Product Studies" document (page 7) and the Panel found that it was reasonable to expect that this would be achieved quite quickly. This would be especially true when using flies in their most active 3-5 days of age period and after adequately starving them for 3-4 hours prior to exposing them to the test animal. In both cases, better-refined definitions and study design recommendations may also increase the power of the resulting data and allow further reduction of vertebrate animal numbers.  PG 61-62
The guideline currently includes some parameters to assist in increasing the power of the data, such as requiring a minimum amount of bloodfeeding or retention of arthropod pests on control animals, which were incorporated into the power versus sample size modeling. 
293
b. If the registrant can reasonably expect that their product will last a full month (as a repellent and/or with insecticidal activity), the Panel found that it may be permissible to allow testing to occur at Day 7 to show the topically applied product has spread over the dog and then eliminate Day 14 exposures. Day 21 and Day 28 exposures should still be included. The Panel noted that this eliminates one time point (Day 14) and still proves month-long residual efficacy.  PG 62
For products lasting 1 month or less, all the time points listed in the guideline will remain because collection at multiple time points helps with the interpretation of a dataset containing the occasional efficacy failure. For a mosquito or biting fly product that is expected to last 30 days, testing had already been omitted on day 7. However, it should be noted that reducing the number of infestations does not reduce the number of animals used nor the overall duration in each study. 
294
b. The Panel recommended that the guidelines address the duration of each experimental exposure. Many of the Panel's concerns about animal welfare related to both sedation and exposure to biting flies could be addressed by limiting the exposure time. A 30-minute to one-hour exposure time for stable flies was suggested.   .............  PG 62

b. The Panel recommended that the guidelines address the duration of each experimental exposure. Many of the Panel's concerns about animal welfare related to both sedation and exposure to biting flies could be addressed by limiting the exposure time. A 30-minute to one-hour exposure time .......... The same exposure times should also be applied to the enrollment phase of these studies.  PG 62
See response to comment #291.

Public Recommendations (docket ID # in parentheses): n/a


Charge Question 20
Comment on the methods for assessing mosquito and biting fly blood-feeding.
a. Are the proposed methods adequate to determine blood feeding for insects where blood feeding status cannot be visually observed through the abdomen?
b. Are there other methods for determining if insects blood-feed? If so, please provide additional methods and references.

Panel Recommendations:
295
a. The Charge Question seems to be asking if sampling half the live flies (where blood feeding cannot be determined by observation) with a blood smash test is acceptable.
This question is subjective and depends on what level of risk is acceptable, particularly where transmission of pathogens is of concern. The remainder of the flies which are to be checked at 48 hours post treatment could be evaluated for blood using one of the more sensitive blood detecting techniques described below, at which time all the flies would be evaluated, thus eliminating uncertainty.  PG 62
The guideline has been revised to determine blood feeding for all mosquitoes (for mortality and repellency claims). 

In regards to using more sensitive techniques, the following language has been added:  "It is preferred that blood-feeding is assessed and confirmed for each individual insect using an enzyme-linked immunosorbent assay (ELISA) (e.g., Beier et al. 1988) or polymerase chain reaction (PCR) based method (e.g., Fitzpatrick et al 2019); however, other methods such as squishing may also be appropriate as long as the method chosen is used consistently throughout the study and applied the same to treated and control groups."
296
Of concern to the Panel was that EPA's definition of "repellency" (has not imbibed blood) differs from the traditional entomological definition of repellency which is directed movement away from repellent. The proposed guidelines (i.e., on page 18) seem to confuse repellency and feeding inhibition. The guideline states:

"i. Repellent effect. Insects should be evaluated to determine if they took a blood meal. Insects that take a blood meal are not repelled. More conservative endpoints such as insect landing may also be considered but should be justified."

"i. Repellency. To determine repellency only, immediately after removing animals from the cage all insects should be aspirated from the cage, knocked down (e.g., using CO2, or frozen), identified to species, crushed on a light background to assess for blood-feeding."

The Panel noted that this does not conform to definitions of repellency put forward by several groups. Repellents cause oriented movements away from the source. In relation to feeding, a better term might be a feeding inhibitor (Deletre et al., 2016). For example, the European guidelines define repellent effect as "a product with a repellent effect will cause the parasite to avoid contact with a treated animal completely and/or to leave a host" (EMEA/CVMP/EWP/005/2000-Rev. 3).

Marchiondo et al. (2013) writes, "Two types of repellency are defined: "sensu stricto" for repellency characterized by an irritant effect, causing the tick to move away from the treated animal or leading it to fall off soon after contact with the treated hair coat within ∼6 - 8 hours and "sensu lato" for all other tick repellency (or expellency) up to ∼24 hours. The first, repellency sensu stricto, may be attributed to the vapor phase of a compound or irritant effect through direct contact (e.g., oil of Citronella but also some synthetic pyrethroids), while the second repellency (or expellency) causes inhibition of attachment or detachment of already attached ticks, (e.g., some synthetic pyrethroids or amitraz)."

Collection of unfed flies means either: 1) the flies simply didn't feed during the test period, 2) the test material was a true repellent, or 3) the test material was a feeding inhibitor.  PG 62-63
We agree with the panel's definition of repellency and have clarified the definition vs. the appropriate end point to test the product. The guideline recommends that blood feeding for mosquitoes and biting flies should be the endpoint since it is easier to demonstrate. However, landings as the endpoint is also an option, if justified. As for fleas, the referenced literature was considered, and a statement was added to the guideline to encourage the submission of a protocol for flea repellency tests. 

 
297
b. The Panel noted that techniques used to determine blood feeding in cat fleas include Drabkin's technique to determine the quantity of blood consumed by adult cat fleas (Rust et al., 2002) and should be applicable to flies. Several other procedures of determining blood are outlined in Marchiondo et al. (2013). A real-time PCR analysis of blood feeding in fleas was 10,000-fold more sensitive than the Drabkin technique (Wang et al., 2012). Wang et al. (2012) concludes, "The HMBS PCR method developed here offers the advantages of both exquisite sensitivity and specificity that make it superior to other approaches for quantification of blood ingested by fleas. The capability to detect minute quantities of blood in single fleas, particularly immediately after colonization of the host, will provide a superior tool for studying flea-host interactions, flea-borne disease transmission, and flea control strategies."  PG 62-63
See response to comment #295. 

Public Recommendations (docket ID # in parentheses):
298
"In section l(5)(ii)(a) I dont see how the abdomen of live mosquitos can be visually assessed and then the mosquitos held for the additional 48 hours post exposure. When they are flying and moving, they cannot be properly assessed in this manner. The only proper way to do this is killing then by freezing and then squashing them to determine blood meal." (0012)
The guideline has been revised to assess all recovered mosquitoes for blood feeding. 


Charge Question 21
To reduce the amount of vertebrate animal testing, we propose to allow assessment of repellency and mortality within the same study by evaluating half of the insects for repellency before determining mortality.
a. Discuss the viability of evaluating both endpoints in the same study by using a subset of the total mosquitoes to determine repellency and provide any additional methods that would determine both endpoints within the same study.
b. The proposed endpoints for mortality are secondary to repellency (i.e., for mortality to be assessed repellency should be observed first) because a mosquito or fly that is not repelled may blood-feed and could potentially transmit pathogens that cause disease. Discuss the potential value of assessing mosquito mortality that are not also repelled. Should mortality of mosquitoes that blood-feed be considered a viable endpoint to show efficacy of a repellent-product? Why or why not? If yes, provide specific rationale as to why.

Panel Recommendations:
299
a. The Panel observed that if the timing of endpoint determinations is different, it seems reasonable to study both repellency and insecticidal activity against mosquitoes and biting flies during the same test. Repellency is immediate; knockdown/mortality occurs up to 48 h after exposure. One is, however, looking to see that none of the insects had a blood meal if the products are to be effective in either category.

As stated in the proposed guidelines (OCSPP 810.3300) Section (l) (4), efficacy endpoints, the "endpoints (e.g., blood feeding, mortality) should be selected with respect to the intended use pattern for the product and mode of action of the a.i."

An active ingredient (a.i.) could have only: (a) a repellent effect; (b) a repellent and a toxic effect (e.g., permethrin); or only a (c) cidal effect (e.g., fipronil).  Thus, the a.i.'s mode of action and its intended use pattern should drive the testing methodology and the efficacy endpoints.

The goal of insecticides/repellents applied topically to dogs/cats is to protect the animal from the biting adult stages of mosquitoes and biting flies and to interrupt blood feeding and thus significantly reduce the potential for disease transmission. Assessment of blood feeding is the critical thing to assess in live, knocked down, moribund and or dead insects after being exposed to a topically treated dog/cat.  Thus, if based on blood feeding >=90% of the insect species being assessed are alive and have no blood in them, then this a.i. has a mode of action that should be considered a repellent. If the a.i. has both a repellent and a mortality effect, then >=90% of the recovered insect species whether they are alive, dead, or moribund should have no blood in them since the a.i. either repelled them and/or killed them before a blood meal could be taken.

Therefore, the Panel concluded that looking at subsets (e.g.-removing half of the insects) of the total number of mosquitoes and biting flies and looking for repellency before assessing mortality is not needed. Blood feeding should be assessed in all recovered insects and this should drive the label claim (use pattern). Depending on the extent of blood feeding the product can claim to be a repellent or an insecticide/repellent or a standalone insecticide that works so fast that insects that land on a treated dog/cat are knocked down or killed so quickly that they can't obtain a blood meal before dying or becoming moribund and subsequently dying.  PG 64
Both repellency and mortality can be assessed in the same study and the guideline has been revised to have all mosquitoes or biting flies assessed for blood feeding. 
300
a. The Panel suggested that some of the automated methods be used to detect the presence of blood in an insect to confirm if blood feeding has occurred. For example, confirmation of a consumed blood meal could be performed using an enzyme-linked immunosorbent assay (ELISA) following the procedure described by Beier et al. (1988).  There are likely other references that could be cited for these methods as well.  PG 64
EPA agrees with the suggestion and has included language indicating that these automated methods including the ELISA method provided are acceptable options to detect blood-feeding.
301
b. The Panel noted that blood feeding should not occur if the test product is effective either as a repellant or as a knockdown/mortality agent. As noted in the Charge Question, repellency is primary and needs to be observed before mortality is determined. However, knowing if insects die quickly could be important. If the biting insect is not repelled and blood feeding occurs, there is potential for disease transmission unless killed quickly. Therefore, even if mortality occurs after insects are on the test animals, mortality/knockdown could also be considered as a viable endpoint.  PG 64-65
EPA agrees that absent blood-feeding, mortality is a viable endpoint and as such, will allow for assessment of mortality if blood-feeding does not occur.  
302
The Panel did not agree with the assumption that the proposed endpoints for mortality are secondary to repellency. As stated above, the mode of action of the a.i. is going to define the endpoint (and label use pattern) and the ultimate goal of the a.i. is to significantly reduce or totally inhibit any blood feeding by mosquitoes/biting flies exposed to a treated dog/cat. In this Charge Question, the Agency states that "a mosquito or fly that is not repelled may blood-feed and could potentially transmit pathogens that cause disease."  This may be true but if the a.i. has a predominant killing or mortality effect, the insect can be killed or incapacitated so quickly after landing on the host that blood feeding doesn't occur.  PG 65
The guideline allows for assessment of mortality if blood-feeding did not occur.  Although mortality may not be secondary in an instance where a killing effect occurs so fast that the insect does not blood feed, the evaluation of this under the guideline study would be the same as for a repellent product that also has insecticidal claims. No changes to the guideline are necessary.
303
To respond to the first bullet point in Charge Question (b),the Panel would again refer the EPA reviewers to the previous discussion points that blood feeding should be assessed in all recovered insects and this should drive the label claim (use pattern) as a repellant, and/or an insecticide/repellent or a standalone insecticide that works so fast that insects that land on a treated dog/cat are knocked down or killed so quickly that they can't obtain a blood meal before dying or becoming moribund and subsequently dying.  PG 65
The Agency agrees and has revised the mosquito/biting fly section to provide a method to evaluate all insects for bloodfeeding and mortality in a single study.
304
To respond to the second bullet part of Charge Question (b), the Panel concluded this question would only apply to an a.i. with a mode of action that has both repellency and a mortality (toxic) effects. Under this scenario, if, based on blood feeding, >= 90% of the insect species being assessed are alive and/or dead (demonstrated mortality) and have no blood in them, then this a.i. has a dual mode of action (repellent combined with mortality effects, such as permethrin). Thus, if <= 10% of live/dead insects do have blood detected in them, then the 90% of insects that don't have blood detected will be making a significant impact on reducing disease transmission and reducing irritation due to the biting process by the insects and potential immune reactions at the bite site. No product in a real-world biological system will consistently provide 100% efficacy, thus, the Panel observed that having a minimum threshold of >= 90% seems appropriate.  PG 65
EPA agrees with the 90% threshold for this type of study.
305
Additional Comments related to Assessing Efficacy in Mosquitoes and Biting Flies
One committee member suggested that the EPA should consider adding a new section to this part of the revised guidelines that assesses knockdown/mortality on topically treated dogs/cats, regardless of their blood feeding status.  This would be analogous to efficacy assessments performed on treated dogs/cats infested with fleas and ticks where they ingest blood prior to being killed by the topically applied a.i. The primary concept is to kill these insects in their local environment which negatively affects reproduction and helps to interrupt the life cycle of the pest and reduces the overall population of the insects in the pet's local environment and also potentially reduces human exposure to bites. It has been documented in the scientific literature that some mosquitoes do not move significant distances from where they breed, thus killing local populations of these insects can be just as important as preventing blood feeding. Since only small numbers of biting insects are actually carrying infectious agents, the impact of significantly reducing blood feeding and reducing transmission of infectious agents is minimal.  PG 65
These products are not intended for controlling mosquito populations in a local area.  Additionally, EPA is not aware of, nor has evidence been provided, that these products have measurable effects on local mosquito populations or offer effective protection against mosquitoes to other animals which did not receive a dose within the local area. For these flying pests, efficacy is based on protection of the animal to which the products are applied. 

Public Recommendations (docket ID # in parentheses):
306
"Yes, repelling mosquitoes is a viable endpoint. Companies need to be given a chance to develop products and formulations that give the end user a plethora of options in order to control mosquito populations. b. This endpoint would have to be defined endpoint in the protocol before trial initiation. Methods to assess repellency can be very similar to mortality assessments. c. Yes, it can be. The initial study question must be defined in the protocol before the initiation of the study. As the methods and endpoints are similar ratings for repellency versus mortality then these need to be tested by two separate trials i.e., one trial to determine mortality effects and one trial to determine repellency effects." (0008)
EPA agrees that repellency is a viable endpoint. EPA has taken the advice of the SAP in that both repellency and mortality can be assessed in the same study by assessing blood feeding for all the mosquitoes or flies. 


Charge Question 22
Efficacy testing is based on the specific label claims proposed for product marketing. Common claims are "kills," "repels," and "controls" (i.e., residual efficacy). For most pesticide products (e.g., direct sprays and residual applications) the typical endpoint to support claims of "kills" or "controls" is based on dead arthropods. To evaluate dead arthropods for shampoo products we propose pests should be shampooed in a tub with a screen over the drain to keep fleas and/or ticks in the tub. This may differentiate the pesticidal effect from the effect of pest removal, which could also occur by using a non-insecticidal shampoo. Discuss the value of using counts of dead arthropods vs. using counts of live arthropods on treated compared to untreated vertebrate animals to determine different endpoints (e.g., kills).
a. Is assessing live arthropods on treated vs. untreated animals an adequate and practical endpoint for determining mortality? Why or why not?
b. Are there alternative methods for determining mortality for shampoo products? If so please provide the methods and supporting references.

Panel Recommendations:
307
a. The Panel noted that the efficacy or effectiveness of treatment is appropriately determined by comparing the proportion of pests killed by the treatment (i.e., proportion of living in the control group - proportion living in the treatment group) with the proportion of pests living in the control group (Abbott, 1925). Determining the proportion of live pests in the control group, and the proportion of live pests in the treatment group, requires counts of living pests on each vertebrate animal and knowledge of the total number of pests applied to each vertebrate animal. Ideally, all the arthropods applied to the vertebrate animals would be counted and classified as live or dead to assess the proportion of live arthropods in the control or treated group. Counts of the dead arthropods would support the claim of killing for the tested product. However, it is unlikely that the total number of arthropod pests applied to each vertebrate animal will be recovered during the counting process. Fleas originally applied to a cat or dog may be ingested or hide within or upon a vertebrate animal. Accordingly, it has become standard practice to use Abbott's formula with live arthropod counts.

In the proposed guidelines, approximately the same total number of arthropods is applied to each animal. It is reasonable to assume that the total number of pests on the vertebrate animal does not change appreciably between the time of application and the time of counting. If all the arthropods originally applied to the vertebrate animal are not9 recovered, the arthropod counts could be subject to measurement error in counting, which more likely would be an issue of undercounting. Bias often becomes apparent when more than 5%-10% of the data are missing. If the extent of undercounting varies considerably between the control and treated groups, or if the application of the treatment leads to very small counts in the treated group (e.g., 0 to 5 pests remain on the animal after application of treatment), the efficacy assessment could be biased. The Panel recommended that measurement error in counting should be investigated by the Agency or taken into account in the modeling framework as discussed in more detail in the response to Charge Question #9.  PG 66
The Agency did not include measurement of error in counting in the models because we do not have an estimate for measurement of error in counting. However, counting live pests is considered the most accurate way to indirectly account for pests. 

Live pest counts on the animals are used for general mortality claims for topical/collar products as well as for shampoo products. 

Utilizing live counts should have no different bias or effect on Abbott's formula being used in the shampoo tests than what is in the basic flea and tick sections.
308
a. Given the discussion above, assessing live arthropods in the bathing tub is an adequate and practical endpoint for determining mortality. However, the Panel noted that the practicality of this recommendation includes the considerations discussed below.

Tick placement for pesticidal shampoo products should follow the same suggestions previously made for mortality and repellency studies. Ticks should be placed on Day-2 (not Day-1) to permit adequate attachment time before the treatment is applied.

Any live pests found in the plugged tub should be added to the total live arthropod counts. The vertebrate animal should be allowed to fully dry for 24 hours before counting the remaining fleas or ticks. Fleas are, however, more likely to be washed off than ticks, while ticks are more likely than fleas to remain attached to a vertebrate animal and not removed by the act of shampooing. Therefore, the Panel recommended that the EPA guidelines should use different approaches to count the fleas and ticks remaining on the vertebrate animals. A fine-toothed comb should be used to recover fleas; ticks still attached to the vertebrate animal may be discovered by carefully palpating the animal.  PG 66-67
The guideline has been revised to recommend infesting with ticks on day -2 instead of day -1. 

The guideline has been revised to allow counting of pests on the animals after the shampooing at up to 24 hours for fleas and 48 hours or ticks. 

All live pests found in the plugged tub that are still alive after the alloted time are added to the total live arthropod counts as well. 
309
a. The Panel noted that more specific guidelines may be needed for how to assess the mortality of fleas or ticks collected after being washed off or recovered from the vertebrate animal after the shampoo treatment as this can be a difficult endpoint to assess. The determination of tick mortality (whether the tick is attached or unattached to the vertebrate animal) can be evaluated by breathing on the pests and looking for movement, but there may be a delayed response.

Fleas and ticks may not die immediately after they are removed from a vertebrate animal, but these pests might die soon after being removed. The death of the removed arthropods may or may not be due to the pesticidal ingredient in the test product. The death of fleas and ticks after detachment from a vertebrate animal may occur because they no longer have access to blood or are not maintained in adequate environmental conditions. Fleas and ticks need appropriate levels of humidity and temperature to ensure that their mortality is not due to environmental factors alone. Conversely, the pest may be moribund and not recover due to the pesticidal treatment. The mortality of the recovered arthropods should be evaluated after the pest is removed from the vertebrate animal to determine if the pest will recover or die from treatment within a short timeframe. Due to the vulnerability of the arthropod while off the animal, the mortality of fleas should be evaluated within 24 hours. Because ticks might take up to 48 hours to recover, the mortality of ticks should be evaluated within a time window of 24-48 hours.  PG 67
The guideline has been revised to add that after the pest holding period, pests should be considered dead if probed with no movement.

The guideline has been revised to allow counting of pests on the animals after the shampooing or in petri dish after being removed from the water up to 24 hours later for fleas and 48 hours or ticks.  

To mitigate flea or tick deaths after removal from the tub due to environmental causes, the following statement has been added to the guideline: "all fleas and/or ticks found in the tub should be removed and moved to a clean container to dry in environmental conditions conducive to their survival for up to 24 hours for fleas and 48 hours for ticks." 
310
a. Soaps have a mortality factor against fleas and ticks. Shampoo may remove the waterproofing ability of the pest and make the arthropod more vulnerable to desiccation. Therefore, the Panel recommended that the guidelines should specify that the appropriate control to compare against the pesticidal shampoo is the shampoo without the active pesticidal ingredient.  PG 67
The guideline has been revised to remove the recommendation to use a non-pesticidal (vehicle) control for shampoos. See responses to comment #s 311 & 317 for more information.
311
a. The use of a control shampoo without the active ingredient is needed to determine whether the pesticidal agent had a greater effect than that of the shampoo alone.  PG 67
The guideline has been revised to remove the recommendation to use a non-pesticidal (vehicle) control for shampoos since the product formulation is registered and not the active ingredient. See response to comment #317 for more information.
312
a. Each vertebrate animal should only be shampooed according to the product claims, and not continuously.  PG 67
This question is for shampoo products that are claiming residual pesticidal claims. Therefore, animals are only shampooed according the label directions. Treatment day is on day zero of the study. 
313
a. Shampoos do not usually have long-term residual efficacy against fleas and ticks. A study by Beugnet et al. (2012) indicates that weekly shampooing of dogs reduced the number of fleas compared to controls, but the overall reduction in flea number never rose above 79.2%, which would not meet the EPA required efficacy of 90.0%. The Beugnet et al. (2012) study also showed that missing a weekly shampoo resulted in a decrease in efficacy from 68.2% to 34.8%. To assess residual activity of the shampoo on the vertebrate animal, it is important to wet and bathe the vertebrate animal, and then allow the vertebrate animal to fully dry for 24 hours before re-infesting with fleas or ticks on the following day. Counts for fleas should be made on the day following re-infestation and counts for ticks should be made two days after re-infestation to confirm that the active ingredient has not been removed from the animal.  PG 67-68
The guideline currently reflects this comment. If a registrant wants to make residual claims, they should re-infest according to the schedule listed in the above sections. The guideline has been revised to indicate that the animals should be dry before reinfestations. 
314
a. The Panel recommended that the guidelines should specify that a vertebrate animal should only be infested when their fur is dry (not with wet fur). In addition, it is not practical to try to comb count a vertebrate animal with wet fur.  PG 68
Both of these suggestions have been incorporated into the guideline. 
315
b. The Panel was not aware of any alternative methods that are adequate for determining mortality of fleas and ticks from shampoo products. However, the Panel recommended that the Agency guidelines be open to reviewing and including alternative methods for determining mortality of pests for shampoo products as these new alternative methods proceed through development and become available.  PG 68
The guideline indicates that novel methods may be discussed with the Agency or submitted as a protocol. 
316
b. In addition, the Agency should review the process that the U.S. Food and Drug Administration (FDA) uses to assess and approve topical insecticidal products available as prescription and over- the-counter drugs used on people and on animals. The FDA has approved insecticides as topical lotions and shampoos for human use against lice and topical insecticidal solutions for pets. Some useful assistance could be found by reviewing FDA protocols for approval of these drugs.  PG 68
The FDA does not have a published guideline to review on effectiveness of shampoo products and instead follows the EMA guideline, which was cited in this guideline.


Public Recommendations (docket ID # in parentheses): n/a


INSECTICIDAL SHAMPOO PRODUCTS (SECTION (M))

Charge Question 23
Typically, efficacy testing for pet products is conducted with a treated group and an untreated control group. To determine efficacy of shampoo products, the mechanical removal of arthropods during the shampoo process should be considered. Efficacy of a product would therefore be considered in relation to a group treated with non-insecticidal shampoo. In this guideline we propose a non-pesticidal/non-medicated shampoo as the control when testing insecticidal shampoo products.
a. Please discuss the value of using a non-pesticidal/non-medicated shampoo as the control 
    i. Does this provide valuable additional information about efficacy beyond an untreated control?
    ii. If so, please provide an appropriate methodology for the comparison of treatment vs. control groups, including references.
b. Should an untreated control also be provided for comparison with the non-pesticidal shampoo? If so, please describe the most appropriate untreated control (e.g., water only, massaging the animal). What is the value of the untreated control and is the informational value of an additional treatment group offset by the increased number of vertebrate animals used for testing?
c. If an untreated control group is included, how should efficacy be determined? Which control group (the non-insecticidal shampoo control or the untreated control) should be compared with the treated group to determine efficacy?

Panel Recommendations:
317
Charge Question #23 focuses on the utility of non-pesticidal/non-medicated shampoo control for determining the efficacy of its pesticidal/medicated shampoo counterpart. Unlike the inert ingredient(s) in other (i.e., non-shampoo) pesticidal products, the ingredients of shampoo are not "truly inert." Specifically, shampoo typically contains a variety of ingredients; these may include water, surfactants (surrounds dirt and oil so water can rinse them away), foaming agents (provide the "suds"), an acid (to balance pH), silicones (create smoothness and add shine), polyquaternium (like a fabric softener), panthenol (fatty alcohols and oils for hydration), fragrances, and preservatives. Since arthropod ectoparasites are relatively small and have large surface area to body weight ratios, anything that disrupts their waterproofing (epicuticular lipids, especially hydrocarbons) could cause desiccation and knockdown even in a relatively high humidity environment. Also, the mechanical action of shampooing can remove the ectoparasites from their hosts.  In other words, the non-pesticidal/non-medicated shampoo may kill the ectoparasites or hamper their ability to adhere onto their hosts. In fact, non-pesticidal/non‐medicated shampoo was shown to reduce the flea count on dogs (Beugnet et al., 2011).  However, in the same study by Beugnet et al. (2011), unlike the pesticidal/medicated shampoo employed, weekly shampooing was needed to maintain the flea count reduction (i.e., continuous control), and the efficacy of shampooing alone on the flea count reduction was shown to be below the performance standard (i.e., >90%) required by different regulatory agencies including the U.S. EPA. Hence, the value of a pesticide in a pesticidal/medicated shampoo product is to provide the continuous control (an important attribute of the product's commercial viability) that cannot be achieved through "treatment" by shampooing only. Accordingly, the utilities of the shampoo control are to control for the short term (i.e., non-continuous) pesticidal action of shampoo "inert" ingredients, as well as to assess the efficacy of pesticidal/medicated shampoo for providing the continuous control. Following a lengthy discussion, the general consensus among most Panel members, except for one Panel member (please see the last paragraph), was that a non-pesticidal/non-medicated shampoo control (i.e., formulation blank) should be used for determining the efficacy of the corresponding pesticidal shampoo product.  PG 69
The use of a non-pesticidal (vehicle) control has been changed to an untreated/negative control for shampoo products, since the Agency typically recommends testing on the end use products. 
318
As mentioned previously, a completely untreated control (i.e., host receiving no shampooing treatment) is not an appropriate control for the efficacy determination of pesticide in a pesticidal/medicated shampoo product because it does not control for the short-term pesticidal effect of shampoo "inert" ingredients. Similarly, control with water only or messaging "treatment" (i.e., placebos) provides no more information than the non-pesticidal/non‐medicated shampoo control (Taenzler et al., 2016).  Accordingly, the inclusion of such untreated control(s) will only increase the number of animals used.  PG 69
See response to comment #317 above. 
319
Most Panel members concluded that for evaluating the performance of the pesticide in a pesticidal/medicated shampoo product, only two groups are needed: a treatment group with the pesticide added to the pesticidal/medicated shampoo product and a shampoo control group with the formulation blank. Because the data collected are counts, a nonparametric test such as Mann-Whitney rank-sum test can be used to compare the treatment versus the control groups (Glantz, 2012).  In conclusion, the efficacy of pesticidal shampoo product should be determined relative to a non-pesticidal shampoo, or as close a formulation as possible to the candidate pesticidal shampoo.  PG 69
See response to comment #317 above.
320
As mentioned above, one Panel member, suggested that the decision on "what is the appropriate treatment of control group?" depends upon the question(s) to be answered and the product claims; specifically, what is (1) the efficacy of the product as sold (shampoo with the pesticide added), (2) the efficacy of the addition of the pesticide to a non-medicated shampoo base, and (or) (3) the need to know the efficacy of each separately - the non-medicated shampoo base alone and the pesticide alone. To address these questions, this Panel member proposed the following experimental designs (in the same order as the questions raised) for the product efficacy testing: (1) the use of water as the control group and the shampoo containing the pesticide as the treatment group, (2) the use of base shampoo without the pesticide for the control group and the shampoo containing the pesticide as the treatment group, and (3) the use of a three group design: water, the base shampoo and the base shampoo containing the pesticide. The last may need additional animals since two effects are being tested.  PG 69-70
A three-part design would increase the number of animals required for the test, and the Agency believes the revised study design, using an untreated control, is sufficient to determine shampoo product efficacy.

Public Recommendations (docket ID # in parentheses):
321
"Application methods. Insecticidal shampoo evaluations should be compared against a non-treated group or a non-treated group and a non-insecticidal shampoo group. This is because noninsecticidal shampoo may have an effect on some of these arthropods. Any effect on the insect population will translate into false positives in the data evaluation. If it must be considered to have a "placebo" than it is suggested that a water only control supplant the placebo. The trial must still include a totally untreated control group to evaluate for confounding of results. This results in a requirement for additional test animals... Change in shampoo testing by use of inert control, will cause issue since a non-medicated shampoo will also wash off some pests making the statistic much more difficult to show statistical significance. It is also a change to an unreal life situation" (0008)
The guideline has been revised to compare the treated shampoo against an untreated control only. 
322
"Testing using a non-pesticidal shampoo as a control results in a test for the effect of the pesticide only, not the pesticidal shampoo. The product that the consumer buys is the pesticidal shampoo, so the pesticidal shampoo should be tested against an untreated control" (0033)
The guideline has been revised to incorporate this comment.
337  -  out of order
"Representative sampling. Including a non-medicated, non-insecticidal shampoo in protocols for insecticidal shampoo testing will artificially impact pest counts and efficacy evaluation. Under field conditions, the user would not shampoo an animal with a non-insecticidal shampoo and then an insecticidal shampoo." (0008)
The guideline has been revised to incorporate this comment.


SIMULATED ENVIRONMENTAL CONDITIONS (SUNSHINE, WATER EXPOSURE AND SHAMPOOING CLAIM) (SECTION (N))

Charge Question 24
Discuss whether the timing for infesting vertebrate test animals after water exposure are adequate.
a. Are the number of infestations presented in the guideline adequate to support bathing and water immersion claims?
b. Discuss appropriate timing of infestations after vertebrate test animals are immersed in water. 
c. Are the methods for determining exposure to sunlight adequate? If not, please provide detailed methods for evaluating the effects of exposure to sunlight along with supporting references.

Panel Recommendations:
323
The Panel recommended that the reinfestation be specified as 24 hours after each exposure to water or bathing (e.g., bathing at day 6 with reinfestation at day 7, bathing at day 13 with reinfestation at day 14, etc.). In this way the animal will be dry prior to the reinfestation, even for animals with thick coats of fur.  PG 70
The guideline already states that "when infestations follow shampooing/wetting they should occur within 48 hours of shampooing/wetting. Animals should be dry before pest infestations." This should provide the lab some flexibility, while explaining the expectation. 
324
The Panel recommended the use of a "non-medicated" shampoo as the control, because this represents a baseline of control that a pet owner could achieve on their own without a shampoo containing a pesticide. The Panel did not recommend the use of the additional water exposure control, as the value of such studies does not justify the number of vertebrate animals needed for this.  PG 70
These tests are intended to demonstrate that a product used to treat an animal will still work after the animal has been bathed or exposed to water. The untreated group that has been bathed or exposed to water is the untreated control and is handled in the same manner as the treated group. Therefore, those test animals are still needed to serve as controls. No changes will be made to the guideline. 
325
a. Yes, the Panel found that the number of infestations presented in the guideline are adequate to support bathing and water immersion claims.  PG 70
No changes to the guideline are needed.
326
b. The Panel recommended that the reinfestation be specified as 24 hours after treatment. In this way the vertebrate animal will be dry prior to the reinfestation, even for animals with thick coats of fur.  PG 70
See response to comment #323. 
327
c. It is difficult to state unequivocally if the methods for determining exposure to sunlight are adequate or not. The current protocol does not really clarify if the product will be stable in sunlight. The vertebrate animals will be outside for part of the day but will be provided shade. Will the animals go into the sunlight? How long will they go into the sunlight? The UV index differs geographically, and this would potentially impact the results. Monitoring the sunlight each animal receives (using an on-collar monitor) would probably be necessary to standardize exposure and back up claims that a product continues to work after some specified amount of sunlight exposure. The Panel recommended that if the current approach is used, that the host be exposed to sunlight for two hours per day. This would potentially offer some standardization, although variation in weather may make such a regulation difficult or impossible to achieve. It would be of some additional value if the daily weather (average daily temperature, high temperature, low temperature, sunshine, rain, etc. was also recorded). While such information may not alter the approval process, these variables would help understand differences that might occur between studies.  PG 70-71
The suggestions for animals to be preferably exposed to direct sunlight for an average of 2 hours a day has been added as well as requesting daily records of the amounts of sunshine and rain to be submitted. 
328
The Panel noted that under the current proposed guidelines, vertebrate animals will spend prolonged periods outdoors and that some of them (i.e., the controls) could pick up additional or new pests while outdoors. As such, it would be valuable if the total number of pests were reported on the controls and treated animals at the end of the study, not just the number of pests that the animal was infested with.  If the outdoor activity of the pets results in accumulation of additional pests of the same species that they were infested with, the number of control animals might need to be increased to get a better assessment of this.  PG 71
 The "infestation period" of these studies are still conducted in a lab, therefore the animals should still be kept separated in outdoor runs during that time as indicated in the mortality testing. However, all animals in the study will have counts conducted, therefore, if control animals picked up" new infestations" it would be reported, and the study would be questionable or invalid. No changes to the guideline are needed. 
329
The Panel also considered using artificial sunlight on caged animals. This has potential for standardization of the "sunlight" exposure but requires protection of the animal's eyes and raises other animal welfare concerns. As such, the Panel did not recommend this approach.  PG 71
No changes to the guideline are needed. 
330
The Panel also suggested that EPA consider claims for sunlight stability using a post-approval field study with client-owned dogs under real world conditions with natural pest infestations in different geographic areas of the US and with a significant portion of the enrolled dogs in the study that spend at least half of their time outdoors with exposure to sunlight (collar monitors to record sunlight would be beneficial).  PG 71 
While the use of post-approval field studies with client-owned dogs would decrease the need of lab animals for this specific test, previously submitted field studies have not yielded useful data. Issues included inadequate or inconsistent pest pressure, lack of appropriate controls, and inaccurate pest counting by owners. However, protocols for field studies may be submitted if desired. 
331
Given the difficulty in understanding the amount of sunlight the animal is exposed to, the Panel recommended consideration of non-vertebrate testing of treated fur.  The treated fur could be exposed to sunlight and the residue of the active ingredient determined after different days or weeks of exposure to sunlight.  PG 71
This is a recommendation to consider in the future, however it is currently unknown how a product breaks down by sunlight in the fur vs. the lipid layer of the skin, where the products are often spread and regenerated into the fur. No changes to the guideline will be made. 
332
Suggested editorial changes:
The Panel noted that in section n(1)(i) for products with efficacy claims of 2-4 weeks, animals should be bathed on Days 6 and 20 so infestations can occur on appropriate Days 7 and 21.  The rationale is that the dogs that are wetted should be allowed to dry for ca. 24 hours prior to being re-infested the next day.  PG 71
The guideline has been revised based on the suggestion.
333
Suggested editorial changes:
Section (n)(1)ii. The Panel suggested the following revision "fur and skin through submerging (except the head) or showering the animal"  PG 71
The guideline has been revised based on the suggestion.

Public Recommendations (docket ID # in parentheses):
334
"General testing for bathing & water exposure claims. Separate considerations should be given to collar formulations to be evaluated at 7 days after the shampooing/wetting. These collars must be allowed enough time to re-release an efficacious amount of active ingredient and for that active ingredient to redistribute across the surface of the animal's body. The current proposed 48-hour challenge does not allow for this. This is significantly different from what has been recently accepted for shampooing/wetting claims for collars. Prior EPA approvals of shampooing/wetting claims for collars have allowed a seven-day interval between washing and re-infestation, this is a reasonable amount of time for the redistribution of active ingredients." (0008)
The purpose of the study is to demonstrate that the product still works after bathing or water exposure. Allowing 7 days for re-evaluation would be similar to reapplying the product and waiting for it to begin working again. If the product does not work within 48 hours of wetting or shampooing, it should not be considered "water-proof" or "shampoo-proof."
335
In section n(1)(i) for products with efficacy claims of 2-4 weeks animals should be bathed on days 6 & 20 so infestations can occur on appropriate days 7 & 21" (0012)
The guideline has been revised based on the suggestion.



References as Indicated by SAP and Public Commenters

Abbott, W.S. 1925. A method of computing the effectiveness of an insecticide. J. Econ. Entomol. 18:265-267.

Akov, S. 1982. Blood digestion in ticks. pp. 97-211. In: Obenchain, F.D., Galun, R. (Eds.), Physiology of Ticks, Pergamon Press, Oxford, New York, 509 p.

Andrade, J.J., Xu, G., Rich, S.M., 2014. A silicone membrane for in vitro feeding of Ixodes scapularis (Ixodida: Ixodidae). J. Med. Entomol. 51, 878-879.

Baker, C. F., McCall, J. W., McCall, S. D., Drag, M. D., Mitchell, E. B., Chester, S. T., and D. Larsen. 2016. Ability of an oral formulation of afoxolaner to protect dogs from Borrelia burgdorferi infection transmitted by wild Ixodes scapularis ticks. Comp. Immuno, Micro, and Infect Dis. 49:65-69.

Baldwin, R.W., and P.G. Koehler. 2007. Toxicity of commercially available household cleaners on cockroaches, Blattella germanica and Periplaneta americana. Fla. Entomol. 90:703-709.

Banks, B.J., Bishop, B.F., Evans, N.A., et al., 2000. Avermectins and flea control: structure-activity relationships and the selection of selamectin for development as an endectocide for companion animals. Bioorg Med. Chem. 8, 2017-2025.

Bar-Zeev, M., and S. Gothilf. 1972. Laboratory evaluation of flea repellents. J. Med. Entomol. 9:215-218.

Becskei, C., Cuppens, O., and S.P. Mahabir. 2018. Efficacy and safety of sarolaner against generalized demodicosis in dogs in European countries: a non-inferiority study. Vet. Dermatol. 29:203 - e72.

Becskei, C., D. Lin, D. Rugg, and T. Geurden. 2017. Speed of kill of a new spot-on formulation of selamectin plus sarolaner for cats against induced infestation with Ixodes ricinus. Vet Parasitol. 238:58-61.

Becskei, C., De Bock, F., Illambas, J., Cherni, J.A., Fourie, J.J., Lane, M., Mahabir, S.P., and R.H. Six. 2016. Efficacy and safety of a novel oral isoxazoline, sarolaner (SimparicaTM), for the treatment of sarcoptic mange in dogs. Vet Parasitol. 222:56-61.

Beier, J.C., Perkins, P.V., Wirtz, R.A., Koros, J., Diggs, D., Gargan, T.P., and D.K. Koech. 1988. Blood meal identification by direct enzyme-linked immunosorbent assay (ELISA) tested on Anopheles (Diptera: Culicidae) in Kenya. J Med Entomol. 25:9 - 16.

Bernillon, P., and F.Y. Bois. 2000. Statistical issues in toxicokinetic modeling: a bayesian perspective. Environmental health perspectives. 108(5):883-893.

Beugnet, F., Fourie, J., and K. Chalvet-Monfray. 2012. Comparative efficacy on dogs of a single topical treatment with fipronil/(S)-methoprene or weekly physiological hygiene shampoos against Ctenocephalides felis in a simulated flea-infested environment. Parasite. 19:153-158.

Beugnet, F., Doyle, V., Murray, M., and K. Chalvet-Monfray. 2011. Comparative efficacy on dogs of a single topical treatment with the pioneer fipronil/(S)-methoprene and an oral treatment with spinosad against Ctenocephalides felis. Parasite. 18:325-331.

Bohme, B., Bauer, B., Clausen, P.H., et al., 2014. In vitro feeding of Dermacentor reticulatus. In: Workshop on Ticks and Tick-borne Diseases. Federal Environment Agency and Federal Researcch Institute for Animal Health, Germany.

Boehme, B., Krull, C., Clausen, P.H., et al., 2014. OAKS: optimization and automation of artificial tick feeding. In: Workshop on Ticks and Tick-borne Diseases. Berlin.

Bonnet, S., Liu, X.Y., 2012. Laboratory artificial infection of hard ticks: a tool for the analysis of tick-borne pathogen transmission. Acarologia 52, 453-464.

Borges, J.C.M., Haddi, K., Oliveira, E., Andrade, B., Nascimento, V., Melo, T.S., Didonet, J., Carvalho, J.C., Rodrigues-Cangussu, A.S., Soares, I., Ascêncio, S., Raposo, R.B.N., and R. Aguiar. 2019. Mosquiticidal and repellent potential of formulations containing wood residue extracts of a Neotropical plant, Tabebuia heptaphylla. Industrial Crops and Products. 129:424-433.

Bosco, A., Leone, F., Vascone, R., Pennacchio, S., Ciuca, L., Cringoli, G., and L. Rinaldi. 2019. Efficacy of fluralaner spot-on solution for the treatment of Ctenocephalides felis and Otodectes cynotis mixed infestation in naturally infested cats. BMC Vet. Res. 15:29.

Bryant, E.H., and D.H. Reed. 1999. Fitness decline under relaxed selection in captive populations, Conservation Biology. 13(3):665-9.

Cavalleri, D., Murphy, M., Seewald, W., Drake, J., Nanchen, S., 2017a. A randomised, blinded, controlled field study to assess the efficacy and safety of lotilaner tablets (Credelio(TM)) in controlling fleas in client-owned dogs in European countries. Parasit. Vectors 10, 526.

Cavalleri, D., Murphy, M., Seewald, W., Drake, J., Nanchen, S., 2017b. A randomized, controlled study to assess the efficacy and safety of lotilaner (Credelio(TM)) in controlling ticks in client-owned dogs in Europe. Parasit. Vectors 10, 531.

Che Lah EF, Yaakop S, Ahamad M, Md Nor S. 2015. Molecular identification of blood meal sources of ticks (Acari, Ixodidae) using cytochrome b gene as a genetic marker. ZooKeys. (478):27 - 43.

Choudhuri, S., Patton, G.W., Chanderbhand, R.F., Mattia, A., and C.D. Klassen. 2018. From Classical Toxicology to Tox21: Some Critical Conceptual and Technological Advances in the Molecular Understanding of the Toxic Response Beginning from the Last Quarter of the 20th Century. Toxicol Sci. 161:5-22.

Chow, S.-C., and J.-P. Liu. 2008. Design and analysis of clinical trials: concepts and methodologies. Vol. 507. John Wiley & Sons.

Coles, T.B., and M.W. Dryden. 2014. Insecticide/acaricide resistance in fleas and ticks infesting dogs and cats. Parasites & Vectors. 7:8.

Croft, B.A., Wagner, S.W., and J.G. Scott. 1982. Multiple-and cross-resistances to insecticides in pyrethroid-resistant strains of the predatory mite, Amblyseius fallacis. Environ. Entomol. 11:161-164.

CVMP, 2006. PRAC-TIC Scientific Discussion. http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Scientific_Discussion/veterinary/000103/WC500062980.pdf.

CVMP, 2011. Comfortis Scientific Discussion. http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Scientific_Discussion/veterinary/002233/WC500110654.pdf.

CVMP, 2016. Guideline for the Testing and Evaluation of the Efficacy of Antiparasitic Substances for the Treatment and Prevention of Tick and Flea Infestation in Dogs and Cats. https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-testingevaluation-efficacy-antiparasitic-substances-treatment-prevention-tick-flea_en-0.pdf.

Dantas-Torres, F., and D. Otranto. 2011. Rhipicephalus sanguineus on dogs: relationships between attachment sites and tick developmental stages. Exp. Appl. Acarol. 53:389: - 397.

Deletre, E.; Martin, T.; Dumenil, C.; and F. Chandre. 2019. Insecticide resistance modifies mosquito response to DEET and natural repellents, Parasites & Vectors, 12:89. https://doi.org/10.1186/s13071-019-3343-9

Deletre, E., Schatz, B., Bourguet, D., Chandre, F., Williams, L., Ratnadass, A., and T. Martin. 2016. Prospects for repellent in pest control: current developments and future challenges. Chemoecology 26:127-142.

de Moura, S.T., da Fonseca, A.H., Nogueira Fernandes, C.G., et al., 1997. Artificial feeding of Amblyomma cajennense (Fabricius, 1787) (Acari: Ixodidae) through silicone membrane. Memorias do Inst. Oswaldo Cruz 92, 545-548.

Diggle, P., Heagerty, P., Liang, K.-Y., and S.L. Zeger, S. 2002. Analysis of longitudinal data - Second Edition. Oxford; New York: Oxford University Press.

Donohue, K., Khalil, S.M., Ross, E., Mitchell, R.D., and R.M. Roe. 2009. Male engorgement factor: Role in stimulating engorgement to repletion in the ixodid tick Dermacentor variabilis. J. Insect Physiol. 55:909 - 918.

Dryden, M., Carithers, D., Solanki, R., and S.J. Gross. 2016. Adulticidal efficacy of a formulation of fipronil/(S)-methoprene/cypermethrin against KS1 (Kansas State 1) Ctenocephalides felis felis (Bouché) fleas. Intern. J. Appl. Res. Vet. Med. 14:203-207.

Dryden, M., Carithers, D., McBride, A., Riggs, B., Smith, L., Davenport, J., Smith, V., Payne, P., and S.J. Gross. 2011. A comparison of flea control measurements methods for tracking flea populations in highly infested private residences in Tampa, FL, following topical treatment of pets with Frontline Plus (fipronil/(S) methoprene). Inter, J. Appl. Res. Vet. Med. 9:356

Dryden, M.W., Payne, P.A., McBride, A., Mailen, S., Smith, V., and D. Carithers. 2008. Efficacy of fipronil (9.8% w/w) + (S)-methoprene (8.8% w/w) and imidacloprid (8.8% w/w) + permethrin (44% w/w) against Dermacentor variabilis (American dog tick) on dogs. Veterinary therapeutics: research in applied veterinary medicine 9:15-25.

Dryden, M.W., Boyer, J.E., and V. Smith. 1994. Techniques for estimating on-animal populations of Ctenocephalides felis (Siphonaptera: Pulicidae). J. Med. Entomol. 31:631-634.

Dumont, P., Chester, T.S., Gale, B., Soll, M., Fourie, J.J., and F. Beugnet. 2015a. Acaricidal efficacy of a new combination of fipronil and permethrin against Ixodes ricinus and Rhipicephalus sanguineus ticks. Parasites & Vectors. 8:51.

Dumont, P., Liebenberg, J., Beugnet, F., and B. Fankhouser. 2015b. Repellency and acaricidal efficacy of a new combination of fipronil and permethrin against Ixodes ricinus and Rhipicephalus sanguineus ticks on dogs. Parasites & Vectors. 8:531.

Environmental Protection Agency (EPA), 1998. Health Effects Test Guidelines OPPTS 870.1200 Acute Dermal Toxicity. https://www.regulations.gov/document?D=EPA-HQ-OPPT-2009-0156-0004.

Environmental Protection Agency (EPA), 2010. Certifect Efficacy Review. https://archive.epa.gov/pesticides/chemicalsearch/chemical/foia/web/pdf/105402/105402-129121-106201-2010-11-03a.pdf.

Environmental Protection Agency (EPA). 2010. Toxicological Review of Dichloromethane (Methylene Chloride). EPA/635/R-10/003.

European Chemicals Agency (ECA). 2016. New Approach Methodologies in Regulatory Science, Proceedings of a Scientific Workshop, Helsinki, Finland, (C)European Chemicals Agency, ISBN: 978-92-9495-397-1

European Medicines Agency (EMA). 2016. Guideline for the testing and evaluation of the efficacy of antiparasitic substances for the treatment and prevention of tick and flea infestation in dogs and cats. http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2016/07/WC500210927.pdf

Falco, R.C., Daniels, T.J., Vinci, V., McKenna, D., Scavarda, C., and G.P. Wormser. 2018. Assessment of duration of tick feeding by the scutal index reduces need for antibiotic prophylaxis after Ixodes scapularis tick bites. Clin Infect Dis. 67:614 - 6.

FDA, 2001. Guidance Document 90. https://www.fda.gov/media/70349/download.

Fehrer, S.L., and R.E. Halliwell. 1987. Effectiveness of Avon's Skin-So-Soft as a flea repellent in dogs. J. Am. Animal Hospital Assoc. 23:217-220.

Ferrero, A.A., Werdin González, J.O., and C. Sánchez Chopa. 2006. Biological activity of Schinus molle on Triatoma infestans. Fitoterapia 77:381-383. https://doi.org/10.1016/j.fitote.2006.03.00

Fourie, J.J., Meyer, L., and E. Thomas. 2019. Efficacy of topically administered fluralaner or imidacloprid/moxidectin on dogs with generalized demodicosis. Parasites & Vectors, 12(1):59.

Fourie, J. J., Horak, I.G., de Vos, C., Deuster, K., and B. Schunak. 2015. Comparative speed of kill, repellent (anti-feeding) and acarcidal efficacy of an imidacloprid/flumethrin collar (Seresto(R)) and a fipronil/(S)-methoprene/Eprinomectin/praziquantel spot-on (Broadline(R)) against Ixodes ricinus (Linné, 1758) on cats. Parasitol Res. 114 (suppl 1):S109-S116.

Fourie, J.J., Joubert, A., Labuschagné, M., and F. Beugnet. 2014. New method using quantitative PCR to follow the tick blood meal and to assess the anti-feeding effect of topical acaricide against Rhipicephalus sanguineus on dogs. Comp Immunol Microbiol Infect Dis. May; 37(3):181-7.

Franta, Z., Frantova, H., Konvickova, J., Horn, M., Sojka, D., Mares, M., and P. Kopacek. 2010. Dynamics of digestive proteolytic system during blood feeding of the hard tick Ixodes ricinus. Parasites & Vectors 3:119.

Gebbia, J.A., Bosler, E.M., Evans, R.D., and E.M. Schneider.1995. Acquired resistance in dogs to repeated infestation with Ixodes scapularis (Acari: Ixodidae) reduces tick viability and reproductive success. Exp. Appl. Acarol. 19:593 - 605.

Gillett, J.D. 1967. Natural selection and feeding speed in a blood-sucking insect. Proceedings of the Royal Society of London. Series B, Biological Sciences. https://doi.org/10.1098/rspb.1967.0029

Glantz, S.A. 2012. Primer of biostatistics. 7th ed. New York: McGraw-Hill Medical. pp. xiv, 306 p.

Gregory, L.M., Zakson, M., Endris, R.G., and W.L. Shoop. 1995. A further comparison of the thumb-counting and comb-counting techniques used to determine Ctenocephalides felis infestation levels on dogs. Vet Parasitol. 56:349-352.

Groff, K., Bishop, P., 2017. Itching for change: embracing modern flea and tick product development. Regul. Toxicol. Pharmacol. 88, 349-355.

Halos, L., Lebon, W., Chalvet-Monfray, K., Larsen, D., and F. Beugnet. 2014. Immediateefficacy and persistent speed of kill of a novel oral formulation of afoxolaner (NexGardTM) against induced infestations with Ixodes ricinus ticks. Parasites & Vectors 7:452-452.

Heckenberg, K., Costa, S.D., Gregory, L.M., Michael, B.F., Endris, R.G., and W.L. Shoop. 1994. Comparison of thumb-counting and comb counting methods to determine Ctenocephalides felis infestations levels on dogs. Vet Parasitol. 53:153-157.

Heller-Haupt, A., and M.G.R. Varma. 1982. The effect of age on susceptibility of two species of African ticks (Ixodidae) to synthetic pyrethroids. Trop. Pest Manag. 28:385 - 392.

Hellmann, K., Knoppe, T., Krieger K., Stanneck, D., 2003. European multicenter field trial on the efficacy and safety of a topical formulation of imidacloprid and permethrin (Advantix(TM)) in dogs naturally infested with ticks and/or fleas. Parasitol. Res. 90, S125-S126.

Hilton, G., Odenkirchen, E., Panger, M., Waleko, G., Lowit, A., Clippinger, A.J., 2019. Evaluation of the avian acute oral and sub-acute dietary toxicity test for pesticide registration. Regul Toxicol Pharmacol. 105, 30-35.

Housenger, 2016. Letter to stakeholders on EPA Office of Pesticide Programs' goal to reduce animal testing. https://www.regulations.gov/document?D=EPA-HQ-OPP-2016-0093-0003.

ICCVAM (Interagency Coordinating Committee on the Validation of Alternative Methods). 2018. A Strategic Roadmap for Establishing New Approaches to Evaluate the Safety of Chemicals and Medical Products in the United States. https://dx.doi.org/10.22427/NTP-ICCVAM-ROADMAP2018

Jones, C.R., Brunner, J.L., Scoles, G.A. and J.P. Owen. 2015. Factors affecting larval tick feeding success: host, density and time. Parasites & Vectors. 8:340

Jones, C.J., Milne, D.E., Patterson, R.S., Schreiber, E.T., and J.A. Milio. 1992. Nectar Feeding by Stomoxys calcitrans (Diptera: Muscidae): Effects on Reproduction and Survival. Environ. Entomol. 21:141-147.

Karadzovska, D., Chappell, K., Coble, S., Murphy, M., Cavalleri, D., Wiseman, S., Drake, J., Nanchen, S. A randomized, controlled field study to assess the efficacy and safety of lotilaner flavored chewable tablets (Credelio(TM)) in eliminating fleas in client-owned dogs in the USA Parasit. Vectors 10, 528.

Kiszewski, E., Matuschka, F.R., and A. Spielmann. 2001. Mating strategies and spermiogenesis in ixodid ticks. Annu. Rev. Entomol. 46:167 - 182.

Koch, H.G., and J.R. Sauer. 1984. Quantity of blood ingested by four species of hard ticks (Acari: Ixodidae) fed on domestic dogs. Ann. Entomol. Soc. Am. 77:142-146.

Kramer, M., Font, E., 2017. Reducing sample size in experiments with animals: historical controls and related strategies. Biol. Rev. 92, 431-445.

Krober, T., Guerin, P., 2007a. The tick blood meal: from a living animal or from a silicone membrane? ALTEX 24 (Spec No), 39-41.

Krober, T., Guerin, P.M., 2007b. An in vitro feeding assay to test acaricides for control of hard ticks. Pest Manag. Sci. 63, 17-22.

Krober, T., Guerin, P.M., 2007c. In vitro feeding assays for hard ticks. Trends Parasitol. 23, 445-449.

Kuhnert, F., Diehl, P.A., Guerin, P.M., 1995. The life-cycle of the bont tick Amblyomma hebraeum in vitro. Int. J. Parasitol. 25, 887-896.

Kuhnert, F., 1996. Feeding of hard ticks in vitro: new perspectives for rearing and for the identification of systemic acaricides. ALTEX 13, 76-87.

Kuzner, J., Turk, S., Fourie, J., Grace, S., Marchiondo, A., and D. Rugg. 2013. Efficacy of a novel fipronil spot-on for the treatment and control of induced infestation of adult cat fleas (Ctenocephalides felis) and castor bean ticks (Ixodes ricinus) on cats. Parasitol. Res. 112:365-372.

Leftwich, P.T., Bolton, M., and T. Chapman. 2015. Evolutionary biology and genetic techniques for insect control. Evol. Appl.10.1111/eva.12280.

Lorusso, V., Dantas-Torres, F., Lia, R.P., Tarallo, V.D., Mencke, N., Capelli, G.C., Otranto, D., 2010. Seasonal dynamics of the brown dog tick, Rhipicephalus sanguineus, on a confined dog population in Italy. Med. Vet. Entomol. 24, 309-315.

Li, A.Y., Meng, H., Palma, K., 2015. In vitro membrane feeding of the lone star tick (Amblyomma americanum) and its use in evaluation of acaricidal compounds. In: Proceedings of Annual AAVPLIWC-ISEP Joint Conference 103.

Macaluso, K.R., Sonenshine, D.E., Ceraul, S.M., et al., 2001. Infection and transovarial transmission of rickettsiae in Dermacentor variabilis ticks acquired by artificial feeding. Vector Borne Zoonotic Dis. 1, 45-53.

Machadoa, M.A., Camposb, D.R., Lopesc, N.L., Barbieri Bastosd, I.P., Botelhoc, C.B., Correiae, T.R., Scotte, F.B., and J.I. Fernandes. 2018. Efficacy of afoxolaner in the treatment of otodectic mange in naturally infested cats. Vet Parasitol. 256:29-31.

Marchiondo, A.A., Holdsworth, P.A., Fourie, L.J., Rugg, D., Hellmann, K., Snyder, D.E., and M.W. Dryden. 2013. World Association for the Advancement of Veterinary Parasitology (W.A.A.V.P.) second edition: guidelines for evaluating the efficacy of parasiticides for the treatment, prevention and control of flea and tick infestations on dogs and cats. Vet Parasitol. 194:84 - 97.

Marchiondo, A.A. Holdsworth, P.A. Green, P., Blagburn, B.L., Jacobs, D.E., 2007. World Association for the Advancement of Veterinary Parasitology (W.A.A.V.P.) guidelines for evaluating the efficacy of parasiticides for the treatment, prevention and control of flea and tick infestation on dogs and cats. Vet. Parasitol. 145, 332-344.

McCall, J.W., Baker, C.F., Mather, T.N., Cheste, S.T., McCall, S.D., Irwin, J.P., Young, S.L., Cramer, L.G., and M.G. Pollmeier. 2011. The ability of a topical novel combination of fipronil, amitraz and (S) methoprene to protect dogs from Borrelia burgdorferi and Anaplasma phagocytophilum infections transmitted by Ixodes scapularis. Vet Parasitol. 179:335-342.

Meadows, C., Guerino, F., Sun, F., 2017. A randomized, blinded, controlled USA field study to assess the use of fluralaner topical solution in controlling canine flea infestations. Parasit. Vectors. 10(1):36.

Meola, R., Meier, K., Dean, S., et al., 2000. Effect of pyriproxyfen in the blood diet of cat fleas on adult survival, egg viability, and larval development. J. Med. Entomol. 37, 503-506.

Meyer, J.A., Disch, D., Cruthers, L.R., Slone, R.L., and R.G. Endris. 2003. Repellency and efficacy of a 65% permethrin spot-on formulation for dogs against Aedes aegypti (Diptera: Culicidae) mosquitoes. Veterinary therapeutics : research in applied veterinary medicine 4:135-144.

Musyoki, J.M., Osir, E.O., Kiara, H.K., et al., 2004. Comparative studies on the infectivity of Theileria parva in ticks fed in vitro and those fed on cattle. Exp. Appl. Acarol. 32, 51-67.

National Research Council. 2007. Division on Earth and Life Studies, Institute for Laboratory Animal Research, Board on Environmental Studies and Toxicology, Committee on Toxicity Testing and Assessment of Environmental Agents, National Academies Press. http://www/nap.edu/catalog/11970.html.

Navarro, D.M.F., Silva, P.C.B., Silva, M.F.R., Napoleao, T.H., and P.M.G. Paiva. 2013. Larvicidal activity of plant and algae extracts, essential oils and isolated chemical constituents against Aedes aegypti. J. Nat. Prod. 3:268-291. https://doi.org/10.2174/221031550304140328113732.

Navarro, C., Nadège, R., Crastes, N., Bonneau, S., 2016. Efficacy and safety of a permethrin-fipronilspot-on solution (Effitix) in dogs naturally infested by ticks in Europe. Biomed. Res. Int. 9498604.

Novartis Animal Health US Inc., 2003. Freedom of Information Summary: PROGRAM(R) (lufenuron) Flavor Tabs (R) and CAPSTAR(R) (nitenpyram) Tablets Flea Management SystemTM. NADA 141-205.

Panahi, Y., Poursaleh, Z., and M. Goldust. 2015. The efficacy of topical and oral ivermectin in the treatment of human scabies. Ann. Parasitol. 61 (1):11-6.

Pap, L., Sarkozy, P., Farkas, R., Bleicher, E., and A. Szego. 1997. Efficacy of some pyrethroids against a strain of the rabbit ear mite (Psoroptes cuniculi): an unusual cross-resistance pattern. Parasitol. Res. 83, 203-205.

Pasay, C.J., Arlian, L., Morgan, M., Vyszenski-Moher, D., Rose, A., Holt, D., Walton, S., and J. McCarthy, J. 2008. High-resolution melt analysis for the detection of a mutation associated with permethrin resistance in a population of scabies mites. Med. Vet. Entomol. 22: 82 - 88.

Pfister, K., and R. Armstrong. 2016. Systemically and cutaneously distributed ectoparasiticides: a review of the efficacy against ticks and fleas on dogs. Parasites & Vectors 9:436.

Portier, C.J., and N.L. Kaplan. 1989. Variability of safe dose estimates when using complicated models of the carcinogenic process. A case study: methylene chloride. Fundamental and applied toxicology: Official Journal of the Society of Toxicology 13:533-544.

Rao, J.N., and A.J. Scott. 1992. A simple method for the analysis of clustered binary data. Biometrics 48:577-585.

Richards, S.L., Langley, R., Apperson, C.S., and E. Watson. 2017. Do tick attachment times vary between different tick-pathogen systems? Environments. 4:37.

Rohdich, N., Roepke, R.K., Zschiesche, E. 2014. A randomized, blinded, controlled and multi-centered field study comparing the efficacy and safety of Bravecto (fluralaner) against Frontline (fipronil) in flea- and tick-infested dogs. Parasit. Vectors. 7, 83.

Ross, P.A., Axford, J.K., Richardson, K.M., Endersby-Harshman, N.M., and A.A. Hoffmann. 2017. Maintaining Aedes aegypti mosquitoes infected with Wolbachia. J Vis Exp. Aug 14:126.

Ross, P.A., Endersby-Harshman, N.M., and A.A. Hoffmann. 2019. A comprehensive assessment of inbreeding and laboratory adaptation in Aedes aegypti mosquitoes. Evolutionary Applications. 12:572-586.

Rupes, V., Chmela, J., Ledvinka, J., and D. Sládková. 1972. Effects of age on the susceptibility of larvae and nymphs of Ixodes ricinus (L.) to insecticides. Folia Parasitol. 19:379 - 382.

Rupes, V., Chmela, J., Ledvinka, J., and M. Privora. 1977. Effectiveness of fenitrothion for area control of ticks Ixodes ricinus (L.). Folia Parasitol. 24:63 - 72.

Rust, M.K., Waggoner, M., Hinkle, N.C., Mencke, N., Hansen, O., and M.B. Vaughn. 2002. Disruption of adult cat flea feeding by imidacloprid. Suppl. Compend. Contin. Educ. Pract. Vet. 24:4-5.

Satyal, P., Chuong, N.T.H., Hung, N.H., Sinh, L.H., The, P.V., Tai, T.A., Hien, V.T., and W.N. Setzer. 2019. Chemical composition, Aedes mosquito larvicidal activity, and repellent activity against Triatoma rubrofasciata of Severinia monophylla leaf essential oil. Parasitology Research. 118(3):733 - 742.

Schofield, S., and S. J. Torr. 2002. A comparison of the feeding behaviour of tsetse and stable flies. Med Vet Entomol 16:177-185.

Sims, S.R., and A.G. Appel. 2007. Linear alcohol ethoxylates: insecticidal and synergistic effects on German cockroaches (Blattodea: Blattellidae) and other insects. J. Econ. Entomol. 100:871-879.

Sims, S.R., Balusu, R.R, Ngumbi, E.N., and A.G. Appel. 2014. Topical and vapor toxicity of saturated fatty acids to the German cockroach (Dictyoptera: Blattellidae). J. Econ. Entomol. 107:758-763.

Sims, S.R., and T.E. O'Brien. 2011. Mineral oil and aliphatic alcohols: toxicity and analysis of synergistic effects on German cockroaches (Dictyoptera: Blattellidae). J. Econ. Entomol. 104:1680-1686.

Six, R.H., Becskei, C., Mazaleski, M.M., Fourie, J.J., Mahabir, S.P., Myers, M.R., and N. Slootmans. 2016. Efficacy of sarolaner, a novel oral isoxazoline, against two common mite infestations in dogs: Demodex spp. and Otodectes cynotis. Vet Parasitol. 222:62.

Six, R.H., Everett, W.R., Young, D.R., Carter, L., Mahabir, S.P., Honsberger, N.A., Myers, M.R., Holzmer, S., Chapin, S., and J.J. Rugg. 2016. Efficacy of a novel oral formulation of sarolaner (Simparica(TM)) against five common tick species infesting dogs in the United States. Vet Parasitol. 222:28-32.

Sojka, D. 2015. Tick blood digestion. In: Melhorn H (ed.) Encyclopedia of parasitology, Springer, Berlin Heidelberg. pp 1 - 4.

Sowilem, M.M., Kamal, H.A., and E. Khater, 2013. l. Life table characteristics of Aedes aegypti (Diptera: Culicidae) from Saudi Arabia. Trop Biomed.; 30(2):301-314.

Stanneck, D., Kruedewagen, E.M., Fourie, J.J., Horak, I.G., Davis, W., and K.J. Krieger. 2012a. Efficacy of an imidacloprid/flumethrin collar against fleas, ticks, mites and lice on dogs. Parasites & Vectors. 5:102.

Stanneck, D., Rass, J., Radelof, I., Kruedewagen, E., Le Sueur, C., Hellmann, K. and K. Krieger. 2012b. Evaluation of the long-term efficacy and safety of an imidacloprid 10%/flumethrin 4.5% polymer matrix collar (Seresto(R)) in dogs and cats naturally infested with fleas and/or ticks in multicenter clinical field studies in Europe. Parasites & Vectors. 5:66.

Su, L.-C., Huang, C.-G., Chang, S.T., Yang, S.-H., Hsu, S.-H., Wu, W.-J., and R.-N. Huang. 2013. An improved bioassay facilitates the screening of repellents against cat flea, Ctenocephalides felis 
(Siphonaptera: Pulicidae). Pest Manage. Sci. 70:264-270.

Szabo, M.P., and G.H. Bechara. 1999. Sequential histopathology at the Rhipicephalus sanguineus tick feeding site on dogs and guinea pigs. Exp Appl Acarol 23:915 - 28.

Szumlas, D.E. 2002. Behavioral responses and mortality in German cockroaches (Blattodea: Blattellidae) after exposure to dishwashing liquid. J. Econ. Entomol. 95:390-398.

Taenzler, J., Gale, B., Zschiesche, E., Roepke, R.K.A., and A.R. Heckeroth. 2016. The effect of water and shampooing on the efficacy of fluralaner spot-on solution against Ixodes ricinus and Ctenocephalides felis infestations in dogs. Parasites & Vectors, 9:233- 233.

Taenzler, J., de Vos, C., Roepke, R.K.A., and A.R. Heckeroth. 2018. Efficacy of fluralaner plus moxidectin (Bravecto(R) Plus spot-on solution for cats) against Otodectes cynotis infestations in cats. Parasites & Vectors, 11:595.

Tajeri, S., Razmi, G.R., 2011. Hyalomma anatolicum anatolicum and Hyalomma dromedarii Acari: Ixodidae) imbibe bovine blood in vitro by utilizing an artificial feeding system. Vet. Parasitol. 180, 332-335.

Thomas, R.S., Yang, R.S., Morgan, D.G., Moorman, M.P., Kermani, H.R., Sloane, R.A., O'Connor, R.W., Adkins, B., Jr., Gargas, M.L., and M.E. Andersen. 1996. PBPK modeling/Monte Carlo simulation of methylene chloride kinetic changes in mice in relation to age and acute, subchronic, and chronic inhalation exposure. Environmental Health Perspectives 104:858-865.

Uspensky, I., and I. Ioffe-Uspensky. 2006. Potential risk of pathogen transmission by acaricidepoisoned ticks. Int. J. Med. Microbiol. 296 (S1):217 - 224.

Vatta, A.F., Young, D.R., Everett, W.R., King, V.L., Cherni, J.A., von Reitzenstein, M., Holzmer, S.J., Chapin, S., and D. Rugg. 2019. Efficacy of a new topical formulation containing selamectin plus sarolaner against three common tick species infesting cats in the United States. Vet Parasitol. 270:S19-S25.

Wada, T., Ishiwata K., Koseki H., Ishikura T., Ugajin T., Ohnuma N., Obata K., Ishikawa R., Yoshikawa S., Mukai K., Kawano Y., Minegishi Y., Yokozeki H., Watanabe N., and H. Karasuyama. 2010. Selective ablation of basophils in mice reveals their nonredundant role in acquired immunity against ticks. J. Clin. Invest. 120:2867 - 2875.

Wade, S.E., Georgi, J.R., 1988. Survival and reproduction of artificially fed cat fleas, Ctenocephalides felis Bouche (Siphonaptera: Pulicidae). J. Med. Entomol. 25, 186-190.

Wang, C., Mount, J., Butler, J., Gao, D., Jung, E., Blagburn, B.L., and B. Kaltenboeck. 2012. Real-time PCR of the mammalian hydroxymethylbilane synthase (HMBS) gene for analysis of flea (Ctenocephalides felis) feeding patterns on dogs. Parasites & Vectors, 5:4.

Wang, H., Paesan, G.C., Nuttall, P.A. and A.G. Barbour. 1998. Male ticks help their mates to feed. Nature. 391:753 - 754.

Waladde, S.M., Young, A.S., Morzaria, S.P., 1996. Artificial feeding of Ixodid ticks. Parasitol. Today 12, 272-278.

Wells, K., Beaucournu, J.C., Durden, L.A., Petney, T.N., Lakim, M.B., O'Hara, R.B., 2012. Ectoparasite infestation patterns of domestic dogs in suburban and rural areas in Borneo. Parasitol. Res. 111, 909-919

Wengenmayer, C., Williams, H., Zschiesche, E., Moritz, A., Langenstein, J., Roepke, R.K., and A.R. Heckeroth. 2014. The speed of kill of fluralaner (Bravecto(TM)) against Ixodes ricinus ticks on dogs. Parasite & Vectors. 7:525.

Williams, H., Young, D.R., Qureshi, T., et al., 2014. Fluralaner, a novel isoxazoline, prevents flea (Ctenocephalides felis) reproduction in vitro and in a simulated home environment. Parasit. Vectors 7 (275).

Yeap, H.L., Mee, P., Walker, T., Weeks, A.R., O'Neill, S.L., Johnson, P., Ritchie, S.A., Richardson, K.M., Doig, C., Endersby, N.M., and A.A. Hoffmann. 2011. Dynamics of the "popcorn" Wolbachia infection in outbred Aedes aegypti informs prospects for mosquito vector control. Genetics. 187(2):583-950.

Yeh, M.T., Bak, J.M., Hu, R., Nicholson, M.C., Kelly, C., and T.N. Mather. 1995. Determining the duration of Ixodes scapularis (Acari: Ixodidae) attachment to tick-bite victims. J. Med. Entomol. 32:853 - 858.

Zackson, M., Gregory, L.M., Endris, R.G., and W.L. Shoop. 1995. Effect of combing time on cat flea (Ctenocephalides felis) recovery from dogs. Vet. Parasitol. 60:149-153.

